Pharmacological studies of malaria in pregnancy, …research-repository.uwa.edu.au/files/3222408/Salman_Sam...Pharmacological studies of malaria in pregnancy, infancy and childhood

Pharmacologicalstudiesofmalariainpregnancy,infancyandchildhoodinPapuaNewGuinea

Sam Salman MBBS (Hons)

This thesis is presented for the degree of Doctor of Philosophy of the University of Western Australia as a component of a MBBS/PhD combined degree

School of Medicine and Pharmacology 2012

II

III

Dedicated to the Baha’i youth in Iran, who continue to be denied access to tertiary education

IV

V

Declaration

This thesis contains the details of five published pharmacokinetic studies performed in Papua

New Guinea. The majority of the work in four of the five studies (Chapters 2, 3, 4 and 5) was

performed by Sam Salman, while his contribution to the fifth (Chapter 6) was substantial. This

contribution is indicated by first authorship on the first four works P

1‐4P and second authorship of

the final P

5P. The contribution of authors to each of the studies is detailed in section i (page VII)

and an acknowledgement to all those involved in the work relating to the thesis is provided in

section ii (page XIII).The co‐authors of these papers have given permission for them to be

included in this thesis.

This thesis is presented for the Doctor of Philosophy component of a combined Bachelor of

Medicine and Bachelor of Surgery/ Doctor of Philosophy (MBBS/PhD) degree at the University

of Western Australia in the School of Medicine and Pharmacology. This degree comprised two

full‐time research years in 2007 and 2008 and then part‐time research combined with the last

three years of the full‐time MBBS course. Academic supervision for this work was provided by

Winthrop Professor Timothy M. E. Davis and Emeritus Professor Kenneth F. Ilett as

coordinating and secondary supervisors, respectively. No part of this thesis has been

presented for a degree at the University of Western Australia or any other university.

In addition to the publications included as a part of this thesis, the candidate also contributed

to other published works during the time of his candidature listed in section ii page (XVIII).

These were not included in this thesis as either the contribution (primarily population

pharmacokinetic analysis) was not as significant as those included in the thesis or the

publications were not related to malaria, the pharmacology of which is the topic of this thesis.

________________________________ __________

Candidate Signature Date

Dr Sam Salman

_________________________________ __________

Coordinating supervisor Signature Date

W/Prof Timothy M E Davis

VI

VII

i 139B139BAuthorcontributionsandethicalapprovalsandfunding

i.a 151B151BPharmacokineticPropertiesofAzithromycininPregnancy

Percentage Author name Nature of contribution

50.0% Sam Salman Responsible for study concept

Responsible for study design

Responsible for ethical approval

Supervised and performed field work

Developed the assay for azithromycin

Performed the azithromycin assays

Developed and interpreted the pharmacokinetic model

Analysed and interpreted the clinical data

Prepared the first draft of the manuscript

4.0% Stephen J Rogerson Responsible for study concept

3.0% Kay Kose Performed field work

3.0% Susan Griffin Performed field work

3.0% Servina Gomorai Performed field work

3.0% Francesca Baiwog Performed field work

3.0% Josephine Winmai Performed field work

3.0% Josin Kandai Performed field work

3.0% Harin A Karunajeewa Responsible for study concept

Assisted with the writing of the manuscript

4.0% Sean J O’Halloran Developed the assay for azithromycin


3.0% Peter Siba Responsible of translation of study findings in to policy in Papua New Guinea

6.0% Kenneth F Ilett

(Secondary PhD supervisor)

Responsible for study concept

Assisted with the pharmacokinetic model


4.0% Ivo Mueller Responsible for study concept


5.0% Timothy M E Davis

(Coordinating PhD supervisor)





Funding sources: National Health and Medical Research Council (NHMRC) of Australia (grant

458555) and was supported and endorsed by the MiP consortium, which is funded through a

grant from the Bill and Melinda Gates Foundation to the Liverpool School of Tropical Medicne.

Ethical approvals: Medical Research Advisory Committee of Papua New Guinea (reference

07.24) and Human Ethics Research Committee at the University of Western Australia

(reference RA/4/1/1871).

VIII

i.b 152B152BPharmacokineticPropertiesofConventionalandDouble‐Dose

Sulfadoxine‐PyrimethamineGivenasIntermittentPreventive

TreatmentinInfancy






Adapted previously published assays for pyrimethamine, sulfadoxine and N‐acetyl‐sulfadoxine

Performed the pyrimethamine, sulfadoxine and N‐acetyl‐sulfadoxine assays






3.0% Nolene Pitus Performed field work

3.0% Josephine Winmai Performed field work

5.0% Brioni Moore Supervised and performed field work

3.0% Peter Siba Responsible for translation of study findings in to policy in Papua New Guinea















Funding: IPTi Consortium and facilities utilised were developed with support from the National

Health and Medical Research Council (NHMRC) of Australia (grant 458555).

Ethical approval: Medical Research Advisory Committee of Papua New Guinea (reference

08.04).

IX

i.c 153B153BPopulationPharmacokineticsofArtemether,Lumefantrine,and

TheirRespectiveMetabolitesinPapuaNewGuineanChildrenwith

UncomplicatedMalaria






Adapted previously published assay for lumefantrine

Developed the assay for desbutyl‐lumefantrine

Performed the lumefantrine and desbutyl‐lumefantrine assays




7.0% Madhu Page‐Sharp Adapted previously published assay for artemether and dihydroartemisinin

Performed the artemether and dihydroartemisinin assays



















Funding: The National Health and Medical Research Council (NHMRC) of Australia (grant

634343).


05.02).

X

i.d 154B154BApharmacokineticcomparisonoftwopiperaquine‐containing

artemisinincombinationtherapiesinPapuaNewGuineanchildren

withuncomplicatedmalaria






Performed the more recent piperaquine assays




10.0% Madhu Page‐Sharp Developed the original assay for piperaquine

Performed the original group piperaquine assays

Developed the assay for artemisinin

Performed the artemisinin assays


4.0% Kevin T Batty Assisted with assay development for artemisinin



















634343).


05.02).

XI

i.e 155B155BArtemisinin‐naphthoquinecombinationtherapyforuncomplicated

paediatricmalaria:Apharmacokineticstudy


35.0% Kevin T Batty Responsible for study concept

Assisted with assay development for artemisinin and naphthoquine

Developed and interpreted the original non‐compartmental pharmacokinetic model


30.0% Sam Salman Responsible for study design


Developed and interpreted the population pharmacokinetic model


4.5% Brioni R Moore Responsible for ethical approval



3.0% John Benjamin Supervised and performed field work


4.5% Sook Ting Lee Developed the original naphthoquine assay

Performed the original naphthoquine assays


4.5% Madhu Page‐Sharp Developed the artemisinin and naphthoquine assay

Performed the artemisinin and naphthoquine assays


2.0% Nolene Pitus Performed field work



Assisted with the population pharmacokinetic model



2.0% Francis W Hombhanje Responsible for study concept









634343).


05.02).

XII

XIII

ii 140B140BAcknowledgements

First and foremost I would like to thank my wife, Lila, who when I embarked on this

combined degree was a friend and has been there for me on every stage of this

journey. Thank you for your patience, support and love, without it I couldn’t have

finished this work.

I would also like to acknowledge my mother and my sister who have always been there

for me and my father, who taught me to always search for new knowledge. A big thank

you to my grandmother, Iran Rezvani, the matriarch of my family, your strength flows

down to us all.

To my extended family and my friends, thank you for your encouragement and well

wishes.

Clinical studies such as those presented in this thesis are not possible with the

contribution of a range of people; I wish to thank them below:

My supervisors Winthrop Professor Timothy M. E. Davis and Emeritus Professor

Kenneth F. Ilett, for your guidance, wisdom and mentorship and for allowing

me to be a part of your research team, it has been a privilege and a pleasure. I

hope that there will be many more years of friendship and work together.

The field team at Alexishafen Health Centre in Papua New Guinea who did the

difficult job of data collection, Susan Griffin, Kay Kose, Servina Gomorai, Nolene

Pitus, Josephine Winmai, Francesca Baiwog, Josin Kandai [nurses], Christine

Kalopo [microscopist] and Bernard (“Ben”) Maamu [driver] for your hard work,

commitment to the work and acceptance into your team.

Dr Madhu Page‐Sharp, research fellow at UWA and Curtin University, for your

guidance with HPLC, assay development and for your generosity in performing

some of the drug assays required for this thesis.

Dr Laurens Manning for your onsite clinical supervision, mentorship with my

medical degree and also for your friendship. You invited me into your home

and made me feel like a part of your own family.

XIV

Dr Harin Karunajeewa for your guidance with respect to the field work and for

your role in co‐ordinating the original piperaquine study.

Dr Sean O’Halloran, Laboratory Manager at PathWest ‐ Clinical Pharmacology

and Toxicology Laboratory, for your guidance with LC‐MS/MS and for allowing

me to perform my assays around the already busy schedule for routine clinical

samples on the LC‐MS/MS equipment.

Dr Moses Laman and Dr Michele Senn for your onsite clinical supervision.

Dr Brioni Moore, Medical Scientist at Curtin University, for your help in

completing recruitment in the infant trial and for co‐ordinating the latter

naphthoquine trial.

Dr Sook Ting Lee for your work in the original Naphthoquine trial and for being

a lab buddy in the first year of the research.

Professor Stephen Rogerson, University of Melbourne, for your expert advice in

the azithromycin trial.

Associate Professor Kevin Batty, Curtin University, for supervising the use of the

LC‐MS at Curtin University School of Pharmacy.

Mary Anne Townsend, Senior Medical Scientist at PathWest ‐ Biochemistry

Department, for your assistance in biochemical assays for the study in infants.

Dr Ivo Muller, the then head of Vector Borne Diseases at Papua New Guinea

Institute of Medical Research, Dr Peter Siba, director of the Papua New Guinea

Institute of Medical Research, and to John Taime, Yagaum site manager for

Papua New Guinea Institute of Medical Research, for welcoming me into your

country, your assistance in providing local resources and support in my time in

Papua New Guinea.

The microscopy and data entry teams at the Yagaum branch of the Papua New

Guinea Institute of Medical Research for your work in providing and managing

data essential to the studies.

Sr Valsi Kurian and the staff of Alexishafen Health Centre for your kind co‐

operation and allowing us to use your facilities for the studies.

The Faculty of Medicine, Dentistry and Health Sciences (in particular Dr Jan

Dunphy) for establishing an A & A Saw Scholarship to assist me in the combined

years of the degree and in conjunction with the Raine Medical Research

XV

Foundation for awarding me with a MBBS/PhD scholarship in the full time years

of the research.

And finally a big thank you to participants of the studies, their families and their

communities for agreeing to be a part of this research, without whom there

would be nothing to report.

My thoughts are with the family of Servina Gomorai, dedicated research nurse, who

died of cancer during the course of this thesis.

My thesis is dedicated to the Baha’i youth in Iran, who continue to be denied access to

tertiary education. Were it not for the courage of my parents to escape and find refuge

in Australia, I would be numbered with them and would not have had this wonderful

adventure.

Some members the field team in the study clinic in Alexishafen. (L‐R) Sitting: Christine Kalopo, Josephine Winmai, Susan Griffin, Kay Kose, Servina Gomorai. Standing: Sam Salman, Nolene Pitus, Ben Maamu.

XVI

XVII

iii 141B141BPublications,presentationsandprizes

iii.a 156B156BPublicationsrelatedtothesis

Salman, S., S. J. Rogerson, K. Kose, S. Griffin, S. Gomorai, F. Baiwog, J. Winmai, J. Kandai, H. A.

Karunajeewa, S. J. O'Halloran, P. Siba, K. F. Ilett, I. Mueller, and T. M. Davis. 2010.

Pharmacokinetic properties of azithromycin in pregnancy. Antimicrob Agents Chemother

54:360‐6. (Chapter 2)

Salman, S., S. Griffin, K. Kose, N. Pitus, J. Winmai, B. Moore, P. Siba, K. F. Ilett, I. Mueller, and T.

M. Davis. 2011. The pharmacokinetic properties of conventional and double‐dose sulfadoxine‐

pyrimethamine given as intermittent preventive treatment in infancy. Antimicrob Agents

Chemother. 55:1693‐700 (Chapter 3)

Salman, S., M. Page‐Sharp, S. Griffin, K. Kose, P. M. Siba, K. F. Ilett, I. Mueller, and T. M. Davis.

2011. Population pharmacokinetics of artemether, lumefantrine, and their respective

metabolites in Papua New Guinean children with uncomplicated malaria. Antimicrob Agents

Chemother 55:5306‐13. (Chapter 4)

Salman, S., M. Page‐Sharp, K. T. Batty, K. Kose, S. Griffin, P. Siba, K. F. Ilett, I. Mueller, and T. M.

E. Davis. 2012. A pharmacokinetic comparison of two piperaquine‐containing artemisinin

combination therapies in Papua New Guinean children with uncomplicated malaria.

Antimicrob Agents Chemother. 56:3288‐97 (Chapter 5)

Batty, K. T., S. Salman, B. R. Moore, J. Benjamin, S. T. Lee, M. Page‐Sharp, N. Pitus, K. F. Ilett, I.

Mueller, F. W. Hombhanje, P. Siba, and T. M. Davis. 2012. Artemisinin‐naphthoquine

combination therapy for uncomplicated pediatric malaria: A pharmacokinetic study.

Antimicrob Agents Chemother. 56:2472‐84 (Chapter 6)

XVIII

iii.b 157B157BOtherpublicationsduringcandidature

Salman, S. *, B. Hullet*, S. J. O'Halloran, D. Peirce, K. Davies, and K. F. Ilett. 2012. Development

of a Population Pharmacokinetic Model for Parecoxib and Its Active Metabolite Valdecoxib

After Parenteral Parecoxib Administration in Children. Anesthesiology 116. (*equal first

authorship)

Salman, S., S. K. Sy, K. F. Ilett, M. Page‐Sharp, and M. J. Paech. 2011. Population

pharmacokinetic modelling of tramadol and its O‐desmethyl metabolite in plasma and breast

milk. Eur J Clin Pharmacol.

Paech, M. J., S. Salman, K. F. Ilett, S. J. O'Halloran, and N. A. Muchatuta. 2012. Transfer of

Parecoxib and Its Primary Active Metabolite Valdecoxib via Transitional Breastmilk Following

Intravenous Parecoxib Use After Cesarean Delivery: A Comparison of Naive Pooled Data

Analysis and Nonlinear Mixed‐Effects Modeling. Anesth Analg 114:837‐44.

Karunajeewa, H. A., S. Salman, I. Mueller, F. Baiwog, S. Gomorrai, I. Law, M. Page‐Sharp, S.

Rogerson, P. Siba, K. F. Ilett, and T. M. Davis. 2009. Pharmacokinetic properties of sulfadoxine‐

pyrimethamine in pregnant women. Antimicrob Agents Chemother 53:4368‐76.

Karunajeewa, H. A., S. Salman, I. Mueller, F. Baiwog, S. Gomorrai, I. Law, M. Page‐Sharp, S.

Rogerson, P. Siba, K. F. Ilett, and T. M. Davis. 2010. Pharmacokinetics of chloroquine and

monodesethylchloroquine in pregnancy. Antimicrob Agents Chemother 54:1186‐92.

Wong, R. P., S. Salman, K. F. Ilett, P. M. Siba, I. Mueller, and T. M. Davis. 2011. Desbutyl‐

lumefantrine is a metabolite of lumefantrine with potent in vitro antimalarial activity that may

influence artemether‐lumefantrine treatment outcome. Antimicrob Agents Chemother

Benjamin, J., B. Moore, S. T. Lee, M. Senn, S. Griffin, D. Lautu, S. Salman, P. Siba, I. Mueller, and

T. M. Davis. 2012. Artemisinin‐naphthoquine combination therapy for uncomplicated

paediatric malaria: A tolerability, safety and preliminary efficacy study. Antimicrob Agents

Chemother. 56:2465‐71

Manning, L., M. Laman, M. Page‐Sharp, S. Salman, I. Hwaiwhanje, N. Morep, P. Siba, I. Mueller,

H. A. Karunajeewa, and T. M. Davis. 2011. Meningeal inflammation increases artemether

XIX

concentrations in cerebrospinal fluid in Papua New Guinean children treated with

intramuscular artemether. Antimicrob Agents Chemother 55:5027‐33.

Karunajeewa, H. A., I. Mueller, M. Senn, E. Lin, I. Law, P. S. Gomorrai, O. Oa, S. Griffin, K. Kotab,

P. Suano, N. Tarongka, A. Ura, D. Lautu, M. Page‐Sharp, R. Wong, S. Salman, P. Siba, K. F. Ilett,

and T. M. Davis. 2008. A trial of combination antimalarial therapies in children from Papua

New Guinea. N Engl J Med 359:2545‐57.

XX

iii.c 158B158BPosterpresentations

Salman, S., M. Page‐Sharp, S. Griffin, K. Kose, I. Mueller, K. F. Ilett, and T. M. E. Davis. 2011.

Population Pharmacokinetics of artemether, lumefantrine and their respective metabolites in

Papua New Guinean children with uncomplicated malaria, Australasian Society of Clinical and

Experimental Pharmacologists and Toxicologists (ASCEPT), Perth.

Salman, S., M. Page‐Sharp, S. Griffin, K. Kose, N. Pitus, J. Winmai, B. Moore, P. Siba, K. F. Ilett, I.

Mueller, and T. M. Davis. 2011. The pharmacokinetic properties of standard and double dose

sulfadoxine‐pyrimethamine(Fansidar®) in infants, Students in Health and Medical Research

Conference (SHMRC), Perth.

Salman, S., M. Page‐Sharp, S. Griffin, K. Kose, N. Pitus, J. Winmai, B. Moore, P. Siba, K. F. Ilett, I.

Mueller, and T. M. Davis. 2011. The pharmacokinetic properties of standard and double dose

sulfadoxine‐pyrimethamine(Fansidar®) in infants, UWA School of Medicine and Pharmacology

Annual Research Symposium, Perth.

Salman, S., H. Karunajeewa, I. Law, I. Muller, T. M. E. Davis, and K. F. Ilett. 2009.

Pharmacokinetics of Chloroquine in Pregnant and Non‐pregnant Women in Papua New

Guinea, Population Approach Group in Australia and New Zealand (PAGANZ), Newcastle.

Salman, S., S. Rogerson, S. J. O’Halloran, I. Muller, T. M. E. Davis, and K. F. Ilett. 2009.

Pharmacokinetic properties of azithromycin in pregnancy, Students in Health and Medical

Research Conference (SHMRC), Perth.

Salman, S., S. Rogerson, S. J. O’Halloran, I. Muller, T. M. E. Davis, and K. F. Ilett. 2009.

Pharmacokinetic properties of azithromycin in pregnancy, UWA School of Medicine and

Pharmacology Annual Research Symposium, Perth.

XXI

iii.d 159B159BPrizes

Finalist for Neville Percy poster prize at Australasian Society of Clinical and Experimental

Pharmacologists and Toxicologists (ASCEPT 2011) for poster: “Population Pharmacokinetics of

artemether, lumefantrine and their respective metabolites in Papua New Guinean children

with uncomplicated malaria”

Best Poster Presentation by a Postgraduate Student/Post doctorate at UWA School of

Medicine and Pharmacology Annual Research Symposium 2011 for poster: “Population

pharmacokinetics of artemether, lumefantrine and their respective metabolites in Papua New

Guinean children with uncomplicated malaria”

Special commendation in Higher Degree by Research Achievements awards (Clinical Medicine

and Dentistry discipline) at UWA for publication: “Pharmacokinetic properties of azithromycin

in pregnancy”

Best Clinical Research Poster Presentation by a Postgraduate Student/Post doctorate at UWA

School of Medicine and Pharmacology Annual Research Symposium 2009 for poster:

“Pharmacokinetic properties of azithromycin in pregnancy”

Best Methodology and Study Design at Students in Health and Medical Research Conference

2009 for poster: “Pharmacokinetic properties of azithromycin in pregnancy”

XXII

XXIII

iv 142B142BAbstract

With half the world’s population still at risk of malaria, it remains one of the most

important global health concerns. In highly endemic areas such as coastal Papua New

Guinea (PNG), adults develop immunity to symptomatic infection, while pregnant

women, infants and children bear the burden of clinical disease. Antimalarial drugs still

play an important role in the treatment and prevention of malaria. In pregnancy and

infancy prevention of disease is afforded, to some extent, by the use of Intermittent

Preventive Treatment (IPT). In childhood, effective treatments are required to prevent

recrudescence and early re‐infection. To enable optimal dosing of pharmacological

therapy, studies performed in these specific at‐risk groups are required.

This primary aims of this thesis were, in samples of at‐risk populations, to describe the

pharmacokinetic (PK) properties of a number of antimalarial drugs using a population

approach and to provide preliminary information regarding their efficacy, safety and

tolerability. These studies were intended to guide future large clinical trials and assist

in determining health policies.

The first of these studies evaluated the PK of azithromycin (AZI) in pregnant and non‐

pregnant women. AZI is one of the few antimalarial drugs known to be safe in

pregnancy and it is conventionally given with chloroquine or sulfadoxine‐

pyrimethamine (SP). The effect of pregnancy on the PK parameters of AZI was not

large enough to justify a dose adjustment. A timed single blood sample that could be

used as a surrogate for overall exposure was identified. The preliminary tolerability

and efficacy data from this study were used in developing the drug regimen for a large

study of IPT in pregnancy currently underway in PNG.

Infants were the participants in the second study in which the PK of conventional and

double dose SP were investigated. Hepatic and renal maturation were incorporated

into the PK model for pyrimethamine, sulfadoxine and N‐acetylsulfadoxine (a

metabolite of sulfadoxine). Exposure was significantly higher in the double dose group

despite a slight reduction in the relative bioavailability of sulfadoxine. Preliminary data

on the safety of a double dose were obtained. The findings support the evaluation of

the efficacy of a double dose regimen for IPT in infants.

XXIV

The last three studies were performed in children and evaluated the use of several

artemisinin combination therapies (ACTs) namely, artemether/lumefantrine (AL),

artemisinin/piperaquine base (ART/PQ)and artemisinin/naphthoquine (ART/NQ).

AL was the first ACT recommended by the World Health Organisation (WHO). The

study found that a subset of children may be under‐dosed as higher mg/kg doses were

required to produce the same exposure to lumefantrine, artemether and

dihydroartemisinin (DHA, an active metabolite of artemether) when compared with

adult doses. It was the first study to describe the population PK of desbutyl‐

lumefantrine, an active metabolite of lumefantrine. The results from this study were

taken into consideration when AL was chosen as first‐line in the treatment of

uncomplicated malaria in PNG.

The next study in this set assessed ART/PQ, an ACT not yet recommended by the WHO

but available in the private sector. The PK of PQ in this combination were compared to

those of a historical study of DHA/PQ phosphate, which had been performed at the

same location. Although there were no clinically significant differences in PQ PK

between formulations, the low ART dose and the reduced ART exposure with

successive doses raise concerns regarding the use of this combination. The results of

this study suggest that an extended dose regimen should be investigated.

The final study was a PK evaluation of ART/NQ, another ACT commercially available

but not yet recommended by the WHO. Three distinct dose regimens were tested in

similar samples of children, and the PK differences between them were analysed. This

study provided the first information of NQ disposition in children as well as providing

additional data on the PK of multiple doses of ART. ART/NQ dosing in a large scale

efficacy trial that is currently being carried out in PNG was based on the PK data from

this study.

In summary, this thesis describes studies in samples of at‐risk individuals in PNG that

add vital information to an often sparse literature on the pharmacology of these

important antimalarial drugs. Future studies of these treatments will be enhanced as a

result.

XXV

XXVI

XXVII

v 143B143BTableofcontents

8Ti8T 8TAuthor contributions and ethical approvals and funding8T .................................................................. VII

8Tii8T 8TAcknowledgements8T .......................................................................................................................... XIII

8Tiii8T 8TPublications, presentations and prizes8T ............................................................................................ XVII

8Tiv8T 8TAbstract8T........................................................................................................................................... XXIII

8Tv8T 8TTable of contents8T ........................................................................................................................... XXVII

8Tvi8T 8TAbbreviations8T .................................................................................................................................. XXXI

8Tvii8T 8TAntimalarial drugs and combinations used in this thesis8T .............................................................. XXXV

8Tviii8T 8TList of Tables8T ................................................................................................................................ XXXVII

8Tix8T 8TList of Figures8T ................................................................................................................................ XXXIX

8Tx8T 8TPreface8T ............................................................................................................................................. XLIII

8T18T 8TGeneral Introduction8T ........................................................................................ 1

8T1.18T 8TMalaria 8T ................................................................................................................................... 1

8T1.1.18T 8TGlobal snapshot 8T .................................................................................................................. 1

8T1.1.28T 8TParasitology8T ........................................................................................................................ 1

8T1.1.38T 8TMalaria in Papua New Guinea 8T ............................................................................................ 3

8T1.1.48T 8TPrevention of Malaria in Pregnancy8T.................................................................................... 5

8T1.1.58T 8TPrevention of Malaria in infancy8T ......................................................................................... 7

8T1.1.68T 8TTreatment of Malaria in childhood 8T ..................................................................................... 9

8T1.28T 8TAntimalarial drugs8T ................................................................................................................ 15

8T1.2.18T 8TArtemisinin and artemisinin derivatives8T ........................................................................... 15

8T1.2.28T 8TArylamino alcohols8T ............................................................................................................ 22

8T1.2.38T 8T4‐aminoquinilones 8T ............................................................................................................ 27

8T1.2.48T 8TAntifolate drugs8T ................................................................................................................ 35

8T1.2.58T 8TAntibiotics8T ......................................................................................................................... 38

8T1.38T 8TPharmacokinetics8T ................................................................................................................. 43

8T1.3.18T 8TIntroduction8T ...................................................................................................................... 43

8T1.3.28T 8TPopulation pharmacokinetics8T ........................................................................................... 49

8T1.3.38T 8TNONMEM8T .......................................................................................................................... 53

8T1.3.48T 8TPharmacokinetic considerations in specific populations8T .................................................. 59

8T1.48T 8TThesis outline8T ....................................................................................................................... 65

8TPREVENTIONOFMALARIAINPREGNANTWOMEN 8T.................................................69

8T28T 8TPharmacokinetic Properties of Azithromycin in Pregnancy 8T ............................ 71

8T2.18T 8TBackground8T ........................................................................................................................... 71

8T2.28T 8TPublication8T ............................................................................................................................ 73

8T2.2.18T 8TAbstract8T ............................................................................................................................. 73

8T2.2.28T 8TIntroduction8T ...................................................................................................................... 74

XXVIII

8T2.2.38T 8TPatients and methods8T ....................................................................................................... 75

8T2.2.48T 8TResults8T ............................................................................................................................... 79

8T2.2.58T 8TDiscussion8T .......................................................................................................................... 84

8T2.2.68T 8TAcknowledgements8T ........................................................................................................... 87

8TPREVENTIONOFMALARIAININFANTS 8T.......................................................................89

8T38T 8TPharmacokinetic Properties of Conventional and Double‐Dose Sulfadoxine‐

Pyrimethamine Given as Intermittent Preventive Treatment in Infancy8T ................... 91

8T3.18T 8TBackground8T ........................................................................................................................... 91

8T3.28T 8TPublication8T ............................................................................................................................ 93

8T3.2.18T 8TAbstract8T ............................................................................................................................. 93

8T3.2.28T 8TIntroduction8T ....................................................................................................................... 94

8T3.2.38T 8TPatients and methods8T ....................................................................................................... 95

8T3.2.48T 8TResults8T ............................................................................................................................. 100

8T3.2.58T 8TDiscussion8T ........................................................................................................................ 108

8T3.2.68T 8TAcknowledgements8T ......................................................................................................... 112

8TTREATMENTOFUNCOMPLICATEDMALARIAINCHILDREN8T.............................115

8T48T 8TPopulation Pharmacokinetics of Artemether, Lumefantrine, and Their

Respective Metabolites in Papua New Guinean Children with Uncomplicated Malaria8T117

8T4.18T 8TBackground8T ......................................................................................................................... 117

8T4.28T 8TPublication8T .......................................................................................................................... 119

8T4.2.18T 8TAbstract8T ........................................................................................................................... 119

8T4.2.28T 8TIntroduction8T ..................................................................................................................... 120

8T4.2.38T 8TPatients and methods8T ..................................................................................................... 121

8T4.2.48T 8TResults8T ............................................................................................................................. 127

8T4.2.58T 8TDiscussion8T ........................................................................................................................ 135

8T4.2.68T 8TAcknowledgements8T ......................................................................................................... 140

8T58T 8TA Pharmacokinetic Comparison of Two Piperaquine‐Containing Artemisinin

Combination Therapies in Papua New Guinean Children with Uncomplicated Malaria8T143

8T5.18T 8TBackground8T ......................................................................................................................... 143


8T5.2.18T 8TAbstract8T ........................................................................................................................... 145

8T5.2.28T 8TIntroduction8T ..................................................................................................................... 146


8T5.2.48T 8TResults8T ............................................................................................................................. 154

XXIX

8T5.2.58T 8TDiscussion8T ....................................................................................................................... 162

8T5.2.68T 8TAcknowledgements8T ........................................................................................................ 167

8T68T 8TArtemisinin‐Naphthoquine Combination Therapy for Uncomplicated

Paediatric Malaria: A Pharmacokinetic Study 8T ......................................................... 169

8T6.18T 8TBackground8T ......................................................................................................................... 169


8T6.2.18T 8TAbstract8T ........................................................................................................................... 171

8T6.2.28T 8TIntroduction8T .................................................................................................................... 173


8T6.2.48T 8TResults8T ............................................................................................................................. 183

8T6.2.58T 8TDiscussion8T ....................................................................................................................... 194

8T6.2.68T 8TAcknowledgements8T ........................................................................................................ 199

8T6.2.78T 8TConflict of interest statement8T ......................................................................................... 199

8T78T 8TGeneral Discussion8T ....................................................................................... 201

8T7.18T 8TSignificance of findings 8T ....................................................................................................... 202

8T7.1.18T 8TPrevention of malaria in pregnancy8T ................................................................................ 202

8T7.1.28T 8TPrevention of malaria in infancy8T ..................................................................................... 203

8T7.1.38T 8TTreatment of uncomplicated malaria in children8T ........................................................... 204

8T7.28T 8TImprovements and future directions8T ................................................................................. 207

8T7.2.18T 8TPrevention of malaria in pregnancy8T ................................................................................ 208

8T7.2.18T 8TPrevention of malaria in infancy8T ..................................................................................... 209

8T7.2.28T 8TTreatment of uncomplicated malaria in children8T ........................................................... 210

8Tx8T 8TReferences8T ....................................................................................................................................... 213

8Txi8T 8TAppendix8T .......................................................................................................................................... 239

XXX

XXXI

vi 144B144BAbbreviations

µg ............... microgram(s)

µl ................ microliter(s)

ACT ............. Artemisinin Combination Therapy

AL ............... artemether/lumefantrine

AQ .............. amodiaquine

ARM ............ artemether

ARTS ........... artesunate

ART ............. artemisinin

AUC ............ area under the curve

AZI .............. azithromycin

BLQ ............. below the limit of quantification

BOV ............ between occasion variability

BSA ............. body surface area

BSV ............. between subject varibility

CI ................ confidence interval(s)

CL ................ clearance

CLRHR .............. hepatic clearance

CLRRR .............. renal clearance

CLRTR .............. total clearance

CQ ............... chloroquine

CWRES ........ conditional wieghted residuals

CysC ........... Cystatin C

DBL ............. desbutyl‐lumefantrine

DHA ............ dihydroartemisinin

DHFR .......... dihydrofolate reductase

DHPS ........... dihydropterate synthase

dl ................ decilitre(s)

ECR50R ............ half maximal effective concentration

FP ................ first‐pass

g .................. gram(s)

GFR ............. glomerular filtration rate

GOF ............ goodness‐of‐fit

h ................. hour(s)

Hb ............... haemoglobin

HPLC ........... high‐performance liquid chromatography

ICR50R ............. half maximal inhibitory concentration

XXXII

IIV ................ inter‐individual variability

IOV .............. inter‐occasion variavility

IPT ............... Intermittent Preventive Treatment

IPTi .............. Intermittent Preventive Treatment in infancy

IPTp ............. Intermittent Preventive Treatment in pregnancy

IQR .............. inter‐quartile range

kRaR................. aborption rate constant

kg ................ kilogram(s)

kRtrR ................ transit compartment rate

l ................... litre(s)

L .................. likelihood

LC‐MS .......... liquid chromatography mass spectrometry

LC‐MS/MS ... liquid chromatography‐tandem mass spectrometry

LRT .............. likelihood ratio test

LUM ............ lumefantrine

M ................. moles per litre

mg ............... milligram(s)

ml ................ millilitre(s)

MQ .............. mefloquine

MTT ............. mean transit time

ng ................ nanogram(s)

NN ............... number of transit compartments

NPC ............. numerical predictive check

NPD ............. naive pooled data

NQ ............... naphthoquine

NSX ............. NR4R‐acetylsulfadoxine

OFV ............. objective function value

PETIA ........... particle enhanced immunoturbidimetry

PCR .............. polymerase chain reaction

pcVPC .......... prediction corrected visual predictive check

PI ................. prediction interval(s)

PK ................ pharmacokinetic(s)

PNA ............. postnatal age

PNG ............. Papua New Guinea

PQ ............... piperaquine

PYR .............. pyrimethamine

QC ............... quality control

XXXIII

QN .............. quinine

R ................. correlation coefficient

RBC ............. red blood cell

RSD ............. relative standard deviation

RSE ............. relative standard error

RUV ............ residual unexplained variability

SD ............... standard deviation

SDX ............. sulfadoxine

SP ................ sulfadoxine/pyrimethamine

STS .............. standard two stage

tR½R ............... half‐life

UPLC ........... ultra high‐performance liquid chromatography

UV ............... ultraviolet

V ................. volume of distribution

VPC ............. visual predictive check

vs. ............... versus

VRssR ............... volume of distribution at steady state

WHO ........... World Health Organisation

WRES .......... weighted residuals

WT .............. body weight

XXXIV

XXXV

vii 145B145BAntimalarialdrugsandcombinationsusedinthisthesis

19TTrade name 19TGeneric name(s) of component(s) 19TAmount per tablet

19TZithromax® 19Tazithromycin 19T500 mg

19TFansidar® sulfadoxine/pyrimethamine 19T500/25 mg

19TCoartem® artemether/lumefantrine 19T20/120 mg

19TDuo‐cotecxin® 19Tdihydroartemisinin/piperaquine phosphate 19T40/320 mg

19TArtequick® 19Tartemisinin/piperaquine base 19T24/144 mgP

a

19TArco® 19Tartemisinin/naphthoquine 19T125/50 mg

P

aPSachets were used.

XXXVI

XXXVII

viii 146B146BListofTables

Table 1‐1 List of some commonly used artemisinin combination therapies .................................... 11

Table 1‐2 Change in relative bioavailability of artemisinin with consecutive dosing calculated from reported AUC values. All comparisons are with day 1 AUC. Artemisinin was given alone unless otherwise specified. ............................................................................................................... 18

Table 1‐3 Fractional difference in tR½R and AUC of artemether and dihydroartemisinin when drugs are coadministed with artemether/lumefantrine .................................................................. 21

Table 1‐4 Manufacturer’s recommended dosing of artemether/lumefantrine in children. Each tablet consists of 20mg of artemether and 120mg of lumefantrine. ............................................... 24

Table 1‐5 Fractional difference in tR½R and AUC of lumefantrine when drugs are coadministed with artemether/lumefantrine ......................................................................................................... 26

Table 1‐6 PCR‐adjusted efficacy of dihydroartemisinin/piperaquine at day 42 or afterwards in various studies. ................................................................................................................................. 29

Table 1‐7 Summary of findings of the pharmacokinetics of naphthoquine in healthy volunteers in Qu et al. P

220P ..................................................................................................................................... 33

Table 1‐8 Changes of enzymes involved in metabolism during pregnancy, adapted from Anderson 2005 P

307P. ............................................................................................................................. 60

Table 2‐1 Baseline characteristics of the study participants by pregnancy status and treatment allocation. Data are mean ± SD, median [IQR] or number (%). ........................................................ 79

Table 2‐2 Side‐effects reported during the first week after initiation of treatment. Data are numbers of patients and (%)............................................................................................................. 80

Table 2‐3 Model building, final parameter estimates and bootstrap results from the AZI population pharmacokinetic modelling. ........................................................................................... 81

Table 2‐4 Secondary pharmacokinetic parameters derived from post hoc Bayesian estimates for pregnant and non‐pregnant study participants (median [IQR]). ................................................ 83

Table 3‐1 Dosing guide for conventional and double‐dose groups with the SDX/PYR doses in mg given in parentheses. ........................................................................................................................ 95

Table 3‐2 Baseline characteristics of study participants. Data are number (%), mean±SD or median [IQR]. .................................................................................................................................. 100

Table 3‐3 Final population PK parameters and bootstrap results for PYR. .................................... 101

Table 3‐4 Post hoc Bayesian predicted PK parameters for PYR for PNG infants given conventional and double doses of SDX/PYR (median [IQR]). ......................................................... 102

Table 3‐5 Final population PK parameters and bootstrap results for SDX and NSX. Parameters for NSX modelling obtained after fixing model parameters for SDX are highlighted in grey. ........ 104

Table 3‐6 Post hoc Bayesian predicted PK parameters for SDX and NSX in PNG infants given conventional and double dosing of SDX/PYR (median [IQR]). ........................................................ 106

XXXVIII

Table 4‐1 Baseline characteristics of study participants. Data are number (%), mean ± SD or median and [inter‐quartile range]. ................................................................................................. 127

Table 4‐2 Final population pharmacokinetic estimates and bootstrap results for lumefantrine and desbutyl‐lumefantrine. ............................................................................................................ 129

Table 4‐3 Final population pharmacokinetic estimates and bootstrap results for ARM and DHA. 131

Table 4‐4 Secondary pharmacokinetic parameters derived from post hoc Bayesian estimates for study participants. Data are median [inter‐quartile range]. .................................................... 133

Table 4‐5 Summary of studies reporting area under the plasma concentration‐time curve (AUC) for lumefantrine. ............................................................................................................................ 136

Table 5‐1 Baseline characteristics of study participants. Data are number (%), mean ± SD or median [IQR]. .................................................................................................................................. 154

Table 5‐2 Final population pharmacokinetic estimates and bootstrap results for piperaquine. ... 156

Table 5‐3 Secondary pharmacokinetic parameters of piperaquine derived from post hoc Bayesian estimates for study participants, and day 7 plasma piperaquine concentrations. Data are median [inter‐quartile range]. .................................................................................................. 158

Table 5‐4 Final population pharmacokinetic estimates and bootstrap results for artemisinin (n=12). ............................................................................................................................................ 159

Table 5‐5 Secondary pharmacokinetic parameters for artemisinin derived from post hoc Bayesian estimates for study participants. Data are median [inter‐quartile range]. ..................... 160

Table 6‐1 Demographic data for children given artemisinin‐naphthoquine for the treatment of uncomplicated falciparum malaria. Data are mean ± SD unless otherwise indicated. .................. 183

Table 6‐2 Population pharmacokinetic parameters and bootstrap results for NQ in children with uncomplicated falciparum malaria. ........................................................................................ 185

Table 6‐3 Post hoc Bayesian parameter estimates and derived secondary pharmacokinetic parameters for NQ in children with uncomplicated falciparum malaria. Data are median [IQR]. 188

Table 6‐4 Population pharmacokinetic parameters and bootstrap results for ART in children with uncomplicated falciparum malaria. ........................................................................................ 191

Table 6‐5 Post hoc Bayesian parameter estimates and derived secondary pharmacokinetic parameters for artemisinin in children with uncomplicated falciparum malaria. Data are median [IQR]. All between‐group comparisons were statistically non‐significant. ....................... 193

XXXIX

ix 147B147BListofFigures

8TUFigure 1‐1 life cycle of Plasmodium in humans and mosquitoes. From http://www.malariasite.com/malaria/LifeCycle.htm.U8T ........................................................................ 2

8TUFigure 1‐2 Map of South Pacific region showing location of PNG. From http://www.wpro.who.int/internet/files/eha/toolkit/web2009/Country%20Profiles/Maps/fiji%20melanesia%20country%20map.jpg. U8T ................................................................................................ 3

8TUFigure 1‐3 Map of Papua New Guinea showing Madang. From: http://geology.com/world/papua‐new‐guinea‐map.gif U8T ..................................................................... 4

8TUFigure 1‐4 Population indices of immunity in an endemic area of P. falciparum transmission (from Langhorne et al 2008 UPU

26UPU). Change over time of various indices of malaria in a population

living in an endemic area of P. falciparum transmission: asymptomatic infection (pink), mild disease (febrile episodes caused by malaria; blue) and severe or life‐threatening disease (green). The data are normalized and are presented as the per cent of maximum cases for each population index. U8T .............................................................................................................................. 10

8TUFigure 1‐5 Artemisinin and its derivatives showing endoperoxide bridge in blue. The different functional groups at the 2‐keto position in red namely oxo for artemisinin, hydroxyl for dihydroartemesisinin, methoxy for artemether, hemisuccinate for artesunate and ethoxy for artemotil. U8T .......................................................................................................................................... 15

8TUFigure 1‐6 Regression line for artemisinin saliva and unbound venous plasma concentrations (ng/ml) in 18 male Vietnamese patients 1‐8 h after the first oral dose of 100 mg or 500 mg artemisinin. Figure 3A in Gordi et al. 2000 UPU

120UP.8T .................................................................................. 17

8TUFigure 1‐7 Measured artemether (■) and dihydroartemisinin (○) concentrations and the model‐fitted curves in a patient who received 80 mg artemether orally at 0, 8, 24 and 48 h demonstrating the time dependant changes seen in artemether and dihydroartemisinin disposition. From van Agtmael et al. UPU

139UP8T ............................................................................................ 20

8TUFigure 1‐8 Arylamino alcohols showing the similarity in structure of halofantrine and lumefantrine in blue. U8T ........................................................................................................................ 22

8TUFigure 1‐9 Some 4‐aminoquinilones antimalarials showing the common 4‐aminoquinilone group in chloroquine, amodiaquine, naphthoquine and piperaquine (a dimer).U8T ............................ 27

8TUFigure 1‐10 Antifolate antimalarial drugs.U8T ........................................................................................ 35

8TUFigure 1‐11 Antibiotics with activity against Plasmodium species.U8T .................................................. 38

8TUFigure 1‐12 One compartment model with a single output rate.U8T..................................................... 45

8TUFigure 1‐13 Two compartment open model with oral dosing and elimination from the central compartment. U8T ................................................................................................................................... 46

8TUFigure 1‐14 Two compartment open model with parenteral dosing and elimination from the central compartment parameterized in terms of clearance and volume parameters.U8T .................... 47

8TUFigure 1‐15 Example of a fitted concentration versus time profile. Observed concentrations over time (red crosses) have been fitting using a curve that is the combination of positive and negative exponentials (black line) that represent absorption and elimination processes respectively.U8T ...................................................................................................................................... 48

XL

8TUFigure 1‐16 A typical Ω matrix with variance terms, the diagonals, in blue (ω1,1 is the variance for η1), covariance terms, the off‐diagonals, in red and black (ω2,1 is the covariance between η1and η2, ω2,1 is the same as ω1,2). U8T ............................................................................................. 55

8TUFigure 1‐17 An example of the changes expected in volume, clearance and half‐life over weight and age using average weight for age data UPU

312UPU. The solid black line represents changes when

only considering allometry while the dashed red line considers both the effects of size and age.U8T 63

8TUFigure 2‐1 Structural model used in the final pharmacokinetic analysis of plasma azithromycin concentrations in the central compartment versus time.U8T ............................................................... 82

8TUFigure 2‐2(A) Population (○) and individual (●) predicted versus observed plasma azithromycin concentrations (µg/l on log10 scale) for the final model. The line of identity is also shown. (B) Weighted residuals vs. time (log scale) for azithromycin final model. U8T ............................................ 82

8TUFigure 2‐3 Visual predicted check plots showing simulated 10 UPU

thUPU (short dashed line), 50 UPU

thUPU (dotted

line) and 90 UPU

thUPU (solid line) percentile concentrations and observed concentration (log scale) data

(grey open circles) versus time (log scale) for non‐pregnant (A) and pregnant (B) participants. U8T .... 84

8TUFigure 3‐1 (A) Population (○) and individual (●) predicted versus observed plasma pyrimethamine concentrations (µg/l on log10 scale) for the final model. The line of identity is also shown. (B) Conditional weighted residuals vs. time for pyrimethamine final model. U8T ........... 103

8TUFigure 3‐2 Visual predicted check plots for PYR showing simulated 10 UPU


thUPU

(dotted line) and 90 UPU

thUPU (solid line) percentile concentrations and observed concentration (log

scale) data (grey open circles) versus time (log scale) for conventional dose (A) and double‐dose (B) participants. U8T ...................................................................................................................... 103

8TUFigure 3‐3 (A) Population (○) and individual (●) predicted versus observed plasma sulfadoxine concentrations (µg/l on log10 scale) for the final model. The line of identity is also shown. (B) Conditional weighted residuals vs. time (log scale) for sulfadoxine final model. U8T .......................... 105

8TUFigure 3‐4 Visual predicted check plots for SDX showing simulated 10 UPU


thUPU

(dotted line) and 90 UPU

thUPU (solid line) percentile concentrations and observed concentration (log

scale) data (grey open circles) versus time (log scale) for conventional dose (A) and double‐dose (B) participants. U8T ...................................................................................................................... 105

8TUFigure 3‐5 Maturation as a fraction of adult clearance for PYR (solid line) and SDX (dashed line) predicted from the PK model plotted against PMA. A box plot of the PMA in the recruited subjects is included to show its distribution in relation to maturation of clearance.U8T .................... 107

8TUFigure 4‐1 Time‐concentration plots showing LUM (○) and DBL () in μg/l on log10 scale. Curves of the median concentration for LUM (solid black line) and DBL (dashed black line) are also shown. U8T ..................................................................................................................................... 128

8TUFigure 4‐2 Population (○) and individual predicted (●) versus observed data for LUM (A) and DBL (B) concentrations (µg/l) for the final model. The lines of identity are also shown. U8T ............. 130

8TUFigure 4‐3 Visual predictive check showing observed 50th (●), 10th () and 90th (○) percen les with the simulated 95% CI for the 50th (solid black line), 10th (grey dotted lines) and 90th (dashed grey lines) percentiles for LUM (A) and DBL (B) concentrations (μg/l on log10 scale) from the final model. U8T ...................................................................................................................... 130

XLI

8TUFigure 4‐4 Population (○) and individual predicted (●) versus observed data for ARM (A) and DHA (B) concentrations (µg/l) for the final model. The lines of identity are also shown. The grey dashed line represents the LOQ of ARM in (A) and DHA in (B). U8T ............................................. 132

8TUFigure 4‐5 Visual predictive check showing observed 50th (●), 10th () and 90th (○) percen les with the simulated 95% CI for the 50th (solid black line), 10th (grey dotted lines) and 90th (dashed grey lines) percentiles for ARM (A) and DHA (B) concentrations (μg/l on log10 scale) from the final model. The fraction of BLQ observations from the data (○ connected with a dotted black line) with the simulated 95% prediction interval are also shown for both ARM and DHA. U8T ................................................................................................................................................ 132

8TUFigure 4‐6 The doses of lumefantrine and artemether in mg/kg given to children 5‐35 kg under current (solid black line) and suggested (dashed grey line) dosing regimens. The horizontal dotted black line represents the dose in mg/kg recommended for a 50 kg adult. U8T ........................ 137

8TUFigure 5‐1 (A) Population predicted (○) and individual (●) predicted versus observed plasma piperaquine concentrations (µg/l on log URU10URU scale) for the final model. The line of identity is also shown. (B) Conditional weighted residuals vs. time (log scale) for piperaquine final model. U8T ....... 157

8TUFigure 5‐2 Visual predictive check showing observed 50 UPU

thUPU (●), 10 UPU

thUPU () and 90 UPU

thUPU (○) percen les

with the simulated 95% CI for the 50 UPU

thUPU (solid black line), 10 UPU

thUPU (grey dotted lines) and 90 UPU

thUPU

(dashed grey lines) percentiles for plasma piperaquine concentrations (µg/l on log URU10URU scale) vs. time (h) for Artequick (A) and Duo‐cotecxin (B) from the final model. The observed data are superimposed as grey crosses. The insert shows data for the first 96 h. U8T ....................................... 158

8TUFigure 5‐3 (A) Population (○) and individual (●) predicted versus observed plasma artemisinin concentrations (µg/l on log URU10URU scale) for the final model. The line of identity is also shown. (B) Conditional weighted residuals vs. time for artemisinin final model. U8T ............................................ 160

8TUFigure 5‐4 Visual predictive check showing observed 50 UPU

thUPU (●), 10 UPU

thUPU () and 90 UPU

thUPU (○) percen les

with the simulated 95% CI for the 50 UPU


thUPU (grey dotted lines) and 90 UPU

thUPU

(dashed grey lines) percentiles for plasma artemisinin concentrations (µg/l on log URU10URU scale) vs. time (h) from the final model. The observed data are superimposed as grey crosses.U8T ................. 161

8TUFigure 6‐1 HPLC‐UV (222 nm) chromatograms showing naphthoquine (N; tURRRU = 9.4 min) and the internal standard, tramadol (T; tURRRU = 6.8 min). Panel A is spiked plasma used in the calibration curve (20 µg/l naphthoquine); Panel B is a patient’s pre‐dose blank sample (with IS) showing no endogenous interference; Panel C is a typical sample (25 µg/l naphthoquine). U8T ....................... 177

8TUFigure 6‐2 LC‐MS chromatograms showing artemisinin (ART; tURRRU = 4.3 min) and the internal standard, artemether (IS; tURRRU = 7.9 min). Panel A is spiked plasma used in the calibration curve (200 µg/l artemisinin); Panel B is a patient’s pre‐dose blank sample (with IS) showing no endogenous interference; Panel C is a typical sample (136 µg/l artemisinin). U8T .............................. 178

8TUFigure 6‐3 Time‐concentration plots of NQ for Group 1 (Panel A), Group 2 (Panel B; milk) and Group 3 (Panel C; water and double‐dose) patients. Inset shows plasma concentration‐time data from 0‐100 h after the dose. U8T .................................................................................................. 184

8TUFigure 6‐4 (A) Population predicted (○) and individual predicted (●) versus observed NQ plasma concentration (µg/l; log scale) for the final model. The line of identity is shown. (B) Conditional weight residuals vs. time (log scale) for NQ final model. U8T ............................................................... 186

8TUFigure 6‐5 Prediction corrected VPC plots for NQ in children with uncomplicated falciparum malaria, showing the observed 50 UPU

thUPU (●), 10 UPU

thUPU and 90 UPU

thUPU (○) percen les with the simulated 95% CI

XLII

for the 50 UPU


thUPU and 90 UPU

thUPU (dashed grey lines) percentiles. Inset shows plasma

concentration‐time data from 0‐100 h after the dose. U8T .................................................................. 189

8TUFigure 6‐6 Time‐concentration plots of ART for Group 2 (Panel A; milk) and Group 3 (Panel B; water and double‐dose) patients. U8T .................................................................................................. 190

8TUFigure 6‐7 (A) Population predicted (○) and individual predicted (●) versus observed ART plasma concentration (µg/l; log scale) for the final model. The line of identity is shown. (B) Conditional weight residuals vs. time (log scale) for ART final model. U8T .......................................... 192

8TUFigure 6‐8 Prediction corrected VPC plots for ART in children with uncomplicated falciparum malaria, showing the observed 50 UPU

thUPU (●), 10 UPU

thUPU and 90 UPU

thUPU (○) percen les with the simulated 95% CI

for the 50 UPU


thUPU and 90 UPU

thUPU (dashed grey lines) percentiles.U8T ................................. 192

XLIII

x 148B148BPreface

The studies in this thesis arose from collaboration between the University of Western Australia

and the Papua New Guinea Institute of Medical Research. They were carried out at Alexishafen

Health Centre just north of Madang on the north coast of Papua New Guinea between

February 2007 and October 2010. Participants were from surrounding villages including Biges,

Haven, Kananam, Maiwara, Malmal, Pau, Rempi and Vidar. The candidate spent nine months

on four separate trips in 2007 and 2008 to co‐ordinate the studies and participate in the bulk

of the data collection. The resultant publications provide the chapters of this thesis that

contain original data. The contribution of the candidate and co‐authors to these articles is

listed in section i.

Drug assays were performed at School of Medicine and Pharmacology, University of Western

Australia, QEII Medical Centre (HPLC‐UV), School of Pharmacy, Curtin University, Bentley

campus (HPLC‐UV and LC‐MS) and Department of Clinical Pharmacology and Toxicology,

PathWest Laboratory Medicine, QEII Medical Centre (UPLC‐LC‐MS/MS). Biochemical analyses

for Chapter 3 were performed at Department of Biochemistry, PathWest Laboratory Medicine,

Fremantle Hospital.

Abbreviations are used throughout the thesis and appear in full when first used. A list is

provided in section vi.

All studies were approved by the Medical Research Advisory Committee of the PNG

Department of Health and the Institutional Review Board of the PNG Institute of Medical

Research. The Medical Research Advisory Committee carries the responsibility of providing

ethical approval for all studies of performed in Papua New Guinea including those involving

international parties. Human research carried out in PNG requires the approval of this body.

The exact nature of, and ethical issues surrounding, each study were presented to these

bodies as is required for studies in humans, particularly vulnerable populations such as young

children and pregnant women.

The following is included as a postscript to this preface:

During the middle of the data collection phase for the infant study six members of the field

team (Susan Griffin, Kay Kose, Servina Gomorai, Nolene Pitus, Christine Kalopo and Bernard

Maamu), a study infant and mother, and myself were involved in a serious car accident.

Although several members of the team were seriously injured there were no fatalities and the

XLIV

study mother and infant received only minor bruising and cuts. I suffered only cuts, bruises

and a dislocated thumb. I am most grateful to the staff of Madang Hospital who provided us

with medical care, and am in awe of the field team most of whom returned to work at the

Papua New Guinea Institute of Medical Research. I am indebted to Dr Laurens Manning and his

wife, Kate, for taking care of me in the days after the accident. It took me a month to regain

full physical and mental capacity, and it had no significant impact on the completion of this

thesis.

1

1 0B0BGeneralIntroduction

1.1 7B7BMalaria

1.1.1 23B23BGlobalsnapshot

Malaria continues to be a serious global health concern with approximately half the

world’s population still at risk.P

6P According to the World Health Organisation (WHO)

there were 107 countries where malaria was still endemic in 2010, eight of which have

interrupted transmission and are in the ‘prevention of reintroduction phase’.P

6P Despite

a reduction from 2005 when there were between 350‐500 million cases a year and

over 1 million deaths, P

7P there are still an estimated 216 million cases and 655 000

deaths attributable to malaria.P

6P In children living in Africa who make up a large

proportion of these deaths, malaria is often complicated with nutritional deficiency.P

8P

Globally there has been a 17% reduction in the incidence of malaria and a 25%

reduction in malaria specific mortality between 2000 and 2010.P

6P

1.1.2 24B24BParasitology

Malaria is an infection due to protozoan parasites from the genus Plasmodium that

target red blood cells (RBCs). There are five known species of Plasmodium that infect

humans. Most of the morbidity and mortality of malaria is attributable to P.

falciparum, which is also the most common globally. P. vivax, P. ovale and P. malariae

are also human malaria parasites. Recently, P. knowlesi, a monkey malaria parasite,

has been found to be an important cause of disease in humans in certain areas of

Asia.P

9P

Malaria is a vector‐borne disease transmitted by the female anopheline mosquito of

which there are more than 30 species. The life cycle of Plasmodium in humans and

mosquitoes is shown in Figure 1‐1. The disease in humans begins after a bite from an

mosquito. During the blood meal, less than 100 sporozoites residing in the mosquito’s

salivary gland PP

10, 11 are injected into the subcutaneous tissues (less frequently into the

blood stream) which then, after a short delay, travel to the liver.12 In the liver,

sporozoites pass through a Kupffer cell and several hepatocytes before beginning to

develop into merozoites inside a hepatocyte.13 In P. vivax, a number of these will enter

2

a dormant stage, and these hypnozoites can reactivate months or years later.14 P.

ovale may also have dormant hypnozoite stage although this hypothesis has recently

been brought into question.15 Each sporozoite will undergo asexual reproduction over

5‐15 days to develop into tens of thousands of merozoites forming a hepatic schizont.

The schizont then ruptures and each merozoite is capable of invading a RBC through a

sequence of receptor interactions, reorientations, vacuole formation and cell entry by

endocytosis.12 Once inside the RBC, the merozoite grows into a trophozoite which

eventually becomes a schizont comprising between 16‐32 merozoites. After 48‐72 h,

the schizont ruptures and each merozoite infects another RBC. Subsequently, another

round of the intra‐erythoctic cycle begins. Instead of undergoing asexual reproduction,

some merozoites in infected RBCs undergo sexual differentiation into gametocytes.

In P. falciparum infections, the surface of the RBC is altered such that asexual parasites

can bind to endothelium and to the placenta, while gametocytes can also adhere to

the endothelium.12 This sequestration of infected RBCs is responsible for cerebral

Figure 1‐1 life cycle of Plasmodium in humans and mosquitoes. From http://www.malariasite.com/malaria/LifeCycle.htm.

3

malaria, other manifestations of organ failure, and pregnancy complications including

maternal anaemia, low birth weight and premature delivery.12, 16 There is growing

evidence that P. vivax is also able to sequester,17 and this may be responsible for the

respiratory and pregnancy‐related complications associated with vivax malaria.17, 18

Once differentiated into micro‐ and macro‐gametocytes, these forms are taken up by

the mosquito during a blood meal. In the mosquito, a zygote is formed, which leads to

the formation of sporozoites. These migrate to the salivary gland within the mosquito

and are then able to infect a human host with the next blood meal of the mosquito.

1.1.3 25B25BMalariainPapuaNewGuinea

Papua New Guinea (PNG) is situated in the southwest of the Pacific Ocean and

occupies the eastern half of the island of New Guinea as well as numerous small

islands (Figure 1‐2). Four human malaria parasites are found in PNG; P. falciparum, P.

vivax, P. ovale and P. malariae. Of these P. falciparum represents 80% of all infections,

while P. vivax is the next most common.6 PNG accounted for 36% of all malaria cases in

the Western Pacific region in 2010.6

Temperature, and therefore altitude, is the main determinant of malaria prevalence in

PNG. The warmer coastal areas are holoendemic and the cooler highlands have low

rates of infections.19 Ninety‐four per cent of the population live in areas of high

Figure 1‐2 Map of South Pacific region showing location of PNG. From http://www.wpro.who.int/internet/files/eha/toolkit/web2009/Country%20Profiles/Maps/fiji%20melanesia%20country%20map.jpg.

4

transmission (more than 1 case per 1,000 population per year), while the remaining six

percent live in an area of low transmission (0‐1 cases per 1,000 population per year).6

Although many countries have seen a decrease in the number of cases from 2000 to

2010, the numbers in PNG have remained stable.6 Malaria is one of the most common

outpatient diagnoses and causes of hospital admission.19

The studies in this thesis were carried out at a small Catholic health centre in

Alexishafen, just north of Madang on the north coast of PNG (Figure 1‐3). There is a

very high rate of malaria transmission in this area, more than 100 cases per 1,000

population each year.6 The local distribution of Plasmodium species is representative

of the country as a whole. P. falciparum infections are the most common, followed by

P. vivax and occasional cases of P. ovale and P. malariae.20

Before DDT spraying programs (1957‐1970), P. vivax was the most common malaria

species in PNG. There was an increase in P. falciparum immediately after DDT spraying

was abandoned.19 For some time after this, the use of antimalarial drugs, particularly

chloroquine (CQ), was central to controlling malaria.19 The diagnosis of malaria is

largely a clinical diagnosis in PNG with an annual blood smear examination rate of <

5%.6 Subsequently, there was extensive, and potentially inappropriate, use of CQ,

Figure 1‐3 Map of Papua New Guinea showing Madang. From: http://geology.com/world/papua‐new‐guinea‐map.gif

5

which led to resistant strains first being reported in 1976.19 By the early 1990s, the

effectiveness of CQ and similar compounds such as amodiaquine (AQ) was greatly

reduced. It was not until 2000 that sulfadoxine/pyrimethamine (SP) was added to CQ

as first‐line for the treatment of uncomplicated malaria.19 The effectiveness of CQ/SP

also declined. In 2008, a large efficacy trial demonstrated that

artemether/lumefantrine (AL), an artemisinin combination therapy (ACT), provided

better treatment outcomes. This treatment is currently being implemented as first‐line

therapy.21 In addition to antimalarial drugs, distribution of insecticide treated bednets

(2000), indoor residual spraying (2010) and intermittent preventive treatment in

pregnancy (ITPp, 1981) have been implemented. Intermittent preventive treatment in

infancy (IPTi) is being evaluated,22 with the efficacy results from the trial pending.23

In areas of high transmission, such as is the case in the lowlands of PNG, it is pregnant

women, infants and children that bear the burden of disease.24, 25 For this reason

greatest attention is given to the control of malaria in these sub‐groups in PNG.

1.1.4 26B26BPreventionofMalariainPregnancy

Adults living in highly endemicity areas develop a degree of immunity to malaria such

that, although infections still occur, they are usually asymptomatic.25, 26 However, all

pregnant women, regardless of the intensity of malaria transmission in the area, are at

risk of a number of complications. These include maternal anaemia and the effect of

placental accumulation of parasites, including low birth weight from prematurity and

intrauterine growth restriction.24 There is also a growing body of evidence that

maternal infection itself increases infant mortality.27 Prior naturally acquired immunity

becomes compromised in pregnancy, particularly during the first and second

pregnancies.25 The effects of malaria during pregnancy are primarily facilitated by the

sequestration of parasites in the maternal placental vascular bed, and are therefore

more common with P. falciparum infections (see 1.1.2 above).24

Antimalarial drug treatment of infected pregnant women is acknowledged to protect

mothers from anaemia and their infants from low‐birth weight and death.28 Treatment

strategies for malaria in pregnancy include ensuring prompt treatment of symptomatic

infections. In endemic areas, the majority of infections are likely to be asymptomatic.

6

Nevertheless, they are still potentially harmful to both mother and infant.29 Another

management strategy is to provide continuous prophylaxis throughout pregnancy.

However, ensuring compliance is difficult.

For these reasons, a novel approach has been developed whereby treatment courses

of antimalarial drugs are administered at regular times during pregnancy regardless of

whether or not the mother is parasitaemic.30, 31 This is called IPTp. Doses are designed

to correspond with usual ante‐natal care, which is often has good compliance in

developing countries.32

Initial studies of IPTp used pyrimethamine (PYR) combined with dapsone. These

studies recorded benefits including a lower parasite density, increased maternal

haemoglobin (Hb) and higher birth weight.33 Currently, SP is drug of choice in many

IPTp programs. Its use is supported by a strong evidence base,28 and it is the only drug

currently recommended by the WHO for this purpose.34 In PNG, a combination of IPT

and chemoprophylaxis is employed. A single treatment dose of CQ with SP (IPT) is

given followed by weekly doses of CQ (chemoprophylaxis).35

Recent findings bring the use of SP for IPTp into question despite a strong evidence

base for the benefits of SP IPTp. In addition to the potential detrimental effects of

antifolate drugs in pregnancy, a major concern relates to the increase in parasite

resistance to SP. In an area of very high SP resistance (68% treatment failure by day 14

in children) there was no benefit of SP IPTp with the suggestion that it may have a

detrimental effect on infant anaemia.36 Another study reported no significant benefit

of SP IPTp in an area where it was previously beneficial, potentially due to a rise in

parasite resistance to SP.37 In contrast to these results, SP IPTp was still efficacious in

an area with high day 14 SP treatment failure in children (8‐39%),38 suggesting that

parasite resistance to SP affects treatment and prophylactic efficacies differentially.

The WHO recommends the use of SP IPTp to continue until the day 14 treatment

efficacy of SP in children falls below 50%.34 The use of SP outside IPTp has decreased,

thereby reducing drug pressure. There is some evidence that drug resistant strains are

decreasing in areas with reduced drug pressure.39, 40 However, no report has

7

demonstrated a positive effect of such reduced drug pressure on treatment or

prophylactic efficacy.

As the efficacy of SP, the only recommended form of IPTp, is under threat, there is a

great need for alternatives.29 However, few antimalarial drugs have proved safe in

pregnancy.41 Additionally, pregnant women are generally excluded from

pharmacokinetic (PK) studies of antimalarial drugs due to fears of adverse maternal

and foetal outcomes.42 Therefore information regarding appropriate dose

adjustments is often lacking. Potential candidates include piperaquine (PQ),

mefloquine (MQ) and azithromycin (AZI).29 The latter is the subject of the publication

in Chapter 2 and its pharmacology is discussed in more detail in section 1.2.5.1 (page

39).

1.1.5 27B27BPreventionofMalariaininfancy

For children born in an endemic area, factors during foetal development influence

their risk of developing malaria in first year of life. For example, maternal malaria

increases the rate of malaria morbidity in infancy,27, 43‐45 an effect that may be greatest

during the later stages of pregnancy.27 The malaria parasite is also capable of crossing

the placenta in utero (or at birth) to cause congenital malaria. 46 This occurs more

commonly in infants of non‐immune mothers, while infants of immune mothers with

congenital malaria are often asymptomatic and can clear the infection

spontaneously.47, 48

During early infancy, there is relative protection against falciparum malaria in endemic

areas, as evidenced by lower than expected rates of infection and low parasitaemia

when malarial infections do occur.49‐51 One explanation for this observation is the

passive transfer of immunity from mother to infant, mediated by maternal

immunoglobulin G passage through the placenta.52 The presence of foetal Hb in infants

could also have a role, with some evidence that it can slow the maturation of malaria

parasites in RBCs.53 Recently, this effect of foetal Hb has been questioned.54 One

report has suggested a combined effect of maternal immunoglobin G and foetal Hb to

impair the cytoadherance of malaria infected RBCs.54 This mechanism of protection

has also been proposed to explain the protective effects of sickle Hb and Hb C.55

8

Although young infants have relative protection against infections with P. falciparum,

no protection has been noted against P. vivax.56

Once this stage of relative protection ends at approximately 3‐6 months of life, the

rate and severity of P. falciparum infections increase.51 The processes of innate

immunity are still developing during this time.57 Peak malaria infection rate occurs at a

younger age in areas of higher transmission, where the acquisition of immunity is

faster.57 Hence, severe falciparum malaria is more common in younger children in

areas of high transmission, and is a significant contributor to the overall morbidity and

mortality in infants.58 In Madang and its surrounds, including Alexishafen where the

studies in this thesis were preformed, the rate of infection in those less than one year

old was approximately 15% for P. falciparum and 5% for P. vivax in the 1980s.20 A more

recent study in a neighbouring province has shown similar results, with a rate of

approximately 14% and 8% for P. falciparum and P. vivax infections, respectively.59 This

study also found the risk of severe malaria to be significantly higher in children below

the age of two compared with older children (odds ratio of 2.2, 95% confidence

interval, CI, 1.8‐2.7).59

Malaria also contributes to the complex interplay of factors, including nutritional

deficiencies, underweight status, and co‐infections, that result in a high infant

mortality rate in developing countries like PNG.8 The effect of malaria on infant

mortality is mediated through a variety of complications, including anaemia, organ

dysfunction and metabolic disturbances.58 In 2008, for children between 1 month and

5 years of life, malaria was the third leading cause of death worldwide after

pneumonia and diarrhoeal diseases, and the second leading cause of death in PNG

after pneumonia.60

Although a number of approaches have been used to deal with the high burden of

malarial disease in infants, IPTi is a relatively new strategy with promising results.61 In

ITPi, infants in endemic areas receive treatment doses of antimalarial drugs at routine

vaccination visits, regardless of clinical and/or parasitological features. SP has been

used as a first‐line agent in evaluating IPTi programs due to its availability, tolerability

9

and relatively low cost. It is also a single dose regimen which facilitates direct

observation of the full treatment.

A recent review of the safety and efficacy data from six trials of SP IPTi in Africa

reported 30% protective efficacy against clinical malaria, 23% protective efficacy

against all‐cause hospital admissions, 31% protective efficacy against malaria‐related

hospital admissions, and 21% protective efficacy against anaemia in the first year of

life.62 This review came to the conclusion that SP IPTi was a valuable tool in the control

of malaria in endemic areas of Africa, despite the emergence of molecular markers of

parasite resistance to SP.62 Results from a similar IPTi trial performed in PNG are

imminent. The results of this trial may differ from those performed in Africa due to the

different host and parasite genetics, species prevalence, and coexisting health

problems.

An additional benefit of IPTi over and above its effect on malaria related morbidity and

mortality may be an increase in immunisation rates. In Mali, immunisation rates

increased from 37% to 54% one year after an IPTi program began.63 Results of a pilot

study of SP IPTi in six African countries, where half a million SP doses were given, are in

preparation.23 Despite the extensive data of the benefits of IPTi, no country currently

has a IPTi policy in place.6

As will be discussed in detail below (section 1.2.4.1, page 36), despite the extensive

use of SP in infants, PK studies are limited. There is some suggestion that an increased

dose is required to ensure adequate exposure. This is the subject of the publication in

Chapter 3.

1.1.6 28B28BTreatmentofMalariainchildhood

Children living in malaria endemic areas continue to have symptomatic infections after

their first year of life. The development of acquired immunity is slow, occurring over

many years, and sterile immunity is yet to be demonstrated.26 Immunity to malaria

occurs in stages. Initially protection against severe malaria is developed followed by

immunity to symptomatic disease, and finally, yet incompletely, to asymptomatic

parasitaemia (Figure 1‐4).26 Acquisition of immunity to non‐cerebral severe infections

10

occurs quic

disease is t

malaria tra

difference

more infec

above, this

malaria is h

disease de

infection is

age of one

Although t

transmissio

approach t

the 1980s

Throughou

opposed to

compound

combinatio

Figure 1‐4 Pop200826). Chantransmissionor life‐threate

ckly in area

therefore ra

ansmission,

can be acco

ctions to ach

s pattern of

higher in yo

spite exper

s therefore

year in PNG

the use of IP

on,65 the tre

to the contr

led to poor

ut the 1990s

o monother

d. In 2001, t

on therapy,

pulation indicenge over time on: asymptomatening disease (

s of high tra

arely seen i

immunity t

ounted for

hieve immu

f acquired im

ounger child

riencing hig

the most co

G.

PT in childre

eatment of

rol of malar

er treatmen

s, combinat

rapy, partic

he WHO pr

with a part

es of immunity of various indicic infection (pingreen). The da

ansmission,

n older chil

to severe m

by a lower

unity in low

mmunity als

dren. Older

her rates of

ommon pre

en is being i

symptomat

ria in childre

nt outcome

tions of ant

ularly those

oduced a re

ticular emp

in an endemic ces of malaria innk), mild diseata are normalizfor each popu

, after only

dren living

malaria occu

rate of infe

transmissio

so exists in

children ar

f infection.5

esentation o

investigated

tic cases is t

en. The rise

es and a nee

imalarials b

e combinati

eport which

hasis on AC

area of P. falcin a populationse (febrile episzed and are preulation index.

one or two

in these are

rs at an old

ction, as we

on settings.

PNG, wher

e less likely

59 Uncompli

of malaria in

d in areas o

the preferre

of parasite

ed for new a

became mor

ions contain

h outlined t

CTs.67 ACTs,

iparum transm living in an ensodes caused byesented as the

infections.6

eas.64 In are

er age.64 Th

ell as the ne

64 As indica

e the risk of

to develop

cated symp

n children a

f seasonal m

ed pharmac

e resistance

antimalaria

re common

ning an arte

he benefit o

where an

ission (from Lademic area of Py malaria; blueper cent of ma

64 Severe

eas of low

his

eed for

ated

f severe

p severe

ptomatic

above the

malaria

cological

to CQ in

l agents.66

nly used as

emisinin

of

anghorne et al P. falciparum e) and severe aximum cases

11

artemisinin compound is partnered with another drug with a longer half‐life (tR½R) , are

either tablet combinations of individual pre‐existing drugs or co‐formulations that

include some partner drugs not commonly used (Table 1‐1).

In order to guide policy makers and health professionals regarding appropriate choices

of antimalarials, the WHO published guidelines for the treatment of malaria in 2006,68

with a follow up second edition in 2010.69 The first of these guidelines recommended

ACTs as the first line treatment of uncomplicated falciparum malaria.68 The four

recommended ACTs were AL, artesunate (ARTS)+AQ, ARTS+MQ and ARTS+SP. The first

of these, AL, is available as a co‐formulation, known as Riamet® or Coartem®. More

detail of the two components of AL is presented in sections 1.2.1.1 (page 16) and

1.2.2.1 (page 23). At the time of the study in Chapter 4, AL was still under investigation

for use PNG, where it is now the first line treatment.

The second edition of these guidelines added dihydroartemisinin (DHA)/PQ, another

co‐formulated ACT, to the list of recommended combinations.69 Around the same time

a Cochrane review of ACTs for the treatment of uncomplicated malaria found DHA/PQ

was at least as effective as the previously recommended ACTs.70 Both the WHO

guidelines and the Cochrane review noted the likely superiority of DHA/PQ to AL in the

Table 1‐1 List of some commonly used artemisinin combination therapies

Artemisinin component

Partner drug Co‐formulated WHO

prequalified WHO

recommended

artesunate amodiaquine Yes Yes

(Coarsucam) Yesa

artesunate mefloquine Yes No Yes

artesunate sulfadoxine‐

pyrimethamine No No Yesa

artesunate chloroquine No No No

artemether lumefantrine Yes Yes (Coartem) Yes

dihydroartemisinin piperaquine phosphate

Yes No Yes

artemisinin piperaquine base Yes No No

artemisinin naphthoquine Yes No No

artesunate pyronaridine Yes No No a Not for multidrug resistant areas

12

treatment of uncomplicated vivax malaria.69, 70 The superiority of DHA/PQ over a

number of other ACTs was also noted in a more recent Cochrane review, which

specifically examined the used of ACTs for vivax malaria.71 A recent efficacy trial in PNG

found a higher efficacy of DHA/PQ compared to AL and ARTS+SP for children with

uncomplicated vivax infection.21 A similar ACT, artemisinin (ART)/PQ base, is also

commercially available and is the subject of the publication in Chapter 5. The

components of these two PQ‐containing ACTs are discussed in more detail in sections

1.2.1.1 (page 16), 1.2.1.2 (page 19) and 1.2.3.1 (page 28).

There is evidence that the newly deployed partner drugs lumefantrine (LUM) and PQ

are vulnerable to the development of parasite resistance.72‐74 Currently there has not

been a need to change drug therapy in response to this. However, there is a need for

more ACTs that are effective and approved by the WHO. Two potential alternative

combinations are ARTS/pyronaridine and ART/naphthoquine (NQ). The former is

backed by the Medicines for Malaria Venture and has been shown to effective against

P. falciparum and P. vivax in Africa and Asia.75 The latter was developed as a single

dose therapy by the manufacturer, in contrast to the WHO recommendations for at

least 3 days of the artemisinin component in combination therapies.69 There are few

data relating to ART/NQ, particularly PK data. This combination is the subject of the

publication in Chapter 6 and NQ is discussed in more detail in section 1.2.3.2 (page 31).

In determining the efficacy of treatment for uncomplicated malaria, the WHO usually

recommends a 28 day post treatment follow‐up.69 For drugs with a longer tR½R such as

PQ this is extended to 42 days.69 Recurrent parasitaemia after treatment can either be

due to re‐infection or recrudescence. For the latter there is persistent erythrocytic

infection after inadequate treatment that becomes apparent either clinically or

parasitologically. The time between treatment and recrudescence is dependent on a

number of factors. Later recrudescence is seen for drugs with longer half‐lives,

including PQ. For P. vivax, the activation of dormant hypnozoites, known as a relapse,

is also a possible mechanism of recurrent parasitaemia. A relapse is unlikely to occur

less than 16 days from the initial infection.14 For falciparum infections, it is possible to

differentiate a re‐infection from a recrudescence using polymerase chain reaction

13

(PCR) technology. In vivax malaria, however, a recrudescence cannot be differentiated

reliably from a relapse, as both originate from the same initial infection.

In providing guidance regarding the choice of first‐line treatment for malaria, the WHO

uses PCR‐adjusted cure rates. These rates only represent those subjects who remained

free from recrudescence and relapse, while not considering re‐infections. The WHO

recommends changing a first‐line treatment with a PCR‐adjusted cure rate of < 90% to

one with a PCR‐adjusted cure rate of > 95%.69 Although a high PCR‐adjusted efficacy is

beneficial in reducing the rise of resistance, a better marker for the impact of

treatment on the individual patient may be the rate of recurrent parasitaemia (non‐

PCR‐adjusted cure rate). Regardless of the aetiology of the infection, the negative

health implications for the patient are similar. The non‐PCR‐adjusted cure rate is

particularly important in areas of high transmission where new infections during

follow‐up are common.21 In these areas, the period of post‐treatment prophylaxis

afforded by the partner drugs in ACTs becomes important.76 This period of prophylaxis

increases with the terminal elimination tR½R of the drug and is shortened with increased

parasite resistance against the drug.76 A recent efficacy trial of four different

antimalarial treatments in PNG found that, although there was a significant difference

in the PCR‐adjusted cure rates, there was no difference in the non‐PCR‐adjusted cure

rate.21 The post treatment prophylaxis efficacy should be considered when two

treatments have similar PCR‐adjusted cure rates.

14

15

1.2 8B8BAntimalarialdrugs

1.2.1 29B29BArtemisininandartemisininderivatives

ART (qinghaosu) is an extract of the leaves of the herb Artemisia annua (sweet

wormwood) and has been used in the treatment of fever in China for many

centuries.77 Although traditional preparations of this compound do produce clinical

effects, such preparations are ineffective when considering the doses required for

adequate treatment outcome.77 A number of derivatives of ART have been produced

by altering the function group at the 2‐keto position (Figure 1‐5). These are DHA,

artemether (ARM), artemotil and ARTS. The first report of ART appeared In Western

literature in 197978 and, since then the use of artemisinin compounds has flourished.

Figure 1‐5 Artemisinin and its derivatives showing endoperoxide bridge in blue. The different functional groups at the 2‐keto position in red namely oxo for artemisinin, hydroxyl for dihydroartemesisinin, methoxy for artemether, hemisuccinate for artesunate and ethoxy for artemotil.

Despite a number of theories and the endoperoxide bridge being essential, the exact

mechanism of antimalarial action of artemisinin compounds remains unknown.79

These compounds are extremely potent as they are able to reduce the parasite load by

10,000 per 2‐day erythrocytic cycle.80 They act on late trophozoite stages as well as

early trophozoites and gametocytes.81 The gametocidal effect has important

16

implications for preventing the transmission of malaria. Despite their short tR½R, there is

growing evidence for the emergence of resistance to these compounds.82

All compounds within this class are generally safe and very well tolerated.83, 84 The only

potentially serious side effects reported have been type I hypersensitivity reactions,

estimated to occur in 1 in every 3,000 patients.85 High doses of ARTS monotherapy for

7 d have been associated with transient neutropaenia.86 From results of in vitro and

animal studies, neurotoxicity has been associated with exposure to artemisinin

compounds.83, 87‐93 The relevance of these finding in humans is uncertain. A number of

case reports,94‐98 case‐control studies of hearing loss,99, 100 and reports of delayed

coma recovery,101, 102 have suggested potential neurotoxicity in human. These reports,

however, essentially represent observational data with no proof of causality.

Prospective studies have been unable to corroborate these findings.103‐107 The extent

of ART that crosses the intact blood brain barrier into CSF is small, only 2% of plasma

concentrations.108 Animal studies have involved allometrically higher doses of slow

release formulations given for longer periods than recommended for the treatment of

human malaria.89, 90, 109 This difference in dosing is thought to be primary reason that

no definitive evidence of neurotoxicity has been found in humans treated with

conventional doses of artemisinin compounds.109

1.2.1.1 75B75BArtemisinin

Despite being the basis for all artemisinin compounds, ART is less commonly used due

to its lower relative potency.110 Both oral and rectal formulations of ART are available.

Initially it was used as a monotherapy where oral dosing for 7 days was required to

avoid recrudescence. At the recommendation of the WHO, the practice of ART

monotherapy has largely ceased.111 Currently combinations with PQ base (Artequick®)

and NQ (ARCO®) are commercially available. The efficacies of these combinations are

discussed in sections 1.2.3.1 (page 28) and 1.2.3.2 (page 31).

The PK of oral ART have been extensively evaluated in healthy adults112‐117 and adults

with malaria,118‐124 while only one study has included children with malaria.125 These

studies used either non‐compartmental analysis112, 113, 117‐121, 123, 124 or compartmental

analysis,114‐116, 122, 125 and reported an elimination tR1/2R between 1.4 hours (h) and 4.8 h

17

in plasma.

minutes in

the first 10

the last re

115 In a nu

eliminatio

and the el

of these st

Although o

concentra

variability

Additiona

to be high

studies wi

sampling (

Figure 1‐6 ReVietnamese p2000120.

. Absorption

n plasma.114

0 h post do

ecordable d

mber of the

on tR1/2R. Ther

limination m

tudies emp

one study r

ations,120 th

between th

lly the auth

her than exp

ith saliva sa

(1.4 h and 4

egression line fopatients 1‐8 h a

n of ART is r

4, 125 The sam

se. In two s

rug concent

ese studies,

refore, aspe

may not hav

loyed saliva

reported a c

e correlatio

he two was

ors noted a

pected.120 T

mpling (0.9

4.8 h).113‐115

or artemisinin safter the first o

rapid with a

mpling dura

studies with

trations wa

, the sampli

ects of the d

ve been acc

a sampling i

correlation

on co‐efficie

s explained

a tendency f

This may exp

9‐2.2 h),116, 1

5, 118, 119, 123, 1

saliva and unboral dose of 100

an absorptio

ation in mo

h longer sam

s much ear

ing period w

distribution

curately est

in the place

between sa

ent was 0.77

by a linear

for concent

plain the tre

121, 124 when

125

ound venous p0 mg or 500 mg

on tR1/2R of ap

st studies h

mpling durat

lier at 8 and

was only tw

phase may

imated. Add

e of venous

aliva and un

7, indicating

relationship

trations in e

end for a low

n compared

plasma concentg artemisinin. F

pproximate

has been res

tions, 24 h

d 12 h respe

wice the rep

y have been

ditionally, a

samples.121

nbound ven

g that only

p (Figure 1‐

earlier saliva

wer tR1/2R rep

d to those w

trations (ng/mlFigure 3A in Go

ely 20

stricted to

and 48 h,

ectively.114,

orted

missed,

a number

1, 122, 124

ous

59% of the

6).

a samples

ported in

with plasma

l) in 18 male rdi et al.

18

In the only study of oral ART PK that included children, 23 children aged 2‐12 years

were given five days of ART dosed according to body weight (WT, approximately 10

milligram/kilogram [mg/kg]) and 31 adults received 500 mg ART daily for 5 days.125

Children were found to have a higher relative clearance (CL) than adults in a model

that used age group as a categorical variable covariate.125 Although the sparse

sampling was probably adequate for developing the population PK model, the last

sample in adults was at 10 h whereas it was 8 h in children.125 An unbalanced study

design is not ideal for comparisons between two groups.

The auto‐induction of ART metabolism has been well characterized, with a primary

effect on the bioavailability of subsequent doses rather than on systemic CL.116 This

reduction in relative bioavailability after multiple dosing has been noted in numerous

studies.113, 117, 119, 121, 123‐125 The results of these studies are summarized in Table 1‐2,

using reported areas under the curve (AUCs) of different doses to determine the

relative bioavailability between doses. The auto‐induction effect is rapid and

associated with a reduction in the exposure of the second dose by 77%.124 A semi‐

physiological model was used to estimate a mean auto‐induction time of 1.9 h.116

Table 1‐2 Change in relative bioavailability of artemisinin with consecutive dosing calculated from reported AUC values. All comparisons are with day 1 AUC. Artemisinin was given alone unless otherwise specified.

Study Day of comparison change in relative bioavailability

Hassan Alin et al. 1996123 six ‐ 83%

Ashton et al. 1998119 five ‐ 75%

Ashton et al. 1998113 four ‐ 66%

seven ‐ 76%

Sidhu et al. 1998125 five ‐ 86%

Svensson et al. 1998117 sevena ‐ 82%

Gordi et al. 2002121 five ‐ 87%

five ‐ 72%

Svensson et al. 2002124 two + 25%b

three ‐ 55%

twoc ‐ 77%

threec ‐ 75% a omeprazole 20 mg was given on days 1 and 7, b non‐significant increase, c mefloquine was given on day 1

19

1.2.1.2 76B76BDihydroartemisinin

DHA is an active metabolite of a number artemisinin compounds, including ARM, ARTS

and artemotil. It also used as an antimalarial compound in its own right. It has an in

vitro activity similar to other artemisinin derivatives, and higher than that of ART.110, 126

Is it available in both tablet and suppository forms. Currently, it is commonly co‐

formulated with PQ tetraphosphate, a combination recently added to the list of WHO

recommended ACTs. The efficacy of this combination is discussed in section 1.2.3.1

(page 28).

When given alone orally, DHA has a rapid absorption with an absorption tR½R of 0.58 ‐

0.83 h.127‐129 It reaches peak plasma concentrations between 1.5‐2 h in adults with or

without malaria.127‐129 In the same samples, the elimination tR½R was found to be

between 0.83 and 1.97 h.127‐129 Following oral administration of ARTS, a similar tR½R is

found for DHA as a metabolite.128‐131 In contrast, the non‐compartmental elimination

tR½R of DHA is longer after the administration of oral ART (1.5‐5.1 h).132‐134 This suggests

formation rate‐limited elimination of DHA after oral administration of ART. No PK

interactions have been reported for DHA.

1.2.1.3 77B77BArtemether

ARM is an artemisinin derivative available for oral dosing as well as intramuscular

injection. Such parental formulations of ARM are important in the treatment of severe

malaria, when oral intake may not be tolerated. ARM is the artemisinin component in

the first approved co‐formulated ACT, AL (Coartem®, Riamet®). The efficacy of this

combination is reviewed in section 1.2.2.1 (page 23).

The PK of ARM and DHA after oral dosing of AL have been extensively investigated.

After oral dosing, peak ARM concentrations are reached after 0.75‐2 h, with DHA

concentrations peaking shortly afterwards.133‐139 This is in contrast to a markedly

slower and variable absorption phase after intramuscular injection.108, 140 Compared to

a fasted state, the AUC of ARM and DHA were increased 2.4 and 1.9 times respectively,

after a standard high‐fat breakfast.134 When simultaneous compartmental analysis of

ARM and DHA has been performed, models using one or two compartments for ARM

linked with one compartment for DHA have adequately described the plasma

20

concentrat

of ARM wit

population

in the CL o

1.5 and 3.6

A number

with LUM,

later time

non‐signifi

potentially

increase in

DHA, due t

DHA AUC (

given with

differences

Figure 1‐7 Mepatient who rseen in artem

tion‐time co

th each dos

n PK analysi

f ARM with

6 h.133, 135‐13

of potentia

with which

compared t

cant decrea

y due to an

n the AUC of

to co‐admin

(see above)

and withou

s in PK. Wh

easured artemeeceived 80 mg ether and dihy

ourse.141, 142

se, associate

s, this phen

h each succe

9, 143, 144

l PK interac

h it is comm

to when it w

ase (0.72 90

increase in

f DHA.134 An

nistration of

were not a

ut MQ136 an

en administ

ether (■) and dartemether or

ydroartemisinin

2 Multiple d

ed with an i

nomenon w

essive dose.

ctions of AR

monly co‐for

was adminis

0% CI 0.41 ‐

first‐pass (F

nother stud

f LUM.139 Th

altered with

nd quinine (

tered with l

ihydroartemisirally at 0, 8, 24 n disposition. F

dosing of AR

increasing A

was adequat

. 142 The elim

RM have als

rmulated, pe

stered alon

‐1.27) in the

FP) metabo

dy found litt

he time dep

h co‐adminis

(QN),137 the

lopinavir/rit

inin (○) concenand 48 h demoFrom van Agtm

RM resulted

AUC of DHA

ely describe

mination tR½

o been asse

eak concen

e.134 One st

e total bioav

lism as ther

tle change i

pendent cha

stered LUM

re were no

tonavir the

trations and thonstrating the tael et al.139

d in a decrea

A.133, 134, 136, 1

ed by a 57%

R of ARM is

essed. Whe

trations occ

tudy also no

vailability of

re was an as

n the AUC o

anges in AR

M.134, 139 Whe

significant

re was a no

he model‐fittedtime dependan

asing AUC

139 In a

% increase

between

n given

curred at a

oted a

f ART,

ssociated

of ART and

RT and

en AL was

on‐

d curves in a nt changes

21

significant decrease in ART AUC (34%), and a significant decrease in DHA AUC (45%).135

As the tR½Rs were similar between the two groups, this effect was likely mediated

through a change in the bioavailability ART rather than induction of metabolism. The

lack of significant results in the above studies may be a consequence of the high

variability in ART and DHA PK parameters with coefficients of variation between 32

and 103%, combined with modest difference in PK (Table 1‐3). Finally, when

ketoconazole, a potent CYP3A4 inhibitor, was co‐administered with a single dose of

ART/LUM, the AUC of ART and DHA increased by 2.39 and 1.66 fold, respectively, and

tR½R of ART and DHA increased by 1.42 and 1.69 fold, respectively. These results

emphasise the role of CYP3A4 in ART disposition.138

Table 1‐3 Fractional difference in tR½R and AUC of artemether and dihydroartemisinin when drugs are coadministed with artemether/lumefantrine

Study Coadministered drug artemether dihydroartemisinin

tR1/2 AUC tR1/2 AUC

Lefevre et al. 2000136 mefloquine ‐ first dose 1.21 1.37 0.95 1.19

last dose N/A 1.04 1.80 1.28

Lefevre et al. 2002138 ketoconazole 1.42 a 2.39 a 1.69a 1.66

Lefevre et al. 2002137 quinine 0.83 0.54b 0.94 0.63b

German et al. 2009135 lopinavir/ritonavir 0.94 0.65 N/A 0.55a a P<0.05. b significant difference (P<0.05) however quinine was administered after the last dose of artemether/lumefantrine when a difference already existed between the groups. The authors cannot therefore be certain that this difference was due to the administration of quinine.

22

1.2.2 30B30BArylaminoalcohols

QN is the oldest member of this group of antimalarial drugs. It was an active

component in powdered bark from the cinchona tree, the first medicine to be used

against malaria in the Western world. In an attempt to synthesize QN in 1856, the first

synthetic textile dye, mauve, was inadvertently produced ‐ later to become the origin

of the 4‐aminoquinilone drugs (page 27). The synthesis of QN was later achieved in

1944. Other arylamino alcohols include MQ, halofantrine and LUM (Figure 1‐8). These

compounds act on asexual blood stages of the parasite. While the mechanism of action

is not clear, there is evidence that they interrupt the handling of heme in the food

vacuole, through a different pathway to the CQ‐like drugs.145 They are not as potent as

the artemisinin compounds, only reducing the parasite load by approximately 100‐

1,000 times with each erythocytic cycle.

Figure 1‐8 Arylamino alcohols showing the similarity in structure of halofantrine and lumefantrine in blue.

The in vitro sensitivities of the drugs in this group are higher in CQ resistant strains

than CQ sensitive strains,146 and are also closely related to each other.147 Therefore,

resistance against LUM, a relatively new drug, may be expected in areas of high

resistance to MQ. Indeed, in these areas, treatment response to LUM has been related

to the degree of this resistance.148 The use of ACTs has seen the return of sensitivity to

MQ in Thailand,149 a positive sign for the future of these compounds.

23

There are a number of concerns with regard to the toxicity of drugs within this group.

QN has many known side‐effects, the more serious of which are ototoxicity, tinnitus,

hypoglycaemia and cardiotoxicity.83 MQ is known to produce neuropsychiatric

symptoms from dizziness and anxiety to convulsions, psychosis and encephalopathy.83

Halofantrine prolongs the electrocardiographic QT interval, increasing the risk of

tachyarrhythmias and death.150 Given the structural similarity between LUM and

halofantrine (Figure 1‐8), it may be expected that LUM would have similar effects on

the heart. In vitro data suggest LUM is safe in this respect, with a wide therapeutic

index. The half maximal effective concentration (ECR50R) for effect on heart muscle is

8,100 nM, many folds higher than half maximal inhibitory concentration (ICR50R) value

for malaria, at 40 nM.151, 152 Additionally, clinical studies have not found a significant

effect of therapeutic doses of LUM on QT interval in either children or adults.153‐158

1.2.2.1 78B78BLumefantrine

LUM, originally known as benflumetol, was developed in the 1970s by the Academy of

Military Sciences in China. It is only available in a combination with ARM, first

registered in China in 1992. After further development by Novartis Pharmaceuticals,

this combination was released to the world in 2001. In vitro, LUM demonstrates

synergy with ARM and DHA against Plasmodium falciparum.151, 159 Desbutyl‐

lumefantrine (DBL), an active metabolite of LUM, has been shown to have a higher

activity than LUM in vitro (between 2 and 7 times)160‐163 and has mild synergy with DHA

in vitro.163

The currently recommended treatment regimen of AL consists of 6 doses taken with

fat over 3 days (at 0, 8, 24, 36, 48, 60 h). Adults receive 4 tables per dose and children

are dosed according to WT (Table 1‐4). This represents an improvement over a

previous less efficacious regimen of 4 doses over two days. A more efficacious

approach of 6 doses over 5 days exists.164 However, it would likely result in poorer

compliance and, therefore, worse outcomes outside the research setting. A simpler

once daily treatment regimen, with doubled doses taken once daily for three days,

resulted in lower LUM AUC and worse efficacy, when compared to the recommended

24

treatment regimen.165 The study was underpowered (n=43) and the two groups were

not well matched in terms of baseline parasitaemia. The median parasitaemia was

1,981/μl in the standard regimen group, compared to 15,549/μl in the once daily

group.165 As a higher baseline parasitaemia can reduce bioavailability,166 the effect of a

simpler dosing schedule on AUC may have been overestimated. A similar once daily

regimen is the recommended regime for a paediatric suspension of the

combination.167 This formulation has promising initial efficacy findings in the treatment

of uncomplicated falciparum malaria in children.168, 169 However, it was not superior to

tablets given as six doses over three days.170 To conclusively accept or reject a

simplified dosing schedule, a larger non‐inferiority trial is required. The available

evidence suggests that, due to the low and variable bioavailability of LUM, multiple

doses of AL are required to allow LUM to reach concentrations that ensure treatment

efficacy.

A recent pooled analysis of seven studies supported by the manufacturer

demonstrated adequate day 28 PCR‐adjusted parasitological cure rates (>97%)

amongst both children and adults with falciparum malaria171. In children, this was

97.3% in the evaluable population and 93.4% in the modified intention to treat

analysis. A Cochrane review reported that, in comparison with other ACTs, AL is at

least as effective in the treatment of falciparum malaria.70 Neither of these reviews

reported the rates of re‐infection, a particular concern with LUM in areas of high

endemicity. LUM has a relatively short tR½R when compared other ACT partner drugs, at

68 ‐ 275 h135‐138, 164 versus 224 ‐ 667 h for PQ.172‐179 Therefore, a shorter period of post‐

treatment prophylaxis would be expected after AL. In PNG, this combination was

found to be superior to DHA/PQ phosphate, ARTS/SP and CQ/SP in treating the P.

falciparum infection in young children with a PCR‐corrected day 42 efficacy of 95.2%.21

Table 1‐4 Manufacturer’s recommended dosing of artemether/lumefantrine in children. Each tablet contains 20 mg of artemether and 120 mg of lumefantrine.

Weight Artemether/lumefantrine dose

5 ‐ <15 kg 1 tablet

15 ‐ <25 kg 2 tablets

25 ‐ <35 kg 3 tablets

>35kg 4 tablets

25

When the rate of recurrent parasitaemia was considered, there was no significant

difference between the various treatment arms in the study.21

The high efficacy of this combination does not extend to the treatment of vivax

malaria. A recent Cochrane review reported AL is probably inferior to DHA/PQ in

preventing recurrent parasitaemia before day 28 after the treatment of vivax

malaria.71 Results from a PNG trial also suggest AL should not be the first choice in the

treatment of uncomplicated vivax malaria.21 In this trial, there was no significant

difference between the 42 day PCR‐adjusted efficacy of AL and CQ+SP, 30.3% and

13.0% respectively (P=0.06).21 These results may be attributed to the relatively short tR½R

of LUM.

The PK of LUM has been evaluated in a number of populations. In all cases, a highly

variable bioavailability has been noted. The bioavailability was found to increase up to

16‐fold when given with food compared to the fasted state.180 Only 1.2g of fat, given

as 36 millilitres (ml) soya milk, was required to reach 90% of maximal absorption in a

study of healthy volunteers.181 In patients with malaria, there is evidence that

absorption improves as they recover from the acute infection.141, 164 In fact,

bioavailability is inversely proportional to both initial parasite load and temperature at

presentation,164 further suggesting there is an independent disease effect on

bioavailability. After a lag‐time of approximately 2 h, absorption occurs relatively

slowly. Maximum concentrations are reached between 4‐6 h after oral dosing,134, 135,

138 and absorption tR½R is between 4‐5 h.141, 164 In compartmental analyses, the PK of

LUM is best described through a two compartment model.141, 143, 164, 182 The exception

is one study in which, due to an inadequate sampling schedule, only a one

compartment model could be supported.142 The terminal elimination tR½R in adults has

been reported between 68 and 275 h.135‐138, 164

Several PK evaluations of LUM in children have been reported.142, 144, 183 However,

methodological issues complicate comparisons with adult data. None of these studies

performed adequate sampling for accurate calculation of the terminal elimination tR½R.

The authors of one of these studies suggested that children may be under‐dosed, as

they found a lower AUC in comparison with healthy adults.144 Due to limitations from

26

their study design, direct comparisons with data from adults with malaria were not

possible.

With regards to the PK of DBL, little has been reported.143, 184 In the two available

studies, the ratio between DBL and LUM was low at between 0.33% and 5.2%,143, 184

and the terminal elimination tR½R of DBL, only reported in one study, was longer than

that of LUM at 137 h vs. 68 h.143

As with ARM, a number of PK interactions for LUM have been investigated (Table 1‐3).

No significant differences in LUM PK were seen when QN was co‐adminstered with

AL.137 MQ decreased the exposure to LUM without affecting its tR½R, potentially through

its effect on reducing bile production and therefore LUM absorption.136 In contrast,

lopinavir/ritonavir increased the AUC of LUM more than two‐fold and this was

associated with a non‐significant increase in the elimination tR½R. This may represent

CYP3A4 inhibition, although the same study found the exposure to ARM, also a CYP3A4

substrate, was decreased.135 Ketoconazole, a potent CYP3A4 inhibitor, also increased

the AUC of LUM. As the elimination tR½R was not significantly altered with ketoconazole

administration, it is likely that increased bioavailability through inhibition of FP

metabolism was the cause of the increased AUC.138

Table 1‐5 Fractional difference in tR½R and AUC of lumefantrine when drugs are coadministed with artemether/lumefantrine

Study Coadministered drug lumefantrine

tR1/2 AUC

Lefevre et al. 2000136 mefloquine 0.74 0.56a

Lefevre et al. 2002138 ketoconazole 0.93 1.65a

Lefevre et al. 2002137 quinine 1.13 1.02

German et al. 2009135 lopinavir/ritonavir 1.14 2.35a a P<0.05.

27

1.2.3 31B31B4‐aminoquinilones

The development of the 4‐aminoquinilone antimalarials began in the textile industry

where methylene blue, one of the original man‐made textile dyes, was used as the first

synthetic drug against malaria. In 1934, modifications of the structure of methylene

blue led to the development of CQ, the first 4‐aminoquinilone. From this time, further

modifications were performed to yield a number of derivatives. These include AQ and

NQ, with an alteration in the side chain, and PQ, the result of dimerising two 4‐

aminoquinilone moieties (Figure 1‐9). They have antimalarial action by interrupting the

process of detoxifying heme in the parasite, a side‐product in the digestion of Hb. This

results in increased oxidative stress and, ultimately, death of the parasite.185

Figure 1‐9 Some 4‐aminoquinilones antimalarials showing the common 4‐aminoquinilone group in chloroquine, amodiaquine, naphthoquine and piperaquine (a dimer).

CQ is considered safe in the doses typically used in the treatment of malaria. When

taken in overdose, CQ can lead to death primarily through cardiotoxicity.83 In usual

doses, pruritus is a particular feature and affects compliance to CQ.186 More serious

side effects are related to long term use of CQ and include neuropathy and

retinopathy.187 Idiosyncratic reactions including rash and bone‐marrow toxicity, have

been reported.187 The toxicity of PQ has not been characterized as well as that of CQ. A

review of more than 2,600 adults and children indicated that it was safe and well

tolerated.188 Less information regarding the toxicity of NQ exists in the literature. A

28

review of 952 adults reported a side‐effect profile similar to that of other antimalarial

drugs.189 This review included data from the manufacturer not available elsewhere in

the literature, with one of the two authors being an employee of the manufacturer.

The same review included a number of studies of NQ have also been performed in

children, with no significant safety concerns thus far.189

1.2.3.1 79B79BPiperaquine

Of the dimeric 4‐aminoquinilones (bisquinilones), PQ is the most commonly used and

well‐researched. It was developed in the 1960s in China, where it was used

extensively. It was initially used as a monotherapy in China, Cambodia and Vietnam. Its

uptake in other parts of the world only began recently. It is available as two forms in

ACT combinations. Its tetraphosphate salt is co‐formulated with DHA (e.g.

Duocotexcin) and its base is co‐formulated with ART (e.g. Artequick). The former has

been extensively researched, and recently added to the list of WHO recommended

ACTs.69 There are few data for the PQ base containing combinations. In vitro, PQ has

activity against P. falciparum similar to that of CQ against CQ sensitive strains.190 A

number of studies have reported a correlation between the ICR50R of CQ and PQ.21, 190‐192

There is also some evidence that the in vitro activity of PQ is independent of resistance

to CQ as summarised in a recent review by Gargano et al.193 This review was written by

an employee of Sigma‐tau, a manufacturer of a formulation of DHA/PQ

tetraphosphate, raising a potential conflict of interest. A number of reports have

shown antagonism in vitro between PQ and DHA.191, 194, 195 The clinical significance of

this observation has not been determined, and is likely to be small.

Gargano et al193 also summarised the results of a number of efficacy studies involving

DHA/PQ, and reported that it is highly effective against both P. vivax and P.

falciparum.193 This review included studies from Asia and Africa that included both

children and adults. Many of the studies reported the 28 day efficacy, rather than the

42 day efficacy. For PQ, given its long tR½R, a longer 42 day efficacy is recommended by

the WHO.69 Table 1‐6 summarises the PCR‐adjusted efficacy of DHA/PQ in studies that

reported efficacy at 42 days or later. Two Cochrane reviews also concluded that

DHA/PQ tetraphosphate was efficacious against both falciparum and vivax malaria.70, 71

29

In contrast to these reports, an efficacy trial in PNG children with malaria found the

combination performed poorly against falciparum malaria, despite a reasonable

efficacy against vivax malaria. In this study, the PCR‐adjusted 42 day efficacy was 88%,

Table 1‐6 PCR‐adjusted efficacy of dihydroartemisinin/piperaquine at day 42 or afterwards in various studies.

Study Country (site) n

Pf d42 efficacy

(PCR‐adjusted) (%)

Denis et al. 2002196 Cambodia (Anlong Veng) 34 94.1a

Cambodia (Snoul) 55 96.4a

Hien et al. 2004197 Vietnam (Ho Chi Minh) 166 98.7a

Ashley et al. 2004198 Thai‐Myanmar border (Mae La, Mawker Thai, Muruchai)

170 100

Ashley et al. 2005199 Thailand (Mae Sot) 179 98.3

Smithuis et al. 2006200 Burma (Dabhine, Mingan) 152 99.3

Mayxay et al. 2006201 Laos (Phalanxay) 105 100

Ratcliff et al. 2007202 Papua (Timika) 289 95.9

Kamya et al. 2007 144 Uganda (Apac) 211 93.1

Janssens et al. 2007 203 Cambodia (Kvav, Anlong Veng) 162 97.5b

Hasugian et al. 2007 202 Papua (Timika) 114 95.2

Zongo et al. 2007204 Burkina Faso (Bobo‐Dioulasso) 172 97.8

Grande et al. 2007205 Peru (Iquitos) 230 98.3

Yeka et al. 2008 206 Uganda (Kanungu) 215 98.0

Karunajeewa et al. 2008 21 Papua New Guinea (Alexishafen, Kunjingini)

100 88.0

Arinaitwe et al. 2009207 Uganda (Tororo) 351 95.8b

Thanh et al. 2009 172 Veitnam (Ninh Thuan) 49 100

Bassat et al. 2009 208 Burkina Faso (Nanoro) 198 90.4

Kenya (Kilifi) 133 89.5

Mozambique (Manhica) 274 90.9

Uganda (Mbarara) 164 93.9

Zambia (Ndola) 192 92.7

total 879 91.5

Valecha et al. 2010 209 Thailand, Laos, India (various) 668 99.3

Mayxay et al. 2010 209 Laos (Phalanxay, Xepon) 197 99.5 a efficacy at day 56, b efficacy at day 63.

30

lower than the average 92.4% and 99% in multicentre studies in Africa and Asia

reported in the review by Gargano et al.193 The results of this trial, and the hypotheses

presented for the lower than expected treatment efficacy against falciparum malaria

including a relationship between the in vitro efficacy of PQ and CQ,192 were brought

into question by Gargano et al.193 No alternative explanation for the result is presented

in the review. The results from the PNG trial are not the lowest reported efficacy of the

combination. At one of the sites of an efficacy trial in Rwanda, Rukara, the 28 day PCR‐

adjusted efficacy was only 87%,210 even lower than the comparable 90.1% 28 day PCR‐

adjusted efficacy in the PNG study. The potential cross‐resistance of PQ with CQ was

also proposed as a potential cause for the low efficacy, as Rukara was in an area well

known for its high CQ resistance.210 In the original analysis of the multicentre study in

Africa, the average 42 day efficacy was only 91.5% (extended per‐protocol),208 as

opposed to the 92.4% stated in the review by Gargano et al.193 Lower rates were found

at individual sites, specifically 89.5%, 90.4% and 90.9% in Kenya, Burkina Faso and

Mozambique, respectively.208 In fact, when considering the results in Kenya and

Mosambique alone, AL compared to DHA/PQ tetraphosphate was significantly more

efficacious at 28 and 42 days as evidenced by a lower PCR corrected failure rate

(P<0.05, Fisher’s exact test, two‐tailed level of significance).208 The studies that have

reported lower efficacies consisted primarily of paediatric subjects. Children have

lower plasma concentrations of PQ on day 7 (see below) and this may, in part, explain

the differences in efficacy between studies. Although an adequate explanation for the

low efficacy of PQ in some areas is yet to be established, the results of the PNG trial

are not as unique as suggested by Gargano et al.193

ART/PQ base is another ACT where PQ is used as the partner drug, albeit in a different

form. This combination is currently marketed as a two day regimen. It therefore is not

in line with current WHO recommendations for at least three days of a artemisinin

compound.69 When three days of an artemisinin compound are given, at least two

erythrocytic cycles are exposed to the drug. A lower number of parasites are therefore

exposed to the partner drug, reducing the risk of parasite resistance. Limited data exist

regarding the efficacy of this combination, with mixed results compared to DHA/PQ

tetraphosphate.211, 212 These two studies in adults demonstrated a 28‐day efficacy of

94 and 100% for the two day regimen. A three day regimen has also been proposed in

31

the literature, and was found to be more efficacious than the two day regimen.213

These early studies, performed in a small number of subjects, represent preliminary

data only and further evaluation is required to assess the absolute and comparative

efficacy of this combination.

The PK of PQ has been primarily investigated after oral administration of PQ

tetraphosphate. The absorption of PQ is relatively slow, with maximum plasma

concentrations reached 2.5 ‐ 4 h post dose172, 176, 179, 214 and an absorption tR½R of 7.7 ‐

9.2 h.174, 177 A number of studies have also demonstrated multiple concentration peaks

following each dose, presumably corresponding to enterohepatic recycling of the

lipophilic drug.177, 215 Discrepancies exists regarding the effect of fat intake on the

bioavailability of PQ tetraphosphate.173, 176, 214, 215 One study of patients with malaria

that used a standard local meal, as opposed to healthy volunteers consuming a high fat

meal,215 found no increase in PQ exposure associated with fat intake.214 The use of

compartmental analysis has often found a two‐compartment model to be suitable.174,

175, 178 One study found that a three‐compartment model better described the time‐

concentration profile. However, there were insufficient data to support it over a two‐

compartment model.177 The disposition of PQ may be more complex and the ability to

detect lower concentrations suggested a triphasic elimination in one study.216 Related

to this finding, the elimination tR½,R of PQ, reported to be between 224 and 667 h,172‐179

is positively associated with the duration of sampling.216 No PK evaluation of PQ base

has been reported.

A limited number of PQ PK studies have been performed in children. In comparison to

adults, children have been found to have lower day 7 concentrations,202 higher relative

CL,174 and lower PQ exposure at critical times.178 These findings suggest that current

dosing in children may be inadequate.

1.2.3.2 80B80BNaphthoquine

NQ, as its phosphate salt, was registered as a medicine in China in 1993 where it was

initially tested as a monotherapy.217 It is structurally similar to AQ, with a large

aromatic side chain on the 4‐aminoquinilone moiety (Figure 1‐9). Currently, it is

available as a co‐formulation with ART, although there is sparse clinical data regarding

32

this combination in the literature. In vitro, NQ has potency similar to PQ. Its in vitro

activity may be correlated with that of CQ,192 although only one report has

investigated its in vitro interactions with other drugs.194 As with PQ (see above),

antagonism between DHA and NQ has been noted.194 In the mouse model, synergism

of NQ with ART against P. berghei was demonstrated.217

The efficacy of ART/NQ has been summarised by Hombhanje et al.,189 who had access

to archival data from the manufacturer, the majority of which were not available

elsewhere in the literature. It should be noted that one of the two authors of this

review is an employee of the manufacturer. The combination is effective against

falciparum malaria in adults and children in Africa and Asia, including PNG, and is

comparable to other ACT combinations including ARM/LUM and DHA/PQ

tetraphosphate.189 One study of adults in this review, however, reported a 28 day

efficacy of 92%, lower than the 95% recommended by the WHO for a new first‐line

treatment. The remainder of the eleven studies reported a 28‐day efficacy of > 96%,

with three of the studies in paediatric populations and two others including children.

The tR½R of NQ is comparable to that of PQ (see below). Therefore, given the WHO

guidance, a 42 day efficacy would also be the recommended end point for efficacy

studies of NQ. Most studies of NQ to date have only performed 28 day follow up.189 No

significant difference in cure rates with a single dose or two doses over 24 h has been

found.189 The recommendation by the WHO to have at least three days of the

artemisinin compound is in relation to reducing the risk of resistance to the partner

drug in ACTs (as discussed above). Therefore, the risk of resistance to NQ is increased

by using a single dose regimen. In vitro data have demonstrated the potential for

parasite resistance against NQ.218

Few data exist regarding the efficacy of this combination against P. vivax. Initial reports

are mixed, ranging from 64% (28 days, non‐PCR‐adjusted)219 to 90% (56 day, non‐PCR‐

adjusted)217 and 99% (42 day, PCR‐adjusted).189 Further data are required to better

define its efficacy against vivax malaria.

The first description of the PK of NQ in Western literature appeared in the conclusion

of a paper in 2004 without reference to the original source of the data.217 After oral

33

administration of NQ alone to 14 healthy volunteers, peak plasma concentration were

achieved 2‐4 h post dose and the elimination tR½R was reported to be between 41 and 57

h.217 Little other PK information was given, other than a sampling duration of 168 h. A

more detailed examination, also in healthy volunteers, was later published.220 The time

to peak plasma concentration was similar to that of the initial report, between 2.5‐3.5

h. Some results were inconsistent within different sections of this study (Table 1‐7).

The range of reported elimination tR½RRsR had significant variability in different sections,

from 156‐299 h.220 For example, when comparing the fed and fasted states in the same

participants, the tR½R was reported to be 156 and 276 h respectively.220 This would

suggest an increased CL over many days caused by the ingestion of a single meal, an

implausible consequence of food intake.220 A more likely cause for this finding would

be problematic study design. Blood samples were only taken out to 216 h, about the

same as the reported tR½R for NQ. Therefore, the study design was not adequate for

determining the elimination tR½R. The results from these two studies would suggest that

NQ has a long tR½R, and a longer sampling duration would be required to characterise it

adequately.

The same study reports that the combination with ART results in a doubled exposure

of NQ, represented by an AUC from 0‐216 h, compared to when the drug is given

alone.220 There was little change in elimination tR½R associated with this change, from

299 h when given alone to 276 h in combination.220 This suggests that NQ has a higher

bioavailability in the combination compared to when administered alone. Different

subjects were used to make this comparison and no patient characteristics were given

to assess if the two groups were comparable. An explanation for this finding was not

provided, and it is not easily explained given the current knowledge of these two

drugs.

Table 1‐7 Summary of findings of the pharmacokinetics of naphthoquine in healthy volunteers in Qu et al.220

Dose Combined/Alone Fasted/Fed Elimination tR½R (h) AUCR0‐216hR (g h l‐1) 200 mg Combined Fasted 256 480

400 mg Combined Fed 156 446

400 mg Alone Fasted 299 424

400 mg Combined Fasted 276 955

600 mg Combined Fasted 233 1875

34

An unusual relationship with coadministed food was also noted. Using a crossover

design, the exposure to NQ was approximately halved compared to a fasted state.220

This finding is not expected for the lipophilic drug, and is in contrast to findings in

PQ,173, 176, 214, 215 and LUM180, 181 where the relative bioavailability either remained

constant or increased with the administration of fat. The washout period used, 168 h,

is too short for a drug with a tR½R of at least 200 h. However, as the fed stage of the

study followed the fasted stage, the short washout period would not explain this

unusual finding. No other studies have evaluated the effect of food on the PK of NQ.

No studies of the PK of NQ in patients with malaria, or children, exist in the literature.

35

1.2.4 32B32BAntifolatedrugs

Antifolate drugs were first used against malaria in the 1930s when sulfachrysoidine, an

antibacterial agent, was used in a trial against malaria (Figure 1‐10). Development was

stagnant until the 1950s when sulfadoxine (SDX), a sulfonamide, was developed and

partnered with PYR, a compound with which sulfonamides were found to have a

synergistic effect.221 Broadly, there are two groups of antifolate drugs categorised with

respect to their targets in the folate synthesis pathway. These are dihydrofolate

reductase (DHFR) inhibitors, such as sulfachrysoidine and SDX, and dihydropterate

synthase (DHPS) inhibitors, such as PYR.145 The mechanism of action of these drugs is

through preventing folate synthesis, and thereby interrupting DNA synthesis in the

malaria parasite. The combination of SP is the most commonly used antimalarial

antifolate treatment today, and contains both a DHFR and a DHPS inhibitor.

Figure 1‐10 Antifolate antimalarial drugs.

These drugs have been used extensively and their safety profiles are well known.

Although humans lack DHPS, DHFR inhibitors do interrupt folic acid synthesis in

humans. Drugs like PYR are known to exacerbate subclinical folate deficiency in

vulnerable subjects. However, in other subjects the antifolate effects are only seen

after long‐term administration.222 SDX, like other sulfonamides, can cause cutaneous

hypersensitivity reactions including the potentially fatal Stevens Johnson Syndrome. In

combination with PYR, this serious side effect is predominately only found when used

36

for long‐term prophylaxis of malaria which is no longer a recommended practice.83 This

side effect is rare when SP is used for the treatment of malaria.83 Other rare

hypersensitivity reactions include vasculitis, myocarditis, acute glomerulonephritis and

pulmonary reactions.83 Haematological toxicity caused by SP, besides that cause by

folate depletion, is thought to be primarily caused by SDX and includes

agranulocytosis, aplastic anaemia and thrombocytopaenia.223 SP has been found to be

safe in the treatment of malaria, with serious side effects being a rare event.

1.2.4.1 81B81BSulfadoxine/pyrimethamine

For antimalarial treatment, SDX and PYR are always used together. Initially, this

combination represented a viable alternative to CQ to which parasite resistance was

growing. After enjoying a brief period of high efficacy, resistance against this

combination occurred quickly, within five years in some areas.224 After this, SP was

used in combination with CQ or ARTS with some success. The most significant role of

SP as an antimalarial has been its use in IPTi and IPTp, as discussed above.

The use of SP in infants older than 2 months has been extensive but it is

contraindicated in younger infants due to the immaturity of their metabolising enzyme

systems.225 There have been few PK evaluations that have included infants to allow

adequate assessment of the exposure to SDX and PYR in this age group. A study of

patients with malaria aged from one year old to adulthood has highlighted the

importance of assessing the PK in the young.226 This study found that children aged 2‐5

years had a double risk of treatment failure in association with a lower exposure to

both SDX and PYR.226 A doubled dose for this age group was, therefore, suggested as

an appropriate alteration to dosing. Although children aged 1‐2 years were also

included in this study, no dose adjustment was found to be necessary.226 These

younger children already received a higher mg per kg dose relative to the other age

groups, and only 11 children were in this age group.226 A population PK study in

children, including infants with congenital toxoplasmosis, found that the elimination tR½R

of both drugs was related to WT.227 A shorter elimination tR½R was found in lighter, and

thus younger, children227 as would be expected from allometry.228 The potential effect

of maturation of elimination processes on the PK of SP in infants has yet to been

established.

37

There is also some evidence that the concentrations of SDX and PYR are lower in

children with malaria than in healthy adults, signifying a potential disease effect of

malarial on PK.229, 230 As these comparisons were made between age groups in

different populations, and very few infants have been included these studies, it is

difficult to draw conclusions regarding the potential impact of malaria on the PK of SDX

and PYR in infants. In IPTi, infants would be in one of three disease states. Some

infants would be afebrile and aparasitaemic, others would be afebrile with a positive

blood smear and the last group would be febrile and slide positive for malaria.

Potentially, PK differences may exist between these three groups. These potential

differences have not yet been investigated.

38

1.2.5 33B33BAntibiotics

Some antibiotics have significant activity against Plasmodium species, although the

parasite is not a prokaryote like bacteria. This antimalarial action is facilitated by the

presence of two organelles in the malaria parasite, the mitochondrion and the

apicoplast. These organelles have their own DNA and have similar processes of protein

production to bacteria.231 The apicoplast is a relic plastid of red algae lineage, devoid of

photosynthetic activity. Apicoplasts are also found in Toxoplasma species.231 It is the

site of antimalarial activity of most antibiotics that have action against Plasmodium

species.231 Tetracyclines, such as doxycycline, also act on the mitochondrion.232 It has

recently been shown that the only essential function of the apicoplast in Plasmodia is

the production of isoprenoid precursors.231 These precursors are involved in a number

of cellular processes including energy production in the mitochondia.231 For all

antibiotics, a ‘delayed death’ phenomenon is seen where the exposed parasites are

able to develop normally while the apicoplast in the next generation of merozoites

cannot divide, and the resultant parasites lack this essential organelle and die.231

Therefore, in the assessment of in vitro antimalarial activity of antibiotic agents, it is

important to allow for longer culture times so that their full effect can be assessed.233

Figure 1‐11 Antibiotics with activity against Plasmodium species.

The toxicity of these compounds has been reported primarily from their use as

antibiotic agents. For example, doxycycline and related tetracyclines are

contraindicated in pregnancy because of their incorporation into teeth and bone234 as

39

well as a small teratogenic risk.235 Unlike doxycycline, AZI is safe to use in pregnancy.236

The Centers for Disease Control and Prevention recommends 1 gram (g) of AZI as first

line treatment of Chylamidia trachomatis in pregnant women. This recommendation is

based on studies confirming its effectiveness, as well as its safety, in pregnancy.237 The

most common side effects of AZI are gastrointestinal, including nausea, diarrhoea and

abdominal pain, but not psudomembranous colitis.238 These side effects are thought to

result from the interaction of AZI with the motilin receptor in the gut wall.239 Some

transient increases in liver enzymes have also been noted.238 No deaths have been

attributed to AZI. Studies of AZI in the setting of malaria have found similar results with

respect to tolerability and safety.240‐242

1.2.5.1 82B82BAzithromycin

AZI is derived from erythromycin, and is a semi‐synthetic azalide antibiotic, a subclass

of the macrolides. It is commonly used in the treatment of respiratory and sexually

transmistted diseases and is available in oral and intravenous formulations.243, 244 In

vitro it has activity against P. falciparum comparable to other compounds in the

antibiotic class of antimalarials245, 246 and is more potent than erythromycin.247 It is

more potent against CQ‐resistant than CQ‐sensitive strains.246

A number of trials have evaluated its use in the treatment of malaria. A Cochrane

review reported that it performed poorly in the treatment of falciparum and vivax

malaria, both as a monotherapy and in combination with CQ or ARTS.248 One trial in

pregnant women found that compared to three days of ARTS with SP and SP alone (the

standard treatment), two days of 1 g AZI with SP resulted in fewer recrudescent

episodes in the treatment of falciparum malaria.249 Therefore, AZI is unlikely to have a

role in the treatment of uncomplicated malaria outside of pregnancy.

The benefits of AZI for IPTp extend past its antimalarial effect. In developing countries,

sexually transmitted infections and, to a lesser extent, pneumococcal infections

contribute to morbidity in pregnancy.250 A large trial of IPTp in pregnant Malawian

women compared three regimens given during pregnancy; i) two single doses of SP

(control group) ii) monthly single doses of SP, and iii) monthly single doses of SP

combined with two single 1g doses of AZI.251 When compared to the control group, the

40

AZI group had a significantly lower rate of preterm delivery, higher birthweight and

fewer postpartum Trichomonas infections.251 The only significant benefit over the SP

monthly group was a decrease in the rate of very preterm birth (<35 weeks

gestation).251 In the AZI group, there was a trend for lower maternal and placental

blood parasitaemia at delivery compared to the other two groups.251 This did not reach

statistical significance, probably due to the small number of events.251 Neonates of the

mothers in the AZI group had significantly lower rates of hospitalization when

compared to the other groups.251 In contrast to these results, a similar IPTp trial found

that there was no additional benefit of 1 g AZI given twice in pregnancy over two doses

of SP in preventing preterm birth, low birthweight, maternal anaemia or maternal

parasitaemia.252 These differences may in part be explained by fewer doses of SP in the

trial without significant results (two doses compared with monthly dosing), a higher

percentage of first pregnancies as well as a greater burden of malaria and syphilis.

These factors could influence the effect of AZI on pregnancy. In addition to the

combination of AZI with SP, a combination of AZI with CQ is also being investigated for

IPTp with results pending.253

The 1g single dose of AZI used in the studies mentioned above is the suggested dose

for sexually transmitted diseases, which may not be suitable in the treatment of

malaria (see below). Neither of these studies assessed the impact of AZI on the

histological appearance of the placenta.251, 252 Histologically, pathological features in

placental malaria can be used to distinguish between active, active‐chronic and past‐

chronic infections,254 each of which is associated with particular clinical outcomes.255

An efficacy trial of SP with AZI against falciparum malaria in pregnancy was the only

study to evaluate this outcome, and was underpowered to assess differences in

placental features.249

Few studies have been published that evaluate the PK of AZI, and no PK data exist for

malaria‐infected patients. Absorption of AZI has been described by both zero256, 257 and

first258, 259258, 259 order kinetics. Both two256, 257, 259 and three compartment models258,

260 have been reported, with an elimination tR½R of 27‐79 h.256‐258, 260‐262 Only one PK

evaluation has been performed in pregnant women,263 but it has significant limitations.

Twenty women were given 1g of AZI at set times before a planned Caesarean section,

41

during which samples for AZI assay were taken.263 Each subject, therefore, only

contributed one sample to the analysis. An elimination tR½R of 12 h was reported,263

significantly shorter than that reported in non‐pregnant adults. Due to the small cohort

and cross‐sectional nature of this study, it is difficult to interpret the significance of this

result. The PK of AZI in earlier stages of pregnancy is likely to differ to those in term‐

gravid women and has not been determined to date. One PK study has shown no

interaction between AZI and CQ;264 but no such data exist for SP and AZI.

In bacterial infections, a single larger dose of AZI has been shown to be as effective as

multiple smaller daily doses.256, 265 This effect of ‘front loading’ is likely due to the

accumulation of AZI in white blood cells to concentrations 100‐1,000 fold higher than

in plasma.250, 252 These high concentrations persist over time and carry the AZI to the

site of infection.256, 258 In contrast, the concentration of AZI in RBCs is essentially

negligible after oral dosing.258 Given the need for relatively prolonged parasite

exposure to therapeutic concentrations in vitro, it is likely that prolonged high

concentrations of AZI in plasma, where the parasites encounter the drug, are required.

Therefore, in the treatment and prevention of malaria, two to three day regimens of

AZI would theoretically be preferable to a single dose.

42

43

1.3 9B9BPharmacokinetics

1.3.1 34B34BIntroduction

PK, put simply, is what the body does with a drug. More formally it is, “the study and

characterization of the time course of drug absorption, distribution, metabolism and

excretion”266 The term was first used in German literature in 1953266 but did not

appear in English literature until 1959.267 PK is a relatively new science and is

considered to have begun in 1937 when Teorell published a pair of papers entitled

“Kinetics of distribution of substances administered to the body”.262, 263 In these, he

attempted to describe the distribution throughout the body through the use of an

approximate system of differential equations.268, 269 Since then, the science (and art) of

PK has grown significantly.

One of the aims of PK is to find the dose at which a drug is therapeutic but not toxic.

This concept existed long before the time course of the concentration of a drug in the

body was characterized. Paracelsus, a German‐Swiss physician (as well as a botanist,

alchemist and astrologer) in the early 1500s wrote, “What is there that is not poison?

All things are poison and nothing (is) without poison. Solely the dose determines that a

thing is not a poison.”270 In the case of antimalarials, there is often a lack of

information to guide the dosing in the most vulnerable populations, namely infants,

children, and pregnant women. Therefore, the study of the PK of antimalarials in these

populations is of paramount importance to ensure that they are both effective and

safe.

In very general terms there are two approaches to PK analysis, non‐compartmental

and compartmental, and these are described briefly below. Essentially, they are both

mathematical tools that help describe the concentration versus time course of a drug

in a bodily fluid.

44

1.3.1.1 83B83BNon‐compartmentalAnalysis

Although some of the equations used in non‐compartmental analysis appeared in

1931, it was not until the 1980s that this method was widely used.266 It uses statistical

moment theory to determine the area under the curve (AUC), area under the moment

curve (AUMC) and the variation of residence times.271 These three properties

correspond to the area under the zeroth, first, and second moment curves

respectively.271 These areas are determined using the trapeozid method with

extrapolation to enable their calculation from time zero to infinity. Other parameters

can then be derived, including the mean residence time, mean transit time (MTT), CL,

volume of distribution (V), rate constants and half‐lives. Additionally, simple

information such as the maximum concentration (CRmaxR) and the time to maximum

concentration (TRmaxR) are determined from simple observation of the concentration

versus time plot.

Although non‐compartmental analyses has sometimes been referred to as ‘model‐

independent’, it is based on a model with a far more restricted structure than

compartmental analysis.272 Although a model is not explicitly stated, there must be at

least one compartment in which the observations being modelled exist.272 An inherit

assumption is that drugs modelled with non‐compartment methods display linear

kinetics.272 These restrictions make these models easier to conceptualize, as only the

compartment in which the observations are made (the central compartment) needs to

be considered.272 Any potential peripheral compartment(s) do not need to be

identified or characterised.272

Non‐compartmental analysis incorporates less input, and therefore less bias, by the

modeller and is able to describe simple PK parameters adequately. However, as the

time course of the drug in the measured fluid is not considered, it has limited use in

predicting future drug concentrations, or in explaining physiological processes.

1.3.1.2 84B84BCompartmentalmethods

Compartmental methods were first utilized by Teorell in 1937, where he used a four

compartment model described by a set of differential equations.273 In these models,

there are a number of defined compartments through which the drug passes according

45

to some d

aim is to m

and offer

renowned

are wrong

The simple

Figure 1‐12 O

In this mo

with an am

compartm

compartm

rate is ind

where is

mass/time

order (line

compartm

where is

/time unit

linear kine

efined rate

make them

an explanat

d statistician

g ‐ some mo

est of these

One compartme

odel, there i

mount of dr

ment). The r

ments are us

ependent o

s the amou

e units (mg/

ear) kinetics

ment:

s the amou

ts (/h for ex

etics for exp

. These mo

close enoug

tion of phys

n, no matte

odels are us

e models, w

ent model with

s a single co

rug, , and

ate of input

sually eithe

of the conce

nt of drug i

/h for exam

s where the

nt of drug i

ample). The

planatory pu

dels only re

gh to enabl

siological pr

r how well

seful".274

with only on

h a single outpu

ompartmen

an output r

ts and outp

r zero‐orde

entration of

n the comp

mple). More

e rate is pro

n the comp

e remainde

urposes.

epresent an

e prediction

rocesses. In

these mode

e compartm

ut rate.

nt in which t

rate, (inpu

puts of comp

r or first‐or

f the drug in

partment,

commonly,

portional to

partment,

r of the disc

approxima

n of future

the words

els are cons

ment, is sho

the observa

uts are direc

partments a

der. In zero

n the compa

is time, and

, processes

o the amou

is time, and

cussion in t

ation to rea

drug conce

of a George

structed, "A

own below:

ations are b

ctly into the

and betwee

o‐order proc

artment:

d is the r

in PK follow

nt of drug i

d is the r

his section

lity. The

ntrations,

e Box, a

All models

eing made

e

en

cesses the

Equation 1‐1

rate in

w first‐

n that

Equation 1‐2

rate in

will utilise

46

The tR½R, tim

as the natu

In the case

body, a on

this case, t

changes in

tissues, pe

required to

between tw

dosing com

Figure 1‐13 Tw

In this mod

respective

2 and 3,

elimination

the drug o

compartm

absorbed.

me for 50% o

ural logarith

e where the

e compartm

the changes

all tissues

rhaps due t

o describe t

wo compar

mpartment)

wo compartme

del , a

ly, and

represen

n ( ). Only

nce it enter

ent that rep

In these mo

of the proce

hm of two d

ere is extrem

ment mode

s in the mea

at all times

to differenc

these. A two

tments in t

) with oral d

ent open mode

and repr

represe

ts the rate o

y two of the

rs the body

presents th

odels, obse

ess to occur

divided by th

½

mely rapid t

l would be a

asured fluid

. Often the

ces in perfus

o compartm

he body wh

dosing could

l with oral dosi

esent the a

ent the rate

of absorptio

e compartm

from the ga

e gastrointe

rvations are

r, can be ca

he first ord

ln 2

ransfer of a

able to desc

d would app

rate of tran

sion, and an

ment open m

hile the thir

d be represe

ing and elimina

mounts in c

es of drug tr

on ( ) and

ments, 2 and

astrointesti

estinal tract

e generally

lculated for

er rate cons

a drug throu

cribe the ch

proximately

nsfer may b

n additiona

model (the

d compartm

ented struc

ation from the

compartme

ransfer betw

d repres

d 3, describ

inal tract. T

t from whic

made in co

r first‐order

stant:

E

ughout all ti

hanged adeq

represent t

e slower in

l compartm

drug is distr

ment repres

cturally as:

central compa

ents 1, 2 and

ween comp

sents the rat

e the distrib

here is an a

h the drug

mpartment

r processes

Equation 1‐3

issues of a

quately. In

the

some

ment is

ributed

sents the

rtment.

d 3

partments

te of

bution of

additional

is

t 2, the

47

central co

amount of

As the con

these mod

compartm

multiplied

compartm

can theref

Figure 1‐14 Tparameterize

where

rate const

distributio

clearance.

and half‐li

mpartment

f drug in ea

ncentration

dels are mo

ment is the a

d by the . F

ments, inter‐

fore be repr

Two compartmeed in terms of c

is the am

tant, is th

on, is the

. From thes

ives.

t. In this cas

ch of the co

of drugs ar

ore informat

amount divi

For the tran

‐compartm

resented in

ent open modeclearance and v

mount of dru

he central vo

e clearance

e primary p

se, the equa

ompartmen

re measured

tively param

ided by the

nsfer of drug

ental cleara

a more fam

el with parentevolume parame

ug in the ga

olume of di

of the drug

parameters

ations descr

nts would be

d in studies

meterized in

concentrat

gs between

ance ( ) is u

miliar form:

eral dosing andeters.

strointestin

istribution,

g, and is t

others can

ribing the ra

e:

of PK, rath

n terms of

tion, while t

n the centra

used. The m

elimination fro

nal tract,

is the pe

the inter‐co

be calculat

ate of chan

her than am

and . T

the is th

al and perip

model in figu

om the central

is the abso

eripheral vo

ompartment

ted, includin

ge of the

Equation 1‐4

Equation 1‐5

Equation 1‐6

ounts,

he of a

he rate

heral

ure 1‐13

compartment

orption

olume of

tal

ng the AUC

48

In a compartmental model, the concentration of the drug in each of the compartments

over time can be linked with a multi‐exponential equation. The concentration is the

integral of the rate of change of the concentration of the drug. For example, the rate

of change in concentration, and its integral (the concentration over time) for a one

compartment model with oral dosing would be:

Equation 1‐7

Equation 1‐8

where is the concentration of the drug at a given time, is the dose of the drug, and

is the bioavailability. To determine the true value of , the AUC from an oral dose

needs to be compared to that of an intravenous dose. When this comparison is not

available, as often is the case, and are relative to , and are known as the

apparent clearance ( / ) and apparent volume of distribution ( / ).

Compartmental analysis, therefore, can be considered as a form of curve fitting. An

appropriate curve (model) is chosen that best represents the observed concentration

versus time data. Figure 1‐15 demonstrates an example of this principle.

Time (h)

Dru

g c

on

cen

trat

ion

(g

/l)

Figure 1‐15 Example of a fitted concentration versus time profile. Observed concentrations over time (red crosses) have been fitting using a curve that is the combination of positive and negative exponentials (black line) that represent absorption and elimination processes respectively.

49

1.3.2 35B35BPopulationpharmacokinetics

Determining the PK parameters in an individual can be relatively easily accomplished

using the aforementioned methods. However, providing information on the PK of a

drug in a population is far more valuable, and requires extension of these methods.

The principles of population PK presented here can readily be found in the

literature.275‐281

Broadly speaking there are two types of PK characteristics that need to be determined

in a population, fixed effects and random effects. Fixed effects refer to the population

average parameters and covariate relationships, while random effects refer to the

variation between individuals, between occasions and within individuals.276 Fixed

effects can be expressed as follows:

Equation 1‐9

where represents an individual, is the clearance of individual , is the

population average of the not related to the covariate, is the value of the

covariate for individual , and is the average proportionality constant

relating to .

The first type of random effect is the variability that exists between individuals within a

population, as no two individuals are perfectly identical. The inter‐individual or

between subject variability (IIV or BSV) can be added to Equation 1‐9 to give:

Equation 1‐10

where expresses the difference between the population expected value of for

individual and . For the population, each is normally distributed with a mean of

0 and a variance of . As two PK parameters within a population may be correlated,

covariance terms also exist between s. This information makes up a matrix Ω, which

has diagonal elements equal to the variance of different , while the other elements

describe the covariance between different s. If multiple occasions exist for each

50

individual, then inter‐occasion or between‐occasion variability (IOV or BOV) also exists

in the population.282 This can be added to Equation 1‐10 to give:

Equation 1‐11

where represents a particular occasion, and represents the difference between

the expected individual clearance ( ) and the clearance for that occasion ( ). Each

is normally distributed with a mean of zero and a variance of . The final random

effect that exists in a population is the variability within an individual, also known as

the error or the residual unexplained variability (RUV). As no biological assay method is

perfect and no study is performed perfectly, there is always a component of error in

each observation that cannot be explained using the other random effects mentioned

above. RUV applies to each observation:

∗ Equation 1‐12

where is the final predicted concentration in patient on occasion at time , ∗

is the model predicted concentration based on all the , , and values of the

population, and , the error, is the difference between and ∗ .

1.3.2.1 85B85BNaïvepooleddatamethod

The simplest method for analysing the concentration versus time curves obtained from

many individuals within a population is the naïve pooled data (NPD) approach. As the

name suggests, this method pools all the observations and them analyses then as if

they had come from a single individual. Using this method, only one random effect,

, is estimated:

∗ Equation 1‐13

has similar properties to the aforementioned RUV, and accounts for all variability

from , and . Therefore, this method assumes that there are no covariate fixed

effects and cannot determine IIV or IOV.

51

Although the NPD approach allows for easy analysis of data with few observations for

each individual, even estimates of can be biased and non‐informative.279

1.3.2.2 86B86BStandardtwostagemethod

The standard two stage (STS) method provides benefits over NPD as it is able to

provide some estimate of IIV. As the name suggests, the analysis is performed in two

steps. In the first, individual estimates of the individual parameters ( for example)

are obtained by analysing the curves of each individual alone. Although it is not used

further, the RUV for each individual is also obtained. The individual parameters are

then used to calculate the population average values ( ), as well as provide an

estimate of IIV ( ). There is potential for covariate relationships to be identified and

IOV ( ) to be estimated at this stage.

Although IIV can be estimated using STS it is often overestimated.283 A limitation of

using this method is that the PK profile of each individual must be able to be analysed

individually.281 Individuals with few observations, or missing vital observations, cannot

be included in the analysis. The data from these individuals, as well as the time and

effort in collecting them, is lost when using this method of analysis. A more complex

shortcoming of STS is that individual estimates for parameter do not all have the same

precision, although they are treated equally when calculating population

parameters.279 This becomes another potential source of bias.279

More complex methods of two stage analysis exist, including global two stage and

iterative two stage methods. These have shown improved performance when

compared to STS.283‐285

1.3.2.3 87B87BNon‐linearmixedeffectsmethod

Non‐linear mixed effects modelling for PK was first described in 1972 and was

proposed as a method for analysing routine patient data.286 It avoids the estimation of

individual parameters, instead s (average population parameters) are estimated

directly. All the fixed effects and random effects expected in a population are

estimated simultaneously, hence the name mixed effects modelling. Using this

methodology, individuals with few data points do not need to be discarded. These

52

individuals can be included in the analysis, and unlike NPD, they maintain their

individuality. The output from this method therefore describes the population as a

whole, rather than the individuals within it.

As sparse data can be analysed, this method is ideal for providing PK information in

patient groups where obtaining full rich sample sets for each individual is difficult

either logistically or ethically.277 Additionally, it is capable of handling complexities in

the model structure, such as transit compartments,287 and challenges in the dataset,

such as observations below the limit of quantification (BLQ).288 For these reasons it is

well suited to the analysis of the antimalarials in the study samples presented in this

thesis, where either sampling was sparse (Chapter 3), data sets were incomplete

(Chapters 2, 4and 6), complexities existed in the structural model (Chapter 5), or

challenges were present in the dataset (Chapter 4).

53

1.3.3 36B36BNONMEM

NONMEM is a computer package that was designed by Beal and Shiener to perform

NONlinear Mixed Effects Modeling.289 It is widely used in analysis of population PK

data and is referred to as the ‘gold standard’ in analyses of this type.

NONMEM employs a likelihood ( ) function, which it maximizes with respect to the

fixed and random effects, and the dataset (a maximum method). To determine for

all the data, however, the marginal for each data point must first be calculated. As

there is considerable difficulty in calculating the marginal exactly, different

approximations are used. These approximations correspond to the various estimation

methods within NONMEM (a detailed derivation of each approximation is provided by

Wang 2007290). A note should be made that, although NONMEM utilises a maximum

method (using approximations), it actually minimizes the value of 2 log , called

the objective function. Therefore, the objective function value (OFV) is used to arrive

at the most likely estimates of PK parameters.

Three estimation method will be discussed here, Laplacian, first‐order conditional

estimates (FOCE) and first order. The Laplacian (first level of approximation) and FOCE

(second level of approximation) methods both use individual parameter estimates to

calculate the OFV at each step. They are therefore ‘conditional’ estimation methods, as

they utilize the random inter‐individual effects in the approximation of the .290, 291

They are more time consuming than the FO (third level of approximation) method,

which only uses the population parameter estimates in the calculation of the OFV.290,

291 Additionally, both the Laplacian and FOCE method can be used with , interaction

(INTERACTION option), where that the effect of on will be considered. This

becomes important when a proportional, or other heteroscedastic, error model is

used. In these error models the size of the effect of on ∗, the model predicted

concentration, will depend on the value(s) of the (s) for the individual, as these effect

the size of ∗.

One important role of the OFV is to allow hypothesis testing of nested models, using

the likelihood ratio test (LRT).292 Two models are nested where the only difference

between them is that one (or more) of the parameters has been fixed (usually to 0) in

54

one model, and is estimated in the other. This applies to covariates (if parameterised

appropriately), and also to structural parameters. The difference between the OFV of

nested models is approximately distributed chi squared ( ), with the degrees of

freedom equal to the number of parameters fixed in the reduced model.291 This is

known as the LRT. For accurate hypothesis testing to take place, FOCE‐I (FOCE with ,

interaction) or Laplace‐I (Laplacian with , interaction) should be used, with a

preference towards Laplace‐I in models with greater non‐linearity.291

1.3.3.1 88B88BModelbuilding

A process of model building needs to take place to arrive at a final PK model,

preferably with the use of hypothesis testing.293 First the data must be checked for

errors of data entry such as impossible or unrealistic times or covariate values. Then,

the structural model is established, with the nature of absorption and elimination

processes as well as the number of compartments determined. In other words the

population parameters, or θ, of the model are established in this step. Alongside this,

the most appropriate model for RUV (ε) should be established.

In general, three types of RUV models exist: additive, proportional and combined.293 Of

these, the combined method is preferred as it takes into account that the size of the

RUV is likely to increase with increasing observations, and that the RUV can never be

zero. If the data is log transformed, an additive RUV model approximates the

proportional RUV model for untransformed data. The main difference is that the

estimated RUV can never be zero when an additive model is used for log transformed

data. The RUV can also be a marker for model selection. A reduction in the RUV

suggests that more of the variability in data is explained.

The IIV and IOV ( and respectively) can then be added to the model, and tested to

assess if they can be estimated (diagonals of the Ω matrix, Figure 1‐16). This depends

on the number of individuals in the population and the number and distribution of the

observations. The correlation between s and s should also be estimated (off‐

diagonals of the Ω matrix, Figure 1‐16), particularly if the model is to be used for

simulation purposes.

55

, , ,

, , ,

, , ,

Figure 1‐16 A typical matrix with variance terms, the diagonals, in blue ( , is the variance for ), covariance

terms, the off‐diagonals, in red and black ( , is the covariance between and , , is the same as , ).

Covariate relationships can then be assessed and tested. Considerations for inclusion

of a covariate relationship include the change in OFV, the biological plausibility of the

relationship, and a decrease in the IIV of the parameter in the relationship. In this way,

part of the IIV can become explained while the other part remains random. One

method for incorporating categorical covariates is as follows:

1 Equation 1‐14

where is the fractional effect of 1 on , and is value of the

covariate for individual . The value of the covariate is usually either 1 or 0, although

covariates with more categories are possible. Potential categorical covariates include

co‐administered substances that induce/inhibit metabolising enzymes or compete at

transporter sites, genetic factors influencing metabolism, disease states (such as

malaria), sex and pregnancy. The latter covariate is discussed in more detail in section

1.3.4.1.

For continuous variables, a number of functions can be tested, such as a centred linear

model:

1 Equation 1‐15

where becomes the fractional effect of a unit increase in on , and is

the median of in the population. A power function is also possible:

Equation 1‐16

56

where now represents the power relationship between and . Given the

flexibility of NONMEM, any covariate relationship can be defined. Nevertheless, the

biological plausibility should be considered before the implementation of any covariate

relationship. Two important types of continuous covariates are those that describe size

and age (discussed further in section 1.3.4.2). Other potential continuous covariates

include gestational age (instead of the categorical pregnancy covariate), parasitaemia,

glomerular filtration rate (GFR) and biochemical indices for an individual.

Some additional aspects of model building and specific covariates particular to parts of

this thesis are discussed below.

1.3.3.1.1 135B135BAbsorptionmodels

The absorption phase of drugs can be complicated, particularly when considering that

some drugs with poor solubility first need to become available for absorption.294

Absorption is often suitably described with first‐order kinetics, while zero‐order

kinetics can also be appropriate in some cases. A lag time between the ingestion of the

drug and its appearance is often required, and can improve the models of datasets

that demonstrate this phenomenon.287 More complex models that include either

combined or sequential zero‐ and first‐order kinetics have also been used.294 The

wrong absorption model had been found to impact the estimation of other parameters

such as and .294 Recently, a model utilizing a mathematical approximation for

transit compartments has been developed. This model allows the estimation of the

number of transit compartments (NN) and the transit compartment rate (kRtrR) as

continuous variables ( s), with their own and .287 This transit compartment model

has been found to be superior in describing the delayed appearance of a number of

drugs after oral administration when compared to those using lag time.287

1.3.3.1.2 136B136BBelowthelimitofquantificationdata

A commonly encountered problem in the analysis of drugs within a biological matrix is

samples with unmeasurable, or even undetectable, concentrations of the drug.288

These BLQ data have been shown to significantly impact the estimates of fixed and

random parameters, resulting in significant bias, even if they only represent 5% of the

entire dataset.295‐297 In 2001, seven ways of dealing with these data in NONMEM were

57

presented. Some of these methods were omission of the BLQ data, setting the BLQ

data to a particular set value, and more complex likelihood‐based methods.288 A

slightly altered version of one of these methods, M3 with F_FLAG,296 has been shown

to consistently perform well in various situations, including those with the percentage

of BLQ data as high as 40%.295‐297 The implementation of this method has been

simplified from version 6.2 of NONMEM onwards and is described in Ahn et al.296

1.3.3.2 89B89BModelevaluation

An important step in the development of a population PK model is the evaluation

phase. Generally speaking, there are internal validation and external validation

techniques.298 The latter is only possible when there is another similar dataset

available to test the final model against. Although a portion of the original data can be

set aside for this purpose, often there are limited numbers of subjects available

initially, which makes this strategy impractical.

1.3.3.2.1 137B137BBasicinternalevaluation

Basic internal validation includes the assessment of goodness‐of‐fit (GOF) plots. GOF

plots include population and individual predictions (PRED and IPRED respectively)

versus observations (OBS), overlaid PRED on OBS versus time, and weighted residuals

(WRES) versus time.293 Recently a new diagnostic, the conditional weighted residual

(CWRES), has been described.299 For models using FOCE estimation, examination of

WRES can suggest an inadequate model is adequate, or that an adequate model is mis‐

specified.299 This is because WRES are calculated using the FO approximation, and

therefore CWRES, calculated based on the FOCE approximation, are more appropriate

when using FOCE estimation.299

1.3.3.2.2 138B138BAdvancedinternalevaluation

One method for obtaining the confidence intervals (CI) of the model estimates is the

non‐parametric bootstrap.300 By sampling with replacement from the study

population, new datasets are obtained which are then analysed with the final model

obtained in the model building stage. The 2.5th and 97.5th percentiles from the

resulting estimates for each of the fixed and random effects can be used to obtain the

empirical 95% CI. Although some reports include the rates of success of these

58

bootstrap results, there is no evidence that these rates indicate model stability or any

other diagnostic.301, 302

The visual predictive check (VPC) originates from the posterior predictive check and is

a simulation based form of evaluation.303 A number of concentrations are simulated

from the final model for each time point in the dataset. The simulated concentrations

are then compared to the observed concentrations to determine the internal

predictive performance of model.303 An extension of this method allows evaluation of

data with BLQ, one of the few tools available in this setting. The 95% CI of simulated

fraction of BLQ data are compared with the actual fraction of BLQ data at set time

periods.296 More recently an improved method for VPCs, known as prediction‐

corrected VPCs (pcVPC), has been developed.304 This method corrects for the

differences in the size of the observations and, therefore, presents a more accurate

visual representation of the predictive performance of the model. 304

A similar method is known as the NPC. NPC use the same simulated data that is

produced with a VPC, yet it presents them in a different way. The number of actual

data below and above different prediction intervals (PI) is compared with 95% CI,

obtained from the simulated data. Unlike a VPC, there is no independent variable,

usually time, in a NPC.

Both VPCs and NPCs can be stratified according a covariate in the model. This is

particularly important when there are two or more groups in the datasets. For

example, when comparing two formulations of the same drug, it is important to assess

if the predictive performance of the model is adequate for each formulation.

59

1.3.4 37B37BPharmacokineticconsiderationsinspecificpopulations

1.3.4.1 90B90BPregnancy

The effects of biological changes that occur during pregnancy which have potential

effects on PK have been reviewed in the literature.305, 306 Changes in gastric emptying

and pH, increased plasma volume with associated decreased concentrations of plasma

proteins, increased body fat, inhibition as well as induction of hepatic enzymes, and

increased glomerular filtration, all have potential implications for PK.305, 306

Pregnancy is known to delay gastric emptying and decrease the acidity in the stomach.

Despite this, the bioavailability of many drugs appears unchanged in pregnancy.306

In pregnancy, the increased plasma volume with associated decrease in drug binding

plasma proteins has a number of effects on PK. One expected change would be an

increase in the V, and therefore a decrease in the peak concentrations.307 Additionally,

due to the higher V, the elimination tR½R would be expected to increase:

½ ln 2 Equation 1‐17

As changes also occur in (see below), the final result may be a decrease, increase or

no change in tR½R.307 These effects, although at times measurable, have not yet proved

to be clinically important.307

Due to the reduced concentration of plasma binding proteins, the fraction of unbound

drug, and therefore the concentration of drug available to act at receptors or be

cleared by the organs of elimination (correlated with ), may be expected to

increase.306 This is particularly important for drugs with high plasma binding and low

extraction ratio, such as phenytoin, where a small decrease in the fraction bound can

produce a large increase in the concentration of free drug.306 Therefore, the effects of

the increased and decreased binding need to be considered simultaneously. Often,

the only solution for such drugs is to measure unbound concentrations or rely on

clinical monitoring.306

60

Changes in metabolism in pregnancy are mediated through changes in hepatic blood

flow, as well as the activity of metabolizing enzymes. Some evidence suggests that

there is an increase in hepatic blood flow in pregnancy.306 In particular, drugs with a

high extraction ratio have an increased .306 The various changes in action of

metabolising enzymes may be more significant than changes in blood flow. These

include changes in phase I, such as cytochrome P450 monooxygenase (CYP) enzymes,

and phase II metabolism, such as uridine 5′‐diphosphate glucuronosyltransferase

enzymes (UGT) and N‐acetyltransferase enzymes (NAT).306, 307 These are summarised in

Table 1‐8 which demonstrates the varied nature of these changes with either

increased, decreased or unchanged action of these enzymes.306 In general, these

changes do not appear to be consistent throughout pregnancy. In fact, there are

differences between the different stages of pregnancy. There is a period of a number

of weeks post‐partum before the activity of these enzymes returns to non‐pregnant

values.306, 307

Just as there are changes in hepatic , so too renal is altered in pregnancy. Renal

is not only determined by the GFR, but also on the active processes of tubular

secretion and reabsorption. Increased blood flow to the kidneys is thought to be

responsible for the higher GFR seen in all stages of pregnancy.306 Interestingly, in the

final six weeks of pregnancy the GFR was found to fall and was no different to non‐

pregnant values from three weeks before delivery.308 The variability seen in the

increased between renally excreted drugs is in part due to active tubular processes.

Table 1‐8 Changes of enzymes involved in metabolism during pregnancy, adapted from Anderson 2005307.

Enzyme Change in pregnancy Example drug substrates

CYP1A2 Decreased caffeine, clozapine

CYP2A6 Increased nicotine

CYP2C9 Increased phenytoin,

CYP2C19 Decreased proguanil, omeprazole

CYP2D6 Increased antidepressants, antiarrhythmics, chloroquine

CYP3A4 Increased >50% of all drugs including lumefantrine, artemether

UGT Increased lamotrigine (UGT1A4), zidovudine (UGT2B7)

NAT2 Nil isoniazid, hydralazine

61

Tubular secretion increases renal , while tubular reabsorption decreases it. These

processes may be differentially altered for different drugs in pregnancy.306

The number and complexity of the interactions between these changes makes

prediction of changes in drug disposition in pregnancy difficult.307 Ideally, a range of

women from different stages of pregnancy, including post‐partum, should be studied

to determine the changes that occur throughout pregnancy and in the weeks

afterwards. In reality, as is the case for the study presented in this thesis, it is more

common that only a small range of gestational ages are present in a study. This makes

defining a continuous relationship between model parameters and pregnancy difficult.

Often, only a categorical relationship characterising the difference between the three

stages of pregnancy with non‐pregnancy, or the postpartum period, is estimated.305, 307

1.3.4.2 91B91BChildhoodandinfancy

The main PK consideration in childhood is the large change in size over a small age

range.309, 310 Using the allometric model, parameters of and are scaled according

to WT, for example:

70

Equation 1‐18

where represents an individual, is the of individual , is the population

average of the for a 70 kg person, is the weight of the individual , and is

allometric coefficient. The effect of this scaling on CL, V and tR½R is presented as a solid

black line in Figure 1‐17.

Although the denominator (70 kg) does not influence model selection,310 a standard

value of 70 kg is chosen to allow easier comparison between studies, and helps

particularly when comparing studies performed in children with studies of adults.

Other size descriptors besides WT, such as fat free mass, ideal body weight, and body

mass index have been used to describe the PK changes associated with size.311 In a

group of obese subjects, changes in were best described using WT, while changes in

were best described using lean body mass.311 In a study population where there is

62

little difference in the body composition of the individuals, and extremes do not exist,

the choice of size descriptor is likely to be less important. Although the allometric

coefficients may be estimated (often with no significant difference with the standard

coefficients),309 the studies in this thesis were not intended, and therefore not

designed or powered, to estimate them accurately. Therefore, allometry, with

standard coefficients of 1 for volume terms and ¾ for clearance terms,309, 310 has been

used a priori in the models presented.

The second consideration after size in studies involving younger patients is that of

organ maturation.309, 310 This is particularly important in infants where the processes of

hepatic and renal maturation are still occurring to a significant extent. A few models

have been used to describe the process of this maturation.310 Of these, a sigmoid ERmaxR

model is the most plausible,309 and has been used to describe this process for a

number of drugs.310 This model allows for gradual maturation in infancy with the

achievement of ‘adult’ levels with increasing age. Using this model, the effect of

maturation on (or ) can then be combined with the effect of size to give:

70

¾

MATCL Equation 1‐19

where is the post menstrual age, is the Hill coefficient for , and

MATCL is the time to 50% maturation of . Similar maturation also takes place

with .309 An example of the changes of , and tR½R seen over a paediatric WT range

(assuming average WT for age312), and incorporating the effect of size and maturity,

are represented as a dashed red line in Figure 1‐17.

63

Weight (kg)

Vo

lum

e o

f d

istr

ibu

tio

n

Weight (kg)

Cle

ara

nc

e

Weight (kg)

Ha

lf-l

ife

Figure 1‐17 An example of the changes expected in volume, clearance and half‐life over weight and age using average weight for age data312. The solid black line represents changes when only considering allometry while the dashed red line considers both the effects of size and age.

In one study of propofol including infants and young children, 80% of the variability in

could be accounted for by size and maturation.310 This example from the literature

highlights the importance of these two covariates, size and age, in children and infants.

64

65

1.4 10B10BThesisoutline

As presented in this chapter, the literature surrounding the pharmacology of a number

of antimalarial compounds used for prevention in pregnancy and infancy and

treatment in childhood has limitations. This thesis presents a number of publications

that add to this literature. Although each chapter with original data in this thesis

relates to a discrete study, all chapters are concerned with the pharmacology of

antimalarial drugs in PNG. The thesis is therefore ordered to follow the different

groups who are at the greatest risk, from the mother who is made vulnerable by her

pregnancy to the infant who, after being affected in the womb, encounters malaria on

their own after birth, and finally the child who, while still developing their own innate

immunity to the infection, suffers from symptomatic malarial infections.

With respect to the prevention of malaria in pregnant women, a study of AZI is

presented in Chapter 2. Although better known for its use as an antibiotic, AZI shows

antimalarial activity like other antibiotics and it is also one of the few antimalarials

with proven safety in pregnancy. Prior to this work only one limited PK evaluation of

AZI in pregnancy had been published in the literature. As antimalarial activity is likely

to be related to the drug concentration in the blood, the optimal AZI dose regimen for

malaria prevention is likely to be different to that for its use as an antibiotic. In

addition, and since there may be a requirement for AZI to be used in combination for

malaria treatment, this study aimed to compare the PK of AZI in plasma of pregnant

women with non‐pregnant controls when given with CQ or SP, and provide preliminary

information on the safety and tolerability of these combinations.

A study of the prevention of malaria in infancy is presented in Chapter 3. This involved

standard and double dose SP in infants and supplements the worldwide investigation

of SP as IPTi. Given suggestions that the standard dose may not be sufficient in this

population and the consequent risk of poor performance of IPTi, there was a clear

need for a higher dose to be evaluated in this population. By employing a sparse blood

sample collection design, with only small volumes collected, this study aimed to

investigate the PK in this population in PNG and to assess the potential for the use of a

double dose with respect to safety and tolerability.

66

The next section of the thesis relates to the treatment of malaria in childhood and

contains three chapters. The methods used in all three of the studies is similar,

however the drugs investigated differ.

In Chapter 4 a publication of the pharmacology of AL, a widely used ACT, is presented.

Although widely used for the treatment of malaria in children, there are concerns

regarding the adequacy of the currently recommended dose regimen. Additionally,

few data exist regarding DBL, a potent active metabolite of LUM. The aims of this

publication were therefore to add to the current literature regarding the PK of the

components of this combination and to compare the PK in children to those in adults

in order to suggest an appropriate paediatric dose regimen.

Chapter 5 presents a publication in which two ACT combinations containing PQ are

compared. A newer combination containing PQ base (ART/PQ), not yet recommended

by the WHO, was investigated and compared with historical data of an older

combination containing PQ tetraphosphate (DHA/PQ), now recommended by the

WHO. Despite being commercially available, there is no published information on the

safety, tolerability or efficacy of this newer combination in children. This study,

therefore, not only aimed to provide a PK comparison of these two combinations but

also to provide preliminary data of the safety, tolerability and efficacy of ART/PQ in

children.

Another new ACT, ART/NQ, is the subject of the publication presented in Chapter 6.

Very few data exist for this combination, particularly in children, and therefore a study

was carried out in two parts. The first was a pilot where paediatric doses were

extrapolated from the adult doses recommended by the manufacturer, as no

recommendations for children were available at that time. The second used this

information to adjust the dose and compare different dose regimens. The aims of this

study were therefore to provide the first PK evaluation of ART/NQ in children and to

assess the impact of different dose regimens. (Safety, tolerability and efficacy data

from this study are presented separately in the literature).

67

Although a conclusion is provided for each chapter, a general conclusion is outlined in

Chapter 7. This summarises the main findings of the each of the chapters, provides

information of how some of this information has already been used, and suggests

some possible future avenues of research.

68

69

162B162BPREVENTIONOFMALARIA

INPREGNANTWOMEN

70

71

2 1B1BPharmacokineticPropertiesofAzithromycinin

Pregnancy

2.1 11B11BBackground

The primary aim of the study presented in this chapter is to compare the PK of AZI of

pregnant women with non‐pregnancy controls when given with CQ or SP. In addition it

aims to provide preliminary information on the safety and tolerability of these

combinations, particularly in pregnancy, as potential IPTp.

This study was performed at a time when an IPTp trial involving AZI was being planned

in PNG. The PNG health policy for managing malaria in pregnancy consisted of

treatment doses of CQ and SP, followed by weekly CQ. In light of increasing parasite

resistance to these drugs in PNG there was a need to investigate other potential

management options, such as AZI IPTp. Although some information existed to guide

the dose to be used in the trial, there was only sparse PK data for AZI in pregnancy and

little tolerability data for the planned regimen (2g daily for two days). Therefore this

study, based on a previous study performed at the same site,313, 314 was developed to

assist in finalising the dose regimen to be used in the AZI IPTp trial.

This study resulted in the publication3 presented in this chapter. Entitled,

“Pharmacokinetic properties of azithromycin in pregnancy” it was published in the

journal Antimicrobial Agents and Chemotherapy (2010. 54(1):p. 360‐6). The

contribution of each of the authors is outlined in section i, which also contains details

of ethical approvals and supporting funding. While the complete publication is

provided in section xi.a below, it has been reformatted to conform to thesis

requirements set by the University of Western Australia. The references have been

combined with those for the thesis as a whole and can be found in section x below.

72

73

2.2 12B12BPublication

Sam Salman,A Stephen J Rogerson,B Kay Kose,C Susan Griffin,C Servina Gomorai,C

Francesca Baiwog,C Josephine Winmai,C Josin Kandai,C Harin A Karunajeewa,A, D Sean J

O’Halloran,E Peter Siba,C Kenneth F Ilett,A,E Ivo Mueller,C Timothy M E DavisA.

ASchool of Medicine and Pharmacology, University of Western Australia, Perth,

Western Australia, Australia

BFaculty of Medicine, University of Melbourne, Melbourne, Australia

CPapua New Guinea Institute of Medical Research, Madang, Papua New Guinea

DWestern Health, Melbourne, Australia

EClinical Pharmacology and Toxicology Laboratory, Path West Laboratory Medicine,

Nedlands, Australia

2.2.1 38B38BAbstract

AZI is an azalide antibiotic with antimalarial activity that is considered safe in

pregnancy. To assess its PK properties when administered as IPTp, two 2 g doses were

given 24 h apart to 31 pregnant and age‐matched 29 non‐pregnant Papua New

Guinean women. All subjects also received single‐dose SP (1500mg/75mg) or CQ (450

mg base daily for three days). Blood samples were taken at 0, 1, 2, 3, 6, 12, 24, 32, 40,

48 and 72 h and then on days 4, 5, 7, 10 and 14 for AZI assay by ultra high‐performance

liquid chromatography‐tandem mass spectrometry (UPLC‐LC‐MS/MS). The treatments

were well tolerated. Using population PK modelling, a three‐compartment model with

a zero‐ followed by first‐order absorption and no lag time provided the best fit. The

AUCR0–∞R (28.7 and 31.8 mg∙h/l for pregnant and non‐pregnant subjects, respectively)

was consistent with results of previous studies, but the estimated terminal elimination

half‐lives (78 and 77 h, respectively) were generally longer. Among a range of potential

covariates including malarial parasitaemia, the only significant relationship identified

was for pregnancy which accounted for an 86% increase in the V of the central

compartment but there was no significant change in AUCR0‐∞R. These data suggest that

AZI can be combined with longer tR½R compounds such as SP in combination IPTp

without the need for dose adjustment.

74


AZI is a semi‐synthetic azalide antibiotic that is structurally related to erythromycin but

which has a broader spectrum of antibacterial activity and more favourable PK

profile.262, 315 It is widely used in the treatment of respiratory and sexually‐transmitted

infections, including those in HIV‐infected patients.243, 244 AZI also inhibits protein

synthesis in the plasmodial apicoplast145, 316 and thus has activity against both

Plasmodium falciparum and vivax.240, 242, 317‐322 Its greatest effect is against the progeny

of parasites that inherit a non‐functioning apicoplast after exposure, with the result

that its antimalarial effect is of slow onset and relatively weak. Therefore, AZI is best

used in combination with other antimalarial compounds as both treatment249, 318, 320

and chemoprophylaxis,240, 323 with likely additive or synergistic effects.246, 319, 321

Malaria in pregnancy can result in adverse outcomes for both mother and foetus.28

Intermittent presumptive treatment in pregnancy (IPTp) aims to reduce the burden of

malaria by administering treatment doses of antimalarial drugs at predetermined

intervals as part of routine antenatal care in endemic areas.324 Because AZI is

considered safe in pregnancy and could have activity against other clinically‐significant

pathogens,236, 325 it has been suggested as a candidate for IPTp. Although the PK of AZI

have been investigated previously,256‐262, 265, 326‐330 only one study included pregnant

women330 and most focused on its antibacterial properties. In addition, AZI is likely to

be partnered with conventional antimalarial drugs if given as IPTp, and there is

evidence that such combinations are safe and well tolerated in studies with CQ in

healthy volunteers327 and with SP in pregnant women.249 Although there does not

appear to be a clinically significant PK interaction with CQ,327 AZI interactions with

other conventional IPTp treatments are unknown. We have, therefore, investigated

the PK properties of AZI in combination with CQ or SP in pregnant and non‐pregnant

women from an area of PNG with intense transmission of both falciparum and vivax

malaria.

75

2.2.3 40B40BPatientsandmethods

2.2.3.1 92B92BStudysite,sampleandapprovals

The present study was conducted at Alexishafen Health Centre, Madang Province on

the north coast of PNG. The pregnant women were recruited at their first antenatal

clinic visit and the age‐matched non‐pregnant volunteers from the same communities

as the pregnant participants. Women were eligible if i) they had not taken any of the

study drugs in the previous 28 days, ii) they had no history of significant allergy to any

study drug, iii) there was no significant co‐morbidity or clinical evidence of severe

malaria, and v) follow‐up was possible for the duration of the study. The study was

approved by the Medical Research Advisory Committee of PNG (reference 07.24) and

the Human Ethics Research Committee at the University of Western Australia

(reference RA/4/1/1871). Written informed consent was obtained from all

participants.

2.2.3.2 93B93BClinicalprocedures

A detailed assessment was performed prior to drug administration including a side‐

effects questionnaire, point‐of‐care haemoglobin and blood glucose (HemoCue®,

Angelholm, Sweden), thick and thin blood films, and (for pregnant participants)

estimation of gestational age by fundal height. A 3 ml blood sample was taken for

subsequent antimalarial drug assay. All women received 2 g AZI (Zithromax®, Pfizer,

New York) both at enrolment and 24 h later. Subjects were also randomised to receive

single‐dose SP (1500mg/75mg; Fansidar®, Roche, Basel, Switzerland) at enrolment

(AZI‐SP arm) or CQ (Chloroquin®, Astra, Sydney, Australia) 450 mg base daily for three

days (AZI‐CQ arm) in accordance with regimens recommended for PNG.331

Administration of all doses was directly observed. The dosing schedule for AZI was

chosen as the simplest regimen that would be likely to ensure effective drug

concentrations during the first 4 days of treatment.316

Following the first dose of AZI (day 0), additional blood samples were taken at 1, 2, 3,

6, 12, 24, 32, 40, 48 and 72 h and then on days 4, 5, 7, 10 and 14 for drug assay. The

exact timing of each blood sample was recorded. All samples were centrifuged

promptly with RBCs and separated plasma stored frozen at ‐80oC. The side‐effects

76

questionnaire was re‐administered at 6 h and then at 1, 2, 3 and 7 days. Haemoglobin,

erect and supine heart rate and blood pressure, respiratory rate, temperature and

blood slides were taken on days 1, 2, 3, 7, 14, 28 and 42, and blood glucose was

measured on days 1, 2 and 3. After completion of follow‐up, pregnant patients were

returned to usual antenatal care.

2.2.3.3 94B94BLaboratoryMethods

Giemsa‐stained thick blood smears were examined independently by at least two

skilled microscopists who were blinded to pregnancy and treatment status. Each

microscopist viewed >100 fields at 1,000 x magnification before a slide was considered

negative. Any slide discrepant for positivity/negativity or speciation was referred to a

third microscopist.

AZI concentrations were measured using a validated UPLC‐LC‐MS/MS method using a

deuterated internal standard. The samples have been retained for subsequent SP

assay. AZI USP was obtained from APAC Pharmaceutical LLC (Ellicott City, MD, USA)

and dR3R‐AZI from Toronto Research Chemicals (North York, Canada). In brief, following

addition of internal standard, AZI was extracted from 5 microliters (µl) of plasma by

protein precipitation. After centrifugation, supernatant (5 µl) was injected onto a

2795/Quattro Premier XE UPLC‐ESI‐MS/MS (Waters Corp, MA, USA) using a Waters

BEH CR18R 1.7m, 2.1 x 100mm column. Gradient elution was performed using mobile

phases A (45/55 v/v comprising 1 g/l ammonium bicarbonate in 50/50 v/v

methanol:water and acetonitrile) and B (50/50 v/v methanol/acetonitrile) at 0.4

ml/min. Adduct transitions were monitored using positive electrospray ionization with

multiple reaction monitoring for AZI and dR3R‐AZI were m/z 749.6‐591.4 and m/z 752.6‐

594.4, respectively. The method was linear to 1012 nanogram (ng)/ml (r2>0.9997) with

a limit of quantification of 2.5 g/l AZI. All inter‐ and intraday coefficients of variation

were <10% and between‐patient variability was <5% when matrix effects were

investigated at three concentrations.

2.2.3.4 95B95BPopulationpharmacokineticanalysis

Concentration‐time datasets were analysed by nonlinear mixed effect modelling using

NONMEM (version 6.2.0, ICON Development Solutions, Ellicott City, MD, USA) with an

77

Intel Visual FORTRAN 10.0 compiler. Linear mammillary model subroutines within

NONMEM (ADVAN4 and 12 used with TRANS4 in the PREDPP library), FOCE with ‐

interaction, and the OFV were used to construct and compare plausible models. Unless

otherwise specified, a difference in OFV of ≥6.63 (2 distribution with 1 d.f., P<0.01)

was considered significant. The R‐based model‐building aid Xpose 6.0 was used for

graphic model diagnosis 332. Secondary PK parameters including volume of distribution

at steady‐state (VRSSR, equal to the sum of the V of all compartments), AUCR0–∞Rs and

elimination tR1/2Rs for the non‐pregnant and pregnant groups were obtained from post

hoc Bayesian prediction in NONMEM using the final model parameters. Macro

constants for the three compartment model were calculated from the modelled

parameters using previously published equations. 333

All volume terms were allometrically scaled with (*(WT/70)1.0) and all CL terms with

(*(WT/70)0.75).309 All V and CL parameters were relative to bioavailability (/F).

Two and three compartment models were compared and then zero and first order

absorption models, with and without a lag time, were assessed alone and in

combination. IIV was added to parameters for which it could be estimated reasonably

from available data. For RUV, both exponential (proportional) and combined

(exponential with additive) error models were tested. In the development of the final

models, we investigated the influence of the covariates pregnancy, dosing group,

fundal height, gestational age, malaria status, initial Hb and blood glucose on model

parameters using Xpose and the generalized additive modelling procedure function.

Relationships between these covariates and individual PK parameters were also

explored by inspection of correlation plots. Covariate relationships identified by this

procedure were evaluated within the NONMEM model. Inclusion of the covariate

required a decrease of ≥3.84 in OFV (2 d.f.=2, P<0.05) and a decrease in the IIV.

Correlations among IIV terms and WRES plots were used in model evaluation.

A bootstrap procedure using Perl speaks NONMEM (PSN) was used to sample

individuals from the original dataset with replacement and generate 1000 new

datasets that were subsequently analysed using NONMEM. The resulting parameters

were then summarized as median and 2.5th and 97.5th percentiles (95% empirical CI) to

78

facilitate validation of the final model parameter estimates. In addition, a stratified

VPC was also performed using PSN with 1000 replicate datasets simulated from the

original. The resulting 80% PI for AZI were plotted with the observed data to assess the

predictive performance of the model.

2.2.3.5 96B96BStatisticalanalysis

SigmaStat® (version 3.10, Systat Software Inc, Chicago, IL, USA) was used for statistical

analysis unless otherwise specified. Data are summarized as mean ± standard

deviation (SD) or median and inter‐quartile range (IQR) as appropriate. Student’s t‐

test or the Mann‐Whitney U test was used for two‐sample comparisons. Categorical

data were compared using either Pearson Chi squared or Fisher’s exact test, and

multiple means by repeated measures ANOVA. A two‐tailed level of significance of

0.05 was used. Drug concentrations at each time point after day 2 were compared to

AUCR0–∞R using Pearson correlation.

79

2.2.4 41B41BResults

2.2.4.1 97B97BPatientcharacteristics

A total of 31 pregnant and 29 non‐pregnant women were recruited between October

2007 and March 2008. All subjects took two AZI doses but two pregnant patients did

not receive either CQ or SP. These women were excluded from initial analyses but

were included subsequently if there was no effect of CQ or SP on AZI PK properties in

the other subjects. Baseline characteristics of the subjects by pregnancy status and

treatment allocation are shown in Table 2‐1. The groups were well matched except

that, consistent with normal physiological changes that occur in pregnancy,307, 334 the

pregnant subjects were significantly heavier and had a lower haemoglobin than the

non‐pregnant subjects for each treatment group (P<0.05). Seven of the pregnant

patients were parasitaemic at baseline compared with only one of the non‐pregnant

subjects (P=0.02).

Table 2‐1 Baseline characteristics of the study participants by pregnancy status and treatment allocation. Data are mean ± SD, median [IQR] or number (%).

Pregnant Non‐pregnant

AZI‐CQ

(n=15)

AZI‐SP

(n=14)

AZI‐CQ

(n=14)

AZI‐SP

(n=15)

Age (years) 26.9 ± 4.1 23.9 ± 5.1 25.7 ± 5.8 27 ± 6.5

Weight (kg) 53.5 ± 7.1a 56.4 ± 7.9a 51.4 ± 5.4 51.9 ± 4.9

Height (cm) 154 ± 7.4 154 ± 7.3 154 ± 6.4 154 ± 2.8

Axillary temperature (°C) 36.4 ± 0.7 36.5 ± 0.6 36.7 ± 0.3 36.4 ± 0.3

P. falciparum parasitaemia 3 (20) 3 (21) 1 (7) 0 (0)

P. vivax parasitaemia 1 (7) 0 (0) 0 (0) 1 (7)

Gestational age (weeks) 24 [22‐27] 21 [19‐24]

Gravidity 3 [2‐5] 2 [1‐4] 1 [0‐3] 2 [0‐3]

Parity 2 [1‐4] 1 [0‐2] 0 [0‐3] 1 [0‐3]

Respiratory rate (/min) 20 ± 1 22 ± 5 20 ± 2 20 ± 1

Supine pulse rate (/min) 91 ± 10 89 ± 7 82 ± 10 88 ± 7

Supine MAP (mm Hg)b 78 ± 7 81 ± 10 79 ± 9 82 ± 7

Haemoglobin (g/dl) 8.5 ± 1.6a 8.2 ± 1.2a 9.3 ± 1.9 10 ± 1.3

Blood glucose (mmol/l) 5.9 ± 1.6 5.7 ± 0.8 6.2 ± 1.1 5.6 ± 2.7 a P<0.05 vs. non‐pregnant subjects; b mean arterial pressure, calculated by adding 1/3 of the pulse pressure (systolic minus diastolic pressure) to the diastolic pressure

80

2.2.4.2 98B98BEfficacy,tolerabilityandsafety

Three of the seven falciparum and one of the two vivax cases at baseline received AZI‐

SP. There was an uncorrected adequate parasitological and clinical response (APCR) of

100% for both treatments. A further eight cases (five of whom were pregnant) became

slide‐positive for P. falciparum and three (two who were pregnant) for P. vivax late in

the 42 day follow‐up period. All received recommended antimalarial therapy.331 All

cases at baseline and during follow‐up were asymptomatic.

Both treatments were well tolerated. Table 2‐2 summarizes the self‐reported side‐

effects in the first week of follow‐up, all of which were mild (defined as not bad

enough to interfere with daily activity) and generally short‐lived (≤2 days). Six patients

reported mild symptoms before drug administration (headache, abdominal pain,

pruritus or dizziness) but these resolved subsequently. Pruritus was only reported in

the AZI‐CQ group (P=0.052). Although not formally assessed, no significant side‐effects

were volunteered at assessments beyond day 7. No patient developed hypoglycaemia

(blood glucose <2.5 mmol/l) or severe anaemia (haemoglobin <5.0 g/decilitre [dl])

Table 2‐2 Side‐effects reported during the first week after initiation of treatment. Data are numbers of patients and (%).

AZI‐CQ

(n=29)

AZI‐SP

(n=29)

Fever 2 (7) 1 (3)

Chills 2 (7) 0 (0)

Headache 6 (21) 4 (14)

Nausea 4 (14) 7 (24)

Vomiting 2 (7) 4 (14)

Diarrhoea 2 (7) 2 (7)

Abdominal pain 4 (14) 3 (10)

Rash 0 (0) 0 (0)

Pruritus 5 (17) 0 (0)

Anorexia 1 (3) 2 (7)

Insomnia 2 (7) 0 (0)

Dizziness 3 (10) 1 (3)

Bone or joint pain 1 (3) 1 (3)

Othera 5 (17) 1 (3) a Cough (2), blocked ear (1), ‘heavy head’ (1), and numbness of calf muscles (1) in the AZI‐CQ group and cough (1) in the AZI‐SP group.

81

during follow‐up. Although postural hypotension (>20 mmHg systolic or >10 mmHg

diastolic fall after standing) occurred 8 times in 7 pregnant (4 from the AZI‐CQ group)

and 7 times in 5 non‐pregnant patients (all 5 in the AZI‐CQ group), differences between

groups were not significant and there were no associated symptoms. After completion

of the study, one of the study participants had a still birth. A medical review of her

case notes by three independent physicians concluded that it was unlikely to be the

result of study medication.

2.2.4.3 99B99BPharmacokineticmodelling

A three compartment model had a lower OFV value than a two compartment model

(8700.058 vs. 8185.104; P<0.001 by 2 test, d.f.=2) and a more favourable distribution

of WRES over time. A zero‐ followed by first‐order absorption without a lag time

provided the lowest OFV and best fit for AZI absorption. The fixed model parameters

Table 2‐3 Model building, final parameter estimates and bootstrap results from the AZI population pharmacokinetic modelling.

Parameter Base model

estimate (%RSE)

Final model

estimate (%RSE)

Bootstrap (n=1000) median [95% CI]

Structural and covariate model parameters

DUR (h) 1.66 (10.4) 1.55 (3.3) 1.56 [1.21‐2.01]

kRa R(/h) 0.513 (3.2) 0.525 (14.8) 0.524 [0.451‐0.623]

VRCR/F (l) 504 (13.9) 384 (17.6) 371 [235‐554]

Pregnancy on VRCR (l) 330 (69.4) 318 [48‐604]

CL/F (l/h) 158 (3.9) 158 (6.7) 158 [145‐171]

VRP1R/FR R(l) 4080 (8.6) 4080 (12.5) 4045 [3402‐4870]

QR1R/F (l/h) 327 (5.7) 325 (12.7) 326 [288‐368]

VRP2R/FR R(l) 5070 (5.6) 5040 (7.3) 5070 [4262‐5730]

QR2R/F (l/h) 67.2 (11.5) 66.4 (12.4) 67.5 [48.5‐84.0]

Random model parameters

IIV VRcR/F 111.4 (20.9) 99.6 (35.5) 99.0 [72.6‐127.6]

IIV CL/F 28.3 (24.1) 28.3 (33.1) 27.9 [21.6‐34.5]

IIV VRP1R/F 35.8 (27.0) 35.6 (27.2) 34.8 [25.7‐45.2]

IIV DUR 73.0 (21.4) 76.9 (22) 75.5 [55.4‐95.6]

RUV (%) 31.3 (9.5) 31.2 (15.1) 30.9 [28.1‐33.8]

OFV in base model: 7999.870, OFV in final model: 7993.646, bootstrap OFV (median [95% CI]) 7974.699 [7756.673‐8201.238]

82

were DUR (the duration of the zero order absorption), kRaR, CL/F, VRCR/F, VRP1R/F, VRP2R/F,

QR1R/F and QR2R/F. The model structure is shown in Figure 2‐1. IIV could be estimated on

DUR, CL/F, VRcR/F and VRP1R/F while a proportional error model was best for RUV. After

Figure 2‐1 Structural model used in the final pharmacokinetic analysis of plasma azithromycin concentrations in the central compartment versus time.

Time (h)

1 10 100 1000

We

igh

ted

re

sid

ua

ls (

azi

thro

my

cin

)

-6

-4

-2

0

2

4

6

Predicted plasma azithromycin (g/l)

10 100 1000

Ob

se

rve

d p

las

ma

azi

thro

my

cin

(

g/l)

10

100

1000

A

B

Figure 2‐2(A) Population (○) and individual (●) predicted versus observed plasma azithromycin concentra ons (µg/l on log10 scale) for the final model. The line of identity is also shown. (B) Weighted residuals vs. time (log scale) for azithromycin final model.

83

testing the various covariates, only pregnancy on VRCR/F produced a significant decrease

in the

OFV (2 d.f.=1, P<0.05) accompanied by a decrease in the IIV of VRCR/F from 111.0 % to

99.6 %.

The results of the parameter estimates and their relative standard errors (RSE) are

summarized in Table 2‐3 and secondary parameter estimates in Table 2‐4. All drug

concentrations after day 2 were strongly correlated with the AUCR0–∞R (r>0.7; P<0.001,)

with 96 h concentrations showing the strongest association (r=0.78). The bootstrap

results (see Table 2‐3) demonstrate robust estimation of both fixed and random

parameters with bias <4% and <5%, respectively. GOF plots of observed versus

population and individual predicted concentration, and WRES versus time are shown in

Figure 2‐2.The VPC results, stratified for pregnancy status, are presented in Figure 2‐3

and show reasonable predictive performance of the model, while demonstrating some

difficulty in capturing post‐absorption plasma concentrations peaks.

Table 2‐4 Secondary pharmacokinetic parameters derived from post hoc Bayesian estimates for pregnant and non‐pregnant study participants (median [IQR]).

Parameter Pregnant (n=31) Non‐pregnant (n=29) P‐value

DUR (h) 1.65 [0.94 – 2.34] 1.75 [1.02 – 2.38] NS a

kRaR (/h) 0.525 [0.525 – 0.525] 0.525 [0.525 – 0.525] NS

VRCR/F (l) 647 [422 – 995] 249 [157 – 363] < 0.001

VRP1R/F (l) 3,620 [2,747 – 3,951] 2,909 [2,296 – 3,586] NS

VRP2R/F (l) 3,888 [3,708 – 4,104] 3,672 [3,456 – 3,888] 0.034

VRSSR/F (l) 8,355 [7,460 – 8,973] 6,875 [6,115 – 7,526] 0.002

tR½αR

b (h) 0.88 [0.57 – 1.36] 0.39 [0.24 – 0.56] < 0.001

tR½βR

b (h) 20.7 [18.3 – 22.8] 18.8 [15.3 – 21] NS

tR½γR

b (h) 78.2 [74 – 82.5] 77.1 [71.5 – 84.5] NS

AUCR0–∞R (g19T∙19Th/l) 28,713 [25,913 – 32,942] 31,781 [28,736 – 38,012] NS a NS, not significant, b tR½αR, tR½βR and tR½γR are the first distribution, second distribution and terminal elimination half‐lives respectively.

84

2.2.5 42B42BDiscussion

The present study is the first PK evaluation of AZI in pregnant and non‐pregnant

women living in a malaria‐endemic area. We found that a three compartment model

with a combined absorption process best described the disposition of AZI in our

subjects. Both two256, 257, 259 and three compartment258, 260 models have been found to

best describe AZI plasma concentration‐time profiles in other contexts. Our ability to

differentiate the tri‐exponential elimination of AZI may have been facilitated by the

relatively long sampling duration. This may also explain why our estimated terminal

elimination tR½R (78 and 77 h for pregnant and non‐pregnant participants, respectively)

was longer than in most previous studies (range 27‐79 h).256‐258, 260‐262 Overall drug

exposure (AUCR0–∞R 28.7 and 31.8 mg∙h/l for pregnant and non‐pregnant subjects,

respectively) was within the range expected from dose‐scaled results from previous

studies in other contexts (26.5‐46.4 mg∙h/l),256, 257, 259‐262, 265, 326, 327 suggesting that the

bioavailability of AZI is not dose‐dependent.

Both zero256, 257 and first258, 259 order absorption have been reported previously for AZI

but neither was appropriate for our data. A combined absorption process where the

drug enters the absorption compartment in a zero order manner and then is absorbed

according to first order kinetics provided the best model in the present study. This is

analogous to the twin processes of i) gastric emptying of the drug into the small

Time (h)

1 4 10 40 100 400

AZ

I (

g/L

)

1

10

100

1000

10000

Time (h)

1 4 10 40 100 400

AZ

I (

g/L

)

1

10

100

1000

10000A B

Figure 2‐3 Visual predicted check plots showing simulated 10th (short dashed line), 50th (dotted line) and 90th (solid line) percentile concentrations and observed concentration (log scale) data (grey open circles) versus time (log scale) for non‐pregnant (A) and pregnant (B) participants.

85

intestine (the zero order process) and ii) absorption in the small intestine proportional

to the amount present (the first order process). Despite this more complex model, AZI

absorption was still not well characterized in our final model. This has been reported

previously257 but is unlikely to be significant in the treatment of uncomplicated malaria

where exposure of the parasite to therapeutic drug concentrations over several life‐

cycles is more important than that immediately after drug administration.

Plasma AZI concentrations appeared to differ between pregnant and non‐pregnant

women only in the first 48 h after the first dose. This was confirmed by the population

PK modelling in which pregnancy, the only significant covariate relationship, accounted

for an 86% increase in VRCR/F. Despite significant differences in the secondary PK

parameters VRCR/F, VRP2R/F, VRSSR/F and tR½αR between pregnant and non‐pregnant subjects,

no difference was seen in tR½γR or AUCR0–∞R. This suggests that drug elimination and

overall exposure are similar in the two groups. A much shorter AZI tR½R (12 h) than in the

present study has been reported previously in pregnant women,330 but this study

employed a shorter sampling duration (168 vs. 336 h), included pregnant women at or

near term, and the analysis was constrained by relatively sparse sampling.

Because of the need for AZI to be combined with other therapies,317, 322 we included

conventional antimalarial drugs currently recommended as part of IPTp in PNG and

other countries.249, 250 There were no significant differences in the disposition of AZI

between the AZI‐CQ and AZI‐SP groups, consistent with a study of the interaction

between CQ and AZI in healthy volunteers.327 We conclude that AZI dose modification

is unnecessary in these combinations. In addition, the lack of an effect of malaria

status as a covariate on AZI disposition suggests that, unlike drugs such as QN,335 the

dose may not have to be adjusted when parasitaemia is present.

The most common side‐effects of AZI, especially with higher doses, are nausea and

vomiting. These symptoms are thought to relate to effect of AZI on the motilin

receptor in the upper gastrointestinal tract.239 However, with the exception of pruritus

which tended to be associated with AZI‐CQ therapy consistent with known CQ

effects,336 there were no differences in the incidence of side‐effects between the two

treatment groups and all reported adverse effects were mild. The AZI dose regimen in

86

both combination therapy groups in the present study (2.0 g daily for two days) was

associated with a similar side‐effect profile to that reported previously after a single

2.0 g dose.260 Use of the sustained release formulation of AZI should reduce side‐

effects including nausea and vomiting.337 However, this formulation has a

bioavailability of 82.8% relative to conventional AZI, suggesting that a higher dose will

be required to achieve the same drug exposure. As well as increasing the cost of AZI

treatment, this could mean that side‐effects are more frequent with higher‐dose

sustained release AZI administration.

Although the present study had limited subject numbers, it is encouraging that both

regimens achieved a 100% uncorrected APCR. The plasma concentrations of AZI

required to achieve cure are unknown as no efficacy trials have included these data.

However, the high correlation between 96 h drug concentrations and AUCR0–∞R in our

patients suggests that a day 4 plasma concentration could be an appropriate surrogate

for overall AZI exposure in efficacy trials in which serial blood sampling is problematic.

It is of interest that prolongation of the in vitro exposure of P. falciparum to 96 h

results in substantially increased potency, suggesting that either AZI renders second‐

generation parasites unable to establish a parasitophorous vacuole upon host cell

invasion or the effect on apicoplast protein synthesis inhibits successful development

of the progeny of drug‐treated parasites.316

Given the need for relatively prolonged parasite exposure to therapeutic plasma

concentrations, it is unlikely that the benefit of “front loading” of AZI used in treating

bacterial infections256, 265 will be relevant in malaria. However, experience with AZI as

an antimalarial agent is growing. A Cochrane review of its efficacy is currently

underway338 and promising results are being seen when used with SP in IPTp, such as

might be given at least twice during pregnancy.249 The present study provides a PK

foundation for the further investigation of AZI as an antimalarial agent in pregnancy,

particularly in combination IPTp. Further data from the present study should also

determine whether AZI influences the disposition of CQ and SP. Although there was a

significant increase in AZI VRCR/F in pregnant women, there was no significant change in

AUCR0‐∞R, and it is therefore likely that no dose adjustments will be required for

pregnant women when AZI is given in combination with CQ or SP.

87

2.2.6 43B43BAcknowledgements

We are most grateful to Sr Valsi Kurian and the staff of Alexishafen Health Centre for

their kind co‐operation during the study. We also thank Christine Kalopo and Bernard

(“Ben”) Maamu for clinical and/or logistic assistance. The study was funded by the

National Health and Medical Research Council (NHMRC) of Australia (grant 458555).

TMED is supported by an NHMRC Practitioner Fellowship.

88

89

163B163BPREVENTIONOFMALARIA

ININFANTS

90

91

3 2B2BPharmacokineticPropertiesofConventional

andDouble‐DoseSulfadoxine‐Pyrimethamine

GivenasIntermittentPreventiveTreatmentin

Infancy


The study in this chapter aims to investigate the PK of conventional and double dose

SP in infants in PNG as well as to assess the safety and tolerability of a double dose as a

potential IPTi.

At the time this study was conceived a trial of SP IPTi was ongoing in the same area.

There is some evidence that the PK of SDX and PYR are different in younger children

when compared to older children or adults. In particular, there was suggestion that

these younger children may be under‐dosed. The role of this PK study was to

investigate the potential use of a higher SP dose for IPTi and to aid in the

interpretation of the results of the SP IPTi efficacy trial.

This study resulted in the publication1 presented in this chapter. Entitled,

“Pharmacokinetic properties of conventional and double‐dose sulfadoxine‐

pyrimethamine given as intermittent preventive treatment in infancy” it was published

in the journal Antimicrobial Agents and Chemotherapy (2011. 55(4):p. 1693‐700). The


of ethical approvals and supporting funding. While the complete publication is

provided in section xi.a below, it has been reformatted to conform to thesis

requirements set by the University of Western Australia. The references have been

combined with those for the thesis as a whole and can be found in section x below.

92

93


Sam Salman,A Susan Griffin,B Kay Kose,C Nolene Pitus,B Josephine Winmai,B Brioni

Moore,A Peter Siba,B Kenneth F Ilett,A,C Ivo Mueller,B Timothy M E DavisA.

ASchool of Medicine and Pharmacology, University of Western Australia, Perth,

Western Australia, Australia

BPapua New Guinea Institute of Medical Research, Madang, Papua New Guinea

CClinical Pharmacology and Toxicology Laboratory, Path West Laboratory Medicine,

Nedlands, Australia


IPTi entails routine administration of antimalarial treatment doses at specified times in

at‐risk infants. SP is a combination that has been used as first‐line IPTi. Because of

limited PK data and suggestions that higher mg/kg paediatric doses than

recommended should be considered, we assessed SP disposition in 70 Papua New

Guinean children aged 2‐13 months randomized to conventional (25/1.25 mg/kg) or

double (50/2.5 mg/kg) dose. Blood samples were drawn at baseline, 28 days and three

time‐points randomly selected for each infant at 4‐8 h, or 2, 5, 7, 14 or 21 d. Plasma

SDX, PYR and NR4R‐acetylsulfadoxine (NSX, principal metabolite of SDX) were assayed by

high‐performance liquid chromatography (HPLC). Using population modelling

incorporating hepatic maturation and cystatin C (CysC)‐based renal function, two‐

compartment models provided best fits for PYR and SDX/NSX plasma concentration

profiles. AUCR0–∞R was greater with double vs. conventional dose for PYR (4,915 vs.

2,844 μg∙d/l) and SDX (2,434 vs. 1,460 mg∙d/l). There was a 32% reduction in SDX

relative bioavailability with double‐dose but no evidence of dose‐dependent

metabolism. Terminal elimination half‐lives (15.6 days for PYR, 9.1 days for SDX) were

longer than previously reported. Both doses were well tolerated without changes in Hb

or hepatorenal function. Five children in the conventional and three in the double‐

dose group developed malaria during follow‐up. These data support the potential use

of double‐dose SP in infancy but further studies should examine the influence of

hepatorenal maturation in very young infants.

94


IPTi is a strategy in which infants in malaria‐endemic areas are given treatment doses

of antimalarial drugs at specified times, regardless of clinical and parasitological status.

Because of its availability, tolerability and relatively low cost, SP has been used first‐

line in IPTi programs, especially in Africa. A recent review of safety and efficacy data

from six trials conducted from 1999 to 2007 revealed that, despite the emergence of

molecular markers of parasite resistance, SP IPTi reduced clinical malaria and malaria‐

related hospital admissions by about one‐third, and anaemia in the first year of life by

15%.339 The duration of effective antimalarial prophylaxis after a dose of SP is 4‐6

weeks.62, 340

There is evidence that the efficacy of SP IPTi is dose‐dependent. When given as a fixed

dose,341 efficacy declines with age as lower mg/kg doses are taken.62 In addition,

studies of older children aged 2‐5 years with falciparum malaria have found higher

CL/F and larger V/F for both SDX and PYR than those in adults.226 Consistent with these

data, a population PK study in children with congenital toxoplasmosis showed that the

elimination half‐lives for both drugs were directly related to WT, with the consequence

that younger, and thus lighter children, had more rapid elimination.227 These studies

suggest that the peak plasma concentration and AUC will be reduced in younger

children and that currently recommended doses of SP of 25 mg/kg and 1.25 mg/kg,

respectively, may be inadequate for full efficacy. Indeed, there is evidence that higher

blood PYR concentrations enhance the ability of paediatric patients to clear resistant

Plasmodium falciparum.342

In view of these data and calls for doubling of the recommended treatment dose in

children aged 2‐5 years,226 we assessed the tolerability, safety and PK properties of SP

given in recommended and double recommended doses to infants living an area of

intense malaria transmission in PNG.

95


3.2.3.1 100B100BStudysite,sampleandapprovals

The present study was conducted at Alexishafen Health Centre, Madang Province on

the north coast of PNG. Infants between the ages of 2 and 13 months from the

surrounding area were eligible for recruitment provided that they i) did not have

features of severe malaria or significant non‐malarial illness, ii) had not been treated

with SDX or PYR in the previous four weeks, iii) did not have a known allergy to either

SDX or PYR, and iv) were available for assessment for the duration of follow‐up.

Written informed consent was obtained from the parents/guardians of all recruited

infants. The study was approved by the Medical Research Advisory Committee of PNG

and the Human Ethics Research Committee at the University of Western Australia.

3.2.3.2 101B101BClinicalprocedures

At enrolment, a clinical assessment was performed that included a standard baseline

symptom questionnaire completed by parents/guardians. A 500 µl finger prick capillary

blood sample was taken for preparation of blood smears for microscopy, baseline drug

assay, biochemical tests and Hb concentration (HemoCue®, Angelholm, Sweden).

Subjects were randomized to receive either the recommended dose of SP (25/1.25

mg/kg Fansidar®, Roche, Basel, Switzerland) or a double dose (50/2.5 mg/kg). Table

3‐1 shows the dose administered based on WT. All dosing was directly observed with

subsequent monitoring and re‐administration of the dose if the infant vomited within

30 minutes. Infants with a positive blood film were also given a three‐day course of AQ

according to PNG national treatment guidelines. 343 All drugs were crushed and mixed

with either water or breast milk before administration by mouth using a syringe.

All infants were re‐assessed on days 1, 2, 3, 5, 7, 14, 21 and 28. A Hb concentration

was determined on each occasion and a repeat symptom questionnaire administered

Table 3‐1 Dosing guide for conventional and double‐dose groups with the SDX/PYR doses in mg given in parentheses.

Body weight Conventional dose Double dose

3‐5.9 kg ¼ tablet (125/6.25) ½ tablet (250/12.5)

6‐11.9 kg ½ tablet (250/12.5) 1 tablet (500/25)

96

at each visit up to day 7. Blood films were repeated on day 28 and/or when fever or a

recent history of a fever was reported. For PK analysis, four additional 500 µl capillary

blood samples were taken from each infant. The first three of these were randomly

selected for each infant from either 4‐8 h or 2, 5, 7, 14 or 21 days post‐dose. A final

sample was taken in all cases on day 28. The exact timing of each blood sample was

recorded. All samples were centrifuged promptly with RBCs and separated plasma

stored frozen at ‐80oC until assay.

3.2.3.3 102B102BLaboratoryMethods

Giemsa‐stained thick blood smears were examined independently by at least two

skilled microscopists who were blinded to dose group. Each microscopist viewed >100

fields at 1,000 x magnification before a slide was considered negative. Any slide

discrepant for positivity/negativity, or speciation was referred to a third microscopist.

CysC concentrations were measured by particle enhanced immunoturbidimetry

(PETIA) using the Tina‐quant CysC kit run on an Elecsys 2010 analyser (Roche,

Indianapolis, USA). Sodium, urea, creatinine, albumin, γ‐glutamyl transferase and

bilirubin were measured using an Integra 800 analyser (Roche) when sufficient plasma

was available.

SDX, sulfamethazine and PYR were obtained from Sigma‐Aldrich (Castle Hill, Australia)

and midazolam hydrochloride from Pfizer (West Ryde, Australia). NSX was synthesized

according to the method of Whelpton et al.344 and found to have a melting point of

230oC and >99.9% purity by HPLC. Acetonitrile was obtained from Merck (Damstadt,

Germany). All other chemicals were of analytical or HPLC grade.

For PYR, SDX and NSX, extraction and separation were performed based on previously

published HPLC‐ultraviolet (UV) methods.227, 314 The internal standards were

midazolam HCl for PYR and sulfamethazine for SDX and NSX. Analytes were assayed

using UV detection at 270 nm. Chemstation Software (version 9, Agilent Technology,

Waldbronn, Germany) was used for analysis of chromatograms. Standard curves were

linear from 5‐1000 microgram/l (µg/l), 0.1‐200 mg/l and 0.02‐10 mg/l for PYR, SDX and

NSX respectively. Intra‐ and inter‐day relative SD (RSDs) were <15% for all analytes at

97

all concentrations. The limits of quantification (LOQ) were 2.5 µg/l, 0.1 mg/l and 0.02

mg/l and the limits of detection (LOD, determined as a signal‐to‐noise ratio of 5) were

1 µg/l, 0.05 mg/l and 0.01 mg/l for PYR, SDX and NSX respectively.

3.2.3.4 103B103BPopulationpharmacokineticanalysis

LogReR concentration vs. time datasets for PYR, SDX and NSX were analysed by nonlinear

mixed effect modelling using NONMEM (version 6.2.0, ICON Development Solutions,

Ellicott City, MD) with an Intel Visual FORTRAN 10.0 compiler. Linear mammillary

model subroutines within NONMEM (ADVAN2/TRANS2 and ADVAN4/TRANS4), FOCE

with ‐ interaction, and OFV were used to construct and compare plausible models.

Unless otherwise specified, a difference in OFV ≥6.63 (2 distribution with 1 d.f.,

P<0.01) was considered significant. Due to the small number of samples with low

concentrations, those below the LOD were not included in the analysis while

concentrations between the LOD and LOQ were kept at their measured

concentrations.

As the subjects were infants with a range of ages, it was important to incorporate

maturation of CL into the model. Therefore total clearance (CLRTR) was defined as the

sum of hepatic clearance (CLRHR) and renal clearance (CLRRR), i.e. CLRTR=CLRHR+CLRRR. The age‐

adjusted hepatic clearance CLRHR was determined using a sigmoid ERmaxR model345 as

TVCLRH Rx (PMAHillCL/(PMAHillCL+MATCLR50R

HillCL)), where TVCLRHR is the population average

value for hepatic clearance, PMA is the postmenstrual age (the age of the infant

recorded from the last menstrual cycle of the mother during pregnancy rather than

birth), HillCL is the Hill coefficient for hepatic clearance and MATCLR50R is the PMA at

which CLRHR is 50% of the mature value. When an accurate PMA could not be obtained it

was estimated from the postnatal age (PNA) and average gestation in PNG.346‐348 CLRRR

was adjusted to a standardized value for an estimated GFR(eGFR) of 120

ml/min/1.76m2, i.e. TVCLRRR x (eGFR/120), where CLRRR is the adjusted renal clearance,

TVCLRRR is the population average value for renal clearance, and the eGFR was

determined from the CysC concentration as 91.62 x (1/CysC1.123).349

Allometric scaling using WT was also used on all volume (*(WT/70)1.0) and clearance

(*(WT/70)0.75) terms. One‐ and two‐compartment models with first order absorption

98

without lag time were assessed for both SDX and PYR. As few data exist to describe the

absorption phase of both drugs, kRaR was fixed to previously published value for

infants.350 IIV was added to parameters for which it could be estimated reasonably

from available data. As logReR concentration data were used, an additive model

(representing proportional error) was used for RUV.

In the development of the final models, we investigated the influence of the covariates

dosing group, relative dose (mg/kg), PMA, malaria status, concomitant treatment with

AQ, and initial Hb concentration using the generalized additive modelling procedure

within Xpose and by inspection of correlation plots. Covariate relationships identified

by this procedure were evaluated within the NONMEM model and inclusion of the

covariate required a significant decrease in OFV accompanied by a decrease in the IIV

of that parameter. Correlations among IIV terms and WRES plots were also used in

model evaluation.

Once a final model for SDX was obtained the parameter estimates were fixed and an

additional compartment was added in order to model NSX concentrations. In order to

allow identifiability in the model, the percentage conversion of SDX to NSX was fixed to

60% based on information from the product information.225 The elimination of NSX

was assumed to be entirely renal.351 The influence of the covariates was assessed on

new model parameters using the method described above.

A bootstrap procedure using Perl speaks NONMEM (PSN) and the resulting parameters

were then summarized as median and 2.5th and 97.5th percentiles (95% empirical CI) to

facilitate validation of the final model parameter estimates. In addition, stratified VPCs

and NPCs were also performed using PSN with 1000 replicate datasets simulated from

the original dataset. NPCs stratified according to PMA were assessed by comparing the

actual with the expected number of data points within the 20, 40, 60, 80, 90 and 95%

PI. For VPCs the resulting 80% PI for drug concentrations were plotted with the

observed data to assess the predictive performance of the model.

99


As previously reported in a study of SP PK in pregnant vs. non‐pregnant women,314 and

using estimates of centrality and variance for PK parameters from previous paediatric

studies226, 227, 230, 342, 352 and an assumed 20% attrition rate, a sample size of 35 in each

group in the present study would be expected to show a >30% increase in the

magnitude of any PK parameter in the double‐dose group at α=0.05 and β=0.1. SPSS

17.0 (SPSS inc. Chicago, IL, USA) was used for all statistical analysis unless otherwise

specified. Data are summarized as mean±SD or median [IQR] as appropriate. Student’s

t‐test or the Mann‐Whitney U test was used for two‐sample comparisons. Categorical

data were compared using either Pearson Chi‐squared or Fisher’s exact test, and

multiple means by repeated measures ANOVA. A two‐tailed level of significance of

0.05 was used.

100

3.2.4 47B47BResults

3.2.4.1 105B105BPatientcharacteristics

Seventy infants were enrolled between April 2008 and December 2008 with equal

numbers in each dose group. Baseline subject characteristics are summarized in Table

3‐2. The double‐dose group received a significantly higher mg/kg dose than the

conventional dose group (P<0.001) and was taller by a mean of 4.3 cm (P=0.015). The

double‐dose group was also older (by a mean of 47 days) and heavier (by 0.4 kg) than

the conventional dose group but these differences were not statistically significant

(P>0.05).

3.2.4.2 106B106BTolerability,safetyandefficacy

Both doses were well tolerated. There were no changes in symptoms in either group

compared to pre‐dose profiles, including an absence of dermatological conditions.

There were no significant changes in Hb, or in plasma urea, creatinine or CysC, over

time. In the conventional dose group, there was a significant but transient mean fall in

plasma albumin of 2 g/l at day 2 (from 38‐36 g/l, P<0.01), but there were no

concomitant increases in plasma bilirubin or hepatic enzymes in either group.

Table 3‐2 Baseline characteristics of study participants. Data are number (%), mean±SD or median [IQR].

Conventional dose

(n=35)

Double dose

(n=35)

Postmenstrual age (days) 454 [383‐513] 501 [428‐532]

Sex (% male) 22 (63%) 24 (69%)

Weight (kg) 6.58 ± 1.31 6.98 ± 1.1

Height (cm) 61.8 ± 6.5 66.1 ± 7.8

Axillary temperature (°C) 36.5 ± 0.6 36.4 ± 0.6

P. falciparum parasitaemiaa 1 (3%) 0 (0 %)

P. vivax parasitaemiaa 3 (9%) 3 (9%)

Respiratory rate (/min) 40 ± 11 42 ± 11

Supine pulse rate (/min) 133 ± 14 133 ± 15

Mean upper arm circumference (cm) 13.2 ± 3.5 13.7 ± 2.6

Haemoglobin (g/l) 9.5 ± 1.3 9.5 ± 1.2

eGFR (ml/min/1.73m2) 80 ± 20 84 ± 16

Sulfadoxine dose (mg/kg)b 35.6 ± 5.6 67.1 ± 12.6

Pyrimethamine dose (mg/kg)b 1.8 ± 0.3 3.4 ± 0.6 a one infant had a mixed P. vivax/falciparum infection, b P<0.001

101

Five infants with vivax malaria and one infant with a mixed P. vivax/ falciparum

infection at enrolment responded to treatment. Three other infants in the

conventional dose group and two in the double‐dose group were administered

antimalarial drugs during follow‐up at an external health care facility and no blood

smears were available for review. No other subjects became symptomatic during the

study. Only two infants in the conventional dosing group and one in the double dosing

group who were aparasitaemic at entry had a positive blood slide on day 28 (all for P.

vivax). All were asymptomatic and each was treated according to PNG national

treatment guidelines.


There were 248, 255 and 247 drug concentration measurements available for PK

modelling for PYR, SDX and NSX respectively. There were four samples with PYR

concentrations between the LOD and LOQ and a further four with concentrations

Table 3‐3 Final population PK parameters and bootstrap results for PYR.

Final model

estimate (RSE%)

Bootstrap (n=1000)

median [95% CI]


kRaR (/h) 0.779 FIXED

VRCR/F (l/70kg) 222 (4) 221 [202‐242]

VRPR/F (l/70kg) 64.1 (24) 63.0 [41.8‐128.5]

Q/F (l/h/70kg) 0.0735 (19) 0.0788 [0.0486‐0.1470]

CLRRR/F (l/h/70kg) 0.416 (64) 0.3820 [0.0621‐0.9868]

CLRHR/F (l/h/70kg) 0.854 (24) 0.878 [0.466‐1.220]

MATCLR50R (days) 318 (8) 326 [286‐367]

HillCL 7.39 (43) 7.80 [3.53‐35.18]


IIV VRCR/F 13.0 (36) 13.6 [3.6‐24.7]

IIV CLRTR/F 27.8 (13) 27.0 [18.2‐35.0]

IIV Q/F 34.1 (32) 36.4 [17.4‐53.4]

Ra (VRCR/F, CLRTR/F) 0.533 (69) 0.563 [‐0.059‐0.826]

R (VRCR/F, Q/F) 1 FIXED

R (CLRTR/F, Q/F) 0.533 (69) 0.563 [‐0.059‐0.826]

RUV (%) 33.6 (23) 32.6 [26.9‐37.4] acorrelation coefficient.OFV in final model: ‐97.384, bootstrap OFV (median [95% CI]): ‐111.812 [‐170.137‐‐62.348]

102

below the LOD for PYR. In addition, seven samples were of insufficient volume for

measurement of PYR after SDX/NSX assay. There were no SDX or NSX concentrations

below the LOQ but NSX concentrations could not be determined in eight samples due

to an unidentified interfering peak. For PYR, a two‐compartment model was superior

to a one‐compartment model with a lower OFV (‐87.081 vs. ‐30.030) and a less biased

WRES vs. time plot. The model parameters were kRaR, CLRHR/F, CLRRR/F, VRCR/F, VRPR/F, Q/F,

HillCL and MATCLR50R. IIV was estimable on CLRTR/F, VRCR/F and Q/F. As the correlation

between the variability of VRCR/F and Q/F was very close to 1, it was subsequently fixed

to unity to assist with successful determination of the covariance matrix. None of the

covariates tested improved the model significantly and therefore the final model

contained only the effects of PMA and WT anticipated from maturation and allometric

scaling, respectively.

The final parameter estimates and the results of the bootstrap procedure for PYR are

shown in Table 3‐3. All model parameters had a bias <11%. GOF plots for PYR are

shown inFigure 3‐1. NPCs of the data showed good predictive performance, as did VPC

plots of the observed drug concentrations and their 80% PI (the 10th and 90th

percentile boundaries) stratified by dosing group (Figure 3‐2 A and B). Post hoc

parameter estimates are shown in Table 3‐4. There was no difference between the

two groups for any of these parameters except for AUCR0‐∞ Rwhich was significantly

higher in the double‐dose group (4,915 vs. 2,844 μg19T∙19Td/l). Median VRSSR for the combined

study sample was 27.8 l, and tR½αR and tR½βR were 72.7 and 374 h, respectively.

Table 3‐4 Post hoc Bayesian predicted PK parameters for PYR for PNG infants given conventional and double doses of SDX/PYR (median [IQR]).

Parameter Conventional dose

(n=35)

Double dose

(n=35) P valuea

CLT/F (l/h) 0.183 [0.13‐0.21] 0.199 [0.164‐0.229] NSb

VC/F (l) 20.2 [17.8‐24.1] 22.1 [19‐25] NS

VP/F (l) 6.18 [5.32‐6.81] 6.49 [5.83‐7.03] NS

VRSSR/F (l) 25.781 [23.319‐30.957] 28.8 [24.9‐31.8] NS

Q/F (l/h) 0.0118 [0.0073‐0.0165] 0.0135 [0.009‐0.0196] NS

tR½αR (h) 73.7 [67.1‐87.7] 70.7 [62.3‐82.3] NS

tR½β R(h) 391 [300‐565] 361 [272‐511] NS

AUC0‐∞ PYR (μg19T∙19Td/l) 2,844 [2,486‐3,571] 4,915 [4,311‐5,681] <0.001 a Mann‐Whitney test, b NS= non‐significant (P>0.05)

103

Initial modelling of SDX revealed that a one‐compartment model was appropriate as

there was minimal bias in the WRES plot that was not improved when a two‐

compartment model was fitted. The model parameters were kRaR, CLRHR/F, CLRRR/F, V/F,

HillCL and MATCLR50R. IIV was able to be estimated on CLRTR/F and V/F. There was a

significant relationship between relative dose (in mg/kg) and relative bioavailability

Time (h)

0 100 200 300 400 500 600 700

Co

nd

itio

na

l we

igh

ted

re

sid

ua

ls (

py

rim

eth

am

ine

)

-4

-2

0

2

4

Predicted plasma pyrimethamine (g/l)

1 10 100 1000O

bs

erv

ed

pla

sm

a p

yri

me

tha

min

e (

g/l)

1

10

100

1000

A

B

Figure 3‐1 (A) Population (○) and individual (●) predicted versus observed plasma pyrimethamine concentrations (µg/l on log10 scale) for the final model. The line of identity is also shown. (B) Conditional weighted residuals vs. time for pyrimethamine final model.

Time (h)

0 200 400 600 800

Pla

sm

a p

yri

me

tha

min

e (

g/l)

0.1

1

10

100

1000

10000 A

Time (h)

0 200 400 600 800

Pla

sm

a p

yri

me

tha

min

e (

g/l)

0.1

1

10

100

1000

10000 B

Figure 3‐2 Visual predicted check plots for PYR showing simulated 10th (short dashed line), 50th (dotted line) and 90th (solid line) percentile concentrations and observed concentration (log scale) data (grey open circles) versus time (log scale) for conventional dose (A) and double‐dose (B) participants.

104

which conformed to a power function, specifically individual relative bioavailability = 1

x ([individual relative dose]/[average relative dose]effect parameter). The value of the

power effect parameter was ‐0.56, indicating that, when the dose is doubled, the

bioavailability falls by 32.2%.

The final parameter estimates and the results of the bootstrap procedure are shown in

Table 3‐5. With the exception of CLRRR, all parameter estimates had bias <13%. The

median bootstrap value for CLRRR was almost double the initial estimate (195%),

demonstrating the difficulty in delineating the difference between and estimating the

hepatic and renal clearance using this methodology. GOF plots for SDX are shown in

Table 3‐5 Final population PK parameters and bootstrap results for SDX and NSX. Parameters for NSX modelling obtained after fixing model parameters for SDX are highlighted in grey.

Final model

estimate (RSE%)

Bootstrap (n=1000)

median [95% CI]


kRa R(/h) 1.23 FIXED

V/F (l/70kg) 24.2 (4) 24.2 [22.5‐26.1]

CLRRR/F (l/h/70kg) 0.0046 (113) 0.0086 [0.0005‐0.0267]

CLRHR/F (l/h/70kg) 0.0458 (16) 0.0427 [0.0290‐0.0640]

MATCLR50R (days) 271 (8) 286 [248‐360]

HillCL 4.07 (52) 4.61 [1.56‐15.54]

Relative dose on relative bioavailability [power]

‐0.56 (14) ‐0.54 [‐0.71‐ ‐0.38]

%NSX (%) 60 FIXED

VRNSXR/F (l/70kg) 11.7 (10.7) 11.7 [9.4‐14.4]

CLRNSXR/F (l/h/70kg) 0.758 (5) 0.756 [0.690‐0.838]


IIV V/F 23.0 (11) 22.2 [17.1‐26.5]

IIV CLRTR/F 23.8 (11) 23.4 [17.9‐28.3]

IIV VRNSXR/F 42.8 (19) 41.7 [21.0‐56.3]

IIV CLRNSXR/F 36.2 (26) 35.6 [25.5‐45.1]

R (V/F, CLRTR/F) 0.644 0.653 [0.439‐0.814]

R (VRNSXR/F, CLRNSXR/F) 0.218 0.226 [‐0.474‐0.729]

RUV SDX (%) 16.5 (11) 16.4 [12.9‐20.1]

RUV NSX (%) 37.1 (9) 37.0 [30.2‐43.1]

OFV in final model: ‐521.177, Bootstrap OFV (median [95% CI]): ‐529.222[‐647.701‐ ‐428.084]

105

Figure 3‐3. NPCs of the data showed good predictive performance, as did VPC plots of

the observed drug concentrations and their 80% PI stratified by dose group in Figure

3‐4.

Time (h)

0 100 200 300 400 500 600 700Co

nd

itio

na

l we

igh

ted

re

sid

ua

ls (

su

lfa

do

xin

e)

-4

-2

0

2

4

Predicted plasma sulfadoxine (mg/l)

10 100 1000

Ob

se

rve

d p

las

ma

su

lfa

do

xin

e (

mg

/l)

10

100

1000A

B

Figure 3‐3 (A) Population (○) and individual (●) predicted versus observed plasma sulfadoxine concentrations (µg/l on log10 scale) for the final model. The line of identity is also shown. (B) Conditional weighted residuals vs. time (log scale) for sulfadoxine final model.

Time (h)

0 200 400 600 800

Pla

sm

a s

ulf

ad

ox

ine

(m

g/l)

1

10

100

1000 A

Time (h)

0 200 400 600 800

Pla

sm

a s

ulf

ad

ox

ine

(m

g/l)

1

10

100

1000 B

Figure 3‐4 Visual predicted check plots for SDX showing simulated 10th (short dashed line), 50th (dotted line) and 90th (solid line) percentile concentrations and observed concentration (log scale) data (grey open circles) versus time (log scale) for conventional dose (A) and double‐dose (B) participants.

106

An additional compartment was added to the final SDX PK model to incorporate the

data for NSX. This resulted in three additional model parameters; VRNSXR/F, CLRNSXR/F and

percentage of total SDX elimination representing conversion of SDX to NSX (%NSX). As

these three parameters cannot be estimated simultaneously, %NSX was fixed to 60%

based on published data.225 The estimates of VRNSXR/F and CLRNSXR/F are directly related to

%NSX and therefore the value of these parameters should be interpreted with caution.

However AUC and tR½R for NSX remain unchanged for different values of %NSX. VRNSXR/F

and CLRNSXR/F were not influenced by any of the available covariates. Final parameter

estimates and results of the bootstrap procedure are shown in Table 3‐5. Bias was <5%

for all NSX parameters and NPCs and VPCs were performed on the NSX dataset and

indicated good predictive performance of the model (data not shown).

There were some significant differences between conventional and double‐dose

groups in the post hoc parameter estimates for both SDX and NSX (Table 3‐6). These

included expected differences in the AUCR0‐∞R for both SDX and NSX, but also differences

in the tR½R and clearance for both drugs which were not revealed by the model covariate

building stage. A higher clearance and lower tR½R in the double‐dose group can be

attributed to organ maturation as these infants were older than those in the

conventional dose group. The median tR½ Rof NSX for the combined study sample was

shorter than that of SDX (8.9 vs. 218 h). The percentage of the AUCR0‐∞R of NSX when

compared to that for SDX was the same for both dose groups (approximately 5%).

Table 3‐6 Post hoc Bayesian predicted PK parameters for SDX and NSX in PNG infants given conventional and double dosing of SDX/PYR (median [IQR]).

Parameter Conventional dosing

(n=35)

Double dosing

(n=35) P‐valuea

CLRT,SDXR/F (l/h) 0.0068 [0.0057‐0.0087] 0.0072 [0.0068‐0.0105] 0.032

VRSDXR/F (l) 2.20 [1.95‐2.53] 2.23 [1.97‐2.64] NSb

tR½ SDXR (h) 232 [203‐252] 207 [179‐232] 0.006

AUCR0‐∞ SDXR (mg19T∙19Td/l) 1,460 [1,167‐1,707] 2,434 [1,881‐2,987] <0.001

CLRNSXR/F (l/h) 0.081 [0.060‐0.094] 0.101 [0.081‐0.116] 0.012

VRNSXR/F (l) 1.14 [1.00‐1.28] 1.17 [0.959‐1.30] NS

tR½ NSXR (h) 10.3 [7.86‐12.2] 8.69 [6.84‐10.6] 0.027

AUCR0‐∞ NSXR (mg19T∙19Td/l) 1,796 [1,397‐2,154] 2,890 [2,482‐3,609] <0.001

AUCR0‐∞ NSXR/ AUCR0‐∞ SDXR (%) 5.0 [4.3‐6.5] 5.0 [4.3‐6.0] NS a Mann‐Whitney test, b NS= non‐significant (P>0.05)

107

Sigmoid ERmaxR curves of hepatic maturity for SDX and PYR by PMA are shown in Figure

3‐5. They are closely related to MATCLR50R values of 318 days and 271 days for PYR and

SDX, respectively. Of the 70 infants, 48 (69%) and 38 (54%) had an estimated hepatic

clearance >90% of adult values for PYR and SDX, respectively.

Figure 3‐5 Maturation as a fraction of adult clearance for PYR (solid line) and SDX (dashed line) predicted from the PK model plotted against PMA. A box plot of the PMA in the recruited subjects is included to show its distribution in relation to maturation of clearance.

PMA (d)

100 200 300 400 500 600

Fra

ctio

n o

f ad

ult

cle

aran

ce

0.00

0.25

0.50

0.75

1.00

SDX PYR

108


The present study is the first to investigate the PK of SP in infants living in a malaria‐

endemic setting in which IPTi is appropriate. It is also the first to investigate the

possibility that a higher dose than conventionally recommended should be given to

achieve therapeutic plasma concentrations in this age group, as has been

recommended for children aged 2‐5 years.226 SP was well tolerated by all infants and

there was no evidence of hepatorenal or bone marrow toxicity even at the higher

dose. The AUCR0‐∞R of both SDX and PYR was significantly higher in the double‐dose

group. However, there was a 32% reduction in the relative bioavailability of SDX when

the dose was doubled, possibly due to saturation of absorption. The percentage of NSX

to SDX exposure (AUC) was the same in both groups suggesting that a double dose

does not affect the metabolic clearance of SDX. The PK properties of PYR were not

dose‐dependent in the present study.

The PK parameters for PYR observed in our children are different to those observed in

other paediatric studies.226, 227, 230, 350, 353 We found a longer tR½βR (15.6 vs. 2.67‐4.46 d)

and a higher conventional dose AUC (2,844 vs. 1,052‐2,607 μg19T∙19Td/l). This may reflect the

fact that most of our children were well. In addition, we employed a relatively long

duration of sampling that facilitated identification of bi‐exponential elimination, a

profile reported previously in studies of adults314, 354, 355 but not children. While one

paediatric study sampled out to 42 d, the drug could not be quantified in 40% of the

samples.226 Although the mean conventional dose PYR AUC in the present study was in

the range of previously reported values in adults (1,602‐3,166 μg19T∙19Td/l),226, 355‐357 these

latter data may have been underestimates because of truncated sampling and/or use

of a relatively insensitive assay. In a study of non‐pregnant PNG women using a similar

sampling profile, assay and PK modelling techniques to those of the present study,314

the mean conventional dose PYR AUC (4,419 μg19T∙19Td/l) was similar to that in the present

double‐dose group. Together with the available tolerability and safety data from the

present study, these considerations suggest that double‐dose PYR is appropriate as

part of SP IPTi.

We found that SDX also had a longer mean elimination tR½R (9.1 vs. 4.1‐8.6 d) and higher

conventional dose mean AUC (1,460 vs. 460‐932 mg19T∙19Td/l) when compared to other

109

studies in children.226, 227, 230, 342, 352 However, the mean AUC was within the range

found in adults (508‐2,757 mg19T∙19Td/l),226, 355‐357 including non‐pregnant women (1,386

mg19T∙19Td/l) from the same location as the present study.314 Although the difference in AUC

compared to other paediatric populations may be explained, as with PYR, by our ability

to detect drug concentrations for longer post‐dose than in previous studies as well as

to the relative health of our subjects, only a few studies have included infants aged <1

year and these formed a minority of the patients recruited. As our sample includes

only children <13 months of age, a limited maturation of elimination processes is likely

to play a role in the longer tR½R and higher AUC observed for both drugs even in the

conventional dose group. Indeed, we found evidence of a slower maturation of these

processes for SDX than PYR.

In the present study we used plasma CysC rather than creatinine to estimate GFR. The

conventional Schwartz creatinine‐based formula relies upon estimates of body

composition358 whereas CysC‐based formulae do not,359 making the estimates more

robust. We used the formula derived by Filler et al.349 as it was derived from a large

paediatric sample and the same PETIA CysC assay used in the present study. CysC

concentrations generated by other assays such as particle‐enhanced

immunonephelometry may differ from those from PETIA.359 The Filler et al. formula is

comparable to others based on CysC derived in children.360‐363

Since hepatic maturation would be still occurring within the age range of our subjects,

it was appropriate to include this phenomenon in our model.228, 364, 365 We used a

sigmoid ERmaxR approach as this has been used previously with a number of other

drugs.228, 366‐369 Our estimates of MATCLR50R, namely 315 and 271 days for PYR and SDX

respectively, fell in the range reported in these latter studies (270‐380 d). The estimate

of the Hill coefficient for SDX was also consistent (4.07 vs. 2.78‐4.6), but the Hill

coefficient for PYR was higher than previously reported (7.39). Although our study age

range captured the process of maturation, most of our infants had clearances >90% of

adult values and very few were <50% (see Figure 3‐5). This limits our ability to

characterize coefficients of maturation which are likely to be inappropriate outside this

age range. For example, adult estimates of tR½ Rfor SDX (333 h) and PYR (tR½αR 113 h, tR½βR

647 h) based on the modelling presented here are higher than previously reported.226,

110

314, 354‐357 Future studies of this type should include a larger range of ages so that the

maturation process from birth to adult activity levels can be determined more

accurately.

Other studies have provided data relevant to the question of whether a higher SP dose

should be given to infants. A PK evaluation of SDX in children aged 6 months to 5 years

with malaria found that those aged <24 months had a lower AUCR0‐336hR than their older

counterparts (12,500 vs. 16,900 mg19T∙19Th/l).352 However all children <24 months of age

received half the dose of older children regardless of WT and no average dose by WT

was reported, thus complicating interpretation of the data. In a similar study,226 an

age‐stratified non‐compartmental analysis of AUCR0‐∞R showed that 1‐2 year‐olds had

sufficient drug exposure while children aged 2‐5 years required a double dose. The

study only had 11 children within the 1‐2 year old age range and, because only whole

tablets were given, the mean dose in this group was almost twice that of ≥12 year olds

(50/2.5 vs. 27.3/1.36 mg/kg). In a population‐based PK analysis of SP in children with

congenital toxoplasmosis aged 1 week to 14 years,227 lighter children had a shorter tR1/2R

and therefore a lower drug exposure. This conclusion was based on use of allometry

since age‐based maturation contributed little to the model, perhaps because of the

small numbers in the younger age groups. Interpreted within their limitations, these

various studies also provide evidence that higher mg/kg SP doses are required in

younger children, including those <1 year of age.

Relatively recent data from the study area indicate that AQ‐SP treatment (until

recently the recommended first‐line antimalarial therapy for young PNG children) is

associated with close to a 90% 28‐day adequate clinical and parasitologic response for

both falciparum (PCR‐corrected) and vivax malaria.370 This is a suboptimal response

but still suggests that either conventional or double‐dose SP treatment in the present

study is likely to have contributed to the relatively small number of infections detected

during follow‐up. Although the present study was not designed to assess relative

efficacy, especially since interpretation of emergent vivax infections remains

problematic371 and given that only one dose was administered rather than the several

scheduled during IPTi, fewer children were treated for symptomatic malaria during

follow‐up or were slide positive on day 28 in the double‐dose group. Indeed, there is

111

evidence from epidemiologic studies utilizing fixed‐dose regimens62, 341 that

appropriate mg/kg doses of SP should be used in IPTi programs to ensure adequate

levels of prevention, especially for symptomatic compared to asymptomatic falciparum

malaria.372

In the light of this dose‐dependency, the fact that no study has shown >60% protective

efficacy during the first year of life,62, 339 evidence that higher blood PYR

concentrations facilitate parasite clearance in paediatric falciparum malaria,342 and the

fact that double‐dose SP in our subjects was safe, well tolerated and associated with

higher exposure to both drug components (especially SDX), the present data argue for

the potential use of double‐dose SP in infancy. As in recent adult studies of PYR

disposition,314 we found that the mean elimination tR½R, of PYR and SDX were longer

than previously reported, a factor that may contribute to the duration of effective

prophylaxis. Although allometric considerations (shorter half‐lives in smaller subjects)

may justify higher SP dosing in infants, we recommend that consideration must be

given to the maturation of hepatorenal elimination processes and the possibility that

increased doses may be inappropriate in very young infants.

112



their kind co‐operation during the study. We also thank Christine Kalopo and Bernard

(“Ben”) Maamu for clinical and/or logistic assistance. The authors note with deep

regret that Servina Gomorrai, who assisted with patient recruitment and data

collection, passed away during the study. The study was funded by a grant from the

IPTi Consortium and utilised facilities developed with support from the National Health

and Medical Research Council (NHMRC) of Australia (grant 458555). TMED is the

recipient of an NHMRC Practitioner Fellowship.

113

114

115

164B164BTREATMENTOF

UNCOMPLICATEDMALARIA

INCHILDREN

116

117

4 3B3BPopulationPharmacokineticsofArtemether,

Lumefantrine,andTheirRespectiveMetabolites

inPapuaNewGuineanChildrenwith

UncomplicatedMalaria


The aims of the study presented in this chapter were to add to the current literature

on the PK of AL in children, particularly those of DBL, and to then compare the PK in

children to those in adults in order to inform an appropriate paediatric dose regimen.

When this study was conceived, an efficacy trial was being undertaken in PNG to

assess a number of newer combinations against conventional CQ/SP21. One of these

was AL. The literature at that time contained few data of the PK of AL in children and

even fewer reports of DBL concentrations and its role in treatment outcome.

Therefore a study of a small group of children in PNG was conducted to aid

interpretation of the results from the efficacy trial and assist with delineating the role

of DBL in treatment outcome.

This study resulted in the publication2 presented in this chapter. Entitled, “Population

PK of artemether, lumefantrine, and their respective metabolites in Papua New

Guinean children with uncomplicated malaria” it was published in the journal

Antimicrobial Agents and Chemotherapy (2011. 55(11):p. 5306‐13). The contribution of

each of the authors is outlined in section i, which also contains details of ethical

approvals and supporting funding. While the complete publication is provided in

section xi.a below, it has been reformatted to conform to thesis requirements set by

the University of Western Australia. The references have been combined with those

for the thesis as a whole and can be found in section x below.

118

119


Sam Salman,A Madhu Page‐Sharp,B Susan Griffin,C Kaye Kose,C Peter M. Siba,C Kenneth

F. Ilett,A Ivo Mueller,C, Timothy M. E. DavisAR.

ASchool of Medicine and Pharmacology, University of Western Australia, Fremantle

Hospital, Fremantle, Western Australia, Australia;

BSchool of Pharmacy, Curtin University of Technology, Bentley, Australia;

CPapua New Guinea Institute of Medical Research, Madang, Papua New Guinea.


There are sparse published data relating to the PK properties of ARM, LUM and their

active metabolites in children, especially DBL. We studied 13 Papua New Guinean

children aged 5‐10 years with uncomplicated malaria who received the six

recommended doses of ARM (1.7 mg/kg) plus LUM (12 mg/kg) given with fat over 3

days. Intensive blood sampling was carried out over 42 days. Plasma ARM, DHA, LUM

and DBL were assayed using liquid chromatography‐mass spectrometry. Multi‐

compartmental PK models for drug plus metabolite were developed using a population

approach that included plasma ARM and DHA BLQ concentrations. Although ARM

bioavailability was variable and its clearance increased by 67.8% with each dose, the

median areas under the plasma concentration‐time curve (AUCR0–∞R) for ARM and DHA

(3,063 and 2,839 µg∙h/l, respectively) were similar to those reported previously in

adults with malaria. For LUM, the median AUCR0–∞R (459,980 µg19T∙19Th/l) was also similar to

that in adults with malaria. These data support the 35% higher mg/kg dose

recommended for children 15‐35 kg vs. a 50 kg adult but question the

recommendation for a lower dose in children weighing 12.5‐15 kg. The median

DBL:LUM ratio in our children was 1.13%, within the range reported for adults and

higher at later time‐points because of the longer DBL terminal elimination tR½R. A

combined DBL plus LUM AUCR0–∞R weighted on in vitro antimalarial activity was

inversely associated with recurrent parasitaemia, suggesting that both parent drug and

metabolite contribute to AL treatment outcome.

120


AL is a fixed‐dose combination therapy used widely for the treatment of malaria.373

ARM is a lipophilic artemisinin derivative that is converted in vivo to DHA, an active

metabolite. Both ARM and DHA have short half‐lives133, 135, 137‐139, 143, 144 but a rapid

effect on parasitaemia. LUM is a highly lipophilic drug with a longer tR½R

135, 137, 138, 143, 164,

182, 374 which is combined with ARM primarily to prevent late recrudescence. Although

the PK properties of ARM, DHA and LUM have been well documented in adults,133‐135,

137, 138, 143, 164, 182, 184, 374‐376 there are scant and inconsistent data relating to the

disposition of DBL, a potent LUM metabolite160, 163, 377, 378 that may influence AL

treatment outcome.163 Reported plasma DBL:LUM concentration ratios after AL dosing

in adults differ >10‐fold,143, 184 while the PK properties of DBL in children are unknown.

In addition, although several studies have attempted to characterize LUM disposition

in children with malaria,142, 144, 183 methodological issues complicate their comparison

with adult data. One study involving a limited sampling schedule suggested that AL‐

treated children with malaria receive an inadequate dose of LUM relative to healthy

adults,144 while the other studies either used pooled plasma concentrations183 or used

a truncated sampling schedule inadequate to characterize LUM PK.142

In view of this situation, we have characterized the population PK of ARM, LUM and

their metabolites in paediatric malaria using a rich sampling schedule to assess

potential differences in disposition between children and adults, and to add to the

limited data on DBL disposition and its role in AL treatment outcome.

121


4.2.3.1 108B108BPatients

We recruited children aged 5‐10 years from Alexishafen Health Centre, Madang

Province on the north coast of PNG. The clinic serves an area where P. falciparum and

P. vivax are hyperendemic, and P. ovale and P. malariae are also transmitted. Children

with an axillary temperature >37.5°C or a history of fever in the previous 24 h were

screened with a Giemsa‐stained thick blood film read by an on‐site trained

microscopist. Those with a mono‐infection of P. falciparum (>1,000 asexual

parasites/microliter), or P. vivax, ovale or malariae (>250/microliter) were eligible

provided that the child’s parents gave informed consent, there were no features of

severe malaria,379 they had not taken any antimalarial drug in the previous 14 days,

there was no evidence of another cause of fever, and there were no features of

malnutrition or other chronic co‐morbidity. The study was approved by the Medical

Research Advisory Committee of the Department of Health, PNG.

4.2.3.2 109B109BClinicalmethods

After enrolment, a standardized history was taken and a clinical examination

performed. A 3 ml blood sample was taken for blood film microscopy, a baseline Hb

and blood glucose, and subsequent drug assay of separated plasma. Urinalysis and

audiometric assessment were performed. Each child was given AL (Coartem, Novartis

Pharma Ltd, Switzerland) at a dose of 1.7 and 10 mg/kg, respectively to the nearest

tablet. This dose was repeated at 8, 24, 36, 48 and 60 h with the exact time of dosing

recorded. All doses were given under direct observation with at least 50 ml of cow’s

milk (equivalent to 2 g of fat). Further venous blood samples were taken from an

indwelling intravenous catheter at 4, 8, 12, 24, 36, 40, 48, 60, 64, 68 and 72 h, and

then by venesection on days 4, 5, 7, 14 and 28. All samples were centrifuged promptly

and RBCs and separated plasma stored frozen at ‐80°C until assay. Detailed clinical

assessment, including a symptom questionnaire, blood film, Hb and blood glucose, was

repeated on days 1, 2, 3 and 7, with additional clinical assessment and blood films on

days 14, 28 and 42.

122

4.2.3.3 110B110BLaboratorymethods

All blood smears taken at baseline and during follow‐up were examined independently

by two skilled microscopists in a central laboratory. Each microscopist viewed 100

fields at 1,000x magnification before a slide was considered negative. Any slide

discrepant for positivity/negativity or speciation was referred to a third microscopist

for adjudication.

For drug assays, HPLC‐grade acetonitrile (Merck, Kilsyth, Australia), tert‐butyl chloride,

ethyl acetate, glacial acetic acid and formic acid (Merck, Darmstadt, Germany), and

ammonium formate (Sigma‐Aldrich, Gillingham, UK) were used. Other solvents and

chemicals were of analytical grade. Stock solutions (1 µg/l in methanol) of ARM (AAPIN

Chemicals, Abingdon, UK), DHA (Sigma, St Louis, MO) and ART (used as an internal

standard; Sigma) were stored protected from light at ‐80°C and used to prepare

working dilutions (0.1, 1, and 10 µg/ml). Calibration curves (2‐200 µg/l) were

constructed for DHA and ARM by spiking blank plasma. Quality control (QC) samples

were prepared in blank plasma at 10, 20, 50 and 200 µg/l and also stored at ‐80 °C

prior to use.

ARM and DHA were extracted as previously described380 with the following

modifications. Briefly, solid phase extraction (SPE) Bond Elut ® PH columns (Varian Inc,

Palo Alto, CA) were pre‐conditioned with 1 ml methanol followed by 1 ml 1 moles/l

(M) acetic acid. Plasma (0.5 ml) was spiked with internal standard (ART 100 µg/l) and

loaded onto the SPE column and drawn through with a medium vacuum. The column

was then washed twice with 1 M acetic acid (1 ml), followed by 20% v/v methanol in

1M acetic acid (1 ml). The column was dried under low vacuum for 30 min and the

retained drugs eluted using 2 ml t‐butyl chloride: ethyl acetate (80:20% v/v). The

eluate was then evaporated under vacuum at 35°C and reconstituted in 50 μl mobile

phase and kept overnight to equilibrate the α and β anomers of DHA.380 Only the α‐

anomer was used for quantification. The injection volume was 10 μl.

The LC‐MS system used was a single quad mass spectrometer (Shimadzu, Kyoto, Japan)

with electrospray ionization (ESI) and atmospheric pressure ionization (APCI) systems.

Assays were performed with 20 mM ammonium formate (pH 5):acetonitrile in 0.1%

123

formic acid (40:60) at a flow rate of 0.2 ml/min, and chromatographic separation

undertaken at ambient temperature on a Synergy fusion‐RP CR18R (150 mm x 2.0 mm

i.d.) column coupled with a 4 mm x 3 mm i.d., 5 µm particle CR18R guard column

(Phenomenex, Lane Cove, Australia). Retention times were 4.5, 7.5 and 12.7 min for

DHA, ART and ARM, respectively. Optimized mass spectra were acquired with an

interface voltage of 4.5 kV, a detector voltage of 1 kV, a heat block temperature of

400°C and a desolvation gas temperature of 250°C. Nitrogen was used as a nebulizer

gas at a flow rate of 1.5 l/min and dry gas flow of 10 l/min. Quantitation was

performed by selected ion monitoring using the dual ionization source mode. The

predominant fragmented ions m/z 221 for ARM and m/z 221 for DHA were used. For

ART, m/z=283 was monitored.

Standard curves were linear (r2 ≥0.999). Chromatographic data (peak area ratio of

DHA:ART and ARM:ART) were processed using LAB Solution software (Version 5,

Shimadzu, Japan). No matrix effect (ion suppression/enhancement) was observed

under methodologies described elsewhere,381 and performance of both assays,

assessed as intra‐ and inter‐day RSD across relevant concentration ranges, was similar

to that published previously.21, 380 Inter‐day accuracies of QC assays were <15% of

nominal values on all occasions. The limits of quantification and detection for DHA and

ARM were 2 and 1 μg/l, and 5 and 2 μg/l, respectively.

LUM and DBL were quantified in plasma using a UPLC‐LC‐MS/MS assay as previously

described.163 The linear range for LUM was 20‐20,000 ng/ml, and inter‐day variability

was 4.94, 4.93, 7.16 and 11.23% and intra‐day variability 2.83, 4.41, 4.11, 9.55% at

20,000, 2,000, 200 and 20 ng/ml, respectively. For DBL, the linear range was 0.5 ‐100

ng/ml, and inter‐day variability was 3.36, 3.47, 9.98 and 6.74% and intra‐day variability

2.47%, 3.46%, 8.16% and 3.48% at 50, 10, 1 and 0.5 ng/ml, respectively. As a LC‐

MS/MS method was used for DBL, matrix effects were assessed where IIV was 3.37%,

4.47% and 9.43% at 50, 10 and 1 ng/ml, respectively.


LogReR (natural log) plasma concentration‐time datasets for LUM with DBL and ARM with

DHA were analysed by nonlinear mixed effect modelling using NONMEM (v 6.2.0, ICON

124

Development Solutions, Ellicott City, MD) with an Intel Visual FORTRAN 10.0 compiler.

The first order conditional estimation with interaction (FOCE‐I) estimation method was

used for the LUM/DBL model and the Laplacian with interaction method for ARM/DHA.

OFV and WRES plots were used to choose suitable models during model‐building. As

FOCE‐I estimation was used, CWRES were also considered in the initial stages of model

building.299 However, as they were similar to WRES, the latter was considered suitable

for further model‐building. Concentrations were modelled as µg/ml with a conversion

factor for all metabolite parameters included into the model to account for the

difference in molecular weight between parent drug and metabolite. Allometric scaling

was used a priori, with volume terms multiplied by (WT/70)1.0 and clearance terms by

(WT/70)0.75.309 RUV was estimated as additive error for the logReR‐transformed data.

Models were parameterized using kRaR, VRCR/F, CL/F, VRPR/F and Q/F.

For the LUM/DBL model, plasma LUM concentrations were initially modelled using

inbuilt 2‐ and 3‐ compartment model structures with first‐order absorption and a fixed

lag time of 2 h134 (ADVAN 4 and 12). Once a suitable (3‐compartment) LUM model had

been determined, the DBL dataset was added and modelled simultaneously. User‐

defined linear mammillary models (ADVAN 5) were constructed testing 1‐, 2‐ and 3‐

compartments with and without FP LUM metabolism. As no data exist regarding the

degree of in vivo DBL conversion from LUM this was set to 100% to allow identifiability.

Therefore, all clearance and volume terms for DBL are relative to LUM bioavailability

(FRLUMR) as well as the degree of metabolic conversion from LUM (FRmet‐DBLR). The term

F*RDBLR (representing FRLUMR x FRmet‐DBLR) will be used for simplicity.

As 45% and 12% of plasma ARM and DHA concentrations, respectively, were BLQ we

used a published method known to produce reliable PK parameters in this situation.382,

383 The method (known as M3297) models continuous and categorical data

simultaneously. Concentrations above the LOQ are included as conventional

continuous data while those BLQ are treated as categorical and the (probability) that

they are BLQ maximized with respect to model parameters. This allows BLQ

observations to contribute to the determination of the OFV and in finalizing model

structure.

125

Initially plasma ARM concentrations were assessed using 1‐ and 2‐compartment

models with first order absorption (ADVAN2 and 4) to obtain a suitable structure. The

kRaR for ARM was fixed to 1 /h 139 as the data did not support its estimation. Once a

suitable (2‐compartment) ARM model had been determined, the DHA dataset was

added and modelled simultaneously using a user‐defined linear mammillary model

(ADVAN 5). For DHA, 1‐ and 2‐ compartments were assessed and the conversion of

ARM to DHA was considered complete for identifiability purposes. Therefore, all

clearance and volume terms for DHA are relative to ART bioavailability (FRARTR) as well as

the degree of metabolic conversion from ART (FRmet‐DHAR). The term F*RDHAR (representing

FRARTR x FRmet‐DHAR) will be used for simplicity.

Once model structure was established, IIV, IOV and their correlations were estimated.

Relationships between model parameters and the covariates age, sex, baseline

parasitaemia and baseline haemoglobin were identified using correlation plots and

subsequently evaluated within NONMEM. Inclusion of the covariate relationship

required a decrease in OFV ≥6.63 (2 distribution with 1 d.f., P<0.01) accompanied by a

decrease in the IIV of that parameter.


A bootstrap using Perl speaks NONMEM (PSN) with 1,000 samples was performed and

the parameters derived from this analysis summarized as median and 2.5th and 97.5th

centiles (95% empirical CI) to facilitate evaluation of final model parameter estimates.

Runs were included in the bootstrap analysis regardless of their minimization status. In

addition, VPCs were performed with 1,000 datasets simulated from the final models.

The observed 10th, 50th and 90th percentiles were plotted with their respective

simulated 95% CI to assess the predictive performance of the model. For the

ARM/DHA model, the observed fraction of BLQ observations was compared with the

median and 95% PI of BLQ observations from these simulated datasets.382

The applicability of the final population models to younger patients from the present

sample was assessed using a NPC. Day 7 plasma LUM concentrations from a previous

study21 from children aged 0.5‐5 years were compared with simulated data from the

126

final models. The actual and simulated number of data points above and below the

20%, 40%, 60%, 80%, 90% and 95% simulated PI were compared.


Changes in Hb, glucose and audiometric data over time were assessed using the

Wilcoxon signed‐rank test. The AUCR0‐∞Rs of DBL and LUM were compared between

subjects with or without recurrent parasitaemia using the Mann–Whitney U test. A

two‐tailed level of significance of 0.05 was considered significant for all comparisons.

127

4.2.4 53B53BResults

4.2.4.1 114B114BClinicalcharacteristicsandcourse

The baseline characteristics of the 13 recruited children are summarized in Table 4‐1.

Eleven had a mono‐infection (9 P. falciparum, 2 P. malariae) on confirmatory expert

microscopy, while 2 had a mixed P. falciparum/vivax infection. AL treatment was well

tolerated and reported symptoms were mild/moderate, short‐lived (<3 days) and

consistent with clinical features of uncomplicated malaria. Initial fever and parasite

clearance were <48 h in all cases.

By 28 days of follow‐up, three children had developed slide‐positive P. vivax (two had

P. vivax at enrolment) and two children had developed P. falciparum (one had P.

falciparum at enrolment). By 42 days of follow‐up, five children had been diagnosed

with P. vivax (two had P. vivax at enrolment) and three with P. falciparum (two had P.

falciparum at enrolment). These data are consistent with the PCR uncorrected results

of a previous larger comparative treatment trial in younger children performed at the

same location.21 The recurrent P. vivax parasitaemia could have resulted from i)

recrudescent infection in those infected with this parasite before treatment, ii)

acquisition of a new P. vivax infection after treatment or, since no primaquine therapy

was administered, iii) appearance of P. vivax from hypnozoites present in the liver at

study entry. For P. falciparum parasitaemia detected during follow‐up, this could have

Table 4‐1 Baseline characteristics of study participants. Data are number (%), mean ± SD or median and [inter‐quartile range].

Children n=13

Age (years) 7.7 ± 1.4

Sex (% male) 8 (62%)

Weight (kg) 19.0 ± 3.5

Height (cm) 112 ± 9

Axillary temperature (°C) 36.8 ± 1.0

P. falciparum parasitaemia 9 (69%)

P. falciparum/vivax parasitaemia 2 (15%)

P. malariae parasitaemia 2 (15%)

Respiratory rate (/min) 28 ± 9

Supine pulse rate (/min) 102 ± 16

Mean upper arm circumference (cm) 16 ± 1

Haemoglobin (g/l) 8.9 ± 1.6

128

represented recrudescence or re‐infection.

The mean Hb concentration was significantly higher on day 28 compared to enrolment

(10.7 vs. 8.9 g/l, P<0.01). There was no significant change in blood glucose over the

first three days of enrolment or audiometric findings over 28 days (data not shown).


LUM and DBL plasma concentration‐time curves are shown in Figure 4‐1. A 3‐

compartment model proved superior to a 2‐compartment model for LUM with a lower

OFV and reduced bias in the WRES plot. The addition of two compartments and the

inclusion of FP metabolism provided the best model once the DBL dataset had been

added. Therefore, the final model comprised 3 compartments for LUM and 2

compartments for DBL. The structural model parameters were kRaR, VRCR /FRLUMR, VRP1R /FRLUMR,

VRP2R /FRLUMR, CL/FRLUMR, QR1R/FRLUMR, QR2R/FRLUMR, FP (percentage contribution of FP metabolism to

DBL metabolic conversion), VRCR/F*RDBLR, VRPR /F*RDBLR, CL/F*RDBLR, Q/F*RDBLR. IIV was able to be

estimated for kRaR, CL/FRLUMR, CRLR/F*RDBLR, VRCR/F*RDBLR and FRLUMR as well as inter‐occasion

variability for FRLUMR (the population value of FRLUMR remained fixed to 1). Variability in

FRLUMR was smaller between individuals than it was between doses in the same

individual (20 vs. 67%). Once IIV and IOV terms were added, inspection of the

Time (h)

0 200 400 600 800

Pla

sm

a lu

me

fan

trin

e o

r d

es

bu

tyl-

lum

efa

ntr

ine

(

g/l)

1

10

100

1000

10000

Figure 4‐1 Time‐concentration plots showing LUM (○) and DBL () in μg/l on log10 scale. Curves of the median concentration for LUM (solid black line) and DBL (dashed black line) are also shown.

129

WRES plot revealed a bias due to the absorption profile of the final dose. Estimation of

a separate kRaR for the 6th (final dose) (kRaD6R) improved the bias and reduced the OFV (‐

7.519 P<0.01). None of the covariates tested improved the model. Residual

unexplained variability (20.8% and 20.9% for LUM and DBL, respectively) was low.

The final model parameter estimates and the bootstrap results are summarized in

Table 4‐2. Bias was <10% for structural and random model parameters. Figure 4‐2and

Figure 4‐3 show GOF plots and VPCs, respectively. The half‐lives and AUC of LUM and

Table 4‐2 Final population pharmacokinetic estimates and bootstrap results for lumefantrine and desbutyl‐lumefantrine.

Parameter Final model

estimate (RSE%)a

Bootstrap

median [95% CI]

Structural model parameters

kRaR (/h) 0.461 (20) 0.442 [0.285‐0.644]

CL/FRLUMR (l/h/70kg) 7.29 (9) 7.21 [5.55‐9.04]

VRCR/FRLUMR (l/70kg) 227 (12) 225 [147‐284]

QR1R/FRLUMR (l/h/70kg) 1.52 (16) 1.57 [0.96‐2.32]

VRP1R/FRLUMR (l/70kg) 115 (19) 109 [57‐214]

QR2R/FRLUMR (l/h/70kg) 0.743 (13) 0.805 [0.208‐1.27]

VRP2R/FRLUMR (l/70kg) 164 (8) 168 [97‐240]

kaRD6R (/h) 1.20 (52) 1.14 [0.50‐3.68]

FP (%) 6.29 (15) 6.45 [4.36‐9.84]

CL/F*RDBLR (l/h/70kg) 701 (10) 694 [561‐851]

VRCR/F*RDBLR (l/70kg) 51,100 (10) 51,200 [42,200‐61,430]

Q/F*RDBLR (l/h/70kg) 439 (19) 424 [305‐632]

VRPR/F*RDBLR (l/70kg) 68,400 (14) 68,000 [51,800‐88,600]


IIV in FRLUMR (%) 19.8 (42) 18.9 [2.5‐29.3]

IIV in ka (%) 55.4 (44) 55.8 [17.1‐92.2]

IIV in CL/FRLUMR (%) 17.7 (20) 16.9 [6.8‐23.7]

IIV in CL/F*RDBLR (%) 26.2 (26) 26.0 [10.4‐37.6]

IIV in VRCR/F*RDBLR (%) 34.1 (22) 33.3 [17.4‐47.8]

IOV in FRLUMR (%) 67.0 (9) 66.4 [53.4‐77.7]

RUV for LUM (%) 20.8 (7) 20.3 [17.4‐22.5]

RUV for DBL (%) 20.9 (7) 20.6 [17.5‐23.0]

a RSE% are the NONMEM produced values from the covariance step. OFV in the final model:‐586.510, bootstrap OFV (median [95% CI]):‐601.901 [‐668.687‐ ‐559.564]

130

Predicted plasma lumefantrine (g/l)

10 100 1000 10000

Ob

se

rve

d p

las

ma

lum

efa

ntr

ine

(

g/l)

10

100

1000

10000

Predicted plasma desbutyl-lumefantrine (g/l)

0.1 1 10 100Ob

se

rve

d p

las

ma

de

sb

uty

l-lu

me

fan

trin

e (

g/l)

0.1

1

10

100A B

Figure 4‐2 Population (○) and individual predicted (●) versus observed data for LUM (A) and DBL (B) concentrations (µg/l) for the final model. The lines of identity are also shown.

DBL are shown in Table 4‐4. The first distribution, second distribution and terminal

elimination half‐lives for LUM had median values of 10.4, 46.6 and 123 h, while DBL

had a median distribution tR½R of 19.7 h and a median terminal elimination tR½R of 141 h.

Overall the metabolite to parent drug ratio was 1.13 % (obtained from AUCR0‐∞R) but

there was a higher ratio at later time‐points. Day 7 LUM concentrations obtained from

younger children were consistent with prediction based on the final model with an

expected number of observations above and below the 20, 40, 60, 80, 90 and 95%

simulated PIs. When the same data for DBL were compared, there was an excess of

points above the 20, 40, 60 80 and 90% PIs and a lack of points below the 20 and 40%

PI, especially at a younger age, demonstrating that the day 7 DBL concentrations in the

younger children were higher than expected from the model.

Time (h)

0 100 200 300 400 500 600 700

Lu

me

fan

trin

e (

g/li

ter)

10

100

1000

10000

A

Time (h)

0 100 200 300 400 500 600 700

De

sb

uty

l-lu

me

fan

trin

e (

g/li

ter)

1

10

100 B

Figure 4‐3 Visual predictive check showing observed 50th (●), 10th () and 90th (○) percen les with the simulated 95% CI for the 50th (solid black line), 10th (grey dotted lines) and 90th (dashed grey lines) percentiles for LUM (A) and DBL (B) concentrations (μg/l on log10 scale) from the final model.

131

Initial modelling of ARM/DHA datasets proved difficult given the large proportion of

BLQ data (45% and 12% for ARM and DHA respectively). Once these data were

incorporated into the model using the method ‘M3’ in Ahn et al. 297, more acceptable

models were obtained. The dispositions of ARM and DHA were best described by a 2‐

compartment model for ARM and a 1‐compartment model for DHA. The structural

model parameters were kRaR, VRCR/FRARMR, VRPR/FRARMR,R RCL/FRARMR, Q/FRARMR, VRCR/F*RDHAR and

CL/F*RDHAR. As with LUM the IIV and IOV of FRARMR was estimated and variability between

doses was larger than between individuals (84.1 vs. 38.1 %). The IIV of CLRARMR was also

estimated. A relationship between CLRARMR and dose number was included and

demonstrated that for each subsequent dose of ARM CLRARMR increased by 67.8%

relative to its value after the first dose. This relationship was accompanied by a

decrease in the OFV (‐82.774, P<0.001) and a reduction in the RUV of both ARM and

DHA. No other covariate relationship improved the model. After the inclusion of

IIV/IOV terms and the covariate relationship, RUV was still high at 51.6% and 53.3% for

ARM and DHA, respectively.

Table 4‐3 Final population pharmacokinetic estimates and bootstrap results for ARM and DHA.

Parameter Final model

estimate (RSE%)a

Bootstrap

Median [95% CI]

Structural model parameters

CL/FRARMR (l/h/70kg) 102 (27) 96.3 [57.0‐167.0]

VRCR/FRARMR (l/70kg) 193 (62) 172 [40‐506]

Q/FRARMR (l/h/70kg) 49.6 (47) 45.8 [19.7‐111.1]

VRPR/FRARMR (l/70kg) 1070 (59) 1220 [593‐3011]

kRaR (/h) 1 [FIXED] 1 [1‐1]

VRCR/F*RDHAR (l/70kg) 440 (40) 417 [69‐826]

CL/F*RDHAR (l/h/70kg) 277 (26) 275 [140‐443]

% increase in CL/FRARMR for each subsequent dose (%)

67.8 (31) 73.3 [40.5‐125]


IIV in FRARMR (%) 38.1 (72) 19.7 [0.3‐58.6]

IOV in FRARMR (%) 84.1 (38) 84.2 [52.2‐113.6]

IIV in CL/FRARMR (%) 84.0 (33) 75.8 [49.1‐108.2]

RUV for ARM (%) 51.6 (12) 50 [37‐61]

RUV for DHA (%) 53.3 (20) 61 [42‐83] a RSE% are derived from the bootstrap. OFV in the final model: 259.853, bootstrap OFV (median [95% CI]):‐ 177.255[77.606‐249.014]

132

The final model parameter estimates and the bootstrap results are summarized in

Table 4‐3. As the covariance step was not successful, NONMEM‐derived RSE could not

be obtained. Bias was <11% for structural and random parameters except IIV for FRARMR

which had a negative 48% bias. Figure 4‐4 and Figure 4‐5 show GOF plots and VPCs,

respectively. The VPCs show all observed 10th, 50th and 90th percentiles within their

simulated 95% CI and the fraction of BLQ data at each time point within its 95% CI for

both ARM and DHA. Secondary parameters for study participants are shown in Table

4‐4. The AUCR0‐∞R and half‐lives of ARM decreased with each dose while the median

DHA to ARM ratio increased.

Predicted plasma artemether (g/l)

1 10 100

Ob

se

rve

d p

las

ma

art

em

eth

er

( g

/l)

1

10

100

Predicted plasma dihydroartemisinin (g/l)

1 10 100

Ob

se

rve

d p

las

ma

dih

yd

roa

rte

mis

inin

(

g/l)

1

10

100

A B

Figure 4‐4 Population (○) and individual predicted (●) versus observed data for ARM (A) and DHA (B) concentrations (µg/l) for the final model. The lines of identity are also shown. The grey dashed line represents the LOQ of ARM in (A) and DHA in (B).

Time (h)

0 10 20 30 40 50 60 70

Fra

ctio

n B

LQ

0.0

0.2

0.4

0.6

0.8

1.0

Pla

sma

art

emis

inin

( g

/l)

0.1

1

10

100

A

Time (h)

0 10 20 30 40 50 60 70

Fra

ctio

n B

LQ

0.0

0.2

0.4

1.0

Pla

sma

dih

yd

roa

rte

mis

inin

( g

/l)

1

10

100

B

Figure 4‐5 Visual predictive check showing observed 50th (●), 10th () and 90th (○) percen les with the simulated 95% CI for the 50th (solid black line), 10th (grey dotted lines) and 90th (dashed grey lines) percentiles for ARM (A) and DHA (B) concentrations (μg/l on log10 scale) from the final model. The fraction of BLQ observations from the data (○ connected with a do ed black line) with the simulated 95% prediction interval are also shown for both ARM and DHA.

133

Table 4‐4 Secondary pharmacokinetic parameters derived from post hoc Bayesian estimates for study participants. Data are median [inter‐quartile range].

Parameter LUM DBL ARM – Dose 1 ARM – Dose 6 ARM – All doses DHA

tR½αR

aR R(h)

10.4

[10.3 – 11.8]

19.7

[18.4 – 22.5]

0.62

[0.60 – 0.64]

0.16

[0.12 – 0.33]

0.80

[0.76 – 0.82]

tR½βR

a (h) 46.6

[44.8 – 48.2]

141

[135 – 150]

16.4

[15.7 – 16.8]

11.9

[11.2 – 13.2]

tR½γR

a (h) 123

[120 – 127]

AUCR0–∞ R(µg∙h/l)b 459,980

[391,330 – 632,730]

5,434

[4,394 – 8,542]

983

[371 – 1,770]

164

[145 – 254]

3,063

[2,357 – 4,513]

2,839

[1,812 – 3,488]

AUCRMETABOLITER/AUCRPARENTR (%) 1.13

[0.93 – 1.55]

36.8

[36.8 – 36.8]

186

[91.8 – 268]

92.7

[59.2 – 94.3]

atR½αR, tR½βR and tR½γR are the first distribution, second distribution and terminal elimination half‐lives respectively for LUM, while for DBL and ARM tR½αR and tR½βR represent the distribution and terminal elimination half‐life respectively and for DHA tR½αR represents the terminal elimination half‐life. brepresents either the AUCR0–∞R for all six doses together or the AUCR0–∞R for individual doses as if they were given alone.

134

4.2.4.3 116B116BRelationshipbetweendrugexposureandtreatment

outcome

The LUM AUCR0‐∞R tended to be lower in children with recurrent parasitaemia on days

28 (n=5, P=0.057) and 42 (n=8, P=0.086), but this was not the case for DBL (P=0.46 and

0.89, respectively). However, a combined AUCR0‐∞R with DBL weighted four times more

than LUM, consistent with its greater antimalarial potency in vitro,160, 163, 377, 378 was

significantly lower in children with recurrent parasitaemia on day 28 (n=5, P=0.028)

and of borderline significance on day 42 (n=8, P=0.063).

135


In the present study of PNG children with uncomplicated malaria treated with a

conventional AL regimen, rich datasets of plasma concentrations of LUM, ARM and

their active metabolites measured during an extended follow‐up period were

successfully analysed using population PK modelling that allowed for a high proportion

of BLQ plasma ARM and DHA concentrations. Our analyses included the first

compartmental PK analysis of plasma DBL concentrations. We found that current dose

recommendations for AL in children result in a LUM AUC similar to that achieved in

adults, despite children receiving a higher average mg/kg dose relative to a 50 kg adult.

However, the subgroup of children weighing 12.5‐15 kg receive the lowest mg/kg dose

and may be at risk of under‐dosing.

Three studies, all from Africa, have examined LUM PK after AL treatment in children.

The first and simplest compared crushed tablets and a dispersible formulation using a

pooled analysis of single blood sample taken at one of six time‐points during a 14‐day

period from 726 children aged <12 years.183 The LUM AUC for both formulations was

higher than in the present study (574,000 and 636,000 vs. 459,980 µg∙h/l). In the

second study,144 six blood samples were taken from children aged 5‐13 years starting

when the last AL dose was given and the LUM AUCR60‐∞R was calculated using non‐

compartmental analysis. When we used our final models to generate an AUCR60‐∞R, this

was higher (257,010 vs. 210,000 µg h/l). Based on their data, the authors reported that

children have lower levels of exposure to LUM than adults using recommended AL

dose schedules.144 A third study of children aged 1‐10 years utilized a population

approach142 but there was no sampling beyond 72 h and no secondary PK parameters

were provided. A comparison with LUM disposition in the present study was,

therefore, not possible.

Comparisons of LUM AUC between studies in adults are also difficult as some report

AUC from the first dose while others use AUCR60‐∞R.Table 4‐5 summarizes the available

data for both measures of drug exposure. There is a difference between LUM exposure

in healthy adults and subjects with malaria, but the AUCs for non‐pregnant adults,

136

pregnant adults and children with malaria are similar. Current AL dose

recommendations for children ensure that those weighing 15‐35 kg receive a 35%

higher average mg/kg dose than a 50 kg adult, but those weighing 12.5‐15 kg receive a

lower mg/kg dose. The AUC data support the higher average mg/kg dose in children

and suggest that those weighing 12.5 ‐ 15 kg should receive 2 tablets rather than 1 to

avoid under‐dosing while not exceeding the highest recommended mg/kg dose (Figure

4‐6). As LUM exposure, measured either as AUC or day 7 concentrations, has

previously been shown to be a prime determinate of efficacy,141, 148 it is important that

under‐dosing is avoided.

The three studies of AL in children also measured plasma ARM/DHA concentrations.142,

144, 183 The first was not able to calculate AUCs from pooled concentration data due to a

sparse sampling schedule.183 The second employed a limited sampling schedule

starting from the last AL dose,144 and the AUCs were therefore lower than those of the

present study (168 vs. 217 µg∙h/l for ARM and 382 vs. 402 µg∙h/l for DHA). The

population approach used in the third study142 produced a similar model of the

disposition of ARM (two compartments) and DHA (one compartment) to that of the

present study. The authors reported a similar increase in CL/FRARM Rwith each dose (57%

vs. 67.8 % in our children) and a higher RUV (61% vs. 51.6% and 82 % vs. 53.3% for

ARM and DHA respectively), the latter observation likely a reflection of the fact that

many plasma concentrations were close to or below the LOQ. As no secondary PK

parameters were provided, a comparison of AUCs could not be performed. However

the half‐lives of ARM, estimated from the PK parameters provided, were longer than

those in our children (0.89 vs. 0.62 h and 32.0 vs. 16.4 h for distribution and

Table 4‐5 Summary of studies reporting area under the plasma concentration‐time curve (AUC) for lumefantrine.

Sample AUCR60–∞/t R(µg∙h/l)a AUCR0–∞/t R(µg∙h/l)a

Healthy adults 383,000‐456,000 135, 137 1,242,000‐2,730,000 135, 374b

Non‐pregnant adults with malaria ‐ 335,000‐758,000 134, 164, 184, 376

Pregnant women with malaria 252,000 143 472,000 182

Children with malaria 210,000 142 572,000‐636,000 183c

Present study 257,000 459,980 aAUC was either a median or mean and was reported either to the last data point or to ∞, bas subjects in Bindschedler et al (10) only received a single dose, the reported AUC has been multiplied by six, cthis study used a pooled approach from single observations in each subject to calculate AUC

137

elimination, respectively, of the first dose), while the elimination tR½R of DHA was

shorter (0.38 vs. 0.80 h).

The AUCs for ARM and DHA in the present study were similar to those reported

previously in adults with malaria133, 143 but higher than those in healthy adults.135, 137,

138 Our terminal elimination tR½R for ARM was longer than those reported in these

studies (16.4 vs. 1.5‐3.9 h) while for DHA it was shorter (0.80 vs. 1.2‐2.1 h). The adult

studies used non‐compartmental methods to determine these half‐lives and this may

account for the differences. Nevertheless, based on these comparisons, exposure to

ARM and DHA in children is adequate with current AL dose recommendations.

Few studies have evaluated the disposition of DBL, an active metabolite of LUM. Our

DBL:LUM ratio (1.13%) falls between values reported in previous treatment studies

(0.33% and 5.2%).143, 184 The lower value (0.33%) was from a study of non‐immune

Columbian adults with malaria that sampled to 168 h and reported AUCR0‐168R. The

higher value (5.2%) was from a study of pregnant Thai women with malaria in which

sampling started after the last dose and AUCR60‐∞R was reported. The difference

between these values can, at least in part, be explained by the study designs as the

metabolite‐to‐parent percentage calculated from AUCR60‐∞R in the present study is more

Figure 4‐6 The doses of lumefantrine and artemether in mg/kg given to children 5‐35 kg under current (solid black line) and suggested (dashed grey line) dosing regimens. The horizontal dotted black line represents the dose in mg/kg recommended for a 50 kg adult.

Weight (kg)

5 15 25 35

Lu

mef

an

trin

e d

os

e (

mg

/kg

)

0

6

12

18

24

Weight (kg)

5 15 25 35

Arte

meth

er d

ose

(mg

/kg

)

0

1

2

3

4

138

than double for AUCR0‐168R (1.96 vs. 0.76 %). However it is likely that ethnicity and

pregnancy contribute to the difference. Age may also influence metabolic conversion

of LUM to DBL as our PK model was able to predict concentrations of LUM but not DBL

in younger children effectively. It is uncertain as to whether malaria itself also

influences the ratio since it was 0.45%, within the range of studies of malaria, after a

single dose of AL in 22 healthy adults.384

As reported previously143 DBL had a longer terminal elimination tR½R than LUM in the

present study (141 vs. 123 h, P <0.001) and therefore the DBL:LUM ratio will increase

with time. Although the ratios found in available studies are low, the in vitro potency

of DBL is between 2.2 and 7.2 times that of LUM160, 163, 377, 378 and it may therefore

contribute to therapeutic outcome. We found a combined weighted LUM/DBL AUC

was more likely to be lower than the AUC of either LUM or DBL alone in subjects with

recurrent parasitaemia at days 28 and 42. This supports the suggestion that DBL may

influence AL treatment outcome.163

Although the variable bioavailability of ARM and LUM has been previously reported,164

it has not previously quantified in children. Given the significant increase in fed vs.

fasted healthy volunteers134 it is recommended that AL is administered with fat in

order to improve absorption. Based on a study in healthy adults who received a single

dose of AL, 1.2 g of fat (equivalent to 35 ml of full cream milk) is required to achieve

90% of maximal LUM bioavailability.375 Although these results may not be directly

applicable to the children with malaria in our study, they ingested 2 g of fat with each

dose and there was still significant between‐dose variability in the bioavailability of

both LUM (67.0%) and ARM (84.1%). We were unable to identify factors that may be

responsible for these observations.

In the analysis of the ARM/DHA dataset, there were a significant number of BLQ

plasma concentrations. This is an issue encountered in PK analyses of a variety of other

antimalarial drugs.142, 226, 385 Traditional approaches to this problem such as excluding

BLQ data from the analysis or setting them to a specific value (such as zero or 50% of

the LOQ) have been shown to bias the PK parameters even when only 10% of the data

are BLQ.297, 382, 383, 386 Our approach was to use a method within NONMEM shown to

139

have little bias in situations with up to 40% BLQ data in population analysis.383 This

method treats BLQ data points as categorical data and maximizes the that its value is

truly below the LOQ.297 Although the implementation of this method has previously

been difficult and time consuming, changes to NONMEM and more efficient data

processing have increased its accessibility. The benefits of this method demonstrated

in relatively simple models are likely to apply to more complex models with parent

drug and metabolite. We were unable to obtain RSE for our parameters in this model

as the covariance step was unsuccessful, a common problem when this method is

used.382, 383 However this does not impact on the reliability of the results obtained and

other methods of model evaluation (such as bootstrap and VPC) can still be used.

Our novel data relating to DBL PK and its favourable pharmacodynamic effects suggest

that future efficacy and PK studies of LUM should include DBL assay to further

elucidate its role. We have also shown that analytical techniques that utilize BLQ data

to refine PK parameter estimates can be applied in this situation. Extended sampling

and a population PK approach allow flexibility in deriving secondary parameters, an

important consideration when comparisons with published non‐standard measures

such as time‐limited AUC are of interest. Our data confirm that current AL dose

recommendations produce similar ARM, DHA and LUM exposure in children to that in

adults with malaria. However, smaller children weighting 12.5‐15 kg are at risk of

under‐dosing and AL doses could be doubled without exceeding the current weight‐

based maximum mg/kg dose in this patient group.

140


We thank the children and their parents/guardians for their participation. We are

most grateful to Sr Valsi Kurian and the staff of Alexishafen Health Centre for their kind

co‐operation during the study. We also thank Jovitha Lammey, Christine Kalopo and

Bernard (“Ben”) Maamu for clinical and/or logistic assistance, and Harin Karunajeewa

for assistance with protocol design. The National Health and Medical Research Council

(NHMRC) of Australia funded the study (grant #634343). TMED is supported by an

NHMRC Practitioner Fellowship.

141

142

143

5 4B4BAPharmacokineticComparisonofTwo

Piperaquine‐ContainingArtemisinin

CombinationTherapiesinPapuaNewGuinean

ChildrenwithUncomplicatedMalaria


The aims of this study presented in this chapter were twofold. Firstly to compare the

PK of ART/PQ base with that of DHA/PQ tetraphosphate and secondly to provide

preliminary data on the safety, tolerability and efficacy of ART/PQ in children.

Although ART/PQ base was being marketed and sold as a treatment for children in a

number of countries at the time of this study, the literature was lacking any

pharmacological data of the combination. Meanwhile, the PK of a closely related

combination, DHA/PQ tetraphosphate, had been previously described in PNG

children.387 There was a need to evaluate ART/PQ base in children to determine if it

had similar properties to DHA/PQ tetraphosphate. This study was conceived and

performed in a similar manner to the study of DHA/PQ tetraphosphate to enable PK

comparisons between the two regimens. It also allowed collection of preliminary

safety, tolerability and efficacy data for ART/PQ.

This study resulted in the publication4 presented in this chapter. Entitled, “A

pharmacokinetic comparison of two piperaquine‐containing artemisinin combination

therapies in Papua New Guinean children with uncomplicated malaria” was published

by the journal Antimicrobial Agents and Chemotherapy (2012. 56(6):p. 3288‐97). The


of ethical approvals and supporting funding. It has been reformatted to conform to

thesis requirements set by the University of Western Australia. The references have

been combined with those for the thesis as a whole and can be found in section x

below.

144

145


Sam Salman,A Madhu Page‐Sharp,B Kevin T Batty,BR RKay Kose,C Susan Griffin,C Peter

Siba,C Kenneth F. Ilett,A Ivo Mueller,C Timothy M. E. DavisA.

ASchool of Medicine and Pharmacology, University of Western Australia, Fremantle

Hospital, Fremantle, Western Australia, Australia;

BSchool of Pharmacy, Curtin University of Technology, Bentley, Australia;

CPapua New Guinea Institute of Medical Research, Madang, Papua New Guinea.


PK differences between PQ base and PQ tetraphosphate were investigated in 34 Papua

New Guinean children aged 5‐10 years treated for uncomplicated malaria with ART/PQ

base or DHA/PQ tetraphosphate. Twelve children received ART/PQ base (two daily

3:18 mg/kg doses as granules) as recommended by the manufacturer with regular

clinical assessment and blood sampling over 56 days. Plasma PQ concentrations from

22 children with malaria of similar age from a previously‐published PK study of

DHA/PQ tetraphosphate (three daily 2.5:20 mg/kg doses as tablets) were available for

comparison. The disposition of ART was also assessed in the 12 children who received

ART/PQ base. Plasma PQ was assayed by HPLC‐UV detection and ART using liquid

chromatography‐mass spectrometry. Multi‐compartment PK models for PQ and ART

were developed using a population‐based approach. ART/PQ base was well tolerated

and initial fever and parasite clearance were prompt. There were no differences in PQ

AUCR0–∞R between the two treatments with medians of 49,451 (n=12) and 44,556 (n=22)

µg∙h/l for ART/PQ base and DHA/PQ tetraphosphate, respectively. Recurrent

parasitaemia was associated with lower PQ exposure. Using a two‐compartment ART

model, the median AUCR0–∞R was 1,652 µg∙h/l. There was evidence of auto‐induction of

ART metabolism (relative bioavailability for the second dose 0.27). These and

previously‐published data suggest that a three‐day ART/PQ base regimen should be

further evaluated, in line with WHO recommendations for all ACTs.

146


The most recent WHO recommendations for the treatment of uncomplicated malaria

include a three‐day course of DHA/PQ as a first‐line ACT.388 Various formulations of

DHA/PQ are marketed in tropical countries (Duo‐cotecxin, Combimal and P‐Alaxin)389

or are in development (Eurartesim) ,390 all of which employ PQ tetraphosphate as the

DHA partner drug. DHA is a semi‐synthetic derivative of ART and its production adds to

the manufacturing cost but, unlike ART, it does not exhibit auto‐induction of

metabolism. In addition, although the tetraphosphate salt of PQ has greater water

solubility and therefore may have better oral bioavailability, incorporation of the lipid‐

soluble PQ base should also simplify production.

Artequick (Artepharm Co Ltd, Guangzhou, China) is an ACT that contains ART in place

of DHA and PQ base rather than PQ tetraphosphate. This combination is formulated as

tablets but also as granules for paediatric use. It is marketed in Cambodia and some

sub‐Saharan African countries. The current manufacturer’s recommendation is for

Artequick to be given as a two‐day regimen391 which contrasts with the three days

recommended for all ACTs by the WHO.388 Although the tolerability, safety, efficacy

and PK properties of DHQ/PQ tetraphosphate have been widely investigated in

children and adults,174, 178, 202, 392, 393 there are limited data relating to the efficacy and

tolerability of ART/PQ base211, 212 and no studies of the PK of this novel combination in

malaria‐infected patients. Concerns have been raised regarding possible under‐dosing

of a number of antimalarial drugs in children144, 226 including PQ.202, 392, 393 Although

children have been included in studies of PQ PK,174, 178 only one PK study of ART has

specifically enrolled paediatric patients.125

We have evaluated the population PK of ART/PQ base (Artequick) in children from PNG

with uncomplicated malaria and compared the data with those of a previously

published study of DHQ/PQ tetraphosphate (Duo‐cotecxin; (Beijing Holley‐Cotec,

Beijing, China) in the same category of patients.387 The primary aims of the present

study were to investigate PK differences between PQ base and PQ tetraphosphate, and

to describe the population PK of ART in PNG children. Secondary aims were to provide

preliminary data relating PK factors to recurrent parasitaemia, and to use both PK and

efficacy data to suggest improved dose regimens for these combinations.

147


5.2.3.1 117B117BPatients

Assessment and recruitment of children for the present and published DHQ/PQ

tetraphosphate studies were as described previously.387 Briefly, all subjects were

children aged 5‐10 years presenting to Alexishafen Health Centre, Madang Province on

the north coast of PNG. The clinic serves an area where Plasmodium falciparum and P.

vivax are hyperendemic, and P. ovale and P. malariae are also transmitted.20 Children

with an axillary temperature >37.5°C or a history of fever in the previous 24 h were

screened with a Giemsa‐stained thick blood film read on‐site by a trained microscopist.

Those with a mono‐infection of P. falciparum (>1,000 asexual parasites µl‐1), or P.

vivax, ovale or malariae (>250 asexual parasites µl‐1) were eligible provided that the

child’s parents gave informed consent, there were no features of severe malaria,379

they had not taken any antimalarial drug in the previous 14 days, there was no

evidence of another cause of fever, and there were no features of malnutrition or

other chronic co‐morbidity. Although the location, population and enrolment

procedures used in the two studies were identical, the DHQ/PQ group was enrolled

between August 2005 and January 2006 while the ART/PQ base group was enrolled

from March 2008 to May 2008.The study was approved by the PNG Institute of

Medical Research Institutional Review Board and the Medical Research Advisory

Committee of the PNG Department of Health.

5.2.3.2 118B118BClinicalmethods

In the present study of ART/PQ base, a standardized history was taken and a clinical

examination was performed. A 3 ml venous blood sample was taken for baseline blood

film microscopy, Hb and blood glucose, and for subsequent drug assay of separated

plasma. Each child treated with granules of ART/PQ base (Artequick) according to WT

(approximately 3:18 mg/kg/day respectively). This dose was repeated at 24 h, as

recommended by the manufacturer, with the exact time of each dose recorded. All

doses were given under direct observation. The full contents of each sachet were

mixed with at least 50 ml of cow’s milk (equivalent to 2 g of fat), as fat has been

reported to increase the bioavailability of PQ tetraphosphate.173, 215 The volume of milk

used was based on previous experience with its palatability and association with

148

nausea in PNG children, as well as the amount of fat found to maximize the absorption

of LUM, another highly lipophilic antimalarial drug, in healthy adults.375

Further venous blood samples were taken from an indwelling intravenous catheter at

1, 2, 4, 12, 24, 28, 36 and 48 h, and then by venesection on days 3, 5, 7, 14, 28, 42 and

56. All samples were centrifuged promptly and RBCs and separated plasma stored

frozen at ‐80°C until assayed. Detailed clinical assessment, including a symptom

questionnaire, blood film, Hb and blood glucose, was repeated on days 1, 2, 3 and 7,

with additional clinical assessment and blood films on days 14, 28, 42 and 56. All blood

smears taken at baseline and during follow‐up were examined independently by at

least two skilled microscopists in a central laboratory. Each microscopist viewed 100

fields at 1,000x magnification before a slide was considered negative. Any slide

discrepant for positivity/negativity or speciation was referred to a third microscopist

for adjudication.

The clinical procedures followed for the DHQ/PQ group have been previously

described387 and were similar to those of the ART/PQ base group. Differences included

i) administration of three days of DHQ/PQ tetraphosphate tablets at a dose of 2.5:20

mg/kg daily (equivalent to 11.5 mg/kg of PQ base daily), ii) drug administration with

water, and iii) blood sampling and clinical follow up to 42 days only.

5.2.3.3 119B119BLaboratorymethods

PQ tetraphosphate reference standard was obtained from Yick‐Vic Chemicals and

Pharmaceuticals, Ltd. (Hong Kong, China); CQ diphosphate and authentic ART were

from Sigma‐Aldrich (St. Louis, USA), and ARM from AAPIN Chemicals Ltd (Abingdon,

UK). Solid phase extraction (SPE) Bond Elut ® PH columns were purchased from Varian

Inc. (Palo Alto, USA). HPLC grade methanol was obtained from Merck Pty Ltd (Kilsyth,

Australia) and LC‐MS grade ammonium formate was from Sigma‐Aldrich (Gillingham,

UK). All other solvents and chemicals were of analytical grade.

For the ART/PQ base group, PQ in plasma was analysed by HPLC as for the original

DHQ/PQ group387 with minor modifications. Briefly, plasma was spiked with CQ as an

internal standard, alkalinized, and extracted into 8 ml of hexane‐isoamyl alcohol (99:1).

149

Baseline samples were assayed for CQ prior to quantification of PQ to ensure no

interference with the internal standard. After centrifugation, the supernatant was

back‐extracted into 100 µl of 0.1 M HCl, aspirated and re‐centrifuged. Aliquots of 80 µl

were injected onto on a Phenomenex CR6R‐phenyl column (Phenomenex, Torrance, CA)

with a mobile phase of 11% acetonitrile in 0.1 M phosphate buffer (pH 2.5) pumped at

1 ml/m. Retention times were 2.5 and 7.3 min for PQ and CQ, respectively and were

detected at 340 nm. The linear assay range was 2‐1,000 µg/l and the intra‐day relative

standard deviations (RSDs) were 10.8%, 8.2%, and 9.4%, and the inter‐day RSDs were

11.6%, 4.4%, and 6.7% at 5, 100, and 1,000 µg/l, respectively. The limits of

quantification and detection were 2 µg/l and 1 µg/l, respectively.

For ART, the extraction procedure used a 1 ml CR18R SPE column as previously

described,380 with following modifications. Briefly, the SPE column was pre‐

conditioned with 1 ml of methanol followed by 1 ml of 1M acetic acid. Plasma samples

(0.5 ml) were spiked with internal standard (ARM, 1,000 µg/l) and loaded onto the pre‐

conditioned SPE column and drawn through using a medium vacuum. The column was

then washed with 1M acetic acid (1 ml x 2), followed by 20% v/v methanol in 1M acetic

acid (1 ml). The column was dried under low vacuum for 30min and retained drugs

were eluted with 2 ml of t‐butyl chloride:ethyl acetate (80:20% v/v). The eluate was

evaporated in a vacuum evaporator at 35°C then reconstituted in 50 µl of mobile

phase and 5 µl aliquots were injected onto the LC‐MS system.

The LC‐MS system used was a single quad mass spectrometer (Model 2020, Shimadzu,

Kyoto, Japan) consisting of a binary pump Mmodel 20AD), vacuum degasser,

thermostated autosampler (model SIL 20ACHT), thermostated column compartment

(Model CTO 20A), photodiode detector (Model SPD M 20A) and mass analyser (Model

MS 2020) with both electrospray ionization (ESI) and atmospheric pressure ionization

(APCI) systems. Analyses were performed in isocratic mode with a mobile phase of

20mM ammonium formate (pH 4.8): methanol (20:80) pumped at a flow rate of 0.2

ml/min. Chromatographic separation was undertaken at 30°C on a Synergy fusion‐RP

CR18R (150 mm x 2.0 mm i.d.) column coupled with a 4 mm x 3 mm i.d., 5 µm particle CR18R

guard column (Phenomenex, Lane Cove, Australia). Retention times were 4.2 min and

7.5 min for ART and ARM respectively. Optimized mass spectra were acquired with an

150

interface voltage of 4.5kV, a detector voltage of 1 kV, a heat block temperature of

400°C and a desolvation gas temperature of 250°C. Nitrogen was used as a nebulizer

gas at a flow rate of 1.5 l/min and dry gas flow of 10 l/min.

Quantification was performed by selected ion monitoring (SIM), using DUIS mode in

which both ACPI and ESI are used simultaneously. All standard curves were linear with

an r2 ≥0.999. Chromatographic data (peak area ratio of ART:ARM) were processed

using the LAB Solution software package (Version 5, Shimadzu, Japan). Responses

from analysis of samples containing three different ART concentrations (5, 200 and

2,000 µg/l) and one ARM concentration (1,000 µg/l) spiked into five separate plasma

samples were used to determine matrix effects (ion suppression/enhancement),

absolute recovery, and process efficiency381 which were between 90 ‐ 98 %, 82‐93%

and 86‐91% respectively. The assay intra‐day RSDs were 9.3, 7.2, and 3.7 % and inter‐

day RSDs were 9.5, 7.1 and 6.5% at 5, 200 and 2,000 µg/l, respectively. The limits of

quantification and detection for ART were 2.5 and 1 µg/l, respectively.


LogReR plasma concentration‐time datasets for PQ and ART were analysed by nonlinear

mixed effects modelling using NONMEM (v 6.2.0, ICON Development Solutions, Ellicott

City, MD, USA) with an Intel Visual FORTRAN 10.0 compiler. The PQ plasma

concentration vs. time data from the published study of DHQ/PQ performed by our

group, which were originally analysed using a patient rather than a population

approach,387 were pooled with the PQ concentration data from the present study. The

FOCE with interaction estimation method was used. OFV and CWRES plots were used

to choose suitable models during the model‐building process. Allometric scaling was

employed a priori, with volume terms multiplied by (WT/70)1.0 and clearance terms by

(WT/70)0.75.394 RUV was estimated as additive error for the log‐transformed data.

Secondary PK parameters including AUCR0–∞R and elimination tR1/2R for the participants

were obtained from post hoc Bayesian prediction in NONMEM using the final model

parameters. Base models were parameterized using kRaR, VRCR/F, CL/F, VRPR/F and Q/F.

For the PQ dataset, two‐ and three‐ compartment models (ADVAN 4 and 12) with first

order absorption with and without lag time were tested. Since inspection of the time

151

concentration curves indicated that there was significant variability in the absorption

phase, a transit compartment model was also tested.287 In this model, the dose passes

through a series of transit compartments before entering the absorption compartment

in order to model the delay often associated with drug absorption. A single rate

constant (kRtrR) describes the entry and exit for all transit compartments. Using a

previously described implementation of the transit compartment model in

NONMEM,287 NN and MTT (equal to (1+NN) / kRtrR) were estimated as continuous

variables. For the ART dataset, 1‐ and 2‐ compartment models (ADVAN 2 and 4) with

first order absorption with and without lag time were evaluated. Once the structure of

the models was established, IIV, IOV, and correlations between IIV terms were

estimated, where supported by the data.

As two different formulations of PQ with different water/lipid solubilities were used,

potential differences in their relative bioavailability were assessed. The difference in

relative bioavailability between first and subsequent doses of PQ and ART was also

investigated. For PQ this was achieved by estimating the difference between the

relative bioavailability of the first dose of PQ phosphate (fixed to 1) and the two doses

of PQ base as well as the two subsequent doses of PQ phosphate. Similarly for ART, the

relative bioavailability of the first dose was fixed to 1 and potential differences

between this and subsequent doses were assessed. The inclusion of an extra

parameter to account for differences in relative bioavailability was only considered if

accompanied by a significant fall in the OFV (>6.63, P<0.01) and an improvement in the

CWRES plot. Differences in absorption parameters (kRaR, NN and MTT) between the two

groups were also assessed within NONMEM. As described below, the effect size of the

difference (%) was estimated. To maintain the extra parameter estimating this

difference, a significant fall in the OFV (>6.63, P<0.01) was required. Differences

between clearance and volume terms between the two formulations were not

assessed, as differences between a salt and base formulation of the same drug are

biologically implausible.

Finally, relationships between model parameters and the covariates age, sex,

log(baseline parasitaemia) and fever were identified through inspection of scatterplots

and boxplots of eta vs. covariate, and subsequently evaluated within NONMEM. The

152

effect size (%) of categorical data (sex, fever) was assessed, while both linear and

power relationships were evaluated for continuous covariates (age, log(baseline

parasitaemia)). For effect size, the individual parameter value = population parameter

value × (1 + effect parameter * covariate value [0 or 1]). For linear relationships, the

individual parameter value = population parameter value × (1 + effect parameter ×

[covariate value for individual]/[average value of covariate]). For power relationships,

the individual parameter value = population parameter value × ([covariate value for

individual]/[average value of covariate]effect parameter). A stepwise forward inclusion and

backward elimination method was used with a significance of P<0.05 required for

inclusion of a covariate relationship and P<0.01 to retain a covariate relationship.

As CQ was used as the internal standard in the PQ assay, the potential impact of

residual CQ in the plasma of the children on PK parameters was assessed through

simulation. A previous study in a similar group of children resident in the same study

area demonstrated that approximately 50% had a measurable plasma CQ

concentration when hospitalized.395 Using plasma CQ concentrations from a previous

PK study of Madang children,387 we simulated i) that half of the children had, at

random, received a treatment course of CQ finishing 14 days prior to the study (just

before to the exclusion period for such treatment) and ii) only children from one of the

treatment groups received CQ treatment 14 days prior to the study. This latter

simulation represents the ‘worst case’ scenario in terms of the effect of residual CQ on

the comparative PK properties of the two PQ formulations through exogenous

augmentation of the internal standard.


Initially, plots of observed vs. individual and population predicted values, and time vs.

CWRES, were assessed. A bootstrap using Perl speaks NONMEM (PSN) with 1,000

samples was performed (for NQ this was stratified according to dose regimen), and the

parameters derived from this analysis summarized as median and 2.5th and 97.5th

percentiles (95% empirical CI) to facilitate evaluation of final model parameter

estimates. In addition, prediction corrected VPCs (pcVPCs)396 and NPCs were

performed with 1,000 datasets simulated from the final models, and these were

stratified according to treatment group for PQ. The observed 10th, 50th and 90th

153

percentiles were plotted with their respective simulated 95% CI to assess the

predictive performance of the model. NPCs were assessed by comparing the actual

with the expected number of data points within the 20, 40, 60, 80, 90 and 95% PI.

These were also stratified according to treatment group for the PQ model.


Comparisons between the baseline characteristics and secondary PK parameters of the

subjects in the DHQ/PQ and ART/PQ base studies were assessed using the Mann‐

Whitney U test for continuous variables and the Fisher exact test for categorical

variables. A two‐tailed level of significance of 0.05 was considered significant for all

comparisons.

154

5.2.4 59B59BResults


The baseline characteristics of all children are summarized in Table 5‐1. Of those who

received ART/PQ base, 11 had a mono‐infection with P. falciparum and one had a

mono‐infection with P. vivax. One child was lost to follow‐up after day 14. ART/PQ

base treatment was well tolerated with reported symptoms were mild, short‐lived (<2

days) and consistent with clinical features of uncomplicated malaria. Initial fever

clearance was <24 h in all cases, and parasite clearance was <48 h in all but one child in

whom it was within 72 h. The child with P. vivax at enrolment cleared parasitaemia

promptly and remained slide negative for the 56 days of follow‐up. Of the 11 children

with P. falciparum, one developed slide‐positive P. falciparum on day 28, another on

day 42 and two more by day 56. As PCR was not performed it was not possible to

determine if these represented recrudescence or re‐infection. Only one child with P.

falciparum at entry became slide positive for P. vivax, on day 56. The mean Hb

concentration, when available, increased as a result of treatment regardless of malaria

status during follow‐up with mean (95% CI) increases from baseline of 1.9 (0.40‐3.3)

(n=9), 1.1 (0.15‐2.5) (n=11) and 1.5 (0.20‐2.5) (n=10) g/dl on days 28, 42 and 56,

respectively (P=0.027, P=0.19 and P=0.041). No cases of hypoglycaemia were

recorded.

Table 5‐1 Baseline characteristics of study participants. Data are number (%), mean ± SD or median [IQR].

19TDuo‐cotecxin387

19T(historical) n=22

19TArtequick

19T(present study) n=12 19TP value

19TAge (years) 19T6.9 ± 1.4 19T7.1 ± 1.5 19T0.790a

19TSex (% male) 19T17 (86%) 19T8 (66%) 19T0.687b

19TWeight (kg) 19T19.1 ± 3.8 19T18.3 ± 3.1 19T0.986a

19TAxillary temperature (°C) 19T37.2 ± 1.2 19T36.3 ± 0.7 19T0.034a

19TP. falciparum parasitaemia 19T19 (86%) 19T11 (92%) 19T1.00b

19TParasite density (/µl whole blood) 19T13,360 [6,900‐51,650] 19T26,270 [3,480‐35,30] 19T0.736a

19TP. vivax parasitaemia 19T2 (9.1 %) 19T1 (8%) 19T1.00b

19TP. malariae parasitaemia 19T1 (4.5 %) 19T0 (0%) 19T1.00b

19THaemoglobin (g/dl) 19T8.6 ± 1.8 19T9.3 ± 2.1 19T0.168a

19TTotal PQ base dose (mg/kg) 19T35.3 ± 4.4 19T38.3 ± 5.8 19T0.136a

19TTotal DHA dose (mg/kg) 19T7.7 ± 1.0

19TTotal ART dose (mg/kg) 19T6.4 ± 1.0 aMann‐Whitney U test, bFisher exact test

155


There were 298 and 174 individual plasma PQ concentrations available from the

DHQ/PQ (n=22) and ART/PQ base (n=12) studies, respectively. No drug concentrations

were BLQ during the 56‐day follow up period. A 3‐comparment model fitted the data

better than a 2‐compartment model with a significant decrease in the OFV (Δ OFV = ‐

109.232, P<0.001). Although the addition of a lag‐time improved the model

significantly (Δ OFV = ‐31.059, P<0.001), the absorption phase was poorly described

with first‐order absorption with or without lag‐time. Therefore a transit compartment

model was tested where NN and MTT through the transit compartments, were

estimated as continuous variables. The transit compartment was significantly better

than a model with lag‐time, resulting in a 37.173 point reduction in the OFV (P<0.001).

Further testing of the combined data sets with models in which the absorption process

of the two formulations of PQ differed (for example, use of a lag‐time model for PQ

base and a transit compartment model for PQ tetraphosphate) were also tested and

offered no advantage over the use of a single transit compartment model. A three‐

compartment model remained superior to a two‐compartment model with the use of

a transit compartment absorption (Δ OFV = ‐57.937, P<0.001).

The structural model parameters were kRaR, NN, MTT, VRCR/FRPQR, VRP1R/FRPQR, VRP2R/FRPQR, CL/FRPQR,

QR1R/FRPQR, QR2R/FRPQR and FR1,ArtequickR. There was poor precision for the estimate of kRaR (%RSE

>100%) as well as a high correlation between kRaR and MTT (>0.95). Therefore, with the

data available in this study, these two parameters could not be estimated

simultaneously and kRaR was set to be the same as kRtrR, i.e. equal to (1+NN) / MTT. IIV was

estimable for MTT, CL/FRPQR, VRCR/FRPQR andVRP1R/FRPQR. Correlation between IIV terms was

estimated for CL/FRPQ Rand VRCR/FRPQR and VRCR/FRPQR andVRP1R/FRPQR. The IOV on FRPQR was also

estimable and accompanied by significant falls in OFV (Δ OFV = ‐69.12, P<0.001) and

RUV (35% to 29%). There was no significant difference between the relative

bioavailability of the two formulations, or between the subsequent doses of PQ base

or tetraphosphate when compared to the first dose. Although inspection of the

concentration‐time curves appeared to indicate a difference in the absorption phase

between the two formulations, when differences in NN and MTT were evaluated, they

did not improve the model. Likewise, none of the tested covariates improved the

model.

156

Table 5‐2 Final population pharmacokinetic estimates and bootstrap results for piperaquine.

19TParameter 19TFinal model

19Testimate (RSE%)

19TBootstrap (n=1000)

19Tmedian [95% CI]

19TStructural and covariate model parameters

19TMTT (h) 19T1.27 (11) 19T1.25 [1.12‐1.58]

19TNN 19T4.20 (19) 19T3.70 [2.77‐5.36]

19TCL/FRPQR (l/h/70kg) 19T40.1 (7) 19T40.7 [36.6‐45.1]

19TVRCR/FRPQR (l/70kg) 19T2,580 (13) 19T2,550 [1,996‐3,142]

19TQR1R/FRPQR (l/h/70kg) 19T113 (21) 19T119.0 [84.3‐166.0]

19TVRP1R/FRPQR (l/70kg) 19T2,760 (24) 19T3,440 [2,750‐5,510]

19TQR2R/FRPQR (l/h/70kg) 19T52.4 (15) 19T52.9 [43.8‐67.1]

19TVRP2R/FRPQR (l/70kg) 19T21,600 (8) 19T22,300 [19,300‐25,320]

19TRandom model parameters

19TIOV in FRPQR (%) 19T46 (14) 19T42 [36‐54]

19TIIV in CL/FRPQR (%) 19T16 (53) 19T16 [5‐29]

19TIIV in VRCR/FRPQR (%) 19T53 (33) 19T45 [31‐71]

19TIIV in VRP1R/FRPQR (%) 19T68 (32) 19T64 [16‐93]

19TIIV in MTT (%) 19T43 (13) 19T42 [34‐52]

19TR (CL/FRPQR, VRCR/FRPQR ) 19T0.33 19T0.272 [‐0.186‐0.710]

19TR (VRCR/FRPQR, VRP1R/FRPQR ) 19T0.85 19T0.874 [0.381‐1.00]

19TRUV (%) 19T29 (5) 19T29 [27‐32]

RSE calculated from bootstrap results. OFV in final model: ‐329.926, bootstrap OFV: (median [95% CI]) ‐316.869 [‐416.930‐‐285.019].

The impact of residual CQ proved to be minimal as assessed using the simulations, with

population PK parameter estimates differing by <9%. When all participants in the same

formulation group were presumed to have taken CQ 14 days prior to the start of the

study, there was still no significant difference between the population PK parameter

estimates of the two PQ formulations.

The final model parameter estimates and the bootstrap results for both PQ

formulations are summarized in Table 5‐2. Bias was <10% for all fixed and random

157

Time (h)

1 10 100 1000Co

nd

itio

na

l we

igh

ted

re

sid

ua

ls (

pip

era

qu

ine

)

-4

-2

0

2

4

Predicted plasma piperaquine (g/l)

1 10 100 1000

Ob

se

rve

d p

las

ma

pip

era

qu

ine

(

g/l)

1

10

100

1000

A

B

Figure 5‐1 (A) Population predicted (○) and individual (●) predicted versus observed plasma piperaquine concentrations (µg/l on logR10R scale) for the final model. The line of identity is also shown. (B) Conditional weighted residuals vs. time (log scale) for piperaquine final model.

model parameters. With the exception of IIV in CL/FPQ all parameters were reasonably

well estimated with RSE of <33%. The correlation between CL/FPQ and VC/FPQ

displayed a wide 95% CI (‐0.186‐0.710). Figure 5‐1 and Figure 5‐2 show GOF plots and

pcVPCs, respectively. The pcVPCs show wide 95% CI for the 10th, 50th and 90th

percentiles due to relatively small numbers of children. The actual 10th, 50th and 90th

percentiles fell into their respective 95% CI for all time‐points for both groups. The

stratified NPCs demonstrated good predictive performance with the expected number

of points above and below the 20, 40, 60, 80, 90 and 95% PIs. The half‐lives, total

AUCR0–∞R and dose adjusted AUCR0–∞R are shown in Table 5‐3. There were no significant

differences in either of these secondary parameters between the two PQ compounds.

The first distribution, second distribution and terminal elimination tR1/2R for all

158

Table 5‐3 Secondary pharmacokinetic parameters of piperaquine derived from post hoc Bayesian estimates for study participants, and day 7 plasma piperaquine concentrations. Data are median [inter‐quartile range].

a Mann‐Whitney U test; b tR½αR, tR½βR and tR½γR are the first distribution, second distribution and terminal elimination half‐lives respectively.

19TParameter 19TPQ (Duo‐cotecxin)

19Tn=22 19TPQ (Artequick)

19Tn=12 19TP valuea

19Tt½α b (h) 19T4.44 [3.43 ‐ 5.30] 19T4.52 [3.76 ‐ 6.41] 19T0.48

19Tt½β b (h) 19T36.1 [33.0 ‐ 45.2] 19T35.3 [28.1 ‐ 58.7] 19T0.82

19Tt½γ b (h) 19T513 [503 ‐ 574] 19T512 [497 ‐ 566] 19T0.82

19TDay 7 concentration (µg/l) 19T39.3 [34.9 ‐ 45.9] 19T42.0 [34.6 ‐ 55.6] 19T0.56

19TAUCR0–∞R (µg∙h/l) 19T49,451 [40,507 – 52,438] 19T44,556 [33,215 – 51,873] 19T0.36

19TAUCR0–∞R (µg∙h/l) / total PQ dose (mg/kg)

19T1.27 [1.06 ‐ 1.50] 19T1.37 [1.09 ‐ 1.65] 19T0.40

participants had median values of 4.5, 36.0 and 512 h respectively. The median PQ

AUCR0–∞Rs for the Artequick and Duo‐cotecxin formulations were 49,451 µg∙h/l and

44,556 µg∙h/l, respectively.

Of the ninety‐six ART drug concentrations (ART/PQ base group, n=12) that were

available for analysis, six (6.25%) were BLQ but above the limit of detection. As these

Time (h)

0 200 400 600 800 1000 1200 1400

Pla

sm

a p

ipe

raq

uin

e (

g/l)

1

10

100

1000

B

Time (h)

0 24 48 72 96

Pla

sm

a p

ipe

raq

uin

e (

g/l

)

1

10

100

1000

Time (h)

0 200 400 600 800 1000 1200 1400

Pla

sm

a p

ipe

raq

uin

e (

g/l)

1

10

100

1000

A

Time (h)

0 24 48 72 96

Pla

sm

a p

ipe

raq

uin

e (

g/l

)

1

10

100

1000

Figure 5‐2 Visual predictive check showing observed 50th (●), 10th () and 90th (○) percen les with the simulated 95% CI for the 50th (solid black line), 10th (grey dotted lines) and 90th (dashed grey lines) percentiles for plasma piperaquine concentrations (µg/l on logR10R scale) vs. time (h) for Artequick (A) and Duo‐cotecxin (B) from the final model. The observed data are superimposed as grey crosses. The insert shows data for the first 96 h.

159

represented a small proportion of the data, they were included at their measured

values. All twelve children has measurable concentrations of ART to 48 h. Initial

modelling of the ART dataset demonstrated a two‐compartment model was

significantly better than a one‐compartment model (ΔOFV = ‐73.417, P<0.001) and that

the absorption phase was best described by a first‐order absorption without a lag time.

Therefore, the structural model parameters were kRaR, VRCR/FRARTR, VRPR/FRARTR,R RCL/FRARTR and

Q/FRARTR. The IIV of VRCR/FRARTR was estimable, as was the IOV on FRARTR. The data supported

the estimation of a relative bioavailability term for the second dose of ART (FR2, ARTR),

with its addition resulting in a significant fall in the OFV (Δ OFV = ‐24.029, P<0.001).

The bioavailability of the second dose was 0.270, relative to the first. No significant

covariate relationships were identified.

The final model parameter estimates and the bootstrap results for ART are

summarized in Table 5‐4. Bias was <10% for all fixed and random parameters. kRaR was

not well estimated with a RSE of 55%, and a four‐range in the non‐parametric 95% CI.

Figure 5‐3 and Figure 5‐4 show GOF plots and pcVPCs, respectively. The pcVPC showed

all observed 10th, 50th and 90th percentiles were within their simulated 95% CI. Due to

the small numbers in the analysis these CI were wide and overlapping. The NPC

Table 5‐4 Final population pharmacokinetic estimates and bootstrap results for artemisinin (n=12).

19TParameter 19TFinal model

19Testimate (RSE%)

19TBootstrap

19TMedian [95% CI]

19TStructural model parameters

19TkRaR (/h) 19T1.67 (55) 19T1.62 [1.01‐4.40]

19TCL/FRARTR (l/h/70kg) 19T124 (12) 19T125 [99‐157]

19TVRCR/FRARTR (l/70kg) 19T590 (30) 19T533 [318‐874]

19TQ/FRARTR (l/h/70kg) 19T43.7 (38) 19T46.4 [19.5‐79.4]

19TVP/FRARTR (l/70kg) 19T435 (26) 19T456 [259‐696]

19TFR2, ART R– relative bioavailability of 2nd dose 19T0.270 (17) 19T0.275 [0.192‐0.368]

19TRandom model parameters

19TIOV in FRARTR (%) 19T43 (27) 19T39 [15‐58]

19TIIV in CL/FRARTR (%) 19T12 (29) 19T12 [4‐18]

19TRUV (%) 19T33 (11) 19T32 [26‐38]

RSE (Relative standard error) calculated from bootstrap results. OFV in final model: ‐63.562, bootstrap OFV: (median [95% CI]‐73.838 [‐110.720‐‐43.043].

160

demonstrated good predictive performance with the expected number of points above

and below the 20, 40, 60, 80, 90 and 95% PIs. The tR1/2R, the AUCR0–∞R of each dose as well

as the total AUCR0–∞R for the study participants are shown in Table 5‐5. The median

distribution and terminal elimination tR1/2R were 1.55 and 7.43 h while the median total

AUCR0–∞R

was

1,652

µg∙h/l.

Predicted plasma artemesinin (g/l)

1 10 100 1000

Ob

se

rve

d p

las

ma

art

em

isin

in (

g/l)

1

10

100

1000A

Time (h)

0 12 24 36 48Co

nd

itio

na

l we

igh

ted

re

sid

ua

ls (

art

em

isin

in)

-4

-2

0

2

4B

Figure 5‐3 (A) Population (○) and individual (●) predicted versus observed plasma artemisinin concentra ons (µg/l on logR10R scale) for the final model. The line of identity is also shown. (B) Conditional weighted residuals vs. time for artemisinin final model.

Table 5‐5 Secondary pharmacokinetic parameters for artemisinin derived from post hoc Bayesian estimates for study participants. Data are median [inter‐quartile range].

19TParameter 19TART (Artequick)

19Tn=12

19TtR½αR

a (h) 19T1.55 [1.49 ‐ 1.60]

19TtR½βR

a (h) 19T7.43 [7.22 ‐ 7.68]

19TAUC (µg∙h/l) – first dose 19T1,347 [1,065 – 1,594]

19TAUC (µg∙h/l) – second dose 19T312 [253 ‐ 438]

19TAUCR0–∞R (µg∙h/l) – total 19T1,652 [1,333 – 2,177] a tR½α Rand tR½βR represent the distribution and terminal elimination half‐life respectively AUCs for each dose has been calculated using standard PK formula to determine the relative contribution of each dose to the total AUCR0–

∞R

161

5.2.4.3 125B125BRelationshipbetweendrugexposureandtreatment

outcome

In the eight children who became slide positive for P. falciparum by day 42 (two in the

ART/PQ base group and six in the DHQ/PQ group), the AUCR0–∞R of PQ was significantly

lower than in those who remained free of P. falciparum infection (median 39,297 vs.

49,776 µg∙h/l, P=0.0060). Clearance and terminal elimination tR1/2R were not significantly

different, however, these children received a lower total mg/kg dose of PQ (median

31.4 vs. 35.7 mg/kg of PQ base, P=0.11). When adjusted for dose, the AUCR0–∞R was no

longer significant between those children with and without slide positivity for P.

falciparum by day 42 (median 1.16 vs. 1.42 µg∙h/l per mg/kg of PQ base, P=0.14).

There was a significant positive correlation between AUCR0–∞R and day 7 drug

concentrations that did reach significance (r=0.70, 95%CI 0.48 – 0.84, P<0.001). Unlike

AUCR0–∞R, day 7 concentrations were not significantly lower in those who developed P.

falciparum slide positivity (n=8) compared to those who did not (n=26) (median 44.1vs.

48.0 µg/l, P=0.22). Similar results were evident when the two children from the

DHQ/PQ group who developed slide positivity for P. vivax by day 42 were included in

the analysis (data not shown). In the child who took >48 hours to clear initial

parasitaemia, the AUCR0–∞Rs of ART and PQ were within the ranges of the other patients.

Time (h)

0 12 24 36 48P

las

ma

art

em

isin

in (

g/l)

1

10

100

1000

Figure 5‐4 Visual predictive check showing observed 50th (●), 10th () and 90th (○) percen les with the simulated 95% CI for the 50

th (solid black line), 10th (grey dotted lines) and 90th (dashed grey lines) percentiles for plasma artemisinin concentrations (µg/l on logR10R scale) vs. time (h) from the final model. The observed data are superimposed as grey crosses.

162


The development of ACTs has seen a variety of different combinations, formulations

and dose regimens introduced into clinical use without a detailed assessment of

tolerability, safety, PK and efficacy. One recently‐marketed ACT, Artequick, appears

relatively inexpensive to produce but uses component drugs that have not been

investigated extensively, especially in a paediatric setting. The present PK and

preliminary efficacy study in PNG children adds to available data 212, 213 suggesting that

there would be benefit in extending the Artequick manufacturer’s recommended two‐

day regimen to three days as this will increase PQ exposure and thus limit late

recurrence of parasitaemia. However, the selection of a relatively low dose of ART (3

mg/kg vs. the 10‐20 mg/kg per dose conventionally recommended), a drug that

induces its own metabolism, may have implications for efficacy, especially in patients

with limited immunity to malaria or where parasite resistance against artemisinin

compounds has started to develop.397

Children in the DHQ/PQ tetraphosphate group were given a mean of 35.3 mg/kg PQ

base over three days387 compared with 38.3 mg/kg PQ base over two days in the

present children treated with ART/PQ base. Overall the exposure to PQ was similar

between the two formulations, and no differences in the post hoc PK parameters were

identified. Although, this suggests that the tablet and granule formulations have

similar bioavailability and that the small amount of fat we administered with each dose

(2 g) is unlikely to influence exposure to PQ, it is not possible to differentiate the

differential influence of food and formulation with the current study design. Two of

three studies involving healthy adults found that fat‐containing foods increased

exposure to PQ tetraphosphate,175, 176, 215 but the volunteers in these studies

consumed relatively large quantities of fat (17‐54 g). Consistent with the present

results, 6.4 g of fat did not increase the exposure to PQ tetraphosphate in adults with

malaria.214 However PQ base is less water soluble than PQ tetraphosphate and

exogenous lipids are known to increase the solubility of lipophilic drugs and thus affect

the extent of absorption.398 PQ base may behave similarly to LUM, another highly

lipophilic drug, and require a smaller amount of fat to maximize absorption.375 Granule

formulations have been reported to increase bioavailability relative to tablets399

consistent with the increased surface area available for dissolution compared to

163

tablets, but our data suggest that this was not a major effect in the case of Artequick.

Future studies evaluating the effect of food and formulation on the disposition of PQ

base in malaria could help refine dose regimens for therapies such as Artequick.

A model with three compartments and a transit sequence prior to absorption best

described the PQ concentration‐time dataset. Most previous studies have used a two‐

compartment model.174, 175, 178 One study in healthy adults found that, although a

three‐compartment model described the post‐administration profile better, there

were insufficient data to support its use over a two‐compartment model.177 The mean

elimination tR1/2R, a parameter influenced by the duration of sampling,216 was 512 h,

within the previously‐reported range of 224‐667 h.172‐179 Since there was substantial

variability in the absorption phase of the plasma PQ concentration profile, a transit

compartment model was tested and proved better than simpler absorption models

that used lag time, as has been found in studies of other drugs.287

It has recently been suggested that children should be given a higher dose of PQ than

adults due to lower day 7 plasma concentrations202 and reduced efficacy.202, 392, 393 This

is supported by comparative PK studies in children and adults that found children had

a higher clearance174 or a lower PQ exposure at critical times during the illness.178

These concerns have also been raised for other antimalarial drugs144, 226 and reflect PK

effects due to the effects of body size, maturation and organ function.394 Although only

children aged between 5 and 10 years were included in the present study, we found

that recurrence of parasitaemia was associated with a lower PQ AUC resulting from a

lower mg/kg dose, consistent with other studies of DHQ/PQ tetraphosphate.196 As PCR

was not performed, these cases may represent either recrudescence (treatment

failure) or reinfection (failure of post‐treatment prophylaxis).

The dose‐adjusted PQ exposure in our children was similar to that found in Caucasian

and Vietnamese healthy adults172, 177, 215 and Thai adults with malaria.214 When

compared to studies of Vietnamese and Chinese healthy adults, the dose‐adjusted PQ

exposure was three and six times lower in our children, respectively,173, 175 suggesting

that there are PQ PK differences between populations. Currently recommended PQ

tetraphosphate doses are 18 mg/kg/day (10 mg/kg/day PQ base) for 3 days.388 A

164

higher average daily dose of PQ base in the Artequick group (19 mg/kg/day) was well

tolerated when given for two days and the same dose given on the third day might

both satisfy WHO recommendations for duration of ACT and address the issue of the

need for higher mg/kg doses in children.

The use of CQ as an internal standard for the PQ is potentially problematic in samples

taken from a malaria‐endemic area where CQ is widely available and used empirically

for treatment of fever. Utilizing CQ usage and PK data from other studies in children

from the Madang area,387, 395 we investigated whether the 14‐day exclusion for prior

antimalarial treatment was sufficient to limit such potential confounding. Even in the

worst case scenario, there was only a small effect on the estimated PK parameters and

one which did not produce falsely significant differences between the two

formulations. Although this is reassuring, it would be best if future similar studies

employed an alternative internal standard.

A number of published studies have evaluated the PK of ART in healthy adults112‐117

and adults with malaria118‐124 but only one has included children with malaria.125 In this

latter Vietnamese study, 23 children aged 2‐12 years were given five days of ART

dosed according to WT (approximately 10 mg/kg) and 31 adults received 500 mg ART

daily for 5 days. Sparse sampling was used to characterize ART population PK in plasma

using NONMEM after the first and final doses, with two samples collected from each

patient on day 1 and a single sample collected on day 5 from some patients. A one‐

compartment model was used, with clearance and volume terms for children and

adults estimated separately. The median WTs and ages of the children were lower in

the present study (18.3 vs. 20 kg and 7.1 vs. 9 years, respectively). Although our value

for kRaR was comparable to that in the Vietnamese study (2.0/h vs. 1.7 /h), a two‐

compartment model provided a better fit in the present study with distribution and

elimination tR1/2Rs of 1.9 h and 8.3 h respectively, compared to a tR1/2R of 1.8 h in the

previous study.125 This difference may reflect the longer sampling duration in the

present study (24 h vs. 8 h post dose) which enabled the identification of a second

exponential phase in the elimination of ART.

165

The elimination tR1/2R of ART has been reported to be between 1.4 and 4.8 h in non‐

compartmental112, 113, 117, 118, 120, 121, 123, 124, 400 and compartmental114‐116, 122, 125 analyses.

The present analysis supports a bi‐exponential disposition for ART, while most previous

compartmental analyses have reported a mono‐exponential disposition. A shorter

sampling duration may be responsible for this difference, as sampling was confined to

<10 h after the last dose in all but one of the studies to date. In addition, assay

sensitivity may also contribute by limiting quantification of ART to those samples taken

<12 h after dose.114, 115 One study of healthy adults given a single dose of ART112 also

reported a bi‐exponential disposition and found longer distribution (2.61 h vs. 1.55 h)

and shorter elimination (4.34 h vs. 7.43 h) tR1/2Rs than in the present study. Although this

study sampled blood to 24 h post dose, ART could only be quantified in samples up to

8 h.

The present median AUCR0–∞R of the first dose of ART (1,347 µg∙h/l) was within the

range reported for healthy adults (1,190‐2,690 µg∙h/l)112‐115, 117 but well below that of

adults with malaria (2,601‐2,780 µg∙h/l)118, 123, 400 who received 500 mg of ART. Our

children received a lower dose of ART 3.2 mg/kg/day and, when adjusted for the

relative dose administered, the AUCR0–∞R for the first dose was above those seen in

adults.118, 123, 400 The auto‐induction of ART metabolism has been well characterized,

with a primary effect on the bioavailability of subsequent doses rather than on

systemic clearance.116 It is likely, therefore, this represents an increase in the activity of

gut wall rather than liver metabolism.

We found a difference in the PK of ART for the second dose that was explained by a

lower relative bioavailability of 0.27 when compared to the first dose. When the AUCs

of different doses have been compared in previous studies, the relative bioavailability

after 4‐7 days was between 0.13 and 0.29.113, 117, 121, 123, 125, 400 One study of African

adults with malaria who received 500 mg ART daily for three days and a single dose of

MQ124 measured ART in saliva and found relative bioavailability was only lower on the

third day (0.45) when MQ was given after the last dose of ART. However, when MQ

was given on the first day at the same time as the first dose of ART, the relative

bioavailability of both the second and third doses of ART was lower, at 0.23 and 0.25

respectively.

166

Our data are in agreement with a rapid mean auto‐induction time of 1.9 h estimated

using a semi‐physiological model for ART116 indicating that all doses after the first have

a lower relative bioavailability. If a third daily dose of ART/PQ base was given, its

relative bioavailability would also be low. The rapid initial parasite clearance in

Artequick‐treated children in the present study despite relatively low and short‐lived

plasma ART concentrations, may reflect the level of malarial immunity in this area of

intense transmission.401 It is likely that relatively low ART doses, even if given over

three days rather than two, might not be as effective where transmission and

consequent immunity are less or where parasite resistance against artemisinin

compounds has started to develop.397

When compared to three days of DHQ/PQ tetraphosphate, the efficacy of two days of

Artequick in adults was equivalent in one study211 and inferior in another.212 A three‐

day Artequick regimen (3.2 and 16.0 mg/kg/day of ART and PQ base, respectively) has

been found to be both well tolerated and more effective than a two day regimen.213

Our preliminary data suggest that the efficacy of two days of Artequick appeared

similar to that of three days of Duo‐cotecxin in PNG children. However, the weight of

evidence from previous studies,212, 213 the low dose of ART in Artequick and its auto‐

induction at a time when the spectre of ART resistance has emerged,397 and the issue

of potential PQ under‐dosing in children would all support further evaluation of a

theoretically more efficacious three‐day Artequick regimen, as recommended by the

WHO for all ACTs.388

167



their kind co‐operation during the study. We also thank Jovitha Lammey, Christine

Kalopo and Bernard (“Ben”) Maamu for clinical and/or logistic assistance. Dr Harin

Karunajeewa is acknowledged for his pivotal role in co‐ordinating the original DHA/PQ

tetraphosphate study. We thank Artepharm Co Ltd for kind provision of Artequick. The

National Health and Medical Research Council (NHMRC) of Australia funded the study

(grant #634343). TMED is supported by an NHMRC Practitioner Fellowship.

168

169

6 5B5BArtemisinin‐NaphthoquineCombination

TherapyforUncomplicatedPaediatricMalaria:

APharmacokineticStudy


The primary aim of the final study presented in this thesis was to provide the first PK,

safety, tolerability and efficacy evaluation of ART/NQ in children. As a number of

different dose regimens were used it also aimed to assess the impact of these different

dose regimens on PK, safety, tolerability and efficacy.

As with ART/PQ base, ART/NQ is a combination readily available for purchase in a

number of countries despite few pharmacological data available in the literature.

Unlike ART/PQ base, however, the partner drug, NQ, is not present in any other

available antimalarial formulation. Data of this combination at the time of this study

was lacking in adults, let alone children. The role of the first part of the study was to

determine the required paediatric dose the recommended adult dose. In the second

part, supported by a preliminary analysis of the data from the first part, two different

dose regimens were tested. This study was designed not only to ensure the safety and

efficacy of a combination not previously studied in children, but also to provide a basis

for future efficacy evaluations (now ongoing in PNG).

This study resulted in the publication5 presented in this chapter. Entitled, “Artemisinin‐

naphthoquine combination therapy for uncomplicated paediatric malaria: A

pharmacokinetic study” it was published by the journal Antimicrobial Agents and

Chemotherapy (2012. 56(5):p. 2472‐2484). The contribution of each of the authors is

outlined in section i, which also contains details of ethical approvals and supporting

funding. It has been reformatted to conform to thesis requirements set by the

University of Western Australia. The references have been combined with those for

the thesis as a whole and can be found in section x below.

170

171


Kevin T. Batty,A,B Sam Salman,C Brioni R. Moore,C John Benjamin,D Sook Ting Lee,C

Madhu Page‐Sharp,A Nolene Pitus,D Kenneth F. Ilett,C Ivo Mueller,E Francis W.

Hombhanje,F Peter Siba,D Timothy M. E. DavisC.

ASchool of Pharmacy, Curtin University, Bentley, Western Australia

BCurtin Health Innovation Research Institute, Curtin University, Bentley, Western

Australia

CSchool of Medicine and Pharmacology, University of Western Australia, Crawley,

Western Australia

DPapua New Guinea Institute of Medical Research, Madang, Papua New Guinea

EInfection and Immunity Division, Walter and Eliza Hall Institute of Medical Research,

Victoria, Australia and Center de Recerca en Salut Internacional de Barcelona (CRESIB),

Barcelona, Spain

FCentre for Health Research, Divine Word University, Madang, Papua New Guinea


ART/NQ is a co‐formulated antimalarial therapy marketed as single‐dose treatment in

PNG and other tropical countries. To build on limited knowledge of the PK properties

of the components, especially the tetra‐aminoquinoline NQ, we studied ART‐NQ

disposition in PNG children aged 5‐12 years with uncomplicated malaria, comparing

single‐dose (15:6 mg/kg) administered with water (Group 1, n=13), single‐dose (22:9

mg/kg) with milk (Group 2, n=17) or two daily 22:9 mg/kg doses with water (Group 3,

n=16). Plasma NQ was assayed by HPLC and ART using liquid chromatography‐mass

spectrometry. Population‐based multi‐compartment PK models for NQ and ART were

developed. NQ disposition was best characterized by a three‐compartment model with

a mean absorption tR½R of 1.0 h and predicted median maximum plasma concentrations

that ranged up to 57 µg/l after the second dose in Group 3. The mean NQ elimination

tR½R was 22.8 days CL/F was 1.1 l/h/kg and VRSSR/F was 710 l/kg. Administration of NQ with

fat (8.5 g; 615 kJ) vs. water was associated with a 25% increased bioavailability. ART

disposition was best characterized by a two‐compartment model with mean CL/F (4.1

l/h/kg) and V/F (21 l/kg) that were similar to those of previous studies. There was a

77% reduction in the bioavailability of the second ART dose (Group 3). NQ has PK

172

properties that confirm its potential as an artemisinin partner drug for treatment of

uncomplicated paediatric malaria.

173


Available data relating to the PK of the antimalarial drug NQ phosphate are limited and

inconsistent. Initial reports suggested that NQ has a high oral bioavailability (>90%)

and a tR½R of 41‐57 h.217 In a more recent study in healthy Chinese men in which NQ was

given alone and co‐formulated with ART,220 the elimination tR½R of NQ was substantially

longer at 250‐300 h. This volunteer study also showed that the AUC for NQ exhibited

an unusual relationship between formulation and co‐administered fat. The mean value

was similar in the fasted NQ monotherapy and fed combination ART‐NQ therapy

groups but more than double this in the fasted volunteers given the fixed

combination.220 The fact that the highest bioavailability was in the fasting state

appears in contrast to the effect of fat on absorption of related drugs such as LUM and

PQ,173, 215, 375, 402 while the apparent beneficial effects of co‐formulation on

bioavailability was difficult to explain.220

It has been shown that NQ is a P‐glycoprotein substrate and that NQ efflux is

saturable,403 suggesting that absorption could be non‐linear at high doses. However,

the Chinese volunteer study of ART‐NQ found dose‐proportional increases in the

maximum plasma concentration (CRmaxR) and AUC for NQ at doses of between 200 and

600 mg.220 The maximum individual value for CRmaxR was just over 100 µg/l in this

study,220 but a CRmaxR of up to 245 µg/l has been reported after a 600 mg dose in

adults.217

The Chinese volunteer study of NQ and ART‐NQ reported a tR½R for ART of 3.6‐4.0 h, a

CL/F of approximately 1.5 l/h/kg and V/F of 8 l/kg.220 By contrast, a number of previous

studies in healthy adult volunteers112, 115, 404, 405 and patients with uncomplicated

falciparum malaria118, 119, 123, 125, 406 have reported lower mean values for tR½R, of 2.0‐2.7

h (mean 2.3 h), higher mean values for CL/F of 5.1‐9.3 l/h/kg (mean 6.7 l/h/kg), and a

higher mean V/F of 16.4‐35.5 l/kg (mean 27 l/kg). The reported mean CL/F and V/F for

ART in children were even greater at 14.4 l/h/kg and 38 l/kg, respectively.125 The

Chinese study did, however, show that the AUC for ART increased with co‐

administered fat,220 consistent with most past reports.407

174

Because of inconsistencies between the few published studies of NQ PK and a lack of

PK data in children, we conducted two PK studies of ART‐NQ in children from PNG with

uncomplicated malaria. An initial pilot study (Study 1), carried out before the

manufacturer had produced a paediatric dosing schedule and utilizing a conservative

(calculated in mg/kg using WT based on the dose for adults) was designed to provide

preliminary PK data relating to NQ disposition in children, while the second study

(Study 2) aimed to characterize the PK of NQ as well as ART in more detail when given

at the manufacturer’s recommended dose with fat (milk) or as a two‐dose regimen.

175


6.2.3.1 126B126BPatientsandclinicalmethods

Full details of the studies have been provided in a separate report.408 In brief, children

aged 5‐10 years who presented with an axillary temperature >37.5°C or a history of

fever in the previous 24 h who were slide‐positive for malaria (P. falciparum >1,000

asexual parasites/μl whole blood or P. vivax >250 parasites/μl) were eligible provided

that had no complications or concomitant illness, no prior treatment with study drugs

no history of allergy to ART or aminoquinoline drugs. Each child’s parents or guardians

gave written informed consent. Approvals were obtained from the PNG Institute of

Medical Research Institutional Review Board and the Medical Research Advisory

Committee of the PNG Health Department.

At enrolment, a history was taken and a full physical examination was performed. An

intravenous cannula was inserted and a baseline venous blood sample was drawn. In

Study 1, all children were administered ARCO™ tablets (50 mg NQ plus 125 mg ART;

Kunming Pharmaceuticals, Kunming, China) orally as a single dose of 2‐4 whole tablets

with water (Group 1). The dose was based on WT as per those recommended by the

manufacturer in mg/kg for adults,409 and represented a dose range of 5.0‐7.5 mg/kg

for NQ phosphate and of 12.5‐16.8 mg/kg for ART. Subjects were not required to fast

prior to, or after, treatment. If the child vomited within one hour the same dose was

re‐administered and the time of re‐administration recorded.

In Study 2, children were randomized by a computer‐generated sequence to receive

ARCO tablets (50 mg NQ plus 125 mg ART) orally based on WT as recommended by the

manufacturer for children409 as either i) a single dose of 3‐6 tablets given with 250 ml

full cream cow’s milk (8.5 g fat) with dose ranges of 6.1‐9.5 mg/kg for NQ and 15.3‐

23.8 mg/kg for ART (Group 2), or ii) the same dose given with water on two occasions

24 h apart (Group 3). Each child was kept fasting under observation and if he/she

vomited within 1 h the same dose was re‐administered and the time of re‐

administration recorded.

176

Group 1 patients had additional venous blood samples drawn through the cannula at

1, 2, 4, 8, 12, 18, 24, 48, 72 h, and by venesection at 4, 7, 14, 28, 42, 56 days. Blood was

collected into lithium heparin tubes, centrifuged at 1,800 g for 5 min and the

separated plasma stored at ‐80°C until analysed for NQ concentration within 8 months

of collection. These children were re‐assessed clinically at 4 and 24 h, and on days 2, 3,

7, 14, 28 and 56.408 Group 2 and 3 patients had further 2.5 ml blood samples for drug

assay taken at 1, 2, 4, 8, 12, 18, 24, 48 and 72 h through the sampling cannula, and by

venesection at 4, 7, 14, 28 and 42 days. Group 3 patients had a second ART‐NQ dose

given with water at 24 h. Groups 2 and 3 had similar post‐treatment clinical and other

monitoring to that performed in Group 1.408

6.2.3.2 127B127BAnalyticalmethods

NQ diphosphate was obtained from ZYF Pharm Chemicals, Shanghai, China, tramadol

hydrochloride and ART were from Sigma‐Aldrich Chemicals, St Louis, MO, USA, and

ARM was from AAPIN Chemicals Ltd, Abingdon, Oxon, UK. All general laboratory

chemicals were of analytical grade (Sigma‐Aldrich Chemicals, St Louis, MO, USA; Merck

Chemicals, Darmstadt, Germany).

NQ in plasma was analysed using a validated HPLC assay, based on established

analytical methods for CQ, PQ and MQ.387, 410 Briefly, plasma samples (500 µl) were

spiked with tramadol as internal standard (500 ng), alkalinized with sodium

tetraborate 2% w/v solution (1 ml) and extracted into 8 ml hexane:ethyl acetate

(80:20) by shaking for 10 min. The samples were then centrifuged at 1300 g for 10 min.

Supernatant (7.5 ml) was back extracted into 0.1 ml of 0.1 M HCl by shaking for 5 min,

followed by centrifugation as above. The HCl layer was transferred to 1.5 ml

microcentrifuge tubes and re‐centrifuged at 1300 g for 25 min to evaporate traces of

organic solvent, after which 70 µl was injected onto the HPLC. Analytes were separated

on a Luna CR18R HPLC column (length, 100 mm; internal diameter[i.d.] 4.6 mm; particle

size 3 µm; Phenomenex, Australia) in series with an Octadecyl CR18R (length, 4 mm; i.d. 3

mm; Phenomenex, Australia) guard column at 30°C with a mobile phase of 18% v/v

acetonitrile in 50 mM KHR2RPOR4R buffer (pH 2.5) pumped at 1 ml/min. Approximate

retention times for NQ and tramadol were 9.4 min and 6.8 min, respectively and the

analytes were detected by UV absorbance at 222 nm (see Figure 6‐1). The linear

177

calibration

were inclu

and 100 µ

and 100 µ

106% and

and 0.5 µg

96% at 5,

ART in pla

Stock solu

methanol)

Figure 6‐1 HPtramadol (T; is a patient’s µg/l naphtho

n range for

uded in each

µg/l respecti

µg/l respecti

91‐98% res

g/l, respecti

20 and 100

sma was an

utions of AR

), and store

PLC‐UV (222 nmtRRR = 6.8 min). Ppre‐dose blank

oquine).

each assay

h batch. The

ively (n=5),

ively (n=15)

spectively. T

ively. Mean

µg/l, respe

nalysed usin

T and ARM

ed in the dar

m) chromatograPanel A is spikek sample (with

was 1‐100

e intra‐day

while inter

). Intra‐day

The limit of

n recoveries

ectively.

ng LC‐MS ba

(internal st

rk at ‐80°C.

ams showing ned plasma usedh IS) showing no

µg/l and QC

RSDs of NQ

r‐day RSDs w

and inter‐d

f quantificat

s of NQ from

ased on an

tandard) we

Working st

aphthoquine (Nd in the calibrato endogenous i

C samples (5

Q were 8.9,

were 7.7, 5.

day accuracy

tion and lim

m plasma w

established

ere prepare

tandard solu

N; tRRR = 9.4 min)tion curve (20 µinterference; P

5 , 20 and 1

3.1 and 4.5

.2 and 3.4 %

y ranges we

mit of detect

were 88%, 98

d assay for A

ed separate

utions were

) and the interµg/l naphthoquPanel C is a typi

100 µg/l)

% at 5, 20

% at 5, 20

ere 92‐

tion were 1

8% and

ARM.108

ly (1 g/l in

e prepared

nal standard, uine); Panel B ical sample (25

5

178

Figure 6‐2 LC‐(IS; tRRR = 7.9 mpatient’s pre‐dµg/l artemisin

from the p

were const

higher con

were re‐an

plasma as

to use in e

The extrac

Elut® PH; V

modificatio

followed b

MS chromatogin). Panel A is sdose blank samnin).

primary stoc

tructed (5‐2

centrations

nalysed follo

above at co

ach batch a

tion proced

Varian Inc.,

ons. Briefly,

by 1 ml of 1M

grams showing spiked plasma mple (with IS) s

ck at 1, 10 a

200 µg/l) fo

s by spiking

owing appro

oncentratio

analysed.

dure used a

Palo Alto, C

, the SPE co

M acetic ac

artemisinin (Aused in the calhowing no end

and 100 mg/

r the lower

into blank

opriate dilu

ns of 5, 200

1 ml CR18R so

CA) as descr

olumn was p

id. Plasma s

RT; tRRR = 4.3 milibration curve dogenous inter

/l. Two sets

r concentrat

plasma. Sam

ution. QC sa

0 and 1,000

olid‐phase e

ribed previo

pre‐conditio

samples (0.

n) and the inte(200 µg/l artemrference; Panel

s of 5‐point

tions and (2

mples abov

mples were

µg/l, and st

extraction (S

ously,380 wit

oned with 1

5 ml) were

rnal standard, misinin); Panel C is a typical sa

calibration

200‐2,000 µ

e the stand

e prepared

tored at ‐80

SPE) column

th minor

1 ml of meth

spiked with

artemether B is a ample (136

curves

µg/l) for

dard curve

in blank

0 °C prior

n (Bond

hanol

h internal

179

standard (ARM, 1 µg), loaded onto the pre‐conditioned SPE column and drawn through

using a medium vacuum. The column was then washed with 1M acetic acid (1 ml; two

washes), followed by 20% v/v methanol in 1M acetic acid (1 ml). The column was dried

under low vacuum for 30 min and retained drugs were eluted with 2 ml of t‐butyl

chloride:ethyl acetate (80:20% v/v). The eluate was evaporated in a vacuum

evaporator at 35°C, reconstituted in 50 µl of mobile phase and 5 µl aliquots injected

into the LC‐MS system.

The single‐quad LC‐MS system (Model 2020, Shimadzu, Kyoto, Japan) comprised a

binary pump (20AD), vacuum degasser, thermostated autosampler (SIL 20ACHT),

thermostated column compartment (CTO 20A), photodiode detector (SPD M 20A) and

mass analyser (MS 2020) with both electrospray ionization (ESI) and atmospheric

pressure chemical ionization (APCI) systems. Analysis was performed in the isocratic

mode with 20 mM ammonium formate (pH 4.8):methanol (20:80) at a flow rate of 0.2

ml/min. Chromatographic separation was undertaken at 30°C on a Synergy Fusion‐RP

CR18R 4 µm column (150 mm × 2 mm i.d.) coupled with a 5 µm CR18R guard column (4 mm ×

2 mm i.d.; Phenomenex, Australia). Retention times were 4.3 and 7.9 min for ART and

ARM respectively (see Figure 6‐2). Optimized mass spectra were acquired with an

interface voltage of 4.5kV, a detector voltage of 1kV, a heat block temperature of

400°C and a desolvation gas temperature of 250°C. Nitrogen was used as the nebulizer

gas at a flow rate of 1.5 l/min and a dry gas flow of 10 l/min.

Both ART and ARM standard solutions were first scanned from (m/z 100‐500) in ESI

and APCI positive mode, as well as combined ESI and APCI (DUIS) mode to identify the

abundance of ions corresponding to respective drugs. The base peak intensity of all

three modes were compared, and showed that DUIS mode gave highest signal

intensity. Therefore, quantitation was performed by selected ion monitoring (SIM),

using DUIS mode. For ART, the parent molecule [M+H]+, (m/z = 283) was used for

quantitation, while for ARM the predominant fragmented ion (m/z =221) was

monitored.411

All standard curves were linear (r20.999). Chromatographic data (peak area ratio of

ART:ARM) were processed using LAB Solution (Version 5, Shimadzu, Japan). Responses

180

from the analysis of three ART concentrations (5, 200 and 2,000 µg/l) spiked into five

separate plasma samples were used to determine matrix effects (ion

suppression/enhancement), absolute recovery and process efficiency.381, 412 Three sets

of matrix solutions were prepared. Set 1 comprised blank plasma spiked first and then

extracted, Set 2 comprised blank plasma extracted first and then spiked post‐

extraction, and Set 3 comprised pure solutions of analyte. The matrix effect (%) was

determined from [Set 2 response × 100]/[Set 3 response], process efficiency (%) from

[Set 1 response × 100]/[Set 3 response] and absolute recovery (%) from [Set 1

response × 100]/[Set 2 response]. The mean ± SD (range) matrix effect for ART was 94

± 12 (79‐105), 90 ± 13 (75‐106) and 91 ± 2 (88‐92)% at 5, 200 and 2,000 µg/l,

respectively. The mean ± SD (range) process efficiency for ART was 93 ± 18 (73‐121),

84 ± 11 (75‐102) and 82 ± 4 (80‐89)% at 5, 200 and 2,000 µg/l, respectively. The mean

± SD (range) absolute recovery for ART was 88 ± 10 (78‐101), 86 ± 9 (77‐102) and 90 ±

7 (87‐101)% at 5, 200 and 2,000 µg/l, respectively. The mean ± SD (range) matrix

effect, process efficiency and absolute recovery for the internal standard, ARM, were

98 ± 10 (87‐113), 88 ± 4 (82‐92) and 91 ± 6 (81‐96)% at 1,000 µg/l. The assay intra‐day

RSDs were 9.3, 7.2 and 3.7% at 5, 200 and 2,000 µg/l, respectively (n=5), while the

inter‐day RSDs were 9.5, 7.1 and 6.5% at 5, 200 and 2,000 µg/l, respectively (n=15).

Inter‐day accuracies determined from the QC samples for each assay batch at 5, 200

and 1,000 µg/l were 108 ± 7% (86‐114), 103 ± 6 (93‐109) and 107 ± 8 (86‐115),

respectively (n=16). The limits of quantification and detection for ART were 2.5 and 1

µg/l, respectively.

6.2.3.3 128B128BPharmacokineticandstatisticalanalysis

The PK properties of NQ were assessed using non‐compartmental analysis (Kinetica

Version 4.4.1; Thermo LabSystems Inc., Philadelphia, PA, USA) for Group 1 subjects and

the data (not shown) were used to refine the study design for Groups 2 and 3. All NQ

data were subsequently pooled and analysed by population PK methods, as were ART

data which were available for Groups 2 and 3.

In the population PK analysis, logReR concentration‐time datasets of NQ and ART were

analysed by nonlinear mixed effect modelling using NONMEM (Version 6.2.0, ICON

Development Solutions, Ellicott City, MD, USA) with an Intel Visual FORTRAN 10.0

181

compiler. NQ data were available for all three groups and ART were available only for

Groups 2 and 3. Linear mammillary model subroutines within NONMEM, FOCE with ‐

interaction, and the OFV were used to construct and compare plausible models.

Unless otherwise specified, a difference in OFV ≥ 3.84 (2 distribution with 1 d.f.,

P<0.05) was considered statistically significant. Secondary PK parameters including VRSSR,

AUCR0–∞R and elimination tR½R for the participants were obtained from post‐hoc Bayesian

prediction in NONMEM using the final model parameters. Macro constants for the

three compartment model were calculated from the modelled parameters using

previously published equations.333 CRmaxR and TRmaxR were estimated by predicting the

concentration of NQ and ART for each individual at six minute intervals to capture the

post‐dose peak.

Allometric scaling was used a priori with all volume terms scaled using (×(WT/70)1.0)

and all clearance terms using (×(WT/70)0.75).309 IIV was added to parameters for which

it could be estimated from the available data. An additive error model was used for

RUV, approximating proportional error as logReR concentration data were used. In the

development of the final models, the influence of the following covariates on the

various model parameters was investigated: dose group, dose occasion, relative dose

(mg/kg), gender, spleen grade, malaria status (by slide positivity), baseline

logR10R(parasitaemia), age, fever and initial Hb concentration. Covariate relationships

identified using the generalized additive modelling procedure within Xpose332 and by

inspection of correlation plots of eta vs. covariate were evaluated within NONMEM.

The potential effect of these covariates on bioavailability, particularly dose group and

occasion, was also considered in cases where a similar relationship was identified for

all volume and clearance terms, given these were relative to bioavailability. The effect

size (%) of categorical data was assessed while both linear and power relationships

were evaluated for continuous covariates. For linear relationships: individual

parameter value = population parameter value×(1+effect parameter×[covariate value

for individual]‐[median value of covariate]), and for power relationships: individual

parameter value = population parameter value×([covariate value for

individual]/[median value of covariate]effect parameter). A stepwise forward inclusion and

backward elimination method was used with a significance level (P value) of <0.05,

accompanied by a decrease in the IIV of the parameter, was required for inclusion of a

182

covariate relationship, and a P value of <0.01 to retain a covariate relationship. For

relationships involving bioavailability, a fall in the IIV of any volume or clearance terms

was required. Correlations among IIV terms were also investigated and CWRES plots

were assessed in arriving at a final model. Two and 3‐compartment models for NQ and

1‐ and 2‐compartment models for ART were compared with first order absorption,

with and without lag time.

A bootstrap procedure in Perl speaks NONMEM (PsN), stratified according to dose

group and WT, was used to sample individuals from the original dataset, and to

generate 1,000 new datasets that were subsequently analysed using NONMEM. The

resulting parameters were then summarized as median and 2.5th and 97.5th percentiles

(95% empirical CI) to facilitate evaluation of the final model parameter estimates. In

addition, pcVPCs304 were performed using PsN with 1,000 replicate datasets simulated

from the original dataset. The observed 10th, 50th and 90th percentiles were plotted

with their respective simulated 95% CI to assess the predictive performance of the

model.304 As a number of covariate effects were found in the model building process

for NQ, NPCs were performed stratified according to those covariates and were

assessed by comparing the actual with the expected number of data points within the

20, 40, 60, 80, 90 and 95% PI.

Data analysis and representation were performed using SigmaPlot version 11 (Systat

Software Inc., San Jose, CA). Data are mean ± SD unless otherwise indicated. Student’s

t‐test (for parametric data) or Mann‐Whitney U test (non‐parametric data) was used

for two‐sample comparisons as appropriate with a significance level of P<0.05.

183

6.2.4 65B65BResults


Thirteen of 15 Group 1 children completed all essential requirements for the PK

component of the study. All of these children had P. falciparum infections at baseline

and one had a mixed infection with P. vivax at low density (160 parasites/µl of blood).

In Groups 2 and 3, there were four children in each group who were considered to

have a low‐grade parasitaemia on screening microscopy at the study site but who were

subsequently found to be slide‐negative on confirmatory expert microscopy. All

recruited Group 2 and 3 children were included in the PK study. Demographic data are

summarized in Table 6‐1.

The content of NQ in the ARCO tablets was determined by dissolving each tablet (n=5)

in 500 ml water using sonication (twice, for 5 min each time), and measuring the

concentration in 8 aliquots. The ART content was determined after dissolving each

tablet (n=6) in 250 ml methanol and following the same procedure. The mean NQ and

ART contents of the ARCO tablets used in the study were 49 ± 5 mg (nominal potency

50 mg NQ) and 129 ± 3 mg (nominal potency 125 mg ART), respectively.

Table 6‐1 Demographic data for children given artemisinin‐naphthoquine for the treatment of uncomplicated falciparum malaria. Data are mean ± SD unless otherwise indicated.

Group 1 Group 2 Group 3

Number 13 17 16

Gender 6 male:

7 female

11 male:

6 female

12 male:

4 female

Age (years) 7.1 ± 1.8 7.7 ± 2.0 6.7 ± 1.6

Weight (kg) 18.0 ± 3.7 18.9 ± 5.2 16.8 ± 3.2

Height (cm) 110 ± 10 117 ± 12 110 ± 9

Admission parasitaemia

(/µl of blood)a

14,757

(5,189‐41,966)

6,674

(2,264‐19,674)

29,416

(12,290‐70,406)

Naphthoquine dose (mg/kg)

6.3 ± 0.9 8.8 ± 1.4 2 × (9.5 ± 0.9)

Artemisinin dose (mg/kg) 15.7 ± 2.3 22.0 ± 3.6 2 × (23.8 ± 2.2) aGeometric mean and 95% confidence interval for children with parasitaemia

184

6.2.4.2 130B130BNaphthoquinepharmacokineticsandpharmacodynamics

The plasma concentration‐time profiles for NQ are shown in Figure 6‐3. For pooled

data from the three groups, a three‐compartment model proved superior to a two

compartment model with a lower OFV (‐404.855 vs. ‐388.736; P<0.01) and no bias in

the CWRES plot in the initial stages of modelling. As there was no evidence of model

misspecification using a three‐compartment model with first order absorption with a

lag‐time, more complex models were not tested. The structural model parameters

(where C refers to the central compartment and P1 and P2 to the two peripheral

compartments) were kRaR, lag time, CL/F, VRCR/F, VRP1R/F, VRP2R/F, and QR1R/F and QR2R/F.

Estimates for the IIV of kRaR, VRCR, VRP2R, CL and QR2R and correlation between some IIV pairs

Time (h)

0 200 400 600 800 1000 1200

Pla

sm

a n

ap

hth

oq

uin

e (

g/l)

1

10

100A

Time (h)

0 200 400 600 800 1000 1200

Pla

sm

a n

ap

hth

oq

uin

e (

g/l)

1

10

100B

Time (h)

0 20 40 60 80 100

Pla

sm

a n

ap

hth

oq

uin

e (

g/l

ite

r)

1

10

100

Time (h)

0 20 40 60 80 100

Pla

sm

a n

ap

hth

oq

uin

e (

g/l

ite

r)

1

10

100

Time (h)

0 200 400 600 800 1000 1200

Pla

sm

a n

ap

hth

oq

uin

e (

g/l)

1

10

100C

Time (h)

0 20 40 60 80 100

Pla

sm

a n

ap

hth

oq

uin

e (

g/l

ite

r)

1

10

100

Figure 6‐3 Time‐concentration plots of NQ for Group 1 (Panel A), Group 2 (Panel B; milk) and Group 3 (Panel C; water and double‐dose) patients. Inset shows plasma concentration‐time data from 0‐100 h after the dose.

185

(kRaR:VRCR/F, VRCR/F:CL/F, CL/F:VRP2R/F and VRP2R/F:QR2R/F) could be obtained (see Table 6‐2).

Significant covariate relationships were added in the following order (written as

covariate‐parameter [relationship type]); fever‐predicted F [negative categorical], first

dose for Group 3‐predicted F[negative categorical] and Hb‐VRCR/F [positive linear].

Although children in Group 2 were estimated to have an approximately 50% lower kRaR,

this

relationship did not satisfy the significance requirements to be included in the final

model (0.01<P<0.05). Fever (axillary temperature >37.3oC) and the first NQ dose in

Table 6‐2 Population pharmacokinetic parameters and bootstrap results for NQ in children with uncomplicated falciparum malaria.

Parameter Final Model

estimate (RSE%)

Bootstrap

median [95% CI]


kRaR (/h) 1.1 (22) 1.0 [0.7‐1.6]

Lag time (h) 0.7 (7) 0.7 [0.6‐0.8]

VRCR/F (l/70kg) 12,500 (15) 12,200 [9,503‐14,958]

VRP1R/F (l/70kg) 15,500 (19) 17,000 [11,843‐83,415]

VRP2R/F (l/70kg 17,200 (8) 16,000 [10,343‐21,600]

CL/F (l/h/70kg) 51.9 (6) 51.5 [30.1‐58.7]

QR1R/F (l/h/70kg) 40.6 (9) 48.2 [24.2‐113.0]

QR2R/F (l/h/70kg) 398 (13) 407 [318‐536]

% decrease in predicted F with fever 31.8 (21) 31.8 [18.3‐47.1]

% decrease in predicted F with 1st dose in Group 3

26.3 (33) 27.7 [9.3‐40.9]

% increase in VRCR/F per g/dl haemoglobin 16.4 (66) 14.9 [3.2‐19.1]


IIV in kRaR (%) 104 (14) 103 [80‐131]

IIV in VRCR/F (%) 77 (9) 77 [63‐90]

IIV in CL/F (%) 32 (13) 31 [23‐57]

IIV in VRP2R/F (%) 37 (17) 40 [25‐59]

IIV in QR2R/F (%) 52 (41) 50 [6‐84]

R (kRaR, VRCR/F) 0.20 0.25 [‐0.08 – 0.58]

R (VRCR/F, CL/F) 0.47 0.46 [0.04 – 0.72]

R (CL/F, VRP2R/F) 0.50 0.51 [0.09 – 0.86]

R (VRP2R/F, QR2R/F) 0.20 0.18 [‐0.62 – 0.91]

RUV (%) 24 (7) 24 [21‐26]

OFV in final model: ‐687.786, bootstrap OFV: (median [95% CI] ‐712.006 [‐817.316 ‐ ‐615.107]

186

Group 3 were associated with 32% and 26% decreases in bioavailability, respectively.

Every 1 g/dl increase in Hb increased VRCR/F by 16%. Presence of slide positivity at

baseline and logR10R(parasitaemia) were not significant covariates in the model. The

residual error for the model was 24% (see Table 6‐2).Goodness of fit and CWRES plots

for NQ are shown in Figure 6‐4. The results of the parameter estimates and the

bootstrap results are summarized in Table 6‐2and post hoc Bayesian parameter

estimates with derived secondary PK parameters in Table 6‐3. The bootstrap

demonstrated reasonable estimates of structural and covariate effect parameters with

a bias of <10% for all parameters except VRP1R for which bias was 19%. Random

parameters had a bias of <7%. AUC was significantly higher in Group 3 (two doses)

when compared to Groups 1 and 2 (P<0.001) and was higher in Group 2 when

compared to Group 1. When the AUC was normalized for total relative (in mg/kg),

Predicted plasma naphthoquine (g/l)

1 10 100

Ob

se

rve

d p

lasm

a n

ap

hth

oq

uin

e (

g/l)

1

10

100

Time (h)

1 10 100 1000

Co

nd

itio

na

l we

igh

ted

re

sid

ua

ls (

na

ph

tho

qu

ine

)

-6

-4

-2

0

2

4

6

A

B

Figure 6‐4 (A) Population predicted (○) and individual predicted (●) versus observed NQ plasma concentra on (µg/l; log scale) for the final model. The line of identity is shown. (B) Conditional weight residuals vs. time (log scale) for NQ final model.

187

there was no longer any significant difference between the groups. The predicted CRmaxR

was <200 µg/l in all children apart from one Group 3 child with a value of 270 µg/l after

188

Table 6‐3 Post hoc Bayesian parameter estimates and derived secondary pharmacokinetic parameters for NQ in children with uncomplicated falciparum malaria. Data are median [IQR].

aP=0.053 and 0.094 for the comparison between Groups 2 and 1, and Groups 2 and 3, respectively. btR½αR, tR½βR and tR½γR are the first distribution, second distribution and terminal elimination half‐lives respectively. cP<0.01 for comparisons between Groups 3 and 1, and Groups 3 and 1. dP<0.01 for comparison between Groups 3 and 1

Parameter Group 1

(n = 13)

Group 2

(n = 17)

Group 3

(n = 16)

kRaR (/h)a 1.3 [0.9 ‐ 1.6] 0.7 [0.4 ‐ 1.0] 1.7 [0.6 ‐ 2.2]

CL/F (l/h) 17.3 [15.3 ‐ 21.8] 19.5 [16.2 ‐ 25.1] 16.6 [14.9 ‐ 19.2]

VRCR/F (l) 2,115 [1,735 – 2,753] 3,494 [1,817 – 7,818] 3,610 [1,383 – 7,030]

VRP1R/F (l) 3,986 [3,432 – 4,318] 3,986 [3,321 – 4,871] 3,543 [3,183 – 3,903]

VRP2R/F (l) 4,392 [3,602 – 5,208] 4,662 [3,816 – 5,347] 4,370 [3,633 – 4,760]

VRSSR/F (l) 10,464 [9,366 – 13,888] 13,161 [10,485 – 14,053] 12,001 [9,390 – 14,882]

tR½αR (h)b 6.8 [4.4 ‐ 9.2] 8.2 [5.7 ‐ 9.7] 7.3 [5.5 ‐ 12.0]

tR½βR (h)b 109 [92 ‐ 121] 115 [103 ‐ 126] 118 [104 ‐ 130]

tR½γR (h)b 525 [490 ‐ 544] 500 [455 ‐ 629] 595 [525 ‐ 624]

AUCR0–∞R (µg∙h/l)c 5,935 [4,776 – 6,551] 7,104 [5,954 – 7,914] 15,385 [13,200 – 18,486]

AUCR1R/dose (µg∙h/l per mg/kg) 917 [822 ‐ 1158] 728 [611 ‐ 1004] 813 [629 ‐ 999]

Relative bioavailabilityd 1.00 [1.00 ‐ 1.00] 1.00 [0.68 ‐ 1.00] 0.75 [0.75 ‐ 0.87]

Observed day 7 concentration (µg/l)c 7.0 [4.9 ‐ 8.3] 8.1 [7.3 ‐ 9.8] 17.9 [12.0 ‐ 22.9]

Predicted CRmax1R ‐ Dose 1 (µg/l) 40.6 [32.6 ‐ 45.5] 33.9 [14.7 ‐ 52.7] 22.9 [14.1 ‐ 49.1]

Predicted TRmax1R ‐ Dose 1 (h) 3.1 [2.7 ‐ 3.7] 4.6 [3.7 ‐ 7.1] 3.3 [2.4 ‐ 4.8]

Predicted CRmax2R ‐ Dose 2 (µg/l) 57.0 [42.2 ‐ 138]

Predicted TRmax2R ‐ Dose 2 (h) 27.3 [26.7 ‐ 28.3]

189

the second dose. The pcVPC for NQ, shown in, demonstrates reasonable predictive

performance of the model. NPCs stratified according to dose group (three strata), Hb

(three strata) and fever (two strata) showed good predictive performance, with the

expected number of data points above and below most PI (data not shown). As the

dose‐corrected PK parameters were consistent across the three groups (Table 6‐3),

data were pooled to provide estimates from the total of 46 patients. Overall, mean±SD

CL/F, VRssR/F, tR½αR, tR½βR and tR½γR for NQ were 1.30 ± 0.45 l/h/kg, 805 ± 256 l/kg, 8.2 ± 3.8 h,

98 ± 16 h and 518 ± 94 h respectively.

Of the 13 of 15 Group 1 patients included in the PK analysis, seven developed

recurrent parasitaemia during the 42‐day follow up period.408 One was a PCR‐

confirmed recrudescence of P. falciparum, four had reinfections with P. falciparum,

and two had an emergence of P. vivax. In Group 2 there was only one emergent P.

vivax and no P. falciparum recurrence, while there were no episodes of slide positivity

during follow‐up in Group 3. The AUCR0‐RRR and Day 7 concentrations of NQ were

significantly lower in the children with any parasitaemia during follow‐up than in those

who remained free of malaria infection (P=0.001 and P=0.005, respectively). However,

the NQ dose in mg/kg was also significantly lower (P=0.001) and the difference in

AUCR0‐RRR was no longer significant when corrected for dose (P=0.97), indicating that the

lower dose rather than individual PK differences was responsible. Day 7 NQ

concentrations correlated significantly with AUCR0‐RRR overall (r=0.91, P<0.001) and in

each of the three groups (r>0.79; P<0.001).

Time (h)

0 200 400 600 800 1000

Pla

sm

a n

ap

hth

oq

uin

e (

g/l

)

1

10

100

Tim e (h)

0 20 40 60 80 100Pla

sm

a n

aph

tho

qu

ine

(

g/l)

1

10

100

Figure 6‐5 Prediction corrected VPC plots for NQ in children with uncomplicated falciparum malaria, showing the observed 50th (●), 10th and 90th (○) percen les with the simulated 95% CI for the 50th (solid black line), 10th and 90th (dashed grey lines) percentiles. Inset shows plasma concentration‐time data from 0‐100 h after the dose.

190

6.2.4.3 131B131BArtemisininpharmacokinetics

Raw plasma ART concentration‐time data are presented in Figure 6‐6. A two‐

compartment model was superior to a one‐compartment model for ART with a lower

OFV (255.146 vs. 122.637, P<0.01) and an improved CWRES. As there was no evidence

of model miss‐specification using a two‐compartment model with first order

absorption and a lag‐time, more complex models were not tested. The structural

model parameters for ART were kRaR, lag time, CL/F, VRCR/F, VRPR/F and Q/F (inter‐

compartmental clearance for VRPR/F). Estimates for the IIV of CL, VRCR/F, kRaR and lag time

could be estimated and a full covariance matrix was obtained. The correlation between

CL/F and V/F was >0.99 and was fixed to 1. As the CWRES plot revealed plasma

concentrations after the second dose were lower than expected, the effect of dose

occasion on observed F was tested as a negative categorical relationship. The addition

of this relationship reduced the OFV by 46.626 (P<0.001) and reduced the residual

Time (h)

0 12 24 36 48

Pla

sm

a a

rte

mis

inin

(

g/l)

1

10

100

1000

10000A

Time (h)

0 12 24 36 48

Pla

sm

a a

rte

mis

inin

(

g/l)

1

10

100

1000

10000B

Figure 6‐6 Time‐concentration plots of ART for Group 2 (Panel A; milk) and Group 3 (Panel B; water and double‐dose) patients.

191

Table 6‐4 Population pharmacokinetic parameters and bootstrap results for ART in children with uncomplicated falciparum malaria.

Parameter Final Model

estimate (RSE%)

Bootstrap

median [95% CI]


kRaR (/h) 1.8 (110) 1.8 [0.6‐6.5]

Lag time (h) 0.7 (31) 0.7 [0.4‐0.9]

VRCR/F (l/70kg) 1160 (31) 1140 [625‐1520]

VRPR/F (l/70kg) 166 (37) 211 [96.1‐1270]

CL/F (l/h/70kg) 178 (12) 176 [141‐216]

Q/F (l/h/70kg) 14.2 (103) 15.3 [6.6‐52.3]

Covariate effect parameters

% decrease in predicted F with 2nd dose 77.0 (9) 78.6 [63.3‐89.9]


IIV in CL/F (%) 57 (25) 56 [43‐67]

IIV in lag time (%) 23 (169) 21 [5‐57]

IIV in kRaR (%) 139 (81) 141 [72‐230]

IIV in VRCR/F : IIV in CL/F (ratio) 0.995 (28) 0.989 [0.760‐1.671]

R (CL/F, lag time) 0.571 0.565 [0.201‐0.997]

R (CL/F, kRaR) 0.0225 0.011 [‐0.550‐0.628]

R (lag time, kRaR) ‐0.340 ‐0.343 [‐0.956‐0.319]

R (CL/F, VRCR/F) 1 FIXED

RUV (%) 51 (31) 50 [37‐65]

OFV in final model: 85.171, bootstrap OFV: (median [95% CI] 73.924 [‐60.386‐178.368]

error of the model by 7%. The second ART dose had 77% lower bioavailability relative

to the first. No other covariate relationships were identified. The residual error in the

final model was 51% (Table 6‐4).

Goodness of fit and CWRES plots for ART are shown in Figure 6‐7. The results of the

final parameter estimates and the bootstrap results are summarized in Table 6‐4 and

post hoc Bayesian parameter estimates with derived secondary PK parameters in Table

6‐5. The bootstrap demonstrated reasonable estimates of structural and random

parameters with a bias of <8% and <10% respectively, with the exception of VRPR/F

where there was a positive bias of 27%. No significant differences were found in

secondary parameters between the groups although there was substantial variability

192

Predicted plasma artemisinin (g/l)

1 10 100 1000 10000

Ob

se

rved

pla

sma

art

emis

inin

(

g/l)

1

10

100

1000

10000

Time (h)

0 10 20 30 40 50

Co

nd

itio

na

l we

igh

ted

re

sid

ua

ls (

art

emis

inin

)

-6

-4

-2

0

2

4

6

A

B

Figure 6‐7 (A) Population predicted (○) and individual predicted (●) versus observed ART plasma concentra on (µg/l; log scale) for the final model. The line of identity is shown. (B) Conditional weight residuals vs. time (log scale) for ART final model.

within each group. The pcVPC for ART is shown in Figure 6‐8 and demonstrates

reasonable predictive performance of the model.

Time (h)

0 12 24 36 48

Pla

sm

a a

rte

mis

inin

(

g/l

)

1

10

100

1000

10000

Figure 6‐8 Prediction corrected VPC plots for ART in children with uncomplicated falciparum malaria, showing the observed 50th (●), 10th and 90th (○) percen les with the simulated 95% CI for the 50th (solid black line), 10th and 90th (dashed grey lines) percentiles.

193

Table 6‐5 Post hoc Bayesian parameter estimates and derived secondary pharmacokinetic parameters for artemisinin in children with uncomplicated falciparum malaria. Data are median [IQR]. All between‐group comparisons were statistically non‐significant.

Parameter Group 2

(n = 17)

Group 3

(n = 16)

kRaR (/h) 2.0 [0.7 ‐ 3.5] 1.1 [0.8 ‐ 4.1]

CL/F (l/h) 82.1 [76.2 ‐ 74.8] 66.9 [61.5 ‐ 62.4]

VRCR/F (l) 348 [246 ‐ 449] 279 [165 ‐ 324]

Q/F (l/h) 5.13 [4.47 ‐ 5.96] 4.69 [4.33 ‐ 5.05]

VRPR/F (l) 42.7 [35.6 ‐ 52.2] 37.9 [34.1 ‐ 41.8]

VRSSR/F (l) 388 [289 ‐ 482] 315 [202 ‐ 362]

tR½αR (h) 2.8 [2.7 – 3.0] 2.7 [2.5 ‐ 2.8]

tR½βR (h) 6.8 [6.2 – 7.0] 6.6 [6.5 ‐ 6.9]

AUCR1R ‐ Dose 1 (µg∙h/l) 5,127 [3,631 – 8,237] 6,770 [5,249 – 10,235]

AUCR1R/dose (µg∙h/l per mg/kg) 267 [170 ‐ 340] 281 [202 ‐ 417]

AUCR2R ‐ Dose 2 (µg∙h/l) 1557 [1207 ‐ 2354]

AUCR0–∞R (µg∙h/l) 8,327 [6,457 – 12,590]

Predicted CRmax1R ‐ Dose 1 (µg/l) 843 [522 – 1,353] 1,105 [736 ‐ 1398]

Predicted TRmax1R ‐ Dose 1 (h) 2.1 [1.6 ‐ 2.9] 2.5 [1.5 ‐ 3.0]

Predicted CRmax2R ‐ Dose 2 (µg/l) 269 [179 ‐ 345]

Predicted TRmax2R ‐ Dose 2 (h) 26.6 [26.0 – 27.4]

The dose‐corrected, first‐dose data indicated that the median AUC for ART was 5%

higher in Group 3 than Group 2. However these and other PK parameters for the two

groups were not significantly different (see Table 6‐5), hence the data were pooled for

total patient group estimates. Overall, mean (±SD) CL/F, VRssR/F and tR½αR for ART were 4.1

± 2.0 l/h/kg, 21 ± 10 l/kg and 2.7 ± 0.3 h respectively. Although the best PK model was

two‐compartment with tR½βR 6.7 ± 0.5 h, this may be a spurious finding due to the

limited concentration‐time data in the present study design and 27% bias in the

bootstrap for VRPR/F.

194


The present study has provided the first paediatric PK data for NQ and additional ART

disposition data to complement the few available for this age‐group. NQ given in the

form of ART‐NQ fixed combination therapy was promptly absorbed (mean absorption

tR½R 1.0 h) and reached a predicted CRmaxR that was <200 µg/l in all but one child even

after the second dose in Group 3. The mean elimination tR½R of NQ (524 h) was longer

than estimates in early reports (41‐57 h)217 and in the recent Chinese adult volunteer

study (156‐299 h).220 There was some evidence of a modest increase in NQ

bioavailability when administered with a small amount of fat, in contrast to the

substantial food‐associated reduction in NQ bioavailability in Chinese adults.220 The

CL/F (1.1 l/h/kg) and V/F (71 l/kg) for NQ in our study were lower than the results

reported in healthy Chinese adults (7.0 l/h/kg and 2,277 l/kg),220 but there are no other

data presently available for direct comparison. In the case of ART, the mean CL/F and

V/F (4.1 l/h/kg and 21 l/kg, respectively) were comparable to those in most previous

studies (means 6.7 l/h/kg and 27 l/kg, respectively).112, 115, 118, 119, 123, 125, 404, 406

The long elimination tR½R and high V/F of NQ in our children were consistent with most

other quinolines and related drugs in clinical use.387, 413, 414 PK modelling indicated that

a three‐compartment model best described the disposition of NQ in the present study.

This finding is consistent with similar PK studies involving CQ 313, 415‐417 and PQ.216 A

number of studies of quinoline and related antimalarial drugs have shown biphasic

drug concentration‐time profiles that can be analysed using a two‐compartment

model.174, 178, 215, 387, 414, 418‐421 Improved PK study design, including more frequent and

longer duration sampling, as well as lower limits of quantification for the analytical

techniques, may explain why recent studies such as ours reveal more complex

elimination kinetics. Indeed, the relatively short NQ elimination tR½R in the Chinese

volunteer study220 could be explained by a short sampling period (216 h) as well as the

use of non‐compartmental methods. In relation to the latter point, we found an

elimination tR½R of 298 h by non‐compartmental methods in Group 1 patients vs. 547 h

in compartmental population analyses of pooled NQ data.

The effect of fat on NQ bioavailability in the present study needs interpretation in light

of the study design. In the preliminary PK study in Group 1 children, there was no

195

requirement for fasting before or after drug administration. It is likely that these

children consumed some fat around the time ART‐NQ was given even though the dose

was administered with water. Group 3 children, who were required to fast throughout

and were given the dose with water, had a 26% lower relative bioavailability compared

to Group 1 and also Group 2 which was the group in which all children were given ART‐

NQ with milk. This evidence of a modest positive effect of fat on bioavailability

contrasts with the observation that the AUC and tR½R of NQ were approximately 50%

lower after food (60% lipid; 2400 kJ) in healthy Chinese adults,220 suggesting an

increased CL and/or reduced oral bioavailability. Studies with quinolines and related

drugs have shown increased absorption with high‐fat meals215, 375, 402 but a standard

Vietnamese meal (17 g fat; 2000 kJ) had little effect on the PK properties of PQ.173 It is

possible that relatively high fat meals such as that used by Qu et al.220 might interfere

with NQ absorption from the gastrointestinal tract, but our experience is that amounts

of fat given as milk greater than that used in the present study (>8.5 g; >615 kJ) have a

high likelihood of inducing significant nausea in an unwell child with malaria. This

observation and our NQ PK data do not suggest that food‐associated under‐dosing will

be problematic in children. The improved NQ bioavailability after the second

compared with the first dose in Group 3 may relate to clinical improvement reflecting

parasite clearance as has been seen with LUM.141

Fever was independently associated with reduced NQ bioavailability, consistent with

PK studies in other contexts.422, 423 There was an independent association between Hb

and VRCR that might suggest NQ accumulation in RBCs but an assessment of partitioning

was beyond the scope of the present study. As is the case for a range of other drugs

including anti‐infectives,424 co‐administration of milk reduced the rate of NQ

absorption.

Relative to previous studies of ART PK112, 115, 118, 119, 123, 125, 404, 406 and compared with

ART monotherapy, Qu et al.220 reported lower values for CL/F (mean 2.7 l/h/kg) in

adults given ART‐NQ while Sidhu et al.125 found a significantly higher value for this

parameter (14.4 l/h/kg) when ART was given to children with uncomplicated

falciparum malaria. While it is difficult to explain the former observation, an

apparently high CL/F may relate to underestimation of the ART AUC. Almost all

196

previous studies have used non‐compartment analysis or one‐compartment models to

determine the PK parameters for ART. A two‐compartment model was, however, the

best fit for the ART concentration‐time data in the present study, probably reflecting

the fact that our limits of quantification (2.5 µg/l) and detection (1 µg/l) were

considerably lower than previous studies (4‐20 µg/l).112, 123, 125, 404, 406 A prolonged

elimination phase may have been undetected in past studies, thus truncating the AUC.

Our assay sensitivity led, in part, to an unexpected limitation of the present study,

namely a lack of sampling >24 h post‐dose. Based on the established PK properties of

ART, we anticipated that ART plasma concentrations would not be detectable beyond

24 h. If prolonged sampling had been performed, this would have allowed a more

definitive multi‐compartment PK characterization.

In a Vietnamese study, co‐ingestion of food was reported to be associated with a non‐

significant 20% reduction in ART AUC after oral ART in healthy adults.404 By contrast,

Qu et al.220 reported that the AUC and tR½R of ART were approximately 75% higher after

co‐administration of food with ART‐NQ combination therapy, suggesting an increased

bioavailability and possible reduction in CL. Our data are consistent with the earlier

study of Dien et al.404, as we also found a non‐significant 5% lower AUC for ART after

food (milk) compared with administration with water. There were no significant

differences when dose group was added as a covariate in the population PK model,

further evidence that fat has no clinically meaningful effect on the PK of ART.

Although the importance of developing paediatric formulations of antimalarial drugs

has been emphasized,425 it is unclear how the manufacturer’s paediatric ART‐NQ dose

recommendations have been developed. Using either a WT‐based equation426,

(DoseRCHILDR (mg) = DoseRADULTR (mg) × [WTRCHILDR/ WTRADULTR]0.75), or a body surface area (BSA)

equation426, 427 (DoseRCHILDR (mg) = DoseRADULTR (mg) × [BSARCHILDR/BSARADULTR])where the

regular adult dose of NQ is 400 mg, adult WT is assumed to be 50 kg and adult BSA is

1.73 m2, the adult dose of 8 mg/kg would scale up to ≥10 mg/kg in children. Our initial

mean conservative dose of 6.3 NQ mg/kg in Group 1 as part of ART‐NQ was associated

with a relatively high late treatment failure rate.408 The regimens used in Groups 2 and

3 (means 9.0 and 9.5 mg/kg NQ per dose) were based on manufacturer’s

197

recommendations of 6.5‐9.5 mg/kg for children up to 40 kg409, 428 which is still short of

the allometrically scaled dose of ≥10 mg/kg.

Efficacy against asexual parasite forms over 42 days of follow‐up in Groups 2 and 3 was

100% but prolonged gametocyte carriage was observed in some patients.408 This latter

observation, concerns regarding the emergence of parasite resistance against

artemisinin compounds in endemic areas with a history of sub‐therapeutic drug use,429

the implication that higher individual paediatric doses than recommended by the

manufacturer can be used, and the safety of the two‐dose ART‐NQ regimen employed

in Group 3,408 are all considerations that make an argument for a three‐day ART‐NQ

regimen in line with WHO recommendations for all ACT.430 We have used the final NQ

model to simulate CRmaxR after three ART‐NQ doses given with milk to 1,000 children

with similar characteristics to the present subjects. The median [95% PI] after three

consecutive daily doses were 36 [19‐76], 69 [44‐128] and 89 [61‐152] µg/l,

respectively, with an absolute range up to 350 µg/l after the third simulated dose. A

predicted CRmaxR >300 µg/l occurred in a small minority of subjects in the simulation. The

Group 3 child with a predicted CRmaxR of 270 µg/l had an uncomplicated clinical course in

the present study and a CRmaxR of 245 µg/l in an adult was not reported to be associated

with toxicity,217 but careful tolerability and safety monitoring would need to be carried

out if a three dose regimen was implemented.

A further argument for multiple dose ART‐NQ relates to the disposition of the ART

component. The conventional adult dose regimen for orally administered ART of 10‐20

mg/kg on the first day followed by 500 mg daily for 4 days407 has been questioned due

to the auto‐induction of ART metabolism that, as in the present study, progressively

and substantially reduces the bioavailability of subsequent doses but does not increase

CL.123 The 15‐24 mg/kg dose of ART used in the present study could, therefore, be

appropriate part of a three‐day ART‐NQ regimen based on the single dose now

recommended by the manufacturer.

The present study had limitations, in part because of the present paucity of PK and

other data relating to ART‐NQ (especially when Group 1 was recruited) but also the

context of a paediatric study in the rural tropics. The sampling schedule could have

198

included more time points after the second dose in Group 3 but relatively robust

estimates for model parameters could still be derived. It was unfortunate that no pure

vivax malaria cases were recruited but the fact that there was only one late P. vivax

infection in Groups 2 and 3 suggests that the long NQ tR½R helps prevent the emergence

of this infection that is seen after other therapies for falciparum malaria in this area

including AL.21

In conclusion, when normalized by WT, the PK parameters for ART in children are

comparable to most previous studies in adults, but CL/F was higher than recently

reported data when ART‐NQ was co‐administered to healthy adults.220 By contrast,

CL/F and V/F for NQ was lower in the present study and the terminal elimination tR½R

was longer at a mean of 21.8 days. Although the predicted bioavailability of the first

dose of NQ was lower in a fasted state, this is unlikely to translate into clinical

meaningful effects. The present PK characterization, as well as associated tolerability,

safety and preliminary efficacy data,408 may justify using the currently recommended

single dose of ART‐NQ for three days in children with uncomplicated malaria.

199


We thank the children and their parents/guardians for their participation. We are also

most grateful to Sister Valsi Kurian and the staff of Alexishafen Health Centre for their

kind co‐operation during the study and to Dr Michele Senn and staff of the Papua New

Guinea Institute of Medical Research for clinical and logistic assistance. Valuable

technical support was provided by Mr Michael Boddy and Mr John Hess, School of

Pharmacy, Curtin University. The study was funded by the National Health and Medical

Research Council (NHMRC) of Australia (grant 634343). STL was the recipient of a

Cranmore Undergraduate Scholarship through the Faculty of Medicine, Dentistry and

Health Science, University of Western Australia, and TMED is supported by an NHMRC

Practitioner Fellowship.

6.2.7 68B68BConflictofintereststatement

FWH has received research funding from Kunming Pharmaceuticals, the manufacturers

of ARCO.

200

201

7 6B6BGeneralDiscussionMalaria continues to be a major global health concern. In PNG there is high

transmission of malaria in coastal areas that puts pregnant women, infants and

children at the highest risk. New pharmaceutical strategies are being employed that

aim to prevent malaria in pregnancy and infancy and treat acute malaria effectively in

childhood. These populations are also those for which the pharmacology of these

treatments is not well characterised. Therefore, it was the general aim of the thesis to

add to the available literature regarding the pharmacology of selected important

antimalarials in pregnancy, infancy and childhood.

202

7.1 21B21BSignificanceoffindings

The aims of this thesis as set out in the outline (section 1.4) were achieved, with five

successful small scale studies performed, interpreted, presented and, in some cases,

used to guide health policy and further investigations. The PK analysis in all studies

used a population approach based on the computer program NONMEM.

7.1.1 69B69BPreventionofmalariainpregnancy

The study presented in Chapter 2 was able to adequately characterise the PK of AZI

when given with CQ or SP in pregnant and non‐pregnant women using a three

compartment model with a sequential zero then first order absorption. When tested

as a categorical covariate, pregnancy was found only to influence the VRCR which was

almost doubled in the pregnant state. This resulted in pregnant women having lower

peak concentrations, but they had similar overall exposure (AUCR0–∞R) and almost

parallel plasma concentration‐time curves after the first few days. Additionally, there

was no difference in AZI disposition between the groups that were co administered CQ

or SP. Therefore, there was no evidence for a dose adjustment of AZI in pregnancy or

when AZI was given with CQ and SP.

The most common side‐effects encountered in this study were nausea and vomiting.

This was not surprising given prior experience with AZI,239 and is a potential barrier to

its implementation in malaria where higher doses are likely required for sufficient

efficacy. The preliminary efficacy data for both groups in this study, with no woman

experiencing parasitaemia in the 42 days after drug administration, are promising,

although by no means a true measure of the efficacy of these combinations. It was also

noted that the 96h concentrations of AZI correlated well with the AUCR0–∞R and could

therefore be used as a surrogate marker of exposure in efficacy trials where extensive

sampling is not feasible.

The results of this study have been used in determining the dose regimen for a large

scale IPTp trial of AZI with SP.431 Given the number of participants experiencing nausea

with 2 g as a single dose in this study, 1 g of AZI is given twice daily for two day in this

ongoing IPTp trial.

203

7.1.2 70B70BPreventionofmalariaininfancy

Chapter 3 contained the results of a study of conventional and double dose SP in

infants. The population PK of PYR, SDX and NSX, a metabolite of SDX, that incorporated

maturational processes were successfully characterised using a sparse sampling

design. A two compartment model with first‐order absorption for PYR was obtained

with an estimated maturation half‐time of 318 days and a Hill coefficient of 7.39. For

SDX, a one compartment model with first‐order absorption and a maturation half‐time

of 271 days with a Hill coefficient of 4.07 was obtained. When the NSX data were

added, only one additional compartment was required to characterize its disposition.

Overall, the AUCR0‐∞R of both SDX and PYR was significantly higher in the double‐dose

group despite a 32% reduction in the relative bioavailability of SDX when the dose was

doubled, possibly due to saturation of absorption.

The double dose was found to be well tolerated and safe in these infants with no

significant symptomatology, as reported by the parent, and no changes in measured

biochemical markers over time. Preliminary efficacy data demonstrated that fewer

children in the double dose group have symptomatic malaria during follow up.

Although double dose of SP resulted in higher exposure to the drugs and was safe and

well tolerated, the study found that the AUCR0‐∞R for both drugs in the conventional

dosing group was higher than previously reported in children and similar to that

reported in adults from other areas. This may be explained, in part, by methodological

differences between this and previous studies, as the AUCR0‐∞R for PYR in the

conventional dose group was approximately half that reported for non‐pregnant

women in the same area.432

Regardless, the findings of this study will assist with the interpretation of an ongoing

IPTi study performed in PNG (results forthcoming) and provide a basis for future

investigations of higher doses of SP in infants.

204

7.1.3 71B71BTreatmentofuncomplicatedmalariainchildren

This thesis contained three studies of ACT treatments for the treatment of

uncomplicated malaria in children.

7.1.3.1 132B132BArtemether‐lumefantrine

In Chapter 4, results from a study of the widely used combination AL were presented.

In this study, the PK of the two components with their active metabolites, DBL for LUM

and DHA for ART, were analysed using NONMEM. The simultaneous modelling of LUM

and DBL was achieved using a model with three compartments for LUM, two

compartments for DBL, first‐order absorption and an element of FP metabolism. This

represented the first population PK model of DBL and the first description of its

disposition in children. For the ARM and DHA model, there were a substantial number

of BLQ observations which called for the use of a likelihood‐based method to allow

them to be retained in the analysis. This method was successfully applied to the data

to obtain a model with two compartments for ARM, one compartment for DHA, first‐

order absorption and a dose dependent clearance for ARM.

Exposure of the children to LUM, ARM and DHA was similar to that previously reported

for adults with malaria, although children receive an average 35% higher mg/kg dose.

Given that a higher dose is appropriate in children, as would be expected from the

principle of allometry, an increased dose is suggested for children weighing 12.5‐15 kg

who currently receive a lower mg/kg dose than a 50kg adult. This change is being

reviewed for use in PNG, where the manufacturers recommended dose regimen is

used.

The metabolic ratio of DBL was similar to that reported previously.137, 178 DBL was

found to play a potential role in determining treatment outcome, outlining the need

for its measurement in future efficacy trials to assess the extent of its contribution to

efficacy. Although the model demonstrated a suitable predictive performance for

LUM, this was not true for DBL where the concentrations were higher in children under

5 years of age than would be predicted.

205

The results from this study were taken into consideration when the results of a large

efficacy trial in the same area resulted in the implementation of AL as the first‐line

treatment for falciparum malaria in PNG.

7.1.3.2 133B133BArtemisinin/piperaquinebase

A comparison of two ACT drugs containing PQ was presented in Chapter 5. A PK model

for PQ was established using historical data from children who received DHA/PQ

tetraphosphate and new data from children who received ART/PQ base. This

comprised of three compartments with a transit compartment model to explain the

varied absorption phase. There were no significant PK differences between the two

formulations and the AUCR0‐∞R of PQ was similar between the two groups. Analysis of

the ART concentrations in the ART/PQ base group found a two compartment model

with first‐order absorption and a dose dependent relative bioavailability to be

adequate. The second dose of ART resulted in a 73% reduction in its bioavailability,

and therefore the contribution of the second dose to the total AUCR0‐∞R was much

smaller than that of the first.

The new combination was found to be well tolerated and reasonably effective,

particularly against appearance of P. vivax parasitaemia.

In light of evidence for a higher dose of PQ in children212, 213 and WHO

recommendations, 69 the evaluation of an extension of the currently recommended

two day regimen for ART/PQ base to a three day regimen is supported by the results of

the study.

7.1.3.3 134B134BArtemisinin/naphthoquine

The final original data chapter of this thesis contained a PK evaluation of ART/NQ, an

ACT for which there is limited current information in the literature. NQ concentrations

were adequately described by a three compartment model with first‐order absorption

and a lag time. Fever at the time of administration reduced bioavailability. Hb was also

found to be a significant covariate with a positive relationship with the VRCR.

Bioavailability was also found to be lower for the first dose in the fasted group,

although there was no difference in the dose‐adjusted total AUCR0‐∞R. ART

206

concentrations were also available for two out of the three groups and were described

by a two compartment model with first‐order absorption and a lag time. Once again

there was a reduction in the relative bioavailability of the second dose, similar to the

finding for ART/PQ base. There were considerable differences in the NQ PK results in

this study when compared to those obtained in healthy adults (the only published PK

available for comparison),220 although there were significant methodological

limitations in the latter study. Results for ART PK were similar to those previously

reported in patients with malaria.

These results, combined with those of the safety, tolerability and efficacy (not a part of

this thesis), were used in determining the dose regimen of ART/NQ for an ongoing

large scale efficacy trial in PNG.

207

7.2 22B22BImprovementsandfuturedirections

With the benefit of time, experience and further consideration, a number of potential

improvements to the analysis of the data presented in this thesis have been

considered by the author. These are presented here alongside potential future

avenues of research stemming from the studies presented here.

208

7.2.1 72B72BPreventionofmalariainpregnancy

Improvement: Although the model presented in Chapter 2 was adequate in describing

the concentration‐time data for AZI after the first few hours, there was still some

model misspecification in regards to the absorption phase. A transit compartment

model, resembling that used for PQ in Chapter 5, may have been more appropriate

and more likely to provide a better account of the absorption phase of AZI.

Improvement: Like many other studies of PK in pregnancy, the categorical relationship

of pregnancy status was evaluated within the covariate testing procedure. In fact, as

the pregnant women were heavier and an allometric model was used, it would be

more appropriate to state that the use of allometry adequately described the PK

difference in pregnancy with the exception of VRCR/F. For the VRCR/F, not only could the

relationship have been presented as a percentage, but the effect of the size difference

between the ideal WT of the pregnant women and their actual WT (ie. the additional

WT due to pregnancy) could have been tested as a continuous covariate. This would

have at least provided a more physiologically related description of the effect of

pregnancy on VRCR/F.

Future direction: Although a correlation between the 96 h concentrations and the

AUCR0–∞R was found, providing a reasonable surrogate in studies where it is not feasible

to determine the latter, Bayesian forecasting is potentially a more informative

alternative. This would provide an estimate of all model parameters for an individual

given the value of their covariates that were influential in the developed model. In an

efficacy trial, not only would this be able to provide a more individualised estimate of

AUCR0–∞R but it would enable more complex comparisons to be made, such as time to

fall below a defined concentration. Nevertheless, the results obtained are only as good

as information available, so if only one concentration measurement was available

some days after the dose, predictions relating the absorption phase would not be as

reliable as those for the elimination phase.

209

7.2.1 73B73BPreventionofmalariaininfancy

Improvement: This paper used a sigmoid ERmaxR model to account for the maturation of

hepatic clearance of the drugs, with excellent parameter precision for estimates of

maturation half‐time (RSE of 8% for PYR and SDX) but not for estimates of the Hill

coefficient (RSE of 43% and 52% for PYR and SDX, respectively. Although a larger age

range may have assisted with these estimates, as stated in the paper, an alternative

would be to use the $PRIOR subroutine within NONMEM to help stabilise the

estimates of these parameters.433 Using this method the Hill coefficient from other

studies reporting hepatic maturation would be included in the NONMEM control

stream and the OFV would be penalized for deviating from this prior information.

Future direction: There is no doubt that there may be a role for a higher dose of SP in

infants. Given the preliminary safety, tolerability and efficacy data presented in this

thesis a comparison of a higher dose with a standard dose IPTi would be warranted.

Unfortunately such studies are difficult and expensive to run and no country has

adopted IPTi as policy despite the current evidence in its favour.

210

7.2.2 74B74BTreatmentofuncomplicatedmalariainchildren

Improvement: In all three studies in children, no calculation of sample size was

performed prior to the study. This is particularly important in the PQ and NQ studies in

which comparisons were made between different groups. It may also have assisted in

the determination of the appropriate sample size for the exploration of covariate

relationships in the PK analysis.

Improvement: In the AL study dose dependent changes in the PK of ARM were

described using a model that account for an increased clearance (including metabolic

conversion to DHA) with each subsequent dose. Although a similar model had been

previously reported, no basis for this relationship is provided.142 In fact, the process

may resemble that of ART, which is best explained by a decreased in relative

bioavailability (presumably by FP metabolism). In the case of ARM this would also be

able to account for the increasing concentrations of DHA seen with subsequent doses.

Although the sample schedule was not ideal for the characterization of ARM

disposition, this alternative explanation is plausible and could have been tested.

Improvement: The use of Bayesian forecasting, as described above, would be an

addition to the data provided in the AL study. Similar to the type of prediction based

forecasting from simulated individuals performed in the ART/NQ study to obtain

estimates of CRmaxR, simulated individuals could be used to compare current and

proposed dose regimens. This would be particularly useful for AL as paediatric

formulations which allow for more precise dosing are available. A more complex dose

regimen based on WT could then be suggested to be used with these formulations.

Similar modelling would also assist with developing paediatric dose recommendations

for the other combination studies in this thesis.

Future direction: As indicated for AZI, Bayesian forecasting of real individuals in

efficacy trials could be performed to provide more informative estimates of PK

parameters. This is particularly pertinent for AL and ART/NQ which are the subject of

an ongoing efficacy trial in PNG. Although the children in these trials are younger, 0.5‐5

years old versus 5‐10/12 years, the data could first be externally validated for the

younger individuals. This was done for LUM for children in an earlier efficacy trial,

211

where the developed model showed good predictive performance (see Chapter 4).

Assessment of this method compared to a single concentration (day 7 for example) will

be required.

Future direction: Fever was found to be associated with low bioavailability for NQ,

despite parasitaemia also being tested as a potential covariate in both of these cases.

This suggests an independent role of fever on the absorption of these lipid soluble

drugs. Currently few data regarding the effects of fever on PK exist in the literature,

with no clear mechanism for the changes noted. There is potential for this covariate to

be assessed in future PK evaluations of these and other antimalarial drugs.

Additionally, there is scope to investigate the mechanism by which fever affects

bioavailability.

Future direction: The very rapid change in relative bioavailability of ART seen in the

two studies of its PK is an interesting phenomenon. As it has been previously shown to

effect bioavailability and not clearance, it is unlikely that it is due to an increase in

hepatic metabolism.122 This effect would likely be an affect of the gut wall, which also

contains metabolising enzymes. This hypothesis could be tested with the use of

grapefruit juice, a potent inhibitor of CYP3A4 in gut wall and not in the liver.

The only other study that presents data for the second dose of the ART found apparent

discordant results.118 When given alone for three days, followed by a dose of MQ,

there was only decreased in the bioavailability if the third, and final, dose of ART.118 In

contrast, when MQ was given with the first dose of ART there was a reduction in the

bioavailability of the second and third doses of ART.118 This result along with the

results of those presented in this thesis, suggest that, when co‐administered with

drugs such as NQ, PQ and MQ, the effects of the autoinduction of ART are accelerated.

A study involving the dosing of ART alone and with various other drugs could provide

insights into potential drug‐drug interactions. These suggested studies could be

performed in healthy volunteers.

212

213

x 149B149BReferences

1. Salman, S., et al., Pharmacokinetic properties of conventional and double‐dose sulfadoxine‐pyrimethamine given as intermittent preventive treatment in infancy. Antimicrob Agents Chemother, 2011. 55(4): p. 1693‐700.

2. Salman, S., et al., Population pharmacokinetics of artemether, lumefantrine, and their respective metabolites in Papua New Guinean children with uncomplicated malaria. Antimicrob Agents Chemother, 2011. 55(11): p. 5306‐13.

3. Salman, S., et al., Pharmacokinetic properties of azithromycin in pregnancy. Antimicrob Agents Chemother, 2010. 54(1): p. 360‐6.

4. Salman, S., et al., A pharmacokinetic comparison of two piperaquine‐containing artemisinin combination therapies in Papua New Guinean children with uncomplicated malaria. Antimicrob Agents Chemother, 2012. 56(6): p. 3288‐97.

5. Batty, K.T., et al., Artemisinin‐naphthoquine combination therapy for uncomplicated pediatric malaria: A pharmacokinetic study. Antimicrob Agents Chemother, 2012. 56(5): p. 2472‐84.

6. World Health Organisation, World Malaria Report: 2011. 2011. 7. World Health Organisation, World Malaria Report: 2005. 2005. 8. Black, R.E., S.S. Morris, and J. Bryce, Where and why are 10 million children

dying every year? Lancet, 2003. 361(9376): p. 2226‐34. 9. Cox‐Singh, J., et al., Plasmodium knowlesi malaria in humans is widely

distributed and potentially life threatening. Clin Infect Dis, 2008. 46(2): p. 165‐71.

10. Rosenberg, R., et al., An estimation of the number of malaria sporozoites ejected by a feeding mosquito. Trans R Soc Trop Med Hyg, 1990. 84(2): p. 209‐12.

11. Ponnudurai, T., et al., Feeding behaviour and sporozoite ejection by infected Anopheles stephensi. Trans R Soc Trop Med Hyg, 1991. 85(2): p. 175‐80.

12. Miller, L.H., et al., The pathogenic basis of malaria. Nature, 2002. 415(6872): p. 673‐9.

13. Ishino, T., et al., Cell‐passage activity is required for the malarial parasite to cross the liver sinusoidal cell layer. PLoS Biol, 2004. 2(1): p. E4.

14. Hulden, L., Activation of the hypnozoite: a part of Plasmodium vivax life cycle and survival. Malar J, 2011. 10: p. 90.

15. Richter, J., et al., What is the evidence for the existence of Plasmodium ovale hypnozoites? Parasitol Res, 2010. 107(6): p. 1285‐90.

16. Berendt, A.R., D.J. Ferguson, and C.I. Newbold, Sequestration in Plasmodium falciparum malaria: sticky cells and sticky problems. Parasitol Today, 1990. 6(8): p. 247‐54.

17. Carvalho, B.O., et al., On the cytoadhesion of Plasmodium vivax‐infected erythrocytes. J Infect Dis, 2010. 202(4): p. 638‐47.

18. Anstey, N.M., et al., Lung injury in vivax malaria: pathophysiological evidence for pulmonary vascular sequestration and posttreatment alveolar‐capillary inflammation. J Infect Dis, 2007. 195(4): p. 589‐96.

19. Muller, I., et al., The epidemiology of malaria in Papua New Guinea. Trends Parasitol, 2003. 19(6): p. 253‐9.

20. Cattani, J.A., et al., The epidemiology of malaria in a population surrounding Madang, Papua New Guinea. Am J Trop Med Hyg, 1986. 35(1): p. 3‐15.

214

21. Karunajeewa, H.A., et al., A trial of combination antimalarial therapies in children from Papua New Guinea. N Engl J Med, 2008. 359(24): p. 2545‐57.

22. Pell, C., et al., Community response to intermittent preventive treatment of malaria in infants (IPTi) in Papua New Guinea. Malar J, 2010. 9: p. 369.

23. IPTI consortium. [cited 2011 18/12]; Available from: 8TUhttp://www.ipti‐malaria.orgU8T.

24. Steketee, R.W., et al., The burden of malaria in pregnancy in malaria‐endemic areas. Am J Trop Med Hyg, 2001. 64(1‐2 Suppl): p. 28‐35.

25. Doolan, D.L., C. Dobano, and J.K. Baird, Acquired immunity to malaria. Clin Microbiol Rev, 2009. 22(1): p. 13‐36, Table of Contents.

26. Langhorne, J., et al., Immunity to malaria: more questions than answers. Nat Immunol, 2008. 9(7): p. 725‐32.

27. Bardaji, A., et al., Impact of malaria at the end of pregnancy on infant mortality and morbidity. J Infect Dis, 2011. 203(5): p. 691‐9.

28. Garner, P. and A.M. Gülmezoglu, Drugs For Preventing Malaria‐Related Illness In Pregnant Women And Death In The Newborn (Cochrane Review), in The Cochrane Library, Issue 3. 2003, Oxford Update Software.

29. Davis, T.M., I. Mueller, and S.J. Rogerson, Prevention and treatment of malaria in pregnancy. Future Microbiol, 2010. 5(10): p. 1599‐613.

30. Parise, M.E. and et al, Efficacy Of Sulfadoxine‐Pyrimethamine For Prevention Of Placental Malaria In An Area Of Kenya With A High Prevalence Of Malaria And Human Immunodeficiency Virus Infection. American Journal of Tropical Medicine and Hygiene, 1998(5): p. 813‐822.

31. van Eijk, A.M. and et al, Effectiveness of Intermittent Preventive Treatment With Sulphadoxine‐Pyrimethamine for Control of Malaria In Pregnancy In Western Kenya: A Hospital‐Based Study. Tropical Medicine and International Health, 2004. 9(3): p. 351‐360.

32. Abou‐Zahr, C.L. and T.M. Wardlaw, Antenatal Care in Developing Countries : Promises, Achievements and Missed Opportunities : An Analysis of Trends, Levels and Differentials, 1990‐2001. 2003, Geneva, Switzerland: World Health Organisation Press.

33. Greenwood, B.M., et al., The effects of malaria chemoprophylaxis given by traditional birth attendants on the course and outcome of pregnancy. Trans R Soc Trop Med Hyg, 1989. 83(5): p. 589‐94.

34. World Health Organisation, Technical Expert Group meeting on intermittent preventive treatment in pregnancy (IPTp). 2008: World Health Organisation Press.

35. Mola, G., Manual of Standard Managements in Obsetrics and Gynaecology for Doctors, HEOs and Nurses in Papua New Guinea. 5 ed. 2002, Port Moresby: SalPress.

36. Harrington, W.E., et al., Intermittent treatment to prevent pregnancy malaria does not confer benefit in an area of widespread drug resistance. Clin Infect Dis, 2011. 53(3): p. 224‐30.

37. Feng, G., et al., Decreasing burden of malaria in pregnancy in Malawian women and its relationship to use of intermittent preventive therapy or bed nets. PLoS One, 2010. 5(8): p. e12012.

38. ter Kuile, F.O., A.M. van Eijk, and S.J. Filler, Effect of sulfadoxine‐pyrimethamine resistance on the efficacy of intermittent preventive therapy for malaria control during pregnancy: a systematic review. JAMA, 2007. 297(23): p. 2603‐16.

215

39. Zhou, Z., et al., Decline in sulfadoxine‐pyrimethamine‐resistant alleles after change in drug policy in the Amazon region of Peru. Antimicrob Agents Chemother, 2008. 52(2): p. 739‐41.

40. Gama, B.E., et al., Chloroquine and sulphadoxine‐pyrimethamine sensitivity of Plasmodium falciparum parasites in a Brazilian endemic area. Malar J, 2009. 8: p. 156.

41. Sevene, E., R. Gonzalez, and C. Menendez, Current knowledge and challenges of antimalarial drugs for treatment and prevention in pregnancy. Expert Opin Pharmacother, 2010. 11(8): p. 1277‐93.

42. White, N.J., R.M. McGready, and F.H. Nosten, New medicines for tropical diseases in pregnancy: catch‐22. PLoS Med, 2008. 5(6): p. e133.

43. Le Hesran, J.Y., et al., Maternal placental infection with Plasmodium falciparum and malaria morbidity during the first 2 years of life. Am J Epidemiol, 1997. 146(10): p. 826‐31.

44. Schwarz, N.G., et al., Placental malaria increases malaria risk in the first 30 months of life. Clin Infect Dis, 2008. 47(8): p. 1017‐25.

45. Malhotra, I., et al., Can prenatal malaria exposure produce an immune tolerant phenotype? A prospective birth cohort study in Kenya. PLoS Med, 2009. 6(7): p. e1000116.

46. Oduwole, O.A., et al., Congenital malaria in Calabar, Nigeria: the molecular perspective. Am J Trop Med Hyg, 2011. 84(3): p. 386‐9.

47. Obiajunwa, P.O., J.A. Owa, and O.O. Adeodu, Prevalence of congenital malaria in Ile‐ife, Nigeria. J Trop Pediatr, 2005. 51(4): p. 219‐22.

48. Mukhtar, M.Y., et al., Congenital malaria among inborn babies at a tertiary centre in Lagos, Nigeria. J Trop Pediatr, 2006. 52(1): p. 19‐23.

49. Covell, G., Congenital malaria. Trop Dis Bull, 1950. 47(12): p. 1147‐67. 50. Petersen, E., et al., Development of immunity against Plasmodium falciparum

malaria: clinical and parasitologic immunity cannot be separated. J Infect Dis, 1991. 164(5): p. 949‐53.

51. Slutsker, L., et al., Malaria infection in infancy in rural Malawi. Am J Trop Med Hyg, 1996. 55(1 Suppl): p. 71‐6.

52. Hviid, L. and T. Staalsoe, Malaria immunity in infants: a special case of a general phenomenon? Trends Parasitol, 2004. 20(2): p. 66‐72.

53. Pasvol, G., et al., Fetal haemoglobin and malaria. Lancet, 1976. 1(7972): p. 1269‐72.

54. Amaratunga, C., et al., A role for fetal hemoglobin and maternal immune IgG in infant resistance to Plasmodium falciparum malaria. PLoS One, 2011. 6(4): p. e14798.

55. Williams, T.N., How do hemoglobins S and C result in malaria protection? J Infect Dis, 2011. 204(11): p. 1651‐3.

56. Poespoprodjo, J.R., et al., Vivax malaria: a major cause of morbidity in early infancy. Clin Infect Dis, 2009. 48(12): p. 1704‐12.

57. Snow, R.W., et al., Risk of severe malaria among African infants: direct evidence of clinical protection during early infancy. J Infect Dis, 1998. 177(3): p. 819‐22.

58. Severe falciparum malaria. World Health Organization, Communicable Diseases Cluster. Trans R Soc Trop Med Hyg, 2000. 94 Suppl 1: p. S1‐90.

59. Genton, B., et al., Plasmodium vivax and mixed infections are associated with severe malaria in children: a prospective cohort study from Papua New Guinea. PLoS Med, 2008. 5(6): p. e127.

216

60. World Health Organisation. Mortality 1‐59 months estimates, 2008, by cause and by country. 2010 [cited 2011 21 December]; Available from: 8TUhttp://www.who.int/healthinfo/statistics/mortality_child_cause/en/index.html U8T.

61. Schellenberg, D., et al., Intermittent treatment for malaria and anaemia control at time of routine vaccinations in Tanzanian infants: a randomised, placebo‐controlled trial. Lancet, 2001. 357(9267): p. 1471‐7.

62. Aponte, J.J., et al., Efficacy and safety of intermittent preventive treatment with sulfadoxine‐pyrimethamine for malaria in African infants: a pooled analysis of six randomised, placebo‐controlled trials. Lancet, 2009. 374(9700): p. 1533‐42.

63. Dicko, A., et al., Increase in EPI vaccines coverage after implementation of intermittent preventive treatment of malaria in infant with Sulfadoxine ‐pyrimethamine in the district of Kolokani, Mali: results from a cluster randomized control trial. BMC Public Health, 2011. 11: p. 573.

64. Gupta, S., et al., Immunity to non‐cerebral severe malaria is acquired after one or two infections. Nat Med, 1999. 5(3): p. 340‐3.

65. Beeson, J.G., et al., Intermittent preventive treatment to reduce the burden of malaria in children: new evidence on integration and delivery. PLoS Med, 2011. 8(2): p. e1000410.

66. Trape, J.F., The public health impact of chloroquine resistance in Africa. Am J Trop Med Hyg, 2001. 64(1‐2 Suppl): p. 12‐7.

67. World Health Organization, Antimalarial drug combination therapy: Report of a WHO technical consultation. 2001, World Health Organization: Geneva, Switzerland.

68. World Health Organization., Guidelines for the treatment of malaria. 2006, World Health Organization: Geneva, Switzerland.

69. World Health Organization, Guidelines for the treatment of malaria ‐ 2nd edition. 2010, World Health Organization: Geneva, Switzerland.

70. Sinclair, D., et al., Artemisinin‐based combination therapy for treating uncomplicated malaria. Cochrane Database Syst Rev, 2009(3): p. CD007483.

71. Sinclair, D., et al., Artemisinin‐based combination therapy for treating uncomplicated Plasmodium vivax malaria. Cochrane Database Syst Rev, 2011(7): p. CD008492.

72. Sisowath, C., et al., The role of pfmdr1 in Plasmodium falciparum tolerance to artemether‐lumefantrine in Africa. Trop Med Int Health, 2007. 12(6): p. 736‐42.

73. Eastman, R.T., et al., Piperaquine resistance is associated with a copy number variation on chromosome 5 in drug‐pressured Plasmodium falciparum parasites. Antimicrob Agents Chemother, 2011. 55(8): p. 3908‐16.

74. Borrmann, S., et al., Declining Responsiveness of Plasmodium falciparum Infections to Artemisinin‐Based Combination Treatments on the Kenyan Coast. PLoS One, 2011. 6(11): p. e26005.

75. Kurth, F., et al., Pyronaridine–artesunate combination therapy for the treatment of malaria. Current Opinion in Infectious Diseases, 2011. 24(6): p. 564‐569 10.1097/QCO.0b013e32834cabdb.

76. White, N.J., How antimalarial drug resistance affects post‐treatment prophylaxis. Malar J, 2008. 7: p. 9.

77. Rath, K., et al., Pharmacokinetic study of artemisinin after oral intake of a traditional preparation of Artemisia annua L. (annual wormwood). Am J Trop Med Hyg, 2004. 70(2): p. 128‐32.

217

78. Antimalaria studies on Qinghaosu. Chin Med J (Engl), 1979. 92(12): p. 811‐6. 79. O'Neill, P.M., V.E. Barton, and S.A. Ward, The molecular mechanism of action of

artemisinin‐‐the debate continues. Molecules, 2010. 15(3): p. 1705‐21. 80. Nosten, F. and N.J. White, Artemisinin‐based combination treatment of

falciparum malaria. Am J Trop Med Hyg, 2007. 77(6 Suppl): p. 181‐92. 81. Chotivanich, K., et al., Transmission‐blocking activities of quinine, primaquine,

and artesunate. Antimicrob Agents Chemother, 2006. 50(6): p. 1927‐30. 82. Stepniewska, K., et al., In vivo parasitological measures of artemisinin

susceptibility. J Infect Dis, 2010. 201(4): p. 570‐9. 83. Taylor, W.R. and N.J. White, Antimalarial drug toxicity: a review. Drug Saf,

2004. 27(1): p. 25‐61. 84. Price, R., et al., Adverse effects in patients with acute falciparum malaria

treated with artemisinin derivatives. Am J Trop Med Hyg, 1999. 60(4): p. 547‐55.

85. Leonardi, E., et al., Severe allergic reactions to oral artesunate: a report of two cases. Trans R Soc Trop Med Hyg, 2001. 95(2): p. 182‐3.

86. Bethell, D., et al., Dose‐dependent risk of neutropenia after 7‐day courses of artesunate monotherapy in Cambodian patients with acute Plasmodium falciparum malaria. Clin Infect Dis, 2010. 51(12): p. e105‐14.

87. Efferth, T. and B. Kaina, Toxicity of the antimalarial artemisinin and its dervatives. Crit Rev Toxicol, 2010. 40(5): p. 405‐21.

88. Brewer, T.G., et al., Fatal neurotoxicity of arteether and artemether. Am J Trop Med Hyg, 1994. 51(3): p. 251‐9.

89. Brewer, T.G., et al., Neurotoxicity in animals due to arteether and artemether. Trans R Soc Trop Med Hyg, 1994. 88 Suppl 1: p. S33‐6.

90. Genovese, R.F., et al., Acute high dose arteether toxicity in rats. Neurotoxicology, 1999. 20(5): p. 851‐9.

91. Stergachis, A., et al., A situational analysis of pharmacovigilance plans in the Global Fund Malaria and U.S. President's Malaria Initiative proposals. Malar J, 2010. 9: p. 148.

92. Petras, J.M., et al., Arteether‐induced brain injury in Macaca mulatta. I. The precerebellar nuclei: the lateral reticular nuclei, paramedian reticular nuclei, and perihypoglossal nuclei. Anat Embryol (Berl), 2000. 201(5): p. 383‐97.

93. Li, Q.G., et al., Neurotoxicity and efficacy of arteether related to its exposure times and exposure levels in rodents. Am J Trop Med Hyg, 2002. 66(5): p. 516‐25.

94. Panossian, L.A., N.I. Garga, and D. Pelletier, Toxic brainstem encephalopathy after artemisinin treatment for breast cancer. Ann Neurol, 2005. 58(5): p. 812‐3.

95. Miller, L.G. and C.B. Panosian, Ataxia and slurred speech after artesunate treatment for falciparum malaria. N Engl J Med, 1997. 336(18): p. 1328.

96. Elias, Z., et al., Neurotoxicity of artemisinin: possible counseling and treatment of side effects. Clin Infect Dis, 1999. 28(6): p. 1330‐1.

97. Franco‐Paredes, C., et al., Neurotoxicity due to antimalarial therapy associated with misdiagnosis of malaria. Clin Infect Dis, 2005. 40(11): p. 1710‐1.

98. Kager, P.A., et al., Arteether administration in humans: preliminary studies of pharmacokinetics, safety and tolerance. Trans R Soc Trop Med Hyg, 1994. 88 Suppl 1: p. S53‐4.

218

99. Toovey, S., Effects of weight, age, and time on artemether‐lumefantrine associated ototoxicity and evidence of irreversibility. Travel Med Infect Dis, 2006. 4(2): p. 71‐6.

100. Toovey, S. and A. Jamieson, Audiometric changes associated with the treatment of uncomplicated falciparum malaria with co‐artemether. Trans R Soc Trop Med Hyg, 2004. 98(5): p. 261‐7; discussion 268‐9.

101. Tran, T.H., et al., A controlled trial of artemether or quinine in Vietnamese adults with severe falciparum malaria. N Engl J Med, 1996. 335(2): p. 76‐83.

102. van Hensbroek, M.B., et al., A trial of artemether or quinine in children with cerebral malaria. N Engl J Med, 1996. 335(2): p. 69‐75.

103. van Vugt, M., et al., A case‐control auditory evaluation of patients treated with artemisinin derivatives for multidrug‐resistant Plasmodium falciparum malaria. Am J Trop Med Hyg, 2000. 62(1): p. 65‐9.

104. Kissinger, E., et al., Clinical and neurophysiological study of the effects of multiple doses of artemisinin on brain‐stem function in Vietnamese patients. Am J Trop Med Hyg, 2000. 63(1‐2): p. 48‐55.

105. McCall, M.B., et al., No hearing loss associated with the use of artemether‐lumefantrine to treat experimental human malaria. Trans R Soc Trop Med Hyg, 2006. 100(12): p. 1098‐104.

106. Davis, T.M., et al., Severe falciparum malaria with hyperparasitaemia treated with intravenous artesunate. Med J Aust, 1997. 166(8): p. 416‐8.

107. Hien, T.T., et al., Neuropathological assessment of artemether‐treated severe malaria. Lancet, 2003. 362(9380): p. 295‐6.

108. Manning, L., et al., Meningeal inflammation increases artemether concentrations in cerebrospinal fluid in Papua New Guinean children treated with intramuscular artemether. Antimicrob Agents Chemother, 2011. 55(11): p. 5027‐33.

109. Gordi, T. and E.I. Lepist, Artemisinin derivatives: toxic for laboratory animals, safe for humans? Toxicol Lett, 2004. 147(2): p. 99‐107.

110. Skinner, T.S., et al., In vitro stage‐specific sensitivity of Plasmodium falciparum to quinine and artemisinin drugs. Int J Parasitol, 1996. 26(5): p. 519‐25.

111. World Health Organization, WHO briefing on Malaria Treatment Guidelines and artemisinin monotherapies 2006, World Health Organization: Geneva, Switzerland.

112. Ashton, M., et al., Artemisinin pharmacokinetics in healthy adults after 250, 500 and 1000 mg single oral doses. Biopharm Drug Dispos, 1998. 19(4): p. 245‐50.

113. Ashton, M., et al., Artemisinin pharmacokinetics is time‐dependent during repeated oral administration in healthy male adults. Drug Metab Dispos, 1998. 26(1): p. 25‐7.

114. Benakis, A., et al., Pharmacokinetics of artemisinin and artesunate after oral administration in healthy volunteers. Am J Trop Med Hyg, 1997. 56(1): p. 17‐23.

115. Duc, D.D., et al., The pharmacokinetics of a single dose of artemisinin in healthy Vietnamese subjects. Am J Trop Med Hyg, 1994. 51(6): p. 785‐90.

116. Gordi, T., et al., A semiphysiological pharmacokinetic model for artemisinin in healthy subjects incorporating autoinduction of metabolism and saturable first‐pass hepatic extraction. Br J Clin Pharmacol, 2005. 59(2): p. 189‐98.

117. Svensson, U.S., et al., Artemisinin induces omeprazole metabolism in human beings. Clin Pharmacol Ther, 1998. 64(2): p. 160‐7.

219

118. Alin, M.H., et al., Clinical efficacy and pharmacokinetics of artemisinin monotherapy and in combination with mefloquine in patients with falciparum malaria. Br J Clin Pharmacol, 1996. 41(6): p. 587‐92.

119. Ashton, M., et al., Artemisinin kinetics and dynamics during oral and rectal treatment of uncomplicated malaria. Clin Pharmacol Ther, 1998. 63(4): p. 482‐93.

120. Gordi, T., et al., Use of saliva and capillary blood samples as substitutes for venous blood sampling in pharmacokinetic investigations of artemisinin. Eur J Clin Pharmacol, 2000. 56(8): p. 561‐6.

121. Gordi, T., et al., Artemisinin pharmacokinetics and efficacy in uncomplicated‐malaria patients treated with two different dosage regimens. Antimicrob Agents Chemother, 2002. 46(4): p. 1026‐31.

122. Gordi, T., R. Xie, and W.J. Jusko, Semi‐mechanistic pharmacokinetic/pharmacodynamic modelling of the antimalarial effect of artemisinin. Br J Clin Pharmacol, 2005. 60(6): p. 594‐604.

123. Hassan Alin, M., et al., Multiple dose pharmacokinetics of oral artemisinin and comparison of its efficacy with that of oral artesunate in falciparum malaria patients. Trans R Soc Trop Med Hyg, 1996. 90(1): p. 61‐5.

124. Svensson, U.S., et al., Population pharmacokinetic and pharmacodynamic modelling of artemisinin and mefloquine enantiomers in patients with falciparum malaria. Eur J Clin Pharmacol, 2002. 58(5): p. 339‐51.

125. Sidhu, J.S., et al., Artemisinin population pharmacokinetics in children and adults with uncomplicated falciparum malaria. Br J Clin Pharmacol, 1998. 45(4): p. 347‐54.

126. Hassan Alin, M., A. Bjorkman, and M. Ashton, In vitro activity of artemisinin, its derivatives, and pyronaridine against different strains of Plasmodium falciparum. Trans R Soc Trop Med Hyg, 1990. 84(5): p. 635‐7.

127. Hung, L.N., et al., Pharmacokinetics of a single oral dose of dihydroartemisinin in Vietnamese healthy volunteers. Southeast Asian J Trop Med Public Health, 1999. 30(1): p. 11‐16.

128. Na‐Bangchang, K., et al., The pharmacokinetics of oral dihydroartemisinin and artesunate in healthy Thai volunteers. Southeast Asian J Trop Med Public Health, 2004. 35(3): p. 575‐82.

129. Binh, T.Q., et al., Oral bioavailability of dihydroartemisinin in Vietnamese volunteers and in patients with falciparum malaria. Br J Clin Pharmacol, 2001. 51(6): p. 541‐6.

130. Ilett, K.F., et al., The pharmacokinetic properties of intramuscular artesunate and rectal dihydroartemisinin in uncomplicated falciparum malaria. Br J Clin Pharmacol, 2002. 53(1): p. 23‐30.

131. Newton, P.N., et al., The pharmacokinetics of intravenous artesunate in adults with severe falciparum malaria. Eur J Clin Pharmacol, 2006. 62(12): p. 1003‐9.

132. Karbwang, J., et al., Determination of artemether and its major metabolite, dihydroartemisinin, in plasma using high‐performance liquid chromatography with electrochemical detection. J Chromatogr B Biomed Sci Appl, 1997. 690(1‐2): p. 259‐65.

133. Lefevre, G., et al., A clinical and pharmacokinetic trial of six doses of artemether‐lumefantrine for multidrug‐resistant Plasmodium falciparum malaria in Thailand. Am J Trop Med Hyg, 2001. 64(5‐6): p. 247‐56.

220

134. Lefevre, G. and M.S. Thomsen, Clinical pharmacokinetics of artemether and lumefantrine (Riamet (R)). Clinical Drug Investigation, 1999. 18(6): p. 467‐480.

135. German, P., et al., Lopinavir/ritonavir affects pharmacokinetic exposure of artemether/lumefantrine in HIV‐uninfected healthy volunteers. J Acquir Immune Defic Syndr, 2009. 51(4): p. 424‐9.

136. Lefevre, G., et al., Pharmacokinetic interaction trial between co‐artemether and mefloquine. Eur J Pharm Sci, 2000. 10(2): p. 141‐51.

137. Lefevre, G., et al., Interaction trial between artemether‐lumefantrine (Riamet) and quinine in healthy subjects. J Clin Pharmacol, 2002. 42(10): p. 1147‐58.

138. Lefevre, G., et al., Pharmacokinetics and electrocardiographic pharmacodynamics of artemether‐lumefantrine (Riamet) with concomitant administration of ketoconazole in healthy subjects. Br J Clin Pharmacol, 2002. 54(5): p. 485‐92.

139. van Agtmael, M.A., et al., Multiple dose pharmacokinetics of artemether in Chinese patients with uncomplicated falciparum malaria. Int J Antimicrob Agents, 1999. 12(2): p. 151‐8.

140. Mithwani, S., et al., Population pharmacokinetics of artemether and dihydroartemisinin following single intramuscular dosing of artemether in African children with severe falciparum malaria. Br J Clin Pharmacol, 2004. 57(2): p. 146‐52.

141. Ezzet, F., R. Mull, and J. Karbwang, Population pharmacokinetics and therapeutic response of CGP 56697 (artemether + benflumetol) in malaria patients. Br J Clin Pharmacol, 1998. 46(6): p. 553‐61.

142. Hietala, S.F., et al., Population pharmacokinetics and pharmacodynamics of artemether and lumefantrine during combination treatment in children with uncomplicated falciparum malaria in Tanzania. Antimicrob Agents Chemother, 2010. 54(11): p. 4780‐8.

143. McGready, R., et al., The pharmacokinetics of artemether and lumefantrine in pregnant women with uncomplicated falciparum malaria. Eur J Clin Pharmacol, 2006. 62(12): p. 1021‐1031.

144. Mwesigwa, J., et al., Pharmacokinetics of artemether‐lumefantrine and artesunate‐amodiaquine in children in Kampala, Uganda. Antimicrob Agents Chemother, 2010. 54(1): p. 52‐9.

145. Schlitzer, M., Malaria chemotherapeutics part I: History of antimalarial drug development, currently used therapeutics, and drugs in clinical development. CHEMMEDCHEM, 2007. 2(7): p. 944‐86.

146. Barnes, D.A., et al., Selection for high‐level chloroquine resistance results in deamplification of the pfmdr1 gene and increased sensitivity to mefloquine in Plasmodium falciparum. EMBO J, 1992. 11(8): p. 3067‐75.

147. Basco, L.K., J. Bickii, and P. Ringwald, In vitro activities of lumefantrine (benflumetol) against clinical isolates of Plasmodium falciparum in Yaounde, Cameroon. Antimicrobial Agents & Chemotherapy, 1998. 42(9): p. 2347‐2351.

148. Price, R.N., et al., Molecular and pharmacological determinants of the therapeutic response to artemether‐lumefantrine in multidrug‐resistant Plasmodium falciparum malaria. Clin Infect Dis, 2006. 42(11): p. 1570‐7.

149. Nosten, F., et al., Effects of artesunate‐mefloquine combination on incidence of Plasmodium falciparum malaria and mefloquine resistance in western Thailand: a prospective study. The Lancet, 2000. 356(9226): p. 297‐302.

221

150. White, N.J., Cardiotoxicity of antimalarial drugs. The Lancet Infectious Diseases, 2007. 7(8): p. 549‐58.

151. Thriemer, K., et al., In vitro interaction of dihydroartemisin and lumefantrine in clinical field isolates from Bangladesh. Wiener Klinische Wochenschrift, 2007. 119: p. 67‐70.

152. Traebert, M., et al., Inhibition of hERG K+ currents by antimalarial drugs in stably transfected HEK293 cells. European Journal of Pharmacology, 2004. 484(1): p. 41‐8.

153. Bindschedler, M., et al., Comparison of the cardiac effects of the antimalarials co‐artemether and halofantrine in healthy participants. American Journal of Tropical Medicine and Hygiene, 2002. 66(3): p. 293‐298.

154. Bindschedler, M., et al., Cardiac effects of co‐artemether (artemether/lumefantrine) and mefloquine given alone or in combination to healthy volunteers. European Journal of Clinical Pharmacology, 2000. 56(5): p. 375‐381.

155. Lefevre, G., et al., Pharmacokinetics and electrocardiographic pharmacodynamics of artemether‐lumefantrine (Riamet) with concomitant administration of ketoconazole in healthy subjects. British Journal of Clinical Pharmacology, 2002. 54(5): p. 485‐92.

156. van Vugt, M., et al., No evidence of cardiotoxicity during antimalarial treatment with artemether‐lumefantrine. American Journal of Tropical Medicine & Hygiene, 1999. 61(6): p. 964‐7.

157. Bakshi, R., et al., An integrated assessment of the clinical safety of artemether‐lumefantrine: a new oral fixed‐dose combination antimalarial drug. Transactions of the Royal Society of Tropical Medicine & Hygiene, 2000. 94(4): p. 419‐24.

158. Makanga, M., et al., Efficacy and Safety of the six‐dose regimen of artemether‐lumefantrine in pediatrics with uncomplicated Plasmodium falciparum malaria: a pooled analysis of individual patient data. American Journal of Tropical Medicine and Hygiene, 2006. 74(6): p. 991‐998.

159. Alin, M.H., A. Bjorkman, and W.H. Wernsdorfer, Synergism of benflumetol and artemether in Plasmodium falciparum. American Journal of Tropical Medicine & Hygiene, 1999. 61(3): p. 439‐445.

160. Noedl, H., et al., Desbutyl‐benflumetol, a novel antimalarial compound: in vitro activity in fresh isolates of Plasmodium falciparum from Thailand. Antimicrob Agents Chemother, 2001. 45(7): p. 2106‐9.

161. Starzengruber, P., et al., Interaction between lumefantrine and monodesbutyl‐benflumetol in Plasmodium falciparum in vitro. Wien Klin Wochenschr, 2008. 120(Suppl 4): p. 85‐89.

162. Starzengruber, P., et al., Specific pharmacokinetic interaction between lumefantrine and monodesbutyl‐benflumetol in Plasmodium falciparum Wien Klin Wochenschr, 2007. 119(Suppl 3): p. 60‐66.

163. Wong, R.P., et al., Desbutyl‐lumefantrine is a metabolite of lumefantrine with potent in vitro antimalarial activity that may influence artemether‐lumefantrine treatment outcome. Antimicrob Agents Chemother, 2011. 55(3): p. 1194‐8.

164. Ezzet, F., et al., Pharmacokinetics and pharmacodynamics of lumefantrine (benflumetol) in acute falciparum malaria. Antimicrob Agents Chemother, 2000. 44(3): p. 697‐704.

222

165. Ashley, E.A., et al., Pharmacokinetic study of artemether‐lumefantrine given once daily for the treatment of uncomplicated multidrug‐resistant falciparum malaria. Tropical Medicine & International Health, 2007. 12(2): p. 201‐8.

166. Ezzet, F., et al., Pharmacokinetics and pharmacodynamics of lumefantrine (benflumetol) in acute falciparum malaria. Antimicrobial Agents & Chemotherapy, 2000. 44(3): p. 697‐704.

167. Dafra Pharm. 01 February 2008]; Available from: 8TUhttp://www.dafra.be/malaria1.html#1U8T.

168. Chanda, P., et al., Assessment of the therapeutic efficacy of a paediatric formulation of artemether‐lumefantrine (Coartesiane) for the treatment of uncomplicated Plasmodium falciparum in children in Zambia. Malaria Journal, 2006: p. 75.

169. Salah, M.T., et al., Efficacy of artemether‐lumfantrine (Co‐Artesiane®) suspension in the treatment of uncomplicated Plasmodium falciparum malaria among children under 5 years in eastern Sudan. Tropical Journal of Pharmaceutical Research, 2006. 5(1): p. 551‐555.

170. Juma, E.A., et al., A randomized, open‐label, comparative efficacy trial of artemether‐lumefantrine suspension versus artemether‐lumefantrine tablets for treatment of uncomplicated Plasmodium falciparum malaria in children in western Kenya. Malar J, 2008. 7: p. 262.

171. Makanga, M., et al., Efficacy and safety of artemether‐lumefantrine in the treatment of acute, uncomplicated Plasmodium falciparum malaria: a pooled analysis. Am J Trop Med Hyg, 2011. 85(5): p. 793‐804.

172. Chinh, N.T., et al., Pharmacokinetics and bioequivalence evaluation of two fixed‐dose tablet formulations of dihydroartemisinin and piperaquine in Vietnamese subjects. Antimicrob Agents Chemother, 2009. 53(2): p. 828‐31.

173. Hai, T.N., et al., The influence of food on the pharmacokinetics of piperaquine in healthy Vietnamese volunteers. Acta Trop, 2008. 107(2): p. 145‐9.

174. Hung, T.Y., et al., Population pharmacokinetics of piperaquine in adults and children with uncomplicated falciparum or vivax malaria. Br J Clin Pharmacol, 2004. 57(3): p. 253‐62.

175. Liu, C., et al., Pharmacokinetics of piperaquine after single and multiple oral administrations in healthy volunteers. Yakugaku Zasshi, 2007. 127(10): p. 1709‐14.

176. Chinh, N.T., et al., Pharmacokinetics of the antimalarial drug piperaquine in healthy Vietnamese subjects. Am J Trop Med Hyg, 2008. 79(4): p. 620‐3.

177. Roshammar, D., et al., Pharmacokinetics of piperaquine after repeated oral administration of the antimalarial combination CV8 in 12 healthy male subjects. Eur J Clin Pharmacol, 2006. 62(5): p. 335‐41.

178. Tarning, J., et al., Population pharmacokinetics of piperaquine after two different treatment regimens with dihydroartemisinin‐piperaquine in patients with Plasmodium falciparum malaria in Thailand. Antimicrob Agents Chemother, 2008. 52(3): p. 1052‐61.

179. Ahmed, T., et al., Safety, tolerability, and single‐ and multiple‐dose pharmacokinetics of piperaquine phosphate in healthy subjects. J Clin Pharmacol, 2008. 48(2): p. 166‐75.

180. Lefevre, G. and M.S. Thomsen, Clinical Pharmacokinetics of Artemether and Lumefantrine (Riamet(R)). Clinical Drug Investigation, 1999. 18(6): p. 467‐480.

223

181. Ashley, E.A., et al., How much fat is necessary to optimize lumefantrine oral bioavailability? Tropical Medicine & International Health, 2007. 12(2): p. 195‐200.

182. Tarning, J., et al., Population pharmacokinetics of lumefantrine in pregnant women treated with artemether‐lumefantrine for uncomplicated Plasmodium falciparum malaria. Antimicrob Agents Chemother, 2009. 53(9): p. 3837‐46.

183. Abdulla, S., et al., Efficacy and safety of artemether‐lumefantrine dispersible tablets compared with crushed commercial tablets in African infants and children with uncomplicated malaria: a randomised, single‐blind, multicentre trial. Lancet, 2008. 372(9652): p. 1819‐27.

184. Hatz, C., et al., Treatment of acute uncomplicated falciparum malaria with artemether‐lumefantrine in nonimmune populations: a safety, efficacy, and pharmacokinetic study. Am J Trop Med Hyg, 2008. 78(2): p. 241‐247.

185. Kumar, S., et al., Antimalarial drugs inhibiting hemozoin (beta‐hematin) formation: a mechanistic update. Life Sci, 2007. 80(9): p. 813‐28.

186. Mnyika, K.S. and C.M. Kihamia, Chloroquine‐induced pruritus: its impact on chloroquine utilization in malaria control in Dar es Salaam. J Trop Med Hyg, 1991. 94(1): p. 27‐31.

187. AlKadi, H.O., Antimalarial drug toxicity: a review. Chemotherapy, 2007. 53(6): p. 385‐91.

188. Myint, H.Y., et al., Efficacy and safety of dihydroartemisinin‐piperaquine. Trans R Soc Trop Med Hyg, 2007. 101(9): p. 858‐66.

189. Hombhanje, F.W. and Q. Huang, Artemisinin‐Naphthoquine Combination (ARCO®): An Overview of the Progress. Pharmaceuticals, 2010. 3(12): p. 3581‐3593.

190. Basco, L.K. and P. Ringwald, In vitro activities of piperaquine and other 4‐aminoquinolines against clinical isolates of Plasmodium falciparum in Cameroon. Antimicrob Agents Chemother, 2003. 47(4): p. 1391‐4.

191. Muangnoicharoen, S., et al., Role of known molecular markers of resistance in the antimalarial potency of piperaquine and dihydroartemisinin in vitro. Antimicrob Agents Chemother, 2009. 53(4): p. 1362‐6.

192. Wong, R.P., et al., In vitro sensitivity of Plasmodium falciparum to conventional and novel antimalarial drugs in Papua New Guinea. Trop Med Int Health, 2010. 15(3): p. 342‐9.

193. Gargano, N., F. Cenci, and Q. Bassat, Antimalarial efficacy of piperaquine‐based antimalarial combination therapies: facts and uncertainties. Trop Med Int Health, 2011. 16(12): p. 1466‐73.

194. Davis, T.M., et al., In vitro interactions between piperaquine, dihydroartemisinin, and other conventional and novel antimalarial drugs. Antimicrob Agents Chemother, 2006. 50(8): p. 2883‐5.

195. Fivelman, Q.L., I.S. Adagu, and D.C. Warhurst, Effects of piperaquine, chloroquine, and amodiaquine on drug uptake and of these in combination with dihydroartemisinin against drug‐sensitive and ‐resistant Plasmodium falciparum strains. Antimicrob Agents Chemother, 2007. 51(6): p. 2265‐7.

196. Denis, M.B., et al., Efficacy and safety of dihydroartemisinin‐piperaquine (Artekin) in Cambodian children and adults with uncomplicated falciparum malaria. Clin Infect Dis, 2002. 35(12): p. 1469‐76.

224

197. Hien, T.T., et al., Dihydroartemisinin‐piperaquine against multidrug‐resistant Plasmodium falciparum malaria in Vietnam: randomised clinical trial. Lancet, 2004. 363(9402): p. 18‐22.

198. Ashley, E.A., et al., Randomized, controlled dose‐optimization studies of dihydroartemisinin‐piperaquine for the treatment of uncomplicated multidrug‐resistant falciparum malaria in Thailand. J Infect Dis, 2004. 190(10): p. 1773‐82.

199. Ashley, E.A., et al., A randomized, controlled study of a simple, once‐daily regimen of dihydroartemisinin‐piperaquine for the treatment of uncomplicated, multidrug‐resistant falciparum malaria. Clin Infect Dis, 2005. 41(4): p. 425‐32.

200. Smithuis, F., et al., Efficacy and effectiveness of dihydroartemisinin‐piperaquine versus artesunate‐mefloquine in falciparum malaria: an open‐label randomised comparison. Lancet, 2006. 367(9528): p. 2075‐85.

201. Mayxay, M., et al., An open, randomized comparison of artesunate plus mefloquine vs. dihydroartemisinin‐piperaquine for the treatment of uncomplicated Plasmodium falciparum malaria in the Lao People's Democratic Republic (Laos). Trop Med Int Health, 2006. 11(8): p. 1157‐65.

202. Price, R.N., et al., Clinical and pharmacological determinants of the therapeutic response to dihydroartemisinin‐piperaquine for drug‐resistant malaria. Antimicrob Agents Chemother, 2007. 51(11): p. 4090‐7.

203. Janssens, B., et al., A randomized open study to assess the efficacy and tolerability of dihydroartemisinin‐piperaquine for the treatment of uncomplicated falciparum malaria in Cambodia. Trop Med Int Health, 2007. 12(2): p. 251‐9.

204. Zongo, I., et al., Randomized comparison of amodiaquine plus sulfadoxine‐pyrimethamine, artemether‐lumefantrine, and dihydroartemisinin‐piperaquine for the treatment of uncomplicated Plasmodium falciparum malaria in Burkina Faso. Clin Infect Dis, 2007. 45(11): p. 1453‐61.

205. Grande, T., et al., A randomised controlled trial to assess the efficacy of dihydroartemisinin‐piperaquine for the treatment of uncomplicated falciparum malaria in Peru. PLoS One, 2007. 2(10): p. e1101.

206. Yeka, A., et al., Artemether‐lumefantrine versus dihydroartemisinin‐piperaquine for treating uncomplicated malaria: a randomized trial to guide policy in Uganda. PLoS One, 2008. 3(6): p. e2390.

207. Arinaitwe, E., et al., Artemether‐lumefantrine versus dihydroartemisinin‐piperaquine for falciparum malaria: a longitudinal, randomized trial in young Ugandan children. Clin Infect Dis, 2009. 49(11): p. 1629‐37.

208. Bassat, Q., et al., Dihydroartemisinin‐piperaquine and artemether‐lumefantrine for treating uncomplicated malaria in African children: a randomised, non‐inferiority trial. PLoS One, 2009. 4(11): p. e7871.

209. Mayxay, M., et al., A phase III, randomized, non‐inferiority trial to assess the efficacy and safety of dihydroartemisinin‐piperaquine in comparison with artesunate‐mefloquine in patients with uncomplicated Plasmodium falciparum malaria in southern Laos. Am J Trop Med Hyg, 2010. 83(6): p. 1221‐9.

210. Karema, C., et al., Safety and efficacy of dihydroartemisinin/piperaquine (Artekin) for the treatment of uncomplicated Plasmodium falciparum malaria in Rwandan children. Trans R Soc Trop Med Hyg, 2006. 100(12): p. 1105‐11.

211. Trung, T.N., et al., A randomized, controlled trial of artemisinin‐piperaquine vs dihydroartemisinin‐piperaquine phosphate in treatment of falciparum malaria. Chin J Integr Med, 2009. 15(3): p. 189‐92.

225

212. Pyar, K.P., et al., Efficacy and safety of artemisinin‐piperaquine (Artequick) compared to dihydroartemisinin‐piperaquine (Artekin) in uncomplicated falciparum malaria in adults. Myanmar Health Sciences Research Journal 2009. 21(2): p. 78‐82.

213. Krudsood, S., et al., Dose ranging studies of new artemisinin‐piperaquine fixed combinations compared to standard regimens of artemisisnin combination therapies for acute uncomplicated falciparum malaria. Southeast Asian J Trop Med Public Health, 2007. 38(6): p. 971‐8.

214. Annerberg, A., et al., A small amount of fat does not affect piperaquine exposure in patients with malaria. Antimicrob Agents Chemother, 2011. 55(9): p. 3971‐6.

215. Sim, I.K., T.M. Davis, and K.F. Ilett, Effects of a high‐fat meal on the relative oral bioavailability of piperaquine. Antimicrob Agents Chemother, 2005. 49(6): p. 2407‐11.

216. Tarning, J., et al., Pitfalls in estimating piperaquine elimination. Antimicrob Agents Chemother, 2005. 49(12): p. 5127‐8.

217. Wang, J.Y., et al., Naphthoquine phosphate and its combination with artemisinine. Acta Trop, 2004. 89(3): p. 375‐81.

218. Wang, H., et al., Plasmodium berghei K173: Selection of resistance to naphthoquine in a mouse model. Experimental Parasitology, 2011. 127(2): p. 436‐439.

219. Li, X.L., L.G. Che, and C.F. Li, Observation on the effectiveness of naphthoquine phosphate tablets on patients with vivax malaria and falciparum malaria <In Chinese>. China Tropical Medicine, 2003(5).

220. Qu, H.Y., et al., Single‐dose safety, pharmacokinetics, and food effects studies of compound naphthoquine phosphate tablets in healthy volunteers. J Clin Pharmacol, 2010. 50(11): p. 1310‐8.

221. Hurly, M.G.D., Potentiation of pyrimethamine by sulphadiazine in human malaria. Transactions of the Royal Society of Tropical Medicine and Hygiene, 1959. 53(5): p. 412‐413.

222. Wickramasinghe, S.N. and R.A. Litwinczuk, Effects of low concentrations of pyrimethamine on human bone marrow cells in vitro: possible implications for malaria prophylaxis. J Trop Med Hyg, 1981. 84(6): p. 233‐8.

223. Bjorkman, A. and P.A. Phillips‐Howard, Adverse reactions to sulfa drugs: implications for malaria chemotherapy. Bull World Health Organ, 1991. 69(3): p. 297‐304.

224. Sibley, C.H., et al., Pyrimethamine‐sulfadoxine resistance in Plasmodium falciparum: what next? Trends Parasitol, 2001. 17(12): p. 582‐8.

225. Roche, Fansidar Product Information. 2005, Dee Why, Australia: Roche Products Pty Ltd.

226. Barnes, K.I., et al., Sulfadoxine‐pyrimethamine pharmacokinetics in malaria: pediatric dosing implications. Clin Pharmacol Ther, 2006. 80(6): p. 582‐96.

227. Trenque, T., et al., Population pharmacokinetics of pyrimethamine and sulfadoxine in children with congenital toxoplasmosis. Br J Clin Pharmacol, 2004. 57(6): p. 735‐41.

228. Anderson, B.J. and N.H. Holford, Mechanistic basis of using body size and maturation to predict clearance in humans. Drug Metabolism & Pharmacokinetics, 2009. 24(1): p. 25‐36.

226

229. Hellgren, U., et al., Standard and reduced doses of sulfadoxine‐pyrimethamine for treatment of Plasmodium falciparum in Tanzania, with determination of drug concentrations and susceptibility in vitro. Trans R Soc Trop Med Hyg, 1990. 84(4): p. 469‐72.

230. Winstanley, P.A., et al., The disposition of oral and intramuscular pyrimethamine/sulphadoxine in Kenyan children with high parasitaemia but clinically non‐severe falciparum malaria. Br J Clin Pharmacol, 1992. 33(2): p. 143‐8.

231. Yeh, E. and J.L. DeRisi, Chemical rescue of malaria parasites lacking an apicoplast defines organelle function in blood‐stage Plasmodium falciparum. PLoS Biol, 2011. 9(8): p. e1001138.

232. Briolant, S., et al., Plasmodium falciparum proteome changes in response to doxycycline treatment. Malar J, 2010. 9: p. 141.

233. Pradines, B., et al., In vitro activities of antibiotics against Plasmodium falciparum are inhibited by iron. Antimicrob Agents Chemother, 2001. 45(6): p. 1746‐50.

234. Knothe, H. and G.A. Dette, Antibiotics in pregnancy: toxicity and teratogenicity. Infection, 1985. 13(2): p. 49‐51.

235. Czeizel, A.E. and M. Rockenbauer, Teratogenic study of doxycycline. Obstet Gynecol, 1997. 89(4): p. 524‐8.

236. Centre for Disease Control and Prevention, Sexually Transmitted Diseases Treatment Guidelines, 2006: Diseases Characterized by Urethritis and Cervicitis. Morbidity and Mortality Weekly Report, 2006. 55(RR‐11): p. 35‐49.

237. Vallely, A., et al., Intermittent Preventive Treatment for Malaria in Pregnancy in Africa: What's New, What's Needed (review). Malaria Journal, 2007. 6(16).

238. Hopkins, S., Clinical toleration and safety of azithromycin. Am J Med, 1991. 91(3A): p. 40S‐45S.

239. Periti, P., et al., Adverse effects of macrolide antibacterials. Drug Saf, 1993. 9(5): p. 346‐64.

240. Anderson, S.L., et al., Successful Double‐Blinded, Randomized, Placebo‐Controlled Feild Trial of Azithromycin and Doxycycline as Prophylaxis for Malaria in Western Kenya. Clinical Infectious Diseases, 1998. 26: p. 146‐150.

241. Taylor, W.R.J., et al., Tolerability of Azithromycin as Malaria Prophylaxis in Adults in Northeast Papua, Indonesia. Antimicrobial Agents and Chemotherapy, 2003. 47(7): p. 2199‐2203.

242. Heppner, D.G., Jr., et al., Randomized, controlled, double‐blind trial of daily oral azithromycin in adults for the prophylaxis of Plasmodium vivax malaria in Western Thailand. Am J Trop Med Hyg, 2005. 73(5): p. 842‐9.

243. Oldfield, E.C., 3rd, et al., Once weekly azithromycin therapy for prevention of Mycobacterium avium complex infection in patients with AIDS: a randomized, double‐blind, placebo‐controlled multicenter trial. Clin Infect Dis, 1998. 26(3): p. 611‐9.

244. Peters, D.H., H.A. Friedel, and D. McTavish, Azithromycin. A review of its antimicrobial activity, pharmacokinetic properties and clinical efficacy. Drugs, 1992. 44(5): p. 750‐99.

245. Noedl, H., et al., Antimalarial Activity of Azithromycin, Artemisinin and Dihydroartemisinin of Fresh Isolates of Plasmodium falciparum in Thailand. Acta Tropica, 2001. 80: p. 39‐44.

227

246. Ohrt, C., et al., Assessment of azithromycin in combination with other antimalarial drugs against Plasmodium falciparum in vitro. Antimicrob Agents Chemother, 2002. 46(8): p. 2518‐24.

247. Nakornchai, S. and P. Konthiang, Activity of Azithromycin or Erythromycin in Combination with Antimalarial Drugs Against Multidrug‐Resistant Plasmodium falciparum In Vitro. Acta Tropica, 2006. 100: p. 185‐191.

248. van Eijk, A.M. and D.J. Terlouw, Azithromycin for treating uncomplicated malaria. Cochrane Database Syst Rev, 2011(2): p. CD006688.

249. Kalilani, L., et al., A randomized controlled pilot trial of azithromycin or artesunate added to sulfadoxine‐pyrimethamine as treatment for malaria in pregnant women. PLoS One, 2007. 2(11): p. e1166.

250. Chico, R.M., et al., Azithromycin‐chloroquine and the intermittent preventive treatment of malaria in pregnancy. Malar J, 2008. 7: p. 255.

251. Luntamo, M., et al., Effect of repeated treatment of pregnant women with sulfadoxine‐pyrimethamine and azithromycin on preterm delivery in Malawi: a randomized controlled trial. Am J Trop Med Hyg, 2010. 83(6): p. 1212‐20.

252. van den Broek, N.R., et al., The APPLe study: a randomized, community‐based, placebo‐controlled trial of azithromycin for the prevention of preterm birth, with meta‐analysis. PLoS Med, 2009. 6(12): p. e1000191.

253. Medicines for Malaria Venture. Azithromycin‐chloroquine (AZCQ). 2011 [cited 2011 22 December]; Available from: 8TUhttp://www.mmv.org/research‐development/project‐portfolio/azithromycin‐chloroquine‐azcqU8T.

254. Bulmer, J.N., et al., Placental malaria. I. Pathological classification. Histopathology, 1993. 22(3): p. 211‐8.

255. Muehlenbachs, A., et al., A novel histological grading scheme for placental malaria applied in areas of high and low malaria transmission. J Infect Dis, 2010. 202(10): p. 1608‐16.

256. Liu, P., et al., Comparative pharmacokinetics of azithromycin in serum and white blood cells of healthy subjects receiving a single‐dose extended‐release regimen versus a 3‐day immediate‐release regimen. Antimicrob Agents Chemother, 2007. 51(1): p. 103‐9.

257. Ripa, S., L. Ferrante, and M. Prenna, A linear model for the pharmacokinetics of azithromycin in healthy volunteers. Chemotherapy, 1996. 42(6): p. 402‐9.

258. Ballow, C.H., et al., Pharmacokinetics of Oral Azithromycin in Serum, Urine, Polymorphonuclear Leucocytes and Inflammatory vs Non‐Inflammatory Skin Blisters in Healthy Volunteers. Clin Drug Investig, 1998. 15(2): p. 159‐67.

259. Mazzei, T., et al., Pharmacokinetics of azithromycin in patients with impaired hepatic function. J Antimicrob Chemother, 1993. 31 Suppl E: p. 57‐63.

260. Pfizer, Zithromax U.S. physician prescribing information 2009. 261. Boonleang, J., et al., Bioavailability and pharmacokinetic comparison between

generic and branded azithromycin capsule: a randomized, double‐blind, 2‐way crossover in healthy male Thai volunteers. Clin Ther, 2007. 29(4): p. 703‐10.

262. Foulds, G., R.M. Shepard, and R.B. Johnson, The pharmacokinetics of azithromycin in human serum and tissues. J Antimicrob Chemother, 1990. 25 Suppl A: p. 73‐82.

263. Ramsey, P.S., et al., Maternal and Transplacental Pharmacokinetics of Azithromycin. American Journal of Obstetrics and Gynecology, 2003. 188(3): p. 714‐718.

228

264. Cook, J.A., et al., Lack of a Pharmacokinetic Interaction Between Azithromycin and Chloroquine. American Journal of Tropical Medicine and Hygiene, 2006. 74(3): p. 407‐412.

265. Lucchi, M., et al., Pharmacokinetics of azithromycin in serum, bronchial washings, alveolar macrophages and lung tissue following a single oral dose of extended or immediate release formulations of azithromycin. J Antimicrob Chemother, 2008. 61(4): p. 884‐91.

266. Wagner, J.G., History of pharmacokinetics. Pharmacol Ther, 1981. 12(3): p. 537‐62.

267. Duncan, W.A.M. and J. Raventos, The pharmacokinetics of halotane (flurothane) anaesthesia. British Journal of Anaesthesia, 1959. 31(7): p. 302‐315.

268. Teorell, T., Kinetics of distribution of substances administered to the body I The extravascular modes of administration. Archives Internationales De Pharmacodynamie Et De Therapie, 1937. 57: p. 205‐225.

269. Teorell, T., Kinetics of distribution of substances administered to the body II The intravascular modes of administration. Archives Internationales De Pharmacodynamie Et De Therapie, 1937. 57: p. 226‐240.

270. Deichmann, W.B., et al., What is there that is not poison? A study of the Third Defense by Paracelsus. Arch Toxicol, 1986. 58(4): p. 207‐13.

271. Riviere, J.E., Comparative pharmacokinetics. Second Edition. ed. 2011, Chichester, West Sussex: Wiley‐Blackwell. x, 443 pages.

272. DiStefano, J.J., 3rd and E.M. Landaw, Multiexponential, multicompartmental, and noncompartmental modeling. I. Methodological limitations and physiological interpretations. Am J Physiol, 1984. 246(5 Pt 2): p. R651‐64.

273. Rescigno, A., Foundations of pharmacokinetics. 2003, New York: Kluwer Academic/Plenum Pub. xvi, 228 p.

274. Box, G.E.P. and N.R. Draper, Empirical model‐building and response surfaces. Wiley series in probability and mathematical statistics. Applied probability and statistics. 1987, New York: Wiley. xiv, 669 p.

275. Yuh, L., et al., Population pharmacokinetic/pharmacodynamic methodology and applications: a bibliography. Biometrics, 1994. 50(2): p. 566‐75.

276. Sheiner, L.B. and T.M. Ludden, Population pharmacokinetics/dynamics. Annu Rev Pharmacol Toxicol, 1992. 32: p. 185‐209.

277. Sheiner, L.B., B. Rosenberg, and V.V. Marathe, Estimation of population characteristics of pharmacokinetic parameters from routine clinical data. J Pharmacokinet Biopharm, 1977. 5(5): p. 445‐79.

278. Mandema, J.W., D. Verotta, and L.B. Sheiner, Building population pharmacokinetic‐‐pharmacodynamic models. I. Models for covariate effects. J Pharmacokinet Biopharm, 1992. 20(5): p. 511‐28.

279. Sheiner, L.B., The population approach to pharmacokinetic data analysis: rationale and standard data analysis methods. Drug Metab Rev, 1984. 15(1‐2): p. 153‐71.

280. Sheiner, L.B. and S.L. Beal, Evaluation of methods for estimating population pharmacokinetics parameters. I. Michaelis‐Menten model: routine clinical pharmacokinetic data. J Pharmacokinet Biopharm, 1980. 8(6): p. 553‐71.

281. Sheiner, L.B. and S.L. Beal, Evaluation of methods for estimating population pharmacokinetic parameters. III. Monoexponential model: routine clinical pharmacokinetic data. J Pharmacokinet Biopharm, 1983. 11(3): p. 303‐19.

229

282. Karlsson, M.O. and L.B. Sheiner, The importance of modeling interoccasion variability in population pharmacokinetic analyses. J Pharmacokinet Biopharm, 1993. 21(6): p. 735‐50.

283. Steimer, J.L., et al., Alternative approaches to estimation of population pharmacokinetic parameters: comparison with the nonlinear mixed‐effect model. Drug Metab Rev, 1984. 15(1‐2): p. 265‐92.

284. Vicini, P. and C. Cobelli, The iterative two‐stage population approach to IVGTT minimal modeling: improved precision with reduced sampling. Intravenous glucose tolerance test. Am J Physiol Endocrinol Metab, 2001. 280(1): p. E179‐86.

285. Patron‐Bizet, F., et al., Assessment of the global two‐stage method to EC50 determination. J Pharmacol Toxicol Methods, 1998. 39(2): p. 103‐8.

286. Sheiner, L.B., B. Rosenberg, and K.L. Melmon, Modelling of individual pharmacokinetics for computer‐aided drug dosage. Comput Biomed Res, 1972. 5(5): p. 411‐59.

287. Savic, R.M., et al., Implementation of a transit compartment model for describing drug absorption in pharmacokinetic studies. J Pharmacokinet Pharmacodyn, 2007. 34(5): p. 711‐26.

288. Beal, S.L., Ways to fit a PK model with some data below the quantification limit. J Pharmacokinet Pharmacodyn, 2001. 28(5): p. 481‐504.

289. Beal, S. and L.B. Sheiner, NONMEM Users Guide part I. 1980, San Fransisco: Division of Clinical Pharmacology, University of California.

290. Wang, Y., Derivation of various NONMEM estimation methods. J Pharmacokinet Pharmacodyn, 2007. 34(5): p. 575‐93.

291. Beal, S. and L.B. Sheiner, NONMEM Users Guide part VII. 1998, San Fransisco: Division of Clinical Pharmacology, University of California.

292. Wahlby, U., E.N. Jonsson, and M.O. Karlsson, Assessment of actual significance levels for covariate effects in NONMEM. J Pharmacokinet Pharmacodyn, 2001. 28(3): p. 231‐52.

293. Beal, S. and L.B. Sheiner, NONMEM Users Guide part V. 1994, San Fransisco: Division of Clinical Pharmacology, University of California.

294. Holford, N.H., R.J. Ambros, and K. Stoeckel, Models for describing absorption rate and estimating extent of bioavailability: application to cefetamet pivoxil. J Pharmacokinet Biopharm, 1992. 20(5): p. 421‐42.

295. Xu, X.S., et al., Impact of low percentage of data below the quantification limit on parameter estimates of pharmacokinetic models. J Pharmacokinet Pharmacodyn, 2011. 38(4): p. 423‐32.

296. Bergstrand, M. and M.O. Karlsson, Handling data below the limit of quantification in mixed effect models. AAPS J, 2009. 11(2): p. 371‐80.

297. Ahn, J.E., et al., Likelihood based approaches to handling data below the quantification limit using NONMEM VI. J Pharmacokinet Pharmacodyn, 2008. 35(4): p. 401‐21.

298. Brendel, K., et al., Are population pharmacokinetic and/or pharmacodynamic models adequately evaluated? A survey of the literature from 2002 to 2004. Clin Pharmacokinet, 2007. 46(3): p. 221‐34.

299. Hooker, A.C., C.E. Staatz, and M.O. Karlsson, Conditional weighted residuals (CWRES): a model diagnostic for the FOCE method. Pharm Res, 2007. 24(12): p. 2187‐97.

230

300. Parke, J., N.H. Holford, and B.G. Charles, A procedure for generating bootstrap samples for the validation of nonlinear mixed‐effects population models. Comput Methods Programs Biomed, 1999. 59(1): p. 19‐29.

301. Gastonguay, M.R. and A. El‐Tahtawy, Effect of NONMEM minimization status and number of replicates on bootstrap parameter distributions for population pharmacokinetic models: A case study. Clin Pharmacol Ther, 2005. 77(2): p. P2‐P2.

302. Holford, N.H., C. Kirkpatrick, and S. Dufful, NONMEM Termination Status is Not an Important Indicator of the Quality of Bootstrap Parameter Estimates, in PAGE. 2006: Bruges.

303. Yano, Y., S.L. Beal, and L.B. Sheiner, Evaluating pharmacokinetic/pharmacodynamic models using the posterior predictive check. J Pharmacokinet Pharmacodyn, 2001. 28(2): p. 171‐92.

304. Bergstrand, M., et al., Prediction‐corrected visual predictive checks for diagnosing nonlinear mixed‐effects models. AAPS J, 2011. 13(2): p. 143‐51.

305. Dawes, M. and P.J. Chowienczyk, Drugs in pregnancy. Pharmacokinetics in pregnancy. Best Pract Res Clin Obstet Gynaecol, 2001. 15(6): p. 819‐26.

306. Anderson, G.D., Using pharmacokinetics to predict the effects of pregnancy and maternal‐infant transfer of drugs during lactation. Expert Opin Drug Metab Toxicol, 2006. 2(6): p. 947‐60.

307. Anderson, G.D., Pregnancy‐induced changes in pharmacokinetics: a mechanistic‐based approach. Clin Pharmacokinet, 2005. 44(10): p. 989‐1008.

308. Davison, J.M., W. Dunlop, and M. Ezimokhai, 24‐hour creatinine clearance during the third trimester of normal pregnancy. Br J Obstet Gynaecol, 1980. 87(2): p. 106‐9.

309. Anderson, B.J. and N.H. Holford, Mechanism‐based concepts of size and maturity in pharmacokinetics. Annu Rev Pharmacol Toxicol, 2008. 48: p. 303‐32.

310. Anderson, B.J. and N.H. Holford, Tips and traps analyzing pediatric PK data. Paediatr Anaesth, 2011. 21(3): p. 222‐37.

311. Green, B. and S.B. Duffull, What is the best size descriptor to use for pharmacokinetic studies in the obese? Br J Clin Pharmacol, 2004. 58(2): p. 119‐33.

312. Kuczmarski, R.J., et al., CDC growth charts: United States. Adv Data, 2000(314): p. 1‐27.

313. Karunajeewa, H.A., et al., Pharmacokinetics of chloroquine and monodesethylchloroquine in pregnancy. Antimicrob Agents Chemother, 2010. 54(3): p. 1186‐92.

314. Karunajeewa, H.A., et al., Pharmacokinetic properties of sulfadoxine‐pyrimethamine in pregnant women. Antimicrobial Agents & Chemotherapy, 2009. 53(10): p. 4368‐76.

315. Lalak, N.J. and D.L. Morris, Azithromycin clinical pharmacokinetics. Clin Pharmacokinet, 1993. 25(5): p. 370‐4.

316. Sidhu, A.B., et al., In vitro efficacy, resistance selection, and structural modeling studies implicate the malarial parasite apicoplast as the target of azithromycin. J Biol Chem, 2007. 282(4): p. 2494‐504.

317. Dunne, M.W., et al., A multicenter study of azithromycin, alone and in combination with chloroquine, for the treatment of acute uncomplicated Plasmodium falciparum malaria in India. J Infect Dis, 2005. 191(10): p. 1582‐8.

231

318. Miller, R.S., et al., Effective treatment of uncomplicated Plasmodium falciparum malaria with azithromycin‐quinine combinations: a randomized, dose‐ranging study. Am J Trop Med Hyg, 2006. 74(3): p. 401‐6.

319. Nakornchai, S. and P. Konthiang, Activity of azithromycin or erythromycin in combination with antimalarial drugs against multidrug‐resistant Plasmodium falciparum in vitro. Acta Trop, 2006. 100(3): p. 185‐91.

320. Noedl, H., et al., Azithromycin combination therapy with artesunate or quinine for the treatment of uncomplicated Plasmodium falciparum malaria in adults: a randomized, phase 2 clinical trial in Thailand. Clin Infect Dis, 2006. 43(10): p. 1264‐71.

321. Noedl, H., et al., In vitro antimalarial activity of azithromycin, artesunate, and quinine in combination and correlation with clinical outcome. Antimicrob Agents Chemother, 2007. 51(2): p. 651‐6.

322. Taylor, W.R., et al., Malaria prophylaxis using azithromycin: a double‐blind, placebo‐controlled trial in Irian Jaya, Indonesia. Clin Infect Dis, 1999. 28(1): p. 74‐81.

323. Kain, K.C., G.D. Shanks, and J.S. Keystone, Malaria chemoprophylaxis in the age of drug resistance. I. Currently recommended drug regimens. Clin Infect Dis, 2001. 33(2): p. 226‐34.

324. White, N.J., Intermittent presumptive treatment for malaria. PLoS Med, 2005. 2(1): p. e3.

325. Sarkar, M., et al., Pregnancy outcome following gestational exposure to azithromycin. BMC Pregnancy Childbirth, 2006. 6: p. 18.

326. Amsden, G.W. and C.L. Gray, Serum and WBC pharmacokinetics of 1500 mg of azithromycin when given either as a single dose or over a 3 day period in healthy volunteers. J Antimicrob Chemother, 2001. 47(1): p. 61‐6.

327. Cook, J.A., et al., Lack of a pharmacokinetic interaction between azithromycin and chloroquine. Am J Trop Med Hyg, 2006. 74(3): p. 407‐12.

328. Luke, D.R. and G. Foulds, Disposition of oral azithromycin in humans. Clin Pharmacol Ther, 1997. 61(6): p. 641‐8.

329. Wildfeuer, A., et al., Comparison of the pharmacokinetics of three‐day and five‐day regimens of azithromycin in plasma and urine. J Antimicrob Chemother, 1993. 31 Suppl E: p. 51‐6.

330. Ramsey, P.S., et al., Maternal and transplacental pharmacokinetics of azithromycin. Am J Obstet Gynecol, 2003. 188(3): p. 714‐8.

331. Gerhardy, C.L. and M. Garrett, Obstetrics and Gynaecology for Nurses and Midwives. 5 ed, ed. G.D.L. Mola and M. Voigt. 2002, Madang, Papua New Guinea: Lutheran School of Nursing.

332. Jonsson, E.N. and M.O. Karlsson, Xpose‐‐an S‐PLUS based population pharmacokinetic/pharmacodynamic model building aid for NONMEM. Comput Methods Programs Biomed, 1999. 58(1): p. 51‐64.

333. Upton, R.N. and G.L. Ludbrook, Pharmacokinetic‐pharmacodynamic modelling of the cardiovascular effects of drugs ‐ method development and application to magnesium in sheep. BMC Pharmacol, 2005. 5: p. 5.

334. Hodge, L.S. and T.S. Tracy, Alterations in drug disposition during pregnancy: implications for drug therapy. Expert Opin Drug Metab Toxicol, 2007. 3(4): p. 557‐71.

335. Krishna, S. and N.J. White, Pharmacokinetics of quinine, chloroquine and amodiaquine. Clinical implications. Clin Pharmacokinet, 1996. 30(4): p. 263‐99.

232

336. Adebayo, R.A., et al., Chloroquine‐induced pruritus in malaria fever: contribution of malaria parasitaemia and the effects of prednisolone, niacin, and their combination, compared with antihistamine. Br J Clin Pharmacol, 1997. 44(2): p. 157‐61.

337. Chandra, R., et al., Clinical pharmacokinetics and gastrointestinal tolerability of a novel extended‐release microsphere formulation of azithromycin. Clin Pharmacokinet, 2007. 46(3): p. 247‐59.

338. van Eijk, A.M. and D.J. Terlouw, Azithromycin for treating uncomplicated malaria. (Protocol). Cochrane Database Syst Rev, 2007(3): p. CD006688.

339. Grobusch, M.P., et al., Intermittent preventive therapy for malaria: progress and future directions. Curr Opin Infect Dis, 2007. 20(6): p. 613‐20.

340. Cairns, M., et al., Duration of protection against malaria and anaemia provided by intermittent preventive treatment in infants in Navrongo, Ghana. PLoS One, 2008. 3(5): p. e2227.

341. Kobbe, R., et al., A randomized controlled trial of extended intermittent preventive antimalarial treatment in infants. Clin Infect Dis, 2007. 45: p. 16‐25.

342. Dzinjalamala, F.K., et al., Association between the pharmacokinetics and in vivo therapeutic efficacy of sulfadoxine‐pyrimethamine in Malawian children. Antimicrob Agents Chemother, 2005. 49(9): p. 3601‐6.

343. Papua New Guinea Department of Health, Standard Treatment of Common Illnesses of Children in Papua New Guinea. 8 ed. 2005, Port Moresby: Papua New Guinea Department of Health.

344. Whelpton, R., G. Watkins, and S.H. Curry, Bratton‐Marshall and liquid‐chromatographic methods compared for determination of sulfamethazine acetylator status. Clinical Chemistry, 1981. 27(11): p. 1911‐4.

345. Anderson, B.J. and N.H. Holford, Mechanism‐based concepts of size and maturity in pharmacokinetics. Annual Review of Pharmacology & Toxicology, 2008. 48: p. 303‐32.

346. Allen, S.J., et al., Causes of preterm delivery and intrauterine growth retardation in a malaria endemic region of Papua New Guinea. Archives of Disease in Childhood Fetal & Neonatal Edition, 1998. 79(2): p. F135‐40.

347. Brair, M.E., et al., Reduced transfer of tetanus antibodies with placental malaria. Lancet, 1994. 343(8891): p. 208‐9.

348. Garner, P., et al., Birthweight and gestation of village deliveries in Papua New Guinea. Journal of Tropical Pediatrics, 1994. 40(1): p. 37‐40.

349. Filler, G. and N. Lepage, Should the Schwartz formula for estimation of GFR be replaced by cystatin C formula? Pediatric Nephrology, 2003. 18(10): p. 981‐5.

350. Corvaisier, S., et al., Population pharmacokinetics of pyrimethamine and sulfadoxine in children treated for congenital toxoplasmosis. Antimicrob Agents Chemother, 2004. 48(10): p. 3794‐800.

351. Hekster, C.A. and T.B. Vree, Clinical pharmacokinetics of sulphonamides and their N4‐acetyl derivatives. Antibiotics & Chemotherapy, 1982. 31: p. 22‐118.

352. Obua, C., et al., Population pharmacokinetics of chloroquine and sulfadoxine and treatment response in children with malaria: suggestions for an improved dose regimen. British Journal of Clinical Pharmacology, 2008. 65(4): p. 493‐501.

353. McLeod, R., et al., Levels of pyrimethamine in sera and cerebrospinal and ventricular fluids from infants treated for congenital toxoplasmosis. Toxoplasmosis Study Group. Antimicrobial Agents & Chemotherapy, 1992. 36(5): p. 1040‐8.

233

354. Weidekamm, E., et al., Plasma concentrations in pyrimethamine and sulfadoxine and evaluation of pharmacokinetic data by computerized curve fitting. Bulletin of the World Health Organization, 1982. 60(1): p. 115‐22.

355. Mansor, S.M., et al., Single dose kinetic study of the triple combination mefloquine/sulphadoxine/pyrimethamine (Fansimef) in healthy male volunteers. Br J Clin Pharmacol, 1989. 27(3): p. 381‐6.

356. Bustos, D.G., et al., Pharmacokinetics of sequential and simultaneous treatment with the combination chloroquine and sulfadoxine‐pyrimethamine in acute uncomplicated Plasmodium falciparum malaria in the Philippines. Tropical Medicine & International Health, 2002. 7(7): p. 584‐91.

357. Green, M.D., et al., Pharmacokinetics of sulfadoxine‐pyrimethamine in HIV‐infected and uninfected pregnant women in Western Kenya. Journal of Infectious Diseases, 2007. 196(9): p. 1403‐8.

358. Schwartz, G.J., L.P. Brion, and A. Spitzer, The use of plasma creatinine concentration for estimating glomerular filtration rate in infants, children, and adolescents. Pediatr Clin North Am, 1987. 34(3): p. 571‐90.

359. Andersen, T.B., et al., Measuring glomerular filtration rate in children; can cystatin C replace established methods? A review. Pediatric Nephrology, 2009. 24(5): p. 929‐41.

360. Bokenkamp, A., et al., Cystatin C‐‐a new marker of glomerular filtration rate in children independent of age and height. Pediatrics, 1998. 101(5): p. 875‐81.

361. Bouvet, Y., et al., GFR is better estimated by considering both serum cystatin C and creatinine levels. Pediatric Nephrology, 2006. 21(9): p. 1299‐306.

362. Grubb, A., et al., Simple cystatin C‐based prediction equations for glomerular filtration rate compared with the modification of diet in renal disease prediction equation for adults and the Schwartz and the Counahan‐Barratt prediction equations for children. Clinical Chemistry, 2005. 51(8): p. 1420‐31.

363. Zappitelli, M., et al., Derivation and validation of cystatin C‐based prediction equations for GFR in children. Am J Kidney Dis, 2006. 48(2): p. 221‐30.

364. Alcorn, J. and P.J. McNamara, Ontogeny of hepatic and renal systemic clearance pathways in infants: part I. Clinical Pharmacokinetics, 2002. 41(12): p. 959‐98.

365. Bjorkman, S., Prediction of cytochrome p450‐mediated hepatic drug clearance in neonates, infants and children : how accurate are available scaling methods? Clinical Pharmacokinetics, 2006. 45(1): p. 1‐11.

366. Allegaert, K., et al., Maturational pharmacokinetics of single intravenous bolus of propofol. Paediatric Anaesthesia, 2007. 17(11): p. 1028‐34.

367. Anand, K.J., et al., Morphine pharmacokinetics and pharmacodynamics in preterm and term neonates: secondary results from the NEOPAIN trial. British Journal of Anaesthesia, 2008. 101(5): p. 680‐9.

368. Anderson, B.J., et al., Vancomycin pharmacokinetics in preterm neonates and the prediction of adult clearance. British Journal of Clinical Pharmacology, 2007. 63(1): p. 75‐84.

369. Potts, A.L., G.R. Warman, and B.J. Anderson, Dexmedetomidine disposition in children: a population analysis. Paediatric Anaesthesia, 2008. 18(8): p. 722‐30.

370. Marfurt, J., et al., Low efficacy of amodiaquine or chloroquine plus sulfadoxine‐pyrimethamine against Plasmodium falciparum and P. vivax malaria in Papua New Guinea. Am J Trop Med Hyg, 2007. 77(5): p. 947‐54.

371. Baird, J.K., Real‐world therapies and the problem of vivax malaria. N Engl J Med, 2008. 359(24): p. 2601‐3.

234

372. May, J., et al., Therapeutic and prophylactic effect of intermittent preventive anti‐malarial treatment in infants (IPTi) from Ghana and Gabon. Malar J, 2008. 7: p. 198.

373. World Health Organization., World malaria report. 2010, World Health Organization: Geneva, Switzerland.

374. Bindschedler, M., et al., Cardiac effects of co‐artemether (artemether/lumefantrine) and mefloquine given alone or in combination to healthy volunteers. Eur J Clin Pharmacol, 2000. 56(5): p. 375‐81.

375. Ashley, E.A., et al., How much fat is necessary to optimize lumefantrine oral bioavailability? Trop Med Int Health, 2007. 12(2): p. 195‐200.

376. Ashley, E.A., et al., Pharmacokinetic study of artemether‐lumefantrine given once daily for the treatment of uncomplicated multidrug‐resistant falciparum malaria. Trop Med Int Health, 2007. 12(2): p. 201‐8.

377. Starzengruber, P., et al., Interaction between lumefantrine and monodesbutyl‐benflumetol in Plasmodium falciparum in vitro. Wien Klin Wochenschr, 2008. 120(19‐20 Suppl 4): p. 85‐9.

378. Starzengruber, P., et al., Specific pharmacokinetic interaction between lumefantrine and monodesbutyl‐benflumetol in Plasmodium falciparum. Wien Klin Wochenschr, 2007. 119(19‐20 Suppl 3): p. 60‐6.

379. World Health Organization Communicable Diseases Cluster., Severe falciparum malaria. Trans R Soc Trop Med Hyg, 2000. 94 Suppl 1: p. S1‐90.

380. Batty, K.T., et al., Selective high‐performance liquid chromatographic determination of artesunate and alpha‐ and beta‐dihydroartemisinin in patients with falciparum malaria. J Chromatogr B Biomed Appl, 1996. 677(2): p. 345‐50.

381. Matuszewski, B.K., M.L. Constanzer, and C.M. Chavez‐Eng, Strategies for the assessment of matrix effect in quantitative bioanalytical methods based on HPLC‐MS/MS. Anal Chem, 2003. 75(13): p. 3019‐30.

382. Bergstrand, M. and M.O. Karlsson, Handling Data Below the Limit of Quantification in Mixed Effect Models. Aaps Journal, 2009. 11(2): p. 371‐380.

383. Bergstrand, M., et al., A comparison of methods for handling of data below the limit of quantification in NONMEM VI, in PAGE 16 Abstr 1201. 2007.

384. Lefevre, G., Personal Communication. 2010. 385. Bell, D.J., et al., Population pharmacokinetics of sulfadoxine and pyrimethamine

in Malawian children with malaria. Clin Pharmacol Ther, 2011. 89(2): p. 268‐75. 386. Yang, S. and J. Roger, Evaluations of Bayesian and maximum likelihood methods

in PK models with below‐quantification‐limit data. Pharm Stat, 2010. 9(4): p. 313‐30.

387. Karunajeewa, H.A., et al., Pharmacokinetics and efficacy of piperaquine and chloroquine in Melanesian children with uncomplicated malaria. Antimicrob Agents Chemother, 2008. 52(1): p. 237‐43.

388. World Health Organization., Guidelines for the treatment of malaria ‐ 2nd edition. 2010, World Health Organization: Geneva, Switzerland.

389. ACTwatch. Malaria Drug Database. 2011 [cited 2011 September]; Available from: 8TUhttp://www.actwatch.info/resources/drugs_home03_search.asp U8T.

390. European Medicines Agency. Eurartesim. 2011 [cited 2011 September]; Available from: 8TUhttp://www.ema.europa.eu/docs/en_GB/document_library/Summary_of_opinion_‐_Initial_authorisation/human/001199/WC500108010.pdf U8T.

235

391. Artepharm Co Ltd. Artequick. 2011 [cited 2011 September]; Available from: 8TUhttp://www.artepharm.com/ProductShow/en.html U8T.

392. Zwang, J., et al., Safety and efficacy of dihydroartemisinin‐piperaquine in falciparum malaria: a prospective multi‐centre individual patient data analysis. PLoS One, 2009. 4(7): p. e6358.

393. Hasugian, A.R., et al., Dihydroartemisinin‐piperaquine versus artesunate‐amodiaquine: superior efficacy and posttreatment prophylaxis against multidrug‐resistant Plasmodium falciparum and Plasmodium vivax malaria. Clinical Infectious Diseases, 2007. 44(8): p. 1067‐74.

394. Anderson, B.J. and N.H. Holford, Mechanistic basis of using body size and maturation to predict clearance in humans. Drug Metab Pharmacokinet, 2009. 24(1): p. 25‐36.

395. Bonsch, C., et al., Chloroquine and its derivatives exacerbate B19V‐associated anemia by promoting viral replication. PLoS Negl Trop Dis, 2010. 4(4): p. e669.

396. Bergstrand, M., et al., Prediction‐corrected visual predictive checks for diagnosing nonlinear mixed‐effects models. Aaps Journal, 2011. 13(2): p. 143‐51.

397. Dondorp, A.M., et al., Artemisinin resistance in Plasmodium falciparum malaria. N Engl J Med, 2009. 361(5): p. 455‐67.

398. Porter, C.J., N.L. Trevaskis, and W.N. Charman, Lipids and lipid‐based formulations: optimizing the oral delivery of lipophilic drugs. Nat Rev Drug Discov, 2007. 6(3): p. 231‐48.

399. Crauwels, H.M., et al., Relative bioavailability of a concept paediatric formulation of TMC278, an investigational NNRTI, in 18th IAS Conference. 2010.

400. Batty, K.T., et al., The pharmacokinetics of artemisinin (ART) and artesunate (ARTS) in healthy volunteers. Am J Trop Med Hyg, 1998. 58(2): p. 125‐6.

401. Djimde, A.A., et al., Clearance of drug‐resistant parasites as a model for protective immunity in Plasmodium falciparum malaria. Am J Trop Med Hyg, 2003. 69(5): p. 558‐63.

402. Crevoisier, C., et al., Food increases the bioavailability of mefloquine. Eur J Clin Pharmacol, 1997. 53(2): p. 135‐9.

403. Crowe, A., et al., Role of P glycoprotein in absorption of novel antimalarial drugs. Antimicrob Agents Chemother, 2006. 50(10): p. 3504‐6.

404. Dien, T.K., et al., Effect of food intake on pharmacokinetics of oral artemisinin in healthy Vietnamese subjects. Antimicrob Agents Chemother, 1997. 41(5): p. 1069‐72.

405. Hien, T.T., et al., Orally formulated artemisinin in healthy fasting Vietnamese male subjects: a randomized, four‐sequence, open‐label, pharmacokinetic crossover study. Clin Ther, 2011. 33(5): p. 644‐54.

406. de Vries, P.J., et al., The pharmacokinetics of a single dose of artemisinin in patients with uncomplicated falciparum malaria. Am J Trop Med Hyg, 1997. 56(5): p. 503‐7.

407. de Vries, P.J. and T.K. Dien, Clinical pharmacology and therapeutic potential of artemisinin and its derivatives in the treatment of malaria. Drugs, 1996. 52(6): p. 818‐36.

408. Benjamin, J., et al., Artemisinin‐naphthoquine combination therapy for uncomplicated pediatric malaria: A tolerability, safety and preliminary efficacy study. Antimicrob Agents Chemother, 2012. 56(5): p. 2465‐71.

236

409. Kunming Pharmaceutical Corporation, Instruction for use of compound naphthoquine phosphate tablets. Product Information Brochure., 2006.

410. Davis, T.M., et al., Assessment of the effect of mefloquine on artesunate pharmacokinetics in healthy male volunteers. Antimicrob Agents Chemother, 2007. 51(3): p. 1099‐101.

411. Shi, B., et al., Quantitative analysis of artemether and its metabolite dihydroartemisinin in human plasma by LC with tandem mass spectrometry. Chromatographia, 2006. 64: p. 523‐530.

412. Rozet, E., et al., Advances in validation, risk and uncertainty assessment of bioanalytical methods. J Pharm Biomed Anal, 2011. 55(4): p. 848‐58.

413. German, P.I. and F.T. Aweeka, Clinical pharmacology of artemisinin‐based combination therapies. Clin Pharmacokinet, 2008. 47(2): p. 91‐102.

414. Price, R., et al., Pharmacokinetics of mefloquine combined with artesunate in children with acute falciparum malaria. Antimicrob Agents Chemother, 1999. 43(2): p. 341‐6.

415. Frisk‐Holmberg, M., et al., The single dose kinetics of chloroquine and its major metabolite desethylchloroquine in healthy subjects. Eur J Clin Pharmacol, 1984. 26(4): p. 521‐30.

416. Gustafsson, L.L., et al., Disposition of chloroquine in man after single intravenous and oral doses. Br J Clin Pharmacol, 1983. 15(4): p. 471‐9.

417. Wetsteyn, J.C., et al., The pharmacokinetics of three multiple dose regimens of chloroquine: implications for malaria chemoprophylaxis. Br J Clin Pharmacol, 1995. 39(6): p. 696‐9.

418. Boudreau, E.F., et al., Mefloquine kinetics in cured and recrudescent patients with acute falciparum malaria and in healthy volunteers. Clin Pharmacol Ther, 1990. 48(4): p. 399‐409.

419. Edwards, G., et al., Pharmacokinetics of chloroquine in Thais: plasma and red‐cell concentrations following an intravenous infusion to healthy subjects and patients with Plasmodium vivax malaria. Br J Clin Pharmacol, 1988. 25(4): p. 477‐85.

420. Obua, C., et al., Population pharmacokinetics of chloroquine and sulfadoxine and treatment response in children with malaria: suggestions for an improved dose regimen. Br J Clin Pharmacol, 2008. 65(4): p. 493‐501.

421. White, N.J., et al., Parenteral chloroquine for treating falciparum malaria. J Infect Dis, 1987. 155(2): p. 192‐201.

422. Beovic, B., et al., Influence of fever on the pharmacokinetics of ciprofloxacin. Int J Antimicrob Agents, 1999. 11(1): p. 81‐5.

423. Mackowiak, P.A., Influence of fever on pharmacokinetics. Rev Infect Dis, 1989. 11(5): p. 804‐7.

424. Welling, P.G., Effects of food on drug absorption. Annu Rev Nutr, 1996. 16: p. 383‐415.

425. Kurth, F., et al., Do paediatric drug formulations of artemisinin combination therapies improve the treatment of children with malaria? A systematic review and meta‐analysis. Lancet Infect Dis, 2010. 10(2): p. 125‐32.

426. Johnson, T.N., The problems in scaling adult drug doses to children. Arch Dis Child, 2008. 93(3): p. 207‐11.

427. Ritschel, W.A. and G.L. Kearns, Handbook of basic pharmacokinetics. 6th ed. 2004, Washington DC: American Pharmacists Association.

237

428. Nigerian‐German Chemicals PLC. Arco Product Information. 2011 [cited 2011 March]; Available from: 8TUwww.ngcplc.com/arco/index.htmlU8T.

429. Dondorp, A.M., et al., Artemisinin resistance: current status and scenarios for containment. Nat Rev Microbiol, 2010. 8(4): p. 272‐80.

430. World Health Organization, Guidelines for the treatment of malaria. 2nd ed. 2010, Geneva: World Health Organization.

431. United States National Institutes of Health. Intermittent Preventive Treatment With Azithromycin‐containing Regimens in Pregnant Women in Papua New Guinea (IPTp in PNG). 2012 [cited 2012 26 March]; Available from: 8TUhttp://clinicaltrials.gov/ct2/show/NCT01136850U8T.

432. Karunajeewa, H.A., et al., Pharmacokinetic properties of sulfadoxine‐pyrimethamine in pregnant women. Antimicrob Agents Chemother, 2009. 53(10): p. 4368‐76.

433. Gisleskog, P.O., M.O. Karlsson, and S.L. Beal, Use of prior information to stabilize a population data analysis. J Pharmacokinet Pharmacodyn, 2002. 29(5‐6): p. 473‐505.

238

239

xi 150B150BAppendix

xi.a 160B160BAppendixA:Fulltextsofpublicationsrelatedtothethesis

240

ANTIMICROBIAL AGENTS AND CHEMOTHERAPY, Jan. 2010, p. 360–366 Vol. 54, No. 10066-4804/10/$12.00 doi:10.1128/AAC.00771-09Copyright © 2010, American Society for Microbiology. All Rights Reserved.

Pharmacokinetic Properties of Azithromycin in Pregnancy�

Sam Salman,1 Stephen J. Rogerson,2 Kay Kose,3 Susan Griffin,3 Servina Gomorai,3 Francesca Baiwog,3Josephine Winmai,3 Josin Kandai,3 Harin A. Karunajeewa,1,4 Sean J. O’Halloran,5 Peter Siba,3

Kenneth F. Ilett,1,5 Ivo Mueller,3 and Timothy M. E. Davis1*School of Medicine and Pharmacology, University of Western Australia, Perth, Western Australia, Australia1; Faculty of Medicine,

University of Melbourne, Melbourne, Australia2; Papua New Guinea Institute of Medical Research, Madang, Papua New Guinea3;Western Health, Melbourne, Australia4; and Clinical Pharmacology and Toxicology Laboratory,

Path West Laboratory Medicine, Nedlands, Australia5

Received 9 June 2009/Returned for modification 17 September 2009/Accepted 19 October 2009

Azithromycin (AZI) is an azalide antibiotic with antimalarial activity that is considered safe in pregnancy.To assess its pharmacokinetic properties when administered as intermittent preventive treatment in pregnancy(IPTp), two 2-g doses were given 24 h apart to 31 pregnant and 29 age-matched nonpregnant Papua NewGuinean women. All subjects also received single-dose sulfadoxine-pyrimethamine (SP) (1,500 mg or 75 mg)or chloroquine (450-mg base daily for 3 days). Blood samples were taken at 0, 1, 2, 3, 6, 12, 24, 32, 40, 48, and72 h and on days 4, 5, 7, 10, and 14 for AZI assay by ultra-high-performance liquid chromatography-tandemmass spectrometry. The treatments were well tolerated. Using population pharmacokinetic modeling, a three-compartment model with zero-order followed by first-order absorption and no lag time provided the best fit.The areas under the plasma concentration-time curve (AUC0–�) (28.7 and 31.8 mg � h liter�1 for pregnant andnonpregnant subjects, respectively) were consistent with the results of previous studies, but the estimatedterminal elimination half-lives (78 and 77 h, respectively) were generally longer. The only significant relation-ship for a range of potential covariates, including malarial parasitemia, was with pregnancy, which accountedfor an 86% increase in the volume of distribution of the central compartment relative to bioavailability withouta significant change in the AUC0–�. These data suggest that AZI can be combined with compounds with longerhalf-lives, such as SP, in combination IPTp without the need for dose adjustment.

Azithromycin (AZI) is a semisynthetic azalide antibiotic thatis structurally related to erythromycin but has a broader spec-trum of antibacterial activity and a more favorable pharmaco-kinetic profile (13, 22). It is widely used in the treatment ofrespiratory and sexually transmitted infections, including thosein HIV-infected patients (32, 34). AZI also inhibits proteinsynthesis in the plasmodial apicoplast (39, 40) and thus hasactivity against both Plasmodium falciparum and Plasmodiumvivax (5, 12, 16, 27–30, 41). It acts mainly against the progenyof parasites that inherit a nonfunctioning apicoplast after ex-posure, with the result that its antimalarial effect has a slowonset and is relatively weak. Therefore, AZI is best used incombination with other antimalarial compounds as both treat-ment (20, 27, 29) and chemoprophylaxis (5, 19), with likelyadditive or synergistic effects (28, 30, 31).

Malaria in pregnancy can result in adverse outcomes forboth mother and fetus (14). Intermittent preventive treatmentin pregnancy (IPTp) aims to reduce the burden of malaria byadministering treatment doses of antimalarial drugs at prede-termined intervals as part of routine antenatal care in areas ofendemicity (44). Because AZI is considered safe in pregnancyand could have activity against other clinically significantpathogens (8, 38), it has been suggested as a candidate forIPTp. Although the pharmacokinetics of AZI have been inves-

tigated (2, 6, 7, 11, 13, 23–26, 35–37, 45), only one studyincluded pregnant women (36), and most focused on its anti-bacterial properties. In addition, AZI is likely to be partneredwith conventional antimalarial drugs if given as IPTp, andthere is evidence that such combinations are safe and welltolerated in studies with chloroquine (CQ) in healthy volun-teers (11) and with sulfadoxine-pyrimethamine (SP) in preg-nant women (20). Although there does not appear to be aclinically significant pharmacokinetic interaction with CQ (11),AZI interactions with other conventional IPTp treatments areunknown. Therefore, we investigated the pharmacokineticproperties of AZI in combination with CQ or SP in pregnantand nonpregnant women from an area of Papua New Guinea(PNG) with intense transmission of both P. falciparum and P.vivax malaria.

MATERIALS AND METHODS

Study site, sample, and approvals. The present study was conducted at Alex-ishafen Health Centre, Madang Province, on the north coast of PNG. The preg-nant women were recruited at their first antenatal clinic visit, and the age-matched nonpregnant volunteers were from the same communities as thepregnant participants. Women were eligible if (i) they had not taken any of thestudy drugs in the previous 28 days, (ii) they had no history of significant allergyto any study drug, (iii) there was no significant comorbidity or clinical evidenceof severe malaria, and (v) follow-up was possible for the duration of the study.The study was approved by the Medical Research Advisory Committee of PNGand the Human Ethics Research Committee at the University of Western Aus-tralia. Written informed consent was obtained from all participants.

Clinical procedures. A detailed assessment was performed prior to drug ad-ministration, including a side effects questionnaire, point-of-care hemoglobinand blood glucose (HemoCue, Angelholm, Sweden), thick and thin blood films,and (for pregnant participants) estimation of gestational age by fundal height. A3-ml blood sample was taken for subsequent antimalarial drug assay. All women

* Corresponding author. Mailing address: Department of Medicine,Fremantle Hospital, P.O. Box 480, Fremantle 6959, Western Australia,Australia. Phone: (618) 9431 3229. Fax: (618) 9431 2977. E-mail:[email protected].

� Published ahead of print on 26 October 2009.

360

by on Decem

ber 18, 2009 aac.asm

.orgD

ownloaded from

received 2 g AZI (Zithromax; Pfizer, New York, NY) both at enrolment and 24 hlater. Subjects were also randomized to receive single-dose SP (1,500 mg or 75mg; Fansidar; Roche, Basel, Switzerland) at enrolment (AZI-SP arm) or CQ(Chloroquin; Astra, Sydney, Australia) (450 mg base daily for 3 days; AZI-CQarm) in accordance with regimens recommended for PNG (15). The adminis-tration of all doses was directly observed. The dosing schedule for AZI waschosen as the simplest regimen that would be likely to ensure effective drugconcentrations during the first 4 days of treatment (40).

Following the first dose of AZI (day 0), additional blood samples were takenat 1, 2, 3, 6, 12, 24, 32, 40, 48, and 72 h and then on days 4, 5, 7, 10, and 14 fordrug assay. The exact timing of each blood sample was recorded. All sampleswere centrifuged promptly, with red cells and separated plasma stored frozen at�80°C. The side effects questionnaire was readministered at 6 h and then at 1,2, 3, and 7 days. Hemoglobin, erect and supine heart rates and blood pressure,respiratory rate, temperature, and blood slides were taken on days 1, 2, 3, 7, 14,28, and 42, and blood glucose was measured on days 1, 2, and 3. After thecompletion of follow-up, pregnant patients were returned to the usual antenatalcare.

Laboratory methods. Giemsa-stained thick blood smears were examined in-dependently by at least two skilled microscopists who were blinded to pregnancyand treatment status. Each microscopist viewed �100 fields at �1,000 magnifi-cation before a slide was considered negative. Any slide discrepant for positivity/negativity or species identification was referred to a third microscopist.

AZI levels were measured using a validated ultrahigh-performance liquidchromatography-tandem mass spectrometry (UPLC-LCMS-MS) method using adeuterated internal standard. The samples were retained for subsequent SPassay. AZI USP was obtained from APAC Pharmaceutical LLC (Ellicott City,MD) and deuterated AZI from Toronto Research Chemicals (North York,Canada). In brief, following the addition of an internal standard, AZI wasextracted from 5 �l of plasma by protein precipitation. After centrifugation,supernatant (5 �l) was injected onto a 2795/Quattro Premier XE UPLC-ESI-MS/MS (Waters Corp, MA) using a Waters BEH C18 1.7-�m, 2.1- by 100-mmcolumn. Gradient elution was performed using mobile phases A (45/55 [vol/vol],comprising 1 g/liter ammonium bicarbonate in 50/50 [vol/vol] methanol-waterand acetonitrile) and B (50/50 [vol/vol] methanol-acetonitrile) at 0.4 ml/min.Adduct transitions were monitored using positive electrospray ionization withmultiple-reaction monitoring for AZI and d3-AZI and were m/z 749.6 to 591.4and m/z 752.6 to 594.4, respectively. The method was linear to 1,012 ng/ml (r2 �0.9997) with a limit of quantification of 2.5 �g/liter AZI. All inter- and intradaycoefficients of variation were �10%, and the between-subject variability (BSV)was �5% when matrix effects were investigated at three concentrations.

Population pharmacokinetic analysis. Concentration-time data sets were an-alyzed by nonlinear mixed-effect modeling using NONMEM (version 6.2.0; IconDevelopment Solutions, Ellicott City, MD) with an Intel Visual FORTRAN 10.0compiler. Linear mamillary model subroutines within NONMEM (ADVAN4and -12 used with TRANS4 in the PREDPP library), first-order conditionalestimation (FOCE) with �-ε interaction, and the objective function value (OFV)(a NONMEM-calculated global goodness-of-fit indicator equal to �2 log-likeli-hood value of data) were used to construct and compare plausible models.Unless otherwise specified, a difference in the OFV of �6.63 (�2 distributionwith 1 df; P � 0.01) was considered significant. The R-based model-building aidXpose 6.0 (http://www.r-project.org/) was used for graphic model diagnosis (18).Secondary pharmacokinetic parameters, including the volume of distribution atsteady state (VSS � V1 � V2 � … � Vn), area under the curve (AUC0–�), andelimination half lives (t1/2), for the nonpregnant and pregnant groups wereobtained from post hoc Bayesian prediction in NONMEM using the final modelparameters. Macro constants for the three-compartment model were calculatedfrom the modeled parameters using previously published equations (42).

All volume and clearance terms were scaled allometrically using [ � (bodyweight/70)1.0] and [ � (body weight/70)0.75], respectively (3), and were expressedrelative to bioavailability (/F). Two- and three-compartment models were com-pared, and then zero- and first-order absorption models with and without a lagtime were assessed alone and in combination. The BSV was added to parametersfor which it could be estimated reasonably from the available data. Both expo-nential (proportional) and combined (exponential plus additive) error modelswere tested for residual unexplained variability (RUV). In developing the finalmodels, we investigated the influence of the covariates pregnancy, treatmenttype, fundal height, gestational age, malaria status, blood glucose, and hemoglo-bin on model parameters using Xpose and the generalized additive modelingprocedure function, as well as inspection of correlation plots. Covariate relation-ships found in this way were evaluated within the NONMEM model. Inclusion ofthe covariate required a decrease of �3.84 in the OFV (�2 df � 2; P � 0.05) and

a decrease in the BSV. Correlations among BSV terms and weighted-residuals(WRES) plots were used in model evaluation.

A bootstrap procedure using Perl speaks NONMEM (PSN) (http://psn.sourceforge.net) was used to sample individuals from the original data set withreplacement and to generate 1,000 new data sets that were subsequently analyzedusing NONMEM. The resulting parameters were then summarized as medianand 2.5th and 97.5th percentiles (95% empirical confidence interval [CI]) tofacilitate validation of the final model parameter estimates. In addition, a strat-ified visual predictive check (VPC) was also performed using PSN with 1,000replicate data sets simulated from the original. The resulting 80% predictionintervals (PI) for AZI were plotted with the observed data to assess the predic-tive performance of the model.

Statistical analysis. SigmaStat (version 3.10; Systat Software Inc., Chicago, IL)was used for statistical analysis unless otherwise specified. Data are summarizedas mean � standard deviation (SD) or median and interquartile range (IQR) asappropriate. Student’s t test or the Mann-Whitney U test was used for two-sample comparisons. Categorical data were compared using either the Pearsonchi-square or Fisher’s exact test, and multiple means were compared byrepeated-measures analysis of variance (ANOVA). A two-tailed level ofsignificance of 0.05 was used. Drug concentrations at each time point after day2 were compared to the AUC0–� using Pearson correlation.

RESULTS

Patient characteristics. A total of 31 pregnant and 29 non-pregnant women were recruited between October 2007 andMarch 2008. All subjects took two AZI doses, but two pregnantpatients did not receive either CQ or SP. These women wereexcluded from initial analyses but were included subsequentlyif there was no effect of CQ or SP on AZI pharmacokineticproperties in the other subjects. Baseline characteristics ofthe subjects by pregnancy status and treatment allocation areshown in Table 1. The groups were well matched, except that,consistent with normal physiological changes that occur inpregnancy (4, 17), the pregnant subjects were significantlyheavier and had lower hemoglobin than the nonpregnant sub-jects for each treatment group (P � 0.05). Seven of the preg-nant patients were parasitemic at baseline compared with onlyone of the nonpregnant subjects (P � 0.02).

Efficacy, tolerability, and safety. Three of the seven P. fal-ciparum and one of the two P. vivax cases at baseline receivedAZI-SP. There was an uncorrected adequate parasitologicaland clinical response (APCR) of 100% for both treatments. Afurther eight cases (five of whom were pregnant) became slidepositive for P. falciparum and three (two who were pregnant)for P. vivax late in the 42-day follow-up period. All receivedrecommended antimalarial therapy (15). All cases at baselineand during follow-up were asymptomatic.

Both treatments were well tolerated, and no patient re-quired medical attention because of side effects. Table 2 sum-marizes self-reported symptoms in the first week of follow-up,�90% of which were mild (not influencing usual daily activity)and short-lived (�2 days). Six patients reported mild pretreat-ment symptoms (headache, abdominal pain, pruritus, or dizzi-ness), but these resolved subsequently. Posttreatment prurituswas reported only in the AZI-CQ group (P � 0.052). Althoughnot formally assessed, no significant side effects were volun-teered at assessments after day 7. No patient developed hypo-glycemia (blood glucose � 2.5 mmol/liter) or severe anemia(hemoglobin � 5.0 g/dl). Although postural hypotension (�20mm Hg systolic or �10 mm Hg diastolic fall after standing)occurred eight times in seven pregnant (four from the AZI-CQgroup) and seven times in five nonpregnant (all five in theAZI-CQ group) patients, the differences between groups were

VOL. 54, 2010 PHARMACOKINETICS OF AZITHROMYCIN IN PREGNANCY 361

by on Decem


.orgD

ownloaded from

not significant and there were no associated symptoms. Aftercompletion of the study, one of the study participants had astillbirth. A medical review of her case notes by three indepen-dent physicians concluded that it was unlikely to be the resultof the study medication.

Pharmacokinetic modeling. A three-compartment modelhad a lower OFV than a two-compartment model (8,700.058versus 8,185.104; P � 0.001 by �2 test; df � 2) and a morefavorable distribution of WRES over time. Zero-order fol-lowed by first-order absorption without a lag time provided thelowest OFV and best fit for AZI absorption. The fixed modelparameters were DUR (the duration of the zero-order absorp-tion); ka (the first-order absorption rate constant); CL/F (clear-ance from the central compartment); VC/F, VP1/F, and VP2/F(volumes of distribution of the central, first peripheral, andsecond peripheral compartments, respectively); and Q1/F andQ2/F (intercompartment clearances for VP1/F and VP2/F, re-

spectively). The model structure is shown in Fig. 1. BSV couldbe estimated for DUR, CL/F, VC/F, and VP1/F, while a pro-portional-error model was best for RUV. After testing thevarious covariates, only pregnancy on VC/F produced a signif-icant decrease in the OFV (�2 df � 1; P � 0.05) accompaniedby a decrease in the BSV of VC/F from 111.0% to 99.6%.

The results of the parameter estimates and their relativestandard errors (RSE) are summarized in Table 3 and second-ary parameter estimates in Table 4. All drug concentrationsafter day 2 were strongly correlated with the AUC0–� (r � 0.7;P � 0.001), with 96-h levels showing the strongest association(r � 0.78). The bootstrap results (Table 3) demonstrate arobust estimation of both fixed and random parameters withbias � 4% and � 5%, respectively. Goodness-of-fit plots ofobserved versus population and individual predicted concen-trations and WRES versus time are shown in Fig. 2 and 3. TheVPC results, stratified for pregnancy status, are presented inFig. 4 and show reasonable predictive performance of themodel while demonstrating some difficulty in capturing post-absorption plasma concentrations peaks.

DISCUSSION

The present study is the first pharmacokinetic evaluation ofAZI in pregnant and nonpregnant women living in a malaria-endemic area. We found that a three-compartment model witha combined absorption process best described the dispositionof AZI in our subjects. Both two-compartment (23, 26, 37) andthree-compartment (6, 35) models have been found to bestdescribe AZI plasma concentration-time profiles in other con-texts. Our ability to differentiate the triexponential eliminationof AZI may have been facilitated by the relatively long sam-pling duration. This may also explain why our estimated ter-minal elimination half-lives (78 and 77 h for pregnant andnonpregnant participants, respectively) were longer than thosein most previous studies (range, 27 to 79 h) (6, 7, 13, 23, 35, 37).The overall drug exposure (AUC0–�, 28.7 and 31.8 mg � hliter�1 for pregnant and nonpregnant subjects, respectively)

TABLE 1. Baseline characteristics of the study participants by pregnancy status and treatment allocation

Parameter

Valuea

Pregnant Nonpregnant

AZI-CQ (n � 15) AZI-SP (n � 14) AZI-CQ (n � 14) AZI-SP (n � 15)

Age (yr) 26.9 � 4.1 23.9 � 5.1 25.7 � 5.8 27 � 6.5Wt (kg) 53.5 � 7.1b 56.4 � 7.9a 51.4 � 5.4 51.9 � 4.9Height (cm) 154 � 7.4 154 � 7.3 154 � 6.4 154 � 2.8Axillary temp (°C) 36.4 � 0.7 36.5 � 0.6 36.7 � 0.3 36.4 � 0.3P. falciparum parasitemia 3 (20) 3 (21) 1 (7) 0 (0)P. vivax parasitemia 1 (7) 0 (0) 0 (0) 1 (7)Gestational age (wk) 24 �22–27� 21 �19–24�Gravidity 3 �2–5� 2 �1–4� 1 �0–3� 2 �0–3�Parity 2 �1–4� 1 �0–2� 0 �0–3� 1 �0–3�Respiratory rate (/min) 20 � 1 22 � 5 20 � 2 20 � 1Supine pulse rate (/min) 91 � 10 89 � 7 82 � 10 88 � 7Supine MAP (mm Hg)c 78 � 7 81 � 10 79 � 9 82 � 7Hemoglobin (g/dl) 8.5 � 1.6b 8.2 � 1.2b 9.3 � 1.9 10 � 1.3Blood glucose (mmol/liter) 5.9 � 1.6 5.7 � 0.8 6.2 � 1.1 5.6 � 2.7

a Data are mean � SD, median �IQR�, or number (%).b P � 0.05 versus nonpregnant subjects.c Mean arterial pressure, calculated by adding 1/3 of the pulse pressure (systolic minus diastolic pressure) to the diastolic pressure.

TABLE 2. Side effects reported during the first week afterinitiation of treatment

ParameterValuea

AZI-CQ (n � 29) AZI-SP (n � 29)

Fever 2 (7) 1 (3)Chills 2 (7) 0 (0)Headache 6 (21) 4 (14)Nausea 4 (14) 7 (24)Vomiting 2 (7) 4 (14)Diarrhea 2 (7) 2 (7)Abdominal pain 4 (14) 3 (10)Rash 0 (0) 0 (0)Pruritus 5 (17) 0 (0)Anorexia 1 (3) 2 (7)Insomnia 2 (7) 0 (0)Dizziness 3 (10) 1 (3)Bone or joint pain 1 (3) 1 (3)Otherb 5 (17) 1 (3)

a Data are numbers (%) of patients.b Cough (2), blocked ear (1), “heavy head” (1), and numbness of calf muscles

(1) in the AZI-CQ group and cough (1) in the AZI-SP group.

362 SALMAN ET AL. ANTIMICROB. AGENTS CHEMOTHER.

by on Decem


.orgD

ownloaded from

was within the range expected from dose-scaled results fromprevious studies in other contexts (26.5 to 46.4 mg � h liter�1)(2, 7, 11, 13, 23, 24, 26, 35, 37), suggesting that the bioavail-ability of AZI is not dose dependent.

Both zero-order (23, 37) and first-order (6, 26) absorptionhave been reported previously for AZI, but neither was appro-priate for our data. A combined absorption process in whichthe drug enters the absorption compartment in a zero-ordermanner and then is absorbed according to first-order kineticsprovided the best model in the present study. This is analogousto the twin processes of (i) gastric emptying of the drug into thesmall intestine (the zero-order process) and (ii) absorption inthe small intestine proportional to the amount present (thefirst-order process). Despite this more complex model, AZI

absorption was still not well characterized in our final model.This has been reported previously (37) but is unlikely to besignificant in the treatment of uncomplicated malaria, whereexposure of the parasite to therapeutic drug concentrationsover several life cycles is more important than that immedi-ately after drug administration.

Plasma AZI concentrations appeared to differ betweenpregnant and nonpregnant women only in the first 48 h afterthe first dose. This was confirmed by the population pharma-cokinetic modeling, in which pregnancy, the only significantcovariate relationship, accounted for an 86% increase in VC/F.Despite significant differences in the secondary parametersVC/F, VP2/F, VSS/F, and t1/2� (first-distribution half-life) be-tween pregnant and nonpregnant subjects, no difference was

FIG. 1. Structural model used in the final pharmacokinetic analysis of plasma azithromycin concentrations in the central compartment versustime. GUT, gastrointestinal tract.

TABLE 3. Model building, final parameter estimates, and bootstrap results from the AZI population pharmacokinetic modeling

Parametera

Value

Base model Final covariatemodel

Bootstrap (n � 1,000)(median �95% CI�)

OFV 7,999.870 7,993.646 7,974.699 �7,756.673–8,201.238�

Pharmacokinetic (estimate �% RSE�)DUR (h) 1.66 �10.4� 1.55 �3.3� 1.56 �1.21–2.01�ka (h�1) 0.513 �3.2� 0.525 �14.8� 0.524 �0.451–0.623�VC/F (liters) 504 �13.9� 384 �17.6� 371 �235–554�Pregnancy on VC/F (liters) 330 �69.4� 318 �48–604�CL/F (liters h�1) 158 �3.9� 158 �6.7� 158 �145–171�VP1/F (liters) 4,080 �8.6� 4,080 �12.5� 4,045 �3,402–4,870�Q1/F (liters h�1) 327 �5.7� 325 �12.7� 326 �288–368�VP2/F (liters) 5,070 �5.6� 5,040 �7.3� 5,070 �4,262–5,730�Q2/F (liters h�1) 67.2 �11.5� 66.4 �12.4� 67.5 �48.5–84.0�

Random (CV % �% RSE�)BSV Vc/F 111.4 �20.9� 99.6 �35.5� 99.0 �72.6–127.6�BSV CL/F 28.3 �24.1� 28.3 �33.1� 27.9 �21.6–34.5�BSV VP1/F 35.8 �27.0� 35.6 �27.2� 34.8 �25.7–45.2�BSV DUR 73.0 �21.4� 76.9 �22� 75.5 �55.4–95.6�RUVProportional error (CV % �% RSE�) 31.3 �9.5� 31.2 �15.1� 30.9 �28.1–33.8�

a CV, coefficient of variation.


by on Decem


.orgD

ownloaded from

seen in t1/2� (terminal elimination half-life) or AUC0–�. Thissuggests that the drug elimination and overall exposures weresimilar in the two groups. A much shorter AZI half-life (12 h)than in the present study was reported previously in pregnant

women (36), but the study employed a shorter sampling dura-tion (168 versus 336 h) and included pregnant women at ornear term, and the analysis was constrained by relatively sparsesampling.

Because of the need for AZI to be combined with othertherapies (12, 41), we included conventional antimalarial drugscurrently recommended as part of IPTp in PNG and othercountries (10, 20). There were no significant differences in thedisposition of AZI between the AZI-CQ and AZI-SP groups,consistent with a study of the interaction of CQ and AZI inhealthy volunteers (11). We conclude that AZI dose modifi-cation is unnecessary in these combinations. In addition, thelack of an effect of malaria status as a covariate on AZI dis-position suggests that, unlike drugs such as quinine (21), thedose may not have to be adjusted when parasitemia is present.

The most common side effects of AZI, especially with higherdoses, are nausea and vomiting. These symptoms are thoughtto be related to the effect of AZI on the motilin receptor in theupper gastrointestinal tract (33). However, with the exceptionof pruritus, which tended to be associated with AZI-CQ ther-apy, consistent with known CQ effects (1), there were no dif-ferences in the incidences of side effects between the twotreatment groups, and most reported adverse effects were mild.

TABLE 4. Secondary pharmacokinetic parameters derived from post hoc Bayesian estimates for pregnant and nonpregnant study participants(median �IQR�)

ParameterValue

Pregnant (n � 31) Nonpregnant (n � 29) P value

DUR (h) 1.65 �0.94–2.34� 1.75 �1.02–2.38� NSa

ka (h�1) 0.525 �0.525–0.525� 0.525 �0.525–0.525� NSVC/F (liters) 647 �422–995� 249 �157–363� �0.001VP1/F (liters) 3,620 �2,747–3,951� 2,909 �2,296–3,586� NSVP2/F (liters) 3,888 �3,708–4,104� 3,672 �3,456–3,888� 0.034VSS/F (liters) 8,355 �7,460–8,973� 6,875 �6,115–7,526� 0.002it1/2�

b (h) 0.88 �0.57–1.36� 0.39 �0.24–0.56� �0.001t1/2�

b (h) 20.7 �18.3–22.8� 18.8 �15.3–21� NSt1/2�

b (h) 78.2 �74–82.5� 77.1 �71.5–84.5� NSAUC0–� (�g h liter�1) 28,713 �25,913–32,942� 31,781 �28,736–38,012� NS

a NS, not significant.b t1/2�, t1/2�, and t1/2� are the first-distribution, second-distribution, and terminal elimination half-lives respectively.

FIG. 2. Observed versus model predicted concentrations (A) andindividual predicted concentrations (B) for AZI. The solid gray linesare the lines of identity, while the dashed black lines are the linearregression lines of best fit.

FIG. 3. Weighted residuals versus time after dose (log scale) plotfor AZI.


by on Decem


.orgD

ownloaded from

The AZI dose regimen in both combination therapy groups inthe present study (2.0 g daily for 2 days) was associated with aside effect profile similar to that reported previously after asingle 2.0-g dose (35). Use of the sustained-release formulationof AZI should reduce side effects, including nausea and vom-iting (9). However, this formulation has a bioavailability of82.8% relative to conventional AZI, suggesting that a higherdose will be required to achieve the same drug exposure. Aswell as increasing the cost of AZI treatment, this could meanthat side effects are more frequent with higher-dose sustained-release AZI administration.

Although the present study had limited subject numbers, it isencouraging that both regimens achieved a 100% uncorrectedAPCR. The plasma concentrations of AZI required to achievecure are unknown, as no efficacy trials have included thesedata. However, the high correlation between 96-h drug levelsand AUC0–� in our patients suggests that a day 4 plasmaconcentration could be an appropriate surrogate for overallAZI exposure in efficacy trials in which serial blood sampling isproblematic. It is interesting that prolongation of the in vitroexposure of P. falciparum to 96 h results in substantially in-

creased potency, suggesting that either AZI renders second-generation parasites unable to establish a parasitophorous vac-uole upon host cell invasion or the effect on apicoplast proteinsynthesis inhibits successful development of the progeny ofdrug-treated parasites (40).

Given the need for relatively prolonged parasite exposure totherapeutic plasma concentrations, it is unlikely that the ben-efit of “front loading” of AZI used in treating bacterial infec-tions (23, 24) will be relevant in malaria. However, experiencewith AZI as an antimalarial agent is growing. A Cochranereview of its efficacy is currently under way (43), and promisingresults are being seen when it is used with SP in IPTp, such asmight be given at least twice during pregnancy (20). Thepresent study provides a pharmacokinetic foundation for thefurther investigation of AZI as an antimalarial agent in preg-nancy, particularly in combination IPTp. Further data from thepresent study should also determine whether AZI influencesthe disposition of CQ and SP. Although there was a significantincrease in AZI VC/F in pregnant women, there was no signif-icant change in the AUC0-�, and it is therefore likely that nodose adjustments will be required for pregnant women whenAZI is given in combination with CQ or SP.

ACKNOWLEDGMENTS

We are most grateful to Sr. Valsi Kurian and the staff of AlexishafenHealth Centre for their kind cooperation during the study. We alsothank Christine Kalopo and Bernard (Ben) Maamu for clinical and/orlogistic assistance.

The study was funded by the National Health and Medical ResearchCouncil (NHMRC) of Australia (grant 458555) and was supported andendorsed by the MiP consortium, which is funded through a grant fromthe Bill and Melinda Gates Foundation to the Liverpool School ofTropical Medicine. T.M.E.D. is supported by an NHMRC PractitionerFellowship.

REFERENCES

1. Adebayo, R. A., G. G. Sofowora, O. Onayemi, S. J. Udoh, and A. A. Ajayi.1997. Chloroquine-induced pruritus in malaria fever: contribution of malariaparasitaemia and the effects of prednisolone, niacin, and their combination,compared with antihistamine. Br. J. Clin. Pharmacol. 44:157–161.

2. Amsden, G. W., and C. L. Gray. 2001. Serum and WBC pharmacokinetics of1500 mg of azithromycin when given either as a single dose or over a 3 dayperiod in healthy volunteers. J. Antimicrob. Chemother. 47:61–66.

3. Anderson, B. J., and N. H. Holford. 2008. Mechanism-based concepts of sizeand maturity in pharmacokinetics. Annu. Rev. Pharmacol. Toxicol. 48:303–332.

4. Anderson, G. D. 2005. Pregnancy-induced changes in pharmacokinetics: amechanistic-based approach. Clin. Pharmacokinet. 44:989–1008.

5. Anderson, S. L., A. J. Oloo, D. M. Gordon, O. B. Ragama, G. M. Aleman,J. D. German, D. B. Tang, M. W. Dunne, and G. D. Shanks. 1998. Successfuldouble-blinded, randomized, placebo-controlled field trial of azithromycinand doxycycline as prophylaxis for malaria in Western Kenya. Clin. Infect.Dis. 26:146–150.

6. Ballow, C. H., G. W. Amsden, V. S. Highet, and A. Forrest. 1998. Pharma-cokinetics of oral azithromycin in serum, urine, polymorphonuclear leuco-cytes and inflammatory vs non-inflammatory skin blisters in healthy volun-teers. Clin. Drug Investig. 15:159–167.

7. Boonleang, J., K. Panrat, C. Tantana, S. Krittathanmakul, and W. Jintapa-korn. 2007. Bioavailability and pharmacokinetic comparison between ge-neric and branded azithromycin capsule: a randomized, double-blind, 2-waycrossover in healthy male Thai volunteers. Clin. Ther. 29:703–710.

8. Centers for Disease Control and Prevention. 2006. Sexually transmitteddiseases treatment guidelines, 2006: diseases characterized by urethritis andcervicitis. MMWR Morb. Mortal. Wkly. Rep. 55:35–49.

9. Chandra, R., P. Liu, J. D. Breen, J. Fisher, C. Xie, R. LaBadie, R. J. Benner,L. J. Benincosa, and A. Sharma. 2007. Clinical pharmacokinetics and gas-trointestinal tolerability of a novel extended-release microsphere formula-tion of azithromycin. Clin. Pharmacokinet. 46:247–259.

10. Chico, R. M., R. Pittrof, B. Greenwood, and D. Chandramohan. 2008.Azithromycin-chloroquine and the intermittent preventive treatment of ma-laria in pregnancy. Malar. J. 7:255.

FIG. 4. Visual predicted check plots showing simulated 10th (shortdashed lines), 50th (dotted lines), and 90th (solid lines) percentileconcentrations and observed concentration (log scale) data (open cir-cles) versus time (log scale) for nonpregnant (A) and pregnant (B) par-ticipants.


by on Decem


.orgD

ownloaded from

11. Cook, J. A., E. J. Randinitis, C. R. Bramson, and D. L. Wesche. 2006. Lackof a pharmacokinetic interaction between azithromycin and chloroquine.Am. J. Trop. Med. Hyg. 74:407–412.

12. Dunne, M. W., N. Singh, M. Shukla, N. Valecha, P. C. Bhattacharyya, V.Dev, K. Patel, M. K. Mohapatra, J. Lakhani, R. Benner, C. Lele, and K.Patki. 2005. A multicenter study of azithromycin, alone and in combinationwith chloroquine, for the treatment of acute uncomplicated Plasmodiumfalciparum malaria in India. J. Infect. Dis. 191:1582–1588.

13. Foulds, G., R. M. Shepard, and R. B. Johnson. 1990. The pharmacokineticsof azithromycin in human serum and tissues. J. Antimicrob. Chemother.25(Suppl. A):73–82.

14. Garner, P., and A. M. Gulmezoglu. 2003. Drugs for preventing malaria-related illness in pregnant women and death in the newborn. CochraneDatabase Syst. Rev. CD000169. Accessed October 2009.

15. Gerhardy, C. L., and M. Garrett. 2002. Obstetrics and gynaecology fornurses and midwives, 5th ed. Lutheran School of Nursing, Madang, PapuaNew Guinea.

16. Heppner, D. G., Jr., D. S. Walsh, N. Uthaimongkol, D. B. Tang, S. Tulyayon,B. Permpanich, T. Wimonwattrawatee, N. Chuanak, A. Laoboonchai, P.Sookto, T. G. Brewer, P. McDaniel, C. Eamsila, K. Yongvanitchit, K. Uhl,D. E. Kyle, L. W. Keep, R. E. Miller, and C. Wongsrichanalai. 2005. Ran-domized, controlled, double-blind trial of daily oral azithromycin in adultsfor the prophylaxis of Plasmodium vivax malaria in Western Thailand. Am. J.Trop. Med. Hyg. 73:842–849.

17. Hodge, L. S., and T. S. Tracy. 2007. Alterations in drug disposition duringpregnancy: implications for drug therapy. Exp. Opin. Drug Metab. Toxicol.3:557–571.

18. Jonsson, E. N., and M. O. Karlsson. 1999. Xpose—an S-PLUS basedpopulation pharmacokinetic/pharmacodynamic model building aid forNONMEM. Comput. Methods Programs Biomed. 58:51–64.

19. Kain, K. C., G. D. Shanks, and J. S. Keystone. 2001. Malaria chemoprophy-laxis in the age of drug resistance. I. Currently recommended drug regimens.Clin. Infect. Dis. 33:226–234.

20. Kalilani, L., I. Mofolo, M. Chaponda, S. J. Rogerson, A. P. Alker, J. J. Kwiek,and S. R. Meshnick. 2007. A randomized controlled pilot trial of azithro-mycin or artesunate added to sulfadoxine-pyrimethamine as treatment formalaria in pregnant women. PLoS One 2:e1166.

21. Krishna, S., and N. J. White. 1996. Pharmacokinetics of quinine, chloro-quine and amodiaquine. Clinical implications. Clin. Pharmacokinet. 30:263–299.

22. Lalak, N. J., and D. L. Morris. 1993. Azithromycin clinical pharmacokinetics.Clin. Pharmacokinet. 25:370–374.

23. Liu, P., H. Allaudeen, R. Chandra, K. Phillips, A. Jungnik, J. D. Breen, andA. Sharma. 2007. Comparative pharmacokinetics of azithromycin in serumand white blood cells of healthy subjects receiving a single-dose extended-release regimen versus a 3-day immediate-release regimen. Antimicrob.Agents Chemother. 51:103–109.

24. Lucchi, M., B. Damle, A. Fang, P. J. de Caprariis, A. Mussi, S. P. Sanchez,G. Pasqualetti, and M. Del Tacca. 2008. Pharmacokinetics of azithromycin inserum, bronchial washings, alveolar macrophages and lung tissue following asingle oral dose of extended or immediate release formulations of azithro-mycin. J. Antimicrob. Chemother. 61:884–891.

25. Luke, D. R., and G. Foulds. 1997. Disposition of oral azithromycin in hu-mans. Clin. Pharmacol. Ther. 61:641–648.

26. Mazzei, T., C. Surrenti, A. Novelli, A. Crispo, S. Fallani, V. Carla, E.Surrenti, and P. Periti. 1993. Pharmacokinetics of azithromycin in pa-tients with impaired hepatic function. J. Antimicrob. Chemother.31(Suppl. E):57–63.

27. Miller, R. S., C. Wongsrichanalai, N. Buathong, P. McDaniel, D. S. Walsh,C. Knirsch, and C. Ohrt. 2006. Effective treatment of uncomplicated Plas-modium falciparum malaria with azithromycin-quinine combinations: a ran-domized, dose-ranging study. Am. J. Trop. Med. Hyg. 74:401–406.

28. Nakornchai, S., and P. Konthiang. 2006. Activity of azithromycin or eryth-

romycin in combination with antimalarial drugs against multidrug-resistantPlasmodium falciparum in vitro. Acta Trop. 100:185–191.

29. Noedl, H., S. Krudsood, K. Chalermratana, U. Silachamroon, W. Leowat-tana, N. Tangpukdee, S. Looareesuwan, R. S. Miller, M. Fukuda, K. Jong-sakul, S. Sriwichai, J. Rowan, H. Bhattacharyya, C. Ohrt, and C. Knirsch.2006. Azithromycin combination therapy with artesunate or quinine for thetreatment of uncomplicated Plasmodium falciparum malaria in adults: arandomized, phase 2 clinical trial in Thailand. Clin. Infect. Dis. 43:1264–1271.

30. Noedl, H., S. Krudsood, W. Leowattana, N. Tangpukdee, W. Thanachartwet,S. Looareesuwan, R. S. Miller, M. Fukuda, K. Jongsakul, K. Yingyuen, S.Sriwichai, C. Ohrt, and C. Knirsch. 2007. In vitro antimalarial activity ofazithromycin, artesunate, and quinine in combination and correlation withclinical outcome. Antimicrob. Agents Chemother. 51:651–656.

31. Ohrt, C., G. D. Willingmyre, P. Lee, C. Knirsch, and W. Milhous. 2002.Assessment of azithromycin in combination with other antimalarial drugsagainst Plasmodium falciparum in vitro. Antimicrob. Agents Chemother.46:2518–2524.

32. Oldfield, E. C., III, W. J. Fessel, M. W. Dunne, G. Dickinson, M. R. Wallace,W. Byrne, R. Chung, K. F. Wagner, S. F. Paparello, D. B. Craig, G. Melcher,M. Zajdowicz, R. F. Williams, J. W. Kelly, M. Zelasky, L. B. Heifets, andJ. D. Berman. 1998. Once weekly azithromycin therapy for prevention ofMycobacterium avium complex infection in patients with AIDS: a random-ized, double-blind, placebo-controlled multicenter trial. Clin. Infect. Dis.26:611–619.

33. Periti, P., T. Mazzei, E. Mini, and A. Novelli. 1993. Adverse effects ofmacrolide antibacterials. Drug Saf. 9:346–364.

34. Peters, D. H., H. A. Friedel, and D. McTavish. 1992. Azithromycin. A reviewof its antimicrobial activity, pharmacokinetic properties and clinical efficacy.Drugs 44:750–799.

35. Pfizer. 2009. Zithromax U.S. physician prescribing information. Pfizer Labs,New York, NY.

36. Ramsey, P. S., M. B. Vaules, G. M. Vasdev, W. W. Andrews, and K. D.Ramin. 2003. Maternal and transplacental pharmacokinetics of azithromy-cin. Am. J. Obstet. Gynecol. 188:714–718.

37. Ripa, S., L. Ferrante, and M. Prenna. 1996. A linear model for the phar-macokinetics of azithromycin in healthy volunteers. Chemotherapy 42:402–409.

38. Sarkar, M., C. Woodland, G. Koren, and A. R. Einarson. 2006. Pregnancyoutcome following gestational exposure to azithromycin. BMC PregnancyChildbirth 6:18.

39. Schlitzer, M. 2007. Malaria chemotherapeutics. Part I: history of antimalar-ial drug development, currently used therapeutics, and drugs in clinicaldevelopment. Chem. Med. Chem. 2:944–986.

40. Sidhu, A. B., Q. Sun, L. J. Nkrumah, M. W. Dunne, J. C. Sacchettini, andD. A. Fidock. 2007. In vitro efficacy, resistance selection, and structuralmodeling studies implicate the malarial parasite apicoplast as the target ofazithromycin. J. Biol. Chem. 282:2494–2504.

41. Taylor, W. R., T. L. Richie, D. J. Fryauff, H. Picarima, C. Ohrt, D. Tang, D.Braitman, G. S. Murphy, H. Widjaja, E. Tjitra, A. Ganjar, T. R. Jones, H.Basri, and J. Berman. 1999. Malaria prophylaxis using azithromycin: a dou-ble-blind, placebo-controlled trial in Irian Jaya, Indonesia. Clin. Infect. Dis.28:74–81.

42. Upton, R. N., and G. L. Ludbrook. 2005. Pharmacokinetic-pharmacodynamicmodelling of the cardiovascular effects of drugs—method development andapplication to magnesium in sheep. BMC Pharmacol. 5:5.

43. van Eijk, A. M., and D. J. Terlouw. 2007. Azithromycin for treating uncom-plicated malaria. Cochrane Database Syst. Rev. CD006688. Accessed Octo-ber 2009.

44. White, N. J. 2005. Intermittent presumptive treatment for malaria. PLoSMed. 2:e3.

45. Wildfeuer, A., H. Laufen, M. Leitold, and T. Zimmermann. 1993. Compar-ison of the pharmacokinetics of three-day and five-day regimens of azithro-mycin in plasma and urine. J. Antimicrob. Chemother. 31(Suppl. E):51–56.


by on Decem


.orgD

ownloaded from

ANTIMICROBIAL AGENTS AND CHEMOTHERAPY, Apr. 2011, p. 1693–1700 Vol. 55, No. 40066-4804/11/$12.00 doi:10.1128/AAC.01075-10Copyright © 2011, American Society for Microbiology. All Rights Reserved.

Pharmacokinetic Properties of Conventional and Double-DoseSulfadoxine-Pyrimethamine Given as Intermittent

Preventive Treatment in Infancy�

Sam Salman,1 Susan Griffin,2 Kay Kose,2 Nolene Pitus,2 Josephine Winmai,2 Brioni Moore,1Peter Siba,2 Kenneth F. Ilett,1,3 Ivo Mueller,2 and Timothy M. E. Davis1*

School of Medicine and Pharmacology, University of Western Australia, Perth, Western Australia, Australia1;Papua New Guinea Institute of Medical Research, Madang, Papua New Guinea2; and Clinical Pharmacology and

Toxicology Laboratory, Path West Laboratory Medicine, Nedlands, Australia3

Received 4 August 2010/Returned for modification 28 December 2010/Accepted 21 January 2011

Intermittent preventive treatment in infancy (IPTi) entails routine administration of antimalarial treatmentdoses at specified times in at-risk infants. Sulfadoxine-pyrimethamine (SDX/PYR) is a combination that hasbeen used as first-line IPTi. Because of limited pharmacokinetic data and suggestions that higher milligram/kilogram pediatric doses than recommended should be considered, we assessed SDX/PYR disposition, ran-domized to conventional (25/1.25 mg/kg of body weight) or double (50/2.5 mg/kg) dose, in 70 Papua NewGuinean children aged 2 to 13 months. Blood samples were drawn at baseline, 28 days, and three time pointsrandomly selected for each infant at 4 to 8 h or 2, 5, 7, 14, or 21 days. Plasma SDX, PYR, and N4-acetylsulfadoxine (NSX, the principal metabolite of SDX) were assayed by high-performance liquid chroma-tography (HPLC). Using population modeling incorporating hepatic maturation and cystatin C-based renalfunction, two-compartment models provided best fits for PYR and SDX/NSX plasma concentration profiles.The area under the plasma concentration-time curve from 0 h to infinity (AUC0–�) was greater with the doubledose versus the conventional dose of PYR (4,915 versus 2,844 �g/day/liter) and SDX (2,434 versus 1,460mg/day/liter). There was a 32% reduction in SDX relative bioavailability with the double dose but no evidenceof dose-dependent metabolism. Terminal elimination half-lives (15.6 days for PYR, 9.1 days for SDX) werelonger than previously reported. Both doses were well tolerated without changes in hemoglobin or hepatorenalfunction. Five children in the conventional and three in the double-dose group developed malaria duringfollow-up. These data support the potential use of double-dose SDX/PYR in infancy, but further studies shouldexamine the influence of hepatorenal maturation in very young infants.

Intermittent preventive treatment in infancy (IPTi) is a strat-egy in which infants in areas in which malaria is endemic aregiven treatment doses of antimalarial drugs at specified times,regardless of clinical and parasitologic status. Because of itsavailability, tolerability, and relatively low cost, sulfadoxine-pyrimethamine (SDX/PYR) has been used as a first-line treat-ment in IPTi programs, especially in Africa. A recent review ofsafety and efficacy data from six trials conducted from 1999 to2007 revealed that, despite the emergence of molecular mark-ers of parasite resistance, SDX/PYR IPTi reduced clinical ma-laria and malaria-related hospital admissions by about one-third and reduced anemia in the first year of life by 15% (23).The duration of effective antimalarial prophylaxis after a doseof SDX/PYR is 4 to 6 weeks (9, 17).

There is evidence that the efficacy of SDX/PYR IPTi is dosedependent. When given as a fixed dose (27), efficacy declineswith age as lower doses (milligrams/kilogram of body weight)are taken (9). In addition, studies of older children aged 2 to 5years with falciparum malaria have found higher clearancerates and larger apparent volumes of distribution for both SDX

and PYR than those in adults (11). Consistent with these data,a population pharmacokinetic (PK) study in children with con-genital toxoplasmosis showed that the elimination half-lives forboth drugs were directly related to body weight, with the con-sequence that younger and thus lighter children had morerapid elimination (37). These studies suggest that the peakplasma concentration and area under the plasma concentra-tion-time curve (AUC) will be reduced in younger children andthat currently recommended doses of SDX/PYR of 25 mg/kgand 1.25 mg/kg, respectively, may be inadequate for full effi-cacy. Indeed, there is evidence that higher blood PYR concen-trations enhance the ability of pediatric patients to clear resis-tant Plasmodium falciparum (19).

In view of these data and calls for doubling of the recom-mended treatment dose in children aged 2 to 5 years (11), weassessed the tolerability, safety, and pharmacokinetic proper-ties of SDX/PYR given in recommended and double recom-mended doses to infants living an area of intense malariatransmission in Papua New Guinea (PNG).


Study site, sample, and approvals. The present study was conducted at Alex-ishafen Health Centre, Madang Province, on the north coast of Papua NewGuinea (PNG). Infants between the ages of 2 and 13 months from the surround-ing area were eligible for recruitment provided that they (i) did not have featuresof severe malaria or significant nonmalarial illness, (ii) had not been treated withSDX or PYR in the previous 4 weeks, (iii) did not have a known allergy to either

* Corresponding author. Mailing address: Department of Medicine,Fremantle Hospital, P.O. Box 480, Fremantle 6959, Western Australia,Australia. Phone: (618) 9431 3229. Fax: (618) 9431 2977. E-mail:[email protected].

� Published ahead of print on 31 January 2011.

1693

SDX or PYR, and (iv) were available for assessment for the duration of follow-up. Written informed consent was obtained from the parents/guardians of allrecruited infants. The study was approved by the Medical Research AdvisoryCommittee of PNG and the Human Ethics Research Committee at the Univer-sity of Western Australia.

Clinical procedures. At enrollment, a clinical assessment was performed thatincluded a standard baseline symptom questionnaire completed by parents/guardians. A 500-�l finger prick capillary blood sample was taken for preparationof blood smears for microscopy, baseline drug assay, biochemical tests, andhemoglobin concentration (HemoCue, Angelholm, Sweden). Subjects were ran-domized to receive either the recommended dose of SDX/PYR (25/1.25 mg/kgFansidar; Roche, Basel, Switzerland) or a double dose (50/2.5 mg/kg). Table 1shows the dose administered based on body weight. All dosing was directlyobserved, with subsequent monitoring and readministration of the dose if theinfant vomited within 30 min. Infants with a positive blood film were also givena 3-day course of amodiaquine according to PNG national treatment guidelines(33). All drugs were crushed and mixed with either water or breast milk beforeadministration by mouth using a syringe.

All infants were reassessed on days 1, 2, 3, 5, 7, 14, 21, and 28. A hemoglobinconcentration was determined on each occasion and a repeat symptom ques-tionnaire administered at each visit up to day 7. Blood films were repeated on day28 and/or when fever or a recent history of a fever was reported. For pharma-cokinetic analysis, four additional 500-�l capillary blood samples were takenfrom each infant. The times for the first three of these were randomly selectedfor each infant from either 4 to 8 h or 2, 5, 7, 14, or 21 days postdose. A finalsample was taken in all cases on day 28. The exact timing of each blood samplewas recorded. All samples were centrifuged promptly, with red cells and sepa-rated plasma stored frozen at �80°C until assay.

Laboratory methods. Giemsa-stained thick blood smears were examined in-dependently by at least two skilled microscopists who were blinded to dosegroup. Each microscopist viewed �100 fields at �1,000 magnification before aslide was considered negative. Any slide discrepant for positivity/negativity oridentification to the species level was referred to a third microscopist.

Cystatin C (CysC) concentrations were measured by particle-enhanced immu-noturbidimetry (PETIA) using the Tina-quant cystatin C kit run on an Elecsys2010 analyzer (Roche, Indianapolis, IN). Sodium, urea, creatinine, albumin,�-glutamyl transferase, and bilirubin were measured using an Integra 800 ana-lyzer (Roche) when sufficient plasma was available.

Sulfadoxine, sulfamethazine, and pyrimethamine were obtained from Sigma-Aldrich (Castle Hill, Australia), and midazolam hydrochloride was obtainedfrom Pfizer (West Ryde, Australia). N4-acetylsulfadoxine (NSX) was synthesizedaccording to the method of Whelpton et al. (39) and found to have a meltingpoint of 230°C and �99.9% purity by high-performance liquid chromatography(HPLC). Acetonitrile was obtained from Merck (Darmstadt, Germany). Allother chemicals were of analytical or HPLC grade.

For PYR, SDX, and NSX, extraction and separation were performed based onpreviously published HPLC-UV methods (26, 37). The internal standards weremidazolam HCl for PYR and sulfamethazine for SDX and NSX. Analytes wereassayed using UV detection at 270 nm. Chemstation software (version 9; AgilentTechnology, Waldbronn, Germany) was used for analysis of chromatograms.Standard curves were linear from 5 to 1,000 �g/liter, 0.1 to 200 mg/liter, and 0.02to 10 mg/liter for PYR, SDX, and NSX, respectively. Intra- and interday relativestandard deviations (RSDs) were �15% for all analytes at all concentrations.The limits of quantification (LOQ) were 2.5 �g/liter, 0.1 mg/liter, and 0.02mg/liter, and the limits of detection (LOD; determined as a signal-to-noise ratioof 5) were 1 �g/liter, 0.05 mg/liter, and 0.01 mg/liter for PYR, SDX, and NSX,respectively.

Population pharmacokinetic analysis. Loge concentration-versus-time datasets for PYR, SDX, and NSX were analyzed by nonlinear mixed effect modelingusing NONMEM (version 6.2.0; ICON Development Solutions, Ellicott City,MD) with an Intel Visual FORTRAN 10.0 compiler. Linear mammillary modelsubroutines within NONMEM (ADVAN2/TRANS2 and ADVAN4/TRANS4),first order conditional estimation (FOCE) with �-ε interaction, and the objective

function value (OFV) were used to construct and compare plausible models.Unless otherwise specified, a difference in OFV of �6.63 (�2 distribution with 1df, P � 0.01) was considered significant. Due to the small number of samples withlow concentrations, those below the LOD were not included in the analysis, whilelevels between the LOD and LOQ were kept at their measured concentrations.

As the subjects were infants with a range of ages, it was important to incorporatematuration of clearance into the model. Therefore, total clearance (CLT) was de-fined as the sum of hepatic clearance (CLH) and renal clearance (CLR), i.e., CLT �

CLH � CLR. The age-adjusted hepatic clearance, CLH, was determined using asigmoid maximum effect (Emax) model (7) as TVCLH � [PMAHillCL/(PMAHillCL �

MATCL50HillCL)], where TVCLH is the population average value for hepatic clear-

ance, PMA is the postmenstrual age (the age of the infant recorded from the lastmenstrual cycle of the mother during pregnancy rather than birth), HillCL is the Hillcoefficient for hepatic clearance, and MATCL50 is the PMA at which CLH is 50% ofthe mature value. When an accurate PMA could not be obtained, it was estimatedfrom the postnatal age (PNA) and average gestation in PNG (3, 15, 21). CLR wasadjusted to a standardized value for an estimated glomerular filtration rate (eGFR)of 120 ml/min/1.76 m2, i.e., TVCLR � (eGFR/120), where CLR is the adjusted renalclearance, TVCLR is the population average value for renal clearance, and theeGFR was determined from the cystatin C concentration (CysC) as 91.62 � (1/CysC1.123) (20).

Allometric scaling using weight (WT) was also used on all volume and clear-ance terms, which were multiplied by (WT/70) and (WT/70)0.75, respectively.One- and two-compartment models with first order absorption without lag timewere assessed for both SDX and PYR. As few data exist to describe the absorp-tion phase of both drugs, the absorption rate constant (ka) was fixed to thepreviously published value for infants (18). Between-subject variability (BSV)was added to parameters for which it could be estimated reasonably from avail-able data. As loge concentration data were used, an additive model (representingproportional error) was used for residual unexplained variability (RUV).

In the development of the final models, we investigated the influence of thecovariates dosing group, relative dose (milligram/kilogram), PMA, malaria sta-tus, concomitant treatment with amodiaquine, and initial hemoglobin concen-tration using the generalized additive modeling procedure within Xpose (http://xpose.sourceforge.net) and by inspection of correlation plots. Covariaterelationships identified by this procedure were evaluated within the NONMEMmodel, and inclusion of the covariate required a significant decrease in OFVaccompanied by a decrease in the BSV of that parameter. Correlations amongBSV terms and weighted residuals (WRES) plots were also used in modelevaluation.

Once a final model for SDX was obtained, the parameter estimates were fixedand an additional compartment was added in order to model NSX concentra-tions. In order to allow identifiability in the model, the percentage conversion ofSDX to NSX was fixed to 60% based on the product information (35). Theelimination of NSX was assumed to be entirely renal (25).The influence of thecovariates was assessed on new model parameters using the method describedabove.

A bootstrap procedure using Perl-speaks-NONMEM (PSN) (http://psn.sourceforge.net) and the resulting parameters were then summarized as medianand 2.5th and 97.5th percentiles (95% empirical confidence interval [CI]) tofacilitate validation of the final model parameter estimates. In addition, stratifiedvisual predictive checks (VPCs) and numerical predictive checks (NPCs) werealso performed using PSN with 1,000 replicate data sets simulated from theoriginal data set. NPCs stratified according to PMA were assessed by comparingthe actual with the expected number of data points within the 20, 40, 60, 80, 90,and 95% prediction intervals (PI). The resulting 80% PI for drug concentrationswere plotted with the observed data to assess the predictive performance of themodel.

Statistical analysis. As previously reported in a study of SDX/PYR pharma-cokinetics in pregnant versus nonpregnant women (26), and using estimates ofcentrality and variance for pharmacokinetic parameters from previous pediatricstudies (11, 19, 32, 37, 40) and an assumed 20% attrition rate, a sample size of35 in each group in the present study would be expected to show a �30%increase in the magnitude of any pharmacokinetic parameter in the double-dosegroup at � � 0.05 and � � 0.1. SPSS 17.0 (SPSS inc. Chicago, IL) was used forall statistical analysis unless otherwise specified. Data are summarized as mean �

standard deviation (SD) or median and interquartile range (IQR) as appropri-ate. Student’s t test or the Mann-Whitney U test was used for two-samplecomparisons. Categorical data were compared using either Pearson chi-squaredor Fisher’s exact test and multiple means by repeated measures analysis ofvariance (ANOVA). A two-tailed level of significance of 0.05 was used.

TABLE 1. Dosing guide for conventional and double-dose groups

Body wtDosage (mg SDX/PYR)

Conventional dose Double dose

3–5.9 kg 1⁄4 tablet (125/6.25) 1⁄2 tablet (250/12.5)6–11.9 kg 1⁄2 tablet (250/12.5) 1 tablet (500/25)


RESULTS

Patient characteristics. Seventy infants were enrolled be-tween April 2008 and December 2008, with equal numbers ineach dose group. Baseline subject characteristics are summa-rized in Table 2. The double-dose group received a signifi-cantly higher milligram/kilogram dose than the conventionaldose group (P � 0.001) and was taller by a mean of 4.3 cm (P �0.015). The double-dose group was also older (by a mean of 47days) and heavier (by 0.4 kg) than the conventional dose group,but these differences were not statistically significant (P �0.05).

Tolerability, safety, and efficacy. Both doses were well tol-erated. There were no changes in symptoms in either groupcompared to predose profiles, including an absence of derma-tological conditions. There were no significant changes in he-moglobin, or in plasma urea, creatinine, or CysC, over time. Inthe conventional dose group, there was a significant but tran-sient mean fall in plasma albumin of 2 g/liter at day 2 (from 38to 36 g/liter; P � 0.01), but there were no concomitant in-creases in plasma bilirubin or hepatic enzymes in either group.

Five infants with vivax malaria and one infant with a mixedPlasmodium vivax/P. falciparum infection at enrollment re-sponded to treatment. Three other infants in the conventionaldose group and two in the double-dose group were adminis-tered antimalarial drugs during follow-up at an external healthcare facility, and no blood smears were available for review. Noother subjects became symptomatic during the study. Only twoinfants in the conventional dosing group and one in the dou-ble-dosing group who were aparasitemic at entry had a positiveblood slide on day 28 (all for P. vivax). All were asymptomatic,and each was treated according to PNG national treatmentguidelines.

Pharmacokinetic modeling. There were 248, 255, and 247drug concentration measurements available for pharmacoki-netic modeling for PYR, SDX, and NSX, respectively. Therewere four samples with PYR concentrations between the LODand LOQ and a further four with concentrations below theLOD for PYR. In addition, seven samples were of insufficient

volume for measurement of PYR after the SDX/NSX assay.There were no SDX or NSX concentrations below the LOQ,but NSX concentrations could not be determined in eightsamples due to an unidentified interfering peak. For PYR, atwo-compartment model was superior to a one-compartmentmodel with a lower OFV (�87.081 versus �30.030) and aless-biased weighted residuals versus time (WRES) plot. Themodel parameters were ka, CLH/F, CLR/F, central compart-ment volume of distribution (V2/F), peripheral compartmentvolume of distribution (V3/F), intercompartmental clearance(Q/F), HillCL, and MATCL50. BSV was estimable on CLT/F,V2/F, and Q/F. As the correlation between the variability ofV2/F and Q/F was very close to 1, it was subsequently fixed tounity to assist with successful determination of the covariancematrix. None of the covariates tested improved the modelsignificantly; therefore, the final model contained only the ef-fects of PMA and WT anticipated from maturation and allo-metric scaling, respectively.

The final parameter estimates and the results of the boot-strap procedure for PYR are shown in Table 3. All modelparameters had a bias of �11%. Goodness-of-fit plots for PYRare shown in Fig. 1. NPCs of the data showed good predictiveperformance, as did VPC plots of the observed drug concen-trations and their 80% PI (the 10th and 90th percentile bound-aries) stratified by dosing group (Fig. 2A and B). Post hocparameter estimates are shown in Table 4. There was no dif-ference between the two groups for any of these parametersexcept for AUC from 0 h to infinity (AUC0–�), which wassignificantly higher in the double-dose group (4,915 versus2,844 �g/day/liter). Median steady-state volume (VSS) for the

TABLE 2. Baseline characteristics of study participants

Parameter

Result for study group

Conventionaldose (n � 35)

Double dose(n � 35)

Postmenstrual age, days �median (IQR)� 454 (383–513) 501 (428–532)Sex �no. (%) male� 22 (63) 24 (69)Weight (kg) (mean � SD) 6.58 � 1.31 6.98 � 1.1Height (cm) (mean � SD) 61.8 � 6.5 66.1 � 7.8Axillary temp (°C) (mean � SD) 36.5 � 0.6 36.4 � 0.6

No. (%) with parasitemiaa

P. falciparum 1 (3) 0 (0)P. vivax 3 (9) 3 (9)

Respiratory rate (per min) (mean � SD) 40 � 11 42 � 11Supine pulse rate (per min) (mean � SD) 133 � 14 133 � 15Mean upper arm circumference (cm)

(mean � SD)13.2 � 3.5 13.7 � 2.6

Hemoglobin (g/liter) (mean � SD) 9.5 � 1.3 9.5 � 1.2eGFR (ml/min/1.73 m2) (mean � SD) 80 � 20 84 � 16Sulfadoxine dose (mg/kg) (mean � SD)b 35.6 � 5.6 67.1 � 12.6Pyrimethamine dose (mg/kg) (mean � SD)b 1.8 � 0.3 3.4 � 0.6

a One infant had a mixed P. vivax/falciparum infection.b P � 0.001.

TABLE 3. Final population PK parameters andbootstrap results for PYR

Parametera

Value

Final modelBootstrap

(n � 1,000)�median (95% CI)�

OFV �97.384 �111.812 (�170.137 to �62.348)

PK parameters �estimate(% RSE)�

ka (per h) 0.779 FixedVC/F (liters/70 kg) 222 (4) 221 (202–242)VP/F (liters/70 kg) 64.1 (24) 63.0 (41.8–128.5)Q/F (liters/h/70 kg) 0.0735 (19) 0.0788 (0.0486–0.1470)CLR/F (liters/h/70 kg) 0.416 (64) 0.3820 (0.0621–0.9868)CLH/F (liters/h/70 kg) 0.854 (24) 0.878 (0.466–1.220)MATCL50 (days) 318 (8) 326 (286–367)HillCL 7.39 (43) 7.80 (3.53–35.18)

Random parameters�% CV(% RSE)�

BSV VC/F 13.0 (36) 13.6 (3.6–24.7)BSV CLT/F 27.8 (13) 27.0 (18.2–35.0)BSV Q/F 34.1 (32) 36.4 (17.4–53.4)

Correlations betweenBSV pairs

R (VC/F, CLT/F) 0.533 (69) 0.563 (�0.059 to 0.826)R (VC/F, Q/F) 1 FixedR (CLT/F, Q/F) 0.533 (69) 0.563 (�0.059 to 0.826)

Residual unexplainedvariability (RUV)

Proportional error�% CV (% RSE)�

33.6 (23) 32.6 (26.9–37.4)

a % RSE, percent relative standard error.

VOL. 55, 2011 SULFADOXINE-PYRIMETHAMINE PHARMACOKINETICS IN INFANCY 1695

combined study sample was 27.8 liters, and the half-lives at �and � phases (t1/2� and t1/2�, respectively) were 72.7 and 374 h,respectively.

Initial modeling of SDX revealed that a one-compartmentmodel was appropriate, as there was minimal bias in the WRES

plot that was not improved when a two-compartment model wasfitted. The model parameters were ka, CLH/F, CLR/F, volume ofdistribution (V/F), HillCL, and MATCL50. BSV was able to beestimated on CLT/F and V/F. There was a significant relationshipbetween relative dose (in milligrams/kilogram) and relative bio-availability which conformed to a power function; specifically,individual relative bioavailability � 1 � ([individual relativedose]/[average relative dose]effect parameter). The value of thepower effect parameter was �0.56, indicating that, when the doseis doubled, the bioavailability falls by 32.2%. The final parameterestimates and the results of the bootstrap procedure are shown inTable 5. With the exception of CLR, all parameter estimates hadbiases of �13%. The median bootstrap value for CLR was almostdouble the initial estimate (195%), demonstrating the difficulty indelineating the difference between and estimating the hepatic andrenal clearance using this methodology. Goodness-of-fit plots forSDX are shown in Fig. 3. NPCs of the data showed good predic-tive performance, as did VPC plots of the observed drug concen-trations and their 80% PI stratified by dose group in Fig. 4.

An additional compartment was added to the final SDX PKmodel to incorporate the data for NSX. This resulted in threeadditional model parameters: volume of distribution of NSX(VNSX/F), clearance of NSX (CLNSX/F), and percentage oftotal SDX elimination representing conversion of SDX to NSX(%NSX). As these three parameters cannot be estimated si-multaneously, %NSX was fixed to 60% based on publisheddata (35). The estimates of VNSX/F and CLNSX/F are directly

FIG. 1. Goodness-of-fit plots for PYR showing observed versusmodel predicted concentrations (A) and individual predicted concen-trations (B) (both log scale) and conditional weighted residuals versustime (C). For panels A and B, the solid gray line represents the line ofidentity, while the dashed black line represents the linear regressionline of best fit; in panel C, the solid gray line represents the locallyweighted scatterplot smoothing (LOESS) smoothed fit.

FIG. 2. Visual predicted check plots for PYR showing simulated10th (short dashed line), 50th (dotted line), and 90th (solid line)percentile concentrations and observed concentration (log scale) data(gray open circles) versus time (log scale) for conventional dose(A) and double-dose (B) participants.


related to %NSX; therefore, the value of these parametersshould be interpreted with caution. However, AUC and t1/2 forNSX remain unchanged for different values of %NSX. VNSX/Fand CLNSX/F were not influenced by any of the available co-variates. Final parameter estimates and results of the bootstrapprocedure are shown in Table 5. Bias was �5% for all NSXparameters, and NPCs and VPCs were performed on the NSXdata set and indicated good predictive performance of themodel (data not shown).

There were some significant differences between conven-tional and double-dose groups in the post hoc parameter esti-mates for both SDX and NSX (Table 6). These included ex-pected differences in the AUC0–� for both SDX and NSX, but

also differences in the half-life and clearance for both drugswhich were not revealed by the model covariate building stage.A higher clearance and lower half-life (t1/2) in the double-dosegroup can be attributed to organ maturation, as these infantswere older than those in the conventional dose group. Themedian t1/2 of NSX for the combined study sample was shorterthan that of SDX (8.9 versus 218 h). The percentage of theAUC0–� of NSX compared to that for SDX was the same forboth dose groups (approximately 5%).

Sigmoid Emax curves of hepatic maturity for SDX and PYRby PMA are shown in Fig. 5. They are closely related toMATCL50 values of 318 days and 271 days for PYR and SDX,respectively. Of the 70 infants, 48 (69%) and 38 (54%) had anestimated hepatic clearance that was �90% of adult values forPYR and SDX, respectively.

DISCUSSION

The present study is the first to investigate the pharmacoki-netics of SDX/PYR in infants living in a setting in whichmalaria is endemic and in which IPTi is appropriate. It is alsothe first to investigate the possibility that a higher dose thanconventionally recommended should be given to achieve ther-apeutic plasma concentrations in this age group, as has beenrecommended for children aged 2 to 5 years (11). SDX/PYRwas well tolerated by all infants, and there was no evidence ofhepatorenal or bone marrow toxicity even at the higher dose.The AUC0–� of both SDX and PYR was significantly higher inthe double-dose group. However, there was a 32% reductionin the relative bioavailability of SDX when the dose was dou-bled, possibly due to saturation of absorption. The percentageof NSX to SDX exposure (AUC) was the same in both groups,suggesting that a double dose does not affect the metabolicclearance of SDX. The pharmacokinetic properties of PYRwere not dose dependent in the present study.

The pharmacokinetic parameters for PYR observed in ourchildren are different from those observed in other pediatricstudies (11, 18, 31, 37, 40). We found a longer t1/2� (15.6 versus2.67 to 4.46 days) and a higher conventional dose AUC (2,844versus 1,052 to 2,607 �g/day/liter). This may reflect the factthat most of our children were well. In addition, we employeda relatively long duration of sampling that facilitated identifi-cation of biexponential elimination, a profile reported previ-ously in studies of adults (26, 28, 38) but not children. Whileone pediatric study sampled out to 42 days, the drug could notbe quantified in 40% of the samples (11). Although the mean

TABLE 4. Post hoc Bayesian predicted PK parameters for PYR for PNG infants given conventional and double doses of SDX/PYR

ParameterMedian result (IQR) for study group

P valuea

Conventional dose (n � 35) Double dose (n � 35)

CLT/F (liters/h) 0.183 (0.13–0.21) 0.199 (0.164–0.229) NSVC/F (liters) 20.2 (17.8–24.1) 22.1 (19–25) NSVP/F (liters) 6.18 (5.32–6.81) 6.49 (5.83–7.03) NSVSS/F (liters) 25.781 (23.319–30.957) 28.8 (24.9–31.8) NSQ/F (liters/h) 0.0118 (0.0073–0.0165) 0.0135 (0.009–0.0196) NSt1/2� (h) 73.7 (67.1–87.7) 70.7 (62.3–82.3) NSt1/2� (h) 391 (300–565) 361 (272–511) NSAUC0–� PYR (�g/day/liter) 2,844 (2,486–3,571) 4,915 (4,311–5,681) �0.001

a Mann-Whitney test. NS, nonsignificant (P � 0.05).

TABLE 5. Final population PK parameters and bootstrap resultsfor SDX and NSXa

Parameter

Value

Final model Bootstrap (n � 1,000)�median (95% CI)�

OFV �521.177 �529.222 (�647.701 to �428.084)

PK parameters �estimate(% RSE)�

ka (per h) 1.23 FixedV/F (liters/70 kg) 24.2 (4) 24.2 (22.5–26.1)CLR/F (liters/h/70 kg) 0.0046 (113) 0.0086 (0.0005–0.0267)CLH/F (liters/h/70 kg) 0.0458 (16) 0.0427 (0.0290–0.0640)MATCL50 (days) 271 (8) 286 (248–360)HillCL 4.07 (52) 4.61 (1.56–15.54)Relative dose on relative

bioavailability (power)�0.56 (14) �0.54 (�0.71 to �0.38)

% NSX (%) 60 FixedVNSX/F (liters/70 kg) 11.7 (10.7) 11.7 (9.4–14.4)CLNSX/F (liters/h/70 kg) 0.758 (5) 0.756 (0.690–0.838)

Random parameters�CV% (% RSE)�

BSV V/F 23.0 (11) 22.2 (17.1–26.5)BSV CLT/F 23.8 (11) 23.4 (17.9–28.3)BSV VNSX/F 42.8 (19) 41.7 (21.0–56.3)BSV CLNSX/F 36.2 (26) 35.6 (25.5–45.1)

Correlations betweenBSV pairs

R (V/F, CLT/F) 0.644 (26) 0.653 (0.439–0.814)R (VNSX/F, CLNSX/F) 0.218 (126) 0.226 (�0.474 to 0.729)

Residual unexplainedvariability�CV% (% RSE)�

Proportional error, SDX 16.5 (11) 16.4 (12.9–20.1)Proportional error, NSX 37.1 (9) 37.0 (30.2–43.1)

a Parameters for NSX modeling obtained after fixing model parameters forSDX are highlighted in bold. % RSE, percent relative standard error.


conventional dose PYR AUC in the present study was in therange of previously reported values in adults (1,602 to 3,166�g/day/liter) (11, 16, 22, 28), the latter data may have beenunderestimates because of truncated sampling and/or use of arelatively insensitive assay. In a study of nonpregnant PNGwomen using a sampling profile, assay, and pharmacokinetic

modeling techniques that were similar to those of the presentstudy (26), the mean conventional dose PYR AUC (4,419�g/day/liter) was similar to that in the present double-dosegroup. Together with the available tolerability and safety datafrom the present study, these considerations suggest that dou-ble-dose PYR is appropriate as part of SDX/PYR IPTi.

We found that SDX also had a longer mean elimination t1/2

(9.1 versus 4.1 to 8.6 days) and a higher conventional dosemean AUC (1,460 versus 460 to 932 mg/day/liter) than those ofchildren in other studies (11, 19, 32, 37, 40). However, themean AUC was within the range found in adults (508 to 2,757mg/day/liter) (11, 16, 22, 28), including nonpregnant women(1,386 mg/day/liter) from the same location as the presentstudy (26). Although the difference in AUC compared to otherpediatric populations may be explained, as with PYR, by ourability to detect drug concentrations for longer time postdosethan in previous studies as well as by the relative health of oursubjects, only a few studies have included infants aged �1 year,and these formed a minority of the patients recruited. As oursample includes only children �13 months of age, a limitedmaturation of elimination processes is likely to play a role inthe longer t1/2 and higher AUC observed for both drugs evenin the conventional dose group. Indeed, we found evidence ofa slower maturation of these processes for SDX than PYR.

In the present study, we used plasma CysC rather thancreatinine to estimate GFR. The conventional Schwartz crea-

FIG. 3. Goodness-of-fit plots for SDX showing observed versusmodel predicted concentrations (A) and individual predicted concen-trations (B) (both log scale) and conditional weighted residuals versustime (C). For panels A and B, the solid gray line represents the line ofidentity, while the dashed black line represents the linear regressionline of best fit; in panel C, the solid gray line represents the LOESSsmoothed fit.

FIG. 4. Visual predicted check plots for SDX showing simulated10th (short dashed line), 50th (dotted line), and 90th (solid line)percentile concentrations and observed concentration (log scale) data(gray open circles) versus time (log scale) for conventional dose(A) and double-dose (B) participants.


tinine-based formula relies upon estimates of body composi-tion (36), whereas CysC-based formulae do not (5), making theestimates more robust. We used the formula derived by Fillerand Lepage (20), as it was derived from a large pediatricsample, and the same PETIA CysC assay used in the presentstudy. CysC concentrations generated by other assays such asparticle-enhanced immunonephelometry may differ from thosefrom PETIA (5). The Filler and Lepage formula is comparableto others based on CysC derived from children (13, 14, 24, 41).

Since hepatic maturation would still be occurring within theage range of our subjects, it was appropriate to include thisphenomenon in our model (1, 8, 12). We used a sigmoid Emax

approach as this has been used previously with a number ofother drugs (2, 4, 6, 8, 34) and our estimates of MATCL50,namely, 315 and 271 days for PYR and SDX, respectively, fellin the range reported in these studies (270 to 380 days). Theestimate of the Hill coefficient for SDX was also consistent(4.07 versus 2.78 to 4.6), but the Hill coefficient for PYR washigher than that previously reported (7.39). Although ourstudy age range captured the process of maturation, most ofour infants had clearances that were �90% of adult values andvery few were �50% (Fig. 5). This limits our ability to char-acterize coefficients of maturation which are likely to be inap-

propriate outside this age range. For example, adult estimatesof t1/2 for SDX (333 h) and PYR (t1/2�, 113 h; t1/2�, 647 h)based on the modeling presented here are higher than thosepreviously reported (11, 16, 22, 26, 28, 38). Future studies ofthis type should include a larger range of ages so that thematuration process from birth to adult activity levels can bedetermined more accurately.

Other studies have provided data relevant to the question ofwhether a higher SDX/PYR dose should be given to infants. Apharmacokinetic evaluation of SDX in children aged 6 months to5 years with malaria found that those aged �24 months had alower AUC0-336 h than their older counterparts (12,500 versus16,900 mg/h/liter) (32). However, all children �24 months of agereceived half the dose of older children regardless of body weightand no average dose by body weight was reported, thus compli-cating interpretation of the data. In a similar study (11), an age-stratified noncompartmental analysis of AUC0–� showed that 1-to 2-year-olds had sufficient drug exposure while children aged 2to 5 years required a double dose. The study had only 11 childrenwithin the 1- to 2-year-old age range, and because only wholetablets were given, the mean dose in this group was almost twicethat of �12-year-olds (50/2.5 versus 27.3/1.36 mg/kg). In a popu-lation-based pharmacokinetic analysis of SDX/PYR in childrenwith congenital toxoplasmosis aged 1 week to 14 years (37), light-er-weight children had a shorter t1/2 and therefore a lower drugexposure. This conclusion was based on the use of allometry, sinceage-based maturation contributed little to the model, perhapsbecause of the small numbers in the younger age groups. Inter-preted within their limitations, these various studies also provideevidence that higher milligram/kilogram SDX/PYR doses are re-quired in younger children, including those �1 year of age.

Relatively recent data from the study area indicate that amo-diaquine-SDX/PYR treatment (until recently the recommendedfirst-line antimalarial therapy for young PNG children) is associ-ated with close to a 90% 28-day adequate clinical and parasito-logic response for both falciparum (PCR-corrected) and vivaxmalaria (29). This is a suboptimal response but still suggests thateither conventional or double-dose SDX/PYR treatment in thepresent study is likely to have contributed to the relatively smallnumber of infections detected during follow-up. Although thepresent study was not designed to assess relative efficacy, espe-cially since interpretation of emergent vivax infections remainsproblematic (10) and given that only one dose was administeredrather than the several scheduled during IPTi, fewer childrenwere treated for symptomatic malaria during follow-up or were

TABLE 6. Post hoc Bayesian predicted PK parameters for SDX and NSX in PNG infants given conventional and double dosing of SDX/PYR

ParameterResult �median (IQR)� for study group

P valuea

Conventional dosing (n � 35) Double dosing (n � 35)

CLT,SDX/F (liters/h) 0.0068 (0.0057–0.0087) 0.0072 (0.0068–0.0105) 0.032VSDX/F (liters) 2.20 (1.95–2.53) 2.23 (1.97–2.64) NSt1/2 SDX (h) 232 (203–252) 207 (179–232) 0.006AUC0–� SDX (mg/day/liters) 1,460 (1,167–1,707) 2,434 (1,881–2,987) �0.001CLNSX/F (liters/h) 0.081 (0.060–0.094) 0.101 (0.081–0.116) 0.012VNSX/F (liters) 1.14 (1.00–1.28) 1.17 (0.959–1.30) NSt1/2 NSX (h) 10.3 (7.86–12.2) 8.69 (6.84–10.6) 0.027AUC0–� NSX (mg/day/liter) 1,796 (1,397–2,154) 2,890 (2,482–3,609) �0.001AUC0–� NSX/AUC0–� SDX (%) 5.0 (4.3–6.5) 5.0 (4.3–6.0) NS

a Mann-Whitney test. NS, nonsignificant (P � 0.05).

FIG. 5. Maturation as a fraction of adult clearance for PYR(dashed line) and SDX (solid line) predicted from the PK modelplotted against PMA. A box plot of the PMA in the recruited subjectsis included to show its distribution in relation to maturation of clear-ance. d, days.


slide positive on day 28 in the double-dose group. Indeed, there isevidence from epidemiologic studies utilizing fixed-dose regimens(9, 27) that appropriate milligram/kilogram doses of SDX/PYRshould be used in IPTi programs to ensure adequate levels ofprevention, especially for symptomatic compared to asymptom-atic falciparum malaria (30).

In light of this dose dependency, the fact that no study hasshown �60% protective efficacy during the first year of life (9,23), evidence that higher blood PYR concentrations facilitateparasite clearance in pediatric falciparum malaria (19), and thefact that double-dose SDX/PYR in our subjects was safe, welltolerated, and associated with higher exposure to both drug com-ponents (especially SDX), the present data argue for the potentialuse of double-dose SDX/PYR in infancy. As in recent adult stud-ies of PYR disposition (26), we found that the mean eliminationt1/2 values of PYR and SDX were larger than previously reported,a factor that may contribute to the duration of effective prophy-laxis. Although allometric considerations (shorter half-lives insmaller subjects) may justify higher SDX/PYR dosing in infants,we recommend that consideration must be given to the matura-tion of hepatorenal elimination processes and the possibility thatincreased doses may be inappropriate in very young infants.

ACKNOWLEDGMENTS

We are most grateful to Valsi Kurian and the staff of AlexishafenHealth Centre for their kind cooperation during the study. We alsothank Christine Kalopo and Bernard (“Ben”) Maamu for clinicaland/or logistic assistance. We note with deep regret that Servina Go-morrai, who assisted with patient recruitment and data collection,passed away during the study.

The study was funded by a grant from the IPTi Consortium andutilized facilities developed with support from the National Health andMedical Research Council (NHMRC) of Australia (grant 458555).T.M.E.D. is the recipient of an NHMRC Practitioner Fellowship.

REFERENCES

1. Alcorn, J., and P. J. McNamara. 2002. Ontogeny of hepatic and renalsystemic clearance pathways in infants: part I. Clin. Pharmacokinet. 41:959–998.

2. Allegaert, K., J. de Hoon, R. Verbesselt, G. Naulaers, and I. Murat. 2007.Maturational pharmacokinetics of single intravenous bolus of propofol. Pae-diatr. Anaesth. 17:1028–1034.

3. Allen, S. J., A. Raiko, A. O’Donnell, N. D. Alexander, and J. B. Clegg. 1998.Causes of preterm delivery and intrauterine growth retardation in a malariaendemic region of Papua New Guinea. Arch. Dis. Child. Fetal Neonatal Ed.79:F135–F140.

4. Anand, K. J., et al. 2008. Morphine pharmacokinetics and pharmacodynam-ics in preterm and term neonates: secondary results from the NEOPAINtrial. Br. J. Anaesth. 101:680–689.

5. Andersen, T. B., A. Eskild-Jensen, J. Frokiaer, and J. Brochner-Mortensen.2009. Measuring glomerular filtration rate in children; can cystatin C replaceestablished methods? A review. Pediatr. Nephrol. 24:929–941.

6. Anderson, B. J., K. Allegaert, J. N. Van den Anker, V. Cossey, and N. H.Holford. 2007. Vancomycin pharmacokinetics in preterm neonates and theprediction of adult clearance. Br. J. Clin. Pharmacol. 63:75–84.

7. Anderson, B. J., and N. H. Holford. 2008. Mechanism-based concepts of sizeand maturity in pharmacokinetics. Annu. Rev. Pharmacol. Toxicol. 48:303–332.

8. Anderson, B. J., and N. H. Holford. 2009. Mechanistic basis of using bodysize and maturation to predict clearance in humans. Drug Metab. Pharma-cokinet. 24:25–36.

9. Aponte, J. J., et al. 2009. Efficacy and safety of intermittent preventivetreatment with sulfadoxine-pyrimethamine for malaria in African infants: apooled analysis of six randomised, placebo-controlled trials. Lancet 374:1533–1542.

10. Baird, J. K. 2008. Real-world therapies and the problem of vivax malaria.N. Engl. J. Med. 359:2601–2603.

11. Barnes, K. I., et al. 2006. Sulfadoxine-pyrimethamine pharmacokinetics inmalaria: pediatric dosing implications. Clin. Pharmacol. Ther. 80:582–596.

12. Bjorkman, S. 2006. Prediction of cytochrome p450-mediated hepatic drugclearance in neonates, infants and children: how accurate are available scal-ing methods? Clin. Pharmacokinet. 45:1–11.

13. Bokenkamp, A., et al. 1998. Cystatin C—a new marker of glomerular filtra-tion rate in children independent of age and height. Pediatrics 101:875–881.

14. Bouvet, Y., et al. 2006. GFR is better estimated by considering both serumcystatin C and creatinine levels. Pediatr. Nephrol. 21:1299–1306.

15. Brair, M. E., B. J. Brabin, P. Milligan, S. Maxwell, and C. A. Hart. 1994.Reduced transfer of tetanus antibodies with placental malaria. Lancet 343:208–209.

16. Bustos, D. G., et al. 2002. Pharmacokinetics of sequential and simultaneoustreatment with the combination chloroquine and sulfadoxine-pyrimethaminein acute uncomplicated Plasmodium falciparum malaria in the Philippines.Trop. Med. Int. Health 7:584–591.

17. Cairns, M., et al. 2008. Duration of protection against malaria and anaemiaprovided by intermittent preventive treatment in infants in Navrongo,Ghana. PLoS One 3:e2227.

18. Corvaisier, S., et al. 2004. Population pharmacokinetics of pyrimethamineand sulfadoxine in children treated for congenital toxoplasmosis. Antimi-crob. Agents Chemother. 48:3794–3800.

19. Dzinjalamala, F. K., et al. 2005. Association between the pharmacokineticsand in vivo therapeutic efficacy of sulfadoxine-pyrimethamine in Malawianchildren. Antimicrob. Agents Chemother. 49:3601–3606.

20. Filler, G., and N. Lepage. 2003. Should the Schwartz formula for estimationof GFR be replaced by cystatin C formula? Pediatr. Nephrol. 18:981–985.

21. Garner, P., et al. 1994. Birthweight and gestation of village deliveries inPapua New Guinea. J. Trop. Pediatr. 40:37–40.

22. Green, M. D., et al. 2007. Pharmacokinetics of sulfadoxine-pyrimethamine inHIV-infected and uninfected pregnant women in Western Kenya. J. Infect.Dis. 196:1403–1408.

23. Grobusch, M. P., A. Egan, R. D. Gosling, and R. D. Newman. 2007. Inter-mittent preventive therapy for malaria: progress and future directions. Curr.Opin. Infect. Dis. 20:613–620.

24. Grubb, A., et al. 2005. Simple cystatin C-based prediction equations forglomerular filtration rate compared with the modification of diet in renaldisease prediction equation for adults and the Schwartz and the Counahan-Barratt prediction equations for children. Clin. Chem. 51:1420–1431.

25. Hekster, C. A., and T. B. Vree. 1982. Clinical pharmacokinetics of sulpho-namides and their N4-acetyl derivatives. Antibiot. Chemother. 31:22–118.

26. Karunajeewa, H. A., et al. 2009. Pharmacokinetic properties of sulfadoxine-pyrimethamine in pregnant women. Antimicrob. Agents Chemother. 53:4368–4376.

27. Kobbe, R., et al. 2007. A randomized controlled trial of extended intermit-tent preventive antimalarial treatment in infants. Clin. Infect. Dis. 45:16–25.

28. Mansor, S. M., et al. 1989. Single dose kinetic study of the triple combinationmefloquine/sulphadoxine/pyrimethamine (Fansimef) in healthy male volun-teers. Br. J. Clin. Pharmacol. 27:381–386.

29. Marfurt, J., et al. 2007. Low efficacy of amodiaquine or chloroquine plussulfadoxine-pyrimethamine against Plasmodium falciparum and P. vivax ma-laria in Papua New Guinea. Am. J. Trop. Med. Hyg. 77:947–954.

30. May, J., et al. 2008. Therapeutic and prophylactic effect of intermittentpreventive anti-malarial treatment in infants (IPTi) from Ghana and Gabon.Malar. J. 7:198.

31. McLeod, R., et al. 1992. Levels of pyrimethamine in sera and cerebrospinaland ventricular fluids from infants treated for congenital toxoplasmosis.Toxoplasmosis Study Group. Antimicrob. Agents Chemother. 36:1040–1048.

32. Obua, C., et al. 2008. Population pharmacokinetics of chloroquine andsulfadoxine and treatment response in children with malaria: suggestions foran improved dose regimen. Br. J. Clin. Pharmacol. 65:493–501.

33. Papua New Guinea Department of Health. 2005. Standard treatment ofcommon illnesses of children in Papua New Guinea, 8th ed. Papua NewGuinea Department of Health, Port Moresby, Papua New Guinea.

34. Potts, A. L., G. R. Warman, and B. J. Anderson. 2008. Dexmedetomidinedisposition in children: a population analysis. Paediatr. Anaesth. 18:722–730.

35. Roche. 2005. Fansidar product information. Roche Products Pty. Ltd., DeeWhy, Australia.

36. Schwartz, G. J., L. P. Brion, and A. Spitzer. 1987. The use of plasmacreatinine concentration for estimating glomerular filtration rate in infants,children, and adolescents. Pediatr. Clin. N. Am. 34:571–590.

37. Trenque, T., et al. 2004. Population pharmacokinetics of pyrimethamine andsulfadoxine in children with congenital toxoplasmosis. Br. J. Clin. Pharmacol.57:735–741.

38. Weidekamm, E., H. Plozza-Nottebrock, I. Forgo, and U. C. Dubach. 1982.Plasma concentrations in pyrimethamine and sulfadoxine and evaluation ofpharmacokinetic data by computerized curve fitting. Bull. World HealthOrgan. 60:115–122.

39. Whelpton, R., G. Watkins, and S. H. Curry. 1981. Bratton-Marshall andliquid-chromatographic methods compared for determination of sulfameth-azine acetylator status. Clin. Chem. 27:1911–1914.

40. Winstanley, P. A., et al. 1992. The disposition of oral and intramuscularpyrimethamine/sulphadoxine in Kenyan children with high parasitaemia butclinically non-severe falciparum malaria. Br. J. Clin. Pharmacol. 33:143–148.

41. Zappitelli, M., et al. 2006. Derivation and validation of cystatin C-basedprediction equations for GFR in children. Am. J. Kidney Dis. 48:221–230.


ANTIMICROBIAL AGENTS AND CHEMOTHERAPY, Nov. 2011, p. 5306–5313 Vol. 55, No. 110066-4804/11/$12.00 doi:10.1128/AAC.05136-11Copyright © 2011, American Society for Microbiology. All Rights Reserved.

Population Pharmacokinetics of Artemether, Lumefantrine, and TheirRespective Metabolites in Papua New Guinean Children with

Uncomplicated Malaria�

Sam Salman,1 Madhu Page-Sharp,2 Susan Griffin,3 Kaye Kose,3 Peter M. Siba,3Kenneth F. Ilett,1 Ivo Mueller,3† and Timothy M. E. Davis1*

School of Medicine and Pharmacology, University of Western Australia, Fremantle Hospital, Fremantle, Western Australia,Australia1; School of Pharmacy, Curtin University of Technology, Bentley, Australia2; and Papua New Guinea Institute of

Medical Research, Madang, Papua New Guinea3

Received 21 June 2011/Returned for modification 20 July 2011/Accepted 20 August 2011

There are sparse published data relating to the pharmacokinetic properties of artemether, lumefantrine, andtheir active metabolites in children, especially desbutyl-lumefantrine. We studied 13 Papua New Guineanchildren aged 5 to 10 years with uncomplicated malaria who received the six recommended doses of artemether(1.7 mg/kg of body weight) plus lumefantrine (10 mg/kg), given with fat over 3 days. Intensive blood samplingwas carried out over 42 days. Plasma artemether, dihydroartemisinin, lumefantrine, and desbutyl-lumefan-trine were assayed using liquid chromatography-mass spectrometry or high-performance liquid chromatog-raphy. Multicompartmental pharmacokinetic models for a drug plus its metabolite were developed using apopulation approach that included plasma artemether and dihydroartemisinin concentrations below the limitof quantitation. Although artemether bioavailability was variable and its clearance increased by 67.8% witheach dose, the median areas under the plasma concentration-time curve from 0 h to infinity (AUC0–�s) forartemether and dihydroartemisinin (3,063 and 2,839 �g � h/liter, respectively) were similar to those reportedpreviously in adults with malaria. For lumefantrine, the median AUC0–� (459,980 �g � h/liter) was also similarto that in adults with malaria. These data support the higher dose recommended for children weighing 15 to35 kg (35% higher than that for a 50-kg adult) but question the recommendation for a lower dose in childrenweighing 12.5 to 15 kg. The median desbutyl-lumefantrine/lumefantrine ratio in the children in our study was1.13%, within the range reported for adults and higher at later time points because of the longer desbutyl-lumefantrine terminal elimination half-life. A combined desbutyl-lumefantrine and lumefantrine AUC0–�

weighted on in vitro antimalarial activity was inversely associated with recurrent parasitemia, suggesting thatboth the parent drug and the metabolite contribute to the treatment outcome of artemether-lumefantrine.

Artemether (ARM)-lumefantrine (LUM) (AL) is a fixed-dose combination therapy used widely for the treatment ofmalaria (33). ARM is a lipophilic artemisinin derivative that isconverted in vivo to dihydroartemisinin (DHA), an active me-tabolite. Both ARM and DHA have short half-lives (14, 19–21,24, 25, 31) but a rapid effect on parasitemia. LUM is a highlylipophilic drug with a longer half-life (11, 13, 14, 19, 20, 24, 30)which is combined with ARM primarily to prevent late recru-descence. Although the pharmacokinetic (PK) properties ofARM, DHA, and LUM in adults have been well documented(4, 5, 11, 13–15, 19–22, 24, 30), there are scant and inconsistentdata relating to the disposition of desbutyl-lumefantrine(DBL), a potent LUM metabolite (26, 28, 29, 32) that mayinfluence AL’s treatment outcome (32). Reported plasmaDBL-to-LUM concentration ratios after AL dosing in adultsdiffer �10-fold (15, 24), while the pharmacokinetic propertiesof DBL in children are unknown. In addition, although severalstudies have attempted to characterize LUM disposition in

children with malaria (1, 16, 25), methodological issues com-plicate the comparison of child data with adult data. One studyinvolving a limited sampling schedule suggested that AL-treated children with malaria receive an inadequate dose ofLUM relative to that for healthy adults (25), while the otherstudies either used pooled plasma concentrations (1) or used atruncated sampling schedule inadequate to characterize LUMpharmacokinetics (16).

In view of this situation, we have characterized the popula-tion pharmacokinetics of ARM, LUM, and their metabolites inpediatric malaria by using a rich sampling schedule to assesspotential differences in disposition between children and adultsand to add to the limited data on DBL disposition and its rolein AL’s treatment outcome.


Patients. We recruited children aged 5 to 10 years from Alexishafen HealthCentre, Madang Province, on the north coast of Papua New Guinea. The clinicserves an area where Plasmodium falciparum and Plasmodium vivax are hyper-endemic and Plasmodium ovale and Plasmodium malariae are also transmitted.Children with an axillary temperature of �37.5°C or a history of fever in theprevious 24 h were screened with a Giemsa-stained thick blood film read by anon-site, trained microscopist. Those with a monoinfection of P. falciparum(�1,000 asexual parasites/microliter) or P. vivax, P. ovale, or P. malariae (�250/microliter) were eligible, provided that the child’s parents gave informed con-sent, there were no features of severe malaria (34), they had not taken anyantimalarial drug in the previous 14 days, there was no evidence of another cause

* Corresponding author. Mailing address: University of WesternAustralia, School of Medicine and Pharmacology, Fremantle Hospital,P.O. Box 480, Fremantle, Western Australia 6959, Australia. Phone: 6189431 3229. Fax: 618 9431 2977. E-mail: [email protected].

† Present address: Centre de Recerca de Salut Internacional deBarcelona (CRESIB), Barcelona, Spain.

� Published ahead of print on 29 August 2011.

5306

of fever, and there were no features of malnutrition or other chronic comorbid-ity. The study was approved by the Medical Research Advisory Committee of theDepartment of Health, Papua New Guinea.

Clinical methods. After enrollment, a standardized history was taken and aclinical examination performed. A 3-ml blood sample was taken for blood filmmicroscopy, baseline hemoglobin and blood glucose measurements were taken,and a subsequent drug assay of separated plasma was performed. Urinalysis andaudiometric assessment were performed. Each child was given artemether-lumefantrine (Coartem, Novartis Pharma Ltd., Switzerland) at a dose of 1.7 and10 mg/kg of body weight, respectively, to the nearest tablet. This dose wasrepeated at 8, 24, 36, 48, and 60 h, with the exact time of dosing recorded. Alldoses were given under direct observation with at least 50 ml of cow’s milk(equivalent to 2 g of fat). Further venous blood samples were taken from anindwelling intravenous catheter at 4, 8, 12, 24, 36, 40, 48, 60, 64, 68, and 72 h andthen by venesection on days 4, 5, 7, 14, and 28. All samples were centrifugedpromptly, and red cells and separated plasma were stored frozen at �80°C untilassayed. A detailed clinical assessment including a symptom questionnaire, ablood film, and hemoglobin and blood glucose measurements was repeated ondays 1, 2, 3, and 7, with additional clinical assessment and blood films on days 14,28, and 42.

Laboratory methods. All blood smears taken at baseline and during follow-upwere examined independently by two skilled microscopists in a central labora-tory. Each microscopist viewed 100 fields at �1,000 magnification before a slidewas considered negative. Any slide discrepant for positivity/negativity or specieswas referred to a third microscopist for adjudication.

For drug assays, high-performance liquid chromatography (HPLC)-grade ace-tonitrile (Merck, Kilsyth, Australia), tert-butyl chloride, ethyl acetate, glacialacetic acid, and formic acid (Merck, Darmstadt, Germany), and ammoniumformate (Sigma-Aldrich, Gillingham, United Kingdom) were used. Other sol-vents and chemicals were of analytical grade. Stock solutions (1 �g/liter inmethanol) of ARM (AAPIN Chemicals, Abingdon, United Kingdom), DHA(Sigma, St. Louis, MO), and artemisinin (used as an internal standard; Sigma)were stored and protected from light at �80°C and used to prepare workingdilutions (0.1, 1, and 10 �g/ml). Calibration curves (2 to 200 �g/liter) wereconstructed for DHA and ARM by spiking blank plasma. Quality control (QC)samples were prepared in blank plasma at 10, 20, 50, and 200 �g/liter and alsostored at �80°C prior to use.

ARM and DHA were extracted as previously described (7) but with thefollowing modifications. Briefly, solid-phase extraction (SPE) Bond Elut PHcolumns (Varian Inc., Palo Alto, CA) were preconditioned with 1 ml of methanolfollowed by 1 ml of 1 M acetic acid. Plasma (0.5 ml) was spiked with an internalstandard (artemisinin, 100 �g/liter), loaded onto the SPE column, and drawnthrough with a medium-suction vacuum. The column was then washed twice with1 M acetic acid (1 ml), followed by 20% (vol/vol) methanol in 1 M acetic acid (1ml). The column was dried with a low-suction vacuum for 30 min, and theretained drugs were eluted using 2 ml of tert-butyl chloride–ethyl acetate (80:20[vol/vol]). The eluate was then evaporated under vacuum at 35°C, reconstitutedin 50 �l of the mobile phase, and kept overnight to equilibrate the � and �anomers of DHA (7). Only the � anomer was used for quantification. Theinjection volume was 10 �l.

The liquid chromatography-mass spectrometry (LC-MS) system used was asingle-quad mass spectrometer (Shimadzu, Kyoto, Japan) with electrospray ion-ization (ESI) and atmospheric-pressure chemical ionization (APCI) systems.Assays were performed with 20 mM ammonium formate (pH 5) and acetonitrilein 0.1% formic acid (40:60) at a flow rate of 0.2 ml/min, and chromatographicseparation was undertaken at ambient temperature on a Synergi Fusion-RP C18

column (inside diameter [i.d.], 150 mm by 2.0 mm) coupled with a 5-�m-particleC18 guard column (i.d., 4 mm by 3 mm; Phenomenex, Lane Cove, Australia).Retention times were 4.5, 7.5, and 12.7 min for DHA, artemisinin, and ARM,respectively. Optimized mass spectra were acquired with an interface voltage of4.5 kV, a detector voltage of 1 kV, a heat block temperature of 400°C, and adesolvation gas temperature of 250°C. Nitrogen was used as a nebulizer gas at aflow rate of 1.5 liters/min and as a dry gas at a flow of 10 liters/min. Quantitationwas performed by selected ion monitoring using the dual-ionization sourcemode. The predominant fragmented ions, m/z 221 for ARM and m/z 221 forDHA, were used. For artemisinin, m/z 283 was monitored.

The standard curves were linear (r2 � 0.999). The chromatographic data (the peakarea ratios of DHA to artemisinin and ARM to artemisinin) were processed usingLabSolutions software (version 5; Shimadzu, Japan). No matrix effect (ion suppres-sion/enhancement) was observed under methodologies described elsewhere (23),and the performance of both assays, assessed as intra- and interday relative standarddeviations across relevant concentration ranges, was similar to that published pre-viously (7, 18). Interday accuracies of QC assays were �15% of nominal values on

all occasions. The limits of quantification and detection were, respectively, 2 and 1�g/liter for DHA and 5 and 2 �g/liter for ARM.

LUM and DBL were quantified in plasma using validated high-performanceliquid chromatography with a UV detection assay (HPLC-UV) and a validatedultra-high-performance liquid chromatography-tandem mass spectrometry (LC-MS/MS) assay, respectively, as previously described (32). The linear range forLUM was 20 to 20,000 ng/ml; interday variability was 4.94%, 4.93%, 7.16%, and11.23% and intraday variability was 2.83%, 4.41%, 4.11%, and 9.55% at 20,000,2,000, 200, and 20 ng/ml, respectively. For DBL, the linear range was 0.5 to 100ng/ml; interday variability was 3.36%, 3.47%, 9.98%, and 6.74% and intradayvariability was 2.47%, 3.46%, 8.16%, and 3.48% at 50, 10, 1, and 0.5 ng/ml,respectively. As an LC-MS/MS method was used for DBL, matrix effects wereassessed where between-subject variability was 3.37%, 4.47%, and 9.43% at 50,10, and 1 ng/ml, respectively.

Pharmacokinetic modeling. Loge (natural log) plasma concentration-time datasets for LUM with DBL and for ARM with DHA were analyzed by nonlinearmixed-effect modeling using NONMEM (version 6.2.0; Icon Development So-lutions, Ellicott City, MD) with an Intel Visual Fortran 10.0 compiler. Thefirst-order conditional-estimation with interaction (FOCE-I) method was usedfor the LUM-DBL model, and the Laplacian with interaction method was usedfor ARM-DHA. The minimum objective-function value (OFV) and weighted-residual (WRES) plots were used to choose suitable models during model build-ing. As FOCE-I estimation was used, conditional weighted residuals were con-sidered in addition to WRESs in the initial stages of model building (17).However, as they were similar, WRESs were considered suitable for furthermodel building. Concentrations were modeled in �g/ml, with a conversion factorfor all metabolite parameters included in the model to account for the differencein molecular weight between the parent drug and the metabolite. Allometricscaling was used a priori, with volume terms multiplied by (weight/70)1.0 andclearance terms by (weight/70)0.75 (3). Residual variability (RV) was estimated asan additive error for the loge-transformed data. Models were parameterizedusing the absorption rate constant (ka), central volume of the distribution (VC/F,where F is bioavailability), clearance (CL/F), and peripheral volume of thedistribution(s) (VP/F) and its respective intercompartmental clearance(s) (Q/F).

For the LUM-DBL model, plasma LUM concentrations were initially modeledusing inbuilt 2- and 3-compartment model structures with first-order absorption anda fixed lag time of 2 h (22) (Advan 4 and 12). Once a suitable (3-compartment) LUMmodel had been determined, the DBL data set was added and modeled simultane-ously. User-defined linear mammillary models (Advan 5) were constructed by testing1-, 2-, and 3-compartment models with and without first-pass LUM metabolism. Asno data exist regarding the degree of in vivo DBL conversion from LUM, this was setto 100% to allow identifiability. Therefore, all clearance and volume terms for DBLare relative to LUM bioavailability (FLUM) as well as the degree of metabolicconversion from LUM (Fmet-DBL). The term F*DBL (representing FLUM timesFmet-DBL) will be used for simplicity.

As 45% and 12% of plasma ARM and DHA concentrations, respectively, werebelow the limit of quantification (BLQ), we used a published method known toproduce reliable pharmacokinetic parameters in this situation (9, 10). Themethod (known as M3) (2) models continuous and categorical data simultane-ously. Concentrations above the limit of quantification (LOQ) are included asconventional continuous data, while those BLQ are treated as categorical, andthe likelihood (probability) that they are BLQ was maximized with respect tomodel parameters. This allows BLQ observations to contribute to the determi-nation of the OFV and the finalizing of the model structure.

Initially, plasma ARM concentrations were assessed using 1- and 2-compart-ment models with first-order absorption (Advan 2 and 4) to obtain a suitablestructure. The ka for ARM was fixed to 1 h�1 (31), as the data did not supportits estimation. Once a suitable (2-compartment) ARM model had been deter-mined, the DHA data set was added and modeled simultaneously using a user-defined linear mammillary model (Advan 5). For DHA, 1- and 2-compartmentmodels were assessed and the conversion of ARM to DHA was consideredcomplete for identifiability purposes. Therefore, all clearance and volume termsfor DHA are relative to ARM bioavailability (FARM) as well as the degree ofmetabolic conversion from ARM (Fmet-DHA). The term F*DHA (representingFARM times Fmet-DHA) will be used for simplicity.

Once the model structure was established, interindividual variability (IIV),interoccasion variability (IOV), and their correlations were estimated. Relation-ships between model parameters and the covariates of age, sex, baseline para-sitemia, and baseline hemoglobin were identified using correlation plots andsubsequently evaluated within NONMEM. Inclusion of the covariate relation-ship required a decrease in OFV of �6.63 (�2 distribution with 1 df, P � 0.01),accompanied by a decrease in the IIV of that parameter.

VOL. 55, 2011 ARTEMETHER-LUMEFANTRINE PHARMACOKINETICS IN CHILDREN 5307

Model evaluation. A bootstrap using Perl-speaks-NONMEM (PSN) with 1,000samples was performed, and the parameters derived from this analysis weresummarized as the median and 2.5th and 97.5th centiles (95% empirical confi-dence interval [CI]) to facilitate evaluation of the final-model parameter esti-mates. Runs were included in the bootstrap analysis regardless of their minimi-zation status. In addition, visual predictive checks (VPCs) were performed with1,000 data sets simulated from the final models. The observed 10th, 50th, and90th percentiles were plotted with their respective simulated 95% CIs to assess thepredictive performance of the model. For the ARM-DHA model, the observedfraction of BLQ observations was compared with the median and 95% predictionintervals (PIs) of BLQ observations from these simulated data sets (9).

The applicability of the final population models to younger patients from thepresent sample was assessed using a numerical predictive check. Day 7 plasmaLUM concentrations (18) from children aged 0.5 to 5 years from a previous studywere compared with the simulated data from the final models. The actual andsimulated numbers of data points above and below the 20%, 40%, 60%, 80%,90%, and 95% simulated prediction intervals were compared.

Statistical analysis. Changes in hemoglobin, glucose, and audiometric dataover time were assessed using the Wilcoxon signed-rank test. The areas underthe concentration-time curves from 0 h to infinity (AUC0–�s) of DBL and LUMwere compared between subjects with or without recurrent parasitemia using theMann-Whitney U test. A two-tailed level of significance of 0.05 was consideredsignificant for all comparisons.

RESULTS

Clinical characteristics and course. The baseline character-istics of the 13 recruited children are summarized in Table 1.Eleven had a monoinfection (9 P. falciparum, 2 P. malariaeinfections) on confirmatory expert microscopy, while 2 had amixed P. falciparum/P. vivax infection. AL treatment was welltolerated, and reported symptoms were mild/moderate, short-lived (�3 days), and consistent with clinical features of uncom-plicated malaria. Times to initial fever and parasite clearancewere �48 h in all cases.

By the 28th day of follow-up, three children had developedslide-positive P. vivax (two had P. vivax at enrollment) and twochildren had developed P. falciparum (one had P. falciparum atenrollment) parasitemia. By the 42nd day of follow-up, fivechildren had been diagnosed with P. vivax (including the twowho had P. vivax at enrollment) and three with P. falciparum(including the one who had P. falciparum at enrollment). Thesedata are consistent with the uncorrected PCR results of aprevious, larger comparative treatment trial in younger chil-dren performed at the same location (18). The recurrent P.

vivax parasitemia could have resulted from (i) a recrudescentinfection in those infected with this parasite before treatment,(ii) the acquisition of a new P. vivax infection after treatment,or, since no primaquine therapy was administered, (iii) theappearance of P. vivax from hypnozoites present in the liver atstudy entry. P. falciparum parasitemia detected during fol-low-up could have represented recrudescence or reinfection.

The mean hemoglobin concentration was significantlyhigher on day 28 than at enrollment (10.7 versus 8.9 g/liter, P �0.01). There was no significant change in blood glucose overthe first 3 days of enrollment or in audiometric findings over 28days (data not shown).

Pharmacokinetic modeling. LUM and DBL plasma concen-tration-time curves are shown in Fig. 1. A 3-compartmentmodel proved superior to a 2-compartment model for LUM,with a lower OFV and reduced bias in the WRES plot. Theaddition of two compartments and the inclusion of first-passmetabolism provided the best model once the DBL data sethad been added. Therefore, the final model comprised 3 com-partments for LUM and 2 compartments for DBL. The struc-tural model parameters were ka, VC/FLUM, VP1/FLUM, VP2/FLUM, CL/FLUM, Q1/FLUM, Q2/FLUM, the percentagecontribution of first-pass metabolism to DBL metabolic con-version (FP), VC/F*DBL, VP/F*DBL, CL/F*DBL, and Q/F*DBL.Interindividual variability was able to be estimated for ka, CL/FLUM, CL/F*DBL, VC/F*DBL, and FLUM, as was interoccasionvariability for FLUM (the population value of FLUM remainedfixed to 1). The variability in FLUM values was smaller betweenindividuals than it was between doses in the same individual(20 versus 67%). Once IIV and IOV terms were added, in-spection of the WRES plot revealed a bias due to the absorp-tion profile of the final dose. Estimation of a separate ka for the6th and final dose (kaD6) improved the bias and reduced theOFV (�7.519, P � 0.01). None of the covariates tested im-proved the model. Residual variability (20.8% and 20.9% forLUM and DBL, respectively) was low.

The final-model parameter estimates and the bootstrap re-sults are summarized in Table 2. Bias was �10% for structural-and random-model parameters. Figures 2 and 3 show good-

TABLE 1. Baseline characteristics of study participants

Characteristic Value (%)a

Age (yr) ....................................................................................... 7.7 � 1.4Male............................................................................................. 8 (62)Wt (kg) ........................................................................................19.0 � 3.5Ht (cm)........................................................................................ 112 � 9Axillary temp (°C)......................................................................36.8 � 1.0

Infection typesP. falciparum parasitemia...................................................... 9 (69)P. falciparum/P. vivax parasitemia........................................ 2 (15)P. malariae parasitemia ......................................................... 2 (15)

Respiratory rate/min.................................................................. 28 � 9Supine pulse rate/min................................................................ 102 � 16Mean upper arm circumference (cm) ..................................... 16 � 1Hemoglobin (g/liter) .................................................................. 8.9 � 1.6

a Data are numbers (percentages), means � standard deviations, or medians �interquartile ranges (n � 13).

FIG. 1. Time-concentration plots showing concentrations of LUM(E) and DBL (ƒ) in �g/liter on a log10 scale. Curves of the medianconcentrations for LUM (solid black line) and DBL (dashed blackline) are also shown.


ness-of-fit plots and VPCs, respectively. The half-lives andAUCs of LUM and DBL are shown in Table 3. The firstdistribution, second distribution, and terminal eliminationhalf-lives for LUM had median values of 10.4, 46.6, and 126 h,respectively, while DBL had a median distribution half-life of19.7 h and a median terminal elimination half-life of 141 h.Overall, the metabolite-to-parent-drug ratio was 1.13% (ob-tained from AUC0–�s), but there was a higher ratio at latertime points. Day 7 LUM concentrations obtained fromyounger children were consistent with predictions based on thefinal model, which resulted in the expected numbers of obser-vations above and below the 20, 40, 60, 80, 90, and 95% sim-ulated PIs. When the same data for DBL were compared,there was an excess of points above the 20, 40, 60, 80, and 90%PIs and a lack of points below the 20 and 40% PIs, especiallyat a younger age, demonstrating that the day 7 DBL levels inthe younger children were higher than expected from themodel.

Initial modeling of ARM-DHA data sets proved difficult,given the large proportion of BLQ data (45% and 12% forARM and DHA, respectively). Once these data were incorpo-rated into the model using the M3 method from Ahn et al. (2),more-acceptable models were obtained. The dispositions ofARM and DHA were best described by a 2-compartmentmodel for ARM and a 1-compartment model for DHA. Thestructural-model parameters were ka, VC/FARM, VP/FARM, CL/FARM, Q/FARM, VC/F*DHA, and CL/F*DHA. As with LUM, theIIV and IOV of FARM were estimated, and the variabilitybetween doses was larger than between individuals (84.1 versus38.1%). The IIV of CLARM was also estimated. A relationshipbetween CLARM and dose number was included and demon-strated that for each subsequent dose of ARM, CLARM in-creased by 67.8% relative to its value after the first dose. Thisrelationship was accompanied by a decrease in the OFV

(�82.774, P � 0.001) and a reduction in the RVs of both ARMand DHA. No other covariate relationship improved themodel. After the inclusion of IIV/IOV terms and the covariaterelationship, the RVs were still high, at 51.6% and 53.3% forARM and DHA, respectively.

The final-model parameter estimates and the bootstrap re-sults are summarized in Table 4. As the covariance step wasnot successful, NONMEM-derived relative standard errorscould not be obtained. Bias was �11% for structural andrandom parameters, except the IIV for FARM, which had anegative 48% bias. Figures 4 and 5 show goodness-of-fit plotsand VPCs, respectively. The VPCs show all observed 10th,50th, and 90th percentiles within their simulated 95% CIs andthe fraction of BLQ data at each time point within its 95% CIfor both ARM and DHA. Secondary parameters for studyparticipants are shown in Table 3. The AUC0–�s and half-livesof ARM decreased with each dose, while the median DHA-to-ARM ratio increased.

Relationship between drug exposure and treatment out-come. The LUM AUC0–� in children with recurrent parasitemiaon days 28 (n � 5) and 42 (n � 8) tended to be lower than thatin children who remained aparasitemic at these times (P � 0.057and 0.086, respectively). There were no differences in the AUC0–�

for DBL (P � 0.46 and 0.89, respectively). However, a combinedAUC0–�, with DBL weighted four times more than LUM (con-

TABLE 2. Final population pharmacokinetic estimates andbootstrap results for LUM and DBL

Parameter Mean(RSE %)a Bootstrap median (95% CI)

Objective-function value �586.510 �601.901 (�668.687 to �559.564)

Structural-model parameterska (/h) 0.461 (20) 0.442 (0.285 to 0.644)CL/FLUM (liters/h/70 kg) 7.29 (9) 7.21 (5.55 to 9.04)VC/FLUM (liters/70 kg) 227 (12) 225 (147 to 284)Q1/FLUM (liters/h/70 kg) 1.52 (16) 1.57 (0.96 to 2.32)VP1/FLUM (liters/70 kg) 115 (19) 109 (57 to 214)Q2/FLUM (liters/h/70 kg) 0.743 (13) 0.805 (0.208 to 1.27)VP2/FLUM (liters/70 kg) 164 (8) 168 (97 to 240)kaD6 (/h) 1.20 (52) 1.14 (0.50 to 3.68)FP (%) 6.29 (15) 6.45 (4.36 to 9.84)CL/F*DBL (liters/h/70 kg) 701 (10) 694 (561 to 851)VC/F*DBL (liters/70 kg) 51,100 (10) 51,200 (42,200 to 61,430)Q/F*DBL (liters/h/70 kg) 439 (19) 424 (305 to 632)VP/F*DBL (liters/70 kg) 68,400 (14) 68,000 (51,800 to 88,600)

Random-model parametersIIV in FLUM (%) 19.8 (42) 18.9 (2.5 to 29.3)IIV in ka (%) 55.4 (44) 55.8 (17.1 to 92.2)IIV in CL/FLUM (%) 17.7 (20) 16.9 (6.8 to 23.7)IIV in CL/F*DBL (%) 26.2 (26) 26.0 (10.4 to 37.6)IIV in VC/F*DBL (%) 34.1 (22) 33.3 (17.4 to 47.8)IOV in FLUM (%) 67.0 (9) 66.4 (53.4 to 77.7)RV for LUM (%) 20.8 (7) 20.3 (17.4 to 22.5)RV for DBL (%) 20.9 (7) 20.6 (17.5 to 23.0)

a RSE percentages are the NONMEM-produced values from the covariance step.

FIG. 2. Population (E) and individual (F) predicted data versusobserved data for LUM (A) and DBL (B) concentrations (�g/liter)from the final model. The lines of identity are also shown.


sistent with its greater antimalarial potency in vitro [26, 28, 29,32]), was significantly lower in children with recurrent parasitemiaon day 28 than in aparasitemic children (P � 0.028) and was ofborderline significance on day 42 (P � 0.063).

DISCUSSION

In the present study of Papua New Guinean children withuncomplicated malaria treated with a conventional AL regi-men, rich data sets of plasma concentrations of LUM, ARM,and their active metabolites measured during an extended fol-low-up period were successfully analyzed using populationpharmacokinetic modeling that allowed for a high proportionof BLQ plasma ARM and DHA concentrations. Our analysesincluded the first compartmental PK analysis of plasma DBLlevels. We found that current dose recommendations for AL inchildren result in a LUM AUC similar to that achieved inadults despite children receiving a higher average mg/kg dosethan a 50-kg adult. However, the subgroup of children weigh-ing 12.5 to 15 kg receives the lowest mg/kg dose and may be atrisk of being underdosed.

Three studies, all from Africa, have examined LUM phar-macokinetics after AL treatment in children. The first andsimplest compared crushed tablets and a dispersible formula-tion by using a pooled analysis of single blood samples taken atone of six time points during a 14-day period from 726 children

FIG. 3. Visual predictive check showing the observed 50th (F),10th (197), and 90th (E) percentiles with the simulated 95% CIs forthe 50th (solid black line), 10th (dotted gray lines), and 90th (dashedgray lines) percentiles for LUM (A) and DBL (B) concentrations(�g/liter on a log10 scale) from the final model.

TABLE 3. Secondary pharmacokinetic parameters derived from post hoc Bayesian estimates for study participantsa

Parameter LUM DBLARM

DHADose 1 Dose 6 All doses

t1/2�b (h) 10.4 (10.3–11.8) 19.7 (18.4–22.5) 0.62 (0.60–0.64) 0.16 (0.12–0.33) 0.80 (0.76–0.82)

t1/2�b (h) 46.6 (44.8–48.2) 141 (135–150) 16.4 (15.7–16.8) 11.9 (11.2–13.2)

t1/2�b (h) 123 (120–127)

AUC0–�c

(�g � h/liter)459,980 (391,330–632,730) 5,434 (4,394–8,542) 983 (371–1,770) 164 (145–254) 3,063 (2,357–4,513) 2,839 (1,812–3,488)

AUCmetabolite/AUCparentdrug (%)

1.13 (0.93–1.55) 36.8 (36.8–36.8) 186 (91.8–268) 92.7 (59.2–94.3)

a Data are medians (interquartile ranges).b For LUM, t1/2�, t1/2� and t1/2� are the first-distribution, second-distribution, and terminal-elimination half-lives, respectively, while for DBL and ARM, t1/2� and t1/2�

represent the distribution and terminal-elimination half-lives, respectively, and for DHA, t1/2� represents the terminal-elimination half-life.c Represents either the AUC0–� for all six doses together or the AUC0–� for individual doses as if they were given alone.

TABLE 4. Final population pharmacokinetic estimates andbootstrap results for ARM and DHA

Parameter Mean(RSE %a)

Bootstrap median(95% CI)

Objective-function value 159.853 177.255 (77.606–249.014)

Structural-model parametersCL/FARM (liters/h/70 kg) 102 (27) 96.3 (57.0–167.0)VC/FARM (liters/70 kg) 193 (62) 172 (40–506)Q/FARM (liters/h/70 kg) 49.6 (47) 45.8 (19.7–111.1)VP/FARM (liters/70 kg) 1,070 (59) 1,220 (593–3,011)ka (/h) 1 (fixed) 1 (1–1)VC/F*DHA (liters/70 kg) 440 (40) 417 (69–826)CL/F*DHA (liters/h/70 kg) 277 (26) 275 (140–443)% increase in CL/FARM

for each subsequentdose (%)

67.8 (31) 73.3 (40.5–125)

Random-model parametersIIV in FARM (%) 38.1 (72) 19.7 (0.3–58.6)IOV in FARM (%) 84.1 (38) 84.2 (52.2–113.6)IIV in CL/FARM (%) 84.0 (33) 75.8 (49.1–108.2)RV for ARM (%) 51.6 (12) 50 (37–61)RV for DHA (%) 53.3 (20) 61 (42–83)

a RSE percentages are derived from the bootstrap.


�12 years of age (1). The LUM AUC for both formulationswas higher than in the present study (574,000 and 636,000,respectively, versus 459,980 �g � h liter�1). In the second study(25), six blood samples were taken from children aged 5 to 13years, starting when the last AL dose was given, and the LUMAUC60–� was calculated using noncompartmental analysis.When we used our final models to generate an AUC60–�, it washigher (257,010 versus 210,000 �g � h liter�1). Based on theirdata, the authors reported that children have lower levels ofexposure to LUM than adults using recommended AL doseschedules (25). A third study of children aged 1 to 10 yearsutilized a population approach (16), but there was no samplingbeyond 72 h and no secondary pharmacokinetic parameterswere provided. A comparison with LUM disposition in thepresent study was, therefore, not possible.

Comparisons of LUM AUCs between studies in adults arealso difficult, as some report the AUC from the first dose, whileothers use the AUC60–�. Table 5 summarizes the available datafor both measures of drug exposure. There is a differencebetween LUM exposure in healthy adults and that in subjectswith malaria, but the AUCs for nonpregnant adults, pregnantadults, and children with malaria are similar. Current AL doserecommendations for children ensure that those weighing 15

to 35 kg receive a 35%-higher average mg/kg dose than a 50-kgadult, but those weighing 12.5 to 15 kg receive a lower mg/kgdose. The AUC data support the higher average mg/kg dosefor children and suggest that those weighing 12.5 to 15 kgshould receive 2 tablets rather than 1 to avoid underdosingwhile not exceeding the highest recommended mg/kg dose(Fig. 6). As LUM exposure, measured as either the AUC orday 7 levels, has previously been shown to be a prime deter-minate of efficacy (12, 27), it is important that underdosing isavoided.

The three studies of AL in children also measured plasmaARM-DHA concentrations (1, 16, 25). The first was not ableto calculate AUCs from pooled concentration data due to asparse sampling schedule (1). The second employed a limitedsampling schedule starting from the last AL dose (25), and the

FIG. 4. Population (E) and individual (F) predicted data versusobserved data for ARM (A) and DHA (B) concentrations (�g/liter)from the final model. The lines of identity are also shown. The dashedgray lines represent the LOQs of ARM (A) and DHA (B).

FIG. 5. Visual predictive check showing the observed 50th (F), 10th(197), and 90th (E) percentiles with the simulated 95% CIs for the 50th(solid black line), 10th (dotted gray lines), and 90th (dashed gray lines)percentiles for ARM (A) and DHA (B) concentrations (�g/liter on a log10scale) from the final model. The fractions of BLQ observations from thedata (E connected with a dotted black line) with the simulated 95%prediction intervals are also shown for both ARM and DHA.


AUCs were therefore lower than those of the present study(168 versus 217 �g � h liter�1 for ARM and 382 versus 402�g � h liter�1 for DHA). The population approach used in thethird study (16) produced models of the dispositions of ARM(two compartments) and DHA (one compartment) that weresimilar to those of the present study. Those authors reported asimilar increase in CL/FARM with each dose (57% versus67.8% in the children in our study) and higher RVs (61%versus 51.6% and 82% versus 53.3% for ARM and DHA,respectively), the latter observation likely a reflection of thefact that many plasma concentrations were close to or belowthe LOQ. As no secondary PK parameters were provided, acomparison of AUCs could not be performed. However, thehalf-lives of ARM, estimated from the pharmacokinetic pa-rameters provided, were longer than those in the children inour study (0.89 versus 0.62 h and 32.0 versus 16.4 h for distri-bution and elimination, respectively, of the first dose), while theelimination half-life of DHA was shorter (0.38 versus 0.80 h).

The AUCs for ARM and DHA in the present study weresimilar to those reported previously in adults with malaria (21,24) but higher than those in healthy adults (14, 19, 20). Ourterminal elimination half-life for ARM was longer than thosereported in these studies (16.4 versus 1.5 to 3.9 h), while forDHA, it was shorter (0.80 versus 1.2 to 2.1 h). The adult studiesused noncompartmental methods to determine these half-lives, and this may account for the differences. Nevertheless,based on these comparisons, exposure to ARM and DHA inchildren is adequate with current AL dose recommendations.

Few studies have evaluated the disposition of DBL, an activemetabolite of LUM. Our DBL/LUM ratio (1.13%) falls be-tween values reported in previous treatment studies (0.33%and 5.2%) (15, 24). The lower value (0.33%) was from a studyof nonimmune Colombian adults with malaria that sampled to168 h and reported the AUC0–168. The higher value (5.2%) wasfrom a study of pregnant Thai women with malaria in whichsampling started after the last dose and the AUC60–� wasreported. The difference between these values can, at least inpart, be explained by the study designs, as the metabolite-to-parent-drug percentage calculated from the AUC60–� in thepresent study is more than double that for the AUC0–168 (1.96versus 0.76%). However, it is likely that ethnicity and preg-nancy contribute to the difference. Age may also influence the

metabolic conversion of LUM to DBL, as our PK model waseffectively able to predict concentrations of LUM, but notDBL, in young children. It is uncertain whether malaria itselfalso influences the ratio, since it was 0.45%, within the range ofvalues from studies of malaria after a single dose of AL in 22healthy adults (G. Lefevre, personal communication).

As reported previously (24), DBL had a longer terminalelimination half-life than LUM in the present study (141 versus123 h), and therefore, the DBL/LUM ratio will increase withtime. Although the ratios found in available studies are low,the in vitro potency of DBL is between 2.2 and 7.2 times that ofLUM (26, 28, 29, 32) and it may therefore contribute to thetherapeutic outcome. We found that a combined weightedLUM-DBL AUC was likely to be lower than the AUC of eitherLUM or DBL alone in subjects with recurrent parasitemia atdays 28 and 42. This supports the suggestion that DBL mayinfluence AL’s treatment outcome (32).

Although the variable bioavailabilities of ARM and LUMhave been previously reported (13), they have not been previ-ously quantified in children. Given the significant increase inthe number of fed versus fasted healthy volunteers (22), it isrecommended that AL be administered with fat in order toimprove absorption. Based on a study in healthy adults whoreceived a single dose of AL, 1.2 g of fat (equivalent to 35 mlof full-cream milk) is required to achieve 90% of the maximalLUM bioavailability (4). These results may not be directlyapplicable to the children with malaria in our study, as theyingested 2 g of fat with each dose, but there was still significantbetween-dose variability in the bioavailabilities of both LUM(67.0%) and ARM (84.1%). We were unable to identify factorsthat may be responsible for these observations.

In the analysis of the ARM-DHA data set, there was asignificant number of BLQ plasma concentrations. This is anissue encountered in pharmacokinetic analyses of a variety ofother antimalarial drugs (6, 8, 16). Traditional approaches tothis problem, such as excluding BLQ data from the analysis orsetting them to a specific value (such as 0 or 50% of the LOQ),have been shown to bias the pharmacokinetic parameters, even

FIG. 6. Doses of lumefantrine and artemether in mg/kg given tochildren weighing 5 to 35 kg under current (solid black line) andsuggested (dashed gray line) dosing regimens. The horizontal dottedblack line represents the dose in mg/kg recommended for a 50-kgadult.

TABLE 5. Summary of studies reporting theAUC for lumefantrinea

Population sampleAUC60–� or AUC60–t

(�g � h/liter)(reference�s�)

AUC0–� or AUC0–t(�g � h/liter) (reference�s�)

Healthy adults 383,000–456,000 (14, 19) 1,242,000–2,730,000b (11, 14)Nonpregnant adults

with malaria335,000–758,000 (5, 13, 15, 22)

Pregnant womenwith malaria

252,000 (24) 472,000 (30)

Children withmalaria

210,000 (16) 572,000–636,000 (1c)

Children in presentstudy

257,000 459,980

a AUCs are either medians or means and are reported either to the last datapoint (t) or to infinity.

b As subjects in the study by Bindschedler et al. (11) received only a singledose, the reported AUC has been multiplied by 6.

c This study used a pooled approach from single observations of each subjectto calculate the AUC.


when only 10% of the data are BLQ (2, 9, 10, 35). Our ap-proach was to use a method within NONMEM shown to havelittle bias in situations with up to 40% of data BLQ in apopulation analysis (10). This method treats BLQ data pointsas categorical data and maximizes the likelihood that theirvalues are truly below the LOQ (2). Although the implemen-tation of this method has previously been difficult and time-consuming, changes to NONMEM and more-efficient dataprocessing have increased its accessibility. The benefits of thismethod when demonstrated in relatively simple models arelikely to apply to more-complex models with parent drugs andmetabolites. We were unable to obtain relative standard errors(RSEs) for our parameters in this model, as the covariancestep was unsuccessful, a common problem when this method isused (9, 10). However, this does not impact the reliability ofthe results obtained, and other methods of model evaluation(such as bootstrap and VPC analyses) can still be used.

Our novel data relating to DBL pharmacokinetics andDBL’s favorable pharmacodynamic effects suggest that futureefficacy and pharmacokinetic studies of LUM should includeDBL assays to further elucidate its role. We have also shownthat analytical techniques that utilize BLQ data to refine phar-macokinetic parameter estimates can be applied in this situa-tion. Extended sampling and a population pharmacokineticapproach allow flexibility in deriving secondary parameters, animportant consideration when comparisons with publishednonstandard measures, such as time-limited AUCs, are of in-terest. Our data confirm that current AL dose recommenda-tions produce ARM, DHA, and LUM exposures in childrenthat are similar to those in adults with malaria. However, smallchildren weighing 12.5 to 15 kg are at risk of being underdosed,and AL doses could be doubled without exceeding the current,weight-based, maximum mg/kg dose in this patient group.

ACKNOWLEDGMENTS

We thank the children and their parents/guardians for their partic-ipation. We are most grateful to Valsi Kurian and the staff of Alex-ishafen Health Centre for their kind cooperation during the study. Wealso thank Jovitha Lammey, Christine Kalopo, and Bernard (“Ben”)Maamu for clinical and/or logistical assistance and Harin Karunajeewafor assistance with protocol design.

The National Health and Medical Research Council (NHMRC) ofAustralia funded the study (grant 634343). T.M.E.D. is supported byan NHMRC Practitioner Fellowship.

REFERENCES

1. Abdulla, S., et al. 2008. Efficacy and safety of artemether-lumefantrine dis-persible tablets compared with crushed commercial tablets in African infantsand children with uncomplicated malaria: a randomised, single-blind, mul-ticentre trial. Lancet 372:1819–1827.

2. Ahn, J. E., M. O. Karlsson, A. Dunne, and T. M. Ludden. 2008. Likelihoodbased approaches to handling data below the quantification limit usingNONMEM VI. J. Pharmacokinet. Pharmacodyn. 35:401–421.

3. Anderson, B. J., and N. H. Holford. 2008. Mechanism-based concepts of size andmaturity in pharmacokinetics. Annu. Rev. Pharmacol. Toxicol. 48:303–332.

4. Ashley, E. A., et al. 2007. How much fat is necessary to optimize lumefantrineoral bioavailability? Trop. Med. Int. Health 12:195–200.

5. Ashley, E. A., et al. 2007. Pharmacokinetic study of artemether-lumefantrinegiven once daily for the treatment of uncomplicated multidrug-resistantfalciparum malaria. Trop. Med. Int. Health 12:201–208.

6. Barnes, K. I., et al. 2006. Sulfadoxine-pyrimethamine pharmacokinetics inmalaria: pediatric dosing implications. Clin. Pharmacol. Ther. 80:582–596.

7. Batty, K. T., et al. 1996. Selective high-performance liquid chromatographicdetermination of artesunate and alpha- and beta-dihydroartemisinin in patientswith falciparum malaria. J. Chromatogr. B Biomed. Appl. 677:345–350.

8. Bell, D. J., et al. 2011. Population pharmacokinetics of sulfadoxine and

pyrimethamine in Malawian children with malaria. Clin. Pharmacol. Ther.89:268–275.

9. Bergstrand, M., and M. O. Karlsson. 2009. Handling data below the limit ofquantification in mixed effect models. AAPS J. 11:371–380.

10. Bergstrand, M., E. Plan, M. C. Kjellsson, and M. O. Karlsson. 2007. Acomparison of methods for handling of data below the limit of quantificationin NONMEM VI, abstr. 1201. Abstr. 16th Annu. Meet. Popul. ApproachGroup Eur., Copenhagen, Denmark, 13 to 15 June 2007. http://www.page-meeting.org/pdf_assets/6427-Poster_PAGE2007_070607_final.pdf.

11. Bindschedler, M., et al. 2000. Cardiac effects of co-artemether (artemether/lumefantrine) and mefloquine given alone or in combination to healthyvolunteers. Eur. J. Clin. Pharmacol. 56:375–381.

12. Ezzet, F., R. Mull, and J. Karbwang. 1998. Population pharmacokinetics andtherapeutic response of CGP 56697 (artemether � benflumetol) in malariapatients. Br. J. Clin. Pharmacol. 46:553–561.

13. Ezzet, F., M. van Vugt, F. Nosten, S. Looareesuwan, and N. J. White. 2000.Pharmacokinetics and pharmacodynamics of lumefantrine (benflumetol) inacute falciparum malaria. Antimicrob. Agents Chemother. 44:697–704.

14. German, P., et al. 2009. Lopinavir/ritonavir affects pharmacokinetic exposureof artemether/lumefantrine in HIV-uninfected healthy volunteers. J. Acquir.Immune Defic. Syndr. 51:424–429.

15. Hatz, C., et al. 2008. Treatment of acute uncomplicated falciparum malariawith artemether-lumefantrine in nonimmune populations: a safety, efficacy,and pharmacokinetic study. Am. J. Trop. Med. Hyg. 78:241–247.

16. Hietala, S. F., et al. 2010. Population pharmacokinetics and pharmacody-namics of artemether and lumefantrine during combination treatment inchildren with uncomplicated falciparum malaria in Tanzania. Antimicrob.Agents Chemother. 54:4780–4788.

17. Hooker, A. C., C. E. Staatz, and M. O. Karlsson. 2007. Conditional weightedresiduals (CWRES): a model diagnostic for the FOCE method. Pharm. Res.24:2187–2197.

18. Karunajeewa, H. A., et al. 2008. A trial of combination antimalarial therapiesin children from Papua New Guinea. N. Engl. J. Med. 359:2545–2557.

19. Lefevre, G., et al. 2002. Interaction trial between artemether-lumefantrine(Riamet) and quinine in healthy subjects. J. Clin. Pharmacol. 42:1147–1158.

20. Lefevre, G., et al. 2002. Pharmacokinetics and electrocardiographic pharmaco-dynamics of artemether-lumefantrine (Riamet) with concomitant administra-tion of ketoconazole in healthy subjects. Br. J. Clin. Pharmacol. 54:485–492.

21. Lefevre, G., et al. 2001. A clinical and pharmacokinetic trial of six doses ofartemether-lumefantrine for multidrug-resistant Plasmodium falciparummalaria in Thailand. Am. J. Trop. Med. Hyg. 64:247–256.

22. Lefevre, G., and M. S. Thomsen. 1999. Clinical pharmacokinetics of arte-mether and lumefantrine (Riamet (R)). Clin. Drug Investig. 18:467–480.

23. Matuszewski, B. K., M. L. Constanzer, and C. M. Chavez-Eng. 2003. Strat-egies for the assessment of matrix effect in quantitative bioanalytical meth-ods based on HPLC-MS/MS. Anal. Chem. 75:3019–3030.

24. McGready, R., et al. 2006. The pharmacokinetics of artemether and lume-fantrine in pregnant women with uncomplicated falciparum malaria. Eur.J. Clin. Pharmacol. 62:1021–1031.

25. Mwesigwa, J., et al. 2010. Pharmacokinetics of artemether-lumefantrine andartesunate-amodiaquine in children in Kampala, Uganda. Antimicrob.Agents Chemother. 54:52–59.

26. Noedl, H., T. Allmendinger, S. Prajakwong, G. Wernsdorfer, and W. H.Wernsdorfer. 2001. Desbutyl-benflumetol, a novel antimalarial compound:in vitro activity in fresh isolates of Plasmodium falciparum from Thailand.Antimicrob. Agents Chemother. 45:2106–2109.

27. Price, R. N., et al. 2006. Molecular and pharmacological determinants of thetherapeutic response to artemether-lumefantrine in multidrug-resistant Plas-modium falciparum malaria. Clin. Infect. Dis. 42:1570–1577.

28. Starzengruber, P., et al. 2008. Interaction between lumefantrine andmonodesbutyl-benflumetol in Plasmodium falciparum in vitro. Wien. Klin.Wochenschr. 120:85–89.

29. Starzengruber, P., et al. 2007. Specific pharmacokinetic interaction betweenlumefantrine and monodesbutyl-benflumetol in Plasmodium falciparum.Wien. Klin. Wochenschr. 119:60–66.

30. Tarning, J., et al. 2009. Population pharmacokinetics of lumefantrine in preg-nant women treated with artemether-lumefantrine for uncomplicated Plasmo-dium falciparum malaria. Antimicrob. Agents Chemother. 53:3837–3846.

31. van Agtmael, M. A., S. Cheng-Qi, J. X. Qing, R. Mull, and C. J. van Boxtel.1999. Multiple dose pharmacokinetics of artemether in Chinese patients withuncomplicated falciparum malaria. Int. J. Antimicrob. Agents 12:151–158.

32. Wong, R. P. M., et al. 2011. Desbutyl-lumefantrine is a metabolite of lume-fantrine with potent in vitro antimalarial activity that may influence arte-mether-lumefantrine treatment outcome. Antimicrob. Agents Chemother.55:1194–1198.

33. World Health Organization. 2010. World malaria report. World HealthOrganization, Geneva, Switzerland.

34. World Health Organization, Communicable Diseases Cluster. 2000. Severefalciparum malaria. Trans. R. Soc. Trop. Med. Hyg. 94(Suppl. 1):S1–S90.

35. Yang, S., and J. Roger. 2010. Evaluations of Bayesian and maximum likeli-hood methods in PK models with below-quantification-limit data. Pharm.Stat. 9:313–330.


Pharmacokinetic Comparison of Two Piperaquine-ContainingArtemisinin Combination Therapies in Papua New Guinean Childrenwith Uncomplicated Malaria

Sam Salman,a Madhu Page-Sharp,b Kevin T. Batty,b Kaye Kose,c Susan Griffin,c Peter M. Siba,c Kenneth F. Ilett,a Ivo Mueller,c andTimothy M. E. Davisa

School of Medicine and Pharmacology, University of Western Australia, Fremantle Hospital, Fremantle, Western Australia, Australiaa; School of Pharmacy, Curtin Universityof Technology, Bentley, Australiab; and Papua New Guinea Institute of Medical Research, Madang, Papua New Guineac

Pharmacokinetic differences between piperaquine (PQ) base and PQ tetraphosphate were investigated in 34 Papua New Guineanchildren aged 5 to 10 years treated for uncomplicated malaria with artemisinin-PQ (ART-PQ) base or dihydroartemisinin-PQ(DHA-PQ) tetraphosphate. Twelve children received ART-PQ base (two daily doses of 3 mg of ART and 18 mg of PQ base asgranules/kg of body weight) as recommended by the manufacturer, with regular clinical assessment and blood sampling over 56days. PQ concentrations in plasma samples collected from 22 children of similar ages with malaria in a previously publishedpharmacokinetic study of DHA-PQ tetraphosphate (three daily doses of 2.5 mg of ART and 20 mg of PQ tetraphosphate as tab-lets/kg of body weight) were available for comparison. The disposition of ART was also assessed in the 12 children who receivedART-PQ base. Plasma PQ was assayed by high-performance liquid chromatography with UV detection, and ART was assayedusing liquid chromatography-mass spectrometry. Multicompartment pharmacokinetic models for PQ and ART were developedusing a population-based approach. ART-PQ base was well tolerated, and initial fever abatement and parasite clearance wereprompt. There were no differences between the two treatments in the values for the PQ area under the concentration-time curvefrom time zero to infinity (AUC0 –�), with medians of 49,451 (n � 12) and 44,556 (n � 22) �g · h/liter for ART-PQ base andDHA-PQ tetraphosphate, respectively. Recurrent parasitemia was associated with lower PQ exposure. Using a two-compartmentART model, the median AUC0 –� was 1,652 �g · h/liter. There was evidence of autoinduction of ART metabolism (relative bio-availability for the second dose, 0.27). These and previously published data suggest that a 3-day ART-PQ base regimen should befurther evaluated, in line with World Health Organization recommendations for all artemisinin combination therapies.

The most recent World Health Organization (WHO) recom-mendations for the treatment of uncomplicated malaria include

a 3-day course of dihydroartemisinin plus piperaquine (DHA-PQ) asa first-line artemisinin (ART) combination therapy (ACT) (48). Var-ious formulations of DHA-PQ are marketed in tropical countries(Duo-cotecxin, Combimal, and P-Alaxin; http://www.actwatch.info/resources/drugs_home03_search.asp) or are in development (Eura-rtesim) (20), and all employ PQ tetraphosphate as the DHA partnerdrug. DHA is a semisynthetic derivative of ART, and its productionadds to the manufacturing cost, but, unlike ART, it does not exhibitautoinduction of metabolism. In addition, although the tetraphos-phate salt of PQ has greater water solubility and therefore may havebetter oral bioavailability, incorporation of the lipid-soluble PQ baseshould also simplify production.

Artequick (Artepharm Co. Ltd., Guangzhou, China) is an ACTthat contains ART in place of DHA and PQ base rather than PQtetraphosphate. This combination is formulated as tablets but alsoas granules for pediatric use. It is marketed in Cambodia and somesub-Saharan African countries. The current manufacturer’s rec-ommendation is for Artequick to be given as a 2-day regimen,which contrasts with the 3 days recommended for all ACTs by theWHO (48). Although the tolerability, safety, efficacy, and phar-macokinetics (PK) properties of DHA-PQ tetraphosphate havebeen widely investigated in children and adults (27, 28, 36, 44, 49),there are limited data relating to the efficacy and tolerability ofART-PQ base (37, 46) and no studies of the pharmacokinetics ofthis novel combination in malaria-infected patients. Concernshave been raised regarding possible underdosing in children for a

number of antimalarial drugs (8, 33), including PQ (27, 36, 49).Although children have been included in studies of PQ pharma-cokinetics (28, 44), only one pharmacokinetic study of ART hasspecifically enrolled pediatric patients (40).

We have evaluated the population pharmacokinetics ofART-PQ base (Artequick) in children from Papua New Guinea(PNG) with uncomplicated malaria and compared the data withthose of a previously published study of DHA-PQ tetraphosphate(Duo-cotecxin; Beijing Holley-Cotec, Beijing, China) in the samecategory of patients (29). The primary aims of the present studywere to investigate pharmacokinetic differences between PQ baseand PQ tetraphosphate and to describe the population pharma-cokinetics of ART in PNG children. Secondary aims were to pro-vide preliminary data relating pharmacokinetic factors to recur-rent parasitemia and to use both pharmacokinetic and efficacydata to suggest improved dose regimens for these combinations.

Received 28 November 2011 Returned for modification 19 January 2012Accepted 26 March 2012

Published ahead of print 2 April 2012

Address correspondence to Timothy M. E. Davis, [email protected].

Copyright © 2012, American Society for Microbiology. All Rights Reserved.

doi:10.1128/AAC.06232-11

3288 aac.asm.org Antimicrobial Agents and Chemotherapy p. 3288–3297 June 2012 Volume 56 Number 6

on May 13, 2012 by guest

http://aac.asm.org/

Dow

nloaded from

MATERIALS AND METHODSPatients. Assessment and recruitment of children for the present andpublished DHA-PQ tetraphosphate studies were as described previously(29). Briefly, all subjects were children aged 5 to 10 years presenting toAlexishafen Health Centre, Madang Province, on the north coast of PNG.The clinic serves an area where Plasmodium falciparum and P. vivax arehyperendemic and where P. ovale and P. malariae are also transmitted(13). Children with an axillary temperature � 37.5°C or a history of feverin the previous 24 h were screened with a Giemsa-stained thick blood filmread on site by a trained microscopist. Those with a monoinfection of P.falciparum (�1,000 asexual parasites �l of whole blood) or of P. vivax, P.ovale, or P. malariae (�250 asexual parasites �l of whole blood) wereeligible provided that the child’s parents gave informed consent, therewere no features of severe malaria (47), they had not taken any antima-larial drug in the previous 14 days, there was no evidence of another causeof fever, and there were no features of malnutrition or other chroniccomorbidity. Although the locations, populations, and enrollment proce-dures used in the two studies were identical, the DHA-PQ group wasenrolled between August 2005 and January 2006 whereas the ART-PQbase group was enrolled from March 2008 to May 2008. The study wasapproved by the PNG Institute of Medical Research Institutional ReviewBoard and the Medical Research Advisory Committee of the PNG Depart-ment of Health.

Clinical methods. In the present study of ART-PQ base, a standard-ized history was taken and a clinical examination was performed. A 3-mlvenous blood sample was taken for baseline blood film microscopy, forhemoglobin and blood glucose, and for subsequent drug assays of sepa-rated plasma. Each child was treated with granules of ART-PQ base (Arte-quick) according to body weight (approximately 3 mg of ART and 18 mgof PQ base/kg of body weight/day). This dose was repeated at 24 h, asrecommended by the manufacturer, with the exact time of each doserecorded. All doses were given under direct observation. The full contentsof each sachet were mixed with at least 50 ml of cow’s milk (equivalent to2 g of fat), as fat has been reported to increase the bioavailability of PQtetraphosphate (25, 41). The volume of milk used was based on previousexperience with its palatability and association with nausea in PNG chil-dren as well as on the amount of fat found to maximize the absorption oflumefantrine, another highly lipophilic antimalarial drug, in healthyadults (5).

Further venous blood samples were taken from an indwelling intrave-nous catheter at 1, 2, 4, 12, 24, 28, 36, and 48 h and then by venesection ondays 3, 5, 7, 14, 28, 42, and 56. All samples were centrifuged promptly andred cells and separated plasma stored frozen at �80°C until assayed. De-tailed clinical assessment, including a symptom questionnaire and deter-mination of blood film, hemoglobin, and blood glucose data, was re-peated on days 1, 2, 3, and 7, with additional clinical assessment anddetermination of blood film data on days 14, 28, 42, and 56. All bloodsmears taken at baseline and during follow-up were examined indepen-dently by at least two skilled microscopists in a central laboratory. Eachmicroscopist viewed 100 fields at �1,000 magnification before a slide wasclassified as negative. Any slide discrepant for positivity or negativity orfor species identification was referred to a third microscopist for adjudi-cation.

The clinical procedures followed for the DHA-PQ group have beenpreviously described (29) and were similar to those followed for theART-PQ base group. Differences in the previous study included (i) ad-ministration of 3 days of DHA-PQ tetraphosphate tablets at a dose of 2.5mg of ART and 20 mg of PQ tetraphosphate/kg of body weight daily(equivalent to 11.5 mg of PQ base/kg of body weight daily), (ii) drugadministration with water, and (iii) blood sampling and clinical follow-uponly until day 42.

Laboratory methods. A PQ tetraphosphate reference standard wasobtained from Yick-Vic Chemicals and Pharmaceuticals, Ltd. (HongKong, China). Chloroquine (CQ) diphosphate and authentic ART werefrom Sigma-Aldrich (St. Louis, MO), and artemether (ARM) was from

AAPIN Chemicals Ltd. (Abingdon, United Kingdom). Solid-phase ex-traction (SPE) Bond Elut PH columns were purchased from VarianInc. (Palo Alto, CA). High-performance-liquid-chromatography(HPLC)-grade methanol was obtained from Merck Pty. Ltd. (Kilsyth,Australia), and liquid-chromatography—mass-spectrometry (LC-MS)-grade ammonium formate was from Sigma-Aldrich (Gillingham,United Kingdom). All other solvents and chemicals were of analyticalgrade.

For the ART-PQ base group, PQ in plasma was analyzed by high-performance liquid chromatography as described for the originalDHA-PQ group (29) with minor modifications. Briefly, plasma wasspiked with CQ as an internal standard, alkalinized, and extracted into 8ml of hexane-isoamyl alcohol (99:1). Baseline samples were assayed forCQ prior to quantification of PQ to ensure no interference with the inter-nal standard. After centrifugation, the supernatant was back extractedinto 100 �l of 0.1 M HCl, aspirated, and recentrifuged. Aliquots of 80 �lwere injected into a Phenomenex C6-phenyl column (Phenomenex, Tor-rance, CA) with a mobile phase of 11% acetonitrile– 0.1 M phosphatebuffer (pH 2.5) pumped at 1 ml/min. Retention times were 2.5 and 7.3min for PQ and CQ, respectively, and PQ and CQ were detected at 340nm. The linear assay range was 2 to 1,000 �g/liter, and the intraday relativestandard deviations (RSDs) were 10.8%, 8.2%, and 9.4% and the interdayRSDs were 11.6%, 4.4%, and 6.7% at 5, 100, and 1,000 �g/liter, respec-tively. The limits of quantification and detection were 2 �g/liter and 1�g/liter, respectively.

For ART, the extraction procedure used a 1-ml C18 SPE column aspreviously described (9), with the following modifications. Briefly, theSPE column was preconditioned with 1 ml of methanol followed by 1 mlof 1 M acetic acid. Plasma samples (0.5 ml) were spiked with an internalstandard (ARM; 1,000 �g/liter), loaded onto the preconditioned SPE col-umn, and drawn through using a medium vacuum. The column was thenwashed with 1 M acetic acid (1 ml used in each of two successive washes)followed by 20% (vol/vol) methanol in 1 M acetic acid (1 ml). The columnwas dried under low-vacuum conditions for 30 min, and retained drugswere eluted with 2 ml of t-butyl chloride:ethyl acetate (80%:20% [vol/vol]). The eluate was evaporated in a vacuum evaporator at 35°C and thenreconstituted in 50 �l of the mobile phase, and 5-�l aliquots were injectedinto the LC-MS system.

The LC-MS system used was a single quadrupole mass spectrometer(model 2020; Shimadzu, Kyoto, Japan) consisting of a binary pump(model 20AD), vacuum degasser, thermostated autosampler (model SIL20ACHT), thermostated column compartment (model CTO 20A), pho-todiode detector (model SPD M 20A), and mass analyzer (model MS2020) with both electrospray ionization (ESI) and atmospheric pressureionization (APCI) systems. Analyses were performed in isocratic modewith a mobile phase of 20 mM ammonium formate (pH 4.8):methanol(20:80) pumped at a flow rate of 0.2 ml/min. Chromatographic separationwas undertaken at 30°C on a Synergy fusion-RP C18 column (150-mmlength by 2.0-mm inner diameter) coupled with a 5-�m-pore-size C18

guard column (Phenomenex, Lane Cove, Australia) (4-mm length by3-mm inner diameter). Retention times were 4.2 min and 7.5 min for ARTand ARM, respectively. Optimized mass spectra were acquired with aninterface voltage of 4.5 kV, a detector voltage of 1 kV, a heat block tem-perature of 400°C, and a desolvation gas temperature of 250°C. Nitrogenwas used as a nebulizer gas at a flow rate of 1.5 liter/min and a dry-gas flowrate of 10 liter/min.

Quantification was performed by selected ion monitoring (SIM), us-ing the DUIS mode, in which ACPI and ESI are used simultaneously. Allstandard curves were linear, with r2 � 0.999. Chromatographic data (peakarea ratio of ART/ARM) were processed using the LAB Solution softwarepackage (version 5; Shimadzu, Japan). Responses from analysis of samplescontaining three different ART concentrations (5, 200, and 2,000 �g/liter) and one ARM concentration (1,000 �g/liter) spiked into fiveseparate plasma samples were used to determine matrix effects (ionsuppression/enhancement), absolute recovery, and process efficiency

Piperaquine and Artemisinin Pharmacokinetics

June 2012 Volume 56 Number 6 aac.asm.org 3289


http://aac.asm.org/

Dow

nloaded from

(32), which were between 90% and 98%, 82% and 93%, and 86% and91%, respectively. The assay intraday RSDs were 9.3, 7.2, and 3.7% andinterday RSDs were 9.5, 7.1, and 6.5% at 5, 200, and 2,000 �g/liter,respectively. The limits of quantification and detection for ART were2.5 and 1 �g/liter, respectively.

Pharmacokinetic modeling. Loge plasma concentration-time datasets for PQ and ART were analyzed by nonlinear mixed effects modelingusing NONMEM (version 6.2.0; ICON Development Solutions, EllicottCity, MD) with an Intel Visual FORTRAN 10.0 compiler. The PQ plasmaconcentration-versus-time data from a previously published study ofDHA-PQ performed by our group, which were originally analyzed using apatient rather than a population approach (29), were pooled with the PQconcentration data from the present study. The first-order conditionalestimation (FOCE) with interaction estimation method was used. Theminimum values of the objective function (OFV) and conditionalweighted residuals (CWRES) plots were used to choose suitable modelsduring the model-building process. Allometric scaling was employed apriori, with volume terms multiplied by (WT/70)1.0 and clearance termsby (WT/70)0.75 (3), where WT represents total body weight. Residualvariability (RV) data were estimated as additive errors for the log-trans-formed data. Secondary pharmacokinetic parameters, including the areaunder the concentration-time curve from time zero to infinity (AUC0 –�)and elimination half-life (t1/2), were obtained for the participants frompost hoc Bayesian predictions using NONMEM and the final model pa-rameters. Base models were parameterized using ka (absorption rate con-stant), VC/F (central volume of distribution), CL/F (clearance), and VP/Fand Q/F (peripheral volumes of distribution[s] and their respective inter-compartmental clearance[s]).

For the PQ data set, two- and three-compartment models (ADVAN 4and 12) with first-order absorption with and without lag time were tested.Since inspection of the time-concentration curves indicated that there wassignificant variability in the absorption phase, a transit compartmentmodel was also tested (39). In this model, the dose passes through a seriesof transit compartments before entering the absorption compartment inorder to model the delay often associated with drug absorption. A singlerate constant (ktr) represents entry and exit for all transit compartments.Using a previously described implementation of the transit compart-ment model in NONMEM (39), the number of transit compartments(NN) and the mean transit time [MTT � (1 � NN)/ktr] were estimatedas continuous variables. For the ART data set, 1- and 2-compartmentmodels (ADVAN 2 and 4) with first-order absorption with and with-out lag time were evaluated. Once the structure of the models wasestablished, interindividual variability (IIV), interoccasion variability(IOV), and correlations between IIV terms were estimated, where sup-ported by the data.

As two different formulations of PQ with different water/lipid solubil-ities were used, potential differences in their relative levels of bioavailabil-ity were assessed. The difference in relative bioavailability levels betweenfirst and subsequent doses of PQ and ART was also investigated. For PQ,this was achieved by estimating the differences between the relative bio-availability levels of the first dose of PQ phosphate (fixed to 1) and the twodoses of PQ base as well as the two subsequent doses of PQ phosphate.Similarly, for ART, the relative bioavailability of the first dose was fixed to1 and potential differences between this and subsequent doses were as-sessed. The inclusion of an extra parameter to account for differences inrelative bioavailability was considered only if accompanied by a signifi-cant (�6.63; P � 0.01) fall in the OFV and an improvement in the CWRESplot. Differences in absorption parameters (ka, NN, and MTT) betweenthe two groups were also assessed within NONMEM. As described below,the effect size (percent) of the difference was estimated. To maintain theextra parameter estimating this difference, a significant (�6.63; P � 0.01)fall in the OFV was required. Differences between clearance and volumeterms for the two formulations were not assessed, as the idea of differencesbetween a salt and base formulation of the same drug is biologically im-plausible.

Finally, relationships between model parameters and the covariatesage, sex, log (baseline parasitemia), and fever were identified throughinspection of scatter plots and box plots of post-hoc values for individualsobtained from IIV distributions versus covariate and were subsequentlyevaluated within NONMEM. The effect size (percent) of categorical data(sex, fever) was assessed, while both linear and power relationships wereevaluated for continuous covariates (age, log [baseline parasitemia]).For effect size, the individual parameter value � population parametervalue � (1 � effect parameter � covariate value [0 or 1]). For linearrelationships, the individual parameter value � population parametervalue � [1 � effect parameter � (covariate value for individual/aver-age value of covariate)]. For power relationships, the individual pa-rameter value � population parameter value � [(covariate value forindividual/average value of covariate)effect parameter]. A stepwise for-ward inclusion and backward elimination method was used, with asignificance of P � 0.05 required for inclusion of a covariate relation-ship and P � 0.01 to retain a covariate relationship.

As CQ was used as the internal standard in the PQ assay, the potentialimpact of residual CQ in the plasma of the children on pharmacokineticparameters was assessed through simulation. A previous study in a similargroup of children resident in the same study area demonstrated thatapproximately 50% had a measurable plasma CQ concentration whenhospitalized (12). Using plasma CQ concentrations from a previouspharmacokinetic study of Madang children (29), we simulated condi-tions such that (i) half of the children had, at random, received atreatment course of CQ finishing 14 days prior to the study (just beforethe exclusion period for such treatment) and (ii) only children fromone of the treatment groups received CQ treatment 14 days prior to thestudy. The latter simulation represents the worst-case scenario interms of the effect of residual CQ on the comparative pharmacokineticproperties of the two PQ formulations through exogenous augmenta-tion of the internal standard.

Model evaluation. Initially, plots of observed versus individual andpopulation predicted values and of time versus CWRES were assessed. Abootstrap analysis using Perl-speaks-NONMEM with 1,000 samples wasperformed (for NQ, this was stratified according to dose regimen), andthe parameters derived from this analysis were summarized as themedian and 2.5th and 97.5th percentiles (with 95% empirical confi-dence intervals [CI]) to facilitate evaluation of final model parameterestimates. In addition, prediction-corrected visual predictive checks(pcVPCs) (11) and numerical predictive checks (NPCs) were per-formed with 1,000 data sets simulated from the final models, and thesewere stratified according to treatment group for PQ. The observed10th, 50th, and 90th percentiles were plotted with their respectivesimulated 95% CIs to assess the predictive performance of the model.NPCs were assessed by comparing the actual with the expected num-bers of data points within the 20%, 40%, 60%, 80%, 90%, and 95%prediction intervals (PI). These were also stratified according to treat-ment group for the PQ model.

Statistical analysis. Comparisons between the baseline characteristicsand secondary pharmacokinetic parameters of the subjects in theDHA-PQ and ART-PQ base studies were assessed using the Mann-Whit-ney U test for continuous variables and the Fisher exact test for categoricalvariables. A two-tailed level of significance of 0.05 was considered signif-icant for all comparisons.

RESULTSClinical characteristics and course. The baseline characteristicsfor all the children in the study are summarized in Table 1. Ofthose who received ART-PQ base, 11 had a monoinfection with P.falciparum and 1 had a monoinfection with P. vivax. One child waslost to follow-up after day 14. ART-PQ base treatment was welltolerated; reported symptoms were mild and short-lived (�2days) and consistent with clinical features of uncomplicated ma-laria. Initial fever clearance occurred in �24 h in all cases, and

Salman et al.

3290 aac.asm.org Antimicrobial Agents and Chemotherapy


http://aac.asm.org/

Dow

nloaded from

parasite clearance occurred in �48 h in all but one child, in whomit occurred within 72 h. The child with P. vivax at enrollmentcleared parasitemia promptly and remained slide negative for the56 days of follow-up. Of the 11 children with P. falciparum, 1developed slide-positive P. falciparum on day 28, another on day42, and 2 more by day 56. As PCR was not performed, it was notpossible to determine if these represented recrudescence or rein-fection. Only one child with P. falciparum at entry became slidepositive for P. vivax, on day 56. The mean hemoglobin concentra-tion, where data were available, increased as a result of treatmentregardless of malaria status during follow-up, with mean (95% CI)increases from baseline of 1.9 (0.40 to 3.3) (n � 9), 1.1 (0.15 to 2.5)(n � 11), and 1.5 (0.20 to 2.5) (n � 10) g/dl on days 28, 42, and 56,respectively (P � 0.027, P � 0.19, and P � 0.041). No cases ofhypoglycemia were recorded.

Pharmacokinetic modeling. There were 298 and 174 individ-ual plasma PQ concentrations available from the DHA-PQ (n �22) and ART-PQ base (n � 12) studies, respectively. No drugconcentrations were below the limit of quantification during the56-day follow-up period. A 3-comparment model fitted the databetter than a 2-compartment model, with a significant decrease inthe OFV (�OFV � �109.232, P � 0.001). Although the additionof a lag time improved the model significantly (�OFV � �31.059,P � 0.001), the absorption phase was poorly described, with first-order absorption determined with or without lag time. Therefore,a transit compartment model was tested where the number oftransit compartments (NN) and the mean transit time (MTT)through the transit compartments were estimated as continuousvariables. The transit compartment model was significantly betterthan a model with lag time, resulting in a 37.173-point reductionin the OFV (P � 0.001). Further testing of the combined data setswith models in which the absorption processes of the two formu-lations of PQ differed (for example, use of a lag-time model for PQbase and a transit compartment model for PQ tetraphosphate)were also tested and offered no advantage over the use of asingle-transit-compartment model. A three-compartmentmodel remained superior to a two-compartment model withthe use of a transit compartment absorption determination(�OFV � �57.937, P � 0.001).

The structural model parameters were ka, NN, MTT, VC/FPQ,

VP1/FPQ, VP2/FPQ, CL/FPQ, Q1/FPQ, and Q2/FPQ. There was poorprecision for the estimate of ka (RSE % � 100) as well as a high(�0.95) correlation between ka and MTT. Therefore, with thedata available in this study, these two parameters could not beestimated simultaneously and ka was set to the same value as ktr,i.e., equal to (1 � NN)/MTT. Interindividual variability was esti-mable for MTT, CL/FPQ, VC/FPQ, andVP1/FPQ. Correlation be-tween IIV terms was estimated for CL/FPQ and VC/FPQ and forVC/FPQ andVP1/FPQ. The IOV for FPQ was also estimable and wasaccompanied by significant falls in OFV (�OFV � �69.12, P �0.001) and RV (35% to 29%). There was no significant differencebetween the relative bioavailabilities of the two formulations orbetween the subsequent doses of PQ base or tetraphosphate andthe first dose. Although inspection of the concentration-timecurves appeared to indicate a difference between the two formu-lations in their absorption phases, when differences in NN andMTT were evaluated, they did not improve the model. Likewise,none of the tested covariates improved the model.

The impact of residual CQ proved to be minimal as assessedusing the simulations, with population pharmacokinetic parame-ter estimates differing by �9%. When all participants in the sameformulation group were presumed to have taken CQ 14 days priorto the start of the study, there was still no significant differencebetween the population pharmacokinetic parameter estimates forthe two PQ formulations.

The final model parameter estimates and the bootstrap resultsfor both PQ formulations are summarized in Table 2. Bias was�10% for all fixed and random model parameters. With the ex-ception of IIV in CL/FPQ, all parameters were reasonably well es-timated, with relative standard errors of �33%. The correlationbetween CL/FPQ and VC/FPQ displayed a wide 95% CI (�0.186 to0.710). Figures 1 and 2 show goodness-of-fit plots and pcVPCs,respectively. The pcVPCs showed wide 95% confidence intervalsfor the 10th, 50th, and 90th percentiles due to relatively smallnumbers of children. The actual 10th, 50th, and 90th percentilesfell into their respective 95% CI ranges for all time points for bothgroups. The stratified NPCs demonstrated good predictive perfor-mance, with the expected number of points above and below the20%, 40%, 60%, 80%, 90%, and 95% PIs. The half-life, totalAUC0 –�, and dose-adjusted AUC0 –� values are shown in Table 3.

TABLE 1 Baseline characteristics of study participants

Parameter

Duo-cotecxin values(29) (historical)(n � 22)a

Artequick values(present study)(n � 12)a P value

Age (yr) 6.9 � 1.4 7.1 � 1.5 0.790b

Sex (% male) 17 (86) 8 (66) 0.687c

Weight (kg) 19.1 � 3.8 18.3 � 3.1 0.986b

Axillary temp (°C) 37.2 � 1.2 36.3 � 0.7 0.034b

No. (%) with P. falciparum parasitemia 19 (86) 11 (92) 1.00c

Parasite density (per �l of whole blood) 13,360 [6,900–51,650] 26,270 [3,480–35,30] 0.736b

No. (%) with P. vivax parasitemia 2 (9.1) 1 (8) 1.00c

No. (%) with P. malariae parasitemia 1 (4.5) 0 (0) 1.00c

Hemoglobin (g/dl) 8.6 � 1.8 9.3 � 2.1 0.168b

Total PQ base dose (mg/kg) 35.3 � 4.4 38.3 � 5.8 0.136b

Total DHA dose (mg/kg) 7.7 � 1.0Total ART dose (mg/kg) 6.4 � 1.0a Data represent number (%), mean � SD, or median [interquartile range {IQR}], as indicated.b Mann-Whitney U test.c Fisher’s exact test.




http://aac.asm.org/

Dow

nloaded from

There were no significant differences between the two PQ com-pounds in any of these secondary parameters. The first distribu-tion, second distribution, and terminal elimination t1/2 values forall participants had median values of 4.5, 36.0, and 512 h, respec-tively. The median PQ AUC0 –� values for the Artequick and Duo-cotecxin formulations were 49,451 �g · h/liter and 44,556 �g ·h/liter, respectively.

Of the 96 ART drug concentrations (ART-PQ base group, n �12) that were available for analysis, six (6.25%) were below thelimit of quantification but above the limit of detection. As theserepresented a small proportion of the data, they were included attheir measured values. All 12 children has measurable levels ofART to 48 h. Initial modeling of the ART data set demonstratedthat a two-compartment model was significantly better than aone-compartment model (�OFV � �73.417, P � 0.001) and thatthe absorption phase was best represented by first-order absorp-tion without a lag time. Therefore, the structural model parame-ters were ka, VC/FART, VP/FART,CL/FART, and Q/FART. The IIV ofVC/FART was estimable, as was the IOV on FART. The data sup-ported the estimation of a relative bioavailability term for the sec-ond dose of ART (F2,ART), with its addition resulting in a signifi-cant fall in the OFV (�OFV � �24.029, P � 0.001). Thebioavailability of the second dose was 0.270 relative to the first. Nosignificant covariate relationships were identified.

The final model parameter estimates and the bootstrap resultsfor ART are summarized in Table 4. Bias was �10% for all fixedand random parameters. ka was not well estimated, with a relative

standard error of 55% and a 4-fold range in the nonparametric95% CI. Figures 3 and 4 show goodness-of-fit plots and pcVPCs,respectively. The pcVPC showed that all observed 10th, 50th, and90th percentile values were within their simulated 95% CIs. Dueto the small numbers used in the analysis, these CIs were wide andoverlapping. The NPC demonstrated good predictive perfor-mance, with the expected number of points above and below the20%, 40%, 60%, 80%, 90%, and 95% PIs. The t1/2 and the AUC0 –�

for each dose as well as the total AUC0 –� for the study participantsare shown in Table 5. The median distribution and terminal elim-ination t1/2 values were 1.55 and 7.43 h, while the median totalAUC0 –� value was 1,652 �g · h/liter.

Relationship between drug exposure and treatment out-come. In the eight children whose samples became slide positivefor P. falciparum by day 42 (two in the ART-PQ base group and sixin the DHA-PQ group), the AUC0 –� value for PQ was signifi-cantly lower than that seen with those who remained free of P.falciparum infection (median, 39,297 versus 49,776 �g · h/liter;P � 0.0060). Clearance and terminal elimination t1/2 values werenot significantly different; however, these children received alower total dose of PQ (median, 31.4 versus 35.7 mg/kg of PQbase; P � 0.11). When adjusted for dose, the differences in theAUC0 –� values were no longer significant between those childrenwith and without slide positivity results for P. falciparum by day 42

TABLE 2 Final population pharmacokinetic estimates and bootstrapresults for piperaquinea

ParameterMean(RSE %)

Bootstrap median[95% CI]

Structural and covariatemodel parameters

MTT (h) 1.27 (11) 1.25 [1.12–1.58]NN 4.20 (19) 3.70 [2.77–5.36]CL/FPQ (liter/h/70 kg) 40.1 (7) 40.7 [36.6–45.1]VC/FPQ (liter/70 kg) 2,580 (13) 2,550 [1,996–3,142]Q1/FPQ (liter/h/70 kg) 113 (21) 119.0 [84.3–166.0]VP1/FPQ (liter/70 kg) 2,760 (24) 3,440 [2,750–5,510]Q2/FPQ (liter/h/70 kg) 52.4 (15) 52.9 [43.8–67.1]VP2/FPQ (liter/70 kg) 21,600 (8) 22,300 [19,300–25,320]

Random model parametersIOV in FPQ (%) 46 (14) 42 [36–54]IIV in CL/FPQ (%) 16 (53) 16 [5–29]IIV in VC/FPQ (%) 53 (33) 45 [31–71]IIV in VP1/FPQ (%) 68 (32) 64 [16–93]IIV in MTT (%) 43 (13) 42 [34–52]Correlation coefficient

(CL/FPQ, VC/FPQ)0.33 0.272 [–0.186–0.710]

Correlation coefficient(VC/FPQ, VP1/FPQ)

0.85 0.874 [0.381–1.00]

Residual variability (%) 29 (5) 29 [27–32]a Parameters are NN (number of transit compartments), MTT (mean transit time),CL/FPQ (clearance), VC/FPQ (central volume of distribution), VP1/FPQ and VP2/FPQ

(peripheral volumes of distribution), Q1/FPQ and Q2/FPQ (intercompartmentalclearance between VP1/FPQ and VC/FPQ and between VP2/FPQ and VC/FPQ

respectively), and F1,Artequick (bioavailability of the first dose of Artequick relative to thefirst dose of Duo-cotecxin). RSE (relative standard error) values were calculated frombootstrap results. OFV in final model, �329.926; bootstrap OFV (median [95% CI]),�316.869 [�416.930 to 285.019].

FIG 1 (A) Population (Œ) and individual (�) predicted versus observedplasma piperaquine concentrations (in micrograms per liter on a log10 scale)for the final model. The line of identity is also shown. (B) Conditional weightedresiduals versus time (log scale) for piperaquine final model.

Salman et al.



http://aac.asm.org/

Dow

nloaded from

(median, 1.16 versus 1.42 �g · h/liter per mg/kg of PQ base; P �0.14). There was a significant positive correlation betweenAUC0 –� values and day 7 drug levels that did not reach signifi-cance (r � 0.70; 95%CI, 0.48 to 0.84; P � 0.001). Unlike AUC0 –�

values, day 7 levels were not significantly lower in those childrenwhose samples showed P. falciparum slide positivity (n � 8) com-pared to those whose samples did not (n � 26) (median, 44.1versus 48.0 �g/liter; P � 0.22). Similar results were evident whenthe two children from the DHA-PQ group whose samples devel-oped slide positivity for P. vivax by day 42 were included in theanalysis (data not shown). For the child who took �48 h to clear

initial parasitemia, the AUC0 –� values for ART and PQ werewithin the ranges of those of the other patients.

DISCUSSION

The development of ACTs has seen a variety of different combi-nations, formulations, and dose regimens introduced into clinicaluse without a detailed assessment of tolerability, safety, pharma-cokinetics, and efficacy. One recently marketed ACT, Artequick,appears to be relatively inexpensive to produce but uses compo-nent drugs that have not been investigated extensively, especiallyin a pediatric setting. The present pharmacokinetic and prelimi-

FIG 2 Visual predictive check showing observed 50th (�), 10th ({), and 90th (Œ) percentiles with simulated 95% CIs for the 50th (solid black line), 10th (graydotted lines), and 90th (dashed gray lines) percentiles for plasma piperaquine concentrations (micrograms per liter on a log10 scale) versus time (h) for Artequick(A) and Duo-cotecxin (B) from the final model. The observed data are superimposed as gray crosses. The inset shows data for the first 96 h.




http://aac.asm.org/

Dow

nloaded from

nary efficacy study in PNG children adds to the available data (30,37), suggesting that there would be benefits in extending the Arte-quick manufacturer’s recommended 2-day regimen to 3 days, asthis would increase PQ exposure and thus limit late recurrence ofparasitemia. However, the selection of a relatively low dose of ART(3 mg/kg of body weight versus the 10 to 20 mg/kg dose conven-tionally recommended), a drug that induces its own metabolism,may have implications for efficacy, especially in patients with lim-ited immunity to malaria or in geographical areas where artemis-inin resistance has started to develop (18).

Children in the DHA-PQ tetraphosphate group were given amean of 35.3 mg/kg PQ base over 3 days (29) compared with 38.3mg/kg PQ base over 2 days in the present children treated withART-PQ base. Overall, the exposures to PQ were similar for thetwo formulations, and no differences in the post hoc pharmacoki-netic parameters were identified. Although this suggests that thetablet and granule formulations have similar bioavailabilities and

that the small amount (2 g) of fat we administered with each dosewas unlikely to influence exposure to PQ, it is not possible todifferentiate the influences of food and formulation with the cur-rent study design. Two of three studies involving healthy adults

TABLE 3 Secondary pharmacokinetic parameters for piperaquinederived from post hoc Bayesian estimates for study participants and day7 plasma piperaquine levels

Parametera

Median [IQR]

P valuebPQ (Duo-cotecxin)n � 22

PQ (Artequick)n � 12

t½� (h) 4.44 [3.43–5.30] 4.52 [3.76–6.41] 0.48t½� (h) 36.1 [33.0–45.2] 35.3 [28.1–58.7] 0.82t½� (h) 513 [503–574] 512 [497–566] 0.82Day 7 plasma

piperaquine level(�g/liter)

39.3 [34.9–45.9] 42.0 [34.6–55.6] 0.56

AUC0–� (�g · h/liter) 49,451 [40,507–52,438] 44,556 [33,215–51,873] 0.36AUC0–� (�g · h/liter)/

total PQ dose(mg/kg)

1.27 [1.06–1.50] 1.37 [1.09–1.65] 0.40

a t½�, t½�, and t½� represent the first distribution half-live, second distribution half-live, and terminal elimination half-live, respectively.b Mann-Whitney U test.

TABLE 4 Final population pharmacokinetic estimates and bootstrapresults for artemisinin (n � 12)a

ParameterMean(RSE %)

Bootstrap median[95% CI]

Structural model parameterska (per h) 1.67 (55) 1.62 [1.01–4.40]CL/FART (liter/h/70 kg) 124 (12) 125 [99–157]VC/FART (liter/70 kg) 590 (30) 533 [318–874]Q/FART (liter/h/70 kg) 43.7 (38) 46.4 [19.5–79.4]VP/FART (liter/70 kg) 435 (26) 456 [259–696]

F2,ART � relative bioavailabilityof 2nd dose

0.270 (17) 0.275 [0.192–0.368]

Random model parametersIOV in FART (%) 43 (27) 39 [15–58]IIV in CL/FART (%) 12 (29) 12 [4–18]Residual variability (%) 33 (11) 32 [26–38]

a Parameters are ka (absorption rate constant), CL/FART (clearance), VC/FART (centralvolume of distribution), VP/FART (peripheral volume of distribution), Q/FART

(intercompartmental clearance between VP/FART and VC/FART), and F2,ART (relativebioavailability of 2nd dose of ART). RSE (relative standard error) values were calculatedfrom bootstrap results. OFV in final model, �63.562; bootstrap OFV (median [95%CI], �73.838 [�110.720 to 43.043].

FIG 3 (A) Population (Œ) and individual (�) predicted versus observedplasma artemisinin concentrations (micrograms per liter on a log10 scale) forthe final model. The line of identity is also shown. (B) Conditional weightedresiduals versus time for artemisinin final model.

FIG 4 Visual predictive check showing observed 50th (�), 10th (�), and 90th(Œ) percentiles with simulated 95% CIs for the 50th (solid black line), 10th(gray dotted lines), and 90th (dashed gray lines) percentiles for plasma arte-misinin concentrations (micrograms per liter on a log10 scale) versus time (h)from the final model. The observed data are superimposed as gray crosses.

Salman et al.



http://aac.asm.org/

Dow

nloaded from

found that fat-containing foods increased exposure to PQ tetra-phosphate (31, 34, 41), but the volunteers in those studies con-sumed relatively large quantities of fat (17 to 54 g). Consistentwith the present results, 6.4 g of fat did not increase the exposureto PQ tetraphosphate in adults with malaria (4). However, PQbase is less water soluble than PQ tetraphosphate, and exogenouslipids are known to increase the solubility of lipophilic drugs andthus affect the extent of absorption (35). PQ base may behavesimilarly to lumefantrine, another highly lipophilic drug, and mayrequire a smaller amount of fat to maximize absorption (5). Gran-ule formulations have been reported to increase bioavailabilityrelative to tablets (15), which is consistent with the increased sur-face area available for dissolution compared to tablets, but ourdata suggest that this is not a major effect in the case of Artequick.Future studies evaluating the effect of food and drug formulationon the disposition of PQ base in malaria could help refine doseregimens for therapies employing drugs such as Artequick.

A model with three compartments and a transit sequence priorto absorption best represented the PQ concentration-time dataset. Most previous studies have used a two-compartment model(28, 31, 44). One study in healthy adults found that, although athree-compartment model represented the postadministrationprofile better, there were insufficient data to support its use over atwo-compartment model (38). The mean elimination t1/2, a pa-rameter influenced by the duration of sampling (45), was 512 h, avalue within the previously reported range of 224 to 667 h (1, 14,25, 28, 31, 34, 38, 44). Since there was substantial variability in theabsorption phase of the plasma PQ concentration profile, a transitcompartment model was tested and proved better than simplerabsorption models that used lag time, as has been found in studiesof other drugs (39).

It has recently been suggested that children should be given ahigher dose of PQ than adults due to lower day 7 plasma concen-trations (36) and reduced efficacy (27, 36, 49). This is supportedby comparative pharmacokinetic studies in children and adultsthat found that children had a higher clearance (28) or a lower PQexposure at critical times during the illness (44). These concernshave also been raised for other antimalarial drugs (8, 33) andreflect pharmacokinetic effects due to the effects of body size, mat-uration, and organ function (3). Although only children aged be-tween 5 and 10 years were included in the present study, we foundthat recurrence of parasitemia was associated with a lower PQAUC resulting from a lower milligram/kilogram dose, consistentwith other studies of DHA-PQ tetraphosphate (16). As PCR wasnot performed, these cases may represent either recrudescence

(treatment failure) or reinfection (failure of posttreatment pro-phylaxis).

The dose-adjusted PQ exposures in our children were similarto that found in Caucasian and Vietnamese healthy adults (14, 38,41) and Thai adults with malaria (4) but were from three to sixtimes lower than those found in studies of Vietnamese and Chi-nese healthy adults, respectively (25, 31), suggesting that there arePQ pharmacokinetic differences between populations. Currentlyrecommended PQ tetraphosphate doses are 18 mg/kg/day (10 mg/kg/day PQ base) for 3 days (48). A higher average daily dose of PQbase in the Artequick group (19 mg/kg/day) was well toleratedwhen given for 2 days, and the same dose given on the third daymight both satisfy WHO recommendations for duration of ACTand address the issue of the need for higher milligram/kilogramdoses in children.

The use of CQ as an internal standard for PQ is potentiallyproblematic for samples taken from an area of malaria endemicitywhere CQ is widely available and used empirically for treatment offever. Utilizing CQ usage and pharmacokinetic data from otherstudies in children from the Madang area (12, 29), we investigatedwhether the 14-day exclusion for prior antimalarial treatment wassufficient to limit such potential confounding. Even in the worst-case scenario, there was only a small effect on the estimated PKparameters, and that effect did not produce falsely significant dif-ferences between the results obtained for the two formulations.Although this is reassuring, it would be best if future similar stud-ies employed an alternative internal standard.

A number of published studies have evaluated the pharmaco-kinetics of ART in healthy adults (6, 7, 10, 19, 23, 43) and adultswith malaria (2, 9, 21, 22, 24, 26, 42), but only one has includedchildren with malaria (40). In the latter Vietnamese study, 23 chil-dren aged 2 to 12 years were given 5 days of ART dosed accordingto body weight (approximately 10 mg) and 31 adults received 500mg of ART daily for 5 days. Sparse sampling was used to charac-terize ART population pharmacokinetics in plasma by the use ofNONMEM after the first and final doses, with two samples col-lected from each patient on day 1 and a single sample collected onday 5 from some patients. A one-compartment model was used,with clearance and volume terms for children and adults esti-mated separately. The median weights and ages of the childrenwere lower in the present study (18.3 versus 20 kg and 7.1 versus 9years, respectively). Although our value for ka was comparable tothat in the Vietnamese study (2.0/h versus 1.7/h), a two-compart-ment model provided a better fit in the present study, with distri-bution and elimination t1/2s of 1.9 h and 8.3 h, respectively, com-pared to a t1/2 of 1.8 h in the previous study (40). This differencemay reflect the longer sampling duration in the present study(24 h versus 8 h postdose), which enabled the identification of asecond exponential phase in the elimination of ART.

The elimination t1/2 of ART has been reported to be between1.4 and 4.8 h in noncompartmental (2, 6, 7, 9, 21, 22, 26, 42, 43)and compartmental (10, 19, 23, 24, 40) analyses. The present anal-ysis supports a biexponential disposition for ART, while most pre-vious compartmental analyses have reported a monoexponentialdisposition. A shorter sampling duration may be responsible forthis difference, as sampling was confined to �10 h after the lastdose in all but one of the studies reported to date. In addition,assay sensitivity may also contribute by limiting quantification ofART to those samples taken �12 h after dose administration (10,19). One study of healthy adults given a single dose of ART (6) also

TABLE 5 Secondary pharmacokinetic parameters for artemisininderived from post hoc Bayesian estimates for study participants

Parametera

ART (Artequick) values(median [IQR])(n � 12)

t½� (h) 1.55 [1.49–1.60]t½� (h) 7.43 [7.22–7.68]AUC (�g · h/liter) � first dose 1,347 [1,065–1,594]AUC (�g · h/liter) � second dose 312 [253–438]AUC0–� (�g · h/liter) � total 1,652 [1,333–2,177]a t½� and t½� represent the distribution half-life and terminal elimination half-life,respectively. AUCs for each dose were calculated using the standard pharmacokineticsformula to determine the relative contribution of each dose to the total AUC0 –� value.




http://aac.asm.org/

Dow

nloaded from

reported a biexponential disposition and found longer distribu-tion (2.61 h versus 1.55 h) and shorter elimination (4.34 h versus7.43 h) t1/2s compared to those determined in the present study.Although this study sampled blood to 24 h postdose, ART couldbe quantified in samples only up to 8 h.

The present median AUC0 –� of the first dose of ART (1,347�g · h/liter) was within the range reported for healthy adults(1,190 to 2,690 �g · h/liter) (6, 7, 10, 19, 43) but well below that ofadults with malaria (2,601 to 2,780 �g · h/liter) (2, 9, 26) whoreceived 500 mg of ART. Our children received a lower dose ofART (3.2 mg/kg/day), and, when adjusted for the relative doseadministered, the AUC0 –� for the first dose was above those seenin adults (2, 9, 26). The autoinduction of ART metabolism hasbeen well characterized, with a primary effect on the bioavailabil-ity of subsequent doses rather than on systemic clearance (23). It islikely, therefore, that this represents an increase in the activity ofgut wall rather than liver metabolism.

We found a difference in the PK of ART for the second dosethat was explained by a lower relative bioavailability of 0.27 com-pared to the first dose. In comparisons of the AUCs of differentdoses in previous studies, the relative bioavailability after 4 to 7days was between 0.13 and 0.29 (7, 9, 22, 26, 40, 43). One study ofAfrican adults with malaria who received 500 mg of ART daily for3 days and a single dose of mefloquine (42) measured ART insaliva and found that relative bioavailability was lower (0.45) onlyon the third day when mefloquine was given after the last dose ofART. However, when mefloquine was given on the first day, at thesame time as the first dose of ART, the relative bioavailabilities ofboth the second and third doses of ART were lower, at 0.23 and0.25, respectively.

Our data are in agreement with a rapid mean autoinductiontime of 1.9 h as estimated using a semiphysiological model forART (23), indicating that all doses after the first had a lower rela-tive bioavailability. If a third daily dose of ART-PQ base were to begiven, its relative bioavailability would also be low. The rapid ini-tial parasite clearance in Artequick-treated children seen in thepresent study despite relatively low and short-lived plasma ARTconcentrations may reflect the level of malarial immunity in thisgeographical area of intense transmission (17). It is likely thatrelatively low ART doses, even if given over 3 days rather than 2,would not be as effective where transmission and consequent im-munity are less or where artemisinin resistance has started to de-velop (18).

Compared to 3 days of DHA-PQ tetraphosphate administra-tion, the efficacy of 2 days of administration of Artequick in adultswas equivalent in one study (46) and inferior in another (37). A3-day Artequick regimen (3.2 and 16.0 mg/kg/day of ART and PQbase, respectively) has been found to be both well tolerated andmore effective than a 2-day regimen (30). Our preliminary datasuggest that the efficacy of 2 days of Artequick administrationappeared similar to that of 3 days of Duo-cotecxin administrationin PNG children. However, the weight of evidence from previousstudies (30, 37), the low proportion of ART in Artequick and itsautoinduction at a time when the specter of artemisinin resistancehas emerged (18), and the issue of potential PQ underdosing inchildren all support further evaluation of a theoretically more ef-ficacious 3-day Artequick regimen, as recommended by the WHOfor all ACTs (48).

ACKNOWLEDGMENTS

We are most grateful to Valsi Kurian and the staff of Alexishafen HealthCentre for their kind cooperation during the study. We also thank JovithaLammey, Christine Kalopo, and Bernard (“Ben”) Maamu for clinicaland/or logistic assistance. Harin Karunajeewa is acknowledged for hispivotal role in coordinating the original DHA/PQ tetraphosphate study.We thank Artepharm Co. Ltd. for kind provision of Artequick.

The National Health and Medical Research Council (NHMRC) ofAustralia funded the study (grant 634343). T.M.E.D. is supported by anNHMRC Practitioner Fellowship.

We have no conflicts of interest to declare.

REFERENCES1. Ahmed T, et al. 2008. Safety, tolerability, and single- and multiple-dose

pharmacokinetics of piperaquine phosphate in healthy subjects. J. Clin.Pharmacol. 48:166 –175.

2. Alin MH, Ashton M, Kihamia CM, Mtey GJ, Bjorkman A. 1996. Clinicalefficacy and pharmacokinetics of artemisinin monotherapy and in com-bination with mefloquine in patients with falciparum malaria. Br. J. Clin.Pharmacol. 41:587–592.

3. Anderson BJ, Holford NH. 2009. Mechanistic basis of using body sizeand maturation to predict clearance in humans. Drug Metab. Pharmaco-kinet. 24:25–36.

4. Annerberg A, et al. 2011. A small amount of fat does not affect piper-aquine exposure in patients with malaria. Antimicrob. Agents Chemother.55:3971–3976.

5. Ashley EA, et al. 2007. How much fat is necessary to optimize lumefan-trine oral bioavailability? Trop. Med. Int. Health 12:195–200.

6. Ashton M, et al. 1998. Artemisinin pharmacokinetics in healthy adultsafter 250, 500 and 1000 mg single oral doses. Biopharm. Drug Dispos.19:245–250.

7. Ashton M, et al. 1998. Artemisinin pharmacokinetics is time-dependentduring repeated oral administration in healthy male adults. Drug Metab.Dispos. 26:25–27.

8. Barnes KI, et al. 2006. Sulfadoxine-pyrimethamine pharmacokinetics inmalaria: pediatric dosing implications. Clin. Pharmacol. Ther. 80:582–596.

9. Batty KT, et al. 1996. Selective high-performance liquid chromatographicdetermination of artesunate and alpha- and beta-dihydroartemisinin inpatients with falciparum malaria. J. Chromatogr. B Biomed. Appl. 677:345–350.

10. Benakis A, Paris M, Loutan L, Plessas CT, Plessas ST. 1997. Pharma-cokinetics of artemisinin and artesunate after oral administration inhealthy volunteers. Am. J. Trop. Med. Hyg. 56:17–23.

11. Bergstrand M, Hooker AC, Wallin JE, Karlsson MO. 2011. Prediction-corrected visual predictive checks for diagnosing nonlinear mixed-effectsmodels. AAPS J. 13:143–151.

12. Bönsch C, et al. 2010. Chloroquine and its derivatives exacerbate B19V-associated anemia by promoting viral replication. PLoS Negl. Trop. Dis.4:e669.

13. Cattani JA, et al. 1986. The epidemiology of malaria in a populationsurrounding Madang, Papua New Guinea. Am. J. Trop. Med. Hyg.35:3–15.

14. Chinh NT, et al. 2009. Pharmacokinetics and bioequivalence evaluationof two fixed-dose tablet formulations of dihydroartemisinin and pipera-quine in Vietnamese subjects. Antimicrob. Agents Chemother. 53:828 –831.

15. Crauwels HM, et al. 2010. Relative bioavailability of a concept paediatricformulation of TMC278, an investigational NNRTI. Abstr. 18th Int. AIDSSoc. Conf., abstr THPE0158.

16. Denis MB, et al. 2002. Efficacy and safety of dihydroartemisinin-piperaquine (Artekin) in Cambodian children and adults with uncompli-cated falciparum malaria. Clin. Infect. Dis. 35:1469 –1476.

17. Djimdé AA, et al. 2003. Clearance of drug-resistant parasites as a modelfor protective immunity in Plasmodium falciparum malaria. Am. J. Trop.Med. Hyg. 69:558 –563.

18. Dondorp AM, et al. 2009. Artemisinin resistance in Plasmodium falcipa-rum malaria. N. Engl. J. Med. 361:455– 467.

19. Duc DD, et al. 1994. The pharmacokinetics of a single dose of artemisininin healthy Vietnamese subjects. Am. J. Trop. Med. Hyg. 51:785–790.

Salman et al.



http://aac.asm.org/

Dow

nloaded from

20. European Medicines Agency. 2011. Eurartesim. http://www.ema.europa.eu/docs/en_GB/document_library/Summary_of_opinion_-_Initial_authorisation/human/001199/WC500108010.pdf. Accessed September 1,2011.

21. Gordi T, Hai TN, Hoai NM, Thyberg M, Ashton M. 2000. Use of salivaand capillary blood samples as substitutes for venous blood sampling inpharmacokinetic investigations of artemisinin. Eur. J. Clin. Pharmacol.56:561–566.

22. Gordi T, Huong DX, Hai TN, Nieu NT, Ashton M. 2002. Artemisininpharmacokinetics and efficacy in uncomplicated-malaria patients treatedwith two different dosage regimens. Antimicrob. Agents Chemother. 46:1026 –1031.

23. Gordi T, et al. 2005. A semiphysiological pharmacokinetic model forartemisinin in healthy subjects incorporating autoinduction of metabo-lism and saturable first-pass hepatic extraction. Br. J. Clin. Pharmacol.59:189 –198.

24. Gordi T, Xie R, Jusko WJ. 2005. Semi-mechanistic pharmacokinetic/pharmacodynamic modelling of the antimalarial effect of artemisinin. Br.J. Clin. Pharmacol. 60:594 – 604.

25. Hai TN, Hietala SF, Van Huong N, Ashton M. 2008. The influence offood on the pharmacokinetics of piperaquine in healthy Vietnamese vol-unteers. Acta Trop. 107:145–149.

26. Hassan Alin M, Ashton M, Kihamia CM, Mtey GJ, Bjorkman A. 1996.Multiple dose pharmacokinetics of oral artemisinin and comparison of itsefficacy with that of oral artesunate in falciparum malaria patients. Trans.R. Soc. Trop. Med. Hyg. 90:61– 65.

27. Hasugian AR, et al. 2007. Dihydroartemisinin-piperaquine versus arte-sunate-amodiaquine: superior efficacy and posttreatment prophylaxisagainst multidrug-resistant Plasmodium falciparum and Plasmodiumvivax malaria. Clin. Infect. Dis. 44:1067–1074.

28. Hung TY, et al. 2004. Population pharmacokinetics of piperaquine inadults and children with uncomplicated falciparum or vivax malaria. Br. J.Clin. Pharmacol. 57:253–262.

29. Karunajeewa HA, et al. 2008. Pharmacokinetics and efficacy of pipera-quine and chloroquine in Melanesian children with uncomplicated ma-laria. Antimicrob. Agents Chemother. 52:237–243.

30. Krudsood S, et al. 2007. Dose ranging studies of new artemisinin-piperaquine fixed combinations compared to standard regimens of arte-misisnin combination therapies for acute uncomplicated falciparum ma-laria. Southeast Asian J. Trop. Med. Public Health 38:971–978.

31. Liu C, et al. 2007. Pharmacokinetics of piperaquine after single andmultiple oral administrations in healthy volunteers. Yakugaku Zasshi 127:1709 –1714.

32. Matuszewski BK, Constanzer ML, Chavez-Eng CM. 2003. Strategies forthe assessment of matrix effect in quantitative bioanalytical methodsbased on HPLC-MS/MS. Anal. Chem. 75:3019 –3030.

33. Mwesigwa J, et al. 2010. Pharmacokinetics of artemether-lumefantrineand artesunate-amodiaquine in children in Kampala, Uganda. Antimi-crob. Agents Chemother. 54:52–59.

34. Nguyen TC, et al. 2008. Pharmacokinetics of the antimalarial drug pip-

eraquine in healthy Vietnamese subjects. Am. J. Trop. Med. Hyg. 79:620 –623.

35. Porter CJ, Trevaskis NL, Charman WN. 2007. Lipids and lipid-basedformulations: optimizing the oral delivery of lipophilic drugs. Nat. Rev.Drug Discov. 6:231–248.

36. Price RN, et al. 2007. Clinical and pharmacological determinants of thetherapeutic response to dihydroartemisinin-piperaquine for drug-resistant malaria. Antimicrob. Agents Chemother. 51:4090 – 4097.

37. Pyar KP, Myint WW, Kyaw MP, Zin T, Than M. 2009. Efficacy andsafety of artemisinin-piperaquine (Artequick) compared to dihydroarte-misinin-piperaquine (Artekin) in uncomplicated falciparum malaria inadults. Myanmar Health Sci. Res. J. 21:78 – 82.

38. Röshammar D, Hai TN, Friberg Hietala S, Van Huong N, Ashton M.2006. Pharmacokinetics of piperaquine after repeated oral administrationof the antimalarial combination CV8 in 12 healthy male subjects. Eur. J.Clin. Pharmacol. 62:335–341.

39. Savic RM, Jonker DM, Kerbusch T, Karlsson MO. 2007. Implementa-tion of a transit compartment model for describing drug absorption inpharmacokinetic studies. J. Pharmacokinet. Pharmacodyn. 34:711–726.

40. Sidhu JS, et al. 1998. Artemisinin population pharmacokinetics in chil-dren and adults with uncomplicated falciparum malaria. Br. J. Clin. Phar-macol. 45:347–354.

41. Sim IK, Davis TM, Ilett KF. 2005. Effects of a high-fat meal on the relativeoral bioavailability of piperaquine. Antimicrob. Agents Chemother. 49:2407–2411.

42. Svensson US, Alin H, Karlsson MO, Bergqvist Y, Ashton M. 2002.Population pharmacokinetic and pharmacodynamic modelling of arte-misinin and mefloquine enantiomers in patients with falciparum malaria.Eur. J. Clin. Pharmacol. 58:339 –351.

43. Svensson US, et al. 1998. Artemisinin induces omeprazole metabolism inhuman beings. Clin. Pharmacol. Ther. 64:160 –167.

44. Tarning J, et al. 2008. Population pharmacokinetics of piperaquine aftertwo different treatment regimens with dihydroartemisinin-piperaquine inpatients with Plasmodium falciparum malaria in Thailand. Antimicrob.Agents Chemother. 52:1052–1061.

45. Tarning J, et al. 2005. Pitfalls in estimating piperaquine elimination.Antimicrob. Agents Chemother. 49:5127–5128.

46. Trung TN, Tan B, Van Phuc D, Song JP. 2009. A randomized, controlledtrial of artemisinin-piperaquine vs dihydroartemisinin-piperaquinephosphate in treatment of falciparum malaria. Chin. J. Integr. Med. 15:189 –192.

47. World Health Organization Communicable Diseases Cluster. 2000.Severe falciparum malaria. Trans. R. Soc. Trop. Med. Hyg. 94(Suppl. 1):S1–S90.

48. World Health Organization. 2010. Guidelines for the treatment of ma-laria— 2nd ed. World Health Organization, Geneva, Switzerland.

49. Zwang J, et al. 2009. Safety and efficacy of dihydroartemisinin-piperaquine in falciparum malaria: a prospective multi-centre individualpatient data analysis. PLoS One 4:e6358.




http://aac.asm.org/

Dow

nloaded from

Artemisinin-Naphthoquine Combination Therapy for UncomplicatedPediatric Malaria: a Pharmacokinetic Study

Kevin T. Batty,a,b Sam Salman,c Brioni R. Moore,c John Benjamin,d Sook Ting Lee,c Madhu Page-Sharp,a Nolene Pitus,d

Kenneth F. Ilett,c Ivo Mueller,e,f Francis W. Hombhanje,g Peter Siba,d and Timothy M. E. Davisc

School of Pharmacy, Curtin University, Bentley, Western Australia, Australiaa; Curtin Health Innovation Research Institute, Curtin University, Bentley, Western Australia,Australiab; School of Medicine and Pharmacology, University of Western Australia, Crawley, Western Australia, Australiac; Papua New Guinea Institute of Medical Research,Madang, Papua New Guinead; Infection and Immunity Division, Walter and Eliza Hall Institute of Medical Research, Victoria, Australiae; Center de Recerca en SalutInternacional de Barcelona (CRESIB), Barcelona, Spainf; and Centre for Health Research, Divine Word University, Madang, Papua New Guineag

Artemisinin-naphthoquine (ART-NQ) is a coformulated antimalarial therapy marketed as a single-dose treatment in Papua NewGuinea and other tropical countries. To build on limited knowledge of the pharmacokinetic properties of the components, espe-cially the tetra-aminoquinoline NQ, we studied ART-NQ disposition in Papua New Guinea children aged 5 to 12 years with un-complicated malaria, comparing a single dose (15 and 6 mg/kg of body weight) administered with water (group 1; n � 13), a sin-gle dose (22 and 9 mg/kg) with milk (group 2) (n � 17), and two daily doses of 22 and 9 mg/kg with water (group 3; n � 16). Theplasma NQ concentration was assayed by high-performance liquid chromatography, and the plasma ART concentration wasassayed using liquid chromatography-mass spectrometry. Population-based multicompartment pharmacokinetic models forNQ and ART were developed. NQ disposition was best characterized by a three-compartment model with a mean absorptionhalf-life (t1/2) of 1.0 h and predicted median maximum plasma concentrations that ranged as high as 57 �g/liter after the seconddose in group 3. The mean NQ elimination t1/2 was 22.8 days; clearance relative to bioavailability (CL/F) was 1.1 liters/h/kg; andvolume at steady state relative to bioavailability (Vss/F) was 710 liters/kg. Administration of NQ with fat (8.5 g; 615 kJ) versuswater was associated with 25% increased bioavailability. ART disposition was best characterized by a two-compartment modelwith a mean CL/F (4.1 liters/h/kg) and V/F (21 liters/kg) similar to those of previous studies. There was a 77% reduction in thebioavailability of the second ART dose (group 3). NQ has pharmacokinetic properties that confirm its potential as an artemisininpartner drug for treatment of uncomplicated pediatric malaria.

Available data relating to the pharmacokinetics of the antima-larial drug naphthoquine phosphate (NQ) are limited and

inconsistent. Initial reports suggested that NQ has a high oralbioavailability (�90%) and a half-life (t1/2) of 41 to 57 h (50). In amore recent study with healthy Chinese men in which NQ wasgiven alone or coformulated with artemisinin (ART) (41), theelimination t1/2 of NQ was substantially longer, at 250 to 300 h.This volunteer study also showed that the area under the concen-tration-time curve (AUC) for NQ exhibited an unusual relation-ship between the formulation and coadministered fat. The meanvalues were similar for the fasted group receiving NQ mono-therapy and the fed group receiving combination ART-NQ ther-apy but more than double this for the fasted volunteers given thefixed combination (41). The fact that the highest bioavailabilitywas in the fasting state appears in contrast to the effect of fat on theabsorption of related drugs, such as lumefantrine and piperaquine(3, 11, 24, 46), while the apparently beneficial effects of coformu-lation on bioavailability were difficult to explain (41).

It has been shown that NQ is a P-glycoprotein substrate andthat NQ efflux is saturable (12), suggesting that absorption couldbe nonlinear at high doses. However, the Chinese volunteer studyof ART-NQ found dose-proportional increases in the maximumconcentration in plasma (Cmax) and AUC for NQ at doses between200 and 600 mg (41). The maximum individual value for Cmax wasjust over 100 �g/liter in this study (41), but a Cmax as high as 245�g/liter has been reported after a 600-mg dose in adults (50).

The Chinese volunteer study of NQ and ART-NQ reported at1/2 for ART of 3.6 to 4.0 h, a clearance relative to bioavailability(CL/F) of approximately 1.5 liters/h/kg, and a volume of distribu-

tion relative to bioavailability (Vz/F) of 8 liters/kg (41). In con-trast, a number of previous studies with healthy adult volunteers(4, 16, 18, 26) and patients with uncomplicated Plasmodium fal-ciparum malaria (1, 5, 15, 25, 45) have reported lower mean valuesfor t1/2 (2.0 to 2.7 h [mean, 2.3 h]), higher mean values for CL/F(5.1 to 9.3 liters/h/kg [mean, 6.7 liters/h/kg]), and a higher meanV/F (16.4 to 35.5 liters/kg [mean, 27 liters/kg]). The reportedmean CL/F and V/F for ART in children were even greater, at 14.4liters/h/kg and 38 liters/kg, respectively (45). The Chinese studydid, however, show that the AUC for ART increased with coad-ministered fat (41), consistent with most past reports (14).

Because of inconsistencies between the few published studiesof NQ pharmacokinetics and a lack of pharmacokinetic data forchildren, we conducted two pharmacokinetic studies of ART-NQin children from Papua New Guinea with uncomplicated malaria.An initial pilot study (study 1), carried out before the manufac-turer had produced a pediatric dosing schedule and utilizing aconservative dose regimen (calculated in milligrams per kilogramof body weight based on the dose for adults), was designed toprovide preliminary pharmacokinetic data relating to NQ dispo-

Received 29 November 2011 Returned for modification 3 January 2012Accepted 1 February 2012

Published ahead of print 13 February 2012

Address correspondence to Timothy M. E. Davis, [email protected].

Copyright © 2012, American Society for Microbiology. All Rights Reserved.

doi:10.1128/AAC.06250-11

2472 aac.asm.org 0066-4804/12/$12.00 Antimicrobial Agents and Chemotherapy p. 2472–2484

sition in children, while the second study (study 2) aimed to char-acterize the pharmacokinetics of NQ as well as ART in more detailwhen these drugs were given at the manufacturer’s recommendeddose with fat (milk) or as a two-dose regimen.

MATERIALS AND METHODSPatients and clinical methods. Full details of the studies have been pro-vided in a separate report (7). In brief, children aged 5 to 10 years whopresented with an axillary temperature of �37.5°C or a history of fever inthe previous 24 h and who were slide positive for malaria (�1,000 asexualP. falciparum parasites/�l of whole blood or �250 Plasmodium vivax par-asites/�l) were eligible provided that they had no complications or con-comitant illness, no prior treatment with study drugs, and no history ofallergy to ART or aminoquinoline drugs. Each child’s parents or guard-ians gave written informed consent. Approvals were obtained from thePapua New Guinea Institute of Medical Research Institutional ReviewBoard and the Medical Research Advisory Committee of the Papua NewGuinea Health Department.

At enrollment, a history was taken and a full physical examination wasperformed. An intravenous cannula was inserted, and a baseline venousblood sample was drawn. In study 1, all children were administered ARCOtablets (50 mg NQ plus 125 mg ART; Kunming Pharmaceuticals, Kun-ming, China) orally as a single dose of 2 to 4 whole tablets with water(group 1). The dose was based on body weight according to those recom-mended by the manufacturer in milligrams per kilogram for adults (33)and represented dose ranges of 5.0 to 7.5 mg/kg for NQ phosphate and12.5 to 16.8 mg/kg for ART. Subjects were not required to fast prior to, orafter, treatment. If the child vomited within 1 h, the same dose was read-ministered and the time of readministration recorded.

In study 2, children were randomized by a computer-generated se-quence to receive ARCO tablets (50 mg NQ plus 125 mg ART) orally basedon body weight as recommended by the manufacturer for children (33) aseither (i) a single dose of 3 to 6 tablets given with 250 ml full-cream cow’smilk (containing 8.5 g fat) with dose ranges of 6.1 to 9.5 mg/kg for NQ and15.3 to 23.8 mg/kg for ART (group 2) or (ii) the same dose given withwater on two occasions 24 h apart (group 3). Each child was kept fastingunder observation, and if he/she vomited within 1 h, the same dose wasreadministered and the time of readministration recorded.

Group 1 patients had additional venous blood samples drawn throughthe cannula at 1, 2, 4, 8, 12, 18, 24, 48, and 72 h and by venesection at 4, 7,14, 28, 42, and 56 days. Blood was collected into lithium heparin tubes andwas centrifuged at 1,800 � g for 5 min, and the separated plasma wasstored at �80°C until analysis for the NQ concentration within 8 monthsof collection. These children were reassessed clinically at 4 and 24 h and ondays 2, 3, 7, 14, 28, and 56 (7). Group 2 and 3 patients had further 2.5-mlblood samples for drug assay taken at 1, 2, 4, 8, 12, 18, 24, 48, and 72 hthrough the sampling cannula and by venesection at 4, 7, 14, 28, and 42days. Group 3 patients received a second ART-NQ dose with water at 24 h.Posttreatment clinical and other monitoring for groups 2 and 3 was sim-ilar to that performed for group 1 (7).

Analytical methods. Naphthoquine diphosphate was obtained fromZYF Pharm Chemicals, Shanghai, China; tramadol hydrochloride andartemisinin were from Sigma-Aldrich Chemicals, St. Louis, MO; and arte-mether was from AApin Chemicals Ltd., Abingdon, Oxon, United King-dom. All general laboratory chemicals were of analytical grade (Sigma-Aldrich Chemicals, St. Louis, MO; Merck Chemicals, Darmstadt,Germany).

The concentration of NQ in plasma was analyzed using a validatedhigh-performance liquid chromatography (HPLC) assay, based on estab-lished analytical methods for chloroquine, piperaquine, and mefloquine(13, 30). Briefly, plasma samples (500 �l) were spiked with tramadol as theinternal standard (500 ng), alkalinized with a 2% (wt/vol) sodium tet-raborate solution (1 ml), and extracted into 8 ml hexane– ethyl acetate(80:20) by shaking for 10 min. The samples were then centrifuged at1,300 � g for 10 min. The supernatant (7.5 ml) was back-extracted into

0.1 ml of 0.1 M HCl by shaking for 5 min, followed by centrifugation asdescribed above. The HCl layer was transferred to 1.5-ml microcentrifugetubes and was recentrifuged at 1,300 � g for 25 min to evaporate traces oforganic solvent, after which 70 �l was injected onto the HPLC. Analyteswere separated on a Luna C18 HPLC column (length, 100 mm; innerdiameter [i.d.], 4.6 mm; particle size, 3 �m; Phenomenex, Australia) inseries with an octadecyl C18 guard column (length, 4 mm; i.d., 3 mm;Phenomenex, Australia) at 30°C with a mobile phase of 18% (vol/vol)acetonitrile in 50 mM KH2PO4 buffer (pH 2.5) pumped at 1 ml/min. Theapproximate retention times (tR) for NQ and tramadol were 9.4 min and6.8 min, respectively, and the analytes were detected by UV absorbance at222 nm (Fig. 1). The linear calibration range for each assay was 1 to 100�g/liter, and quality control (QC) samples (5 �g/liter, 20 �g/liter, and 100�g/liter) were included in each batch. The intraday relative standard de-viations (RSDs) of NQ were 8.9, 3.1, and 4.5% at 5 �g/liter, 20 �g/liter,and 100 �g/liter, respectively (n � 5), while interday RSDs were 7.7, 5.2,and 3.4% at 5 �g/liter, 20 �g/liter, and 100 �g/liter, respectively (n � 15).Intraday and interday accuracy ranges were 92 to 106% and 91 to 98%,respectively. The limit of quantification and limit of detection were 1�g/liter and 0.5 �g/liter, respectively. Mean levels of recovery of NQ fromplasma were 88%, 98%, and 96% at 5 �g/liter, 20 �g/liter, and 100 �g/liter, respectively.

The concentration of ART in plasma was analyzed using liquid chro-matography with mass spectrometry detection (LC-MS) based on an es-tablished assay for artemether (36). Stock solutions of ART and arte-mether (the internal standard) were prepared separately (1 g/liter inmethanol) and were stored in the dark at �80°C. Working standard so-lutions were prepared from the primary stock at 1, 10, and 100 mg/liter.Two sets of 5-point calibration curves were constructed (5 to 200 �g/literfor the lower concentrations and 200 to 2,000 �g/liter for the higher con-centrations) by spiking into blank plasma. Samples above the standardcurve were reanalyzed following appropriate dilution. QC samples wereprepared in blank plasma as described above at concentrations of 5, 200,and 1,000 �g/liter and were stored at �80°C prior to use for each batchanalyzed.

The extraction procedure used a 1-ml C18 solid-phase extraction(SPE) column (Bond Elut PH; Varian Inc., Palo Alto, CA) as describedpreviously (6), with minor modifications. Briefly, the SPE column waspreconditioned with 1 ml of methanol followed by 1 ml of 1 M acetic acid.Plasma samples (0.5 ml) were spiked with the internal standard (arte-mether; 1 �g), loaded onto the preconditioned SPE column, and drawnthrough by using a medium vacuum. The column was then washed with 1M acetic acid (1 ml; two washes), followed by 20% (vol/vol) methanol in1 M acetic acid (1 ml). The column was dried under a low vacuum for 30min, and retained drugs were eluted with 2 ml of t-butyl chloride– ethylacetate (80:20%, vol/vol). The eluate was evaporated in a vacuum evapo-rator at 35°C and was reconstituted in 50 �l of the mobile phase, and 5-�laliquots were injected into the LC-MS system.

The single-quadrupole LC-MS system (model 2020; Shimadzu,Kyoto, Japan) comprised a binary pump (20AD), a vacuum degasser, anautosampler with a thermostat (SIL-20AC HT), a column compartmentwith a thermostat (CTO 20A), a photodiode detector (SPD M 20A), and amass analyzer (MS 2020) with both electrospray ionization (ESI) andatmospheric pressure chemical ionization (APCI) systems. Analysis wasperformed in the isocratic mode with 20 mM ammonium formate (pH4.8)–methanol (20:80) at a flow rate of 0.2 ml/min. Chromatographicseparation was undertaken at 30°C on a Synergy Fusion-RP C18 column(length, 150 mm; i.d., 2 mm; particle size, 4 �m) coupled with a C18 guardcolumn (length, 4 mm; i.d., 2 mm; particle size, 5 �m; Phenomenex,Australia). The retention times were 4.3 min and 7.9 min for ART andartemether, respectively (Fig. 2). Optimized mass spectra were acquiredwith an interface voltage of 4.5 kV, a detector voltage of 1 kV, a heat blocktemperature of 400°C, and a desolvation gas temperature of 250°C. Nitro-gen was used as the nebulizer gas at a flow rate of 1.5 liter/min and a dry gasflow of 10 liter/min.

Naphthoquine and Artemisinin Pharmacokinetics

May 2012 Volume 56 Number 5 aac.asm.org 2473

Both ART and artemether standard solutions were first scannedfrom m/z 100 to 500 in ESI and APCI positive mode, as well as in thecombined ESI-and-APCI (DUIS) mode, to identify the abundance ofions corresponding to the respective drugs. The base peak intensities of

all three modes were compared and showed that the DUIS mode gavethe highest signal intensity. Therefore, quantitation was performed byselected ion monitoring (SIM) using the DUIS mode. For ART, theparent molecule [M � H]� (m/z 283) was used for quantitation, while

FIG 1 HPLC-UV (222 nm) chromatograms showing naphthoquine (N) (tR, 9.4 min) and the internal standard, tramadol (T) (tR, 6.8 min). (A) Spiked plasmaused in the calibration curve (20 �g/liter naphthoquine); (B) a patient’s blank predose sample (with the internal standard) showing no endogenous interference;(C) a typical sample (25 �g/liter naphthoquine).

Batty et al.


for artemether, the predominant fragmented ion (m/z 221) was mon-itored (44).

All standard curves were linear (r2, �0.999). Chromatographic data(peak area ratio of ART to artemether) were processed using LAB Solution(version 5; Shimadzu, Japan). Responses from the analysis of three ARTconcentrations (5, 200, and 2,000 �g/liter) spiked into five separate

plasma samples were used to determine matrix effects (ion suppression/enhancement), absolute recovery, and process efficiency (37, 43). Threesets of matrix solutions were prepared. Set 1 comprised blank plasmaspiked first and then extracted; set 2 comprised blank plasma extractedfirst and then spiked; and set 3 comprised pure solutions of the analyte.The matrix effect, process efficiency, and absolute recovery were ex-

FIG 2 LC-MS chromatograms showing artemisinin (ART) (tR, 4.3 min) and the internal standard (IS), artemether (tR, 7.9 min). (A) Spiked plasma used in thecalibration curve (200 �g/liter artemisinin); (B) a patient’s blank predose sample (with the IS) showing no endogenous interference; (C) a typical sample (136�g/liter artemisinin).



pressed as percentages. The matrix effect was calculated as (set 2 re-sponse � 100)/(set 3 response), the process efficiency as (set 1 response �100)/(set 3 response), and the absolute recovery as (set 1 response �100)/(set 2 response). The mean matrix effects � standard deviations(SD) for ART were 94% � 12% (range, 79 to 105%), 90% � 13% (range,75 to 106%), and 91% � 2% (range, 88 to 92%) at 5, 200, and 2,000�g/liter, respectively. The mean process efficiencies � SD for ART were93% � 18% (range, 73 to 121%), 84% � 11% (range, 75 to 102%), and82% � 4% (range, 80 to 89%) at 5, 200, and 2,000 �g/liter, respectively.The mean absolute recoveries � SD for ART were 88% � 10% (range, 78to 101%), 86% � 9% (range, 77 to 102%), and 90% � 7% (range, 87 to101%) at 5, 200, and 2,000 �g/liter, respectively. The mean matrix effect,process efficiency, and absolute recovery � SD for the internal standard,artemether, were 98% � 10% (range, 87 to 113%), 88% � 4% (range, 82to 92%), and 91% � 6% (range, 81 to 96%) at 1,000 �g/liter. The intradayRSDs for the assay were 9.3, 7.2, and 3.7% at 5, 200, and 2,000 �g/liter,respectively (n � 5), while the interday RSDs were 9.5, 7.1, and 6.5% at 5,200, and 2,000 �g/liter, respectively (n � 15). Interday accuracies deter-mined from the QC samples for each assay batch at 5, 200, and 1,000�g/liter were 108% � 7% (range, 86 to 114%), 103% � 6% (range, 93 to109%), and 107% � 8% (range, 86 to 115%), respectively (n � 16). Thelimits of quantification and detection for ART were 2.5 and 1 �g/liter,respectively.

Pharmacokinetic and statistical analyses. The pharmacokineticproperties of NQ were assessed using noncompartmental analysis (Ki-netica, version 4.4.1; Thermo LabSystems Inc., Philadelphia, PA) forgroup 1 subjects, and the data (not shown) were used to refine the studydesign for groups 2 and 3. All NQ data were subsequently pooled andanalyzed by population pharmacokinetic methods, as were ART data,which were available for groups 2 and 3.

In the population pharmacokinetic analysis, loge concentration-timedata sets for NQ and ART were analyzed by nonlinear mixed-effect mod-eling using NONMEM (version 6.2.0; Icon Development Solutions, Elli-cott City, MD) with an Intel Visual Fortran (version 10.0) compiler. NQdata were available for all three groups, while ART data were availableonly for groups 2 and 3. Linear mammillary model subroutines withinNONMEM, first-order conditional estimation (FOCE) with �-� interac-tion, and the objective function value (OFV; a NONMEM-calculatedglobal goodness-of-fit indicator equal to �2 times the log of the likeli-hood) were used to construct and compare plausible models. Unless oth-erwise specified, a difference in the OFV of �3.84 (�2 distribution with 1df; P � 0.05) was considered statistically significant. Secondary pharma-cokinetic parameters, including the volume of distribution at steady-state(Vss; calculated as the sum of volumes of individual compartments), thearea under the curve from 0 h to infinity (AUC0 –�), and the eliminationt1/2 for the participants, were obtained from post hoc Bayesian predictionin NONMEM using the final model parameters. Macro constants for thethree-compartment model were calculated from the modeled parametersusing previously published equations (49). Cmax and the time to Cmax

(Tmax) were estimated by predicting the concentrations of NQ and ARTfor each individual at 6-min intervals to capture the postdose peak.

Allometric scaling to a standard adult body weight (WT) was used apriori with all volume terms scaled using �(WT/70)1.0 and all clearanceterms scaled using �(WT/70)0.75 (2). Between-subject variability (BSV)was added to parameters for which it could be estimated from the avail-able data. An additive error model was used for residual unexplainedvariability (RUV), approximating proportional error as loge concentra-tion data were used. In the development of the final models, the influenceof the following covariates on the various model parameters was investi-gated: dose group, dose occasion, relative dose (in milligrams per kilo-gram), gender, spleen grade, malaria status (by slide positivity), baselinelog10 parasitemia, age, fever, and initial hemoglobin concentration. Covariaterelationships identified using the generalized additive modeling procedurewithin Xpose (29) and by inspection of correlation plots of � versus a cova-riate were evaluated within NONMEM. The potential effect of these cova-riates, particularly the dose group and occasion, on bioavailability wasalso considered in cases where similar relationships was identified for allvolume and clearance terms, given that these were relative to bioavailabil-ity. The effect size (expressed as a percentage) of categorical data wasassessed, while both linear and power relationships were evaluated forcontinuous covariates. Linear relationships were calculated as follows: indi-vidual parameter value � population parameter value � {1 � effect param-eter � [(covariate value for individual) � (median value of covariate)]}.Power relationships were calculated as follows: individual parametervalue � population parameter value � [(covariate value for individual)/(median value of covariate)effect parameter]. A stepwise forward inclusionand backward elimination method with a significance level (P value) of�0.05, accompanied by a decrease in the BSV of the parameter, was re-quired for the inclusion of a covariate relationship, and a P value of �0.01was required to retain a covariate relationship. For relationships involvingbioavailability, a decrease in the BSV of any volume or clearance wasrequired. Correlations among BSV terms were also investigated, and con-ditional weighted residuals (CWRES) plots were assessed in arriving at afinal model. Two- and three-compartment models for NQ and one- andtwo-compartment models for ART were compared with first-order ab-sorption, with and without a lag time.

A bootstrap procedure in Perl-speaks-NONMEM (PsN), stratified ac-cording to dose group and weight, was used to sample individuals fromthe original data set and to generate 1,000 new data sets that were subse-quently analyzed using NONMEM. The resulting parameters were thensummarized as the median and 2.5th and 97.5th percentiles (95% empir-ical confidence interval [CI]) to facilitate evaluation of the final modelparameter estimates. In addition, prediction-corrected visual predictivechecks (pcVPCs) (9) were performed using PsN with 1,000 replicate datasets simulated from the original data set. The observed 10th, 50th, and90th percentiles were plotted with their respective simulated 95% confi-dence intervals to assess the predictive performance of the model (9)Because a number of covariate effects were found in the model-buildingprocess for NQ, numerical predictive checks (NPCs), stratified accordingto those covariates, were performed and were assessed by comparing theactual with the expected number of data points within the 20, 40, 60, 80,90, and 95% prediction intervals (PI).

TABLE 1 Demographic data for children given artemisinin-naphthoquine for the treatment of uncomplicated P. falciparum malaria

Characteristica Group 1 Group 2 Group 3

No. of children 13 17 16Gender 6 male, 7 female 11 male, 6 female 12 male, 4 femaleAge (yr) 7.1 � 1.8 7.7 � 2.0 6.7 � 1.6Wt (kg) 18.0 � 3.7 18.9 � 5.2 16.8 � 3.2Ht (cm) 110 � 10 117 � 12 110 � 9Parasitemia (no. of parasites/�l of blood)b upon admission 14,757 (5,189–41,966) 6,674 (2,264–19,674) 29,416 (12,290–70,406)Naphthoquine dose (mg/kg) 6.3 � 0.9 8.8 � 1.4 2 � (9.5 � 0.9)Artemisinin dose (mg/kg) 15.7 � 2.3 22.0 � 3.6 2 � (23.8 � 2.2)a Data are means � SD unless otherwise indicated.b Geometric mean (95% confidence interval) for children with parasitemia.

Batty et al.


Data analysis and representation were performed using SigmaPlot,version 11 (Systat Software Inc., San Jose, CA). Data are means � SDunless otherwise indicated. Student’s t test (for parametric data) or theMann-Whitney U test (for nonparametric data) was used for two-samplecomparisons as appropriate with a significance level (P value) of �0.05.

RESULTS

Thirteen of 15 group 1 children completed all essential require-ments for the pharmacokinetic component of the study. All ofthese children had P. falciparum infections at baseline, and onehad a mixed infection with P. vivax at low density (160 para-sites/�l of blood). Four children in group 2 and four in group 3were considered to have low-grade parasitemia on screening mi-croscopy at the study site but were subsequently found to be slidenegative on confirmatory expert microscopy. All group 2 and 3children recruited were included in the pharmacokinetic study.Demographic data are summarized in Table 1.

The content of NQ in the ARCO tablets was determined bydissolving each tablet (n � 5) in 500 ml water by using sonication(twice, for 5 min each time) and measuring the concentrations in8 aliquots. The ART content was determined after dissolving eachtablet (n � 6) in 250 ml methanol and following the same proce-dure. The mean NQ and ART contents of the ARCO tablets usedin the study were 49 � 5 mg (nominal potency, 50 mg NQ) and129 � 3 mg (nominal potency, 125 mg ART), respectively.

Naphthoquine pharmacokinetics and pharmacodynamics.The plasma concentration-time profiles for NQ are shown in Fig.3. For pooled data from the three groups, a three-compartmentmodel proved superior to a two-compartment model, with alower OFV (�404.855 versus �388.736; P, �0.01) and no bias inthe CWRES plot in the initial stages of modeling. Because therewas no evidence of model misspecification by use of a three-com-partment model with first order-absorption with a lag time, more-complex models were not tested. The structural model parameters(where C refers to the central compartment and P1 and P2 to thetwo peripheral compartments) were the absorption rate constant(ka), lag time, CL/F, VC/F, VP1/F, VP2/F, and Q1/F and Q2/F (in-tercompartment clearances between VP1/F and VC/F and betweenVP2/F and VC/F, respectively). Estimates for the BSV of ka, VC, VP2,CL, and Q2 and the correlation between some BSV pairs (ka andVC/F, VC/F and CL/F, CL/F and VP2/F, and VP2/F and Q2/F) couldbe obtained (Table 2). Significant covariate relationships wereadded in the following order [given as covariate–parameter (rela-tionship type)]): fever–predicted F (negative categorical), firstdose for group 3–predicted F (negative categorical), and hemoglo-bin–VC/F (positive linear). Although children in group 2 wereestimated to have an approximately 50% lower ka, this relation-ship did not satisfy the significance requirements for inclusion inthe final model (0.01 � P � 0.05). Fever (axillary temperature,�37.3°C) and the first NQ dose in group 3 were associated with32% and 26% decreases in bioavailability, respectively. Every1-g/dl increase in the hemoglobin level increased VC/F by 16%.Slide positivity at baseline and log10 parasitemia were not signifi-cant covariates in the model. The residual error for the model was24% (Table 2).

Goodness-of-fit and CWRES plots for NQ are shown in Fig. 4.The results of the parameter estimates and the bootstrap resultsare summarized in Table 2, and post hoc Bayesian parameter esti-mates with derived secondary pharmacokinetic parameters aregiven in Table 3. The bootstrap demonstrated reasonable esti-

mates of structural and covariate effect parameters with a bias of�10% for all parameters except VP1, for which the bias was 19%.Random parameters had a bias of �7%. The AUC was signifi-cantly higher in group 3 (two doses) than in groups 1 and 2 (P �0.001) and was higher in group 2 than in group 1. The predictedCmax was �200 �g/liter for all children, apart from one group 3child with a value of 270 �g/liter after the second dose. When theAUC was normalized for the total relative dose (in milligrams per

FIG 3 Concentration-time plots for NQ in plasma for group 1 (A), group 2(milk) (B), and group 3 (water and double dose) (C) patients. (Insets) Plasmaconcentration-time data from 0 to 100 h after the dose.



kilogram), there was no longer any significant difference betweenthe groups. The pcVPC for NQ, shown in Fig. 5, demonstrate thereasonable predictive performance of the model. NPCs stratifiedaccording to dose group (three strata), hemoglobin (three strata),and fever (two strata) showed good predictive performance, withthe expected number of data points above and below most predic-tion intervals (data not shown).

Since the dose-corrected pharmacokinetic parameters wereconsistent across the three groups (Table 3), data were pooled toprovide estimates for the total of 46 patients. Overall, the mean �SD CL/F, Vss/F, t1/2�, t1/2�, and t1/2� for NQ were 1.30 � 0.45liters/h/kg, 805 � 256 liters/kg, 8.2 � 3.8 h, 98 � 16 h, and 518 �94 h, respectively.

Of the 13 (of 15) group 1 patients included in the pharmaco-kinetic analysis, 7 developed recurrent parasitemia during the 42-day follow-up period (7). One had a PCR-confirmed recrudes-cence of P. falciparum; four had reinfections with P. falciparum;and two had an emergence of P. vivax. In group 2, there was onlyone emergent P. vivax recurrence and no P. falciparum recurrence,while there were no episodes of slide positivity during follow-up ingroup 3. The AUC0-� and day 7 concentrations of NQ were sig-nificantly lower for the children with any parasitemia during fol-low-up than for those who remained free of malaria infection (P,0.001 and 0.005, respectively). However, the NQ dose in milli-grams per kilogram was also significantly lower (P, 0.001), and thedifference in AUC0-� was no longer significant when corrected fordose (P, 0.97), indicating that the lower dose, rather than individ-

ual pharmacokinetic differences, was responsible. Day 7 NQ con-centrations correlated significantly with AUC0-� overall (r, 0.91;P, �0.001) and in each of the three groups (r, �0.79; P, �0.001).

Artemisinin pharmacokinetics. Raw plasma ART concentra-tion-time data are presented in Fig. 6. A two-compartment modelwas superior to a one-compartment model for ART, with a lowerOFV (255.146 versus 122.637; P, �0.01) and an improvedCWRES. Since there was no evidence of model misspecification byuse of a two-compartment model with first-order absorption anda lag time, more-complex models were not tested. The structuralmodel parameters for ART were ka, lag time, CL/F, VC/F, VP/F,and Q/F (intercompartmental clearance for VP/F). Estimates ofthe BSV of CL, VC/F, ka, and lag time could be made, and a fullcovariance matrix was obtained. The correlation between CL/Fand V/F was �0.99 and was fixed at 1. Since the CWRES plotrevealed that plasma ART concentrations after the second dosewere lower than expected, the effect of dose occasion on observedF was tested as a negative categorical relationship. The addition ofthis relationship reduced the OFV by 46.626 (P, �0.001) and re-duced the residual error of the model by 7%. The second ART dosehad 77% lower bioavailability than the first. No other covariaterelationships were identified. The residual error in the final modelwas 51% (Table 4).

Goodness-of-fit and CWRES plots for ART are shown in Fig. 7.The results of the final parameter estimates and the bootstrapresults are summarized in Table 4, and post hoc Bayesian param-eter estimates with derived secondary pharmacokinetic parame-

TABLE 2 Population pharmacokinetic parameters and bootstrap results for NQ in children with uncomplicated P. falciparum malaria

Parameter Mean (RSEa [%]) in the final model Bootstrap median (95% CI)

Objective function value �687.786 �712.006 (�817.316 to �615.107)

Structural model parameterska (h�1) 1.1 (22) 1.0 (0.7 to 1.6)Lag time (h) 0.7 (7) 0.7 (0.6 to 0.8)VC/F (liters/70 kg) 12,500 (15) 12,200 (9,503 to 14,958)VP1/F (liters/70 kg) 15,500 (19) 17,000 (11,843 to 83,415)VP2/F (liters/70 kg) 17,200 (8) 16,000 (10,343 to 21,600)CL/F (liters/h/70 kg) 51.9 (6) 51.5 (30.1 to 58.7)Q1/F (liters/h/70 kg) 40.6 (9) 48.2 (24.2 to 113.0)Q2/F (liters/h/70 kg) 398 (13) 407 (318 to 536)

Covariate effect parameters (%)Decrease in predicted F with fever 31.8 (21) 31.8 (18.3 to 47.1)Decrease in predicted F with 1st dose in group 3 26.3 (33) 27.7 (9.3 to 40.9)Increase in VC/F per g/dl hemoglobin 16.4 (66) 14.9 (3.2 to 19.1)

Random model parametersBSV (%)

ka 104 (14) 103 (80 to 131)VC/F 77 (9) 77 (63 to 90)CL/F 32 (13) 31 (23 to 57)VP2/F 37 (17) 40 (25 to 59)Q2/F 52 (41) 50 (6 to 84)

Correlation coefficientka, VC/F 0.20 0.25 (�0.08 to 0.58)VC/F, CL/F 0.47 0.46 (0.04 to 0.72)CL/F, VP2/F 0.50 0.51 (0.09 to 0.86)VP2/F, Q2/F 0.20 0.18 (�0.62 to 0.91)

RUV (%) 24 (7) 24 (21 to 26)a RSE, relative standard error.

Batty et al.


ters are given in Table 5. The bootstrap demonstrated reasonable12/$12.00 Antimicrobial Agents and Chemotherapy p.2472–2484estimates of structural and random parameters with bi-ases of �8% and �10%, respectively, except for VP/F, where therewas a positive bias of 27%. No significant differences in secondaryparameters were found between the groups, although there was sub-stantial variability within each group. The pcVPC for ART is shown inFig. 8 and demonstrates the reasonable predictive performance of themodel.

The dose-corrected, first-dose data indicated that the medianAUC for ART was 5% higher in group 3 than group 2. However,these and other pharmacokinetic parameters for the two groupswere not significantly different (Table 5); hence, the data werepooled for total patient group estimates. Overall, mean (�SD)CL/F, Vss/F, and t1/2� for ART were 4.1 � 2.0 liters/h/kg, 21 � 10liters/kg, and 2.7 � 0.3 h, respectively. Although the best pharma-cokinetic model was a two-compartment model with a t1/2� of6.7 � 0.5 h, this may be a spurious finding, due to the limitedconcentration-time data in the present study design and the 27%bias in the bootstrap for VP/F.

DISCUSSION

The present study has provided the first pediatric pharmacoki-netic data for NQ and additional ART disposition data to comple-ment the few available for this age group. NQ given in the form ofART-NQ fixed combination therapy was promptly absorbed(mean absorption t1/2, 1.0 h) and reached a predicted Cmax of�200 �g/liter in all but one child even after the second dose ingroup 3. The mean elimination t1/2 of NQ (524 h) was longer thanestimates in early reports (41 to 57 h) (50) and in the recent Chi-nese adult volunteer study (156 to 299 h) (41). There was someevidence of a modest increase in NQ bioavailability when it wasadministered with a small amount of fat, in contrast to the sub-stantial food-associated reduction in NQ bioavailability in Chi-nese adults (41). The CL/F (1.1 liters/h/kg) and V/F (71 liters/kg)for NQ in our study were lower than the results reported forhealthy Chinese adults (7.0 liters/h/kg and 2,277 liters/kg) (41),but no other data are presently available for direct comparison. Inthe case of ART, the mean CL/F and V/F (4.1 liters/h/kg and 21liters/kg, respectively) were comparable to those in most previousstudies (means, 6.7 liters/h/kg and 27 liters/kg, respectively) (1, 4,5, 15, 16, 18, 25, 45).

The long elimination t1/2 and high V/F of NQ in our childrenwere consistent with those for most other quinolines and relateddrugs in clinical use (22, 30, 40). Pharmacokinetic modeling indi-cated that a three-compartment model best described the dispo-sition of NQ in the present study. This finding is consistent withsimilar pharmacokinetic studies involving chloroquine (21, 23,32, 52) and piperaquine (48). A number of studies of quinolineand related antimalarial drugs have shown biphasic drug concen-tration-time profiles that can be analyzed using a two-compart-ment model (10, 19, 27, 30, 39, 40, 46, 47, 53). Improved pharma-cokinetic study design, including more-frequent sampling oflonger duration, as well as lower limits of quantification for theanalytical techniques, may explain why recent studies such as oursreveal more-complex elimination kinetics. Indeed, the relativelyshort NQ elimination t1/2 in the Chinese volunteer study (41)could be explained by a short sampling period (216 h) as well as bythe use of noncompartmental methods. In relation to the latterpoint, we found an elimination t1/2 of 298 h by noncompartmentalmethods in group 1 patients versus 547 h in compartmental pop-ulation analyses of pooled NQ data.

The effect of fat on NQ bioavailability in the present studyneeds interpretation in light of the study design. In the prelimi-nary pharmacokinetic study in group 1 children, there was norequirement for fasting before or after drug administration. It islikely that these children consumed some fat around the timeART-NQ was given, even though the dose was administered withwater. Group 3 children, who were required to fast throughoutand were given the dose with water, had a 26% lower relativebioavailability than children in group 1 and also group 2, in whichART-NQ was administered with milk. This evidence of a modestpositive effect of fat on bioavailability contrasts with the observa-tion that the AUC and t1/2 of NQ were approximately 50% lowerafter food (60% lipid; 2,400 kJ) in healthy Chinese adults (41),suggesting an increased CL and/or reduced oral bioavailability.Studies with quinolines and related drugs have shown increasedabsorption with high-fat meals (3, 11, 46), but a standard Viet-namese meal (17 g fat; 2,000 kJ) had little effect on the pharmaco-kinetic properties of piperaquine (24). It is possible that relatively

FIG 4 (A) Population predicted (Œ) and individual predicted (�) versusobserved plasma NQ concentrations (�g/liter; log scale) for the final model.The line of identity is shown. (B) Conditional weight residuals versus time (logscale) for the final NQ model.



high fat meals such as that used by Qu et al. (41) might interferewith the absorption of NQ from the gastrointestinal tract, but ourexperience is that amounts of fat given in milk greater than thatused in the present study (�8.5 g; �615 kJ) have a high likelihoodof inducing significant nausea in an unwell child with malaria.This observation and our NQ pharmacokinetic data do not sug-gest that food-associated underdosing will be problematic in chil-

dren. The improved NQ bioavailability after the second dose rel-ative to that after the first dose in group 3 may relate to clinicalimprovement reflecting parasite clearance, as has been seen withlumefantrine (20).

Fever was independently associated with reduced NQ bioavail-ability, consistent with pharmacokinetic studies in other contexts(8, 35). There was an independent association between hemoglo-bin and VC that might suggest NQ accumulation in red blood cells,but assessment of partitioning was beyond the scope of the presentstudy. As is the case for a range of other drugs, including anti-infectives (51), coadministration of milk reduced the rate of NQabsorption.

Qu et al. (41) reported CL/F values (mean, 2.7 liters/h/kg) foradults given ART-NQ that were lower than those obtained in pre-vious studies of ART pharmacokinetics (1, 4, 5, 15, 16, 18, 25, 45)and with ART monotherapy, while Sidhu et al. (45) found a sig-nificantly higher value for this parameter (14.4 liter/h/kg) whenART was given to children with uncomplicated P. falciparum ma-laria. While the former observation is difficult to explain, an ap-parently high CL/F may relate to underestimation of the ARTAUC. Almost all previous studies have used noncompartmentalanalysis or one-compartment models to determine the pharma-cokinetic parameters for ART. A two-compartment model was,however, the best fit for the ART concentration-time data in thepresent study, probably reflecting the fact that our limits of quan-tification (2.5 �g/liter) and detection (1 �g/liter) were consider-ably lower than those in previous studies (4 to 20 �g/liter) (4, 15,16, 25, 45). A prolonged elimination phase may have been unde-tected in past studies, thus truncating the AUC. Our assay sensi-tivity led, in part, to an unexpected limitation of the present study,namely, a lack of sampling �24 h postdose. Based on the estab-lished pharmacokinetic properties of ART, we anticipated that

TABLE 3 Post hoc Bayesian parameter estimates and derived secondary pharmacokinetic parameters for NQ in children with uncomplicated P.falciparum malaria

Parametera Group 1 (n � 13) Group 2 (n � 17) Group 3 (n � 16)

ka (h�1)b 1.3 (0.9–1.6) 0.7 (0.4–1.0) 1.7 (0.6–2.2)CL/F (liters/h) 17.3 (15.3–21.8) 19.5 (16.2–25.1) 16.6 (14.9–19.2)VC/F (liters) 2,115 (1,735–2,753) 3,494 (1,817–7,818) 3,610 (1,383–7,030)VP1/F (liters) 3,986 (3,432–4,318) 3,986 (3,321–4,871) 3,543 (3,183–3,903)VP2/F (liters) 4,392 (3,602–5,208) 4,662 (3,816–5,347) 4,370 (3,633–4,760)Vss/F (liters) 10,464 (9,366–13,888) 13,161 (10,485–14,053) 12,001 (9,390–14,882)t½� (h)c 6.8 (4.4–9.2) 8.2 (5.7–9.7) 7.3 (5.5–12.0)t½� (h)c 109 (92–121) 115 (103–126) 118 (104–130)t½� (h)c 525 (490–544) 500 (455–629) 595 (525–624)AUC0–� (�g · h/liter)d 5,935 (4,776–6,551) 7,104 (5,954–7,914) 15,385 (13,200–18,486)AUC1/dose (�g · h/liter per mg/kg) 917 (822–1,158) 728 (611–1,004) 813 (629–999)Relative bioavailabilitye 1.00 (1.00–1.00) 1.00 (0.68–1.00) 0.75 (0.75–0.87)Observed day 7 level (�g/liter)d 7.0 (4.9–8.3) 8.1 (7.3–9.8) 17.9 (12.0–22.9)

Dose 1Predicted Cmax1 (�g/liter) 40.6 (32.6–45.5) 33.9 (14.7–52.7) 22.9 (14.1–49.1)Predicted Tmax1 (h) 3.1 (2.7–3.7) 4.6 (3.7–7.1) 3.3 (2.4–4.8)

Dose 2Predicted Cmax2 (�g/liter) 57.0 (42.2–138)Predicted Tmax2 (h) (h) 27.3 (26.7–28.3)

a Data are medians (interquartile ranges).b P, 0.053 and 0.094 for the comparison between groups 2 and 1 and groups 2 and 3, respectively.c t½�, t½�, and t½� are the first distribution, second distribution, and terminal elimination half-lives, respectively.d P, �0.01 for comparisons between groups 2 and 1 and groups 3 and 1.e P, �0.01 for comparison between groups 3 and 1.

FIG 5 Prediction-corrected VPC plots for NQ in children with uncompli-cated P. falciparum malaria, showing the observed 50th percentile (�) and the10th and 90th percentiles (Œ) with the simulated 95% CIs for the 50th percen-tile (solid black line) and the 10th and 90th percentiles (dashed gray lines).(Inset) Plasma concentration-time data from 0 to 100 h after the dose.

Batty et al.


plasma ART concentrations would not be detectable beyond 24 h.If prolonged sampling had been performed, it would have alloweda more definitive multicompartment pharmacokinetic character-ization.

In a Vietnamese study, coingestion of food was reported to beassociated with a nonsignificant 20% reduction in the ART AUCafter oral ART administration to healthy adults (16). In contrast,Qu et al. (41) reported that the AUC and t1/2 of ART were approx-imately 75% higher after coadministration of food with ART-NQcombination therapy, suggesting increased bioavailability and apossible reduction in CL. Our data are consistent with the earlierstudy of Dien et al. (16), in that we also found a nonsignificant 5%lower AUC for ART after ingestion of food (milk) compared withadministration with water. There were no significant differenceswhen the dose group was added as a covariate in the populationpharmacokinetic model, further evidence that fat has no clinicallymeaningful effect on the pharmacokinetics of ART.

Although the importance of developing pediatric formulationsof antimalarial drugs has been emphasized (34), it is not clear howthe manufacturer’s pediatric ART-NQ dose recommendationshave been developed. Using either a weight-based equation (28)(dose for a child [mg] � dose for an adult [mg] � [weight of achild/weight of an adult]0.75) or a body surface area (BSA) equa-

tion (28, 42) (dose for a child [mg] � dose for an adult [mg] �[BSA of a child/BSA of an adult]), where the regular adult dose ofNQ is 400 mg, adult weight is assumed to be 50 kg, and adult BSAis 1.73 m2, the adult dose of 8 mg/kg would scale up to �10 mg/kgin children. Our initial, conservative mean dose of 6.3 mg/kg NQfor group 1 as part of ART-NQ was associated with a relativelyhigh late-treatment failure rate (7). The regimens used for groups2 and 3 (means, 9.0 and 9.5 mg/kg NQ per dose) were based on themanufacturer’s recommendations of 6.5 to 9.5 mg/kg for childrenweighing as much as 40 kg (33, 38), doses that still fall short of theallometrically scaled dose of �10 mg/kg.

Efficacy against asexual parasite forms over 42 days of fol-low-up for groups 2 and 3 was 100%, but prolonged gametocytecarriage was observed in some patients (7). The latter observation,together with concerns regarding the emergence of artemisininresistance in areas of endemicity with a history of subtherapeuticdrug use (17), the implication that higher individual pediatricdoses than those recommended by the manufacturer can be used,and the safety of the two-dose ART-NQ regimen employed forgroup 3 (7), supports an argument for a 3-day ART-NQ regimenin line with WHO recommendations for all artemisinin combina-tion therapies (54). We have used the final NQ model to simulateCmax after three ART-NQ doses given with milk to 1,000 childrenwith characteristics similar to those of the present subjects. Themedian Cmax values (95% prediction intervals) after three consec-utive daily doses were 36 (19 to 76), 69 (44 to 128), and 89 (61 to152) �g/liter, respectively, with an absolute range up to 350 �g/liter after the third simulated dose. A predicted Cmax of �300

TABLE 4 Population pharmacokinetic parameters and bootstrap resultsfor ART in children with uncomplicated P. falciparum malaria

Parameter

Mean (RSEa

[%]) in thefinal model Bootstrap median (95% CI)

Objective function value 85.171 73.924 (�60.386–178.368)

Structural model parameterska (h�1) 1.8 (110) 1.8 (0.6–6.5)Lag time (h) 0.7 (31) 0.7 (0.4–0.9)VC/F (liters/70 kg) 1,160 (31) 1,140 (625–1,520)VP/F (liters/70 kg) 166 (37) 211 (96.1–1,270)CL/F (liters/h/70 kg) 178 (12) 176 (141–216)Q/F (liters/h/70 kg) 14.2 (103) 15.3 (6.6–52.3)

Covariate effect parameter: %decrease in predictedF with 2nd dose

77.0 (9) 78.6 (63.3–89.9)

Random model parametersBSV

CL/F (%) 57 (25) 56 (43–67)Lag time (%) 23 (169) 21 (5–57)ka (%) 139 (81) 141 (72–230)VC/F to CL/F (ratio) 0.995 (28) 0.989 (0.760–1.671)

Correlation coefficientCL/F, lag time 0.571 0.565 (0.201–0.997)CL/F, ka 0.0225 0.011 (�0.550–0.628)Lag time, ka �0.340 �0.343 (�0.956–0.319)CL/F, VC/F 1 Fixed

RUV (%) 51 (31) 50 (37–65)a RSE, relative standard error.

FIG 6 Plasma ART concentration-time plots for group 2 (milk) (A) andgroup 3 (water and double dose) (B) patients.



�g/liter occurred in a small minority of subjects in the simulation.The group 3 child with a predicted Cmax of 270 �g/liter had anuncomplicated clinical course in the present study, and a Cmax of245 �g/liter in an adult was not reported to be associated withtoxicity (50), but careful tolerability and safety monitoring wouldneed to be carried out if a three dose regimen were implemented.

A further argument for multiple-dose ART-NQ relates to thedisposition of the ART component. The conventional dosage reg-imen for orally administered ART of 10 to 20 mg/kg on the firstday, followed by 500 mg daily for 4 days (14), has been questioneddue to the autoinduction of ART metabolism that, as in the pres-ent study, progressively and substantially reduces the bioavailabil-ity of subsequent doses but does not increase CL (25). The 15- to24-mg/kg dose of ART used in the present study could, therefore,be an appropriate part of a 3-day ART-NQ regimen based on thesingle dose now recommended by the manufacturer.

The present study had limitations, in part because of the pres-ent paucity of pharmacokinetic and other data relating toART-NQ (especially when group 1 was recruited) but also becauseof the context of a pediatric study in the rural tropics. The sam-pling schedule could have included more time points after the

second dose in group 3, but relatively robust estimates for modelparameters could still be derived. It was unfortunate that no pureP. vivax malaria cases were recruited, but the fact that there wasonly one late P. vivax infection in groups 2 and 3 suggests that thelong NQ t1/2 helps prevent the emergence of this infection, whichis seen after other therapies for P. falciparum malaria in this area,including artemether-lumefantrine (31).

In conclusion, when normalized by body weight, the pharma-cokinetic parameters for ART in children are comparable to those

FIG 7 (A) Population predicted (Œ) and individual predicted (�) versusobserved plasma ART concentrations (�g/liter; log scale) for the final model.The line of identity is shown. (B) Conditional weight residuals versus time (logscale) for the final ART model.

TABLE 5 Post hoc Bayesian parameter estimates and derived secondarypharmacokinetic parameters for artemisinin in children withuncomplicated P. falciparum malariaa

Parameter Group 2 (n � 17) Group 3 (n � 16)

ka (h�1) 2.0 (0.7–3.5) 1.1 (0.8–4.1)CL/F (liters/h) 82.1 (76.2–74.8) 66.9 (61.5–62.4)VC/F (liters) 348 (246–449) 279 (165–324)Q/F (liters/h) 5.13 (4.47–5.96) 4.69 (4.33–5.05)VP/F (liters) 42.7 (35.6–52.2) 37.9 (34.1–41.8)Vss/F (liters) 388 (289–482) 315 (202–362)t½� (h) 2.8 (2.7–3.0) 2.7 (2.5–2.8)t½� (h) 6.8 (6.2–7.0) 6.6 (6.5–6.9)AUC1 (�g · h/liter) (dose 1) 5,127 (3,631–8,237) 6,770 (5,249-10,235)AUC1/dose (�g · h/liter per

mg/kg)267 (170–340) 281 (202–417)

AUC2 (�g · h/liter) (dose 2) 1,557 (1,207–2,354)AUC0–� (�g · h/liter) 8,327 (6,457–12,590)

Dose 1Predicted Cmax1 (�g/liter) 843 (522–1,353) 1,105 (736–1,398)Predicted Tmax1 (h) 2.1 (1.6–2.9) 2.5 (1.5–3.0)

Dose 2Predicted Cmax2 (�g/liter) 269 (179–345)Predicted Tmax2 (h) 26.6 (26.0–27.4)

a Data are medians (IQR). All between-group comparisons were statisticallynonsignificant.

FIG 8 Prediction-corrected VPC plots for ART in children with uncompli-cated P. falciparum malaria, showing the observed 50th percentile (�) and the10th and 90th percentiles (Œ) with the simulated 95% CIs for the 50th percen-tile (solid black line) and the 10th and 90th percentiles (dashed gray lines).

Batty et al.


obtained in most previous studies with adults, but CL/F washigher than that in data recently reported when ART-NQ wascoadministered to healthy adults (41). In contrast, CL/F and V/Ffor NQ were lower in the present study, and the terminal elimina-tion t1/2 was longer, at a mean of 21.8 days. Although the predictedbioavailability of the first dose of NQ was lower in a fasted state,this is unlikely to translate into clinically meaningful effects. Thepresent pharmacokinetic characterization, as well as associatedtolerability, safety, and preliminary efficacy data (7), may justifyusing the currently recommended single dose of ART-NQ for 3days for children with uncomplicated malaria.

ACKNOWLEDGMENTS

We thank the children and their parents/guardians for their participation.We are also most grateful to Sister Valsi Kurian and the staff of AlexishafenHealth Centre for their kind cooperation during the study and to MicheleSenn and the staff of the Papua New Guinea Institute of Medical Researchfor clinical and logistic assistance. Valuable technical support was pro-vided by Michael Boddy and John Hess, School of Pharmacy, Curtin Uni-versity.

This study was funded by the National Health and Medical ResearchCouncil (NHMRC) of Australia (grant 634343). S.T.L. was the recipient ofa Cranmore Undergraduate Scholarship through the Faculty of Medicine,Dentistry, and Health Science, University of Western Australia, andT.M.E.D. is supported by an NHMRC Practitioner Fellowship.

F.W.H. has received research funding from Kunming Pharmaceuti-cals, the manufacturer of ARCO.

REFERENCES1. Alin MH, Ashton M, Kihamia CM, Mtey GJ, Bjorkman A. 1996. Clinical

efficacy and pharmacokinetics of artemisinin monotherapy and in com-bination with mefloquine in patients with falciparum malaria. Br. J. Clin.Pharmacol. 41:587–592.

2. Anderson BJ, Holford NH. 2008. Mechanism-based concepts of size andmaturity in pharmacokinetics. Annu. Rev. Pharmacol. Toxicol. 48:303–332.

3. Ashley EA, et al. 2007. How much fat is necessary to optimize lumefan-trine oral bioavailability? Trop. Med. Int. Health 12:195–200.

4. Ashton M, et al. 1998. Artemisinin pharmacokinetics in healthy adultsafter 250, 500 and 1000 mg single oral doses. Biopharm. Drug Dispos.19:245–250.

5. Ashton M, et al. 1998. Artemisinin kinetics and dynamics during oral andrectal treatment of uncomplicated malaria. Clin. Pharmacol. Ther. 63:482– 493.

6. Batty KT, et al. 1996. Selective high-performance liquid chromato-graphic determination of artesunate and alpha- and beta-dihydroartemisinin in patients with falciparum malaria. J. Chromatogr. BBiomed. Appl. 677:345–350.

7. Benjamin J, et al. 2012. Artemisinin-naphthoquine combination therapyfor uncomplicated pediatric malaria: a tolerability, safety, and preliminaryefficacy study. Antimicrob. Agents Chemother. 56:2465–2471.

8. Beovic B, et al. 1999. Influence of fever on the pharmacokinetics ofciprofloxacin. Int. J. Antimicrob. Agents 11:81– 85.

9. Bergstrand M, Hooker AC, Wallin JE, Karlsson MO. 2011. Prediction-corrected visual predictive checks for diagnosing nonlinear mixed-effectsmodels. AAPS J. 13:143–151.

10. Boudreau EF, et al. 1990. Mefloquine kinetics in cured and recrudescentpatients with acute falciparum malaria and in healthy volunteers. Clin.Pharmacol. Ther. 48:399 – 409.

11. Crevoisier C, Handschin J, Barre J, Roumenov D, Kleinbloesem C.1997. Food increases the bioavailability of mefloquine. Eur. J. Clin. Phar-macol. 53:135–139.

12. Crowe A, Ilett KF, Karunajeewa HA, Batty KT, Davis TM. 2006. Role ofP glycoprotein in absorption of novel antimalarial drugs. Antimicrob.Agents Chemother. 50:3504 –3506.

13. Davis TM, et al. 2007. Assessment of the effect of mefloquine on artesu-nate pharmacokinetics in healthy male volunteers. Antimicrob. AgentsChemother. 51:1099 –1101.

14. de Vries PJ, Dien TK. 1996. Clinical pharmacology and therapeuticpotential of artemisinin and its derivatives in the treatment of malaria.Drugs 52:818 – 836.

15. De Vries PJ, et al. 1997. The pharmacokinetics of a single dose of arte-misinin in patients with uncomplicated falciparum malaria. Am. J. Trop.Med. Hyg. 56:503–507.

16. Dien TK, et al. 1997. Effect of food intake on pharmacokinetics of oralartemisinin in healthy Vietnamese subjects. Antimicrob. Agents Che-mother. 41:1069 –1072.

17. Dondorp AM, et al. 2010. Artemisinin resistance: current status andscenarios for containment. Nat. Rev. Microbiol. 8:272–280.

18. Duc DD, et al. 1994. The pharmacokinetics of a single dose of artemisininin healthy Vietnamese subjects. Am. J. Trop. Med. Hyg. 51:785–790.

19. Edwards G, et al. 1988. Pharmacokinetics of chloroquine in Thais: plasmaand red-cell concentrations following an intravenous infusion to healthysubjects and patients with Plasmodium vivax malaria. Br. J. Clin. Pharma-col. 25:477– 485.

20. Ezzet F, Mull R, Karbwang J. 1998. Population pharmacokinetics andtherapeutic response of CGP 56697 (artemether � benflumetol) in ma-laria patients. Br. J. Clin. Pharmacol. 46:553–561.

21. Frisk-Holmberg M, Bergqvist Y, Termond E, Domeij-Nyberg B. 1984.The single dose kinetics of chloroquine and its major metabolite deseth-ylchloroquine in healthy subjects. Eur. J. Clin. Pharmacol. 26:521–530.

22. German PI, Aweeka FT. 2008. Clinical pharmacology of artemisinin-based combination therapies. Clin. Pharmacokinet. 47:91–102.

23. Gustafsson LL, et al. 1983. Disposition of chloroquine in man after singleintravenous and oral doses. Br. J. Clin. Pharmacol. 15:471– 479.

24. Hai TN, Hietala SF, Van Huong N, Ashton M. 2008. The influence offood on the pharmacokinetics of piperaquine in healthy Vietnamese vol-unteers. Acta Trop. 107:145–149.

25. Hassan Alin M, Ashton M, Kihamia CM, Mtey GJ, Bjorkman A. 1996.Multiple dose pharmacokinetics of oral artemisinin and comparison of itsefficacy with that of oral artesunate in falciparum malaria patients. Trans.R. Soc. Trop. Med. Hyg. 90:61– 65.

26. Hien TT, et al. 2011. Orally formulated artemisinin in healthy fastingVietnamese male subjects: a randomized, four-sequence, open-label,pharmacokinetic crossover study. Clin. Ther. 33:644 – 654.

27. Hung TY, et al. 2004. Population pharmacokinetics of piperaquine inadults and children with uncomplicated falciparum or vivax malaria. Br. J.Clin. Pharmacol. 57:253–262.

28. Johnson TN. 2008. The problems in scaling adult drug doses to children.Arch. Dis. Child. 93:207–211.

29. Jonsson EN, Karlsson MO. 1999. Xpose—an S-PLUS based populationpharmacokinetic/pharmacodynamic model building aid for NONMEM.Comput. Methods Programs Biomed. 58:51– 64.

30. Karunajeewa HA, et al. 2008. Pharmacokinetics and efficacy of piper-aquine and chloroquine in Melanesian children with uncomplicated ma-laria. Antimicrob. Agents Chemother. 52:237–243.

31. Karunajeewa HA, et al. 2008. A trial of combination antimalarial thera-pies in children from Papua New Guinea. N. Engl. J. Med. 359:2545–2557.

32. Karunajeewa HA, et al. 2010. Pharmacokinetics of chloroquine andmonodesethylchloroquine in pregnancy. Antimicrob. Agents Chemother.54:1186 –1192.

33. Kunming Pharmaceutical Corporation. 2006. Instruction for use of com-pound naphthoquine phosphate tablets. Product information brochure.Kunming Pharmaceutical Corporation, Kunming, Yunnan Province,China.

34. Kurth F, et al. 2010. Do paediatric drug formulations of artemisinincombination therapies improve the treatment of children with malaria? Asystematic review and meta-analysis. Lancet Infect. Dis. 10:125–132.

35. Mackowiak PA. 1989. Influence of fever on pharmacokinetics. Rev. Infect.Dis. 11:804 – 807.

36. Manning L, et al. 2011. Meningeal inflammation increases artemetherconcentrations in cerebrospinal fluid in Papua New Guinean childrentreated with intramuscular artemether. Antimicrob. Agents Chemother.55:5027–5033.

37. Matuszewski BK, Constanzer ML, Chavez-Eng CM. 2003. Strategies forthe assessment of matrix effect in quantitative bioanalytical methodsbased on HPLC-MS/MS. Anal. Chem. 75:3019 –3030.

38. Nigerian-German Chemicals Plc. January 2012, accession date. ARCOproduct information. Nigerian-German Chemicals Plc, Lagos, Nigeria.www.ngcplc.com/arco/index.html.

39. Obua C, et al. 2008. Population pharmacokinetics of chloroquine and



sulfadoxine and treatment response in children with malaria: suggestionsfor an improved dose regimen. Br. J. Clin. Pharmacol. 65:493–501.

40. Price R, et al. 1999. Pharmacokinetics of mefloquine combined withartesunate in children with acute falciparum malaria. Antimicrob. AgentsChemother. 43:341–346.

41. Qu HY, et al. 2010. Single-dose safety, pharmacokinetics, and food effectsstudies of compound naphthoquine phosphate tablets in healthy volun-teers. J. Clin. Pharmacol. 50:1310 –1318.

42. Ritschel WA, Kearns GL. 2004. Handbook of basic pharmacokinetics, 6thed. American Pharmacists Association, Washington DC.

43. Rozet E, Marini RD, Ziemons E, Boulanger B, Hubert P. 2011. Ad-vances in validation, risk and uncertainty assessment of bioanalyticalmethods. J. Pharm. Biomed. Anal. 55:848 – 858.

44. Shi B, et al. 2006. Quantitative analysis of artemether and its metabolitedihydroartemisinin in human plasma by LC with tandem mass spectrom-etry. Chromatographia 64:523–530.

45. Sidhu JS, et al. 1998. Artemisinin population pharmacokinetics in chil-dren and adults with uncomplicated falciparum malaria. Br. J. Clin. Phar-macol. 45:347–354.

46. Sim IK, Davis TM, Ilett KF. 2005. Effects of a high-fat meal on the relativeoral bioavailability of piperaquine. Antimicrob. Agents Chemother. 49:2407–2411.

47. Tarning J, et al. 2008. Population pharmacokinetics of piperaquine aftertwo different treatment regimens with dihydroartemisinin-piperaquine inpatients with Plasmodium falciparum malaria in Thailand. Antimicrob.Agents Chemother. 52:1052–1061.

48. Tarning J, et al. 2005. Pitfalls in estimating piperaquine elimination.Antimicrob. Agents Chemother. 49:5127–5128.

49. Upton RN, Ludbrook GL. 2005. Pharmacokinetic-pharmacodynamicmodelling of the cardiovascular effects of drugs—method developmentand application to magnesium in sheep. BMC Pharmacol. 5:5.

50. Wang JY, et al. 2004. Naphthoquine phosphate and its combination withartemisinine. Acta Trop. 89:375–381.

51. Welling PG. 1996. Effects of food on drug absorption. Annu. Rev. Nutr.16:383– 415.

52. Wetsteyn JC, De Vries PJ, Oosterhuis B, Van Boxtel CJ. 1995. Thepharmacokinetics of three multiple dose regimens of chloroquine: impli-cations for malaria chemoprophylaxis. Br. J. Clin. Pharmacol. 39:696 –699.

53. White NJ, Watt G, Bergqvist Y, Njelesani EK. 1987. Parenteral chloro-quine for treating falciparum malaria. J. Infect. Dis. 155:192–201.

54. World Health Organization. 2010. Guidelines for the treatment of ma-laria, 2nd ed. World Health Organization, Geneva, Switzerland.

Batty et al.


289

xi.b 161B161BAppendixB:NONMEMcodeforpopulationpharmacokineticmodels

inthethesis

290

291

$PROB Azithromycin in Pregnancy Final Model

$DATA AZIDATA.CSV IGNORE=#

$INPUT # ID TIME CONC=DV DOSE=AMT CMT RATE MDV PREG DSGP FDHT GEST MALS HB GLUC

$SUB ADVAN12 TRANS4

$PK

TVCL=THETA(1)*(WT/70)**(.75)

TVV2=THETA(2)*(WT/70)+PREG*THETA(9)

TVQ3=THETA(3)*(WT/70)**(.75)

TVV3=THETA(4)*(WT/70)

TVQ4=THETA(5)*(WT/70)**(.75)


TVKA=THETA(7)

TVD1=THETA(8)

CL=TVCL*EXP(ETA(1))

V2=TVV2*EXP(ETA(2))

Q3=TVQ3

V3=TVV3*EXP(ETA(3))

V4=TVV4

Q4=TVQ4

KA=TVKA

D1=TVD1*EXP(ETA(4))

S2=V2

S3=V3

S4=V4

$ERROR

Y=F*(1+ERR(1))

$THETA

(0,137.6943) ; CL

(0,431.3440) ; Vcentral

(0,681.0347) ; Q3

(0,2928.8) ; Vperi

(0,42.8039) ; Q4

(0,7019.6) ; Vperi 2

(0,0.4800,2) KA

(0,1.5549) ; DUR

(0,28.5617) ; COV

$OMEGA

0.0545 ; IIV‐CL

0.9630 ; IIV‐V2

0.1174 ; IIV‐V3

0.0313 ; IIV‐DUR

292

$SIGMA

0.2037 ; Proportional error

$EST MAX=9990 SIG=3 METHOD=COND INTER POSTHOC PRINT=5

$COV PRINT=E

293

$PROB Pyrimethamine in Infants Final Model

$DATA PYRDATA.CSV IGNORE #

$INPUT # ID TIME CONC=DV DOSE=AMT MDV GFR DSGR DSKG EPMA MALS AQTR HB

$SUB ADVAN4 TRANS4

$PK

TVHILL=THETA(5)

HILL=TVHILL

TVMA50=THETA(6)

MA50=TVMA50

FCL=((EPMA)**HILL)/(((EPMA)**HILL)+MA50**HILL)

TVCLH=THETA(7)*FCL

CLH=TVCLH

RGFR=EGFR/120

TVCLR=THETA(8)*RGFR

CLR=TVCLR

TVCL=(CLH+CLR)*((WT/70)**0.75)

TVKA=THETA(1)


TVQ=THETA(4)*((WT/70)**0.75)


CL=TVCL*EXP(ETA(1))

KA=TVKA

V2=TVV2*EXP(ETA(2))

Q=TVQ*EXP(ETA(2)*THETA(9))

V3=TVV3

S2=V2

S3=V3

$ERROR

IPRED =LOG(F+0.001)

IRES=DV‐IPRED

IWRES=IRES/1

Y=LOG(F)+ERR(1)

$THETA

(0, 0.779) FIX ; KA

(0, 222) ; V2

(0, 64.1) ; V3

(0, 0.0735,20) ; Q

(0, 7.39,40) ; hill

(0, 318,400) ; ma50

(0, 0.854,10) ; CLh

(0, 0.416,10) ; CLr

294

(0.1, 6.95,50) ; ETA‐R

$OMEGA BLOCK(2)

0.0772 ; IIV‐CL

0.0192 ; R (CL, V2)

0.0168 ; IIV‐V2

$SIGMA

0.113 ; Proportional Error

$EST MAX=0 SIG=3 METHOD=COND INTER POSTHOC PRINT=5 MSFO=MSFILE

$COV PRINT=E

295

$PROB Sulfadoxine in Infants Final Model

$DATA SDXDATA.CSV IGNORE #

$INPUT # ID TIME CONC=DV DOSE=AMT MDV GFR DSGR DSKG EPMA MALS AQTR HB

$SUB ADVAN2 TRANS2

$PK

TVHILL=THETA(4)

HILL=TVHILL

TVMA50=THETA(5)

MA50=TVMA50


TVCLH=THETA(6)*FCL

CLH=TVCLH

RGFR=GFR/120

TVCLR=THETA(7)*RGFR

CLR=TVCLR

TVKA=THETA(1)

TVCL=(CLH+CLR)*((WT/70)**0.75)

TVV=THETA(2)*(WT/70)

KA=TVKA

CL=TVCL*EXP(ETA(1))

V=TVV*EXP(ETA(2))

F1=1*((DSKG/60)**THETA(3))

S2=V

$ERROR

IPRED =LOG(F+0.001)

IRES=DV‐IPRED

IWRES=IRES/1

Y=LOG(F)+ERR(1)

$THETA

(0,1.23) FIX ; KA

(0,20,30) ;V

(‐.8,‐.2,0) ;DSKG‐F

(0,6,100) ;HILL

(0,280,400) ;MA50

(0,.03,.5) ;CLH

(0,0.003,.5) ;CLR

$OMEGA BLOCK(2)

0.1 ; IIV‐CL

0.05 ; R (CL,V)

0.1 ; IIV‐V

296

$SIGMA

0.5 ; Proportional error

$ESTIM MAX=9990 SIG=3 METHOD=COND INTER POSTHOC PRINT=5

$COV PRINT=E

297

$PROB Sulfadoxine + N‐acetylsulfadoxine in Infants Final Model

$DATA SDXNSXDATA.CSV IGNORE #

$INPUT # ID TIME CONC=DV DOSE=AMT CMT MDV GFR DSGR DSKG EPMA MALS AQTR HB

$SUB ADVAN5 TRANS1

$MODEL

NCOMPARTMENTS=3

COMP=(GUT,DEFDOSE) ;Gut compartment

COMP=(PARENT) ;Observation compartment for sulfadoxine

COMP=(METAB) ;Observation compartment for n‐acetylsulfadoxine

$PK

TVHILL=THETA(4)

HILL=TVHILL

TVMA50=THETA(5)

MA50=TVMA50


TVCLH=THETA(6)*FCL

CLH=TVCLH

RGFR=GFR/120

TVCLR=THETA(7)*RGFR

CLR=TVCLR

TVKA=THETA(1)

TVCL2=(CLH+CLR)*((WT/70)**0.75)


TVCL3=THETA(8)*RGFR*((WT/70)**0.75)


KA=TVKA

CL2=TVCL*EXP(ETA(1))

V2=TVV2*EXP(ETA(2))

CL3=TVCL3*EXP(ETA(3))

V3=TVV3*EXP(ETA(4))

F1=1*((DSKG/60)**THETA(3))

K12=KA

K20=(CL2*.4)/V2

K23=(CL2*.6)/V2

K30=CL3/V3

S2=V

$ERROR

IF (CMT.EQ.2) THEN

IPRED=LOG((A(2)/V2)+0.0001)

IRES=DV‐IPRED

IWRES=IRES/1

298

Y=LOG(F)+ERR(1)

ENDIF

IF(CMT.EQ.3) THEN

IPRE=LOG((A(3)/V3)+.0001)

IRES=DV‐IPRE

IWRE=IRES/1

Y=LOG(F)+ERR(2)\

ENDIF

$THETA

(0,1.23) FIX ; KA

(0,24.2,30) FIX ; V2

(‐1,‐.56,1.5) FIX ; DSKG‐F

(0,4.07,100) FIX ; HILL

(0,271,400) FIX ; MA50

(0,.0458,.5) FIX ; CLh

(0,0.00439,.5) FIX ; CLr

(0,0.3,30) ; CL3

(0,10,30) ; V3

$OMEGA BLOCK(2)

0.1 ; IIV‐CL2

0.05 ; R (CL2,V2)

0.1 ; IIV‐V2

$OMEGA BLOCK(2)

0.1 ; IIV‐CL3

0.05 ; R (CL3,V3)

0.1 ; IIV‐V3

$SIGMA

0.0272 ; Proportional error for sulfadoxine

0.1 ; Proportional error for n‐acetylsulfadoxine

$ESTIM MAX=9990 SIG=3 METHOD=COND INTER POSTHOC PRINT=5

$COV PRINT=E

299

$PROB Lumefantrine + Desbutyl‐lumefantrine in Children Final Model

$DATA LUMDBLDATA.CSV IGNORE=#

$INPUT # ID TIME CONC=DV DOSE=AMT CMT MDV AGE SEX PARA HB

$SUB ADVAN5 TRANS1

$MODEL

NCOMPARTMENTS=6


COMP=(PAR1) ;Observation compartment for lumefantrine

COMP=(MET1) ;Observation compartment for desbutyl‐lumefantrine

COMP=(PAR2) ;Peripheral compartment 1 for lumefantrine

COMP=(PAR3) ; Peripheral compartment 2 for lumefantrine

COMP=(MET2) ; Peripheral compartment for desbutyl‐lumefantrine

$PK

CL2=THETA(1)*((WT/70)**(.75))*EXP(ETA(2))

V2=THETA(2)*(WT/70)

Q4=THETA(3)*((WT/70)**(.75))

V4=THETA(4)*(WT/70)

Q5=THETA(5)*((WT/70)**(.75))

V5=THETA(6)*(WT/70)

KA=THETA(7)*EXP(ETA(1))

IF(DOSN.EQ.6) KA=THETA(13)*EXP(ETA(1))

ALAG1=2.0

F1=(1)*EXP(ETA(3)+ETA(6))

IF (DOSN.EQ.2) F1=1*EXP(ETA(3)+ETA(7))





MWRATO=472.83/528.939

CL3=THETA(8)*MWRATO*((WT/70)**(.75))*EXP(ETA(4))

V3=THETA(9)*MWRATO*(WT/70)*EXP(ETA(5))

Q6=THETA(10)*MWRATO*((WT/70)**(.75))

V6=THETA(11)*MWRATO*(WT/70)

FP=THETA(12)

S2=V2

S3=V3

S4=V4

S5=V5

S6=V6

K12=KA

K13=KA*(FP)/(1‐FP)

300

K24=Q4/V2

K42=Q4/V4

K25=Q5/V2

K52=Q5/V5

K20=0

K23=CL2/V2

K30=CL3/V3

K36=Q6/V3

K63=Q6/V6

$ERROR

IF (CMT.EQ.2) THEN

IPRED =LOG(F+0.001)

IRES=DV‐IPRED

IWRES=IRES/1

Y=LOG(F)+ERR(1)

ENDIF

IF(CMT.EQ.3) THEN

IPRE =LOG(F+0.001)

IRES=DV‐IPRE

IWRE=IRES/1

Y=LOG(F)+ERR(2)

ENDIF

$THETA

(0,7.87,) ;CL2

(0,257,) ; V2

(0,1.45,) ; Q4

(0,109,) ; V4

(0,.874,) ; Q5

(0,184,) ; V5

(0,.409,) ; KA

(0, 873) ; CL3

(0, 35400) ; V3

(0, 757) ; Q6

(0, 68200) ; V6

(0, .1,1) ; FP

(0,.409,) ; KA‐DOSE6

$OMEGA

0.1 ; IIV‐KA

$OMEGA BLOCK(2)

0.1 ; IIV‐CL2

0.05 ; R (CL2, F)

0.1 ; IIV‐F

$OMEGA BLOCK(2)

0.1 ; IIV‐CL3

0.05 ; R (CL3,V3)

301

0.1 ; IIV‐V3

$OMEGA BLOCK(1)

.1 ; IOV‐F

$OMEGA BLOCK SAME ; IOV‐F





$SIGMA

0.0565 ; Proportional error for lumefantrine

0.565 ; Proportional error for desbutyl‐lumefantrine

$EST MAX=9990 SIG=3 METHOD=COND INTER POSTHOC PRINT=1

$COV PRINT=E

302

$PROB Artemether + Dihydroartemisinin in Children Final Model

$DATA LUMDBLDATA.CSV IGNORE=#

$INPUT # ID TIME CONC=DV DOSE=AMT CMT MDV AGE SEX PARA HB

$SUB ADVAN5 TRANS1

$MODEL

NCOMPARTMENTS=6


COMP=(PAR1) ;Observation compartment for artemether

COMP=(MET1) ;Observation compartment for dihydroartemisinin

COMP=(PAR2) ;Peripheral compartment for artemether

$PK

CL2=THETA(1)*((WT/70)**(0.75))*(1+THETA(7)*EXP(ETA(8))*(DSCL‐1))

V2=THETA(2)*(WT/70)

Q4=THETA(3)*((WT/70)**(0.75))

V4=THETA(4)*(WT/70)

KA=THETA(5)

SIG=THETA(6)

F1=1*EXP(ETA(1)+ETA(2));*EXP(ETA(1))






MWRAT= 284.35/298.37

V3=THETA(8)*(WT/70)*(MWRAT)

CL3=THETA(9)*((WT/70)**0.75)*(MWRAT)

SIG2=THETA(10)

K12=KA

K20=0

K30=CL3/V3

K23=CL2/V2

K24=Q4/V2

K42=Q4/V4

S2=V2

S3=V3

S4=V4

$ERROR

LOQ=LOG(5)

IPRED=LOG(F+0.001)

IRES=DV‐IPRED

303

IWRES=IRES/1

DUM=(LOQ‐IPRED)/SIG

CUMD=PHI(DUM)

IF(TYPE.EQ.1.AND.CMT.EQ.2) THEN

F_FLAG=0

Y=LOG(F)+SIG*ERR(1)

ENDIF


F_FLAG=1

Y=CUMD

ENDIF

LOQ2=LOG(2)

IPRED=LOG(F+0.001)

IRES=DV‐IPRED

IWRES=IRES/1

DUM=(LOQ2‐IPRED)/SIG2

CUMD=PHI(DUM)


F_FLAG=0

Y=LOG(F)+SIG2*ERR(1)

ENDIF


F_FLAG=1

Y=CUMD

ENDIF

$THETA

(0, 64.2) ; CL

(0, 53.1) ; V2

(0, 15) ; Q

(0, 349) ; V3

(0, 1) FIX ; KA

(0, 0.587) ; Proportional error for artemether

(0, 0.628) ; CLDOSN

(0,360) ; V

(0,260) ; CL

(0,.6) ; Proportional error for dihydroartemisinin

$OMEGA

0.385 ; IIV‐F

$OMEGA BLOCK(1)

.1 ; IOV‐F






$OMEGA

304

0.537 ; IIV‐CL2

$SIGMA

1 FIX

$EST MAX=9990 SIG=3 METHOD=COND INTER LAPLACIAN

$COV PRINT=E

305

$PROB Piperaquine in Children Final Model

$DATA PQDATA.CSV IGNORE=#

$INPUT # ID TIME CONC=DV DOSE=AMT AGE SEX PARA FEV

$SUB ADVAN6 TOL=3

$MODEL

NCOMPARTMENTS=4


COMP=(PAR1) ;Observation compartment for piperaquine

COMP=(PAR2) ; Peripheral compartment 1 for piperaquine

COMP=(PAR3) ;Peripheral compartment 2 for piperaquine

$PK

CL=THETA(1)*(WT/70)**(.75)*EXP(ETA(4))

V2=THETA(2)*(WT/70)*EXP(ETA(5))

Q3=THETA(3)*(WT/70)**(.75)


Q4=THETA(5)*(WT/70)**(.75)

V4=THETA(6)*(WT/70)

MTT=THETA(7)*EXP(ETA(7));Mean transit time

NN=THETA(8);*EXP(ETA(4)) ;Number of transit compartments

FPQ=1

IF(DOSN.EQ.1) FPQ=1*EXP(ETA(1))



S2=V2

S3=V3

S4=V4

IF(TIME.EQ.0) THEN

PD1=0

TDOS1=0

PD2=0

TDOS2=200

PD3=0

TDOS3=200

ENDIF

IF(AMT.GT.0.AND.DOSN.EQ.1) THEN

PD1=AMT*FPQ

TDOS1=TIME

ENDIF


PD2=AMT*FPQ

TDOS2=TIME

306

ENDIF


PD3=AMT*FPQ

TDOS3=TIME

ENDIF

F1=0

KTR=(NN+1)/MTT

KA=KTR

L=LOG(2.5066)+(NN+.5)*LOG(NN)‐NN ;Sterling

$DES

X=0.00001

RATE1=EXP(LOG(PD1+X)+LOG(KTR+X)+NN*LOG(KTR*T+X)‐KTR*T‐L)

DADT(1)=RATE1‐KA*A(1)

IF(T.GE.TDOS2) THEN


RATE2=EXP(LOG(PD2+X)+LOG(KTR+X)+NN*LOG(KTR*(T‐TDOS2)+X)‐KTR*(T‐TDOS2)‐L)

DADT(1)=RATE1+ RATE2‐KA*A(1)

ENDIF

IF(T.GE.TDOS3) THEN




DADT(1)=RATE1+ RATE2 + RATE3 ‐KA*A(1)

ENDIF

DADT(2)=KA*A(1)‐Q3/V2*A(2)+Q3/V3*A(3)‐Q4/V2*A(2)+Q4/V4*A(4)‐CL/V2*A(2)

DADT(3)=Q3/V2*A(2)‐Q3/V3*A(3)

DADT(4)=Q4/V2*A(2)‐Q4/V4*A(4)

$ERROR

IPRED =LOG(F+0.001)

IRES=DV‐IPRED

IWRES=IRES/1

Y=LOG(F)+ERR(1)

$THETA

(1,70,) ; CL

(1,4000,) ; V2

(1,400,) ; Q3

(1,4000,) ; V3

(1,120,) ; Q4

(1,40000,) ; V4

(.1,1.1,20) ; MTT

(.1,1.02,) ; NN

$OMEGA BLOCK(1)

307

0.6 ; IOV‐F



$OMEGA BLOCK(3)

0.1 ; IIV‐CL

0.05 ; R (CL,V2)

0.1 ; IIV‐V2

0

0.1 ; R (V2,V3)

0.5 ; IIV‐V3

$OMEGA

0.6 ; IIV‐MTT

$SIGMA

0.15 ; Proportional error for piperaquine

$ESTIM MAX=9990 SIG=2 METHOD=COND INTER POSTHOC

$COV PRINT=E

308

$PROB Artemisinin (from ART‐PQ) in Children Final Model

$DATA ART_PQ_DATA.CSV IGNORE #

$INPUT # ID TIME CONC=DV DOSE=AMT AGE SEX PARA FEV

$SUB ADVAN4 TRANS4

$PK

CL=THETA(1)*((WT/70)**(0.75))*EXP(ETA(1))

V2=THETA(2)*(WT/70)

Q=THETA(3)*((WT/70)**(0.75))

V3=THETA(4)*(WT/70)

KA=THETA(5)

F1=1*EXP(ETA(2))

IF(DOSN.EQ.2) F1=THETA(6)*EXP(ETA(3))

S2=V2

S3=V3

$ERROR

IPRED =LOG(F+0.001)

IRES=DV‐IPRED

IWRES=IRES/1

Y=LOG(F)+ERR(1)

$THETA

(0,100,) ;CL

(200,500,) ; V2

(0,50,) ; Q

(0,500,) ; V3

(1,3,) ; KA

(0,.2,2) ; F2

$OMEGA

0.5 ; IIV‐V2

$OMEGA BLOCK(1)

0.1 ; IOV‐F


$SIGMA

0.1 ; Proportional error for artemisinin

$ESTIM MAX=9990 SIG=3 METHOD=COND INTER POSTHOC

$COV PRINT=E

309

$PROB Napthoquine in Children Final Model

$DATA NQDATA.CSV IGNORE #

$INPUT # ID TIME CONC=DV DOSE=AMT GRP DOSN FVR DSKG SEX SPLN MAL PARA AGE FEV HB

$SUB ADVAN12 TRANS4

$PK


CL=THETA(2)*(WT/70)**0.75*EXP(ETA(3))

V2=THETA(3)*(WT/70)*EXP(ETA(2))*(1+THETA(11)*(HB‐10.3))

Q3=THETA(4)*(WT/70)**0.75

V3=THETA(5)*(WT/70)

Q4=THETA(6)*(WT/70)**0.75*EXP(ETA(5))


ALAG1=THETA(8)

F1=1*(1‐THETA(9)*(FEV))

IF (GRP.EQ.3.AND.DOSN.EQ.1) F1=1*(1‐THETA(9)*(FEV))*(1‐THETA(10))

S2=V2

S3=V3

$ERROR

IPRED =LOG(F+0.001)

IRES=DV‐IPRED

IWRES=IRES/1

Y=LOG(F)+ERR(1)

$THETA

(0,0.4107,5) ; KA

(0,70.8289,600) ; CL

(0,10502.4,100000) ; V2

(0,44.8867,500) ; Q3

(0,28499.2,100000) ; V3

(0,273.5786,4000) ; Q4

(0,13137.0,120000) ; V4

(0,0.0307,1) ; LAG

(0,0.0280,.8) ; F‐FEV

(0,0.0584,.5) ; F31

(0,0.0156,.2) ; V2‐HB

$OMEGA BLOCK(5)

0.21 ; IIV‐KA

0.017 ; R (KA,V2)

0.29 ; IIV‐V2

0

0.14 ; R (V2,CL)

0.1789 ; IIV‐CL

0

310

0

0.01 ; R (CL,V4)

0.078 ; IIV‐V4

0

0

0

0.06 ; R (V4,Q4)

0.10 ; IIV‐Q4

$SIGMA

0.0376 ; Proportional error for naphthoquine

$EST MAX=9990 SIG=3 METHOD=COND INTER POSTHOC

$COV PRINT=E

311

$PROB Artemisinin (from ART‐NQ) in Children Final Model

$DATA ART_NQ_DATA.CSV IGNORE #

$INPUT # ID TIME CONC=DV DOSE=AMT GRP DOSN FVR DSKG SEX SPLN MAL PARA AGE FEV HB

$SUBROUTINES ADVAN4 TRANS4

$PK

CL=THETA(1)*((WT/70)**(0.75))*EXP(ETA(1))

V2=THETA(2)*(WT/70)*EXP(ETA(1)*THETA(8))

Q=THETA(3)*((WT/70)**(0.75))

V3=THETA(4)*(WT/70)


F1=1

IF(DOSN.EQ.2) F1=1*(1‐THETA(6))

ALAG1=THETA(7)*EXP(ETA(2))

S2=V2

S3=V3

$ERROR

IPRED =LOG(F+0.001)

IRES=DV‐IPRED

IWRES=IRES/1

Y=LOG(F)+ERR(1)

$THETA

(0,544.3046) ; CL

(0,202.6390) ; V2

(0,5.2472) ; Q

(0,61.5488) ; V3

(0,1.5661) ; KA

(0.01,0.7709,.9) ; F2

(0,0.3301,1) ; LAG

(0,1.1865,10) ; ETA‐R

$OMEGA BLOCK(3)

0.2 ; IIV‐CL

0.1 ; R (CL,LAG)

0.6 ; IIV‐LAG

0.1 ; R (CL,KA)

.07 ; R(LAG,KA)

.23 ; IIV‐KA

$SIGMA

0.22 ;Proportional Error for artemisinin

$EST MA

312

X=9990 SIG=3 METHOD=COND INTER POSTHOC

$COV PRINT=E

Documents

Pharmacological studies of malaria in pregnancy, …research-repository.uwa.edu.au/files/3222408/Salman_Sam...Pharmacological studies of malaria in pregnancy, infancy and childhood