Retinal Origins of Vigabatrin Toxicity In Infantile Spasms

by

Julianna Sienna

A thesis submitted in conformity with the requirements for the degree of Master of Medical Science

Institute of Medical Science University of Toronto

© Copyright by Julianna Sienna 2011

ii

Retinal Origins Of Vigabatrin Toxicity In Infantile Spasms

Julianna Sienna

Master of Science

Institute of Medical Science University of Toronto

2011

Vigabatrin (VGB) is an anti-epileptic drug used to treat children with Infantile

Spasms (IS). The 3.0 flicker amplitude of the electroretinogram (ERG) is

currently used to monitor visual function changes in infants on VGB. To find

a more specific marker of permanent changes due to VGB, sedated ERGs

were performed on 31 IS patients and 13 retinally normal controls to isolate

components of the cone pathway. ERG growth curves, for each component,

recorded from children with IS were generated using data recorded pre-VGB

treatment and for controls. Only the cone off response (from Off bipolar

cells) and cone photoreceptor sensitivity were associated with decreased

flicker amplitude. Twenty nine percent of patients had an abnormal cone off

response. No patient had an abnormal cone off response at baseline. No

patient with an abnormal cone off response recovered normal function. The

cone off response could serve as a marker VGB retinal toxicity.

iii

Acknowledgments

This thesis would not have been possible without the help and support of many people. First and foremost, I would like to thank Dr. Carol Westall for her unending patience, commitment to my learning and for sharing so much of her knowledge and expertise with me. Dr. Westall has given me so many opportunities to grow both as a scientist and as a person over the last two years and I will always be grateful for that. Thank you for standing in my corner as I pursued my dreams, believing in my project and truly serving as a role model for me.

I would also like to thank Tom Wright for his support in the development and execution of this project and Carole Panton and Melissa Cotesta for performing ERGs. I have appreciated your guidance, your technical knowledge. Tom, Carole & Melissa’s participation in this project and my life made coming to the hospital everyday a treat. Thank you for always believing I could pull it out in the end and always helping me to get there.

Thanks must also be given to Dr. Raymond Buncic, and the sedation nurses Beverly Griffiths and Yasmin Sherrif for their help in testing. Thank you for being teachers as well as team members. I have appreciated the support from all the members of the Ophthalmology department at Sick Kids.

The supervision and advice from the members of my program advisory committee must be acknowledged. Drs. Agnes Wong, Carter Snead & Gideon Koren were instrumental in ensuring the quality of this research project.

Lastly, thank you to my family and friends whose constant support and encouragement helped me to stay focused.

iv

Table of Contents

Acknowledgments .............................................................................. iii  

Table of Contents ................................................................................ iv  

List of Tables ...................................................................................... vi  

List of Figures .................................................................................... vii  

List of Appendices .............................................................................. ix

List of Abbreviations………………………………………………………………………………………xii

Introduction 1  

  Infantile Spasms ............................................................................. 1  1

1.1  Etiology 2  

1.2  Treatment 3  

  Vigabatrin ....................................................................................... 4  2

2.1  Clinical Pharmacology of Vigabatrin 6  

2.2  Vigabatrin Associated Visual Field Loss 8  

2.2.1  Animal Studies ................................................................... 8  

2.2.2  Adult Human Studies ......................................................... 10  

2.2.3  Studies in Children and Infants ............................................ 13  

  Vision ........................................................................................... 18  3

  Electroretinography ......................................................................... 21  4

  Monitoring Retinal Function in IS Patients with ERGS ........................... 25  5

Methods

  Subjects and Controls Patients ......................................................... 32  6

6.1  Subjects 32  

v

6.1.1   INCLUSION CRITERIA ........................................................ 32  

6.1.2  EXCLUSION CRITERIA ........................................................ 33  

6.2  Control Patients 33  

6.2.1   INCLUSION CRITERIA: ....................................................... 33  

6.2.2  EXCLUSION CRITERIA: ....................................................... 33  

  Testing & Data collection ................................................................. 34  7

7.1  Scoring 36  

7.1.1  Flicker .............................................................................. 36  

7.1.2  Photopic negative response ................................................. 36  

7.1.3  Cone Photoreceptors Parameters ......................................... 37  

7.1.4  Cone Off Response ............................................................ 38  

  Statistical Analysis .......................................................................... 39  8

Results

  Group Demographics ....................................................................... 41  9

9.1  Subjects 41  

9.2  Controls: 46  

 Analysis ....................................................................................... 48  10

10.1   Developmental curves 48  

 Discussion .................................................................................... 79  11

11.1   Clinical Implications 95  

11.2   Problems and Considerations 96  

11.3   Future Directions 98  

 References ................................................................................. 101  12

Appendices ..................................................................................... 115

vi

List of Tables

1. Symptomatic causes of Infantile Spasms by category.......................2 2. Initial case reports of vigabatrin-associated visual field loss.............11 3. Case reports of two IS patients with visual field loss.......................14 4. Summary of studies investigating visual field defects

in children on VGB using perimetry...............................................16 5. Summary of ERG findings in adults taking VGB…............................26 6. Summary of ERG results in children and infants taking

vigabatrin.................................................................................29 7. Additional stimulus condition testing parameters............................35 8. Group demographics of patients with IS at baseline........................42 9. Drug history and visual acuity in Infantile Spasms patients at

baseline....................................................................................44 10. Group demographics of controls...................................................46 11. Longitudinal drug information and visual acuity in IS patients..........56 12. Diagnostic characteristics of abnormal tests using flicker, cone off,

and sensitivity…………………………………………………………………………………....68 13. Individual patient data for all those with at least one

abnormal test. ........................................................................76 14. Distribution of sex and mean daily VGB dose for normal vs.

abnormal test using different definitions of abnormality.................78

vii

List of Figures

1. Structure of vigabatrin.............................................................5 2. Schematic diagram of the retina..............................................18 3. Light adapted (photopic) 3.0 ERG............................................23 4. 3.0 Flicker response from IS patient........................................36 5. Photopic negative response from patient..................................37 6. Hood & Birch equation to describe the leading edge of

the a-wave………………………………………………………………………………………38 7. Cone off response.................................................................38 8.

a. Control flicker amplitude plotted by age in months............48 b. Baseline IS flicker amplitude plotted by age in months.......49

9. a. Control PhNR amplitude plotted by age in months.............50 b. Baseline IS PhNR amplitude plotted by age in months........50

10. a. Control cone sensitivity plotted by age in months..............51 b. Baseline IS cone sensitivity plotted by age in months.........51

11. a. Control cone maximum response plotted by age in

months. ………………………………………………………………………………..52 b. Baseline IS cone maximum response plotted by age in

months. …………………………………………………………………………………52 12. Control (solid line) and subject (dashed line) developmental curves

for: a. Flicker amplitude……………………………………………………………………53 b. PhNR amplitude……………………………………………………………………..54 c. Cone sensitivity……………………………………………………………………..54 d. Cone maximum response……………………………………………………..55

13. Boxplot comparing cone off response amplitude in controls with IS subjects at baseline.................................................55

14. a. Box plot comparing adjusted flicker amplitude over time on

vigabatrin.. …………………………………………………………………………..60 b. Box plot comparing adjusted PhNR amplitude over time on

vigabatrin.……………………………………………………………………………..61

viii

c. Box plot comparing adjusted maxiumum response (Rmax) over time on vigabatrin.. …………………………………………………….62

d. Box plot comparing adjusted cone sentivity over time on vigabatrin………………………………………………………..63

15. Cone off response in normal and abnormal test................65 16.

a. Adjusted Flicker amplitude for patients with normal vs. abnormal cone off response...........................................66

b. Adjusted Flicker amplitude for patients with normal vs. abnormal photopic negative response .............................66

c. Adjusted Flicker amplitude for patients with normal vs. abnormal Cone Rmax ..................................................67

d. Adjusted Flicker amplitude for patients with normal vs. abnormal cone Sensitivity .............................................67

17. Survival plots for cone off response, flicker amplitude and cone sensitivity ..................................................................70

18. Venn Diagram of the number test points where patients had overlap between each test. ..................................................71

19. Mosaic plots of agreement in classifying tests between: a. flicker and cone off.......................................................72 b. flicker and sensitivity ...................................................73 c. cone off and sensitivity ................................................73 d. flicker, cone off, and sensitivity .....................................74

20. Structure of D- alpha amino adipic acid (left) and vigabatrin (right)................................................................93

21. Structure of 4-methryl glutamic acid (left) and GYKI 52466 (right).……………………………………………………………………..93

ix

List of Appendices

1. Current lab protocol …………………………………………………………………………….115

2. Vigabatrin and infantile epilepsy subject consent form………………………116

3. Vigabatrin and infantile epilepsy control consent form……………………….122

4. Patient report form……………………………………………………………………………….129

5. R software for Hood and Birch model………………………………………………….131

x

List of Abbreviation

Ab………………………………………………………………………………………..………………abnormal ACTH……………………………………………………………………Adrenocoritcotropic hormone AHD…………………………………………………………………….…….aldehyde dehydrogenase α-AAD…………………………………………………………………………. alpha aminoadipic acid amp………………………………………………………………………………………………….…amplitude AMPA………..α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor APB ……………………………………………………………….….2-amino-4-phosphonobutyrate ATP……………………………………………………………………………….Adenosine triphosphate CB……………………………………………………………………………………………………..…Clobazam CI…………………………………………………………………………………………Confidence Interval CL………………………………………………………………………………………….…………Clonazepam CNS…………………………………………………………………………..…Central Nervous System CO………………………………………………………………………………………………….Corticotropin CPS…………………………………………………………………..…………Complex Partial Seizures CS………………………………………………………………………………………………controlled study CYP P450………………………………………………………………………………..Cytochrome P450 CZ………………………………………………………………………………………………Carbemazepine D/C………………………………………………………………………………….………………discontinued Dx…………………………………………………………………………………………………………diagnosis EEG…………………………………………………………………………………Electroencephalogram EOG…………………………………………………………………………………………electroretinogram ERG…………………………………………………………………………………………Electroretinogram FR……………………………………………………………………………………………….Folate reduced G…………………………………………………………………………………………………………Goldmann GABA…………………………………………………………………….…Gamma-aminobutyric acid GABA-T……………………………………………………………………………….GABA-transaminase HF………………………………………………………………………………………………Humphrey field IS……………………………………………………………………………………………Infantile Spasms IT………………………………………………………………………………………………….…implicit time L………………………………………………………………………………………………….……longitudinal LA…………………………………………………………………………………………………….Lamotrigine mo……………………………………………………………………………………………………….……month OAT…………………………………………………………….……………ornithine aminotransferase OCT……….………………………………………………………….Optical coherence tomography OP…………………………………………………………………………………..……oscillatory potential OX………………………………………………………………………………………….……Oxcarbazepine PB…………………………………………………………………………………………….……Phenobarbital PDA …………………………………………………………….…. 2,3 piperidine dicarboxylic acid PH…..……………………………………………………………………………………………………Phenytoin PhNR……………………………………………………………………….Photopic negative response phot………………………………………………………………………………………………………photopic

xi

pts…………………………………………………………………………………………………….……patients RAO………………………………………………………………………………….retinal amine oxidase RPE……………………………………………………………………………retinal pigment epithelium Rx………………………………………………………………………………………….…………prescription scot…………………………………………………………………………………………………….…scotopic SSAO…………………………………………………..Semicarbazide-sensitive amine oxidase SV……………………………………………………………………………………………Sodium valproate td…………………………………………………………………………………………………………..trolands TSC…………………………………………..………………………… Tuberous Sclerosis Complex TTX………………………………………………………………………………………….….… Tetrodotoxin VA……………………………………………………………………………………………………visual acuity VEP…………………………………………………………………………………visual evoked potential VEU ………………………………………………………………………visual electrophysiology unit VFD…………………………………………………………………………………………visual field defect VFL………………………………………………………………………………………………visual field loss VGB………………………………………………………………………..……………………….…Vigabatrin yr…………………………………………………………………………………………………….……………year

1

Introduction

Infantile Spasms 1Infantile Spasms are a specific type of seizures that affect children with a disorder

called West Syndrome. Infantile Spasms (IS) is classically associated with

hypsarrythmia, an EEG abnormality characterized by Gibbs, Fleming, and, Gibbs

(1954) as ‘a chaotic and disorganized background pattern consisting of high

voltage slow waves and spikes that are diffuse, non-rhythmic and variable in

duration and location’. Infantile spasms are characterized by massive myoclonic

jerks which consist of very brief flexor and /or extensor spasms of the trunk, head

and / or neck that present within the first year of life. These individual spasms

typically last between 1 and 5 seconds often occur in clusters of 3-20. Wong and

Trevathan (2001) estimated that IS affects 2-5 in every 10,000 births annually

worldwide based on data from six studies (Riikonen & Donner, 1979; Cowan &

Hudson, 1991; Ludvigsson, Olafsson, Sigurthadttir & Hauser, 1994; Sidenvall &

Eeg-Olofsson, 1995; Trevathan, Murphy & Yeargin-Allsopp, 1999; Rantala &

Putkonen, 1999). Between 70-90% of patients with IS have mental retardation,

and IS often is associated with intractable epilepsy, and severe developmental

delay and / or cognitive impairment (Zupanc, 2003). The prognosis for these

patients is very poor; IS is associated with a 5-30% mortality rate at 9 years of

age (Wong & Trevathan, 2001), in part due to the fact that between 20-50% of

these patients will develop Lennox-Gastaut syndrome (Zupanc, 2003). Treatment

2

is usually aggressive and immediate in order to achieve the best long term

outcomes in this disease (Zupanc, 2003).

1.1 Etiology

Though the etiology of seizures cannot be identified in every patient; there are

several known causes of Infantile Spasms. The etiology may be characterized as

either symptomatic or cryptogenic. Cryptogenic spasms are believed to be caused

by age-related multifactorial genetic predispositions however, there are also

believed to be other reasons that have not yet been identified (Wong & Trevathan,

2001). Symptomatic spasms can be further delineated into three groups: prenatal,

perinatal and postnatal. Table 1 lists some common events in each category

(Wong & Trevathan, 2001).

Table 1. Symptomatic causes of Infantile Spasms by category.

Symptomatic Causes

Pre-natal Perinatal Post-natal (50%)

Intrauterine insults and infections

Hypoxic-ischemic encephalopathy

Infection

Malformations of cortical development

Obstetric trauma Trauma

Neurocutaneous syndromes

Labour complications Hypoxic-Ischemic insult

Metabolic disorders Other asphyxia events Tumors

3

1.2 Treatment

There are both surgical and pharmacological treatments for IS. Surgery has been

shown to have an 80% cure rate and positive long term outcomes (Yum et al.,

2011). However, such surgeries can only be performed in the small proportion of

IS patients with unilateral or focal congenital or early-acquired cortical lesions in

whom the epileptic zone can be accurately and reliably defined and removed.

Adrenocorticotropic hormone (ACTH) has historically been a first line

pharmacological treatment for IS as it has shown to produce reduction in seizures

and improved EEG’s in IS patients (Mackay et al., 2004; Baram et al., 1996).

Early treatment (within the first month of onset) with ACTH has even been shown

to lead to a more favourable prognosis in terms of seizure control, mental and

motor development (Willoughby, Thurston & Holowach, 1966). In a study

performed by Cohen-Sadan and colleagues (2009), IS patients were treated with

either early ACTH (within first month), late ACTH (any time after one month) or

vigabatrin (without delay) and short and long term outcomes were compared.

Short term outcomes (seizure cessation and EEG normalization) were comparable

between all three groups. At long term follow-up (range 5-16 years) those in the

early ACTH group were significantly more likely to achieve normal cognitive

outcomes than the VGB group. None of the early ACTH group experienced seizures

at follow-up, however 14% of late ACTH and 54% of vigabatrin subjects

experience seizures (Cohen-Sadan et al., 2009). It appears that in most cases,

4

ACTH and vigabatrin have comparable efficacy at reducing seizure frequency. One

large randomized trial that excluded patients with tuberous sclerosis complex

(TSC), found that although at 14 days of anti-epileptic treatment, ACTH had

superior seizure cessation rates, by 14 months this benefit disappeared because

ACTH had higher relapse rates (Lux et al., 2005). Mackay et al. (2004) is the

American Academy of Neurology Practice Parameter which assigns levels of

evidence and is currently the most authoritative source of information regarding

efficacy profiles of vigabatrin and ACTH. Mackay et al. (2004) ascertained based

on a literature review that ACTH is probably effective for short-term treatment of

IS and resolution of hypsarrhythmia and that vigabatrin is possibly effective in

treating IS. Mackay et al. (2004) noted that there is a lack of sufficient higher

level evidence on efficacy, dosage and treatment schedule for either drugs. ACTH

is also associated with hypertension, metabolic abnormalities, osteoporosis sepsis,

congestive heart failure and cushingoid features (Riikonen & Donner, 1980). Given

this, VGB is favoured in cases of IS with TSC and used in many other cases of

Infantile Spasms, this is especially true at our centre.

Vigabatrin 2Vigabatrin is an anti-epileptic drug used to treat both adults with refractory

complex partial seizures and as first line therapy for children with Infantile Spasms

in Canada, the UK and Europe (National Institute for Health and Clinical Excellence

[NICE], 2010; Scottish and Intercollegiate Guidelines Network [SIGN], 2005;

5

Wheless, Clarke, Arzimanoglou & Carpenter, 2007). VGB was recently approved

for use in the USA. Vigabatrin, γ-vinyl GABA (figure 1), marketed as Sabril, is

produced in the USA by Lundbeck Pharmaceuticals.

Figure 1. Structure of vigabatrin.

Vigabatrin was first developed in the 1975 by the Merrell Dow Research Institute

(Richens 1991; Lundbeck Inc., 2010). VGB is a structural analog of GABA and is

believed to act by irreversibly inhibiting GABA transaminase (GABA-T). By

inhibiting GABA-T, VGB causes less GABA to be broken down, increasing GABA

levels in the CNS (Grove et al, 1981, 1980). GABA also accumulates in the retina

to 5 times the baseline levels in rats 18 hours after administration of vigabatrin

(Neal, Cunningham, Shah & Yazulla, 1989) and levels are found to be 5 time

higher in the retina than in the brain (Sills et al., 2001). GABA is especially

detected in Muller cells (Neal et al., 1989). Levels of vigabatrin also increase in

the retina to 260% of control levels in Sprague Dawley rats (Sills et al, 2001), and

in this experiment brain levels of vigabatrin averaged 61% of levels in the retina.

VGB is produced as a racemic mixture (equal proportions) of R(-) and S(+)-

enantiomers. Enantiomers are the two mirror image forms that a chiral molecule

can take and can be thought of like a right and left hand. In vigabatrin, the chiral

carbon (the carbon that is asymmetric in that it is bonded to four different groups)

6

is Carbon 4. The R-enantiomer of vigabatrin is completely inactive (Meldrum &

Murugaiah, 1983). In infants with tuberous sclerosis, between 78-100% of

patients experience a reduction in spasm frequency (Lerner, Salamon & Sankar,

2010). A recent review showed that in studies of IS patients with a range of

etiologies, seizure control was achieved in 18-81% of patients (Lerner et al.,

2010).

2.1 Clinical Pharmacology of Vigabatrin

Vigabatrin crosses the blood brain barrier and produces a dose-dependent increase

in neuronal GABA which is reflected in GABA concentrations in CSF (Ben-

Menachem, Persson, Mumford, Haegele & Huebert, 1991; Petroff, Rothman,

Behar, Collins, & Mattson, 1996). Vigabatrin is water soluble, non-plasma protein

bound and absorbed rapidly after oral administration, reaching peak plasma

concentrations in 2 hours (Richens, 1991; Rey, Pons, & Olive, 1992). VGB has a

large volume of distribution and is found throughout the body (Richens, 1991).

Sixty to eighty percent of the dose of vigabatrin is excreted unchanged in urine

within 24 hours following first-order kinetics (Schechter, 1986). Little to no hepatic

metabolism occurs (Richens, 1991; French, 1999). S(+) enantiomer has an

elimination half-life in children between 5-6 hours (Rey et al., 1990). The effects

of vigabatrin however, last much longer than the time the drug is in the body, as

they are dependent on the synthesis of new GABA-T enzyme which may take

several days (Lundbeck Inc., 2010; Richens 1991). VGB is not known to affect CYP

7

P450 enzymes, a system of enzymes known to metabolize drugs (Van Parys,

Meijer, & Edelbroek, 1995). In clinical trials, no relationship between plasma

concentration of VGB and therapeutic effect has been shown (Lundbeck Inc.,

2010; Richens, 1991). In children, VGB may need to be given in higher mg/kg

doses because it may be less bioavailable (Rey et al., 1990). VGB has also been

shown in rats to cause GABA accumulation in the retina (Neal et al., 1989).

In general, vigabatrin is well tolerated, though side effects include hypotonia,

irritability, weight gain and lethargy (Chiron et al., 1991). Visual field defects were

first reported in adults taking VGB for localized related epilepsy by Eke, Talbot,

and Lawden in 1997. It is estimated that in those adult patients who are

prescribed VGB, 52% develop visual field defects (Maguire, Hemming, Wild,

Hutton & Harson, 2010). The mechanism of this toxicity is unknown. The defect is

characterized as a bilateral and concentric peripheral visual field constriction (Wild,

Martinez, Reinshagen, & Harding, 1999). In general, the nasal field is affected

more than the temporal field. In adults, behavioural visual field testing is used to

monitor peripheral visual field defects while on vigabatrin. In infants, this type of

testing is often unfeasible and thus electrophysiological testing is undertaken.

8

2.2 Vigabatrin Associated Visual Field Loss

2.2.1 Animal Studies

VGB retinal toxicity has been described in albino rats (Butler, Ford & Newberne,

1987), rabbits (Ponjavic, Granse, Kjellstrom, Andreasson, & Bruun, 2004) and

mice (Wang et al, 2008). Severe disorganization of peripheral retinal

photoreceptor layer was been reported (Butler et al., 1987) and cone

photoreceptors were found to degenerate (Duboc et al., 2004; Wang et al., 2008).

Duboc et al. (2004) and Ponjavic et al. (2004) showed that these changes

preceded decreases in the 30 Hz flicker response of the ERG.

Several mechanisms have been proposed based on results seen in animals. Butler

et al. first suggested that VGB is a mediator of phototoxicity (1987). This was

suggested because only albino rats were susceptible to the toxic effects. This

theory was later supported by work by Izumi et al. (2004) and Jammoul et al.

(2009). Jammoul et al. (2009) demonstrated that albino rats kept in darkness did

not develop retinal toxicity (defined as decreased photopic ERG amplitudes, length

of disorganization of photoreceptor layer, and number of cone segments).

Jammoul et al. (2009) also put forth a novel theory that taurine levels were

related to VGB phototoxicity. They noted that VGB treated animals had lower

levels of taurine than control animals, taurine levels were correlated with both

photopic ERG amplitudes and cone density (measured using histology), and

9

furthermore, that taurine supplementation may prevent the development of retinal

toxicity. Jammoul et al. (2010) also demonstrated in a neonatal rat model which

more closely approximates a model of IS, that VGB treatment caused cone

photoreceptor damage, disorganization of the photoreceptor layer, gliosis, and

retinal ganglion call loss, and that these effects could be partially prevented by

supplementing with taurine.

While the studies of phototoxicity and taurine levels in animal studies are

promising, it is still very unclear by what mechanism VGB causes visual field loss.

The localization of the deficit to the peripheral retina in humans presents some

major questions given that this is a rod dominated area and electrophysiologically

assessed damage has primarily been seen to cone system responses. Krauss,

Johnson, Sheth and Miller (2003) suggested that VGB may cause abnormal

integration of receptive fields by increasing GABA. GABA inhibits lateral inhibition

in the inner retina. It is suggested that this could explain the photopic visual loss

in a low-density cone area of the peripheral retina. It remains unclear whether an

accumulation of GABA or VGB in the retina is responsible for the deficits. It is

difficult to assert that the mechanism is the same in animals as in humans as a)

relevant animal model for Infantile Spasms have only recently been developed

(Stafstrom et al, 2011) b) the effects seen in animals are at doses much higher

than those typically administered to IS patients and c) the VGB mediated effects in

10

animals are not selective – all animals develop the same pattern of toxicity, which

is not seen in infants or adult humans treated with VGB.

2.2.2 Adult Human Studies

Vigabatrin-Associated Visual Field Defects were first described in detail by Eke et

al. in 1997. This case series reported symptomatic constriction of visual field in

three patients taking vigabatrin (details in Table 2). In all cases, patients reported

either tunnel vision or general visual defects after 2- 3 years of vigabatrin use.

One previous case was described by Faedda, Giallonardo, Marchetti and Manfredi,

in 1993. Eke et al. noted 9 cases (including the three that he reported) that had

been reported to the Committee on Safety of Medicines in the United Kingdom at

the time of publication and 28 cases of visual field abnormalities reported to the

manufacturers, (Hoechst Marion Roussel, now known as Sanofi-Aventis) by

January 1997. A fourth case was also reported by Wilson and Brodie in 1997 (table

2). Since that time, many reports have been conducted on visual field defects in

adult patients taking VGB. A recent systematic review estimated that based on 22

studies, the mean proportion of vigabatrin treated adults with visual field loss was

52% (95% CI: 46%-59%) (Maguire et al., 2010).

11

Table 2. Initial Case Reports of Vigabatrin-Associated Visual Field Loss

Case 1 – F (Eke) Case 2 – M (Eke)

Case 3 – F (Eke)

Case 4 – M (Wilson)

Dx Complex partial epilepsy

Temporal lobe epilepsy

Complex partial epilepsy

Refrac. Partial onset epilepsy

Dose 2000 mg/d 4000 mg/d 3500 mg/d 3000 mg/d

Other anti-epileptics

CZ 600 mg/d SV 1400 mg/d

PH 400 mg/d SV 3000 mg/d

CZ 600 mg/d CZ 1200 mg/d SV 5000 mg/d

Length on VGB before

37 mo 28 mo 38 mo ~ 6 yr

Visual Field Loss?

Severe, but normal acuity

Contracted peripheral fields particularly nasally normal acuity

Peripheral vision defects

Visual deterioration, blurring, loss of peripheral vision impaired VA (6/9 both eyes)

Electrophysiological findings

OPs subnormal, low Arden ratio (139% L, 167% R), VEP normal

Low Arden ratio (185%), VEP normal, OPs not assessed

ERG: reduced Ops

Flat EOG, sub-normal cone and rod ERG, normal VEP

Other Findings

Pale discs peripheral retina atrophic, MRI – left hippocampal atrophy, normal blood tests, Fluorescein angiography – spotty in RPE

MRI – hippocampal asymmetry, blood tests and fluorescein angiography normal, pale discs

MRI, fundoscopy and fluorescein angiography normal

Bilateral optic atrophy, maculopathy, MRI normal,

VGB D/C No improvement Remained Stable

No improvement

No improvement

12

Legend: CZ - Carbemazepine, SV - Sodium Valproate, PH - Phenytoin, F – Female, M- Male, Dx – Diagnosis, mo – months, yr – years, VA – visual acuity, OP – oscillatory potential, L - left, R - right, VEP – visual evoked potential, ERG - electroretinogram, EOG -electrooculogram, RPE – retinal pigment epithelium, D/C -discontinued

Some adult studies have also attempted to elucidate the mechanism of VGB visual

field defects. Further support for the decreased taurine theory was given by Sorri

et al. (2010) and Roubertie, Bellet and Echenne (1998). Roubertie et al. (1998)

suggested that the visual field loss seen in adults could be due to VGB inhibiting

ornithine transferase. Sorri et al. (2010) demonstrated that in patients age 14-78

who had stopped VGB for more than 1 year (n=21), ornithine δ aminotransferase

(OAT) activity was reduced in those with visual field defect (n=11) compared to

those without (n=10). This difference was not found between those with (n=4)

and without (n=6) visual field defects who were still taking the drug. Jammoul et

al presented evidence from 5 of 6 infants treated with vigabatrin who had

undetectable or below normal range levels of taurine supported this theory

(Jammoul et al., 2009).

Other anti-epileptic agents that elevate CNS levels of GABA such as tiagabine, do

not produce the same visual field defects as VGB (Krauss et al., 2003).

13

2.2.3 Studies in Children and Infants

Three reports published in 1999 described visual field loss in children taking VGB.

Wohlrab, Boltshauser, Schmitt, Schriever, and Landau (1999) described concentric

visual field loss, measured by Goldman perimetry, in 5 of 12 patients taking VGB

who would cooperate for assessment. Four of these patients had stopped

vigabatrin treatment at the time of assessment. All patients with field loss were

noted to be asymptomatic, and had been taking concurrent medications (Valproic

acid or oxcarbazepine). Of note, in an age matched control population with

complex partial or generalized epilepsy one patient also had concentric visual field

loss; this patient had been taking valproic acid and lamotrigine (Wohlrab et al.,

1999). In a report by Bjelajac, Gautam, and Logan (1999), 41 of 158 children

taking VGB had ‘ophthalmic abnormalities’. Of these patients, two were found to

have visual field defects, two had pale optic discs, 2 had retinal dysfunction, two

had refractory errors, 1 had optic atrophy and 1 had transient vision loss, however

none of these patients had a baseline test for comparison. No data was given on

length of treatment, etiology of seizures, or concurrent seizure medication use.

Lastly, Vanhatalo, Pääkkönen, and Nousiainen (1999) report two cases of infants

treated for IS with VGB who experience visual field constriction. Their details are

in table 3.

14

Table 3. Case Reports of two IS patients with Visual Field Loss

Legend: F – Female, M- Male, Dx – Diagnosis, mos – months, Rx – Prescription, CZ - Carbemazepine, OX – Oxcarbazepine, SV - Sodium Valproate, CO – Corticotropin, PB – Phenobarbital, CL – Clonazepam, LA – Lamotrigine, CB - Clobazam

Fifteen studies have investigated visual field defects in children on VGB using

perimetry. Results of these studies are summarized in table 4. Those studies not

indicated with an asterisk were included in a review by Maguire et al. (2010). The

percentage of children on VGB who experience visual field loss has been estimated

to be 34% (95% CI 25-42%) in comparison to unexposed controls who experience

visual field loss 7% of the time (Maguire et al, 2010). Children in these studies

were 2.5 years – 21 years. These studies describe long term effect in IS

populations on VGB, there has not yet been an accurate description of VGB on

visual fields in infants. This is because perimetry is difficult to perform in infants,

Case 1 – 10 years/ F Case 2 – 15 years / F

Dx Epilepsy since 30 mos IS @ 3 mos, complex partial epilepsy

Previous Rx

Monotherapy with CZ, OX or SV Varying combinations of CO, PB, CZ, SV, CL, LA, and/or CB

Rx at time of event

VGB for 33 months - 2,500 to 2,750 mg/d (for first 5 months together with CZ)

VGB for 57 mos - 1,000 to 3,500 mg/d (as add-on with LA and CB)

Event Bilateral visual field constriction at 33 mos tx

Severe constriction of the peripheral visual fields at 57mos tx

15

especially those with developmental delay as many IS patients have. Two studies

used an altered technique to measure visual fields. The first, Werth and Schadler

(2006), used a test which did not require the participant to understand the

instructions, thus increasing its success in young and developmentally disabled

children. The other, White Sphere Kinetic Perimetry (WSKP) was used by Agrawal,

Mayer, Hansen & Fulton (2009) in VGB treated children aged 1-19 years (median

6 years). WSKP provided better compliance rates than Goldmann perimetry (28/31

vs 9/31). There was acceptable agreement between both tests in 9 VGB treated

patients and in 10 control subjects (ages 4-7 years). Caution was advised in

interpreting the results as those VGB treated patients who could not complete

Goldmann perimetry tended to have smaller fields than those who did cooperate

(Agrawal et al., 2009). The number of children that could not be included because

of difficulty with perimetry is listed for each study in table 4. In one study, patients

less than 6 years old or with mental handicap were excluded from visual field

analysis (Ascaso, Lopez, Mauri & Cristobal, 2003).

In order to monitor vision in this population, electroretinograms are routinely

used. When VGB was re-approved for use in the USA, it was done under the

SHARE (Support, Help and Resources for Epilepsy) program, where physicians

must take part in the REMS (Risk Evaluation and mitigation Strategy). On top of

requiring routine assessments of effectiveness, this strategy also incorporates

baseline and regular vision monitoring using ERGs when and where possible.

16

Several additional studies have shown results of both visual field and ERG

monitoring in this population (Agrawal et al., 2009; Gaily, Johnson, & Lappi, 2009;

Camposano, Major, Halpern & Thiele, 2008). Monitoring in infants less than 2

years old who cannot complete perimetry was also recommended in a recent

review by Sergott (2010).

Table 4. Summary of studies investigating visual field defects in children on VGB using perimetry

Author (year) Design Control grp

Test Method

Mean VGB duration

% VFD (exp’d)

Cant do VFs

Wohlrab (1999)

CS Y G -- 42% 92%

Gross-Tsur (2000)

CS N G & HF 3.0 years 65% 29%

Ianetti (2000) CS N G & HF -- 19% 30%

Pelosse(2001) CS N G 3.4 years 55% --

Roccella (2001)

L N HF -- 33% --

Vanhatalo (2002)

L N G 2.2 years 24% Only inc if could do G

Spencer (2003)*

CS N G & HF 3 mos-9 years

36% 72%

Ascaso(2003) L N HF 3.5 20% --

Pojda-Wilczek (2005)

L N HF -- 53% --

Werth (2006) CS Y G -- 33% 46%

You (2006) L N HF 4.0 years 22% Only inc if

17

could do HF

Wild (2009) 8-12 y only

L Y G + HF 29.3 mos 20% 91-92.7%

Camposano (2008)*

CS N G 17.7 mos resp. 6.3 mos Non-resp

4% 58%

Gaily (2009)* CS N G 21.0 mos 7% 0

Agrawal (2009)*

CS Y WSK + G

Not reported

WSK – (29%)

71% (G) 10% (WSK)

Legend: grp – group, exp’d – exposed, VFD – Visual Field Defect, VFs – Visual Field Testing, CS – Control Study, L – Longitudinal, Y – Yes, N – No, G -Goldmann Perimetry, HF – Humphrey Field Analyzer, mos – months, inc - included, resp – responder, WSK – White Sphere Kinetic Perimetry

18

Vision 3

Figure 2. Schematic diagram of the retina (Webvision, http://webvision.med.utah.edu/, Simple organization of the retina, available under a Attribution, Noncommercial, No Derivative Works Creative Commons License © 2011)

To understand the principles of electrophysiology, it is important to first

understand retinal anatomy and physiology. Figure 2 depicts the anatomy of the

retina. When an object is viewed, light enters the eye, and is refracted by the

cornea, aqueous humour and lens, creating a focused image on the retina

(assuming no refractive error). Cells in the retina respond to changing patterns of

illumination. The first synapse is at the back of the retina where the

photoreceptors lay abutting the retinal pigment epithelium. There are two different

types of photoreceptors – rods and cones, named as such for their morphology

Outer Nuclear Layer

Outer Plexiform Layer Inner Nuclear Layer Inner Plexiform Layer Ganglion Cell Layer

19

(see Figure 2). Rods are responsible for vision in low level light whereas cones are

responsible for colour vision and vision in lighted conditions. In the outer segment,

the segment closest to the retinal pigment epithelium, both rod and cones contain

a visual pigment called a chromophore made up of opsin and retinal. In the dark, a

steady current of Na+ flows into open channels cause both rod and cone

photoreceptors to be partially depolarized to a resting potential of -40 mV.

Depolarized photoreceptors release glutamate. When light stimulates

photoreceptors, a photon is absorbed causing retinal to photoisomerize from 11-

cis form to an all trans active form, leading to a conformational change. This also

causes the closure of cGMP-gated cation channels of the photoreceptor membrane,

leading to the hyperpolarization of photoreceptor cell membrane and stopping the

flow of glutamate. This hyperpolarization activates the dendrites of bipolar cells

and horizontal cells. There are 11 types of bipolar cells, 10 are for cone inputs,

and 1 type for rod inputs. Each bipolar dendrite may be in contact with between 2

and 20 cones or up to 50 rods depending on retinal location. Different types of

bipolar cells have different receptors of glutamate. Bipolar cells connecting with

rod photoreceptors and some cone bipolar cells depolarize in response to an

increase in retinal illumination. These are “ON” bipolar cells. In the cone pathways

there are also bipolar cells which depolarize in response to decreased retinal

illumination; the “OFF” bipolar cells. The difference in the response results from

differences in how the ON and OFF bipolar cells respond to changes in glutamate.

20

Decreasing retinal illumination increases glutamate release from cone

photoreceptors which results in depolarization of the OFF pathway, whilst the ON

pathway responds in the opposite way to an increase in glutamate (Werblin &

Dowling, 1969; Werblin, 1991).

Simultaneously, horizontal cells receive inputs from many rod and cone

photoreceptors. In response to a target with a dark centre and light surround,

glutamate release from the central cones will increase resulting in reduced activity

to “ON” bipolar cells. At the same time the cones responding to light surround will

decrease glutamate release. Upon receiving reduced flow of glutamate, horizontal

cells reduce the amount of GABA that they feedback to the cone photoreceptor

which hyperpolarize. The hyperpolarization is greater than the response to a global

decrease in retinal illumination. This is the basis of lateral inhibition which allows

increased resolution to gratings and more complex environments. In the inner

plexiform layer, bipolar axons contact and transmit signals to ganglion cell

dendrites. Ganglion cells also receive input from amacrine cells, another form of

interneuron, and transmit the summation of these signals to the brain, specifically

to the primary visual cortex, via the optic nerve.

21

Electroretinography 4

Changes in visual function have also been measured using visual

electrophysiological tests in adults and results correlated to measures of visual

fields. The electroretinogram is a visual electrophysiological test. Electrophysiology

uses electrodes to measure the electrical activity produced by the body. All visual

electrophysiological tests use changing visual stimuli to elicit electrical responses.

The electroretinogram measures the response of the retina in response to light

stimulation. Specifically, these responses may be generated by retinal neurons or

‘as a result of the effect on retinal glia of changes in extracellular potassium

concentrations brought about by the activity of these neurons’ (Frishman, 2006).

Depending on the stimulus conditions used, ERGs can provide information from

different parts of the retina. Both the rod and cone systems can be tested. The

rod system is tested in a dark-adapted state, this is called scotopic testing. The

cone system is tested in a light-adapted state, this is called photopic testing. The

International Society of Clinical Electrophysiology of Vision (ISCEV) provides

standard conditions under which the scotopic and photopic responses of the ERG

are recorded (Marmor et al., 2009).

1) Dark adapted 0.01 ERG – maximal response of dark adapted retina to dim light

(scotoptic)

22

2) Dark adapted 3.0 ERG – maximal response of dark adapted retina to bright

light (scotopic)

3) Dark adapted 3.0 Oscillatory potentials – scotopic and photopic wavelets on

characteristic waveforms

4) Light adapted 3.0 ERG – response of light adapted retina to bright light

(photopic)

5) Light adapted 3.0 flicker ERG – response from light adapted retina to 30 Hz

light (photopic)

It is common practice for clinicians to test more than the basic 5 tests which will

be described below. A list of the conditions used in our lab to clinically monitor

patients is provided in Appendix 1 including details of stimuli strength, recording

time, length of flash and filter parameters. The test goes through 14 different

steps and takes approximately 45 minutes to complete.

The ERG is a complex waveform over time composed of characteristic peaks and

troughs. This project investigates changes in the cone system and thus this study

uses the light adapted, photopic ERG. The form of the response follows the

physiologic processes that are responsible for vision. The retina is responsible for

converting light energy into a neural response: this process is known as

phototransduction. The two major components of the ERG waveform are the a-

and b- waves.

23

Figure 3. Light adapted 3.0 ERG.

Legend: a- a-wave, b- b-wave, i – i-wave

As in figure 3, a waveform from a light adapted 3.0 ERG, the first negative peak is

the a –wave. The leading edge of the a-wave corresponds to the hyperpolarization

of cone photoreceptors in response to a flash stimulation. The a-wave also

includes contributions from the OFF pathway. ‘The a-wave is truncated by the rise

‘of b –wave. In order to describe photoreceptor function, Lamb and Pugh

developed a way to relate the leading edge of the a-wave to the leading edge of

the photoreceptor response to light (Lamb & Pugh, 1992; Pugh & Lamb, 1993).

Hood and Birch adjusted this model to take into account the biochemical

transduction cascade to model the capacitance of the cone membrane (1995). To

assign appropriate values for Td (1-5 ms) and tau (1-5ms), these parameters

were varied using control data to find the minimum Root Mean Square of the fit

and these values were used throughout. A Td of 3.3 and tau of 5 ms were used.

221

Time (ms)

1

Am

plitu

de (

uV)

a

b

i

24

The next positive peak is the b-wave which is generated by the depolarization of

ON bipolar and Muller cells. The b-wave is followed by another negative going

wave. The trough of this wave is called the photopic negative response (PhNR);

there is however a brief spike on the negative wave called the i-wave – which is

believed to be generated by off bi-polar cells (Frishman, 2009; Sieving, Murayama

& Naarendorp, 1994; Xu & Karwoski, 1994, 1995). The photopic negative

response is believed to be a response of the spiking activity of retinal ganglion

cells and their axons. Experiments in monkeys and cats have shown that using

tetradoxin (TTX), which is known to block the Na+ dependant action potentials

that occur in all ganglion cells and some amacrine cells, the PhNR response is

eliminated (Viswanathan, Frishman, Robson, Harwerth, & Smith, 1999).

When using a long duration stimuli (>100ms), an i-wave is usually not present,

however a d-wave can be seem at light offset. This wave represents the ‘transient

depolarization of hyperpolarizing cone bipolar cells in combination with the positive

going termination of the cone photoreceptor response. Sieving and colleagues

demonstrated that 2,3 piperidine dicarboxylic acid (PDA), which blocks responses

of off bipolar cells, reduced or eliminated the d-wave at light offset (1994). PDA

also eliminated the i-wave in the brief flash ERG (Sieving et al, 1994).

The light adapted 3.0 flicker response is also a photopic response but is elicited by

a very quick, 30 Hz frequency repeated flash and thus does not resemble the other

25

photopic responses. The resultant 30 Hz response can be reduced to implicit time

and amplitude of the response. It is measured from the peak (top of the wave) to

the trough (bottom of that wave). It is the amplitude, the vertical distance

between these two points, which is of interest for the current study. The flicker

response represents the integrity of the whole cone pathway, including cone

photoreceptors and postreceptoral cells (Bush & Sieving, 1996). Using 2-amino-4-

phosphonobutyrate (APB) in a macaque monkey model, the b-wave was

eliminated from the dark adapted 3.0 ERG but the flicker response was still

present, albeit greatly reduced. They identified a residual contribution from the d-

wave. When PDA was used in addition to APB, both the b-wave and the flicker

response were eliminated (Bush & Sieving, 1996).

The amplitude of the flicker response has been shown to be related to visual field

loss due to vigabatrin. In adults with CPS, a flicker amplitude <52 uV was 100%

sensitive and 75% specific for detecting visual field loss (Harding, 2000a). It has

been reported to decrease in a proportion of infants on vigabatrin (McCoy, Wright,

Weiss, Go, & Westall. 2011; Westall et al., 2002; Spencer & Harding, 2003).

Monitoring Retinal Function in IS Patients with 5ERGS

ERGs have been used in adults and children to monitor changes due to vigabatrin.

Table 5 summarizes all studies reporting ERG results in adults on vigabatrin, and

26

where possible, the relation to visual field loss (VFL).

Table 5. Summary of ERG findings in adults taking VGB.

Reference Pts with Ab ERG

ERG abnormalities, amount with ERG Ab and VFL

Krauss (1998) 4/38 Reduced inner retinal cone response, reduced OPS + VFL (3/4)

Arndt (1999) 9/20 Reduced Ops only + VFL (7/9)

Daneshvar (1999)

4/10 Ab scotopic (4/4), Ab photopic (1/4), VFL (4/4)

Lawden (1999) 0/31 VFL (12/31)

Harding (2000a)

5/8 Reduced flicker amp, delayed flicker b wave latency, VFL (5/5)

Harding (2000b)

18/18 Latency of the 3.0 flicker b-wave and a–b amp, <52 uV flicker amp – VFL (18/18)

Miller (1999) 32/32 Reduced OP amps (32/32), reduced photoreceptor sensitivity (9/32), reduced rod and cone b waves (32/32), reduction in cone flicker responses correlated strongly with the degree of VFL as measured by kinetic perimetry, 7/20 VFL

Hardus (2001) 15/30 b-wave abnormalities in 15 pts with VFL

Coupland (2001)

18/76 eyes

Reduced 3.0 flicker amp (18/76), reduced photopic (30/76) and scotopic b wave (30/76), reduced Ops (22/76), VFL not measured

Ponjavic (2001)

7/12 Reduced 3.0 flicker amp (7/7), reduced photopic b-wave amp and latency (4/7), VFL (7/7)

27

Legend: pts – patients, Ab – Abnormal, VFL – Visual field loss, Ops – Oscillatory potentials, amp – amplitude, IT – Implicit time, inc - inclusion

Our group has previously published ERG results in infants. In a case series, we

showed peripheral retinal nerve fibre layer atrophy associated with decreased 3.0

flicker response on 3/3 patients (Buncic et al., 2004). Currently in our lab, we

define an abnormal 30 Hz flicker as a significant reduction (more than accounted

for by variation) in age expected values of flicker amplitude on two consecutive

tests (testing conducted at 3-5 month intervals) (McCoy et al., 2011). The results

Besch (2002) 18/20 Altered OPs (18/18) – 18/18 VFL, delayed b-wave (6/18) – 6/6 VFL

Jensen (2002) 9/10 Abnormal responses both in scotopic, photopic conditions and in OPs (5/9), 3/9 – VFL

Van Der Torren (2002)

11/19 Correlation between reduced cone and rod b-wave amp (11/19) and OP latency (2 & 3) and VFL, 20/29 VFL

Comaish (2002)

14/14 Reduced cone b-wave amp, reduced OP amp correlated with VFL (8/14)

Hardus (2003) 9/11 Reduced photopic b wave amp (7/11) related to amount of VFL, Decreased photopic a wave latency (9/11), 11/11 VFL

McDonagh (2003)

19/32 Reduced 3.0 flicker amp, OP2, cone a wave IT, cone b-wave amp significantly different in VGB with VFL vs. without, (19/32) VFL

Bourcier (2004)

12/12 Reduced flicker amplitude, 12/12 VFL

Kjellstrom (2008)

8/8 (inc. criteria)

Reduced 3.0 flicker amp and latency, reduction in amp of rod cone resp, reduced rod isolated response, VFL 8/8

28

of this study, and others investigating ERG changes in children taking vigabatrin

are summarized in table 6.

We have found that baseline measurements in infantile spasms can be variable.

Some infantile spasms patients have an abnormal result compared to normally

developing children at baseline, before they have been initiated on anti-epileptic

medication (McFarlane, Westall & Wright, 2011). The pre-existing abnormality of

ERG and contrast sensitivity (Mirabella et al., 2007; Morong et al., 2003) in

children with infantile spasms also suggest that the retina may be a peripheral

marker of GABAergic dysfunction in these children. Therefore, the Westall lab

compares changes in ERG with each child`s results before starting the drug.

Toxicity is defined as two consecutive occasions of reduced (greater than typical

inter-visit variation) flicker amplitude from that individual child’s baseline value.

29

Table 6. Summary of ERG results in children and infants taking vigabatrin

Reference Subjects Age

Length of VGB Tx

Visual field findings

ERG findings

Gross-Tsur (2000)

3.5-18 years (mean 13 + 4.1)

3.0 +1.6 years

11/17 VFD 6 ND ERG 2 Ab ERG 3 N ERG

4/11 Ab ERG reduced amp of cone, rod of mixed cone-rod b wave

Pelosse (2001) 9.6 years 41 months

7/14 VFD 6/7 ND ERG 1/7 N ERG

3/5 Ab ERG Reduced b wave amp, raised a/b ratio

Westall (2002) 1.5-180 months at first visit

0-18 months

ND Reduced flicker amp, reduced cone b wave amp

Harding (2002) 3-15 years - 18/26 VFD Photopic a wave IT, 3.0 flicker IT, Reduced 3.0 flicker assoc. with severe VFD

Spencer (2003) 3-15 years - 4/11 VFD 8/26 Ab ERG Reduced 3.0 flicker, Ab OPs 3.0 flicker <70uV = 75% sensitive, 71.4% spec for detecting VFD

Westall (2003) 5-26 months (baseline)

5-42 months

ND 17/17 Ab ERG

Reduced photopic OPs 2+3, improvement in 12/17 when VGB stopped

Buncic (2003) 2.5-15 years

21 months – 6 years

3/3 VFD 3/3 reduced 3.0 flicker amp, 1/3 reduced cone response

30

Legend: Tx – treatment, VFD – visual field defect, ND – Not done, Ab – Abnormal, N – Normal, amp – amplitude, IT – Implicit time, assoc. – associated, R – maximum response, pts – patients, VF - visual field testing

The current research was designed to reduce the pre-test variability and increase

sensitivity and specificity for an electrophysiological test for VA- function loss by

using tests that isolates targeted areas of the retina. The 30 Hz flicker response is

a measure of integrity of the entire cone pathway. We have identified three ERG

tests used to isolate different parts of the cone pathway: cone a wave modeling,

the photopic negative response and the photopic off response. The first, cone a

wave modeling, isolates the response of the cone photoreceptors. The photopic

1/3 delayed mixed rod cone

Pojda-Wilczek (2005) Abstract only

8-20 years old

- More than half VFD

Decreased or borderline b-wave amp “after flicker 3.0”

Eklund (2006) Abstract only

Med 22 months (6-69 months)

- ND R rod and R cone reduced in majority of pts, photopic b wave IT prolonged

Camposano (2008)

- - 1/25 VFD 1/20 Ab ERG, Ab photopic and scotopic responses, prolonged 3.0 latency (no VF testing)

McCoy (2011) 1 month – 18 years

- ND 18/160 patients reduced flicker amp on 2 consecutive occasions

Kjellstrom (2011) Abstract only

12-228 months

- ND Reduced b wave amp of rod, rod-cone and 3.0 flicker, altered rod-cone a wave IT and amp

31

negative response, is thought to isolate the response of ganglion cells. The last,

the photopic off response is a marker of cone off bipolar cells. The aim of this

study is to observe whether changes in any of these parameters contribute to the

decreases observed in the 3.0 flicker amplitude with VGB use. If we can identify a

particular part of the cone pathway that is reduced, this may help to elucidate a

possible mechanism of the toxicity and may represent a more specific marker of

changes due to VGB.

Lastly, in a long flash protocol, the d wave may be present. This represents off

bipolar cells and gives an easier way to estimate its response (from short flash)

because it is not a part of the negative wave after the b-wave. It is the peak after

the photopic negative response when using a long flash. It is measured as the

difference in amplitude from 200ms (the time at with the light stimulus is turned

off) to the highest peak following this point.

32

Methods

Subjects and Controls Patients 6

6.1 Subjects

Infants taking prescribed vigabatrin at the Hospital for Sick Children (Sick Kids)

are routinely followed by the Ophthalmology clinic for signs of vigabatrin toxicity.

Infants are scheduled for one test at baseline (within 2 weeks before or after

starting the drug), and follow-up testing every 3-4 months while on vigabatrin and

at least one test once the drug is discontinued. The parents of all patients taking

vigabatrin referred to electrophysiological testing were approached for consent to

participate in the study. Between sedation and the start of testing, the study was

explained to interested parents and questions were taken. If the infant met

inclusion criteria listed below and agreed to participate in the study, they were

asked to sign a consent form (Appendix 2 & 3). Those who consented agreed to

allow the additional steps required in the study to be appended to the ERG they

had scheduled for that day as well as any future ERG tests scheduled by their

ophthalmologist.

6.1.1 INCLUSION CRITERIA

• Diagnosed Infantile Spasms

• Recently on (<2 weeks) or expected to start on vigabatrin

• Four years of age or younger at study enrolment

33

6.1.2 EXCLUSION CRITERIA

• Other retinal disease

• Prematurity <30 weeks

• History of drugs known to affect the retina (except seizure medication)

6.2 Control Patients

The parents of children with idiopathic motor nystagmus (EOHN) referred for

electrophysiological testing were approached for consent to participate in the

study. Consent was obtained in the same way as for the subjects. It was explained

to parents that the data would only be used if their child was found to have all

parameters in the standard ERG falling within normal age limits. These patients

were only seen for one testing.

6.2.1 INCLUSION CRITERIA:

• Four years of age or younger at study enrolment

6.2.2 EXCLUSION CRITERIA:

• Known retinal disease

• retinal origins of EOHN

• Prematurity <30 weeks

• History of drugs known to affect the retina (except seizure medication)

34

Testing & Data collection 7All testing was performed in the Visual Electrophysiology Unit at Sick Kids.

Approval for the study was obtained from the Research Ethics Board at Sick Kids.

Consenting patients underwent current standard clinic protocol for ERG of children

under four years of age and where possible, visual acuity was assessed using

Teller acuity cards or Cardiff acuity test. Information about drug dosage,

concurrent drug usage, co-morbidities, age at diagnoses, etiology of seizures etc.

was collected from electronic patient charts (See Patient Data Collection form,

appendix 4). ISCEV Standard ERGs were performed on subjects: at baseline

(before initiating VGB) and at 3-6 month intervals on the drug and after

discontinuation and on controls. Infants were sedated by a sedation nurse using

chloryl hydrate (80 mg.kg body weight; maximum dose 1g). Pupils were dilated

with 1% cyclopentolate and 2.5% phenylephrine in children and with 0.5%

cyclopentanate in infants <4 months old. Subjects were dark-adapted for 30

minutes using both adhesive eye patches and plastic eye patches. After 30

minutes dark adaptation and once the infant was reliably asleep, they were taken

to the ERG exam room. In a dark adapted room, patches were removed and

appropriately sized Burien-Allen electrodes were placed on the cornea. A reference

electrode was placed on the centre of the forehead using electrode paste. ERGs

were recorded using a Ganzfield ColourDome stimulator. Data were collected using

the Espion software. For the purposes of this study, two stimulus conditions were

35

added to our lab’s standard protocol (See Appendix 1). This extended the time of

the test, which typically takes 45 min – 1 hr, by four minutes (at the most).

Details of the added stimulus conditions are below in table 7. If the child woke up

during the procedure and could not be coaxed back to sleep, the additional steps

were abandoned. Infants were monitored throughout testing for vital signs by the

sedation nurse and for adequate light exposure by the camera mounted in the

dome, visible to the tester.

At the end of testing, electrodes were removed and the infant was brought to an

ophthalmologist for sedated fundus examination.

Table 7. Additional stimulus condition testing parameters

Legend: λ – wavelength, PhNR – photopic negative response, cd – candela, avg’d – averaged, scot – scotopic

Back-ground λ

Background Intensity

Stimulus λ

Stimulus Intensity (cd.s/m2)

Stimulus Duration

# trials avg’d

Long Flash

White 30 cd / m2 White 1250 200 ms 3

PhNR Blue (465nm)

100 scot cd/m2

Red (635nm)

6.3 <5 ms 10

36

7.1 Scoring

Each step was assessed and any blinks or noisy recordings were removed.

Average waveforms were produced by the Espion software.

7.1.1 Flicker

Flicker amplitude was hand scored from peak to trough. Amplitude (y-axis) and

implicit time (y-axis) were recorded (figure 4).

Figure 4. 3.0 Flicker response from IS patients

7.1.2 Photopic negative response

Amplitude and implicit time of photopic negative response were hand-scored by

selecting the lowest trough after the b-wave, as shown by the arrow in figure 5

below and recorded in a coded excel spreadsheet. The amplitude of the a-wave,

the first negative peak, was also measured and recorded.

-100

-150

Am

plitu

de (

uV)

Time (ms)

37

Figure 5. Photopic negative response from

patient.

Arrow identifies trough of response, where

amplitude is measured as the difference from

zero.

7.1.3 Cone Photoreceptors Parameters

Data from three ERG steps using a white flash stimulus on a white background

(already a part of clinic protocol – details appendix 1) were extracted to Microsoft

excel from the Espion software. Using a program developed in R statistics software

(appendix 5), these data were used to fit waves modeling the leading edge of the

a-wave. The R program automatically detects the peaks and troughs of the a-wave

at each intensity for each test and uses those values in the Hood and Birch model.

This model (see Figure 6) was developed by Hood and Birch, adapted from Lamb

and Pugh, to generate values for R (maximum response) and S (sensitivity). If the

model was not able to provide an accurate fit for a test (defined as fit >0.5, where

1.0 is worst, 0.0 is best), these data were examined individually. If blinks or other

artifacts were present, these data were excluded. If not, the data were re-run

200

100

0

-100

-200

Time (ms)

Am

plitu

de (

uV)

0 50 100 150 200

38

individually without automatic peak detection. If this resulted in an adequate fit

the data were included.

Figure 6. Hood & Birch equation to describe the leading edge of the a wave.

7.1.4 Cone Off Response

Cone off response (see Figure 7), was initially measured as the difference between

the amplitude at 200 ms (point b), when the light stimulus turns off, to the top of

the peak following 200 ms (point c). Waves were only scored if the amplitude at

200ms (point b) was higher than that of the a-wave (point a) as to not include

those results in which the patient blinked as per a protocol used by Horn et al.

(2011).

)exp(-t/ * ))²]}Rt-(t5.0exp[1({),( CONEd τCONEIStiR −−=

0 100 200 300

-100

0

100

200

-200

(a) (b) (c)

Am

plitu

de (

uV)

Time (ms)

Figure 7. Cone off response.

Left most arrow (a) identifies a-

wave, middle arrow (b) identifies

amplitude at 200ms, right most

arrow (c) identifies d-wave where

off response is measured. Dotted

line indicates amplitude of a- wave.

39

Statistical Analysis 8

The ERG is still developing in infants up to about 2 years of age. To account for the

effect of age in these previously unstudied markers in an infant population,

developmental curves were developed. The data from each patient was plotted

against the age (in months) at baseline for each parameter (flicker, PhNR, Cone S,

Cone Rmax, Cone off). The growth curve for each group was explained by a

logarithmic regression function. The developmental curves for subjects and

controls for each parameter were compared.

Using the growth curves, all data were converted to a value that represented the

difference from the value expected for age. The value expected at the age of test

was derived from the logarithmic developmental curve and then subtracted from

the observed value at each test point, giving the age-adjusted value.

Some functions increase positively with age, including flicker amplitude, Cone Off

and Cone S. In these cases, a negative value represents a reduction from normal.

The other parameters, Cone R and PhNR, increase in a negative direction with age.

In these cases, a positive value represents a reduction from normal.

Evidence from animal research indicates that functional changes in the ERG after

vigabatrin treatment are often a result of structural changes in the retina. While

these ERG changes may be reversible, from a clinical perspective the interest lies

40

in detecting those ERG changes that are related to the irreversible visual field loss;

we would expect these to persist even after the drug is discontinued. Therefore,

vigabatrin toxicity was defined as any significant change in a targeted biomarker

from expected seizure-affected development that persists after drug

discontinuation. Adverse drug reactions (changes in biomarkers while on the drug

that disappear when the drug is discontinued) will also be documented. If changes

do not persist after discontinuation, this suggests that electrophysiological markers

have identified changes before structural damage or another different mechanism

of toxicity than that proposed from animal models.

We investigated whether individual markers changed over the course of drug

treatment using ANOVA for each parameter. Then, for those parameters that were

affected by vigabatrin use, we investigated whether changes in these markers

were associated with a change in flicker amplitude using linear models (using

continuous value or normal / abnormal). For any markers, or characteristics (time

on drug, age of seizure onset etc), that were associated with flicker abnormalities,

they were used as a diagnostic tool to investigate: abnormalities at baseline,

survival curves, whether changes were permanent or reversible.

41

Results

Group Demographics 9

9.1 Subjects

Group demographics for subjects are presented in table 8. Thirty one children

with Infantile Spasms (IS) were tested using ERGs. Their ages at first test ranged

from 3.18-23.6 months, with a mean age of 9.6 months (median 8.6 months).

There were 18 male and 13 female subjects tested. Baseline testing was done

between 0-34 days after initiating vigabatrin (mean 10 days, med 7 days). Two

subjects did not have a baseline test because they were not followed by Sick Kids

and thus were only referred for ERG testing at 7 or 11 months after initiating

vigabatrin.

Age of Infantile Spasms diagnosis ranged from 2.7 – 23.4 months, mean 8.7

months (median 8.4). The age that seizures were first noted was not recorded in

all cases, for those that were, time between first noticed seizure and diagnosis of

IS ranged from 0-10.49 months, mean 1.82 months.

At baseline, subjects underwent ophthalmoscopy exams and behavioural visual

assessments. Ophthalmoscopy results included an examination of the fundus,

macula and disc. Results of ophthalmoscopy were normal.

42

Table 8. Group demographics of patients with IS at baseline

STUDY ID

Sex Age of 1st sx (mos)

Age dx IS (mos)

Age VGB init

(mos)

Co-morbidities

VIE01 F * 7.38 7.38 *

VIE02 M * 9.21 9.57 *

VIE03 F 3.31 3.87 3.87 Bilateral perisylvian polymicrogyria

VIE04 M 11.93 12.43 12.43 Developmental delay

VIE05 F * 4.26 5.48 *

VIE06 M * 7.28 7.28 Trisomy 21

VIE07 M 5.61 8.85 8.85 Delayed milestones

VIE08 M 13.02 16.03 16.03 Trisomy 21

VIE09 M 5.74 5.84 5.97 Developmental delay, abnormal MRI, lissencephaly

VIE10 M * 8.36 8.79 Twin A

VIE11 M 17.74 18.46 18.56 Neonatal sx

VIE12 M 11.93 13.67 13.67 L MCA, ACA stroke 2* to embolism, Galen malformation

VIE13 M * 4.16 4.16 Delayed milestones

VIE14 M 22.13 23.38 23.38 *

VIE15 F 5.08 5.87 5.34 Lissencephaly

VIE16 F * 8.49 8.49 Trisomy 21,

VIE17 M * 11.11 11.80 36 wks gestation, cyst in brain post infection, delayed milestones

43

VIE18 M * 8.07 8.07 L MCA infection

VIE19 M 6.30 6.82 6.89 *

VIE20 M * 8.85 8.85 35 wks gestation

VIE21 F * 2.66 3.02 Lissencephaly

VIE22 F * 7.25 7.38 Tuberous sclerosis

VIE23 F * 6.00 6.00 37 wks gestation

VIE24 F * 8.16 8.16 Trisomy 21

VIE25 F 3.54 4.10 4.20 36 wks gestation

VIE26 F 9.64 9.64 9.64 *

VIE27 F * 9.67 9.67 Failure to thrive, developmental delay, dextrocardia

VIE28 M 5.51 5.77 5.77 *

VIE29 F * 6.23 6.23 CP 2* to HIE

VIE30 M * 9.15 9.15 32 wks gestation, sturge webber syndrome

VIE31 M 6.00 16.49 16.49 CP, developmental delay, interventricular hemorrhage, microcephaly, non-IS sx

Legend: * - not reported or unknown; Wks- weeks; sx – seizure; mos – months; M – male; F – female; init – initiated; MCA- middle cerebral artery; ACA- Anterior cerebral artery; 2*- secondary; L – left; CP-Cerebral Palsy; HIE – Hypoxic ischemic event

Subjects were tested within one month of starting Vigabatrin using sedated ERGs.

This data was used to create developmental curves. All fundus exams were

44

normal. Drug and Visual acuity information at baseline testing is provided in table

9.

Table 9. Drug history and visual acuity in Infantile Spasms patients at baseline.

STUDY ID

Mos on VGB

Age at test

Other AED meds

Dose of VGB VA (Binoc) logMAR

VA test

VIE01 0.23 7.61 600 mg BID No co-op

VIE02 0.23 9.80 800 mg BID No co-op

VIE03 0.13 4.00 CO, PH 600 mg BID NT

VIE04 0.07 12.49 600 mg BID >1.6 T

VIE05 0.23 5.70 625 mg bid LP only

VIE06 0.23 7.51 600 mg bid NT

VIE07 0.16 9.02 1.1 T

VIE08 0.00 16.03 625 mg bid 1.3 T

VIE09 0.26 6.23 750 mg bid No co-op

VIE10 0.82 9.61

500 mg am, 750 mg pm

NR

VIE11 0.23 18.79 CO 1000 mg bid 0.3 C

VIE12 0.26 13.93 650 mg bid 0.1 C

VIE13 CO 750 mg bid 1.1 T

VIE14 0.26 23.64 650 mg bid 0.8 T

VIE15 0.33 5.67 450 mg bid NR

VIE16 0.26 8.75 600 mg bid Fixes and

45

follows

VIE17 1.11 12.92 1000 mg bid 1.1 T

VIE18 0.23 8.30 ACTH 625 mg bid 1.1 T

VIE19 0.13 7.02 750 mg bid 1.3 T

VIE20 0.59 9.44 600 mg bid NT

VIE21 0.16 3.18

PH, ACTH 500 mg bid Fixes and follows

VIE22 0.20 7.57 720 mg bid 1.0 T

VIE23 0.23 6.23 650 mg bid 1.0 T

VIE25 0.43 4.62 750 mg bid NR

VIE26 0.39 10.03 LE 500 mg bid >1.1 T

VIE27 0.20

9.87

375 mg bid Slow to pick up fixation

VIE28 0.46 6.23 750 mg bid 0.2 C

VIE29 0.66 6.89 300 mg bid NT

VIE30 0.46 9.61

650 mg bid No attention

VIE31 1.05 17.54 PH, LE 725 mg bid >1.1 T

Legend: mos – months; AED – anti-epileptic drug; VA – visual acuity; binoc – binocular; T – Teller, C – Cardiff, co-op – cooperation, NT – Not tested; cpd – cycles per degree; LP – Light Perception; NR – No response; CO – Clobazam; LE – Levetiracetam; ACTH –Adrenocorticotrophic hormone; PH – Phenobarbital

46

9.2 Controls:

Group demographics for controls are presented in table 10. Seventeen children

were tested in the VEU to rule out retinal reasons for early onset horizontal

nystagmus. Six were tested prospectively and data from eleven patients were

gathered retrospectively. Three patients were excluded because ERGs showed

retinal abnormalities that were responsible for their nystagmus. Thus, 9 male and

4 female patients, ages ranging from 5.6 – 47.02 months, with a (mean age of

18.7 months) were included.

At baseline, visual acuity was assessed when possible and subjects underwent

cycloplegic exams and ophthalmoscopy exams.

Ophthalmoscopy results included an examination of the fundus, macula and disc.

Table 10. Group demographics of controls

ID Referred for Age (mos)

Normal ERG

Comorbidities VA (binoc) logMAR

VA test

C01 EOHN 41.54 No 0.1 C

C02 EOHN with vertical component

13.84 No Left head tilt 0.4 C

C03 EOHN 9.21 Yes 0.2 C

C04 EOHN, photophobia

13.28 Yes 0.0 C

47

Legend: #=retrospective; mos – months; VA – visual acuity; binoc – binocular; EOHN – early onset horizontal nystagmus; C – Cardiff; T – Teller; LE – left eye; RE – right eye; LX(T)- left

C05 R jerk nystagmus

31.80 No Trisomy 21, roving eye mvts, poor visual attention

1.4 T

C06 EOHN, photophobia

23.18 Yes Left head turn 0.1 C

#C07 EOHN, poor fixation, no following

5.93 Yes Plagiocephaly, proximal hypotonia

1.4 T

#C08 EOHN, LX(T) c LH(T)

5.61 Yes 1.7 T

#C09 RE nystagmus

15.51 Yes R strab, limited adduction LE, R face turn, intranuclear ophthalmoplegia

0.2 C

#C10 EOHN 18.36 Yes Trisomy 21 0.2 C

#C11 EOHN 16.26 Yes Twin A, 38 wks gestation, ON hypoplasia, hypopigmented fundi

0.2 C

#C12 EOHN 15.51 Yes Delayed milestones 0.1 C

#C13 EOHN 14.85 Yes R face turn 0.0 C

#C14 EOHN 10.39 No 1.3 T

#C15 Nystagmus LE

14.03 Yes L cataract extraction, anterior vitrectomy, wears L contact lens

1.0 T

#C16 EOHN, photophobia

22.07 Yes LE only

LET, decreased stereopsis 1.6 T

#C17 EOHN 47.02 Yes AHP c chin elevation, high hyperopia

0.1 C

48

intermittent exotropia ; LH(T) left intermittent hypertropia- ; mvts – movements; strab – strabismus; ON – optic nerve; LET – left esotropia; AHP-abnormal head posture; c –with

Analysis 10

10.1 Developmental curves

In both controls and subjects, flicker amplitude increased with age. Flicker

amplitude increase from baseline approximately 25% over 24 months in patients

(baseline IS) and 15% over the first 24 months in controls (Figure 8a and b).

Three patients have abnormal flicker (falling greater than 45 uV below the

developmental curve). The two curves follow a similar pattern. The development

curve for subjects is only 2-3 uV lower than controls over the entire line.

Controls

Figure 8 a. Control flicker amplitude plotted by age in months. Solid line is the line of best fit and represents developmental curve.

49

In both patients and controls, PhNR amplitude increases with age, this is reflected

in the curve becoming more negative over time (Figure 9a & b). The curves follow

a similar trend though subjects increase approximately 400% over 24 months

while controls increase approximately 240% over the first 24 months. The

absolute value of PhNR is on average between 15 - 18 uV less in patients than

controls at any point along the curve.

y  =  8.82ln(x)  +  67.629  R²  =  0.01761  

0  20  40  60  80  

100  120  140  160  180  

0.00   10.00   20.00   30.00   40.00   50.00  

Flicker  a

mplitu

de  (u

V)  

Age  (months)  

Subjects  

Figure 8 b. Baseline IS flicker amplitude plotted by age in months. Dashed line is the line of best fit and represents developmental curve.

50

y  =  -­‐16.19ln(x)  +  17.821  R²  =  0.42643  

-70

-60

-50

-40

-30

-20

-10

0

10

20

30

0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00 45.00 50.00

PhNR  Am

plitu

de  (μ

V)  

Age  (months)  

Controls  

Figure 9 a. Control PhNR amplitude plotted by age in months. Solid line is the line of best fit and represents developmental curve.

y  =  -­‐15.46ln(x)  +  24.628  R²  =  0.12962  

-­‐70  

-­‐60  

-­‐50  

-­‐40  

-­‐30  

-­‐20  

-­‐10  

0  

10  

20  

30  

0.00   5.00   10.00   15.00   20.00   25.00   30.00   35.00   40.00   45.00   50.00  

PhNR  Am

plitu

de  (u

V)  

Age  (months)  

Subjects  

Figure 9 b. Baseline IS PhNR amplitude plotted by age in months. Dashed line is the line of best fit and represents developmental curve.

51

Patients and controls have quite different development curves for sensitivity

(Figure 10a and b). In controls, there is an increase with age, with the curve

y  =  2.009ln(x)  +  39.241  R²  =  0.01802  

0  

20  

40  

60  

80  

100  

120  

0.00   5.00   10.00   15.00   20.00   25.00   30.00   35.00   40.00   45.00   50.00  

SensiQvity  (p

hot  td-­‐

1  s-­‐3)  

Age  (months)  

Controls  

Figure 10a. Control cone sensitivity plotted by age in months. Solid line is the line of best fit and represents developmental curve.

y  =  15.564ln(x)  +  9.8971  R²  =  0.06027  

0  

20  

40  

60  

80  

100  

120  

0.00   5.00   10.00   15.00   20.00   25.00   30.00   35.00   40.00   45.00   50.00  

SensiQvity  (p

hot  td-­‐

1  s-­‐3)  

Age  (months)  

Subjects  

Figure 10b. Baseline IS cone sensitivity plotted by age in months. Dashed line is the line of best fit and represents developmental curve.

52

increasing approximately 5% from baseline over the first 24 months. In subjects,

there appears to be a fairly large effect of age, with the curve increasing 85%

from baseline over 24 months.

y  =  -­‐4.961ln(x)  -­‐  47.343  R²  =  0.07372  

-­‐100  

-­‐90  

-­‐80  

-­‐70  

-­‐60  

-­‐50  

-­‐40  

-­‐30  

-­‐20  

-­‐10  

0  

0.00   5.00   10.00   15.00   20.00   25.00   30.00   35.00   40.00   45.00   50.00  

Rmax  (u

V)  

Age  (months)  

Controls  

Figure 11a. Control cone maximum response plotted by age in months. Solid line is the line of best fit and represents developmental curve.

y  =  -­‐5.711ln(x)  -­‐  47.249  R²  =  0.03371  

-­‐100  -­‐90  -­‐80  -­‐70  -­‐60  -­‐50  -­‐40  -­‐30  -­‐20  -­‐10  0  

0.00   10.00   20.00   30.00   40.00   50.00  

Rmax  (u

V)  

Age  (months)  

Subjects  

Figure 11b. Baseline IS cone maximum response plotted by age in months. Dashed line is the line of best fit and represents developmental curve.

53

y  =  8.82ln(x)  +  67.629  R²  =  0.01761  

y  =  7.3404ln(x)  +  74.923  R²  =  0.09322  

0  

20  

40  

60  

80  

100  

120  

140  

0.00   10.00   20.00   30.00   40.00   50.00  

Flicker  a

mplitu

de  (u

V)  

Age  (months)  

Subjects and controls have similar developmental curves for maximum response

(figure 11 a and b) with maximum response amplitude increasing with age; this is

reflected in the line becoming more negative over time. The absolute value of the

maximum response in subjects increases from baseline approximately 22% over

24 months. Controls maximum response increase approximately 13% from

baseline to 24 months.

Developmental curves for subjects and controls for each marker, overlayed for

comparison are shown in Figure 12 a-d. In each figure, the dotted line represents

subjects with IS and the solid line represents control subjects.

Figure 12. Control (solid line) and subject (dashed line) developmental

curves for, a) flicker amplitude, and b) PhNR Amplitude.

Figure 12 a.

54

y  =  -­‐15.46ln(x)  +  24.628  R²  =  0.12962  

y  =  -­‐16.19ln(x)  +  17.821  R²  =  0.42643  

-­‐70  

-­‐60  

-­‐50  

-­‐40  

-­‐30  

-­‐20  

-­‐10  

0  

10  

20  

30  

0.00   10.00   20.00   30.00   40.00   50.00  

PhNR  Am

plitu

de  (u

V)  

Age  (months)  

y  =  15.564ln(x)  +  9.8971  R²  =  0.06027  

y  =  2.8424ln(x)  +  36.672  R²  =  0.02539  

0  

20  

40  

60  

80  

100  

120  

0.00   10.00   20.00   30.00   40.00   50.00  

SensiQvity  (p

hot  td-­‐

1  s-­‐3)  

Age  (months)  

Figure 12 b.

Figure 12. Control (solid line) and subject (dashed line) developmental curves for c) cone sensitivity, and d) cone maximum response. Figure 12 c

55

4060

8010

012

0

Con

e O

ff R

espo

nse

Ampl

itude

(uV)

Controls Subjects Baseline

Cone off response amplitude did not display any effect of age in either subjects or

controls. In figure 11, box plots display the range (whisker to whisker), median,

(solid line through square), 25th and 75th percentile (bottom and top of square) of

cone off response amplitudes. Controls range from 39 uV – 87 uV (median67 uV).

Subjects range from 49 uV – 118 uV (median 83 uV).

Figure 13. Boxplot comparing cone off response amplitude in controls with IS subjects at baseline. Dots represent outliers (greater than two standard deviations away from the mean).

y  =  -­‐4.961ln(x)  -­‐  47.343  R²  =  0.0737  

y  =  -­‐5.711ln(x)  -­‐  47.249  R²  =  0.03371  

-­‐100  

-­‐90  

-­‐80  

-­‐70  

-­‐60  

-­‐50  

-­‐40  

-­‐30  

-­‐20  

-­‐10  

0  

0.00   5.00   10.00   15.00   20.00   25.00   30.00   35.00   40.00   45.00   50.00  

Rmax  (u

V)  

Age  (months)  

Figure 12 d.

56

Longitudinal data:

Data regarding visual acuity, drug dosage and other AED use in longitudinal tests is provided in table 11. Table 11. Longitudinal drug information and visual acuity in IS patients

STUDY ID

Test #

Age (mos)

Other AED meds

Dose of VGB VA (binoc) (logMAR)

VA test

VIE01 2 11.4 400 mg BID NT T

VIE02 2 15.0 800 mg BID No attention

VIE02 3 20.9

VIE04 2 15.7 TO, CO, LE 1875 mg per day > 1.6 T

VIE04 3 21.7 TO, CO, LE 1875 mg per day NR, fixes and follows

T

VIE04 4 25.3 TO, CO, LE 1875 mg per day >1.4 T

VIE05 2 9.7 CO, LE D/C 1.4 T

VIE05 3 14.1 CO, LE D/C 0.1 C

VIE07 2 12.8 750 mg bid 0.1 C

VIE07 3 18.9 750 mg bid 0.2 C

VIE08 2 19.8 635 mg bid 0.8 T

VIE08 3 24.1 NT

VIE08 4 29.3 D/C 0.1 C

VIE09 2 10.4 625 mg bid NT

VIE09 3 15.4 D/C 0.1 C

VIE10 2 13.2 1250 mg per day 0.4 C

VIE10 3 18.6 D/C 0.2 C

57

VIE11 2 22.4 1000 mg bid 0.2 C

VIE11 3 28.9 D/C 0.2 C

VIE12 2 17.9 TO 750 mg bid 0.8 T

VIE12 3 21.3 TO, CO 750 mg bid 0.9 T

VIE12 4 25.2 CO, VA 750 mg bid 0.8 T

VIE13 1 16.1 CO 750 mg bid 1.1 T

VIE13 2 21.1 250 mg bid NR

VIE13 3 27.3 D/C 1.6 T

VIE14 2 27.5 575 mg bid 0.8 T

VIE14 3 31.2 575 mg bid 0.2 C

VIE14 4 38.1 D/C 0.0 C

VIE15 2 9.9 450 mg bid NLP

VIE15 3 13.5 450 mg bid LP only

VIE15 4 17.2 CO 300 mg bid fixes, no following

VIE16 2 13.1 ACTH (d/c) 200 mg bid 1.0 T

VIE16 3 16.8 D/C 0.3 C

VIE16 4 19.7 D/C 0.2 C

VIE17 2 16.7 1000 mg bid 1.1 T

VIE17 3 20.1 NT

VIE17 4 24.4 1000 mg bid 0.0 C

VIE17 5 28.1 1000 mg bid 0.0 C

VIE17 6 35.9 1000 mg bid 0.2 C

58

VIE17 7 40.3 D/C 0.0 C

VIE19 2 10.2 750 mg bid 0.8 T

VIE19 3 13.6 750 mg bid 0.7 T

VIE19 4 18.0 250 mg bid 0.2 C

VIE19 5 21.9 D/C 0.0 C

VIE20 2 13.1 750 mg bid 1.4 T

VIE20 3 17.2 750 mg bid NR T

VIE20 4 22.1 VA 750 mg bid NR T

VIE21 2 9.0 PH 375 mg bid NT

VIE22 2 11.9 1320 mg per day 0.8 T

VIE22 3 15.9 CL, CZ 1320 mg per day 0.8 T

VIE22 4 20.4 CL, CZ 450 mg bid 0.1 C

VIE22 5 26.1 TO, CO 200 mg bid 0.1 C

VIE22 6 30.3 TO, CO, LE, VA

D/C 0.0 C

VIE23 2 9.9 650 mg bid 0.0 C

VIE23 3 15.6 D/C 0.0 C

VIE25 2 8.5 TO 400 mg bid NT

VIE27 2 17.5 TO, LE 375 mg bid NT

VIE27 3 22.0 TO, LE 200 mg bid no eye movement

VIE28 2 10.1 750 mg bid 0.2 C

Legend: TO – Topiramate; CO – Clobazam; LE – Levetiracetam; VA – Valproic Acid; ACTH –Adrenocorticotrophic hormone; PH – Phenobarbital; CL –

59

Clonazepam; CZ – Carbamazepine; bid – twice daily; NT – Not tested; cpd – cycles per degree; NLP – No light perception; LP – Light Perception; NR – No response; C – Cardiff; T- Teller

Age-adjusted values for flicker, cone S, cone Rmax and PhNR were calculated

using the method described in Methods section 3.0. Time was divided into

timebands in: baseline , 3 months (0-4.5 months on VGB), 6-9 months (4.51-11.5

months on VGB), 12 months (11.51-13.50 months on VGB) and 15+ months

(13.51 +). The longest duration of VGB treatment included in this study was 28.5

months. Off VGB was also included as a category for those children who had ERGs

after stopping VGB treatment. Data are presented as box plots and change over

the course of drug treatment (not including ‘off drug’ time band) was compared

using ANOVA (figure 12 a-d). If there was no effect of the drug, one would expect

that the box plots would centre around zero.

60

Figure 14a. Box plot comparing adjusted flicker amplitude over time on vigabatrin. Dots represent outliers.

Adjusted flicker amplitude decreases over time (P=0.020). In those patients

treated with vigabatrin for 15 months or more, the median adjusted flicker

amploitude is 62uV less than expected for age (figure 12a). There appears to be

recovery in the median after the drug is discontinued, however this difference is

not significant in patients who were on the drug for at least 9 months before

discontinuing vigabatrin (P=0.2).

-100

-50

050

100

Adju

sted

Flic

ker A

mpl

itude

(uV)

Time on Vgb

Baseline 3 mos 6-9 mos12 mos 15 + mos Off vgb

Adj

uste

d Fl

icke

r Am

plitu

de (

uV)

61

Figure 14b. Box plot comparing adjusted PhNR amplitude over time on vigabatrin. Dots represent outliers.

Adjusted photopic negative response amplitude decreases over time (p=0.013)

(figure 12 b). There appears to be recovery when the drug is stopped, as the

median PHNR amplitude in those tested once they have stopped the drug only is 1

uV smaller than expected for age. The difference between the last test on the drug

and off the drug results is not significant for patients treated at least 9 months on

vigabatrin (P=0.6)

-100

-50

050

Adju

sted

PhNR

Am

plitu

de (u

V)

Time on Vgb

Baseline 3 mos 6-9 mos12 mos 15+ mos Off vgb

Adj

uste

d Ph

NR A

mpl

itude

(uV

)

62

-40

-20

020

40

Adjus

ted

Rmax

(uV)

Time on Vgb

Baseline 3 mos 6-9 mos12 mos 15+ mos Off vgb

Figure 14 c. Box plot comparing adjusted maximum response (Rmax) over time on vigabatrin. Dots represent outliers.

Adjusted maximum response increases over time (p = 0.004) (Figure 12 c). This

means that the amplitude is becoming larger over time. When the drug is

discontinued, the median value returns to close to what is expected for age,

however this difference is not significant (P=0.5). This may be due to a drug

effect, which is why it returns to value expected for age when the drug is stopped.

Adj

uste

d Rm

ax (

uV)

63

-50

050

100

150

Adjus

ted

Sens

itivity

(pho

t td-

1 s-

3)

Time on Vgb

Baseline 3 mos 6-9 mos12 mos 15+ mos Off vgb

Figure 14 d. Box plot comparing adjusted cone sentivity over time on vigabatrin. Dots represent outliers.

Adjusted sensitivity decreases over time on vigabatrin (p=0.027) (figure 12 d). In

those patients who are treated with vigabatrin for 15 months or more, the median

sensitivity is 62 phot td-1 s-3 less than expected for age. There appears to be some

degree of recovery when the drug is stopped but the median remains 30 phot td-1

s-3 less than expected for age at that time. The difference between adjusted

sensitivity at last test on the drug and off the drug for patients treated with

vigabatrin for at least 9 months is not significant (P=0.7).

Adj

uste

d Sen

sitiv

ity (

phot

td-1

s-3

)

64

For each response, linear models were used to investigate whether adjusted flicker

amplitude was associated with a normal or abnormal test at any point. An ideal

marker would be assosciated with decreased age adjusted flicker response.

Abnormal versus normal was delineated by establishing the 95th percentile of

distance from the developmental curve at baseline and rounding to the nearest

whole number for all markers except cone off response. As cone off response did

not demonstrate a developmental curve, the 95%ile was calculated based on raw

values. Values for abnormal cutoff points are listed below.

Flicker response was considered abnormal if amplitude was >42 uV less than

expected for age. Photopic negative response was considered abnormal if

amplitude was > 25 uV than expected for age. Cone maximum response was

considered abnormal if amplitude was >24uV than expected for age. Cone

Sensitivity was considered abnormal if value was >40 scot td-1 s-3 less than

expected for age. Cone Off response was considered abnormal if amplitude was

<40 uV. In all cases, where cone off response was abnormal, the peak was

unmeasureable (Figure 15).

65

Figure 15. Cone off response in normal and abnormal test. Left panel shows normal cone off response ( 65 uV). Right panel shows abnormal cone off response in an Infantile Spasms patient.

Abnormal cone off respones were significantly assosciated with reduced age

expected flicker amplitude (p<0.001) (figure 14a). The median adjusted flicker

amplitude in those with an normal cone off response is 0uV (expected for age),

whereas in those with an abnormal cone off response the median adjusted flicker

is -50uV. Furthermore, the maxiumum adjusted flciker response in those patients

with an abnormal cone off response is – 20 uV.

0 100 200 300

-100

-50

0

50

100

Time (ms)

Am

plitu

de (

uV)

0 100 200 300

-100

-50

0

50

100

Time (ms) Am

plitu

de (

uV)

66

Figure 16a.

Adjusted Flicker amplitude for patients with normal vs abnormal cone off response

Adjusted flicker amplitude is not associated with a normal or abnormal photopic

negative response amplitude (P=.49) (figure 16b), or cone maximum response (P

=0.25) (figure 16c).

Figure 16b.

Adjusted Flicker amplitude for patients with normal vs abnormal photopic negative response

Adj

uste

d Fl

icke

r Am

plitu

de (

uV)

Cone Off Response

Adj

uste

d Fl

icke

r Am

plitu

de (

uV)

Photopic Negative Response

67

Figure 16c.

Adjusted Flicker amplitude for patients with normal vs abnormal Cone Rmax

Decreased adjusted flicker amplitude is significantly associated with abnormal

cone sensitivity (P<0.001) (figure 16 d). The median adjusted flicker amplitude in

patients with abnormal cone sensitivity is -45uV, where as in patients with normal

cone sensitivity, median adjusted flicker is 0uV (expected for age).

Figure 16d. Adjusted Flicker amplitude for patients with normal vs abnormal cone Sensitivity.

Adj

uste

d Fl

icke

r Am

plitu

de (

uV)

Cone Rmax

Adj

uste

d Fl

icke

r Am

plitu

de (

uV)

Cone Sensitivity

68

Table 12 is a comparison of each marker’s ability to detect abnormal tests, using

the guidelines set forth above. Patients were only considered abnormal if they had

abnormalities beyond baseline. Therefore, patients were only included if they had

at least two tests (Eight patients excluded).

Table 12. Diagnostic characteristics of abnormal tests using flicker, cone off, and sensitivity Legend: pts – patients

Flicker response identified, 46% of patients (26 % of 81 tests) as abnormal, cone

off identified 29% of patients (14% of 49 tests) and sensitivity identified 46% of

patients (24% of 76 tests) as abnormal.

In the one patient who recovered cone sensitivity once the drug had been stopped,

sensitivity was still reduced (-34 phot td-1 s-3) compared with that expected for

Flicker Cone off Sensitivity

# abnormal tests 21 (11 pts) 9 (7 pts) 18 (11 Pts)

Time of peak abnormality 9 months 9 months 6-9 months

# Abnormal at baseline 3 0 1

Pts with 2 consecutive Abnormal tests

5 2 4

# with abnormal test followed by normal test

3 0 4

Maintain abnormality once drug stopped?

Yes (4) Yes (5) Yes (4), No (1)

69

age. The patient who had an abnormal sensitivity test at baseline was not one of

the three patients with an abnormal baseline flicker.

Figure 17. shows survival curves for each marker up to 15 months.

Cone off response has no abnormal tests at baseline and reaches a 69% survival

rate at 15 months. Flicker is 5% abnormal at baseline and reaches 64% survival

rate at 15 months. Cone sensitivity is 5% abnormal at baseline and reaches 48%

survival at 15 months.

70

Figure 17. Survival plots for cone off response (top), flicker amplitude (middle) and cone sensitivity (bottom). Solid lines indicate survival curves. Dashed lines indicate 95th confidence intervals. Y-axis represents proportion ‘survived’, in this case with normal test. X-axis is months on vigabatrin.

0 2 4 6 8 10 12 14

1.0 0.8 0.6 0.4 0.2 0.0

1.0 0.8 0.6 0.4 0.2 0.0

1.0 0.8 0.6 0.4 0.2 0.0

71

Mosaic plots were created to represent the overlap between the three markers in

classifying a particular patient’s test as normal or abnormal.

2 1 Normal 1 Abnormal

11 (a) 38 (d) 38 (b)

0 (c) 0

Figure 18. Venn Diagram of the number test points where patients had overlap between each test. Flicker and cone off (a), flicker and sensitivity (b), sensitivity and cone off (c), or all three tests (d) conducted.

In each mosaic plot, the bottom line represents one marker, in the first plot (figure

19a), flicker. The square is divided vertically into two sections. The left hand

section represents the proportion of tests that had a normal flicker response and

the right hand section represents the proportion of tests that had an abnormal

flicker response. On the right hand of the square cone off response is represented.

The two boxes that begin at the top side of the square had an abnormal cone off

response and the two boxes that meet the bottom side of the square had a normal

cone off response. In this way, four categories are created.

1) Normal Flicker – Normal Cone Off (Blue): 73%

2) Normal Flicker – Abnormal Cone Off (Green, top left): 11%

3) Abnormal Flicker – Normal Cone Off (Green, bottom right): 3%

4) Abnormal Flicker – Abnormal Cone Off (Yellow): 13%

Flicker

Sensitivity

Cone off

72

From this plot, it is also clear that of all patients who had both flicker and cone off

response tests performed:

a) Flicker was normal in 84% of cases and abnormal in 16%.

b) Cone off response was normal in 76% and abnormal in 24%.

Figure 19. Mosaic plots of agreement in classifying tests between (a) flicker and cone off, (b) flicker and sensitivity, (c) cone off and sensitivity, (d) flicker, cone off and sensitivity.

Abnormal

24%

Cone Off

76%

Normal 84% 16% Normal Flicker Abnormal

Figure19a.

73%

13%

3%

11%

73

Abnormal

20%

Cone Sens

80%

Normal

80% 20% Normal Flicker Abnormal

Abnormal

14%

Cone Sens

86%

Normal

75% 25%

Normal Cone Off Abnormal

Figure 19b.

9%

16%

5%

70%

Figure 19c.

10%

10%

10%

70%

74

Figure 19d. shows the agreement between all three tests. This plot is the same as the others except that there are two columns of normal and abnormal cone sensitivity, creating four columns. In the case where flicker response was normal and cone off response was abnormal, each test had a normal sensitivity; this explains why there is only one column in that section.

Sensitivity

Normal Abnormal Normal Abnormal

71% 19%

Abnormal

16%

Cone Off

84%

Normal

76% 24% Normal Flicker Response | Abnormal

Figure 19d

61% 13%

5% 5%

11% 2.5%

2.5%

75

Between all three markers, there is perfect agreement in 72% of tests. For cone

off and flicker, sensitivity and flicker, and sensitivity and cone off, there is 86%,

80% and 79% agreement between markers respectively.

Case Reports

Table 13 illustrates results of flicker, cone off and cone sensitivity for all patients

who had at least one abnormal test. All patients, with the exception of one

(patient 8), were initiated on vigabatrin within 3 weeks of seizure onset. In patient

8, there was a 3 month delay because they were being seen at another centre

where the type of seizure was not identified. Of the twelve patients who showed

any abnormal tests, only four had been treated with other AED’s (patient 13 -

Clobazam, 15 - Clobazam, 16 - ACTH, and 27 - topiramate, levetiracetam)

compared to 12 of twenty patients without abnormal test who were taking other

AEDs (see table 9 & 11).

76

Table 13. Individual patient data for all those with at least one abnormal test

0 3 6 9 12 15 18 21 24 27

F

1 O

S

F

2 O

S

F x

8 O x

S x

F x

9 O x

S x

F x

10 O x

S x

F

12 O

S

F x

13 O x

S x

77

F

15 O

S

F x x

16 O x x

S x x

F

17 O

S

F x

19 O x

S x

F

27 O

S

F

28 O

S

All patients with at least one abnormal test on at least one of the three markers were included except for two patients (number 14 & 29) who were excluded because their only abnormal test was abnormal flicker at baseline.

Legend: F – flicker; O - cone off; S - cone sensitivity; Green colouring identifies a normal test result; red identifies an abnormal test result; White spaces identify that a test was not done; X’s indicate the patient was off vigabatrin at that time.

78

Table 14 below shows the distribution of sex for normal and abnormal groups for

different criteria (no values are significantly different by a t test between normal

and abnormal for each definition).

Table 14. Distribution of sex and mean daily VGB dose for normal vs abnormal test using different definitions of abnormality

% Male Mean Daily VGB dose (mg/d)

Abnormal (Ab) Definition

Abnormal Normal Abnormal Normal

Any Ab tests 9/13 = 61% 5/10 = 50% 1240 1305

Ab by flicker (excluding Baseline

only)

5/8 = 62% 9/15 = 60% 1200 1300

Ab for Cone off 5/7 = 71% 9/16 = 56% 1210 1290

Ab for Sens 7/ 11 = 64% 7/12 = 58% 1200 1330

79

Discussion 11The present study investigated four potential new electrophysiological markers of

changes due to vigabatrin: photopic negative response, cone sensitivity, cone

maximum response and cone off response. These measures were compared to the

3.0 flicker amplitude, which is currently the most sensitive measure of Vigabatrin

retinal toxicity. The major findings were that both cone sensitivity and cone off

response were negatively affected by Vigabatrin use over time and were correlated

with results from 3.0 flicker amplitudes. Cone maximum response was not altered

significantly with drug treatment and changes in the photopic negative response

were not related to changes in the 3.0 flicker amplitude. It is still unclear which

response may be the best marker of change but the cone off response represents

a promising marker because it is not abnormal at baseline and does not return to

normal after having an abnormal test. The cone off response identifies 30% of

patients as abnormal, a value similar to the estimated 34% of children who

experience visual field loss, (Maguire et al., 2010).

It has been widely established that in adult patients taking vigabatrin, visual field

loss and some degree of visual function loss occurs. While correlations between

visual field loss and retinal dysfunction have been problematic even in an adult

population, it is clear that there are major changes that happen to the

electroretinogram in some patients taking vigabatrin in both adults and children.

Vigabatrin continues to be used as a first-line treatment for Infantile Spasms and

80

its use has become even more widespread since it’s reintroduction into the

American market in 2009. The use of the electroretinogram to monitor visual

function in pediatric vigabatrin users has been agreed upon by many, including in

the official REMS strategy, however the most appropriate and reliable marker has

not been agreed upon.

It has become clear in research done by our lab that infants with IS may have

some degree of altered visual function, even before initiating vigabatrin (Mirabella

et al., 2007; McCoy et al., 2011; McFarlane et al., 2011). Thus, the ability of our

lab to take baseline measures in these children is key to the delineation of the

effects of seizures versus that of the drug.

This study also described the development of photopic negative response and cone

off response in retinally normal control patients and Infantile Spasms patients,

which has not previously been done. As well, while cone sensitivity and cone

maximum response have been studied in normally developing infants up to 10

weeks (Hansen & Fulton, 2005), they have not been studied previously in IS

patients as demonstrated in this study or normal controls from 10 weeks to 4

years old.

Cone off response and cone sensitivity have not been studied in adults taking

vigabatrin or children old enough to reliably conduct visual field testing, so it is

difficult to know if these changes are related to loss of visual fields. However,

81

though we cannot currently know whether these changes are directly responsible

for visual field loss, it is clear that there are significant, lasting defects in these

parameters in some patients taking vigabatrin.

The loss of cone off response could be explained by damage to the cone OFF

bipolar pathways. Off bipolar cells respond to glutamate by depolarizing. The loss

in function could either represent a blockage or alteration in synaptic transmission

from photoreceptors to off bipolar cells, or by direct damage to bipolar cells

themselves.

Reduction in PhNR amplitude, over time might also be contributing to dysfunction

at the level of the bipolar cells. Changes in adjusted PhNR amplitude may not be

related to changes in flicker because a lag exists between initial bipolar cell

damage and the downstream effect on ganglion cells.

Decreases in cone sensitivity, but not cone maximum response, suggests that

vigabatrin affects primarily the process of phototransduction and may not, at least

in the early stages, be the cause of degenerating cones. Cone degeneration has

been seen in animal models, however these studies employ more acute doses and

sensitivity of cones has not previously been reported in these vigabatrin treated

animals. In normal controls aged 8-40 years, cone sensitivity values were found

to range from 55 – 120 phot td-1 s-3, with a mean (+ SE) of 81 + 5.5 phot td-1 s-3

by Hansen and Fulton (2005). Although in our study, toxicity was defined as a

82

difference from age expected value, it is notable that the absolute values are also

well below these values (mean = 6.03 phot td-1 s-3). There are no normal values

for children aged 4 months to 48 months. However, in infants 10 weeks old,

Hansen and Fulton demonstrated average cone sensitivity to be 59 + 3.9 phot td-1

s-3 and the mean sensitivity in our control population was 52.37 phot td-1 s-3. It is

puzzling that the control patients did not experience a significant effect of

development, while the IS patients did. This is probably related to the small

sample size of controls (n=10). As well, the cone sensitivity was greatly reduced in

the youngest of infants with IS and the developmental curve will reflect a catch up

in early infancy.

That fact that alterations in cone off response and cone sensitivity are related to

changes in the 3.0 flicker amplitude is consistent with work by Bush and Sieving

(1996). Bush and Sieving conclude that along with the contributions of cone

photoreceptor potentials, post receptoral cells that normally produce the b and d

waves, i.e. On and Off Bipolar cells, are strong contributors to the photopic fast

flicker response. It was also noted that the flicker response is independent of inner

retinal responses.

There is not complete agreement between the flicker amplitude, cone off response

and cone sensitivity in identifying abnormal tests; it is still unclear which is the

optimum marker for diagnosing true vigabatrin-induced retinal toxicity. While

83

flicker is the only test that has been correlated to visual field loss in adults with

100% sensitivity, it has only been shown to be 75% specific (Harding et al.,

2000b). This suggests that flicker amplitude may over-diagnose visual field loss.

This might also explain why some patients have an abnormal flicker test followed

by a normal test. If this is true, it is promising that the cone off response identifies

fewer patients as abnormal (29% vs. 46% for flicker and cone sensitivity). While

there is overlap in the patients in which both flicker and cone off response are

deemed abnormal (13% of cases), in the cases where only cone off response is

abnormal (11%), it may be that cone off response is identifying damage earlier

(i.e. VIE 9 and 28). It may also be that the cone off response is under-diagnosing

the problem. It is difficult to confirm this for two reasons: the lack of visual field

testing and the lack of a complete data set for all patients (cone off response

protocol only performed from June 2010 – June 2011). Further research is needed

to confirm this.

The question of the true mechanism of vigabatrin-induced retinal toxicity still

remains. It would be ideal to have a marker of the first or direct source of toxicity

in these infants, however, any marker that correlates to visual field loss will be

useful in screening IS patients on Vigabatrin. As visual field testing is not currently

feasible in this population, it is interesting to consider what these results indicate

about a possible mechanism of toxicity.

84

Animal studies have identified two key phenomena in the development of retinal

damage with vigabatrin use. First, it is clear that light exposure is key to the

development of retinal dysfunction in this population. This has been shown in both

rats (Butler et al., 1987; Izumi et al., 2004; Jammoul et al., 2009) and mice

(Jammoul et al., 2009). As well, it seems that taurine levels may play some role in

the toxicological mechanism.

In adult (Jammoul et al., 2009), and neonatal rats (Jammoul et al., 2010) treated

with vigabatrin who develop retinal dysfunction, taurine levels have been shown to

be depleted. Jammoul also demonstrated that six Vigabatrin–treated IS patients

ranging from 8.5 months – 3 years of age had reduced taurine levels. Visual fields

and ERG findings were not presented for these patients, therefore it is unclear

whether reduced taurine is related to decreases in retinal function (Jammoul et al.,

2009). It has been suggested that taurine levels are normal pre-treatment and are

reduced by vigabatrin treatment, however this has only been shown in one patient

with Infantile Spasms (Jammoul et al., 2009). It is unclear what role the

reduction of taurine levels plays in the toxicological mechanism. It may be that

either:

a) Vigabatrin à Decreased taurine à Retinal damage

b) Vigabatrin à Retinal damage à Decreased taurine

85

In mechanism (a), it is hypothesized that an increase in GABA could decrease

taurine levels as GABA is a competitive inhibitor of the taurine transporter (Lee

and Kang, 2004; Jammoul et al., 2009) or VGB may directly affect taurine uptake

and release. Alternatively, taurine levels might be reduced as a result of decreased

taurine synthesis via a decreased cysteine pool. This would involve increased

enzymatic conversion, or by use of taurine as a free radical scavenger. For

example, if antioxidant glutathione levels were reduced as a result of oxidative

stress, this would lead to a reduction in cysteine, resulting in reduced taurine

synthesis (Hayes & Sturman, 1981). In Sprague-Dawley rats, dietary taurine

supplementation decreases malondialdehyde levels in the retina, and increase

retinal taurine levels, superoxide dismutase and glutathione peroxidase. This

supplementation prevented photochemical damage caused by fluorescent light.

Supplementation with taurine in VGB treated adult and neonatal rats partially

prevented retinal lesions (Jammoul et al., 2009, 2010).

It is clear that in general, there are no damaging effects of natural light exposure

to the eye / retina, as light serves as a key component in the process of vision and

is not known to cause retinal toxicity in healthy normal eyes (Roberts, 2001). In

this review, Roberts identifies seven factors that may affect whether light is

damaging:

a) Intensity

In general, the greater the intensity, the more likelihood a light will damage the

86

eye. Cumulative light damage has been generally recognized to occur as a result

of lower long term exposure when there is a loss of protective mechanisms in the

eye (often because of age).

b) Wavelength

Shorter wavelengths have greater potential to cause damage. The young human

retina may be exposed to light as short as 320nm-400nm. If light exposure plays

a role, it may be expected that higher levels of phototoxicity occur in areas that

receive greater ultra violet light exposure (higher elevation, closer to equator).

c) Site of damage

Damage in specific components of the retina may be able to be repaired.

d) Oxygen tension

The greater the oxygen content of an ocular tissue the more susceptible is to

oxidative and photoxidative damage. The retina has high oxygen content in

different tissues. Among retinal cells, Muller cells are least susceptible, bipolar

cells are more so, and ganglion cells are most susceptible to the damaging effects

of ischemia (Hayreh & Weingeist, 1980).

e) Chromophores

Several endogenous molecules may act as chromophores in the retina including

rhodopsin, opsin and melanin and A2E. Chromophores are molecules that change

their conformational shape when hit by light. These may either absorb light or

quench reactive oxygen species. An exogenous chemical may also act as a

87

chromophore. There are several factors which may make a chemical act more

likely to act as a chromophore. These include: having a tricyclic, heterocyclic or

porphyrin ring, absorbing visible light, being able to cross the blood-ocular barrier

and ability to bind ocular tissues.

f) Defense system(s)

Antioxidant enzymes and antioxidants found in the eye help defend again

oxidative and photo-induced damage.

g) Repair.

The effect of light exposure on mediating vigabatrin’s toxic effects was first

postulated when it was noted that chronic oral administration of VGB led to outer

retina disorganization in Sprague-Dawley rats (albino) but not Lister Hooded

(pigmented rats) (Butler et al., 1987). Not all albino rats develop VGB toxicity, as

demonstrated by Gibson et al. (1990) in CD[SD]BR rats. Light remained implicated

in the toxicity mechanism given evidence that administration of VGB to isolated

retinas and via intraperitoneal injection of S-D rats induced light dependent acute

retinotoxicity (Izumi et al., 2004). Retinotoxicity did not result from injection of

GABA or tiagabine and furthermore, light by itself did not induce retinal toxicity

(Izumi et al., 2004). Brief and subacute systemic administration of vigabatrin

caused damage to photoreceptors and Muller cell dysfunction. Importantly, Izumi

88

et al. (2004), also concluded based on the inner retinal site of damage, that

neither GABA nor glutamate, mediate acute VGB toxicity. Izumi suggested that

vigabatrin sensitizes photoreceptors and Muller cells to light-induced damage.

In consideration of the mechanism of vigabatrin retinotoxicity, it should be

considered whether reduced taurine levels may be related to light exposure.

Wasowicz, Morice, Ferrari, Callebert & Versaux Botteri (2002), demonstrated that

light exposure caused photoreceptor degeneration, decreased retinal taurine levels

and increased vitreal taurine levels in only albino Wistar rats and not Long – Evans

pigmented rats.

Bulley & Shen (2010) noted that off bipolar cells in salamander retina may release

taurine as well as glutamate. Taurine was found primarily in the Off bipolar

terminals in the IPL, but not amacrine or ganglion cells. It is believed that taurine

suppresses glutamate-elicited Ca2+ in third order neurons by ionotropic glutamate

receptors. Decreased taurine levels may account for bipolar cell abnormalities. If

so, taurine supplementation may help to recover bipolar cell function even after

the drug has been stopped.

Not all patients who receive vigabatrin develop retinal toxicity. Several

mechanisms could be responsible:

a) Dietary taurine levels

b) Light exposure levels

89

c) Metabolizing enzymes

d) Vigabatrin dose

e) Genetic background factors

f) Any combination of these factors

Dietary taurine levels and light exposure levels may affect the susceptibility of

patients to vigabatrin retinal toxicity. It may be that a combination of low dietary

taurine intake, a retinal antioxidant, and increased light exposure, an oxidizing

agent, lead to an increased chance of vigabatrin damage. Retinal cells would need

to be sensitized to the light damage by something other than low taurine levels. It

is unlikely that low taurine levels themselves would sensitize retinal cells to

phototoxicity; if this were the case we would expect to see these visual field defect

in malnourished but non vigabatrin treated children and adults.

A difference in metabolizing enzymes could explain the effects in targeted retinal

cells and would fit with the premise of oxidative damage. Clinical pharmacological

studies have identified that between 60-80% (Schechter, 1986; Haegele &

Schechter, 1986; Rey et al., 1990) of the active S enantiomer of vigabatrin is

excreted unchanged in urine. If vigabatrin were undergoing retinal metabolism,

the product of this reaction may account for the 20-50% of vigabatrin that was not

detected in previous pharmacokinetic studies. The structure of vigabatrin does not

appear to be a prime suspect for ocular phototoxicity, however metabolic

90

conversion could change that. It is plausible that vigabatrin undergoes local retinal

metabolism to convert it to a toxic metabolite, which sensitizes retinal cells to light

damage. A similar mechanism happens in methanol toxicity.

Methanol is converted to formaldehyde by alcohol dehydrogenase, which is further

converted to formic acid by aldehyde dehydrogenase (AHD-2), an enzyme that

metabolizes retinaldehyde to retinoic acid in the normal human retina. When

formate (formic acid salt) accumulates, retinal toxicity may occur. After

biotransformation, a combination of direct oxidative stress at the Muller cell,

combined with decreased anti-oxidants enzyme activity, leads to methanol-

induced retinal toxicity. The Muller cells are believed to be the site of

biotransformation and initial insult (Garner, Lee, & Louis Ferdinand, 2002). This

may be because a unique aldehyde dehydrogenase isoform, AlDH-2, has been

shown to be present almost exclusively in the Muller cells in the adult retina of

mice. Levels of this enzyme were highest in dorsal retina, with still many in the

temporal peripheral retina and very few in the central retina (McCaffery, Tempst,

Lara, & Drager, 1991).

A decrease in ATP is seen after methanol administration in Folate-reduced rats,

which has been shown to be a good model of human methanol toxicity (Eells,

Henry, Lewandowski, Seme, & Murray, 2000), and this corresponds to changes in

the ERG (Garner & Lee, 1994). Decreases in the b-wave and loss of potassium

91

induced Muller cell depolarization were similar between methanol treated folate

reduced rates and alpha AAD (a Muller cell toxin) in folate reduced rats. Seme,

Summerfelt, Henry, Neitz & Eells (1999) showed the ERG flicker amplitude of M-

cones, cone that respond to medium wavelengths of light (450-630nm), decreased

in a formate concentration and time dependent manner. In FR methanol treated

rats, reductions of glutathione have also been reported Rajamani, Muthuvel,

Senthilvelan, & Sheeladevi (2006). Formic acid also inhibits cytochrome oxidase, a

mitochondrial enzyme, in cultured Muller cells (Eells et al., 2003). This arrests

electron transport chain activity (Nicholls, 1975, 1976), which in turn stops

regeneration of ATP and thus leads to cell death (Treichel, Henry, Skumatz, Eells &

Burke, 2004). Photoreceptors and the RPE also accumulate formate and cytotoxic

effects are seen in both types of cells (Treichel et al. 2004b). It has been

postulated that photoreceptors undergo more methanol toxicity than RPE because

they have higher level of antioxidant enzymes including catalase.

A moderating effect of light has not been seen in methanol toxicity. It is possible

that because methanol toxicity happens quite quickly, the mediating effects of

light exposure would not be observed. There is typically a latent period between

ingestion and symptom initiation followed by ocular symptoms that accompany the

systemic symptoms of methanol toxicity within 48 hours.

92

ERG effects of methanol are not the same as those observed in this study due to

vigabatrin. To understand what type of metabolism might turn vigabatrin into a

compound to sensitize retinal cells to oxidative damage, it could be useful to

understand the structure and property of other molecules that demonstrate similar

ERG effects. One other toxin, D-alpha aminoadpic acid (D-α AAA), which is a

glutamate analogue, although not subject to retinal metabolism, shows a similar

pattern of ERG changes. When D-α AAA is intravitreally administered to carp

retina, a reduction in glutamine synthetase activity, which is exclusively localized

to Muller cells, occurs within hours (Kato, Sugawara, Matsukawa, & Negishi,

1990). Of note, of the three isomers of α AAA (D, DL, L), D caused the least

reduction (28% vs. 45-65%) in GST activity. D-α AAA also caused the least

difference in the protein profile of the retina compared to L and DL. D-α AAA

caused a reversible decrease in the ERG b-wave and an increase in the a-wave

(Kato et all, 1990). In further experiments in Mudpuppy retinae, 5 mM of D-α AAA

preferentially reduced the d-wave (versus b-wave) of the ERG and the off

response of the Muller cells, but did not cause Muller cell damage (Zimmerman &

Corfman, 1984). L-α AAA however, caused preferential reversible b-wave and on

response reduction and was accompanied by sustained histological damage to

Muller glial cells. It is believed that D-α AAA may act as an antagonist to synaptic

receptors in the “off” pathway. As this xenobiotic selectively reduced the d –wave

93

(and also causes an increase in the a-wave, similar to the increase in maximum

response in this study), it is interesting to consider the structure, see figure 20.

D- alpha amino adipic acid vigabatrin

Figure 20. Structure of D- alpha amino adipic acid (left) and vigabatrin (right).

Bipolar cells primarily have two ionotropic glutamate receptors: AMPA which is

suppressed by AMPA antagonist GYKI 52466 (a 2,3-benzodiazepine) and Kainate

which is suppressed by SYM2081 (4-methyl glutamic acid), see figure 21.

4-methyl glutamic acid GYKI 52466

Figure 21. Structure of 4-methyl glutamic acid (left) and GYKI 52466 (right).

Given that D-AAA is more similar to 4-methyl glutamic acid than GYKI 52466, and

this structure is also much more similar to vigabatrin vigabatrin (or its metabolite)

might suppress the kainate receptor of bipolar cells.

94

Retinal metabolism may occur by CYP1A1/1A2, CYP4A, MOA A and B and retinal

specific enzymes ‘retinal amine oxidase’ (RAO) and xanthine oxidase (oxidative

stress in retinopathy). RAO is a human retinal specific enzyme, encoded by the

gene ACO2, and known to have an alternatively spliced variant which can lead to

different isoforms (Imammura et al., 1998), converts amines to aldehydes and

ammonia.

R–CH2–NH2 + O2 + H2Oà R–CHO + H2O2 + NH3

This could convert vigabatrin to 4-oxo 5-hexenoic acid and simultaneously create

hydrogen peroxide. While this step in itself would not lead to a structure similar to

those of D-AAA or 4-methyl glutamic acid, further metabolism of vigabatrin could

do that, with the initial RAO metabolism stage causing oxidative damage, and

further metabolism leading to bipolar cell signal blockade.

RAO is localized in mouse models to retinal ganglion cells (Imammura et al., 1998)

though no ganglion cell staining was detected when similar techniques were used

in human retinal cells. Authors noted as that when assessing AOC-2-like SSAO

activity, the retinal samples exhibited dramatic individual variation (Kaitamieni et

al., 2009).

The initial insult may be at the level of the cone off bipolar cell due to retinal

metabolism by RAO (or other retinal specific enzymes), and photoreceptor damage

results more from oxidative damage (as photoreceptors are very sensitive to this)

95

after sensitization. This would potentially make cone bipolar cells an earlier marker

of VGB toxicity. More research is needed to investigate this potential theory.

11.1 Clinical Implications

This study highlights several important issues for clinicians treating children with

Infantile Spasms on vigabatrin. The first is that abnormalities appear in these

children as early as three months of age. It was also demonstrated that some

children did not show an abnormal test result until coming off vigabatrin. These

two findings give further support to our current protocol of testing within three to

four months after the baseline test, and testing after the drug was discontinued. It

is important to assess each child as early as possible after an abnormal test.

Currently, because of the variability in the flicker response, clinicians must wait

until two tests have been conducted, at an average of four months apart; until

they can confidently say the child has an abnormal ERG. Ideally, we would see

children with an abnormal ERG within one month after the first abnormal test. This

however is unrealistic given the requirements for a sedated ERG (orthoptist, ERG

time, sedation nurse, physician assessment etc.). In this study, there were no

cases where the cone off response was abnormal and went on to be normal again.

If it can be confirmed that the cone off response is a valid marker of toxicity then

clinicians would not need to wait for two tests (four months) in order to diagnose

vigabatrin attributed retinal toxicity. It is however still unclear if stopping the drug

96

after an abnormal cone off response test would lead to a reversal, or stop

progression of damage due to vigabatrin and therefore further research is needed.

11.2 Problems and Considerations

One of the major difficulties in using the cone off response as a marker is the

propensity for infants and children (if not sedated) to blink over the course of the

200 ms light stimulus. A blink would result in a different waveform, which might

alter the amplitude of the response. To use the cone off response as a valid

marker, the average waveform should return to the level of the a-wave to ensure

that the amplitude of the off response is most accurate (Horn et al., 2011). In our

data, even in those cases in which the waveform did not return to the level of the

a-wave at 200 ms, a d-wave could still be easily identified, however these tests

were still excluded. As a result, the number of useable recordings was

approximately halved. The reduction in useable tests because of blink artifacts

could present a problem if this response was used as a marker in these infants. In

this study, we only performed 3 repetitions of the cone off response. The minimal

amount of repetitions had to be used in order to comply with ethical standards in

the hospital of minimizing the time a child is sedated. If this response were added

to the clinical protocol for vigabatrin monitoring, it would be prudent to allow more

than 3 repetitions to ensure that some recordings could be made without blinks.

As well, it may be worth moving the step earlier in the photopic sequence, as the

infants eyes may be less tired or aversive to the light, and less likely to blink.

97

Another problem with the cone off response as a potential marker is that it thus

far appears to identify non-reversible changes due to vigabatrin, as after one

abnormal test, the cone off response remains abnormal and does not recover after

the drug has stopped. It is noted that the longest time after drug cessation that a

child may be seen in our clinic is 1.5 years, it is possible that after this point some

recovery may be seen. More importantly, while the changes may not be

permanent, early identification may be able to stop the progression of these

changes. If the functional changes are as a result of damage by free radicals, early

identification could signal the need for treatment with anti-oxidants and / or

vigabatrin cessation. This may halt further structural changes.

Lastly, the study was hindered by the use of a convenience sample. Our recruiting

pool existed of all IS patients referred to SickKids for ERG testing. We were unable

to ask children to come in for extra visits to monitor them more frequently, nor

were we able to recruit from outside of SickKids. This was true as well for normal

controls, where we could only approach those patients who were already

scheduled to be seen for ERG testing to participate. This is primarily due to the

risk and cost associated with sedation, which is necessary for the procedure. These

IS patients included in this study will continue to be studied to gain more

longitudinal data. This study supports a need for seeing children, especially those

who have an abnormal test, more regularly than every 3-4 months.

98

11.3 Future Directions

This study provides preliminary data to support the use of the cone off response as

a marker of retinotoxic changes due to vigabatrin. Further studies at our centre

and others, are needed to confirm in a larger population if cone off response is

indeed a preferable marker to 3.0 flicker amplitude.

Another step in validating the cone off response as a marker of changes due to

vigabatrin is to recall these study patients when they are four years of age and

older. These children would undergo repeat ERG’s and visual field testing when

they are old enough to complete behavioural visual field tests. This would allow

correlation between visual field loss and ERG results and give information as to

whether ERG dysfunction can predict visual field dysfunction.

It would also be prudent to investigate changes in the cone off response in adults

with CPS taking vigabatrin. In these patients, ERG responses could be serially

measured with visual fields to see whether cone off response or cone sensitivity

dysfunction correlate with visual field loss.

Optical coherence tomography could also be used to investigate early structural

changes in IS patients. Early damage to middle retinal layers on OCT would help to

validate the cone bipolar cells as a marker. There are similar difficulties in testing

infants with OCT as with visual field testing. One group has used a technique in

which sedated infants are held up to the OCT for testing, however this is not

99

commonly performed and results did not correlate entirely with ERG results (Mets

et al., 2011).

Further support could be garnered for this theory if it were established that

vigabatrin was metabolized by retinal enzymes and by investigating where these

enzymes are localized. Local retinal metabolism should be studied in animal

models. Some enzymes have been found only in human retinas and this would

make the problem more difficult to investigate. Histological studies of deceased

vigabatrin users may help to discover which enzymes might be involved.

Identification of specific enzymes in either animal or human specimens could lead

to a target for genetic screening. It is clear that there is often genetic variation in

enzymes, including drug metabolizing enzymes. Genetic polymorphisms can be

identified using directed screens of different genes. Genetic polymorphisms may

either be responsible for a patient developing toxicity, or may make them more

susceptible to damage. It could be any combination of specific isoforms of an

enzyme, a certain threshold of light exposure, reduced taurine intake and

increased vigabatrin doses that are related to developing toxicity. Until it is clear

whether retinal enzymes play a role in vigabatrin metabolism and whether they

are susceptible to genetic variation, this theory cannot be confirmed.

Lastly, controlled studies in IS patients investigating the effects of taurine

supplementation and decreased light exposure on the development of vigabatrin

100

associated retinal dysfunction would help to understand the mechanism of toxicity

in humans.

In the interim, it is important to continue monitoring flicker response, cone off

response and sensitivity.

101

References 12Agrawal, S., Mayer, D. L., Hansen, R., & Fulton, A. (2009). Visual fields in young

children treated with vigabatrin. Optometry and Vision Science, 86(6), 767-773.

Arndt, C. F., Derambure, P., Defoort Dhellemmes, S., & Hache, J. C. (1999). Outer retinal dysfunction in patients treated with vigabatrin. Neurology, 52(6), 1201-1205.

Ascaso, F., Lopez, M., Mauri, J., & Cristobal, J. (2003). Visual field defects in pediatric patients on vigabatrin monotherapy. Documenta Ophthalmologica, 107(2), 127-130.

Baram, T. Z., Mitchell, W. G., Tournay, A., Snead, O. C., Hanson, R. A., & Horton, E. J. (1996). High-dose corticotropin (ACTH) versus prednisone for infantile spasms: A prospective, randomized, blinded study. Pediatrics, 97(3), 375-379.

Ben Menachem, E., Persson, L. I., Mumford, J., Haegele, K. D., & Huebert, N. (1991). Effect of long-term vigabatrin therapy on selected neurotransmitter concentrations in cerebrospinal fluid. Journal of Child Neurology, Suppl 2, S11-S16.

Besch, D., Kurtenbach, A., Apfelstedt Sylla, E., Sadowski, B., Dennig, D., Asenbauer, C., . . . Schiefer, U. (2002). Visual field constriction and electrophysiological changes associated with vigabatrin. Documenta Ophthalmologica, 104(2), 151-170.

Bjelajac, A., Gautam, M., & Logan, W.J. (1999). Vigabatrin and ophthalmologic abnormalities in pediatric patients. Neurology, 52, A236.

Bourcier, F.N., De Toffol, B., Majzoub, S., & Delplace, M. (2004). Electrophysiologic and visual field disturbances evolution in patients treated with vigabatrin. Investigative Ophthalmology & Visual Science, 45, A255.

Bulley, S., & Shen, W. (2010). Reciprocal regulation between taurine and glutamate response via Ca2 dependent pathways in retinal third-order neurons. Journal of Biomedical Science, 17 Suppl 1, S5-S5.

Buncic, J. R., Westall, C., Panton, C., Munn, J. R., MacKeen, L., & Logan, W. (2004). Characteristic retinal atrophy with secondary "inverse" optic atrophy identifies vigabatrin toxicity in children. Ophthalmology, 111(10), 1935-1942.

102

Bush, R. A., & Sieving, P. A. (1996). Inner retinal contributions to the primate photopic fast flicker electroretinogram. Journal of the Optical Society of America.A, Optics, Image Science, and Vision, 13(3), 557-565.

Butler, W. H., Ford, G. P., & Newberne, J. W. (1987). A study of the effects of vigabatrin on the central nervous system and retina of sprague dawley and lister-hooded rats. Toxicologic Pathology, 15(2), 143-148.

Camposano, S. E., Major, P., Halpern, E., & Thiele, E. A. (2008). Vigabatrin in the treatment of childhood epilepsy: A retrospective chart review of efficacy and safety profile. Epilepsia, 49(7), 1186-1191. doi:10.1111/j.1528-1167.2008.01589.x

Chiron, C., Dulac, O., Beaumont, D., Palacios, L., Pajot, N., & Mumford, J. (1991). Therapeutic trial of vigabatrin in refractory infantile spasms. Journal of Child Neurology, Suppl 2, S52-S59.

Cohen-Sadan, S., Kramer, U., Ben-Zeev, B., Lahat, E., Sahar, E., Nevo, Y., . . . Goldberg-Stern, H. (2009). Multicenter long-term follow-up of children with idiopathic west syndrome: ACTH versus vigabatrin. European Journal of Neurology, 16(4), 482-487.

Comaish, I. F., Gorman, C., Brimlow, G. M., Barber, C., Orr, G. M., & Galloway, N. R. (2002). The effects of vigabatrin on electrophysiology and visual fields in epileptics: A controlled study with a discussion of possible mechanisms. Documenta Ophthalmologica, 104(2), 195-212.

Coupland, S. G., Zackon, D. H., Leonard, B. C., & Ross, T. M. (2001). Vigabatrin effect on inner retinal function. Ophthalmology, 108(8), 1493-6.

Cowan, L. D., & Hudson, L. S. (1991). The epidemiology and natural history of infantile spasms. Journal of Child Neurology, 6(4), 355-364.

Daneshvar, H., Racette, L., Coupland, S. G., Kertes, P. J., Guberman, A., & Zackon, D. (1999). Symptomatic and asymptomatic visual loss in patients taking vigabatrin. Ophthalmology, 106(9), 1792-1798.

Duboc, A., Hanoteau, N., Simonutti, M., Rudolf, G., Nehlig, A., Sahel, J. A., & Picaud, S. (2004). Vigabatrin, the GABA-transaminase inhibitor, damages cone photoreceptors in rats. Annals of Neurology, 55(5), 695-705.

103

Eells, J. T., Henry, M. M., Lewandowski, M. F., Seme, M. T., & Murray, T. G. (2000). Development and characterization of a rodent model of methanol-induced retinal and optic nerve toxicity. Neurotoxicology, 21(3), 321-330.

Eells, J. T., Henry, M. M., Summerfelt, P., Wong Riley, M. T. T., Buchmann, E. V., Kane, M., . . . Whelan, H. T. (2003). Therapeutic photobiomodulation for methanol-induced retinal toxicity. Proceedings of the National Academy of Sciences of the United States of America, 100(6), 3439-3444.

Eke, T., Talbot, J. F., & Lawden, M. C. (1997). Severe persistent visual field constriction associated with vigabatrin. BMJ (Clinical Research Ed.), 314(7075), 180-181.

Eklund, S.E., Moskowitz, A., Barnaby, A.M., Hansen, R.M., & Fulton, A.B. (2006). The Effect of Vigabatrin on the ERG in patients with Infantile Spasms. Investigative Ophthalmology & Visual Science, 47, A1657.

Faedda, M.T., Giallonardo, A.T., Marchetti, A. & Manfredi, M. (1993). Terapia con vigabatrin nelle epilessie parziali resistenti. G Neuropsicofarmacol, 15, 105–8.

French, J.A. (1999). Vigabatrin. Epilepsia, 40(Suppl 5), S11-6.

Frishman, L.J. (2006). Origins of the electroretinogram. In J.R. Heckenlively & G.B. Arden (Eds.), Principles and practice of clinical electrophysiology of vision (pp. 39-184). Cambridge, MA: The MIT Press.

Gaily, E., Jonsson, H., & Lappi, M. (2009). Visual fields at school-age in children treated with vigabatrin in infancy. Epilepsia, 50(2), 206-216.

Garner, C. D., & Lee, E. W. (1994). Evaluation of methanol-induced retinotoxicity using oscillatory potential analysis. Toxicology, 93(2-3), 113-124.

Garner, C. D., Lee, E. W., & Louis Ferdinand, R. T. (1995). Müller cell involvement in methanol-induced retinal toxicity. Toxicology and Applied Pharmacology, 130(1), 101-107.

Gibbs, E. L., Fleming, M. M., & Gibbs, F. A. (1954). Diagnosis and prognosis of hypsarhythmia and infantile spasms. Pediatrics, 13(1), 66-73.

Gibson, J. P., Yarrington, J. T., Loudy, D. E., Gerbig, C. G., Hurst, G. H., & Newberne, J. W. (1990). Chronic toxicity studies with vigabatrin, a GABA-transaminase inhibitor. Toxicologic Pathology, 18(2), 225-238.

104

Gross Tsur, V., Banin, E., Shahar, E., Shalev, R. S., & Lahat, E. (2000). Visual impairment in children with epilepsy treated with vigabatrin. Annals of Neurology, 48(1), 60-64.

Grove, J., Schechter, P. J., Tell, G., Koch Weser, J., Sjoerdsma, A., Warter, J. M., . . . Rumbach, L. (1981). Increased gamma-aminobutyric acid (GABA), homocarnosine and beta-alanine in cerebrospinal fluid of patients treated with gamma-vinyl GABA (4-amino-hex-5-enoic acid). Life Sciences, 28(21), 2431-2439.

Grove, J., Tell, G., Schechter, P. J., Koch Weser, J., Warter, J. M., Marescaux, C., & Rumbach, L. (1980). Increased CSF gamma-aminobutyric acid after treatment with gamma-vinyl GABA. Lancet, 2(8195 pt 1), 647-647.

Haegele, K. D., & Schechter, P. J. (1986). Kinetics of the enantiomers of vigabatrin after an oral dose of the racemate or the active S-enantiomer. Clinical Pharmacology Therapeutics, 40(5), 581-586.

Hansen, R., & Fulton, A. (2005). Development of the cone ERG in infants. Investigative Ophthalmology Visual Science, 46(9), 3458-3462.

Harding, G. F., Robertson, K., Spencer, E. L., & Holliday, I. (2002). Vigabatrin; its effect on the electrophysiology of vision. Documenta Ophthalmologica.Advances in Ophthalmology, 104(2), 213-229.

Harding, G. F., Wild, J. M., Robertson, K. A., Lawden, M. C., Betts, T. A., Barber, C., & Barnes, P. M. (2000). Electro-oculography, electroretinography, visual evoked potentials, and multifocal electroretinography in patients with vigabatrin-attributed visual field constriction. Epilepsia, 41(11), 1420-1431.

Harding, G. F., Wild, J. M., Robertson, K. A., Rietbrock, S., & Martinez, C. (2000). Separating the retinal electrophysiologic effects of vigabatrin. treatment versus field loss. American Journal of Ophthalmology, 130(5), 691.

Hardus, P., Verduin, W. M., Berendschot, T. T., Kamermans, M., Postma, G., Stilma, J. S., & van Veelen, C. W. (2001). The value of electrophysiology results in patients with epilepsy and vigabatrin associated visual field loss. Acta Ophthalmologica Scandinavica, 79(2), 169-174.

Hardus, P., Verduin, W., Berendschot, T., Postma, G., Stilma, J., & van Veelen, C. (2003). Vigabatrin: Longterm follow-up of electrophysiology and visual field examinations. Acta Ophthalmologica Scandinavica, 81(5), 459-465.

105

Hardus, P., Verduin, W., Berendschot, T., Postma, G., Stilma, J., & van Veelen, C. (2003). Vigabatrin: Longterm follow-up of electrophysiology and visual field examinations. Acta Ophthalmologica Scandinavica, 81(5), 459-465.

Hayes, K. C., & Sturman, J. A. (1981). Taurine deficiency: A rationale for taurine depletion. Adv Exp Med Biol, 139, 79-87.

Hayreh, S. S., & Weingeist, T. A. (1980). Experimental occlusion of the central artery of the retina. IV: Retinal tolerance time to acute ischaemia. British Journal of Ophthalmology, 64(11), 818-825.

Hood, D. C., & Birch, D. G. (1995). Phototransduction in human cones measured using the alpha-wave of the ERG. Vision Research, 35(20), 2801-2810.

Horn, F., Gottschalk, K., Mardin, C., Pangeni, G., Jnemann, A., & Kremers, J. (2011). On and off responses of the photopic fullfield ERG in normal subjects and glaucoma patients. Documenta Ophthalmologica, 122(1), 53-62.

Iannetti, P., Spalice, A., Perla, F. M., Conicella, E., Raucci, U., & Bizzarri, B. (2000). Visual field constriction in children with epilepsy on vigabatrin treatment. Pediatrics, 106(4), 838-842.

Imamura, Y., Noda, S., Mashima, Y., Kudoh, J., Oguchi, Y., & Shimizu, N. (1998). Human retina-specific amine oxidase: Genomic structure of the gene (AOC2), alternatively spliced variant, and mRNA expression in retina. Genomics, 51(2), 293-298.

Izumi, Y., Ishikawa, M., Benz, A., Izumi, M., Zorumski, C., & Thio, L. (2004). Acute vigabatrin retinotoxicity in albino rats depends on light but not GABA. Epilepsia, 45(9), 1043-1048.

Jammoul, F., Dégardin, J., Pain, D., Gondouin, P., Simonutti, M., Dubus, E., ... Picaud, S. (2010). Taurine deficiency damages photoreceptors and retinal ganglion cells in vigabatrin-treated neonatal rats. Molecular & Cellular Neurosciences, 43(4), 414-421.

Jammoul, F., Wang, Q., Nabbout, R., Coriat, C., Duboc, A., Simonutti, M., . . . Picaud, S. (2009). Taurine deficiency is a cause of vigabatrin-induced retinal phototoxicity. Annals of Neurology, 65(1), 98-107.

Jensen, H., Sj, O., Uldall, P., & Gram, L. (2002). Vigabatrin and retinal changes. Documenta Ophthalmologica, 104(2), 171-180.

106

Kaitaniemi, S., Elovaara, H., Grn, K., Kidron, H., Liukkonen, J., Salminen, T., . . . Elima, K. (2009). The unique substrate specificity of human AOC2, a semicarbazide-sensitive amine oxidase. Cellular and Molecular Life Sciences, 66(16), 2743-2757.

Kato, S., Sugawara, K., Matsukawa, T., & Negishi, K. (1990). Gliotoxic effects of alpha-aminoadipic acid isomers on the carp retina: A long term observation. Neuroscience, 36(1), 145-153.

Kjellstrm, U., Andrasson, S., & Ponjavic, V. (2011). Electrophysiological evaluation of retinal function in children receiving vigabatrin medication. Journal of Pediatric Ophthalmology and Strabismus, , 1-9.

Kjellstrom, U., Lovestam-Adrian, M., Andreasson, S., & Ponjavic, V. (2008). Full-field ERG and visual fields in patients 5 years after discontinuing vigabatrin therapy. Documenta Ophthalmologica.Advances in Ophthalmology, 117(2), 93-101. doi:10.1007/s10633-007-9108-3

Krauss, G. L., Johnson, M. A., Sheth, S., & Miller, N. R. (2003). A controlled study comparing visual function in patients treated with vigabatrin and tiagabine. Journal of Neurology, Neurosurgery and Psychiatry, 74(3), 339-343.

Krauss, G. L., Johnson, M. A., & Miller, N. R. (1998). Vigabatrin-associated retinal cone system dysfunction: Electroretinogram and ophthalmologic findings. Neurology, 50(3), 614-618.

Krauss, G. L., & Miller, N. R. (1999). Vigabatrin: An effective antiepilepsy drug--balancing the risk of visual dysfunction. The Annals of Pharmacotherapy, 33(12), 1367-1368.

Lamb, T. D., & Pugh, E. N. (1992). A quantitative account of the activation steps involved in phototransduction in amphibian photoreceptors. Journal of Physiology, 449, 719-758.

Lawden, M. C., Eke, T., Degg, C., Harding, G. F., & Wild, J. M. (1999). Visual field defects associated with vigabatrin therapy. Journal of Neurology, Neurosurgery and Psychiatry, 67(6), 716-722.

Lee, N., & Kang, Y. (2004). The brain-to-blood efflux transport of taurine and changes in the blood-brain barrier transport system by tumor necrosis factor-alpha. Brain Research, 1023(1), 141-147.

107

Lerner, J., Salamon, N., & Sankar, R. (2010). Clinical profile of vigabatrin as monotherapy for treatment of infantile spasms. Neuropsychiatric Disease and Treatment, 6, 731-740.

Ludvigsson, P., Olafsson, E., Sigurthardttir, S., & Hauser, W. A. (1994). Epidemiologic features of infantile spasms in iceland. Epilepsia, 35(4), 802-805.

Lundbeck Inc. (2010). Learn about Sabril. Retrieved from http://www.sabril.net/hcp/about-us.aspx

Lux, A. L., Edwards, S. W., Hancock, E., Johnson, A. L., Kennedy, C. R., Newton, R. W., . . . United Kingdom Infantile Spasms, S. (2005). The united kingdom infantile spasms study (UKISS) comparing hormone treatment with vigabatrin on developmental and epilepsy outcomes to age 14 months: A multicentre randomised trial. Lancet Neurology, 4(11), 712-717.

Mackay, M. T., Weiss, S. K., Adams-Webber, T., Ashwal, S., Stephens, D., Ballaban-Gill, K., . . . Child Neurology, S. (2004). Practice parameter: Medical treatment of infantile spasms: Report of the american academy of neurology and the child neurology society. Neurology, 62(10), 1668-1681.

Maguire, M., Hemming, K., Wild, J., Hutton, J., & Marson, A. (2010). Prevalence of visual field loss following exposure to vigabatrin therapy: A systematic review. Epilepsia, 51(12), 2423-2431.

Marmor, M. F., Fulton, A. B., Holder, G. E., Miyake, Y., Brigell, M., & Bach, M. (2009). ISCEV standard for full-field clinical electroretinography (2008 update). Documenta Ophthalmologica, 118(1), 69-77.

McCaffery, P., Tempst, P., Lara, G., & Drager, U. C. (1991). Aldehyde dehydrogenase is a positional marker in the retina. Development, 112(3), 693-702.

McCoy, B., Wright, T., Weiss, S., Go, C., & Westall, C. (2011). Electroretinogram changes in a pediatric population with epilepsy: Is vigabatrin acting alone? Journal of Child Neurology, 26(6), 729-733.

McDonagh, J., Stephen, L. J., Dolan, F. M., Parks, S., Dutton, G. N., Kelly, K., . . . Brodie, M. J. (2003). Peripheral retinal dysfunction in patients taking vigabatrin. Neurology, 61(12), 1690-1694.

108

McFarlane, M.T., Westall, C.A., & Wright, T. (2011). Investigating Infantile Spasms aetiologies and their effects on retinal dysfunction. Investigative Ophthalmology & Visual Science, 52, A271.

Meldrum, B. S., & Murugaiah, K. (1983). Anticonvulsant action in mice with sound-induced seizures of the optical isomers of gamma-vinyl GABA. European Journal of Pharmacology, 89(1-2), 149-152.

Mets, R.B., Geddie, B., Trimboli-Heidler, C., Elling, N., Jaafar, M., McClintock, W., & Madigan, W.P. (2011). Utility of Optical Coherence Tomography in monitoring Vigabatrin retinal toxicity. Investigative Ophthalmology & Visual Science, 52, 3683.

Miller, N. R., Johnson, M. A., Paul, S. R., Girkin, C. A., Perry, J. D., Endres, M., & Krauss, G. L. (1999). Visual dysfunction in patients receiving vigabatrin: Clinical and electrophysiologic findings. Neurology, 53(9), 2082-2087.

Mirabella, G., Morong, S., Buncic, J. R., Snead, O. C., Logan, W. J., Weiss, S. K., . . . Westall, C. A. (2007). Contrast sensitivity is reduced in children with infantile spasms. Investigative Ophthalmology & Visual Science, 48(8), 3610-3615. doi:10.1167/iovs.06-0755

Morong, S., Westall, C. A., Nobile, R., Buncic, J. R., Logan, W. J., Panton, C. M., & Abdolell, M. (2003). Longitudinal changes in photopic OPs occurring with vigabatrin treatment. Documenta Ophthalmologica.Advances in Ophthalmology, 107(3), 289-297.

National Institute for Health and Clinical Excellence (NICE). Newer drugs for epilepsy in children, December 2010. Available from www.nice.org.uk/nicemedia/live/11532/32858/32858.pdf . Accessed March 22, 2011.

Neal, M. J., Cunningham, J. R., Shah, M. A., & Yazulla, S. (1989). Immunocytochemical evidence that vigabatrin in rats causes GABA accumulation in glial cells of the retina. Neuroscience Letters, 98(1), 29-32.

Nicholls, P. (1975). Formate as an inhibitor of cytochrome c oxidase. Biochemical and Biophysical Research Communications, 67(2), 610-616.

Nicholls, P. (1976). The effect of formate on cytochrome aa3 and on electron transport in the intact respiratory chain. Biochimica Et Biophysica Acta, 430(1), 13-29.

109

Pelosse, B., Momtchilova, M., Roubergue, A., Doummar, D., Billette de Villemeur, T., & Laroche, L. (2001). [Study of visual field and vigabatrin treatment in children]. Journal Français d'Ophtalmologie, 24(10), 1075-1080.

Petroff, O. A., Rothman, D. L., Behar, K. L., Collins, T. L., & Mattson, R. H. (1996). Human brain GABA levels rise rapidly after initiation of vigabatrin therapy. Neurology, 47(6), 1567-1571.

Pojda-Wilczek, D., Emich-Widera, E., Herba, E., Pojda, S. M., & Marszal, E. (2005). [The function of eye and vision system in children and youth treated with vigabatrin--our own experiences]. [Funkcja narzadu wzroku u dzieci i mlodziezy leczonych wigabatryna ze szczególnym uwzglednieniem spostrzezen wlasnych.] Klinika Oczna, 107(10-12), 654-657.

Ponjavic, V., & Andreasson, S. (2001). Multifocal ERG and full-field ERG in patients on long-term vigabatrin medication. Documenta Ophthalmologica.Advances in Ophthalmology, 102(1), 63-72.

Ponjavic, V., Granse, L., Kjellstrom, S., Andreasson, S., & Bruun, A. (2004). Alterations in electroretinograms and retinal morphology in rabbits treated with vigabatrin. Documenta Ophthalmologica.Advances in Ophthalmology, 108(2), 125-133.

Pugh, E. N., & Lamb, T. D. (1993). Amplification and kinetics of the activation steps in phototransduction. Biochimica Et Biophysica Acta, 1141(2-3), 111-149.

Rajamani, R., Muthuvel, A., Senthilvelan, M., & Sheeladevi, R. (2006). Oxidative stress induced by methotrexate alone and in the presence of methanol in discrete regions of the rodent brain, retina and optic nerve. Toxicology Letters, 165(3), 265-273.

Rantala, H., & Putkonen, T. (1999). Occurrence, outcome, and prognostic factors of infantile spasms and lennox-gastaut syndrome. Epilepsia, 40(3), 286-289.

Rey, E., Pons, G., & Olive, G. (1992). Vigabatrin. clinical pharmacokinetics. Clinical Pharmacokinetics, 23(4), 267-278.

Rey, E., Pons, G., Richard, M. O., Vauzelle, F., D'Athis, P., Chiron, C., . . . Olive, G. (1990). Pharmacokinetics of the individual enantiomers of vigabatrin (gamma-vinyl GABA) in epileptic children. British Journal of Clinical Pharmacology, 30(2), 253-257.

110

Richens, A. (1991). Pharmacology and clinical pharmacology of vigabatrin. Journal of Child Neurology, Suppl 2, S7-10.

Riikonen, R., & Donner, M. (1980). ACTH therapy in infantile spasms: Side effects. Archives of Disease in Childhood, 55(9), 664-672.

Riikonen, R. S. (2000). Steroids or vigabatrin in the treatment of infantile spasms? Pediatric Neurology, 23(5), 403-408.

Roberts, J. E. (2001). Ocular phototoxicity. Journal of Photochemistry and Photobiology.B, Biology, 64(2-3), 136-143.

Roccella, M., Parisi, L., & D’Iapico, N. (2001). Analisi del campo visivo nei bambini con epilessia parziale in monoterapia con Vigabatrin. Boll Lega It Epil, 114, 243–244.

Roubertie, A., Bellet, H., & Echenne, B. (1998). Vigabatrin-associated retinal cone system dysfunction. Neurology, 51(6), 1779-1781.

Schechter, P. J. (1986). Vigabatrin. In New anticonvulsivant drugs, eds Meldrum, B. S. & Porter, R. J. pp 265-275. London: John Libbe.

Scottish and Intercollegiate Guidelines Network (SIGN). Diagnosis and management of epilepsies in children and young people: a national clinical guideline, March 2005. Available from www.sign.ac.uk/pdf/sign81.pdf. Accessed March 22, 2011.

Seme, M. T., Summerfelt, P., Henry, M. M., Neitz, J., & Eells, J. T. (1999). Formate-induced inhibition of photoreceptor function in methanol intoxication. The Journal of Pharmacology and Experimental Therapeutics, 289(1), 361-370.

Sergott, R. (2010). Recommendations for visual evaluations of patients treated with vigabatrin. Current Opinion in Ophthalmology, 21(6), 442-446.

Sidenvall, R., & Eeg Olofsson, O. (1995). Epidemiology of infantile spasms in sweden. Epilepsia, 36(6), 572-574.

Sieving, P. A., Murayama, K., & Naarendorp, F. (1994). Push-pull model of the primate photopic electroretinogram: A role for hyperpolarizing neurons in shaping the b-wave. Visual Neuroscience, 11(3), 519-532.

111

Sills, G. J., Patsalos, P. N., Butler, E., Forrest, G., Ratnaraj, N., & Brodie, M. J. (2001). Visual field constriction: Accumulation of vigabatrin but not tiagabine in the retina. Neurology, 57(2), 196-200.

Sorri, I., Brigell, M., Mlyusz, M., Mahlamki, E., de Meynard, C., & Klviinen, R. (2010). Is reduced ornithine-δ-aminotransferase activity the cause of vigabatrin-associated visual field defects? Epilepsy Research, 92(1), 48-53.

Spencer, E. L., & Harding, G. F. (2003). Examining visual field defects in the paediatric population exposed to vigabatrin. Documenta Ophthalmologica.Advances in Ophthalmology, 107(3), 281-287.

Stafstrom, C., Arnason, B. G. W., Baram, T., Catania, A., Cortez, M., Glauser, T., . . . Swann, J. (2011). Treatment of infantile spasms: Emerging insights from clinical and basic science perspectives. Journal of Child Neurology,

Treichel, J., Henry, M., Skumatz, C. M. B., Eells, J., & Burke, J. (2004). Antioxidants and ocular cell type differences in cytoprotection from formic acid toxicity in vitro. Toxicological Sciences, 82(1), 183-192.

Treichel, J., Henry, M., Skumatz, C. M. B., Eells, J., & Burke, J. (2004). Antioxidants and ocular cell type differences in cytoprotection from formic acid toxicity in vitro. Toxicological Sciences, 82(1), 183-192.

Treichel, J., Murray, T., Lewandowski, M., Stueven, H., Eells, J., & Burke, J. (2004). Retinal toxicity in methanol poisoning. Retina, 24(2), 309-312.

Trevathan, E., Murphy, C. C., & Yeargin Allsopp, M. (1999). The descriptive epidemiology of infantile spasms among atlanta children. Epilepsia, 40(6), 748-751.

van der Torren, K., Graniewski Wijnands, H., & Polak, B. C. P. (2002). Visual field and electrophysiological abnormalities due to vigabatrin. Documenta Ophthalmologica, 104(2), 181-188.

Van Parys, J.A.P., Meijer, J.W.A., & Edelbroek, P.M. (1995). Comparison of enzyme induction by various antiepileptic drugs including oxcarbazepine and vigabatrin. Epilepsia, 36, S61.

Vanhatalo, S., Pääkkönen, L., & Nousiainen, I. (1999). Visual field constriction in children treated with vigabatrin. Neurology, 52(8), 1713-1714.

112

Vanhatalo, S., Nousiainen, I., Eriksson, K., Rantala, H., Vainionp, L., Mustonen, K., . . . Granstrm, M. (2002). Visual field constriction in 91 finnish children treated with vigabatrin. Epilepsia, 43(7), 748-756.

Viswanathan, S., Frishman, L. J., Robson, J. G., Harwerth, R. S., & Smith, E. L. (1999). The photopic negative response of the macaque electroretinogram: Reduction by experimental glaucoma. Investigative Ophthalmology Visual Science, 40(6), 1124-1136.

Wang, Q., Jammoul, F., Duboc, A., Gong, J., Simonutti, M., Dubus, E., . . . Picaud, S. (2008). Treatment of epilepsy: The GABA-transaminase inhibitor, vigabatrin, induces neuronal plasticity in the mouse retina. The European Journal of Neuroscience, 27(8), 2177-2187.

Wasowicz, M., Morice, C., Ferrari, P., Callebert, J., & Versaux Botteri, C. (2002). Long-term effects of light damage on the retina of albino and pigmented rats. Investigative Ophthalmology Visual Science, 43(3), 813-820.

Werblin, F. S., & Dowling, J. E. (1969). Organization of the retina of the mudpuppy, necturus maculosus. II. intracellular recording. Journal of Neurophysiology, 32(3), 339-355.

Werblin, F. (1991). Synaptic connections, receptive fields, and patterns of activity in the tiger salamander retina. A simulation of patterns of activity formed at each cellular level from photoreceptors to ganglion cells [the friendenwald lecture]. Investigative Ophthalmology Visual Science, 32(3), 459-483.

Werth, R., & Schadler, G. (2006). Visual field loss in young children and mentally handicapped adolescents receiving vigabatrin. Investigative Ophthalmology Visual Science, 47(7), 3028-3035.

Westall, C. A., Logan, W. J., Smith, K., Buncic, J. R., Panton, C. M., & Abdolell, M. (2002). The hospital for sick children, toronto, longitudinal ERG study of children on vigabatrin. Documenta Ophthalmologica.Advances in Ophthalmology, 104(2), 133-149.

Westall, C., Nobile, R., Morong, S., Buncic, J. R., Logan, W., & Panton, C. (2003). Changes in the electroretinogram resulting from discontinuation of vigabatrin in children. Documenta Ophthalmologica, 107(3), 299-309.

Wheless, J. W., Clarke, D. F., Arzimanoglou, A., & Carpenter, D. (2007). Treatment of pediatric epilepsy: European expert opinion, 2007. Epileptic Disorders, 9(4), 353-412.

113

Wild JM. Chiron C. Ahn H. Baulac M. Bursztyn J. Gandolfo E. Goldberg I. Goñi FJ. Mercier F. Nordmann JP. Safran AB. Schiefer U. Perucca E. (2009). Visual field loss in patients with refractory partial epilepsy treated with vigabatrin: Final results from an open-label, observational, multicentre study. CNS Drugs, 23(11), 965-982.

Wild, J. M., Martinez, C., Reinshagen, G., & Harding, G. F. (1999). Characteristics of a unique visual field defect attributed to vigabatrin. Epilepsia, 40(12), 1784-1794.

Willoughby, J. A., Thurston, D. L., & Holowach, J. (1966). Infantile myoclonic seizures: An evaluation of ACTH and corticosteroid therapy. The Journal of Pediatrics, 69(6), 1136-1138.

Wilson, E. A., & Brodie, M. J. (1997). Severe persistent visual field constriction associated with vigabatrin. chronic refractory epilepsy may have role in causing these unusual lesions. BMJ (Clinical Research Ed.), 314(7095), 1693.

Wohlrab, G., Boltshauser, E., Schmitt, B., Schriever, S., & Landau, K. (1999). Visual field constriction is not limited to children treated with vigabatrin. Neuropediatrics, 30(3), 130-132.

Wong, M., & Trevathan, E. (2001). Infantile spasms. Pediatric Neurology, 24(2), 89-98.

Xu, X., & Karwoski, C. J. (1994). Current source density analysis of retinal field potentials. II. pharmacological analysis of the b-wave and M-wave. Journal of Neurophysiology, 72(1), 96-105.

Xu, X., & Karwoski, C. (1995). Current source density analysis of the electroretinographic d wave of frog retina. Journal of Neurophysiology, 73(6), 2459-2469.

You, S., Ahn, H., & Ko, T. (2006). Vigabatrin and visual field defects in pediatric epilepsy patients. Journal of Korean Medical Science, 21(4), 728-732.

Yum, M., Ko, T., Lee, J., Hong, S., Kim, D., & Kim, J. (2011). Surgical treatment for localization-related infantile spasms: Excellent long-term outcomes. Clinical Neurology and Neurosurgery, 113(3), 213-217.

Zimmerman, R. P., & Corfman, T. P. (1984). A comparison of the effects of isomers of alpha-aminoadipic acid and 2-amino-4-phosphonobutyric acid on

114

the light response of the müller glial cell and the electroretinogram. Neuroscience, 12(1), 77-84.

Zupanc, M. L. (2003). Infantile spasms. Expert Opinion on Pharmacotherapy, 4(11), 2039-2048.

115

Appendices (if any)

Appendix 1. List of current lab protocol

For all steps: Background colour: White Stimulus colour: White

Stimulus Intensity

Sweeps/result

Time between results

Background Intensity

Adaptation Time

Unit cd*s/m2 ms s cd*s/m2 s

1 0.00039 8 2 0p

2 0.00151 8 2 0p

3 0.00245 10 2 0p

4 0.00632 8 2 0p

ISCEV (1)

5 0.01578 6 5 0p

6 0.04 6 6 0p

7 0.097 6 10 0p

ISCEV (2)

8 2.291 6 15 0p

9 7.6 4 15 0p

10 10 3 30 0p

ISCEV (4)

11 2.291 15 2 29 600

12 4.1 20 2 29

ISCEV (5) -

flicker

13 2.291 30 0, continuous phase

29

14 10 8 2 29

116

Appendix 2. Vigabatrin and Infantile Epilepsy Subject Consent Form

THE HOSPITAL FOR SICK CHILDREN

Department of Ophthalmology

Visual Electrophysiology Unit

Phone (416) 813-6516

Hospital for Sick Children (SickKids)

RESEARCH CONSENT FORM (For Parents of Subjects Prescribed Vigabatrin)

Title of Research Project: Vigabatrin and Infantile Epilepsy

Investigators

Director of Electrophysiology: Dr.Carol Westall 416-813-6516

Responsible Individual: Dr. Carol Westall 416-813-6516

Senior Orthorptist: Carole Panton O.C. (C) 416-813-6133

Orthoptist: Melissa Cotesta OA 416-813-7789

Research Manager: Thomas Wright 416-813-7790 Ophthalmologist: Dr. J. Raymond Buncic 416-813-6508

Graduate student: Julianna Sienna 416-813-7654 ext 3606

Student: Ananthavalli Kumarappah

Student: Ashna Patel

Name: D.O.B.:

Hosp#:

117

Purpose of the research:

The drug vigabatrin is used to help control seizures. In some people the drug might cause problems with vision. This might be related to some small changes of the retina. The retina is the inner lining of the eye that makes a picture of what we see (like film in a camera).

We want to better understand what is happening to the eyes in children on vigabatrin. The following tests may be performed before, during vigabatrin therapy and after its withdrawal.

The electroretinogram (ERG) is an electrophysiological test to measure the electrical response of the retina. ERGs are a routine clinical test used to assess retinal function when a retinal disease is suspected or known. We want to better understand what is happening to the eyes in people undergoing vigabatrin treatments.

Description of the research:

The following tests will be performed once your child’s neurologist or ophthalmologist has referred them to the Visual Electrophysiology Unit and an appointment has been made. Sedated ERGs are clinically indicated for children with Infantile Spasms taking the antiepileptic vigabatrin to find out whether any changes to the retina have taken place. The tests will take one hour to be performed.

Electroretinogram (ERG): This study will involve the addition of two extra steps to clinic protocol. These extra steps are intended to isolate the response from a specific part of the retina and will extend the testing time by approximately 1 minute.

Patient’s health records will be reviewed for purposes of this study for information about drug history, co-morbidities etc. Standard clinic intake tests may be performed including vision, colour vision, refractive error, and / or ophthalmoscopy.

Potential Harms (Injury, Discomforts or Inconvenience):

There are minimal harms associated with participation in this study. Under exceptional circumstances there is a slight risk that your child may receive a minor scratch to the front of her/his eye. This scratch would feel similar to having a piece

118

of sand in your eye and the discomfort may last for 2-3 days. We will check for this and provide any necessary treatment. The eye drops cause slight stinging, but this resolves within 10 seconds. The drops which we use to dilate your child’s pupils may cause his/her vision be blurred up close for 4-8 hours. The risks involved in this study are no greater than those for normal clinic protocol.

Potential Benefits:

To the individual: Your child may not benefit directly from participating in this study. Ophthalmological and neurological care will continue whether your child continues in this study or not. Our research team will send you a letter detailing our findings when the study is completed.

To Society: Knowledge gained from this study will hopefully allow physicians to optimize vigabatrin therapy to ensure patients receive the maximum benefit whilst minimizing visual toxicity. Better control of the risks associated with this powerful therapy will make its use in a wider patient population more feasible.

Confidentiality:

We will respect you and your child’s privacy. No information about who you are will be given to anyone or be published without your permission, unless the law makes us do this.

For example, the law could make us give information about you • If a child had been abused • If you have an illness that could spread to others • If you or someone else talks about suicide (killing themselves), or • If the court orders us to give them the study papers

Sick Kids Clinical Research Monitors or the regulator of the study may see your health record to check on the study. By signing this consent form, you agree to let these people look at your records. We will put a copy of this research consent form in your patient health record and give you a copy as well.

The data produced from this study will be stored in a secure, locked location. Only members of the research team (and maybe those individuals described above) will

119

have access to the data. Following completion of the research study the data will be kept as long as required by the Sick Kids “Records Retention and Destruction” policy. The data will then be destroyed according to this same policy.

Reimbursement

Compensation will be provided at a rate of $20.00 for each testing session in recognition of your time and effort. If you stop taking part in the study, you will be compensated for those tests your child has undergone up until that point.

Participation:

Participation in research is voluntary. Your child is likely to be refered for testing every 3-6 months once they are on vigabatrin. If you do choose to participate you will only be asked to sign the consent form once. By signing this consent form you agree to have testing extended by 1 minute every time your child comes in for testing. You can withdraw your child from the study at any time. The care you get at Sick Kids will not be affected in any way by whether you take part in this study.

New information that we get while we are doing this study may affect your decision to take part in this study. If this happens, we will tell you about this new information ask you again if you still want to be in the study.

During this study we may create new tests, new medicines, or other things that may be worth some money. Although we may make money from these findings, we cannot give you any of this money now or in the future because you took part in this study.

We will give you a copy of this consent form for your records.

In some situations, the study doctor or the company paying for the study may decide to stop the study. This could happen even if the medicine given in the study is helping you. If this happens, the study doctor will talk to you about what will happen next.

If your child becomes ill or harmed because your child took part in this study, we will treat your child for free. Your signing this consent form does not interfere with your child’s legal rights in any way. The staff of the study, any people who gave money for the study, or the hospital are still responsible, legally and professionally, for what they do.

Sponsor / Funder of the study

120

The sponsor of this research is Sick Kids Hospital. The funder of this research is Lundbeck Pharmaceuticals.

Conflict of Interest

Some of the people doing this study may have a conflict of interest. That means that they may benefit personally, financially, or in some other way from this study .

Dr. Westall (Principal Investigator) has received or may receive for research related to the present study money, or one or more of the following other benefits: speaker's fees, travel assistance, industry-initiated research grants, investigator- initiated research grants, consultant fees, honoraria, gifts, intellectual property rights such as patents, etc. from sponsor(s) that have activities related to the present study.

Consent

•By signing this form, I agree that:

1) You have explained this study to me. You have answered all my questions.

2) You have explained the possible harms and benefits (if any) of this study.

3) I know what I could do instead of having my child take part in this study. I understand that I have the right to refuse to let my child take part in the study. I also have the right to take my child out of the study at any time. My decision about my child taking part in the study will not affect my child’s health care at SickKids.

4) I am free now, and in the future, to ask any questions about the study.

5) I have been told that my child’s medical records will be kept private. You will give no one information about my child, unless the law requires you to.

6) I understand that no information about my child will be given to anyone or be published without first asking my permission.

7) I have read and understood pages 1 to 5 of this consent form. I agree, or consent, that my child ________________________ may take part in this study.

121

_____________________________________________________________

Printed Name of Parent/Legal Guardian Parent/Legal Guardian’s signature & date

_____________________________ ______________ _ ________

Printed Name of person who explained consent Signature & date

________________________________ ______________________

Printed Witness’name (if the parent/legal Witness’ signature & date

Guardian does not read English)

If you have any questions about this study, please call Julianna Sienna at (416)-813-7654 ext. 3606.

If you have questions about your rights as a subject in a study or injuries during a study, please call the Research Ethics Manager at (416)813-5718.

122

Appendix 3. Vigabatrin and Infantile Epilepsy Control Consent Form

THE HOSPITAL FOR SICK CHILDREN

Department of Ophthalmology

Visual Electrophysiology Unit

Phone (416) 813-6516

Hospital for Sick Children (SickKids)

RESEARCH CONSENT FORM (For Parents of Controls)

Title of Research Project: Vigabatrin and Infantile Epilepsy

Investigators

Director of Electrophysiology: Dr.Carol Westall 416-813-6516

Responsible Individual: Dr. Carol Westall 416-813-6516

Senior Orthoptist: Carole Panton O.C. (C). 416-813-6133

Orthoptist: Melissa Cotesta OA 416-813-7789

Research Manager: Thomas Wright 416-813-7790 Ophthalmologist: Dr. J. Raymond Buncic 416-813-6508

Graduate student: Julianna Sienna 416-813-7654 ext 3606

Name: D.O.B.:

Hosp#:

123

Student: Ananthavalli Kumarappah

Student: Ashna Patel

Purpose of the research:

The drug vigabatrin is used to help control seizures. In some people the drug might cause problems with vision. Your child does not have seizures and is not taking vigabatrin.

To better understand what is happening to the eyes of people taking the drug vigabatrin, we need to record these tests in normal children like your child who do not have visual abnormalities due to medication. The following tests may be performed to find out how much these responses change in children with normal retinas.

The electroretinogram (ERG) is an electrophysiological test used to measure the electrical response of the retina. Sedated ERGs are a routine clinical test used to assess retinal function when a retinal disease is suspected or known. Sedated ERGs are clinically indicated for children with suspected idiopathic nystagmus. We want to better understand what is happening to the eyes in people undergoing certain drug treatments.

Description of the research:

The following tests will be performed once your child’s neurologist or ophthalmologist has referred them to the Visual Electrophysiology Unit and an appointment has been made. These tests are clinically indicated to find out whether any changes to the retina have taken place. The tests will take one hour to be performed.

Electroretinogram (ERG): The ERG will be administered according to standard clinic protocol with the addition of two extra steps. These extra steps are intended to isolate the response from a specific part of the retina and will extend the testing time by approximately 1 minute.

124

Patient’s health records will be reviewed for purposes of this study for information about drug history, co-morbidities etc. Standard clinic intake tests may be performed including vision, colour vision, refractive error, and / or ophthalmoscopy.

Potential Harms (Injury, Discomforts or Inconvenience):

There are minimal harms associated with participation in this study. Under exceptional circumstances there is a slight risk that your child may receive a minor scratch to the front of her/his eye. This scratch would feel similar to having a piece of sand in your eye and the discomfort may last for 2-3 days. We will check for this and provide any necessary treatment. The eye drops cause slight stinging, but this resolves within 10 seconds. The drops which we use to dilate your child’s pupils may cause his/her vision be blurred up close for 4-8 hours. The risks involved in this study are no greater than those for normal clinic protocol.

Potential Benefits:

To the Individual: Your child will not benefit directly for participating in this study. Ophthalmological and neurological care will continue whether your child continues in this study or not. Our research team will send you a letter detailing our findings when the study is completed.

To Society: Knowledge gained from this study will hopefully allow physicians to optimize vigabatrin therapy to ensure patients receive the maximum benefit whilst minimizing visual toxicity. Better control of the risks associated with this powerful therapy will make its use in a wider patient population more feasible.

Confidentiality:

125

We will respect you and your child’s privacy. No information about who your child is will be given to anyone or be published without your permission, unless the law makes us do this.

For example, the law could make us give information about you • If a child had been abused • If you have an illness that could spread to others • If you or someone else talks about suicide (killing themselves), or • If the court orders us to give them the study papers

Sick Kids Clinical Research Office Monitor or the regulator of the study may see your child’s health record to check on the study. By signing this consent form, you agree to let these people look at your child’s records. We will put a copy of this research consent form in your patient health records.

The data produced from this study will be stored in a secure, locked location. Only members of the research team (and maybe those individuals described above) will have access to the data. Following completion of the research study the data will be kept as long as required by the Sick Kids “Records Retention and Destruction” policy. The data will then be destroyed according to this same policy.

Reimbursement

Compensation will be provided at a rate of $20.00 for each testing session in recognition of your time and effort. If you stop taking part in the study, you will be compensated for those tests your child has undergone up until that point.

Participation:

Participation in research is voluntary. If you choose to participate in this study you can withdraw your child from the study at any time. The care you get at Sick Kids will not be affected in any way by whether you take part in this study.

New information that we get while we are doing this study may affect your decision to take part in this study. If this happens, we will tell you

126

about this new information. And we will ask you again if you still want to be in the study.

During this study we may create new tests, new medicines, or other things that may be worth some money. Although we may make money from these findings, we cannot give you any of this money now or in the future because you took part in this study.

We will give you a copy of this consent form for your records.

In some situations, the study doctor or the company paying for the study may decide to stop the study. This could happen even if the medicine given in the study is helping you. If this happens, the study doctor will talk to you about what will happen next.

If your child becomes ill or is harmed because they took part in this study, we will treat your child for free. Your signing this consent form does not interfere with your child’s legal rights in any way. The staff of the study, any people who gave money for the study, or the hospital are still responsible, legally and professionally, for what they do.

Sponsor / Funder of the study

The sponsor of this research is Sick Kids Hospital. The funder of this research is Lundbeck Pharmaceuticals.

Conflict of Interest

Some of the people doing this study may have a conflict of interest. That means that they may benefit personally, financially, or in some other way from this study .

Dr. Westall (Principal Investigator) has received or may receive for research related to the present study (money, or one or more of the following other benefits: speaker's fees, travel assistance, industry-initiated research grants, investigator- initiated research grants, consultant fees, honoraria, gifts, intellectual property rights such as

127

patents, etc.) from sponsor(s) that have activities related to the present study.

Consent

“By signing this form, I agree that:

1) You have explained this study to me. You have answered all my questions.

2) You have explained the possible harms and benefits (if any) of this study.

3) I know what I could do instead of having my child take part in this study. I understand that I have the right to refuse to let my child take part in the study. I also have the right to take my child out of the study at any time. My decision about my child taking part in the study will not affect my child’s health care at SickKids.

4) I am free now, and in the future, to ask questions about the study.

5) I have been told that my child’s medical records will be kept private. You will not give anyone information about my child, unless the law requires you to.

6) I understand that no information about my child will be given to anyone or be published without first asking my permission.

7) I have read and understood pages 1 to 5 of this consent form. I agree, or consent, that my child___________________ may take part in this study.

_________________________________

Printed Name of Subject & Age Subject’s signature & date (if applicable)

__________________________________ __________________________________

128

Printed Name of Parent/Legal Guardian Parent/Legal Guardian’s signature & date

_________ _________________________________

Printed Name of person who explained consentSignature of Person who explained consent & date

____________________________________________________________________

Printed Witness’ name (if the subject/legal guardian Witness’ signature & date

does not read English

If you have any questions about this study, please call Julianna Sienna at (416)-813-7654 ext. 3606.

If you have questions about your rights as a subject in a study or injuries during a study, please call the Research Ethics Manager at (416) 813-5718.

129

Appendix 4. Patient report form.

130

131

Appendix 5. R software for Hood and Birch model

library(tcltk)

loadWaves<-function(folder,eye=NA){

#load rod waves, file is expected to be a csv #each column represents a wave #supporting file (*.info) contains start time, sample frequency, intensities if(is.na(eye)){ inp<-readline('Which eye (l/r)?') }else{ inp<-eye }

if(inp=='l'){ fname<-'LEwaves.csv' }else{ fname<-'REwaves.csv' }

waves<-read.csv(file=paste(folder,fname,sep='/'),header=FALSE) info<-read.csv(file=paste(folder,'info.csv',sep='/'),header=FALSE)

info<-data.frame(info) names(info)<-c('Intensity','Start','Deltat')

if(length(unique(info$Start))>1 | length(unique(info$Deltat))>1 ) { #need to resample cat('Resampling Not Implemented',sep='\n') stop() }

waves<-ts(waves/1000,start=unique(info$Start),deltat=unique(info$Deltat)) return(list(data=waves,info=info)) }

zeroWaves<-function(data){

132

newData<-data

matplot(x=time(newData$data),y=newData$data,type='l',lwd=2,col=rainbow(ncol(data$data)))

legend('bottomright',paste(data$info$Intensity,'(',c(1,2,3),')'),col=rainbow(ncol(data$data)),lwd=2) # inp<-tolower(substr(readline('Do you wish to zero these waves (y/n) ?'),1,1)) inp<-'y' if(inp=='y'){

prestim<-window(data$data,end=0)\ vals<-apply(prestim,2,mean) for(iloc in 1:ncol(data$data)){

newData$data[,iloc]<-data$data[,iloc]-vals[iloc] } }

matplot(x=time(newData$data),y=newData$data,type='l',lwd=2,col=rainbow(ncol(data$data))) legend('bottomright',paste(data$info$Intensity,'(',c(1,2,3),')'),col=rainbow(ncol(data$data)),lwd=2)

#continue<-tolower(substr(readline('Accept (y/n) ?'),1,1)) continue<-'y’ while(continue=='n'){ inp<-readline('Enter the waves to be zero\'d (indexes separated by ,) or 0 for all:') inp<-as.integer(strsplit(inp,',')[[1]]) if(inp[1]==0){ inp<-seq(from=1,to=ncol(data$data)) } if(max(inp)>ncol(data$data) | min(inp)<1){ cat('Wave not found',sep='\n') next } else{ targets<-inp } inp<-readline('Enter the new zero factors (comma seperated list, or times separated by a colon):')

133

if(length(grep(':',inp))>0){ cat('Using new window for 0\n') inp<-as.numeric(strsplit(inp,':')[[1]]) prestim<-window(data$data,start=inp[1],end=inp[2]) vals<-apply(prestim[,targets],2,mean) } else{ vals<-as.numeric(strsplit(inp,',')[[1]]) if(length(vals)<length(targets)){ cat('Must provide a factor for each wave to be zeroed\n') next } }

for(iloc in 1:length(targets)){ newData$data[,targets[iloc]]<-data$data[,targets[iloc]]-vals[iloc] }

matplot(x=time(newData$data),y=newData$data,type='l',lwd=2,col=rainbow(ncol(data$data))) legend('bottomright',paste(data$info$Intensity,'(',c(1,2,3,4),')'),col=rainbow(ncol(data$data)),lwd=2)

#continue<-tolower(substr(readline('Accept (y/n) ?'),1,1)) }

return(newData) }

truncateWaves<-function(data){ startIdx<-rep(min(which(time(data$data)>0))-1,dim(data$data)[2]) minIdx<-rep(min(which(time(data$data)>11))-1,dim(data$data)[2])

return(cbind(startIdx,minIdx)) }

findTimeIdx<-function(times,target){ #support function to find nearest time to a target minIdx<-max(which(times<target)) maxIdx<-min(which(times>target))

134

idx<-minIdx+round((target-times[minIdx])/(times[maxIdx]-times[minIdx]))

return(idx) }

extractAWaves<-function(data,idx){

for(iloc in 1:ncol(data)){ data[1:(idx[iloc,1]-1),iloc]<-NA data[(idx[iloc,2]+1):nrow(data),iloc]<-NA }

data<-window(data,start=time(data)[min(idx[,1])],end=time(data)[max(idx[,2])]) return(data) }

selectIntensities<-function(intensities){ #cat(paste(' (',c(1:3),') ',intensities,'\n',sep='')) #inp<-readline('Which intensities should be used in the model (enter index numbers seperated by comma, or 0 for all):') #ans<-as.integer(strsplit(inp,',')[[1]]) #if(ans[1]==0){ # ans<-1:length(intensities) # } #if(max(ans)>length(intensities)){ # cat('Unidentified intensity') # stop() # } ans<-c(1,2,3)

return(ans) }

estimateRmax<-function(awaves){ return(min(apply(awaves,2,min,na.rm=TRUE))) }

generateSimWavesRod<-function(intensities,window,S,Td,Rmax){ #times<-seq(from=window[1]*10,to=(window[2]*10)+1)

times<-seq(from=0,to=(window[2]*10)+1) out<-matrix(ncol=length(intensities),nrow=(length(times)+(Td*10))) #convert flash intensities to correct units (Td.ms)

135

intensities<-(10^intensities)*0.0001^2 for(iloc in 1:length(intensities))

{

out[,iloc]<-c(rep(0,Td*10),{1-exp(-intensities[iloc]*S*(times)^2)}*Rmax)

}

#out<-ts(out,start=window[1],frequency=10)

out<-ts(out,0,frequency=10)

return(out)

}

generateSimWavesCone<-function(intensities,window,S,Td,Rmax,tau2){

waves<-generateSimWavesRod(intensities,window,S,Td,Rmax)

times<-seq(from=0,to=((end(waves)[1]+end(waves)[2]/frequency(waves))*10)-1)

tmpwaves<-matrix(ncol=ncol(waves),nrow=length(times)*2-1)

for(iloc in 1:ncol(waves))

{

#tmpwaves[,iloc]<-handConvolve(exp(-times/tau2),waves[,iloc])

tmpwaves[,iloc]<-convolve(exp(-times/tau2),rev(waves[,iloc]),type='o')

#need to find which window to look at here

}

waves<-window(ts(tmpwaves,start=0,frequency=frequency(waves)),end=end(waves))

waves<-waves/(min(waves)/Rmax)

return(waves)

136

}

calcFit<-function(awaves,simwaves){

if(all(is.na(simwaves))){

return(Inf)

}

if(!is.ts(awaves)){

cat('awaves must be a time series\n')

stop()

}

if(!class(awaves)[1]==class(simwaves)[1]){

cat('awaves and sim waves must be the same class\n')

stop()

}

if(class(awaves)[1]=='mts')

{

if(!ncol(awaves)==ncol(simwaves)){

cat('Must have the same number of waves in observed and predicted data\n')

stop()

}

}

if(frequency(simwaves)<frequency(awaves)){

cat('simwaves must have a higher sample frequency than observed waves\n')

stop()

137

}

op<-getOption('warn')

options(warn=-1)

simwaves<-window(simwaves,start=start(awaves),end=max(time(awaves)),frequency=frequency(awaves))

options(warn=op)

stat<-sqrt(sum(as.vector(awaves-simwaves)^2,na.rm=TRUE)/sum(as.vector(awaves-mean(awaves,na.rm=TRUE))^2,na.rm=TRUE))

#cat(paste('stat=',stat,'\n'))

return(stat)

}

fitAwaves<-function(folder,eye=NA){

#ans<-tolower(substr(readline('Are these rod or cone responses (r/c)?\n'),1,1)[[1]])

ans<-'c'

if(ans=='r'){

type<-'rod'

}

else{

type<-'cone'

}

data<-loadWaves(folder,eye)

data<-zeroWaves(data)

138

windowIdx<-truncateWaves(data)

intensityIdx<-selectIntensities(data$info[,1])

awaves<-extractAWaves(data$data,windowIdx)

idx<-windowIdx[intensityIdx,]

awaves<-awaves[,intensityIdx]

windowTimes<-c(min(time(awaves)),max(time(awaves)))

#ans<-readline('Enter an estimate for Rmax (a=auto):\n')

ans<-'a'

if(tolower(substr(ans[[1]],1,1))=='a'){

Rmax<-estimateRmax(awaves)

}

else{

Rmax<-as.numeric(ans[[1]])

}

#ans<-readline('Enter a value for Td (a=auto):\n')

ans<-3.3 #CHANGE THIS FOR A SET Td VALUE (no quotes if numeric), 'a' for auto

if(tolower(substr(ans[[1]],1,1))=='a'){

Td<-'auto'

}

else{

139

Td<-as.numeric(ans[[1]])

}

if(Td=='auto'){

params<-c(Rmax,10)

lbounds<-c(Rmax-100,0.1)

ubounds<-c(0,200)

}

else{

params<-c(Rmax,10)

lbounds<-c(Rmax-100,0.1)

ubounds<-c(0,200)

TdSet<-Td

}

#ans<-readline('Enter a value for Tau (a=auto):\n')

ans<-5 #change this for a set value to change Tau (no quotes if numberic), 'a' for auto

if(tolower(substr(ans[[1]],1,1))=='a'){

Tau<-'auto'

}else{

Tau<-as.numeric(ans[[1]])

TauSet<-Tau

}

intensities<-data$info[intensityIdx,1]

140

cat('Fitting waves, this may take some time\n')

if(type=='rod'){

result<-nlm(fitWaves,params,intensities=intensities,window=windowTimes,awaves=awaves,tdSet=Td,'rod')

if(Td=='auto'){

Td<-result$estimate[3]

}

#else{

# Td<-

# }

simWaves<-window(generateSimWavesRod(intensities,windowTimes,result$estimate[2],Td,result$estimate[1]),frequency=frequency(awaves))

bestFit<-list(Td=Td,Rmax=result$par[1],S=result$par[2],Fit=result$objective)

}

else{

params<-c(params)

lbounds<-c(lbounds)

ubounds<-c(ubounds)

#result<-nlm(fitWaves,params,intensities=intensities,window=windowTimes,awaves=awaves,tdSet=Td,'cone')

if(Td=='auto'){

TdRange<-seq(from=0.1,to=4.5,by=0.1)

TauRange<-seq(from=1,to=5,by=0.1)

141

results<-matrix(nrow=length(TdRange)*length(TauRange),ncol=6)

for(iloc in 1:length(TdRange)){

for(nloc in 1:length(TauRange)){

cat(paste('iloc=',iloc,' nloc=',nloc,'\n'))

Td<-TdRange[iloc]

Tau<-TauRange[nloc]

result<-nlminb(params,fitWaves,intensities=intensities,window=windowTimes,awaves=awaves,tauSet=Tau,tdSet=Td,type='cone',lower=lbounds,upper=ubounds)

results[iloc,]<-c(Tau,Td,result$objective,result$par[1],result$par[2],result$convergence)

}

}

minIdx<-which.min(results[,3])

cat('\n')

if(results[minIdx,6]<1){

cat('\nSUCCESS\n')

}

simWaves<-window(generateSimWavesCone(intensities,windowTimes,results[minIdx,5],results[minIdx,2],results[minIdx,4],results[minIdx,1]),frequency=frequency(awaves))

bestFit<-list(Td=results[minIdx,2],Tau=results[minIdx,1],Rmax=results[minIdx,4],S=results[minIdx,5],tau2=results[minIdx,1],Fit=results[minIdx,3])

}

142

else{

result<-nlminb(params,fitWaves,intensities=intensities,window=windowTimes,awaves=awaves,tdSet=Td,tauSet=Tau,type='cone',lower=lbounds,upper=ubounds)

cat('\n')

if(result$convergence<1){

cat('\nSUCCESS\n')

}

simWaves<-window(generateSimWavesCone(intensities,windowTimes,result$par[2],Td,result$par[1],Tau),frequency=frequency(awaves))

bestFit<-list(Td=Td,Rmax=result$par[1],S=result$par[2],tau2=Tau,Fit=result$objective)

}

}

matplot(x=time(awaves),y=awaves,lty=1,lwd=2,type='l')

matlines(x=time(simWaves),y=simWaves,lwd=2,lty=2)

return(list(fit=bestFit,awaves=awaves,simwaves=simWaves))

}

fitWaves<-function(factors,intensities,window,awaves,tdSet,tauSet=NA,type){

#if(!isTRUE(all.equal(Td,3,tollerance=0.1))){browser()}

143

Td<-round(tdSet,1)

Rmax<-factors[1]

#tau2<-factors[1]

S<-factors[2]

#cat('.')

if(type=='cone'){

tau2<-tauSet

simwaves<-generateSimWavesCone(intensities,window,S,Td,Rmax,tau2)

}else{

simwaves<-generateSimWavesRod(intensities,window,S,Td,Rmax)

}

return(calcFit(awaves,simwaves))

}

handConvolve<-function(x,h){

out<-vector(length=(length(x)+length(x)-1))

for(iloc in 1:length(out)){

out[iloc]<-0

for(jloc in 1:length(h)){

if(iloc - jloc < 1){

next

}

144

if(iloc - jloc > length(x)){

next

}

out[iloc]<-out[iloc] + h[jloc] * x[iloc-jloc]

}

}

return(out)

}

processFolder<-function(folder=NA)

{

if(is.na(folder)){

folder<-tk_choose.dir()

}

output<-matrix(ncol=8)

output<-data.frame(output)

names(output)<-c('Patient','Test','Eye','Td','Rmax','S','tau','fit')

patients<-dir(folder)

iloc<-0

for(pat in patients){

cat(paste('Processing patient:',pat,'\n'))

tests<-dir(paste(folder,pat,sep='/'))

for(test in tests){

cat(paste('Processing test:',test,'\n'))

145

if(file.exists(paste(folder,pat,test,'LEwaves.csv',sep='/'))){

#process LE

iloc<-iloc+1

x<-fitAwaves(paste(folder,pat,test,sep='/'),'l')

output[iloc,1]<-pat

output[iloc,2]<-test

output[iloc,3]<-'l'

output[iloc,4]<-x$fit[1]

output[iloc,5]<-x$fit[2]

output[iloc,6]<-x$fit[3]

output[iloc,7]<-x$fit[4]

output[iloc,8]<-x$fit[5]

}

if(file.exists(paste(folder,pat,test,'REwaves.csv',sep='/'))){

#process LE

146

iloc<-iloc+1

x<-fitAwaves(paste(folder,pat,test,sep='/'),'r')

output[iloc,1]<-pat

output[iloc,2]<-test

output[iloc,3]<-'r'

output[iloc,4]<-x$fit[1]

output[iloc,5]<-x$fit[2]

output[iloc,6]<-x$fit[3]

output[iloc,7]<-x$fit[4]

output[iloc,8]<-x$fit[5]

}

}

}

write.csv(output,file=paste(folder,'output.csv',sep='/'))

cat('Done\n')

return(output)

1