TmciCg

sCmAHS

21

2

Hemodialysis Membrane With a High-Molecular-Weight Cutoff andCytokine Levels in Sepsis Complicated by Acute Renal Failure:

A Phase 1 Randomized Trial

Michael Haase, MD,1,2 Rinaldo Bellomo, MD, FRACP, FJFICM,1 Ian Baldwin, RN,1

Anja Haase-Fielitz, BPharm,1,2 Nigel Fealy, RN,1 Piers Davenport, MD,3 Stanislao Morgera, MD,2

Hermann Goehl, PhD,4 Markus Storr, PhD,4 Neil Boyce, MD,5 and Hans-Hellmut Neumayer, MD2

Background: Sepsis is the leading cause of acute renal failure. Intermittent hemodialysis (IHD) is acommon treatment for patients with acute renal failure. However, standard hemodialysis membranesachieve only little diffusive removal of circulating cytokines. Modified membranes may enable bothsuccessful IHD treatment and simultaneous diffusive cytokine removal.

Study Design: Double-blind, crossover, randomized, controlled, phase 1 trial.Setting & Participants: Tertiary intensive care unit. 10 septic patients with acute renal failure

according to RIFLE class F.Intervention: Each patient was treated with 4 hours of high-cutoff (HCO)-IHD and 4 hours of high-flux

(HF)-IHD.Outcomes & Measurements: We chose relative change in plasma interleukin 6 (IL-6) concentrations

from baseline to 4 hours as the primary outcome for effective cytokine removal. We measured plasmaand effluent concentrations of cytokines (IL-6, IL-8, IL-10, and IL-18) and albumin.

Results: Median age was 53 years (25th to 75th percentiles, 43 to 71 years). Both treatmentsachieved equal control of uremia. Four hours of HCO-IHD accomplished a greater decrease in plasmaIL-6 levels (�30.3%) than 4 hours of HF-IHD (1.1%; P � 0.05). HCO-IHD, but not HF-IHD, achievedsubstantial diffusive clearance of several cytokines (IL-6, 14.1 mL/min; IL-8, 75.2 mL/min; and IL-10,25.5 mL/min). Such clearance also was associated with greater relative decreases in plasma IL-8 andIL-10 levels in favor of HCO-IHD (P � 0.02, P � 0.04). We found significantly greater relative changesfrom prefilter to postfilter plasma IL-6, IL-8, and IL-10 values in favor of HCO-IHD (P � 0.02, P � 0.01, P� 0.01). During HCO-IHD, cumulative albumin loss into the effluent was 7.7 g (25th to 75th percentiles,4.8 to 19.6) versus less than 1.0 g for HF-IHD (P � 0.01).

Limitations: Small phase 1 trial.Conclusion: In septic patients with acute renal failure, HCO-IHD achieved simultaneous uremic

control and diffusive cytokine clearances and a greater relative decrease in plasma cytokine concentra-tions than standard HF-IHD.Am J Kidney Dis 50:296-304. © 2007 by the National Kidney Foundation, Inc.

INDEX WORDS: Sepsis; acute renal failure; intermittent hemodialysis; high-cutoff-point membranes;polyamide; cytokines.

ctm

auc

he combination of sepsis and acute renalfailure (ARF) is associated with very high

orbidity and mortality.1 Sepsis is the leadingause of ARF.1 Intermittent hemodialysis (IHD)s a common treatment for patients with ARF.linical and molecular biology research sug-ested that cytokines and other septic mediators

From the 1Intensive Care Unit, Austin Hospital, Univer-ity of Melbourne, Australia; 2Department of Nephrology,harité University Medicine, Berlin, Germany; 3Depart-ent of Medicine, Monash Medical Centre, Melbourne,ustralia; 4Gambro Dialysatoren GmbH, Device Research,echingen, Germany; and 5Australian Red Cross Bloodervice, University of Melbourne, Australia.

Received March 7, 2007. Accepted in revised form May 8,007. Originally published online as doi:

0.1053/j.ajkd.2007.05.003 on July 5, 2007.

American Journal of K96

ontribute to the pathogenesis of sepsis and mul-iple organ failure, including ARF.2-5 Their re-oval might be desirable.However, standard hemodialysis membranes,

lthough effective in diffusively clearing manyremic toxins, showed little diffusive removal ofirculating cytokines (water soluble; molecular

Trial registration: www.clinicaltrials.gov; study number:CT00333593.Address correspondence to Rinaldo Bellomo, MD, FRACP,

JFICM, Director of Intensive Care Research, Austin Hospi-al, 3084 Heidelberg, Victoria, Australia. E-mail: rinaldo.ellomo@austin.org.au© 2007 by the National Kidney Foundation, Inc.0272-6386/07/5002-0015$32.00/0doi:10.1053/j.ajkd.2007.05.003

N

Ftb

idney Diseases, Vol 50, No 2 (August), 2007: pp 296-304

wab

wokhmntttk

bgm(w

S

tfk

cdd

toddr(

I

tTdtpsaCmttad

H

cFMflb

raDeehmis54

oar(ma[dgua

M

8iipmh

bmIM

cIp

chuMopw

High Cutoff Point Hemodialysis in Sepsis 297

eight, 8 to 55 kd).6,7 This most likely is attribut-ble to the limited pore size of standard mem-ranes.Recently, high-cutoff (HCO)-point membranes

ith moderately increased pore size were devel-ped. These HCO membranes had greater cyto-ine removal capacity compared with standardigh-flux (HF) membranes in ex vivo experi-ents,8-10 animal experiments,11,12 and prelimi-

ary clinical studies.13,14 These membranes nowheoretically make it possible to treat septic pa-ients with ARF and simultaneously provide con-rol of uremia and fluid status, as well as cyto-ine removal.Accordingly, we conducted a phase 1, double-

lind, randomized, controlled study to investi-ate the biological effect of combining an HCOembrane and intermittent hemodialysis mode

HCO-IHD) in the treatment of septic patientsith ARF.

METHODS

tudy Population

The Human Research Ethics Committee of Austin Hospi-al (Melbourne, Australia) approved the study. Written in-ormed consent was obtained from each patient’s next ofin.Inclusion criteria comprised the simultaneous presence of

onsensus criteria for sepsis15 and dialysis-dependent ARFefined using the Risk, Injury, Failure, Loss, End-stage renalisease (RIFLE) criterion F.16

We excluded patients for whom there were limitations onherapy, for whom written informed consent could not bebtained, those who were not expected to survive 24 hoursespite full treatment, patients participating in another ran-omized controlled study, and patients who were likely toequire a diagnostic procedure outside the intensive care unitICU) on the study day.

ntervention

This study was a phase 1 investigation of a new experimen-al therapy for patients with sepsis associated with ARF.hus, we enrolled 10 adult patients with sepsis and ARF in aouble-blind, crossover, randomized, controlled trial at aertiary ICU from June 2006 to November 2006. Eachatient received 4 hours of HCO-IHD and 4 hours oftandard HF-IHD in random order. Randomization waschieved by means of a Microsoft Excel–based (Microsoftorp, Redmond, WA) random-number generator. Study he-odialysis membranes were labeled A or B by the manufac-

urer and could not be differentiated on inspection. Alloca-ion assignment to membrane type was concealed until datanalysis was complete. We ensured a 4-hour washout period

uring which no renal replacement therapy was applied.

emodialysis Technique

Vascular access was obtained using 13.5-Fr dual-lumenatheters (Niagara; Bard, Ontario, Canada). We used theresenius 4008S ARrT Plus dialysis machine (Freseniusedical Care AG, Bad Homburg, Germany). We set blood

ow at 200 mL/min and dialysate flow at 300 mL/min. Fluidalance was kept even during both study treatment periods.Both membranes were Polyflux filters of the same mate-

ial (polyarylethersulfone) and surface area (1.1 m2; Poly-mide; Gambro Dialysatoren GmbH, Hechingen, Germany).uring HCO-IHD, we used custom-made HCO-point dialyz-

rs, and during HF-IHD, commercially available HF dialyz-rs. The membrane of the HCO-point dialyzers (HCO 1100)as an estimated pore size of 10 nm, resulting in an esti-ated in vitro cutoff point of 100 to 150 kd and an estimated

n vivo cutoff point of 50 to 60 kd. The membrane of thetandard HF dialyzers (PF11-S) has an estimated pore size ofnm, resulting in an estimated in vitro cutoff point of 35 to

5 kd and an estimated in vivo cutoff point of 15 to 20 kd.Dialysate was produced online through a portable reverse

smosis unit. The composition of the dialysate fluid wasdjusted according to the specific needs of each patient withespect to sodium (135 to 145 mEq/L [mmol/L]), potassium3.0 to 4.0 mEq/L [mmol/L]), calcium (2.5 mEq/L [1.25mol/L]), chloride (105 to 110 mEq/L [mmol/L]), bicarbon-

te (30 to 35 mEq/L [mmol/L]), and magnesium (1.2 mg/dL0.5 mmol/L]). Phosphate was administered intravenouslyuring study treatment to maintain a serum phosphate levelreater than 3.1 mg/dL (�1.0 mmol/L). When necessary,nfractionated heparin was administered for both techniquest a rate of 500 U/h to prevent filter clotting.

easurements and Calculations

We measured cytokine (interleukin 6 [IL-6], 26 kd; IL-8,kd; IL-10, 2 � 18 kd; and IL-18, 18 kd), urea, and albumin

n plasma before (baseline) and at 4 hours (end) afternitiation of HCO-IHD or HF-IHD. Additional prefilter andostfilter measurements of all plasma cytokines and measure-ents of all cytokines in effluent were made at 1 (start) and 4

ours (end) of HCO-IHD and HF-IHD treatment.IL-6, IL-8, IL-10, and IL-18 were measured in duplicate

y means of enzyme-linked immunosorbent assay usingatched antibody pairs and recombinant standards (IL-6,

L-8, and IL-10; R & D Systems, Minneapolis, MN; IL-18;BL Int, Woburn, MA).Assay sensitivities were 1.6 to 3.0 pg/mL, and the coeffi-

ient of variation for each assay was less than 10% for IL-6,L-8, IL-10, and IL-18 concentrations of 15.6 to 1,000g/mL.Plasma albumin was measured using a multichannel bio-

hemical analyzer (Hitachi 747 Analyser; Boehringer Mann-eim, Indianapolis, IN). Effluent albumin was measuredsing immunoturbidometry (Turbimeter; Behringwerke AG,arburg, Germany). Intra-assay and interassay coefficients

f variation were 4.5% and 3%, respectively.17 Effluent/lasma concentration ratio (E/P ratio) and clearance ratesere calculated as follows:

Ce ⁄ (Cpi � Cpo)

E ⁄ P ratio �

2

wCfim(

aMcIO

A

pc

cw

aps

sahtgccrfiso

FTsoev7y

Haase et al298

Clearance � E ⁄ P ratio � Qd

here Ce is effluent concentration (picograms per milliliter),pi is plasma concentration (picograms per milliliter) atlter inlet, Cpo is plasma concentration (picograms perilliliter) at filter outlet, and Qd is dialysate flow rate

milliliters per minute).We recorded mean arterial pressure (MAP) hourly. Also,

djustments in norepinephrine (NE) dose to maintain anAP greater than 75 mm Hg were recorded hourly. We

alculated Acute Physiology and Chronic Health EvaluationI (APACHE II) score18 on ICU admission and Sequentialrgan Failure Assessment (SOFA) score19 on the study day.

nalysis of Data

The primary outcome is defined as relative change inlasma IL-6 concentrations from baseline to end of IHDomparing both types of treatment.

Other outcomes include group comparison of relativehanges in plasma IL-8, IL-10, IL-18, and albumin levels, asell as clearances from baseline to end of study treatment.Because data lacked normal distribution, we present them

s median and 25th to 75th percentiles. Relative changes inrefilter to postfilter plasma cytokine concentrations at the

tart and end of each study treatment were pooled to increase u

ample power. Sampling for measurement of cytokines andlbumin was missed at the end of 1 HCO-IHD treatment;owever, baseline values for this treatment were included inhe analysis. Values were compared between and withinroups using Wilcoxon signed-rank test and Bonferroniorrection, when appropriate. We present time effect toontrast it with treatment effect. To evaluate a time effect onelative change in cytokine concentrations, values during therst 4 hours of treatment were compared with those of theecond treatment period using Wilcoxon signed-rank test. Pf 0.05 or less is considered significant.

RESULTS

Patient enrollment into the study is shown inig 1. Patient baseline characteristics are listed inable 1. All patients presented with sepsis andevere ARF (RIFLE class F) and were on continu-us venovenous hemofiltration therapy beforenrollment. Median duration of such continuousenovenous hemofiltration was 2 days (25th to5th percentiles, 1 to 3). Median age was 53ears (25th to 75th percentiles, 43 to 71). Median

Figure 1. Patient enroll-ment into the study. HCO-IHD,high-cutoff-point intermittenthemodialysis; HF-IHD, stan-dard high-flux intermittent he-modialysis.

rinary output during the 24-hour period before

e0ncta2(o

cfoofpaTItPdts8

P

(wpt

I

wtc74t

ttcwp0b0

DR

12ptmHtm1lasmI

aH

High Cutoff Point Hemodialysis in Sepsis 299

nrollment was 180 mL (25th to 75th percentiles,to 530). Severe sepsis was caused by pneumo-

ia (5 patients), peritonitis (2 patients), and pan-reatitis, mediastinitis, or endocarditis (3 pa-ients). Median APACHE II score on ICUdmission was 24 (25th to 75th percentiles, 21 to8), and SOFA score on the study day was 1425th to 75th percentiles, 10 to 16). Sixty percentf patients survived hospital admission.Of 10 study patients, 5 were randomly allo-

ated to receive 4 hours of HCO-IHD first,ollowed by 4 hours of standard HF-IHD, and thether 5 patients, vice versa. In 3 patients, continu-us heparin infusion at constant dose was usedrom start to end of study treatment. The other 7atients did not require anticoagulation with hep-rin. Data for plasma urea levels are listed inable 2. During both study treatments, HCO-HD and HF-IHD, plasma urea levels substan-ially decreased from baseline to end (P � 0.01,

� 0.01). Patient fluid balance was kept evenuring both study treatments by using ultrafiltra-ion rates between 0 and 150 mL/h, which, onceet, were maintained in each patient for the entire-hour treatment period.

rimary Outcome: IL-6

At baseline, plasma IL-6 level was 177 pg/mL25th to 75th percentiles, 85 to 255) in patientsho were treated first with HCO-IHD and 147g/mL (25th to 75th percentiles, 24 to 422) forhose treated first with HF-IHD (P � 0.9).

We found a greater relative decrease in plasma

Table 1. Baseline Characteristics, Mode

PatientNo.

Age(y)

Sex(M/F)

Weight(kg)

APACHE IIScore

1 78 M 96 282 69 M 67 203 48 M 65 234 77 M 75 225 34 F 55 206 55 M 95 267 52 M 105 218 44 M 75 289 69 M 80 25

10 40 F 50 34

Abbreviations: M, male; F, female; APACHE II, Acute Phdmission; SOFA, Sequential Organ Failure Score on stF-IHD, standard high-flux intermittent hemodialysis.*Urinary output during 24 hours before the study.

L-6 levels after 4 hours of HCO-IHD compared g

ith HF-IHD (Fig 2; P � 0.05). There was noime effect on relative change in plasma IL-6oncentrations (first 4 hours, �14.6%; 25th to5th percentiles, �51.0 to 2.7 versus the secondhours, �13.6%; 25th to 75th percentiles, �28.7

o 18.8; P � 0.5).Figure 3 shows individual plasma IL-6 concen-

rations for patients on HCO-IHD and HF-IHDreatment at baseline and end of treatment. Whenomparing baseline plasma IL-6 concentrationsith values at the end of HCO-IHD treatment,lasma IL-6 levels decreased significantly (P �.02), whereas values remained unchanged fromaseline to the end of HF-IHD treatment (P �.3).

iffusive Cytokine Clearances andelative Changes

At the start, diffusive clearances for IL-6 were4.1 mL/min (25th to 75th percentiles, 5.0 to6.9) for HCO-IHD and 6.6 mL/min (25th to 75th

ercentiles, 0 to 34.8) for HF-IHD (P � 0.02). Athe end of treatment, IL-6 clearances were 9.6L/min (25th to 75th percentiles, 8.2 to 11.7) forCO-IHD and 1.6 mL/min (25th to 75th percen-

iles, 0 to 11.1) for HF-IHD (P � 0.03). Theajority of sample concentrations of IL-8, IL-

0, and IL-18 in effluent during HF-IHD wereess than the assay detection limits. At the startnd end of HCO-IHD treatment, median diffu-ive clearance values for IL-8 (75.2 and 68.3L/min), IL-10 (25.5 and 10.5 mL/min), and

L-18 (11.5 and 6.5 mL/min) were significantly

y Treatment Applied First, and Survival

FAore

Urinary Output*(mL)

Study TreatmentFirst

HospitalSurvival

9 180 HF-IHD Yes5 550 HF-IHD Yes2 175 HCO-IHD Yes2 3,000 HCO-IHD No0 520 HF-IHD Yes7 0 HCO-IHD No5 85 HF-IHD Yes5 250 HCO-IHD No6 0 HF-IHD Yes9 0 HCO-IHD No

y Chronic Health Evaluation Score on intensive care unity; HCO-IHD, high-cutoff-point intermittent hemodialysis;

of Stud

SOSc

11111111

ysiologudy da

reater compared with HF-IHD. Changes in cyto-

kw0

IcHIcH�swfdl

cpI

Tab

le2.

Pla

sma

Cyt

oki

ne,

Ure

a,an

dA

lbu

min

Lev

els

atB

asel

ine

and

En

do

fTre

atm

enta

nd

Th

eir

Rel

ativ

eC

han

ges

Fro

mB

asel

ine

toE

nd

ofH

igh

Cu

toff

and

Hig

h-F

lux

Inte

rmit

ten

tHem

od

ialy

sis

Var

iabl

eV

alue

sat

Bas

elin

eP

Val

ues

atE

ndP

Rel

ativ

eC

hang

e(%

)P

HC

O-I

HD

HF

-IH

DH

CO

-IH

DH

F-I

HD

HC

O-I

HD

HF

-IH

D

8(p

g/m

L)61

.0(2

8.6-

97.4

)58

.2(3

2.2-

97.7

)0.

351

.6(2

0.8-

102.

9)64

.3(2

8.3-

107.

5)�

0.01

�20

.5(�

28.5

-�2.

4)15

.2(�

2.0-

21.2

)0.

0210

(pg/

mL)

56.2

(6.8

-214

.5)

56.5

(6.6

-227

.5)

0.6

33.2

(6.1

-75.

4)51

.4(6

.2-2

46.9

)0.

03�

19.4

(�53

.3-3

.7)

6.3

(�14

.1-2

8.9)

0.04

18(p

g/m

L)1,

628

(1,2

04-4

,347

)1,

418

(1,2

06-4

,302

)0.

61,

349

(908

-3,6

83)

1,61

5(1

,012

-4,0

95)

0.4

�4.

5(�

12.6

-0.3

)�

2.5

(�8.

3-3.

6)0.

3a

(mg/

dL)

51.6

(34.

2-84

.2)

55.8

(19.

2-10

3.7)

0.9

29.9

(18.

3-57

.3)

34.5

(9.8

-45.

9)0.

9�

41.0

(�53

.6-�

34.9

)�

40.5

(�50

.9-�

34.1

)0.

7um

in(g

/L)

26.0

(23.

3-31

.0)

26.5

(20.

0-31

.5)

0.6

25.5

(22.

5-29

.0)

26.5

(20.

8-31

.0)

0.3

�4.

8(�

6.7-

1.7)

0.0

(�4.

5-3.

5)0.

5

ote:

Pla

sma

cyto

kine

leve

lspr

esen

ted

asm

edia

n(2

5thto

75th

perc

entil

es).

To

conv

ertp

lasm

aur

eain

mg/

dLto

mm

ol/L

,mul

tiply

by0.

166;

albu

min

ing/

Lto

g/dL

,mul

tiply

by. bb

revi

atio

ns:I

L-6,

inte

rleuk

in6;

HC

O-I

HD

,hig

h-cu

toff

inte

rmitt

enth

emod

ialy

sis;

HF

-IH

D,s

tand

ard

high

-flux

inte

rmitt

enth

emod

ialy

sis.

(htdt�

Haase et al300

ine clearances from start to end of HCO-IHDere not significant (IL-6, P � 0.3; IL-8, P �.3; IL-10, P � 0.5; and IL-18, P � 0.2).

In Table 2, we list plasma IL-8, IL-10, andL-18 levels at baseline and end and their relativehanges from baseline to end of treatment forCO-IHD and HF-IHD. Treatment with HCO-

HD achieved a significantly greater relative de-rease in plasma IL-8 levels compared withF-IHD (P � 0.02). No time effect was found (P0.2). The relative decrease in IL-10 levels was

ignificantly greater during HCO-IHD comparedith HF-IHD (P � 0.04), and no time effect was

ound (P � 0.4). We observed no significantifference in relative decreases in plasma IL-18evels (P � 0.3).

For IL-6, IL-8, and IL-10, we found a signifi-antly greater relative decrease in prefilter toostfilter plasma cytokine levels during HCO-HD compared with HF-IHD (Table 3).

Plasma albumin levels at baseline and end and

Figure 2. Relative changes in plasma interleukin 6IL-6) concentrations during high-cutoff (HCO) intermittentemodialysis (IHD) and high-flux (HF)-IHD from baselineo end of treatment. Relative changes in plasma IL-6 levelsuring HCO-IHD, �30.3% (25th to 75th percentiles, �53.4o �8.3) versus HF-IHD, 1.1% (25th to 75th percentiles,14.3 to 32.8; P � 0.05).

their relative changes are listed in Table 2. At theIL-

IL-

IL-

Ure

Alb N

0.1 A

smw(tlpH

vmtm

pp

Hembtd�b0Ntpw

tWpcpdrcwl

dbi

aB 3.

(p

im

High Cutoff Point Hemodialysis in Sepsis 301

tart of HCO-IHD, albumin clearance was 1.9L/min (25th to 75th percentiles, 1.5 to 2.7),hich decreased significantly to 0.8 mL/min

25th to 75th percentiles, 0.5 to 1.6; P � 0.01) athe end. During HCO-IHD, cumulative albuminoss into the effluent was 7.7 g (25th to 75th

ercentiles, 4.8 to 19.6) versus less than 1.0 g forF-IHD (P � 0.01).During each type of study treatment, 1 indi-

idual patient had 2 bottles of 20% human albu-in (100 mL each bottle) prescribed by the

reating clinician for hemodynamic manage-ent.During HCO-IHD, there was a trend for im-

roved MAP from 76.5 mm Hg (25th to 75th

ercentiles, 69.5 to 87.3) at baseline to 81.5 mm

Figure 3. Individual plasma interleukin 6 (IL-6) concennd high-flux (HF)-IHD at baseline and end of treatment. Baseline of HF-IHD compared with end of HF-IHD, P � 0.0

Table 3. Relative Changes From Prefilter to PostfilterPlasma Cytokine Levels During High-Cutoff and

High-Flux Intermittent Hemodialysis

PlasmaCytokine HCO-IHD HF-IHD P

IL-6 �13.5 (�22.1-�0.7) 0.9 (�5.6-12.9) 0.02IL-8 �7.2 (�10.9-1.8) 2.2 (�1.1-7.9) 0.01IL-10 �17.5 (�24.3-7.6) 0.8 (�8.8-6.5) �0.01IL-18 �17.1 (�25.0-8.2) 1.8 (�8.3-6.2) 0.2

Note: Relative changes presented as median percent25th to 75th percentiles). Values at start and end wereooled.Abbreviations: IL-6, interleukin 6; HCO-IHD, high-cutoff

ntermittent hemodialysis; HF-IHD, standard high-flux inter-

aittent hemodialysis.

g (25th to 75th percentiles, 72.5 to 96.5) at thend (P � 0.09). During HF-IHD, MAP was 77.0m Hg (25th to 75th percentiles, 72.0 to 89.3) at

aseline and 78.0 mm Hg (25th to 75th percen-iles, 72.0 to 87.8) at the end. Furthermore,uring HCO-IHD, NE dose changed from 8.0g/min (25th to 75th percentiles, 0.0 to 12.3) ataseline to 2.0 �g/min (25th to 75th percentiles,.0 to 12.3) at the end, whereas during HF-IHD,E dose was stable at 4.0 (25th to 75th percen-

iles, 0.0 to 12.8) and 4.0 �g/min (25th to 75th

ercentiles, 0.0 to 12.3). However, these changesere not significant.

DISCUSSION

We conducted a phase 1 randomized clinicalrial of HCO-IHD in patients with septic ARF.

e tested whether HCO-IHD can affect levels oflasma IL-6 and other cytokines and whether itan achieve diffusive cytokine clearance by com-aring it with HF-IHD. We found evidence ofiffusive clearance for several cytokines and aelative decrease in plasma IL-6, IL-8, and IL-10oncentrations. These findings were associatedith stable plasma albumin concentrations and

imited albumin loss.Diffusive clearance of cytokines during hemo-

ialysis with commercially available HF mem-ranes is low, and adsorptive clearance is lim-ted. Accordingly, although control of uremia

during high-cutoff (HCO) intermittent hemodialysis (IHD)of HCO-IHD compared with end of HCO-IHD, P � 0.02.

trationsaseline

nd fluid status can be achieved using IHD or

ocmwctbpmhcaumokaatwfiaHsfsakdcArmvAba

Hpiadiiribimi

tlit

obkblebamcmdi

atAcccatmatso

Istseebcs

noetct

Haase et al302

ther forms of renal replacement therapy, plasmaytokine levels often remain unchanged.6,7,20 Si-ultaneous control of uremia and fluid status, asell as extracorporeal removal of circulating

ytokines, appears theoretically desirable in sep-ic patients with ARF.6,20 Recently, HCO mem-ranes were developed with the aim of removinglasma cytokines.21,22 When used in convectiveode (hemofiltration), HCO membranes achieve

igher cytokine removal capacity compared withonventional membranes.13 Unfortunately, theylso cause moderate to severe albumin losses ofp to 15 g/4-hour session.10 However, if HCOembranes were combined with a diffusive mode

f renal replacement therapy, a degree of cyto-ine clearance (cytokine dialysis) could bechieved while simultaneously delivering fluidnd uremic control. We recently confirmed theseheoretical assumptions ex vivo.8,9,21 In this study,e further tested these assumptions in vivo. Ourndings confirm diffusive clearance of cytokiness seen in ex vivo experiments.9,21 IntermittentCO-HD was not investigated previously in

eptic patients with ARF. In this study, we showor the first time that during only 4 hours, diffu-ive therapy with HCO-point membranes couldchieve the diffusive removal of several cyto-ines and was associated with a greater relativeecrease in plasma concentrations of severalytokines simultaneously in septic patients withRF (cytokine dialysis). In the only previous

andomized controlled trial investigating HCOembranes at low dialysate flow (continuous

enovenous hemodialysis) in septic patients withRF, no overall decrease in plasma IL-6 valuesy renal replacement therapy was seen, evenfter 48 hours of treatment.23

Despite previous publications on the use ofCO membranes in septic patients with ARF, theresent study is the first double-blind random-zed clinical trial of this technology, as well as ofny other blood purification technology. Suchouble blinding is important in eliminating biasn patient selection and choice of timing ofntervention. The findings are objective, usingobust and well-established assays for cytokinesn plasma, and differences are significant androadly consistent across several cytokines. Thentervention was technically simple and used aembrane of a material (Polyamide) used safely

n innumerable dialysis treatments worldwide. s

Following the concept of the “peak concentra-ion hypothesis,” we aimed to decrease all circu-ating mediators at high plasma concentrations,ncluding proinflammatory and anti-inflamma-ory substances.24

With the 4 cytokines measured, we provide 1f the broadest selections of cytokines in septiclood purification studies. We chose these cyto-ines for proof of concept, ie, because of theiriological importance and because their molecu-ar weight was within the range of a potentiallyffective elimination by HCO dialysis mem-ranes. In a recent study, high plasma IL-6, IL-8,nd IL-10 levels were associated with increasedortality in human septic ARF.25 Finally, we

hose change in plasma IL-6 levels as the pri-ary outcome because there is sufficient evi-

ence that high IL-6 levels are associated withncreased risk of death.25,26

However, our study has limitations. As withll phase 1 studies, it carries a greater chance ofype I and type II error than larger studies.ccordingly, our findings might be caused by

hance alone. The observation of consistenthanges in plasma concentrations and relativehanges in concentration of several cytokinescross the HCO filter (relative change in prefiltero postfilter values) make this less likely. Further-

ore, given the potential for harm (albumin loss)nd the experimental nature of this therapy (firstime in humans), we considered that a phase 1tudy was necessary before applying this technol-gy to a larger cohort of patients.Although there was an effect on plasma IL-6,

L-8, and IL-10 concentrations, we could nothow a significant effect on plasma IL-18. Givenhat IL-18 is smaller in molecular size than IL-6,ize-dependent exclusion from effluent does notxplain our findings. A high level of IL-18 gen-ration sufficient to exceed removal might haveeen responsible for our observations. The highoncentrations of IL-18 seen in our patientsupport this possibility.

The IL-6 and IL-10 clearances reported mayot appear sufficiently large to explain a decreasef 20% to 30% in plasma concentrations. How-ver, measurement of effluent cytokines is likelyo have limited accuracy because cytokine con-entrations in effluent decrease to levels less thanhe optimal performance range for standard as-

ays at dialysate flow of 300 mL/min. In keeping

wchpmwotm

stprotaags

rathma

flimnetwc

ttrHdcHodmtd

b

cc

asspkepHw

c

RF

fs

N

C

ks

s

cpd

haT

hI

d

acp

pti

h

High Cutoff Point Hemodialysis in Sepsis 303

ith this concept, the interquartile range for ourlearances was wide and we therefore mightave underestimated true clearance. It also isossible that other molecules were removed byeans of HCO-IHD that we did not measure, buthich might have participated in the generationf IL-6, IL-8, and IL-10. Their removal, ratherhan a direct effect on the cytokine themselves,ight have been responsible for our findings.There was loss of albumin, which may be a

ource of concern to some clinicians. However,his loss did not translate into a decrease inlasma albumin levels or significant albumineplacement. The clinical significance of the lossf albumin in the present study remains uncer-ain. However, our observations invite caution inpplying this therapy and suggest the need forlbumin monitoring should HCO-IHD be used atreater intensity or for longer periods in futuretudies.

Adsorption may have contributed to cytokineemoval in the present study. However, poly-mide membranes previously achieved little cy-okine adsorption.27 In addition, both treatmentsad equal adsorptive surfaces because they usedembranes of the same chemical composition

nd surface area.There were no differences in fluid balance or

uid removal during both study treatments, mak-ng it unlikely that differences in fluid manage-ent account for the physiological observations

oted in the study. Uremic control also wasquivalent, making it unlikely that differences inhe clearance of small-molecular-weight toxinsere responsible for the differences in cytokine

oncentrations.The washout period during the crossover to

he second therapy was short. It is possible thathe therapy applied first would have had a car-yover effect on the therapy applied second.owever, because no effect was seen with stan-ard HF-IHD, no conceivable carryover effectan be logically expected during subsequentCO-IHD. If HCO-IHD had a carryover effectn standard IHD, this would have potentiallyecreased plasma cytokine levels during it, thusaking detection of an effect less likely. We note

hat no time effect on plasma cytokine levels wasetected during the study.Finally, additional larger studies are needed to

etter understand the relationship between such 2

ytokine extraction, cytokine blood levels, andlinical effects.

In conclusion, HCO-IHD applied for 4 hourschieved equivalent uremic and fluid control ineptic patients with ARF, but was associated withimultaneous diffusive cytokine removal and ap-eared more effective in decreasing plasma cyto-ine concentrations than standard HF-IHD. Thisffect was not associated with a decrease inlasma albumin levels. Additional studies ofCO-IHD in the treatment of septic patientsith ARF seem justified.

ACKNOWLEDGEMENTSThe authors thank the intensive care nursing staff for their

ooperation with this study.Support: Dr. Haase holds a postdoctoral Feodor-Lynen

esearch Fellowship from the Alexander von Humboldt-oundation, Germany.Financial Disclosure: None.

REFERENCES1. Uchino S, Kellum JA, Bellomo R, et al: Acute renal

ailure in critically ill patients: A multinational, multicentertudy. JAMA 294:813-818, 2005

2. Parrillo JE: Pathogenetic mechanisms of septic shock.Engl J Med 328:1471-1477, 19933. Bone RC: Sir Isaac Newton, sepsis, SIRS, and CARS.

rit Care Med 24:1125-1128, 19964. Pinsky MR, Vincent JL, Deviere J, et al: Serum cyto-

ine levels in human septic shock. Relation to multiple-ystem organ failure and mortality. Chest 103:565-575, 1993

5. Hack CE, Aarden LA, Thijs LG: Role of cytokines inepsis. Adv Immunol 66:101-195, 1997

6. Kellum JA, Johnson JP, Kramer D, et al: Diffusive vs.onvective therapy: Effects on mediators of inflammation inatients with severe systemic inflammatory response syn-rome. Crit Care Med 26:1995-2000, 19987. Mariano F, Fonsato V, Lanfranco G, et al: Tailoring

igh-cut-off membranes and feasible application in sepsis-ssociated acute renal failure: In vitro studies. Nephrol Dialransplant 20:1116-1126, 20058. Lee WC, Uchino S, Fealy N, et al: Super high flux

emodialysis at high dialysate flows: An ex vivo assessment.nt J Artif Organs 27:24-28, 2004

9. Uchino S, Bellomo R, Morimatsu H, et al: Cytokineialysis: An ex vivo study. ASAIO J 48:650-653, 200210. Morgera S, Klonower D, Rocktaschel J, et al: TNF-

lpha elimination with high cut-off haemofilters: A feasiblelinical modality for septic patients? Nephrol Dial Trans-lant 18:1361-1369, 200311. Lee PA, Weger GW, Pryor RW, et al: Effects of filter

ore size on efficacy of continuous arteriovenous hemofiltra-ion therapy for Staphylococcus aureus-induced septicemian immature swine. Crit Care Med 26:730-737, 1998

12. Kline JA, Gordon BE, Williams C, et al: Large-poreemodialysis in acute endotoxin shock. Crit Care Med

7:588-596, 1999

enM

ta

CsgC

ftIQ2

hA

I1

tfplC

trp1

rv

flI

rcs

tp2

cf

ct4

Ct

Haase et al304

13. Morgera S, Haase M, Kuss T, et al: Pilot study on theffects of high cutoff hemofiltration on the need for norepi-ephrine in septic patients with acute renal failure. Crit Careed 34:2099-2104, 200614. Morgera S, Rocktaschel J, Haase M, et al: Intermit-

ent high permeability hemofiltration in septic patients withcute renal failure. Intensive Care Med 29:1989-1995, 2003

15. Members of the Society of Crit Care Med Consensusonference Committee: Society of Crit Care Med Consen-

us Conference: Definitions for sepsis and organ failure anduidelines for the use of innovative therapies in sepsis. Critare Med 20:864-874, 199216. Bellomo R, Ronco C, Kellum JA, et al: Acute renal

ailure—Definition, outcome measures, animal models, fluidherapy and information technology needs: The Secondnternational Consensus Conference of the Acute Dialysisuality Initiative (ADQI) Group. Crit Care 8:R204-R212,00417. Haase M, Bellomo R, Morgera S. A pilot study of

igh-adsorption haemofiltration in human septic shock. Int Jrtif Organs 30:108-117, 200718. Knaus WA, Draper EA, Wagner DP, et al: APACHE

I: A severity of disease classification system. Crit Care Med3:818-829, 198519. Vincent JL, de Mendonca A, Cantraine F, et al: Use of

he SOFA score to assess the incidence of organ dysfunction/ailure in intensive care units: Results of a multicenter,rospective study. Working group on “sepsis-related prob-ems” of the European Society of Intensive Care Medicine.

rit Care Med 260:1793-1800, 1998 1

20. Sander A, Armbruster W, Sander B, et al: Hemofiltra-ion increases IL-6 clearance in early systemic inflammatoryesponse syndrome but does not alter IL-6 and TNF alphalasma concentrations. Intensive Care Med 23:878-884,99721. Uchino S, Bellomo R, Goldsmith D, et al: Cytokine

emoval with a large pore cellulose triacetate filter: An exivo study. Int J Artif Organs 25:27-32, 200222. Uchino S, Bellomo R, Goldsmith D, et al: Super high

ux hemofiltration: A new technique for cytokine removal.ntensive Care Med 28:651-655, 2002

23. Morgera S, Slowinski T, Melzer C, et al: Renaleplacement therapy with high-cutoff hemofilters: Impact ofonvection and diffusion on cytokine clearances and proteintatus. Am J Kidney Dis 43:444-453, 2004

24. Ronco C, Bonello M, Bordoni V, et al: Extracorporealherapies in non-renal disease: Treatment of sepsis and theeak concentration hypothesis. Blood Purif 22:164-174,00425. Simmons EM, Himmelfarb J, Sezer MT, et al: Plasma

ytokine levels predict mortality in patients with acute renalailure. Kidney Int 65:1357-1365, 2004

26. Oberholzer A, Souza SM, Tschoeke SK, et al: Plasmaytokine measurements augment prognostic scores as indica-ors of outcome in patients with severe sepsis. Shock 23:488-93, 200527. Bouman CSC, van Olden RW, Stomenbeek CP, et al:

ytokine filtration and adsorption during pre- and postdilu-ion hemofiltration in four different membranes. Blood Purif

6:261-268, 1988