Gut1993; 34:1714-1717
Prediction of the response of chronic hepatitis C to interferon alfa: a statistical analysis of pretreatment
variables
JCamps, S Crisostomo, MGarcia-Granero, J I Riezu-Boj, M P Civeira, J Prieto
Abstract
Pretreatment variables that could predictthe response of chronic hepatitis C to interferon alfa treatment have not been fully assessed.
Eighteen baseline variables were evaluated in a seriesof100consecutivepatientstreated with a 12 monthcourse of interferonalfa. For the purposes of this study, responsewas defined as the return to normal of aminotransferase activities before the third month of treatment.
Seventypercentof thepatientsrespondedto treatment. Sixvariables wereassociatedwith an increased likelihood ofresponse assessed by univariateanalysis. With stepwise multiple regressionanalysisassessment,however,only three variables remained independently pre- dictive ofresponse:lowyglutamyltransferase (yGT) activities(p<0.001),absenceofobesity (p=0.005),and absenceofcirrhosis(p=0-01).
The responserateinpatientswithyGTactivi- ties <066 iikat/l (n=55)was 78%and 60% in patients with values >066 ikat/l (n=45) (p=.0048). Response was attained in 75% of non-obese patients (n=80), compared with only 50% of obese patients (n=20)(p=003).
Finally, 80% of patients without cirrhosis (n=76) responded, while among those with cirrhosis (n=24) the response rate was only 37%(p<0001).All 23patients without cirrho- sis, <40 years old, and with yGT activities
<0-66 tkat/l responded to treatment, while only28-5% of 14patients with cirrhosis, >40 years old, and with yGT activities >0 66 pkat/l respondedtointerferon affa(p<0-001).
Thosefindingsmaybe useful whenevaluating interferonalfa trialsand itissuggestedthat this treatmentshould beappliedearlyin thecourse of chronichepatitisC.
(Gut1993; 34: 1714-1717)
LiverUnit, Department ofInternalMedicine, ClinicaUniversitaria, UniversidaddeNavarra, Pamplona,Spain
JCamps SCris6stomo MGarcia-Granero
JIRiezu-Boj M PCiveira J Prieto Correspondenceto:
ProfessorJ Prieto, DepartmentofInternal Medicine,Clinica Universitaria,31008- Pamplona,Spain Accepted for publication 22April 1993
Interferon alfa has been assessed in thetreatment
ofchronichepatitisC in several trials.`6Inmost
studies normalisation of aminotransferase activi- ties has been attained in about 50% of the patients and relapses after stopping interferon alfa are usuallyseen in halfof theresponders.6 Pretreatment parameters that might predict
responsetointerferonalfa havenotbeenclearly determined. Insometrialspredictiveparameters
ofresponse were not found.'257 Other studies have suggested, however, that the absence of cirrhosis andayoungeragepredictedresponseto
interferon alfa.8-2 Differentcriteria when defin- ingresponsetointerferonalfawereused in these
reports, which might account for the discrep-
ancy. The knowledge of factors predictive of
responseinchronichepatitis Chasaremarkable
clinical interest, as was the case in chronic hepatitis B.'3
This study was aimed at assessing several pretreatment clinical variables as predictors of response to interferon alfa, in a series of 100 consecutive patients treated with a 12 month course of lymphoblastoid interferon alfa.
Univariate analysis was followed by astepwise multiple regression analysis to evaluate the independent predictivevalue of each variable.
Patients and methods
PATIENTS
One hundredpatients with chronic hepatitis C consecutively treated with the same schedule of lymphoblastoid interferon alfa (Wellferon, Wellcome Research Foundation, Beckhenham, UK) were included in thisstudy. Eighty four of them hadparticipatedin a randomised control- led trial of 12 monthinterferonalfa v no treat- ment'and 16patientsweretreatedwiththe same dose and schedule in an uncontrolled study. All patients were treated between 1988 and 1991.
Interferonalfawasgiven inastepdown schedule:
3million units (MU)daily for two months, 3 MU threetimesperweekforthree months, and 1 5 MU three times per week for seven months.
Before treatment all patients had a clinical evaluation, including a medical history with specialattention toepidemiological data, analyti- cal blood tests, and a percutaneousliver biopsy performed within three months before treat- ment. Other causes of chronic hepatitis were carefully excluded. All patients, except three, had anti-HCV antibodies by a second genera- tion enzyme immunoassay (Ortho Diagnostic Systems, Raritan, NJ). Inall casesaminotrans- ferase activities had beenraised for atleast six months before starting treatment and the liver biopsyshowedchronichepatitis withorwithout cirrhosis. Patientswithanti-HIV antibodiesand other serious medical illnesses were excluded.
All patients were instructed to stop alcohol intake atleast six monthsbeforestarting treat- ment. The trials were approved by the Local Ethics Committee and written informed consent wasobtained from allpatients.
Aswell aspretreatmentvariables, neutralising anti-lymphoblastoid interferon alfa antibiodies wereassessedby aspecific bioassay inasample collected duringthe last weekof treatment in 15 of 30patientswhodidnotrespondtotreatment to investigate their potential influence in non- responsiveness. The bioassay is based on the protective effect of lymphoblastoid interferon alfaonmonkeykidneyV3 cellschallengedwith 1714
group.bmj.com on August 2, 2012 - Published by
gut.bmj.com Downloaded from
Prediction oftheresponseofchronic hepatitis Ctointerferon alfa:astatistical analysis ofpretreatmentvariables
Semliki Forest virus and was performed as
described elsewhere.'4 Purified lymphoblastoid interferon alfa of standard potency and con-
trol anti-serum samples (kindly supplied by Wellcome Research Foundation, Beckenham, UK) were used to confirm the specificity and
sensitivityoftheassay.
DEFINITION OF RESPONSE
Responsewasdefined as a complete normalisa- tion of aminotransferase activities before the third month oftreatmentandpersisting atleast until the fifth month of theinterferon alfacourse.
The initiation of theresponse beyond the third month oftreatmentwasnotseeninany case.For the present analysis those patients who had responded initially and relapsed when the inter- feron dosewastaperedto1-5 MU three timesper weekatfivemonthswereconsidered responders.
All other patients, including 16 whose alanine aminotransferase (ALT) activities decreased by
more than 50% but did not return to normal,
wereconsidered non-responders. Our definition ofresponse wasintendedtoassessthecapacity of interferon alfa forinducingabiochemical remis- sion of thedisease.
PRETREATMENT VARIABLES
Eighteenpretreatmentvariableswereassessedin eachpatient: age, sex,historyofacutehepatitis andtransfusions,presenceofobesitydefinedby
a body weight >20% the ideal weight (deter- minedwithanage/sexbased standardtable),and theestimatedtime of evolution consideredasthe interval between anobvious percutaneous risk factor (for example, transfusion) or the first detected ALT rise and the start oftreatment,
wereallrecordedfromtheclinicalhistoryof each patient. Liver biopsyspecimens were classified intotwo groupsbyanexperiencedliverpatholo- gist: chronic hepatitis (either persistent (five cases)oractive (71 cases))and active cirrhosis.
Presence of fatty degeneration of hepatocytes
was also assessed. Before starting treatment ALT(normal range: 0 10-0 48
[tkat/l),
alkaline phosphatase(normalrange: 1 21-3-45[tkat/l),
y glutamytransferase (yGT) (normalrange: 0-08- 0-63 pkat/l), albumin,andyglobulin values, aswell as neutrophil and platelet counts, were
determined in each case. Procollagen type III
aminoterminal peptide,amarker of fibrosis and inflammatory activity,'5 was assessed using a commercial radioimmunoassay kit ( RIA-gnost PIIIP, Behringwerke AG, Marburg, Germany).
Anti-nuclear and anti-smoothmuscle antibodies
were assessed by indirect immunofluorescence and titres higher than 1/80 were considered positive.
STATISTICS
Univariateanalysiswasappliedtoeach variable to assess its association with response to inter-
feron alfatreatment. X2and Fisher's exact tests were used for dichotomous variables, and Student'st testfor continuousvariables. Timeof
evolution and yGT were logarithmically trans-
formed before analysis. Cathegorical explana-
tory variables weretransformed to 0/1 dummy variables.'6 Stepwise multiple linear regression analysis (Stat View SE+graphics, 1988 Abacus Concepts Inc, Berkeley, CA) was subsequently applied to all variables to assess theirindepen- dentpredictivevalue.Statisticalsignificance was assumedwhere p-005 andall p values are two tailed.
Results
UNIVARIATE ANALYSIS
Seventy per cent of the patients responded to treatment (as defined in the methods section), while theremaining 30% did not. Table I shows the results of the univariate analysis. Six vari- ables were found to be significantly associated with treatment response. Response tointerferon alfawassignificantlymorefrequent inthe young (p<0-001), in patients without cirrhosis (p<0001), and in patients withlow yGT values (p=0 003). Theremainingthree predictive vari- ables were lessstronglyassociated with response tointerferon alfa (TableI). Absenceofobesity, male sex, and a high platelet count predicted response to interferon alfa. Only 37% of the patients with cirrhosis (n=24) responded to interferonalfa, whilein 80%ofthe patients with chronic hepatitis (n=76)
aminotransferase
activities returned to normal (p<0-001). The response rateamongpatients <45 years (n=39) was 92%, while itdecreased to 56% in patients>45years(n=61) (p<0-001).Patientswith yGT values 60-66
[ikat/l
hadaresponse rateof78%, while response wasobtainedinonly60%of the patients with higher yGT values(p=0048). All patients (n=23) without cirrhosis, <40 years, and withyGTvalues<0'66 [tkat/l
respondedto interferon alfawhile only 64% of non-cirrhotic patients, >53 years or with yGT values >1-00 iikat/l (n=28), or bothresponded to this treat- ment. In contrast, the response rate was only 28-5%amongpatients with cirrhosis,>40years and withyGT values>0 66ikat/l
(n=14).When univariate analysis was applied to the patients divided by liver histological tests into
TABLE I Univariate analysis ofpretreatment variables
Non- Responders responders
Variable (n=70) (n=30) pValue
Age(y)* 42-7 53-5 <0001
Sex(M/F) 46/24 13/17 004
Obesity (%) 14 33 0 03
Acutehepatitis(%) 30 27 NS
Transfusions (%) 53 40 NS
Timeofevolution(months)* 46-6 58-7 NS
ALT(Rkat/l)*at 2-66 2-36 NS
Alkalinephosphatase
(Rkat/l)*bt 2-61 2-36 NS
yGT([tkat/l)*ct 0-61 0-91 0 003
Neutrophils(109/l)* 3-08 270 NS
Platelets(109/1)* 179 152 0-02
Album,in(g/l)* 42-9 41-1 NS
yGlobulins
(g/l),*
13-7 15-3 NSPIIIP(ng/ml)* t 18-7 219 NS
ANApositive(%) 9 3 NS
SMApositive(%) 17 13 NS
Chronichepatitis/cirrhosis 61/9 15/15 <0-001
Fattydegeneration (%) 34 50 NS
*Resultspresentedasmeans.tNormalranges:(a)010-048
stkat/l,(b)1-21-3-45 skat/l,(c)0-08-063
[skat/l,
(d)6-12ng/mL.ALT=alanineaminotransferase;yGT=glutamyltransferase;
PIIIP=procollagenaminoterminalpeptide;ANA=anti-nuclear antibodies; SMA=anti-smoothmuscleantibodies.
1715
group.bmj.com on August 2, 2012 - Published by
gut.bmj.com Downloaded from
Camps,Cris6stomo,Garcfa-Granero,Riezu-Boj, Civeira,Prieto TABLE II Univariateanalysis ofpretreatmentvariables in
patientswith chronichepatitis
Non- Responders responders
Variable (n=61) (n=15) pValue
Age (y)* 40-9 51-6 0-008
Sex (M/F) 40/21 6/9 NS
Obesity (%) 16 47 0-01
Acutehepatitis(%) 33 20 NS
Transfusions (%) 56 40 NS
Timeof evolution (months)* 42-9 59 0 NS
ALT(,tkat/l)* 2-66 2-38 NS
Alkalinephosphatase(Rkat/l)* 2-55 2-16 NS
yGT (,ukat/l)* 0-56 0 90 0-03
Neutrophils(109/l)* 3-12 2-92 NS
Platelets(109/l)* 188 158 0-05
Albumin (g/l)* 43-3 42-7 NS
yGlobulins(g/l)* 13-5 13-4 NS
PIIIP(ng/ml)* 17-8 16-6 NS
ANApositive(%) 7 7 NS
SMA positive(%) 16 7 NS
Fattydegeneration(%) 34 47 NS
*Resultspresentedasmeans.
Abbreviations the same as in Table I.
twogroups (chronichepatitis/cirrhosis), among patients with cirrhosis (n=24), none of the variables predicted response, while among patients without cirrhosis (n=76), young age, absenceof obesity, low yGT values,and a high plateletcount wereallassociated withresponse (Table II).
Neutralising anti-lymphoblastoid interferon alfa antibodies were detected in only one of 15 patients tested who did not respond (7%).
STEPWISE MULTIPLE LINEAR REGRESSION ANALYSIS
Stepwise multiple regression analysis showed that yGT values (p<0-001), obesity (p=0-005) and histological tests (p=0 01) were the only independently predictive variables. Low yGT values,the absenceofobesity,and the absenceof cirrhosis wereallindependently associatedwith a positive response to interferon alfa. The remainingpretreatmentvariables found predic- tive on univariate analysis added no further improvement in discrimination. Total body weight did not influence the response to treat- mentbecausethemeanweight of responders and non-responders was similar (kg; responders:
mean(SEM) 66-9 (12-7), non-responders: 67-5 (9-2)). Therefore, it was obesity itself, rather than total weight, which seemed to influence response.
Stepwisemultiple regression analysiswasalso applied to the group ofpatients with cirrhosis (n=24) without finding significantly predictive variables,althoughobesity, fatty degeneration of hepatocytes, andhigh yglobulinvalues tended to correlate with non- response. None of the responders were obese while 20% of the non- responders were obese; fatty degeneration was presentin 33%oftherespondersandin 53% of thenon-responders;mean((SD), g/l)yglobulin valuesinresponderswere: 15-1(4-9),compared with 17-3 (4-8) in non-responders. Among patients without cirrhosis (n=76) a low yGT value (p<0 001), the absence of obesity (p=
0-008), andahigh plateletcount(p-002)were independentlyassociated with responsetointer- feron alfa.
Discussion
It has not been clearly established if pretreat- mentvariables thatmightpredict theresponseof chronic hepatitisC tointerferon alfa do exist.'2 Some studies did not find predictive vari- ables,'2I 7 while others gave quite disparate results.8'- 17 These discrepanciesmay beattrib- utedtothesmallnumberofpatients includedin some studies, to different criteria for defining response, andto shortcomings in the statistical analyses. Some of these analyses were only univariate. Moreover, some studies have com- pared patients treated with different doses of interferon alfa and it has been establishedthat different schedules of interferon alfamayachieve different response rates.3 In this study, we clearly show,inalargeseries ofpatientsconsecu- tively treated with the same schedule of inter- feronalfa, thatsomepretreatmentvariablesmay independently predict response to this treat- ment. The criteriawehave used for the defini- tion of response were intended to assess the ability of interferon alfa to achieve a clear cut biochemical response, which is usually associ- ated with a transient or sustained clearance of serum HCV
RNA.'5
Spontaneous normalisa- tions ofALT activities have only been seen in about 1% of controls in randomised trials.6 Relapsesduringtreatment, which wereseenin 15 patients when receiving the lower doses of interferon alfa(1-5MU, threetimesperweek), and relapses occurring after interferon alfawas stoppedwere nottaken into account.The finding that most young patients (<45 years) without cirrhosis will respond to treat- ment (95%), whileonlyone third (36%)of the patients>45 years and withestablishedcirrhosis willbenefit fromtreatmentis clinically relevant.
This finding emphasises the need for early treatment in chronic hepatitis C. In addition, this might explain the discrepancies in the response rates among different clinical trials, which could be because of differences in the characteristics ofthe populations studied. The reason why yGT values influence response to interferon alfais not known. It could be specu- lated that a rise in yGT values might merely reflect an as yet undefined change in liver function, which mightsomehowimpairresponse to treatment. In fact some preliminary studies havesuggested thatursodeoxycolic acid,which decreasesyGT and alkalinephosphatase values in somecholestatic liver diseases,might enhance interferonalfa activityin chronichepatitis C.'9A preliminaryreporthas alsofound yGT valuesto be the best predictor of response tointerferon alfa.20 Unfortunately, most studies have not evaluated the influence of this variable.
Thereasonwhypatients withcirrhosis do not respond as well to interferon alfa as patients without cirrhosis is unknown. It can be specu- lated, however, that host and virological factors could both influenceresponse tointerferon alfa.
Cirrhosis has been associated with immune changes2'22 and it isusuallyseenafter many years of infection. Immune selected HCV mutants resistant to interferon alfa treatment might appear in cases with a long evolution of the disease.2324 It has been shown that HCV, like most RNA viruses, circulates as a population
1716
group.bmj.com on August 2, 2012 - Published by
gut.bmj.com Downloaded from
Prediction ofthe response ofchronichepatitis Ctointerferon alfa: a statistical analysisofpretreatment variables 1717
of closely related but none the less distinct genomes, a findingthat implies a high capacityto generate escape mutants.25 In addition, cirrhosis maychange thepharmacokinetics of some drugs and that could be thecase withinterferon.
The mechanism by which obesity may adversely influence response also remains unknown. One study found significant differ- ences in the mean weight between responders and non-responders.26 We found no differ- ences, however, when assessing the total body weight. Interferon alfa bioavailability could be diminished in obese patients because of a decreaseinthe absorption of the drug when the subcutaneous route is used, as in this study.
Other studies have shown that the skinfold thickness inversely correlates with the sub- cutaneousabsorption of some drugs.27 Pharma- cokinetic studies should solve this issue.
Meanwhile,wewouldfavour the intramuscular routein obese patients.
Neutralising anti-lymphoblastoid interferon alfa antibodies do notseemtoplay asignificant part in non-response to treatment, as they were detected inonly one of 15non-responders tested.
This lowfrequency of detection is in accordance witharecentstudy, which detected neutralising antibodies in one of 78 treated patients, and contrastswith thehigherfrequencies seen with recombinant interferon alfapreparations.28
We thank Dr Ignacio Monreal and Carmen de Carlos for performingthePIIIPRIA, and the Department of Pathology (Clinica Universitaria,Pamplona,Spain)forexaminingthe liver biopsyspecimens.WearealsogratefultoDrJuanIEsteban for helpfulcommentsonthemanuscript. This workwassupportedin partbythe Fundaci6n Ram6nAreces,Spain,andbyagrantfrom theComisi6n Interministerial deCienciayTecnologfa(SAL90- 0625),Spain.
1 DavisGL, Balart LA, SchiffER,LindsayK, Bodenheimer HC, Perrillo RP,etal.Treatmentof chronichepatitisC with recombinant interferon alfa. A multicenter randomized, controlled trial. NEnglJMed1989; 321: 1501-6.
2 DiBisceglie AM, Martin P, Kassianides C, Lisker-Melman M,Murray L, Waggoner J,etal. Recombinant interferon alfatherapy for chronichepatitis C. Arandomized,double- blind,placebo-controlledtrial. NEngl JMed1989;321:
1506-10.
3Davis GL. Recombinantalfa-interferontreatmentofnon-A, non-B(typeC)hepatitis:reviewof studies and recommenda- tionsfortreatment. Hepatol 1990;11:S72-7.
4Camps J, Castilla A, Ruiz J, Civeira MP, Prieto J. Random- ised trial of lymphoblastoid alfa interferon in chronic hepatitis C: Effects on inflammation, fibrogenesis and viremia.JHepatol 1993;17:390-6.
5 MarcellinP,BoyerN,GiostraE, DegottC,CourouceAM, Degos F, et al. Recombinant human alfa-interferon in patientswith chronicnon-A,non-Bhepatitis:amulticenter randomizedcontrolled trial from France.Hepatology1991;
13: 393-7.
6TineF, Magrin S, CraxiA, PagliaroL.Interferon fornon-A, non-B chronic hepatitis. Ameta-analysis of randomised clinicaltrials.JHepatol1991;13:192-9.
7 DavisGL,LindsayK,AlbrechtJ,BodenheimerHC,Balart LA, PerrilloRP,etal.Predictorsof responsetorecombinant alpha interferon in patients with chronic hepatitis C [Abstract].Hepatology1990;12:905.
8 Causse X, Godinot H, Chevallier M, Chossegros P, ZoulimF, Ouzan D, et al. Comparison of 1 or 3 MU of interferon alfa- 2b and placebo in patients with chronic non-A, non-B hepatitis. Gastroenterology1991;101:497-502.
9 Saracco G, Rosina F, Lattore V, Torrani-Cerenzia MR, Chiandussi L, Gallo V, et al. Factors influencing long-term responseof chronic non A, non B hepatitis to interferon treatment[Abstract].Hepatology 1990;12: 437.
10 Ideo G,Bellati G,Pedraglio E, Leandro G. One year therapy of non-A, non-B/C chronic hepatitis with recombinant alpha- 2ainterferon orlymphoblastoid alpha interferon [Abstract].
J Hepatol 1990; 11 (suppl 2): S31.
11 RealdiG, Diodati G, Bonetti P, Scaccabarozzi S, AlbertiA, Ruol A, etal. Recombinant human interferon alpha-2a in community-acquired non-A, non-B chronic hepatitis.
Preliminary results of a randomized, controlled trial.
J Hepatol1990;11(suppl 1): S68-71.
12DiBisceglie A,Hoofnagle JH. Therapy of chronic hepatitis C with a-interferon: the answer? or more questions?
Hepatology1991;13:601-3.
13Brook MG,KaryiannisP,Thomas HC. Which patients with chronic hepatitis B virus infection will respond to a- interferon therapy? A statistical analysis of predictive factors.Hepatology1989; 10:761-3.
14Weck PK,LeventhalBG, Brand C, Finter N. Detection and incidenceof neutralizing antibodies tointerferon-alpha-nI.
Interferon Res1989;9(suppl 1): S37-43.
15McCulloughAJ,Stassen WN, Wiesner RH, Czaja AJ. Serum typeIIIprocollagenpeptide concentrations in severe chronic active hepatitis:relationship to cirrhosis and disease activity.
Hepatology 1987; 7: 49-54.
16Jobson JD. Qualitative responseregression models and logistic regression. In: Jobson JD, ed.Applied multivariate data analysis.Vol 2. New York:SpringerVerlag, 1992: 278-328.
17Marcellin P, Giostra E, Boyer N, Loriot M-A, Martinot- Peignoux M, Benhamou JP. Is the response to recombinant alpha interferon relatedto the presenceof antibodies to hepatitis C virus in patients with chronic non-A, non-B hepatitis?JHepatol 1990; 11: 77-9.
18 Shindo M, Di Bisceglie AM, Cheung L, Shih W-K,Cristiano K, FeinstoneSM, et al. Decrease inserumhepatitis C viral RNAduringalpha-interferon therapy for chronic hepatitis C. Ann Intern Med1991;115: 700-4.
19 Angelico M, Gandin C, Goffredo F, Pescarmona E, Del VecchioC, CapocacciaL.A combinationofinterferon-alfa andursodeoxycolic acid ismoreeffective than interferon- alfa aloneinanti-HCVpositive chronic hepatitis:arandom- ized, histology-controlled, clinical trial. Gastroenterology 1992; 102:775A.
20Mazzella G, SalzettaA, BazzoliF,FestiD, Aldini R, Morelli C, et al. Efficacy and factors predictive of response to treatmentwithlymphoblastoidinterferon in chronicnon-A, non-Bhepatitis. Gastroenterology 1992; 102: 661A.
21 HassnerA, KletterY, Shiag D, Yedvab M,Aronson M, ShiboletS.Impairedmonocytefunction in liver cirrhosis.
BMJ1981; 282: 1262-3.
22Taylor KB, Thomas HC. Gastrointestinal and liver diseases.
In:Stites DP, Stobo JD,WellsJV,eds. Basicandclinical immunology. Appleton & Lange,1987:468-80.
23Houghton M, Weiner A, Han J, Kuo G, Choo Q-L.
Molecularbiology ofthehepatitis C viruses: implications for diagnosis, development and control of viral disease.
Hepatology1991; 14: 381-8.
24OkadaS-I,AkahaneY, Suzuki H, Okamoto H, Mishiro S.
The degreeof variability intheaminoterminalregionof the E2/NS1 protein of hepatitis C virus correlates with responsiveness to interferon therapy in viremic patients.
Hepatology 1992;16:619-24.
25 MartellM, EstebanJI, Quer J, Genesca J, Weiner A, Esteban R, etal.HepatitisC virus(HCV) circulatesas apopulation of different butclosely related genomes:quasispeciesnatureof HCV genomedistribution.JVirol1992;66:3225-9.
26 SchvarczR, Weiland0,WeistalR, NorkransG,Fryden A, FobergU. A randomized controlled openstudy ofinterferon alpha-2btreatmentof chronicnon-A,non-Bposttransfusion hepatitis:nocorrelationof outcometopresence ofhepatitis C virus antibodies.ScandlInfectDis1989;21:617-25.
27 KroonC, De BoerA,KroonJM, SchoenmakerHC,Meer FJM,CohenAF.Influence of skinfold thicknessonheparin absorption. Lancet 1991; 337: 945-6.
28 AntonelliG,CurrentiM,TurrizianiO,DianzaniF.Neutraliz- ing antibodiestointerferon-a: relativefrequency in patients treated withdifferent interferonpreparations.JInfectDis 1991;163:882-5.
group.bmj.com on August 2, 2012 - Published by
gut.bmj.com Downloaded from
