Research Article

FibroTest is an independent predictor of virologic response in

chronic hepatitis C patients retreated with pegylated interferon
alfa-2b and ribavirin in the EPIC3 programq
Thierry Poynard1,⇑, Mona Munteanu1, Massimo Colombo2, Jordi Bruix3, Eugene Schiff4, Ruben Terg5,
Steven Flamm6, Ricardo Moreno-Otero7, Flair Carrilho8, Warren Schmidt9, Thomas Berg10,
Thomas Mcgarrity11, E. Jenny Heathcote12, Fernando Gonçales13, Moises Diago14, Antonio Craxi15,
Marcelo Silva16, Navdeep Boparai17, Louis Griffel17, Margaret Burroughs17, Clifford Brass17,
Janice Albrecht17, 
1
APHP-UPMC Liver Center, Paris, France; 21st Division of Gastroenterology, Fondazione IRCCS Maggiore Hospital, University of Milan, Milan,
Italy; 3Centro de Investigación Biomédica en Red de Enfermedades Hepaticas y Digestivas, Liver Unit, Hospital Clinic, University of Barcelona,
Barcelona, Spain; 4University of Miami School of Medicine, Miami, FL, USA; 5Hospital Municipal de Gastroenterologia Dr. Bonorino Udaondo,
Capital Federal, Argentina; 6Northwestern University, Chicago, IL, USA; 7Hospital Universitario de la Princesa and Centro de Investigación
Biomédica en Red de Enfermedades Hepaticas y Digestivas (Instituto de Salud Carlos III), Madrid, Spain; 8Department of Gastroenterology,
University of São Paulo School of Medicine, São Paulo, Brazil; 9University of Iowa Hospitals and Clinics, Iowa City, IA, USA; 10Charité, Campus
Virchow Klinikum, Universitätsmedizin Berlin, Germany; 11Milton S. Hershey Medical Center, Hershey, PA, USA; 12University Health Network,
Toronto, ON, Canada; 13Department of Medical Clinical, Faculty of Medical Sciences, University of Campinas, Campinas, Brazil; 14Hospital
General Universitario de Valencia, Valencia, Spain; 15GI and Liver Unit, DIBIMIS, University of Palermo, Palermo, Italy; 16Hospital Universitario
Austral, Pilar, Argentina; 17Schering-Plough Research Institute, Kenilworth, NJ, USA

Background & Aims: EPIC-3 is a prospective, international study Methods: Of 2312 patients enrolled, 1459 had an available base-
that has demonstrated the efficacy of PEG-IFN alfa-2b plus line FT, biopsy, and complete data. Uni- (UV) and multi-variable
weight-based ribavirin in patients with chronic hepatitis C and (MV) analyses were performed using FT and biopsy.
significant fibrosis who previously failed any interferon–alfa/ Results: Baseline characteristics were similar as in the overall
ribavirin therapy. The aim of the present study was to assess population; METAVIR stage: 28% F2, 29% F3, and 43% F4, previous
FibroTest (FT), a validated non-invasive marker of fibrosis in relapsers 29%, previous PEG-IFN regimen 41%, high baseline viral
treatment-naive patients, as a possible alternative to biopsy as load (BVL) 64%. 506 patients (35%) had undetectable HCV-RNA at
the baseline predictor of subsequent early virologic (EVR) and TW12 (TW12neg), with 58% achieving SVR. The accuracy of FT
sustained virologic response (SVR) in previously treated patients. was similar to that in naive patients: AUROC curve for the diagno-
sis of F4 vs F2 = 0.80 (p <0.00001). Five baseline factors were asso-
ciated (p <0.001) with SVR in UV and MV analyses (odds ratio: UV/
MV): fibrosis stage estimated using FT (4.5/5.9) or biopsy (1.5/
Keywords: Hepatitis C; Non-responder; Relapser; Cirrhosis; Fibrosis; Biomarkers, 1.6), genotype 2/3 (4.5/5.1), BVL (1.5/1.3), prior relapse (1.6/1.6),
Treatment failure; Early virologic response. previous treatment with non-PEG-IFN (2.6/2.0). These same fac-
Received 29 March 2010; received in revised form 16 June 2010; accepted 22 June tors were associated (p 60.001) with EVR. Among patients
2010; available online 15 September 2010 TW12neg, two independent factors remained highly predictive
q
Presented in part at: the 55th (2004), 56th (2005), and 57th (2006) Annual
Meetings of the American Association for the Study of Liver Diseases; the 40th
of SVR by MV analysis (p 60.001): genotype 2/3 (odds ratio = 2.9),
(2005) and 43rd (2008) Annual Meetings of the European Association for the fibrosis estimated with FT (4.3) or by biopsy (1.5).
Study of the Liver; Digestive Diseases Week 2005; and HEP DART 2007: Frontiers Conclusions: FibroTest at baseline is a possible non-invasive
in Drug Development for Viral Hepatitis. alternative to biopsy for the prediction of EVR at 12 weeks and
⇑ Corresponding author. Address: Hôpital La Pitié Salpêtrière, Service d’Hépatol-
SVR, in patients with previous failures and advanced fibrosis,
ogie, 47–83 Boulevard de l’Hôpital, 75651 Paris Cedex 13, France. Tel.: +33 142 16
10 02; fax: +33 142 16 14 25. retreated with PEG-IFN alfa-2b and ribavirin.
E-mail address: [email protected] (T. Poynard). Ó 2010 European Association for the Study of the Liver. Published
 
The complete list of the EPIC3 Study Group membership appears in Appendix by Elsevier B.V. All rights reserved.
1. Study participants were told of investigators’ conflicts of interests.
Abbreviations: CI, confidence interval; COPILOT, Colchicine versus PegIntron Long-
Term; EPIC3, Evaluation of PegIntron in Control of Hepatitis C Cirrhosis; EVR, early Introduction
virologic response; FT, FibroTest; HALT-C, Hepatitis C Antiviral Long-term Trea-
tment against Cirrhosis; HCV, hepatitis C virus; IFN, interferon; LLD, lower limit of
detection; OR, odds ratio; PCR, polymerase chain reaction; PEG-IFN, peginterfer-
The assessment of fibrosis stage is useful in the treatment of
on; RT-PCR, reverse transcriptase PCR; SVR, sustained virologic response; WBD, patients with chronic hepatitis C (CHC), both for the decision to
weight-based dose. treat and in follow-up [1]. Because of the potentially untoward

Journal of Hepatology 2011 vol. 54 j 227–235

Research Article
complications of liver biopsy, several non-invasive methods have Assessments
been developed as possible alternatives [2].
FibroTest (FT), a set of non-invasive biomarkers of liver fibro- Serum samples and biochemical markers
FibroTest was measured at screening. Serum samples were collected and cen-
sis and activity, has been extensively validated in patients with
trally stored. Samples were blindly assessed without knowledge of any patient
CHC [2,3]; with similar diagnostic and prognostic value as a characteristics and according to recommended procedures [3,13–15]. FT com-
biopsy of 25-mm length [4,5], FT has been approved by the health bined the following five markers: alpha2-macroglobulin, haptoglobin, gamma
authority in France as a possible alternative for the initial assess- glutamyl transpeptidase (GGT), total bilirubin, and apolipoprotein A1. Apolipo-
ment of fibrosis and cirrhosis in patients with CHC[6,7]. protein A1, alpha2-macroglobulin, and haptoglobin were determined using
serum samples stored at 80 °C. An automatic nephelometer (Beckman Instru-
Several studies of FT during standard hepatitis C virus (HCV)
ments, Brea, CA, USA, or Dade-Behring, Deerfield, IL, USA) and Roche Diagnostics
treatment in treatment-naive patients have been performed. reagents (Roche Diagnostics, Indianapolis, IN, USA), Siemens Healthcare Diagnos-
These studies demonstrated that FT is similar to paired liver biop- tics reagents (Siemens Healthcare Diagnostics, Deerfield, IL, USA), or Beckman
sies in demonstrating reduced necrosis and fibrosis in sustained Instruments reagents (Beckman Instruments, Brea, CA, USA) were used. The coef-
ficient of variation of all assays was lower than 3%. GGT, alanine aminotransferase
virologic responders [5,8–11]. However, no specific study of FT
(ALT), and total bilirubin were assessed prospectively during the trial period,
has been performed in previous non-responders to interferon/ using Hitachi 747 or 911 automates or Roche Modular Analysers.
ribavirin treatment. ActiTest (AT) combined the same five markers as FT plus ALT. It has a high
The aim of this analysis was to validate the utility of FT as a predictive value for the diagnosis of significant activity features [3,14]. SteatoTest
possible alternative to biopsy for staging fibrosis in non-respond- combines the same 6 markers as AT plus aspartate aminotransferase (AST), serum
triglycerides, cholesterol, fasting glucose and body mass index. It has a high pre-
ers using the first phase of the EPIC-3 trial (Evaluation of PegIn-
dictive value for the diagnosis of significant steatosis [16].
tron in Control of Hepatitis C Cirrhosis) – a large, prospective,
multiphase clinical program – evaluating the retreatment of
Virologic markers
patients with CHC with significant fibrosis/cirrhosis in whom Plasma HCV-RNA was measured at screening, at weeks 12, 24, and 48 of treat-
previous treatment with non-pegylated or peginterferon (PEG- ment and at 12 and 24 weeks post-treatment. HCV-RNA analyses (TaqManÒ;
IFN) alfa plus ribavirin was ineffective (i.e., virologic non- Applied Biosystems, Foster City, CA, USA; lower limit of detection [LLD] for 95%
sensitivity of 125 IU/ml) were performed centrally at the Schering-Plough
responders or relapsers) [12].
Research Institute (SPRI) Laboratory (Kenilworth, NJ, USA). Samples below the
LLD for which a signal was detected were characterized as low positive or detect-
able; those for which no signal was detected were characterized as negative or
Patients and methods undetectable. Confirmatory testing using TaqMan (Quest Nichols Laboratory,
San Juan Capistrano, CA, USA) was performed on a subset of samples tested at
Patient selection the SPRI Laboratory. Early virologic response (EVR) was defined as HCV-RNA
below the LLD (125 IU/ml) at TW12. The primary efficacy end point of the thera-
peutic study was SVR, defined as undetectable serum HCV-RNA 24 weeks post-
Patients enrolled in EPIC3 were included in the present FT evaluation if they had treatment. A secondary efficacy end point was the difference in SVR rates by
12 weeks of virology (TW12) results available, interpretable baseline FT and liver fibrosis score.
biopsy. Details of inclusion criteria as well as the results of the first study of the pro-
gram have been published elsewhere. Twelve patients 18–65 years of age with CHC
and significant hepatic fibrosis/cirrhosis (METAVIR score F2, F3, or F4), who failed Histological criteria
combination therapy with non-pegylated or PEG-IFN alfa/ribavirin therapy, were Pretreatment liver biopsy specimens were scored by a single pathologist using
eligible to enroll. All patients in this study had previously received a minimum of METAVIR criteria [17]. The pathologist was blinded to historical biopsy reports
12 weeks of combination therapy and did not achieve sustained virologic response and other clinical data. All the patients were biopsied after their prior failed ther-
(SVR). Patients were categorized according to the previous response (non-respon- apeutic regimen.
der, relapser, or treatment failure) to combination therapy based on documented
HCV-RNA polymerase chain reaction (PCR) results. Non-responders had detectable Statistical analysis
HCV-RNA at the end of therapy. Relapsers had undetectable HCV-RNA at the end of
treatment (EOT) and had subsequent detectable HCV-RNA during post-treatment
Calculating FibroTest diagnostic values
follow-up. Patients, who did not meet the protocol definition for non-responder
The diagnostic values of FT were assessed by the receiver-operating characteris-
or relapser because documentation of HCV-RNA assays did not fulfill these defini-
tics (ROC) curves, which plot sensitivity versus 1 – specificity [14,18–19]. The
tions but had detectable HCV-RNA more than 1 week after the end of their previous
respective overall diagnostic values were compared using the area under the
therapeutic regimen, were designated as treatment failures. These patients likely
ROC curves (AUROCs). The AUROCs were compared to 0.50 (no diagnostic value)
represented a mixture of relapsers and non-responders. Additional major inclusion
and to AUROCs observed in an integrated database of patients with CHC who
criteria were HCV-RNA positivity, hepatic fibrosis documented by historical biopsy
were never treated (naive population) [14].
showing at least portal fibrosis with few septa (METAVIR score F2), compensated
Estimates of AUROCs and comparisons between AUROCs used an empirical
liver disease (Child-Pugh class A), hemoglobin P12 g/dl for women and P13 g/
nonparametric method [18,19]. Two factors are strongly associated with the AUR-
dl for men, absolute neutrophil count P1500/mm3, platelet count P80,000/mm3
OCs of fibrosis biomarkers and biopsy: the prevalence of the different fibrosis
and body weight of 40–125 kg. Major exclusion criteria included known coinfec-
stages that define advanced and non-advanced fibrosis [20,21] and the length
tion with HIV or hepatitis B virus, decompensated liver disease and history of or
of biopsy [21,22].
current hepatocellular carcinoma.
To address this risk of spectrum bias due to different prevalences of fibrosis
stages between studies we used two previously validated method of standardiza-
Study design and conduct tion [20–22]. For the same test, if only F3 and F4 patients are included, AUROCs
are mathematically lower than if only F2 and F4 patients are included [20,23].
The retreatment trial of the EPIC-3 program is a prospective, open-label, clinical Using an equal proportion of each fibrosis stage permitted a standardized AUROC
trial and was conducted at 133 sites in North America, Europe, Latin America, Tai- expression. In this standard prevalence distribution, the difference between the
wan and Australia. Patients received PEG-IFN alfa-2b 1.5 lg/kg/wk and daily mean fibrosis stage of advanced fibrosis minus the mean fibrosis stage of non-
weight-based dose (WBD) ribavirin (800 mg for 665 kg; 1000 mg for >65– advanced fibrosis (DANA) is 2.5. The first step of such a standardization is to
85 kg; 1200 mg for >85–105 kg; and 1400 mg for >105–125 kg) for up to estimate the relationship between DANA and observed AUROCs (ObAUROC).
48 weeks. Patients with detectable HCV-RNA at treatment week (TW) 12 were For each population, DANA is calculated as: mean advanced fibrosis estimated
offered randomization into maintenance studies. The study was approved by by {[(prevalence F2 2) + (prevalence F3 3) + (prevalence F4 4)]/(prevalence
the ethics committee at each participating institution and was conducted accord- F2 + prevalence F3 + prevalence F4)} minus mean non-advanced fibrosis esti-
ing to good clinical practice and the Declaration of Helsinki. All patients provided mated by [prevalence F1/(prevalence F0 + prevalence F1)]. From the regression
written informed consent [12]. formula linking the ObAUROC to DANA, one can calculate an AUROC standardized

228 Journal of Hepatology 2011 vol. 54 j 227–235

JOURNAL OF HEPATOLOGY
(StAUROC) at the DANA value of 2.5. Contrary to the ObAUROC, the StAUROC esti- to be included in the present study. Four patients were excluded
mate for a given test for the diagnosis of advanced fibrosis is independent of the due to un-interpretable FT results (Fig. 1).
prevalence of advanced (F4) and non-advanced fibrosis (F2/F3) stages. We used
the previously validated formula in patients with CHC [StAUROC = ObAUROC +
Baseline characteristics of the diagnostic population were
(0.1056) (2.5 ObDANA)]. The corresponding StAUROC for F2 versus F4 similar to the safety population (Table 1): 70% male; median
(ObDANA = 2) was ObAUROC + (0.1056)(0.5) = ObAUROC + 0.0528. The corre- age 51 years; 28% with few fibrous septa (METAVIR F2), 29% with
sponding StAUROC for F2 versus F3 (ObDANA = 1) was ObAUROC + many septa (F3) and 43% with cirrhosis (F4) at biopsy; previous
(0.1056)(1.5) = ObAUROC + 0.1584.
relapses 29%; previous PEG-IFN regimen 41%; genotype 1 in
To prevent the risk of considering a short biopsy sample as the gold stan-
dard, we used the previously validated adjusted FT AUROC as the gold standard 82%; and high (>6 105 IU/ml) baseline viral load (BVL) in 64%.
for the given biopsy length (AlAUROC) [21] For instance a biopsy of 16-mm The median time difference between FT sampling and historical
length for the diagnosis of F3 versus F4 has an AUROC of 0.82 versus the gold biopsy was 161 days (95% CI 150–168).
standard (entire liver) (GsAUROC). The FibroTest AlAUROC = ObAUROC/GsAU-
ROC. An ObAUROC for FT of 0.71, using biopsy of 16 mm, has an
A1AUROC = 0.71/0.82 = 0.87.22. Diagnostic value of FT

Obuchowski measure The accuracy of FT for the diagnosis of fibrosis in the present pop-
Lambert et al. proposed in order to overcome both spectrum effect and ordinal ulation of non-responder patients was similar to that of previous
scale, to use the Obuchowski measure [21]. Furthermore this measure allows to validations in treatment-naive patients for all the observed and
compare two biomarkers with a single test, avoiding appropriate correction for
adjusted AUROCs, according to prevalence of stages or biopsy
the type I error when comparing two biomarkers for different stages or grades.
This measure is a multinomial version of the AUROC. With N categories of the length (Table 2) [3,23]. The unique test estimating all pairwise
gold standard outcome (histological fibrosis stage) and AUROCst, the estimate performances using Obuchowski measure was significant versus
of the AUROC of diagnostic tests for differentiating between categories s and t, random performance (Table 2). The mean biopsy length was sim-
the Obuchowski measure, is a weighted average of the N(N 1)/2 different ilar in the present study to that in the previously published inte-
AUROCst corresponding to all the pairwise comparisons between two of the N
categories. Each pairwise comparison has been weighted to take into account
grated database (13 mm vs 16 mm) [14]. The observed AUROCs
the distance between fibrosis stages (i.e., the number of units on the ordinal were 0.75 for the diagnosis between F2 and F4, 0.65 between
scale). A penalty function proportional to the difference in METAVIR units adjacent stages F2 and F3 and 0.62 between F3 and F4. These
between Stages was defined: the penalty function was 0.25 when the difference
between stages was 1, 0.50 when the difference was 2, and 1 when the difference
was 3. The Obuchowski measure can be interpreted as the probability that the
non-invasive index will correctly rank 2 randomly chosen patient samples from Patients screened
different fibrosis stages according to the weighting scheme, with a penalty for n = 2333
misclassifying patients [21].

Prognostic value of FibroTest Patients included in

Uni- (UV) and multivariable (MV) analyses for SVR were performed using a com- safety study
plete model (genotype, viral load, age, sex and previous PEG-IFN) [12] that also n = 2312
included FT and biopsy. Logistic regression analyses, with SVR as the response
variable and key baseline and demographic variables as explanatory variables,
were performed to assess the effect of the prognostic factors on SVR rates. Step- Patients included in
wise regression methods were used to build prediction models. In the modeling diagnostic study
FT was entered as a continuous variable and biopsy as METAVIR stages. SVR rates Baseline FibrotestTM and biopsy
were also summarized by baseline METAVIR fibrosis score (F2, F3 and F4), esti- n = 1459
mated both by FT and biopsy. The Armitage S trend test was used to compare
the proportion of patients who attained SVR among these patients. Number
Cruncher Statistical Software was used [24].
HCV-RNA 12 weeks
Diagnostic and prognostic value of ActiTest and SteatoTest
Both the degree of necroinflammation and steatosis were also evaluated using
validated biomarkers, ActiTest and SteatoTest, previously validated in naive
Undetectable Detectable
patients [9,10,16].
The predetermined following cutoffs of FT values were used for the corre- n = 527 n = 932
sponding METAVIR stages: 60.27: F0; 60.48: F1; 60.58: F2; 60.74: F3;and
>0.74: F4. No No
Significant activity was defined as presumed grade A2/A3 (METAVIR scoring end-of-follow up end-of-follow up
system: moderate/severe necroinflammatory activity) according to the 0.52 lab- HCV-RNA PCR HCV-RNA PCR
oratory predetermined cutoff [3]. The predetermined following cutoffs of AT val-
n = 21 n = 760
ues were used for the corresponding METAVIR grades: 60.29: A0; 60.52: F1;
60.62: F2;and >0.62: A3.
Significant steatosis was defined as presumed stage S2/S3/S4 (steatosis
between 5% and 100%) according to the laboratory predetermined cutoff [16].
Treated 48 weeks

Results
Undetectable RNA <2-log RNA drop ≥2-log RNA drop
Patient characteristics n = 506 n = 43a n = 108
SVR 58% SVR 0%b SVR 13%b
In all, 2333 patients were screened, 2312 patients were enrolled
(treatment and safety population) and 1459 patients (diagnostic Fig. 1. Flow diagram of patients included in the therapeutic study and in the
population) had available baseline FT, biopsy and complete data diagnosis study.

Journal of Hepatology 2011 vol. 54 j 227–235 229

Research Article
Table 1. Baseline Patient Characteristics of the ‘‘Safety’’ and ‘‘Diagnostic able serum HCV-RNA at TW12 (TW12neg); 678 (46%) were trea-
Population’’. ted for 48 weeks with 24 weeks’ follow-up. In the 1459 patients
Safety Diagnostic enrolled (intention-to-treat population), the rate of SVR was
populationa population 21% (312/1459). The rate of SVR was 58% among the 506 patients
n = 2312 n = 1459 who were TW12neg, 13% among 108 patients with detectable but
P2-log drop in viral load (all of whom had an HCV-RNA of
Male, n (%) 1650 (71) 1015 (70)
<750 IU/ml and the majority had <125 IU/ml) and 0% among 43
Caucasian, n (%) 1932 (84)
patients with less than 2-log drop (Fig. 1).
Mean age, y (SD) 49.2 (12.4) 50.9 (8.3)
Fibrosis stage (METAVIR) Prognostic value of baseline FT
Few septa (F2) 658 (29) 410 (28)
Many septa (F3) 676 (29) 424 (29) Baseline fibrosis stage estimated using FT had the same prognos-
Cirrhosis (F4) 974 (42) 625 (43) tic value as that estimated using biopsy for SVR and for EVR.
As observed in the therapeutic population, five baseline fac-
Necroinflammatory
tors were significantly associated with SVR (Table 3A) and EVR
activity (METAVIR)
(Table 3B) in UV and MV analyses: fibrosis stage estimated using
No activity (A0) 157 (7) 92 (6) FT or biopsy, genotype 2/3, BVL, prior relapse and previous treat-
Minimal activity (A1) 1742 (75) 1107 (76) ment with non-PEG-IFN.12 Among patients who were TW12neg
Moderate activity (A2) 389 (17) 245 (17) (n = 506), only three factors remained highly predictive of SVR
Severe activity (A3) 22 (1) 15 (1) by MV analysis: fibrosis estimated with FT or by liver biopsy,
genotype 2/3 and BVL (Table 3C).
Mean weight, kg 81.1 80.5
A graded decrease in SVR was observed among all patients
Genotype, n (%)
(n = 1459), as well as the subset with EVR (n = 506), as the META-
1 1859 (80) 1203 (82) VIR fibrosis score increased by FT or biopsy analysis. SVR
2 75 (3) 47 (3) decreased in the diagnostic population from 40% to 15% (Armit-
3 294 (13) 168 (12) age test S for trend = 78,554; p <0.00001) and from 75% to 52%
4 68 (3) 37 (3) (S = 8391; p = 0.004) in the 12-week responder subset according
to FT score for fibrosis and from 27% to 16% (S = 56,637;
Missing 17 (1) 4 (<1)
p = 0.00002) and from 63% to 48% (S = 9224; p = 0.001) for liver
Viral load
biopsy scoring of fibrosis, respectively (Fig. 2). This trend was also
600,000 IU/ml 853 (37) 524 (36) consistent across patients classified as F0 and F1 with FT (Fig. 2).
>600,000 IU/ml 1451 (63) 935 (64) When discordance between baseline fibrosis estimates was
Missing 8 (<1) 0 (0) entered as a covariate in the model, FT retained a significant pre-
dictive value for EVR (odds ratio [OR] = 0.24; p = 0.009 without
Previous combination
significance for discordant cases OR = 1.02, p = 0.93) and SVR
therapy, n (%)
(OR = 0.13; p = 0.003); however, biopsy scoring had a weaker pre-
IFN alfa + ribavirin 1425 (62) 857 (59) dictive value for EVR (OR = 0.83; p = 0.03 with significant predic-
PEG-IFN alfa + ribavirin 865 (37) 602 (41) tive value for discordant cases OR = 1.67; p = 0.003); a similar
No combination therapy 22 (1) 0 (0) trend was observed for SVR and biopsy (OR = 0.67; p = 0.004 with
Previous response, n (%) borderline significance for discordant cases OR = 1.41; p = 0.10).
Nonresponse 1401 (61) 880 (60)
Relapse 647 (28) 424 (29) Discordant patients according to FT and biopsy fibrosis estimates
b
Treatment failure 264 (11) 155 (11) A total of 292 patients (94 presumed F0 and 198 F1 with FT) were
a
All patients enrolled in the EPIC3 trial [12]. suspected to be either false negative of FT or false positive of
b
There was no significant difference for the characteristics of patients between the
biopsy (144 F2, 80 F3, and 68 F4 with biopsy).
safety population and the diagnostic population.
Comparison between biopsy length, inflammatory scores, and
ObAUROCs were equivalent to AUROCs of 0.78–0.81 after stan- ALT levels for these 292 discordant patients versus the 1167 con-
dardization on DANA and to AUROCs of 0.72–0.82 after standard- cordant patients observed: lower METAVIR inflammatory score at
ization to biopsy length (Table 2). The AUROC for stage F4 vs F2F3 biopsy:1.03 (95%CI 0.99–1.09) vs 1.15 (1.12–1.18) (p = 0.001);
was 0.69 (95% CI 0.66–0.71), higher than random value lower median ALT: 58 UI/L (52–63) vs 94 (91–97); there was
(p <0.0001). no significant difference between biopsy length: median 13 mm
(12–14) vs 12 mm (12–12) respectively (p = 0.18).
Virologic response to treatment
Diagnostic and prognostic value of baseline biomarkers of activity
Results were comparable to those observed in the overall popu- and steatosis
lation [12]. One thousand four hundred and fifty nine patients
received PEG-IFN alfa-2b 1.5 lg/kg/wk plus WBD ribavirin The accuracy (AUROCs) of ActiTest for the diagnosis of activity
(800–1400 mg/days) for 12–18 weeks; 506 (35%) had undetect- grades was significant (p <0.001), but consistently lower in the

230 Journal of Hepatology 2011 vol. 54 j 227–235

JOURNAL OF HEPATOLOGY
Table 2. Diagnostic Value of FibroTest (AUROCs and Obuchowski measures) for the Diagnosis Between Each Fibrosis Stage Observed in the Present Study (Non-
responders) and in Previous Integrated Data (Naive Patients).

FibroTest vs. Biopsy

Comparison Nonresponders Controls, Naivee

F4 vs F2
Number of patients 1035 376
a
Observed AUROC 0.75 (0.72-0.78) 0.75 (0.70-0.80)
b
Standardized AUROC 0.80 (0.77-0.83) 0.80 (0.75-0.85)
Adjusted on biopsy length AUROCc 0.82 (0.79-0.85) 0.82 (0.77-0.87)
Weighted Obuchowski measured 0.71 (0.68-0.74) 0.84 (0.80-0.86)
F3 vs F2
Number of patients 834 364
Observed AUROCa 0.65 (0.61-0.68) 0.63 (0.58-0.68)
b
Standardized AUROC 0.81 (0.77-0.85) 0.79 (0.74-0.84)
Adjusted on biopsy length AUROCc 0.76 (0.72-0.80) 0.73 (0.68-0.78)
Weighted Obuchowski measure 0.59 (0.56-0.62) 0.63 (0.61-0.65)
F4 vs F3
Number of patients 1049 234
Observed AUROCa 0.62 (0.59-0.66) 0.65 (0.59-0.71)
Standardized AUROCb 0.78 (0.75-0.81) 0.81
c
Adjusted on biopsy length AUROC 0.72 (0.69-0.75) 0.76
Weighted Obuchowski measure 0.58 (0.55-0.61) 0.63 (0.61-0.65)
All pairwise comparisons (Obuchowski measure) 0.63 (0.61-0.65) 0.70 (0.68-0.72)
All AUROCs were significant (p <0.001) versus random AUROC (0.50). There was no significant difference between the AUROCs of non-responders and naive patients. There
were lower weighted Obuchowski measures for non-responders compared to controls.
a
Observed area under the receiver-operating characteristics curves (AUROCs) and 95% confidence interval.
b
Standardized AUROC preventing spectrum bias for eventual comparisons between studies, corresponding to a standard difference of fibrosis stage of 2.5 METAVIR unit
[20].
c
Gold standard–adjusted AUROC for the given biopsy length preventing the bias of considering biopsy as a gold standard calculated as the ratio between ObAUROC and
GsAUROC of 16-mm biopsy versus gold standard (0.82 for F3 vs. F4 and 0.86 for F2 vs. F3) [21].
d
Obuchowski measure takes into account the spectrum effect (weighted according to difference between stages) and the overall measure takes into account the multiple
testing.
e
Integrated database of naive patients with chronic hepatitis C [14].

present study than in the integrated database of naive patients 2 months apart, the AUROC was 0.77 (95% CI, 0.64–0.85) versus
(Supplementary file 1) [14]. The Spearman correlation coefficient 0.68 (95% CI, 0.65–0.71) among patients tested 2 months or
between ActiTest and activity grade at biopsy was 0.17 more apart (p = 0.12).
(p <0.0001), with an obvious spectrum bias among patients with We observed a significantly lower SVR in patients with steato-
low activity scores (A0/A1) and high activity scores (A2/A3), as sis using liver biopsy estimate or SteatoTest. The SVR was lower
76% of patients had A1 and 17% had A2. Among 76 patients in those with high steatosis (>10%; S2) graded by liver biopsy
who had a biopsy and ActiTest within 2 months of each other, (18.8% SVR; 106/564) compared to those with steatosis 610%
the AUROC of A2/A3 versus A0/A1 was 0.71 (95% confidence (23.9% SVR; 203/850; p = 0.02). Those with more hepatic steatosis
interval [CI], 0.54–0.83) versus 0.60 (95% CI, 0.56–0.64) for those by SteatoTest (>0.57) had an SVR of 18.1% (95/525) compared to
patients with two or more months between their liver biopsy and SteatoTest <0.57 (24.1% SVR; 214/889; p = 0.009).
ActiTest (p = 0.16). There was no prognostic value of METAVIR
activity grades either estimated using ActiTest or biopsy (data
not shown). Discussion
The accuracy (AUROC) of SteatoTest (n = 1415) for the diag-
nosis of S2/S3/S4 (steatosis between 5% to 100%; prevalence The results of this study illustrate both the diagnostic and prog-
40%) versus S0/S1 (steatosis less than 5%; prevalence 60%) nostic utility of liver injury biomarkers (FT, ActiTest and Steato-
was 0.68 (95% CI, 0.65–0.71) (p <0.0001 versus AUC of 0.50). Test) for clinicians and suggest that they could serve as a
This was lower than AUROCs observed in 171 HCV naive possible alternative to liver biopsy in patients with CHC for
patients, performed 40 days apart: 0.80 (95% CI, 0.74–0.86; whom previous combination therapy failed. The diagnostic value
p = 0.02). The Spearman correlation coefficient between Steato- of FT [2,3], Actitest, [14], and SteatoTest [16] has already been
Test and steatosis grade at biopsy was 0.32 (p <0.0001). Among validated in patients naive to HCV treatment. In this study, FT
76 patients who had a biopsy and SteatoTest less than had the same prognostic value as liver biopsy for predicting the

Journal of Hepatology 2011 vol. 54 j 227–235 231

Research Article
Table 3A. Prognostic value of FibroTest versus biopsy for early virologic response.

Factor Univariate Odds Ratio Multivariate Odds Ratio

n = 1459 Significance n = 1304 Significance
Fibrosis stage
FibroTest 4.9 (2.9-8.1) <0.0001 4.2 (2.2-7.9) <0.0001
Biopsy 1.2 (1.1-1.4) 0.001 1.3 (1.1-1.5) 0.001
Genotype 2/3 11.5 (7.9-16.6) <0.0001 8.9 (5.8-13.6) <0.0001
Baseline viral load <log6 1.9 (1.5-2.4) <0.0001 1.5 (1.2-2.0) 0.003
Prior relapse (77 missing) 6.2 (4.8-8.0) <0.0001 6.5 (4.9-8.8) <0.0001
Previous non-PEG-IFN 1.3 (1.05-1.6) 0.02 2.0 (1.5-2.7) <0.0001

likelihood of a SVR to treatment for hepatitis C as has been pre- tory activity grades is also possible with ActiTest9 [14], as are ste-
viously observed in naive patients [8–10]. atosis grades using SteatoTest, both for viral or nonviral steatosis
[16]. Other biomarkers (HFE gene, transferrin saturation, mag-
Limitations netic resonance imaging) could also provide non-invasive possi-
ble alternatives to biopsy for the diagnosis of hemochromatosis
The main limitations of this study were the non-simultaneous [27].
measurement of the biomarkers and biopsy, the FT was not One limitation in the validation portion of this study was the
assessed in all included patients, and the length of biopsy sam- relatively long mean duration between biopsy and serum sam-
ples was suboptimal. pling (161 days). This limitation relates to the use of historical
There were no differences in demographic characteristics, biopsies for studies, which minimized the invasiveness of this
baseline clinical parameters and virologic responses between trial. This delay between FT and biopsy could be a factor that
the patients who had the measurements required to accurately may explain the lower accuracy observed for AT for the diagnosis
assess FT and thus be included in the present study, and of those of activity grade and for SteatoTest for the diagnosis of steatosis
patients recruited to the EPIC3 retreatment study (Table 1). grade in comparison with the accuracy observed in studies of
The variability of FT and its components have been exten- naive patients in which the median interval between biopsy
sively investigated. The assays for this study were centralized and FT was only 40 days [16]. These features of activity and ste-
in two CLIA laboratories (LabCorp, Raritan, NJ, USA and Covance, atosis are less stable than the fibrosis stage, and this hypothesis is
Indianapolis, IN, USA) following the recommended pre-analytic supported by the increase of ActiTest and SteatoTest AUROCs in
and analytic procedures [13–15]. Only four (3/1000) patients subpopulations with shorter intervals between biopsy and sero-
were excluded because of a high-risk profile of false positive or logic testing. Despite these limitations (also present for biopsy),
false negative. The usual main confounders were observed: this was a unique opportunity to validate these biomarkers as
hemolysis of the sample, acute inflammation and Gilbert syn- prognostic indicators of subsequent viral clearance in a large pop-
drome [3,7,13,14]. ulation of prior non-responders to antiviral therapy. They are
One disadvantage of employing FT alone versus biopsy to fewer validations of SteatoTest in patients with chronic hepatitis
evaluate hepatic fibrosis would be the possible inclusion of C than for FT [3–11] and ActiTest, [3,9,10,28] and the SteatoTest
patients with additional causes of liver disease (e.g., due to alco- performance must be confirmed by other studies.
hol, nonalcoholic steatohepatitis, or hemochromatosis). However, Another limitation of this study is the relatively short length
this risk is reduced because FT has the same diagnostic value in of biopsy sample (16 mm) in comparison with the recommended
the most common causes of liver diseases [3] and the same prog- length of 25 mm [29]. But as most studies of liver biopsy per-
nostic value as biopsy in patients with chronic hepatitis B [25] formed in large populations fail to achieve this optimal length
and alcoholic liver disease [26]. The diagnosis of necroinflamma- [22], it may be argued that using more reproducible serologic

Table 3B. Prognostic value of FibroTest versus biopsy for sustained virologic response.

Factor Univariate Odds Ratio Multivariate Odds Ratio

n = 678 Significance n = 601a Significance
Fibrosis stage
FibroTest 4.5 (2.2-9.0) <0.0001 5.0 (2.3-11.0) <0.0001
Biopsy 1.5 (1.2-1.8) <0.0001 1.6 (1.3-2.0) <0.0001
Genotype 2/3 4.5 (3.1-6.6) <0.0001 4.0 (2.6-6.0) <0.0001
Baseline viral load <log6 1.7 (1.2-2.3) 0.0009 1.6 (1.2-2.3) 0.005
Prior relapse 1.6 (1.1-2.2) 0.007 1.7 (1.1-2.4) 0.007
Previous non-PEG-IFN 1.4 (0.99-1.9) 0.06 1.6 (1.1-2.3) 0.02
a
For 77 patients prior type of response (relapse or non-responder) was missing.

232 Journal of Hepatology 2011 vol. 54 j 227–235

JOURNAL OF HEPATOLOGY
Table 3C. Prognostic value of FibroTest versus biopsy for sustained virologic response among patients with early virologic response.

Factor Univariate Odds Ratio Multivariate Odds Ratio

n = 506 Significance n = 444a Significance
Fibrosis stage
FibroTest 3.5 (1.5-8.0) 0.003 4.0 (1.6-10.0) 0.003
Biopsy 1.4 (1.1-1.7) 0.003 1.5 (1.2-1.9) 0.0007
Genotype 2/3 3.0 (2.0-4.6) <0.0001 2.9 (1.8-4.5) <0.0001
Baseline viral load <log6 1.7 (1.2-2.4) 0.004 1.6 (1.08-2.4) 0.02
Prior relapse 0.99 (0.7-1.4) 0.96 1.06 (0.7-1.6) 0.80
Previous non-PEG-IFN 1.3 (0.93-1.9) 0.11 1.3 (0.8-2.0) 0.23
a
For 62 patients prior type of response (relapse or non-responder) was missing.

estimates of liver injury achieve the same or better results with- values of biomarkers for EVR as well as the prognostic value of
out the attendant risks of a liver biopsy. FT in patients with undetectable HCV-RNA at 12 weeks of
therapy.
Advantages of serologic markers One advantage of biomarkers compared to biopsy assessment
is that it enables to both evaluate liver disease severity and to
The main advantages of this study are the large population of anticipate the treatment outcome in patients with contraindica-
non-responder patients, the multicenter, multinational and pro- tions to biopsy or those who refuse it. Additionally biomarkers
spective nature of the population, and the centralized assessment can be safely used to provide long-term follow-up of liver disease
of histology. The analyses of this study were made with an inde- severity without the constraint of repetition of an invasive proce-
pendent and blinded assessment of biomarkers, fibrosis stage, dure. Patients are spared the risks associated with liver biopsy.
activity and steatosis grades. During the trial, all sera were pro- An important advantage already demonstrated in naive
spectively stored. In this study we were able to demonstrate patients [5,9,10] is that FT is at least as accurate as biopsy for pre-
the reproducibility of FT’s accuracy for the diagnosis of fibrosis dicting virologic response, either SVR or SVR among patients with
stage, the reproducibility of baseline FT for the prediction of EVR. As with treatment-naive patients, baseline METAVIR fibrosis
SVR and for the first time the demonstration of the prognostic score estimated using biopsy or FT was the second strongest

F0
100 F1
S = 8391 F2
p = 0.004 S = 9224 F4
p = 0.001 F3
75
75
63 65
S = 78554 61 60
57
SVR (%)

p <0.00001 52
48
50 S = 56637
40 p = 0.00002

29 27
27
24
25 20
15 16

0
All patients Undetectable All patients Undetectable
(N = 1459) HCV-RNA at 12 weeks (N = 1459) HCV-RNA at 12 weeks
(n = 506) (n = 506)

FibrotestTM Biopsy

Fig. 2. Sustained virologic response (SVR) rate according to baseline fibrosis stage estimated using either FibroTest or biopsy in overall diagnostic population and in
patients with undetectable HCV-RNA at 12 weeks (EVR). When estimated using biopsy, patients with baseline stages F0 (no fibrosis) and F1 (portal fibrosis) were
excluded from the overall study. When estimated using FibroTest the baseline fibrosis stage was F0 and F1 in 94 and 198 patients, respectively. These discordant cases could
be either false negatives of the FibroTest or false positives of the biopsy. In the absence of a true gold standard, the fact that the virologic responses (either SVR or SVR
among EVR) of patients classified as F0 by FT were significantly higher than those classified F2 by biopsy (40% vs 27% for SVR, 75% vs 63% for SVR among EVR) strongly
suggest that they could be biopsy false positives.

Journal of Hepatology 2011 vol. 54 j 227–235 233

Research Article
predictor, after genotype and before BVL, in the present study. from Valeant and serves on the Speakers Bureau for Novartis; M.
These data establish TW12 as a simple and effective point at Colombo, E. Schiff, A. Craxi, J. Bruix, and E.J. Heathcote are consul-
which to decide on retreatment with PEG-IFN alfa-2b/ribavirin tants for Schering-Plough; A. Craxi and E.J. Heathcote are consul-
of patients who previously failed interferon alpha/ribavirin treat- tants for Roche; L. Griffel, M. Burroughs, C. Brass, J.K. Albrecht,
ment. After considering patient and disease factors, including and N. Boparai are employees of Schering-Plough Research Insti-
fibrosis stage using FT, patients with HCV-RNA below the LLD tute and stockholders of Schering-Plough; R. Moreno-Otero, F.
at TW12 – those with negative or near negative results – should Carrilho, R. Terg, and T. McGarrity have nothing to disclose. Thi-
continue therapy (>50% chance of SVR), whereas others can be erry Poynard is the inventor of FibroTest–ActiTest–SteatoTest
spared further drug exposure because the likelihood of attaining with a capital interest in Biopredictive, the company marketing
an SVR is low. Alternatively, these patients may consider long- the tests. The patents belong to ‘‘Assistance Publique Hôpitaux
term low-dose maintenance therapy [12,30–33]. de Paris’’ a public organization. Mona Munteanu is a full-time
Although other studies [34,35] have shown that significant employee of Biopredictive.
fibrosis and cirrhosis are negative predictors of SVR, EPIC3 is
the first large study to clearly demonstrate a graded decrease in
Funding source(s)
SVR rates as fibrosis score progresses from F2 to F3 to F4 both
using biopsy and FT [12]. Furthermore, as suggested in previous
Schering-Plough Research Institute fully supported the study
studies [32], it seems possible that FT could be even better than
except for the FibroTest algorithm supported by Biopredictive.
short-length biopsy for the prediction of SVR [4,5,9]. In the pres-
ent study 302/1459 (20%) patients were classified at baseline as
F0 (n = 94) or F1 (n = 198) by FT and these same patients were Role of study sponsor
classified by biopsy as F2 or F3 or F4 by biopsy. As already dem-
onstrated these discordant cases could be either false negatives of Schering-Plough Research Institute played a major role in the
the FT or false positives of the biopsy [4,5,9]. In the absence of a study design of the EPIC3 Program. They collected, analyzed and
true gold standard, the fact that the virologic responses (either performed the initial interpretation of all virologic and clinical
SVR, EVR, or SVR among EVR) of patients classified as F0 by FT data. Biopredictive performed and interpreted the FibroTest assay
were significantly higher than of those classified as F2 by biopsy and analyzed the FibroTest data. All the authors provided input
(40% vs 27% for SVR, 54% vs 42% for EVR, 75% vs 63% for SVR and agreed to the final interpretation. The decision to submit
among EVR) strongly suggest that the liver biopsy scores were the report was made by Schering-Plough Research Institute in
incorrect. The multivariate analysis reinforced this hypothesis conjunction with the publication committee.
as the knowledge of discordant cases significantly increased the
prognostic value of fibrosis staging using biopsy but not using FT. Acknowledgments
Despite the smaller number of validations for ST, the present
study suggests that similar to findings in treatment-naive We would like to thank Weiping Deng, Nicole Stauffer, Luminita
patients, steatosis scored by SteatoTest is another predictor of Justice, Michele Sauer, Barbara Kapelan, Misti Linaberry, Kather-
SVR (albeit less accurate) in previously non-responder patients. ine Giordano, Stefanie Alfano, Peter Savino, Michael Salman,
However, the respective roles of metabolic and viral steatosis as and Becky Liou of Schering-Plough Research Institute for their
independent prognostic factors need further study [36]. contributions to the conduct of the study.

Conclusions Supplementary data

Biomarkers such as FibroTest can be used as a possible alterna- Supplementary data associated with this article can be found, in
tive to liver biopsy for fibrosis staging and thus simplify the man- the online version, at doi:10.1016/j.jhep.2010.06.038.
agement of patients with CHC who failed their first treatment. As
previously suggested in treatment-naive patients, these validated
biomarkers should also facilitate the design of trials in non-treat- References
ment-naive patients [8,9]. They could also be used as surrogate
[1] NIH consensus statement on management of hepatitis C: 2002. NIH Consens
markers in trials evaluating the risk–benefit of maintenance ther-
State Sci Statements 2002;19:1–46.
apy, without increasing the risk and the cost of repeated liver [2] Manning DS, Afdhal NH. Diagnosis and quantitation of fibrosis. Gastroen-
biopsies but with an increase of power through repeated non- terology 2008;134:1670–1681.
invasive measures of biomarkers. [3] Halfon P, Munteanu M, Poynard T. FibroTest–ActiTest as a non-invasive
marker of liver fibrosis. Gastroenterol Clin Biol 2008;32:22–38.
[4] Poynard T, Munteanu M, Imbert-Bismut F, Charlotte F, Thabut D, Le Calvez S,
Financial support et al. Prospective analysis of discordant results between biochemical
markers and biopsy in patients with chronic hepatitis C. Clin Chem
2004;50:1344–1355.
T. Poynard, M. Colombo, S. Flamm, T. Berg, E. Schiff, E.J. Heathco- [5] Ngo Y, Munteanu M, Messous D, Charlotte F, Imbert-Bismut F, Thabut D,
te, A. Craxi, M. Silva, and W. Schmidt are members of the Speak- et al. A prospective analysis of the prognostic value of biomarkers (FibroTest)
ers Bureau for Schering-Plough; A. Craxi, E.J. Heathcote, and W. in patients with chronic hepatitis C. Clin Chem 2006;52:1887–1896.
Schmidt are members of the Speakers Bureau for Roche; T. Poy- [6] Fontaine H, Petitprez K, Roudot-Thoraval F, Trinchet JC. Guidelines for the
diagnosis of uncomplicated cirrhosis. Gastroenterol Clin Biol 2007;31:
nard, M. Colombo, E. Schiff, M. Diago, S. Flamm, T. Berg, A. Craxi,
504–509.
M. Silva, E.J. Heathcote, and F. Gonçales receive research support [7] La Haute Autorité de Santé (HAS) in France. The HAS recommendations for the
from Schering-Plough; A. Craxi and E.J. Heathcote receive management of the chronic hepatitis C using non-invasive biomarkers. <http://
research support from Roche; S. Flamm receives research support www.has-sante.fr/portail/display.jsp?id 1=4 c_476486> [accessed 08.08].

234 Journal of Hepatology 2011 vol. 54 j 227–235

JOURNAL OF HEPATOLOGY
[8] Poynard T, Imbert-Bismut F, Ratziu V, Chevret S, Jardel C, Moussalli J, et al. [22] Poynard T, Halfon P, Castera L, Charlotte F, Le Bail B, Munteanu M, et al.
Biochemical markers of liver fibrosis in patients infected by hepatitis C virus: Variability of the area under the receiver operating characteristic curves in
longitudinal validation in a randomized trial. J Viral Hepat 2002;9:128–133. the diagnostic evaluation of liver fibrosis markers: impact of biopsy length
[9] Poynard T, McHutchison J, Manns M, Myers RP, Albrecht J. Biochemical and fragmentation. Aliment Pharmacol Ther 2007;25:733–739.
surrogate markers of liver fibrosis and activity in a randomized trial of [23] Poynard T, Munteanu M, Morra R, Ngo Y, Imbert-Bismut F, Thabut D, et al.
peginterferon alfa-2b and ribavirin. Hepatology 2003;38:481–492. Methodological aspects for the interpretation of liver fibrosis non-invasive
[10] D’Arondel C, Munteanu M, Moussalli J, Thibault V, Naveau S, Simon A, et al. A biomarkers: a 2008 update. Gastroenterol Clin Biol 2008;32:8–21.
prospective assessment of an ‘a la carte’ regimen of PEG-interferon alpha2b [24] Hintze JL. NCSS 2007 user guide. Kaysville, UT: Number cruncher statistical
and ribavirin combination in patients with chronic hepatitis C using systems software NCSS; 2007.
biochemical markers. J Viral Hepat 2006;13:182–189. [25] Ngo Y, Benhamou Y, Thibault V, Ingiliz P, Munteanu M, Lebray P, et al. PLoS
[11] Patel K, Benhamou Y, Yoshida EM, Kaita KD, Zeuzem S, Torbenson M, et al. An One 2008;3:e2573.
independent and prospective comparison of two commercial fibrosis marker [26] Naveau S, Gaudé G, Asnacios A, Agostini H, Abella A, Barri-Ova N, et al.
panels (HCV FibroSURE and FIBROSpect II) during albinterferon alfa-2b Diagnostic and prognostic values of non-invasive biomarkers of fibrosis in
combination therapy for chronic hepatitis C. J Viral Hepat 2009;16:178–186. patients with alcoholic liver disease. Hepatology 2009;49:97–105.
[12] Poynard T, Colombo M, Bruix J, Schiff E, Terg R, Flamm S, et al. The Epic Study [27] Brissot P, de Bels F. Current approaches to the management of hemochro-
Group. Peginterferon alfa-2b and ribavirin: effective in patients with matosis. Hematol Am Soc Hematol Educ Program 2006:36–41.
hepatitis C who failed interferon alfa/ribavirin therapy. Gastroenterology [28] Poynard T, Munteanu M, Ngo Y, Castera L, Halfon P, Ratziu V, et al. ActiTest
2009;136:1618–1628. accuracy for the assessment of histological activity grades in patients with
[13] Imbert-Bismut F, Messous D, Thibaut V, Myers RB, Piton A, Thabut D, et al. chronic hepatitis C, an overview using Obuchowski measure. Gastroenterol
Intralaboratory analytical variability of biochemical markers of fibrosis Clin Biol 2010;34:388–396.
(Fibrotest) and activity (Actitest) and reference ranges in healthy blood [29] Bedossa P, Dargère D, Paradis V. Sampling variability of liver fibrosis in
donors. Clin Chem Lab Med 2004;42:323–333. chronic hepatitis C. Hepatology 2003;38:1449–1457.
[14] Poynard T, Imbert-Bismut F, Munteanu M, Messous D, Myers RP, Thabut D, [30] Di Bisceglie AM, Shiffman ML, Everson GT, Lindsay KL, Everhart JE, Wright
et al. Comp Hepatol 2004;3:8. EC, et al. Prolonged therapy of advanced chronic hepatitis C with low-dose
[15] Imbert-Bismut F, Messous D, Raoult A, Poynard T, Bertrand JJ, Marie PA, et al. peginterferon. N Engl J Med 2008;359:2429–2441.
Results transferability on RXL, ARX, X-Pand, BN2 (Dade Behring) and [31] Poynard T, Ratziu V. Prolonged therapy for hepatitis C with low-dose
Modular DP (Roche Diagnostics) analysers: application to component assays peginterferon. N Engl J Med 2009;360:1152.
of Fibrotest and Actitest. Ann Biol Clin (Paris) 2005;63:305–313. [32] Afdhal NH, Levine R, Brown R, Freilich B, O’Brien M, Brass C. Colchicine
[16] Poynard T, Ratziu V, Naveau S, Thabut D, Charlotte F, Messous D, et al. Comp versus peg-interferon alfa 2b long-term therapy: results of the 4-year
Hepatol 2005;4:10. COPILOT Trial. J Hepatol 2008;48:S4 (abstract).
[17] Bedossa P, Poynard T. For the METAVIR cooperative study group. An [33] Bruix J, Poynard T, Colombo M, Schiff E, Reichen J, Burak K, et al. PegIntron
algorithm for the grading of activity in chronic hepatitis C. Hepatology maintenance therapy in cirrhotic (METAVIR F4) HCV patients, who failed to
1996;24:289–293. respond to interferon/ribavirn (IR) therapy: final results of the EPIC3
[18] DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two cirrhosis maintenance trial. J Hepatol 2009;50 (Suppl 1):S22.
or more correlated receiver operating characteristic curves: a nonparametric [34] Poynard T, McHutchison J, Goodman Z, Ling MH, Albrecht J. For the
approach. Biometrics 1988;44:837–845. ALGOVIRC Project Group. Is an ‘‘a la carte’’ combination interferon alfa-2b
[19] Zhou X, Obuchowski N, McClish D. Statistical Methods in Diagnostic plus ribavirin regimen possible for the first line treatment in patients with
Medicine. New York, NY: John Wiley and Sons; 2002. p. 235. chronic hepatitis C? Hepatology 2000;31:211–218.
[20] Poynard T, Halfon P, Castera L, Munteanu M, Imbert-Bismut F, Ratziu V, et al. [35] Veldt BJ, Heathcote EJ, Wedemeyer H, Reichen J, Hofmann WP, Zeuzem S,
Standardization of ROC curve areas for diagnostic evaluation of liver fibrosis et al. Sustained virologic response and clinical outcomes in patients with
markers based on prevalences of fibrosis stages. Clin Chem 2007;53: chronic hepatitis C and advanced fibrosis. Ann Intern Med 2007;147:
1615–1622. 677–684.
[21] Lambert J, Halfon P, Penaranda G, Bedossa P, Cacoub P, Carrat F. How to [36] Poynard T, Ratziu V, McHutchison J, Manns M, Goodman Z, Zeuzem S, et al.
measure the diagnostic accuracy of noninvasive liver fibrosis indices: the Effect of treatment with peginterferon or interferon alfa-2b and ribavirin on
area under the ROC curve revisited. Clin Chem 2008 Aug;54 (8):1372–1378. steatosis in patients infected with hepatitis C. Hepatology 2003;38:75–85.

Journal of Hepatology 2011 vol. 54 j 227–235 235