Predictability of HCV treatment outcome
J Gastrointestin Liver Dis
September 2006 Vol.15 No.3, 213-219
Address for correspondence:Mimidis Konstantinos, MD, PhD
8, Chrisostomou Smirnis str.
GR-68100 Alexandroupolis, Greece
Hepatitis C Virus Survival Curve Analysis in Naïve Patients
Treated with PegInterferon α α α α α-2b Plus Ribavirin. A Randomized
Controlled Trial for Induction with High Doses of PegInterferon
and Predictability of Sustained Viral Response from Early
Konstantinos Mimidis1, Vassilios P Papadopoulos1, Ioannis Elefsiniotis2, Dimitrios Koliouskas3, Ioannis Ketikoglou4,
Emmanouil Paraskevas5, Stylianos Kanatakis6, Dimitrios Chrysagis7, Georgios N Dalekos8, Charalambos Tzathas9,
Andreas Protopapas10, Eleni Gigi11, Epameinondas Tsianos12, Georgios Kartalis1
1) 1st Department of Medicine, Democritus University of Thrace, Alexandroupolis. 2) Helena Venizelou Maternal Hospital,
Athens. 3) ”Ahepa” Hospital, Aristotelion University of Thessaloniki, Thessaloniki. 4) Hippokration Hospital, Athens. 5)
Savvas Hospital, Athens. 6) Red Cross Hospital, Athens. 7) Loimodon Hospital, Thessaloniki. 8) Department of Medicine,
Academic Liver Unit, University of Thessaly, Larissa. 9) Attikon University Hospital, Athens. 10) Giannitsa General
Hospital, Giannitsa. 11) Aristotelion University of Thessaloniki, Thessaloniki. 12) Department of Medicine, University of
Aim. To evaluate the significance of induction with high
doses of pegylated interferon α-2b (Peg-IFNα-2b) and the
predictability of sustained virologic response (SVR) in naïve
patients with chronic hepatitis C. Methods. 188 consecutive
naïve patients with chronic hepatitis C were enrolled in a
randomised controlled clinical trial. Patients were randomised
to receive either Peg-IFNα-2b 3.0 mcg/kg QW x 12 weeks
followed by 1.5 mcg/kg QW x 36 weeks plus 800-1200 mg
ribavirin (Arm A) or Peg-IFNα-2b 1.5 mcg/kg QW x 48 weeks
plus 800-1200 mg ribavirin (Arm B). HCV-RNA was obtained
at 0, 4, 8, 12, 16, 24, 48 and 72 weeks. Differences between
schemes were evaluated by Kaplan-Meier curves.
Predictability of SVR was assessed by two-way contingency
table analysis and ROC curve analysis. Results. From 176
patients, 75 had genotype 1, 15 genotype 2, 75 genotype 3
and 11 genotype 4. No statistical significance emerged in
HCV-RNA positivity, side effects and withdrawals between
schemes. Patients with genotype 1 achieved lower SVR
(46.6%) in comparison to patients with genotypes 2/3 (94.1%,
p<0.001) and 4 (90.9%, p=0.002). The most appropriate time
for estimation of SVR for genotype 1 is week 8 (accuracy=
0.84, AUC=0.90) while predictability increases with time in
genotypes 2/3, reaching maximum accuracy=0.93 and
AUC=0.76 at week 16. Conclusion. Induction with high
doses of Peg-IFNα-2b does not preclude better outcome
and rapid virologic response at 4 weeks of treatment
sufficiently predicts SVR. These findings might be useful in
an attempt to gain supportive evidence for decision making
in difficult-to-treat patients.
Chronic hepatitis C - pegylated interferon α-2b - ribavirin
- sustained virologic response
Chronic hepatitis C virus (HCV) infection is a major
healthcare challenge worldwide. Over 150,000 000
individuals are infected. Current therapeutic directives
underline that the combination of pegylated interferon (Peg-
IFN) (either α-2a or α-2b) and ribavirin is the best approach
for the time being (1). No substantial differences have been
detected between the two Peg-IFNs (2). Nevertheless, as
Peg-IFN α-2b is characterized by a larger volume of
distribution than α-2a, it should be adjusted for body weight
(3). So far, the clinical significance of the administration of
larger doses of Peg-IFNα-2b in the initial phase of treatment
has not been clarified.
It has been demonstrated that patients with chronic
hepatitis C treated with Peg-IFNα-2a/b and rivabirin achieve
higher rates of sustained virologic response (SVR;
undetectable HCV-RNA in serum 24 weeks after the end of
treatment) in cases of HCV genotypes 2 and 3. In contrast,
HCV genotype 1 has been correlated with poorer prognosis
at the end of treatment (48 weeks – EOT) and reduced
frequency of SVR (72 weeks) (4,5).
A number of clinical studies have tried to evaluate the
predictability of the final outcome based on HCV-RNA data.
It has been proposed that the decision to continue or aban-
don therapy can be made as early as week 12 (6,7). Further-
more, rapid virologic response (defined as 2 log reduction
of HCV-RNA at week 4) has been proposed to correlate
with a better outcome; this was clinically evaluated in 16-,
14- and 12-week duration studies for genotypes 2/3 (8-10).
The present study aims to evaluate the usefulness of
induction with higher doses of Peg-IFNα-2b (3.0 μcg/kg)
Mimidis et al 214
during the initial trimester of a 48-week therapy in
combination with ribavirin and the predictability of SVR from
early virological data.
Material and methods
The present study is a randomised, controlled clinical
trial that was conducted in 8 clinical centres. The sample
size (at least 79 individuals in each group) was chosen to
allow detection of 20% difference (baseline treatment 70%
effective, proposed treatment 90% effective) with a statistical
significance of 0.95 and a power of 0.90. The total sample
size was further increased by 20% for the fear of case losses
during the study.
Inclusion criteria included detectable HCV-RNA in
serum with a qualitative PCR method, liver biopsy indicating
chronic hepatitis within 6 months before enrollment, elevated
alanine aminotransferase at entry and at least once during 6
months before screening and patients’ informed consent.
Exclusion criteria included hepatitis B infection,
hemochromatosis, α-1 anti-trypsin deficiency, Wilson’s
disease, autoimmune hepatitis, alcohol-, drug-, or obesity-
induced liver disease, HIV infection, active substance abuse
and any known pre-existing medical condition that could
interfere with a patient’s participation. Laboratory values
that excluded patients for enrollment included a serum
creatinine > 1.5 mg/dl, an absolute neutrophile count < 1,000/
μl, a platelet count < 50,000 or hemoglobin < 11 g/dl.
To achieve an analysis of at least 158 patients (two
groups consisting of at least 79 patients each), 188 naïve
patients with chronic hepatitis C were initially enrolled in
the study. Each patient was randomised to receive either
Peg-IFNα-2b 3.0 mcg/kg QW x 12 weeks followed by Peg-
IFNα-2b 1.5 mcg/kg QW x 36 weeks plus 800-1200 mg ribavi-
rin (Arm A) or Peg-IFNα-2b 1.5 mcg/kg QW x 48 weeks plus
800-1200 mg ribavirin (Arm B). Ribavirin dose was adjusted
to body weight (11 mg/kg) (5). HCV-RNA was tested at week
0, 4, 8, 12, 16, 24, 48 and 72 with the use of COBAS Amplicor
HCV test, v.2.0 assay (Roche Molecular Systems,
Branchburg, NJ), a qualitative method with sensitivity of
100 copies/ml. HCV genotype was determined with the
INNO-LiPA HCV II kit (Bayer Diagnostics, Emeryville, CA).
Adherent subjects were defined as those who received
80% or more of their total IFN dose, who received 80% or
more of the ribavirin dose, and were treated for 80% or more
of the expected duration of therapy (80/80/80 criterion).
Nonadherent patients included those who underwent dose
reduction (to < 80% of either drug) or were prematurely
withdrawn from treatment (treated for < 80% of the prescribed
Kaplan-Meier HCV survival curve analysis based on
HCV-RNA data was evaluated according to treatment
received (Arm A or Arm B) and genotype (1 or 2/3 or 4). The
subgroups on which Kaplan-Meier was applied were tested
for gender and age to document uniformity of these potential-
ly confounding parameters. The duration of the observation
period arbitrarily ended in non-responders at the last
(positive) HCV-RNA point. Similarly, the HCV death event
(response) was arbitrarily computed as the mean between
the last positive and the first negative HCV-RNA result,
under the prerequisite that the patient remained negative till
the end of the observation period. Additionally, Kaplan-
Meier analysis enabled integration of data from patients
who discontinued treatment (intentionally or due to adverse
effects) before 48 weeks.
In an effort to assess the predictive value of early
virological data concerning the final outcome (SVR), ROC
curve analysis was performed and area under curve (AUC)
was computed for all possible pairs of HCV-RNA data at 4,
8, 12 and 16 weeks from the one side and SVR from the other
using the internet statistic tool at http://www.rad.jhmi.edu/
jeng/javarad/roc/JROCFITi.html. The response to therapy,
in terms of achievement of undetectable HCV-RNA levels
was considered as a positive event. Four categories were
used to describe the final outcome (definitely positive
outcome: undetectable HCV-RNA at week 72; possibly
positive outcome: undetectable HCV-RNA at weeks 24/48
and unknown HCV-RNA at week 72; possibly negative
outcome: detectable HCV-RNA at weeks 24/48 and unknown
HCV-RNA at week 72; and definitely negative outcome:
detectable HCV-RNA at week 72). When the data did not
allow the construction of a fitted curve, an empiric curve
was alternatively plotted. All fitted curves are accompanied
by their 95% confidence intervals.
Two-way contingency table analysis including accuracy,
sensitivity, specificity, positive predictive value, negative
predictive value, number needed to treat, number needed to
diagnose and risk ratio (along with P) was performed for all
possible pairs of HCV-RNA data at 4, 8, 12 and 16 weeks
from the one side and SVR from the other using the internet
statistic tool http://members.aol.com/johnp71/ctab2x2.html.
This analysis included all patients who completed
successfully the follow-up period.
Continuous parameters were tested using the Student’s
t-test (in case of two groups) and ANOVA (in case of more
than two groups). When the sample size fell between 10 and
40, Kolmogorov-Smirnov and Lilliefors tests were used to
ascertain comparability. Discrete parameters were tested
using Chi-square test. When expected frequencies were
found to be smaller than 5, the Fisher’s exact test was
alternatively preferred. All means and risk ratio values are
accompanied by their 95% confidence limits. All tests were
two-tailed. The level of statistical significance was set to
Twelve patients were excluded from the study, as they
had not fulfilled the 80/80/80 criterion. These patients were
totally ignored in the analysis process.
From the remaining adherent 176 patients, 75 had HCV
genotype 1 (42.6%), 15 had HCV genotype 2 (8.5%), 75 had
HCV genotype 3 (42.6%) and 11 had HCV genotype 4 (6.3%).
Predictability of HCV treatment outcome 215
During therapy, 20 patients withdrew consent and 12
discontinued therapy due to side effects. These patients
were incorporated in the Kaplan-Meier analysis for the time
they were followed and in the ROC curve analysis if their
follow-up was at least 24 months. In contrast, two-way
contingency table analysis included only those patients
who had a complete follow-up.
Twenty-nine patients failed one or more times to be
present throughout follow-up. These patients, along with
genotype 4 patients (who were too few to form a separate
group) were also omitted from analysis for assessment of
All data concerning response to treatment (HCV-RNA
negativity), side effects and withdrawals are presented in
detail in Table I. The overall response in the group of 133
patients who reached EOT was 78.1%. The majority of them
(70.8%) also achieved SVR. A survival curve for the total
group is presented in Fig.1.
Table I Descriptive statistics of the present study regarding response rates, side effects and withdrawals. GEN: Genotype, Till EOT:
Followed till End Of Treatment, SE: Side effects, WD: Withdrawals, TO: Total. Total numbers are lower than expected as some
patients missed being present once or more times during the follow-up period
Week 48 (EOT)
Week 72 (SVR)
Fig.1 Kaplan-Meier plot (expressing HCV-RNA positivity) for
the total group of patients.
A ROC analysis for the total group is presented in Fig.2.
Area under curve is increased according to time.
Patients with HCV genotype 1 versus genotypes 2/3
From the patients with HCV genotype 1/2/3 (100 men, 65
women of mean age 39.97 ± 2.11 years), 83 followed Arm A
treatment and 82 Arm B treatment. Neither gender ratio nor
age differs between subgroups formed according to Arm
and Type. Additionally, type ratio was comparable between
Arms (Table II).
Twenty patients withdrew consent before EOT (15.9%).
Fourteen patients encountered side effects (8.0%) and
12 of them were been obliged to abandon therapy. All side
effects during the study period are presented in Table III.
No statistically significant result was found when subgroups
of patients according to Arm or genotype were compared.
In detail, the comparison between Arm A subgroup and
Arm B subgroup showed similar side effects rate (X2 = 0.77,
p = 0.38) and withdrawal rate (X2 = 0, p ~ 1).
When HCV-RNA positivity was concerned, a significant
difference between Arm A and Arm B was neither observed
in the total group (RR=0.89 with ±95% confidence intervals
0.55-1.23, P ~ 1), nor in the subgroup of genotype 1 patients
(RR=0.98 with ±95% confidence intervals 0.31-1.65, p ~ 1),
nor in the genotype 2/3 patients (RR=0.96 with ±95%
confidence intervals 0.52-1.40, p ~ 1). In contrast, a strong
difference was documented between genotype 1 and
genotype 2/3 (RR = 3.63 with ±95% confidence intervals
2.12-5.14, p < 0.001 – Fig.3).
The two-way contingency table analysis for the
estimation of SVR based on the viral load status at 4, 8, 12
and 16 weeks after the initiation of treatment is summarized
in Table IV. Risk ratios are all statistically significant
independent of genotype and week of observation. Both
best accuracy (0.84) and best AUC (0.90) are observed at
week 8 for genotype 1. In contrast, accuracy as well as AUC
increases with time in case of genotypes 2/3 (Fig.2).
Patients with HCV genotype 4
As there is little evidence regarding treatment of HCV
genotype 4 in Greece, it is believed that even the small group
of the present study (n=11) is informative.
Mimidis et al216
Fig.2 ROC curve analysis for predictability of SVR from early virologic data (weeks 4, 8, 12 and 16) for genotype 1, genotypes 2/3
and all patients. Gray lines represent 95% confidence intervals.
Predictability of HCV treatment outcome217
Table II Demographical data of patients enrolled in the study along with group comparability data
p for age
p for gender
p for type ratio
Arm A Type 1
Arm B Type 1
Arm A Type 2/3
Arm B Type 2/3
Table III Side effects according to Arm and Type. Two patients
continued treatment till EOT
Low back pain
Myelotoxicity, weight loss
Fig.3 Comparison of Kaplan-Meier plots between type 1 and
type 2/3 patients (dotted line: Type 2/3, straight line: Type 1).
A summary of the two treatment groups, qualitative HCV-
RNA results (at 4, 8, 12, 16, 24, 48 and 72 months), intentional
discontinuations and side effects are given in Table I.
The Kaplan-Meier analysis reveals that HCV type 4
presents limited survival compared with type 1 (p=0.002,
Fig.4) but not with genotypes 2/3 (P=0.63, Fig.5).
Fig.4 Comparison of Kaplan-Meier plots between type 1 and
type 4 patients (dotted line: Type 4, straight line: Type 1).
Fig.5 Comparison of Kaplan-Meier plots between type 2/3 and
type 4 patients (dotted line: Type 2/3, straight line: Type 4).
IFN-based regimens for the treatment of chronic hepatitis
C have become increasingly effective and currently are able
to eradicate the virus in more than 50% of cases. Peg-IFN α-