Liver International ISSN 1478-3223
VIRAL HEPATITIS
A tool for selecting patients with a high probability of sustained
virological response to peginterferon alfa-2a (40kD)/ribavirin
Peter Ferenci1, Rodrigo Aires2, Ioan Ancuta3, Andrew Arohnson4, Hugo Cheinquer5, Dragan Delic6,
Michael Gschwantler7, Dominique Larrey8, Ludovico Tallarico9, Manuela Schmitz10, Fernando Tatsch11* and
Denis Ouzan12
1 Medical University of Vienna, Vienna, Austria
2 The Goi
as Federal University Clinical Hospital, Gio^
ania, Brazil
3 Hospital “Dr. I. Cantacuzino”, Bucharest, Romania
4 University of Chicago Medical Center, Chicago, IL, USA
5 Hospital de Clınicas de Porto Alegre, Porto Alegre, Brazil
6 Clinic for Infectious Disease, Clinical Center Serbia, Belgrade, Serbia
7 4th Department of Internal Medicine, Wilhelminenspital, Vienna, Austria
8 Liver Unit, Saint Eloi Hospital, Montpellier, France
9 Divisione di Medicina, Ospedale Elena D’Aosta, Naples, Italy
10 IST GmbH, Mannheim, Germany
11 F. Hoffmann-La Roche Ltd, Basel, Switzerland
12 Institut Arnault Tzanck, Saint-Laurent-du-Var, France
Keywords
Caucasian – hepatitis C – peginterferon alfa/
ribavirin – predictive scoring – PROPHESYS –
sustained virological response
Correspondence
Peter Ferenci, Department of Internal
Medicine III, Gastroenterology and
Hepatology, Medical University of Vienna,
Vienna, Austria
Tel: +43 1 40400 4741
Fax: +43 1 404004735
e-mail: [email protected]
Received 14 May 2013
Accepted 7 December 2013
DOI:10.1111/liv.12439
Abstract
Background & Aims: Pretreatment identification of patients likely to achieve
a sustained virological response (SVR) with peginterferon alfa-2a/ribavirin
would be useful for individualizing treatment choices. The aim of this analysis was to devise a simple scoring system to identify patients with high probability of achieving an SVR with peginterferon alfa-2a/ribavirin. Methods:
Using data from 2109 Caucasian treatment-naive hepatitis C virus (HCV)
genotype 1 mono-infected patients from the PROPHESYS cohorts, the relationship between favourable baseline characteristics and SVR was explored
using generalized additive model analysis, and a scoring system was devised
to predict SVR. Results: Points were assigned for: age (years) (≤35: 2; >35,
≤45: 1; >45: 0); body mass index (kg/m2) (≤20: 2; >20, ≤22: 1; >22: 0); HCV
RNA (IU/ml) (≤100 000: 3; >100 000–400 000: 2; >400 000–800 000: 1;
>800 000: 0); platelets (>150 9109/l: 1; ≤150 9109/l: 0); alanine aminotransferase [9upper limit of normal (ULN)] (>3: 1; ≤3: 0); serum aspartate aminotransferase (9ULN) (≤1: 1; >1: 0). 1029, 698 and 382 patients had scores
of 0–2, 3–4 and ≥5, respectively, among whom SVR rates were 35.0, 54.9 and
76.7%. SVR in patients with scores ≥5 and undetectable HCV RNA by week
4 was 86.7%. The score was tested against two databases of patients who
received peginterferon alfa-2a/ribavirin in other clinical trials; similar high
SVR rates in patients with scores ≥5 were reported. Conclusions: The scoring
system can reliably identify treatment-naive HCV genotype 1 mono-infected
Caucasian patients who have a high probability of achieving an SVR with peginterferon alfa-2a/ribavirin and will be particularly useful where protease
inhibitors are not readily available.
Treatment with a hepatitis C virus (HCV) protease
inhibitor (PI) plus peginterferon alfa/ribavirin produces
higher sustained virological response (SVR) rates among
genotype 1-infected treatment-naive patients than dual
*Present address: AbbVie, Chicago, IL, USA
Clinicaltrials.gov: NCT01070550 (PROPHESYS 1) NCT01066793
(PROPHESYS 2) NCT01066819 (PROPHESYS 3).
Liver International (2014)
© 2013 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
peginterferon alfa/ribavirin therapy (1, 2). However, PIbased triple therapy increases the adverse event burden
compared with dual therapy (3, 4). PIs also predispose
patients to many potential drug interactions, which
increases the complexity of treatment. PI-based triple
therapy is recommended for genotype 1-infected
patients in the US (3). In contrast, treatment
recommendations are nuanced in several European (5,
1
Prediction of sustained virologic response
6) and Asian countries (7), where dual therapy remains
an acceptable treatment option for genotype 1
patients with undetectable HCV RNA at week 4 of dual
therapy.
Patients with HCV genotype 1 infection with undetectable HCV RNA at week 4 of treatment with peginterferon alfa-2a (40kD)/ribavirin achieve SVR rates of
88% compared with overall SVR rates up to 75% among
patients treated with PI-based triple therapy (1, 2, 8). It
must be noted that SVR rates are considerably lower in
non-rapid virological response (RVR) patients treated
with dual therapy, that treatment duration is generally
shorter in patients receiving PI-based triple combination regimens, and that easier-to-treat patients (i.e.,
those with an IL28B CC genotype) have been shown to
achieve SVR rates of 100% with telaprevir plus peginterferon alfa-2a (40kD)/ribavirin (9). Nevertheless, there is
a subset of patients that achieve high SVR rates when
treated with dual therapy.
If patients likely to achieve an SVR with dual therapy
could be reliably identified on the basis of pretreatment
characteristics, PIs could be reserved for the most difficult-to-treat patients, and the overall adverse event burden could be reduced. The database from the large
PROPHESYS cohorts, which includes data from more
than 4500 treatment-naive genotype 1-infected patients
(10), provides the opportunity to explore the feasibility
of a simple scoring system to identify patients most
likely to achieve an SVR with peginterferon alfa/ribavirin.
The objective of this analysis was to devise a simple
scoring system using a few readily obtainable, minimally
invasive baseline factors that would identify patients
with a high probability of achieving an SVR after treatment with standard doses of peginterferon alfa-2a
(40kD) (PEGASYSâ; Roche, Basel, Switzerland) plus
ribavirin.
Methods
Patients included in this analysis were enrolled in the
prospective international non-interventional PROPHESYS cohorts (see Supporting Information).
This analysis was restricted to treatment-naive
Caucasian patients with HCV genotype 1 mono-infection who were enrolled outside the USA and prescribed
peginterferon alfa-2a (40kD) 180 lg/week plus ribavirin
1000/1200 mg/day for an intended duration of
48 weeks. Only patients with complete records for key
baseline characteristics were included. Patients were
excluded if they had an end-of-treatment response, no
HCV RNA test result showing relapse, and missed an
HCV RNA test ≥140 days after end of treatment for
reasons unrelated to safety or efficacy.
Treatment success (SVR) was defined as serum HCV
RNA <50 IU/ml at or after the end of the 24-week
untreated follow-up period. RVR was defined as HCV
RNA <50 IU/ml by week 4.
2
Ferenci et al.
A unique scoring system, the dual therapy response
predictor (DTRP) score, was devised through a fourstep process. As a first step, baseline patient characteristics and disease-related factors that were included in
the scoring system were identified by multiple logistic
regression (MLR) analysis, with SVR as the dependent
variable. First, a univariate logistic regression analysis
was done in which factors associated with SVR with a
P-value of <0.25 were selected for consideration in
the MLR analysis. Next, the MLR analysis was done
using a backward elimination process; only factors
with a significance level of 0.05 remained in the
model. Finally, the remaining predictors were
included in the final model only if the factors were
selected in more than 50% of 500 backwards selection
runs using bootstrap samples of the analysis population. Factors considered in the analysis were age,
weight, body mass index (BMI), serum HCV RNA
level, gender, liver fibrosis, mode of infection, hepatic
steatosis, serum alanine aminotransferase (ALT) ratio,
serum aspartate aminotransferase (AST) ratio and
platelet count.
Step 2 involved generalized additive models (GAM)
using non-parametric smoothing splines. These were
applied to explore the structure of the relationship
between each of the continuous variables identified in
the MLR analysis and SVR. Furthermore, the values of
each factor were grouped by deciles and plots of the
empirical probabilities for the midpoints of the deciles
were included in the graphics.
In step 3, the thresholds used in the DTRP score were
determined. If the probability of achieving an SVR, as
presented graphically in step 2, was above 0.5, 0.6 and
0.7 for a given patient value of a factor, then 1, 2 or 3
points, respectively, were assigned to the value. The
actual cut-offs were then refined by selecting round
values to ensure ease of use.
If in the MLR analysis categorical variables (e.g., gender) had been selected, the SVR rates for each category
would have been used to assign 1, 2 or 3 points in accordance with the approach for continuous variables.
Finally, in step 4, the DTRP score assigned to a given
patient comprised the sum of all points for the selected
predictors.
The DTRP score was validated by comparing the
results with those from two cohorts of Caucasian, treatment-naive HCV genotype 1 mono-infected patients
who received treatment with standard doses of peginterferon alfa-2a (40kD)/ribavirin. Patients in one cohort
(validation set 1) had received treatment for a fixed
duration of 48 weeks in four randomized international
clinical trials (11–14). Patients in the second cohort
(validation set 2) were HCV genotype 1-infected
patients who had been treated for 24, 48 or 72 weeks in
a multicentre randomized controlled trial of responseguided therapy in Austria (15, 16). IL28B genotype data
were available for a subset of patients in validation set 2
(17).
Liver International (2014)
© 2013 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
Ferenci et al.
Prediction of sustained virologic response
The positive predictive value (PPV) of an RVR for
SVR, defined as the probability that a patient with an
RVR will achieve an SVR, was calculated for DTRP score
categories. The Cochran–Armitage test was used, as
indicated, to assess whether a higher DTRP score was
associated with a higher virological response rate. All
statistical tests were considered exploratory.
Results
A total of 7163 treatment-naive HCV mono-infected
patients were enrolled in PROPHESYS, of whom 2109
HCV genotype 1 mono-infected patients met the selection criteria and were included in the analysis (Fig. S1).
All patients were Caucasian, 50.7% were male, the mean
age was 47.0 years, the mean BMI was 25.7 kg/m2, the
mean serum HCV RNA level was 6.0 log10 IU/ml and
35.6% of patients who had a pretreatment liver fibrosis
assessment with a valid result were diagnosed with
bridging fibrosis or cirrhosis (Table 1).
Step 1. MLR analysis to identify predictive factors
Predictive factors retained in the final MLR model for
SVR were age, BMI, serum HCV RNA level, hepatic
fibrosis, ALT ratio, AST ratio and platelet count. Factors
retained in the final MLR model were used to construct
Table 1. Baseline characteristics of patients with complete records
included in the analysis
Characteristic
All patients (N = 2109)
Male, n (%)
1069 (50.7)
Mean age ± SD, years
47.0 ± 11.8
Mean body weight ± SD, kg
73.7 ± 14.7
Mean body mass index ± SD, kg/m2
25.7 ± 4.2
Mean ALT ratio ± SD†
2.3 ± 1.9
Mean AST ratio ± SD‡
1.8 ± 1.5
Mean platelet count ± SD, 9109/l
214.2 ± 69.2
Platelet count 9109/l, n (%)
≥150
1736 (82.3)
<150
373 (17.7)
Method used to assess hepatic fibrosis, n (%)
Biopsy
1321 (62.6)
Non-invasive test
290 (13.8)
Not assessed
498 (23.6)
Patients with bridging fibrosis or
573/1609 (35.6)
cirrhosis, n/N (%*)
Mean HCV RNA level ± SD, 9106 IU/ml
3.1 ± 5.1
HCV RNA level (IU/ml), n (%)
≤400 000
525 (24.9)
>400 000–800 000
250 (11.9)
>800 000
1334 (63.3)
*Only patients with a valid result from a fibrosis assessment by biopsy
or by a non-invasive method were included in the calculation.
†ALT ratio = patient’s ALT value divided by ULN for the local laboratory.
‡AST ratio = patient’s AST value divided by ULN for the local laboratory.
ALT, alanine aminotransferase; AST, aspartate aminotransferase; SD,
standard deviation; ULN, upper limit of normal.
Liver International (2014)
© 2013 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
the DTRP score. Liver histology (hepatic fibrosis) was
the only factor that was not included in the score. This
decision was taken so as not to impose a requirement
for a pretreatment biopsy.
Steps 2 and 3. Graphic analysis and selection of cut-offs
The graphic displays of the GAM analysis for the six
factors that comprise the DTRP score are shown in
Fig. 1 and the scoring system and cut-offs that were
arrived at with the aid of these graphic displays are
shown in Table 2.
Step 4. Derivation of patient scores
To derive the total score for a given patient, the individual’s baseline characteristics are assigned points as
shown in Table 2 and the sum is calculated. The total
score ranges from 0 to 10 for a given patient and higher
scores are associated with a higher chance of treatment
success.
The distribution of DTRP scores in the 2109 patients
included in this analysis is shown in Fig. 2. The median
score was 3 (interquartile range: 1–4). Values for baseline characteristics included in the DTRP score are
shown in Table 3. As expected, SVR rates for each predictor were related to the score for that predictor, with
higher scores being associated with higher rates of SVR
(Table 4).
In the 1609 patients with invasive or non-invasive
fibrosis data, the percentage of patients with bridging
fibrosis or cirrhosis and baseline DTRP score of 0, 1, 2,
3, 4 and ≥5 were 69.8% (44/63), 46.7% (148/317),
34.3% (132/385), 33.2% (99/298), 33.3% (83/249) and
22.6% (67/297) respectively.
Virological response according to DTRP score
Sustained virological response rates increased with
DTRP score and were 35.0, 54.9 and 76.7%, respectively,
in patients with scores of 0–2, 3–4 and ≥5 points (Cochran–Armitage test, P < 0.0001). The PPV of baseline
DTRP score for SVR is shown in Fig. S2). The trend
towards higher SVR rates with increasing DTRP score
remained intact when patients were grouped according
to the extent of hepatic fibrosis; however, within each
DTRP category SVR rates were consistently 10–13%
higher in patients without bridging fibrosis/cirrhosis
than in patients with bridging fibrosis/cirrhosis
(Fig. S3). The difference in SVR rates between females
and males within each category was <2% (data not
shown).
Rapid virological response rates also increased with
DTRP score and were 12.7, 27.2 and 55.2% in patients
with scores of 0–2, 3–4 and ≥5 points (Cochran–Armitage test, P < 0.0001; Fig. 3). In contrast with the large
between-group differences in overall SVR rates in the
three categories, SVR rates among patients who
3
Ferenci et al.
Prediction of sustained virologic response
0.9
0.8
0.8
Probability (mSVR)
(B) 1.0
0.9
Probability (mSVR)
(A) 1.0
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0.7
0.6
0.5
0.4
0.3
0.2
0.1
10
20
30
40
50
60
70
80
0.0
90
10
30
20
0.9
0.8
0.8
Probability (mSVR)
(D) 1.0
0.9
Probability (mSVR)
(C) 1.0
0.7
0.6
0.5
0.4
0.3
0.2
1
2
3
4
5
6
7
0.6
0.5
0.4
0.3
0.2
0.0
8
0
100
200
0.8
Probability (mSVR)
Probability (mSVR)
0.9
0.8
0.6
0.5
0.4
0.3
0.2
400
500
600
700
1.0
0.9
0.7
300
Baseline platelets x 109/l
(F)
1.0
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.1
0.0
60
0.7
HCV RNA at BL in log10(IU/ml)
(E)
50
0.1
0.1
0.0
40
BMI in kg/m2
Age in years
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17
Baseline ALT ratio
0.0
0
1
2
3
4
5 6
7
8
9 10 11 12 13 14 15 16 17 18 19
Baseline AST ratio
Fig. 1. Relationship between predictors included in the dual therapy response predictor (DTRP) scoring system and sustained virological
response (SVR). (A) Age; (B) body mass index; (C) hepatitis C virus RNA; (D) platelets; (E) ALT/ULN*; (F) AST/ULN*. Short vertical lines indicate
SVR (1.0) and no SVR (0.0). Open circles represent probability of SVR according to the GAM analysis. Closed black circles represent empirical
probability for midpoints of deciles (used in steps 2 and 3 of the process for deriving DTRP score). *ALT ratio and AST ratio = patient’s ALT
and AST value, respectively, divided by ULN for the local laboratory. ALT, alanine aminotransferase; AST, aspartate aminotransferase; ULN,
upper limit of normal.
achieved an RVR were much more similar, and were
almost identical in patients with a score of 3–4 and ≥5
points (Fig. 3). The PPV (95% CI) of an RVR for SVR
for patients with scores of 0–2, 3–4 and ≥5 points was
74.8% (66.5%, 82.0%), 86.8% (81.2%, 91.3%) and
86.7% (81.4%, 91.0%) respectively. The trend towards
higher SVR rates in patients with higher DTRP scores
was also apparent in patients who did not achieve an
RVR, although SVR rates were considerably lower in
these individuals (28.5, 42.2 and 62.7% in patients with
4
DTRP scores of 0–2, 3–4 and ≥5 points respectively;
Fig. 3).
Sustained virological response rates for patients with
DTRP scores of 0–2, 3–4 and ≥5 points in the PROPHESYS (training) and at non-US study sites of validation
set 1 were very similar (Fig. 4). When the DTRP scoring
system was applied to Caucasian treatment-naive, genotype 1 mono-infected patients enrolled at US study sites
in the four studies in validation set 1 (N = 252),
the trend in SVR rates was also similar, although the
Liver International (2014)
© 2013 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
Ferenci et al.
Prediction of sustained virologic response
scores of 0–2, 3–4 and ≥5, respectively (Fig. 4). Among
the subset with a known IL28B genotype, SVR rates were
consistently higher in patients with a CC than a non-CC
genotype; however, SVR rates were similar in those with
scores of 3–4 and ≥5 points both among individuals
with a CC genotype (85%) and those with a non-CC
genotype (65%) (Fig. S4).
Table 2. DTRP scoring system
Predictor
Score
Age, years
≤35
>35 but ≤45
>45
BMI, kg/m2
≤20
>20 but ≤22
>22
HCV RNA, IU/ml
≤100 000
>100 000 but ≤400 000
>400 000 but ≤800 000
>800 000
Platelet count 9109/l
>150
≤150
ALT/ULN*
>3
≤3
AST/ULN*
≤1
>1
2
1
0
2
1
0
Discussion
This analysis suggests that it is possible to identify
Caucasian patients with HCV genotype 1 infection who
have a high probability of achieving an SVR with dual
peginterferon alfa-2a/ribavirin therapy using a simple,
inexpensive and easy-to-calculate score. The score uses
readily available demographic (BMI, age) and laboratory data (ALT, AST, platelet count and HCV RNA
level). The score could be easily incorporated into routine patient encounters, either as a simple paper-based
form or as a component of an electronic medical record.
The score does not include certain well established predictors of response such as liver fibrosis, host IL28B
genotype or RVR status. Exclusion of these factors from
the DTRP score does not imply that they should be
ignored if available. Rather the DTRP score is one of
several important tools that can be used to inform treatment decisions in patients with chronic hepatitis C. For
example, the combination of a DTRP of 3 or more and
an RVR in a patient who has received 4 weeks of dual
therapy can be used to identify patients likely to achieve
an SVR rate of approximately 85%. Furthermore, the
presence of advanced fibrosis is an important indicator
of an immediate need for anti-HCV therapy; thus the
DTRP can be complemented by other information such
as invasive or non-invasive fibrosis assessments when
available.
3
2
1
0
1
0
1
0
1
0
*ALT ratio and AST ratio = patient’s ALT and AST value, respectively,
divided by the ULN for the local laboratory.
ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI,
body mass index; DTRP, dual therapy response predictor; HCV, hepatitis
C virus; ULN, upper limit of normal.
absolute SVR rates were somewhat lower: 30.1% (49/
163), 50.0% (37/74) and 73.3% (11/15), respectively, in
patients with scores of 0–2, 3–4 and ≥5 points. Similarly
SVR rates in Austrian patients with genotype 1 infection
(validation set 2, N = 394) were 40.5% (68/168), 67.8%
(101/149) and 80.5% (62/77) in individuals with DTRP
25
23.2
21.2
Patients (%)
20
18.3
14.8
15
10
5
8.8
5.8
4.4
2.0
0
93
2109
447
2109
489
2109
385
2109
0
1
2
3
0.9
0.5
313
2109
186
2109
122
2109
43
2109
20
2109
11
2109
4
5
6
7
8
9
DTRP score
Fig. 2. Distribution of dual therapy response predictor score.
Liver International (2014)
© 2013 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
5
Ferenci et al.
Prediction of sustained virologic response
Table 3. Baseline characteristics included in the score for the training and validation sets
Parameter
Age, years
BMI, kg/m2
HCV RNA, 9106 IU/ml
Platelets, 9109/l
ALT/ULN*
AST/ULN*
Training set
(from PROPHESYS) (N = 2109)
Validation set 1
(non-US) (N = 720)
Validation set 2
(Austria) (N = 394)
47.0 (11.8)
[48, 18–83]
25.7 (4.2)
[25.1, 15–56]
3.12 (5.08)
[1.39, <0.1–77.0]
214 (69)
[211, 39–609]
2.27 (1.86)
[1.69, 0.2–16.4]
1.83 (1.47)
[1.37, 0.2–18.1]
43.0 (10.2)
[44, 19–74]
27.0 (4.9)
[26.5, 17–64]
3.30 (4.57)
[1.84, <0.1–35.0]
222 (63)
[217, 82–528]
2.62 (2.24)
[1.95, 0.4–17.1]
1.62 (1.25)
[1.24, 0.4–12.5]
44.8 (10.7)
[46, 18–67]
25.6 (4.0)
[25.3, 18–41]
1.83 (3.68)
[0.71, <0.1–38.9]
221 (61)
[218, 63–438]
2.17 (1.69)
[1.66; 0.4–12.9]
1.85 (1.61)
[1.43; 0.5–24.4]
*ALT ratio and AST ratio = patient’s ALT and AST values, respectively, divided by the ULN for the local laboratory. The upper limit of normal for
patients from Austria were 45 IU/l for ALT and 35 IU/l for AST.
Values are expressed as mean (SD) or [median, min–max].
ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; HCV, hepatitis C virus; SD, standard deviation; ULN, upper
limit of normal.
Table 4. SVR rate by categorized predictors included in the DTRP scoring system
Predictor
Age, years
≤35
>35 but ≤45
>45
BMI, kg/m2
≤20
>20 but ≤22
>22
HCV RNA, IU/ml
≤100 000
>100 000 but ≤400 000
>400 000 but ≤800 000
>800 000
Platelet count, 9109/l
>150
≤150
ALT/ULN*
>3
≤3
AST/ULN*
≤1
>1
Score
Training set
SVR % (95% CI)
Validation set 1 (non-US)
SVR % (95% CI)
Validation set 2 (Austria)
SVR % (95% CI)
2
1
0
69.9 (64.9, 74.5)
50.8 (46.2, 55.4)
42.3 (39.6, 45.1)
69.1 (61.7, 75.9)
50.8 (44.3, 57.3)
38.7 (33.2, 44.4)
79.2 (68.0, 87.8)
61.2 (51.7, 70.1)
50.0 (43.0, 57.0)
2
1
0
63.6 (54.2, 72.2)
59.2 (52.5, 65.7)
46.9 (44.5, 49.2)
65.5 (45.7, 82.1)
59.7 (47.5, 71.1)
48.3 (44.3, 52.3)
85.2 (66.3, 95.8)
68.1 (52.9, 80.9)
55.0 (49.4, 60.5)
3
2
1
0
77.5 (71.5, 82.8)
59.4 (53.6, 65.0)
51.6 (45.2, 57.9)
41.5 (38.9, 44.2)
83.7 (70.3, 92.7)
63.7 (53.6, 73.0)
52.4 (41.1, 63.6)
43.5 (39.1, 48.1)
81.3 (63.6, 92.8)
71.1 (59.5, 80.9)
56.1 (45.7, 66.1)
51.1 (43.7, 58.4)
1
0
53.1 (50.7, 55.5)
30.6 (25.9, 35.5)
52.0 (48.1, 56.0)
35.4 (25.1, 46.7)
60.7 (55.4, 65.9)
42.2 (27.7, 57.8)
1
0
52.7 (48.0, 57.3)
48.1 (45.7, 50.6)
58.1 (50.8, 65.2)
47.3 (42.9, 51.6)
64.7 (52.2, 75.9)
57.4 (51.8, 62.8)
1
0
55.6 (51.7, 59.5)
46.3 (43.7, 48.9)
52.4 (46.0, 58.8)
48.9 (44.3, 53.6)
65.1 (54.1, 75.1)
56.8 (51.1, 62.4)
*ALT ratio and AST ratio = patient’s ALT and AST value, respectively, divided by ULN for the local laboratory.
ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; CI, confidence interval; HCV, hepatitis C virus; ULN, upper
limit of normal.
Liver fibrosis was associated with treatment outcomes
in the MLR analysis; however, the results of a liver
biopsy were excluded from the score because it is invasive, and not acceptable to many patients. The use of the
score does not preclude non-invasive assessment of
fibrosis (or biopsy for that matter), which provides useful information (18). Indeed, it is prudent to assess liver
6
fibrosis prior to initiating therapy, and reasonable to
assume that the presence of advanced fibrosis will
reduce the chance of achieving an SVR even in patients
with a high DTRP score.
Individual patient DTRP scores ranged from 0 to 10,
with higher scores indicating a higher probability of
treatment success. SVR rates in patients with scores ≥5
Liver International (2014)
© 2013 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
Ferenci et al.
Prediction of sustained virologic response
0–2 points
3–4 points
≥5 points
100
86.8 86.7
74.8
Patients (%)
75
P < 0.0001
62.7
55.2
50
42.2
28.5
27.2
25
12.7
0
131
1029
190
698
211
382
RVR
98
131
165
190
183
211
SVR in patients
with an RVR
239
838
194
460
89
142
SVR in patients
without an RVR
Fig. 3. Rapid virological response (RVR) rates and sustained virological response (SVR) rates in patients who did and did not achieve an RVR
according to dual therapy response predictor (DTRP) score. The Cochran–Armitage test was used to test whether a higher DTRP score was
associated with a higher RVR rate.
0–2 points
3–4 points
≥5 points
100
Patients with an SVR (%)
P < 0.0001
80.5
76.8
76.7
75
67.8
57.3
54.9
50
40.5
37.7
35.0
25
0
360
1029
383
698
293
382
Training set (N = 2109)
141
374
134
234
86
112
Validation set 1 (N = 720)
(Non-US sites)
68
168
101
149
62
77
Validation set 2 (N = 394)
Fig. 4. Sustained virological response (SVR) rates in patients with a dual therapy response predictor (DTRP) score of 0–2, 3–4 and ≥5 points
in PROPHESYS (training dataset) and in the validation datasets. The Cochran–Armitage test was used to test whether a higher DTRP score
was associated with a higher SVR rate.
were 76.7% in the PROPHESYS (training) dataset,
76.8% in validation set 1 and 80.5% in validation set 2.
These results are in remarkably close agreement, especially when one considers that PROPHESYS was a large
and diverse cohort study and the validation data were
Liver International (2014)
© 2013 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
taken from large randomized controlled trials that were
conducted over a period of approximately one decade
(11–16).
The results of the present study cannot be directly
compared with the results of phase III trials with
7
Prediction of sustained virologic response
telaprevir or boceprevir. In a phase III study with boceprevir, in which treatment-naive genotype 1 patients
received a 4-week lead-in with dual therapy, only 66/938
(7%) of patients had undetectable HCV RNA after
4 weeks (1). This is in contrast with many European
and Asian randomized controlled studies with dual
therapy showing RVR rates between 25 and 65% (19).
Among patients with undetectable HCV RNA at week 4
who received 48 weeks of dual peginterferon alfa/ribavirin therapy, the SVR rate was 97%, which was similar
to the 90% SVR rate in patients who were HCV RNAnegative at week 4 and subsequently received boceprevir-based triple combination therapy (1).
As noted above, the overall SVR rate was 77% among
patients of the training set with a score of ≥5 points and,
among patients with a score of ≥5 points who achieved
an RVR, the SVR rate was 87%. This suggests that there
is a subset of genotype 1 patients (16–20% in the three
analysis sets had DTRP scores ≥5) that could be identified for whom dual peginterferon alfa-2a (40kD)/ribavirin therapy can achieve a comparatively high SVR rate.
The DTRP score may be particularly useful in practice settings in which PIs are accessible only to a small
proportion of patients; for example, in countries that
have universal health coverage but lack the financial
resources to pay for expensive treatments, or in countries that lack universal healthcare, in which only individuals with sufficient financial resources will have
access to PIs (20). In these settings peginterferon alfa/
ribavirin will continue to be used as a first-line treatment and PIs are likely to be used selectively. A recent
study shows how RVR can be used to identify patients
that will not benefit from the addition of a PI (21). This
study recruited non-cirrhotic genotype 1-infected
patients with a baseline HCV RNA level <600 000 IU/
ml and randomized those who achieved an RVR to
complete an additional 20 weeks of dual therapy or an
additional 24 weeks of boceprevir-based triple therapy;
SVR rates were similar in the two groups (88% vs. 90%,
respectively) (21).
In resource-limited settings, the DTRP score could be
incorporated into treatment algorithms for genotype 1
patients. All patients would receive 4 weeks of dual peginterferon alfa/ribavirin therapy after which the choice
of therapy would be based on the DTRP score and the
virological response. Patients with low DTRP scores
(0–2) would be likely to have advanced disease; this is a
group that has poor tolerability for triple therapy (22)
and is therefore not the best candidate for triple therapy.
Should such patients not achieve a >1-log drop in HCV
RNA after a 4-week lead-in phase, consideration should
be given to stopping treatment and waiting for more
effective regimens. Such a strategy presumes that the
patient understands the risks and benefits of delaying
treatment (informed deferral) and that waiting does not
put them at risk of hepatic decompensation (23).
In contrast, patients who have high DTRP scores
(≥5) and who achieve an RVR at the end of the 4-week
8
Ferenci et al.
lead-in phase with peginterferon alfa/ribavirin could be
good candidates to continue with dual therapy because
our analysis shows that 87% of such patients achieved
an SVR. Treatment can even be shortened in such
patients to 24 weeks (15). In comparison, the SVR rate
was 82% in non-black genotype 1 patients who had at
least a 1-log10 reduction in HCV RNA at the end of the
4-week lead-in phase and who then completed up to
44 weeks of treatment with boceprevir-based triple therapy in a phase III clinical trial (1).
Management of patients with intermediate scores
(score 3–4) depends on the circumstances. If cost is not
an issue, triple therapy is likely the best option; otherwise, dual therapy can be an option. Whichever therapy
is used, clinicians should adhere to the standard stopping rules, which in the case of dual therapy is less than
a 2-log drop in HCV RNA at week 12.
Using the DTRP score to identify patients likely to
respond to dual therapy would avoid the toxicity and
potential for drug interactions associated with PIs and
would prevent the selection of PI-resistant variants in
patients who do not achieve an SVR, the clinical significance of which is unknown. Using dual therapy rather
than triple therapy in some patients may also reduce
costs. For example, a cost-effectiveness analysis showed
that drug-related costs per SVR are 12–30% lower when
all patients start dual peginterferon alfa/ribavirin therapy
and switch to PI-based triple therapy only if they achieve
at least a 1-log10 drop in HCV RNA by week 4 (24).
Strengths of the DTRP score include ease of use, the
lack of necessity for a liver biopsy and the fact that it
was derived from a real-world cohort study, rather than
a homogenous group of patients enrolled in a randomized clinical trial. The score is easy to calculate because
it requires routine clinical data that are usually available
when physicians are discussing treatment options with a
given patient. The tool does not require IL28B genotype
testing, which is unlikely to be readily accessible in
resource limited settings. At present, the score can only
be applied to treatment-naive genotype 1-infected Caucasian patients. The DTRP score was developed using
data from non-US patients; thus, the utility in US-based
patients remains to be established. The reason for the
exclusion of US-based patients from the modelling exercise is that SVR rates are generally markedly different in
US-based patients than in patients who reside in other
countries.
This analysis has certain limitations because of the
retrospective nature of the analysis and because the data
are drawn from a cohort study. It is important to highlight that, although SVR rates are high in the highest
DTRP category, patients included in the current exploratory analysis received dual therapy with an intended
treatment duration of 48 weeks, which contrasts with
the shorter durations of therapy that are recommended
for patients who achieve an RVR with dual or PI-based
triple therapy. Other limitations include the use of data
only from Caucasian patients and the lack of informaLiver International (2014)
© 2013 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
Ferenci et al.
tion on host genetics. Genetic differences are a likely
cause of the lower SVR rates in US-based patients. Host
IL28B genotype is the most important pretreatment factor in determining response to interferon-based therapy
(25). Host genotype was not determined in the PROPHESYS cohorts or included in the DTRP score because its
impact on treatment outcome was not discovered until
after the trial was initiated (26). However, the percentage of patients with an IL28B CC genotype in validation
cohort 2 did not differ greatly across the DTRP categories, which suggests that the score contains predictors
that are not surrogates for IL28B genotype. Patients with
a CC genotype are more likely to achieve an RVR;
however, the presence of a non-CC genotype does not
preclude achievement of an RVR or an SVR with peginterferon alfa/ribavirin (26). The results in validation
cohort 2 suggest that, when IL28B genotype is available,
the score might be most useful in individuals with a
score of 0–2, among whom there was a large difference
(47%) between SVR rates in those with CC and non-CC
genotypes. In comparison the difference in SVR rates
between CC and non-CC patients was much smaller in
those with scores of 3–4 (21%) or ≥5 (17%). RVR is a
better predictor of SVR than IL28B genotype (27). Estimates of SVR rates in this retrospective analysis can only
be indirectly compared with those obtained with
PI-based triple therapy in randomized trials because of
the inherent limitations of cross-study comparisons. It
is only possible to put the results in perspective and it is
not possible to draw definitive conclusions.
In conclusion, the DTRP score is a simple measure
that uses readily available baseline characteristics without requiring a liver biopsy, and can be used to identify
treatment-naive genotype 1-mono-infected Caucasian
patients who have a high probability of achieving an
SVR with dual peginterferon alfa-2a (40kD)/ribavirin
therapy. The score should be validated in a prospective
study and tested and adapted for use in non-Caucasian
populations. This scoring system will be particularly
useful in practice settings where PIs are not readily
available. In addition, the DTRP score can be used to
inform treatment decisions in the light of the efficacy
and safety profile of these new therapies.
Acknowledgements
This study was supported by F. Hoffmann-La Roche
Ltd, Basel, Switzerland. Support for third-party writing
assistance for this manuscript was provided by F. Hoffmann-La Roche Ltd, Basel, Switzerland.
Conflict of interests: PF: Global Advisory Board:
Roche/Genentech, Merck, Janssen, Boehringer-Ingelheim and Rottapharm-Madaus; Advisor: GSK, Achilleon, Idenix; Speaker’s bureau: Roche, MSD Austria,
Janssen Austria, BMS Austria; Unrestricted research
grant: Roche Austria and MSD Austria. RA, IA, DD, LT
and DO: No conflict of interests; AA: Vertex: Advisory
board/ speaking; Merck: advisory board, grant support,
Liver International (2014)
© 2013 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd
Prediction of sustained virologic response
speaking; Novartis: advisory board; HC: Research Grant
and Speaker: Roche, MSD. Advisory Board: Roche; MG:
Has attended advisory committees for Janssen, Schering-Plough, Gilead and Bristol-Myers Squibb and
received speaking/teaching fees from Roche, ScheringPlough, Gilead, Bristol-Myers Squibb, Janssen and
Bayer. DL: Advisory board or participation to a HCV
protocol with: BMS, Gilead, Roche, Merk, Boerhinger
Ingelheim, Abbvie and Janssen-Cilag. MS: was an
employee of IST GmbH. IST GmbH have a contract
with Roche to provide statistical support. FT: employee
of AbbVie, Chicago IL; was an employee of F. Hoffmann-La Roche Ltd, Basel, Switzerland at the time the
study was conducted.
References
1. Poordad F, McCone J Jr, Bacon BR, et al. Boceprevir for
untreated chronic HCV genotype 1 infection. N Engl J
Med 2011; 364: 1195–206.
2. Jacobson IM, McHutchison JG, Dusheiko G, et al. Telaprevir for previously untreated chronic hepatitis C virus
infection. N Engl J Med 2011; 364: 2405–16.
3. Ghany MG, Nelson DR, Strader DB, Thomas DL, Seeff
LB. An update on treatment of genotype 1 chronic hepatitis C virus infection: 2011 practice guideline by the American Association for the Study of Liver Diseases. Hepatology
2011; 54: 1433–44.
4. FDA Drug Safety Communication: Serious skin reaction
after combination treatment with the hepatitis C drugs Incivek (telaprevir), peginterferon alfa, and ribavirin (issued
December 19, 2012). Available at http://www.fda.gov/
Drugs/DrugSafety/ucm332731.htm (accessed 21 January
2013).
5. Sarrazin C, Berg T, Cornberg M, et al. [Expert opinion on
boceprevir- and telaprevir-based triple therapies of
chronic hepatitis C]. Z Gastroenterol 2012; 50: 57–72.
6. Leroy V, Serfaty L, Bourliere M, et al. Protease inhibitorbased triple therapy in chronic hepatitis C: guidelines by
the French association for the study of the liver. Liver Int
2012; 32: 1477–92.
7. Omata M, Kanda T, Yu M-L, et al. APASL consensus
statements and management algorithms for hepatitis C
virus infection. Hepatol Int 2012; 6: 409–35.
8. Fried MW, Hadziyannis SJ, Shiffman ML, Messinger D,
Zeuzem S. Rapid virological response is the most important predictor of sustained virological response across
genotypes in patients with chronic hepatitis C virus infection. J Hepatol 2011; 55: 69–75.
9. Bronowicki J, Hezode C, Bengtsson L, et al. 100% SVR in
IL28B CC patients treated with 12 weeks of telaprevir,
peginterferon and ribavirin in the PROVE2 trial [abstract
1094]. J Hepatol 2012; 56 (Suppl. 2): S430–1.
10. Marcellin P, Cheinquer H, Curescu M, et al. High sustained virologic response rates in rapid virologic response
patients in the large real-world PROPHESYS cohort confirm results from randomized clinical trials. Hepatology
2012; 56: 2039–50.
11. Fried MW, Shiffman ML, Reddy KR, et al. Peginterferon
alfa-2a plus ribavirin for chronic hepatitis C virus infection. N Engl J Med 2002; 347: 975–82.
9
Ferenci et al.
Prediction of sustained virologic response
12. Hadziyannis SJ, Sette H Jr, Morgan TR, et al. Peginterferon-alpha2a and ribavirin combination therapy in chronic
hepatitis C: a randomized study of treatment duration and
ribavirin dose. Ann Intern Med 2004; 140: 346–55.
13. Roberts SK, Weltman MD, Crawford DH, et al. Impact of
high-dose peginterferon alfa-2A on virological response
rates in patients with hepatitis C genotype 1: a randomized
controlled trial. Hepatology 2009; 50: 1045–55.
14. Reddy KR, Shiffman ML, Rodriguez-Torres M, et al.
Induction pegylated interferon alfa-2a and high dose ribavirin do not increase SVR in heavy patients with HCV
genotype 1 and high viral loads. Gastroenterology 2010;
139: 1972–83.
15. Ferenci P, Laferl H, Scherzer TM, et al. Peginterferon alfa2a and ribavirin for 24 weeks in hepatitis C type 1 and 4
patients with rapid virological response. Gastroenterology
2008; 135: 451–8.
16. Ferenci P, Laferl H, Scherzer TM, et al. Peginterferon alfa2a/ribavirin for 48 or 72 weeks in hepatitis C genotypes 1
and 4 patients with slow virologic response. Gastroenterology 2010; 138: 503–12.
17. Scherzer TM, St€attermayer AF, Strasser M, et al. Impact of
IL28B on treatment outcome in hepatitis C virus G1/4
patients receiving response-guided therapy with peginterferon alpha-2a (40kD)/ribavirin. Hepatology 2011; 54:
1518–26.
18. Ferenci P, Aires R, Beavers KL, et al. Predictive value of
FIB-4 and APRI versus METAVIR on sustained virologic
response in genotype 1 hepatitis C patients. Hepatol Int
2014; doi:10.1007/s12072-013-9484-6 [Epub ahead of
print].
19. Ferenci P. Optimal treatment duration for patients with
HCV genotype 1 infection. J Viral Hepat 2011; 18(Suppl.
2): 7–13.
20. Ferenci P, Reddy KR. Impact of HCV protease-inhibitorbased triple therapy for chronic HCV genotype 1 infection.
Antivir Ther 2011; 16: 1187–201.
21. Pearlman BL, Ehleben C. Hepatitis C genotype 1 virus
with low viral load and rapid virologic response to peginterferon/ribavirin obviates a protease inhibitor. Hepatology
2014; 59: 71–7.
22. Hezode C, Dorival C, Zoulim F, et al. Safety of telaprevir
or boceprevir in combination with peginterferon alfa/ribavirin, in cirrhotic non responders. First results of the
French early access program (ANRS CO20-CUPIC)
[abstract 8]. J Hepatol 2012; 56(Suppl. 2): S4.
10
23. Aronsohn A, Jensen D. Informed deferral: a moral requirement for entry into the hepatitis C virus treatment warehouse. Hepatology 2012; 56: 1591–2.
24. Hunyady B, Makara M, Gervain J, et al. Treating naive
chronic hepatitis C patients with dual (peginterferon/ribavirin, SOC) or triple (SOC + protease inhibitor, PI) therapy in Hungary. An approach with limited resources
[abstract 1118]. J Hepatol 2012; 56(Suppl. 2): S440–1.
25. Ge D, Fellay J, Thompson AJ, et al. Genetic variation in
IL28B predicts hepatitis C treatment-induced viral clearance. Nature 2009; 461: 399–401.
26. Thompson AJ, Muir AJ, Sulkowski MS, et al. Interleukin28B polymorphism improves viral kinetics and is the
strongest pretreatment predictor of sustained virologic
response in genotype 1 hepatitis C virus. Gastroenterology
2010; 139: 120–9.
27. St€attermayer AF, Stauber R, Hofer H, et al. Impact of
IL28B genotype on the early and sustained virologic
response in treatment-naive patients with chronic hepatitis
C. Clin Gastroenterol Hepatol 2011; 9: 344–50.
Supporting information
Additional Supporting Information may be found in the
online version of this article:
Fig. S1. Selection of patients for this analysis. Reasons
for premature withdrawal are shown for patients
included in the analysis.
Fig. S2. Positive predictive value (95% confidence
interval) of DTRP score for SVR.
Fig. S3. SVR rates in patients with a DTRP score of
0–2, 3–4 and ≥5 points according to the extent of hepatic fibrosis at baseline. Only patients who had previously
undergone an invasive or non-invasive fibrosis assessment were included.
Fig. S4. SVR rates in patients with a DTRP score of
0–2, 3–4 and ≥5 points according to host IL28B genotype in a subset of patients in validation dataset 2, which
contains data from treatment-naive genotype 1 monoinfected Austrian patients who received response-guided
therapy for 24, 48 or 72 weeks with standard doses of
peginterferon alfa-2a (40kD)/ribavirin.
Liver International (2014)
© 2013 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd