Response-guided peg-interferon plus ribavirin treatment duration in chronic hepatitis C: Meta-analyses of randomized, controlled trials and implications for the future

Authors


  • Potential conflict of interest: Dr. Sanchez-Tapias is a consultant for and is on the speakers' bureau of Novartis. He is in the speakers' bureau of and received grants from Schering-Plough and Roche. He is a consultant for and received grants from Bristol-Myers Squibb. He is also in the speakers' bureau of Gilead. Dr. Ferenci advises, is on the speakers' bureau of, and received grants from Roche. Dr. Mangia received grants from Merck and Roche. She is a consultant for Janssen. Dr. Jensen is a consultant for Globeimmune. He is a consultant for and advises Abbott Labs. He advises Roche, Merck, Pfizer, Pharmasset, Tibotec, Vertex, Bristol-Myers Squibb, Boehringer/Ingelheim, and Genentech. Dr. Cervoni received grants from Schering-Plough. Dr. Buti is on the speakers' bureau of Merck. Dr. Martino is a consultant for Merck and Gilead.

  • This meta-analysis was not supported by any pharmaceutical company or governmental agency.

Abstract

Response-guided pegylated interferon (peg-IFN) plus ribavirin (P/R) therapy trials on genotype (G)1 and G2/G3 hepatitis C virus–infected patients provide contradictory results. We conducted meta-analyses of randomized, controlled trials to address (1) the benefit of a 72-week extended-duration therapy in G1-slow responders and (2) adequate shortened duration therapy in G1 and G2/G3-rapid responders. Seventeen trials were selected, including 624 G1 rapid responders, 570 G1 slow responders, and 2,062 G2/G3 rapid responders. Virologic outcomes and treatment discontinuation data were collected from published articles and by asking investigators. Pooled estimates of sustained virologic response (SVR), relapse, and dropouts were calculated using the random effects model, considering the variability of shortened duration, ribavirin dose, genotype, and baseline viral load. In G1 slow responders, a 72-week extended duration increased SVR (+10.7%; 95% CI [confidence interval]: +4.4% to + 17.1%), decreased relapse (−12.3%; 95% CI: −25.4% to 0%), and did not significantly increase drop-out rates (+4.5%; 95% CI: −0.6% to + 9.6%). The benefit of extended duration was lower when using a weight-based ribavirin regimen (+8.7%; 95% CI: +1.7% to + 15.8%). In G1 rapid responders, a 24-week shortened duration decreased SVR (−12.5%; 95% CI: −19.2% to −5.8%) and increased relapse rates (+8.8%; 95% CI: +2.9% to + 14.8%). Such differences were not significant in patients with baseline viral load <400,000 UL/mL (−4.4%; 95% CI: −9.8% to + 1%). In G2/G3 rapid responders, SVR was more common for standard 24-week duration than for shortened durations (+4.1%; 95% CI: +0.1% to + 8.5), but this benefit was not significant when ribavirin was weight-adjusted and the short duration was 16 weeks (−1.7%; 95% CI: −6.1% to + 2.7%) and for G2 patients (+1.6%; 95% CI: −0.2% to + 5.5%). Conclusion: Long durations of P/R therapy improve SVR, regardless of genotype. This effect is nonetheless negligible in rapid responders, with the most favorable conditions for SVR (G2, G1 with low viral load, and G3 with weight-adjusted ribavirin regimen). (HEPATOLOGY 2011;)

The standard of care for chronic hepatitis C is the combination of peg-interferon plus ribavirin that leads to hepatitis C virus (HCV) eradication in 60% of cases, with variations depending on the characteristics of host and virus and, in particular, viral genotype.1,2 Patients infected with genotype 1 (G1) are the most difficult to cure: they achieve sustained virologic response (SVR) in under 50% of cases in pivotal studies.1, 2 This explains the considerable recent efforts to develop new antiviral molecules specific to G1 strains, including boceprevir3 and telaprevir.4 However, despite their virologic efficacy, concerns are raised regarding the induction of viral cross-resistance,5 which could impede long-term viral elimination. This leaves the systematic use of these molecules as first-line treatments in naïve G1 patients open to debate, particularly in patients with genetic susceptibility for SVR with standard treatment, such as the CC polymorphism of the interleukin-28B (IL-28B) gene.6 Moreover, no similar therapeutic advances have been made for HCV genotype 2 (G2) and 3 (G3) strains, which are reputed to be the most sensitive to standard treatment, with SVR rates of 80%-90%. With standard pegylated interferon (peg-IFN) plus ribavirin therapy, the concept of response-guided duration therapy (i.e., adjusting treatment duration according to the outcome of HCV viral load under treatment) has been widely investigated, particularly in two situations. In the 70% of G2/G3 patients and the 15% of G1 patients who develop a rapid virological response (RVR), as defined by undetectable HCV viral load at week 4, the aim was to reduce treatment duration to lower the risk of adverse events and cost of treatment. G2 and G3 patients were thus tested for receiving a 12-16-week treatment duration versus the standard 24-week therapeutic regimen, whereas G1 patients were tested for receiving a 24-week duration versus the standard 48-week regimen. The second situation involved G1 patients who develop slow virologic response, as defined by a decrease in viral load of over 2 log between baseline and week 12, followed by subsequent HCV RNA loss before week 24. In these patients, the aim was to extend treatment duration beyond the standard 48 weeks to increase the probability of SVR. In the two cases mentioned (i.e., decrease or increase of treatment duration), controlled trials produce contradictory results. Initial studies showed that the increase in treatment duration in slow responders helped increase the probability of SVR,7, 8 but two larger scale studies published recently9, 10 were negative and controversial. In G2/G3 rapid responders, the trials were also contradictory. The largest trial (ACCELARATE study11) recommended maintaining treatment for a 24-week period, whereas all other trials suggested that treatment duration could be reduced, particularly in G2 patients or in G3 patients when the initial viral load was low.12-17 In G1 rapid responders, it was reported that treatment duration could be reduced to 24 weeks in the event of low baseline viral load, without apparent controversy.18-22 Nonetheless, because few patients were included in the randomized trials, which focused on this comparison, the conclusion of nonsignificant difference in SVR rate between the short arm and the standard duration arm remains questionable.

Therefore, the most effective method of administering peg-IFN and ribavirin therapy still remains unknown. This issue must be resolved to specify the therapeutic gain expected with future therapeutic regimens, because the concept of response-guided therapy will probably apply with the new antiviral agents. We thus decided to explore these controversial issues by a systematic review with meta-analyses, evaluating as thoroughly as possible the effects of modulating the duration of peg-IFN plus ribavirin therapy on SVR and relapse rates in G2 and G3 rapid responders, G1 rapid responders, and G1 slow responders.

Abbreviations

95% CI, 95% confidence interval; cEVR, complete early virologic response; G1, genotype 1; G2/G3, genotype 2 and 3; HCV, hepatitis C virus; IFN, Interferon; IL-28B, interleukin-28B; ITT, intention-to-treat method LVL, low viral load; PCR, polymerase chain reaction; peg-IFN, pegylated interferon; P/R, peg-interferon plus ribavirin; RCTs, randomized, controlled clinical trials; RVR, rapid virologic response; SRs, slow responders; SVR, sustained virologic response.

Patients and Methods

Meta-analyses were conducted according to a predetermined protocol, following the recommendations of Sacks et al.23

Study Objective.

The primary end-point was to identify effects of different durations of peg-IFN plus ribavirin combination therapy on SVR. Secondary end-points were relapse rate (as defined by percentage of patients with detectable HCV viral load within 6 months of the end of treatment) and safety, as assessed by percentage of treatment discontinuations resulting from adverse events.

Literature Search.

Studies were retrieved using MEDLINE, the Cochrane library database, and manual searches. The key words, “chronic hepatitis C,” “pegylated interferon,” and “peg-interferon,” were used as free text words and/or combined with “randomized controlled clinical trials (RCTs)” and “clinical trials.” General reviews, meta-analyses, and references from published RCTs and presented to the International Association for the Study of the Liver, the European Association for the Study of the Liver, and the American Association for the Study of Liver Diseases meetings from 2004 to 2010, were also searched manually for additional citations.

Criteria for Inclusion.

We selected RCTs published as full articles in English in peer-reviewed journals and RCTs published as abstracts in landmark international, European, or American meetings since 2004, comparing different treatment durations in naïve patients with chronic hepatitis C. All included RCTs evaluated peg-IFN alpha 2a or alpha 2b and ribavirin combination therapy. The intervention tested was treatment duration, which should have been determined as a fixed duration from the onset of treatment or at week 4. For studies comparing extended (i.e., 72 weeks) versus standard (i.e., 48 weeks) duration in G1 slow responders, the outcome of virologic response at 12 weeks and 24 weeks was mandatory. For studies comparing short (i.e., 24 weeks) versus standard (i.e., 48 weeks) duration in G1, or short versus standard (i.e., 24 weeks) duration in G2 or G3 patients with rapid virologic response, HCV RNA might be undetectable at week 4.

Criteria for Exclusion.

Nonrandomized trials, trials that included patients who had already received antiviral therapy, and trials that did not use peg-IFN and ribavirin combination therapy in all the studied arms were excluded. RCTs were also excluded when the rapid virologic responders and the slow responders could not be identified (i.e., when the outcome of virologic response at week 4, week 12, and week 24 was not available). In the event of unpublished virologic data, a request for information was made to investigators before excluding the trials from this report.

Range of Patient Characteristics, Diagnoses, and Treatments.

The following items were recorded as potentially useful in assessing clinical heterogeneity between RCTs: baseline viral load, HCV genotype, type of peg-IFN therapy (i.e., alpha 2a or alpha 2b), and ribavirin regimen (i.e., fixed dose or weight-adjusted dose).

Selection and Data-Extraction Bias.

Three reviewers first screened titles and abstracts, yielding 25 potentially eligible publications. All RCTs considered for inclusion were analyzed independently by the three reviewers, who conferred with the others in cases of disagreement. Reviewers contacted investigators for clarification, when necessary. The decision on inclusion or exclusion was not related to the results or conclusions of each manuscript.

Statistical Analysis.

All analyses were performed with the intention-to-treat (ITT) method; they included all randomized patients, and patients without the endpoint were considered as a treatment failure. When not given in the publication, the response rate according to the ITT method was recalculated. We used the rate ratio (relative risk) and the risk difference for SVR, relapse, and dropout between the different durations tested in the RCTs. We used the DerSimonian and Laird model for random-effects meta-analysis to obtain summary estimates across studies. We tested for heterogeneity by using the Cochran Q test, which follows a chi-square distribution, to make inferences about the null hypothesis of homogeneity (considered significant at P < 0.10). This report follows the PRISMA guidelines24 and the Cochrane collaboration guideline for reporting meta-analyses.

Results

Effect of Extended 72-Week Duration in G1 Slow Responders.

Eight trials including naïve G1 patients comparing extended versus standard duration of peg-IFN plus ribavirin combination therapy were considered for this meta-analysis.7-10, 19, 21, 25-27 One trial was excluded because the longest extended treatment duration was 68 weeks.25 The DITTO-HCV study was also excluded because virologic outcome was not provided by the investigators.27 The six other trials published as full articles fulfilled the inclusion criteria. In two trials, patients received a fixed 800-mg/day ribavirin regimen,7, 8 whereas in the four other trials, patients received a weight-based ribavirin regimen. The study design was variable regarding the randomization procedure at baseline, week 4, week 12, week 24, or week 36. Several studies reported the results of G1 and G4 patients together, so we restricted our report to G1 patients alone. The main characteristics of the selected trials are shown in Table 1. Of a total of 3,599 G1-naïve patients treated by peg-IFN plus ribavirin, 567 slow-responder patients were randomized to receive 48 versus 72 weeks of combination therapy. Slow virologic response was defined by detectable HCV RNA at week 12 despite a log drop in viral load of more than 2 from baseline and undetectable HCV RNA at week 24.

Table 1. Description of the Selected Trials
Trials comparing 48 vs. 72 weeks treatment duration in genotype 1 patients
Studies (Ref)Peg-IFN alphaRibavirin regimen (mg/d)GenotypePCR limit of detection (IU/mL)Time of RandomisationNo. of randomised patientsNo. of slow responders (SR) (%)No. of patients with cEVR at week 12
Berg 2006 (7)2a800150baseline45597 (21)163 (36)
Sanchez-Tapias 2006 (8)2a800150week 437156 (15)132 (36)
Pearlman 2007 (26)2b800-1400110week 24361112 (31)ND
Mangia 2008 (21)2a or 2b1000-1200150baseline69674 (11)265 (38)
Ferenci 2010 (10)*2a1000-1200150week 1228969 (15)162 (36)
Buti 2010 (9)2b800 -1400130week 361427159 (11)589 (41)
Trials comparing 24 vs. 48 weeks treatment duration in genotype 1 patients who achieved RVR
Studies (Ref)Peg-IFN alphaRibavirin regimen (mg/d)GenotypePCR limit of detection (IU/mL)Time of RandomisationNo. of randomised patientsNo. of patients with RVR (%)No. of patients with RVR and LVL (<40000 IU/mL)
Jensen 2006 (19)2a1000-1200150post-hoc analysis38988 (23)49
Mangia 2008 (21)2a or 2b1000-1200150baseline696185 (27)69
Yu 2008 (22)2a1000-1200125baseline20087 (43)52
Liu 2008 (20)2a1000-1200150baseline308201 (65)91
Berg 2009 (18)2b800-140015.3baseline43363 (15)ND
Trials comparing 24 weeks vs. shortened treatment duration in genotype 2 or 3 patients who achieved RVR
Studies (Ref)Peg-IFN alphaRibavirin regimen (mg/d)GenotypePCR limit of detection (IU/mL)Time of RandomisationNo. of randomised patientsNo. of patients with RVR (%)shortened regimen duration
  1. Data restricted to genotype 1.

  2. *IFN, interferon; RVR, rapid virologic response; SR, slow responders; cEVR, complete early virologic response.

Mecenate 2007 (15)2a800-12002-3600Week 4180120 (67)12 weeks
Lagging 2008 (13)2a8002-3600baseline382231 (61)12 weeks
Dalgard 2008 (12)2b800-14002-350Week 4428302 (71)14 weeks
Mangia 2005 (14)2b1000-12002-350baseline283178 (63)12 weeks
Von Wagner 2005 (16)2a800-12002-3600Week 4153142 (93)16 weeks
Yu 2007 (17)2a1000-1200250baseline150130 (87)16 weeks
Shiffman 2007 (11)2a8002-350baseline1426959 (67)16 weeks

Pooled Analysis in Genotype 1 Slow Responders.

Data from clinical trials comparing SVR after 48 or 72 weeks of combination therapy in G1 patients with slow virologic response are shown in Table 2. The 72-week extended duration of peg-IFN plus ribavirin therapy was associated with a significant increase in the rate of SVR, compared with the standard 48-week duration of therapy (39.4% versus 30.3%; risk ratio: 1.40; 95% CI: 1.11-1.77; P = 0.003). The increase in SVR rate associated with extended duration was 10.7% (weight-adjusted risk difference; 95% CI: +4.4% to +17.1%; P = 0.0009). This benefit was observed in trials that used a fixed 800-mg/day ribavirin regimen with an increase in SVR rate of 19.6% (95% CI: +4.8% to +34.3%; P = 0.009) and also in trials using a weight-based ribavirin regimen. The increase in SVR associated with extended duration was then 8.7% (95% CI: +1.7% to +15.8%; P = 0.014). This analysis is summarized in Fig. 1A. Rate of relapse was lower in the group treated for 72 weeks (16.0% versus 33.0%; risk ratio: 0.54; 95% CI: 0.37-0.77; P = 0.008). The weight-adjusted risk difference was –12.3% (−25.4% to 0%; P = 0.005). Rate of dropouts was not statistically different between the extended duration and the standard duration groups, despite a trend toward higher dropouts in the extended-duration group (16.6% versus 10.4%; risk ratio: 1.46; 95% CI: 0.98-2.19; P = 0.065). The weight-adjusted risk difference was +4.5% (95% CI: −0.6% to +9.6%; P = 0.082). The SUCCESS study9 was associated with the higher risk of dropout with the extended duration, although it was not significant (risk ratio: 2.18; 95% CI: 0.62-7.71).

Table 2. Meta-Analytical Data of the Selected Trials
Trials comparing 48 vs. 72 weeks treatment duration in genotype 1 slow respondersMeta-analyses for SVR
Studies (Ref) No. of patients with SVR/SR (%)No. of patients with relapse/SRNo. of patients with drop out/SR% of SVR increase with 72 weeks regimen (95%CI of risk difference)Random effect weight for risk-ratio (%)
48 weeks72 weeks48 weeks72 weeks48 weeks72 weeks
Berg 2006 (7) 15/45 (33)25/52 (48)NANA11/4521/5214 (−5 to 34)20.3
Sanchez-Tapias 2006 (8) 5/31 (16)11/25 (44)NANA3/313/2527 (4 to 50)6.8
Pearlman 2007 (26) 9/49 (18)20/52 (38)13/495/527/498/5220 (3 to 37)11.5
Mangia 2008 (21) 0/21 (0)4/53 (7)1/216/532/217/536 (−4 to 16)0.6
Ferenci 2010 (10) 15/35 (43)19/34 (56)17/357/342/353/3413 (−10 to 36)21.4
Buti 2010 (9) 37/86 (43)35/73 (48)32/8616/733/866/735 (−11 to 220)39.3
Trials comparing 24 vs. 48 weeks treatment duration in genotype 1 patients who achieved RVRMeta-analyses for SVR
Studies (Ref)SubgroupNo. of patients with SVR/RVR (%)No. of patients with relapse/RVRNo. of patients with drop out/RVR% of SVR increase with 48 weeks regimen (95%CI of risk difference)Random effect weight for risk-ratio (%)
24 weeks48 weeks24 weeks48weeks24 weeks48weeks
Jensen 2006 (19)all viral load29/33 (91)50/55 (88)3/331/551/334/553 (−10 to 17)17.6
LVL24/27 (89)21/22 (95)NANANANA6 (−10 to 22)10.9
Mangia 2008 (21)all viral load95/123 (77)54/62 (87)22/1236/626/1232/6210 (−2 to 21)21.4
LVL38/45 (84)20/24 (83)NANANANA−2 (−20 to 17)6.8
Yu 2008 (22)all viral load40/45 (89)42/42 (100)5/450/420/450/4211 (1 to 21)27.9
LVL27/28 (96)24/24 (100)NANANANA3 (−6 to 13)8.9
Liu 2008 (20)all viral load79/104 (76)95/97 (98)NANANANA22 (13 to 30)27.5
LVL46/49 (94)42/42 (100)NANANANA6 (−2 to 14)10.5
Berg 2009 (18)all viral load12/15 (80)13/14 (93)NANANANA12 (−13 to 37)5.6
Trials comparing 24 weeks vs. shortened treatment duration in genotype 2 or 3 patients who achieved RVRMeta-analyses for SVR
Studies (Ref)SubgroupNo. of patients with SVR/RVR (%)No. of patients with relapse/RVRNo. of patients with drop out/RVR% of SVR increase with 24 weeks regimen (95%CI of risk difference)Random effect weight for risk-ratio (%)
shortened regimen24 weeksshortened regimen24 weeksshortened regimen24 weeks
  1. NA, data not available on published articles and after queries to investigators; SVR, sustained virologic response; RVR, rapid virologic response; SR, slow responder; LVL, low viral load (<400,000 IU/mL). G2: genotype 2; G3: genotype 3.

Mecenate 2007 (15)G2 or G344/60 (74)50/60 (84)2/601/60NANA9.8 (−4.8 to 24.5)8.0
Lagging 2008 (13)G2 or G385/120 (71)101/111 (91)31/1204/1111/1206/11120.0 (10.2 to 29.7)12.4
Dalgard 2008 (12)G2 or G3120/148 (81)136/150 (91)16/1488/1509/1484/1509.5 (1.7 to 17.4)15.9
G227/29 (93)30/31 (97)NANANANA3.7 (−8.7 to 15.9)7.9
G393/110 (84)106/115 (92)NANANANA7.6 (−0.8 to 16.0)12.9
Mangia 2005 (14)G2 or G3113/133 (85)41/45 (91)13/1331/451/1332/455.5 (−5.0 to 16.0)13.3
G289/102 (87)31/35 (89)NANANANA0.6 (−12.0 to 13.2)6.8
G324/31 (77)10/10 (100)NANANANA18.9 (−0.3 to 38.1)2.9
Von Wagner 2005 (16)G2 or G358/71 (82)57/71 (80)9/713/713/716/71−1.4 (−14.3 to 11.5)9.9
G218/19 (95)18/19 (95)NANANANA0 (−16.3 to 16.3)4.7
G339/51 (76)39/52 (75)NANANANA−1.4 (−17.9 to 15.1)3.2
Yu 2007 (17)G2 or G343/43 (100)85/87 (98)0/432/870/430/87−1.7 (−6.4 to 3.0)20.8
G243/43 (100)85/87 (98)0/432/870/430/87−1.7 (−6.4 to 3.0)29.8
Shiffman 2007 (11)G2 or G3387/489 (79)400/470 (85)NANA19/48923/4706.0 (1.1 to 10.8)19.7
G2200/257 (78)210/247 (85)NANANANA7.2 (0.4 to 13.9)16.0
 G3185/230 (80)187/219 (85)NANANANA4.9 (−2.0 to 11.9)15.8
Figure 1.

Effect of extended (72 weeks) versus standard (48 weeks) treatment duration on SVR in G1-naïve patients. (A) Meta-analysis on slow responders according to ribavirin regimen. (B) Meta-analysis on patients who developed complete early virologic response (cEVR).

Comparison With Early Responders.

Of the clinical trials considered for this systematic review, four compared SVR with 48 and 72 weeks of combination therapy in G1 patients who achieved complete early virologic response (cEVR), as defined by undetectable HCV RNA at week 12.7, 8, 10, 21 Two studies could not be exploited. In the study by Pearlman et al., only results from slow-responder patients were available.26 In the SUCCESS study including 1,427 G1 patients, 813 patients had undetectable HCV RNA at week 12 (224 of them already had undetectable HCV RNA at week 4). All of them were treated for 48 weeks, as only slow-responder patients were randomized to compare extended versus standard duration of treatment.9

Pooled analysis did not show any significant benefit for 72 weeks of treatment duration versus 48 weeks for patients with undetectable HCV RNA at week 12: SVR rate was no different between the extended-duration versus the standard-duration groups (69.3% versus 64.5%; risk-ratio: 1.06; 95% CI: 0.95-1.18; not significant). The weight-adjusted risk difference was +4.4% (95% CI: −3.1% to +11.8%; not significant). Forest plots are shown in Fig. 1B.

Effect of Shortened 24-Week Duration in Genotype 1 Rapid Responders.

In patients with RVR, the outcome of a 24-week shortened duration was assessed in nine trials. Two trials were excluded because they were nonrandomized.28, 29 One additional trial was excluded because its definition of RVR did not meet the usual criteria.27 Another was not considered because ribavirin was given at a fixed dose of 800 mg per day, instead of a weight-based regimen.30 Five trials18-22 fulfilled the inclusion criteria. The study of Jensen et al.19 involved post-hoc analysis of data collected during a randomized, multinational, phase III study.31 The four other studies18, 20-22 were designed to optimize treatment duration according to virologic outcome. The five trials included 2,026 G1 patients who received a weight-based ribavirin regimen and were randomized to receive 24 versus 48 weeks of combination therapy. The main characteristics of the selected trials and the meta-analytical data are shown in Table 1.

Pooled Analysis Irrespective of Baseline HCV Viral Load.

Of the 2,026 G1 patients treated with peg-IFN and weight-based ribavirin regimen, 624 (31%) patients achieved RVR, but only 590 patients were tested for receiving 48 versus 24 weeks of combination therapy. Of these patients, 48 weeks of therapy was associated with a significantly higher rate of SVR, compared with 24 weeks of therapy (94.1% versus 79.7%; risk ratio: 1.15; 95% CI: 1.07-1.24; P < 0.0001), with a weight-adjusted risk difference of +12.5% (95% CI: +5.8% to +19.2%; P < 0.0001; Table 2). Forest plots are shown in Fig. 2A. Rate of relapse was lower in the group treated for 48 weeks (4.4% versus 14.9%; risk ratio: 0.45; 95% CI: 0.22-0.93; P = 0.031). The weight-adjusted risk difference was –8.8% (95% CI: −14.8% to −2.9%; P = 0.004). Rate of dropouts could be calculated in 360 patients from three studies.17, 19, 21 Three patients (3.6%) dropped out. Dropout rate was similar in the group treated for 48 weeks (3.8% versus 3.5%; risk ratio: 1.06; 95% CI: 0.36-3.11; not significant). The weight-adjusted risk difference was +0.9% (95% CI: −3.3% to +3.5%; not significant).

Figure 2.

Comparison of SVR in standard (48 weeks) versus shortened (24 weeks) treatment duration in G1 rapid virologic responders. (A) Meta-analysis irrespective of baseline viral load. (B) Meta-analysis restricted to patients with baseline viral load <400,000 IU/mL.

Pooled Analysis in Patients With Low Viral Load.

Four of the five selected trials19-22 provided the comparison of SVR rates in G1 rapid virologic responders, according to baseline viral load. This comparison was directly available in three published articles20-22 and recorded by calling the investigator of one additional study.19 Of the 590 patients with RVR, 261 patients had a low viral baseline load defined as less than 400,000 IU/mL. Meta-analytical data are shown in Table 2. For patients with RVR and low baseline viral load, rates of SVR were not statistically different when comparing 24 and 48 weeks of therapy, despite a trend toward better results in the 48-week group (95.5% versus 90.6%; risk ratio: 1.05; 95% CI: 0.99-1.11; not significant; the weight-adjusted increase in SVR associated with 48 weeks was 4.4%; 95% CI: −1.0% to +9.8%; not significant). Forest plots are shown in Fig. 2B. A sensitivity analysis showed that a significant difference would have been observed if 110 additional G1 patients with RVR and low viral load (LVL) had been included in the trials.

Effect of Shortened Duration in G2 and G3 Rapid Responders.

Nine trials including G2 and/or G3 patients compared shortened versus standard duration of peg-IFN plus ribavirin combination therapy and were considered for this meta-analysis. The REDD 2/3 trial32 was excluded because data on rapid virologic responders were not available. One additional study33 was also excluded because 34% of the included patients had received a previous course of antiviral therapy and individual data on naïve patients were not available. Six trials fulfilled the inclusion criteria, involving 3,002 patients, including 2,062 who developed an RVR. The main characteristics of the selected trials are shown in Table 1. The shortened duration of treatment was 12 weeks in three studies, 14 weeks in one study, and 16 weeks in three studies. Ribavirin dose ranged between 800 and 1400 mg/day according to body weight in five studies, whereas the two others used an 800-mg/day ribavirin regimen irrespective of body weight.

Pooled Analyses.

Of the 2,062 rapid virologic responders considered for the meta-analysis, 1,720 (83.4%) achieved SVR. Overall, the standard 24-week duration of peg-IFN plus ribavirin therapy was significantly associated with a higher rate of SVR (87.5% versus 79.9%; risk ratio: 1.08; 95% CI: 1.01-1.15; P = 0.004) with a weight-adjusted risk difference of +6.4% (95% CI: +0.9% to +12.0%; P < 0.001). However, this analysis showed significant heterogeneity (Cochran Q = 19.68; P = 0.0032). We thus conducted a sensitivity analysis by removing the study by Lagging et al.,13 which showed the greatest difference between the two groups (see Table 2). This second analysis solved the problem of heterogeneity (Cochran Q = 9.33, P = 0.11) and did not change the overall results: The rate of SVR was still higher in the standard-duration group (87.1% versus 81.0%; risk ratio: 1.05; 95% CI: 1.00-1.11; P = 0.039), with a weight-adjusted risk difference of +4.1% (95% CI: 0.1% to +8.5%; P = 0.020).

Rate of relapse could be studied in only six of the seven trials: The ACCELERATE study11 did not provide rate of relapse among rapid responders, and the investigators did not reply to our query. Rate of relapse was lower in the standard-duration group (3.6% versus 12.3%; risk ratio: 0.35; 95% CI: 0.21-0.61; P < 0.0001), with a weight-adjusted risk difference of –6.6% (95% CI: −12.7% to −0.4%; P = 0.001). Rate of dropouts could be studied in all the trials published as full articles. It was no different between the standard 24-week duration and the shortened-duration groups (4.5% versus 3.3%; risk ratio: 1.41; 95% CI: 0.78-2.53; not significant). The weight-adjusted risk difference for dropouts was +1.1% (95% CI: −0.9% to +3.2%; not significant).

Sensitivity Analysis According to Short Duration and Ribavirin Regimen.

Because trials were heterogeneous regarding duration in the short arm (12, 14, or 16 weeks) and the ribavirin regimen (fixed dose of 800 mg/day or weight-based ribavirin regimen, i.e., 800-1,200 mg/day), we separated the trials into two categories. The first included the four trials in which the shortened duration was 12 or 14 weeks and/or the ribavirin regimen was a fixed dose of 800 mg/day. Those trials were called trials with a “suboptimal short arm.” They included 1,559 RVR patients. Of these, 1,291 (82.8%) achieved SVR. The second category included the two trials designed with a shortened duration of 16 weeks and a weight-adjusted ribavirin regimen. These trials were called trials with an “optimal short arm.” They included 272 RVR patients. Of these, 243 (89.3%) achieved SVR. The results of the meta-analysis were different in the two categories of trials. In trials with a “suboptimal short arm,” the standard 24-week duration was associated with higher SVR rates (86.4% versus 80.0%; risk ratio: 1.09; 95% CI: 1.04-1.14; P < 0.001). The weight-adjusted risk difference for SVR was +6.9% (95% CI: +3.2% to +10.6%; P < 0.001). In trials with an “optimal short arm,” SVR rates were similar in the standard-duration and shortened-duration arms (89.9% versus 88.6%; risk-ratio: 0.98; 95% CI: 0.94-1.03; not significant). The weight-adjusted risk difference was –1.7% (95% CI: −6.1% to +2.7%; not significant), without any trend toward higher SVR rate in the standard-duration arm. Forest plots are shown in Fig. 3A.

Figure 3.

Comparison of SVR in standard (24 weeks) versus shortened treatment duration in G2/G3 rapid virologic responders. (A) Meta-analysis according to short duration and ribavirin regimen. Studies with an optimal short arm were designed with both a 16-week short duration and a weight-based ribavirin regimen, whereas studies with a suboptimal short arm were designed with either a short duration of 12 or 14 weeks or a fixed 800-mg/day ribavirin dose. (B) Meta-analysis according to genotype.

Sensitivity Analysis by Genotype.

A sensitivity analysis by genotype (G2 or G3) was conducted in four of the six trials for which this data were available. This included 739 G2 rapid virologic responders and 843 G3 rapid virologic responders. Forest plots are shown in Fig. 3B.

SVR was achieved in 623 (84.3%) G2 rapid virologic responders, with no significant difference between standard (89.3%) or shortened (83.8%) duration: The risk ratio was 1.02 (95% CI: 0.97-1.06; not significant), and the weight-adjusted risk difference was +1.6% (95% CI: −0.2% to +5.6%; not significant). However, a trend toward better SVR rates was observed with standard treatment duration in G2 patients included in trials using a suboptimal short arm (86.6% versus 81.4%; risk ratio: 1.06; 95% CI: 0.99-1.13; P = 0.059). The weight-adjusted risk difference was +5.3% (95% CI: 0% to +10.7%; P = 0.052). Conversely, no benefit was observed with standard duration in G2 patients from the two trials with an optimal short arm (weight-adjusted risk difference: −1.6%; 95% CI: −6.1% to +2.9%; not significant).

SVR was achieved in 683 (81%) G3 rapid virologic responders and was more frequent in cases of standard duration, compared with shortened duration (86.4% versus 76.3%; risk ratio: 1.08; 95% CI: 1.01-1.14; P = 0.014). The weight-adjusted risk difference was +6.2% (95% CI: 1.3% to +11.1%; P = 0.014). Similarly to that observed in G2 patients, the benefit of standard duration was only observed in G3 patients included in trials using a suboptimal short arm (88.1% versus 81.4%; risk ratio: 1.08; 95% CI: 1.02-1.15; P = 0.038), conversely to that observed in the study by Von Wagner et al.16 (Table 2). The weight-adjusted risk difference was +6.9% (95% CI: 1.8% to +11.1%; P = 0.032).

Discussion

This meta-analysis comparing the duration of peg-IFN–ribavirin treatment in hepatitis C leads to three main conclusions: (1) It is beneficial to pursue treatment for 72 weeks in G1 slow responders; (2) in G1 rapid responders, treatment must be maintained for 48 weeks when the viral load is high, whereas a slight decrease in SVR rate is observed for a 24-week duration when the initial viral load is lower than 400,000 mIU/L, but is not significant; and a (3) a reduction in treatment duration does not lower the chances of curing G2 and G3 rapid responders, as long as the duration is at least 16 weeks and the ribavirin dose is weight-adjusted.

Through data gathering, the results of the different trials were homogenized to identify comparable populations and early virologic events (response at week 4, week 12, and week 24). The only persistent heterogeneity was the viral-load positivity threshold, which lowered over time as a result of improvements in molecular biology techniques (Table 1). However, these differences had little effect on our results. Another important point was that individual data and/or answers to our queries could have been obtained from the investigators for the majority of the trials, providing accurate comparisons of virologic outcomes and safety profiles. Such feedback was not necessary for trials reported in detail and was not a condition for including the trials in the meta-analyses if there was sufficient information, despite no answer from the investigator on specific points.7, 11

The results for G1 slow responders encourage treatment to be continued for 72 weeks. These results may be considered as obsolete, as G1 patients should undergo triple-drug therapy containing specific protease inhibitors in the near future. Nonetheless, considering the issues of toxicity and concerns regarding the induction of long-term viral resistance with these drugs,5 a substantial proportion of patients may continue to receive a standard bitherapy in the future. These results are, therefore, still useful. In this population, the positive impact of extending treatment duration was obvious when patients received only 800 mg of ribavarin per day7, 8 and was limited when ribavirin dose was weight-adjusted (Fig. 1A). This suggests that extending treatment duration is particularly useful when the drug has a moderate antiviral effect. The more recent trials that produced negative results may also have taken other parameters influencing response to treatment into account, such as insulin resistance,34 that may explain the lack of therapeutic benefit obtained by a single increase of treatment duration. Our meta-analysis could not explore all these factors, as measures associated with antiviral treatment were not specified in the trials. Another important point concerned treatment discontinuations that negatively affected ITT results, particularly in the extended-duration arm of the SUCCESS study.9 The majority of these dropouts were not related to severe adverse events, but to the patients' wish to discontinue treatment. The delayed randomization (at week 36) may have discouraged the patients from completing this trial, especially when they were randomly assigned to the extended-duration group (8.2% versus 1.2%).9 Nevertheless, increasing treatment duration at 72 weeks did not significantly increase dropout rate related to severe adverse events, as demonstrated in our meta-analysis.

In G1 rapid virologic responders, we observed that the different trials comparing treatment durations included a small number of patients (Table 1). This was particularly true for patients with an initial viral load of under 400,000 IU/mL. For these patients, the meta-analysis showed that the rate of SVR was only 4% higher when the duration was maintained at 48 weeks, but the difference was nonsignificant. This was the consequence of a type 2 error resulting from only 110 missing patients. This can be illustrated by making a comparison with G2/G3 patients: A decrease in SVR rate of similar magnitude was, indeed, observed in the ACCELERATE trial in the 16-week treated arm, instead of the standard 24 weeks. However, in this study, this difference was significant as the result of the high number of patients included. We, therefore, acknowledge that G1 rapid virologic responders should continue to receive 48 weeks, regardless of the baseline viral load. However, in the event of low baseline viral load, individual patient considerations, such as cost or side effects, could support the case for 24 weeks of peg-IFN and ribavirin therapy, in view of the modest increment in SVR to be gained with a therapy duration of 48 weeks. The role of protease inhibitors in this population remains to be defined, as the phase III boceprevir trial with a lead-in phase (SPRINT-2) showed no increase in SVR in naïve G1 rapid virologic responders who received triple combination therapy versus standard treatment.35

The results for G2 and G3 provided by this meta-analysis are the most useful for future clinical practice, as no phase III trial using new molecules has yet been carried out on this population. Only the question of reducing treatment duration in rapid responders has been thoroughly explored and remains controversial: The ACCELERATE study contradicts previous trials, but stands as the reference trial because of the number of patients studied and the quality of the methodology. It is, nonetheless, interesting to observe that the results of our meta-analysis include this trial, but contradict its findings on some key points. The significant differences obtained in this trial were so small that the clinical significance of its statistically significant differences must be challenged. This question was particularly relevant in the subpopulation of G2 rapid virologic responders, as already mentioned36: The difference in SVR obtained with a 24-week duration was limited in the ACCELERATE study and was no longer significant in our meta-analysis. However, our most interesting conclusion is that the effect of duration fades when antiviral treatment is more intense. This is shown by the difference obtained when analyzing the subgroup of patients receiving a weight-adjusted ribavirin regimen, instead of a fixed dose of 800 mg/day. When patients received 16 weeks of treatment with weight-adjusted ribavirin, no trend toward better SVR rates in the 24-week arm was seen (Fig. 3A). A nonsignificant difference was also observed in a post-hoc analysis of the ACCELERATE study focusing on patients <65 kg body weight (i.e., in which the 800-mg/day ribavirin dose was adequate37.) We assume that G2/G3 rapid virologic responders must, therefore, receive a sufficient dose of ribavirin as far as tolerance allows, which is more important than maintaining duration at 24 weeks.

Our study does not answer all questions on optimizing classic bitherapy in hepatitis C. The first question involves taking the viral load into account at week 8. Mangia's study shows that this could determine the probability of a SVR perhaps more effectively than the response at week 12.21 Other trials did not study this issue, so we could not examine it. The second question concerns the administration of ribavirin. Recent studies suggest that plasma concentration, erythrocyte ribavirin, or drop in hemoglobin under ribavirine38 correlated with ribavirin concentration, are predictive factors for SVR. No trial in our meta-analysis involved pharmacological follow-up of ribavirin or a sequential measurement of hemoglobin, or even an identical procedure for a decrease in ribavirin or for erythropoietin administration in the event of adverse effects. This element was, therefore, not studied. Furthermore, treatment observance in all trials included were unknown. The third point relates to the population. Predictive genetic factors of virologic response, such as IL-28β6 or equilibrative nucleoside transporter 139 gene polymorphisms, could not be studied, as their predictive value was not known when the trials were started. However, because all studies considered were randomized, controlled trials, such genetic factors may have accounted for differences in SVR rates between trials, rather than between the studied arms themselves. Last, we did not carry out sensitivity analyses restricted to groups of patients according to their age or fibrosis stage because of the large amount of missing data. The majority of the studies considered in the present meta-analysis, particularly those focused on shortened treatment durations, included a small number of patients with extensive fibrosis or cirrhosis. The extrapolation of our conclusions to these specific patients thus remains questionable.

In conclusion, our meta-analysis shows that adjusting peg-IFN–ribavirin treatment duration is a significant therapeutic option. Increasing treatment duration limits the risk of relapse in all populations, especially when there are unfavorable conditions for antiviral treatment efficacy, either because of the treatment itself (non-weight-adjusted ribavirin dose) or because the population is difficult to cure (G1 slow responders). It may be reasonable to propose a 16-week treatment duration for G2 and G3 patients who receive weight-adjusted ribavirin regimen.

Ancillary