SVR12 is higher than SVR24 in treatment-naïve hepatitis C genotype 1 patients treated with peginterferon plus ribavirin.

BACKGROUND: Randomized clinical trials (RCTs) of interventions for the hepatitis C virus have historically used sustained virological response (SVR) at 24 weeks after treatment (SVR24) as the key effect measure. However, recent RCTs investigating the efficacy of new direct acting agents (DAAs) have used SVR at 12 weeks after treatment (SVR12). While there is evidence to suggest SVR24 and SVR12 are similar in patients receiving new DAAs, this is unlikely to be true for patients receiving backbone peginterferon-ribavirin control treatment. Establishing the difference between SVR12 and SVR24 for patients receiving peginterferon-ribavirin treatment is therefore necessary to avoid biased interpretations of the benefits of newer DAAs.
METHODS: We searched the MEDLINE®, Embase™, and Cochrane CENTRAL for RCTs with a peginterferon-ribavirin arm that used SVR24 and/or SVR12. As no RCTs reported on both, we pooled SVR12 and SVR24 proportions using conventional meta-analysis. Proportions were pooled separately for peginterferon alpha-2a and alpha-2b. Further, a Bayesian meta-regression model was employed to estimate the difference between SVR12 and SVR24.
RESULTS: Thirty-five RCTs including a peginterferon arm were identified. Twenty-four trials included a peginterferon alpha-2a plus ribavirin arms, of which 20 reported SVR24 and five reported SVR12. Seventeen trials included a peginterferon alpha-2b plus ribavirin arm, of which 16 reported SVR24 and one reported SVR12. Using Bayesian meta-regression, the pooled SVR12 was 6% higher than SVR24 with peginterferon alpha-2a (53% versus 47%) and 5% higher with peginterferon alpha-2b (45% versus 40%) and 95% credible intervals (CrIs) were only marginally overlapping. The meta-regression also demonstrated a marginally significant relative risk of 1.13 (95% CrI 0.99-1.26) of SVR12 versus SVR24. The conventional pairwise meta-analyses were consistent with these findings.
CONCLUSION: Considering the relatively large difference observed between SVR12 and SVR24, it seems reasonable to insist that future clinical trials report both to allow for complete transparency and clarity in their interpretation.

Background

Historically, Phase II and Phase III clinical trials of hepatitis C virus (HCV) treatments have defined sustained virological response (SVR) as an undetectable HCV RNA 24 weeks after end of treatment (SVR24). This definition of SVR has been used in all key randomized clinical trials (RCTs) of peginterferon plus ribavirin, telaprevir, and boceprevir. However, due to the high efficacy of newer direct acting agents (DAAs) (eg, faldaprevir, simeprevir, and sofosbuvir), clinical trials assessing these treatments have used SVR at 12 weeks after end of treatment (SVR12). In patients receiving these newer agents, some RCTs have demonstrated nearly identical SVR12 and SVR24,1,2 and this has left trial researchers comfortable using SVR12 as the primary efficacy outcome. In addition, the impressively high SVRs observed in the recent clinical trials, along with the interferon-sparing properties of the newer agents, is surely sufficient to convince drug regulatory authorities. However, as the newer DAAs are slated to enter the market, decision-makers and clinicians will be faced with the challenge of deciding which agents are most likely to clear HCV, and for many Western countries, which agent(s) will be the most cost-effective to reimburse.

Since none of the newer agents have been compared head-to-head in RCTs, the best approach for establishing comparative efficacy between all agents is a technique commonly referred to as indirect treatment comparison (ITC) meta-analysis. ITC meta-analysis is a technique that is now widely recognized by health technology agencies worldwide. A key premise of ITC meta-analyses, however, is that all considered agents have been compared with the same control intervention using the same outcome measure. In the case of the DAAs, most newer agents have been compared with peginterferon plus ribavirin, but the efficacy of the first generation DAAs, telaprevir and boceprevir, have been established using SVR24; whereas the efficacy of the newer DAAs have predominantly been established using SVR12. Since decisions regarding the reimbursement of the newer DAAs will rely heavily on the results of ITC meta-analyses in most countries, and since the validity of ITC meta-analysis hinge on the similarity of SVR24 and SVR12 in patients receiving peginterferon plus ribavirin, it is highly important to establish whether this similarity in fact holds true. Further, since peginterferon plus ribavirin is considerably less potent than newer DAAs, and the end-of-treatment response is typically substantially higher than SVR24, there is reason to believe that SVR24 and SVR12 differ importantly in patients receiving peginterferon plus ribavirin. The extent to which SVR24 and SVR12 may differ needs to be established so decision-makers can appropriately account for the potential difference in their decision-making processes. In Table 1, we illustrate the potential bias that can occur on the relative efficacy estimates between DAAs and control peginterferon-ribavirin, when assuming that SVR12 and SVR24 are equal.

Table 1

Illustration of potential bias associated with assuming SVR12 and SVR24 are equal

One trial has compared telaprevir with peginterferon alpha-2a plus ribavirin.

The SVR24 for telaprevir 75% and the SVR24 in the control arm is 47%.

The resulting relative risk is RR =75%/47% = 1.60.

One trial has compared a new direct acting agent with peginterferon alpha-2a plus ribavirin.

The SVR12 for the new direct acting agent is 90% and the SVR12 in the control arm is 53%.

The resulting relative risk is RR =90%/53% = 1.70.

The relative risk for telaprevir and the new direct acting agent thus appear highly similar.

However, assuming second trial had also measured SVR24 and the control group SVR24 was also 47%, the resulting relative risk would be

RR =90%/47% = 1.91, which is considerably higher than 1.60.

Abbreviations: SVR12, sustained virological response at 12 weeks after treatment; SVR24, sustained virological response at 24 weeks after treatment; RR, relative risk.

Currently, no clinical trials including a peginterferon plus ribavirin arm have reported both SVR12 and SVR24. However, several trials have reported SVR24, and a number of trials investigating newer DAAs include a peginterferon plus ribavirin arm and report on SVR12. To compare SVR24 and SVR12 in patients receiving peginterferon plus ribavirin we therefore undertook a systematic review of clinical trials including a peginterferon plus ribavirin arm that reported either SVR24 or SVR12.

Methods

Trial eligibility criteria

We included RCTs involving standard doses of peginterferon alpha-2a or alpha-2b administered in combination with ribavirin for 48 weeks to treatment-naïve patients. Standard doses were determined according to EASL (European Association for the Study of the Liver) (alpha-2b 1.5 μg per kg subcutaneously once weekly, alpha-2a 180 μg subcutaneously once weekly, ribavirin total daily dose of 600–1400 mg depending on patient weight). We included clinical trials that reported SVR for HCV genotype 1. We arbitrarily allowed for combined reporting for genotype 1 and 4 if the proportion of genotype 4 patients did not exceed 10%. We only considered studies conducted in North America or Europe and excluded trials administering non-standardized doses, including patients with co-infections (eg, HIV [human immunodeficiency virus]) or comorbidities (eg, cirrhosis).

Search strategy

A search strategy was developed in consultation with a medical librarian. The included search terms were peginterferon OR peg-interferon OR pegylated interferon AND ribavirin AND hepatitis C. The search was limited to randomized trials in humans. We searched the following databases (from inception to week 31 [July 29–August 4], 2013): MEDLINE®, Embase™, and Cochrane CENTRAL. We also identified relevant published systematic reviews in our search and scanned their bibliographies for additional relevant trials. Lastly, we scanned the abstract books from the 2012 and 2013 annual meetings of AASLD (American Association for the Study of Liver Disease) and of EASL, respectively.3,4

Study selection and data extraction

Two investigators (KT and ED) scanned all abstracts and potentially eligible full text articles independently for eligibility. Disagreements were resolved by discussion with a third reviewer (EJM). Data extraction was also performed independently and in duplicate by two investigators (KT and ED). Data items included the trial region, year of publication, peginterferon type, peginterferon dose, ribavirin dose, number of patients, and number of patients achieving SVR (either 12 or 24).

Data analysis

To explore the potential differences between SVR24 and SVR12, we plotted the extracted proportions (and 95% confidence intervals) of each outcome for all eligible clinical trials using forest plots. Forest plots were constructed separately for peginterferon alpha-2a and alpha-2b as these two peginterferon treatments have previously been demonstrated to yield different SVR proportions.5–7 We also pooled the proportions across trials using a DerSimonian-Laird random-effects model.8 This was done by pooling logit transformed proportions to preserve symmetry and normality of the mean. To assess combinability of the retrieved proportions, we estimated the degree of heterogeneity in each pairwise meta-analysis using the I2 measure for heterogeneity.9 To estimate whether SVR12 and SVR24 were different, we performed a Bayesian random-effects meta-regression of SVR proportions (and associated 95% credible intervals [CrIs]).10 (Note that the code used for the Bayesian analyses is provided in the Supplementary material.) This Bayesian regression model also concomitantly controlled for type of peginterferon. All analyses were conducted using WinBUGS version 1.4.3 (WinBUGS, Cambridge, UK).11

Results

A total of 35 clinical trials were eligible and included in the analyses.12–45 Twenty-four trials included a peginterferon alpha-2a plus ribavirin arms,12–33 of which 20 reported SVR2412–28 and five reported SVR1229–33 (one reported on both29). Seventeen trials include a peginterferon alpha-2b plus ribavirin arm,31,34–45 of which 16 reported SVR2434–45 and one reported SVR12.31 The characteristics of all clinical trials and peginterferon treatments are presented in Table 2. Figure 1 shows a schematic of the trial selection process.

Table 2

Characteristics of the additional single-arm peginterferon alpha-2a and alpha-2b data included in the analyses

Trial	No of patients	Genotype	Experimental intervention	Peginterferon arm	SVR12	SVR24
Fried et al12	298	1	Peginterferon alpha-2a	Alpha-2a	–	46.3%
Hadziyannis et al13	271	1	Peginterferon alpha-2a	Alpha-2a	–	52.0%
Ferenci et al14	95	1	Peginterferon alpha-2a	Alpha-2a	–	51.6%
Diago et al15	475	1	Peginterferon alpha-2a	Alpha-2a	–	47.8%
von Wagner et al16	352	1	Peginterferon alpha-2a	Alpha-2a	–	52.8%
Zeuzem et al17	114	1	Peginterferon alpha-2a	Alpha-2a	–	57.9%
Hezode et al18	82	1	Peginterferon alpha-2a	Alpha-2a	–	46.3%
Roberts et al19	438	1	Peginterferon alpha-2a	Alpha-2a	–	50.0%
Marcellin et al20	212	1	Peginterferon alpha-2a	Alpha-2a	–	43.9%
Zeuzem et al21	441	1	Peginterferon alpha-2a	Alpha-2a	–	51.0%
Ascione et al22	93	1,4	Both peginterferons	Alpha-2a	–	54.8%
Mchutchison et al23	1,035	1	Both peginterferons	Alpha-2a	–	40.9%
Rumi et al24	91	1	Both peginterferons	Alpha-2a	–	48.4%
Yenice et al25	40	1	Both peginterferons	Alpha-2a	–	45.0%
McHutchison et al26	75	1	Telaprevir	Alpha-2a	–	41.3%
Jacobson et al27	361	1	Telaprevir	Alpha-2a	–	43.8%
Bronowicki et al28	11	1	Asunaprevir	Alpha-2a	–	45.4%
Bronowicki et al29	53	1	Asunaprevir	Alpha-2a	52.8%	45.0%
Sulkowski et al30	71	1	Faldaprevir	Alpha-2a	56.3%	–
Manns et al31	50	1	Simeprevir	Alpha-2a	62.0%	–
Jacobson et al32	132	1	Simeprevir	Alpha-2a	50.0%	–
Lawitz et al33	26	1	Sofosbuvir	Alpha-2a	57.7%	–
Benhamou et al34	226	1	Peginterferon alpha-2b	Alpha-2b	–	41.6%
Berg et al35	225	1	Peginterferon alpha-2b	Alpha-2b	–	48.0%
Brady et al36	311	1,4	Peginterferon alpha-2b	Alpha-2b	–	29.6%
Buti et al37	86	1	Peginterferon alpha-2b	Alpha-2b	–	43.0%
Jacobson et al38	1,313	1	Peginterferon alpha-2b	Alpha-2b	–	38.7%
Kumada et al39	31	1	Peginterferon alpha-2b	Alpha-2b	–	49.2%
Poordad et al40	70	1	Peginterferon alpha-2b	Alpha-2b	–	27.1%
Scotto et al41	26	1	Peginterferon alpha-2b	Alpha-2b	–	50.0%
Shiffman et al42	48	1	Peginterferon alpha-2b	Alpha-2b	–	29.2%
Sjogren et al43	29	1	Peginterferon alpha-2b	Alpha-2b	–	41.4%
Ascione et al22	93	1,4	Both peginterferons	Alpha-2b	–	39.8%
McHutchison et al23	1,019	1	Both peginterferons	Alpha-2b	–	39.8%
Rumi et al24	87	1	Both peginterferons	Alpha-2b	–	32.2%
Yenice et al25	40	1	Both peginterferons	Alpha-2b	–	32.5%
Kwo et al44	104	1	Boceprevir	Alpha-2b	–	37.5%
Poordad et al45	363	1	Boceprevir	Alpha-2b	–	37.8%
Manns et al31	80	1	Simeprevir	Alpha-2b	42.5%	–

Abbreviations: SVR12, sustained virological response at 12 weeks after treatment; SVR24, sustained virological response at 24 weeks after treatment.

Figure 1

Study flow diagram.

Abbreviations: peginf, peginterferon; SVR12, sustained virological response at 12 weeks after treatment; SVR24, sustained virological response at 24 weeks after treatment.

Table 3 presents the pooled SVR12 and SVR24 proportions associated with peginterferon alpha-2a and peginterferon alpha-2b under the conventional DerSimonian-Laird pairwise meta-analysis and the Bayesian meta-regression. Figure 2 displays the forest plot of SVR24 and SVR12 proportions associated with peginterferon alpha-2a arms, as well as the pooled SVR24 and SVR12 proportions for these subgroups. The pooled SVR12 was 6% higher than SVR24 with peginterferon alpha-2a, and 95% confidence intervals were not overlapping. The degree of heterogeneity in both meta-analyses was I2=0% in the meta-analysis of SVR12 proportions and I2=65% in the meta-analysis of SVR24 proportions. The pooled SVR12 was 3% higher than SVR24 with peginterferon alpha-2b in the conventional meta-analysis and 5% higher in the Bayesian meta-regression. In the conventional meta-analysis where all data were analyzed separately, the 95% confidence intervals were overlapping. However, in the Bayesian meta-regression where all data were analyzed simultaneously and the relative risk between SVR12 and SVR24 was assumed equal for the two peginterferons, the 95% CrIs were only slightly overlapping. From the Bayesian meta-regression the relative risk between SVR12 and SVR24 (assumed equal for the two peginterferons) was RR =1.13 (95% CrI 0.99–1.26). That is, the Bayesian meta-regression provided marginally statistically significant evidence that the SVR12 is, on average, 13% relatively higher than SVR24.

Table 3

The pooled SVR12 and SVR24 proportions associated with peginterferon alpha-2a plus ribavirin and peginterferon alpha-2b plus ribavirin

SVR time point	Peginterferon type	Conventional	Bayesian
		Meta-analysis	Meta-regression
SVR12	Alpha-2a	54% (49%–59%)	53% (46%–59%)
SVR24	Alpha-2a	49% (46%–51%)	47% (45%–49%)
SVR12	Alpha-2b	43% (33%–53%)	45% (39%–53%)
SVR24	Alpha-2b	40% (38%–41%)	40% (38%–43%)

Notes: Proportions obtained with conventional pairwise meta-analysis of proportions, and with Bayesian meta-regression. The 95% confidence intervals and credible intervals are presented in parenthesis for the conventional meta-analysis and Bayesian meta-regression, respectively.

Abbreviations: SVR, sustained virological response; SVR12, SVR at 12 weeks after treatment; SVR24, SVR at 24 weeks after treatment.

Figure 2

Forest plots of RCT peginterferon alpha-2a arms informing SVR12 (A) and SVR24 (B) proportions.

Abbreviations: RCT, randomized controlled trial; SVR12, sustained virological response at 12 weeks after treatment; SVR24, sustained virological response at 24 weeks after treatment.

Discussion

Our findings indicate that SVR12 is higher than SVR24 in HCV genotype 1 patients treated with peginterferon plus ribavirin. This difference holds true for both peginterferon alpha-2a and peginterferon alpha-2b. These findings do not come as a surprise as it has long been known that end-of-treatment response with peginterferon-ribavirin treatment is larger than SVR24, and that the loss of viral response in patients after end of treatment occurs gradually during the 24-week follow-up period.

Our findings have implications for the interpretation of current and future RCTs comparing new DAAs with backbone peginterferon-ribavirin alone. In particular, newer direct acting agents may erroneously be interpreted as only exhibiting moderate benefit over backbone peginterferon-ribavirin treatment if SVR is assessed at 12 weeks after treatment and the assumption is made that SVR12 and SVR24 are identical. ITC meta-analysis and network meta-analysis informing comparative efficacy of newer DAAs may also become biased if a mixture of SVR12 and SVR24 control responses are used in the analysis.

There are strengths and limitations to consider in our analysis. Strengths include the systematic inclusion of all relevant clinical trials allowing for a well informed analysis. Further, our analysis was designed specifically for this project using a Bayesian approach that recognizes differences in effect sizes across arms, thereby allowing greater precision to detect differences. Weaknesses of our study include the fact that publication bias inherently exists as the trials have been completed but do not report on both 12- and 24-week SVR. Newer DAAs in particular have had their trials completed but are, as yet, unavailable in published form. Lastly, our results are based on separated single-arm evidence, and while the considerable amount of data adds to the reliability of our results, confirmations of our findings will be required from randomized clinical trials that report both SVR12 and SVR24. In this vein, we should note that our findings stand in contrast to a recent retrospective observational study by Martinot-Peignoux which found a close to 100% agreement between SVR12 and SVR24 in patients receiving either of the two peginterferons.46 This study, however, included several genotype 2 and 3 patients as well and previous relapsers and non-responders, and so, the populations are not fully comparable. The SVR24 in the study by Martinot-Peignoux was 71%, in contrast to the 35%–55% in all the RCTs included in our analysis. The study by Martinot-Peignoux used a highly sensitive assay with a lower level of detection of 9.6 IU/mL, and one can thus speculate that use of older less sensitive assays (eg, Copas Amplicor with 50 IU/mL lower level of detection) may be the cause of differences between SVR12 and SVR24. However, a subgroup analysis by RCTs using highly sensitive assays (defined as <20 IU/mL) versus “moderately” sensitive assays (actualized as 50 IU/mL or higher, although defined as >20 IU/mL) showed nearly identical SVR24 (results not shown). All trials reporting SVR12 used highly sensitive assays. In fact, the pooled proportions of highly sensitive assay SVR24 estimates were further from the SVR12 estimates than the pooled proportions of the “moderately” sensitive assays. Considering that in the only RCT that reported both SVRs,29 the SVR12 was 51.8% and the SVR24 was 45.0%; and considering that our meta-analytic evidence is based on a large set of homogenous trials including a total of over 10,000 patients, it seems reasonable to suggest that limited confidence should be placed on the findings of the study by Martinot-Peignoux.

Considering the relatively large difference observed between SVR12 and SVR24, it seems reasonable to insist that future clinical trials report both outcome measures to allow for complete transparency and clarity in their interpretation.

Supplementary material

WinBUGS code and data for the Bayesian meta-regression.