Meta-analysis: Ribavirin Plus Interferon vs. Interferon Monotherapy for Chronic Hepatitic C – an Updated Cochrane Review
Authors and Disclosures
Posted: 12/08/2010; Alimentary Pharmacology & Therapeutics. 2010;32(7):840-850. © 2010 Blackwell Publishing
Discussion Only
The present review suggests that the rationale for performing a number of the recent randomized trials on ribavirin plus interferon vs. interferon alone for chronic hepatitis C is debatable. Several short-term trials on patients who were treatment naïve, relapsers and nonresponders have been conducted after a convincing benefit was found of adding ribavirin on sustained virological response. By contrast, our review shows that more evidence from long-term trials is needed regarding the effects of ribavirin plus interferon on clinical outcomes.
Strengths and Weaknesses
We did a comprehensive search of several databases and tried to minimize publication bias and language bias by including trials regardless of publication status and language. Furthermore, two authors independently assessed trials and data to avoid errors. We may not have identified all randomized trials and we were unable to obtain data from some trials. Accordingly, the strength of the combined evidence may be underestimated.
The included patients and intervention regimens varied considerably, which likely causes some of the substantial inter-trial heterogeneity. This heterogeneity can be considered both as a weakness and strength. Limiting our analysis to include trials on one specific intervention regimen or subgroup of patients would have provided a more focused answer. However, the variety of intervention regimens and patients increases the external validity of the results and we adjusted our analysis for heterogeneity.[17]
We based our trial sequential analysis information size estimation on event proportions in the control group from previous trials on interferon monotherapy[24] and on estimated relative risk reduction (10% for sustained virological response and 50% for morbidity and mortality) plus type 1 error of 5% and type 2 error of 20%. Thus, our findings are appropriate only under these assumptions. It could be judged that other prespecified intervention effects or type 1 and 2 errors are more appropriate for this clinical situation. However, in the analysis for sustained virological response, realistic alteration in the parameters of effect size and error rates would not have changed the overall conclusion. Regarding morbidity and mortality, the use of relative risk reduction of 50% may seem too optimistic. A choice of smaller effect size would have made the analysis even more restrictive and lead to the same conclusion: more evidence is needed regarding clinical outcomes.
Trial Sequential Analysis on Sustained Virological Response
Adding ribavirin to interferon was recommended by consensus in Europe in 1999, and in the US in 2002.[6, 27] We found that after 1998, several redundant randomized trials have been conducted on adding ribavirin to interferon reporting the outcome-sustained virological response (Figure 3). The results of the 50 subsequent trials, which enrolled additional 7349 patients since 1998, did not affect the effect size, but simply narrowed the CI and reduced the P-value. We acknowledge that trial sequential analysis was performed according to publication date and some trials were initiated before the accumulated evidence (or consensus recommendations) was available. Thus, some of the trials that seem redundant by our analysis would have been appropriate at the time of their launch. Furthermore, we acknowledge that it could be argued that it was fair to assess the effect of ribavirin in some subgroups. Having said this, we still find that adding ribavirin to interferon does increase the sustained virological response in all relevant subgroups and, in retrospect, many trials have been conducted since the time at which this association was statistically convincing (Figures 4 and 5).
Trial Sequential Analysis on Clinical Outcomes
According to our 'traditional' cumulative meta-analysis, there is some evidence (P = 0.03) of benefits on clinical outcomes when adding ribavirin to interferon. However, this result is not adjusted for the risks of random error and multiple testing. According to trial sequential analysis, it may be a spurious result and we need more evidence to detect (or reject) a beneficial effect on clinical outcomes (Figure 6). This could be obtained either by conducting new trials with a sample size equalling the difference between the accrued number of patients and the calculated desired information size (n = 3812 patients; Figure 6) or by performing long-term evaluation of patients enrolled in the previous trials. Potential new trials should have long-term follow-up to obtain protracted clinical events, but are logistically very difficult and dropouts are likely to occur. Such trials may also be considered unethical as controls are randomized to a treatment (monotherapy) assumed inferior to that currently recommended.[6, 28] Long-term evaluation of patients enrolled in the previous trials may, however, be invalid because these patients may have been retreated with different antiviral regimens because of participation in new trials or as part of clinical practice.
We contacted all trial authors and asked whether any long-term clinical outcomes had been assessed subsequent to the trial publication. Most authors did not reply, many informed that it was impossible to obtain valid data, but a few authors provided long-term data or were planning to do so. Hopefully, these additional clinical data will provide valuable information, within a reasonable time. Although a beneficial effect of ribavirin added to interferon will become established, we need to remember that ribavirin is associated with several severe adverse reactions, it is expensive and the number needed to treat to prevent one patient from developing liver-related morbidity is likely to be large.[11, 12] Such considerations may sometimes be ignored when adding ribavirin as standard care and conflict current recommendations.[5]
Sustained Virological Response as a Surrogate Outcome
Sustained serological and virological response is considered the target for chronic hepatitis treatment, but it is still only a putative surrogate outcome.[8, 9, 29] Some criteria that influence the validity of a surrogate outcome are as follows: first, that the surrogate is biological plausible; second, if a correlation is established between the surrogate and the clinical outcome, e.g. in observational studies; third, if one or, even better, many randomized trials on similar drug classes have established that the intervention's effect on the surrogate captures the effect on the pertinent clinical outcome.[30]
Interestingly, the mechanism of ribavirin is unclear but the rationale for using ribavirin is biologically plausible and epidemiological studies have showed a correlation between sustained virological response and liver-related morbidity.[4] Unfortunately, the third criterion has never been confirmed, i.e. randomized trials or meta-analyses have never established, for any antiviral intervention, that sustained virological response captures a beneficial effect on liver-related morbidity or mortality. We may therefore risk that the intervention causes a positive response on a virological response but has no beneficial effect or even harmful effect on clinical outcomes.
Most of the included trials were partly or fully sponsored by for-profit organizations, which significantly influence research on hepatitis. These organizations endorse surrogate outcomes as they provide more events, the events appear faster and are easier to assess. The trial process by using, e.g. sustained virological response is thereby shortened, speeding up the implementation of antiviral interventions. However, currently the area of hepatitis C clinical research seems weakened with too many redundant small trials with short-term follow-up on potentially useless surrogate outcomes.
Review Complete Data @ Medscape
No comments:
Post a Comment