The prospective, multicentered, randomized, double-blinded, placebo-controlled trial by Romero-Gómez and coworkers1 demonstrating that adding metformin to peginterferon and ribavirin was safe and improved insulin sensitivity in treatment-naïve patients without diabetes who are chronically infected with genotype 1 hepatitis C virus (HCV) raises a number of issues.
First, the cutoff used for insulin resistance in this study (i.e., HOMA [homeostasis model assessment] index > 2) was somewhat low, and the reasons behind this choice are not sufficiently explained by the authors. In clinical research, the decision on what cutoff points to use is crucial to give optimal enlightenment of individuals in need of pharmacological interventions because of increased insulin resistance. This choice must be made by considering the overall impact on health due to intervention (e.g., improvement of insulin resistance) and potential burden of overtreatment if a low cutoff point is employed. Second, the analysis by Romero-Gómez et al. is limited to patients with genotype 1 chronic hepatitis C. Because insulin resistance in patients with chronic hepatitis C does not differ among HCV genotypes,2 the reasons for this strict inclusion criteria is not sufficiently elucidated. Third, serum HCV RNA levels do not seem to predict insulin resistance in patients with chronic hepatitis C,2 and thus it is unclear whether the increased viral decline as seen in the metformin arm is the cause of improved HOMA values. Finally, the positive effects of metformin on sustained virological response and viral decline were confined to female patients. Notably, these findings were mainly derived from a sex-based subgroup analysis. Meaningful information from subgroup analyses within a randomized trial is restricted by low statistical power and the issue of multiple testing. In order to minimize the risk of accepting and publishing false positives, it has been proposed that the results of subgroup analysis be accepted merely as hypotheses.3 Given the risks of false positive findings when multiple subgroup analyses are done, it is not surprising that a subgroup-specific test may show a significant (P < 0.05) or suggestive (P = 0.05 to P = 0.10) effect of treatment, even when the trial failed to do so overall.3, 4 Of note, this is exactly the case of the study by Romero-Gómez et al.1 In order to support the validity of a claimed subgroup effect, there are some potential strategies, including the availability of a biological explanation (that is clearly missing in the published trial) or replication in another independent study.3 Taken together, the evidence published by Romero-Gómez et al. is likely insufficient to guide clinical care in patients with chronic hepatitis C who have insulin resistance.