We thank the authors for their interest in our article1 on treatment of insulin resistance in patients with hepatitis C virus (HCV) infection. The authors make several points that require addressing. First, “the choice of a cutoff for HOMA > 2 could be somewhat low”. We elected this cutoff based on our previous results2 showing that a homeostasis model assessment (HOMA) score > 2 had a significant, and independent, negative impact on the sustained virological response (SVR) rate. These results encouraged us to include this subpopulation in the present study. Second, “the inclusion was limited to patients with genotype 1” infection. Currently, there is a need to increase therapeutic efficacy in patients infected with genotype 1. They represent the most prevalent genotype, and in which we had demonstrated a negative impact of baseline insulin resistance on SVR.2 Studies addressing the impact of treating insulin resistance in other genotypes are urgently required. Third, “the relationship between insulin resistance and HCV replication is controversial”. In our study, viral decline during the first 12 weeks was higher in females receiving metformin, and was related to better SVR rate. These findings support the hypothesis that metformin works better in females by improving antiviral activity of peginterferon plus ribavirin. Fourth, “sex-based subgroup analysis”. Despite sex-based subgroup analysis having not been an “a priori” issue in the study plan, the distribution of the variables included in the model would probably not change even when increasing the sample size and, as such, the significant difference induced by metformin would be maintained. The statistical power of analysis of the cohort of females is shown in Fig. 1. (nQuery Advisor 4.0; Statistical Solutions, Saugus, MA). Indeed, women were not underrepresented in this trial, because there was no significant differences in sex distribution in the studied population sample which, in all respects, reflects the overall distribution of hepatitis C prevalence by sex in Spain.3 The impact of sex on SVR was evaluated using the chi-squared test in the univariate analysis and in the multivariate logistic regression analysis. Interactions need to be studied between independent variables, but lack meaning if applied to an independent versus a dependent variable.4
Although our trial was not designed to evaluate potential sex differences, a potential type-2 error plus the power analysis and the doubled SVR in females receiving metformin prompted our hypothesis that metformin improves antiviral efficacy of treatment with peginterferon alfa-2a plus ribavirin. We do agree, however, that confirmatory studies of this novel finding are warranted.