Purpose: Chronic hyperinsulinemia may play a role in breast cancer etiology. We performed a meta-analysis examining whether serum concentrations of insulin and C-peptide are associated with increased breast cancer risk.
Methods: We restricted our analyses to prospective studies. After a systematic literature search, we computed summary relative risks (SRRs) and 95 % confidence intervals (95 % CIs) using random effect models applied to the relative risk associated with the highest versus lowest quantile of serum concentrations. We also graphically examined results in order to identify whether dose-response relationships were present.
Results: Six articles including 1,890 cases were retrieved for serum insulin levels and five for serum C-peptide levels including 1,759 cases. SRR and 95 % CI were 1.08 (0.66-1.78) for insulin and 1.04 (0.77-1.41) for C-peptide. Heterogeneity of results between studies was high for insulin and inexistent for C-peptide. Restricting the analysis to women diagnosed with breast cancer before or after menopause did not alter results. In insulin studies, SRR computed from relative risks not adjusted for body mass index (and other risk factors) was 1.22 (0.91-1.63). The SRR fell to 1.02 (0.53-1.97) in studies that adjusted for body mass index and other factors. Similar drops occurred in C-peptide studies, from 1.11 (0.87-1.41) to 1.06 (0.70-1.61). No consistent dose-response relationship was apparent in either pre- or post-menopausal cancers.
Conclusions: Our meta-analysis of observational studies found no evidence of an association between serum insulin or C-peptide concentrations and breast cancer risk. Increased risk found by some studies may have been due to inadequate control for adiposity.