Hyperinsulinemia, insulin resistance and colorectal adenomas: A meta-analysis

Background. Insulin stimulates cell proliferation and inhibits apoptosis. While epidemiologic studies have investigated associations between markers of insulin resistance/hyperinsulinemia (i.e., circulating insulin, homeostasis model assessment-insulin resistance (HOMA-IR), C-peptide) and risk of colorectal adenoma (CRA), the effect size has not yet been quantified. Objective. We aimed to summarize the association between hyperinsulinemia/insulin resistance and risk of CRA, including whether the association is independent of adiposity. Materials and Methods. Pubmed and Embase were searched through April, 2015 to identify observational studies investigating the associations between insulin, C-peptide and HOMA-IR and CRA risk. Using the highest versus lowest category meta-analysis and dose response meta-analysis based on a random-effects model, we estimated summary odds ratio (OR) and the corresponding 95% confidence interval (95% CI). Results. A total of 27 studies (insulin: 16 studies including 14,007 cases; C-peptide: 11 studies including 8639 cases; HOMA-IR: 8 studies including 11,619 cases) were included in this meta-analysis. The summary ORs of CRA comparing the highest with the lowest quantile were 1.33 for insulin (95% CI = 1.12-1.58, 12 = 73.9%, P-heterogeneity < 0.001), 1.44 for C-peptide (95% CI = 1.13-1.83, I-2 = 63.5%, P-heterogeneity = 0.003), and 1.33 for HOMA-IR (95% CI = 1.10-1.60, I-2 = 69.1%, P-heterogeneity = 0.004). Upon stratification by ethnicity, higher levels of insulin and C-peptide were significantly associated with increased risk of CRA in non-Asian ethnicity (summary OR for insulin = 1.67 [95% CI = 1.28-2.17], I-2 = 34.9%, P-heterogeneity = 0.16; summary OR for C-peptide = 1.59 [95% CI = 1.22-2.08], I-2 = 21.5%, P-heterogeneity = 0.27) but not in Asians (summary OR for insulin = 1.10 [95% CI = 0.92-1.33], I-2 = 76.6%, P-heterogeneity = 0.001; summary OR for C-peptide = 1.27 [95% CI = 0.84-1.91], I-2 = 72.6, P-heterogeneity = 0.01). We observed evidence for the existence of publication bias for insulin (P = 0.01 by Egger test) and HOMA-IR (P = 0.05 by Egger test). The results were confirmed in linear dose-response meta-analysis. These significant positive associations generally persisted even after adjustment for adiposity, although the effect size was substantially attenuated. Conclusions. Independent of adiposity, higher levels of insulin, C-peptide, and HOMA-IR were significantly associated with increased risk of CPA. The weaker associations and high heterogeneity in Asian studies warrant further research. These results indicate that insulin resistance and hyperinsulinemia may contribute in part to the association between obesity and CPA risk (C) 2015 Elsevier Inc. All rights reserved.