Circulating bilirubin levels and risk of colorectal cancer: serological and Mendelian randomization analyses.

Background Bilirubin, a byproduct of hemoglobin breakdown and purported anti-oxidant, is thought to be cancer preventive. We conducted complementary serological and Mendelian randomization (MR) analyses to investigate whether alterations in circulating levels of bilirubin are associated with risk of colorectal cancer (CRC). We decided a priori to perform analyses separately in men and women based on suggestive evidence that associations may differ by sex. Methods In a case-control study nested in the European Prospective Investigation into Cancer and Nutrition (EPIC), pre-diagnostic unconjugated bilirubin (UCB, the main component of total bilirubin) concentrations were measured by high-performance liquid chromatography in plasma samples of 1386 CRC cases and their individually matched controls. Additionally, 115 single-nucleotide polymorphisms (SNPs) robustly associated (P < 5 x 10(-8)) with circulating total bilirubin were instrumented in a 2-sample MR to test for a potential causal effect of bilirubin on CRC risk in 52,775 CRC cases and 45,940 matched controls in the Genetics and Epidemiology of Colorectal Cancer Consortium (GECCO), the Colon Cancer Family Registry (CCFR), and the Colorectal Transdisciplinary (CORECT) study. Results The associations between circulating UCB levels and CRC risk differed by sex (P-heterogeneity = 0.008). Among men, higher levels of UCB were positively associated with CRC risk (odds ratio [OR] = 1.19, 95% confidence interval [CI] = 1.04-1.36; per 1-SD increment of log-UCB). In women, an inverse association was observed (OR = 0.86 (0.76-0.97)). In the MR analysis of the mainUGT1A1SNP (rs6431625), genetically predicted higher levels of total bilirubin were associated with a 7% increase in CRC risk in men (OR = 1.07 (1.02-1.12);P = 0.006; per 1-SD increment of total bilirubin), while there was no association in women (OR = 1.01 (0.96-1.06);P = 0.73). Raised bilirubin levels, predicted by instrumental variables excluding rs6431625, were suggestive of an inverse association with CRC ...