Abstract
The association between vitamin D and total and colorectal cancer risk was inconsistent in observational studies. We conducted Mendelian randomization approach in which the effect of confounding might be reduced. 110 single nucleotide polymorphisms (SNPs) associated with 25-hydroxyvitamin D concentrations were systematically selected according to the "GWAS Catalog" from all ethnic populations. For the SNP-vitamin D concentration association, 3978 individuals from two Japanese cohorts were included. Regarding SNP-total and colorectal cancer association, 4543 cancer cases and 14,224 controls and 7936 colorectal cancer cases and 38,042 controls, respectively were included from the Japanese Consortium of Genetic Epidemiology and other studies in Japan. There was no significant association between the genetically predicted plasma 25-hydroxyvitamin D concentration and total or colorectal cancer in any of the MR analyses. Odds ratios per doubling in vitamin D concentration were 0.83 (95% confidence interval [CI] 0.63-1.09) for total cancer and 1.00 (95% CI 0.80-1.24) for colorectal cancer in inverse variance weighted method, 0.83 (95% CI 0.57-1.19) for total cancer and 1.01 (95% CI 0.75-1.37) for colorectal cancer in MR-Egger method. Consistent with previous MR analyses among European ancestries, there was no significant association identified between 25-hydroxyvitamin D levels and total or colorectal cancer among Asians.