LAUSR

The associations between blood lipids, including total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), triglycerides, and low-density lipoprotein cholesterol (LDL-C), and colorectal cancer risk are controversial. We evaluated potential causal relationships between blood lipids and colorectal cancer risk. Using the baseline data from the Japanese Consortium of Genetic Epidemiology studies, we estimated the single-nucleotide polymorphism (SNP)-exposure associations (n=34,546 for TC, n=50,290 for HDL-C, n=51,307 for triglycerides, and n=30,305 for LDL-C). We also estimated the SNP-outcome associations in another Japanese dataset (n=7,936 colorectal cancer cases and n=38,042 controls). We conducted Mendelian randomization analyses for the association between each blood lipid type and the risk of colorectal cancer using an inverse variance-weighted method. The total variances explained by the selected SNPs in TC (68 SNPs), HDL-C (50 SNPs), log-transformed triglycerides (26 SNPs), and LDL-C (35 SNPs) were 7.0%, 10.0%, 6.2%, and 5.7%, respectively. The odds ratios for colorectal cancer were 1.15 (95% confidence interval 1.01-1.32) per 1 standard deviation (SD) (33.3 mg/dL) increase in TC, 1.11 (0.98-1.26) per 1 SD (15.4 mg/dL) increase in HDL-C, 1.06 (0.90-1.26) per 1 SD (0.5 log-mg/dL) increase in log-transformed triglycerides, and 1.17 (0.91-1.50) per 1 SD (29.6 mg/dL) increase in LDL-C. Sensitivity analyses consistently suggested the positive association between TC and colorectal cancer, whereas results of each lipid component were inconsistent. In conclusion, this large Mendelian randomization study of a Japanese population showed a potentially causal association between high TC and colorectal cancer risk, although the association between each lipid component and colorectal cancer remained inconclusive.