Abstract
Background: The causal relationship between blood lipids and colorectal cancer (CRC) risk has been preliminarily explored in previous Mendelian randomization (MR) studies, but these investigations were limited to conventional or partial metabolic lipid profiles. Recent advancements in genome-wide association studies of plasma lipidomics have expanded our understanding of lipid categories, underscoring the need to evaluate the causal associations between a broader range of lipid types and CRC risk to enhance risk assessment. Methods: This MR study utilized 179 lipid phenotypes across 13 lipid classes to investigate their causal associations with CRC risk. Genetic variants significantly associated with lipid traits at the genome-wide level (P<5×10(-8)) were used as instrumental variables for MR analysis. Initial analyses were conducted using a discovery dataset (n=321,040), followed by validation in an independent replication dataset (n=185,616). Meta-analysis was then employed to determine the strength of causal evidence. The inverse-variance weighted (IVW) method and Wald ratio were the primary MR approaches, complemented by up to nine methods for multidimensional validation. Sensitivity analyses included tests for pleiotropy, heterogeneity, Steiger directionality, and Bayesian colocalization analysis, among others. Results: After Bonferroni correction and rigorous validations, 9 significant causal associations were identified. Specifically, genetically predicted levels of sterol ester (27:1/20:5) (OR(IVW) = 1.214, 95% CI 1.119-1.317), phosphatidylcholine (20:4_0:0) (OR(IVW) = 1.147, 95% CI 1.077-1.222), phosphatidylcholine (16:0_22:4) (OR(IVW) = 1.312, 95% CI 1.170-1.472), phosphatidylcholine (16:0_22:5) (OR(IVW) =1.181, 95% CI 1.093-1.277), and phosphatidylcholine (18:0_20:5) (OR(IVW) = 1.198, 95% CI 1.104-1.300) were significantly associated with an increased risk of CRC. Conversely, levels of phosphatidylcholine (18:1_20:2) (OR(IVW) = 0.832, 95% CI 0.771-0.898), phosphatidylethanolamine (18:2_0:0) (OR(IVW) = 0.804, 95% CI 0.732-0.882), phosphatidylcholine (16:0_18:0) (OR(Wald ratio) = 0.611, 95% CI 0.481-0.777), and phosphatidylcholine (O-18:1_18:2) (OR(Wald ratio) = 0.723, 95% CI 0.620-0.840) were significantly associated with a decreased risk of CRC. Colocalization analysis revealed posterior probabilities for hypothesis 4 exceeding 90%, identifying rs174546 and rs28456 as shared causal variants. Additionally, 14 suggestive causal associations were observed. Conclusions: This study establishes a causal link between specific lipid species and CRC risk. These findings suggest new avenues for CRC prevention and treatment strategies.