Abstract MP58: Association Between Plasma Low-Density Lipoprotein Cholesterol Level and Risk of Abdominal Aortic Aneurysm in ARIC: a Mendelian Randomization Study

Abstract

Introduction: Abdominal aortic aneurysm (AAA) is one of the leading causes of death in the United States. Observational epidemiological studies demonstrated that circulating level of LDL cholesterol is positively associated with AAA risk. However, a causal link between LDL and risk of AAA is still unclear. Hypothesis: There is a causal relationship between plasma LDL and risk of AAA, which can be demonstrated by a Mendelian randomization (MR) approach. Methods: Using a gene score for LDL as an instrumental variable (IV), we conducted a large MR study to investigate the association between plasma level of LDL and risk of AAA in 8,967 European Americans in the ARIC Study. Twelve SNPs that were reported in a published genome-wide association study for LDL and not associated with other lipids or total cholesterol (TC) were selected. A weighted LDL gene score was created as the sum of the number of risk alleles of the selected SNPs multiplied by the SNP-specific effect size on LDL obtained from the literature and then divided by the total number of SNPs. For participants who took antihyperlipidemic medication, a constant was added to their plasma LDL value to account for the effect of medication. A total of 262 clinical AAA cases were ascertained through hospital discharge diagnoses or death certificates during 23 years of follow-up. The IV-predicted risk for AAA per 1 SD increment of LDL was assessed using an IV probit model adjusted for age, sex, field center, and 10 principal components to account for population stratification. A probit regression model adjusted for known risk factors and other lipids was also performed to confirm the association between LDL and AAA risk at the phenotype level. Results: Using the LDL gene score as an IV for LDL, a positive association was observed between plasma LDL and the predicted risk of AAA. Each 1 SD increase in LDL (instrumented by the LDL gene score) was associated with an average increase of 10% (95% CI=0.1%- 20%) in the risk of AAA. The strength of the LDL gene score as an IV was adequate (F-statistic = 20), and this score was not related to other risk factors for AAA, including other lipids. Similar results were observed after we removed participants who took antihyperlipidemic medications. The positive association based on the LDL gene score was concordant in direction with the association between LDL level and AAA without gene score (average increase in the risk of AAA per 1 SD increment in LDL = 0.6%, 95%CI=0.3 %-1%). We conducted a sensitivity analysis by adding SNPs related to TC. This new score was marginally correlated with triglycerides and reduced the adequacy of MR model to below statistical significance. Conclusion: Our finding supports a causal relationship between plasma level of LDL and AAA risk, with higher LDL level leading to a higher AAA risk.