Plasma lipid levels and risk of retinal vascular occlusion: A genetic study using Mendelian randomization

Abstract

The causal effects of plasma lipid levels and the risk of retinal vascular occlusion (RVO) have not been clearly identified, especially for high-density lipoprotein-cholesterol (HDL-C) and low-density lipoprotein-cholesterol (LDL-C). Here, we try to identify these causal risk factors using a two-sample Mendelian randomization (MR) analysis. Single nucleotide polymorphisms (SNPs) were chosen as instrumental variables (IVs). We obtained genetic variants associated with lipid exposure at the genome-wide significance (P<5×10^-8) level from a meta-analysis of GWAS from the Global Lipids Genetics Consortium (GLGC) based on 188,577 individuals of mostly European ancestry for MR analyses. Meanwhile, we used lipid GWAS from UK Biobank (UKB) with a sample size of 115,078 individuals as a supplement. We obtained genetic predictors of RVO from a FinnGen biobank study. We conducted both univariable and multivariable MR (MVMR) analyses to identify the causal effects of RVO. Although inverse variance weighted (IVW) was the primary method used for MR analyses, MR-Egger and weighted-median methods were used as supplements to IVW. We determined the heterogeneity of IVs using Cochrane's Q test and I² , and used the MR-Egger intercept and MR-PRESSO Global test to detect horizontal pleiotropy. A leave-one-out sensitivity analysis was conducted by removing a single variant from the analysis. Genetically predicted increased HDL-C level was associated with decreased risk of RVO from GLGC [OR=0.806; 95% CI=(0.659, 0.986); P=0.036], which was consistent with UKB results [OR=0.766; 95% CI=(0.635, 0.925); P=0.005]. MVMR analysis for plasma lipids [adjusted OR=0.639; 95% CI=(0.411, 0.992); P=0.046] or diabetes [adjusted OR=0.81; 95% CI=(0.67, 0.979); P=0.029] suggested that low HDL-C may be an independent risk factor for RVO. However, there was no evidence to support a causal association between LDL-C {GLGC [adjusted OR=1.015; 95% CI=(0.408, 2.523); P=0.975], UKB [OR=1.115; 95% CI=(0.884, 1.407); P=0.359]}, total cholesterol {GLGC [adjusted OR=0.904; 95% CI=(0.307, 2.659); P=0.854], UKB [OR=1.047; 95% CI=(0.816, 1.344); P=0.716]} or triglycerides {GLGC [OR=1.103; 95% CI=(0.883, 1.378); P=0.385], UKB [OR=1.003; 95% CI=(0.827, 1.217); P=0.098]} and RVO. Using two-sample MR analysis, our study suggested that dyslipidemia was a risk factor for RVO. Furthermore, our results indicated that a low HDL-C level may be an independent risk factor for RVO, suggesting that controlling HDL-C level may be effective in RVO development.

Keywords: high-density lipoprotein-cholesterol; low-density lipoprotein-cholesterol; mendelian randomization; plasma lipid; retinal vascular occlusion.