Abstract
Purpose :
Previous studies have suggested dyslipidemia as a potential risk factor for diabetic retinopathy (DR). However, evidence of this association has been less consistent compared to other systemic risk factors such as hyperglycemia and hypertension. This study evaluated the association between lipids and DR using a genetic approach of Mendelian randomization (MR).
Methods :
The effects of previously reported lipid-associated single nucleotide polymorphisms (SNPs) at 157 loci were tested in East Asian populations from the Asian Genetic Epidemiology Network consortium. 31 SNPs showing associations (p<0.05) with high-density lipoprotein (HDL) cholesterol, 16 with low-density lipoprotein (LDL) cholesterol and 15 with triglycerides (TG) were selected as instrumental variables (IVs) in MR analyses. We also selectively examined 9, 6 and 6 SNPs with genome-wide association with HDL, LDL and TG respectively (p<5x10-8) to increase the IV strength and thus minimise potential bias. The effect of each lipid-associated SNP on DR was assessed based on a genome-wide association study of 725 DR cases and 2141 controls from the Singapore Epidemiology of Eye Diseases study. Genotyping was performed using Illumina 610 and OmniExpress chips. DR was evaluated using a standard protocol based on retinal photographs graded according to a modified Airlie House classification system. The association between each lipid and DR was estimated by pooling individual SNP effects using fixed-effect meta-analysis.
Results :
Raised LDL levels had a protective effect on the development of DR (odds ratio [OR]=0.47, 95% CI=0.24-0.93; p=0.03) when all 16 LDL SNPs were used. However, no significant association was observed when using only 6 genome-wide associated LDL SNPs as IVs (OR=0.48, 95% CI=0.22-1.05; p=0.07). There was no association between HDL and TG with DR in both analyses using unrestricted or strong IVs. The ORs (95% CI; p-value) were 1.47 (0.91-2.38; 0.12) and 1.0 (0.61-1.63; 0.99) using unrestricted IVs and 1.48 (0.87-2.52; 0.15) and 0.96 (0.55-1.66; 0.88) using strong IVs for HDL and TG respectively.
Conclusions :
Our results showed no significant association between lipids and DR. Further studies with increased sample sizes may be needed to validate these results. We will also further explore the association between lipids and non-proliferative DR and vision-threatening DR which has been shown in previous studies.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.