Proliferative diabetic retinopathy and center-involving diabetic macular edema display strong genetic associations with common variants in Asian populations

Abstract

Purpose : To identify genetic variants associated with diabetic retinopathy (DR) and center-involving diabetic macular edema (ciDME) assessed by optical coherence tomography (OCT) in South Indian patients with type 2 diabetes (T2D).

Methods : T2D patients recruited at Sankara Nethralaya in South India had both eyes imaged with fundus photographs, which were graded for DR. A subset of patients was imaged with structural OCT. DNA was genotyped on the Expanded Multi-Ethnic Genotyping Array (MEGA^EX) array, and imputation to 1000 Genomes was performed. Genome-wide association analyses were performed using generalized linear mixed models (GLMM) in R using the GENESIS package. For the analyses, we compared patients with and without DR, with and without proliferative DR (PDR), and with and without ciDME. All analyses were adjusted for HbA1c and duration of diabetes. Replication was performed in a Singapore cohort with fixed effects meta-analysis.

Results : In the discovery cohort, 2261 patients were included in the DR (n=737) vs. no DR (n=1524) analysis, and there were no genome-wide significant associations. For the PDR outcome, we compared cases with PDR (n=289) to controls without DR (n=1524). The A allele of rs11199996 in an intergenic region on chromosome 10 was associated with PDR at a genome-wide significant level [Odds Ratio (OR)=4.1, P=2.88 X 10^-8]. In the Singapore replication cohort (PDR cases=55 and controls without DR=767), the effect estimate was in the same direction (OR=2.0, P=0.08) and meta-analysis yielded a final OR of 3.4 (P=1.88 X 10^-8). This same variant was also genome-wide significant when we compared cases with PDR (n=289) to controls without PDR (e.g., those with nonproliferative DR or no DR, n=1909), with an OR=3.0 in meta-analysis with the Singapore replication cohort (P=4.39 X 10^-8, n=55 PDR cases and n=937 controls without PDR). 385 participants had OCT scans gradable for ciDME. The C allele of rs10169007 in SGPP2 was associated with ciDME (OR = 0.02, P = 1.47 X 10^-8). Replication for ciDME was not available as the Singapore cohort did not have OCT data.

Conclusions : A genetic variant on chromosome 10 was significantly associated with PDR in a meta-analysis including both South and East Asians. In addition, a genetic variant in SGPP2 was associated with the endophenotype of ciDME.

This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.