Diabetic retinopathy (DR), an ocular complication of diabetes mellitus (DM), is known to have a strong genetic component independent of duration of diabetes and extent of diabetic control, but to date, very few genes have been reliably identified. Further understanding of the molecular pathogenesis is required in order to develop better strategies for prevention and treatment of DR.
Patients with type 1 or type 2 DM of at least 5 years duration were recruited through multiple Australian hospitals. A genome-wide association study was conducted using 1522 patients, 740 of whom have significant grades of DR (proliferative DR, non-proliferative DR or macular edema). Genotyping was performed on Illumina OmniExpress arrays. Population stratification was assessed with Eigenstrat and outliers removed. Association analysis was conducted in Plink.
The most significant associations were found in the subset of type 2 DM patients with proliferative DR (n=224), compared to 624 type 2 DM patients with no DR. Under univariate analysis, three genome-wide significant loci were identified; chr2, rs3738918, p=1.2x10-8; chr4, rs12501601, p=2.4x10-8; and chr8, rs7839922, p=1.2x10-9. When data was adjusted for 10 principal components, no loci reached genome-wide significance. The top ranked SNP was rs6499191 on chr16, p=5.9x10-7. The second ranked SNP was rs7839922 on chr8, p=4.6x10-6, the same SNP as in the univariate analysis. These 4 SNPs were typed in a replication cohort consisting of 90 similar cases versus 240 similar controls. The chr1 locus showed clear replication (p=0.03), but showed no evidence of association in the discovery cohort following correction for 10 principal components. Chr4 showed no evidence of replication (p=0.99) while chr 8 and 16 showed borderline (p=0.06) and significant (p=0.015) replication respectively.
These two loci harbour plausible candidate genes for involvement in proliferative DR and are under further investigation to elucidate the role of these loci in blinding diabetic eye disease.