| dc.description.abstract | The goal of this dissertation was to utilize techniques to understand the genetic architecture of diabetic retinopathy with three aims: 1) develop a phenotyping algorithm to identify diabetic retinopathy cases in electronic health records (EHR), 2) examine the associations between genetic variants and diabetic retinopathy, and 3) validate results from genetic predictors of diabetic retinopathy. By developing algorithms to identify individuals with diabetes, with and without diabetic retinopathy, we show the potential of leveraging existing EHR systems to advance our understanding of etiology of diabetic retinopathy. We demonstrate the EHR-based algorithms to be portable and accurate across separate EHR systems and this represents a vast improvement over existing high-throughput method such as PheWAS. We used summary statistics from a previous GWAS along with results from the MVP, BioVU, UKBB, and the MGBB to evaluate genetic associations with diabetic retinopathy. We identified nine novel loci, localizing to MFSD4A, TENM2, GRHL2, TCF7L2, FBXW8, COL4A2, SLC16A3, TMPRSS6, and G6PD, that either increase or decrease risk for diabetic retinopathy. Additionally, there was a high genetic correlation between African- and European-ancestry meta-analyses, even though some loci only appeared in a single ancestry. Validation of our findings raise several practical and clinical implications for management of diabetes. Given the common occurrence of G6PD deficiency and its impact on diabetic complications, screening African Americans with diabetes for this deficiency may be of high priority. If screening African Americans for this variant is cost-prohibitive, then concurrently checking blood glucose levels along with HbA1c could be another strategy. Our estimates suggest that with comprehensive genetic screening of African Americans and subsequent standard-of-care treatment, possibly aimed at glucose rather than HbA1c targets, ~12% of diabetic retinopathy cases could be avoided in U.S. African Americans alone. Based on the evidence from this study, we suggest that implementation of this screening would mitigate disparity of prevalence and severity for other diabetes sequelae among African Americans | |
