Identification of Genetic Variants for Diabetic Retinopathy Risk Applying Exome Sequencing in Extreme Phenotypes

Abstract

Background: Diabetic retinopathy (DR) risk has been shown to vary depending on ethnic backgrounds, and thus, it is worthy that underrepresented populations are analyzed for the potential identification of DR-associated genetic variants. We conducted a case-control study for the identification of DR-risk variants in Mexican population.

Methods: We ascertained 60 type 2 diabetes mellitus (T2DM) patients. Cases (n = 30) were patients with advanced proliferative DR (PDR) with less than 15 years after a T2DM diagnosis while controls (n = 30) were patients with no DR 15 years after the diagnosis of T2DM. Exome sequencing was performed in all patients, and the frequency of rare variants was compared. In addition, the frequency of variants occurring in a set of 169 DR-associated genes were compared.

Results: Statistically significant differences were identified for rare missense and splice variants and for rare splice variants occurring more than once in either group. A strong statistical difference was observed when the number of rare missense variants with an aggregated prediction of pathogenicity and occurring more than once in either group was compared (p = 0.0035). Moreover, 8 variants identified more than once in either group, occurring in previously identified DR-associated genes were recognized. The p.Pro234Ser KIR2DS4 variant showed a strong protective effect (OR = 0.04 [0.001-0.36]; p = 0.04).

Conclusions: Our study showed an enrichment of rare splice acceptor/donor variants in patients with PDR and identified a potential protective variant in KIR2DS4. Although statistical significance was not reached, our results support the replication of 8 previously identified DR-associated genes.

Figure 1

Variant frequencies showing statistically significant differences among groups. (a) Missense variants with population frequency less than 1%. (b) Missense variants with a population frequency less than 1%, which were found in at least two subjects, either in cases or controls, and which had in silico aggregate prediction of pathogenicity. (c) Splice acceptor/donor variants with a population frequency less than 1% and observed in at least two subjects, either in cases or controls.