Relationship between circulating metabolites and diabetic retinopathy: a two-sample Mendelian randomization analysis

Abstract

Diabetic retinopathy (DR) is the most frequent microvascular complication of diabetes mellitus, however, its underlying biological mechanisms remain poorly understood. We examined single nucleotide polymorphisms linked to 486 blood metabolites through extensive genome-wide association studies conducted on individuals of European ancestry. The FinnGen Biobank database served as a reference to define DR. Two-sample MR analysis was conducted to reveal the association between the levels of genetically predicted circulating metabolites and the susceptibility to DR. To validate the robustness of the obtained findings, sensitivity analyses with weighted median, weighted mode, and MR-Egger were conducted. 1-oleoylglycerophosphoethanolamine (odds ratio [OR] (OR per one standard deviation [SD] increase) = 0.414; 95% confidence interval [CI] 0.292–0.587; P = 7.613E−07, P_FDR = 6.849E−06), pyroglutamine (OR per one SD increase = 0.414; 95% confidence interval [CI] 0.292–0.587; P = 8.31E−04, P_FDR = 0.007), phenyllactate (PLA) (OR per one SD increase = 0.591; 95% confidence interval [CI] 0.418–0.836; P = 0.003, P_FDR = 0.026), metoprolol acid metabolite (OR per one SD increase = 0.978; 95% confidence interval [CI] 0.962–0.993; P = 0.005, P_FDR = 0.042), 10-undecenoate (OR per one SD increase = 0.788; 95% confidence interval [CI] 0.667–0.932; P = 0.005, P_FDR = 0.049), erythritol (OR per one SD increase = 0.691; 95% confidence interval [CI] 0.513–0.932; P = 0.015, P_FDR = 0.034), 1-stearoylglycerophosphoethanolamine (OR per one SD increase = 0.636; 95% confidence interval [CI] 0.431–0.937; P = 0.022, P_FDR = 0.099), 1-arachidonoylglycerophosphoethanolamine (OR per one SD increase = 0.636; 95% confidence interval [CI] 0.431–0.937; P = 0.030, P_FDR = 0.099) showed a significant causal relationship with DR and could have protective effects. stachydrine (OR per one SD increase = 1.146; 95% confidence interval [CI] 1.066–1.233; P = 2.270E−04, P_FDR = 0.002), butyrylcarnitine (OR per one SD increase = 1.117; 95% confidence interval [CI] 1.023–1.219; P = 0.014, P_FDR = 0.062), 5-oxoproline (OR per one SD increase = 1.569; 95% confidence interval [CI] 1.056–2.335; P = 0.026, P_FDR = 0.082), and kynurenine (OR = 1.623; 95% CI 1.042–2.526; P = 0.041, P_FDR = 0.097) were significantly associated with an increased risk of DR. This study identified metabolites have the potential to be considered prospective compounds for investigating the underlying mechanisms of DR and for selecting appropriate drug targets.