Abstract
Frailty and diabetes mellitus (DM) are closely linked, but their causal relationship remains unclear. This study aims to determine the bidirectional causal relationship between frailty and different DM subtypes using Mendelian randomization (MR). We performed a 2-sample MR analysis using summary statistics from large-scale genome-wide association studies. The inverse-variance weighting method was the primary analytical approach, with MR-Egger regression and weighted median methods for sensitivity analysis. Horizontal pleiotropy and heterogeneity were assessed using MR-PRESSO and Cochran Q test. Genetically predicted frailty was significantly associated with an increased risk of type 2 diabetes (T2DM) and gestational diabetes (GDM) (odds ratio [OR] = 2.142, 95% confidence interval [CI]: 1.751-2.621, P < .001; OR = 2.280, 95% CI: 1.368-3.800, P = .002), but no causal relationship was observed for type 1 diabetes or glycemic traits (P > .05). Conversely, genetically predicted type 1 diabetes, T2DM, GDM, and postprandial glucose levels (2-hour post-load glucose) increased the risk of frailty (OR = 1.026, 95% CI: 1.014-1.038, P < .001; OR = 1.046, 95% CI: 1.033-1.058, P < .001; OR = 1.068, 95% CI: 1.040-1.096, P < .001; OR = 1.095, 95% CI: 1.049-1.144, P < .001). Sensitivity analyses confirmed the robustness of these findings. This study provides genetic evidence supporting a bidirectional causal relationship between frailty and diabetes, particularly T2DM and GDM. These findings highlight the need for early frailty screening in diabetic patients and better metabolic management in frail populations.