Abstract
AIMS: We intended to characterize the superiority of triglyceride glucose-body mass index (TyG-BMI) in predicting type 2 diabetes mellitus (T2DM) compared with triglyceride glucose (TyG) and homeostatic model assessment for insulin resistance (HOMA-IR). METHODS: A total of 699 nondiabetic participants in the Da Qing IGT and Diabetes Study were involved in the present analysis and classified according to the median of baseline TyG-BMI, namely the G1 (low TyG-BMI) and G2 (high TyG-BMI) groups. Information on developing diabetes was assessed from 1986 to 2020. RESULTS: During the 34-year follow-up, after adjustment for confounders, the G2 group had a higher risk of developing type 2 diabetes than the G1 group (hazard ratio [HR]: 1.92, 95% confidence interval [CI]: 1.51-2.45, p < 0.0001). Restricted cubic spline analyses showed that increased TyG-BMI was linearly related to higher risks of type 2 diabetes (p for non-linearity>0.05). Time-dependent receiver operator characteristics curves suggested that TyG-BMI exhibited higher predictive ability than TyG (6-year: area under the curve [AUC](TyG-BMI) vs. AUC(TyG), 0.78 vs. 0.70, p = 0.03; 34-year: AUC(TyG-BMI) vs. AUC(TyG), 0.79 vs. 0.73, p = 0.04) and HOMA-IR (6-year: AUC(TyG-BMI) vs. AUC(HOMA-IR), 0.78 vs. 0.70, p = 0.07; 34-year: AUC(TyG-BMI) vs. AUC(HOMA-IR), 0.79 vs. 0.71, p = 0.04) in both short and long terms, and the thresholds of TyG-BMI to predict type 2 diabetes were relatively stable (195.24-208.41) over the 34-year follow-up. CONCLUSIONS: In this post hoc study, higher TyG-BMI was associated with an increased risk of type 2 diabetes and demonstrated better predictability than TyG and HOMA-IR, favoring the application of TyG-BMI as a potential tool for evaluating the risk of type 2 diabetes in clinical practice.