Association between the triglyceride-glucose index and risk of type 2 diabetes mellitus and the mediating effect of BMI: a comparative analysis in Chinese and Japanese populations

Objective The triglyceride–glucose (TyG) index is a surrogate marker of insulin resistance and metabolic dysregulation. We aimed to examine its association with incident type 2 diabetes mellitus (T2DM) in Chinese and Japanese adults and to quantify the mediating role of body mass index (BMI). However, ethnic differences in the TyG—diabetes association, population-specific thresholds, and potential mediating mechanisms—remain unclear. This study aimed to evaluate the association between the TyG and incident T2DM in Chinese and Japanese adults and to quantify the mediating role of BMI in these associations. Methods We conducted a retrospective cohort study using data from the China Rich Healthcare Group (n=199,050) and the Japanese NAGALA database (n=15,464). TyG was calculated as lnfasting triglycerides (mg/dL) ×fasting plasma glucose(mg/dL)/2. Incident T2DM was defined according to American Diabetes Association criteria. Multivariable Cox models, restricted cubic splines, and two-piecewise regression were used to characterize linear and nonlinear associations between TyG and diabetes risk. Predictive performance was assessed using receiver operating characteristic (ROC) curves. Mediation analysis with 5,000 bootstrap replications quantified the proportion of the TyG–diabetes association mediated by BMI. Results During a median follow-up of 5.0 years, 2,563 participants developed T2DM. Recent reviews and experimental studies indicate that Japanese adults often develop type 2 diabetes at relatively modest levels of adiposity, frequently with insufficient insulin secretion and a disproportionately high burden of visceral and ectopic fat (fully adjusted HR per 1-unit increase, 2.32; 95% CI, 2.16–2.47), with clear gradients across TyG quartiles in both cohorts. BMI partially mediated the TyG–diabetes association (19.06% in Chinese vs 14.22% in Japanese adults), supporting adiposity-related pathways and population differences in metabolic mediation. Nonlinear analyses suggested cohort-specific inflection points, with risk rising more steeply above TyG ≈8.98 in Chinese and ≈7.88 in Japanese adults. TyG showed moderate discrimination for 5-year diabetes risk (AUC ≈0.74), outperforming triglycerides alone; fasting plasma glucose (FPG) remained more discriminative, and ROC results are reported descriptively. Conclusions Higher baseline TyG was associated with incident T2DM in both Chinese and Japanese adults, with BMI partially mediating the TyG–diabetes association. These findings suggest that TyG captures triglyceride–glucose dysregulation beyond overall adiposity, with population-specific differences in metabolic pathways. The identification of nonlinear patterns underscores the need for population-tailored risk stratification based on TyG.