Anthropometry-adjusted TyG indices predict diabetes, graft failure, and mortality in kidney transplant recipients

The triglyceride-glucose (TyG) index is a widely used surrogate marker of insulin resistance and associated with cardiometabolic outcomes. However, the comparative prognostic value of TyG-derived indices, particularly those incorporating anthropometric measures or long-term variability, for adverse outcomes after kidney transplantation remains unclear.

To compare the prognostic performance of 15 TyG-derived indices for predicting new-onset diabetes after transplantation (NODAT), graft failure, and all-cause mortality in kidney transplant recipients (KTRs).

This prospective observational study was conducted within the TransplantLines Biobank and Cohort study and included 464 KTRs (median age, 56.9 years; 65.1% male) followed for a median of 71 months. Fifteen TyG-derived indices were classified into metabolic, anthropometry-adjusted, and variability-based domains. Their prognostic performance was assessed using an analytical framework integrating machine learning with conventional statistical methods.

Comparative analyses identified 4 indices, TyG-WWI, TyG-WHR, TyG mean, and TyG-BMI, as having superior predictive performance. Among them, anthropometry-adjusted indices (TyG-WHR and TyG-WWI) showed the most stable associations across 3 outcomes. In multivariable analyses, TyG-WHR independently predicted NODAT (HR 2.30, 95% CI 1.21-4.39), graft failure (HR 2.22, 95% CI 1.18-4.18), and mortality (HR 1.87, 95% CI 1.19-2.96), while TyG-WWI independently predicted mortality (HR 2.07, 95% CI 1.11-3.86). Cross-validation showed no consistent advantage of combined models over single well-performing indices.

Among adult KTRs from the TransplantLines cohort with a functioning graft, TyG-WWI and TyG-WHR showed the most consistent associations with adverse post-transplant outcomes. Their potential role as complementary metabolic risk markers requires validation in larger and more diverse transplant cohorts.