Cardiovasc Diabetol. 2026 Sep 23;25(1):285. doi: 10.1186/s12933-026-03386-8.
ABSTRACT
BACKGROUND: The triglyceride-glucose (TyG) index and non-traditional lipid parameters (nTLPs) are inexpensive markers of overlapping metabolic and lipoprotein burden. Their CVD-specific longitudinal associations and incremental prognostic contribution in adults with cardiovascular-kidney-metabolic (CKM) stages 0-3 remain uncertain.
METHODS: We conducted a 2015 landmark analysis in the China Health and Retirement Longitudinal Study. TyG was evaluated jointly with the atherogenic index of plasma (AIP), non-HDL cholesterol (non-HDL-C), Castelli risk indices I and II (CRI-I and CRI-II), and the lipoprotein combine index (LCI). We examined antecedent 2011 values, 2015 landmark values, and a two-visit time-weighted cumulative measure from 2011 to 2015. The primary analysis family comprised five cumulative dual-high versus dual-low comparisons for incident CVD. Cause-specific Cox and Fine-Gray models were fitted, with Benjamini-Hochberg correction. Secondary analyses addressed heart disease, stroke, dose response, effect modification, robustness, and censoring-aware five-year prediction.
RESULTS: Among 4,420 participants with CKM stages 0-3 and valid post-landmark CVD follow-up, 827 developed CVD; endpoint-specific analyses included 4,411 participants with 571 heart-disease events and 4414 with 328 stroke events. For cumulative dual-high exposure, adjusted Cox HRs for CVD ranged from 1.29 (95% CI 1.09-1.52) to 1.45 (1.20-1.74), and Fine-Gray SHRs ranged from 1.28 (1.09-1.50) to 1.43 (1.19-1.72); all primary FDR q values were < 0.003. Secondary stroke estimates were generally larger. The covariate-only C-index was 0.616 and five-year IPCW AUC was 0.634 (0.608-0.659). The best expanded model, TyG plus CRI-II, yielded a C-index of 0.628 and AUC of 0.665 (0.640-0.690), corresponding to changes of 0.012 and 0.031, respectively.
CONCLUSIONS: Higher two-visit cumulative TyG-nTLP burden was associated with incident CVD after accounting for competing death. However, incremental discrimination was small, and shared laboratory components limit claims of biological independence or synergy. These indices may provide additional information for epidemiologic risk characterization, while their incremental clinical predictive value appears limited and warrants further validation.
PMID:42778923 | DOI:10.1186/s12933-026-03386-8