Reassessment-stage metabolic indices and subsequent cardiorenal outcomes in adults with diabetes: CHARLS and an independent hospital validation cohort.
Liu M, Kong Y, et al. • Frontiers in endocrinology • 2026
Among adults with diabetes who remained free of overt cardiorenal outcomes at reassessment, higher TyG, TCBI, and METS-IR were associated with subsequent composite cardiorenal outcomes, but predictive improvement over conventional models was limited.
Key Findings
Results
Each 1-SD increase in TyG index was associated with significantly higher composite cardiorenal risk in the CHARLS reassessment cohort.
HR 1.26 (95% CI: 1.06–1.49) per 1-SD increase in TyG index
Analysis performed in glycemia-adjusted complete-case Cox models with 787–800 participants
The eligible CHARLS reassessment cohort included 1,104 participants and 311 composite cardiorenal outcomes
Participants were identified with diabetes in 2011, attended the 2015 survey free of overt cardiorenal outcomes, and had post-2015 follow-up
Proportional-hazards tests showed no violations
Results
Each 1-SD increase in TCBI was associated with significantly higher composite cardiorenal risk.
HR 1.19 (95% CI: 1.03–1.37) per 1-SD increase in TCBI
Estimate derived from glycemia-adjusted complete-case Cox analyses
Selection-weighted and discrete-time estimates were similar to primary Cox results
Association was reproduced in the independent hospital validation cohort
Results
Each 1-SD increase in METS-IR was associated with significantly higher composite cardiorenal risk.
HR 1.20 (95% CI: 1.04–1.38) per 1-SD increase in METS-IR
Estimate derived from glycemia-adjusted complete-case Cox analyses in CHARLS cohort
Association was reproduced in the independent hospital validation cohort (N=1,578; 296 events)
Selection-weighted and discrete-time sensitivity analyses supported the primary estimate
Results
The CHG index was not significantly associated with composite cardiorenal outcomes in the primary CHARLS analysis and showed a weaker and graphically inconsistent pattern in the validation cohort.
CHG did not reach statistical significance in the glycemia-adjusted Cox analyses in CHARLS
In the independent hospital validation cohort, CHG showed 'a weaker and graphically inconsistent pattern'
This contrasted with TyG, TCBI, and METS-IR, which showed consistent associations across both cohorts
Results
Adding TyG, TCBI, or METS-IR to conventional models provided only minimal and non-statistically significant improvements in discrimination.
In the wave-approximated 4-year prediction sample (N=749; 173 events), apparent AUC gains were 'small and not statistically significant'
10-fold cross-validated AUC increased only from 0.665 to 0.669–0.671 when adding the metabolic indices
There was no consistent decision-curve advantage for any of the indices
Authors concluded 'predictive improvement was limited'
Results
Associations of TyG, TCBI, and METS-IR with composite cardiorenal outcomes were reproduced in an independent hospital validation cohort.
Independent hospital validation cohort included N=1,578 participants with 296 events
Post-landmark follow-up was administratively censored at 4 years
Positive composite-outcome associations were reproduced for TyG, TCBI, and METS-IR
CHG showed a weaker and graphically inconsistent pattern in this cohort as well
Results
Robustness analyses including selection weighting, discrete-time models, and proportional-hazards diagnostics supported the primary associations.
Selection-weighted estimates were similar to primary Cox estimates for TyG, TCBI, and METS-IR
Discrete-time model estimates were also consistent with primary results
Proportional-hazards diagnostic tests showed no violations for any index
Restricted cubic splines and Kaplan-Meier curves were among the analytic approaches used
Conclusions
The study characterized TyG, TCBI, and METS-IR as exploratory, low-burden adjunctive markers at reassessment rather than clinically actionable predictors.
Authors stated these indices 'may serve as exploratory, low-burden adjunctive markers at reassessment, with supportive independent clinical validation'
Authors noted 'clinical utility requires further study'
The indices were evaluated in participants who had already survived to a reassessment wave free of overt cardiorenal outcomes, reflecting a survivor/selection context
The paper explicitly stated the indices capture 'residual cardiorenal risk beyond conventional glycemic markers'
What This Means
This research suggests that three simple blood-based metabolic scores — the triglyceride-glucose (TyG) index, the triglyceride-total cholesterol-body weight index (TCBI), and the metabolic score for insulin resistance (METS-IR) — are linked to higher risk of future heart or kidney problems in people with diabetes who appeared healthy at a follow-up medical check. The study followed over 1,100 people with diabetes from a large Chinese national survey who were re-examined in 2015 and had not yet developed heart or kidney disease. Those with higher scores on these three metabolic indices were more likely to experience a cardiovascular or kidney event afterward. A fourth index, the CHG index, did not show a consistent association. These findings were confirmed in a separate group of more than 1,500 hospital patients, lending credibility to the results.
However, the research also found that adding these metabolic indices to existing prediction models did not meaningfully improve the ability to identify who would develop heart or kidney disease. The improvement in prediction accuracy was very small — the model's discrimination score (AUC) only nudged from 0.665 to about 0.670 — and there was no clear benefit shown in decision-curve analyses, which test whether using a tool in practice would actually help clinical decision-making.
This research suggests that TyG, TCBI, and METS-IR could serve as simple, low-cost supplementary markers to flag elevated cardiorenal risk in people with diabetes at routine medical check-ups, since they require only standard lab measurements and body weight. However, the authors caution that whether acting on these scores would actually improve patient care remains unknown, and further studies are needed before these indices could be recommended for routine clinical use.
Liu M, Kong Y, Chen S, Wu S, Sun B, Chen L. (2026). Reassessment-stage metabolic indices and subsequent cardiorenal outcomes in adults with diabetes: CHARLS and an independent hospital validation cohort.. Frontiers in endocrinology. https://doi.org/10.3389/fendo.2026.1926018