Cardiovascular Diabetology · Oct 20, 2026
Joint association of estimated glucose disposal rate and normalized creatinine-to-cystatin C ratio with incident cardiovascular disease
By Zhong S, He L, Xiao L et al.
Background: Estimated glucose disposal rate (eGDR) reflects insulin-resistance burden, whereas normalized creatinine-to-cystatin C ratio (NCCR) is a creatinine-cystatin C and body-weight-based biomarker originally proposed as a biochemical correlate of body-weight-adjusted skeletal muscle mass. Whether their combination identifies cardiovascular risk beyond either marker alone remains uncertain. Methods: We included 4,935 participants aged ≥ 45 years from the China Health and Retirement Longitudinal Study who were free of CVD at baseline in 2011, had eGDR and NCCR values within their respective 1st–99th percentile ranges, and had complete data on Model 3 covariates. Incident CVD was defined as the first self-reported heart disease or stroke through 2020, with event time assigned to the survey year of the first positive report. Cox regression, restricted cubic splines, joint-exposure analyses, prediction evaluation, subgroup analyses, and sensitivity analyses were performed. Multiple imputation was additionally conducted as a sensitivity analysis. Results: During a median follow-up of 9 years, 1,235 incident CVD events occurred. Using participants with low eGDR and low NCCR as the reference, those with high eGDR and high NCCR had a fully adjusted HR of 0.629 (95% CI 0.538–0.734). CVD risk decreased progressively across participants with zero, one, and two high markers (P for trend < 0.001). At 9 years, the AUCs were 0.618 for Model 3, 0.638 for Model 3 plus eGDR, 0.623 for Model 3 plus NCCR, and 0.638 for Model 3 plus both markers. Adding NCCR to the model containing eGDR did not produce a discernible improvement in discrimination. Conclusions: The combination of low eGDR and low NCCR identified a phenotype with a higher risk of incident CVD, whereas increasing numbers of high markers were associated with progressively lower risk. These findings support the use of eGDR and NCCR for joint epidemiological risk characterization, although the incremental predictive value of NCCR beyond eGDR alone was limited.