Cumulative estimated glucose disposal rate and longitudinal eGDR patterns in relation to incident cardiovascular disease among middle-aged and older adults with metabolic dysfunction-associated steatotic liver disease: a prospective cohort study

Estimated glucose disposal rate (eGDR) is a surrogate measure of insulin sensitivity derived from routinely available clinical variables. However, whether repeated eGDR measurements provide additional information for cardiovascular risk assessment among individuals with metabolic dysfunction-associated steatotic liver disease (MASLD) remains unclear. We examined the associations of cumulative eGDR and longitudinal eGDR patterns with incident cardiovascular disease (CVD) and evaluated their incremental risk-discrimination value. Using data from the China Health and Retirement Longitudinal Study, we conducted a prospective cohort analysis with 2015 specified as the landmark. Participants were aged ≥ 45 years in 2011, met an operational TyG-WC–based definition of MASLD in 2011, had calculable eGDR in both 2011 and 2015, were alive and free of CVD at the 2015 landmark, and had valid post-2015 outcome and follow-up data. Cumulative eGDR was calculated from the 2011 and 2015 measurements and, because all participants shared the same four-year interval, was mathematically proportional to the mean of the two measurements; K-means clustering was used exploratorily to identify longitudinal eGDR patterns, operationally characterised as two-wave eGDR level–change patterns based on the two observed measurements. Participants were followed from 2015 to 2020, and incident CVD was defined as the first report of physician-diagnosed heart disease or stroke. Cox proportional hazards models and restricted cubic splines were used to estimate associations and dose–response relationships. Competing-risk, subgroup, and sensitivity analyses, including alternative handling of CVD event times, were performed to assess robustness. Incremental predictive performance was evaluated using conventional and internally validated discrimination measures, including paired out-of-bag C-index comparisons. A total of 2,059 participants were included, of whom 457 developed incident CVD during follow-up. Each 1-SD lower cumulative eGDR was associated with a 39% higher risk of CVD after full adjustment (hazard ratio [HR] 1.39, 95% confidence interval [CI] 1.26–1.55). Compared with participants in the highest cumulative eGDR quartile (Q1), those in the lowest quartile (Q4) had an HR of 2.40 (95% CI 1.77–3.25; P for trend < 0.001). The dose–response relationship was approximately linear (P for non-linearity = 0.915). K-means clustering identified four exploratory longitudinal eGDR patterns based on the two observed measurements. Using the high eGDR–slight decline pattern as the reference, the moderate-low eGDR–slight rise, high eGDR–marked decline, and low eGDR–slight decline patterns were associated with HRs of 1.43 (95% CI 1.07–1.90), 1.64 (1.20–2.23), and 2.14 (1.67–2.74), respectively. Lower cumulative eGDR was also associated with higher risks of heart disease and stroke, and the main findings remained stable across multiple sensitivity analyses, including alternative handling of event times. Adding cumulative eGDR to the conventional risk-factor model increased the paired out-of-bag C-index by 0.033 (95% CI 0.009–0.056); however, when 2015 eGDR was already included, adding cumulative eGDR produced essentially no further improvement (ΔC-index − 0.001, 95% CI − 0.007 to 0.006). Among middle-aged and older Chinese adults meeting an operational TyG-WC–based MASLD definition, lower cumulative eGDR and longitudinal eGDR patterns characterised by low levels at both observed measurements or a marked decline between measurements were associated with a higher risk of incident CVD. Cumulative eGDR provided modest incremental discrimination beyond conventional risk factors but essentially no additional discrimination beyond the 2015 eGDR measurement, indicating substantial overlap with the most recent eGDR value.