Jing Chen, Zhonghua Sun, Jiajun Qiu, Zhongqing Zhou, Kaiqi Ma, Yongxin Liang, Chunqun Li, Shicong Liang
In middle-aged and older adults, most CTI-related adiposity indices were significantly associated with the newly identified surrogate-defined CLMM phenotype, except for CTI-ABSI and CTI-WWI, which were not independently associated after full adjustment. Among them, CTI-BMI showed the strongest statistical association and the highest relative discrimination among the evaluated indices, although its absolute discrimination was modest. These findings suggest that both metabolic-inflammatory burden and obesity-related phenotypes may provide complementary information for population-level risk characterization of the surrogate-defined CLMM phenotype. Given the heterogeneous ascertainment and potential misclassification of its component conditions, these findings should not be interpreted as estimates of the risk of clinically confirmed CLMM.
BACKGROUND: The C-reactive protein-triglyceride-glucose index (CTI) is a comprehensive marker that reflects inflammation and insulin resistance. Cardiovascular-liver-metabolic (CLM) diseases caused by multiple factors may be related to insulin resistance, obesity, metabolism, etc. Our aim is to explore the association between the fat index derived from CTI and a newly identified surrogate-defined CLM multimorbidity (CLMM) phenotype, while also comparing the relative discriminatory performance of these indices and exploring potential biological domains.
METHODS: From the China Health and Retirement Longitudinal Study (CHARLS), a total of 6534 individuals with full baseline records were included in our analysis. Nine CTI-derived adiposity indices were calculated, including CTI, CTI-BMI, CTI-WC, CTI-WHtR, CTI-BRI, CTI-WWI, CTI-CVAI, CTI-ABSI, and CTI-CI. CVD was identified primarily from self-reported physician diagnoses, whereas MASLD was inferred using the lipid accumulation product (LAP) rather than imaging, histology, or clinical adjudication. Because the component conditions were ascertained using different methods and levels of diagnostic accuracy, CLMM was operationalized as a surrogate-defined composite phenotype rather than clinically confirmed CLMM. Associations as well as potential nonlinear relationships were examined using multivariable logistic regression, restricted cubic spline (RCS) modeling, and two-piece segmented regression analyses. Apparent discrimination and exploratory incremental model performance were quantified using receiver operating characteristic (ROC) curve analysis and reclassification measures, specifically the net reclassification improvement (NRI) and integrated discrimination improvement (IDI). Subgroup and sensitivity analyses were performed to assess stability.
RESULTS: During the approximately four-year follow-up interval, 406 participants (6.2%) met the criteria for the newly identified surrogate-defined CLMM phenotype. Most CTI-derived adiposity indicators were significantly associated with elevated odds of this phenotype, although CTI-ABSI and CTI-WWI showed null associations after full adjustment. In fully adjusted models, CTI-BMI and CTI-CVAI (Q4 OR 3.46 for CTI-BMI; Q4 OR 2.37 for CTI-CVAI) showed the strongest associations (both P <0.001). RCS analyses further indicated significant nonlinear exposure-response relationships for all indices, except for CTI. CTI-BMI had the highest AUC among the evaluated CTI-derived indices (AUC = 0.638, 95% CI 0.611-0.664), although its absolute discriminatory ability was modest. However, CTI-BMI did not materially outperform BMI alone in a direct same-sample comparison (AUC = 0.638 vs 0.638; DeLong P = 0.924), and the IDI for CTI-BMI versus BMI was only 0.02%, despite a positive NRI. When CTI-BMI was added to the fully adjusted baseline model, the AUC increased only slightly from 0.617 to 0.627. Although the NRI indicated improved reclassification (NRI = 28.63%, 95% CI 18.63-38.64%), the IDI was 0.03% (95% CI 0.01-0.06%), indicating a minimal improvement in average discrimination. Compared with participants with low CTI and low BMI, those with low CTI and high BMI (OR 2.49, 95% CI 1.80-3.48) and those with high CTI and high BMI (OR 2.53, 95% CI 1.87-3.45) had similarly elevated odds of the newly identified surrogate-defined CLMM phenotype. The two effect estimates were numerically similar, and no statistically significant multiplicative or additive interaction was observed. Significant effect modification was observed for the association between CTI-BMI and this phenotype, with stronger relationships identified in men, current-smokers, and current-drinkers (all P for interaction <0.001). Sensitivity analyses were generally directionally consistent, with CTI-BMI remaining significantly associated with the surrogate-defined phenotype after additional adjustment for baseline CLM component status, despite a reduction in the effect estimate.
CONCLUSIONS: In middle-aged and older adults, most CTI-related adiposity indices were significantly associated with the newly identified surrogate-defined CLMM phenotype, except for CTI-ABSI and CTI-WWI, which were not independently associated after full adjustment. Among them, CTI-BMI showed the strongest statistical association and the highest relative discrimination among the evaluated indices, although its absolute discrimination was modest. These findings suggest that both metabolic-inflammatory burden and obesity-related phenotypes may provide complementary information for population-level risk characterization of the surrogate-defined CLMM phenotype. Given the heterogeneous ascertainment and potential misclassification of its component conditions, these findings should not be interpreted as estimates of the risk of clinically confirmed CLMM.