Association between insulin resistance-related indices and cardiovascular disease in patients with atopic dermatitis: mutual validation using real-world and NHANES data.

Background: The number of patients with atopic dermatitis (AD) who develop cardiovascular disease (CVD) has been increasing year by year. However, reliable biomarkers for identifying AD patients with a higher burden of prevalent CVD remain lacking. Methods: This study included patient data from the First Affiliated Hospital of Harbin Medical University from 2014 to 2023 and data from the U.S. National Health and Nutrition Examination Survey (NHANES). Multivariable logistic regression analyses were performed to evaluate the associations of triglyceride-glucose index (TyG), TyG-body mass index (TyG-BMI), C-reactive protein-triglyceride-glucose index (CTI), and CTI-body mass index (CTI-BMI) with prevalent CVD in patients with AD. Restricted cubic spline (RCS) analyses, subgroup analyses, and sensitivity analyses were conducted to evaluate the robustness of the findings. Seven machine learning (ML) models and SHAP analysis were applied to assess the discriminative performance of these indices for prevalent CVD status. Results: In the real-world dataset, 145 patients had prevalent CVD, whereas 75 participants had prevalent CVD in the NHANES dataset. Multivariable logistic regression analyses showed that, after adjustment for clinical covariates, all four indices were significantly and positively associated with prevalent CVD among patients with AD. These associations remained consistent across most subgroup analyses. Restricted cubic spline analyses showed no evidence of nonlinear associations, and sensitivity analyses supported the robustness of the findings. Among the evaluated ML models, CTI-BMI demonstrated the highest discriminative performance for prevalent CVD status in both datasets and contributed most substantially to model discrimination. Conclusion: This study demonstrated that TyG, TyG-BMI, CTI, and CTI-BMI were significantly associated with prevalent CVD among patients with AD. CTI-BMI showed the strongest discriminative performance among the evaluated indices and provided modest incremental discrimination when incorporated into machine learning models. These findings suggest that these indices may serve as potential adjunctive markers for cardiovascular burden assessment in patients with AD; however, further prospective studies are required to determine their value for predicting future cardiovascular events.