Development and validation of a novel metabolic health-related nomogram to improve predictive performance of cardiovascular disease risk in patients with prediabetes.

Abstract

Objective: The prevalence of prediabetes among adults in the U.S. is three times higher than that of diabetes, highlighting a greater disease burden. Both diabetes and prediabetes have been demonstrated to be associated with an increased risk of cardiovascular disease (CVD). However, research has primarily focused on diabetes, with limited attention to CVD risk prediction in prediabetes. Emerging 13 metabolic health-related indicators have been proposed to optimize the predictive effect on CVD risk in patients with prediabetes. This study aimed to compare the predictive efficacy of these biomarkers and further develop a nomogram to improve predictive performance of the CVD risk in patients with prediabetes.

Methods: All eligible participants in the National Health and Nutrition Examination Survey (NHANES) 1999-2020 were enrolled in this study and randomly assigned to the development and validation cohorts in a ratio of 7:3. In the development cohort, the efficacy of 13 indicators used to predict the CVD risk was assessed by receiver operative characteristic (ROC) curves. Independent risk predictors identified by multivariate logistic regression were used to construct a nomogram, and internal and external validation were further implemented.

Results: The ROC curve demonstrated that the triglyceride-glucose (TyG) index was an effective predictor of CVD risk [area under the curve (AUC) = 0.694] and exhibited the best predictive performance among the 13 metabolic health-related indices. Based on independent risk factors identified by multivariate logistic regression, the CVD risk nomogram [including age, gender, hypertension, TyG, stress hyperglycemia ratio (SHR), and neutrophil-to-lymphocyte ratio (NLR)] was successfully constructed and validated with good performance (AUCs/C-indexes > 0.70 for all).

Conclusion: This study developed a reliable nomogram for predicting CVD risk in patients with prediabetes. The model demonstrated robust performance and offered a simple yet individualized approach for predicting the CVD risk in patients with prediabetes.