1346-P: Identification of Prediabetes Subtypes and Their Association with Outcomes Using Six Variables—A Data-Driven Cluster Analysis

Introduction and Objective: Clinical features of prediabetes show heterogeneity. In this study, we aimed to provide a refined classification of prediabetes and identify individuals at increased risk of outcomes. Methods: We did a data-driven cluster analysis of prediabetes (n = 3340) from the China Health and Retirement Longitudinal Study (2011-2020). The clustering was based on six variables: systolic blood pressure, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, fasting plasma glucose, glycated hemoglobin, and body mass index. Replication was performed in two cohorts: the English Longitudinal Study of Aging (2008-2018; n = 1529) and the China Health and Nutrition Survey (2009-2015; n = 2769). Cox models were used to compare the risks of type 2 diabetes mellitus (T2DM) and cardiovascular disease (CVD). Results: We found seven replicable clusters of prediabetes patients that exhibited significant differences in metabolic features and risks of T2DM and CVD. Individuals in clusters 4, 5, 6, and 7 had significantly higher risks of T2DM, and Cluster 3 (hypertension) had the highest CVD risk. Conclusion: We grouped patients into seven clusters, which exhibited differences in metabolic features and risks of T2DM and CVD. This classification may help identify populations that could benefit most from early intervention for risk factors. Disclosure X. Chen: None. Q. Pan: None.