A1C-Based Subgroups of Type 2 Diabetes and Their Association with Microvascular Complications—Cluster Analysis of a Large Italian Cohort

Introduction and Objective: Individuals with type 2 diabetes (T2DM) exhibit significant heterogeneity, reflected by different expression of complications and responses to therapy. Existing clustering approaches integrate multiple parameters. This study aims to classify T2DM subjects into clusters solely based on longitudinal A1c trends, applying artificial intelligence and machine learning techniques. Methods: This retrospective study utilized data from the dataset of the Italian Association of Medical Diabetologists (AMD) extracted from Electronic Medical Records of over 200 diabetes clinics during a 12-year-observation period (2006-2018). Inclusion criteria required consistent annual A1c measurements over a 10-year period. Using the k-means algorithm with dynamic time warping as the distance metric, clusters were identified based on A1c trajectories. Demographic and anthropometric data, glucose-lowering therapies, and prevalence of nephropathy and retinopathy were compared across clusters. Results: In a population of 68,486 subjects, four distinct clusters were identified: well-controlled, moderately controlled, uncontrolled with continuous improvement (WCI), and uncontrolled. The uncontrolled cluster showed the highest prevalence of nephropathy and retinopathy (52.73% and 64.18%, respectively), while the well-controlled cluster exhibited the lowest (33.62% and 23.34%, respectively). The moderately controlled and uncontrolled WCI clusters had complication rates in between these extremes. Differences in demographic and clinical characteristics further highlighted the heterogeneity among the clusters. Conclusion: A1c trajectories provide a practical basis for clustering T2DM subjects into four subgroups with different prevalence of microvascular complications. This methodology, relying on a single parameter, might help in optimizing therapeutic outcomes, minimizing complications, and advancing precision medicine in T2DM management.