Classification of glucose profile in Japanese patients with type 1 diabetes and its association with complications.

Abstract

The importance of utilizing continuous glucose monitoring (CGM) to optimize glycemic profiles and thereby prevent the onset of diabetic complications in patients with type 1 diabetes has been increasingly recognized. Nevertheless, studies evaluating the risk of diabetic complications in patients with type 1 diabetes through machine learning approaches based on CGM data remain limited. To classify the glycemic profiles of Japanese patients with type 1 diabetes using a data-driven cluster analysis based on CGM and clarify the association between these clusters and diabetic complications. In this cross-sectional study, a cluster analysis using glycemic metrics from CGM of 153 Japanese patients with type 1 diabetes was performed. Logistic regression analysis adjusted for age, sex, and duration of diabetes was performed to compare the risk of diabetic complications by cluster. The cluster analysis identified four clusters. Cluster 1 (n = 53) exhibited an optimal glycemic profile. Cluster 2 (n = 46) demonstrated an extended duration of hyperglycemia and a higher risk of elevated brachial-ankle pulse wave velocity than Cluster 1. Cluster 3 (n = 39) demonstrated an extended duration of hypoglycemia and a higher risk of severe hypoglycemia than Cluster 1. Cluster 4 (n = 15) demonstrated large glycemic variability associated with hyperglycemia and hypoglycemia. Cluster 4 had higher risks of polyneuropathy, elevated brachial-ankle pulse wave velocity, and higher cardiovascular disease risk scores than Cluster 1. High-risk diabetic complications were identified for each cluster classified by glycemic profile.