Classification of Japanese type 1 diabetes based on clinical phenotypes and its association with diabetic complications: Across-sectional study.

Introduction: Despite the increasing number of studies using machine learning to develop individualized treatment strategies, only a few have been conducted in patients with type 1 diabetes. This study aimed to identify the characteristics of Japanese patients with type 1 diabetes, classified into subgroups using data-driven cluster analysis based on pancreatic beta-cell function, obesity, and glycemic control, and clarify the association between these subgroups and diabetic complications.

Materials and methods: In this cross-sectional study, a cluster analysis using three variables (C-peptide, body mass index, and glycated hemoglobin) in 206 Japanese patients with type 1 diabetes was performed. Multivariate logistic regression analysis was performed to compare the risk of diabetic complications by subgroup.

Results: The cluster analysis identified four subgroups. Group 2 (n = 58), characterized by high body mass index levels, had a higher risk of hepatic steatosis than the control group (Group 1, n = 90). Meanwhile, Group 3 (n = 44), characterized by high glycated hemoglobin levels, had higher risks of retinopathy, polyneuropathy, elevated brachial-ankle pulse wave velocity, and hepatic steatosis than Group 1 and Group 4 (n = 14), characterized by residual endogenous insulin, had a higher risk of chronic kidney disease than Group 1.

Conclusions: The risks of diabetic complications differed between subgroups of Japanese patients with type 1 diabetes. Tailored treatment approaches based on subgroup characteristics are a potential treatment option for reducing the risks of diabetic complications in this population.