Differences in imeglimin response in subgroups of patients with type 2 diabetes stratified by data‐driven cluster analysis: A post‐hoc analysis of imeglimin clinical trial data

Aim To explore differences in imeglimin response among type 2 diabetes (T2D) patient clusters using data‐driven cluster analysis. Methods Data‐driven cluster analysis (non‐hierarchical k‐means clustering) was performed on randomized, double‐blind, imeglimin monotherapy and adjunctive (to insulin) therapy trials based on four baseline variables: (1) disease duration; (2) body mass index (BMI); (3) HbA1c; and (4a) homeostatic model assessment of β‐cell function (HOMA‐β) (monotherapy trials) or (4b) insulin total daily dose (adjunctive trial). Results Four clusters were identified with distinct clinical characteristics in both monotherapy (1‐4) and adjunctive therapy (I‐IV) trials; clusters 1 and I had lower values across all four indices versus the overall population, clusters 2 and II had a longer diabetes duration, cluster 3 had higher baseline BMI and HOMA‐β, and cluster III had higher baseline BMI and insulin total daily dose, while clusters 4 and IV had higher baseline HbA1c. Between‐group differences in HbA1c change (95% confidence interval) and effect size (ES) at week 24 varied considerably by cluster (cluster 1: −0.82 [−1.00, −0.63], ES = 1.47; cluster 2: −0.64 [−0.89, −0.39], ES = 1.18; cluster 3: −0.86 [−1.38, −0.33], ES = 0.84; cluster 4: −1.27 [−1.73, −0.82], ES = 1.44). For imeglimin adjunctive therapy, HbA1c improvements were significant versus placebo at week 16, excluding cluster III (cluster I: −0.63 [−0.95, −0.31], ES = 0.88; cluster II: −0.66 [−1.02, −0.30], ES = 1.13; cluster III: −0.31 [−0.73, 0.11], ES = 0.46; cluster IV: −0.82 [−1.29, −0.35], ES = 0.99). Conclusions Differences in imeglimin response were observed among T2D patient clusters. Patient stratification may help with selection of those most probable to respond to imeglimin.