Immune Cell-Based Clustering Reveals Clinically Distinct Endotypes in Type 2 Diabetes

Type 2 diabetes (T2D) is a heterogeneous disease, yet current classifications fail to capture its complexity or enable precision care. Here, we introduce a clinically scalable, immune-based endotyping strategy using routine blood immune cell counts - neutrophils, lymphocytes, and monocytes - from over 1,500 individuals with newly diagnosed T2D across three independent longitudinal European cohorts. Unsupervised clustering identified four consistent and clinically meaningful immune endotypes: Severe Inflammatory Diabetes (SIND), Mild Inflammatory Diabetes (MIND), Lymphocyte-Rich Diabetes (LYRD), and Lymphocyte-Deficient Diabetes (LYDD), that named immunotypes. These endotypes were associated with divergent long-term outcomes, with SIND and LYDD showing increased cardiovascular, renal, and mortality risks, and MIND and LYRD linked to more favorable trajectories. Multi-omic profiling revealed endotype-specific inflammatory signatures, including selective expansion of CCR2hi and CD39hi classical monocytes in the SIND endotype. Integration with single-cell RNA-seq uncovered distinct monocyte subsets in SIND, enriched for chemotaxis and myeloid activation transcriptional programs, alongside skewed adaptive immunity. Strikingly, this pro-inflammatory immune profile was attenuated by IL-1β antagonism and bariatric surgery-induced diabetes remission, underscoring its therapeutic relevance. These findings position immune-based endotyping as a robust and accessible tool to stratify risk, uncover disease mechanisms, and guide personalized intervention strategies in T2D.