Mechanistic heterogeneity in diabetic cardiomyopathy: a framework for endotype discovery and validation

Diabetic cardiomyopathy (DbCM) refers to diabetes-associated myocardial dysfunction or remodeling. In this condition, diabetes-related biological processes are considered central after competing and coexisting contributors have been evaluated. Clinical phenotypes remain essential for diagnosis, staging, risk assessment, and management. However, they may not fully capture the biological heterogeneity underlying myocardial injury in diabetes. In this Review, we organize current evidence around five overlapping candidate mechanistic domains. These domains are metabolic-energy remodeling, inflammatory-immune activation, fibrotic remodeling, microvascular-endothelial dysfunction, and mitochondrial-organelle stress. They are not treated as five validated patient endotypes. We examine their biological architecture and cross-talk, their relationships with clinical phenotypes and disease stage, and candidate measurements derived from cardiovascular imaging, circulating biomarkers, and multi-omics approaches. We further consider how multimodal data may be integrated into continuous mixed-membership profiles that preserve biological overlap and uncertainty. We propose a staged framework in which candidate profiles progress from analytically robust measurement and profile discovery to model locking, external replication, longitudinal evaluation, and validation for a prespecified intended use. Prognostic, monitoring, trial-enrichment, and treatment-selection claims require distinct forms of evidence. Current guideline-based care remains independent of research profiles. Mechanism-matched treatment should remain a hypothesis for prospective testing. This framework therefore provides a roadmap for DbCM endotype discovery and validation rather than an established taxonomy or an actionable precision-medicine system.