Data-Driven Diabetes Clustering in East Asians: Pathological Heterogeneity and Personalized Medicine

Diabetes is highly heterogeneous, rendering traditional type 1 and type 2 classification insufficient for optimizing treatment strategy. This mini-review examines data-driven clustering-categorizing adult-onset diabetes into five subtypes: SAID (severe autoimmune), SIDD (severe insulin-deficient), SIRD (severe insulin-resistant), MOD (moderate obesity-related), and MARD (mild age-related)-within the context of East Asian populations. East Asians exhibit a distinct "Asian phenotype" characterized by impaired insulin secretion and visceral fat accumulation at lower body mass index levels. Meta-analyses reveal a significantly higher proportion of the SIDD subtype in East Asians compared to Caucasians, driven by ancestry-specific genetic variants affecting β-cell function and tissue-specific gene expression. Furthermore, complication risks differ between regions; while the SIRD cluster remains the primary risk for metabolic dysfunction-associated steatotic liver disease (MASLD) and nephropathy across populations, East Asian SIDD patients face significantly higher risks of chronic kidney disease and sarcopenia than their Caucasian counterparts. These findings emphasize that while clustering provides a robust framework for risk stratification, clinical application in East Asians requires modifications accounting for unique body compositions and pathophysiology. Integrating subtype-guided strategies-such as early intensive insulin therapy for SIDD and multifaceted insulin sensitivity improvement for SIRD-represents a critical step toward personalized medicine to eradicate diabetes complications in East Asians.