Genome‐Wide Association Analyses Identify Distinct Genetic Architectures for Extreme Early‐Onset and Late‐Onset T2D

AIMS: Type 2 diabetes (T2D) is a heterogeneous disorder with substantial variation in age at onset (AAO). This study aimed to characterize the distinct genetic architectures and biological mechanisms underlying extreme AAO-defined T2D subtypes. MATERIALS AND METHODS: Using 74 795 European-ancestry participants from the UK Biobank, we performed genome-wide association studies (GWAS) of relatively early-onset T2D (eoT2D; AAO < 55 years) and late-onset T2D (loT2D; AAO ≥ 70 years). We investigated subtype-specific genetic loci, SNP-based heritability, genetic correlations, Mendelian randomization (MR)-based relationships, polygenic risk scores (PRS) and phenome-wide association studies (PheWAS). Single-cell transcriptomic data from human pancreatic tissues were further used to evaluate cell-type-specific expression patterns of candidate genes. RESULTS: SNP-based heritability was substantially higher for eoT2D than loT2D (11.2% vs. 6.4%), with eoT2D displaying distinct genetic loci related to β-cell function and insulin regulation, including SLC30A8 and IRS1. By contrast, loT2D showed a comparatively lipid-related genetic profile, featuring APOE-associated signals and expression patterns in immune-related cell populations. Linkage disequilibrium score regression (LDSC) and MR analyses further underscored this divergence: eoT2D exhibited broader genetic overlap with cardiometabolic traits, whereas loT2D showed stronger relationships with traditional metabolic risk factors. Finally, subtype-specific PRSs improved risk discrimination beyond conventional covariates, although their clinical utility warrants further evaluation. CONCLUSIONS: Extreme AAO-defined T2D subtypes exhibit partially distinct genetic architectures, highlighting AAO as an important dimension of T2D heterogeneity and providing a framework for future age-stratified genetic risk assessment.