Abstract
<p>Brain age gap (BAG) has emerged as a widely used measure of brain health, with applications in neuropsychiatric prognostication and clinical trial enrichment. Over the past eight years, 23 genome‑wide association studies (GWAS) in more than 62,000 adults have advanced the current understanding of genetic contributions to brain ageing. BAG is moderately heritable, with common variants explaining up to 29% of phenotypic variance in the population. This systematic review and re-analysis consolidates current knowledge of the genetic architecture of brain ageing and highlights the divergence in identified variants across brain age estimation methodologies. Across studies, 195 genes have been associated with brain age, with seven (PLEKHM1, MAPT, KANSL1, CRHR1, SPPL2C, ARHGAP27, and STH) consistently identified. These converge on pathways related to neurodevelopment, cytoskeletal stability, stress responsivity, and cellular maintenance. Differences in BAG modelling produce distinct genetic profiles: shallow‑learning and multimodal approaches tend to implicate genes linked to neurodegeneration (e.g., APOE), whereas deep learning of T1-weighted images more often highlights microstructural and intracellular signalling pathways. Functional annotation further suggests that BAG reflects broadly distributed biological processes rather than brain‑exclusive mechanisms, instead capturing aspects of systemic integrity that are relevant to central nervous system function. Accordingly, BAG may be best understood not as a disease‑specific biomarker, but as a quantitative endophenotype reflecting cumulative deviation from normative brain integrity.</p>