Genetics-informed bidirectional mapping of glycemic traits and brain phenotypes

Dysglycaemia is linked to brain atrophy, white-matter disruption and dementia risk, yet the directionality and mechanisms of glycemic-brain coupling remain unclear. Here we integrate large-scale GWAS of fasting glucose, fasting insulin, 2-hour glucose, HbA1c and type 2 diabetes with multimodal brain imaging phenotypes and major brain disorders using bidirectional two-sample Mendelian randomization. Across 4,040 tests, we identify 54 glycaemia and brain causal associations (FDR P < 0.05) and uncover a timescale-dependent hierarchy: short-term glycemic traits map to distributed macrostructure and functional-network, whereas long-term glycemic burden preferentially implicates long-range association and commissural white-matter pathways with focal prefrontal-paracentral vulnerability. Reverse analyses reveal 44 brain IDPs and glycaemia effects. Multi-trait Bayesian colocalization and tissue/cell-type eQTL integration nominate shared causal loci enriched in glial and neurovascular pathways. Individual-level UK Biobank analyses validated glycaemia-associated brain structural alterations and non-linear risk patterns for incident outcomes. Together, our results provide a genetically anchored atlas of bidirectional glycemic-neural relationships, linking metabolic dysregulation to multiscale brain vulnerability and disorder risk.