Sex-specific differences in nonlinear associations between glycaemia and brain health in UK Biobank

AimsDiabetes and hyperglycaemia are associated with poorer brain health, but most studies focus on diagnosed diabetes and rarely examine non-linear or sex-specific associations between glycaemia and neuroimaging biomarkers. We investigated whether glycaemia shows non-linear associations with brain structure and whether these differ by sex. MethodsIn 36,321 UK Biobank participants, sex-stratified fractional polynomial models assessed associations of random glucose and HbA1c with brain MRI volumes, adjusted for covariates. Restricted cubic splines and range-restricted linear models assessed robustness and estimated sex-specific slopes. ResultsAssociations of glycaemia with whole brain and grey matter volumes were inverted J-shaped in both sexes, with lower volumes at higher glycaemia and less consistent evidence of lower volumes at the lower end. Predicted whole brain volume peaked at glucose values between 4.4 and 4.7 mmol/L and HbA1c values between 33 and 37 mmol/mol. Restricted cubic splines confirmed non-linearity for glucose with whole brain volume (p_non-linearity=0.006) and grey matter (p_non-linearity<0.001), with evidence of sex interaction (p_interaction=0.02-0.03). For HbA1c, splines supported non-linearity for whole brain volume (p_non-linearity=0.006) and grey and white matter volumes (both p_non-linearity<0.001), with little evidence of sex interaction (p_interaction[&ge;]0.4). White matter hyperintensity volume showed a J-shaped association for HbA1c (p_non-linearity=0.002), with no evidence of sex interaction. Linear models in the descending range indicated steeper glucose-associated declines in whole brain and grey matter volumes in females than males. ConclusionsGlycaemia-brain imaging associations were non-linear. Higher glycaemia was associated with less favourable brain structural profiles, with stronger associations for glucose in females. These findings indicate that associations between glycaemia and brain structure extend across the glycaemic spectrum rather than being limited to clinically defined diabetic states. What is already known about this subject?O_LIDiabetes is associated with adverse brain outcomes including reduced brain volume and greater small vessel disease burden and is an established risk factor for dementia. C_LIO_LIMost neuroimaging studies have focused on diabetes status rather than the full glycaemic continuum and have relied on sex-pooled analyses. C_LIO_LIEmerging evidence suggests glycaemia may relate to health outcomes nonlinearly and that these relationships may differ by sex. C_LI What is the key question?O_LIIs glycaemia associated with structural brain health across its full range, including below diabetes diagnostic thresholds, and do these associations show nonlinear or sex-specific patterns? C_LI What are the new findings?O_LIWe found evidence of nonlinear associations between glycaemia and brain structure in both sexes, with inverted J-shaped associations for whole brain and grey matter volumes. C_LIO_LIHigher glycaemia was associated with less favourable brain structural profiles across the glycaemic continuum, including within the non-diabetic range. C_LIO_LISex differences were observed particularly for glucose, with steeper declines in brain volumes at higher glycaemic levels in females than males. C_LI How might this impact on clinical practice in the foreseeable future?O_LIThese findings reinforce the importance of considering glycaemia as a continuous exposure rather than relying on diagnostic thresholds when evaluating metabolic contributions to brain ageing. C_LI