Brain structure age in obstructive sleep apnea: a cross-sectional deep learning analysis

Brain-age methods may summarize structural variation, but it is unclear whether brain structure age (BSA) differs between adults with obstructive sleep apnea (OSA) and controls. We tested global and region-specific BSA differences and associations with clinical measures. This cross-sectional observational analysis included 45 treatment-naive adults with OSA and 45 healthy controls. Participants underwent T1-weighted 3 T MRI and Montreal Cognitive Assessment; OSA participants additionally underwent polysomnography. Images were processed with the volBrain BrainStructureAges pipeline. The primary analysis compared global BSA between groups. Regional comparisons and clinical/polysomnographic correlations were secondary analyses, with Benjamini–Hochberg false discovery rate (FDR) correction applied within each family. A sensitivity regression adjusted the global BSA comparison for chronological age, sex, and BMI. Global BSA did not differ significantly between groups (49.94 ± 11.98 years in OSA vs. 46.89 ± 11.75 years in controls; mean difference, 3.05 years; 95% CI, − 1.92 to 8.02; p = 0.226). The adjusted OSA group effect was 2.05 years (95% CI, −0.67 to 4.78; p = 0.138). Of 125 regional comparisons, 16 had nominal unadjusted p < 0.05, but none survived FDR correction (minimum q = 0.248). Global BSA was nominally associated with Epworth Sleepiness Scale score ( n = 34, r = 0.424, p = 0.013) and REM latency ( n = 34, r = −0.388, p = 0.023), but neither survived FDR correction across 15 clinical/polysomnographic correlations ( q = 0.175 for both). In this cohort, global BSA did not differ between OSA and controls, and no regional or clinical/polysomnographic association remained significant after multiplicity correction. The findings provide no evidence that BSA distinguishes OSA from controls or functions as a clinical biomarker for OSA. Larger longitudinal studies with polysomnography-confirmed controls are needed to determine whether smaller effects were missed.