Accelerated brain aging as a transdiagnostic biomarker: A lifespan MRI brain age study across seven disorders.
Authors
Affiliations (1)
Affiliations (1)
- Department of Psychosomatic Medicine and Psychotherapy, Jena University Hospital, Jena, Germany.
Abstract
Accelerated brain aging has been associated with several neuropsychiatric disorders; however, findings remain inconsistent due to limited lifespan modeling and inadequate age-bias correction. This study aimed to establish normative brain-aging trajectories across the lifespan and quantify disorder-specific deviations in brain-predicted age difference (brain-PAD) in a large, age-diverse sample. Brain-age models were trained on structural MRI data from 25,425 healthy individuals (aged 2-95 years) and validated in 270 independent cross-sectional and 188 longitudinal participants from the Dallas Lifespan Brain Study. Clinical analyses included 1,737 patients with schizophrenia (SZ), major depressive disorder (MDD), bipolar disorder (BD), attention-deficit/hyperactivity disorder (ADHD), chronic substance use disorder (CSUD), Alzheimer's disease (AD), and frontotemporal dementia (FTD), compared with 1,793 age-matched controls. Brain-PAD was computed after age-bias correction, regional morphometric differences were assessed using ANCOVA, and model interpretability was evaluated using SHapley Additive exPlanations (SHAP). The model achieved a mean absolute error of ~5-7 years across validation and independent cohorts. Healthy individuals showed median brain-PAD values near zero, and longitudinal analyses tracked within-person aging. Significantly elevated Δbrain-PAD values were observed in SZ (+7.64 years), FTD (+7.61 years), AD (+3.24 years), BD (+2.08 years), MDD (+2.76 years), and CSUD (+2.91 years), but not in ADHD. Regional analyses revealed shared and disorder-specific morphometric alterations, and SHAP highlighted key contributions from ventricular enlargement and fronto-parietal cortical features. Brain-PAD captures biologically meaningful structural variation across disorders and provides a quantitative framework for investigating lifespan brain aging in psychiatric and neurodegenerative conditions.