Neurology AdvisorEstimating Brain Age with MRI Helps Stratify Dementia Risk

ℹ️ Observational Association Only Evidence

Researchers evaluated whether brain-predicted age difference, or brain-PAD, could improve dementia-risk estimation beyond cognitive testing and standard visual MRI assessments. The retrospective cohort included 799 memory-clinic patients, comprising 412 with subjective cognitive decline (SCD) and 387 with mild cognitive impairment (MCI). Brain age was calculated from baseline MRI using two pretrained models and compared with chronological age. During a median 3.2-year follow-up, 235 patients progressed to dementia. Higher brain-PAD remained independently associated with progression in the SCD subgroup, but not in patients with MCI after adjustment for visual MRI findings.

Clinical Considerations

• Overall, 235 of 799 patients, or 29.4%, progressed to dementia, including 39 with SCD and 196 with MCI.
• In the full cohort, each higher unit of BrainageR-derived brain-PAD remained associated with progression after adjustment for visual MRI measures and amyloid status.
• Among patients with SCD, brain-PAD retained prognostic significance after adjustment for visual MRI findings (HR, 1.09; 95% CI, 1.04-1.15).
• Adding brain-PAD to the SCD model produced a modest C-index increase from 0.77 to 0.79; after amyloid adjustment, the C-index remained 0.84.
• Brain-PAD was not independently associated with progression in the MCI subgroup after adjustment for visual MRI findings.
• A data-derived SCD threshold had high negative predictive values but low positive predictive values, limiting its readiness for individual clinical prediction.
• The relatively young memory-clinic population and absence of external validation constrain generalizability.

Practice Applications

• Interpret brain-PAD as an emerging prognostic measure rather than an independently validated dementia-prediction tool.
• Recognize that its added value appeared concentrated in patients with SCD rather than those with established MCI.
• Integrate MRI-derived risk estimates with cognitive testing, visual MRI interpretation, amyloid status, and longitudinal clinical assessment.
• Avoid using the reported threshold for clinical decision-making before independent external validation.
• Monitor prospective studies evaluating calibration, reproducibility, generalizability, and effects on patient outcomes.