Background and Purpose—MRI biomarkers play an important role in the diagnostic work-up of dementia, but their prognostic value is less well-understood. We investigated if simple MRI rating scales predict mortality in a memory clinic population.

Methods—We included 1138 consecutive patients attending our memory clinic. Diagnostic categories were: subjective complaints (n=220), mild cognitive impairment (n=160), Alzheimer disease (n=357), vascular dementia (n=46), other dementia (n=136), and other diagnosis (n=219). Baseline MRIs were assessed using visual rating scales for medial temporal lobe atrophy (range, 0–4), global cortical atrophy (range, 0–3), and white matter hyperintensities (range, 0–3). Number of microbleeds and presence of infarcts were recorded. Cox-regression models were used to calculate the risk of mortality.

Results—Mean follow-up duration was 2.6 (±1.9) years. In unadjusted models, all MRI markers except infarcts predicted mortality. After adjustment for age, sex, and diagnosis, white matter hyperintensities, and microbleeds predicted mortality (white matter hyperintensities: hazard ratio [HR], 1.2; 95% CI, 1.0–1.4; microbleeds: HR, 1.02 95% CI, 1.00–1.03; categorized: HR, 1.5; 95% CI, 1.1–2.0). The predictive effect of global cortical atrophy was restricted to younger subjects (HR, 1.7; 95% CI, 1.2–2.6). An interaction between microbleeds and global cortical atrophy indicated that mortality was especially high in patients with both microbleeds and global cortical atrophy.

Conclusion—Simple MRI biomarkers, in addition to their diagnostic use, have a prognostic value with respect to mortality in a memory clinic population. Microbleeds were the strongest predictor of mortality.