Validation of the brain health index in the European Prevention of Alzheimer's Dementia cohort

•The Brain Health Index combines different magnetic resonance imaging sequences to quantify vascular and non-vascular aspects of brain health.•We assessed the clinical utility of this novel biomarker in a large-scale cohort.•The BHI and certain measures of cognition are significantly correlated, as are BHI and risk factors for cognitive decline, both implying good construct validity.•However, there is not a significant change in BHI over time in a longitudinal sub-cohort.•The BHI may perform differently in clinical populations, and this is a target for future work. Brain Health Index (BHI) assimilates various MRI sequences, giving a quantitative measure of brain health. To date, BHI validation has been cross-sectional and limited to selected populations. Further large-scale validation and assessment of temporal change is required to understand its clinical utility. Assess 1) relationships between variables associated with cognitive decline and BHI 2) associations between BHI and measures of cognition and 3) longitudinal changes in BHI and relationship with cognitive function. BHI computation involved Gaussian mixture-model cluster analysis of T1, T2, T2*, and T2 FLAIR MRI data from participants within the European Prevention of Alzheimer's Dementia (EPAD) cohort. Group differences (gender- and health-based) were evaluated using independent samples Welch's t-tests. Relationships between BHI, age and cognitive tests used linear regression. Longitudinal analysis (12/24 months) utilised mixed linear regression models to examine BHI changes, and paired BHI/cognition associations. Data from N = 1496 predominantly Caucasian participants (50–88 years old, 43.32% male) were used. BHI scores were lower in those with diabetes (p < 0.001, d = 0.419), hypertension (p < 0.001, d = 0.375), hypercholesterolemia (p < 0.001, d = 0.193) and stroke (p < 0.05, d = 0.512). APOE was not significantly related to BHI scores. After correction for age, cross-sectional BHI scores were significantly associated with all measures of cognitive function in males, but only the Four Mountains Test (4MT) in females. Longitudinal change in BHI and cognition were not consistently related. BHI is a valid marker of cognitive decline and relatively stable over 1-2 year follow-up periods. Further work should assess temporal changes over a longer duration and determine relationships between BHI and cognition in more diverse populations.