Brain patterns and risk factors in the FINGER RCT multimodal lifestyle intervention.

Importance: Despite the emergence of anti-amyloid therapies for Alzheimer's disease, targeting modifiable risk factors remains the most effective primary prevention strategy for dementia. While cognitive benefits of multimodal lifestyle interventions have been demonstrated, the underlying effects on brain structure remain unclear, likely due to heterogeneity in brain structure among at-risk individuals.

Objective: To investigate how distinct subgroups of at-risk individuals, defined by cortical and subcortical grey matter (GM) patterns, differ in their response to the FINGER intervention, as well as in their demographic, vascular, and lifestyle profiles.

Design: Observational study employing unsupervised clustering of MRI-based cortical thickness and subcortical volume metrics, followed by longitudinal assessment of a lifestyle intervention.

Setting: The FINGER randomized controlled trial (RCT), a population-based, multidomain lifestyle intervention targeting older adults (aged 60-77) with elevated cardiovascular risk (CAIDE score ≥ 6) and average to slightly below-average cognitive performance.

Participants: A total of 120 participants (61 intervention, 59 control) with available baseline MRI data.

Intervention: Participants were randomly assigned (1:1, double-blind) to a 2-year multidomain lifestyle intervention group - targeting diet, physical activity, cognitive training, social engagement, and metabolic and vascular risk management - or to a control group receiving standard health advice.

Main outcomes and measures: Sociodemographic, vascular, and lifestyle factors, medical comorbidities, and cognitive performance, were assessed at baseline (pre-intervention). Additionally, brain structural outcomes (mean cortical thickness, Alzheimer's disease and resilience-related cortical signatures, hippocampal volume), and cognition (global, executive function, processing speed, memory) were analysed post-intervention using hierarchical linear models stratified by GM cluster.

Results: Clusters with diffuse or frontal-predominant cortical thinning, but with more favourable vascular profiles, characterized by lower blood pressure and reduced obesity, showed significantly less cortical thinning (mean thickness, AD-signature, and resilience-signature regions; all p < 0.05) following the intervention.

Conclusions and relevance: Stratifying at-risk individuals by GM patterns and vascular risk revealed differential brain responses to the FINGER intervention. These findings underscore the value of brain-based subtyping to optimize personalized dementia prevention strategies in heterogeneous at-risk populations.

Trial registration: ClinicalTrials.gov Identifier: NCT01041989.