Neuropsychiatric symptom clusters across the cognitive impairment spectrum in advanced age: a data-driven analysis integrating AD biomarker and clinical profiling

Abstract

Neuropsychiatric symptoms (NPS) are among the most common manifestations of cognitive impairment, yet their heterogeneity complicates understanding of how they relate to Alzheimer’s disease (AD) pathology. While individual NPS have been linked to AD biomarker burden, the interplay between co-occurring neuropsychiatric profiles and AD biomarkers in shaping cognitive and functional outcomes remains insufficiently understood. This study used data-driven clustering to identify NPS clusters in 300 individuals (156 mild cognitive impairment, 112 AD dementia, 32 non-AD dementia) using the 12 domains of Neuropsychiatric Inventory. K-means clustering identified three clusters: Minimal/No NPS (55.7%), Apathy-Behavioral Dysregulation (17.7%), and Depression-Anxiety/Affective (26.7%). Clusters differed significantly in age, sex, education, cognitive status, and diagnostic composition (all p < .05). The Apathy-Behavioral cluster showed advanced cognitive impairment and AD/non-AD dementia predominance, whereas the Depression-Anxiety cluster showed preserved cognition, female predominance, and highest vascular risk factor prevalence. Measured AD pathology (plasma pTau217, CSF amyloid and tau) did not differ across clusters, whereas APOE ε4 carriership was higher in the two symptomatic clusters (both 44.4% vs. 26.0%; p = .032), an association that did not persist after multivariable adjustment. Hierarchical regression revealed five of eight biomarker-outcome associations were largely independent of NPS clusters (change < 10%), with exceptions involving functional outcome: pTau217 (11.4%), tau (17.7%), and APOE ε4 (29.6%). Findings are consistent with a parallel-pathway hypothesis: measured AD pathology and NPS clusters showed largely independent associations with cognition, with only partial overlap in their associations with functional status.