Association of late‐onset dementia risk scores and Alzheimer’s disease pathology status in preclinical older adults

Abstract Background Dementia risk scores (DRS) for the prediction of Alzheimer’s disease (AD) have the potential to identify at‐risk individuals for secondary prevention trials. To date, risk scores have focused primarily on predicting dementia symptoms rather than AD histopathology. Prediction of biological disease status could inform more precise targeting of at‐risk populations for prevention strategies. In this study, we compare the predictive value of three late‐life DRSs, namely the revised CAIDE (younger and older old) and the Reitz DRS (Reitz et al. 2010). Capacity for distinguishing pathogenic amyloid‐beta status and relation to multi‐domain cognitive performance was explored. Methods 1378 cognitively‐unimpaired participants aged 60‐85 years from the UK Cognitive Health in Ageing Register: Investigational, Observational, and Trial Studies in Dementia Research: Prospective Readiness cOhort (CHARIOT‐PRO) Sub‐study had cerebral amyloid load dichotomized as positive vs negative. The predictive value of the Reitz and CAIDE risk scores for determining amyloid status were tested in separate logistic regression models. Associations of the risk scores with global and multi‐domain cognitive performance (assessed via the RBANS) were further evaluated. Finally, the contributions of individual components of the risk scores on amyloid status and cognitive performance were tested in multiple logistic and linear regression models. Results 19.3% of the participants were amyloid‐positive. The Reitz and CAIDE young‐old (age<75) scores were higher in the amyloid positive versus negative group (median(IQR): 20 (10) vs 15 (11), 1 (2) vs 0 (2) respectively; P<0.05). Higher Reitz and CAIDE younger‐old scores were associated with increased risk of amyloid positivity (OR=1.05 (1.03‐1.08) and 1.17 (1.03‐1.34) respectively; P<0.05). Reitz score alone was negatively correlated with global cognition, memory and visuospatial indices ( r ranges from ‐0.10 to ‐0.14; P<0.05). The potential contributing factors for predicting amyloid status were age, ApoE‐e4, and ethnicity; while diabetes and education predicted cognitive performance. In contrast, nil associations of CAIDE older‐old (age≥75) score with amyloid status or cognitive performance were noted (P>0.05). Conclusions The risk scores tested revealed differential predictive abilities in relation to amyloid pathology status and late‐life cognitive performance. Risk composites focusing on identified genetic and metabolic factors may be useful for more precise participant selection in targeted prevention and risk‐reduction trials.