Neuropsychiatric symptoms (NPS) — including aggression, psychosis, anxiety, apathy, and depression — affect a large proportion of people with Alzheimer’s disease (AD) and drive disability, faster decline, institutionalization, mortality, and healthcare costs. Prior characterizations of NPS prevalence have typically relied on self-reported race/ethnicity, but whether NPS burden differs when groups are defined by genetically inferred ancestry is unclear. If self-reported race misaligns with genetic ancestry, population-level estimates, risk stratification, screening strategies, and trial enrollment could be biased. The authors sought to determine how genetic ancestry assignment versus self-reported race affects inferences about NPS prevalence in AD.
The analysis used whole-genome sequencing data from 7,118 participants drawn from the Alzheimer's Disease Sequencing Project (ADSP). Neuropsychiatric symptoms were assessed using the NPI-Q and grouped into three clusters for analysis: early psychosis (clinical dementia rating, CDR 0.5–1), late psychosis (CDR 2–3), and affective symptoms. The study leveraged available clinical and demographic data linked to the sequencing dataset; no new participant recruitment occurred for this secondary analysis.
Genetic ancestry was inferred by principal component clustering of whole-genome sequence data, producing six ancestry groups identified by the authors: EUR (European), AFR (African), EAS (East Asian), SAS (South Asian), AMR (Admixed American), and ADMIXED. The authors compared these genetically derived assignments with participants’ self-reported race/ethnicity to quantify discordance and to examine how assignment method changed prevalence estimates for NPS clusters.
NPS measures from the NPI-Q were organized into three clinically meaningful clusters reflecting psychosis and affective symptom domains and stratified by dementia severity using the CDR scale. Early psychosis corresponded to milder dementia stages (CDR 0.5–1), and late psychosis to more advanced stages (CDR 2–3). Affective symptoms were treated as a separate cluster for prevalence comparisons across ancestry groups.
Prevalence of NPS clusters was compared across genetically inferred ancestry groups and across genetic ancestry versus self-reported race/ethnicity. The authors used Fisher’s exact tests and regression models to evaluate differences in NPS prevalence between groups and to assess the impact of ancestry assignment on inferences about symptom burden. Specific model covariates and full regression specifications were described in the full manuscript and supplementary material.
Genetic ancestry assignment differed markedly from self-reported race/ethnicity in this ADSP cohort, and this discordance materially affected estimates of NPS prevalence. Prevalence of affective symptoms and psychosis varied across genetic ancestry groups. Reported prevalence values in the abstract include: affective symptoms highest in EAS (90%) and SAS (77%) and lowest in AFR (66%); psychosis prevalence highest in EAS (74%) and SAS (70%) and lowest in AMR (55%) and EUR (56%). Similar patterns were observed for both early and late psychosis clusters. The authors conclude that using genetically defined ancestry yields different population-level estimates compared with relying on self-reported race.
The authors interpret these findings to mean that standard race categories based on self-report may obscure true differences in NPS burden across population groups. The observed ancestry-associated differences imply that partly distinct genetic and environmental drivers may underlie variation in NPS prevalence. From a research and clinical perspective, the results support incorporating genetic ancestry into AD research design, risk stratification, screening approaches, and clinical trial planning to avoid misestimation of disease burden and to improve group-specific understanding of neuropsychiatric manifestations.
This work is reported as a preprint and has not been peer reviewed; the authors explicitly note the preprint status and caution that the findings should not be used to guide clinical practice until validated. The analysis used only de-identified, previously consented data from multiple contributing studies; original studies obtained informed consent and IRB approvals. Details of full regression models, effect estimates, and potential confounder adjustments are provided in the full manuscript and supplementary materials. Individual-level ADSP data are controlled-access and available through the NIAGADS Data Sharing Service with an approved data access request.
Funding sources reported include National Institute on Aging grants (P30AG066462, U01AG079850, R01AG064614, U24AG074855, R01AG062268) and a National Institute of Mental Health grant (R01MH120794). The authors declared no competing interests. Ethical oversight for the secondary analysis was not required because only de-identified data were used; each primary study contributing data had IRB approval and informed consent.
Note: All factual details in this summary are taken from the source preprint. The manuscript is a preprint and has not completed peer review; additional methodological detail and full results are available in the full preprint and supplementary material.