Blood-Based Biomarkers Predict Differential Longitudinal Decline in Alzheimer's Disease Psychosis: Evidence from Two...

Psychosis affects 40% of individuals with Alzheimer's disease (AD) and is associated with accelerated cognitive decline. Blood-based biomarkers, particularly plasma phosphorylated tau (ptau), have demonstrated utility in predicting cognitive decline in AD, with ptau217 showing superior performance in many studies. However, whether these biomarkers predict differential cognitive trajectories in AD with psychosis (ADP) remains unknown.

Two independent cohorts were analyzed: Alzheimer's Disease Neuroimaging Initiative (ADNI; n=659: 172 cognitively unimpaired [CU], 406 AD, 81 ADP) and Litwin-Zucker Research Center (LZ; n=142: 68 CU, 57 AD, 17 ADP) with 6-year follow-up. Psychosis was defined by non-zero Neuropsychiatric Inventory delusions or hallucinations scores. In ADNI, plasma ptau181, ptau217, ptau231, amyloid-β42/40, GFAP, and NfL were quantified using NULISA. In LZ, ptau181, ptau205, ptau212, ptau217, amyloid-β42/40, GFAP, and NfL were quantified using Simoa. Linear mixed-effects models assessed prediction of cognitive decline across memory, language, visuospatial, and executive function domains.

In ADNI, baseline ptau181 predicted differential ADP decline in language (p<0.05), visuospatial (p<0.05), and executive function (p<0.05); ptau217 predicted language (p<0.05) and visuospatial (p<0.05) decline; GFAP predicted language (p<0.05) and visuospatial (p<0.05) decline; and NfL visuospatial decline (p=0.01). In LZ, ptau181 predicted decline in memory (p<0.05), language (p<0.0001), visuospatial (p<0.05), and executive function (p<0.05); ptau217 predicted memory (p<0.05) and visuospatial (p<0.05) decline; and GFAP predicted language decline (p<0.05). Johnson-Neyman analyses revealed ADP-AD divergence at low ptau181 thresholds in ADNI, while LZ showed crossover patterns with steeper ADP decline at low biomarker levels that attenuated at high levels where AD decline was steeper.

ADP exhibited accelerated cognitive decline across domains driven by a distinct biomarker landscape compared to non-psychotic AD. Plasma ptau181 demonstrated broader domain-specific associations with decline in ADP than other blood-based biomarkers and associated exclusively with executive function impairment, indicating its unique utility for predicting cognitive trajectories in this pathophysiological subtype.