Alzheimer’s disease spectrum profiled by CSF and imaging biomarkers: Tau PET best predicts cognitive decline

Abstract

AbstractBackgroundThe concept of pathophysiologically profiling individuals based on their biomarker status for Aβ42 (A) or tau (T) deposition and neurodegeneration (N) is attracting interest in the Alzheimer’s disease research field. All three processes can be monitored by either cerebrospinal fluid (CSF) sampling or imaging (PET/MRI). We aimed to test the comparability of the CSF and imaging biomarkers in each AT(N) category and their ability to predict longitudinal cognitive decline.MethodA subset of data from 254 patients who had had PET investigations with amyloid‐β and tau tracers, and CSF sampling at the same time point, and a structural MRI within one year was selected from the Alzheimer’s Disease Neuroimaging Initiative trial data. The participants were grouped by clinical diagnosis at that time: cognitively normal (n=88), subjective memory concern (n=82), early (n=35) or late (n=31) mild cognitive impairment, and Alzheimer’s disease (n=18). Agreement between the modalities for A, T and N was assessed with Cohen’s Kappa statistic. Linear mixed‐effects models were used to predict subsequent decline in the episodic memory composite score.ResultBiomarkers for A in PET and CSF showed moderate agreement (Kappa=0.42‐64). Biomarkers for T in PET and CSF showed fair agreement (Kappa<0.40), except in the Alzheimer’s disease group where agreement was moderate (Kappa=0.64). The N biomarkers (CSF vs atrophy in MRI) showed evidence of discordance across all groups (Kappa<0.26). Baseline tau PET positivity predicted longitudinal decline in episodic memory irrespective of the p‐Tau181 (CSF) status (estimate=‐0.02, p≤0.05), but the same was not seen for isolated p‐Tau181 (CSF) positivity (estimate=0.00, p=0.58). Baseline tau and amyloid‐β PET positivity biomarkers predicted decline in episodic memory (estimate=‐0.02, p≤0.0001), but isolated amyloid‐β PET positivity did not (estimate=‐0.01, p=0.09). Isolated tau PET positivity was not observed in the absence of amyloid‐β PET positivity, with the exception of one participant with PET values close to the cutoff point.ConclusionWhile results for amyloid‐β are relatively similar using CSF or imaging, CSF and imaging results for tau and neurodegeneration are not interchangeable. Tau PET positivity was superior to phosphorylated tau and amyloid‐β PET in predicting cognitive decline in the Alzheimer’s disease continuum.