Abstract
AbstractBackgroundWhile the temporal dynamics of biomarkers in Alzheimer’s disease (AD) have been studied in much detail, the lack of established methods for accurately staging pre‐dementia individuals in a time‐consistent manner has resulted in uncertain trajectories and lack of accurate estimates on the temporal relations between biomarkers and clinical milestones.MethodsA novel latent‐time disease progression model was developed to estimate a multivariate time‐consistent trajectory describing the typical timeline of clinical scales (CDR‐SB, ADAS‐cog, MMSE) and amyloid PET along the AD continuum based on data from amyloid‐negative (A‐) cognitively unimpaired (CU) and the full cognitive spectrum of A+. For each subject, the model predicted amyloid‐cognition (AC) scores that represented the progression along the estimated timeline. Correlation to unseen variables was used to validate the AC score construct. AC scores were then used as a natural timescale for comparing a range of biomarker trajectories.ResultsUsing 1424 individuals (6236 visits, 4581 follow‐up years) from the Alzheimer’s Disease Neuroimaging Initiative, we estimated a timeline of Alzheimer’s disease that spanned almost 20 years from the earliest stages of CU amyloid PET positivity (AC score 0) to severe dementia (Figure 1).The predicted subject‐level AC scores had the best cross‐domain correlation among all compared staging methods (Table 1).CSF biomarkers (Aβ1‐42/p‐tau181) showed the earliest changes, followed by amyloid PET, plasma p‐tau181 and finally tau PET (Figure 2). In terms of abnormality relative to A‐ CU, CSF Aβ1‐42 reached a population‐level mean value above the A‐ CU 95%‐percentile 1.5 years before amyloid PET. Tau PET reached this level 8 years after amyloid PET (Figure 3). In early pre‐clinical stages, CSF Aβ1‐42 was the biomarker that was most sensitive to change. In the later pre‐clinical and early symptomatic stages amyloid PET was most sensitive followed by tau PET in prodromal stage, and measures of cognition and function in the dementia stage (Figure 4).ConclusionThis analysis supports the AC score as a data‐driven approach for studying biomarker trajectories across the Alzheimer’s continuum. The analysis highlighted the value in using different biomarkers to detect subject‐level changes across disease stages. Replication analyses in the BioFINDER study are ongoing.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have