Baseline regional Tau distribution predicts fast cognitive decline in subjects with mild cognitive impairment. (20th December 2022)

• Record Type: Journal Article
• Authors: Charil, Arnaud; Devanarayan, Viswanath; Murali, Leema; Qi, Xin; Krause, Stephen; Sachdev, Pallavi; Reyderman, Larisa; Koyama, Akihiko; Dhadda, Shobha; Irizarry, Michael C
• Is Part Of: Alzheimer's & dementia. Volume 18(2022)Supplement 5
• Page Start: n/a
• Page End: n/a
• Publication Date: 2022-12-20
• Subjects: Alzheimer's disease -- Periodicals; Alzheimer Disease -- Periodicals; Dementia -- Periodicals; Démence; Maladie d'Alzheimer; Périodique électronique (Descripteur de forme); Ressource Internet (Descripteur de forme); 616.83
• DOI: 10.1002/alz.068154
• Languages: English
• ISSNs: 1552-5260

Abstract

Abstract: Background: Neurofibrillary tangles represent a pathological hallmark of AD that spreads throughout the brain in a relatively stereotypical spatio‐temporal fashion along Braak stages. Understanding the impact of regional Tau distribution and spread on future cognitive decline would be highly valuable for targeted patient selection and treatment strategy in drug development. Method: Regional flortaucipir (FTP) PET SUVR in 77 brain regions, and Braak stage regions I‐VI, were available from 638 ADNI subjects (89 AD, 117 MCI, 432 age‐matched normal controls). Baseline FTP data, processed at UC Berkeley, was partial volume corrected to reduce off‐target binding contamination. Braak stage II (i.e., hippocampus) was excluded from the analysis due to known off‐target binding in this region. A machine‐learning (ML) approach via stochastic gradient boosting was applied to determine the characteristics of regional Tau distribution that uniquely reflect AD relative to control subjects. This signature was then applied to PET scans of an independent group of 117 MCI subjects to predict their 24‐month cognitive decline, as defined by an increase in the clinical dementia rating sum of boxes (CDR‐SB). MCI subjects predicted to have 'control‐like' Tau deposition signature were considered as slow cognitive decliners while those predicted to have 'AD‐like' signature were considered as fast decliners. Prediction performance was evaluated and the regions of Tau deposition indicative of faster cognitive decline were identified. Result: MCI subjects predicted as fast cognitive decliners had odds ratio of 7.3 relative to those predicted as slow decliners (7.3‐fold greater odds of decline), with the median and 75 th percentile increase in CDR‐SB of 1 and 3 respectively, compared to an increase of 0 and 0.5 among those predicted as slow decliners. Subjects with Tau deposition in Braak I (entorhinal cortex) and Braak III at baseline, and particularly in the amygdala, were predicted to be fast decliners. Further presence of Tau in parahippocampal, fusiform and precuneus regions at baseline were strongly associated with fast cognitive decline. Conclusion: The nature of baseline regional Tau distribution in MCI subjects is a strong indicator of fast cognitive decline that can help inform targeted patient selection and treatment strategy in drug development.