Cluster analysis of neuropathologic features in the National Alzheimer's Coordinating Center data set. (December 2021)

• Record Type: Journal Article
• Authors: Culhane, Jessica E.; Teylan, Merilee A.; Chen, Yen‐Chi; Chan, Kwun Chuen Gary; Mock, Charles; Kukull, Walter A.
• Is Part Of: Alzheimer's & dementia. Volume 17(2021)Supplement 3
• Page Start: n/a
• Page End: n/a
• Publication Date: 2021-12
• 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.054624
• Languages: English
• ISSNs: 1552-5260

Abstract

Abstract: Background: Presence of multiple pathologies associated with neurodegenerative disease at autopsy is common. We investigated whether unsupervised cluster analysis of neuropathologic features resulted in distinct neuropathologic profiles, and compared demographic and clinical features between clusters. Method: We included autopsied participants from the National Alzheimer's Coordinating Center Uniform Data Set whose last visit was within two years of death. Participants were excluded if they had rare neuropathologic diseases or were not evaluated for each neuropathology variable used for clustering. We applied a hierarchical clustering algorithm using amyloid plaques (Thal phase, neuritic plaque density), tau neurofibrillary tangles (Braak stage, frontotemporal lobar degeneration (FTLD) tau presence), vascular pathology (infarcts, microinfarcts, arteriosclerosis, atherosclerosis, cerebral amyloid angiopathy), Lewy bodies (LB), and transactive response DNA‐binding protein 43 (TDP‐43) inclusions. We compared neuropathologic, demographic, and clinical features of the resulting clusters. Result: The algorithm identified seven distinct clusters (Figure 1) of the 1, 156 participants that met inclusion criteria. Four clusters included high levels of Alzheimer's disease neuropathologic change (ADNC), characterized as ADNC+TDP, ADNC+LB, ADNC+FTLD‐tau, and ADNC alone. Two clusters were characterized by high levels of TDP‐43 or FTLD‐tau, but mild burden of other pathologies, and the remaining cluster had mild burden of all pathologies. Vascular pathologies were prevalent across all clusters. In general, primary etiologic diagnosis among those who were cognitively impaired at their last visit corresponded well with the defining neuropathologies of each cluster (Figure 2). In the ADNC+LB cluster, primary clinical diagnosis was rarely Lewy body dementia (LBD). Even in clusters with mild ADNC, Alzheimer's disease was commonly the primary diagnosis. Overall, ADNC+TDP and ADNC+LB clusters had the most impairment at their last visit, with the highest prevalence of cognitive symptoms and lowest scores on cognitive tests. The cluster with mild pathologic burden, while having the highest proportion of participants with normal cognition, also had the highest proportions with mild cognitive impairment. Conclusion: Despite the diverse combinations of pathologies present at autopsy, unsupervised cluster analysis revealed distinct pathologic profiles of neurodegenerative disease. Clusters with mixed pathologies had higher cognitive impairment. Primary clinical diagnoses largely reflected neuropathologic burden, although Alzheimer's disease was commonly diagnosed in clusters with low ADNC burden.