Year: 2025
Type: article
Abstract: BACKGROUND: Late-life cognitive trajectories display heterogeneity between individuals and across cognitive domains. We aimed to identify latent classes of cognitive trajectories among three cognitive domains (memory, executive function, and language) and to investigate their association with baseline demographic and clinical characteristics. METHOD: Harmonized data were obtained from 14 longitudinal cohorts of cognitive aging and dementia. Participants were restricted to those with at least 3 observations for each domain (memory, language and executive function). We developed three latent class linear mixed models, whereby each composite domain was modeled as both a linear and quadratic function of age. Separate models were developed with 1 through 5 classes. We assessed goodness of fit through a comparison of class proportions, AIC, BIC, and entropy. Baseline variables were compared across classes, with ANOVA for continuous and Chi-square for categorical variables. RESULT: We included 38,358 participants (73% non-Hispanic white, 41% male, age=73 ± 9 years, education = 15 ± 4, mild cognitive impairment=19%, Alzheimer's disease = 14%, APOE-ε4=31%, APOE-ε2=13%) (Table 1). Across domains, the 4-class model displayed the best performance (Figure 1) including non-decliners, slow decliners, steady decliners, and rapid decliners. The memory classes deviated from the other two domains with the rapid decliners class displaying poor performance at younger ages with a less precipitous decline, whereas the rapid decline group in the other two domains declined rapidly with age. Across domains, the non-decliners had the highest proportion of non-Hispanic Black, cognitively unimpaired, ε2 carriers, and lowest proportion of ε4 carriers. Slow decliners were older and included more ε2 carriers. Steady decliners had the second highest proportion of AD at baseline and had the highest or second highest proportion of ε4 carriers. Rapid decliners had the highest proportion of AD at baseline, and the highest or second highest proportion of ε4 carriers. CONCLUSION: While multiple subgroups move from cognitively normal to dementia, our latent class approach highlights subtypes that decline at different times and at different rates. Future work will seek to integrate disease time into models, clarify the stability of the observed subgroups of decliners, and characterize the neuropathological and neuroimaging features that contribute to the distinct etiologies of decline across domains.
Source: Alzheimer s & Dementia
Authors: Melisa Lara Gomez, Phoebe Scollard, Sarah Biber, Michael L Cuccaro, Jesse Mez, Seo‐Eun Choi, Brandon Klinedinst, Michael L. Lee, Arthur W. Toga, Aditi Sathe, Michelle J. Clifton, Jane Zyski, Alaina Durant, Shannon Turner, Walter W. Kukull, Andrew J. Saykin, Kwangsik Nho, C. Dirk Keene, Shubhabrata Mukherjee, Eric B Larson, Anja Soldan, Marilyn S. S. Albert, Lori L Beason‐Held, Keenan A. Walker, Luigi Ferrucci, Murat Bilgel, Susan M. Resnick, Yang An, Jason J. Hassenstab, Carlos Cruchaga, Jigyasha Timsina, Corinne D. Engelman, Sterling C Johnson, Richard Mayeux, Katherine A. Gifford, Angela L. Jefferson, Mary Ellen I. Koran, Dandan Liu, Kimberly R. Pechman, Adam Brickman, Annie J. Lee, Jennifer J. Manly, Miguel Arce Renteria, Badri N. Vardarajan, David A. A. Bennett, Julie A Schneider, Lisa L. Barnes, Derek B. Archer, Paul K Crane, Logan Dumitrescu, Timothy J. Hohman, Jo Ellen Wilson
Institutions: Vanderbilt University Medical Center, University of Washington, University of Miami, Boston University, University of Southern California, Vanderbilt University, Indiana University Indianapolis, Indiana University School of Medicine, Indiana University – Purdue University Indianapolis, Johns Hopkins University, Johns Hopkins Medicine, National Institutes of Health, National Institute on Aging, Washington University in St. Louis, University of Wisconsin–Madison, NewYork–Presbyterian Hospital, Columbia University Irving Medical Center, Columbia University, Stanford Health Care, Stanford Medicine, Stanford University, Rush University Medical Center, VA Tennessee Valley Healthcare System
Language: English
Subfield: Psychiatry and Mental health
Field: Medicine
Domain: Health Sciences
Open Access status: hybrid
Estimated APC paid: $4,200
OpenAlex ID: W7117321726
FWCI:
Cites:
Cited by:
Related to: