HAI Book 2025 - Flipbook - Page 378
Liu, Peiwei
87
The role of cohort heterogeneity in predicting AD/ADRD cognitive
decline
Peiwei Liu1, Jacinda Taggett1, Theresa Harrison1, Heather Snyder3, Charles DeCarli4,
Prashanthi Vemuri5, Danielle Harvey6, William Jagust1, Laura Baker2, Susan Landau1
1
Department of Neuroscience, University of California, Berkeley, Berkeley, CA, US
Gerontology and Geriatric Medicine, Wake Forest University, Winston-Salem, NC, US
3
Alzheimer’s Association, Chicago, IL, US
4
Department of Neurology, University of California, Davis, Sacramento, CA, US
5
Department of Radiology, Mayo Clinic, Rochester, MN, US
6
Public Health Sciences, University of California, Davis, Sacramento, CA, US
2
Background: We previously reported that tau PET was more predictive of cognition in ADNI (a clinic-based
sample low in comorbidities and ethnoracial diversity) than U.S. POINTER (a community-based sample with high
cardiovascular risk and ethnoracial diversity), and this difference was driven by A´+ individuals. Here, we
investigate the roles of cortical thickness and cardiovascular risk contributing to cohort differences among A´+
older individuals.
Methods: U.S. POINTER (N = 775; 68.9 ± 5.2 years, 62.8% female, 30.6% A´+) and ADNI (N = 405; 69.9 ± 5.4 years,
58% female, 31.1% A´+) were matched by age, sex, Clinical Dementia Rating Scale Sum of Boxes score, and APOE4
status (Table 1). Among the A´+ subset, generalized linear models in each cohort examined baseline whole-brain
tau PET (MK6240 in U.S. POINTER, FTP in ADNI) and Preclinical Alzheimer9s Cognitive Composite (PACC) scores.
We also examined cohort-specific cortical thickness associations with PACC scores and cardiovascular risk
(Framingham Risk Score (FRS)). Models with PACC controlled for age, sex, education, and PACC exposures, while
FRS related models were controlled for education only.
Results: Tau throughout temporal and posterior cingulate regions (Figure 1), and lower cortical thickness in
temporal pole, precuneus, and frontal regions (Figure 2A) were associated with poorer cognition (PACC) in ADNI,
but these associations were much weaker in U.S. POINTER. However, there were more robust cortical thickness
associations throughout frontal, cingulate, parietal, and occipital regions with cardiovascular risk in U.S. POINTER
than ADNI (Figure 2B).
Conclusion: In ADNI, cognition is linked to cortical thickness and tau in AD-related regional patterns that have
been previously observed in predominantly clinic-based samples. In POINTER, however, these associations were
much weaker, and thickness was more closely linked to cardiovascular risk. Cohort recruitment differences,
particularly in diversity and cardiovascular risk factors, impact the generalizability of AD-related brain-cognition
relationships.
Keywords: Alzheimer’s disease, Tau, Cortical thickness, Cognition, Cardiovascular risk
HAI2025 - 378
