Author ORCID Identifier
https://orcid.org/0009-0001-4309-8883
Date Available
1-10-2027
Year of Publication
2025
Document Type
Doctoral Dissertation
Degree Name
Doctor of Philosophy (PhD)
College
Public Health
Department/School/Program
Epidemiology and Biostatistics
Advisor
Erin Abner
Abstract
Project 1 - The characteristics associated with parental history of dementia for participants in the National Alzheimer’s Coordinating Center (NACC) are not well documented.
Included participants (n=27,723) were stratified by parental history of dementia and frequencies and means were tabulated by factors of interest. Associations of APOE4 dose with maternal and paternal dementia history were assessed using multinomial regression analysis. We used binary logistic regression to estimate the association between participant cognitive status and parental history of dementia. Among those with diagnosed cognitive impairment, we used multiple linear regression to test the hypothesis that participants with parental history of dementia would have earlier age at onset.
The highest frequency of cognitive impairment diagnosis was found among those reporting no history of parental dementia (61%), and the lowest frequency was found among those with a history of both parents (50%). In the logistic regression analysis, participants reporting both parents with dementia had lower adjusted odds of cognitive impairment (OR 0.64, 95% CI: 0.56-0.72) compared to those with no reported parental history. Consistent with reporting no family history, these participants also had the lowest frequency of 1 or more ε4 alleles (35%), compared to volunteers with maternal (47%), paternal (47%), or a history of both parents (52%).
Researchers should strongly consider including parental history of dementia in addition to APOE genotype or e4 dose as control variables when using the NACC UDS data, as our results suggest each provides different information about genetic risk for dementia.
Project 2 - Parental history of dementia may lead to participation bias and has been largely replaced in analyses with genetic markers such as APOEe4. We studied how the inclusion of parental history of dementia affects associations between depression and cognitive status.
Participants without dementia were stratified by cognitive and depression status at their initial study visit (“baseline”): normal cognition without depression (n=7,592), normal cognition with depression (n=2,346), MCI without depression (n=2,280), and MCI with depression (n=1,525). Logistic regression and Fine and Gray models, with death of the participant treated as a competing event, were used to analyze whether parental history of dementia affects the association between depression and cognitive status at baseline (n=20,796) and over time. Fine and Gray models were used for participants with normal cognition at baseline and then again for NACC participants with MCI at baseline to estimate associations for the conversion to MCI or dementia respectively. Models included APOEe4 dose, participant demographics (age, sex, race, and education), in addition to self-reported depression and depressive symptoms.
The adjusted odds of MCI were increased for those with depression at baseline, OR=2.51 (95% CI: 2.36-2.68), and inclusion of parental history of dementia in the logistic regression analysis slightly strengthened the association (OR=2.55 [2.39-2.73]). Selection bias was evident, as participants reporting both parents with dementia had the lowest adjusted odds of cognitive impairment (OR 0.65, 95% CI: 0.57-0.73) compared to no reported parental history. In the competing risk analysis, among those with normal cognition at baseline, those with depression had an increased risk of converting to MCI or dementia, subdistribution hazard ratio (sHR)=1.47 (95% CI: 1.31-1.64). However, inclusion of parental history of dementia did not alter the estimates significantly in any of the Fine and Gray models.
There may be substantial heterogeneity among NACC participants with and without parental history of dementia. In our longitudinal analysis, parental history of dementia did not alter the association between depression and conversion to worse cognitive status. This suggests parental history of dementia may lead to differences in participation but may not confound longitudinal associations of interest.
Project 3- This study included deceased participants from the UK-ADRC who consented to a postmortem brain autopsy. A cross-sectional analysis was performed on two autopsy cohorts, one group of subjects were autopsied from 1990-2011 (n=560) using different types of silver stains and the second group of cases were autopsied from 2012-present (n=374) using immunohistochemical stains and digital pathology counting methods. Subjects were classified as either normal or demented at autopsy based on the most recent ADRC clinician assessment and autopsy Braak staging data. Quantitative assessment of NP, DP, NFTs, amyloid density, NP burden, and tau burden was performed in all of the following brain regions: frontal, temporal, parietal, occipital, amygdala, entorhinal, hippocampus, subiculum, anterior cingulate, and posterior cingulate. Sex differences of the log transformed counts and densities of brain pathologies were analyzed using multiple linear regression models while controlling for APOEe4 dose, age at death, and education. After analysis, the adjusted least squared means and confidence intervals returned to the anti-log scale for interpretation.
We found similar levels of brain pathology among men and women with normal cognition, independent of the neuropathological assessment methods. Average MMSE scores were similar among men and women with dementia in both autopsy cohorts, but we still found women with dementia had more AD pathological changes related to amyloid density, NP, and pathological Tau accumulation in many of the brain regions associated with clinical signs of dementia. We found women vs men had higher counts of NFTs under the previous autopsy protocol in all brain regions except the subiculum and under the current autopsy protocol women had significantly more tau burden in all areas of the brain except for the amygdala and entorhinal regions. Under the new autopsy protocol, women had an approximately 50% higher NP burden and Tau burden in the frontal, temporal, parietal, and occipital brain regions compared to men as well as less pronounced burdens in other brain regions.
These results suggest sex differences in neurofibrillary pathology accumulation over time and may be connected to overt signs of dementia. Survival bias may explain the discrepancy between the increased prevalence of AD among women without a conclusive increase in risk of developing AD. Further research is needed to determine if these sex differences are in fact related to AD brain pathology or attributable to selection or survival bias.
Digital Object Identifier (DOI)
https://doi.org/10.13023/etd.2024.540
Recommended Citation
Guilliams, Chad, "PARTICIPANT CHARACTERISTICS, SELECTION BIAS, AND SEX DIFFERENCES IN BRAIN AGING AND DEMENTIA STUDIES." (2025). Theses and Dissertations--Epidemiology and Biostatistics. 46.
https://uknowledge.uky.edu/epb_etds/46