Statistical Methods for Kidney Markers as Shared Determinants of Dementia and Physical Disability in Older Adults

点击翻译按钮获取中文摘要

Dementia affects over 44 million adults worldwide, and Alzheimer’s disease (AD) and related dementias (ADRD) account for 60-80% of all cases among older adults. Physical disability is often the final consequence of dementia before death. One-third of dementia cases may be attributable to modifiable factors, and due to unclear benefit of approved AD treatment, there is a need to identify intervention targets to prevent dementia and physical disability. Since both conditions may be preceded by declining cognitive and physical perform- ance by over a decade, shared biological determinants of dual cognitive-physical decline that impact neuro- logical, musculoskeletal, and other organ systems may inform therapeutic targets to prevent dementia and physical disability. Separate research on these endpoints indicates that declining kidney function, which is a model of premature aging, relates to both cognitive and physical decline. Trials of treatments to reverse kidney disease are underway, but prevention may be more effective. Thus, identifying early markers of kidney decline that are related to dual cognitive-physical decline and joint dementia-disability onset in initially health older adults is a key step toward this goal. Beyond known kidney effects on bone, a kidney-brain axis and kidney- muscle axis posit multiple mechanisms by which declining kidney health may lead to dementia and physical disability such as kidney aging, impaired mineral homeostasis, and accumulation of uremic markers. However, epidemiologic studies have not rigorously investigated whether kidney markers of these biological mechanisms affect relations and dynamics between cognitive and physical endpoints. Plus, studies of longitudinal cognitive and physical endpoints are vulnerable to survival bias and unmeasured confounding. Limitations of current statistical methods are a key barrier to accurately quantifying the strength of relations between cognitive and physical endpoints and to identifying markers of shared biological mechanisms to explain these relations. Thus, new statistical methods are needed to overcome these barriers. Specific aims of this proposal are to: 1) quantify relations between cognitive and physical endpoints over time; 2) test relations of kidney markers with cognitive-physical endpoints; and 3) develop/validate a biomarker risk score to jointly and dynamically predict dementia and disability. To this end, we propose to extend novel structural models for multivariate longitudinal and time-to-event outcomes and apply them to harmonized data from 8 cohort studies of >22,000 community- dwelling adults aged at least 65 years. We hypothesize that after addressing survival bias, cognitive and physical endpoints will have stronger relations that are explained and predicted, in part, by markers of kidney aging, impaired mineral homeostasis and by accumulation of uremic markers. New statistical methods developed as essential tools to jointly study cognitive and physical endpoints will be made available to the scientific community. The ultimate public health impact of this project is the potential for novel candidate biomarkers to inform development of new strategies to prevent ADRD and physical disability in older adults.