Cognitive Heterogeneity in those with high Alzheimer's Disease Risk

点击翻译按钮获取中文摘要

The lack of an effective treatment for Alzheimer's disease (AD) has led to a call to detect the disease earlier in its course but AD's insidious onset that can span many years, adds complexity to doing so. As a result, the National Institute on Aging (NIA) has identified to understand the heterogeneity of AD, particularly at the asymptomatic stages as a research priority. While there are well-documented high AD risk factors (e.g., age, apolipoprotein E4, cardiovascular risk, amyloid and tau pathology), diagnosis is not inevitable, but it remains unknown why only some of those with high AD risk progress to disease and others do not. We contend that one challenge for answering this question is that by the time traditional AD preclinical symptoms of memory decline and/or hippocampal atrophy emerge, the neurodegenerative trajectory is already on a near irreversible course. We further hypothesize that traditional measurement methods produce crude measures that mask the broader range of clinical expression in the preclinical period and preclude the earliest opportunity to detect the beginning of the neurodegenerative trajectory. In this updated application, we seek to leverage the Framingham Heart Study (FHS) cognitive aging and dementia database, acquired through nearly 7 decades of prospective examination. Unique to FHS since 2005 has been the collection of novel NP indices (error responses, digital metrics such as item-level latencies, fragmented responses). Baseline data were collected at a time when the vast majority of these participants appeared asymptomatic, including those who are at high AD risk, a subset of which have since progressed to incident AD as well as similarly high AD risk subgroups who did not. Through a one year R56, we provide new preliminary data in support of our aims to 1) characterize the cognitive heterogeneity of these high AD risk groups as they do and do not progress to disease, 2) determine whether traditional neuroimaging biomarkers differentiate between progressors and non- progressors and 3) develop novel machine learning methods to identify neuroimaging indices even earlier than traditional MRI measures. We predict that with additional analyses we will identify unique cognitive profiles that better differentiate those at high AD risk who do and do not progress to AD, that the NP profiles of high AD risk progressors will be associated with AD neuroimaging markers (e.g. decline in total brain and hippocampal volume, increase in white matter hyperintensities) while the NP profiles of high AD risk non-progressors will not show similar evidence of brain structure changes. We will further build on our preliminary work of developing an adversarial learning framework to enhance baseline MRI images to serve as better predictors of high AD risk progressor and non-progressor groups than the original images. Results will lead to identification of a broader spectrum of preclinical presentation in those with high AD risk than has been previously recognized and thus better characterize the heterogeneity of NP performance, particularly earlier in the disease course, potentially identifying a critical period in which intervention strategies can mitigate disease risk.