Evaluating Longitudinal Changes in the Human Structural Connectome in Relation to Cognitive Aging

评估与认知衰老相关的人体结构连接组的纵向变化

• 批准号: 9925718
• 负责人: Elliot Max Tucker-Drob
• 金额: $ 47.05万
• 依托单位: UNIVERSITY OF TEXAS AT AUSTIN
• 依托单位国家: 美国
• 财政年份: 2017
• 资助国家: 美国
• 起止时间: 2017-09-01 至 2022-04-30
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/9925718
• 关键词: 3 year old; 3-Dimensional; Adult; Age; Aging; Algorithms; Birth; Brain; Brain region; Cognitive; Cognitive aging; Coupled; Coupling; Data; Data Analyses; Data Collection; Data Set; Dementia; Diffusion Magnetic Resonance Imaging; Elderly; Ensure; Equation; Genetic; Graph; Health; Human; Image; Impaired cognition; Individual; Intervention; Left; Life Style; Link; Longevity; MRI Scans; Magnetic Resonance Imaging; Measures; Mediating; Medical; Memory; Methods; Modeling; Nerve Degeneration; Outcome; Participant; Physiological; Process; Property; Quality of life; Reaction Time; Reproducibility; Research; Risk Factors; Sampling; Scanning; Self Care; Speed; Structure; System; Testing; Theoretical model; Time; Validation; Variant; Visuospatial; Work; aged; biobank; cognitive ability; cognitive change; cognitive performance; cognitive testing; cohort; connectome; feature selection; functional independence; genetic predictors; improved; indexing; interest; longitudinal analysis; machine learning algorithm; machine learning method; mortality; network architecture; novel; physical conditioning; predictive test; prevent; social; sociodemographic predictors; sociodemographics; theories; tool

PROJECT SUMMARY Progressive aging-related cognitive declines are associated with limitations in self-care and functional independence, deteriorating physical health, and impending dementia and mortality, even among the otherwise healthy. Identifying and understanding the neurodegenerative processes that underlie cognitive aging is key to developing interventions to prevent or ameliorate cognitive decline. Disconnection theories of aging specifically implicate weakening of structural brain connectivity as a key mechanism of cognitive decline, but until recently, diffusion MRI data and connectomic methods needed to rigorously test such theories have been lacking. To expedite understanding how aging-related changes in the human structural connectome relate to aging-related cognitive declines, we will apply the latest connectomic and multivariate data analysis methods to existing data from two highly unique datasets: (1) The UK Biobank, a cross-sectional sample of ~10,000 40-75 year old adults, who have undergone diffusion MRI scanning, have been measured with multiple cognitive tests, and have provided extensive sociodemographic and medical information; and (2) The Lothian Birth Cohort of 1936, a narrow-age cohort of older adults (baseline age = 73 years; N = 731) who have undergone diffusion MRI scanning, have been measured with multiple cognitive tests, and have provided extensive sociodemographic and medical information on each of three separate occasions, each separated by three years. Using recently developed graph-theoretic models, we will construct structural brain connectome networks for each participant's diffusion MRI data at each wave and extract indices reflective of network topology within several specific networks of interest (NOIs) identified ex ante. We will also identify topologically central hub regions that disproportionately govern efficiency within each individual's connectome network. We will apply cross-sectional and longitudinal structural equation models to examine aging-related transformations in network indices, examine concurrent and longitudinal coupling between network indices and cognitive abilities, and test predictors of levels and changes in network indices and cognitive abilities. This will allow us to contrast the predictive utility of the selected NOIs for cognitive aging and to identify specific features of network architecture involved in cognitive aging and mediate the effects of demographic, medical, and lifestyle risk factors for cognitive aging. We additionally implement machine- learning methods to estimate an upper bound of prediction of cognitive aging from network indices, and identify novel features of network topology as candidate mechanisms of cognitive decline. The availability of two uniquely large and well-characterized datasets will allow us to ensure that findings are rigorous and reproducible using within sample (holdout) and between sample cross-validation. For all aims, we will place considerable emphasis on testing for incremental validity of network indices relative to both conventional structural neuroanatomical measures and topologically naïve summary indices of network integrity.

项目摘要 渐进的与衰老相关的认知下降与自我保健和功能的局限性有关 独立，身体健康恶化以及即将来临的痴呆症和死亡率，即使在其他 健康。识别和理解认知衰老的基础的神经退行性过程是 开发干预措施以预防或改善认知能力下降。衰老的断开理论 实施结构性大脑连接性的弱化，作为认知能力下降的关键机制，但直到最近， 缺乏严格检验此类理论所需的扩散MRI数据和连接方法。到 加快了解与衰老相关的与衰老相关的人类结构连接的变化如何变化 认知下降，我们将应用最新的连接元和多元数据分析方法 来自两个高度独特数据集的现有数据：（1）英国生物库，一个横截面样本 经过扩散MRI扫描的〜10,000 40-75岁的成年人已通过 多次认知测试，并提供了广泛的社会人口统计学和医学信息；和 （2）1936年的Lothian出生队列，狭窄的老年人队列（基线年龄= 73岁； n = 731）经过扩散MRI扫描，已通过多个认知测试测量， 并在三个单独的每个人中提供了广泛的社会人口统计学和医学信息 场合，每人分离三年。使用最近开发的图理论模型，我们将 在每个波浪和 提取索引反映了几个特定感兴趣网络（NOI）的网络拓扑的索引 赌注。我们还将确定拓扑中心的中心区域，这些区域不成比例地控制着每个区域 个人的连接网络。我们将应用横截面和纵向结构方程模型 检查网络指数中与衰老相关的转换，检查并发和纵向耦合 在网络指数和认知能力之间，以及测试网络指数水平和变化的预测指标 和认知能力。这将使我们能够对比所选NOI的认知衰老的预测效用 并确定与认知衰老有关的网络体系结构的特定特征，并调解 认知衰老的人口，医疗和生活方式风险因素。我们还实施机器 - 学习方法以估算网络指数认知老化预测的上限，并确定 网络拓扑的新特征是认知能力下降的候选机制。两个的可用性 独特的大型且特征良好的数据集将使我们能够确保发现很严格，并且 可在样品（保留）和样品交叉验证之间可重现。对于所有目标，我们将放置 相对于两个常规的网络指数的增量有效性的测试非常强调 网络完整性的结构神经解剖学测量和拓扑幼稚的摘要指标。