In-vivo MRI-based prediction of TDP43 pathology in aging

摘要 反式反应DNA结合蛋白43(TDP43)的病理，一种罕见的主要蛋白质异常 肌萎缩侧索硬化症和额颞叶变性，现在被公认为一种常见的 与年龄相关的神经病理学，在尸检中发现约50%的老年人。根据最近的说法 证据表明，TDP43在衰老过程中的病理变化与认知功能下降更快和罹患 痴呆症，除了阿尔茨海默氏症和其他与年龄相关的神经病理造成的影响之外。尽管 由于其高频和有害影响，TDP43只能在尸检中被诊断出来，目前有 没有一种方法能够提供关于TDP43在衰老过程中存在的病理的体内准确信息。 拟议项目的总体目标是开发和测试基于体内MRI的TDP43分类器 多模式MRI与病理相结合的老年人衰老病理研究 社区队列。具体地说，我们提出了a)基于EX-C的机器学习来训练TDP43分类器。 老年人脑宏观、微结构和化学特征的活体MRI测量 人，b)在体内翻译这些分类器，以及c)使用纵向临床、体内MRI在体内测试它们， 以及没有痴呆症的老年人的病理数据。我们已经产生了一种独特的体外和体内 Rush Memory and Aging Project(MAP)基础设施中的多模式MRI-病理数据库 (R01AG17917)和宗教秩序研究(ROS)(P30AG010161)，两个纵向、流行病学临床- 招募非痴呆者并具有高随访率和高随访率的老年病理队列研究 尸检率。利用我们的数据库，我们已经产生了令人信服的初步结果，以支持我们的目标。 首先，我们已经证明了体外脑MRI数据可以与体内MRI数据相关联。第二，我们展示了 TDP43的病理与特定的脑MRI特征有关，独立于其他年龄相关的 病理学。第三，基于体外MRI，我们显示出高的TDP43分类性能(AUC=0.81 患有和不患有阿尔茨海默氏病(一种常见的并存病理)的人的特征。 第四，我们证明了从体外MRI分类获得的置信度分数与 TDP43的沉积病理进展(即TDP43分期)。最后，我们翻译了一个初步的体外实验 体内使用的TDP43分类器，并在小样本中证明了它具有很高的体内分类能力 表现，并独立地与较低的认知能力相关联。我们建议进一步开发和测试这一点 TDP43衰老病理的在体MRI分类有希望。

ABSTRACT Transactive response DNA-binding protein 43 (TDP43) pathology, a primary protein abnormality in the rare diseases amyotrophic lateral sclerosis and frontotemporal lobar degeneration, is now recognized as a common age-related neuropathology, detected at autopsy in approximately 50% of older persons. According to recent evidence, TDP43 pathology in aging is associated with more rapid cognitive decline and higher odds of dementia, above and beyond contributions from Alzheimer’s and other age-related neuropathologies. In spite of its high frequency and deleterious effects, TDP43 can only be diagnosed at autopsy, and there is currently no approach that provides in-vivo accurate information about the presence of TDP43 pathology in aging. The overall goal of the proposed project is to develop and test an in-vivo MRI-based classifier of TDP43 pathology in aging by combining multimodal MRI and pathology in the same older persons from large community cohorts. Specifically, we propose to a) train TDP43 classifiers using machine learning based on ex- vivo MRI measurements of macro-structural, micro-structural and chemical brain characteristics of older persons, b) translate these classifiers in-vivo, and c) test them in-vivo using longitudinal clinical, in-vivo MRI, and pathology data on older adults enrolled without dementia. We have generated a unique ex-vivo and in-vivo multimodal MRI-pathology database within the infrastructure of the Rush Memory and Aging Project (MAP) (R01AG17917) and Religious Orders Study (ROS) (P30AG010161), two longitudinal, epidemiologic clinical- pathologic cohort studies of aging that recruit non-demented individuals and have high follow-up rates and high autopsy rates. Using our database, we have produced compelling preliminary results in support of our aims. First, we have demonstrated that ex-vivo brain MRI data can be linked to in-vivo MRI data. Second, we show that TDP43 pathology is related to specific brain MRI characteristics independent of other age-related pathologies. Third, we show high TDP43 classification performance (AUC=0.81) based on ex-vivo MRI features in persons with as well as without comorbid Alzheimer’s pathology (a common coexisting pathology). Fourth, we demonstrate that the confidence score obtained from ex-vivo MRI classification is linked to the progression of deposition of TDP43 pathology (i.e. TDP43 stages). Finally, we translated a preliminary ex-vivo TDP43 classifier for use in-vivo and demonstrated in a small sample that it has high in-vivo classification performance, and is independently associated with lower cognition. We propose to further develop and test this promising in-vivo MRI classifier of TDP43 pathology in aging.