School of Biomedical Informatics, University of Texas Health Science Center at Houston, Houston, Texas, USA.
Department of Diagnostic and Interventional Imaging, the McGovern Medical School, University of Texas Health Science Center at Houston, Houston, Texas, USA.
Sci Rep. 2020 Mar 26;10(1):5527. doi: 10.1038/s41598-020-62263-w.
Alzheimer's disease (AD) varies a great deal cognitively regarding symptoms, test findings, the rate of progression, and neuroradiologically in terms of atrophy on magnetic resonance imaging (MRI). We hypothesized that an unbiased analysis of the progression of AD, regarding clinical and MRI features, will reveal a number of AD phenotypes. Our objective is to develop and use a computational method for multi-modal analysis of changes in cognitive scores and MRI volumes to test for there being multiple AD phenotypes. In this retrospective cohort study with a total of 857 subjects from the AD (n = 213), MCI (n = 322), and control (CN, n = 322) groups, we used structural MRI data and neuropsychological assessments to develop a novel computational phenotyping method that groups brain regions from MRI and subsets of neuropsychological assessments in a non-biased fashion. The phenotyping method was built based on coupled nonnegative matrix factorization (C-NMF). As a result, the computational phenotyping method found four phenotypes with different combination and progression of neuropsychologic and neuroradiologic features. Identifying distinct AD phenotypes here could help explain why only a subset of AD patients typically respond to any single treatment. This, in turn, will help us target treatments more specifically to certain responsive phenotypes.
阿尔茨海默病（AD）在症状、测试结果、进展速度和磁共振成像（MRI）上的萎缩方面存在很大的认知差异。我们假设，对 AD 的临床和 MRI 特征进行无偏分析，将揭示出许多 AD 表型。我们的目标是开发并使用一种计算方法，对认知评分和 MRI 体积的变化进行多模态分析，以检验是否存在多种 AD 表型。在这项共纳入 857 名受试者（AD 组 n=213，MCI 组 n=322，对照组 CN 组 n=322）的回顾性队列研究中，我们使用结构 MRI 数据和神经心理学评估来开发一种新的计算表型方法，以无偏的方式对 MRI 中的脑区和神经心理学评估的子集进行分组。表型方法是基于耦合非负矩阵分解（C-NMF）构建的。结果，该计算表型方法发现了四种具有不同神经心理和神经影像学特征组合和进展的表型。这里确定不同的 AD 表型可以帮助解释为什么只有一部分 AD 患者通常对任何单一治疗有反应。这反过来又将帮助我们更有针对性地将治疗方法针对某些有反应的表型。
