Insel Philip S, Mattsson Niklas, Donohue Michael C, Mackin R Scott, Aisen Paul S, Jack Clifford R, Shaw Leslie M, Trojanowski John Q, Weiner Michael W
Department of Veterans Affairs Medical Center, Center for Imaging of Neurodegenerative Diseases, San Francisco, CA, USA.
Department of Veterans Affairs Medical Center, Center for Imaging of Neurodegenerative Diseases, San Francisco, CA, USA; Department of Radiology and Biomedical Imaging, University of California, San Francisco, CA, USA; Clinical Neurochemistry Laboratory, Institute of Neuroscience and Physiology, The Sahlgrenska Academy, University of Gothenburg, Mölndal, Sweden.
Alzheimers Dement. 2015 Oct;11(10):1171-9. doi: 10.1016/j.jalz.2014.11.002. Epub 2014 Dec 9.
Alzheimer's disease (AD) is characterized by the accumulation of β-amyloid (Aβ) associated with brain atrophy and cognitive decline. The functional form to model the association between Aβ and regional brain atrophy has not been well defined. To determine the relationship between Aβ and atrophy, we compared the performance of the usual dichotomization of cerebrospinal fluid (CSF) Aβ to identify subjects as Aβ+ and Aβ- with a trilinear spline model of CSF Aβ.
One hundred and eighty-three subjects with mild cognitive impairment and 108 cognitively normal controls with baseline CSF Aβ and up to 4 years of longitudinal magnetic resonance imaging data from the Alzheimer's Disease Neuroimaging Initiative were analyzed using mixed-effects regression. Piecewise-linear splines were used to evaluate the nonlinear nature of the association between CSF Aβ and regional atrophy and to identify points of acceleration of atrophy with respect to Aβ. Several parameterizations of CSF Aβ were compared using likelihood ratio tests and the Akaike information criterion. Periods of acceleration of atrophy in which subjects transition from CSF Aβ negativity to CSF Aβ positivity were estimated from the spline models and tested for significance.
Spline models resulted in better fits for many temporal and parietal regions compared with the dichotomous models. The trilinear model showed that periods of acceleration of atrophy varied greatly by region with early changes seen in the insula, amygdala, precuneus, hippocampus, and other temporal regions, occurring before the clinical threshold for CSF Aβ positivity.
The use of piecewise-linear splines provides an improved model of the nonlinear association between CSF Aβ and regional atrophy in regions implicated in the progression of AD. The important biological finding of this work is that some brain regions show periods of accelerated volume loss well before the CSF Aβ42 threshold. This implies that signs of brain atrophy develop before the current conventional definition of "preclinical AD".
阿尔茨海默病（AD）的特征是β-淀粉样蛋白（Aβ）的积累，伴有脑萎缩和认知衰退。用于模拟Aβ与区域脑萎缩之间关联的功能形式尚未得到很好的定义。为了确定Aβ与萎缩之间的关系，我们将脑脊液（CSF）Aβ的常用二分法用于识别Aβ阳性和Aβ阴性受试者的表现，与CSF Aβ的三线样条模型进行了比较。
对来自阿尔茨海默病神经影像倡议组织的183名轻度认知障碍受试者和108名认知正常对照者进行分析，这些受试者有基线CSF Aβ以及长达4年的纵向磁共振成像数据，采用混合效应回归分析。分段线性样条用于评估CSF Aβ与区域萎缩之间关联的非线性性质，并确定相对于Aβ的萎缩加速点。使用似然比检验和赤池信息准则比较CSF Aβ的几种参数化。从样条模型估计受试者从CSF Aβ阴性转变为CSF Aβ阳性时的萎缩加速期，并检验其显著性。
与二分模型相比，样条模型对许多颞叶和顶叶区域的拟合更好。三线模型表明，萎缩加速期因区域而异，在岛叶、杏仁核、楔前叶、海马体和其他颞叶区域出现早期变化，发生在CSF Aβ阳性的临床阈值之前。
使用分段线性样条为AD进展中涉及区域的CSF Aβ与区域萎缩之间的非线性关联提供了一个改进的模型。这项工作的重要生物学发现是，一些脑区在CSF Aβ42阈值之前就显示出体积损失加速期。这意味着脑萎缩的迹象在当前“临床前AD”的传统定义之前就已出现。
