Department of Neurology, Washington University in St. Louis, St. Louis, Missouri, USA.
Malinckrodt Institute of Radiology, Washington University in St. Louis, St. Louis, Missouri, USA.
Ann Clin Transl Neurol. 2021 May;8(5):1096-1109. doi: 10.1002/acn3.51354. Epub 2021 May 4.
To enable use of clinical magnetic resonance images (MRIs) to quantify abnormalities in normal appearing (NA) white matter (WM) and gray matter (GM) in multiple sclerosis (MS) and to determine associations with MS-related disability. Identification of these abnormalities heretofore has required specialized scans not routinely available in clinical practice.
We developed an analytic technique which normalizes image intensities based on an intensity atlas for quantification of WM and GM abnormalities in standardized MRIs obtained with clinical sequences. Gaussian mixture modeling is applied to summarize image intensity distributions from T1-weighted and 3D-FLAIR (T2-weighted) images from 5010 participants enrolled in a multinational database of MS patients which collected imaging, neuroperformance and disability measures.
Intensity distribution metrics distinguished MS patients from control participants based on normalized non-lesional signal differences. This analysis revealed non-lesional differences between relapsing MS versus progressive MS subtypes. Further, the correlation between our non-lesional measures and disability was approximately three times greater than that between total lesion volume and disability, measured using the patient derived disease steps. Multivariate modeling revealed that measures of extra-lesional tissue integrity and atrophy contribute uniquely, and approximately equally, to the prediction of MS-related disability.
These results support the notion that non-lesional abnormalities correlate more strongly with MS-related disability than lesion burden and provide new insight into the basis of abnormalities in NA WM. Non-lesional abnormalities distinguish relapsing from progressive MS but do not distinguish between progressive subtypes suggesting a common progressive pathophysiology. Image intensity parameters and existing biomarkers each independently correlate with MS-related disability.
利用临床磁共振成像（MRI）定量多发性硬化症（MS）患者正常表现的白质（WM）和灰质（GM）异常，并确定与 MS 相关残疾的相关性。到目前为止，这些异常的识别需要专门的扫描，而这些扫描在临床实践中通常不可用。
我们开发了一种分析技术，该技术基于 WM 和 GM 异常的强度图谱对标准化 MRI 进行图像强度归一化，这些 MRI 是通过临床序列获得的。高斯混合模型用于总结来自 5010 名 MS 患者多国家数据库的 T1 加权和 3D-FLAIR（T2 加权）图像的图像强度分布，该数据库收集了影像学、神经表现和残疾测量数据。
基于归一化的非病变信号差异，强度分布指标可区分 MS 患者和对照参与者。该分析揭示了复发型 MS 与进行性 MS 亚型之间的非病变差异。此外，我们的非病变测量值与残疾之间的相关性大约是总病变体积与残疾之间相关性的三倍，后者是使用患者衍生的疾病步骤测量的。多元模型显示，额外病变组织完整性和萎缩的测量值独特地、大约相等地对 MS 相关残疾的预测有贡献。
这些结果支持这样一种观点，即非病变异常与 MS 相关残疾的相关性比病变负担更强，并为正常表现 WM 中的异常基础提供了新的见解。非病变异常可区分复发型和进行型 MS，但不能区分进行型亚型，提示存在共同的进行性病理生理学。图像强度参数和现有的生物标志物都与 MS 相关残疾独立相关。
