Department of Neurology, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430030, China.
Department of Neurology, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430030, China.
Mult Scler Relat Disord. 2021 Jul;52:102965. doi: 10.1016/j.msard.2021.102965. Epub 2021 Apr 17.
Brain lesions in patients with myelin oligodendrocyte glycoprotein antibody-associated disease (MOGAD) are indistinguishable from those with relapsing-remitting multiple sclerosis (RRMS) and aquaporin-4 antibody-positive neuromyelitis optica spectrum disorder (AQP4-Ab NMOSD).
Patients with MOGAD, RRMS, and AQP4-Ab NMOSD with abnormal brain lesions were retrospectively reviewed and divided into training and validation sets. Discriminatory models using brain images and demographics were generated to identify optimal predictors using orthogonal partial least square discriminant analysis after principal component analysis (PCA) of clinico-radiological data without a diagnosis. Constructed models were tested in an independent cohort.
PCA of 51 brain scans and demographics from patients (13 MOGAD, 24 RRMS, and 14 AQP4-Ab NMOSD) demonstrated that RRMS was distinct from antibody-mediated conditions. The best predictors between MOGAD and AQP4-Ab NMOSD were poorly demarcated lesions, large abnormalities (both predictive for MOGAD), female sex, disease duration, linear lesions adjacent to the lateral ventricle, and cerebellum involvement (all predictive for MOGAD). Periventricular, ovoid/round, juxtacortical, and callosal lesions; Dawson's fingers; T1 hypointensity (all predictive for RRMS); and fluffy as well as large lesions (for MOGAD) were the best predictors of MOGAD and RRMS. RRMS versus MOGAD and RRMS versus AQP4-Ab NMOSD models exhibited a high predictive value and perfect accuracy (100%), which was validated in an independent cohort. The model of patients with AQP4-Ab NMOSD and MOGAD exhibited lower predictive power but still achieved an accuracy of 90%.
Brain MRI characteristics combined with demographics enables the distinction of MOGAD from RRMS and AQP4-Ab NMOSD. Fluffy and large lesions are relatively specific MRI characteristics in patients with MOGAD with brain abnormalities in Asian countries.
髓鞘少突胶质细胞糖蛋白抗体相关疾病（MOGAD）患者的脑部病变与复发缓解型多发性硬化症（RRMS）和水通道蛋白 4 抗体阳性视神经脊髓炎谱系障碍（AQP4-Ab NMOSD）患者的脑部病变无法区分。
回顾性分析了具有异常脑部病变的 MOGAD、RRMS 和 AQP4-Ab NMOSD 患者，并将其分为训练集和验证集。使用脑图像和人口统计学数据，通过对临床放射学数据进行主成分分析（PCA）后进行正交偏最小二乘判别分析，生成鉴别模型，以确定无诊断的最佳预测指标。构建的模型在独立队列中进行了测试。
对 51 例脑部扫描和患者的人口统计学数据（13 例 MOGAD、24 例 RRMS 和 14 例 AQP4-Ab NMOSD）进行 PCA 分析后，结果显示 RRMS 与抗体介导的疾病有明显区别。MOGAD 和 AQP4-Ab NMOSD 之间的最佳预测指标为边界不清的病变、大病变（两者均提示 MOGAD）、女性、疾病持续时间、毗邻侧脑室的线性病变和小脑受累（均提示 MOGAD）。脑室周围、卵圆形/圆形、皮质下和胼胝体病变；道森氏指；T1 低信号（均提示 RRMS）；绒毛状和大病变（提示 MOGAD）是 MOGAD 和 RRMS 的最佳预测指标。RRMS 与 MOGAD 以及 RRMS 与 AQP4-Ab NMOSD 模型的预测值均较高，准确性均为 100%，并在独立队列中得到验证。AQP4-Ab NMOSD 和 MOGAD 患者模型的预测能力较低，但仍达到 90%的准确率。
脑 MRI 特征结合人口统计学特征可用于区分 MOGAD 与 RRMS 和 AQP4-Ab NMOSD。在亚洲国家，具有脑异常的 MOGAD 患者中，绒毛状和大病变是相对特异的 MRI 特征。
