复发缓解型多发性硬化患者的连接指纹：基于脑连接模式判别的磁共振研究

For multiple sclerosis (MS), a devastating autoimmune disease, it still is a challeges in individual precise diagnosis and objective evaluation. In recent years, the improvement of magnetic resonance imaging (MS) has significantly improved the understanding of this disease: different structural and functional pattern with neuromyelitis optica (NMO) and clinically isolated syndromes (CIS) in MS patients. The recent studies suggest that an individual’s functional brain connectivity profile is unique and reliable, analogous to a connectivity fingerprint. Moreover, the connectivity fingerprint can be used as a disease marker, for the analysis of disease. Base on recents and our studies, we hypothesis that: relapsing-remitting multiple sclerosis (RRMS) have unique connection pattern as " connectivity connections fingerprint ", it could provide information brain damage and plasticity for individual diagnosis and clinical evaluation. In this study, we used this connectome feature vector as a fingerprint to quantify similarities and difference between two cerebral regions by high resolution DTI and resting-state fMRI. (1) To establish the relevance of these connectivity profiles by intrinsic connection and/or white matter fiber, connectivity probability as vector to construct the "connectivity fingerprint ", linear support vector machines for classification and screening group- and individual-level characteristic "connectivity fingerprint " with highly reliability and validity which used to diagnose. (2) To investigate the predictive value between structural and functional "connectivity fingerprint". (3) To investigate the relationship between structural "connectivity fingerprint", functional "connectivity fingerprint" and clinical assessments, to modeling for disease discrimination. Decoding "connectivity fingerprint "，could reveal the pathological mechanisms among the cerebral structure – cerebral function - clinical function.

多发性硬化(MS)的精准个体诊断和病情客观评估仍面临挑战。MRI的进步显著提高对该疾病的认识：患者存在不同于NMO和CIS的脑结构和功能改变模式。而最新研究表明，人脑MR连接模式和指纹一样具有个体独特性，可用于疾病分析。在本组和相关研究基础上，假设：常见的复发缓解型MS患者存在独特的连接模式即“连接指纹”，可提供脑损害和可塑性信息用于疾病判别。为此，本研究拟采集此类患者高分辨DTI和静息态fMRI大样本数据，1)以脑区(节点)间内源性功能连接和白质纤维束结构连接概率为向量构建“连接指纹”，使用线性支持向量机等分类和筛选组/个体水平上具有较高信度和效度的特征“连接指纹”用于诊断；2)建模探讨结构“连接指纹”对功能“连接指纹”的预测价值；3)最后探讨结构和功能“连接指纹”与患者临床功能指标的相关性，建模用于疾病客观评估。从“连接指纹”角度出发，有助于揭示患者脑结构-脑功能-临床功能的改变机制。

多发性硬化(MS)的精准个体诊断和病情客观评估仍面临挑战。MRI的进步显著提高对该疾病的认识：患者存在不同于NMO和CIS的脑结构和功能改变模式。而最新研究表明，人脑MR连接模式和指纹一样具有个体独特性，可用于疾病分析。项目在前期研究基础上，扩大患者采集，含临床孤立综合征(CIS)、多发性硬化(MS)和视神经脊髓炎谱系疾病(NMOSD)患者的高分辨T1WI和静息态功能磁共振成像(rs-fMRI)及DTI等序列数据和临床评估，对多中心（神经免疫疾病多中心）的大样本数据进行“连接指纹”构建及分析，研究发现：（1）在MS急性期，区域性的脑活动趋于规律性，可随着脑萎缩或病变负荷逐渐消失，最终在缓解期呈现出随机性。这一发现部分解释了发生在多发性硬化症患者功能可塑性和临床残疾；（2）多发性硬化存在脑自发活动的异常，在左侧中央前回活动趋于更规则而小脑活动时变性更大；这种变化和病程及脑实质分数具有相关性；（3）9个特征可以用来区分MS的病变和类似的缺血性脱髓鞘病变，LASSO logistic回归模型是表现最好的模型，训练集的AUC、敏感性和特异性分别为0.900 (95% CI: 0.883-0.918)、87.0%、58.9%和95.2%，验证集的AUC、敏感性和特异性分别为0.828 (95% CI: 0.791-0.864)、87.7%、53.6%和94.4%；（4）提取了11个与区分MS和NMOSD有关的特征构成模型，其曲线下面积值、准确度、敏感度和特异度在训练集中分别为1.00，100.0%，100.0% 和100.0%；在测试集中分别为0.944，86.7%，92.1% 和76.9%；在独立的外部验证集中分别为0.902，81.6%，79.7%和86.0%。结果表明，基于脑病灶T2WI的影像组学特征构建的随机森林模型可以有效地区分MS和NMOSD；（5）MS组内局部连接组指纹被试间差异较HC组内被试间差异显著。该项目研究显著提高了多发性硬化患者的诊断，鉴别诊断能力，在大脑功能活动的发现为理解多发性硬化脑功能重塑提供了理论基础。有助于揭示患者 脑结构-脑功能-临床功能的改变机制。