基于多元神经影像表征的帕金森病患者脑深部电刺激疗效的预测研究

Deep brain stimulation (DBS) is an extensively used surgical treatment for advanced Parkinson's disease (PD), but the surgical outcomes following DBS varies from patient to patient. It is difficult to predict patient’s response to DBS, and accurate prediction of response is the major concern in clinical decision. The L-dopa impact test is currently used to predict patient’s response to DBS, but its correlation with clinical improvement is weak. It is urgent to develop new objective quantitative biomarkers to predict the patient’s response DBS. The heterogeneity in pathological mechanisms and progress of disease in PD patients are closely related to the difference in patient’s response to DBS. Our previous results have shown that the new multimodal magnetic resonance technique permits to extract quantitative imaging features reflecting the heterogeneity and progress of disease in PD patients. This project aims to develop and validate a predictive model for response to DBS in patients with PD based on preoperative multivariate neuroimaging features, with the aid of machine learning techniques. The achievements of this project will provide important theoretical references and new strategies for individualized treatment and maximizing DBS benefits for PD patients.

脑深部电刺激（DBS）是中晚期帕金森病（PD）外科治疗的首选方案，但不同患者手术收益差异较大。DBS疗效的精准预测是临床决策中的重点和难点。目前左旋多巴冲击试验是判断DBS疗效的主要指标，但与临床改善程度的相关性不强，亟需开发新的客观定量指标预测DBS疗效。PD患者内在病理机制的异质性和病情进展与DBS疗效差异密切相关。申请者前期研究证实：新型多模态磁共振技术可以发现PD隐匿微观的脑结构功能改变，提供大量反映PD患者内在病理机制异质性及病情进展的神经影像表征。本项目拟采用新型多模态磁共振成像技术，构建PD患者DBS术前的脑结构功能网络，并融合高性能机器学习技术，提取反映DBS术前PD患者内在病理机制异质性和病情进展的神经影像表征，结合DBS术前临床信息与术后疗效信息建立基于多元神经影像表征的DBS疗效预测模型，为实现PD患者个体化治疗及DBS手术收益最大化提供重要理论参考及新策略。

脑深部电刺激（DBS）是中晚期帕金森病（PD）外科治疗的首选方案，但不同患者手术收益差异较大。DBS疗效的精准预测是临床决策中的重点和难点。新型多模态磁共振技术可以发现PD隐匿微观的脑结构功能改变，提供大量反映PD患者内在病理机制异质性及病情进展的神经影像表征。本项目采用新型的多模态磁共振成像技术，构建PD患者DBS术前的脑功能控制网络，提取反映疾病内在病理机制异质性和病情进展的功能控制网络指标，并构建DBS疗效预测模型，发现这可以较好地预测PD患者DBS术后的运动功能改善情况。此外，姿势步态不稳为主型（PIGD)和震颤为主型（TD）是PD常见的运动亚型。PIGD往往表现出更严重的认知问题和更快的进展速度。我们的研究发现PIGD和TD运动亚型海马亚区的萎缩模式不同；PIGD亚型的海马亚区体积降低更明显，并且随着年龄的增加而加速。海马的萎缩介导了两个运动亚型认知的差异。