Comprehensive multimodal analysis of patients with neuroimmunological diseases

中枢神经系统（CNS）的神经免疫性疾病代表了广泛的诊断范围，其中大多数被认为是罕见的疾病。这些疾病的病理生理学知之甚少，有效的治疗方法是零星的。此外，各种神经免疫性疾病的诊断及其与最常见的多发性硬化症（MS）的区别并不是微不足道的。事实上，目前MS的诊断标准缺乏分子特异性，需要排除所有替代诊断。 从现象学诊断成功过渡到分子诊断的医学领域（例如肿瘤学或传染病）加速了有效疗法的发展。这是因为疾病的分子疾病分类学和诊断工具的开发，允许测量个体患者的不同致病机制是精确医学的先决条件。在不了解疾病的患者特异性驱动因素的情况下，临床医生可能会开出患者缺乏靶点的治疗方案，这不仅会导致不可接受的社会成本，而且还会使患者暴露于所应用治疗的副作用，而没有潜在的获益。类似地，如果患者的疾病表现包括一种以上的病理生理机制，则此类患者需要靶向所有贡献过程的组合治疗。因此，我们认为神经免疫疾病的分子疾病分类学的建立和分子测试的发展，可以区分和量化不同的致病过程，导致破坏的中枢神经系统组织的神经免疫学和神经学的治疗进展是必不可少的。 本项目研究神经免疫学疾病科（NDS）神经免疫学CNS疾病诊断性检查患者的鞘内免疫反应和致病机制。本研究的目的是确定免疫介导的CNS疾病中残疾发展的潜在机制，并将其与免疫和/或神经系统对CNS损伤的生理（通常是有益的）反应区分开来。我们已经制定了自然历史协议（09-I-0032），根据该标准，所有未经治疗的疑似免疫介导的CNS疾病患者在NDS接受详细评价，包括收集疾病活动性、严重程度和残疾的临床和临床旁定量指标，标准和新的定量神经成像标记物以及新的免疫学标记物来源于脑脊液（CSF）和血液的（细胞和分子）生物标志物。这些离体研究补充了体外共培养模型，使我们能够定义新的、临床验证的生物标志物的细胞来源，以及理想情况下它们在疾病过程中的功能和作用。 对所有患者进行编码，并以无偏（即盲法）方式进行临床旁、神经成像和分子生物标志物的分析，以确定哪些生物标志物与特定神经免疫疾病或表型相关。我们使用机器学习和数学建模应用于训练队列，并在独立的验证队列中验证新的诊断分类器，模型，临床或成像量表。 本研究的长期目标是获得以下知识：1。允许精确确定个体患者中神经免疫性疾病表达的致病过程;和2.允许对每种已确定的疾病机制进行精确的治疗靶向。这是精确神经学的基础，据此我们可以在治疗上抑制所有致病机制并增强修复机制，以最大限度地减少CNS组织损伤的程度。目前，我们专注于MS的研究，与其他神经免疫疾病作为MS研究的对照。 短期目标是鉴定MS活动的基因、蛋白质、途径（即，MS复发的存在和频率），进展（即，通过成像生物标志物/量表和临床失能反映的CNS组织破坏的量），以及最重要的，MS严重性（即，残疾或CNS组织破坏的增加速度）。我们还努力了解目前FDA批准的MS治疗是否具有疗效，残留疾病活动的机制是什么，如何最好地测量它，以及如何治疗靶向它。MS残疾和严重性的临床测量）仅仅是太不敏感而不能允许检测和验证个体基因或蛋白质对疾病过程施加的小效应大小。因此，我们暂时重新关注临床实用性量表的开发（和验证），通过将新技术（如移动的平台（智能手机和平板电脑）上的应用程序开发）与机器学习策略相结合，显著提高灵敏度和准确性。

Neuroimmunological diseases of the central nervous system (CNS) represent a broad spectrum of diagnoses, most of which are considered rare disorders. The pathophysiology of these diseases is poorly understood and effective therapies are sporadic. Additionally, the diagnosis of varied neuroimmunological disorders and their differentiation from the most common one, multiple sclerosis (MS), is not trivial. Indeed, current diagnostic criteria of MS lack molecular specificity and require excluding all alternative diagnoses. Medical fields that successfully transitioned from phenomenological to molecular diagnoses (e.g. oncology or infectious diseases) accelerated development of effective therapies. This is because molecular nosology of diseases and development of diagnostic tools that allow measurements of different pathogenic mechanisms in individual patients are prerequisites for precision medicine. Without understanding patient-specific driver(s) of the disease, clinicians may prescribe therapies for which a patient lacks target(s), incurring not only unacceptable societal costs, but also exposing patients to side-effects of applied therapies without their potential for benefit. Analogously, if a patient's expression of a disease encompasses more than one pathophysiological mechanism, such patient requires combination treatments that target all contributing processes. Thus, we view establishment of molecular nosology of neuroimmunological diseases and development of molecular tests that can distinguish and quantify different pathogenic processes that lead to destruction of the CNS tissue as essential for therapeutic advances in neuroimmunology and neurology in general. This project studies intrathecal immune responses and pathogenic mechanisms in patients referred to Neuroimmunological Diseases Section (NDS) for diagnostic work-ups of neuroimmunological CNS disorders. The goal of this study is to define the mechanisms underlying the development of disability in immune-mediated disorders of the CNS and to distinguish these from physiological (and often beneficial) responses of the immune and/or nervous systems to CNS injury. We have established natural history protocol (09-I-0032) under which all untreated patients with suspected immune-mediated disorders of the CNS undergo detailed evaluation at NDS, consisting of the collection of clinical and paraclinical quantitative measures of disease activity, severity and disability, standard and novel quantitative neuroimaging markers and novel immunological (cellular and molecular) biomarkers originating from cerebrospinal fluid (CSF) and blood. These ex-vivo studies are supplemented with in-vitro co-culture models that allow us to define cellular origin of novel, clinically-validated biomarkers and, ideally, their function and role in disease process. All patients are coded and analysis of paraclinical, neuroimaging and molecular biomarkers are performed in an unbiased (i.e. blinded) fashion to define which biomarkers are associated with specific neuroimmunological disease or phenotype. We use machine learning and mathematical modeling applied to training cohorts and validate novel diagnostic classifiers, models, clinical or imaging scales, in the independent validation cohorts. The long-term objective of this study is to acquire knowledge that: 1. Allow precise determination of pathogenic processes that underlie expression of a neuroimmunological diseases in individual patients; and 2. Allows precise therapeutic targeting of each identified disease mechanism. This is the basis of precision neurology, whereby we can therapeutically inhibit all pathogenic mechanisms and enhance repair mechanisms to minimize the extent of CNS tissue damage. Currently we focus on studies of MS, with other neuroimmunological diseases serving as controls for MS studies. Short-term goal is to identify genes, proteins, pathways that underlie MS activity (i.e., presence and frequency of MS relapses), progression (i.e., the amount of CNS tissue destruction reflected by imaging biomarkers/scales and clinical disability) and, most importantly, MS severity (i.e., speed of accrual of disability, or CNS tissue destruction). We also strive to understand whether current, FDA-approved treatment of MS are curative, what mechanisms underlie residual disease activity, how to best measure it, and how to therapeutically target it. This work led us to conclude that current clinical scales (i.e., clinical measured of MS disability and severity) are simply too insensitive to allow detection and validation of small effect sizes that individual genes or proteins exert on disease process. Therefore, we temporarily re-focused on development (and validation) of scales of clinical utility with significantly enhanced sensitivities and accuracies, by combining novel technologies, such as App development on mobile platforms (smartphones and tablets) with machine learning strategies.