Mechanism-Based Prediction of Chronic Post-Stroke Pain Using Advanced Neuroimaging Techniques

中枢性中风后疼痛(CPSP)是一种慢性疾病，发生在8%至18%的中风患者中(Andersen等人，1995；Liampass等人，2020)。CPSP在发病、表现和强度上是不同的，不幸的是，很难治疗。药理学试验发现结果喜忧参半：抗抑郁药、抗惊厥药和麻醉剂显示出轻微到中度的改善，但受到作用规模小和重复性差的限制(oh&Seo，2015)。药物治疗也有麻烦的副作用，对患者的生活质量产生负面影响。非药物治疗，如脑深部刺激和重复经颅磁刺激(RTMS)初级运动皮质(M1)可以改善CPSP，但在缓解疼痛方面存在较大的个体差异。一些患者没有受益，但另一些患者几乎完全缓解了疼痛。这种差异的原因尚不清楚，但可能与M1的通用靶向有关。针对与疼痛产生机制相关的大脑区域的治疗可能是更好的选择。然而，这将需要对CPSP的病理生理学有详细的了解。该博士建立在初级躯体感觉皮质的细胞构造区3A、3B和1参与疼痛处理的新发现的基础上，旨在确定有效治疗CPSP的新靶点(Whitsel等人，2019年)。假设当3A过度活跃时，它会导致异常的疼痛处理，导致慢性疼痛。在CPSP中，当对3A的投射变得过度兴奋时，当3B/1区受损，不再通过皮质内连接抑制3A时，或者由于中风引起的胶质细胞和小胶质细胞功能的改变，导致网络水平的异常兴奋性，就可能发生这种情况。由于3A的位置与M1直接相邻，rTMS对M1的刺激可能无意中针对3A，这可能解释了rTMS的一些治疗反应。更有效地抑制3A的干预可能导致CPSP的长期慢性疼痛减轻。目前，关于3A在CPSP发展中的作用的文献尚有空白。该项目旨在确定涉及3A的大脑回路中断是否在CPSP的发生中起重要作用，以及3A是否代表治疗和预防CPSP的有效靶点。具体目标是：(1)识别引起CPSP的功能障碍的大脑回路，(2)使用病变标测和功能遗传学分析来寻找生物标记物，用于预测向CPSP的转变和rTMS的治疗反应，这将有助于精确的医学治疗方法，以及(3)建立机器学习预测算法来预测CPSP的发展和治疗反应。本项目和学生分阶段组织：(1)R Project、Python和MatLab的培训，以进行MRI和fMRI扫描的数据驱动分析，以及学习建立机器学习算法来预测中风患者与CPSP相关的结果，(2)对当前CPSP文献进行系统分析，(3)CPSP患者、非CPSP中风患者和健康对照的数据收集，(4)与CPSP的发展和结果有关的病变标测的数据驱动分析，(5)基于机器学习的预测CPSP的发展和结果的生物标志物的发展，(6)对CPSP中3A传入作用的假设驱动的调查，(7)针对3A区域的实验性rTMS，以减轻中风患者的疼痛。参与者将通过当地的NHS信托基金(安特里大学医院和皇家利物浦大学医院)招募，并可能通过其他能够参与的信托基金招募。由于MRI或CT扫描是中风患者的标准做法，因此将使用患者记录中的一些数据。

Central post-stroke pain (CPSP) is a chronic condition which develops in 8% to 18% of stroke patients (Andersen, et al., 1995; Liampass, et al., 2020). CPSP is heterogenous in onset, presentation, and intensity, and is, unfortunately, difficult to treat. Pharmacological trials have found mixed results: antidepressants, anticonvulsants, and anaesthetics have shown mild to moderate improvements but are limited by small effect sizes and poor reproducibility (Oh & Seo, 2015). Pharmacological treatments also have troublesome side-effects which negatively impact on patient quality of life. Nonpharmacological therapies such as Deep Brain Stimulation and repetitive Transcranial Magnetic Stimulation (rTMS) to primary motor cortex (M1) can improve CPSP but there is large inter-individual variability in pain relief. Some patients experience no benefit but others attain almost complete pain relief. The cause of this variability is unknown but may relate to generic targeting of M1. Treatments targeting brain regions that are mechanistically relevant to pain generation may represent a better option. However, this will require a detailed understanding of CPSP's pathophysiology.This PhD builds on the novel finding of primary somatosensory cortex's cytoarchitectonic areas 3A, 3B and 1 involvement in pain processing, and aims to identify novel targets to effectively treat CPSP (Whitsel, et al., 2019). 3A is hypothesised to contribute to abnormal pain processing, leading to chronic pain, when it is overactive. In CPSP this could happen when projections to 3A become hyperexcitable; when 3B/1 areas are damaged and no longer inhibit 3A through intracortical connections or because of stroke-induced alterations in glia and microglia cell function that cause abnormal network level excitability. As the location of 3A is directly adjacent to M1, rTMS stimulation of M1 could inadvertently target 3A, which could explain some therapeutic responses to rTMS. Interventions which more effectively inhibit 3A could lead to long-term chronic pain reduction in CPSP.There is currently a gap in literature examining the role of 3A in the development of CPSP. This project aims to determine whether disruption of brain circuits involving 3A are important in causing CPSP and also whether 3A represents a valid target for treating and preventing CPSP. The specific aims are to: (1) identify dysfunctional brain circuits that cause CPSP, (2) use lesion mapping and functionnectonomic analysis to find biomarkers for predicting the transition to CPSP and treatment response to rTMS, which will contribute to precision medicine approach to treatment, and (3) build a machine learning prediction algorithm to predict CPSP development and treatment response.This project and studentship is organised in stages: (1) training in R Project, Python and MatLab to undertake data-driven analyses of MRI and fMRI scans, as well as learning to build a machine learning algorithm to predict stroke patient outcomes in relation to CPSP, (2) conducting a systematic analysis of current CPSP literature, (3) data collection of CPSP patients, non-CPSP stroke patients, and healthy controls, (4) data-driven analysis of lesion mapping as it relates to development and outcomes of CPSP, (5) development of machine learning based biomarkers predicting development and outcomes of CPSP, (6) hypothesis-driven investigation of 3A afferents role in CPSP, (7) experimental rTMS targeting of 3A area to alleviate pain for stroke patients.Participants will be recruited through the local NHS Trusts (Aintree University Hospital and Royal Liverpool University Hospital) and potentially through other Trusts that are able to participate. As MRI or CT scanning is standard practice in stroke patients, some data will be used from patients' records.