Combining biological and non-biological markers to develop a model predictive of treatment response for individuals with depression

该学生进行了一项研究，调查炎症生物标志物和临床标志物是否会预测抑郁和焦虑心理治疗的治疗反应，作为他们MRes轮换项目的一部分。研究结果表明，儿童期创伤和精神药物使用率较高，预示着焦虑和抑郁的治疗无反应。因此，该研究得出结论，需要进一步研究，以确定治疗无反应风险较高的患者，并预先将其分配给不同或更密集的治疗。两位导师在更广泛的重性抑郁症和情感障碍的转化研究方面都有丰富的专业知识。更具体地，Cleare教授已经进行了研究，检查反应的临床预测因子，特别是治疗抗性或严重性的标志物（Fekadu et al.，2009）和临床结果的各种生物标志物预测因子。这些包括炎性蛋白（Strawbridge et al.，2015; 2019）、生长因子（Pisoni等人，2018）和遗传多态性（Fischer et al.，2018年）。Cleare教授和Juruena博士都检查了患者的儿童创伤史和下丘脑-垂体-肾上腺（HPA）轴的生物标志物，预测抑郁功能缓解的未来严重程度（例如Fischer等人，2018; Tunnard等人，2014; Juruena等人，2009年; 2010年; 2013年）。该信息也可用于博士研究中心的博士培训赠款的电子提交。该项目的第一部分是一个系统的审查（如适用，采用随机效应荟萃分析）已发表的研究，招募了接受心理治疗和心理与药物联合治疗抑郁症的个体样本，这些研究测量了潜在的非生物学因素。（临床，人口统计学或心理社会学）和生物学（例如炎症和内分泌生物标志物）的治疗前预测因子。最近，全面的系统综述已经检查了抗抑郁药治疗反应的生物学和非生物学预测因子（Perlman等人，2019）和心理治疗反应的生物标志物预测因子（Cristea et al.，2019年）。这些结论表明，总体而言，反应预测因子并不特定于评估的个体治疗，可能代表抑郁症预后不良的一般风险因素。然而，心理治疗是一个相对忽视的领域，最近没有评论评估心理治疗反应的临床预测因素。作为项目的第一部分，进行系统性审查将使学生能够确定哪些因素最有可能在预测模型开发阶段进行测试。整个审查过程分配了一年时间，包括文献范围界定、审查问题定稿和具体方法（10月至12月;协议将于1月注册），运行系统搜索和筛选研究纳入（一月至二月），确定研究的纳入资格（3月至4月），提取数据，为审评结果做准备，并进行质量评估（5月至6月），完成结果（包括任何数据分析; 7月至8月），解释结果的含义，并撰写论文以供发表（9月至10月）。在第一年，学生还将参加由生物统计部门组织的各种研讨会，这些研讨会将发展他们在纵向预测建模和统计编程方面的技能和理解（见下文部分;培训）。第二年的重点是：（1）根据上述审查结果建立模型;（2）协助完成第二组数据收集工作（LQD研究）;（3）准备所有数据以供分析。

The student conducted a study investigating whether inflammatory biomarkers and clinical markers would predict treatment response to psychological therapy in depression and anxiety as part of their MRes rotation project. The study's findings suggested that higher rates of both childhood trauma and use of psychotropic medications were predictive of treatment non-response for both anxiety and depression. Thus, the study concluded that further research is needed in order to identify patients who are at higher risk for treatment non-response and pre-emptively allocate them to different or more intense treatments. Both supervisors have substantial expertise in translational research into major depressive disorder and affective disorders more broadly. More specifically, Professor Cleare has conducted research examining clinical predictors of response, particularly markers of treatment resistance or severity (Fekadu et al., 2009) and various biomarker predictors of clinical outcomes. These include inflammatory proteins (Strawbridge et al., 2015; 2019), growth factors (Pisoni et al., 2018) and genetic polymorphisms (Fischer et al., 2018). Both Professor Cleare and Dr Juruena have examined patients' history of childhood trauma and biomarkers of hypothalamic-pituitary-adrenal (HPA) axis predicting future severity of remission of depression function (e.g. Fischer et al., 2018; Tunnard et al., 2014; Juruena et al., 2009; 2010; 2013). This information may also be used by the Centre for Doctoral Studies for the electronic submission of Doctoral Training Grants. The first part of the project constitutes a systematic review (and if appropriate, with random-effects meta-analysis) of published studies recruiting a sample of individuals undertaking a psychological therapy and combined psychological and pharmacological treatments for depression which measure potential non-biological (clinical, demographic or psychosocial) and biological (e.g. inflammatory and endocrine biomarkers) pre-treatment predictors of subsequent therapy response. Recently, comprehensive systematic reviews have examined biological and non-biological predictors of antidepressant treatment response (Perlman et al., 2019) and biomarker predictors of psychological therapy response (Cristea et al., 2019). These conclude that response predictors are, overall, not specific to individual treatments assessed and may represent generalised risk factors for poor depression prognosis. However, psychological therapies are an area of relative neglect and no recent review has assessed clinical predictors of psychotherapy response. Undertaking this systematic review as the first part of the project will allow the student to determine which factors have the most potential to be tested during a predictive model development stage. One year is allocated for the entire review process, which covers literature scoping, finalising the review question and specific methodology (October - December; protocol to be registered by January), running the systematic search and screening studies for inclusion (January - February), finalising the eligibility of studies for inclusion (March - April), extracting data in preparation for review results and undertaking quality assessment (May - June), completing results (including any data analyses; July - August), interpreting the meaning of results and writing up a manuscript for publication (September - October). During the first year the student will also attend various workshops organised by the biostatistics department which will develop their skills and understanding in longitudinal prediction modelling and statistical programming (see below section; Training). The second year will be focused on: (1) model development, in response to the results of the above review; (2) contributing to the completion of data collection for the second cohort (the LQD study); and (3) preparing all data for analyses.