Personalizing COPD therapies by exploring heterogeneity of treatment effects

 描述(由申请人提供)：通过确定治疗效果的异质性来个性化COPD治疗项目摘要/摘要慢性阻塞性肺疾病(COPD)是美国第三大主要死亡原因，影响到近2400万美国人1。慢性阻塞性肺病是一种异质性疾病，很难个性化治疗决定。现有的治疗建议3受到气流阻塞严重程度的严重影响，而气流阻塞不能很好地预测结果，如病情恶化和死亡3。用于预测疾病结果的多维指数4、5和治疗建议说明了病情加重和呼吸困难3，但没有考虑慢性阻塞性肺疾病常见的合并症、虚弱和与年龄相关的情况。例如，两个FEV1%预测和恶化频率相似的患者将接受相同的治疗，而不考虑使用长效支气管扩张剂6的心血管事件或使用吸入类固醇7的肺炎的风险。实际上，建议忽略了可能存在从某些治疗中受益或多或少超过随机临床试验(RCT)报告的平均结果的个人。这种治疗效果的异质性是一个新的概念9，10，这是因为随机对照试验报告的是平均参与者的平均结果，忽略了基线风险的差异10以及治疗受益因基线风险而异的可能性。事实上，随着基线风险的相对风险不断降低，对于风险最低的患者来说，治疗的绝对益处将会更低，因此风险最低的患者可能会从治疗中产生更多的不利影响。这一概念以前没有在COPD中探索过，它与确定不同COPD疗法的合适候选者有关。例如，尽管阿奇霉素减少了RCT11的恶化，但由于担心副作用12，13，它没有被治疗指南采用。亚组分析和对COPD表型的研究已经确定了影响阿奇霉素和罗氟司汀疗效的患者特征。然而，使用多变量模型评估治疗效果的异质性将提供更多微妙的信息来指导个人治疗，正如其他疾病16、17所证明的那样。类似地，现有的治疗指南和多维指数没有解决COPD18-20中“过早衰老”的越来越多的证据，这是一个重要的考虑因素，因为COPD患者经常在视觉、听力、记忆、大小便失禁、跌倒、骨质疏松症方面存在依赖日常生活活动的风险(以下均称为老年性疾病21)。这些因素改变了基线风险，可能导致治疗反应的异质性22，但到目前为止，这些因素一直被忽视。大多数老年人的情况可以在办公室进行评估，而骨质疏松症可以在临床指示的胸部影像检查中进行评估。将老年疾病纳入治疗效果异质性的模型中，将使我们能够最大限度地提高COPD患者个体的治疗效益。与NHLBI将研究转化为实践的目标相一致，我们建议重新分析COPD治疗的随机对照试验，以确定治疗效果的异质性，并测试增加老年疾病数据将产生稳健和适用的个性化风险模型的假设。 我们假设，使用COPD预后的多变量模型将使我们能够确定最有可能从COPD干预中受益且不太可能遭受伤害的个人，老年状况将独立地增加COPD预后不良的风险，并且纳入老年状况将导致更可靠的风险模型。这项以患者为中心的建议有两个关键目标。首先，我们将使用COPD结局的多变量模型重新分析随机对照试验，以评估治疗效果的异质性，估计COPD治疗对个体的益处和风险是如何变化的。其次，我们将前瞻性地确定老年疾病在COPD中的负担，并测试它们对更准确的结局风险模型的贡献，这些模型可以在未来用于类似的分析，并设计更有效和更有信息量的随机对照试验。我们的具体目标是：1)使用慢性阻塞性肺疾病结果的风险模型来确定慢性阻塞性肺疾病治疗效果的异质性；2)确定≥老年疾病在慢性阻塞性肺疾病中的频率；以及3)确定老年条件对2年慢性阻塞性肺疾病结果的独立贡献。这些研究的完成将产生COPD结果的新的和强大的模型，允许识别最有可能从干预中受益或遭受伤害的受试者；这是制定个性化决策的第一步。马丁内斯博士作为一名受过公共卫生、卫生服务和转化研究培训的肺科医生，拥有独特的背景，并致力于研究和学术生产力。这项建议包括一项教育计划，使他能够弥补培训中的空白：专注于临床试验和纵向研究的新颖分析、肺部疾病的老年方面以及结果预测方面的专业知识。这些经历将为他成为一名以老年和个性化医学为重点的转化研究的独立研究员奠定基础。马丁内斯博士将得到一个杰出的多学科指导团队的支持：桑迪普·维扬博士是为个性化药物的交付开发模型的先驱，而梅兰·K·韩博士是慢性阻塞性肺疾病表型鉴定的领导者。老年医学的领导者Neil B.Alexander博士、Caroline R.Richardson博士和Christine T.Cigolle博士以及COPD治疗方法的领导者Fernando Martinez博士将担任顾问。

 DESCRIPTION (provided by applicant): Personalizing COPD therapies by determining heterogeneity of treatment effects Project Summary/Abstract Chronic obstructive pulmonary disease (COPD) is the 3rd leading cause of death in the country, affecting nearly 24 million Americans1. COPD is a heterogeneous disease2, making it difficult to personalize treatment decisions. Existing treatment recommendations3 are heavily influenced by severity of airflow obstruction, a poor predictor of outcomes such as exacerbations and death3. Multidimensional indices to predict disease outcomes4,5 and treatment recommendations account for exacerbations and dyspnea3, but do not account for comorbidities, frailty and age-related conditions that are common in COPD. For example, two patients with similar FEV1% predicted and exacerbation frequency will receive the same therapy, regardless of risk for cardiovascular events with long-acting bronchodilators6 or pneumonia with inhaled steroids7,8. Indeed, the recommendations overlook that there are likely to be individuals who will benefit more or less from certain therapies than the average outcome reported from randomized clinical trials (RCTs). This heterogeneity of treatment effects is a novel concept9,10, explained by the fact that RCTs report average results for the average participant, ignoring differences in baseline risk10 and the possibility that treatment benefit varies by baseline risk. Indeed, with a constant relative risk reduction across baseline risk, the absolute benefit of treatment will be lower for te lowest risk patients, so that those at lowest risk may have more adverse than beneficial effects from treatment. This concept, not previously explored in COPD, is relevant to identifying appropriate candidates for different COPD therapies. For example, although azithromycin reduced exacerbations in a RCT11, it has not been adopted by treatment guidelines3 due to concerns about side effects12,13. Subgroup analysis and research on COPD phenotypes have identified patient characteristics impacting outcomes with azithromycin and roflumilast14,15; however, evaluating heterogeneity of treatment effects using multivariate models will provide more nuanced information to guide individual treatment, as has been demonstrated for other diseases16,17. Similarly, existing treatment guidelines and multidimensional indices do not address the growing evidence of "premature aging" in COPD18-20, an important consideration as COPD patients frequently have impairments in vision, hearing, memory, incontinence, falls, sarcopenia and are at risk for dependency for activities of daily living (all hereafter termed geriatric conditions21). These factors change the baseline risk and are probable contributors to heterogeneity of treatment response22, but have been, so far, overlooked. Most geriatric conditions can be assessed in the office, and sarcopenia can be assessed in clinically-indicated thoracic imaging studies. Incorporating geriatric conditions into models of heterogeneity of treatment effects will allow us to maximize treatment benefits for individual COPD patients23,24. Aligned with the NHLBI goal of translating research into practice, we propose to re-analyze RCTs of COPD treatments to identify heterogeneity of treatment effects and test the hypothesis that adding data on geriatric conditions will yield robust and applicable personalized risk models. We hypothesize that using multivariate models of COPD outcomes will allow us to identify individuals most likely to benefit and less likely to experience harm from COPD interventions and that geriatric conditions will independently contribute to risk of poor COPD outcomes, and that inclusion of geriatric conditions will result in more robust risk models. This patient-orientd proposal has two key objectives. First, we will use multivariate models of COPD outcomes to re-analyze RCTs to assess heterogeneity of treatment effects, to estimate how the benefits and risks of COPD treatments vary for individuals. Second, we will prospectively determine the burden of geriatric conditions in COPD and test their contribution to more accurate risk models of outcomes, which can be used in the future for similar analysis and to design more efficient and informative RCTs. Our specific aims, focused on individuals ≥50 years old with COPD, are: 1) Use risk models of COPD outcomes to identify heterogeneity of treatment effects in COPD; 2) Determine the frequency of geriatric conditions in COPD; and 3) Determine the independent contribution of geriatric conditions to 2-year COPD outcomes. Completion of these studies will generate novel and robust models of COPD outcomes, allowing for identification of subjects most likely to benefit or experience harm from interventions; a first step in developing personalized decisions. Dr. Martinez has a unique background as a Pulmonologist trained in Public Health, Health Services and Translational Research, with a proven commitment to research and scholarly productivity. This proposal includes an educational plan that will allow him to develop the gaps in his training: focused expertise in novel analysis of clinical trials and longitudinal studies, geriatric aspects of pulmonary disease, and outcome prediction. These experiences will provide him with the foundation to become an independent investigator in translational research with emphasis on the elderly and personalized medicine. Dr. Martinez will be supported by an outstanding multi-disciplinary mentorship team: Dr. Sandeep Vijan, a pioneer in developing models to inform the delivery of personalized medicine and Dr. MeiLan K. Han, a leader in COPD phenotyping. Drs. Neil B. Alexander, Caroline R. Richardson, and Christine T. Cigolle, leaders in geriatrics, and Dr. Fernando Martinez, a leader in developing treatments for COPD, will serve as advisors.