Strengthening evidence on optimal multidrug-resistant tuberculosis treatment regimens through improved epidemiologic methods

项目摘要/摘要 2017年，超过55万人感染了一种对利福平耐药的结核病(TB) 和异烟肼，这是标准一线结核病方案中最有效的两种药物。从历史上看，对病人的治疗 耐多药(MDR)结核病一直是长期、有毒和无效的：平均而言，只有55%的接受治疗的患者 在常规的18至24个月方案结束时治愈，其中包括每天注射治疗 大约八个月。2012年和2013年，监管部门有条件地批准了首例新结核病 50年来的药物，贝达奎兰(BDQ)和地拉曼(DLm)，为更有效和更低毒性提供了希望 耐多药结核病的治疗。2019年初，根据现有的最佳证据，世界卫生组织(世卫组织) 发布了耐多药结核病治疗的新指南。这些指南包括一个新的优先药物排名 耐多药结核病方案的组成，该方案将BDQ确立为18至24个月免费注射的支柱 方案，并核准了一种标准化的九个月方案，其中包含一种可注射的。 尽管传统的无注射和缩短的含有MDR-TB的注射方案代表 巨大的改进，这些新的指南推迟了对一些关键问题的指导 缺乏证据，使得许多国家结核病项目不确定如何实施耐多药结核病治疗。 例如，尽管建议指导整个疗程，但没有建议。 关于每种药物的最佳使用时间。此外，目前还没有关于这种疗法的安全性或有效性的数据。 由五种最优先的药物组成的养生法。缺乏高质量的耐多药-结核病队列数据也阻碍了 关于其他紧急问题的指导，包括(1)BDQ的普通非标签使用的安全性和有效性 和DLm，(例如，联合给药和延长药物使用时间)；以及(2)BDQ或DLm是否可以 被代入推荐的九个月疗程。在缺乏解决这些问题的临床试验数据的情况下 从观察性耐多药结核病队列数据中得出的问题、稳健、有效的因果推断对于 确保耐多药结核病的最佳治疗，并将死亡率、发病率和疾病传播降至最低 美国和世界各地的患者。 由我们小组实施的EndTB倡议提供了产生高质量证据的独特机会 向耐多药结核病的治疗和护理提供信息。2,600例耐多药肺结核患者的EndTB观察队列 17个国家的BDQ和/或DLM--这是此类数据集中规模最大、系统收集的前瞻性数据集。详细 纵向数据将允许进行强有力的分析，为世界卫生组织优先考虑的挥之不去的关键问题提供信息 组织。我们将利用这一独特的资源，在耐多药结核病高负担国家建立强有力的伙伴关系，并 在流行病学方法方面的高级专业知识，以确定最佳的耐多药结核病方案；告知最佳使用和 BDQ和DLm的不良事件监控，并说明稳健的纵向数据和 新的方法可以减少这些偏差。

PROJECT SUMMARY/ABSTRACT In 2017, more than 550,000 people became sick with a strain of tuberculosis (TB) that was resistant to rifampin and isoniazid, the two most potent drugs in the standard first-line TB regimen. Historically, treatment for patients with multidrug-resistant (MDR) TB has been long, toxic, and ineffective: on average, only 55% of treated patients were cured at the end of the conventional 18- to 24-month regimen, which included a daily injection for approximately eight months. In 2012 and 2013 regulatory authorities conditionally approved the first new TB drugs in fifty years, bedaquiline (BDQ) and delamanid (DLM), offering hope for more effective and less toxic MDR-TB treatment. In early 2019, based on the best available evidence, the World Health Organization (WHO) issued new guidelines for MDR-TB treatment. These guidelines included a new priority drug ranking for the composition of MDR-TB regimens, which established BDQ as an anchor of an injectable-free 18- to 24-month regimen and endorsed a standardized nine-month regimen containing an injectable. Although conventional injectable-free and shortened injectable-containing MDR-TB regimens represent enormous improvements, these new guidelines deferred guidance on a number of key questions for which evidence is lacking, leaving many national TB programs uncertain about how to implement MDR-TB treatment. For example, although recommendations guide the overall duration of treatment, there was no recommendation on the optimal duration of use of each drug. Moreover, there is no data on the safety or effectiveness of the regimen comprising the five top priority drugs. A lack of high quality MDR-TB cohort data also prevented guidance on other urgent questions, including (1) the safety and effectiveness of common off-label uses of BDQ and DLM, (e.g., co-administration and extended use of the drugs); and (2) whether BDQ or DLM could be substituted into the recommended nine-month regimen. In the absence of clinical trial data to address these questions, robust, valid causal inference from observational MDR-TB cohort data is of paramount importance to guaranteeing optimal treatment—and minimizing death, morbidity, and disease transmission—for MDR-TB patients throughout the United States and the world. The endTB initiative, implemented by our group, offers a unique opportunity to generate high-quality evidence to inform MDR-TB treatment and care. The endTB observational cohort of 2,600 MDR-TB patients treated with BDQ and/or DLM in 17 countries—the largest, systematically-collected prospective dataset of its kind. Detailed longitudinal data will permit robust analyses to inform lingering, critical questions, prioritized by the World Health Organization. We will exploit this unique resource, strong partnerships in high-MDR-TB-burden countries, and advanced expertise in epidemiologic methods to identify optimal MDR-TB regimens; inform optimal use of and adverse event monitoring for BDQ and DLM, and illustrate how combinations of robust, longitudinal data and novel methods can reduce these biases.