Mathematical Modeling in Clinical Trials of Tuberculosis Therapeutics

描述（由申请人提供）：控制结核病伙伴关系已优先发展较短的治疗方案，作为其全球消除结核病计划的一部分。为了实现这一目标，许多实体已经进行了II期临床试验，使用2个月的培养转化作为替代终点，以测试新药方案的灭菌能力。然而，所有拟议的方案的顺序测试是耗时和成本高昂的，因此需要更有效的策略来选择方案进行测试。此外，如果没有对新疗法的成本进行事先评估，就有可能开发出一种在经济上不可行的结核病新疗法。数学建模可以通过综合先前人类和小鼠研究的数据来提供这个问题的解决方案，以生成可用于临床试验设计的新方案的复发率估计值。首先，将开发TB治疗和复发的通用马尔可夫模型，其中TB患者将被分配到几个治疗组中的一个，然后通过在1、2、3或4个月时对培养阳性进行中期评估，每月进行一次。然后假设他们完成了巩固治疗期，之后将对他们进行为期两年的复发随访。将在每个阶段考虑毒性和耐受性。我们将对所有相关参数（特别是培养转化率和复发率）使用适当的分布，并使用概率敏感性分析来产生复发率的点估计值（95%置信区间），可用于确定哪些方案值得进一步测试。然后，该模型将用于实现3个具体目标：目标1将使用该模型检查在标准和缩短治疗持续时间的情况下，在TB治疗的前两个月内用替诺氟沙星替代异烟肼对TB复发率的潜在影响。目标2将确定高剂量利福喷丁（在结核病治疗的前2个月内给药）对复发率的影响和缩短结核病治疗持续时间的可能性。目标3将是模拟利奈唑胺对耐多药结核病治疗的潜在作用，特别注意增加有效性和毒性之间的权衡。将计算每种模式的成本和成本效益。

DESCRIPTION (provided by applicant): The Stop TB Partnership has prioritized the development of shorter regimens as part of their global plan to eliminate TB. In pursuit of that goal, many entities have undertaken Phase II clinical trials using 2-month culture conversion as a surrogate endpoint to test the sterilizing ability of new drug regimens. However, the sequential testing of all proposed regimens is time-consuming and cost-prohibitive, so more efficient strategies for selecting regimens for testing are required. Furthermore, without prior evaluation of the costs of new therapies, there is a risk of developing a new treatment for TB that is not economically viable. Mathematical modeling can provide a solution to this problem by synthesizing data from prior human and mouse studies to generate estimates of relapse rates for new regimens that could be used in the design of clinical trials. First, a generic Markov model of TB treatment and relapse will be developed in which individuals with TB will be assigned to one of several treatment arms, then followed in monthly cycles through an interim evaluation of culture positivity at 1, 2, 3, or 4 months. They will then be assumed to complete a consolidation phase of therapy, after which they will be followed for two years for relapse. Toxicity and tolerability will be considered at each stage. We will use appropriate distributions for all relevant parameters, particularly culture conversion and relapse rates, and use probabilistic sensitivity analysis to produce point estimates of relapse rates with 95% confidence intervals that can be used to determine which regimens are worthy of further testing. This model will then be used to accomplish 3 specific aims: Aim 1 will be to use the model to examine the potential impact of substituting moxifloxacin for isoniazid during the first two months of TB treatment on TB relapse rates, given standard and abbreviated treatment durations. Aim 2 will be to determine the impact of high-dose rifapentine (administered during the first 2 months of TB treatment) on relapse rates and potential for reduction of the duration of TB treatment. Aim 3 will be to model the potential effect of linezolid on the treatment of MDR-TB, paying particular attention to the tradeoff between increased effectiveness and toxicity. Costs and cost-effectiveness will be calculated in each model.