Tradeoffs in Introduction Policies for the Anti-Tuberculosis Drug Bedaquiline: A Model-Based Analysis

BACKGROUND New drugs for the treatment of tuberculosis (TB) are becoming available for the first time in over 40 y. Optimal strategies for introducing these drugs have not yet been established. The objective of this study was to compare different strategies for introducing the new TB drug bedaquiline based on patients' resistance patterns. METHODS AND FINDINGS We created a Markov decision model to follow a hypothetical cohort of multidrug-resistant (MDR) TB patients under different bedaquiline use strategies. The explored strategies included making bedaquiline available to all patients with MDR TB, restricting bedaquiline usage to patients with MDR plus additional resistance and withholding bedaquiline introduction completely. We compared these strategies according to life expectancy, risks of acquired resistance, and the expected number and health outcomes of secondary cases. For our simulated cohort, the mean (2.5th, 97.5th percentile) life expectancy from time of initiation of MDR TB treatment at age 30 was 36.0 y (33.5, 38.7) assuming all patients with MDR TB received bedaquiline, 35.1 y (34.4, 35.8) assuming patients with pre-extensively drug-resistant (PreXDR) and extensively drug-resistant (XDR) TB received bedaquiline, and 34.9 y (34.6, 35.2) assuming only patients with XDR TB received bedaquiline. Although providing bedaquiline to all MDR patients resulted in the highest life expectancy for our initial cohort averaged across all parameter sets, for parameter sets in which bedaquiline conferred high risks of added mortality and only small reductions in median time to culture conversion, the optimal strategy would be to withhold use even from patients with the most extensive resistance. Across all parameter sets, the most liberal bedaquiline use strategies consistently increased the risk of bedaquiline resistance but decreased the risk of resistance to other MDR drugs. In almost all cases, more liberal bedaquiline use strategies reduced the expected number of secondary cases and resulting life years lost. The generalizability of our results is limited by the lack of available data about drug effects among individuals with HIV co-infection, drug interactions, and other sources of heterogeneity, as well as changing recommendations for MDR TB treatment. CONCLUSIONS If mortality benefits can be empirically verified, our results provide support for expanding bedaquiline access to all patients with MDR TB. Such expansion could improve patients' health, protect background MDR TB drugs, and decrease transmission, but would likely result in greater resistance to bedaquiline.