In 2016, tuberculosis (TB) caused 10.4 million new infections. Of these, 600,000 were drug resistant, yet the treatment success rate for drug resistant strains was only 54%, pointing to the potential for emergent epidemics of drug resistant TB. Previous studies have analyzed the effect of drug resistance on the number of new active infections, but not on the prevalence of the latent form of the disease. In this study, we examine how the long-term interaction of four levels of drug resistant Mycobacterium tuberculosis affects the prevalence of latent drug resistant TB in the population of a city. Specifically, we developed a deterministic SEIS (susceptible-latent-infected-susceptible) multi-strain model, to investigate the interaction of four different levels of drug resistance in the presence of treatment. We fit the model to data from the New York City TB epidemic of the late 1980s and early 1990s, and then ran simulations over a range of plausible parameter sets to assess the impact of these dynamics on the prevalence of latent TB in the population. We also considered the consequences of different treatment strategies (antibiotics versus a vaccine) on the outcomes.
Results/Conclusions
We estimated the total latently infected population at 27-32% as of 1949, when the first successful treatment of TB with streptomycin took place. Our simulations for the treatment era predict that approximately 100-150 years after the introduction of antibiotics, or 30-80 years from now, drug resistant latent TB strains will replace 95% of drug sensitive TB strains in the latently infected population, triggering an epidemic of drug resistant TB. Despite the fitness cost of drug resistance, in an environment where treatment with antibiotics is standard protocol, fitness of resistant strains appears to be relatively high, possibly enhanced by strain competition. Small treatment strategy decisions today could have a large impact on the future dynamics of these M. tuberculosis strains.