The epidemiological fitness cost of drug resistance in Mycobacterium tuberculosis

Abstract

The emergence of antibiotic resistance in Mycobacterium tuberculosis has raised the concern that pathogen strains that are virtually untreatable may become widespread. The acquisition of resistance to antibiotics results in a longer duration of infection in a host, but this resistance may come at a cost through a decreased transmission rate. This raises the question of whether the overall fitness of drug-resistant strains is higher than that of sensitive strains--essential information for predicting the spread of the disease. Here, we directly estimate the transmission cost of drug resistance, the rate at which resistance evolves, and the relative fitness of resistant strains. These estimates are made by using explicit models of the transmission and evolution of sensitive and resistant strains of M. tuberculosis, using approximate Bayesian computation, and molecular epidemiology data from Cuba, Estonia, and Venezuela. We find that the transmission cost of drug resistance relative to sensitivity can be as low as 10%, that resistance evolves at rates of approximately 0.0025-0.02 per case per year, and that the overall fitness of resistant strains is comparable with that of sensitive strains. Furthermore, the contribution of transmission to the spread of drug resistance is very high compared with acquired resistance due to treatment failure (up to 99%). Estimating such parameters directly from in vivo data will be critical to understanding and responding to antibiotic resistance. For instance, projections using our estimates suggest that the prevalence of tuberculosis may decline with successful treatment, but the proportion of cases associated with resistance is likely to increase.

Fig. 1.

The epidemiological model. As described in Materials and Methods, the number of individuals in each subpopulation of the model changes according to a stochastic linear birth–death process. The parameters are defined as follows: α is the transmission rate per individual per unit time, c is the transmission cost due to resistance, ρ is the rate of evolution of resistance per individual per unit time, τ is the rate of detection (and treatment commencement) per individual per unit time, and ε_S and ε_R are the rates of cure due to treatment for sensitive and resistant strains, respectively. The mutation rate of the marker μ and the rate of death or natural recovery δ are not shown in this figure; these occur in every state. Mutation of the marker does not affect the phenotype but gives rise to a new genotype.

Fig. 2.

Marginal posterior distributions. (A) The transmission cost c. (B) The rate of evolution of resistance ρ. (C) The relative fitness Φ_RS of resistant strains compared with sensitive strains. The isolates from Venezuela were genotyped by using spoligotyping (spol.); isolates from Estonia were typed by using IS6110; isolates from Cuba were typed by using both spoligotyping and IS6110-typing. (D) The proportion of resistant cases in the population that arose through treatment failure leading to the evolution of resistance from sensitivity.

Fig. 3.

Projection of cases resistant to drugs by simulating the model forwards by using the parameter estimates from the Estonia data. (A and C) Moderate cure rates of ε_S = 0.52 and ε_R = 0.202 (50% treatment success). (B and D) Optimistic cure rates of ε_S = 2.95 and ε_R = 0.47 (85% treatment success). A and B show the posterior median total number of cases of tuberculosis over time (black lines) and the median number of cases with drug-resistant M. tuberculosis (red lines), starting with the estimated posterior distribution of parameters. C and D show the proportion of cases that are drug resistant (black lines) and the cumulative proportion of simulations that undergo extinction (red lines). In both scenarios, the case detection and treatment rate is τ = 1.21, which corresponds to the WHO target of 70% of cases detected and treated. Solid lines represent median values, and error bars indicate minimum 95% credibility intervals at 10-y intervals.