Within patient microevolution of Mycobacterium tuberculosis correlates with heterogeneous responses to treatment.

It is now commonly accepted that the MTB population within individual tuberculosis (TB) patients can be more heterogeneous than was traditionally thought1,2,3. Heteroresistance, which refers to the coexistence of both drug-resistant and drug-sensitive strains or several drug-resistant strains with discrete drug resistance conferring mutations, has been described in clinical isolates of MTB previously3,4,5,6. In some cases, the genetic heterogeneity was due to mixed infections (polyclonal infections) by two distinct MTB strains with different drug susceptibilities that could be easily distinguished by standard fingerprinting1,2,7. In other cases, it was described as an intermediate or transitory state during the evolution of a single clonal strain to drug-resistance3,6,8. For instance, sequencing of serial isolates from a given patient revealed multiple resistance-conferring mutations that were transiently detected during development of drug resistance only one of which was ultimately fixed3,9,10. This phenomenon may have been the result of clonal interference because different drug-resistant mutations might result in different fitness costs3,9,10. An earlier study investigated different lesions obtained following lung surgery and found discrete pulmonary lesions carried strains with different drug-resistant mutations, which had evolved from a single parental strain11. This finding was further confirmed by analyzing the drug susceptibility profile of multiple cavity isolates from 5 patients that underwent pulmonary resection surgery12. In these situations, clonal interference does not apply because the strains reside in physically discrete lesions and genetic heteroresistance might therefore be a long-term phenomenon in vivo. In this study, taking advantage of the high resolution of deep whole genome sequencing together with molecular typing method, we tracked the population genetic dynamics of MTB population in serial sputum samples and characterized this phenomenon in more detail. Meanwhile, we provided the possible correlation between lesions’ disparate responses and heterogeneous bacterial population.