Browse the corpus

to the editor: Multidrug-resistant tuberculosis (MDR-TB) is a growing public health problem.1 Recently, the World Health Organization endorsed new regimens for the treatment of MDR-TB that rely on new and repurposed antibiotic agents (i.e., bedaquiline, pretomanid, and linezolid with or without moxifloxacin [BPaL/M]).2 Much of the global burden of MDR-TB is driven by patient-to-patient transmission of drug-resistant Mycobacterium tuberculosis,3 and since BPaL/M is used to treat MDR-TB, resistance to these drugs that is emerging in patients with MDR-TB could spread. On the other hand, experimental studies have shown that resistance to even a single drug can reduce the competitive (i.e., Darwinian) fitness of M. tuberculosis, as compared with drug-susceptible strains (in the absence of the drug),4 and because most M. tuberculosis strains that acquire resistance to BPaL/M are already resistant to many first- and second-line drugs, such strains might be less transmissible due to a cumulative reduction in competitive fitness. Yet, the contribution of patient-to-patient transmission to the growing burden of BPaL/M resistance has not been determined.

trains that acquire resistance to BPaL/M are already resistant to many first- and second-line drugs, such strains might be less transmissible due to a cumulative reduction in competitive fitness. Yet, the contribution of patient-to-patient transmission to the growing burden of BPaL/M resistance has not been determined. Here, we analyzed the genomes of a 13-year nationwide collection of 6926 clinical M. tuberculosis isolates from the country of Georgia, a known hotspot of MDR-TB.5 We identified 60 M. tuberculosis strains that were resistant to at least isoniazid and rifampicin (i.e., the two most important first-line drugs), along with fluoroquinolones and one of the other BPaL/M compounds; henceforth, we refer to these strains as “highly drug-resistant.” Drug resistance was inferred with the use of whole-genome sequencing data and not by phenotypic testing. With the exclusion of mixed infections, we found that 16 of 58 (28%) of these highly drug-resistant strains belonged to one of four genomic clusters. Within each of these clusters, all strains had an identical mutational profile, a finding that indicates patient-to-patient transmission (Fig. 1).

by phenotypic testing. With the exclusion of mixed infections, we found that 16 of 58 (28%) of these highly drug-resistant strains belonged to one of four genomic clusters. Within each of these clusters, all strains had an identical mutational profile, a finding that indicates patient-to-patient transmission (Fig. 1). To determine if transmission of highly drug-resistant strains also occurs in other clinical settings, we analyzed an additional 81,576 M. tuberculosis genomes from global sources. We identified an additional 454 highly drug-resistant strains in 26 countries. With the exclusion of mixed infections, we found that among 420 of these strains, 117 (28%) from 10 countries in the Americas, Europe, Africa, and Asia occurred in 1 of 41 genomic clusters. In addition, we identified 9 highly drug-resistant strains from 4 countries that carried resistance mutations to all BPaL/M compounds, thus potentially classifying them as “totally drug-resistant.” Further details are provided in the Supplementary Appendix, available with the full text of this letter at NEJM.org. Our results show that despite the recent introduction of BPaL/M, resistance to these new antituberculosis drugs has already developed in at least 27 countries across four continents. Furthermore, a quarter of these cases involved patient-to-patient transmission. Our findings call for improvements in diagnostic capacity, infection control, and surveillance, without which the longevity of BPaL/M is at risk.