Over the recruitment period, 342 patients (194 with RR and 148 with RS M. tuberculosis) participated in the study; 33 case-patients were excluded because of laboratory contamination or failed sputum culture and DNA extraction. Of the final 309 GeneXpert-positive included participants, 200 (65%) were male (Table 1), 118 were RS and 191 were RR, and all had phenotypic DST successfully completed. RR was strongly associated with a history of TB treatment (p<0.0001) (Table 1).

Figure 1

Figure 1. Genomic profiling of Mycobacterium tuberculosis strains, Myanmar, comparing discriminatory power offered by GeneXpert (Cepheid, https://www.cepheid.com) and additional genotypic testing, such as line-probe assay or whole-genome sequencing....

We compared the results of the GeneXpert, phenotypic DST, and genomic analyses to further evaluate drug resistance (Figure 1). Of 118 cases diagnosed as RS by using GeneXpert, 16 (14%) were identified as isoniazid resistant on the basis of genomic analyses (Figure 1); resistance was conferred either by a mutation in the katG gene (S315T; 12 [75%] of 16) or in the promoter region of the inhA gene (c-15t; 4 [25%] of 16) (Table 2; Figure 1). All 16 cases were phenotypically confirmed as isoniazid resistant (Table 2).

Figure 2

Figure 2. Prevalence of resistance-conferring mutations in genes of phenotypically resistant isolates of Mycobacterium tuberculosis strains, Myanmar. A) katG and inhA; B) embB; C) ...

All 191 RR isolates identified by GeneXpert were phenotypically resistant; the S450L mutation in the rpoB gene was the dominant mutation (137 [72%] of 191 (Table 2; Figure 2). WGS further identified that 10 (5%) were only rifampin resistant (pre-MDR), 144 (75%) were MDR, 31 (16%) were pre-XDR, and 6 (3%) were XDR; results were confirmed by phenotypic DST (Figure 1). All pre-XDR isolates harbored mutations in the gyrA gene, and D94G was most prevalent (12 [39%] of 31), followed by A90V (8 [26%] of 31) (Table 2; Figure 2). Resistance to aminoglycoside injectable drugs (XDR) was predominantly associated with rrs A1401G (4 [6 [67%] of 6) and G1484T (2 [33%] of 6) mutations. Mutations in the embB gene (M306V; 42 [53%] of 80), the M306I mutant (34 [43%] of 80), and mutations in the rpsl gene (K43R; 126 [90%] of 140) were present in all ethambutol-resistant and streptomycin-resistant isolates, resulting in a sensitivity of 70.2% and a specificity of 79.5% for ethambutol and a sensitivity of 84.3% and a specificity of 55.9% for streptomycin (Table 2; Figure 2). Known mutations conferring resistance to new and repurposed drugs, such as bedaquiline, delamanid, pretomanid, linezolid and clofizamine, were not identified by WGS in the 207 drug-resistant isolates (26 pre-MDR, 144 MDR, 31 pre-XDR, 6 XDR).

Figure 3

Figure 3. Maximum-likelihood tree based on whole-genome analysis of 309 Mycobacterium tuberculosisstrains from Myanmar. Lineages and drug resistance status of isolates are shown. MDR indicates multidrug-resistant to 2 first-line drugs...

Figure 4

Figure 4. Maximum-likelihood tree of Mycobacterium tuberculosisstrains, Myanmar, within 9 clusters and their drug resistance profiles. Dotted lines indicate boundaries of individual clusters. An outgroup (#108) differs by >100 single-nucleotide...

Using specific SNP barcodes, we classified the M. tuberculosis isolates as either lineage 1 (73, 24%), lineage 2 (201, 65%), lineage 3 (16, 5%) or lineage 4 (19, 6%) (Figure 3; Appendix Table 1). Most isolates were identified as belonging to sublineage 2.2.1, Beijing strain (Appendix Table 1). Isolates linked to TB lineage 2 were more commonly drug resistant than those belonging to other lineages (175 [85%] of 207 vs. 32 [15%] of 207; p<0.0001). In contrast, the other lineages were more commonly associated with drug susceptibility (76 [75%] of 102 vs. 26 [25%] of 102; p<0.001) (Table 1; Appendix Table 2). Drug-resistant isolates were also more commonly found in the East and West districts of Yangon (124 [60%] of 207; p<0.0001) (Table 1; Appendix Table 2) and to be associated with patients who had previously received treatment (112 [54%] of 207; p<0.0001) (Appendix Table 2). A total of 181 [87%] of 207 isolates were genomically unlinked on the basis of a standard pairwise distance threshold. The remaining 26 (13%) of 207 drug-resistant isolates formed 9 potential transmission chains (Figure 4).

Cases within most of these groups were located within the same districts (Figure 4), and each group contained a combination of new and previously treated TB patients. In 6 groups, all isolates had the same resistance profile; the remaining 3 (i.e., groups 5, 6, and 8) groups, had different resistance profiles. In group 6, an XDR isolate appears to have developed from an isoniazid-resistant (pre-MDR) isolate (Figure 4).

This WGS study from Myanmar provides new insights into the landscape of drug-resistant TB in the country’s largest city. A large proportion of isolates with high-level drug resistance, including pre-XDR and XDR, were identified. However, there was no resistance to new and repurposed drugs, such as bedaquiline, pretomanid, delamanid, and linezolid. Most drug-resistant cases were associated with previous treatment, and few were clearly associated with community transmission. These findings suggest an additional diagnostic tool, such as the Xpert MTB/XDR cartridge or line-probe assay (LPA), in addition to Xpert MTB/RIF, and new oral regimens, including bedaquiline and pretomanind, are needed for effective surveillance and treatment/management of MDR TB in Mynamar. Further studies are also required to investigate apparent cases of independent emergence and community transmission of MDR TB in Yangon.

Consistent with previous reports on lineage 2 from neighboring countries, this study identified a strong association between lineage 2 M. tuberculosis and drug resistance (14–17). There was strong agreement between WGS (presence of resistance-conferring mutations) and phenotypic DST to isoniazid and rifampin in this study (18). These findings indicate quality assurance in the TB laboratory diagnostic service provided by the Myanmar National Tuberculosis Reference Laboratory.

The Xpert MTB/RIF assay has been effective in the simultaneous detection of TB and resistance to rifampin. Because it can provide a diagnosis for a patient within 2 hours, GeneXpert is critical in TB control in high-burden settings. One of the limitations of current cartridges for Xpert is that resistance to isoniazid is assumed when rifampin resistance is detected. This approach captures a large portion of drug-resistant TB cases during diagnosis. However, for a few case-patients, which includes patients who have isoniazid resistance without concurrent rifampin resistance (14% in this study and 9.4% in the recently reported multicountry study [19]), treatment with a first-line regimen can contribute to the emergence of further drug resistance.

We previously reported that a patient with undiagnosed isoniazid resistance without concurrent rifampin resistance received a first-line treatment regimen that resulted in development of MDR TB (20). This finding highlights the limitations and real-world consequences of basing treatment decisions solely on results of the GeneXpert MTB/RIF system in a high-burden setting, where hundreds of cases are reported daily. This limitation is a serious impediment to controlling the spread of more extensive drug resistance (21). For example, although Xpert MTB/RIF can correctly diagnose RR MDR cases, it cannot detect pre-XDR and XDR cases. As identified in this study, 20% of rifampin-resistant cases identified by GeneXpert were pre-XDR (17%) and XDR (3%) cases, suggesting that ≈1 of 5 patients received limited treatment on the basis of treatment guidelines at the time of the study.

Most case-patients (including pre-XDR and XDR patients) in this study had drug-resistant isolates that were not closely related (genomically unlinked), which is suggestive of independent emergence of drug resistance because of limited diagnosis or treatment, as well as patient noncompliance. This finding is in contrast to previous studies from other high burden settings, such as China and South Africa, which showed a high proportion of drug-resistant cases that were genomically linked, suggesting community transmission (15,21–23).

Although it is possible that we simply did not have a high enough sampling fraction of all drug-resistant cases in the population under study, a high number of unclustered drug-resistant cases could be caused by differences in population density; the North, East, and West sections of Yangoon are in an urban industrial setting. These districts have a considerable factory-based workforce and thus draw in highly mobile migrant populations (internal migration), including members from neighboring states and regions, for employment (24). This finding enables a continuous flow of persons from outside Yangon, which could be independently introducing infections into the region. In addition, their status as migrants means they might have limited access to healthcare services, which is a barrier to rapid diagnosis and appropriate treatment for TB, underscoring the effect of migration on the TB burden in cities in Myanmar, particularly Yangon (20,25).

In addition to internal migration, cross-border migration has occurred in recent years, such as ≈6 million persons from fellow Greater Mekong Subregion (GMS) countries Cambodia, Laos, Thailand, and Vietnam. Therefore, the Myanmar NTP is collaborating with nongovernmental organizations and NTPs from other GMS countries to reduce the TB burden among Myanmar migrants. Further WGS studies outside Yangon and along these GMS borders are required to provide an insight into the transmission patterns of MDR TB in migrants. Coupling this collaboration with TB-related health education and increase access to care could ultimately reduce the TB burden among migrants.

Further studies are also required to clarify the limitations and roles of both public and private healthcare providers in current treatment pathways for TB in Yangon, which might be contributing to the high rates of MDR TB. In our study, WGS showed a chain of infection, leading to the progression of pre-MDR cases toward XDR and subsequent transmission events, highlighting the need for effective diagnosis. This finding has implications for public health policies and also shows the need for local data to drive effective intervention.

Our study has major implications for clinical practice in Myanmar. First, effective treatment for MDR TB cases requires identification of the high proportion of pre-XDR and XDR TB, which cannot be achieved by current Xpert MTB/RIF testing (27,28). The drug resistance‒conferring mutations reported in this study can be detected by first-line and second-line LPA, such as GenoType MTBDRplus and MTBDRsl (Hain Lifescience GmbH, https://www.hain-lifescience.de), or the recently launched Xpert MTB/XDR (28,29). These platforms can provide clinicians with an expanded drug-susceptibility report without the need for culturing and WGS. Recently, the Myanmar National Tuberculosis Programme diagnostic algorithm has been updated to extend first-line LPA for patients with a history of previous treatment. Second, several new or repurposed drugs (i.e., bedaquiline, delamanid linezolid, and pretomanid) are drugs already available in Myanmar. The apparent absence of preexisting mutations that confer resistance to these drugs justifies their introduction into treatment regimens for drug-resistant TB in Myanmar, as per WHO recommendations (30–32).

Our study has limitations that could lead to overestimation and underestimation of the true magnitude of the MDR epidemic and might not reflect the national situation. First, a large cohort of MDR TB, pre-XDR TB, and XDR TB cases was identified. The study region in Yangon is known to be a high-burden setting compared with other regions of Myanmar, accounting for ≈50% of all national cases (26). The 3 diagnostic centers in this study are also the major drug-resistant TB treatment centers in Yangon. Therefore, it is likely that the landscape of infections is not representative of all Myanmar. Another limitation is that the study timeframe and size make it unlikely that we captured a full spectrum of MDR TB strains in the population and, as noted earlier, we might have missed identification of some transmission links, thus overestimating the proportion of resistant isolates that are independent.

Our study has major implications for clinical practice in Myanmar. First, effective treatment for MDR TB cases requires identification of the high proportion of pre-XDR and XDR TB, which cannot be achieved by current Xpert MTB/RIF testing (27,28). The drug resistance‒conferring mutations reported in this study can be detected by first-line and second-line LPA, such as GenoType MTBDRplus and MTBDRsl (Hain Lifescience GmbH, https://www.hain-lifescience.de), or the recently launched Xpert MTB/XDR (28,29). These platforms can provide clinicians with an expanded drug-susceptibility report without the need for culturing and WGS. Recently, the Myanmar National Tuberculosis Programme diagnostic algorithm has been updated to extend first-line LPA for patients with a history of previous treatment. Second, several new or repurposed drugs (i.e., bedaquiline, delamanid linezolid, and pretomanid) are drugs already available in Myanmar. The apparent absence of preexisting mutations that confer resistance to these drugs justifies their introduction into treatment regimens for drug-resistant TB in Myanmar, as per WHO recommendations (30–32).

Our study is useful for public health officials in designing interventions for an evidence-based approach for early detection of cases (active case finding) with optimized diagnosis and treatment. Introducing additional diagnostic methods, such as routine LPA or Xpert MTB/XDR in tandem with Xpert MTB/RIF, and treatment regimens with new oral drugs would further assist in controlling and containing MDR TB in Myanmar. In addition, this study underscores the need for local data, rather than being based on general information from similar studies that have different healthcare delivery systems to drive public health policies for effective intervention.