The all-oral BPaL(M) regimen, consisting of bedaquiline, pretomanid (Pa), linezolid, and moxifloxacin (moxifloxacin is stopped if fluoroquinolone resistance is detected), is becoming the preferred option for treating rifampin-resistant tuberculosis (TB) (
1,
2). Pa poses two challenges in this context. First, Bateson et al. (
3) described unprecedented differences in the intrinsic susceptibility of different
Mycobacterium tuberculosis complex (MTBC) lineages to Pa using the Mycobacteria Growth Indicator Tube (MGIT) system. Most notably, lineage 1 (L1), which accounts for 28% of TB cases globally, was found to be intrinsically less susceptible than the other major MTBC lineages [lineage 2 (L2), lineage 3 (L3), and lineage 4 (L4)], raising the question whether L1 responds equally well to BPaL(M) compared with L2–4 (
3).
Second, clinical strains with high Pa MICs due to mutations in known Pa resistance genes were identified without known nitroimidazole exposure, suggesting genetic drift or yet unknown selective pressures (
3–5). In the few settings with good surveillance or routine antimicrobial susceptibility (AST) results, these mutants are rare (
5,
6). However, because these mutants are known to be transmissible, it is plausible that some settings exist in which an intrinsically Pa-resistant cluster is frequent, underlining the need for routine AST (
4,
7). Yet, rapid AST directly from clinical samples is currently impossible as no commercial genotypic AST assay exists that interrogates
ddn (
Rv3547),
fbiA (
Rv3261),
fbiB (
Rv3262),
fbiC (
Rv1173),
fbiD (
Rv2983), and
fgd1 (
Rv0407), the six genes required for the activation of the pro-drug Pa [no resistance mutations have been described in
dprE2 (
Rv3791)], the target of Pa, to date) (
8–10). Although efforts are underway to address this diagnostic gap (e.g., Genoscreen is evaluating Deeplex Myc-TB XL, an updated version of its WHO-endorsed targeted next-generation sequencing assay), the interpretation of genotypic AST results will remain a persistent challenge as the aforementioned resistance genes are non-essential and, consequently, thousands of different loss-of-function mutations can theoretically confer resistance (
9,
11,
12).
The goal of this study was twofold. First, we used MGIT to refine the current understanding of the effect of the MTBC diversity on susceptibility, with a particular focus on the less frequent lineage 5 (L5), lineage 6 (L6), lineage 7 (L7), lineage 8 (L8), and lineage 9 (L9) that were not tested or were underrepresented in the literature (
3,
13,
14). Second, we used Middlebrook 7H11 (7H11) as an alternative medium to investigate whether the differences observed with MGIT were media specific.