CC and MIC DST

CC DST was carried out on 208 MTBC strains. A total of 130 (130/208; 62.5%) strains were resistant to RIF/INH, as per MGIT and proportion method (Additional file: Table S2), thus meeting the MDR definition. Furthermore, 185 strains had their MICs determined using MYCOTBI plates and were stratified by CC DST results (Fig. 1). The mode MICs were > 16 µg/mL (≤ 0.12–>16 µg/mL) and ≤ 0.12 µg/mL (≤ 0.12–0.5 µg/mL) for RIF-resistant and RIF-susceptible strains, respectively. As for INH, the MICs for INH-resistant strains varied widely (mode MIC of 2 µg/mL; ≤0.03–>4 µg/mL), whereas for INH-susceptible strains the MICs were within a narrow range of 0.03–0.06 µg/mL (mode MIC of ≤ 0.03 µg/mL).

Fig. 1

Distribution of RIF (A)/INH (B) MICs stratified by CC DST results. RIF rifampicin; INH isoniazid; MIC minimum inhibitory concentration; DST drug susceptibility testing; CC critical concentration

Molecular DSTLPA

LPA identified 125 strains as MDR (Additional file 1: Table S2). Resistance to RIF was detected in 128 (128/208; 61.5%) strains. The most prevalent mutation pattern was S531L (MUT3) found in 108 (108/208; 51.9%) strains, coupled with either ΔWT8 (S531L; 107/208; 51.4%) or with a WT8 band present (S531L; 1/208; 0.5%).

Resistance to INH was detected in 125 (125/208; 60.1%) strains (Additional file 1: Table S2). All but two were found to carry mutations in the codon 315 of the katG gene. Of these, the majority (122/208; 58.7%) had S315T1 (MUT1) with a WT band present (4/208; 1.9%) or absent (118/208; 56.7%), followed by one (1/208; 0.5%) strain with both WT and MUT bands absent.

Mutations in the inhA gene (19/208; 9.1%) were rare; 17 (17/208; 8.2%) strains harboured single C-15T polymorphism, one (1/208; 0.5%) strain had T-8 C polymorphism, and one (1/208; 0.5%) missed both WT1 and MUT bands in -15 region of the inhA promoter.

Mutation profiling by WGS

A total of 124 (124/208; 59.6%) strains were identified as MDR using WGS (Additional file 1: Table S2). Overall, 125 (125/208; 60.1%) strains were found RIF-resistant, while 124 (124/208; 59.6%) were INH-resistant.

The most prevalent RIF resistance conferring mutation was rpoB S450L accounting for a total of 105 (105/208; 50.9%) strains, while the remaining mutations were as folows: rpoB H445D (6/208; 2.9%), rpoB D435V (5/208; 2.4%), rpoB H445Y (3/208; 1.4%), rpoB S450W (3/208; 1.4%), rpoB H445L (1/208; 0.5%), rpoB H445C (1/208; 0.5%), rpoB H445N (1/208; 0.5%) (Additional file 1: Table S2).

The majority (124/208; 59.6%) of INH-resistant variants harboured single katG S315T (106/208; 51.0%) mutation. The remaining strains appeared to be double mutants containing katG S315T and either inhA C-15T (15/208; 7.2%) or inhA T-8 A (1/208; 0.5%) substitution. Single inhA C-15T substitution occurred in two (2/208; 1%) strains only (Additional file 1: Table S2).

Comparison of methodsCC DST vs. MIC DST

Overall, the agreement between CC and MIC DST was equally high for RIF (97.8%) and INH (97.3%) (Table 1). Only 4 (4/185; 2.2%) strains gave discordant results for RIF exhibiting low MICs for strains DLT80 (≤ 0.12 µg/mL), KR-PLM-15 (0.5 µg/mL), DLT48, and DLT76 (0.25 µg/mL). Noteworthy, the latter two strains (DLT76 and DLT48) required and additional week of incubation to reach higher MICs of > 16 µg/mL and 1 µg/mL, respectively. A total of 5 (5/185; 2.7%) strains demonstrated discordant results for INH. Two of these strains (DLT33, DLT107) exhibited MICs of 0.12 µg/mL, while another two (KR-PLM-15, DLT49) had MICs of 0.06 µg/mL. One strain (DLT80) had an MIC of ≤ 0.03 µg/mL.

Table 1 Comparison of MIC DST and CC DST results for RIF/INHSequence analysis vs. software platforms

The results of drug resistance profiling performed with WGS and two software platforms (TB Profiler and Mykrobe) showed a concordance of 100% for RIF and 99% for INH (Table 2). The latter result was due to two strains that were identified as INH-susceptible upon WGS, yet resistant by either TB Profiler and Mykrobe (G48A mutation in the promoter of the ahpC gene; 1 strain) or Mykrobe (L302L mutation in the fabG1 gene; 1 strain).

Table 2 Performance characteristics of different methods and tools for the prediction of resistance to RIF/INHPerformance of molecular methods

The overall performance of LPA was satisfactory with a sensitivity of 98.5% and 96.1%, and a specificity of 100% for detecting resistance to RIF and INH, respectively (Table 2). Nevertheless, LPA failed to detect resistance in 5 (5/130; 3.8%) phenotypically INH-resistant and two (2/130; 1.5%) RIF-resistant strains.

Compared to LPA, WGS demonstrated slightly lower sensitivity (96.2%) for RIF and (95.4%) for INH, while the specificity for both drugs was exactly the same (100%). Interestingly, rpoB S450L (strains DLT80, DLT48, DLT76), rpoB D435V (strain D-PL-22), and katG S315T (strain DLT80) mutations were not detected by WGS, though they were determined by LPA. The concordance between LPA and WGS was 97.6% and 99.5% for RIF and INH, respectively (Table 2).

Phenotypic vs. molecular methods

A total of 11 (11/208; 5.3%) MDR strains (determined by CC DST) had discordant results with either WGS, LPA or MIC DST. The discordance concerned resistance to RIF in two (2/208; 1%) strains, INH – in 5 (5/208; 2.4%) strains, and to both drugs – in four (4/208; 1.9%) strains (Fig. 2).

Fig. 2

Phenotypic and genotypic drug susceptibility profiles of RIF/INH in 11 strains with discordant results. RIF rifampicin; INH isoniazid; MGIT Bactec MGIT 960 system; LJ Löwenstein-Jensen; MIC minimum inhibitory concentration; MYCOTBI 96-well MYCOTBI® plates for MIC testing (ThermoFisher Scientific, USA); The MICs taken after 14 days were used for the Fig. 3; LPA line-probe assay; WT wild type; Δ – absence of hybridization signal with wild-type probes; WGS whole-genome sequencing; § – mutations of uncertain significance; * – strains became RIF-resistant after additional week of incubation on day 21

The LPA failed to detect RIF-resistance in strain (DLT102) carrying rpoB S450W. However, this was not observed for the other two rpoB S450W mutants (DLT41, DLT89). In case of DLT41 and DLT89 strains, mutation rpoB S450W resulted in absence of both rpoB WT8 and MUT bands which indicates inferred resistance. However, the WT8 band was present in strain DLT102 which let it to be considered as RIF wild type.

As for the strain KR-PLM-15, a MIC testing revealed possible flaws with the CC DST, as the MICs for RIF and INH only increased slightly reaching 0.5 µg/mL and 0.06 µg/mL. Moreover, no resistance-conferring mutations were detected by LPA and WGS.

One katG S315T mutant (DLT49) increased MIC only to 0.06 µg/mL, possibly due to technical errors, as CC DST results were congruent with LPA/WGS. No INH-resistance associated mutations were detected by LPA/WGS for strains KR-PLM-8 and KR-PLM-20. For strain KR-PLM-8, both CC and MIC DST results were congruent. A similar comparison for strain KR-PLM-20 was not possible due to contamination with non-tuberculous mycobacteria.

Strains for which resistance was detected by LPA but not by WGS (DLT80; DLT48) had MICs below the CLSI breakpoints for RIF/INH (Fig. 2).

MIC DST vs. WGS

Among RIF-resistant strains, all but two (DLT33–8.0 µg/mL, DLT60–16.0 µg/mL) with rpoB S450L mutation (97/185; 52.4%) had their MICs > 16 µg/mL, as well as strains harbouring the rpoB H445D (6/185; 3.2%), H445Y (3/185; 1.6%), and S450W (3/185; 1.6%) mutations (Fig. 3). Five (5/208; 2.4%) strains with rpoB D435V mutation had their MICs spanned over a wide range of values (i.e. from 4.0 µg/mL to > 16 µg/mL). So far, seven mutations conferring so-called borderline RIF-resistance, expected to exhibit lower MIC values, have been identified [11]. Two such mutations (rpoB H445L and H445N) were detected and resulted in noticeably high MICs of 4.0 µg/mL and > 16 µg/mL, respectively. Single rpoB H445C mutant (1/185; 0.5%) had MIC of 8.0 µg/mL. Single genotypically susceptible strain (1/185; 0.5%) demonstrated MIC of 16 µg/mL. The MICs of the remainder of the genotypically RIF-susceptible (65/185; 35.1%) strains fell within a range of ≤ 0.12–0.5 µg/mL.

Fig. 3

Distribution of RIF (A)/INH (B) MICs stratified by the presence of resistance-conferring mutations (WGS). RIF rifampicin; INH isoniazid; MIC minimum inhibitory concentration; DST drug susceptibility testing; CC critical concentration; WGS whole-genome sequencing

The MIC distribution among INH-resistant strains harbouring katG S315T mutation (103/185; 55.7%) ranged from 0.06 to > 4 µg/mL, with a mode MIC of 2 µg/mL. The mode MIC was two-fold higher (> 4 µg/mL) among double mutants harbouring katG S315T and either inhA C-15T (14/185; 7.6%) or inhA T-8 A (1/185; 0.5%) mutation compared to single katG mutants. Single (1/185; 0.5%) inhA C-15T promoter mutant had its MIC of 1 µg/mL. The MIC of genotypically INH-susceptible strains ranged from ≤ 0.03 to 0.6 µg/mL (63/185; 34.0%) with three strains showing unexpectedely high MICs of 1 µg/mL (1/185; 0.5%) and > 4 µg/mL (2/185; 1.1%).