Increasing prevalence of resistance to second-line drugs among multidrug-resistant Mycobacterium tuberculosis isolates in Kuwait

Molecular methods detect genetic mutations associated with drug resistance. This study detected resistance-conferring mutations in gyrA/gyrB for fluoroquinolones and rrs/eis genes for second-line injectable drugs (SLIDs) among multidrug-resistant Mycobacterium tuberculosis (MDR-TB) isolates in Kuwait. Fifty pansusceptible M. tuberculosis and 102 MDR-TB strains were tested. Phenotypic susceptibility testing was performed by MGIT 960 system using SIRE drug kit. GenoType MTBDRsl version 1 (gMTBDRslv1) and GenoType MTBDRsl version 2 (gMTBDRslv2) tests were used for mutation detection. Results were validated by PCR-sequencing of respective genes. Fingerprinting was performed by spoligotyping. No mutations were detected in pansusceptible isolates. gMTBDRslv1 detected gyrA mutations in 12 and rrs mutations in 8 MDR-TB isolates. gMTBDRsl2 additionally detected gyrB mutations in 2 and eis mutation in 1 isolate. Mutations in both gyrA/gyrB and rrs/eis were not detected. gMTBDRslv1 also detected ethambutol resistance-conferring embB mutations in 59 isolates. Although XDR-TB was not detected, frequency of resistance-conferring mutations for fluoroquinolones or SLIDs was significantly higher among isolates collected during 2013–2019 versus 2006–2012. Application of both tests is warranted for proper management of MDR-TB patients in Kuwait as gMTBDRslv2 detected resistance to fluoroquinolones and/or SLIDs in 3 additional isolates while gMTBDRslv1 additionally detected resistance to ethambutol in 58% of MDR-TB isolates.

The gMTBDRslv2 assay also detected N538D mutation in 1 isolate and a non-specific mutation indicated by lack of hybridization with wild-type (WT) gyrB probe in another isolate. PCR-sequencing of gyrB identified T539N mutation in the latter isolate. The remaining 100 isolates contained wild-type gyrB. Furthermore, 1 isolate contained an eis mutation detected only by lack of hybridization with WT2 eis probe. PCR-sequencing of eis identified G-10A mutation in this isolate. The remaining 101 isolates contained wild-type eis. Both eis and gyrB mutations were detected in isolates resistant to all 4 SIRE drugs ( Table 2). The prevalence of mutations conferring resistance to FQs (gyrA/gyrB) or SLIDs (rrs/eis) was significantly higher in MDR-TB strains resistant to 3 drugs versus 2 drugs (12 of 36 versus 2 of 32, P = 0.007) as well as for all 4 SIRE drugs versus 2 drugs (9 of 34 versus 2 of 32, P = 0.045). Interestingly, the prevalence of mutations conferring resistance to FQs (gyrA/gyrB) or SLIDs (rrs/ eis) was also higher in MDR-TB strains with an rpoB S450L mutation versus isolates with other rpoB mutations (19 of 73 versus 4 of 29, P = 0.182), however, the difference was not statistically significant.
The distribution of mutations in gyrA/gyrB or rrs/eis genes conferring resistance to FQs and SLIDs, respectively, among 23 MDR-TB strains collected over a 14-year period (2006 to 2019) together with mutations in rpoB, inhA and embB genes and fingerprinting data by spoligotyping are presented in Table 3. No isolate contained a mutation in both gyrA/gyrB + rrs/eis genes, i.e., XDR-TB was not detected in Kuwait ( Table 3). The spoligotyping data showed that the majority (13 of 21, 62%) of MDR-TB strains with a mutation in gyrA/gyrB or rrs/eis genes Table 2. Phenotypic resistance by MGIT 960 system to SIRE drugs and genotypic screening of mutations in rpoB, gyrA, gyrB, rrs and eis among 152 clinical M. tuberculosis isolates. INH isoniazid, RIF rifampin, EMB ethambutol, STR streptomycin. a Other mutations included H445Y/D/R, n = 12; D435V, n = 4; Q432E/K/L/P, n = 5; S450W, n = 3; V170F, n = 2; M434I + D435Y, n = 2 and D435G + H445Q, n = 1. b Two isolates with rpoB mutation S450W. c One isolate with rpoB mutations D435G + H445Q. d One isolate with rpoB mutation H445R. www.nature.com/scientificreports/ belonged to Beijing genotype. All isolates with identical gyrA mutation were genotypically distinct strains based on other mutation patterns and/or by spoligotyping (Table 3). However, 6 of 8 isolates with A1401G mutation in rrs were genetically identical. The year-wise data on the prevalence of gyrA/gyrB mutations conferring resistance to FQs, rrs/eis mutation conferring resistance to SLIDs and embB codon 306 mutation conferring resistance to EMB in MDR-TB strains during the 14-year study period are shown in Table 4. When the total 14-year period of the study was split into two 7-year periods, the prevalence of mutations conferring resistance to FQs (9 of 47 versus 5 of 55, P = 0.160) or SLIDs (7 of 47 versus 2 of 55, P = 0.077) alone was higher in MDR-TB strains collected during 2013 to 2019 compared to 2006 to 2012, however, the differences were not statistically significant (Table 4). On the contrary, the prevalence of mutations conferring resistance to FQs or SLIDs (pre-XDR-TB) was significantly higher during 2013 to 2019 compared to 2006 to 2012 (16 of 47 versus 7 of 55, P = 0.016) ( Table 4).

Discussion
Kuwait is an Arabian Gulf country in the Middle East. The total population of 4.8 million individuals in 2019 comprised 30% Kuwaitis and 70% expatriate workers or their dependents (https:// www. paci. gov. kw/ Defau lt. aspx). Most expatriates in Kuwait originate from TB endemic countries of South/Southeast Asia and Africa (such as India, Bangladesh, Pakistan, Philippines, Egypt, Sudan and Ethiopia) 31,33,34 . Kuwait has a low (~ 24 cases per 100 000 population) incidence of TB and a low (~ 1.1%) incidence of MDR-TB 31,35 . Nearly 85% of all TB cases and > 90% of MDR-TB cases occur in expatriate population 31,35 . These cases mainly arise due to reactivation of latent TB infection acquired previously by TB patients in their respective countries even though all expatriates are screened for the evidence of active TB disease at the time of their entry into Kuwait [33][34][35] . All clinical specimens from suspected TB patients are tested by GeneXpert MTB/RIF assay for rapid diagnosis of active TB disease and detection of RIF resistance in addition to routine processing for smear microscopy and culture by MGIT 960 system 30,31 . Although phenotypic DST against first-line drugs and STR is performed on all M. tuberculosis isolates for optimal patient management, phenotypic DST against second-line drugs is not performed, mainly due to low rate of MDR-TB 31,35 .
In this study we evaluated the performance of two line probe assays (gMTBDRslv1 and gMTBDRslv2) for rapid detection of mutations conferring resistance to FQs and SLIDs in M. tuberculosis and the results were extended by PCR-sequencing of respective gene loci. These assays identified 5 different gyrA mutations at 2   42 . Several studies have shown that A90V and D94A mutations cause low-level resistance (minimum inhibitory concentrations, MICs of 0.5 mg/L to 4 mg/L) to ofloxacin/levofloxacin while D94G and D94N mutations mostly cause high-level resistance to FQs (MICs of 4 mg/L to 32 mg/L) 27,[37][38][39]43 . Phenotypic DST of M. tuberculosis for first-line anti-TB drugs; RIF, EMB and PZA is problematic and many isolates with low-level yet clinically significant resistance to RIF and EMB are scored as drug-susceptible strains [15][16][17][18][19][20]28,29,44 . Similarly, phenotypic DST for FQs is also not perfect as some isolates with low-level resistance and containing specific gyrA (mainly A90V and D94A) mutations yield discordant results 25,27,38 . Taken together, the findings suggest that detection of gyrA mutations in few FQ-susceptible isolates is also very likely due to faulty phenotypic DST data rather than lack of association of gyrA mutations with FQ resistance.
All 23 isolates with gyrA/gyrB or rrs/eis mutation contained S315T mutation in katG and 19 of 23 isolates contained S450L mutation in rpoB. The mutation S315T in katG has minimal effects on fitness of M. tuberculosis and isolates with this mutation are more likely to remain virulent, are amenable to further transmission and so acquire resistance to additional drugs 3,8,45 . Similarly, the mutation S450L in rpoB has minimal effects on fitness as M. tuberculosis isolates with this mutation usually contain compensatory mutations in rpoA/rpoC, remain infectious and thus acquire resistance to additional drugs in new patients 3,8,45,46 . Although the prevalence of mutations conferring resistance to FQs (gyrA/gyrB) and SLIDs (rrs/eis) in MDR-TB strains was higher in isolates with an rpoB S450L mutation versus isolates with all other mutations (19 of 73 versus 4 of 29), the difference did not reach statistical significance (P = 0.182). High prevalence of gyrA and rrs mutations conferring resistance to FQs and SLIDs, respectively, in MDR-TB strains carrying S450L mutation in rpoB gene was also reported from India 47 . The mutation patterns and fingerprinting data by spoligotyping showed that nearly all 14 isolates with a gyrA/gyrB mutation were genotypically distinct strains. These 14 isolates were obtained from expatriate patients with 13 of 14 patients originating from the Indian subcontinent (India, n = 11, Bangladesh, n = 1; Nepal, n = 1). The data are consistent with previous observations showing that most MDR-TB strains in Kuwait are genotypically different as they arise mostly as a result of reactivation of latent M. tuberculosis infection acquired previously by expatriate patients in their respective countries 34,48,49 . Furthermore, the occurrence of gyrA mutations in MDR-TB strains isolated mostly from Indian patients is consistent with high prevalence of resistance of multidrug-resistant M. tuberculosis isolates from India to FQs 40,47,50,51 . Also, the detection of D94G (n = 5) and A90V (n = 3) mutations in 8 of 10 FQ-resistant strains is consistent with the high prevalence of these mutations in MDR-TB strains from India 40,47,50,51 . www.nature.com/scientificreports/ Eight MDR-TB isolates contained A1401G mutation in rrs that confers resistance to SLIDs and 6 of 8 isolates (isolate no. 3762/322, 463/57, 818/153, 6633/640, 10,246/708 and 16,359/764) were genetically similar and were recovered from Kuwaiti patients, likely as a result of local transmission of MDR-TB in Kuwait even though the index case remained elusive 34 . Although phenotypic DST against SLIDs was not performed in this study, all MDR-TB isolates with a defined rrs mutation in a recent study from South Africa showed phenotypic resistance to SLIDs 39 . A multicenter cohort study carried out in seven high TB burden countries has recently shown that inaccurate DST by phenotypic methods leads to under-treatment of drug-resistant TB and increased patient mortality 52 . The study further showed that rapid molecular DST in place of phenotypic DST for first-line and second-line drugs is required for improved outcome for patients with MDR-TB, pre-XDR-TB or XDR-TB 52 .
Although XDR-TB was not detected in Kuwait, the prevalence of mutations conferring resistance to FQs alone (9 of 47 versus 5 of 55, P = 0.160) or SLIDs alone (7 of 47 versus 2 of 55, P = 0.077) was higher, though not statistically significant, in MDR-TB strains collected during 2013 to 2019 compared to 2006 to 2012. However, the prevalence of mutations conferring resistance to FQs or SLIDs (pre-XDR-TB) was significantly higher during 2013 to 2019 compared to 2006 to 2012 (16 of 47 versus 7 of 55, P = 0.016) strongly suggesting that the prevalence of these mutations is increasing in Kuwait. The prevalence of resistance to FQs has also increased among MDR-TB strains in recent years in India 53-56 and China 43 . These findings are relevant as most (11 of 14, 79%) MDR-TB strains in Kuwait with gyrA/gyrB mutations were obtained from expatriate patients of Indian origin.
Application of gMTBDRslv1 assay also detected an embB codon 306 mutation in 59 of 102 (58%) MDR-TB strains. Interestingly, only 25 of these 59 isolates were phenotypically resistant to EMB while the remaining 34 isolates were EMB-susceptible. The embB codon 306 mutations confer low-level resistance to EMB which is often missed by the faulty phenotypic DST methods [20][21][22]24,44 . Thus rapid detection of EMB resistance in MDR-TB strains by gMTBDRslv1 test will avoid treatment with an ineffective drug, evolution of further resistance and drug-related toxicity 57 .
Our study has a few limitations.
(1) Phenotypic DST of pansusceptible and MDR-TB strains against FQs and SLIDs was not performed, (2) The presence of mutations in gyrA/gyrB genes conferring resistance to FQs and in rrs/eis genes conferring resistance to SLIDs by gMTBDRslv1 and gMTBDRslv12 tests was carried out in culture isolates and not directly on clinical samples obtained from TB patients and (3) The outcome of infection among MDR-TB patients was not available as most patients were expatriates who were sent back to their respective countries after the initial objective of sputum smear-negative status was achieved.
In conclusion, our results show that combined use of gMTBDRslv1 and gMTBDRslv2 assays in 102 MDR-TB strains detected mutations in 12 and 2 isolates in gyrA and gyrB, respectively conferring resistance to FQs and in 8 and 1 isolate in rrs and eis, respectively, conferring resistance to SLIDs. Mutations in embB were also detected in 59 MDR-TB isolates. These mutations were absent among 50 pansusceptible M. tuberculosis isolates. Although XDR-TB was not detected, an increasing trend in the frequency of pre-XDR-TB was evident in Kuwait as the frequency of mutations conferring resistance to FQs or SLIDs was significantly higher among isolates collected during 2013-2019 versus 2006-2012. Application of both tests is warranted for proper management of MDR-TB patients in Kuwait, a low TB/MDR-TB setting, as gMTBDRslv2 detected resistance to FQs and/or SLIDs in 3 additional isolates while gMTBDRslv1 additionally detected resistance to EMB in 58% of MDR-TB isolates. The embB mutations confer low-level resistance to EMB which is often missed by the faulty phenotypic DST methods and their detection in MDR-TB strains by gMTBDRslv1 will avoid treatment with an ineffective drug and drug-related toxicity.

Materials and methods
Clinical specimens and M. tuberculosis isolates. A total of 102 MDR-TB isolates cultured during 2006 to 2019 from 102 TB patients (representing all available MDR-TB strains) and 50 randomly selected M. tuberculosis isolates fully susceptible to first-line anti-TB drugs (pansusceptible strains) collected from 50 TB patients were analyzed. The MDR-TB strains were grown from 78 respiratory and 24 non-respiratory specimens while pansusceptible strains were cultured from 34 respiratory and 16 non-respiratory clinical specimens. All clinical specimens were obtained from suspected TB patients after obtaining verbal consent only as part of routine patient care, diagnostic work-up and resistance surveillance at Kuwait National TB Control Laboratory (KNTCL) before initiation of treatment with anti-TB drugs. The study was approved by the Health Sciences Center Ethical Committee, Kuwait University (Approval no. VDR/EC/3451 dated 18-12-2018) and all experimental procedures and investigations were performed in accordance with their guidelines and regulations. Since the study did not involve direct contact with patients and the results are reported on deidentified samples without revealing patient identity, the need for informed consent was waived by the Health Sciences Center Ethical Committee.
Sterile clinical specimens were processed directly for the cultivation of mycobacteria whereas non-sterile samples were processed by using N-acetyl-L-cysteine and sodium hydroxide (NALC/NaOH) at KNTCL for culture. The cultures were grown in liquid media-based automated mycobacteria growth indictor tube (MGIT) 960 system according to manufacturer's instructions (Becton Dickinson, Sparks, MD, USA) and as described previously 31,44 . All MGIT cultures were positive for the presence of acid-fast bacilli (AFB) by Ziehl-Neelsen smear microscopy and for the presence of M. tuberculosis complex DNA by AccuProbe DNA probe assay, performed as described previously 31 . Phenotypic DST by Bactec MGIT 960 system. Phenotypic DST of all M. tuberculosis isolates was performed with the automated MGIT 960 system for RIF, INH, EMB and STR by using the SIRE drug kit used according to manufacturer's instructions and as described previously 31,44 . The KNTCL regularly participates in drug susceptibility proficiency testing. Detection of mutations conferring resistance to RIF and INH. All isolates were tested by commercial GenoType MTBDRplus line probe assay (Hain Lifesciences, Nehren, Germany) for the detection of mutations in hot-spot region of rpoB gene for RIF resistance and for detection of mutations at katG codon 315 (katG) and inhA regulatory region (inhA) for INH resistance, according to kit instructions and as described previously 59 . Water was used in place of DNA for negative controls. PCR-sequencing of hot-spot region and N-terminal region of rpoB gene and katG and inhA was carried out for isolates showing no mutation or a nonspecific mutation indicated by lack of hybridization with a wild-type probe only. DNA sequencing was performed as described previously 59 . M. tuberculosis codon numbering system was used for the rpoB gene 60 .

Detection of mutations conferring resistance to FQs and SLIDs
The gMTBDRslv1 contains probes targeting gyrA for FQ resistance, rrs for resistance to SLIDs and embB gene codon 306 for EMB resistance. The gMTBDRslv2 contains probes targeting gyrA and gyrB for FQ resistance, rrs for resistance to SLIDs and eis for resistance to KAN 36 . Both gMTBDRslv1 and gMTBDRslv2 were obtained commercially (Hain Lifesciences) and were used according to kit instructions. Lack of detection of specific mutation or detection of resistance only by lack of hybridization with wild-type probes was confirmed by PCR amplification followed by DNA sequencing (PCR-sequencing) of the corresponding gene fragment.