Background
Widespread occurrence of drug-resistant tuberculosis (TB) and multidrug-resistant (MDR)-TB (infection with
Mycobacterium tuberculosis strain resistant at least to rifampicin, RIF and isoniazid, INH; the two most effective first-line anti-TB drugs) is a serious threat to TB control success worldwide. According to global annual surveys conducted by World Health Organization (WHO), an estimated 490,000 cases of MDR-TB occurred among 10.4 million new active TB cases in 2016 [
1]. Compared to drug-susceptible TB, treatment of MDR-TB is more expensive, drug regimens are more toxic and require longer (18–24 months) duration of treatment which often results in clinical failure or disease relapse [
1‐
3]. Unsuccessful treatment of MDR-TB is a risk factor for extensively drug-resistant TB (XDR-TB, infection with MDR-TB strains additionally resistant to a fluoroquinolone and injectable agent such as kanamycin, amikacin or capreomycin) which is often fatal in developing countries [
2‐
4]. Accurate drug susceptibility testing (DST) of
M. tuberculosis in clinical specimens and culture isolates to first-line drugs is crucial for rapid diagnosis of MDR-TB for proper patient management, for limiting further transmission of MDR-TB and development of XDR-TB [
2,
5]. Although rapid liquid culture-based phenotypic DST methods are considered as the gold standard by WHO for identifying resistance to RIF, INH and other first-line drugs, these methods still require 1–2 weeks to report results [
5,
6]. Molecular DST methods rapidly detect genetic mutations associated with drug resistance [
2,
7].
Resistance of
M. tuberculosis to RIF in 95–97% isolates is due to mutations in an 81-base pair (bp) hot-spot region (HSR) of the
rpoB gene (HSR-
rpoB) [
8]. The remaining 3–5% isolates contain mutations in N-terminal or cluster II region of the
rpoB gene or in other genes [
8,
9].
M. tuberculosis isolates with canonical (undisputed) HSR-
rpoB mutations (like Q513P, Q513K, H526R, S531 L or S531 W,
Escherichia coli numbering system, [
8]) as well as isolates with mutations (such as V146F) in the N-terminal end of the
rpoB gene exhibit high-level resistance to RIF which are readily detected by rapid phenotypic DST methods [
8,
9]. Some molecular assays targeting HSR-
rpoB are not specific as silent mutations in this region may occasionally lead to detection of false-positive RIF resistance [
10]. Recent studies have also shown that rapid liquid culture systems such as Mycobacteria growth indicator tube (MGIT) 960 system as well as the proportion method with shorter (4 weeks) incubation time often fail to detect strains exhibiting low-level (minimum inhibitory concentration, MIC of 0.5–2.0 μg/ml) resistance to RIF [
11‐
14]. These low-level RIF-resistant strains with increased MICs below the critical concentration mostly contain specific mutations within HSR-
rpoB, particularly at codon 511 (such as L511P), codon 516 (such as D516Y), codon 526 (such as H526N, H526L and H526S), and codon 533 (such as L533P) [
11‐
14]. Mutation I572F in cluster II region of the
rpoB gene also increases MICs below the critical concentration conferring low-level resistance to RIF [
11‐
14]. Nearly 30% RIF-resistant
M. tuberculosis isolates from Swaziland contained this (disputed) mutation and rapid liquid culture systems failed to accurately detect strains with this mutation [
15]. The clinical significance of some (D516Y and I572F) of these disputed (generally missed by rapid phenotypic DST methods) mutations in conferring resistance to RIF is indicated by gene replacement studies [
16]. Low-level resistance to RIF is clinically significant as patients infected with
M. tuberculosis strains with disputed
rpoB mutations often fail treatment or relapse [
17‐
20]. The prevalence of
M. tuberculosis isolates with disputed
rpoB mutations is largely unknown since phenotypic DST in low TB incidence, high income countries is usually carried out by rapid liquid culture-based methods. This study determined the occurrence of disputed mutations in HSR-
rpoB as well as I572F mutation in cluster II region of the
rpoB gene in clinical
M. tuberculosis strains phenotypically susceptible to RIF in Kuwait, a country with low (24 per 100,000) incidence of TB as well as a low (~ 1%) incidence of MDR-TB [
1,
21]. Common mutations conferring resistance to INH were also detected [
7,
9]. For isolates with an
rpoB mutation, molecular detection of resistance for two other first-line drugs for which rapid culture-based DST methods are either cumbersome (pyrazinamide, PZA) [
5,
6,
22,
23] or unreliable (ethambutol, EMB) [
5,
6,
24‐
26] was also performed.
Discussion
Kuwait is a low (24 cases per 100,000 population) TB incidence country [
21]. Nearly 80% of all TB cases and > 90% of drug-resistant TB (including MDR-TB) cases in Kuwait occur in expatriate subjects mainly originating from TB endemic countries of South/Southeast Asia (such as Bangladesh, India, Pakistan and Philippines) [
21,
27,
30]. Phenotypic DST of
M. tuberculosis isolates in Kuwait during 2002–2010 was carried out simultaneously by BACTEC 460 TB system as well as by MGIT 960 system, however, BACTEC 460 TB system was discontinued on January 1, 2011 and phenotypic DST has been performed only by MGIT 960 system since 2011. Resistance rates for any first-line drug, INH, 2 or more drugs (excluding INH + RIF with/without additional resistance, polydrug resistance) and INH + RIF (with/without additional resistance, MDR-TB) were reported as 12.4, 9.1, 2.5 and 0.9%, respectively [
21].
Rapid phenotypic methods (such as MGIT 960 system) are reliable for the detection of RIF-resistant strains with canonical
rpoB mutations, however, they often fail to detect
M. tuberculosis isolates with disputed
rpoB mutations that exhibit low-level resistance to RIF [
11‐
14]. In this study, we detected the presence of disputed
rpoB mutations in 242
M. tuberculosis isolates phenotypically susceptible to RIF. While mutations in HSR-
rpoB were detected by a line probe assay, mutation I572F in cluster II region of
rpoB gene was detected by developing a simple agarose gel-based MAS-PCR assay. Previously, I572F mutation was detected either by PCR-sequencing or by a real-time PCR assay [
15,
20,
29].
Our data showed that 4 of 242 (1.7%) RIF-susceptible isolates contained a disputed HSR-
rpoB mutation while I572F mutation was not detected. The H526N mutation was found in 2 isolates while D516Y and S531C mutations were found in 1 isolate each. The D516Y and H526N mutations have previously been shown to increase RIF MIC of
M. tuberculosis by 2–10 fold [
13,
16,
17]. Two of 4 patients infected with
M. tuberculosis strains carrying D516Y mutation in one study failed to respond to standard first-line treatment [
17]. Similarly, 4 of 6 patients infected with
M. tuberculosis strains carrying H526N mutation in another study either relapsed or failed treatment [
19]. The S531C is a rare HSR-
rpoB mutation and an isolate with S531C mutation was previously detected as RIF-resistant by BACTEC 460 TB system and/or agar dilution method [
32]. However, the MGIT 960 system failed to detect RIF resistance in
M. tuberculosis isolate (D4) with S531C mutation in this study. Interestingly, all 4 isolates with a disputed HSR-
rpoB mutation in our study were polydrug-resistant (resistant to 2 or more drugs excluding RIF) strains resistant at least to INH (MDR-TB strains).
The occurrence of disputed
rpoB mutations has been reported in few studies. The occurrence of disputed mutations in one study (carried out on
M. tuberculosis isolates from Bangladesh and Democratic Republic of Congo) was found to be much higher (in 13.1% of all isolates from Bangladesh and in 10.6% of isolates from Democratic Republic of Congo), however, unlike our study,
M. tuberculosis isolates from both these countries were cultured from retreatment (failure and relapse/reinfection after primary treatment) patients [
12]. Two previous studies have reported the occurrence of disputed mutations among
M. tuberculosis strains isolated from new TB patients [
18,
20]. The data obtained in archived samples from Bangladesh showed that RIF resistance varied significantly among
M. tuberculosis strains in different years (0.4% in 2005 versus 2.1% in 2010) and 7 of 1022 (0.7%) sputum samples contained
M. tuberculosis with a disputed
rpoB mutation [
18]. Since the prevalence of RIF resistance varied significantly in different years and the fact that Bangladesh is among the top 30 high TB burden countries [
1], the data may not be applicable for low TB incidence countries. The data from Australia showed that 5 of 214 drug-resistant isolates contained disputed HSR-
rpoB mutations [
20]. Among these 5 isolates, 4 isolates were monoresistant to INH and 1 isolate was monoresistant to PZA. None of 202 drug-susceptible
M. tuberculosis isolates contained an HSR-
rpoB mutation [
20]. In contrast, all our isolates with a disputed HSR-
rpoB mutation were polydrug-resistant strains resistant at least to INH but none of our 66 monodrug-resistant isolates contained an HSR-
rpoB mutation.
Although the occurrence of RIF (and INH) resistance in 4 of 242 (1.7%) isolates appears to be higher than the reported occurrence [
20] of MDR-TB in ~ 1% of
M. tuberculosis isolates in Kuwait, the data should be interpreted with caution as the yearly occurrence of disputed
rpoB mutations will likely be much lower (< 0.1%). This is because the proportion of polydrug-resistant strains was much higher (112 of 242, 46.3%) than normally reported (~ 3%) and the proportion of pansusceptible strains was much lower (64 of 242, 26.4%) than normally reported (~ 85%) among clinical
M. tuberculosis isolates in Kuwait [
21]. The rare occurrence of disputed
rpoB mutations is probably also related to the fitness cost associated with these mutations. Isolates with mutations at codon 526 and 531 (except S531 L mutation) generally exhibit significantly decreased fitness which may lead to their removal from circulation and replacement by strains with greater fitness [
33].
The clinical significance of disputed
rpoB mutations is indicated by gene replacement studies and patients infected with such strains often fail treatment or relapse [
17‐
20]. The clinical significance of disputed
rpoB mutations in our study is also indicated by the following observations. i) All 4 isolates were resistant to INH and contained S315 T mutation in
katG gene, an alteration that is strongly associated with acquisition of additional drug resistance leading to MDR-TB due to its minimal effects on fitness of tubercle bacilli [
34,
35], ii) Molecular screening for EMB resistance showed that 3 of 4 (including 1 isolate phenotypically susceptible and 2 isolates phenotypically resistant to EMB) isolates contained M306 V mutation in
embB gene that confers low-level resistance to EMB and is also strongly associated with MDR-TB phenotype [
36‐
38], iii) Molecular screening for PZA resistance showed that 2 of 4 isolates (including 1 of 2 isolates for which phenotypic DST for PZA could not be performed) contained a
pncA mutation that is either previously described in PZA-resistant strains [
22,
23] or is highly suggestive of PZA resistance due to nonsynonymous mutation affecting the structure of pyrazinamidase, iv) Molecular screening for resistance conferring mutations showed that 3 of 4 isolates with a disputed
rpoB mutation were actually resistant to three or all four first-line drugs (excluding streptomycin), and v) The isolates were genotypically unrelated as they were isolated at different (1 isolate each in 2005, 2010, 2011 and 2016) time points and all 4 isolates analysed by spoligotyping belonged to different
M. tuberculosis lineages. Unfortunately, the final treatment outcome for the patients yielding
M. tuberculosis isolates with disputed
rpoB mutations was not available as the isolates were recovered from expatriate TB patients who were sent back to their respective country after initial treatment objective (sputum smear-negative status) was achieved. Similar observations regarding lack of final treatment outcome involving expatriate patients have also been recorded at other geographical locations [
39].
Two other observations are noteworthy in our study. Among INH-resistant isolates, the frequency of
inhA-RR mutations was higher than
katG315 mutations (42% versus 30%) in INH-monoresistant strains, however, this ratio was reversed in isolates with acquisition of additional phenotypic resistance to one more drug (15% versus 77% for isolates with INH + SM resistance and 8% versus 40% for isolates with INH + EMB resistance, Table
1). Furthermore, the
inh-RR mutations were absent among INH-resistant isolates resistant to two other drugs. Our data support previous observations that fitness is not adversely affected by
katG315 mutation and they are more likely to acquire resistance to additional drugs [
34,
35]. Our data also support recent observations showing that
katG315 mutations are significantly associated with additional resistance to SM and EMB and are more likely to cause unfavourable treatment outcome while
inhA-RR mutations are mainly associated with additional resistance to SM only [
40]. Secondly, our data also reiterate the limitations of phenotypic DST for EMB as isolates with low-level EMB resistance are often missed by these tests [
25,
26] and phenotypic DST for PZA which often yields unreliable or no result [
5,
6]. In this regard, one isolate (D2) for which phenotypic PZA susceptibility could not be determined contained a non-synonymous (R2P) mutation in
pncA. To the best of our knowledge, this (R2P) mutation has not been described previously in the literature [
22,
23]. Substitution of arginine by proline at amino acid position 2 is likely involved in conferring resistance to PZA as many non-synonymous
pncA mutations described in the literature involve replacement of a wild-type amino acid with proline [
22,
23]. Thus, molecular screening provides drug resistance profiles that help in proper management of MDR-TB patients as ineffective drugs are not included in drug regimens [
3,
9,
41,
42].
Our study has a few limitations. (i) The MIC values of the isolates with a disputed rpoB mutation to RIF were not determined and (ii) The details of the patient’s treatment history and clinical outcome were not available as all 4 patients were expatriate subjects who were sent back to their respective country after initial treatment objective (sputum smear-negative status) was achieved.