Prevalence of and Molecular Basis for Tuberculosis Drug Resistance in the Republic of Georgia: Validation of a QIAplex System for Detection of Drug Resistance-Related Mutations

Abstract

We developed a QIAplex system for the simultaneous detection of 24 Mycobacterium tuberculosis gene mutations responsible for resistance to isoniazid (INH), rifampin (RIF), streptomycin (STM), and ethambutol (EMB) in 196 M. tuberculosis isolates recovered in the Republic of Georgia. In comparison to phenotypic susceptibility tests, the QIAplex showed sensitivity and specificity of 85.4% and 96.1% for INH, 94.4% and 99.4% for RIF, 69.6% and 99.2% for STM, 50.0% and 98.8% for EBM, and 86.7% and 100.0% for multidrug resistance, respectively. The dominant resistance mutations revealed were a mutation in katG resulting in S315T (katG S315T), rpsL K43R, and rpoB S531L. Mutations katG S315G and S315T and rpoB S531L were detected with higher frequencies in pretreated patients than in naive patients (P < 0.05). Simultaneous detection of 24 common drug resistance-related mutations provides a molecular tool for studying and monitoring M. tuberculosis resistance mechanism and epidemiology.

Tuberculosis (TB) has emerged as a major public health problem in the former Soviet republics. Rates of TB in the Republic of Georgia increased dramatically following the breakup of the Soviet Union due to an economic decline, the general failure of TB control, and the reduction in other health care-associated services (1, 16). Additionally, this situation was exacerbated by a civil war that followed independence in 1992 to 1993; the war resulted in the internal displacement of several thousand people in whom high rates of TB were detected (16). According to the World Health Organization (WHO) Global TB Report 2006, there were 3,717 new TB cases in the Republic of Georgia in 2004; indeed, Georgia has one of the highest TB burdens in eastern Europe (1, 16).

Although a number of published clinical trials have demonstrated TB cure rates of over 95% with minimal relapse rates, the global rates of multidrug-resistant TB (MDR-TB) are increasing, mainly due to the human immunodeficiency virus epidemic as well as the lack of funding for health infrastructure. This, in turn, leads to incorrect or incomplete treatment, which increases the resistance rate (11). According to a global survey performed by WHO and the Centers for Disease Control and Prevention (CDC) between 2000 and 2004, 20% of 17,690 Mycobacterium tuberculosis isolates were considered MDR-TB isolates (2). Among eastern European countries, Georgia contributed 7% of the MDR-TB new smear-positive cases and 27% of the retreatment smear-positive cases (16). The molecular mechanisms responsible for the high incidence of MDR-TB in Georgia have not been previously studied.

A prompt determination of antimicrobial susceptibility profiles of clinical isolates of M. tuberculosis is critical for choosing effective drug therapy and also for preventing the propagation of drug-resistant strains (11). However, as the conventional antimicrobial susceptibility testing of mycobacteria depends on the microorganism being cultured, such testing requires 4 to 6 weeks after primary isolation. Thus, molecular biology tools have been developed in order to provide a rapid susceptibility profile (5, 7, 9, 12, 15). In particular, multiplex formats have been developed in order to detect a number of TB drug resistance-related mutations (6, 10, 17).

We report here the development of a QIAplex system for determining the prevalence and the molecular basis of TB drug resistance in the Republic of Georgia. This technology uses a target-enriched multiplex PCR (4) to simultaneously amplify and detect 24 mutations in the M. tuberculosis genome responsible for resistance to isoniazid (INH), rifampin (RIF), streptomycin (STM), and ethambutol (EMB) (7, 9, 10, 12, 15, 17). The performance of the technology was validated in comparison to standard antimicrobial susceptibility methodology previously performed with clinical M. tuberculosis isolates recovered from patients referred to the National Center of Tuberculosis and Lung Diseases, Republic of Georgia.

(This study was presented in part at the 107th General Meeting of the American Society for Microbiology, Toronto, Canada, 21 to 25 May 2007.)

TB isolates and characterization.

Consecutive isolates of M. tuberculosis recovered from patients referred to the National Center of Tuberculosis and Lung Diseases in Tbilisi, Republic of Georgia, between January and March 2006 were evaluated. Collected clinical specimens were processed by standard methods and cultured on Lowenstein-Jensen medium for isolation of mycobacteria. Sample demographic and treatment information was also documented.

Phenotypic antimycobacterial susceptibility testing.

Susceptibility testing for antimycobacterial drugs was performed by the method of absolute concentration (10) and was considered to be the “gold standard” for this study. MDR-TB isolates were defined when they were resistant to INH and RIF. Mycobacterial suspension was performed from the primary culture and the turbidity adjusted to 1 McFarland standard with sterile saline. A series of 10-fold dilutions were prepared, and 0.2 ml was inoculated onto media containing the following first-line TB drugs: STM (4 μg/ml), RIF (40 μg/ml), and EBM (2 μg/ml). The INH (0.2 μg/ml)-containing media were inoculated with 0.2 ml of a 100-fold dilution of the suspension. All inoculated sets were incubated at 37°C in an atmosphere of 5 to 10% CO₂ for 28 days.

Nucleic acid extraction.

A loopful of colony from the Lowenstein-Jensen medium was suspended in 200 μl RNase-free water and boiled for 10 min as previously described (14). The suspension was centrifuged at 13,000 rpm for 10 min, and the supernatant was stored at −20°C until used.

Target-enriched multiplex technology.

The QIAplex TB assay (catalog no. 015-01-S) was used in a 50-μl reaction mixture composed of 6 μl of QIAplex SuperPrimers (mixture of QIAplex SuperPrimers and gene-specific nested primers for the amplification of katG, inhA, kasA, rpoB, rpsL, rrs, and mabA loci), 25 μl of Multiplex Master Mix (Qiagen Inc., Valencia, CA), 5 μl of extracted DNA, and 14 μl of water. Amplification was carried out with the five-stage QIAplex cycling program, and the PCR products were further characterized using a suspension array for multiplex detection on a Luminex 100 instrument (Luminex, Austin, TX) (4). Each targeted region and its respective wild-type sequence along with the allelic variation are listed in Table 1. Results for each channel are expressed as the median fluorescent intensity (MFI) value. The presence of a mutation was indicated by the difference in the hybridization MFI signal between the wild-type probe and the mutant probe. A hybridization MFI signal for a mutant ≥30% more than the wild-type signal was considered positive for the presence of the mutation.

TABLE 1.

TB resistance-related mutations detected by the QIAplex TB assay and prevalence between pretreated and new cases^a

Drug	Target locus	Nucleotide change	Amino acid change	No. detected	No. of isolates (% detection) recovered from:		P
					New cases (n = 132)	Pretreated cases (n = 64)
INH	katG	AGC→ACC	S315T	28	9 (6.8)	19 (29.7)	0.001
	katG	AGC→AAC	S315N	0
	katG	AGC→ATC	S315I	0
	katG	AGC→AGG	S315R	0
	katG	AGC→GGC	S315G	4	0 (0.0)	4 (6.3)	0.039
	ndh	CGC→TGC	R268H	0
	mabA	−15 C→Tb	NA	9	6 (2.3)	3 (4.7)	0.610
RIF	rpoB	Δ of GAC	Δ at 516	0
	rpoB	GAC→GTC	D516V	1	0 (0.0)	1 (1.6)	0.327
	rpoB	CAC→TAC	H526Y	0
	rpoB	CAC→GAC	H526D	2	1 (0.8)	1 (1.6)	0.545
	rpoB	CAC→CGC	H526R	0
	rpoB	CAC→CCC	H526P	0
	rpoB	TCG→TGG	S531W	0
	rpoB	TCG→TTG	S531L	15	5 (3.8)	10 (15.6)	0.005
STM	rpsL	AAG→AGG	K43R	37	20 (15.2)	17 (26.6)	0.056
	rrs	C→T at 491	NA	5	4 (3.0)	1 (1.6)	0.472
	rrs	C→T at 512	NA	0
	rrs	A→T at 513	NA	3	1 (0.8)	2 (3.1)	0.249
	rrs	C→T at 516	NA	4	1 (0.8)	3 (4.7)	0.103
EMB	embB	ATG→ACA	M306I	5	2 (1.5)	3 (4.7)	0.197
	embB	ATG→ACC	M306I	0
	embB	ATG→ACT	M306I	0
	embB	ATG→GTG	M306V	0

^aNA, not applicable; Δ, deletion.

^bAt the promoter region.

A total of 196 M. tuberculosis isolates were recovered during the study period. Most of the isolates were from pulmonary sites, including sputum (95.5%) and bronchoalveolar lavage (2%) and pleural (1.5%) fluid. Only two isolates were obtained from urine (1%). Patients enrolled in this study were between 16 and 85 years of age (mean ± standard deviation: 39.3 ± 14.5 years), and 162 (82.7%) were male. Among them, 131 (66.8%) had not received any previous anti-TB treatment and 65 (33.2%) had been treated previously; they were considered new and retreatment cases, respectively. All the new patients and 35 (17.9%) of the retreatment patients received a relevant standardized treatment regimen (second category) using first-line TB drugs. The remaining 30 (15.3%) retreatment patients received nonstandardized (chaotic) treatment using both first- and second-line TB drugs.

Phenotypic testing for susceptibility to anti-TB drugs INH, RIF, STM, and EMB was performed by the absolute-concentration method on solid agar medium. Among all 196 isolates tested, 82 (41.8%) were resistant to at least one antimycobacterial agent. Among the resistant isolates, 41 (20.9%), 18 (9.2%), 49 (25.0%), and 5 (2.6%) were resistant to INH, RIF, STM, and EMB, respectively. None of the isolates tested showed resistance to EMB alone, and four isolates were resistant to all four drugs tested (Table 2). Fifteen (7.7%) isolates were considered multidrug resistant due to phenotypic resistance to both INH and RIF.

TABLE 2.

Sensitivity and specificity of the QIAplex TB assay in comparison to phenotypic antimycobacterial susceptibility results^a

Antimycobacterial drug(s)	No. of isolates with result (standardb/ QIAplex):				Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)
	R/R	R/S	S/R	S/S
INH	35	6	6	149	85.4	96.1	85.4	96.1
RIF	17	1	1	177	94.5	99.4	94.7	99.4
STM	48	21	1	126	69.6	99.2	98.6	85.8
EMB	3	3	2	188	50.0	98.9	60.0	98.4
INH + RIF (MDR-TB)	13	2	0	181	86.7	100.0	98.9	100.0

^aPPV, positive predictive value; NPV, negative predictive value; R, resistant; S, susceptible.

^bPhenotypic susceptibility profiles determined by the absolute-concentration method.

The performance of the QIAplex TB assay was validated against the phenotypic susceptibility test results. Among phenotypically resistant isolates, TB resistance-related mutations were detected in 41 (20.9%) for INH, 18 (9.2%) for RIF, 69 (35.3%) for STM, and 6 (3.1%) for EMB, giving sensitivities of 85.4%, 94.5%, 69.6%, and 50% and specificities of 96.1%, 99.4%, 99.2%, and 98.9%, respectively. The positive and negative predictive values were 85.4% and 96.1% for INH, 94.7% and 99.4% for RIF, 98.6%, 85.8% for STM, and 60.0% and 98.4% for EMB, respectively (Table 2). The sensitivity, specificity, and positive and negative predictive values were 86.7%, 100.0%, 100.0%, and 98.9%, respectively, when the QIAplex was used to detect MDR-TB isolates (Table 2). The QIAplex provided excellent specificities, i.e., small major-error rate, for drug resistance prediction for all four drugs. While satisfactory sensitivities for predicting RIF resistance (94.5%) were observed, relatively poor sensitivities, i.e., a very major-error rate, for predicting resistance to STM and EMB were observed (Table 2).

The molecular mechanisms for anti-TB drug resistance in Georgia were analyzed. Among 24 TB resistance-related mutations, 11 mutations were detected in our study, with frequencies from 1 to 37 (Table 1). The QIAplex detected mutations in 35 of 41 isolates (85.4%) phenotypically resistant to INH, with none of these mutations detected in the remaining 6. The most common mutations found were as follows: a mutation in katG resulting in S315T (katG S315T; 26 isolates), followed by katG S315G (4 isolates) and a mabA mutation (5 isolates) (Table 3). Among 18 isolates phenotypically resistant to RIF, rpoB S531L (14 isolates) was the dominant mutation detected, and only one isolate with none of these mutations was detected by the QIAplex (Table 3). Among 69 isolates phenotypically resistant to STM, rpsL K43R (37 isolates) was the dominant mutation detected, while none of the mutations were detected in 21 resistant isolates (Table 3). The embB M306I mutation was detected in only three of six isolates phenotypically resistant to EMB (Table 3). Both katG S315G (86.7%) and rpoB S531L (80.0%) were the dominant mutations detected in 15 MDR-TB isolates (Table 3).

TABLE 3.

Molecular basis of antimycobacterial resistance in phenotypically resistant clinical isolates in the Republic of Georgia

Phenotype	No. of resistance-related mutations detected	Resistance-related mutations detected (no.)
INH resistance	41	katG S315T (26), katG S315G (4), mabA (5), unknown (6)
RIF resistance	18	rpoB S531L (14), rpoB D516V (1), rpoB H526D (2), unknown (1)
STM resistance	69	rpsL K43R (37), rrs 491 C→T (4), rrs 513 A→T (3), rrs 516C→T (4), unknown (21)
EMB resistance	6	embB M306I (3), unknown (3)
INH + RIF resistance (MDR-TB)a	15	katG S315T only (1), katG S315T + rpoB S531L (11), katG S315T + rpoB H526D (1), mabA + rpoB D516V (1), rpoB S531L only (1)

^aINH and RIF resistance-related mutations are listed. Two isolates possessed only INH or RIF resistance single mutations.

The frequencies of TB drug resistance-related mutations correlated with the treatment status of these cases (Table 1). Except for EMB, drug resistance-related mutations were detected at a significantly higher rate in the pretreated cases than in the new cases (P = 0.014, P = 0.000, and P = 0.001 for STM, RIF, and INH, respectively) for the antimycobacterial drugs tested. Among the individual drug resistance-related mutations, three mutations, including S531L (rpoB) and S315G and S315T (katG gene), were detected at a statistically higher rate in the pretreated cases (P = 0.005, 0.039, and 0.001, respectively). S315G (katG) and D516V (rpoB) were detected only in isolates recovered from pretreated cases (Table 1). In 69 isolates phenotypically resistant to STM, the QIAplex TB assay detected 23 resistance-related mutations in 27 isolates (85.2%) recovered from the pretreated patients; these values are significantly higher than those for the new patients (Table 1).

The QIAplex TB assay was able to detect 94.5% of the RIF-resistant isolates tested, with only two discrepant cases against the phenotypic susceptibility assay being observed. In 95% of the RIF-resistant isolates, the phenotypic resistance is due to mutation in the rpoB gene encoding the β-subunit of the RNA polymerase (6), making this gene a good marker for molecular assays for detecting RIF drug resistance. Previous studies have shown that specific mutations in the rpoB gene may also be associated with low-level RIF resistance, which is not detected by the phenotypic methodology (13). In our study, rpoB S531L was the dominant mutation detected, which was related to RIF resistance. In addition, one isolate presenting the rpoB S531L mutation was determined to be RIF susceptible by the standard phenotypic susceptibility assay, which suggests that it was one false-negative result.

The INH resistance mechanism is more complex and not as well understood as the mechanism responsible for RIF resistance. Currently only 80 to 90% of INH resistance phenotypes have been discovered to be associated with mutations in katG, inhA, ahpC, and ndh genes (15). In our study, the QIAplex was able to detect INH resistance-related mutations in 85.4% of phenotypically resistant isolates, showing a good correlation with the phenotypic method; nevertheless, detection of INH resistance remains a challenge for the molecular assay. Resistance to EMB was uncommonly detected, by either phenotypic methods or the QIAplex TB assay. This may reflect the local pattern of antimicrobial use, in which STM is frequently used instead of EMB as part of initial anti-TB therapy.

The QIAplex TB assay presented an overall low sensitivity (69.6%) for detecting STM-resistant isolates. It has been advocated that high-level resistance to STM is strongly associated with alteration in the drug target (3). However, 25 to 35% of the STM-resistant isolates do not show mutations in the rpsL and rrs genes, suggesting that additional resistance mechanisms exist mainly in isolates showing low levels of STM resistance (16). This suggests that further studies are required in order to identify these additional mechanisms to improve the diagnostic sensitivity of the molecular assays. Interestingly, the QIAplex TB assay detected significantly more STM resistance-related mutations (85.2%) in pretreated patients than in new patients. This may be probably due to specific selections of rpsL and rrs mutant strains during antimycobacterial therapy and suggests that this assay may be valuable in determining an anti-TB therapy regimen for pretreated patients.

The frequency of drug-related mutations found in the tested isolates recovered from the Republic of Georgia was in agreement with previous reports. Most of the INH-resistant isolates (91 to 94%) recovered from patients from the post-Soviet countries have shown alteration in the katG gene (8, 10, 12), which is similar to the frequency found by the QIAplex TB assay (86%). Additionally, this study also revealed a higher frequency of the rpoB S531L (82.4%) and rpsL K43R (77.1%) mutations among the RIF- and STM-resistant isolates, respectively. These alterations are more commonly found in phenotypically resistant isolates, varying from 55 to 75.7% for rpoB S531L and 54 to 61% for rpsL K43R, according to the studied population (6, 8, 12). In 15 MDR-TB isolates included in our study, 13 (86.7%) were positive for both katG S315T and rpoB S531L mutations, suggesting that a combination of the two genetic mutations is the dominant molecular basis of MDR-TB and an accurate marker for predicating MDR-TB in the Republic of Georgia.

This study also showed that drug resistance-related mutation rates for pretreated and new cases were significantly different. Antimycobacterial drug resistance-related mutations, which included three individual mutations, rpoB S531L and katG S315G and S315T, were detected in significantly higher numbers in pretreated cases than in new cases. Two mutations, katG S315G and rpoB D516V, were detected only in isolates recovered from pretreated cases. It has been shown that katG, rpsL K43R, and rpoB alterations are more frequently observed in isolates showing high-level INH, STM, and RIF resistance phenotypes, playing a role as secondary mutation, which are selected during drug therapy (5). Additionally, it has been postulated that mutations such as those in katG may also provide some survival advantage during patient treatment (5). In our study, among 69 isolates phenotypically resistant to STM, the QIAplex TB assay detected significantly higher resistance-related mutations in isolates recovered from pretreated patients than from new patients, supporting the theory that selections happened frequently during antimycobacterial drug treatment. An increased rate of detection of STM resistance-related mutations in pretreated patients suggests that this biomarker may be valuable when managing patients with treatment failure.

The QIAplex technology has proved to be a powerful tool for the rapid detection of multiple targets in a single reaction (4). This feature makes the detection of several specific drug resistance-related mutations in the M. tuberculosis genome possible. Although genetic tests are more accurate than analysis of resistance phenotypes for determining antimycobacterial drug resistance, our study confirmed one of the main disadvantages of genetic assays. Molecular mechanisms for antimycobacterial drug resistance remain incomplete, while novel resistance genes and mutations continue to emerge. Looking for known resistance-related mutations can miss new mechanisms of resistance, thereby resulting in the occurrence of very major errors in the clinical setting. This was indicated in our study when the QIAplex was used to predict STM and EMB resistance. It is important that new molecular mechanisms associated with resistance to antimycobacterial agents, especially STM and EMB, continue to be identified and considered if genetic methods are to become a reliable guide for decisions regarding initial therapy for TB. Until such knowledge is available, the currently used phenotypic methods for identifying resistance will continue to play an invaluable role in optimizing the therapy of people with TB.