First Evaluation of an Improved Assay for Molecular Genetic Detection of Tuberculosis as Well as Rifampin and Isoniazid Resistances

Abstract

The commercially available line probe assay MTBDRplus 2.0 (Hain Lifescience, Nehren, Germany) was evaluated for its ability to detect Mycobacterium tuberculosis complex (MTBC) and mutations conferring resistance to rifampin (RMP) and isoniazid (INH) directly in smear-negative and smear-positive pulmonary clinical specimens under routine laboratory conditions. A total of 348 samples originating from Moldova, a high-incidence country for tuberculosis (TB), were investigated. Two hundred fifty-seven (73.9%) were smear negative, 12 samples were excluded, and 81 (23.3%) were smear positive. Two DNA extraction methods were applied. Compared to culture and clinical data as the reference standard (adapted from Vadwai V et al., J. Clin. Microbiol. 49:2540–2545, 2011), overall sensitivity and specificity were 87.6 and 99.2%, respectively. One hundred four of the 257 smear-negative samples turned out to be culture positive, and 20 were MTBC culture negative but were positive based on clinical symptoms. The combined sensitivity and specificity in the subgroup of smear-negative samples were calculated to be 79.8 and 99.2%, respectively. MTBDRplus 2.0 detected RMP and INH resistance with sensitivity and specificity of 94.3 and 96.0%, respectively. In conclusion, the MTBDRplus 2.0 assay is a rapid and highly sensitive test for the detection of M. tuberculosis strains from smear-positive and -negative clinical specimens and provides additional information on RMP and INH resistance status, which can easily be included in routine laboratory work flow.

INTRODUCTION

Every year, there are almost nine million new cases of infection with Mycobacterium tuberculosis. There are approximately two million deaths per year due to tuberculosis (TB) infection, and TB remains the leading cause of death among individuals infected with HIV (http://www.cdc.gov/hiv/resources/factsheets/).

Multidrug-resistant tuberculosis (MDR-TB), defined as resistance to at least rifampin (RMP) and isoniazid (INH), threatens TB control programs in many parts of the world (20). Increases of drug resistance in TB can lead to virtually untreatable extensively drug-resistant TB (XDR-TB), defined as MDR-TB plus additional resistance to fluoroquinolones and one of the injectable second-line drugs kanamycin, amikacin, and capreomycin (7, 20).

MDR-TB and TB/HIV coinfection are threats to the Republic of Moldova's efforts to reach global targets. The National DRS (2006) confirm the extent of drug-resistant TB in the Republic of Moldova. From 2001 to 2006, primary MDR-TB increased from 6.3 to 19.4% and acquired MDR-TB increased from 36.7 to 51%. The prevalence of TB/HIV coinfection was 4.7% (264 TB/HIV cases in a total of 5,591 TB cases) in 2009. In 2010, MDR-TB was detected in 25% of the new TB cases and in 61% of the retreatment TB cases (22).

The National TB Reference Laboratory (NRL) is located at the Phthisiopneumology Institute and functions as the coordinating body for microbiological services for TB diagnostics in Moldova. The NRL performs microscopy, the culture of M. tuberculosis, drug susceptibility testing (DST) for first- and second-line drugs, molecular resistance testing, the training of laboratory network staff, the standardization of laboratory methods for the country, the epidemiological survey of TB drug resistance, and external quality control for other laboratories. The laboratory annually performs 75,000 investigations for TB diagnostics. The laboratory participates in external quality assurance (EQA) (Supranational Reference Laboratory [SNRL], Borstel, Germany). The last results of external quality control (EQC) demonstrated 100% coincidence of DST for first-line drugs and 96% for second-line drugs.

The rapid detection of MDR-TB ensures treatment with appropriate drugs, thereby reducing morbidity, mortality, economic costs, the further transmission of infection, and the emergence of XDR strains (3).

Line probe assays (LPA) for the fast and reliable detection of M. tuberculosis complex (MTBC) and resistance to RMP and INH have recently been endorsed by the WHO (19, 21). The commonly used MTBDRplus 1.0 (Hain Lifescience, Nehren, Germany), which detects RMP and INH resistance, was evaluated with culture samples and smear-positive specimens in a variety of settings (6, 9, 10). As the sensitivity for the detection of mycobacteria by microscopy is known to be low (16), a substantial number of bacteria are not detected, and culture methods have to be performed to detect mycobacteria sufficiently in smear-negative samples. To overcome this limitation, the MTBDRplus 2.0 test (Hain Lifescience, Nehren, Germany) was further improved to detect mycobacteria and their resistance status against RMP and INH in smear-positive, smear-negative, and culture-positive specimens.

Here, we describe the evaluation of this new MTBDRplus 2.0 test for smear-positive and smear-negative pulmonary specimens in a high-TB-prevalence country.

MATERIALS AND METHODS

Study design.

The main purpose of this study was to determine the usefulness of MTBDRplus 2.0 for detecting Mycobacterium tuberculosis and its resistance to RMP and INH in smear-negative pulmonary samples in a routine setting. Therefore, we did not exclude patients under treatment.

Diagnostic performance characteristics were determined using 338 sputum samples from the daily laboratory routine by comparing MTBDRplus 2.0 to smear microscopy, culture (cultivation in Bactec MGIT960 and Löwenstein-Jensen agar), mycobacterial identification of positive cultures using GenoType Mycobacterium CM (Hain Lifescience), and phenotypic drug susceptibility testing (DST). Since technicians performing molecular and reference tests were not aware of patient data or results of other tests, biases were minimized.

The study site was located in a country with a high prevalence of MDR-TB. Microscopy and cultivation methods were performed onsite. DNA extractions from decontaminated sputum samples followed by MTBDRplus 2.0 testing were carried out in a second laboratory at the same time. The DNA-based PCR/LPA test was compared to the following microbiological methods: microscopy, culture, and DST from liquid and solid media.

To prevent DNA contamination, we used strict room separation, including a work flow from setting up the PCR to hybridization. Furthermore, a negative control was included in every run and was always negative, giving a strong indication of no laboratory contamination.

Patient specimen processing.

Until decontamination, sputum specimens were kept for a maximum of 2 days in sterile plastic containers at 4°C. All samples were processed with N-acetyl-l-cysteine and sodium hydroxide (NALC-NaOH) decontamination (8). After centrifugation, pellets were resuspended in 1.5 ml of phosphate buffer (pH 6.8).

Microscopy, culture, and DST.

The samples were processed according to international guidelines using the NALC-NAOH decontamination procedure (final NaOH concentration, 1%) and the Ziehl-Neelsen (ZN) technique for smear staining (8). One mycobacterial growth indicator tube (MGIT; Becton Dickinson) was prepared for each resuspended sample pellet by adding 0.8 ml oleic acid-albumin-dextrose-catalase (OADC)-PANTA (Becton Dickinson). A 0.5-ml aliquot of the sample was inoculated into the tube and incubated at 37°C in an automated Bactec MGIT960 (Becton Dickinson) device for a maximum of 42 days with continuous monitoring. Positive MGIT tubes were removed from the machine, and the time to positivity (TTP) was recorded. The presence of acid-fast bacilli was confirmed by Ziehl-Neelsen microscopy. Positive cultures were confirmed as MTBC using the GenoType Mycobacterium CM line probe test (Hain Lifescience, Nehren, Germany). Indirect DST was performed using the proportion method on Middlebrook 7H11 agar slants with 1.0 μg/ml RMP and 0.2 μg/ml INH.

DNA extraction.

To compare the efficiency of different DNA extraction methods and to determine the sensitivity and specificity of MTBDRplus 2.0, two different extraction methods were applied for DNA/RNA extraction from clinical specimens: (i) an automated extraction method using the GenoXtract instrument (Hain Lifescience, Germany) in combination with the GXT DNA/RNA extraction kit (Hain Lifescience, Germany), and (ii) a manual method using the GenoLyse kit (Hain Lifescience, Germany) for DNA extraction. The automated DNA/RNA extraction device GenoXtract is based on magnetic beads and handles up to 12 samples in parallel using prefilled cartridges. DNA extraction with GenoXtract and the GTX DNA/RNA extraction kits were performed according to their instructions for use. Briefly, after the loading of the GenoXtract with the prefilled cartridges and disposable pump units, 700 μl of decontaminated sample was added. Ten μl eluate was directly used for PCR.

DNA extraction with GenoLyse was performed according to the manufacturer's protocol. Briefly, 500 μl of decontaminated sample material was transferred into a 1.5-ml screwcap tube and centrifuged for 15 min at 10,000 × g in a standard tabletop centrifuge. The pellet was resuspended in 100 μl of an alcalic lysis buffer and incubated for 5 min at 95°C in a water bath. Subsequently, 100 μl of neutralization buffer was added to lysate, vortexed, and centrifuged for 5 min at full speed in a tabletop centrifuge. Five μl supernatant was directly used for PCR.

PCR.

To reduce workload and increase sensitivity, MTBDRplus 1.0 was changed as follows. Two amplification mixes contained all necessary PCR components, including Taq polymerase. Primers for rpoB, AC (amplification control), and inhA were redesigned, as was the sequence of one rpoB wild-type (WT) probe. The PCR protocol was modified, but the layout of the strips remained unchanged (Fig. 1).

Fig 1.

Example of a developed MTBDRplus 2.0 strip. Line 1, MDR strain (ΔrpoB WT8, rpoB MUT3, ΔkatG WT, katG MUT1, ΔinhA WT1, inhA MUT1); line 2, MDR strain (ΔrpoB WT3, ΔrpoB WT4, rpoB MUT1, ΔkatG WT, katG MUT1); line 3, MDR strain (ΔrpoB WT8, rpoB MUT3, ΔkatG WT, katG MUT1, ΔinhA WT1, inhA MUT1); line 4, MDR strain (ΔrpoB WT8, rpoB MUT3, ΔkatG WT, katG MUT1); and line 5, WT strain.

Briefly, 10 μl amplification mix A and 35 μl amplification mix B were mixed, and 5 μl of manually extracted or 10 μl of automatically extracted DNA was added. The amplification mixes are ready to use and consist of all necessary components, including polymerase. The PCR was performed according to the following protocol: 15 min at 95°C, where the activation of the polymerase takes place; 20 cycles of 30 s at 95°C and 2 min at 65°C; and 30 cycles of 25 s at 95°C, 40 s at 50°C, and 40 s at 70°C. The final cycle consisted of 8 min at 70°C.

Hybridization.

Hybridization was performed automatically using the GTBlot 48 device (Hain Lifescience). Hybridization and stringent washing buffers were preheated to 45°C. Twenty μl of denaturation buffer was mixed thoroughly in a plastic 48-well tray with 20 μl of amplified sample and incubated at room temperature for 5 min. Subsequently, 1 ml of hybridization buffer was added to each well and mixed. One prelabeled test strip was added into each well, and the wells were incubated for 30 min at 45°C. All solutions were completely aspirated following incubation. One ml stringent buffer was then added to each strip and incubated for 15 min at 45°C. After the aspiration of the solution, 1 ml rinse buffer was added to each strip and incubated at room temperature for 1 min. The following incubation steps took place at room temperature. The rinse buffer was completely removed, 1 ml of diluted conjugate buffer was added to each strip, and they were incubated for 30 min. After incubation, all solutions were removed and the test strips were rinsed twice with rinse buffer for 1 min, followed by a washing step with distilled water for 1 min. One ml substrate buffer was added to each strip, and they were incubated for 5 to 8 min. All solutions were removed, and the reaction was stopped by two rinses with distilled water. The test strips were dried and then taped to the MTBDRplus 2.0 assay worksheet for interpretation.

Interpretation.

Five controls are set on the DNA strip: the conjugate control (CC) area, the AC band, and the rpoB, katG, and inhA locus control zones. The CC area documents the efficiency of conjugate binding and substrate reaction and always has to be developed. If the AC band is developed, mistakes during extraction and amplification setup and the carryover of amplification inhibitors can be excluded. The rpoB, katG, and inhA locus control zones detect a region specific for the respective locus and must always stain positive when the TUB zone has indicated the presence of an M. tuberculosis complex strain.

Twenty-two probes for the characterization of the tested strain are located on the DNA strip. TUB represents an MTBC-specific region of the 23S rRNA gene. This zone hybridizes with amplicons generated from all MTBC members. Eight rpoB wild-type probes (WT1 to WT8) encompass the region of the rpoB gene coding for amino acids 505 to 534. Four other probes are specific for the most common mutations: D516V, H526Y, H526D, and S531L (rpo MUT1, MUT2, MUT3, and MUT4, respectively). Three probes are specific for the codon 315 region of katG. One is the wild-type probe (katG WT), while two others (katG MUT1 and MUT2) are designed for the AGC-to-ACC (S315T1) and the AGC-to-ACA (S315T2) mutations, respectively. Six probes are designed for the promoter region (−8, −15, and −16) of the inhA gene. Two wild-type probes, inhA WT1 and WT2, cover −15, −16, and −8 nucleic acid positions, respectively; four others (inhA MUT1, MUT2, MUT3A, and MUT3B) detect mutations of C15T, A16G, T8C, and T8A, respectively. The MTBDRplus 2.0 results were interpreted as described in the manufacturer's instructions.

Statistical analysis.

Statistical analysis was performed as described previously (1, 14).

RESULTS

Culture, which is the gold standard, is known to be a suboptimal reference standard (4). Therefore, a category named clinical positive, which is similar to the term CRS (for composite reference standard [17]), was included based on the patient's history of TB infection, on clinical symptoms, and on previously culture-confirmed TB. A patient is assessed as clinical positive when at least clinical symptoms, based on X-ray, etc., and a positive previous TB culture result are available. A patient is assessed as clinical negative when the patient shows TB symptoms but has had no previously confirmed TB infection and no actual positive TB culture result.

Diagnostic performance in pulmonary specimens.

MTBDRplus 2.0 was designed to detect and identify MTBC bacteria in smear-positive and -negative specimens and to determine RMP and INH resistance in a single test. To evaluate both, first the performance of MTBDRplus 2.0 was compared to that of conventional culture and clinical findings and secondly with conventional DST and respective clinical findings for RMP and INH.

For DNA isolation, two different methods were applied, the manual DNA extraction with the GenoLyse kit and the automated method using the GXT DNA/RNA extraction kit and the GenoXtract device. In total, 348 pulmonary specimens were processed.

From 166 samples extracted manually with the GenoLyse kit, 4 samples (2.4%) had to be excluded (3 had no DST results, and 1 had insufficient volume for extraction). Hence, 162 specimens were used for the performance evaluation of MTBDRplus 2.0 with GenoLyse. Of these isolates, 39 (24.1%) were smear and clinical positive, one (1.2%) was smear positive but clinical negative, 61 (37.7%) were smear negative and clinical positive, and 61 (37.7%) were smear and clinical negative (Table 1).

Table 1.

MTBDRplus 2.0 test results for detection of MTBC compared to culture and clinical data^a

DNA extraction method (n) and smear result (n)	MTBDRplus 2.0 result
	MTBC culture positive, clinical positive (n = 181)		MTBC culture negative, clinical positive (n = 21)		MTBC culture negative, clinical negative (n = 134)
	Pos	Neg	Pos	Neg	Pos	Neg
GenoLyse (162)
Pos (40)	38	0	1	0	0	1
Neg (122)	40	12	9	0	1	60
GenoXtract (174)
Pos (39)	39	0	0	0	0	0
Neg (135)	39	13	11	0	0	72
Combined (336)
Pos (79)	77	0	1	0	0	1
Neg (257)	79	25	20	0	1	132
Pos + Neg	156	25	21	0	1	133

^aPos, positive; Neg, negative.

In this group of 162 specimens processed with the GenoLyse kit, 78 of the 90 MTBC culture-positive specimens were also detected by MTBDRplus 2.0 (Table 1). Seventy-two specimens remained negative by culture, while among this group 11 specimens showed a positive MTBC result with MTBDRplus 2.0. Thus, the overall sensitivity and specificity for all included specimens with culture as the reference standard were estimated to be 86.7 and 84.7%, respectively. Positive (PPV) and negative (NPV) predictive values were 87.6 and 83.6%, respectively. Sensitivity and specificity for smear-positive and smear-negative samples were comparable to those of culture and clinical findings (Table 2).

Table 2.

MTBDRplus 2.0 detection of MTBC with GenoLyse and GenoXtract isolation methods compared to culture and clinical data^a

Extraction method and smear status	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)
GenoLyse
Pos + Neg	88.0	98.4	98.9	83.6
Neg	80.3	98.4	98.0	83.3
GenoXtract
Pos + Neg	87.3	100	100	84.7
Neg	79.4	100	100	84.7
Combined
Pos	87.6	99.2	99.4	84.1

^aPos, positive; Neg, negative.

From 182 samples automatically processed with the GenoXtract DNA extraction device, 6 samples (3.3%) had to be excluded: 4 had no DST result, 1 was of insufficient quality after NALC-NaOH decontamination (too viscous for pipetting), and 1 failed due to a GenoXtract handling problem. An additional two (1.1%) had to be excluded, since no clinical data were available. From 174 specimens used for the performance evaluation of MTBDRplus 2.0 in the GenoXtract DNA isolation group, 39 (22.4%) were smear and clinical positive, 63 (36.2%) were smear negative and clinical positive, and 72 (41.4%) were smear and clinical negative.

From these 174 GenoXtract isolates, 102 (58.6%) were clinical positive and 72 (41.4%) were clinical negative. Seventy-eight of the 91 culture-positive samples were also found to be MTBC positive with MTBDRplus 2.0 (Table 1). In the subgroup of 52 smear-negative and culture-positive samples, 39 were detected as MTBC positive with MTBDRplus 2.0. Eighty-three specimens remained negative by culture, while among this group 11 specimens showed a positive MTBC result with MTBDRplus 2.0. All 11 could be classified as clinical positive by former patient data.

Results similar to those for GenoLyse regarding overall sensitivity and specificity (with culture and clinical data as the reference standards) were calculated for the GenoXtract isolation method and combined (Table 2). MTBDRplus 2.0 results with both extraction methods were similar.

To assess MTBDRplus 2.0 as a tool to diagnose TB, the results were correlated with clinical data (Table 3). The combined overall sensitivity, specificity, PPV, and NPV of smear-positive and smear-negative samples for both extraction methods compared to clinical findings were 87.6, 99.2, 99.4, and 84.1%, respectively.

Table 3.

MTBDRplus 2.0 test results for detection of MTBC compared to clinical data with combined GenoLyse and GenoXtract results^a

Clinical data by smear type	MTBDRplus 2.0 result for TB status
	Pos	Neg
Pos
TB patient	78	0
Non-TB patient	0	1
Neg
TB patient	99	25
Non-TB patient	1	132

^aPos, positive; Neg, negative.

Detection of rifampin resistance.

For 156 of 336 samples tested with MTBDRplus 2.0, phenotypic DST results also were available (Table 4). From all 106 RMP-resistant strains, 100 could be correctly detected by MTBDRplus 2.0, as well as 48 out of 50 susceptible strains, resulting in a sensitivity and specificity of 94.3 and 96.0%, respectively. Only one sample (0.94%) was RMP monoresistant; it was detected because it was missing WT band 8.

Table 4.

MTBDRplus 2.0 test results for detection of RMP and INH resistance compared to DST applied to all isolates having phenotypic and genotypic results

MTBDRplus 2.0 result (n = 156)	DST result
	RMPr	RMPs	INHr	INHs
RMPr	100 (51/49)	2 (1/1)
RMPs	6 (1/5)	48 (24/24)
INHr			115 (57/58)	4 (1/3)
INHs			5 (1/4)	32 (18/14)

RMP^r, resistant to rifampin; RMP^s, susceptible to rifampin; INH^r, resistant to isoniazid; INH^s, susceptible to isoniazid. Data in parentheses are the smear status (smear-positive/smear-negative specimens) of samples on which the MTBDRplus 2.0 was performed.

Detection of isoniazid resistance.

Similarly to rifampin, results for 156 isoniazid resistance cases were available (Table 4). From all 120 INH-resistant strains, 110 could be correctly detected by MTBDRplus 2.0, as could be 32 out of 36 susceptible strains, resulting in sensitivity and specificity of 95.8 and 88.9%, respectively. Twenty-two samples (18.33%) were INH monoresistant.

In comparisons between smear-negative and smear-positive specimens, results of MTBDRplus 2.0 for the sensitivity of RMP and INH resistance detection do not differ significantly (Table 5).

Table 5.

MTBDRplus 2.0 detection of RMP and INH resistances regarding smear status compared to culture and clinical data

Drug and smear statusa	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)
RMP
Pos	98.1	96.0	98.1	96.0
Neg	90.7	96.0	98.0	82.7
INH
Pos	89.3	94.7	98.2	94.7
Neg	93.5	82.3	95.1	77.7

^aPos, positive; Neg, negative.

Prevalence of mutations.

The prevalence of mutations was calculated on the basis of all MTBDRplus 2.0 results (n = 177), including culture-negative, MTBDRplus 2.0-positive cases (n = 21). In total, 114 strains were RMP resistant by MTBDRplus 2.0 and carried a mutation in rpoB; 135 strains carrying a total of 201 mutations (66 isolates with 2 mutations; 48.9%) were INH resistant by MTBDRplus 2.0.

The most prevalent mutation in rpoB resulting in an amino acid exchange was S531L, found in 86.8% (n = 99) of cases, followed by mutations in codons 530 to 533 (n = 7; 6.1%), D516V (n = 5; 4.4%), H526Y (n = 1; 0.9%), mutations in codon 510 to 513, and a deletion. The most prevalent mutation in katG resulting in an amino acid exchange was S315T in 128 out of 135 strains (94.8%), 1 strain had a katG deletion, 71 (52.6%) had an inhA C15T exchange, and 1 had an inhA T8C exchange.

DISCUSSION

The first version of MTBDRplus was validated for smear-positive pulmonary samples, and it could be proven that both the detection of MTBC and of RMP and INH resistance was high (sensitivity of 98.1% for the detection of RMP resistance and 84.3% for the detection of INH resistance [10]).

However, the usefulness of smear-negative samples with this test was limited. The new MTBDRplus 2.0 test is dedicated to more convenience, with a master mix that is ready to use after combining two components provided with the kit, and to a higher analytical sensitivity, which can be assumed by the redesigned amplification program. The DNA purification methods recommended by the manufacturer may, in principle, also be applied to extrapulmonary specimens (this remains to be demonstrated). We concentrated our work on pulmonary samples, as these are by far the most important kind of specimen in our laboratory.

In this study, the new, redesigned MTBDRplus 2.0 was evaluated for the first time. In principle, the functionality of the new MTBDRplus 2.0 test was demonstrated independently of the DNA extraction method (sensitivity of 88.0 and 87.3% and specificity of 98.4 and 100% with GenoLyse and GenoXtract methods, respectively).

To determine the performance of the test, a special focus was put on smear-negative pulmonary specimens. Therefore, the majority of tested specimens were smear negative (73.9%; 257 out of 348). The new MTBDRplus 2.0 test showed a high sensitivity (76.0%; combined data from both extraction methods, with culture as the reference standard) to detect MTBC in smear-negative pulmonary specimens with a single test. This is high compared to the sensitivity of the Cepheid GenXpert system (72.5%) as described by Boehme et al. (2). The overall specificity of 99.3% for the detection of M. tuberculosis in sputum samples of non-TB patients also is comparable to that of the Cepheid system (99.2%).

However, both studies (2 and this study) are not directly comparable, since Boehme et al. excluded patients under treatment from the final analysis, whereas in this study all samples were included regardless of their treatment status, reflecting the routine laboratory situation.

When the data for MTBDRplus 2.0 were correlated with the clinical data, the sensitivity increases to 79.8% for smear-negative specimens. However, for the interpretation of these study results, it must be considered that patients under treatment are included that harbor DNA of dead bacteria that could be detected by PCR/LPA, as is known for any nucleic acid-based detection method, but their samples could be microscopic and culture negative at the same time. Looking into the clinical data could clarify the amount of false-positive results. Of the 22 false positives, 18 patients were already under treatment. Only 4 samples out of 182 were not in agreement with clinical data (2.2%). Nevertheless, the detection of TB-DNA from dead bacteria led to a higher rate of false-positive results than did culture results, and this reduces the specificity value. Patients of this cohort were monitored after 1, 2, 3, and 5 months of treatment with conventional DST, and the treatment was changed according to the results. As the aim of this study was to determine the suitability of this test for use in a routine laboratory, we did not exclude these patients under treatment.

In various other studies, it could be shown that LPAs achieved a high sensitivity and specificity for RMP and INH resistance (10, 11). The results in this study differ slightly, with a sensitivity and specificity for RMP of 94.3 and 96% and for INH of 95.8 and 88.9%, respectively. The sensitivity for the detection of RMP resistance is slightly lower than that published in the meta-analysis (10) and is higher than that for INH. Despite the fact that only a small number of resistant strains were investigated, clustering may be a reason for the deviation from the mean values. Some cases of resistance also may be caused by mutations not detectable with MTBDRplus 2.0, e.g., in other regions of rpoB, katG, or unknown genes (5, 12).

Another explanation for these false resistance results could be the existence of mutations resulting in low-level RMP resistance (18). In fact, the respective patterns showed no mutation band occurring on the strip, but wild-type band 8 was missing. Sequencing data could identify a mutation in codon 533 leading to the amino acid exchange L533P. This mutation is known to have a doubtful level of resistance (13).

A second infection with a nosocomial strain during the stay in the hospital could be a further explanation for discrepant results. The phenomenon of the nosocomial transmission of MDR/XDR-TB between patients being determined by the investigation of M. tuberculosis DNA diversity using DNA fingerprinting in TB hospitals from Moldova was demonstrated in earlier studies (V. Crudu et al., presented at the Annual Congress of the European Respiratory Society, Barcelona, Spain, 18 to 22 September 2010). No further analysis of discrepant results was undertaken, since this was not the focus of this study.

The main aim of this study was to show the functionality of the new test with routine specimens in a country with high TB prevalence and a high burden of TB drug resistance, and we also tested the capability of the assay to detect MTBC in smear-negative specimens. The need for the rapid and reliable identification of TB in smear-negative samples, and especially in HIV-infected patients, seems still to be a challenge. For example, a sensitivity of only 47% for the GenXpert system to detect MTBC from smear-negative, culture-positive samples from HIV-infected patients was described (15). Even though new technologies claim to have a high sensitivity for smear-negative samples, further efforts are necessary to overcome these limitations.

It is worth noting that 18.3% of the INH-resistant strains are INH monoresistant. This reflects the importance of a test being able to discriminate between MDR and INH monoresistance.

One approach could be to apply more than one specimen to a PCR-based analysis. Boehme et al. (2) clearly demonstrated an increase in sensitivity when up to 3 specimens were analyzed. This would give great improvement to TB diagnosis when the speed of nucleic acid methods and the highest sensitivity and specificity of liquid culture diagnostic could be matched. Therefore, automation and a price level similar to those of liquid culture systems are needed. This new version could be a first step in this direction.