Background

Despite the recent advances in management, tuberculosis (TB) remains a leading cause of mortality and morbidity worldwide. The emergence of multidrug-resistant tuberculosis (MDRTB) further constitutes a serious threat to the control of TB. In Taiwan, the incidence of TB was 63.2/100,000 population in 2007 and 62.0/100,000 population in 2008 (1). The Taiwan drug resistance surveillance program revealed that the combined drug resistance rates in 2005 were 10.1% for isoniazid (INH), 6.2% for rifampicin (RIF), and 4.0% for MDRTB (2). Among the various genotypes of Mycobacterium tuberculosis (MTB), the Beijing genotype is the dominant strain in Taiwan (3, 4).

Pyrazinamide (PZA) is an important first-line anti-tuberculosis (anti-TB) drug that is used in short-course chemotherapy and is one of the cornerstone drugs in the treatment of MDRTB (5). Through pyrazinamidase (PZase) which is constitutively expressed in Mycobacterium tuberculosis (MTB), PZA is hydrolyzed to toxic pyrazinoic acid (POA) intracellularly (6, 7). Although the exact mechanism is unknown, it has been postulated that POA exerts an inhibitory effect on cellular metabolism in acidic conditions (8-10). The PZase enzyme is encoded by the pncA gene (11). Mutations in the pncA gene may cause a reduction in PZase activity which may be the major mechanism of PZA resistance in MTB (12, 13). Previous reports have well characterized the mutations of the pncA gene in PZA-resistant MTB isolates, however the correlation varies between different geographical areas (14-17). A great diversity of pncA gene mutations has also been described, including missense mutations, one or more base insertions or deletions, and complete deletion (12, 13, 18). Meanwhile, a substantial proportion of resistant isolates have been found to have intact PZase activity without pncA gene mutations, suggesting an alternative mechanism of PZA resistance.

Due to the inhibitory effect of low pH on in vitro growth of MTB, conventional PZA drug susceptibility testing on a solid medium is of limited value and not routinely done in many areas (19). Considering the unique bactericidal effect of PZA among first-line anti-TB drugs, it is important to identify PZA resistance in clinical practice, especially in dealing with MDRTB cases. The purpose of the present study was to identify the characteristics of pncA mutations in clinical MDRTB isolates, and evaluate the effectiveness of pncA gene analysis in identifying PZA-susceptible isolates. The impact of genotyping in PZA resistance was also investigated.

Methods

Mycobacterial isolates

This study was conducted in six hospitals in Taiwan, including five referral medical centers and one regional hospital that specialize in pulmonary diseases. MDRTB clinical isolates were collected from newly diagnosed tuberculosis patients from January 2007 to December 2009. The demographic profiles and clinical characteristics of the patients were obtained from medical records, and the chest radiograms were interpreted by the in-charge doctors of each hospital. The institutional review boards of all six hospitals approved the study and informed consent was obtained from each patient before enrollment.

Drug susceptibility testing (DST) and PZase assay

The drug susceptibilities of isoniazid and rifampicin were performed by the proportion method and their critical concentrations for resistance were as follows: isoniazid, 0.2 μg/ml and 1.0 μg/ml; rifampicin, 1.0 μg/ml and 5.0 μg/ml (20). MDRTB isolates were defined as those resistant to low drug concentration levels in both isoniazid and rifampicin. The susceptibility of PZA was assessed by the non-radiometric BACTEC Mycobacteria Growth Indicator Tube (MGIT) 960 method (BD Biosciences, Sparks, MD, USA). The recommended critical concentration of 100 μg/ml PZA was used to discriminate between PZA-susceptible and PZA-resistant isolates. The PZase activity test was performed according to Wayne's procedure (21).

DNA extraction, pncAgene sequencing

Genomic DNA was extracted from Middlebrook 7H11 cultures as described previously (22). Entire pncA genes and 82 bp of an upstream putative regulatory sequence were amplified by polymerase chain reaction (PCR) with forward (P1) and reverse (P6) primers as described previously (23). Briefly, the PCR mixture (100 μL) contained ~1 ng of DNA template, final concentrations of 1.0 μM of each set of primers, 200 μM deoxynucleoside triphosphate (dATP, dCTP, dGTP and dTTP; Pharmacia, Uppsala, Sweden) and 10 μL Taq buffer and 5.0 U/μL Taq polymerase (Gibco-Bethesda Research Laboratories). Amplification was performed for 40 cycles (1 min at 94°C, 1 min at 56°C and 1 min at 72°C) by a GeneAmp PCR system 9700 thermocycler (Perkin Elmer, Applied Biosystems, CA). To determine the pncA sequence, PCR products were purified with a QIAquick PCR purification kit (Qiagen GmbH, Hilden, Germany). The PCR products were directly sequenced using an ABI 377 automatic DNA sequencer (Perkin Elmer, Applied Biosystems, CA).

Genotyping

All clinical isolates were genotyped by a commercial spoligotyping kit (Isogen Bioscience B.V., Maarssen, Netherlands). The "Beijing strain" was defined as deletion from spacer 1 to spacer 34 in the direct repeat region and the presence of (at least 3) spacers 35-43.

Data analysis

Comparisons of demographic and clinical characteristics were done using the chi-square test or Fisher's exact test for categorical variables, and the two-tailed independent t test for continuous variables. Binary logistic regression analysis was performed to determine the independent variables, and odds ratios with their 95% confidence intervals were presented. The sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of the pncA gene mutations and PZase activity test were calculated according to the PZA susceptibility results. Significance was defined as p < 0.05 (two-tailed). Statistical analysis was performed using a statistical software package (SPSS version 17.0, SPSS Inc., Chicago, IL, USA).

Results

Patient characteristics

During the study period, a total of 66 MDRTB isolates were collected from newly diagnosed tuberculosis patients in six hospitals in Taiwan. Among these MDRTB isolates, 36 (54.5%) were resistant to PZA and 30 (45.5%) were susceptible to PZA according to the DST results assessed by the MGIT 960 method. The demographic data of these patients are shown in Table 1. MDRTB isolates with concomitant PZA resistance were more likely to be sputum smear negative at the diagnosis of tuberculosis (52.8% vs. 26.7%, p = 0.032), and more likely to be associated with ethambutol (58.3% vs. 33.3%, p = 0.043) and streptomycin resistance (66.7% vs. 36.7%, p = 0.015). Other demographic profiles, including age, gender, previous anti-TB treatment, smear positivity, and the presence of cavities in radiographs, were comparable between patients infected with PZA-resistant or susceptible MDRTB isolates. Of these MDRTB isolates, 37 (56.1%) were Beijing and 29 (43.9%) non-Beijing strains. The proportion of PZA-resistant isolates was similar between the Beijing and non-Beijing strains.

Table 1 Demographic data of the multidrug-resistant tuberculosis patients with or without pyrazinamide (PZA) resistancea
Full size table

PZA susceptibility results and mutations in the pncAgene

The mutations in the pncA gene and the correlation with PZA susceptibility testing are shown in Table 2. Of 66 clinical MDRTB isolates, 36 were wild type and the other 30 were associated with at least one pncA gene mutation. Among these 30 isolates, we found 36 unique mutations out of a total of 39 different mutations. The majority of these were point mutations that resulted in nucleotide substitutions (37/39, 94.9%). There was one deletion (1/39, 2.6%) and one insertion (1/39, 2.6%). In addition, three isolates were identified to possess more than one site of pncA mutation in a single isolate, and all of these isolates were resistant to PZA. The mutations were scattered along the pncA gene and no highly conservative region could be identified. Seven isolates with the wild-type pncA gene were found to be resistant to PZA. Meanwhile, one isolate which was susceptible to PZA had a pncA gene mutation (A98C, Asp(33)→Ala).

Table 2 pncA gene mutation analysis, PZase activity test, and spligotyping of the MDRTB isolates
Full size table

Genotyping and PZA susceptibility results

The comparison of PZA susceptibility testing and pncA mutations and the impact of genotyping are shown in Table 3. Among 36 isolates with PZA resistance, 29 were documented with pncA gene mutations. Among 30 isolates which were susceptible to PZA, 29 were associated with the wild type pncA gene. Overall, the sensitivity and specificity of pncA gene mutations in predicting PZA resistance were 80.6% and 96.7%, respectively. In isolates belonging to Beijing genotypes (37 isolates), the sensitivity and specificity were 76.2% and 93.8%, respectively. In non-Beijing strains (29 isolates), the sensitivity and specificity were 86.7% and 100%, respectively.

Table 3 Concordance of pncA gene analysis and PZA susceptibility testing by the BACTEC MGIT 960 method
Full size table

The concordance of PZA susceptibility testing and PZase activity and the impact of genotyping are shown in Table 4. Among 36 PZA-resistant isolates, 21 had no detectable PZase activity as judged by Wayne's method. In 30 PZA-sensitive isolates, 28 had positive PZase activity. The overall sensitivity and specificity of PZase activity testing in predicting PZA resistance were 58.3% and 93.3%, respectively. In 37 Beijing strain isolates, the sensitivity and specificity were 66.7% and 93.8%, respectively. In 29 non-Beijing strain isolates, the sensitivity and specificity were 46.7% and 92.9%, respectively.

Table 4 Concordance of PZase activity and PZA susceptibility testing by the BACTEC MGIT 960 method
Full size table

Discussion

PZA is an important and effective first-line anti-TB drug both in MDRTB and fully susceptible isolates. However, traditional susceptibility testing of PZA is of limited value and not routinely performed. Detection of pncA gene mutations may provide rapid and reliable information about PZA susceptibility profiles. In our analysis, more than half of the MDRTB isolates were also resistant to PZA. No clinical characteristics have a good correlation with PZA resistance, except for concomitant resistance to other first-line anti-TB agents. The analysis of pncA gene mutations is a test of high specificity and moderate sensitivity in predicting PZA resistance. The correlation between pncA gene mutations and PZA resistance is slightly better in non-Beijing strains as compared with Beijing strains. The present study suggests that pncA gene analysis is useful in identifying PZA susceptibility profiles in MDRTB isolates, especially in non-Beijing strains. However, the presence of wild type pncA genes in PZA-resistant isolates also suggests that mechanisms other than pncA gene mutations leading to PZA resistance may exist, and this deserves further exploration.

Classified as a first-line oral anti-TB agent, PZA is widely used in the intensive phase of anti-TB treatment and plays a pivotal role in the treatment regimen of MDRTB and extensively drug-resistant tuberculosis (XDRTB) disease. Unfortunately, concomitant resistance to other first-line anti-TB drugs, including PZA, is not uncommon in MDRTB and XDRTB isolates. Due to difficulties in performing PZA susceptibility tests and the unreliable results of the traditional method, information about PZA resistance in MDRTB isolates is not routinely obtained in clinical settings. Recent studies from South Africa and Thailand reported a PZA resistance rate of around 50% in MDRTB isolates (24, 25). In line with these reports, the PZA resistance rate was 54.5% in the present study by the BACTEC MGIT 960 method. The underlying mechanism leading to the high PZA resistance rate in MDRTB isolates remains to be identified. However, an increased incidence of previous PZA exposure in these isolates is a possible cause (26). Assuming that approximately half of MDRTB isolates are resistant to PZA, then nearly half of them are PZA-susceptible. If we can identify the PZA-susceptible MDRTB isolates timely and accurately, it would be possible to add PZA to the combination of anti-TB drugs for MDRTB patients.

The early identification of drug susceptibility by drug resistance gene analysis has been widely used for first-line anti-TB drugs in recent years, especially rifampicin and isoniazid (27). The PZase enzyme, encoded by the pncA gene, plays a vital role in the bactericidal effect of PZA against MTB isolates (11-13). Theoretically, pncA gene analysis should enable clinicians to identify PZA resistance early. However, the correlation between pncA gene mutations and PZA susceptibility varies among previous studies (14-17). As compared with the BACTEC MGIT 960 method, the overall sensitivity and specificity of pncA gene analysis in the present study were 80.6% and 96.7%, respectively, which is comparable with previous reports (17, 24). The high specificity suggests that pncA gene analysis is an ideal method to identify PZA-resistant MDRTB isolates. Meanwhile, the relatively lower sensitivity means that false negative results are possible and implies that mechanisms other than pncA mutations are involved in PZA resistance. Although pncA gene analysis is a rapid test with a high positive predictive value, it cannot completely replace the phenotypic susceptibility testing of PZA.

We found a high diversity of pncA gene mutations without major hot spots in the MDRTB isolates. Among 36 PZA-resistant MDRTB isolates, 38 different mutations scattered along the whole pncA gene were identified out of 29 isolates. Meanwhile, one mutation at nucleotide 98(A→C) was found in one isolate among 30 PZA-susceptible MDRTB isolates. No clustered isolates were found to share identical pncA gene mutations and spoligotyping in our analysis. Our findings of scattered mutations support similar observations in previous studies (16, 24). The high diversity of pncA gene mutations could also be used as a supplement to current genotyping methods in epidemiological investigations.

In the present study, we also evaluated PZase activities by Wayne's test, which is based on the detection of POA. The specificity of PZase activity in predicting PZA susceptibility was comparable with that of pncA gene analysis, but the sensitivity was much lower. Moreover, a significant proportion of isolates with pncA gene mutations possessed detectable PZase activity (14/30, 46.7%). According to previous reports, the sensitivity of the PZase assay ranged from 79% to 96% and the specificity was generally above 95% (28, 29). The cause of the remarkable discordance between the PZase assay test, pncA gene analysis, and PZA susceptibility in the present study could not be readily clarified in the study design. The geographic differences in MDRTB isolates may have contributed to the differences between our findings and previous reports. It is possible that some pncA gene mutations may lead to a reduction, but not loss, of PZase activity that can still be detected by Wayne's test. Although the PZase activity is not totally lost in these isolates, it results in phenotypical resistance to PZA. Unlike previous studies that enrolled both MDRTB and non-MDRTB isolates, only MDRTB isolates were included for analysis in the present study. The different composition of MTB isolates, including various genotypes and drug susceptibility profiles, may affect the predictive value of PZase activity testing and causes of the discordant results.

The difference of PZA susceptibility and pncA gene mutations among various genotypes of MTB isolates has rarely been evaluated before. The Beijing strain is the dominant strain in Eastern Europe and Southeastern Asia, including Taiwan (30). Differences in clinical presentations and treatment outcomes have been reported, although with some controversy among various geographical areas (3, 31). In the present study, we compared the correlation of pncA gene analysis, PZase activity, and PZA susceptibility between Beijing and non-Beijing strains. We demonstrated that the Beijing strain had a slightly lower sensitivity in pncA gene analysis, but much higher sensitivity in PZase activity test, as compared with non-Beijing strains. By comparison, the specificity was similar between Beijing and non-Beijing strains in both pncA gene analysis and PZase activity tests. No specific mutation was found in Beijing or non-Beijing genotype isolates. Our data demonstrated different genetic and phenotypic characteristics of Beijing and non-Beijing strains in PZA resistance. The differences between Beijing and non-Beijing strains in PZA susceptibility warrant further verification with a larger sample size, and the underlying mechanism remains to be clarified.

In the present study, we found that MDRTB patient with concomitant PZA resistance were more likely to be sputum smear negative. The cause of significant correlation between PZA susceptibility and sputum smear results cannot be easily identified in the study design. A recent study showed that exposure of lowly active MTB bacilli to anti-TB agents led to the emergence of resistant mutants and it required higher drug concentrations to eliminate metabolic inactive mycobacteria (32). Therefore, the MDRTB isolates with PZA resistance are probably less metabolic active which may lead to the lower sputum smear positive rate. Further in-vivo and in-vitro studies will be needed to verify the issue and elucidate the underlying mechanism.

Our study has several limitations. Only MDRTB isolates were included for analysis and the predictive value of pncA gene analysis in non-MDRTB isolates could not be evaluated in the present study. We did not evaluate the mutation of other important drug resistant gene, such as rpoB, inhA, and katG, in these isolates. The relatively small case numbers may also lead to a lower power of our analysis. More importantly, this study was performed in a TB endemic area with a predominance of the Beijing strain. It remains to be determined if our findings can be applied to areas with a lower TB incidence or without predominance of the Beijing strain.

Conclusions

Our study demonstrated that the PZA resistant rate in MDRTB isolates was high, and the majority of the PZA-susceptible isolates presented with wild type pncA genes. Analysis of pncA genes is a test of high specificity in predicting PZA resistance, enabling clinicians to add PZA timely and accurately to treatment regimens for MDRTB patients. The concordance between pncA gene analysis and PZA resistance was better in non-Beijing as compared with Beijing strains, suggesting that pncA gene analysis is more suitable for MDRTB isolates belonging to non-Beijing strains.