Fast test for assessing the susceptibility of Mycobacterium tuberculosis to isoniazid and rifampin by real-time PcR

Mycobacterium tuberculosis is the bacterium that causes tuberculosis (TB), a leading cause of death from infectious disease worldwide. In 2008, approximately two million deaths and nine million new cases of TB were reported and 5% of these were resistant to anti-TB drugs (WHO 2009a, b). In Brazil, TB was the fourth leading cause of death from infectious disease in 2008 and the first for acquired immune deficiency syndrome patients (MS/ SVS 2010), with an incidence of 48 per 100,000 inhabitants (SESSP/SVE 2006, MS/SVS 2009). From the 11,900 isolates processed in our laboratory during the last three years (2007-2009), 15% exhibited resistance to at least one drug and 5.5% were multidrug-resistant (MDR). Rifampin (RIF) resistance is an excellent marker for MDR TB. Approximately 95% of all RIF-resistant strains contain mutations localised in an 81-bp core region of the rpoB gene (Telenti et al. 1993, Riska et al. 2000) and all mutations that occur in that region result in RIF resistance (Telenti et al. 1993, El-Hajj et al. 2001). In contrast to RIF, 75% of isoniazid (INH)-resistant strains have mutations either in codon 315 of the katG gene or in the inhA ribosomal binding site accompanied by mutations in the ahpC gene and/or the oxyR intergenic region (Piatek et al. 2000, Rossetti et al. 2002). In Brazil, similar percentages of mutations in the rpoB and katG genes have been described (Valim et al. 2000, Rossetti et al. 2002). Rapid diagnosis of resistant strains is important for the control of TB. The isolation and subsequent susceptibility testing of M. tuberculosis remain the gold standard, but this process takes many weeks to complete (Wada et al. 2004, MS/SVS/DVE 2008). The real-time polymerase chain reaction (RT-PCR) assay is faster and may detect all mutations that occur in the M. tuberculosis rpoB 81-bp core region, the katG 315 codon and the inhA ribosomal binding site (Piatek et al. 2000, El-Hajj et al. 2001, Marín et al. 2004, Ruiz et al. 2004, Wada et al. 2004, Yesilkaya et al. 2006, Boehme et al. 2010). In this study, we investigated an RT-PCR assay based on molecular beacon probes to determine susceptibility and/or resistance of M. tuberculosis isolates to INH and RIF in comparison with the phenotypic gold standard method that, in our study, was considered to be the automatised BACTEC MGIT 960 susceptibility assay (Becton Dickinson. Available from: finddiagnostics.org/export/sites/default/resource-entre/ find_documentation/pdfs/ mgit_manual_nov_2007.pdf). The goal was to assess whether RT-PCR is able to produce sensitive, specific and faster results than the traditional method for routine laboratory drug-susceptibility testing.

Mycobacterium tuberculosis is the bacterium that causes tuberculosis (TB), a leading cause of death from infectious disease worldwide.In 2008, approximately two million deaths and nine million new cases of TB were reported and 5% of these were resistant to anti-TB drugs (WHO 2009a, b).In Brazil, TB was the fourth leading cause of death from infectious disease in 2008 and the first for acquired immune deficiency syndrome patients (MS/ SVS 2010), with an incidence of 48 per 100,000 inhabitants (SESSP/SVE 2006, MS/SVS 2009).From the 11,900 isolates processed in our laboratory during the last three years (2007)(2008)(2009), 15% exhibited resistance to at least one drug and 5.5% were multidrug-resistant (MDR).
Rifampin (RIF) resistance is an excellent marker for MDR TB.Approximately 95% of all RIF-resistant strains contain mutations localised in an 81-bp core region of the rpoB gene (Telenti et al. 1993, Riska et al. 2000) and all mutations that occur in that region result in RIF resistance (Telenti et al. 1993, El-Hajj et al. 2001).In contrast to RIF, 75% of isoniazid (INH)-resistant strains have mutations either in codon 315 of the katG gene or in the inhA ribosomal binding site accompanied by mutations in the ahpC gene and/or the oxyR intergenic region (Piatek et al. 2000, Rossetti et al. 2002).In Brazil, similar percent-ages of mutations in the rpoB and katG genes have been described (Valim et al. 2000, Rossetti et al. 2002).
Rapid diagnosis of resistant strains is important for the control of TB.The isolation and subsequent susceptibility testing of M. tuberculosis remain the gold standard, but this process takes many weeks to complete (Wada et al. 2004, MS/SVS/DVE 2008).The real-time polymerase chain reaction (RT-PCR) assay is faster and may detect all mutations that occur in the M. tuberculosis rpoB 81-bp core region, the katG 315 codon and the inhA ribosomal binding site (Piatek et al. 2000, El-Hajj et al. 2001, Marín et al. 2004, Ruiz et al. 2004, Wada et al. 2004, Yesilkaya et al. 2006, Boehme et al. 2010).In this study, we investigated an RT-PCR assay based on molecular beacon probes to determine susceptibility and/or resistance of M. tuberculosis isolates to INH and RIF in comparison with the phenotypic gold standard method that, in our study, was considered to be the automatised BACTEC MGIT 960 susceptibility assay (Becton Dickinson.Available from: finddiagnostics.org/export/sites/default/resource-entre/find_documentation/pdfs/ mgit_manual_nov_2007.pdf).The goal was to assess whether RT-PCR is able to produce sensitive, specific and faster results than the traditional method for routine laboratory drug-susceptibility testing. 1 mL of H 2 O.The suspension was boiled for 20 min and frozen at -70ºC overnight.The thawed suspension was than centrifuged for 10 min at 10,000 g and the supernatant was diluted 1/10 in H 2 O.The lysate suspensions were kept at -20ºC until used.

M. tuberculosis isolates -
Automatised BD Bactec™ MGIT™ 960 test -We used the BACTEC™ MGIT™ 960 system (Becton, Dickson and Company, Maryland, USA) as the gold standard method for phenotypic drug susceptibility testing, following the manufacturer's instructions.The INH and RIFsusceptible M. tuberculosis strain H37Rv (ATCC 27294) was used as a wild type control for susceptibility testing, for RT-PCR assays and for rpoB, katG and inhA DNA sequencing analysis.
PCR amplification and DNA sequencing -The amplification conditions and nucleotide sequences of the primers used to amplify the 555-bp segment of the katG gene containing codons 313-318 (GenBank access X68081 -katG 315), a 248-bp segment of the inhA gene (GenBank access Z79701) and a 189-bp segment of the 81-bp core of the rpoB gene (GenBank access L27989) were previously described (Telenti et al. 1997, Piatek et al. 1998, 2000, El-Hajj et al. 2001, van Doorn et al. 2003, Wada et al. 2004) (Table I).DNA amplification was verified on 2% agarose gels.The amplified PCR products were purified using a PureLink™ PCR Purification Kit (Invitrogen Corporation, Carlsbad, CA, USA).DNA sequencing was performed using the Big Dye Terminator Cycle Sequencing Kit (Applied Biosystems, Foster City, CA, USA) with an ABI Prism 3130xl DNA Sequencer (Applied Biosystems).The selection of M. tuberculosis isolates for DNA amplification and sequencing was based on the phenotypic susceptibility testing results for INH and RIF.All M. tuberculosis isolates that were phenotypically resistant to INH (120) or RIF (74) were submitted to gene sequencing for katG and inhA or rpoB, respectively (Supplementary data).For comparison purposes, additional numbers of M. tuberculosis INH and RIF-sensitive isolates were also submitted for katG and inhA and rpoB gene sequencing, approximately 10% of the 868 (n = 90) and 914 (n = 95), respectively (Supplementary data).
RT-PCR -All 988 M. tuberculosis isolates were tested for katG, inhA and rpoB using single-target RT-PCR assays.The sensitivity and specificity of the RT-PCR assays were calculated using the phenotypic resistance test results as the gold standard.To test for INH resistance, two sets of primers and probes were used to hybridise to two different amplicon targets (Table I).One molecular beacon probe was complementary to codons 313-318 of the katG gene and the other was complementary to the inhA ribosomal binding site.To test for RIF resistance, one set of primers and five probes (SW143, SW89, SW111, SW112 and SW182) were used to hybridise to a single rpoB amplicon such that their probe sequences spanned the entire core region, with overlapping sequences of one to three nucleotides (Table I).The RT-PCR probes were molecular beacons designated to hybridise to targets that did not contain mutations associated with antibiotic resistance.Drug resistance was indicated by the absence of a characteristic increasing fluorescent signal during PCR amplification.All assays were performed in duplicate and in sealed wells on a 96-well microtitre plate (Roche Diagnostics, Indianapolis, Ind).Reactions were performed in a Roche LightCycler 480 II (Roche Diagnostics, Indianapolis, Ind).Each reaction mixture contained 5 µL of lysed suspension diluted 1/10 in water, 2 µL of each primer, 2 µL of probe and 12.5 µL of 2X TaqMan ® Universal master mix (Applied Biosystems).PCR-certified water (Roche Diagnostics, Indianapolis, Ind) was added to bring the reaction volume to 25 µL.PCR mixtures were first incubated for 2 min at 50ºC and 10 min at 95ºC and then for 45 cycles of 30 s at 95ºC, annealing at 1 min at 50ºC (55ºC for the inhA reaction) and 30 s at 60ºC.
All reactions for RIF testing were initially performed as a single-target RT-PCR assay with one set of primers and one FAM-labelled probe per reaction tube for all 988 isolates.After all probes had been individually tested, all phenotypically RIF-resistant and the same ~10% additional RIF-sensitive isolates mentioned above were reevaluated by dual-target RT-PCR assay.We prepared two sets of dual-target RT-PCR assays: (i) the HEX-labelled probe SW182 plus the FAM-labelled probe SW112 and (ii) the HEX-labelled probe SW89 plus the FAM-labelled probe SW111.Each combination was used in a different RT-PCR assay.Probe SW143 was not used for dualtarget assays because we did not identify any mutation in that region based on a single-target RT-PCR assay or by DNA sequencing.
C T values from 18-39 were considered positive, meaning that no mutation was present on the gene fragment being tested, therefore indicating susceptibility.C T values equal to zero or equal to or greater than 40 were considered negative, indicating the presence of a drug resistance mutation.C T values between 1-17 were not found among the tested isolates.
Ethics -This study was approved by the ethical review board of the São Paulo University (CONEP/265/08).
RIF -Of the 988 isolates analysed, 92.4% (913/988) were sensitive and 7.6% (75/988) were resistant as determined by RT-PCR (Supplementary data).The overall sensitivity of the RT-PCR assay in detecting RIF resistance using all five different probes was 99% (73/74) (95% CI: 93, 100) when compared to the phenotypic assay.One of the RIF phenotypically resistant isolates was RIF-sensitive by RT-PCR assay.The nucleotide sequence analysis of this isolate revealed that a mutation at codon 522 (TCG-TTG), covered by probe SW111, was not detected by RT-PCR assay.We found a discrepancy between the phenotypic and genotypic results (RT-PCR and DNA sequencing) for RIF resistance for two M. tuberculosis isolates: one at codon 516 (GAC-TAC) and the other at codon 526 (CAC-CTC), using probes SW89 and SW112, respectively.No discrepant results were obtained between duplicates.

Dual-target RT-PCR assay -
The combination of two probes for each of the dual-target RT-PCR assays for RIF was based on the prevalence of mutations at the 81-bp core region of the rpoB gene.We excluded probe SW143 from both combinations because there were no mutations present on the DNA region complementary to probe SW143.All 74 phenotypically RIF-resistant isolates and 95 (10%) of phenotypically RIF-sensitive isolates were used to evaluate the two dual-target RT-PCR assays (SW182 + SW112 and SW89 + SW111).There was 100% correlation between the results of single and dual-target RT-PCR assays.No discrepant results were obtained between duplicates.

DIScuSSION
Because of the slow growth of M. tuberculosis, long delays in the diagnosis of drug-resistant TB occur when conventional culture-based susceptibility assays are used (the time required is, on average, 2-4 weeks).More rapid approaches that use genetic analysis for the detection of mutations associated with drug resistance have been described (Telenti et al. 1993, Kapur et al. 1994, Torres et al. 2000, 2003, El-Hajj et al. 2001, Garcia de Viedma et al. 2002, Garcia de Viedma 2003, van Doorn et al. 2003, Marín et al. 2004, Wada et al. 2004, Yesilkaya et al. 2006, Boehme et al. 2010) and the RT-PCR assay has been proposed as an alternative approach to detect drug-resistant organisms (Kapur et al. 1994, Piatek et al. 2000, Garcia de Viedma et al. 2002, Marín et al. 2004, Ruiz et al. 2004, Kocagoz et al. 2005, Boehme et al. 2010).In this study, we assessed reported RT-PCR assays to predict the drug resistance of M. tuberculosis isolates.
The RT-PCR results and DNA sequencing analysis obtained in this study were consistent for all katG and inhA mutations, except for one rpoB mutation.In this isolate, probe SW111 was not able to detect a mutation at codon 522 (TCG-TTG).We are unable to explain this discrepancy despite numerous repetitions of MGIT 960, DNA sequencing and RT-PCR assays.The mutation missed by probe SW111 is close to the 3' end of its sequence, only three nucleotides before the molecular beacon hairpin arm.It is possible that the binding capacity of probe SW111 is affected when mismatches are too close to that area.We also identified discrepancy between phenotypic and genotypic results of the RIF resistance of two M. tuberculosis isolates.DNA sequencing predicted that RIF resistance of these two isolates was consistent after three rounds of DNA sequencing and RT-PCR testing, but they were phenotypically RIFsensitive, even after repetitions.According to our data, RT-PCR and MGIT failed to detect 1% (1/74) and 2% (2/95) of RIF-resistant isolates, respectively.
If we consider that any mutation on the 81-bp core region of the rpoB gene confers RIF resistance, these two cases suggest that sometimes RT-PCR may be more accurate than the phenotypic assay.Similar discrepancies have been described, such as when a mutation at codon 533 was compatible with a clinically RIF-resistant case, but was overlooked by the BACTEC method (Riska et al. 2000).
All 75 (100%) RIF-resistant isolates obtained by genetic sequencing in this study had mutations in the rpoB gene.This contrasts with previous data showing that RIF-resistant isolates in Brazil exhibited rpoB mutation frequencies varying from 80-89% (Valim et al. 2000, de Miranda et al. 2001, Clemente et al. 2008).Previous data on katG reported that mutations occur in approximately 60-85% of INH-resistant isolates (Rossetti et al. 2002, Cardoso et al. 2004, Hofling et al. 2005, Clemente et al. 2008) but, in our study, mutation of the katG gene was seen in 55% (65/118) of the INH-resistant isolates.The reasons for these discrepancies are not clear, but may be related to different methods used to perform the phenotypic susceptibility testing, differences in year and geographical site of M. tuberculosis isolation and the use of different sizes of sample collections.
RT-PCR cannot completely replace culture-based susceptibility tests because conventional testing can detect resistance due both to genetic mechanisms that are not yet understood and to a larger number of drugs.There are a number of limitations to the widespread use of PCR-based techniques; resistance to anti-TB drugs may involve changes in multiple genes and at multiple possible locations within a gene.However, the fact that the majority of the INH and RIF resistance of M. tuberculosis isolates are due to point mutations on the katG, inhA and rpoB genes, respectively, molecular approaches such as RT-PCR are promising.
Our results confirm that RIF-resistant M. tuberculosis can be detected in less than 4 h using a very specific and sensitive assay.Moreover, the assay is a powerful tool because it is simple to perform and readily automatable for high-throughput screening and the results enable faster diagnosis and immediate decision-making to determine appropriate drug therapy.
The rapid detection of M. tuberculosis resistance continues to be an important public health matter.Commercial assays, such as GeneXpert (Cepheid), which detects M. tuberculosis and resistance to RIF, demonstrated good performance with clinical specimens (Helb et al. 2010, Lawn & Nicol 2011, Zeka et al. 2011).In our study, the use of RT-PCR to determine M. tuberculosis resistance to RIF was also very promising and its application in clinical samples will be explored in future studies.
The cost of in-house molecular designs could be equivalent to, or even less expensive than, the costs of phenotypic antibiograms if the new liquid-culture media are used.However, cost, speed and professional expertise are not the only factors to consider.The molecular analysis of resistance provides additional information and essential data that are not available if only phenotypic antibiograms are obtained.Genotypic analysis also specifies the resistance mutation involved, which helps to identify strains with high and/or broad levels of resistance (van Soolingen et al. 2000, Garcia de Viedma 2003).The optimal approach for drug susceptibility testing of MTB will likely involve both methodologies.
We selected the first 988 isolates of M. tuberculosis received in our laboratory between October 30 2008-March 13 2009 for phenotypic drug-susceptibility testing from TB patients living in greater São Paulo, Brazil.M. tuberculosis lysates were prepared by transferring colonies from a Löwenstein-Jensen slant into a 2 mL screw-cap tube containing Financial support: CNPq (573830/2008-8), FIC/Global Infectious Diseases Research Training Program, NIH, University of Pittsburgh (D43TW006592)

TABLE I
Primers and probes for gene amplification, sequencing and real-time polymerase chain reaction (RT-PCR)

TABLE II Gene
mutation found associated with anti-tuberculosis drug resistance in São Paulo isolates Susceptibility and resistence to isoniazid (INH), rifampin (RIF) or both in 988 isolates obtained by phenotypic and molecular methods [real-time polymerase chain reaction (RT-PCR)]