Emergence of rifampin-resistant strains of Mycobacterium tum sediments within 6 hours. The procedure utilizes a single

Transcription

1 446 Evaluation of a Polymerase Chain Reaction Based Universal Heteroduplex Generator Assay for Direct Detection of Rifampin Susceptibility of Mycobacterium tuberculosis from Sputum Specimens Diana L. Williams, Laynette Spring, Thomas P. Gillis, Max Salfinger, and David H. Persing From the Molecular Biology Research Department, Laboratory Research Branch, Gillis W. Long Hansen s Disease Center at Louisiana State University, Baton Rouge, Louisiana; Wadsworth Center, New York State Department of Health and Albany Medical College, Albany, New York; and Medical Microbiology Department, Mayo Clinic, Rochester, Minnesota In a double-blind study, 655 sputum specimens were obtained from individuals suspected of having tuberculosis and were analyzed for the presence of Mycobacterium tuberculosis and rifampin susceptibility with use of a polymerase chain reaction (PCR) based universal heteroduplex generator assay (PCR/UHG-Rif). Of the specimens containing viable M. tuberculosis, 100% of the smearpositive (n Å 41) and 50% of the smear-negative (n Å 6) specimens tested positive for the organism by PCR/UHG-Rif. Nineteen of 537 culture-negative specimens tested positive for M. tuberculosis by PCR/UHG-Rif and were from patients with confirmed tuberculosis who were receiving antituberculosis therapy at the time of specimen collection. Thirty-five specimens contained nontuberculous mycobacteria and were negative by PCR/UHG-Rif. Genotypic evidence of rifampin resistance in five of six culture-confirmed, rifampin-resistant isolates was obtained by PCR/UHG-Rif, yielding a sensitivity and specificity for the assay of 83% and 98.2%, respectively. These results demonstrate the feasibility of using a PCR-based assay directly on sputum specimens for simultaneous detection of M. tuberculosis and rifampin susceptibility, and they suggest that patients with smear-positive, untreated tuberculosis and those presenting with suspected drug-resistant tuberculosis are the most appropriate groups for testing by PCR/UHG-Rif. Emergence of rifampin-resistant strains of Mycobacterium tum sediments within 6 hours. The procedure utilizes a single tuberculosis has threatened the usefulness of this drug, a critical tube, heminested PCR amplification to generate an M. tubercu- component of the short-course multidrug therapy for tuberculo- losis specific, rpob gene fragment [11] that is annealed to a sis [1 3]. The need for new tools for the rapid detection of synthetic UHG derived from the Rif region of the rpob gene drug-resistant M. tuberculosis directly in clinical specimens is of M. tuberculosis [10]. Detection of the rifampin genotype great because of the lengthy time required to culture the organism, to identify it, and to perform drug susceptibility testing heteroduplexes separated on polyacrylamide minigels. is based on analysis of electrophoretic patterns of resultant by conventional culture-based techniques [1 5]. Recently, the In this report we demonstrate the feasibility of using this novel molecular basis of resistance of M. tuberculosis to rifampin technology for detecting rifampin susceptibility of M. tuberculosis from direct analysis of a large collection of clinical sputum has been shown to be associated with deletion, insertion, or missense mutations in the rpob gene, which encodes the B- specimens from patients suspected of having tuberculosis. subunit of the DNA-dependent RNA polymerase [6 9]. Using this information, we developed a rapid, DNA-based assay for detecting rifampin susceptibility of M. tuberculosis Materials and Methods directly in clinical specimens, called the PCR/universal heteroduplex generator assay (PCR/UHG-Rif) [10]. This assay simulbacteriology Laboratory of the Wadsworth Center (Albany, Sputum specimens (n Å 655) received at the Clinical Myco- taneously detects the presence of M. tuberculosis and identifies the rifampin susceptibility of the organism directly from spuevaluation. These specimens were submitted from chest clinics, NY) from January through August 1994 were included in this correctional facilities, or institutions enrolled in the New York State Fast Track Program for rapid tuberculosis testing. Fast Track specimens were of sputum from patients whose acid- Received 11 March 1997; revised 6 October fast-bacilli (AFB) stained smears were known to be positive. Grant support: National Institutes of Health (grant no. AI-35274). Reprints or correspondence: Dr. Diana L. Williams, Molecular Biology Specimens were shipped by overnight mail. Information on Research Department, Laboratory Research Branch, G. W. L. Hansen s Dis- patient treatment history was not available in most cases. ease Center, LSU, P.O. Box 25072, Baton Rouge, Louisiana Specimens were homogenized, decontaminated, and concen- Clinical Infectious Diseases 1998;26: trated according to recommendations of the Centers for Disease 1998 by The University of Chicago. All rights reserved /98/ $03.00 Control and Prevention [12]. Each resuspended pellet was inoc-

2 CID 1998;26 (February) Evaluation of PCR/UHG-Rif Assay 447 ulated into one 12B BACTEC (Becton Dickinson, Sparks, MD) Table 1. Detection of Mycobacterium tuberculosis in sputum speci- vial (0.5 ml) and onto one Lowenstein-Jensen-Gruft slant mens by PCR/UHG-Rif. (0.25 ml). In addition, specimens from the Fast Track Program Result of PCR/UHG-Rif and other AFB-smear-positive specimens were subcultured (no. of specimens) onto one Middlebrook 7H10/selective 7H11 biplate (0.25 ml). BACTEC vials were checked for elevated growth index twice weekly the first 2 weeks and weekly thereafter, for a total of Isolated in culture Positive Negative M. tuberculosis weeks. Fast Track specimens were read thrice weekly for 3 M. tuberculosis/mycobacterium avium 3 0 weeks then weekly for a total of 8 weeks. Any smear-positive NTM 0 35 specimens without growth at 8 weeks were incubated an addi- No growth tional 4 weeks. Biplates and the Lowenstein-Jensen-Gruft NOTE. PCR/UHG-Rif Å PCR/heteroduplex-based assay for detection of slants were checked for visible colonies weekly for 6 and 8 M. tuberculosis and susceptibility to rifampin. weeks, respectively. Specimens were analyzed for mycobacterial growth in BACTEC 12B and Smears from broth cultures with an elevated growth index Lowenstein-Jensen media. Nontuberculous mycobacteria: M. avium complex (19 specimens), M. foror visible colonies from solid media were checked for AFB tuitum (4), M. xenopi (2), and M. gordonae (7); 3 specimens contained acidwith the Ziehl-Neelsen stain. Identification of M. tuberculosis fast bacilli not further identified. complex in AFB-positive cultures was accomplished with the AccuProbe M. tuberculosis complex DNA probe assay (Gen- Probe, San Diego) according to the manufacturer s recommen- DNA sequencing of PCR products from all specimens condations. In addition, performance of a set of biochemical tests taining rifampin-resistant M. tuberculosis was accomplished enabled the final identification to the species level [12]. by cycle sequencing with use of the fmol DNA Sequencing As part of routine susceptibility testing, BACTEC radiomet- System (Promega, Madison, WI) as previously described [8]. ric rifampin-susceptibility testing was performed on all initial M. tuberculosis positive cultures, with use of a 2-mg/mL concentration of rifampin. If tubercle bacilli were resistant to rifampin, Results then initial cultures were tested for the percentage of rifam- Detection of M. tuberculosis. Of 655 clinical specimens ana- pin-resistant bacilli by the proportion method, with use of lyzed, 47 were culture-positive for M. tuberculosis, including Middlebrook 7H10 agar containing a 1-mg/mL concentration three specimens yielding mixed cultures of M. tuberculosis of rifampin [13]. and Mycobacterium avium; 35 were culture-positive for only Sputum sediments (250 ml) were added to 583 ml of abso- nontuberculous mycobacteria (NTM); and the remaining 573 lute ethanol (final concentration, 70%) in sterile, 1.5-mL microfuge specimens were culture-negative (table 1). Forty-four speciand tubes and incubated for at least 2 hours at 25 C to sterilize mens containing viable M. tuberculosis tested positive by preserve specimens for PCR analysis [14]. Ethanol-fixed PCR/UHG-Rif. Forty-one of these specimens were smear-posi- sediments were then stored at 25 C for 1 12 months until tive, including three specimens yielding mixed cultures of processed for PCR. For preparation of DNA template from M. tuberculosis and M. avium. Three of six specimens that were ethanol-fixed sediments prior to amplification, M. tuberculosis AFB smear negative and culture-positive for M. tuberculosis cells were lysed by means of a quick-boil lysis method [10]. tested positive by PCR/UHG-Rif. A 193-bp fragment of the rpob gene was amplified from Of the 35 specimens containing NTM (M. avium complex, crude cell lysates with use of a heminested PCR with primers Mycobacterium fortuitum, Mycobacterium gordonae, and My- and conditions previously described [10]. Positive controls con- cobacterium xenopi) none tested positive for M. tuberculosis sisted of PCR reactions with 10 ml of crude cell lysate containing by PCR/UHG-Rif (table 1). Nineteen of 573 culture-negative Ç10 3 M. tuberculosis H37Rv ATCC (rifampin- specimens tested positive for M. tuberculosis by PCR/UHG- susceptible), and negative controls consisted of PCR reactions Rif (table 1). Patient records showed that all 19 specimens were with 2 ml of ethanol-fixed pooled sputum sediments from 10 obtained from patients with culture-confirmed (prior culture) individuals who had no evidence of tuberculosis and were PCRnegative tuberculosis who were receiving antituberculosis therapy when for the IS6110 repetitive element of M. tuberculosis the specimens were taken for PCR/UHG-Rif and were therefore [15]. Two positive and two negative controls were included considered true positives in this evaluation (table 1). These with each group of 20 specimens. data indicated that the overall sensitivity and specificity of Heteroduplexes were generated and banding profiles were detected PCR/UHG-Rif for detecting M. tuberculosis in sputum speci- by electrophoresis on polyacrylamide minigels (Novex, mens were 96% and 100%, respectively. The sensitivity for San Diego) as previously described [10]. The gels were stained smear-positive and smear-negative culture-positive specimens in ethidium bromide (1 mg/ml) for 15 minutes and destained in was 100% and 50%, respectively. H 2 O for 10 minutes, and then heteroduplexes were visualized Detection of rifampin resistance. Of 44 specimens that were and photographed under ultraviolet transillumination. culture-positive for M. tuberculosis and positive for M. tuber-

3 448 Williams et al. CID 1998;26 (February) Table 2. Rifampin susceptibility of Mycobacterium tuberculosis in Discussion sputum specimens, as determined by PCR/UHG-Rif or conventional drug susceptibility testing. Rifampin is an important component of the effective multidrug No. of isolates therapy for tuberculosis [1 3, 16, 17]. Resistance to this drug is a problem of increasing importance in industrialized as Test for M. tuberculosis M. tuberculosis well as developing countries, particularly among HIV-positive resistance culture positive culture negative individuals [1 3, 16, 17], and is a marker for multiple drug resistant tuberculosis. Infections with rifampin-resistant strains BACTEC of M. tuberculosis require accurate and rapid identification so Susceptible 38 NA that appropriate therapy can be initiated to reduce morbidity Resistant 6 NA Total 44 and mortality and to reduce the spread of rifampin-resistant PCR/UHG-Rif tuberculosis. In the present study, the PCR/UHG-Rif was com- Susceptible pared to conventional culture methodologies for the detection Resistant 5 1 of M. tuberculosis and rifampin susceptibility with use of a Total large collection of clinical sputum specimens. NOTE. PCR/UHG-Rif Å PCR/heteroduplex-based assay for detection of The design of this study was to analyze consecutive speci- M. tuberculosis and rifampin susceptibility; NA Å not applicable. mens referred to the New York State Tuberculosis Laboratory. BACTEC M. tuberculosis rifampin-susceptibility testing with a 2-mg/mL Since this laboratory employs the Fast Track Program, a disproconcentration of rifampin. portionate number of specimens that are smear-positive for AFB are submitted for identification of M. tuberculosis and determination of susceptibility to first-line drugs. Of the 47 culosis by PCR/UHG-Rif, 38 demonstrated the rifampin-susstudy, specimens that were culture-positive for M. tuberculosis in this ceptible phenotype in BACTEC drug susceptibility testing, 29 (61%) were submitted for the Fast Track analysis while 39 demonstrated the rifampin-susceptible genotype in (i.e., were smear-positive) and 18 (39%) were submitted as PCR/UHG-Rif (table 2). Among the 19 M. tuberculosis culsmear-negative. routine specimens. Only six of the routine specimens were ture negative, PCR/UHG-Rif-positive specimens, 18 showed Therefore, this study primarily focuses on the the rifampin-susceptible genotype (table 2). Taken together, the utility of genotypic testing of smear-positive specimens. sensitivity and specificity of the PCR/UHG-Rif for detecting Results of this study demonstrated that PCR/UHG-Rif accu- rifampin resistance in culture-positive specimens was 83% and rately detected both the presence and rifampin susceptibility 98.2%, respectively. of M. tuberculosis in clinical specimens. Excellent sensitivity The specimen that tested genotypically rifampin-susceptible and specificity values were obtained for detecting M. tuberculoby PCR/UHG-Rif but phenotypically rifampin-resistant by sis, and the overall predictive value of a positive and negative BACTEC did not contain mutations within the Rif region of the rpob gene targeted by PCR/UHG-Rif, as determined by DNA sequencing (table 3). The rifampin-susceptibility finding Table 3. Analysis of rifampin-resistant M. tuberculosis from clinical for this isolate was repeated with use of the proportion method, specimens of patients with tuberculosis. and it was found to be resistant to a 2-mg/mL concentration of Percentage of rifampin. When a single colony from this isolate grown on Specimen rifampin PCR/UHG- Mutant amino 7H10 agar containing rifampin (1 mg/ml) was fixed in ethanol no. resistance* Rif genotype acid and crude cell lysates were analyzed for the presence of mutations in the rpob Rif region by direct DNA sequencing, no S None mutations were found R Leu R Tyr 526 The other five specimens containing rifampin-resistant R Phe 514 (insertion) M. tuberculosis showed altered heteroduplex patterns and x ND R Tyr 526 contained mutations within the 81-bp Rif region of rpob, resulting in substitutions within either the serine 531 or histi R Phe 514 (insertion) * Per proportion plate method (7H10 agar, 1 mg/ml). dine 526 codons or an insertion of a phenylalanine codon R Å rifampin-resistant M. tuberculosis; S Å rifampin-susceptible M. tuberbetween codons 513 and 514 (table 3). All of these mutations culosis. have been previously found to be associated with the develop- Numbering system based on the b-subunit of the RNA polymerase of Escherichia coli. ment of rifampin resistance in M. tuberculosis [8]. Propor- Rifampin susceptibility results were not obtained for this specimen but are tion-method analysis demonstrated that these rpob mutants from M. tuberculosis cultures of specimens taken from the same patient at the were detected with PCR/UHG-Rif in specimens that con- same time. tained between 10% and 55% rifampin-resistant M. tuberculosis organisms (table 3). x ND Å not done; specimen was culture-negative. Other specimens from same patient contained rifampin-resistant M. tuberculosis (% of rifampin resistance unknown).

4 CID 1998;26 (February) Evaluation of PCR/UHG-Rif Assay 449 test was 100% and 99.5%, respectively. In addition, with a significant prevalence of multiple drug resistant PCR/UHG-Rif detected the presence of M. tuberculosis DNA M. tuberculosis and HIV coinfection. in 19 culture-negative specimens. Further analysis of the re- Our results also suggest that smear-positive, untreated cases cords for these specimens showed that they originated from of tuberculosis are the most appropriate group for testing by patients with culture-confirmed tuberculosis who were being PCR/UHG-Rif. However, we feel that patients receiving treatment treated when specimens were collected for PCR/UHG-Rif. and suspected of having drug-resistant disease or who Accuracy of the PCR/UHG-Rif test for detecting susceptibil- are noncompliant with therapy should be tested. Genotypic ity to rifampin was good, as judged by sensitivity (83%) and information on rifampin susceptibility in this group of patients specificity (98.2%). Predictive values for a positive test (rifam- may provide particularly salient information to guide the clinician pin resistance; 83%) and negative test (rifampin susceptibility; in continuing therapy. 97%) were also high and reflected the low percentage of false Caution should be taken with interpretation of PCR/UHGnegatives and false positives in the testing. Rif results for treated patients, because clinical histories in Since PCR/UHG-Rif is based on the detection of alter- many of the culture-negative suspected cases were incomplete ations in the nucleic acid sequence of the Rif region of the in our study; therefore, the suitability of PCR/UHG-Rif for rpob gene to determine rifampin resistance [10], specimens culture-negative specimens is unclear. Accordingly, we would that exhibit the rifampin-resistant phenotype but do not contain suggest that conventional culture should be run in parallel on mutations within the Rif region of rpob will appear as specimens from patients receiving treatment, to determine if rifampin-susceptible in this assay. One such specimen was viable M. tuberculosis is present in these specimens. In addi- found in this study. This specimen was found repeatedly to tion, our results indicate that conventional culture methodologies, contain 50% rifampin-resistant M. tuberculosis by the proportion such as BACTEC, should be used concurrently on AFB method but contained the rifampin-susceptible geno- smear negative specimens to reduce the risk of missing type in PCR/UHG-Rif and by DNA sequencing, indicating M. tuberculosis positive specimens that may test negative by that the mechanism of resistance may be attributed to muta- PCR/UHG-Rif. tions within rpob outside the region analyzed or to a mechanism The most cost-effective implementation of this assay would not involving rpob. be at the centralized laboratory level, where large numbers of Incorrect susceptibility results could occur with use of specimens would be available for daily testing. In this manner, PCR/UHG-Rif if a specimen contains a rifampin-susceptible large numbers of specimens could be evaluated at the same strain of M. tuberculosis containing a silent mutation within time, thus reducing the overall cost of the assay per specimen. the Rif region of the rpob. A recent review by Musser suggests In summary, PCR/UHG-Rif is a powerful method for the that this occurs infrequently [18]. For example, of Ç500 rifampin-susceptible simultaneous detection of M. tuberculosis complex and rifam- and rifampin-resistant M. tuberculosis strains pin susceptibility directly from clinical sputum specimens. evaluated for the presence of mutations in the Rif region, õ2% PCR/UHG-Rif requires minimal time (Ç6 hours) to perform, have been shown to contain silent mutations. None were requires less time and sophisticated equipment than does direct observed in the strains evaluated in the present study. DNA sequencing, and should be applicable to most centralized The ability of PCR/UHG-Rif to detect the rifampin-resistant clinical laboratories in which PCR technology is available. In phenotype in specimens containing as few as 10% resistant addition, since ethanol-fixed sputum sediments are used in this M. tuberculosis organisms (table 3) is very important because procedure, the hazards associated with working with live many specimens do not contain 100% rifampin-resistant organisms, M. tuberculosis are eliminated [14]. By testing sputum speciof as demonstrated by drug susceptibility testing with use mens directly, PCR/UHG-Rif can reduce the time between the proportion method. In a recent study of 158 rifampinresistant acquisition of the specimen, determination of rifampin suscepti- M. tuberculosis isolates obtained in New York, 21% of bility, and initiation of appropriate chemotherapy for those specimens contained 10% 25% rifampin-resistant organisms, individuals with rifampin-resistant tuberculosis. In addition, 32% contained 30% 50%, 26% contained 55% 75%, and resistance to rifampin is often associated with resistance to 18% contained 80% 100% in the original cultures (M. Salfinger, isoniazid [3, 17] and may serve as a marker for multiple drug unpublished data, ). No specimens were en- resistant tuberculosis. Therefore, rapid detection of rifampin countered in this study with õ10% rifampin-resistant organisms. resistance may play a key role in the monitoring and control Therefore, it is unclear how this assay would perform of multiple drug resistant tuberculosis. for the detection of rifampin resistance in specimens containing õ10% resistant M. tuberculosis organisms. It is our impression that PCR/UHG-Rif may be useful in a setting where rapid detection of M. tuberculosis and rifampin Acknowledgments susceptibility profiles would improve management and restrict The authors thank Marie Fitzgerald for her excellent data manthe spread of multiple drug resistant M. tuberculosis strains. agement, Christy Limbers for her technical support, and Teresa One such setting would be laboratories serving populations Felmlee for helpful suggestions.

5 450 Williams et al. CID 1998;26 (February) References 10. Williams DL, Limbers C, Spring L, Jayachandra S, Gillis T. PCR-heteroduplex detection of rifampin-resistant Mycobacterium tuberculosis. In: 1. Frieden T, Sterling T, Pablos-Mendez A, Kilburn G, Cauthen M, Dooley Persing D, ed. PCR protocols for emerging infectious diseases. 1996: S. The emergence of drug-resistant tuberculosis in New York City. N Engl J Med 1993;328: Whelen AC, Felmlee TA, Hunt JM, et al. Direct genotypic detection of 2. Goble M, Iseman M, Madsen L, Waite D, Ackerson L, Horsburgh C. Mycobacterium tuberculosis rifampin resistance in clinical specimens Treatment of 171 patients with pulmonary tuberculosis resistant to isoni- by using a single tube heminested PCR assay. J Clin Microbiol 1995; azid and rifampin. N Engl J Med 1993;328: : Ellner JJ, Hinman AR, Dooley SW, et al. Tuberculosis symposium: emerg- 12. Kent PT, Kubica GP. Public health mycobacteriology a guide for the ing problems and promise. J Infect Dis 1993;168: level III laboratory. Atlanta: U.S. Department of Health and Human 4. Salfinger M, Pfyffer GE. The new diagnostic mycobacteriology laboratory. Services, Centers for Disease Control, Eur J Clin Microbiol Infect Dis 1994;13: Inderlied CB, Salfinger M. Antimicrobial agents and susceptibility tests: 5. Huebner RE, Good RC, Tokars JI. Current practices in mycobacteriology: mycobacteria. In: Murray PR, ed. Manual of clinical microbiology. 6th results of a survey of state public health laboratories. J Clin Microbiol ed. Washington, DC: American Society for Microbiology, 1995: 1993;31: p Telenti A, Imboden P, Marchesi F. Detection of rifampin-resistance muta- 14. Williams DL, Gillis TP, Dupree WG. Ethanol fixation of sputum sediments tions in Mycobacterium tuberculosis. Lancet 1993;341: for DNA-based detection of Mycobacterium tuberculosis. J Clin Micro- 7. Telenti A, Imboden P, Marchesi F, Schmidheini T, Bodmer T. Direct, biol 1995;33: automated detection of rifampin-resistant Mycobacterium tuberculosis 15. Eisenach KD, Cave MD, Bates JH, Crawford JT. Polymerase chain reacby polymerase chain reaction and single-strand conformation polymor- tion amplification of a repetitive DNA sequence specific for Mycobactephism analysis. Antimicrob Agents Chemother 1993;37: rium tuberculosis. J Infect Dis 1990;161: Williams DL, Waguespack C, Gillis TP, et al. Characterization of rifampin 16. Mitchison DA, Nunn AJ. Influence of initial drug resistance on the reresistance in pathogenic mycobacteria. Antimicrob Agents Chemother sponse to short-course therapy of pulmonary tuberculosis. Am Rev 1994;38: Respir Dis 1986;133: Kapur V, Li L, Iordanescu S, et al. Characterization by automated DNA 17. Vareldzis BP, Grosset J, de Kantor I, et al. Drug-resistant tuberculosis: sequencing of mutations in the (rpob) encoding the RNA polymerase laboratory issues. Tuber Lung Dis 1994;75:1 77. B subunit in rifampin-resistant Mycobacterium tuberculosis strains from 18. Musser JM. Antimicrobial agent resistance in mycobacteria: molecular New York City and Texas. J Clin Microbiol 1994;32: genetic insights. Clin Microbiol Rev 1995;8: