Direct Identification of Mycobacteria in Primary Liquid Detection Media by Partial Sequencing of the 65-Kilodalton Heat Shock Protein Gene
http://www.100md.com
微生物临床杂志 2006年第1期
Laboratory Services, British Columbia Centre for Disease Control
Department of Pathology and Laboratory Medicine, University of British Columbia, Vancouver, British Columbia, Canada
ABSTRACT
We investigated extending the use of direct partial hsp65 gene sequencing for the identification of mycobacteria to isolates in primary liquid detection media as an economical, feasible, and more rapid means of identification. During the course of the study, the hsp65 sequence-based identifications for isolates from 670 primary liquid detection media determined to be positive for acid-fast bacilli were compared to the identifications derived from Accuprobes, biochemical test panels, or 16S rRNA gene sequencing. Preliminary analysis indicated a 97.6% concordance, with a final agreement of 99.1% between the identification algorithms. hsp65 sequencing costs (US$32.84) were greater than the cost of identification with Accuprobe (US$19) but less than the cost of the biochemical test panel identification (average cost, US$98.90) and equivalent to the cost of 16S rRNA sequencing, although there was a referral cost (US$59.85) for the shipping of isolates to another reference laboratory. Analysis indicated that our laboratory would have recognized a cost savings of approximately $12,000 by using hsp65 sequencing to identify isolates from specimens with a negative fluorescent- smear status and would have achieved further savings by using it as an alternative to biochemical panel testing for fluorescent-smear-positive specimens. The time to identification by hsp65 gene sequencing was slightly longer than that required by the Accuprobe assay (1 versus 2 days), shorter than that required by the biochemical test panels (2 days versus 26 days on average), and more rapid than referral for 16S rRNA gene sequencing.
INTRODUCTION
The identification of mycobacteria has traditionally been accomplished by determining their ability to utilize particular compounds, their growth characteristics, and their colonial morphologies. This testing is performed with isolates derived from primary cultures or subcultures from primary liquid detection media (PLDM), such as Bac T Alert 3D bottles (bioMerieux, Durham, N.C.), BACTEC 12B bottles (Becton Dickinson Diagnostic Instrument Systems, Sparks, Md.), MGIT bottles (Becton Dickinson Diagnostic Instrument Systems), or Myco/F bottles (Becton Dickinson Diagnostic Instrument Systems) (10, 21). Regardless of the source, all isolates must be incubated for sufficient growth to occur, which, depending on the organism, may take several days to several weeks. This means of identification requires expertise and is time-consuming, expensive, and labor-intensive, but it is still used today by many mycobacteriology laboratories to identify at least some Mycobacterium species.
The introduction of the Accuprobe system (GenProbe, San Diego, Calif.) greatly accelerated the identification of some mycobacteria, as testing could be performed directly from PLDM. Unfortunately, probes specific for only four species and two complexes have been developed. Several "home-brew" restriction enzyme analysis methods have been developed as a means for the rapid identification of mycobacteria from both solid and liquid culture media, but they are not without problems (3-9, 11, 14, 16, 17, 18, 19, 23). More recently, a commercially available line probe assay (Inno-Lipa Mycobacteria; Innogenetics, Ghent, Belgium) has become available for the identification of Mycobacterium species. This assay identifies a larger number of species than Accuprobe tests (16 species versus 4 species and two complexes) and has the advantage of testing for all species within its database at one time (13). Previously, we reported on the use of partial hsp65 gene sequencing as a means for mycobacterial identification and found it to be an extremely rapid identification method but required isolates from solid culture media, which lengthened the time from organism detection in primary liquid detection media until the final identification (12). Therefore, we explored the utility of hsp65 gene sequencing directly from primary liquid detection media determined to be positive for acid-fast bacilli as an alternative and cost-effective means of identification of mycobacteria. We report here on the results of that study.
MATERIALS AND METHODS
Mycobacterial strains. A total of 670 bottles of primary liquid detection media (BACTEC 5, MGIT 96, Myco/F 17, and Bac T Alert 3D 552 bottles) not included in our previous study and determined to be positive for the presence of acid-fast rods were investigated (12). Conventional identification of the isolates by the use of our current identification algorithm spanned 37 Mycobacterium species and taxonomic groups and unique species, as well as Nocardia and Tsukamurella species (Table 1).
Conventional identification. Mycobacterial isolates were identified directly from fluorescent-smear-positive specimens as Mycobacterium tuberculosis complex by using the Amplified Mycobacterium tuberculosis Direct test (AMTD; GenProbe, San Diego, Calif.) and confirmed by an M. tuberculosis complex-specific Accuprobe assay from PLDM or were identified from PLDM known to be positive for acid-fast organisms by using M. tuberculosis complex-, Mycobacterium gordonae-, or Mycobacterium avium-Mycobacterium intracellulare complex-specific Accuprobes (GenProbe). These identifications were confirmed by colonial morphology, as in the case of M. gordonae and the M. avium-M. intracellulare complex, or by nitrate and niacin testing for the M. tuberculosis complex. When organisms in PLDM were Accuprobe negative or not tested, aliquots were subcultured to appropriate solid culture media and the isolates were identified by using further Accuprobes or biochemical test panels or, when it was required, by partial or complete 16S rRNA gene sequencing, as described previously (12).
DNA extraction for hsp65 gene sequencing. For all PLDM confirmed to contain acid-fast organisms, a 1.5-ml aliquot was transferred to a microcentrifuge tube and the DNA was extracted as published previously (2, 12).
DNA amplification and partial hsp65 gene sequencing. Partial amplification and hsp65 gene sequencing, identification of all isolates, as well as PCR and sequencing of controls were carried out as described previously (12).
Determination and identification of isolates from mixed sequences. The optimization of the cycle sequence reactions during our previous study ensured that the signal-to-noise ratio was very high and that the presence of a second signal predicted to exceed 20% of the major signal could be detected. All sequences were examined for the presence of subordinate base peaks (2, 12). When subordinate peaks occurred, two sequences were generated and subjected to BLAST searches. One consisted of the sequence derived from all major base call peaks, and the second sequence was derived by substituting the minor base calls for the major peaks where they occurred, which we speculated should represent the major and minor organisms present, respectively.
Mycobacterial hsp65 sequence database. Our previous database, which consisted of 140 hsp65 sequence entries (111 valid and putative species and groups plus 29 distinct sequences), was supplemented with sequence entries downloaded from the GenBank database, further unique sequences detected in our laboratory, and type strains of Nocardia and Tsukamurella species: M. avium complex group B8; Mycobacterium chimaera (GenBank accession number AJ548480); Mycobacterium cosmeticum (GenBank accession number AY449731); Mycobacterium florentinum (GenBank accession number AJ616230); Mycobacterium massilense (GenBank accession number AY596465); Mycobacterium parascrofulaceum sequevars I, II, III, IV, and V (GenBank accession numbers AY337274, AY337275, AY337276, AY337277, and AY337278, respectively); Mycobacterium saskatchewanense (GenBank accession number AY208859), Mycobacterium shinshuense ATCC 33728T, Mycobacterium species strain MCRO 24; Mycobacterium vanbaalenii (GenBank accession number AY438091); Nocardia asteroides ATCC 19247T; Nocardia brasiliensis ATCC 19296T; Nocardia carnea ATCC 6847T; Nocardia farcinica ATCC 3318T; Nocardia nova ATCC 33726T; Nocardia otitidiscaviarum ATCC 14629T; Nocardia transvalensis ATCC 6865T; Tsukamurella inchonensis DSM 44067T; Tsukamurella paurometabola DSM 20162T; Tsukamurella pulmonis DSM 44142T; Tsukamurella spumae DSM 44113T; Tsukamurella strandjordii DSM 44573T; and Tsukamurella tyrosinosolvens DSM 44234T. The database was also updated during the course of the study with Mycobacterium kansasii types 2, 3, 4, and 5 (GenBank accession numbers AY438086, AY438087, AY438088, and AY438089, respectively).
Cost analysis. The costs associated with facility space (lighting and heating or air conditioning), primary specimen examination (stained smears and solid and/or liquid media), sequencer acquisition, and depreciation were not included in the overall cost calculations. The overall costs of identification of isolates identified by the AMTD test, Accuprobes, biochemical test panels, or 16S rRNA sequencing were determined by using a subset of 158 positive specimens. These samples were allocated into one of two groups: samples with a positive direct acid-fast smear result (n = 39) or samples with a negative direct acid-fast smear result (n = 119). An AMTD test was performed for samples from the former category, based on a decision by a technologist following a prescribed identification algorithm. If the AMTD test detected M. tuberculosis complex in the patient's specimen, the identification was confirmed with an M. tuberculosis complex-specific Accuprobe when the PLDM became positive, and the species of the isolate was determined from solid media by using nitrate and niacin tests. Specimens that had a negative AMTD test result (n = 19) or for which the test was not performed (n = 119) were tested by using further Accuprobes when the corresponding PLDM was determined to be positive for acid-fast bacilli. All Accuprobe-negative organisms were subjected to biochemical test panel identification, and those not identified were referred to the Canadian National Reference Centre for Mycobacteriology (NRCM), National Microbiology Laboratory, Winnipeg, Manitoba, Canada, for partial 16S rRNA gene sequencing. Cost analysis of all tests, which were based on a conversion rate of 0.73 Canadian dollars to 1.00 U.S. dollar, included technologist time and all consumables and were calculated to be as follows (all values are in U.S. dollars): AMTD, $19; Accuprobe, $19; biochemical test panels for slowly growing biochemically active and inactive organisms, $97.63; biochemical test panels for rapidly growing organisms, $102.12; processing and shipping to NRCM, $59.85; and hsp65 sequencing, $32.84. The cost of partial 16S rRNA gene sequencing was not included as a cost to our facility.
RESULTS
Mycobacterium species were the only organisms present in 613 of 670 (91.5%) of the PLDM indicated to contain acid-fast organisms (Table 1). The remaining PLDM (Table 1) contained organisms other than Mycobacterium species (19 PLDM), were mixed cultures (22 PLDM), or failed to grow any acid-fast organism (16 PLDM). We were unable to amplify the hsp65 gene from any of the last set of cultures but were able to amplify the hsp65 gene from all other samples, with a single exception, and confirmed the identities of all M. tuberculosis complex isolates, all M. gordonae isolates, and 316 of 318 M. avium-M. intracellulare complex isolates, as well as all isolates from 25 other species (Table 1) (12). Mixed cultures represented 3.3% of all PLDM examined, with 12 of 22 (54.5%) detected by direct hsp65 sequencing and 10 of 22 detected only upon culture to solid media (Table 2). We confirmed the identities of 43 isolates from the 22 mixed cultures directly from the PLDM or from solid culture media, bringing our overall identification agreement to 99.1% (650 of 656).
Cost analysis. A subset of 158 specimens for which the results of all tests were known was used to develop a "per identification cost," which was later applied to all samples in the study (Table 3). The subset was divided into a direct fluorescent-smear-positive group (group 1) and a direct fluorescent-smear-negative group (group 2). Group 1 consisted of 39 samples, with 20 having a positive AMTD test result and 19 having a negative AMTD test result. This represented a total of 39 tests for the identification of 20 isolates at a cost to our laboratory of $37.05 per direct identification (39 tests x $19) of M. tuberculosis complex. Of the 20 AMTD test-positive specimens, all had a corresponding positive PLDM result and the M. tuberculosis complex identification was confirmed by the Accuprobe assay. Two AMTD test-positive PLDM were also tested by an M. avium-M. intracellulare complex-specific Accuprobe assay, but both were negative. We estimated our cost to be $20.90 per confirmed identification (22 tests x $19 per test). Of the 19 AMTD test-negative cultures, 14 were identified as M. avium-M. intracellulare complex and 1 was identified as M. gordonae by the Accuprobe assay; but 5 negative Accuprobe tests were performed, and we estimated the cost to be $20.00 per identification (20 tests x $19 per test). Four cultures did not react with Accuprobes and were identified by in-house biochemical test panels at $102.12 per isolate. The total cost required to confirm the identification of M. tuberculosis complex or to identify other species detected by PLDM from fluorescent-smear-positive specimens was estimated to be $30.90 per isolate. The cost required to identify isolates by hsp65 sequencing remained constant at $32.84 per identification. Of the 119 organisms allocated to group 2, 87 organisms were identified by the Accuprobe assay. In addition, 53 tests gave negative results, for a total of 140 Accuprobe tests performed to identify 87 isolates, with an estimated identification cost of $30.57 per isolate (140 tests x $19 per test). The 32 isolates not identified by Accuprobe tests were subjected to identification by biochemical test panels. Eighteen isolates were rapidly growing mycobacteria (identification cost, $102.12), and 14 were slowly growing mycobacteria (identification cost, $97.63). Twenty-two isolates identified at an average cost of approximately $145.68 per identification and the 10 remaining isolates required 16S rRNA gene sequencing, with an incurred shipping cost of $59.85 per shipment. We estimate that the identification costs for group 2 isolates were approximately $51.29 per isolate by our current algorithm and $32.84 per identification by hsp65 sequencing.
DISCUSSION
A recent study conducted by our laboratory demonstrated the validity of using partial hsp65 sequencing from solid culture media as a rapid means for the identification of Mycobacterium species (12). In an attempt to further decrease the time required to identify isolates and reduce costs, we investigated the use of an aliquot of PLDM as a source of mycobacterial DNA (12). To date, we have analyzed 656 isolates from 670 PLDM and determined the correlation between our current identification algorithm and hsp65 sequencing to be 99.1%. Similar to our previous study, Mycobacterium species capable of being identified by Accuprobe as M. avium-M. intracellulare complex (n = 329), M. gordonae (n = 82), or M. tuberculosis complex (n = 73) constituted approximately 70% of the isolates detected in our cultures. As noted previously, most identification discrepancies were limited to a few species (12). Of interest in this study was the fact that examination of 34 M. abscessus isolates revealed that 9 were identical to our database entry for M. abscessus variant VM585. These nine isolates had a 98.8% sequence match (396 of 401 nucleotides) with M. abscessus ATCC 19977T and a 100% sequence match with the newly proposed species M. massiliense (1). These organisms were single isolates from respiratory specimens from eight different patients, indicating that M. massiliense is common in our patient population (26.4% of all M. abscessus isolates), but unlike the isolate used to describe the new species (1), they may not have had a pathogenic role. As well, examination of 79 isolates that we identified as M. intracellulare indicated that a group of 9 isolates from different patients had identical hsp65 sequences and were only 99.0% related to the M. intracellulare type strain (unpublished data) but were identical to the recently published species M. chimaera (20). Full 16S rRNA gene sequencing of one of our isolates indicated that it was identical to the M. chimaera type strain (GenBank accession number AJ548480), and we assumed that our other eight isolates were identical. This indicates that M. chimaera was relatively common in our patient population, representing 2.8% of all our M. avium-M. intracellulare complex isolates and 10.5% of our purported M. intracellulare isolates.
Our cost for the identification of isolates by use of our standard identification algorithm varied from a low of $20.00 to confirm the results for direct fluorescent-smear-positive, AMTD test-positive, and PLDM-positive samples to in excess of $145.68 to confirm the results for isolates that were direct fluorescent-smear negative, Accuprobe negative, and not identified by biochemical test panels and that required 16S rRNA gene sequencing. By our identification algorithm, 167 isolates were not identified by Accuprobes and required biochemical identification. These Mycobacterium species, as well as Nocardia and Tsukamurella species, constituted 25.5% of the isolates in our study (Table 1). Ninety-seven isolates met the identification criteria for in-house biochemical testing, and 70 isolates were referred to NRCM for identification or confirmation of the identity in 31 shipments, at a total cost to our facility of $1,810. Of the 70 isolates submitted to NRCM, 18 isolates (25.7%) were identified as Tsukamurella species, a genus not included in our biochemical test panel database. We had encountered Tsukamurella species in our previous study, and to aid in their identification we had included sequences from six Tsukamurella species type strains in our most recent hsp65 sequence database. However, examination of the hsp65 sequence divergence for these species (unpublished data) indicated that, like the partial 16S rRNA gene sequence, hsp65 sequences may not be sufficiently divergent to report the results for isolates to the species level or that the current taxonomy is not adequate and we reported them only as Tsukamurella species, a cost-effective alternative to the use of our primary identification algorithm (15, 22).
Mixed cultures represented 3.3% (22 of 670) of the PLDM indicated to contain acid-fast bacilli. Twelve mixed cultures were detected by direct hsp65 sequencing, and the remainder were detected only upon subculture to solid media. Since the presence of mixed cultures was immediately reported to our Mycobacteriology Laboratory at the British Columbia Centre for Disease Control, we are not able to determine the impact that these results had on the solid medium culture report. Two incidences occurred where M. tuberculosis complex isolates in mixed culture were detected only upon subculture of positive PLDM to solid culture media. We had postulated that for cost containment, hsp65 sequencing would allow us to forgo subculture of PLDM believed to contain organisms of dubious pathogenicities, but these results indicate that all positive PLDM must be subcultured to substantiate the presence of a single Mycobacterium species.
PLDM submitted during this study contained 656 isolates that spanned 37 Mycobacterium species or taxonomic groups and 5 unique Mycobacterium species (as determined by partial 16S rRNA gene sequencing), as well as Nocardia and Tsukamurella species (Table 1). Mycobacterium species can be identified by Accuprobes for $19 per identification, under ideal conditions. Analysis of a subset of our PLDM indicates that this does not always occur and that the cost per identification within this group varied. One hundred seventy isolates not able to be identified by Accuprobes required identification by biochemical test panels at an average cost of approximately $98.00 per attempt. Only 97 of these total isolates were identified by in-house biochemical test panels, although some were identified incorrectly, at an approximate total cost of $10,000. The same isolates were identified by hsp65 sequencing at a total approximate cost of $3,250. The identification of 70 isolates sent to NRCM for confirmation or not identified by biochemical test panels and referred for 16S rRNA sequencing was estimated to cost our facility $7,400, as opposed to an identification cost of $2,300 by hsp65 gene sequencing. This equates as a total savings to our laboratory of approximately $12,000 for the hsp65 identification of Accuprobe-negative isolates, as well as a significant time savings per identification. Our latest study supports the results of our previous study and has proven the usefulness and cost-efficiency of hsp65 gene sequencing directly from primary liquid detection media as a cost-effective means for the identification of mycobacteria (12).
ACKNOWLEDGMENTS
We thank Nicola DiTomaso, Mycobacteriology Laboratory, British Columbia Centre for Disease Control, for excellent technical assistance; all members of the Mycobacteriology Laboratory at the British Columbia Centre for Disease Control for their assistance and cooperation during the course of this study; as well as Joyce Wolfe and the staff at the National Reference Centre for Mycobacteriology for 16S rRNA gene sequencing.
REFERENCES
Adekambi, T., M. Reynaud-Gaubert, G. Greub, M. J. Gevaudan, B. La Scola, D. Raoult, and M. Drancourt. 2004. Amoebal coculture of "Mycobacterium massilense" sp. nov. from the sputum of a patient with hemoptoic pneumonia. J. Clin. Microbiol. 42:5493-5501.
Applied Biosystems Inc. 2001. Users manual, ABI Prism 3100 genetic analyzer. Applied Biosystems Inc., Foster City, Calif.
Beggs, M. L., R. Stevanova, and K. D. Eisenach. 2000. Species identification of Mycobacterium avium complex isolates by a variety of molecular techniques. J. Clin. Microbiol. 38:508-512.
Brunello, F., M. Ligozzi, E. Cristelli, S. Bonora, E. Tortoli, and R. Fontana. 2001. Identification of 54 mycobacterial species by PCR-restriction fragment length polymorphism analysis of the hsp65 gene. J. Clin. Microbiol. 39:2799-2806.
Da Silva Rocha, A., C. da Costa Leite, H. Magarinos Torres, A. B. de Miranda, M. Q. Pires Lopes, W. M. Degrave, and P. N. Suffys. 1999. Use of PCR-restriction fragment length polymorphism analysis of the hsp65 gene for rapid identification of mycobacteria in Brazil. J. Microbiol. Methods 37:223-229.
de Magalhaes, V. D., P. de Melo Azevedo Fda, J. Pasternak, and M. D. Valle Martino. 2002. Reliability of hsp65-RFLP analysis for identification of Mycobacterium species in cultured strains and clinical specimens. J. Microbiol. Methods 49:295-300.
Devallois, A., K. S. Goh, and N. Rastogi. 1997. Rapid identification of mycobacteria to species level by PCR-restriction fragment length polymorphism analysis of the hsp65 gene and proposition of an algorithm to differentiate 34 mycobacterial species. J. Clin. Microbiol. 35:2969-2973.
Ergin, A., T. Kocagoz, and D. Us. 2000. Evaluation of 120 mycobacterial strains isolated from clinical specimens to the species level by polymerase chain reaction-restriction enzyme analysis. Scand. J. Infect. Dis. 32:657-662.
Hernandez, M. S., G. P. Morlock, W. R. Butler, J. T. Crawford, and R. C. Cooksey. 1999. Identification of Mycobacterium species by PCR-restriction fragment length polymorphism analyses using fluorescence capillary electrophoresis. J. Clin. Microbiol. 37:3688-3692.
Hosojima, A. C. S., Y. Fukasawa, Y. Kazumi, M. Takahashi, K. Hirano, and T. Mori. 1992. Comparison of MB-Check, Bactec, and egg-based media for recovery of mycobacteria. J. Clin. Microbiol. 30:878-881.
Levi, M. H., J. Bartell, L. Gandolfo, S. C. Smole, S. F. Costa, L. M. Weiss, L. K. Johnson, G. Osterhout, and L. H. Herbst. 2003. Characterization of Mycobacterium montefiorense sp. nov., a novel pathogenic mycobacterium from moray eels that is related to Mycobacterium triplex. J. Clin. Microbiol. 41:2147-2152.
McNabb, A., D. Eisler, K. Adie, M. Amos, M. Rodrigues, G. Stephens, W. A. Black, and J. Isaac-Renton. 2004. Assessment of partial sequencing of the 65-kilodalton heat shock protein gene (hsp65) for routine identification of Mycobacterium species isolated from clinical sources. J. Clin. Microbiol. 42:3000-3011.
Padilla, E., V. Gonzalez, J. M. Manterola, A. Perez, M. D. Quesada, S. Gordillo, C. Vilaplana, M. A. Pallares, S. Molinos, M. D. Sanchez, and V. Ausina. 2004. Comparative evaluation of the new version of the Inno-Lipa mycobacteria and genotype mycobacterium assays for identification of Mycobacterium species from MB/BacT liquid cultures artificially inoculated with mycobacterial strains. J. Clin. Microbiol. 42:3083-3088.
Pai, S., N. Esen, X. Pan, and J. M. Musser. 1997. Routine rapid Mycobacterium species assignment based on species-specific allelic variation in the 65-kilodalton heat shock protein gene (hsp65). Arch. Pathol. Lab. Med. 121:859-864.
Patel, J. B., R. J. Wallace, Jr., B. A. Brown-Elliott, T. Taylor, C. Imperatrice, D. G. B. Leonard, R. W. Wilson, L. Mann, K. C. Jost, and I. Nachamkin. 2004. Sequence-based identification of aerobic actinomycetes. J. Clin. Microbiol. 42:2530-2540.
Ringuet, H., C. Akoua-Koffi, S. Honore, A. Varnerot, V. Vincent, P. Berche, J. L. Gaillard, and C. Pierre-Audigier. 1999. hsp65 sequencing for identification of rapidly growing mycobacteria. J. Clin. Microbiol. 37:852-857.
Sanguinetti, M., B. Posteraro, F. Ardito, S. Zanetti, A. Cingolani, L. Sechi, A. De Luca, L. Ortona, and G. Fadda. 1998. Routine use of PCR-reverse cross-blot hybridization assay for rapid identification of Mycobacterium species growing in liquid media. J. Clin. Microbiol. 36:1530-1533.
Steingrube, V. A., J. L. Gibson, B. A. Brown, Y. Zhang, R. W. Wilson, M. Rajagonpalan, and R. J. Wallace, Jr. 1995. PCR amplification and restriction endonuclease analysis of a 65-kilodalton heat shock protein gene sequence for taxonomic separation of rapidly growing mycobacteria. J. Clin. Microbiol. 33:149-153.
Taylor, T. B., C. Patterson, Y. Hale, and W. W. Safranek. 1997. Routine use of PCR-restriction fragment length polymorphism analysis for identification of mycobacteria growing in liquid media. J. Clin. Microbiol. 35:79-85.
Tortoli, E., L. Rindi, M. J. Garcia, P. Chiaradonna, R. Dei, C. Garzelli, M. Kroppenstedt, N. Lari, R. Mattei, A. Mariottini, G. Mazzarelli, M. I. Murcia, A. Nanetti, P. Piccoli, and C. Scarparo. 2004. Proposal to elevate the genetic variant MAC-A, included in the Mycobacterium avium complex, to species rank as Mycobacterium chimaera sp. nov. Int. J. Syst. Evol. Microbiol. 54:1277-1285.
Vincent, V., B. A. Brown-Elliott, K. C. Jost, Jr., and R. J. Wallace, Jr. 2003. Mycobacterium: phenotypic and genotypic identification, p. 560-584. In P. R. Murray, E. J. Baron, J. H. Jorgensen, M. A. Pfaller, and R. H. Yolken (ed.), Manual of clinical microbiology, vol. 1, 8th ed. ASM Press, Washington, D.C.
Woo, P. C. Y., A. H. Y. Ngan, S. K. P. Lau, and K. Yuen. 2003. Tsukamurella conjunctivitis: a novel clinical syndrome. J. Clin. Microbiol. 41:3368-3371.
Zeller, V., A. Nardi, C. Truffot-Pernot, W. Sougakoff, B. Stankoff, C. Katlama, and F. Bricaire. 2003. Disseminated infection with a Mycobacterium related to Mycobacterium triplex with central nervous system involvement associated with AIDS. J. Clin. Microbiol. 41:2785-2787.(Alan McNabb, Kathy Adie, )
Department of Pathology and Laboratory Medicine, University of British Columbia, Vancouver, British Columbia, Canada
ABSTRACT
We investigated extending the use of direct partial hsp65 gene sequencing for the identification of mycobacteria to isolates in primary liquid detection media as an economical, feasible, and more rapid means of identification. During the course of the study, the hsp65 sequence-based identifications for isolates from 670 primary liquid detection media determined to be positive for acid-fast bacilli were compared to the identifications derived from Accuprobes, biochemical test panels, or 16S rRNA gene sequencing. Preliminary analysis indicated a 97.6% concordance, with a final agreement of 99.1% between the identification algorithms. hsp65 sequencing costs (US$32.84) were greater than the cost of identification with Accuprobe (US$19) but less than the cost of the biochemical test panel identification (average cost, US$98.90) and equivalent to the cost of 16S rRNA sequencing, although there was a referral cost (US$59.85) for the shipping of isolates to another reference laboratory. Analysis indicated that our laboratory would have recognized a cost savings of approximately $12,000 by using hsp65 sequencing to identify isolates from specimens with a negative fluorescent- smear status and would have achieved further savings by using it as an alternative to biochemical panel testing for fluorescent-smear-positive specimens. The time to identification by hsp65 gene sequencing was slightly longer than that required by the Accuprobe assay (1 versus 2 days), shorter than that required by the biochemical test panels (2 days versus 26 days on average), and more rapid than referral for 16S rRNA gene sequencing.
INTRODUCTION
The identification of mycobacteria has traditionally been accomplished by determining their ability to utilize particular compounds, their growth characteristics, and their colonial morphologies. This testing is performed with isolates derived from primary cultures or subcultures from primary liquid detection media (PLDM), such as Bac T Alert 3D bottles (bioMerieux, Durham, N.C.), BACTEC 12B bottles (Becton Dickinson Diagnostic Instrument Systems, Sparks, Md.), MGIT bottles (Becton Dickinson Diagnostic Instrument Systems), or Myco/F bottles (Becton Dickinson Diagnostic Instrument Systems) (10, 21). Regardless of the source, all isolates must be incubated for sufficient growth to occur, which, depending on the organism, may take several days to several weeks. This means of identification requires expertise and is time-consuming, expensive, and labor-intensive, but it is still used today by many mycobacteriology laboratories to identify at least some Mycobacterium species.
The introduction of the Accuprobe system (GenProbe, San Diego, Calif.) greatly accelerated the identification of some mycobacteria, as testing could be performed directly from PLDM. Unfortunately, probes specific for only four species and two complexes have been developed. Several "home-brew" restriction enzyme analysis methods have been developed as a means for the rapid identification of mycobacteria from both solid and liquid culture media, but they are not without problems (3-9, 11, 14, 16, 17, 18, 19, 23). More recently, a commercially available line probe assay (Inno-Lipa Mycobacteria; Innogenetics, Ghent, Belgium) has become available for the identification of Mycobacterium species. This assay identifies a larger number of species than Accuprobe tests (16 species versus 4 species and two complexes) and has the advantage of testing for all species within its database at one time (13). Previously, we reported on the use of partial hsp65 gene sequencing as a means for mycobacterial identification and found it to be an extremely rapid identification method but required isolates from solid culture media, which lengthened the time from organism detection in primary liquid detection media until the final identification (12). Therefore, we explored the utility of hsp65 gene sequencing directly from primary liquid detection media determined to be positive for acid-fast bacilli as an alternative and cost-effective means of identification of mycobacteria. We report here on the results of that study.
MATERIALS AND METHODS
Mycobacterial strains. A total of 670 bottles of primary liquid detection media (BACTEC 5, MGIT 96, Myco/F 17, and Bac T Alert 3D 552 bottles) not included in our previous study and determined to be positive for the presence of acid-fast rods were investigated (12). Conventional identification of the isolates by the use of our current identification algorithm spanned 37 Mycobacterium species and taxonomic groups and unique species, as well as Nocardia and Tsukamurella species (Table 1).
Conventional identification. Mycobacterial isolates were identified directly from fluorescent-smear-positive specimens as Mycobacterium tuberculosis complex by using the Amplified Mycobacterium tuberculosis Direct test (AMTD; GenProbe, San Diego, Calif.) and confirmed by an M. tuberculosis complex-specific Accuprobe assay from PLDM or were identified from PLDM known to be positive for acid-fast organisms by using M. tuberculosis complex-, Mycobacterium gordonae-, or Mycobacterium avium-Mycobacterium intracellulare complex-specific Accuprobes (GenProbe). These identifications were confirmed by colonial morphology, as in the case of M. gordonae and the M. avium-M. intracellulare complex, or by nitrate and niacin testing for the M. tuberculosis complex. When organisms in PLDM were Accuprobe negative or not tested, aliquots were subcultured to appropriate solid culture media and the isolates were identified by using further Accuprobes or biochemical test panels or, when it was required, by partial or complete 16S rRNA gene sequencing, as described previously (12).
DNA extraction for hsp65 gene sequencing. For all PLDM confirmed to contain acid-fast organisms, a 1.5-ml aliquot was transferred to a microcentrifuge tube and the DNA was extracted as published previously (2, 12).
DNA amplification and partial hsp65 gene sequencing. Partial amplification and hsp65 gene sequencing, identification of all isolates, as well as PCR and sequencing of controls were carried out as described previously (12).
Determination and identification of isolates from mixed sequences. The optimization of the cycle sequence reactions during our previous study ensured that the signal-to-noise ratio was very high and that the presence of a second signal predicted to exceed 20% of the major signal could be detected. All sequences were examined for the presence of subordinate base peaks (2, 12). When subordinate peaks occurred, two sequences were generated and subjected to BLAST searches. One consisted of the sequence derived from all major base call peaks, and the second sequence was derived by substituting the minor base calls for the major peaks where they occurred, which we speculated should represent the major and minor organisms present, respectively.
Mycobacterial hsp65 sequence database. Our previous database, which consisted of 140 hsp65 sequence entries (111 valid and putative species and groups plus 29 distinct sequences), was supplemented with sequence entries downloaded from the GenBank database, further unique sequences detected in our laboratory, and type strains of Nocardia and Tsukamurella species: M. avium complex group B8; Mycobacterium chimaera (GenBank accession number AJ548480); Mycobacterium cosmeticum (GenBank accession number AY449731); Mycobacterium florentinum (GenBank accession number AJ616230); Mycobacterium massilense (GenBank accession number AY596465); Mycobacterium parascrofulaceum sequevars I, II, III, IV, and V (GenBank accession numbers AY337274, AY337275, AY337276, AY337277, and AY337278, respectively); Mycobacterium saskatchewanense (GenBank accession number AY208859), Mycobacterium shinshuense ATCC 33728T, Mycobacterium species strain MCRO 24; Mycobacterium vanbaalenii (GenBank accession number AY438091); Nocardia asteroides ATCC 19247T; Nocardia brasiliensis ATCC 19296T; Nocardia carnea ATCC 6847T; Nocardia farcinica ATCC 3318T; Nocardia nova ATCC 33726T; Nocardia otitidiscaviarum ATCC 14629T; Nocardia transvalensis ATCC 6865T; Tsukamurella inchonensis DSM 44067T; Tsukamurella paurometabola DSM 20162T; Tsukamurella pulmonis DSM 44142T; Tsukamurella spumae DSM 44113T; Tsukamurella strandjordii DSM 44573T; and Tsukamurella tyrosinosolvens DSM 44234T. The database was also updated during the course of the study with Mycobacterium kansasii types 2, 3, 4, and 5 (GenBank accession numbers AY438086, AY438087, AY438088, and AY438089, respectively).
Cost analysis. The costs associated with facility space (lighting and heating or air conditioning), primary specimen examination (stained smears and solid and/or liquid media), sequencer acquisition, and depreciation were not included in the overall cost calculations. The overall costs of identification of isolates identified by the AMTD test, Accuprobes, biochemical test panels, or 16S rRNA sequencing were determined by using a subset of 158 positive specimens. These samples were allocated into one of two groups: samples with a positive direct acid-fast smear result (n = 39) or samples with a negative direct acid-fast smear result (n = 119). An AMTD test was performed for samples from the former category, based on a decision by a technologist following a prescribed identification algorithm. If the AMTD test detected M. tuberculosis complex in the patient's specimen, the identification was confirmed with an M. tuberculosis complex-specific Accuprobe when the PLDM became positive, and the species of the isolate was determined from solid media by using nitrate and niacin tests. Specimens that had a negative AMTD test result (n = 19) or for which the test was not performed (n = 119) were tested by using further Accuprobes when the corresponding PLDM was determined to be positive for acid-fast bacilli. All Accuprobe-negative organisms were subjected to biochemical test panel identification, and those not identified were referred to the Canadian National Reference Centre for Mycobacteriology (NRCM), National Microbiology Laboratory, Winnipeg, Manitoba, Canada, for partial 16S rRNA gene sequencing. Cost analysis of all tests, which were based on a conversion rate of 0.73 Canadian dollars to 1.00 U.S. dollar, included technologist time and all consumables and were calculated to be as follows (all values are in U.S. dollars): AMTD, $19; Accuprobe, $19; biochemical test panels for slowly growing biochemically active and inactive organisms, $97.63; biochemical test panels for rapidly growing organisms, $102.12; processing and shipping to NRCM, $59.85; and hsp65 sequencing, $32.84. The cost of partial 16S rRNA gene sequencing was not included as a cost to our facility.
RESULTS
Mycobacterium species were the only organisms present in 613 of 670 (91.5%) of the PLDM indicated to contain acid-fast organisms (Table 1). The remaining PLDM (Table 1) contained organisms other than Mycobacterium species (19 PLDM), were mixed cultures (22 PLDM), or failed to grow any acid-fast organism (16 PLDM). We were unable to amplify the hsp65 gene from any of the last set of cultures but were able to amplify the hsp65 gene from all other samples, with a single exception, and confirmed the identities of all M. tuberculosis complex isolates, all M. gordonae isolates, and 316 of 318 M. avium-M. intracellulare complex isolates, as well as all isolates from 25 other species (Table 1) (12). Mixed cultures represented 3.3% of all PLDM examined, with 12 of 22 (54.5%) detected by direct hsp65 sequencing and 10 of 22 detected only upon culture to solid media (Table 2). We confirmed the identities of 43 isolates from the 22 mixed cultures directly from the PLDM or from solid culture media, bringing our overall identification agreement to 99.1% (650 of 656).
Cost analysis. A subset of 158 specimens for which the results of all tests were known was used to develop a "per identification cost," which was later applied to all samples in the study (Table 3). The subset was divided into a direct fluorescent-smear-positive group (group 1) and a direct fluorescent-smear-negative group (group 2). Group 1 consisted of 39 samples, with 20 having a positive AMTD test result and 19 having a negative AMTD test result. This represented a total of 39 tests for the identification of 20 isolates at a cost to our laboratory of $37.05 per direct identification (39 tests x $19) of M. tuberculosis complex. Of the 20 AMTD test-positive specimens, all had a corresponding positive PLDM result and the M. tuberculosis complex identification was confirmed by the Accuprobe assay. Two AMTD test-positive PLDM were also tested by an M. avium-M. intracellulare complex-specific Accuprobe assay, but both were negative. We estimated our cost to be $20.90 per confirmed identification (22 tests x $19 per test). Of the 19 AMTD test-negative cultures, 14 were identified as M. avium-M. intracellulare complex and 1 was identified as M. gordonae by the Accuprobe assay; but 5 negative Accuprobe tests were performed, and we estimated the cost to be $20.00 per identification (20 tests x $19 per test). Four cultures did not react with Accuprobes and were identified by in-house biochemical test panels at $102.12 per isolate. The total cost required to confirm the identification of M. tuberculosis complex or to identify other species detected by PLDM from fluorescent-smear-positive specimens was estimated to be $30.90 per isolate. The cost required to identify isolates by hsp65 sequencing remained constant at $32.84 per identification. Of the 119 organisms allocated to group 2, 87 organisms were identified by the Accuprobe assay. In addition, 53 tests gave negative results, for a total of 140 Accuprobe tests performed to identify 87 isolates, with an estimated identification cost of $30.57 per isolate (140 tests x $19 per test). The 32 isolates not identified by Accuprobe tests were subjected to identification by biochemical test panels. Eighteen isolates were rapidly growing mycobacteria (identification cost, $102.12), and 14 were slowly growing mycobacteria (identification cost, $97.63). Twenty-two isolates identified at an average cost of approximately $145.68 per identification and the 10 remaining isolates required 16S rRNA gene sequencing, with an incurred shipping cost of $59.85 per shipment. We estimate that the identification costs for group 2 isolates were approximately $51.29 per isolate by our current algorithm and $32.84 per identification by hsp65 sequencing.
DISCUSSION
A recent study conducted by our laboratory demonstrated the validity of using partial hsp65 sequencing from solid culture media as a rapid means for the identification of Mycobacterium species (12). In an attempt to further decrease the time required to identify isolates and reduce costs, we investigated the use of an aliquot of PLDM as a source of mycobacterial DNA (12). To date, we have analyzed 656 isolates from 670 PLDM and determined the correlation between our current identification algorithm and hsp65 sequencing to be 99.1%. Similar to our previous study, Mycobacterium species capable of being identified by Accuprobe as M. avium-M. intracellulare complex (n = 329), M. gordonae (n = 82), or M. tuberculosis complex (n = 73) constituted approximately 70% of the isolates detected in our cultures. As noted previously, most identification discrepancies were limited to a few species (12). Of interest in this study was the fact that examination of 34 M. abscessus isolates revealed that 9 were identical to our database entry for M. abscessus variant VM585. These nine isolates had a 98.8% sequence match (396 of 401 nucleotides) with M. abscessus ATCC 19977T and a 100% sequence match with the newly proposed species M. massiliense (1). These organisms were single isolates from respiratory specimens from eight different patients, indicating that M. massiliense is common in our patient population (26.4% of all M. abscessus isolates), but unlike the isolate used to describe the new species (1), they may not have had a pathogenic role. As well, examination of 79 isolates that we identified as M. intracellulare indicated that a group of 9 isolates from different patients had identical hsp65 sequences and were only 99.0% related to the M. intracellulare type strain (unpublished data) but were identical to the recently published species M. chimaera (20). Full 16S rRNA gene sequencing of one of our isolates indicated that it was identical to the M. chimaera type strain (GenBank accession number AJ548480), and we assumed that our other eight isolates were identical. This indicates that M. chimaera was relatively common in our patient population, representing 2.8% of all our M. avium-M. intracellulare complex isolates and 10.5% of our purported M. intracellulare isolates.
Our cost for the identification of isolates by use of our standard identification algorithm varied from a low of $20.00 to confirm the results for direct fluorescent-smear-positive, AMTD test-positive, and PLDM-positive samples to in excess of $145.68 to confirm the results for isolates that were direct fluorescent-smear negative, Accuprobe negative, and not identified by biochemical test panels and that required 16S rRNA gene sequencing. By our identification algorithm, 167 isolates were not identified by Accuprobes and required biochemical identification. These Mycobacterium species, as well as Nocardia and Tsukamurella species, constituted 25.5% of the isolates in our study (Table 1). Ninety-seven isolates met the identification criteria for in-house biochemical testing, and 70 isolates were referred to NRCM for identification or confirmation of the identity in 31 shipments, at a total cost to our facility of $1,810. Of the 70 isolates submitted to NRCM, 18 isolates (25.7%) were identified as Tsukamurella species, a genus not included in our biochemical test panel database. We had encountered Tsukamurella species in our previous study, and to aid in their identification we had included sequences from six Tsukamurella species type strains in our most recent hsp65 sequence database. However, examination of the hsp65 sequence divergence for these species (unpublished data) indicated that, like the partial 16S rRNA gene sequence, hsp65 sequences may not be sufficiently divergent to report the results for isolates to the species level or that the current taxonomy is not adequate and we reported them only as Tsukamurella species, a cost-effective alternative to the use of our primary identification algorithm (15, 22).
Mixed cultures represented 3.3% (22 of 670) of the PLDM indicated to contain acid-fast bacilli. Twelve mixed cultures were detected by direct hsp65 sequencing, and the remainder were detected only upon subculture to solid media. Since the presence of mixed cultures was immediately reported to our Mycobacteriology Laboratory at the British Columbia Centre for Disease Control, we are not able to determine the impact that these results had on the solid medium culture report. Two incidences occurred where M. tuberculosis complex isolates in mixed culture were detected only upon subculture of positive PLDM to solid culture media. We had postulated that for cost containment, hsp65 sequencing would allow us to forgo subculture of PLDM believed to contain organisms of dubious pathogenicities, but these results indicate that all positive PLDM must be subcultured to substantiate the presence of a single Mycobacterium species.
PLDM submitted during this study contained 656 isolates that spanned 37 Mycobacterium species or taxonomic groups and 5 unique Mycobacterium species (as determined by partial 16S rRNA gene sequencing), as well as Nocardia and Tsukamurella species (Table 1). Mycobacterium species can be identified by Accuprobes for $19 per identification, under ideal conditions. Analysis of a subset of our PLDM indicates that this does not always occur and that the cost per identification within this group varied. One hundred seventy isolates not able to be identified by Accuprobes required identification by biochemical test panels at an average cost of approximately $98.00 per attempt. Only 97 of these total isolates were identified by in-house biochemical test panels, although some were identified incorrectly, at an approximate total cost of $10,000. The same isolates were identified by hsp65 sequencing at a total approximate cost of $3,250. The identification of 70 isolates sent to NRCM for confirmation or not identified by biochemical test panels and referred for 16S rRNA sequencing was estimated to cost our facility $7,400, as opposed to an identification cost of $2,300 by hsp65 gene sequencing. This equates as a total savings to our laboratory of approximately $12,000 for the hsp65 identification of Accuprobe-negative isolates, as well as a significant time savings per identification. Our latest study supports the results of our previous study and has proven the usefulness and cost-efficiency of hsp65 gene sequencing directly from primary liquid detection media as a cost-effective means for the identification of mycobacteria (12).
ACKNOWLEDGMENTS
We thank Nicola DiTomaso, Mycobacteriology Laboratory, British Columbia Centre for Disease Control, for excellent technical assistance; all members of the Mycobacteriology Laboratory at the British Columbia Centre for Disease Control for their assistance and cooperation during the course of this study; as well as Joyce Wolfe and the staff at the National Reference Centre for Mycobacteriology for 16S rRNA gene sequencing.
REFERENCES
Adekambi, T., M. Reynaud-Gaubert, G. Greub, M. J. Gevaudan, B. La Scola, D. Raoult, and M. Drancourt. 2004. Amoebal coculture of "Mycobacterium massilense" sp. nov. from the sputum of a patient with hemoptoic pneumonia. J. Clin. Microbiol. 42:5493-5501.
Applied Biosystems Inc. 2001. Users manual, ABI Prism 3100 genetic analyzer. Applied Biosystems Inc., Foster City, Calif.
Beggs, M. L., R. Stevanova, and K. D. Eisenach. 2000. Species identification of Mycobacterium avium complex isolates by a variety of molecular techniques. J. Clin. Microbiol. 38:508-512.
Brunello, F., M. Ligozzi, E. Cristelli, S. Bonora, E. Tortoli, and R. Fontana. 2001. Identification of 54 mycobacterial species by PCR-restriction fragment length polymorphism analysis of the hsp65 gene. J. Clin. Microbiol. 39:2799-2806.
Da Silva Rocha, A., C. da Costa Leite, H. Magarinos Torres, A. B. de Miranda, M. Q. Pires Lopes, W. M. Degrave, and P. N. Suffys. 1999. Use of PCR-restriction fragment length polymorphism analysis of the hsp65 gene for rapid identification of mycobacteria in Brazil. J. Microbiol. Methods 37:223-229.
de Magalhaes, V. D., P. de Melo Azevedo Fda, J. Pasternak, and M. D. Valle Martino. 2002. Reliability of hsp65-RFLP analysis for identification of Mycobacterium species in cultured strains and clinical specimens. J. Microbiol. Methods 49:295-300.
Devallois, A., K. S. Goh, and N. Rastogi. 1997. Rapid identification of mycobacteria to species level by PCR-restriction fragment length polymorphism analysis of the hsp65 gene and proposition of an algorithm to differentiate 34 mycobacterial species. J. Clin. Microbiol. 35:2969-2973.
Ergin, A., T. Kocagoz, and D. Us. 2000. Evaluation of 120 mycobacterial strains isolated from clinical specimens to the species level by polymerase chain reaction-restriction enzyme analysis. Scand. J. Infect. Dis. 32:657-662.
Hernandez, M. S., G. P. Morlock, W. R. Butler, J. T. Crawford, and R. C. Cooksey. 1999. Identification of Mycobacterium species by PCR-restriction fragment length polymorphism analyses using fluorescence capillary electrophoresis. J. Clin. Microbiol. 37:3688-3692.
Hosojima, A. C. S., Y. Fukasawa, Y. Kazumi, M. Takahashi, K. Hirano, and T. Mori. 1992. Comparison of MB-Check, Bactec, and egg-based media for recovery of mycobacteria. J. Clin. Microbiol. 30:878-881.
Levi, M. H., J. Bartell, L. Gandolfo, S. C. Smole, S. F. Costa, L. M. Weiss, L. K. Johnson, G. Osterhout, and L. H. Herbst. 2003. Characterization of Mycobacterium montefiorense sp. nov., a novel pathogenic mycobacterium from moray eels that is related to Mycobacterium triplex. J. Clin. Microbiol. 41:2147-2152.
McNabb, A., D. Eisler, K. Adie, M. Amos, M. Rodrigues, G. Stephens, W. A. Black, and J. Isaac-Renton. 2004. Assessment of partial sequencing of the 65-kilodalton heat shock protein gene (hsp65) for routine identification of Mycobacterium species isolated from clinical sources. J. Clin. Microbiol. 42:3000-3011.
Padilla, E., V. Gonzalez, J. M. Manterola, A. Perez, M. D. Quesada, S. Gordillo, C. Vilaplana, M. A. Pallares, S. Molinos, M. D. Sanchez, and V. Ausina. 2004. Comparative evaluation of the new version of the Inno-Lipa mycobacteria and genotype mycobacterium assays for identification of Mycobacterium species from MB/BacT liquid cultures artificially inoculated with mycobacterial strains. J. Clin. Microbiol. 42:3083-3088.
Pai, S., N. Esen, X. Pan, and J. M. Musser. 1997. Routine rapid Mycobacterium species assignment based on species-specific allelic variation in the 65-kilodalton heat shock protein gene (hsp65). Arch. Pathol. Lab. Med. 121:859-864.
Patel, J. B., R. J. Wallace, Jr., B. A. Brown-Elliott, T. Taylor, C. Imperatrice, D. G. B. Leonard, R. W. Wilson, L. Mann, K. C. Jost, and I. Nachamkin. 2004. Sequence-based identification of aerobic actinomycetes. J. Clin. Microbiol. 42:2530-2540.
Ringuet, H., C. Akoua-Koffi, S. Honore, A. Varnerot, V. Vincent, P. Berche, J. L. Gaillard, and C. Pierre-Audigier. 1999. hsp65 sequencing for identification of rapidly growing mycobacteria. J. Clin. Microbiol. 37:852-857.
Sanguinetti, M., B. Posteraro, F. Ardito, S. Zanetti, A. Cingolani, L. Sechi, A. De Luca, L. Ortona, and G. Fadda. 1998. Routine use of PCR-reverse cross-blot hybridization assay for rapid identification of Mycobacterium species growing in liquid media. J. Clin. Microbiol. 36:1530-1533.
Steingrube, V. A., J. L. Gibson, B. A. Brown, Y. Zhang, R. W. Wilson, M. Rajagonpalan, and R. J. Wallace, Jr. 1995. PCR amplification and restriction endonuclease analysis of a 65-kilodalton heat shock protein gene sequence for taxonomic separation of rapidly growing mycobacteria. J. Clin. Microbiol. 33:149-153.
Taylor, T. B., C. Patterson, Y. Hale, and W. W. Safranek. 1997. Routine use of PCR-restriction fragment length polymorphism analysis for identification of mycobacteria growing in liquid media. J. Clin. Microbiol. 35:79-85.
Tortoli, E., L. Rindi, M. J. Garcia, P. Chiaradonna, R. Dei, C. Garzelli, M. Kroppenstedt, N. Lari, R. Mattei, A. Mariottini, G. Mazzarelli, M. I. Murcia, A. Nanetti, P. Piccoli, and C. Scarparo. 2004. Proposal to elevate the genetic variant MAC-A, included in the Mycobacterium avium complex, to species rank as Mycobacterium chimaera sp. nov. Int. J. Syst. Evol. Microbiol. 54:1277-1285.
Vincent, V., B. A. Brown-Elliott, K. C. Jost, Jr., and R. J. Wallace, Jr. 2003. Mycobacterium: phenotypic and genotypic identification, p. 560-584. In P. R. Murray, E. J. Baron, J. H. Jorgensen, M. A. Pfaller, and R. H. Yolken (ed.), Manual of clinical microbiology, vol. 1, 8th ed. ASM Press, Washington, D.C.
Woo, P. C. Y., A. H. Y. Ngan, S. K. P. Lau, and K. Yuen. 2003. Tsukamurella conjunctivitis: a novel clinical syndrome. J. Clin. Microbiol. 41:3368-3371.
Zeller, V., A. Nardi, C. Truffot-Pernot, W. Sougakoff, B. Stankoff, C. Katlama, and F. Bricaire. 2003. Disseminated infection with a Mycobacterium related to Mycobacterium triplex with central nervous system involvement associated with AIDS. J. Clin. Microbiol. 41:2785-2787.(Alan McNabb, Kathy Adie, )