INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The rapidly growing pathogenic mycobacteria Mycobacterium abscessus, Mycobacterium chelonae, and Mycobacterium fortuitum (and related species) cause several forms of clinical disease of varying severity, most commonly skin and soft tissue infections but also skeletal, pulmonary, and disseminated disease (1, 5-7, 15, 16, 19, 20). Data from several studies have shown that these species vary in susceptibility to antimicrobial agents useful for therapy (1, 2, 4, 5, 11-15, 17, 20). For this reason, antimicrobial susceptibility testing of isolates considered clinically significant is recommended. Various methods of testing susceptibility of the rapidly growing mycobacteria have been described, including agar disk elution, broth microdilution, and the E-test (2, 3, 11, 17). Currently, however, no standardized testing method for this group of organisms exists, nor has the interlaboratory reproducibility of any method been assessed. The primary goal of the present multicenter study was to evaluate the broth microdilution method for its ability to provide reproducible MIC endpoints and interpretive categories in several laboratories with different levels of experience with susceptibility testing of rapidly growing mycobacteria. A secondary goal was to identify a clinical isolate of one of these rapidly growing mycobacteria that would be an acceptable quality control organism for the microdilution test.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Organisms. Ten clinical isolates, previously studied at the University of Texas Health Center in Tyler, were selected for testing. Of the four M. fortuitum group isolates (three M. fortuitum and one Mycobacterium peregrinum), one (1353) was chosen because of its susceptibility to tetracyclines and the low clarithromycin MIC for it, a second (1359) was chosen because of the low clarithromycin MIC for it, a third (1351) was chosen because the clarithromycin endpoint was indeterminate (i.e., trailing), and a fourth (1352) was chosen because the clarithromycin MIC for it was high. Of the three M. chelonae isolates, one (1866) was chosen because it had mutational resistance to clarithromycin (18) and one (1814) was chosen because it was susceptible to tetracyclines but only moderately susceptible to tobramycin. Of the three M. abscessus isolates, one (1802) was chosen because it had mutational resistance to clarithromycin (18). The remaining isolates were selected because the MICs for them were typical of their species based on previous broth MIC results. Isolates on trypticase soy agar slants were mailed from the University of Texas Health Center at Tyler to the other three participating sites, where they were maintained at room temperature until tested.

Antimicrobial agents and microdilution trays. The antimicrobial agents evaluated were amikacin, cefoxitin, ciprofloxacin, clarithromycin, doxycycline, imipenem, sulfamethoxazole, and tobramycin (against M. chelonae only). A single lot of dried and sealed MIC trays containing twofold serial dilutions of each drug was provided by Trek Diagnostics (formerly AccuMed International, Inc., Westlake, Ohio). The final concentration ranges were 1 to 128 µg/ml for amikacin, 2 to 256 µg/ml for cefoxitin, 0.125 to 16 µg/ml for ciprofloxacin, 0.03 to 64 µg/ml for clarithromycin, 0.25 to 32 µg/ml for doxycycline, 1 to 64 µg/ml for imipenem and sulfamethoxazole, and 1 to 16 µg/ml for tobramycin. Each tray also contained a positive growth control well. The trays were stored at ambient temperature until they were used in the study.

Inoculum preparation. Each isolate was subcultured once onto a common lot of sheep blood agar plates provided by Remel (Lenexa, Kans.) and incubated in ambient air at 30°C for 72 h. Inocula were prepared by swabbing the confluent portion of growth on the blood agar plate with a sterile cotton swab. Growth on the swab was transferred to a tube containing 4.5 ml of sterile water and glass beads (Trek Diagnostic Systems), and the turbidity was adjusted until it matched that of a 0.5 McFarland standard by visual examination or by using a nephelometer. The growth suspensions were mixed vigorously on a vortex mixer for 15 to 20 s. The final inoculum (approximately 5 × 10⁵ CFU/ml) was prepared by transferring 50 µl of the suspension to a tube containing 10 ml of cation-adjusted Mueller-Hinton broth (Trek Diagnostic Systems) and inverting the tube 8 to 10 times prior to use.

Susceptibility test method. Broth microdilution MIC testing was performed within 30 min after final inoculum preparation as described by Brown et al. (3). Final inoculum suspensions were poured into plastic troughs (Trek Diagnostic Systems), and 100-µl aliquots were transferred to each well of the MIC tray with a multichannel pipettor. The inoculated trays were covered with an adhesive seal and incubated at 30°C in ambient air. A blood agar plate was also inoculated with a loopful of the final inoculum to check for purity. The trays were first examined after 72 h. If growth (appearing as turbidity or a deposit of cells at the bottom of the well) in the growth control well was sufficient (i.e., at least 2+, based on the following scale: ± to 1+ growth, a few flecks in the bottom of the well; 2+, moderate growth for the particular species in the well; and 3+ to 4+, a readily visible button in the bottom of the well), the MICs were recorded. Otherwise, the trays were reincubated and read daily thereafter (for up to 5 days) until moderate growth was visible. For all but sulfamethoxazole, the MIC was recorded as the lowest concentration of a drug that inhibited visible growth. For sulfamethoxazole, the endpoint or MIC was defined as the concentration of the drug in the well with approximately 80% inhibition of growth compared to the growth in the control well with no drug. Susceptible and resistance breakpoints are listed in Table 1 (3).

Quality control. Staphylococcus aureus ATCC 29213 and Pseudomonas aeruginosa ATCC 27853 were tested at each site at the beginning of the study. Quality control was considered acceptable if the results were within ranges recommended by the National Committee for Clinical Laboratory Standards (NCCLS) (8a).

Study design and analysis. Four laboratories participated in the study; one had extensive experience with susceptibility testing of rapidly growing mycobacteria (site A), one had some experience (site B), and two (sites C and D) had no experience. The testing personnel at site D also had very limited experience with the microdilution method in general. Before testing was begun, personnel at site A reviewed the MIC procedure and interpretation of the results with the other sites via conference call to ensure standardization of the protocol. During the study, testing personnel at sites B, C, and D consulted personnel at site A if questions about the test procedure or interpretation arose.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

The day on which the MICs were considered interpretable differed among the sites. At site A, which had the most experience with testing the rapidly growing mycobacteria, all MICs were read on day 3, and at site B, all were read on day 4. At site C, most MICs were read on day 4 but a few were read on day 5, and at site D, about half were read on day 4, most of the other half on day 5, and a few on day 3.

In general, the MICs for the organisms tested in this study were similar to those previously reported by other investigators (Table 2) (4, 9, 11-15). Tables 3 through 5 summarize the microdilution MIC results of the seven antimicrobial agents tested for M. abscessus and M. fortuitum and the eight drugs tested for M. chelonae and the percent agreement among the four participating laboratories. Agreement varied considerably for the different isolate-drug combinations. Overall, agreement was best for cefoxitin, ranging from 97.2 to 100% for all isolates except M. fortuitum 1359, for which there was 91.7% agreement. Agreement also was excellent for doxycycline (97.2 to 100%) for all isolates except M. chelonae 1814, for which agreement was 61.1%. For amikacin and ciprofloxacin, agreement was good to excellent for isolates of the M. fortuitum group (100% for amikacin and 97.2 to 100% for ciprofloxacin) and M. abscessus (amikacin, 97.2 to 100%; ciprofloxacin, 91.7 to 100%) but was much lower for the isolates of M. chelonae (66.7 to 97.2% for amikacin and 66.7 to 94.4% for ciprofloxacin). For tobramycin, agreement was excellent (94.4 to 100%) for all isolates of M. chelonae except one (77.8% agreement). Agreement was excellent for sulfamethoxazole and M. abscessus (97.2 to 100%) and good for M. chelonae (91.7 to 97.2%) but was poor, ranging from 41.7 to 86.1%, for isolates of the M. fortuitum group. For clarithromycin, agreement was excellent (97.2 to 100%) for two isolates each of M. abscessus and M. chelonae, including the isolates (1802 and 1866) with known mutational resistance to the drug, but was much lower for all isolates of the M. fortuitum group (27.8 to 88.9% agreement). For all other isolate-drug combinations, agreement varied widely, and in over half of the cases, it was less than 80%.

To assess the potential impact of the variability in MIC results on patient management, we also evaluated percent agreement based on the interpretive category (Table 6). Again, agreement varied considerably, but the results were different from those based on MIC values. Agreement was excellent for doxycycline (97.2 to 100%); it was >94% for ciprofloxacin and 9 of the 10 isolates and was 100% for clarithromycin and 8 isolates. For the aminoglycosides, agreement was 100% for the M. fortuitum group but varied from 50 to 100% for M. abscessus and M. chelonae. Agreement was lowest with imipenem.

Further analysis of the data revealed some possible reasons for the broad MIC ranges, poor agreement by interpretive category, or both for certain isolate-drug combinations. For sulfamethoxazole, MICs that were lower (isolates 1801 and 1814) and higher (isolates 1353 and 1359) than the mode, causing wide MIC ranges and poor agreement by interpretive category, were reported by site D, where the testing personnel not only had no experience with the rapidly growing mycobacteria but also had limited experience with broth microdilution testing in general. Findings with clarithromycin were similar. The lack of reproducibility of MIC values for several isolates (i.e., 1801, 1831, 1353, and 1359) was due to very low MICs (i.e., 0.03 or 0.06 µg/ml compared to the modal MIC [Tables 3 to 5]) reported by site D. For isolates 1801 and 1353 the range was made even broader due to MICs higher than the modal MIC reported by site B (i.e., 4 µg/ml for isolate 1801 for all nine testing events). Another problem with clarithromycin occurred with two isolates of M. fortuitum that had trailing endpoints, similar to that observed when testing sulfonamides. This phenomenon caused difficulty in interpretation at all sites.

The problem of trailing endpoints was observed with ciprofloxacin only against isolates of M. chelonae. The lack of reproducibility (isolates 1814 and 1831) and the lower agreement by interpretive category (isolate 1831) were primarily due to reports from site D, suggesting lack of familiarity with the growth pattern as the cause of the problem. For both isolates, site D reported MICs lower than the mode (i.e., 2 and 4 µg/ml compared to >16 µg/ml for isolate 1814 and 2 µg/ml for six testing events compared to 8 µg/ml for isolate 1831).

For two isolates of the M. fortuitum group and one isolate of M. abscessus, cefoxitin MICs clustered at 32 and 64 µg/ml, and this twofold dilution variability caused a difference in interpretation: currently, 32 µg/ml is considered intermediate whereas 64 µg/ml is considered resistant (3). This same clustering at the breakpoint between intermediate and resistant was responsible for the poor agreement by interpretive category for tobramycin and one isolate of M. chelonae and for amikacin and one isolate of M. abscessus and two isolates of M. chelonae.

Suggestions for susceptibility testing of M. abscessus, M. chelonae, and the M. fortuitum group, based on the results of this study, are outlined in Table 7.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Susceptibility testing of clinically significant isolates of the rapidly growing mycobacteria is recommended because these organisms differ in susceptibility to the antimicrobial agents commonly used for therapy (2, 4, 11-16). Based on data from the Centers for Disease Control and Prevention, many laboratories in the United States do such testing by a variety of methods (21). No standardized susceptibility test method currently exists for the rapidly growing mycobacteria, but investigators who have studied M. abscessus, M. chelonae, and the M. fortuitum group extensively recommend broth microdilution testing (3, 11, 12). Data concerning other species of rapidly growing mycobacteria are very limited. Because of variability in the appearance of growth of M. abscessus, M. chelonae, and the M. fortuitum group in microdilution trays, unlike most bacterial isolates, interpretation of the MIC may be difficult. The primary focus of our study, therefore, was to evaluate the reproducibility of the broth microdilution method in several laboratories where persons performing the test have different levels of experience with this technique for testing the rapidly growing mycobacteria.

We found that reproducibility of MICs and agreement by interpretive category varied considerably among the different isolates and the different drugs. The wide range of MICs observed with several isolate-drug combinations was rarely seen with results reported by site A, where the rapidly growing mycobacteria have been studied for many years. This suggests that inexperience was in part responsible for the poor reproducibility and/or poor agreement by interpretive category.

The wide MIC ranges for several drugs, especially those which have a trailing endpoint, such as sulfamethoxazole and ciprofloxacin with isolates of M. chelonae, were predominantly due to results from site D, which had no experience with the rapidly growing mycobacteria and very limited experience with microdilution testing in general. Excluding site D's data from analysis, however, has minimal impact on the overall results. The most noticeable change is 100% agreement by interpretive category for sulfamethoxazole and isolates of the M. fortuitum group. The only other positive effect was better reproducibility and agreement by category for ciprofloxacin and M. chelonae 1814 and 1831.

The recently revised Statement on Diagnosis and Treatment of the Nontuberculous Mycobacteria from the American Thoracic Society suggests that a minimum of seven drugs (amikacin, cefoxitin, ciprofloxacin, clarithromycin, doxycycline, imipenem, and a sulfonamide) should be tested against rapidly growing mycobacteria (16). Some modifications or additions to this recommendation are needed, however. For isolates of M. chelonae only, we recommend including tobramycin, because it has a much better therapeutic margin than amikacin (12) and most consider it the aminoglycoside of choice for this species (16).

We also believe that a sulfonamide need not always be tested. Data from previous studies have shown that virtually all isolates of M. chelonae and M. abscessus are resistant to sulfamethoxazole (MIC  64 µg/ml), whereas all isolates of M. fortuitum are susceptible (MIC  32 µg/ml) (11, 12, 17). In addition, because of a major inoculum effect and use of an 80% inhibition-of-growth endpoint, testing can be problematic (3). Therefore, if the isolate has been identified (i.e., at least as belonging to the M. fortuitum group versus the M. chelonae-abscessus group), testing a sulfonamide may not be necessary. If the drug is tested and the MIC differs from the expected values, that result should be withheld until testing has been repeated. If the repeat result again differs from the expected values, we recommend reporting that result with a comment indicating that (i) the MIC is greater or less than that expected for the particular species and (ii) if the drug is being considered for therapy, the laboratory should be notified so the isolate can be sent to a reference laboratory for confirmation of the susceptibility test result and the identification.

Additional suggestions for modifications of susceptibility methods involve breakpoints for doxycycline and cefoxitin. The establishment of resistance breakpoints for doxycycline was relatively easy compared to those for most other drugs for the rapidly growing mycobacteria, as the distribution of MICs is primarily bimodal. In an early study (1979) comprised mostly of M. fortuitum isolates, Wallace et al. compared agar dilution MICs and disk diffusion (both done in Mueller-Hinton agar) and found that for doxycycline, 65 of 66 isolates (98%) had disk zones of inhibition of either 30 or 15 mm in diameter (17). The doxycycline MICs of all isolates of M. fortuitum with zone diameters of 15 mm were 8 µg/ml in agar, and those for which the MICs were 1 µg/ml all had disk zone diameters of 30 mm. The doxycycline MICs of only 14 of 66 (21%) isolates were between 2 and 8 µg/ml (17). In a subsequent study with broth microdilution, the doxycycline MICs of only 6 of 96 (6%) isolates of M. fortuitum were in the 2- to 8-µg/ml range (12). This same study demonstrated a similar bimodal distribution of MICs for M. chelonae and M. abscessus (12).

In these early studies, MIC testing was done by agar dilution, while most studies conducted since 1985 have utilized broth. Data from some studies have suggested that MICs of doxycycline against the rapidly growing mycobacteria are lower in broth than in agar. In a study by Swenson et al. (11), a comparison of broth and agar MICs for 18 strains of M. fortuitum showed that isolates generally were more susceptible in broth (e.g., the concentration of drug that inhibited 50% of the strains was 8 µg/ml in broth and 32 µg/ml in agar). The one laboratory in the current study that has been performing susceptibility testing of the rapidly growing mycobacteria for many years has utilized disk diffusion to help with interpretation of doxycycline results for isolates for which the MICs in broth are 2 to 8 µg/ml. Such isolates with disk zone diameters of 15 mm have been reported as resistant, those with zone diameters of 30 mm have been reported as susceptible, and those with zone diameters of 16 to 29 mm have been reported as intermediate. This laboratory reviewed the disk diffusion results for 118 isolates of M. fortuitum for which the MICs in broth were 2 to 8 µg/ml; all but 3 (97%) had disk zone diameters of <30 mm, and all but 14 (88%) had zones of inhibition with diameters of 15 mm (20a). This suggests that MICs for the isolates for which the MICs in broth were 2 to 8 µg/ml would likely have been higher (8 µg/ml) if tested in agar. Several studies have demonstrated the success of doxycycline monotherapy in the treatment of disease caused by rapidly growing mycobacteria when the infecting organism is susceptible in vitro to concentrations of 1 µg/ml (1, 5, 20). We are aware of no clinical data regarding the efficacy of doxycycline therapy for isolates of the M. fortuitum group for which MICs are 2 to 8 µg/ml when tested by either agar or broth dilution.

Based on these findings, the proposed breakpoints for doxycycline are 1 µg/ml (susceptible), 2 to 8 µg/ml (intermediate), and 16 µg/ml (resistant). These recommended breakpoints apply only to doxycycline and not to minocycline or tetracycline and are the same as those suggested by two of the investigators in the latest edition of the Clinical Microbiology Procedures Handbook (3). They differ from the breakpoints listed in the current NCCLS document for aerobic bacteria, which has one set of values for all tetracyclines: 4 µg/ml for susceptible, 8 µg/ml for intermediate, and 16 µg/ml for resistant (8a).

The other drug for which breakpoint modifications are recommended is cefoxitin (Table 1). The problem with the existing cefoxitin breakpoints is that the resistance breakpoint (64 µg/ml) is in the middle of the normal MIC range for untreated isolates of several of the rapidly growing mycobacteria. Previous studies have demonstrated that the cefoxitin MICs for more than 90% of isolates of M. fortuitum and M. abscessus range from 16 to 64 µg/ml, with a mode of 32 µg/ml (2, 12, 14). In three studies from three different laboratories the MICs for 232 of 239 (97%) M. fortuitum isolates and 243 of 258 (94%) M. abscessus isolates were within this range (2, 12, 14). With the usual recommended dosing, peak serum cefoxitin levels above 100 µg/ml can be achieved (10). Additionally, the clinical response of isolates for which the MIC is 32 µg/ml to treatment with cefoxitin does not differ from the response of isolates for which the MIC is 64 µg/ml (20a). Based on this information, we recommend changing the cefoxitin interpretive breakpoints as follows: 16 µg/ml, susceptible; 32 to 64 µg/ml, intermediate; 128 µg/ml, resistant. This differs from the Clinical Microbiology Handbook (3), which has breakpoints of 16, 32, and 64 µg/ml, respectively. Using these new breakpoints, the percent agreement by interpretive category was 100% for all isolates in our study except 1802 and 1352, for which agreement was 97.2 and 86.1%, respectively.

Several problem areas for reproducibility of testing of clarithromycin, imipenem, tobramycin, and amikacin were identified in this study. The difficulty with clarithromycin occurred with some isolates of M. fortuitum for which the endpoint was trailing. In our study, these isolates were problematic for all sites. Currently, there are no clinical data with which to correlate the MIC interpretation in these cases. Given the lack of clinical information and the availability of other oral drugs with which to treat most isolates of M. fortuitum (i.e., quinolones and sulfonamides), we recommend a conservative interpretation. In our opinion, isolates of M. fortuitum that have a trailing endpoint with clarithromycin should be considered resistant to the drug until clinically relevant information that refutes this approach is available.

With regard to imipenem, reproducibility was poor at all sites. Although the reason(s) for the lack of reproducibility is not known, we believe that drug instability is at least partially responsible. Based on data from previous studies (2, 12, 14), all isolates of M. fortuitum are susceptible or intermediate to imipenem in vitro (MIC 8 µg/ml). For the isolates of M. fortuitum included in our study, all MICs of >8 µg/ml, with the exception of two reports of 16 µg/ml from site A, were reported from the three laboratories with the least experience (primarily site B). In all three of these laboratories MICs were interpreted on day 4 or 5 (compared to consistent reading on day 3 at site A). Based on these findings, we hypothesize that for isolates of M. fortuitum the problem with imipenem can be avoided by strict adherence to a 3-day incubation period, which, in the experience of one of the authors (R.W.), is virtually always sufficient for M. fortuitum. If the imipenem MIC for an isolate of M. fortuitum is >8 µg/ml on day 3, we recommend repeating the test. If the repeat result is >8 µg/ml, it should be reported with a comment indicating that (i) the MIC is greater than that expected for M. fortuitum and (ii) if the drug is being considered for therapy, the laboratory should be notified so the isolate can be sent to a reference laboratory for confirmation. With isolates of M. abscessus and M. chelonae, on the other hand, growth often is not adequate until day 4. Given the instability of imipenem and the need for more prolonged incubation when testing the latter two species, we recommend either not testing isolates of M. abscessus and M. chelonae against imipenem or not reporting the result if the organism is resistant until the problem with reproducibility is resolved.

The last drugs with reproducibility problems were tobramycin and amikacin. For these drugs, lack of agreement occurred predominantly with isolates of M. abscessus and M. chelonae that had modal MICs close to the breakpoints for resistance. The specific reasons for the problems are unknown. Until this issue is resolved, we suggest the following. Because therapeutically tobramycin is recommended only for M. chelonae infections, in our opinion, results should be reported only for isolates of this species, not for isolates of the M. fortuitum group or M. abscessus. In addition, isolates of M. chelonae for which the tobramycin MIC is >4 µg/ml should be retested before the result is reported. If the repeat result is >4 µg/ml, we recommend reporting that result with a comment indicating that (i) the MIC is greater than that expected for M. chelonae and (ii) if the drug is being considered for therapy, the laboratory should be notified so the isolate can be sent to a reference laboratory for confirmation of both resistance and identification. It is possible that the isolate belongs to a newly recognized species, Mycobacterium immunogen, for which the MICs of both cefoxitin and tobramycin (22) are high, in contrast to M. chelonae, which usually is susceptible to tobramycin.

With regard to amikacin, the most significant problem is lack of agreement based on the interpretive category. As with several other drugs, the amikacin MICs fall within a narrow range (2, 12). In our study, this was an issue for M. abscessus 1802 and M. chelonae 1831 and 1866. For 1802 and 1866, all MICs of 64 µg/ml, which is the currently recommended breakpoint for resistance, were reported by sites B and C, one of which tended to report higher than modal MICs for other drugs. For 1802, only three of the nine results at both sites were 64 µg/ml; most of the other six results were 32 µg/ml, and one result from site C was 16 µg/ml. For 1866, only one result from site B and two results from site C were 64 µg/ml; the other results ranged from 8 to 32 µg/ml. MICs for aminoglycoside-treated isolates of M. abscessus and M. chelonae which develop mutational resistance to amikacin will be >1,024 µg/ml (9). Based on these data, to avoid potential reporting errors and consequent failure to add an important supportive agent to the therapeutic regimen, we recommend that isolates of M. abscessus for which the amikacin MIC is 64 µg/ml be retested. If the repeat result is 64 µg/ml, it should be reported with a comment indicating that (i) the MIC is greater than that expected for M. abscessus and (ii) if the drug is being considered for therapy, the laboratory should be notified so the isolate can be sent to a reference laboratory for confirmation of resistance. Because tobramycin is the aminoglycoside of choice for isolates of M. chelonae, amikacin results need to be reported only if the isolate is resistant to tobramycin. In such cases, the guidelines suggested above for M. abscessus should be followed.

A secondary goal of our study was to identify a candidate clinical isolate to serve as a quality control organism for susceptibility testing of the rapidly growing mycobacteria. Although none of the isolates evaluated was perfect for this role, M. peregrinum 1353, ATCC 700686, was closest to optimal and is our choice for a quality control organism.

In summary, our data suggest that broth microdilution testing of the common rapidly growing pathogenic mycobacteria requires skill acquired through experience with the test method and knowledge of the expected susceptibility patterns of the different species. For laboratories that infrequently encounter isolates of rapidly growing mycobacteria for which susceptibility testing is clinically indicated, referring those isolates to an experienced laboratory may be most reasonable. If a laboratory chooses to perform testing in house, however, several issues must be addressed. The drugs recommended by the American Thoracic Society (16), plus, in our opinion, tobramycin for isolates of M. chelonae, should be tested at concentrations appropriate for these organisms. Because commercial panels do not provide adequate concentrations and/or drugs for testing these organisms, in-house-prepared or custom-made commercial panels must be used. Test performance must be validated. At present no proficiency testing service (such as the College of American Pathologists) regularly includes the rapidly growing mycobacteria, although during the past year the Centers for Disease Control and Prevention performance evaluation program for susceptibility testing of Mycobacterium tuberculosis included one isolate of M. fortuitum (21). The best alternative at present would be comparison of results with those of an experienced reference laboratory. This should be done with initial validation of the test system and again on a regular basis to demonstrate continued proficiency. This almost certainly requires identification of the isolate to the species level or, at a minimum, differentiation of the M. fortuitum group from the M. chelonae-M. abscessus group. Additional pathogenic species, such as Mycobacterium mucogenicum and Mycobacterium smegmatis, were not evaluated in this study but may be encountered among clinical isolates. It is not anticipated that these other rapidly growing mycobacteria will perform differently than the three species or taxa evaluated in the present study, although the recommended drugs to be tested may differ.