INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The rapidly growing mycobacteria Mycobacterium abscessus, Mycobacterium chelonae, and the Mycobacterium fortuitum group cause various forms of clinical disease, most frequently skin and soft tissue infections but also skeletal, pulmonary, catheter-related, and disseminated disease (1, 5, 6, 8, 16, 18-20). These different species vary in their susceptibilities to antimicrobial agents useful for therapy (1, 2, 4, 5, 13, 14, 16, 17, 19); therefore, antimicrobial susceptibility testing of isolates considered clinically significant is recommended (18). The most frequently described methods of testing susceptibilities of the rapidly growing mycobacteria are agar disk elution and broth microdilution, the latter of which is recommended by investigators who have extensively studied the rapidly growing mycobacteria (3, 13, 14). Few studies have evaluated the Etest (2, 7, 9), and among these, in only one was the issue of reproducibility addressed. The goals of the present multicenter study were twofold: (i) to evaluate the Etest for its ability to provide reproducible endpoints and interpretive categories by several laboratories with different levels of experience with regard to susceptibility testing of rapidly growing mycobacteria and (ii) to assess the accuracy of Etest results by comparing them to those obtained by broth microdilution (21).

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Organisms. Ten clinical isolates (four of the M. fortuitum group, three of M. chelonae, and three of M. abscessus) previously studied at the University of Texas Health Center at Tyler were selected for testing, as described elsewhere (11, 21). Isolates were identified using PCR-restriction analysis of a 439-bp segment of the 65-kDa heat shock protein gene (12). 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 on the slants at room temperature until tested.

Antimicrobial agents. Single lots each of amikacin, cefoxitin, ciprofloxacin, clarithromycin, doxycycline, imipenem, and trimethoprim-sulfamethoxazole Etest strips (AB Biodisk, Piscataway, N.J.) were evaluated. Final concentration ranges were 0.016 to 256 µg/ml for amikacin, cefoxitin, clarithromycin, and doxycycline and 0.002 to 32 µg/ml for ciprofloxacin, imipenem, and trimethoprim-sulfamethoxazole. The strips for all agents except clarithromycin have been cleared by the Food and Drug Administration for in vitro diagnostic use, but the application for testing rapidly growing mycobacteria has not been cleared for any.

Susceptibility test method. Each isolate was subcultured once onto a common lot of sheep blood agar plates (Remel, Lenexa, Kans.) and then incubated in ambient air at 30°C for 72 h. Suspensions were prepared by emulsifying colonies in 5 ml of Mueller-Hinton broth (Remel) to achieve a density equal to a 1.0 McFarland turbidity standard by visual examination or by using a nephelometer. Suspensions were mixed vigorously on a vortex mixer for 15 to 20 s and then used to inoculate the entire surface of two 150-mm-diameter Mueller-Hinton blood agar plates (Remel). A blood agar plate was also inoculated with a loopful of the final inoculum to check for purity.

Quality control. Staphylococcus aureus 29213 and Enterococcus faecalis 29212 were tested at each site each time the Etest was performed. Quality control was considered acceptable if results were within ranges recommended by the National Committee for Clinical Laboratory Standards (10).

Study design and analysis. Four laboratories participated in the study; one had extensive experience with the Etest method of susceptibility testing of mycobacteria (site C); the other three sites (A, B, and D) had minimal to no experience with the method. All laboratories tested each isolate three times on each of three separate days. MIC results and day of reading were recorded on data sheets and mailed to a coinvestigator (M.P.) for entry of data into a database. Each test at each site was considered a separate result. Agreement was determined by calculating the percentage of MICs within a 3-dilution range (i.e., mode ± 1 twofold dilution) for each drug. The range of concentrations evaluated for each drug was the one that corresponded to the range of concentrations used in the broth microdilution test (21). High off-scale MICs (with respect to the highest concentration tested by broth microdilution) were converted to the next highest concentration, whereas low off-scale MICs were left unchanged. Breakpoints for determining susceptibility and resistance (Table 1) are those recommended by Woods et al. (21), except for trimethoprim-sulfamethoxazole, for which the breakpoint for sulfamethoxazole was used. To compare Etest results to those of broth microdilution, Etest MICs that were in between dilutions tested by broth microdilution were rounded to the next highest dilution included in the broth microdilution assay (e.g., if the Etest MIC was 12 µg/ml, it was adjusted to 16 µg/ml).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Tables 2 through 4 summarize the Etest MIC results of the seven antimicrobial agents tested for M. abscessus, M. chelonae, and the M. fortuitum group and the interlaboratory percent agreement among the four participating laboratories. Intralaboratory reproducibility is shown in Table 5. Both intra- and interlaboratory agreement varied considerably for the different isolate-drug combinations. Interestingly, the laboratory with the least variability in MIC results had no experience using the Etest method for susceptibility testing of rapidly growing mycobacteria.

Overall, interlaboratory agreement was best for trimethoprim-sulfamethoxazole, followed by doxycycline. Of note, agreement for doxycycline and M. chelonae strain 1814 (selected based on known susceptibility to tetracyclines) was only 36.1%. For ciprofloxacin, agreement was good for M. abscessus, the M. fortuitum group, and all but one M. chelonae isolate. With some drugs, agreement was excellent (>94%) for all isolates of one species or group of organisms and considerably lower for the other two. This was true for cefoxitin and imipenem and M. chelonae and for amikacin and the M. fortuitum group. With clarithromycin, agreement was good for M. abscessus but was lower for M. chelonae and for the M. fortuitum group.

To assess the potential impact of the variability in MIC results on patient management, we also evaluated inter- and intralaboratory percent agreement based on interpretive category (Tables 6 and 7). Again, agreement varied, but results differed somewhat from those based on MICs. Interlaboratory agreement was 100% for doxycycline with all 10 isolates, for trimethoprim-sulfamethoxazole with nine isolates, and for ciprofloxacin with eight isolates. For imipenem and clarithromycin, interlaboratory agreement was 100% for M. chelonae and 91.7 to 100% for M. abscessus but varied more widely for the M. fortuitum group, especially with imipenem. Amikacin agreement was 100% for the M. fortuitum group but was considerably lower for M. chelonae and M. abscessus, a pattern similar to the amikacin reproducibility results. Overall, for every drug-isolate combination except cefoxitin and M. abscessus strain 1801, percent interlaboratory agreement by interpretive category was equal to or greater than percent agreement by MIC. The poorer agreement by interpretive category for M. abscessus strain 1801 and cefoxitin resulted from the fact that the modal MIC was at the upper limit of the intermediate range.

Intralaboratory percent agreement by interpretive category was 100% for many drug-organism combinations. Only for amikacin against M. abscessus 1802 and M. chelonae 1866, and for imipenem against M. fortuitum 1351, did agreement fall to as low as 55.6% for one laboratory in each case.

Further analysis of the data revealed that for many drug-isolate combinations, results that were considerably higher or lower than the mode or resulted in a categorical interpretation considered atypical for that species were reported by one or two sites. For amikacin and M. abscessus strains 1801 and 1807, for both of which the modal MIC was 16 µg/ml, all results of 256 µg/ml were reported by site C, where the testing personnel had appreciable experience with the Etest. Sites B and C reported all amikacin results of 256 µg/ml for M. abscessus strain 1802 and M. chelonae strains 1831 and 1866, although the interpretive category was not affected because for each isolate the modal MIC was 64 µg/ml, the breakpoint for resistance. For cefoxitin, sites B and C reported results considered resistant (i.e., >64 µg/ml) for M. fortuitum group strains 1351 and 1359, for which the modal MICs were 64 and 32 µg/ml, respectively (the M. fortuitum group typically is susceptible or intermediate to cefoxitin [13, 17]). The five ciprofloxacin results of <4 µg/ml (considered susceptible or intermediate) for M. chelonae strain 1831 (modal MIC, 8 µg/ml) were reported by sites A and D. Site C reported all nine clarithromycin results considered susceptible or intermediate (i.e., <8 µg/ml) for M. fortuitum group strain 1351 (modal MIC, 64 µg/ml), which was selected because it was known to have a trailing endpoint for this drug. Sites B and D reported all doxycycline results of 2 to 8 µg/ml (considered intermediate) for M. chelonae strain 1814 (modal MIC, 0.25 µg/ml), which was chosen because it was known to be susceptible to tetracyclines. With imipenem and M. fortuitum strain 1359, all but 1 of the 16 results of >8 µg/ml (considered resistant) were reported by sites B and D (for M. fortuitum group strains, MICs of imipenem typically are <8 µg/ml [13, 17]). Site D was responsible for all nine trimethoprim-sulfamethoxazole results of >32 µg/ml (considered resistant) reported for M. fortuitum strain 1351 (M. fortuitum group strains typically are susceptible to sulfonamides [13]).

Interlaboratory modal MICs and percent agreement by interpretive category by Etest were compared to those obtained by broth microdilution (21) for these same 10 isolates. For doxycycline and sulfonamides, the modal MICs were equivalent (i.e., within ± 1 twofold dilution) for all isolates, although for sulfonamides the percent agreement for one isolate differed by 25% between Etest (75%) and broth microdilution (100%). For all other drugs, the modal MICs by the two methods differed by more than ± 1 twofold dilution for one or more isolates (Table 8), and in all cases the Etest MIC was higher. Table 8 also shows the drug-isolate combinations for which the difference in percent category agreement for the two methods was 25%. Given the tendency for Etest MICs to be higher than broth microdilution MICs, there was better percent agreement by interpretive category by Etest for organisms for which MICs were closer to the resistant range.

View this table: [in this window] [in a new window]   TABLE 8.   Drug-isolate combinations for which the modal MICs by Etest and broth microdilution differed by more than ± 1 twofold dilution and/or for which the difference in percent category agreement for the two methods was 25%a

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Susceptibility testing of clinically significant isolates of M. abscessus, M. chelonae, and the M. fortuitum group is recommended because these organisms differ in their susceptibilities to the antimicrobial agents commonly used for therapy (2, 4, 13, 14, 16-18). According to the American Thoracic Society, amikacin, cefoxitin, ciprofloxacin, clarithromycin, doxycycline, imipenem, and a sulfonamide should be tested (18). Investigators who have studied the rapidly growing mycobacteria most extensively recommend broth microdilution testing (3, 13, 14). Use of this method, however, is somewhat problematic. Trays must be prepared in house or custom made by a manufacturer because no commercial panels contain either doxycycline or cefoxitin at sufficiently high concentrations. Additionally, interpretation of the MIC for the rapidly growing mycobacteria is not always obvious, especially for persons who have minimal experience with microdilution testing of these organisms. Growth of the rapidly growing mycobacteria in microdilution trays often does not appear as a crisp, well-defined button in the bottom of the well, as is true of rapidly growing bacteria such as Escherichia coli and S. aureus. Additionally, some rapidly growing mycobacteria have trailing endpoints in broth. For these reasons, an alternative method that yields results that are comparable to broth microdilution, for which the components are commercially available, and for which endpoints are no more difficult to read, would be useful. The Etest has the potential to fulfill these requirements; but before it can be recommended, its accuracy and reproducibility must be documented. These were the goals of our study.

We found that reproducibility of Etest MICs and agreement by interpretive category varied among the different isolates and the different drugs but was less variable than what we observed with broth microdilution for these same isolates (21). To our knowledge, Biehle et al. are the only other investigators who have published data concerning the reproducibility of the Etest for testing the rapidly growing mycobacteria (2). They tested 25 isolates, apparently only once at two sites. Additionally, they did not indicate which species were tested or how the isolates were selected, and they did not include trimethoprim-sulfamethoxazole. In their study, the agreement among MICs differed for the different isolate-drug combinations and was 81% overall, using our definition of mode ± 1 twofold dilution. Their overall agreement by interpretive category was 92%, ranging from 76% for cefoxitin to 100% for amikacin and ciprofloxacin. In our study, interlaboratory agreement by interpretive category was 100% for ciprofloxacin and 8 of 10 isolates and 100% for amikacin and the M. fortuitum group. Cefoxitin results cannot be compared because different breakpoints were used in the two studies.

The major issue with the Etest was that MICs of a number of drugs, including amikacin, cefoxitin, and imipenem, were higher than broth microdilution MICs (21), putting the Etest MICs into the resistant category. Though the broth microdilution MIC breakpoints listed in reference 21 are not yet approved by the National Committee for Clinical Laboratory Standards, they have been in use (with minor exceptions) in in vitro studies for more than 10 years. Clinical data have been accumulated by one of us (R.J.W.) on many of the agents tested, demonstrating that reasonable correlation exists between these in vitro susceptibility results and clinical outcome. In particular, data from several studies have validated that amikacin and cefoxitin are active clinically against isolates of M. fortuitum and M. abscessus (5, 20).

Etest results for certain drugs (e.g., cefoxitin and imipenem) in the present study also are higher than those reported by others who have examined the Etest for susceptibility testing of the rapidly growing mycobacteria (2, 7, 9). The exact reasons for the discrepancies are not known, but it is certainly possible that differences in the performance or interpretation of the Etest played a role, because the Etest has not been standardized for the rapidly growing mycobacteria. Of the variables (i.e., inoculum density, incubation time, temperature and atmosphere, agar medium, and interpretation of the endpoint), only incubation time (i.e., 72 h) has been consistent.

With regard to inoculum, Biehle et al. (2) used a density equal to that of a 0.5 McFarland standard; all other investigators (7, 9), including us, used a density equal to a 1.0 McFarland standard. In the present evaluation, the plates were incubated at 30°C because isolates of the M. abscessus-chelonae group grow best at this temperature; however, in all other studies incubation has been at 35 to 37°C. Incubation atmosphere has included both ambient air and CO₂. With regard to agar medium, Mueller-Hinton agar with blood (9), Mueller-Hinton agar with OADC (oleic acid-albumin-dextrose-catalase) (2), and PDM ASM II agar (7) have been used in previous studies. In the current evaluation, Mueller-Hinton blood agar was chosen because it is readily available and relatively inexpensive and, in the experience of one of the authors (A.W.), allows good growth of the rapidly growing mycobacteria. With regard to interpretation, we found that determining the endpoint was difficult when the MIC approached the upper concentration limit on the strip, which unfortunately is near the breakpoint for some of the isolate-drug combinations. Interpretation also was problematic for isolates that had trailing endpoints with certain drugs, especially clarithromycin, doxycycline, and trimethoprim-sulfamethoxazole. Another variable that could potentially influence interpretation is the initial inoculum preparation. If the inoculum suspension is not smooth (i.e., the microparticles are in clumps rather than uniformly dispersed), which is not unusual with certain of the rapidly growing mycobacteria, the edge of the ellipse may be hazy rather than sharp, which makes reading difficult.

In addition to performance, cost must be considered when selecting a susceptibility test method. Each Etest strip for the drugs needed to test rapidly growing mycobacteria costs $2.05, and Mueller-Hinton blood agar plates cost about $0.50. To have custom broth microdilution trays containing the eight drugs (i.e., the seven drugs tested in this study plus tobramycin) recommended for testing rapidly growing mycobacteria (21) prepared by a commercial manufacturer (e.g., Trek Diagnostics), the cost per drug (for microtiter trays and other required reagents and materials) is about $1.12 for a minimum order of 500 panels. Microtiter trays can be stored at room temperature for 2 years after the date of manufacture. The technical time to perform the test and interpret the results is about the same for both methods.

In summary, use of the Etest for susceptibility testing of the rapidly growing mycobacteria shows sufficient promise to warrant further study. Results of this evaluation suggest that experience and proficiency testing programs are essential. However, before the Etest can be recommended, the method must be optimized such that results are comparable to those obtained by broth microdilution. In the current evaluation, Etest MICs were consistently higher than those obtained by broth microdilution, which in many cases would have caused reports of false resistance. Further studies are needed to determine the optimal medium and incubation temperature and atmosphere. Additionally, explicit instructions concerning interpretation of the endpoint, including pictures, are essential, especially for clarithromycin, doxycycline, and trimethoprim-sulfamethoxazole.