Comparison of agar dilution, broth microdilution, disk diffusion, E-test, and BACTEC radiometric methods for antimicrobial susceptibility testing of clinical isolates of the Nocardia asteroides complex.

An evaluation was undertaken to determine the optimal method for the in vitro susceptibility testing of 26 Nocardia asteroides complex isolates to the following antimicrobial agents: amikacin, ampicillin, amoxicillin-clavulanate, ceftriaxone, ciprofloxacin, erythromycin, imipenem, minocycline, and trimethoprim-sulfamethoxazole. Five testing methods were studied including the agar dilution, broth microdilution, and disk diffusion methods, the epsilometer test (E-test), and the BACTEC radiometric method. Results for each antimicrobial agent and each testing method were interpreted as indicating susceptibility, intermediate susceptibility, or resistance according to current guidelines of the National Committee for Clinical Laboratory Standards (NCCLS) for bacteria that grow aerobically and were then compared to a "gold standard" susceptibility test result. The gold standard result for each Nocardia isolate was established by a consensus of the results of the majority of testing methods used in the study. When the results were combined for all antimicrobial agents tested against all Nocardia isolates by all methods, the BACTEC radiometric method produced the highest level of agreement (97.9%) with the consensus results and had the fewest very major (n = 1), major (n = 2), and minor (n = 2) errors. In contrast, the results of the agar dilution method were in least agreement (93.2%) with the consensus results, and this method also produced the most very major (n = 8), major (n = 4), and, along with the disk diffusion method, minor (n = 6) errors. For all test methods, interpretive errors were most frequent when testing ampicillin or amoxicillin-clavulanate. Moreover, for all Nocardia nova isolates tested, ampicillin susceptibility results by any of the testing methods were not in agreement with the results of testing for beta-lactamase by the nitrocefin (Cefinase) disk method. We conclude that among the methods evaluated, the BACTEC radiometric method appeared to be the best for determining the in vitro susceptibilities of members of the N. asteroides complex to a panel of nine antimicrobial agents. However, none of the test methods, including the BACTEC method, accurately predicted the ampicillin resistance of the N. nova isolates tested, all of which produced beta-lactamase. Presuming that this beta-lactamase hydrolyzes ampicillin, this disparity may relate to the NCCLS breakpoints that were used, which may require modification for this antimicrobial agent when tested against N. nova isolates.