Wide variability in Pseudomonas aeruginosa aminoglycoside results among seven susceptibility testing procedures

Abstract

Seven commonly used antimicrobial susceptibility testing methods were used to test the susceptibility of 150 isolates of Pseudomonas aeruginosa against gentamicin, tobramycin, amikacin, carbenicillin, and piperacillin. Results were compared with respect to the susceptibility characteristics of the population of isolates as defined by each method. Conventional methods included agar disk diffusion and agar dilution, carried out in accordance with current recommendations of the National Committee for Clinical Laboratory Standards, as well as broth microdilution testing with cation-supplemented Mueller-Hinton broth (CSMHB). Methods in which instrumentation was used for result determination included the Autobac I, Avantage, Sensititre Autoreader (using a breakpoint panel at 18 h of incubation), and Vitek (AMS-240, using the GNS susceptibility card). When necessary for comparison, MIC data were converted to categorical interpretations (susceptible, intermediate, and resistant). With respect to gentamicin, no significant differences were noted among the results of disk diffusion, broth microdilution, Sensititre Auto breakpoint, or Vitek methods which characterized 60 to 67% of isolates as susceptible, 16 to 22% as intermediate, and 13 to 17% as resistant. In contrast, agar dilution, Autobac, and Avantage, although yielding gentamicin results similar to those of one another, were each significantly different in result reporting from the other four methods above for gentamicin results, and they characterized the Pseudomonas population largely as susceptible (88 to 97%), with 0 to 6% intermediate and only 3% to 6% resistant. More isolates were characterized as being resistant to gentamicin in the Avantage test if an assay broth supplemented with increased amounts of calcium was used. Cation impregnation of Autobac disks did not appreciably change Autobac results. The geometric mean MIC of gentamicin was 4 micrograms/ml lower in the agar dilution method than in the CSMHB microdilution method, despite monitoring of the agar for cation content through performance disk diffusion testing with P. aeruginosa ATCC 27853. Tobramycin activity was greater than gentamicin activity, and susceptibility to tobramycin ranged from 89 to 97%, with few statistically significant differences noted among the seven methods studied. Differences in MIC distribution and geometric mean MIC between agar dilution and CSMHB microdilution testing were minimal and suggested less of a cation influence on tobramycin than gentamicin results. Although amikacin was also more active than gentamicin (83 to 99% of isolates were susceptible), differences in the amikacin results among methods tended to reflect the same trends in reporting as seen with gentamicin testing, with the exception that results of Avantage testing were similar to those of disk diffusion, CSMHB microdilution, Sensititre, and Vitek. A difference in geometric mean MIC of 5 micrograms/ml between CSMHB testing and agar dilution testing suggested the influence of divalent cations on amikacin results. Few highly significant differences were noted among methods when isolates were tested against carbenicillin and piperacillin, except that Avantage piperacillin results (66% susceptible) and Autobac piperacillin results (98% susceptible) were noticeably different from the percent piperacillin susceptibility (range, 85 to 92%) measured by the other methods. Method-dependent variability among aminoglycoside susceptibility results, particularly when testing gentamicin, prevents meaningful comparison of Pseudomonas susceptibility trends among hospitals when different methods are used and promotes confusion and frustration among clinical microbiologists and clinicians owing to the uncertainties of clinical meaning of these data.