Investigation of Staphylococcus aureus Isolates Identified as Erythromycin Intermediate by the Vitek-1 System: Comparison with Results Obtained with the Vitek-2 and Phoenix Systems

Abstract

We identified 69 Staphylococcus aureus isolates that were erythromycin intermediate as reported by the Vitek-1 system using the GPS-105 card. Of the 57 strains that were available for further testing, all were erythromycin resistant by broth microdilution and the Phoenix system, while the Vitek-2 system identified 55 of 57 strains (96%) as erythromycin resistant. The majority of isolates (54 of 57 [95%]) exhibited the inducible MLS (macrolide-lincosamide-streptogramin family) phenotype, as shown by the double-disk test. We recommend that all S. aureus strains determined as erythromycin intermediate by the Vitek-1 system be interpreted as resistant to erythromycin.

Erythromycin and clindamycin belong to the macrolide-lincosamide-streptogramin (MLS) family of antibiotics, for which there are three different mechanisms of resistance: target site modification, enzymatic antibiotic inactivation, and macrolide efflux pumps (3, 7, 8). The best-known and most common mechanism is target modification caused by methylation of adenine nucleotides in the 23S ribosomal subunit. The erm genes encode these resistance enzymes (10). Expression of MLS resistance can be either constitutive or inducible (2). Previous studies have shown that inducible MLS resistance is common among erythromycin-resistant Staphylococcus aureus strains (4, 11). Such strains remain susceptible to clindamycin, since they do not induce erm expression. However, the use of clindamycin for the treatment of an infection due to an inducibly resistant strain of S. aureus can result in the emergence of resistance. Constitutive mutants can occur at frequencies of 10⁻⁷ (6). Therefore, if the staphylococcal inoculum at the site of infection is high, resistance can emerge during the course of clindamycin therapy (1, 12).

Many microbiology laboratories today use automated systems to determine antibiotic susceptibilities in S. aureus. Because the majority of MLS resistance in S. aureus is erm mediated, erythromycin-resistant strains are routinely reported as clindamycin resistant in Canada (9), although this is currently not recommended by the National Committee for Clinical Laboratory Standards (NCCLS) guidelines. However, no studies have looked at strains identified as erythromycin intermediate.

One of the widely used systems in Canada and the United States is the Vitek-1 system by bioMerieux (Hazelwood, Mo.). From the results generated in two different laboratories in the Toronto region, we identified 69 S. aureus isolates that were erythromycin intermediate according to the Vitek-1 system. These included strains from a genetically diverse collection of Canadian and U.S. methicillin-resistant S. aureus (MRSA) and methicillin-susceptible S. aureus (MSSA) stored at the Microbiology Department at Mount Sinai Hospital. The isolates had been collected between 1997 and 2002 and were stored in buffered glycerol at −70°C. Of the 69 S. aureus isolates, 57 (55 MRSA and 2 MSSA) were available for further antibiotic susceptibility testing, which included broth microdilution and disk diffusion. These results were also compared to those from the bioMerieux Vitek-2 and Becton Dickinson (Sparks, Md.) Phoenix systems.

Broth microdilution was done according to NCCLS guidelines. Cation-adjusted Mueller-Hinton broth was prepared with media from Difco Laboratories (Detroit, Mich.). Erythromycin and clindamycin powders were obtained from Sigma (St. Louis, Mo.). Organisms were screened for inducible MLS resistance with two 2-μg clindamycin disks placed precisely 15 and 20 mm (edge-to-edge) from a 15-μg erythromycin disk on Mueller-Hinton plates (antimicrobial disks and plates obtained from Oxoid, Nepean, Canada). Blunting of the zone of inhibition around the clindamycin disk after 18 to 24 h of incubation was interpreted as positive for inducible MLS resistance (5). GPS-105, AST-P526, and PMIC/ID-14 cards were used in the Vitek-1, Vitek-2, and Phoenix systems, respectively, as per the manufacturers' instructions.

All 57 strains had been tested initially on the Vitek-1 system by using the system software version 8.4. The original erythromycin MICs were 1 μg/ml for 44 isolates (77%), 2 μg/ml for 5 isolates (9%), and 4 μg/ml for 8 isolates (14%) (Table 1). The clindamycin MIC was ≤0.5 μg/ml by the Vitek-1 system for all of the isolates, and these had been interpreted as susceptible. When broth microdilution was performed, all strains were erythromycin resistant (MIC, ≥8 μg/ml) and clindamycin susceptible (MIC, ≤0.5 μg/ml).

TABLE 1.

S. aureus erythromycin susceptibilities according to different techniques

Vitek-1 erythromycin MIC (μg/ml)a (n isolates)	No. of isolates with inducible MLS resistance by disk diffusion	Erythromycin MIC (μg/ml)a
		Broth microdilution	Phoenix	Vitek-2
1 (44)	44	≥8 (44)	≥8 (44)	1 (1), 4 (1), ≥8 (42)
2 (5)	5	≥8 (5)	≥8 (5)	≥8 (5)
4 (8)	5	≥8 (8)	≥8 (8)	≥8 (8)

^aValues in parentheses are numbers of isolates.

Antimicrobial susceptibilities by disk diffusion showed 56 of the 57 strains (98%) were erythromycin resistant (inhibition zone diameter of 13 mm or less), while 1 strain was erythromycin intermediate, with an inhibition zone diameter of 16 mm. The double-disk method showed 54 of 57 strains (95%) had the inducible MLS phenotype, with the remaining 3 strains truly susceptible to clindamycin. The results were equivalent whether the clindamycin and erythromycin disks were 15 or 20 mm apart. However, blunting of the zone of resistance was more evident when disks were 15 mm apart, with mean reductions in radius of 7.6 ± 2.1 mm with disks 15 mm apart and 5.5 ± 1.5 mm with disks 20 mm apart.

In comparison, the Phoenix system correctly identified all isolates as erythromycin resistant (MIC, ≥8 μg/ml) and the Vitek-2 system found 55 of 57 isolates (96%) were erythromycin resistant (MIC, ≥8 μg/ml). For the two remaining isolates, the erythromycin MICs determined by the Vitek-2 system were 1 and 4 μg/ml. These two isolates had the inducible MLS phenotype. Both the Phoenix and Vitek-2 systems found all the strains to be clindamycin susceptible, and neither had expert rules overriding the susceptibility results. When subsequent testing of 55 isolates was performed with the Vitek-1 system, 13 isolates (24%) continued to be erythromycin intermediate and 1 (2%) was susceptible.

The erythromycin-intermediate S. aureus strains identified in this study represent anomalous results generated by the Vitek-1 system. Of 57 isolates from across North America, all were erythromycin resistant by the NCCLS reference method of broth microdilution, and 56 were resistant by the reference disk diffusion method. Repeat testing with the Vitek-1 system gave variable results that continued to be discordant from the other methods for 25% of the isolates. However, erythromycin-intermediate S. aureus strains would not normally be retested in a routine clinical laboratory before results are reported. The Vitek-2 and Phoenix automated systems have improved detection of erythromycin resistance in S. aureus, identifying 96 and 100% of isolates as resistant, respectively. The reason for the difference in results between the different systems is not readily apparent. The Vitek-1 system uses two concentrations of erythromycin at 0.5 and 4.0 μg/ml and interpolates or extrapolates the actual MIC based on the growth rate of the organism in those two concentrations. The Vitek-2 system incorporates 0.25-, 0.5-, and 2.0-μg/ml erythromycin wells into its antimicrobial susceptibility card and calculates the actual MIC based on the growth characteristics at those concentrations. The Phoenix system most resembles that of broth microdilution, having erythromycin wells containing all concentrations from 0.25 to 4.0 μg/ml.

Double-disk diffusion testing for the MLS resistance phenotype found that 95% of the strains were inducible, consistent with the presence of an erm gene. The newer Vitek-2 and Phoenix systems have more accurate determinations of the erythromycin MIC for S. aureus. However, the small proportion of S. aureus isolates that do have intermediate susceptibility to erythromycin on the Vitek-2 system should be confirmed by double-disk diffusion, because these may also be anomalous results.

Given our results and the high rates of inducible MLS resistance, one could argue that routine screening of all Vitek-1 erythromycin-intermediate S. aureus isolates by a methodology such as the double-disk method is not warranted, since this would delay reporting of results. Rather, all erythromycin-intermediate S. aureus isolates identified by the Vitek-1 system could be automatically interpreted as resistant to erythromycin. More controversial is how to report clindamycin susceptibility in these isolates, since they will be reported as susceptible by the Vitek-1 system. Options include reporting all such isolates as resistant to clindamycin or issuing a warning with the susceptibility test report. For example, the NCCLS document currently suggests warning physicians that Enterobacter, Citrobacter, and Serratia may develop resistance during prolonged therapy with broad-spectrum cephalosporins. Therefore, isolates that are initially susceptible may become resistant 3 to 4 days after initiation of therapy, and testing of repeat isolates may be warranted. In addition, neither the Vitek-2 nor the Phoenix system currently employs expert rules to reflect the inducible nature of clindamycin resistance, and this results in a major reporting error each time such an isolate is identified.

Similar problems may exist for antimicrobial susceptibility testing of other gram-positive organisms, such as group B Streptococcus in the Vitek-1 system, and we are currently investigating this issue.