LETTER

Top Letter References

The accurate detection of oxacillin resistance among staphylococcal isolates in the clinical laboratory is of vital importance in order to guide therapy and possibly avoid treating patients unnecessarily with vancomycin. Oxacillin-susceptible staphylococcal infections are more effectively treated with a -lactam antimicrobial than with vancomycin for reasons of rapid bactericidal activity, better tissue absorption, fewer complications, lower cost, and limiting the possibility of selecting vancomycin resistance (2, 8). The accurate detection of these strains is directly dependent upon the quality of the manual or automated susceptibility testing system used in the clinical laboratory.

In recently reported studies, correlations of oxacillin MICs for coagulase-negative staphylococci (CoNS) with the presence or absence of the mecA gene have prompted the National Committee for Clinical Laboratory Standards (NCCLS) to modify the oxacillin interpretive criteria for this organism group (2, 3, 7, 8). The NCCLS currently recommends applying a susceptible MIC breakpoint of 0.25 µg/ml and a disk diffusion test breakpoint of 18 mm when testing oxacillin against CoNS, changed from the previous breakpoints of 2 µg/ml and 13 mm, respectively (6).

This letter outlines the results from a preliminary assessment of the ability of the Vitek System (bioMerieux Vitek, Inc., Hazelwood, Mo.) to accurately classify CoNS as oxacillin susceptible or resistant by the new NCCLS interpretive criteria. A total of 125 strains of CoNS were tested by the reference broth microdilution and disk diffusion methods (4, 5), and by the reference quality E test (AB BIODISK, Solna, Sweden) (7). All strains were initially isolated from patients with bloodstream infections and were selected from geographically diverse hospitals in the United States during the period 1995 to 1996 (2). Strains were further characterized as mecA positive (99 strains) or mecA negative (26 strains) as determined by the PCR method described by Geha et al. (1). Nine Staphylococcus species, identified at the institution of origin, were tested including (mecA positive/negative): Staphylococcus epidermidis (68/11 strains), S. haemolyticus (13/0 strains), S. hominis (10/3 strains), S. warnerii (2/6 strains), S. saprophyticus (3/1 strains), S. simulans (3/0 strains), S. capitis (0/2 strains), S. lugdunenisis (0/2 strains), and S. auricularis (0/1 strain). This organism collection was concurrently tested by using specialized Vitek cards with an oxacillin dilution range that adhered to the lowered breakpoint concentration (MIC range, 0.25 to 8 µg/ml). Quality control was maintained by testing S. aureus ATCC strains 25923, 29213, and 43300 for all methods; all results were within expected control ranges.

Table 1 summarizes the categorical phenotype results obtained by all methods compared to the organism's mecA status. Two strains of S. lugdunenisis (a species not assessed by the Vitek system) were excluded from the analysis. Agar-based systems showed absolute (100.0%) sensitivity for detecting mecA-positive CoNS strains while maintaining acceptable specificity (disk diffusion, 87.5%; E test, 79.2%) for detecting mecA-negative strains. Broth-based systems also possess high sensitivity for detecting mecA-positive strains (Vitek, 98.0%; broth microdilution, 97.0%) with specificities comparable to other methods (87.5 and 91.7%, respectively). A total of 10 strains were retested (triplicate determinations) by all methods in order to confirm the initial discord between their genotypic and phenotypic results. Retests of two strains, one strain by the disk diffusion test (S. hominis) and the other by the broth microdilution method (S. epidermidis), resulted in a categorical change, leading to agreement with the mecA genotype. The remaining eight strains demonstrated reproducible discrepancies between their genotypic and phenotypic results.

When the broth microdilution method was used as the reference phenotypic test (Table 2), all other methods were essentially equal and capable of categorizing oxacillin-resistant strains with absolute (disk diffusion and E test) or near-perfect (Vitek) accuracy. A single strain of mecA-positive S. epidermidis was repeatedly misclassified as susceptible by the Vitek system (MIC, 0.25 µg/ml). However, repeated broth microdilution results from both our laboratory and that of bioMerieux Vitek for this strain ranged from 0.25 to 1 µg/ml, i.e., values spanning the new oxacillin-resistant breakpoint of 0.5 µg/ml. Specificities for correctly categorizing oxacillin-susceptible strains as determined by broth microdilution ranged from 76.0% (E test) to 88.0% (Vitek).

While the agar-based systems tested in this study (disk diffusion and E test) were slightly more accurate predictors of a mecA-positive CoNS strain, the Vitek and broth microdilution systems performed at an acceptable level (98.0 and 97.0%, respectively). Errors from all methods were secondary to the infrequently isolated strains with susceptibility results at or near the new oxacillin breakpoint. Most clinical laboratories must depend on reference or commercial phenotypic methods because of the lack of resources to perform molecular (PCR or DNA probe) tests. Although Vitek cards with an extended oxacillin range remain developmental and are currently undergoing Food and Drug Administration clinical trials, this initial evaluation indicates a favorable role for the Vitek system in the accurate determination of oxacillin resistance in CoNS.