TEXT

Top Abstract Text References

The oxacillin agar screen test for detection of oxacillin resistance in Staphylococcus aureus was first included in the NCCLS dilution methods document, M7, in 1990 (11), although neither that edition nor the current edition (14) provides details on how the test should be inoculated. Furthermore, a variety of inoculation methods are recommended on the product labels of commercially prepared media. Despite this, many studies have shown that the oxacillin screen test performs well for detection of S. aureus strains that contain the mecA gene, which mediates oxacillin resistance (1-4, 6-8, 15-17, 20). However, when studies have included strains whose resistance is heterogeneous, the test has been shown to perform less well (2, 15).

Recommendations for quality control of the test were not included in NCCLS documents until 1997 (13), when S. aureus ATCC 43300, a mecA-positive strain that is very heterogeneous in its expression of oxacillin resistance (10), was suggested as the quality control strain. The reason for recommending a strain that was difficult to detect was that, when adequate growth was obtained, the user would be assured that the test could detect heterogeneously resistant strains. Recent comments addressed to the NCCLS Subcommittee on Antimicrobial Susceptibility Testing (14) have questioned the appropriateness of this strain for this procedure. Other references (9) and package inserts from commercial plates have recommended the use of S. aureus ATCC 33591, a strain which expresses homogeneous resistance.

The present study was undertaken both to clarify the best inoculation methods for the NCCLS oxacillin agar screen test and to verify that S. aureus ATCC 43300 is adequate for use as a positive quality control strain for the test.

Study design. The study was conducted in two laboratories, the Centers for Disease Control and Prevention (CDC) and Massachusetts General Hospital (MGH), in three phases. Phase 1 evaluated several inoculation methods in order to select a method that could be used in the subsequent phases. Phase 2 was undertaken to confirm that commercial lots of agar performed well with the chosen inoculation method. Phase 3 was performed to document the specificity of the test using a set of more commonly encountered susceptible organisms (see below).

Organisms. In phases 1 and 2, the organisms chosen for testing represented a challenge set from the CDC culture collection. For the resistant organisms, those of a low expression class were selected (19), with oxacillin MICs ranging from 4 to 128 µg/ml; for the susceptible strains, organisms for which the oxacillin MICs were 1 to 8 µg/ml when tested by agar dilution using 4% NaCl were included. All strains used in phases 1 and 2 had been tested by PCR for the presence of the mecA gene and by population analysis (for oxacillin-resistant strains only) to determine the level of expression of oxacillin resistance (19). The organisms fell into four groups. Group 1 (n = 19) comprised mecA positive strains determined to be in expression class 1 or 2, i.e., they shared a high degree of heterogeneity. Group 2 (n = 16) comprised mecA-negative strains for which agar dilution MICs of oxacillin were 4 µg/ml when tested with 4% salt (5) but were 2 µg/ml with 2% salt. Group 3 (n = 5) comprised mecA-negative strains for which the oxacillin MICs were 4 µg/ml with both 4 and 2% salt. The five strains in group 3 were considered to have the "MOD" phenotype; two of the strains had been characterized previously and had penicillin-binding proteins (PBPs) with modified affinity to penicillin but did not contain PBP2a, the mecA gene product (18). Three additional strains in group 3 were classified as "MOD" only phenotypically, i.e., the oxacillin MICs for these strains were >2 µg/ml and were not lowered with the addition of clavulanic acid, and these strains did not contain mecA. The "MOD" strains were included to see how they would test; however, it was decided before beginning the study that because of the unknown prevalence of such strains, their unknown clinical significance, and the lack of molecular characterization of all the strains, the results of their testing would be considered separately in data analysis. The last group (group 4 [n = 20]) comprised mecA-negative strains for which the oxacillin MICs were 1 to 2 µg/ml with 4% salt.

Quality control strains. Four S. aureus strains were used as controls on each day of testing: ATCC 43300, a heteroresistant strain; ATCC 33591, a homogeneously resistant strain; and two oxacillin-susceptible strains, ATCC 25923 and ATCC 29213.

Oxacillin screen agar medium. For phase 1, plates containing 4% salt and 6 µg of oxacillin (Sigma, St. Louis, Mo.)/ml were prepared at CDC using the standard reference lot of Mueller-Hinton agar that is used by manufacturers to standardize their production lots (12). For phase 2, commercial plates were purchased from five different manufacturers, i.e., BD Biosciences (BBL, Cockeysville, Md.), PML Microbiologicals (Wilsonville, Oreg.), Hardy Diagnostics (Santa Maria, Calif.), Gibson Laboratories, Inc., (Lexington, Ky.), and Remel (Lenexa, Kans.); three manufacturers' plates were tested at MGH, and two manufacturers' plates were tested at CDC. In addition, for phases 2 and 3, plates were prepared and tested at CDC using Mueller-Hinton II agar (BBL). All plates were stored at 4 to 8°C for no longer than 4 weeks or until the manufacturer's expiration date. In the latter part of phase 2 and in phase 3, additional lots of commercial plates were obtained from the two most commonly used manufacturers (BBL and Remel).

Inoculation methods. Inoculum suspensions were prepared from 18- to 24-h cultures grown on TSA-SB and adjusted to equal a 0.5 McFarland standard. Four inoculation methods were studied during phase 1: (i) spotting an area 10 to 15 mm in diameter using a cotton swab that was dipped in the suspension and from which excess fluid was expressed, (ii) streaking a quadrant of the plate using a swab prepared as above, (iii) spotting an area 10 to 15 mm in diameter using a 1-µl disposable loop, and (iv) spotting 10 µl using a micropipette. Four organisms were tested per plate, with only one inoculation method used per plate to avoid biasing the reading between plates. For phases 2 and 3, testing was done by spotting an area 10 to 15 mm in diameter using a 1-µl disposable loop. All plates were incubated at 35°C and read after 24 h.

Reading. All readings were done by two independent readers using transmitted light at both sites, except for the testing of additional lots of media at the end of phase 2, where one reader only was used. All readers had had previous experience in reading oxacillin screen plates. Plates were read as positive (if confluent growth was observed), weak positive (if the growth was almost confluent but hazy or >20 colonies), the number of colonies (if <20), or negative. However, for data analysis purposes, any growth of >1 colony was considered positive.

Phase 1. In phase 1, the abilities of the four inoculation methods to detect resistant strains were essentially equivalent (Table 1), with the method that would be expected to deliver the largest inoculum (the 10-µl pipette) achieving 100% sensitivity among all readers and laboratories. All the mecA-positive strains were detected by the other three methods by all readers in both laboratories, except for two strains (strains 107 and 351) which were not detected by all three methods in one laboratory. For these strains oxacillin MICs were 4 µg/ml by both conventional broth microdilution (2% salt) and agar dilution testing with increased salt (4%), and these strains were determined to be among the most heterogeneous of those tested (expression class 1; data not shown). Although the 10-µl-pipette inoculation method gave the best sensitivity across laboratories and readers, it resulted in the growth of a large number of susceptible strains (i.e., false positives), which resulted in a specificity of 88.9% in the best case and as low as 58% in the worst case. The swab and loop methods performed better in terms of specificity, ranging from 94 to 97%. The best combination of sensitivity (95 to 100%) and specificity (97%) among all readers and all laboratories was obtained with the 1-µl loop. However, the performances of all methods, except the 10-µl pipette, were not significantly different (P > 0.05) from each other (McNemar's chi-square test).

Phase 2. Of the five commercial lots tested (Table 2), the lowest sensitivity (84%) was obtained with the Gibson medium, although the performance of the in-house lot was similar (84 to 90%). There was no statistically significant difference among the sensitivity results when readers or medium lots were compared (P > 0.05 by McNemar's chi-square test). If the two highly heterogeneous strains (strains 107 and 351) are excluded, all the phase 2 media detected the remaining 17 resistant strains except for the lot from Gibson, which failed to grow one of the remaining 17 resistant strains (as noted by both readers), and the in-house lot, which was read as negative for one strain by one reader only. The Gibson lot also failed to grow the resistant control strain, ATCC 43300, and therefore results from that round of testing on that medium should not have been reported in a clinical laboratory.

Phase 3. In order to determine how well the media would perform in a routine fashion (presumably with less-challenging strains), additional lots of both BBL and Remel media (and a new in-house lot) were tested against a group of susceptible organisms for which the oxacillin MICs did not cluster around the resistance breakpoint and a subset of the susceptible challenge organisms from phases 1 and 2. For the susceptible strains, the specificity of the Remel lot improved but still remained below 90% (range, 84 to 88%). For eight mecA-positive strains, the sensitivity was 100% with all three media tested (data not shown).

Testing strains with modified PBPs. Results for the MOD strains tested in phases 1 and 2 were dependent on the oxacillin MIC. Strains for which MICs were 8 µg/ml were more likely to give positive results than those for which MICs were 4 µg/ml and which displayed variable growth depending on the laboratory and reader (data not shown).

Quality control. In phase 1, both oxacillin-resistant organisms grew with all inoculation methods. On two occasions, weak growth of one of the susceptible control strains, S. aureus ATCC 25923 (the strain recommended for disk diffusion quality control), was recorded as positive by one of the readers in one of the laboratories when the 10-µl inoculum was used. In phase 2, one of the commercial lots failed to grow S. aureus ATCC 43300, and growth of ATCC 25923 was detected on the PML lot by one reader. Growth of ATCC 25923 was also detected during phase 3 in one laboratory by both readers on one of two days of testing using Remel's medium. On this day growth was noted for both the quality control strain and the blinded quality control strain.