The use of a cefoxitin disk test to detect staphylococci that are likely to contain the mecA
gene has been widely advocated since the test was first suggested (3
) and has been adopted by antimicrobial susceptibility testing organizations worldwide (http://www.bsac.org.uk/_db/_documents/version_6.1.pdf
). For laboratories that do not use disk diffusion as their primary testing method, performing disk diffusion requires additional expense and reagents. Consequently, we performed studies to determine if a cefoxitin broth microdilution MIC breakpoint would correlate with the presence of the mecA
gene in staphylococci.
Cation-adjusted Mueller-Hinton broths (CAMHB) from three manufacturers (BBL, BD Diagnostic Systems, Sparks, MD; Difco, BD Diagnostic Systems, Sparks, MD; and Oxoid, Basingstoke, Hampshire, England) (cation content adjusted if necessary), were used to prepare frozen microdilution panels containing cefoxitin concentrations of 0.5 to 32 μg/ml (including 6 μg/ml) and oxacillin concentrations of 0.06 to 16 μg/ml (in BBL CAMHB only). The frozen panels were shipped to all participants, along with 30-μg cefoxitin disks (BBL). Each laboratory used its current lot of Mueller-Hinton agar for disk diffusion testing. Participating laboratories included the Centers for Disease Control and Prevention (CDC), Atlanta, GA; BD Diagnostic Systems, Sparks, MD; bioMérieux, Inc., Hazelwood, MO; Duke University Hospital, Durham, NC; JMI Labs, North Liberty, IA; Massachusetts General Hospital, Boston, MA; Siemens Healthcare Diagnostics MicroScan, Inc., West Sacramento, CA; University of Rochester, Rochester, NY; Robert Wood Johnson Medical School, New Brunswick, NJ; Statens Serum Institut, Copenhagen, Denmark; and Trek Diagnostic Systems, Cleveland, OH.
Each laboratory tested approximately 50 unique clinical isolates from their own collection (30 Staphylococcus aureus isolates and 20 coagulase-negative staphylococci [CoNS]) and quality control strains for each test day (S. aureus ATCC 29213 [tested by MIC only] and S. aureus ATCC 25923 [tested by disk diffusion only]). Additional strains were tested at the CDC, totaling 167 S. aureus isolates and 22 CoNS. The total number of strains tested in all labs was 479 S. aureus isolates and 204 CoNS. All results were read after both 18-h and 24-h incubations. Endpoints were read as the lowest concentration at which no visible growth was observed. At 18 h of incubation, 97.3% of the MIC quality control results (for S. aureus ATCC 29213) and 97.9% of the disk diffusion quality control results (for S. aureus ATCC 25923) were within the range considered acceptable by the Clinical and Laboratory Standards Institute (CLSI).
After testing, all strains were shipped to the CDC for further analysis. PCR testing for mecA
was performed as previously described (8
) on the following strains: (i) any S. aureus
strain for which the cefoxitin MIC was <16 μg/ml in any of the three media tested (n
= 180); (ii) any S. aureus
strain for which the cefoxitin MIC was >8 μg/ml and the oxacillin MIC was <4 μg/ml in any of the three media tested (n
= 5); and (iii) a sample of the 304 isolates (n
= 16 [5%]) for which the cefoxitin MIC was >8 μg/ml and the oxacillin MIC was >2 μg/ml. All other S. aureus
strains for which the cefoxitin MIC was >8 μg/ml and the oxacillin MIC was >2 μg/ml at 24 h of incubation were presumed to be mecA
positive. All CoNS were tested by PCR for the presence of mecA
and identified to species level by using previously described methods (1
). For the purposes of data analysis, mecA
or presumed mecA
was used as the gold standard.
Cefoxitin MIC results for the S. aureus group (which included eight isolates of Staphylococcus lugdunensis) showed a clear separation of mecA-positive and mecA-negative strains between 4 and 6 μg/ml; most isolates with cefoxitin MICs of 6 μg/ml contained the mecA gene (Table ). Among 13 strains for which the cefoxitin MIC was 6 μg/ml in any medium at either incubation time period, 11 were mecA positive. All the S. lugdunensis strains tested were mecA negative and had cefoxitin MICs of ≤4 μg/ml. Using a cefoxitin breakpoint of ≤4 μg/ml for oxacillin-susceptible or mecA-negative isolates and that of ≥8 μg/ml (rounding 6 μg/ml up to 8 μg/ml) for oxacillin-resistant or mecA-positive isolates, sensitivity and specificity were excellent, 99.7 to 100%. If these breakpoints are applied to the isolate population in this study, there are some category discrepancies between cefoxitin susceptibility testing and oxacillin susceptibility testing. These were limited to isolates for which the cefoxitin MIC was ≥6 μg/ml and oxacillin MIC was <4 μg/ml. The number of isolates varied depending on the medium and time of incubation, with the lowest being three strains at 24 h in BBL CAMHB and the highest being seven strains at 18 h in Difco CAMHB. All were mecA positive except for two isolates. For the mecA-positive isolates, the oxacillin MICs were 1 or 2 μg/ml, and the cefoxitin MICs ranged from 6 to >32 μg/ml.
Cefoxitin MICs for 479 isolates of Staphylococcus aureus tested in three different CAMHBa
The sensitivity and specificity of cefoxitin disk diffusion testing in this study were also excellent, 99.7 and 100%, respectively (data not shown). These results are similar to those from a previous study when cefoxitin MIC testing was done to determine cefoxitin disk diffusion breakpoints (9
). In that 10-laboratory study using panels prepared with CAMHB of two manufacturers and the MIC breakpoints proposed here, the sensitivity was 98 to 99%, and specificity was 99 to 100% (9
Cefoxitin MIC results for CoNS showed no clear separation of mecA-positive and -negative strains (Table ). However, test sensitivity using breakpoints of ≤2 μg/ml and ≥4 μg/ml differed substantially between the media used, as follows: ≥99% at both 18 h and 24 h for Difco and Oxoid CAMHB, but only 89% at 18 h and 94% at 24 h for BBL CAMHB. However, specificity for all media at both times was ≤80%.
Cefoxitin MICs for 203 CoNS tested in three different CAMHB
Comparing the performance for CoNS grouped by Staphylococcus epidermidis
versus non-S. epidermidis
isolates (Table ), sensitivities were similar for both organisms groups (at 24 h, 94 to 99% for S. epidermidis
and 91 to 100% for non-S. epidermidis
). However, the specificity of the test was much better for S. epidermidis
than that for non-S. epidermidis
isolates (at 24 h, 85 to 91% for S. epidermidis
versus 54 to 69% for non-S. epidermidis
). Although the performance of the cefoxitin MIC test in this study for the prediction of mecA
in CoNS is not ideal, its use might be considered if its performance were no worse than that of the oxacillin MIC test. Although our results and those from the previous CLSI study (9
) showed that the two tests to have similar sensitivities (Table ), the specificity of cefoxitin (69 to 90%) was not as good as that of oxacillin (89 to 91%).
Comparison of sensitivity and specificity of cefoxitin MICs of ≤2 μg/ml for mecA negativity and ≥4 μg/ml for mecA positivity for three groups
Comparison of sensitivity and specificity of oxacillin and cefoxitin MICs for prediction of mecA presence in CoNS read at 24 h in two CLSI studies
For many laboratories, the cefoxitin MIC test can be incorporated into routine testing of S. aureus
when the drug is included in panels of automated antimicrobial susceptibility testing systems. Several automated susceptibility testing manufacturers have begun using cefoxitin MICs to predict mecA
-mediated resistance in Staphylococcus
spp. In a recent report, the cefoxitin MIC was a better predictor of mecA
-mediated resistance in CoNS than is reported here (4
). However, performance differences among commercial systems may reflect difference in the systems’ use of both oxacillin and cefoxitin data as well as the ability of device manufacturers to use a single-source optimized medium, whereas a reference method needs to work in media from multiple manufacturers.
Recently, two groups of investigators demonstrated that moxalactam may be a better indicator of mecA
in CoNS than is cefoxitin. One group recommended the use of a combination of oxacillin MIC and moxalactam MIC tests (7
); the other recommended the use of a combination of cefoxitin disk and moxalactam disk tests (5
). The use of moxalactam tests in combination with cefoxitin and/or oxacillin tests to predict the mecA
status of CoNS may deserve further study, given that the use of cefoxitin MICs alone is not optimal for this purpose.
In summary, in an 11-laboratory study, we found that cefoxitin broth microdilution breakpoints of ≤4 μg/ml for the prediction of mecA
-negative results and those of ≥6 or 8 μg/ml for mecA
-positive results in S. aureus
isolates read at 18 h of incubation were highly sensitive and specific. These breakpoints for S. aureus
were approved by the CLSI Subcommittee on Antimicrobial Susceptibility Testing and were included in the M100-S18 tables published in January 2008 (2
). For CoNS, we found that cefoxitin MIC levels were sensitive but not specific in identifying the presence of mecA
and, thus, recommend that they not replace oxacillin MICs at this time.