Misclassification of Susceptible Strains of Staphylococcus aureus as Methicillin-Resistant S. aureus by a Rapid Automated Susceptibility Testing System

Abstract

Eight Staphylococcus aureus strains initially identified by Vitek GPS-BS or GPS-SA cards as resistant to oxacillin, but susceptible to most non-beta-lactam antibiotics, were found on further testing to be susceptible to oxacillin and ceftizoxime by disk diffusion tests. For all these strains, the MICs of oxacillin were ≤0.5 μg/ml by agar dilution tests, and the strains were oxacillin susceptible when tested by the BBL Crystal MRSA ID and a Vitek machine with GPS-101 cards. None grew on oxacillin-salt agar screening plates. None were positive for mecA gene sequences by PCR. When S. aureus strains tested by Vitek GPS-SA or GPS-BS cards appear resistant to only penicillin and oxacillin, a confirmatory test such as the oxacillin-salt agar screening method should be performed.

For many years, the multidrug resistance pattern exhibited by most methicillin-resistant Staphylococcus aureus (MRSA) strains has helped differentiate them from mecA-negative strains of S. aureus that appear resistant to oxacillin by disk diffusion or have borderline susceptibility to oxacillin (MICs of 2 μg/ml), which are seldom multidrug resistant. However, MRSA strains resistant to penicillin and oxacillin but susceptible to agents such as erythromycin, clindamycin, gentamicin, and ciprofloxacin have been described recently (9, 18). The occurrence of such strains emphasizes the need for methods that differentiate MRSA from non-MRSA strains.

During a recent survey of S. aureus colonization among 600 emergency room patients at a tertiary hospital in Brazil, 10 of 149 S. aureus isolates recovered from anterior nares cultures were initially identified by a Vitek machine as MRSA (resistant to penicillin, oxacillin, and other beta-lactams) (17). Only 4 of the 10 isolates classified as MRSA were multidrug resistant and yielded growth on oxacillin-salt agar screen plates. The remaining six isolates initially classified as MRSA were susceptible to most non-beta-lactam antibiotics including clindamycin, gentamicin, and ciprofloxacin. During the same time, a hospital in California reported to us that two isolates initially identified as MRSA (oxacillin resistant) by a Vitek machine also were susceptible to most non-beta-lactam agents. Because these isolates had antimicrobial susceptibility patterns that are not seen frequently among MRSA strains, we questioned whether these isolates were truly MRSA.

The six S. aureus isolates recovered from six emergency room patients in Brazil and the two isolates recovered from a patient in California were included in the study. Control strains included S. aureus ATCC 25923, ATCC 29213, and ATCC 43300. Three multidrug-resistant clinical MRSA isolates (one from Miriam Hospital [Providence, R.I.] and two from Brazil) were included as control strains.

Initial susceptibility tests on the six S. aureus isolates recovered at the tertiary hospital in Brazil were performed with a Vitek machine (version WSVTK R05.03) with GPS-BS gram-positive test panels according to the manufacturer’s instructions. Quality control strains were tested routinely and yielded expected results. The two isolates from California were initially tested by using a Vitek machine (version WSVTK R05.03) and GPS-SA gram-positive test panels. Subsequently, study isolates were tested at the Veterans Affairs Medical Center in Providence, R.I., with a Vitek machine (version WSVTK R05.03) and GPS-101 gram-positive test panels.

Additional antimicrobial susceptibility tests were performed on study isolates at the tertiary hospital in Brazil and at Miriam Hospital by using commercially prepared disks and disk diffusion methods recommended by the National Committee for Clinical Laboratory Standards (15). The six study isolates from Brazil were also tested at the tertiary hospital in Brazil by using the BBL Crystal MRSA ID system according to the manufacturer’s recommendations. The MICs of oxacillin for the eight study isolates were determined at Miriam Hospital by standard methods (16). Isolates were screened for oxacillin resistance on Mueller-Hinton agar plates containing 6 μg of oxacillin per ml plus 4% NaCl by recommended methods (16). Agar screening plates used in Brazil were produced in-house, while those used at the Miriam Hospital were commercially prepared (Remel, Lenexa, Kans.). Disk diffusion, agar dilution, and oxacillin-salt screening assays were all incubated at 35°C for 24 h.

Four of the six study isolates from Brazil and both isolates from California were tested for the presence of mecA gene sequences by PCR methods. Two separate sets of PCR primers described earlier were used: AAT CGA TGG TAA AGG TTG GC and TTC TGC AGT ACC GGA TTT GC (14) and CTT TGC TAG AGT AGC ACT CG and GCT AGC CAT TCC TTT ATC TTG (9). Control strains included ATCC 25923, ATCC 29213, ATCC 44300, and three multidrug-resistant MRSA clinical isolates. PCR assays were performed by the Department of Pathology, Miriam Hospital.

All eight study isolates initially classified as MRSA by the Vitek system were resistant to penicillin but susceptible to oxacillin, amoxicillin-clavulanic acid, and ceftizoxime by standardized disk diffusion tests. Oxacillin zone diameters ranged from 16 to 18 mm when tested in Brazil and from 14 to 19 mm when tested at Miriam Hospital. All isolates were susceptible to clindamycin, gentamicin, and ciprofloxacin. All study isolates from Brazil tested by using the BBL Crystal MRSA ID system were susceptible to oxacillin. For all study isolates, MICs of oxacillin were <2 μg/ml (susceptible) when tested in a Vitek machine with GPS-101 test panels at the Veterans Affairs Medical Center. MICs for MRSA control strains were 8 μg/ml or greater. Agar dilution oxacillin MICs for the study isolates ranged from 0.25 to 1.0 μg/ml (susceptible), with oxacillin MICs for all but one isolate being 0.5 μg/ml or less. None of the eight study isolates yielded growth on oxacillin-salt agar screening plates. The four study isolates from Brazil and two isolates from California that were tested were negative for mecA gene sequences by PCR methods. All MRSA clinical control strains yielded amplicons the same as those of S. aureus ATCC 43300, while known methicillin-susceptible control strains were negative.

Differentiating MRSA strains from other strains of S. aureus is important because serious MRSA infections often require treatment with vancomycin, whereas infections caused by methicillin-susceptible strains and mecA-negative borderline-resistant strains can be treated with beta-lactam agents (3). Misidentification of methicillin-susceptible S. aureus as MRSA will promote unnecessary (and inappropriate) use of vancomycin. Also, misclassification of methicillin-susceptible S. aureus strains as MRSA will result in implementation of special MRSA isolation and barrier precautions, which are not warranted for patients with susceptible strains.

The heterogeneous resistance to methicillin and oxacillin manifested by many strains of MRSA made detection of these organisms problematic for early versions of rapid automated susceptibility testing systems (1, 2, 6, 10, 20). Early versions of the Vitek system had difficulty detecting MRSA (4, 6, 13, 20), but modifications led to improved detection (7, 8). However, some modifications of GPS-M cards led to false reports of oxacillin resistance among S. aureus strains (7). Early evaluations comparing a DNA probe for the mecA gene with Vitek GPS-SA cards revealed that false resistance to oxacillin among methicillin-susceptible strains occurred with only 0.4% of strains (19). Revisions to the software programs used with the Vitek system resulted in improved detection of MRSA by GPS-SA cards (11). However, one study of mecA-negative S. aureus strains with borderline susceptibility to oxacillin revealed that 10% of such strains were classified as MRSA by Vitek GPS-SA cards (12). For these strains, microdilution broth oxacillin MICs were ≥4 μg/ml, and the strains also yielded growth on oxacillin-salt screening plates.

In our study, a majority of the isolates misclassified as MRSA by Vitek GPS-BS or GPS-SA cards were not borderline resistant. Although a relatively small number of S. aureus strains were misclassified by Vitek GPS-BS cards in the hospital in Brazil, 6 (60%) of the 10 isolates initially classified as MRSA by the GPS-BS cards were shown not to be MRSA. As a result, the prevalence of MRSA among outpatients visiting the emergency room was exaggerated by the false resistance to oxacillin. If this phenomenon occurs in other hospitals where GPS-BS or GPS-SA cards are used, it could lead to inappropriate use of vancomycin among patients suspected of having community-acquired MRSA infections.

Our study did not determine the cause of the false oxacillin resistance results obtained with the Vitek system. The aberrant results were not due to inadvertent use of outdated GPS-BS or GPS-SA cards. The problem appears to be due to the specific card being used, rather than to the main components of the Vitek system, since all eight study isolates were accurately classified as oxacillin susceptible by Vitek GPS-101 cards. Also, except for unusual mecA-positive S. aureus strains for which MICs of oxacillin are ≤2 μg/ml, MRSA strains are detected accurately by Vitek GPS-503 cards (5).

These findings suggest that S. aureus isolates that appear resistant to oxacillin but susceptible to erythromycin, clindamycin, gentamicin, and ciprofloxacin should be tested further by a confirmatory method. PCR assays for the mecA gene are currently considered a “gold standard” for identifying MRSA (5). However, in laboratories where such assays are not available, oxacillin-salt agar screen plates are probably the best alternative (3, 5).