Antimicrobial Susceptibility Testing for Staphylococcus lugdunensis

ABSTRACT

Evaluation of penicillin and oxacillin susceptibility testing was conducted on 200 Staphylococcus lugdunensis isolates. Disc diffusion with penicillin 1 IU (P1, EUCAST) and penicillin 10 IU (P10, CLSI) was compared with nitrocefin discs (Cefinase) and automated broth microdilution (Vitek 2). Oxacillin susceptibility was extrapolated from cefoxitin (FOX; 30 μg) disc diffusion and compared with Vitek 2 results. The reference methods were blaZ and mecA PCR. Penicillin zone diameter and zone edge correlated with blaZ PCR results in all except two P10-susceptible isolates (very major error [VME]) and one P1-resistant isolate (major error [ME]). A total of 148 isolates were blaZ negative, of which 146 and 149 isolates were susceptible by P1 and P10, respectively. A total of 127 were penicillin susceptible by Vitek 2. Vitek 2 overcalled resistance in 21 blaZ-negative, 20 P1-susceptible, and 22 P10-susceptible isolates (Vitek 2 ME rate, 14.2%). Two mecA-positive isolates were oxacillin resistant by FOX disc and Vitek 2 methods (categorical agreement). However, 18 FOX-susceptible mecA-negative isolates tested resistant by Vitek 2. In conclusion, Vitek 2 overestimated penicillin and oxacillin resistance compared with disc diffusion and PCR results. In our study, disc diffusion with zone edge interpretation was more accurate and specific than automated broth microdilution for S. lugdunensis.

INTRODUCTION

Staphylococcus lugdunensis is a skin commensal that can resemble Staphylococcus aureus in virulence and pathogenicity (1). Although it is a common cause of skin and soft tissue infections (2, 3), the disease spectrum of S. lugdunensis can also include bacteremia with endocarditis (4), native and prosthetic osteoarticular infections (5), central nervous system infections (6–8), peritonitis (9), pneumonia (10), endophthalmitis (11), and urinary tract infections (12). S. lugdunensis is a coagulase-negative, Gram-positive coccus that is frequently susceptible to narrow spectrum β-lactams such as benzylpenicillin (3, 13–15). The worldwide prevalence of β-lactamase-producing, blaZ-positive, penicillin-resistant S. lugdunensis isolates is 7% to 40% (16). Oxacillin resistance mediated by penicillin-binding protein (PBP2A) encoded by mecA is uncommon (17–21).

Antimicrobial susceptibility testing (AST) for S. lugdunensis uses the same breakpoints as those for S. aureus, as this method better reflects the presence of mecA and oxacillin resistance. EUCAST and CLSI guidelines differ; however, with respect to the penicillin concentration for disc diffusion. EUCAST recommends 1-IU discs (P1) while CLSI recommends 10-IU discs (P10). Evaluation of the penicillin disc zone edge in both S. aureus and S. lugdunensis is recommended by both EUCAST and CLSI and is a surrogate marker for blaZ-mediated penicillin resistance (22–24). Due to low sensitivity, the use of chromogenic cephalosporin-based tests for the detection of β-lactamases is not recommended by EUCAST (25). Oxacillin susceptibility is extrapolated from disc diffusion using cefoxitin (FOX) 30 μg for both S. lugdunensis and S. aureus. The performance of automated broth microdilution (BMD) for S. lugdunensis is unclear. False-positive penicillin and oxacillin results by Vitek 2 (bioMérieux, Durham, NC) have been reported; however, these studies were limited by small sample sizes and did not include data on disc zone edge interpretation (26, 27).

To determine the optimal method for AST in S. lugdunensis, we evaluated penicillin susceptibility testing by comparing disc diffusion at two different concentrations (P1 and P10) with automated BMD (Vitek 2) and chromogenic cephalosporinase testing using nitrocefin discs (Cefinase; Becton, Dickinson and Company, Sparks, MD). Oxacillin susceptibility testing was evaluated by comparing FOX disc diffusion with automated BMD (Vitek 2 oxacillin MIC and Vitek 2 cefoxitin screen). blaZ and mecA PCR were used as the reference methods.

(Preliminary data from this study were presented at the Australian Society for Antimicrobials [ASA] Annual Scientific Meeting in Melbourne, Australia, 27 to 29 February 2020 [Poster 23, awarded the Best Poster Prize].)

MATERIALS AND METHODS

Staphylococcus lugdunensis isolates.

Two hundred nonduplicate, clinical isolates identified as S. lugdunensis by matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS; Bruker Daltronics, Billerica, MA) between February 2018 and February 2020 at the SA Pathology microbiology laboratory at the Flinders Medical Centre in Adelaide, South Australia, were included. Retrospectively evaluated isolates collected from February 2018 to August 2019 were stored at −80°C in cryopreservation fluid and glycerol beads (Protect microorganism preservation system; Thermo Fisher Scientific, Cleveland, OH). Stored isolates were recultured onto horse blood agar (HBA; bioMérieux) and incubated at 35°C under aerobic conditions for 18 h. Organism identity was reconfirmed by MALDI-TOF MS prior to AST. Isolates not identified as S. lugdunensis were excluded. Isolates collected between September 2019 and February 2020 were tested and evaluated prospectively.

AST via disc diffusion using the Kirby-Bauer method.

Suspensions (0.5 McFarland standard) of S. lugdunensis from HBA plates were inoculated onto Mueller-Hinton agar (MHA; bioMérieux), and antibiotic-impregnated discs including P1 and P10 (Oxoid, Thermo Fisher Scientific) were applied. Following incubation at 35°C under aerobic conditions for 18 h, zones of inhibition were measured using electronic calipers and interpreted with reference to both EUCAST and CLSI criteria (28, 29). Isolates with a P1 zone diameter of ≥26 mm or a P10 zone diameter of ≥29 mm and fuzzy (“beach”) zone edges were regarded as penicillin susceptible. Isolates with a P1 zone diameter of <26 mm or a P10 zone diameter of ≤28 mm with sharp (“cliff”) zone edges were regarded as penicillin resistant. All isolates with a disc zone diameter of >40 mm were recorded as “40 mm.”

β-Lactamase testing.

Bacterial growth from around the P1 disc edge was applied to Cefinase discs moistened with sterile water. β-Lactamase activity was present if hydrolyzation of the amide bond in the β-lactam ring led to a color change from cream to red after up to 1 h at room temperature, per the manufacturer’s instructions. The quality control strains Haemophilus influenzae (ATCC 49766, β-lactamase negative) and S. aureus (ATCC 29213, β-lactamase positive) were concurrently tested for β-lactamase activity.

AST by the Vitek 2 automated microbial identification system.

Automated BMD using Vitek 2 AST-P656 cards was performed on all S. lugdunensis isolates per the manufacturer’s instructions. Isolates with a Vitek 2 penicillin MIC of >0.125 μg/mL were penicillin resistant per EUCAST and CLSI criteria (28, 29).

blaZ PCR testing.

All 200 S. lugdunensis isolates were placed in MagNA Pure 96 external lysis buffer (Roche Diagnostics, Indianapolis, IN), extracted using a MagNA Pure 96 system (Roche), and examined for blaZ via real-time PCR with methods adapted from Pereira et al. (30). S. aureus ATCC 29213 and S. aureus ATCC 25923 were included as blaZ-positive and blaZ-negative controls, respectively.

Oxacillin susceptibility testing.

Methods and results for oxacillin susceptibility testing by FOX disc diffusion, Vitek 2, and mecA PCR are included in the supplemental material.

Quality control.

Quality control was performed in accordance with the National Association of Testing Authorities (NATA) Laboratory Practice Guidelines and was within expected limits for P1, P10, FOX, and nitrocefin discs, for the AST cards, and for the selective culture media used in our study.

Resolution of discrepancies.

Disc diffusion was repeated for isolates with discordant zone diameter and zone edge results and when zone diameter readings were within 1 mm of the recommended EUCAST and CLSI breakpoints. Disc diffusion was repeated for isolates with fuzzy zone edges but P1 zone diameters of 25 mm (2 isolates, resistant), 26 mm (10 isolates, susceptible) and 27 mm (7 isolates, susceptible), and for one isolate with a sharp zone edge and a P10 zone diameter of 28 mm (resistant). On retesting, two isolates with fuzzy zone edges and an initial P1 zone diameter of 25 mm were reclassified as susceptible with repeat zone diameters of 32 mm and 36 mm. These two isolates were included in the analyzed data set. All other isolates maintained categorical agreement with respect to disc zone diameter and disc zone edge on retesting. Six blaZ-negative isolates that were resistant by all other tested methods were retested. One isolate that remained blaZ-negative on retesting was subsequently found to be mecA positive. The remaining five isolates were blaZ positive on repeat testing and were included in the analyzed data set.

Data collection, entry, and analysis.

Data relating to the type and site of our isolates was collected. AST results were recorded and interpreted by two independent evaluators (J.S.K.T. and I.P.). Major error (ME) and very major error (VME) rates were calculated using cross-tabulation. Articles published in a foreign language were interpreted with Google Translate.

RESULTS

For the 200 isolates included in our study, S. lugdunensis was predominately cultured from the lower limbs (72 isolates, 36%), axillae/torso (38 isolates, 19%), and groin/perineum (32 isolates, 16%). A total of 11 isolates (5.5%) were from sterile sites and included blood cultures and hip, bursa, and pleural tissue.

Penicillin susceptibility testing.

Penicillin susceptibility differed by method (Fig. 1).

FIG 1.

Results of penicillin susceptibility testing by method.

Nineteen isolates were resistant by Vitek 2 with a penicillin MIC of 0.25 μg/L. Of these isolates, 18 were blaZ negative and susceptible by P1 and P10, and only one isolate was blaZ positive. A total of 54 isolates were resistant by Vitek 2 with a penicillin MIC of 0.5 μg/L. Of these isolates, 51 had concordant Vitek 2, P1, and P10 results and were blaZ positive. Of the three discordant results, one involved a blaZ-positive isolate that was resistant by P1 (15 mm, sharp) and Vitek 2 (penicillin MIC of 0.25 μg/mL) but susceptible by P10 (30 mm, fuzzy; very major error [VME]) and two involved blaZ-negative isolates that were susceptible by P1 (both 28 mm, fuzzy) and P10 (31 mm and 36 mm, fuzzy) but resistant by Vitek 2 (penicillin MIC of ≥0.25 μg/mL; major error [ME]). In summary, 20 P1-susceptible, 22 P10-susceptible, and 21 blaZ-negative isolates were incorrectly identified as resistant by Vitek 2.

Figure 2 illustrates the distribution in penicillin disc zone diameters by blaZ PCR status. For blaZ-negative isolates, the median P1 zone diameter was 30 mm, compared with ≥40 mm for P10. For blaZ-positive isolates, the median P1 zone diameter was 10 mm, compared with 18 mm for P10.

FIG 2.

Penicillin 1 IU and penicillin 10 IU compared using blaZ PCR. A total of 66 P10 isolates had disc diameters of >40 mm (indicated by a star; data not shown). The vertical dotted line indicates the zone diameter breakpoint. Susceptible, P1 ≥ 26 mm, P10 ≥ 29 mm; resistant, P1 < 26 mm, P10 ≤ 28 mm. P1, penicillin 1 IU disc (EUCAST); P10, penicillin 10 IU disc (CLSI).

Susceptibility results by P1 and P10 demonstrated 98% concordance with blaZ results. Table 1 lists the assigned susceptibility result by P1, P10, Vitek 2, and Cefinase, compared with blaZ PCR as a reference method, and the associated ME and VME rates. P1 correctly identified all blaZ-positive isolates with no VME. Only two MEs were seen with P1 (ME rate of 1.4%). The first P1 ME involved a blaZ-negative isolate that was susceptible by P10 (30 mm, fuzzy) and Vitek 2 (penicillin MIC of 0.125 μg/mL) but resistant by P1 (24 mm, sharp). The second P1 ME involved an isolate that was penicillin resistant due to the presence of mecA despite being blaZ-negative that was resistant by P1 (6 mm, sharp). This isolate was susceptible by P10 (30 mm, fuzzy), which would constitute a VME if mecA-mediated penicillin resistance was considered. Other VMEs involving P10 included two blaZ-positive, P1 (15 mm and 23 mm, sharp) and Vitek 2-resistant (penicillin MIC of ≥0.25 μg/mL) isolates that were P10 susceptible (30 mm and 32 mm, fuzzy; VME rate of 3.8%).

TABLE 1.

Penicillin susceptibility testing methods compared via blaZ PCR for detection of penicillin resistance

Penicillin susceptibility	Penicillin susceptibility by testing methoda:
	P1 disc diffusion			P10 disc diffusion			Vitek 2 penicillin MIC			Cefinase
	CA	ME	VME	CA	ME	VME	CA	ME	VME	CA	ME	VME
Susceptible (blaZ negative, n = 148)	146	2 (1.4%) b	0	147	0	1 (0.7%) b	127	21 (14.2%)	0	148	0	0
Resistant (blaZ positive, n = 52)	52	0	0	50	0	2 (3.8%)	52	0	0	38	0	14 (26.9%)

^aBoldface entries correspond to the very major errors and major errors for the different susceptibility testing methods. CA, categorical agreement; ME, major error (false resistance); VME, very major error (false susceptibility).

^bPenicillin resistance mechanism for one blaZ-negative isolate that was resistant by P1 (ME) and susceptible by P10 (VME) was attributed to mecA.

A total of 148 isolates (74%) were Cefinase- and blaZ-negative. Of these isolates, 147 were also susceptible by P1, P10, and Vitek 2. A total of 14 blaZ-positive isolates that were penicillin resistant by all other methods were Cefinase negative (VME rate of 26.9%).

DISCUSSION

In our local review on susceptibility testing for S. lugdunensis, we detected penicillin resistance and blaZ positivity rates of 25% among 200 isolates, which correlates with previously published rates from Denmark (3), Sweden (13, 31), and Singapore (19). Higher rates of blaZ positivity have been reported elsewhere. A subanalysis of 92 isolates by McHardy et al. reported blaZ positivity of 39% and penicillin resistance of up to 45% in the United States (15). Hagstrand Aldman et al. and Batista et al. reported blaZ positivity of 33% from 112 isolates in Sweden (32) and 40% from 60 isolates in Spain, respectively (26). Kachrimandiou et al. reported penicillin resistance of 51% from 55 isolates in Greece; however, they did not perform blaZ PCR testing (33).

With respect to penicillin susceptibility testing for S. lugdunensis, our study is unique from the perspective of the comparison we have made between two different penicillin disc concentrations per EUCAST (P1) and CLSI (P10), the inclusion of penicillin disc zone edge evaluation in AST, and the use of molecular diagnostics, namely, blaZ PCR, as a reference method. Both EUCAST and CLSI no longer endorse the use of chromogenic cephalosporin-based tests for the detection of β-lactamase and instead recommend evaluation of the penicillin disc zone edge in both S. aureus and S. lugdunensis as a surrogate marker for penicillin resistance and blaZ positivity (25). This recommendation is corroborated by the Cefinase VME rate of 26.9% in our study. Regarding penicillin disc zone edge evaluation, Gill et al. demonstrated that a sharp (“cliff”) penicillin disc zone edge correlated with β-lactamase activity in 100% and 93% of S. aureus and Staphylococcus epidermidis isolates, respectively (23). Others have reported lower accuracy of the P10 zone edge (sensitivity, 89%) and zone diameter (sensitivity, 66%) compared to that of the P1 zone edge and zone diameter (sensitivity, 100%) for the detection of β-lactamase activity and blaZ in S. aureus (22). A specificity of 9% was noted by McHardy et al. with respect to the P10 zone edge. This was attributed to a large number of sharp P10 zone edges, regardless of blaZ PCR status, in S. lugdunensis isolates (15). In contrast, Hagstrand Aldman et al. found 100% concordance between P10 disc zone edge appearances and blaZ PCR status. Our study demonstrated 98% concordance between the penicillin disc zone edge appearance (P1 and P10) and blaZ PCR status. This supports current practice at our institution, where routine assessment of the penicillin disc zone edge is regarded as a reliable screening tool for penicillin resistance. We did not specifically record instances where the interpretation of the disc zone edge differed between readers. However, P1 had more definite demarcation of the zone edge, thus rendering interpretation of the zone edge and zone diameters easier than that for P10, especially when close to the EUCAST and CLSI breakpoints. The median P1 and P10 zone diameters for susceptible isolates incorrectly labeled resistant by Vitek 2 were both 30 mm and were much closer to the breakpoint for P10 (29 mm; 1-mm difference) than to the breakpoint for P1 (26 mm; 4-mm difference).

Overall, P1 demonstrated slightly superior accuracy to that of P10, with no VME. P1 correctly identified all penicillin-resistant isolates except for one isolate that was blaZ-negative but penicillin-resistant due to mecA. After excluding this isolate, the corrected P1 ME rate of 0.7%, as well as the uncorrected P1 ME rate of 1.4%, were both within the accepted standard of <3%. In contrast, P10 had a VME rate of 3.8%, attributed to two blaZ-positive, P1-resistant isolates that were incorrectly labelled susceptible by P10, which was higher than the accepted standard VME rate of <1.5% (34). Reporting and selection bias were minimized in our study, as the investigators were unaware of the blaZ PCR status of our isolates until after disc diffusion, automated BMD, and Cefinase testing had been performed.

Overestimation of penicillin resistance by automated BMD using Vitek 2 compared to that determined by blaZ PCR was evident. The Vitek 2 ME rate for penicillin susceptibility testing among our S. lugdunensis isolates was 14.2%. Batista et al. also noted a trend in which Vitek 2 overestimated the penicillin MIC by 41% compared with conventional broth microdilution (26). Similarly, we demonstrated that the discrepancy between disc diffusion results and automated BMD was most pronounced when the Vitek 2 penicillin MIC was 0.25 μg/L. This discrepancy was less apparent when the Vitek 2 penicillin MIC was 0.5 μg/L. At this dilution, only two blaZ-negative isolates that were susceptible by P1 and P10 were misclassified as Vitek 2 resistant (ME rate of 1.4%).

Evaluation of oxacillin susceptibility testing for S. lugdunensis was a secondary aim of our study (refer to the supplemental material). We acknowledge that the applicability and generalizability of our oxacillin data are limited by the small subset of 40 isolates for which mecA PCR was performed and by the low rate of oxacillin resistance of 1% at our institution. Rates of oxacillin resistance above 8% have been reported elsewhere (15, 19, 31, 33). Detection of oxacillin resistance using the Vitek 2 oxacillin MIC, Vitek 2 cefoxitin screen, and mecA PCR are thought to have sensitivities and specificities of above 95% in coagulase-negative staphylococci. John et al. reported 100% sensitivity and specificity of all three aforementioned methods in 41 isolates of S. lugdunensis (35). In our study, Vitek 2 overcalled oxacillin resistance in 18 FOX-susceptible, mecA-negative isolates. Batista et al. reported on similar findings involving Vitek 2 oxacillin testing for mecA-negative S. lugdunensis isolates (26). We demonstrated that a positive Vitek 2 cefoxitin screen and a Vitek 2 oxacillin MIC of 4 μg/L were both poorly predictive of oxacillin resistance and of the presence of mecA. Instead, a correlation between FOX and mecA PCR was detected that requires validation.

The ME and VME identified in our study with respect to susceptibility testing for S. lugdunensis deviate from the expected performance of automated BMD and warrant further evaluation (28). Given the overestimation of penicillin resistance by automated BMD, we propose that in the absence of readily available confirmatory blaZ testing, a penicillin MIC result of 0.25 μg/L obtained via automated BMD may benefit from retesting to minimize the risk of ME, particularly if noted to be discordant with a susceptible P1 zone diameter and zone edge result. It is uncertain if automated BMD also overestimates penicillin and oxacillin resistance in other staphylococci, such as S. aureus, or if the same discrepancy is seen with non-β-lactam antibiotics in S. lugdunensis. Penicillin susceptibility of 75% and low oxacillin resistance of 1% were reported in our study, which should encourage preferential use of narrow-spectrum β-lactams for the treatment of S. lugdunensis infections in Australia. The clinical implications of reported false resistance include inappropriately broad prescribing for an otherwise susceptible organism and emergence of antimicrobial resistance through selective pressure.

A key finding from our study was excellent correlation (98%) between P1 and P10 zone diameter, zone edge appearance, and blaZ PCR. Based on our data, we advise that for isolates of S. lugdunensis that are penicillin susceptible by disc diffusion and have fuzzy (“beach”) disc zone edges, subsequent testing via automated BMD to confirm penicillin susceptibility may not be required. Instead, automated BMD may be more useful to confirm penicillin resistance when isolates of S. lugdunensis are resistant by disc diffusion and have sharp (“cliff”) disc zone edges. Where possible, establishing blaZ status is prudent to adjudicate discordance between phenotypic methods and to confirm β-lactam susceptibility, particularly for invasive infections involving S. lugdunensis, in which time above the MIC is critical to achieving a clinical and microbiological cure.