Comparative Performances of Vitek-2, Disk Diffusion, and Broth Microdilution for Antimicrobial Susceptibility Testing of Canine Staphylococcus pseudintermedius

ABSTRACT

Staphylococcus pseudintermedius is the primary cause of canine cutaneous infections and is sporadically isolated as a pathogen from humans. Rapidly emerging antibiotic-resistant strains are creating serious health concerns so that accurate and timely antimicrobial susceptibility testing (AST) is crucial for patient care. Here, the performances of the AST methods Vitek-2, disk diffusion (DD) and broth microdilution (BMD) were compared for the determination of susceptibility of 79 S. pseudintermedius isolates from canine cutaneous infections and one from human pyoderma to oxacillin (OXA), amoxicillin/clavulanate (AMC), cephalothin (CEF), gentamicin (GEN), enrofloxacin (ENR), doxycycline (DOX), clindamycin (CLI), inducible clindamycin resistance (ICR), mupirocin (MUP), and trimethoprim-sulfamethoxazole (SXT). Overall, the agreement of DD and Vitek-2 using the veterinary AST-GP80 card with reference BMD was ≥90%, suggesting reliable AST performances. While DD generated mainly minor errors and one major error for OXA, Vitek-2 produced one very major error for GEN, and it failed in identifying one ICR-positive isolate. Moreover, five bacteria were diagnosed as ICR-positive by Vitek-2, but they showed a noninduction resistance phenotype with manual methods. All S. pseudintermedius isolates were interpreted as susceptible or intermediately susceptible to DOX using CLSI breakpoints for human staphylococci that match the DOX concentration range included in AST-GP80. However, this could lead to inappropriate antimicrobial prescription for S. pseudintermedius infections in companion animals. Considering the clinical and epidemiological importance of S. pseudintermedius, we encourage updating action by the system manufacturer to address AST for this bacterium.

INTRODUCTION

Staphylococcus pseudintermedius is the bacterial agent most frequently found in canine clinical samples, mainly from skin and ears (1). Additionally, this bacterium can also occasionally cause infections in humans, which is more likely in dog owners (2). The high resistance to methicillin and other antimicrobials observed among S. pseudintermedius isolates is of concern because it worryingly limits treatment options (3). In such a scenario, accurate and timely antimicrobial susceptibility testing (AST) is crucial for patient care.

Advanced laboratories working in the field of human or veterinary diagnostics use automated systems for AST, including Vitek-2 (bioMérieux), due to easy application and cost-effectiveness.

Automated methods are preferred over the more laborious broth microdilution (BMD) and disk diffusion (DD) (4), although these manual techniques are validated as reference procedures for AST by the Clinical and Laboratory Standards Institute (CLSI) and the European Committee on Antimicrobial Susceptibility Testing (EUCAST) (5, 6). The Vitek-2 system is associated with the Advanced Expert System (AES), which is designed to analyze results generated by Vitek-2 according to the bacterial species. The reliability of Vitek-2 for AST was confirmed for several staphylococcal species, including Staphylococcus aureus and Staphylococcus epidermidis (4, 7, 8). However, Vitek-2 may produce erroneous results when testing antibiotic molecules against certain staphylococci (9, 10).

To our knowledge, the diagnostic accuracy of Vitek-2 has never been investigated on S. pseudintermedius, for which CLSI reports some unique breakpoints and guidelines to interpret the antibiotic efficacy, and neither has the intermethod concordance between manual and automated AST techniques.

Therefore, here, we evaluated the performance of the automated system Vitek-2, DD, and BMD in pairwise comparison for the efficacy determination of seven antibiotic classes against S. pseudintermedius isolates from cutaneous infections.

MATERIALS AND METHODS

Bacterial isolates.

A total of 80 S. pseudintermedius isolates collected in clinical microbiology laboratories in Central Italy from 2018 to 2020 were included in this study, 79 from dogs with cutaneous infections and 1 from a dog owner with pyoderma. Only one isolate per subject was collected. All isolates were cultured from −20°C storage onto mannitol salt agar supplemented with 5% (vol/vol) egg yolk emulsion. Identification of the isolates was performed to the species level using a PCR-restriction fragment length polymorphism (RFLP) approach. RFLP is based on the detection of the single restriction MboI site in locus pta discriminating S. pseudintermedius from other Staphylococcus intermedius group microorganisms (11). Before AST was performed, each bacterium was subcultured on cation-adjusted Mueller-Hinton agar (CAMHA). The quality-control strains used for AST were Staphylococcus aureus subsp. aureus Rosenbach ATCC 25923 and ATCC 29213.

Antimicrobial susceptibility testing.

Antimicrobial susceptibility testing was carried out to determine the susceptibility of S. pseudintermedius to β-lactams (oxacillin, OXA; amoxicillin/clavulanate, AMC; cephalothin, CEF), aminoglycoside (gentamicin, GEN), fluoroquinolone (enrofloxacin, ENR), tetracycline (doxycycline, DOX), sulfonamide (trimethoprim-sulfamethoxazole, SXT), lincosamide (clindamycin, CLI), and mupirocin (MUP).

Broth microdilution. BMD was performed in triplicate according to CLSI standards M07 (12) and VET01 (13) to define the lowest antimicrobial concentration that inhibits visible bacterial growth (MIC). First 96-well plates were inoculated with 100 μl of 2-fold serial dilutions of antibiotic in cation-adjusted Mueller-Hinton broth (CAMHB). For comparative purposes, final concentrations of the drugs covered the standard range tested by the Vitek-2 AST-GP80 card—OXA, 0.25 to 4 μg/ml; AMC, 2 to 32 μg/ml; CEF, 2 to 32 μg/ml; GEN, 0.5 to 16 μg/ml; ENR, 0.5 to 4 μg/ml; DOX, 0.5 to 16 μg/ml; SXT, 10 to 320 μg/ml (Sigma, St. Louis, MO, USA), CLI, 0.125 to 4 μg/ml; MUP, 0.125 to 256 μg/ml (Jinlan Pharma-Drugs Technology, Hangzhou, China). S. pseudintermedius colonies were resuspended in sterile medium, and the suspension turbidity was measured spectrophotometrically at an optical density at 600 nm (OD₆₀₀). The bacterial suspension was then adjusted in CAMHB to 5 × 10⁵ CFU per ml and vigorously vortexed. Plates were inoculated with 100 μl of bacterial suspension and incubated at 35°C for 16 to 20 h. To determine the MIC of OXA, the medium was supplemented with 2% (wt/vol) NaCl, and plates were further incubated at 35°C in ambient air until 24 h. Purity and inoculum size were checked by seeding the inoculum and its dilutions onto blood agar plates containing 5% sheep blood and subsequently counting CFU after 16 to 24 h. Negative (sterile CAMHB) and positive (S. pseudintermedius isolate in CAMHB without antibiotics) controls were included in each plate.

Disk diffusion. The antimicrobial susceptibility profiles of S. pseudintermedius isolates were investigated using the DD method performed according to CLSI document VET01 (13). S. pseudintermedius colonies were resuspended in sterile broth at a turbidity equivalent to the spectrophotometric 0.5 McFarland standard. The bacterial suspension was swabbed onto CAMHA plates within 15 min with a sterile cotton-tipped swab. Plates were allowed to dry for 5 min, and antimicrobial disks were placed on them. The disks contained the following amounts of antimicrobials: 1 μg OXA, 2 μg CLI, 5 μg ENR, 10 μg GEN, 25 μg SXT, 30 μg AMC, 30 μg CEF, 30 μg DOX, and 200 μg MUP (Liofilchem, Roseto degli Abruzzi, Italy). Zones of inhibition were read manually after incubation at 35°C for 16 to 18 h under aerobic conditions. The plates were then reincubated at 35°C in ambient air until 24 h to obtain final results for OXA.

Vitek-2. Each isolate was tested by using the Vitek-2 system (version 8.02) according to the manufacturer’s instructions. Antimicrobial susceptibility of the isolates was tested on a Vitek-2 instrument (bioMérieux, Inc., Durham, NC) by using a single disposable AST-GP80 card per isolate. The AST-GP80 card was automatically inoculated with a bacterial suspension prepared in 0.45% saline at a spectrophotometric turbidity of 0.5, sealed, and loaded into the Vitek-2 instrument for incubation and reading. The AST-GP80 card contained OXA, AMC, CEF, GEN, ENR, CLI, DOX, and SXT at the same concentration ranges tested by BMD and described above. Categorical interpretations of AST results from Vitek-2 were based on the AES.

Inducible clindamycin resistance.

Inducible clindamycin resistance (ICR) was assayed with three methodologies. (i) BMD was carried out in single wells containing a combination of CLI and erythromycin at final concentrations of 0.5 μg/ml and 4 μg/ml, respectively. The inoculum consisted of 100 μl of bacterium suspension at 5 × 10⁵ CFU per ml in CAMHB. Microplates were then incubated at 35°C in ambient air for 18 to 24 h and examined for any growth (5, 14). (ii) A 0.5 McFarland standard suspension of each S. pseudintermedius isolate was inoculated onto the surface of a CAMHA plate. A CLI disk (2 μg) and an erythromycin disk (15 μg) were placed at least 15 mm apart measured edge to edge. Plates were incubated at 35°C in ambient air for 18 h and examined for a flattening of the zone of inhibition around the clindamycin disk (D-test) or hazy growth within the zone of inhibition (5, 14). (iii) The Vitek-2 AST-GP80 cart contains an ICR test which compares growth in a 0.5-μg/ml CLI well with growth in a well containing 0.25 μg/ml CLI plus 0.5 μg/ml erythromycin.

Interpretative criteria.

Zone diameters and MICs were interpreted using the breakpoints listed in Table 1. To parallelly interpret the results from DD, BMD, and Vitek-2, the selection of the susceptibility cutoff for each antibiotic reflected the fixed concentration range included in the veterinary Vitek-2 card AST-GP80. Breakpoints used for zone diameters and MICs of OXA, GEN, CLI, and SXT were recommended by both of the most current CLSI documents, M100 and VET08, at the time of the study (5, 14), which have been dedicated to drug selection for human and veterinary treatment, respectively. Breakpoints for ENR were extrapolated from the VET08 (14), while those for DOX zone diameter and MIC and CEF MIC were based on M100 criteria (5). Cutoffs for AMC and CEF DD matching AST-GP80 were derived from the archived CLSI document M100-S22 (15). MUP MICs were interpreted according to EUCAST clinical breakpoints for antimicrobial agents that are used topically (6).

TABLE 1.

MIC and zone diameter breakpoints for Staphylococcus pseudintermedius susceptibility categorization^a

Drug	MIC breakpoints (μg/ml)			Zone Diam breakpoints (mm)			Reference(s)
	S	I	R	S	I	R
OXA	≤0.25		≥0.5	≥18		≤17	5, 14
AMC	≤4/2		≥8/4	≥20		≤19	15
CEF	≤2	4	≥8	≥18	17–15	≤14	14, 15
GEN	≤4	8	≥16	≥15	14–13	≤12	5, 14
ENR	≤0.5	1–2	≥4	≥23	22–17	≤16	14
DOX	≤4	8	≥16	≥16	15–13	≤12	5
CLI	≤0.5	1–2	≥4	≥21	20–15	≤14	5, 14
SXT	≤2/38		≥4/76	≥16	15–11	≤10	5, 14
MUP	≤1	2–128	>256	≥30	29–18	<18	6

^aAbbreviations: OXA, oxacillin; AMC, amoxicillin/clavulanate; CEF, cephalothin; GEN, gentamicin; ENR, enrofloxacin; DOX, doxycycline; CLI, clindamycin; MUP, mupirocin; SXT, trimethoprim-sulfamethoxazole; S, sensitive; I, intermediate; R, resistant.

Data analysis.

Analysis of the results was based on the current consensus standards for conducting a performance evaluation of AST methods (16). For all antibiotics, except for MUP, categorical agreement (CA) was estimated between BMD (reference) and DD, BMD (reference) and Vitek-2, and DD (reference) and Vitek-2. CA was calculated as the percentage of isolates producing the same category result (sensitive, intermediate, or resistant) compared to the reference method. Minor error (MiE) describes the case when the reference test results are intermediate while the results of the other method under evaluation are sensitive or resistant, or vice versa. Major error (ME) refers to the eventuality that reference test results are sensitive while the method under analysis results are resistant. Conversely, a very major error (VME) indicates that reference test results are resistant but the results of the method under evaluation are sensitive. Essential agreement (EA) was evaluated between BMD (reference) and Vitek-2 for all antibiotics. Since the Vitek-2 card AST-GP80 does not include MUP, comparative diagnostic accuracy for this antibiotic was evaluated exclusively between DD and BMC (reference). For Vitek-2, EA was reported when the obtained MIC is within ± 1 doubling dilution (1 log₂) of that obtained by BMD. The acceptable rate for EA and CA was established at ≥90%.

Data normality for statistical analysis was evaluated with Shapiro-Wilk and Kolmogorov-Smirnova tests using SPSS software. The Kappa coefficient (κ) was determined for categorical comparisons and interpreted according to McHugh (17). Spearman’s correlation coefficient (r_s) was used to describe MIC agreement (18). A P value of ≤0.05 was considered statistically significant.

Investigation of discrepancies.

As recommended (16), isolates displaying MiE, ME, or VME were retested. A new culture was subcultured from the frozen stock and tested using both of the methods under evaluation and the reference test. EA and CA were calculated following resolution of discrepant results after twice repeating the assay. Data obtained from the investigation of the errors were used to calculate the reproducibility of DD, BMD, and Vitek-2 for S. pseudintermedius. Two measures of reproducibility were applied: (i) MIC agreement as a percentage of all antimicrobial-organism combinations that were within ± 1, 2, or 3 log₂ dilutions and (ii) average standard deviation of the zone diameter for each antibiotic.

RESULTS

Susceptibility test results.

Based on the results from BMD (Table 2), 30 out of 80 (37.5%) S. pseudintermedius isolates were resistant to OXA; thus, they were phenotypically classified as methicillin-resistant. As a rule, all 30 OXA-resistant S. pseudintermedius isolates were considered resistant to other β-lactam agents from the diagnostic angle. However, by using the specific breakpoints for AMC and CEF listed in Table 1, the number of bacteria resistant to AMC and CEF appeared lower than that resistant to OXA. Based on such results, shown in Table 2, we built intermethod comparisons. All S. pseudintermedius isolates were susceptible to MUP and susceptible or intermediately susceptible to DOX. There was a variable resistance to GEN, ENR, SXT, and CLI, as shown in Table 2.

TABLE 2.

Susceptibility profiles of 80 S. pseudintermedius isolates to the tested antibiotics determined by broth microdilution and interpreted according to Table 1^a

Drug	Susceptibility categorization (no. [%])
	S	I	R
OXA	50 (62.50)	0 (0.00)	30 (37.50)
AMC	60 (75.00)	0 (0.00)	20 (25.00)
CEF	60 (75.00)	2 (2.50)	18 (22.50)
GEN	53 (66.25)	4 (5.00)	23 (28.75)
ENR	50 (62.50)	6 (7.50)	24 (30.00)
DOX	52 (65.00)	28 (35.00)	0 (0.00)
CLI	46 (57.50)	2 (2.50)	32 (40.00)
ICR	67 (83.75)	0 (0.00)	13 (16.25)
SXT	43 (53.75)	0 (0.00)	37 (46.25)
MUP	80 (100.00)	0 (0.00)	0 (0.00)

^aAbbreviations: OXA, oxacillin; AMC, amoxicillin/clavulanate; CEF, cephalothin; GEN, gentamicin; ENR, enrofloxacin; DOX, doxycycline; CLI, clindamycin; ICR, inducible clindamycin resistance; MUP, mupirocin; SXT, trimethoprim-sulfamethoxazole; S, sensitive; I, intermediate; R, resistant.

Comparison of broth microdilution and disk diffusion.

Both OXA DD and BMD are recommended by CLSI as methods to phenotypically detect mecA-mediated β-lactam resistance in S. pseudintermedius. In this regard, CA between the two methods was achieved for 79/80 (98.8%) isolates (κ = 0.97). Only one ME was detected, which was the result of a measured 17-mm OXA zone, while the MIC was ≤0.25 μg/ml (Fig. 1A). The majority (23/30, 76.7%) of OXA-resistant isolates grew at the edge of the oxacillin disk (zone diameter of 6 mm). The growth of the same bacteria was inhibited by an OXA MIC of ≥4 μg/ml. Conversely, the majority of OXA-susceptible isolates showed an OXA MIC of ≤0.25 μg/ml and an inhibition diameter of between 20 and 25 mm. BMD and DD showed a 100% agreement on AMC, SXT, and MUP (κ = 1.00). Except for OXA, the comparison between BMD and DD provided only MiEs for CEF (κ = 0.93), GEN (κ = 0.90), ENR (κ = 0.84), CLI (κ = 0.89), and DOX (κ = 0.92) (Fig. 1C to G). Overall, CA between DD and BMD was >90%. Agreement on ICR between DD and BMD was 100%. The kappa coefficient of between 0.90 and 0.84 indicated strong agreement for GEN, ENR, and CLI (P ≤ 0.0001), while κ between 1.00 and 0.95 suggested an almost perfect agreement for the rest of the antibiotics and ICR (P ≤ 0.0001).

FIG 1.

Scatterplots comparing the results of MIC (μg/ml) obtained by broth microdilution with the zone diameter derived from disk diffusion (mm) for (A) oxacillin, (B) amoxicillin/clavulanate, (C) cephalothin, (D) gentamicin, (E) enrofloxacin, (F) doxycycline, (G) clindamycin, (H) trimethoprim-sulfamethoxazole, and (I) mupirocin. Red lines represent the applied resistance breakpoints. *, minor error; a, major error.

Comparison of broth microdilution and Vitek-2.

The agreement between AST performances of Vitek-2 and BMD for S. pseudintermedius is described in Table 3. All Vitek-2 and BMD analysis terminated successfully due to the acceptable growth in the control wells. CAs and EAs between BMD and Vitek-2 were ≥90% for all antibiotics considered. The majority of the errors were MiEs, but seven VMEs were found, one for GEN, five for CLI, and one for ICR. Vitek-2 reported a MIC of ≤0.5 μg/ml against an S. pseudintermedius isolate that resulted as GEN-resistant by BMD and a CLI MIC between 0.25 and 0.5 μg/ml against five isolates that showed hazy growth, a constitutive resistant phenotype, by manual methods. Vitek-2 diagnosed this phenotype as positive ICR, resulting in five ICR MEs. Method comparison detected two MEs for OXA and one for CLI. The two MEs for OXA were found when testing two isolates that were sensitive to ≤0.25 μg/ml by BMD, while Vitek-2 detected a MIC of 1 μg/ml. Regarding ME for CLI, one isolate had a MIC of ≤0.125 μg/ml by BMD, but it was diagnosed as CLI resistant by the Vitek-2 system. The kappa coefficient of between 1.00 and 0.95 suggested an almost perfect agreement for OXA, AMC, CEF, GEN, DOX, and SXT (P ≤ 0.0001). Strong agreement (κ = 0.88 to 0.80) was found for ENR and CLI, while the intermethod reliability for ICR (κ = 0.76) was substantial (P ≤ 0.0001). A very strong correlation was found between MIC values provided by BMD and Vitek-2 for all antibiotics tested (r_s ≥ 0.88, P ≤ 0.01).

TABLE 3.

Performance of Vitek-2 AST-GP80 card for Staphylococcus pseudintermedius in comparison to broth microdilution method^a

Drug	EA (no. [%])	rs	CA (no. [%])	κ	VME (no. [%])	ME (no. [%])	MiE (no. [%])
OXA	76/80 (95.00)	0.97	78/80 (97.50)	0.95	0/30 (0.00)	2/50 (4.00)	0/80 (0.00)
AMC	80/80 (100.00)	0.98	80/80 (100.00)	1.00	0/20 (0.00)	0/60 (0.00)	0/80 (0.00)
CEF	79/80 (98.75)	0.98	79/80 (98.75)	0.97	0/18 (0.00)	0/60 (0.00)	1/80 (1.25)
GEN	79/80 (98.75)	0.94	78/80 (97.50)	0.95	1/23 (4.35)	0/53 (0.00)	1/80 (1.25)
ENR	79/80 (98.75)	0.95	76/80 (95.00)	0.88	0/24 (0.00)	0/50 (0.00)	4/80 (5.00)
DOX	79/80 (98.75)	0.97	78/80 (97.50)	0.95	0/0 (NA)	0/52 (0.00)	2/80 (2.50)
CLI	72/80 (90.00)	0.88	72/80 (90.00)	0.80	5/32 (15.63)	1/46 (2.17)	2/80 (2.50)
ICR	NA	NA	74/80 (92.50)	0.76	1/13 (7.69)	5/67 (7.46)	0/80 (0.00)
SXT	79/80 (98.75)	0.93	80/80 (100.00)	1.00	0/37 (0.00)	0/43 (0.00)	0/80 (0.00)

^aAbbreviations: EA, essential agreement; r_s, Spearman’s correlation coefficient; CA, categorical agreement; κ, kappa coefficient; VME, very major error; ME, major error; MiE, minor error; OXA, oxacillin; AMC, amoxicillin/clavulanate); CEF, cephalothin; GEN, gentamicin; ENR, enrofloxacin; DOX, doxycycline; CLI, clindamycin; ICR, inducible clindamycin resistance; SXT, trimethoprim-sulfamethoxazole; NA, not applicable.

Comparison of disk diffusion to Vitek-2.

The agreement between AST performances of Vitek-2 and DD for S. pseudintermedius is described in Table 4. As observed comparing Vitek-2 and BMD, the comparison between Vitek-2 and DD generated one VME for ICR and five for CLI alongside one ME for CLI and five for ICR. Additionally, Vitek-2 recorded two isolates as susceptible to GEN, while inhibition zones were 12 mm. Overall, the agreement of Vitek-2 with DD was comparable to that with BMD; of the eight antibiotics tested, seven showed CA of ≥90%. CA for CLI was lower (85%), but half of the discrepancies were minor. Accordingly, κ equal to 0.70 for CLI and 0.76 for ICR indicated substantial intermethod agreement (P ≤ 0.0001). Strong agreement was found for CEF, GEN, ENR, and DOX (κ = 0.90 to 0.80, P ≤ 0.0001). An almost perfect agreement was suggested by a κ of ≥0.97 determined for OXA, AMC, and SXT (P ≤ 0.0001).

TABLE 4.

Performance of Vitek-2 AST-GP80 card for Staphylococcus pseudintermedius in comparison to the disk diffusion method^a,b

Drug	CA (no. [%])	κ	VME (no. [%])	ME (no. [%])	MiE (no. [%])
OXA	79/80 (98.75)	0.97	0/31 (0.00)	1/49 (2.04)	0/80 (0.00)
AMC	80/80 (100.00)	1.00	0/20 (0.00)	0/60 (0.00)	0/80 (0.00)
CEF	77/80 (96.25)	0.90	0/17 (0.00)	0/61 (0.00)	3/80 (3.75)
GEN	76/80 (95.00)	0.89	2/21 (9.52)	0/53 (0.00)	2/80 (2.50)
ENR	72/80 (90.00)	0.80	0/23 (0.00)	0/55 (0.00)	8/80 (10.00)
DOX	75/80 (93.75)	0.87	0/0 (NA)	0/53 (0.00)	5/80 (6.25)
CLI	68/80 (85.00)	0.70	5/32 (15.63)	1/40 (2.50)	7/80 (8.75)
ICR	74/80 (92.50)	0.76	1/13 (7.69)	5/67 (7.46)	0/80 (0.00)
SXT	80/80 (100.00)	1.00	0/37 (0.00)	0/43 (0.00)	0/80 (0.00)

^aAgreements of below 90% are shown in boldface type.

^bAbbreviations: CA, categorical agreement; κ, kappa coefficient; VME, very major error; ME, major error; MiE, minor error; OXA, oxacillin; AMC, amoxicillin/clavulanate; CEF, cephalothin; GEN, gentamicin; ENR, enrofloxacin; DOX, doxycycline; CLI, clindamycin; ICR, inducible clindamycin resistance; SXT, trimethoprim-sulfamethoxazole; MUP, mupirocin; NA, not applicable.

Method reproducibility.

A total of 42 isolates displayed discordant intermethod results for one or more antibiotics. This subgroup was used to assess the intralaboratory reproducibility of DD, BMD, and Vitek-2 for S. pseudintermedius. As detailed in Table 5, BMD had minimal variation after test repeats, with no drug showing more variation than another. Vitek-2 provided a higher percentage of readings outside the mode ± 1 dilution than BMD, particularly when analyzing OXA, ENR, CLI, and SXT. Standard deviations of DD zone diameters ranged from ±1.15 (OXA) to ±4.05 mm (SXT).

TABLE 5.

Intralaboratory reproducibility of broth microdilution (BMD), Vitek-2, and disk diffusion (DD) for S. pseudintermedius^a

Drug	Percentage MIC agreement (%) for:						Avg SD (mm) for DD
	BMD (log2 dilutions)			Vitek-2 (log2 dilutions)
	± 1	± 2	± 3	± 1	± 2	± 3
OXA	98.42	0.79	0.79	84.92	3.97	11.11	1.15
AMC	98.41	0	1.59	93.65	3.97	2.38	2.47
CEF	96.83	0.79	2.38	92.86	1.59	5.55	2.83
GEN	98.41	0	1.59	92.86	2.38	4.76	2.00
ENR	97.62	0.79	1.59	85.71	10.32	3.97	1.37
DOX	98.41	0	1.59	92.07	4.76	3.17	2.73
CLI	98.41	0	1.59	88.89	0	11.11	2.21
MUP	99.21	0	0.79				2.59
SXT	96.83	0	3.17	84.12	1.59	14.29	4.05

^aAbbreviations: Avg SD, average standard deviation; OXA, oxacillin; AMC, amoxicillin/clavulanate; CEF, cephalothin; GEN, gentamicin; ENR, enrofloxacin; DOX, doxycycline; CLI, clindamycin; SXT, trimethoprim-sulfamethoxazole; MUP, mupirocin.

DISCUSSION

Previous studies described the performances of Vitek-2 with Staphylococcus spp. targeting single antibiotics (19–24). Moreover, automated AST methods were comprehensively assessed on staphylococcal species other than S. pseudintermedius (4, 25). To the best of our knowledge, we compared for the first time AST results generated by BMD, DD, and automated Vitek-2 specifically for S. pseudintermedius, focusing on antibiotic classes used to treat skin infections (26).

To parallelly interpret the results provided by the different AST methods, we selected the breakpoints listed in Table 1. The concentration range tested by the Vitek-2 AST-GP80 card is concordant with both human and veterinary breakpoints recommended by the most recent CLSI guidelines for the majority of the molecules used in this study. However, the CLSI document VET08 (14) recommends breakpoints for certain antibiotics against canine S. pseudintermedius that do not currently fit with the concentration range included in the AST-GP80 card. The cutoff selected for AMC was derived from the archived CLSI document M100-S22 as well as the DD breakpoint for CEF (15). These references correlate with the concentration range proposed in the veterinary Vitek-2 AST-GP80 card, but they are intended for human treatment. Conversely, the CLSI standard VET08 recommends reporting the sensitivity of canine Staphylococcus spp. to AMC if the MIC is ≤0.25/0.12 μg/ml, while the minimum AMC concentration tested through AST-GP80 is 2/1 μg/ml. We encountered a comparable drawback in interpreting the response of S. pseudintermedius to DOX, for which the VET08 document suggests susceptibility if the MIC is ≤0.125 μg/ml and resistance if the MIC is ≥0.5 μg/ml. Incompatibly, the minimum DOX concentration tested by the Vitek-2 AST-GP80 is 0.5 μg/ml. Therefore, interpretive criteria for the DOX MIC and zone diameter were necessarily selected from CLSI standard M100 (5), dedicated to guidelines for the treatment of human staphylococci. Using M100 breakpoints for DOX, all S. pseudintermedius were susceptible or intermediately susceptible to DOX, but this could be an interpretive bias potentially affecting the decision making of antimicrobial prescribing for companion animals.

Once the analysis breakpoints were established, the core of the study was the evaluation of intermethod agreement among widely used AST techniques for S. pseudintermedius. This information is crucial to ensure that a correct methodology is applied. Despite this interest, previous studies have been limited to assay the concordance of cefoxitin/OXA disk zones and MIC with mecA/mecC PCR and to evaluate the diagnostic test accuracy of DD relative to BMD for canine clinical S. pseudintermedius (19, 27). As put forward, the evidence we found points to DD performing comparably to BMD for all the antibiotic classes tested. Both DD and Vitek-2 showed a satisfactory agreement of >90% with BMD in identifying both susceptible and resistance phenotypes. However, there were a few exceptions. As observed earlier (4), Vitek-2 generated one VME for GEN, and it failed in identifying 1/13 ICR-positive isolates. Therefore, a Vitek-2 negative ICR should be confirmed by CLSI D-test or BMD before reporting CLI as susceptible, as previously suggested (28). Unexpectedly, Vitek-2 improperly diagnosed five bacteria as CLI-susceptible/ICR-positive, but they had a noninduction resistance phenotype with an inner ring of reduced growth up to the edge of the CLI disk (29). In contrast, Jorgensen et al. noted that staphylococci showing such a phenotype were reported as CLI resistant by Vitek-2 (30).

We included MUP in the testing panel because it is a last-resort molecule used topically to decolonize skin and soft tissue of both methicillin-susceptible and methicillin-resistant staphylococci (31). However, as far as we know, MUP is available as a veterinary product only in the United States for topical treatment of canine pyoderma (32). This may be the reason the veterinary card AST-GP80 does not include MUP. Due to the lack of this antibiotic in the AST-GP80 card panel, the intermethod agreement for MUP, which was 100%, was evaluated solely between DD and BMD. By using the AST-P549 card, Malaviolle et al. (33) detected MUP-resistant strains by Vitek-2. However, the development of a card that can distinguish between low-level and high-level MUP resistance would be even more helpful, as the response of low-level MUP-resistant staphylococci to the highly concentrated ointment is dubious.

Although the inclusion of bacteria showing different responses to the majority of antibiotics allowed us to analyze the performance of AST methods in different susceptibility contexts, the skewed distribution of S. pseudintermedius susceptibility to MUP and DOX and the use of geographically similar S. pseudintermedius isolates may limit the generalizability of our results. Future intermethod comparisons should concentrate on S. pseudintermedius collected from different hosts and geographical areas and showing a wider range of antibiotic susceptibility patterns. This is important because the mechanisms underlying antimicrobial resistance in the genus Staphylococcus are complex, and these could generate phenotypic responses detected differently by manual AST methods compared to automated ones.

Overall, the evidence from this study suggests reliable performances of DD and Vitek-2 with the veterinary AST-GP80 card in comparison to the reference BMD for S. pseudintermedius of animal origin. However, while DD generated minor errors, Vitek-2 could not reliably detect constitutive and inducible CLI resistance. Interpretive bias could arise from the use of Vitek-2 AST-GP80, including concentration ranges for DOX and AMC that do not match CLSI clinical breakpoints for canine S. pseudintermedius.

Considering the clinical and epidemiological importance of S. pseudintermedius and breakpoints recommended to interpret its antibiotic susceptibility, we encourage updating action by the system manufacturer to address AST for this bacterium and improve antibiotic prescribing in companion animals.