Thirty-four clinical MRSA isolates were randomly selected from the TIST study that collected clinical isolates from 22 hospitals between 2006 and 2010 (Hsueh, 2008). Staphylococci were identified by the colonial morphology, Gram stain, and coagulase test. MRSA isolates were further confirmed by the tube coagulase test and their growth on 6 μg/mL oxacillin salt agar screen plates. Isolates were stored at −70°C in Protect Bacterial Preservers (Technical Service Consultants Limited, Heywood, UK) until use. Their genetic relatedness was examined by pulse-field gel electrophoresis (PFGE) as previously described (Tenover et al., 1995; Lai et al., 2017).

The tested antibiotics included oxacillin, erythromycin, gentamicin, clindamycin, rifampin, minocycline, cefazolin, cefmetazole, cefotaxime, cefepime, vancomycin (Sigma, St Louis, MO), fosfomycin (Ercros, Barcelona, Spain), linezolid, tigecycline (Pfizer, New York, NY), fusidic acid (Leo Pharma, Ballerup, Denmark), teicoplanin (Sanofi-Aventis, Bridgewater, NJ), ciprofloxacin (Bayer, Leverkusen, Germany), and daptomycin (Cubist Pharmaceuticals, Lexington, MA). The MIC determination by the agar dilution method and interpretation criteria were based on the recommendations of the Clinical and Laboratory Standards Institute (CLSI), Food and Drug Administration (FDA), and British Society for Antimicrobial Chemotherapy (Andrews, 2001; Clinical and Laboratory Standards Institute, 2012). For fosfomycin susceptibility tests, glucose-6-phosphate (25 μg/mL) was added to the agar plates. Daptomycin MIC was studied in Mueller-Hinton broth that was adjusted to 50 μg/mL calcium. S. aureus ATCC 29213 was used as a control strain in each run of MIC measurements. According to cefazolin MIC, MRSA were categorized into two groups, the high-cefazolin MIC (HCM, MIC > 128 μg/mL) and low-cefazolin MIC (LCM ≤ 128 μg/mL) groups.

Genomic DNA from MRSA isolates was purified and used as a template for PCR amplification. PCR amplification of PVL genes, i.e., lukF-PV and lukS-PV, was performed as previously described (Montanaro et al., 2016). The primer sequences for the PVL genes were luk-PV-1 and luk-PV-2. SCCmec typing was tested by multiplex PCR (M-PCR) according to a previously published protocol (Zhang et al., 2005). The M-PCR assay used 4 primers for mec (mecI-F, mecI-R, IS1272-F, and mecR1-R) and ccr (ccrAB-α2, ccrAB-α3, ccrAB-α4, and ccrAB-β2) complexes. A single-target PCR was used to detect type 5 ccr by ccrC-F and ccrC-R primers.

To evaluate the in vitro effect of G/C combinations, the microdilution checkerboard method was used to calculate the fractional inhibitory concentrations (FICs), as recommended by the CLSI (Clinical and Laboratory Standards Institute, 2012; Lai et al., 2016). The following formulas were used to calculate the FIC index, which equals the FIC of drug A (MIC of drug A in combination/MIC of drug A alone) + FIC of drug B (MIC of drug B in combination/MIC of drug B alone). Synergism was defined as a FIC index of ≤ 0.5, an indifference FIC index of >0.5 but ≤ 4, and an antagonism FIC index of >4 (Lai et al., 2016).

Two MRSA isolates were randomly selected for another in vitro measurement of the inhibitory effect of combination regimens, as recommended by the CLSI (National Committee for Clinical Laboratory Standards, 1999). In brief, bacterial suspensions were diluted to 5.0 × 10⁵ colony-forming units (CFUs)/mL in fresh Mueller–Hinton broth. Drug concentrations of vancomycin or teicoplanin were adjusted to 1xMIC, 1/2xMIC, and 1/4xMIC. For each of four cephalosporins, the drug concentrations of 1- and 1/2-fold susceptible breakpoints were used for the combination with a glycopeptide. Bacterial counts were measured at 4, 8, and 24 h by enumerating the colonies in 10-fold serially diluted specimens of 100-μL aliquots plated on the nutrient agar (Difco Laboratories, Sparks, MD) at 37°C. All experiments were performed in duplicate.

Synergism was defined as a ≥2 log₁₀ decrease in CFU/mL between the combination regimen and its most active constituent after 24 h, as well as the number of surviving organisms in the combination regimen, which must be ≥2 log₁₀ CFU/mL below the starting inoculum. In addition, at least one of the combination drugs must be present at a concentration that does not affect the growth of the test organism (American Society for Microbiology, 2014). Bacteriostatic and bactericidal activities were defined as < 3 log₁₀ and ≥3 log₁₀ reductions in CFU/mL at 24 h, respectively, relative to the starting inoculum (National Committee for Clinical Laboratory Standards, 1999).

The MICs of vancomycin or teicoplanin alone and combined with 1x or 1/2x susceptible breakpoint concentration (SBC) of a cephalosporin were determined by the agar dilution method. A MIC ratio indicates the fold of the MIC decline of the C-G combination vs. a glycopeptide alone.

Data analyses were performed using SPSS for Windows 17.0 (SPSS Inc., Chicago, Illinois, USA). Because of the small sample sizes and violation of the normal distribution assumption of the OD ratio, the Mann–Whitney U-test was used to compare the differences between groups. The Kruskal–Wallis H and Dunn's tests were applied for multiple comparisons. The statistical significance was set to a P < 0.05.

Of 34 randomly selected MRSA isolates, 21 were LCM isolates, and 13 were HCM isolates. PVL genes were found in 10 isolates, which were LCM isolates. None of the HCM isolates harbored lukF-PV or lukS-PV (Figure 1). SCCmec typings of the LCM isolates included type IV (13 isolates) and V (8 isolates), and all 13 HCM isolates were type III. The pulsotype number of 13 HCM isolates was 6 (pulsotype A-F, one strain was untypeable), which was 10 for 21 LCM isolates (pulsotype G-P; Figure 2).

The MIC₉₀ values of cefazolin, cefmetazole, cefotaxime and cefepime for HCM isolates (512, 512, 128, and 512 μg/mL, respectively) were higher than those of LCM isolates (32, 16, 64, and 64 μg/mL, respectively). A similar trend was also noted for oxacillin, minocycline, fusidic acid, ciprofloxacin, fosfomycin, and daptomycin. Of note, the MIC₉₀s for HCM isolates was lower than for LCM isolates (16 vs. 128 μg/mL) for clindamycin. Additionally, for vancomycin or teicoplanin, the MIC ranges and MIC₉₀ values were similar in the HCM and LCM isolates (Table 1).

With the checkerboard method, the FIC₉₀ of vancomycin-based combinations ranged from 0.53 to 0.75, and teicoplanin-based combinations ranged from 0.51 to 0.62 in the HCM isolates (Table 2). Likewise, in the LCM isolates, the FIC₉₀ of vancomycin-based combinations ranged from 0.51 to 0.56, and teicoplanin-based combinations ranged from 0.51 to 0.75. No antagonism was noted with either vancomycin or teicoplanin in combination.

Among vancomycin-based combinations, synergism was noted in 30.8% of the HCM isolates for cefepime and 69.2% for cefotaxime. In the LCM isolates, synergism was present in 13.6% for cefepime and 66.7% for cefazolin (Table 2). Among the teicoplanin-based combinations, 38.5% of the HCM isolates exhibited synergism that was lower for cefepime and cefotaxime, which was 84.6% for cefazolin. Synergism was not observed in any isolates for the cefepime-teicoplanin combination, but it was 47.6% for cefotaxime-teicoplanin combinations (Table 2).

With the checkerboard method, we found in vitro inhibitory activity of the combinations of 1/2x MIC of vancomycin or teicoplanin and 1/2x MIC of a cephalosporin in all 13 HCM isolates (Table 3). The inhibition effect can be observed in 9 of 13 HCM isolates even if cefazolin is used at a concentration of 1/128x MIC combined with vancomycin at a concentration of 1/2x MIC.

In time-killing studies using 1x or 1/2x SBCs of a cephalosporin combined with 1/2x or 1/4x MIC of vancomycin or teicoplanin, bacterial loads are shown in Table 4. For a randomly selected HCM isolate, TIST-5, synergism was noted in any of four cephalosporins at a concentration of 1x SBC in combination with vancomycin at a concentration of 1/2x MIC, and there was a decline of 2.5–3.0 log₁₀CFU/mL. At the concentration of 1/2x SBC, all cephalosporins, except cefotaxime, combined with vancomycin at the concentration of 1/2 MIC, leading to a bacterial load reduction of 2.8–3.6 log₁₀ CFU/mL. The synergism was observed at combinations of 1/2x SBC cefmetazole and 1/4x MIC vancomycin with a colony count reduction of 3.12 log₁₀ CFU/mL. For teicoplanin-based combinations, antibacterial activity was similarly potent, and the combination of 1/2x SBC cefmetazole and 1/4x MIC teicoplanin can cause a bacterial load reduction of 2.8 log₁₀ CFU/mL.

For another randomly selected LCM isolate, TIST-10, synergism was present in any of four cephalosporins at a concentration of 1x SBC combined with 1/2x MIC vancomycin or teicoplanin, which can cause bacterial load reductions of 2.4–2.7 and 2.1–2.5 log₁₀ CFU/mL, respectively (Table 4). Moreover, a lower cephalosporin concentration, 1/2x SBC, in combination with 1/2x MIC vancomycin or teicoplanin can cause a bacterial load reduction of 2.3–2.44 or 2.2–2.5 log₁₀ CFU/mL, respectively. When combined with 1/4x MIC vancomycin, 1x or 1/2x SBC cefmetazole can exhibit synergistic activity, resulting in a bacterial load reduction of 3.3 and 2.7 log₁₀ CFU/mL, respectively, at 24 h of incubation. For teicoplanin-based regimens, 1/4x MIC teicoplanin can show synergism when combined with 1x or 1/2x SBC cefazolin or cefmetazole or with 1x SBC cefotaxime.

The last way to examine the in vitro effect of C-G combinations is to evaluate the glycopeptide MIC change in the presence and absence of a cephalosporin. The MICs of vancomycin or teicoplanin were lower if there was any cephalosporin at the concentration of 1/2 SBC in culture media, which was independent of the cefazolin MIC. The MIC ratio of glycopeptides (glycopeptide MIC in the presence of a cephalosporin and glycopeptide) ranged from 1/6 to 3/4 (Table 5). If the cephalosporin concentration was fixed at 8 μg/mL, such changes in the glycopeptide MIC were most obvious with cefmetazole (Figure 3).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.