Bactericidal action of nafcillin, vancomycin, and three cephalosporins against nafcillin-susceptible and nafcillin-resistant coagulase-negative

Coagulase-negative staphylococci (S. epidermidis, 43 strains; S. warneri, 16 strains; S. haemolyticus, five strains; and others, four strains) were tested by the agar dilution method for nafcillin susceptibility: 53 were susceptible with a minimal inhibitory concentration (MIC) of less than or equal to 2 micrograms/ml; four were of indeterminate susceptibility, MIC = 4-16 micrograms/ml; and 11 were resistant, MIC greater than or equal to 32 micrograms/ml. The bactericidal activities from 0 to 24 hr for nafcillin, vancomycin, cephalothin, cefazolin, and cefamandole, each at 16 micrograms/ml in broth, were determined for all the isolates. The data indicate that a nafcillin agar dilution susceptibility test result of resistance does not consistently predict lack of killing activity by the cephalosporins. It is likely that each cephalosporin would have to be tested against individual coagulase-negative staphylococci in order to determine a suitable therapeutic or prophylactic cephalosporin, if a cephalosporin were to be used. Vancomycin was bactericidal for all the nafcillin-resistant coagulase-negative organisms tested.


INTRODUCTION
This study was undertaken to compare the differential bactericidal action of nafcillin, vancomycin, cephalothin, cefazolin, and cefamandole against nafcillin-susceptible and nafcillin-resistant coagulase-negative staphylococci. The methods of comparative analyses used in the study quantified static [minimal inhibitory concentration (MIC) and extent of kill) as well as kinetic (rate of kill) indices for all five drugs to study possible advantages for cephalosporins compared with the other drugs.

Bacteria
Sixty-eight coagulase-negative staphylococci were obtained from patients' specimens submitted for culture. The 68 isolates included 43 strains of S. epidermidis, 16 strains of S. warneri, five strains of S. haemolyticus, and one strain each of S. capitis, S. hyicus, S. sciuri, and S. simulans. (The one strain of S. hyicus was coagulase-negative). These isolates were recovered from blood, cerebrospinal fluid, orthopedic wounds, pleural fluid, peritoneal fluid, trachea, or intravenous catheter tip. Each organism was isolated in pure culture or was the predominant organism in the culture. The organisms were considered to be potentially meaningful in terms of clinical infection; however, patient histories were not reviewed to determine if the isolates were truly causes of infection.

Antimicrobial Agents
Nafcillin, vancomycin, cephalothin, cefazolin, and cefamandole standard powders of known potency were obtained from the manufacturers.

Minimal Inhibitory Concentrations
The MICs for the five drugs were determined by agar dilution using methods of the National Committee on Clinical Laboratory Strandards (1983), except that MIC endpoints were read after 24 and 48 hr incubation at 35°C. The 48 hr incubation at 35°C was used to detect heteroresistance (Thornsberry et al., 1973). The 48-hr endpoints were used for the final data tabulations. Streptococcus faecalis ATCC 29212 was used as the control strain.

Bactericidal Assays
The bactericidal action of each antibiotic against the 68 organisms was measured in brain heart infusion broth with drug concentrations of 16 p,g/ml. The concentration was chosen because it represented clinically achievable serum concentrations for each antibiotic. A drug-free growth control was included for each organism. Inocula, which had final concentrations of I × 10 B CFU/ml, were made from overnight growth of the organisms on blood agar plates--an acceptable procedure for staphylococci (Baker et al., 1983). Test cultures were incubated at 37°C, and 0.1 ml samples were taken, diluted, and plated on Mueller-Hinton agar at 0, 2, 4, 6, 8, and 24 hr. Colonies formed by the surviving bacteria were counted after 48 hr incubation at 37°C. Data from a total of 408 bactericidal and control assays are reported in this study.
The rate of killing for each antibiotic was assessed by linear regression analysis on the lOglo CFU/ml versus time data for the 2, 4, and 6 hr time points. These time points were chosen because the slopes of the time-kill data lines from 2 to 6 hr represented the initial rate of killing for each antibiotic.
The extent of bactericidal action was assessed by comparing the number of viable organisms remaining after 24 hr incubation. A 99.9% reduction (3 loglo decrease} at 24 hr was interpreted as bactericidal; a <99.9% reduction in CFU/ml at 24 hr was considered bacteriostatic; an increase in CFU/ml at 24 hr compared with zero time was considered to be noninhibition (e.g., growth).
Data analysis was performed using the National Institutes of Health-sponsored PROPHET computer network (Castleman et al., 1975). For small sample sizes (n ~< 50), the Wilk-Shapiro test for normality was used to test the hypothesis that a sample came from a population that was normally distributed. For larger sample sizes (n > 50), the D'Agostino test for normality was used (NIH Publication No. 80-2169. The Kruskal-Wallis test was used to compare data that were not normally distributed. Multiple pairwise comparisons were performed whenever a significant difference was detected with the Kruskal-Wallis test, using a procedure that paralleled the methodology of the Neuman-Kuels test, except that sample rank sums were used instead of means (NIH Publication No. 80-2169 A p value of <0.05 was used for overall significance, and a p value of <0.01 was used for detecting differences in pairwise comparisons.

AntibioUc Inhibitory Concentrations
The agar dilution susceptibility tests revealed 53 nafcillin-susceptible organisms, four of indeterminant susceptibility, and 11 resistant organisms ( Figure 1). The resistant organisms were predominantly S. epidermidis, but included S. warneri and S. haemolyticus. There was no significant differential susceptibility pattern between species or for isolates from different anatomic sites. Organisms that were susceptible   to nafcillin were susceptible to the other antibiotics; some nafcillin-resistant organisms, however, were susceptible to some of the cephalosporins. All of the isolates were susceptible to vancomycin.

Bactericidal Assay Data
The plots of log10 CFU/ml versus time were approximately linear from 2 to 6 hr. An equation of the form y =mx + b was used to describe the data as follows: logao (CFU/ml), = loglo (CFU/ml)to -K (t ° -t), where the subscripts t and t ° refer to the time of the CFU/ml determinations, K (the rate constant of antibiotic killing) is the slope of the loglo CFU/ml versus time plot, and (t ° -t) is the elapsed time in hours between the CFU/ml determinations. Taking the antilog of both sides, the equation reduces to (CFU/ml)t = (CFU/ml)to e -z.3oa r Ct o -O, where 2.303 is the factor needed for converting the log~o equation to natural logarithms. The rate constants for bactericidal action (K) for nafcillin-susceptible, -indeterminant, and -resistant organisms were not normally distributed; therefore, median values are shown in Table 1. The Kruskal-Wallis test was used to compare the rate of antibiotic killing between treatment groups within the staphylococcal nafcillinsusceptibility subclasses (columns of Table 1) and to compare the rate of antibiotic killing between nafcillin-susceptibility subclasses within each treatment trial (rows of Table 1). The rate constants for antibiotic killing of nafcillin-susceptible strains were comparable for all the antibiotics, with only the controls being significantly different. Both nafcillin and cefazolin had low rate constants for nafcillin-resistant strains, whereas the rate constants for vancomycin, cephalothin, and cefamandole changed very little; only the controls were significantly different. When comparisons were made of the rates of antibiotic killing within treatment groups, only nafcillin Nafcillin-susceptible isolates and cefazolin showed significant decreases in killing rates between the nafcillinsusceptible and nafcillin-resistant strains. Figure 2 shows the extent of bactericidal action at 24 hr. A one-way analysis of variance was used to compare the number of viable bacteria remaining at 24 hr in each nafcillin-susceptibility subclass. For nafcillin-susceptible and -indeterminate organisms, the decrease in colony counts at 24 hr was similar for all five drugs and significantly different than the controls. For nafcillin-resistant organisms, vancomycin showed significantly greater bactericidal activity when compared with the other four drugs; furthermore, all the drugs except nafcillin were significantly different from the controls. On an individual basis, the 24 hr bactericidal actions of the drugs were vancomycin > cephalosporins > nafcillin.
An agar dilution susceptibility test result of nafcillin-susceptible predicted that nafcillin and the three cephalosporins were bactericidal for the majority of strains and bacteriostatic for the remaining strains. The results for the 15 strains of indeterminate or resistant susceptibility to nafcillin were less clear cut, with the four drugs being bactericidal, bacteriostatic, or noninhibitory ( Table 2).
The MICs indicated all 68 strains were susceptible to vancomycin. Vancomycin was bactericidal for 45 of the 53 strains that were susceptible to nafcillin and bacteriostatic for the remaining eight strains--a result similar to that of the other four aDetermined by 48-hr agar dilution minimal inhibitory concentration (MIC) tests. bThe nafcillin agar dilution MIC for this strain was 32 p.g/ml; the broth dilution MIC was 2 p.g/ml. The strain also exhibited heteroresistance.
drugs. Vancomycin was bactericidal for 14 of the remaining 15 strains and was bacteriostatic against one strain of indeterminate nafcillin susceptibility (Table 2).

DISCUSSION
In vitro susceptibility testing of Staphylococcus epidermidis-group organisms and Staphylococcus aureus, for the semisynthetic penicillinase-resistant penicillins, is subject to major difficulties in interpretation of results. Much more data are available for S. aureus than for the other species of staphylococci (Barry and Badal, 1977; McDougal and Thornsberry, 1984;Sabath, 1977;Thornsberry et al., 1973;Thornsberry and McDougal, 1983). Staphylococcus aureus, which are clearly resistant to nafcillin (or oxacillin or methicillin), also show resistance to cephalosporins by in vitro bactericidal assays and in the clinical setting; it is, however, not clear from published data that the same statements can be made for the coagulase-negative staphylococci (Ein et al., 1979;Frongillo et al., 1984;Hansen, 1983;John and McNeill, 1980;Laverdiere et al., 1978).
The isolates used in this study were not preselected for known nafcillin resistance and were a representative sample of typical patient isolates. The anatomic sites and species distribution of the isolates were representative of coagulase-negative organisms in general (Gill, 1983;Marsik and Brake, 1982;Price and Flournoy, 1982;Sewell et al., 1982). S. epidermidis and S. haemolyticus have been reported to show a wider spectrum of antibiotic resistance than other species in the group (Gill, 1983;Marsik and Brake, 1982;Price and Flournoy, 1982). We found no obvious differences in nafcillin susceptibility by species, but tested relatively few strains of species other than S. epidermidis and S. warneri.
The addition of 2%-5% NaC1 to the agar or broth has been advocated for susceptibilty testing of S. aureus. The utility of added NaC1 in susceptibility testing of coagulase-negative staphylococci is less clear and has not been standardized for bactericidal assays of these bacteria. Therefore, we chose not to add NaCI to our assay system. Cephalothin and cefamandole killed nafcillin-resistant organisms as swiftly as they killed nafcillin-susceptible strains. A likely explanation for this observation is that several of the nafcillin-resistant strains were susceptible to cephalothin and cefamandole. Comparison of the agar dilution results indicated this to be true. Similarly, comparison of the 24-hr bactericidal assay results indicated that all three cephalosporins were bactericidal for some of the nafcillin-resistant strains. This result could not be predicted by the nafcillin susceptibility test result, but did correlate with the specific cephalosporin susceptibility test result.
Cephalothin cumulative MICs were slightly lower than cefazolin and cefamandole cumulative MICs. Siebert et al. (1979) noted differences in cumulative MICs and minimal bactericidal concentrations against methicillin-resistant S. epidermidis, with cefazolin less active than cephalothin. We found that cefazolin killed the nafcillinresistant strains more slowly than cephalothin or cefamandole when the drug concentrations were 16 p,g/ml. The results of the 24-hr bactericidal assays, however, were not significantly different for cefazolin compared with cephalothin or cefamandole. We conclude that the bactericidal activities of the three cephalosporins were comparable even though the rates of killing were different. To choose one of the three cephalosporins over the others based on putative differences in potential bactericidal activity seems unwarranted.
The primary objective of this study was to determine whether or not in vitro susceptibility tests of coagulase-negative staphylococci for nafcitlin accurately reflect concomitant resistance to cephatothin, cefazolin, and cefamandole. Nafcillin resistance apparently does not uniformly predict cephalosporin resistance. A secondary objective was to determine if the three cephalosporins showed differential activity against the nafcillin-resistant strains. Although cefazolin killed the organisms more slowly than cephalothin or cefamandole, each at 16 p,g/ml, there was no difference in the bactericidal action at 24 hr. The data from the agar dilution MIC tests and the bactericidal assays indicated that the objectives were met, in terms of in vitro results.
Extrapolation of the results to the in vivo setting would require a parallel trial in animals, but we are not convinced such a study should be done. Animal model experiments, although sometimes yielding useful information for human infection, often produce data that are difficult to apply to human S. epidermidis infections (Archer et al., 1980). Differences in serum protein binding of drug, pharmacokinetics of drug, and methods used to initiate infection in animal models introduce complexities that are sometimes difficult to interpret. Furthermore, not all coagulasenegative organisms have a similar virulence in animals. Baddour et al. (1984) consistently produced experimental endocarditis in rats with S. epidermidis, whereas they were rarely successful with S. hominis.
A clinical trial with cephalosporin therapy of selected patients infected with nafcillin-resistant coagulase-negative staphylococci could be done, but may not be warranted given the availability of acceptable alternative therapy. If a cephalosporin were to be used, each isolate would need to be tested for susceptibility to the lethal action of the cephalosporins in order to select the most appropriate drug.