In vitro Antimicrobial Activity of Ampicillin-Ceftriaxone and Ampicillin-Ertapenem Combinations Against Clinical Isolates of Enterococcus faecalis with High Levels of Aminoglycoside Resistance

This paper reports on the in vitro antimicrobial activity of ampicillin-ceftriaxone and ampicillin-ertapenem combinations against five strains of E. faecalis with high-level aminoglycoside resistance recovered from blood of septicemic patients. Double disk diffusion test and time killing curves were used. A bacteriostatic synergistic effect between ampicillin and ceftriaxone was detected using the disk diffusion assay for three of the five enterococcal strains studied. With the same three isolates enhanced bactericidal activity was also observed using time killing experiments. Overall, for these three strains, after 24 hr of contact, a decrease ≥ 2 log10 from the initial bacterial inoculum was registered with most ampicillin-ceftriaxone combinations, reaching with some of them a colony reduction ≥ 3 log10. This bactericidal interaction was negatively influenced increasing the bacterial inoculum. In all five isolates neither a bacteriostatic nor a bactericidal cooperation was observed for ampicillin combined with 2 mg/l of ertapenem. This investigation broadened the evidence of antimicrobial synergism in vitro between ampicillin and ceftriaxone in selected strains of Enterococcus faecalis with high-level aminoglycoside resistance.


BACKGROUND
Enterococci have been reported as major cause of acquired nosocomial infections including urinary tract infections, intra-abdominal and pelvic infections, bacteremia, endocarditis and meningitis. Enterococcus faecalis is responsible for about 80-90% of all enterococcal infections [1]. Severe enterococcal infections require a bactericidal therapy, which is usually a combination of a cell wall active agent such as ampicillin, penicillin or a glycopeptide plus gentamicin or streptomycin. However, the increasing resistance to high levels of aminoglycosides (HLAR) and to other antimicrobials leads to the loss of effective synergisms and thus bactericidal activity [1,2].
Enterococci are considered to be resistant to cephalosporins. Nonetheless, a synergistic effect between penicillin and cephalosporins has been reported in several in vitro studies. Mainardi observed an in vitro synergism between amoxicillin and cefotaxime in 48 out of 50 E. faecalis strains due to the saturation of PBPs 2 and 3 by cefotaxime in addition to the partial saturation of PBPs 4 and 5 with amoxicillin [3]. In another experiment, a synergistic effect between the two antimicrobials was not obtained in E. faecalis rabbit endocarditis because of the brief period of time in which amoxicillin and cefotaxime concentrations remained in the *Address correspondence to this author at the Infectious Disease Section, Department of Experimental Medicine and Biochemical Sciences, University of Perugia, Perugia, Italy; Tel +390755784359; Fax +390755784334; Email: pasticci@unipg.it range required to obtain synergism [4]. Gavaldà demonstrated a synergistic effect between ampicillin and ceftriaxone in vitro as well as in an experimental model of E. faecalis endocarditis [5][6][7]. The same author reported at three months an overall cure rate of 67.4% among 43 patients with E. faecalis endocarditis, 21 HLAR and 22 non-HLAR [6]. Others reported controversial results [8,10].
Enhanced bactericidal activity has been shown for ampicillin combined with imipenem in E. faecium experimental endocarditis [11]. Ertapenem is another carbapenem antibiotic with a broad spectrum of antibacterial activity against anaerobic and aerobic Gram negative bacteria, excluding the non-fermenters, and anaerobic and aerobic Gram positives, excluding methicillin resistant Staphylococcus aureus and enterococci [12][13][14]. In contrast to imipenem, it is less active against Pseudomonas spp., Acinetobacter spp. and Grampositive pathogens due to the poor affinity of the PBPs of these micoorganisms [12][13][14].
The purpose of this work was to further analyse the in vitro interaction between amipicillin and ceftriaxone and to evaluate that of ampicillin-ertapenem combination against five high-level aminoglycoside resistant E. faecalis strains isolated from patients with bacteremia.

METHODS
Bacterial strains. Five strains of E. faecalis isolated from bacteremic patients were studied. Bacterial identification was verified by standardized procedures (API 20 Strep, Bio-Merieux, France). Strains were stored at -70°C until use.
Before each experiment aliquots were thawed and subcultured in 5% sheep Columbia agar plates.
Antibiotic susceptibility studies. Beta-lactamase production was assessed with the nitrocephin disk test (Becton Dickinson, Sparks, Maryland, USA). Antibiotic susceptibility was determined by agar diffusion [15]. Etest (AB-BIODISK, Solna, Sweden), and with the Phoenix microdilution (Becton Dickinson) methods. For ampicillin, ceftriaxone and ertapenem MICs and MBCs were determined also using the standard macrodilution method in Müller-Hinton broth (MHB) with a bacterial inoculum of 2-5x10 5 CFU/ml [16][17]. Stock solutions of ampicillin, ceftriaxone or ertapenem were prepared and stored at -20°C until the day of use. To determine bactericidal activity of antibiotics 0.1 ml of the bacterial suspensions from each dilution tube were subcultured in Müller-Hinton agar (MHA) plates [18]. Plates were incubated at 37°C for 24 and 48 hours and colonies were counted at the same times. An effect was considered bactericidal when killing of 10 3 CFU/ml from the initial inoculum was recorded [18]. E. faecalis ATCC 29212 was used as control strain.
Time killing studies were performed in MHB supplemented with several concentrations of ampicillin, ceftriaxone, and ertapenem [5,19]. The final antibiotic concentrations were the following: ampicillin 0.5, 1, 2 and 4 mg/l (between MIC and sub-inhibitory concentrations); ceftriaxone 10, 20, 40 and 80 mg/l, between peak and trough values during patient treatment; ertapenem 2 mg/l, trough value. The bactericidal activity was evaluated for ampicillin alone or in combination with ceftriaxone or ertapenem. Test tubes were inoculated with a suspension of enterococcal strains at a final concentration of 2-5x10 5 CFU/ml or 2-5x10 7 CFU/ml and incubated at 37°C. A tube without antibiotics was used as positive growth control. The number of CFU/ml was obtained at time 0, after 4 and 24 hours of incubation plating on MHA a number of different dilutions from each tube. Data are the mean of CFU recorded for each plate. To reduce the carry over effect the inoculum was spread over the entire plate surface after the broth had dried [19].
To define antimicrobial interactions the criteria previously applied [5,20] have been utilized. Antimicrobial interactions were defined as additive when a 2 log 10 CFU/ml decrease between the combination and its most active agent alone was registered after 24 hr. If the reduction reached 3 log 10 the combination was considered to have a synergistic effect [5,20].

RESULTS
The five E. faecalis strains in the study were isolated from blood. Three were from patients with central line related sepsis, one from a patient with a liver abscess and another from a patient with prosthetic valve endocarditis ( Table  1). All the five strains were beta-lactamase negative. MIC values indicate that they were susceptible to ampicillin, linezolid, glycopeptides and resistant to ceftriaxone and ertapenem ( Table 1). All the isolates were highly resistant to aminoglycosides. Isolate 2927 had gentamicin MIC >500 mg/l, and streptomycin MIC 64 mg/l, while strain 2980 had gentamicin and streptomycin MIC respectively 8 mg/l and >1000 mg/l. The other three strains were both gentamicin and streptomycin resistant ( Table 1).
After 24 hours of incubation MBCs of ampicillin varied from 4 to above 32 mg/l ( Table 2), producing a bactericidal effect (99.9% killing of the initial bacterial inoculum) in four out of five strains. Only one strain (2980) appeared to be tolerant to ampicillin. Intermediate concentrations of ampicillin (2 and 4 mg/l) produced more efficacious bacterial killing for strains 2927 and 2980 ( Table 2). MICs of ceftriaxone and ertapenem were consistent with the resistance break points, therefore sub-cultures for MBC evaluations were not obtained.
A bacteriostatic synergistic effect between ampicillin and ceftriaxone, recorded with two + in Table 2, was observed with the disk diffusion assay for three of the five enterococcal strains evaluated ( Table 3). In broth, ampicillin MICs did not change or decreased of only one tube dilution when either ceftriaxone or ertapenem at concentrations of 10 mg/l and 2 mg/l, respectively, were added ( Table 3).
The time killing curves of ampicillin alone or in combination with 10 mg/l ceftriaxone showed after four hours of contact that 4 mg/l ampicillin produced a bacteriostatic effect with colony counts ranging from minus 0.3 to minus 2.9 log 10 in all the strains ( Table 4). At the same intervals, a bacteriostatic effect was also observed with 2 mg/l ampicillin for strains 2929 and 2980 and 1 mg/l against strain 2980. Ceftriaxone alone (10 mg/l) did not produce bacteriostatic effect. The addition of ceftriaxone (10 mg/l) to ampicillin did not modify the antibacterial activity of ampicillin for strains 2557 and 2648 either after 4 or 24 hr. On the contrary, a positive bactericidal interaction resulted for strains 2927, 2929 and 2980. Indeed, the combination 4 mg/l ampicillinceftriaxone produced a bactericidal effect for strains 2927 and 2980, 2 mg/l ampicillin-ceftriaxone for strains 2929 and 2980, 1 mg/l ampicillin-ceftriaxone for strain 2929. Overall, after 24 hr a synergism was recorded for the combinations 4 mg/l, 1 mg/l, 2 mg/l ampicillin-ceftriaxone against strains 2927, 2929 and 2980 respectively, while an additive effect was seen with the combinations 0.5 mg/l ampicillinceftriaxone against isolates 2927, 2929, and 2 mg/l ampicillin-ceftriaxone against strain 2927. Using an higher bacterial inoculum (10 7 CFU/ml) the synergism ampicillin-ceftriaxone was still detected only with 2 mg/l ampicillin for the isolate 2980 (data not shown). The synergism between ampicillinceftriaxone was not affected when 2 mg/l of ampicillin was combined with increasing ceftriaxone concentrations ( Table  4).
In all the experiments the bactericidal activity of ampicillin was never decreased by ceftriaxone.
When ampicillin was combined with 2 mg/l of ertapenem non consistent positive or negative interactions were registered against the strains tested ( Table 4).
In this study, the antimicrobial activity of ampicillin alone and in combination with ceftriaxone or ertapenem in five strains of high-level gentamicin and streptomycin resis- Values are change in log 10 CFU/ml With the broth macromethod MBC test, ampicillin alone produced a bactericidal effect (99.9% killing of the initial bacterial inoculum) in four out of five strains at concentrations equal or above those used for time killing experiments. Only one strain appeared to be tolerant to ampicillin. It is widely known that most enterococci are less likely killed at elevated concentrations of beta-lactam antibiotics [5,9,19]. Thus, the bactericidal effect of ampicillin in these experi-ments may have been due to the tested ampicillin concentrations which were very close to the MICs. Consequently, we may have been unable to evaluate the ampicillin tolerance of these strains [19]. Nevertheless, the purpose of the study was not to evaluate the ampicillin tolerance but rather to evaluate if any in vitro synergistic activity existed between ampicillin and ceftriaxone or ampicillin and ertapenem for these isolates.
Coming to this point, in broth MIC of ampicillin did not changed when a 10 mg/l concentration of ceftriaxone was added. On the contrary an increased antimicrobial activity of ampicillin and ceftriaxone was evidenced with the double disk test and the time killing curves in three out of five strains. With this last technique, non bactericidal concentrations of ampicillin produced an effective bacterial killing when ceftriaxone was added and some of the ampicillinceftriaxone combinations showed a real synergistic interaction. Other researchers have shown that saturation of different bacterial PBPs with ampicillin and ceftriaxone is responsible for the enhanced bacterial killing of this antimicrobial combinations [5] and that the effect was persistent for all the isolates tested [5]. Further experiments with an increased number of HLAR E. faecalis isolates will help to confirm strain variability to the ampicillin-ceftriaxone combination. More over, better standardized definitions of antimicrobial interactions will help to reduce variability among different studies [21].
All together, the results obtained on ampicillinceftriaxone interactions with this small investigation broadened the evidence of bacteriostatic and bactericidal synergism in vitro between ampicillin and ceftriaxone. This effect was evident only for three of the five strains evaluated and was not influenced by the ceftriaxone concentrations. Until further in vitro and clinical data are collected, this combination could be used only in selected clinical situations such as infections caused by highly antimicrobial resistant E. faecalis or the occurrence of treatment adverse events [10].
The data on ampicillin-ertapenem combination did not show in vitro synergism either with disk diffusion and time killing curves when an ertapenem concentration of 2 mg/l (concentration observed at trough during patient treatment) was added to ampicillin for all the five strains in the study. Given this, the bactericidal interactions with ertapenem concentrations higher than 2 mg/l was not investigated. Differences in ertapenem affinity for enterococcal PBPs may be a reason for the absence of antibacterial activity of this combination [22][23][24][25].
In conclusion, the results of this in vitro study broadened the evidence of bacteriostatic and bactericidal synergism in vitro between ampicillin and ceftriaxone. The effect was evident only for three of the five strains evaluated and was not influenced by the ceftriaxone concentrations. Double disk diffusion was predictive of time killing curve results.