High-Dose Ampicillin-Sulbactam Combinations Combat Polymyxin-Resistant Acinetobacter baumannii in a Hollow-Fiber Infection Model

ABSTRACT

Acinetobacter baumannii is emerging with resistance to polymyxins. In 24-h time-kill experiments, high-dose ampicillin-sulbactam in combination with meropenem and polymyxin B achieved additivity or synergy against 10⁸ CFU/ml of two clinical A. baumannii isolates resistant to all three drugs (maximum reductions, 1.6 and 3.1 logs). In a 14-day hollow-fiber infection model, high-dose ampicillin-sulbactam (8/4 g every 8 h, respectively) in combination with meropenem (2 g every 8 h) and polymyxin B (1.43 mg/kg of body weight every 12 h with loading dose) resulted in rapid (96 h) eradication of A. baumannii.

TEXT

Acinetobacter baumannii is a troubling nosocomial pathogen with an exceptional propensity for acquiring resistance mechanisms against commonly used antimicrobials (1). Although carbapenems have traditionally been the drug of choice for treating A. baumannii infection, enzyme-mediated hydrolysis of carbapenems has forced clinicians to utilize polymyxins as a drug of last resort against extensively drug-resistant A. baumannii (2, 3). Unfortunately, the emergence of A. baumannii strains resistant to polymyxins has prompted the search for novel dosing schemes and combination regimens that overcome such extreme levels of drug resistance (4–6).

One proposed strategy for combating drug resistance in A. baumannii is the use of high-dose ampicillin-sulbactam regimens that have been evaluated in exploratory clinical studies (7, 8). The combination of ampicillin-sulbactam with a carbapenem and a polymyxin has also demonstrated a promising mortality benefit against colistin-resistant A. baumannii (9); however, the pharmacodynamics of high-dose ampicillin-sulbactam alone and in combination with other agents has yet to be fully defined. In the present study, time-kill experiments were utilized to investigate the level of killing by high-dose ampicillin-sulbactam, meropenem, and polymyxin B alone and in double/triple combinations against A. baumannii resistant to each of these antibiotics based on MIC testing. A hollow-fiber infection model (HFIM) was subsequently used to fully define the time course of A. baumannii killing over 14 days.

Time-kill experiments were conducted on two clinical A. baumannii strains (N5406 and 03-149-2) using an established methodology (10) that was adapted to accommodate a high bacterial burden of 10⁸ CFU/ml to approximate the burden involved with a life-threatening A. baumannii infection, as conducted previously (11). Both isolates were resistant to ampicillin-sulbactam (MIC, 32/16 mg/liter), meropenem (MIC, 64 mg/liter), and polymyxin B (MICs, 64 mg/liter [strain N5406] and 32 mg/liter [strain 03-149-2]). Briefly, overnight cultures of N5406 and 03-149-2 were added to Mueller-Hinton broth adjusted with magnesium (12.5 mg/liter) and calcium (25 mg/liter) to achieve a 10⁸ CFU/ml starting inoculum (20 ml total volume). Fresh stocks of each antibacterial were prepared on the day of each experiment to achieve clinically achievable concentrations of 132/70 mg/liter of ampicillin-sulbactam (AK Scientific, Union City, CA), 55 mg/liter of meropenem (AK Scientific, Union City, CA), and 1.5 mg/liter of polymyxin B (Sigma-Aldrich, St. Louis, MO) (7, 12, 13). Samples were collected at 0, 1, 2, 4, 8, and 24 h, serially diluted in saline, plated on Mueller-Hinton agar, and counted after 24 h of incubation. Due to consistent colony formation regardless of the magnitude of the saline dilution, centrifugation of samples to prevent antibiotic carryover was not performed. All experiments were performed in duplicate on separate days. In comparison to the most active single agent in a combination, additivity was said to occur if an additional ≥1 log reduction was achieved by the combination, whereas synergy occurred when an additional ≥2 log reduction was achieved by the combination, at any time point (14).

To fully define the time course of A. baumannii killing, strain N5406 was subsequently investigated in a 14-day HFIM as previously described (15). Using a 10⁸ CFU/ml starting inoculum, bacterial samples were collected at 0, 1, 2, 3, 4, 6, 24, 26, 28, 30, 48, 50, 52, 54, 72, 74, 76, 78, 96, 144, 192, 240, 288, and 336 h for viable cell counting. To simulate the human kinetics observed during aggressive antibiotic treatment, the dose of ampicillin-sulbactam was derived from a prospective study of intensified ampicillin-sulbactam dosing (7, 16), meropenem concentrations were based on the highest dose supported by the manufacturer (12), and polymyxin B dosing was adapted from a population pharmacokinetic study as detailed previously (13, 17). The following regimens were simulated in the HFIM as monotherapies, double combinations, and a triple combination: ampicillin-sulbactam, administered at 8/4 g every 8 h (maximum concentration of free drug in serum [fC_max], 132/70.2 mg/liter; half-life [t_1/2], 1.5 h), meropenem, 2 g every 8 h (fC_max, 54.8 mg/liter; t_1/2, 1.5 h; 3-h prolonged infusion), and 3.33 mg/kg of body weight of polymyxin B at 0 h (fC_max, 3.61 mg/liter; t_1/2, 8 h) followed by 1.43 mg/kg every 12 h thereafter (fC_max, 2.41 mg/liter). During combination experiments, polymyxin B was supplemented to maintain a profile consistent with an 8-h half-life (18).

The pharmacokinetics of polymyxin B were confirmed using a previously validated liquid chromatography single-quadrupole mass spectrometry (LC-MS) method that resulted in observed concentrations ≤11.0% from target concentrations, on average (see “Supplementary Pharmacokinetic Validation” in the supplemental material for expected versus observed concentration profiles) (19). Meropenem concentrations were also quantified using a liquid chromatography tandem mass spectrometry method (LC-MS/MS) that utilized an Agilent (Santa Clara, CA) 1200 and Agilent 6430 LC-MS systems. The meropenem calibration curve was linear with an R² value of >0.999 with good reproducibility (coefficient of variation, ≤3.57%) and accuracy (99.7 to 109.4%). Expected versus observed concentrations were within 6% for all samples, with an R² value of 0.99. Sulbactam concentrations were determined via triple-quadrupole LC-MS/MS that utilized an Agilent 640 series with an Agilent 1260 binary pump autosampler. The sulbactam calibration curve was linear over a range of 1 to 100 mg/liter (R² > 0.999) and 109% accuracy for control samples. The expected versus observed concentrations were in good agreement with a mean ratio of observed/expected concentrations of 1.038 and an R² value of 0.963.

The results of the time-kill experiments are displayed in Fig. 1. Against strain N5406, none of the individual agents were able to reduce bacterial counts by ≥1 log at any time point. Similarly, none of the combinations achieved a ≥2-log reduction by 24 h. The most active double combination was polymyxin B–meropenem, which achieved a 1.1-log reduction by 8 h. The ampicillin-sulbactam–polymyxin B–meropenem triple combination was the only regimen that achieved sustained additivity with a 1.4-log reduction at 8 h, followed by a 1.6-log reduction at 24 h. The antibiotic combinations were more active against strain 03-149-2 than against strain N5406. The double combinations of ampicillin-sulbactam–meropenem and polymyxin B–meropenem achieved synergy by 8 h with 2.3- and 2.9-log reductions, respectively. Overall, the triple combination was the most active combination against 03-149-2, with synergetic killing at 8 and 24 h (reductions of 3.1 and 2.2 logs, respectively).

FIG 1.

Time-kill experiments of ampicillin-sulbactam (132/70 mg/liter), meropenem (55 mg/liter), and polymyxin B (1.5 mg/liter) and their double and triple combinations against 10⁸ CFU/ml of two clinical A. baumannii strains resistant to each antibiotic. Experiments were performed in duplicate, and standard deviations are indicated by error bars.

To further define the pharmacodynamics of the antibiotics alone and in combination, N5406 was investigated in an HFIM. The results are summarized in Fig. 2. Both polymyxin B and meropenem alone failed to achieve a ≥1-log reduction by 6 h, with counts for the duration of the experiment mirroring that of the growth control. The polymyxin B–meropenem combination reduced counts by ≥2 logs at 6 h. Stasis ensued for 24 h, but by 48 h, counts had risen above 10⁸ CFU/ml. In contrast, not only did the double combinations of ampicillin-sulbactam with either meropenem or polymyxin B achieve sustained bactericidal activity, but ampicillin-sulbactam alone also eradicated N5406 by 144 h, albeit with killing over the first 72 h that was slower than that with the ampicillin-sulbactam double combinations. The triple combination displayed the most dramatic killing with a 3.2-log reduction by 6 h, whereas ampicillin-sulbactam alone and the ampicillin-sulbactam–polymyxin B and ampicillin-sulbactam–meropenem double combinations achieved 0.1-, 0.6-, and 2.3-log reductions, respectively. By 96 h, counts for the triple combination fell below the limit of detection (∼100 CFU/ml), while ampicillin-sulbactam alone and the ampicillin-sulbactam–polymyxin B and ampicillin-sulbactam–meropenem double combinations maintained counts at <10³ CFU/ml until A. baumannii was eradicated later in the experiments. Although the ampicillin-sulbactam–meropenem and ampicillin-sulbactam–polymyxin B double combinations initially killed more rapidly than ampicillin-sulbactam alone, the two combinations took longer to push bacterial counts below the limit of detection.

FIG 2.

The 14-day HFIM results of growth control (A), ampicillin-sulbactam (B), meropenem (C), and polymyxin B (D) alone, in double combinations (E, F, and G), and in a triple combination (H) against 10⁸ CFU/ml of strain N5406. The limit of detection was ∼100 CFU/ml. Antibiotic regimens were 8/4 g of ampicillin-sulbactam every 8 h (fC_max, 132/70.2 mg/liter; t_1/2, 1.5 h), 2 g of meropenem every 8 h (fC_max, 54.8 mg/liter; t_1/2, 1.5 h; 3-h prolonged infusion), and 3.33 mg/kg of polymyxin B at 0 h (fC_max, 3.61 mg/liter; t_1/2, 8 h) and then 1.43 mg/kg every 12 h thereafter (fC_max, 2.41 mg/liter).

Strains of A. baumannii that are capable of mounting resistance mechanisms against nearly all of the antimicrobials available in the global armamentarium are emerging. Here, we have defined the pharmacodynamics of high-dose ampicillin-sulbactam in combination with meropenem and/or polymyxin B against two strains of A. baumannii that were resistant to all of the agents used in the investigation. The ampicillin-sulbactam killing of multidrug-resistant A. baumannii is attributed predominantly to the intrinsic activity of sulbactam, which binds to penicillin-binding protein 1 (PBP1) and PBP3 (20), with pharmacokinetics/pharmacodynamics best described by the percentage of time in which the free drug concentration remains above the MIC (T>MIC) (21). In this study, the large dose of sulbactam investigated in the HFIM maintained a high T>MIC of ∼74% that likely explains the drastic killing observed during ampicillin-sulbactam monotherapy. Accelerated killing observed in ampicillin-sulbactam combination regimens was likely due to the increased permeability of the outer membrane conferred by polymyxin B (3) or the strong affinity of meropenem for complementary PBPs, such as PBP2 (12). The lack of killing achieved by ampicillin-sulbactam alone in static time-kill experiments may suggest that rapid division of A. baumannii (as observed in the HFIM) is necessary for optimal killing.

The rapid eradication of A. baumannii achieved by the triple combination in the HFIM is consistent with the clinical use of ampicillin-sulbactam in combination with a carbapenem and a polymyxin. In a retrospective study that evaluated 17 patients treated for colistin-resistant A. baumannii infection (13/17 had ventilator-associated pneumonia [VAP]), 7 patients received ampicillin-sulbactam–carbapenem–colistin and survived for at least 30 days following their diagnoses, whereas 6/10 of the patients who received other antibiotic combinations died (P = 0.3) (9). Of the 6 patients who died, 3 had received either ampicillin-sulbactam alone, ampicillin-sulbactam–tigecycline, or ampicillin-sulbactam–tigecycline–colistin. In our study, the combination of ampicillin-sulbactam with meropenem and polymyxin B was the most active regimen in both time-kill and HFIM experiments, highlighting that ampicillin-sulbactam may be best utilized in combination with a carbapenem and a polymyxin.

Similarly, the eventual eradication of A. baumannii resistant to ampicillin-sulbactam in the HFIM after exposure to the high-dose ampicillin-sulbactam regimen (8/4 g every 8 h) is consistent with the clinical use of high-dose ampicillin-sulbactam regimens. In a prospective study of VAP due to ampicillin-sulbactam-resistant A. baumannii (MIC, >32/16 mg/liter in all patients), 14 patients were administered 6/3 g of ampicillin-sulbactam every 8 h, whereas a second group of 13 patients was treated with 8/4 g of ampicillin-sulbactam every 8 h (7). Despite high levels of antibiotic resistance (64.3% and 69.2%, respectively) patients in the two groups achieved clinical success, and no major adverse events were reported. In a subsequent investigation, patients with VAP due to multidrug-resistant A. baumannii were treated with either 6/3 g of ampicillin-sulbactam every 8 h or 3 MIU of colistin every 8 h (8). Although the A. baumannii strains were resistant to ampicillin-sulbactam, 28-day mortality rates were comparable between the ampicillin-sulbactam and the colistin groups (30% versus 33%, respectively), and rates of nephrotoxicity were lower in the ampicillin-sulbactam group (15% versus 33%, respectively). However, because the pharmacokinetics of the patients enrolled in the high-dose ampicillin-sulbactam studies were not evaluated (7, 8), the drug concentrations used in our study were based on ampicillin-sulbactam concentrations in healthy volunteers (16) and were scaled for the higher dose, potentially obscuring the translation of the current results into the clinic.

In conclusion, high-dose ampicillin-sulbactam in combination with meropenem and polymyxin B was the most active of all of the regimens evaluated against A. baumannii resistant to all three drugs in 24 h time-kill experiments and a 14 day HFIM. The eradication observed in the HFIM suggests that clinicians may be able to overcome A. baumannii resistance mechanisms through dose manipulation and the strategic selection of combination regimens, thus expanding treatment options for polymyxin-resistant A. baumannii infections. Further experiments in which the dose of ampicillin-sulbactam is titrated to establish a dose-response curve are required to fully inform rational dosing of ampicillin-sulbactam in the face of extensively drug-resistant A. baumannii strains.