MIC Profiling of Ceftazidime-Avibactam against two Carbapenemase producing Klebsiella pneumoniae Isolates

Abstract

Objective:

To correlate results of a modified susceptibility testing method to outcomes of ceftazidime-avibactam (CAZ/AVI) therapy.

Methods:

Two bloodstream K. pneumoniae isolates (CAZ/AVI susceptible) from an abdominal source were recovered from 2 unrelated patients. Both patients were treated with CAZ/AVI but had discordant outcomes: KP118 (eradication within 24h) and KP286 (persistent bacteremia for over 30 days). Carbapenemase production in the two isolates was confirmed via Carba NP test. CAZ minimum inhibitory concentration (MIC) was determined with escalating avibactam concentration (0 – 16 mg/L). The concentration-response was characterized by the sigmoid inhibitory maximum effect model. The best-fit parameter values were used to predict %T>MICi associated with CAZ/AVI exposures expected in peritoneal fluid after standard dosing (2.5g every 8h). These CAZ/AVI exposures were simulated in the hollow-fiber infection model (HFIM), and the bacterial responses were correlated to observed clinical outcomes.

Results:

The AVI-dependent reduction in CAZ MIC was well characterized in both bacterial isolates (r² > 0.98). In HFIM, sustained suppression of KP118 (T>MICi = 100%) was observed over 5 days, but not with KP286 (T>MICi < 100%). These observations are consistent with the clinical courses of the patients.

Conclusions:

The discordant patient outcomes could be potentially explained by MIC profiling of CAZ/AVI. This method appears to be more robust than conventional susceptibility testing in predicting positive clinical outcome of CAZ/AVI therapy, and the clinical utility of this approach should be further investigated.

1. Introduction

Antibiotic resistance is a global issue which threatens the outcomes of patients with severe infections. Multiple resistance mechanisms reducing the activity of antibiotics have been reported and they have the potential to spread rapidly among different bacteria. In the United States, multidrug-resistant organisms account for more than 2.8 million each year with an associated mortality of 35,000 deaths [1]. In addition, the Center of Disease Control estimated the cost of antibiotic resistance as $55 billion per year overall: $20 billion in excess for direct healthcare costs, with additional society burden for lost productivity as high as $35 billion per year [2]. Bacteria are versatile and have evolved over time to thrive in the presence of different antibiotic classes. Antibiotic resistance may be expressed intrinsically due to chromosomal gene mutations as well as acquired through horizontal gene transfer. The production of β-lactamase(s) is an example of enzymatic drug degradation and represents the most common resistance mechanism of β-lactams in Enterobacteriaceae. To regain activity, β-lactams have been coupled with β-lactamase inhibitors. Patient care is routinely guided by susceptibility testing with a single inhibitor concentration (e.g., 4 mg/L of tazobactam or avibactam). However, this methodology does not take into consideration varying inhibitor concentration observed in vivo and may not reliably predict positive clinical outcomes. We report two complex bacteremia cases due to carbapenemase-producing K. pneumoniae secondary to an abdominal source. We also explored a modified susceptibility testing method to predict outcome to ceftazidime-avibactam therapy.

2. Patient clinical courses

Two bloodstream K. pneumoniae isolates from an abdominal source were recovered from 2 unrelated patients. Both patients were treated with ceftazidime-avibactam but had discordant outcomes. KP118 was collected from a 73-year-old female with a medical history of pancreatic cancer. Approximately 11 months prior to admission, the patient had a pancreaticoduodenectomy procedure which was complicated by biliary stricture requiring a biliary drain placement. On admission day 3, patient had bacteremia due to carbapenem resistant Klebsiella pneumoniae (presumed source was cholangitis). The isolate was ceftazidime-avibactam susceptible (MIC = 1/4 mg/L) and she received ceftazidime-avibactam (1.25g IV every 8h - adjusted for estimated creatinine clearance of 38 mL/min). Patient underwent biliary drain replacement and metal stent placement for a hepatic pseudoaneurysm. Blood cultures cleared within 24 hours on therapy and remained negative, patient was discharged to complete a total of 14 days of therapy. KP286 was collected from a 40-year-old male with a medical history of abdominal gunshot wound who underwent pancreaticoduodenectomy roughly 20 years ago, requiring a chronic percutaneous transhepatic cholangiography drain. Seven days upon hospital admission, patient had bacteremia due to carbapenem resistant Klebsiella pneumoniae. The isolate was ceftazidime-avibactam susceptible (MIC = 2/4 mg/L); the patient was treated with ceftazidime-avibactam (1.25g IV every 8h - adjusted estimated for creatinine clearance of 43 mL/min) and the drain was replaced. Bacteremia was cleared within five days but recurred eight days later. Imagining of the abdomen was unremarkable and the right subclavian catheter was removed. Patient cleared bacteremia within two days and was later discharged to complete 14 days of therapy with ceftazidime-avibactam. Approximately two months later, patient was readmitted with abdominal pain and tenderness at the site of the drain with imaging consistent with a bile leak. All repeat blood cultures persistently revealed carbapenem resistant and ceftazidime-avibactam susceptible (MIC 2/4 mg/L) K. pneumoniae. The patient was again treated with ceftazidime-avibactam and metronidazole, but was ultimately discharged to hospice for comfort care.

3. Methods

Carbapenemase production in the 2 isolates was confirmed via CarbaNP test, and ceftazidime susceptibility (MIC) was determined with escalating avibactam concentration (0 – 16 mg/L) in the clinically relevant range. The concentration-response (i.e., MIC profiling) was characterized by the sigmoid inhibitory maximum effect (Emax) model [3]. The best-fit parameter values were used to predict %T>MICi associated with ceftazidime-avibactam exposures expected in peritoneal fluid after standard dosing (2.5g every 8h) [4,5]. These ceftazidime-avibactam exposures were subsequently simulated in the hollow-fiber infection model for each isolate, as previously described [3]. A high baseline inoculum (approximately 1 × 10⁸ CFU/mL) was used to mimic bacterial burden commonly encountered in intra-abdominal infections. Serial samples were obtained from the infection model to ascertain the pharmacokinetic simulations; ceftazidime and avibactam concentrations were assayed using a validated liquid chromatograph tandem mass spectrometry method [3]. Viable bacterial burden was also tracked over time by quantitative culture, and correlated to observed clinical outcomes.

3. Results

The avibactam-dependent reduction in ceftazidime MIC was well characterized in both isolates (r²> 0.98) (data not shown). In the hollow-fiber infection model, simulated ceftazidime-avibactam exposures were satisfactory. Typical pharmacokinetic profiles are shown in Figure 1. Bacterial responses are as shown in Figure 2. Considerable reduction in bacterial burden was observed within 24h for both isolates. Sustained suppression of KP118 (T>MICi = 100%) was observed over 5 days, but not with KP286 (T>MICi < 100%). Reduced susceptibility (>3× baseline MIC) to ceftazidime-avibactam was not detected over time. These observations appeared consistent with the clinical courses of the patients.

Figure 1.

Typical simulated ceftazidime (A) and avibactam (B) exposures in the hollow-fiber infection models

Note: open symbols represent experimental observationssolid lines depict best-fit models

Figure 2.

Bacterial responses to simulated ceftazidime-avibactam exposures

Note: data shown as mean ± SD

4. Discussion

We observed discordant outcomes to ceftazidime-avibactam therapy in two cases of bacteremia due to carbapenemase-producing K. pneumoniae. These observations corroborated well with bacterial responses to fluctuating ceftazidime-avibactam exposures (mimicking clinical dosing in humans) in the hollow-fiber infection model. The Clinical Laboratory and Standards Institute (CLSI) Enterobacteriaceae breakpoints for ceftazidime and ceftazidime-avibactam are 1 mg/L and 8/4 mg/L, respectively. While the KP286 isolate was susceptible with an MIC of 2/4 mg/L, this case may suggest a higher probability of ceftazidime-avibactam monotherapy treatment failure with MICs ranging from 2/4 to 8/4 mg/L. As such the role of ceftazidime-avibactam combination therapy warrants further investigations. Since both isolates were (categorical) susceptible to ceftazidime-avibactam, clinical outcomes observed would not have been predicted by conventional susceptibility testing. A limitation of conventional methods is that they do not fully capture different levels of enzymatic activity displayed by the isolates. Time above MIC (%T>MIC) is the most common pharmacokinetic / pharmacodynamic (PK/PD) index utilized to guide the optimal dosing of β-lactam antibiotics. However conventional indices may not be directly applicable to therapy involving β-lactamase inhibitors, as these inhibitors are routinely administered in combination with the β-lactam component. In contrast, our approach better reflects bacterial response to a range of β-lactamase inhibitor concentrations encountered in vivo. The relative magnitude of the projected %T>MICi might offer a pharmacological basis to explain the discordant clinical outcomes. The modified susceptibility testing appeared to be more robust than conventional methods in predicting favorable patient outcome of CAZ/AVI therapy.

Several limitations should be recognized. Most notably, patient outcomes observed could be influenced by severity of illness and adequacy of source control. Moreover, the study was limited to only two carbapenemase-producing isolates exposed to ceftazidime-avibactam. Additional isolates harboring other β-lactamases should be examined in future studies to ascertain if there is a robust threshold exposure to predict clinical success, and for other β-lactam/β-lactamase inhibitor combinations.

In conclusion, the discordant patient outcomes may be (at least partially) explained by MIC profiling of ceftazidime-avibactam. This method appears to be more robust than conventional susceptibility testing, and the clinical utility of this approach should be further investigated.

Highlights.

• We reported two bacteremia cases due to carbapenemase-producing K. pneumoniae.

• Both isolates were susceptibile to ceftazidime-avibactam by conventional testing.

• Discordant clinical outcomes were observed with ceftazidime-avibactam therapy.

• MIC profiling was used to predict %T>MICi exposures of ceftazidime-avibactam.

• MIC profiling appears robust in predicting positive clinical outcomes.