Piperacillin Pharmacokinetics and Pharmacodynamics in Paediatric Patients Who Received High Frequency Piperacillin/Tazobactam Dosing Intra-Operatively

Abstract

Piperacillin/tazobactam (PTZ) is a broad-spectrum antibiotic, typically dosed every six hours (q6h). Guidelines recommend dosing PTZ every 2 hours (q2h) intra-operatively for complex abdominal surgeries, including liver transplant. The data supporting the guidelines for intra-operative dosing are sparse and the pharmacokinetics/pharmacodynamics (PK/PD) of q2h dosing has never been studied by simulation or in humans. We compared PK/PD parameters of high-frequency intra-operative dosing and q6h post-operative dosing in critically ill children.

Paediatric patients who received PTZ during complex abdominal surgery or transplants and who had intra-operative and post-operative opportunistic samples were included. Using a published PK model and observed concentrations, individual piperacillin PK/PD parameters were estimated using Bayesian estimation. We simulated alternative post-operative dosing strategies using the patients with the highest and lowest estimated piperacillin clearance.

Thirteen patients were included (median age: 3.1 years, 85% liver transplant). PK parameters between intra-operative and post-operative phases were not significantly different (clearance: 15.8 +/− 7.2 vs. 12.6 +/− 6.3 L/h/70 kg, p =0.070; central volume: 13.4 [13.1, 13.8] vs 15.2 [12.2, 16.0] L/70 kg, p=0.22). At an individual level, intra-operative clearance values were - 35% to 139% of the post-operative values, while central volume intra-operative values were - 40% to 77% of the post-operative values. Intra-operative piperacillin exposures were higher during high-frequency dosing compared to the post-operative period (AUC/hour: 109 [93.4, 127] vs. 62.8 [41.6, 78.3] mg/L, p=0.002). Simulations showed great variation in optimal dosing strategies that would minimize toxicity and maximize efficacy, demonstrating a role for individualized dosing in paediatric surgical populations.

1. Introduction

Piperacillin/tazobactam (PTZ), whose active antibacterial component is piperacillin, is a broad-spectrum antibiotic typically dosed every 6 hours (q6h) in critically ill children [1]. PTZ has been associated with nephrotoxicity at high exposures in critically ill children and young adults [2]. The antibiotic is used as prophylaxis during complex abdominal surgeries and solid-organ transplants, such as liver and pancreatic transplants, due to its activity against Staphylococcus, Enterococcus and gram-negative aerobic bacilli [3]. When used in patients who undergo complex abdominal surgeries or transplants, intra-operative PTZ is administered every 2 hours (q2h) based on international guidelines [4]. This higher frequency dosing regimen is based on a study of adult liver transplant patients, which found that trough concentrations were too low to maintain target attainment (64mg/L), when dosed every 4 hours (q4h) [4]. This study suggested that intervals less than 4 hours would be sufficient. Therefore, the consensus guidelines for PTZ use intra-operatively during liver transplant recommends a dosing interval of 2 hours [3, 4]. The paediatric (>2 months) dosing regimen is 100-112.5 mg/kg q2h PTZ (80-100mg/kg piperacillin component q2h) intra-operatively [3].

Higher frequency dosing has not been studied by simulation or in humans. This lack of data is concerning due to reports of piperacillin-associated acute kidney injury (AKI) in children [2, 5]. Adult studies have found that PTZ treatment is associated with increased risk of developing AKI, as defined by creatinine-based criteria, especially when used with other nephrotoxic medications [6]. The relationship between PTZ and nephrotoxicity in children are not as clear [7] but an association between high piperacillin concentrations and creatinine-defined AKI has been shown [2]. While increasing frequency of dosing during surgery would allow for target attainment, based on time of free piperacillin concentrations over a certain minimum inhibitory concentration (MIC), there is risk of increased exposure and therefore AKI, which has been associated with increased hospital length of stay (LOS), cost, morbidity and mortality in children [8, 9].

Characterizing the pharmacokinetics/pharmacodynamics (PK/PD) parameters of piperacillin in patients who receive high frequency dosing intra-operatively would allow for assessment of the risk of developing AKI. Our primary aim was to describe piperacillin PK/PD in paediatric patients who received high-frequency q2h PTZ dosing while undergoing complex abdominal surgeries and transplants and to compare to piperacillin PK/PD post-operatively when receiving standard q6h dosing. Our secondary aim was to understand how higher frequency dosing can affect the development and incidence of AKI.

2. Patients and Methods

2.1. Patient Population

We conducted a retrospective study with data collected from a larger prospective β-lactam study [10] that was approved by the Cincinnati Children’s Hospital Medical Center (CCHMC) Institutional Review Board. Our patient population included patients admitted to CCHMC where PTZ was administered during an abdominal surgical procedure between October 2018 and June 2021. These patients received high-frequency PTZ (q2-3h) dosing intra-operatively during complex abdominal surgeries, including liver transplant and total pancreatectomy with islet autotransplantation (TPIAT). Post-operatively, patients received q6h dosing while in the paediatric intensive care unit (PICU). In our PICU, patients receive post-operative PTZ prophylaxis for at least 48-72 hours but this duration may be extended for patients who have open fascia after initial surgery. Patients who were on extracorporeal support devices that are known to affect drug clearance, such as Continual Kidney Replacement Therapy (CKRT) or Extracorporeal membrane oxygenation (ECMO), were excluded. Our initial patient population meeting inclusion criteria were 24 patients.

2.2. Piperacillin concentration measurements

We used a scavenged opportunistic sampling approach to measure piperacillin concentrations. Blood from clinical samples that were collected in the first seven days of the abdominal surgery and within 24 hours after a dose of PTZ were obtained to measure free piperacillin concentrations. Opportunistic samples that were collected during the 30-minute infusion time of the antibiotic were excluded. All other samples were included. For the intra-operative period when dosing was every 2 hours, samples could be obtained 0.5-2 hours postdose; for the post-operative period when dosing was every 6 hours, samples could be obtained 0.5-6 hours post-dose. Our method of collection and storage of opportunistic samples has been described previously [10, 11]. Free and total piperacillin concentrations were measured using a previously validated high-performance liquid chromatography (HPLC) assay [10].

2.3. PK/PD parameter estimates

We used a previously published two-compartment piperacillin PK model in paediatric critically ill children [12] and observed concentrations from opportunistic sampling to estimate clearance and central volume (V1) and piperacillin exposure metrics for the intra-operative and post-operative states using Bayesian estimation in MwPharm++ (Mediware, Czech Republic). The model was selected since it incorporated weight as a covariate on all four parameters (allometric scaling for clearance) and maturation effect on clearance. A visual check was performed to ensure that the fit of the concentration versus time profile was appropriate with respect to the concentrations measured and concentrations were within the 95^th% confidence interval generated by the program. For each patient, we estimated PK/PD parameters in the intra-operative state using only concentrations obtained while the patients were in the operating room. Post-operative parameters were estimated using concentrations collected during the second through ninth dose of PTZ treatment in the PICU, or when PICU PTZ treatment ended if this occurred before the ninth dose (Supp. Figure 1). We excluded concentrations collected when patients transitioned between the operating room and PICU and between the first and second doses in the PICU, since this is a highly dynamic and unstable period, during which there may be multiple interventions affecting PK/PD, including fluid boluses or blood administration. Concentrations collected after the first 48 hours in the PICU were excluded because after this period patients are stabilized and are returning to pre-surgical states.

We estimated clearance and V1, scaled to 70 kg (allometric for clearance, linear for volume), to allow for comparison between patients [13]. We estimated piperacillin area under the curve (AUC) normalized per hour and the concentrations from the minimum and maximum peaks and troughs as metrics of piperacillin PK exposure. As a marker of antibacterial effectiveness, we estimated the percentage of time that free concentrations were above the minimum inhibitory concentration (%T>MIC) [14]. The reported MIC values were based on CLSI breakpoints for Enterococcus/Enterobacterales and Pseudomonas aeruginosa bacteria, which are 8 mg/L and 16 mg/L respectively [15]. There remains a lack of consensus of the actual PD target in critically ill children [16]. Therefore, we chose two stringent targets and measured the %T>MIC at the CLSI breakpoint and 4x the breakpoint for both Enterococcus/Enterobacterales and Pseudomonas aeruginosa.

2.4. PTZ-AKI determination

Acute kidney injury (AKI) associated with PTZ exposure (PTZ-AKI) was determined based on our previous study of a larger cohort that included the patients in this study [2]. Briefly, we used three established criteria for drug-associated AKI: (i) AKI present for 24 hours to 7 days after the first PTZ dose, (ii) AKI that is at least Kidney Disease Improving Global Outcomes (KDIGO) stage 2, which is measured by creatinine or urine output criteria and (iii) For patients with baseline creatinine <0.5 mg/dL, increased creatinine must be greater than 0.5 mg/dL. Adjudicators determined which PTZ-AKI criteria were met [2]. If a patient met the first two or all three of the criteria, the probability of PTZ-AKI was evaluated using the Naranjo Adverse Drug Reaction Probability Scale [17]. We evaluated patients who met at least the first two criteria since the third criterion would exclude patients of younger ages. Those who met the first two PTZ-AKI criteria and evaluated possible, probable or definite on the Naranjo scale were considered as having PTZ-AKI.

2.5. Clinical Data Collection

We conducted a chart review for data regarding demographics, hospitalization characteristics and outcomes. Demographic data included age, sex, race, weight, height, and body mass index (BMI) of the patient. Hospitalization characteristics and outcomes collected included the length of stay (LOS) in the hospital and in the PICU, days on piperacillin and comorbid conditions. Comorbid condition was defined as receiving medication or requiring subspecialty care for at least one condition. Clinical data were recorded and stored in REDCap [18].

2.6. Statistical Analysis

Student’s t-tests and Wilcoxon sign rank tests were performed to compare PK parameters between the intra-operative and post-operative states. The %T>MIC was measured as both a dichotomous and continuous variable. McNemar’s test was used to compare the dichotomous outcome and Wilcoxon sign rank was used to compare the continuous measurements.

2.7. Target Attainment Evaluation

We evaluated target attainment using model-informed simulations for the post-operative period using the PK parameters of the two patients with the highest clearance value and lowest clearance value to define the range of dosing regimens to consider for this patient population. We simulated continuous infusions of 100 mg/kg, 200 mg/kg, 300 mg/kg, and 400 mg/kg of PTZ given over 24 hours and dosing regimens based on a previous study in critically ill patients [12], which included 75 mg/kg (based on piperacillin) given over 2 hours every 4 hours, 100 mg/kg given over 1 hour every 4 hours, and a loading dose of 75 mg/kg followed by a continuous infusion of 300 mg/kg over 24 hours. For the continuous infusion regimens, our steady state concentration goal was between 32 mg/L (4x Enterococcus/Enterobacterales CLSI breakpoint) and 50.1 mg/L, the value which was previously reported to be the mean trough among patients with PTZ-AKI [2]. For non-continuous dosing regimens, we aimed for trough concentrations to be above 4xMIC for Enterococcus and below 50.1 mg/L.

3. Results

3.1. Demographic and hospitalization characteristics

From the initial cohort of 24, patients were excluded for lacking opportunistic samples from both periods, receiving doses of PTZ from other institutions without documentation of timing, or if modeling and simulation of antibiotic concentrations was inadequate, such as in the case of post-operative kidney failure that resulted in eventual CKRT (Figure 1). The final patient population consisted of 13 patients. Demographics, intra-operative characteristics and hospitalization characteristics are shown in Table 1 and Supplementary Table 1. No patients died within 30 days of surgery. All patients had a decrease in their estimated glomerular filtration rate (eGFR) based on creatinine (Supplementary Table 1) during their surgery (range: 18-63% decrease). There were 93 concentrations among the 13 patients (average 7.2 samples/patient), with an average of 1.7 samples/patient in the intra-operative period and 5.5 samples/patient in the post-operative period.

Figure 1. Diagram of reasons for exclusion.

CKRT: Continuous kidney replacement therapy.

Table 1.

Demographics and Hospitalization Characteristics of Final Cohort of Patients who Received High Frequency PTZ Dosing Intra-Operatively.

Number of patients	N=13
Reason for surgery
Liver Transplant	10 (76.9%)
TPIAT	1 (7.69%)
Other Abdominal Surgery	2 (15.4%)
Age (years)
Median [IQR]	3.11 [1.51, 12.8]
Weight (kg)
Median [IQR]	13.8 [9.3, 34.2]
Sex Assigned at Birth
Female	10 (76.9%)
Male	3 (23.1%)
Self-Identified Race patients may identify with more than one race
White	3 (23.1%)
Black	6 (46.2%)
Hispanic	4 (30.8%)
Unknown	1 (7.69%)
PICU Length of Stay (days)
Median [IQR]	4 [3, 4]
Hospital Length of Stay (days)
Median [IQR]	28 [16, 72]
Days on Piperacillin
Median [IQR]	5 [3, 8]
Number of patients with positive culture and description of culture	1, Enterobacter in urine, MIC of 8 mg/L
Number of intra-operative samples taken
Mean +/− SD	1.7 +/− 0.75
Number of post-operative samples taken
Mean +/− SD	5.5 +/− 2.5

IQR: interquartile range; TPIAT: total pancreatectomy with islet autotransplantation; PICU: pediatric intensive care unit; SD: standard deviation; MIC: minimum inhibitory concentration

3.2. Pharmacokinetic differences between both states

Despite a decrease in eGFR during surgery, piperacillin clearance and central volume were not significantly different between the intra-operative and post-operative groups (clearance: 15.8 +/− 7.2 vs. 12.6 +/− 6.3 L/h/70 kg, p =0.07; central volume: 13.4 [13.1, 13.8] vs 15.2 [12.2, 16.0] L/70 kg, p=0.22) (Table 1). There was, however, large intra-individual variability between the two states. Clearance intra-operative values were −35% to 139% of the post-operative values, meaning some patients had lower intra-operative clearance than post-operative clearance and others had higher intra-operative clearance. Central volume intra-operative values were −40% to 77% of the post-operative values.

3.3. Differences in estimated free piperacillin exposure between both states

There were significant differences in exposure metrics between both dosing regimens. Specifically, the AUC normalized per hour and maximum free trough concentrations were higher in the intra-operative state with higher frequency dosing compared to the post-operative state (AUC per hour: 109 [93.4, 127] vs. 62.8 [41.6, 78.3] mg/L, p=0.002; maximum trough: 81.8 [40.8, 109] vs. 14.3 [7.79, 28.5] mg/L, p<0.001) (Table 2).

Table 2.

Comparison of PK/PD data and target attainment during intra-operative and post-operative periods patients (n=13).

	Intra-op (n=13)	Post-op (n=13)	p-value
Clearance (CL) of Piperacillin (L/h/70kg), mean ± SD†	15.8 ± 7.2	12.6 ± 6.3	0.070
Central Volume (V1) of Piperacillin (L/70kg), median [IQR]‡	13.4 [13.1, 13.8]	15.2 [12.2, 16.0]	0.22
Maximum Peak Concentration (mg/L), Median[IQR]‡	276 [255, 336]	244 [220, 273]	0.017
Minimum Peak Concentration (mg/L), Mean ± SD†	236 ± 54.4	221 ± 38.7	0.29
Maximum Trough Concentration (mg/L), Median[IQR]‡	81.8 [40.8, 109]	14.3 [7.79, 28.5]	<0.001
Minimum Trough Concentration (mg/L), Median[IQR]‡	54.4 [37.3, 68.3]	5.95 [2.32, 17.2]	<0.001
AUC normalized per hour (mg/L), Median[IQR]‡	109 [93.4, 127]	62.8 [41.6, 78.3]	0.002
Percent of dosing interval free piperacillin concentrations remain above 8 mg/L (4xMIC for Enterococcus) (%) Median [IQR]†	100 [100, 100]	97.8 [84.5, 100]	0.014
Percent of dosing interval free piperacillin concentrations remain above 32 mg/L (4xMIC for Enterococcus) (%) Median [IQR]†	100 [100, 100]	70.4 [39.5, 83.0]	0.003
Percent of dosing interval free piperacillin concentrations remain above 16 mg/L (4xMIC for Pseudomonas aeruginosa) (%) Median [IQR]†	100 [100, 100]	82.4 [59.3, 100]	0.014
Percent of dosing interval free piperacillin concentrations remain above 64 mg/L (4xMIC for Pseudomonas aeruginosa) (%) Median [IQR]†	89.5 [66.1, 100]	33.6 [23.3, 60.6]	<0.001

^†t-test;

^‡Wilcoxon Sign Rank.

IQR: interquartile range; SD: standard deviation; AUC: area under the curve; MIC: minimum inhibitory concentration; intra-op: intra-operative; post-op: post-operative.

3.4. Target attainment compared between both states

For both Enterococcus and Pseudomonas aeruginosa, the %T>1x MIC and 4x MIC was greater during the intra-operative state than in the post-operative state (Table 2). Attainment of free piperacillin concentrations for 100% over the MIC occurred more frequently in the intra-operative state. When using the breakpoint for Enterococcus, 13 patients (100%) met 100%T>MIC in the intra-operative state compared to 5 patients (38.5%) in the post-operative state. For 100%T>4xMIC for Enterococcus, 11 patients (84.6%) met target for 100% of the dosing interval in the intra-operative state and 1 patient (7.7%) met the target for the post-operative period. When using the breakpoint for Pseudomonas aeruginosa, 13 patients (100%) met 100%T>MIC in the intra-operative state compared to 5 patients (38.5%) in the post-operative state. For 100%T>4xMIC for Pseudomonas aeruginosa, 4 patients (30.8%) met target in intra-operative state and no patients met the target for the post-operative period.

3.6. Piperacillin-tazobactam-associated acute kidney injury

In our previous study to determine the prevalence of PTZ-AKI in a larger cohort of 107 critically ill children [2], which included the surgical patients in this study, we found that 35% of patients met at least the first two criteria for PTZ-AKI and AKI was probably or possibly due to piperacillin. When we removed the 13 surgical patients from the previous cohort of 107 patients, 29 of the remaining 94 non-surgical critically ill patients (31%) met at least the first two PTZ-AKI criteria and were considered as possible or probable on the causality scale. When specifically evaluating the 13 surgical patients in this cohort, we found that 3 patients (23.1%) met the first two criteria for PTZ-AKI only and were considered as possible or probable, while 3 patients (23.1%) met all three criteria for AKI, indicating that 6 patients (46.2%) met at least the first two criteria for AKI, in comparison to 31% of the non-surgical critically ill patients.

3.5. Simulation to evaluate different dosing regimens across spectrum of clearance values

Since target attainment was consistently low in the post-operative state, we simulated various dosing regimens to maximize efficacy and reduce toxicity. For the patient with the lowest clearance of 4.37 L/h/70kg, we found that none of the simulated dosing regimens met our criteria of being above 32 mg/L to ensure efficacy and below 50.1 mg/L to minimize toxicity as all trough concentrations and steady state concentrations were above 50.1 mg/L. However, for the patient with the highest clearance of 23.5 L/h/70kg, we found a continuous infusion of 400 mg/kg over 24 hours would provide steady state concentrations between 32 mg/L and 50.1 mg/L.

4. Discussion

In our surgical cohort, we found no significant difference between clearance and V1 between the intra-operative and post-operative states. We did note that eGFR did decrease throughout surgery with patients having lower eGFR’s at the time of admission to the PICU compared to prior to surgery. There was a trend towards a decrease in piperacillin clearance between intra-operative and post-operative periods (15.8 +/− 7.2 vs. 12.6 +/− 6.3 L/h/70 kg, p=0.07) but not significant, which may be due to the small cohort size and that piperacillin clearance is not solely dependent on eGFR based on creatinine. We also found that there was great individual variability in the clearance and V1 within our patient population in that piperacillin clearance and volume could be higher or lower intra-operatively compared to the post-operative period.

Piperacillin exposure levels were higher during the intra-operative period when higher frequency dosing occurs. While the higher exposures may prevent infection, there are concerns for increased risk of nephrotoxicity. The estimated intra-operative exposure was concerning for developing PTZ-AKI as the median maximum trough concentration in this study was 81.8 mg/L and our previous study showed that patients who developed PTZ-AKI had a mean maximum trough value of 50.1 mg/L [2]. Compared to a general population of non-surgical critically ill children where 31% developed PTZ-AKI by meeting at least two criteria [2], we found that 46% of our patients met the definition of PTZ-AKI. Conversely, the lower post-operative exposures found in this study may put patients at risk for infections.

When evaluating simulated dosing regimens between the patient with the highest and lowest clearance in the intra-operative state (5-fold difference), we could not find a consistent dosing regimen that would maximize efficacy while minimizing toxicity. Developing population PK models for piperacillin in similar patients to use in Bayesian estimation for individualized PK/PD profiles in patients undergoing abdominal surgery would be warranted.

There are limitations to this study. Our small cohort contained a high level of heterogeneity which likely impacted our ability to describe the relationship between clearance and V1 between the two periods. This study did not use targeted, densely collected samples, but instead relied on opportunistic samples to reduce blood draws and to enroll more patients. However, we used Bayesian estimation to estimate PK parameters from opportunistic samples and a prior population PK model to develop individual PK/PD profiles.

5. Conclusion

Our surgical cohort demonstrated high inter-individual variability in clearance and central volume within the intra-operative and post-operative states, with high piperacillin PK exposure in the intra-operative period that may place patients at risk for piperacillin-associated acute kidney injury. The high variability may suggest a role for model-informed precision dosing to achieve efficacy while minimizing toxicity at an individual patient’s level.