Study design and population
This prospective observational PK/PD study was conducted at the cardiac intensive care unit (CCU) in Severance Cardiovascular Hospital between November 2015 and January 2019. The protocol for the study was approved by the Institutional Review Board (IRB No. 4-2014-0919) of Severance Hospital and was registered at ClinicalTrials.gov (NCT02581280). Written informed consent was obtained from each patient or the legal representative of unconscious patients prior to enrolment. This study included critically ill patients aged 19 years or older who were admitted CCU for venoarterial -ECMO and were prescribed piperacillin/tazobactam. The exclusion criterion was the use of drugs known to change piperacillin/tazobactam plasma concentrations.
Details of ECMO and CRRT
The ECMO circuit included a centrifugal blood pump with a pump controller (Capiox® SP-101, Terumo Inc., Tokyo, Japan), an air-oxygen mixer (Sechrist® Ind., Anaheim, CA, USA), and conduit tubing (Capiox® EBS Circuit with X coating, Terumo Inc., Tokyo, Japan).
Patients underwent independent CRRT utilizing continuous venovenous haemodiafiltration (CVVHDF, Prismaflex®; Baxter Inc., IL, USA). Prismaflex® ST100 set configurations were used with the polyacrylonitrile AN 69 membrane of 1m2 surface area. All CVVHDF settings were established at the discretion of the treating physician.
Dosing and sampling procedure
The standard empirical dose used at our institution for patients with normal renal function is 4/0.5 g of intravenous piperacillin/tazobactam every 6 h (q6h) or q8h infused over 40 min per dose. However, when creatinine clearance (CrCL) was <40 mL/min or the patient was receiving CRRT, the piperacillin/tazobactam dosage was reduced to 3/0.375 g or 2/0.25 g q6h.
Blood samples were collected from an existing arterial catheter just before piperacillin/tazobactam administration (time 0) and then 0–0.5, 0.5–1, 1–2, 2–4, 4–6, and 6–8 h afterward. The samples were collected between days 2 and 4 of VA ECMO. In the case of concomitant CRRT, the first sample was obtained at least 24 hours after CRRT initiation. Fourteen subjects received piperacillin/tazobactam after ECMO weaning, and blood samples were collected as indicated above between days 2 and 4 of VA ECMO weaning.
plasma concentration assay
Blood samples were added to EDTA tubes and centrifuged at 1,500 × g for 15 min within 1 h of collection and stored at –80 °C until analysis. Validated liquid chromatography–tandem mass spectrometry assays were used to quantify piperacillin and tazobactam concentrations . The linear concentration ranges of piperacillin and tazobactam were 1 to 220 mg/L and 0.5 to 28 mg/L, respectively. The lower limit of quantification was 0.5 mg/L for both piperacillin and tazobactam. The intra-assay accuracies of the quality control samples (1, 3, 30, and 160 mg/L) ranged from 94.9% to 96.5% for piperacillin and from 90.0% to 94.7% for tazobactam.
Population PK analyses
Population PK analyses were conducted using non-linear mixed-effects modeling (NONMEM®; version 7.4; ICON Development Solutions, Ellicott, MD, USA) and the first-order conditional estimation with interaction method. We explored 1-, 2-, and 3-compartment models. Interindividual variability (IIV) was modeled exponentially, and additive, proportional, and combined residual error models were evaluated. Model selections were based on the decrease in objective function value (OFV), goodness-of-fit (GOF) plots, and the relative standard errors of the parameters. When adding another parameter to the model, a decrease in OFV greater than 3.84 points between two nested models was significant at P-values of 0.05.
The relationship between individual PK parameter estimates and the covariates (sex, age, weight, total plasma protein, glomerular filtration rate, blood urea nitrogen, serum creatinine, CrCL, total bilirubin, use of CRRT, CRRT blood flow rate, CRRT dialysate flow rate, CRRT duration, use of ECMO, ECMO pump speed, ECMO flow rate, ECMO duration) were investigated in scatter plots and a biologically plausible explanation for altering PKs were explored. Continuous covariates were centered at the median value of the population. Based on the ꭓ2 test, stepwise covariate modeling with forward inclusion (P <0.05, OFV decrease of 3.84 points) and backward deletion (P <0.01, OFV increase of 6.64 points) was performed. Reduced inter-individual and residual variabilities, precision of the parameter estimates, and diagnostic plots and shrinkage were also considered. The extent of shrinkage, as a measure of model over-parameterization, was calculated for each PK parameter with associated IIV.
Validation of the final model was conducted using the bootstrap method (n = 5,000 runs) and prediction-corrected visual predictive checks (pc–VPCs). The median and 95% confidence intervals for the bootstrap results were compared with the estimated PK parameters from the final model. By using pc-VPCs, 1000 datasets were simulated from the final model parameter estimates, and the 95% CIs for the 5th, 50th, and 95th percentiles based on the simulated datasets were calculated and overlaid with prediction corrected observed concentration for visual inspection .
Monte Carlo simulations
Monte Carlo simulations (n = 1,000) were conducted for segmented CrCL level (20–150 mL/min) in four cases: concomitant use of ECMO and CVVHDF, receiving ECMO only, weaning from ECMO and receiving CVVHDF, weaning from ECMO and not receiving CVVHDF. The dosing regimen was created based on a total daily piperacillin/tazobactam dose of 16/2, 12/1.5, 9/1.125, 8/1, or 6/0.75 g, with an inter-dose interval of 6 or 8 h and duration of infusion of 0.5 h (intermittent bolus), 3 or 4 h (extended infusion equal to half of the dosing interval). Proportion of time between two infusions with piperacillin concentrations above the concentration threshold (ranging from 2 to 64 mg/L) was determined. The PK/PD target for piperacillin was set to 50% fT >MIC of 16 mg/L, which is the clinical breakpoint for P.aeruginosa defined by the European Committee on Antimicrobial Susceptibility Testing , and considered successful if the PTA ≥90% [14, 15]. The PTA for 63% fT >2 mg/L of free tazobactam was also evaluated, and the acceptable PTA level was set at ≥50% [16, 17].