Estimation of piperacillin clearance with different glomerular filtration rate formulas in critically ill children

External Evaluation of Population Pharmacokinetic Models of Piperacillin in Preterm and Term Patients from Neonatal Intensive Care

BACKGROUND AND OBJECTIVES

Piperacillin/tazobactam is extensively used off-label to treat late-onset neonatal sepsis, but safety and pharmacokinetic data in this population are limited. Additionally, the organic immaturity of the newborns contributes to a high piperacillin pharmacokinetic variability. This affects the clinical efficacy of the antibiotic treatment and increases the probability of developing drug resistance. This study aimed to evaluate the predictive performance of reported piperacillin population pharmacokinetic models for their application in a model-informed precision dosing strategy in preterm and term Mexican neonatal intensive care patients.

METHODS

Published population pharmacokinetic models for piperacillin which included neonates in their study population were identified. From the reference models, structured models, population pharmacokinetic parameters, and interindividual and residual variability data were extracted to be replicated in pharmacokinetic software (NONMEM® version 7.4). For the clinical study, a sampling schedule was designed, and 2-3 blood samples of 250 µL were taken from neonates who met the inclusion criteria. Piperacillin plasma concentrations were determined by liquid chromatography/tandem mass spectrometry. The clinical treatment data were collected, and piperacillin plasma concentrations were estimated using reference pharmacokinetic models for an a priori or Bayesian approach. Statistical methods were used in terms of bias and precision to evaluate the differences between observed and estimated neonatal piperacillin plasma concentrations with the different approaches and to identify the pharmacokinetic model that best fits the neonatal data.

RESULTS

A total of 70 plasma samples were collected from 25 neonatal patients, of which 15 were preterm neonates. The overall median value (range) postnatal age, gestational age, body weight, and serum creatinine at the sampling collecting day were 12 (3-26) days, 34.2 (26-41.1) weeks, 1.78 (0.08-3.90) Kg, 0.47 (0.20-0.90) mg/dL, respectively. Three population pharmacokinetic models for piperacillin in infants up to 2 months were identified, and their predictive performance in neonatal data was evaluated. No pharmacokinetic model was suitable for our population using an a priori approach. The model published by Cohen-Wolkowiez et al. in 2014 with a Bayesian approach showed the best performance of the pharmacokinetic models evaluated in our neonatal data. The procedure requires two blood samples (predose and postdose), and, when applied, it predicted 66.6% of the observations with a relative median absolute predicted error of less than 30%.

CONCLUSIONS

The population pharmacokinetic model developed by Cohen-Wolkowiez et al. in 2014 demonstrated superior performance in predicting the plasma concentration of piperacillin in preterm and term Mexican neonatal intensive care patients. The Bayesian approach, including two different piperacillin plasma concentrations, was clinically acceptable regarding bias and precision. Its application for model-informed precision dosing can be an option to optimize the piperacillin dosage in our population.

Estimation of cefepime, piperacillin, and tazobactam clearance with iohexol-based glomerular filtration rate in paediatric patients

PURPOSE

Estimated glomerular filtration rate (eGFR) equations reflect kidney function imprecisely. We aimed to describe whether iohexol-based GFR or eGFRs predict clearance of cefepime, piperacillin, and tazobactam in pharmacokinetic (PK) models in this population and its clinical significance.

METHODS

Hospitalized patients (0.5-25 years) with haemato-oncological disease and infection receiving cefepime or piperacillin/tazobactam were included. PK samples were collected at a steady state concomitantly with samples for iohexol-based GFR. PK models were developed in NONMEM. Weight, postmenstrual age, iohexol-based GFR, different eGFR equations (Schwartz updated, Lund-Malmö revised, CKD-EPI, Bouvet, Schwartz cystatin C-based) were tested as covariates. Probabilities of neurotoxic/therapeutic concentrations were assessed by simulations.

RESULTS

Fifteen patients receiving cefepime and 17 piperacillin/tazobactam were included (median (range) age 16.2 (1.9-26.0) and 10.5 (0.8-25.6) years, iohexol-based GFR 102 (68-140) and 116 (74-137) mL/min/1.73 m², respectively). Two-compartment model provided the best fit for all drugs. Weight was covariate for central and peripheral compartment, clearance and intercompartmental clearance (only tazobactam), and postmenstrual age for clearance (excluding cefepime). Iohexol-based GFR was the best predictor of clearance. The model of cefepime without vs with iohexol-based GFR underestimated the probability of neurotoxic concentrations (28.3-28.6% vs 52.1-69.3%) and overestimated the probability of therapeutic concentrations (> 90% vs 81.9-87.1%) in the case of iohexol-based GFR 70-80 and 130-140 mL/min/1.73 m², respectively.

CONCLUSION

Iohexol-based GFR can predict better than eGFRs the clearance of cefepime, piperacillin, and tazobactam in children and young adults with haemato-oncological disease and infection, warranting further investigation as an indicator of renal function to improve targeting of therapeutic window.

TRIAL REGISTRATION NUMBER AND DATE OF REGISTRATION

EudraCT 2015-000,631-32, EudraCT 2016-003,374-40 (24.10.2016).

Scoping review of augmented renal clearance in critically ill pediatric patients

Augmented renal clearance (ARC), a phenomenon of enhanced elimination of renal solutes, has been described in adult critically ill patients, but little is known about the phenomenon in children. The aim of this scoping review was to gather and summarize all evidence on ARC in pediatric patients to examine its breadth and depth including prevalence, risk factors, and pharmacokinetic alterations and identify any gaps for further areas of inquiry. PubMed, Embase, and Web of Science were searched for titles, abstracts, or keywords that focused on ARC. Non-English studies, reviews, and nonhuman studies were excluded. Reporting followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses for Scoping Reviews (PRISMA-ScR) guidelines. Data were extracted on article type, study details, patient population, ARC definition and prevalence, methods of renal function assessment, and study results. A total of 215 citations were found with 25 citations meeting the criteria for inclusion in pediatrics (2102 total patients); the majority of studies (84%) focused on pharmacokinetics (PK) of antimicrobial agents. The median/mean age range was 1.25-12 years. There were a total of 10 different definitions of ARC. The prevalence of ARC ranged from 7.8% to 78%. The most common method for documenting creatinine clearance (CrCl) was the modified Schwartz equation (64%). Only 20% of studies reported risk factors for ARC including low serum creatinine, increasing age, febrile neutropenia, male, septic shock, and treatment with antibiotics. Glycopeptide antimicrobials were the most evaluated class (42.9%) among the 21 antimicrobial drug studies. All studies reported increased drug clearance and/or poor probability of achieving target concentrations of the agents studied. ARC showed variable prevalence in pediatric patients likely due to the lack of a standard definition and many studies not considering age-related changes in CrCl with pediatric intensive care unit (PICU) patients. ARC was shown to impact PK of antibiotics commonly administered to pediatric patients, which may necessitate changes in standard dosing regimens.