Ibrahim El-Haffaf (1), Romain Guilhaumou (2), Lionel Velly (3), Amélie Marsot (1)
(1) Faculty of Pharmacy, Université de Montréal, Montreal, QC, Canada, (2) Service de Pharmacologie Clinique et Pharmacovigilance, Assistance Publique des Hôpitaux de Marseille, (3) Service d’anesthésie-réanimation, Hôpital de la Timone, Assistance Publique des Hôpitaux de Marseille et Institut de neurosciences de la Timone, CNRS, Aix Marseille Université
Introduction: Piperacillin-tazobactam is an extended-spectrum β-lactam/β-lactamase inhibitor antibiotic frequently prescribed in intensive care units. In clinical practice, the unbound concentration is usually determined in order to assess the efficacy of this antibiotic [1]. This can either be done by directly measuring the unbound concentration through analytical methods or by applying a factor to the measured total concentration. In many analyses, an unbound fraction of 0.70 is assumed for piperacillin to calculate the unbound concentration from the total concentration [2]. However, in critically ill patients, hypoalbuminemia is a frequent phenomenon that impacts the binding degree of drugs to proteins [3]. This suggests that a higher unbound fraction for piperacillin may be observed in critically ill patients, indicating that the application of a factor of 0.70 could be inappropriate. We sought to showcase different hypothetical scenarios of how a variation in the unbound fraction can impact the external evaluation of a population pharmacokinetic model.
Objectives: Evaluate the impact of different central unbound fraction values on the predictive performance of an externally validated piperacillin-tazobactam population pharmacokinetic model.
Methods: A dataset of 1000 patients receiving piperacillin-tazobactam by continuous infusion was simulated. The dosing regimens simulated were based on the creatinine clearance, ranging from 8/1 to 16/2 g per day. The simulation generated concentration-time profiles per dosing regimen. After total concentrations were obtained, free concentrations were calculated according to a central unbound fraction value of 0.70, with a variation of ± 0.02, following a normal distribution. Prediction error was calculated to determine model bias and imprecision. These steps were then repeated for central unbound fraction values of 0.75, 0.80 and 0.85. Bias values of ± 20% and imprecision values of less than 30% were considered acceptable. A previously externally validated model was used for this evaluation. This evaluation was performed using NONMEM version 7.5.
Results: A total of 4000 virtual patients were simulated in this dataset. The model used for evaluation was the model by Klastrup et al. [4]. There was a negative trend found between the unbound fraction and bias. For a central unbound fraction of 0.70, the bias was 23.1% and the imprecision was 32.3%, whereas the bias and imprecision for a central unbound fraction of 0.85 were 1.3% and 26.6%, respectively.
Conclusions: This simulation showed that applying a central unbound fraction of 0.85 to total concentrations in the dataset resulted in the favorable validation of the model by Klastrup et al., compared to using a value of 0.70, indicating a potential impact of the unbound fraction on the predictive performance of a population pharmacokinetic model for piperacillin-tazobactam. Since this analysis was based solely on simulated data and with assumptive unbound fraction values, it would be ideal to verify these findings with real patient data from a large cohort to determine whether the unbound fraction truly impacts the external evaluation process of a model.
References:
[1] Craig WA. Pharmacokinetic/pharmacodynamic parameters: rationale for antibacterial dosing of mice and men. Clin Infect Dis. 1998;26(1):1–10
[2] Pfizer.2012. Piperacillin sodium and tazobactam sodium (Zosyn) prod-uct information. Pfizer, New York, NY.
[3] Ulldemolins M, Roberts JA, Rello J, Paterson DL, Lipman J. The effects of hypoalbuminaemia on optimizing antibacterial dosing in critically ill patients. Clin Pharmacokinet. 2011 Feb;50(2):99-110.
[4] Klastrup V, Thorsted A, Storgaard M, Christensen S, Friberg LE, Öbrink-Hansen K. Population pharmacokinetics of piperacillin following continuous infusion in critically Ill patients and impact of renal function on target attainment. Antimicrob Agents Chemother. 2020;64(7):e02556-e2619.
Reference: PAGE 29 (2021) Abstr 9770 [www.page-meeting.org/?abstract=9770]
Poster: Methodology - Model Evaluation