Gastón GarcÃa-Orueta (1); Natalia Riva (1,2); Iñaki F Trocóniz (1,2,3); Zinnia P Parra-Guillen (1,2)
(1) Department of Pharmaceutical Sciences, School of Pharmacy and Nutrition, University of Navarra, Pamplona, Spain. (2) Navarra Institute for Health Research (IdiSNA), Pamplona, Spain. (3) Institute of Data Science and Artificial Intelligence, DATAI, University of Navarra, Pamplona, Spain.
Introduction:
Over the past decades, there has been an increasing awareness and use of population pharmacokinetic (PK) models as a relevant tool to guide and individualized dosing regimens in clinical practice [1]. This approach becomes especially relevant in critically ill patients, due to the impact of their pathophysiological conditions in the different drug pharmacokinetic processes, and even more if we are dealing with pediatric populations, commonly understudied. The primary constraint in this setting lies in the scarcity of subjects, added to the limited blood sampling tolerance in children compared to adults [2]. In addition, the economic expense as well as the hospital human resources needed for each sample represent a considerable limitation. In pharmacokinetic studies, the quantity of samples, both in terms of number of patients as well as longitudinal measurements, is crucial for constructing robust models. For pediatric patients undergoing CRRT, the Different number of patients, samples per patient or proportion of samples from each matrix (plasma, prefilter, postfilter, effluent) may have serious effect in model developing. In this context, optimal design theory can certainly play a key role in PK studies to maximize the information collected with the effectiveness of the smallest sample size possible.
Objectives:
The aim of this study is to explore the implications of different optimal design scenarios using as an example a previously developed population pharmacokinetic model for teicoplanin in critically ill children undergoing continuous renal replacement therapy (CRRT).
Methods:
Briefly, a two compartment model accounting for renal clearance, as well as clearance via CRRT was previously developed using data from 26 children (13 with and 13 without CRRT) from which plasma (in the case of no CRRT) or prefilter, postfilter, and effluent samples had been taken (ca. 5 sample time points per patient) [3]. The impact of different design elements (number of matrices, number of sampling time points and number and proportion of patients) was explored and optimized. Covariates were not considered in these first analyses. The criterion for assessing and refining the designs was the evaluation of the population Fisher Information Matrix (FIM) with block-diagonal modality [4] and attained parameter precision, measured as relative standard error in percentage (RSE). The software utilized for this study was NONMEM 7.5 [5], incorporating the $DESIGN option [6].
Results:
First, the original study design above described was evaluated, confirming the identifiability of the model with RSEs below 25 % and 40 % for typical and inter-individual variability parameters, respectively. Subsequently, different combinations of available matrices from CRRT patients were tested (prefilter, postfilter and effluent; prefilter and postfilter/effluent; prefilter only). Omitting postfilter samples had only a minor impact on overall performance, as the precision of the most affected parameter (intercompartmental clearance, Q) went from 23.9% to 24.7%. In contrast, a significantly worsening in precision (D >20 %) was however observed when only prefilter or prefilter and postfilter measurement were collected, affecting to parameters from the peripheral compartment (Q and total volume, V2).
These results suggest that effluent samples are more meaningful than postfilter ones in this context. Similarly, designs with different number of samples per patient were explored, revealing that the expected RSEs in Q and V2 were also the most affected when decreasing the number of sampling time points, and going from 23.9 and 16.5% up to 67.2 and 43.1%, respectively, in the design with three samples per patient. Precision in the rest of the parameters was also worsen, but still below 25 %. Optimizing sampling times resulted in a substantial improvement of the precision in parameters and in the FIM evaluation. As anticipated, precision of variability parameters was only slightly affected by changes in the number of matrices considered or sampling times, and mainly dependent on the number of individuals included in the study.
Conclusions:
A study design optimization showed that it is possible to develop a population PK model of teicoplanin in pediatric patients undergoing CRRT with less samples without compromising parameter precision but decreasing study burden as well as inconvenience for blood sampling in children. This work can be used as a reference to guide new studies in this special population.
References:
[1] Manolis, E., Osman, T. E., Herold, R., Koenig, F., Tomasi, P., Vamvakas, S., & Saint Raymond, A. (2011). Role of modeling and simulation in pediatric investigation plans. Paediatric anaesthesia, 21(3), 214–221.
[2] Roberts, J. K., Stockmann, C., Balch, A., Yu, T., Ward, R. M., Spigarelli, M. G., & Sherwin, C. M. (2015). Optimal design in pediatric pharmacokinetic and pharmacodynamic clinical studies. Paediatric anaesthesia, 25(3), 222–230.
[3] García-Orueta et al. PAGE 32 (2024) Submitted Abstract.
[4] Retout, S., Duffull, S., & Mentré, F. (2001). Development and implementation of the population Fisher information matrix for the evaluation of population pharmacokinetic designs. Computer methods and programs in biomedicine, 65(2), 141–151.
[5] Beal SL, Sheiner LB, Boeckmann AJ, Bauer RJ, eds. NONMEM 7.5 Users Guides. ICON plc; 1989–2020. https://nonmem.iconplc.com/nonmem750.
[6] Bauer, R. J., Hooker, A. C., & Mentré, F. (2021). Tutorial for $DESIGN in NONMEM: Clinical trial evaluation and optimization. CPT: pharmacometrics & systems pharmacology, 10(12), 1452–1465.
Reference: PAGE 32 (2024) Abstr 10974 [www.page-meeting.org/?abstract=10974]
Poster: Drug/Disease Modelling - Paediatrics