NONMEM versus nlmixr2 : An example of external evaluation of gentamicin Pop-PK models
Alexandre Duong (1,2), Amélie Marsot (1,2,3)
(1) Laboratoire de Suivi Thérapeutique Pharmacologique et Pharmacocinétique, Montréal (Qc), Canada, (2) Faculté de Pharmacie, Université de Montréal, Montréal (Qc), Canada, (3) Centre de Recherche CHU Sainte-Justine, Montréal (Qc), Canada.
Introduction: Population-based pharmacokinetic modeling (popPK) consists of estimating and quantifying the different sources of variability that can influence the behavior of drugs, such as antibiotics, in the body. Furthermore, the models obtained allow to estimate pharmacokinetic (PK) parameters of a patient from concentrations obtained from therapeutic drug monitoring and thus calculate the individualized dose most likely to reach the exposure of interest. PopPK modeling is usually performed using NONMEM or MONOLIX which are the reference programs in the academic and industrial community. In the last few years, several other software packages have been developed to perform the same analyses, notably the nlmixr package, available on the R software. Nlmixr2 is an open-source package and was developed to perform nonlinear mixed-effects modeling [1, 2]. Moreover, nlmixr2 has shown similar performances to NONMEM for stochastic approximation expectation-maximization (SAEM) and first order-conditional estimation with interaction (FOCEI) algorithms with sparse and rich sampling data [3].
Objective: The objective of this project is to compare the results of the same study carried out on NONMEM and nlmixr2. This analysis consists of evaluating the predictive performance of a previously published popPK models of gentamicin in our population of interest.
Methods: Based on a previous study, we performed external evaluations using NONMEM® version 7.5 of four gentamicin models developed for critically ill patients [5]. For the external datasets, gentamicin dosing data, information on the treatment, the patient and the bacteria were collected retrospectively in two Quebec establishments (Hôpital du Sacré-Coeur de Montréal (HSCM) and Institut Universitaire de Cardiologie et Pneumologie de Québec (IUCPQ) from 2009 to 2019 and 2014 to 2020, respectively. Model equations were re-written on R and external evaluations were performed using the second version of nlmixr (nlmixr2) [6]. For both external evaluations, predictive performance was assessed based on the estimation of bias (MDPE) and imprecision (MADPE) with the prediction error (PE%).
Results: Population bias and imprecisions were evaluated for gentamicin popPK models of Rea et al. and Bos et al. [7, 8]. External evaluations for both models of Hodiamont et al. are in progress. Demographic characteristics of the combined datasets (n=87 and 498 observations) were previously described [5]. For Rea et al.’s model, the population bias and imprecision estimated with nlmixr2 were respectively 44.2% and 54.1%. Identical results obtained with NONMEM with population bias and imprecision values of 44.2% and 54.6%, respectively. For Bos et al.’s model, the population bias and imprecisions estimated with nlmixr2 were -44.0% and 47.8%, respectively. Identical population values were obtained with NONMEM (differences of les than 1%).
Conclusion: We believe that this is the first work that compares NONMEM and nlmixr2 performances with therapeutic drug monitoring data collected during a clinical setting. External evaluations performed on both software resulted in the same interpretation in terms of population predictive performance. However, a limitation that should be considered is that inter-occasion variability could not be implemented, as of November 2022.[9] Further studies include the evaluation of nlmixr2’s clinical impact by comparing a priori dosing nomograms simulated by both software.
References:
[1]: Fidler M, Xiong Y, Schoemaker R, Wilkins J, Trame M, Hooijmaijers R, Post T, Wang W (2022). nlmixr: Nonlinear Mixed Effects Models in Population Pharmacokinetics and Pharmacodynamics. R package version 2.0.7, https://CRAN.R-project.org/package=nlmixr.
[2]: Fidler M, Wilkins J, Hooijmaijers R, Post T, Schoemaker R, Trame M, Xiong Y, Wang W (2019). “Nonlinear Mixed-Effects Model Development and Simulation Using nlmixr and Related R Open-Source Packages.” CPT: Pharmacometrics \& Systems Pharmacology, 8(9), 621–633.
[3]: Schoemaker R, Fidler M, Laveille C, Wilkins J, Hooijmaijers R, Post T, Trame M, Xiong Y, Wang W (2019). “Performance of the SAEM and FOCEI Algorithms in the Open-Source, Nonlinear Mixed Effect Modeling Tool nlmixr.” CPT: Pharmacometrics \& Systems Pharmacology, 8(12), 923–930. https://doi.org/10.1002/psp4.12471 .
[4]: Duong A, Simard C, Williamson D, Marsot A. Model Re-Estimation: An Alternative for Poor Predictive Performance during External Evaluations? Example of Gentamicin in Critically Ill Patients. Pharmaceutics. 2022; 14(7):1426. https://doi.org/10.3390/pharmaceutics14071426.
[5]: https://github.com/nlmixr2
[6]: Rea RS, Capitano B, Bies R, Bigos KL, Smith R, Lee H. Suboptimal aminoglycoside dosing in critically ill patients. Ther Drug Monit. 2008;30(6):674-681. doi:10.1097/FTD.0b013e31818b6b2f.
[7]: Bos JC, Prins JM, Mistício MC, et al. Population Pharmacokinetics with Monte Carlo Simulations of Gentamicin in a Population of Severely Ill Adult Patients from Sub-Saharan Africa. Antimicrob Agents Chemother. 2019;63(4):e02328-18. Published 2019 Mar 27. doi:10.1128/AAC.02328-18.
[8]: Schoemaker R. “nlmixr: pharmacometrics modelling for the people” presented at Pharmacokinetics UK (PKUK) on November 2nd 2022. https://www.occams.com/assets/pdf/PKUK2022_221108.pdf .