Ground-breaking software for modelling, simulation, optimization, estimation and validation applications in drug development timeline processes.
Sánchez-Herrero, Sergio (1); Serna, Jenifer (1); Diego García-Álvarez (2); Rueda-Ferreiro, Almudena (1).
(1) Simulation Department, Empresarios Agrupados Internacional S.A., Madrid, Spain. (2) Computer Department, University of Valladolid, Valladolid, Spain.
Introduction:
With the rapidly changing health care and research environments, it has become essential that the drug development process achieve greater safety, efficiency, cost effectiveness and timeliness drug testing for humans. Modelling & Simulation (M&S) is one of the best innovative technologies to achieve them [1]. A growing number of regulatory submissions include mathematical models that simulates the concentration of a drug over time in tissue(s) and blood, by taking into account pharmacokinetics mechanisms (Liberation, Absorption, Distribution, Metabolism and Elimination (LADMET) and pharmacology processes (flows, heart rate, volumes or blood perfusion) to describe the vascular system for individual, populations or subpopulations [2]. To complete an M&S drug development timeline the use of specialized software tool has become essential to manage Non-Compartmental analysis, Compartmental PK/PD models and physiologically based pharmacokinetic (PBPK) estimation and simulation models.
Objectives:
- Combination of technical analysis (NCA, QSP, PK/PD/PBPK) for every phase in drug development processes in a unique flexible simulation user-friendly tool.
- Reusable and multiscale flexible models with complete mathematical description.
- Export and connect with external tools like excel or python.
Methods:
PhysPK software (version 2.4) is a M&S platform based on the EcosimPro Language (EL), which is an multilevel object-oriented a-causal simulation language. PhysPK consists in a set of libraries for NCA analysis and M&S complex pharmacokinetics/pharmacodynamics and physiological systems based on top-down programming capacity for continuous-discrete systems. The L-ADME and physiological mechanisms are modelled inside PhysPK through differential-algebraic equations (DAE) and discrete events [4]. In addition, any other mechanism based on DAE could be implemented. As consequence, NCA, PK/PD and PBPK approach could be developed from cell to the whole body ensuring mass conservation law of each chemical compound inside all the spatial regions [5, 6, 7].
PhysPK analysis and models could characterize the observed data to develop models with acceptable degree of precision [8]. Thereby, first Order Conditional Estimation methods FOCE-I or Bayesian estimation methods are used [9] with intra-subject and inter-subject variability and covariance matrix. On the other hand, PhysPK apply validation methods to performance and follow a fit-for purpose approach like Montecarlo, goodness of fit (GOF), Virtual prediction check, Virtual population or Bootstrap [10].
To complete this M&S environment models can be exported to re-use models for different compounds (Bioequivalence analysis) and connect with external tools like excel or python.
Results:
Analysis were developed for NCA bolus single dose, infusion single dose, extravascular first dose and steady state and single oral dose. Analysis were performed in comparison with Winnonlin, gPKPDSim and IQnca R package [11, 12]. Spaghetti plots and NCA metrics were obtained the same value for PK parameter related to Lambda_Z and error ratio <0.5% for predicted PK parameter.
On the other hand, a two-compartment extra-basal model fit, and multi-dose projection for PK of a large molecule (Monoclonal Antibody) with error ratio <0.5% were replicated [11]. In addition, PK profile of a 10 mg/kg 4qw IV dosing regimen for three simulations were performed, CL1=6.89 and CL2=0.5*CL1. Plots show similar results in simulations.
Predictive engines based on pharmacokinetics modelling for tacrolimus personalized dosage in paediatric renal transplant patients were published by PhysPK [13]. PhysPK developed two new population pharmacokinetic models (PK and PBPK approach). The PBPK model showed 95% CIs for the predicted percentiles for TAC concentrations, as well as AUC24 dispersions for normalised TAC doses 4-fold lower than in the PK model (Monte Carlo studies) for real patient data.
PhysPK models could be handled with Excel and Python [14].
Conclusions:
NCA analysis and PK/PD/PBPK/QSP modelling, simulation, optimization, estimation and validation could be performed by PhysPK. This unique approach could achieve greater efficiency, cost effectiveness and timeliness in drug development process. In addition, export models and connect with python will allow us to engage with new cutting-edge technologies like artificial intelligence algorithms.
References:
[1] Holford, Nicholas HG, et al. "Simulation in drug development: good practices." Draft Publication of the Center for Drug Development Science (CDDS). Draft version 1 (1999): 23.
[2] Morrison, Tina M., et al. "Advancing regulatory science with computational modeling for medical devices at the FDA's office of science and engineering laboratories." Frontiers in medicine 5 (2018): 241.
[3] Garrett, A., O'Kelly, M., Walp, D., & Berry, N. S. (2015). Lifecycle modeling and simulation in clinical trials. Applied Clinical Trials, 24(6/7), 30.
[4] Reig-Lopez, Javier, et al. "A multilevel object-oriented modelling methodology for physiologically-based pharmacokinetics (pbpk): Evaluation with a semi-mechanistic pharmacokinetic model." Computer Methods and Programs in Biomedicine 189 (2020): 105322.
[5] Thelen, Kirstin, et al. "Evolution of a detailed physiological model to simulate the gastrointestinal transit and absorption process in humans, part 1: oral solutions." Journal of pharmaceutical sciences 100.12 (2011): 5324-5345.
[6] Thelen, Kirstin, et al. "Evolution of a detailed physiological model to simulate the gastrointestinal transit and absorption process in humans, part II: extension to describe performance of solid dosage forms." Journal of pharmaceutical sciences 101.3 (2012): 1267-1280.
[7] Prado-Velasco, Manuel. "III-58: Manuel Prado-Velasco Bridging the gap between open and specialized modelling tools in PBPK/PK/PD with PhysPK/EcosimPro modelling system: PBPK model of methotrexate and 6-mercaptopurine in humans with focus in reusability and multilevel modelling features."
[8] Robinson, S. (1997, December). Simulation model verification and validation: increasing the users' confidence. In Proceedings of the 29th conference on Winter simulation (pp. 53-59).
[9] Dartois, Céline, et al. "Evaluation of uncertainty parameters estimated by different population PK software and methods." Journal of pharmacokinetics and pharmacodynamics 34.3 (2007): 289-311.
[10] Brendel, Karl, et al. "Are population pharmacokinetic and/or pharmacodynamic models adequately evaluated?." Clinical pharmacokinetics 46.3 (2007): 221-234.
[11] Hosseini, Iraj, et al. "gPKPDSim: a SimBiology®-based GUI application for PKPD modeling in drug development." Journal of pharmacokinetics and pharmacodynamics 45.2 (2018): 259-275.
[12] The IQnca R package. https://iqnca.intiquan.com/
[13] Prado-Velasco, Manuel, Alberto Borobia, and Antonio Carcas-Sansuan. "Predictive engines based on pharmacokinetics modelling for tacrolimus personalized dosage in paediatric renal transplant patients." Scientific reports 10.1 (2020): 1-18.
[14] Evans, Marina V., et al. "A physiologically based pharmacokinetic model for intravenous and ingested dimethylarsinic acid in mice." Toxicological sciences 104.2 (2008): 250-260.
Video link:
https://drive.google.com/file/d/1Mh-nrS8sY7unGyWmgtALcGqLYxvWeHkD/view?usp=sharing
