Igor Vasyutin (1), Victoria Kulesh (1), Alina Sofronova (1), Holly Kimko (2), Conchi Villar Moas (3), Matthew Bridgland-Taylor (3), Mike Rolf (4), Amy Pointon (5), Kirill Peskov (1, 6), Viсtor Sokolov (1), Ravindra Alluri (7)
(1) M&S Decisions LLC, Moscow, Russia (2) CPQP, CPSS, BioPharmaceuticals R&D, AstraZeneca, Gaithersburg, USA (3) Regulatory Toxicology and Safety Pharmacology, CPSS, BioPharmaceuticals R&D, AstraZeneca, Cambridge, UK (4) Regulatory Toxicology and Safety Pharmacology, CPSS, BioPharmaceuticals R&D, AstraZeneca, Gothenburg, Sweden (5) Safety Innovation, CPSS, BioPharmaceuticals R&D, AstraZeneca, Cambridge, UK (6) Sechenov University, Moscow, Russia (7) CPQP, CPSS, BioPharmaceuticals R&D, AstraZeneca, Cambridge, UK
Objectives: Predicting potential drug-induced unintended effects on the cardiovascular (CV) system remains a critical component of drug discovery process. To enable detection of these effects prior to clinical development a range of approaches are applied including preclinical telemetry studies followed by model-based analyses. Most often they include quantification of the link between plasma pharmacokinetic (PK) and telemetry signal via empirical pharmacokinetic-pharmacodynamic (PKPD) modeling [1]. However, such approach, among other limitations, ignores the intricate crosstalk between different CV effects [2]. Snelder et al. has developed a semi-mechanistic integrative model of the CV system which distinguishes between primary (i.e. drug) and secondary (i.e. physiological responses) effects of a drug on hemodynamic parameters [3]. The aim of the current work is to validate the semi-mechanistic model using telemetry data for compounds with known mechanism of action, previously unused in the model development.
Methods: The semi-mechanistic model includes 16 parameters from three turnover equations with homeostatic feedback from mean arterial pressure (MAP) for three primary hemodynamic measurements: heart rate (HR), stroke volume (SV) and total peripheral resistance (TPR). In addition, the model includes functional relationship between HR and SV, circadian variations, and handling effects.
Both internal and external data were used in the analysis. Internal data included data from a study in 8 Wistar-Kyoto rats treated with vehicle and one or two doses of glycopyrrolate, clonidine, nifedipine and atenolol followed by a week of washout period between compounds. HR and MAP were collected every 1 minute and averaged in 15 minutes bins throughout a period of 0.5 hours prior to and 22 hours after treatment.
A systematic review was performed to identify published rat PK models or data for the compounds under investigation. Circadian variations were modeled with cosine function while handling effects were modeled with exponential function with negative argument. Several structural and statistical models were tested to describe the effect of a drug. The optimal model was selected based on minimal AIC, goodness of fit plots and estimated parameter precision.
Monolix (version 2020R1) and R software (version 4.0.2) were used for data visualization, parameter estimation, post-processing of the model outputs and simulations.
Results: Sequential modeling approach was taken: (1) PK data from internal sources and nine external studies were used to estimate PK parameters per compound, except for atenolol, for which a published model was used [4], (2) placebo models were developed to describe typical changes in HR and MAP following vehicle treatment, and (3) the primary effects of treatment were evaluated by combining PK and placebo model parameters into a single framework, followed by a stepwise evaluation of the drug effect with various types of models at each hemodynamic variable individually or in combinations.
In total, 816 models were tested for four compounds. Optimal models show positive linear effect on HR for glycopyrrolate with a slope of 2.33 uM-1 (95% CI: 1.92 – 2.73), Imax-type effect on TPR for nifedipine and on HR for atenolol with maximum effect of -42.6% (95% CI: -45.2% – -40.0%) and -22.3% (95% CI: -24.8% – -19.8), respectively, and negative linear effect on HR and SV and positive linear effect on TPR for clonidine with the slopes of -6.43 uM-1 (95% CI: -7.60 – -5.25), -19.61 uM-1 (95% CI: -25.13 – -14.09) and 12.70 uM-1 (95% CI: 10.30 – 15.11) %), respectively.
Predicted primary effects for all compounds are in agreement with the expected effects, except for clonidine: as an alpha-2 adrenoreceptor agonist, clonidine is expected to cause decrease in HR and TPR leading to decrease in MAP. However, in our study in conscious rats, clonidine increased MAP which cannot be described by a decline in HR and TPR.
Conclusions: We have successfully identified the primary mechanisms driving CV effects for four compounds (glycopyrrolate, clonidine, nifedipine, atenolol) based on rat HR and MAP data using the semi-mechanistic modeling approach originally proposed by Snelder et al. The alignment of the primary drug effects predicted by the model with known empirical targets confirms the validity of the approach and sets a basis for translation of the model to other species.
References:
[1] V. Sokolov, A. Sofronova, L. Stolbov, et al. Model-based analyses of preclinical telemetry data in support of decision-making in early drug development: an exemplar workflow.
[2] Collins TA, Bergenholm L, Abdulla T, et al. Modeling and Simulation Approaches for Cardiovascular Function and Their Role in Safety Assessment. CPT Pharmacometrics Syst Pharmacol. 2015 Mar;4(3):e00018. doi: 10.1002/psp4.18.
[3] Snelder N, Ploeger BA, Luttringer O, et al. Drug effects on the CVS in conscious rats: separating cardiac output into heart rate and stroke volume using PKPD modelling. Br J Pharmacol. 2014 Nov;171(22):5076-92. doi: 10.1111/bph.12824.
[4] van Steeg TJ, Freijer J, Danhof M, de Lange EC. Pharmacokinetic-pharmacodynamic modelling of S(-)-atenolol in rats: reduction of isoprenaline-induced tachycardia as a continuous pharmacodynamic endpoint. Br J Pharmacol. 2007 Jun;151(3):356-66. doi: 10.1038/sj.bjp.0707234.
Reference: PAGE 30 (2022) Abstr 10105 [www.page-meeting.org/?abstract=10105]
Poster: Drug/Disease Modelling - Safety