Yu Fu (1), N. Snelder (2), H. Taghvafard (1), P.H. van der Graaf (1,3), J.G.C. van Hasselt (1)
(1) System Biomedicine and Pharmacology, LACDR, Leiden University, the Netherlands (2) LAP&P Consultants BV, Leiden, the Netherlands (3) Certara QSP, Canterbury, UK
Objectives:
The cardiovascular hemodynamic system is complex and highly regulated. Drugs can affect hemodynamic function through different modes of action. Early prediction and quantification of cardiovascular drug effects during preclinical drug development is crucially important to support dose selection and decision making during drug development. Previously a minimal mathematical systems model to quantify drug effects on key hemodynamic variables was developed using rat experimental data for eight compounds by Snelder et al (“Snelder model”) [1,2]. The model characterizes drug effects on the interrelationship between five hemodynamic biomarkers including heart rate, stroke volume, total peripheral resistance, cardiac output and mean arterial pressure. The current study aims to further evaluate if the mode of action of new drugs can be identified while keeping the system-specific parameters fixed, and identify the minimal required data to translate the model to other species by evaluating if the system-specific parameters can be identified using data from only one compound with known mechanism of action (MoA).
Methods:
We performed a structural identifiability analysis using the Matlab toolbox GenSSI 2.0 with different combinations of observations, including heart rate, cardiac output and mean atrial pressure [3]. Practical identifiability was evaluated using stochastic simulation and estimation (SSE) using Perl-speaks-NONMEM to determine if the model can be used to identify and quantify drug mode of action. We used 100 samples to perform each SSE analysis. Datasets consisted of 5 animals with densely sampled hemodynamic biomarkers, reflecting a typical experimental design. We simulated and re-estimated several study designs evaluating drug effects with different magnitudes for EC50 in association with three potential sites of action (heart rate, stroke volume and total peripheral pressure). The simulated study designs utilized informatively sampled observations. Using the SSE analysis we quantified the number of models that correctly identified drug mode of action, and we evaluated bias and precision of drug effect parameters.
Results:
The cardiovascular-hemodynamic model proposed by Snelder is structurally locally identifiable based on observations of heart rate, cardiac output and mean arterial pressure. The models is also structurally and globally identifiable based on observations for heart rate and mean arterial pressure. The SSE analyses indicated that a true drug effect could be identified for different EC50 and Emax based on successful minimization and statistical significance of p<0.05. The power to identify a significant drug effect was 100% for all values of EC50 and Emax. Both system-specific parameters and drug-specific parameters can be estimated precisely with a minimal bias. Although our structural identifiability analysis indicated that the model is structurally identifiable based on heart rate and mean arterial pressure, we found that models including observations for heart rate, cardiac output and mean atrial pressure showed an increased percentages of successful minimization compared to models that only include heart rate and mean atrial pressure.
Conclusions:
Our analysis supports the use of the Snelder model to identify and quantify drug mode of action in preclinical cardiovascular experiments. MAP, HR and CO measurements following administration of one compound with known MoA provide a good starting point for translating this model to other species and guide study design.
References:
[1] Snelder, N. et al. PKPD modelling of the interrelationship between mean arterial BP, cardiac output and total peripheral resistance in conscious rats. Br J Pharmacol 169, 1510-1524, doi:10.1111/bph.12190 (2013).
[2] Snelder, N. 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 171, 5076-5092, doi:10.1111/bph.12824 (2014).
[3] Ligon, T. S., et al. GenSSI 2.0: multi-experiment structural identifiability analysis of SBML models. Bioinformatics 34(8): 1421-1423, doi: 10.1093/bioinformatics/btx735 (2018).
Reference: PAGE 28 (2019) Abstr 8965 [www.page-meeting.org/?abstract=8965]
Poster: Drug/Disease Modelling - Other Topics