Giovanni Di Veroli
AstraZeneca
Objectives: Pharmaceutical companies routinely screen compounds in vitro and in vivo to understand hemodynamics related safety risks. Our current model of screening is rather linear: in vitro studies are initially used to prioritize compounds at early stage based on secondary pharmacology. Later in development, in vivo studies allow observing changes in pre-clinical species (rodents and large animal species) and translation to humans is mostly based on these pre-clinical findings. However, a recent review by Pfizer highlighted strong limitations in this approach [1]. A significant challenge in the prediction of translation is the incorporation of molecular findings in the animal-based toxicological risk assessment. We hypothesized that a mathematical model which includes major pathways of hemodynamics regulation and can capture pharmacology known to impact on cardiovascular function could be developed in order to bridge secondary pharmacology and telemetry studies.
Methods: An extensive literature review was carried to review all the involved physiological processes and the role of a pre-defined list of 50 pharmacological sites in affecting these processes. We then designed and built two models which share a common structure (13 model variables and 24 interactions) but different parametrizations for rat and dog species. The models were built in a bottom-up fashion via parametrization of isolated interactions or groups of 2 or 3 interactions by leveraging on appropriate in vitro, in vivo and ex-vivo experiments. The models were designed to recapitulate known hemodynamics regulation processes at levels of granularity consistent with pre-clinical screens readouts and the need of incorporating secondary pharmacology sites of action.
Results: Following model design, interactions parametrization and assembly, we verified that the model integrated behaviour would correctly predict a number of settings not necessarily used in its construction. Simulations were able to predict daily changes in blood pressure, heart rate, stroke volume and total peripheral resistance in rat and dog species within standard deviations (or 25% when standard deviations were not available). Further simulations were carried where endothelin or dopamine were injected, or where baroreceptor, sympathetic or parasympathetic nerves were stimulated. Simulations of blood pressure and heart rate showed that the model was also able to correctly recapitulate available quantitative (again within standard deviations or 25%) or qualitative published results for these additional experiments in these two pre-clinical species. We subsequently tested the model in its capacity to predict correct directionality of changes in blood pressure or heart rate with a pre-defined list of 50 pharmacological sites related to hemodynamics regulation, showing that changes predicted here also agreed with the reported outcomes.
Conclusions: We believe these virtual dog and rat hemodynamics regulation models could constitute important milestones in hemodynamics safety assessment via a Quantitative Systems approach (QSP or QST). While they were designed to investigate safety risk and assess translation, they could also potentially be used to explore new therapies. Several additional steps could be taken to further develop these pre-clinical species models or to develop a human version.
References:
[1] BHATT, S., NORTHCOTT, C., WISIALOWSKI, T., LI, D. & STEIDL-NICHOLS, J. 2019. Preclinical to clinical translation of hemodynamic effects in cardiovascular safety pharmacology studies. Toxicological Sciences, 169, 272-279.
Reference: PAGE 32 (2024) Abstr 10782 [www.page-meeting.org/?abstract=10782]
Poster: Methodology - Other topics