Virtual heart for drug screening

Research details

Principal Investigator: Professor Blanca Rodriguez

Organisation: University of Oxford

Award type: Infrastructure for Impact grant

Start date: 2016

Duration: 5 years

Amount: £512k

Case study

Drug-induced cardiotoxicity is one of the leading causes for attrition during pharmaceutical development and can also result in drugs being withdrawn after market approval. International guidelines (e.g. ICH S7B) require compounds to be evaluated for their effects on ventricular repolarisation and proarrhythmic risk using in vitro and in vivo tests. Assessment is carried out during the early stages of drug development in in vitro assays using heart tissue and cells, typically from guinea pigs, rabbits, dogs or non-human primates. The primary function of these assays is to identify changes to electrophysiological properties (e.g. action potential duration and/or cardiac ionic currents). Later in development, cardiovascular safety including electrophysiology and mechanical read-outs of cardiac function (e.g. ECG, blood pressure and heart rate) are assessed as part of a core battery of regulatory pharmacology tests usually performed in dogs or non-human primates. The number of animals used per drug at this stage is four or eight animals depending on whether the study is conducted in Europe or the USA respectively. In some cases, it may be necessary to conduct further animal studies to understand potency and mechanisms of action.

It is estimated that 20% to 50% of drug candidates are abandoned due to cardiotoxicity despite in vitro and in vivo testing. Worldwide there are a number of efforts to provide more predictive tools that exploit the potential of human-induced pluripotent stem cell-derived cardiomyocytes and in silico models in order to avoid the current reliance on animal models.

3Rs benefits (actual and potential)

The Computational Cardiovascular team at Oxford initially generated the Virtual Assay software, with EPSRC funding, to simulate the effect of drugs on the electrophysiology, and calcium dynamics linked with contractility, of populations of human cardiomyocytes. The software has been enhanced by the availability of a comprehensive database of human in silico models, funded by the NC3Rs, for specific disease conditions such as heart failure, myocardial ischaemia, genetic disorders and cardiomyopathies.

With NC3Rs funding, the team have gone on to compare predictions using simulations from Virtual Assay with data from clinical trials, animal models, and in vitro methods, helping to build confidence in in silico approaches. A publication in Frontiers in Physiology in 2017 describes how Virtual Assay simulations predict clinical risk of drug-induced Torsade de Pointes with higher accuracy than animal experiments for more than 60 reference compounds. The simulated repolarisation abnormalities were shown to be a more accurate, specific and sensitive biomarker for arrhythmia risk than action potential duration measured in animals – with the Virtual Assay simulations having an 89% accuracy compared to 75% in the rabbit isolated heart model. The paper has been viewed more than 10,000 times.

Virtual Assay is being evaluated by four major pharmaceutical companies for use in early drug development to assess arrhythmic risk, with initial estimates of a 30% to 33% reduction in animal use per year across the sector.  The software is also freely available with an academic licence and is currently being used by six research groups.

Scientific and technological benefits

Typically, computer models of heart function are based on one generic model representative of a generic cell and therefore may not be representative of the wider population where cardiac response to drugs may differ. An advantage of the Virtual Assay is that rather than this ‘one model fits all’ approach, the software generates a large population of simulated human cardiac cells with unique profiles, allowing better modelling of disease state, drug responses and genetic variation. The population featured in the Frontiers in Physiology publication consisted of 1,213 unique cardiac cellular simulations, which were then used to run an in silico drug trial with reference compounds.

To improve Virtual Assay’s physiological relevance the team are also adapting it to account for other cardiac comorbidities. For example, cardiotoxic adverse events are more likely to occur in patients where electromechanical function of the heart is impaired. Such comorbidities are, however, rarely reproduced in in vivo studies which use healthy animals. A computer simulation for in silico drug trials in single cell electromechanical models is currently being validated, with plans to extend the model to simulate multiple cells and eventually the whole heart.

Added value

In 2017, Dr Elisa Passini, a senior postdoctoral scientist in the team, was awarded the inaugural Safety Pharmacology Society Technological Innovation Award as well as the NC3Rs International 3Rs Prize which is sponsored by GSK. Blanca was awarded the MPLS Impact Award for Commercial Impact in 2018 by the University of Oxford. Blanca and the team have given presentations on Virtual Assay at a range of national and international scientific meetings. In 2018, with funding from the NC3Rs, they hosted a symposium titled ‘In Silico Drug Safety and Efficacy’ in Oxford with 100 delegates from academia, industry and regulatory authorities. There have also been extensive outreach activities which include Elisa presenting at the Hay Festival and organising an event through The Royal Institution, where school pupils used the Virtual Assay in a masterclass in computational biology. An article published in The Conversation, ‘Why computer simulations should replace animal testing for heart drugs’, has been viewed more than 20,000 times.

This case study was published in our 2019 Research Review.