In silico testing aimed at evaluating the cardiovascular toxicity of medicines during development

Date :
Changed on 18/05/2020
In medical research, both in vitro (in the laboratory) and in vivo (in the organism) testing is employed. Increasingly, however, in silico methods - i.e. digital simulation - are also being used. Within the framework of the EU project INSPIRE**, the Commedia team from Inria Paris is helping specialists in safety pharmacology to better evaluate and predict cardiovascular toxicity in drug candidates. We caught up with Damiano Lombardi, coordinator of the project for Inria, to talk about the project.

What is the goal of this EU project and what is Inria’s role in it?

Damiano Lombardi: The aim of INSPIRE is to improve the evaluation of cardiovascular toxicity in potential medicines in order to be able to eliminate those with side effects judged severe enough to warrant development being stopped or to warrant being withdrawn from the market at the earliest available opportunity. There are ten partners from six different countries, including pharmacological research teams, pharmaceutical laboratories, suppliers of software and sensors and biotechnology companies. Inria is developing in silicomethods, i.e. digital models as a complement to the in vitro and in vivo testing carried out in order to evaluate toxicity. 

Why is it not always possible to detect toxic side effects from the outset?  

D.L.:  Toxicity studies follow strict methodologies, are highly advanced and are able to identify the majority of risks, but not all of them: the heart is an incredibly complex system. Determining how it will react to a treatment both short- and long-term remains a real challenge from a scientific perspective, especially when you’re talking about millions of potential patients. 

What aspects of the heart are you going to be focusing on?

D.L.: Primarily, we will be looking at electrophysiology, i.e. the electrical functioning of heart cells, or cardiomyocytes. Each one is home to a complex system of canals, pumps and exchangers in which ionic species flow, such as sodium, potassium and calcium. We have to model the impact of new drugs at this cell level, progressing up to the auricles and ventricles before finally looking at the heart as a whole.

The second aspect we focus on is haemodynamics, i.e. the dynamics of blood flow. Will the new molecule affect blood pressure as it comes out of the heart? If so, to what extent? Will it make vessels more rigid? Will these issues become worse over time?

At what stage of the development of drugs will your models be used?

D.L.: During the initial screening phase, i.e. during the first evaluation of thousands of candidate molecules. Toxicity is currently quantified using manual methods, but these are limited and extremely slow. 

The aim of INSPIRE is to upgrade to reliable, high-speed tools, enabling molecules that are toxic for the heart to be identified and eliminated at an early stage. This will require finding a way to handle the enormous amount of data produced during the evaluation. It is another Inria team, Delys (see inset), that will be responsible for tackling this side of the project.

Two Inria teams and three PhDs in the pipeline

Two Inria teams are involved in the INSPIRE project. Commedia, which Damiano Lombardi belongs to, is focused on the algorithms used to evaluate both the heart’s electrical system and its haemodynamics (blood flow); two PhDs will be devoted to this. Delys, meanwhile - an Inria Paris team which specialises in distributed systems - will oversee a PhD on managing the massive volumes of data generated by the studies into cardiovascular toxicity. By way of an example, MEAs make 25,000 measurements a second over a period of several hours and can be combined with other methods such as fluorescence imaging. The input flows of information are enormous and come from different devices: it’s a real challenge.   

Is safety pharmacology a new subject for your Inria team?

D.L.: Commedia, the team to which I belong, specialises in modelling and simulating the cardiovascular and respiratory systems. We have been working on safety pharmacology for a number of years. The team devised a method combining in vitro and in silico testing for predicting blockages in ionic canals and certain types of cardiac arrhythmia; it was on the strength of this project that we were called upon for INSPIRE.  

ITNs or Innovative Training Networks are EU programmes which provide 4 years of funding for training networks in Europe. In the context of INSPIRE, they will be funding PhDs. The PhD students taken on will be able to - indeed, will be required to - visit different research bodies in EU countries, in addition to short placements with partners (whether public, industrial or otherwise) of the research structure hosting them.  ITNs promote research as a way of training the scientists of the future, giving them the opportunity to experience mobility. The PhD students who emerge from the Inspire programme will thus benefit from multidisciplinary and cross-sectoral training.

Where will you get the data you need for your models?

D.L.: The data will be produced during in vitro testing. In electrophysiology, microelectrode arrays (MEAs) are used, which were invented in the 1970s to study neurons. Several hundred cardiomyocytes are put on a microchip. A current is injected into this sort of cardiac tissue, which is then measured as an output, as you would with an electrocardiogram. The molecule to be tested is then added and the operation is repeated in order for the signals to be compared.

In a similar vein, for haemodynamics, partners of the INSPIRE project will supply us with data on blood flow and pressure at different points of the vascular tree, with and without the molecule.

Will this information be sufficient to make reliable evaluations?

D.L.: We will add experimental data from previous research into older, established molecules, which will provide us with a basic learning tool. In electrophysiology, for example, research is capable of telling us how a blockage in a potassium channel is translated into MEA signals. In other words, we have solid a priori knowledge on the unwanted side effects to look for, which will help make our predictions more reliable.  

Could you give us some examples of the phenomena you’re looking to identify?

D.L.: In haemodynamics, the overarching ambition is to be able to predict the long-term evolution of how the heart works. A molecule which is non-toxic when first taken can become toxic over time, causing disorders such as hypertension. If we were to find a way of making reliable predictions over a number of weeks or even a number of months, then that would constitute real progress.

In electrophysiology, our aim is to be able to determine whether or not a drug candidate will affect one or more of the three ionic channels, with precise dose-response graphs. We also want to identify potential cumulative effects: the molecule might be a calcium blocker at a low dose, for example, but might also be a potassium blocker at a higher dose.


*Silicon is the most widely used material in computer chips.

**INSPIRE: INnovation in Safety Pharmacology for Integrated cardiovascular safety assessment to REduce adverse events and late stage drug attrition.

This project is funded under H2020-EU.1.3.1.