Could a Virtual Human Replace Animal Models?

Dutch scientists are working toward animal-free safety testing using every modern tool of the trade

This is part of Eureka's ongoing series, 3Rs in Focus, which focuses on innovative projects, partnerships and collaborations that are helping to responsibly reduce, refine and replace research animals. 

According to Juliette Legler, Professor of Toxicology at Utrecht University and coordinator of VHP4Safety, 30 million animals are used globally each year in legally required drug and chemical safety testing. What if many, or even all, of those animals could be replaced with a cascade of assays and machine learning algorithms designed to mimic human biology?

Virtual Human Platform for Safety Assessment

The Dutch-funded Virtual Human Platform for Safety Assessment (VHP4Safety) aims to do just that. A consortium of universities, government agencies, and companies, they are funded mainly by the Dutch Research Agenda: Research on Routes by Consortia (NWA-ORC) as well as through donations, totaling 11.2 million Euros. According to a paper published this March, VHP4Safety has the following mission:

1. An in silico description of human biology and elucidation of adverse outcome pathways (AOPs) in humans, meaning any pathway involved in toxic effects caused by a chemical.
2. Come up with a series of assays in silico and in vitro to measure critical events in those pathways, taking animals out of the equation.
3. A “human-based holistic approach” that could better predict human reactions to chemicals, rather than relying on the imperfect animal model, including human-relevant scenarios such as differences in susceptibility due to age, gender, and disease state.
4. Use machine learning to do what it does best: condense and analyze the data from the “batteries of assays” to accurately predict human health effects.

How Does VHP4Safety Work?

“You can explain by skin sensitization, that’s the easiest one,” said Sylvia Pelgrom, Section Head Regulatory Toxicology at Charles River’s Den Bosch site. “When you have an allergic reaction, you have some key events that need to happen before you have an effect. By covering those key events with in silico and in vitro investigations, you can predict whether the endpoint of an allergic reaction will happen or not.”

In another example submitted to the Dutch Research Council, researchers from Utrecht University want to discover a method for predicting drug-hypersensitivity responses (DHR) in patients. DHR is a dangerous reaction that is not able to be modeled in animals and therefore often goes undetected until the first clinical tests in humans. The researchers claim that three key events (cell stress, cytokine release, and cell migration) often occur during the lead up to an inflammatory response. They wish to determine whether these key events also relate to DHR, and whether a combination of biomarkers and in vitro tests can identify the likelihood that an individual patient will develop DHR. If the research project is funded, it will provide another key testing strategy to add to the VHP.

Sylvia is part of the VHP project from the regulatory angle, communicating with government regulators and the general public about the progress of the project to ensure they can meet the safety concerns of regulators. The hope is for the project to eventually be accepted as a valid alternative in as many countries as possible. This may be the biggest hurdle the project will face: can the platform prove itself to not only Dutch regulators, but also to the EU, US, or other countries? 

“I think it takes many years, but [regulators] are more willing than before, because public opinion is that we have to replace animal studies,” she said. “On the other hand, regulating authorities have to be sure that it is safe, because if a product comes on the market that is not safe, everybody will point to them.”

It is not only public opinion on the side of reducing animal research; there are scientific arguments to make as well. Rats and mice have never been a perfect model for human beings, which has occasionally led to safety concerns still cropping up in clinical trials or after a chemical is released in the environment. On the other hand, the reason animal models are still prevalent is due to the overwhelming complexity of biological systems. Never before have scientists been able to completely accurately predict what could happen in a living body. It is only recently, with the advent of new technology in testing and data analysis software that we have been able to come close.

Walter Westerink is a colleague of Sylvia’s from Den Bosch and part of one of the case studies for thyroid mediated developmental neurotoxicity, including the development of AOP and in silico/in vitro testing. He explained that it is taking an army of postdocs and PhDs to run enough in vitro and in silico tests to feed the algorithm. Even still, he says they know already that it is unlikely every new chemical will be able to be tested this way. Non-soluble chemicals, for example, will be difficult to test in vitro because they might not reach the target cell.

So, how close are we? In the March paper published in Current Opinion in Toxicology, researchers admit that any safety testing regimen “will never be complete in any system in the sense that all information that could be wished for will be available.” Meaning that although this enormous undertaking will attempt to mimic human biological responses, it is possible to miss something. However, the same is true for animal models. Both systems have limitations and a certain level of uncertainty, but not in ways that are easy to compare.

“Ultimately, our goal is to have animal studies replaced by alternatives, but we know this is a long way to go,” said Sylvia. “For now, it is important to find ways to involve regulatory authorities in the new developments for safety testing, because at the end they have to accept the alternative safety approach. By involving regulatory authorities in this process at an early stage and providing them alternative data, we have to demonstrate that alternatives are at least as safe as the current (animal based) data.”

Stay tuned for our next story about two tools that are transforming cancer research and treatment and helping us meet our 3Rs objectives