Image courtesy of Pixabay

The pandemic has forced a rethink of clinical research, but the pharma industry continues to rely on animal testing. While pundits have observed that computer modeling and techniques such as microdosing can reduce animal testing, animal testing continues to be integral in preclinical studies. 

But computer models are now sufficiently accurate to predict the response of many drugs, said David Harel, CEO of CytoReason (Tel Aviv, Israel). “We are getting to the point that computer models of certain diseases can generate better predictions than animal models,” he said. 

But there are caveats. It could take longer to move from animal-based safety testing, which often involve rodents. Such animal trials tend to be limited in size. “They’re not a big burden. And they’re not super expensive,” Harel said. And regulators frequently consider animal data when evaluating drug safety. But animal studies don’t always correlate to human toxicity, leading some drug developers to question their value.  

In any case, the path for computer models to replace animal-based efficacy studies is clearer, Harel said. Computing techniques such as machine learning can translate animal data to human data and integrate data from multiple human clinical trials. 

Ultimately, computer models are not just a more humane alternative to animal studies; they are potentially “faster, cheaper, and more accurate,” Harel said. 

One of the reasons Cytoreason is initially focused on modeling the immune system is the considerable difference between animals’ immune response and that of humans. “There are certain diseases that you can’t model in an animal, and the quality of the predictions you generate in an animal is not as good as the predictions you can generate on computer models relying on human data,” Harel said. 

One traditional hurdle to running computer models for human disease is the difficulty of obtaining non-animal data. “How do you get data from humans before you started the human trials? That’s a Catch-22,” Harel said. 

But the field of regulatory affairs has simplified processes for sharing human data. Traditionally, individual companies tended to have individual policies, making such data aggregation complex. 

The healthcare industry’s embrace of cloud computing is also helping. “The reality is that the healthcare/life science industry is just getting introduced to the capabilities of the cloud,” Harel said. “A lot more data is being collected and aggregated these days than eight years ago.”

In the pandemic, regulators are also growing more flexible. FDA, for instance, has scrutinized computer simulations of acute respiratory distress syndrome (ARDS) resulting from COVID-19 infection, which has a high mortality rate. “Based on our data, FDA approved an efficacy trial with a major pharmaceutical company without any animal models,” Harel said. 

The pandemic could ultimately spur further use of computer models in drug development, Harel said. “As we’ve learned for history, things that are done in emergency times — if they work — will be accepted overall.”