The end of animal testing? Experts discuss whether AI and emerging tech might eliminate suffering

07 Jun 2023 --- Artificial Intelligence (AI), big data and organ-on-a-chip technologies might reduce or even put an end to animal testing for scientific experiments, including nutrition studies, experts forecast. Furthermore, the Humane Society International (HSI) states that many animal science experiments fail in human trials.

However, there is still a sense that animal testing keeps us safe “despite data to the contrary,” and the use of animal subjects is still standard procedure, though more ethical, humane and accurate approaches are emerging.

NutritionInsight speaks with several leading thinkers and experts about the controversies surrounding animal experimentation.

Peter Singer, the author of Animal Liberation Now and the Ira W. DeCamp professor of bioethics at the University Center for Human Values at Princeton University, tells us that “Animals have been widely used in many areas of science since the 19th century. They continue to be used because that is how people are trained in many of the biological sciences – the older generation teach the younger to do it in the way they themselves did it.”

Singer details that methods that do not involve animals are gradually being adopted: “I suspect that better AI will also reduce, if not eliminate, the need for animals in this area.”

Aviva Vetter, cosmetics senior manager at HSI, tells us that there is a need for more familiarity with non-animal assessment methods as they are relatively new and confidence is still low.

“However, awareness of the limitations of animal testing is growing, as is familiarity with non-animal methods. The science is improving quickly, updating regulations takes more time, and often, changing peoples’ minds takes a bit longer – but change is coming,” Vetter adds.

Singer suspects that better AI might reduce, if not eliminate, the need for animals in this area.Saved by technology?
Chris Magee, head of policy and media at Understanding Animal Research, details the future prospect of replacing animal testing with technology.

“There are many exciting new technologies on the horizon, as well as incremental improvements to existing approaches, including animal models, and the trend toward reducing animal numbers that started in the 1970s, is likely to accelerate rapidly in certain areas like regulatory testing while barely moving in others.”

“AI has great potential, but boring-sounding changes can be just as meaningful. Big data requires big storage. We can’t collect or mine meaningful data off a 1.44 MB floppy disk. We also have vastly better systems of recording and cross-checking health outcomes as clinics move to digital records and citizens self-surveil with wearable health tech.” 

“We’re increasingly able to generate new health insights in the same way that your phone’s map app can spot a traffic jam from the aggregate data of all its users,” Magee suggests.

“Like building New York out of Lego”
Magee further exemplifies the liver-on-a-chip technology – natural miniature tissues grown inside a microfluidic chip – that detects 87% of liver effects that animal models would miss.

“This would equate to about 9 out of every 100 drugs that would ultimately fail animal or human testing being scrapped before they even got that far, which is great news. Is it the end of animal testing though? Of course not – it still leaves out 91 of 100 drugs.” 

Magee says that liver and heart arrhythmia is the biggest reason for safety failures, so the liver chips’ success would have to be replicated across many other target organs.

“However, we will still need animal models for certain things despite all this. The processes we’re mapping are often sub-microscopic, vastly complex, too poorly understood to model in silico and don’t present themselves in partial systems like human cells on a plastic strip,” he adds. 

“Parts of it are on the scale where biology interfaces with atomic and quantum physics. It’s like trying to build New York out of Lego.” 

Reducing animal trials in nutritionCryptobiotix aims to tackle main drivers of human clinical trial failure while reducing the need for animal trials in nutrition science.
We also speak with Aurélien Baudot, CEO and co-founder of Cryptobiotix, about how animal findings for the science of the human gut microbiome are not fully representable.

He co-founded Cryptobiotix to tackle two main drivers of human clinical trial failure in the field – the absence of predictive preclinical models and poor understanding of interpersonal responses early on.

“This is what we do in the laboratory with the SIFR (Systemic Intestinal Fermentation Research) technology – an ex-vivo gut microbiome model technology – while reducing the need for animal trials in nutrition science.”

“In the gut microbiome field, we see marked discrepancies in the gastrointestinal ecosystem of mice and humans, both in terms of presence and balance of bacteria. When exposed to nutritional ingredients, such dynamic ecosystems will react differently, different bacteria will grow and different functions of that bacterial ecosystem will be enriched.”

Baudot says that mice studies are still being used, typically in complex, immunity-related research questions. Scientists use them mainly to understand the downstream impact of their products after the gut microbiome has metabolized them. 

“Even then, they [researchers] know to consider the results with caution. Each model can answer a specific set of questions and has specific limitations. You would, for example, not investigate the development of bird wings in a mouse model.”

Question of accuracy
Magee explains that the accuracy of animal experiments for science depends on what is being tested and stresses the importance of study designs.

“Did they use a mouse for a gastrointestinal study? Well, that does not give you the best results, but equally, maybe they used a mouse to avoid using a monkey.”

“Sometimes that sort of trade-off is considered worth it, or maybe the study was looking at three generations of animals, so they wanted the rapid aging only found in the mouse, not the monkey and certainly not the human,” says Magee.

“Animal models are best thought of as a tool that must be used correctly to be effective, and that’s what good study design is all about. If a builder tried to drive a nail in with a screwdriver, it doesn’t mean there’s anything wrong with the tools or they can’t be used effectively, just that the particular builder’s results will be poor.”

“I think there is also an aspect of old habits dying hard. When we published our validation paper – comparing human clinical data to SIFR data, one of the reviewers lamented the absence of animal model validation for our findings. One can only appreciate the irony,” says Baudot.

The more animal testing is seen as critical for human health – the less opposition there is, says Vetter.Regulations and laws
Vetter at HSI details that it’s working on implementing non-animal methods for chemical and product safety by helping legislators to modernize regulations worldwide, collaborating with Animal-Free Safety Agreement.

She explains that animal testing is often required by law when involving chemicals potentially threatening humans and the environment.

“Many people are unaware of the type or amount of animal testing required for different types of consumer goods. Generally, polls indicate that the more the public is informed about animal testing, the more opposed they are.”

She explains that the more animal testing is seen as critical for human health, such as in drug testing, the less opposition there is despite data showing that animal testing does not provide the most reliable results.

Baudot details that in the US, the Senate recently enacted the FDA Modernisation Act 2.0, a bill signifying an evolution away from animal testing requirements in nutrition science. 

“In Europe, the EU issued Directive 2010/63, aiming to limit animal testing to scientific or educational purposes if no alternative is available.”

He argues that those are steps in the right direction. “Nevertheless, as scientists, we must not exclude the validity of animal models in specific cases and I believe animal testing will always be a part, albeit diminishing, of preclinical testing.”

“We are at a turning point in the industry, where both our customers and regulatory bodies gradually move away from animal trials toward more ethical data generation approaches,” Baudot concludes. 

By Beatrice Wihlander