Credit: Emulate

Lorna Ewart has long been frustrated by the mismatch between animal and human toxicity data. All too often preclinical studies in rats and dogs will suggest a drug is safe, only for clinical trial data to raise a red flag. In 2012, in the early days of organ-on-a-chip technology, she pitched a project to her colleagues at AstraZeneca to assess whether an engineered system that mimics the liver could better interpret cross-species safety data and tease out real toxicity concerns. “I got laughed out of the room initially,” jokes Ewart, who is now Chief Scientific Officer at Emulate, a leading organ-on-chip company. It took a further 12 months to get real traction and internal funding for the project.

The tune has since changed. In April, the FDA announced plans to phase out animal testing for new drug candidates approaching the clinic. The goal is to first overhaul toxicity testing for monoclonal antibodies, but other modalities are also on the agenda. US regulators are embracing organs-on-chips, organoids and an array of in silico and computational tools, with a goal of making animal studies “the exception rather than the norm for pre-clinical safety/toxicity testing” within the next 3–5 years.

Weeks later, the NIH said it was also deprioritizing animal models in biomedical research.

Ewart is excited about these moves.

The toxicology community has made massive strides with organs-on-chips, she explains. Emulate’s Liver-Chip can detect many hepatotoxicity signals that are missed by animal testing, the company reported in 2022. Liver-Chips alone, as an adjunct to animal testing, could save the biopharmaceutical sector US$3 billion in R&D productivity, Ewart and colleagues calculated at the time. When other organ-on-chip systems were factored in, the savings ballooned to $24 billion.

But many of these other chip models still need to be developed, characterized and validated for toxicology testing, adds Ewart. Drug developers are embracing the approach, but not in all programmes. And regulators still need to qualify these tools. Only then can the community work to fully implement these technologies alongside animal testing, where still required, to assemble comprehensive safety packages.

“There's a tremendous amount of work to be done but some huge opportunities ahead,” says Ewart.

What was your reaction to the FDA’s plan to phase out animal toxicity studies?

First of all, I was really happy to see it.

Regulatory acceptance is an obstacle that's put in my path every single day. I speak to lots of different pharmaceutical scientists who say: “we like what you're doing, but we're not quite sure how to use it. What does the data mean, and importantly, if I put it into an IND [investigational new drug] application is it going to slow down my application review?”

I understand their perspective, because I've been on the other side of the table. So I was really pleased that this announcement is a concrete move forward.

Let's not forget that this announcement comes on the back of a lot of advocacy and previous work. There was the FDA Modernization Act 2.0, which was signed in 2022 by President Biden, which changed the requirement from ‘pre-clinical’ to ‘non-clinical’ to allow for more use of human-relevant technologies. This was a way for the FDA to underline their interest in alternatives and encourage sponsors to submit data from organ-chips, organoids and in silico models, although without qualification for these tools it was unclear whether they could be used in an IND decision-making arena. Enter the FDA’s ISTAND pilot programme, which is aimed at supporting the qualification of novel drug discovery tools. [Emulate is in phase 2 of the ISTAND programme to qualify its Liver-Chip for the identification of drug-induced liver injury for certain drug candidates.] These were good moves, but they were never about complete animal replacement.

But I was also taken aback by the boldness of the roadmap, as it is hugely ambitious. However, without getting into politics, I will say that I am a strong believer that to effect change you have to get stuck in and create a little bit of positive disruption.

Is the goal of phasing out animal models in 3–5 years realistic?

It could be.

The current guidelines state a sponsor should do safety and toxicity testing in a small and a large animal. I've personally seen multiple projects read out with differences between those two animals, and that causes problems. The key questions are, which one predicts human response and does the second species identify toxicity signals that the first one doesn’t?

So in 3–5 years, the regulatory norm could go from requiring two species to one. And then we've still got this toolbox concept, this broad risk assessment strategy, but you've eliminated potentially one of the animal models.

I'm excited by the goal, I like to move at a fast pace. But there is a large voice of toxicologists who have raised a flag. The Society of Toxicology just released a statement that they are concerned about the speed at which this is going.

The focus is initially on monoclonal antibodies. What do you make of that?

I think it's a sensible place to start.

I don't have a hotline into the origins of the roadmap, but my suspicion is that this decision was multi-factorial given monoclonal antibody testing relies on non-human primates as the second species.

Firstly, there is a scientific strand to the argument. Not all human antigens are conserved in primates, binding affinity can be different and in some cases the target is absent. In such cases you can raise an antibody homologue to the animal antigen, but you still aren’t working with the actual antigen and immunogenicity remains a challenge. It's questionable as to why we've been taking this route.

Then, there is the very high financial cost of primate studies, which can take up to 36 months from study design to execution. So, there is probably a push for a more human relevant approach that is cost effective and likely a quicker read out on safety.

Lastly there is an ethical consideration. China was the major exporter of non-human primates to the US but this stopped during COVID. Countries like Cambodia saw an opportunity and started shipping primates, but sadly they were taking primates from the wild and didn't understand how to ship them appropriately. Consequently, many arrived dead on the shores of the US.

However, the problem with the FDA’s new approach is that most of the toxicities for monoclonal antibodies are immune-mediated. On one hand, it makes sense to move away from animals because animal immune systems behave differently to human immune systems. But I would suggest that there's still a lot of R&D needed to build good, immune-competent in vitro human-relevant models.

What about on the small-molecule side?

For small-molecule toxicities, there's around 8–10 organs that frequently present dose-limiting toxicities. So if one was to try to build, evaluate, validate and qualify models of those organs, and start using them within the next 3–5 years, that would give some comfort that one could legitimately reduce animal use.

It took you and colleagues 7 years to develop and publish your Liver-Chip. How long will it take to get the other models up to speed?

The hard part of that project was developing the dog liver chip, and in particular getting fresh dog hepatocytes. That exacerbated the timeline. Nowadays, it would take around 18–24 months to get a model up and running, and then you would start validation and qualification.

The heart and brain are other major target organs of toxicity. There are very good “heart-chips” that have a similar degree of sensitivity and specificity to what we reported with the Liver-Chip. In terms of neurological models, there's a lot of interest in mini brains, effectively neural organoids.

One of my hopes is that with these announcements the regulatory community can start to address a few of the loose ends that slow down the qualification process. For example, when we were submitting our Liver-Chip to ISTAND for FDA qualification, it took us a degree of time to write our letter of intent. If there was a template that we could follow, such time stealers could be erased and the teams can focus on doing the science.

The FDA has talked about launching a series of workshops, and the NIH has created an Office of Research, Innovation, Validation, and Application. I think that will help too.

The NIH’s push would require models that can be used for efficacy testing too. How different are the chip development considerations for efficacy versus toxicology?

For now most toxicology is still done in so-called healthy model systems. All of the cells that we use in our Liver-Chip are from donors that met strict criteria — no alcohol usage, no smoking, no illicit drug use, and so on.

This is starting to change with an active debate on using disease-specific models in toxicology testing. Pfizer recently halted a phase I trial of its GLP-1 antagonist [danuglipron] because of raised liver enzymes. While there is no publicly available information indicating that the affected participant had fatty liver disease, at least 70% of obese individuals have some fat accumulation in the liver and this is a known risk factor for greater susceptibility to toxicity.

But when you are looking at a disease model for efficacy, part of the reason it is extremely challenging is because there's never a consensus around a disease. Diseases are hugely complex and heterogeneous. Sticking with fatty liver, I could make a fatty liver model by giving hepatocytes additional fructose, for example, and that would be perfectly acceptable. But someone else might say I need to have fatty acids in there too. To exemplify that point, there's over 10 different mouse models of fatty liver. There isn't an animal model consensus.

The pharmacologist in me would say that when you’re building a model for efficacy, you have to pick what axis you're looking at and characterize that well. Don't try and be an all-comer for all things.

That's where you need the basic biology, and that’s the type of thing that I can see the NIH funding going forward.

Is there sufficient funding to get the needed model development done?

Funding is a big problem.

Since the FDA’s announcement, I haven't seen any shift in funding. And of course a lot of American academics are concerned that the NIH is shrinking its budget and so there is a legitimate question mark around where that money is going to come from.

What I would like to see is for money to be put forward to fund studies like the one that we did on our Liver-Chip. We didn't get any grant funding for that. As a small company that was a hefty investment, and I'm hugely grateful that our investors approved the spend. Not all companies can do that.

If we want to get that 3–5-year timeline to come alive, we're going to have to put some money behind it, because chip developers are often created from university spin outs.

We also need to put more funding into training future scientists. The field of bioengineering is still young, so we need to train cross-disciplinary scientists who understand biology but also have an engineering mindset, and vice versa. They are quite different skill sets.

While the FDA and NIH have announced ambitious plans to overhaul non-clinical testing, both organizations are also reeling from layoffs, re-organizations and funding cuts. How do you expect that to play out?

I'm not in either of those organizations, so I can't directly comment.

From a leadership perspective, it’s always critical to build and maintain high-performing teams. If the recent changes were driven by a desire for greater efficiency, I may have approached the situation slightly differently. My concern is that, as a result of these changes, some exceptionally talented and experienced individuals may no longer be part of either organization moving forward.

I think another concern, taking it out of the US, is how the rest of the world is going to respond to this?

Recently, there was a debate in Parliament in the United Kingdom about the use of dogs in medical research. It doesn't look like there's going to be any changes in the near term here, which is sad. The EU has said it's going to be 2026 before it puts a position out on animals in medical research.

The counter argument is that as long as the US retains around 48% of the pharmaceutical market, it will always lead the regulatory way.