NAMs: Can they really replace animals?

Welcome back to Tox & Trials, a newsletter that gives you a first-hand, raw, educational view of toxicology in the pharmaceutical industry.

NOM NOM NOM.

No, that’s not right.

NAM NAM NAM.

Better!

NAMs. New Approach Methodologies. A trending buzzword in the industry. Some might even say the future of medicine. But where are we actually going with it? And how do we get there?

The greatest draw to NAMs is the opportunity to reduce and even replace animal testing. If we can rely more heavily on in vitro and in silico methods, we can start to shift the paradigm away from relying so heavily on animal methods. And have more accurate, more translational data to boot.

The dream is great, the goal is lofty - but there is potential:

In the current landscape, there have been some significant leaps in the NAM world. A company called Emulate, Inc. has been quite successful in the organ-on-a-chip space, even creating a chip that can help predict Drug-induced Liver Injury (DILI). There are several companies, like Lhasa Limited , that use QSAR to predict ADME properties, simple toxicity and more. VeriSIM Life uses big data to computationally create new compounds to de-risk their development. Nova In Silico created a solution called jinkō to simulate clinical trials. The solutions are coming and they’re coming fast.

In the nonclinical toxicology space, one of the biggest ventures when it comes to NAMs is the work on virtual control animals, something that Charles River Laboratories and Sanofi recently partnered together on to bring to the forefront of nonclinical safety work.

But the question remains - is it enough? Will it be enough? Can it be enough?

The fact of the matter is that a lot of NAMs work is being done on either side of nonclinical safety in the pipeline, yet nonclinical safety is where the majority of animals are used in drug development. As of now, the virtual control animal initiative is the only nonclinical toxicology dedicated computational solution to date.

Why is this?

For one, I would argue that toxicology - your IND-enabling GLP studies that determine if you get to go first-in-human (FIH) - are your most sensitive studies. They take the most money, they take the most time, and they take the most risk. So, if OncoBio wants to get their new oncology drug into patients as fast and as capital-efficient as possible then they are going to focus more on doing what the regulatory agencies want them to do to progress the drug along rather than attempting some fancy AI for the first time. In other words, a lot of companies like to play it safe but typically because they can only afford to play it safe.

For two, nonclinical safety and toxicology studies are, by design, heavily siloed. What I mean by this is that the regulatory decisions made on drug safety are made by comparing data in a tox study with data in that same study. This makes breaking the silo open even harder.

Let me break it down:

What is boils down to is the first dose of a new drug that goes into people is, for the most part, based on the findings of ONE toxicology study engrained in a process that has barely been updated in years.

Now imagine a world where the study is run as usual, but the findings are also cross-referenced with external data that encompasses a multitude of angles: in silico data, in vitro data, in vivo data, human data, animal data, all of it. Imagine the accuracy of assessing toxicity of a new drug this way, versus the “siloed-way”. Imagine the translatability to the clinic. Imagine not having to run a tox study for 28 days or 3/6/9 months. Imagine the amount of animals we wouldn’t have to use. Imagine the money saved.

Imagine the risk saved.

I hate to burst your bubble, but this technology does not exist. However, the point is that’s where we should be headed, that’s where we should be thinking.

Virtual control animals are a great start, and they will help to reduce, refine and replace animals (the 3R’s). But I fully believe we can make a bigger impact, that there’s potential for a greater good. Perhaps virtual control animals are a Proof of Concept in a way, where we can show development of a computer-based model integrated into the toxicology space with regulatory support and high effectiveness. Maybe virtual control animals really are just the beginning.

And maybe all of the work being done on either side of nonclinical safety and toxicology will help bolster the work that needs to be done inside it. If you asked me when I started my career if we would be able to use AI to accurately read histopathology slides I would’ve thought you were writing a movie script for year 2124. But here we are, in 2024, doing just that.

Personally, will continue to work on NAMs in nonclinical safety, hopeful that one day we can truly make the impact I see in my mind. I encourage others to do the same. The need for in-the-weeds scientists is critical in developing these types of technology.

To that end, are some resources in case you want to join the front lines:

ASCCT - American Society for Cellular and Computational Toxicology

Society of Toxicology (SOT) - CTSS: Computational Toxicology Specialty Section

PHUSE : uses data to advance technology (data is queen)

American College of Toxicology (ACT) (educational)

Do you think NAMs will be able to live up to their potential?

Dessi McEntee is a board-certified toxicologist with extensive experience in bringing new medicines to the clinic and beyond through well-executed nonclinical development programs and strategy. She delivers Tox & Trials in an effort to help educate on the nuances of navigating nonclinical safety and toxicology within drug development, a commonly misunderstood or under-understood piece of the pharmaceutical pipeline.