Episode#124| Live Event Recording: NAMs-From Theories to Validations

Speaker bio: https://3dheals.com/new-approach-methodologies-from-theory-to-validation/

On demand video link: Coming soon

YouTube: Coming soon 

Subscribe to future events

What does it take for a biologically sophisticated model to become a tool that pharma, CROs, and regulators can actually trust?

In this 3DHEALS panel, experts in human-relevant in vitro models, 3D bioprinting, organoids, microfluidics, and drug development explore the transition from promising New Approach Methodologies (NAMs) to decision-ready assays. The discussion moves beyond the simplistic question of whether NAMs can replace animal models and instead focuses on a more practical challenge: how can these systems generate reliable evidence that improves drug-development decisions earlier?

Mike Clements examines the path from stem-cell-derived cardiac safety assays to regulatory-facing tools, including the importance of functional readouts, reproducibility, standardization, and clear context of use. Graham Craig brings a commercial perspective on why NAMs create value when they help teams advance, deprioritize, or stop programs before expensive downstream studies. Andrew Lee discusses fit-for-purpose 3D-bioprinted cardiac tissues designed to measure contractility, conduction, calcium dynamics, and arrhythmogenic behavior. Pranav Joshi explains how organoids can become assay-ready through engineering control, lifecycle quality control, standardized recovery, and transferable workflows. Alexandre Civiere shares practical skin-on-chip applications spanning permeation, wound healing, and tissue-aging models.

The panel also addresses adoption bottlenecks: model complexity versus utility, physiologic relevance, cell density, operator variability, QC gates, automation, standardization, regulatory engagement, and the role of cross-sector consortia.

Featuring:
Dr. Mike Clements, Axion BioSystems
Graham Craig, Voxell Bio
Andrew Lee, FluidForm Bio
Pranav Joshi, Bioprinting Laboratories
Alexandre Civiere, Revivo Biosystems
Moderated by Dr. Lowry Curley, Luna LifeSci

00:01:20 - 3DHEALS Mission And Networking
00:03:11 - Why Most Stem Cell Models Stall
00:05:40 - Cardiac MEA Assays And Regulators
00:10:32 - Standardization Roadmap From CIPA
00:14:33 - The Commercial Case For NAMs
00:19:20 - Fit For Purpose Beats More Complexity
00:27:08 - 3D Bioprinted Cardiac Tissues In Practice
00:35:12 - Arrhythmia Patterns And Disease Geometry
00:39:30 - Organoids To Assay Ready Workflows
00:45:57 - QC Gates Across The Organoid Lifecycle
00:50:02 - Skin Chips Permeation And Wound Healing
00:56:43 - Aging Models And Regulatory Reality
00:59:29 - Panel What Changed In Five Years
01:07:55 - Adoption Where It Moves Fast
01:13:20 - Physiological Relevance Versus Reproducibility
01:15:03 - Standardization Who Must Drive It
01:21:30 - Operator Variability And Automation Fixes
01:27:30 - Contamination And Process Traceability
01:30:12 - Why Organ Chips Still Lag
01:35:55 - AI That Helps Versus AI Theater
01:41:50 - How Much Animal Testing Can Shift
01:44:10 - Final Hopes And How To Help

Transcript

Copy

SPEAKER_04 0:00

Hello, hello. Good morning, everyone. Okay, let me just make sure the zoom is working and people are trickling in. Okay. Why is there no attendees? That's my question. Hold on one sec. Okay, awesome. I see two attendees. That's great. That means the technology is working. Okay. We just need N1 to make sure the technology works every time. It gives me it makes things really exciting for me. Okay, my name is Jenny Chen. I'm the founder and CEO of 3D Heals. You probably saw me quite a few times now if you attend these kind of virtual events. So 3D Heals has three missions. One is to educate the public about various topics relevant to 3D printing in healthcare. And today, I think is a relevant field that a lot of 3D bioprinting companies are interested in. And number two is to create network opportunities. And we do both in-person and virtual events. So even though virtual is not as powerful for networking, I would suggest that you put your social media links in the chat box. And I mean, if you want to, and share with your uh audience, share with the audience and then speaker. And the speaker can also share your uh connection, uh your links or your contact information in the in the chat box. And also tell us a little bit about yourself. You know,

3DHEALS Mission And Networking

SPEAKER_04 1:20

don't be shy, tell us where you are. I know we have about 20 countries registered for this event. So you must be from all over the place. So tell us where you're from, what you're looking for, particularly in this event. We got your questions and we're prepared to answer them. And the final mission is pitch 3D. We've been doing this since 2018. We have helped more than 100 startups to connect them with institutional investors, which means venture capitalists. So if you're in the space of 3D technologies and healthcare, connect with me. We are looking for early stage startups, which means you have to be before Series A to be considered. Okay, without further ado, I'd like to introduce the moderator of today's event, Dr. Laurie Curly. In fact, I recently did a podcast with him. So if anyone wants to learn how a neural organoid can play Pac-Man, that's a good episode. Slightly misleading, but it's a good episode. So, Lori, I'll let you take over.

SPEAKER_01 2:17

Okay, perfect. Well, yeah, thank you, Jenny. First, excited to help moderate this event. Space very close to my heart, spent a lot of time hearing, you know, the neurospace, as Jenny mentioned. Just a little bit of how we'll run this. We're gonna have each person here present. And while they're doing that, please, everyone in the audience, submit your questions, and those can be both specific to that presentation and along the broader topic. And so after that person presents, we will try to get to a few of the presentation specific. We're gonna save the broader ones for the end, but we'll be relatively quick with QA until we get to the panel because that's you know, I think where we're really gonna roll up our sleeves. And uh I'm excited to learn from everyone here. So without further ado, I want to introduce and let Mike Clements go ahead and kick us

Why Most Stem Cell Models Stall

SPEAKER_01 3:11

off.

SPEAKER_07 3:11

Great. Hi, everyone. Let me just uh share my uh share my screen. Can everyone see my screen? Great. Well, hello everyone, and thanks for joining us. Every year, hundreds of scientific papers describe exciting new stem cell-dered models, organoids, co-cultures, new differentiation protocols, and increasingly sophisticated human in vitro systems. Yet only a handful of those assays become ones that pharmaceutical companies routinely use or the regulators are willing to consider when making drug development decisions. But the question is why? The answer isn't simply better biology. It's productivity, reproducibility, standardization, and demonstrating that that model is fit for its intended purpose. Ultimately, the value of a stem cell derived model isn't determined by how closely it resembles human tissue in the laboratory. It's determined by whether it generates reliable, interpretable data that can support a real decision. And now we're seeing some human stem cell derived assays make exactly that transition from promising research models to technologies that influence drug discovery and safety assessment. I'd like to illustrate that using two examples I've been fortunate to work on. The first is the human IPSC cardiomysa MEA assay for assessing prarrhythmia risk. Or in other words, it's the risk that a drug could trigger a dangerous abnormal heart rhythm. The second is the emerging human IPSC neural MEA assay for seizure liability. There are different stages of maturity, but together they illustrate how a promising model can evolve into a trusted NAM. And by NAM I mean a new approach methodology, broadly a technology or approach designed to answer a specific hazard or risk assessment question more efficiently than existing methods. The key is that a NAM should be fit for purpose, able to generate relevant interpretable evidence for the decision being made. For human safety assessment, that often means improving our ability to predict human biology and understand the mechanism. And when a NAM does that, it can also reduce reliance on traditional animal testing, for example, by identifying high-risk compounds earlier, reducing repeat studies, or supporting decisions where animal models are less relevant. And regulatory thinking has evolved alongside the science. This shift didn't suddenly begin with recent legislation. Regulatory engagement with NAMS has been building for more than a decade through initiatives such as SIPA and FDA-led validation studies. The FDA Modernization Act 2.0 provided additional flexibility around how non-clinical evidence can be generated. But the key point is that the scientific bar hasn't been lowered. Regulators still need confidence that an assay is relevant, reproducible, and appropriate for its proposed context of use. So

Cardiac MEA Assays And Regulators

SPEAKER_07 5:40

what does that look like in practice? Cardiac safety provides perhaps one of the clearest examples. A major concern is whether a drug could provoke a dangerous arrhythmia, particularly tossard de points or TDP, a rare but potentially fatal ventricular arrhythmia. TDP is associated with prolongation of the QT interval, which reflects delayed electrical recovery or repolarization of the ventricles. Because QT prolongation can signal increased TDP risk, it became a key focus of ICHS7B and E14 guidelines. The S7B nonclinical guidance relies heavily on HERG potassium channel inhibition, which can delay repolarization and prolong the QT interval, together with QT measurements in animal studies, typically in beagle dogs. Those approaches have been very effective at protecting patients, but they also have limitations because they focus largely on surrogate markers of risk rather than measuring the integrated human cardiac electrophysiology that ultimately determines whether arrhythmia occurs. Human IPSC-derived cardiomyocytes offer a potential solution. Rather than measuring a single ion channel, these cells incorporate multiple interacting ion channels within a human cardiomyocyte, so we're moving towards an integrated functional system. When these cardiomyocytes are cultured on multielectrode arrays or MEAs, we can not invasively record their electrical activity. We can measure parameters such as field potential duration, beat rate, conduction, and rhythm. And importantly, we can detect arrhythmia-like behavior directly. That's the distinction I want to emphasize. The assay measures function. But compelling biology alone wasn't enough to drive adoption. A critical step was the comprehensive in vitro proarrhythmia assay, or CIPRA initiative. CIPA brought regulators, pharmaceutical companies, and academic groups together to develop and evaluate a new framework for assessing proarrhythmia risk. Human IPSC cardiomyocytes provided the integrated biological check within that framework. Crucially, multi-sensor studies then demonstrated that the assay could produce meaningful results across laboratories, and those studies helped establish confidence that the technology wasn't simply working in one expert laboratory. It could become a reproducible assay. That evidence ultimately contributed to regulatory recognition of the human IPSC cardiomyocite data as supportive evidence in prorogmia risk assessment. Perhaps even more important is what happened next. A recent FDA analysis reviewed IND submissions between 2012 and 2025 that included human IPSC cardiomyocite data. An investigational new drug, IND, is a regulatory application submitted to the US FDA and allows a sponsor to begin clinical trials in humans with a new drug or biologic that has not yet been approved. Forty-eight IND submissions have been submitted to the FDA CEDA with human IPSC derived cardiomyocite MEA impedance studies as supporting data. These human IPSC cardiomyocite studies demonstrated higher overall accuracy in predicting clinical QT prolongation risk compared to standard HERG and in vivo non-rodent QT studies alone. Notably, to achieve comparable accuracy to human IPSC cardiomyocytes, additional animal studies beyond the standard in vivo QT assessment were needed. Even more interesting, the authors from the FDA went on to propose several scenarios where human IPSC cardiomyocite studies could be considered as a replacement for in vivo rodent QT studies, shown here in this table. So we're moving from interesting biology to validation to multi-sensor reproducibility to regulatory facing decision support. And once that happens, commercial adoption follows. Today, multiple CROs routinely offer SIPA-style human IPSC cardiomyas assays. In other words, this is no longer simply an experimental technology. The FDA's ISTAN program provides a pathway for the formal qualification of innovative drug development tools. In 2025, Axian submitted a letter of intent for the Human IPSC cardiomyas MEA essay, and that letter of intent was accepted into the ISTAN qualification program. I want to emphasize that the acceptance into ISTAN does not mean the assay is already qualified. Rather, it marks the beginning of the qualification process. It provides a formal pathway to work with the FDA to define and generate the evidence required to demonstrate the assay is fit for its proposed context of use. So in a little more than a decade, we've seen this assay move from an emerging stem cell model to a broadly used regulatory facing NAM. And now it's into a formal qualification process. I think the lesson from this journey is extremely important. Standardization may not be the most exciting part of science, but it's often what determines whether a technology becomes widely adopted. Human relevant biology is important, but industry and regulators also need confidence that an assay will produce reproducible interpretable data wherever it's performed. So can we use this same roadmap elsewhere? I think neural safety provides an interesting test

Standardization Roadmap From CIPA

SPEAKER_07 10:32

case. Drug-induced seizure liability remains difficult to assess preclinically. Traditionally, approaches rely heavily on behavioral observations in animals, which can be difficult to interpret and don't always translate cleanly to human risk. MEA technology offers an alternative because it directly measures neuronal electrical activity and network synchronization. Historically, primary rodent neurons have been widely used because they form robust, reproducible neural networks. But there's growing interest in moving towards human IPSC derived neural cultures because species differences in iron channels, receptor pharmacology, and seizure thresholds can complicate translation from animals to humans. Human neural models therefore offer the potential to measure seizure-related network activity in a more directly human-relevant biological system. But the neural field currently faces the same challenges that the cardiac safety faced earlier, standardization. Different laboratories use different cells, media, maturation periods, and analysis parameters. So one of the efforts we've been involved with has been defining practical functional acceptance criteria for these assays. For example, before asking whether a compound disrupts a neural network, we first need evidence that the neural network itself is sufficiently mature and active. That means defining quantitative baseline criteria and appropriate positive controls. These details might sound mundane, but they're exactly what makes results comparable between laboratories. And we're beginning to see encouraging translational evidence. Studies from groups including ACI and BMS have demonstrated how MEA neural assays can provide useful information about seizure liability and help identify safety risks during development. So the assay is beginning to move from academic exploration towards practical decision support. I don't want to overstate where the field is today. The human IPSC neural seizure assay isn't at the regulatory maturity of the cardiomyocyte assay, but the trajectory looks very familiar. In other words, we're beginning to see the same roadmap emerge that transformed the cardiac assay into a regulatory facing technology. Standardization, reproducibility, collaboration, and then shared data sets. So what can we take from these two examples? I suggest three things. First, NAMS don't need to be to recreate the entire human body. They need a clearly defined context of use and enough biological relevance to answer that particular question. Second, standardization and reproducibility are essential. Without them, even outstanding biology struggles to move beyond specialist laboratories. And third, reducing reliance on animal testing is an important benefit of NAMS, but go but it goes hand in hand with other major opportunities, enabling earlier, more human-relevant decisions. That can mean identifying problematic compounds sooner, gaining better mechanistic insight, reducing late-stage surprises, and ultimately making better decisions about which compounds should should progress into more resource intensive studies. So I'll finish where I started. Why do only a handful of stem cell models become assays that industry adopts and regulators are willing to consider? It's not because they're the most complex. It's because they're functionally relevant, reproducible, standardized, and fit for purpose. And I think that's an important distinction as we develop increasingly sophisticated organoids, microphysiological systems, and other human models. The challenge isn't simply to build better models, it's to turn those models into reliable evidence that people can trust to make decisions. And ultimately, I think that's how we move NAMS from theory to validation and from validation into real-world drug development. Thank you very much.

SPEAKER_01 14:00

Perfect. Well, thank you very much, Mike. Appreciate that. Great intro and very informative. Next, we are going to bring on Andrew Lee, co-founder, senior scientist at Fluidform Bio. Great.

SPEAKER_05 14:13

Let me uh share my screen. Actually, could we uh I think I'm having some trouble seeing my content. Could we go to someone else first?

SPEAKER_01 14:21

We can certainly do that. We will skip forward to the next person, come back to you. So, Graham Craig, uh Chief Commercial Officer at Voxcel, please go ahead and take it away.

SPEAKER_06 14:31

I will see my screen?

The Commercial Case For NAMs

SPEAKER_06 14:33

Perfect. Great, thank you. Uh thanks, Laurie. Thanks uh 3D Heels and uh Jenny for organizing this uh fantastic event today. I really appreciate being uh invited and feel quite humbled to be joining such a great group here for this discussion. So thank you. So yeah, my name is Graham Craig. I'm uh Chief Commercial Officer at Boxel, and I want to approach today's topic from perhaps a slightly different perspective. And you know, I come from uh quite a commercial background and spent a lot of the early part of my career as a strategy consultant actually sort of interviewing RD scientists to try and understand, you know, the timelines, the budgets, the probabilities of sex success for their clinical programs and preclinical programs. And what I what I saw firsthand was a consistent sort of struggle, wrestling with building confidence that a drug program was actually uh going to translate and that a drug actually might work in the way expected once it went into humans. And so subsequently, there's quite often uh a bit of a leap of faith in terms of jumping from preclinical into the clinical setting. And obviously, as we all know, uh it still remains a very sort of risky project overall. So what I want to talk today about today is really the opportunity where I think NAMS sit in terms of drug development, what this means from a commercial perspective, and what it's gonna take really for newer technologies like this to be embraced. So I don't think I I kind of agree with Mike's closing comments there, that I don't think drug developers necessarily need more models for the sake of uh sort of a broader repertoire of data or more sophistication. I think what we're looking at here and what what the market needs is sort of greater confidence that the decisions that are being made are actually going to translate into humans and therefore, you know, better says better decisions can be made earlier. Now, there's a tremendous amount of interest in this space right now. Over the last few years in particular, we've seen significant shift in the regulatory environment. The the sort of milestones and the timeline that I've laid out on this slide should be familiar to anyone who's been looking at this space. But I think to understand the gravity of this, you really have to sort of think about the how typically slowly bureaucracy moves, especially when it comes to health and life science. And so things like the Modernization Act, the roadmap that the FDA produced for reducing animal testing, these are all sort of significant signals, I think, that the regulators were trying to sort of stimulate a shift in thinking and mindset. 2026 is looking like a real inflection point for the NAMS field with the draft guidance that was released, with the NIH announcing this huge $150 million in funding to develop things like Ariva. And then a real signal that stood out for me very recently, a couple of weeks back, was the announcement from the National Clinical Institute that for grant applications, they're going to ask applicants to actually submit explicit reasons why they want to include animals in their studies and why alternatives are inadequate. So when you consider that the vast majority of drugs actually, you know, their origins stem from work in academic labs, I think this is a real important signal of that change in mindset that's being encouraged. Where I think, you know, where the regulators and and more broadly the industry are asking people to think, you know, not necessarily what do what, you know, what is my drug going to be indicated for, what is the right disease model that I can I can test this on. So more of, you know, what's the questions I want to answer and what's the best uh about my drug and what's the best model to answer that particular question. And I think why that mindset change is important to understand that we need to sort of look more commercially at the context, the business context of drug development and the environment that we're working in. So what I'm showing you on this chart is sort of the archetypical business case profile for a particular drug development path, right? So we we know that there's 10 to 12 years of heavy investment into drug programs, and then drug developers have a limited window of opportunity to go away and commercialize that drug before patent expiry. So it's it's a long, it's an expensive, and it's a very risky investment. And as an industry, we've spent decades generating increasingly sophisticated preclinical data. We invested mind-blowing amounts of capital into the infrastructure around this, but our drug development programs still have this brutal level of attrition, you know, 5% or less chance that your drug's gonna make it. So for me, the fundamental issue and the opportunity for NAMS is that it's not necessarily we can generate more preclinical data or be more technologically intensive. It's more that, you know, we have an opportunity to now to generate better data earlier in development, and specifically an opportunity to learn more about human biology before we make these increasingly expensive decisions to advance programs to the clinic. So if we

Fit For Purpose Beats More Complexity

SPEAKER_06 19:20

sort of zoom in and look at that earlier part of the development pathway, there's some fundamental questions that always need to be answered. So, you know, does my therapeutic engage its intended target? Does that engagement actually trigger the downstream biological changes that you're expecting? Can you actually reach the target? What are the liabilities around treating that target? These issues, historically, unfortunately, have only really been fully resolved and answered once drugs have gone into clinical trial. And so what NAMS, I think, potentially allow us to do is pull some of those questions and the answers to those questions much further to the left and much earlier. And that leads to really one of the main points that I wanted to make in today's discussion, and that's that the real value of a NAM is basically proportional to the cost of the decision that it can bring forward. So if my model can give me sufficiently credible information to advance or even deprioritize or stop a program before I make a big investment in an animal study, you know, that is tremendously valuable to the industry. So, you know, rather than looking at NAMS principally by asking what animal model does this replace, I think it's important to ask, you know, what's the decision category that this improves? So I've grouped sort of four key decision areas here. The first is target engagement. You know, not simply does the molecule bind a receptor or a target, but you know, do we see meaningful engagement in a human relevant context? So, you know, I've spent a lot of time in the field of antibody discovery and development. And so we know now that antibody-based therapies are getting increasingly targeted, going after very novel epitopes and targets that frankly might not exist in that form in animals. And we're also, you know, generating human antibodies, and so you know, to expect them to go and be a species cross-reactive, they may not also always bind in rodent models, for example. And so people will spend hundreds of thousands of dollars genetically modifying a rodent, for example, to express that specific epitope, and it might take a couple of years to get there, by which time the target is no longer novel, and uh, you know, you've lost that advantage of your novel target identification. Second is, you know, the functional efficacy, once you've engaged that target, does it produce that downstream biology changes that we expected? Third is really, you know, we can answer questions about liability identification, I think, much sooner. You know, are there unwanted biological effects that appear alongside the efficacy? I think that's becoming increasingly important as we develop more complex things like ADCs and biospecifics and T cell engagers, where you know, relevance, the species relevance can become increasingly challenging and liabilities you know increasingly harder to predict. And finally, I think, you know, exposure, tissue interaction, you know, does your drug actually get into the relevant compartment? Can it penetrate tissue? Can it pass through blood vessels and get into uh the surrounding tissue? You know, these are real four key areas where I think NAMS can materially change and improve development decisions. But there's, you know, there's an important qualification to all of this. I think more complexity is not automatically better for a model. You know, if you want to understand simple target binding, then a simple biochemical assay might be perfectly fine. Or you know, a cytotoxicity question, just putting your target or your therapy into a 2D culture might also be perfectly sufficient. But once you get into things like immune trafficking, vascular toxicity, questions where you know spatial biology becomes more important, you know, you have to choose the right model for the right question. So I don't think the future is one where we simply kind of make models more complex. I think it's ones where you know the we start to appreciate that sort of the least complex model that can reliably answer a question that we actually need answered is the one that we need to use. And that's really what fit for purpose should mean. So and this brings us to a validation, which which Mike also spoke a little bit about. And that's really the heart of today's discussion, I think. There's there's lots of scientifically interesting and beautiful models out there, but being uh interesting and cool is not uh, you know, necessarily what's needed when it comes to the appropriate tool for drug development. It's really first biological relevance. You know, does this model have the components that we need to answer the question? You know, can we characterize it reliably and reproducibly? And are we able to benchmark it against known human biology and really have that clear context of use? And then, you know, what is the actual decision that we're we're trying to qualify from this and inform this model for? And then finally, and I think most importantly, is adoption. So can my model actually be used in another laboratory? Can they reproduce the same results? Does it easily fit into their workflows? Is the you know the cost-benefit analysis and the economics of using that model make sense? You know, validation, I think, doesn't necessarily just mean proving that the model reproduces the biology you expect. It's it also means you know reliably answering those questions, those specific questions that we want them to answer. And then that means, you know, I think we needed a lot of things to come together to make the NAM space evolve. And I think we're seeing signs now that this is moving beyond just experimentation and that we're actually now getting adoption. So the NIH putting all that uh capital into the standardization of these tools, you know, reportedly last year Charles River generated $200 million of revenue from their NAMS portfolio. And you know, my other breadcrumb here that I've shared is from a syndicated report saying that, you know, that the you know, I take these with a pinch of salt, by the way, these kind of reports, but the fact that people are estimating billion dollar market for organs on a chip and you know, such huge growth rates like this, you know, you know, grabs attention. So clearly, you know, we've got the capital in place, we've got regulatory support and some evidence of commercial momentum behind this. And I think, you know, this is all fantastic, but what's really going to determine whether these technologies become actual key infrastructure is gonna be sort of a bit of the wood chopping activities, right? It's gonna be showing the reproducibility, standardizing things, showing that they're transferable. So the sophistication will get attention for NAMS, but I think it's the adoptability that is really key to getting these technologies used. So I'll finish on three thoughts. First, I think the regulatory environment has moved decisively here and it's definitely encouraging use of these kind of models. Second, I think the greatest near-term opportunity for NAMS is not necessarily in replacing animals. I don't think we'll ever wholesale replace animal studies, but I think they can significantly improve our ability to make better decisions earlier. You know, can we can we validate target engagement? Can we show those good functional responses? And third, I think widespread adoption is going to ultimately depend on the validation and adoptability as much as it does on any scientific sophistication. But I think perhaps the real sign that this field will have succeeded is when we start referring to them as alternative models at all and simply see them as the most appropriate models for the question that we're trying to answer. So those are my thoughts on this. Uh thank you very much, everyone, and I'm I'm looking forward to the rest of the discussion today.

SPEAKER_01 26:30

Perfect. Thank you, Graham. I like that point when they're no longer alternative, they're just methods. Yeah. Uh so quick reminder, everybody, please ask questions. There's a lot, I'm sure, that not everyone knows. And even if it's maybe a point you have that you want to discuss later that your colleagues may not know, we want to have a discussion that is informative. I know we will, but yeah, let's let's answer what we need to. So Andrew, would you are you ready to go? Or okay, perfect. Next we'll go to Andrew Lee, co-founder, senior scientist of Fluid Form Bio.

SPEAKER_05 27:07

Thanks,

3D Bioprinted Cardiac Tissues In Practice

SPEAKER_05 27:08

Lori. Yeah, I just want to share my perspective on NAMS. I think my discussion will be a little bit more application-based. So we are a company that focuses on creating replacement tissues using 3D bioprinting. And so I think a lot of the points that the other panelists have mentioned really aligns well with you know how we're thinking about going about creating and building these models for neocessions. So, you know, I think for this talk, I'll be focusing specifically on cardiac, you know, cardiovascular diseases. And um, you know, I think of course this field extends, be extends way beyond just a specific, right, specific niche. But in this case, you know, there's a huge unmet need that exists in the cardiovascular space in terms of the patients that have needs that need to be addressed. So highlighted in red here, there's about 30% of unmet need in the cardiovascular space, and really only about 5% of that of the existing innovation in bar pharma is targeted towards cardiovascular diseases. You know, and you know, the question bears to uh you know, we can ask why are there so few pipeline assets where there's such a significant need. And really it comes down to there is a it's a very challenging field uh all across the board, right? So there is a very low approval rate across disease areas, and really the transition time from phase to phase is extremely long, as we all have, you know, those of us have experience in this field can tell. So high failure rates takes a long time and really costs much more than many other disease areas. And you know, I think really as Mike and Graham have mentioned, really that's there's this sort of transition translational problem where we want to gather as much data early on as we can, but we have a problem going from these cellular target level information data sets towards these animal models, human, human models. So, you know, we really don't have we're not really passing the best candidates from our pre uh from our lead optimization phase into our preclinical testing, where we're really investigating more of the system, you know, systemic organ level, animal level effects. And really, you know, we we don't we don't collect the key data that we need to understand these disease modifying effects on human tissues. And so in in the cardiovascular space, specifically related to contractility, conduction velocity, metabolism, and calcium handling. And so, you know, I think as I mentioned, you know, really the current current paradigm is focused on these binding affinities, target engagements, and I think there could be a much better way of improving that translation using NAMS, right? Going directly from our lead optimizations and then better understanding how that translates to human specifically, and then eventually replacing the animal part of of the of the pipeline. So for us, you know, we've developed this platform where we're able to build, you know, fit-for-purpose, disease-specific tissues in any sort of geometry that you could think of in different well-played formats, and then and then really design the functional readouts uh specific to the to the tissues themselves. So tissue and organ scale specifically, right? Uh not just on the molecular and cellular level where we're looking at these tissue level functions. So you can see these these different tissues. I'll play some of these videos here. Um, you know, we're able to build a wide variety in of geometric complexity and also cellular and compositional complexity tissues. And then so we'll take you know any sort of human cells, an ACM matrix, build a start with sort of computational design tissue architecture, and we publish this in some high-profile journals, uh, and you know, take these and directly three bioprint them using these native biomaterials. And this is really what enables, I think, these NAMS for uh the specific functions that we're looking at of them. And then we'll take our robotic biofatification. We have the ability to control deposition any way we want it, and then really design, as I said, design analytical capabilities and physiological readouts specific to these three-dimensional tissues through to this systemic level information. You know, and I think some of the highlights is really we're able to create the native tissue structure, right? So at multiple length scales, we're able to deposit these, you know, cells and cardiac tissues at native like cell density and alignment. Um, and you'll see with some of the data I'll show is that we can build these healthy cardiac tissues and design the tissue specific to the imaging modality, the data that we want to collect. You know, we want to demonstrate responsiveness to test compounds, tool compounds, and then eventually we can apply this toward disease cardiac tissues, whether it's you know, mimicking the you know composition or mimicking the disease structure phenotypes that are demonstrated in these disease tissues, and then be able to uh reproduce a model of that disease phenotype. So here, you know, here are some of the again, some of the tissues that we've shown that we can fabricate. And this is just a small sub small sample of what we can create. And you can see from some of these more two-dimensional two and a half D structures to three, you know, true 3D, you know, some of these ventricular structures. And I think being able to start to create some of these ventricular level readouts where you not only have contractility readouts, but you start to get you know volume, pressure, ejection, fraction type information that then starts to um get us closer to collecting the data that we we might want. Right. And then I think another key component to to the approach moving forward is really starting to replicate the the you know physiological scale cellular density that exists. You know, a lot of the approaches today really use utilizes a fraction of the cellular population and r only creates a small, a lower density tissue. And I think for us being able to replicate of upwards of hundreds of millions per ml density, that's that's really where I think the magic lies, right? Being able to get the cells in there, have them do their thing, and then uh I think that's where the function is primarily derived. And you know, we're this is just some of the analytical pipeline, right? Taking these tissues, being able to show that we can readout some of the calcium dynamics out of them as well as the contractility readouts. We are able to plot some of these optical mapping and then be able to calculate out conduction velocity and the conduction and and and isotropy. And these are starting to get more of these three-dimensional readouts that are, I think, primarily relevant in in things like, for example, rhythm arithmogenic behaviors, where you not only have a dysfunction in the rhythm, but also you start to see dysfunction in the conduction pattern as well as the propagation of these action potentials. And so being able to visualize that in a dish ahead of time, that's that's getting us, I think, one step closer to that system level information. We're able to demonstrate some of these functional readouts. I'll kind of go through these pretty quickly. So to isopotteranol, classic beta adrenergic drug, looking at is there a functional cyclic AMP signaling, and then also a variety of different key pathways of interest that respond well. Um and then this is this part here is really starting to to think about how we can engineer,

Arrhythmia Patterns And Disease Geometry

SPEAKER_05 35:12

specifically program in some of the, you know, some of these erythmogenic behaviors that that are reminiscent of disease states, right? So you can see some of these multiple wavefronts that are signatures of re-entrance circuits in in a human heart. Uh you have these rotor activities that also are pinned to a specific defect that in a disease state might be a fibrotic region that is that is leading to this sort of rotor behavior. We can also induce the arrhythmogenic behavior, so chemically induce them and then you know go about fixing them with a lidocaine treatment, which is very common for treating erythmogenic behaviors. And then lastly, I want to share, you know, some of these structural structural patterns can also be recreated. So in some of these fibrotic myocardiums, as you as you start to age and disease states develop, a lot of times you get excessive deposition of collagen, and we can go in and start to computationally design some of these tissues and their uh substrate phenotypes, and then also then recreate them using the rebile printing, specifically our fresh printing technique, and then create some of these patterns reminiscent of a disease state. So this really extends what you know extends well towards complex geometries, complex compositions, and then being able to create that substrate from which we can investigate and gather the data that we that we intend. So, you know, I think the key to building some of these translational disease models and also NAMS in general really comes down to some of these five points, right? So better physiology that's mimetic, truly memetic to the human tissues that we are looking to build, being able to tune these designs, having access to both the cell and material inputs, but also having the fabrication strategies that come in and matches the ability to process these materials. And then we have a controllable process, right? That is based on first principles based on engineering and parametric design, where we have 3D robotic control and the fabrication to infinitely iterate on on this on this process. And then I think then creating these testable tissues that that we are gathering data on in a longitudinal study, not just at acute studies that have you know fixed time points, right? So we can monitor them over time. So I think these are some of the key key principles that I that would be moving forward, we can rely on to create useful models. And I think that really is the key is going beyond creating models for creating for creating them sake, but really thinking about how how are these models actually useful, specifically in a you know commercial setting, in a drug delivery setting drug, sorry, drug development setting. And so I will end my presentation there and uh looking forward to sharing more and answering any questions.

SPEAKER_01 38:10

Perfect. Thank you, Andrew. Very, very insightful and interesting there on the cardiac side. And I think you made a good point at the end also, you know, one spectrum is building these for you know the sake of building them. And I think from let's say an academic perspective, that is important. All of these start here, right? And there may be some more unique biologies, some more unique uh tools that you develop there, but most of us here are coming from the commercial perspective, and you have to pretty quickly drill down into you know what do customers want, context abuse, right? Something people talk a lot about here, and then the data that you showed, I think was very compelling, and you need that to convince pharma, right? So when I'm working with companies on my consulting side, that is a lot of it. It's focus, figure out what your customer wants to see. Uh, and there's so much we can do with these things, focus isn't easy. So thank you. Yeah, that was a really good point there at the end. Okay, so now we're gonna move to Pranav Joshi, and he is a senior scientist at bioprinting laboratories. Take it away.

SPEAKER_00 39:19

Yep. Thank you, Lori, and thank you, 3D Hills, uh, for this opportunity to be part of this discussion. And I can see the screen.

unknown 39:28

Okay, please.

SPEAKER_00 39:29

So,

Organoids To Assay Ready Workflows

SPEAKER_00 39:30

yes. Today I want to focus on a very practical question. How do we move from a biologically promising organoid model to a NAM that another lab can actually use with confidence? Organoids can provide remarkable biological relevance, but biological sophistication alone does not make a model useful for a drug development. To become decision ready, the biology has to be reproducible, scalable, technically characterized, transferable, and validated for a clearly defined purpose. I'll discuss how engineering control, fit-for-purpose validation, and ultimately an essay-ready delivery model can help close that translation gap. So, one of the major motivations for NAM development is the persistent difficulty of translating preclinical findings into human outcomes. The problem is that species differences, simplified biology, and variable experimental workflows can reduce confidence in human predictivity. So for NAMS, the goal should not simply be to replace one model with another. The more important question is: does this model improve confidence in the specific decision we are trying to make? That idea becomes increasingly important as our models become more biologically complex. So as we move from simple 3D culture towards organoids and even more advanced organoid systems, we can capture more human-relevant biology. We gain human genetic context, multicellular architecture, and a biology that evolves over time. But complexity also introduces new challenges. We see greater variability, more demanding handling, increased operator dependence, maturation difference, and sometimes more difficult interpretation. So there is an important trade-off. The goal should not be to build the most complex model possible, it should be to use the level of complexity required to improve the intended decision. That is the principle I want to carry through the rest of this presentation. A promising research model and a decision-ready NAM are not the same thing. A promising organoid may show an interesting biological phenotype, whereas a decision-ready NAM needs a defined context of use. An expert-dependent protocol needs to become a transferable SOP. Representative images needs to be complemented by quantitative acceptance criteria, and a successful experimental batch needs to become reproducible performance across multiple batches. Moreover, biological plausibility needs to translate into demonstrated performance for the intended decision. This is also why fit-for-purpose validation is so important. The validation burden should be proportional to the decision being supported, not to the visual complexity of the model. So, to put it in other words, the best model is not necessarily the most complex model. It is the least complex model that captures the biology needed for the intended decision. And we recently had an opportunity to examine this problem from another perspective through NIH's iCOR customer discovery program. And one finding changed how we thought about the product. Users were not primarily telling us that they needed another platform, they were describing what happens before even the experiment started. Months of culture, specialized personnel, failed batches, metrics handling, transfers, and variable recovery. That shifted our question from how do we provide a better culture platform to can we shift the customer from building the biology to running the assay? That leads towards an assay-ready approach, qualified organoids on plate, standard recovery protocol, batch release information, reference controls, and an automation compatible workflow. And I think that lesson extends beyond our own platform. For NAM adoption, model performance is only part of the problem. The operational burden matters just as much. So our pillar and perfusion platform gives us one example of how engineering can reduce some of that operational variability. The objective is not simply to grow a complex organoid, it is to control where the cells are placed, how the initial structures are generated, how the organoids are cultured, how nutrients are delivered, and how they are measured, and eventually how they can be stored and distributed. 3D bioprinting helps control initial loading. The standard 384 well footprint supports automation. Palm free perfusion provides dynamic culture without external pumps or tubing, and in situ imaging reduces disruptive transfer steps. The platform itself does not define the context of use. It provides the process control infrastructure needed to develop different organoid models into more reproducible assays. So our liver organoid works. So this example is from our liver organoid work. This provides three complementary pieces of evidence toward that goal. So the first is manufacturing control. We demonstrated control 3D bioprinting and scalable organoid expansion. Second is functional response. So using an injury and recovery paradigm, the liver organoids demonstrated regenerative behavior. And third is initial on-plate cryopreservation feasibility. We demonstrated the first phase of preserving liver organoids directly on the plate and recovering them after thaw. Together, these results support feasibility across manufacturing, biological response, and cryopreservation. But this is not yet a finished acety product. Post-thaw identity, function, batch consistency, and assay performance still need to be systematically established. And that distinction leads directly to what we mean by acety product or acety models. Because cryopreservation by itself does not make an organoid acet RAD. Viability after thaw is only the beginning. An acet AD workflow starts with reproducible manufacturing. Then we need to qualify the identity, morphology, viability, uniformity. And next comes the cryopreservation and distribution with defined freezing conditions, storage stability, and ultimately shipping robustness. Recovery also has to be standardized. We need a defined recovery protocol and a predictable recovery window. Then the customer should be able to move directly into the assay using reference controls and an automation compatible workflow. And finally, the recovered biology has to regenerate or generate reproducible functional data that meet the acceptance criteria for their intended decision.

QC Gates Across The Organoid Lifecycle

SPEAKER_00 45:57

So the transition is not simply from fresh organoid to frozen organoid. It is manufacture, qualify, preserve, recover, essay, and then decide. This is what we mean by moving toward an SRAD NAP. The brain organoid program allows us to apply this framework to a much more specific context of use. Here the application is developmental neurotoxicity hazard identification and prioritization. And defining that context changes how we think about the model. A brain organoid does not need to create an entire human brain. It needs to reproducibly capture the developmental biology necessary to detect relevant disruption. Our workflow therefore focuses on control spheroid generation, plate uniformity, maturation, and response to reference chemicals. The validation package is being built around morphology, molecular markers, functional endpoints, reference chemical response, reproducibility, and RNA sequencing and other multimodal integration. The key question that we need to ask here is can the model reliably detect decision-relevant developmental disruption? So this brings us to one of the most important aspects of reproducibility. We cannot wait until the final assay to discover that the model failed. If an endpoint fails at the end of a multi-week differentiation, we need to know whether the problem originated with the starting cells, employed body formation, transfer, maturation, cryo preservation, or the assay itself. That is why QC has to be distributed across the organoid lifecycle. At the process level, that includes starting cell quality, EV uniformity, transfer efficiency, maturation and recovery. And at the assay level, we need plate uniformity, morphology, viability, lineage markers, functional endpoints, reference chemical response, and reproducibility. And each QC gate needs two things: a predefined acceptance criteria and a clear corrective action before advancing to the next stage. Ultimately, all of these measurements should answer one higher level question. Does this NAM provide sufficiently reliable evidence for its defined context of use? That is the decision performance that we need to consider. So even a well-validated and technically strong NAM will not automatically be adopted. Several things have to converge. The context of use has to be narrow and credible. Manufacturing has to be reproducible enough to support meaningful batch release criteria. Biological and technical validation have to be fit for purpose. The workflow itself has to transfer across operators and eventually across labs. And the model has to be compatible with the infrastructure users already have for their storage, recovery, imaging, automation, and analysis. So biological relevance is necessary, but adoption requires confidence in both the biology and the operation. For developers of NAMS, that operational layer needs to be considered much earlier in the development process. So I would like to close with three points. First is this human relevance is necessary, but complexity must serve the decision. And the second is the context of use definitely defines the validation burden. And third is the adoption requires qualified, transferable, assay-ready workflows, not the isolated models. The opportunity is to combine human-relevant biology with engineering control, lifecycle QC, predictable recovery, and an evidence package that allows another lab to use the model with confidence. That, in my opinion, is how we move from promising organoids toward decision-ready NAMS. Thank you.

SPEAKER_01 49:35

Perfect. Thank you very much for that. We've got a lot of questions coming in, which is fantastic. We're, again, I'm going to save that for the end. Some of them are specific, but I know it's going to apply to everyone, having done this long enough. So keep them coming. So last but certainly not least, we have Alex Savier. He is sales business manager at Revivo Biosystem.

Skin Chips Permeation And Wound Healing

SPEAKER_03 50:02

Hi, hello everyone. Thank you, Jenny, for inviting us today. Happy to join the conversation. You'll see we we specialize in two things here. We have a microfluidic platform that we use with this type of chips, the size of a roughly of a credit card. And we create as well our own 3D reconstructed tissue. We focus mainly on barrier tissue and we focus on skin and dermatology application. And as well, we are developing some more advanced technology called fiber, which is a cell engineering technology to be able to track any biomarker and to allow the cell to release a fluorescent biomarker for any proteomic and transcriptomic activity. Today I will focus a bit more on practical application that we have with our hardware M3D tissue. And you will see we work with a different type of industry, with of course the pharmaceutical industry, but the cosmetic industry and as well the chemical industry and the medical devices industry. So we have different regulatory landscape, a different level of maturity, and I will go through that. A bit of workflow, so that's the way we work. We place a tissue of both chips, and the platform uh is connected to a pump system, and we have a constant dynamic flow to perfuse the tissue. This allows us to collect any biomarkers released by the tissue for longitudinal study and for uh very precise kinetics, and then we have as well a software solution to generate efficacy curve and efficacy data. And as well, so with this workflow, we have developed ref skin, which is our proprietary 3D reconstructed skin model based on the co culture of keratinocycle and fibroblasts. We use the human primary cells. Um and here's quite important that the collagen is 100% animal free and produced by the human fibroblasts. Now we'll jump into the different applications that we have. You see, so one of the big things that we do is we modernize in vitro permeation study to understand how a drug penetrates and permeates through the skin barrier. So we use our platform Relego, which is an automated and programmable supernatant sampling and collection device. We mimic the physiological condition because we control the temperature, and so we we can mimic the body temperature and we have a constant nutrient supply going on the basal side of the tissue. This is uh so it's a unidirectional flow driven by a pump, and this allows us to collect anything that is either released by the tissue or that penetrates permitted through the tissue for a topical application, and this is automatically collected on into 96 volt plates on the machine. So we allow the user and uh to gain in operational efficiency because you don't need to be there to collect a different time point, the machine does it for you automatically, and then you can come back 24 hours later, 48 hours later, to collect the L weight and for the analysis. And for basically a web kind of curve, this is uh just for illustration, it's a caffeine permeation. We we use different types of skin model from proconcerative tissue uh to uh synthetic membrane to human biopsies, even to porcine skin as well. And we can see actually that animal model here doesn't necessarily behave well, so it's not necessarily an even a good model. And I wanted to illustrate this this case study with a regulatory landscape. So for this one, we are in a field that is very mature, and those types of uh flow-through cell operators, so microphysic chips, are already part of the guideline. So there is the OECD 428 regulating this type of study, and the it's part of the guideline, and for the USP chapter 1724, uh it has been updated recently last year in the in the latest iteration of the guideline. So here we're in a case where the regulatory framework is quite clear, those types of technology, microfluidic chips, have now been recognized by the by the regulators. Um less tricky in terms of regulation. You will see the next case study is a bit different. So we work as we focus on dermatology, we have developed um a wound healing model. So it's based on our 3D reconstructed tissue. We create a standardized wound on the epidermis only, and it's a self-healing model. It takes four days for the model to heal. It's based on the cell-to-cell communication between the epidermis and the dermis. You can see here when we get an histology slide, we have this very neat and clean wound only on the epidermis, always three millimeter diameter. It's a highly reproducible and high throughput. We can generate up to 50 wounds in in less than 10 minutes. So we all for screening of different types of drugs or dressing medical devices. And what we do in in terms of analysis, of course, histology is one of the endpoints to follow the healing and the repetitalization of the wound, but we track as well uh the biomarker. One of the key biomarkers here is, for example, the interlockin' 6 that uh drastically increase once the wound is generated and then decrease over time as the as the wound starts to close and the carotinocyte start to migrate. Uh we do as well the thing that is proper to our uh macrophytic platform, we use caffeine, caffeine permeation as a functional test to assess the barrier restoration. So it's very easy to understand when there when the barrier is compromised, the caffeine diffuses very fast through the through the skin, and when the barriers start to to to close to reform, we have a lesser and lesser uh speed of diffusion of the of the caffeine. Here it's a very different regulatory contest for wound care. There is no uh nothing related to a CKC study. This the the main market is mainly related to medical devices and to dressing. And the only regulatory guideline that exists is this ISO 10993, but it's only for irritation tests replacing a rabbit model, and this has been done already a few years back, and it's using 3D reconstructed tissue instead of rabbits. So for efficacy studies that what we do here with this model, it's it's a blank page, and we are trying to build a new standard. So we are working with some leaders in this market of woon care and and and dressing technology to make it as the next standard for efficacy studies. So this was the second application, and the third one you'll see it's

Aging Models And Regulatory Reality

SPEAKER_03 56:43

totally different. As we work a lot in dermatology, we have a lot of clients specializing in cosmetic products. So this is a model that we have created leveraging on microfluidics, and it's a very unique way to age the skin tissue. So in general, to age the skin tissue, we use traditionally UV exposure, uh, but it's a very limited way to mimic aging. Uh what we do here is we expose the skin through the microphone flow to a hormonal supplement for seven days, and we can see that when we explore we expose the tissue, we gradually age the tissue and see a change, a shift in almost as this of the tissue. And this is how it looks like. First, when we do an histology, we can see that the the thickness of the tissue has drastically decreased for the epidermis and the dermis. Key biomarker expression, like A67, which is a proliferation biomarker, has drastically decreased as well. And linked to the shrinkage of the of the dermis, we can see that collagen-1 synthesis has drastically decreased as well. But this this model, as of now, we use it mainly for cosmetic purposes. So a bit of context for cosmetic research, there is a ban which is owned for more than 10 years in this field. Uh, there is a ban on animal models, so all the cosmetic companies they have to find alternative solutions. And when they want to claim that this product has a given efficacy, in terms of regulation, there is this EU regulation 6555, and it's very open, they just the the the company just need to find the right model to prove that uh the the the c the that their their evidence is uh is there for this specific tissue, for this specific product. So so yes, so that's what we use it mainly for cosmetics, but what is interesting recently in the in the past few months we see that this type of uh microfluoridic stimulation and exposure to age tissue got as as well traction with other types of of field, and we got some uh universities who want to try it on other types of tissue more for pharmaceutical application. So we will try as well to build some case study for pharmaceutical research using this approach. So, yeah, so that was my my presentation today, just to tell you that we can see for us it's really a case by case. We are working with different types of industry, cosmetic, medical devices, pharmaceutical, chemical, with different types of a regulatory framework, different level of maturity. Um when there is no clear regulatory framework, we are trying to build it with a key market leader or with an academic institution.

SPEAKER_01 59:20

Thank you. Fantastic, thank you, Alex. Another great presentation. So I think that is everyone.

Panel What Changed In Five Years

SPEAKER_01 59:29

Um a lot of really good questions came in and um some specific, you know, some some general, so obviously this stimulated a lot of thought, conversation, and curiosity from people. So what I'm gonna do, uh, because some people's specific questions came in, you know, a couple presentations later, I'm going to kind of weave them into the discussion because I think pretty much every single one of them is is relevant for everyone else. So I am going to start broadly, ask you all some questions about the field, and then we'll move into some of the specifics. A lot of them weave us into topics that we were going to talk about anyway, reproducibility, scalability for where instance. Uh, and then we have some more kind of general topics topics, excuse me, at the end. So to start succinctly, not that it's an easy answer question to answer, but for everyone, what does your organization do today that couldn't be done five years ago? So, what's the the first thing, I guess, that that comes to mind uh for each of you? And I can kind of go from person to person, whoever wants to jump in first and and you know, make sure everybody gets a chance to to weigh in. Anybody want to volunteer? Uh I can go ahead. Go ahead, fair enough.

SPEAKER_00 1:00:46

All right, no, it's uh yeah. So for us at Bioprinting Lab saying today we can basically combine 3D bioprinting, the mult 3D bioprinting, like pump-free dynamic perfusion and in-seo imaging, along with on-plate cryo preservation, basically in a multi-wheel format that that can just like move these promising organoids to like SA ready, like uh decision-ready NAMS that was not possible like five years ago. So yeah.

SPEAKER_01 1:01:25

Perfect. I like that that combination is is critical. Graham, are you hand something as well if you'd like to jump in?

SPEAKER_06 1:01:31

Yeah, for sure. So, you know, so a voxel we've built at these 3D bioprinted tissue models, but importantly, they're vascularized. So we have functional vasculature within these tissues. And so we're able to perfuse them uh without the need for any peripheral pumps or any equipment, which you know speaks to the adoptability issue that I was I was speaking to earlier. And so we're trying to really solve problems that before were difficult for people, questions that were difficult for people to answer. So, you know, I mentioned target engagement in my presentation, which is a real key one for us. So, you know, people would go away and make these transgenic animals to see if they could engage that target in a rodent. And that's the kind of thing that now that we we are bringing to the market for our partners to help them answer those sorts of questions. I'm gonna hijack very quickly the screen if you don't mind, because I didn't kind of get to show this in my presentation. But this is an example of one of our tissues that we've endothelialized, and we have a prostate uh cancer cells incorporated into our tissues. And in this case, we were able to take uh cell therapy from our partner and actually show them that by perfusing the tissue, we could observe their targeted therapy penetrating the vasculature going into the tissue and engaging with a cancer cell, which you know, five years ago would not have been possible to have this level of like human-relevant validation that your cell therapy was working so early on in the development process. And that's that's given them a lot of confidence to uh to move forward to their program. So, you know, that that's one of the big changes I think that NAMS can can bring.

SPEAKER_01 1:03:06

Nice. Very, very important. But anyone else? I don't want to just call people out for the the sake of calling them out, but Andrew Mike, uh Yeah, I I can go next, uh Larry.

SPEAKER_07 1:03:16

Thanks. You know, Axiom, you know, obviously we're we're very interested in new models as as well, you know, organoids, microtropic systems, etc. But I think one sorry about that.

SPEAKER_01 1:03:28

Yeah, go ahead, Mike. That was a timer, but um they'll give you some more time.

SPEAKER_07 1:03:32

No problem. One of the things I think the the the biggest advancements I think relative uh to this kind of conversation is that push towards more kind of regulatory facing. And I think that's something that hopefully the the NAMS field is is starting to see as as well, is is that you know those it's that additional confidence, right? That you're willing to take that data now and use it to support an IND, you know, not just having confidence internally, but in externally to justify those decisions you're making. Perfect. Thank you.

SPEAKER_05 1:04:05

Yeah, and I think for us it's a true fit-for-purpose design of NAMS. It's going from kind of the end in mind, talking to the customer, understanding what their specific need is for you know that exists within a tissue that needs to be created, um, rather than you know building in every single feature that then we might think recreates biology. And then, you know, understanding what is the true disease or that they're trying to mimic, or is it a liability that they're trying to investigate, and then um being able to match it up to our platform and technology where we can incorporate the right cells, the right ECMs, the right structure, geometry, density, and then because we have that capability, we we can match up the two and then truly leverage I think that that connection.

SPEAKER_01 1:04:56

Nice, thank you. Alex, I'll give you a chance and we have plenty of topics, but if you Yeah, go ahead.

SPEAKER_03 1:05:02

I think for us at Rival Biosystem, it's another application that I didn't show, but we work a lot with um skin biopsies coming from a patient from abdominoplasty. And actually, thanks to this dynamic flow when we place the biopsy on a chip, we we can keep the tissue alive for up to three weeks, which was not feasible five years five years ago because um when you keep the tissue on a static culture, it starts to die after six to seven days. So it really changed the framework for our experimentation, longer exposure to to active substances, repeated exposure. So this is one of the key uh takeaway approaches. Very nice, very nice.

SPEAKER_01 1:05:39

Okay, so one more broad question, then we'll we'll get into some specifics. This one is around understanding where you all are seeing adoption occurring the fastest, and let's say anywhere you might see that it's lagging behind the most. And I'm excited to hear about the the answers to this because I know it varies from technology from area of biology. And so look forward to each of your questions. And just to frame it, you know, this can be anything from the stage of drug development that you're seeing, particular applications, disease areas, just wherever you're seeing the most traction and kind of the biggest opportunities, but that are lagging.

SPEAKER_05 1:06:16

I can I can go. I think we're, you know, we're see for us, we're seeing, you know, we worked with a number of different partners and we're seeing across the board, you know, lots of different areas. So in safety, in vitro safety pharmacology is one classical example of how we might be able to apply NAMS and the drug development process. But then I think more recently some of the external innovation folks at large pharma have been reaching out, engaging with us. And that's where and that could that really also range across many different functions within you know pharma organizations. But you know new drug targ new drug development is is I think one of the key areas that are really starting to understand how we might be able to engage with NAMS. And then and then on the other side of disease modeling being able to replicate some of the you know pathophysiology of diseases that's I think that's also an interesting area where say for example a company might have you know specialty engineered disease cells or they might have you know drugs that that target you know targets specific uh disease pathways then they might be able to engage with someone like us or you know anyone in this field where you know we could recreate a disease phenotype and then have them utilize that as a way to study the disease. Thank you.

SPEAKER_01 1:07:39

And so yeah plenty to get to if no one else has specific thoughts there but jump in if you do give you one more second. Okay. We'll move into more detailed questions or more specific questions, let's say but as I mentioned before these are broadly applicable.

Adoption Where It Moves Fast

SPEAKER_01 1:07:55

These were topics that you know I really wanted to dive into with you all as it was kind of moving into reproducibility being a part of this and the other spectrum is physiologically uh physiological relevance. So those can be a push and a pull from my experience the more complex sometimes the harder to reproduce. And so let's start with physiological relevance. Andrew this one was directed towards you earlier you described a roadblock related to ensuring physiologically relevant cell density or just getting to a better cell density. How did you solve that?

SPEAKER_05 1:08:34

Yeah I mean I think it's relatively simple. We we simply take a bunch of cells and put it into syringe and we print we print it. I think it's no more complex than that but and that primarily because the reason we can do that is primarily derived from the fact that we can we can take this sort of almost a pellet of cells that we're printing and be able to deposit it into a support material that holds the cells in place while gelation of the material takes place. So it it's it's kind of a function of the technology that we're using. So be able to then you know deposit and work with these high density slurries of cells that previously really there's no other way of you know of working with because you would have to pipette if you were to try to pipette this it would all get left behind in the pipette and then you also don't have any way to deposit or handle it you know once you're trying to get into the tissue that you're trying to create. So yeah it's it's really just concentrating them and printing with them with their technology you call it you call it simple.

SPEAKER_01 1:09:39

I'm not sure if that's the word I would use but so yeah for others I'm sure you know you also had the same problem. Maybe problem is the wrong word because this was the whole point of what we're addressing. How did you ensure your cells you know represented the model appropriately could be cell density again organization morphology what were some of the big things that that you had to do to ensure that I think for us um you know we've been doing a lot of work in the sort of the novel therapeutic space for things like T cell engages, bi specifics.

SPEAKER_06 1:10:10

And so this that question of cell density as well was incre uh you know very important for us. You know, have you got the right number of T cells actually inside your tissue so that you can actually go and show a cytotoxic effect, for example. So you know optimizing that, looking at reference uh just the actual physiological data from human tissue was important. But also things like measuring a stiffness. So so we bioprint our tissues and and sort of tune the ECM to make sure we're matching the stiffness of of the tissues that we're trying to model as well. So there's a there's quite a bit of iteration needed to to get through that but you know and and trial and error but that's that's all through the optimization process that we do on each model that we develop.

SPEAKER_01 1:10:51

Perfect what about anyone else? Anything particular that comes to mind?

SPEAKER_03 1:10:55

Yeah just for for us quickly as we work with with skin and skin is a is a barrier tissue our our main focus was to to recreate this barrier to have the ability to have a quite good barrier function, something as close as possible to a skin biopsy. So we've worked a lot on that and we're not too far from the skin biopsy now. So we have a very good model from pharmacology and to understand the topical application of of drug.

SPEAKER_05 1:11:20

And I guess I'm also interested in asking the other panelists, you know, how I think we've we've gone a long way in creating different lineages of cells, right, whether it's IPSCs or primary cell types. How I guess in your field of interest how do you feel like are we ready in terms of our cell development process to then translate what we're putting what the cells are we're putting in and you know how will they perform once they get into you know clinical studies or actual field of use. I love that ask each other questions. Go ahead I like that idea.

SPEAKER_07 1:11:57

I think it's a symbiotic relationship right between uh the platform and and the cells and so the system is only as good as as each of those those components. And you know for the two examples that I used I mentioned the the Fuji cells because they're very well validated, right? And then you know that you get the same result every single time. And so for us that's been very important because you know what when we're talking about the complexity of training new operators and and looking at the same result over time or in different labs, you don't need the biology changing on you. That's so I think it comes back back to a little bit what Larry was saying about that trade-off between complexity and and it's almost like how do you know when good is good enough? Right? And and a lot of the time I think people will trade off perfection for I know it's gonna keep working every single time. Right. And and so I I would say for where we are at the moment for like the pro rhythmic essays and the seizure essays we're in a pretty good place, I think, for the type to but that doesn't mean that you know there won't be developments in the future that is going to mean that there's going to be uh new cell models that can answer different questions. But I think the to come back to your your question, I think the the quality of the cells are absolutely fundamental to to to answering the question.

SPEAKER_06 1:13:16

Yeah I I agree I think you know we're at the point where we're getting

Physiological Relevance Versus Reproducibility

SPEAKER_06 1:13:20

you know useful data to help people make decisions. But you know I think over time we'll see more questions come up around you know the representativeness of cell lines to actual patients. And I think you know there is a world in the future maybe where this can be taken in a bit of a sort of personalized medicine direction as well where you can you know can actually generate a NAM from a patient and help determine you know the right treatment for that particular patient because we know that you know cancer phenotypes you know differ. Like there's a lot of diversity. There's not necessarily one specific treatment that's going to work for everybody. So yeah there's a I think we've got some of the early questions we've made fantastic progress of some of the early questions for NAMS. Still a long way to go. Perfect.

SPEAKER_01 1:14:05

So I will say there are more very specific questions for the panelists. Jenny are are we going to be able to share everyone's contact information you know at at at a certain point you know they get specific enough I think maybe one-on-one uh outreach would be would be more helpful is is that possible Jenny not through us the the speaker if they're open to share their email address and you know want to get in touch so you you're welcome to share that in the in the chat box.

SPEAKER_04 1:14:35

But um and also their LinkedIn is already public anyways so so if you have very detailed questions you can contact them on social media or email if people are open to share that contact. And so yeah it is really up to the speaker right now.

SPEAKER_01 1:14:52

Perfect. I just want to make sure as much as possible everyone gets their their question answered so that that works. All right so moving on to a topic that we've had a lot of questions about is standardized standardization.

Standardization Who Must Drive It

SPEAKER_01 1:15:03

So you know this is often cited as as one of the biggest you know hurdles blockers do broad acceptance of this couple questions I guess starting broad who do you all think should be the one driving this? And I think that can be both from the physically proving it perspective and then the uh you know to use something Mike said when is good enough uh you know when is good good enough and so that can be regulators and users you know there's some consortia vendors themselves you know give me give me your thoughts on that yeah I I would like to jump in on this one.

SPEAKER_00 1:15:40

So uh and from our perspective and uh when when we talk with all these different like uh partners the users perspective so it's like every pretty much everyone's responsibility in terms of let's say the developers or the vendors needs to basically qualify their their their process right so the process that they use to manufacture develop it it all needs to be like accusated and then the in terms of the users or the customers perspective they need to define like the the performance criteria basically like okay what what type of data what type of functions will be enough for them to make that decision or to move the the the program or the the drug forward and then that's where all these like the the consortium comes in to basically standardize the because there are like multiple different platforms available for the same type of work. So this cross platform standardizations or cross cross platform workflow integration needs to be considered and then yeah definitely the regulatory from regulatory standpoint they need to as we came up with this FEA draft guidance they need to come up with certain specific guidelines that will help everyone to just like come to one like basically a simple conclusion to move forward. So it's it's like yeah everyone needs to contribute to a certain extent towards this standardization.

SPEAKER_06 1:17:15

Yeah I I agree with that I think there's potentially a bit of a trap for drug developers if they're thinking that they're gonna sort of wait for the perfect NAM to be validated and approved by the regulators and then you know jump on that one. I think what this needs is sort of users to actually tell us what the specific question they're trying to answer and then developers like ourselves actually making that model and showing that it can you know reliably answer the question and then the regulators saying okay yeah we accept you know this this approach for this context of use and and that's you know that that's how I think fundamentally we'll we'll get acceptance but like I say I think waiting for one particular method to be approved and and and moving with that is is kind of a potentially a trap. It's gonna take a bit of from everyone. Yeah.

SPEAKER_07 1:18:05

So I I covered in my talk at the beginning about the CIPR initiative and I think uh the the HESI program provides I think a blueprint for other uh particular kind of research areas and that you know it was a significant problem and so it was able to attract you know pharmaceutical companies, CROs, academics and importantly you know the FDA, the regulators and bring them around the table and define what the problem is and on what how well the it wasn't called a NAM back then, but you know how well how well does the assay have to perform to be able to be useful, right? You know what and now it's called kind of context of use right and and so I think those kind of consortia are really really important because if you're coming at this as a technology provider and you're trying to guess you know what the customer wants is going to be very very difficult. And so I I I would also say I think a lot of this is we're learning together, right? The industry is learning with us and and we're and so getting everyone around the table I think is the fastest way to learn.

SPEAKER_05 1:19:08

Mike do you have any uh sort of insight of how the SIPA initiative all came together? Because I I'm well familiar with you know with the SIPA initiative is just kind of how is that a conference? Is it just there's a couple key stakeholders that started the conversation.

SPEAKER_07 1:19:23

Yeah so it it really started I think back in 2012, 2013 and there was this kind of idea that there was an overemphasis on one particular IN channel and that maybe the that focus although it'd been very good at screening out particularly prorhythmic drugs, it was also potentially weeding out potentially useful drugs unnecessarily. And so from that point of view there was an appreciation in industry there's an appreciation at the FDA and so they kind of brought people together guided by and just come and try and answer that that problem. And so I think it was an issue that industry appreciated needed some some help and and that was kind of the catalyst but it was very useful having a group like HESI to to to kind of pull people to together. So I think having that you know major problem is is is is something that helps bring everyone around the table right the clearer problem definition.

SPEAKER_01 1:20:22

So continuing along the lines of standardization kind of getting more into the you know how how the where the rubber meets the road let's say this person's asked a question specific to cardiac organoids but I think this is for everyone really in this person's experience they can generate very consistent cardiac organoids when they do the protocol but when another student does it using the exact same protocol exact same reagents there's often differences. You know how have you all dealt with that you know how did you you solve that maybe you know what any thoughts you have on that because I know that is a very common issue.

SPEAKER_00 1:20:59

Yeah so uh we we have experienced that a lot and when we try to commercialize this this technology the the the platform and so it's it all and we we're still working on that it's like it's never ending. So the what we try to do is try to because we know this like the operator to operator manual handling it it creates a lot of variability. So we try to reduce as much as this manual handling possible with like if you

Operator Variability And Automation Fixes

SPEAKER_00 1:21:30

can incorporate automation in and basically trying to also we try to simplify the the workflow, the the SOPs because if there are like multiple steps that the users need to do from taking the model to plating them to moving them in a different system for imaging for all kinds of analysis then that creates a lot of variabilities and reproducibility issues. So we are trying to deal with that or address that in terms of like okay how can we simplify the workflow? How can we incorporate some of this like automation to avoid the the manual handling and then also in terms of like okay how will like how standard or how like QC validated system we we can create so that the the users and the the from perspective of this SARADI model is was our this was our biggest motivation to even think about the the SRAD system because we hear this every time from the user that like okay the same protocol doesn't work for them. And so how can we address that? So if if we can just like provide this SRAD models which is already like well validated then that will solve a lot of issues. But even if not towards that if if it's just about the the protocol and trying to do that. So yeah simplifying the the workflow is is I think the the biggest considerations or the biggest factors.

SPEAKER_06 1:23:04

You know for us for for for our models you know vasculature is incre uh like a very important component is like the center of of what we're doing with our models and and uh Carolina and our founders sort of recognized early on that you know having consistency in the vasculature would be very important for any data that we generated in our tissues and and their approach to this was to you know typical engineers you know they do things very consistently and um in a standardized way was to really apply robotics to this. So that's a bit part of the reason why we've gone down the very controlled 3D bioprinting it's a two PP system that we've developed at Boxel which just means that that vasculature is very consistent all the time and it helps us to address that that reproducibility issue. And you know throwing VEGF for example onto a tissue isn't necessarily going to give you a consistent structure every time but we know that you know with high resolution bioprinting we can get that consistency.

SPEAKER_01 1:24:01

So then another aspect of this is quality control because I think on some level right the reproducibility is is going to happen, let's say and so how have you all incorporated steps towards ensuring that I know one of the presenters even said you know you want to scrap things early if it's not working. So I guess kind of two two sided question number one, how have you all addressed that? And then also you know when you're talking to customers of people using your model with their particular assays maybe how do you tell them or how do you suggest that they address quality control so we've built that into our assay right so we're measuring spontaneous cardiomyocide activity and spontaneous neural activity.

SPEAKER_07 1:24:45

And so we can tell people who are using our assay you know we've defined what that should be right what does it what does activity need to look like to tell you it's ready to start right or potentially now's the time to to stop. So I think that's fundamental with these kind of new more complicated technologies because the endpoints are going to become very sensitive to the starting uh condition. So the more that you can build that quality control metric into the start to decide whether it's good enough to start, I think that's going to be very important.

SPEAKER_00 1:25:17

Yeah I I totally agree with Mike so that's what like in in our case and in my presentations I I included that specific uh the the QC uh workflow the QC gate that because from our platform perspective we are not like creating a specific organoid model but we are creating this entire like the integrated workflow that the users can use. So we we try to include this QC criteria, QC gates in from the beginning, from the the starting cells to the differentiation process to the assays, the the imaging part so that we make an informed decision before we move to each and every stage. And so avoid those because especially for organoids and there's like complex in vitro models it takes weeks and months of process culture. So if in the end you you identify some issues then that's like yeah you you you're wasting a lot of resources. So yeah from the or if you implement this from early on that's uh much better.

SPEAKER_06 1:26:24

Yeah I think it's you know the infrastructure question is key there. You know you know for us it's you know every tracking and logging data around every tissue that we produce. So you know what would who was the operator of the bioprinter that day what was the you know the cell aligned lot that was used what was the the media what was the bioink like we've we spent a lot of time and effort sort of building that QC system into our process so that we sort of have end-to-end qc control and that that's that's really critical because it also helps you learn as well like if if something does need to be optimized, you know what you know controlling elements is important so you're not just sort of shooting in the in the dark trying to you know problem solve.

SPEAKER_01 1:27:05

I like that a lot you know I think my first thought on QC is testing the cells are they you know the right phenotype are they uh you know firing at the right rate but your point about track it every step of the way if there is a failure failure you can go back to that I I I like that. All right so one last specific question and then we have some more broader topics to talk about.

Contamination And Process Traceability

SPEAKER_01 1:27:30

So this one is highly specific I I want to ask this because I think unfortunately maybe we've all dealt with this and it is an incredibly painful thing. Contamination control maybe this pig piggybacking off of QC you know when someone does this and you know spends a long time on it and then they get contamination maybe only in a few wells, right? Not in all of it. You know, what have you all done to address that? Um you know is that something you see is just a little bit Inherent. Um, you know, what are some of the sources for this and and your approach? We could also flip that. Okay, go ahead, Andrew. I think you came off mute.

SPEAKER_05 1:28:08

Yeah, I mean, I think it's unavoidable it happens sometimes. I think you you want to obviously minimize the risk as much as possible. I I think a lot of times it's difficult to difficult to find out. I mean you might want to figure out where contamination might come from, but it's quite challenging to go through every single reagent and every single step of the process to try to find out why contamination happened. So I think it's it really piggybacks off of having the right process, having a standardized process, well-trained personnel. And then when it does happen, I think you just have to accept it and kind of move on. Otherwise you'll be spending a lot of time trying to figure out what happened. And then you I guess the way that we resolved it is we go with brand new regions, kind of review some of the protocols, and then just hone in on some of the key steps that contamination might happen and pay some more attention to that.

SPEAKER_06 1:29:05

I think that also speaks to the sort of the QC infrastructure that we spoke to in the last question as well, because that gives you then the the traceability to try and figure out where that contamination is is coming from. For us, the way we've bridged it is really containment. So, you know, what our our tissue handling is in environments that are as clean as possible to sort of minimize the risk of any contamination and uh you know having the right uh you know air handling systems, etc., and containing you know, keeping the rooms clean is is really critical, critical for you. Yeah.

SPEAKER_05 1:29:38

And I think just going back to the previous question, I you know, I think we probably spend as much time looking at our QC set of data, like our process tracking data set, just as much time looking at those as our actual, you know, functional outcome data set. So I think it's both are critically important.

SPEAKER_01 1:29:58

Perfect. Thank thank you all for that. All right. So towards big picture thinking, uh, and want to hear, you know, from all of you, I I I think everyone will have a nuanced thought on this, even if you might not think so. You all

Why Organ Chips Still Lag

SPEAKER_01 1:30:12

see different aspects of this. What is the biggest caveat you all see today that's keeping organ chip out of routine preclinical workflows? So I'll keep it broad, but you know, this can be anything science, costs, internal politics within pharma. What what do you see?

SPEAKER_03 1:30:29

For us, it's really changing habits, how people work. When they are used to work with a method, it can take a lot of time to convince them to uh to shift to to uh to another uh type of uh ethical methodology. As we work with different industries for us, with chemical, cosmetic, pharma we see there are different dynamics. For example, the the cosmetic industry likes to try new things because they they want to claim uh new efficacy on a yearly basis. So we can see that the adoption is a bit faster on that field. For pharma we can see they're more conservative, which is uh totally normal because people's life is uh at stake. So yeah, different dynamics depending on the on the industry, but it's really the adoption and the uh the effort made to educate the the the user.

SPEAKER_00 1:31:18

Yeah, and so like it's it's from from our perspective or for from my perspective, working with this like uh complex in vitro models and in terms of the users for this, like what we have come to know is this workflow integration, the the operational challenges in terms of implementing certain models into their infrastructure is is like one of the the major issue or other factors that that plays. I mean science, yeah, it it's like definitely uh it can be excellent. Price is also relevant, but then what we have seen throughout the the this process is yeah, the when it's not compatible with their existing infrastructure, then that's like the the kill. So uh it it needs to be aligned with whatever the infrastructure they have, and as well as like, okay, if this new model is going to just like answer some of the the questions to for their need.

SPEAKER_01 1:32:21

Okay. Uh maybe uh another variation of that for anyone here, you know, how have you seen your champions perhaps that you're working with? What do they need to do to convince their higher-ups? You know, where have you seen this work? Maybe it's data, you know, maybe it's something else, and where has this perhaps broken down?

SPEAKER_05 1:32:41

Yeah, I guess commonly it's data. I think we as scientists, I guess a lot of our counterparts are scientists, and usually scientists are data driven by nature. And so, you know, I think that's kind of a double-edged sword. It's it's nice to demonstrate data, but also at times, you know, more data incurs questions of, well, can you show me what's next, right? Can you show me more data? So more data might be getting more data, and then it might become a pretty prolonged process of trying to convince our champions higher ups. I think, yeah, I I think maybe the other the other key factor really is just I think enthusiasm and true understanding of of the technology from the developer side, right? So from the champion, understanding what is what what is a value that the developer is is proposing and then having that the enthusiasm to then convey convince to go in and really working at to convince their you know their their higher ups or their internal team that this is a technology worth adapt ad adopting. I think it's a to me, I think it's a pretty nebulous uh kind of you have to tread a pretty fine line between between data and then also providing a compelling story and understanding of of what it is that we're we're offering.

SPEAKER_06 1:34:00

Yeah, I I think to add to that, I think the other thing, you know, I I've I've been fortunate to work on the other side of the table and and work within external innovation in biopharma as well. And so what I recognize is that that champion has to be willing to put their hand up and go internally to leadership and say, hey, we should go and spend X hundred thousand dollars on a proof of concept study with Company X. And you know, if that doesn't yield good results, you know, that that's a risky career move for them. And so they have to be very convinced that the use case data is there and it's gonna work, um, and that takes time. But there's also a bit of a risk of conflict because are you prepared to deal with the situation of this new technology coming in and giving you a different answer to those animal models which have just become embedded for years and years. Um, and then that takes a whi a while to recognize because the fundamental thing everybody needs to recognize is that you know, when when 95% of your programs are failing in clinic, you know, there's gotta be a clear message, a clear signal that the tools you've got preclinically aren't doing the best job at predicting that. So, you know, I think we need uh we definitely need more champions within the industry, but you know, in order to convince them, we also need very specific use cases and and compelling data.

SPEAKER_05 1:35:15

Yeah, and we've also heard pretty directly, you know, we were working with some of the in vitro extra innovation teams, and some of their feedback is, you know, love to share this with our internal team, but our some of our in vivo teams have a process and workflow that they are very familiar with, and it is going to be difficult to convince them to move away from that. You you would need to be pretty compelling, and how can we how can we together craft a compelling you know argument to share with those counterparts on the in vivo side of the of the of the table?

SPEAKER_01 1:35:50

Anyone else? I think that covered quite a lot. All right, so a question

AI That Helps Versus AI Theater

SPEAKER_01 1:35:55

that I don't know, have to ask about AI. So everyone is using it, right? Everyone has AI, it's in all their decks, it's in all their websites. You know, where do you see that this genuinely adds value? Where are people truly using it uh, you know, to to do something innovative, not just incremental? And where do you see that it's maybe just decorative, right? Love to hear your thoughts on that because everyone's doing it apparently, but I don't think that's the case.

SPEAKER_07 1:36:26

So we've had users taking our data and using it in machine learning models for several years now. And I think to me, the biggest challenge here is volume of data and consistency and quality of data, right? I think that's gonna be the biggest challenge, I think, for the for the field is collecting those data-rich quality data sets to make good models. I think that's gonna be so I I would say I think people are already doing it, and I think it's just gonna continue to grow.

SPEAKER_00 1:36:58

Yeah, and uh I I agree with Mike and just like would like to add that so as we move towards more like complex models, so we will be generating a lot of data that might be like difficult to reproducibly interpret by a scientist or human scientist. So in that case, like we see in our case, we see the use of AI in in in that stage and in that context where you generate all this like multimodal data that needs to be analyzed and interpreted in terms of like okay, how it can help us predict those clinical outcomes. So yeah, and if it's just like used from the beginning, then I think like, yeah, I as you mentioned, it will just be a decorative addition rather than getting some use out of those AI models.

SPEAKER_06 1:37:53

I think AI's got you know tremendous potential for us to you know discover new medicines and make advances in medicine. But I always come back to thinking of it as like a you know, it's a multivariate regression analysis, and my data helps me get a better R-squared value and better predictive value from my model. Um, you know, it's still need to do experiments in the lab and actually prove that those predictions you know work out. So I think another area where NAMS can benefit is is you know quickly turning around data and testing those hypotheses and shortening the learning loops for people working in AI-driven drug discovery.

SPEAKER_01 1:38:29

Yeah, I I kind of agree, you know, data analysis, analytics, processing is definitely a big one, you know, being able to pull signal out of the noise. And if you are doing the multivariate aspect, how do you weigh all those together? You know, so it is exciting. And I'm being a little facetious in in you know making fun of it. It's it's obviously incredibly powerful, but you do see instances where maybe people aren't actually using it. But you know, I I I've seen a way where you can also double your workflow, for instance, start doing some of these things. There's an image processing company I'm I'm working with where basically you can do two markers in a given channel, and with AI, you can separate those two markers incredibly accurately. So basically, you know, you're able to half the time of screening with high content imaging, and you know, there are areas like that. So it's exciting to see as it picks up in in more and more creative ways, but definitely have seen a lot of progress on the data processing side. So we're you know, coming close to the end. I would like to give each of you a chance to maybe ask each other questions or you know, make a a statement of something we haven't talked about, you know, your thoughts, right? What you think the most important thing is coming up in the next year, anything to that effect. I want to hear from you.

SPEAKER_04 1:39:48

Can I just say something about the AI front? I mean, everyone here has a product to sell essentially, but I think I'm more on the buyer side. Like just imagine I'm the farmer. How do I decide who should I use and what kind of methodology should I use? I would love to have AI to help me to figure that out. Like some of you guys, you know, help design the whole workflow. That's incredible. But that's just one company. Like, I wonder if I can use AI to figure out who would work the best for me for my intended purposes. That's something I would I would do and and and design for. I wonder if you know, companies like Charles River is already doing that. I'm curious about that, actually.

SPEAKER_05 1:40:33

I guess it sounds like a parallel to that is essentially any investor who is in deep tech might want a tool that helps them understand what is a company worth investing in, right? Kind of uh I think a similar, probably a similar tool exists for for some of these investors, could be applied towards uh you know as a partnership, how who who might make an impact on my on my pipeline, on my process?

SPEAKER_04 1:40:58

Yeah. I mean, I think a related question, and that and that's the the last question I will have for me, for me, is how big, how far do you think NAMS collectively can reach? In other words, how much percentage of current animal testing for pharma can actually be replaced in NAMS? And also relevant question would be how soon? Because that's definitely an investor question. If it's like a little more than a lifetime, probably not for me. It should be sooner than bioprinted organ, I would imagine.

SPEAKER_07 1:41:31

Yeah, I think sooner than that. I think it's it's a difficult question because it did the question I think relates to where the animal usage is taking place and in the process, right? And so there are certain stages where almost uh you know it's a the the animal model has been used as

How Much Animal Testing Can Shift

SPEAKER_07 1:41:50

a kind of a catch-all, right? To to find things that you potentially were gonna miss. But I think where NAMS are very strong is answering very specific questions. So if if along the the way in in a discovery where you're looking to answer a very specific question, that is where I think we're gonna be most successful and getting traction.

SPEAKER_03 1:42:11

Yeah, absolutely. I'm excited. I'm thinking for any systemic safety study, it will take a bit of time when you want to test the safety on multiple of organs. But when, for example, in immuncology you are tracking a very specific biomarker, numbers are already quite efficient. So it really depends on the on the field and uh on the what is the question to be answered.

SPEAKER_06 1:42:31

I think that it's hard to put a percentage on that already, Jenny, but I think when you I think we almost need to look at like phase one attrition rates and say, like, of this, you know, of these failures, what proportion of them could have been avoided if that signal had been seen sooner preclinically. Um and I and I would, you know, I'd I'd I'd put a bet that a significant proportion of them could have been seen sooner in the right NAM. But it's it's gonna be use case by use case to build up what proportion of all the animal test testing can be replaced. I think it's a significant percentage, but I don't think it's you know the the whole organism effect of a drug is also still very important and it'll take a while before we have a you know a human on a chip.

SPEAKER_04 1:43:15

Yeah. Today is the first time I understand why people use beagles as a testing animal. I mean, actually, the main reason for animal motto is because 1937 there's this drug called elixir that killed tons of children, and that's when the regulator decided that we must have some kind of regulatory process to not kill our people. So that's why we had animal models, which is, you know, not something I like either. And also the rabbit skin irritation test, that's something I did not know until today. This sounds horrible. But on the other hand, you know, how how do you protect people? So, anyways, something to ponder.

SPEAKER_01 1:43:56

That's such a good point. I know we're at time, but yeah, sometimes we think, oh, the FDA give us, you know, we want to replace animals. They're there for a reason, right? And and you have to prove yourself before we risk, yeah, something bad getting to the the

Final Hopes And How To Help

SPEAKER_01 1:44:10

population.

SPEAKER_04 1:44:11

Okay. I think anyone has a final request, shout-out, you know, future hopes, conclusion remarks. Now is time. Laurie, I'll I'll start with you. What do you hope for?

SPEAKER_01 1:44:23

My hope, I mean, I think the context of use and and the field thing is such a good part of it. You know, my my hope is that the fields that are leading, you know, liver, some of these applications that have been around a long time, you know, long. I hope we start seeing some discrete wins. And I think we are with ICN, we are with validation qualification network. And so once we start seeing those, look, all the rest of us are working on perhaps more difficult organ systems, let's say brain, right? And those wins, then investors get more interested, and that snowball effect is is what I'm excited about.

SPEAKER_04 1:44:56

Andrew?

SPEAKER_05 1:44:56

Yeah, I think I I my hope is is kind of for the field to I think we're seeing a lot of that now, but for the field to really shift towards the focus of how do we utilize these NAMs. I think we spent the last decade developing the materials and tools and you know platforms that we really I think now is the time to start to focus on, you know, what is the true problems that we need to solve. And then I think more of that open conversation with with people that own and understand the problems, I think that would also be another part of that is you know, how can we connect the builders with the people that really need it? The the the companies, the farmers, you know. And I think that would be that would be great if if we can mash up the two. And then I think shout out to Jenny for continuing to put on these talks.

SPEAKER_04 1:45:45

Maybe I'll create an AI agent to figure out how to match you guys. Graham, uh, what do you think?

SPEAKER_06 1:45:52

Yeah, I think you know, my hope is that you know over the next three or four years, you know, NAMs start to be seen as really part of the toolkit before any, you know, animal testing happens to give people you know the confidence and some of those key questions that I that I laid out. And that is just seen as a as a standard, a gold standard is to to test in a NAM before you you do any further preclinical work. And I hope that pharmaceutical companies are also sort of embracing that mindset as well, because I think it's a lot of value that can be can be brought. And thank you as well, Jenny, for organizing this event. It's been fantastic.

SPEAKER_04 1:46:28

Yep. Thank you. And Alex, final words?

SPEAKER_03 1:46:32

Yeah, um, I guess my hope is to see a greater adoption in the in the next year, the more as well complex model with a multiple organ, even if there are some case studies already happening now, but pushing more on that. And I guess that's all. And thank you, Jenny, for organizing that.

SPEAKER_04 1:46:50

You're welcome. Pranav?

SPEAKER_00 1:46:52

Yes, uh first of all, thank you, Jenny and 3D Hills and uh Laura for for this wonderful discussion towards decision-ready NAMS. And my hopes that like we can collectively work towards getting to the the end goal that we all are striving for. And if we can have this platform, the the AI tool you mentioned where we can that doesn't exist yet. Yeah, the where we can come together and find the right partners to collaborate with. That yeah, that that would be really great.

SPEAKER_04 1:47:26

Awesome remarks. Um, and thank you, Laurie, for wonderful moderating today. There's no way I could have done this kind of level of depth and professionalism. So thank you very much for doing a wonderful job. And this video will be on demand for a couple of weeks for the public for free. So invite your friends and colleagues to check it out. But we will also post process them into little clips, and hopefully one of them can go viral. That's that's my hope.

SPEAKER_01 1:47:52

I like it. Thank you, Jenny.

SPEAKER_04 1:47:53

Yeah, you can help though. You can help with going viral. All right, all right, thank you. Thank you. Bye guys, yeah.

SPEAKER_01 1:47:59

Thanks, everyone. Enjoyed your your feedback.