Mary Parker:
I'm Mary Parker, and welcome to this episode of Eureka's Sounds of Science.
Today I'm joined by Dr. Madhu Lal Nag, chief science officer for InSphero. She joins me to share her expertise on spheroids, which are a promising alternative in in vitro models. As researchers and regulators gain more confidence in non-animal models, spheroids are poised to make a big difference in drug development. Welcome, Madhu.
Madhu Lal Nag:
Thank you, Mary. It's a pleasure to be here.
Mary Parker:
I'm really glad to have you here. I will admit I do not know much about spheroids, so I'm expecting to get a little bit educated today.
Madhu Lal Nag:
Then this should be exciting for both of us.
Mary Parker:
So, can you tell me first a little bit about yourself, a little bit about your background?
Madhu Lal Nag:
Yes, absolutely. It would be a pleasure. So, as you introduced me, I'm Madhu and I serve as the chief scientific officer for InSphero. And I'm trained as a molecular oncologist. I got my PhD at the George Washington University. And I've spent the last 17 to 18 years developing three-dimensional models. So, I definitely have a special love for everything three-dimensional.
Mary Parker:
Nice.
Madhu Lal Nag:
And I've sort of carried that theme on in my research career. I spent nine years at the NIH and three-and-a-half years at the FDA, and then made my way over to the dark side of drug development, and then came home to three-dimensional models here at InSphero, and I'm really, really enjoying it. I think what I really enjoy about my job right now is that as a community, as a market, we are really ripe for using non-animal models now. And I think the community, pharma, biotech, regulators, and vendors like us, we are really coming together to understand how we can move the field further. So, this podcast is really timely.
Mary Parker:
Perfect. I always like to hear that. What got you interested in oncology to begin with?
Madhu Lal Nag:
I've been interested in oncology since I was seven years old. My father, when he was a resident in medical school, he had a seven-year-old that he could not save. He had leukemia. And this is obviously many, many, many years ago when we really didn't have any medicines to treat children. And I think that that affected him so much. I still remember that dinner table conversation, and I think for me it was, I have to do something that somehow alleviates this pain, and it sort of stuck in my head forever.
Mary Parker:
Yeah. No, that's a pretty common refrain amongst a lot of researchers. They have some early core memory of dealing with disease, whether themselves or a friend or who knows, and it-
Madhu Lal Nag:
And it just sticks with you.
Mary Parker:
Yeah.
Madhu Lal Nag:
Exactly.
Mary Parker:
Yeah. And it would have to, to get you through medical school. I think you'd need that kind of motivation because from what I hear, it's a little difficult. So, what is Insphero? What so they do?
Madhu Lal Nag:
So, InSphero, we've been around for 15 years now and we're the global leader in 3D in vitro models. We provide physiologically relevant human in vitro models to pharma and biotechs all over the world, both in the US, Europe, and Asia, to help them make better informed decisions in the drug development process. So. We have a large suite of toxicology models as well as very complex disease models in oncology, in metabolic diseases like MASH and diabetes. And then, outside of that, we are really growing our portfolio into other safety, pharmacology, and disease models as well.
Mary Parker:
And how does InSphero work with Charles River?
Madhu Lal Nag:
So, InSphero and Charles River have partnered together to provide in vitro solutions for liver toxicology in the human relevant, physiologically relevant model that we have. And this truly is our biggest business ... This is our core business, our human liver toxicology model. And I'll talk a little bit more it once I tell you a little bit more about how InSphero and Charles River are working together.
So, one aspect of our partnership is with the human liver model. The other aspect of our partnership is with our cross-species model. So, we also have mouse and rat spheroid models as well. And the idea here, Mary, really is to be able to show that the animal models are not as predictive as the human in vitro models. Everything in an in vitro system, at the end of the day, nothing speaks as loudly and clearly as data, and to be able to show in an entire in vitro system that animal models don't work and human models do. And then the comparison between an in vitro animal system and an in vivo animal and then how that compares to a human being is very powerful. So, we also are looking to collaborate with Charles River over that.
Mary Parker:
That's perfect. I mean, nothing is going to prove it more than data-driven research showing a direct comparison between the two. So, I think that that's really smart plan.
Madhu Lal Nag:
Exactly. And I think Charles River is such a big player in the field. Right? So, if you think about everything that the FDA demands from any biotech or pharma that's developing drugs, you have to do a safety and a toxicology animal model. And most people are going to come to a couple big names in the industry, Charles River being one of them. So, I think what Charles River is doing with their focus on the three R's really, and really their four-R initiative is very forward-thinking, right? Because they're now bringing in these human relevant in vitro models with people like us to show people that you can establish this continuum of testing.
And once the in vitro models are actually more predictive than the in vivo models, they're able to show their customers or clients that you're able to reduce the amount of time and resources that you actually need to perform a preclinical study.
Mary Parker:
Yeah, I like that. So, we've covered some background. What are microphysiological systems, which is a mouthful, according to the FDA?
Madhu Lal Nag:
So, the FDA, you know, it's interesting you asked me that question because the FDA did us all a big favor by making the definition extremely broad. So, the FDA defines microphysiological systems as in vitro tools that model the function of human or animal organs and tissues. So, as you can tell, this is very broad. It takes on spheroids, organoids, organ on a chip system, bio-printed tissues, anything with vasculature flow or anything like that.
So, it is hard because there is no established metric in the community for what a microphysiological system is. There is another term. There is another term that a lot of people in the community use, it's CIVM, complex in vitro models, and this is a little bit more prescriptive, if you will. And this brings together all the in vitro models and delineates them a little bit more than just microphysiological systems. And I think what the FDA sort of wants the microphysiological systems umbrella to cover, depending on whether they're complex in vitro models or not, is to be able to give you better data on disease modeling, especially when a good animal model does not exist. To be able to give you good toxicity data. And then, to really be able to tie the efficacy and safety of a drug to post-market approval, because that's where a lot of data ... There's this gap in the safety data with post-market approval studies. And I think this is where complex in vitro models can also play a big role.
Mary Parker:
I like that, complex in vitro models. I also like it when researchers call anything complex, because then you know it's really complex.
Madhu Lal Nag:
It's really complex.
Mary Parker:
I mean, yeah. So, what are spheroids exactly? When you talk about the different kinds of models, what are spheroids?
Madhu Lal Nag:
Now you're speaking to my heart and it might take us a day to get through this. No.
Mary Parker:
Give me the elevator pitch for spheroids.
Madhu Lal Nag:
I'll give you the elevator pitch. So, spheroids are multicellular 3D structures that self-assemble from a variety of cell types. And the simplest form of a spheroid can form from one cell type, and then, as you increase complexity, you can have a spheroid built out of multi-cell types. But the idea is that you're not pushing them to form anything. They self-aggregate into a spheroidal structure, hence the name spheroids.
Obviously they're different from organoids which are derived from stem cells. They recapitulate some of the tissue functions, but they're not tissue-specific functions. So, what they do recapitulate are the ability of different cell types to interact with each other. They recapitulate certain biological functions that happen because you don't have cells that are fixed onto a two-dimensional plastic layer, which is not the in vivo scenario in any case. And then, they're extremely good at being able to mimic physiological scenarios of the micro environment. Now, that is an extremely important part of what spheroids actually mimic, that you can get in in throughput, that a lot of other systems don't give you. And what I mean by that really is when you think about being able to mimic drug discovery to a particular target organ, you have to think about gradients of oxygen like hypoxia, gradients of acidity. All of those things can easily be mimicked in a spheroid without bringing in the complexity of organ-specific functions that you would have to do with an organoid.
Mary Parker:
That does sound like it would be uniquely qualified to deal with issues of toxicity and efficacy. How close would you say spheroids are to humans?
Madhu Lal Nag:
Excellent question. Right? Because here again, I think I'll go back to what I think the most unique thing about spheroids is. It's number one, they do mimic the micro environment pretty closely and you can use primary cells for this. Right? So, historically, people have tended to think about spheroids as being dealt with with cell lines. And that has been true for a lot of screening propositions and the utility of spheroids historically. But actually, at InSphero, we use very few cell lines and most of our spheroid models are from primary human cells. So, they really truly are a patient in a spheroid, so to speak.
But from the data that you get out of a spheroid, it's incomparable because they really are robust and reproducible. So, instead of getting one data point from one microphysiological system in a 384 well plate, you get 384 spheroids and you get all that data in a reproducible and robust manner. So, I think, sort of the USP of using primary human tissue, so patient tissue, which is obviously as close as you can get to a patient in a 384 well plate, and get all that data from it. I think it gives you a lot of data that is as close to a patient as you can get at, I wouldn't say at a 50,000-foot level, but I would say at a 500-foot level.
Mary Parker:
Okay. It sounds like they might be useful for personalized medicine where you're trying to find the exact right treatment for a particular patient.
Madhu Lal Nag:
Yes. Exactly. And especially when it comes to screening large libraries of compounds, this is where you'd want to go in with a spheroid model, because you can get so much data. And then when you want to validate mechanisms of actions, et cetera, you can use more complex models.
Mary Parker:
So, what niche do they fill in the drug discovery and development space?
Madhu Lal Nag:
In the drug discovery and development space, I think different groups use them for different things. I'll take liver toxicology as an example. Some people use it for go/no-go decisions earlier on. Some people use it as part of a suite of assays to make go/no-go decisions. And then, some people use it to understand the mechanism of action for compounds that have failed further down the line.
So, I think it's interesting how people think about where the utility really is. But for the human-relevant toxicology models, we find a lot of utility in flagging or de-risking compounds earlier on, because when you de-risk compounds earlier on, you fail faster, right?
Mary Parker:
Yeah.
Madhu Lal Nag:
And that's the dogma of pharma, and it's a really good one. On the disease model side, I think the bar is a lot higher, but it also serves a purpose because for a lot of diseases, like with neurological diseases, there are no good predictive animal models. So, you actually do need a human physiologically relevant model, and this is where spheroids form a perfect niche because there is an unmet need where nothing exists. And then you have these human primary cell spheroids that fill that gap.
Mary Parker:
Can you give an example of an aspect of a disease that a spheroid can emulate? I mean, we talked a little bit about liver. Is there anything else you can say on that subject?
Madhu Lal Nag:
Yeah, absolutely. I can talk a little bit. I'll take the liver and talk about how it can emulate de-risking both on the toxicology side but also on the disease model side.
So, a typical example would be once a set of say, of small molecules has gone through a set of screening for efficacy and they've come through from say, 500 compounds to 10 compounds that they want to set through for validation studies. And then, from 10 compounds they come down to five compounds. And then, these five compounds have to go through a battery of ADME-Tox studies. And then, typically this is where they would send it out for an animal model. Right? And that takes time, and it takes time to come back. And the results are not always very cut and dry.
So, this is where you would have compounds being put through liver tox tests. And if they come back with being flagged for a tox concern, then they would go into not just our primary drug-induced liver injury assay, but a whole suite of assays that we have to really pinpoint the mechanism of action of DILI, so to speak. So, it's not just the fact that a compound has or can cause drug-induced liver injury, but the fact that you can also pinpoint using a multi-stage step, the mechanism of action, and then look for a gene expression signature for that particular mechanism of action. So, from a toxicity perspective in the liver model, that is sort of the workflow that we would use.
And from the disease model area, in fact, we have had two companies that have used our system to put forward candidates in MASH for a phase 2B trial, because there was no other model that could actually emulate the disease for the mechanism action of drug that they had developed.
Mary Parker:
Do you have any examples that you can give us of where spheroids have been able to be used as a model where no other model was even possible?
Madhu Lal Nag:
Yes. For one of the areas where spheroids are heavily used that a lot of people come to us for is for type 2 diabetes, because there are no human-relevant models for islets. And we are the only people who do beta-cell proliferation assays and then mechanistic assays in human-relevant diabetes models. So, we actually have a spheroid model for human islets, and we have a very strong islet discovery team, disease model team. So, this is one area where, in fact, I can't give out confidential information-
Mary Parker:
Oh, sure.
Madhu Lal Nag:
... but we actually, we have a few people who will be submitting INDs using our data. And this is precisely because there's no in vivo model or no other in vitro model that can really recapitulate any aspect of the disease as compared to our model.
Mary Parker:
That's fantastic. And a good segue into talking about the regulatory end of things. How are you working to de-risk spheroids to hopefully gain more adoption? I know you talked about this a little by running the concurrent animal model studies. Is there anything else that you guys are doing?
Madhu Lal Nag:
Yes. And it's interesting because the FDA will always be close to my heart, having worked there, right? It's funny to be on the other side of the fence.
I think what has been really heartening to see, and I sort of alluded to this in the beginning, has been that the FDA has opened its doors, so to speak, to the community. So, the FDA wants people to have a conversation with them, they want people to submit data. But the issue has been that people didn't know that they wanted this. So, there is this organization called the Critical Part Institute. They operate out of Arizona. And they've brought together the three big players in this space. The regulators like the FDA, and actually, they're working on regulatory convergence between the FDA and the EMA. They've brought together pharma and biotech, which are basically the end users of these human in vitro models, the biggest end users. There's also academics obviously. And then method providers, like us and other companies that are making these complex in vitro models or microphysiological systems.
And the idea really is to establish a continuous loop, if you will, between these three groups, so that in a pre-competitive space, the pharma and biotech community can generate data and deposit that with the FDA without fear that it'll interfere with any one of their actual drug submissions. And then, the FDA can look at that data and say, "You know what? For this particular indication, this data looks like it's very predictive and it looks much better than the animal model." And then, the method provider that developed this model, to answer this particular question, which is known as context of use, this particular assay or this particular model should be qualified for this context of use.
So, I think this is where we are moving with each model, working with the FDA to ask the FDA to qualify it for that particular context of use. And by that, I mean whatever readout you have from a model or from an assay that a method provider develops, that particular readout is going to give you an answer about one aspect of a disease or one aspect of an adverse effect. And as long as you are able to tie that in vitro readout to that particular aspect of an adverse effect or to that particular aspect of the disease phenotype and establish that context of use, then the assay or the model is qualified for that.
Mary Parker:
God, that sounds like a lot of work being done to get these up to the point where they're going to be more adopted and more integrated into the regular process, which is excellent. Do you think that the industry in general is becoming more receptive to not just spheroids, but to these other models as well?
Madhu Lal Nag:
Yes. I think the industry is becoming more receptive to the idea of them. I definitely see much more adoption by bigger companies. I think they still come at a higher price point, number one. Number two, I think the FDA has to embrace it with a little bit more enthusiasm. And as I said, I think they're ready to do it, but they just need to see more data.
And now, as far as the FDA is concerned, their primary goal is safety. So, if they are not convinced that a model cannot give you 100% ... There's no room to fail. It cannot be predictive 99%. It has to give you 100% predictive accuracy on safety. And unless they're convinced of that, they're not going to qualify a model to be predictive of safety.
Mary Parker:
Yeah, that makes perfect sense. I mean, at the end of the day, the idea is to make things more and more safer and more predictive of their effects in humans.
Madhu Lal Nag:
Exactly.
Mary Parker:
And I think we are getting there. And it kind of leads into my last question of do you think that combining spheroids with other alternatives like organoids and even combining them with animal model data, just forming this sort of super model, will that get all of these closer to regulatory acceptance?
Madhu Lal Nag:
I could evangelize about this all day long. I think, yes. In a nutshell, yes. I think what we really need to do is establish this ecosystem of microphysiological models and be able to demonstrate that this whole continuum of microphysiological models or complex in vitro models answers a particular question in drug development that can be used by pharma or biotech.
And so, what I mean by that really is, if you take it as an end-to-end solution, when you use spheroids at the beginning, they're not the most physiologically complex. They are physiologically relevant, but they don't have a huge level of complexity in it. But they do have a lot of throughput. You can generate a lot of robust and reproducible data that is physiologically relevant, and you can screen many compounds or many antibodies or many modalities and get a lot of information. You then move to a more complex model like organoids, and you validate some of that biology there. These are much more complex models, so you don't get throughput, but that's okay because you're testing fewer compounds or fewer hits, so to speak. And then you move into a system with vasculature and flow, like bio-printed tissue, et cetera.
So, in that way, you move to more and more complex models and each model gives you a different set of answers that builds upon each other, and I like to call it the Lego model of physiology because even if you think about any disease, right? It's complex. It's multi-layered. And even when you think about how to treat it, you're not always treating it with one drug, you're treating it with a combination of drugs over time. Similarly, I think when you look at this ecosystem or continuum of different assays, they're giving you different answers about different aspects of that disease. And I think you do need a combination of these to answer a particular question about how you would tackle that disease.
So yes, I do think that being able to exist in that ecosystem and working as an end-to-end solution will be a huge advantage to all involved.
Mary Parker:
It sounds like there's going to be a lot of collaboration and cooperation between these different model companies in the future, which is also exciting for me.
Madhu Lal Nag:
That is exciting.
Mary Parker:
Love scientific collaboration. Well, thank you so much for joining me. This has been really fascinating. I appreciate your insight into this topic.
Madhu Lal Nag:
No, thank you for having me. This has been a pleasure.