Lantern Pharma Inc Aktie

Hi. This is Craig Brelsford with RedChip Companies. Thank you for joining us for what promises to be an exciting session with Lantern Pharma. Today's session is centered around a live real-time demonstration of withZeta.ai, Lantern Pharma's next-generation AI platform designed to transform how oncology drugs are discovered, particularly in rare cancers. Rather than just talking about the technology, you'll see it in action, executing research workflows, synthesizing complex scientific data and generating insights in real-time. This is a rare opportunity to observe how AI is being applied at the front lines of drug development.

Lantern Pharma, which trades on the NASDAQ under the ticker LTRN, is positioning this platform not only as a scientific engine, but also as a scalable subscription-based business with meaningful commercial potential.

Joining us today is Panna Sharma, Chief Executive Officer, President and Director of Lantern Pharma, who will guide us through the demonstration and discuss the broader implications of this technology. We will begin with the presentation and demo momentarily followed by a Q&A session. [Operator Instructions]

Before we begin, please allow me to read the safe harbor statement. This call may contain forward-looking statements within the meaning of the Private Securities Litigation Reform Act of 1995. All statements pertaining to future financial and/or operating results, along with other statements about the future expectations, beliefs, goals, plans or prospects expressed by management constitute forward-looking statements. Any statements that are not historical facts should also be considered forward-looking statements. And of course, forward-looking statements involve risks and uncertainties.

Panna, go right ahead.

Thank you. Thank you, everyone, for joining me this afternoon. I really appreciate you guys taking time to learn about our AI platform withZeta. I'm going to give you a little bit of background before we dive into the actual platform. But the most important thing we're going to do today is everyone on this call is going to witness live the research on a very specific rare cancer, understanding what drugs may or may not work well in that rare cancer, understanding the gaps and issues using our AI architecture and then actually developing a drug for that rare cancer and perhaps even time allowing developing a budget to make that molecule or to run the trial.

That's a lot in half an hour or 45 minutes, but that is the power of this multi-agentic tool. We were focused, I'll take a little backstory on how we got to where we are today withZeta. WithZeta started -- I'm going to share my screen also. Let's see how we can do this. I share the screen. Okay. Hopefully, everyone can see the screen. So we started withZeta. It started really as an internal project initially. As all of you may or may not know, Lantern Pharma was founded on the premise that big data and AI could help us better understand how molecules work or don't work in cancer and be able to help us also decide which cancers to go after and help us decide what combinations of drugs.

To me, these are all essentially data problems. And if you get a lot of the data, you have a better shot at success. It doesn't mean it's going to be binary like it's going to succeed or it's going to fail, but you have a better shot. You can constantly make better incremental changes and you can better refine what you're going after through data.

And so the premise of Lantern was to use lots and lots of data. We use hundreds and hundreds of billions of data points from wherever we can find them, and we had algorithms on top of that and disease models and insights and of course, tons of curated knowledge bases, knowledge bases that help us be better at developing drugs as an AI drug developer first.

And then I was at a conference a little over a year ago, and the question came from the audience about what if -- what we were doing internally, what if they could use it. And I thought for a minute, well, we've got a great team. We can do a collaboration. But collaborations are lengthy. They're expensive. They take time. And at the time, AI was really beginning to blossom, cloud-based computing, real-time SQL queries, agentic architectures to translate natural language to machine executable code.

And the thought came to me what if we can start doing that at a level and scale where we can actually create our portions of our AI and put it out into the public. Let the public use it. And I thought this would be a wonderful, wonderful AI for good. If we could discover therapies and develop insights not only just for our small emerging biotech, but we could do it for the broad oncology cancer community, especially in cancers where they're so challenging, rare cancers. The data is sparse. It's hard to collect. It's hard to put together.

Experts are hard to get on the phone and models break. What if we could do this in rare cancers because we are sitting now at Lantern on 12 FDA designations, 6 orphan, 4 rare pediatric, 2 Fast Track. And the thought was if a small company like ours can have all these targeted FDA designations and have a real conviction that these are going to work and these -- and they all did.

They work from in silico to then in vitro to then in vivo and now like all of our drugs or first generation of molecules are all in trials where the mechanisms have been validated. And very importantly, we've seen some very good impact on patients and the drugs are tolerable and they're working the way we imagined that they would work according to our early models. So we had a really good, I would say, library of success doing drug development, the AI win, the Lantern win.

We said, why don't we do this now and bring this architecture and bring our ways of thinking to the entire rare cancer community, where it's accessible to any investigator anywhere, 24 hours a day. You can bring the power of all these experts. And in rare cancers, it's very critical because the market has largely failed. There's over 438 distinct and rare cancers in our curated knowledge base.

And very importantly, there's about 5 million rare cancer patients globally that have very few options. About 30% of deaths every year occur in rare cancers. So although every rare cancer can be very small, there could be some that only affects 50, 60 people a year in the U.S. and maybe less than 400 or 500 globally and some even smaller than that. But collectively, rare cancers account for 30% of deaths. And it's still, regardless of the patient population size, it costs nearly $1 billion to $2 billion to go and solve for these cancers. We want to make a change in that.

It shouldn't cost that much. It shouldn't take that long, and these cancer patients deserve better outcomes. So what if we can compress that time. And we were already doing it at Lantern. We compress the time line significantly. On average, from an idea to an IND filing to dosing of first patient can take upwards of 3 to 5 years for pharma companies, sometimes even longer.

For us, we can do it at a fraction of the time and cost. And now we can even do it faster because of Zeta. We can do real-time literature searches like literally in seconds and actually query complex data. We can do pathway analysis, both molecular and permeability questions around the blood-brain barrier, cheminformatic questions, [ acne ] predictions, tox predictions. And we can do -- again, do all that within seconds and minutes.

We have the world's fastest performing molecular analysis and molecular intelligence tool, where we can analyze over 99 features of a molecule literally within a second or 2. Nowhere else is that available. And we said, why don't we make this available not just in rare cancers, but for all molecules, and we did that as well. We also created and implemented a de novo molecular structures engine called ether0, where we implemented it to actually create novel molecules, generate new first-in-kind molecules or even optimize existing molecules. And then all of this can be embedded in a real living knowledge graph so that you can retain the information and it's built as you search.

So it's like a real living brain that's growing literally in front of you as you explore and go deeper and you can share this knowledge graph with others in your organization or even other collaborators. The other great thing this tool has -- we got this from feedback -- early feedback from the community is they wanted to have various research modes. They wanted to have in-depth investigator mode where you can recurse up to 10 investigation rounds with lots of parallel tools.

You could go slightly lighter with an explorer mode, where you can have an interactive dialogue, a quick dialogue and maybe only go 3 tools deep and only a few rounds of recursive thinking. And so it's optimized for speed rather than comprehensive depth. And then we created a synthesized mode, a reporter mode where it's structured into a comprehensive report that you can share with others or share with other organizations and use for transforming your dialogue and your insight with your co-scientists withZeta into more formal documentation.

And that's the basis of also of a lot of other future features where we're going to be able to actually generate final regulatory and potentially even FDA-compliant and publishable documents directly from the withZeta platform. As we're creating Zeta, we realized that not all scientists really are the same, right? They're all very different. Just like any industry, there are subtleties between what a medicinal chemist does and the tools they think about, the tools they use, the way they approach a problem versus, say, a clinical oncologist versus perhaps a biomarker and translational scientists versus a clinical trial design person.

They're all trying to solve for the same problem, that rare cancer or that cancer or that disease, but they're bringing their collective multidisciplinary skills in different ways. Some people focus on molecular design, some on endpoint selection, some on, is this biomarker available clinically or am I going to have to develop it? Some focus on the treatment landscape, some focus on lines of therapy. And so we thought, well, depending on the user, depending on the problem they're trying to solve, depending on where they are in the journey of developing and understanding the rare cancer, they may want to bring one of those personas and heighten it. And so what we did is we trained them based on each of these personas. And each of these personas had very specific things that they were great at and very importantly, tools that they used.

Remember, underlying all of Zeta is a highly curated architecture of knowledge bases, tools and prioritization of tools for recursive thinking. And so that's what these personas are, these personalities. It's like you get the same information, but if you're an accountant, you're going to think slightly different than a marketer, than say, a strategy person. You guys are all solving the same business problem, but you're going to approach it using different lens. And that's what these co-scientist personas do.

In future versions, we're actually thinking about having an environment in which all these co-scientists actually all try to solve your problems and then get together to create a Meta answer to your problem. So you get the perspective of all deep perspective from all the personalities. So imagine it's almost like a virtual biotech scientific team collapsing onto your problem and solving it literally within minutes or hours and not weeks and months.

So what you're about to see today real-time, and we'll dive into this next is a little bit about the platform, just a general orientation. We're going to query a rare cancer. We're going to see how the knowledge graph evolves. We're going to actually do some molecular analysis on one of the potential molecules for this rare cancer. And if any of you want to put into the chat window or suggest a rare cancer or a question for a rare cancer, I'd be happy to take something live because, again, this is a live demo. I want to show you this is not canned. This is real-time. And we're doing this real-time. Nothing here is going to be canned and hopefully have time to actually generate a novel molecule and actually even generate perhaps even a budget or a route to create the molecule.

So I'll come back after the demo and talk a little bit about what's next on the platform. Like any platform, like any tool, we're already excited and thinking about how can we do multiuser investigation? How can we do role-based access for larger organizations? How can we do white label deployment for some of these pharmas that are very excited to be using this and use Zeta to augment what they're doing.

These tools in the future are not going to be, are we going to use them, it's going to be how much are we using them. And these tools are going to get better and smarter. And the evolution of science will be with co-scientists.

So with that, I'm going to go ahead and move over to the platform. There's a great question. I know it's a little early, but there was a great question. So it's why only cancer? And it's a very valid question, and I don't mean to kind of change or be little any of those. But the reason we focus on cancer is that that's what we're good at.

So it would be very difficult for us to have the credibility and more importantly, the know-how. I mean we were already building these tools. We didn't create Zeta because we were trying to be creating an AI platform. We created Zeta because we're a cancer company. Everyone in our team is maniacally focused on creating cancer drugs and insights faster and cheaper. We know that there's a new way to use data to solve these complex biological problems.

And all the cancers that we're going after are challenging. They're rare or orphan and there's some of the more difficult cancers, and we enjoy that. And that's the need. We want to go after this white space like one of our trials is for people who don't smoke and they get lung cancer, non-small cell lung cancer. It's a very different disease. The pathways are different. The mutational profile is different. And our drug seems to have some great results. That's in a Phase II trial in Japan, Taiwan, United States, multiple countries.

And we've got some great data on that, and we've been able to really make an impact in a lot of these patients in the trial. Another one of our drugs is focused on very aggressive cancers that have what's called DNA damage repair mutation, and they overexpress a certain enzyme called PTGR1. In that drug, LP-184 is so unique that it actually activates inside the cancer cell because PTGR1 is only expressed -- not only expressed, it's highly overexpressed in aggressive cancer cells, and that's what activates LP-184 into being these 2 really potent molecules, one that attacks the nuclear -- the DNA inside the nucleus of the cancer cell and breaks it and one that attacks in the cytoplasmic region, we believe.

And then LP-284, which is another synthetically lethal molecule, works very differently. It tends to really prefer blood cancer cells and cells that overexpress CD19 and CD20 and also breaks apart the DNA. Another molecule that we're developing that has a very unique payload. It's an ADC that has a very unique payload and a very, very potent, cryptophycin payload attached to an antibody.

And again, going after some pretty challenging cancers. And so we're cancer drug developers. That's kind of almost everyone's background in the company. And we're developing and building these tools not because we just wanted to develop tools. We're doing them because we wanted -- we needed to do it. It came from the essential problem is that we're a small company with limited budget, but we're big believers in data and AI and using big data to solve big complex problems in oncology.

And so as we're developing these tools, we realized that in this agentic world, and our team was uniquely positioned and knowledgeable that we can actually share these tools with the world. And that's what we've done, and we're very excited about it. We think the underlying platform is such a massive force multiplier. And it makes sense for us in cancer to do it first because that's where we built the models. That's where we built the algorithms.

That's where we built the knowledge base. And so it was very straightforward, if you can call it straightforward to go from an internal set of tools and systems to now a publicly facing one. And whoever asked that question, you're absolutely right. We can do it for many other diseases. And that's our long-term ambition with the platform is maybe to partner with other pharma companies and other disease companies and stamp it out for cardiac, for immunologic disease, other rare diseases.

But rare cancers to us is such a prime need, very expensive, very costly, very time consuming and 30% of deaths and no one solved for it. So this is the only platform. So we want to do something that's big, meaningful and unique. And it's -- we have the credibility to do it in this category. So that's why here. So let's go through the platform. Hopefully, you guys can -- I don't know if you can see my screen. Can everyone see it? Yes.

Okay. So let's look at the screen. So when we get to Zeta, we go to the sidebar, you can have all your prior chats in the sidebar. We can go to a new chat. And so we'll go to a new chat. You can also have an entire ontology. So if you're ever interested in how these cancers link, you can dig into the ontology of liver cancer and go to liver angiosarcoma and understand kind of what the subtypes are, get the classification, get all kinds of great information.

We're launching a trial in bladder cancer, for example, in the next few months. It will be in a certain subtype of bladder cancer. So you can see all the various subtypes. Inverted urothelial papilloma, which, again, I don't know anything about, but it is a subtype. It's one of those ultra-rare cancers. So this is -- gives us an ontological framework for all the 439 cancers. You can see the toolkit.

So we have a number of proprietary databases, our rare cancer knowledge base, standard of care database, a rare cancer taxonomy, all the clinical trials over 570,000 clinical trials, including their outcomes, a cancer drug database with all the FDA-approved drugs and all the standard of care protocols as well as our own drug database and then a massive research document library. And so that way, the key for the proprietary database is that gives a structure and guide for how Zeta think. So this is your space. This is your domain in which your lab to function and think about.

And then we also give it external resources, medical literature, rare disease ontology, cell line references so that can generate new ideas from biomarkers and of course, FDA drug labels, you're going to see an open FDA to look at all the approved indications, safety warnings, dosing information, pharmacology, et cetera, as it gets updated. So both live and proprietary databases, and that governs the space in which Zeta thinks and that keeps it very, very different than any of the kind of the GPTs that are out there today because it's only thinking in this space.

So let's go back to -- this is my account. Let's go back to the tool. Now you can pick a co-scientist or you can pick general or just start with general from now, and you can go into various levels of depth, like I said, explorer mode, which is default, investigator mode, et cetera. And you can also pick tools. So imagine you're a Merck or BMS and you want to attach your internal tools, you can easily go to the toolkit, go to your external resources and you can attach your own internal tools. And that's something that will be quite exciting as well.

So does anyone have a rare cancer or a cancer they'd like to discuss? Or let's go ahead and ask Zeta, we can actually also voice activated. So let's go ahead and do it with voice since I'm standing here. Can you tell me which rare pediatric blood cancers have the highest unmet need and which drugs are promising in Phase III for those cancers? So a lot of times when you're researching, you're typing or sometimes you don't want to type all that, so give it a nice cool voice interface, so you can actually just chat with Zeta.

The answers are somewhat complicated. So it's not going to chat back to you, not quite yet, but it's now going to go and do a systematic investigation. And so it's going to use multiple tools. You can see which tools it's going through and running through. You can then see how much it's reading. So it's already read through about 36 papers, 20 clinical trials that's gone through, 7 outside references and 9 PubMed references.

So you can see a massive amount of data. So this is like real-time reading all this and beginning to connect dots and systematically evaluate the knowledge and to try to get us to the answer. So as it looks and executes on the tools, you can see which tools it's using. And then you can also -- one of the most important things from a lot of these AIs is we want the AI to be transparent. So if you go to a lot of tools today, they may be -- they tell you perhaps about the reference, but not always. But the challenge that they oftentimes have is that they don't tell you how they arrived at the conclusion or what they're making up because they're generative AI.

So their focus on coming up with like the next word, the next chunk, the next vector embedding may or may not actually have to do with what you're talking about and may be poisoned or informed by knowledge that's not as relevant. So as you can see, Zeta has seen several rare pediatric blood cancers with significant unmet need, and it's an -- identified specifically now the Phase III one. So it's identified the blood cancers. It's letting you know as a colleague or co-scientist, hey, I'm almost there.

I'm now cutting through all the Phase III data, and it's in a search for the specific Phase III therapies that are the most challenging. So again, something like this is very common. You would do this in a CRO, you might do in a pharma company, maybe as an investment bank, maybe as a biotech analyst. And it may take you hours. This has now taken us probably less than 2 minutes.

And you can see this color coding on the screen, which is very important. We'll come back to color coding. But you can see that Zeta has given us things right here. And this is very important because it's never, it's trained as a drug developer, a summation of all these scientists, scientists. You're not training it to be a scientist as you might have to if you go to like a ChatGPT or a Claude. Its personality is that of the ultimate kind of Lantern scientist that it's going to talk to you in bullet points, specific direct. And it's going to put things into structures because that's very important as a scientist.

So Tier 1, most critical unmet need, no Phase III therapies available. So it automatically thinks about what's critical and what the promising early therapies are. Again, drugs are in orange, diseases are in red, biomarkers are in green or purple. Papers that are referenced are in the inverse white and gray. And it says unmet need severity critical for non-Down syndrome AML -- AMKL actually.

And then, of course, it gives us the subtypes terrible survival, high relapse rate and then it gives us Phase Tier 2, where there's a high unmet need, but there's Phase III therapies emerging. And then also moderate need where there's multiple Phase III therapies. So obviously, there's some promise and then it gives us the treatment gap paradox. So the most striking finding is that the pediatric blood cancer, the worst outcomes have no Phase III therapies.

So 5-year survival in some of these pediatric blood cancers, Phase III therapies, none, none, none and some very low. So therapeutic landscape is inverted because people are going after the disease with a better baseline outcome, and they have multiple Phase III options, and it gives us the conditions and it has a recommendation. So just like a scientist, it's not going to just sit there and regurgitate. It's driven to action. That's very, very important.

It wants to get to an answer. It wants to help you actually solve for a problem. And so for the highest unmet need, JMML, non-DS-AMKL, accelerated development pathways are critical. So expand trametinib evaluation, investigate the Hedgehog pathway, develop the cooperative trial and explore novel targets identified through comprehensive genomic profiling.

And so this is one of the things we could do. So probably not -- we don't have time on this call, but users of Zeta would be actually able to do comprehensive genomic profiling by pulling up all the data sets that we have and actually look at what's going on. But let's go ahead and look at are there -- so on Zeta, we can also just do orally the adjuvant typing. Are there any Hedgehog/GLI pathway inhibitors that are currently in development for rare cancers or other cancers that we can learn from?

So very broad question. I think it translated GLI to July. So -- yes, but it corrected it, which is great. So it knew what we were talking about. So now it's going to investigate those inhibitors that are currently in development to identify strategies applicable to pediatric AMKL, which is very, very high need. And again, it will give you total transparency in how it's getting there.

And as you can also see, as it does it, the number of publications is jumped up to 87. So a lot of publications now it's worked through. And this is very important because as a scientist, you're going to read -- you always have a stack of publications and you're going to be able to read and maybe remember the last 2, 3, 4. But imagine having all of these 87 all embedded into your brain and live and available and you're connecting the dots and all of them real-time, like that is pretty phenomenal.

And more importantly, you're connecting things that you may not even know about. So the clinical experience, now it's giving us why this drug failed in Phase III, why it matters, why this drug succeeded. So now it's going to do exactly what a scientist is going to do. It's going to look at failures, successes and look at why these different programs failed or succeeded and perhaps come up with a potentially superior strategy.

And so it's going to look at the GLI inhibitors in development, what the lessons are for what we should be doing, even giving us a combination therapy, which is likely, which is great because that's how we think as drug developers as you want to attack these cancers from multiple pathways, multiple combinations, and it's already coming to that conclusion as well. This kind of gives us the most actionable strategy, which combines XYZ, would you like to explore specific mechanism of this or investigate potential resistance mechanisms?

And as I mentioned, we're going to look at the knowledge graph. So as it develops this, this knowledge graph, it's creating a knowledge graph of these concepts of the drug, of the protein, of the disease, of the trial that it's involved with. And as we -- and you can drill into any area of the knowledge graph. And as the knowledge graph evolves, it's going to start connecting more and more dots. Remember, we have about -- we have about 60 nodes in it now. You can click on any of the nodes.

You can share these with others, so you can export it as an interactive HTML file or you can export it as a JSON object, whichever you prefer, which is really exciting. And going back here, these are the next steps. But what we mentioned is that we want to -- this is an interesting drug. This ruxolitinib plus glasdegib, investigator-initiated trial feasible. Priority is medium, doesn't give it a high priority. It gave arsenic trioxide and chemotherapy because it's already FDA approved for one of those indications, a higher priority.

But let's go with this combination. It comes with a question, can we improve this combination on this combo regimen? This is a question that would take companies months and months and maybe even years. And they look at a combination like this and oftentimes in a combination today, a cancer company thinks it's an excellent question, we investigate the target alternatives to the glasdegib-ruxolitinib combo focusing on precision inhibitors and ADCs.

Oftentimes, they'll take those and say, can we actually combine the drug to a new type of drug or to a conjugate like an antibody drug conjugate, which is exactly what I asked Zeta, my co-scientist to think about and come back to me. Now historically, and again, just from our own ADC program, when we ask questions like that, it takes weeks of research. You may have ideas, but of course, everyone is always biased and so you want to explore out your first idea, but then you want to validate it. You want to have data. You want to have mechanistic rationale. You want to look at the literature. You want to do some takedown of actual data sets and explore what makes sense.

And so this is what Zeta is doing in the background. So all that work that a scientist would do to review research, do primary data, look at structures, analyze molecules, look at making predictions on how those molecules may or may not work in various cancers, look at what's being overexpressed or targeted, all that work is being done in the background now. And also it's reviewing the literature as well. So it's going to continue updating the literature.

And then you can actually go through the reasoning process. And it's a deep and meaningful question. And so it's -- and if it doesn't have an answer, unlike a lot of GPTs, it's not going to force an answer just to make you happy. It's going to say, I can answer it or I don't have sufficient information or here's what I need. Now the other thing with this particular question because it's so targeted, maybe I'm a biotech that actually has an ADC that's targeting this Hedgehog pathway. And so it may not be in the public domain. So what I could easily do is I could attach a file.

I can go here to the attachment piece right here, and I can attach a file and say, take a look at this file and tell me what you think. And so that's also very, very possible. When I go into light mode, it might be easier to see.

Wow, it's a great, great answer, detailed. So this is wonderful. So precision target folate receptor alpha discovered a fusion-specific surface target. It's referencing a discovery made in 2022, the folate receptor specifically and highly expressed on the surface of this positive AMKL cell, making an ideal precision target is great. So it's going to target this for the ADC, limited normal tissue expression, very important. That's a rule that we always look at. So even as it goes and starts targeting this, it already knows fundamental disease models and rules for how you actually think about targets.

And you want targets that are highly overexpressed in your disease and highly underexpressed or not expressed or a normal expression level in healthy tissue. And then also you want to make sure you have selective targeting. So you want the right therapeutic window. You want to make sure that the target is accessible on the surface. All these fundamental things that the typical GPTs don't fully understand or may or may not arrive at. That knowledge is already built into Zeta.

So if any of us have some targets, Tier 1, floor-directed therapies, [indiscernible], which is great, gives us the mechanism, preclinical efficiency, what its advantages are over the combination. Why trial -- terminated trial, raised questions about off-target, pediatric tolerability, efficacy threshold not met. And then it looks at the [ 4-1 ] CAR-T therapy, what the advantages are, what the challenges are, what the development pathway and then gave us some other ideas around Tier 2, BC-XL selective inhibitors mechanistically superior. This is great.

So now we're getting into options that we didn't even ask about. It reached its own conclusion after looking at the ADCs to say it's going to look at another selective inhibitor, something that we hadn't even prompted it to, but it made and revealed a critical therapeutic target by looking at a very, very recent finding that it's dependent on BC-XL, not BCL, BCL-2, which is great because, again, there's so many rare cancers. There's no way to be up on every single possible mechanism and biomarker, could easily overlook it.

And then Tier 3, which is CD123-directed immunotherapy, which has been tried as well. So you can see it now tiers all these various approaches and actually gives us an alternative just to the ruxolitinib drug as well. And so what is it proposing? An improved combination strategy, the [ 4-1 ] ADC plus BC-XL most targeted, that's great.

A dual fusion-driven vulnerability, potential synergy challenges, thrombocytopenia, overlapping chemotherapy. So again, this is very important because remember, it has access to all the standard of care data as well. So it's going to know what is going on with the side effects are, et cetera. And it gives us 3 specific strategies and also tells us you can actually do a GLI inhibitor, long-term investment, strongest biological rationale.

So let's say the strategy, this is long-term investment, inhibitor development program and then it ranks it. So again, puts it into a nice table, how clinically ready is it, what are the advantages and then gives us some really exciting immediate recommendations. So this is for newly diagnosed superiority to venetoclax. So let's ask it about this, although it didn't say that we should do a strategy for GLI inhibitor, I'm kind of intrigued by that.

So let's ask it about this GLI inhibitor and ask a very specific question. Given some of the challenges with GLI inhibitors and safety in the past, can you help us develop a novel GLI inhibitor that overcomes some of the issues in prior drug candidates? Let's see what it says. Of course, it said July, it has said GLI in the second line. So let's see how it thinks about this. So I like the challenge. So we're actually now -- just so we know, I should have put this into medicinal chemist mode. I didn't do that, but that's okay.

But we're here explicitly asking it to go come up with a potential new design for a long-term GLI inhibitor program that overcome historical pharmacological liabilities. And then someone asked a question about safety and tolerability, we can definitely touch on that as well. It's beginning to touch on some of those because the first generation of GLI and the second generation had some of those issues. So we're asking it explicitly to evaluate and overcome some of those liabilities.

We can also -- if we wanted to go into -- I urge you guys all to sign up for withZeta, and you could go into withZeta and actually ask it in these historical trials, what safety and tolerability issues have you seen the most in pediatric blood cancer trials? And you can zero in and say, what about for target inhibitors or what about for immunotherapy. So you can even give it by class, which would be a great problem. So it has a molecular structure of one of the drugs. It has critical structure liabilities revealed.

So it analyzes what's out there today, tells you what the pharmacological problems are, looks at case studies, poorly stable. It's a diamine derivative, which is interesting. So it's a prodrug, and it causes dosing challenges and pharmacokinetic is unpredictable, doesn't have great aqueous solubility. So you can see it's calculating all these values. These values do not come from a database. It's actually calculating it. So Zeta is handing this off to our molecular intelligence LQM and it's looking at the SMILE structure and the overall chemical formula and saying what is the logP value?

What is the TPSA? What is the [indiscernible]? And those are things that we've taught and trained it to think about when it tries to create a good medicine. And it also tells it has excessive molecular flexibility, which makes sense, too, because look at the structure, probably very bendy. And then it says the drug-likeness, it failed 3 out of the 6 filters, which you can still have a drug and fail 3 out of 6. But again, the more -- the fewer you fail, the more likely is you can have a good drug.

So now these are the things that we look at when we want to look at drugs as we look at weight, bond donors, [indiscernible] bonds, it's potential to cross the blood-brain barrier. So BBB penetrability to us is very important because if you're making these molecules, it's great because cancer oftentimes travels to the brain. So if you have a molecule can then both challenge tumors, but also then travel to the brain in case there are brain mets, that's like a really holy grail.

It tells us it needs improvement on BBB penetration. So that's something maybe we can live with, very few molecules cross. So maybe we look at that, and it's actually improved the drug-likeness from 3 filters to 4. And the molecule is now firmly in CNS drug space with optimal logP, et cetera. So these are the first-generation improvements. And now it's going to refine it further. And the way it thinks about it is that it's going to constantly do 2 or 3 levels of refinement, just like the scientists would.

You're never going to be happy with the first iteration. And they're going to say, okay, well, what do I like? What do I not like? And you're not going to get that like with a general Groq or OpenAI. I mean it doesn't know these things yet. It could definitely learn them, but it's going to learn them in a very awkward way. This is -- we've trained it explicitly to think as like a medicinal chemist or a clinical trial person or a clinical oncologist. And so it has a view of the world that is very different than these general purpose LLMs and isn't just randomly guessing at what the next thing it needs to say or do is.

And we have a question from -- can you save an investigation and get alerted? Yes, we are going to talk about some of the features like alerting and saving, sharing knowledge graphs, et cetera. But yes, those are all things that are on the road map, and we'll talk about that after we finish this investigation. And as you can see, as we mentioned, the knowledge graph continues to grow. So this is a live living knowledge graph. And so if I wanted to talk to my colleague, [ Barrett ] or Rick or Reed and say, hey, this is something really interesting.

Let's explore this further. Let's do some more sessions. I can actually just export it as a JSON file, which is machine readable, which is really cool, or I can export it as an HTML file that's interactive. And they cannot have to read through the text, they can actually see and click on it and see what is the thing that I'm trying to research.

And like any knowledge space that are graphing your head, you're going to have some explicit areas of high density, but then you're also going to have these concepts that flowed out there that you know somehow are involved, but you're not quite sure exactly how. And so you can see it's created this little node around STAT3, which is involved with many cancers. Not quite connected yet and even has this thing floating out here, we'll see what that is. The SMO protein, okay? That's also involved in the Hedgehog pathway and it's involved in some of these GLI drugs, but again, kind of not fully connected into the core knowledge graph.

So let's see what the final. So here's what the analysis is, the optimized candidate. So it refined it. The V2 iteration shows improvement. Here's a new drug after 3 iterative design cycles, a comprehensive comparison of a novel GLI inhibitor candidate. Now I want to put this in perspective. This is now -- we're about maybe 44 minutes into a dialogue with this AI. This is a novel drug for an ultra-rare cancer that we've never talked about before. This is not something that I trained Zeta on. This is something that in a traditional biotech environment, including at Lantern, would very likely take upwards of 4 to months to a year.

Let that sink in, 44 minutes with one instance of Zeta. The future of Zeta is I can have 5 or 6 of these Zetas. I can have all of them collaborating on this kind of problem and actually compete with one another for different types of questions. So it created this molecule. It gave me the optimization results, told me what it likes and what doesn't like. again, each iteration so I have complete transparency in what it changed and tweaked. Iteration 3, what it's called the final, tells them what test it pass to be a good drug or not a good drug, what it solved for chemical stability, aqueous solubility, molecular flexibility.

The BBB penetration is only partial success, but that's okay. We may want to go ahead and say, we don't need that with this molecule because we need to target the systemic tumor and not necessarily the brain mets if they arise. And then it gives us a candidate, what the advantages are for this candidate, what the development challenges are that we need to still look for and then what the key advantages are over the current combination regimen that we first started with.

Remember, this was the combination regimen that was initially proposed for some of these cancers that we were intrigued about because we're targeting both the Hedgehog pathway as well as an indirect pathway inhibition.

And we said, can we design something that does both. Gives us a summary of what we've achieved, what this represents. I look at 21 tools. That's a lot of tools. 21 tools. It's referenced over 140 different publications and trials and PubMed results and drug studies. And it gives you an answer about how it actually arrived at it. So you can sit there and go through and read it or you can even ask Zeta and say Zeta, what are the flaws in this reasoning process and it can explore alternative delivery strategies. So you can see Zeta is a very, very comprehensive tool for drug development.

And again, we focus initially in cancers and rare cancers because that's what we're really great at. That's what we know a lot about. But the same architecture technically with different disease models and some updated data sets, we believe we can do this for multiple other categories. So there, I'm going to pause and go back to some of the future features. Thank you guys for going through this. We know this is incredibly complicated stuff. And -- but this is the only AI that does it.

And again, it's a direct result of the work that we've been doing, but now with a wonderful natural language interface. And I can save all this as a PDF file even. So like I can -- there you go, there's a PDF file. I can pop up in the PDF file and also talk about what's next. Let's -- so here's the PDF file. I don't know if you guys can see that. But here's the PDF file, of course, the disclaimer, not medical advice, consult health care professional, et cetera. And then I can share this internally.

Some people like knowledge graph, some people like the PDF, some people want both. But again, we made it so that it thinks like a scientist because you may want these tables. You may want to look at what the flaws are. You may want to look at it and come back and say, hey, I don't like this ethyl plus ethanol stability property. Let's tweak that. Or let's look at how this can be slightly different. And so these are all the things that you would want to do.

And let's go ahead now, let's go back to the future piece. What else we're doing with Zeta in the future, which is very important, of course. Okay, there we go. So what's next? Let's talk about what -- where this is headed. We've got a lot of exciting biotech companies using it, top researchers and institutions in Europe at Fox Chase at UT Southwest, a lot of different great research groups.

We have had over 100-plus people interested at the American Association of Cancer Research. We think this could be a massive multi-hundred million dollar opportunity to sell subscriptions. Obviously, we're coming at a very low price initially to get -- to drive usage, to drive adoption.

And we think many people said this has been a great tool. Some of the biggest users so far of the tool are actually consultants, analysts, some CROs and academic researchers. So what's next? We're going to have enterprise features such as team workspaces, shared knowledge graphs, API access, white label deployment potentially and then also social features to make it more sticky, personalized feeds, multiuser investigation sessions so that myself and maybe Barrett or someone else on the call can -- we can all work together toward a problem and have shared research environments.

We also plan on deeper biology and smarter tools. We're going to have a mobile optimized interface, so people can look at it in other devices. We're going to expand the rare cancer disease ontology with more molecular and pathway annotation. We're going to incorporate more deeply our antibody drug conjugate module, our antibody assessment development characterization also within Zeta and more pathway mechanism knowledge. So we -- our goal is to make this smarter and smarter and bring the best possible information, knowledge and disease modeling to the power of everyone's desktop.

And we think this can crush and collapse the time lines involved in drug development, at least initially in cancer, but eventually lots of diseases. And yes, we can definitely save an investigation and get alerted that those are all part of the features that we're going to have. But we urge you guys to go and sign up, sign up, start using it, give us feedback. And I think this multi-agentic AI is really a fundamental reset. It's a massive force multiplier. And it's our conviction that some of the biggest breakthroughs in the future of medicine are going to come from machine and human expertise coming together.

These autonomous co-scientist systems, these multi-agentic, multi-tool, kind of autonomous scientists that are iterating and thinking on their own because we've taught them how to do it. These are -- and we say a co-scientist, and we call it withZeta because it essentially -- it's making us smarter and better. Like every scientist wants to wake up and be the best. But can they read 39 papers in 2 minutes? Probably not.

Can they store in their mind space 500 different biomarkers and 200,000 trials and last 12 years of failures and 3 years of successes, maybe there are definitely geniuses that can do that, but they're far and few between. And so this becomes really a tool and especially as it becomes harder and harder to become a great scientist because the knowledge is just piling up, we need these systems to really make scientific intelligence and put it at a different level.

It becomes a force multiplier and becomes a fast accelerant to actually using AI for what it's meant to do. It's meant to be doing great things to actually lift us to new levels, not make cat videos and spoofs and have people's bank accounts. It's really here to make humanity better. And scientists, especially, we want to do great things, and now we have the tools to do it.

So very excited about this intersection of autonomous scientific intelligence with great minds, and that's why we call Zeta co-scientist. So I'll pause here and see if there are any more questions from the audience.

So question, what does the feedback look like from other biotechs? I have to say the feedback so far has been really very exciting to us. It's been wonderful feedback. They always first ask, well, how is this different from the other GPTs. And then after they start using it for about 15 minutes, 10 minutes, say now, it's very different. And that question never comes up again. We've got people sit down and say, oh, I've been studying this rare cancer for 12 years, and they pull out a paper research and they say, what about these things?

And they try to -- and then they walk away like, wow, I need this tool. So the feedback so far has been quite, quite exciting. I mean a lot of the features that we put out like the multiple roles and the recursive thinking and the tool sets all came from very early users and feedback from our early data that we had. The subscriber traction is just beginning. Again, we just launched the public debut at AACR. I think the next 6 months is going to be basically building up a subscriber base and getting this out there into as many hands and into many people as we possibly can.

And then we're really going to see the flywheel, the revenue flywheel, I think, more after the first 12 months. But yes, there's a lot of commercial interest. Near-term milestones are obviously continue to get users, to land a couple of biopharma deals, especially larger biopharma deals and very importantly, start building out the enterprise toolkit so that pharma does not have any reason not to subscribe. We want to make it as plain and simple and easy as possible for people to get on to the tool and just be the best at developing great new science.

So thank you, everyone, for participating. I always get a joy out of these sessions. It's wonderful to see what these tools can do. And so thank you all for your time this afternoon.

That concludes today's webinar.

Hello. This is Paul Kuntz with RedChip Companies. I want to thank everyone for joining us for what promises to be an exciting session with Lantern Pharma.

Today's session is centered around a live real-time demonstration of withZeta.ai, Lantern Pharma's next-generation AI platform designed to transform how oncology drugs are discovered, particularly in rare cancers.

Rather than just talking about the technology, you'll actually see it in action, executing research workflows, synthesizing complex scientific data, and generating insights in real time.

This is a rare opportunity to observe how AI is being applied at the front lines of drug development.  

Lantern Pharma, which trades on the NASDAQ under the ticker LTRN, is positioning this platform not only as a scientific engine but as a scalable subscription-based business with meaningful commercial potential.

Joining us today is Panna Sharma, Chief Executive Officer, President, and Director of Lantern Pharma, who will guide us through the demonstration and discuss the broader implications of the technology. 

We'll begin with the presentation and demo momentarily, followed by a Q&A session.

[Operator Instructions].

Before we begin, please allow me to read the safe harbor statement.  

This call may contain forward-looking statements within the meaning of the Private Securities Litigation Reform Act of 1995.

All statements pertaining to future financial and/or operating results, along with other statements about the future expectations, beliefs, goals, plans, or prospects expressed by management, constitute forward-looking statements.

Any statements that are not historical facts should also be considered forward-looking. And of course, forward-looking statements involve risks and uncertainties.  

With that, Panna, please go ahead.

Paul, thank you very much, and thank you for attending this morning. I'm going to give a little bit of background.

Hopefully, everyone can see my screen okay that I'm sharing. But we launched withZeta internally as a tool to actually help us better understand which indications to pursue and which drug combinations made sense in cancers.

It started really as a genesis when I was presenting at a rare disease conference, and I was talking about how Lantern was using data to better understand indications and how we were bringing indications and pinpointing them with our drugs with precision.

We ended up with 6 orphan drug designations, 4 rare pediatric disease designations, and 2 fast track.

And there were some in the audience who asked, really, how we were able to do this so quickly. That got me thinking, well, what if we could do it not just for ourselves, what if we could do it for every drug developer?  

What if we could use the promise of cloud computing and these new emerging AI architectures that can understand natural language and generate real-time queries, and that could go and digest complex publications? 

What if we could bring all that technology, integrate it in a way that is trained exclusively in the domain of understanding disease and developing medicine? 

So we embarked on a journey at Lantern to say, well, what's one of the most complex areas? We said rare cancers. They're by nature complex.

There's not enough data. It's very scattered. And we said, let's go on this mission to try to create a multi-agentic, multi-tool architecture that obviously serves our needs first as drug developers, but then can serve the needs of the tens of thousands of biotechs, cancer centers, drug developers, scientists, students, and even patients trying to better understand rare cancers and coming up with more cost-effective and more timely insights. 

So we started this mission, and the market, though, for rare cancers is quite significant.  

WithZeta has over 438 highly curated cancers that have complex information, bioinformatic, the current state of the disease, the biomarkers of interest, disease progression, trials that have succeeded, trials that have failed, and drugs that have succeeded and failed.

It's probably the most comprehensive collection of curated. And on top of that, there are still about 5 million patients globally who are affected by rare cancers. 

So although they're rare, in totality, though, it's about 30% of all deaths every year in cancer that are rare cancers. And so we thought, what if we could do all the literature research in real time?  

What if we can look at all the pathway analysis to predict anything from blood-brain barrier penetrability of a drug to the hem informatic nature of a drug to its topology to ADME predictions. 

On top of that, what if we can create new compounds, what if we can optimize compounds, what if we can create new molecular structures, and analyze their features? 

And on top of that, what if we can actually, as we're doing the research, create a living knowledge graph that is able to package the information and share it across our internal institution or across our colleagues.  

As we got that, we saw that the Agentic multitool architecture wasn't just repackaging information. It was actually creating complex connections and generating new scientific knowledge. 

It was advancing our understanding in ways that really accelerated the path from what could take hours, days, weeks, literally into minutes, and we'll see this slide today.

We'll actually go and solve for a real cancer today. Then, what we did on top of that, we said, well, not everyone wants the depth of information. 

So we got feedback from early users that we should have, like a super deep mode, call it investigator mode. We should have a mode that's maybe quick, more conversational, more based on speed rather than comprehensive depth, call that explorer mode.

You can switch between these modes. And then ultimately, even a reporter mode, which can take very complex pages and pages or volumes of information on your research and compress it into a report or a dialogue that you can share as more formal documentation.  

On top of that, what we thought is as we got deeper into this, the needs of different scientists are different. Some want to get general information, but there's a person who wants to actually design a trial even in our own team, and they want to go deep on trial design. 

There may be someone who wants to go deep on biomarker selection and design experiments to validate biomarkers, look at CRISPR data, and look at other information from biomarker studies.

There may be someone who's more focused on the therapy of translating molecular insight into actual clinical activity. 

So we created scientists that are tuned to how scientists really think, whether it be a medicinal chemist who does de novo drug design, or a clinical trial strategist, or a clinical oncologist, or a biomarker person.

We created these personas inside of Zeta, and you can select a persona and change it. So you may want to design a drug, and then you may want to go right into better understanding its clinical implications.

They may want to switch to designing a trial. And the great thing about all these is that they can actually all talk to one another. 

So they're not individual. They all can interact with one another. And in future versions of Zeta, we'll actually have teams of these co-scientists. 

Again, we'll see that in action today. So multiple research modes, a network of agents and tools, and co-scientists. So we have a lot of these different parameters now.  

What we're going to see today, in real time as we dig into it, is that we're going to query for rare cancers. We're going to see the knowledge graph in action. We're going to actually do some molecular analysis and then generate a new novel molecule here this morning. 

So, all over everyone's cup of coffee or so in the next 20 to 30 minutes, we're going to go deep on cancer and then actually get into designing a drug and looking at combinations. 

I'll come back to this presentation and talk about what's next withZeta. And, but for now, let's dive into the tool itself and then switch over to my actual computer screen.

This is from an earlier session I was doing, but let's go ahead and start a new session.  

So as I said, this is where you typically land when you go withZeta. And I have a screen a little bit larger than normal just to make sure everyone can see it.

Again, it defaults to the general scientist, and we can ask any question, just in natural language, you can even use your voice.

But before we dive in, let me give you a couple of features. In the side bar, you can start new chat like most you can go through various chats that you've had before and go through them and select something from a prior session.  

You can switch into different modes if you want to do light mode. In fact, we can do that today, or you can go to dark mode, or you can send it to your default.

Then, you can also go into a toolkit. And the toolkit is very important because this is the structure of how Zeta thinks.

We put a knowledge base, proprietary knowledge base across rare cancers, as well as gave it information about the standards of care, then we guard railed it with a cancer taxonomy, giving all the information from hundreds of thousands of clinical trials and their outcomes, drug database, and then a massive research document library that continues to grow. 

 It's got over 1.2 million knowledge objects.

On top of that, we have a very large molecular analysis engine, a large quantitative model, as we mentioned earlier this week, a model to do blood-brain barrier permeability, and then also to design new molecules from scratch.

And all these agents, they talk to one another. So we pass information from the natural language interface to the tools, the proprietary databases, and also to external resources that are highly selective. 

So PubMed, rare disease oncologies like the NCI Forus, certain cell line references, and of course, the open FDA to look at approved indications, safety warnings, dosing information, et cetera. 

So this is very comprehensive. This is exactly what any team in biotech would want at their disposal. And the great thing about it is that it's available 24 hours a day.

Just log in just like you would to Claude or GPT, but in this case, very, very deep in cancer drug development. So let's ask a question. In fact, let's ask it through audio.  

What rare cancers, what rare blood cancers, are in high patient need in children? And as it starts thinking through, you're going to see all the literature that it digests quickly.

And of course, you've got credits here, so people who subscribe will have credits. And then this gives you a sense of the context window, meaning what is the percent of the maximum context window that is being used. And what our team has done is made it dynamic, so it allows the conversations to continue fairly indefinitely.

And so you never really run out of a context window. So as you can see now, it's digested over 34 sources from PubMed publications, clinical trials, and it's going through, trying to now make a systematic investigation.

One of the things that we wanted to do, which is very important in science, is transparency. So as Zeta works on this, it actually shows you which tools it's using. And then, very importantly, how it's integrating the information.

And so you can always go back in here and actually look at how it arrives at the conclusions. It's not a black box. And that's one of the more important things when you're doing kind of detailed knowledge and scientific work, is to really have the transparency and how did we arrive at the conclusions that we arrived at.

And do we want to tweak them? Do we want to change the investigation? And just like any team, these tools are not going to always be 100% right, but they're going to be vastly more right and have all the information than wrong. 

And they're going to speed up the whole process of actually developing a drug, doing investigations, coming up with cutting-edge information, and synthesizing. 

So as you can see, now it's beginning to stream the information. So it gave us the blood cancers, pediatric blood cancers, with a high need. Infant leukemia with certain rearrangements. And as you can see, these genes are highlighted in green, just the way a drug developer would want them. 

So we've trained to think like a drug developer or a Lantern scientist, where certain genes are highlighted, papers are inversely highlighted, like in this gray and white. And then diseases are in red.

This is very important because when you scan through literature information, you want these things to jump out, genes, proteins, tissues, diseases, and it highlights that in real time. This is very important. 

We'll come back to it because it's also part of the knowledge graph that gets created, which you can share. And this is what guides the knowledge graph creation as well.

So as you can see, it's got a number of diseases that are rare blood cancers that it highlights, and it gives us what the common threads are for these young age, poor survival, and also gives us an idea of some new emerging trends. 

So we can ask it very importantly, we can switch over to a medicinal chemist mode. Let's do that.  

Can you discuss the challenges with Menin inhibitors and rank some of the preclinical drug candidates in Phase II and III trials? 

That's typically if you're researching the space, you would take hours to figure that out. It's not very likely you're going to have something ready to go that has a deep dive on menin inhibitors in blood cancers for pediatric indications and then rank order which ones are interesting.

And so that's what this is going to do in real time. And again, it's going to access a number of tools and then also recurs through them and rank the preclinical. So clearly, it's going to go through and start looking at some of these and then also rank them.  

That's great. Again, these are not questions that are precanned. So this is now really thinking about the challenges with menin inhibitors, differentiation syndrome, primary safety concern, brings up safety from other studies on these menin inhibitors.

So you can see both these drugs have 15% to 25% incidence and severity in some of these blood cancers, 10% to 20% across all menin inhibitors tells you what happens to patients, what the management requirements are, various options for dosing and then eventually, of course, acquired resistance as the cancer changes, limited response, drug and then it's going to wen them. 

So Tier 1, obviously, is the one that's already approved. So that's going to have the highest rank. And it gives us the strength of that and the weaknesses.

Then it gives us the Phase II registrational trial is the #2 rank. This is KO-539, regulatory submission anticipated, great. It tells us what the results were from the Phase I and the Phase II, again, gives us some of the strengths and some of the weaknesses, including some of the CAM2A activity, which is interesting and very limited pediatric data.

And then goes and continues with other menin inhibitors as well.

Tier 2 drugs and then some of the early-stage drugs. And again, exactly what we wanted. It gives us a ranking summary, which we can cut and pace so we can take this table. 

Again, it tries to keep everything very brief and concise. Because that's the way scientists want to think. We want really the data and so it puts it into a nice table.

It tells us some of the insights. And then very importantly, it updates the knowledge graph. And so you can see now that it has a knowledge graph that it updates.

You can play with the knowledge graph and look at the relationships. The drugs are in orange, disease in red, the genes in green. You can see how this gene relates to this CAM2A rearrangement acute leukemia. 

You can go dig into that specific gene, that disease and that drug. The wonderful thing about this is you can actually then also export this knowledge graph either as an HTML file or which is interactive actually or as a machine readable JSON.

And so this is great because this allows an enterprise to save all the information across all your scientists as they're doing this work. So let's go back and ask Zeta to dive deeper into a new menin inhibitor.  

Can you help me design a novel menin inhibitor that has better potential to be combined with other drugs and reduces side effects?

The great thing about this is that if it cannot do something, it's not going to do it. So unlike a lot of chat agents, which try to make up stuff in order to basically be sycophantic, it's actually going to tell you whether this challenge is really doable or not. 

In this case, since it's a pretty significant challenge, it's going to probably use 16 tools or maybe even more, and it's going to think.  

So, imagine your team is thinking about Menin inhibitors, and you have a lot of detailed knowledge about the biology of Menin inhibition. You want to really target that mechanism.

And you can now do a deep dive and say, " Is it possible to look at some of the existing scaffolds and structures and come up with something that is novel?" But then also is going to be combined because we know that combinations oftentimes have more durable and deeper responses rather than a single monotherapy agent. 

And so, the first thing you would do as a team is you take a look at the scaffolds, you look at side effect issues, you look at what pathways maybe you want to trigger or not trigger, and, of course, look at all of the existing strengths and weaknesses in the existing drugs.

And that's exactly what it's doing.  

It's conducting a systematic investigation of the current landscape, identifying the limitations, looking at some of the current limitations like efflux, AGP efflux problem, drug-drug interactions, and potential issues with combination therapies. And so this is great. 

So it identifies the flaws and the challenges, and now it's going to start looking at designing an optimized analog using another one of its agents and tools. And so this is exactly the way a scientist would think. Imagine this is all being done in real time.  

And as we do it, we can now see that this now has access to over 128 different sources, 43 trials down to the open FDA, 74 different papers, and 5 other outside publications.

And it's doing all this in real time. So, just in the moment that we've started the webinar, now it's identifying the key problems, identifying the structural challenges, and actually, it is now iterating on actually creating a drug, which is quite interesting.

So, create the first drug.  

Obviously, this is the smile string, the weight, some of the key characteristics, total polar surface area, log p values, Roke table bonds, but it doesn't pass all the drug likeness filters, okay? So it says it's going to retry.

So, it says the ethylene glycol modification backfired. It added a hydroxyl group that increased TPSANHBond donors. And so, it's going to now iterate yet again.  

So this is really important because it tells you also if it fails in creating something new or if it at least thinks it failed because now you have that history of, okay, this was attempted, but it seems not to get through the in-silico filters.

And so we've trained its generative chemistry modules and its molecular analysis filters to think about creating molecules that have more drug likeness and constantly trying to correct itself.  

And so it's going to do one final iteration. So again, it passes certain filters, continues to have a couple of issues, and wants to have optimal flexibility and wants to have lower H-bond donors, which could be a trade-off.  

You could actually tell it, "Hey, Zeta, I'm fine with decreased CNS penetration. So go ahead and show me the drug." 

Again, we like CNS penetration or getting through the blood-brain barrier. We think it's key. It's a great characteristic to have if you can do it. But again, we can always tell Zeta, hey, we want to take that, claw that back, and not worry about HBond donors as much, and show me the iteration.

It's going to actually come up with a new SmileSting in real time.  

That's one of the great, great features of this. And again, as it does, it also updates the Knowledge Graph as well. So, we can start this conversation.

So here we go, the final phenyl hydroxyl changes. It's designed as a next-generation inhibitor. Here's the molecular structure, the smile string.

It compares it to the original and what the optimization is, decreased molecular weight, fixed and passed many of these rules, increased bioavailability, has excellent HBAN donors and acceptors, and then tells you how to design it, which is great.  

Current design, final metabolic optimization, predicted therapeutic advantage over at least some of the existing drugs, and elimination of dangerous drug-drug interactions.

Again, this is just one instance of Zeta. Imagine now you have multiple instances of Zeta, multiple Zetas that are trying to work on this problem in an afternoon. 

This is something to go after these Menin inhibitors and design a new drug, and then actually look at the combination potential. This is something that would take months, literally, and probably hundreds of thousands of dollars for biotech.  

And here, we're doing it in real time. I mean, literally in 20 or so minutes. And it gives us exactly how to now test this drug and what it's doing, and also gives us a development path. 

Computational validation, which we would do here at Zeta, deals with how to do chemical synthesis and in vitro validation.  

And we can actually ask Zeta now, as, let's say, a general research scientist,  

"Can you provide me with a detailed budget to develop this molecule from research grade to full GMP, along with an aggressive schedule?"

So, we've made our drug. We've investigated this need in Menin inhibition in pediatric. We know there's a high need. We targeted some scaffolds and ideas that we thought we could play around with.

We have the potential molecule that was created after 3 iterations, which is fairly quick. And now we're also going to have a budget.  

Again, all this is within 30 minutes. Imagine you're a biotech and you're doing the same thing. This is typically what takes months and months of work, including a budget.

This is a realistic budget we've trained it, and we've given anything that we do at Lantern for all of our drugs, and again, we've dosed over 100 patients across 3 different trials and some very, very challenging cancers.  

We've gone from ideas on a whiteboard all the way through manufactured drug products, all the way to dosing people in the clinic, to writing investigator brochures secure 12 FDA designations.

All that work that our team has done, all that is instilled inside Zeta. And Zeta has learned from it and improved it and does it faster. And then we've trained it from all the outside collaborations that we've done with our institutions. And we've trained all the outside literature.  

So, it's like you have inside of Zeta this unbelievable amount of expertise and resources and the ability to generate new knowledge.

Very importantly, you can take what has typically taken weeks, months, and sometimes years and compress that to days. And so now it's going to actually generate working right now on a budget. 

And so, it's going to use 11 different tools. It's creating a detailed budget that we can actually put into reporter mode and then take that budget and submit it to our Board, submit it to CROs, actually look at the budget, disagree, or agree with it.

And again, these budgets for some of these molecules literally can take weeks, having gone through this myself. And it's oftentimes pulling teeth from CROs to get these kinds of detailed budgets.  

So here, again, we can do this in a matter of minutes. So, with that, I know we're coming up against some time limitations. And I'm going to go ahead and ask folks to maybe provide some questions.

We have some questions already. So Matt, we have a question here.  

Could this look at failed clinical trials and explain why they may not have worked?  

Yes, that's a great question. Why don't we actually... Go to that specifically, start a new chat. So, I'm going to go ahead and ask it.  

"Can you review some of the most recent Phase II failures in rare cancer clinical trials and provide insight as to their limitations and why they failed?"

So, as it works with that, I'm going to go ahead and ask another question. Great question. Do we use this internally?  

Yes, we've been using it internally. We actually have probably have 50 going on 100 external users. It's growing pretty rapidly. 

So the knowledge graphs, people love the knowledge graphs actually, and love the ability to budget and optimize molecules, et cetera.

But yes, we use it internally, and we'll have some press about how we've been using it internally and some of the new molecules that we're doing, actually, in some very, very interesting novel biology as well.  

So going back... And now you can see it's getting information. And as you can see, it's also going through all these trials. And again, has a library of over 560,000, some trials at its disposal. So it's going to go through and start ranking these failed trials.  

Another question, I'm going to answer live, which works on this, is whether multiple researchers can collaborate on the same question. We have very exciting collaboration capabilities.  

So like I said, the knowledge graph, you can share the knowledge graph as an interactive file. And we are going to have an enterprise feature where we can have teams work on the same problems or sets of problems.

Yes, we'll have a lot of enterprise and team collaboration features that are built in, so people can all look at that. And also you can share it. You can download it as a PDF, which I'll also show you. So here we go, just got detailed answers on the failure question.  

Wow, this is really more interesting than I thought. I'm just giving probably more detail than I would have even thought about, which is great, and then giving us future recommendations on embracing basket trials, and I see cell therapy issues, which I would have said also, but I was not thinking in the domain of cell therapy.

I was mostly thinking, obviously, in small molecules, antibodies, and ADCs. But yes, cell therapy failures are massive. That could be a whole separate chapter. We can talk and go into Zeta. 

So immunotherapy response, heterogeneity, and sarcomas, manufacturing bottleneck in cell therapies, and molecular subtyping, statistical design match reality.  

Yes, it's a great answer, a very good detailed answer, and some good conclusions are not random events, but reflect systematic mismatches between trial assumptions and rare disease realities. This is definitely worth a whole article on this topic right here. 

It is a category 1 trial design flaw and category 2 enrollment barriers. So it's interesting. 40% of the failures are trial design flaws, which it gets into, including statistical power of this calculation.

Category 2 enrollment barriers, which are always tough, very, very tough. 

And so let me talk a little bit about some in the future. But yes, obviously, we can get into the trials. This is a very big question about how Zeta knows when it has enough evidence to stop researching and provide an answer.

Yes, that's a great question. I mean, obviously, it exhausts its own knowledge tree. I think you can always improve it.  

So there's a bit of juggling in terms of whether it's going to start filling the context window. But yes, there's definitely a lot of engineering behind the scenes in terms of the data and the context window, the number of matches, and the number of tools being used. 

So again, it has various modes. It has an investigator mode, which recurs more and uses more tools in parallel, and then the explorer mode, which is about 3 recursive steps and uses fewer tools in parallel. 

So a lot of the knowing when to stop comes from the mode that you're going to be in. But again, you can always go back and say, I want you to go deeper or less or I want you to go into a specific tool.  

The other way that we design Zeta in terms of enough is the scientist role. Do we have enough in terms of whether you're playing the role of a clinical oncologist or a biomarker?

Do I have enough to draw conclusions about what a translational scientist needs to put out or a medicinal chemist, et cetera? So a lot of it is driven partly by role and then partly here by the depth of the question.  

But again, if you're a large pharma or a big biotech, you may also want to add your own external resources to it. You may have your own library, may have your own compendium, your own pharmacy and cyclopedias, your own data sets.

And so in this version, you can actually add your own files. So if you want to teach Zeta something that it doesn't know, you can add a file and say, Zeta, please read this and tell me what you think, which is a great thing. You can also output from here.

So again, it has the knowledge graph that you can share, but very importantly, it also has a PDF writer, so you can actually get the PDF document.  

So let me go ahead and stop sharing and go back to some of the enterprise features and discuss that. 

So we're going to have more team workspaces to answer that question, where you automatically share knowledge graphs, we have collaborative annotation, and better session history.

We're going to have social features where you can have credential researcher profiles linking their affiliations and their work, a multiuser investigation session, so that if, let's say, myself, John, Rick, and Shelley all want to work together, we'll be able to work together remotely.  

We'll have personalized feeds so that as you come back, it will know, hey, Panay, we're deep into these menin inhibitors.

There are some other things that are relevant, recent, or breakthrough. We'll also be able to white-label this and allow API access and custom integrations. 

So these are all the future. And we really think, obviously, drug development is a collaborative exercise. And Zeta is built for that. And Zeta really is to make scientists great.  

Every scientist wants to become better and better and stronger and faster and have more impact in the work that they're doing. And these kinds of multi-agentic tools, these co-scientist tools, are really a wonderful example of how AI can be used for good.

And we continue to enhance it. We'll have deeper data modalities. We'll have more pathway mechanistic knowledge, greater biology models, and also mobile optimization and more tools to support IND preparation, filings, et cetera.  

So this tool will become all enhanced. Our vision is that for this to be almost like the Bloomberg for medicine. Anyone involved in drug development or in the development of medicine and biomedical research, at least in cancer initially and perhaps then in other categories, will have Zeta at their disposal.  

So I want to thank all of you guys. I know I've gone a few minutes over, but thank you very much for walking through this. And I think multi-aggentic AI isn't just the next step. 

It's really a fundamental reset of what is possible. WithZeta is on the leading edge of representing how this can be a breakthrough in rare cancers, and eventually, we think in medicine, and how we can bring human insight and expertise, and autonomous scientific intelligence.

And this is how this will coevolve together and bring us, we think, medicines faster, cheaper, and with greater precision.  

So with that, I'd like to conclude this morning's webinar and urge you guys to go to withzeta.ai, sign up, try it out, get your companies to subscribe, and let's push Zeta to become even better and better and a more powerful tool, especially in rare cancers where these drugs are definitely needed.  

All right. Thank you very much, and thank you all again for joining this morning.

This concludes today's event. Thank you for joining us today.

Well, first of all, I want to thank everyone for joining us at 8:30 Eastern to see what I think is going to be probably the first real public unveiling of this next-generation withZeta AI platform. Many of you have had some of the fortune to actually see it in the past in demos. Some of you are actually users, which is even better. But we've now come up with the next generation. And let me walk you through some of the features.

Just as a reminder, withZeta.ai started as an initiative at Lantern Pharma for us to think about rare cancers. And we've been particularly, I would say, both gifted and focused on trying to take our therapies and focus them on challenging or rare cancers, partly as part of a development strategy, but partly also their white space. There are a lot of cancers that basically have no standard of care or a highly unmet need -- have high unmet needs. Zeta is actually one of the rare stars. It's a type of star, Zeta star, and they're very rare. And so as we kind of thought about this project, we codenamed it withZeta, initially Zeta and then withZeta because as the power of the platform increased, it became more than just a big data platform. It became more than just a RADR platform. It became more than just a platform for going out and gathering information and putting it into nice tables. It does all that.

But it actually started having the ability to reason. We use natural language processing and we tied all our tools together. And it actually became almost like a colleague, the co-scientist. And so we said this is really the withZeta project. And we focused on rare cancers. And there's over 438 rare cancer areas that we'd like to conquer. And they're going to be accessible to any investigator, any scientist anywhere in the world. And that is very, very important.

So the market around rare cancers has largely been not very successful. It's usually done as an afterthought of other drugs. You'll have a drug, you'll have a molecule and then maybe you're fortunate enough to think about the mechanism, perhaps working in a rare cancer. And what makes rare cancer challenging is that the markets are tiny. But the deaths in aggregate are almost 30%, and there's about 5 million rare cancer patients globally, very often without any therapies. And oftentimes, that trials, oftentimes that enough data for someone to convincingly go after something, and what we did is we accumulated all that sparse data, all that information, all the disease models, all the approved drugs, all the clinical trials that have failed, all the clinical trials that have succeeded, all the literature that really makes sense and put that all together in one place to help us reason through all this.

So as the tool became more powerful, largely because of Reed and his team doing the data engineering, we're able to develop a tool that became a platform that became a co-scientist. And it really works with us, and we'll talk a little bit later about what it's done not only for Lantern, but for others. And on top of that, we're able to not just repackage information, but generate net new scientific knowledge, and we'll see this today. We can bring in real-time literature search, do molecular pathway analysis, do all kinds of molecular analysis around a molecule, actually generate new molecules using a 24 billion parameter LLM focused just on molecular structures that are medicinally appropriate. And then actually also live and have all that knowledge as a knowledge graph. And so -- and it builds on that knowledge graph, and we'll see that.

We designed it so oftentimes, people want really, really in-depth information. So one of my colleagues, Chief Scientific Officer, he may not want 1 or 2 paragraphs. We may want 6 pages. Maybe there's others that want 19 pages. Some people may want to report. Some people may want to have quick answers that allow them to then have more investigations later. And so we created research modes, much like you'll see in many GPTs. You'll have a quick explorer mode, which is actually very efficient and there's multiple rounds. So it's really a scientist. It's really not, hey, how's the weather? It's really still thinking through some complex initial exploration. An investigator mode that recurses through 10 investigation rounds with lots of parallel tools.

And to put that into context, imagine if you're -- any of you who are chess players or game players, thinking through like 10 steps ahead and oftentimes 3, 4, 5, 6 games at a time and 10 steps ahead, it becomes quite complicated. I know I can't do it. Maybe there are some super humans that can do that kind of stuff. But imagine now being able to do that in some complex disease questions in cancer. It's really an amazing thing to see. And it's not hours. This is literally minutes. And then finally, you can put all this together and you can share the knowledge graph with others, but you can also share the report.

And on top of that, what we've done is thought about what are the different types of tools, what are the different kinds of people that it takes to build great medicine. And it takes a lot. Not only do you need general research scientists, but you need people who are highly trained at molecular feature analysis and scaffold design and ad lead predictions. You need the medicinal chemists. You need to have people who think about the trials. How do you develop a protocol? What's working, what's not working? What are things that are really endpoint strategies? Or what's appropriate from a regulatory standpoint? Then you also need the concept of a real person in the clinic, an oncologist who understands the landscape, knows what standard of care, when to deviate, when biomarkers are useful, when they're overkill, and really what is the strategy to go from a molecular concept to a clinical translation. And clinical oncologists are so wonderfully gifted at that, and they're seeing patients.

And then finally, the translational scientists. What is the biomarker validation that we need? What's the correlations that make sense? What's already been done? What's worked, what hasn't worked? And so all these co-scientists are there at your disposal.

One of the things we try to do also on top of that is we built proprietary knowledge bases, things that we curated that we as Lantern scientists and Lantern developers think that this is the way we want to make drugs. This is what we would expect from our colleague. And so we've kept the framework simple, bulleted, detailed enough and encompassing always pushing forward to get to patients. And so we'll talk -- we'll take a look at the demo today, but now you can actually tune Zeta to take on the personalities of these various co-scientists. And maybe, Reed, you can expound a little bit about how we did that and then where that's going next?

Yes. So for those of you all in attendance that have been fortunate enough and we're grateful that you've been with us through the beta experiment, you've been discussing with the general research scientists. And a lot of the feedback that we got was contradictory in some ways, where some users wanted super in-depth clinical trial protocol design, while others wanted just quick, generate a molecule and let's iterate on the molecule itself.

And so it wasn't that one of those responses or feedbacks was more accurate than the other, but it really highlighted the need that there are these different co-scientist personas that personify or embody the type of scientists that you want to be discussing with. So the way that we've done this is, all of these co-scientist personas have access to the same underlying databases and tool sets and skills. What we've done is tune which tools that they are preferential towards and how they then organize those tools into sequential and coherent workflows to actually get to answers.

And so the medicinal chemist is going to focus on actual scaffold based or de novo molecular generation with very rigorous chemical -- actually computing the molecular features of those generated compounds to validate, whereas the clinical trial strategists and clinical oncologists are going to rely far more heavily on AACT and our clinical trials database and our rare cancer knowledge base. So the same information is available to all of them, but it tunes how specific they are to various domains.

And this is really what happens in the real world. So if you're a knowledge worker and you have a lot of things, you are multidisciplinary, everyone is going after the same problem, but the tool set that you bring and the knowledge base that you bring oftentimes is slightly different.

Think of a house, the person in charge of plumbing obviously knows how a house works, but they're going to focus on the plumbing set and the tools. The electrician, same thing, they're going to focus on their tool set. And so in any multidisciplinary knowledge world, they all are focused on the same problem, but the tools they bring out and the tools they prioritize are slightly different. And that gives you some really interesting different responses, especially as the responses cascade over multiple iterations. And we'll talk about, this is very theoretical, but what we're thinking about in the future with these different scientists that are powered this way, is to actually have the scientists talk to each other. And maybe, Reed, you can expand upon this concept that we're beginning to play around within build.

Yes. So now that we've expanded the platform to embody these various co-scientist personas, the next step for us is, two kind of scopes. The first one is that if you're talking to the general research scientist, there might be a question where the general research scientist wants to go off and consult the medicinal chemist. So in the general workflow without any intervention by the user, Zeta can kind of route the question towards the appropriate expert, essentially at a higher level of abstraction, but emulating the mixture of experts model in LLM engineering.

But then beyond that, what we're considering is this idea of the simulacrum, which by a strict definition, it means that the simulation or emulation of something becomes so real that it replaces or appears just as real as the original thing that it was set to imitate. And our intention with this is to create kind of a roundtable conversation of panelists where these various co-scientist experts are talking to each other with a high-level prompt to kind of seed question of what you'd like to discover or ask about. And then the human is still there in the loop. The human is one member at that round table. But now instead of you with a single co-scientist, it's you with a collective of intelligences.

Yes. And this is very important because we really think this eventually will allow companies and teams to have an entire biopharma development team in a box.

So imagine you as a Chief Scientific Officer or as a Lead Research Scientist or even as a CEO, and your job is to create some unique novel therapy or to find something interesting, not only can you have one team of scientists at Zeta, our eventual vision is you can have multiple teams going after these problems. And we'll talk a little bit about that later on. But let's get down into some of the things that we're going to talk about today in the demo and what you're going to see real time. We're going to give you -- Reed and I are going to provide you an orientation to the broad platform, features, things, buttons, just how do we use it? We're going to do some queries on some rare cancers to show how the investigation works. We're going to see how the knowledge graph can be used and how it's a great feature, especially for teams and organizations. We're going to do some real-time molecular analysis, including some really detailed predictions and then actually maybe think about generating a novel molecule, either based on a de novo structure that we uncovered today or a scaffold that we think is promising or some new target that we think needs to be interrogated. And now we're open to that.

Now remember, I want to remind all of you that this is not canned. This is a live demo. Now some of the questions, of course, Reed and I did go over in advance, but this is live. You're going to see the system live, and we've engineered it also with a lot of cost in mind. So everything that you're seeing is being terraformed real time, and that's very important. This infrastructure-as-code is also what allows us in many ways to have some controls around performance and cost. So let's go through for our attendees now to the live and get rid of the PowerPoint, and move over to the system.

[Presentation]

Okay. So this is where you would land. So this is where -- if I've set up an account, I would go here. And I could pick the persona or keep it general, and then I could ask a certain question. So is there a question anyone wants to type that they want to ask today about a rare cancer, anything specific?

Not sure if we have a question box for the attendees or any way to pull that. I would say just run with it, Panna.

No problem. So let's ask it about over-expression of EGFRvIII. What rare cancers over-express EGFRvIII and still are in need of improvement in therapy? So let's ask you that question. I'd like to go into investigator mode. And as you can see here, we're doing the general scientists. I could pick any specific scientist. But we're going to do general initially, investigator. So these are some of the things that you may want to think about as you type your question, like what mode, et cetera. And of course, you want to be cognizant about your credits. So you'll have your credits here. And obviously, the more depth, the more credits that will get used.

But let's go ahead and ask this first question. And EGFRvIII is a topic. It's definitely a target. It's definitely a concept that cancer scientists think about. And so now it's going to start thinking. And unlike a lot of AIs, what we're trying to do is walk you through the reasoning process and what tools are being used and how it's investigating it. So just imagine if you had someone in your team and you say, "Hey, give me what's going on in EGFRvIII in these hundreds of rare cancers because we want to prioritize it because we have a molecule that we think really does something in that space.

Great. And then more importantly, it's going to read all of these papers or review these papers real-time. This is something that would take hours, maybe days, unless you happen to have a colleague that already knew everything about EGFRvIII in rare cancers. But now it's going through and it's giving you kind of real-time what it's thinking about, what it's researching and reviewing. And then it's going to use tools to execute across that in terms of searching, in terms of reasoning, and then it's also going to give you the reasoning process. And this is very important because the reasoning process, if you're a company, let's say, you're Amgen or Servier or BMS, you may want to tweak it. You may want to say, "Hey, this is the way we like to think about things. We'd like to put this higher, this lower. We'd like to add these concepts or we'd like to give it a certain flavor and index these internal data sources first". And these are all things that can be quite readily tweaked for the enterprise features, and we'll talk a little bit about that as well.

So now based on the initial findings, it's going to give us, again, an answer.

Panna, by the way, if you want to collapse those sources, you can click the source on the top right there.

Yes. So we can see it's gone through about, what, 30...

41 in this case. So those are all the external resources that it's querying. That's going to be the list of clinical trials that it's referenced, PubMed articles that it's found the abstract for and DOIs.

And then additionally, if you wanted to see other internal proprietary databases and AI modules that were queried you could go into the research queries and see exactly what tools were executed in this process.

And so now you have a fairly detailed answer from Zeta that says here are the therapeutic gaps, primarily affected GBM in rare cancers, giant cell GBM and gliosarcoma, similar GFR fusion patterns, overlapping mutation landscape.

And what you can notice is the colors, the colors are very important because as drug developers and people in science, we'd like to have things that pop out as we read thousands and thousands of words, right? And so things that are disease are in red, things that are biomarkers are in green, literature references are reversed in the gray and white, drugs are in orange. And so quickly, as you see these colors, as you scan the landscape, you can see, this is a drug and we look at that, and we dig into a disease. And we look at a trial that's being referenced. This is a [indiscernible] vaccine, which is spectacular Phase III failure, okay, immune checkpoint. So it's going through the litany of drugs being used, the mechanisms driving some of the resistance that it sees. So this is, again, thinking like a drug developer. It's not just giving you a summary, it's thinking about what can I do to improve the landscape. And part of that is always looking at what is resistance, what's not working, lysosomal sequestration and emerging strategies to overcome resistance, multi-antigen targeting, combination immunotherapy, TME modulation and CD47.

And then tells you what the unmet need is, young patients, bad prognosis. So it gives you a pretty good overview. And again, it gives you insight into that space. Now what it did as it did all this, which is very important, is it generates a knowledge graph. So much like any knowledge worker, you think about these concepts and how they're related to it. Sometimes there's concepts that float around that aren't perfectly related. But you can see EGFRvIII is over-expressed in GBM. GBM is treated with these drugs in the orange. So orange is drugs. There are certain proteins like LAG-3 or SERP or TIM-3 that may be involved. So again, you start creating the knowledge graph, and this may then get linked to other concepts like this disease, giant cell glioblastoma or gliosarcoma. And so this landscape continues evolving.

So let's ask it about -- one of the new features that some of you that have been using Zeta in the past will see, we'll demo this next Reed is, we now have given Zeta, much like a lot of the AI tools, the ability to actually prompt you with questions, ways to clarify or improve the question. So sometimes if there's something that it wants clarity on or it says there's an overwhelming abundance of directionality in terms of the tree or the -- it will come and say, do you mean A, B, C, D or can you clarify? And that, I think, is also a great improvement because that's how people interact, right? You ask a colleague question like, Mike, hey, Mike, what do you think about this target, it was like, well, do you mean about it in the context of the intra-tumoral environment? Or are you talking about it as a therapeutic target? Those are the questions that any kind of co-scientists, your colleagues would have. And so now Zeta does that as well today.

So now it's telling you it's looking at the late-stage landscape. No gliosarcoma-specific trials, although it does find some current standard of care. So this is great, and it has some evidence. This is going way back to the SCUBE trials.

Those are glioblastoma trials.

Yes. And then this is [indiscernible] it's including it, recent failed approaches. That's great. So let's say -- let's look at one of these in detail. In fact, this is a great place maybe for Zeta to say, I'm missing something, right?

Also probably a good chance to switch the co-scientists you're talking to.

Yes. Are there any deeper questions I should be asking or that I am missing? And let's switch to -- should we switch over to a biomarker and translational scientist. Let's try that.

That is an interesting thing. You can switch between depth and the co-scientists you're discussing with, middle of a conversation, seamlessly. So you don't have to start a new conversation to change the co-scientist persona that you're discussing with, kind of the hat that Zeta is wearing in that moment. So it makes it pretty easy to evolve a conversation thread in the way that we're doing here.

And again, as you can see, the knowledge graph continues to evolve. And so if I want to export the knowledge graph as an HTML file, as a JSON file automatically in store, which is very important because then it's machine readable for any future use. So now we ask it to -- there are certain deeper missing questions and we gave it the persona of the translational biomarker scientist. And we'll see what it comes back with. And then after that, what we'll do is we'll dig right into actually designing -- potentially designing a molecule or designing a trial and actually giving us a budget and time line and a protocol.

So any of those venues -- and the important thing about the way we provided the tools and capabilities of Zeta is that we thought about it in the landscape of what do we as drug developers need to do. We need to put together disparate results. We need to think about scaffolds. We need to optimize molecules. We need to look at how molecules are made. We would love to do de novo design, and we do that also now. We'd love to maybe benchmark that. Future versions will include integrated bio-informatic tools, which obviously people will use and pay for and also be able to do it on any of the data that we've already pulled or even your own data. So now there's a mode you can actually upload information to Zeta as well, and we'll talk a little bit about that feature after this question.

Currently limited to PDFs and text files, but this will quickly be expanded to actual source data that you might want to have uploaded to enhance some of these questions for your own sake.

And you'll be able to do it eventually into your own secure Amazon or Google Cloud as well. So let's imagine you have a folder full of data on pancreatic neuroendocrine tumors. You want to say, "Hey, Zeta, I want you to go ingest this proprietary data. It will just be between you and Zeta. It won't train the rest of the -- machine won't learn on it. It will be your own private instance. And so those are things and features that we're building out now for future enterprises as well.

So as you can see, this is great, deeper strategic questions you should be asking, therapeutic resistance questions, translational strategy, which is key mechanistic biology questions, clinical trial questions. And you can see here, it's used over 14 tools. So a lot of tool usage in this one to kind of give us these types of questions. And again, no dedicated drug development, lump with GBM despite being -- having distinct biology, worst treatment response, unique molecular features, no biomarker-directed trials, wow.

So here's a clear gap, right? Prognosis reality, real-world outcome data, we can go look at it, terrible MOS and it gives us outcome in some of the trials, EGFR TKIs, ADC, that's an interesting ADC. That's pretty late-stage. Again, gliosarcoma is very, very challenging cancer.

So let's look at this drug here, Afatinib. It's an EGFR TKI, which we -- a lot of people know about. And we'll ask Zeta a really important question, can we create an improved version of Afatinib that has combination potential with a checkpoint?

Might want to switch to the medicinal chemist for this mode, and that brings up the valid point what I mentioned before, if we're working on having Zeta actually do the routing across the various co-scientists appropriately on its own.

Now the future of this isn't just your one-on-one interaction with Zeta, I'm looking at this question, but also now you've created kind of a unique knowledge pathway that you can share internally in the organization. You've gotten tons of sources. So you have your own, what is this now, 90 some...

97, it looks like, yes.

Yes, that it's now looking at and ingesting. And more importantly, you've got a huge knowledge graph that's growing. And so all this can be shared and you can come back to it as well. So Zeta thinks it's an excellent question. Good. I'm glad. This is precisely the rational drug design challenge. I'm glad that's why we gave it to you Zeta.

We like to think so.

So now if I had to ask this question, let's say, a CDMO or my medicinal chemist or team, they'll go off for weeks or maybe months and think about it, think about the scaffold, what works, what doesn't work, what are the features that they would want to improve? What's going to make it work best with the checkpoint? How do the checkpoint inhibitors work? Which one features in the checkpoint would I want to synergize with so that I can have the kind of synergy scores and values that combination regimens have. Big complex questions, and we're giving it like buckets of different kinds of questions that can take drug developers months to even come up with answers that they feel are rational, grounded and have the ability to get validated in the lab.

And that is very important because you're going to ultimately ask the question, well, can I take these concepts to the lab? And we check for that. So anything -- any molecule that comes up, it checks, can it actually make it and you design around a translational biomarker strategy. It things like a scientist and says, is there a proxy or are there methods? Is there a biology? Is there a publication that, hey, this is something that I can go take off and do through proteomic work, tissue microarray work, standard sequencing work. And so again, validation is very important. So it has the knowledge or awareness that the things that I'm going to provide to you as a user need to eventually be put and tested into the real world.

So as you can see, it's beginning to use lots of tools, validating SMILES strings, computing molecular features, and it's going to do that, again, real time, its doing this in minutes, not days. And after it does all that, it also will update the knowledge graph and update the bibliography and so you can share that. As we look at that, we'll go through a couple of other features. So you can start your conversation. So if it's a conversation you want to have, you can see EGFRvIII therapeutic resistance. I've started that. And so that's kind of top of mind. All your recent conversations will be here, as well and then your research toolkit and then kind of what rare cancers.

And so we have a rare cancer tree, oncology tree as well that's built into the tool. So you can see how all the rare cancers are grouped, classified and what families they belong to. Quick start guide as well, which we urge you guys to take a look at and also certain account features. So while we wait for it, we can go to the account features. And so the account features, you can see how you've been using it, how your credits break down, what conversations, persona et cetera. So all this is all available. So let's go back to our chat.

That's a different one.

There we go. All right. So did it create the molecule here?

I think because you went to the account page, it stopped the stream.

No. Okay.

So just ask the question again. Yes, I should have warned you that the toolkit, rare cancers, all of the other modals pop-up and will keep the conversation stream running. The account page will kill the web socket, unfortunately.

Well we're learning real-time. It makes sense the account page would kill the web socket, yes. Makes sense. Can you probably want to do a -- do you want to go to account page?

There's a way to make the account page, not break the web socket, future version update. So yes, while this is going, unfortunately, for the second time, one thing to note as well is the way in which this medicinal chemist is going to go about this.

So as a scientist, you are not going to one-shot a brand-new molecule, whether it's de novo-based or scaffold based. It's going to take multiple iterations of trying a modification, recomputing its molecular features and seeing what dials got turned. Once you know that, then iterate a second time. So Zeta is going to do both of those. First of all, it's going to -- we have several tools, one in particular, which is computing the molecular features of it. And this is a subcomponent of our PredictBBB application, but it computes over 90 physiochemical properties of the chemical itself, such as molecular weight, hydrogen bonds, rotatable bonds, total polar surface area, a whole bunch of them as well as drug likeness scores and filters and then use those as guardrails for prompting Ether Zero, which is the smaller open source large language model, which is actually doing the SMILES modifications and reasoning in English.

And so it's this back and forth from real-time computation of molecular features, validating how a compound actually behaves and what it looks like physically to then modifying it with Ether Zero and then repeating the loop until you've arrived at optimal features and targetability.

And as you can see now, it's up to 8 tools that's going through. And it obviously has it marching orders to try to use the scaffold around this targeted TKI, but improve the BBB of it and more importantly, also improve its potential to be synergistic with checkpoint inhibitors. So kind of given us some concepts, and it's going to come back and tell us if that's doable, not doable. And the thing about the co-scientist personality also here is that it likes to tell you when it can't do something.

So unlike a lot of LLMs or chat agents that try to force an answer to please you, Zeta is not in that category. If they can't do something or if the drug is good enough or it doesn't seem like there's something doable, it will come back and tell you exactly that, which is great because it's more efficient that way and now you can go through a different strategy.

As it streams the answer, you can see how it's thinking about it, is actually decent, real barriers are -- it's molecular weight too high, LogP too high, excessive flexibility, the rotatable bonds. So it's adding methoxy group to increase total polar surface area compared to completely wrong, then it says it wrong. New optimization strategy, keep the halogens, reduce [indiscernible] dimethyl, preserve the core, change the warhead. So now it's going to actually look at all of these very, very specific things. And it came on the optimization strategy by looking at all the molecular features, as Reed talked about.

So the molecular feature tool that we have that PredictBBB, but also predicts tons of other stuff is looking at that analysis, feeding it to the other LLM. The LLM is then incorporating that for potential de novo design or scaffold improvements. And then it's going to then continue iterating also because we want the synergy with checkpoint. And just in terms of time, I want to make sure that we hit on a few other things. So it's going to continue working on this in the background. It's going to rephrase with some stronger therapeutic context, which is great. It's now up to 16 tools.

Yes, it's still going.

So as it continues to move, which is really fantastic, why don't we go over to some of the future things that we're going to do? I'm going to leave this in the background, Reed.

Yes. Just go to the PowerPoint.

Yes, I go to the PDF. So I'm going to do that. And then I'll keep an eye on when it's got done. It's making -- it's now -- it's through a second iteration actually it says. So it may be done quickly.

So again, now we're generating a novel molecule, kind of the end of some of our real-time discovery. But what's next? So as we mentioned, enterprises, we're going to have features to have team workspaces, role-based access and audit trails, custom integrations with their internal R&D platforms and then white label deployment for big partners because they'll have their own cloud, they'll have their own data resources. And of course, they don't want to share it in the public with Zeta. We're going to have social features to have shared research environments for teams, credential research profiles, multiuser investigation sessions and then perhaps even feeds that are personalized that tells you what new data and information is available.

We continue to enhance our tools, broader data modality support, potential IP landscape analysis and then increased integration with some of our proprietary tools like our pathway mechanism knowledge base, expanding the literature around the -- access around the biology models, specifically our ADC squared model, more disease oncology around rare cancers and then synthetic lethality and combination therapy prediction, which is a very important category. And of course, key things, optimized for mobile, structured on-boarding so that people can understand all the tools and capabilities and perhaps even in the future, regulatory tools supporting IND prep, therapy designation framing and actual document production. So these are all the things that Zeta will do. Let's go back and see how Zeta is doing in the molecular generation. Let me stop that and go back.

Great. Well, that's good. So in the time it took to make a couple of -- show a couple of slides, it's created an optimized analog with enhanced BBB penetration and checkpoint inhibitor. So as you can see, it benchmarks it against the original improvement, what its targets were and did it achieve the targets. It's at the warning threshold for rotatable bonds, which Afatinib already was anyway because, in fact, that's one of the challenges with it, but it's improved the drug likeness. It passed 4 out of the 6 drug likeness filters that we've got. It made some modifications. It tells you the modifications just like a scientist would tell you. It tells you why it's going to be synergistic with checkpoint, which is great. This is really key, what the mechanism and activation pathways are.

And then very importantly, it will give you some combination ideas. Dosing and sequencing strategy. So again, we've really designed this to, not just give you a one-off answer, but then push the answer into getting it to patients and trials. And again, it may be 100% right, it may be only 70% right. So if you have multiple people that look at this and you share the knowledge graph, a few hours of work with Zeta in a couple of days, you can do what traditionally has taken months. And more importantly, you're not going to miss key ideas. And it will tell you how a biomarker-driven patient strategy. So I can take that patient strategy. Imagine taking this, cutting and pasting it and putting it into the biomarker mode and saying what are some disadvantages of this approach or advantages of this approach. And I can also now take all of this and create a PDF of this entire thing. And so that's also very cool. So I can take this and just save it as a PDF. And so it will save it as a PDF. And as you can see, it's also now updated the knowledge graph also. So the PDF just downloaded. Let me bring it up, which is the PDF just downloaded onto my desktop somewhere. And since I have a very organized desktop, it will be very easy to find.

I'm not sure if we're seeing the window.

No, I got to switch over to the window.

Yes. While Panna is pulling that up, one thing I just add to the mention of this is that Ether Zero has now proposed this scaffold-based modification of -- for a new molecule. And then there are many different ways that we could take this, one of which being use Ether Zero again for actual retro-synthesis analysis and propose within the medicinal chemist a way to actually synthesize the molecule. Otherwise, as Panna said, we can go into biomarker strategies or create a whole clinical trial protocol around it. And again, that's where you can switch between the various co-scientists for the type of response you're hoping to get.

And then it will feed out a nicely laid out PDF with Zeta. It will have its appropriate disclosures. You know that this is really not a medical advice. It's not diagnostic information. It's really for specialized research and its information. So it tells you what its goal was, what its approach was really just a lot of the things that we covered. And again, keep preserving some of the highlights that you can tie all that back to the knowledge graph that you've got. And it's got the strategic recommendations as well and dosing and sequencing strategy. And again, every one of those areas can be improved.

So you can go back and then test that. And again, our vision is that you'll have maybe an initial response with this kind of depth of 9, 15, 20 pages. And then you feed that to the entire community of scientists and you have them discuss and talk about and come back to you. So imagine if you're a team working on the future of metabolic inhibitors or metabolic pathway drugs, you can have multiple teams, all working like literally within hours and days to go attack problems that traditionally have taken scientists years. And our vision is what's taken 50 years in the past in cancer research and cancer drug development. Well, now in the future in the world of these multi-agentic infrastructure only take a few years.

And this is really the CEO of Anthropic pushed out this concept, but being able to do in 50 years -- be able to do in 5 years, what used to take 50 years. And that's what's possible with these tools. And so with that, I'd like to thank you guys all for joining this morning. And we'll maybe open it up to any questions that people have or anything that we've got a few minutes remaining that people want to discuss, but I urge you guys to take a look at withZeta and withZeta.ai. We have a couple of tiers of subscription available. And for those of you that on the demo and knowledge accounts, it's great, keep using it. We'll have obviously the new version in production now. Is that right, Reed?

Correct. It is fully live in production with all of the new -- the latest and greatest you saw today as well as full payment integration. So if you were a user in our beta tier for early access, your credits and all of your access has been retained. You don't lose access now that we have launched to live. There is a grace period of about 2-months that you still have full access to the platform. And I'm not sure if we have it set up to receive -- we do have Q&A. Okay. Great.

Yes. I want to take some of these questions and Reed, I'll let you know.

So we have an anonymous attendee. AI tokens can be very expensive, how profitable are they?

The subscription plans are very, very, very profitable. Again, as I mentioned early on, we didn't dig into it. We really optimized around the token use, and we're -- we've got a lot of room we can continue improving it, but it's a lot of it has to do with credits and knowing what tools use, which credits. Also, a lot of our work isn't in inference. It's really in large quantitative models, LQMs. So all those molecular feature analysis that is the underpinning of molecular design, a lot of that is not done on inference chips like NVIDIA or Grok, those are done on traditional chipsets that are actually much faster and cheaper. So things like RAG and response and a lot of that uses inference chips, but a lot of the hardcore chemistry and some of the biological work where it's data-driven doesn't need the same kind of tokenization and is fairly inexpensive. But good answer. But yes, the subscription plans, again, are also guard-railed through credits. So that's been answered.

Next question, does it access information and data on devices? Good question. I think in the future, we could, but the current version does not access information on devices. One could upload it. And Reed, do you want to talk a little bit about the upload feature?

Yes. We do have -- this is again, the latest version that is now on production withZeta.ai, but you can upload text documents. So document files, text files or PDFs can be uploaded and included in the thread of a conversation. So it's not added to the global Zeta knowledge base or like a personal knowledge base of your own, yet. But what it does is it just adds the context or the content of whatever text file you've uploaded into the current context of that conversation thread. So that's how we've handled this at the very starting point for right now of just allowing users to upload PDFs if they want to upload an academic literature and ask questions and cross-reference against it. But that's currently the extent to which Zeta has any access to on-device information.

Good question. But in the future, you can imagine a big drug development team may have a sequencer or a proteomic installation or something where they're generating data on the daily, that can all be stored in their cloud. And then eventually, Zeta would have access to that in a real-time way just for that company or for that team. But yes, right now, we've guard-railed it from going in the broader system because we don't want the chance of actually any information not being used properly for any of our users, but more importantly, people wanting to try to poison the system, too. So the Zeta knowledge base is very, very closed. But good question. Any more questions? If not...

Yes. Panna, there was one more. Do you mind disclosing current subscription revenue and prediction of it in your next quarter? Disclosing subscription revenue.

We're just launched, so probably too early, but we do have some AI revenue already. So that's exciting. But we'll talk about the prediction and the growth, I think as the AI platform fully launches. We have a major launch coming up at the American Association of Cancer Research in San Diego later this month, and we've got a lot of demos already set up there. So I think this will probably do very, very well given the extraordinarily flexible pricing plans that we've launched with. And again, these will only increase as the tool set increases rapidly.

Again, we've also -- are hiring in our Center of Excellence in Bangalore to increase and hit all the road map functionality. And again, I think for the future of drug discovery, co-scientists are really going to be the way that teams work and get the maximum amount of productivity and the maximum amount of knowledge efficiency and use.

So again, thank you, everyone, for joining us this morning. We'd be happy to answer questions. Again, you can reach me at ps@lanternpharma, and we'll be happy to do customized demos and give you more details. Thank you very much again. And Reed, thanks a lot for joining.

Yes. Thank you all. There is also a contact form on withZeta.ai, so you can reach us that way as well.

Thank you.

Thanks, everyone.

Good afternoon, and welcome to our fourth quarter and year-end 2025 earnings call. As a reminder, this call is being recorded. [Operator Instructions] A webcast replay of today's conference call will be available on our website at lanternpharma.com shortly after the call.

We issued a press release after market close today, summarizing our financial results and progress across the company for the fourth quarter and year ended December 31, 2025. A copy of this release is available through our website at lanternpharma.com, where you will also find a link to the slides management will be referencing on today's call.

We would like to remind everyone that remarks about future expectations, performance, estimates and prospects constitute forward-looking statements for purposes of safe harbor provisions under the Private Securities Litigation Reform Act of 1995. Lantern Pharma cautions that these forward-looking statements are subject to risks and uncertainties that may cause actual results to differ materially from those anticipated. A number of factors could cause actual results to differ materially from those indicated by forward-looking statements, including results of clinical trials and the impact of competition.

Additional information concerning factors that could cause actual results to differ materially from those in the forward-looking statements can be found in our annual report on Form 10-K for the year ended -- March 30, 2026 and Lantern Pharma does not intend to update any of these forward-looking statements to reflect events from circumstances that occur after today, unless required by law. A webcast replay of the conference call and webinar will be available on Lantern's website.

On today's webcast, we have Lantern Pharmacy, EO, Panna Sharma; and CFO, David Margrave. Panna will start things off with introductions and an overview of Lantern strategy, business model and highlight recent achievements in our operations, after which David will discuss our financial results. This will be followed by some concluding comments from Panna, and then we'll open the call for Q&A.

I'd now like to turn the call over to Panna Sharma, President and CEO of Lantern Pharma. Panna, please go ahead.

Good afternoon, and thank you for joining us today to hear about our fourth quarter and fiscal year 2025 results and corporate progress. As many of you have heard me say past, computation and AI-driven approaches are increasing their presence and usage at both large and emerging pharma companies for all facets of drug discovery and fundamental biomedical research. The future of medicine is going to be intimately involved with AI technologies and AI models. Our leadership and the innovative use of and machine learning to transform the process of developing precision oncology therapies should yield significant returns for investors and patients as our industry matures and adopts an AI-centric data-first approach to drug development. .

2025 was a defining year for Lantern Pharma. We achieved clinical validation, we believe, across multiple programs while establishing the foundation for our next phase of growth. We believe that we had encouraging and a unique development with LP-300 in the Phase II HARMONIC observations, combined with also a successful Phase Ia completion for LP-184 clinical trial and most recently, an FDA IND clearance for a pediatric CNS cancer program through Starlight Therapeutics. We believe all of these represent a formational milestones that validate and strengthen our ad driven approach to precision oncology.

Today, we're sitting at a point in time where all of our initial ideas and concepts regarding our molecules have now been dosed to patients successfully in some manner in both Phase I and Phase II trials. Also, our full year financial results reflect disciplined execution with a 19% reduction in total operating expenses year-over-year, even as we advanced multiple clinical programs through key inflection points and also introduced a highly unique multi-agentic system aimed at contouring cancers. As we move into 2026, we are positioning to advance our clinical program expand our RADR platform's commercial reach and revenue potential globally through our new AI center of excellence in India and further strengthen our balance sheet.

Our AI-driven clinical pipeline now encompasses multiple drug candidates across solid tumors, blood cancers and pediatric oncology with a combined estimated annual market potential exceeding $15 billion and approaching $20 billion. On average, our newly developed drug programs have been advanced from initial AI insights or concepts to first in human clinical trials in 2.5 to 3 years and at approximately a few million dollars per program. It is very important to note that we have dosed over 100 patients of prostate programs and seen clear linkage to mechanisms and patient value that we believe can yield future medicinal opportunities in a range of cancers that we're continuing to advance.

Before moving on, I want to take a moment to directly address some malicious and fake news that has been circulated online falsely claiming that I'm departing Lantern Pharma or have stepped down as CEO. This is categorically untrue and appears to be rooted in a deliberate and perhaps malicious attempt to manipulate our stock price. This disinformation has caused real harm to our company to the mission we are pursuing on behalf of cancer patients and to our investors. And we intend to pursue all appropriate civil, criminal legal recourse against those responsible. Let me share with you now more importantly, the more notable achievements over the last past year and quarter and where we are heading into 2026.

Let me start with our LP-300 program, the HARMONIC trial, which addresses a significant and growing unmet need in lung cancer. HARMONIC focus exclusively on never smokers and non-small cell lung cancer who have progressed after treatment on TKIs. In Asia, never smokers represent now close to 40% of all non-small cell lung cancer cases compared to about 15% to 17% in the U.S. and Europe. The market opportunity here is substantial. Over $4 billion, we believe, annually in spend on people who are not smokers or never smokers and get non-small cell lung cancer. There are currently no therapies approved specifically for this patient population.

[Technical Difficulty] HARMONIC trial continued to advance through the fourth quarter and into early 2026, with ongoing patient enrollment and follow-up -- Japan and Taiwan. Last year, we completed the target enrollment in Japan ahead of schedule across 5 clinical sites including the National Cancer Center of Tokyo. During Q4, clinical investigators presented data at the 66th Annual Meeting of the Japan Lung Cancer Society from both Asian and U.S. cohorts. The trial has previously demonstrated an 86% clinical benefit rate and a 43% objective response rate in its initial safety leading cohort, including one patient with a durable complete response and survival continuing for nearly 2 years.

Let's talk a little bit about our upcoming Type C meeting. We're getting more involved with the FDA. And in March, we submitted a Type C meeting package to the FDA for LP-300, the meeting scheduled now for mid-May 2026. We are seeking FDA feedback on 3 proposed protocol amendments that came out as a direct result of our observations from the trial -- eGFR exon 21 L858R mutation where our preliminary analysis suggests greater clinical benefit from the LP-300 regimen in combination with the chemo doublet for these L858R mutation patients. Second, increasing maximum LP-300 treatment cycles from 6 to 8 based on established safety data and the mechanism -- 2 stage design reflecting the evolving treatment landscape that has made continued randomization to the control arm, increasingly challenging due to changed control protocols. We are actively exploring collaboration and partnering opportunities globally to maximize LP-300's commercial potential.

We're in discussions with several regional and global pharma companies around the future of this exciting treatment and we expect additional clinical updates in the coming weeks, along with insights on the exon 21 L858R population of patients. Turning out to what I believe remains one of our most significant assets. In Q4 of 2025, we reported additional positive LP-184 Phase I results, showing durable disease control in heavily pretreated advanced cancer patients. The trial enrolled 63 patients achieved all primary endpoints of the 48% clinical fit rate at or above the therapeutic threshold. So unique and promising signal of activity in this patient population. The data validated our synthetic lethal hypothesis. We saw marked tumor reductions that are observed in [indiscernible] some DNA damage repair mutations, including CHEK2, ATM, BRCA 1, STK11. And these are all alterations that were initially flagged or signaled through radar-driven insights.

We also established a recommended Phase II dose of 0.39 milligrams per kilogram with a favorable safety profile and saw notable clinical benefits in some very difficult-to-treat cancers, including relapsed GBM, gastrointestinal, stromal tumors and thymic carcinoma. Many of these were these patients are now getting clinical benefit for over a year into their treatment cycles. These are typically tumors with sub 6-month PFS and very poor OS as well. The Phase Ib Phase IIa development plan are building on these results, and we're positioning these into multiple precision oncology trials.

Let me walk you through those. First, triple breast cancer, where over $4 billion are spent. We have an FDA reviewed protocol for a combination study with olaparib and we hold fast track designation. Second, non-small cell lung cancer with patients that have KEAP1/STK11 mutations, we believe about a $1.5 billion opportunity in patients typically fail immunotherapy and are not good responders for chemotherapy. Third, an investigator-led bladder cancer study planned in Denmark targeting PTGR1 overexpressing tumors with DNA damage repair mutations. All three are precision oncology trials, where they're being driven by mechanistic insights, biomarkers and very focused patient populations that we believe have been value from the outcomes in our Phase I and also in our extensive preclinical work.

These trials are subject, of course, to additional funding, which we're actively pursuing and whether it be through grants or other mechanisms. What distinguishes our synthetic lethal approach is [indiscernible] precision. Unlike conventional chemotherapies that indiscriminately target dividing cells, both our first in-human drugs, LP-184 and 284 exploit specific genomic vulnerabilities in cancer cells, particularly those with deficiencies in DNA damage repair. The pharmacokinetic data from these trials suggest we're approaching concentration levels that correlate with the nanomolar potency that we've already observed in clinical models, a critical inflection point that we believe has shown a proof of mechanism in patients and it may pave the way for future trials and more importantly, pharma partnerships.

During our collaboration last year with MD Anderson, it was also revealed that LP-184 had a very unique and remarkability to transform immunologically cold tumors, especially in TNBC into hot tumors, a breakthrough with profound in bookings through expanding immunotherapy benefits to previously unresponsive patients. This isn't merely additive efficacy. It represents a mechanistic synergy that addresses one of immunotherapies most significant limitations and it opens up additional co-development opportunities and new indication expansion where PD-1 and PD-L1 checkpoint inhibitors have stopped working.

Let me move on to Starlight Therapeutics. At Starlight Therapeutics, we cleared for a planned Phase I pediatric CNS cancer trial. We announced this last week on Friday. This is an innovative trial design that we unveiled at the Society for Neuro-Oncology and it features a unique combination of spironolactone and exemplifies the power of computational biology. We're approaching -- I'm sorry we're exploiting the synthetic lethality of our drug NGBM through a mechanistically elegant interaction. Spironolactone degrades ERCC, a critical DNA repair protein that causes further vulnerability that the STAR-001 exploits with precision in these brain tumors. The IND being cleared for this trial is a milestone that I'm particularly proud of, and I want to spend some more time on it because Starlight Therapeutics, our CNS oncology franchise is now well positioned for that.

In early 2026, the FDA cleared the IND for Starlight Therapeutics in not only recurring CNS tumors but in ATRT and rare pediatric tumors. With the clearance, we now have INDs cleared for both our adult and our pediatric programs, positioning pursue clinical development across the full patient spectrum. This is a pivotal regulatory milestone for our wholly-owned success-focused subsidiary. Starlight -- has received both rare pediatic disease designation and orphan drug designation from the FDA for ATRT, along with additional rare pediatric disease designations for hepatoblastoma, rhabdomyosarcoma and malignant rhabdoid tumors. These dedications provide pathways for FDA priority review vouchers upon a mutual approval.

PRVs have been sold or transferred for a significant value historically with recent transactions in the range of $150 million to $200 million and our drug has 4 of these. Currently, each of these rare pediatric disease designations independently qualifies upon potential FDA approval and meeting other program conditions for these PRVs. That's multiple shots on goals from a single molecule representing a potentially meaningful source of nondiluted value for Lantern and its shareholders, independent of the commercial potential of the underlying therapy. Now the scientific rationale for combination with spironolactone is compelling, it's unique and novel.

Preclinical studies demonstrated a 3- to 6-fold increase in GBM cell sensitivity when combining with these agents with most preclinical models showing complete tumor eradication and minimal recurrence. This can be especially critical in the most sensitive patients such as children, the elderly or those that have gone undergone multiple prior lines of therapy. Even more interesting is that [indiscernible] has shown antitumor activity in GBM regardless of the MGMT status. So let's talk a little bit about why this mechanism is distinctive in first in class. This is where our RADR AI platform and novel mechanistic biology really comes to life. The planned trial includes a dedicated combination cohort, evaluating STAR-001 with spironolactone.

And again, this was initially identified with our platform. and we believe that this combination creates a unique synthetic lethality in these challenging brain tumors. Now once our drug is activated when over -- when PTGR1 is overexpressed. It induces DNA double-stranded breaks that are lethal to the cancer cell, if left unrepaired, as a critical insight. The cancer cell has a repair escape route, and we found a way to basically shut it down. We identified that we can degrade ERCC3 excision repair complementation group -- that's a key heli case in the repair pathway. It's a central repair mechanism that's used in some of these very aggressive tumors.

Now we can shut it down by delivering spinal lactone. And basically, this is how you block the cancer cells from trying to come back. And that's where spinal lactone enters the picture. It's brain penetrated. It can be orally administered as a long safety record in adults and also now in pediatric and it degrades the ERCC3 protein through targeted proteosomal degradation. In our preclinical models, we saw ERCC protein levels reduced by at least 50%. And when we actually through some dosing optimization, we got even more reduction and by reducing that ERCC3, we grew the ability for the repair to happen. So this is a rationally designed identified validated combination that's been validated in the clinic that creates enhanced synthetic lethality that amplifies the tumor killing activity or STAR-001.

So I want to underscore some things to make this combination strategy unique. The ERC was identified and validated through our analysis, not through just additional screening. The combination partner, spironolactone is already well characterized and derisked the safety profile of this combination. The mechanism, precision bioactivation with targeted repair path inhibition, we believe, represents a first-in-class and unique approach to how to treat these devastating brain cancers. And now that the IND is cleared, Starlight Therapeutics is positioned to move rapidly into the clinic, of course, subject to more funding. Starlight, which is 100% of my Lantern will have the potential to be another very positive impact on our investors as we monetize this unique asset, monetize the patents and potentially monetize the PRVs.

Now this computational capability doesn't really enhance our existing programs. It opens up entirely new therapeutic possibilities as well. We'll talk about that. a little later. In Q1 of 2026, we also received FDA orphan drug designation for soft tissue sarcoma. Adding to the existing designations in mantle cell and high-grade B-cell lymphoma. We also had a patient in Q4 that represent -- that we presented clinical data at the 25th Leukemia Lymphoma and Myeloma Congress in New York, and we confirmed a complete metabolic response in a heavily pretreated diffuse large B-cell patient who has remained cancer-free since we initially reported this result.

LP-284 benefits from composition of matter patents through 2039 across major global markets and also, of course, the orphan drug designation, and we continue to explore LP-284 beyond lymphoma, including as a potential therapeutic for autoimmune disorders such as lupus and SLA, where our preclinical data have showed significant potency in reducing clonal B cells actually CD19 and CD20 B cells. And this work could be -- could dramatically expand the commercial opportunity for this asset. We're beginning active dialogue to look and seek partners for this unique drug on the back of the compelling Phase I data that's being put together and the responses that we're beginning to see.

Now let me shift to what I believe has become an increasingly important value driver for us and one that are commercial, RADR AI platform is commercial opportunities independent of our drug programs. RADR integrates 2 -- 300 billion plus oncology folk data points, hundreds of advanced machine learning algorithms and prediction success validated in the natural clinical trials, not only for ourselves but also for our partners. In early '26, we needed an AI Center of Excellence in India to help us grow [indiscernible] focus more on the RADR platform and with Zeta.ai, giving us the ability to develop capabilities and features around the clock.

We're beginning to recruit world-class ML talent and also give us additional scalability to support additional biopharma partnerships and feature development. We also continue to lead with BBB. BBB which holds 5 of the top 11 positions in the therapeutic data commons also has been enhanced significantly over the last month or 2. And we also are beginning now to commercialize our LBX AI, which is our liquid biopsy AI. We've highlighted in our roles, the amount of money we've put into all our programs. Last year, we spent about $1 million across our AI technologies and platforms. And we're also, at the same time, able to develop what we believe is another key aspect for the future of AI-driven drug development. We're at an inflection point with zeta.ai because it's not only an inflection point because the system has now been launched to multiple demo partners, but it's really how science itself will be conducted. Agentic AI systems that reason, collaborate and act autonomously -- these are poised to become the standard infrastructure for drug discovery and scientific R&D. This is not a question of if but when. Lantern through [indiscernible] data intends to be the standard bearer for this kind of shift, especially in rare cancers.

Now think about how most people use AI and drug development today. They ask a single model of question. They get an answer. They typically do it in concert with a series of engineers and computational biologists. And it's really almost really a one-off event. And they may ask it in parallel, they ask, may ask it several times, they may develop tools to look at the same question. But with Zeta, you're doing it in natural language and you're getting the facility of doing it as an orchestra, a multigenic architecture, where specialized tools trained on [indiscernible] , pathway analysis, clinical trial design biomarker identification, molecular feature assessment, novel chemistry generation, collaborate, challenge assumptions and cross-validated findings all in real time before delivering hardened insights.

Many of you have been able to see this in person and actually see how we've been able to go from ideas to inscience to actually potentially powerful new medicinal concepts in under an hour. Now the true power is not in any single agent, but the intelligent orchestration, a true AI coscientist, and we've built that for helping to conquer rare cancers. This approach fundamentally inverts the traditional drug development paradigm. Before a single experiment is run, where they can rigorously stress test hypotheses through computational analysis and recursive reasoning, interrogating literature of modeling pathways, analyzing historical trial data, feeding on your own private unique insights and data evaluating biomarker strategies, stress testing medicinal concepts, looking at molecules against known patient populations and then only advancing the most heartened of the ideas -- failed experiments by 80% or 90%, we can allocate precious R&D resources and time to the most promising opportunities and do it faster.

We can test dozens of hypotheses in parallel while Lantern would still be designing the first experiment. This platform also from being fundamentally new, a persistent interactive organizational memory. Every interaction, every insight, every hypothesis tested is stored and instantly clearable. And also, you generate knowledge graphs. It's like having your own entire scientific advisory experts, your full research team and comprehensive access to questions and answers available 24 hours a day, 7 days a week, for any question in your domain. Scientific R&D already arriving now. Since late 2025 [indiscernible] has been active demo beta testing with over 25 biotech companies, cancer research centers, biopharma consultants and even some CROs and investment banks, where we're generating significant early engagement that validates both demand and differentiation. We've designed with zeta that with the multi-tier commercial architecture that serves the entire drug development ecosystem.

And at the foundation, we'll have accessible academic tier that brings early career researchers and university teams into the platform. and also individual subscriptions, institutional licenses and they'll continue to help validate platform and create the network effect, which makes data increasingly valuable. We'll have a professional tier that serve emerging biotech and midsized developers through usage-based licensing, and then we'll also have an enterprise level, large pharma, where they can deploy in their own private clouds and also add to their proprietary knowledge graphs and deepen their internal data integration and perhaps even deploy customized ontologies and use it for unique configurations.

So this will be a multitiered commercial architecture and we believe it could also be used over time in multiple other disease areas beyond cancer. The beauty of this model is the natural progression researchers discover with Zeta in an academic setting and they'll carry that experience as they continue deployment. At every level, the platform gets smarter as more users and data flow through the system. And we believe that this global rare disease and rare cancerous therapeutic market is projected to exceed about $300 billion by 2028. And the broader AI-enabled drug discovery market represents we believe an additional $20 billion to $50 billion long-term opportunity for with Zeta and our multi-agentic AI architecture.

Our longer-term plan is to scale with Zeta beyond rare cancers and into other complex therapeutic [indiscernible], each presenting the same fundamental problems of fragmented knowledge slow experimental cycles, expensive failures. And we believe this represents a potential near-term market opportunity of $20 billion to $50 billion and that with Zeta.ai is our first Agentic commercial product designed to capture a meaningful share of that. When you connect the dots, clinically validated radar, commercially ready AI modules and a multitiered revenue model. You see a business model that extends well beyond our pipeline. We believe our AI tools and services represent several hundred million dollars in stand-alone market potential, and that's a powerful complement to our drug development strategy.

So now I'll turn the call over to David Margrave to discuss our financials and our other key metrics. David?

Thank you, Panna, and good afternoon, everyone. I'll now share some financial highlights from our fourth quarter and full year ended December 31, 2025. I'll start with a review of the fourth quarter. Our general and administrative expenses were approximately $1.5 million for the fourth quarter of 2025 compared to approximately $1.6 million in the prior year period. R&D expenses were approximately $2.7 million for the fourth quarter of 2025. In the fourth quarter reported a net loss of approximately $4.1 million for the fourth quarter of 2025 or $0.36 per share compared to a net loss of approximately $5.9 million or $0.54 per share for the fourth quarter of 2024.

For the full year 2025, our R&D expenses were approximately $11.5 million down from approximately $16.1 million in 2024. This decrease was primarily attributable to an approximate $4 million reduction in research studies and materials relating to the conduct and support of our clinical trials in part due to a $0.6 million decrease in payroll and compensation expenses and an $81,000 decrease in consulting expenses. Our general and administrative expenses for the full year 2025 were approximately $6.5 million up slightly from approximately $6.1 million for 2024. The increase was primarily attributable to increases in business development and investor relations expenditures, of approximately $436,000 increases in patent costs of approximately $55,000 and an increase in corporate insurance of approximately $51,000. Our R&D expense continued to exceed our G&A expenses by a strong margin, reflecting our focus on advancing our product candidates and pipeline -- [Technical Difficulty] or $1.57 per share compared to approximately $20.8 million -- $0.93 per share for 2024. Our loss from operations in the 2025 calendar year was partially offset by interest income and other income net totaling approximately $0.9 million.

Our cash position, which includes cash equivalents and marketable securities, was approximately $10.1 million as of December 31, 2025. Based on our currently anticipated expenditures and capital commitments, believe that our existing cash, cash equivalents and marketable securities as of the date of this call will enable us to fund our anticipated operating expenses and capital expenditure requirements until at least approximately late July 2026 to mid-September 2026. We will need to raise substantial additional funding in the near future, and we are actively evaluating and pursuing potential funding alternatives. As of December 31, 2025, we had 11,254,697 shares of common stock outstanding. No outstanding warrants to purchase shares and outstanding options to purchase 1,296,126 shares. These options combined with our outstanding shares of common stock give us a total fully diluted shares outstanding of approximately 12.6 million shares as of December 31, 2025.

I'll now turn the call back over to Panna for an update on some of our development programs. Panna?

So our leadership in the innovative use of AI machine learning to transform costs and time lines in the development of precision oncology therapies has allowed us to bring 3 molecules into clinical trials with teams, cost and efficiency that are almost unheard of in oncology biotech. And even that, we're actually seeing massive year-over-year improvements in our spend and the output that we're seeing. During 2025, we achieved our goal of integrating generative AI to transform our platform into a system of autonomous, agentic co scientists and put together a model that we believe can be future for how scientists create value.

So looking ahead, how do we expect to see value creation catalysts. We have a Type C meeting coming up with the FDA on focusing enrollment in the HARMONIC trial on EGFR exon 21 L858R patients. These are patients that do very poorly, and we've seen some meaningful improvement as a result of being dosed with our drug in combination with the chemo doublet. We've also seen the same in extending LP-300 treatment cycles and we believe that the current environment because of the changes in standard of care really require converting the current design to a single-arm Simon 2-stage design. We'll have some data around the 858 population in the near future. Our planned investigator-sponsored trial with LP-300 in combination with osimertinib in chemo in frontline with [indiscernible] driver mutations is also advancing.

We also have planned initiation of an LP-184 phase, an LP-184 trial in bladder cancer in Denmark, which is paid for by the Danish government and the Danish Cancer Research group. We expect to start that for PTGR1 overexpressing bladder cancers, the DNA damage repair mutations. We also have planned initiation, again, subject to funding of our LP-184 Phase Ib/II in TNBC and in CNS cancers as well. Additionally, we'll have a major launch of our Wood Zeta platform at AACR coming up next month. and we'll be converting a lot of beta engagements to commercial partnerships and actually also launch the full multi-tiered subscription frame. We'll also be pursuing additional funding, including potential grant revenue to fund plant operations and clinical advancement of our precision oncology trials.

We're not just building better tools for ourselves. We're fundamentally reimagining what's possible in precision oncology and building tools that the entire community can actually use. And as we continue this journey, our agentic RADR platform positions us at the forefront of an entirely new paradigm in drug development, one where AI doesn't really assist human resource but actively drives drug discovery forward through autonomous continuous learning and insights that can be tested in labs and deployed into the clinic for patients.

So the golden age in medicine isn't just beginning. It's so accelerating exponentially. We've seen a lot of activity in the past 4 to 6 months. The intelligent always on Symphony is actually here. cancer patients, especially reg cancer patients can't wait another 50 years for the typical 50 years of progress we've seen and we believe that this next 50 years of progress can happen in the next 5. And this is something we believe very strongly that AI is going to accelerate the development and the use of knowledge in a way that we haven't seen in medicine.

As we advance into 2026, we're laser-focused on executing our dual-engine strategy, enhancing our clinical assets through key inflection points and then out-licensing or partnering them while simultaneously scaling our AI platform for commercial deployment, each clinical milestone validates our AI platform's predictive power, while every platform enhancement accelerates our pipeline and creates new partnership opportunities. Also now with zeta.ai, we believe we're setting the standard for how multi-agentic tools and AI systems can be used in drug development and we're bringing that into commercial setting, and we see multiple paths to create value using that platform. We're not just building better tools. We believe we're fundamentally reimagining what's possible in the time line and capabilities of precision oncology, and we're building it to be the standard that hopefully, the rest of the industry also follows.

I want to thank our exceptional team, our partners and our shareholders for your continued support, and also our team internally for helping put today's call together. So thank you, and I hope we can continue improving outcomes for cancer patients, while all transforming the economics of drug development.

If you'd like to ask any questions, you can do so in 1 or 2 ways. You can type a question using the QA tool, and we'll get back to you shortly or you can send us an e-mail to investor at Lantern Pharma, and we'll get back to any questions that you might have.

Thank you, everyone, for your time this afternoon.

Thank you very much.

Okay. Thank you, everyone, for joining us today for our LP-184 clinical trial results webinar. We'll be talking in depth today with some of our Lantern's team and also with our collaborator and partner in developing the trial and some of the early science behind LP-184. Dr. Igor Astsaturov from Fox Chase is also on with us as well. I'm going to be talking about an overall, the drug just to give everyone an overview of the molecule. And then we'll be talking about the scientific background and some of the early development to give people an appreciation for the molecule. And then we'll be digging right into the clinical results from the trial and some of the observations that we've made.

We've got a lot of information to present and walk through today. So I appreciate everyone's patience, and we'll take some time for Q&A toward the end. If you have questions, definitely send them into the chat window, and we'll be selecting questions to go through. With that, I'm going to introduce our panelists today. We've got our Chief Scientific Officer, Dr. Kishor Bhatia. We have our Head of Clinical Development, Dr. Reggie Ewesuedo. And as I mentioned, also, we have Dr. Igor Astsaturov from Fox Chase who's been our collaborator in the development of the molecule.

So with that, let's talk a little bit about 184. 184 is being aimed primarily for the treatment of advanced solid tumors, largely focused on tumors that are deficient in DNA damage repair potential. These include tumors such as TNBC, tumors such as GBM and perhaps even bladder cancer. In fact, in bladder, pancreatic, TNBC, a very large percentage of those tumors have DNA damage repair pathway mutations or deficiencies, which make them exceptionally vulnerable to this drug. Our Phase Ia has been completed. We had very good safety and tolerability data coming out of that and the profile is acceptable, but also some very encouraging clinical activity across a number of solid tumor types and most importantly, validation that tumors that have DNA damage repair efficiency tend to have exceptional sensitivity to the drug and is a great tee-up for the future Phase Ib, Phase II trials that are being planned.

We have 3 orphan designations for the drug. We have 2 Fast Track designations, both of which we'll be pursuing in trials coming into next year. The market potential for this drug is quite exceptional, blockbuster-like potential because it seems to work in a wide range of cancers, including, as we mentioned, the ones where we have Fast Track. About 1 in 4 to 1 in 5 cancers harbor some type of DNA damage repair deficiency. And again, those make it particularly sensitive. What we also saw, as Reggie will point out, in the trial is that a very high percentage of patients that had advanced recurrent solid tumors had high levels of PTGR1. And as you've seen and heard us talk about that before, PTGR1 is really a smoking gun for potentially transforming the prodrug into its active cancer killing molecule. There's about 170,000-plus cases and nearly 400,000 global. So it's a very large market.

And the way it works is that it alkylates or breaks apart the DNA through double-stranded breaks at a very specific location at the 3 prime position of adenine. And again, we validated this in early work and now we're beginning to see it now in the clinical work. This molecule is purely synthetic. Kishor will talk a little bit about what inspired the molecule from a natural substance, but the molecule that we see in the trials is fully synthetic route in manufacturing. And it can be combined with checkpoint inhibitors, PARP inhibitors, potentially chemotherapy and radiation therapy and also a totally novel molecule, spironolactone.

So it makes a wonderful combination, which also we believe increases the potential, but also gives us some flexibility in how we design and think about getting to the right kind of efficacy that's going to get us above and beyond the current standards of care. Over 10 patents have been issued or pending for this drug, and we have claims that extend through the early 40s. So with that, I'm going to turn it over to Kishor and Igor to talk a little bit about the backdrop of the molecule and about the early development and insights on LP-184. Let's go to the next slide. I'm going to turn over Kishor to you and Igor.

Thank you, Panna. Thank you, everyone, for taking the time to participate in our webinar. As Panna told you, our molecule, LP-184, was actually inspired from a naturally occurring molecule called iludin that occurs in the Jack-O'-Lantern mushrooms. The derivation of LP-184 from iludin is an interesting story, but I think the key message that I'd like to share with you is that, the design of the molecule now bestows upon these iludins and bestows upon LP-184 property of synthetic lethality. And I will go through some of the evidence on how we came about it.

Panna has already talked about the application of LP-184 into various solid tumors, so I won't spend more time on it. But let's actually go to understand the mechanism of action of LP-184. And our story actually began by looking at correlations between LP-184 sensitivity in a large panel of cell lines. And the question we asked is, can we figure out what cell lines are sensitive? What are resistant? And what are the reasons why certain cells are sensitive to LP-184? Two things immediately stood apart. One was that the response of a tumor cell line to LP-184 is predicated by the level of expression of a gene called PTGR1. And the graph on the left shows you the very strong correlation between PTGR1 expression and LP-184.

The table on the right shows you various genes whose expression made LP-184 exquisitely sensitive of the pathways in which they were expressed correlated with exclusive sensitivity to LP-184. What that -- the mixture of these 2 basically told us that the story is likely that tumors that have high expression of LP-184 but have deficiencies in repair related to one of these pathways where these genes are involved are likely the tumors that are most indicative of response to LP-184. Taking this further from in silico to actually asking questions on the bench, validation of some of these studies, we collaborated with Fox Chase and Igor's team conducted some excellent studies. One of these studies was to knock out using CRISPR, PTGR1 in cell lines.

And as you can see in the graph in the middle, if you knock out PTGR1 in a tumor cell line and you can look at the green line compared to, for example, the red line, knocking out PTGL1 as in the green line totally obliterates cytotoxicity of LP-184, further proving our in silico-based hypothesis that PTGL1 is very essential to the activity of LP-184. Similar results were obtained when we moved from cell line data to xenograft data, and that graph is shown on the right-hand side.

The basic takeaway message here is that indeed, even in vivo and in vitro, our in silico prediction confirms that PTGR1 is essential for LP-184 activity. But what that means in practical terms is that because PTGR1 is expressed -- overexpressed in tumor cell lines, but not in normal cell lines, this provides LP-184 with a window of selectivity, which is very critical for the therapeutic index of any antitumor molecule. Now moving further along, what this tells us is that it might be important eventually to have an assay where we can measure the level of PTGR1 in tumors. And this is precisely what we have done. We developed RNA RT -- RQ-PCR assay to look at the expression of PTGR1 in various tumor settings, allowing us the possibility of providing tumor stratification and moving LP-184 into a precision medicine approach.

I will now move to the next slide, which gives the other facet that makes LP-184 particularly exciting. And one has to think of combining these 2 properties that are required for tumors to be sensitive. One we talked about is PTGR1, and I showed you some data. Here, I'm showing you data of cell lines that have mutations in specific genes, which are in a pathway called nucleotide excision repair. So these are cell lines that have -- and you can see this on the graph on the left-hand side, tumors that have mutations in ERCC1, ERCC2 or ERCC6 and therefore, are unable to carry out the nucleotide excision repair when exposed to LP-184 are highly sensitive to LP-184.

On the right-hand side, the graph tells you about another DNA damage repair pathway, which is the homologous recombination pathway. And here, if you compare the light blue and the dark blue graphs, the dark blue graph is from a tumor that has loss of function for BRCA2, a gene involved in homologous recombination pathway. And we'll speak about homologous recombination pathways further. But basically, again, what this message tells us is what we hypothesized using in silico data is indeed true on the bench. That is tumors that have mutations in DNA damage repair pathways, be it homologous recombination deficient or nucleotide excision repair are highly sensitive to LP-184.

There are a couple of other points that the in silico data provided. And what they basically said is that in addition to nucleotide excision repair and homologous recombination repair, there are additional genes such as those that are involved in replication stress pathways that are highly correlated with sensitivity to LP-184. This is critical because what this tells you is that we are not just looking at tumors that are deficient in nucleotide excision repair pathway, for example, bladder cancer or homologous recombination pathways, for example, triple-negative breast cancers, but also tumors that have damaged pathways in the replication stress.

Why this is important, and I'll show you that in some slides later, is that it means that LP-184 might be able to overcome resistance to PARP inhibitors, but it also means that LP-184 might be a very good agent to turn cold tumors to hot tumors for immunotherapy. And although I won't have time to present some of the data, I thought it was worthwhile mentioning some of these properties of LP-184. Here's another piece of data that was very useful in us trying to develop what indications we would focus upon as we move clinically. And what this data is doing is it's looking at precise quantitation of double-strand breaks after tumor cells are exposed to LP-184.

In addition, what it provides is some of the reasons why homologous recombination deficient tumors might be more sensitive. And when we quantitate the amount of double-strand breaks in tumors that have deficiency in homologous recombination, for example, by nonfunctional BRCA2, the total amount of double-strand breaks in those tumors is twofold compared to tumors where BRCA2 is functional. Now if you put certain things together, PTGR1 overexpression, deficiency of DNA damage repair, these 2 starts will mostly align where there's a tumor environment.

Normal cells express low PTGR1, are rarely deficient in DNA repair pathways and LP-184 will basically be ineffective in killing normal cells. But tumor cells that have these defects are going to be highly sensitive. So in that sense, LP-184 becomes tumor selective. And the story that gets built up is depicted in this schema, starting with LP-184 in the presence of PTGR1, the cyclopropyl ring of LP-184 gets opened, allowing the cyclopropyl ring to become reactive to a nucleophilic attack on DNA. And therefore, the amount of DNA damage caused by LP-184 is proportional to the levels of PTGR1.

In addition, once damaged, this damaged DNA cannot be repaired in tumors that are deficient in DNA damage pathway and therefore, they die, which is very different than what would happen to normal cells that are exposed to LP-184. We further validated some of the concepts of the nucleotide excision repair pathways. Remember earlier, I showed you a slide with cells that are mutant in specific nucleotide excision repair pathways. We extended these studies further, again, in collaboration with Igor, looking at ERCC4 knockouts. And the panel on the top shows you both the knockout of ERCC4 as well as the data that knocking out ERCC4 reduces the IC50 of LP-184 by twofold. And this can also be replicated in an in vivo model of these tumors.

But beyond doing genetic engineering to create a depletion of NER, there's a direct application of this principle to define a combinatorial agent for LP-184. And this combinatorial agent is spironolactone. A couple of years ago, there was some very interesting data, which showed that spironolactone can degrade a protein called ERCC3, which is a critical protein for nucleotide excision repair. And the panel in the bottom shows that this pharmacological inhibition of ERCC3 by spironolactone in presence of LP-184 is able to totally regress glioblastoma tumors in mouse models.

Now having some better handle on how NER deficiency affects LP-184, we moved some of our attention to understanding what happens in real-life tumors that have homologous recombination deficiency. In this table, what we are showing you are 3 different tumor classes. One is non-small cell lung cancer, the middle one is pancreatic cancer and the bottom is prostate cancer. And these are tumors derived from patients with -- where the tumors have specific mutated HR genes shown in the last column. The synthetic lethality of LP-184 in a variety of these tumors is clearly evident. In addition, what you can appreciate is that LP-184 inhibits 77% to 90% of cell growth in these tumors compared to 29% to 80% in Olaparib.

So LP-184 is able to perform the same therapeutic levels as a PARP inhibitor in HR tumors. But I will show you additional data that will tell you that LP-184 is actually superior to the PARP inhibitors. Now clearly, when you speak of homologous recombination deficiency, the first tumors that come to mind are triple-negative breast cancer. Breast cancer is a paradigm of BRCA1, BRCA2 and other homologous recombination deficiency. Since the data clearly pointed that the synthetic lethality of LP-184 is unmasked by DNA repair deficiency, we now looked at triple-negative breast cancers.

Again, these triple-negative breast cancer tumors that I'm showing you the data for are real-life clinical tumors obtained from patients that have -- that are either PARP inhibitor sensitive or PARP inhibitor resistant. Irrespective of what the clinical picture of these tumors were in terms of treatment with PARP across all the 10 triple-negative breast cancer PDX mouse models, we saw that LP-184 totally regressed all these tumors. This was very, very exciting. It was very exciting that our hypothesis starting from in silico going further to in vitro and eventually to xenografts and now to PDX models appears to stay validated. But there's something else that's important here. And while there isn't sufficient time to discuss the details of the mechanism of why PARP inhibitor resistant tumor would be sensitive to LP-184, there are a couple of important aspects that arise from this.

One is that LP-184 could be an exemplary partner with PARP inhibitor, not only to overcome resistant but also to diminish resistance when used in combination. In this graph, what I'm showing you is data both from a PARP sensitive -- PARP inhibitor sensitive and a PARP inhibitor resistant model. And if you focus on the graph on the left-hand side, going from left to right, you have the tumor growth with control alone, which is vehicle. And you can see that LP-184 has totally wiped out this tumor at 4 milligrams per kg and almost wiped this tumor out at a lower dose of 2 milligram per kg, clearly showing a dose response relationship.

You can also see the dose response relationship with Olaparib in this, but the most exciting part is the last 4 graph parts on the left-hand side which show you that a combination of LP-184 at very low levels of 0.75 milligrams per kg and Olaparib at subtherapeutic levels totally wipe out the tumor, be it PARP sensitive, such as on the left-hand side or PARP resistant such as on the right-hand side. Again, further validation of the fact that LP-184 induces complete tumor regression in HR-deficient triple-negative breast cancer. And this applies not just to tumors that are resistant to PARP inhibitors, but also tumors that are resistant to doxorubicin and cyclophosphamide.

So this is where I was mentioning that LP-184 is similar to PARP, but more superior than PARP. There are several reasons for it. PARP inhibitors are not indicative in tumors that are nucleotide excision repair, but LP-184 is. So the spectrum of DNA repair deficient tumors that LP-184 can affect is far more than PARP inhibitors, and it's a great partner for combating resistance to PARP inhibitors and resistance to other standard of care agents. Here are additional data now moving from breast cancer to pancreatic cancer and the sensitivity of HR-deficient tumors in -- DNA damage repair pancreatic models was studied in Dr. Igor's lab. And as you can see that LP-184 is highly sensitive that tumors with mutations such as in ATR or BRCA1 are highly sensitive to LP-184.

When you compare the sensitivity and the potency of LP-184 to other standard of care agents in real-life PDX models, pancreatic PDX models, you find that LP-184 demonstrates 20x to 400x higher potency compared to, for example, gemcitabine. So obviously, all these preclinical data provide us with significant excitement of how to position LP-184 in multiple areas in solid tumors. I will now ask Dr. Reggie to discuss some of the early clinical data. Reggie?

Yes. Thank you, Dr. Kishor. If I might move to the next slide, I think my presentation starts with the clinical review of the program. I thought for this webinar, I should, at a very high level, summarize the message. The first is to announce that we've completed the first-in-human Phase Ia study. This is, as stated, our very first time of getting into the clinic with LP-184. Obviously, it's not the first time of getting in with a drug that was in the same class, but I will speak to that later on.

The second is to let you know our planned Phase Ib/II clinical trials. Panna spoke about some of those. And Essentially, the 3 areas we're trying to focus on initially in our own with LP-184 ADAs monotherapy or combination and triple-negative breast cancer based on the preclinical set of data that we've generated, I think, is on top of that list. There's non-small cell lung cancer. This will be combined with standard of care and dual IO combination in a subset of patients, the KEAP1/STK11, PD-L1 low tumor patients. These patients have a very, very poor prognosis. They don't do well with the current standard of care regimen.

And then we have an investigator-sponsored trial that we're doing with Professor Helle in Denmark. This is in bladder cancer patients that have PTGR1 positive tumors as well as NER deficient tumors. And then I will speak to additional clinical development opportunities that we're also exploring and the scientific rationale for that. Could you go on to the next slide, please. Now there are 2 things that informed of going into the clinic, the Phase I trial. The first one, obviously, is to ask the question based on the totality of the preclinical data that we have, the toxicity, the PK and what have you, what would be an optimal regimen that we could get into the clinic as a first-in-human regimen.

And we settled on day 1, day 8 every 21 days. As is typical, you get into the clinic, you look for signs of whether your regimen is optimal, what safety and other correlative studies that might help you improve on that regimen or what are you satisfied with the current regimen. So this is the initial regimen, and I will speak to some of the thoughts based on the data that we've got so far on additional opportunities that we're looking at.

The other part is to ask the question around time lines. We all know that Phase I studies sometimes take a very long time, sometimes based on those traditional approach and designs, we decided to take an adaptive approach to BOIN regimen. I think the BOIN strategy design. This allows us really to move fast so that we can answer questions without having any break between dose levels. So we're able to do that by using the BOIN design, and that is reflected on the slide. So could you go on to the next slide, please?

So the objectives are pretty standard for Phase I, at least the primary objective. I will emphasize again, is first-in-human. So looking at the safety, but beyond safety, now based on the preclinical experience, this is a drug that we believe very strongly is not something that is a typical chemotherapy cytotoxic where you reduce the length of the regimen to maybe 6 cycles and what have you. We believe that this was a drug that we could take for a very long time in the clinic, patients could stay on it as far as they are benefiting. So the tolerability was also key to us to understand that before we declare the recommended Phase II dose. Certainly, the maximum tolerated dose is a given.

And then secondarily, to inform that primary objective, we wanted to look at the PK of the drug muscle. Again, it's a prodrug, nevertheless. We also wanted to look at preliminary activity with a focus on a set of patients for which we had predicted might derive benefit from the drug based on their DNA damage in the repair alterations. And of course, we wanted to understand just going in, do we need a biomarker strategy early on for a subset of patients based on PTGR1 or is our initial hypothesis correct, which is to say patients that are coming into this Phase I study, they obviously have advanced disease. I mean most of them have aggressive disease as tends to be the case with this population of patients.

Would we observe very, very elevated, a good number of the patients having a very high prevalence of PTGR1 overexpression in these tumor samples. So those were really the objectives as we got into the study. So could you go to the next slide? I will take on the first primary objective. Well, this is typical demographics. There was really nothing unusual here for us. So please go to the next slide. I'll just mention nothing unusual, but from a PK perspective for the clinical pharmacologists in the audience, there was a balance of male and female and the age, obviously was the normal adult, but there was a little bit of spread between young adults and the elderly population.

So from a PK perspective, we thought it would be interesting to interrogate that data a little bit later on to see whether there was any variability or source of variability based on some of those variables. So what did we see as a primary objective? We achieved in full the primary objectives for the study, which was exciting for us. I need not remind you that sometimes we get into the clinic first-in-human. There are a lot of molecules that are in the graveyard because they couldn't survive based on safety or some unfortunate around tolerability and so on. So that's why I made the statement that we achieved our primary objective in full.

The drug demonstrated a very, very favorable safety profile. We observed mostly grade 1, grade 2 adverse events, and these were manageable. And based on the mechanism of action and the class of the drug, we were expecting nausea, vomiting, which happened. This was very manageable. In fact, to the extent that we could have only want -- a single dose of antiemetics was really required when it was needed to control emesis in the patients. The other thing we observed, which is very encouraging for us was that unlike Irofulven, a predecessor in this class, we didn't have any visual or ocular toxicities. That was really an actually standard for Irofulven.

The dose-limiting toxicities, acute onset, reversible transaminitis in 2 patients at dose level 12. Thus, we had a 40% DLT rate using the BOIN design. So we declared that dose level 12, we weren't going to go any higher than that. And we had a lower than 30% DLT rate at dose level 11. So subsequently, that became the maximum tolerated dose. Again, those are the elements of the BOIN design. We observed on grade 2 toxicities, and we declared that dose level 10 actually will be a recommended Phase II dose.

Now in terms of the PK, very, very interesting for us, and we really observed good news here that the drug actually demonstrated a very high therapeutic ratio potential. From dose level 7 onwards, we're able to achieve drug concentrations that we believe if you look at the right-hand side of what we have, looking at the projected therapeutic drug concentrations, you begin to see despite variability into individual variability from dose level 7 onwards, we're able to achieve that therapeutic above therapeutic concentrations. And we think this enhances the opportunities for monotherapy as well as combination treatment for regimens.

We go on to the next slide. Now let me spend some time on what we're beginning to observe in terms of preliminary antitumor activity. This was very, very interesting to us and encouraging. Despite the fact that this was a heterogeneous patient population, we enrolled 63 patients. There were 52 that were evaluable for tumor response. As I speak to you, today, we still have 4 that are still on treatment, and I will detail those very soon. 28 patients were dosed of these evaluable patients above the therapeutic dose level were [ well ] for tumor response. And the majority of those patients obviously achieved stable disease.

Four patients had stable disease that was durable for greater than 6 months, and they are reflected below each GBM patient, a patient with GIST thymic carcinoma as well as non-small cell lung cancer patient. Now we decided to also do a deeper dive into dose level 10, which is the recommended Phase II dose, the way we see it based on this regimen. There were 4 patients at that dose that achieved disease control, that was 44% and 22% of those patients really maintained control beyond 6 months, and some of them are still on. Now we did something else, which was to ask the question, what is happening with patients that have the DDR alterations of interest. And those are reflected, ATM, [ CHK ], BRCA, like Kishor had talked about, KEAP1/STK11.

We are fortunate as we looked at this, that 8 of those patients, 57% really had stable disease. But more importantly, 21% of them maintained that for much longer time. Some -- in fact, most of them are still on for more than a year. Could you go to the next slide? So I thought I should highlight for you why we're excited and experience with some of these patients. So we take the first patient. It's a 50-year-old patient came into the study with obviously advanced standard carcinoma. Thymic carcinoma, obviously, is a very rare carcinoma. But nevertheless, this patient was diagnosed in December, about 13 years ago, completed 4 lines of therapy, including our standard of care regimen and some investigational product. And lastly, a protein degrader before coming into our study.

At study entry, we decided to look at what will be the expected PFS for patients like this. Now it is range in the literature depending on how many lines of therapy, but you end up with 3.8 to 14.7 months. Luckily for us, there was liquid biopsy done and this patient showed a CHK2 alteration. We started on treatment, LP-184 in November of 2024 and is currently on cycle 17 of treatment and have continued to have benefit beyond 12 months. And what was notable to us was obviously the patient has several lesions. There was this mixed response, but there was a maximum target lesion reduction of 26%. So gradually, we're beginning to see this reduction in individual tumors, but 26% is the most we observed till date.

The second patient is a 62-year-old female Stage 4 GIST diagnosed in 1994 and the natural history of this disease is that patient [ has been stable ] for a long time. But nevertheless, the patient has been on several lines of therapy, including, again, investigational agents and those that are in the standard of care. At study entry, the median PFS was 4.8 to 6.3 months. Again, the patient had an ATM alteration, and we're very interested to see how the patient does. This patient is now on cycle 19 beyond 1 year on study. We've had this very stable disease and the target lesion that was in the maximum reduction is about 9%.

And then the last patient I want to tell you about is a 62-year-old male, again, stage 4 squamous non-small cell lung cancer that was diagnosed about 3 years ago. He completed 2 lines of therapy, including standard of care and IO monotherapy prior to study entry progressed very rapidly through the IO agent. At study entry, again, the median PFS was expected to be 2.5 in 7.6 months. This patient has a BRCA1 alteration, started treatment about 2 years ago and currently is on cycle 34 of treatment and continues to have clinical benefit, very notable target lesion reduction of 22%, approaching a partial response.

Now the reason I mentioned the number of cycles of treatment is just to correlate or validate what I said, looking at LP-184, the mechanism of action and obviously, our preclinical data, you could see this is not your traditional cytotoxic chemotherapy. Patients are staying on for a very long time. And I will mention these patients that I've mentioned to you, none of them have had DLTs. There's been no discontinuation. We had very low rates of discontinuation interruptions in our Phase I study, including these patients. In fact, 2 of them had to go on to a higher dose level. So intra-patient dose escalation above [indiscernible] recommended Phase II dose.

Again, we wanted to understand that tolerability of the regimen as we prepare for that development in other indications. We talked about the PTGR1. Again, we are fortunate to have a lot of samples from some of these patients. And what we found out was that 87.5%, about 90% of our patients for which we're able to get samples had PTGR1 expression levels that were above what is required for bioactivation of LP-184. We also try to look at the patients that have prior therapy, but they had alkylating agents or prior DNA damaging agents and what was really happening. Again, we found that in majority of those patients regardless, had this very high level of expression of PTGR1, which is good news for us in terms of LP-184 at this stage of disease that we are going after for patients.

We go to the next slide. So let me spend some -- few minutes on our planned Phase I, Phase II clinical trials. Next slide, please. This is just another slide showing at a very high level. I think I will explain it in the next subsequent slide telling you about the market potential, the trial sizes that we're going after in this -- and the trial highlights as well. I think the subsequent slide will take each of these in some detail. So the first one is the monotherapy in relapsed -- in advanced pretty much metastatic triple-negative breast cancer patients with HR deficiency.

Now if you are familiar with what the FDA has been talking about this new Project Optimus, we talked with the FDA, they encouraged us to do this. So we have single agent 2 dose levels where we're looking at our recommended Phase II dose and perhaps another dose that we agree with the agency and then move on from there. The next study is -- okay, I'm looking at this, somebody went -- sorry, somebody went way too fast for me. Now the Phase II, obviously, is the Simon 2-stage. Again, we are trying to be very pragmatic and ask questions very early. Is this drug really doing what we expect it to do based on all our preclinical data and what we've seen in the clinic. And if the answer is no, we want to be able to pivot very quickly.

So the Simon 2-stage is an approach and the design for which that we are taking across our programs. The primary objective, again, in this case is to look for the right dose to be project Optimus and then preliminary activity. And secondarily, we look at how the longevity of those activities and also try to validate PK in a particular indication of interest as well. The next slide. And then there's the combination. I don't want to dwell too much on this. I think Dr. Kishor talked about it, the scientific rationale. But the only thing is, remember, when we looked at that preclinical data, there was synergy, not additive action.

So we're taking a very -- trying to take 2 things, try to see for a lower dose of LP-184 in this study, but also the dosing regimen to make it a little bit more frequently and ask the question whether that will really be the optimal dose for us to use in combination with Olaparib. The good news is, I told you about the safety profile of LP-184 in a single Phase I study. So we don't anticipate any overlapping toxicities that will make the regimen difficult. And then after the initial phase, we go on to again Simon 2-stage for the Phase II.

The next slide, please. The same is true for the non-small cell lung cancer with the KEAP1/STK11 mutation, and these are patients, obviously, that have a low PD-L1 expression. They do very poorly with standard of care, and we're trying to figure out the right regimen initially before we go on to the Phase II portion of the study, again, using the Simon 2-stage. The next slide, please.

And lastly, the IST, the investigator-sponsored trial that we're doing with Dr. Helle Pappot in Denmark. This is quite an interesting study. It's a biomarker-driven study. We know this is a very difficult population. And the one thing we know about this set of patients is that even when you have a dose from patients that don't have advanced bladder cancer, they tend not to still tolerate such doses too well. So the approach here from the investigator was to learn from what we have done and use a dose level 10, which is really a very soft spot. It's not a maximum tolerated dose, dose level 11 is. But -- so go below that and ask the question whether that will be okay for this patient population before we then go into a Simon 2-stage design again and ask the question of activity of this regimen in this patient population.

All right. The next slide. So let me conclude by spending a few minutes on the additional clinical development opportunities that we see. Of course, they are more than just what I will highlight here in terms of combinations and what have you. But the first thing to highlight is a post-radiation treatment opportunity for patients with pancreatic adenocarcinoma. What we do know and Dr. Igor is on the call as well is of interest to him also this strategy is that if you expose these tumors to radiation, you upregulate PTGR1, so they overexpress that. And the second thing we want to do, obviously, is to optimize tumoral exposure and those are now Phase 1 via alternative therapeutic dosing schedules. So I will stop there. I think that's my last slide.

Thank you, Reggie. Perhaps this is a good segue to also request some comments from Dr. Igor, particularly trying to understand how as a clinician you see these 2 set of results, the preclinical results, you have been a part of several of those studies. You know them quite well. And you've been aware of some of these upcoming clinical results. How would you tie these together in terms of where do you see the drug? How do you see it?

Thank you, Kishor and Reggie and Panna for first inviting me and also for pulling this off. This was -- when I started collaborating with you, it was kind of a remote lofty goal to develop this compound that was abandoned by drug industry like decades ago. And now we're at the point where when we're discussing how to move forward for specific indications, the fact that you have completed Phase I is remarkable that you see in this heavily pretreated patients months and months beyond 12 months activity and stable disease is clearly talking to -- speaking to the fact that this is a mechanism-based therapeutic. And I firmly believe that drugs that succeed in the clinic are the ones that are mechanistically based.

So the rationale for PDGR1 testing is very solid. I think it's clearly a way to convert the prodrug to a fully effective, fully functional alkylating agent. I think it's very interesting, the sort of idea of combination with other DNA damaging agents since the way it works on ERCC family mediated DNA repair mechanism is entirely different from double-stranded DNA damage repair that is dismantled by PARP inhibitor, or platinum. So in heavily pretreated breast cancer patients or any cancer patient has BRCA or homologous recombination deficiency, that will be really a niche where you can develop this drug for clinical use.

This also encompasses a significant subset of pancreatic cancer patients, about 10% to 15%. I do also think that the opportunity is with bladder cancer where the prevalence of ERCC3, ERCC4 mutations is pretty high. So I think this is something to look forward to see the results of bladder cancer trial from your collaborators in Copenhagen.

Thank you, Igor. If I understand correctly, Reggie, most of the patients we had in Phase I were those who had experienced at least 3 or 4 previous lines of therapy. There were none that had lower than that, correct?

Very small numbers, especially we have just a few numbers of patients with GBM, for example.

Yes. In GBM, yes, of course. I mean that was a very different...

Other than GBM, the other ones that came in a typical Phase I, if you have 1 or 2 that have 2 lines of therapy, that is the exception.

That's the thing. So how patients who are -- who have become refractory to multiple lines of therapy, still being stabilized with LP-184 monotherapy is pretty remarkable. I think. I mean, one would expect that tumors that have not been subjected to additional lines are likely going to be more sensitive.

That should be the case. Again, that's why we're trying to -- if you go in combination therapy like we're trying to do with non-small cell lung cancer, that obviously is first-line treatment, we expect to see better results. But nevertheless, it's almost like the flip side of the coin when it comes to safety, right? Despite the fact that these patients were heavily pretreated, I go back to the primary objective. We still saw the drug did so well in terms of the safety. So one is very good. In combination with Olaparib, which those patients will probably be second line first line.

Yes, they could be second or third. So one of the -- couple of observations, and we'll take some more questions. We do have Fast Track designation in triple-negative breast cancer. So that's one of the things that we're particularly excited about because if we see some -- based on the design of the Phase Ib, Phase II, the BOIN design, if we start seeing some really good indications, there's a lot of opportunity to go from there to a pivotal kind of approval. And so we can comment on that. But what we're also seeing, and I think we touched upon it slightly is that PARP inhibitors and LP-184 are really as predicted, of course, in silico, but they're really almost perfectly synergistic agents.

PARP inhibits the repair of the DNA of the cancer cells. And so when you have this PARP inhibition, even if you break apart, what our drug does, it works on the other side of the barbell. It destroys and creates a double-stranded breaks that then PARP works -- the PARP inhibitors work to delay or destroy the ability to repair it. So one is completely destroying. The other side is avoiding it from repair. So it's a very synergistic agent. We hope to see some really good data. But very importantly, because of the synergy we've seen preclinically, we hope to see some of that. And again, we've seen evidence in the BRCA patients.

So if you look at the BRCA patients, the BRCA patients had much lower dose levels, which then points to the fact that we were getting sufficient damage at a lower dose level because, again, the preclinical data suggested that there was synergy at lower doses of BRCA with lower doses. Now the dose that we used preclinically that we saw, again, complete tumor inhibition was at 0.75, which translates into almost a suboptimal dose in the humans. So it's kind of an interesting observation. And again, BRCA over time, as you take these BRCA inhibitors, PARP inhibitors, you start creating tolerability issues. And we've seen already in some of the ovarian cancers, you've already seen backing off of some of the approvals because of the tolerability issues that we've seen clinically.

So there's a real opportunity. Now again, we'd like to think of 184 as almost like as a pipeline in this molecule. Look, we've seen a number of cancers. We've seen a number of different ways to combine this drug with other drugs by combining mechanistically the ways that they would work together. And again, a lot of this is all data-driven. So...

Yes. If I may, there's another point that's, I think, particularly important, at least for me, it's been a very exciting aspect of LP-184 with respect to TNBC. A large proportion of triple-negative breast cancers, and there are 2 clinicians here, so correct me if I'm wrong, actually often present with brain mets. Now LP-184 also has a property that it crosses the blood-brain barrier. And in fact, some of the studies we did, which I didn't discuss, were asking, would LP-184 prevent brain mets. So we used animal models and we're basically able to show that, yes, it does. And therefore, as we move further along in TNBC, there's another aspect of LP-184 that most of the other drugs don't have.

There is a good penetration to the brain. Most of the metastases are settled as micrometastases until they become clinically apparent. So if you come in with this agent that can reduce the burden of these micrometastases, then sure, the answer is yes.

Kishor, Igor, Reggie, any closing comments or about 184?

I personally, I would just quickly -- I'll close by saying that given my experience, we didn't get into that. It's not about my background, but the reality is that when I look at a drug like this and the experience so far, the fact that you're able to dose this long in patients and not find any kind of cumulative toxicity. I think to me, that's very, very encouraging because that's a problem with most drugs.

One other thing that I pleasantly encouraged PDGR1 is expressed in hepatocytes and actually a pretty significant level. So I was worried about potential hepatotoxicity and remarkable is you don't see much of it. It's really credit to you, first of all, pulling off this amazing Phase I trial to demonstrate the tolerability of the agent and really it kind of paves the path forward to its full clinical exploration.

Yes. That lack of hepatic toxicity is a particular strong evidence of the dual aspects that require LP-184 to be cytotoxic, which is not just PTGR1, not just DNA repair deficiency, having both these together. And that's why I think it becomes such a great molecule for precision therapy, where you can understand, does the tumor have high PTGR1? Does the tumor have HRD, NER, replication stress. Then yes, in that tumor, this molecule is going to kill it.

Well, great. Kishor, Dr. Igor, Dr. Reggie, thank you guys all for your time. I want to thank our audience for participating. We're able to get this done exactly in 1 hour. So that's pretty good, very good rehearsal. So thank you all for joining us. Please send us your questions, and we look forward to future trials. And more importantly, we look forward to answering more questions about ideas and where to take this molecule in other interesting combinations as well. Thank you, everybody.

All right.

Thank you.

Good morning, and welcome to our third quarter 2025 earnings call. As a reminder, this call is being recorded [Operator Instructions]. A webcast replay of today's conference call will be available on our website at lanternpharma.com shortly after the call.

We issued a press release before the market opened today, summarizing our financial results and progress across the company for the third quarter ended September 30, 2025. A copy of this release is available through our website at lanternpharma.com, where you will also find a link to the slides management will be referencing on today's call.

We would like to remind everyone that remarks about future expectations, performance, estimates and prospects constitute forward-looking statements for purposes of safe harbor provisions under the Private Securities Litigation Reform Act of 1995. Lantern Pharma cautions that these forward-looking statements are subject to risks and uncertainties that may cause actual results to differ materially from those anticipated. A number of factors could cause actual results to differ materially from those indicated by forward-looking statements, including results of clinical trials and the impact of competition.

Additional information concerning factors that could cause actual results to differ materially from those in the forward-looking statements can be found in our annual report on Form 10-K for the year ended December 31, 2024, which is on file with the SEC and available on our website. Forward-looking statements made on this conference call are as of today, November 13, 2025, and Lantern Pharma does not intend to update any of these forward-looking statements to reflect events from circumstances that occur after today, unless required by law. The webcast replay of the conference call and webinar will be available on Lantern's website.

On today's webcast, we have Lantern Pharma's CEO, Panna Sharma; and CFO, David Margrave. Panna will start things off with introductions and an overview of Lantern's strategy and business model and highlight recent achievements in our operations, after which David will discuss our financial results. This will be followed by some concluding comments from Panna, and then we'll open the call for Q&A.

I'd now like to turn the call over to Panna Sharma, President and CEO of Lantern Pharma. Panna, please go ahead.

Good morning, everyone, and thank you for joining us to hear about our third quarter 2025 results and corporate progress. As many of you have heard me say in the past, computational and AI-driven approaches are increasing their presence and usage at both large and emerging pharma companies for all facets of drug discovery and development.

Lantern's leadership in the innovative, efficient and pragmatic use of AI and machine learning to transform the process of developing precision oncology therapies should yield significant returns for investors and for patients as our industry matures and adopts an AI-centric data-first approach to drug development.

This past quarter has been transformative in many respects for Lantern Pharma, a quarter where we have met many clinical, regulatory and validation milestones. And we have also significantly advanced the commercial availability and launch of our AI modules.

The third quarter of 2025 represents a pivotal inflection point for Lantern Pharma. We've made significant advancements across our clinical stage portfolio, while simultaneously expanding the capabilities of our proprietary AI platform, RADR. And we've also set up the future of our CNS-focused subsidiary, Starlight Therapeutics. These achievements position us well for multiple value-creating catalysts in the coming quarters and years.

Let me share with you some of the more notable achievements this past quarter. Let me start with what I believe is our most significant milestone to date clinically. Our LP-184 Phase Ia clinical trial successfully achieved all primary endpoints, demonstrating a 48% clinical benefit rate in evaluable cancer patients who received doses at or above the therapeutic threshold. What's particularly exciting is that we observed marked tumor reductions in patients harboring DNA damage repair mutations, specifically in CHK2, ATM, and STK11/KEAP1 genes. This validates our AI-driven precision medicine approach and the hypothesis of synthetic lethality and DNA damage repair that guided this program from the start.

On the regulatory front, we completed a productive FDA Type C meeting for our subsidiary, Starlight Therapeutics, a company is focused entirely on CNS cancers. The agency provided clear guidance and pathway clarity for our planned pediatric CNS cancer trial targeting an ultra-rare brain cancer, ATRT. Importantly, the FDA confirmed our strategy to combine LP-184, which we will call STAR-001 in this indication with spironolactone based on our preclinical synergy data.

We also made important progress across our broader pipeline. Preliminary Phase II data from our LP-300 HARMONIC trial were presented at the 66th Annual Meeting of the Japan Lung Cancer Society. We're planning a more comprehensive data update via webinar this December.

For LP-284, our non-Hodgkin's lymphoma program, we showcased clinical data at the 25th Annual Lymphoma, Leukemia and Myeloma Congress. The presentation generated interest from both biopharma companies and clinical investigators, and we've initiated several discussions around combination therapy opportunities.

Building on the Phase Ia results from LP-184, we're now positioned to advance LP-184 into multiple targeted Phase Ib, Phase II trials. Our precision biomarker-driven strategy will focus on 4 high-value indications, triple-negative breast cancer, non-small cell lung cancer with KEAP1 or STK mutations, bladder cancer with DNA repair deficiencies and first recurrent GBM. Collectively, these indications represent a combined annual market potential exceeding $7 billion.

To provide additional insight into the LP-184 data and our development plans, we're hosting a KOL-led scientific webinar on November 20 at 4:30 Eastern. Dr. Igor Astsaturov from Fox Chase Cancer Center will join us to discuss the clinical results and what they mean for the future of this program.

Beyond our clinical programs, we demonstrated the commercial readiness of our RADR AI platform at the inaugural AI for Biology and Medicine Symposium. We showcased several platform modules as deployable, highly scalable web accessible AI tools that can be licensed to biopharma partners and research centers. It's an important step in our strategy to monetize the technology that powers our drug discovery efforts.

Finally, I want to emphasize our continued commitment to disciplined capital management. As of September 30, we had approximately $12.4 million in cash, cash equivalents and marketable securities. Based on our current operating plans, we expect this provides runway into approximately the third quarter of 2026.

Before we turn to the financials, let me provide some color and details around our programs, both our drug programs and our growing program of AI modules, which we believe have the market potential of several hundred million on their own as AI tools and services.

First, some context on the Phase Ia trial. This is a first-in-human study that enrolled 63 patients, a fairly large number given that we started at a very low dose and escalated upwards. This was in advanced solid tumors who had exhausted all standard treatment options, which is fairly normal for Phase I studies. These are heavily pretreated cancers, oftentimes in very difficult to treat tumors. The trial, which you can find on clinicaltrials.gov as NCT05933265, successfully met all of its primary endpoints. The headline number that I want you to focus on is this. We observed clinical benefit with 48% of evaluable patients who were treated at or above the therapeutic dose threshold.

In a Phase Ia trial in heavily pretreated patients with advanced disease, that's a unique and promising signal of activity. But what's even more compelling is where we saw that activity. The data validated our core hypothesis about synthetic lethality. Patients whose tumors harbor specific DNA damage repair mutations, particularly in CHEK2, ATM and also STK/KEAP1 and actually also BRCA showed marked tumor reductions. This is what exactly what our RADR platform predicted well before starting this trial. And seeing it play out in actual patients is tremendously validating, but also very uplifting for our team where we can see how AI is being used for good and having a real-world impact on improving and changing outcomes. For us, this also gives us a very clear safety standpoint.

LP-184 demonstrated a favorable profile with minimal dose-limiting toxicities. This is critical because it gives us flexibility. We can now pursue both monotherapy approaches and combinations with agents that we have identified as synergistic such as PARP inhibitors and immunotherapy, also spironolactone. Both -- these all have been predicted through our AI platform, again, as I note, before the trials even began.

Let me give you a few clinical examples that really illustrate the potential here. In recurrent GBM, one of the most aggressive in treatment-resistant cancers, 2 out of 16 patients showed disease stabilization despite prior exposure to multiple therapies such as TMZ, lomustine and radiation. In GBM, as you will learn during our webinar on the 20th, we have the flexibility to modulate and enhance the efficacy of LP-184 by a factor of 3 to 6x, a potentially game-changing improvement. Even more encouraging, 2 patients at a dose level 10 have now maintained disease control for over 8 months and remain on treatment today. This is much more durable than has been expected for most Phase I studies.

We also saw durable clinical benefit in other notoriously difficult tumor types, gastrointestinal stromal tumors and thymic carcinoma. These aren't common cancers, but they're devastating when they occur and options are extremely limited. Our work in these rare cancers has also encouraged us to double down on our desire to transform the world of rare cancers and develop an open access tool for rare cancer drug development, codenamed withZeta, which I'll talk about a little later this morning.

Transitioning to clinical expansion. So the obvious question is this, what do we do with these results? And this is where our AI-driven development strategy really shines and demonstrates its value. Rather than pursuing a traditional broad Phase II basket type trial, we're taking a precision medicine approach. We're positioning to launch in 4 targeted Phase Ib, Phase II trials. Each one focused on a specific biomarker-defined patient population, where LP-184 has the highest probability of success and the best synergy agent for that particular tumor indication. One of these trials in Denmark in recurrent advanced bladder cancer is an investigator-led study. We have made this molecule into a portfolio of opportunities using data and precision oncology approaches.

So let me walk through these quickly. The first one is in triple-negative breast cancer. It's our largest market opportunity, almost $4 billion. We're pursuing 2 parallel approaches, one in monotherapy with DNA repair gene mutations and a combination study with a PARP inhibitor, olaparib, specifically in BRCA-mutated patients. We've already received FDA Fast Track designation, which will expedite our development time line. We expect to enroll approximately 60 patients across both arms upon full enrollment.

Second, non-small cell lung cancer with KEAP1 or STK11 mutations. This is a genetically defined subset of lung cancer who typically have very poor responses to immunotherapy. We're combining LP-184 with nivolumab and ipilimumab, 2 checkpoint inhibitors in patients with low PD-L1 expression. This represents, we believe, just in the U.S., close to $2 billion and probably closer to $3-plus billion globally. Again, we have an FDA Fast Track designation submission in process, and this trial will enroll approximately 34 patients.

Third, an investigator-led trial in bladder cancer, recurrent advanced bladder cancer. This is being led by Dr. Pappot at Rigshospitalet in Denmark. It's focused on patients with advanced urothelial carcinoma who have specific markers indicating DNA repair deficiency. This represents, we believe, about a $500 million-plus global market opportunity, and we expect to enroll about 39 patients.

Finally, first recurrent GBM, which we're pursuing through Starlight Therapeutics. Here, we're combining LP-184, which we will call STAR-001 and CNS indications with spironolactone. This combination showed synergistic activity in our preclinical models. We have both FDA Fast Track and Orphan Drug Designation for this indication. This trial will use a Simon 2-stage design with 2 separate arms based on IDH mutation status. We expect to enroll about 38 to 40 patients and represents what we believe is about $1 billion in U.S. market and probably closer to $2 billion globally.

When you add up these indications, they represent a combined market opportunity exceeding $7 billion. And critically, each trial is designed with biomarker-driven enrollment criteria that increase our probability of success. In fact, as you've probably heard me say in the past, biomarker-driven cancer trials increased the success by 4 to 12x.

Now rather than pursuing broad basket-like development, we're taking a very directed approach investing our resources exclusively in patient populations where the Phase I data and our AI-driven RADR insights predict meaningful clinical benefit and where there is real commercial opportunity and patient need. This is precision oncology at its best, using AI to identify the right patients in the right indications with the right combination drugs. And it all flows directly from what we learned in the Phase Ia trial, which was also heavily supported and predicted by the in silico AI work of our team and with multiple publications prior to that.

Now let me turn to our LP-300 program and the HARMONIC trial, which addresses a significant growing need in lung cancer, lung cancer and never smokers that have progressed after treatment with TKIs. This is an important distinction. In Asia, never smokers represent 33% to 40% of all cases compared to only about 15% to 16% in the U.S. and Europe.

This demographic reason is one of the reasons why we expanded this trial into Japan and Taiwan. It gives us access to the patient population, and it gives access to pharmas who want to develop therapies for this population. The market opportunity here is substantial globally, approaching $4 billion annually, and there are no current therapies approved for this patient population. But it is a space that more companies are interested in and are developing interested -- and are developing interest and are trying to approach it with various targeted combination opportunities. There's a real white space here that we're going after and the potential even to get to an earlier line of treatment. We completed enrollment in Japan this past quarter at 5 clinical sites, and we presented data at the 66th Annual Meeting in the Japan Lung Cancer Society, which was presented by Dr. Jonathan Dowell from UT Southwest.

Now the preliminary data from this trial, which we've already shared publicly, showed 86% clinical benefit rate, which is very encouraging. And we have one patient who has demonstrated a durable complete response with survival continuing for nearly 2 years, a remarkable outcome. I think we have another patient, which is now approaching a year. Now we're planning a more comprehensive webinar in December before the year closes where we'll present additional patient follow-up data and clinical readouts from both the Asian and U.S. cohort. This will give us an opportunity to discuss the data in much greater depth and provide regulatory strategy insight and positioning moving forward.

I should also mention that during the third quarter, we made a strategic change in our clinical operations in Asia. We transitioned our CRO services in Taiwan with a specific focus on cost reduction and operational efficiency. In Japan, we supplemented our team by bringing more activity in-house. This is part of a broader commitment to disciplined capital management and efficiency while maintaining the quality and integrity of the trial. The strategic positioning of Harmonic also opens doors for potential regional partnerships in Asia and co-development opportunities where the never smoker population is most prevalent.

Now let me turn to LP-284, a program targeting recurrent non-Hodgkin's lymphoma, which has generated interest from clinical communities and also from biopharma to approach combination approaches. This is our first in-human trial for LP-284, which we expect to enroll about 30 to 35 patients with aggressive recurrent non-Hodgkin's lymphoma, including mantle cell and high-grade B-cell, where we have orphan indications for both. This represents a global market opportunity of about $3 billion and with patients who have failed multiple prior lines of therapy and have very limited options.

In fact, in October, we presented clinical data from this ongoing trial at the 25th Annual Lymphoma Leukemia Myeloma Congress in New York City. The cornerstone of that presentation was a heavily pretreated patient with aggressive grade 3 B-cell lymphoma, specifically DLBCL, who had exhausted standard therapies, and we saw a complete metabolic response with LP-284 as monotherapy after 2 doses, 2 cycles. This is exactly the kind of signal we're hoping to see and validates many of our preclinical hypothesis for this drug. It also validates the mechanistic insight, and we saw complete metabolic response and the lesions around the hips and spine completely went away. This patient has now remained cancer-free since we initially reported this result in July Q2 of this year.

LP-284 has a novel mechanism of action. It demonstrates particular lethality in cells with DDR, a targetable vulnerability that's common in non-Hodgkin's lymphoma. This mechanistic differentiation is what's driving interest from partners. Now following this presentation, we've started discussions with investigators and companies around opportunities for combination therapy development with existing FDA-approved agents, post-immunotherapy treatment strategies and leveraging the 284 mechanism where current therapies are failing, especially in what's exciting indications beyond lymphoma.

Based on preclinical data, we're evaluating 284 and rituximab as a potential alternative to cyclophosphamide and methotrexate in lupus, systemic lupus SLE. Our preclinical models showed that 284 reduced urinary microalbumin and kidney damage -- which is a key marker of kidney damage in lupus by approximately tenfold and depleted B cells by fourfold when combined with rituximab. We saw even greater B-cell depletion when both agents were used together. This suggests LP-284 could become a next-generation B-cell depleting therapy in a number of autoimmune diseases, which would dramatically expand the commercial opportunity for this asset.

LP-284 also benefits from strong intellectual property protection. We have composition of matter patents granted in U.S., Europe, Japan, India and Mexico, providing exclusivity through at least 2039. The molecule, as I mentioned, also has Orphan Drug Designation in mantle cell and high-grade B-cell lymphomas. We're now focused on recruiting additional sites with a focus on non-Hodgkin's lymphoma and high-grade B-cell lymphomas.

The momentum we're seeing with LP-284, both clinically and in terms of partner interest reinforces our view that this asset has significant opportunity, both stand-alone as a wholly owned program or as part of a strategic collaboration. And we're very open to those discussions, both again in combination in non-Hodgkin's or in other autoimmune categories.

Now transitioning to our AI platform discussion. As I mentioned earlier, I want to shift gears and talk about what I believe is an increasingly important value driver for Lantern, our RADR AI platform and the commercial opportunities it represents independent of our drug development programs. For those less familiar with RADR, it's our proprietary AI and machine learning platform. And it's not just a tool we use internally. It's now a commercial asset with its own revenue potential, which is growing.

The platform has demonstrated over 80% prediction success across multiple use cases, and now it's been validated in natural clinical trials through programs like LP-184, LP-284 and also with Actuate Therapeutics. In all cases, it's correctly predicted biomarker responses and in many cases, combination synergies before we've even actually enrolled a patient. We've developed 8 distinct AI-powered modules that address critical pain points in oncology drug development. And we've developed cases for these pain points, which we're now developing into modules for the broader drug development community.

In October, we showcased the commercial readiness of 2 RADR modules at the inaugural AI Biology and Medicine Symposium. We demonstrated that our AI platform, PredictBBB achieves a 94% accuracy for BBB permeability prediction and can screen 200,000 molecular candidates in under a week. To put that in context, our algorithms currently hold 5 of the top 11 positions on the therapeutic data comments, and that's a best-in-class performance.

We also presented our LBx-AI liquid biopsy platform, which has achieved 86% to 90% accuracy in predicting treatment response initially in non-small cell lung cancer, which will be very useful for us. And now we're extending it through collaborations with research centers into other indications as well.

Both of these opportunities, we believe, are significant. Blood-brain barrier technology market alone is predicted to be close to $1 billion. And when you consider a very few percentage, 2% to 6% of the molecules actually cross the BBB, there is a need for better predictive tools, one that don't take weeks or months and end up destroying animals. So the need there is obvious and urgent. The interesting thing, PredictBBB is it also gives us access to a lot of other molecular characteristics of that compound, and we can predict a lot of other drug-like features that are important, both for drug manufacturing and also predicting potential drug activity once delivered internally.

Now let me introduce Zeta. It's our multi-agentic AI platform for rare cancers. Very excited about this and what connects directly to our experience with both LP-184 and 284 in rare tumors like gastrointestinal and thymic carcinoma. It's our newest initiative. We're calling it withZeta. It's a multi-agentic co-scientist.

Now here's the fundamental challenge. In rare cancer research and drug development, which often comes after molecule developed that often comes much later, the critical insights in rare cancers are scattered across disconnected data sources. A researcher or a clinician trying to understand treatment options for a patient with a rare sarcoma or a rare pediatric brain tumor has to manually search through clinical trial databases, PubMed genomic databases, drug interaction databases, molecular feature databases. It's fragmented, time-consuming and inevitably incomplete.

For drug developers, this fragmentation slows discovery, increases cost and often means that promising connections between existing molecules or indications and rare cancer vulnerabilities are [Audio Gap].

What Zeta does is a multi-agentic AI system. Think of it as a co-scientist that addresses this problem head on or actually a series of co-scientists. We've integrated curated rare cancer databases and ontologies across over 500,000 clinical trials, 250,000 publications with over 1.2 million knowledge objects into an agentic large language model architecture that uses recursive reasoning loops to transform fragmented biomedical knowledge and insights into an interconnected investigational platform. And it interacts with you in plain English. So -- and it's an AI system that thinks like a scientist, connects dots across disparate data sources and can answer complex questions in minutes about rare cancers. These are things that would otherwise take researchers weeks or months to investigate manually.

We'll dig into more of the details about Zeta in the coming days, and we'll have more information as -- but the key is that it will help you design and improve and optimize molecules that can target vulnerabilities or mechanisms across these hundreds of rare cancers.

So you can ask questions to Zeta like what existing molecules with blood-brain barrier penetration have shown activity against mutations commonly found in a specific pediatric brain tumor and will search, reason and provide evidence-based answers with citations, and you'll be able to have it quickly pick potential combination regimens as well for that rare cancer benchmarked against successful and not successful trials across drug classes that you can help Zeta understand. And it can actually also predict potential efficacy in subtypes of that rare cancer and give you considerations that can then be taken to the lab.

From an industry and business value perspective, withZeta delivers several things: Speed, smarter decision-making, novel discovery and potential for improved patient outcomes faster and most importantly, massive cost and time savings across the rare cancer drug development cycle.

I'd like to think about withZeta strategically is that we're positioning Lantern as a unified team of AI co-scientists, always available, always updated for rare cancer research and drug development, a unified AI interface for complex scattered data and models that accelerates and improves novel therapy discovery and trial design. This is a tool that can shorten development timelines by months and years, particularly in rare cancers where that data is sparse and every delay means challenges and time and lives lost.

By making Zeta available to researchers and clinicians over the next month, we'll establish Lantern as a central hub for rare cancer drug development and insights. This creates network effects, brings more users and data into our ecosystem and positions us as a trusted partner when those researchers need to take the next step, whether it's preclinical development, biomarker validation, clinical trial design or co-development.

Now let me briefly discuss how we're scaling our AI infrastructure to support Zeta and RADR's continued expansion. We're establishing dedicated machine learning and data engineering teams in India, which will allow us to double or triple our technical team size while maintaining our current cost structure. This gives us round-the-clock development capabilities, access to world-class machine learning talent at reduced costs and the scalability to support multiple drug programs and additional biopharma partnerships simultaneously. When you connect the dots, our clinically validated RADR platform, the commercial-ready modules we're deploying and withZeta positions us as a hub for not only rare cancer, but cancer drug development and the infrastructure to scale.

Now we believe our AI tools and services in the future can represent several hundred million dollars in stand-alone market potential and will attract a lot of interest in the broader big tech community, and but most importantly, lower the risks and costs associated with creating cancer drugs. And that's a very powerful complement to our drug development strategy.

Now I'll turn over the call to our CFO, David Margrave, who will provide details on our financial results for the quarter.

Thank you, Panna, and good morning, everyone. I'll now share some financial highlights from our third quarter ended September 30, 2025. Our R&D expenses were approximately $2.4 million for the third quarter of '25, down from approximately $3.7 million for the third quarter of 2024. The decrease was primarily due to decreases in research study and materials expenses relating to the conduct and support of clinical trials as well as decreases in consulting expenses and in payroll and compensation expenses.

Our general and administrative expenses were approximately $1.9 million for the third quarter of 2025 compared to approximately $1.5 million in the prior year period. The increase was primarily attributable to increases in business development and investor relations expenditures as well as increases in other professional fees and increases in patent costs.

We recorded a net loss of approximately $4.2 million for the third quarter of 2025 or $0.39 per share compared to a net loss of approximately $4.5 million or $0.42 per share for the third quarter of 2024.

Our cash position, which includes cash equivalents and marketable securities was approximately $12.4 million as of September 30, 2025. We believe our cash, cash equivalents and marketable securities on hand as of the date of this earnings call will enable us to fund our anticipated operating expenses and capital expenditure requirements into approximately Q3 2026. We will need substantial additional funding in the near future, and one of our key objectives is to pursue additional funding opportunities.

In July of this year, we entered into an ATM sales agreement with ThinkEquity as sales agent, pursuant to which Lantern may offer and sell up to $15.53 million of its common stock from time to time in at-the-market offerings to or through our sales agent.

During the quarter ended September 30, 2025, we sold 212,444 shares of common stock under the ATM for gross proceeds of approximately $989,000. Between October 1, 2025, and the date of this earnings call, we've sold an additional 144,204 shares of common stock under the ATM for gross proceeds of approximately $634,000.

As of September 30, 2025, we had 11,040,219 shares of common stock outstanding with outstanding options to purchase 1,218,828 shares and no warrants outstanding. These outstanding options, combined with our outstanding shares of common stock, give us total fully diluted shares outstanding of approximately 12.26 million shares as of September 30.

And I'll now cover some near-term milestones that we think will accelerate value for investors. And these are several value-creating catalysts that we see in the near future. In the immediate near term, in this November, and Panna talked about this earlier, and we're very excited about this discussion. Next week, November 20, at 4:30 p.m. Eastern, we're going to have a KOL-hosted scientific webinar on LP-184 Phase Ia details from the clinical study and clinical development strategy. And in December of this year, we'll be giving for LP-300, an interim patient follow-up and additional clinical data.

And then also in this upcoming quarter, we'll be discussing continued commercial developments for the AI platform modules, including the multi-agentic system that Panna discussed about withZeta for rare cancer development.

And I'll now turn things back to Panna for some closing remarks.

Thanks, David. As you know, we've had a number of catalysts and objectives that continue to '26, which you can see on the slide, but we'll be talking about those in follow-up meetings with investors as well. But as you can see, by integrating our capabilities in AI and bringing them to the public, we're not just building better tools. We're actually fundamentally reimagining what's possible in precision oncology, an era that I call the golden age of AI in medicine.

As we advance into 2026, we're laser-focused on executing our dual engine strategy. We got really 2 powerful engines of the company. One is the ability to generate new molecules that are very precise and focused on very unique cancers. And the second engine is the engine of our AI platform that we're now ready to commercialize and make available. So we're advancing our clinical assets while simultaneously scaling our platform for commercial deployment.

So I want to thank our exceptional team, our partners, our shareholders for their continued support. Together, we're lighting the way toward precision oncology solutions, solutions that can improve outcomes for cancer patients while very importantly, transforming the economics of drug development.

With that, I'd like to open the call to questions and also thank our team for helping to prepare us for these calls and preparing the content.

So we've a question in about tracking toward an interim event analysis for LP-300 trial.

At the December webinar, we do not believe we'll be at the 31 events, which is good news because that means that patients are coming off of the trial. So the positive news is that patients are on the trial longer, but we will report out data, clinical data and insights that have resulted. We expect 31 events right now, we're tracking to be sometime in early '26, which we think is actually a very positive news.

We do expect to see the Denmark trial. There's a question for the Denmark trial. That has now been approved. IRBs are set. project manager has been assigned. We expect that to start sometime either in late December or early January at one site, which is investigator-led in Denmark.

Another question is that we've guided for an IND submission for the pediatric CNS program.

Yes, now that the FDA is kind of back in business and looking and renewing new INDs, we're already prepared to submit that, and I expect that submission to happen here in the next few weeks. In terms of when we anticipate initial patient dosing, hard to say. We're already beginning discussions with sites, but I expect that to be sometime in early '26.

There's a question about the withZeta portion of our AI platform. We will have additional news next week on withZeta, which is very exciting. Like most software, we expect the early rollout to be interesting and bumpy. We'll learn a lot from it. We've already begun using it internally. And in fact, we'll talk about this next week, but we've got a number of really exciting programs that have already been designed and are now being tested as a result of withZeta. But it will be available as select demo to collaborators and select partners. And so December will be a lot of demo and learning and broader rollout throughout January and February and Q1.

Next question is for 184.

Yes, for the indications, we do plan on figuring out what is the best of those indications where we're getting the biggest impact and move that into larger scale trials ideally with partners. As I mentioned, [ Boris ], all those indications are very exciting indications, and we've had interest from pharma companies. Of course, they want to see some of the early Phase Ib, Phase II data, but all of those are potentially partnerable.

Next question is Zeta.

Yes, Zeta was initially developed as a culmination of our internal efforts to develop drugs initially 184 and 284 for rare cancers. We wanted to go after categories where there was no therapy approved, categories there was high need, categories where we thought the mechanism would work and could be exploited. As we did that and we gathered information about some of these cancers, we said, well, we can do it for all rare cancers. There's no tool out there. In fact, when we talk to other rare cancer experts, many of the cancers we're pursuing, it was scattered. Papers were hard to get, hard to get in front of experts, hard to get data. Trials were oftentimes took way too long and standards of care often changed or the best drug often changed. And we said, this is part of the frustration in these cancers, and that's why they take time or too much money. What if you could actually have one source and then train that source to think in the way that a drug developer thinks. So yes, it was an internal effort, and now it's going to be a front-facing natural language interface tool. And I'm happy to give you [ Boris ], if you'd like peek at it and even early demo, happy to provide that to you.

Another great question on STAR-001 trial design for pediatric brain tumors.

Yes, I do believe that the trial design allows for inclusion of other pediatric high-grade gliomas. Yes, we designed it to allow for that, including specifically diffuse midline gliomas.

Okay. If there are no further questions. I want to thank everyone for joining and very importantly, for listening in this morning. We know it's a little past the market open. So I appreciate all of you staying online. Thank you very much for your time, and I appreciate everyone's effort and also more importantly, your support as Lantern Pharma continues to transform drug development in oncology.

Thanks a lot.

Hello, everyone. This is Paul Kuntz with RedChip Companies. I want to thank you for joining today's event with Lantern Pharma, which trades on the NASDAQ under the ticker LTRN. With us today, we have Panna Sharma, Chief Executive Officer, President and Director of Lantern Pharma. We will begin with a brief presentation in a moment, and then we'll open up the event to your questions. You may submit your question at any time by simply clicking the Q&A button at the Zoom window.

Before we begin, I will read the safe harbor statement. This call may contain forward-looking statements within the meaning of the Private Securities Litigation Reform Act of 1995. All statements pertaining to future financial and/or operating results, along with other statements about the future expectations, beliefs, goals, plans or prospects expressed by management constitute forward-looking statements. Any statements that are not historical facts should also be considered forward-looking. And of course, forward-looking statements involve risks and uncertainties.

With that, I will now turn the webinar over to Panna. Please go ahead.

Paul, thank you very much, and I appreciate everyone taking time this afternoon to learn a little bit about our company. I'm going to talk a little bit today about how we've developed Lantern Pharma and also give you appreciation of the 2 major engines that drive our growth and uniqueness. A lot of you have heard about AI. A lot of you have heard about how drug development is becoming increasingly data-driven. And I'm going to give you real-life examples of how we're doing it.

Lantern Pharma, as Paul pointed out, we're publicly traded, LTRN. We're about 24 people headquartered in Dallas, Texas. And we're focused on using AI for one thing, AI for good. And we're focused on developing cancer medicines. Our methods, the way we do it is to develop new first-in-human drugs, of which we have 2 now that are being dosed in patients every week and also to repurpose or actually rescue drugs. So drugs that have failed have never gotten approved, but actually have some promise. And of those, we have one, LP-300, that's currently in the Phase II. Our other 2 drugs is LP-184. That currently just finished enrollment in a Phase I, a pretty large-scale Phase I, about 65 patients. And also LP-284, which is a sister drug to LP-184 that's in a trial in blood cancers.

So we've got a pretty robust portfolio. And on top of that, we actually have gained 11 FDA designations. 5 of those are orphan designations, 4 of these are rare pediatric designations and 2 Fast Track. And so again, this is a small company. Our burn rate is fairly small for a biotech, about $4.5 million a quarter. We're managing 3 trials and developing an AI engine. And so we've always taken the approach that we want to be able to develop drugs not only faster, but also more cost effectively and with greater precision. Without data and without AI, that's something that just is not possible.

And so when I joined the company, our big focus was to better understand the molecules that were in front of us that we could be developing, but also to understand how they worked, where they worked, where they didn't work, what molecules they work best in, what cancer indications do they seem most likely to actually make an impact? Where are they going to work better than other drugs that are approved or in late-stage development.

These are all the kind of questions that traditional drug developer has. Oftentimes, they'll spend years in a lab or years in different groups trying to work these problems out. Imagine being able to do those thousands of times over in silico. Imagine being able to have thousands of algorithms compete to give you an answer that you can get within days or weeks and not months or years. That's the power of how to use AI in the development of cancer medicine.

But in order to do that, you have to have the right data. And not only the data, you actually have to have the right models. Not only do you have to have models, but you have to have models that can be learning constantly that you can iterate and have a recursive process so that you're not just using the AI once off the shelf and putting it back on the shelf. The AI has to be a living, breathing system. Our AI platform has grown tremendously, and it continues to grow, and we're going to be rolling out some aspects of it publicly. In fact, one of the most recent things we did was we announced the public release of one of the modules in our platform that predicts blood brain barrier penetrability.

We also, at the end of our second quarter, when we announced second quarter results, we also talked about the completion of our Phase I trial with LP-184, which is a great milestone for us. We also talked about some very exciting complete responses that we had seen in 2 of our trials, one in aggressive B-cell lymphoma that was recurring with LP-284, another complete response in the primary lesions of a lung cancer patient with LP-300, and that's a trial for never smokers.

So again, the 2 engines, both of which are very important, but the vast majority of our spend really is on the trials and on the biotech side is to develop precision oncology medicines that are derisked, that we know have a certainty and have a clear mechanism that is driving action in the tumors that we want -- in the way that we want so we can predict outcomes and then also then to have an AI platform that is able to do this and grow this on its own over and over and over. And we've seen that not only in our own drugs, but we've seen that in our collaborators.

We have some great collaborators, Oregon Therapeutics, Actuate Therapeutics, companies developing very unique drugs on their own. Actuate is public. We own some stock in Actuate as a result of allowing them to use our platform. We have some potential IP with Oregon Therapeutics and their very unique drug as well.

So part of our business model is to continue to mature the AI platform, RADR, but also to then develop these medicines and then license them out to biotech and larger pharma companies. Lantern wants to focus on our mission, which is to develop innovative new insights develop those insights into molecules, bring those molecules into trials and do it faster and cheaper than ever before.

We think this is the model of the future of drug development to do it using data, to use it using insights that can be replicated from theoretical cancer biology to real-world patients and then more importantly, understand how drugs can be combined because oftentimes in cancer, it's not just one mechanism that will destroy cancer. You've got to use multiple mechanisms to actually drive a more durable, more deepened response. And we've seen that over and over in trials. And so one of our big things is to drive combination therapies. And so a lot of our AI work has been focused on combination therapies and finding how combinations will work together. So that's a little bit of an overview.

I'm going to talk a little bit about each of the drugs because the cancer indications that we're going after are really exciting, but more importantly, very much needed. The first indication is in Phase II. It's a $4 billion to $5 billion opportunity globally in annual sales. These are patients who are not smokers. They're never smokers, but they still get non-small cell lung cancer.

And ironically, there's no wonderful outcome for these patients. After they fail kinase therapy, if they're eligible for kinase therapy -- kinase inhibitors, there's really not a lot of good options. For us, it's white space. So we've developed a program that focuses on exploiting these kinase mutations. And so we have the only pan-kinase modulator that's in a Phase II clinical trial that also is very synergistic with chemotherapy. So this drug targets and kind of denatures the kinase receptor. So it slows down the growth of the cancer. And then once it's inside the cancer cell, it resets the redox cycle and allows the chemotherapy to kill off the cell.

So far, we've seen some great results in the early readout of the first cohort, we saw an 86% clinical benefit rate. We also saw a patient move from a partial response to a complete response, which is also very exciting. This is not just a temporary kind of complete response. This is fairly durable. We've seen this patient now have a complete response in their primary lesions now going on 2 years.

So that's very exciting. I mean it changed that kind of outcome for those patients. And we will have a readout with this patient group that now has expanded to Asia specifically Japan and Taiwan, where about 33% to 40% of new cases in non-small cell lung cancer are people who don't smoke. People who don't smoke have a totally different biological profile of their cancer. The mutations they have are different. The responses to chemotherapy are different. They don't respond to immunotherapy. And eventually, they do fail some of the targeted kinase therapies, and that's where our drug comes in.

So we have a very clear clinical path. We've got excellent initial data from the cohort. We've finished enrollment now in Japan and are looking for partners in Asia, and we expect to then have more data readouts later this fall, but also perhaps toward the end of the year as well. And so that is an asset that can be partnered out. We think there's no other drug targeting this never smoker population. And again, this is a $4 billion to $5 billion a year spent on this patient group.

The next drug that I'm going to talk about, LP-184 is a very unique molecule, first in human, and it's targeting a large range of solid tumors. We just completed enrollment in the Phase I trial, large trial, but 65 patients across a wide range of cancers, including GBM and brain cancer, including some lung cancers, et cetera. So a big range of cancers. Like most Phase Is, these will be fairly late-stage patients, meaning these patients have exhausted existing therapies because the focus of the Phase I really is to find an optimal dose that we can proceed with.

And the reason for this is really clear because the drug is very potent. And we feel we have a very good maximum tolerated dose and a recommended go-forward or Phase II dose. And this is a very small dose, but the drug is nanomolar potent, meaning very tiny amounts, nanograms per ml of this drug seem to kill off these tumors.

But it does it under certain very unique conditions. So like most precision oncology therapies, when the conditions are right in the tumor, and this happens to be in about 20% to 25% of cancers have either high levels of PTGR1, that's an enzyme that activates this drug inside of the cancer cell or they have what's called a deficiency in their DNA repair pathway. Either one of those conditions is around for LP-184, this drug really lights up.

And so we find that a wide range of tumors respond to this drug and -- but it's very potent. And so finding the right dose, finding the right tolerability level, understanding the pharmacokinetics of this drug is very important. And we also have now clearance on 2 Phase Ib, Phase II trials for this drug, including in triple-negative breast cancer and also in STK11, KEAP1 mutated lung cancers.

We also have an IST, an investigator-led study in bladder cancer. And all those cancers I mentioned, TNBC, bladder, STK11, KEAP1 mutated lung cancer, these are all very large indications multibillion-dollar indications. But most importantly, there are all categories in which their therapies are needed. And so for each of those, we have very clear signals that this drug works in that cancer, but actually works even better in synergy with other drugs. And so many of these will actually be combination trials.

Our last drug, LP-284 that's now in clinical trials is also very exciting. It's a sister drug to LP-184. And it targets B-cells, which we have 2 orphan indications in mantle cell lymphoma and in high-grade B-cell lymphomas. We think about $3 billion to $4 billion are spent every year. And again, clinical positioning is really important. So what we've discovered is that when people fail first-line or second-line mantle cell or high-grade B-cell therapy, there's not a lot of great options, and our drug seems to work really well and continues to drive a response in those tumors.

We actually saw this. This was all theorized, and we actually saw this in a recent patient. And again, it's really a wonderful moment for the company, kind of transformative because a lot of the theories and ideas that we had about how does the drug work? Where is it best positioned? Will it work in later-stage tumors that have become resistant to other therapies. All these publications and all these theories that we have, now we're seeing in patients. And there was a patient that had failed 3 prior lines of therapy. And in fact, internally, there was some debate whether this was going to be an ideal patient because they had failed some really state-of-the-art therapy. A bispecific antibody made by Janssen, great bispecific. They had failed stem cell transplant, they failed a CAR-T or they had some partial response and but not very durable.

I think it was only a month or so. And so there's the question, is this -- are we going to be able to make an impact? Well, after 2 doses on our LP-284 drug, this is in a high-grade B-cell lymphoma, recurrent and the patient had complete metabolic response. And this patient had lesions, cancer lesions up and down their spine and into their pelvis, and we saw complete resolution of those lesions.

And that's very exciting. We know it's only one patient, but it gives us a very clear signal that this drug is working. It's doing something. Now we have to get stats on our side and get a lot of these types of patients and drive similar kinds of response. And hopefully, the responses are durable enough and meaningful enough that we'll be able to get to potential accelerated or even breakthrough in that indication because when patients in these high-grade B-cell lymphomas and mantle cell lymphomas fail, the outcome is pretty poor.

And so again, to be able to make an impact in that patient group is very exciting. So that's why I think about our company as being a company that's doing AI for good. We're trying to take all this great data, all this great algorithmic capability, all the great infrastructure of cloud computing and leverage it to focus on the development of precision cancer therapies. We focus on all sorts of problems every day at the company. It could be a problem around manufacturing, improving manufacturing. It could be a problem around combinations, what combinations. It could be a problem around what other drugs -- sorry, cancer indications can this drug be pointed at that we're excited about.

All those are wonderful problems and can be solved not just once or twice, but hundreds of times or thousands of times using data and algorithms. And each time you learn something unique that then you can go do experiments, you can gather more data and recurs. Again, our model is not use AI and get an answer. Our model is use AI hundreds of times, thousands of times, get a library of answers, have those answers compete, enrich those answers with real-world data and iterate. And that's what allows us to have the model that we have.

We're maniacally focused on the next generation of cancer therapies. Today, we have 3 small molecules in the clinic. We have already a new generation of molecules that we're working on preclinically. Most of these are antibody drug conjugates or other forms of drug conjugates, which we think will be revolutionary in the market. That's a completely exciting new modality.

But our business model is a focus on that innovation, do early execution and then sell the assets off to larger biotech and larger pharma partners. Again, our AI platform, we plan on making that public. I'll talk a little bit more about that, what I call taking a chapter out of kind of the DeepSeek playbook. We publicly released our very first module called predictBBB.ai. You can go there today, sign up for an account. And you can predict any molecule and predict whether it's going to actually cross the blood brain barrier.

We have probably the most reliable and most scalable algorithm to predict any small molecule's ability to penetrate the blood brain barrier. Only about 2% to 6% of molecules actually cross the blood brain barrier. And it's an area that can take tens or hundreds of thousands or millions of dollars to figure out properly, and we can do it in seconds. And that's the revolutionary potential of these kinds of approaches of doing drug development in a totally different level, scale and time frame.

We'll have many more modules. This is a very tiny, tiny taste. We have a very important module coming out later this fall that's going to be focused on a much larger range of what's called a multi-agentic system. And we're going to take something that we do really well. And we have 11 FDA designations, 5 of them in orphan designations, 4 of them in rare pediatric designations, 2 of them that are fast track. So if you think about that, for a company of 22, 24 people, we have 11 designations.

So it's something we do very well, something we think about, something we enjoy thinking about, and we're going to bring that out to the public and allow that to see the light of day and allow people to start developing molecules and understanding rare cancers. And so that would be very exciting that will come out sometime in September. So the platform, major engine of growth and value in each of the molecules also. So -- and again, we're very focused on maintaining a very disciplined fiscal profile. We burned about $4.5 million a quarter, even with our 3 trials going on. And we'll have data from all the trials. We'll have data from our AI platform, and we have cash into middle of 2026.

So we're pretty well managed. No warrants, no toxic overhang, no debt. We've got a total of about 10.8 million, 10.9 million shares outstanding and about 12 million shares in total on a fully diluted basis. And lots of news coming out over the next several months. So with that, I'll take a quick pause. Again, A ticker LTRN publicly traded as Lantern, and I'll love to take questions from everyone today.

Thanks, Panna. Great presentation. We are now going to open up the call for questions. [Operator Instructions] One of the first questions we had, Panna, was, is there a business reason for making the modules available free to the public? And how will that increase revenues and shareholder value?

Yes. So I think part of the challenge with any kind of AI is that it's so black box and until tools like ChatGPT and DeepSeek and other tools became publicly available and open, it was hard for people to imagine and spend on this kind of AI black box. And so I think we're going to take a premium kind of approach where we open up these modules. We allow people to take and test these modules, use it 5x, use it 10x for free and then start purchasing tokens or enter into collaborations.

And so we can talk until you're blue in the face that our AI does X or Y or Z. But when you actually have pharma companies using it or developers using it, it becomes a whole different game. And so we want to be disruptive, and we think there's plenty of opportunity to do collaborations and charge for tokens, charge for use and allow people to transform their own work.

Great. Thank you. And we had another question. You have over 100 issued and pending patents. Where do you see the strongest moat, composition of matter, methods or RADR?

Yes. I think we have about 140 issued and pending applications now. Composition of matter is always important, but also a lot of -- is very important in terms of the methods, how do you plan on using that composition, where, how, under what circumstances, meaning what's biomarker signature with what other drugs. And so we've patented all that -- all those findings that we've had. And we also have patented several aspects of RADR as well.

I think each of the molecules has a wonderful patent estate around it. It's very important. Obviously, the most -- the longest patents are the ones that we've filed in the last few years. So those would be around 284 and also some of the findings for 300 and 184. So I don't think there's the strongest. I think they're all important and strong, but 284 since it's the newest probably is very strong, but it's also the smallest of the indication. So you got to juggle size of the market and duration of the patent, all those matter.

Software patents and algorithm patents are a little bit more challenging. They take longer to eventually issue. And by the time some of these will issue, there's going to be constant new things. I mean I think one of the things that we're very excited about is that we're thinking about is how is -- we see the wonderful world of how the current chip technologies has changed computing and computing cost. But imagine now what certain aspects of quantum computing, and I know that a lot of people think it's hype.

Yes and no. I mean I think the certain chip aspects are definitely much further along than we think. But the software embedded quantum computing is actually here now. And I think there's a lot that can be done in terms of being able to simultaneously design different states of molecules all at once. And that really can be a game changer because you can take that same concept and do it to simultaneously modeling different outcomes in cancer patients all at once instead of serial or even in parallel.

Imagine a system being able to think about 2 or 3 or 4 of these things and do it in a multiplex way. I mean you can -- I mean, it's really unreal what quantum computing is going to open us up into for really deep biology modeling. And so I think that's not 20 years out. I think that's much more near term. I think 2 to 5 years. And so I think that's going to be game-changing in terms of where AI can go next to be able to predict biology.

Thank you, Panna. And with your PredictBBB at around 94% accuracy, is that opening any doors in CNS, either for Starlight or for external partnerships?

Yes. We're in discussion with a couple of companies to use the algorithm to screen libraries. We're actually in discussion with groups to talk about using it as part of a trial as well to select in a basket trial that's going on for CNS cancers. So yes, it's generated a lot of interest. The challenge always and part of why it's public is that historically, these public tools have never been much more than about 70% accurate, maybe in the 60s, and they have limitations.

And our algorithm has limitations, too. But I don't expect $100 million of flown as a result overnight. So just to gauge -- I mean, I expect partnerships to occur. And I expect that PredictBBB will be a game changer in the way people start thinking about opening up AI systems for large-scale development.

And again, followed up with lot of as well. So I don't want to get into details, but when you predict a molecule's blood brain barrier penetrability and if you look at the details that we put on the BBB white paper and the details of the actual algorithms, we have algorithm cards that they can look at if people register, is that you can see that the -- we're taking into account thousands and thousands of parameters.

These parameters are molecular features. And so we have all those features of millions of molecules. So we're just using those features to give one piece of data, which is our prediction on BBB, but we can use all those features to predict lots of other features on a molecule.

And so imagine the next iteration of the predict BBB will not just predict its BBB, but it's going to predict all kinds of others. It's going to predict its kinetics. It's going to predict whether it's lipophilic or lipophobic. It's going to predict its bonding strength. It's going to predict how it donates electrons. It's going to predict how many open rings it has. It's going to predict its drug ability. It's going to predict potential safety issues.

So BBB is just a gateway to now being able to predict dozens of things about any molecule. And so that's really the game plan is BBB is just a gateway. And again, if you -- people who actually know the space when we look at, okay, well, how are they predicting it, they'll realize that in order to predict it, we've looked at, and it's in the white paper, over 8,742 molecular features of any compound in real time. And being able to do algorithms and machine learning across those features, again, in real time, across the web and then be able to give those is potentially revolutionary in the development of medicines.

And another question we had, why expand the LP-300 study to Japan and Taiwan? And what should we watch for from those new sites?

Yes, great question. So never smokers are fairly understudied historically group of patients. They historically have not responded very well to chemotherapy. They don't respond very well at all to immunotherapy. In fact, there's an immunotherapy trial that just launched and it's on the label, it says -- sorry, on the criteria, inclusion/exclusion criteria, not label. It says exclude people who are never smokers because never smokers have a very poor response to immunotherapy.

But now in the United States, about 15% to 20% of new cases of non-small cell lung cancer are never smokers. That's 30,000, 40,000 patients a year. It's not a -- it's a pretty sizable population. Globally, it's closer to 170,000 to 200,000. In Japan, Taiwan, South Korea, parts of China, the never smoker population, people who have less than 100 cigarette events or tobacco events in their lifetime is a much larger percentage approaching 35% to 40-plus percent.

So it's double that in the United States. It's a known issue there. A lot of KOLs have been trying to study it. Many of the never smoker population has EGFR mutations or other kinase mutations, and that's how this drug works. This drug works by denaturing the kinase receptor. It's one of the unique drugs that actually has a pan-kinase capability. It modulates multiple kinase receptors and denatures them to slow down the growth. And then once it's inside the cancer cell, it resets the thyodoxone glutaphyone cycles and makes the drug -- cancer cell sensitive to chemotherapy drugs.

Why Japan, Taiwan? The incidence rate is there. We know that in Asia, the pharma companies are looking for this kind because the population has it as a higher percentage and Japan, in particular, and also Taiwan have studied this. And so they have the patient population. They studied it. They have a need for a drug.

So doing that in their backyard is a very much letting them know that we're open to partner the asset out. And we've actually just completed enrollment in Japan. So that's also -- enrollment happened pretty quickly ahead of schedule. And now we're focused on in Taiwan and also continuing enrollment in the U.S. But yes, very good question. Thank you for that.

And our next question, could you share your latest thinking around combination therapy opportunities, particularly in triple-negative breast cancer and non-small cell lung cancer?

Yes. Very, very good question. Yes. So both of those, we have new INDs that were approved. These are for Phase Ib, Phase II trials. I think they can -- we can hopefully even get to accelerated or even fast track because we have fast track now for TNBC already. The way our drug works for 184 is that it causes really high amounts of double-stranded DNA breaks inside the cancer cell.

That's important because once this drug gets inside the cancer cell and PTGR1 is there to activate it, it breaks apart the cancer cell by causing breaks in its DNA. But like most cells, it can try to repair it. And so what we thought about early on is what if we could find a mechanism that we could stop that repair from happening? Well, that drug exists. Those are PARP inhibitors. These are drugs that inhibit the PARP enzyme and PARP inhibitors are about, I don't know, about $1.8 billion to $3-plus billion drug class out there. There are a number of PARP inhibitors already in the market. And so our thought was we cause the DNA to break.

On the other side, totally synergistic is that we're giving it with a drug that blocks its repair. So it's kind of a one-two punch. And again, this is theoretical, just very much theory, but then we put it in the lab, and it was really, really brilliant done work by our team. And the early work that we saw is that we reduced the amount of our drug, we reduced the amount of PARP and the response is even better, nearly 100% tumor growth inhibition, meaning we destroyed tumors at 100%.

So like our drug alone, say, it was 75% or 60% or it's dose dependent. And just the PARP alone was a certain percentage, but we can reduce the amount of both drugs, therefore, actually actually having a better safety profile, which is really important, especially with PARP inhibitors because PARP start getting tolerance issues more than safety issues. And so we can reduce the amount, and we actually have a better overall impact, again, theory. And so now we modeled that using lots of different models, both models for binding, using the data we got.

And we said, can we do a trial where in triple-negative breast cancer, we use the PARP with our drug and see this kind of exquisite response. And so far, what we've seen in all the preclinical studies and all the models that we've done in multiple sites, multiple labs, it's very, very promising. And they're very synergistic. Like I said, one is completely destroying the DNA and the other is stopping it from repairing. So it's kind of like 2 bookends. We're like attacking the problem. And that's what combination therapy is supposed to be. Can we attack this problem from multiple places so that it's -- you're getting the maximum impact.

And as we see in a lot of cancers, you want to get as much of that impact upfront because that will give us a much longer, more durable response. Very similar in non-small cell lung cancer that has what's called KEAP1 or STK11. These are mutations that are really, really bad. Having an STK11 mutation or a KEAP1, these are very aggressive cancers. And in lung cancer, particularly, they tend to come back. They tend to be multidrug resistant.

And we saw is that even PD-1 drugs, which have an impact, they don't really have a durable impact. And the only current method that's used is to give a combination of checkpoint inhibitors, nivolumab and ipilimumab, 2 of them. And they actually tend to make the patients survive about a year, sometimes less. And so -- but that's not great.

And so our thought was how can we synergize with PD-1 drugs, these checkpoint inhibitors because the mechanisms of our drug and PD-1 are so wildly different. What we found is that we actually -- when our drug is given with PD-1 drug, it actually makes the PD-1 much more durable, and we drive cancer death to -- because of our drug because the cancer cell tends to have more upregulated PTGR1.

So again, we try to find those signals. Again, this is all data. It's all data that's available. We dissect that data. We model it. We then put it in the lab. We bring it back on the exquisite combination we found due to some of the KOLs that we work with, not just our own because we're a very collaborative company.

But some of our collaborators said, yes, this STK11 and KEAP1 is really important, and we think it makes sense because you're going to reenergize the tumor to be a hot tumor. We think we'll get a better, more meaningful response. And we think you're introducing a new mechanism, which also will be favorable. And that's how that STK11 KEAP1 mutant lung cancers trial is organized. So again, it's the power of data and the power of modeling.

And in all these combinations, we try to create unique models did it for DNA damage repair agents. We did it with checkpoint inhibitors. These are models that we've built. These are models that we've enriched with our own data. These are then models that we continue to enrich by taking preclinical lab results and even patient results that we can get and putting it back into the model to learn from. And so that's the power of the future of AI and data-driven drug development. It's not just take it off the shelf, have a prediction, go run after it. It's constantly enriching and learning from it.

And so those 2 trials, that's how those combinations were created. Excellent, excellent question. And again, very meaningful categories. Again, we think, hopefully, we get some good early responses and can move toward some accelerated pathways to get these drugs and combinations to patients that need them.

Excellent. We had another question. I assume that the LP-184 study of 65 patients would have 20 patients at the highest last dose escalation. What is the length of follow-up planned? And when will results be made public?

We don't have a time line public, probably sometime in the next few months as we dissect, we still have patients that are on the trial. And so that's great. That's good news actually. So -- but no, it's not 20 or so patients in the last cohort. Each cohort is between 3 and 6 patients. Our maximum -- as we said, our maximum dose level achieved was in cohort 12, dose level 12. And our recommended Phase II dose will probably be dose level 10, which is 0.39 mg per kg -- milligrams per kilogram. We're not going to have 20 patients in dose level 12. Dose level 12, we backed off on and we're going to go dose level 10. That's going to be our recommendation, and we'll see how the Drug Safety Review committee hopefully agrees to that recommendation.

But 10, we're very comfortable because if we get good responses at 10, we can go to 11. And we know 12 that we're getting into pushing the boundaries of what's going to be tolerated. But 10 is great because you can go up to 11, but you can also back down to 9 and that kind of 9 to 11 range, we still have enough drug substance, we think, to be biologically active.

We actually have another LP-184 question. Which tumor types are you most focused on first? And how does your PTGR1 biomarker help pick the patients most likely to benefit?

That's exactly how it helps. So ask that question. Great question. So if you have high levels of PTGR1, what we've mapped is a threshold we published that if you have a 4.2x or 4.5x, I believe, fold log higher of that enzyme in the tumor, there's a very high chance you're going to be hyperresponsive. We may use that eventually as a companion diagnostic.

But the 2 signatures that we're going to look for is, one, DNA damage response mutations. And there are about 13 to 15 genes that remember that are in that DNA response signature. And if you have a mutation or aberration in either the nucleotide excision repair pathway or the homologous repair pathways, you're going to be a great candidate for this drug. And TNBC, GBM, bladder and STK11, KEAP1, which are part of those global repair pathway genes are excellent candidates for LP-184.

We do have a bladder cancer trial that we'll talk about that we believe will be an IST, which means it will be investigator-sponsored and will be paid for by a different research institution. So that will be very exciting. In those bladder cancers that we intend on studying with LP-184, about almost 40% of those bladder cancers have what's called DNA damage repair mutations. It's a very high percentage. In lung cancer, it's about 7%. So every cancer has its own percentages.

We're going to go after those in which there's really, we believe is to be a clear path towards having a meaningful impact and getting to a white space, meaning there's either no drug approved in that line of therapy or there is a need for one or both. And that's how we think about indications. Like where can we -- where is the earliest possible place where we can get this drug to an approval? Excellent question.

Great. Thank you, Panna. And with that, I mean, I'll just pass it back to you real quick for do you have any final comments you would like to leave for the audience?

No, great questions from your audience. So this is great. Again, we've got a lot of milestones, a lot of data that we're going to be releasing over the next several months. We've got a good runway in front of us, and we also have the AI platform that we'll be making more and more publicly available. So I hope you guys have learned a little bit from today's webinar.

I hope you guys participate in the upside and see how AI can be used for good, not just making pretty pictures and writing research papers for people, but AI and these frameworks can actually be used to totally transform the development of cancer therapies. So thank you for taking time out on your afternoon to listen to our webinar.

Very exciting. Thank you, Panna. And for our audience, for more information on Lantern Pharma, you can always call us here at RedChip. That's 1-800-REDCHIP or you can e-mail us at LTRN that's the ticker symbol [email protected]. You can also visit ltrninfo.com where you can download the investor presentation, fact sheet and even sign up for news alerts on Lantern. With that, I want to thank everyone for joining us today. Thank you again, Panna.

Thank you, everybody.