Could a cure for your disease already be sitting on a pharmacy shelf? In this episode, we explore the revolutionary work of Every Cure, a nonprofit on a mission to unlock the hidden potential of existing drugs through a groundbreaking approach to drug repurposing. We sit down with co-founder Dr. David Fajgenbaum, a physician-scientist who saved his own life by discovering a new use for an old drug after being diagnosed with the rare Castleman disease. His journey sparked a global movement to find treatments for the millions of patients suffering from diseases with no approved therapies.
This episode dives deep into the systemic gaps in the pharmaceutical industry and how Every Cure is using cutting-edge technology to fill them. Dr. David Fajgenbaum shares his incredible personal story, from near-fatal illness to pioneering a new field of medical research. We discuss how his team is leveraging AI for drug discovery and rediscovery, building an AI platform that systematically matches thousands of approved drugs against thousands of diseases to find promising new treatments. This data-driven approach, which Dr. Fajgenbaum calls "computational pharmacophenomics," is already yielding incredible results for both common and rare disease treatments.
We explore the nonprofit's unique model, fueled by major funding from partners like ARPA-H and the TED Audacious Project, which allows them to pursue promising treatments that lack financial incentives for traditional development. You'll hear the stunning success stories of this drug repurposing strategy in action—from Leucovorin, a vitamin that is helping nonverbal children with autism speak their first words, to lidocaine, a common numbing agent showing a 29% reduction in mortality for breast cancer patients in a large clinical trial. This conversation reveals how technologies like biomedical knowledge graphs and large language models are accelerating the search for cures, turning a 100-day analysis into a 17-hour process and bringing hope to patients worldwide.
Introduction
Dr. Trevor Royce: Hey, Paul. How are you?
Dr. Paul Girard: Fine. How are you?
Dr. Trevor Royce: Great holiday weekend. I've got four small kids, and I feel I spent half the year recovering from Christmas and the second half of the year recovering from Easter.
Dr. Paul Girard: I'm in the same boat with you. Recovery Monday.
Dr. Trevor Royce: Sounds about right. We've got a full holiday, Patriots' Day.
Dr. Paul Girard: That's right. And I live up here in New England, so a Boston Marathon happening just a little south of me.
Dr. Trevor Royce: Amazing. A great slice of Americana. Should we dive in?
Dr. Paul Girard: Let's pontificate.
Dr. Trevor Royce: Let's do it. Welcome to HealthTech Remedy, the show where three physician leaders in health technology tell the stories of new and established companies and interview leaders from the industry. I'm Trevor Royce, radiation oncologist and researcher with experience in real-world evidence, informatics, and AI diagnostics.
Dr. Paul Girard: I'm Paul Girard. I started off as a physical medicine and rehabilitation physician before focusing on reimbursement policy, molecular diagnostics, and market access for AI products.
Dr. Trevor Royce: And missing today, the third leg of our stool, Timothy Showalter. He's a radiation oncologist and medical device entrepreneur who's focused on bringing AI advances to cancer patients. This is going to be a great episode, Paul. We don't have Tim today, which means we can be brutal to him.
Introducing EveryCure and the Mission to Repurpose Drugs
And we'll have a lot of fun discussing about our next company, which is EveryCure. They are a nonprofit whose mission is to unlock the full potential of existing medicines to treat diseases in patients.
Dr. Paul Girard: And then we'll speak with Dr. David Fajgenbaum, co-founder of EveryCure.
Dr. Trevor Royce: Yeah, super excited about today's episode and our interview with Dr. David Fajgenbaum later. This journey that he's had and this nonprofit he started, EveryCure, has just been incredible to watch. Just taking off like a rocket ship, incredible impact. And it has been inspiring to so many, from the patients to the providers, and just really captured the nation's imagination in terms of what can be done in the modern world of healthcare.
EveryCure has been profiled on Good Morning America, USA Today, Wall Street Journal, New York Times. They've all had large profiles of what they're doing. They were recognized by getting this audacious project grant from TED, the lecture series organization. And Dr. Fajgenbaum himself was actually awarded as Philadelphia Citizen of the Year in 2024 for this work.
You can read in detail about his work with his book that has made some headlines called Chasing My Cure, where he describes his own journey with a rare disease and the pursuit of treatment for that disease. And so we'll hear more about that, of course, from David later. Taking a step back, to summarize, in 2010, Dr. Fajgenbaum was diagnosed with Castleman disease. He had several rushes with near death and then discovered a new use for an existing drug. It had already been approved for another disease. Again, we'll hear more from David about all this. It had never been used for his rare disease that he was diagnosed with Castleman disease.
And now he's been in remission for 11 years, gone on to have this incredible career as a physician, researcher, and entrepreneur, and built EveryCure, which is dedicated to scaling this approach, finding treatments that exist for conditions that may have been underappreciated, that this drug could be used for that, and trying to benefit patients with rare diseases from repurposing existing drugs. So as I mentioned, EveryCure is a nonprofit. They're on this mission to unlock the full potential of existing medications to treat diseases in patients.
And just a little bit of background, the National Institutes of Health define the rare disease as those which affect fewer than 200,000 people in the U.S. But there are thousands of rare diseases, which altogether affect tens of millions of Americans and hundreds of millions of people around the world. And now after the big profiles in the Wall Street Journal and the New York Times, I can only imagine what Dr. Fajgenbaum's email inbox looks like with millions, hundreds of millions of people wanting how their disease can be cured.
More than 90% of these rare diseases have no approved treatments. And the financial incentives for the big innovators in drug development like big pharma, it's hard for them to make the business case to commit resources for these rare diseases, which may in isolation have a limited impact, but it broadly affects many people when you pull all the rare diseases together. And so it's going to be incredible to talk to David later to think about how he's going to solve for this and take these approaches or techniques to scale so that you can impact many patients with many of these different, heterogeneous diseases. What do you think, Paul?
Dr. Paul Girard: The fact that what he's doing has such a need and he's willing to give time to come on this podcast, I think makes us lucky. I think this is a really interesting concept. You think about a lot of drug development and finding indications for drugs comes from pharmaceutical companies. And as you were just referencing, it takes a lot of time and money that pharmaceutical companies have to devote to this.
There is, of course, research being done by independently funded research that looks for use of drugs outside of drug studies funded by pharma. In general, just finding an indication for a drug, showing the drug works for that indication, takes years. No matter who's funding it, whether it's pharma, whether it's the NIH, whether it's some other body, it's money that could go to something else. And so you want to make sure that when you start to think down the lines of, “Hey, this drug could be effective for something,” you're going to be right. And sometimes people aren't.
And so, it's just to me, the concept of how can we match drugs that we already know are absorbed in people, that we already know have a side effect profile that is tolerable in people? How can we match those with treatments or with diseases that don't have a treatment? I think that's really exciting. And I guess a big question is, in the long run, how is this going to be used? Is it going to be used to just handle rare diseases or will this be used even more broadly to say, hey, we have a treatment and we have a disease that's either rare or not so rare. Could we start to match this existing treatment with this disease and might it give a benefit that we didn't know about in the past, that we didn't think of and that we're not getting from our existing on-label therapies? So to me, I think that's really exciting. It's obviously rare diseases are the first place where something like this makes sense. But I think that the methodology and the approach could be applied much more broadly.
Dr. Trevor Royce: Yeah, absolutely. And I think an important piece of this is how it highlights just how complicated pathogenesis is and the human condition and all the diseases. We often, even in things like prostate cancer, use this label of prostate cancer. But when you really get under the hood, no two prostate cancers are really the same when you get down to the genetic level. And if there's a way to solve for all these different diseases and all the unique therapies that are needed for them, in particular, by using drugs that already exist, I think it's a pretty amazing pursuit.
Dr. Paul Girard: It's rare disease meets precision medicine.
Dr. Trevor Royce: Exactly right. That's exactly right. I think that's very well put. It's interesting when you think about this idea of using existing drugs for a new purpose or for a condition that hadn't been used before. It's been around for a long time. And the Wall Street Journal highlighted this, the FDA approves the drug. Doctors can prescribe it off-label to patients with other conditions if they think it'll help. And this happens all the time in the delivery of care. And the example used by the Wall Street Journal was that Ozempic was originally approved for people with type 2 diabetes. And now, of course, it's used by millions of people without the disease for weight loss.
And when I was reading about EveryCure, what came to mind was the 1990s movie, Lorenzo's Oil. I'm not sure if you're familiar with that. I'm sure Tim would know it's got Nick Nolte. I think Susan Sarandon in it, maybe 1992. And anyways, it's a story where parents have a child with a rare condition and they discover an oil that's commercially available that can be used to help their child with this rare disease live longer. And it's based on a true story and an amazing success story.
So I feel Dr. David Fajgenbaum is taking this Lorenzo's oil. It happened to him personally. So it's much bigger than that. As a spectator, just watching the movie, which I enjoyed, and then applying it at scale to help all these patients. And it'll be cool to hear the details there.
EveryCure's Nonprofit Model: Tackling the Financial Gaps in Drug Development
So in terms of EveryCure and their structure, I think it's interesting because they're a nonprofit and we haven't really spent much time talking about that. And Paul would really value your thoughts on some of this. But basically, I'll give you a summary. They secured over $110 million in committed funding through philanthropic support through organizations and so forth since their launch in September of 2022. And these include two big keynote funders. One, as I mentioned, the TED Audacious Project, which was a five-year, $60 million commitment for them to pursue their mission. And then two, ARPA-H was a $48 million three-year contract. And ARPA-H is interesting. It's often positioned as a healthcare version of DARPA, which was the big audacious project for the defense side of things for the U.S. government. And DARPA is what led to the creation of the internet. And so this ARPA-H has committed this great contract to EveryCure, and it'll be amazing to see how they get that support.
Another interesting piece is they have already made such an impact in different diseases and had such publicity accompany this. I think it's noteworthy. We haven't spent a lot of time on our show highlighting this, but their board of directors has some really impressive people on there. So I just wanted to kind of give that a shout out because it does show the gravitas that this company's mission encapsulates, including Janet Woodcock, who served as the acting commissioner of the FDA, was at the Center for Drug Evaluation and Research, or CDER, for many years. And so, this really driving force and health on the regulatory side in the U.S. and the FDA and will be an amazing contributor to EveryCure's ability to execute their mission.
Powering the Platform with AI and Strategic Partnerships
Dr. Paul Girard: It's impressive all around. In terms of the approach they take, what they try to do is systematically identify how drugs and diseases can match each other. And their platform has about 4,000 drugs that they try to match against over 18,000 diseases. And they try to figure out how the biological effects of the drug might impact the diseases and might be used to treat the diseases using a ranking system that uses a wide variety of evidence sources like studies, case reports, clinical trial data, health records. And interestingly, they leverage Atropos Health's GENEVA OS™ research. From another company that we recently had on the podcast. So it'll be interesting to hear about, we heard about the potential technology on a prior podcast. It'll be interesting to hear about its application in this company to solve a real world problem. And it sounds like most of their revenue comes from grants and contributions since they are a nonprofit.
Dr. Trevor Royce: Accompanying this big fundraise, some high profile press, they've had some pretty impressive collaborations get stood up with partners. And this is another thing that we should hear from David is how they work with partners and tech industry, the pharma industry, nonprofits, researchers, and so forth, how they can put all this together. But some examples are a collaboration with Google Cloud. Again, the backbone of all this drug repurposing is using AI to comb through and identify patterns and the incredible amount of medical research data out there. And so they intend to use Google Cloud's AI tools, most famously [Could not verify with context] or large language models, to power what they call their matrix platform. And that is the foundation of what will be used to accelerate their discovery, their validation, and dissemination of these new treatments that they identify.
So in terms of their progress to date, they have advanced over a dozen repurposed treatments for diseases that weren't initially intended for. Obviously, we've heard, and most famously, about Castleman's disease, which Dr. Fajgenbaum went through himself. POEMS syndrome, angiosarcoma were some of the success stories and case studies that I'd read about when we were preparing for this. And I'm sure David will share some additional examples from us. Any final words or thoughts from you, Paul, on EveryCure before we jump over to David?
Dr. Paul Girard: No, I'm just excited to hear from him and learn about what he's doing and what the future looks like for him and EveryCure.
Dr. Trevor Royce: There you have it. Well, that's an overview of EveryCure. Let's hear from the man himself. We'll go over to Dr. David Fajgenbaum for the interview here shortly. Welcome to the show, David.
Dr. David Fajgenbaum: Thanks so much for having me, Trevor. Excited for this.
Dr. Trevor Royce: Great. Well, we're going to make good use of our time. We only have about 30 minutes together. Why don't we start from the top and we'll just turn it over to you, David. You've got an incredible story. Why don't you tell us a little bit about your background, the story of EveryCure, and your own journey?
Dr. David Fajgenbaum’s Journey: From Patient to Pioneer
Dr. David Fajgenbaum: I grew up in Raleigh, North Carolina, and ended up going to college at Georgetown. And my dream was to always play college football. And I got to Georgetown to play college football. And then my life just got really turned upside down and really shattered when my mom was diagnosed with brain cancer a couple of weeks after I got to college. And that just fundamentally changed how I wanted to focus my life. And I immediately said, I want to go into medicine. I want to take care of patients like my mom. I want to take on horrible diseases like cancer. She passed away about a year after her diagnosis. And that just lit a fire in me to say, I want to dedicate my life to finding treatments for horrible diseases.
Fast forward to my third year of med school. I was well on my way to being that doctor that I wanted to be. And then I personally got very ill with a disease called Castleman's and spent many months in the ICU, had my last rights read to me, nearly died on five different occasions. But thankfully, chemotherapy, a number of different treatments were able to save my life. But that then refocused me into say, OK, if I want to take on diseases and treat patients like I'd hoped to, I got to save my own life first.
Went down this journey to try to find a drug that could maybe save my life and did experiments of my own blood samples and analyzed a bunch of data and ended up discovering that a drug that's been around for decades called sirolimus, it was made for organ transplantation. I thought that it might actually be able to treat my disease, Castleman's. And I began testing it on myself, and it's now been over 11 years. I'm doing great on this drug. And this experience for me was so eye-opening. Wait a minute, this drug wasn't made for my disease. And somehow it was just sitting there at the pharmacy shelf and it could treat me and it has and it saved my life. And so I've been on a mission these last 11 years to try to figure out how many more drugs are already approved and made for one condition, maybe at your local pharmacy, that could actually be repurposed and save lives today.
Dr. Trevor Royce: It's incredible. An incredible story. Were you pretty much going down the oncology pathway before this happened? Did you see yourself as a cancer doctor in the future because of your mom?
Dr. David Fajgenbaum: That was the plan. Oncology doctor, I saw myself as a, providing patient care, but also doing research. And I dreamed of one day being a part of maybe clinical trials or laboratory work that led to a drug for a cancer. And that would be the pinnacle, the dream. Never imagined that I would go down this path of looking to repurpose drugs, drugs that are on the one yard line that maybe we could get to patients right away.
Defining Drug Repurposing: Unlocking the Potential of Existing Medicines
Dr. Trevor Royce: Tell us a little bit about this concept of repurposed drugs. How would you define that or characterize that?
Dr. David Fajgenbaum: Repurposing is the term to describe when you take a drug that's approved for one disease and you try it in a new disease. And it's not a new concept. I'm sure all the listeners have heard about Viagra and how it was repurposed from heart disease to its very well-known use. But most people may not know that it's also been repurposed for a rare pediatric lung disease. And so kids with pulmonary child hypertension were dying before their teenage years, and now they can live full lives on Viagra.
And of course, there's many other examples like thalidomide, which is now an amazing drug for the cancer multiple myeloma, and also it's approved for leprosy, just shows that though diseases may appear different, like leprosy and myeloma, they can oftentimes share the same underlying problem and therefore can be treated with the same drug. And then of course, every drug or many drugs, particularly small molecules, can have more than one mechanism in action. So that creates the perfect ingredients for drugs to be able to be used in many diseases.
But the system we're in is such that drug companies develop a drug for a disease and maybe one or two more diseases. And they oftentimes consider dozens of possibilities, but they have to pick the one or two or three diseases to pursue, which means that us humans don't get a benefit from all the amazing drugs that are developed and manufactured and are available at the pharmacy. And this just became so clear to me because I was dying from this disease, Castleman disease, and I didn't have more than a few months to live. And I had to find a drug to save my life. And so I didn't have a billion dollars to make a new drug and I didn't have a decade to wait for that drug to be developed.
So it was just this very simple concept of, well, I guess the only way I can survive is if I find something that's already out there. That could treat me. And of course, I'd been getting chemotherapies before that that also weren't made for my disease. And that was part of the realization for me. I was like, wait a minute, Adriamycin and cyclophosphamide, they weren't made for Castleman's. They're saving my life. It's not permanent, but they're doing something. So maybe there's something else out there. And yeah, it turns out there's tremendous opportunities to save lives with the drugs we already have.
Dr. Trevor Royce: It's incredible.
Building EveryCure: Why a Nonprofit is Essential for an All-vs-All Approach
Dr. Paul Girard: A lot of our listeners have the itch to do something beyond just routine clinical care, whether it's entrepreneurship or it's starting something for public health. I guess it'd be interesting if you could tell us a little bit about the structure of EveryCure. You talked about the vision, but how did you get this off the ground? It's a nonprofit, but clearly what you're doing requires some resources. How do you get revenue? How do you keep the lights on? And how did you really make this dream a reality?
Dr. David Fajgenbaum: Yeah, it's great questions. And maybe I'll even go back before then and talk about what we were doing the last 10 years, which then led into EveryCure. And that's that I joined the faculty at Penn. And here in my center, we focus on trying to find new uses for existing medicines within rare inflammatory diseases and cancers. And over these years, we've helped to advance 13 more repurposed drugs for diseases they weren't intended for, like POEMS syndrome and DADA2 and Castleman's, of course, angiosarcoma.
And so we had been building the processes and the ways to take a drug that's approved for one thing and find a new use for it, get it into treatment guidelines, educate doctors about it, and really change clinical practice without actually ever changing the label. So not actually getting the FDA to change the label, but getting that drug to be used and help patients. And so we've been doing that for the last decade. And about three years ago, we decided we just had this responsibility that there's this massive gap in our system where the drugs that are already approved, there's no one that's taking responsibility for making sure they're used for all the diseases possible.
And simultaneously, there's no financial incentive to do that. And then also simultaneously, there was obviously incredible progress utilizing artificial intelligence that could make it become a theoretically feasible thing to do to match every drug versus every disease. And so that's what we set out to do is to build an AI platform to look across all FDA approved drugs and all diseases and then to identify the most promising opportunities.
So we thought, well, what does this need to be? Do we structure it as a for-profit or a non-profit? What do we need to build here? And we spent a lot of time thinking about it, and we felt that the greatest opportunity for impact and filling this gap would be through a non-profit organization. We wanted to have the flexibility to go after the non-profitable opportunities, because drug companies are actually pretty good, and biotechs and entrepreneurs are pretty good at finding, hey, I think this drug could be used in a new way, and maybe if I change the dose or the formula, maybe I can figure out this new use. But the real gap is when there's no dose change, there's no form of change, and there is no financial upside, there is no profit to be made, which actually is the vast majority of opportunities, because 80% of drugs are already generic.
And so the drugs that we know well, that are relatively safe, that are already manufactured, on the one yard line, those are the very drugs that none of us work on, no one in medicine works on. And so it felt like, okay, we got to have some nonprofit that doesn't need to be profitable, that goes after the nonprofit opportunities. So we structured it as a nonprofit, we built out this AI platform. And to do that, we had to raise a significant amount of funding. So we got a almost $50 million contract from the federal government from an agency called ARPA-H, and then more recently announced about $60 million in philanthropic funding from philanthropists who want to fix the system and recognize that these drugs are falling through the cracks.
And early on, we realized that fundraising would be really tough for what we're doing because everyone has a disease that they care about. I have a disease I care about. It's Castleman's and ALS and a few others. We all have a disease that we care about. So every time we talk to a philanthropist about funding our work, they'd say, love what you're doing. We want you to work on this disease. We'll give you $10 million if we work on that disease. And we actually had to turn down tens and tens of millions of dollars saying, well, we would love to find a drug for that disease. And I actually really hope we do.
But the innovation here and the gap we have here is that no one is looking across the whole forest to find the lowest hanging fruit across the whole forest and then moving those things forward. And I really hope that there's a drug for your condition or the condition you care about that is low hanging fruit, but someone's got to start looking for the low hanging fruit. And so we turned down a lot of money and we didn't raise very much money early on, but we eventually were able to connect with philanthropists who realized that the system's got to be fixed. And we were able to raise the funds to get started.
And of course, you both know clinical trials are expensive. A single clinical trial could cost $50 million or $100 million. We're not going to do trials that are that expensive. But to prove these drugs work in a rigorous fashion, which is what we have to do, requires a lot of funding. And we believe that the impact we've made thus far and the impact that we're going to make over the next few years is going to lead to other individuals who are going to say, I want to support this kind of work. No one's going to make any money, but I want to help people.
Success Stories: From Leucovorin for Autism to Lidocaine for Breast Cancer
Dr. Paul Girard: That's a really interesting model that you were talking about there with the, you're looking to repurpose drugs across indications and looking for the lowest hanging fruit. And I feel that's different than how we typically hear about things, say, NIH funding and various ways that people prioritize research. It's not, hey, I'm going to look across diseases at the lowest hanging fruit. It's rather, I'm going to look at some very specific indication and see if I can make breakthroughs and discoveries there. Hearing this model that I think almost doesn't exist outside of EveryCure, are there examples that you can talk about of where you've had an impact? And I'm going to say, if you've had an impact, that implies to me that the more conventional model hasn't, or at least not yet.
Dr. David Fajgenbaum: Yeah, it's the right question to ask. So I'm not aware of any other groups that are taking this all-versus-all approach. We actually had to come up with a term for it. So drug repurposing has been around forever, but it's always been in two directions. So one is you've got a disease you care about, like Castleman's, and you're trying to find a drug for it. That's traditional drug repurposing. The other is you've got a drug. You're a drug company. You want to find an indication for it, indication expansion. But what we're doing, we call computational pharmacophenomics, which is a mouthful. We probably could have done a better job with our marketing.
But that's where instead of starting with a single drug or single disease, you start with all drugs and all diseases. And you quantify everything versus everything. And then you look for the low-hanging fruit across the spectrum. And so, yeah, we've taken this approach. we started building out the team really in earnest about a year ago and then started running our full pipeline about six months ago. From running this full pipeline from our tech team to our med team, we've come across treatments that have been, look so promising. So we categorize things into three categories of repurposing. We have drugs that look promising from our AI platform, but there's a lot more lab work to be done to really understand the mechanism. We have some that the mechanism looks really solid, but now we need to actually study in clinical trials or at least in clinical practice. And then last, the trials have been definitive. The drug just needs to be given to people. And we call those unsung heroes, but they're not being utilized. And it's typically because there's some sort of financial challenge.
And so I can go through all three. In terms of this unsung hero, which is this last category, we identified leucovorin for children with autism who have speech delays as being this really promising opportunity. Turns out a lot of kids with autism have antibodies that prevent folate from getting into their brain. They have a cerebral folate deficiency. But if you give them leucovorin, it basically bypasses through this reduced folate channel. And these kids get folate in their brain. There've been three randomized controlled trials, the most recent one published this fall in India, that demonstrated improvement in verbal communication for these kids, but no utilization whatsoever of leucovorin. It costs about 17 cents a pill to make leucovorin. It's been around for decades. No one will make money off of leucovorin for kids with autism, but a lot of kids will be able to speak.
I've gotten a chance to meet some parents of kids that have gotten leucovorin. And I learned from this one family that their son, Mason, hadn't spoken. He was three years old. He hadn't spoken before. And he started taking leucovorin. And within three days, he said his first word. And his first word was more. He wanted more. And as a parent, we've got two kids. And I think about how oftentimes they want more from me, more food, more something. And then being able to tell me more is actually one of the most important things they can say. So that really hit home. And actually, another family of a kid named Ryan, he was nonverbal for five years. He could say occasional words, but they were always repeating what someone said back to him. But then within a couple of weeks, the start of leucovorin began to speak and actually could share coherent sentences and feelings because he'd had a folate division. He'd not gotten folate into his brain in the five years he was alive. So that's an unsung hero. So the next step for something like that is actually to raise awareness and advocate for there's a test that can be done to test whether kids have antibodies against the folate receptor and to educate about the potential for this in this subgroup of kids with autism.
Dr. Trevor Royce: And how do you bring that all the way to the patient then?
Dr. David Fajgenbaum: Yeah. So we ran a story on CBS Evening News a couple of months ago that led to a massive uptick in testing for the autoantibody test. It's led to, we presume, increase in prescriptions. That's one of the levers that we can pull, but we think about engaging with physicians. Here we are right now. There's probably a physician, there might be a pediatrician on the line, are listening in that's going to hear about this and say, maybe I'll learn about cerebral folate deficiency. And so it's going to be a bit of an unusual process because no nonprofit drug company has ever existed. We don't own the drug.
It's an interesting model. So the other end of the spectrum is what we call Frontier Explorer. That's where the platform has identified something promising, but more work is needed. And I'll share a couple of examples. One of them is that we had a patient who was not responding to any treatments for Castleman disease. And so we turned to the algorithm and looked for the number one predicted drug for Castleman's. And the number one drug that came up was adalimumab, a TNF inhibitor. It had never been used before for Castleman's, but mechanistically made some sense. We did some laboratory work. We validated both there's a TNF signature in Castleman's patients, but also increased T-cell production of TNF. And so we gave this drug to a patient. He responded really well. We published this paper in the New England Journal of Medicine. So this is the earliest side where AI is pointing us to something that we humans hadn't seen before.
And I have to say, when we started our platform or we started building this out, I thought we would find a lot more of those or that we'd be mostly in the realm of AI giving us new predictions. But actually what's been surprising is AI is just helping us find a lot of things that us humans have already found, but us humans have just discarded because it's not profitable. But so that's an example where it's a true de novo discovery. We validated in the lab. So that's what we call frontier explorer. We have to do lab work afterwards.
In the middle category, we call clinical gems. And that's where there's some really promising mechanistic data, but we need more clinical data. And an example of that is with lidocaine and breast cancer. There has been a large randomized control trial of 1,600 patients in India that were randomly assigned to get peritumorally injected. So basically lidocaine bathing the tumor in half the patients and then randomly did not get any sort of lidocaine in the other half. And the patients who had a peritumoral injection of lidocaine had a 29% reduction in mortality of five years versus those patients who didn't. Published in the journal Clinical Oncology, one of the best oncology journals in the world. But there's really been next to no uptake of lidocaine injection for breast cancer.
And that's in part because the mechanism is still not fully understood. And so we're doing some laboratory work there. But it also is in part because a lot of medical oncologists want to do another randomized control trial, follow another group of patients for five or more years. I will say, when I talk to surgical oncologists and I tell them and I share the data with them, they usually just say, I'm going to go inject some lidocaine into the next patient's tumor that I resect. And so this is an interesting debate. Where would you categorize this? There's been a randomized controlled trial, 600 patients, survival benefit with lidocaine, which of course is going to be used in the middle of the procedure anyway. It's going to be used at the site of the incision should we do another trial, and we likely will. But in some cases, as I said, doctors are going right to using it. So these are the kinds of things we're finding, which I have to admit, I never imagined that a vitamin, folinic acid, might be something that could help the kids with autism and ever imagined a numbing medicine might be able to treat a cancer. These are things that we never could have dreamed of, but I think that's the power of taking this completely disease and drug agnostic approach is saying, I don't know what the disease is or what the drug is going to be, but I just want to look across everything versus everything and let the data tell me what looks promising. And then, and then we can take it forward.
Dr. Trevor Royce: Yeah. Related to that, it sounds like this work impacts well beyond rare diseases. Some of the diseases you mentioned are incredibly common with high incidence and massive impacts. And here you're just scratching the surface and discovering new connections and therapies that are already widespread.
Dr. David Fajgenbaum: That's right. And that was the other thing I wasn't expecting. I was convinced that we were going to only find things for very rare diseases because I figured the math of it was that, an inexpensive drug for a rare disease, the math doesn't work. So therefore, we're going to have to go after those rare diseases. But I think that both of the conditions I mentioned, breast cancer and autism, obviously, are not rare diseases, but they're very inexpensive treatments. And so I think for the very, very inexpensive treatments, there's even a market gap there, even for the very common diseases.
Under the Hood: The AI Technology Driving Drug Rediscovery
Dr. Trevor Royce: Yeah, that makes that makes a lot of sense. And you can tell us a little bit what's under the hood on the tech side and how this error of AI has really let you guys take off.
Dr. David Fajgenbaum: Sure. So there have been a few it feels like a few lifetimes worth of AI generations just in the last three years early on. and there's still some elements of it that's part of our foundational platform. Early on, we're using something called XGBoost, which is very similar to a random forest approach, where basically, I'll take a step back even further. We utilize what are called biomedical knowledge graphs, which are a two-dimensional representation of everything the world knows about every drug, every disease, every gene, every protein. Imagine you were to say, I want every biomedical concept, I'm going to put it on a wall. And then I'm going to put edges, which are connections between them. I'm going to define the connections. So IL-6 is increased in Castleman disease. Siltuximab inhibits IL-6. But imagine doing that across every single biomedical concept. There's millions of concepts. There's tens of millions of edges connecting them. So we build these knowledge graphs.
And the good news is, is the US government actually spent a lot of money to construct really high quality biomedical knowledge graphs over the last decade. They didn't know how they were going to be used, but a very forward-thinking leader of NCATS named Chris Austin decided that they should be built. So these knowledge graphs serve as really great ingredients for us to then apply machine learning algorithms. So I was mentioning early on that we utilize XGBoost, which is similar to Random Forest, on top of these knowledge graphs. And we basically train the algorithms on known treat relationships. So we know GLP-1 agonists treat diabetes. We know that siltuximab treats Castleman disease. Imagine training on every single drug for every disease, training all these relationships. And then what we do is we ask the algorithm to score every other relationship that isn't known.
So does a GLP-1 treat Parkinson's disease? Does siltuximab treat GVHD, whatever it might be, but give a score from zero to one. So if it's close to zero, it's unlikely to work. If it's close to one, that means it's likely to work because of the similarity between the pattern of connections to what we know works. So it's a lot of its mechanistic data. So, there's a shared mechanism. The drug inhibits this, and there's too much of that in that disease. So we use a lot of mechanistic data, but also clinical data gets integrated in that as well. So it's a really exciting way to use a lot of different data sets and use old school artificial intelligence, something like an XGBoost to get these scores.
Now, over the last couple of years, we've been supplementing that in a lot of ways. So using LLMs to do something we call embedding so that each of those nodes, instead of just being interleukin-6, it's interleukin-6 with all of the world's knowledge about interleukin-6, which of course, before it would have taken way too much time for humans to generate. But now with LLMs, we can actually improve the embeddings of what does that node mean? It's not just a word, but actually what's all the information about IL-6 that comes with it. And then where we're looking to in the future is using LLMs even more and creating even foundational models that are specific for drug repurposing.
So I mentioned that we've gone through multiple generations the last couple of years because I think everyone listening is aware of the incredible progress in AI. And I'll just give a quick snippet of it in terms of compute. When we ran the algorithm for the first time two years ago, it took us 100 days to score every drug versus every disease because it's 4,000 drugs and it's 18,000 diseases, so it's about 75 million scores. It took 100 days to do that. Our most recent run last month, it took 17 hours to do that same analysis. And what's amazing about a 17-hour run is that we can look at results and we can tweak and improve things, we can change things, we can integrate more data, then we can run the same analysis again the next day or the next week, and we can continue to learn and iterate. When you've got 100-day cycles to iterate, you can't really iterate too well. And we were doing a lot of waiting and hoping, and there's a lot of edge of our seats excitement, but this is just so exciting when we can iterate the way that we can now.
Dr. Trevor Royce: Yeah, I guess one quick follow-up question on that, and then Paul, pass it over to you. As a nonprofit, what's your approach to getting access to the compute you need for that type of turnaround or the GPUs and so forth? I assume you've got some pretty nice partnerships coming together that people want to help to this mission.
Dr. David Fajgenbaum: Yeah, well, we do have some nice partners. I will say that I was convinced when we first launched that we would never pay for any compute and that there are so many companies that do this that we would most certainly never pay a dollar for it. We haven't quite gotten to that level. We've certainly gotten support. We've gotten some compute and some cloud. But I'm hopeful that maybe one of those companies is listening to this podcast is maybe going to come in and give us the free compute that I think we need. And then I think that at the end of the day, no one's going to make any money off anything we're doing. and we're just going to help people. So, so, so like you, Trevor, I was optimistic. I'm hopeful that we'll get some more compute.
Dr. Paul Girard: That was really interesting. I think one of the most interesting things to hear about is ever since I was in medical school, we had people talking about AI helping in drug discovery and companies coming out with that idea. And it sounds like you're not quite at the point of discovering molecules, but you're certainly at the point of using AI to help find how to use existing drugs for purposes. So it's like the dream that we've been hearing about for, I don't know, 25 years is finally becoming a reality.
Dr. David Fajgenbaum: Yeah. I appreciate you saying that, Paul. And I totally agree. 99.999% of AI in medicine is around drug discovery, new molecules, because that's where all the money is. But we do drug rediscovery. We're trying to rediscover new uses for the medicines we already have. And there's no money in it. So that's why no one does it. But there's so much impact. These drugs are already at our CVS. We've got almost 4,000 approved compounds that are at your local pharmacy that could help you or someone you love and no one using AI to find those uses. And so we're doing it. And it's so exciting. And every patient we're able to reach, it just powers us to keep going.
The Human Impact and How You Can Help
Dr. Trevor Royce: Pretty powerful visual, the idea of thinking of all these things, just at your local pharmacy for whatever rare disease, or as you put it, the disease that's important to you might be able to help that. And kind of tongue in cheek, but earlier we were joking, I'm going to throw some numbers out, but, 300 million people globally with a disease with no FDA approved treatment, 18,000 recognized disease, but only 4,000 have approved treatments. And I cannot imagine what your inbox is like. There must be so many people reaching out to you with their disease of interest, whatever, either a loved one or for themselves, asking for you guys to work on this problem.
Dr. David Fajgenbaum: It is, and it's both heartbreaking and also very humbling and very, it really creates focus because I get hundreds, maybe even thousands of emails on a daily basis from people with horrible conditions that need treatments. And as I shared earlier, our model is that we use AI to find the best opportunities. And so we can't say, we're going to work on this disease or that disease or this drug or that drug. We've got to let the data drive us there. But it certainly highlights the power of our platform and the importance of what we're doing and the heartbreaking need that's out there. So it very much locks us in on the work we got to do.
Dr. Trevor Royce: Thanks so much, David. I don't think it's an understatement to say you've got hundreds of thousands of people pulling for you. And we certainly are as well. We look forward to seeing the journey continue. Really appreciate you taking the time to chat with us today.
Dr. David Fajgenbaum: Thanks so much, Trevor. So appreciate you. And for anyone listening that wants to support, you can go to everycure.org/ideas. If you have an idea for a repurposed drug that you think could help patients, maybe you've treated someone with it and it's helped them. Let us know everycure.org/ideas. If you want to help be an expert and help to review the next repurposing opportunity in your disease area, you can go to everycure.org/experts. And if you just want to support our work, you can go to everycure.org/donate and help us to keep driving this forward.
Dr. Trevor Royce: That's fantastic. Thank you so much, David.
Dr. David Fajgenbaum: Thanks so much, Trevor. Thanks, Paul.
