RARE Daily

Bringing Precision to the Treatment of Rare Cancers

December 28, 2023

Despite the prevalence of cancer, the vast majority of known cancers are rare and face the same type of treatment challenges as other rare diseases. David Hysong was diagnosed at the age of 27 with adenoid cystic carcinoma, a rare head and neck cancer, and that set him on the path to address the needs he saw in patients with these diseases. Hysong, founder and CEO of Shepherd Therapeutics, discusses his company’s use of AI to analyze individual patient’s tumor RNA, its efforts to match rare cancer patients to their best therapeutic options, and how it is using data captured from the transcriptome to develop new therapies for people with rare cancers.

 

 

Daniel Levine: David, thanks for joining us.

David Hysong: Thanks so much for having me. It’s a pleasure.

Daniel Levine: We’re going to talk about rare cancers, Shepherd Therapeutics, and its efforts to develop treatments for these conditions. Before we get into the specifics of what Shepherd’s doing though, I wanted to start with your own journey and how it led to the creation, both of Shepherd Therapeutics and the Shepherd Foundation. When you were 27, you were a graduate student at Harvard, you earned a master’s degree in Divinity. You were diagnosed with adenoid cystic carcinoma, a rare head and neck cancer. How did you come to get diagnosed?

David Hysong: It’s a great question, Daniel. It’s definitely a non-linear story and path to get here, but one that is all too familiar to every other patient in the country, right? None of us expect this to come. I was the exact same. I was at Harvard, as you said, doing a graduate degree. I’d ended up there after another near death experience. I had actually been hit by a bus on a motorcycle while I was working against human trafficking in Southeast Asia. Ended up at Harvard to rehab and had a package together to hopefully become an officer in the SEAL teams in the US Navy. So I’d gone out to officer selection for that in Coronado at their base out there. I got really sick during training, finished up, but was advised that I had dealt with something pretty serious. So came back and it turned out that I had had a rare head of neck cancer, as you said, the entire time—so came at the moment that I least expected it.

Daniel Levine: What were you told about the condition when you were diagnosed in terms of both treatments and prognosis?

David Hysong: Treatments—that there was nothing. I think that’s on my journey of education to do what I do now. Part of that was starting with the same level of information that I think most patients do, which is relatively minimal. Thought of chemotherapy as one type, one thing that they did, radiation as another, surgery—cut, burn, poison. So, diving in and realizing that there were more than a hundred different types of chemotherapy approved by the FDA, none of them worked in my disease; that there are close to 150 more modern targeted treatments—none of those were for my cancer either. The reality in front of me, which was that there’s really not much that they could do for me, hit me. And it was diving into that and recognizing that the fate that stared me in the face stared into the face of hundreds of thousands of new diagnoses every year in America and millions of patients, all of whom collectively had various forms of rare cancer. But, all told, we make up about a third of the U.S. cancer population. So it was, sorry, there’s not much we can do for you. If it comes back, there won’t be many options. Again, no approved drugs, no approved chemos. So, the initial treatment was I had multiple surgeries—first one, actually, I had my tumor cut out while I was awake under local anesthesia on accident—so not the most auspicious start. Did transfer over to Mass General. After that, I had a second successful surgery, [and] would’ve had local radiation of my head and neck where they just basically blasted me from the neck up, which would’ve been awful. I did get to miss that to date. And so, they sent me on my way and said, typically, patients are healthy for the first five years or so. As you get closer to 10, and I’m about eight years out now, typically that’s when you start to see this thing come back. It has a pretty high recurrence rate. So, it was kind of like a Sword of Damocles hanging over my head. I had time, but I didn’t know whether it was another five years or another 50.

Daniel Levine: Going through this experience, you got to learn a lot about the way rare cancers are treated. What problems did you come to recognize?

David Hysong: First off, just the sheer prevalence. One that, again, we talk of rare cancer as if it’s a rare thing, and it’s not. Three hundred eighty or so out of the approximately 400 types of cancer recognized in the literature today constitute a rare form. And beyond that, again, we’re still talking about cancer, which is a 2000 year old word. We’re talking about lung cancer or brain cancer, breast cancer, where cancer simply is located in the body, all of which is just so outdated. So the overall thinking behind this certainly needed to move to understand that rare is not rare, that there are tons of forms, hundreds of forms of rare cancer. And then beyond that, that really, again, every form of cancer, every single individual patient has a unique “fingerprint,” genomic signature that is uniquely theirs. So really, we should be thinking about this in terms of rare not being rare, but then almost the ability to then regroup patients in a way that we have not done yet. So, I think that was the first recognition. The second is that for the majority of those cancers, and that’s all forms of pediatric are rare, a tremendous number of adult and common cancers, that there simply aren’t effective therapeutics. There aren’t approved therapeutics, there aren’t effective chemotherapies. And for many of these, there aren’t even treatment guidelines for doctors to follow or they’re decades old. So it’s kind of a wasteland out there for a number of patients. I think that we do things, we give drugs to patients that we know won’t work in the hopes that they will, but the truth was that very few patients actually had effective, precise, targeted, personalized medicine.

Daniel Levine: How did you go from having what’s this uncertain future with a diagnosis and a disease that’s lurking out there without treatment to founding a biotech?

David Hysong: Danny, I’m a literary guy more than a science guy by training in education. And I always tell this story, there’s a favorite poet of mine, and he wrote about what it feels like and what it looks like to die of nothing but a rage to live. And that kind of hit me deep down in my gut somewhere, that idea of dying of nothing but a rage to live. I just made the decision that if this thing was ever going to win, that it was not going to be without a heck of a fight, and that I was going to wage that war and fight that fight and that battle not just on behalf of myself, but on behalf of as many patients as possible. It was just kind of a blind determination, diving in knowing that I knew so little about that fight, but that I believe that I was willing to set out on that journey, that others would be willing to join me, and thank God that turned out to be the case.

Daniel Levine: Who is Gene Williams and how did you connect with him?

David Hysong: Yeah, Gene was my original co-founder. He was a Harvard and Harvard Business School grad. He had been an early employee at Genzyme, the first rare disease biotech under Henry Termeer, and had also really helped pioneer and push some of the patient-centric initiatives. So, everyone is, or not everyone but many people are familiar with John Crowley and the Novazyme story, great movie called Great Expectations, or Extraordinary Measures, excuse me, about the father John who founds a biotech to save his children. Again, a rare disease like mine, some promising science, not much capital to take it forward. Gene had been an early believer, an early witness, to the power of patients, really willing to dive in and own their fate. So, I sat down for a coffee with Gene and said, “Gene, I’ve had this idea, what if we did a Genzyme but for rare cancer, specifically in oncology?” And I expected that I would take a few years to go to business school and kind of shop and build the idea. And he just said, “David, the world needs this. You should do this. Do this right now and I’ll help you.” And that was that.

Daniel Levine: There are a couple of different entities you founded. There’s Shepherd Therapeutics. There’s also Shepherd Health and the Shepherd Foundation. I want to focus on the approach of Shepherd Therapeutics, but can you separate out those different entities and explain what they do and what, if any, interplay exists between them?

David Hysong: Yeah, absolutely. So the experience in Cambodia honestly taught me a lot about what you do when you’re trying to root out some sort of systemic injustice or solve a systemic issue. And so, from my standpoint, I saw the lack of treatments for rare cancer patients, which is what it is. It’s a market-driven problem. Companies have a responsibility to develop the drugs that will make the most money for their shareholders. It is what it is. It’s American capitalism in the 21st century. So I wanted to take a holistic approach, one that leveraged and harnessed market forces to compete in that for-profit arena. But I also wanted to build something that would really focus on helping to write policy and advocate and do all of the more philanthropic legislative initiatives. So the foundation really focuses on that. We have a piece of federal legislation in the House and the U.S. Senate right now that would mandate insurance coverage for sequencing, for genomic sequencing for every patient in the country—so very proud of that. We’ve done a lot of advocacy work on the Hill, every congressional Senate office, all the major agencies—so really being involved at that level of decision making. And then again, Shepherd Therapeutics and Shepherd Health are the for-profit entities, one focuses more on drug development, and then one focuses much more on patient care, on being the partner of choice for both doctors and physicians, able to identify the very highest potential, most promising personalized medicine recommendations for a patient based off of their RNA and genomic signature.

Daniel Levine: I suspect people who looked at your bio would think it was among the most unusual that they’ve seen for a biotech CEO. You worked as an investigator in Cambodia, as you referenced, investigating sexual slavery. You have several academic degrees, but among those is a divinity degree from Harvard. What gave you the confidence to establish and lead a biotech company, engage in AI drug discovery?

David Hysong: I don’t think that anyone truly knows what they’re doing the first time they set up their first company. I think that I had a lot of leadership experience, and I had a lot of experience with learning how to talk about a vision and a story in a way that made people want to be a part of it. And so, I think that’s really what I leveraged. Honor and surprise garnered us a number of awards, some recognition by Forbes and Charles River and whatnot, a number of articles people took a real interest in what we were doing, and it also really garnered the attention of some really top tier individuals. Dr. Keith Flaherty is a dear friend and advisor to Shepherd, and we are actually setting up a drug development company right now specifically to target drug development in India and Africa and beyond, kind of a global scope. He’s found a number of companies. He’s the director of clinical research at Mass General, just an amazing, amazing guy. And Keith and I really connected off a shared level of literature, which we’ll talk about in a different conversation, but from a mission standpoint. He said, “David, look, there aren’t a lot of people who are really trying to prioritize patients right now the way that you are.” And so he came on board, wanted to help, and thank God the team was able to build just some really wonderful and revolutionary technology that helped back up that vision and those claims. So I’d say that the non-traditional experiences have honestly been a real boon and a real help, and that they really helped you to think through an issue from different angles and an outside the box approach and to come at an industry from the standpoint of, okay, what can I draw from other industries? What hasn’t been done here? Whereas I think if you grew up in that industry and you come from a more traditional background, you might not be as privy to.

Daniel Levine: One of the other interesting things to me is the way you’ve gone about building a team, the people you’ve selected have a very personal connection to the company’s mission. Can you explain the thinking there and how that’s played out?

David Hysong: It did, yeah. And I think that a lot of it was a willingness to bring on people from different backgrounds. The belief that from a systematic standpoint, we have the resources, the knowledge, the technology to make a difference in these diseases. And then [with] the Indy One individual with enough motivation, passion, and intelligence, we can solve the problems we decide to solve, as the common refrain there. And so, I very intentionally built a team of misfits, and there were pharma execs that had been in drug development for 30 years, Bristol Myers Squibb, and Sanofi, and Otsuka, and Genzyme. There were physicians like Dr. Keith Flaherty, and there were computer scientists, engineers, and previous college dropouts turned entrepreneurs, all of whom were united by this undying determination to make a difference. And so, when you can sit down in a room and tell people that the goal is to give patients decades of extra life, not days, and to drive for 60, 70, 80 percent response rates not 20, 30 or 40, and they don’t blink it up, they don’t tell you that that’s dumb or it hasn’t been done before. They say, yeah, that’s exactly what the goal should be. Let’s figure it out together. So I think that’s really been the common denominator and has enabled us to build what we’ve built.

Daniel Levine: Shepherd has a precision oncology platform called DELVE. Rather than looking at a single target, the idea here is to identify complex and interconnected mechanisms. What’s the case for doing that in cancer?

David Hysong: Absolutely. I think it started from simply two data points. One, it was recognizing that the biomarkers that are being used in drug development are almost never effective. The single point biomarkers, when you take them, especially when you take them outside the original cancer or organ of interest, we recognized in a number of analyses that they’re oftentimes no better than a coin flip. A drug has an infinite biological impact. It’s doing so many other things. And so we really wanted to take a step back and also add to that the fact that we were working on rare refractory metastatic pediatric cancers where a lot of these had no basic scientific spending, NIH hadn’t spent on them. So we didn’t have targets of interest. We sometimes didn’t even have models to work with. So, it forced us to go back to raw patient data, and I say raw patient data, not limited panels with a few hundred genes, but the full transcriptome. And so, we really made a bet on RNA data. We incorporate DNA, [undecipherable], et cetera, but we are really focused on RNA as our starting point. And so we look at a drug’s activity across as many types of cancer as possible. I think that our hypothesis is that a cancer drug can work in different patients for very, very different reasons. And we see that play out in the clinic, and it might work in you and I, Daniel, for very different reasons, whether with the same cancer or different. So, its overall activity is dependent on the microenvironment, on a number of off-target and unstudied impacts as much as on target and studied. What the platform does is we focus on a huge data spread. So, let’s look at this cancer and the drug in as many types of cancer as possible, let’s do a good data distribution of models or patients if it’s in the immunotherapy space that were really responsive and those that were not at all. And then the middle area where patients were somewhat or somewhat not. And we really were able to zero in on what do the highest responding patients have in common, and so, set mathematics loose on that. You’re looking at the whole transcriptome at 20,000 plus genes via the RNA, and then we’re able to mathematically extract all of those highest responding, all of those in that group, what do they have in common? And so then instead of a single marker, we’re pulling out usually several dozen to several hundred genes that actually matter—what we like to call the fingerprint of the drug or the RNA signature of the drug. And then you’re just matching that to an individual patient, whether you’re doing that on an individual, N-of-1 basis, for a patient that has cancer today, or you’re looking at, say, a group of patients for drug development. You’re just saying, does the genomic fingerprint of this patient match the fingerprint of the highest responding group? And if the answer is yes, then you recommend the drug, and if not, then no. So, think of it as instead of whether a small, medium, or large on a piece of clothing, it’s leveraging 20,000 different measurements that could be even more robust than that. And then really tightening in on what precision medicine can.

Daniel Levine: And you are focusing on the human transcriptome, specifically the tumor RNA. Why do the RNA rather than say proteins that may be over expressed in a tumor or the DNA drivers of the tumor?

David Hysong: Absolutely. So, from our standpoint RNA, it’s DNA put into action, right? So, DNA is starting to get to work, so a little bit more robust information there. We think that it kind of pulls in a number of other factors that, especially when you’re looking at it and people familiar with the conversation, when you’re looking at bulk RNA rather than single cell, you’re pulling in a lot of different inputs in terms of everything that’s factoring into what makes that tumor grow. So that’s DNA and RNA, the protein level connections—we leverage those—but we wanted something, we wanted to take a hypothesis agnostic approach and set mathematics loose on raw genomic data. And the transcriptome is wonderful because it’s more robust than DNA, but you can cap it and then you can unleash mathematics on it, whereas the proteome is highly complex, millions of proteins, we don’t even really know how many, so it’s a little bit more difficult to book in, put a cap on and say, okay, interrogate this. So, we really like RNA as the kind of central starting point, and then we do map forwards and backwards and integrate other information, but that combination of highly complex, robust information, but still able to be interrogated mathematically. One final point is obviously lots of buzz around AI and machine learning and all this. And so we absolutely leverage artificial intelligence in our work. But the core platform, the core mathematics behind Shepherd is white box, and it is, again, advanced math. And so we’re able to unpack why the platform pulls out what it pulls, and then map that forwards and backwards so it’s not a black box that just spits out an answer. It allows us to unpack and understand, oh wow, there’s just a high degree of complex connections at work that oftentimes connect to the clinical targets of the drug, but do so in a much, much more robust manner and integrate all of the off-target effects as well.

Daniel Levine: So, there is a lot of excitement around AI. In the end, the results are only going to be as good as the data the system’s using. Can you walk us through in a little bit more detail about the inputs, where they come from, how unique they might be, and how much of it is derived from experiments you might conduct beyond that?

David Hysong: Yeah, absolutely. The input data, so about 500 drugs, 509 drugs in the platform to date. Those are FDA approved. Those are repurposed agents. They have all been seeded with, again, we try to evaluate the activity of a drug across at least a hundred different models or types of cancer, as diverse as possible, ideally even more than that. And so we’re able to optimize, we need to have response data and then the associated transcriptomic data, the raw RNA data associated with that patient or that model. And so again, it’s that diversity that’s really important. And so, you get a nice data spread of responders, non-responders, and those in the middle, and then the platform algorithms that are then unleashed on that. It’s Bayesian based, Bayesian statistics based, all of it built in-house. It’s not only taking into account which genes via the RNA are mutated and dysregulated, it’s saying, okay, what is the direction? What is the degree of that mutation? And then what is the overall contribution of that specific gene? And that equation is built for every single gene in the dataset, so again, for more than 20,000. And so then the output, as I mentioned, is this kind of full signature with those three associated variables that I just stated. And then you’re able to again, take that signature and then map it on. So that’s the kind of initial input data that is most important, that it’s the starting point. And then we’re able to look again, like you said, at DNA, at the proteome, look at when we are working on an individual patient, how that signature maps to the patient’s dysregulations. It has led to some remarkable insights, certainly in terms of new drugs for individual patients or indications that had not been explored systematically. I think we always find that when the platform pulls a high recommended drug that there is robust evidence in the literature, oftentimes in a small clinical trial or an N-of-1 study or publication. So, it’s really helping to connect all those dots. And then again, when we present to a physician or a patient, then we’re making certain that we incorporate all of that literature evidence, all the bioinformatic evidence, all the things that a physician would typically expect to see, so that again, they can understand, “okay, I wouldn’t have arrived at this drug combination or this drug for this patient, but man, this makes so much sense.” There’s great evidence in the literature and it’s a rational approach to drug identification as well.

Daniel Levine: And is the goal here to use this as a way to determine a precise treatment for an individual patient, or is it to develop therapeutics that could be used more broadly for patients with these rare cancers?

David Hysong: Very much both. So Shepherd Health, as you mentioned, that is the patient facing company. I always wanted to put this power directly in the hands of patients. I think that a lot of times as patients, we feel like we’re on defense, that we don’t have a lot of autonomy. There’s, I think, just a remarkably positive and healthy wave coming that started in consumer health and now is moving more and more into actual acute care and healthcare, which is that patients are taking responsibility and they’re choosing their doctors, they’re choosing their hospitals, they’re getting second opinions, they’re identifying additional companies. So certainly, wanted Shepherd to be partnered with as we have been—we’ve been part of a national consortium of some of the best hospitals out there. So wanted to go top down via traditional distribution channels, but also wanted to build that patient first company so that a patient at a community hospital or in a community setting where 85 percent of patients are being treated can come and say, “I know that I need Shepherd”, order the test, take it back to their treating oncologist, and then we’re able to work with the treatment team again to identify that. So, then the output is the short list of therapeutics that are personalized to the patient’s transcriptomic signature and status at that specific moment in time. Again, as a patient mutates and changes that fingerprint or that signature changes, we like to have as up-to-date sequencing as possible. And so, it is highly bespoke, highly personalized therapeutic identification for the patient at that moment in time.

Daniel Levine: What’s been done to validate the approach?

David Hysong: Absolutely. So again, very first, we built it as a drug development tool. I mentioned that it’s really the same approach from a drug development perspective. We’re taking therapeutics. And so again, very interested in that approach and doing that as well. So, biopharma partnerships in the works as well as the drug development company I mentioned. And so, it’s taking, instead of 500 drugs for one patient, it’s saying, okay, let’s take this one drug and look at where the patients are located and identify across types of cancer. And what we typically see is that we’re able to identify dozens of forms of rare cancer that would benefit, and oftentimes the patients are distributed. So, there are a handful of indications where the majority of patients might be located, and we can help identify what those criteria are, but we’re able to include a dozen or more other types of cancers, oftentimes rare, that would never have been included or thought of for the drug development. The very first patient we treated on an N-of-1 basis was at the request of a Mayo Clinic physician, and it was a four-year-old with metastatic refractory rhabdomyosarcoma, intubated, ICU, failed 12 therapeutics. Parents had been told to put her on hospice three separate times. So, the physician at Mayo and another treating physician at Cleveland Clinic requested Shepherd. We had her sequencing run, ran it through the platform and identified two drugs, a two drug combination that they ended up selecting. And three days later, she was out of the ICU. And two weeks later she flew home. And this was a patient that again, her parents had been told, let her go, put her on hospice. So based on the strength of that initial outcome, we joined a national consortium of hospitals—a Beat Childhood Cancer Consortium. So, we’ve been sitting on a national tumor board for the past two and a half years where we typically focus on and help treat patients, pediatric patients that have exhausted standards of care. The doctors don’t want to give up, but they don’t know what else to do. So, they run sequencing, bring it to this tumor board, and so then we’re able to run the platform and then treat patients. I think to date something around 38 distinct types of cancer have come through, and then again, [we] just soft launched a couple months ago, broader than the consortium, making this available to patients around the country and then gearing up here in 2024 to really drive that growth and really hard launches.

Daniel Levine: How can patients or physicians access the service?

David Hysong: Really simple, come to the website. I mean, we really tried to streamline everything through there. So certainly [we] have building more and more physician relationships and direct referrals, et cetera, but have made it really simple for patients or physicians to come just direct to the Shepherd website, shepherd.bio, input their information, and then basically we work with them so we can get whatever sequencing might need to be completed. Usually that’s already been done, but then get their data so that then we can run the platform and generate a report.

Daniel Levine: In terms of the drug discovery efforts, where are you in that? Do you have a pipeline at this point?

David Hysong: To date, primarily with biopharma partnerships, both in the small molecule and immunotherapy space. As I mentioned, again, setting up a drug development company with Dr. Keith Flaherty, so focused on global markets and also the U.S. So that’ll be the first initiative that brings clinical stage assets in-house at Shepherd. So, we’re hoping that that can be just a replicable model where, again, we can distribute drugs across the globe, but also make an impact here at home. So I think that’s the focus of the two. One final note to make is I think one of the most remarkable things has been when we’ve worked on an individual indication, like we’ve had a lot of rhabdomyosarcoma patients, for instance, because of that connection with Mayo Clinic physicians, it has never failed that the platform, , identifies really interesting novel targets, novel drug classes for indications that have not been explored. So Shepherd is not in the business of designing new molecules to date, but in terms of uncovering new insights, much, much better biomarkers in novel targets for specific indications, pretty much that’s par for the course and a pretty standard part of our output.

Daniel Levine: It strikes me that maybe one of the bigger challenges here, particularly since you’re dealing with combinations of therapies that may not have been used together, are really safety concerns. How does your system account for safety? Is that built into it in any way?

David Hysong: So right now, no. There is a project underway to do that. So basically you should be able to computationally predict whether a drug will be toxic to specific organ systems or not. That will take additional validation. So right now, the safest thing for patients is for us to rely on the physician’s input, known toxicity. So, when we have a combination come up, I think we have 612 known combinations identified that are in our database—so known oncology indications, we’re always looking to see whether the drug has been combined in oncology for pediatric or adult patients. So that’s the first gate and [we] put that right on the report. Second to that, right now, very much relying on the treating physician and the care team. They’re going to know their patient best. Again, we might say that a drug wouldn’t or would be toxic in general or for that specific patient. They’re going to know what the patient, if they’ve had a maximum dosage of a specific therapeutic or if the patient has other toxicity concerns that they would be concerned about. So right now, leaving that in the hands of the physicians. I am hopeful that we help build a world and are one of the companies that in a couple years’ time can actually predict likelihood of metastasis, likelihood of treatment resistance, and yes, actual model the toxicity for any specific therapeutic and/or combination.

Daniel Levine: And how’s the company funded to date and what’s the plan for raising additional capital?

David Hysong: Yeah, absolutely. So today, funded by three primary family offices. The first was Dr. Ruth Shaber. She’s a Forbes Top 50 global impact investor, former physician at Kaiser Permanente, runs her own venture firm, philanthropic fund, et cetera, so really amazing, amazing lead angel investor that helps build Shepherd. And then two large family offices that were impacted by cancer very specifically and very personally. So right now, moving into two things—finalizing what will be our first large institutional round to really drive on the biopharma partnership piece, the patient side of things. I’ve always wanted Shepherd to also be by patients for patients for a number of reasons. I always really was invested in having patients at the table, even having patients as part of our cap table and give them the opportunity to invest. We are doing a community round where we’re asking and inviting patients and doctors to also have a hand in building the early patient version of Shepherd, Shepherd Health—so a combination of institutional rounds with classic venture capitalists and some wonderful investors as well as the community rounds. And then again, I mentioned the separate company with Dr. Flaherty that will have a separate group of funders as well. It’s been really wonderful to see the type of investors that are out there, ones that are truly, truly interested in making a long-term impact for patients. They’re not always the easiest to find, but fortunately we have been successful in finding some of those.

Daniel Levine: David Hysong, co-founder and CEO of Shepherd Therapeutics. David, thanks so much for your time today

David Hysong: Daniel. Thank you. Appreciate it.

This transcript has been edited for clarity and readability.

Stay Connected

Sign up for updates straight to your inbox.

FacebookTwitterInstagramYoutube