First in Human Episode #2 featuring Dr. Keith Flaherty

Episode 2 of First in Human features Dr. Keith Flaherty, Founder at Scorpion Therapeutics. First In Human is a biotech podcast that interviews industry leaders and investors to learn about their journey to in-human clinical trials. Presented by Vial, a tech-enabled CRO. Episodes launch every other Tuesday.

Simon Burns: Thank you so much for joining us today.

Dr. Keith Flaherty: My pleasure. Thanks for having me, Simon.

Simon Burns: Awesome. Let’s kick it off with the quick round of introductions. I’m co-founder and CEO of Vial. We launched this company with the mission to reimagine clinical trials, to make them far more efficient and faster for sponsors. Dr. Flaherty, world renowned Oncologist barely needs an introduction. But maybe for the purposes of the audience, would love a quick background. 

Dr. Keith Flaherty: Yeah. I’ll maybe just kind of highlight a few elements of my background that I think are relevant to our conversation. So medical oncologist of 22 years, vintage now. That number itself, I suppose, is a little bit relevant because that’s as long as the concept of targeted therapy in oncology has existed. And that’s, you know, kind of, I guess, by design in the sense that I came into the field specifically because I had a sense that genetic insights were starting to give way to therapeutics.

And I spent nine years at the University of Pennsylvania finishing my training in medical oncology, then building [inaudible 00:01:26] in my academic career there. Moved to Massachusetts General Hospital 13 years ago to build out a therapeutic development group broadly across cancer with a particular emphasis on early phase clinical trials. So sort of what nowadays kind of fits into the large category of phase I/II clinical trials.

At any given time we’re conducting dozens and dozens of trials in that realm, phase I/phase II. And in cutting across, you know, various modalities, targeted therapy, immunotherapy. We have a salt therapy group as well. And guess maybe two other things that I would mention. One is that I’ve had just about for the entirety of my career, I’ve also been involved in NCI funded work. Both on my own and in institution but also through the cooperative groups, so ECOG-ACRIN is my rapome.

Cooperative group which gives me a kind of fat lens, I suppose in terms of publicly funded clinical research. And then lastly, starting nine years ago I co-founded a series of companies starting with Loxo Oncology. And most recently, Scorpion Therapeutics which has just given me another dimension or view on the world of clinical trials. Kind of looking at things from the sponsor’s side and not only from my long standing position as an academic investigator.

Simon Burns: Fantastic. Before we jump into talking to clinical trials, ESMO just wrapped lots of great data bits, lots happening. I’m curious what your takeaways were from the conference.

Dr. Keith Flaherty: Well, as I just mentioned, my career long focus has always been on early therapeutic development and so I guess for someone like me it was a little bit of down year at ESMO. I’ve been saying this about ASCO for a long time too that we seem to never get like two consecutive ASCOs where there’s lots of really kind of eye popping new results.

And so it’s like an every other year type phenomenon. I feel like ESMO was a little bit of a down year also. Again, just from like kind of a like new drug, first in class, you know, like wow. We’re now at the beginning of a new era with you know, a given type of therapy. Rather, ESMO had a lot of data that I think was important but just a little bit more confirmatory kind of later stage stuff.

So certainly in the targeted therapeutic space, the D12 E inhibitor data. Was nice to see more evidence and now, earlier line of therapy evidence in non-small cell lung cancer. The colorectal cancer data certainly I think captured a lot of attention. And the response rates like monotherapy, not so great. Combination would be EGFR antibody encouraging, but the duration of response like really quite short and these are, you know, genetically complex tumors. So perhaps we should be totally surprised.

Also, suggests maybe there’s room for improvement in terms of the underlying KRAS therapeutics. There’s just a whole wave of those coming forward beyond the Mirati and Amgen compounds. And then I guess the last thing I would say is that antibody drug conjugates, wow, are they here to stay. [laughs]

Simon Burns: Yeah.

Dr. Keith Flaherty: I guess you could say that within HER2, right, that’s just you know, like that asset is really makes a very powerful statement. But of course, Trastuzumab, earlier the past couple meetings triple negative breast cancer non-small cell lung cancer data, all supports this. But just in terms of the number of early antibody drug conjugates as reflected in ESMO data. It’s just carpet bombing right?

There’s every surface targeted for which one can direct an antibody and attach a payload on the backend. I’d say that I think there’s a loud and clear signal from ESMO and other major meetings in the last year that antibody drug conjugates, we’re gonna see complete saturation in that field pretty soon. 

Simon Burns: Great breakdown. Definitely feels like the era of the ADC is upon us. Let’s talk clinical trials. You’ve seen them from every angle obviously, administrator, PI, sponsor last nine years. [00:05:00] What’s broken? Is the obvious question. What do you think is the key also clear here to kind of billing clinical trial structure that scales , and supports sponsors?

Dr. Keith Flaherty: Well, how much time do you we have for this? I mean, it’s yeah. It’s kind of a long list. I guess maybe I would start by breaking down that my thoughts into kind of two domains. One is kind of operational and so basically there I would say we reinvent the wheel constantly, constantly. From the academic site side, in interacting with sponsors and CROs we’re constantly reinventing the wheel.

 And yet our pharmacy capacity and our pharmacy SOPs and our, you know, clinical laboratory, operations, and our clinical trial infrastructure and capabilities. They’re not reinvented every month , or two months. And yet, we’re just constantly having to kind of, you know, present ourselves. I mean, this is true across the field right? All types of sites large and small, the same.

So this idea of having to qualify as a site over and over again like that, that feels like something that’s broken. We should just be able to simply deposit if you will and update our capacities and the specifications of, you know, do we measure troponin I or troponin T. I mean, super simple stuff like that, right, where we just shouldn’t have to be filling out forms over and over again. Rather we should have a way of working with a CRO partner where they can catalog what sites can do, how they do it. And be able to present that, if you will to drug developers. 

Those are very much in the weeds issues. I guess maybe I would jump straight from there over to a bigger picture issue which is that, biomarker guided therapeutic development. Obviously has been a thing in the quote unquote targeted therapies space, but increasingly is going to be a thing in antibody drug conjugates per our exchange. And immunotherapy, believe it or not.

I mean, you know, precision medicine principles are gonna ultimately inform all of our development concepts. And then not just in the future, and so we’ve been increasingly biomarker characterizing our patient population like with next generation sequencing, right, is the obvious starting point. But I would even start by saying we ourselves don’t really know who we have. And I’ll blame this a bit on the electronic health record environment in that which we find ourselves. Here I don’t just mean Mass General.

When I say we, I generally mean actually the whole field of academic and non academic therapeutic developers. So that’s a starting point. Do CROs know what we have in terms of molecular characterized patient populations? Like no, I think not. Not remotely. That pipeline hasn’t even really been built in a significant way. Do sponsors know what we have? No. So ultimately trying to understand just from an execution perspective of where are the patients? And in biomarker defined terms … have they, at any given institution or across the whole population, how many of those patients are in the hoster section. You know, sort of adjuvant situation may or may not relapse. How many patients are on the first line setting, second line setting, third line setting? How are currently available therapies even serving these patients? This is critical information to guide drug development and execution of clinical trials. But the inefficiency of how that data is jailed up I mean in our incredibly antiquated systems on our end, but I would say also without any kind of pass through CROs as sponsors.

 Like to me, to call it broken, it’s like what? It’s never even been built, I guess. So like, you know, can it be broken if it hasn’t even been built? But these are the kinds of things that are just described as like massive inefficiencies in terms of how we operate.

Simon Burns: We think a lot about everything you just said. We think a lot about how paper source is still kind of the dominant way that research visits captured. It’s certainly not intelligent or forward. There’s no logic. We think a lot about how the data cells you, you talk about hold back research. It’s still surprising to me that CRO’s had been built and the infrastructure built in the large database. And then kind of accelerate the process at that way.

Maybe on this there’s a lot of ways technology can be applied to impact clinical trials. Do you have a short list? Do you have a favorite list of applications where you think technology could have the biggest impact?

Dr. Keith Flaherty: Yeah. Let me describe in a patient journey way. But at that I mean not so much through the course of their cancer diagnosis, but really kind of more like as they’re navigating treatment and the context of the clinical trial. Patients are home more than they’re with us, right? So the typical treatment cycle of 28 days, right, 27 days or 28 days, patients are at home.

How do we interact with them? By email and phone calls. Like that’s kind of preposterous, right? So you have going back to the kind of the concept of things being kind of jailed up. You have what people refer to as patient reported outcome tools. But I don’t wanna call them by a conventional term. You have tools by which patients could be describing for us how they’re feeling, how they’re functioning in the context of therapy.

So I want to know about how they’re feeling and functioning as it pertains to their cancer, of course. But in particular, I would like to know how are they feeling and functioning as it pertains to the treatment that we’re giving them. Right? So I want not a 104 question confrontation for a patient to have to deal with every morning because even a highly motivated clinical trial patient’s probably not gonna torture themselves through that interface.

I want a nimble dynamic question architecture that patients can respond to, you know, bespoke for a [00:10:00] therapeutic. Right? So whether it’s a targeted therapy, it’s gonna cause rash and diarrhea. Whether it’s a immunotherapy that’s gonna cause autoimmune type. If it’s a cell therapy, it’s gonna cause cytokine release syndrome. I want bespoke tools, right, that where a patient if they answer no to the first five questions as in they’re feeling fine. Great, they’re done. Like that’s it, end of the day.

One question yes, like builds out to a tree of seven or eight follow on questions that delve into more details. Anyway, that’s the beginning, right. So kind of this high resolution data that we’re lacking that would need to be no more than even daily, I think to just absolutely explode our understanding and our knowledge regarding the therapeutics that we’re investigating and what they do. And particularly focusing on toxicity of course.

now coming to the patient journey as it pertains to like now, on day 28 or day one. They come into your clinic. Well, let’s go with day 28. So now they’re back in the clinic. So basically, because we don’t have the tools that I just described, what’s happening is they’re interacting with never less than five site staff including myself. You know, clinical research staff and clinical staff, right. So we have two pockets of people, never fewer than five people a patient’s going to be interacting with.

And information’s going to be gathered from each of those five people over the course of a few hours in which they’re with us, producing conflicting information. Right? So we don’t have an account from the patient for the previous 27 days, or for that matter, just the day before to which we can each respond by asking clarifying questions. And just confirming and corroborating. We don’t have that. Rather, we’re scrambling to get partial account number one from the first interaction. Then account number two produces conflicting information and non overlapping information from the first account.

Then the physician oftentimes is in the fourth position and then, let’s say the patient’s getting an infusion. Now, the infusion nurse interacts and produces some more information, if you will, later interacting with the patient. And then after that clinic day, over the course of the next about two weeks, we seem to reconcile what it is that we heard from the patient. Well like, why do we not have the patient actually telling us, right, using a PRO type tool, like how they’ve been doing. The morning of the visit, we have a representation of up to the moment basically how they’ve been doing, to which we can, you know hone and clarify and mold that account.

So that everyone including the treating physician has ultimately gotten a chance to essentially sign off and corroborate that account. And then all of it would be data, right, as opposed to this mad scramble that begins after the patient has left the clinic, where we’ve now got this rather ridiculous you know. I wouldn’t call it paper trail but this kind of scattered attempt to actually capture the patient experience. All of that is just, I mean, it’s maddening to me, right.

But like my patients and I can’t actually sort of share information about how they’re doing. And where I can just simply review, corroborate. My research nurse can, you know, kind of reconcile a couple points, get more detail as needed. Right. What I’m getting at is not just missed data but we have bad data as a consequence of these lack of use of tools. That frankly have been developed in just little kind of pockets here and there, but just haven’t been brought together.

Simon Burns: Totally. It’s a little shocking that the average dentist visit that you can go through that. The you could flow on your phone a few hours beforehand and get clinical research. No comparative modern technology in place. Taking a broader lens… did I get it company number nine now? Biotech company number nine? Is that where we’re at?

Dr. Keith Flaherty: No, no, no. Those are still in my head I think. Seven companies over the past nine years. So nine is the number of years . Still stuck on seven companies right now.

Simon Burns: Sounds good. Either way, you know, remarkable on the number of companies you’ve created and the impact you’ve had. I’m sure there are tons of early stage, other grad students leaving or early stage biotechnology teams asking you. You know, you run these successful trials. How can I run a successful trial? What’s key to running successful trial and I’m sure you’re asked everything from CRO selection to site selection, how to pick. How to think about protocol design. What advice do you have for these biotech companies?

Dr. Keith Flaherty: Yeah. Well, I mean, I guess I’d mostly answer that question relating to that kind of perspective of things being broken. And because I consider kind of the relationship ecosystem between sponsors, CROs, and sites to be not very functional what I usually advise is that a sponsor, my framing of course, is small biotech sponsor, right? Like program number one, program number three, program number five but like, you know, that, kind of sponsor’s world view.

 Each program is your be all and end all. It’s your precious value of creating opportunity. You wanna reach right out to the key sites that you’re gonna be collaborating with, particularly looking at the phase I/II. Like as you’re aiming towards first clinical investigation, you wanna reach out directly to build those relationships because there’s no other trusted way of actually having a dialogue.

The way in which we can have dialogues, right. You know, outside of medicine, outside of biomedical research, outside of clinical trials. They’re incredibly dynamic, right, but we don’t have that [00:15:00] sadly in our world of clinical investigation. And so that means that new outreach has to be made all the time to try to find champions. Find people who agree with you that this science is compelling. Agree that you need to attempt to translate it to medicine and have that conviction and sort of shared notion that this is one of the most important things they can be working on. That’s, that’s the key point, right.

You need to find those partners. But then, back to the topic we talked about a little while ago. It’s great to have that conviction. Do you have the patients, right? And I don’t just mean do you have non-small cell lung cancer patients. I mean, do you have the type of patients I’m actually looking for ? And that’s a molecular feature issue. That is a line of therapy issue. That is a brain metastases, yes, no. And if yes, you know, more details, right. The, the back to data. I mean no one carries around in their head a detailed understanding of the number of patients that they might truly have available once you’re filtering through 40 eligibility criteria.

And so what I’m trying to get at here is that, what I would advise is that to some reasonable approach to approximation, the sponsor needs to try to make sure that they hit their ground running by virtue of doing the work I’m describing manually. That’s the 2022 answer to your question. We basically just reach out by email. We arrange the Zoom meeting and we start to have an interaction that is not even semi-quantitative. It’s like fully qualitative about whether there’s scientific alignment. And then is there feasibility alignment between sponsor and site. And then we have CROs to come along and plug in the operational gaps if you will, get site onboarded.

And you know, kinda help us get up to speed in terms of actually operating the study. But any case, this is what I’m getting at here is a bit of a mad scramble. When you know you’ve got a development candidate and IND, you know, is kind of ID filings. A date on the calendar. It just over and over again feels like, to reuse a phrase I used before, we are reinventing the wheel every single time. That’s been true in my role within the companies that I’ve founded as we get near to clinical development. And it’s how I see it all the time as an academic, interacting with many more companies who are coming to us.

Simon Burns: Totally. It’s a huge problem with evidence sort of alignment, these, these large CROs who certainly don’t have alignment incentives in a lot of these cases. I wanna go before clinic /pre clinical, with Scorpion, you’re now, you know, advancing precision oncology 2.0 as they’re calling it. I’m curious to understand some of the key opportunities there are. Huge amount of need for data. We’re now moving into a bigger data environment.

What are some infrastructure companies or some data sets? What could you and the Scorpion team use to amplify and enhance the mission that you’re working on?

Dr. Keith Flaherty: Well I think it really is just a bundling of a few of the things we’ve talked about. Our lead lead program is like PI3 kinase, mutant selected PI3 kinase inhibitor. PI3 kinase mutations are the most common activated in all of cancer. And most commonly found in breast cancer but a kind of broad distribution over the cancer types in which these mutations are found.

Alpelisib is the only FDA approved PI3 kinase inhibitor in solid tumors. And it’s alpha selected but not mutant selected, and only approved in breast cancer. So my questions as like that program is moving briskly towards clinical development really pertain to who out there has interest? Who has so much interest that they’re actually working with a couple of the companies that are moving very similar assets into the clinic, right? So competitive landscape questions. I wanna know who’s doing what.

You know, self-reported. Now PI3 kinase having seen such a common mutation that a number of sites can actually support more than one of these trials . So this is not just a yes/no exchange here but really wanting to understand the broad definition of competitive landscape. Not just focused on mutant selective guide kinase inhibitors. Right, because breast cancer patients and head and neck cancer patients, endometrial cancer patients and non-small cell lung cancer patients who have PI3 kinase mutations, they’re eligible for other trials as well.

So I actually wanna know really the full then diagram representation. Okay. I kind of need to know where these patients being drawn within your own clinical trial portfolio at a center. And where would you place this in your priority scale? Like be honest because I’d rather not find out six months after we’ve activated at the site, that it’s actually the 12th priority in their clinical trial portfolio. So this dance needs to be done where basically we simply find alignment. where are the champions who have the patients who actually consider this study to be a priority in the broader landscape.

Again, not you know, beyond just the individual [inaudible 00:22:36] that we’re actually targeting in its trial. But then back to the point that I raised before, how many PI3 kinase mutant patients do they have? Where are they in their journey? How many of them have actually been cured with adjuvant therapy, aren’t going to be available for a trial that’s enrolling metastatic patients. These are very real filtering questions where right now, we don’t have any ascertainment of them at all.

So what I want in a perfect world is I want partnership outside of the therapeutic development companies. So, as a site. Now putting my MGH hat back on. I want [00:20:00] partnership with an entity that’s going to work with us to build the data, data warehouse, databases that are dynamic with our patients as they’re going through their treatment journey. So that we ourselves know what we have and so that we can communicate that with trial sponsors. And as a trial sponsor, that’s what I want on the other side. 

I want to be able to find those partners. And to call such partners CROs, you know, just to me feels wrong because that’s just not the terminology that we used for what I’m referring to. This is some future looking thing that combines functionalities that have never existed in a CRO to date. And so, I suppose Vial could be the first of those ever. But you know I am just as game for just inventing some new term for what I’m describing.

Simon Burns: A really clever name let us know. We’ll be sure to adopt it. But totally agree with you. I think the bar’s low for data and having systems talk to each other. There’s certainly a lot to be done there. Okay. Segment for you overrated, underrated, I give you a topic that’s in the news, very well talked about in the oncology circles. You tell me if it’s overrated or underrated in terms of its impact in the future of therapeutics development. First off, machine learning, AI, deep learning. Various approaches here. Tech bio is the new hot kind of category dujour. What do you think about their impact on the future of therapeutics development?

Dr. Keith Flaherty: So machine learning and AI in drug development , definitely overrated. I mean in 2022, lots of promises made, nothing delivered. And that goes to first principles in terms of even just the sort of learnings within the chemistry, medicinal chemistry as in chemo informatics. Component of a life cycle, although I truly see that as being where there’s probably going to be nearest term opportunity. Understanding cancer cell biology in a way that actually allows us to crystal ball novel points of intervention and develop next generation precision medicine principles. Definitely overrated . Not even remotely there yet in terms of actually translating a single observation forward and having clinical validation.

But something that was intuitive from AI or machine learning. Umm is gonna move the needle. That’s gonna change. I mean, like no doubt someday. It just as you’re asking the question right now, overrated.

Simon Burns: Great controversial opinion. I like it. Circulating toward DNA. Obviously different approaches for different types of tumors. But I’m curious, what your lens is there. One is a diagnostic, you know, active kind of guiding of care. The two increasingly discussions about it being used in clinical trials either as an end point or an early signal, an observational additional end point in trials. What’s your sense of ctDNA?

Dr. Keith Flaherty: Yeah. Massively underrated. So circulating tumor DNA of course has been subject to academic work now for years. But wow, have we just started to see the beginnings of our ability to actually leverage that. And monitoring disease like response to disease to treatments in the medistic setting, let’s say. And understanding well not just surrogacy for conventional outcomes, which I think could be very powerful. But also in tumor evolution and resistance mechanisms, I think that’s also going to be important.

But what I really am focused is in the post operative setting. So now with very, very sensitive methods for actually being able to detect circulating tumor DNA in patients who are in the post op setting, AKA adjuvant. Adjuvant, you know, was like I’ve been saying for years before we had circulating tumor DNA to my patients, which is adjuvant just means we’re not smart enough to know whether you actually have cancer or you don’t. All right. But we’re getting smarter.

Now these assays aren’t so sensitive that we can completely sort patients who have microscopic residual disease or MRD or not. But with the technologies definitely advancing. And so my point is we are on the verge right now of defining a new cancer stage for the purposes of clinical trials, which is patients basically with solid tumors who have curative intense surgery, but are MRD positive, you know, two weeks later. Those patients need therapy. They actually still have cancer. We know they still have cancer, right. So this is an adjuvant like you might or you might not.

And the efficiency of conducting clinical trials in that population and be able to really score clinical benefit and not treat patients who don’t need to be treated in the way that we’ve classically done in the adjuvant setting. That is to me just a huge moment in the field. And it’s brand new. I mean, literally this is we’re talking about like a new category of clinical trials, different than quote unquote full blown metastatic disease. or overt metastatic disease. It’s radiographically slash clinically evident. And out of this kind of dark ages of adjuvant therapy, which has just been an incredibly inefficient way of actually trying to guide patients to therapy.

Simon Burns: Last question. You’ve certainly been a huge driver in continued to get the shift towards bio techs driving innovation. Are we in the early days of bio techs driving innovation? Where’s the dial go in between pharma and bio techs driving innovation? We’ve seen both sides now.

Dr. Keith Flaherty: Yeah. I guess it’s hard to choose an overrated or underrated. I’d say my sense, mind you I live in Cambridge and I work in Boston. I think we generally feel like biotech companies are the driver of innovation oncology. Right. That’s just like, you know, uncontroversial. And my strong sense is that in the 22 years that I’ve been in oncology that’s been the nature of the beast.

I mean, innovation, risk taking, new approaches are adopted in [00:25:00] biotech. Right, and kind of try it out and honed, refined and over time. And then become standards just broadly across the field. And certainly in the hands of large pharma companies the same. But when you ask that question about biotech companies, this is not just therapeutic developers. It’s not just diagnostic companies. And data and capturing data and optimally utilizing leveraging data , that’s its own super broad area.

But what I’m trying to describe ultimately is an ecosystem that actually combines all those things . Now, in what instances can you say that that can and should be combined in one company, right? Shouldn’t there be therapeutic development companies that are also diagnostic companies? Yeah, I’m talking about small bio techs here, like nimble small bio techs. And also have their hands on large amounts of data. That’s probably getting to be a little too much to expect all domains to fit under one roof, at least, in the immediate near term. But but I think it’s absolutely critical that you have to be able to create that triangulation.

And so the question is just, what are the company units that are needed to be able to do that? Is that just two companies that can cover those bases collaboratively? And obviously, I’m thinking more of a hub where you’ve got one company that is a hub and around that hub you’ve got the spokes feeding out to a bunch of therapeutic and diagnostic developers that create one new ecosystem. I think we’re just really starting to have a view on how to create some real efficiency here. But man, we’re coming out of an era of just enormous inefficiency when it comes to the siloing of these components.

Simon Burns: Well, Dr. Flaherty, it’s been a total thrill getting to know you, having you advise us on building Vial and the conversation today was great. Thanks so much for jumping on.

Dr. Keith Flaherty: Appreciate the time.

Contact Us

By submitting, you are agreeing to our terms and privacy policy
This field is for validation purposes and should be left unchanged.