Jeff Elton (00:01):
Welcome to the ConcertAI Podcast. This is Jeff Elton, CEO of ConcertAI. Today, we’re being joined by Luke Dunlap. Luke leads ConcertAI’s Digital Clinical Trial Solutions. During this session, we’re going to cover a little bit of the experience and the transformation about how clinical trials are being conducted, some of the newest technologies and the value that’s being realized and how we can transform and actually accelerate biomedical innovations over the course of the next two to five years.
Welcome to the podcast room here in Cambridge, Massachusetts. My recollection is that your early professional career was actually in biomedical research, and I’m always wondering what’s really the motivation to get in the field, the attractiveness. Actually, you’ve stayed close to research and clinical development ever since, but maybe just take us back to the motivation, the origins.
Luke Dunlap (00:57):
Thank you, Jeff. It’s really nice to be here and a part of this. Science has been a part of my background and a part of my history since I was a child. I’ve always had a passion for biology, for science, in general, for chemistry. I did my undergraduate education at Purdue University, got a biochemistry degree from there, did my advanced degree at the University of Florida with a degree in molecular and cellular genetics. So it’s always been a part of who I am.
My grandfather died when I was in high school of esophageal cancer. I remember that was a pivotal point for me. It wasn’t so much of a calling for me to be in oncology as it was to understand how the science relates to healthcare and to patient’s lives and what could you do with the knowledge that we had to change people’s and patients’ lives. So it’s always been really a part of my background in that way.
When I went to graduate school, I actually learned more about myself and, really, it’s that intersection of data and technology with the healthcare that I really love to do. So I’ve made a number of decisions around my career around starting up new technologies when I was more junior in my career working with big consulting technology organizations so that I could learn the technology. Now with ConcertAI, I’m super excited to be leading this program for clinical development technologies today.
Jeff Elton (02:19):
So Luke, I can see that those were some super powerful impressions, and I really like your point about the intersection between data and technology because clinical development and research oftentimes is about that acquisition of the data, but in a way that you can have confidence in the data and it actually can be used and progress for regulatory decision, evidence generation. I want to spend a little time on that and I want to take advantage of the experience that you have in the industry.
So maybe what you can do is talk a little bit to me about there were an array of different technologies that were really the mainstay of doing clinical trials and clinical development in the industry, and they evolved over probably the better part of a 25, 30-year period of time, and some big manufacturers and some big technology-centric consulting firms brought those forward, but as you think about that, what really worked well about those and what were some of the limitations, the impediments, and maybe the burden that they may have brought?
Luke Dunlap (03:20):
Yeah. I think there’s a lot to unpack there. I think in a pre-COVID perspective, you had a patchwork of technologies. It was a very innovative space, right? You had the opportunity to get capital and build new innovations, but those mainstay technologies and clinical trial management systems, IRT systems, practice management systems, eCOA, eConsent, those were the mainstays and there were small number of very large players in the industry. They weren’t particularly motivated to work with each other. They didn’t work particularly well together, and it created a lot of issues, frankly, when you ran international studies as well because it didn’t work with everybody’s rules and regulations, and it didn’t work in certain languages.
So it created a whole lot of issues, but there wasn’t that big push to say this has to change, right? There was, like I said, a lot of innovation in this space, but things like decentralized studies, which were called virtual studies pre-COVID, those were things that were relatively immature technologies. Using new devices for ePRO, the concept of long-term consenting for patients was really more of a tissue bank type of thing. So these technologies were there and innovating, but the push to do it came hard and fast with COVID.
Jeff Elton (04:34):
We have a tendency in these fields to use a lot of abbreviations. So ePRO is patient reported outcomes, eCOA is clinical outcomes assessment tools. So I just want to make sure because we’ve got broad audiences that sometimes come in into this area, but I want to come back to the pandemic because the pandemic was like the ultimate crucible in stress test. You said something a moment ago. You started talking about a patchwork was the term you used of different technologies. When I think about patchwork, I used to think about, boy, there’s effort being placed on somebody or some entity to make the patchwork really work as a system. I mean, that’s what gets evoked. So when we think about what happened in the pandemic, what worked, what didn’t work, what accelerated into practice that may have just been used in more nascent pilot scale before?
Luke Dunlap (05:31):
So I think you’re exactly right, right? When we talk about a patchwork, I have the same mental picture in my mind of how do you make one system work with another systems that actually touch the clinical sites, systems that actually touch the payment systems and the billing and some of the things are on site management and some of the systems that you have around the analytics, the submissions. All of those things did require a lot of effort and it required a lot of expense, to be honest with you, by the sponsors and by the contract research organizations to be able to pull those things together.
They did work. That’s why there were a small number of players in that space because even though they weren’t perfect, they weren’t very innovative, they worked, and when they could bank on them, you’re going to use something that you have a lot of trust and confidence in as opposed to more innovative technologies that could actually probably do things better.
When you get into the push and the complete change in direction that happened with COVID, all of that had to change because, first off, the protocols weren’t actually designed to be able to support anything other than patient visits to sites. The concept was virtual clinical trials was there. The concept of decentralized clinical trials wasn’t.
Jeff Elton (06:41):
What’s a decentralized clinical trial? I mean, I could make something up when I listen to the word, but what does it really mean to be decentralized?
Luke Dunlap (06:48):
Fundamentally, it means that you’re designing a protocol that is going to fundamentally reduce the burden on the patient. That’s why 30% of patients drop out of clinical trials because of the burden to them by making things easier so they’re not traveling to sites. They can do things within their home. They can take and provide information about their patient reported outcomes over the phone or through technologies to capture it. So it was never designed to be a complete replacement for going in. That doesn’t work especially for oncology, but it does mean that you’re simplifying the process through the use of technologies and allowing them to have a lower burden to participate.
Jeff Elton (07:29):
So Luke, what I’m hearing from you as I’m listening to this, that the study designs themselves actually have to fit with the technology because certain technologies couldn’t be done with all study designs. I want to come back to that as we have our conversation, this connection between the study design and the technologies, but the other thing I heard from you is that trial participation was hard. It’s always been hard, but it may have gotten even harder during the pandemic, and this idea of burden, however burden gets defined, may be a big differentiator as to whether or not a patient’s even going to be willing and can sustain that participation.
When I listen to your description, I’m acquiring the data in the home. It says to me that certain studies and therapeutics that probably works really well for in certain other studies in therapeutics, it may not work so well. Where did we learn it worked well and where did we learn it didn’t work so well?
Yeah, great question. Going to your original point about the protocol design, April of 2020 actually came to fruition. You had hospitals not doing anything but non-elective procedures. You had clinics closing and shuttering, which means patients couldn’t be compliant to the protocol because they couldn’t get in the office to have a visit. You had the investigators and doctors themselves actually being repurposed to COVID wards. So you had all parts of the clinical trial just came to a halt. You had clinical monitors that couldn’t go and visit to the sites.
So there were so many different aspects of this that broke. For studies that required less insight visits were the ones that actually stayed open. It was easier to replace a visit with a Zoom call as long as they could make the adjustment to the protocol, but make no mistake, there was 75% decrease in enrollment in patients in cancer alone over 2020 and 2021. So it was a huge impact.
I think the things that didn’t work were the legacy technologies and their ability to support different modalities of data capture. I think that’s a big thing that we’ve seen over the last two or three years. The whole area around decentralized studies is so that you can use simple, easy, fungible technologies to piecemeal your clinical trial and data capture together that doesn’t require you to go in, which is the fundamental break at COVID.
Jeff Elton (09:56):
I’m hearing that these may have been less mature technologies, but because we couldn’t work the way we used to work, they became adequate enough in certain categories of studies to do the data collection, and we learned maybe how to make them a bit more dependable on the data collection, but the data I could collect was bounded by either something I can do in a teleconsult, observable collection or I’m assuming a remote device cloud type infrastructure.
You made the comment that oncology trials went down by 75%, and I’m not completely surprised by that. When you think about it, oncology oftentimes requires phlebotomy and imaging to understand whether you’ve got tumor response, et cetera. So I would imagine there were some barriers around that. So when we started returning back, now people going back into a clinical settings and patients were coming back, I saw there was a press release between Bristol Myers Squibb and ConcertAI that talked about a new digital clinical trial process. What was learned from that prior period that’s now being applied to the next generation of oncology trials? How has that been informed by both the legacy and then the more recently used decentralized and what I’ll call lighter weight clinical development technologies to now start informing that next generation approaches?
Luke Dunlap (11:28):
For the BMS announcement for digital trial solutions, we couldn’t be more proud to have done that, and that’s a culmination of a lot of work over the past two to three years directly over that same time span that we’re talking about that evolution that’s going on in clinical development over the COVID period. I think the drivers really were straightforward is that going forward, trials had to be designed in a different way. So we start there with the protocol because that’s how you have to conduct the study and that’s in the first place.
Jeff Elton (11:57):
So you’re rethinking the design of the trial itself.
Luke Dunlap (11:59):
The very nature of the design of the trial, that’s right, and to use not just remote technologies, but technologies that the patients already have. They already have a phone. Stop building new websites and things that don’t always scale and don’t always work internationally. Everybody has a phone. Everybody can use data transfer, the use of fungible, simple, easy technologies that don’t cost a lot and can support decentralized patient visits and subsequently allow you to conduct the trials much more simpler than before.
Digital trial solutions is actually designed around that mentality, but the key for us is around how do we use that information in the first place so that we can identify the right patients. We’re predominantly in oncology, so our patients have to be in a particular place. Maybe it’s a particular line of therapy, but there’s a lot of complex inclusion and exclusion criteria.
Jeff Elton (12:55):
This sounds like you’re doing work right in the workflow of the care delivery process.
Luke Dunlap (13:01):
So Jeff, you’re right. Clinical development technologies was designed to work exactly within the clinical workflow of the provider organizations, and it’s about allowing the providers, the investigators, and ultimately the sponsors as well with a very comprehensive detailed view of the patients that are part of that clinical trial so that they can match them against literally hundreds of inclusion, exclusion criteria and be able to assess which ones are most suitable to actually start the trial in the first place.
Jeff Elton (13:31):
I hadn’t really thought about it the way you’re saying it, but I have lots of stakeholders here, so I have the provider research site, I have the patient themselves, and I have the sponsor. So if I’m the provider research site, what’s different for me and why would I be pleased or happy with a newer alternative approach?
Luke Dunlap (13:49):
Yeah, that’s right. The provider sites have clinical research staff that have to do a lot of very manual, very granular work, and that is to ensure that the data captured around the patient, whether that’s from the EMR, from the patient visit itself, the labs from prior visits, each time that patient comes in, all of that information that’s required has to be pushed over into eCRFs or case report forms. Those case report forms take up a lot of time.
So part of working within the clinical workflow is our ability to use technology to automate that. So we can pull that data from multiple systems, we can use that from both structured data, unstructured data. We can pull information from centralized labs. We can pull information from genetics and next generation sequencing panels, and we can put all of that together. So we’re reducing the burden on the site because they don’t have to go do that manual anymore. We’re actually doing that. We’re pulling that data together, we’re presenting it to them. So the clinical research staff is literally doing a validation of what we’ve done and a submission of what we’ve done instead of doing all the labor itself. So it really simplifies that.
Jeff Elton (14:59):
This sounds like I’m doing an intelligent automation. I’m recognizing the normal work that that clinical team would have to do, but since I know technology, I can make the technology work and take the work away from the clinician, but they’re still in control of the key decisions in that workflow, in that process.
Luke Dunlap (15:21):
That’s right, and that’s just for the execution of the patient visits and schedules and forms themselves. There’s an additional burden on the site as well for around patient identification. If you think pre-COVID, you’re effectively relying on the memory of a good physician, “I know these patients. They sort of fit. I want to go look at them more closely. I’m going to do a clinical review of them and I’m going to identify the ones and list them an enrollment.” For digital trial solutions, we actually do that for them. We’re able to digitize in machine executable languages that decompose the inclusion, exclusion criteria of a protocol and are able to apply that to the EMR records of the patients, as well as the unstructured data and all the other things we just said so that you can actually create an eligibility score. So that way, you’re not just looking at one or two or three patients where you think they might fit. You’re actually screening all of your population.
Jeff Elton (16:17):
So you are covering 100%. You’re covering every patient and doing the screening so they can’t miss them.
Luke Dunlap (16:22):
Everybody’s got a score. There’s going to be obviously patients that are higher eligibility, much more suitable, things that the physician investigator would’ve already known to go get.
Jeff Elton (16:31):
This is all patients, all trials?
Luke Dunlap (16:33):
All patients, all trials. The method works the same, but it really allows you to get into that gray area of patients where you think they might or you wouldn’t have thought to associate them with this study or there are things that maybe they are missing labs or maybe they’re missing values. It doesn’t fail those. It puts them into wait lists. So you can actually take the time and say, “I need more information about these patients,” but they stay on your forward looking radar so you have a healthier pool of patients that are available for that clinical setting.
Jeff Elton (17:04):
Sounds like I have an intelligent digital CRA that just never sleeps while I’m working in this particular process.
Luke Dunlap (17:10):
That’s right.
Jeff Elton (17:11):
So let’s go over to the sponsor for a moment. So this sounds like if I’m the research site, I’ve got this integrated, I think you used the word cockpit sometime ago around that, and I’ve got this confidence and assurance that I’m not missing people. There’s a lot of work. You’re making that work a lot easier so I can have more confidence to engage in research, but what’s different as a sponsor because they have some very demanding requirements, obviously, regulatory consideration being, what are they seeing different?
Luke Dunlap (17:43):
The sponsors benefit in a lot of different ways and in very different ways than the sites themselves do. The sites benefit from organization screening, automation of workflow. Some of the key ways that are most important to sponsors are really acceleration of the clinical study itself. You can use fewer sites for that clinical study.
Jeff Elton (18:02):
Fewer sites. why?
Luke Dunlap (18:04):
Because the sites will overperform.
Jeff Elton (18:06):
What does overperform mean?
Luke Dunlap (18:07):
It means they can enroll at higher rates than what was traditionally allowed for them.
Jeff Elton (18:11):
Okay. Then without, if they didn’t have the technologies, the infrastructure available to them.
Luke Dunlap (18:16):
That’s right, and this is the state-of-the-art in a lot of the industry today is you’re using retrospective data to look at where patients have historically been. We are looking at realtime information or near realtime information about real patients that exist at sites today. So there’s less guesswork. You’re not going based on historical performance. You’re going based on what those records actually say today, and that’s a powerful differentiator for digital trial solutions.
Jeff Elton (18:42):
So Luke, I want to go back to this concept that you brought up at the very beginning of the podcast of patchwork. I’m always struck by that because I’ve got this visualization of my mind of just gaps and things and manual force. So how does that play out in the new model?
Luke Dunlap (18:55):
Yeah. Part of that is, really, we talked about all of the time and effort to keep all of those individual threads bound. With this, that is eliminated because you’re actually getting a mirror image of what those sites, what your study is capable from not just a patient potential, but the quality and the performance of those studies that can actually give the sponsors a better landscape to make decisions on what sites they go to, how they use those sites, and how they can use the composition of sites to meet their diversity goals for the study itself.
Jeff Elton (19:33):
So Luke, when we talked about it earlier on, there seemed like there was very manual and at times, as I’m listening to you, arduous labor-intensive process. In fact, you used even onsite personnel. So on this one, which sounds like for the provider side, you’ve provided a lot of automation tools and integrality to that workflow. What’s the sponsor see for the consented patients when you’re actually running that study now?
Luke Dunlap (20:03):
So great question, Jeff. I think I’d go right back to our digital trial solutions and being embedded in the clinical data workflow processes. Since we’ve automated those processes coming out of all the different source systems, the EMRs, the clinical trial management systems, the labs, all of the things that we talked about before, you’ve automated the process for moving that data into eCRFs and the simplification at the site level for submitting that data into the EDC system. Because you’ve got all of that, the sponsor doesn’t have to do the same level of source data validation, the same level of monitoring. It’s simpler for them and it’s easier. So it brings down the cost of actually operating that clinical trial.
Jeff Elton (20:48):
Wow. So I’m actually going to do my abbreviation decoder ring here again, which I do out of necessity. So CRFs are case report forms and EDC is electronic data captures, and these are just two systems of record that have been there. I want to go back to this notion of speed and acceleration here because that’s super important in clinical trials. You said something else also about fewer sites being necessary, but the sites might perform more, but I want to capture what I heard you say, and I’m not sure that’s accurate, so I want to check this out with you, is that there were activities and structure of things, whether it was source data verification, I believe you said, and others that may not need to be done. So this idea that not only are you doing the integration of the data, but you may be eliminating areas and categories of tasks because they’re not necessary. Am I hearing that in the right way?
Luke Dunlap (21:41):
Yeah, essentially that’s correct. So you’re accelerating through automation, and because of the automation, fewer human hands touch the data, which means the quality of the data is so much better, which means your obligation for source data validation is lower. I think getting back to your point of using fewer sites because those sites are overperforming, and I think accelerating the trial overall means that for accelerating the studies themselves, you have all of those things putting together to reduce the timelines, whether it’s timeline for site identifications, the timelines for site initiations, the timeline for initiating the studies themselves. The recruitment period is shorter because we have the entire patient population on a radar screen for the investigators, and you’ve got the automated data workflow-
Jeff Elton (22:29):
Well, hang on, because what you just said is super important because what I was hearing is I’ve taken the compartmentalized stages of the process and I’ve literally rethought the process, and this idea that you sometimes can’t gain benefit by just not adding technologies to an existing flawed, slow, patchworked process, but it sounds like what you did is you actually reconsidered the process in addition to the now powering that reconsidered process with new technologies.
Luke Dunlap (23:04):
That’s right. I think the only thing I would add to that is that in the same way that study technologies had to become much more flexible during COVID to account for these, we’ve designed that way from the start, right? So regardless of whether studies continue go to fully decentralized or what I think we’re all seeing in the market now is a little bit of a backtrack on that and what you’re seeing more of what we would consider a hybrid trial model, you’ve rethought all the elements of that clinical research development and trial execution process and being flexible enough to add and subtract as we need to support the study.
Jeff Elton (23:41):
So I feel like I have to ask this question. Anytime somebody rethinks, reconsiders something that’s a regulatory process that has a long history and this works. It works?
Luke Dunlap (23:53):
It works. Yeah, that was the whole nature of the BMS publication.
Jeff Elton (23:57):
Well, that’s tremendous. So I want to come back to why we’re all doing what we do. There was a third part of that process when I started that question, one being the research site and the provider, one being the sponsor, but really, it’s also about the patient.
Luke Dunlap (24:11):
Absolutely.
Jeff Elton (24:12):
So if I’m a patient, what is it about this that I care about? Why would I care and why would I say, “I’m so happy this is now available and accessible”?
Luke Dunlap (24:23):
So a lot of good reasons. I think I’ll stay with oncology first, right? Oncology clinical trials tends to be the last line of therapy for cancer patients. So getting as many participants involved in clinical research, it extends lives. So it has a direct impact on being able to do that. We do that because more patients that may have been borderline and discarded are now considered.
Now, with bringing in different aims over the last couple of years to really support stated diversity goals, you’re much more inclusive of the population as a whole. So it benefits patients that wouldn’t normally participate. It benefits patients that are frankly on their last line of therapy. So in a lot of different ways, I think this is good for the patient.
Jeff Elton (25:11):
It sounds like you have the ability to make research available at sites that may have been pretty conservative in sparing in the amount of research activity we’re able to take on. So in the same way you’re aiding the productivity, if you will, on the sponsor side, it sounds like you may be aiding the productivity on that side, which that in itself I would imagine pushes more of it to the community and to the boundary where it would be more accessible to those patients.
Luke Dunlap (25:40):
That’s a benefit to both the sponsor who can now use different sites and include community-based sites that are typically better at providing a more diverse population than traditional large academic medical centers that are the mainstay of oncology research today.
Jeff Elton (25:55):
So this last question you get to just do a personal riff on. If you take a look out over the next two, three, maybe four years at the outside and you’re in a technology field that tends to move pretty quickly, if you’re going to say the one or two things you’re most excited about that has the highest potential to move needed therapeutics to the patients with the highest medical needs, what are those one or two things?
Luke Dunlap (26:24):
The thing I am the most excited about, I think there are a few things, actually, not just one big one. I think we are ConcertAI. AI is in our name for a reason, and AI has the ability to further the game-changing efforts that we’re doing today. We’re doing things right now about predicting patient eligibility and enrollment lines of therapy. There’s a lot of really core elements of machine learning that we’re doing today that are exciting, but when I think about what’s on the horizon with the concept of digital therapeutics, digital biomarkers, a lot of these things are going to be really transformative in this space because it’s going to actually help us even make more broad the patient population pool out there to support clinical research. So I think AI is probably the thing I’m most excited about.
The second piece would really be how digital trial solutions and our technology products are really going to contribute to that overall concept of the learning healthcare system. Again, some of that is, again, machine learning and AI, but a lot of that is conquering the information and data gap that sits in healthcare today.
We are rapidly breaking down silos, and the tokenization of patients, microservice and API technologies, and I’m going into some more acronyms for you, but there’s so many things that right now that are accelerating what we’re doing and making the barrier to entry lower, the costs lower, the efficiencies higher. So I think those are the two things that are most exciting to me.
Jeff Elton (27:56):
Out of responsibility, I’m going to take the bait application programming interface and it’s how one program can talk to another program. So one is I’m really happy that you have vented on the point of breaking down the silos because everything you said so far, even the intrasite silos, were actually really important. So you’re breaking down the intrasite silos. It sounds like you’re actually taking down the silos for the collection of data. So this notion of making the data play together as a system for the benefit of the provider, the research, and obviously here the patient and actually for conducting that, that’s super important. That’s very difficult to do, and that is actually what will transform actually our ability to transform healthcare and clinical research.
Luke Dunlap (28:43):
Absolutely.
Jeff Elton (28:43):
So Luke Dunlap, thanks so much for being here in the podcast room today. Very important work. Anything we can do to accelerate needed biomedical innovations is the thing that we should be applying ourself to. So thanks so much for the time today.
Luke Dunlap (28:56):
Thank you for having me.
Jeff Elton (28:58):
I want to again thank Luke for his insights into the area of digital clinical trial technologies. We learned how the world is transformed from digital technologies that were put together in a bit of a patchwork, and how now through a more integrated solutions that were integral both to the operating model, to the providers and the biopharma innovators we’re able to lower burden on the sites, burden on the patients, and accelerated needed new medical innovations. We’re super optimistic as a result of many of Luke’s observations of achieving the 30 to 50 percent improvement in time and cost going forward.
Thanks for tuning in to this month’s podcast and we hope you’ll join us again next month. Until next time, good morning, good afternoon, good night.