Rich McCormick: Hi, I’m Rich McCormick, Vice President of Oncology here at Vial. We’re discussing some of the common challenges that exist within oncology research trials. Joining me today are two members of our Vial Oncology Scientific Advisory Board, Dr. Nathan Pennell and Dr. Brian Rini. Welcome to you both.
Let’s start with introductions. How about you, Dr. Pennell?
Nathan Pennell: Thank you for having me. My name is Nate Pennell, I’m a thoracic medical oncologist at the Cleveland Clinic in Cleveland, Ohio, where I’m also the Vice Chair of Clinical Research.
Brian Rini: I’m Brian Rini, chief medical oncologist at Vanderbilt in Nashville. I’m also chief of clinical trials for the Vanderbilt-Ingram Cancer Center.
Rich McCormick: Welcome to you both. So Dr. Rini, we’ll start with you. What are the top three challenges you encounter within your ongoing oncology trials?
Brian Rini: There’s probably a top 30, but I’ll give you the top three. Challenge number one is, and these are in no particular order, complexity. The complexity of taking from concept to written trial to first patient in, through last patient, and publication is an enormously complex multi-step process. It’s more complex than it needs to be. But it’s definitely complex.
Number two is the human resource required to execute that. Which is difficult, especially post-COVID, but it’s always been difficult. It’s only been made more difficult, so finding the quantity and quality of people to execute that process is exceedingly difficult.
Number three is, more than ever, we need to be selective in the trials we’re doing. Gone are the days when we’re just going to open any trial we want and do whatever we want. We don’t have the resources to do that. It’s getting investigators to understand, and to be disciplined about only doing the trials with the highest impact, the highest credit, the highest benefit, the best science, et cetera.
Rich McCormick: Thank you. Dr. Pennell, would you like to add?
Nathan Pennell: I would second the complexity of the trials. The regulatory burden upon trying to get trials open with the time it takes to get a trial from when you want to open it to opening it, in terms of, generating budgets, negotiations, dealing with legal contracting. While it’s opening, all the complexity of the toxicity monitoring, queries from CROs, dealing with constant flow of amendments that sometimes seem like new trials every few months which stress the personnel who are already somewhat constrained.
Once you have a trial open, one of the biggest things is then identifying the patients and being able to accrue them to the trial. We still need to do a better job of identifying people and being able to properly screen them. It takes a lot of time and personnel as well. Especially, when it ends up either not accruing and then you’ve wasted everyone’s time. Or, you’re constantly getting screen fails, which wastes a lot of resources without helping anyone.
Rich McCormick: Insightful. Staying with you, Dr. Pennell, to what extent does the volume of ongoing oncology trials influence your decision-making process when it comes to evaluating treatment options for your patients?
Nathan Pennell: In most academic centers, especially in the cancer field, we’re relatively unique in that we think clinical research is part of standard treatment for patients with cancer. Often enrolling on a clinical trial is the best available option for someone even if there is a standard available treatment.
We try to have clinical trials available for almost every relatively common patient situation we see. I’m a lung cancer doc, so we try to have something available for most of the first line patients we might see across different stages of lung cancer. But, the volume of trials is limited, that any one place can open.
We have to also be very cognizant of opening the right trials. Not spending a lot of time and resources on studies we’re not going to be able to accrue to. That doesn’t allow us to spend enough time on the trials that really would be important options for patients.
Rich McCormick: Dr. Rini?
Brian Rini: I totally agree. It’s what I was alluding to earlier. I’ve never been in a clinical research environment where we can just do everything. At least in 25 years that hasn’t happened. So, I’m guessing it’s probably not going to happen in how many ever I have left. The hard part is saying, “Okay, we’re going to prioritize trials, this is the best science, this is where we’re the most involved, and we get the most credit, and this is what the patients needs are.”
That complex equation, we probably all have equations in our head of here’s what we want to do and here’s what we can do and how do we marry those? The more you have this, in my opinion, unrealistic view of, “Well we’re going to open every trial that comes along because we want to have a trial for every patient.” You’re not going to end up doing anything.
You’re going to get bogged down in trials. By the time you open them, those trials will no longer be relevant, you won’t accrue patients, and you haven’t helped anybody. It’s complex. I’m not at all saying it’s easy. Especially now with COVID and resource restrictions, that’s especially a cue to get that right.
Rich McCormick: Do either of you encounter patients you cared for a while who are maybe interested in a research trial they found online, that isn’t offered at your clinic? Do you then play a role in helping connect them with where that trial happens?
Brian Rini: Absolutely. I want patients to go on trials period, full stop, right? If it’s at my place, great. If it’s down the road at one of our competitors, that’s fine. Ideally it’s with me at my place, but honestly, sometimes as patients, especially as they get more refractory and more maybe phase one or they have specific genetic mutations, again we can’t do everything.
Even though we may compete against hospitals on some level, we all have the same mission we want to take the best care of patients. Nate, and I, and many others, we wouldn’t be in these positions if we didn’t believe trials were the best option. Or, at least, a very good option for most patients. I have no problem calling somebody down the street and saying, “Hey do you have a trial for this patient?” It works both ways.
Nathan Pennell: I completely agree. Getting the best option for the patient is the most important thing . Whether it’s an option here or not. Cancer is so vulcanized, now, into different diseases. There’s not just one trial for non-small cell lung cancer, for example, or renal cell carcinoma, there’s multiple different sub-groups with bio-markers which might help guide a patient to an optimal treatment.
You just can’t have trials for every single patient that might need it. It’s part of our job to understand what might be best for that patient and help guide them to a center where they might have a better option that might work for that patient.
Rich McCormick: Dr. Rini, what sort of pain points do you face when monitoring patient safety and protocol compliance throughout the lifecycle of trial?
Brian Rini: Trials have become these 100, 150, 200-page complex documents. We have 250 open clinical trials, right? So you do the math. There’s medical school, scientific career or other places. You’re dealing with complex documents increasingly quantitatively and qualitatively complex with relatively inexperienced staff that turns over. It’s a recipe for disaster.
I don’t know what the solution is. I’d love for protocols to be much less complicated in monitoring and tox gathering. We collect a lot of data nobody ever sees or does anything with. We could go down the rabbit hole if you wanted, but, we’re doing all sorts of things whether it’s under the guise of compliance or legal. We do a lot of things that aren’t helping anybody, but, they’re almost like habit.
There’s this inertia that’s developed that protocols have only become more complex in my time. I’ve never been around when there’s been less complex protocols and things have gotten simpler. There becomes this tipping point where they’re so complex they’re almost impossible to execute.
Rich McCormick: Dr. Pennell?
Nathan Pennell: This is such a huge, important topic. The truth is almost every stakeholder agrees clinical trials are un-necessarily complicated. The FDA says we don’t need all this data. They’re putting out support for trials like Pragmatic-a lung, which really is collecting one end point, which is overall survival and only serious adverse events related to the treatment.
That’s all the data anyone needs to enter. If that really ends up being a successful trial, I hope that will lead towards a movement of de-escalating complexity on studies. Now if you want to write a NCI-sponsored cooperative group trial, you get feedback on if your eligibility is reasonable and just related to the outcome or the safety of the patient and not un-necessarily complex.
They want to minimize complexity, but once you get to a registrational pharma-run trial, it’s just ridiculous the level of complexity that they have. Not just the protocol that’s 300 pages, but also the manuals that come with it and everything gets amended every sixty days or so. It’s difficult to keep up with this. It leads to tons of deviations. When you finally get someone who knows what they’re doing, they turn over and you have to train someone new, which leads to a lot of other repeat problems again. So, yes, as I said, we could do an entire podcast easily on the complexity of clinical trials [laughs].
Rich McCormick: That’ll be the next one. We’ll definitely have to do that. [laughs] As Vial is a tech-enabled CRO, the next few questions are going to move into the technology aspect of a trial. Dr. Pennell, we’ll stay with you. What technology gaps do you observe trial management? What solutions do you believe could effectively address those issues?
Nathan Pennell: Medicine and clinical research is a weird hybrid now. We have all these fancy eMRs. Everything is web-enabled. But we’re really constrained by patient privacy laws, individual propriety software, and EMRs that don’t want to talk to one another. Everything should be smooth, and integrate so perfectly. You should be able to pull all the data out automatically and verify things.
Ultimately, we end up printing out pages of EMRs and showing them to someone over a webcam for source verification. The different not inter-operable software. The fact every portal you work with on a trial is a different software company, I have to retrain on it over and over again. Our personnel have to retrain on things. Every time we want to use a device, we have to reinvent the wheel.
Our IT has to come in, make sure it’s not some sort of corporate espionage trying to steal people’s data, which it never is. And yet, it still takes four months for them to sign off on it. There’re a lot of opportunities for standardization and working inter-operability of technology in clinical research that would just vastly simplify running clinical research.
Brian Rini: With all our data in the EMR, a trial should be able to pull demographics, labs and all sorts of things, even path report. But, pulling natural language stuff is difficult even if it’s, a discrete field like a lab, it seems to be inordinately complex. Someday, a patient goes on trial and everything that’s automated and standard is already pulled.
It’s really only the new, more complex data that needs to be put in. But, we’re a long way from that as far as I know. There is technology out there but it’s let’s just say, far from common use. But that would cut down on the errors and the deviations and all that stuff that Nate talked about.
I’m sure there’s a hundred examples of this, but the other one is toxicity capture. Patients should just have a tablet and they say, “What symptoms have you had today? Diarrhea. How many times?” It should just take them through, straight from the source. And then it’s automatically uploaded into some EDC as opposed to me asking them in clinic, my nurse overhears it, she writes this down, it goes here, into my notes, and the data person interprets my notes, which are very brief.
There’s 18 layers from the patient’s mouth and reality to the data. I’m sure tox data capture on trials is not great. When I see these tox tables, it’s with a huge grain of salt. Because you know what goes into it, right? [laughs] The quality of data that goes into it. Again, it’s eliminating all that redundancy. There’s so much redundancy in clinical research. It’s just habit. We don’t know how else to do it or we haven’t invested in the technology to do it.
Rich McCormick: So final question, Dr. Pennell, when considering oncology research, how might the incorporation of advanced analytics AI, or machine learning solutions improve both efficiency and accuracy of trial outcomes?
Nathan Pennell: If you think about it, almost everything we deal with is large data sets. Clinical data, laboratory data, imaging data, genomic data. We’re just picking through the little bits we think are important and reporting them out. This should be a perfect situation for AI and machine learning to be able to solve all the problems of identifying patients that may be eligible for a clinical trial based upon what’s already in there.
If they had natural language processing, and could read the notes, the path reports, the imaging, and tell tell if a patient is eligible and pop up and say this patient should be enrolled on this trial. Your CROs could integrate with the EMR so all of the data is coming directly from the source documents, into the data collection with no middle man in between that needs to be monitored.
Technically, the imaging is just data. They could monitor whether it’s truly responding or not without counting on me with my little calipers on a screen measuring, then writing it down with illegible pen, crossing it out, and initialing it. There’s so many things that go into clinical research people don’t understand quite how manual it is. There is no need for it to be manual if we had sufficient power to be able to collect and use all of that data. There’s a lot of potential. Some of what I’ve seen with AI has been so exciting and promising. I feel we’re closer than I might’ve thought we were a year ago. But still a lot of work to be done to make that actually work.
Rich McCormick: Thanks, Dr. Pennell. Dr. Rini?
Brian Rini: I’m no AI expert, but I agree. There’s just these huge gaps. I hope we look back in a few years almost like we’ve looked back at blood-letting medicine hundreds of years ago and think, “how did they do that?”
Nate’s right. We’ve measured tumors with this little electronic thing. The edges are fuzzy, we write it down. It’s like I talked about with toxicity, I tell my fellows there’s nothing objective about an objective response. It’s human eyeballs measuring fuzzy things on a screen and adding them up and doing the math. It’s kind of silly, right? It’s really silly.
All the data’s there. It’s all there. It just takes a higher order function to analyze it automatically. We can focus on the things that really matter, right? We spend a lot of our collective resource on these redundancies which don’t require an MD license to measure tumors. It’s not really a doctor function. It should be an automated thing. But it’s not, so we take a whole lot of people’s time and effort to do that. When if we could automate all that stuff, we can spend our time, thinking about new treatments and biomarkers. We could spend our time with the really creative stuff that we should be.
Rich McCormick: Dr. Pennell and Dr. Rini, thanks for taking time today to talk about the challenges that you’ve faced when running oncology trials at your sites. I hope you have a great rest of your week.
Brian Rini: Thanks.
Nathan Pennell: Thanks so much for having us on.
