Prevention and early detection trials have been especially vulnerable to being disrupted by the COVID-19 lockdown, and a comparison of two regimens for monitoring pancreatic cysts—EA2185—was more vulnerable than most.
“If you wanted to design the perfect study to be interrupted by COVID, this is it, because pancreatic cysts are not an emergency,” David S. Weinberg, a gastroenterologist at Fox Chase Cancer Center and the principal investigator on the study, said to The Cancer Letter.
“So, if patients are less likely to be going to their doctor because of COVID-related concerns, particularly a year ago—as we all know, there was many fewer visits to doctors, many fewer visits to emergency rooms, and many fewer incidental identifications of cysts,” Weinberg said.
“Further, if a family member called me and said, ‘Hey, I had a CAT scan, I did not have a kidney stone,’ but they told me, ‘I had a pancreatic cyst, do I have to deal with that right away?’ the answer, except under rare circumstances, is ‘No, you can wait,’” Weinberg said.
“So, in our trial, of course, we’ve got a protocol. And the protocol is that the cyst needs to be identified within the last six months. So, if a patient doesn’t get back to their doctor for six-and-a-half months after this CAT scan, because there’s nothing emergent on it, then, unfortunately, the patient can’t be a participant in our trial. COVID, of course, has also made it difficult to find research staff and keep research offices fully humming.”
A video of the conversation is available here.
Weinberg spoke with Paul Goldberg, editor and publisher of The Cancer Letter.
Paul Goldberg: Dr. Weinberg, thank you for agreeing to talk with me about this. Why don’t we focus on the whole series of scientific questions arising from your trial, which is EA2185. What is the main question that you’re trying to answer?
David S. Weinberg: The main question is how to optimally take care of a very large group of patients, who I think are, for the most part, under-appreciated.
Pancreatic cysts are extraordinarily common, particularly in older people. The overwhelming majority will never cause any of them a problem.
However, there’s a small fraction of cysts that over time will turn into pancreatic cancer, and pancreatic cancer, obviously, is a terrible disease for which we have, at this point anyway, no way to reliably prevent or, some might even argue, diagnose early. So, there’s a small percentage of the population, patients with cysts, maybe patients with Type 2 diabetes, whom you could target as a group who are at higher risk to develop cancer.
So, there are a variety of recommended surveillance strategies for pancreatic cysts, and our trial is comparing the two major strategies. One is based on the so-called Fukuoka guidelines. And the other is based upon the guidelines promulgated by the American Gastroenterological Association. In some ways, the recommendations for surveillance are the same in that both focus on the available tools we have today for surveillance, and that is mostly cross-sectional imaging supplemented by endoscopic ultrasound in selected patients.
The difference between the two surveillance strategies rests upon the intensity of surveillance: How often are MRIs or CAT scans of the abdomen recommended, what characteristics of the cyst need to be present to recommend endoscopic ultrasound? In our study, in part to avoid confusing anyone by saying we are rigorously and completely following the Fukuoka guidelines or rigorously and completely following the AGA guidelines, each of which has strong adherents, I might add, we’ve instead called one arm of the trial high intensity surveillance, similar to Fukuoka, and the other low intensity, similar to the AGA.
Bottom line is, if you’re in the high intensity surveillance arm, you’re more likely to get cross-sectional imaging more frequently, you’re more likely to get an endoscopic ultrasound. What we purposely are not doing for this trial is comparing indications to proceed with surgery. That’s a completely different issue from surveillance, and we’ve adopted the standard changes either on CAT/MRI scans or endoscopic ultrasound that would prompt surgery and held them equivalent for both arms.
I see, but to establish kind of the stakes of this thing, the surgery is the Whipple?
DSW: Right. So, the reason this matters, once you get past the practical aspects of surveillance, is that, frankly, neither of these surveillance strategies is great. And that’s really the problem, because what you’re of course trying to do is identify patients at high enough risk for cancer to justify operating on them.
The problem is that you don’t want to operate on people who aren’t going to benefit, namely patients who don’t have cancer, or, unfortunately, patients whose life expectancy will not be meaningfully changed by surgery.
You don’t want to operate on more people than you need to. On the other hand, as we were talking about a few minutes ago, there are very few ways to prevent pancreatic cancer. So, if you happen to find that group of patients who will in fact benefit from surgery, patients with presumably either early-stage cancer, that not only is resectable, but is resectable in a way that five-year survival after surgery is meaningful.
Unfortunately, that’s still a small group even of resectable patients. We’re also considering, for our trial, a success when a patient at surgery has high-grade dysplasia in their pancreatic cyst. Nobody knows with confidence what’s the transformation rate from high-grade [dysplasia] to cancer.
But if this part of the GI tract is similar to other parts of the GI tract, high-grade dysplasia is certainly a major steppingstone to development of malignancy. And if you can operate on patients and prevent them from getting pancreatic cancer that will kill them, at least in 2021, that’s the best that we can hope for.
Can I interrupt for a second? Maybe I could ask you to summarize in a nutshell what the benefit would be from having conducted this trial. What will you know then that you don’t know now?
DSW: Right. So, the basic tension, the take-home messages. Nobody, patients or doctors, want to operate on people who aren’t going to benefit. By the same token, they don’t want to withhold surgery from patients who will.
Our ability to risk-stratify needs to improve because missing the chance to prevent cancer is a terrible thing. On the other hand, subjecting someone to a major operation with major complications, at least potentially, is also a terrible thing. So, how to balance those two poles through a surveillance strategy is the question.
We want to figure out which of the surveillance strategies works better, and better for this trial is defined clinically, meaning we have a series of outcomes, which we’re defining as either good or bad, but there also are questions of resource utilization. So, if you get twice as many CAT scans and MRIs and your clinical outcomes are unchanged, then you spend a lot of money and resources without getting any benefit.
And that also doesn’t make any sense.
Okay, but it’s also, what is the prevalence of pancreatic cysts? Does anybody know?
DSW: The prevalence of pancreatic cysts, all comers, if you’re not worried about the size of the cyst, depends on how you do the imaging, but MRI is the most sensitive. And in patients over 60, the studies range anywhere from 5% of the entire population over 60 to 50% of the population over 60.
Our trial focuses on one centimeter or greater pancreatic cysts, which are certainly less common and might be seen in two to 5% of the over-60 population.
And you also have that in kidney; right? Or not?
DSW: Well, I’m not as familiar with the kidney literature, but certainly, incidentaloma findings in various intra-abdominal organs are common, liver cysts, pancreatic cysts, kidney cysts. I suspect that for the patients and the doctors who care for those individuals, everybody wants a good way to know that this finding on the CAT scan is either important or it’s not.
So, you’re talking about risk stratification that may have implications for other organs? Or not?
DSW: No, in this trial we’re focusing only on pancreas.
Yes, but in terms of methodology [and] approach.
DSW: I think the question of how to take available tools and do the best method of risk stratification, of course, it’s a common theme across any organ system. Where pancreas, anyway, at this point, perhaps lags behind some other areas is that we don’t have good biomarkers, and one of the goals of our trial through correlate studies is to identify new and better markers, whether they are blood-based, something you can pick up in the DNA if you do a buccal swab etc.
We also have a large component of radiomics in this trial, or taking the available images and analyzing them in more sophisticated ways, all looking for markers that either would add to our current surveillance strategy to come up with a better risk stratification tool, or if we really hit a home run, replace some of those existing studies.
What is, maybe we should just do the numbers here, how many patients are you trying to accrue? And what is the budget? How much would it cost to do this study?
DSW: Well, the number of patients we’d like to accrue is about 4,600 evenly divided in that patients will be randomized to either the high intensity or low intensity surveillance strategy, and then regardless of arm, each group would be followed for five years.
This is an expensive trial. The actual budget is a little hard to calculate in that, like many cooperative group trials, it’s being conducted at, hopefully, at least in terms of places that have signed up, more than 200 sites around the United States, and we’ve recently added Canada, and we’re looking for a couple of other international sites.
There’s obviously a tremendous amount of money that could be spent as the number of participants grows, because we’re banking biosamples, we’re banking radiomics data, and those are costs that in theory are budgeted for the trial, but won’t be expended unless we have the samples.
So, what is the amount of money do you think it’s going to cost? Does anyone know this at this point, is there an estimate or a range of estimates?
DSW: I think the NCI has money set aside, and it’s certainly in the tens of millions of dollars for a trial this size.
Okay, so the numbers aren’t really known, right? Or are they?
DSW: The numbers, an interesting feature of cooperative group trials is even if you are the overall chair of the trial, which would be me, the construction of the budget is a complicated dance between the cooperative groups, NCI and NIH, I suspect.
So, unlike when I submit an R01, where I control all aspects of the budget, there are parts of this one that I don’t have any particular reason to need to know about, as long as I can be assured that every time a blood sample shows up at the centralized biobank for the trial, it’s appropriately handled.
Oh, interesting. So, who is diagnosing these cysts, for the most part ,and what kind of docs will be putting patients on your trial? These are not oncologists?
DSW: Not generally. And that’s one of the challenges of trying to do this trial through the cooperative groups. The cooperative groups are a wonderful entity that science in general and medicine in particular has benefited from for years.
But the way the cooperative groups are set up is, of course, they emphasize oncology treatment trials. So, prevention trials, or early detection trials like this, in terms of practical aspects, if you weren’t in the trial, patients are generally seen by gastroenterologists or surgeons.
Their cysts are generally identified by radiologists, and in many ways, the providers most immediately responsible for the patient’s care, or primary care doctors, because the most common scenario since these cysts are nearly always completely asymptomatic is that they’re identified serendipitously.
Somebody has abdominal pain, their primary care doctor gets a CAT scan of their abdomen looking for a kidney stone, they might find a kidney stone, but lo and behold, the radiologist picks up a 1.1-centimeter cyst in the head of the pancreas.
That information typically then goes back to the primary care doctor who has to figure out what to do about that. Generally, the response, at least hopefully, is referral to someone who is more familiar with the longitudinal management of cysts, and that tends to be gastroenterologists or surgeons. And as you’ve already suggested, generally not medical oncologists, because these patients for the most part, thankfully, will never get pancreatic cancer and wouldn’t need a medical oncologist to take care of them.
But NCTN, for example, does not really deal with primary care physicians very much, if ever, I don’t know about gastroenterologists. I haven’t heard of many being involved, and radiologists are another category, and they do trials, so how do you get them into this thing? And then the next question would be how do institutions set up the structures so the money follows the service?
DSW: Those are great questions that need better answers than the ones we have now.
Certainly, we have struggled in our trial, and I know in other large prevention trials that struggle with the basic premise of how to make sure that the physicians who direct the cancer trial programs, whether that’s in the NCTN centers, or whether that’s in the NCORP centers, as you suggested, generally those are medical oncologists as we’ve already discussed. These are not patients who typically would see medical oncologists.
I think that problem is increasingly recognized by NCI and by DCP, the Division of Cancer Prevention. My hope, and certainly all evidence suggests to date, that knowledge that this is a particular problem, at least for EA2185, hopefully will drive a variety of novel solutions. I don’t think that without a change to some of these basic premises—NIH’s shorthand for who can put patients on studies is basically called rostering—and the question is how to roster gastroenterologists and surgeons through NCTN or NCORP sites, so that they can be more active participants in these trials.
Certainly, above my pay grade is the question of how the money gets distributed from NIH to NCI to cancer centers, whether that’s NCORP or NCTN, but in general, a problem that we’ve seen in this study is a concern by some sites that the money for the trial is controlled by the cancer center, but the physicians and other investigators participating in the trial might not even be members of the cancer center. How do you fix that problem so that you incorporate more of the providers who care for these patients to facilitate trials like this.
Well, you haven’t mentioned primary care physicians. I can’t even figure that one out, how does that work?
DSW: So, primary care physicians are a real challenge.
I think that radiologists are too. Remember the name of one of the cooperative groups is ECOG-ACRIN ,and ACRIN, of course, are radiologists.
So, how to probably use information systems more effectively than we use them now to identify patients is one of the solutions that, certainly for EA2185, we want to look at and want to look at very aggressively.
Electronic medical records, in theory, allow for relatively easy searchability, at least they should, or software can be put on top of them that allows for that searchability. It isn’t difficult for me to imagine, but so far it’s been difficult, understandably, to execute, to set up a data search system for a radiology department where once a week anyone who had a pancreatic cyst on a CAT scan or an MRI is culled from the radiology records. And, of course, with all the appropriate HIPAA protections, those patient names are forwarded to the research team.
How to then approach those patients, roping in, as you said, the involved docs. If I were a primary care physician and somebody started contacting my patients without me knowing about it, I would find that potentially problematic.
On the other hand, if it’s clear that primary care doctors are eager participants in the trial, they don’t have to work real hard to get their patients in, they just have to give approval that if there’s a patient who might be eligible, it’s okay to approach those patients, then that may solve a problem.
Some places have tried to do that. I work at a smaller cancer center, Fox Chase, and it’s easier for us to search the radiology database, but to search the radiology database of a large academic medical center, that’s a lot of moving pieces, where everybody has to agree that we’re going to do it and what are we going to do with the information.
That said, that is certainly one way that we could find a lot of these patients more effectively than now, which is essentially opportunism. If a patient is referred to a doctor who’s involved in the trial, the patient might hear about the study. And generally patients when they hear about the study are willing to participate.
When does the study begin? What’s the status of it right now?
DSW: We started about a year ago. If you wanted to design the perfect study to be interrupted by COVID, this is it, because pancreatic cysts are not an emergency. So, if patients are less likely to be going to their doctor because of COVID-related concerns, particularly a year ago—as we all know, there were many fewer visits to doctors, many fewer visits to emergency rooms, and many fewer incidental identifications of cysts.
Further, if a family member called me and said, ‘Hey, I had a CAT scan, I did not have a kidney stone,’ but they told me, ‘I had a pancreatic cyst, do I have to deal with that right away?’ the answer, except under rare circumstances, is ‘No, you can wait,’ Weinberg said.
So, in our trial, of course, we’ve got a protocol. And the protocol is that the cyst needs to be identified within the last six months. So, if a patient doesn’t get back to their doctor for six-and-a-half months after this CAT scan, because there’s nothing emergent on it, then unfortunately the patient can’t be a participant in our trial. COVID, of course, has also made it difficult to find research staff and keep research offices fully humming.
That’s slowly coming back as well.
So, it’s a long way to get to the answer to your question, which is we’ve been open for about a year, but our enrollment is far below what we had hoped for, a fraction of it, honestly. And I think COVID on a list of one to 10, is problem one through eight. So as COVID disappears, hopefully this will become easier.
But as we were talking about a few minutes ago, the challenges of getting patients enrolled, at least on a large scale, in cancer prevention trials through the cooperative groups remains. And I think there is great interest and effort on the part of NCI and DCP, working in concert with cooperative groups like ECOG-ACRIN to come up with solutions for how to fix that. It’ll benefit all of us.
If not cooperative groups then who? Who the heck cares about this scientific problem, because there are no drugs involved, it’s not an industry issue, it’s a total publicly funded clinical trials question.
DSW: Right. And because the size of the trials and the budgets are way bigger than the budgets for any individual grant that most people could submit either to a federal agency or anywhere else. So, there are very few mechanisms except federal funding to mount trials of this size, to answer public health questions, which, as you put it, are not going to be ones that industry has a stake in.
The only potential exception to that, and it’s not a trivial one, is there are plenty of people, both in the federal government and in private groups who are interested in the development of biomarkers that would allow for better diagnosis.
This trial, not to blow our own horn, could put together a prospectively acquired cohort of 4,500 people or so, where we have all of the clinical data, accrued prospectively: blood, and other biosamples, radiology data that could be analyzed, and all of that was available to link back with clinical outcomes…
At this point, we don’t have a good enough biomarker or a good enough radiographic marker. Even if an industry partner came to us, there just isn’t enough data to suggest that you take a finite resource like that, meaning all of the data and biosamples that we acquire, and indiscriminately, let people use it.
I absolutely want to let people use the data we’ve got, but pancreatic cyst fluid, something that we would collect as part of this trial, is measured in milliliters, and once you use up that resource, you can’t get it back.
So, we need a mechanism, and we’re certainly discussing what it should be, to be able to prioritize biomarkers described already or described over the next couple of years as the trial rolls on, that seem to have the highest likelihood of being a winner, because no one else will have the data we’ve got. Most trials in pancreatic cysts are done in patients retrospectively. They’re almost always done in patients who go to surgery, which is a fraction of the patients with pancreatic cysts, and we don’t know anything about the ones who don’t.
So, we have, or we hope to have, a database which is really a great resource for all investigators interested in this area, we just need to get the patients in the trial.
Oh, fascinating. This really should have been my first question, but I’m saving it to closer to the end. How did you come across this issue as the PI? What is the genesis of your interest in this?
DSW: So for full disclosure, while I did not write the AGA’s pancreatic cyst surveillance guideline, I was the chairman of the AGA’s committee on guideline composition at the time. So, when the AGA’s guideline was released, it created in the world that worries about pancreatic cysts, tremendous discussion, shall we say, because there are some very strongly held views about what to do with these patients, and it’s entirely understandable why.
If I were taking care of someone and I had a surveillance strategy that I didn’t think was rigorous enough and a patient slipped through my fingers and developed a cancer that couldn’t be cured, I would feel like I had failed that patient.
The back of the envelope estimate is that we spend at least a billion dollars a year in the United States alone on radiology studies associated with pancreatic cysts. If I’m going to spend the billion dollars, I want to spend it well.
As we talked about a while ago, the AGA guidelines, the low-intensity surveillance arm in our study, is less resource-intensive. Patients do generally go longer. The recommendation for how long to wait to repeat a CAT scan or an MRI is generally longer. And that makes a lot of people nervous, understandably. On the other hand, there is absolutely zero evidence that one strategy works better than another.
So, this is, in fact, an important question at a perfect time to study it, where any investigator who participates in this trial clearly can be at clinical equipoise. They don’t know what to do. And if we don’t know what to do, it makes sense to do a trial where you actively compare two guidelines, which incidentally is a very unusual thing in medicine.
Guidelines get promulgated in all sorts of settings, by all sorts of groups and they’re never compared, or they’re rarely compared. So, we thought that this trial was important because it allowed not only the opportunity to address an important question, it goes to the heart of one of NCI’s, and that should be anybody’s, concerns about over-diagnosis versus under-diagnosis.As we’ve discussed the stakes of missing the cancer are high and the stakes of overreacting to a CAT scan or an MRI are also high.
So, getting a good answer makes sense. The ability to compare guidelines makes sense- and cost effectiveness is not a specific goal of this trial, but we would be foolish not to consider how much resources of all sorts are used here. When you think about it, if there are roughly 50 million people in the United States over the age of 60, and even a fraction of that, 2%, 3%, has a cyst that in theory requires some form of serial cross-sectional imaging.
The back of the envelope estimate is that we spend at least a billion dollars a year in the United States alone on radiology studies associated with pancreatic cysts. If I’m going to spend the billion dollars, I want to spend it well.
Right, right. You were involved in the guideline development though, not directly, of course, so, what happens next? Do you come back to work and say, ‘Hey, let’s come up with a trial and compare the guidelines?’ How did that work for you?
DSW: So I initially went to PCORI with the trial protocol, because as we’ve already discussed, there are not a lot of mechanisms to allow for such a large trial with a bigger budget than most individual grants would allow. PCORI has its own set of priorities, and they’re good priorities, PCORI’s.
I think this is a good priority too, but in their system of what they wanted to emphasize, at least when we went to them, this didn’t rank high enough to be an area that they wanted to emphasize.
They had plenty of good areas. I then started talking with colleagues who were involved in cooperative groups because as a gastroenterologist, even when working at a cancer center, I don’t have much to do with their cooperative groups.
So, it’s been an interesting learning experience to work in this framework. It gives you the opportunity to do remarkable studies, but as we’ve talked about, not just for me, because I’m a gastroenterologist, but for anybody who’s doing prevention trials, it’s clearly an area of emphasis for NCI, there’s no doubt about it, but the rules to get these trials done, meaning what do you have to do to get patients enrolled at an adequate rate, are probably different than the rules required when you compare chemotherapy A to chemotherapy B, and how to make sure that the substantial resources that are spent on prevention trials result, in a completed trial with valuable data.
I think everybody wants to figure out how to do that better, in that everybody starts from anyone at NIH and NCI down to investigators like me.
Oh, that’s fascinating. Is there anything we forgot? Anything you’d want to mention?
DSW: I think we covered everything that was of top importance.
Well, thank you so much for talking with me.
