Chapters Transcript Pancreatic Cystic Lesions: When to Worry, When to Stop? Koushik K. Das, MD, presents on the epidemiology of pancreatic cystic lesions. People still can come on. So just uh thank you so much to everybody for joining us today. Uh especially since it is a Thursday and we usually do hold these on Wednesday, but um some of our specialists uh do have clinic on Wednesday. So um that's why we're having it on Thursday. So uh also I wanted to address uh at the end of the program, an evaluation will pop up in your browser. So if you could, wouldn't mind filling that out, um That would be great. Also, I'm gonna send out an email at like I always do at the very end of the session and I'll have the link for that evaluation in there as well. Um I will also be taking attendance at the beginning and the end of the program for see me credits. So if you are calling in on a cell phone or you are using doctor doss uh link, if you could just type your name and to the chat, that would be great because then that way I make sure I get everybody, I will send those DME credits out uh at the end of the week, probably beginning of next week. If you just, if you don't see one for the next week or over the next week, just go ahead and email me and let me know. Um Let's see what else each speaker is gonna go ahead and give their talk. So if you have questions that you want to ask, feel free during their talk to type those into the chat and Q and A and then we will go ahead and answer those at the end of the program. Um And so our first speaker is Doctor Kushida. He earned his medical degree at Columbia University College of Physicians and Surgeons in New York and his residency in internal medicine at Massachusetts General Hospital in Boston. He then went to the University of Pennsylvania in Philadelphia to complete his fellowships in gastroenterology and advance endoscopy. Doctor doss is currently an assistant professor with the division of gastroenterology with the Washington University Physicians. He sees patients at Barnes West County Hospital, Missouri Baptist and Barnes Jewish Hospital for diagnosis and treatment of biliary and pancreatic disorders, pancreatic cancer, pancreatic cysts, Barre's esophagus, er CPSE us interval, stinting, endoscopic mucosal resectioning and I'm sure others as well. Those are just some highlights of what he does. So without further ado let doctor doss take it from here. Thank you. Thanks Nicole. Um Very nice to meet you guys. I'm Koshi Das. Um II I know it's difficult to be informal over the context of zoom, but I do want to make this as useful to you all as possible. So please feel free. Um It'll be hard for me to monitor the chat sometimes when I'm in the middle of the conversation. But if Nicole, if you see something in the chat and you want to interrupt on, on somebody else's behalf, please feel free. Or if you guys wanna just um blurt out something that's also totally fine. Um I'm gonna be talking to you all about pancreatic cystic lesions, which I'm sure as a PC P, my sister is a PC P. This comes up very frequently um as an incidental finding in the context of other kinds of workups. And um and so we have, you know, we, we have to deal with this in our everyday lives. And so I'll try to make this as um useful as possible. Um These are my disclosures, none of which are particularly relevant to today. So, in an overview, so we're gonna talk first about pancreatic cyst, what are they, the epidemiology of pancreatic cyst, their classification and prognosis, clinical guidelines and where the clinical guidelines have some holes. Um And how good we'll look at. How good are we at preoperative um Assessments will ask very importantly for a group of people who are, you know, in a primary care setting. When to stop, when the, you know, that when to stop conversation is important to have in all contexts, when to stop colonoscopy. When to stop, you know, um cervical cancer screening or, or breast cancer screening or so many other things. Um We'll talk about is five years enough for surveillance or do we need to go longer or shorter? We'll talk a little bit. Um I have to, you know, being at wash U and doing a lot of science and having, having a lab background, I have to sneak in some science in some way. So we'll talk a little bit about some translational insights um from what we have now learned over the last 10 years about pancreatic cyst and how this has implications for how we survey these patients in from a bench to bedside kind of clinical way. And then we'll talk about some of the diagnostics that we have. Um the tools that we use currently like C AM L A cytology. Some of the tools that are um you know, on the horizon like uh uh telomerase methylation and a biomarker. I'm developing do s one as well as things that are, you know, imminently available like next generation sequencing and how that kind of fits into things. So first talking about prevalence of cystic pancreatic lesions against something I probably don't need to reiterate just how common these things are. Um This is an old study looking at 2832 patients who underwent um sequentially went, underwent abdominal CT scan at Hopkins or Hopkins affiliate. Um over the course of a year for all comers. Um And they found essentially of everybody who showed up 2.6% of all patients are found with a cyst. And it tracked very closely with A I DS. So patients above the age of 88.7% of them were found with an incidental cyst having nothing to do with their chief complaint in replicated studies, you know, down the road and in, in a, in a big meta analysis that was published recently, a global meta analysis published recently, these numbers were much higher and even much higher with MRI S which tend to be more sensitive for small cystic lesions. And you can routinely see patients uh upwards of 30 plus percent of patients uh above the age of 80 having a cyst in their pancreas. What I frequently tell patients in the clinic is that if in autopsy studies from Japan, if you look at patients who died of non pan causes, just died, about 40% of those dead bodies have a cyst in their pancreas. And so that is probably the ground truth true incidence of of cysts in in patients. And so obviously, 40% of people are not dropping dead of pancreas cancer. And so, trying to, you know, understand how this incredibly common, you know, uh phenomenon is is associated and it is not associated with, with cancer or will will cause a problem will not cause a problem is really the central issue that we have. So, pancreatic cyst can be broadly broken down into pseudocysts which tend to be in the context of a prior episode of pancreatitis, nonneoplastic cysts, cystic neoplasm and then some hereditary associated conditions that have cysts like von Hippel, Lindau and Polycystic kidney disease. We're gonna primarily focus today in talking about cystic neoplasms because that's the majority of what you're gonna see in clinical practice. They're broken down broadly into mucin cysts and non cysts. Mucinous cysts um include intro papillary, mucinous neoplasms or I PM NS uh which uh form the, the, the lion's share of what we have to deal with in clinical practice and mucinous CS neoplasms, non mucinous cysts are cerro cyst Anoma and solids Coppi neoplasms. The reason we make this distinction is cerys Anoma um for the most part, um do not have significant uh cancer potential. And so being able to distinguish something as being definitively non mucinous has some benefits, mucinous cysts while they do have some malignant potential, um the vast majority of which will never, you know, um you know, develop into a cancer. And so being able to make the differentiation between something being mucinous and non mucous. And then within the mucous category to being able to risk stratify that as being high risk or low risk is kind of one of the central issues that we we deal with clinically. So, cyst Anoma as I sort of was um alluding to these um are histopathological aligned with simple glycogenic um epithelium. They typically occur in, in more in women and in, in older women in the seventh decade of life, they have characteristic um imaging findings that you can kind of see here with a central calcification or central scar in the top left um uh top left image. Um And they can see that this is on AC T and then this is on the MRI scan um as well. And on endoscopic ultrasound, they have this very complex honeycombed appearance uh that you can see here. Um This is an example of what they look like on excision and on pathology. You can see that what that central scar looks like on gross. And then this uh simple um glycogenic um pili um on, on pathology. And this is an example of what an E US FN A is uh when we're trying to sample, you know, one of these lesions um mucin a cystic neoplasm. Um these typically occur in younger uh women, again, female predominant, fifth decade of life. They uh histopathological have um mesenchymal um like ovarian stroma. There's no communication with the pancreatic duct and they frequently express estrogen and progesterone receptors. Um They have um some malignant potential. So, in resected cohorts, about 44 percent are benign. Um 7% borderline, 15% have carcinoma side two and 33% have out and out cancer for the most part because they're found in um younger healthier people and we know that they have an eventual development, development potential towards cancer. Um If you suspect an MC N, we typically will um talk with a surgeon about surgical resection. Um uh this is like what it looks like on gross and again, what it looks like uh pathologically in interductal papillary music as neoplasms of the pancreas um are broken down anatomically as side branch I PM NS or branch duct I PM NS which you can see here in the cartoon and the MRI picture below as being involved with one of the branch ducts of the pancreas. Um mixed type I PM NS that involve both the main pancreatic duct, which you can see here in the middle as well as in the branch duct and main duct I PM NS which are primarily um in uh primarily the main pancreatic duct itself that's involved endoscopically. You can see um at the pilla which is where the bi the biol elect and the pancreatic duct drain into the intestine. Um mucin extruding from the pan from the actual pancreatic duct. In these, in these two pictures, um demonstrating how these these um these lesions are involving the main pancreatic duct and secreting mucin into the main pancreatic duct. And in the second picture, you can actually see the frond like kind of pili growing from the main pancreatic duct and actually coming out into the, into the G I tract. Um This is an example of us kind of taking a look at one of these I PM NS under endoscopic ultrasound, you can see there's an anechoic structure in there and there's something kind of solid within there which is um uh a mural nodule or the development of an uh of likely an early cancer in that context. And what we're trying to do in these situations is both sample the fluid as well as if we can sample some of that tissue to see whether or not there's um high grade dysplasia or cancer in there that helps in preoperatively kind of assessing these uh preoperative, assessing these lesions. Um So um on a pathologic level side branch I PM N can be broken down into gastric type or pancreatic Abili type and main duct I PM NS are broken down into oncotic, oncotic and intestinal types. Those types tend to develop into a colloid uh looking cancer and the side brans tend to develop into a tubular um looking can. Um but one of the fundamental questions is how many of these cysts are actually in a plastic and how many are benign. Um This is um a data that's 10 and 10 and 20 years old. Now looking at this uh distribution of symptoms versus a uh uh not symptoms because a lot of these patients are found, you know, asymptomatically. And what we basically find is that in the asymptomatic population in surgical cohorts. So this is obviously people who have been referred to a surgeon. Um a a small percentage of patients are malignant and about twice as many uh in, in patients who are symptomatic um as opposed to asymptomatic. Um how many are actually found, you know, in the context of, of ongoing workups to have cancer? So this is data from Benin Wu from um the Kaiser Group looking at 1800 patients who were surveyed over the course of five years. Um And in that context, 53 were diagnosed with the pancreatic cancer cancer. Um 39 essentially were coincidental with the um with the initial identification of the cyst, the cyst and the cancer kind of were recognized at the same time. Same time and 14 were one, they were actually being surveyed that developed into a cancer. And so the incidence of cancer in this kind of broad group of patients from um an entire health system was about 0.4% per year during surveillance, which is obviously not a very big percentage. But if you compare it to the baseline risk of pancreatic cancer in that population, it's significantly higher 35 fold. Um SI R versus um non cystic patients. Um These patients do um seem to have a higher, you know, uh prevalence of um germline genetic mutation. This is data from a Multi Center consortium um looking at um patients with resected pancreatic cysts um and doing germline genetic testing on them. And what they found doing a um a 94 panel, germline testing panel was that 23 out of the 315 resected patients had germline mutations. Um Nine of those 23 actually were associated with pancreatic cancer susceptibility. Again, not big numbers 2.9% but more than you would necessarily expect in the general population. And why this is, you know, relevant to us is that patients who have uh one of these germline gene mutations and have pancreatic cyst may have a different um uh risk of development of, of cancer or, or, or behavior in their cyst that we need to kind of think about. So I PM NS in high risk individuals are more likely to grow more likely to develop worrisome features and have a higher rate of malignant transformation. Um as compared to people, um you know, average people or people who have just hereditary um risks for pan for pancreas cancer. So there's no shortage of guidelines uh on a subject that, you know, really, you know, we are lacking in a lot of primary data. I always joke when a new guideline comes out that it, it sometimes essentially muddies the water sometimes more than it than it helps. Um The, the ones that are probably most cited or or discussed are the International Association of Pancreat guidelines. Probably are the dominant guideline in this space. The radiologist will will hew to the ac R guidelines probably more. Um And we also have a G A uh and American college guidelines as well. I'll talk to you a little bit about the evolution of the guidelines that helps you kind of generally see what are the recurring themes in all of these guidelines just to understand what are the things that we get worried about when we see on imaging in, in a patient. So the very first set of guidelines from the IAP were the Sendai guidelines. Um and these were all the way back in 2006. And they basically said patients who had symptoms attributable to the cyst dilation of the pancreatic duct more than ac or a cyst, more than three centimeters, the presence of intramural nodules or cyst fluid cytology that showed concern for cancer should go to the operating room and you'll see these some flavor of these kind of criteria kind of continue to float through all these different guidelines. They tried to break it down relatively simply that if a cyst was less than a centimeter, then you should get MRI in a year if it was 1 to 3 centimeters and you should add E us to that, you know, make sure if it's more than three centimeters, then you should talk of, you know, resecting the lesion. Um and they tried to validate this in clinical practice. And while they, they found that this is incredibly sensitive in picking up cancers 97 to 100%. It was incredibly non specific, 20 to 30%. And so you're finding a lot in series after series of patients with benign cysts being resected, um sometimes significantly more than the number of malignant cysts that are being resected. Um you know, in these groups. So you need to try to improve your, your specificity, not trying not to lose your sensitivity. And so that's essentially what these fall follow up guidelines in 2012 were where they added worrisome features. Again, you're seeing that cyst greater than three centimeters change to the pancreatic duct caliber, mural nodules, pancreatic duct, 5 to 9 millimeters thickened or enhanced cyst walls, lymphadenopathy. And then some of the same high risk stigmata have stuck, stuck around having a pancreatic duct, more than 10 millimeters and enhanced solid component imaging or presenting, you know, with painless jaundice obviously are all very concerning that there's a cancer, you know, underlying this. And as these guidelines, you know, in every field, you know, slowly iteratively change, you see that these flow diagrams get, you know, progressively more complicated as they add more features and more more options. So the red at the top are basically those high risk stigmata. If any of those are present, then go to see a surgeon. If the worrisome features are present, which is in orange, it'll help you dictate whether or not um you should go do an upfront endoscopic or if not. If you just uh should do size gated um surveillance, but this surveillance can get pretty intense in this study. They're suggesting that if you have a cyst that's over three centimeters, you should be getting surveillance alternating with MRI and the E US every 3 to 6 months, which is, you know, an incredibly intensive surveillance strategy which even in Japan, which is where this uh much of this originated when they actually present their real world data, very few patients are actually able to adhere to such an intensive strategy. This was revised in 2017 and they added some more worrisome features, namely lymphadenopathy C A 99 elevation and some and the rate of cyst growth, the A G A kind of jumped in the middle of all of this, the American Gastrology Association and put a much more conservative spin out there considering at the time they were talking about doing surveillance every 3 to 6 months in some patients. And what they basically said was that you should repeat an MRI kind of regardless of the size in a year and then biennial 8 to 5 years. And if nothing has happened after after five years, you should just probably stop and they, you should consider surgery only if there's positive cytology or if there's a dilated main pancreatic duct or um a uh a solid component on imaging. So a much more conservative strategy, people, you know, really reacted negatively to that when they came out, they said, how dare you stop after five years and how could you have such high bars to, you know, go to surgery? You know, this is going to lead to a lot of problems, blah, blah, blah. And there was a lot of debate in the field about, you know, the utility of that, the European guidelines came out. I'm only bringing this up because in their guidelines, endoscopic ultrasound doesn't nearly play as high of a role because they've been more skeptical about, you know how the endoscopic ultrasound moves the needle in terms of surveillance. Um you know, strategies, but I'll say is that in all of these guidelines, you know, and paper after paper center after center that tried to validate these retrospectively, they all, you know, perform the Fukuoka guidelines or the follow up to Sendai guidelines are better in terms of specificity, but they lose some sensitivity as opposed to Sendai guidelines. The A G A guidelines similarly have good specificity, but their sensitivity drops off. And the, the fundamental too is that they're all operating on information or inputs that are, you know, imperfect. And so you're trying to parse through a lot of things that are not, um you know, that are not, you know what you hope, you know, that, that are not, you know, 100% predictive. So, you know, you could be a nihilist and be like, well, what if you did absolutely nothing? And so this paper actually came out, um, a year and a half ago asking that exact question essentially what is benign neglect to do for you for pancreatic cyst. So they, um, kind of used a very opportune cohort of patients that they had, this was in Vienna and they basically found patients who were found with the cyst, but nobody did anything about it, um, for more than 10 years and they excluded people who they knew they were going to have, you know, a higher risk of a cancer. People with main duct I PM N pancreatic cancer, chronic pancreatitis patients who died within a year of their pancreatic cystic lesion who are basically looking at people who like, for example, had a, had a, you know, motor vehicle accident, had ac T scan in the er there was a five millimeter cyst on the, on the ct that nobody really paid any attention to because the person was there, you know, for a trauma and then said, you know, what happens if that, you know, system was lost to follow up. And then we kind of found them, you know, 10 years later. Um and they used a control group of patients from Verona in Italy, which is one of the big centers for I PM NS in Europe who were being extraordinarily closely surveyed. Um And these were all low risk lesions less than 15 millimet, they were on protocol surveillance and they had 376 non surveyed piece uh cysts and 299 surveyed piss cyst. And what was fascinating was that in this low risk cohort of, of cysts where you've taken out people with these obviously high risk features at at presentation over the course of 10 years. If you did nothing, the there was a similar incidence of pancreatic cancer, a similar disease specific mortality, pancreatic cancer, similar to cumulative risk of pancreatic cancer and similar survival from the pancreatic cancer if it was to develop, suggesting that, you know, there is certainly a group of these cysts that are probably not benefiting from us being super intense and it's probably a group of these cysts that you know, may actually be of some benefit. Um Why don't we just use AC T and MRI? Because there are limitations um in terms of accurately predicting what the cyst is and in differentiating aggressive and non aggressive behavior. And you know, these cysts on imaging, you know, look like this. They all, all four of these images show you cysts on CT and MRI, but they're all different pathologies. The one on the top left is a mucinous cyst neoplasm on the right, a cerys adenoma on the bottom left, uh pseudocyst and on the bottom, right, a solid pseudo pai tumor. And so obviously, you know, you can't fully, you know, determine what is going on in cyst just by um you know, appearance alone. So one of the questions I had asked with my research group here at Wash U is how good are we using like we meaning not artificial intelligence but actual intelligence. How are we as a as experts at Wash U? Um And then subsequently, I did the same study um on, on participants from Johns Hopkins. Um How good are we as experts at a pre assessing a pancreatic cyst knowing that all the um information that we are getting is imperfect but waiting it in real time appropriately, heuristically so that we can try to make better decisions about stuff. This is the data that kind of preceded the study that I had done and this is looking at uh multidisciplinary groups um from, from Hopkins and from um from China as well. And what they basically show is that the diagnostic accuracy for, for being able to stratify things using endoscopic ultrasounds, MRI S et cetera is about 63.5 to 78% and 30% of the time when you refer patients to these big centers and they have um evaluation here, we actually alter the management of those patients when they were rereviewed versus when they are reviewed um you know, by, by physicians in the community who are obviously, you know, good and well trained, but just may not have as much experience, you know, dealing with this every single day. Um and that clinically relevant um erroneous diagnosis were only made less than 10% of the time when they were actively reviewed in these settings. So we tried to model this as I was alluding to um in a multi dis fashion, both done here. And then I replicated a study this year at, at Hopkins and we took patients um 40 patients who were being presented at our pancreatic multidisciplinary panel. And we broke them up into two groups. Um Essentially cohort, one had 10 high risk lesions and 10 low risk lesions. And cohort two had 10 high risk lesions and 10 low risk lesions. And we created a mock or fake multidisciplinary panel where we presented the cases, presented the imaging to a radiologist, to a gastroenterologist, to a surgeon, to an oncologist, all working together, speaking together um without um you know, biasing them or telling them, you know what the cases were, what, how, how they ended up, um you know, panning out, we also half the group got the endoscopic ultrasound data upfront. And half the end half the group did not have the endoscopic ultrasound data upfront. And for those cases where they didn't have the endoscopic ultrasound data upfront, we then provided them the endoscopic ultrasound data and see did the endoscopic ultrasound change your opinion or not? And so we generated basically these three groups, one group where you just have imaging and you're not doing any endoscopic ultrasound, one group where you have imaging and endoscopic ultrasound and one group where you initially didn't have the endoscopic ultrasound. But then you got the ultrasound after the fact to see, hey, did that change, you know what your thoughts were? And the purpose of this is basically to study how good are we are we biased in terms of how we think about endoscopic ultrasound data? And you know what is the actual ceiling and floor of where we are, where we should be thinking about, you know, additional diagnostics in this space. And so this is a busy slide. So I apologize, but essentially the yellow on the left. Is it? Oh, sorry, somebody, somebody just has a question or maybe they just were talking. Um the the yellow group on the left is the, the the assessment of, of, of our, of our, you know, mock pan pancre multidisciplinary groups um without the benefit of the US, the seconds with E US. And the third is if you were initially uh blinded and then unblinded, there's a lot of data here. But the take home points is that with, with an endoscopic ultrasound in hand, the expert, the expert panel review in both of these centers was pretty darn good. We were 97% sensitive and 70% specific for figuring out if something was a mucinous cyst, 80% sensitive and 83% specific for something that was advanced in the aplasia, meaning high grade dysplasia or cancer. And 85% sensitive. And 85% specific for appropriately referring something to surgery that absolutely needed surgery. Um And so, you know, while each of these individual features, the size, the presence of a nodule, the cytology, you know, um, the pancreatic duct diab are, may be imperfect. The point of this is to show that, you know, if you are working with groups of people who see a lot of this stuff and see a lot of these patterns on a day to day basis. Um, we ultimately do a pretty good job of figuring out, you know, which patients are going to benefit from surgery or not. Obviously not. Perfect. We'd like all these numbers to be 100 and 100. But um better than the sum of its parts is what I'm gonna get at. So, um, I'm gonna shift gears a little bit to, you know, a subject that comes up a lot in primary care, which is when to stop, when to de escalate. Um, when is it that these cysts are, are something you're gonna die of and not die from as I tell my patients. So this is paper, this is a paper that came from MG A looking at a retrospective valuation of 725 patients that were being carefully monitored with these cysts over years and years and they utilize a Charleston comorbidity index which maybe many of you may be familiar with, but it is a really easy bedside. Um, you know, tool that you can plug in um patient's medical history to kind of calculate numbers on. And they used a threshold of of seven. And they basically said that if a patient had a Charleston morbid morbidity index of equal to or more than seven, their median survival was 43 months and that there were 11 times more likely to die within three years of a non I PM N related outcome than an I PM related outcome. This, you know, I frequently will bring this up with patients in clinic. And sometimes they're shocked to know that they have a median survival of 43 months or shocked to know that they're 11 times more likely to die of something else than, than of this. Um, but it's a helpful kind of, um, starting point sometimes to kind of talk to patients who are more elderly or more frail and more com comorbid that essentially if you did find a cancer in them, it would be the, the exception to the rule. The other thing I also talk about with them is that if you ever talk morbidity and its of 789 or 10, how well are you going to do with a whipple if I actually find a cancer? Because it's always this question of like, you know, a dog chasing the mailman. If a dog keeps chasing and chasing and chasing and actually finds a cancer, you know, what are you gonna do? With that information, similar data um from, from Europe, from that Italian Verona cohort combined with the MGH data um showed that in 281 patients undergoing surveillance, the five year disease specific survival in patients, even with those worrisome features was still very high. 96.2%. And with higher stigmata which remember are things like presenting with jaundice or a big pancreatic duct was also not as high but you know, 60 60%. Um this is a patient uh paper done out of um out of um the Kaiser group again and they looked at 1800 patients followed over 10 years and they just find a high risk patient much more loosely. I think having a Charleston coma index of more than three and a high risk cyst just very broadly as one being more than three centimeters or having a mildly dilated pancreatic duct to five millimeters which, you know, is not, you know, for sure a cancer by any stretch. But what they, what they found was that essentially if you have a high risk patient and a low risk cyst, your chance of of a pancreas cancer um is 0.3%. And your chance of developing a non pancreas cancer associated death is 38.8%. So, you know, more than 100 times more likely to die of something non pancreatic cancer related. Why bring this up? Because we have umpteen patients who are coming in dragging in an oxygen tank in a wheelchair who have a six millimeter or 10 millimeter cyst that's being surveyed for a very long period of time and sometimes under well meaning but not clinically informed recommendations from AIST who are just looking at scans and not a patient and not interpreting whether or not ongoing surveillance, you know, makes sense or not. So it was in this kind of vein that people started thinking like, you know, when should we kind of de escalate or not? Think about following patients muddying the waters is that there have been several series from different places from Europe and the United States suggesting that there are some cysts that even beyond five years of stability may go on to develop cancer in, in this top group, you see 18% develop worrisome features or high risk stigma. After five years in the MGH cohort, they showed that cyst there were less than a centimeter and a half over the course of five years, had less than 1% chance of developing a cancer, which is an important um you know, thing to note, but there was some risk of developing cancer beyond five years. And this data was similarly seen in, in Japan and in Korea. Um though a little bit higher, I I would say in the Asian cohorts and in the American cohorts in this context, this paper came last year which I want to highlight uh because I think it, it bears um in a conversation in a primary care setting or in a patient in a shared decision making setting when you're talking about with the patient, what this study was essentially looking at patients across the world, um looking at 3844 patients that were being surveyed for at least 12 months and looking at branch duct and tribal I PM NS and basically seeing what be, you know, became them. And so um 20% of them had worrisome features. 1.8% of them had higher stigmata. They defined a trivial pancreatic cystic lesion as cysts less than 15 millimeters after five years of surveillance. And what they found was that the rates of cancer of pancreatic cancer, I should say were similar after five years to the overall population control in patients above the age of 75 who had cysts less than three centimeters or above the age of 65 with cysts less than 15 millimeters. And that you should probably consider discontinuing surveillance of these two groups of patients. The argument being that if you're not gonna, you know, survey or screen every single patient off the street for pancreatic cancer, then why would you do that in a patient who has the same risk of pancreas cancer as somebody just walking down the road? Um and they kind of even drilled down a little bit further to see. Well, what is the harms of continuing because like, you know, a lot of the time the patients, like, well, what's the harm? Just do another MRI another year? You know, I don't mind paying the copay. Like, what's the big deal? Um, so if you look at the trivial cysts, those are the ones less than 15 millimeters that were stable over five years and looked at what the all cause mortality versus the I PM N specific mortality was. These patients had a 4.9% all cause mortality and a 0.3% disease specific mortality. So they were 16 times more likely to die of something else. Addition, there may be actually a harm to continuing surveillance of this group because in trivial I PM NS that were without worrisome features or high stigmata. 25 of them actually eventually underwent resection. Um, you know, and that's something that definitely can happen because you keep looking at the cyst and you get nervous and is there something there? Well, we should probably do something, you know, you could see how a person could get talked into that given the imperfect nature of our preoperative evaluation of these patients. And what they found was amongst those 25 per uh patients who did undergo surgery 8% died. Um whereas the, the non resected mortality in that trivial group was 4.9% suggesting that those who actually underwent the surgery did worse than the ones who, you know, who didn't. Um, and, you know, and surgery is not trivial as you probably all have remembered from your training or seen in your experience amongst those who had surgery in this, you know, 3800 patient group, 100 and 64 went to surgery and only 32% actually had cancer or high grade dysplasia. Close to 70% of them didn't and probably didn't need the surgery at all. But all of them faced the same complication rate of 50%. Um with a 30 day mortality of 2.4% which is obviously not insignificant. So we have to really be, you know, thinking about this and this is kind of where these new guidelines, you know, that came out this year in um February March, um came to be this Kyoto guidelines, um which are updates of those prior Fukuko guidelines. And what you'll see is obviously, the worrisome feature list has kind of grown a little bit more to include new onset diabetes. For example, C A 19 9 has been added into their lymphadenopathy. The cyst growth rate has gone to 2.5 millimeters per year. But very importantly in this, you know, what was previously a very conservative group of people who said that people should get surveyed forever and ever and ever. There is a recognition that if you have a cyst that's less than 20 millimeters in size and you've been surveying it for five years, you could continue, um you could consider stopping surveillance, especially in a person, you know, who has a much higher likelihood of dying of something else. Um So I'm gonna switch gears a little bit from something that, you know, from a very clinical discussion to something a little bit more on the bench to kind of talk about how that informs how we think about these cysts. So one of the questions that comes up frequently, you know, when we're talking in, in pancreatic circles is do I PM NS and pancreas cancers have a similar cell of origin? So, meaning are I PM NS and are, are all I PM NS destined to go towards cancer? Do cancer show up separate from I PM NS? Are they, are they, are they neighbors? Are they cousins? Are they brothers? And you know, what are the mutations that drive I PM NS versus one of the mutations that drive, you know, pancreas cancer? Is this something where we're looking at like a polyp in a colon that's slow will develop in inherently into a cancer if it's given enough time or is this something that, you know, you might see a polyp in one spot, but the cancer shows up from something totally different on in another area. And then you know, why do certain types of I PM NS are more likely to develop into cancers than other types? And you know, work that's now been stretching on for about 15 years um in looking at which mutations are found and which types of cysts have really revealed a lot of, you know, important and interesting insights where basically over the course of all these different papers, I'm kind of showing you on the left and just giving you the reader's digest version. But essentially, if you do genomic analysis, a parent could exist in I PM NS. It shows you that, you know, Kras is not an uncommon mutation in a lot of these um uh lesions, gene mutations are very commonly seen in I PM. NS VHL mutations are seen in cerys, adenoma kine and B one mutations are seen in um solid pseudo neoplasms and the combination of P 53 and P I three Kase or P 10 were seen in patients who have, you know, uh something more aggressive um or, or advanced. But you know, what are the conceptual limitations of sampling? You know, one cyst, a lot of the times you'll see patients who have, you know, cysts spread out throughout the pancreas. Do all those cysts have the same risk of cancer or is the biggest one, the one you need to worry about is the cancer gonna show up in the biggest cyst or is it gonna show up in the smallest cyst? So this has actually been looked at um this is a work from Laura Wood from, from Hopkins where they did laser capture micro dissection um from 20 patients who had resected I PM NS. And they found that basically there was a lot of heterogeneity in, in the, in the mutations in the, in the I PM NS with low grade dysplasia. The ones um the I PM NS had multiple different clones, meaning they were very heterogeneous in terms of their genetics. Not every single part of that cyst is the same and multiple high graded I PM NS did not have a specific mutation in a driver gene that they could identify. Um And similar group. Uh Similar data has also shown that in patients who actually developed a a pancreas cancer with an adjacent I PM N, they found that 51% of the cour I PM NS and pancreas cancers were related but close to 20 percent of the I PM NS and pancreas cancer were genetically completely different. Meaning this person developed a solid pancreas cancer, you know, in the body of the pancreas, but their cyst that you've been paying so much attention to and putting a needle in and you know, and addressing over and over again is in the head of the pancreas suggesting that I PM NS encompass multiple subclans all progressing in parallel and then conta uh concurrent metachronous solid pancreas cancers are frequent and frequently not derived from this. The pancreas cyst itself. Therefore, assessing the specific cyst may give you a false sense of security in some ways that the person is totally fine, you know, and that that's in the entire story. So I'm gonna shift in the last part of this talk to talking about diagnostics um that we use, you know, in the clinic and that might be on in the near horizon. Um First, I'm just talk about the stuff that we use every single day. That's been a, a backbone of the traditional so to speak cyst fluid studies that we use. Um to try to help figure out what kind of cyst we're dealing with. The first is amylase, which are all you know, should be familiar with in, in clinical studies because you said in clinical practice because you send this all the time from the serum in patients when you're worried about pancreatic inflammation, all that Emily in the cyst fluid indicates is a connection between the cyst and the ductal system. Um And the level can't really differentiate between a pseudocyst or an I PM N or an MC N or prognostic prognosticated at all about malignant risk where it helps is to try to figure out if something is a cerys adenoma or immunocyte neoplasm because those are not associated with the pancreatic duct and should have a low um amylase as opposed to an I PM N, which is inherently associated with the duct. And so therefore, have a high amylase ce A is something that's been looked at, you know, now for close to 20 years um to help try to predict whether something is mucinous or not mucinous. That is if something is an MC N or an I PM N or if something is a cerys adenoma or the pseudo capillary neoplasm, um This, you know, has had, has strong utility in the extreme. So if it is very, very low or zero, it has a positive predictive value of excluding uh mucinous uh pancreatic cyst. um you know, definitely if it's extraordinarily positive, so more than 800 then it's positive predictive value is 94 percent confirming that it's something mucinous. The problem is everything in between five and 800. And so this is why you had lots of different papers. The 105 1 is from our group, the 192 is the original from MGH, the 30.7 is from uh U Penn. I think um you know, people giving you all kinds of different cutoffs with different sensitive and specificities. Um and trying to, you know, uh help define as something use or not use. The other thing is even if you have an A ce a of 1000 there's not really association with malignant potential. Um And so that, you know, it's something also to kind of uh conceptually, you know, wrap your head around cytology has been around since the beginning of cyst fluid analysis. Um and has a very broad kind of overall assessment of its accuracy in practice, the overall accuracy is probably somewhere around 50%. It's very specific. So obviously, if you see cancer cells on the cytology, you know that there's something cancer is there, but it's not sensitive at all. And it's, and the sensitivity is as low as 50% or lower. So what I tell patients is just because I don't find, you know, cancer in the, in that in the cyst, for example, that I identified, it does not at all exclude that possibility. Um which is obviously not a very reassuring thing, you know, to tell a patient. Um cyst fluid glucose is something that's been um evaluated a lot more in recent. Um in recent years. Um originally coming from Walter Kim, um who's out at Stanford and what they found is that reduced glucose levels in cyst fluid was actually strongly associated with mucin cyst. It's not clear why this is whether it's because the cysts have a lot of metabolism because they're growing a lot and then using up that glucose or if there's displacement of fluid with mucin and the mucin just doesn't have a lot of glucose in it. But what's clear is that a low cyst fluid glucose across many, many studies has a pretty good sensitivity of 90% and specificity of 85% for picking up a mucin cystic lesion. Um and you can even do this with point of care. Um glucose for us. And you know, the endoscopy lab just putting, dropping a little bit of the cyst fluid, you know that as soon as we get out of the patient onto a test strip, um even not sending it down to the lab just using a, a point of care, you actually do pretty well. Um And in this multi center study done by one of our, our former fell um, Zach Smith, he basically showed that if you have a laboratory glucose of less than 25 you have a very good um 88% sensitivity and 91% specificity for picking up a mucous um am minus cyst. So, in many ways, I think a lot of us are, are now shifting to use glucose a lot more than, than, than ce A. Um just because it's much, it's, it can be a lot easier to interpret. Certainly, certainly is less, less expensive and it is extremely accessible. Um In assessing these cystic lesions, we also have through the forceps needle biopsies uh for the last several years, sometimes called moray forceps. Um These are miniature biopsy, extraordinarily small biopsy forceps actually can fit within a 19 gauge needle and then be passed inside of a cyst to biopsy the cyst wall. The purpose has been to look essentially at the cyst wall tissue to try to assess if it's something mucinous or not mucinous or potentially cancerous or not. You have to do about three passes, you know, for specimen adequacy in data from Dennis Yang, from um Florida, looking at 100 and 14 prospective patients, they were successful most of the time, 100 and 11 at 100 and 14 cases. And it did increase the the the yield of the classification of a mucinous cyst. Um in the 14 patients where they actually had resection pathology to compare it against the forceps was predictive of what they ultimately found of it being either mucinous or not mucinous as opposed to just cytology a lot. The cost you pay though is an advance, a significant adverse event rate of bleeding up to 6.1% which I've personally seen and pancreatitis of up to 55 5.5% again, which I've personally seen. So the question with all these, you know, interventions is the juice worth the squeeze. Um There have been multiple studies now and meta analysis looking at this through, through the needle force of biopsy and it is definitely um pretty good for telling us thing is mucinous or not mucinous. But in terms of telling you whether what type of assist it is, it's not all that great. And then cancer potential also not particularly that great. So I would say in the last 23 years, um you know, people have been less excited about using using this technology, especially since this mucous versus non mucous sensitivity specificity is not that much different than glucose, which is obviously easier, faster and cheaper, um, you know, in, in most situations, um I'm gonna quickly just uh go through a couple of the things that are on the horizon, um or, or just becoming available um to, to kind of give you some sense of the dynamism that exists in this space uh in attempts to try to risk stratify. Um you know, the different pancreas cysts using, you know, pancreas cyst fluid. Um One is uh telomerase, which is from Mike Goggins group in Hopkins. And they found that um the elevated fluid telomerase activity can accurately predict cancer and high grade dysplasia. Um Though there's issue with freestyle with the utilization of that, this is from a group um out of Scandinavia looking um at doing mass spec uh performed on, on patients um with cysts and there's also with also with pancreatic cancer and they found uh two specific targets to mucin proteins. MC five A and muck two. And using a quant, a sophisticated targeted, targeted quantitative mass spec, they were actually able to try to uh delineate um a high grade versus low grade cysts. Um And they, you know, suggested that looking at these specific peptides had a 97% spec spec specificity at least for high grade dysplasia and invasive cancer. Um The lower sensitivity um work out of mayo clinic um from um um Doctor Muja Dar um was, has been looking at uh methylated DNA markers to discriminate advanced neoplasia and pancreatic cyst. And they used a whole methyl discovery approach um and its validation in tissue um and then evaluated cyst fluid from E US and surgical um aspiration um patients and found um you know, pretty good sensitive specificity. About 92% sensitivity, about 90 percent. My own work has been on this on a biomarker called MV DOS one which we have found recently with my collaborator Jeff Brown here at WASU um recognizes a, a very unique and well um um protected suation epitope three prime sulfate Lewis AC which is distinct from C A 19 9, which you can see in the middle there. Um But it seems to be very unique for the development of and we found almost 10 years ago now when I was at MGH, um that this was a very sensitive marker for identifying in pathology as well as in a CIS fluid pilot um lesions of I PM N lesions that were more likely to develop into cancer. Um And I put together a multi center consortium around the country including Hopkins, um Sloan Kettering MGH and ourselves. Um and looked at 100 and 81 patients who had perioperatively aspirating cysts. And we found that um this biomarker was able to predict in cysts um advanced ne aplasia with the sensitivity of 88% and 90 specificity of 99%. Um And so we're in the process of, of, you know, bringing that from the bench to the bedside, which is obviously, you know, a very different thing when you do something on a 96 well plate in a, in a lab, as opposed to when you're trying to deliver it, um, you know, to, to in a, in a clinical environment, but hopefully some someday soon in the effort to try to find needles and haystacks, um There's been a lot of work, especially coming out of Pittsburgh. Um looking at next generation sequencing and next generation sequencing panels. This is the first paper talking about um uh uh uh the first sort of multi multi analy NGS panel that came out of a sings work. Um And what they were invest basically trying to do is say we have all these basic, you know, signs and um you know, evaluation showing all these driver mutations that drive people drive. Um you know, cyst towards cancers, can we assess that in cyst fluid? And, you know, predicting in real, in real time knowing that P 53 and P I three Kase and P 10 are associated with, you know, advanced neoplasia. Um And so basically, they looked in this initial study at preoperative pancreas cyst fluids, uh cystic lesions that were aspirated. Um For next generation sequencing, 93% had specimen adequacy pathology was only available about 17% of patients. So obviously, there's gonna be a bias in the patients actually get, you know, um surgery for us being able to validate this. But, you know, this was initially pretty, pretty good that you could use uh KRS and GS mutations with pretty good sensitivity for knowing that you're dealing with an if PM N and P 53 mutations, P A three Kase and P 10 mutations were also pretty good, 88% sensitivity, 95% specificity for figuring out that you had an advanced K ne aplasia. Um And so these essentially are the, the combinations of things that they kind of added together to try and um cobble together, you know, reasonable sensitivity and specificity for figuring out if something was cancerous or non cancerous. And this is iteratively improved, you know, o over the years. Um This is the most latest um you know, version of that panel, I, what I would essentially cause the, the 3.0 of that next generation sequencing panel that just got published in gastroenterology last year. This looked uh initially looked at a cohort of 97 lesions um that they then just to kind of optimize the panel, add additional uh genes that were kind of going to be investigated. Now, they got went up to 22 genes. Um And then they used a very large real world uh prospective cohort of patients who are having cyst fluid analyzed and sent for commercially um you know, commercial study. And this, this, this wasn't free. The patients were paying for the for the cyst fluid analysis um from 31 institutions around the country and they had a good clinical follow up in about 251 or 21% underwent surgery. Um And then in a sub subset of these patients, they actually did on comin to see, you know whether in the tissue they could find the same mutations that they were finding in the cyf fluid. Um These are the sort of overall um clinical as well as um genomic mutations that were associated um with I PM NS IP NS with advanced nepl. Um and I PM NS and M CNN S with advanced neoplasia and so cutting to the, to the chase because I don't want to get too lost. I know this. Um not all of you are gonna be in the weeds of, of, of, of genetics and pancreatic cyst, but map kindness and gene S mus. Um in combination with TP 53 S Mad four and Katin and B one and M two alterations had an 88% sensitivity and 95% specificity advanced ne aplasia. VHL was sensitive and 100% specific for cyr cyst Anoma men. One alterations were sensitive and very specific for neuroendocrine tumors. Um and these are, you know, these, these numbers and findings are not dramatically different, I would say than the previous, but iteratively, like I said, you know, slightly better than what it was before. Um You know, there are issues obviously that come up when you do a multi analyst next generation sequencing panel, um essentially, you know, the vast majority are being picked up by the things that they found in the 2017 paper, the P 53 and SMD four, and you're adding a lot of complexity to that uh analysis to pick up, you know, a very small number of patients. Ultimately, um you know, again, the, these two genes are doing the heavy lifting and then these, you're adding every gene, you're adding, you're getting less and less and less, you know, out of it and obviously, additional cost and complexity. Um I'll skip over this issue. Um We obviously need long term prospective data. There's been a national study called the E A 2185 study, which we're one center for, we're basically following patients perspectively randomizing them to high intensity and low intensity surveillance to see, you know, who, whether or not, you know, it actually behooves us to be looking at these cysts every six months or 12 months or if we can look at them every two or 2 to 5 years. Um There's some technical issues, you know about, you know, what happens if you do one of these genes, uh these gene analysis and you find a very low frequency but high risk uh mutation, say, for example, you do one of these and you find a P 53 mutation which you know is associated with cancer but it's only one analytic or very, very low menal frequency. Do you follow that patient more closely? Do you resect that patient? You know, I think a assessing exactly quantitatively how much or how many clones are, are, are too bad and which patients need to go surgery is something we need to still figure out in that space. But it's ultimately very exciting that there's a lot of dynamism here in trying to be able to both figure out which patients need more surveillance, which patients need less surveillance. And then what tools we can use to, to, to employ in those patients who are at highest risk to actually make sure that we um you know, find the cancer, you know, before, you know, before it turns into something that we can't um you know, address with surgery or chemotherapy. Um So I'll finish up with the conclusions and try to leave 5, 10 minutes for us to, to, to have questions, but essentially cyst fluid amylase and ce A especially in the extremes can be helpful to identify what cyst you're dealing with. Cytology has a very high positive predictive value but or negative predictive value, reduced cyst glucose is an inexpensive, easy and accurate assay for assessing mucous and non mucous lesions. Multidisciplinary review of pancreatic cystic lesions in the data from our group um is really a gold standard um and is actually quite good at assessing for advanced neoplasia not that different than a lot of these, you know, more um elaborate tests um through the needle forces. Biopsy is good at mucinous versus non mucinous pancreas cyst. But um has significant adverse events that limits its um you know, utilization. Newer cyst fluid studies like telomerase mass spec panels, methylation targets the dos one antibody that I showed you all are very promising with a similar um advanced and neoplasia detection rates in the 85 to 90% sensitivity, 90 to 100% specificity range. Um Genomic analysis of pancreas cystic lesions have demonstrated that there are can mutation that seem to be clearly associated with certain pancreatic cyst types. And that next generation sequencing allows for the identification of very low frequency mutational targets. These are have increased costs and complexity and there's issues that still need to work be worked out in research and, and, and practical clinical um you know, um settings to figure out how much of a bad mutation is a bad mutation um which patients need more aggressive surveillance, which you know, which genomics, you know, fingerprints suggest that, you know, we can throttle down the frequency of, of, of surveillance. Um and all that hopefully will be, you know, continue to be worked out over the next five or 10 years. I'll stop there. Um This is our entire interventional endoscopy group. Um as it stands, it's, it's quite a large group. Um as you can see here, um we have um uh a faculty uh uh eight faculty now um that are full time uh and uh three A PPS as well as an excellent um cadre of, of nurses um that really help us um help patients with complicated and difficult problems, um like pancreatic cyst, but so many others. Um if you ever have any questions or ever have any issues or um issues with uh patients that you need, um you know, us to talk through, you can give a call to the number up on the screen. 314747 9843, which is why UIE, um or you can, you know, take my cell phone number or any of our cell phone numbers and we're always happy to help troubleshoot things. Invariably, somebody shows up jaundiced with a PK mask in your office on a Friday afternoon and you're really trying to figure out what the best and smartest thing is to do for them and we're always happy to help troubleshoot that stuff or to get imaging brought over. Um, you know, from your neck of the woods over here to help kind of guide with the next decision is recognizing that it is obviously a burden for patients have to travel, you know, long distances. Um So on that note, um I just want to open it up to any questions that people may have. Um, you know, about what we're talking about or anything else that you may um have in this space. Created by Presenters Koushik K. Das, MD Gastroenterology. Interventional Endoscopy View full profile
