Novel Approaches to Enhancing the Diagnosis of Pancreatic Cancer: New Horizons and Emerging Technologies
Clinical Dialogues™ in Pancreatic Cancer
Novel Approaches to Enhancing the Diagnosis of Pancreatic Cancer:
New Horizons and Emerging Technologies
|Anil K. Rustgi, MD
T. Grier Miller Professor of Medicine and Genetics
Chief, Division of Gastroenterology
University of Pennsylvania, School of Medicine
|Koushik K. Das, MD|
Division of Gastroenterology
University of Pennsylvania
My name is Anil Rustgi. I am the chief of gastroenterology at the University of Pennsylvania. Our division and program works intensively in pancreatic cancer from basic science research to clinical care. My colleague, Dr. Koushik Das, is also a member of the University Of Pennsylvania, Division of Gastroenterology. Today we will be asking Dr. Das questions related to pancreatic cancer in terms of diagnosis and the opportunities at hand and on the horizon for enhancing the screening of pancreatic cancer.
Dr. Anil Rustgi (AR): Let’s start with some of the epidemiological aspects of pancreatic cancer. How many people get it every year in the United States, and from a clinical view point, how do they present?
Dr. Koushik Das (KD): With regard to epidemiology, pancreatic cancer is currently, according to the SEER database, the 12th most common cancer in America with an estimated 45,220 cases a year. It is currently the fourth leading cause of cancer mortality with over 38,000 deaths per annum, but it is expected to become the second leading cause of cancer mortality by 2020. The majority of patients present at an older age (about 85+% over the age of 55) and unfortunately only around 9% of cases present with localized disease amenable to surgical resection; over half of the patients present with distant metastases usually to the liver, and over a third with regional spread to local lymph nodes making upfront surgical management inappropriate.
AR: If we as a community were able to diagnose pancreatic cancer at an earlier stage, would we anticipate improvement in outcomes for our patients with pancreatic cancer?
KD: That is a good question. Survival is certainly better with earlier stage disease than it is with widely spread disease. However, it is not dramatically so the way it is, say for example for melanoma, where stage 1A patients have a better than 95% 5-year survival and metastatic patients have less than 15% survival. Instead, even amongst patients with stage 1A pancreatic cancer, the patients have under a 20% survival at 5 years and stage 4 patients have a much poorer, less than 5% survival at 5 years. With that said, as of 2014, surgery remains the only option for a cure. Thus, if the disease is caught early and 20% of patients survive to 5 years, this is certainly better than the alternative. I will also add that these long-term post-surgical survival studies have not been looked at with recent advances of adjuvant chemoradiation, or with regimens like Gem, FOLFIRINOX, or the immunologic therapies that are close on the horizon. I will also make a point that several studies have demonstrated that all Whipple’s resections for pancreatic cancer are not made the same and that the center in which the surgeries are done and the multidisciplinary team dedicated to those patients make a big difference in outcomes for the patients. In a low-volume center that is doing few surgeries, the perioperative mortality and complication rate can be three to five times higher than that seen in a high-volume center with high-volume surgeons like we have here at the University of Pennsylvania. Thus, diagnosing pancreatic cancer in an early, surgically resectable stage is not a guarantee of cure, but it certainly does dramatically improve outcomes for a disease that generally presents quite late and with a few definitive management options.
|“Diagnosing pancreatic cancer in an early, surgically resectable stage is not a guarantee of cure, but it certainly does dramatically improve outcomes for a disease that generally presents quite late and with a few definitive management options.”|
AR: Do you think that it is feasible or appropriate to expect to have a screening test for pancreatic cancer available in the general population in the near future or the distant future?
KD: Most of the common screening tests that we think of for cancer in the general population are tools to examine groups for commonly occurring cancers like mammograms for breast cancer, colonoscopies for colorectal cancer, or the PSA test for prostate cancer. The important thing in thinking about creating a screening test for a population is to understand the underlying incidence and prevalence of a disease process to guide you in what test characteristics are required to maximize accuracy and minimize false positives. The incidence of breast or prostate cancer is around 130 to 140 per 100,000 people. Meanwhile, thanks to effective screening programs and advances that have been put in place in the last 10 or 15 years, the incidence of colorectal cancer has fallen from about 55 down to 35 in a 100,000 people. I only gave you these numbers to show that, in comparison, pancreatic cancer is still quite uncommon, at 12 to 14 per 100,000. What this means is that because the incidence is so low in the general population, we are essentially looking for a needle in a haystack. Furthermore, if we were to find something suspicious, our current armamentarium consists primarily of relatively radical, morbid surgery to remove the pancreas which, while life-saving in many situations, can have many important life-changing side effects. Thus, given the low overall incidence of pancreatic cancer in the general population, we would need to have a test with near 100% sensitivity and specificity to be effective so we do not find a lot of false positives which may lead to unnecessary and morbid surgeries. As this bar is so high owing to the low incidence of disease in general population, it is unlikely that we will find a test that will perform to that degree of accuracy to allow for safe screening amongst the general population.
AR: Given that, which population of patients in your opinion would benefit the most from screening in pancreatic cancer?
KD: Over the last 5 to 10 years, screening efforts have primarily focused on populations of patients who have significantly higher risk of pancreatic cancer than that seen in the general population. There are several common environmental factors that increase the risk of pancreatic cancer, most notably alcohol and tobacco use, though these are not sufficient to warrant screening. The two overall situations where the risk of pancreatic cancer is sufficiently elevated to warrant screening are in 1) patients with known or suspected genetic predisposition to pancreatic cancer or 2) in those with a history of hereditary pancreatitis.
|Risk Factors for Pancreatic Cancer
About 5% to 10% of all cases of pancreatic cancer are thought to have a genetic basis. Outside of known cancer syndromes, it is estimated that having one first-degree relative with pancreatic cancer has no increased risk of pancreatic cancer, whereas having two first-degree relatives has about a 6.5-fold increased risk from the general population, and having three first-degree relatives increases that risk to about 32-fold over the general population.
There are several important genetic syndromes in which pancreatic cancer risk is increased, specifically, familial atypical multiple mole melanoma (FAMMM), BRCA1 and 2 mutation carriers, Peutz-Jeghers syndrome, and hereditary nonpolyposis colorectal cancer (HNPCC) previously known as Lynch syndrome.
Among patients with FAMMM, there is between 13 and 22 times increased risk of developing pancreatic cancer; close to 20% of these patients develop pancreatic cancer by the age of 75. FAMMM is due to germline mutations in p16 or CDK4 which inhibit interactions of the cyclin D1-CDK4/CDK6 complex. The classical presentation in this syndrome is obviously with melanoma in two or more family members, but pancreatic cancer is an important component of this syndrome as well.
Amongst the patients with BRCA1 or 2 mutations, there is an approximately three to four times increased risk over the general population for pancreatic cancer. In this syndrome, mutations are passed on in an autosomal dominant fashion. The BRCA1/2 genes are integral in the normal physiologic repair of double-strand breaks in DNA and their impaired function allows for a loss of heterozygosity and translocations in the genome.
In Peutz-Jeghers syndrome (PJS), patients have a close to 132 times risk of developing pancreatic cancer, and interestingly, in this syndrome, the patients actually develop a wide array of pancreatic lesions, not just traditional pancreatic ductal adenocarcinoma. Specifically, PJS patients have been described to develop cystic lesions of the pancreas, ampullary lesions, and others. PJS is due to mutations in the LKB1 (STK11) gene that has important interactions with p53 and BRG1. The classic clinical presentation for PJS is hamartomatous polyps in the GI tract as well as characteristic mucocutaneous pigmentation.
Finally in HNPCC, the risk of developing pancreatic cancer is about two to four times the baseline population risk. HNPCC is due to germline mutations in mismatch repair genes affecting either the MSH2/MSH6 complex or the MLH1/PMS2 complex. These mutations interfere with the normal “proofreading” function in DNA replication. This syndrome is also autosomal dominant, with a high, 80% penetrance.
In addition to genetic syndromes, patients with hereditary pancreatitis are the other major group of people who have a risk of pancreatic cancer and may merit surveillance. Overall, the risk of pancreatic cancer is estimated to be about 35 times higher than the general population.
These syndromes are autosomally dominant with about an 80% penetrance. Mutations in the PRSS1 gene, a cationic trypsinogen, which cause premature activation of the protein is thought to be central to the etiology of the recurrent bouts of pancreatitis that are a hallmark of this condition. In the setting of chronic inflammation, pancreatic cancer can ultimately emerge. In addition to PRSS1, there are several other rarer mutations in genes like chymotrypsin C, SPINK1, and CFTR that have also been associated with hereditary pancreatitis. Work out of the University of Pittsburgh and others, has suggested a very interesting relationship between environmental factors, like cigarette smoking, which can dramatically increase the risk of pancreatic cancer in patients with hereditary pancreatitis and several other associated mutations.
There was an international consortium of experts that convened a few years ago to come up with a consensus of who should be screened for pancreatic cancer. Their recommendations have been released recently in Gut and efforts to screen these patients are underway in the CAPS (Cancer of the Pancreas Screening) study. Essentially the “who” that was defined by this study was patients who have two or more affected blood relatives, at least one of which are first degree with pancreatic cancer, patients who have Peutz-Jeghers syndrome, and those patients who have p16 mutations, BRCA 2 mutations, PALB2 mutations, and HNPCC associated mutation who have at least one first- or second-degree relative with pancreatic cancer. In terms of when to start screening, this was obviously in the realm of expert opinion, but the thought is at age 45 for familial pancreatic cancer or age 30 for Peutz-Jeghers syndrome patients. Finally, in terms of “how to screen,” the general consensus is via screening programs of MRI/MRCP and endoscopic ultrasound.
AR: What are the traditional modes of diagnostics for screening pancreatic cancer and how widely available are they?
KD: MRI and endoscopic ultrasound are the two traditional means of evaluating for pancreatic cancer. Endoscopic ultrasound has emerged as an extremely useful tool to delineate mass lesions utilizing sonography through the wall of the stomach or duodenum. In addition to delineating even small lesions, it can also be diagnostic by allowing a platform to perform fine needle aspiration (FNA). It is a relatively low-risk procedure and provides helpful confirmatory information in cases of atypical pancreatic ductal adenocarcinoma or in the desire to exclude inflammatory lesions that may be masquerading as a pancreatic mass. The accuracy ranges somewhere between 63% to 93% in assessing for preoperative resectability, and greater than 90% sensitivity in detecting even small pancreatic lesions. The limitations of the screening tool is that in chronic pancreatitis, cystic diseases of the pancreas or in patients who have abnormal or postsurgical anatomy, there may be difficulty in performing the test. In addition, there is inherently a degree of operator dependence with this technique. The second modality to image the pancreas is with MRI. There has been some debate in the radiology literature in comparing triple-phase CT to MRI/MRCP, and, depending on the center, MR may actually be better in finding smaller tumors or assessing resectability. There may be advantage for MRCP vs CT given the ability to image the duct with high resolution. Especially in cases of cystic lesions of the pancreas which may or may not communicate with the duct (as in mucinous cystic neoplasms or intraductal papillary mucinous neoplasms), being able to delineate associations with the main ducts or branch ducts can be extremely helpful both in a pre-surgical planning perspective and also in a diagnostic perspective. The benefit of MRI over EUS is that it is obviously less invasive and also that it provides serial imaging that can be examined via direct comparison year to year to assess for growth or change in any particular lesion.
I will say that the weakness of both of these general approaches is that neither address the molecular underpinnings of this disease process, that is to say the early onset of KRAS mutations, the emergence of preneoplastic microscopic lesions in the pancreas called PanINs (pancreatic intraepithelial neoplasms) or the possibility of early dissemination of pancreatic cancer cells into the bloodstream.
AR: And so if we go from the traditional modes, are there novel biomarkers that are available now or again in the horizon for pancreatic cancer screening?
KD: This has been a fertile ground of investigation. There have been numerous studies to validate various markers via immunohistochemistry in primary tumors or metastasis. Broadly speaking, these markers can be grouped as growth signals, growth inhibitory signals, factors that may help invasion, apoptosis, or angiogenesis, factors that may affect escape from immune surveillance, or important epigenetic modifications. Though numerous different markers have been examined, nothing has really emerged with significantly elevated hazard ratio to suggest that it is particularly effective as a screening tool.
In terms of the serum markers, the classic serum markers in clinical use today is sialylated Lewis blood group antigen CA19-9. However, its use is really limited to monitoring response to therapy and not as a diagnostic marker. Its limitations are ultimately due to its poor sensitivity and specificity. Elevations in CA19-9 are very common in patients with benign pancreatobiliary disorders, making it unsuitable as a screening test especially in these populations. Approximately 5% to 10% of the population does not express Lewis antigens and thus do not have detectable CA19-9. Furthermore, among resectable pancreatic cancer cases, only 65% of patients have elevations in CA19-9. It also does not do a good job in discerning patients with chronic pancreatitis from those with pancreatic ductal adenocarcinoma as 40% of chronic pancreatitis patients have an elevated CA19-9.
MUC1 carbohydrate antigens and related antibodies like the PAM4 antibody have been investigated but have not panned out as ideal screening targets.
There was excitement several years ago about the utility of screening for mesothelin, which is a protein that is found in most pancreatic cancer tissues but not normal pancreatic tissue. An ELISA has been developed to look at circulating protein levels in patients with other cancers as well like ovarian cancer. This differential expression of mesothelin has also been exploited as a potential target of novel immunotherapy by Dr. Carl June at the University of Pennsylvania. One aspect of his work has been on creating targeted T-cell therapies directed against mesothelin. Ultimately, the utility of mesothelin as a screening test is still quite investigational and the final data validating its use still remains to be fully reported.
In addition, several groups have looked at proteomic approaches as well as microRNAs in screening but a clear positive hit has not emerged from this. There is also a lot of interest in looking at pancreatic juice analysis, proponents of which say that the benefit of this technique is that it inherently evaluates not just a specific lesion but rather the entire gland, sampling for field defects of any cell in communication with the duct in theory. In this fashion, researchers from Johns Hopkins and elsewhere have been looking for genetic and epigenetic markers like mutant KRAS, p53, and GNAS, but specificity and sensitivity remains suboptimal. Furthermore, obtaining pancreatic juice is somewhat complicated as it generally requires stimulation of the pancreas with secretin and an invasive procedure to collect the material.
AR: Do you think it is possible to develop a blood test looking for pancreatic cancer cells, and if so, how would one go about doing it?
KD: Dr. Andy Rhim and Dr. Ben Stanger first demonstrated the possibility of evaluating for circulating pancreatic cancer cells in a mouse model of pancreatic cancer. In an elegant paper from Cell in 2012 Dr. Rhim and Dr. Stanger demonstrated that in approximately half of lineage labeled mice with no apparent pancreatic ductal adenocarcinoma, but only preneoplastic lesions (PanIN lesions), there were already single cell liver metastases present and circulating pancreatic cells in the bloodstream of these animals. This startling finding opened the doors to the possibility that the transition from an epithelial to a mesenchymal state, and the dissemination of these cells into the periphery may actually be an early event in pancreatic ductal adenocarcinoma which somewhat jives with what we see in the clinic. Patients present primarily with quite advanced disease, and furthermore, many patients blossom metastases that are not clearly evident preoperatively in the postoperative period. Perhaps the reason that so many patients present with metastatic pancreatic cancer is not because it is the end-stage of the disease process, but rather something that starts to occur at the earliest stages of the disease and progresses as the primary tumor progresses. Similarly, perhaps patients with no evident metastasis by imaging modalities have micro or “nano” metastases already present that only become evident with time. Subsequently, work by Dr. Rhim and Dr. Stanger have tried to translate these findings into human patients, to see if this observation of circulating tumor cells are something that can be seen in at-risk populations. In a recent paper in Gastroenterology from earlier this year, they looked at patients with cystic lesions of the pancreas, and in about a third of those patients, they found identifiable circulating pancreatic cells. In the same study, they found that the majority of pancreatic ductal adenocarcinoma patients also had circulating pancreatic cancer cells in comparison to control patients undergoing routine colonoscopy screening who had no noticeable circulating cells. This is very exciting as this technique may be a means to identify those individuals with lesions that have dissemination potential and may in fact be the ones who we should focus our attention and our potential therapy for. However, the limitation is that we still do not fully know what having these cells in the periphery actually means. We do not know if in those one-third of patients with cystic lesions of the pancreas whether the presence of circulating cells actually correlates to a pathophysiologic process. For that matter, we do not know whether or not someone can go on having a few circulating cells in the periphery for years and years without any sort of meaningful risk of disease progression, or what number of cells in the periphery is “too many.” We are still very early in this field of circulating tumor cells not just in pancreas cancer, but in I would say all cancer types, but it is certainly very exciting that with new microfluidic technologies and nano technologies, we are able to identify single cells that are able to escape into the periphery. Hopefully, this will provide information not only useful for our basic understanding of how cancer can spread from a primary site to a distant site but also for our patients in translating these techniques, hopefully, into the dream of a blood test that would be able to identify those patients who are at risk for developing pancreatic cancer.
|“We are still very early in this field of circulating tumor cells not just in pancreas cancer, but in I would say all cancer types, but it is certainly very exciting that with new microfluidic technologies and nano technologies, we are able to identify single cells that are able to escape into the periphery.”|
AR: It is very exciting about the technology underlying circulating tumor cells and circulating epithelial cells. Are there other blood-based tests that are being evaluated for patients with pancreatic cancer and/or at risk for pancreatic cancer?
KD: I would say that the most promising technology in addition to circulating tumor cells or circulating epithelial cells is circulating tumor DNA or ctDNA. This work has primarily been pioneered at Johns Hopkins by Luis Diaz. In several different epithelial cancer types and non-epithelial cancer types researchers have investigated for the presence of ctDNA. Dr. Diaz recently published some of his data in Science Translational Medicine from earlier this year. Essentially, what they found was that in nearly 100% of patients with bladder, colorectal, or gastroesophageal cancers there was detectible circulating tumor DNA that was able to be found. The ctDNA that they were looking at were normally about 180 base pairs, and about 2 mL of blood is all they needed for their test. In patients with pancreatic cancers, non-metastatic tumors had ctDNA detectable in only about 48% of tumors whereas metastatic patients with widely spread tumors had a much higher percentage of circulating tumor DNA. Interestingly in terms of understanding the biology of why there is a disparity of ctDNA in different epithelial tumors, gliomas and other brain tumors had very, very low circulating tumor DNA, suggesting that the blood brain-barrier and sanctuary sites may prevent dissemination. It has been clear that the biggest possible benefit of this technology may be amongst patients who are postoperative. In looking at ctDNA and CEA in colorectal cancer, these researchers found undetectable ctDNA was 100% sensitive and specific for recurrence-free survival. This would be a great technology to be able to utilize in pancreatic cancer patients as it always remains a question after a patient undergoes a Whipple surgery as to whether or not they will recur. In addition, these researchers were able to utilize the ctCNA as essentially a liquid biopsy. For example, in colorectal cancer patients they were able to identify K-RAS mutations to predict EGFR escape emerging; one could similarly see how this could be highly useful in finding pancreatic cancer patients who may become resistant to gemcitabine for example. Overall, I think that circulating tumor DNA will probably not be useful for the general population and is probably of limited sensitivity in patients who have localized pancreatic cancer, however it may definitely have a role in patients with metastatic disease or particularly in those individuals who are being treated and may be needed to be monitored while on treatment.
AR: That is very comprehensive. I hope our audience will find these answers to be extremely useful in terms of the landscape for diagnostics in pancreatic cancer for patients at risk, for patients with average risk, what we have been using over time, and where the opportunities lie going forward. Switching gears, what advances are emerging for novel diagnostics in patients with cystic lesions of the pancreas? These were quite different than what we have been talking about in terms of pancreatic ductal adenocarcinoma, the most common type of pancreatic cancer. Could you describe some more about what is meant by cystic lesions of the pancreas? What is the likelihood that they may become malignant or cancerous and how can we diagnose these cystic lesions?
KD: Cystic lesions of the pancreas are increasingly being identified as essentially a different disease entity entirely from pancreatic ductal adenocarcinoma, which is what we traditionally think of as pancreatic cancer. They are an entirely distinct set of lesions that are, broadly speaking, divided between benign lesions like serous cystadenoma (SCA) or simple cysts and those with malignant potential like mucinous cystic neoplasms (MCNs), IPMNs (intraductal papillary mucinous neoplasms) as well as other rare cystic lesions of the pancreas, including syndromically associated pancreatic cysts (ie, Von Hippel-Lindau) and those associated with cystic neuroendocrine tumors. As opposed to traditional PDAC, IPMNs for example have been shown to develop into two types of cancers – tubular carcinoma and colloid carcinomas of the pancreas. Being able to assess first which type of cystic lesion you are dealing with and second, assessing that lesion’s individual risk for developing cancer is emerging to be an extremely challenging clinical task. Clinical criteria, notably the Sendai Guidelines first published in 2006 that were updated in 2012 as the Fukuoka Guidelines, have attempted to address the question of how to manage IPMN that may or may not warrant surgical resection. What is becoming clear is that while there is certainly a malignant potential for patients with IPMNs, their definitive malignant fate is anything but certain. There is an increasing appreciation for the biology that underlies these lesions. There are epithelial subtypes of IPMN, each of which carries a different risk of development of pancreatic cancer and anatomic subtypes that have demonstrated that lesions associated with the main duct have an increased risk of neoplasia. There has been an intense desire to augment the adopted clinical criteria to identify those lesions that would particularly benefit from surgical resection. These attempts have included things like pancreatic juice analysis that I had alluded to earlier for things like KRAS, TP53, and GNAS mutations. In addition, investigators have looked at methods to evaluate a cyst fluid aspirated via endoscopic ultrasound for different markers, proteins, gene mutations, and other targets but no clear definitive panel has emerged to be truly effective. Furthermore, traditional cytology has also been a mainstay in cyst fluid analysis and while highly specific, it lacks sufficient sensitivity to be used alone. So, this actually remains a very hotly debated, discussed, and researched field to try and find an effective screening test. The need for improved diagnostics in this space is particularly dire because as patients in the community are getting more and more CT scans for many unrelated indications, the incidental identification of IPMNs is becoming increasingly common. In fact, studies of asymptomatic patients undergoing CT scans of the abdomen have suggested an overall incidence of IPMN as high as 2-3% of the general population and over 8% of patients over 80. However, only a tiny fraction of these lesions harbor any malignant potential. Thus, more effective tools are needed to stratify those patients that would benefit from surgical management, Whipple surgery, from those that would benefit from watchful waiting or quite frankly no watchful waiting at all. This is still an active area of interest certainly of mine and many others in the field.
AR: I am appreciative on behalf of providers and patients for your very thoughtful and comprehensive answers. You have conveyed a great deal of information especially the excitement surrounding the future of diagnostics in pancreatic cancer as well as information about cystic lesions, and we will look forward to your participation and contributions in the research underlying these domains in the future. Thank you.
- Howlader N, Noone AM, Krapcho M, et al. eds. SEER Cancer Statistics Review, 1975-2010, National Cancer Institute. Available at: https://seer.cancer.gov/csr/1975_2010/
- Winter J, Cameron J, Campbell K, et al. 1423 Pancreaticoduodenectomies for Pancreatic Cancer: A Single-Institution Experience. J Gastrointest Surg. 2006;10:1199-1211.
- Birkmeyer JD, Siewers AE, Finlayson EVA, et al. Hospital volume and surgical mortality in the United States. N Engl J Med. 2002;346:1128-1137.
- Ho V, Heslin MJ. Effect of hospital volume and experience on in-hospital mortality for pancreaticoduodenectomy. Ann Surg. 2003;237:509-514.
- Rustgi AK. Familial pancreatic cancer: genetic advances. Genes Dev. 2014;28:1-7.
- Canto MI, Harinck F, Hruban RH, et al. International Cancer of the Pancreas Screening (CAPS) Consortium summit on the management of patients with increased risk for familial pancreatic cancer. Gut. 2013;62(3):339-347.
- Dewitt J, Devereaux BM, Lehman GA, et al. Comparison of Endoscopic Ultrasound and Computed Tomography for the Preoperative Evaluation of Pancreatic Cancer: A Systematic Review. Clin Gastroenterol Hepatol. 2006;4:717-725.
- Soriano A, Castells A, Ayuso C, et al. Preoperative staging and tumor resectability assessment of pancreatic cancer: prospective study comparing endoscopic ultrasonography, helical computed tomography, magnetic resonance imaging, and angiography. Am J Gastroenterol. 2004;99:492-501.
- Gress FG, Hawes RH, Savides TJ, et al. Role of EUS in the preoperative staging of pancreatic cancer: a large single-center experience. Gastrointest. Endosc. 1999;50:786-791.
- Andersson M, Kostic S, Johansson M, et al. MRI combined with MR Cholangiopancreatography versus helical CT in the evaluation of patients with suspected periampullary tumors: a prospective comparative study. Acta Radiol. 2005;46:16-27.
- Waters JA, Schmidt CM, Pinchot JW, et al. CT vs MRCP: optimal classification of IPMN type and extent. J Gastrointest Surg. 2008;12:101-109.
- Ansari D, Rosendahl A, Elebro J, et al. Systematic review of immunohistochemical biomarkers to identify prognostic subgroups of patients with pancreatic cancer. Br J Surg. 2011;98:1041-1055.
- Goggins M. Identifying molecular markers for the early detection of pancreatic neoplasia. Semin Oncol. 2007;34:303-310.
- Rhim AD, Mirek ET, Aiello NM, et al. EMT and dissemination precede pancreatic tumor formation. Cell. 2012;148:349-361.
- Rhim AD, Thege FI, Santana SM, et al. Detection of circulating pancreas epithelial cells in patients with pancreatic cystic lesions. Gastroenterology. 2014;146:647-651.
- Bettegowda C, Sausen M, Leary RJ, et al. Detection of circulating tumor DNA in early- and late-stage human malignancies. Sci Transl Med. 2014;6:224ra24.
- Tanaka M, Fernández Del Castillo C, Adsay V, et al. International consensus guidelines 2012 for the management of IPMN and MCN of the pancreas. Pancreatology. 2012;12(3):183-197.
- Tanaka M, Chari S, Adsay V, et al. International Consensus Guidelines for Management of Intraductal Papillary Mucinous Neoplasms and Mucinous Cystic Neoplasms of the Pancreas. Pancreatology. 2006;6:17-32.
- Mino-Kenudson M, Fernandez-del Castillo C, Baba Y, et al. Prognosis of invasive intraductal papillary mucinous neoplasm depends on histological and precursor epithelial subtypes. Gut. 2011;60:17121-1720.
- Sahora K, Mino Kenudson M, Brugge W, et al. Branch Duct Intraductal Papillary Mucinous Neoplasms. Annals of Surgery. 2013;258:466-475.