Chapter 61A. Perspective on Pancreatic Neoplasms

Dr Wolfgang, Dr Schulick, and Dr Cameron provide an extremely comprehensive chapter on the evaluation and treatment of patients with cancer of the periampullary region and specifically the pancreatic head. The multidisciplinary working group from Johns Hopkins Hospital has made numerous contributions to the field of pancreatic cancer biology as well as the clinical management of patients with this disease. As all of you know, over the past two to three decades, Dr Cameron has demonstrated how to surgically manage patients with pancreatic cancer to achieve optimal outcome. Importantly, there has been a tremendous advance in both the understanding of the molecular biology of pancreatic cancer as well as our ability to accurately image the pancreas and periampullary region prior to surgery. Advances in both computed tomography (CT) and magnetic resonance imaging (MRI) have allowed for accurate assessment of critically important tumor-vessel relationships. Such accurate assessment of the relevant anatomy is important for both pretreatment staging and for planning the technical steps in performing pancreaticoduodenectomy, especially if vascular resection and reconstruction may be indicated. Although very experienced surgeons such as the authors can accurately assess resectability at the time of laparotomy, the ability to preoperatively classify patients as resectable, borderline resectable or locally advanced allows for the appropriate triage of patients for optimal treatment sequencing (surgery first or after neoadjuvant therapy), the evaluation of patients for investigator-initiated and cooperative group clinical trials, and for the referral of patients to higher volume centers. Indeed, to the extent that outcome is improved for patients with localized disease at high-volume centers (by high-volume surgeons), patients will need to be accurately staged (CT imaging) and, when necessary, have biliary stents placed safely in order to facilitate referral to a specialty center. The ability to perform endoscopic ultrasound (EUS)–guided fine-needle aspiration (FNA) biopsy will prevent diagnostic uncertainty and allow for medical oncology consultation and multidisciplinary care.

Fortunately, the last decade has witnessed the development of consensus for the CT staging of pancreatic cancer. In an attempt to clarify the anatomy of resectable, borderline resectable, and locally advanced disease, Varadhachary and colleagues from the University of Texas M.D. Anderson Cancer Center proposed an objectively defined, CT-based classification that distinguished borderline resectable from both resectable and locally advanced pancreatic cancer.1 The Varadhachary definitions considered venous abutment and encasement (without occlusion) to be resectable, in the absence of tumor extension to the celiac or superior mesenteric (SMA) arteries. However, this definition was developed for the conduct of clinical trials of neoadjuvant treatment sequencing and was not intended to support a surgery-first strategy for patients who may require vascular resection and reconstruction. The Varadhachary definitions also assumed the technical capability to resect and reconstruct the superior mesenteric-portal vein (SMPV) confluence when necessary and that the major determinant of margin status (R status) was the tumor-artery (celiac, hepatic, SMA) relationship (Table 61A-1). Katz and colleagues in 2008 reported 160 patients with borderline resectable disease (using the Varadhachary definition) and introduced three subtypes of the borderline category, often referred to as Katz types A, B, and C.2 Type A patients were those with borderline resectable tumor anatomy as defined in the Varadhachary manuscript. Type B patients were borderline resectable because of a concern for possible extrapancreatic metastatic disease and included those with CT findings suspicious for, but not diagnostic of, metastatic disease. One could also add to this group those patients with very high cancer antigen 19-19 (CA19-9) levels (important that the CA19-9 be measured when the serum bilirubin is normal). Type C patients were borderline resectable due to a marginal performance status or significant preexisting medical comorbidity thought to require protracted evaluation that precluded immediate surgery. By definition, type C patients were thought to have reversible causes of their current symptoms such as hyperbilirubinemia-induced anorexia and fatigue. Katz and colleagues provided compelling data in support of induction chemotherapy (followed by chemoradiation) for patients with borderline resectable disease. Of equal importance, they defined borderline resectable disease in all three forms that we see clinically: anatomic (local tumor anatomy), oncologic/biologic (possible advanced disease not fully apparent on imaging), and physiologic (marginal performance status).

For those surgeons who recommend a surgery-first strategy to patients with localized, potentially resectable pancreatic cancer, the CT definition of what should be considered “resectable” for which immediate surgery may be considered is often more limited than the M.D. Anderson definitions above. The AHPBA-SSO-SSAT consensus panel definition of resectable (as distinct from borderline resectable) included only those tumors with no evidence (on CT) of even abutment or distortion of any aspect of the SMPV confluence as well as no extension to the adjacent mesenteric arteries.3 At present, there appears to be growing consensus for a fairly narrow definition of “resectable” when a surgery-first strategy is planned. In the absence of neoadjuvant therapy, resectable disease is defined as the absence of tumor extension to the adjacent visceral arteries (SMA, celiac, hepatic) and the absence of tumor-induced unilateral shift, distortion, or narrowing of any aspect of the SMPV confluence; such patients may be taken directly to surgery if a suitable clinical trial is not available.3 In contrast, patients with borderline resectable pancreatic cancer, as defined by the AHPBA-SSO-SSAT consensus panel, should be treated with induction therapy before surgery.

What is not clear at present is how best to treat patients with borderline resectable pancreatic cancer before consideration of surgery. Many incorporate a period of induction systemic therapy, especially in those with arterial abutment, to include at least 2 months of chemotherapy before chemoradiation.2 Emerging clinical trials, as well as off-protocol therapy, will likely include what may prove successful in metastatic disease. For example, gemcitabine plus nab-paclitaxel (Abraxane) or FOLFIRINOX (5-flourouracil [5-FU], leucovorin, irinotecan, and oxaliplatin).4,5 The length of induction systemic therapy, the timing and dose of radiation therapy, and the optimal postoperative systemic therapy remain undefined. The published experience to date with neoadjuvant treatment sequencing for resectable and borderline resectable pancreatic cancer has largely used chemoradiation as a component of the treatment program. Experience using chemotherapy alone is relatively untested, the obvious concern being the risk for local tumor recurrence. The first national trial of neoadjuvant therapy for resectable pancreatic cancer (the American College of Surgeons Clinical Oncology Group Z5041 [ACOSOG Z5041]) does not incorporate radiation therapy; the local recurrence rate will be an important endpoint with major therapeutic implications.

Probably the most important technical aspect of pancreaticoduodenectomy is the dissection of superior mesenteric artery.6 In general, exposure of the SMA is facilitated by complete mobilization of the SMPV confluence to the patient's left. This allows for careful separation of the uncinate process from the jejunal branch of the SMV and, ultimately, exposure of the SMA. Our current understanding of the pathophysiology of local recurrence following pancreaticoduodenectomy (with or without multimodality therapy) is microscopic infiltration of the autonomic neural sheath that surrounds the SMA. Adenocarcinoma of the pancreas has a predisposition to spread along neural tissue, and this is likely responsible for the high frequency of local recurrence. As our systemic therapies become more effective, local recurrence may become a more dominant pattern of failure. It is impossible to argue with the technique described by Dr Wolfgang as the operation of pancreaticoduodenectomy is arguably done better at Johns Hopkins Hospital than any other institution in the world. Students, residents, and even experienced surgeons will benefit greatly by careful review of this chapter.

Chapter 58 by Dr Maley and Dr Yeo is equally comprehensive in their superb discussion of cystic neoplasms of the pancreas. They focus predominantly on serous cystadenoma, mucinous cystic neoplasm (MCN), and intraductal papillary mucinous neoplasm (IPMN). Regarding serous cystadenoma, this histology demonstrates fascinating tumor biology. As mentioned by the authors, it is generally felt that serous cystadenomas do not have the biologic ability to metastasize to distant organs or regional lymph nodes. However, they can be locally invasive and erode into adjacent bowel (duodenum, transverse colon, stomach) and occasionally can obstruct the splenic vein (resulting in sinistral portal hypertension) or the superior mesenteric and/or portal veins (resulting in extrahepatic portal hypertension). Importantly, the diagnosis of (microcystic) serous cystadenoma can usually be made on high-quality CT imaging with or without the additional benefit of EUS because of its characteristic imaging appearance (unless the serous cystadenoma is macrocystic). When referring a patient for EUS to confirm a diagnosis of serous cystadenoma, we would recommend an FNA biopsy if the EUS is not consistent with this diagnosis or there appears to be discrepancy between CT or MRI imaging and the EUS appearance. As noted by the authors, serous cystadenomas are characterized by a cyst fluid CEA level that is usually undetectable or very low (<5 ng/mL). In 2011, the diagnosis of a serous cystadenoma is usually not difficult; however, knowing when to intervene with surgery is often challenging. As patient age and operative risk (medical comorbidities) increase, the benefit to surgery in an otherwise asymptomatic patient may be low. For example, it is relatively easy to understand a recommendation for surgery in an otherwise completely healthy 60-year-old patient with a 5- to 6-cm serous cystadenoma.7 However, the same pancreatic tumor in a 75-year-old patient with one or two coronary stents and a relatively sedentary lifestyle may not be the correct approach. In our practice, we try to carefully weigh risk versus benefit in asymptomatic patients.8 In addition, for serous cystadenomas that are less than 4–5 cm in size, we usually require that they demonstrate growth, over a period of observation, before proceeding with surgery. Our underlying philosophy is to avoid surgery-related mortality and major morbidity, especially in patients who are asymptomatic with a tumor histology such as serous cystadenoma that poses no risk for distant metastases.

The discussion on MCN is particularly well done and as emphasized my Dr Maley and Dr Yeo, ovarian stroma is now required to secure a diagnosis of MCN. While the relationship between size of the MCN and malignant potential is not perhaps as well worked out as with IPMN, it is reasonable to utilize the same 3-cm rule. For example, in the absence of a solid component or mural nodule, mucinous neoplasms less than 3 cm in diameter, that are radiographically consistent with MCN, can usually be safely observed. A recommendation for observation is all the more compelling in patients of advanced stage and with medical comorbidities. Obviously, it is sometimes difficult to differentiate an MCN from a unifocal branch duct IPMN in a woman. Importantly, as mentioned by the authors, MCNs occur in the pancreatic body and tail and are exceedingly rare in the pancreatic head or uncinate process. In our practice we would consider an MCN in the pancreatic head or uncinate process to represent an IPMN. What remains controversial is the surgical approach to MCNs of presumed low malignant potential. For example, consider an otherwise healthy 50-year-old woman with a 3-cm presumed MCN in the pancreatic body who has undergone EUS-guided FNA biopsy and the mucinous nature of the cyst fluid has been confirmed. In this patient with no evidence of a solid component or mural nodule on CT or EUS, surgery is recommended because of the patient's young age as well as the size of the MCN. Because this is a premalignant neoplasm, it would be very difficult to support a recommendation of observation in a patient of this age. If her serum level of CA19-9 was also normal, this finding adds further support to there being no invasive component to the tumor at this time. The compelling question then is whether she should undergo middle segment pancreatectomy, distal pancreatectomy (with or without splenic preservation), or enucleation. In patients without cancer, we do need to pay more attention to preservation of islet cell mass in an effort to avoid the intermediate and long-term complications of insulin-dependent diabetes (in addition to the lifestyle changes introduced with insulin dependence). Our choice for operation in this patient would be a middle segment pancreatectomy with pancreaticojejunostomy for the distal pancreas and creation of a serosal patch sewn to the proximal pancreatic transection site. This would hopefully minimize the risk for anastomotic leak, preserve islet cell mass, and ensure that the lesion is completely excised with negative margins. We have not yet adopted enucleation as a routine part of our practice when dealing with mucinous (premalignant) neoplasms.

When dealing with a patient who has presumed IPMN, especially those with branch duct disease, the Sendai guidelines have now been widely incorporated into clinical practice.9 As noted by the authors, use of these guidelines will result in a slightly higher-risk (for invasive adenocarcinoma) population being considered for surgery. By definition, the low-risk patients would be treated with at least a period of observation until the size of the cyst or the CT characteristics (suggestion of a solid component) prompt surgical intervention. Such a strategy is designed to avoid surgery, and its associated risk for mortality and morbidity, in patients with small cystic neoplasms who have no chance of harboring an invasive cancer. In our practice, enucleation would rarely be considered as we do not operate on the cystic neoplasms that would be considered most appropriate for enucleation (those of very low risk). However, the increasing use of cross-sectional imaging has resulted in many more patients being diagnosed with cystic neoplasms of the pancreas. Would it be reasonable to consider a lesser procedure (enucleation or endoscopic alcohol ablation) in smaller branch duct IPMNs that may be diagnosed in younger patients? Is there a role for enucleation or ablative therapies in patients where the risk for invasive carcinoma is approaching zero? This is now a subject of debate at national meetings, and I suspect represents a fruitful area for cooperative group clinical trials.

Fortunately, when dealing with a patient who has IPMN and requires surgery, the need for total pancreatectomy is uncommon. If the right or left side of the pancreas requires resection, we commonly send the pancreatic transection margin for frozen-section analysis. Work from our group and others has demonstrated that it is probably unnecessary to chase a margin with low-grade dysplasia (PanIN-1) as this can result in the unnecessary resection of additional pancreas.10 Importantly, at many institutions, there is only modest expertise in the interpretation of frozen-section evaluation of pancreatic transection margins. Surgeons should be cautioned to avoid overaggressive resection of grossly normal pancreatic parenchyma based on frozen-section evaluation of pancreatic transection margins especially when dealing with IPMN. Another area of operative/technical challenge includes IPMN involving the neck of the pancreas; should one resect the right or the left side of the gland? In this situation we would typically divide the pancreas to the right of the neoplasm at the junction of the head and neck of the pancreas. We would then send this margin for frozen-section evaluation before committing the patient to an extended distal pancreatectomy (in those cases where middle segment pancreatectomy is not preferred). When performing an extended distal pancreatectomy, one needs to be certain that the proximal pancreatic transection margin will be negative, especially if the patient could also be treated with an extended pancreaticoduodenectomy and thereby preserve some islet cell mass. In general, the preservation of some islet cell mass does facilitate improved blood sugar control even if not obviating the need for insulin. Last, it is important to note that patients who undergo surgery for IPMN (in contrast to those who undergo pancreatectomy for MCN) do require long-term follow-up. We typically obtain an MRI of the abdomen 2–4 months following surgery and our next scan would typically occur 1 year later. If there is no evidence of an abnormality in the remaining pancreas, our MRI imaging interval is in the range of 12–24 months depending on the histology of the previous resection, patient age, and the general health (performance status) of the patient.

As described by Dr Riall and Dr Evers, pancreatic neuroendocrine tumors (pNETs) are usually low- to intermediate-grade tumors arising from the pancreatic islets. They are also known as pancreatic endocrine tumors, islet cell carcinoma, and pancreatic carcinoid tumors. The current preferred nomenclature is pancreatic neuroendocrine tumors or pNETs.11 The biology of this class of tumors is both unique and fascinating. For example, why should sporadic, nonmetastatic insulinomas virtually never develop distant recurrence and only very rarely recur locally (virtually all local recurrences are secondary to incomplete enucleation)? Metastatic insulinoma is very rare and, when seen, it is always synchronous at the time of diagnosis; we have not seen a case of metachronous metastases. In contrast, patients with multiple endocrine neoplasia type I (MEN1) who have nonfunctioning pNETs have a risk of metastatic spread that appears related to the size of the primary tumor in the pancreas. Patients with primary tumors less than 2.5 cm in size rarely have associated liver metastases.12 When weighing the risk of long-term insulin-dependent diabetes with the risk for distant metastases, we often observe small (<1–1.5 cm) nonfunctioning pNETs in young MEN1 patients. Further, why should the biology of patients with Zollinger-Ellison syndrome differ based on whether the primary tumor is in the pancreas or the duodenum and what determines where the tumor arises? Equally mystifying is why duodenal gastrinomas are so small, often less than 1 cm in diameter and rarely associated with liver metastases. Gastrinomas, when located in the pancreas, are usually found within the pancreatic head or uncinate process (gastrinoma triangle) and those 3 cm in size and larger are frequently associated with liver metastases. Consistent with the biology of duodenal gastrinoma, patients with carcinoid tumors of the duodenum also rarely have synchronous or metachronous liver metastases even though lymph nodes metastases are very common.13 Indeed, the biologic explanation for the varied metastatic potential of functioning and nonfunctioning pNETs is an area of active investigation. Recent studies suggest that pancreatic, but not duodenal, gastrinomas express pancreatic-duodenal homeobox1 (Pdx1) and that only duodenal gastrinomas express sonic hedgehog signifying a different molecular origin for the two tumors.14

When evaluating a patient with hypergastrinemia, it is important to remember that the major cause of hypergastrinemia is parietal cell dysfunction, resulting in achlorhydria and pernicious anemia. Such patients can be differentiated from those with Zollinger-Ellison syndrome by the absence of gastric acid production. In the outpatient center, placement of a nasogastric tube with aspiration of gastric juice for pH testing will easily make this diagnosis. We frequently see patients who have elevated serum gastrin levels from either concomitant administration of a proton pump inhibitor or because of parietal cell dysfunction. A pancreatic or duodenal producing tumor is a much less frequent cause of hypergastrinemia. Importantly, consistent with the optimal operative management of virtually all pNETs, regional lymphadenectomy is an important part of the operative procedure for patients with gastrinoma. Patients with functioning or nonfunctioning pNETs can have persistent or recurrent disease in regional lymph nodes in the absence of liver, bone, or lung metastases. Careful attention to regional lymphadenectomy is an underemphasized and very important component of their surgical management.

As noted by the authors, patients with insulinoma virtually always have an insulin level greater than 3 mIU/mL (usually >6 mIU/mL) when the blood glucose is less than 40–45 mg/dL and an insulin to glucose ratio of 0.3 or less reflects the inappropriate secretion of insulin at the time of hypoglycemia. The conclusion of any carefully done observed fast, performed to confirm the diagnosis of hyperinsulinism, includes the administration of 1 mg of glucagon given intravenously. It is critically important to confirm the diagnosis of insulinoma by allowing the glucose to decline to a level of less than 45 mg/dL (at which point the patient is usually symptomatic) and observing the relief of symptoms with the administration of glucagon. Intravenous glucagon is associated with an elevation of serum glucose of approximately 20 mg/dL. The reversal of hypoglycemia with glucagon confirms that hypoglycemia is insulin mediated. In contrast to gastrinomas that usually occur in the duodenum and pancreatic head/uncinate, insulinomas may develop anywhere throughout the pancreas and do not arise in the duodenum. In the absence of MEN1, the overwhelming majority of insulinomas are unifocal. As previously mentioned, if metastatic disease is not seen at the time of diagnosis, metachronous distant metastases from a presumed benign insulinoma does not occur. However, if the insulinoma is incompletely enucleated, a local recurrence can develop; such local recurrences may not occur for years after the primary operation. When we perform an enucleation of a benign insulinoma, we typically use bipolar forceps and are extremely careful to avoid violation of the tumor capsule. The advantage of bipolar cautery is that the operative field remains dry and that one can appreciate the junction of the tumor capsule and the normal pancreatic parenchyma. Because the pancreas is highly vascular, it is critically important to keep the operative field as dry as possible. Proper technique for enucleation is much more important than whether the operation is done laparoscopically or open. In the event of a pancreatic fistula, the presence or absence of an abdominal incision becomes insignificant. When an enucleation is performed, the anatomy of the primary tumor in relation to the pancreatic duct should be appreciated on preoperative imaging, and, if needed, this important anatomic relationship can be confirmed with intraoperative ultrasound. If one performs a very large enucleation or injures the pancreatic duct, a Roux-en-Y limb of jejunum can be used to create a pancreaticojejunostomy. For large defects in the pancreas, we have made liberal use of Roux-limbs for internal drainage.

Regarding MEN1, all at-risk patients should undergo genetic testing. In addition to MEN1, pNETs can occur in association with tuberous sclerosis, neurofibromatosis, and von Hippel-Lindau (vHL) syndrome. As previously mentioned, there is a defined association between tumor size and risk for liver metastases in MEN1 patients with nonfunctioning pNETs. Balancing the risk for insulin-dependent diabetes with the risk for metachronous liver metastases is indeed a difficult challenge. In our practice, the timing of pancreatic surgery in MEN1-associated nonfunctional pNETs is based on the tumor biology seen within the family (frequency of metastases and death secondary to metastatic pNET), the age of the patient, and the size of the pancreatic tumors. For patients who have a family history of metastatic neuroendocrine carcinoma, we begin to discuss surgery when the pNET becomes greater than 1 cm in size and argue against continued observation when the largest pNET reaches 1.5–2 cm in size. Dr Norman Thompson was the first to recommend a partial pancreatectomy for patients with MEN1.15 He described the combination of distal subtotal pancreatectomy with enucleation of identified lesions in the pancreatic head and uncinate process. This operation also included regional lymphadenectomy. The “Thompson procedure” aimed to decrease the risk of metachronous liver metastases while preserving some islet cell function, hopefully enough to avoid insulin dependent diabetes. Patients with aggressive MEN1-associated pNETs will often eventually require completion of total pancreatectomy, as one would expect metachronous recurrence within the remaining pancreas after the first operation. However, one would hope that the second operation to complete the total pancreatectomy would not occur until many years after the first operation, thereby avoiding insulin dependence until later in life at which time the long-term complications of type 1 diabetes are less unlikely. Because the natural history of MEN1-associated pNETs becomes clear only after decades of follow-up, we still have much to learn in the management of these patients.

Importantly, the last decade has witnessed an explosion in the understanding of the biology of pNETs and their potential response to targeted therapies. Tyrosine kinase inhibitors such as sunitinib and the mammalian target of rapamycin (mTOR) inhibitor everolimus have shown activity in patients with metastatic pNETs and have stimulated renewed interest in translational research and novel therapies for this disease.16,17 Importantly, many patients who develop metachronous recurrence years after pancreatectomy for a nonfunctioning pNET will have somewhat indolent disease. The majority of these patients will be treated with octreotide, and disease stabilization is frequently seen.18 The optimal management of patients with low-volume metastatic pancreatic neuroendocrine carcinoma requires thoughtful multidisciplinary input from oncologists, interventional radiologists, and frequently surgeons. Liver resection, systemic therapy (to include octreotide), and ablative therapies both intra-arterial and percutaneous can all add to improved length and quality of life for affected patients.

Last, in our practice we rarely proceed with surgery in a patient with a nonfunctioning pNET if a gross, complete resection cannot be performed. However, in the setting of known metastatic disease or a large, borderline resectable primary tumor, we frequently use neoadjuvant therapy (cytotoxic and/or biologic). In the setting of synchronous liver metastases, a one- or two-stage surgical approach can be used with or without neoadjuvant therapy. In contrast to exocrine pancreatic cancer, we frequently resect neuroendocrine liver metastases. When dealing with a resectable primary tumor and resectable liver metastases, we usually remove the pancreatic tumor first; if the pancreatectomy goes well, some or all of the liver disease can be addressed at the same operation. If the magnitude of surgery required for the pancreatic primary is too large, the liver surgery should be performed at a second stage. As one can imagine, there are various degrees of complexity regarding combined pancreas-liver resections. For patients with more advanced liver metastases, in whom a future liver remnant can be cleared at the time of the initial pancreatectomy, one can even consider portal vein embolization and second-stage extended hepatectomy. This of course assumes excellent health of the patient, a reasonable age category, and the absence of medical comorbidities. Indeed, future options to combine novel systemic therapies with advanced surgical techniques hold great promise for patients with pNETs, even those with disease that may have been considered nonoperable in the past.