Abstract
Context.—Pancreatic endocrine neoplasms (PENs) are diagnostically challenging tumors whose natural history is largely unknown. Histopathology allows the distinction of 2 categories: poorly differentiated high-grade carcinomas and well-differentiated neoplasms. The latter include more than 90% of PENs whose clinical behavior varies from indolent to malignant and cannot be predicted by their morphology.
Objectives.—To review the literature and report on additional primary material about the clinicopathologic features, classification, staging, grading, and genetic features of PENs.
Data Sources.—Literature review of relevant articles indexed in PubMed (US National Library of Medicine) and primary material from the authors' institution.
Conclusions.—The diagnosis of PEN is generally easy, but unusual features may induce misdiagnosis. Immunohistochemistry solves the issue, provided that the possibility of a PEN has been considered. Morphology allows the distinction of poorly differentiated aggressive carcinomas from well-differentiated neoplasms. The World Health Organization classification criteria allow for the discernment of the latter into neoplasms and carcinomas with either benign or uncertain behavior. The recently proposed staging and grading systems hold great promise for permitting a stratification of carcinomas into clinically significant risk categories. To date, inactivation of the MEN1 gene remains the only ascertained genetic event involved in PEN genesis. It is inactivated in roughly one-third of PENs. The degree of genomic instability correlates with the aggressiveness of the neoplasm. Gene silencing by promoter methylation has been advocated, but a formal demonstration of the involvement of specific genes is still lacking. Expression profiling studies are furnishing valuable lists of mRNAs and noncoding RNAs that may advance further the research to discover novel markers and/or therapeutic targets.
GENERAL FEATURES
Pancreatic endocrine neoplasms (PENs) are epithelial tumors with endocrine differentiation. They account for approximately 2% of all pancreatic neoplasms and commonly affect adults between the ages of 40 and 60 years with no sex predilection.1,2 At variance with the fast proliferating and deadly pancreatic ductal adenocarcinoma, the typical PEN grows slowly and impairs patient quality of life only very late in the course of the disease, even when metastatic. It is thus important to distinguish PEN from ductal adenocarcinoma, because its prognosis is largely more favorable.1,2
Pancreatic endocrine neoplasms are usually sporadic but may be part of hereditary syndromes mostly including multiple endocrine neoplasia type 1 (MEN-1) and, more rarely, von Hippel-Lindau (VHL) syndrome, neurofibromatosis type 1 (NF1), and tuberous sclerosis complex (TSC). Sporadic PENs are solitary, whereas the hereditary forms may be multifocal.
Patients seek medical assistance because of symptoms resulting from either hormonal hypersecretion or mass. Thus, PENs are clinically defined as functioning (syndromic) (F-PENs) or nonfunctioning (nonsyndromic) (NF-PENs), depending on the presence of a syndrome related to inappropriate hormone secretion. The clinical syndromes associated with the different types of F-PENs are summarized in Table 1. The patients presenting with metastatic or mass-related symptoms are those with NF-PENs and report abdominal pain, nausea, weight loss or, exceptionally, jaundice. Half of clinically observed PENs and more than 50% of surgically resected cases are NF-PENs.1–5
Main Clinicopathologic Characteristics of Functioning Pancreatic Endocrine Neoplasms (PENs)

Imaging procedures address the diagnosis by recognizing the characteristic hypervascular pattern of these lesions, which is present in about 70% of cases (Figure 1). Large size and smooth regular margins with lack of dilation of the biliary and main pancreatic ducts contribute to the differentiation of PEN from adenocarcinoma. Most PENs express receptors for somatostatin and can be easily identified by somatostatin receptor scintigraphy.4,6–8 Moreover, the current and more sensitive imaging techniques increasingly detect small and asymptomatic tumors, the so-called pancreatic endocrine incidentalomas.9–11
Hypervascular appearance of a pancreatic head mass (asterisk) on computed tomography scan. Figure 2. Left pancreatectomy. Pancreatic endocrine neoplasm presenting as solid, homogeneous, large mass in the tail of the pancreas. Figure 3. Left pancreatectomy. Pancreatic endocrine neoplasm with evidence of gross vascular invasion of the splenic vein (asterisk).
Hypervascular appearance of a pancreatic head mass (asterisk) on computed tomography scan. Figure 2. Left pancreatectomy. Pancreatic endocrine neoplasm presenting as solid, homogeneous, large mass in the tail of the pancreas. Figure 3. Left pancreatectomy. Pancreatic endocrine neoplasm with evidence of gross vascular invasion of the splenic vein (asterisk).
Most F-PENs, with the exception of insulinomas, are diagnosed when they are already malignant diseases, and liver metastases are common.12–15 Nonfunctioning PENs are also frequently malignant, as more than 50% of patients have liver metastases at diagnosis and almost 40% are not candidates for radical surgery because of either locally advanced disease or unresectable metastases.16–18 Patients with well-differentiated NF-PENs have a 5-year survival rate of approximately 65% and a 10-year survival rate of 45%.2,19
PATHOLOGIC FEATURES
Macroscopy
Pancreatic endocrine neoplasms can be located anywhere within the pancreas.
Common Type
Pancreatic endocrine neoplasms usually present as solitary, solid, homogeneous masses, usually from 1 to 5 cm in diameter, with rounded and sharp borders (Figure 2), rarely surrounded by a fibrotic pseudocapsule. Their expansive pattern of growth differs from the infiltrative pattern of the ductal adenocarcinoma. When located in the pancreatic head, they usually compress and deviate, but not infiltrate, the main pancreatic and biliary ducts. Also, PENs extending outside the pancreas usually displace rather than invade the adjacent structures including large vessels. Their color and consistency depend on the amount of stroma and vascularization; the usual PEN is rich in small vessels and poor in fibrotic stroma. The color varies from brown to reddish and the consistency is slightly firmer than that of the surrounding parenchyma. Rare masses appear hemorrhagic. Necrotic yellowish foci can be observed in large masses. Features of malignancy evident at macroscopic examination include involvement of perivisceral fat, mainly as satellite nodules, and invasion of duodenal wall or adjacent organs, common bile duct, spleen, or large vessels (Figure 3). Involvement of splenic vessels with thrombosis can cause splenic infarctions.
Uncommon Types
Pancreatic endocrine neoplasms may have unusual macroscopic aspects. They may mimic cystic or fibrotic tumors (Figure 4, A and B). Pancreatic endocrine neoplasms with cystic appearance may be either unilocular or multilocular and are filled with a clear or hemorrhagic fluid20–23 (Figure 4, A). These tumors are most difficult to diagnose preoperatively and are often mistakenly interpreted as being cystic neoplasia. Pancreatic endocrine neoplasms with fibrotic appearance have considerable fibrosis that results in firm consistency, whitish color, and ill-defined borders (Figure 4, B). These PENs look like a ductal adenocarcinoma. Cases of black or yellow PENs have been described. The “pigmented black PEN”24 is composed of cells with abundant intracytoplasmic lipofuscin. This tumor may mimic metastatic melanoma. The yellow “lipid-rich PEN” can mimic adrenal cortical neoplasia.25
Pancreaticoduodenectomy. A, Pancreatic endocrine neoplasm presenting as large unilocular cyst with fibrous pseudocapsule. B, Pancreatic endocrine neoplasm resembles ductal adenocarcinoma. Figure 5. A, Pancreatic endocrine neoplasm growing around and causing stricture of the main pancreatic duct and upstream ductal dilatation (asterisk). B, Nests of neuroendocrine cells in abundant stroma (hematoxylin-eosin, original magnifications ×40 [A] and ×100 [B]).
Pancreaticoduodenectomy. A, Pancreatic endocrine neoplasm presenting as large unilocular cyst with fibrous pseudocapsule. B, Pancreatic endocrine neoplasm resembles ductal adenocarcinoma. Figure 5. A, Pancreatic endocrine neoplasm growing around and causing stricture of the main pancreatic duct and upstream ductal dilatation (asterisk). B, Nests of neuroendocrine cells in abundant stroma (hematoxylin-eosin, original magnifications ×40 [A] and ×100 [B]).
PEN and the Main Pancreatic Duct
Pancreatic endocrine neoplasms rarely grow within the duct itself and, even more rarely, infiltrate the duct and cause strictures. Intraductal PENs are NF-PENs that grow within the main pancreatic duct without invading its epithelium and may be connected to an extraductal lesion. They have been described in case reports26–29 in which most cases were malignant. Pancreatic endocrine neoplasms causing duct strictures are lesions growing around the main pancreatic duct with extensions into its wall that cause abrupt-type strictures30,31 (Figure 5, A and B).
Microscopy
Most PENs are well-differentiated tumors whose highly characteristic histologic features permit the recognition of their endocrine nature. Broad variations in the architectural and cytological aspects can be encountered, particularly in larger tumors. Tumors smaller than 0.5 cm are called microadenomas, generally show a trabecular pattern, and are observed more frequently in patients with genetic disease.32,33
The Architectural Aspects
Pancreatic endocrine neoplasm typically has an organoid pattern of growth, characterized by solid nests and macrotrabecular or microtrabecular/gyriform patterns with cords, festoons, and ribbons (Figure 6, A). Glandular (Figure 6, B), acinar, and cribriform features can also be observed. Although one pattern is generally prevalent, more than one can be seen in different regions of the same tumor. A rich vascularization characterizes most PENs and is responsible for their hyperdense radiologic appearance. Typically, numerous small vessels encircle the neoplastic nests. These vessels are embedded in a variable amount of stroma (Figure 6, C), which rarely forms sclerohyaline bands and only occasionally shows calcified foci. The presence of amyloid is frequent in insulinomas. Necrosis can be present either as large and confluent areas (“infarctlike”), especially in large tumors, or as punctate foci recognized at microscopic observation in the center of neoplastic nests. No conclusion may ever be reached by tumor histologic analysis alone concerning the functional state or hormone type produced. The 2 exceptions are the stroma with amyloid deposits identifying insulinoma and the glandular pattern with psammoma bodies identifying somatostatinoma.
Pancreatic endocrine neoplasm with (A) trabecular and (B) glandular growth pattern. C, Pancreatic endocrine neoplasm rich in stroma (hematoxylin-eosin, original magnifications ×200 [A and B] and ×100 [C]). Figure 7. Pancreatic endocrine neoplasm composed of small- to medium-sized cells with salt-and-pepper chromatin and inconspicuous nucleolus (hematoxylin-eosin, original magnification ×400). Figure 8. Pancreatic endocrine neoplasm composed of cells with (A) prominent nucleoli and (B) atypical features (hematoxylin-eosin, original magnifications ×400).
Pancreatic endocrine neoplasm with (A) trabecular and (B) glandular growth pattern. C, Pancreatic endocrine neoplasm rich in stroma (hematoxylin-eosin, original magnifications ×200 [A and B] and ×100 [C]). Figure 7. Pancreatic endocrine neoplasm composed of small- to medium-sized cells with salt-and-pepper chromatin and inconspicuous nucleolus (hematoxylin-eosin, original magnification ×400). Figure 8. Pancreatic endocrine neoplasm composed of cells with (A) prominent nucleoli and (B) atypical features (hematoxylin-eosin, original magnifications ×400).
The Cytologic Aspects
Regardless of the growth pattern, the neoplastic cells have similar cytologic features. The classic PEN is composed of small- to medium-sized cells with eosinophilic to amphophilic and finely granular cytoplasm. The nuclei are usually centrally located, round or oval, uniform in size and show finely stippled “endocrine chromatin” referred to as salt-and-pepper (Figure 7). Nucleoli are inconspicuous or absent; rarely, they are prominent and incorrectly suggest an acinar cell carcinoma (Figure 8, A). In some tumors, the neoplastic cells show a plasmacytoid appearance due to peripherally located nuclei. When the tumor is made up of small cells with minimal cytoplasm, PEN may be confused with a high-grade small cell endocrine carcinoma, but the mitotic activity is low and necrosis is undetected. Sometimes nuclei show variable shapes and atypia (Figure 8, B), which is not a criterion for aggressiveness as it is in adenocarcinomas, where it correlates with prognosis.34
Unusual cytologic features can be occasionally seen in otherwise conventional PENs; when they become prevalent, the tumor is considered a morphologic variant. Oncocytic, lipid-rich, clear cell, and rhabdoid variants, as well as tumors with marked nuclear pleomorphism, have been reported2 (Figure 9, A and B). The clinical outcome of these tumors does not differ significantly from that of conventional PENs. The main reason for recognizing these variants is to avoid misdiagnosis with other pancreatic or extrapancreatic neoplasms. In fact, these unusual morphologies may obscure the endocrine nature of the neoplasm that can be readily established by immunohistochemistry, provided that the possibility of a PEN has been considered.
Pancreatic endocrine neoplasm with unusual cell types. A, Oncocytic cells. B, Lipid-rich cells (hematoxylin-eosin, original magnifications ×400). Figure 10. Poorly differentiated pancreatic endocrine neoplasm characterized by cells with irregular nuclei, coarse chromatin, nuclear molding, and scant, ill-defined cytoplasm (A) and by large cells with marked atypia (B) (hematoxylin-eosin, original magnifications ×400 [A] and ×1000 [B]). Figure 11. Immunolabeling of pancreatic endocrine neoplasm. A, Immunolabeling for chromogranin in the basal region of the cells. B, Diffuse immunolabeling for synaptophysin (original magnifications ×400).
Pancreatic endocrine neoplasm with unusual cell types. A, Oncocytic cells. B, Lipid-rich cells (hematoxylin-eosin, original magnifications ×400). Figure 10. Poorly differentiated pancreatic endocrine neoplasm characterized by cells with irregular nuclei, coarse chromatin, nuclear molding, and scant, ill-defined cytoplasm (A) and by large cells with marked atypia (B) (hematoxylin-eosin, original magnifications ×400 [A] and ×1000 [B]). Figure 11. Immunolabeling of pancreatic endocrine neoplasm. A, Immunolabeling for chromogranin in the basal region of the cells. B, Diffuse immunolabeling for synaptophysin (original magnifications ×400).
The Morphologic Signs of Malignancy
Tumoral infiltration of the duodenum and/or the biliary duct wall and lymph node metastases identify the malignant forms, and they should be carefully searched because they are not always evident macroscopically. The involvement of the peripancreatic fat can be more difficult to evaluate when the tumor has expansive growth margins and a pseudocapsule. Perineural and vascular invasion are most easily recognized in the peritumoral nerves and vessels within or adjacent to the pseudocapsule, if present; these are prognostic parameters included in the World Health Organization (WHO) classification1 (Table 2). The slowly growing tumor may entrap normal pancreatic structures such as ducts and acini, giving the false impression that the neoplastic cells have taken multiple differentiation paths; these aspects should never be considered as evidence of aggressive behavior.
Poorly differentiated endocrine carcinomas present with ill-defined solid masses and extensive necrosis, and histologically resemble small cell carcinomas or large cell endocrine carcinomas of others organs2 (Figure 10, A and B). They typically have a high mitotic rate, with a proliferative activity of more than 20%, and frequently stain for p53 protein.1,2
Immunohistochemistry
Immunohistochemistry serves primarily to confirm the endocrine nature of the neoplasia and thus to differentiate PENs from other pancreatic, extrapancreatic, and metastatic neoplasms. It is also useful for determining the type of hormones produced by the neoplastic cells. In addition, immunohistochemical markers are used or, at least, have been proposed for assessment of prognosis.
General Endocrine Markers
Although examination of hematoxylin-eosin sections in general allows the diagnosis of endocrine tumor, the endocrine differentiation has to be proved by immunohistochemical labeling with antibodies to at least 1 general endocrine marker, synaptophysin35 or chromogranin A (CgA),36 which are located in small neurotransmitter-storing synaptic vesicles and in neurosecretory granules, respectively (Figure 11, A and B). The cytosolic neuron-specific enolase37 and protein gene product 9.538 are less specific, and their diagnostic utility is limited.
Chromogranin A stain is proportional to the content of neurosecretory granules and may be patchy and of variable intensity. Up to 20% of PENs show only focal CgA positivity, in contrast with the generally diffuse labeling for synaptophysin. Poorly differentiated endocrine tumors are usually negative for CgA, while synaptophysin persists in the neoplastic cells. Circulating CgA is also used as a tumor marker for PENs. It is elevated in 60% to 80% of cases and correlates with disease burden, and, as such, is very useful for diagnosis, follow-up, and monitoring of response to therapy.4,39,40
Additional Markers for Differential Diagnosis
Different cytokeratins show variable immunoreactivities in PENs. Cytokeratins 8 and 18 are constantly positive, whereas cytokeratins 7 and 20 are usually negative; cytokeratin AE1/AE3 labels 50% of PENs.41 β-Catenin displays a membranous pattern of staining, in contrast with the abnormal nuclear positivity observed in solid-pseudopapillary neoplasm.42 Neural markers CD56 and S100 protein can be expressed in PENs. The first is a cell adhesion molecule and is identified in most cases, whereas S100 immunoreactivity is present in only a few tumors.34
Pancreatic endocrine neoplasms may focally express trypsin, a marker of acinar differentiation; if more than 25% of the neoplastic cells are positive, the neoplasm should be classified as a mixed acinar-endocrine carcinoma.43–45 Labeling for glycoproteins DUPAN-2 or carbohydrate antigen 19-9, markers of ductal differentiation,16,45 can be evident in cases with pseudoglandular pattern, but this is not sufficient for a diagnosis of mixed ductal-endocrine carcinoma, unless a typical adenocarcinoma component is recognized.
Identification of Hormones Produced by the Neoplasia
Pancreatic endocrine neoplasms may express normally produced pancreatic hormones (insulin, glucagon, somatostatin, pancreatic polypeptide), hormones of ectopic origin (gastrin, vasoactive intestinal polypeptide, adrenocorticotrophic hormone), and bioamines (serotonin). The pattern of labeling for these hormones varies widely; some PENs show strong and diffuse positivity, while others only show focal and faint staining; the positivity for more than one hormone is common. In F-PENs, the hormone responsible for the clinical syndrome can be demonstrated, but staining intensity or the number of positive cells does not correlate with the severity of symptoms because of impairment in the storage and secretion capability of the cells. Nonfunctioning PENs produce and secrete a wide range of hormones whose serum levels are elevated. The potential reasons for the lack of a syndrome include inadequate secretion of hormone (eg, amount is too small to cause symptoms) or secretion of a hormone in an inactive, functionally inert form. However, the type of hormone produced may be useful as a serologic marker for the early recognition of relapse or metastases.
Prognostic Markers
Several recent studies have suggested a prognostic significance for a variety of immunohistochemical markers, such as COX2,46 p27,47 and CD9948 or some expressed by normal islet cells, like progesterone receptors.49 However, these results need to be verified in independent patient cohorts.50,51 The expression of cytokeratin 19, regarded as a marker of ductal epithelium, has been suggested as a marker of aggressiveness48,52,53 that gives prognostic information independently from that obtained by WHO criteria.54 As proliferative activity has a recognized prognostic value,1,55 its assessment by Ki-67 immunostaining is a routine practice in several institutions, including ours. The WHO classification considers a Ki-67 labeling index of 2% as a discriminant, among well-differentiated PENs, between those with benign and those with uncertain behavior.1
Differential Diagnosis
The differential diagnosis of PENs includes the other “solid cellular” pancreatic neoplasms, some extrapancreatic tumors, and, more rarely, cystic tumors and nonneoplastic lesions of the pancreas.
Solid Cellular Pancreatic Neoplasms
These are acinar cell carcinomas,56,57 pancreatoblastomas,58 and solid-pseudopapillary neoplasms,59,60 which frequently share the clinical scenario with PENs because of the presence of a slow-growing, pushing mass. However, each of these neoplasms can be differentiated from PENs because of its distinctive clinicopathologic features. Acinar cell carcinoma generally affects adults; solid-pseudopapillary neoplasm is more common in young females; pancreatoblastoma mostly arises in children. Pancreatic endocrine neoplasms often show a wide variety of different organoid patterns that generally are not present in the other solid cellular tumors, and their cells exhibit the classic salt-and-pepper chromatin. An acinar pattern, prominent nucleoli, and a high mitotic rate should suggest acinar cell carcinoma; degenerative pseudopapillae, cells with clear cytoplasm, cytoplasmic hyaline globules, and oval nuclei with longitudinal grooves are diagnostic for solid pseudopapillary tumor. Squamoid nests are a hallmark of pancreatoblastoma. Immunohistochemistry is needed to demonstrate the endocrine (chromogranin and synaptophysin) or acinar (trypsin) differentiation. Solid-pseudopapillary neoplasms generally do not express specific cell-lineage markers and only focally express keratin in a minority of cases. They express nonspecific markers such as vimentin, neuron-specific enolase, and CD56, but typically they show abnormal nuclear positivity for β-catenin, and the cells immunolabel with CD10 but never with chromogranin.
Other Pancreatic and Extrapancreatic Neoplasms
Differential Diagnosis of the Morphologic Variants of PEN
The morphologic variants of PENs are prone to be misdiagnosed for more aggressive neoplasms. Biopsy or cytologic material represents the greatest diagnostic challenge for the surgical pathologist in these cases, particularly in the case of hepatic metastatic disease.
Clear cell PENs should be differentiated from all the clear cell lesions occurring in the pancreas. Among metastatic neoplasms,65 renal cell carcinoma is by far the most relevant66–68; immunohistochemical analysis demonstrating coexpression of CD10+, vimentin, and cytokeratin, together with lack of endocrine markers, allows its identification. Less frequent primary pancreatic lesions to be considered include tumors rich in lipid, glycogen, or mucin foamy-gland adenocarcinoma,69 ductal adenocarcinoma with clear cell features,70,71 serous adenoma, especially its solid variant, clear cell variant of solid-pseudopapillary neoplasm,72 and clear cell sugar tumor.63 Those occurring in the peripancreatic region comprise adrenal tumors, steroid-secreting tumors of the ovary, clear cell hepatocellular carcinoma, and clear cell sarcoma or melanoma.73
Oncocytic74 and lipid-rich PENs25 may resemble hepatocellular and adrenal cortical tumors. Nuclear pleomorphism, glandular formation, glassy cytoplasmic inclusions (“rhabdoid features”), and occasional abundance of fibrotic stroma are features erroneously suggesting a diagnosis of pancreatic adenocarcinoma or metastatic carcinoma. Pleomorphic PEN34,75 is frequently diagnosed as ductal adenocarcinoma and rhabdoid PEN76–78 as anaplastic carcinoma79,80 or metastases with rhabdoid phenotype, particularly of melanoma. The absence of necrosis and a low mitotic rate should suggest a low-grade neoplasm.
Pancreatic endocrine neoplasms presenting as cystic lesions,21–23,81 either macrocystic or microcystic, are difficult to diagnose preoperatively. Most are nonfunctioning tumors and the identification of their endocrine nature is usually delayed until the histopathologic examination of the surgical specimen.
Ductuloinsular tumor of the pancreas82 is an endocrine tumor that may be misdiagnosed as mixed ductal-endocrine neoplasm. It is characterized by the presence of several small, cytologically bland ductules intimately admixed with the endocrine component. It has been suggested that this neoplasm be named pancreatic endocrine tumor with entrapped ductules83 to describe the nonneoplastic nature of the ductules. The clinical behavior of this neoplasm is similar to typical well-differentiated PEN.
Nonneoplastic Conditions
The aggregation of nonneoplastic islets in chronic pancreatitis may be mistaken for a PEN. In this case, endocrine elements are grouped in vaguely insular structures characterized by the persistence and normal distribution of the 4 cell types normally present in the pancreatic insulae.84 The presence of endocrine elements in perineural spaces, observed in rare cases of severe chronic pancreatitis, must not be considered a sign of neoplasia.
CLASSIFICATION
The WHO classification, proposed in 2000 and modified in 2004 (Table 2), identifies 3 main categories of PENs: well-differentiated endocrine tumor, well-differentiated endocrine carcinoma, and poorly differentiated endocrine carcinoma.1
Morphology allows the distinction between poorly differentiated carcinomas and well-differentiated neoplasms. The former are invariably high-grade malignancies. The latter include more than 90% of PENs, whose clinical behavior, varying from indolent to malignant, cannot be predicted by their tissue architecture or cytologic features. Well-differentiated endocrine carcinomas are morphologically similar to well-differentiated endocrine tumors and are identified only when there is either invasion of adjacent structures or metastasis. As a number of well-differentiated endocrine tumors may eventually metastasize, the WHO classification has introduced 2 subcategories—benign and of uncertain behavior—that are assigned to a tumor according to the presence of a series of pathologic parameters, widely recognized as prognostic factors (Table 2). The presence of vascular and perineural invasion, a tumor size larger than 2 cm, a mitotic rate higher than 2 mitoses per 10 high-power fields, and a Ki-67 proliferative index above 2% are considered a sign of potential aggressive behavior.16,55,85–89
In our experience, an accurate examination of the peripancreatic adipose tissue to search for lymph node metastasis is mandatory to avoid the erroneous assignment of a number of well-differentiated endocrine carcinomas to the well-differentiated endocrine tumor category. In fact, metastasis does not always cause an increase in nodal size (Figure 12). In this respect, a recent study demonstrated that tumor size was not associated with the probability of lymph node metastasis, and positive nodes were identified in 5 of 19 patients with tumors smaller than 2 cm.90 Recent studies by our group and others have validated the WHO classification and demonstrated that it is a useful tool for clinical purposes.18,54,87,90–95 The WHO criteria efficiently discriminate PENs into different categories that show a statistically significant different outcome and permit the planning of surgical or medical strategies.
Small lymph node (2 mm) with metastasis (whole mount section) (hematoxylin-eosin, original magnification ×40). Figure 13. Pancreatic endocrine neoplasm in von Hippel-Lindau disease. The pancreas is completely occupied by a serous cystic neoplasm.
Small lymph node (2 mm) with metastasis (whole mount section) (hematoxylin-eosin, original magnification ×40). Figure 13. Pancreatic endocrine neoplasm in von Hippel-Lindau disease. The pancreas is completely occupied by a serous cystic neoplasm.
STAGING AND GRADING FOR PROGNOSTIC STRATIFICATION
The WHO classification has defined the features that discriminate benign behavior or low-risk, well-differentiated endocrine tumors from low-grade, malignant, well-differentiated endocrine carcinomas. This has been an important step and several recent publications have proven its effectiveness.18,54,87,90–95 However, the WHO classification does not allow a prognostic stratification of patients affected by well-differentiated endocrine carcinomas, which represent the true clinical challenge among PENs.
As the traditional TNM staging system does not include endocrine neoplasms, the European Neuroendocrine Tumor Society (ENETS) has proposed a TNM-based staging system for gastroenteropancreatic endocrine tumors. The TNM classification proposal reported in Table 3 is the result of a consensus conference that included 62 experts in the field of digestive endocrine tumors from 20 countries.96 The TNM system is based on the evaluation of the following parameters: size, extrapancreatic invasion, and lymph node and liver metastasis. These are the very same parameters used in the WHO classification to assign a tumor to a benign or malignant category, whereas in the TNM system they are independently evaluated and scored. Each of these parameters has a diverse weight in the assignment of the stage grouping, which results in a measure for the extent of the disease with prognostic significance.
TNM Classification and Staging Proposed by \[cb1.8\]the European Neuroendocrine Tumor Societya\[cb0\]
![TNM Classification and Staging Proposed by \[cb1.8\]the European Neuroendocrine Tumor Societya\[cb0\]](https://allen.silverchair-cdn.com/allen/content_public/journal/aplm/133/3/10.5858_133.3.350/1/m_i1543-2165-133-3-350-t03.gif?Expires=1742358773&Signature=sjl5-Oa3XUDRxmRdHL5K84UTp3OQgUM74GyLU3yEXIKasL0XQrW7yryP1ulZfRC8rppOWe2gDXiUEsUj7dNkaKzrg8CiF6jiJQ0GPh8CY77IpR1iXAVQ3rUN3Vd7HD9ENiT8xbnsqTPeq32xFcI2Ekpa6TyoFQ0JwS02jvGcVPvjzaizZYp7uTh5oc5ADVQ4zz0zd0Rw9XAsQtU4NVliYN-y8xH4GfLfulkHgM5TzwuY6afPXKtjb3aDm-tCn3zWkn2Xm5nhTdkQD5D9PsIIbZtpPjPAouRt8DcBpVVAz1P2pwnZB1trGF0AFcoIXqc7HW8K7IekwUfPJaU95kynKw__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
However, the clinical need to differentiate between carcinomas belonging to the same stage is not solved by the TNM assignment. Therefore, in the same conference, the potential role of a grading system applied to PENs that would provide this information was widely discussed. Cytologic criteria were considered of no help in this matter, so it was decided to use the proliferation activity as a grading system based on mitotic count and/or Ki-67 index (Table 4).
Both the TNM staging and grading systems have been shown to be valid tools for prognostic stratification of PENs in clinical practice.93,97 The TNM staging of PENs for 274 patients observed at our institution in the last 15 years has allowed the assignment of a relative risk of death proportional to the TNM stage at the time of diagnosis. Patients with stages II, III, and IV disease showed a respective risk of death of 7, 29, and 58 times higher than did patients with stage I tumors (A.S., unpublished data, 2008). Moreover, grading is useful for therapeutic decisions, particularly in patients with advanced stage disease, as we have recently reported in a prospective single center study.98 Also, Aparicio et al99 reported that a slow tumor growth rate is the only predictive factor for response to therapy with somatostatin analogues.
A different proposal for a staging and grading system applied to PENs, which is reported to provide more accurate prognostic information than that of the WHO, comes from a group at the Memorial Sloan-Kettering Cancer Center (New York, New York).90 This combined system uses tumor size and metastasis to stage PENs into 3 categories (<2 cm, primary; ≥2 cm, primary; or metastases) and uses mitosis and necrosis to grade PENs into 2 categories (low grade, no necrosis with <2 mitoses per 50 HPFs; or intermediate grade with necrosis and/or ≥2 mitoses per 50 HPFs).
THERAPEUTIC CHOICES
The therapy of choice for PENs is surgical resection with curative intent. The issue of which is the best treatment for advanced, well-differentiated endocrine carcinomas not suitable for radical surgery remains controversial3,98 and consensus guidelines have been proposed.4,13–15 As most cases express receptors for somatostatin, somatostatin analogues have been used and shown to achieve frequent stabilization of the disease. The therapeutic options for the more aggressive diseases include either chemotherapy, usually based on doxorubicin and streptozotocin, or peptide receptor radionuclide therapy. The open question is how we can best identify which patients to treat with somatostatin analogues and which will benefit from more aggressive therapies as first-line treatment.
GENETIC FEATURES
Pancreatic endocrine neoplasms may arise in the context of 4 hereditary cancer syndromes, the most common of which is MEN-1, where PENs occur at an earlier age than they do in sporadic forms, may precede other manifestations of the syndrome, and determine the prognosis. Nonfunctioning PENs are also a rare, yet well recognized part of VHL (Figure 13), whereas they seldom occur in NF1 and TSC. However, most PENs occur as sporadic diseases.
Hereditary Neoplasms
Multiple Endocrine Neoplasia Type 1
Multiple endocrine neoplasia type 1 is an autosomal dominant disease with greater than 95% penetrance, characterized by multifocal endocrine tumors affecting multiple organs. MEN1 gene encodes a 68-kDa protein of 610 amino acids named menin. MEN1 germline mutations generally cause the inactivation of menin and, consistent with the role of MEN1 as a tumor suppressor gene, most MEN-1–associated tumors show somatic loss of the second wild-type allele (loss of heterozigosity). The pancreas is involved in roughly 60% of cases with multiple endocrine tumors of variable size, mostly NF-PEN, either synchronous or metachronous.100,101
The finding of “pancreatic microadenomatosis” in more than 80% of patients with MEN-1, that is, the presence of numerous microadenomas, suggests that the latter are precursor lesions of MEN-1–associated PENs.33 A microadenoma is a lesion up to 5 mm in diameter with the following typical features: (1) a trabecular growth pattern, (2) a distinct stromal component, and (3) immunopositivity for 1 hormone, with prevalence of glucagon. Microadenomatosis is often associated with 1 or more macrotumors (>5 mm) with frequent immunohistochemical expression of glucagon and pancreatic polypeptide or, rarely, somatostatin, and never causes hormonal syndromes.32,100 Moreover, the detection of loss of heterozygosity in the so-called monohormonal isletlike endocrine cell clusters found in MEN-1 pancreas, has identified these as forerunners of microadenomas.102 The single enlarged islets with an increased number of glucagon cells found in MEN-1 pancreas are also a potential precursor of microadenomas. However, retention of the normal MEN1 allele in these cells indicates that they are still nonneoplastic in nature. Although pancreatic microadenomatosis is a feature of MEN-1 syndrome, this condition is not specific, as multiple glucagon- or insulin-producing microadenomas have also been found in patients with no evidence of hereditary syndromes.33
von Hippel-Lindau Disease
von Hippel-Lindau syndrome is an autosomal dominant disease with age-dependent penetrance and a marked phenotypic variability,103 characterized by predisposition to a variety of malignant and benign neoplasms in various organs, including central nervous system hemangioblastomas, renal cell carcinomas, and pheochromocytomas. The VHL tumor suppressor gene is located on chromosome 3p25. The VHL protein is a component of a complex involved in the degradation of hypoxia-inducible factor, a transcription factor that plays a central role in the regulation of gene expression by oxygen.104 In up to 77% of cases, the pancreas is involved by serous microcystic adenomas and in up to 16% of cases, by PENs.101,103 Pancreatic endocrine neoplasms are multiple in about 40% of cases and most are nonfunctioning. Two histological variants of PEN occur in VHL: the conventional type and, more commonly, the clear cell type that appears to be specific for the disease.105 The latter type can be differentiated from metastatic renal cell carcinoma by using immunohistochemical staining for CD10+, vimentin, and cytokeratin together with endocrine markers. Combined PEN/serous cystadenomas also occur, sometimes involving the whole pancreas. von Hippel-Lindau– related PENs grow as slowly as their sporadic counterpart and no precursor lesions have yet been identified.101
Neurofibromatosis Type 1
Neurofibromatosis type 1 is an autosomal dominant disease with high penetrance106 caused by mutations of the NF1 gene, localized on chromosome 17q11.2, which encodes neurofibromin that acts as a negative regulator of the Ras-related G-proteins by increasing Ras GTPase activity. Our knowledge of NF1-associated intestinal neuroendocrine tumors (NETs) is based on case reports and small series.101 They occur in 1% of patients with NF1 and most arise in the region of the ampulla of Vater, show glandular structures containing periodic acid-Schiff–positive psammoma bodies, and consistently express somatostatin (SOM-NETs). Neurofibromatosis type 1–related SOM-NETs may be associated with gastrointestinal stromal tumors. Rare intrapancreatic SOM-NETs or insulinomas have been reported in patients with NF1.107,108 Precursor lesions of NF1-associated endocrine neoplasms have not been identified.
Tuberous Sclerosis Complex
Tuberous sclerosis complex is an autosomal dominant disorder with almost complete penetrance, characterized by hamartomatous lesions caused by inactivating mutations in either the TSC1 gene at 9q34 or the TSC2 gene at 16p13.3. These genes encode the proteins hamartin and tuberin, respectively, which form a complex that participates in the regulation of cell proliferation, growth, and differentiation.109 The TSC1/ TSC2 dimer mediates a key step in the phosphoinositide 3-kinase (PI3K) signaling pathway. Thus, the TSC1/TSC2 complex is involved in the regulation of mTOR activity, a master controller of protein translation that integrates information on growth stimuli, cellular energy levels, nutrient availability, hypoxia, and cell growth.109 The few cases of PENs described in patients with TSC were insulinomas or NF-PENs, some of which were malignant.101
Sporadic Neoplasms
The genetic aberrations associated with PENs include anomalies in the “anatomy” and function of DNA. The anatomic lesions consist of chromosomal alterations (numerical and structural) and epigenetic changes that regulate gene transcription. The latter changes include DNA methylation and acetylation or other modifications of DNA-associated histones. The third anatomic lesion of DNA is mutation in single genes. The functional genomic alterations are those found at the gene expression level and involve 2 main types of RNAs: protein-coding mRNAs and regulatory small RNAs known as “noncoding RNAs” (microRNAs and small noncoding RNAs or sncRNAs).110
Chromosomal Anomalies
Chromosomal analysis has identified numerous regions of chromosome loss and gain in sporadic PENs,111 suggesting the existence of 2 subgroups: those showing frequent allelic imbalances and those showing a low number of allelic imbalances.112,113 It has been suggested that these 2 molecular subgroups, which correspond to aneuploid or near-diploid tumors, respectively, have prognostic value.113 More recent studies based on comparative genomic hybridization array technology confirmed the previous observation that the total number of genomic changes per tumor appears to be associated with both tumor burden and stage of the disease, suggesting that genetic alterations accumulate during tumor progression.112 A recent study concluded that the DNA copy number status is the most sensitive and efficient marker of clinical outcome in insulinomas and is of potential interest in noninsulinoma PENs.114
These findings point to chromosomal instability as an important mechanism associated with tumor progression. In particular, comparative genomic hybridization and microsatellite analysis (loss of hetrozygosity) have shown that chromosomal losses occur more frequently than gains, whereas amplifications are seemingly uncommon.112,113 The most frequent gains are on chromosomes 5q (25%), 7pq (41%), 9q (28%), 14q (32%), 17pq (31%), and 20q (27%). The most frequent losses include 1p (21%), 3p (19%), 6q (28%), 11q (30%), Y (31%), and X (31%). An important observation from these studies is that the alterations are not randomly distributed on chromosomes but are particularly common in distinct chromosomal regions. Furthermore, genomic alterations including losses of chromosomes 1 and 11q, as well as gains of 9q, are already present in many small tumors (<2 cm), thus suggesting that these may be early events in the development of PENs.115–118 A strong correlation has been found between sex chromosome loss and aggressive behavior of PEN, namely, the presence of local invasion or metastasis.119
Microsatellite instability, that is, the occurrence of widespread mutations in microsatellite sequences, due to inactivation or silencing of mutator genes, is a very rare phenomenon; it was not found in 76 published cases of PENs,120 and was found in only 1 of 113 PENs in our institution (A.S., unpublished data, 2008).
Epigenetic Anomalies
Epigenetic abnormalities occur in PENs, but as yet, genes targeted by these changes and with a definite role in PEN tumorigenesis have not been identified.
Two studies on the methylation status of several candidate tumor suppressor genes have been published to date.120,121 The first analyzed 11 genes in 46 PENs and revealed hypermethylation in the following 8 genes: RASSF1A (75%), P16/INK4A (40%), MGMT (40%), MLH1 (23%), APC (21%), E-cadherin (23%), P73 (17%), and RARB (25%); no methylation was found in TIMP3, P14, and GST.121 The second study also included 46 PENs that were analyzed for 11 genes, 7 of which were common to both reports and showed the following methylation rates: RASSF1A (80%), P16/INK4A (0%), MGMT (17%), MLH1 (0%), APC (48%), E-cadherin (2%), TIMP3 (0%); the remaining 4 genes were MEN1 (19%), HIC-1 (93%), RUNX3 (7%), and PTEN (0%).120,121
Two findings were consistent in the analysis of the 7 genes common to both studies: the first is the lack of methylation in the TIMP3 gene, a finding that contrasts with a report showing that methylation is associated with loss of expression of TIMP3 in 8 of 18 PENs (44%)122; the second is the high rate of methylation in the RASSF1A gene, a finding confirmed in other PEN series in which the reported rate ranged from 60% to 100% of cases.121,123–125
Because RASSF1A inhibits tumor growth in both in vitro and in vivo systems,123 it has been suggested that its inactivation may be a crucial event in PEN pathogenesis.120,123–125 However, this assumption has never been confirmed by data demonstrating the loss of RASSF1A expression in the presence of gene methylation. Although RASSF1A has been considered a RAS inhibitor, and, as such, its inactivation is considered to result in the indirect activation of the RAS pathway in PENs, recent studies assign a more complex role to this gene in cell life.126
Beyond methylation studies regarding single genes, 10 PENs have also been evaluated for the global methylation status of DNA by analysis of long interspersed nucleotide elements (LINE) 1 and ALU sequences.127 In particular, PENs were found to have a global hypomethylation with respect to normal pancreas, but less commonly than is seen in intestinal endocrine tumors versus normal intestine.
A recent report by our group has suggested that chromatin remodeling by histone acetylation might play a role in pancreatic endocrine neoplasms, as the histone-deacetylase inhibitor trichostatin A strongly inhibits cell growth of different pancreatic endocrine carcinoma cell lines.128
Single-Gene Mutations
To date, mutation of MEN1 and allelic loss of chromosome 11q, which encompasses the region containing the MEN1 locus, are the most common genetic alterations found in PENs. These occur in NF-PENs and gastrinomas with a frequency higher than that in insulinomas.102,129,130 Indeed, mutations are detected in about 30% of NF-PENs, whereas insulinomas, gastrinomas, glucagonomas, and VIPomas show a mutation rate of 7%, 36%, 67%, and 44%, respectively.129
In contrast to the low rate of MEN1 mutations, more than 50% of all PENs exhibit losses at 11q13 and/or more distal parts of the long arm of chromosome 11. This implies that a haploinsufficiency of the MEN1 gene may be sufficient as an initiating factor, which is possibly backed by the loss of additional oncosuppressor genes lying distally on this chromosomal arm.131–133
Promoter methylation has been suggested as an additional mechanism of inactivation of MEN1, but the existing data are controversial.120,121
Menin associates with and modulates the histone methyltransferase activity of a nuclear protein complex that regulates gene expression. Menin-dependent histone H3 lysine-4 methylation maintains the in vivo expression of cyclin-dependent kinase inhibitors to prevent pancreatic islet tumors. Therefore, menin activates an epigenetic mechanism of tumor suppression by promoting histone modifications and maintaining transcription at multiple loci encoding cell cycle regulators, essential for endocrine growth control.134 Interestingly, tumors of Men1 β-cell mutant mice exhibit deregulation of genes involved in cell proliferation and cell cycle control.135
Mutations of the VHL tumor suppressor gene have been detected only in rare sporadic PENs, despite a high loss-of-heterozygosity rate of this chromosomal region, suggesting the involvement of potential tumor suppressor genes telomeric to the VHL locus.129,136
Other tumor suppressor genes or oncogenes including P16, PTEN, KRAS, DPC4, and P53 are only occasionally mutated, mostly in malignant tumors.2,129,137–140
Aberrant expression of both β-catenin and E-cadherin correlated strongly with lymph node spread and liver metastases in gastrointestinal endocrine tumors, but only 4 of the 165 PENs displayed nuclear staining and had a higher Ki-67 proliferation index.141,142 No mutation was detected in the β-catenin gene.142
Gene Expression Alterations: Protein-Coding RNAs
Gene expression profiling with microarray-based technology has produced a long list of differentially expressed genes in PENs that may, in the future, lead to the identification of prognostic markers and therapeutic targets.143–146 One study has suggested Serpine10 and BIN1 as potential markers and BST2 and LCK as potential therapeutic molecular targets.146 Another study has compared benign and metastatic PENs and has identified differentially expressed genes involved in pathways related to angiogenesis and remodeling, signal transduction through tyrosine kinases, calcium-dependent cell signaling, and response to drugs.145 A third study with 19 PENs has suggested the existence of “benign” and “malignant” clusters, corresponding to the WHO categories of well-differentiated endocrine tumor and well-differentiated endocrine carcinoma, respectively. Moreover, because the platelet-derived growth factor receptor β (PDGFR-β) is in the active, phosphorylated state in 83% of all PENs, it has been suggested as a candidate therapeutic target.147
Several studies have addressed the analysis of expression levels of particular genes in PENs. Some factors like VEGF-C, MAGE-1, P27/KIP1, thrombomodulin, and SRC kinases have been described as being involved in metastatic spread of PEN, thus suggesting new therapeutic approaches.47,148–150 Other molecules recently proposed as possible targets for novel therapies of PENs are CDK4, PDGFR-β, CLDN3, and CXCL-12.151–154 Finally, the expression of ARHI seems to be a prognostic factor for disease outcome in pancreatic endocrine neoplasms,155 whereas expression of clusterin, ghrelin receptor, utrophin, and cyclin D1 do not relate to tumor aggressiveness.156–158
Gene Expression Alterations: Regulatory MicroRNAs
MicroRNAs are small noncoding RNAs that regulate gene expression by targeting specific mRNAs for degradation or translation inhibition. Recent evidence indicates that microRNAs contribute to tumor development and progression and may have diagnostic and prognostic value in several human malignancies.110 An investigation of global microRNA expression patterns in normal pancreas, PENs, and acinar cell carcinomas showed that a particular pattern of microRNA expression distinguishes PEN from normal pancreas and acinar carcinoma, suggesting that this set of microRNAs might be involved in PEN tumorigenesis. This study also showed that miR-204 is primarily expressed in insulinomas and correlates with immunohistochemical expression of insulin and that the overexpression of miR-21 is strongly associated with both a high Ki-67 proliferation index and the presence of liver metastases. These results suggest that alteration in microRNA expression is related to endocrine neoplastic transformation and progression of malignancy and might prove useful in distinguishing tumors with different clinical behavior.159
CONCLUSIONS
An accurate pathology report must include the necessary information to allow a correct classification, staging, and grading of the disease for prognostic evaluation, which will drive therapeutic choices.
At variance with many other tumor types, the genes and pathways involved in hereditary neoplastic syndromes, whose spectrum comprise PENs, do not seem to be involved in the genesis of the sporadic PENs. The exception is MEN1 gene inactivation. The degree of genomic instability, as measured by the number of chromosomal anomalies or changes in DNA copy number, correlates with the aggressiveness of the neoplasm.
The differentially expressed genes on the lists produced by expression profiling studies are pieces of the fascinating puzzle of PEN pathogenesis, a puzzle that awaits the discovery of keystone findings to be constructed.
Acknowledgments
Supported by the European Community Grant FP6 MolDiagPaca and by the Fondazione Cariverona, Verona, Italy.
References
The authors have no relevant financial interest in the products or companies described in this article.
Author notes
Reprints: Paola Capelli, MD, Department of Pathology, Section of Anatomical Pathology, Policlinico G. B. Rossi, 37134 Verona, Italy ([email protected])