Pancreatic neuroendocrine tumors (PNETs) are a rare group of neoplasms, especially in children and adolescents, with an estimated incidence of 0.18 cases per million in individuals younger than 19 years of age, approximately one-tenth of that observed in adults. As a subset of pancreatic tumors, however, PNETs represent an important clinical entity in the pediatric population, comprising approximately 13% to 25% of all pediatric pancreatic neoplasms. These tumors are heterogeneous in their presentation, malignant potential, and treatment, which vary depending on their clinical functionality, histopathologic characteristics, and whether they occur as part of a hereditary syndrome. This chapter will discuss the evaluation and management of PNETs, with particular emphasis in relation to the pediatric population.
Embryologic Origin of PNETs
Whether clinically functional or nonfunctional, PNETs arise from the various subtypes of pancreatic hormone–expressing cells, which are ultimately derived from the endodermal pancreatic epithelium. Subsequent to fusion of the dorsal and ventral pancreatic buds between the sixth and seventh weeks of gestation (G6-7w), compartmentalization of the tip and trunk domains occurs, which in mice, gives rise to tip-derived exocrine acinar cells, as well as trunk-derived ductal and endocrine cells. The first endocrine progenitor cells, which express the neurogenin-3 transcription factor, appear at G7.5w, and subsequently delaminate from the ductal epithelium to give rise to various types of hormone-expressing cells that eventually coalesce to form islets. Islet clusters in rats are organized with a central core of insulin-secreting β cells, comprising 60% to 80% of the total islet cell population, which are peripherally surrounded by other cell types. These include glucagon-secreting α-cells, somatostatin-secreting delta-cells, pancreatic polypeptide-secreting (PP) cells, and ghrelin-secreting ε-cells, comprising 15% to 20%, 5% to 10%, 2%, and less than 1% of total pancreatic islet cells, respectively. Other minor cell populations include gastrin-secreting G-cells, and vasoactive intestinal polypeptide (VIP) secreting D1 cells. The ratio of particular endocrine cell types appears to vary both within individual islets, as well as across certain anatomic regions of the pancreas. Although information on the spatial distribution of pancreatic endocrine cells in humans is limited, several immunohistochemical and immunofluorescence studies have been conducted in rats and mice, which roughly correlate with the anatomic distribution of sporadic PNETs in humans. There appears to be, for example, a PP-rich, glucagon-poor islet characteristic of the ventral lobe, and a PP-poor, glucagon-rich islet characteristic of the dorsal lobe. Beta-cells, however, seem to be evenly distributed across anatomic regions within the gland. Hence, sporadic insulinomas tend to occur throughout the pancreas, whereas glucagonomas usually occur in tail, and PPomas are often localized to the uncinate region.
Classification of PNETs—Functional and Nonfunctional, Sporadic and Hereditary
PNETs are categorized broadly as either functional or nonfunctional based on their clinical presentation, and as sporadic or hereditary depending on whether they occur as part of an inherited syndrome. Functional PNETs (FPNETs) are associated with various clinically apparent syndromes resulting from hormonal hypersecretion, such as observed in patients with insulinomas, gastrinomas, glucagonomas, VIPomas, and certain other rare functional tumors. Those PNETs that are not associated with a clinical syndrome are termed “non-functional” PNETs (NFPNETs), although they may still demonstrate evidence of hormonal secretion, either biochemically or on immunohistochemical staining of tumor specimens. These hormonally “silent” tumors often secrete or stain positive for pancreatic polypeptide (60% to 85%), somatostatin, or ghrelin, and are discovered incidentally or present with symptoms such as abdominal pain or weight loss. Independent of functionality, PNETs can often be followed biochemically with various other hormonal biomarkers, including serum chromogranin A (CgA), synaptophysin, and neuron-specific enolase (NSE). As a tumor marker for NFPNETs, CgA is often the most clinically useful in order to gauge efficacy of treatment and monitor for evidence of recurrence.
Grading and Staging
Prognostic stratification by grade and stage has been recently proposed in classification systems developed by the World Health Organization (WHO), the European Neuroendocrine Tumor Society (ENETS), and the American Joint Commission on Cancer (AJCC). A common grading system has been adopted universally by all three classification systems, whether directly (WHO, ENETS) or indirectly (AJCC), which is based on mitotic count and Ki67 index ( Table 84-1 ). Although additional subtleties exist with respect to the presence of other pathologic features such as cellular morphology, well-differentiated G1 and G2 PNETs often have a much better overall prognosis than poorly differentiated G3 neuroendocrine carcinomas (NECs). Staging determined using the ENETS and AJCC systems is based on the familiar TNM scheme, with the latter using the same staging system as that used for exocrine pancreatic cancers. Although the AJCC system may be more clinically familiar, the ENETS system appears to be superior in its ability to stratify patients based on survival ( Table 84-2 ).
| Grade | Mitosis/10 HPF | Ki67 |
|---|---|---|
| G1 | <2 | <3% |
| G2 | 2-20 | 3% to 20% |
| G3 | >20 | >20% |
| ENETS System | AJCC System | ||
|---|---|---|---|
| Primary Tumor | Primary Tumor | ||
| TX | Primary tumor cannot be assessed | T1 | Tumor limited to the pancreas and size ≤2 cm |
| T0 | No evidence of primary tumor | T2 | Tumor limited to the pancreas and size >2 cm |
| T1 | Tumor limited to the pancreas and size <2 cm | T3 | Peripancreatic spread without major vascular involvement (celiac axis or superior mesenteric artery) |
| T2 | Tumor limited to the pancreas and size 2 to 4 cm | T4 | Major vessel involvement |
| T3 | Tumor limited to the pancreas and size >4 cm OR invasion of the duodenum and/or bile duct | ||
| T4 | Tumor invasion of adjacent organs or major vessels | ||
| Regional Nodes | Regional Nodes | ||
| NX | Nodal status cannot be assessed | NX | Nodal status cannot be assessed |
| N0 | No regional lymph node metastasis | N0 | No regional lymph node metastasis |
| N1 | Regional lymph node metastasis | N1 | Regional lymph node metastasis |
| Distant Metastases | Distant Metastases | ||
| M0 | No distant metastasis | M0 | No distant metastasis |
| M1 | Distant Metastasis | M1 | Distant Metastasis |
| Staging | Staging | ||
| I | T1 N0 M0 | IA | T1 N0 M0 |
| IIA | T2 N0 MO | IB | T2 N0 M0 |
| IIB | T3 N0 M0 | IIA | T3 N0 M0 |
| IIA | T4 N0 M0 | IIB | T1-3N1M0 |
| IIB | Any T N1 M0 | III | T4 any N M0 |
| IV | Any T Any N M1 | IV | M1 |
Inherited Syndromes
The incidence of PNETs is higher among children with certain hereditary endocrinopathies, and therefore appropriate biochemical and potentially radiologic screening should be considered in this patient population.
Multiple Endocrine Neoplasia Type 1 (MEN1)
Multiple endocrine neoplasia type 1 (MEN1) or Wermer’s syndrome is a hereditary endocrinopathy with a prevalence of approximately 1 to 2 per 40,000 resulting from dominant inactivating mutations in the MEN1 tumor suppressor gene located at chromosome 11q13. This gene encodes the nuclear protein Menin, which has been implicated in the transcriptional regulation of a variety of genes involved in both endocrine cell proliferation and survival. Clinically, patients classically develop tumors of the parathyroid, endocrine pancreas, and anterior pituitary, but will also have an increased incidence of adrenocortical hyperplasia, thymic, bronchial, and gastric carcinoid tumors, as well as tumors of the central nervous system (CNS), smooth muscle, and skin.
Diagnosis is best made via complete MEN1 gene sequencing, or in the absence of a known mutation, clinically, by the presence of two of the three main endocrine tumors (parathyroid, pancreas, pituitary), or by one of these tumors in the setting of a family history of MEN1.
Malignant PNET is the most common cause of mortality in patients with MEN1, responsible for 38% to 44% of total deaths according to data from pooled literature series, undoubtedly contributing to the reduced mean life expectancy of 55 years for patients with this syndrome.
Although 30% to 80% of patients with MEN1 will be diagnosed with a PNET, often in the second or third decade of life, autopsy series demonstrate the presence of multiple foci of either microscopic or macroscopic PNETs, usually diffusely distributed throughout the pancreas, in essentially all of these individuals.
Although pathologic evidence of NFPNETs is ubiquitous in patients with MEN1, most are small or microscopic, with clinical symptoms related to local compression occurring in only 0 to 15% of affected individuals. However, more recent studies demonstrate that NFPNETs are being diagnosed more frequently than functional tumors in these individuals, likely related to earlier and more regular biochemical and radiologic screening, which has been suggested to begin before 10 years of age. When and how to treat these nonfunctional tumors, however, in terms of tumor size and extent of pancreatic resection, is still controversial, as will be discussed in the section on NFPNETs.
Despite the increasing awareness of NFPNETS in MEN1, FPNETs of varying sizes are nevertheless still frequently diagnosed in these patients. It is generally agreed that all of these tumors should be treated, regardless of size, given the negative effect of hormonal hypersecretion on both morbidity and mortality. Indeed, although not as prevalent as in the past, 8% to 14% of patients with MEN1 still die of a PNET-related hormonal excess. The two most common FPNETs in MEN1 are gastrinomas and insulinomas, occurring in approximately 50% to 60% and 20% of patients, respectively. In addition to tumor multiplicity, FPNETs in patients with MEN1 have a younger age of onset and carry a similar or even higher risk of malignancy relative to their sporadic counterparts occurring in the general population. In the setting of a gastrinoma, in which 50% to 60% of patients present with malignant disease, this can have a profound effect on morbidity and mortality. In a recent pooled analysis of literature series, the majority of deaths secondary to PNETs in patients with MEN1 (approx. 60%) were a result of malignant gastrinomas, independent of gastrin or other hormone-related excess. Insulinomas, however, have a more favorable outcome, with only 3% to 18% demonstrating evidence of locoregional or distant disease. Other rare FPNETs, including glucagonoma and VIPoma, although diagnosed in less than 5% of patients with MEN1, are associated with a high risk of malignancy.
Current clinical practice guidelines for PNETs in patients with MEN1 recommend annual biochemical screening for insulinoma with fasting glucose and serum insulin beginning at age 5, and for gastrinoma with serum gastrin with or without gastric pH measurement beginning at age 20. Screening for NFPNETs with serum CgA and pancreatic polypeptide (PP), as well as for rare FPNETs with serum glucagon and VIP, should begin before 10 years of age, in addition to suggested imaging in the form of magnetic resonance imaging (MRI), computed tomography (CT), or endoscopic ultrasound (EUS).
Once a diagnosis of a PNET is made in the setting of MEN1, management is often nuanced and can vary significantly from that of sporadic PNETs.
Although certain FPNETs show a predilection toward anatomic regionalization in sporadic tumors, in patients with MEN1, both functional and nonfunctional tumors often arise diffusely across the entire pancreas. This inevitably influences subsequent surgical management, as preoperative localization of a functional tumor or tumors responsible for a clinical syndrome is often crucial, and does not necessarily correlate with visualized tumors on simple cross-sectional imaging or EUS. As a result of this disseminated gland involvement, the clinical objective for endocrine pancreatic disease in MEN1 is often functional and locoregional control rather than cure.
Moreover, given the multiple endocrinopathies present in MEN1, treatment sequencing is also essential, based on the timeline of presentation and potential interdependence of these neoplasms. A suspected diagnosis of a gastrinoma, for instance, can be clarified by initial treatment of a patient’s hyperparathyroidism, which can also secondarily ameliorate the symptoms associated with Zollinger-Ellison syndrome. A further discussion on diagnosis, localization, treatment, outcome, and follow-up for specific PNETs associated with MEN1 is presented in subsequent sections.
Von Hippel-Lindau (VHL)
von Hippel-Lindau (VHL) is an autosomally dominant inherited syndrome with a prevalence of one per 100,000 resulting from inactivating mutations in the VHL tumor suppressor gene on chromosome 3p. The wild-type vHL protein, which is a component of the E3 ubiquitin ligase, is involved in the inhibition of angiogenesis and cell growth through degradation of hypoxia-inducible factor 1a (HIF1a). Clinical manifestations of the disease include renal cell carcinomas (RCCs), CNS and retinal hemangioblastomas, endolymphatic sac tumors, pheochromocytomas, and pancreatic tumors.
Patients can be diagnosed via VHL gene sequencing, or clinically by the presence of two or more of the associated tumors, or one of these tumors in the setting of a family history of VHL.
Pancreatic neoplasms in VHL are present in 33% to 77% of patients and consist mainly of cystic lesions and neuroendocrine tumors, although the main cause of mortality is RCCs and CNS tumors, resulting in a median life expectancy of less than 50 years. Rarely (0.4% to 3.0%) are metastatic lesions from RCC observed in the pancreas. The cystic lesions, comprising mostly simple cysts and serous cystadenomas, do not usually require surgical extirpation unless there is evidence of an incidental mucinous cystic neoplasm, which carries a higher risk of malignancy than its serous counterparts, or compressive symptoms related to local effects of the tumor.
PNETs are diagnosed in 10% to 17% of VHL patients, and are essentially clinically nonfunctional, with few reports to date of PNET-associated hormonal syndromes.
The NFPNETs in VHL, which in contrast to MEN1 are often solitary in nature, have been reported to have a more indolent natural history than their sporadic counterparts, although approximately 10% to 15% of these tumors are malignant, with distant disease still detected in 11% to 20% of individuals. The less aggressive nature of these tumors, however, may also be a manifestation of the selection and lead-time bias associated with screening efforts in VHL patients, where PNETs are often detected earlier in the course of their progression. Indeed, annual biochemical and radiologic screening for PNETs in patients with VHL has been suggested to start between 15 and 16 years of age.
Various clinicopathologic features have also been associated with increased risk of malignant behavior. Patients with VHL exon 3 mutations, as well as those with tumors that are larger than 3 cm or have a doubling time of less than 500 days appear to have a more aggressive pancreatic tumor phenotype relative to VHL disease carriers without these features. Individuals with any one of these features have been suggested to undergo more frequent semi-annual or annual follow-up, whereas patients with two or more of these features should be considered for surgical resection.
Treatment sequencing also has importance for NF1, the same as for MEN1, based on the timeline of presentation and nature of the tumor. Pheochromocytomas, for instance, if detected concurrently with other tumors, should be resected first, to avoid a potential intraoperative hypertensive crisis.
Neurofibromatosis Type 1 (NF1)
Neurofibromatosis type 1 (NF1), also known as von Recklinghausen’s disease, is an autosomally dominant inherited disorder, with an incidence of one per 3000 live births, resulting from inactivating mutations in the NF1 gene on 17q11.2, which encodes the neurofibromin tumor suppressor protein. Wild-type neurofibromin, in its function as a Ras-GTPase activating protein, inhibits cell proliferation and survival by attenuating flux through Ras-signaling cascades.
In the absence of positive genetic testing, a diagnosis is made clinically by the presence of any two of the following clinical manifestations: neurofibromas, café-au-lait spots, intertriginous freckling, Lisch nodules, optic pathway gliomas, bony lesions, or a first-degree family relative with NF1.
PNETs occur rarely in patients with NF1, although they frequently manifest as malignant duodenal somatostatinomas (SSoma) when present. Metastasis occurs in up to 30% of these duodenal SSomas, which often present as an NFPNET, given the rarity of the somatostatinoma “inhibitory” syndrome, which classically presents as diabetes, cholelithiasis, weight loss, steatorrhea, and achlorhydria.
Although even less common, other NFPNETS, as well as gastrinomas and insulinomas have also been reported to occur in patients with NF1.
Tuberous Sclerosis (TSC)
Tuberous sclerosis is an autosomally dominant inherited disease, with an incidence of approximately 10 in 100,000 live births, resulting from mutations in either the TSC1 or TSC2 genes, which encode the hamartin and tuberin tumor suppressor proteins, respectively. These two proteins form a complex that acts to sequester Rheb GTPase and inhibit cell proliferation and survival via mammalian target of rapamycin (mTOR) signaling pathways. Clinical manifestations include diffuse hamartomas, with associated neurologic and dermatologic features.
A total of 26 cases of pancreatic tumors have been reported to be associated with TSC to date in the literature, 20 of which are PNETs (7 insulinomas, one gastrinoma, and 12 NFPNETs). Patients presented between the ages 6 and 51, most of whom were asymptomatic, with the discovery of their PNETs made on routine imaging. Although upregulation of mTOR signaling does appear to impart a more aggressive phenotype to these tumors, it is unclear currently whether PNETs should truly be considered a feature of TSC.
Nesidioblastosis
Hyperinsulinism in the pediatric population can be secondary to either an insulinoma or nesidioblastosis, a congenital dysmaturation or deregulation of the pancreatic islets. Both present with similar clinical and biochemical features, and must be distinguished from each other preoperatively, owing to variation in their surgical management.
Initially described by Laidlaw in 1938, nesidioblastosis most commonly occurs in neonates younger than 18 months of age, and presents with hypoglycemic hyperinsulinism and associated sequelae, including seizures. There are two main histologic subtypes, the more common diffuse subtype, comprising 60% to 70% of cases, and the focal subtype making up the remaining 30% to 40% of cases.
The diffuse or inherited subtype has been associated with compound heterozygote or recessive mutations in the genes encoding the sulfonylurea receptor 1 (ABCC8) and potassium inward rectifier (KCNJ711), two components of the pancreatic β-cell ATP sensitive K+ channel. The focal form, conversely, results from a paternally inherited mutation and secondary somatic loss/silencing of the maternal allele on 11p.
The epigenetic silencing or somatic mosaic paternal uniparental disomy (UPD ) at 11p15 that is frequently observed in Beckwith-Wiedemann syndrome has been hypothesized to be related to the preponderance of congenital hyperinsulinism in these patients, although no mutations in ABCC8 or KCNJ711 with concurrent paternal UPD been observed to date.
Optimal management of focal congenital hyperinsulinism is surgical enucleation, whereas the diffuse form is initially treated medically with somatostatin analogues (SSAs) and diazoxide. Unfortunately, more than 50% of patients with the diffuse subtype are refractory to maximal medical treatment, and require near total-pancreatectomy, which is complicated by short-term persistent postoperative hypoglycemia in 40% to 60% of infants, and on long-term follow-up almost certain endocrine insufficiency in the form of insulin-dependent diabetes mellitus.
Insulinoma
Epidemiology
Insulinomas are the most common functional sporadic PNET, with an incidence of 0.7 to 4 per million people per year. Although the median age of diagnosis for sporadic tumors is 50 years, insulinomas present earlier in the setting of inherited syndromes, such as MEN1, where the diagnosis usually occurs in the third decade of life. The majority of these tumors are sporadic and benign; however, traditionally, 10% of insulinomas are associated with MEN1, and 10% are malignant.
Diagnosis
Historically, the diagnosis of an insulinoma was made clinically by the Whipple’s triad, first described by Allen O. Whipple in 1938. This included the presence of neuroglycopenic (seizures, confusion, lethargy, visual disturbances, and motor deficits) and/or adrenergic symptoms (palpitations, diaphoresis, and anxiety), fasting hypoglycemia, and resolution of symptoms with glucose administration.
Diagnosis is currently made biochemically in the context of a supervised 72-hour fast, with the observation of an inappropriate elevation in plasma insulin or proinsulin, with a relative suppression of ketone bodies for a given set of blood glucose levels. Patients with proinsulin-dominant secreting tumors may remain euglycemic for a longer period during the fast given the decreased biologic activity of the prohormone. A proinsulin/insulin ratio greater than 25%, however, is considered diagnostic of proinsulinoma. Factitious hypoglycemia and confounding with certain oral hypoglycemics should also be excluded with the measurement of C-peptide levels and urine sulfonylureas, respectively.
Localization
Once a diagnosis is confirmed biochemically, the tumor is localized with various preoperative and intraoperative imaging studies. Although most sporadic benign insulinomas tend to be small (0.5 to 2 cm), localization is not usually problematic unless the tumor is smaller than 1 cm. For patients with MEN1 where insulinomas and other PNETs tend to be scattered diffusely throughout the pancreas, however, accurate localization of the appropriate functional tumors and choice of resection can become challenging.
Preoperative imaging includes noninvasive transabdominal ultrasound, high resolution CT or MRI, as well as 111 In-pentetreotide scan (OctreoScan, Figure 84-1 ). Transabdominal ultrasound has a low and variable sensitivity (25% to 70%) for detecting insulinomas, and tends to more useful for tumors larger than 1 to 3 cm. Dual-energy spectral CT has a sensitivity of ~95% for localizing sporadic insulinomas, which are often observed as a blush in the late arterial phase owing to the hypervascularity of this tumor. MRI, which has a sensitivity similar to CT (71% to 95%), and obvious benefit for minimizing radiation exposure in the pediatric population, will often demonstrate a low signal intensity on T1- and a moderate to bright signal on T2-weighted images. Octreotide scans should be used selectively, as they have a relatively low sensitivity (40% to 60%) for localization, consistent with the finding that only approximately 50% of insulinomas express the type II somatostatin receptor (SSTR2). More invasive imaging, in the form of EUS and calcium-stimulated arteriography is useful when noninvasive imaging fails to localize the tumor, or for patients with multiple PNETs, such as in MEN1, where anatomic regionalization of the clinically relevant functional PNETs is essential. EUS combined with fine needle aspiration is highly user dependent, with an overall sensitivity of 70% to 94% for tumor localization, although when performed by expert endoscopists can detect tumors as small as 2 to 3 mm. Small insulinomas in the pancreatic tail near the splenic hilum are more difficult to visualize using this modality.
