Gastrointestinal Carcinoid Tumors

Biology of the Neuroendocrine System

The neuroendocrine system consists of a glandular (solid) system and a diffuse system (DNES). The glandular system is composed of the pituitary, parathyroids, paraganglia, and adrenal medulla. The diffuse system comprises cells dispersed throughout the skin, thyroid, lung, thymus, pancreas, gastrointestinal (GI) tract, biliary tree, and the urogenital system. The multiplicity of neuroendocrine cells in the GI tract makes it the largest single endocrine organ.

A characteristic feature of neuroendocrine cells is the production of a diversity of biogenic amines, peptides, prostaglandins, and tachykinins. These secretory products are stored in both large dense-core vesicles and small synaptic-like vesicles. Among other functions, it appears that the protein, chromogranin A (CgA), is essential in the genesis of vesicles and regulates the biogenesis of the dense-core granules. Such granules are then secreted under control of a complex variety of stimuli. The type of secretory granule produced depends in part on the cell type. Currently there are 16 different types of endocrine cells, producing more than 100 secretory products, that have been described. Although some of these cell types are present throughout the GI tract, others have a more restricted topography ( Table 80.1 ). Some cell types are located in one organ and produce specific products that result in well-described clinical syndromes (e.g., beta cells in the pancreas and insulinomas), whereas others produce a spectrum of disease states and are in a variety of locations (e.g., enterochromaffin [EC] cells and somatostatin in the stomach, small bowel, and colon). Curiously, the common nonfunctional NET, frequently found in the pancreas, has no clear cell of origin and no obvious secretory product.

There has been significant debate and misunderstanding about the developmental origin of gut neuroendocrine cells. Gut endocrine cells have many similarities to neural cells—they produce substances with transmitter functions, have secretory granules, and have similar cellular antigens (synaptophysin and neuron-specific enolase). For this reason, it was initially postulated that they were of neuroectodermal origin and were thus initially termed neuro endocrine cells. However, it has since been convincingly demonstrated that these cells come not from the ectoderm but rather have endodermal origins. Furthermore, it has been demonstrated more recently that all four intestinal epithelial cell types (enteroendocrine cells, goblet cells, Paneth cells, and enterocytes) differentiate from common pluripotent stem cells in the intestinal crypts. These enteroendocrine cells renew themselves from a large reservoir of stem cells and continue the process of differentiation throughout life. The regulatory controls for this process of enteroendocrine cell differentiation are under active investigation and are incompletely understood. Interestingly, it appears that the regulation of differentiation of enteroendocrine cells is similar to that for cells within the nervous system. Both cell types are controlled by similar genes encoding basic helix-loop-helix (bHLH) transcription factors under the control of the Notch signaling pathway, which may have potential therapeutic implications for NETs.

Nomenclature

The term carcinoid was coined before there was a full understanding of the tumor type. However, this term has remained in use clinically and is in part responsible for the confusion surrounding the terminology of NETs. First, the term neuroendocrine is not entirely accurate. As described previously, this term was used because it was assumed these cells derived from the neuroectoderm. Because it has since been shown that this is not the case, many have suggested the term be abandoned in favor of enteroendocrine . However, because there are many shared structural and regulatory properties between neural cells and endocrine cells, the term neuroendocrine tumor, or NET, seems appropriate, and most international groups have accepted its use.

An older term used for these enteroendocrine tumors was APUDoma (amine precursor uptake and decarboxylation), which describes a common biologic function of the cell type. These cells also have been referred to as EC or argentaffin cells because of their staining affinity for chromium and silver salts, respectively. Furthermore, those cells that required a reducing agent to stain with silver salts were termed argyrophilic cells. Those tumors that produce an active product are often referred to by the specific term for each product (glucagonoma, insulinoma, etc.) or by the term for the group that produces these products most often—islet cell tumors.

The number of terms in use for this cell type and their respective tumor types has served to generate significant confusion in the field. In fact, a group of experts participating in a summit of the National Cancer Institute identified this as one of the major hurdles to progress and commented “semantic issues continue to obfuscate the field.” For example, the term carcinoid unfortunately continues to give the inaccurate impression of the relative benignity of these lesions although many do not behave as such. The more accurate term for this group of tumors currently advocated by the World Health Organization (WHO), the European Neuroendocrine Tumor Society (ENETS), and the North American Neuroendocrine Tumor Society (NANETS) is gastroenteropancreatic neuroendocrine tumors (GEP-NETs). However, the term carcinoid continues to be in use and is best understood as a synonym for well-differentiated neuroendocrine neoplasms of the luminal GI tract.

Incidence

The clinical incidence of NETs ranges from approximately 1.3 per 100,000 to 5.25 per 100,000 in a study using the 2004 data from Surveillance, Epidemiology and End Results (SEER). However, detailed autopsy studies have shown that the majority of carcinoid tumors are discovered after death from another cause. A Swedish study, from 1958 to 1969, demonstrated that only 10% of all cases were identified on surgical specimens, with 90% discovered incidentally at autopsy. In fact, 1.2% of people had a carcinoid tumor at autopsy. A second smaller but more recent autopsy study in Japan demonstrated that a similar fraction (1.6%) of patients had endocrine tumors in their pancreas on random sectioning, but when the pancreas was more carefully examined with serial sections, a full 10% of patients had endocrine tumors.

Multiple studies have documented a significantly increasing incidence of NETs over the past several decades, with an estimated 3% to 10% increase per year over the past 35 years. The lack of consensus in nomenclature results in differences in classification and incidence reports. Although a fraction of this rise may be related to a true increase in disease frequency, much of this is probably more because of the improvements in clinical awareness of the disease, the improvement in diagnostic and imaging technology, and the changing classification systems. For example, because databases such as SEER included malignant disease only, lower-grade carcinoid tumors were not reported until 1986. Furthermore, the distribution of the primary sites of tumor has changed with time as well. For example, rectal carcinoids represented only 9% of tumors in the early SEER cohort (1973 to 1991) compared with 18% in the later cohort (1992 to 1999). Similarly, 7% of the early cohort had appendiceal carcinoids compared with 2% in the later cohort. These changes could reflect the increase in the use of sigmoidoscopy and the decrease in incidental appendectomies, respectively, rather than a true change in the sites of disease.

Although carcinoid tumors are often considered rare, they are actually the second most prevalent GI cancer, after colorectal cancer. The estimated 29-year limited-duration prevalence of NETs in 2004 was approximately 103,000, making them significantly more common than esophageal, gastric, pancreatic, and hepatobiliary cancers.

Risk Factors

The risk factors for the development of sporadic NETs are still unknown. Not surprisingly, the risk factor shown to be most significantly associated with development of a NET is the parental history of a carcinoid tumor in an extrapulmonary site. A similar pattern was seen with history of carcinoid tumor in a sibling. In addition, there was an increase in the incidence of carcinoid tumors in the offspring of parents with cancers of the brain, breast, liver, endocrine system, and urinary tract. This finding of an excess number of cases of carcinoid tumors in offspring of parents with any history of cancer was also demonstrated in a US case-control series. In addition, this study demonstrated that a long-term history of diabetes mellitus was a risk factor for the development of gastric carcinoids, especially in women. For environmental exposures, the bulk of the evidence suggests that there is not a strong association with cigarette smoking. A large European population-based case-control study suggested an increased risk of small bowel carcinoid with occupational exposure to organic solvents and rust-preventive paint containing lead, although more evidence is needed.

Although little is known about the risk of sporadic NETs, there are four well-established genetic disorders that are strongly associated with the development of pancreatic neuroendocrine tumors (PNETs). The most recognized is multiple endocrine neoplasia type 1 (MEN1), which is associated with PNETs in approximately 65% of cases; 7% have gastric carcinoids. The PNETs in MEN1 patients are most commonly nonfunctional, but gastrinomas and insulinomas are frequently reported. The treatment strategies in these patients are complicated given the multifocality of pancreatic tumors and the need to balance a morbid operation against the potential for benefit in the context of multiple other competing medical issues.

The second genetic disorder known to be associated with PNETs is the von Hippel-Lindau (VHL) syndrome, caused by a mutation in the VHL tumor suppressor gene. Pancreatic cysts and tumors develop in the majority of patients. PNETs develop in approximately 15% of patients with VHL, are usually nonfunctional, and asymptomatic. In general, the tumors in VHL patients show indolent growth and rarely metastasize if less than 3 cm in size. For these reasons, many authors recommend routine resection only when tumors are larger than 3 cm. The third genetic disorder associated with PNETs is von Recklinghausen syndrome (neurofibromatosis 1). These patients are affected in only 10% of cases, with nearly all patients having a duodenal somatostatinoma. These generally are large periampullary tumors and generally require resection to treat local complications. The final genetic association with PNETs is tuberous sclerosis. Only a small fraction of these patients are affected, and both functional and nonfunctional tumors occur.

Staging and Classification Schemes

A large part of the difficulty in arriving at a standard classification scheme for GEP-NETs has been their widely variable biologic behavior coupled with their locations throughout the GI tract. To further complicate the situation, until recently the WHO classification for NETs used a system that incorporated both grading and staging information in a single system. Although it provided some prognostic information, this schema did not allow for more advanced tumor stages with metastatic disease to benefit from the significant amount of clinical information that derives from grade alone. The 2000 version of the WHO classification used the terms well-differentiated NET with benign behavior , well-differentiated NET with uncertain behavior , well-differentiated NE carcinoma with low-grade malignancy , and poorly differentiated NE carcinoma . In contrast, the 2004 version described well-differentiated endocrine tumor , well-differentiated neuroendocrine carcinoma , and poorly differentiated carcinoma . These terms are still encountered in pathology reports because the protocols for NET specimen examination published from the College of American Pathologists in 2010 still advocate the use of these earlier staging systems.

Diagnosis of Neuroendocrine Tumors

Radiographic Evaluation

If clinically suspected, a gastric, duodenal, or rectal carcinoid is best evaluated endoscopically, both with standard techniques and endoscopic ultrasound. However, standard axial imaging with computed tomography (CT) and magnetic resonance imaging (MRI) remains the mainstay of the evaluation of intraabdominal complaints. Although routine abdominopelvic CT imaging may detect large NETs, it is much less sensitive for the detection of small bowel luminal lesions. However, this can be improved by performing CT enterography, magnetic resonance (MR) enterography, or enteroclysis. The value of CT and MRI is that they can simultaneously detect hepatic metastases, examine regional lymph node basins, and specify the mesenteric vasculature that is often involved ( Fig. 80.1 ). Their ability to detect PNETs is also excellent, with hypervascular masses often being seen best on arterial phase imaging. For these reasons, contrast-enhanced CT scan is the imaging modality of choice for the majority of intraabdominal NETs.

Somatostatin receptor scintigraphy (Octreoscan) involves the use of ¹¹¹ In–diethylenetriaminepentaacetic acid (DTPA)-labeled octreotide to locate small tumors that overexpress somatostatin receptors (particularly subtypes 2 and 5). This is most useful for PNETs, with the exception of insulinomas. This has more recently been coupled with CT imaging to provide improved localization of areas of uptake. It is thought to be the most sensitive imaging modality for detecting metastatic disease. Other imaging modalities to consider include the use of intraoperative or endoscopic ultrasound for difficult-to-locate pancreatic lesions.

Neuroendocrine Tumors by the Organ of Origin