Between 1966 and 1968, Pearse⁸, ⁹ described a group of widely dispersed endocrine cells whose products appeared to be amines and peptide hormones that were linked together by their possession of a common set of cytochemical characteristics. These included nonspecific esterase and cholinesterase, alpha-glycerophosphate dehydrogenase in the cytoplasm, and the ability to produce certain biogenic amines such as adrenalin, dopamine, and 5-hydroxytryptamine (serotonin).¹⁰ Pearse coined the term APUD system for this group of dispersed endocrine cells, an acronym for their common ability to engage in Amine Precursor Uptake and Decarboxylation. The APUD system was expanded to include most endocrine cells producing polypeptide hormones, carotid body type I cells, melanoblasts, and the clear cells of the urogenital tract.¹¹, ¹², ¹³, ¹⁴ The common cytochemical features of the APUD cells were interpreted by Pearse to reflect a common embryologic origin, which he postulated was the neural crest. Subsequently, Pearse concluded that his APUD system corresponded to Feyrter’s earlier description of the diffuse endocrine system.³ Pearse’s hypothesis did not hold up in experimental studies. Thus, while allograft studies of the neural crest conclusively showed the C cells of the thyroid, the parathyroids, the adrenal medulla, and the ganglion cells of the myenteric plexus were derived from the neural crest, they were equally conclusive in showing that the gastric and pancreatic endocrine cells originated from endoderm.¹⁵, ¹⁶ Other studies by Andrew¹⁷ and Pictet et al.¹⁸ showed that removing the neural crest from developing rabbit embryos does not prevent the development of insulin-producing B cells, which appeared to have a common endodermal precursor with pancreatic exocrine cells. Furthermore, cell turnover data pointed to the crypt base as the site of origin for all epithelial cell types, including the endocrine cells.¹⁹ Finally, studies on gastrointestinal tumors demonstrated an increasing number with mixed epithelial and endocrine cell components and rare tumors in which dense core granules and zymogenic granules were found within the same cells, while cell line cultures suggested that there are common endoderm precursor cells for both gut epithelial and endocrine cells.

The Pearse hypothesis linking all endocrine cells and neural cells presumably led to the designation of the endocrine cells as “neuroendocrine cells” and to the designation of the neoplasms containing such cells as “neuroendocrine neoplasms.” This seems cumbersome and even pedantic, since the cells of these tumors in the gut have no neural features. They look like endocrine cells. So, we will drop the neuro part of the adjective, and, in this chapter, we will refer to the cells as endocrine cells and to the neoplasms containing endocrine cells as endocrine neoplasms. After all, this is five letters shorter and takes nothing away from the subject. In the 200 World Health Organization book covering tumors of the digestive system, all the tumors containing these cells are included in chapter entitled “Endocrine tumors,” rather than “Neuroendocrine tumors,” so a precedent for this shorter name has already been established. However, the seventh edition of AJCC cancer staging manual has retained the term “neuroendocrine” and the newest WHO classification does the same, thus the saga continues.

Over the years, the gut endocrine cells have been characterized using different techniques which looked at them from different perspectives. First, there were the light microscopic basigranulated and clear cells. Later, as mentioned above, they were separated by the reactions of their granules with silver stains into argentaffin and argyrophil types. Next, they were analyzed by the electron microscopic appearances of their secretory granules which tended to be distinctive, resulting in an alphabet soup of cells called D, G, L, M, N, P, and S cells and there were also cells designated by various combinations of two or more letters. More recently, they have been characterized by the specific substances their granules contain based upon immunocytochemical techniques that employ antibodies directed against each of these substances. From a practical point of view, at this stage in our analyses, we rarely use the silver staining techniques, nor do we use the electron microscope. We have enormous amounts of data based on these analyses, but when we really want to find out what the cells are doing, that is, what they are making and possibly secreting, we have specific antibodies with which to stain them. Some cells communicate with the lumen of the glands or crypts or pits, while others are buried deep in these structures with no such communication.

Carcinoid tumor is a name from pathology antiquity coined in German by Oberndorfer in 1907 in a study of intestinal endocrine tumors as “karzinoide,” which meant carcinoma like.²⁶ Later, similar tumors in other sites in the gut and elsewhere, such as the lung, were given the same name. The classic carcinoid tumors were in the terminal ileum and appendix and produced serotonin. The term “carcinoid tumor” remains popular in spite of the attempt to replace it with the designation “well-differentiated endocrine tumor/neoplasm,” and published reports of these tumors continue to refer to them as carcinoid tumors, not by the more cumbersome and less catchy name. We know that carcinoid tumors are usually harmless, but we also know that some metastasize, so, the use of the carcinoma-like designation is as acceptable as is the term “adenocarcinoma” for pleomorphic, proliferating, often extensive, necrotic epithelial tumors that commonly metastasize, although many do not. Therefore, in this chapter, the term “carcinoid tumor” is used without synonyms and without apology. In fact, often they will be referred to simply as “carcinoids,” dropping the “tumor” word.

There is a certain lack of uniformity among clinicians and pathologists regarding the use of the term “carcinoid tumor.” Is this a specific type of endocrine tumor that is defined by its product, namely serotonin, or is it a generic type of tumor, based only on the low-grade cell type, regardless of the secretory product? The use of the term is clearly site related, since we use it for low-grade endocrine tumors of the gut, the lung, the thymus, and a few other sites, but very similar tumors in the endocrine pancreas have different names. Carcinoid tumors of the gut are among the most histologically uniform tumors anywhere in the body and they are comparable to low-grade tumors that arise in other endocrine organs, such as
the pituitary, the parathyroid glands, and the islets of Langerhans. Regardless of the sites in which they arise, they are composed of much the same type of bland-appearing cuboidal cells or short columnar cells that are, in general, remarkably uniform, except for a few nuclei that may be slightly bigger than their neighbors, or occasional cells that may have more fine cytoplasmic granules than do other cells (Fig. 5-3). Mitoses are usually infrequent. Most tumors have little or no necrosis. The only histologically interesting aspect is that these uniform cells are arranged in several patterns, including in large nests referred to as the insular pattern, in ribbons of cells that often interconnect; and in tubules or glands in which the cells are arranged around small lumens (Fig. 5-4). Certain growth patterns are more common in certain tumor locations, although there is considerable overlap among sites. Thus, although ileal tumors usually have the nest or insular pattern, every so often the cells are arranged in ribbons or tubules. Most textbooks and published papers also spend a lot of words describing the growth patterns of carcinoid tumors in different sites, with trabecular, glandular, or insular patterns occurring in this or that site, but these differences are not important in clinical practice. If we see a carcinoid tumor of the small intestine that has a trabecular growth pattern when the literature tells
us it is supposed to have an insular pattern, will we panic or deny its existence because it is growing the wrong way? Of course not! We will accept this aberrant growth pattern and wonder what it means in terms of behavior or treatment, or if it is an unusual cell type. In this chapter, we will spend few words about growth patterns. The carcinoid tumors differ from one site to another in the background mucosa in which they arise, in their relation to other hormones on which their constituent cells are dependent, and in the substances they produce.

After an exhaustive literature review of reported carcinoid tumors, Modlin and colleagues concluded that 38% of the carcinoid tumors occurred in the small intestine, 18% in the appendix, 21% in the rectum, 12% in the colon, and 6% in the stomach.²⁷ A subsequent Japanese study of 6,799 gastrointestnal carcinoid showed the distribution to be 24.5% in rectum, 18.4% in stomach, 13.6% in duodenum, 19.4% in jejunum and ileum, 15.5% in appendix, 4% in colon, and 0.7% in esophagus.²⁸ This kind of information gives us a rough estimate or location, but these numbers are likely to change with greater use of upper endoscopy with detection of small carcinoid tumors complicating autoimmune gastritis and lower endoscopy with detection of small rectal carcinoid tumors. The often quoted demographic information on asymptomatic tumors is useless. What difference does it make if appendiceal carcinoids are more common in males or females, or if they occur at one age or another? They are often incidental findings in appendectomies, which are usually performed for the clinical features of acute appendicitis, especially since incidental appendectomies performed during other abdominal operations are becoming less common.

The carcinoid tumors differ from one site to another in terms of the types of mucosa in which they arise, in their relation to other hormones on which their constituent cells are dependent, and in the substances they produce. In fact, many contain multiple substances including hormones and amines such as serotonin.

Almost all carcinoids in the appendix and rectum are asymptomatic as are most of the gastrin-induced tumors in the stomach that accompany both autoimmune atrophic gastritis and the Zollinger-Ellison syndrome (ZES), although there are a few exceptional larger metastasizing tumors in these settings. Symptoms caused by carcinoid tumors are the results of their bioactive products, an uncommon situation, or structural changes that the products induce.

Prognostically, metastatic risk for carcinoid tumors at all sites is size, site, and product dependent. This probably explains the differences in metastases between ileal and appendiceal or rectal tumors. At the time of diagnosis, ileal carcinoids as a group are larger than those at other sites, presumably because they are found when they produce symptoms, and in order to produce symptoms, they need to grow large enough to either metastasize or kink the bowel and cause obstruction. In contrast, appendiceal and rectal tumors are discovered by accident when they are small. Of course, this may also indicate inherent differences in growth capabilities. Ileal carcinoid tumors may have more growth potential than those in the other two sites. With regard to site and product, duodenal carcinoids producing gastrin (gastrinomas) are more likely to metastasize when they are much smaller than carcinoids at other sites producing other substances, and the same situation holds, although less so, for somatostatin-producing duodenal carcinoids. There is little information on the prognostic importance of other parameters such as the number of mitoses, nuclear pleomorphism, and vascular invasion.

There is an overwhelming desire among clinicians and pathologist alike to designate tumors as benign or malignant for good reasons; however, we have learnt from many situations in other organ systems and certain tumor types that this is not always possible, and this also holds true for endocrine tumors of the GI tract. For example, size and site are probably the most reliable criteria in predicting outcome. However, it is interesting that most texts, reviews, staging manuals, or studies on endocrine tumors do not even mention the practical problems associated with measurement of size of endocrine tumors. For example, most of the appendeceal or rectal carcinoids are detected incidentally and the size of the tumor is evaluated by measuring it on the slides. Every pathologist recognizes
that while measuring the two dimensions of a lesion on the slide is not problematic, constructing the three dimensions of a tumor is inaccurate, especially when dealing with measurements in millimeter range. This almost implies that the entire appendix has to be submitted for histology, the cross sections of the appendix have to be submitted in some form of sequence, and tissue blocks have to be exactly of the same thickness.

The issue with the rectal carcinoids detected incidentally in polyps is that when the tumor is present at the base of the excision biopsy, there is no way of knowing what is left behind. The only solace one can derive from this is that most of the data in the literature come from such inaccurate measurements! The other issue with size is that while carcinoids smaller than 1 cm at most sites behave in a benign fashion, identical tumors in the small intestine even when less that 1 cm are known to metastasize. In our own experience as well as in the literature, well-documented cases of metastatic carcinoids are described where extensive search failed to reveal a primary site. This raises a suspicion of either a very small primary or difficult-to-evaluate primary sites, including appendix. For these reasons, despite the fact that most carcinoids behave in a benign manner, they should be considered potentially malignant as a group and hence the need for a staging system.

The other confounding factor with these tumors is that histologically identical tumors have widely variable outcomes depending on site, location, and the nature of background disease. While less differentiated tumors have generally more aggressive behavior, well-differentiated histology does not guarantee a benign behavior. Thus histologic grading is also an issue and many different classification schemes are currently in use. The 2010 WHO classification divides the neuroendocrine neoplasms into two groups: (a) Well-differentiated neuroendocrine tumors that include the conventional carcinoids (well-differentiated neuroendocrine neoplasm), and (b) neuroendocrine carcinomas that include the atypical carcinoid/moderately differentiated endocrine neoplasms and poorly differentiated neuroendocrine carcinomas (small cell and large cell type). The idea appears to be that by using the term “tumor” as opposed to “carcinoma,” one can imply a more benign behavior, a concept that all may not agree with. The American Joint Committee on Cancer (AJCC, 2010) grading system further puts well-differentiated neuroendocrine tumors (carcinoids) in three grades (G1-G3) taking mitotic activity and proliferative index (ki67 index) into account, even though there are (currently) no data to support these of being of value outside of the pancreas.²⁹ Other systems have used histologic differentiation, mitotic activity, and tumor necrosis in classifying the endocrine tumors into well-, moderately, and poorly differentiated categories. In the absence of overwhelming data specific to gastrointestinal endocrine tumors, it is difficult to favor any one system over the other, and local/regional requirements may guide the reporting of these tumors.

The other problem with staging of endocrine tumors is that progressively increasing depth of invasion in the bowel wall does not necessarily imply a progressively poor outcome unlike adenocarcinomas. Despite many studies that show some trends, it is not clear whether some of the histologic features that in other cancers imply an aggressive behavior for example, angioinvasion, perineural invasion, or infiltrative growth pattern, indicate a poor outcome in endocrine tumors of the GI tract and would benefit from a more aggressive therapy. Thus formulating a staging system for these endocrine tumors has been difficult, but at last the AJCC has published a staging system in 2010 that takes the site, size, and extent of tumor spread into account.²⁹ Tumors from appendix, stomach, small intestine, and large bowel are all staged differently, making the system somewhat cumbersome and complex. However, having a formal staging system like this is likely to result in more uniform reporting of these tumors with better comparison of data across geographical borders, and one cannot argue against its utility.

The seventh edition of the cancer staging manual from the AJCC (2010) includes staging protocols for all endocrine (neuroendocrine) tumors of stomach, small intestine, colon and rectum, and appendix.²⁹ The definition of the T stages is site specific and is mainly based on size, involvement of muscularis propria and serosa, and involvement of adjacent structures. There is only a single N designation, meaning involvement of regional nodes is considered N1 regardless of the total number of nodes. This should be emphasized that this staging is applied only for the well-differentiated endocrine tumors/carcinoids, atypical carcinoids (well-differentiated neuroendocrine carcinoma), and gangliocytic paragangliomas. Moderately to poorly differentiated endocrine carcinomas (small cell and large cell endocrine carcinomas) and mixed tumors are staged just like carcinomas at that site. It still remains to be seen whether this staging system would have any practical implications in guiding treatment strategies, beyond what is achieved by accounting for site-specific management guidelines.

The esophagus is a rare site for carcinoid tumors. In fact, esophageal carcinoid tumors are so rare that there are many more small cell carcinomas in comparison.
As a result, there is little important information that has accumulated about esophageal carcinoid tumors.³⁰, ³¹ The few that have been reported are in the lamina propria and deeper, and they look like carcinoid tumors elsewhere, although some of them have been reported as atypical or malignant carcinoid tumors. They tend to produce single lumps or masses, but some of them have been found in association with Barrett’s mucosa, even as separate foci adjacent to invasive Barrett’s adenocarcinomas. In fact, it is possible that the Barrett’s setting is more common than the sporadic setting. It appears that some of the reported cases are not typical carcinoid tumors with uniform cells and nuclei, but atypical forms or even small cell carcinomas. So the esophageal carcinoid literature may be contaminated with other tumors. There is little long-term follow-up information, so it is impossible to get hard data about metastases, survival, and features of the tumors that correlate with adverse outcome. However, the limited published data suggest that esophageal carcinoid tumors are likely to have a favorable outcome, but survival seems to be stage related and possibly also related to whether they are histologically typical carcinoids or if they have atypical features, such a pleomorphism, mitoses, or necrosis. They do not make anything that results in any clinical syndrome.

The stomach is a rich source of carcinoid tumors, but most, by far, are clinically insignificant in terms of long-term patient survival. Gastric carcinoid tumors occur in three well-established different settings, with a potential fourth, all of which are described below³² (Table 5-3).

Type I carcinoid tumors. These arise in the body or fundic mucosa that has been ravaged by (autoimmune) atrophic gastritis, or simply “autoimmune gastritis,” since atrophy is simply one of several microscopic characteristics. All patients with pernicious anemia (PA) have this form of atrophic gastritis, but so do a lot of other patients, almost certainly resulting from “creeping atrophy” secondary to Helicobacter gastritis that extends by direct extension from the antrum with age in Hp+ patients.

PA is a macrocytic anemia that is caused by vitamin B₁₂ deficiency, as a result of intrinsic factor deficiency. Patients with autoimmune gastritis also commonly have autoantibodies to parietal cells, an intrinsic factor, or both. Increasingly it appears that many of these occur on a background of long-standing H. pylori gastritis, in whom parietal cell antibodies are relatively common. Indeed, it is now unclear whether the “typical” variant of PA seen in patients, e.g., in patients of Scandinavian decent, really needs H. pylori as a triggering or potentiating agent. However, there appear to be genuine hereditary forms of the disease, such as that associated with autoimmune polyendocrine syndrome type 1, or the X-linked poly-endocrinopathies that include the immune dysfunction, polyendocrinopathy, and enteropathy, X-linked (IPEX) syndrome that do not require H. pylori.

Autoimmune gastritis is a condition in which parietal cells are destroyed, leading to lack of acid production, which, in turn, leads to loss of acid inhibition of gastrin production by the antral G cells, leading to hypergastrinemia (the pathology of this is described further in Chapter 13). Normally, the EC-like cells or ECL cells on the basement membrane of the glands are stimulated by gastrin to secrete histamine, but hypergastrinemia also leads to proliferation of these cells, especially when there are no parietal cells which are also targets of gastrin. This proliferation takes several patterns, but the most obvious is the nodular pattern in which tiny nodules of ECL cells develop at the base of the mucosa (Fig. 5-6). Other changes also occur in autoimmune gastritis, including glandular atrophy; pyloric, pancreatic, and intestinal metaplasias; and inflammation, which can direct the observer to look for the ECL-cell nodules. It is thought that such nodules progressively enlarge and eventually develop neoplastic features, including infiltration of mucosa and submucosa. The type I gastric carcinoid tumors account for at least three-fourths of all gastric
carcinoids. Thus any generic discussion of gastric carcinoid tumors is likely to be heavily weighted toward these type I tumors. This number is likely to increase as more cases of unsuspected autoimmune gastritis are uncovered as a result of increasing upper endoscopy. At what point is an ECL-cell proliferation considered to be a carcinoid tumor? There are some seemingly arbitrary definitions that simplify this issue, which are listed in the WHO book on tumors of the digestive system.³³ First, there is a phase referred to as “dysplasia” in which the cells become relatively atypical and the nodule enlarge and fuse with micro-invasion of newly formed stroma (Fig. 5-7). Then, when the nodules reach a size larger than 1/2 of a millimeter or
invade into the submucosa, they are classified as carcinoid tumors/NETs (Figs. 5-8 and 5-9). Type I tumors are invariably asymptomatic, commonly multiple, and small and often microscopic. These tumors tend to be easily treatable by local endoscopic excision if large or forming an endoscopic polyp, or by simple follow-up to ensure tumors are not enlarging, which is rarely seen, with a generally excellent outcome. They rarely metastasize, probably because they are not large enough when we find them. Only when a tumor reaches about 2 cm across, is metastasis a concern and type I tumors of this size are rare. Lymph node metastases from these tumors are rare, and deaths almost unrecorded.

Any time a gastric carcinoid/NET is biopsied incidentally, usually as the result of the finding of a small endoscopic nodule or polyp in the body or fundus, the chances are at least 3 to 1 that it is a type I tumor, and almost all these patients need no additional therapy. Periodic surveillance endoscopy is advisable in patients with autoimmune gastritis primarily to screen for dysplasia or carcinomas, for which the risk is increased by about a factor of 7. We have seen several cases in which total gastrectomies were performed for type I carcinoid tumors, which was unnecessary and a therapeutic overkill. Therefore, whenever a gastric carcinoid tumor/NET is found in a biopsy, it is critical that the flat body mucosa be
examined for autoimmune gastritis. Unfortunately, in most such situations, the endoscopists usually only biopsy the small polyp and the status of the adjacent mucosa remains unknown. In such cases, the pathology report should discuss this issue that there is insufficient adjacent mucosa to tell if the carcinoid is arising in autoimmune gastritis, and that this determination is critical for patient management. Often, a repeat upper endoscopy is required to sample the flat body mucosa from greater curvature, lesser curvature, and incisura. Surprisingly, antral biopsies can also be of value in showing a G-cell hyperplasia in atrophic gastritis, but usually no G-cell hyperplasia when there is a gastrinoma; however, antral biopsies alone in the absence of sampling from the body mucosa are insufficient to render a diagnosis of autoimmune gastritis. The presence of Helicobacter or extensive multifocal atrophy with metaplasia might indicate current or prior H. pylori infection. An ancillary test is serum gastrin, which in these patients is very high, excluding the diagnosis of sporadic (type III) carcinoids; however, one should ensure that the patient is off any proton pump inhibitors for sufficient time before measuring serum gastrin levels.

Type II carcinoid tumors. These arise as a result of excessive gastrin production by gastrin-producing endocrine tumors, known as gastrinomas, resulting in the ZES. Presumably, they evolve much the way the type I carcinoids evolve, through a hyperplasia-neoplasia sequence of ECL cells. However, the background body mucosa is not atrophic, but thicker than normal due to hypertrophy of the parietal cells, also caused by the hypergastrinemia. Although the ZES occurs both sporadically and as a consequence of MEN-I, virtually all of the patients with type II carcinoid tumors have MEN-I. Gastric carcinoids arising in the sporadic ZES are extremely rare.

Although the bulk of G cells are in the gastric antrum, peculiarly, the gastrinomas are almost always in the duodenum in MEN-I, not in the stomach, whereas the pancreas is a more common site for sporadic gastrinomas. Very rarely, the cause of the hypergastrinemia is hyperplasia of the gastric antral G cells. This is discussed in more detail later in this chapter under hyperplasias. Type II carcinoid tumors only account for about 5% of all gastric carcinoid tumors. The type II ECL tumors, those that arise secondary to the gastrinomas in the MEN-I syndrome, are like the type I tumors, small and multiple, and usually occur in the body and fundus, but a few have been found in the antrum.³⁴, ³⁵, ³⁶ In contrast to the type I tumors, they are likely to be more aggressive and metastasize more often to lymph nodes and liver, so they may need more aggressive therapy. Possibly this is due to some effect of the somatic defect due to abnormalities in the Menin gene. The combined data concerning type I and II tumors suggest that hypergastrinemia by itself can cause ECL-cell hyperplasia, but not neoplastic transformation and the development of carcinoid tumors. This additional step to neoplasia apparently requires other genetic defects. In addition to the MEN-I genetic change, Bcl2, p53, and MMP9 have been implicated.³²