Chapter 5, Page 2 
Chapter 5, Page 2
BARRETT'S ESOPHAGUS, COLUMNAR DYSPLASIA, AND ADENOCARCINOMA OF THE ESOPHAGUSBARRETT'S ESOPHAGUSThe most common predisposing factor for Barrett's esophagus is chronic gastroesophageal reflux. Evidence in support of this comes from the following observations: 1) patients with clinical gastroesophageal reflux have a much higher prevalence of Barrett's esophagus when compared with patients undergoing endoscopy for other reasons (13, 54, 76); 2) patients with reflux-associated conditions, such as scleroderma and achalasia after esophagomyotomy, have an increased prevalence of Barrett's esophagus (1, 46, 73, 88), and one patient with Zollinger-Ellison syndrome developed Barrett's esophagus and esophageal adenocarcinoma (125); 3) most children with Barrett's esophagus have a history of regurgitation in infancy and reflux symptoms, and many have gastroesophageal reflux as documented on radiography or by pH monitoring (21, 47); and 4) the geographic variation in the prevalence of Barrett's esophagus parallels a similar variation in gastroesophageal reflux. Other factors may also be associated with Barrett's esophagus, albeit much less commonly. These include esophageal bile reflux associated with the postgastrectomy state (73), esophageal injury following lye ingestion, and genetic predisposition. The common link appears to be esophageal acid reflux as it has been shown that alkaline reflux parallels acid reflux and can be largely abolished with potent acid antisecretory therapy (5, 20, 122, 133a). It has also been suggested that there is a relationship between Helicobacter pylori infection in the stomach and Barrett's esophagus. However, reports regarding this relationship are contradictory and remain unproven (7, 80, 128). It is still possible that in some patients, Barrett's esophagus has a congenital basis. Congenital rests of gastric epithelium usually occur in the cervical epithelium but occasionally may be found in the distal esophagus (89). It is theoretically possible that some cases of Barrett's esophagus occur when these small rests expand in response to gastroesophageal reflux. The sequence of events leading to Barrett's esophagus has not been clearly defined. It is probable that ulceration of the squamous epithelium occurs in response to gastroesophageal reflux of acid, bile, and duodenal contents and that reepithelialization occurs via multipotential stem cells which in turn differentiate into the variety of epithelial cells found in Barrett's esophagus. There may, however, be other mechanisms. It is possible, although unproven, that metaplasia occurs simply by the upward migration and overgrowth of columnar epithelium from the gastric cardia, in response to gastroesophageal reflux. Alternatively, as was postulated in the past, Barrett's esophagus may arise from congenital rests of gastric columnar epithelium, which usually occur in the cervical esophagus but have also been documented in the distal esophagus. Although these rests may account for rare cases of childhood Barrett's esophagus, there is no evidence of this mechanism in adults (25, 63, 84). Columnar Dysplasia in Barrett's Esophagus and Its Association with Esophageal Adenocarcinoma. There is good evidence that the most important risk factor for esophageal columnar dysplasia is prolonged gastroesophageal reflux complicated by Barrett's esophagus. Most patients with dysplasia have a history of clinical gastroesophageal reflux (20, 35, 46, 139), and dysplasia almost invariably develops in Barrett's epithelium (68). The international variation in the prevalence of esophageal columnar dysplasia parallels a similar variation in gastroesophageal reflux. Dysplasia, especially high grade, is considered to be a precursor lesion for adenocarcinoma, based largely on retrospective histologic studies of adenocarcinoma in which dysplasia is commonly found in the adjacent mucosa. In addition, some patients with Barrett's esophagus followed by serial endoscopy appear to progress to low-grade dysplasia, then to high-grade dysplasia, and on to cancer. It is presumed that low-grade dysplasia reflects a heightened cancer risk but this risk is less well defined than for high-grade dysplasia (15). Thus, the finding of dysplasia in a given patient may not only be a marker of heightened cancer risk but may itself already be associated with early invasive carcinoma elsewhere (68, 96). However, it should be stressed that the time sequence for the progression of dysplasia to carcinoma is variable, that it is not inevitable, and that even when dysplasia develops, it commonly does not progress further (44, 68, 96). The pathogenic factors responsible for the transformation of epithelium in Barrett's esophagus to dysplasia and carcinoma remain unclear. It is most likely that they are multifactorial and include hereditary factors, infections with H. pylori, and continued reflux injury. It remains unclear whether it is the exposure to refluxing gastric acid or to the bile which predisposes this epithelium to dysplasia and carcinoma. There are some reports which have suggested that patients with Barrett's esophagus and complications, ulcers, strictures, or dysplasia, are more likely to have alkaline reflux than those with uncomplicated Barrett's disease (5). Some patients have developed dysplasia and carcinoma following antireflux therapy (16a, 43, 84, 97, 107, 110, 124). However, this could be because acid suppression was incomplete. In fact, we and others have seen columnar-lined mucosa showing squamous metaplasia (suggesting possible regression of Barrett's esophagus) with maximal medical therapy or antireflux surgery (16a, 43, 55, 107, 113, 114, 115, 125). Whether such therapies reduce the risk of adenocarcinoma in the esophagus is as yet unknown. Regression of Dysplasia in Barrett's Esophagus. There have been sporadic reports of the regression of dysplasia in Barrett's esophagus (15, 31, 90, 91, 97, 107, 115). Since it has been shown that the development of dysplasia is accompanied by genomic instability and progressive genetic changes, it is hard to visualize regression of this process. To us it seems more likely that if dysplasia truly disappears, it may result from the body's immunologic response or the progression of the clonal process to a point of nonviability (i.e., necrosis). Other possible explanations for apparent regression of dysplasia are that the histologic changes diagnosed as dysplasia were in fact reactive atypia rather than true dysplasia; that dysplasia is still present but not detectable endoscopically; and that dysplasia was entirely removed by biopsy. DNA and Oncogene Markers. Flow cytometry studies of epithelial DNA content in Barrett's esophagus with dysplasia (and also occasionally without) and carcinoma have clarified the sequence of events leading up to carcinoma at the cellular level. Barrett's esophagus is only rarely associated with abnormal nuclear DNA content (aneuploidy) or cell cycle abnormalities; however, such abnormalities are common in dysplasia and carcinoma (31, 71, 71a, 94). Studies by Rabinovitch et al. (85) and others (31) have shown that the dysplasia of Barrett's esophagus commonly encompasses several different populations of aneuploid cells, but that early adenocarcinomas usually have only a single clone of aneuploid cells. These findings indicate that Barrett's esophagus is prone to genomic instability, which can result in multiple aneuploid populations; some patients develop a clone of cells that is capable of invasion (15, 33, 41, 44, 45). Only a few oncogene studies have been performed on patients with Barrett's esophagus with and without dysplasia and carcinoma. Studies of p53 gene mutations have demonstrated different clones in mucosae from patients with Barrett's dysplasia and patients with adenocarcinoma (9, 18, 55a, 59a, 109). In addition, point mutations affecting the adenomatous polyposis coli (APC) suppressor gene, which were previously described for familial polyposis, have now also been demonstrated in both adenocarcinoma and squamous cell carcinoma of the esophagus (11). These findings support the concept that several genetic alterations are required for tumorigenesis to occur in Barrett's esophagus. In another oncogene study, it was found that c-myc expression was consistently observed in all grades of dysplasia and carcinoma and the H-ras gene was consistently expressed in higher grades of dysplasia and carcinoma but absent in nondysplastic epithelium. H-ras expression may be a helpful marker for identifying dysplastic lesions that can progress to carcinoma (78). However, the potential clinical utility of specific oncogenes in predicting heightened cancer risk remains to be clarified (40). Prevalence and Incidence. Barrett's esophagus is a complication that occurs in 10 to 20 percent of all patients with prolonged gastroesophageal reflux (15, 80, 110, 118); these statistics come from studies of patients who had endoscopic evaluations for gastroesophageal reflux disease (120). However, a number of studies have indicated that Barrett's esophagus is probably more common than was previously thought. One study based on a review of all the records of patients with known Barrett's esophagus treated at the Mayo Clinic and other health providers in Olmsted county, Minnesota, found that the incidence of clinically apparent Barrett's esophagus was 18 cases per 100,000 population (17). Another study, based upon a prospective review of all autopsies performed at the Mayo Clinic, found that the frequency of Barrett's esophagus was 376 per 100,000 population (118). These two studies suggest that for each case of Barrett's esophagus being treated, there may be as many as 20 unrecognized cases in which dysplasia or carcinoma never develop (118). Many cases are not detected because only one third of patients with Barrett's esophagus in some series are symptomatic ( 19, 116, 124) and there is some evidence the Barrett mucosa is less sensitive to refluxed acid than is normal squamous mucosa (52, 116). The reported prevalence of dysplasia in Barrett's esophagus varies from 14 percent to as high as 40 percent (22). However, true prevalence figures for dysplasia are as difficult to obtain as they are for Barrett's esophagus, for the reasons given above. Furthermore, difficulty arises because most studies tend to lump together the findings of dysplasia and carcinoma. Thus, although it is true that most cases of adenocarcinoma of the esophagus are associated with dysplasia, there are a significant number of cases of dysplasia of the esophagus, especially low-grade dysplasia, without carcinoma. The frequency of malignant change in Barrett's esophagus remains unclear. The prevalence rate of adenocarcinoma at the time of diagnosis of Barrett's esophagus is 7 to 15 percent (79). Based on retrospective studies, the incidence of adenocarcinoma of the esophagus developing in patients with Barrett's esophagus without evidence of dysplasia has been estimated at between 1 in 440 and 1 in 52 patient years, a risk that is 30 to 125 times greater than that in the population without Barrett's esophagus. Speckler (119, 121) has extrapolated these data to calculate an incidence of 500 cancers per 100,000 patients with Barrett's esophagus per year. These figures imply that the risk of developing carcinoma in the lifetime of a patient who has Barrett's esophagus at age 50 is 5 to 10 percent (16a, 50, 84, 120, 121, 124). The wide range of estimates reflects the retrospective nature of most of the studies or the bias related to the selection process of patients sent to referral centers (118, 120). For example, in one referral population of 50 patients, 7 had dysplasia at the beginning of the study, 13 developed dysplasia, and 5 developed invasive carcinoma in a follow-up period averaging 5.2 years, an incidence of carcinoma 125 times over that of the general population. It is also possible that the patients who developed cancer already had it at the initial presentation but it was histologically undetected (42). There are only a few prospective studies evaluating the true cancer incidence in Barrett's esophagus. In a study conducted by the American College of Gastroenterology, 6 of 220 patients with Barrett's esophagus developed adenocarcinoma (1 case per 150 person years). The mean length of follow-up was about 4 years (range, 0.1 to 19.1 years) (84). In two other studies, the incidence was 1 per 175 person years (120) and 1 per 441 person years (16a). Thus, although high-grade dysplasia is a good indicator for high risk of adenocarcinoma, better prospective long-term follow-up is needed to determine the proportion of patients who do develop adenocarcinoma. The risk of carcinoma developing in short-segment Barrett's esophagus, defined as a 1- to 2-cm affected segment in the distal esophagus, is unknown (37, 41). The risk of developing cancer appears to be similar to that with usual cases of Barrett's esophagus and is unrelated to the length of the columnar lined segment (45, 68). Another problem lies in separating these cases from carcinomas of the gastric cardia (see Differential Diagnosis of Esophageal Adenocarcinoma). However, increasing evidence suggests that adenocarcinoma of the gastric cardia belongs to the same spectrum of disease as Barrett-associated adenocarcinoma (53, 66, 137). There is marked geographic variation in the prevalence of Barrett's esophagus with dysplasia, which parallels the prevalence of gastroesophageal reflux. It is much less common in Japan and China, where reflux esophagitis is rare, than in Europe and North America. Clinical Features. Barrett's esophagus, with or without dysplasia, is most commonly diagnosed between the ages of 40 and 60 years (23), but it can also occur in children (47). There is a clear- cut predominance in white males over females, in the order of 4 to 1 (84). The disorder is uncommon in blacks: less than 2 percent of blacks with gastroesophageal reflux have Barrett's esophagus (84). There is an association with heavy cigarette smoking and alcohol abuse. In one study, 85 percent of patients with Barrett's esophagus were heavy cigarette smokers and 16 percent were alcohol abusers (120). Most patients with Barrett's esophagus and dysplasia are asymptomatic. Those with symptoms exhibit no unique clinical features beyond those of chronic gastroesophageal reflux disease. These clinical features include regurgitation, heartburn, dysphagia, and odynophagia (pain on swallowing) which is usually due to ulceration or stricture. A change in symptoms, such as progressive dysphagia, is likely to portend the development of advanced esophageal cancer (116, 124). Squamous cell carcinoma and adenosquamous tumors of the esophagus have been reported in patients with Barrett's esophagus (103, 105). In one study, Barrett's esophagus was found in 13 of 42 cases (40 percent) of squamous cell carcinoma of the esophagus, however, the dysplasia and carcinoma were separated from the Barrett's esophagus by normal squamous epithelium, supporting the view that the two are not pathogenetically related (105). It is probable that the simultaneous occurrence of Barrett's esophagus and squamous cell cancer is related to common risk factors, such as alcohol and smoking. Data on whether the incidence of colonic adenomas and adenocarcinomas is increased in patients with Barrett's esophagus is conflicting (86, 99, 117, 132). Gross and Endoscopic Findings. Barrett's esophagus almost always occurs in conjunction with a sliding esophageal hernia (fig. 5-1). The endoscopic features and the gross appearance of resected specimens are quite distinctive. The squamocolumnar junction is displaced, with a salmon- pink mucosa extending proximally from its usual location at the esophageal sphincter (figs. 5-4A,B). Whitish squamous islands are commonly seen in the proximal few centimeters (figs. 5-4C,D). Ulcerations may be seen anywhere in the zone of metaplasia. Segments of Barrett mucosa shorter than 5cm may not exhibit circumferential involvement but may appear as tongues of red mucosa extending cephalad (figs. 5-4E,F). In almost all cases, the endoscopic and gross appearances of Barrett's esophagus with dysplasia or early cancer cannot be distinguished from those of simple Barrett's metaplasia, so multiple biopsies are usually necessary. However, on rare occasions dysplasia has presented as small, sometimes multiple, pedunculated adenomas (44, 57, 70, 130, 137). Originally it was thought that dysplasia was a multifocal process. However, mapping studies that we and others have done show that it is commonly a unifocal process, as is the invasive carcinoma when present. The dysplastic process may involve the entire Barrett's mucosa or a limited area (68, 96). Microscopic Findings. By definition, Barrett's esophagus is a metaplastic process in which columnar epithelium, usually accompanied by underlying mucous glands, replaces the normal esophageal squamous epithelium for variable lengths from the lower esophageal sphincter region cephalad (figs. 5-5, 5-6). The metaplastic epithelium is typically characterized by the presence of a variety of cells such as goblet cells, gastric and intestinal (small bowel and colonic) columnar cells, intermediate cells (with mixed features of gastric and intestinal cells), Paneth cells, and endocrine cells. Pancreatic acinar metaplasia may also be found in about 10 percent of individuals, primarily in the region of the lower esophageal sphincter (figs. 5-6, 5-7, 5-8, 5-9, 5-10, 5-11A,B, 5-11C,D,E,)5-12) (55b, 61, 143). The presence of goblet cells is the most useful feature for diagnosis of Barretts esophagus since these cells are not normally present in gastric cardiac or fundic mucosa. This is especially important since the gastroesophageal junction, which is the point at which the tubular esophagus joins the saccular stomach, does not always coincide with the squamocolumnar junction. The latter may lie anywhere within the region of the lower esophageal sphincter which occupies approximately the distal 2 cm of the esophagus. As a result, the finding of gastric cardiac or fundic mucosa within the lower esophageal sphincter should not be considered indicative of Barrett's metaplasia whereas goblet cells should. Goblet cells also help distinguish between an eccentric esophagogastric junction and a short Barrett segment. However, from a practical point, the exact site of the goblet cells within the gastroesophageal region is not that critical since this metaplastic change denotes an increased risk of cancer both in the esophagus and the gastric cardia (36, 37, 122a). The metaplastic epithelium has been given a variety of names such as specialized epithelium (45), which we favor; distinctive epithelium (80); and Barrett's mucosa. Histologically, it is identical to intestinal metaplasia of the incomplete type (type II and III) in the stomach (69). Another problem sometimes results from misinterpretation of the finding of fundic- type mucosa. Fundic mucosa does not occur in Barrett's esophagus but is the result of a hiatus hernia, which is almost invariably associated with Barrett's esophagus. In some patients the landmarks of the lower esophagus may not be clearly visible especially if the lower esophagogastric sphincter is patulous (36). In such patients the location of the lower esophageal sphincter may be identified by gently insufflating air to demarcate the saccular hernia from the tubular esophagus and to demonstrate the proximal margins of the gastric folds which end at the sphincter (36, 37). There are two major epithelial components of Barrett's esophagus: the surface epithelium and the underlying glands (fig. 5-6). Additionally, it has been shown that there may be reduplication of the muscularis mucosae with interposition of lamina propria between the two muscular layers. This may have clinical implications if invasion of the first layer of muscularis mucosae does not involve the submucosa because of new layers of lamina propria and muscularis mucosae. However, this remains to be confirmed (127). Histologic Findings. The mucosal surface is usually flat and composed of pits or crypt-like structures as in the stomach but can be villiform especially if inflamed (fig. 5-7). The epithelium covering the surface and the pits is lined by a blend of goblet cells and columnar cells (fig. 5-6). The goblet cells characteristically contain acidic mucins, usually consisting of admixtures of sialomucins and lesser amounts of sulfomucins, which stain positively with Alcian blue at a low pH (2.5) (figs. 5-11A,B). The columnar cells are often heterogeneous: some resemble normal gastric foveolar epithelium, intestinal absorptive cells, or gastric mucous neck cells, whereas others are atypical (5-11C,D,E). For example, some of gastric- type columnar cells have an eosinophilic rather than clear cytoplasm. A significant number of these cells stain for acid mucins with Alcian blue at low pH (77a). A number of intestinal- type columnar absorptive cells have only a partially developed brush border. The mucous cell metaplasia of Barrett's esophagus may also include colon-like cid- sulfated mucins, demonstrable with the high iron diamine stain (fig. 5-11E) (28). Approximately half or more of biopsy specimens from Barrett's esophagus have scattered cells containing these mucins (59, 143). The question is frequently raised as to whether the Alcian blue stain is required for the diagnosis of Barrett's esophagus. In most instances goblet cells are numerous and easily found and therefore the special stain is unnecessary. However, Alcian blue stain is useful in detecting isolated goblet cells when histologically differentiating a short- segment Barrett's esophagus and an eccentric gastroesophageal junction. Glandular Component of Barrett's Esophagus. In addition to metaplasia of surface and pit epithelium, there is metaplasia of mucous glands (figs. 5-5, 5-6). These glands are usually present as a thin band overlying the muscularis mucosae, but often they are sparse or even absent. They are usually composed of pure mucous glands but occasionally contain scattered parietal cells and resemble cardiac- type glands. As previously mentioned, we believe that if fundic- type mucosa is found in the esophagus, it almost always represents a hiatus hernia rather than Barrett's esophagus. In addition, Barrett's esophagus should not be confused with congenital islands of ectopic gastric mucosa, usually of fundic type. These occur mainly in the cervical esophagus (the inlet patches) and are characteristically separated from the stomach by a zone of intact squamous epithelium (see Heterotopic Gastric Mucosa at the end of this chapter). Sometimes the glands become cystically dilated and may insinuate themselves between the muscle fibers of the muscularis mucosae and thus mimic early adenocarcinoma, especially if there is dysplasia. Differentiation between benign, dysplastic, and malignant cysts depends upon a careful examination of the cytology of the cells lining the cysts. In the case of Barrett's esophagus, the cells lack features of dysplasia and carcinoma. Other gastrointestinal elements may be present within the glands. These include Paneth cells (fig. 5-12 ) (39, 56, 98, 112), endocrine cells, and pancreatic acini. Pepsinogen I, normally found in chief cells, and pepsinogen II, also found in chief cells and in antral and cardiac glands, have been demonstrated immunohistochemically in Barrett's mucosa (72). There is usually only a mild accompanying inflammatory infiltrate of mononuclear cells in the lamina propria of Barrett's esophagus. However, sometimes the inflammation is intense, with heavy neutrophilic infiltration and erosions. H. pylori may be found in the esophagus of some patients with Barrett's mucosa, primarily in the cardiac- type mucosa, especially in mucosa that is actually inflamed (65, 81, 128, 129). The organisms are present only in the distal 4 cm of the esophagus and only when they occur in the stomach as well (81, 129). Thus, their presence appears to represent a consequence of reflux rather than a primary infection. Biopsies from eroded or ulcerated areas reveal a superficial fibrino- purulent exudate, underlying granulation tissue, and occasionally, bizarre stromal cells (fig. 5-9). Sometimes biopsies from what appears to be an ordinary Barrett's mucosa show large foci of granulation tissue when an erosion is not apparent endoscopically. Exuberant granulation tissue may on occasion mimic glandular epithelium and be confused with adenocarcinoma (figs. 5-10), particularly at frozen section; serial sections help differentiate the microvessels of granulation tissue from malignant glandular formation and factor VIII immunostaining helps confirm the vascular nature of the lesion. The bizarre stromal cells can sometimes be confused with sarcoma. Biopsy specimens sometimes show squamous epithelium overlying specialized columnar epithelium (110), especially from the proximal zone of Barrett's esophagus. This change has been designated as "pseudoregression" or squamous metaplasia of the columnar epithelium (fig. 5-13). Presumably this represents an attempt to reverse the metaplastic process. This finding may be seen in cases of partial regression of Barrett's mucosa after antireflux medication or antireflux surgery (fundoplication) (16a, 43, 55, 107, 113, 114, 115, 125). |
