The normal esophagus is lined with squamous epithelium.

.. ALAN J. CAMERON, M.D. In Barrett's esophagus, the squamous lining of the lower esophagus is replaced by columnar epithelium. Barrett's esophagus is associated with gastroesophageal reflux and an increased risk of the development of esophageal cancer. Endoscopy shows red columnar epithelium in the lower esophagus. Biopsy is needed to confirm intestinal metaplasia. Some cases progress from dysplasia to invasive adenocarcinoma. Medical or surgical anti reflux treatment controls symptoms and esophagitis, but Barrett's esophagus remains. Patients are usually followed up by endoscopy for detection of dysplasia or early cancer. For patients with low-grade dysplasia, follow-up is adequate; however, for those with high-grade dysplasia, esophagectomy or experimental endoscopic mucosal ablation is advised. Mayo Clin Proc 1998;73:457-461 H,RAs = Histamine H 2 receptor antagonists; PPIs = proton pump inhibitors The normal esophagus is lined with squamous epithelium. In Barrett's esophagus, the squamous lining of the lower esophagus, damaged by gastroesophageal reflux of acid in conjunction with pepsin and bile, becomes replaced by columnar epithelium. The two clinically important aspects of Barrett's esophagus are its association with symptomatic gastroesophageal reflux disease and the increased risk of the development of esophageal cancer (adenocarcinoma). PATHOPHYSIOLOGY Patients with Barrett's esophagus usually have severe gastroesophageal reflux. Esophageal motility testing in patients with Barrett's esophagus shows a weak lower esophageal sphincter that, in conjunction with a hiatal hernia, allows reflux to occur. Esophageal peristalsis is often impaired; thus, refluxed acid remains in the esophagus longer and causes more damage. Reflux esophagitis, with superficial erosions in the squamous epithelium above the columnar-lined segment, is often present. Benign craters resembling gastric ulcers may occur in the columnar-lined area. Inflammation and resultant fibrosis may lead to an esophageal stricture, often at the upper end of the Barrett's esophagus. SYMPTOMS Most patients with Barrett's esophagus have a history of heartburn and acid regurgitation. Typically, heartburn is described as a burning discomfort behind the sternum, From the Division of Gastroenterology and Hepatology and Internal Medicine, Mayo Clinic Rochester, Rochester, Minnesota. Address reprint requests to Dr. A. J. Cameron, Division of Gastroenterology and Hepatology, Mayo Clinic Rochester, 200 First Street SW, Rochester, MN 55905. Mayo Clin Proc 1998;73:457-461 457 rising upward toward the throat. Heartburn often occurs after meals and when a patient is lying down at night. Temporary relief from an antacid is common. Regurgitation of sour-tasting fluid may occur when the patient bends forward, or it may awaken the patient at night because of coughing and a choking sensation. A few patients do not have reflux symptoms, and the diagnosis of Barrett's esophagus is an incidental finding. Strictures cause difficulty in swallowing solid food. Less frequent symptoms of Barrett's esophagus include chest pain, hematemesis, or melena. Only about 4 to 10% of patients with frequent reflux symptoms have a long segment of Barrett's esophagus (defined as at least 3 em of columnar epithelium in the esophagus). Distinguishing patients with this condition from those with uncomplicated reflux disease is not possible on the basis of symptoms alone; symptoms are similar whether Barrett's esophagus is present or not. DIAGNOSTIC TESTS Upper gastrointestinal roentgenography with use of barium is an insensitive method for the detection of Barrett's esophagus. In most cases, it shows a hiatal hernia, and the radiologist often notes free reflux of barium into the esophagus, but these are nonspecific findings. An esophageal ulcer crater or a benign-appearing mid-esophageal stricture is more suggestive of Barrett's esophagus, but most cases do not have either of these findings. Endoscopy is needed to establish the diagnosis. In the normal esophagus, the junction between the red gastric columnar epithelium and the pink esophageal squamous epithelium is found at the extreme lower end of the esophagus. In Barrett's esophagus, this junction is displaced upward, and red columnar epithelium extends into the 1998 Mayo Foundation for Medical Education and Research

458 esophagus and is easily distinguished from the pale pink squamous epithelium seen proximally. After detecting Barrett's esophagus, the endoscopist searches for any associated reflux esophagitis, esophageal ulcer, stricture, or hiatal hernia and especially for any evidence of esophageal carcinoma such as nodularity or a mass. Previously, one definition of Barrett's esophagus necessitated that at least 3 em of columnar epithelium had to be present in the esophagus. Investigators now recognize that shorter segments or tongues of columnar epithelium can be associated with adenocarcinoma of the esophagogastric junction. A short Barrett's esophagus may be diagnosed if endoscopy shows a definite area of columnar-appearing red epithelium in the lower esophagus and a biopsy shows intestinal metaplasia. A biopsy is needed to confirm the diagnosis of Barrett's esophagus. The columnar epithelium of the upper stomach, immediately below the esophagogastric junction, is of the gastric fundic or cardiac type. The histologic hallmark of Barrett's esophagus is the finding of intestinal metaplasia (also called specialized columnar epithelium) in the esophagus. Mucin-containing goblet cells are in this epithelium. The shape of the goblets within the cell cytoplasm is easily seen with standard hematoxylin-eosin-stained sections and can be seen more vividly with alcian blue staining. Goblet cell intestinal metaplasia constitutes most of the area of Barrett's esophagus. This histologic type is found in more than 95% of cases in which the endoscopist has seen a long (greater than 3 em) segment of Barrett's esophagus. This type of epithelium is associated with esophageal adenocarcinoma. When multiple biopsies show no intestinal metaplasia but only normal gastric cardiac or fundic epithelium, the diagnosis of Barrett's esophagus should be questioned. The biopsy specimens may have been obtained from columnar epithelium within a diaphragmatic hernia, not from the esophagus. With no intestinal metaplasia, the patient probably does not have an increased risk of cancer and therefore does not need to undergo the anxiety, discomfort, and expense associated with future endoscopic follow-up. Ambulatory (24-hour) esophageal ph monitoring, a clinically available test, usually shows severe acid reflux in patients with Barrett's esophagus. It does not, however, distinguish between patients with and those without Barrett's esophagus and thus is rarely needed. The same is true for esophageal motility testing, except that a motility test may be indicated before an antireflux operation as a guide to the type of procedure that should be performed. TREATMENT OF REFLUX IN BARRETT'SESOPHAGUS As a general principle, the treatment of reflux is the same whether Barrett's esophagus is present or not. Traditional advice to patients with symptomatic reflux disease has been to elevate the head of the bed at night by using blocks or a foam wedge. Patients are also advised to lose weight, to avoid eating within 3 hours of bedtime, and to cut back on large or high-fat meals. Chocolate, peppermint, and raw onions may aggravate reflux and are best avoided. Antacids taken as needed are useful for control of symptoms. These simple and inexpensive measures are sufficient in many patients with mild reflux. Modem drug therapy is more effective, however, and is often needed for patients with Barrett's esophagus. For the treatment of gastroesophageal reflux disease, with or without Barrett's esophagus, proton pump inhibitors (PPIs) are the most effective drugs. The two currently available PPIs in the United States are omeprazole (Prilosec) and lansoprazole (Prevacid). They are similar in efficacy. The usual initial dose is as follows: omeprazole, 20 mg once daily; lansoprazole, 30 mg once daily. In a few patients, a larger dose is needed to control symptoms or heal esophagitis, and sometimes a lower dose is sufficient for long-term maintenance. Histamine Hz receptor antagonists (HzRAs) include cimetidine (Tagamet), ranitidine (Zantac), famotidine (Pepcid), and nizatidine (Axid). HzRAs are less potent suppressors of gastric acid secretion than are PPIs. Many control trials have shown that H2RAs are effective treatment of reflux disease but that PPIs are more effective. Cisapride (Propulsid), a promotility drug, is also effective, but even the combination of this drug with an H2RA is not as effective as a PPI alone; the routine addition of cisapride to a PPI is unnecessary. PPIs are extremely effective in controlling reflux symptoms. Many patients find that their symptoms disappear when they are taking this type of medication; however, many report that when use of the drug is discontinued, symptoms recur within days. Thus, they often need to continue the PPI on a long-term basis. Contrary to earlier concerns, use of PPIs has not been associated with any increased risk of gastric carcinoid or carcinoma in patients treated for at least 7 years. Determining Helicobacter pylori status or serum gastrin levels before or during PPI therapy is unnecessary. One drawback to long-term PPI treatment is the cost. At the time of the writing of this report, the retail cost of omeprazole or lansoprazole taken once daily for a year is $1,000 to $1,500. A less effective alternative that sometimes may be adequate is a nonprescription HzRA; the cost is about half that of a PPI for a standard dose. Of note, treatment with PPIs controls symptoms and esophagitis but does not result in regression of Barrett's esophagus. In trials of treatment administered for a few years, regenerative patches (islands) of squamous epithelium have often been detected on endoscopy, but the overall length of columnar epithelium has not changed. There-

fore, the risk of cancer is not eliminated, and the need for continued surveillance is unchanged. If a patient with no symptoms or esophagitis is found incidentally to have Barrett's esophagus, prescribing a PPI or other medication is unnecessary. For some patients with severe reflux disease, with or without Barrett's esophagus, an operation is an alternative to long-term use of medication. Usually, a laparoscopic fundoplication of the Nissen or Toupet type is performed. The laparoscopic approach has substantially reduced postoperative discomfort in comparison with open thoracotomy or laparotomy; the hospital stay is now only 1 to 2 days. Excellent results are obtained in 80% or more of cases. Again, control of reflux by an operation rarely results in regression of Barrett's esophagus, and the subsequent risk of cancer and need for surveillance are not reduced. ADENOCARCINOMA The normal esophagus is lined with squamous epithelium. Approximately 40 or 50 years ago, squamous cell cancers constituted the majority of esophageal malignant tumors. During that era, only about 2 to 4% of esophageal cancers were adenocarcinomas. Since about 1970, a striking and continuing increase in adenocarcinoma has occurred; adenocarcinomas now account for more than 50% of esophageal cancers. In Olmsted County, Minnesota, we showed that the incidence of adenocarcinoma of the esophagus and of the esophagogastric junction has increased fivefold to sixfold since about 1970. 1 Most adenocarcinomas of the esophagus and many adenocarcinomas of the junction arise in Barrett's esophagus. A previous history of gastroesophageal reflux disease is a risk factor for adenocarcinoma, although reflux is so common in the general population that the risk to the normal person with reflux symptoms is minimal, unless Barrett's esophagus is present. The risk of adenocarcinoma is greater in smokers; however, no satisfactory explanation exists for the substantial increase in this disease in recent decades. Patients with Barrett's esophagus have a 30- to 125-fold increased risk of the development of esophageal cancer in comparison with the general population. The individual risk is about 1 cancer per 150 person-years. The disease is most common in white male persons. Currently, more than 90% of patients with adenocarcinoma are diagnosed after a brief history (weeks or 1 or 2 months) of tumor-related symptoms, such as progressive dysphagia, weight loss, and chest pain. At this stage, with a mass in the esophagus often in association with lymph node and sometimes distant metastatic disease, many patients are not candidates for surgical resection. For those in 459 whom resection is possible, the chance of surviving for 5 years is only 15 to 20%. In a patient with a tumor mass, endoscopy that shows the tumor often also demonstrates a previously unsuspected Barrett's esophagus. In such patients, Barrett's esophagus has almost certainly been present for years, and the cancer is a late complication. When adenocarcinoma is detected at an early, usually presymptomatic stage in patients with Barrett's esophagus, the chance of surgical cure is substantially higher-50 to 80%. Adenocarcinoma in Barrett's esophagus is believed to develop in a sequence of changes, from nondysplastic (metaplastic) columnar epithelium, through low-grade and then high-grade dysplasia, and finally invasive cancer. In high-grade dysplasia, the cells appear malignant histologically but have not yet invaded deeper layers of the esophageal wall. The cell nuclei are enlarged, hyperchromatic, and variable in size and shape. The nuclei are no longer in an orderly arrangement at the bases of the cells but are stacked above one another. Cell cytoplasm is reduced. The glandular architecture may be disorganized. SCREENING Endoscopy is usually indicated in patients with reflux who have dysphagia, chest pain, weight loss, or gastrointestinal bleeding. In light of the high prevalence of reflux symptoms and the relatively low incidence of adenocarcinoma, it is not reasonable to attempt to perform endoscopy in every person with heartburn in order to search for Barrett's esophagus in the hope that later cancer deaths can be prevented. One suggestion is to perform screening endoscopy in patients who have had reflux symptoms for 5 years or more and who have not previously undergone endoscopy. SURVEILLANCE Patients with Barrett's esophagus are usually advised to undergo periodic surveillance by endoscopy and biopsy for the detection of carcinoma or high-grade dysplasia at a stage when treatment has an excellent chance of preventing later cancer death." At endoscopy, our standard practice is to obtain biopsy specimens from any suspicious area such as a nodule or stricture and also to obtain 4-quadrant biopsy specimens every 2 em in the columnar-lined segment. Surveillance is not recommended in elderly patients or in those who have other medical problems that would cause them to be poor candidates for a major surgical procedure such as esophagectomy. No prospective controlled study has been done to prove that surveillance increases life expectancy.' The investigators of a decision-analysis study" concluded that annual surveillance might actually decrease life expectancy. Surveillance every 2 years improved survival

460 but was expensi ve for the gain expected; surveillance every 5 years was as cost-effective as some other accepted screening programs. DYSPLASIA High-grade dysplasia is rare, but low-grade dysplasia is common. In 270 consecutive patients with Barrett's esophagus, we found high-grade dysplasia in 5%, lowgrade dysplasia in 34%, and no dysplasia in 44%; in 17%, findings were indefinite. Interpretation of the grade of dysplasia is difficult, and even experts may disagree. If high-grade dysplasia is suspected, confirmation by a second pathologist is recommended. Different grades of dysplasia often occur in different areas of the same patient's esophagus. The endoscopist should obtain systematic biopsy specimens from the entire columnar-lined esophagus to find the highest (worst) grade of dysplasia present. For the patient with no dysplasia, we currently recommend follow-up examination at I year and after that every 2 to 3 years. For the patient with low-grade dysplasia, follow-up should be done yearly. DETECTION OF HIGH-GRADE DYSPLASIA The finding of high-grade dysplasia on biopsy is an ominous sign. High-grade dysplasia is not recognizable endoscopically. If nodularity or stricture is evident and a pathology report describes high-grade dysplasia, the clinician should be especially suspicious that a cancer is already present. In reported series, 10 to 52% of patients who underwent an operation after an endoscopic biopsy diagnosis of high-grade dysplasia already had invasive cancer in their esophagus; it was found by the pathologist after resection. In patients with no invasive cancer at the time of the diagnosis of high-grade dysplasia, two studies found that 19 to 26% had development of cancer in the next 2 to 7.5 years. Surgical resection of the esophagus with esophagogastric anastomosis is a major procedure; the mortality rate is about 3% in specialized centers, and complications are frequent. Resection is the preferred management of adenocarcinoma and is the most frequent approach in patients with high-grade dysplasia. As an alternative to surgical resection for dysplasia, usually highgrade dysplasia, interest in mucosal ablation techniques that preserve the esophagus is substantial. At endoscopy, the esophageal columnar epithelium is removed by using laser coagulation, electrocautery, heater probe, an argon beam coagulator, or laser in combination with a photosensitizing drug such as hematoporphyrin derivative. The lastmentioned approach is the preferred modality at our institution.' A PPI is administered to prevent further acid reflux, and squamous epithelium replaces the denuded area. Much or all of the Barrett's epithelium can thus be removed; whether the risk of cancer is reduced has not yet been proved. SHORT BARREn'S ESOPHAGUS AND INTESTINAL METAPLASIA The incidence of adenocarcinoma of the esophagogastric junction is increasing at a rate similar to that of adenocarcinoma of the esophagus, whereas the incidence of adenocarcinoma of the distal stomach is decreasing. Adenocarcinoma of the esophagogastric junction may arise in short lengths (less than 3 ern) of intestinal metaplasia in the distal esophagus." The risk of cancer for patients with short lengths of Barrett's type of epithelium is unknown. Currently, investigators suggest that; in patients with definite and endoscopically visible short segments or tongues of intestinal metaplasia in the esophagus, follow-up should be the same as in those with long segments of Barrett's esophagus. Biopsy specimens obtained from just below a normally located squamocolumnar junction, with no endoscopic evidence of Barrett's esophagus, will show intestinal metaplasia in approximately 20% of patients, with or without reflux symptoms, who undergo endoscopy; in older patients, the percentage is higher.' A diagnosis of Barrett's esophagus should not be made in these patients. Clearly, we cannot follow up one-fifth of the older population with histologic intestinal metaplasia in order to detect a rare case of adenocarcinoma of the junction. If the endoscopist does not detect a definite Barrett's esophagus, obtaining biopsy specimens to search for intestinal metaplasia serves little purpose. POPULATION PERSPECTIVE The following estimates are based on several Mayoauthored articles on the population of Olmsted County, Minnesota: " total population, 100,000; adults, 70,000 ; those experiencing heartburn each week, 9,000; those (with or without heartburn) in whom intestinal metaplasia is detected at the esophagogastric junction, 7,000; and those with Barrett's esophagus (3 em or longer), in most of whom it is unrecognized, 400. Barrett's esophagus has been clinically diagnosed in 25 persons. Two persons develop an adenocarcinoma each year. These numbers show that the risk of an adenocarcinoma developing in a person with heartburn or intestinal metaplasia (no visible Barrett 's esophagus) is very small. The risk for such development is higher in patients with Barrett's esophagus, and thus surveillance in that group is justified. In the general population, Barrett's esophagus is seldom recognized; thus, our current efforts will have minimal effect on the death rate from adenocarcinoma for the population as a whole.

Mayo Clln Proc, May 1998, Vol 73 461 RECOMMENDATIONS The following approaches are recommended in the management of Barrett's esophagus. 1. For patients with a more than 5-year history of reflux symptoms, one endoscopic screening should be considered. 2. For the control of symptoms and esophagitis, PPls or fundoplication is effective, but Barrett's epithelium and the risk of cancer remain. 3. For patients with visible Barrett's esophagus and intestinal metaplasia but no dysplasia, examination at 1 year and then every 2 to 3 years is recommended. 4. Annual examination is recommended for those with low-grade dysplasia. 5. Surgical resection should be considered in those with high-grade dysplasia; if resection is not possible, endoscopic mucosal ablation is an option. 6. For patients in whom biopsy specimens show intestinal metaplasia but no visible Barrett's esophagus, surveillance is probably unnecessary. REFERENCES 1. Pera M, Cameron AJ, Trastek VF, Carpenter HA, Zinsmeister AR. Increasing incidence of adenocarcinoma of the esophagus and esopnagogastrtc junction. Gastroenterology 1993;104:510 513 2. Tytgat GNJ. Does endoscopic surveillance in esophageal columnar metaplasia (Barrett's esophagus) have any real value? Endoscopy 1995;27:19-26 3. Spechler SJ. Barrett's esophagus: should we brush off this ballooning problem? [editorial]. Gastroenterology 1997;112:2138-2142 4. Provenzale D, Kemp JA, Arora S, Wong JB. A guide for surveillance of patients with Barrett's esophagus. Am J Gastroenterol 1994;89:670 680 5. Laukka MA, Wang KK. Initial results using low-dose photodynamic therapy in the treatment of Barrett's esophagus. Gastrointest Endosc 1995;42:59-63 6. Cameron AJ, Lomboy CT, Pera M, Carpenter HA. Adenocarcinoma of the esophagogastnc junction and Barrett's esophagus. Gastroenterology 1995;109:1541-1546 7. Spechler SJ, Goyal RK. The columnar-lined esophagus, intestinal metaplasia, and Norman Barrett. Gastroenterology 1996;110:614 621 8. Cameron AJ. Epidemiology of columnar-lined esophagus and adenocarcinoma. Gastroenterol Clin North Am 1997 Sep;26:487 494 Questions About Barrett's Esophagus (See article, pages 457 to 461) 1. Which one of the following is correct regarding Barrett's esophagus? a. Medical treatment of reflux usually causes regression of Barrett's esophagus b. An antireflux operation usually causes regression of Barrett's esophagus c. If an untreated patient with Barrett's esophagus has no reflux symptoms, the Barrett's esophagus will probably regress d. Treatment of reflux does not eliminate the need for surveillance in patients with Barrett's esophagus e. Barrett's esophagus protects against development of esophageal cancer 2. Which one of the following demonstrates epithelium characteristically seen in Barrett's esophagus? a. Gastric cardia b. Gastric fundus c. Skin d. Intestine e. Gastric antrum 3. Which one of the following class of drugs is most effective in control of reflux symptoms? a. Histamine Hz receptor antagonists b. Anticholinergics c. Promotility agents d. Antacids e. Proton pump inhibitors 4. Which one of the following features of Barrett's esophagus has the highest risk for later development of adenocarcinoma? a. High-grade dysplasia b. Low-grade dysplasia c. Metaplasia d. Chronic inflammation e, Ulcer 5, Which one of the following is correct regarding adenocarcinoma of the esophagus? a. It accounts for 2 to 4% of esophageal cancers b. It usually arises in Barrett's esophagus c. Incidence is equal in both sexes d. It is preventable by appropriate follow-up of all patients with Barrett's esophagus e. It has an increased risk in patients taking Hz receptor drugs Correct answers: 1. d, 2. d, 3. e, 4. a, 5. b For personal use, Mass reproduce only with permission from Mayo Clinic Proceedings,