Management of Barrett's esophagus.

There have been major recent advances in the understanding of the pathogenesis and epidemiology of Barrett's esophagus and adenocarcinoma of the esophagus. The advent of potent acid suppression with proton pump inhibitors and safe, minimally invasive antireflux procedures has made alleviating symptoms and eliminating peptic complications achievable goals for the vast majority of patients. Endoscopic surveillance of Barrett's esophagus is considered the standard of care and is widely used in clinical practice. Neither medical nor surgical antireflux procedures, however, result in the regression of Barrett's esophagus in any consistent manner. Thermal and chemical endoscopic ablation techniques show promise in both the management of high grade dysplasia and the reversal of Barrett's esophagus, but these techniques are still of unproven benefit, and can be costly and risky. Therefore, prospective and controlled studies with long term follow-up are needed before incorporating ablative techniques into routine clinical practice. Management of high grade dysplasia remains controversial. Alternative management strategies include surveillance, resection or ablation, tailored to the individual patient and the available expertise. Targets for future research include defining appropriate surveillance intervals; finding biological markers that identify patients at higher risk of progressing to cancer; defining the cancer risk and the appropriate management of patients with short segment Barrett's esophagus; understanding the natural history of dysplasia and comparing alternatives for the management of high grade dysplasia; and studying whether surgical management can delay or prevent the progression to dysplasia and adenocarcinoma.

during the past two decades in the United States and the Western world (1)(2)(3).
The definition of Barrett's esophagus has evolved considerably over the years (4). Currently, the most widely accepted definition of Barrett's esophagus is a "change in the esophageal epithelium of any length that can be recognized by endoscopy and is confirmed to have intestinal metaplasia by biopsy" (5). This recognizes several key points. First, intestinal-type metaplasia needs to be present, because it is the lesion that is associated with adenocarcinoma and requires surveillance (6). Second, a major emphasis is placed on endoscopic recognition, allowing a distinction between Barrett's esophagus and intestinal metaplasia at the gastroesophageal junction. The latter condition is very common (7) and does not have a clear-cut association with gastroesophageal reflux (8,9), and its progression to dysplasia and adenocarcinoma is not well defined (9). Third, the length of the Barrett's segment is not crucial; even short segment Barrett's cases carry a substantial (though possibly lower) risk of progression toward dysplasia and adenocarcinoma (10)(11)(12).

EPIDEMIOLOGY OF BARRETT'S ESOPHAGUS
AND ADENOCARCINOMA Most epidemiological studies of Barrett's esophagus have focused on patients with a long segment of columnar epithelium. Studies on the prevalence of Barrett's esophagus have yielded variable results because of differences in design and referral bias, and because patients with Barrett's esophagus often have minimal symptoms (13). Although retrospective reviews demonstrated a prevalence of Barrett's esophagus of 10% to 20% of patients undergoing endoscopy for reflux symptoms, prospective data revealed that only 3% to 5% of patients with weekly or daily heartburn have Barrett's esophagus (14). The autopsy rate for Barrett's esophagus was 1% in Olmsted County, Minnesota, United States; this translates into an age-adjusted 'true' prevalence that is 17 times higher than the clinically diagnosed prevalence in the same county during the same period (15). These data infer that the vast majority of patients with Barrett's esophagus remain undiagnosed.
The prevalence of Barrett's esophagus and esophageal adenocarcinoma increases with age. The median age for developing Barrett's esophagus has been estimated at 40 years, while the mean age of diagnosis is 63 years (16). Both Barrett's esophagus and esophageal adenocarcinoma are predominantly male diseases, with a male to female ratio of approximately two to one and eight to one, respectively (14). Esophageal adenocarcinoma is more prevalent in white than in black Americans (1).
The vast majority of esophageal adenocarcinomas appear to arise from Barrett's esophagus. Reports are variable as to the incidence of adenocarcinoma in patients with Barrett's esophagus. A review of 18 available series estimated the incidence of cancer in patients with Barrett's esophagus to be one/100 years of patient follow-up (17). More recent studies with a longer duration of follow-up have found a cancer incidence of one/180 to 208 years of patient follow-up (18)(19)(20).

PATHOGENESIS OF BARRETT'S ESOPHAGUS
The controversy as to the etiology of Barrett's esophagus persisted until a landmark study demonstrated columnar regeneration in a dog model following severe mucosal injury from chronic gastroesophageal reflux (21). Further studies suggested that the population of metaplastic cells in Barrett's esophagus develop from multipotential cells present in the basal layer of the esophageal mucosa (22,23). The exact pathogenesis of Barrett's esophagus and esophageal adenocarcinoma, however, remains unknown. Substantial experimental and clinical evidence support the importance of gastroesophageal reflux in causing esophageal mucosal injury and predisposing the development of Barrett's esophagus (21,24,25). Barrett's esophagus represents the severe end on the spectrum of reflux disease. However, studies have shown that the degree of acid reflux and esophageal motor abnormalities is similar in patients with Barrett's esophagus and patients with severe esophagitis without Barrett's esophagus (26). This suggests that additional factors are important in the development of Barrett's esophagus.
Reflux of duodenal contents, particularly bile acids, may contribute to the development of Barrett's esophagus. It is well recognized that bile, particularly in conjunction with acid and pepsin, can cause disruption of the mucosal barrier in the esophageal epithelium and accentuate the degree of esophagitis (27,28). The majority of patients with Barrett's esophagus have increased exposure to both acid reflux and reflux of bile acids (29)(30)(31). Furthermore, the severity of bile reflux was found to parallel acid reflux, rising in a graded manner from low levels in controls, to higher levels in patients with mild esophagitis, to even higher in patients with severe esophagitis and Barrett's esophagus (32). Laboratory studies have shown that bile acids can become concentrated in the esophageal mucosa, up to 14 times the luminal concentration while in the presence of acid. This process is driven by the physiochemical forces related to the presence of a weak acid (bile) at an interface between an acid (luminal) and a neutral (cellular) interface (33). The role of bile reflux in the absence of excessive acid reflux is unclear. Although animal experimental models have suggested that isolated bile reflux may increase the risk of adenocarcinoma (34), studies of symptomatic postgastrectomy patients have shown that bile reflux does not cause endoscopic esophagitis unless there is concomitant acid reflux (35). Both medical therapy for reflux disease with potent antisecretory agents (36,37) and surgical therapy for reflux (38) can lower the amount of bile reflux.
Genetic factors may play a role in the development of Barrett's esophagus. It has been recently reported that relatives of patients with Barrett's esophagus and esophageal adenocarcinoma have a greater likelihood than controls of having gastroesophageal reflux disease (GERD) or Barrett's esophagus (39,40).
Newer insight into the pathogenesis of Barrett's esophagus and esophageal adenocarcinoma may arise from the study of growth factors. The most studied growth factor system is that of the epidermal growth factor (EGF) (41). EGF, the related transforming growth factor-alpha and their receptor are all overexpressed in patients with Barrett's esophagus, and particularly in patients with dysplasia and adenocarcinoma (42,43). The levels of EGF and transforming growth factor-alpha in the gastric juice are three times higher in patients with Barrett's esophagus compared with patients with nonulcer dyspepsia (44). It is not yet known whether this increased exposure to EGF causes or is merely associated with Barrett's esophagus.
Another breakthrough came from the recent discovery of a transcription factor that is specific for the intestinal-type epithelium CDX-1. Despite its absence of expression in normal gastric and esophageal epithelium, CDX-1 is excessively transcribed in intestinal metaplasia of the esophagus and stomach (45). Further understanding of the biological activity of CDX-1 may provide key information on the pathogenesis of Barrett's esophagus.
Another recent area of interest is the association of Barrett's esophagus with the impairment of the mechanism for bioactive compound detoxification. Barrett's epithelium has been shown to have a significantly lower glutathione content and lower glutathione-S-transferase (GST) enzyme activity compared with normal esophageal epithelium (46). The same research group recently studied polymorphism in genes for enzymes controlling drug metabolism and detoxification (including the GSTP1 gene). Patients with Barrett's esophagus and esophageal adenocarcinoma were more likely to express the polymorphic variant GSTP1b (47) than healthy blood donors. The occurrence of GSTP1b resulted in significantly lower GST enzyme activity.

PROGRESSION FROM METAPLASIA
TO ADENOCARCINOMA The progression from Barrett's esophagus to esophageal adenocarcinoma is also poorly understood. It is well accepted that esophageal adenocarcinoma is the result of progressive cellular and biological changes from nondysplastic Barrett's esophagus, to advancing degrees of dysplasia, and finally to invasive adenocarcinoma. The biological mechanisms underlying this progression are the subject of intense research efforts. Although several biomarkers, such as proliferation indexes, DNA-ploidy and p53 abnormalities, are being investigated (48)(49)(50), very limited data are currently available to make any recommendations as to their potential clinical management utility. It is hoped that a prospective tissue bank for Barrett's esophagus (51) and the subsequent testing of biomarkers will help to identify those markers useful in the management of Barrett's.

MANAGEMENT OF BARRETT'S ESOPHAGUS
There are several goals involved in the management of Barrett's esophagus. These include: first, control of reflux symptoms; second, healing of esophagitis; and third, surveillance for early detection of dysplasia and cancer (5). Goals that remain elusive include stopping the progression of Barrett's esophagus and promoting the reversal of Barrett's esophagus and dysplasia. The tools available to contribute to achieving these goals are endoscopic surveillance, medical therapy, surgical therapy and endoscopic ablative techniques. It should be remembered that reversal of Barrett's esophagus must be complete to be meaningful, because the ultimate goal is the prevention of adenocarcinoma, and dysplasia can progress to adenocarcinoma even with partial regression of the Barrett's mucosa (52). Recent preliminary studies have shown that full reversal to squamous mucosa resulted in normalization of both the cell cycle and the proliferation abnormalities associated with Barrett's esophagus. However, when only partial regression of the columnar epithelium was achieved, these cell cycle and proliferation abnormalities persisted (53,54). It has also been reported that residual Barrett's mucosa following 'ablative' therapy progresses to dysplasia in one-third of the cases over an average of two years (55). Surveillance: The rationale for surveillance is that patients with Barrett's esophagus are at an increased risk of developing adenocarcinoma. Adenocarcinoma of the esophagus has a dismal prognosis when diagnosis is delayed until patients develop symptoms such as dysphagia. Preliminary reports suggest that those adenocarcinomas detected during surveillance for Barrett's are at an earlier stage, affording patients better survival (56,57). Because the goal of surveillance is to detect the presence of dysplasia, multiple, systematic (four-quadrant) biopsies of the metaplastic segment should be obtained. This is best done after healing of erosive esophagitis to minimize the effect of active inflammation, because inflammation can cause histological atypia, which can be misinterpreted as dysplasia. In addition to the biopsies obtained from normal appearing Barrett's epithelium, any mucosal abnormality should be extensively biopsied. Newer techniques such as mucosal staining with methylene blue may increase the detection of dysplasia and, therefore, make surveillance more efficient (58). The appropriate intervals for endoscopic surveillance are still empirical because they have not been rigorously tested (Table 1). Recent practice guidelines sponsored by the American College of Gastroenterology recommend follow-up endoscopy every two to three years after two yearly endoscopies fail to  show dysplasia. If low grade dysplasia is diagnosed, endoscopic surveillance is recommended every six months for one year and then yearly if there is no progression (5). Although the value of surveillance in patients with Barrett's has been criticized (18), endoscopic surveillance has been adopted in clinical practice (59). The follow-up of patients with short segment Barrett's (ie, less than 3 cm) is controversial. The surveillance intervals are unclear, and firm recommendations await additional information on the cancer risk in patients with short segment Barrett's (5). The management and follow-up of patients with high grade dysplasia (see below) are even more controversial because there are no controlled studies comparing the different options. Medical therapy: Medical therapy has focused primarily on the control of GERD symptoms and the healing of esophagitis. Because Barrett's esophagus appears to be a severe form of chronic GERD, aggressive management of GERD is appropriate. Lifestyle changes are extremely important in this patient population. These include the avoidance of large and late meals; weight management; avoidance of tobacco, alcohol and coffee; and elevation of the head of the bed. Abstinence from alcohol and tobacco is particularly important because both may be associated with a higher incidence of adenocarcinoma (60). Treatment with H 2 receptor antagonists can improve reflux symptoms and heal the majority of erosive esophagitis and ulceration cases (61). More recently, proton pump inhibitors were shown to be superior to H 2 receptor antagonists in the treatment and maintenance of patients with Barrett's-associated ulcers and severe esophagitis (62,63). Patients with Barrett's esophagus have, as a group, greater esophageal exposure to reflux than usual GERD patients (64) and can require higher doses of potent acidlowering medications for the control of acid reflux (65). Even though medical therapy can control GERD symptoms and heal esophagitis, it does not appear that a significant regression of Barrett's can be achieved with either H 2 receptor antagonists or proton pump inhibitors (66)(67)(68). Surgical antireflux therapy: Using antireflux therapy for GERD has increased considerably in the past five years, after the introduction of laparoscopic techniques. Surgical therapy is particularly appealing for patients with Barrett's esophagus because many of them require long term treatment with potent medical therapy. Surgical therapy has a theoretical advantage in that it can control reflux of all forms of gastric contents into the esophagus; for example, surgical therapy has been shown to be superior to medical therapy in controlling duodenogastroesophageal reflux (69,70). Antireflux procedures have been shown, in most studies, to offer very good symptomatic relief in patients with Barrett's mucosa and to heal erosive esophagitis (71). Although initial reports suggested that antireflux procedures could cause regression of Barrett's esophagus (72), recent studies have failed to confirm this finding (73)(74)(75). In most series, the regression of the Barrett's changes was incomplete. Measuring the regression of Barrett's may be inaccurate, because antireflux operations typically bring the esophagogastric junction into the abdomen, thus shifting the squamocolumnar junc-tion distally. The effect of antireflux procedures on the fate of dysplasia has also been variable, with studies showing progression of dysplasia in some patients and regression or apparent prevention in others (20,76). There remains interest in the possibility that antireflux surgery can prevent or retard the development of adenocarcinoma in patients with Barrett's esophagus. Nonetheless, it is well documented that adenocarcinoma can develop after antireflux procedures (72,73,(77)(78)(79)(80), and it is recommended that patients with Barrett's remain enrolled in endoscopic surveillance after surgical management. Investigators from the Mayo Clinic have reported long term follow-up on patients with Barrett's esophagus who underwent antireflux surgery. Although three patients developed adenocarcinoma, all cancers were diagnosed within the first three years following surgery. There were no additional cancers that developed despite a follow-up period of up to 18 years (mean 6.5 years) (79). This suggests that the patients in the Mayo series who developed cancer already had severe dysplasia or even carcinoma in situ before surgical intervention. In another report, adenocarcinomas developed years after antireflux surgery in patients with recurrent reflux after surgical failures (80). The authors believe, based on the available evidence, that the indications for antireflux surgery in patients with Barrett's esophagus should be similar to those for patients with severe GERD: surgical therapy is indicated for symptomatic relief in patients with documented reflux disease and poor response to medical therapy, as well as in young patients as an alternative to long term acid suppressive therapy. It is important to continue endoscopic and histological surveillance postoperatively for these patients. It is conceivable that the indications for surgical management may broaden as more information is obtained on the pathogenesis of Barrett's esophagus and if prospective studies demonstrate that surgical management affects cancer development. Such prospective clinical investigations are urgently needed. In addition, surgical antireflux procedures, which reduce esophageal acid exposure to normal, may be coupled with endoscopic ablative techniques as a novel approach to reversing Barrett's metaplasia (81).

Management of high grade dysplasia:
The optimal management of Barrett's esophagus patients with high grade dysplasia remains particularly controversial. Most surgical series suggest esophagectomy once high grade dysplasia is documented, citing a 40% to 50% incidence of finding invasive adenocarcinoma in the surgical specimens (82)(83)(84)(85)(86)(87). Surgical resection is generally curative for mucosal and submucosal (T 1 ) lesions. The mortality rate for esophagectomy by experts in Barrett's esophagus is acceptably low at 4% or less, although there is still a high morbidity rate (85). Late complications often involve anastomotic strictures, which can be managed with dilation (84). Resection also eliminates the subsequent need for endoscopic surveillance or medical acid suppression. Esophagectomy in patients with Barrett's and high grade dysplasia is arguably more appropriate than for the majority of patients with invasive cancer, for whom the prognosis remains dismal despite surgery. Either a transthoracic or transhiatal approach may be used. Transhiatal esophagectomy has achieved excellent results in this patient population and is the procedure of choice when resective surgery is required (86). Most of these studies, however, were retrospective in nature, did not provide a full description of the endoscopic findings and did not follow a strict preoperative endoscopic biopsy protocol. Some series included patients with dysphagia and endoscopic findings such as nodularity. Other investigators have challenged these findings by reporting that a strict endoscopic biopsy protocol can successfully differentiate between adenocarcinoma and high grade dysplasia (88). Investigators have reported that patients defined as having high grade dysplasia by this protocol can undergo endoscopic surveillance and have a lower rate of cancer (88)(89)(90). The American College of Gastroenterology practice guidelines recommend expert pathological confirmation for high grade dysplasia and give two management alternatives: surgical resection or endoscopy every three months (with surgery reserved for frank malignancy) (5). The recent introduction of endoscopic ablation techniques further complicates the management of high grade dysplasia.
The authors suggest that the decision to proceed with surgical resection for high grade dysplasia should consider the patient's physiological condition, and the expertise of available pathological, surgical and gastroenterological care. Patients with high grade dysplasia detected in a nodular lesion should undergo resective surgery, younger patients with minimal comorbidity should probably undergo resective surgery, and older patients with comorbid illnesses may be enrolled in strict surveillance or endoscopic ablation as part of an ongoing research trial. Endoscopic ablation: The rationale for endoscopic ablation is relatively straightforward: the goal is to ablate the columnar mucosa and allow it to heal as normal squamous mucosa. GERD control, either by medical or surgical therapy, is necessary as an adjunct to ablation. Endoscopic ablation has been used in the esophagus to reverse intestinal metaplasia, eliminate dysplasia, cure early cancer and palliate advanced cancer. It is important to note that endoscopic ablation for the therapy of Barrett's esophagus is still considered experimental and should be used only as part of research protocol or in patients who have comorbidities judged to preclude surgery or who refuse surgery (91,92). A variety of techniques have been described, and these can be broadly divided into thermal, chemical or mechanical ablation. Thermal ablation can be accomplished with the use of a variety of lasers (neodymium:yttrium-aluminum-garnet laser, the potassiumtitanyl-phosphate laser, the argon laser), multipolar electrocoagulation, argon beam plasma coagulation and cryotherapy using a cryospray catheter. Chemical ablation has been performed with photodynamic therapy (PDT), using primarily porfimer sodium or 5-aminolevulonic acid. Mechanical ablation is performed with endoscopic mucosal resection (EMR). EMR can be achieved by a 'suck and cut' or 'lift and cut' method (93). EMR has been mostly used in the management of early esophageal cancer and has limited usefulness in the management of Barrett's esophagus and dysplasia unless high grade dysplasia is very focal in nature. Because of these limitations, the remainder of the discussion will focus on the thermal and chemical modalities.
The choice of an ablative technique depends on the characteristics of the technique (ie, depth of injury), the lesion to treat (ie, length of the Barrett's esophagus and presence of dysplasia), the costs and the available expertise (92). Table 2 summarizes the different characteristics of each method. PDT is, so far, the only method that allows for therapy of a large area in a single firing; all other methods can treat only a small area with each firing. PDT is, therefore, easier to use and requires fewer sessions in patients with longer segments of metaplasia, particularly with the advent of elongated cylindrical diffusers with centring balloons. Depth of injury is paramount in determining both the efficacy and side effect profile of each technique. Deep injury is most likely needed to eliminate high grade dysplasia or early cancer. It is, however, associated with a higher incidence of complications, particularly stricture formation. Because the mucosa is thicker in the presence of advanced dysplasia, PDT or laser therapy is probabably superior to argon beam plasm coagulation or multipolar electrocoagulation in patients with high grade dysplasia.
Interpretation of the results of endoscopic ablation is limited by the absence of well conducted studies. The available reports are limited by a lack of controls, small numbers of patients, variable and generally short durations of follow-up, poor characterizations of patient characteristics and the combination of patients with and without dysplasia. Because the goals of treatment differ in patients with and without high grade dysplasia, the results are summarized separately where possible. Tables 3 and 4 summarize the results of selected studies on ablative therapies in patients with and without high grade dysplasia. The largest published study with endoscopic ablation describes using PDT with porfimer sodium as the photosensitizer. Overholt   CONCLUSIONS Despite major advances in the understanding of Barrett's esophagus, its management remains difficult and many areas are still controversial. Advances in medical and surgical man-agement have made alleviating symptoms and eliminating peptic complications achievable goals for the vast majority of patients. Neither medical nor surgical antireflux therapy, however, has resulted in the regression of Barrett's esophagus or the prevention of adenocarcinoma in any consistent manner. The possibility that antireflux surgery retards the progression from Barrett's esophagus to adenocarcinoma needs to be tested in a well designed, clinical trial. Endoscopic ablation techniques show promise in both the management of high grade dysplasia and the reversal of Barrett's esophagus. Doctors should wait for the results of prospective and controlled studies with longer follow-up before incorporating ablative techniques into routine clinical practice. Surgical resection remains the standard against which modalities must be measured when high grade dysplasia or adenocarcinoma are present. Management of high grade dysplasia, as noted, requires the consideration of surveillance or resection tailored to the   individual patient and the available expertise. Future research is also needed to define better appropriate surveillance intervals, to find biological markers that identify patients at higher risk of progression toward cancer, and to define the cancer risk and the appropriate management of patients with short segment Barrett's esophagus.