Barrett's esophagus (BE) is an acquired condition in which the squamous lining of the distal esophagus is replaced by a specialized intestinal metaplastic columnar epithelium. Although in the past BE was defined as the presence of columnar-lined esophagus on endoscopy, it is now accepted that confirmation of intestinal metaplasia by histology is required to make the diagnosis. The American College of Gastroenterology summarizes the most widely accepted definition as "a change in the esophageal epithelium of any length that can be recognized at endoscopy and is confirmed to have intestinal metaplasia by biopsy of the tubular esophagus and excludes intestinal metaplasia of the cardia."¹ The importance of BE is its risk of transforming into esophageal adenocarcinoma.

The incidence of BE over the past 20 years has risen sharply, and with it, there has been a dramatic increase in the incidence of adenocarcinoma.² Reflux without BE can be treated medically (acid suppression) and surgically (fundoplication), with surgical therapy gaining increasing attention. Indeed, with the development of minimally invasive surgery, there has been an increased willingness to consider surgical options for reflux disease.³ From 1990 to 1997, the rate of antireflux surgery increased from 4.4 to 12.0 per 100,000 adults, and the proportion of operations performed laparoscopically increased from 25% to 76%.³ Similar trends have been noted in other countries as well.^4,5 In this context, we review the available evidence for both medical and surgical approaches to the management of BE and their respective roles in halting disease progression.

EPIDEMIOLOGY

The prevalence of BE in the population has been difficult to assess. Cameron and Lomboy at the Mayo Clinic found BE in 0.73% of 51,311 patients having endoscopy.⁶ In another endoscopy screening report, a prevalence rate of 1% was found in residents of Olmsted County, Minnesota.⁷ Still other studies have found much higher rates. Studies of patients undergoing screening sigmoidoscopy or colonoscopy who agreed to an upper endoscopy revealed BE in 10-25% of these patients.^8–10 Seven cases were found in 733 autopsies (approximately 1%). The adjusted prevalence rate in this autopsy series was 376 cases per 100,000 population, significantly higher than the clinically diagnosed prevalence of 22.6 per 100,000.^7,11 While many have suggested that the rate of BE has increased in the population, the incidence of BE has increased in parallel with the use of endoscopy.^12,13 Whether there is a true increase in the incidence of BE or the increase simply reflects the increased use of endoscopy is unclear. In contrast, the incidence of esophageal adenocarcinoma has increased 10-fold in the 25-year period 1974–1997. Many adenocarcinomas occur in patients without a previous diagnosis of BE, suggesting that many people with this condition remain undiagnosed.¹³ The rate of malignant transformation of BE itself remains low. The overall risk of developing adenocarcinoma in patients with BE ranges from 0.2% to 1.9% per year.¹⁴

PATHOGENESIS

Norman Barrett first described the condition that bears his name over half a century ago, believing it to be a short esophagus. We now know that duodenogastric reflux is a primary factor in metaplasia of the distal esophagus. Patients with Barrett's metaplasia are known to reflux considerably greater acid and bile into their esophagus than patients with esophagitis alone.^15,16 The abnormal acid exposure is primarily because of a structurally defective lower esophageal sphincter (LES), with defects in length, pressure, or both. Such sphincter defects can be demonstrated manometrically in more than 90% of patients with Barrett's metaplasia.¹⁷Additionally, esophageal motility in patients with gastroesophageal reflux disease (GERD) is often compromised, leading to delayed clearance of refluxate from the esophagus.¹⁸

Historical dogma has held that BS develops to its full extent with little subsequent increase in length.^6,19 This belief is based on data from patients with only long segments of Barrett's metaplasia within the tubular esophagus (i.e., >3 cm). Intestinal metaplasia actually occurs in short segments (<3 cm) and even can occur microscopically just below the squamous epithelium of an endoscopically normal-appearing gastroesophageal junction (GEJ).¹⁹ BS does seem to progress in stepwise fashion upward.^10,19 Only after intestinal metaplasia occurs in the esophagus is there an increased cancer risk. The finding of intestinal metaplasia thus is required for the diagnosis of BE. It is important to realize that intestinal metaplasia in the cardia is not pathologically equivalent to Barrett's metaplasia because the former does not carry the same risk of malignant transformation.²⁰ Many use 3 cm as the dividing line between short- and long-segment metaplasia, although this division is fairly arbitrary. While the literature reports trends of increased cancer risk with longer segments, there is no hard evidence that a true risk stratification occurs at a particular segment length.¹⁰

The composition of refluxate likely plays a role in the metaplastic sequence. Patients who reflux both gastric and duodenal contents have a higher prevalence of BE than do patients with reflux gastric acid alone.¹⁷ Analysis of the composition of refluxate in 281 patients with GERD demonstrated that the patients with the greatest degree of mucosal injury were more likely to have both duodenal and gastric reflux as opposed to pure gastric reflux.²¹ Moreover, the ability of duodenal refluxate to induce mucosal injury may be pH-dependent. Both ionized bile salts and the lipophilic, unionized bile acid are present in the 3–5 pH range. Unionized bile salts can cross membranes rapidly, entering the alkaline cytoplasm, where the ionized form becomes trapped within the cell, causing toxicity.^17,22 The combined importance of both types of reflux in the pathogenesis of BE is important when discussing therapeutic options.

The role of refluxate in the metaplastic process is mediated by a range of molecules, including cyclooxygenase-2, c-myc, and mitogen-activated protein kinase signaling molecules.²² Additionally, other molecular features implicated in the neoplastic sequence include increased expression and changes in pattern of expression of the proliferation marker (Mib-1) Ki-67 antigen; overexpression of p53 protein; overexpression of growth factors such as epidermal growth factor, c-erbB2, and transforming growth factor alpha, decreased and abnormal expression of the cell adhesion molecule E-cadherin, and in carcinomas, increased expression of serine proteases.^23,24 The role of cyclooxygenase has sparked interest in nonsteroidal anti-inflammatory agents as a possible modality of chemoprevention.²² The literature supports the possibility of metaplastic regression in early or short-segment Barrett's metaplasia, but with varying degrees after both medical therapy and surgical intervention.^{15,17,25–33}

TREATMENT OF BARRETT'S METAPLASIA

Ideally, there are four goals in the treatment of BE: (1) control of symptoms, (2) cessation of gastroduodenal reflux into the esophagus, (3) elimination of Barrett's metaplasia, and (4) decreasing the risk of adenocarcinoma. The tools available to achieve these goals include acid-suppression therapy, antireflux surgery, endoscopic ablative techniques, and surveillance. Discussion regarding how to best achieve these goals and whether these goals are indeed attainable generates considerable controversy.

Controlling Symptoms

Medical therapy is very effective for controlling symptoms in patients with uncomplicated reflux. However, standard doses of available medications are less likely to control symptoms in patients with BE because BE generally is associated with severe reflux complaints.^10,26 Few reports specifically address symptom relief with acid suppression in BE. Most series measure other outcomes, such as regression of metaplasia. In a series of small studies, between 20% and 30% of patients with BE continued to have heartburn symptoms while taking proton pump inhibitor (PPI) medication.^31,34,35 Better results were seen in a long-term randomized, prospective trial of surgical versus medical therapy, in which 91% of patients after a median follow-up of 5 years reported excellent to good symptom control, similar to the results with surgical therapy.³⁶ Symptom relief with acid-suppression therapy in BE is similar to that in patients with Los Angeles grades C and D esophagitis but worse compared with lesser grades.¹⁰

Interestingly, some patients with BE have relatively minor complaints when compared with patients who have similar amounts of reflux but no metaplasia. It is possible that the altered sensitivity of the metaplastic epithelium to refluxed acid may explain the poor predictive value of reflux symptoms for diagnosing BE.^17,37,38 For most patients with BE, medical therapy can provide good symptom relief, but many of these patients have incomplete relief and seek other treatment.

Surgical intervention, most recently in the form of laparoscopic antireflux surgery (LARS), is the alternative for controlling symptoms. In a literature review of 25 articles with 1021 patients from 1980 to 2003, there was a mean clinical success rate of 77% with long-term follow-up.³⁹ In one report, the results of antireflux therapy were so inferior to those with uncomplicated GERD that the authors stopped performing standard antireflux operations in patients with BE.⁴⁰ While most surgeons agree that these are often more difficult operations, this report appears to be the exception. At the University of Washington, we have experienced excellent results in patients with BE. In a series of 106 patients with a median follow-up of 4 years, 96% of patients had improvement in heartburn, 84% had improvement in regurgitation, and 81% had improvement in symptoms of dysphagia. Similarly good results were seen in extraesophageal symptoms²⁶ (Table 33-1). Studies also have demonstrated that LARS improves quality-of-life scores and returns patients to levels of global functioning similar to those of their age-matched norms^41,42 (Table 33-2 and Fig. 33-1).

From Oelschlager BK, Barreca M, Chang L, et al: Clinical and pathologic response of Barrett's esophagus to laparoscopic antireflux surgery. Ann Surg 238:460, 2003.

When directly compared with medical therapy, surgery is at least as good, if not better, at controlling symptoms of reflux. In a prospective but nonrandomized study reported on 45 patients undergoing either medical or surgical treatment of BE, symptoms of heartburn or dysphagia recurred in 88% of patients treated with medical therapy alone and in only 21% after antireflux surgery.⁴³ In a prospective, randomized trial of medical versus surgical management of patient's with BE, symptoms were controlled equally (over 90%) in both treatment arms.³⁶ The excellent results of surgery in most case series, which are largely composed of patients referred in instances of medical failure, highlight the efficacy of surgical therapy. Taken together, the literature suggests that surgery is equivalent to or better than medical treatment alone for management of the symptoms of patients with BE.

Controlling Reflux

Currently, the American College of Gastroenterology states that the goal of treatment of BE is symptom relief.¹ However, the elimination of symptoms in medically treated patients cannot be equated with eradication of reflux.⁴⁴ Katzka and Castell demonstrated that up to 80% of patients whose symptoms of reflux are under control with medical therapy continue to have abnormal acid exposure.⁴⁵ Yeh and colleagues showed that after controlling symptoms with esomeprazole, 62% of patients with BE had abnormal pH profiles, with loss of control particularly at night.⁴⁶ Other investigators also have demonstrated poor control of acid exposure in the esophagus.^31,47,48

The risk of incomplete control of acid reflux in BE is the potential of producing a metaplastic effect. Fitzgerald and colleagues, in an ex vivo study, showed that pulsed contamination with acid triggers cellular proliferation of Barrett mucosal cells, whereas continuous acid exposure blocks cellular proliferation.⁴⁹ Intermittent acid exposure, as occurs often at night in medically managed patients, may lead to cellular hyperproliferation and promote the metaplastic-carcinoma sequence.⁵⁰ Hence, whereas PPIs may control symptoms, the harmful effects of reflux on the esophageal mucosa are still present.

Duodenal reflux is another important consideration, given its synergistic role with acid in the promotion of BE.⁵¹ Data on the efficacy of medical therapy to control bile reflux are conflicting. It has been suggested that PPI therapy also may reduce duodenal reflux.^52–54 One study demonstrated a decrease in total esophagus-to-bile exposure from 28.9% to 2.4%.⁵² It is known that omeprazole 20 mg twice daily reduces the volume of gastric secretion by about 40%.⁵² One explanation for this effect may be that even though the volume of duodenal reflux remains the same, the decrease in gastric secretions caused by PPI therapy reduces the total volume of fluid in the esophagus, hence decreasing total esophageal exposure to bile.⁵³ Yet other authors have observed that medical therapy actually is rather poor at controlling bile reflux.^36,55 In one series of patients who underwent Bilitec (Medtronic, Inc., Minneapolis, MN) monitoring, 75% in the medical arm had pathologic levels of bile reflux.³⁶ Additionally, others have hypothesized that PPIs may paradoxically facilitate bile-induced injury while the patient remains asymptomatic.

PPIs create an alkaline environment and enable partial dissociation of bile salts into their unionized and lipophilic form. This occurs most readily at the critical pH values between 3 and 5. The lipophilic form of salts can pass through membranes of mucosal cells causing injury. At higher pH of 7, bile salts are ionized and remain harmlessly in solution. At very low gastric pH, bile salts precipitate and are of no significance.¹⁷ Patients on twice-daily omeprazole, however, spend more than 30% of the day at a pH of less than 4.⁵⁶ Strict acid suppression to keep pH levels above 7 would require significantly higher doses than would be practical. In the final analysis, incomplete therapy may prove to be worse than no therapy by facilitating the dissociation of bile salts into their unionized state, which is injurious to the mucosa.¹⁷

By correcting the anatomy of the cardia and creating a barrier to reflux, LARS provides control of both gastric and duodenal refluxate. Hofstetter and colleagues reported normal pH monitoring in 83% of patients after surgery.²⁸ In a prospective, randomized trial of medical versus surgical therapy in BE patients, 91% of the surgical arm had normal pH monitoring versus 75% in the medical arm. In patients who underwent Bilitec monitoring, 75% in the medical arm had pathologic levels of bile reflux in contrast to 8% in the surgical arm.³⁶

The University of Washington series demonstrated a mean preoperative DeMeester score of 100.6, which improved to a mean 17.1 postoperatively. Among patients who underwent postoperative pH monitoring, 26% had abnormal reflux. This failure rate may be explained in part by the fact that 64% of patients in this study had other complicating features, such as reoperation, paraesophageal hernia, ineffective motility preoperatively requiring a partial wrap, and peptic strictures.²⁶ These results also emphasize the difficulty of operating on BE patients, who often have severe associated anatomic and physiologic abnormalities. Still, the majority of our BE patients at the University of Washington experience long-term control of symptoms and reflux after LARS.²⁶ Indeed, many other authors have reported good outcomes for BE compared with uncomplicated GERD patients after LARS.^{28,50,57–59}

Regression of BE

The third goal of therapy is to halt and potentially reverse the development of intestinal metaplasia. Aggressive medical therapy has in a few cases been able to reduce the surface area of BE, although the amount of regression is very modest, and the results are not consistent.^{26,29–31,60} In a randomized, double-blinded study of omeprazole versus ranitidine, the omeprazole arm achieved a modest regression in the total area of BE by approximately 8% after 2 years treatment.³⁰ In one of the most rigorous longitudinal follow-up studies, 309 patients on medical therapy were followed for an average of 3.8 years with no evidence of complete BE regression.⁶¹ Aggressive medical therapy, while suppressing symptoms of reflux, seems to have little affect on the natural history of BE.

In contrast, there is compelling evidence for regression of BE after surgical therapy. In 1980, Brand and colleagues were the first to report regression of BE after surgery in 4 of 10 patients.²⁵ Since then, many authors have reported small series with modest rates of regression ranging between 10% and 47% after antireflux surgery.^{15,25–28,41,43,59,62} With the development of LARS, larger series of patients are demonstrating regression of metaplastic tissue. At the University of Washington, 106 BE patients with LARS were followed for a median of 40 months. Of these patients, 33% had complete regression of intestinal metaplasia. All patients with complete regression had short segments of BE (<3 cm) before the operation. None of the patients with long-segment BE preoperatively had complete regression²⁶ (Fig. 33-2). Bowers and colleagues, in a series of 104 patients, similarly noted regression of short-segment BE in 59% of patients compared with 30% in patients with long-segment disease.⁴¹ The series of DeMeester and colleagues series showed a similar trend.⁶³

The complete regression in patients with short-segment disease is not surprising given that this may represent early disease and certainly less disease burden. Additionally, there are also reports of reversal of dysplasia. In a prospective, randomized study, Ortiz and colleagues showed that dysplasia developed in 22% of patients on medical therapy and in only 2.3% of patients after reflux surgery with a median follow-up of 5 years.⁶⁴Hofstetter and colleagues observed in their series that low-grade dysplasia regressed to nondysplastic BE in 7 of 16 patients (44%), and intestinal metaplasia regressed to cardiac mucosa in 9 of 63 patients (14%).²⁸ Despite the lack of randomized, controlled trials, these results are significantly better than the results demonstrated with medical therapy alone. These studies suggest that earlier referrals to surgery for patients with BE may yield the best chance for cure of the disease (see Table 33-2).

Progression to Cancer

There is much debate in the literature over whether any therapy can decrease the incidence of adenocarcinoma in patients with BE. Certainly one might expect that if BE could be reversed by surgery, the risk of malignant degeneration also might be decreased. Multiple authors have found data to support this view. Katz and colleagues, for example, found that 9-year dysplasia- and cancer-free survival was 100% in patients treated surgically versus only 50% in those receiving medical therapy.⁶⁵ One meta-analysis that reviewed 4121.7 patient-years of follow-up after antireflux surgery found the yearly incidence of adenocarcinoma to be 0.3%, lower than the study's estimated incidence of patients treated medically.⁵⁰ However, other studies have failed to show a clear difference.¹⁴ A meta-analysis comparing 4678 patient-years of follow-up in the surgical group versus 4906 patient-years in the medical group found the incidence of cancer in the surgical group to be 3.8 per 1000 patient-years compared with 5.3 in the medical group. Although the findings were lower in the surgical group, the difference was not statistically significant (p = 0.29)⁶⁶ (see Table 33-2).

The main difficulty in determining which is the more effective treatment lies in the very low incidence of cancer progression in patients with BE. It is estimated that up to 6 years may be required for adenocarcinoma to develop from BE. Cases of adenocarcinoma diagnosed within a few years of surgery may represent patients who already had neoplastic transformation at or before the time of surgery. Hence the development of adenocarcinoma during the first few years after surgery may represent unavoidable progression of disease.⁵⁰ Indeed, in multiple follow-up studies, most cancers developed in the first 5 years postoperatively.⁵⁰ Even though some studies have failed to demonstrate statistically significant trends, they do consistently show a tendency for surgery to be protective.⁵⁰ The sample size needed to create a study of sufficient power to achieve statistical significance is hampered by the low rate of neoplastic transformation. Some have argued that even if antireflux surgery could prevent esophageal adenocarcinoma, its use solely for this purpose could not be sanctioned because the surgical mortality rate likely exceeds the annual incidence of cancer.¹⁴ However, the rate of cancer transformation ranges from 0.2% to 1.9% per year with a higher cumulative lifetime risk, whereas the mortality rate associated with laparoscopic surgery as a single event ranges from 0.07% to 0.5%.^14,67,68 Taking into account the excellent control of symptoms provided by surgery, a compelling argument can be made that antireflux surgery for early-stage disease can restore the antireflux barrier and has a favorable effect on the natural history of BE.⁵⁰

COMPLICATIONS OF LARS

It is important to consider potential complications from surgery when discussing the surgical management of a patient with BE. Pessaux and colleagues, in a retrospective study of 1470 patients who underwent laparoscopic fundoplication, found an intraoperative complication rate of 2.1%. Conversion to an open procedure occurred in 6.5% of patients reviewed, most often because of poor exposure and bleeding. Postoperative complications were seen in 2.9% of patients during the first month after surgery. There was one death in the series (0.07%) in a patient who developed necrosis of the wrap⁶⁷ (Table 33-3).

From Pessaux P, Arnaud JP, Ghavami B, et al: Morbidity of laparoscopic fundoplication for gastroesophageal reflux: A retrospective study about 1470 patients. Hepatogastroenterology 49:448, 2002.

Rantanen and colleagues in Finland reviewed 5502 operations with specific attention to life-threatening complications in both open and laparoscopic cases. There were 9 fatal and 13 life-threatening complications after 3993 open cases (mortality 0.2%) and 1 fatal and 14 life-threatening complications after 1162 laparoscopic cases (morality 0.1%). There was no statistical difference in mortality rate between the two approaches; however, the rate of life-threatening complications was higher in the open group.⁶⁸

Flum and colleagues, evaluating data from the Nationwide Inpatient Sample of the Health Care Utilization Project and the Washington State Comprehensive Hospital Abstract Reporting System, found a higher mortality rate of 0.8%. They also noted in Washington state that there was a wide variation in the experience of surgeons performing LARS and a direct relationship between case order and adverse outcome, with better outcomes associated with improved experience. They found that the risk of morbidity and mortality dropped significantly after 15 cases.⁶⁹

In summary, the rates of complications are low but not negligible, and the rates of complications are correlated with surgeon experience.

ALTERNATIVE APPROACHES IN OBESE PATIENTS AND IN PATIENTS WITH LONG-SEGMENT BE

Earlier we focused our attention on classic antireflux surgery involving fundoplication as the principal operative technique to create a reflux barrier. Some patients with BE, such as those who are obese or perhaps those who have long-segment disease, require special consideration because these conditions adversely affect the outcome of laparoscopic fundoplication.^39,70,71 Obesity is a risk factor for GERD, and with its growing prevalence, obesity has become a significant factor in managing patients with GERD.⁷⁰ Surgeons often refrain from surgery in these patients because they are technically difficult and the outcomes are inferior.^70,71 An excellent alternative to a fundoplication in obese patients is the Roux-en-Y gastric bypass.^70–73 First, patients benefit from the numerous effects of weight loss. Second, it provides excellent relief of GERD symptoms.^70–73 Third, it may positively affect the natural history of BE. Others have advocated the diversion of biliopancreatic secretions away from the esophagus in patients with intractable reflux and BE.^39,74 At the University of Washington, we have had a patient with BE regression after a Roux-en-Y gastric bypass.

Because some have experienced high recurrence rates and failure to control reflux with classic antireflux surgery, especially in patients with long-segment BE, alternative approaches have been considered.³⁹ Csendes has argued that an acid-suppression–duodenal-diversion procedure (e.g., a duodenal switch in which an antireflux procedure is performed together with highly selective vagotomy), coupled with a Roux-en-Y loop with or without a partial gastrectomy, can durably control symptoms, eliminate reflux, and decrease the risk of cancer.^39,75,76 More investigation is needed before advocating more aggressive surgery to prevent cancer. If such surgery were applied widely, the associated mortality might outweigh the lifetime risk of neoplastic transformation. If one considers the 1.9% 30-day mortality rate for gastric bypass surgery in obese patients, it is hard to justify its use solely to prevent cancer.⁷⁷ However, one still might be justified in employing these larger surgeries when managing debilitating symptoms not well addressed by standard fundoplication procedures, in addition to improving morbidity and mortality from obesity-related complications.

ENDOSCOPIC SURVEILLANCE AFTER ANTIREFLUX SURGERY IN PATIENTS WITH BE

The consensus view of the American Gastroenterological Association holds that patients with BE should have the same endoscopic surveillance after surgery as medically treated patients.¹⁰ While antireflux surgery is often associated with regression of metaplasia, a variable proportion of BE patients (1–5%) can progress to high-grade dysplasia and carcinoma.^{10,14,17,40,78–80} The ultimate goal of endoscopic surveillance is to reduce mortality related to cancer, but long-term prospective studies looking at efficacy are lacking. A few uncontrolled retrospective studies suggest that survival in patients with esophageal cancer is better in those who had endoscopic surveillance.^81–85 However, the outcomes of these studies clearly may have been influenced by selection bias, lead-time bias, and length bias.¹⁰ Notwithstanding the inherent limitations of available data, earlier detection of high-grade dysplasia and carcinoma should have a positive effect on survival, as in other cancers. Thus, regardless of the surgical outcome, BE patients require endoscopic surveillance postperatively.^10,50

SUMMARY

Surgery, compared with medical therapy, more consistently achieves the four ideal objectives of treating BE: (1) controlling symptoms, (2) cessation of reflux of gastroduodenal contents into the esophagus, (3) elimination of Barrett's metaplasia, and (4) decreasing the risk of progression to adenocarcinoma. Laparoscopic antireflux procedures are very safe to perform and are associated with few complications and low morbidity and mortality. While surgical therapy may not be suitable for all patients, it should be considered and discussed with all patients who have BE.

EDITOR'S COMMENT

One reason for the higher failure rates of LARS in patients with BE is that these patients are more likely to have foreshortened esophagus. I find that I am more likely to include a Collis extension in these patients. Also, one should note that performing LARS does not prevent or unduly complicate a future esophagectomy should it be required.

–RB

REFERENCES