Barrett’s Esophagus: Diagnosis, Surveillance, and Medical Management

Introduction

Barrett's esophagus (BE) is an acquired condition resulting from severe esophageal mucosal injury. It is unclear why some patients with gastroesophageal reflux disease (GERD) develop BE whereas others do not. The diagnosis of BE is established if the squamocolumnar junction is displaced more than 1 cm proximal to the esophagogastric junction, and intestinal metaplasia is detected by biopsy. BE would be of little importance if not for its well-recognized association with esophageal adenocarcinoma (EAC). The incidence of EAC continues to increase, and the 5-year survival rate for this cancer remains very poor. However, the overall disease burden of esophageal cancer and cancer risk for an individual patient with BE remains low.

Current strategies for improved survival in patients with EAC focus on cancer detection at an early and potentially curable stage. Early detection can be accomplished either by screening more patients for BE, or with endoscopic surveillance of patients with known BE. However, current screening and surveillance strategies are inherently expensive, inefficient, and of unproven benefit. New techniques to improve the efficiency of cancer surveillance continue to evolve and hold promise to change clinical practice in the future. Treatment options include aggressive acid suppression, antireflux surgery, and chemoprevention with endoscopic eradication therapy reserved for patients with dysplasia or early EAC.

Epidemiology

The incidence of BE has increased markedly since the 1970s. This increase was previously thought to be due to the increased use of diagnostic upper endoscopy combined with the change in the definition of BE to include shorter segments of columnar-lined epithelium. However, more recent data from a Northern Ireland population based study found that the incidence of BE has increased from 23.9/100,000 between 1993 and 1997 to 62.0/100,000 between 2002 and 2005. These findings have also been encountered in the Netherlands and Australia.

It is estimated that BE is found in approximately 5% to 15% of patients undergoing endoscopy for symptoms of GERD. Population-based studies suggest that the prevalence of BE is approximately 1.3% to 1.6%. Interestingly, most of these patients in the general population have short segments of BE, and approximately 45% have no reflux symptoms. The prevalence of BE in white men increases with age until a plateau is reached at approximately age 50.

Various risk factors have been identified for the presence of BE, including frequent and long-standing reflux episodes, erosive esophagitis, smoking, male gender, white race, older age, central obesity, absence of H. pylori infection and absence of statin use. Familial aggregation of BE is seen in less than 10% of patients with BE and is felt to be caused by an incompletely penetrant autosomal dominant genetic variant. Genome wide association studies have also identified several candidate susceptibility loci for BE.

Diagnosis

BE is currently best defined as a metaplastic change in the lining of the tubular esophagus extending more than 1 cm above the gastroesophageal junction with biopsy confirmation of intestinal metaplasia. Endoscopically, this metaplastic change is characterized by displacement of the squamocolumnar junction proximal to the esophagogastric junction defined by the proximal margin of gastric folds ( Fig. 25.1 ). At the time of endoscopy, landmarks should be carefully identified, including the diaphragmatic pinch, the esophagogastric junction as best defined by the proximal margin of the gastric folds seen on partial insufflation of the esophagus, and level of the squamocolumnar junction. It is commonly accepted that the proximal margin of the gastric folds is the most useful landmark for the junction of the stomach and the esophagus. However, the precise junction of the esophagus and the stomach may be difficult to determine endoscopically because of the presence of a hiatal hernia, the presence of inflammation, and the dynamic nature of the esophagogastric junction, all of which may make targeting of biopsy specimens problematic.

Endoscopists identify landmarks necessary for the diagnosis of the columnar-lined esophagus inconsistently (see Fig. 25.1 ) ; this leads to inconsistencies in the diagnosis of BE. The Prague classification was developed to standardize the description of BE. This classification scheme describes the circumferential extent (C value) and maximum extent (M value) of columnar mucosa above the proximal margin of the gastric folds (see Fig. 25.1 ). The Prague classification does not include columnar islands; however, reliability coefficients for both criteria for segments greater than 1 cm in length, as well as for other endoscopic landmarks, including the diaphragmatic hiatus and proximal margin of the gastric folds, are excellent. Recognition of less than 1 cm of columnar metaplasia even with this scoring system is still problematic, pointing out the difficulties in measuring such short segments. The Prague classification has been validated in clinical settings in North America, Europe, and Asia.

If the squamocolumnar junction is 1 cm or more above the level of the esophagogastric junction, as defined by the proximal margin of the gastric folds using partial insufflation, biopsy specimens should be obtained for confirmation of intestinal metaplasia. There is ongoing debate regarding the need for intestinal metaplasia for the diagnosis of BE. The professional societies of North America all require intestinal metaplasia for the diagnosis of BE, whereas the British Society of Gastroenterology does not require the presence of intestinal metaplasia for the diagnosis. Although one study suggested that nongoblet cell columnar metaplasia shows DNA content abnormalities indicative of neoplastic risk similar to neoplasia encountered in intestinal metaplasia, a large population based study from Northern Ireland found that the risk of EAC was higher in patients with intestinal metaplasia than in those without.

How many biopsy specimens are needed to detect intestinal metaplasia? Detection of intestinal metaplasia is related to several factors, including location of biopsies, length of columnar-lined segment, number of biopsy specimens obtained, male gender, and increasing age. Intestinal metaplasia is more commonly found in biopsy specimens obtained in the proximal portion of the columnar-lined esophagus where goblet cell density is also greater. Data suggests that the detection of intestinal metaplasia increases with increasing number of biopsy specimens per endoscopy; four biopsy specimens have a yield of 34.7%, whereas eight biopsy specimens have a yield of 94% for intestinal metaplasia. Taking more than eight biopsy specimens does not seem to enhance the yield of intestinal metaplasia. This has led the current iteration of the American College of Gastroenterology Barrett's Esophagus (ACG BE) guidelines to suggest obtaining a minimum of eight biopsies to maximize the yield of intestinal metaplasia.

It may be difficult to determine endoscopically where the esophagus ends and the stomach begins for the reasons outlined earlier. It is impossible to reliably distinguish columnar metaplasia of the distal esophagus from columnar metaplasia of the stomach. All current practice guidelines and quality metrics recommend that biopsy specimen of the normal squamocolumnar junction should not be routinely obtained in clinical practice if it is at the level of the esophagogastric junction. Several studies indicate that cancer risk of intestinal metaplasia of the normal gastroesophageal junction is substantially lower compared to those with BE and thus that routine biopsies should not be obtained.

Screening

Only 5% to 7% of patients newly diagnosed with EAC have undergone endoscopy and received a prior diagnosis of BE. As such, surveillance of patients with BE would be expected to have limited impact on the 95% of EAC patients who are not known to have BE. One potential strategy to decrease the mortality rate of EAC is to identify more patients at risk, namely those with BE. Population-based studies suggest that in patients with newly diagnosed EAC, a prior endoscopy and diagnosis of BE are associated with both early stage cancer and improved survival. Current professional society practice guidelines now all recommend considering screening for BE in selected patients at high risk for BE. The target population includes patients with chronic GERD symptoms and one or more risk factors including male gender, age over 50 years, Caucasian ethnicity, central obesity, and a positive family history of BE or EAC.

High-definition white light endoscopy (HD-WLE) with biopsy is still the only validated technique to diagnose BE. However, it has clear limitations as a screening tool, including cost, risk, complexity, and diagnostic inconsistencies. For example, a 2014 study found that only 68% of patients diagnosed with BE in the community had confirmed BE, whereas the remainder had the diagnosis of BE reversed due to the absence of a columnar lined esophagus, intestinal metaplasia, or both. If screening with endoscopy and biopsy were applied to the estimated 20% of the population with regular GERD symptoms, the cost implications would be staggering. Unsedated upper endoscopy using small-caliber instruments still has the potential to change the economics of endoscopic screening because this technique may decrease sedation-related complications and costs. Unsedated small-caliber endoscopy detects BE and dysplasia with sensitivity comparable to conventional endoscopy. Although both procedures are well tolerated by patients, a major hurdle for unsedated endoscopy is continued patient resistance to undergoing a test without sedation. Otherwise, there are still no validated alternative techniques to screen for BE that overcome the cost and risks associated with conventional upper endoscopy.

There is clearly a need to develop either a better profile of patients at high risk for BE or a far less expensive tool to provide mass population screening. A variety of risk prediction models have been studied to accomplish this by examining risk factors such as frequency and duration of GERD symptoms, age, smoking, waist to hip ratio, H. pylori status, and inflammatory cytokines. Models that include variables in addition to GERD frequency show encouraging performance characteristics but need further validation prior to application into clinical practice.

A variety of alternatives to both sedated and unsedated endoscopy are also now under investigation around the world. Much interest is directed at a nonendoscopic Cytosponge currently undergoing clinical trials. This Cytosponge capsule, administered by a nonphysician in an outpatient setting, had a sensitivity of 80% for BE and a specificity of 92% in a United Kingdom trial. Another nonendoscopic device under study is tethered endomicroscopy, but data to date are limited. Exciting new screening strategies based on the microbiome and exhaled volatile organic compounds are also under study.

After a normal initial upper endoscopy, some clinicians wonder if a repeat screening upper endoscopy should be undertaken in symptomatic GERD patients at a later date. Several studies have addressed this point with consistent results. In patients with nonerosive reflux disease at the index endoscopy, BE is rarely found if the repeat endoscopy is performed within 5 years. BE may be present in 9% to 12% of patients with erosive esophagitis at the time of index endoscopy, and higher grades of esophagitis are associated with a higher case finding rate of BE on repeat endoscopy. Screening for BE in GERD patients should take place only after initial therapy with a proton pump inhibitor (PPI). A negative endoscopy at baseline makes it highly unlikely to find BE if endoscopy is repeated.

There are still no data from randomized controlled trials or observational studies to evaluate the strategy of screening. A decision analysis model by Inadomi et al (2009) examined screening of 50-year-old white men with chronic GERD symptoms for BE and found that one-time screening is probably cost-effective if subsequent surveillance is limited to patients with dysplasia on initial examination. This strategy would result in a cost of $10,440 per quality-adjusted life year saved compared with a strategy of no screening or surveillance. Other modeling studies support screening in patients with chronic GERD symptoms as well, but only if the following conditions are met: patients at high risk for BE, high-grade dysplasia (HGD), or EAC; high sensitivity and specificity of endoscopy with biopsy; and little or no reduction in quality of life with esophagectomy. Any variation of these ideal conditions quickly made this strategy cost-ineffective. Problems inherent in showing the utility of a screening program, such as healthy volunteer bias, lead time bias, and length time bias, all need to be addressed.

Surveillance