Controversies in the Definition of Barrett Esophagus

Current Definitions of Barrett Esophagus

The definition of BE according to the current American Gastroenterological Association (AGA) medical position statement on the management of BE is “the condition in which any extent of metaplastic columnar epithelium that predisposes to cancer development replaces the stratified squamous epithelium that normally lines the distal esophagus.” The manuscript goes on to state, “Intestinal metaplasia is required for the diagnosis of Barrett's esophagus because intestinal metaplasia is the only type of esophageal columnar epithelium that clearly predisposes to malignancy.” Therefore the diagnosis of BE requires a combination of endoscopic and histologic findings, in that the columnar metaplasia must involve the tubular esophagus on endoscopy and must contain goblet cells, which define the presence of intestinal metaplasia (IM), on biopsy specimens.

Agreement does not exist worldwide on the requirement for goblet cells to diagnose BE. The British Society of Gastroenterology mandates only histologic proof of columnar mucosa on biopsies taken from the tubular esophagus; goblet cells do not need to be documented. In Japan the presence of any endoscopically detectable columnar lining of the distal esophagus is adequate to prove the diagnosis of BE; biopsy confirmation is not mandated. The main controversies underlying these differing definitions are the potential for development of EAC in the setting of a columnar-lined esophagus (CLE) with or without the presence of goblet cells, as well as in IM developing distal to the tubular esophagus, so-called cardia intestinal metaplasia (CIM).

Historical Perspective

Fundamental to the definition of BE is the definition of the esophagus. In 1950 Norman Barrett ( Fig. 31.1 ) determined the distal boundary of the esophagus to be the squamocolumnar junction (SCJ), believing that an organ should be defined by its mucosa. Because of this interpretation, a columnar-lined tubular structure subjacent to the esophagus was not recognized as CLE but rather was thought to represent an intrathoracic tubular stomach caused by a congenitally foreshortened esophagus ( Fig. 31.2A ). An irony is that Barrett initially misdiagnosed the condition later bearing his name!

In 1953, based on the recognition that columnar-lined intrathoracic tubular structures as described by Barrett lacked a peritoneal lining (characteristic of the stomach) and contained esophageal submucosal glands, Allison and Johnstone concluded that the tubular structures were of esophageal, as opposed to gastric, origin. Barrett later agreed with Allison and Johnstone, and introduced the term columnar-lined esophagus in 1957 (see Fig. 31.2B ). Over the next 6 decades, the definition of BE evolved as the genesis of CLE, and its relationship to EAC, became better understood. In the early 1980s Haggitt and Dean first recognized the association between CLE with IM and the risk of developing EAC. Since then, documentation of the presence of IM has been necessary to establish the diagnosis of BE in the United States and most of the world.

With the introduction and popularization of the flexible fiberoptic upper endoscope came the need for a reliable method to identify the gastroesophageal junction (GEJ) from within the esophageal lumen. External landmarks, such as the location of the peritoneal reflection, and the presence of esophageal glands are not discernible at the time of endoscopy. The first endoscopic definition of the GEJ was proposed by Hayward in 1961 and determined the stomach to start where the esophageal tube flared into a gastric pouch. In this publication, Hayward also described what he believed to be the normal histology spanning the GEJ. Specifically, he stated that the lower 1 to 2 cm of the esophagus could be lined by columnar epithelium and that this “cardiac (or junctional) mucosa” extended 3 cm onto the stomach ( Fig. 31.3 ).

Despite the lack of strong evidence supporting his conclusions, Hayward's observation that a 5-cm segment of cardiac epithelium normally lined the distal 2 cm of the esophagus and proximal 3 cm of the stomach was accepted as fact for the next 30 years. He hypothesized that this “junctional” epithelium served to buffer the esophageal squamous lining from the noxious effects of gastric acid. The theory was based, at least in part, on the premise that the juxtaposition of gastric oxyntic mucosa directly with squamous esophageal mucosa was not teleologically sound due to the potential for inducing erosive reflux esophagitis; a buffer zone was thought necessary. The dogma that the distal 2 cm of the esophagus could be lined by cardiac epithelium led to the dictum that at least 3 cm of CLE had to be present to establish the diagnosis of BE. To this day, the 3-cm cutoff has persisted in distinguishing long-segment BE (LSBE), defined as CLE 3 cm or more in length, from short-segment BE (SSBE), defined as CLE less than 3 cm.

In 1987 McClave and colleagues first suggested that the GEJ was delineated by the proximal edge of the gastric rugal folds. No empiric data were provided to support the claim. Despite the lack of evidence, this endoscopic landmark of the GEJ has persisted and is commonly used currently. At the 2004 AGA Chicago Workshop on the Management and Diagnosis of BE, a number of statements were formulated, including statement 7: “The proximal margin of the gastric folds is a reliable endoscopic marker for the gastroesophageal junction.” The evidence behind this statement was determined to be grade IV-C, or from the “opinions of respected authorities based on clinical experience, descriptive studies, or reports of expert committees” and with “poor evidence to support the statement.” Furthermore, the manuscript stated, “The group universally favored using the proximal margin of the gastric folds to identify the GEJ but recognized that there are scant data that validate it.”

Given the lack of a universally accepted endoscopic criterion for the location of the GEJ, the most recent position of the AGA (in 2011) was, “A majority of published studies on Barrett's esophagus conducted over the past 20 years have used the proximal extent of the gastric folds as the landmark for the GEJ. In the absence of compelling data for the use of alternative markers, we advocate the continued use of this landmark.” Thus in the United States the definition of the GEJ continues out of tradition and convenience rather than out of scientific proof of the validity of the proximal rugal folds as an appropriate landmark. To the contrary, the GEJ is defined in Japan by the distal limit of palisade vessels visualized in the lower esophagus.

Countering the validity of the rugal folds as an accurate determinant of the GEJ is the theory that, as the lining of the esophagus is damaged by reflux of gastric contents, the squamous epithelium undergoes metaplasia to a columnar cell type with the SCJ migrating proximally as a result. In addition, with progressive GERD the distal esophagus dilates and loses its tubular configuration, becomes part of the gastric sac, and develops rugal folds. Therefore what had once been tubular esophagus lined with squamous epithelium becomes gastric in appearance with rugal folds. The use of the proximal limit of the rugal folds as the criterion for determining the GEJ in this model would shift it inappropriately cephalad.

Histologic Determination of the Gastroesophageal Junction

Given the importance of the location of the GEJ in determining the presence of BE, alternative definitions of the junction have been sought. Due to the inability to rely on external landmarks at the time of endoscopy, and the lack of data confirming the validity of the proximal extent of the gastric rugal folds as an appropriate marker, histologic assessment has been proposed as the definitive manner to denote this junction.

Critical to assessing histologic abnormalities in the region of the GEJ is an understanding of the epithelia residing within the esophagus and stomach in health and disease. The normal esophagus is lined with stratified squamous epithelium; this epithelium is never found in the stomach. On the other hand, gastric oxyntic epithelium is always present and confined to the stomach. In the junctional zone between the esophagus and stomach, three types of epithelium may be found and are variably present: pure cardiac, oxyntocardiac, and intestinalized cardiac. These three subtypes of columnar epithelium are distinguishable by the presence of mucous, parietal, and goblet cells.

In a study from 1976 assessing the histology of CLE, with the GEJ defined by manometry, Paull et al. detected these three variable epithelial types. Cardiac (“junctional type”) epithelium ( Fig. 31.4 ), comprised exclusively of mucus-secreting cells, and oxyntocardiac (“gastric fundic type”) epithelium ( Fig. 31.5 ), comprised of mucus-secreting cells as well as some parietal and chief cells, were found in all patients; intestinalized cardiac (“specialized type”) epithelium ( Fig. 31.6 ), containing mucus-secreting cells and prominent goblet cells, was found in some. Of note, these three cell types have been detected without visible CLE in the region just distal to the GEJ, as defined by the proximal extent of the gastric rugal folds.

When present, these epithelia always reside between the squamous lining of the esophagus and gastric oxyntic mucosa, creating what Chandrasoma and colleagues have termed a squamo-oxyntic gap ( Fig. 31.7 ). Of note, when only one epithelium is present, such as is found only in short gaps generally less than 5 mm, it is oxyntocardiac. In longer gaps, cardiac and oxyntocardiac mucosa may both be present, with cardiac in a more proximal location. When all three epithelial types are present, intestinalized cardiac epithelium is proximal, cardiac epithelium is in the middle, and oxyntocardiac epithelium is distal, although admixing may occur. Thus the distribution of types of columnar epithelium within CLE is not random, with IM residing at the most proximal region of CLE and extending downward while nonintestinalized columnar epithelia occupy a more distal position. In patients with CLE the squamo-oxyntic gap traverses the GEJ as defined by the rugal folds, whereas in patients who have a normal-appearing GEJ on endoscopy, the gap is entirely subjacent to the tubular esophagus.

An extensive body of literature has emerged from the University of Southern California disproving the dogma that 5 cm of cardiac mucosa normally reside within the distal esophagus and proximal stomach. The competing theory proposed by Chandrasoma and co-authors from that institution is that cardiac mucosa is not present at the normal GEJ but rather represents metaplastic esophageal epithelium resulting from pathologic exposure to refluxed gastric juice.

The proof of their theory comes in several forms. An assessment of biopsies taken just distal to the normal-appearing GEJ, using a retroflexed endoscope, found cardiac epithelium usually less than 5 cm in length and occasionally absent. Patients with cardiac and oxyntocardiac epithelium were more likely to have an abnormal 24-hour ambulatory esophageal pH study and a mechanically defective lower esophageal sphincter on manometry, both suggesting the presence of GERD, than patients lacking those epithelial types.

An autopsy study reported in 2000 assessed the histology of the GEJ in adult subjects who had no evidence of GERD throughout their lives. Cardiac epithelium was absent in more than half of specimens, and the length of cardiac and oxyntocardiac epithelia ranged from 0.4 to 8.05 mm, with a squamo-oxyntic gap of less than 5 mm in the majority. Half of subjects were found to have a direct transition from squamous to gastric oxyntic mucosa in at least part of the circumference of the SCJ. These findings were validated by a later independent study. An additional autopsy study in children found cardiac epithelium in all subjects, but limited to a length between 1 and 4 mm, with a median of 1.8 mm, far less than the 50-mm length previously thought to be normal.

Based on these findings, a direct transition from esophageal squamous mucosa to oxyntic gastric mucosa does exist in some individuals, without a buffer zone of cardiac epithelium. When cardiac epithelium is present in normal individuals, its length is short, generally limited to a few millimeters or less. Additional studies have demonstrated that the presence of cardiac epithelium distal to the endoscopic GEJ, as defined by the rugal folds, is associated with GERD, with the length of the segment being a sensitive indicator of GERD severity. Finally, cardiac mucosa residing distal to the SCJ generally demonstrates chronic inflammation in the lamina propria, consistent with GERD, and likely not related to Helicobacter pylori infestation of the stomach, the other potential cause. Mucosal inflammation is the major stimulus for the development of IM, both in the esophagus and stomach.

If the premise is accepted that cardiac epithelium at the GEJ is pathologic, related to GERD-induced metaplasia of the esophageal mucosa rather than a normal lining of the proximal stomach, an accurate histologic determination of the GEJ becomes possible. In this model the GEJ is defined by the proximal border of gastric oxyntic mucosa as assessed on biopsies. In a normal subject the GEJ is located where esophageal squamous mucosa transitions to gastric oxyntic mucosa. In patients with GERD the GEJ is located where metaplastic columnar epithelium residing within the squamocolumnar gap transitions to gastric oxyntic mucosa. Because this latter transition is not discernible at the time of endoscopy, the true GEJ in such circumstances can only be determined by extensive, meticulous, and accurately recorded biopsies.