barrett’s esophagus: histopathologic definitions and diagnostic criteria

7
World J. Surg. 28, 148-154, 2004 DOI: 10.1007/s00268-003-7050-4 O WORLD Journal of SURGERY © 2004 by the Societe Internationale de Chirurgie Barrett's Esophagus: Histopathologic Definitions and Diagnostic Criteria James Mueller, M.D.,' Martin Werner, M.D., Z Manfred Stolte, M.D. 3 'Department of Surgery, Klinikum Rechts der Isar, Technical University of Munich, Ismaningerstrasse 22, D -81675 Munich, Germany 2Institute of Pathology, Klinikum Rechts der Isar, Technical University of Munich, Ismaningerstrasse 22, D -81675 Munich, Germany 3 Institute of Pathology, Klinikum Bayreuth, Preuschwitzerstrasse 101, D-95445 Bayreuth, Germany Published Online: January 20, 2004 Abstract. Adenocarcinoma of the distal esophagus is rising more rapidly in incidence than any other visceral malignancy in the Western world. It is well established that most, if not all, of these tumors develop in Barrett's esophagus via the metaplasia-dysplasia-carcinoma sequence and could theoretically be detected at an early stage, but despite this, the majority of these tumors are still detected late in their course. This highlights the fact that the goal of effective surveillance for patients at risk for developing an adenocarcinoma of the distal esophagus is still far off. In addition, adeno- carcinomas of the esophagogastric junction and gastric cardia are also ris- ing in incidence, but their carcinogenesis and their relation to Barrett's esophagus are still being defined, as are the meaning and significance of the relatively new entities "short-segment Barrett's" and "ultra-short- segment Barrett's". This review attempts to clarify the main histopatbolog- ic issues concerned with the definition of Barrett's esophagus, its distinc- tion from intestinal metaplasia of the gastric cardia, as well as the criteria for the histologic diagnosis of dysplasia and carcinoma in Barrett's esopha- gus. Adenocarcinomas of the distal esophagus and esophagogastric junction are rising more rapidly in incidence than any other visceral malignancy in the industrialized Western world, despite the fact that the incidence of esophageal squamous cell carcinoma is con- stant and the incidence of gastric carcinoma is falling in this region [1, 2]. These tumors represent a particularly serious problem be- cause most are detected late in their course and the patients have correspondingly poor prognoses [3, 4]. Because the treatment of locally advanced and metastatic tumors of the distal esophagus and esophagogastric junction is not likely to have a decisive impact on this situation, early detection and treatment of these tumors is in- creasingly recognized as the crucial factor for improving the situa- tion. However, at present, the role, design, and effectiveness of sur- veillance programs for patients with Barrett's esophagus, or at risk for the development of Barrett's esophagus (i.e. reflux symptoms), including endoscopy with systematic biopsies, is still being debated [5, 6] and has not yet been firmly established in clinical practice. For some time it has been recognized that adenocarcinoma of This article is part of the World Progress in Surgery symposium on Bar- rett's esophagus and esophageal cancer, which was published in Volume 27, Number 9 of World Journal of Surgery. Correspondence to: James Mueller, M.D., Department of Pathology, Bay- state Medical Center, 759 Chestnut Street, 01199 Springfield, Massachu- setts, USA, e-mail: [email protected] the distal esophagus nearly always arises in an epithelium that has previously undergone glandular metaplasia ("Barrett's esopha- gus") [7]. A histopathologic pathway (the metaplasia-dysplasia- carcinoma sequence) has been defined and characterized that pro- vides the theoretical basis for the surveillance and early detection of esophageal adenocarcinoma at a point at which it can be success- fully treated with a low rate of morbidity and mortality [8]. The precursor lesions and carcinogenetic pathways for more distal tu- mors in the esophagogastric junction and proximal stomach (gas- tric cardia and subcardial region) are presently not as well charac- terized. Although a great deal has been learned about Barrett's esopha- gus since Norman Barrett first described glandular epithelium in the lower esophagus in 1950 [9], many questions regarding the his- topathology, classification, molecular biology, and clinical manage- ment of Barrett's esophagus and adenocarcinoma of the distal esophagus are still unanswered. The molecular biology of Barrett's carcinoma and its precursor lesions, and to a lesser extent, carci- noma of the gastric cardia, is increasingly better understood on the basis of a series of studies of protein expression (through immuno- histochemistry [10-13]; genomic changes—e.g. comparative geno- mic hybridization [14, 15], fluorescence in situ hybridization [FISH] [16], and cytogenetic studies [17, 18]; and gene expression level [e.g., reverse transcriptase polymerase chain reaction for mRNA determinations). These studies are based on the hope that, in the future, specific molecular changes might serve as reliable markers for the early identification of patients who are likely to develop carcinoma [19, 20]. At present, however, our best predictor of the future development of a carcinoma in a given patient with Barrett's esophagus is still the histologic identification of dysplasia. The pur- pose of this review is to present an overview of the histopathology of Barrett's esophagus and carcinoma, with particular emphasis on those features that have an impact on the clinical diagnosis and management of Barrett's esophagus and carcinoma. Changing Definition of Barrett's Esophagus Over the past 50 years, the definition of Barrett's esophagus has gone through three major phases. The first of these began when Norman Barrett, a surgeon, described red glandular mucosa in the distal esophagus of some of his patients that he assumed to be a

Upload: james-mueller

Post on 10-Jul-2016

216 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Barrett’s esophagus: Histopathologic definitions and diagnostic criteria

World J. Surg. 28, 148-154, 2004DOI: 10.1007/s00268-003-7050-4 O WORLD

Journal ofSURGERY© 2004 by the Societe

Internationale de Chirurgie

Barrett's Esophagus: Histopathologic Definitions and Diagnostic Criteria

James Mueller, M.D.,' Martin Werner, M.D., Z Manfred Stolte, M.D. 3

'Department of Surgery, Klinikum Rechts der Isar, Technical University of Munich, Ismaningerstrasse 22, D -81675 Munich, Germany2Institute of Pathology, Klinikum Rechts der Isar, Technical University of Munich, Ismaningerstrasse 22, D -81675 Munich, Germany3Institute of Pathology, Klinikum Bayreuth, Preuschwitzerstrasse 101, D-95445 Bayreuth, Germany

Published Online: January 20, 2004

Abstract. Adenocarcinoma of the distal esophagus is rising more rapidly inincidence than any other visceral malignancy in the Western world. It iswell established that most, if not all, of these tumors develop in Barrett'sesophagus via the metaplasia-dysplasia-carcinoma sequence and couldtheoretically be detected at an early stage, but despite this, the majority ofthese tumors are still detected late in their course. This highlights the factthat the goal of effective surveillance for patients at risk for developing anadenocarcinoma of the distal esophagus is still far off. In addition, adeno-carcinomas of the esophagogastric junction and gastric cardia are also ris-ing in incidence, but their carcinogenesis and their relation to Barrett'sesophagus are still being defined, as are the meaning and significance ofthe relatively new entities "short-segment Barrett's" and "ultra-short-segment Barrett's". This review attempts to clarify the main histopatbolog-ic issues concerned with the definition of Barrett's esophagus, its distinc-tion from intestinal metaplasia of the gastric cardia, as well as the criteriafor the histologic diagnosis of dysplasia and carcinoma in Barrett's esopha-gus.

Adenocarcinomas of the distal esophagus and esophagogastricjunction are rising more rapidly in incidence than any other visceralmalignancy in the industrialized Western world, despite the factthat the incidence of esophageal squamous cell carcinoma is con-stant and the incidence of gastric carcinoma is falling in this region[1, 2]. These tumors represent a particularly serious problem be-cause most are detected late in their course and the patients havecorrespondingly poor prognoses [3, 4]. Because the treatment oflocally advanced and metastatic tumors of the distal esophagus andesophagogastric junction is not likely to have a decisive impact onthis situation, early detection and treatment of these tumors is in-creasingly recognized as the crucial factor for improving the situa-tion. However, at present, the role, design, and effectiveness of sur-veillance programs for patients with Barrett's esophagus, or at riskfor the development of Barrett's esophagus (i.e. reflux symptoms),including endoscopy with systematic biopsies, is still being debated[5, 6] and has not yet been firmly established in clinical practice.

For some time it has been recognized that adenocarcinoma of

This article is part of the World Progress in Surgery symposium on Bar-rett's esophagus and esophageal cancer, which was published in Volume 27,Number 9 of World Journal of Surgery.

Correspondence to: James Mueller, M.D., Department of Pathology, Bay-state Medical Center, 759 Chestnut Street, 01199 Springfield, Massachu-setts, USA, e-mail: [email protected]

the distal esophagus nearly always arises in an epithelium that haspreviously undergone glandular metaplasia ("Barrett's esopha-gus") [7]. A histopathologic pathway (the metaplasia-dysplasia-carcinoma sequence) has been defined and characterized that pro-vides the theoretical basis for the surveillance and early detectionof esophageal adenocarcinoma at a point at which it can be success-fully treated with a low rate of morbidity and mortality [8]. Theprecursor lesions and carcinogenetic pathways for more distal tu-mors in the esophagogastric junction and proximal stomach (gas-tric cardia and subcardial region) are presently not as well charac-terized.

Although a great deal has been learned about Barrett's esopha-gus since Norman Barrett first described glandular epithelium inthe lower esophagus in 1950 [9], many questions regarding the his-topathology, classification, molecular biology, and clinical manage-ment of Barrett's esophagus and adenocarcinoma of the distalesophagus are still unanswered. The molecular biology of Barrett'scarcinoma and its precursor lesions, and to a lesser extent, carci-noma of the gastric cardia, is increasingly better understood on thebasis of a series of studies of protein expression (through immuno-histochemistry [10-13]; genomic changes—e.g. comparative geno-mic hybridization [14, 15], fluorescence in situ hybridization [FISH][16], and cytogenetic studies [17, 18]; and gene expression level[e.g., reverse transcriptase polymerase chain reaction for mRNAdeterminations). These studies are based on the hope that, in thefuture, specific molecular changes might serve as reliable markersfor the early identification of patients who are likely to developcarcinoma [19, 20]. At present, however, our best predictor of thefuture development of a carcinoma in a given patient with Barrett'sesophagus is still the histologic identification of dysplasia. The pur-pose of this review is to present an overview of the histopathologyof Barrett's esophagus and carcinoma, with particular emphasis onthose features that have an impact on the clinical diagnosis andmanagement of Barrett's esophagus and carcinoma.

Changing Definition of Barrett's Esophagus

Over the past 50 years, the definition of Barrett's esophagus hasgone through three major phases. The first of these began whenNorman Barrett, a surgeon, described red glandular mucosa in thedistal esophagus of some of his patients that he assumed to be a

Page 2: Barrett’s esophagus: Histopathologic definitions and diagnostic criteria

Mueller et al.: Definition and Diagnosis of Barrett's Esophagus 149

portion of the stomach which had been pulled into the distalesophagus as a result of the scarring following ulceration [9]. (Hisdescription, however, was not the first report of this finding; in 1906a pathologist, Tileston, had described gastric-like mucosa in thedistal esophagus of patients who had an "esophageal peptic ulcer"[211.) On the basis of these early descriptions Barrett's esophaguswas almost entirely macroscopically defined and was also consid-ered by many authors to be a congenital, rather than an acquired,condition [22].

As Barrett's esophagus became the object of histopathologic andadditional clinical studies, it began to be appreciated that it couldbe acquired and that it was not necessarily composed of a gastric-type epithelium. In this second phase, three types of glandular epi-thelium were identified as being characteristic of Barrett's esopha-gus: (1) specialized intestinal epithelium (with goblet cells), (2)junctional epithelium (or cardia-antral type), without goblet cells,and (3) fundic or oxyntic epithelium, also without goblet cells [23].In retrospect, the identification of these three types of epitheliummay have been due to uncertainty about where the biopsies hadcome from—i.e., from the esophagus or from a hiatal hernia—because this could not be distinguished by the techniques used atthat time (barium swallow with radiography) [23]. The diagnosis ofBarrett's esophagus was thus based on two criteria: (1) the histo-logic identification of one of these types of epithelium and (2) thefinding of a segment with columnar metaplasia in the esophagus ofat least 3 cm length (according to some authors, 5 cm [24]). Short-segment Barrett's was later defined as 2-3 cm in length, with Bar-rett's epithelium < 2.0 cm in length considered to be a normal vari-ant [25]. Pathology reports based on this concept identified the typeof glandular epithelium that was present and used phrases such as"consistent with Barrett's esophagus," because the diagnosis wasnot based on histopathology alone.

The third, and present, phase began with the increasingly com-mon use of endoscopy and the recognition that specialized colum-nar epithelium was not a normal finding in the distal esophagus,and that only this type of epithelium was associated with an in-creased risk for the development of adenocarcinoma [5]. In 1994,Spechler et al. reported that 18% of patients with normal endo-scopic findings (regardless of the indication for the endoscopy pro-cedure) had specialized columnar epithelium in the vicinity of thegastroesophageal junction [26]. Subsequent studies in larger num-bers of patients in similar clinical settings found frequencies ofabout 12% of this epithelium near the esophagogastric junction inpatients without endoscopic findings [27, 28]. Clearly then, this wasnot a normal finding. Because it has become firmly established thatthe term `Barrett's esophagus" denotes a precancerous lesion,most authors now recommend that pathologists only use the termwhen specialized intestinal epithelium is found. Simultaneously,the diagnosis of Barrett's esophagus is increasingly based on histo-pathologic findings alone, regardless of the length of the columnarepithelium [29]. Although this concept seems to be more straight-forward, it is also associated with a few new problems [30], as will beshown below.

Barrett's Epithelium: Histopathology

Normal Esophageal Epithelium, Changes in GastroesophagealReflux Disease

Based on epidemiologic and clinical studies, a clear link has beenestablished between gastroesophageal reflux disease (GERD) and

the development of esophageal specialized columnar epithelium.Up to 20% of patients with GERD have "long segment" (greaterthan 3 cm length) Barrett's esophagus [31], and the areas of theworld where GERD is common also have the highest incidence ofBarrett's esophagus [32]. Because not all patients with GERD haveBarrett's esophagus, and because the components of the refluxatevary among patients, both the components of the refluxate (i.e.,pancreatic, biliary juice) and the role of the lower esophagealsphincter (LES) in promoting metaplasia are areas of active re-search [33].

The normal mucosa of the esophagus is composed of a stratifiedsquamous epithelium resting upon a lamina propria. Changes con-sistent with reflux esophagitis include hyperplasia of the basal celllayer, broadening of the basal zone, extension of the papillae intothe lamina propria, erosion and ulceration of the mucosa, and in-traepithelial eosinophils and neutrophils, which are less common,although frequently cited in the literature [34]. Mucosal capillarycongestion and hemorrhage are taken by some authors to be par-ticularly important because they correlate with the signs of redden-ing seen by endoscopy [35]. However, none of these changes arediagnostic of GERD and must be carefully compared with endo-scopic and clinical findings.

Specialized Intestinal Metaplastic Columnar Epithelium

Barrett's epithelium is a form of metaplasia (replacement of onemature tissue type by another) in which the normal squamous lin-ing is replaced by a glandular lining of the specialized intestinaltype. Although earlier it was believed that this epithelium was theresult of upward migration of gastric epithelium, it is now generallyaccepted that it originates from a multipotential cell in the esopha-gus itself, which could be found in the basal layer of squamous lin-ing or among the cells lining the ducts of submucosal glands. Stud-ies using electron microscopy have identified cells in Barrett'sesophagus with features intermediate between squamous and glan-dular cells—i.e., cells with both microvilli characteristic of glandu-lar cells and desmosomes, which are normally found only in squa-mous cells [36]. It has also been reported that, in some cases ofBarrett's esophagus, a hybrid epithelium can be found which hasthe appearance of stratified squamous epithelium but expresses cy-tokeratins (e.g., cytokeratin 19), which are characteristic of glandu-lar epithelium [37]. An additional study demonstrated that whilethe normal gastric cardia expresses cytokeratins 8 and 19 and thenormal stratified squamous epithelium expresses cytokeratins 4and 13, the columnar epithelium of Barrett's esophagus expressesall four of these cytokeratin types [38], fitting well with the conceptthat this is a true hybrid epithelium. The reason for the formation ofthe glandular epithelium is not known, but presumably this type ismore resistant to damage resulting from gastroesophageal reflux.

Histologically, this specialized intestinal metaplastic columnarepithelium (see Fig. 1) is characterized by a columnar epitheliumthat contains several cell types. At the surface, (1) goblet cells, (2)blue columnar, (3) clear columnar, and (4) enterocyte-like cellswith brush borders are found. Scattered Paneth cells can also rarelybe seen, and also rarely, mature enterocytes are observed [32]. Be-low the surface, coiled, mucus-type glands are present. AlthoughBarrett's epithelium is thought to form as a result of gastroesopha-geal reflux, there is seldom a marked inflammatory component[32]. It is now generally agreed that the goblet cell is the key celltype for the identification of Barrett's epithelium. Usually goblet

Page 3: Barrett’s esophagus: Histopathologic definitions and diagnostic criteria

150 World J. Surg. Vol. 28, No. 2, February 2004

Long, Short, and Ultra-short-segment Barrett's Esophagus

Fig. 1. The specialized intestinal metaplastic columnar epithelium of Bar-rett's esophagus showing prominent goblets cells that can be identified eas-ily on this H&E stain (H&E stain, original magnification 100x).

cells are easy to identify in a routine H&E stain, but at times theyare not easily seen and an Alcian blue stain at a pH of 2.5 is recom-mended to help make them more apparent [29]. With this stain,neutral mucin (as found in the gastric cardia) is negative, but acidicmucin (found in the intestine or in Barrett's epithelium) stains darkblue. In the past, some authors accepted positive staining of colum-nar cells with the Alcian blue stain as sufficient for the diagnosis ofBarrett's esophagus [32], but it has become apparent that the cardiacan also express acidic mucins and stain positively in columnar cellsin some cases, which could lead to a misinterpretation of biopsiesnear the esophagogastric junction (see also below). Usually, how-ever, when adequate biopsies (at least three particles) are taken,most experienced pathologists can reliably identify the specializedcolumnar epithelium of Barrett's esophagus on a routine H&Estain.

Based on histopathologic findings and the type of mucin pro-duced, intestinal metaplasia throughout the gastrointestinal tractcan be classified as "complete" (type I) or "incomplete" (type II orIII) [39]. In type I intestinal metaplasia, columnar cells that havethe features of small intestine absorptive cells are present, togetherwith goblet cells and significant numbers of Paneth cells. Incom-plete intestinal metaplasia may be either of the gastric (tType II) orcolonic type (type III), in which the columnar cells seen along withthe goblet cells resemble gastric foveolar or colonic absorptivecells, respectively. Paneth cells are only occasionally seen in incom-plete intestinal metaplasia. The columnar epithelium characteristicof Barrett's esophagus is classified as intestinal metaplasia of theincomplete type (types II or III) and has been found to be associ-ated with GERD, in contrast to the complete form of intestinalmetaplasia, usually seen in the stomach, that is associated with atro-phic gastritis [40]. In the columnar cells, the Alcian blue stain at pH2.5 is negative in complete intestinal metaplasia, but positive in in-complete metaplasia [41]. A further distinction between mucintypes can be made using the high-iron diamine-PAS stain (HID-PAS) in which sialomucin stains blue (typical of type II metaplasia)and sulfomucin stains brown black (typical of the type III form).There have been reports that loss of sialomucin expression in favorof sulfomucin and the type III form of intestinal metaplasia mightbe more closely related to the development of carcinoma, but this iscontroversial [42, 43].

Depending on the extent of the metaplastic change, Barrett'sesophagus may be classified as "long segment" (at least 3 cm inlength), "short segment" (involving < 3 cm), and "ultra-short-segment" (not visible endoscopically, but histologically confirmedcolumnar epithelium) [44, 45]. Although this descriptive approachseems straightforward, it is also associated with a few practicalproblems and open questions [5]. First, although the connectionbetween Barrett's esophagus (conventional, or long-segment Bar-rett's) and the development of adenocarcinoma is well defined,with overall risk estimates for patients with Barrett's esophagusvarying from 1/52-441 patient years, or a lifetime risk for an indi-vidual patient of about 5% [28, 46], the risk of carcinoma develop-ment for short-segment or ultra-short-segment Barrett's is less welldefined [8]. That risk, however, appears to be < that associatedwith long-segment Barrett's, because a smaller surface area is in-volved. Studies have found that 8% to 12% of patients with short-segment Barrett's have dysplasia, whereas dysplasia is evident in19% to 24% of patients with long-segment Barrett's [47, 48]. Evenless is known about the risk of dysplasia or carcinoma developmentin ultra-short-segment Barrett's. Because biopsy with the intent offinding short or ultra-short-segment Barrett's is not routine,, theoverall incidence of these types is not known. It is estimated, how-ever, that short-segment Barrett's is about 6 to 20 times more fre-quent than long-segment Barrett's [26] and that ultra-short-segment Barrett's may be even more frequent. Therefore, despitethe lower risk of carcinoma development in an individual patientwith these shorter forms of Barrett's esophagus, the contribution ofthese forms to the overall frequency of adenocarcinoma could beconsiderable because they are more frequent.

The term "ultra-short-segment Barrett's esophagus" is not uni-versally accepted because there is uncertainty about whether thebiopsy came from the esophagus can easily lead to confusion withintestinal metaplasia of the cardia (see below). Some authors pre-fer the term "intestinal metaplasia of the gastroesophageal junc-tion" because this does not rely on precise localization of the biopsy[44]. The risk of cancer development and even the etiology of ultra-short-segment Barrett's esophagus or metaplasia of the gastro-esophageal junction continues to be debated; some authors evenconsider it to be a finding without pathologic significance [49]. An-other concept is that cardia epithelium is an acquired pathologicepithelium that extends into the distal esophagus and is the precur-sor of Barrett's epithelium [50, 51].

Barrett's Esophagus versus Metaplasia of the Gastric Cardia

As indicated by the foregoing discussion, an important questionthat has received increasing attention because of the emphasis onhistopathology for the diagnosis of Barrett's esophagus, is its dis-tinction from intestinal metaplasia of the gastric cardia. Cardia in-testinal metaplasia (CIM) is thought to have a lower cancer riskthan Barrett's esophagus (at least long-segment Barrett's), but it isalso to be frequent, because the number of cases of cardia carci-noma is rising rapidly [51]. One systematic study of 225 patientswho underwent endoscopy for reasons other than reflux symptomsfound that 4% had long-segment Barrett's esophagus, 7% had theshort-segment form, 15% had intestinal metaplasia at the esopha-gogastric junction (EGJ) (or ultra-short-segment Barrett's) and4% had CIM. Unlike Barrett's, CIM does not appear to be related

Page 4: Barrett’s esophagus: Histopathologic definitions and diagnostic criteria

Mueller et al.: Definition and Diagnosis of Barrett's Esophagus 151

to GERD, and many cases appear to be associated with. Helicobac-terpylori infection [52, 35]. In addition, the demographics of pa-tients with CIM differ from those of patients with Barrett's [53].Barrett's esophagus patients are likely to be white men who smokeand have GERD, whereas patients with CIM seldom have thesefeatures, but may be more likely to have Helicobacter pylori infec-tion [54].

The best criterion for the distinction between Barrett's esopha-gus and CIM is the precise knowledge that the biopsy specimencame from the esophagus or the stomach, but because this is notalways possible at the EGJ, a distinction based on the biopsy itself isalso desirable. Such a distinction is difficult, but histologic criteriado exist by which one can be made [54]. Histologically, Barrett'sepithelium is characterized not only by goblet cells but also by Al-cian blue-positive immature goblet cells, or "pre-goblet cells." Incontrast, CIM tissues include no pre-goblet cells, but they are fre-quently characterized by enterocytes. Especially important is thefinding that the areas adjacent to CIM show normal foveolar epi-thelium where those of Barrett's epithelium contain pre-gobletcells.

Special stains have also been used to help distinguish CIM.Metaplasia of the EGJ (or ultra-short-segment Barrett's esopha-gus) has been found to more frequently express sulfomucins (high-iron diamine/Alcian blue [AB/HID]) (p < 0.0001) and to involvethe surface glandular epithelium (p < 0.0001) than intestinal meta-plasia of the cardia [55]. It has also been reported that immunohis-tochemistry may be helpful in making the distinction. For example,one study found that the combination of cytokeratin antibodiesCK7 and CK20 can identify Barrett's esophagus (superficial mu-cosa with CK20 staining and strong CK7 staining of the entiregland) in 95%-100% of cases, while gastric intestinal metaplasiacases were negative, including those of the cardia [56]. These re-sults have not been confirmed by subsequent reports [57]. Morerecently, the monoclonal antibody MAbDAS-1 has been reportedto reliably identify Barrett's esophagus, at least the form having thecharacteristic histology of the colonic type of intestinal metaplasia(incomplete intestinal metaplasia type III) [58].

Other helpful information in making the distinction betweenBarrett's esophagus and CIM are symptoms and the status of theremainder of the gastric mucosa. On the one hand, when intestinalmetaplasia is found in biopsies near the EGJ in a patient known tohave H. pylori gastritis without reflux symptoms, this is likely CIMdue to H. pylori. On the other hand, in a patient with reflux symp-toms in whom normal corpus and antral biopsies or only C-gastritisin the antrum are found, the diagnosis is more likely Barrett'sesophagus, especially when the histopathologic findings are alsoconsistent with this diagnosis. In this connection, a recent reporthas described findings in biopsies of the cardia that are said to dis-tinguish between gastroesophageal reflux and H. pylori as the eti-ology [59].

Dysplasia and Carcinoma in Barrett's Epithelium

Dysplasia

As already mentioned, dysplasia is the most reliable marker for theidentification of patients who will develop an esophageal carci-noma [8]. As in the rest of the GI tract, dysplasia in Barrett'sesophagus is defined as an unequivocally neoplastic process. Twogrades, low grade and high grade, are used, with an additional class

Fig. 2. Low grade dysplasia in Barrett's esophagus. Characteristic featuresinclude depletion of cytoplasmic mucin, hyperchromatic, more prominentnuclei, and beginning pseudostratification (H&E stain, original magnifica-tion 200x).

"indefinite for dysplasia" applied in uncertain cases [32]. Low-grade dysplasia (LGD) in Barrett's epithelium (Fig. 2) is character-ized by hyperchromatic, enlarged nuclei and depletion of cytoplas-mic mucin. The process often extends to the surface of theepithelium with architectural changes that include budding ofglands, and pseudostratification. In high-grade dysplasia (HGD)the changes are similar to LGD, but are more marked (Figs. 3 A, Band 4). In HGD the glands become complex, with bifurcation andbranching, the cells are hyperchromatic and have only scant cyto-plasm. By definition, no invasion of the lamina propria is present.

There is a relatively high rate of interobserver variability amongpathologists for the identification of LGD and its distinction fromHGD. When asked to classify a given lesion as LGD or HGD, therate of disagreement among pathologists ranges from 50% to 70%)[60]. For HGD and carcinoma, the figures are somewhat better, atabout 85% to 90%. However, because the consequences of the di-agnosis are profound, in case of a dysplasia diagnosis, the recom-mendation that the biopsies be reviewed by a second pathologistwith specific experienced in this area is a component of nearly allrecommendations for endoscopic surveillance of Barrett's esopha-gus [8, 6].

Carcinoma

For the diagnosis of carcinoma (Fig. 4), invasion of the lamina pro-pria is required (UICC- International Union Against Cancer [61]).However, because the incidence of carcinoma in the esophagealresection specimen after a biopsy diagnosis of HGD is very high (upto 90%), in most centers the consequence of the diagnosis of HGDis the same as that of carcinoma—i.e., esophageal resection or,more recently, ablation [6, 62]. Among patients who are diagnosedwith carcinoma confined to the mucosa by biopsy, 7% already havesubmucosal invasion in the resection specimen, and as many as halfof those with submucosal invasion have regional lymph node me-tastases [51]. For patients with carcinoma confined to the mucosain the resection specimen, 2% or less have regional lymph nodemetastases [63, 64]. An important aspect of the evaluation of thedepth of invasion of carcinoma in Barrett's esophagus involves theBarrett's epithelium, which may have its own, newly formed mus-

Page 5: Barrett’s esophagus: Histopathologic definitions and diagnostic criteria

152

World J. Surg. Vol. 28, No. 2, February 2004

Fig. 3. High-grade dysplasia in Barrett's esophagus. Here the featuresshown in Figure 2 are accentuated and the nuclei are enlarged, pleomor-phic, and hyperchromatic with loss of polarity within the epithelium. No

invasion of the lamina propria is identified (H&E stains, original magnifi-cations: A, 200x, B, 400x).

Fig. 4. Adenocarcinoma arising in Barrett's esophagus, with infiltration ofthe moderately differentiated tumor throughout the lamina propria (H&Estain, original magnification 100x).

cularis mucosa lying superficial to the original muscularis mucosa.In this case, only invasion through the original, deeper muscularismucosa is defined as "submucosal" invasion [65].

Histology after Laser Ablation

A special situation, one that confronts pathologists with increasingfrequency, is the evaluation of biopsies from Barrett's esophagusthat has been treated for the removal of the Barrett's epitheliumwith laser (argon-plasma coagulation) ablation, photodynamictherapy, or medications including proton pump inhibitors (PPI).One study of the histologic changes after these types of therapyfound varying degrees of replacement of the Barrett's epitheliumby squamous epithelium, formed either by ingrowth of the adjacent

normal squamous epithelium of the esophagus or by squamous is-lands, perhaps from the ducts of submucosal glands or by appar-ently squamous metaplasia of the Barrett's epithelium itself [66].This epithelial replacement should result in a marked reduction ofthe risk of carcinoma development by virtue of a reduction in thesurface area of Barrett's epithelium. Although some researchershave reported that complete replacement of the Barrett epitheliumby squamous epithelium occurs in most cases after ablation [67],others have found that re-replacement of the Barrett's epitheliumby squamous epithelium is not complete, so that islands of glandu-lar epithelium remain—in some cases even areas of dysplasia-beneath the newly formed squamous epithelium [66]. This obvi-ously complicates any post-therapeutic endoscopic surveillanceprogram directed toward detecting the development of dysplasia orcarcinoma. It is particularly important in such a situation that ad-equate, deep biopsies be obtained during endoscopy.

Endoscopy Biopsy Protocol

Two factors place significant pressure on the endoscopist takingbiopsies in possible cases of Barrett's esophagus: (1) the conceptthat any length of columnar epithelium within the esophagus con-stitutes Barrett's esophagus and (2) the importance of the differen-tial diagnosis of Barrett's esophagus versus CIM. The endoscopistmust be absolutely certain about the location of the biopsie speci-mens that are taken, ensuring that the entire area of potentialmetaplasia is adequately sampled. In addition, the endoscopistmust convey this information to the pathologist who will interpretthe histology. Including short-segment and ultra-short-segmentBarrett's esophagus, it is estimated that 80% of all cases of Barrett'sof all types are missed on routine endoscopy. One recommendationthat has been made to help overcome this shortcoming is that, forbasic surveillance, four-quadrant biopsies be taken at 2-cm inter-vals beginning within the proximal stomach (2 cm distal to the most

Page 6: Barrett’s esophagus: Histopathologic definitions and diagnostic criteria

Mueller et al.: Definition and Diagnosis of Barrett's Esophagus 153

proximal gastric folds), through the segment of suspected metapla-sia to the proximal stratified squamous epithelium, plus biopsies ofany macroscopically identifiable lesions [68, 69]. In cases in whicheither low-grade or high-grade dysplasia is identified, a second setof biopsies at intervals of 1 cm is recommended to avoid samplingerror, which might miss an area of high-grade dysplasia or carci-noma. Some authors are of the opinion that because the risk ofcarcinoma in CIM or ultra-short-segment Barrett's must be low,patients with a normal-appearing Z-line and esophagus should un-dergo biopsy, even though such testing would be needed to esti-mate the frequency of ultra-short-segment Barrett's esophagus.

Summary

The histopathologic definition of Barrett's esophagus in use todayis increasingly based on the histopathologic identification of colum-nar epithelium with goblet cells in the esophagus, regardless of itslength. Although this has simplified the diagnosis of Barrett'sesophagus, the question of the diagnostic criteria for and the clini-cal significance of short-segment, ultra-short-segment Barrett'sand intestinal metaplasia of the cardia are new issues that are notyet resolved. Although the metaplasia-dysplasia-carcinoma path-way is now firmly established and the guidelines for the clinicalmanagement of dysplasia and carcinoma are well defined, prob-lems remain with respect to the histopathologic detection and di-agnosis of dysplasia, as well as with the effective use of screeningprograms. These problems are highlighted by the high proportionof patients who present with locally advanced and metastatic tu-mors [3], and by the fact that few of these patients were previouslyknown to have had Barrett's esophagus. Much remains to be donebefore a significant impact on the rising incidence and overall poorprognosis of these tumors can be expected. An increased awarenesson the part of general practitioners and gastroenterologists that allpatients with reflux symptoms could have a potentially serious con-dition that needs to be investigated would be an important stepforward. In addition, new epidemiologic, clinical, and molecularbiologic data that is now accumulating will provide a better basis forthe estimation of risk in a given patient who has been found to haveclassical, short-segment or ultra-short-segment Barrett's esopha-gus or CIM.

Resume. L'incidence de l'adenocarcinome de l'oesophage distal augmenteplus rapidement que celle de toute autre affection maligne viscerale dans lemonde occidental. I1 est bien demontre que la plupart des tumeurs, sinontoutes, se developpent it partir de l'oesophage de Barrett en passant par lasequence metaplasie-dysplasie -carcinome et pourraient en theorie titredetectees a an stade precoce, mais malgre cela, la majorite de ces tumeurssont toujours detectees tardivement dans leer evolution. Ceci souligne lefait que Pefficacite de la surveillance chez le patient a risque de developperun adenocarcinome de l'oesophage distal est toujours loin d'être atteinte.De plus, les adenocarcinomes de la jonction esogastrique et du cardia sontaussi en hausse, mais leur carcinogenese et leur rapport a I'oesophage deBarrett ne sont toujours pas bien definis, de mime que de relativementnouvelles entites que sont la definition et la signification des oesphages deBarrett < court * (<< short-segment Barrett's ») et de Barrett << ultra-court(<< ultra-short-segment Barrett's *). Cette revue essaie de clarifier les pointsd'interets histopathologiques principaux en ce qui concerne la definition del'oesophage de Barrett, ce qui la distingue de la metaplasie intestinale ducardia gastrique, de mime que les criteres histologiques pour le diagnostic dedysplasie et de carcinome de l'oesophage de Barrett.

Resumen. En los paises industrializados la frecuencia del adenocarcinomadel esofago distal es cada vez mayor, sobrepasando ampliamente laincidencia de otras neoplasias malignas viscerales. Sc sabe que la mayoria,

si no todos estos adenocarcinomas se desarrollan a partir de un esofago deBarrett, de acuerdo con la conocida secuencia metaplasia-displasia-carcinoma y por consiguiente, teoricamente podrian ser diagnosticados enestadios precoces; sin embargo, la mayoria de estos tumores se detectan enestadios avanzados, to que implica el fracaso de una vigilancia eficaz deaquellos pacientes con riesgo a desarrollar un adenocarcinoma de esofagodistal. Tambien, se ha registrado un incremento de la frecuencia de losadenocarcinomas de la union esofagogastrica y del cardias gastrico, perosu carcinogenesis y su relacion con el esofago de Barrett todavia estan pordeterminar, como tambien ocurre con la interpretacion y transcendenciade entidades relativamente nuevas tales como "Barrett de segmento corto"y "Barrett de segmento ultra-corto ". En esta revision se intentan aclarar:los hechos histopatologicos mas importantes relacionados con lascaracteristicas del esofago de Barrett y sus diferencias con la metaplasiaintestinal del cardias gastrico y establecer criterios histologicos para eldiagnostico de displasia o carcinoma en el esofago de Barrett.

References

1. Blot WJ, Devesa SS, Kneller RW, et al. Rising incidence of adenocar-cinoma of the esophagus and gastric cardia. J.A.M.A. 1991;265:1287-1289

2. Pera M, Cameron AJ, Trastek VF, et al. Increasing incidence of adeno-carcinoma of the esophagus and esophagogastric junction. Gastroen-terology 1993;104:510-513

3. Hoff SJ, Sawyers JL, Blanke CD, et al. Prognosis of adenocarcinomaarising in Barrett's esophagus. Ann. Thorac. Surg. 1998;65:176-180

4. Siewert JR, Stein HJ. Barrett's cancer: indications, extent, and resultsof surgical resection. Semin. Surg. Oncol. 1997;13:245-252

5. Spechler SJ. The columnar-lined esophagus. History, terminology, andclinical issues. Gastroenterol. Clin. North Am. 1997;26:455-466

6. Stein HJ. Esophageal cancer: screening and surveillance. Dis. Esopha-gus 1996;9:3-19

7. Hamilton SR, Smith RR. The relationship between columnar epithelialdysplasia and invasive adenocarcinoma arising in Barrett's esophagus.Am. J. Clin. Pathol. 1987;87:301-312

8. Sampliner RE. Practice guidelines on the diagnosis, surveillance, andtherapy of Barrett's esophagus. The Practice Parameters Committee ofthe American College of Gastroenterology. Am. J. Gastroenterol.1998;93:1028-1032

9. Barrett NR. Chronic peptic ulcer of the oesophagus and "oesophagi-tis". Br. J. Surg. 1950;38:175-182

10. Swami S, Kumble S, Triadafilopoulos G. E-cadherin expression in gas-troesophageal reflux disease, Barrett's esophagus, and esophageal ad-enocarcinoma: an immunohistochemical and immunoblot study. Am. J.Gastroenterol. 1995;90:1808-1813

11. Hamelin R, Flejou JF, Muzeau F, et al. TP53 gene mutations and p53protein immunoreactivity in malignant and premalignant Barrett'sesophagus. Gastroenterology 1994;107:1012-1018

12. Yacoub L, Goldman H, Odze RD. Transforming growth factor-alpha,epidermal growth factor receptor, and MiB-1 expression in Barrett's-associated neoplasia: correlation with prognosis. Mod. Pathol. 1997;10:105-112

13. Hong MK, Laskin WB, Herman BE, et al. Expansion of the Ki-67 pro-liferative compartment correlates with degree of dysplasia in Barrett'sesophagus. Cancer 1995;75:423-429

14. Walch A, Specht K, Bink K, et al. Her-2/neu gene amplification, el-evated mRNA expression and protein overexpression in the metapla-sia-dysplasia-adenocarcinoma sequence of Barrett's esophagus. Lab.Invest. 2001;81:791-801

15. van Dekken H, Geelen E, Dinjens WN, et al. Comparative genomichybridization of cancer of the gastroesophageal junction: deletion of14Q31-32.1 discriminates between esophageal (Barrett's) and gastriccardia adenocarcinomas. Cancer Res. 1999;59:748-752

16. Walch AK, Zitzelsberger HF, Bruch J, et al. Chromosomal imbalancesin Barrett's adenocarcinoma and the metaplasia- dysplasia-carcinomasequence. Am. J. Pathol. 2000;156:555-566

17. Palanca-Wessels MC, Barrett MT, Galipeau PC, et al. Genetic analysisof long-term Barrett's esophagus epithelial cultures exhibiting cytoge-netic and ploidy abnormalities. Gastroenterology 1998;114:295-304

18. Menke-Pluymers MB, van Drunen E, Vissers KJ, et al. Cytogeneticanalysis of Barrett's mucosa and adenocarcinoma of the distal esopha-gus and cardia. Cancer Genet. Cytogenet. 1996;90:109-117

Page 7: Barrett’s esophagus: Histopathologic definitions and diagnostic criteria

154

World J. Surg. Vol. 28, No. 2, February 2004

19 Werner M, Mueller JD, Walch AK, et al. The molecular biology ofBarrett's esophagus. Histol. Histopathol. 1998;14:553-559

20. Mueller J, Werner M, Siewert JR. Malignant progression in Barrett'sesophagus: pathology and molecular biology. Recent Results CancerRes. 2000;155:29-41

21. Tileston W. Peptic ulcer of the esophagus. Am. J. Med. Sci. 1906;132:240-265

22. Allison PR, Johnstone AS. The oesphagus lined with gastric mucousmembrane. Thorax 1953;8:87-101

23. Paull A, Trier JS, Dalton MD, et al. The histologic spectrum of Bar-rett's esophagus. N. Engl. J. Med. 1976;295:476-480

24. Rothery GA, Patterson JE, Stoddard CJ, et al. Histological and histo-chemical changes in the columnar lined (Barrett's) oesophagus. Gut1986;27:1062-1068

25. Hayward J. The lower end of the esophagus. Thorax 1961;16:36-4126. Spechler Si, Zeroogian JM, Antonioli DA, et al. Prevalence of meta-

plasia at the gastro-oesophageal junction. Lancet 1994;344:1533-153627. de Mas CR, Kramer M, Seifert E, et al. Short Barrett: prevalence and

risk factors. Scand. J. Gastroenterol. 1999;34:1065-107028. Cameron AJ. Epidemiology of columnar-lined esophagus and adeno-

carcinoma. Gastroenterol. Clin. North Am. 1997;26:487-49429. Weinstein WM, Ippoliti AF. The diagnosis of Barrett's esophagus: gob-

lets, goblets, goblets. Gastrointest. Endosc. 1996;44:91-9530. Spechler SJ, Goyal RK. The columnar-lined esophagus, intestinal

metaplasia, and Norman Barrett. Gastroenterology 1996;110:614-62131. Winters C, Spurting Ti, Chobanian Si, et al. Barrett's esophagus. A

prevalent, occult complication of gastroesophageal reflux disease. Gas-troenterology 1987;92:118-124

32. Lewin K. Barrett's esophagus, columnar dysplasia, and adenocarci-noma of the esophagus. In Tumors of the Esophagus and Stomach,Washington, DC., Armed Forces Institute of Pathology, 1985;99-139

33. Stein HJ, Hoeft S, DeMeester TR. Functional foregut abnormalities inBarrett's esophagus. J. Thorac. Cardiovasc. Surg. 1993;105:107-111

34. Lee RG. Esophagus. In Sternberg, SS, editor, Diagnostic Surgical Pa-thology, New York, Raven Press, 1989;917-936

35. Riddell RH. The biopsy diagnosis of gastroesophageal reflux disease,"carditis," and Barrett's esophagus, and sequelae of therapy. Am. J.Surg. Pathol. 1996;20:531-S50

36. Shields HM, Zwas F, Antonioli DA, et al. Detection by scanning elec-tron microscopy of a distinctive esophageal surface cell at the junctionof squamous and Barrett's epithelium. Dig. Dis. Sci. 1993;38:97-108

37. Peck RM, Puspok A, Potzi R, et al. Histological findings after routinebiopsy at the gastro-oesophageal junction. Eur. J. Gastroenterol. Hepa-tol. 1999;11:1265-1270

38. Seitz G. Bioptische Differentialdiagnostik der gastroosophagealen Re-fluxkrankheit (GERD). In Kirchner T, editor, Verhandlungen derDeutschen Gesellschaft fiir Pathologie: 83. Tagung, Munich, Urbanand Fischer, 1999;27-36

39. Boch JA, Shields HM, Antonioli DA, et al. Distribution of cytokeratinmarkers in Barrett's specialized columnar epithelium. Gastroenterol-ogy 1997;112:760-765

40. Bettendorf U. Osophagus. In Remmele W, editor, Pathologie, Berlin,Springer-Verlag, 1984;85-140

41. Voutilainen M, Farkkila M, Juhola M, et al. Complete and incompleteintestinal metaplasia at the oesophagogastric junction: prevalences andassociations with endoscopic erosive oesophagitis and gastritis. Gut1999;45:644-648

42. Jass JR, Filipe MI. The mucin profiles of normal gastric mucosa, intes-tinal metaplasia and its variants and gastric carcinoma. Histochem. J.1981;13:931-939

43. Peuchmaur M, Potet F, Goldfain D. Mucin histochemistry of the co-lumnar epithelium of the oesophagus (Barrett's oesophagus): a pro-spective biopsy study. J. Clin. Pathol. 1984;37:607-610

44. Pera M, Trastek VF, Pairolero PC, et al. Barrett's disease: pathophysi-ology of metaplasia and adenocarcinoma. Ann. Thorac. Surg. 1993b;56:1191-1197

45. Sharma P, Morales TG, Sampliner RE. Short segment Barrett'sesophagus-the need for standardization of the definition and of en-doscopic criteria. Am. J. Gastroenterol. 1998;93:1033-1036

46. Spechler Si. Short and ultrashort Barrett's esophagus-what does itmean? Semin. Gastrointest. Dis. 1997a;8:59-67

47. Stein HJ, Siewert JR. Barrett-Osophagus_was ist gesichert? Chir. Gas-troenterol. 1992;8:200-205

48. Clark GWB, Ireland AP, Peters JH, et al. Short-segment Barrett'sesophagus: a prevalent complication of gastroesophageal reflux diseasewith malignant potential. J. Gastrointest. Surg. 1997;1:113-122

49. Weston AP, Krmpotich PT, Cherian R, et al. Prospective long-termendoscopic and histological follow-up of short segment Barrett'sesophagus: comparison with traditional long segment Barrett's esopha-gus. Am. J. Gastroenterol. 1997;92:407-413

50. Trudgill NJ, Suvarna SK, Kapur KC, et al. Intestinal metaplasia at thesquamocolumnar junction in patients attending for diagnostic gastros-copy. Gut 1997;41:585-589

51. Chandrasoma PT, Der R, Ma Y, et al. Histology of the gastroesopha-geal junction: an autopsy study. Am. J. Surg. Pathol. 2000;24:402-409

52. DeMeester SR, DeMeester TR. The diagnosis and management ofBarrett's esophagus. Adv. Surg. 1999;33:29-68

53. Spechler SJ. The role of gastric carditis in metaplasia and neoplasia atthe gastroesophageal junction. Gastroenterology 1999;117:218-228

54. El SH, Sonnenberg A, Jamal MM, et al. Characteristics of intestinalmetaplasia in the gastric cardia. Am. J. Gastroenterol. 1999;94:622-627

55. Hackelsberger A, Gunther T, Schultze V, et al. Intestinal metaplasia atthe gastro-oesophageal junction: Helicobacter pylori gastritis or gastro-oesophageal reflux disease? Gut 1998;43:17-21

56. Byrne JP, Bhatnagar S, Hamid B, et al. Comparative study of intestinalmetaplasia and mucin staining at the cardia and esophagogastric junc-tion in 225 symptomatic patients presenting for diagnostic open-accessgastroscopy. Am. J. Gastroenterol. 1999;94:98-103

57. Ormsby AH, Goldblum JR, Rice TW, et al. Cytokeratin subsets canreliably distinguish Barrett's esophagus from intestinal metaplasia ofthe stomach. Hum. Pathol. 1999;30:288-294

58. El-Zimaity HMT, Graham DY. Cytokeratin subsets for distingushingBarrett's esophagus from intestinal metaplasia of the cardia using en-doscopic biopsies. Lab. Invest. 2000.

59. Griffel LH, Amenta PS, Das KM. Use of a novel monoclonal antibodyin diagnosis of Barrett's esophagus. Dig. Dis. Sci. 2000;45:40-48

60. Lembo T, Ippoliti AF, Ramers C, et al. Inflammation of the gastro-oesophageal junction (carditis) in patients with symptomatic gastro-oesophageal reflux disease: a prospective study. Gut 1999;45:484-488

61. Reid BJ, Haggitt RC, Rubin CE, et al. Observer variation in the diag-nosis of dysplasia in Barrett's esophagus. Hum. Pathol. 1988a;19:166-178

62. UICC- International Union Against Cancer. TNM Classification ofMalignant Tumours, 5th edition, New York, John Wiley and Sons,1997;1-213

63. van den Boogert J, van Hillegersberg R, Siersema PD, et al. Endoscopicablation therapy for Barrett's esophagus with high-grade dysplasia: areview. Am. J. Gastroenterol. 1999;94:1153-1160

64. Stein HJ, Feith M, Muller J, et al. Limited resection for early adeno-carcinoma in Barrett's esophagus. Ann. Surg. 2000;232:733-742

65. Nigro JJ, Hagen JA, DeMeester TR, et al. Prevalence and location ofnodal metastases in distal esophageal adenocarcinoma confined to thewall: implications for therapy. J. Thorac. Cardiovasc. Surg. 1999;117:16-23

66. Takubo K, Sasajima K, Yamashita K, et al. Double muscularis mucosaein Barrett's esophagus. Hum. Pathol. 1991;22:1158-1161

67. Biddlestone LR, Barham CP, Wilkinson SP, et al. The histopathologyof treated Barrett's esophagus: squamous reepithelialization after acidsuppression and laser and photodynamic therapy. Am. J. Surg. Pathol.1998;22:239-245

68. Gossner L, May A, Stolte M, et al. KTP laser destruction of dysplasiaand early cancer in columnar-lined Barrett's esophagus. Gastrointest.Endosc. 1999;49:8-12

69. Levine DS, Haggitt RC, Blount PL, et al. An endoscopic biopsy proto-col can differentiate high-grade dysplasia from early adenocarcinomain Barrett's esophagus. Gastroenterology 1993;105:40-50

70. DeMeester TR. Surgical treatment of dysplasia and adenocarcinoma.Gastroenterol. Clin. North Am. 1997;26:669-684