CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2012;10:1232 1236 ALIMENTARY TRACT Effects of Dropping the Requirement for Goblet Cells From the Diagnosis of Barrett s Esophagus MARIA WESTERHOFF,* LINDSEY HOVAN, CHRISTINE LEE, and JOHN HART *Department of Anatomic Pathology, University of Washington Medical Center, Seattle, Washington; and Department of Pathology, University of Chicago Medical Center, Chicago, Illinois See editorial on page 1237. BACKGROUND & AIMS: The 2011 American Gastroenterological Association diagnostic criteria for Barrett s esophagus (BE) require the presence of goblet cells in biopsy specimens of columnar mucosa within the esophagus. In other countries, patients can be diagnosed with BE based on evidence of columnar epithelium, regardless of the presence of goblet cells. We examined the effects that a broader criteria would have on diagnoses of patients with endoscopically suspected BE. We also compared the clinical outcomes of patients with and without goblet cells in esophageal biopsy samples. METHODS: We analyzed the University of Chicago Medical Center database to identify 690 patients with no previous history of BE who underwent endoscopic biopsy analysis for BE from 1987 to 2008. We collected endoscopy reports, histology results, and chart reviews. RESULTS: Of biopsy specimens analyzed, samples from 258 patients contained goblet cells and 379 did not (53 of the biopsy samples contained only squamous mucosa). Patients whose biopsy samples contained goblet cells had longer endoscopic columnar segments (mean, 4.6 cm) and more biopsy specimens taken (mean, 5 biopsy specimens) than those without (mean length, 1.6 cm; mean, 4 biopsy specimens). Of patients whose biopsy specimens did not contain goblet cells, 35% underwent additional endoscopy; goblet cells continued to be absent in 88% of these (mean follow-up time, 5.8 y; 2.8 additional procedures; mean total biopsy specimens, 12). Goblet cells were identified in only 19% of all patients with columnar mucosa less than 2 cm. No patient without goblet cells developed adenocarcinoma. CONCLUSIONS: Decreasing the requirement for goblet cells would increase the diagnosis of BE by 147%. Among patients with short columnar segments, subsequent endoscopy generally does not reveal goblet cells, so the columnar mucosa might represent proximal stomach. Decreasing the requirement for goblet cells would cause many patients to be inaccurately labeled as BE. Keywords: Intestinal Metaplasia; Nongoblet Cell Mucosa; Esophageal Adenocarcinoma; Diagnostic and Prognostic Factors. In 1998, the American College of Gastroenterology issued practice guidelines for the diagnosis of Barrett s esophagus (BE). 1 In this document, Barrett s esophagus was defined by 2 criteria: (1) endoscopic evidence of a segment of columnar mucosa within the tubular esophagus, and (2) biopsy confirmation of intestinal metaplasia (goblet cells [GCs]) within the columnar segment. It was well known at that time that intestinal-type mucosa was not the only type of metaplastic columnar mucosa that could occur in Barrett s esophagus because gastric cardiac and fundic type of epithelium also had been documented in Barrett s segments. 2 However, intestinal metaplastic mucosa had been conventionally accepted as the only neoplastic precursor of esophageal adenocarcinoma. 3 5 The British Society of Gastroenterology subsequently published their criteria for the diagnosis of BE, and specifically omitted the requirement for biopsy confirmation of intestinaltype metaplasia (GCs). 6 Although the endoscopic requirements are similar to the previous American College of Gastroenterology guidelines, histologically, any type of columnar epithelium is regarded as acceptable to diagnose BE. They maintain that intestinal metaplasia can be missed by poor endoscopic sampling via inadequate numbers of biopsy specimens. Moreover, it is their contention that intestinal metaplasia always can be shown if enough biopsy specimens are taken over an adequate period of time. The controversy regarding the requirement for the presence of intestinal metaplasia has been addressed in the updated practice guidelines recently published by the American Gastrointestinal Association. 7 In this document, a distinction is made between the definition of BE and the criteria for the diagnosis of the condition. The definition of Barrett s has been broadened in the following manner, as follows:... the condition in which any extent of metaplastic columnar epithelium that predisposes to cancer development replaces the stratified squamous mucosa that normally lines the distal esophagus. 7 However, the guidelines go on to state that the diagnosis of BE currently requires biopsy confirmation of intestinal-type metaplastic epithelium because... intestinal metaplasia is the only type of esophageal columnar epithelium that clearly predisposes to malignancy. 7 This point is also a matter of debate because recent studies have contended that nonintestinal metaplastic Barrett s mucosa has the same cancer risk as that of intestinal type. 8 11 In the United States, broadening the criteria of BE to include all types of columnar mucosa would have a substantial impact Abbreviations used in this paper: BE, Barrett s esophagus; GC, goblet cell. 2012 by the AGA Institute 1542-3565/$36.00 http://dx.doi.org/10.1016/j.cgh.2012.05.013
November 2012 OVERDIAGNOSING BARRETT S ESOPHAGUS 1233 on medical costs and quality of life. The cost of life insurance for an otherwise healthy, nonsmoking man has been reported to have an average increase of 118% with a diagnosis of BE. 12 In a woman, this increase is averaged to be 177%. Most of such patients with BE live a normal life expectancy and die of causes other than esophageal adenocarcinoma. 13 A recent Dutch cohort study of more than 42,000 patients showed that the overall annual risk of cancer development was only 0.14%; low-grade dysplasia on initial endoscopic biopsy, older age, and male sex were the strongest predictors in progression to adenocarcinoma. 14 This retrospective analysis sought to determine how a broader histologic definition of BE would have affected the diagnosis of BE in patients with endoscopically identified columnar-lined mucosa. Second, we compared the neoplastic progression in these patients with and without GCs. Methods Patient information and tissue specimens were collected according to a protocol approved by the Institutional Review Board at the University of Chicago. Cases A search of a computerized pathology database for all adult patients between 1987 and 2008 identified 690 patients with the following specifications: (1) no previous diagnosis of BE (undergoing initial endoscopies), (2) endoscopically identified columnar-lined mucosa with clinical suspicion of BE (including 24 patients with irregular Z-lines), and (3) biopsy specimens obtained from the columnar-lined segment. Patients included 277 women and 413 men between the ages of 20 and 91 years. Pathology results and endoscopic information, including the length of the columnar segment and the number of biopsy specimens taken, were recorded. Chart review was performed on all patients to determine whether additional endoscopic procedures, esophagectomies, or mucosal resections were performed and for causes of death. Figure 1. Endoscopically measured length of columnar mucosa in the esophagus. Overall, the endoscopic length of columnar segments in patients with GCs were longer than in those without. The majority of patients without GCs had segments of only 1 or 2 cm in length. Table 1. Average Endoscopic Length and Number of Biopsy Specimens Taken Among Patients With and Without GC Patients with GC Patients without GC P value Average endoscopic length, cm 4.60 1.60.05 Average number of biopsy specimens 5.00 4.00.30 taken on initial endoscopy Statistics Statistical analysis was performed by using GraphPad Prism version 4.00 (GraphPad Software, San Diego CA) (2- tailed paired t tests) and Analyse-it for Microsoft Excel (version 2.20; Analyse-it Software, Ltd, http://www.analyse-it.com/, 2009) (Mann Whitney U test and the Fisher exact probability test). Results Including Nongoblet Cells in the Definition of Barrett s Esophagus Would Have Increased the Initial Diagnosis by 147% A total of 258 patients had GCs identified in their initial endoscopic biopsy specimens. In contrast, 379 patients (who each also had a columnar-lined esophagus identified on endoscopy) had columnar mucosa devoid of GCs on biopsy. In these patients the biopsy specimens generally revealed gastric cardia type mucosa, although in some, oxyntic mucosa also was present. There were also 53 patients who did not show any evidence of columnar mucosa on biopsy. In these patients, biopsy specimens taken from what was reported to be columnar lining had only squamous epithelium present and this group was separated from further comparison. The lengths of the endoscopically identified columnar segments were longer in those with GC than those without (P.001; Figure 1, Table 1). The number of biopsy specimens taken on the initial endoscopy was higher for those with GCs, but the difference was not statistically significant (P.3; Figure 2, Table 1). Based on the American College of Gastroenterology guidelines, 37% (258 of 690) of all the patients with endoscopically visualized columnar mucosa were diagnosed with BE. However, using the British guidelines, 92% (637 of 690) of these patients would have been diagnosed with BE, increasing the diagnosis of Barrett s esophagus by 147%. Majority of Patients Without Goblet Cells Continue to Lack Them on Subsequent Endoscopies A total of 163 patients with GCs and 133 patients without GCs underwent at least 1 repeat endoscopic procedure at the University of Chicago Medical Center. Among those who did not have GCs on initial endoscopic biopsy specimens, 15 patients (12%) eventually had GCs upon repeat endoscopies (Table 2). However, the remaining 118 patients (88%) continued to lack GCs. There was no statistical difference between these 2 groups in regard to the number of additional endoscopic procedures (P.5) or the number of additional biopsy specimens taken (P.26). The mean follow-up period for the group
1234 WESTERHOFF ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 10, No. 11 Figure 2. Number of biopsy specimens taken on initial endoscopic biopsy. The number of biopsy specimens taken on initial endoscopy was generally higher for those who had intestinal metaplasia; they also had longer segments. These initial endoscopies were for diagnostic purposes and, hence, did not necessarily follow surveillance protocol for biopsies. without GCs was 5.8 years; for the group with GCs, the mean number of years followed up was 4.9 (P.5). The length of the endoscopic columnar segment was the only significant difference between these 2 groups (P.001). Overall, GCs were identified in only 19% of all patients with columnar segments less than 2 cm. Patients With Goblet Cell Mucosa Develop Dysplasia and Cancer Progression to dysplasia and cancer was compared in all patients with GCs (N 178) with those without GCs (N 118) over a range of 3 to 25 years (mean, 4.8 y in patients with GCs, and 5.8 y in those without GCs) (Figure 3). Dysplasia developed in 14 patients with GCs (Table 3). However, among the patients without GCs, none developed dysplasia or adenocarcinoma. Of the 15 patients who originally lacked GC on initial biopsy but had GC detected in a follow-up endoscopy, none developed dysplasia or adenocarcinoma. Two patients in whom GCs had been identified developed adenocarcinoma. One of these patients had high-grade dysplasia on a previous biopsy, and the other patient was discovered to have adenocarcinoma on an endoscopy after a period of 10 years lost to surveillance from the original procedure that had diagnosed BE. Overall, 78% (11 of 14) of the patients with dysplasia had endoscopic segments longer than 3 cm. The shortest segment with dysplasia was 2 cm. Thus, having short segment BE was Figure 3. Kaplan Meier curve for progression to dysplasia. There is a statistically significant difference between patients with GCs and patients without. not completely without risk because 22% (3 of 14) of those who developed dysplasia had segments less than 3 cm. Also, none of the patients described as having irregular Z lines developed dysplasia or cancer. The majority of patients who developed dysplasia or cancer were men (71% of those who developed dysplasia; 100% of those who developed cancer) and aged in their late 50s (mean age, 56.5 y). Among the patients with dysplasia, the average follow-up period was 10.13 years. Of the patients with low-grade dysplasia, 63% regressed upon follow-up evaluation (5 of 8) and 37% remained the same grade (3 of 8). Of the patients with high-grade dysplasia, 17% progressed to cancer (1 of 6), 33% regressed (2 of 6), and 50% remained the same grade (3 of 6). There were 2 deaths among the group without GCs, but none were related to esophageal diseases. Among 3 deaths in the patients with GCs, one was related to esophageal cancer, but was secondary to the esophagectomy and not the tumor itself. Also, separate from the rest of this study, there were 10 patients with prevalent dysplasia and 2 patients with cancer discovered on initial endoscopy. All of these patients had endoscopic columnar segments longer than 3 cm (average, 6.2 cm) and had GCs present in their biopsy specimens. Discussion This study addressed the potential impact of broadening the criteria of BE. Our findings show that broadening the criteria of BE by decreasing the requirement for GCs increases the percentage of patients who would have been labeled as having BE from 37% to 92%. However, based on both chart Table 2. The Average Number of Additional Procedures and Biopsy Specimens Taken in Patients With No GCs on Initial Biopsy, Who Underwent Subsequent Endoscopy No GCs on subsequent biopsy specimens (n 118) GCs identified on subsequent biopsy specimens (n 15) P value Average number of additional endoscopic procedures 2.80 2.07.50 Average number of additional biopsies 7.00 6.20.26 Average years of follow-up evaluation 5.80 4.90.50 Average endoscopic length of columnar mucosa, cm 1.60 4.10.001 NOTE. There were 133 patients with no GCs on initial biopsy specimens who underwent follow-up endoscopy.
November 2012 OVERDIAGNOSING BARRETT S ESOPHAGUS 1235 Table 3. Neoplastic Outcome of Patients With No GCs vs All Patients With GCs No GCs (n 118) GCs (n 178) P value Average number of additional 2.80 2.50.50 endoscopic procedures Average years of follow-up 5.80 4.80.07 evaluation Progression to dysplasia, n 0 14.00.00 Progression to adenocarcinoma, n 0 2.00.16 review and follow-up endoscopy, the patients lacking GCs did not develop dysplasia or esophageal adenocarcinoma. There are several possible explanations for the group of patients with endoscopic and histologic columnar mucosa but without intestinal metaplasia. First, the columnar mucosa lacking GCs could represent proximal gastric mucosa because it is possible that endoscopists occasionally mistake gastric cardia as a segment of BE. In our study, only 19% of all patients with segments less than 2 cm had GCs. In one study, 84 trained endoscopists evaluated a set of 30 endoscopic cases and were asked to locate where the esophagus ended. 15 By using a Japanese endoscopic criteria where the distal end of the esophageal vessels demarcates where the esophagus ends, an interobserver value of only 0.17 was achieved, whereas by using the Prague criteria an interobserver value of 0.35 was observed. This study illustrates the difficulty in accurately identifying the esophagogastric junction endoscopically in patients with possible short-segment Barrett s esophagus. In such cases, decreasing the requirement for GCs from the definition of BE may erroneously increase the diagnosis based on biopsy specimens of columnar mucosa from the proximal stomach. The second possibility is that the group without GCs may represent poor sampling during endoscopy. In our study, 12% of the patients who originally lacked GCs were found to have them on subsequent endoscopy. These patients either had GCs initially that were missed by poor sampling the first time around, or developed GCs during the course between repeat endoscopies. However, in contrast to the British Society of Gastroenterology, which reasons that intestinal metaplasia always can be shown in patients with esophageal columnar mucosa over time, 88% of our patients without GCs continued to lack them subsequently. They had comparable numbers of endoscopic procedures, numbers of biopsy specimens, and years of follow-up evaluation as the group with GCs. These results are similar to the findings of data from an Irish population that concluded that GCs seem to either be present from the start or absent. 16 Finally, the columnar mucosa in the patients lacking GCs may represent metaplastic mucosa without intestinal metaplasia that is reported to harbor precancerous abnormalities. 11,17 Although a recent study has shown DNA content abnormalities from biopsy specimens of non-gc columnar mucosa in patients with irregular Z-lines, none of the patients in our study with irregular Z-lines developed dysplasia or cancer. 11 This raises the question of whether abnormalities in DNA content are clinically significant in nondysplastic mucosa that likely is chronically inflamed. Our finding of an absence of risk for dysplasia or cancer in patients without GCs is in agreement with the previously mentioned Irish data (2969 patients), in which the risk of esophageal adenocarcinoma in patients with non-gc mucosa was found to be similar to that of the general population. 16 Our findings are, on the other hand, in direct contrast to reports from the United Kingdom, which concluded that there is no difference in cancer development among patients with or without GCs. 10 These studies, however, were based largely on chart review, whereas our report also analyzed biopsy specimens from follow-up endoscopies. For example, the article by Kelty et al 10 categorized their patients into those with and without GCs by reviewing index biopsy specimens, of which there was an average of only 4. They excluded cases of patients who had undergone follow-up endoscopies with further biopsies. There was also no mention of endoscopic segment length. They then followed up patients outcomes by chart review and cancer registry data. It is possible that their results may be based on poor sampling and that the adenocarcinoma that developed from patients without GCs were in fact reflective of poor sampling. In summary, this study of American patient data analyzed the neoplastic outcome of patients with and without GC. It also quantifies the marked increase in the diagnosis of BE that would occur if the broader criteria for BE was adopted. An alternative approach would be to target patients who are at highest risk of cancer development, such as dysplasia on initial endoscopic biopsy, male sex, and older age, which may aid in protocols for endoscopic surveillance. References 1. Sampliner RE. Practice guidelines on the diagnosis, surveillance, and therapy of Barrett s esophagus. The Practice Parameters Committee of the American College of Gastroenterology. Am J Gastroenterol 1998;93:1028 1032. 2. Paull A, Trier JS, Dalton MD, et al. The histologic spectrum of Barrett s esophagus. N Engl J Med 1976;295:476 480. 3. Theisen J, Stein HJ, Dittler HJ, et al. Preoperative chemotherapy unmasks underlying Barrett s mucosa in patients with adenocarcinoma of the distal esophagus. Surg Endosc 2002;16:671 673. 4. Chandrasoma P, Wickramasinghe K, Ma Y, et al. Is intestinal metaplasia a necessary precursor lesion for adenocarcinomas of the distal esophagus, gastroesophageal junction and gastric cardia? Dis Esophagus 2007;20:36 41. 5. Krishnadath KK. Novel findings in the pathogenesis of esophageal columnar metaplasia or Barrett s esophagus. Curr Opin Gastroenterol 2007;23:440 445. 6. Playford RJ. New British Society of Gastroenterology (BSG) guidelines for the diagnosis and management of Barrett s oesophagus. Gut 2006;55:442. 7. American Gastroenterological Association, Spechler SJ, Sharma P, et al. American Gastroenterological Association medical position statement on the management of Barrett s esophagus. Gastroenterology 2011;140:1084 1091. 8. Chaves P, Crespo M, Ribeiro C, et al. Chromosomal analysis of Barrett s cells: demonstration of instability and detection of the metaplastic lineage involved. Mod Pathol 2007;20:788 796. 9. Gatenby PA, Ramus JR, Caygill CP, et al. Relevance of the detection of intestinal metaplasia in non-dysplastic columnar-lined oesophagus. Scand J Gastroenterol 2008;43:524 530. 10. Kelty CJ, Gough MD, Van Wyk Q, et al. Barrett s oesophagus: intestinal metaplasia is not essential for cancer risk. Scand J Gastroenterol 2007;42:1271 1274. 11. Liu W, Hahn H, Odze RD, et al. Metaplastic esophageal columnar epithelium without goblet cells shows DNA content abnormalities
1236 WESTERHOFF ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 10, No. 11 similar to goblet cell-containing epithelium. Am J Gastroenterol 2009;104:816 824. 12. Shaheen NJ, Dulai GS, Ascher B, et al. Effect of a new diagnosis of Barrett s esophagus on insurance status. Am J Gastroenterol 2005;100:577 580. 13. Sikkema M, de Jonge PJ, Steyerberg EW, et al. Risk of esophageal adenocarcinoma and mortality in patients with Barrett s esophagus: a systematic review and meta-analysis. Clin Gastroenterol Hepatol 2010;8:235 244; quiz, e32. 14. de Jonge PJ, van Blankenstein M, Looman CW, et al. Risk of malignant progression in patients with Barrett s oesophagus: a Dutch nationwide cohort study. Gut 2010;59:1030 1036. 15. Amano Y, Ishimura N, Furuta K, et al. Which landmark results in a more consistent diagnosis of Barrett s esophagus, the gastric folds or the palisade vessels? Gastrointest Endosc 2006;64:206 211. 16. Murphy SJ, Johnston BT, Murray LJ. British Society of Gastroenterology guidelines for the diagnosis of Barrett s oesophagus: are we casting the net too wide? Gut 2006;55:1821 1822. 17. Riddell RH, Odze RD. Definition of Barrett s esophagus: time for a rethink is intestinal metaplasia dead? Am J Gastroenterol 2009;104:2588 2594. Reprint requests Address requests for reprints to: Maria Westerhoff, MD, 1959 NE Pacific Street, Seattle, Washington 98195. fax: (206) 598-3803. e-mail: mariawesterhoff@gmail.com Conflicts of interest The authors disclose no conflicts. Funding Supported by the University of Chicago Department of Pathology.