Barrett's esophagus: Its prevalence and association with adenocarcinoma in patients with symptoms of gastroesophageal reflux

Barrett's Esophagus: Its Prevalence and Association With Adenocarcinoma In Patients With Symptoms of Gastroesophageal Reflux Michael G. Sarr, MD, Baltimore, Maryland Stanley R. Hamilton, MD, Baltimore, Maryland Gary C. Merrone, MD, Baltimore, Maryland John L. Cameron, MD, Baltimore, Maryland

The presence of columnar epithelium in the lower esophagus has stimulated considerable interest since its original description by Barrett [1] in 1950. More recently, the apparent association of adenocarcinoma and Barrett's esophagus has led many to consider this columnar epithelium as a premalignant condition. The prevalence of Barrett's esophagus and its association with adenocarcinoma in various patient populations needs to be determined. Therefore, we studied patients whom we could identify as being at risk for Barrett's esophagus (that is, patients with symptoms of gastroesophage~alreflux deemed severe enough by their gastroe~erologists to warrant esophagoscopy and biopsy). The aims of our study were threefold: to determine the prevalence of Barrett's esophagus in this population, to compare patients with and without Barrett's esophagus for the clinical and objective findings of gastroesophageal reflux, and to compare patients with Barrett's esophagus with and without adenocarcinoma. Clinical M a t e r i a l

From 1975 to 1981, 1,020 esophageal biopsies in patients with symptoms compatible with gatroesophageal reflux were identifiedin the aurgic~ pathology filesof The Johns Hopkins Hospital. Symptoms of gastroesophageal reflux included regurgitation,heartburn, epigastricdistress,and dysphagia. Subsequent biopsieson the same patient in any given year were excluded; however, no attempt was made to exclude the same patient ifrebiopsied in another year. Once a patient was found to have Barrett's esophagus, future esophageal biopsies on the patient were excluded. Patients with squamous carcinoma of the esophagus, adenocarcinoma of the stomach, and infectiousesophagitis were excluded. Mean age was 50 years, ranging from 2 to 86 years.Men comprised 48 percent of the group. The main Frome~e~ of Surge~andPae~o~o~,TheJOhmHor~s Medk~ InstRu~ons.B~dtimors.~ . P,eq~JeStsfor repr|rds8houkJbe addres.~edtOMichaelSerf,kiD,~ mereo! Surgery,l~kx~, 6~, JohnsHopklr4Hos¢~, E~Irnoce, 21205.

Presentedat ~ 25zh ~ Merino of ~ Societyfor Sur0eryof the A I ~ Tract, New Orleans,Louislan~May 22-23, 1984.

¥~

145,January1SOS

symptom of gastroesophageal reflux prompting esophagoscopy was epigastric distress (31 percent of patienL~), heartburn (23 percent of patients), regurgitation (23 percent of patients), and dysphagia (23 percent of patients).

For each biopsy, the surgical pathology records were reviewed in conjunctionwith the endoscopicreport. The diagnosis of Barrett's esophagus was based on both the endoscopic description and the histopathologic findings. The distinctive, specialized type of Barrett's mucosa showed villiformcolumnar-lined mucosa with goblet cells characteristically interspersed among columnar mucous cells. Mucosa of the cardiac type and fundic type histopathologically resembling mucosa of the stomach was interpreted as Barrett's mucosa only when the biopsy specimen was stated clearly by the endoscopist to be from the esophagus and not from a hiatus hernia. Patients with equivocal histopatho!ogic or endoscopic findings were excluded from the Barrett's group. The 362 patients who underwent esophageal biopsy in the last2 years of the study (1980 and 1981) were analyzed in greater detail.The duration of symptoms, the history of previous gastricor esophageal surgery, and the presence of achalasia,scleroderma, a hiatal hernia,or an esophageal stricture se~n on endoscopy were recbrded. Moreover, in addition to the gross findings at esophagoscopy, the results of objective tests of esophageal function, includil~g esophageal reflux of barium on upper gastrointestinalseries,the Bernstein test,lower esophageal sphincter pressure on esophageal manometry, and the standard acid reflux test were noted. Statisticalcomparisons of patients with and without Barrett's esophagus were performed using chi-square analysis. During the years 1975 to 1981, 13 patients undergoing esophagoscopy for symptoms compatible with gastroesophageal reflux showed adenocarcinoma arising in a Barrett's esophagus either from an esophageal biopsy specimen or from a subsequent surgical specimen. The majority of these patients were believed initially,by find. ings on upper gastrointestinalseries,endoscopy, or both, to have gastric adenocarcinoma of the cardia or of the fundus; only on review of the pathologic specimens was adenocarcinoma found to have arisen in a segment of Barrett's esophagus. The prevalence of adenocarcinoma of the esophagus in patients with Barrett's esophagus and

167

Sarr et al

TABLE I

Prevalence of Barrett's Esophagus In Biopsy Specimens From Patients With Symptoms of Gastroesophagesl Reflux"

Year

n

1975 1976 1977 1978 1979 1980 1981 Total

7/114 13/115 5/130 5/133 10/166 24/171 20/191 84/1020

Preval?nce -

% 6 11 4 4 6 14 11 8

• Regurgitation, heartburn, eplgastrlc distress, and dysphagia.

symptoms compatible with gastroesophageal reflux were calculated for the study period. Moreover, these patients with adenocarcinoma were analyzed in depth and were compared with the group of 39 patients with Barrett's esophagus without carcinoma seen from 1980 to 1981. Because only five patients with coexistent Barrett's esophagus and adenocarcinoma were seen from 1980 to 1981, all 13 patients from-1975 to 1981 ~vere included to

TABLE II

increase the number and make the comparison more meaningful.

Results .The overall prevalence of Barrett's esophagus in patients undergoing esophagoscopy and biopsy for evaluation of symptoms compatible with gastroesophageal reflux was 8 percent (84 of 1,020 patients). The yearly prevalence of Barrett's esophagus ranged from 4 to 14 percent (Table 2), In 1980 and 1981, 362 esophageal biopsies were performed in patients with symptoms compatible with gastroesophageal reflux. Forty-four of these patients (12 percent) had Barrett's esophagus identified histopathologically. The mean age of the patients with Barrett's esophagus (56 years) was not statistically different than the age of patients without Barrctt's esophagus (51 years) (Table II). The sex distributions and race distributions were similar. When the main symptoms of gastroesophageal reflux prompting esophagoscopy were compared, patients with Barrett's esophagus had a higher prevalence of dysphagia (34 percent versus 17 percent, p <0.05) and a lower prevalence of epigastric distress (11

Profile of Patients With Symptoms of Esophageal Reflux Undergoing Esophageal Biopsy: 1980 Through 1981" Patients Without Barrett's Esophagus (n = 318) '

Patients With Barrett's Esophagust (n = 44)

52

51

56

48 52

47 53

52 48

83/ 17

81 19

91 9

24 32 25 19

24 32 27 17

25 30 11 34

28 30 42 49 8 2 6 26 3 2

27 31 42 48 4 <1 4 6 3 1

36 25 39 70 31 11 20 14 0 9

All Patients (n --- 362) Mean age (yr) Sex Male Female Race White Black Chief symptom RegUrgitation Heartburn Eplgastrlc distress Dysphagia Duration of syml~torns < 1 year 1-5 years > 5 years Hiatal hernia Esophageal stricture Mid esopl~Lgus Distal esophagus Esophageal ulceration Achalasia ScleroderrrB Previous surgery Gastric Esopha~leal . . . . . . . . . . . . . . . . . . . . . . .

7,4 101

7 8

................

5 7

* All values other than mean age are porcentages. t Values for eplgastrlc distress, dysphagla, hla~al hernia, and esophageal stricture were-all significantly different from the.values In patients without Barrett's esophagus (p <0.05). t Hemlgnstrectomy (24 patients) and vagotomy and wloroplasty (2 patients). I Esophagogastrectomy (5 patients), Holler esophagogastrlc myotomy (3 patients) and antlreflux procedures (28 patients).

4ss

~ ~Jmm~

~ ¢,Wtmv

Barrett's Esophagus, Adenocarcinoma, and Gastroesophageal Reflux

percent versus 27 percent, p <0.05). The prevalence of regurgitation (25 percent versus 24 percent) and heartburn (30 percent versus 32 percent), and the duration of symptoms were not significantly different (Table II). Patients with Barrett's esophagus more commonly had a hiatal hernia (70 percent versus 48 percent, p <0.01), an esophageal stricture (31 percent versus 4 percent, p <0.01), or both (Table II). Mid esophageal strictures were associated almost exclusively with Barrett's esophagus (five of six patients). Eleven patients were found to have achalasia, none of whom had Barrett's esophagus. In contradistinction, eight patients had scleroderma, four of whom had Barrett's esophagus. Previous gastric or esophageal surgery had no apparent effect on the presence or absence of Barrett's esophagus (Table II); however, the total number of such patients was too small for meaningful comparison. The esophagoscopist suspected Barrett's esophagus in only 34 percent of those patients found to have Barrett's esophagus on esophageal biopsy (Table III). Two patients (5 percent) had what was interpreted endoscopically as a normal appearing esophageal mucosa. Erythema of the esophageal mucosa was a more frequent endoscopic finding in patients with Barrett's esophagus when compared with those without Barrett's esophagus (84 percent versus 63 percent, p <0.05). The routine clinical tests of gastroesophageaI reflux and esophageal function were compared in patients with and without Barrett's esophagus (Table IV). Gastroesophageal reflux of barium on upper gastrointestinal series was found in 28 percent of the patients with Barrett's esophagus and in 29 percent of the patients without Barrett's esophagus. When performed, results of the Bernstein test were positive in two thirds of the patients in both groups. All 11

TABLE IV

TABLE III

Esophagoscoplc Findings in Patients With Symptoms of Gastroeaophageal Reflux: 1980 Through 1981

Principle Findings Normal Erythema Ulcers Pallor Barrett's esophagus Stricture

Prevalence (%) Patients ~ithout Barrett's All Patients Esophagus (n = 362) (n = 318)

Patients Wi~ Barrett's Esophagus" (n -- 44)

14 65 7 21 5

16 63 6 23 1

5 84 14 7 34

8

4

31

" Values for findings of erythema, pallor, Barrett's esophagus, and stricture were all significantly different from the values In patients without Barrett's esophagus (p <0.05).

patients with Barrett's esophagus had abnormally low lower esophageal sphincter pressures {less than 10 mm Hg), whereas only 32 of 60 patients (53 percent) without Barrett's esophagus had a hypotonic lower esophageal sphincter. This represents a statistically greater prevalence of hypotonic lower esophageal sphincter in patients with Barrett's esophagus (p <0.0I). Reflux on the standard acid reflux test was found in the three patients with Barrett's esophagus who were studied and in 17 of 29 patients without Barrett's esophagus. Because of the small number of patients studied, this difference lacked statistical significance. From 1975 to 1981, 13 patients with symptoms compatible with gastroesophageal reflux were found to have adenocarcinoma of the esophagus arising in Barrett's mucosa. During this time interval, a total of 84 patients with both Barrett's esophagus and

Objective Tests of Esophageal Function in Patler'ts With Symptoms of Gastroesophageal Reflux (GER) Undergoing Esoph/~goscopy °

Test Upper GI series GER No GER Bernsteln test Positive results Negative results Esophageal manomeb'y Low LES (<10 mm Hg) Normal LES (>10 mm Pig) Standard acid reflux text Reflux No reflux

Patients Without Barrett's Esophagus (n : 318) 211

Patients With Barrett's Esophagus (n = 44) 32

261 (29) 150 (71) 58

9 (28) 23 (72) 6

38 (65) 20 (34) 60

4 (67) 2 (33) 11

32 (53) 28 (47) 29

11 (100)t 0... 3

17 (59) 12 (41)

~ (100) O...

• Values In parentheses are percentages. t Values significantly different from thOse of patients without Barrett's esophagus (p <0.05). LIES == lower esophageal sphincter. Volume 149, J ~ a r y 1915

18g

Sarr et al

TABLE V

Comparison of Patients With Barrett's Esophagus With (1975 through 1981) and Without (1980 through 1981) Adenocarclnoma" BarroWs Esophagus Without Adenocarctnoma (n = 39)

Moan age (yr) Sex Male Female Race White Black Chief symptom Regurgitation Heartburn Eplgastrlc distress Dysphagla Duration of symptoms < 1 year 1-5 years > 5 years Hiatal hernia Esophageal ulceration Previous surgeryl Gastric Esopha~ea! . . . . . . . . . . . . .

Barrett's Esophagus & Adenocarotnomat t (n = 13)

56

57

51 49

77 23

90 20

100 0

28 31 10 31

8 8 15 69

34 28 38 69 13

67 13 25 31 10

0 5. . . . . . . . . . . . . . . . . . .

8 8

• All values other than mean ego are percentages. Vatues for percentage of male patients and less than 1 year duration of symptoms were statistically different from the values in patients with Barrett's esophagus without adenocarclnoma (p <0,I). Values for dysphagla and more than 5 years duration of symptoms were statistically different from the values in patients with Barrstt's esophagus without adenocarcinoma (p <0.05). ¶ Previous gastric surgery Included hemlgastrectomy in two patients, one of whom had adenocarcinoma; previous esophageal surgery Included antireflux procedures In three patients, one of whom had adenocarcinoma, and one other patient had a colonic interposition.

symptoms of gastroesophageal reflux were seen. Thus, 15 percent of the symptomatic patients with Barrett's esophagus (13 of 84 patients) had a coexistent adenocarcinoma arising from their Barrett's mucosa. The 13 patients with adenocarcinoma and Barrett's esophagus seen between 1975 and 1981 differed in several respects from the 39 patients with Barrett's esophagus but without associated adenocarcinoma seen between 1980 and 1981 (Table V). These 13 patients tended to more often be male (77 percent versus 51 percent, p = 0.1). They more commonly presented with dysphagia (69 percent versus 31 percent, p <0.05), and tended to have a shorter duration of symptoms (less than 1 year, 67 percent versus 33 percent, p = 0.57). Hiatal hernia was less common in the 13 patients with adenocarcinoma (31 percent versus 69 percent, p <0.05). 190

Comments

Since the original description of Barrett's esophagus over 30 years ago [1], there has been debate over its pathogenesis, significance, and treatment. Many investigators believed that the columnar-lined lower esophagus ~represented failure of the columnar epithelium of the embryonic esophagus to regress completely as usually occurs during gestation [2-4]. However, present evidence strongly supports the acquired nature of Barrett's esophagus. Experimental work by Bremner et al [5] has demonstrated columnar regeneration of the denuded canine distal esophagus in the presence of gas~roesophageal reflux. In patients with Barrett's esophagus, manometric studies have shown a hypotonic lower esophageal sphincter, and 24 hour pH monitoring of the distal esophagus has demonstrated marked gastroesophageal reflux [6,7]. Moreover, several investigators have observed temporally the orad extension of columnar epithelium in patients with Barrett's esophagus [8-10]. All these studies suggest that Barrett's esophagus represents a columnar metaplasia of the lower esophagus in response to chronic acid-peptic reflux [I0-12]. Nevertheless, many questions remain unanswered about its prevalence and natural history. In patients undergoing esophagoscopy for symptoms compatible with gastroesophageal reflux, we found a prevalence of Barrett's esophagus of 8 percent (84 of 1,020 patients). Our prevalence compares closely with that of Naef and colleagues [I1] who found a prevalence of 11 percent (140 of 1,225 patients) in a group selected with endoscopic changes of peptic esophagitis. The prevalence of Barrett's esophagus in patients without symptoms of gastroesophageal reflux is unknown. Of interest, however, were six additio~nal patients found over the same time interval of our study (1975 through 1981) to have Barrett's esophagus on endoscopy performed for other unrelated symptoms. Because these six patients had no symptoms of gastroesophageal reflux, they were not included in our study. This observation demonstrates clearly that not all patients with Barrett's esophagus have symptoms of gastroesophageal reflux. Our patients were purposely selected to include a population that could be identified to be at risk for Barrett's esophagus (that is, those patients with symptoms consistent with gastroesophageal reflux). Furthermore, by selecting this population, we hoped to determine which symptoms or clinical findings were associated more commonly with Barrett's esophagus and thereby define which subgroup of patients should be evaluated more closely. Although the clinical complex of dysphagia, hiatal hernia, and esophageal stricture within our overall group of patients with gastroesophageal reflux tended to differentiate.the patients with Barrett's esophagus, these findings alone were of little specificity because The Amedcan Jo~nal o( Surgery

BarroWs Esophagus,Adenocarcinoma,and GastroesophagealReflux

of their frequent occurrence. No symptom of gastroesophageal reflux or associated clinical finding reliably selected those patients with Barrett's esophagus or excluded those without Barrett's esophagus. Patients with Barrett's esophagus, when subjected to the routine clinical tests for gastroesophageal reflux and esophageal function, were often found to have objective findings of esophageal dysfunction. Esophageal reflux on either upper gastrointestinal series or on standard acid reflux testing or a positive Bernstein test result failed to reliably differentiate those patients with Barrett's esophagus from those without it. As suggested by others [7,13,14], a hypotensive lower esophageal sphincter was a frequent finding in patients with Barrett's esophagus, but it also occurred in about half of those patients with symptoms of gastroesophageal reflux without Barrett's esophagus (Table IV). One of the most important findings from this study was the frequency with which Barrett's esophagus was not recognized by esophagoscopy. Only 34 percent of the patients with Barrett's esophagus were suspected of having this condition by the endoscopist; therefore, we believe that all patients with symptoms of gastroesophageal reflux subjected to esophagoscopy should have esophageal biopsies performed routinely in an effort to identify all those with Barrett's esophagus. Numerous reports have suggested an association between Barrett's esophagus and adenocarcinoma of the esophagus. This apparent association was recognized both by examination of patients with Barrett's esophagus [4,11,12,15,16] and by pathologic review of patients with presumed primary adenocarcinoma of the esophagus or gastric cardia [17-19]. Allison and Johnstone [12] were the fLrst to describe esophageal adenocarcinoma arising from the columnar epithelium of a patient with recognized Barrett's esophagus. Some patients with Barrett's esophagus can be found to have dysplasia of Barrett's mucosa [20,21]. Similarly, Haggitt et al [17] found degrees of dysplasia ranging from slight to carcinoma in situ in 10 of 12 patients with adenocarcinoma arising from Barrett's esophagus. Dysplasia occurred not only adjacent to the hlvasive tumor mass but also in remote areas of Barrett's mucosa. Similar histopathologic changes of dysplasia have also been recognized in several of our patients. The reported prevalence of adenocarcinoma in Barrett's esophagus has ranged from 0 to 15 percent [4,11,12,14-16,22]. We found a prevalence of 15 percent in patients with Barrett's esophagus who had symptoms consistent with gastroesophageal reflux. As in other reports [I5,22,23], male patients predominated (10 of 13 patients). Dysphagia was the most common symptom and was usually of short duration (less than 1 year). Dysphagia was probably Vokmw 149, JamNule')r1985

related to the development of the tumor mass and not secondary to gastroesophageal reflux. There are little data on the temporal development of dysplasia or adenocarcinoma in Barrett's esophagus [24]. At the present time, we concur with other groups who recommend regular surveillance of patients with known Barrett's esophagus [15,17,22,23,25), although the risk may be present over a period of decades. Surveillance should include not only esophagoscopy with routine esophageal biopsies, but also esophageal cytologic examination, which may be a more sensitive screening technique [26]. The question of appropriate treatment of patients with Barrett's esophagus remains unresolved. Some investigators have claimed at least partial regression of Barrett's epithelium after successful antireflux procedures [8,10,13,22]; others have not seen any regression [27]. Although we and others [13,22] advocate an antireflux procedure for most patients with Barrett's esophagus, it is not known whether the occurrence of adenocarcinoma can be prevented. Similarly, should dysplasia be recognized in tile columnar epithelium of Barrett's esophagus, it remains unproved whether or not prophylactic esophagectomy is warranted. Summary

The pathologic reports of all 1,020 esophageal biopsy specimens obtained between 1975 and 1981 in patients with symptoms of gastroesophagea| reflux were reviewed. Barrett's esophagus was identified in 84 patients (8 percent). The 362 patients seen between 1980 and 1981 were reviewed in detail. The symptoms in patients with Barrett's esophagus differed from those of the patients without Barrett's esophagus. Dysphagia was more often present in the former group (34 percent versus 16 percent, p <0.05) and epigastric distress was less frequent (11 percent versus 27 percent, p <0.05). Objective findings of hiatal hernia, esophageal stricture, and esophageal ulcers occurred more commonly in patients with Barrett's esophagus than in those without Barrett's esophagus (70 percent versus 48 percent, 31 percent versus 4 percent, and 14 percent versus 6 percent, respectively, p <0.05). Mid esophageal strictures were associated almost exclusively with Barrett's esophagus (five of six patients). At esophagoscopy, erythema was seen more commonly with Barrett's esophagus. The diagnosis was suspected by the endoscopist in only 34 percent of patients subsequently demonstrated histopathologically to have Barrett's esophagus. There was no significant difference in the prevalence of a positive Bernstein test result or gastroesophageal reflux on upper gastrointestinal series in patients with and without Barrett's esophagus. However a hypotensive lower esophageal sphincter was found more commonly in patients with Barrett's esophagus (100 percent versus 53 percent, p <0.05). Thirteen of the 84 patients with Barrett's esophagus 191

San" et al

(15 percent) had a coexistent adenocarcinoma arising from Barrett's mucosa. These patients, when compared with the patients with Barrett's esophagus without carcinoma, were more often male (77 percent versus 51 percent, p = 0.1), more often had dysphagia (69 percent versus 34 percent, 1~ <0.05), and more frequently had a comparatively short duration of symptoms (67 percent versus 36 percent, p <0.05). Our findings suggest that patients with Barrett's esophagus have a high risk of development of carcinoma. Because the entity is often not recognized at endoscopy, routine esophageal biopsy should be performed on all patients undergoing esophagoscopy for symptoms of gastroesophageal reflux. Patients with known Barrett's esophagus should be followed closely with repeated endoscopy and biopsy. Acknowledgment: We thank Drs. Frank Milligan and Marshall Bedine for their help in providing the medical records of patients included in this study. References 1. Barrett NR. Chronic peptic ulcer of the oesophagus and "cesophagitts." Brit J Surg 1950;38:175-82. 2. Barrett NR. The lower esophagus lined by columnar epithelium. Surgery 1957;41:881-94. 3. Wolf BS, Marshak RH, Som ML. Peptic esophagltis and peptic ulceration of the esophagus. Am J Roentgenol Radlol Ther Nucl Meal 1958;79-741-59. 4. Naef AP, Savary M. Conservative operations for peptic esoph~gltis with stenosls In COlumnar-Ilnadlower esophagus. Ann Thorac Surg 1972;13:543-51. 5. Brenner CG, Lynch VP, Ellis FH Jr. Barrett's esophagus: Congenital or acquired? An experimental study of eosphageal mucosal regeneration in the dog. Surgery 1970;68:20916. 6. Cohen BR, Wolf BS, Som M, Janowltz HD. Correlation of manometric, oesophagoscoplc, and radiologlc findings in the columnar-lined gullet (Barrett syndrome). Gut 1963;4: 406-12. 7. iascone C, OeMeestar Tit, Little AG, Skinner DB. Barrett's esophagus: functional assessment, proposed pathogenests and surgical therapy. Arch Surg 1983;118:543-9. 8. EndoM, KobayashiS, Kozu T, Taremoto 1",NakayamaN. A case of Ban-ett's epithelium followed up for five years. Endoscopy 1974;6:48-51. 9. Mossberg SM. The columnar-lined esophagus (Barrett syndrorne): an acquired condition? Gastroenterology 1966; 50:671-6. 10. Goldman M, Beckman RC. Barrett syndrome: case report with discussion about concepts of pathogenesis. Gastroenterology 1960;39:104-10. 11. Naef AP, Savary M, Ozzelio L Colut'rmar-llnodlower esophagus: An acquired lesion with malignant potential. J Thorac Car. diovasc Stag 1975;70:826-35. 12. Allison PR, Johnstone AS. The oesophagus lined with gastric mucous membrane. Thorax 1953;8:87-101. 13. Brand DL, YlyisakerJT, Gelfand M, Pope CE IlL Rep. ;sion of columnar esophageal (Barrett's) epithelium after -.:-reflux surgery. N Engl J Med 1980;302:844-8. 14. Robbins AH, Hermos JA, Schimrnel EM, Frledl,'~.:-sr DM, Messlan RA, The columnner.lined esophagus: az,'~lyslsof 26 cases. Radiology 1977;123:1-7. 15. Hat~ A, Payne WS, We|land I.H, FontanaRS. Ade~mmclnoma in the columnar-lined lower (Barrett's) oesophagus. Thorax 1973;28:511-4.

192

16. Borrle J, Goldwater L. Columnar cell-lined esophagus: Assessment of etiology and treatment. J Thorac Cardlovasc Surg 1976;71:825-34. 17. Haggltt RC, Tryzelaar J, Ellis FH, Colcher H. Adenocarclnoma complicating columnar eplthelium-llned (Barrett's) esophagus. Am J Clin Pathol 1978;70:1-5. 18. RaphaelHA, Ellis FH Jr, Dockerty MB. Primary adenocarclnoma of the esophagus: 18 year rovlew and review of literature. Ann Surg 1966; 164:785-96. 19. Reyes CV, Wang T. Primary adenocarcinoma of the esophagus: a review of 12 cases. J Surg Oncol 1981;18:153-8. 20. Adler RH. The lower esophagus lined by columnar epithelium: Its association with hiatal hernia, ulcer, stricture, and tumor. J Thorac Cardlovasc Surg 1963;45:13-34. 21. PaullA, Trier JS, Dalton MD, Camp RC, Loeb P, Goyal RK. The hlstologio spectrum of Baffett's esophagus. N Engl J Med 1976;295:476-80. 22. Skinner DB, Walther BC, RIddeU RH, Schmldt H, lascone C, DeMeester TR. Barrett's esophagus: comparison of benign and malignant cases. Ann Surg 1983;198:554-66. 23. Parsa C. AdenoGarclnomaassociated with Barrett's esophagus. JAOA 1980;79:572-6. 24. Cameron AJ, Ott BJ, Payne WS. Barrett's oesophagus: Incidence of adenocarclnoma during long-term follow up. Gut 1983;24 (abstr): 1007-8. 25. Dorsch ER, Graham DY, Lie JT, Estrada RG. The columnarcell-lined (Barrett's) esophagus: a premallgnantcondition? South Med J 1977;70:505-7. 26. Belladonna JA, Hajdu SI, Balns MS, Wlnawer SJ: Adenocarclnoma in sltu of Barrett's esophagusdiagnosedby endoscopic cytology. N Engl J Med 1974;291:895-6. 27. Mangla JC. Barrett's epithelium: regression or no regression. N Engl J Med 1980;303:529.

Discussion Martin H. Max (Norfolk, VA): Dr. Sarr, please state for us your definition of Barrett's esophagus for purposes of this study. Is it related to the lower esophageal sphincter, Z line, or centimeters from the incisors?

Keith A. Kelly (Rochester, MN): Dr. Sarr, is the true incidence of Barrett's esophagus even higher than the alarming incidence indicated in this study. You implied that only those patients who underwent biopsy were studied. The bottom line here might be that biopsy ought to be a routine part of every endoscopy for esophageal reflux. It might provide evidence of Barrett's esophagus, even though the clinician may not recognize it endoscopically. I suspect that some cases of Barrett's esophagus are missed because the endoscopist did not obtain a biopsy specimen. Would you comment on this? D a v i d W. B u t s c h (Moretown, VT): Dr. Sarr, how often do you perform endoscopy, cytologic examination, and biopsy in the patients with Barrett's esophagus?

Michael G. Sarr (closing): Dr. Max, our definition of Barrett's esophagus was a pathologic definition. We were careful to include only patients who had esophageal biopsy specimens obtained above the manometrically defined lower esophageal sphincter or from an area t h a t was described by the endoscopist as part of the esophagus and not the hiatal hernia. Any patient who did not fulfill either of these criteria was not included in the group with Barrett's esophagus. Dr. Kelly, in terms of the incidence of Barrett's esopha ~ s , yes, quite possibly the incidence of the abnormality

Tire ~

JklUll~l ~ SuqleCy

Barrett's Esophagus, Adenocarclnoma, and Gastroesophageal Reflux

might be even higher. Unfortunately we don't have the denominator. We studied only those patients who had symptoms suggestive of gastroesophageal reflux. The symptoms were broad enough to include most patients who, indeed, had symptomatic reflux, but probably also included a group of patients who did not have reflux as well. We did find that there were some patients who had Barrett's esophagus who did not have any sympt/oms of reflux. They were excluded from the study. Interestingly, two of these patients also had adenoeareinoma, and they were evaluated for symptoms not of gastroesophageal reflux. One had upper gastrointestinal bleeding and the other had nonspecific symptoms. Thus, some patients with Barrett's esophagus will present with an associated adenocarcinoma without having had previous symptoms. Finally, Dr. Butsch, how closely to follow patients with

Vott~ 16, ~

lSSS

Barrett's esophagus is a good question. There have been a number of studies recently that have reviewed patients with Barrett's esophagus to determine how long it took for adenocarcinoma to develop. The group at the Mayo Clinic reviewed a group of patients retrospectively and found that if they excluded the patients who presented with all adenocarcinoma and just included patients who had Barrett's esophagus, in only one in every 400 patient years would a carcinoma develop. A paper has been presented at this meeting that suggests that one in 175 patient years was the incidence of adenocarcinoma developing. From an economic standpoint, I don't know what the appropriate length of follow-up should be--probably endoscopy every 6 months or every year with multiple biopsies. There has even been a suggestion that esophageal cytologic examination might be most sensitive.

193