Management of Adenocarcinoma in a Columnar-Lined Esophagus

Transcription

1 Management of Adenocarcinoma in a Columnar-Lined Esophagus I. A. Harle, M.D., R. J. Finley, M.D., F.R.C.S.(C), M. Belsheim, M.D., F.R.C.P.(C), D. C. Bondy, M.D., F.R.C.P.(C), M. Booth, M.D., F.R.C.P.(C), D. Lloyd, M.D., F.R.C.P.(C), J. W. D. McDonald, F.R.C.P.(C), S. Sullivan, M.D., F.R.C.P.(C), L. S. Valberg, M.D., F.R.C.P.(C), W. C. Watson, M.D., F.R.C.P.(C), J. V. Frei, M.D., F.R.C.P.(C), R. Slinger, M.D., F.R.C.P.(C), M. Troster, M.D., F.R.C.P.(C), G. E. Meads, M.D., F.R.C.S.(C), and J. H. Duff, M.D., F.R.C.S.(C) ABSTRACT Of 89 patients diagnosed between 1973 and 1983 as having at least 3 cm of columnar-lined esophagus, 22 were found to have adenocarcinoma. There was no difference in sex ratio, smoking, or the use of alcohol between the benign and adenocarcinoma groups. The patients with adenocarcinoma were older (63 years versus 57 years) and had a higher frequency of dysphagia (64% versus 46%), gastrointestinal bleeding (36% versus 24%), extended columnar-lined esophagus (94% versus 28% ), and epithelial dysplasia (68% versus 10%). Heartburn was less frequent in the adenocarcinoma group (59% versus 79%), but when it occurred, it was of longer duration (mean, 18.8 years versus 10.9 years). In 2 patients, progression from benign columnar-lined esophagus to early adenocarcinoma was observed. Of the patients with adenocarcinoma, 2 received palliative treatment without resection and died four and nine months later. Six underwent partial esophagogastrectomy with 1 postoperative death. Four had residual columnarlined esophagus at the resection margins. In one of them, stricture developed and in one, anastomotic recurrence of adenocarcinoma; 1-year survival was 50%. Fourteen patients underwent total thoracic esophagectomy with no operative deaths, strictures, or anastomotic recurrences; 1- year survival was 5 of 6. Surgical staging revealed that 63% had transmural spread and 55%, lymph node involvement. Patients with columnar-lined esophagus, particularly with epithelial dysplasia, extended columnar-lined esophagus, and a long duration of heartburn, are at risk of developing adenocarcinoma. Adenocarcinoma may be diagnosed at an early stage if patients with epithelial dysplasia are followed with regular endoscopy. Total versus partial removal of the esophagus for adenocarcinoma in columnar-lined esophagus decreases the risk of complications from residual columnar-lined esophagus. From the Departments of Surgery, Medicine, and Pathology, University of Western Ontario and St. Joseph s University, and Victoria Hospitals, London, Ont, Canada. Presented at the Twenty-first Annual Meeting of the Society of Thoracic Surgeons, Phoenix, AZ, Jan 21-23, Address reprint requests to Dr. Harle, do R. 1. Finley, M. D., Department of Surgery, Victoria Hospital, 375 South St, London, Ont, Canada N6E 2H3. Columnar-lined esophagus is defined as a condition in which the distal tubular esophagus is lined with columnar-type epithelium regardless of the presence or absence of a hiatus hernia. Skinner and associates [l] added a further specification to this definition by suggesting that greater than 3 cm of distal tubular esophagus must be lined with columnar-type epithelium to be termed a columnar-lined esophagus. This entity was described by Barrett [2] in 1950 as a condition of peptic ulceration arising in a zone of gastric epithelium located above the cardiac sphincter. He [2, 31 conjectured that the majority of patients represented cases of a congenital short esophagus with part of the stomach extending upward into the mediastinum. In 1953, Allison and Johnstone (41 differed with Barrett in their description of the entity. They wrote that Barrett s esophagus was an esophagus lined with columnar epithelium and remarked on its association with ulceration and stenosis of the esophagus. In 1953, Morson and Belcher [5] described the case of an adenocarcinoma arising in a columnar-lined esophagus. Since that time, the pathogenesis of this adenocarcinoma has often been disputed. It has been suggested that examples of both an acquired and congenital etiology exist. The congenital theory states that rests of columnar epithelium persist from the fetal esophagus into the adult esophagus [6] and that these columnar cells undergo neoplastic change and develop into adenocarcinoma. Today, the acquired theory is the most accepted. This hypothesis states that injured squamous epithelium adapts to chronic injury by undergoing destruction and subsequent metaplastic changes to become columnar epithelium that is more acid resistant. The metaplastic epithelium has the potential to undergo dysplasia and eventually become an adenocarcinoma [ Various investigators have assessed numerous risk factors for their role in the malignant degeneration of columnar-lined esophagus. Some of these factors are smoking, alcohol, gastroesophageal reflux, and epithelial dysplasia (1, 131. Management of adenocarcinoma arising in a columnar-lined esophagus and its premalignant states is not well established. The purpose of this study is to identify clinical and pathological risk factors that contribute to the development of adenocarcinoma in Barrett s esophagus. In addition, methods of treatment of established adenocarcinoma in columnar-lined esophagus are com- 330

2 331 Harle et al: Adenoma in Columnar-Lined Esophagus pared in an attempt to determine which treatment is best. Material and Methods For this study, Barrett s esophagus was defined as columnar-type epithelium lining 3 cm or more of the distal tubular esophagus regardless of the presence or absence of a hiatus hernia. Patients with islands of gastric mucosa within the esophageal mucosa were not included in the series, but those with tongues of gastric mucosa extending 3 cm or more above the anatomical gastroesophageal junction were included. Biopsy-proven columnar-type epithelium lining the distal esophagus was necessary; endoscopic visualization alone was an insufficient criterion for inclusion in the study. Eightynine patients met the criteria for this series. Of these 89 people, 22 were identified as having biopsy-proven adenocarcinoma arising in a columnar-lined esophagus. As described by Paul1 and associates (141, three distinct types of epithelium were found within a columnarlined esophagus. These included a gastric fundic type of epithelium with parietal cells and chief cells, and a junctional type of epithelium containing mucus-secreting columnar cells and pyloriclike glands. The most predominant and important type identified was the specialized columnar epithelium with a villiform surface structure and evidence of intestinal metaplasia with goblet cells and Paneth s cells. Metaplasia was defined as the process in which the squamous cell of the esophageal mucosa undergoes a change to, or replacement by, a cell type that is not normal for this tissue, namely, the columnar cell. For the purposes of this study, dysplasia referred to changes in size, shape, and organization of the mucosa of the columnar-lined esophagus that are suggestive of atypia and neoplastic alteration. Dysplasia was graded as follows: 1 = negative for dysplasia; 2 = mild; 3 = moderate; and 4 = severe. Malignant columnar-lined esophugus was defined as the condition in which adenocarcinoma occurs above the anatomical gastroesophageal junction and arises from columnar epithelium lining the distal tubular esophagus. In this series, cases of benign and malignant columnar-lined esophagus were also classified as being either extended or limited columnar-lined esophagus, as proposed by Ransom and associates (151. Limited columnar-lined esophagus refers to columnar epithelium extending upward from the anatomical gastroesophageal junction to a maximum of 30 cm from the incisors. The extended variety is columnar-lined esophagus extending upward from the anatomical gastroesophageal junction to a level higher than 30 cm from the incisors. The charts of patients diagnosed as having esophagitis, hiatus hernia, Barrett s esophagus, and adenocarcinoma of the esophagus and gastroesophageal junction from the records of the gastroenterology and surgery departments of University of Western Ontario Hospitals between the years 1973 and 1983 were reviewed. In addition, gastroenterology and pathology reports from 1973 to 1983 were examined to identify patients with a pathological diagnosis of benign or malignant columnarlined esophagus. From these sources, 89 patients were diagnosed as having benign or malignant columnar-lined esophagus. Their charts were reviewed and assessed for social habits, symptoms, endoscopic descriptions, medications, and surgical procedures involving the esophagus or stomach. Alcohol was considered a potential risk factor if a person ingested alcoholic beverages socially or more frequently. People who had a previous history of alcohol abuse but had since reformed were also included. Smoking was the other social habit considered, and those who had stopped smoking more than 10 years prior to presentation were considered nonsmokers. A review of the symptomatology in patients with columnar-lined esophagus established three main categories of symptoms. Symptoms of the first category, gastroesophageal reflux, were manifested as heartburn or regurgitation or both, and were aggravated by postural change. The second major category, dysphagia, was related to obstruction. The third category was evidence of upper gastrointestinal bleeding and included melena, hematemesis, or anemia. Of the 89 patients, 67 had the benign condition of Barrett s esophagus. The remaining 22 were identified as having adenocarcinoma of the esophagus arising in a Barrett s esophagus. The diagnosis of malignant columnar-lined esophagus was made from endoscopically directed biopsies and confirmed following pathological examination of the surgical resection specimen. All surgical specimens were reexamined by a single pathologist at each of the three hospitals to assess the presence or absence of the following four variables: columnar or squamous epithelium at the proximal resection margin; dysplasia of the columnar epithelium; degree of transmural spread; and evidence of metastases to resected lymph nodes. All adenocarcinomas arising in a columnar-lined esophagus were staged using the TNM system for carcinoma of the esophagus as described by the International Union against Cancer in Geneva. Patients who had not undergone endoscopy within six months of December, 1983, were contacted by letter or by telephone and requested to return for repeat endoscopy and biopsy. Biopsy specimens were taken from above, below, and within the columnar-lined esophagus. To compare risk factors between the benign and malignant columnar-lined esophagus, chi-square analysis was used. Other data were statistically analyzed using the unpaired Student t test. Results Risk Factors Among the 67 patients with benign columnar-lined esophagus, there were 45 male and 22 female patients, a male: female ratio of 2.04: 1. Among the 22 patients with adenocarcinoma, there were 17 men and 5 women, a higher male: female ratio of 3.4: 1. The mean age was 57

3 332 The Annals of Thoracic Surgery Vol 40 No 4 October years for the group with benign disease (range, 17 to 85 years) and years for the group with malignant disease (range, 35 to 83 years). The social habits of smoking and alcohol consumption were reviewed for both groups. Twenty-seven (40%) of the 67 patients with columnar-lined esophagus had a smoking history, and only 7 (32%) of the 22 with malignant columnar-lined esophagus were smokers. Likewise, alcohol consumption was not a significant risk factor between the two groups; 20 (30%) of the 67 patients with benign columnar-lined esophagus and 9 (41%) of the 22 patients with adenocarcinoma consumed alcoholic beverages. Fifty-three (79%) of the 67 patients with benign columnar-lined esophagus complained of symptoms suggestive of reflux. In the group with adenocarcinoma, reflux symptoms were not as common (13 of 22 patients, 59%). When present, however, reflux occurred longer in the group with malignant disease (18.8 years) compared with the group with benign disease (10.9 years). We differentiated between the patients in each group on the basis of extended or limited columnar-lined esophagus. We were unable to determine whether 5 patients with adenocarcinoma had an extended or limited columnar-lined esophagus from review of the pathological data, and hence only 17 could be assessed from this aspect. It was significant that 16 of 17 (94%) of the patients with malignancy had an extended columnarlined esophagus, compared with 16 of 57 patients (28%) with benign columnar-lined esophagus (p < 0.05). Dysplasia was an important feature in columnar epithelium surrounding adenocarcinoma. Dysplastic changes were evident in 15 (68%) of 22 patients with adenocarcinoma, while only 7 (10%) of 67 patients with the benign condition of Barrett's esophagus showed dysplasia (p < 0.05). All patients demonstrating dysplastic alterations had associated intestinalization of the columnar epithelium. Presenting Features The review of the symptomatology in patients with columnar-lined esophagus revealed three major areas of symptoms. Fifty-three (79%) of the 67 patients with benign columnar-lined esophagus complained of symptoms suggestive of reflux, but only 59% of the group with malignancy (13/22) had such symptoms. Dysphagia was present in 64% of the patients with adenocarcinoma (14/22) versus only 46% (31167) of those with benign columnar-lined esophagus. A highly significant finding was the more frequent evidence of upper gastrointestinal bleeding in the group with adenocarcinoma. At presentation, it was found in 36% of that group (8122) but in only 24% of the group with benign columnar-lined esophagus (16/67) (p < 0.05). A pair of 56-year-old identical male twins were seen with adenocarcinoma in a columnar-lined esophagus within one month of each other. The first brother complained of dysphagia but had no history of gastroesophageal reflux. The resected specimen showed a transmural adenocarcinoma in an extended columnarlined esophagus with spread to the regional lymph nodes plus numerous areas of carcinoma in situ throughout dysplastic columnar epithelium. The second brother had epigastric pain and a 10-year history of heartburn. The resected specimen showed a transmural adenocarcinoma in a dysplastic, extended columnarlined esophagus with spread to the regional lymph nodes. Both patients were smokers and drinkers, and are alive following total esophagectomy and gastric interposition. Follow-up of Patients with Benign Disease Two patients who were originally diagnosed as having benign columnar-lined esophagus were followed for one month and eighteen months and showed progression from a metaplastic intestinalized epithelium to a dysplastic lesion with early carcinoma in situ. Both of these patients underwent a total esophagectomy and a gastric pull-up, and are alive and well as of this writing. Of 23 patients with benign columnar-lined esophagus who returned for follow-up, 13 had received anti-reflux procedures. One of them was asymptomatic following a Nissen fundoplication and had no evidence of residual columnar-lined esophagus on endoscopic biopsy and examination a year after operation. Nine were asymptomatic following the antireflux procedure, but still demonstrated persistent columnar-lined esophagus. Three of the 13 patients had continued reflux symptoms as well as continued evidence of columnar-lined esophagus on endoscopic biopsy. Treatment of Patients with Malignant Disease Patients with adenocarcinoma arising in columnar-lined esophagus were managed by one of three modes of therapy: palliative treatment without resection; partial esophagogastrectomy with anastomosis in the chest; or total thoracic esophagectomy with a gastric pull-up and anastomosis in the neck. Two of the 22 patients with adenocarcinoma were not considered for resection. One was an 83-year-old woman not medically suitable for operation. She was treated with radiation therapy and died four months after diagnosis of adenocarcinoma in columnar-lined esophagus. The other individual was a 69-year-old man who was diagnosed with Barrett's esophagus in 1976 and did not undergo endoscopy again until 1981, at which time he had biopsy-proven adenocarcinoma of the esophagus in columnar-lined esophagus. He had been asymptomatic while being treated with antacids; then, dysphagia suddenly developed. Metastases to liver and lung were found, and the patient was treated by insertion of a Mousseau-Barbin tube for palliation. He died nine months after the diagnosis. Six patients underwent partial esophagogastrectomy with an en bloc resection of local nodal tissue and a gastroesophageal anastomosis within the thorax. The condition of a 35-year-old man deteriorated in the postoperative period, and he died three months after opera-

4 333 Harle et al: Adenoma in Columnar-Lined Esophagus tion. His death was considered to have occurred in the postoperative period. Four of the 6 patients had residual columnar epithelium at the proximal resection margin. In 1 of them, an anastomotic stricture requiring repeat dilations developed. The patient is still alive 4 years after esophagogastrectomy. In another patient, an anastomotic recurrence developed 2 years following esophagogastrectomy; he underwent a total thoracic esophagectomy with a gastric pull-up. This man survived only 3 years from the time of the original diagnosis of adenocarcinoma. Cumulative one-year survival for this group of patients was 50% (3/6). The remaining 14 patients underwent en bloc total thoracic esophagectomy with esophagogastric anastomosis in the neck. There were no operative deaths in this group, and no anastomotic strictures or anastomotic recurrences have developed. Of the 6 patients who have been followed for more than a year, 5 are alive and free from recurrent disease. Staying Staging was performed postoperatively on all surgical resection specimens. Among the 6 patients who underwent partial esophagogastrectomy, 1 had a Stage I tumor, 2 had Stage I1 tumors, and 3 had Stage I11 tumors. Among the 14 patients who had total esophagectomy, 4 had Stage I, 2 had Stage 11, and 8 had Stage 111 tumors. Sixty-three percent had transmural spread, and 55% showed lymph node involvement. Comment The reported incidence of adenocarcinoma arising in a columnar-lined esophagus shows wide variability. The incidence ranges from 2.5% at the Mayo Clinic [16], 8.6% at Lausanne (121, 26% at Indianapolis (171, 30% at Rotterdam [18], and 46% at the University of Chicago [l]. In this study, 25% of the patients with at least 3 cm of columnar-lined esophagus had adenocarcinoma. The large discrepancy in the incidences reported is accounted for not by the true rate of adenocarcinoma, that is, malignancies that are usually symptomatic and easily identified in each center, but by the real incidence of benign columnar-lined esophagus. This incidence can be affected by the definition of a columnar-lined esophagus, visual versus biopsy proof of the columnar-lined epithelium, and the source of the patient data base used for analysis. Since the incidence of gastric mucosa occurring within the tubular esophagus ranges from 0.67 to 70% [7, 191, a strict definition of Barrett s esophagus is necessary. The strict criteria used by Skinner and colleagues [l] for inclusion of benign columnar-lined esophagus may account for the high incidence of adenocarcinomas in their series. Although their inclusion criteria for a columnarlined esophagus were used in this study, the incidence of adenocarcinoma was 25%, not 46%. Compared with that study, the lower incidence of adenocarcinoma in our report could be accounted for by a broader patient data base from not only the department of surgery and pathology but also the department of medicine, which sees more patients with benign columnar-lined esophagus. The pathogenesis of adenocarcinoma in columnarlined esophagus is a poorly understood topic but one for which many theories have been advanced. Some investigators [ 1, 6, 121 support injurious factors such as acid, bile, smoking, or alcohol as important stimuli for the development of adenocarcinoma in columnar-lined esophagus. Skinner and associates [l] promoted smoking as a significant risk factor in the development of adenocarcinoma in Barrett s esophagus. Data from the present study suggest that neither smoking nor alcohol consumption is more common in the group with malignant columnar-lined esophagus than in patients with benign columnar-lined esophagus. According to acquired theory, adenocarcinoma develops in a columnar-lined esophagus in response to the long exposure of esophageal squamous epithelium to acid and bile. This injured squamous epithelium undergoes destruction and subsequent metaplasia to become columnar epithelium that is more acid resistant [20]. If injury is allowed to continue, the metaplastic cell has the potential to undergo dysplasia and, eventually, malignant degeneration [6]. In support of this theory are the cases of 2 patients in this study whose disease progressed from benign columnar-lined esophagus to carcinoma in situ. One of these 2 patients gave a past history of gastroesophageal reflux for more than 20 years, for which she underwent an unsuccessful antireflux procedure 14 years prior to diagnosis of adenocarcinoma in Barrett s esophagus. She also complained of progressive dysphagia for 10 years before diagnosis of malignancy. The other patient was seen with esophageal ulceration, hematemesis, and melena at the time of diagnosis of Barrett s esophagus. Over the ensuing year and a half, the patient had symptoms of severe reflux esophagitis, and eventually a carcinoma developed. Although the cases of these 2 patients suggest that gastroesophageal reflux may be an important factor in the development of adenocarcinoma in a columnar-lined esophagus, only 59% of patients with malignant columnar-lined esophagus compared with 79% of those with benign columnar-lined esophagus complained of symptoms suggestive of reflux. Therefore, what could be the stimulus for the development of adenocarcinoma in the 41% of patients with malignancy without symptomatic reflux? Certainly some of the patients could have gastroesophageal reflux without symptoms, but there also could be some other cause for malignant degeneration in the esophagus. The group with benign columnar-lined esophagus had an average duration of reflux of 10.9 years compared with 18.8 years in the group with adenocarcinoma, even though fewer of these patients complained of reflux. Perhaps the longer the duration of reflux, the greater the likelihood of progression from squamous epithelium to

5 334 The Annals of Thoracic Surgery Vol 40 No 4 October 1985 columnar epithelium to intestinalization and then dysplasia and finally frank adenocarcinoma. If gastroesophageal reflux is an important stimulus for this sequence of events, then abolishing reflux by an antireflux procedure should prevent further progression of changes and perhaps even allow regression toward normal [6, 20, 211. Only 1 of 13 patients with benign columnar-lined esophagus showed regression of columnar epithelium to normal squamous epithelium 1 year following an antireflux procedure. Our data do not support regression of columnar epithelium following an antireflux procedure, but it should be pointed out that we did not perform any 24-hour ph studies on these patients to assess whether or not they had continuing gastroesophageal reflux. If reflux cannot be implicated as a causative factor lending support to the acquired theory, then the congenital theory of columnar-lined esophagus with or without associated reflux must be examined. The congenital theory postulates that rests of columnar epithelium persist from the fetal esophagus into the adult esophagus. In the embryo, the esophagus is lined with stratified columnar epithelium. Between the 28- and 34-mm stage, changes occur. The esophageal lumen becomes temporarily and partially occluded by epithelial proliferation and bridging; later, by the process of vacuolation, the lumen is reestablished. In the 130-mm embryo, stratified columnar epithelium is replaced by stratified squamous epithelium beginning in the middle third of the esophagus and extending proximally and distally [6, 7, 22, 231. It is not confirmed [22] whether the ciliated cells are transformed into squamous cells [24] or whether the ciliated cells undergo desquamation [25]. The last portion of the esophagus to lose its ciliated columnar lining is the upper third where columnar epithelium may exist at birth and persist for a short time thereafter [ll, 221. Support for the congenital theory comes from the examination of 200 esophageal autopsy specimens from neonates by Postlethwait and Musser [26], who found one esophagus completely lined with columnar epithelium, which is similar to the complete, sheetlike covering formed in an extended columnarlined esophagus. From this 10-year study at the University of Western Ontario, we have found several features that provide support for the congenital theory. First, only 59% of patients with malignant columnar-lined esophagus had a history of reflux, a finding that suggests some factor other than acid injury in the development of adenocarcinoma in a columnar-lined esophagus. Second, 2 of the patients with adenocarcinoma in columnar-lined esophagus were identical twins and both were seen at the age of 56 years within one month of each other. Both had identical lesions with an adenocarcinoma arising in an extended columnar-lined esophagus of the lower esophagus with positive paraesophageal nodes, both underwent total thoracic esophagectomy with gastric pull-up, and are alive and well at present. It is important that only 1 of the twins was symptomatic for mild reflux, the other had no symptoms of reflux. Third, an extended columnar-lined esophagus usually forms a complete sheetlike covering with minimal inflammation or stricture formation, thereby suggesting a congenital origin. In this study, a columnar-lined esophagus extending proximal to the aortic arch was significantly more common in patients with the malignant columnar-lined esophagus and was present in all but 1 of the patients with adenocarcinoma. Militating against the congenital theory is the fact that usually the cervical esophagus of the fetus is the last area of columnar epithelium to be replaced by squamous epithelium, and in the extended type the upper esophagus is where the squamous epithelium exists. Another feature contradicting a congenital origin for the extended columnar-lined esophagus is that it is accompanied by severe reflux with very low lower esophageal sphincter pressures [15]. Our experience suggests that it is unlikely that columnar-lined esophagus is an exclusively acquired entity and that consideration should be given to a congenital etiology in many instances. The management of adenocarcinoma in columnarlined esophagus has varied in the past. In the presence of a resectable adenocarcinoma in a columnar-lined esophagus, two possible surgical methods should be considered. Our results do not favor the use of partial esophagogastrectomy with anastomosis in the chest, because the presence of persisting columnar epithelium above the anastomosis allows it to be subject to the ongoing risks of recurrent malignancy, stricture formation, or both. Our experience supports complete removal of the intrathoracic esophagus with anastomosis of the stomach to the cervical esophagus in the neck. This technique resulted in low morbidity and no mortality in our series. Although longer follow-up of these patients is necessary, early survival has been excellent and there has been no evidence of recurrent anastomotic cancer or stricture formation. How can survival with adenocarcinoma in columnarlined esophagus be improved? We suggest close, endoscopically directed biopsy follow-up every six months to 1 year in patients with columnar-lined esophagus. This frequency should be increased to every three months if there is any evidence of dysplasia or of columnar-lined esophagus above the aortic arch. As in the study of Skinner and colleagues [l], dysplasia is strongly associated with adenocarcinoma in a columnar-lined esophagus. Frequent surveillance had resulted in the early diagnosis and management of a more favorable lesion-carcinoma in situ-in 2 of our patients. In this series of 89 patients with columnar-lined esophagus, there was a 25% incidence of adenocarcinoma of the columnar-lined esophagus. Our series differed from others in that the incidences of smoking and alcohol consumption were not significantly different between the patients with benign columnar-lined esophagus and those in whom adenocarcinoma in columnar-lined esophagus developed. We believe that the columnar-lined esophagus and its development of

6 335 Harle et al: Adenoma in Columnar-Lined Esophagus adenocarcinoma cannot be attributed exclusively to an acquired pathogenesis because of our cases of identical male twins in whom similar malignant lesions developed and because of the absence of reflux symptoms in 41% of patients with malignant columnar-lined esophagus; both of these features suggest that perhaps a congenital pathogenesis plays some part in this entity. The presence of an extended columnar-lined esophagus and evidence of dysplasia warrant aggressive follow-up, as these patients are at a greater risk to undergo malignant degeneration within the columnar-lined esophagus. We followed 2 patients who progressed from benign columnar-lined esophagus to adenocarcinoma, and they are alive and well today following total thoracic esophagectomy for Stage I and Stage I1 lesions. Management of adenocarcinoma in columnar-lined esophagus should be aggressive. In this series, total thoracic esophagectomy was associated with the lowest morbidity and mortality and excellent cumulative 1-year survival. References 1. Skinner DB, Walther BC, Riddell RH, et al: Barrett s esophagus: comparison of benign and malignant cases. Ann Surg 198:554, Barrett NR: Chronic peptic ulcer of the oesophagus and oesophagitis. Br J Surg 38:175, Barrett NR The lower esophagus lined by columnar epithelium. Surgery 41:881, Allison PR, Johnstone AS: The oesophagus lined with gastric mucous membrane. Thorax 8:87, Morson BC, Belcher JR: Adenocarcinoma of the oesophagus and ectopic gastric mucosa. Br J Cancer 6:127, Sjogren RW Jr, Johnson LF: Barrett s esophagus: a review. Am J Med 74:313, Berardi RD, Devaiah KA: Barrett s esophagus: collective review. Surg Gynecol Obstet 1W521, Bremner CG: Lynch VP, Ellis FH: Barrett s esophagus: congenital or acquired? Surgery 68:209, Goldman MC, Beckman RC: Barrett syndrome: case report with discussion about concepts of pathogenesis. Gastroeneterology 39:104, Hamilton SR, Yardley JH: Regeneration of cardiac type mucosa and acquisition of Barrett mucosa after esophagogastrostomy. Gastroenterology 72669, Mossberg SM: The columnar-lined esophagus (Barrett syndrome): an acquired condition? Gastroenterology 50:671, Naef AP, Savary M, Ozzello L, Pearson FG: Columnarlined lower esophagus: an acquired lesion with malignant predisposition. J Thorac Cardiovasc Surg 75:826, Kalish RJ, Clancy PE, Orringer MB, Appelman HD Clinical, epidemiologic, and morphologic comparison between adenocarcinomas arising in Barrett s esophageal mucosa and in the gastric cardia. Gastroenterology 86:461, Paul1 A, Treir IS, Dalton MD, et al: The histologic spectrum of Barrett s esophagus. N Engl J Med 295:476, Ransom JM, Patel GK, Clift SA, et al: Extended and limited types of Barrett s esophagus in the adult. Ann Thorac Surg 33:19, Hawe A, Payne WS, Weiland LH, Fontana RS: Adenocarcinoma in the columnar epithelial-lined lower (Barrett) esophagus. Thorax 28511, Radigan LR, Glover JL, Chipley FE, Shoemaker RE: Barrett esophagus. Arch Surg 112:486, Dees J, Van Blankenstein M, Frendel M: Adenocarcinoma in Barrett s esophagus: a report of 13 cases. Gastroenterology 74:1119, Taylor AL: The epithelial heterotopias of the alimentary tract. J Pathol Bacteriol 30415, Bremner CG: The columnar-lined (Barrett s) esophagus. Surg Ann 9:103, Brand DL, Ylvisaker JT, Gelfand M, Pope CE 11: Regression of columnar esophageal (Barrett s) epithelium after antireflux surgery. N Engl J Med 302:844, Johns BAE: Developmental changes in the esophageal epithelium in man. J Anat 86:431, Johnson FP: The development of the mucous membrane of the esophagus, stomach and small intestine in the human embryo. Am J Anat 10521, Neumann E: Die Metaplasie des fotalen esophagusepithel. Fortschr Med 15:366, Schaffer J: Uber die drusen der menschlichen speiserohre. Sb Akad Wess Wien 106:175, Postlethwait RW, Musser AW: Changes in the esophagus in 1,OOO autopsy specimens. J Thorac Cardiovasc Surg 68:953, 1974 Discussion DR. TOM R. DeMEESTER (Omaha, NE): This study presented by Dr. Finley is both affirmative and controversial regarding the current state of knowledge about Barrett s esophagus. It is controversial in that the investigators have interpreted the observation that symptoms suggestive of reflux occurred in 79% of the group with benign disease and only 59% of the group with adenocarcinoma to indicate that there should be a shift away from an acquired and a shift toward a congenital pathogenesis for Barrett s esophagus. This conclusion is based on soft data. In other reported series, an observation supporting an acquired etiology is that in more than 95% of patients with Barrett s esophagus, severe esophageal exposure to gastric juice has been documented with 24-hour esophageal ph monitoring (Arch Surg , 1983). In many of these patients, the symptoms of heartburn and regurgitation were not present. This is apparently due to individual differences in esophageal sensitivity to gastric juice and the possibility that Barrett s epithelium may not be sensitive to an acid ph. Furthermore, manometric studies on these patients showed that a severe and mechanically defective distal esophageal sphincter is almost universally a cause of increased gastroesophageal reflux and that the degree of Barrett s involvement is related to the amount of esophageal exposure to gastric juice. The authors garner further support for their congenital hypothesis from a unique experience with identical twins who both had malignant lesions within Barrett s esophagus at 56 years of age. The current pathogenesis of carcinoma appears to be related to both a genetic predisposition toward the development of a specific malignancy and prolonged exposure to a carcinogen or epithelial irritant. Consequently, an alternative explanation for this unique observation could be that with the same genetic background, the continuous exposure of the esophagus to gastric juice resulted in a malignant transformation of an acquired Barrett s epithelium in each twin at the same time. It is unfortunate that one of the twins did not encounter a pediatric surgeon and obtain a fundoplication at an early age. If this had been so, a natural experiment would have emerged.

7 336 The Annals of Thoracic Surgery Vol 40 No 4 October 1985 The study is affirmative in that it underscores the emerging concept that the presence of dysplasia in the biopsy specimen of Barrett s mucosa is an ominous sign and merits aggressive follow-up, even in the absence of symptoms. My colleagues and 1 believe that if the dysplasia is high grade, the patient should be strongly advised to undergo esophagectomy. Dr. Finley, at which point would you advise similar action? Regarding the surgical management of adenocarcinoma in Barrett s esophagus, we would encourage, as did the authors, a near-total thoracic esophagectomy. 1 would extend this to a near-total gastrectomy because of the tendency of these lesions to develop near the cardia and extend throughout the submucosal lymphatic system of the proximal stomach. Reconstruction is done with a colon interposition between the antrum or duodenum and the esophagus with the esophagocolonic anastomosis just inside the thoracic inlet. DR. w. SPENCER PAYNE (Rochester, MN): 1 enjoyed this presentation, and will allude to only one aspect. This relates to the recommended cancer detection program for patients with benign Barrett s esophagus who are at risk for developing subsequent cancers. As in the present study, most reports, including our own, indicate an 8 to 15% prevalence of esophageal adenocarcinoma among all patients identified as having Barrett s esophagus. Although this is an impressive prevalence, one should not interpret it as indicating that cancer of the esophagus will develop in up to 15% of patients with benign Barrett s esophagus. Alan Cameron and I recently attempted to quantify the cancer risk in patients with Barrett s esophagus through a retrospective analysis of 102 patients seen at the Mayo Clinic over a 20-year period. In only 2 of these patients did esophageal adenocarcinoma subsequently develop during an average follow-up of 8.5 years. This was an incidence of 227 cancers per 100,000 patients per year, approximately 30 times greater than the incidence expected in our general population. Another way of looking at these data, however, is that 1 cancer was detected for every 441 patient-years of observation. Thus, if these patients had undergone semiannual cancer detection, only 1 cancer would be detected out of 882 examinations. Despite the cancer incidence, the actuarial survival of these 102 patients was identical to that of the general population. These comments are made not to suggest there is no problem, but rather to place the problem in a more realistic perspective. In addition, there are potentially serious problems of patient compliance, costs, and the effectiveness of current methods of detecting early cancers in patients with Barrett s esophagus. DR. FINLEY: I would like to thank the discussants for their comments. I agree with Dr. DeMeester that it is difficult to determine if patients with an adenocarcinoma in a columnar-lined esophagus have gastroesophageal reflux either prior to or at the time of the cancer diagnosis. Even with 24-hour ph studies, these data will be hard to obtain except in a prospective study. In our study, identical twins had development of adenocarcinoma in a columnar-lined esophagus within months of each other. One twin had a long history of gastroesophageal reflux, and one did not. I agree that the second twin may have had gastroesophageal reflux without symptoms. However, 1 believe that this clinical observation gives credence to the congenital theory of Barrett s esophagus Dr. Payne, 1 am very concerned when I find dysplasia at the time of esophageal biopsy. If the dysplasia is severe, we either follow that patient very closely with multiple biopsies and brushings of the esophageal mucosa every three to six months or, more frequently, recommend esophagectomy if the patient is fit. Although Dr. Payne s excellent report has more than 100 patients, it points out the need for a multi-institutional study of a large number of patients with a columnar-lined esophagus. The question that needs to be answered is, Does the early diagnosis of adenocarcinoma in patients with a columnar-lined esophagus by regular endoscopic examination reduce crude mortality in this disease process? In the meantime, I recommend that patients with an extended columnar-lined esophagus, mucosal dysplasia, or a long history of gastroesophageal reflux be followed by careful endoscopic examination and esophageal biopsies every six months. 1 thank Dr. Griff Pearson for sponsoring this paper.