Cancer of Esophagus and Esophagogastric Junction: Analysis of Results of 1,025 Resections after 5 to 20 Years

Cancer of Esophagus and Esophagogastric Junction: Analysis of Results of 1,025 Resections after 5 to 20 Years Yun Kan Lu, M.D., Yueh Min Li, M.D., and Yue Zhi Gu, M.D. ABSTRACT Resection was carried out in 1,025 of 1,654 patients with cancer of the esophagus or esophagogastric junction at the Peking Medical College Hospitals in China from 1953 through 1973. All cancers of the esophagus were squamous cell carcinomas except for five adenocarcinomas. A lesion localized within the esophageal wall was found in 55% and lymph node metastasis in 41.3% of the patients undergoing resection. All cancers of the esophagogastric junction were adenocarcinomas. The tumor had invaded beyond the boundaries of the stomach in 76.7% of these patients, and positive nodes were found in 61% of the patients. The rate of resectability was 81.2% for esophageal cancer and 74% for cancer of the esophagogastric junction. Surgical mortality after resection was 4.9% (50/1,025). The 5-year survival after resection was 20.9% (214/1,025). Better results were found following complete resection: 24% (210/ 875) for all patients, 28.2% (162/575) for patients with cancer of the esophagus, and 16% (48/300) for patients with cancer of the esophagogastric junction. Late survival at 10, 15, and 20 years after resection of esophageal cancer was 20%, 12%, and 7.4%, respectively. The favorable prognostic factors after resection of esophageal cancer were tumor of the lower third of the esophagus, the absence of lymph node involvement, and the presence of a localized lesion. The 5-year survival for patients with cancer of the lower third of the esophagus was 32.7%. It was 64.2% for patients with a localized lesion with negative nodes in this subgroup. From 1953 through 1973, a total of 1,654 patients with cancer of the esophagus or esophagogastric junction were admitted to the Department of Thoracic Surgery of the Beijing (Peking) Medical College Hospitals in Beijing, China. In this report, we evaluate the long-term results in the 1,025 patients who underwent resection. Material and Methods Table 1 categorizes all the patients in this series. Of the 1,654 patients admitted, 1,306 underwent thoracotomy and 1,025, resection. Hence, resection was performed in 62% of all the patients admitted (1,025/1,654) and in From the Department of Cardio-thoracic Surgery, First Hospital and People s Hospital, Beijing Medical College, Beijing, People s Republic of China. Accepted for publication Mar 14, 1986. Address reprint requests to Dr. Lu, Department of Cardio-thoracic Surgery, First Hospital, Beijing Medical College, Beijing, People s Republic of China. 78.5% of the patients having thoracotomy (1,025/1,306) Of the patients undergoing resection, 664 had esophageal cancer and 361, cancer of the esophagogastric junction. The male to female ratio was 3.4: 1 for esophageal cancer and 7.6: 1 for cancer of the esophagogastric junction. The mean age was 59 years. The duration of symptoms (mainly dysphagia) ranged from 10 days to 25 months (mean, 4.2 months for esophageal cancer and 6.4 months for cancer of the esophagogastric junction). All patients were residents of northern China. Operation The operation in this series was a standard esophagogastrectomy with primary esophagogastrostomy. No staged procedure was used. The operation was performed through a generous left posterolateral thoracotomy with removal of a rib so that the tumor and the left upper abdominal organs were accessible. In the case of esophageal cancer, after exploration and mobilization of the tumor, the diaphragm was opened. The greater and lesser curvatures of the stomach were fully mobilized as far as the pyloric region in order that the stomach could be brought up to any desired level in the chest. Sufficient blood supply of the gastric remnant was maintained by careful preservation of the right and left gastroepiploic vessels. Usually 5 to 6 cm of esophagus above the gross tumor was resected. The level of the anastomosis was determined by the location of the tumor. The esophagus was not freed more than 2 to 4 cm above the level chosen for anastomosis. Handsewn technique was used. After construction of the anastomosis, the stomach was brought up over the site of the anastomosis by attaching it to the surrounding mediastinal structures so as to cover the stoma as well as to relieve any possible tension. The diaphragm was sutured to the stomach. Closed-chest drainage was instituted in the usual manner. For cancer of the esophagogastric junction, a single thoracotomy instead of a laparotomy was used for exploration and for resection. In this series, a standard resection was considered as complete as one in which all gross tumor together with appropriate proximal and distal esophagus (or stomach or both) and local enlarged lymph nodes were removed. When known tumor or metastasis was left behind in the chest, abdomen, or cut edges of the specimen by microscopic examination, the resection was considered incomplete. Logan and Skinner s en bloc dissection was not used for esophageal cancer in this series. A similar procedure was performed for selected instances of advanced cancer of the esophagogastric junction. It consisted of total gastrectomy, distal esophagectomy, splenectomy, distal pancreatectomy and en bloc lymph node dissection. 176 Ann Thorac Surg 43:176-181, Feb 1987

177 Lu, Li, Gu: Cancer of Esophagus and Esophagogastric Junction Table 1. Categorization of Patients in This Series Esophagogastric Esophageal Junction Category Cancer Cancer Total Admission 1,079 575 1,654 Thoracotomy 818 488 1,306 Resection (standard) 664 361 1,025 Complete 575 300 875 Incomplete 89 61 150 Pathological Characteristics All resected specimens were examined pathologically. Particular attention was directed to the length of the tumor (measured in a fresh specimen), the invasion level of the lesion, and local lymph nodes. In this series, the thoracic esophagus was divided into three segments. The segment between the dome of the pleural cavity and the undersurface of the aortic arch was defined as the upper third. Between the aortic arch and the esophageal hiatus, the esophagus was equally divided into two portions, the middle third and lower third, at approximately the level of the inferior pulmonary vein. Pathological staging was done according to the degree of invasion by the lesion and the status of local lymph nodes in the surgical specimen. The term localized was used when the tumor was confined to the esophageal wall without lymph node involvement. The term advanced was used when the lesion had invaded the periesophageal tissues with or without lymph node metastasis. The term degree of invasion referred to the anatomical layers that the tumor may invade: the mucosa, submucosa, muscularis, serosa, or adjacent tissues. A complete record of the microscopic examination was available for 504 of 575 resected specimens of esophageal cancer, of these, 279 (55%) were classified as localized and 225 (44.6%), as advanced. Lymph nodes were involved in 41.3% (208/504) of the specimens. There was a significant (p <.05) increase in lymph node metastasis with increasing degree of invasion (Table 2). All of the cancers of the esophagogastric junction were adenocarcinomas. Of these, 76.7% (211/275) were invasive. Lymph nodes were positive in 61% (164269). Table 2. Degree of Invasion and Lymph Node Status of 504 Resected Specimens of Esophageal Cancer No. of Resected Lymph Node Degree of Invasion Specimens Metastasis % Submucosa 1 0 0 Muscularis 175 52 29.7" Full thickness 273 118 43.2" Adjacent tissue 55 38 69. la Total 504 208 41.3 "Proportion is significantly different at a p level of less than.05 by Tukey's w procedure for multiple comparisons (121. Table 3. Surgical Deaths after Standard Resection No. of No. of Mortality Resection Resections Deaths (%) Esophageal cancer 664 37 5.6 Complete resection 575 32 5.6 Incomplete resection 89 5 5.6 Esophagogastric cancer 361 13 3.6 Complete resection 300 10 3.3 Incomplete resection 61 3 4.9 Total 1,025 50 4.9 Complete resechon 875 42 4.8 Incomplete resection 150 8 5.3 Results Resectability The rate of resectability was 78.5% (1,025/1,306), 81.2% (664/818) for esophageal cancer and 74% (361/488) for cancer of the esophagogastric junction. Operative Mortality Any death occurring in the hospital after resection, regardless of cause, was considered an operative death. The overall mortality was 4.9% (50/1,025); it was similar for complete and incomplete resections (Table 3). Follow-up and Long-Term Survival Data for this follow-up study were obtained by questionnaires and clinical visits, and the study was completed at the end of 1979. Twenty-six patients have been lost to follow-up (and were calculated as dead). Hence, followup was complete for 97.5% of the patients. The 5-year survival was 20.9% (214/1,025), 24.4% (162/ 664) for patients with esophageal cancer and 13.3% (48/ 361) for patients with cancer of the esophagogastric junction. Better results were found with complete resections: 28.2% (162/575) for cancer of the esophagus and 16% (48/ 300) for cancer of the esophagogastric junction (Table 4). There were only 4 (2.7%) 5-year survivors among the 150 patients who had incomplete resection. The 5-year survival expressed as a percentage of the total number of patients admitted was 12.9% (214/1,654) and of the number of patients having exploratory thoracotomy, 16.4% (2141,306). Late survival at 10, 15, and 20 years was 20%, 12%, and 7.4%, respectively. CANCER OF UPPER THIRD OF ESOPHAGUS. Only 3 (10.7%) of 28 patients lived longer than 5 years. The duration of symptoms was 3 to 4 months in this subgroup. The tumor was 3 to 5 cm in length, and it invaded the muscularis layer in 2 patients and the submucosa in 1 patient. All of them received preoperative radiotherapy and subsequent resection in 1963. No tumor cell could be found in 2 of the 3 specimens. Both of those patients are alive and well without recurrence. The third patient died of an unrelated disease in the eighth year after resection. The 10-year survival of this subgroup was 7.7% (2/26). CANCER OF MIDDLE THIRD OF ESOPHAGUS. The 5-year survival was 26.6% (87/327). The length of the tumor

178 The Annals of Thoracic Surgery Vol 43 No 2 February 1987 Table 4. Long-Term Survival after Standard Complete Resection Esophagogastric Esophageal Junction Survival Total Cancer Cancer 5 Years Resections 875 575 300 Survivors 210 162 48 % 24 28.2 16 10 Years Resections 737 495 242 Survivors 119 99 20 % 16.1 20 8.3 15 Years Resections 503 382 121 Survivors 53 46 7 % 10.5 12.0 5.8 20 Years Resections 253 189 64 Survivors 16 14 2 % 6.3 7.4 3.1 ranged from 2.5 to 8 cm (average, 4.8 cm). Mean duration of symptoms was 4.5 months. A localized lesion was found in 58.7% of the resected specimens (168/286) and positive nodes in 38.1% (109/286). The 5-year survival for patients with a localized lesion in this subgroup was 42.9% (72168); it was only 2.8% (3/109) when the lymph nodes were involved. The 10-year survival was 17.2%. CANCER OF LOWER THIRD OF ESOPHAGUS. The 5-year survival was 32.7% (72/220). The length of the tumor ranged from 2 to 13 cm (mean, 5.4 cm). The duration of symptoms was 5 months in most patients. The tumor was localized in 50% of patients (951190). The highest 5- year survival in the whole series appeared in this subgroup, 64.2% (61/95), and involved patients with a localized lesion. Tables 5 and 6 show the relationship between the long-term results and the degree of invasion by tumor and lymph node status at different locations of esopha- geal cancer. CANCER OF ESOPHAGOGASTRIC JUNCTION. The 5-year survival was 16% (48/300) following complete resection. It was 3.3% for incomplete resections (2/61). A localized lesion within the stomach was found in 45.8% (22/48) of the patients surviving 5 years after complete resection. The 10-year survival was 8.3%, and none of the 10-year survivors died of recurrence. Early and Lute Deaths Following radical (complete) resection of esophageal cancer, there were 543 survivors; 162 of them were 5- year survivors. The other 381 patients died of the disease within 5 years after resection, in 244 of them (64%, 244/ 381), the cause of death was recurrence within the chest, mainly in the mediastinum. There were 82 late deaths (death after fifth postoperative year) among the 162 5-year survivors (Table 7). Five patients died of local recurrence. Two patients died of a second supraaortic cancer, proven to be squamous cell carcinoma, in the thirteenth or fourteenth postoperative year. Radiotherapy was instituted, to no avail. Seven patients died of other malignancies, and 52 died of unrelated diseases. The cause was uncertain in the remaining 16 patients. Comment In view of the fact that the number of long-term survivors after resection of esophageal cancer and cancer of the esophagogastric junction is still disappointingly small in the past two decades, one would assume that the value of the surgical treatment is limited and essentially palliative. Reports [3-61 of recent large series, however, have documented encouraging survival results coupled with a decreased operative mortality. These findings indicate that the pessimism prevailing in the literature might no longer be justified. Surgical Mortality and Anastomotic Leak A lower rate of death is one of the major prerequisites for improving the long-term results. Cardiopulmonary complications, empyema, and anastomotic leak were the main causes of surgical death in our earlier years [7]. Nowadays most of them are considered to be preventable. With experience in surgical techniques and preoperative and postoperative treatment, the incidence of the first two complications has been markedly decreased. The incidence of anastomotic leak has been progressively diminished by us to quite a low level [8], although it remains the main cause of surgical death. We used a handsewn technique instead of a stapler device from 1953 to 1973. The anastomosis is performed with two-layer interrupted silk sutures. The inner layer approximates the mucosa of the stomach and the esophagus, and the outer one unites the muscularis of the Table 5. Lymph Node Status in Long-Term Survivors of Esophageal Cancer 5-Year Result Lymph Node S u rv i v a 1 a Metastasis Resections Survivors @) Resections 10-Year Result Absent 305 146 47.9 257 87 33.9 Present 208 13 6.3 180 9 5 "Significance: p <.01. Survivors Survival" (%)

~ ~~ 179 Lu, Li, Gu: Cancer of Esophagus and Esophagogastric Junction Table 6. Five-Year Survival with Esophageal Cancer and Pathological Categories of 504 Resected Specimens Lesion within Esophageal Wall Lesion with Extraesophageal Tissues Negative Positive Negative Positive All Lesions Lymph Nodes Lymph Nodes Lymph Nodes Lymph Nodes Location (%) (%) (%) (%) (%) Upper third 10.7 (28) 19 (16) 0.0 (12)... (0)... (0) Middle third 26.6 (286)d 43 (168)d 4 (85) 11 (9) 0 (24) Lower third 37.9 (190) 64 (95) 11 (73) 12 (8) 14 (14) Numbers in parentheses indicate the number of resected specimens. ball lesions comparison is separated from lesionflymph node group comparisons.,d, Figures with dissimilar letters are significantly different at a p level of less than.05 by Tukey s w procedure for multiple comparisons [12] Table 7. Causes of Late Deaths after Standard Resection of Esophageal Cancer More than Cause 5-10 Years 10-15 Years 15-20 Years 20 Years Total Noncancerous disease 18 17 12 5 52 Recurrence 4 0 1 5 Other primary cancers 7 Bladder 1 0 0 0 Cervix 2 0 0 0 Liver 0 1 0 0 Lung 2 0 0 0 Kidney 0 0 1 0 Second primary cancer 0 2 0 0 2 of esophagus Uncertain 12 2 2 0 16 Total 39 22 16 5 82 esophagus and the serosa of the stomach. Mattress stitches are preferred for the outer layer. The sutures are made to bite through the muscularis of the esophagus and deep into the submucosa lest the fragile muscular coat of the esophagus become lacerated. Perhaps the most important factor contributing to anastomotic leak is tension. A tense anastomosis is potentially a leaking anastomosis. To prevent a leak by ways consistent with the principles of healing, three important points should be emphasized: accurate approximation of stoma1 surfaces, coverage of the anastomosis with serosa, and avoidance of excessive tension on the suture line. The risk of leak with the handsewn technique in 526 resections (1950 to 1963) was reviewed [8]. The overall incidence was 2.1%, and the lowest incidence was found in the supraaortic anastomosis (0.8%, 2/239). We have continued to be satisfied with the results achieved since then. It has always been our policy to use whatever anastomosis appears to have the lowest incidence of leak. Consequently, we put the anastomosis more frequently in the chest than in the neck. However, it should be stressed that the neck anastomosis is safe and is usually indicated for a lesion of the middle-third or upper-third segment of the esophagus. On the contrary, the construction of a supraaortic anastomosis is more demand- ing in terms of technique and sometimes may be very difficult, especially if the exposure is not adequate. Under this circumstance, a neck anastomosis is a wise alternative. Moreover, a leak from a neck anastomosis may often be closed after adequate drainage, but a leak from an intrathoracic anastomosis is a disaster carrying a very high mortality. We believe a neck anastomosis is a better choice when it is indicated. However, routine total thoracic esophagectomy with a neck anastomosis has not been our preference (see also section Role of Resection). Poor general condition and hypoproteinemia, which often are present in most patients, are well-recognized factors unfavorable to satisfactory healing of the anastomosis. However, we believe, these factors are not the main cause of leak, as sound healing may often be achieved in poor-risk patients. The importance of good technique in performing an esophagogastric anastomosis cannot be overemphasized. Left Thoracotorny The choice of the right or left side for esophageal resection is somewhat controversial. In our practice, a left thoracotomy is used exclusively for exploratory or palliative procedures for or resection of all esophageal cancers. This approach affords an optimum exposure of the esophagus and, more importantly, also greatly facilitates

180 The Annals of Thoracic Surgery Vol 43 No 2 February 1987 the assessment and mobilization of the stomach and its surrounding structures. For the same reason, we prefer this approach rather than a laparotomy for exploration and resection of a tumor at the cardiac end of the stomach. When an extensive resection or total gastrectomy is indicated, the left thoracotomy incision may be readily extended to a combined thoracoabdominal incision. A right thoracoabdominal approach or simultaneous right thoracotomy and laparotomy with a semilateral position results in poor exposure of either the thoracic or abdominal organs. Neither of these incisions has been our preference. In our practice, a single right thoracotomy for cancer of the middle third of the esophagus has infrequently been used because of its limited exposure of the stomach. For high-level esophageal cancer, a left thoracotomy and an oblique left neck incision are employed. They can be performed without changing the position of the patient. The cancer located at the level of the aortic arch (i.e., the "retroaortic" cancer) may be an exception to a left thoracotomy, but this cancer is rare. The retroaortic cancer is overlain by the aortic arch and produces an obstacle in the left approach. In such instances, we use a right posterolateral thoracotomy for exploration and dissection of the tumor. After the tumor is removed and the incision is closed, the patient is turned to the supine position. The stomach is mobilized through a midline laparotomy and the proximal esophagus, with a neck incision. The stomach is then delivered up through a retrosternal tunnel for a neck anastomosis. In the case of an unresectable lesion, a palliative bypass procedure is usually considered, and the left thoracotomy incision gives the most ready approach to the abdominal access. Stomach, colon, or jejunum can be transplanted to the chest through a left diaphragmatic incision. The merits of a left thoracotomy have not been fully appreciated. Exploration and Resectability As preoperative staging of esophageal cancer and cancer of the esophagogastric junction is a less accurate clinical assessment, which often cannot distinguish between localized and advanced disease prior to thoracotomy, and since no single diagnostic study governs the decision regarding resectability, exploratory thoracotomy often becomes necessary in most patients in whom contraindications are not present. It should be emphasized that the purpose of exploration is to stage and to determine the resectability of the tumor. The goal of resection is cure or palliation. Exploration of the lesion must be dealt with in an aggressive and careful manner. Resectability cannot ultimately be determined until the relationship of the tumor to the important adjacent structures is well demonstrated. It should be noted that inadequate mobilization can often give a misleading picture that suggests an unresectable tumor. Exploratory dissection may be the most challenging part of the whole procedure. The criteria of resectability have been elusive but lie somewhere between the one extreme of an overly ag- gressive attitude and the other extreme of accepting only the most favorable tumor pathology for resection. Early Diagnosis and Adjuvant Treatment A high index of suspicion and a cytological examination might be the best approach to an early diagnosis. Best results can be achieved if resection is carried out in a very early stage (more than 90% of 5-year survivors), as indicated by the experience from Lien County of Honan Province in China [9]. However, it seems that widespread screening techniques for detecting an early lesion may not be available in most medical centers. Even in China, the percentage of patients seen early in most hospitals has been small. Screening techniques were not used in this series. In this study, only 1 patient with esophageal cancer was seen early. The search for a better method that can result in a readily obtainable early diagnosis in most hospitals has been pursued with great vigor. However, we predict that in the near future, the patient with a lesion who is seen early will continue to be the minority in the case series. Our retrospective review shows that the most common cause of early death (death within 5 years after standard resection) has been local recurrence within the chest, mainly in the mediastinum. This fact implies that a more aggressive attitude toward resection and adjuvant radiotherapy may be a reasonable approach to treatment of advanced disease. Although there is little evidence to support this, preoperative radiotherapy can always sterilize the mediastinal tumor cells; some clinical reports [lo, 11) suggest that preoperative irradiation produces substantial benefit and improvements both in resectability and long-term results without causing higher surgical mortality and morbidity. Factors Affecting Long-Term Survivors After careful study of all the data available on our patients, we found that for the long-term survivors, there was no correlation between survival and duration of symptoms, age, or sex. The length of the tumor shown in the esophagogram was one of the major criteria in the preoperative decision for exploration, and resectability was higher for patients with a tumor of less than 5 to 6 cm in length. But there was no correlation between tumor length and the long-term results. A tumor of shorter length does not negate a poor prognosis. The favorable prognostic variables in intrathoracic esophageal cancer are lower location, absence of lymph node involvement, and a localized lesion. The patient with cancer of the upper third of the esophagus has a most unfavorable prognosis after resection, even when the lesion is localized or without lymph node involvement. By contrast, the patient with a localized lowerthird cancer may demonstrate a better outcome after standard resection. The presence of lymph node metastasis, either intranodal or perinodal, represents an ominous prognosis, regardless of degree of invasion or location of the lesion. A positive lymph node may be present at any degree of invasion, and there is a significant increase in lymph

181 Lu, Li, Gu: Cancer of Esophagus and Esophagogastric Junction node metastasis with increasing degree of invasion. However, degree of invasion within the esophageal wall may not by itself be a factor influencing long-term results. Tumor invasion of the superficial or deep muscularis layer may carry a similar survival. Role of Resection It is not easy to define the exact value of resection for esophageal cancer (or cancer of the esophagogastric junction) in the literature. A standard partial esophagogastrectomy with esophagogastrostomy has been the choice for almost every patient in our practice. From the data available, we found its value in terms of long-term results is limited only to localized lesions of the lower or middle-third of the esophagus for esophageal cancer (or cancer of the esophagogastric junction) without lymph node metastasis. The 5-year survival was about 50% for esophageal cancer and less than 20% for cancer of the esophagogastric junction. When the tumor invades extraesophageal tissues or there is lymph node metastasis, the prognosis is poor. It should be pointed out that there was a higher incidence of recurrence in the chest after standard resection in our patients, 64% in 381 patients who died within 5 years after resection. Recurrence could be anywhere in the chest. The usual presentation was a mass in the mediastinum, and it could interfere with the anastomosis. Advanced disease was demonstrated in the majority of patients. It is suggested that this higher incidence during that period might be due to the advanced pathological condition combined with a conservative attitude toward resection of the proximal esophagus (the esophagus between the upper margin of gross tumor and the anastomosis). Our policy has been modified since then. We use a supraaortic anastomosis after resection of a lower-third lesion and a neck anastomosis for a middlethird tumor to obtain enough normal proximal esophagus. We hope the results will show improvement in a future study. En bloc resection consists of a total esophagectomy with en bloc dissection of the mediastinal lymph nodes and resection of a substantial portion of the stomach with en bloc dissection of celiac lymph nodes. This operation was described by Logan [l] in 1963. Of 251 resections for cancer of the esophagogastric junction or lower third of the esophagus, 5-year survival was 16%, which was one of the best records at that time. The operative mortality was 21%, and one-third of the operative deaths were caused by anastomotic leak. Probably because of its difficult technique and higher mortality, Logan's procedure was not widely adopted. In 1983, Skinner [2] reported the results of 80 resections performed using Logan's procedure from 1969 to 1981. Skinner extended the surgical indication to include middle-third and cervical esophageal cancers, lowered the surgical mortality to 11%, and improved the longterm results (18% 5-year actuarial survival). The best finding is its effectiveness in control of local disease. Only 3 (4%) of 71 surgical survivors had local recurrence at the anastomotic site. We do not think en bloc dissec- tion is necessary for a localized lesion without lymph node metastasis because the 5-year survival after standard resection for a localized lesion in patients with cancer of the lower or middle third of the esophagus was 50.6% (133/263) and the operative mortality, 5.6%. In contrast, in patients with an advanced lesion, the results after standard resection have been very poor. The overall 5-year survival was less than 8%. It was 7.3% (4/55) for patients with a lesion with extraesophageal invasion with or without lymph node metastasis, and 6.3% (13/208) for patients with a tumor with lymph node metastasis. It is obvious that standard resection has little effect in the advanced disease. We do not know if the long-term results may be further improved by extensive resection. More patients and time are needed to prove its beneficial effect on 5-year survival. To improve the long-term results, efforts should be directed toward accurate surgical and pathological staging (to define the localized lesion), decreasing surgical mortality, combined treatments, and en bloc resection in certain instances. We thank David G. Bryant, M.Sc., Ph.D., Associate Professor of Biostatistics, Memorial University of Newfoundland, Nfld, Canada, for his expert advice and assistance in the statistical analysis. References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. Logan A: The surgical treatment of carcinoma of the esophagus and cardia. J Thorac Cardiovasc Surg 46:150, 1963 Skinner DB: En bloc resection for neoplasms of the esophagus and cardia. J Thorac Cardiovasc Surg 85:59, 1983 Ellis FH Jr, Gibbs SP: Esophagogastrectomy for carcinoma: current hospital mortality and morbidity rates. Ann Surg 190:699, 1979 Wu YK, Chen PT, Fang JP, Lin SS: Surgical treatment of esophageal carcinoma. Am J Surg 1392305, 1980 Huang GJ, Zhang DW, Wang GQ: Surgical treatment of carcinoma of the esophagus: report of 1647 cases. Chin Med J [Engl] 94:305, 1981 Xu L-T, Sun Z-F, Li Z-J, Wu L-H: Surgical treatment of carcinoma of the esophagus and cardiac portion of the stomach in 850 patients. Ann Thorac Surg 35:542, 1983 Lu YK, Li YM, Chen TM, Keng CC: An analysis of mortality and postoperative complications of esophageal resection for cancer. Chin Med J [Engl] 83:39, 1964 Lu YK, Wang TH, Ku YC, Chen TM: Surgical consideration in anastomotic leakage following esophageal resection for cancer. Chin Med J [Engl] 84:612, 1965 Coordinating Group for Research on Esophageal Cancer: Early diagnosis and surgical treatment of esophageal cancer under rural conditions. Chin Med J [Engl] 2:113, 1976 Huang GJ, Gu XZ, Zhang RG: Combined preoperative irradiation and surgery in esophageal carcinoma: report of 408 cases. Chin Med J [Engl] 94:73, 1981 Marks RD Jr: Preoperative radiation therapy for carcinoma of the esophagus. Cancer 38:84, 1976 Steel RGD, Torrie JH: Principles and Procedures of Statistics. New York, McGraw-Hill, 1960, pp 109-110