In most Western countries, esophageal adenocarcinoma

Transcription

1 Nodal Metastasis and Sites of Recurrence After En Bloc Esophagectomy for Adenocarcinoma Geoffrey W. B. Clark, FRCS(Ed), Jeffrey H. Peters, MD, Adrian P. Ireland, FRCS(I), Afshin Ehsan, BS, Jeffrey A. Hagen, MD, Milton T. Kiyabu, MD, Cedric G. Bremner, FRCS, and Tom R. DeMeester, MD Departments of Surgery and Pathology, University of Southern California, Los Angeles, California The operative specimens from 43 patients undergoing en bloc esophagectomy for adenocarcinoma of the lower esophagus or cardia were analyzed. Depth of invasion of the tumor and extent and location of lymph node metastases were determined. Postoperative recurrence was identified from positive findings on successive 3-month computed tomographic scans. Positive nodes occurred in 33% (2/6) of intramucosal tumors, 67% (6/9) of intramural tumors, and 89% (25/28) of transmural tumors (p < 0.01). Commonly involved nodes were those in the lesser curve of the stomach (42%), parahiatal nodes (35%), paraesophageal nodes (28%), and celiac nodes (21%). Excluding perioperative deaths, follow-up was complete for 38 patients. Twenty patients had recurrence. Fifteen patients (40%, 15/38) had nodal recurrence: cervical, 7.9% (3/38); superior mediastinal, 21% (8/38); and abdominal, 24% (9/38) (retropancreatic in 7 and retrocrural in 2). Of 5 patients with nodal recurrence alone, 3 (%) had recurrence at sites outside the margins of resection. Patients with four metastatic nodes or less had a survival advantage over those with more than four (p < 0.05). There was no difference in survival according to location of nodal metastases. Two (22.2%) of 9 patients with celiac node metastases survived longer than 4 years. Adenocarcinoma of the lower esophagus and cardia spreads widely to mediastinal and abdominal nodes, and death can occur from nodal disease. Rates of lymph node metastases increase with the depth of the primary tumor. Patients with lymphatic metastases can be cured particularly if there are fewer than four nodes involved. Curative surgical therapy necessitates wide lymph node resection to ensure removal of all metastatic nodes. (Ann Thorae Surg ) In most Western countries, esophageal adenocarcinoma has replaced squamous cell carcinoma as the most common esophageal malignancy [1, 2]. Surgical resection remains the only means of cure. Controversy over the ideal surgical treatment centers on the benefit of extended mediastinal and abdominal lymphadenectomy [3]. Several authors [4-7] have reported a survival advantage with extended lymphadenectomy, whereas others [8, 9], less convinced, continue to pursue a more palliative approach. Relevant to this controversy is information regarding patterns of nodal metastases at the time of resection. Critical analysis of the patterns of lymph node metastases in the surgical specimens after transhiatal resection is inappropriate, as a limited nodal dissection is performed [10], and detailed analysis of patients undergoing en bloc resection has been reported only for esophageal squamous cell carcinoma [11, 12]. We performed a detailed analysis of the extent of nodal spread in the operative specimens from patients undergoing curative en bloc resection for adenocarcinoma of the distal esophagus or cardia. Complete follow-up for all patients was obtained, and the presence of recurrent disease was assessed at regular intervals, allow- Presented at the Thirtieth Annual Meeting of The Society of Thoracic Surgeons, New Orleans, LA, Jan 31-Feb 2,1994. Address reprint requests to Dr DeMeester, Department of Surgery, University of Southern California School of Medicine, 1510 San Pablo St, Suite 514, Los Angeles, CA by The Society of Thoracic Surgeons ing correlation of patterns of recurrence with original operative findings. Material and Methods The study population comprised 43 patients who had an en bloc esophagectomy for adenocarcinoma of the distal esophagus or cardia. Surgical therapy was the primary treatment. None of the patients had preoperative radiotherapy or chemotherapy. Postoperative chemotherapy was reserved for patients with recurrent disease. The age range of the patients was 25 to 77 years (median age, years). Forty patients were male and 3 were female. Patient Selection Patients were selected for en bloc resection based on potential for cure. All patients were physiologically fit: less than 75 years of age with a forced expiratory volume in 1 second of greater than 1.25 L and a cardiac ejection fraction of greater than Patients with poor physiologic variables, advanced local disease, or systemic metastases were not considered suitable for the en bloc procedure. Advanced local disease was defined by the presence of transmural spread and clear evidence of more than four macroscopically enlarged malignant nodes. The decision to perform a curative en bloc esophagectomy or a palliative transhiatal esophagectomy was made either preoperatively or intraoperatively /94/$7.00

2 Ann Thorae Surg CLARK ET AL 647 Operative Approach The operation begins with an abdominal exploration to assess the extent of the primary tumor and the regional nodal status. If there is evidence of late-stage disease or involvement of the porta-hepatis or subpancreatic nodes on frozen section, a transhiatal resection is performed. If the abdominal findings are favorable, the abdomen is closed, the patient is repositioned, and a right posterolateral thoracotomy is performed. Biopsy specimens of the lower paratracheal nodes are obtained, and if positive, the esophagus is removed by a simple dissection. Metastases in the portal nodes or the lower paratracheal nodes signify disease at the margins of the resection, thereby precluding a curative intent. If the nodes are negative, the intrathoracic esophagus is removed with an en bloc dissection of the adjacent bronchial, subcarinal, paraesophageal, and parahiatal lymph nodes. The block of tissue removed is bounded superiorly by the upper border of the azygos vein and the lower border of the aortic arch, inferiorly by the superior border of the pancreas, anteriorly by the membranous portion of the main bronchi and pericardium, laterally by the right and left mediastinal pleura, and posteriorly by the vertebral column and the aorta. After the thoracic dissection, the thoracotomy is closed. The patient is repositioned, and the procedure is continued through the reopened abdominal incision with an en bloc resection of the proximal three quarters of the stomach, the greater omentum, and the spleen and splenic artery with systematic removal of the hepatic, celiac, left gastric, and splenic artery lymph nodes. Included in the resection is the posterior areolar tissue from behind the crura and a rim of the diaphragmatic muscle around the hiatus. Restoration of gastrointestinal continuity was achieved by a left colon interposition in the majority of patients. Postoperative Staging The length of the tumor was recorded from the operative specimen, and the depth of invasion was determined histologically. Intramucosal tumors involved the lamina propria but were limited by the muscularis mucosa. Intramural tumors had spread through the submucosa and into the muscularis propria but had not penetrated through the wall. Transmural tumors had penetrated through the esophageal wall into the adventitia. The total number of lymph nodes in each specimen and their locations were recorded according to the sites shown in Figure 1. Paraesophageal nodes were those lymph nodes located adjacent to the subcarinal intrathoracic esophagus and encompassed both middle and lower mediastinal node groups. Parahiatal or paracardial nodes were those nodes located in close proximity to the gastroesophageal junction. Histologic examination of all resected nodes was performed to identify the presence of metastatic disease. Barrett's esophagus was diagnosed when specialized intestinal metaplasia was present adjacent to the tumor on histologic examination of the resected esophagus. The tumors were classified according to a staging system proposed by Skinner and colleagues [13]: Lower paratracheal Subcarinal Paraesophageal ParahiatalfParacardial Splenic Hilum Splenic Artery Greater Curve Lesser Curve Left Gastric Celiac Hepatic Portal Right Gastric Retropancreatic Fig 1. Locations of lymph nodes. Frozen section biopsies are performed at the beginning of the procedure to exclude disease at the margins of the resection (lymph node groups 12, 14, and 1). When these nodes are negative, en bloc resection is performed with complete removal of lymph node groups 2 through 11 and 13. Included in the resection are a small number of tracheobronchial nodes (not shown), portal nodes, and retropancreatic nodes. Wall penetration WO = Tumor limited to mucosa WI = Tumor penetration through submucosa and into but not through muscle layers W2 = Tumor penetration through muscle layers Lymph node involvement NO = No lymph node involvement N1 = One to four lymph nodes involved N2 = Five or more lymph nodes involved To calculate the effect of location of the metastatic nodes on survival, patients were stratified according to the level of the most distant involved nodes: Level 0 No nodes Level I Paraesophageal Parahiatal Lesser curve Level II Splenic artery Splenic hilum Greater curve Left gastric artery

3 648 CLARK ET AL Ann Thorac Surg % OF TUMORS WITH LYMPH NODE METASTASIS o 2/6 INTRAMUCOSAL 619 INTRAMURAL TUMOR DEPTH * 25/28 TRANSMURAL Fig 2. Prevalence of lymph node metastases according to depth of esophageal wall penetration by tumor. (* p < 0.01;K = 9.3,2 dfj Level III Portal Hepatic Celiac Subcarinal Tracheobronchial Subpancreatic Paratracheal Right gastric artery Follow-up Hospital survivors were followed up with laboratory studies, a chest roentgenogram, and a chest and abdominal computed tomographic scan at 3-month intervals for the first 3 years and then every 6 months. Evidence of nodal recurrence was determined by the presence of enlarging nodes on sequential computed tomographic scans. Statistical Analysis The K- test or Fisher's exact test was used for comparisons of proportions. Nonparametric data were compared by the Mann-Whitney U test. Survival estimates were calculated using the Kaplan-Meier method, and comparisons between survival curves were tested by the log-rank test. Significance was taken at the 0.05 level. Results Characteristics of Resected Specimens The average number of lymph nodes examined after en bloc resection was 42 ± 20 (range, 16 to 98). A total of 1,784 lymph nodes were examined, and 206 (11.5%) were found to be positive. Lymph node metastases were present in 33 (76.7%) of the 43 patients. In 29 (67.4%) of 43 tumors, the epicenter was in the lower esophagus and in 14 (32.6%), in the cardia. The depth of the primary tumor was intramucosal in 6 patients, intramural in 9, and transmural in 28. Relationship of Nodal Metastases to Depth and Length of Primary Tumor Figure 2 shows the prevalence of lymph node metastases related to the depth of the primary tumor. Of importance, % OF TUMORS WITH NODAL METASTASES o (n=8) (n=9) (n=6) (n=5) (n=15) <2 Node Location 2 ~ ~ ~ - 5 LENGTH OF TUMOR (CM) Fig 3. Prevalence of nodal metastases according to length of tumor in pathologic specimen. 33% of intramucosal adenocarcinomas (2/6) had regional nodal metastases, one with parahiatal node involvement and the other with lesser curve node involvement. Increased tumor length correlated directly with the prevalence of nodal metastases (Fig 3). Pattern of Nodal Metastases in Esophageal Adenocarcinoma Nodal spread from adenocarcinoma of the lower esophagus or cardia was preponderantly to the paraesophageal (27.9%), parahiatal (34.9%), lesser curve (41.9%), and celiac nodes (20.9%). Results of analysis of lymph node positivity according to total number of nodes examined in each location is shown in Table 1. The results indicate that metastatic nodes are clustered around the lower esophagus Table 1. Pattern of Lymph Node Spread in Resected Tumors of Lower Esophagus and Cardia' Tracheobronchial Subcarinal Paraesophageal Parahiatal Splenic hilum Splenic artery Greater curve Lesser curve Left gastric Celiac Hepatic Portal Right gastric Retropancreatic a Numbers in parentheses are percentages. No. of Patients With Positive Nodes Total No. of (n = 43) Nodes Resected 4 (9.3) 12 (27.9) 15 (34.9) 2 (4.7) 4 (9.3) 18 (41.9) 3 (7.0) 9 (20.9) 0(0) >5 No. of Positive Nodes 1 (2.4) 9 (2.3) 37 (11.7) 35 (14.2) 3 (7.7) 2 (2.8) 10 (11.2) 64 (24.5) 8 (8.5) 25 (41.7) 6 (28.6) 2 (2.4) 4 (8.5) 0(0)

4 Ann Thorae Surg 1994;58: CLARK ET AL 649 Table 2. Pattern of Lymph Node Spread According to Tumor Depth Intramucosal Adenocarcinoma (n = 6) Intramural Adenocarcinoma (n = 9) Transmural Adenocarcinoma (n = 28) No. of Patients No. of Patients No. of Patients With Positive Percentage With With Positive Percentage With With Positive Percentage With Node Location Nodes Positive Nodes Nodes Positive Nodes Nodes Positive Nodes Tracheobronchial Subcarinal Paraesophageal Parahiatal Splenic hilum Splenic artery Greater curve Lesser curve Left gastric Celiac Hepatic Portal Right gastric and cardia but may spread in small numbers to more distant locations. The location of the esophageal tumors was determined endoscopically according to the distance from the incisors to the proximal and distal margins of the lesion. There were eight adenocarcinomas arising between 25 and 34 em and 35 between 35 and 45 em. Higher esophageal tumors showed a slight preponderance to spread to paraesophageal nodes and lower tumors to abdominal nodes, but differences were not significant. Similarly, there was no significant difference in the distribution of nodal metastases when lower esophageal tumors were compared with tumors whose epicenter was in the cardia. Table 2 shows that the distribution of lymph node metastases fans out according to the depth of tumor invasion. The frequency and the extent of nodal metastases increase with greater wall penetration. % SURVIVAL I ~ : ; - : ~ - " " : ; { Comparison was made between the 23 patients with adenocarcinomas arising in Barrett's mucosa and the 20 patients in whom Barrett's mucosa could not be found. The patients with adenocarcinomas arising in Barrett's mucosa were seen at an earlier stage than patients without Barrett's adenocarcinomas 00/23 versus 18/20 transmural tumors; p < 0.01 by Fisher's exact test). Despite this, the pattern of nodal metastases in the two groups was similar. Survival According to Distribution of Nodal Metastases Patients with four involved lymph nodes or less had a survival advantage over patients with more than four metastatic nodes (,i = 4.7, P< 0.05) (Fig 4). Figure 5 shows that patients with intramucosal tumors and those with tumors 2 em in diameter or smaller were more apt to have four or fewer nodal metastases and hence a better survival. When patients were stratified according to the four levels of location of nodal involvement, there was no significant difference between the survival curves. Outcome in Patients With Celiac Node Metastases Celiac node involvement has been considered by some authors [l4] to be synonymous with M-stage disease, thus precluding long-term survival. Figure 6 shows that the ::;:4 nodes, n=27 INTRAMUCOSAL INTRAMURAL TRANSMURAL PERCENT I (7) >4 nodes, n=16 0% 11.1 % 53.6% (6) (3) x'=6.4. p<o.oi PERCENT 46.7% MONTHS Fig 4. Kaplan-Meier survival curve according to number of lymph node metastases: four or less versus more than four. Bars indicate the standard error. (K = 4.7; p < 0.05 by log-rank analusis.) >5 (em) Difference not significant Fig 5. Correlation between depth and length of primary tumor with more than four nodal metastases.

5 650 CLARK ET AL Ann Thorac Surg ] 994;58: % SURVIVAL (7) (6) (5) (3) Celiac node (+), n=9 OL o MONTHS Fig 6. Kaplan-Meier survival curve of patients with celiac node metastases compared with that of patients without eeliae node involvement. The difference was not significant by log-rank analysis CY! = 0.9; P = 0.3). Bars indicate the standard error. survival curve of patients with celiac node involvement was not significantly different from the curve of those without celiac involvement. Although most patients with celiac node metastases have recurrence, celiac metastases did not preclude long-term survival, as 2 patients survived for 56 and 68 months. Patterns of Recurrence Five patients who underwent en bloc resection did not survive the perioperative period (operative mortality rate, 11.6%). The median duration of follow-up for the 38 remaining patients is 16.5 months (range, 3 to 107 months), and 18 patients are alive and disease free. Twenty patients (52.6%) have had development of recurrent disease after resection (Table 3), 15 of whom have died. NODAL RECURRENCE. Fifteen patients (39.5%) had nodal recurrence at one or more locations (see Table 3). The sites of nodal recurrence are shown in Figure 7. Nine patients (23.7%) had abdominal nodal recurrence, 8 (21.1%) had intrathoracic nodal recurrence, and 3 (7.9%) had cervical nodal recurrence. In only 3 of the 15 patients was the nodal recurrence located within the area of dissection: in retrocrural nodes, an area in which it is difficult to achieve a complete nodal clean-out, in 2 and in the region of the right hilum in 1. Four patients free from systemic disease died of nodal metastases, and 1 patient underwent a successful resection of recurrent nodal disease. The location of these nodal recurrences was outside the area of the dissection in 3 of the 5 patients. All 9 patients with abdominal nodal recurrences had involved abdominal nodes at the time of operation, whereas no patient with negative abdominal nodes at operation had development of an abdominal recurrence. The main site of abdominal nodal recurrence was in the undissected retropancreatic region (7 patients). Two patients had recurrence in the retrocrural nodes. Intrathoracic nodal recurrence developed in 8 patients, only 4 of whom had positive thoracic nodes at the time of resection. All patients with thoracic nodal recurrence had involvement of the upper mediastinal nodes, a finding suggesting that these recurrences may have arisen from nodes situated along the recurrent laryngeal nerve that were not resected. One patient had superior mediastinal node recurrence with a tumor mass extending into the middle mediastinum, and 1 patient had recurrence around the right hilar region. Three patients (7.9%) had cervical nodal recurrence. SYSTEMIC RECURRENCE. Fifteen patients (39.5%) had systemic recurrence, with the most common sites being the liver (26.3%, 10/38), lungs 03.2%,5/38), and bone 00.5%, 4/38) (see Table 3). Ten of the 15 had concurrent nodal recurrence, whereas 5 had systemic metastases only. Of 10 patients who had negative nodes at operation (3 with intramural and 3 with transmural wall penetration), only 1 had development of systemic metastases, which suggests that esophageal adenocarcinoma is not a systemic disease from the outset. ANASTOMOTIC RECURRENCE. Four patients 00.5%) had recurrence of tumor at the cervical anastomotic site between 30 and 56 months after resection. The esophageal resection margin was free from microscopic tumor in each patient. The length of the esophagus proximal to the gross tumor in the fixed specimen was 6, 6.5, 8, and 11.5 ern. The distance in the fixed specimen is approximately half the distance in the fresh unfixed specimen. In 1 patient there were microscopic foci extending to within 1 ern of the proximal resection margin. Two of these patients had development of cervical nodal recurrence before the anastomotic recurrence, and the latter was considered to be secondary to direct invasion from the nodal metastases rather than a true anastomotic recurrence. The rate of true anastomotic recurrence was 2/38 (5%). Comparison Between Patients With and Without Recurrence Seven of the 18 disease-free patients have been followed for more than 3 years (median survival, months) (Table 4). They were compared with the 20 patients who had development of recurrent disease. Patients who were disease free had less esophageal wall penetration than patients with recurrence (2/7 with W2 tumors versus 16/20; p < 0.05 by Fisher's exact test). Six of the 7 disease-free patients had four or fewer involved nodes in the resected specimen compared with only 7 of 20 with recurrence (p < 0.05 by Fisher's exact test). The ratio of metastatic lymph nodes to the total number of resected lymph nodes was also lower in patients who remained disease free compared with those who showed recurrence 0:20 versus 1:6; p < 0.05 by Mann-Whitney U test) (Fig 8). Comment The aim of this study was to describe the extent of nodal metastases in patients undergoing extended thoracic and abdominal lymphadenectomy during curative en bloc re-

6 Pathologic Tumor Stage and Location of Metastasis in 20 Patients With Recurrence Operative Findings Nodal Recurrence Systemic Recurrence Local Rec Survival Abdominal Thoracic No. of No. % Stage (mo) Nodes Nodes Nodes Positive Positive Abdominal Thoracic Cervical Liver, Lung, Bone Anasto WONO W2Nl WI Nl WI Nl W2Nl W2Nl W2Nl W2N W2N W2N W2N W2N WI N W2N W2N W2N W2N W2N W2N W2N n

7 652 CLARK ET AL Ann Thorae Surg 0.7 RATIOOF INVOLVEDvs. TOTAL NODES RESECTED Retropancreatic (7/38) 18l1J1I - Superior mediastinuml (8138) 21% IF,,,,,,,,,;;;;,.r7\ Aortopulmonary -" '. -,\ Fig 7. Sites of lymph node recurrence after en bloc esophagectomy. Margins of the resection are indicated by the broken line. section of adenocarcinoma of the distal esophagus or cardia. The results provide the basis for several useful clinical principles in the management of this disease. Adenocarcinomas of the lower esophagus and cardia spread widely to regional nodes, most commonly to nodes along the lesser curvature, celiac axis, and parahiatal regions. However, 9% of patients had involved subcarinal nodes and 16% had involved nodes of the splenic hilum, along the splenic artery, or along the greater curvature of the stomach. It is likely that the subcarinal and splenic nodes would remain after transhiatal resection. Further, the involved nodes along the greater curvature of the stomach would be, by necessity, transposed into the chest.' :20 DISEASEFREE>3 YEARS (n=7) 1:6 RECURRENT DISEASE (n=20) Fig 8. Ratio of involved nodes to total number of resected nodes in disease-free survivors (overall survival, >3 years; median survival, months) compared with patientswith recurrence (p < 0.05 by Mann Whitney U test). The horizontal line represents the mean. if the stomach was used to reestablish gastrointestinal continuity. In addition, more than 20% of patients had positive celiac axis nodes, an area not dissected by most surgeons during transhiatal esophagectomy. The observations that 7 patients with nodal involvement have survived longer than 3 years and that nodal disease was responsible for the death of 4 of 23 patients free from systemic metastases suggest that an en bloc resection has some advantages for patients with early disease [6]. Our findings confirm previous reports [13] showing that patients with metastases to four lymph nodes or less had a significant survival advantage. We could find no relationship between location of the nodal metastases and survival. It is interesting to note that all patients without metastatic abdominal nodes identified during abdominal lymphadenectomy remained free from abdominal recurrence. This was not so for thoracic disease; 4 patients who were free from thoracic nodal involvement at the time of resection had intrathoracic recurrence. This emphasizes that upward lymphatic spread occurs and stresses the importance of the intrathoracic dissection, the portion where liberties are taken during transhiatal dissection. In 80% of patients who had nodal recurrence, the site Table 4. Pathologic Tumor Stage in Patients Who Survived Disease Free Longer Than 3 Years Patient Survival Abdominal Thoracic No. of No. % No. Stage (mo) Nodes Nodes Nodes Positive Positive 1 WONO WI NO WONO WONI WI Nl W2Nl W2N Median Mean

8 Ann Thorac Surg CLARK ET AL 653 was outside the limits of the resection. All patients with thoracic nodal recurrence had disease in their upper mediastinum or aortopulmonary window, a finding suggesting that the recurrences arose from nodes lying along the recurrent laryngeal nerve that are not routinely removed by the en bloc dissection. Extended mediastinal and neck dissection has been advocated by some authors [12] for the treatment of esophageal squamous cell carcinoma, with improved survival. There are no data on the benefit of this approach for esophageal adenocarcinoma. The increased morbidity and the possibility of permanent hoarseness are factors that have discouraged us from enlarging the operation. We will continue to review this situation as more data on location of recurrence emerge. Perhaps a staged approach should be used where a superior mediastinal and cervical node dissection is done after the patient demonstrates freedom from systemic disease and imaging investigation shows enlarged mediastinal or cervical nodes. Tumor depth was a good indicator of nodal involvement. The finding that 2 of 6 patients with intramucosal lesions had lymph node metastases is in agreement with Japanese investigators [12] who reported a 30% nodal involvement in 40 patients with esophageal squamous cell carcinoma limited to the esophageal mucosa. This finding has implications in the management of high-grade dysplasia in patients with Barrett's esophagus, where the prevalence of unexpected adenocarcinoma is up to 50% [15-17]. If one third of patients with mucosal adenocarcinomas have metastatic nodes and if patients who have less than four nodes involved can have the same outcome as those without nodal involvement [6], then simple esophagectomy may be inappropriate treatment. Rather, an en bloc esophagogastrectomy may be necessary to assure a survival advantage for patients who have benefited from early detection. It has been reported that the finding of metastatic disease involving the celiac nodes precludes survival beyond 2 years [14]. In defiance of this statement, 2 patients with celiac axis involvement survived long term (>4 years) before recurrence developed. This suggests that celiac axis lymphadenopathy may not imply total despair. This experience has impressed on us that the characteristics of patients who remained free from disease were not predictable, although these patients tended to have less wall penetration and a fewer number of metastatic nodes. In conclusion, this report documents the pattern of nodal spread in patients undergoing en bloc resection of adenocarcinoma of the distal esophagus or cardia. The extent of nodal spread can be wide, which emphasizes the need of wide resection of the lesion if cure is to be achieved. The presence of four or fewer nodal metastases was associated with a lower risk of recurrence, but the precise location of the involved nodes had no predictive value. In contrast to currently prevailing opinion, this emphasizes that a more extensive operation should be done for patients with disease detected early. References 1. Lund 0, Hasenkam JM, Aagaard MT, Kimose HH. Timerelated changes in characteristics of prognostic significance of carcinomas of the esophagus and cardia. Br J Surg 1989;76: Blot WI, Devesa SS, Kneller RW, Fraumeni JF. Rising incidence of adenocarcinoma of the esophagus and gastric cardia. JAMA 1991;265: Wong J. Esophageal resection for cancer: the rationale of current practice. Am J Surg 1987;153: DeMeester TR, Zaninotto G, Johansson KK. Selective therapeutic approach to cancer of the lower esophagus and cardia. J Thorac Cardiovasc Surg 1988;95: Skinner DB, Ferguson MK, Soriano A, Little AG, Staszak VM. Selection of operation for esophageal cancer based on staging. Ann Surg 1986;204: Hagan JA, Peters JH, DeMeester TR. Superiority of extended en-bloc esophagectomy for carcinoma of the lower esophagus and cardia. J Thorac Cardiovasc Surg 1993;106: Lerut T, DeLeyn P, Coosmans W, Van Raemdonck D, Scheys I, Lesaffre E. Surgical strategies in esophageal carcinoma with emphasis on radical lymphadenectomy. Ann Surg 1992;216: Orringer MB. Transhiatal esophagectomy without thoracotomy for carcinoma of the thoracic esophagus. Ann Surg 1984;200: Orringer MB, Orringer JS. Esophagectomy without thoracotomy. A dangerous operation? J Thorac Cardiovasc Surg 1983;85: Orringer MB, Marshall B, Stirling MC Transhiatal esophagectomy for benign and malignant disease. J Thorac Cardiovasc Surg 1993;105: Akiyama H, Tsurumaru M, Kawamura T, Ono Y. Principles of surgical treatment for carcinoma of the esophagus: analysis of lymph node involvement. Ann Surg 1981;194: Isono K, Sato H, Nakayama K. Results of a nationwide study on the three field lymph node dissection of esophageal cancer. Oncology 1991;48: Skinner DB, Dowlatshahi KD, DeMeester TR. Potentially curable cancer of the esophagus. Cancer 1982;50: Aikou T, Altorki NK, Skinner DB, Baba M, Shimazu H. Lymph node dissection for carcinoma of the cardia: is it worthwhile? In: Nabeya K, Hanaoka T, Nogami H, eds. Recent advances in diseases of the esophagus. Tokyo: Springer-Verlag, 1994: Hameeteman W, Tytgat GNI, Houthoff HJ, van den Tweel JC Barrett's esophagus: development of dysplasia and adenocarcinoma. Gastroenterology 1989;96: Pera M, Trastek VF, Carpenter HA, Allen MS, Deschamps C, Pairolero PC Barrett's esophagus with high-grade dysplasia: an indication for esophagectomy? Ann Thorac Surg 1992;54: Altorki NK, Sunagawa M, Little AG, Skinner DB. High grade dysplasia in the columnar lined esophagus. Am J Surg 1991; 161:97-9. DISCUSSION DR ROBERT J. GINSBERG (New York, NY): It did not seem to matter what the dissection was. It was the number of nodes involved that ultimately caused the recurrence. So how do you conclude the value of a wide en bloc resection? DR CLARK: We think that the en bloc resection is important in that we achieve a complete nodal clearance. The problem is that many of the lymph nodes that were removed were not obviously enlarged at the time of operation, and there is probably quite a

9 654 CLARK ET AL Ann Thorae Surg considerable group of patients who have micrometastases within the nodes. Therefore, we believe that the chances of survival are heavily based on the surgeon's ability to perform an effective cancer operation in that he or she must remove all the potentially involved lymph nodes. DR GINSBERG: So, if you have four lymph nodes involved within the region of the primary tumor, would you suggest a wide resection or not? It appeared as though if involved lymph nodes were at either end of your resection, you did not do a wide dissection. Why was that? DR CLARK: The problem is that when we do the resection, these patients are considered to have early cancer; we clinically consider them to have early cancer from the computed tomographic scans and from the endoscopic ultrasound images. By early cancer, we intend to select patients who have four lymph node metastases or less, have relatively small tumors, and are clearly free from systemic metastases. However, we have previously shown a stage-for-stage survival advantage for en bloc esophagectomy compared with transhiatai resection, which included patients who had more than four lymph node metastases. We also do intraoperative staging. If the patient has gross lymph node involvement that was not suspected preoperatively, we convert to either a simple esophagectomy or a transhiatal esophagectomy. Once the operation has commenced, you are not going to know the precise number of lymph nodes involved until you get the operative specimen back. If we know for certain the patient has more than four lymph node metastases, then we do not recommend en bloc esophagectomy. DR MICHAEL BOUSAMRA (Milwaukee, WI): My question concerns the subcarinallymph nodes, the greater curvature lymph nodes, and the splenic lymph nodes. Obviously if they are involved, they may not be removed with a transhiatal resection. I wonder whether they were involved as an isolated entity or whether they were always involved with other lymph nodes. DR CLARK: I could not say for certain, but as a general rule, patients had other sites of lymph node metastases. I do not think there was any patient who had involvement of only the greater curve nodes or only the subcarinal nodes. DR GINSBERG: There already are two staging classifications for nodes, involving the numbering of nodes, that of the Japanese and that proposed by the intergroup. Why do you have a third? Also, we have a T status from the TNM classification that is exactly the same as your depth of wall. Why not refer to it as the T status and make it uniform throughout the world? DR CLARK: We have traditionally used Dr Skinner's classification of wall penetration being Wand nodal involvement being NO, Nl, and N2. We prefer to use a non-tnm staging system because the TNM system lacks an intermediate stage for limited nodal disease, which we think is important. We have found that our staging is useful in the analysis of clinical data and that the system is a good predictor of outcome, but we could have used a different system. Bound volumes available to subscribers Bound volumes of the 1993 issues of The Annals of Thoracic Surgery are available only to subscribers from the Publisher. The cost is $96.00 (outside US add $20.00 for postage) for volumes 55 and 56. Each bound volume contains a subject and author index, and all advertising is removed. The binding is durable buckram with the name of the journal, volume number, and year stamped on the spine. Payment must accompany all orders. Contact Elsevier Science Inc, 655 Avenue of the Americas, New York, NY 10010; or telephone (212) (facsimile: (212) ).