Impact of Radical Systematic Mediastinal Lymphadenectomy on Tumor Staging in Lung Cancer

Transcription

1 Impact of Radical Systematic Mediastinal Lymphadenectomy on Tumor Staging in Lung Cancer Jakob R. Izbicki, MD, Bernward Passlick, MD, Ortrud Karg, MD, Christian Bloechle, MD, Klaus Pantel, MD, Wolfram T. Knoefel, MD, and Olaf Thetter, MD Department of Surgery and Institute of Immunology, University of Munich, Munich, and Division of Thoracic Surgery, Central Hospital Gauting, Gauting, Germany The extent of lymphadenectomy in the treatment of non-small cell lung cancer is still a matter of controversy. While some centers perform mediastinal lymph node sampling with resection of only suspicious lymph nodes, others recommend a radical, systematic mediastinal lymphadenectomy (LA) to improve survival and achieve a better staging. Herein we report on the impact of LA on tumor staging in a controlled, prospective, randomized clinical trial comparing lymph node sampling and LA in a total of 182 patients with operable non-small cell lung cancer. Regardless of the type of lymphadenectomy performed, the percentage of patients with pathologic N1 or N2 (sampling: n = 23, 23.0%; LA: n = 22, 26.8%) disease was very similar in both groups, indicating that systematic radical lymphadenectomy is not an essential prereq- uisite to determine the N stage of a patient. In contrast, the number of patients detected to have lymph node involvement at multiple levels was significantly increased by LA. In the lymph node sampling group only 4 of 23 patients (17.4%) with N2 disease were found to have more than one lymph node level involved, whereas LA results in the detection of excessive N2 disease in 12 of 21 patients (57.2%; p = 0.007), which was associated with a shorter distant metastases-free (p = 0.021) and overall survival. In conclusion, LA is not essential to determine the N stage of a patient, but results in a more detailed staging of the N2 region, which is of prognostic significance. Therefore, it might be useful to identify patients with a higher risk for tumor relapse. (Ann Thorac Surg 1995;59:209-14) he extent of lymphadenectomy in surgical treatment T of non-small cell lung cancer (NSCLC) is still a matter of controversy. Although some authors claim that a radical systematic mediastinal lymphadenectomy should be performed in almost every patient with a resectable primary tumor [1-3], others, including those of the American Lung Cancer Study Group, recommend mediastinal lymph node sampling as the treatment of choice in order to reduce perioperative morbidity and mortality [4-6]. Postulated arguments for a radical lymphadenectomy are better local tumor control with a prognostic benefit, especially for patients with N2 disease, and a complete nodal staging of the tumor. We performed a prospective, randomized, controlled clinical trial to compare the results of conventional lymphadenectomy (LS) with radical systematic lymphadenectomy (LA) in patients undergoing curative resection of NSCLC. Recently we were able to report that radical lymphadenectomy can be performed with an acceptable morbidity and mortality, but we were not able to show a significant effect on survival [7]. Herein we report on the effect of radical lymphadenectomy on nodal staging of NSCLC. Accepted for publication Aug 12, Address reprint requests to Dr Izbicki, Department of SurgeD', University of Hamburg, Martinistra/?e 52, Hamburg, Germany. Patients and Methods This trial was approved by the Ethical Research Committee of the Medical Faculty, University of Munich. Patients were evaluated, randomized, treated and followed up at the Department of Surgery, University of Munich; Division of Thoracic Surgery, Central Hospital Gauting; and Department of Pulmonary Medicine, Central Hospital Gauting. Eligibility Criteria Patients of any age and sex with a curatively resectable NSCLC were eligible for this study. Tumor-associated exclusion criteria were evidence of distant metastasis (M1 disease), contralateral or supraclavicular nodal involvement (N3 disease) and confirmation of extensive N2 stage by computed tomographic scan (more than one station of the N2-region involved). Patient-associated exclusion criteria were previous or coexistent malignant disease, severe heart failure, renal insufficiency (creatinine level greater than two times normal upper limit), myocardial infarction less than 6 months earlier, liver cirrhosis, and insuffécient pulmonary reserve. After randomization, patients were excluded if they had evidence of intrapulmonary metastases. Patients whose resection specimen exhibited residual tumor at the resection margin also were excluded, as were patients 1995 by The Society of Thoracic Surgeons /95/$ (94)00717-L

2 210 IZBICKI ET AL Ann Thorac Surg IMPACT OF LYMPHADENECTOMY ON TUMOR STAGING 1995;59: whose tumor was subsequently classified as small cell lung cancer. Diagnostic Evaluation Each patient was assessed by plain chest roentgenogram, bronchoscopy with peribronchial or intraluminal biopsies, computed tomographic scan of thorax and abdomen, abdominal ultrasound, and bone scan. Only patients with enlarged contralateral lymph nodes (more than 1 cm in short-axis diameter) underwent a preoperative mediastinoscopy/biopsy to exclude or confirm N3 disease. For assessment of functional operability, body plethysmography, preoperative arterial blood gas analysis, and bicycle ergonometry were performed. Randomization and Surgical Technique At thoracotomy, eligible patients were assigned randomly (using computer-generated random numbers) into a regional lymphadenectomy with mediastinal lymph node sampling group (LS) or radical systematic lymphadenectomy group (LA). The surgical approach was by anterolateral thoracotomy in the fourth intercostal space. The technique of resection of the primary lung tumor was the same in both groups, consisting of a classic lobectomy or pneumonectomy, in some cases combined with bronchoplastic or sleeve-resecting procedures. Tumors that exhibited adherence to neighboring structures or organs such as pericardium, thoracic wall, diaphragm, or pulmonary artery were treated by extended resections with en-bloc removal of lobe or lung with adhering structures. Confirmation of tumorous invasion of these structures was not attempted intraoperatively. The bronchial stump routinely was closed with a linear stapler TA 30 (United States Surgical Instruments, Norwalk, CT) and covered with a pericardial or pleural flap. In the LS group the resection was combined with a regional lymphadenectomy of interlobar, peribronchial, and hilar nodes representing nodes 10, 11, and 12 according to the lymph node mapping of the American Thoracic Society [1]. A mediastinotomy was performed by longitudinal incision of the mediastinal pleura and nodes of regions 2 to 9 were explored. Any nodes suspicious of cancer were removed and submitted to pathohistologic analysis. In modification to suggestions of the American Lung Cancer Study Group [4, 5] only nodes of regions 4, 5, and 7 were removed in all patients routinely. In the LA group resection was combined with a radical systematic en-bloc mediastinal lymphadenectomy as described by Naruke and colleagues [8] and Martini and Flehinger [9] and described by us previously [7]. Briefly, the superior mediastinal compartment contained between trachea, superior vena cava from the level of the azygos vein to the right subclavian artery, and right recurrent laryngeal nerve was dissected and the trachea, azygos vein, superior vena cava, and ascending aorta were completely freed from all tissue. Azygos vein and vagus nerve were generally spared and the right laryngeal nerve was exposed. As a modification of the technique given by Martini and Flehinger [9], the mediasti- num anterior to the superior vena cava was routinely included in the dissection including thymectomy, dissecting the left and right brachiocephalic vein, the phrenic nerve, and the ascending aorta. Subcarinal, paraesophageal, and inferior pulmonary lymph nodes were removed en bloc, exposing the entire thoracic esophagus and the vagal nerve. The thoracic duct usually was ligated at the height of the main carina. For sampling reasons, contralateral hilar nodes were also excised. In left-sided cancer the subaortic compartment--contained between the left pulmonary artery, the aortic arch, the left recurrent laryngeal nerve, and the phrenic nerve-- was dissected by completely freeing the left vagal nerve and the recurrent laryngeal nerve. Thereafter the aortopulmonary ligament of Botalli was ligated and divided and the aortic arch was mobilized anteriorly to facilitate dissection of paratracheal nodes (nodes 2, 3, and 4). Routinely, contralateral hilar nodes were removed for sampling. Pathohistologic Analysis and Postoperative Adjuvant Therapy Tumors were classified according to the staging classification suggested by the International Union for the Control of Cancer. Each lymph node was sectioned at three different levels and classified as positive or negative for tumor. Frozen sections of lymph nodes were not used as a routine procedure during operation. Lymph nodes were mapped by the surgeon according to the mapping scheme of the American Thoracic Society [1]. Patients whose primary tumor was classified by the pathologist as a T3 or T4 tumor (stages IIIa and IIIb) received adjuvant postoperative percutaneous radiation therapy of the tumor bed with 50 Gy. All patients with involvement of nodes of the N2 region (stages IIIa and IIIb) received percutaneous radiation therapy of the entire mediastinum with 50 Gy. This adjuvant therapeutic regimen was followed in both groups. Follow-up and Definitions of Tumor Recurrence Patients were followed up at 6 month intervals. Routine examinations included a plain chest roentgenogram, a computed tomographic scan of thorax and abdomen, an abdominal ultrasound, a bone scan, and bronchoscopy with biopsy if warranted. The median follow-up period was 26.8 months (range, months). Local recurrence was defined as evidence of tumor within the same lung, at the bronchial stump, or manifest disease in ipsilateral mediastinal lymph nodes. Distant metastatic disease was defined as disease in the contralateral lung or outside the hemithorax, including supraclavicular lymph node metastases as well as metastases to distant organs, eg, brain and adrenal glands. Statistical Analysis The primary and main criteria for assessment were overall survival, local recurrence-free, and distant metastases-free survival. The secondary criteria were the postoperative mortality and morbidity rates and the effect of radical lymphadenectomy on tumor staging. Normal dis-

3 Ann Thorac Surg IZBICKI ET AL ;59: IMPACT OF LYMPHADENECTOMY ON TUMOR STAGING tribution of data was tested with a normal probability plot and Kolmogorov-Smirnov test. Frequency of events was tested with a )(2 test. Probability of cumulative cancer-related survival exclusive of 30-day mortality metastasis-free interval and local recurrence-free interval were calculated by the Kaplan-Meier-method and compared using log-rank-analysis. The level of significance was set at p less than Results Two hundred one patients were recruited for the study and were randomly assigned to be treated either by radical systematic mediastinal LA (n = 100) or by mediastinal LS (n = 101). After randomization, 19 patients (LS: n = 1; LA: n = 18) were excluded from analysis due to residual tumor or classification as small-cell lung cancer. Thus, 182 patients remained, with 82 in the LA and 100 in the LS group. After randomization there were no statistically significant differences between both groups with respect to age and sex of the patient or tumor site, tumor histology, T stage (Table 1), and type of primary tumor resection (Table 2). In the LS group a total of 393 lymph node levels (4.4 levels/patient) were removed with metastatic involvement in 55 levels (13.9%). In patients treated by LA, 590 lymph node levels (7.2 levels/patient) were resected and 78 levels (13.2%) were found to be positive by histopathological examinations. Table 1. Patient Characteristics Procedure LS LA p Characteristic (n - 100) (n 82) Value Age (y) Range NS Mean Sex (M/F) 80/20 56/26 NS Location (% in parens) Right 57 (57) 53 (64.6) NS Left 43 (43) 29 (35.4) NS Tumor histology (% in parens) Squamous cell carcinoma 46 (46) 27 (32.9) NS Adenocarcinoma 33 (33) 39 (47.6) NS Large cell carcinoma 8 (8) 4 (5.0) NS Adenosquamous carcinoma 4 (4) 3 (3.6) NS Other types 9 (9) 9 (10.9) NS T stage (% in parens) 1 17 (17) 18 (21.9) NS 2 65 (65) 52 (63.4) NS 3 7 (7) 9 (10.9) NS 4 11 (11) 3 (3.6) NS N stage (% in patens) NO 55 (55) 48 (58.5) NS N1 20 (20) 9 (10.9) NS N2 23 (23) 22 (26.8) NS N3 2 (2) 3 (3.7) NS LA= ymphadenectomy; LS = lymph node sampling; NS = not significant. Table 2. Types of Primary Tumor Resection Procedure LS LA (n 100) (n = 82) Resection Type n % n % Not extended Extended Upper lobectomy Bronchoplastic UL Extended UL Middle lobectomy Lower lobectomy Bronchoplastic LL Extended LL Upper bilobectomy Lower bilobectomy Pneumonectomy Extended Pneumonectomy LA - lymphadenectomy; LL - lower lobectomy; LS lymph node sampling; UL = upper lobectomy. Whereas in LA all ipsilateral lymph nodes are removed, in LS only the hilar, interlobar, peribronchial, tracheobronchial, and subcarninal lymph node levels are routinely resected. Superior mediastinal, paratracheal and pretracheal, paraesophageal, and pulmonary ligament lymph node levels are resected only if the surgeon is suspicious of cancer. Therefore, LS is justified only if intraoperative surgical staging of lymph nodes is correct or the surgeon at least does not underestimate the nodal status. Hence, we analyzed the accuracy of surgical lymph node staging on a level-by-level basis (Table 3). In the LS group underestimation of lymph nodes occurred in only 3.6% of the nodes, 52.9% were staged correctly, and 27.7% were overestimated. Similar results were obtained in the LA group. Taking into account that lymph nodes of regions 4 through 7 and 10 through 13 were resected in all cases, the interest has to focus on lymph nodes of region 1 through 3 and 8 through 9. Here underestimation was found in only one out of 124 lymph node levels (0.8%) in the LS group and in 6 out of 257 node levels of the LA group (2.3%). Regardless of the type of lymphadenectomy performed, the percentage of patients pathologically staged as having N1 or N2 disease was very similar in both groups (see Table 1). In contrast, the number of patients with lymph node involvement at multiple levels was significantly increased by LA (Table 4). In the LS group only 17.4% of the patients with N2 disease were found to have more than one lymph node level involved, whereas LA resulted in the detection of excessive N2 disease in 59.1% of patients (Table 4; p = 0.007). Overall, the type of lymphadenectomy did not influence the risk of local tumor recurrence, distant metastasis, and survival. During the follow-up period 21 patients of the LS group and 17 patients of the LA group had local

4 212 IZBICKI ET AL Ann Thorac Surg IMPACT OF LYMPHADENECTOMY ON TUMOR STAGING 1995;59: Table 3. Surgical Staging of Lymph Nodes a Lymph Node Sampling Underestimated b Correct Overestimated Lymph Node Levels n % n % n % LN 1 0/ / / LN 2, 3 0/ / / LN 4 1/ / / LN 5, 6 2/ / / LN 7 1/ / / LN / / LN 9 0/ / / LN / / / LN contralateral 0/ / Total no. of lymph / / node levels Radical Systematic Lymphadenectorny Underestimated Correct Overestimated n % n % n 4 2/ / / / / / / / / / / / / / / / / / / / According to the lymph node mapping of the American Thoracic Society. refers to the histopathologic results of lymph node levels. LN = lymph node level. b Underestimation, correct staging and overestimation of surgical staging recurrence (NS). Thirty-three patients of the LS and 24 of the LA group had distant metastasis (NS), and 26 LS patients and 24 of the LA group died from recurrent lung cancer (NS). Because radical lymphadenectomy results in the detection of excessive N2 disease in a higher percentage of patients, next we asked whether this might be of prognostic significance in this subgroup of patients. There were no significant differences between patients with lymph node involvement at one level or more than one level with respect to T stage (p = 0.332) and tumor histology (p = 0.342; X 2 test). Although the analysis was limited by the small number of patients in each subgroup, there were more distant metastases (p = 0.021) and cancer-related deaths (p = 0.162) in patients with lymph node involvement of more than one level (see Table 5 and Fig 1). Finally, we analyzed whether the type of lymphadenectomy might influence the detection of patients with skip metastases. Skip metastases were defined as lymph node metastases in the N2 region without positive lymph nodes in the N1 region. As shown in Table 6, the number of patients with skip metastases in the lymph node sampling group was even higher (30.4%) than in the group with radical lymphadenectomy (18.2%). Comment The extent of lymphadenectomy in treatment of NSCLC is still controversial. In addition to a prognostic benefit, advocates of radical LA claim a better staging of the tumor [1, 3, 9-12]. Opponents of the radical approach postulate a higher morbidity and mortality due to the extent of the operation and even a negative effect on long-term prognosis due to impaired local immune response. However, for staging reasons they recommend LS [4, 5]. In LA, all ipsilateral lymph nodes are resected routinely, whereas in LS, with the exception of hilar, interlobal, tracheobronchial, and subcarinal lymph nodes, only lymph nodes suspicious of cancer are removed. Therefore, LS seems to be an acceptable technique only if the surgeon is able to perform an adequate intraoperative lymph node staging. In our study, intraoperative surgical staging tended to overestimate lymph nodes; underestimation of lymph node levels was a rare event (Table 3). This is consistent with the results of our previous study, during which we observed that underestimation of the N stage by surgical intraoperative staging occurred in only 9 out of 108 patients (8.3%) [13]. In addition, LA does not automatically increase the percent- Table 4. Number of Lymph Node Levels With Metastatic Involvement LS Disease Stage 1 level >1 level 1 level >1 level N1 (% in parens) N2 (% in parens) LA versus LS, p = (~ test). LA = radical lymphadenectomy; LA 16/20 (80.0) 4/20 (20.0) 819 (88.8) 119 (11.1) 19/23 (82.6) 4 23 (17.4) 9 22 (40.9) 13/22 (59.1) a LS = lymph node sampling. Table 5. Prognosis of Patients With N2 Disease and Lymph Node Involvement at Multiple Levels a No. of Levels Cancer- Local Tumor Distant Related Recurrence Metastases Death 1 level 4 (50.0) 2 (25.0) 3 (37.5) (n = 8; % in parens) >1 level 3 (33.3; 6 (75.0; 6 (75.0; (n = 9; % in parens) p 0.717) b p ) p = 0.162) a Analysis is based on a reduced number of patients. In total, 5 patients were lost to follow-up, b Values of p are based on a log-rank test.

5 Ann Thorac Surg IZBICKI ET AL ;59: IMPACT OF LYMPHADENECTOMY ON TUMOR STAGING 100% A 80%- 60%- 40%" 20%- 0% o 100% 80% 60% 40% 2 % t o%l o Postoperative Months I '2 1 '8 2'4 3'0 Postoperative Months p=o.021 p = O. 162 Fig 1. Distant metastases-free (A) and overall survival (B) in patients treated by systematic radical lymphadenectomy with metastatic involvement of I (n - 8) or more than 1 (n = 9) lymph node level of the N 2 region. Kaplan-Meier curves were calculated and a log-rank test was carried out. (Solid line = one N2 level involved; broken line = more than one N2 level involved.) age of patients in whom N2 disease was detected (Table 1). Hence, LS with resection of suspicious lymph nodes seems to be a sufficient operation to determine the N stage of a patient and to remove involved lymph nodes, at least those with gross involvement. The most striking difference between LA and LS was that in the LA group the percentage of patients with N2 Table 6. Skip Metastases in N2 Patients No. of Patients No. of Patients Detection With N2 With Skip Method Disease Metastases a,b LS 23 7 (30.4) LA 22 4 (18.2) a Skip metastases were defined as metastases to lymph nodes of the N2 region without N1 involvement by histopathologic examinations. b Numbers in parentheses are percentages. LA radical lymphadenectomy; LS lymph node sampling. 36 3'6 involvement at multiple levels was significantly greater than that in the LS group (Table 4). Therefore, it is justified to assume that there must have been a substantial number of lymph nodes in patients treated by LS that were macroscopically not suspicious of cancer and therefore not resected but would have contained tumor cells by histopathologic examinations. Thus, the most interesting question is: do patients benefit from a resection of those lymph node levels with respect to an improved survival or a complete nodal staging? As recently reported by our group, the prognosis of patients with LA was not significantly improved compared with those treated by LS even by analyzing subgroups of patients with respect to T and N stage [7]. When comparing survival rates of patients with NSCLC from centers advocating LA as a routine measure and centers supporting LS, the most striking difference in survival rates is observed in patients with N2 disease. Martini and Flehinger [9] have described 5-year survival rates of up to 30%; others describe rates of 9% [14]. The critics of Martini and Flehinger's series have pointed out that patients in their series represented a highly selected group, with no clinical evidence of N2 disease [15]. In such a highly selected group, 5-year survival rates of 42% can be reached even without a radical lymphadenectomy [16]. Patients with clinically evident N2 disease in Martini and Flehinger's series exhibited a 5-year survival rate of 9% after LA [9], which is comparable with other series after LS [14]. Therefore, in patients with obvious N2 involvement, the disease is presumably too disseminated to be cured by operation alone. This is supported by our recent observations that in 21.6% of the patients with NSCLC, bone marrow micrometastases can be detected at the time of operation [17] and that even in patients histopathologically staged as NO, lymph node micrometastases have already occurred in about 15% of these patients [18]. In our study, all patients with N2 disease received postoperative percutaneous radiotherapy to the mediastinum, which was considered to be the treatment of choice at our institution during that period. It was therefore not possible to exclude the study patients from our standard regimen, although we are aware of the fact that the comparison between LA and LS might be influenced by radiotherapy. Although LA is not obligatory to determine the N stage of the patient (Table 1), it significantly increases the percentage of patients in whom N2 involvement at multiple levels can be detected (Table 4). As demonstrated in Figure 1 and shown by other studies, involvement of several N2 levels is associated with a worse prognosis with respect to the appearance of distant metastases and overall survival [9, 16]. Therefore, these patients have to be identified and admitted postoperatively to an adjuvant therapeutic regimen. Because these patients at risk can be recognized only by the resection of a sufficient number of lymph nodes, routine radical mediastinal lymphadenectomy seems to be essential for staging reasons.

6 214 IZBICKI ET AL Ann Thorac Surg IMPACT OF LYMPHADENECTOMY ON TUMOR STAGING 1995;59: References 1. Martini N, Flehinger BJ, Zaman MB, et al. Results of resection in non-oat cell carcinoma of the lung with mediastinal lymph node metastases. Ann Surg 1983;198: Martini N, Flehinger BJ, Zaman MB, et al. Results of surgical treatment in N2 lung cancer. World J Surg 1981;5: Naruke T, Goya T, Tsuchiya R, et al. The importance of surgery, to non-small cell carcinoma of lung with mediastinal lymph node metastasis. Ann Thorac Surg 1988;46: Thomas PA, Piantadosi S, Mountain CF, et al. Should subcarinal lymph nodes be routinely examined in patients with non-small cell lung cancer? J Thorac Cardiovasc Surg 1988; 95: Thomas P, Rubinstein L. Cancer recurrence after resection: T1 N0 non-small cell lung cancer. Ann Thorac Surg 1990;49: Sorensen JB, Badsberg JH. Prognostic factors in resected stages I and II adenocarcinoma of the lung. J Thorac Cardiovasc Surg 1990;99: Izbicki JR, Thetter O, Habekost M, et al. Radical systematic lymphadenectomy in non-small cell lung cancer. Br J Surg 1994;81: Naruke T, Suemasu K, Ishikawa S. Surgical treatment for lung cancer with metastasis to mediastinal lymph nodes. J Thorac Cardiovasc Surg 1976;71: Martini N, Flehinger BJ. The role of surgery in N2 lung cancer. Surg Clin North Am 1987;67: , Naruke T, Goya T, Tsuchiya R, et al. Prognosis and survival in resected lung carcinoma based on the new international staging system. J Thorac Cardiovasc Surg 1988;96: Hata E, Hayakawa K, Miyamoto H, et al. Rationale for extended lymphadenectomy for lung cancer. Theor Surg 1990;5: Martini N, Flehinger BJ, Zaman MB, et al. Prospective study of 445 lung carcinomas with mediastinal lymph node metastases. J Thorac Cardiovasc Surg 1980;80: Izbicki JR, Thetter O, Karg O, et al. Accuracy of computed tomographic scan and surgical assessment for staging of bronchial carcinoma. J Thorac Cardiovasc Surg 1992;104: Pearson FG, DeLarue NC, Ilves R, et al. Significance of positive superior mediastinal nodes identified at mediastinoscopy in patients with resectable cancer of the lung. J Thorac Cardiovasc Surg 1982;83: Shields TW. The significance of ipsilateral mediastinal lymph node metastasis (N2 disease) in non-small cell carcinoma of the lung. J Thorac Cardiovasc Surg 1990;99: Patterson GA, Piazza D, Pearson FG, et al. Significance of metastatic disease in subaortic lymph nodes. Ann Thorac Surg 1987;43: Pantel K, Izbicki JR, Angstwurm M, et al. Immunocytological detection of bone marrow micrometastasis in operable nonsmall cell lung cancer. Cancer Res 1993;53: Passlick B, Izbicki JR, Kubuschok B, et al. Immunohistochemical assessment of individual tumor cells in lymph nodes of patients with non-small-cell lung cancer. J Clin Oncol 1994;12: INVITED COMMENTARY This is a unique report describing a well-designed randomized trial for comparing two methods of evaluating mediastinal lymph nodes at the time of presumably curative resection for primary lung cancer: lymph node sampling consisted of obtaining samples from stations 4, 5 and 7 in all patients, and examining grossly abnormal nodes identified in the other stations from 2 to 9. Radical lymphadenectomy was indeed a radical resection, including the periesophageal nodes and ultimately ligation of the thoracic duct. The current American Thoracic Society map was used to define the location of nodal stations. There were no survival differences observed in the two groups. As might be anticipated, a higher incidence of multiple station involvement was identified in the radical lymphadenectomy group. There are some features of the study that may have impaired a clear interpretation of results. (1) All patients with N2 disease received radical postoperative irradiation, which may have modified the interpretation of recurrence rates and survival data in the small number of N2 cases reported for each category. (2) Intraoperative frozen section was not employed. Had it been used, the incidence of N2 disease by lymph node sampling may have been modified because "false negatives" would have been identified. (3) Had mediastinoscopy been used in all cases, the accuracy of the information would have been better still. It would have been of interest if Izbicki and associates had provided more detailed information on the nature and incidence of "skip lesions" in the mediastinal nodal metastases. The information is undoubtedly available in their records and could have provided a report on nodal stations involved in relation to the site of the primary tumor itself--at least by lobe and preferably by segment. F. Griffith Pearson, MD Division of Thoracic Surgery Department of Surgery University of Toronto The Toronto Hospital Eaton Wing Elizabeth St Toronto, ON M5G 2C4 Canada