Survival After Resection of Stage I1 Non-Small Cell Lung Cancer Nael Martini, MD, Michael E. Burt, MD, PhD, Manjit S. Bains, MD, Patricia M. McCormack, MD, Valerie W. Rusch, MD, and Robert J. Ginsberg, MD Division of Thoracic Surgery, Memorial Sloan-Kettering Cancer Center and Cornell University Medical College, New York, New York From 1973 to 1989, 214 patients with stage I1 non-small cell lung cancer were treated by resection and complete mediastirial lymph node dissection. There were 116 adenocarcinomas and 98 squamous cancers. There were 35 T1 N1 and 179 T2 N1 tumors. Whereas T1 tumors were mainly adenocarcinomas (83%), this difference was not apparent in T2 lesions. Regardless of histology, half of the patients had a single involved N1 lymph node. Lobectomy was performed in 68% of the patients, pneumonectoniy in 31%, and wedge resection or segmentectomy in 1%. Lobectomy was sufficient to encompass all disease in 34 of 35 T1 N1 tumors. Only 48 patients (22%) received postoperative external irradiation and 11 patients (5%) received chemotherapy. The overall 5-year disease-free survival was 39%. The best survival rates were in patients who had a single node involved and tumors 3 cm or less in diameter (48%). The pattern of recurrence differed by histology. Local or regional recurrence was more frequent in patients with squamous carcinoma whereas distant metastases were more commonly seen in adenocarcinomas (87%) with brain as the most frequent site (adenocarcinoma, 52%; squamous, 34%). It is concluded that in stage I1 carcinomas, resection remains the treatment of choice, that mediastinal lymph node dissection provides the most accurate staging, and that the best adjuvant treatment to improve survival is yet to be determined. (Ann Tkoruc Surg 1992;54:460-6) atients with stage I carcinomas of the lung are pre- P sumed to have early lung disease with a favorable prognosis after resection. Patients with stage I11 and IV carcinomas are viewed to have advanced tumors and a poor prognosis although some of them benefit from operation or combined modality therapy that includes operation. There remains a small, intermediate group of patients who present with early invasion into hilar or pulmonary nodes (Nl) as the sole site of tumor spread, ie, stage I1 patients, in whom resection is currently the treatment of choice but recurrence is frequent and the overall survival not encouraging. Efforts have been made to supplement the surgical treatment in this group of patients by radiation therapy, chemotherapy, or both. To date there has been no concrete evidence that any adjuvant therapy is beneficial. In 1984, we reported our experience with 75 patients with completely resected stage I1 non-small cell lung cancer [l]. Although we identified several prognostic factors in this group of patients, the small number of patients prevented us from identifying what adjuvant treatment might prolong survival. We have since nearly tripled our case material and present the results of our analysis. Presented at the Twenty-eighth Annual Meeting of The Society of Thoracic Surgeons, Orlando, FL, Feb 55, 1992. Address reprint requests to Dr Martini, 1275 York Ave, New York, NY 10021. 0 1992 by The Society of Thoracic Surgeons Material and Methods From 1973 to 1989, 214 patients with stage I1 non-small cell lung cancer were treated at Memorial Sloan-Kettering Cancer Center by resection and complete mediastinal lymph node dissection. This represented 9% of all resections for lung cancer during that period. The disease was staged postsurgically by the international TNM criteria for cancer staging [2]. The surgical and pathologic reports of all patients were carefully reviewed to ensure that the resection was complete, that no residual tumor, gross or microscopic, was left behind, and that no involved mediastinal lymph nodes were present. The records were analyzed to assess the prognostic significance of N1 involvement. Each tumor was evaluated for size, histology, visceral pleural involvement, and location. N1 nodes were evaluated for location and number of involved nodes. The surgical treatment rendered and all adjuvant treatment administered were also recorded. Follow-up status and recurrent site, if any, were also recorded. There were 146 men and 68 women, a ratio of 2.2: 1. The age range was 34 to 85 years with a median age of 62 years. Forty-four patients (21%) were 70 years of age or older. The tumor was located in the right lung in 107 patients and in the left lung in 107. Ten percent of the tumors were in the main bronchus beyond 2 cm of the carina, 51% in the upper lobes, 36% in the lower lobes, and 4% in the right middle lobe. There were 116 adenocarcinomas and 98 squamous cancers (Table 1). Included under adenocarcinomas were 18 bronchoalveolar carcino- 0003-4975/92/$5.00
Ann Thorac Surg 1992;54:460-6 MARTINI ET AL 461 Table 1. Resected Sfage I1 Lung Cancer by Sex and Histology His tology Male Female Total Squamous carcinoma 76 22 98 Adenocarcinoma 56 33 89 Bronchiolar carcinoma 7 11 18 Large cell carcinoma 7 2 9 Total 146 68 214 mas and 9 large cell carcinomas. Based on bronchoscopic and radiographic presentations, the primary tumor was considered central in 56 patients and peripheral in 158. There were 35 T1 N1 MO tumors and 179 T2 N1 MO. T1 tumors were mainly adenocarcinomas (83%). In T2 tumors, squamous carcinomas were as frequent as adenocarcinomas (squamous carcinomas 92, adenocarcinomas 87). The surgical treatment in the 214 patients studied consisted of lobectomy, bilobectomy, sleeve lobectomy, pneumonectomy, and wedge resection or segmentectomy (Table 2). The extent of pulmonary resection was influenced by the size and location of the tumor. Lobectomy was performed in 68% of the patients, pneumonectomy in 31%, and wedge resection or segmentectomy in 1%. The 3 patients treated by a lesser resection than a lobectomy had a peripheral tumor and severely limited pulmonary reserve. Nevertheless, a complete mediastinal node dissection was carried out in all 3 patients. Pneumonectomy was performed in 22% of patients with peripheral tumors and in 55% of patients with central tumors. Lobectomy was sufficient to encompass all disease in 34 of 35 T1 tumors. The specific tumor size in centimeters was recorded for all patients. There were 68 patients with a tumor size of 3 cm or less and 146 patients with tumors greater than 3 cm, 53 of which were greater than 5 cm. Of the 68 patients with tumors 3 cm or less in diameter, only 35 had T1 lesions; 14 had T2 lesions by reason of extension into the main bronchus and 19 had visceral pleural involvement. Four levels of N1 nodes were identified: hilar (level lo), interlobar (level ll), lobar or peribronchial (level 12), and segmental (level 13). Of the 214 resected tumors, one half (107) had metastasis in one N1 node, 86% had metastasis in only one level of N1 nodes, 13% in two levels, and 1% in three (Table 3). Lobar or peribronchial lymph nodes Table 3. Number and Level of Involved N1 Nodes Number of Levels Involved One N1 Total N1 Level Node 1 2-3 No. (%) 10 29 44 18 62 (28) 11 12 21 15 36 (17) 12 61 112 26 138 (64) 13 5 6 5 11 (5) Total 107 (50%) 183 (86%) 31 (14%) 214 (100%) (level 12) were involved by tumor in 138 patients (64%) and hilar nodes (level 10) in 61 (28%). Bilobectomy or pneumonectomy was performed in 55% of patients with tumor involvement in levels 10 or 11. Lobectomy was the most common procedure performed for tumors involving only levels 12 and/or 13; 75% of these patients had a lobectomy, and 23% had a pneumonectomy. There were seven postoperative deaths, an incidence of 3.3%, all in patients with T2 lesions. The causes of death included myocardial infarction in 3, respiratory failure in 2, pulmonary embolism in 1, and pneumonia in 1. Postoperatively, 48 patients (22%) received external irradiation and 11 patients (5%) received chemotherapy. Of 35 patients with T1 N1 lesions, 8 received radiation therapy and 1, chemotherapy. Of 179 patients with T2 N1 lesions, 36 received radiation therapy, 6 received chemotherapy, and 4 received both irradiation and chemotherapy. Results Survival Survival was measured by the Kaplan-Meier method considering postoperative deaths and deaths from lung cancer [3]; comparisons were made by the log rank method. The overall 5-year survival of patients with N1 disease was 39% (median follow-up: 37 months) (Fig 1). There was no difference in survival between patients with T1 and T2 disease (40% for T1, 38% for T2) (Fig 2). However, there was a significant difference in survival by tumor size. Patients with tumors 3 cm or less in diameter I,,I Table 2. Surgical Treatment of Stage 11 Lung Cancer Procedure T1 T2 Total Lobectomy 34 74 108 Bilobectomy... 30 30 Sleeve lobectomy... 7 7 Pneumonectomy 1 65 66 Wedgekegment... 3 3 Total 35 179 214 a 0.1 1 4 0.1 01 lo 0 12 24 36 40 60 Fig 1. Overall survival after resection in stage ZZ lung cancer (n = 214).
~~ 102 462 MARTINI ET AL Ann Thorac Surg 1992;54:46M - AdenoCa (n=116) - 0.9 + Sq Cell Ca (n=98) - 0.8-0.7 z e 0 12 24 36 48 60 Fig 2. Siirvival from resection of stage 11 lung cancer in Tl versus T2 tumors. (n = 68) had a 5-year survival of 47% as compared with patients with tumors 5 cm or greater (n = 72), in whom the 5-yex survival was 29% (p = 0.021) (Fig 3). This difference was not observed when the cutoff point for tumor size was 3 cm. Patients with tumors greater than 3 cm (n := 146) had a 5-year survival of 35% (p = 0.212). There was no significant difference in survival between patients with squamous carcinoma versus adenocarcinoma (44% at 5 years for squamous carcinoma and 34% for adenocarcinoma; p = 0.238) (Fig 4), between patients with T1 adenocarcinomas and T1 squamous carcinomas ( p = 0.288), or between patients with T2 adenocarcinomas and T2 squamous carcinomas (p = 0.329). Although bronchoalveolar carcinomas and large cell cancers were grouped under adenocarcinomas, the difference in survival among patients in these three subgroups was not significant. Patients who received postoperative radiation therapy did significantly worse than those who did not (Fig 5). However, the majority of the patients had received no irradiation after resection and the decision to give irradiation was arbitrary and not standardized. Our interpretation of this was that radiation therapy was selected for patients with larger tumors, multiplicity of nodes, central tumors, or pleural involvement (Table 4) because most of > z 0 t U 2 n 0.5-0.4-0.2 0.3 I (p=0.238) 34% 0.3 0.1 1,,,, j;.1 0 0 12 24 36 48 60 Fig 4. Survival from resection of stage 1Z lung cancer by histology. (AdenoCa = adenocarcinoma; Sq Cell Ca = squamous cell carcinoma.) them died of their disease. However, when radiation therapy was added to the Cox model as a factor, we found that postoperative radiation therapy was a highly significant ( p = 0.0005) predictor of poor outcome, independent of these variables. There was a statistically significant difference in survival rates between patients with involvement of a single N1 node versus multiple N1 nodes (45% at 5 years for a single node versus 31% for multiple nodes; p = 0.016) (Fig 6). The best survival rates were in patients who had a single node involved and tumor 3 cm or less in diameter (48%) (Fig 7). There was no significant difference in the 5-year survival rates between patients whose tumors had visceral pleural invasion compared with those who did not (pleura negative, 41%; pleura positive, 33%; p = 0.294), nor was there a difference in survival by location of the tumor (peripheral, 39%; central, 37%). However there was some difference in survival between patients who had small tumors (3 cm or less) and no pleural involvement versus patients whose tumors were greater than 3 cm in diameter and who also had pleural involvement (46% versus 25%; p = 0.073). There was no difference in survival by the extent of resection performed (lobectomy, 40%; pneumonectomy, 35%; p = 0.480). + >3 to c5 (n=74) - 0.9 - RT(n=48) + No RT (n=166) - 0.8-0.7 (c=3 vs >=5: p=0.021) Fig 3. Survival from resection of stage 11 lung cancer by tumor size (in centimeters). Fig 5. Survival from resection of stage 11 lung cancer in patients who received postoperative radiation therapy (RT).
Ann Thorac Surg 1992:54:460-6 MARTINI ET AL 463 Table 4. Characteristics of Patients Who Received Postoperative Radiation Therapy. Single <=3 (n=40) - 0.9 Characteristics RT TI 8 (17%) Tumor size 5 3 cm 11 (23%) 57 (34%) 5 0.5 - One N1 node 17 (35%) 90 (54%) F 0.4 - + Mult >3 (n=79) Squamous carcinoma 17 (35%) 81 (49%) 0.3 - Central tumor 9 (19%) 46 (28%) 0 0.2 - p=0.025 27% - 0.2 Pleura involved 16 (33%) 45 (27%) 0.1-0.1 53 cm, one node 5 (10%) 35 (21%) 0 0 >3 cm, multiple nodes 25 (52%) 54 (33%) 0 12 24 36 48 60 5-year survival 13% 45% Total 48 166 Fig 7. Survival from resection of stage 11 lung cancer by tumor size (in centimeters) and number of involved N1 nodes. (Mult = multi- RT = radiation therapy. ple.) - 0.8-0.7-0.6 sions, 42 of 87 patients (48%) had recurrence, 13 of 42 (31%) locoregionally and 29 of 42 (69%) at distant sites, 34% of which were in brain. In adenocarcinoma, 18 of 29 patients (62%) with T1 lesions had recurrence, 2 of 18 (11%) locoregionally, 1 of whom had received postoperative irradiation, and 16 of 18 (89%) at distant sites, 62% of which were in brain. In T2 disease, 52 of 85 patients (61%) had recurrence, 7 of 52 (13%) locoregionally and 45 of 52 (87%) at distant sites, 49% of which were in brain. The Patterns of Recurrence overall incidence of recurrence was greater in patients There were 114 patients (53%) with N1 disease who had with adenocarcinoma (p = 0.06). The difference in the recurrence after treatment (57% with T1 lesions and 52% incidence of local recurrence between histologies was not with T2 lesions). We have defined local recurrence as significant ( p = 0.10), but the difference in the incidence of evidence of tumor within the same lung or at the bron- distant metastases was significant (p = 0.001), with chial stump and regional recurrence as clinically mani- greater frequency in patients with adenocarcinoma. fested disease in mediastinal nodes despite mediastinal The effect of postoperative irradiation on local or relymph node dissection at the original operation, lymph gional recurrence was evaluated (Table 6). There were 21 node metastasis in contralateral lymph nodes, and lymph local recurrences in 166 patients who received no radianode metastasis in supraclavicular regions. Distant recur- tion treatment (12%) and 3 local recurrences in 48 patients rence was defined as disease in the contralateral lung or in who received radiation treatment (6%). The data sugdistant lymph nodes or distant organs. gested that postoperative radiation therapy reduces local Using a multivariant analysis based on the Cox proportional hazard model [4], favorable prognostic factors of significance in this study have been tumor size of 3 cm or less (p = 0.0082) and solitary involvement of N1 nodes (p = 0.0129). A trend toward improved survival is noted if the histology is a squamous carcinoma (p = 0.0716), but there is little or no impact on survival of the location of the tumor or the presence of pleural invasion (p = 0.3). The patterns of recurrence differed by histology (Table 5). In patients with squamous carcinoma, 2 of 6 with T1 lesions had recurrence, both locally, 1 of whom had received postoperative irradiation; in those with T2 le- Table 5. Sites Of ~m~rence by Histology" Squamous Variable Overall carcinoma Adenocarcinoma 1 No Recurrence 93 (45) 49 (53) 44 (39) - Single (n=107) - 0.9 Recurrence 114 (55) 44 (47) 70 (61) + Multiple (n=107) - 0.8 Localhegional 24 (21) 15 (34) 9 (13) - 0.7 Distant 90 (79) 29 (64) 61 (87) Brain 42 10 32 Bone 19 10 9 Lung 16 4 12 Liver 13 4 9 Other 12 4 8 - - 01 lo 0 12 24 36 48 60 Postop deaths 7 5 2 a 0.1 0.1 Subtotal 207 93 114 Total 214 98 116 Fig 6. Survival from resection of stage 11 lung cancer by number of involved Nl nodes. a Numbers in parentheses are percentages, excluding postoperative deaths.
464 MARTINI ET AL!STAGE 11 LUNG CANCER Ann Thorac Surg 1992;54:460-6 Table 6. lncidence of LocallRegional Recurrence After Postoperative Radiation Therapy by Histology Overall Squamous Adenocarcinoma Group No. LR (%) No. LR (%) No. LR (%) Radiation 48 3 (6%) 17 1(6%) 31 2 (6%) No Radiation 166 21 (12%) 81 14 (17%) 85 7 (8%) Total 214 98 116 LR = IocaVregional recurrence. and regional recurrence in patients with squamous carcinoma but has no detectable effect in patients with adenocarcinoma. The n.umber of patients who received chemotherapy was far too small (5%) to lend any meaningful conclusion on the impact of this form of adjuvant treatment. We calculated the probabilities of recurrence in patients with resected N1 tumors to be 30% within 1 year of treatment, 45% within 3 years, and 52% within 5 years. Ten percent are believed to remain at risk for recurrence after 5 years. There were 16 new cancers that developed after treatment of the primary lung carcinoma in the 214 patients studied; 7 of these were new primary lung carcinomas. A colon carcinoma was found in 4 patients, a malignant lymphoma in 2, and an epiglottic carcinoma in 1. To date, 60 patients in this study are alive and well 1 to 16 years after treatment, with a median follow-up of 60 months. Three additional patients were lost to follow-up, 2 at 2 months and 1 at 9 months. Six are alive with disease 2 to 5 years after treatment (median follow-up, 46 months). There were 148 late deaths, 34 of unrelated causes and 114 of the original carcinoma. Deaths from unrelated causes were due to a new cancer in 7 patients, cardiorespiratory problems in 25 (chronic obstructive pulmonary disease, pulmonary embolism, myocardial infarction, heart failure, or stroke), cirrhosis in 1, and old age in 1. Comment A stage It lung carcinoma is defined as a T1 or T2 tumor of the lung with metastasis solely to intrapulmonary (including hilar) lymph nodes. Resection is still the most important treatment of this group of patients. The overall 5-year survival is 39% with no observed survival advantage by histology, location of the tumor, or by T1 versus T2. Although T1 N1 lesions had been classified as stage I1 disease by the Japanese, since 1966, physicians in the US adhered to the American Joint Committee staging, which included this subset under stage I carcinoma. Shields and associates [5] were the first to suggest to the American Joint Committee in 1980 that T1 N1 lesions should also be classified as stage I1 disease. In our earlier report [l] on 75 patients with resected stage I1 cancers, pleural involvement had a significant impact 011 survival (p = 0.01). However, this factor does not appear to be of significance in the current review. On the other hand, the number of involved N1 nodes is significant. Exactly one half of the patients had metastases in only one N1 node and 85% had lymph node involvement in only one N1 level. A significant difference in survival was noted in patients with metastasis to only one N1 lymph node compared with those with more than one lymph node involved (45% versus 31%; p = 0.016). Interestingly, a similar observation was noted by us in patients with N2 disease treated by resection [6, 71. N2 lymph node metastasis was confined to one lymph node in one third of the patients, and these had a significantly better survival compared with those with more than one N2 lymph node involvement (p = 0.005). According to the TNM staging system, a tumor with visceral pleural invasion or proximal extension into the main bronchus despite its small size is classified as a T2 lesion [2]. Based on this classification, we observe no survival difference between T1 and T2 tumors. However, if we were to assess survival based on tumor size alone, tumor size becomes a significant prognostic factor. The pattern of recurrence differed by histology. Local or regional recurrence was more frequent in patients with squamous carcinomas (squamous cancer, 34%; adenocarcinoma, 13%) whereas distant metastases were more commonly seen in adenocarcinomas (adenocarcinoma, 87%; squamous, 64%). The most frequent site of distant metastasis was the brain in both major histologic cell types. Distant metastases were found in brain in 46% of recurrent adenocarcinomas and 23% of recurrent squamous cancers. If the site at which a tumor is most likely to recur can be predicted, appropriate adjuvant treatment may be used more effectively. In the Northwestern Universit? study, most N1 patients had local recurrence [8, 91. Therefore postoperative radiation therapy was recommended. However, several centers, including ours, report that distant metastases are more frequent than local or regional recurrences in resected stage I1 carcinomas [l, 2, 10-121. The effect of postoperative irradiation on local or regional recurrence was evaluated. The data suggest that postoperative radiation therapy reduces local or regional recurrence in patients with squamous carcinoma but has no detectable effect in patients with adenocarcinoma. Because of the relatively small number of patients who have stage I1 lung cancer, it has been difficult to assess whether they do represent variants of early stage disease or should be viewed as advanced cancers and treated with combined modality therapy. This group of patients benefits from complete surgical resection with mediastinal lymph node dissection. However, more than one half of them have recurrence in the course of their follow-up. Few had been treated with irradiation or chemotherapy postoperatively, but no single institution has been able to recommend adjuvants to enhance survival in these patients. Two large cooperative studies have attempted to address this group of patients. The Lung Cancer Study Group randomized 189 resected stage I1 patients by his-
Ann Thorac Surg 1992;54460-6 MARTINI ET AL 465 tology to receive adjuvant treatments [lo, 13, 141. Those with resected squamous cancers were treated by postoperative external irradiation or received no further treatment [lo]. In resected stage I1 adenocarcinomas, the randomized arms of the trial were cyclophosphamide, doxorubicin, and cisplatin chemotherapy or immunotherapy [14]. These studies concluded that in squamous carcinoma, postoperative radiation therapy reduced local and regional recurrence but had no impact on survival. In adenocarcinoma, chemotherapy had a survival advantage over immunotherapy. The drawbacks of these studies were that stage I1 and I11 resected lung cancers were grouped together and no separate analysis of patients with stage I1 carcinomas was reported. The Lung Cancer Study Group reported a survival difference between patients with squamous carcinoma and adenocarcinoma, and between T1 and T2 lesions [15]. The second large cooperative trial was conducted by the Ludwig Lung Cancer Study Group from Europe [16]. It evaluated resected stage I1 carcinomas for survival and recurrence. There were 253 patients with stage I1 disease entered into the study (81 T1 N1 and 172 T2 Nl). Only median survivals were reported, indicating that patients with T1 lesions lived twice as long as patients with T2 disease (median T1, 4.8 years; T2, 2.3 years). Naruke and associates [17] from the National Cancer Center Hospital of Japan reported on 221 patients with completely resected stage I1 lung cancer and showed an overall 5-year survival of 43% and a 5-year survival of 52% for T1 lesions and 38% for T2 lesions (p = 0.05). There was no difference in survival by histology in their series. In an earlier study by the Veterans Administration Surgical Oncology Cooperative Group 152 patients with resected stage I1 lung cancer had shown no significant difference in survival at 3 years between T1 and T2 lesions (Tl, 37%; T2, 40%) [4]. A more recent report of the MD Anderson experience combined with that of the Lung Cancer Study Group reported on 317 patients with resected stage I1 lung cancer [2]. The 5-year survival was 54% for T1 lesions (n = 67) and 40% for T2 lesions (n = 250). These differences in observation among the various series may reflect differences in case material and in the method of lymph node assessment among the various institutions. We conclude that in patients with stage I1 non-small cell lung cancer: 1. 2. 3. 4. 5. The best treatment is surgical, and complete resection is essential for long-term survival. The rate of recurrence is high despite surgical treatment. The survival difference by histology is not significant, but tumor size and number of involved N1 nodes are the two most favorable prognostic factors. Location of the primary tumor, location of the N1 nodes, involvement of the visceral pleura, extent of resection, and T1 versus T2 are not major prognostic factors. The patterns of recurrence differ by histology (more local recurrences in patients with squamous carcinoma and more distant metastases in patients with adenocarcinoma). 6. Brain metastasis is the most common form of recurrence. 7. Postoperative external radiation therapy seems to reduce the incidence of local and regional recurrence, at least in squamous carcinoma, without affecting the ultimate survival. 8. Most recurrences are distant metastases, suggesting the need for effective systemic treatment once nodal involvement is present. The specific regimens that might benefit these patients require further study. References 1. Martini N, Flehinger BJ, Nagasaki F, Hart B. Prognostic significance of N1 disease in carcinoma of the lung. J Thorac Cardiovasc Surg 1983;86:646-53. 2. Mountain CF. A new international staging system for lung cancer. Chest 1986;89(Suppl):22533. 3. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53:457-81. 4. Cox DR. Regression models and life tables. J R Stat SOC B 1972;34: 187-220. 5. Shields TW, Humphrey EW, Matthews M, Eastridge CE, Keehn BS. Pathological stage grouping of patients with resected carcinoma of the lung. J Thorac Cardiovasc Surg 1980;80:400-5. 6. Martini N, Flehinger BJ, Zaman MB, Beattie EJ Jr. Results of resection in non-oat cell carcinoma of the lung with mediastinal lymph node metastases. Ann Surg 1983;198:38&97. 7. Martini N, Flehinger BJ. The role of surgery in N2 lung cancer. Surg Clin North Am 1987;67:103749. 8. Shields TW, Yee J, Conn JH, Robinette CD. Relationship of cell type and lymph node metastasis to survival after resection of bronchial carcinoma. Ann Thorac Surg 1975;20: 501-10. 9. Immerman SC, Vanecko RM, Fry WA, Head LR, Shields TW. Site of recurrence in patients with stages I and I1 carcinoma of the lung resected for cure. Ann Thorac Surg 1981;32:23-7. 10. The Lung Cancer Study Group (prepared by Weisenburger TH, Gail M). Effects of postoperative mediastinal radiation in completely resected stage I1 and stage 111 epidermoid cancer of the lung. N Engl J Med 1986;315:1377-81. 11. Newman SB, DeMeester TR, Golomb HM, Hoffman PC, Little AG, Raghavan V. Treatment of modified stage I1 (TlNlMO, T2NlMO) non-small cell bronchogenic carcinoma. A combined modality approach. J Thorac Cardiovasc Surg 1983;86:180-5. 12. Downey RS, Sewell CW, Mansour K. Large cell carcinoma of the lung: a highly aggressive tumor with dismal prognosis. Ann Thorac Surg 1989;4780&8. 13. Holmes EC, Hill LD, Gail M, for the Lung Cancer Study Group. A randomized comparison of the effects of adjuvant therapy on resected stages I1 and 111 non-small cell carcinoma of the lung. Ann Surg 1985;202:33540. 14. Holmes EC, Gail M, for the Lung Cancer Study Group. Surgical adjuvant therapy for stage I1 and stage Ill adenocarcinoma and large-cell undifferentiated carcinoma. J Clin Oncol 1986;4:71c5. 15. Holmes EC. Treatment of stane I1 cancer (TlN1 and T2N1). Surg Clin North Am 1987;67:c945-9. 16. The Ludwig Lung Cancer Study Group. Patterns of failure in patients with resected stage I and I1 non-small cell carcinoma of lung. Ann Surg 1987;205:67-71. 17. Naruke T, Goya T, Tsuchiya R, Suemasu K. Prognosis and survival in resected lung carcinoma based on the new international staging system. J Thorac Cardiovasc Surg 1988;96: 440-7.
466 MARTINI ET AL Ann Thorac Surg 1992;5446&6 DISCIJSSION DR THOMAS R. J. TODD (Ottawa, Canada): It is my honor to be invited to discuss this report, particularly as it has been prepared and written by one of the giants in our field who is known lor his integrity and his thoroughness, as we have seen here today, and indeed for his precise evaluation of his results. I had the privilege as well of reviewing the manuscript ahead of time, for which I thank Dr Martini. I think that Dr Martini meant to challenge us with this report rather than try to determine for us what the ideal therapy and outcome will be in patients with N1 bronchogenic carcinoma. And as we go through this, perhaps my reasons for that will become apparent. If we look at the study critically, the first thing is it is retrospective, and we have all had it hammered into us what the problems are with retrospective studies. Nonetheless, they can be very descriptive. However, in a retrospective study the populatison is not homogeneous. And certainly from a therapeutic point of view this population is not homogeneous in that about 27% of them had some form of adjuvant therapy postoperatively. And we do not know how that adjuvant therapy was distributed between those who had multiple nodes versus single nodes and those who had one level versus more than one level of nodes involved. In addition, the patients in a retrospective study differ in other ways. For example, did these patients have mediastinoscopy ahead of time? This might be particularly important for left-sided lesions when biopsy of the superior mediastinum at thoracotomy, particularly for the N3 contralateral nodes in left lower lobe tumors, is difficult. The second problem is that the N1 designation is somewhat diffuse, and we all know this. In other words, does a lymph node involved at the segmental orifice of the bronchus have the same prognostic significance as an involved single lymph node at the pleural reflection? I do not think we know that. In my own hospital we have attempted to redefine and subgroup the N1 nodes, and we are examining this in a prospective fashion to try and answer that question. Now, based on this critique I have two questions for Dr Martini. 'The first is, was there a difference in mortality based on whether the single involved node was at the 10, 11, 12, or 13 station? Second, was mediastinoscopy performed routinely, and if it was not, was the superior mediastinum meticulously examined at thoracotomy, particularly on the left side and for left lower lobe lesions in particular? Now, it is my contention that the next major advance in the therapy of lung cancer is going to be in this group of patients with N1 disease. In my hospital this accounts for 21% of our tumors, not 9%. And just as we have seen a change in not only therapy but prognosis in subgroups of carcinoma of the colon based on this type of analysis, I believe that we are going to see new therapies enhancing the already reasonable surgical results in N1 bronchogenic carcinoma. In that regard, therefore, we are indebted to~ Dr Martini, who, as he has so many times in the past, has set the stage for us very accurately and issued us a challenge. DR MARTINI: Admittedly this is a retrospective study. Nearly 8 years ago, as I presented a handful of patients seen at our institution, I emphasized that no single institution has the capability of including enough stage I1 patients to be able to do a prospective study let alone a randomized one to answer the many questions raised. Mediastinoscopy was not done routinely in these patients. However, at operation a meticulous lymph node dissection was carried out, and a lymph node dissection gives us the most accurate staging, most importantly in this group of patients, in ruling out the presence of N2 disease that may have been overlooked. In left-sided lesions, lymph node dissection is limited, particularly in the superior mediastinum. Perhaps in the future mediastinoscopy would be helpful, particularly if these patients are to be treated by combined modality therapy. The difference in mortality by location of the N1 nodes had no survival advantage. The main point of this report is that however early these lesions are presumed to be, the presence of any lymph node involvement is a bad prognostic sign. DR JOHN S. CHAMBERS (San Diego, CA): I, too, wanted to ask the question about staging mediastinoscopy, because the point has been raised that perhaps the surgical trauma of invasive staging might tend to affect the prognosis by spreading disease. I wonder if you investigated whether the patients who had mediastinoscopy had a worse prognosis. The other question was how many of your patients preoperatively had NO disease diagnosed and how many had N2 disease; in other words, were up-staged and down-staged? You must have explored quite a few patients with N2 disease to find these N1 patients, and what happened to those N2 patients who had, I assume, no preoperative adjuvant therapy? If their prognosis was poor, which it must have been, would you alter your approach to this kind of advanced patient by using neoadjuvant therapy of one form or another, radiotherapy or drugs? DR MARTINI: As you all well know, I have not been the staunchest advocate of mediastinoscopy, and therefore most of those patients did not have routine mediastinoscopy. I cannot answer your question on whether mediastinoscopy had an adverse impact as an invasive procedure. I would hope it would not. It is my opinion that the essential question is not whether you do a mediastinoscopy or not. The question is what do you do with the information after you obtain it from mediastinoscopy and what treatment do you choose for your specific patient. It is true that in this subgroup of patients it is difficult to clinically stage whether they have NO or N2 disease. In fact in our series as well as that of the Japanese, the error in the preoperative assessment as compared with the intraoperative was very high; only 30% of patients we had assumed to have had clinically N1 disease remained N1 patients. The remainder were either upgraded or downgraded. What do you do with a patient when you are confronted at operation with N2 disease not previously assessed? The answer is very simple: you resect what you can with a complete node dissection. Whether neoadjuvant therapy can or cannot be used in this group of patients remains to be studied in the future.