Surgical resection is the first treatment of choice for

Predictors of Lymph Node and Intrapulmonary Metastasis in Clinical Stage IA Non Small Cell Lung Carcinoma Kenji Suzuki, MD, Kanji Nagai, MD, Junji Yoshida, MD, Mitsuyo Nishimura, MD, and Yutaka Nishiwaki, MD Division of Thoracic Oncology, National Cancer Center Hospital East, Kashiwa, Chiba, Japan Background. The feasibility of limited surgical resection for clinical stage IA non small cell lung cancer still remains controversial. Methods. From July 1987 through April 1998, 389 patients with clinical stage IA disease underwent major lung resection and complete mediastinal lymph node dissection. Univariate and multivariable analyses were performed to determine predictors of local or regional tumor spread: pathologic lymph node involvement, intrapulmonary metastases, and lymphatic invasion. Results. Of the 389 patients, 88 (23%) had lymph node involvement or intrapulmonary metastases pathologically. According to multivariable analyses, grade of differentiation and pleural involvement were significant predictors of local or regional tumor spread (p < 0.01). Based on these results, more than 40% of clinical stage IA non small cell lung cancer patients showed pathologic lymph node involvement or intrapulmonary metastases, or both, if the patients had both of the predictors of pathologic local or regional involvement: moderate or poor differentiation of the primary tumor and pleural involvement by tumor cells. Conclusions. Limited surgical resection is not feasible for clinical stage IA non small cell lung cancer, especially when the tumor shows moderate or poor differentiation, or pleural involvement. (Ann Thorac Surg 2001;72:352 6) 2001 by The Society of Thoracic Surgeons Accepted for publication April 13, 2001. Address reprint requests to Dr Suzuki, Division of Thoracic Surgery, National Cancer Center Hospital, 1-1, Tsukiji 5 cho-me, Chuo-ku, Tokyo, Japan; e-mail: kjsuzuki@ncc.go.jp. Surgical resection is the first treatment of choice for clinical stage I non small cell lung cancer patients. Anatomical lobectomy with mediastinal node sampling has been generally accepted as the standard surgical intervention [1]. The rationale for the procedure was clearly demonstrated by the Lung Cancer Study Group (LCSG) based on a randomized trial [2], and by other retrospective studies [3, 4]. In contrast, limited surgical resection, defined as removal of a pulmonary tumor and surrounding lung by a technique designed to preserve lung parenchyma encompassing less than a lobectomy [5], has been performed in compromised patients with limited pulmonary function [6]. Several institutes have also reported that limited surgical resection could be an alternative to anatomic lobectomy, especially for clinical stage IA non small cell lung cancer [7 9]. Limited surgical resection may be a reasonable approach for a carcinoma without invasive spread. However, several researchers have concluded that, even in small-sized lung carcinoma, there could be local or regional spread of cancer cells: pathologic nodal involvement, intrapulmonary metastases, or lymphatic involvement [10]. Determining the indications for limited operation based on only the tumor size appears questionable. To investigate reasonable and reliable indications of limited operation, we attempted to identify simple clinicopathologic predictors of local or regional tumor spread with respect to occult lymph node involvement, intrapulmonary metastases, and lymphatic permeation. and Methods From July 1987 through April 1998, 1,142 non small cell lung cancer patients underwent surgical resection at our institute. Among them were 389 patients with clinical stage IA disease, who underwent major lung resection and complete mediastinal lymph node dissection. Their characteristics were reviewed retrospectively. All patients underwent thoracic computed tomographic (CT) scan preoperatively using an X-vision/SP system (Toshiba, Tokyo, Japan) and 10-mm thick contiguous sections to evaluate nodal status. The clinical diagnosis of nodal involvement was determined by diagnostic radiologists and based on the CT findings: mediastinal or hilar lymph nodes 1.0 cm or larger in their shortest axis were diagnosed as metastatic [11]. Histologic typing was determined according to the World Health Organization classification [12]. The stage of the disease was based on the TNM classification of the International Union Against Cancer [13]. The mediastinal lymph node dissection was performed according to the methods described by Na- 2001 by The Society of Thoracic Surgeons 0003-4975/01/$20.00 Published by Elsevier Science Inc PII S0003-4975(01)02748-5

Ann Thorac Surg SUZUKI ET AL 2001;72:352 6 STAGE MIGRATIONS IN STAGE I LUNG CANCER 353 ruke and colleagues [14] All resected specimens were formalin fixed and sliced at 5-mm to 10-mm intervals. Primary lung neoplasms, nodules, and lymph nodes were evaluated microscopically by conventional hematoxylin and eosin stain. The medical record of each patient was reviewed for age, sex, pack-years smoking, symptoms (symptomatic versus asymptomatic), laterality of tumor (left versus right), lobar distribution (upper/middle lobe versus lower lobe), maximum tumor dimension (larger than 2.0 cm versus 2.0 cm), histology (adenocarcinoma versus others), grade of tumor differentiation (good versus moderate or poor), serum carcinoembryonic antigen (CEA) level (5.0 ng/ml versus less than 5.0 ng/ml), pleural involvement (P0 versus P1 to P3), lymphatic invasion (positive versus negative), pathologic N status (positive versus negative), and intrapulmonary metastasis (present versus absent). was classified as P0, P1, P2, and P3: P0 tumor did not extend beyond the elastic pleural layer; P1 tumor invaded the elastic layer of the visceral pleura but did not expose itself on the pleural surface; P2 meant tumor exposed on the pleural surface; and P3 tumor invaded parietal pleura or chest wall. Lymphatic invasion indicated tumor cells identifiable in the lymphatic vessel lumen [15]. Intrapulmonary metastasis was defined as an independent mass, isolated from the primary malignancy, that had histopathologic features identical to the primary tumor. Univariate analyses were performed by the logistic regression procedure [16] on StatView 5.0 (SAS Institute, Cary, NC) with a Power Macintosh G3 300 MT to determine predictors of lymph node involvement or intrapulmonary metastases, which could be potential risks for local recurrence after limited surgical resection. The significant predictors of these factors in univariate analyses were also evaluated using multiple regression analyses. Clinical predictors of lymphatic invasion by tumor cells were also investigated using univariable and multivariable analyses. We also specifically investigated smallsized lung cancers, measuring 2.0 cm or less, because some authors have discussed limited operation among this population [7, 9]. Because CEA values were not available for some patients, we initially performed multivariable analysis in all patients with clinical stage I lung cancer using measurements other than serum CEA levels. We also performed multivariable analysis among patients in whom serum CEA levels were available. Carcinoembryonic antigen levels were examined in 337 (87%) of 389 clinical stage IA lung carcinoma patients. As there were no differences in clinical background between patients with and without available CEA values, the results of multivariable analysis among patients who had available CEA values and overall cases were identical. The 2 test was used to compare the probability of pathologic nodal involvement, intrapulmonary metastases, and lymphatic invasion by tumor cells between patient subgroups. Differences were considered to be statistically significant when the p value was less than 0.05. Table 1. Relationship Between Clinical Features and Pathologic Local or Regional Nodal Involvement (pn) or Intrapulmonary Metastases (pm) Among Clinical Stage IA Lung Cancer a Results pn or pm Total 389 88 (23)... Age c Range (median) 28 85 (64)... 0.33 Sex Male 202 40 (20) 0.17 Female 187 48 (26) Pack-years smoking c Range (median) 0 216 (10)... 0.19 Symptoms Asymptomatic 323 70 (22) 0.32 Symptomatic 66 18 (27) CEA (ng/ml) 5.0 232 50 (22) 0.30 5.0 105 28 (27) Side of tumor Left 152 27 (18) 0.07 Right 237 61 (26) Lobar distribution Upper or middle lobe 276 66 (24) 0.34 Lower lobe 113 22 (19) 2.0 187 26 (14) 0.001 2.0 202 62 (31) Histology Squamous 63 10 (16) 0.17 Adenocarcinoma 316 77 (24) 0.09 Others 10 1 (10) Good 204 28 (14) 0.001 Moderate or poor 185 60 (32) P0 300 58 (19) 0.004 P1 3 89 30 (34) a Numbers in parentheses are percentages. analyses in a logistic regression model. variable into the logistic regression model. CEA carcinoembryonic antigen. Among 389 resected clinical stage IA non small cell lung cancer patients, 202 were men and 187 were women. Their ages ranged from 28 to 85 years (median 64 years). Eighty-one (21%) and 16 (4%) patients had pathologic lymph node involvement and intrapulmonary metastases, respectively. The incidence of local or regional lymph node involvement or intrapulmonary metastases in each patient subgroup is shown in Table 1. Eighty-eight (23%) patients had local or regional nodal involvement or intrapulmonary metastases pathologically. In univariate analyses, maximum tumor dimension larger than 2.0 cm, moderate or poor differentiation, and positive pleural involvement by tumor were significant predictors of b Probability in univariate c Entered as a continuous

354 SUZUKI ET AL Ann Thorac Surg STAGE MIGRATIONS IN STAGE I LUNG CANCER 2001;72:352 6 Table 2. Multivariable Analyses of Factors That Predict Carrying Pathologic Nodal Involvement or Intrapulmonary Metastases Among Clinical Stage IA Lung Cancer Moderate/poor versus good 2.64 1.57 4.43 0.001 P1 3 versus P0 2.40 1.42 4.06 0.001 2.0 versus 2.0 1.73 0.99 3.00 0.053 pathologic local or regional nodal involvement or intrapulmonary metastases (p 0.001, p 0.001, p 0.004, respectively). According to multivariable analyses of these three factors (Table 2), grade of differentiation and pleural involvement were significant predictors (p 0.001, p 0.001), and maximum tumor dimension was marginally significant (p 0.053). By combining these predictors, we identified subgroups with a high probability of potentially incomplete limited resection among clinical stage IA resectable non small cell lung cancer (Table 3). More than 40% of patients showed pathologic local or regional nodal involvement or intrapulmonary metastases, if patients had more than two of the following three predictors: maximum tumor dimension larger than 2.0 cm in size, moderate or poor differentiation, and pleural involvement by tumor cells. Table 3. Potential Risks of Pathologic Nodal Involvement (pn) or Intrapulmonary Metastases (pm) in Clinical Stage IA Non Small Cell Lung Cancer a Subgroup pn or pm Total clinical stage IA disease 389 88 (23)... Tumor size 2 cm, moderate or poor differentiation, and pleural involvement With all of the factors 36 19 (53) 0.001 With any single factor 256 60 (23) With none of the factors 97 9 (9) Tumor size 2 cm and moderate or poor differentiation With both of the factors 108 45 (42) 0.001 With either of the factors 171 32 (19) With neither of the factors 110 11 (10) Tumor size 2 cm and pleural involvement With both of the factors 55 24 (44) 0.001 With either of the factors 178 44 (25) With neither of the factors 156 20 (13) Moderate or poor differentiation and pleural involvement With both of the factors 54 23 (43) 0.001 With either of the factors 164 44 (27) With neither of the factors 171 21 (12) Table 4. Relationship Between Clinical Features and Lymphatic Vessel Invasion by Tumor Cells Among all Surgically Resected Clinical Stage IA Lung Cancer a Lymphatic Invasion Total 389 107 (28)... Age c Range (median) 28 85 (64)... 0.51 Sex Male 202 51 (25) 0.30 Female 187 56 (30) Pack-years smoking c Range (median) 0 216 (10)... 0.23 Symptoms Asymptomatic 323 87 (27) 0.58 Symptomatic 66 20 (30) CEA (ng/ml) 5.0 232 66 (28) 0.84 5.0 105 31 (30) Side of tumor Left 152 45 (30) 0.46 Right 237 62 (26) Lobar distribution Upper or middle lobe 276 78 (28) 0.60 Lower lobe 113 29 (26) 2.0 187 44 (24) 0.09 2.0 202 62 (31) Histology Squamous 63 12 (19) 0.10 Adenocarcinoma 316 92 (29) 0.14 Others 10 3 (30) Good 204 40 (20) 0.001 Moderate or poor 185 67 (36) P0 300 69 (23) 0.001 P1 P3 89 38 (43) b Probability in univariate analyses in logistic regression model. c Entered as a continuous variable into the logistic regression model. CEA carcinoembryonic antigen. The relationshietween clinicopathologic factors and pathologic lymphatic invasion by tumor cells is summarized in Tables 4 and 5. Among the 389 clinical stage IA patients, 107 (28%) had lymphatic invasion by tumor Table 5. Multivariable Analyses of Factors That Predict Lymphatic Vessel Invasion by Tumor Cells Among Surgically Resected Lung Cancer Moderate/poor versus good 2.11 1.32 3.37 0.002 P1 P3 versus P0 2.34 1.40 3.90 0.001 b 2 test.

Ann Thorac Surg SUZUKI ET AL 2001;72:352 6 STAGE MIGRATIONS IN STAGE I LUNG CANCER 355 Table 6. Probability of the Presence of Lymphatic Invasion by Tumor Cells in Clinical Stage IA Resected Non Small Cell Lung Cancer a Subgroup cells. Univariable and multivariable analyses revealed two significant predictors of the invasion: moderate or poor differentiation and pleural involvement by tumor cells. More than half of the patients with these two factors showed pathologic lymphatic invasion (Table 6). There were 187 clinical stage IA non small cell lung cancers 2.0 cm or less in maximum tumor dimension (Table 7). Multivariable analysis revealed that pleural involvement by tumor cells (odds ratio [OR] 3.53, 95% confidence interval [CI] 1.54 to 8.08, p 0.003), and moderate or poor differentiation (OR 3.06, 95% CI 1.55 to 6.04, p 0.001) were significant predictors of the three risk factors of local recurrence. Twelve (67%) of 18 patients with both of the two predictors had at least one of the three risk factors of local recurrence, whereas only 15 (15%) of 97 patients with neither of the predictors had either of the risk factors (p 0.003). Comment Lymphatic Invasion Total clinical stage IA disease 389 107 (28)... Moderate or poor differentiation and pleural involvement with both of the factors 54 28 (52) 0.001 with either of the factors 164 49 (30) with neither of the factors 171 30 (18) b 2 test. Major lung resection, lobectomy, or pneumonectomy, combined with mediastinal lymph node dissection, has been considered to be the standard treatment for clinical stage IA lung cancer [1], which is supported by the results of the LCSG trial [2]. Limited surgical resection has been suggested as being indicated in compromised patients with limited pulmonary function [6]. Some authors have reported that limited pulmonary resection was feasible when a non small cell lung cancer was small [7, 9]. However, the LCSG randomized trial showed no difference in local or regional recurrence rates, regardless of the tumor size [2]. Although segmentectomy was reported to be preferred to wedge resection of the lung Table 7. Multivariable Analyses of Factors That Predict Pathologic Nodal Involvement, Intrapulmonary Metastases, or Lymphatic Invasion by Tumor Cells Among Stage IA With Lung Cancer 2.0 cm or Smaller in Size Moderate/poor versus good 3.06 1.55 6.04 0.001 P1 P3 versus P0 3.53 1.54 8.08 0.003 because of its presumed higher local control rates, even segmentectomy resulted in more frequent local or regional recurrence than lobectomy in the LCSG trial [2]. The higher local or regional recurrence rate after limited resection was observed even though a negative surgical margin had been confirmed pathologically in their study [2]. This finding was probably because of micrometastasis in local or regional lymph nodes, pulmonary parenchyma, or microscopic occult lymphatic invasion [17]. These micro-involvements were reported to be frequently left undetected [18]. Thus, limited resection is not feasible for every small-sized lung cancer. Resection is contraindicated for lung cancer harboring a high probability of local or regional tumor spread, no matter how small the tumor. Limited surgical resection may be a reasonable approach for a lung carcinoma without invasive spread, whereas major lung resection and complete local or regional lymph node dissection is necessary to determine whether the tumor has local or regional invasive spread. No clinicopathologic factor has been accepted as a predictor of negative local or regional tumor spread, and some investigators have reported that even innocentlooking mediastinal lymph nodes should be dissected routinely [10, 19]. Noguchi and colleagues [20] reported the prognostic significance of central collapse region in small adenocarcinoma of the lung. They concluded that type A or B adenocarcinoma, defined as localized bronchioloalveolar carcinoma with or without central collapse region, were thought to be in situ carcinoma [20]. If their conclusion is true and intraoperative frozen section differential diagnosis is reliably made as to their subtypes, limited lung resection can be indicated for patients with type A or B tumor. However, this differentiation is not as easy to diagnose as other conventional pathologic factors, and an experienced pathologist, who is not available at every institute, is essential. Therefore, simpler predictors of local or regional tumor spread are desired in selecting limited surgical resection. According to our results, lung cancers with pleural involvement, moderate or poor differentiation, or tumors 2.0 cm or larger are considered to have a high probability of local or regional tumor spread. Although we attempted to identify factors predictive of noninvasive lung cancer patients, who are possible candidates for limited resection, no conventional clinicopathologic factors proved to be an acceptable predictor of negative local or regional tumor spread. Instead, we showed that lung cancers showing a pleural tail on plain chest roentgenogram or CT scan, evident pleural indentation on thoracotomy, moderate or poor differentiation, or 2.0 cm or larger in tumor size, were highly likely to show invasive growth. In these populations, limited surgical resection may often result in local or regional incomplete resection and in possible local or regional recurrence. Because our results were based on the routine pathologic examination, prospective study using a molecular marker, such as p53 immunohistochemical staining, might be needed for the confirmation of our results. Based on our results in 187 clinical stage IA non small

356 SUZUKI ET AL Ann Thorac Surg STAGE MIGRATIONS IN STAGE I LUNG CANCER 2001;72:352 6 cell lung cancer cases, 2.0 cm or less in maximum tumor dimension, pleural involvement, or moderate or poor differentiation had a significant association with pathologic lymph node involvement, intrapulmonary metastasis, or lymphatic invasion by tumor cells. Even in lung cancers smaller than 2.0 cm, there was a high probability of local or regional recurrence, when a limited lung resection was performed for lung cancer with pleural involvement, or moderate or poor differentiation. This finding is consistent with previous reports [2 4]. Although some investigators reported that limited surgical resection would be feasible for small sized non small cell lung cancer [7, 9], we conclude that limited surgical resection should be avoided for lung cancer with pleural involvement or moderate or poor differentiation. In conclusion, our results showed that major lung resection should be the treatment of choice for clinical stage IA non small lung cancer when the tumor shows moderate or poor differentiation or pleural involvement, or when the tumor is larger than 2.0 cm in size. We thank Satoshi Sasaki, MD, Epidemiology and Biostatistics Division, National Cancer Center Research Institute East, for his technical support in the statistical analyses. We also thank Prof J. Patrick Barron, International Medical Communications Center, Tokyo Medical University, for reviewing the English usage in the manuscript. Supported in part by a Grant-in-Aid for Cancer Research from the Ministry of Health and Welfare, Japan. References 1. Beattie EJ. The surgical treatment of lung tumors. Pneumonectomy or lobectomy. Surgery 1957;42:1124 8. 2. Ginsberg RJ, Rubinstein LV. Randomized trial of lobectomy versus limited resection for T1 N0 non-small cell lung cancer. Lung Cancer Study Group. Ann Thorac Surg 1995;60: 615 23. 3. Warren WH, Faber LP. Segmentectomy versus lobectomy in patients with stage I pulmonary carcinoma. Five-year survival and patterns of intrathoracic recurrence. J Thorac Cardiovasc Surg 1994;107:1087 94. 4. Landreneau RJ, Sugarbaker DJ, Mack MJ, et al. Wedge resection versus lobectomy for stage I (T1 N0 M0) non-smallcell lung cancer. J Thorac Cardiovasc Surg 1997;113:691 700. 5. Luketich JD, Ginsberg RJ. Limited resection versus lobectomy for stage I non-small cell lung cancer. In: Pass HI, Mitchell JB, Johnson DH, Turrisi AT, eds. Lung cancer: principles and practice. Philadelphia: Lippincott-Raven, 1996:561 6. 6. Miller JI Jr. Limited resection of bronchogenic carcinoma in the patient with impaired pulmonary function. Ann Thorac Surg 1993;56:769 71. 7. Tsubota N, Ayabe K, Doi O, et al. Ongoing prospective study of segmentectomy for small lung tumors. Study Group of Extended Segmentectomy for Small Lung Tumor. Ann Thorac Surg 1998;66:1787 90. 8. Jensik RJ. Miniresection of small peripheral carcinomas of the lung. Surg Clin North Am 1987;67:951 8. 9. Kodama K, Doi O, Higashiyama M, Yokouchi H. Intentional limited resection for selected patients with T1 N0 M0 nonsmall-cell lung cancer: a single-institution study. J Thorac Cardiovasc Surg 1997;114:347 53. 10. Asamura H, Nakayama H, Kondo H, Tsuchiya R, Shimosato Y, Naruke T. Lymph node involvement, recurrence, and prognosis in resected small, peripheral, non-small-cell lung carcinomas: are these carcinomas candidates for videoassisted lobectomy? J Thorac Cardiovasc Surg 1996;111: 1125 34. 11. Aronchick JM. CT of mediastinal lymph nodes in patients with non-small cell lung carcinoma. Radiol Clin North Am 1990;28:573 81. 12. World Health Organization. Histological typing of lung tumors. Geneva: World Health Organization, 1981. 13. Sobin LH, Wittekind C. International Union Against Cancer: TNM classification of malignant tumours. New York: Wiley- Liss, 1997. 14. Naruke T, Suemasu K, Ishikawa S. Lymph node mapping and curability at various levels of metastasis in resected lung cancer. J Thorac Cardiovasc Surg 1978;76:832 9. 15. Kurokawa T, Matsuno Y, Noguchi M, Mizuno S, Shimosato Y. Surgically curable early adenocarcinoma in the periphery of the lung. Am J Surg Pathol 1994;18:431 8. 16. Dawson-Saunders B, Trapp RG. Statistical methods for multiple variables. In: Dawson-Saunders B, Trapp RG. Basic and clinical biostatistics. Norwalk, CT: Appleton & Lange, 1994: 210 31. 17. Ichinose Y, Yano T, Yokoyama H, Inoue T, Asoh H, Katsuda Y. The correlation between tumor size and lymphatic vessel invasion in resected peripheral stage I non-small-cell lung cancer. A potential risk of limited resection. J Thorac Cardiovasc Surg 1994;108:684 6. 18. Izbicki JR, Passlick B, Hosch SB, et al. Mode of spread in the early phase of lymphatic metastasis in non-small-cell lung cancer: significance of nodal micrometastasis. J Thorac Cardiovasc Surg 1996;112:623 30. 19. Graham AN, Chan KJ, Pastorino U, Goldstraw P. Systematic nodal dissection in the intrathoracic staging of patients with non-small cell lung cancer. J Thorac Cardiovasc Surg 1999; 117:246 51. 20. Noguchi M, Morikawa A, Kawasaki M, et al. Small adenocarcinoma of the lung. Histologic characteristics and prognosis. Cancer 1995;75:2844 52.