A Prospective Study of Indications for Mediastinoscopy in Lung Cancer With CT Findings, Tumor Size, and Tumor Markers

GENERAL THORACIC A Prospective Study of Indications for Mediastinoscopy in Lung Cancer With CT Findings, Tumor Size, and Tumor Markers Hideki Kimura, MD, PhD, Naomichi Iwai, MD, PhD, Soichiro Ando, MD, PhD, Kimitaka Kakizawa, MD, PhD, Naoyoshi Yamamoto, MD, Hidehisa Hoshino, MD, and Takashi Anayama, MD, PhD Division of Thoracic Diseases, Chiba Cancer Center, and Department of Thoracic Surgery, Graduate School of Medicine, Chiba University, Chiba City, Japan Background. Biopsies by mediastinoscopy remain the most reliable preoperative staging method for N2 lung cancer. Because it is neither practical nor economical to recommend mediastinoscopy for all candidates for surgery, we developed indicational criteria for videoassisted mediastinoscopy (VAM) and carried out a prospective study to validate its usefulness. Methods. Patients with resectable primary lung cancer were chosen for VAM when at least one of three clinical indicators was present: (1) computed tomographic evidence of mediastinal adenopathy, (2) elevated levels of serologic tumor markers, and (3) diameters of primary cancers (> 2 to 3 cm). Patients without positive nodes (group 2) underwent thoracotomy, and patients with positive nodes (group 3) received induction therapy. When none of these criteria were met (group 1), thoracotomy with R2b lymph node dissection was performed without VAM. Results. One hundred twenty-one men and 82 women (total, 203) were eligible for the study. The mean age of For the management of lung cancer, it is important to identify patients who have mediastinal lymph node metastasis (N2 or N3). Mediastinal node metastasis is identified in approximately 10% of patients having negative computed tomographic (CT) results [1]. Such patients may have a single positive node or microscopic involvement of a single level of mediastinal nodes. When this is the case, as much as a 30% 5-year survival can be expected after complete surgical resection [2, 3], and these patients may further benefit from induction therapy [4 6]. Thus it is important for optimum treatment planning to develop a clinical pathway for detecting mediastinal node metastases. Relatively new diagnostic modalities such as positron emission tomography (PET) scan [7, 8] and ultrasound guided trans-tracheal bronchoscopic biopsies [9, 10] have improved the accuracy of diagnosis compared with that of CT scans. However the limited availability of these Accepted for publication Dec 31, 2002. Address reprint requests to Dr Kimura, Division of Thoracic Diseases, Chiba Cancer Center, 666-2, Nitona-cho, Chuoh-ku, Chiba, 260-8717 Japan; e-mail: hkimura@internet.chiba-cc.pref.chiba.jp. the patients was 64.4 years (range, 39 to 75 years) with primary lung cancer. The patients were comprised of 135 adenocarcinomas, 46 squamous cell cancers, and 22 other carcinomas. There were 78 patients in group 1, 87 in group 2, and 38 in group 3. The stages of group 2 patients were more advanced ( 2 63.2668; p < 0.001) than those of group 1. As the incidence of positive indicators for VAM increased, the ratios of N2 patients increased from 2.5% (all negative) to 90.4% (triple positive: p < 0.001). The correlation of our criteria with the pathology findings revealed a diagnostic sensitivity of 95.8% and a negative predictive value of 97.4%. Using three indicators for N2 prediction, we selected 96% (46 of 48) pn2, N3 patients and avoided 37% (76 of 203) unnecessary VAMs. Conclusions. We established and validated currently useful criteria for VAMs in the management of primary lung cancer. (Ann Thorac Surg 2003;75:1734 9) 2003 by The Society of Thoracic Surgeons techniques and their unproven accuracy restrict their practicality and reliability. Biopsies by mediastinoscopy remain the most reliable preoperative staging method [11, 12]. The efficacy of mediastinoscopy may have been enhanced by the relatively recent advent of videoassisted mediastinoscopy (VAM) [13]. Because it is neither practical nor economical to recommend mediastinoscopy to all candidates for surgery, we have developed indication criteria for cervical mediastinoscopy. We report the outcome of a prospective study to evaluate these criteria. Patients and Methods None of the patients in this study had received anticancer therapy before enrollment. All patients were staged based on the TNM classification of the New International Staging System for Lung Cancer [14], and dissection of lymph nodes was performed in accordance with the lymph node map proposed by the Japan Lung Cancer Society [15]. All patients had to meet the following eligibility criteria for the study: resectable primary lung 2003 by The Society of Thoracic Surgeons 0003-4975/03/$30.00 Published by Elsevier Inc PII S0003-4975(03)00035-3

Ann Thorac Surg KIMURA ET AL 2003;75:1734 9 INDICATIONS FOR MEDIASTINOSCOPY IN LUNG CANCER Table 1. Criteria for Video-Assisted Mediastinoscopy 1. Tumor size (maximum diameter) Adenocarcinoma or undefined histology ( 2 cm) Squamous-cell or large-cell carcinoma ( 3 cm) Small cell carcinoma of any size 2. Size of mediastinal lymph nodes by computed tomography Shortest axis ( 1.0 cm) 3. Elevated levels of tumor markers CEA 5.0 ng/ml (chemiluminescence immunoassay) CA19-9 37 U/mL (chemiluminescence immunoassay) CA-125 35 U/mL (radioimmunoassay) NSE 10 ng/ml (radioimmunoassay) Pro-GRP 46 pg/ml (enzyme-linked immunoadsorbent assay) AFP 10 ng/ml (chemiluminescence immunoassay) CYFRA 3.5 ng/ml (electrochemoluminescence immunoassay) SCC 1.5 ng/ml (enzyme immunoassay) (153-9-1) was approved by the Chiba Cancer Center Ethical Committee of Human Studies and Treatment, and informed consent was obtained from all patients before enrollment. Sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were calculated as follows: Sensitivity (%) true positive/(true positive false negative) Specificity (%) true negative/(true negative false positive) Positive predictive value (%) true positive/(true positive false positive) Negative predictive value (%) true negative/(true negative false negative) Accuracy (%) (true positive true negative)/total patients 1735 GENERAL THORACIC AFP alpha feto protein; CA19-9 cancer antigen 19-9; CA125 cancer antigen 125; CEA carcinoembryonic antigen; CYFRA cytokeratin 19 fragment; NSE neuron-specific enolase; Pro- GRP pro-gastrin releasing peptide; SCC squamous cell carcinoma. cancer in clinical stages I to IIIA as demonstrated by histology or cytology, less than 75 years of age, and performance status 0-1 (World Health Organization classification). Advance informed consent was required to enroll in this study. For staging and pathologic diagnosis of lung cancer, the following were mandatory: chest radiography, CT scan, bronchofiberscopy, and sputum cytology. When the possibility of extrathoracic metastasis was indicated clinically, brain magnetic resonance imaging, abdominal ultrasonography, and 67Ga scan or 99m Tc bone scan were done. Indication criteria for VAM are shown in Table 1. Any that was satisfied was referred to as positive, and any that was not satisfied was referred to as negative. Small cell lung cancer was also an indication for VAM. Thoracotomy with R2b lymph node dissection was recommended when all criteria for VAM were negative. Video-assisted mediastinoscopy was performed using the 10970BR Karl Storz mediastinoscope (Karl Storz, Tuttlingen, Germany), OTV-SX Olympus video-system (Olympus, Tokyo, Japan) and a harmonic scalpel (Ethicon Endo-Surgery, Cincinnati, Ohio). Usually, all the accessible lymph nodes from no. 1 superior mediastinal to no. 7 subcarinal were sampled. Video-assisted mediastinoscopy was stopped either when a positive node was identified by frozen section histologic examination or when sampling had reached position the no. 7 subcarinal lymph nodes without having encountered any apparent cancer. Statistical Analysis All data were managed and analyzed by the Statistics Division of the Chiba Cancer Center. The 2 test was used to detecting differences in statistical analysis, and odds ratios were calculated by statistical program Stat Mate III for Macintosh (ATMS, Tokyo, Japan). This study The terms true positive, true negative, false positive, and false negative, in which pn2, N3 was confirmed either by pathology using VAM or by using thoracotomy, and pn0, N1 was confirmed by using thoracotomy with R2b lymph node dissections were defined as follows: True positive: pn2, N3 patients who satisfied each True negative: pn0, 1 patient who did not satisfy each False positive: pn0, 1 patient who satisfied each False negative: pn2, N3 patients who did not satisfy each When all three indicators were combined, the definitions were: True positive: pn2, N3 patients who satisfied at least one True negative: pn0, N1 who satisfied none of the criteria False positive: pn0, N1 who satisfied at least one False negative: pn2, N3 patients who satisfied none of the criteria Numbers of positive indicators for VAM were expressed as follows: Single positive: one of three indicators was satisfied. Double positive: two of three indicators were satisfied. Triple positive: all indicators were satisfied. Results Number of Patients Five hundred fifty-nine primary lung cancer patients were referred to us between October 1998 and April 2002 (Fig 1). However, 352 of these patients were excluded from the study because of stage, performance status,

GENERAL THORACIC 1736 KIMURA ET AL Ann Thorac Surg INDICATIONS FOR MEDIASTINOSCOPY IN LUNG CANCER 2003;75:1734 9 T4N0 (pleural dissemination). There were 2 N2 patients (2 of 78; 2.6%), as a result of pre-tracheal lymph node metastasis (station no. 3) and pretracheal and tracheobronchial (no. 4) lymph node metastasis. Among the thoracotomy patients in group 2, there were 52 stage I patients, 15 stage II, 6 stage IIIA, 10 stage IIIB, and 4 stage IV (Table 2). The stages of group 2 patients were significantly more advanced ( 2 63.2668; p 0.001) than those of group 1. There was 1 pn3 patient and 6 pn2 false negative patients. Reasons for the false negatives indicators were insufficient lymph node dissection caused by technical error in 1 N3 patient; no. 5 subaortic lymph node metastasis in 5; and micrometastasis to subcarinal lymph nodes in the others. The causes for stage IIIB patients were pulmonary metastasis (PM1: T4) in 8, infiltration to the left atrium (T4) in 1, and the 1 N3 patient previously mentioned. The cause for stage IV (4 patients) was pulmonary metastasis to other lobes (PM2). The diagnostic sensitivity, specificity, accuracy, positive predictive value, and negative predictive value for VAMs of mediastinal involvement were 84.4% (38 of 45), 100% (80 of 80), 94.4% (118 of 125), 100% (38 of 38), and 91.9% (80 of 87), respectively. Fig 1. Primary lung cancer patients were allocated to resectable and nonresectable groups according to the eligibility criteria for the study of video-assisted mediastinoscopy (VAM). Eligible patients were divided into VAM and thoracotomy (group 1) patients using the indication criteria for VAM. The VAM group was further divided into negative (group 2) and positive (group 3) on the basis of the pathological results of mediastinal lymph node metastasis. Group 2 patients underwent thoracotomy and group 3 patients underwent chemotherapy or radiotherapy. advanced age, and because of miscellaneous other reasons. Therefore, initially 207 patients were enrolled in the study, but 4 became ineligible because of the presence of unresectable cancer caused by left atrial or pulmonary arterial invasion in 2, and because of avoidance of thoracotomy in response to cardiac insufficiency or patient s refusal in the other 2. Thus, 203 eligible patients remained for analysis. There were 121 men and 82 women with a mean age of 64.4 years (range, 39 to 75 years). There were 135 patients with adenocarcinomas, 46 with squamous cell carcinomas, 7 with adenosquamous cell carcinomas, 8 with small cell carcinomas, 3 with large cell carcinomas, and 4 with other cancers. According to the indication criteria for VAM (Table 1), 125 patients underwent VAM and 78 had thoracotomy without VAM (group 1) (Fig 1). Of the VAM patients, 87 patients were negative for mediastinal involvement and received thoracotomy (group 2). Thirty-eight patients were positive for metastasis and were given chemotherapy or radiation therapy (group 3). Stages of Patients Of those who had thoracotomy without VAM (group 1), there were 74 stage I patients, 1 stage II, 2 stage IIIA, and 1 stage IIIB (Table 2), including 2 T2N2 patients and 1 Correlations Between Three Clinical Indicators for VAM and Pathologic Results of Lymph Node Metastasis There were 155 pn0, N1 and 48 pn2, N3 patients confirmed by pathology after VAM or thoracotomy. Three clinical factors for VAM indication were compared with the pathologic N stages. TUMOR SIZE. There were 101 patients whose tumor size was greater than the criteria comprising 35 pn2, N3 and 66 pn0, N1. There were 102 patients whose tumors were smaller than the criteria including 13 pn2, N3 and 89 pn0, N1 (Table 3). The odds ratio was 3.63 (95% confidence intervals, 1.78 to 7.39) signifying that the probability of pn2, N3 in patients with tumors larger than the criteria was 3.6 times that in those with smaller tumors. When tumor size was used as the indicator to predict mediastinal involvement, the sensitivity, specificity, and accuracy were 72.9%, 57.4%, and 61.0%, respectively (Table 4). Table 2. Postsurgical Stages With and Without Video- Assisted Mediastinoscopy Stage Without Video-Assisted Mediastinoscopy (group 1) With Video-Assisted Mediastinoscopy (group 2) I 74 52 II 1 15 IIIA 2 6 IIIB 1 10 IV 0 4 Total 78 87 Group 1 versus group 2: 2 63.2668; p 0.001. Stages confirmed by pathologic examinations.

Ann Thorac Surg KIMURA ET AL 2003;75:1734 9 INDICATIONS FOR MEDIASTINOSCOPY IN LUNG CANCER Table 3. Correlation Between Tumor Size and Pathologic Results of Lymph Node Metastasis Tumor a size ( ) Tumor b size ( ) Total Table 5. Correlation Between Tumor Markers and Pathologic Results of Lymph Node Metastasis Marker a ( ) Marker b ( ) 1737 Total GENERAL THORACIC pn0, N1 89 66 155 pn2, N3 13 35 48 Total 102 101 203 a Tumor size ( ): tumors smaller than the of tumor size; b Tumor size ( ): tumors larger than the. See Table 1 for details. Odds ratio 3.63; 95% confidence intervals 1.78 7.39. pn0, N1 106 49 155 pn2, N3 13 35 48 Total 119 84 203 a Marker ( ): no tumor marker elevated; b Marker ( ): at least one tumor markers elevated above the normal level. Odds ratio 5.82; 95% confidence interval 2.83 11.9. TUMOR MARKERS. There were 84 patients who had at least one tumor marker that was elevated above the normal level. Of those patients, 35 were pn2, N3 and 49 were pn0, N1 (Table 5). The 119 patients in which none of the tumor markers was found to be elevated consisted of 13 pn2, N3 and 106 pn0, N1 (Table 5). The odds ratio was 5.82 (95% CI, 2.83 to 11.9). When tumor markers were used as the indicator to predict N2 disease, the sensitivity, specificity, and accuracy were 72.9%, 68.4%, and 69.4%, respectively (Table 4). COMPUTED TOMOGRAPHIC FINDINGS. There were 34 patients whose mediastinal lymph nodes were larger than the criteria of the CT findings (CT ); these were comprised of 30 pn2, N3 patients and 4 pn0, N1 patients (Table 6). In 169 patients, CT findings were negative (CT ), consisting of 18 pn2, N3 and 151 pn0, N1 (Table 6). The odds ratio was 62.9 (95% CI, 19.8 to 199.1). The sensitivity, specificity and accuracy of the CT findings were 62.5%, 97.4%, and 89.1%, respectively (Table 4). Correlations Between Three Clinical Indicators for VAM and Lymph Node Metastasis There were 125 patients in whom at least one of the indicators was positive. These patients were examined by VAM. There were 79 pn0, N1 patients and 46 pn2, N3 patients in the VAM group (groups 2 and 3; Table 7); of the pn2, N3 patients, 38 were confirmed by VAM and 8 by thoracotomy. Seventy-eight patients were negative for any indicators and had thoracotomy without VAM (group 1). There were only 2 pn2 patients, whereas the other 76 were pn0 or pn1. The odds ratio was 22.1 (95% CI, 5.18 to 94.3). When all three indicators were used for the prediction of mediastinal involvement, the sensitivity, specificity, accuracy, positive predictive value, and Table 4. Sensitivity, Specificity, and Accuracy of Each Clinical Indicator for Video-Assisted Mediastinoscopy Sensitivity Specificity Accuracy Computed 62.5% (30/48) 97.4% (151/155) 89.1% (181/203) tomographic scan Tumor size 72.9% (35/48) 57.4% (89/155) 61.0% (124/203) Tumor 72.9% (35/48) 68.4% (106/155) 69.4% (141/203) markers All three 95.8% (46/48) 49.0% (46/155) 60.0% (122/203) negative predictive value were 95.8%, 49.0%, 60.0%, 36.8%, and 97.4%, respectively (Table 4). Correlations Between Positive Incidence of 3 Clinical Indicators for VAM and Lymph Node Metastasis There were 21 patients in whom all criteria were positive (triple positive) (Table 8); among these, 19 patients (90.4%) were pn2 or pn3. There were 48 patients in whom two of three criteria were satisfied (double positive), 56 patients who had only one that was positive (single positive) and 78 patients in whom all of the indicators were negative (all negative; group 1). Seventy-six of the 78 all-negative patients (97.4%) and 45 of the 56 single positive patients (80.3%) were pn0 or pn1. As the incidence of positive indicators for VAM increased, the ratios of mediastinal involvement increased from 2.5% (all negative) to 90.4% (triple positive). The differences were statistically significant (all-negative versus single positive; p 0.001; double vs triple p 0.001). Correlation of Histologic Types and Treatment Groups There were 135 patients with adenocarcinomas, 46 patients with squamous cell carcinomas and 22 patients with other histologic types of tumors as detailed in Table 9. Most of the VAM positive patients had adenocarcinomas (ie, 29 of 38 [76%]). There were no N2 squamous cell carcinomas after negative VAMs. Two patients in group 1 with adenocarcinomas who did not have VAMs and 6 patients in group 2 (negative VAMs) were found to have N2 metastases at thoracotomy. Comment As induction therapy followed by thoracotomy with the intent to resect lung cancer has become more standard Table 6. Correlation Between Computed Tomographic Findings and Pathologic Results of Lymph Node Metastasis CT a CT b Total pn0, N1 151 4 155 pn2, N3 18 30 48 Total 169 34 203 a Computed tomography : shortest axis of mediastinal lymph nodes 1.0 cm; b Computed tomography : shortest axis of mediastinal lymph nodes 1.0 cm. Odds ratio 62.9; 95% confidence interval 19.8 199.1.

GENERAL THORACIC 1738 KIMURA ET AL Ann Thorac Surg INDICATIONS FOR MEDIASTINOSCOPY IN LUNG CANCER 2003;75:1734 9 Table 7. Correlations Among Three Clinical Indicators for Video-Assisted Mediastinoscopy and Pathologic Results of Lymph Node Metastasis Negative a Positive b Total pn0, N1 76 79 155 pn2, N3 2 46 48 Total 78 125 203 a Negative: all three indicator types (tumor size, tumor markers, and computed tomographic findings) negative; b Positive: at least one indicator type positive. Odds ratio 22.1, 95% confidence interval 5.18 94.3. Table 8. Correlations Between Positive Incidences of Three Clinical Indicators for Video-Assisted Mediastinoscopy and Pathologic Results of Lymph Node Metastasis pn2, N3 pn0, N1 Total Statistical Difference All negative 2 (2.5%) 76 (97.4%) 78 p 0.001 Single positive 11 (19.6%) 45 (80.3%) 56 ] 2 10.85 Double positive 16 (33.3%) 32 (66.6%) 48 p 0.001 Triple positive 19 (90.4%) 2 (9.5%) 21 ] 2 19.085 Totals 48 155 203 Table 9. Correlation of Histologic Types of Tumors and Treatment Groups Group 1 Group 2 Group 3 Case Totals Adenocarcinoma 57 49 29 135 Squamous cell 15 28 3 46 carcinoma Others 6 10 6 22 Case total 78 87 38 203 Group 1: thoracotomy without video-assisted mediastinoscopy; Group 2: thoracotomy after video-assisted mediastinoscopy (videoassisted mediastinoscopy negative patients); Group 3: chemotherapy or radiotherapy after video-assisted mediastinoscopy (video-assisted mediastinoscopy positive patients). for selected patients with N2 or stage IIIA lung cancer [4 6], strict staging becomes increasingly essential for planning therapy. Mediastinal node sampling by mediastinoscopy is required for patients entering cooperative trials of induction therapy. Along with the increasing use of CT screening for evaluating patients who have possible lung cancers, small cancers are now being detected more often at the early stages of the disease [16, 17]. Mediastinoscopy for all early stage cancer patients is probably not practical nor economical. Therefore, it is urgent to establish proper indications for mediastinoscopy in the management of lung cancer, and it is necessary to evaluate the efficacy of VAM. Although we have not made a formal comparison between VAM and standard mediastinoscopy, VAM offers better visualization and makes possible surgery with 2 surgeons in cooperation, thus facilitating safe dissection, biopsy, and hemostasis [13]. There have been relatively few reports that focus on the indications for mediastinoscopy. Mediastinoscopy has been used to identify mediastinal invasion by T3, T4 tumors as well as to provide access to mediastinal lymph nodes. Most authors have discussed mediastinoscopy in conjunction with indications for thoracotomy. Some authors recommend mediastinoscopy for all patients who are being considered for thoracotomies [11, 18], whereas others insist that mediastinoscopy is indicated only when mediastinal involvement is suspected [19, 20]. We do not agree with either of these viewpoints. It is not practical to perform VAMs in all patients being considered for thoracotomy, because small sized early cancers are being found quite frequently, and it is for this reason that we undertook this study to seek the best available indications for mediastinoscopy. If VAM is recommended only for patients in whom N2 disease is suspected, minimal metastatic N2 disease is likely to be overlooked. The criteria we evaluated and validated may change and evolve as still newer and better noninvasive diagnostic methods become available. Our choice of three clinical indicators as predictors for N2 disease was based in part on our previous findings [21]. Suzuki and colleagues [22] reported that adenocarcinoma patients with negative chest CT scans tumors larger than 2 cm and high carcinoembrionic antigen (CEA) levels, had 43% pathologic N2 disease. Oda and colleagues [23], in their analysis of mediastinal node metastasis in clinical stage I nonsmall cell lung cancer, found that 13% of 524 patients had pathological N2 disease when systematic nodal dissections were done. The incidences of N2/N3 disease in patients with adenocarcinoma smaller than 21 mm and in those with squamous cell carcinoma smaller than 31 mm in diameter were 11.3% (13 of 115) and 2.3% (2 of 86), respectively. These findings guided us in developing our criteria for VAM. Our findings show that the number of positive serologic tumor markers examined correlate closely with the stages of cancer. As the number of positive tumor markers increases, the stage of the patient advances [20]. Most patients whose CEA, cancer antigen 125 (CA125), cytokeratin 19 fragment (CYFRA), or a combination thereof, is elevated have advanced cancers, and their tumors were more resistant to treatment. We tested eight tumor markers in the first screening. There were 31 marker-positive N2 patients among which CEA (27 patients, 87.0%), CA-125 (10 patients, 32.2%), CYFRA (8 patients, 25.8%), and neuron-specific enolase (NSE) (7 patients, 22.5%) were positive. At least one of these markers were positive in marker-positive N2 patients. Therefore it is sufficient to test for these four markers to verify indications for mediastinoscopy. The costs of mediastinoscopy, including that of general anesthesia, histologic examination, and hospital stay are estimated to be at least 10 times the cost of a four-tumormarker examination. Use of three indicators to predict mediastinal involvement is better than that of single indicator. Had we used CT findings as the only predictor of N2 disease, we would

Ann Thorac Surg KIMURA ET AL 2003;75:1734 9 INDICATIONS FOR MEDIASTINOSCOPY IN LUNG CANCER have not done VAMs in 37.5% (18 of 48) of the N2, N3 patients; had we used only tumor size or tumor markers as indications for VAM, we may have failed to select 27.0% (13 of 48) of the N2 patients prior to thoracotomy. Also, if we had used two of the three indicators, 3 to 6 N2 patients would have been exempted from VAMs. However, by using three indicators for N2 prediction, we selected 96% (46 of 48) of the pn2, N3 patients and avoided 37% (76 of 203) of unnecessary VAMs. Pathologic examinations after VAM and thoracotomy indicate that the diagnostic sensitivity and negative predictive value for our postulated indication criteria were 95.8% and 97.4%, respectively. Furthermore, the diagnostic sensitivity, specificity, accuracy, positive predictive value, and negative predictive value for VAMs of mediastinal involvement were 84.4%, 100%, 94.4%, 100%, and 91.9%, respectively. Positron emission tomography was less effective than our criteria, yielding 70% (diagnostic sensitivity), 86% (specificity), and 84% (accuracy), respectively [8]. Thus our criteria for recommending VAM helped us to find patients with N2 cancers without radiographic evidence of mediastinal involvement and are useful for screening N2 patients in the clinical management of lung cancer. As with virtually all diagnostic methods, ours is not perfect. There were 2 patients with right upper lobe carcinomas of N2 disease among the 78 patients, in whom none of the three indicators for VAM was present. In both patients, complete resection of the primary tumors and lymph nodes was accomplished, but these patients may have been better served with induction chemotherapy. Although our criteria for VAMs are still imperfect, they correctly avoided 76 VAM examinations that would have been negative; only 2 patients who would have benefited from VAM did not have the procedure. Our indication criteria for VAMs have proven useful for the selection of patients with minimal N2 disease. We believe they should be further refined and are now ready to use for the collaborative study of multimodality treatment. We thank Dr John R. Benfield for his invaluable help and encouragement and Dr Robert L. Thurer for his insightful comments in preparing this article. We are also grateful to C. W. P. Reynolds for the linguistic assistance with the article. This study (2001-75) is supported in part by the Smoking Research Foundation of Japan. References 1. Patterson GA, Ginsberg RJ, Poon PY, et al. A prospective evaluation of magnetic resonance imaging, computed tomography, and mediastinoscopy in the preoperative assessment of mediastinal node status in bronchogenic carcinoma. J Thorac Cardiovasc Surg 1987;94:679 84. 2. Mountain CF. Surgery for stage IIIa-N2 non-small cell lung cancer. Cancer 1994;73:2589 98. 3. Andre F, Grunenwald D, Pignon JP, et al. Survival of patients with resected N2 non-small-cell lung cancer: evidence for a subclassification and implications. J Clinic Oncol 2000;18: 2981 9. 1739 4. Rosell R, Gomez-Codina J, Camps C, et al. A randomized trial comparing preoperative chemotherapy plus surgery with surgery alone in patients with non-small-cell lung cancer. N Engl J Med 1994;330:153 8. 5. Roth JA, Fossella F, Komaki R, et al. A randomized trial comparing perioperative chemotherapy, and surgery with surgery alone in resectable stage IIIA non-small-cell lung cancer. J Natl Cancer Inst 1994;86:673 80. 6. Choi NC, Carey RW, Daly W, et al. Potential impact on survival of improved tumor down staging and resection rate by preoperative twice-daily radiation and concurrent chemotherapy in stage IIIA non-small-cell lung cancer. J Clin Oncol 1997;15:712 22. 7. Kernstine KH, McLaughlin KA, Menda Y, et al. Can FDG- PET reduce the need for mediastinoscopy in potentially resectable nonsmall cell cancer? Ann Thoracic Surg 2002;73: 394 402. 8. Kernstine KH, Stanford W, Mullan BF, et al. PET, CT, and MRI with Combidex for mediastinal staging in non-small cell lung carcinoma. Ann Thoracic Surg 1999;68:1022 8. 9. Wallace MB, Silvestri GA, Sahai AV, et al. Endoscopic ultrasound-guided fine needle aspiration for staging patients with carcinoma of the lung. Ann Thorac Surg 2001;72: 1861 7. 10. Gress FG, Savides TJ, Sandler A, et al. Endoscopic ultrasonography, fine-needle aspiration biopsy guided by endoscopic ultrasonography, and computed tomography in the preoperative staging of non-small-cell lung cancer: a comparison study. Ann Intern Med 1997;127:604 12. 11. Luke WP, Pearson FG, Todd TR, et al. Prospective evaluation of mediastinoscopy for assessment of carcinoma of the lung. J Thorac Cardiovasc Surg 1986;91:53 6. 12. Rendina EA, Venuta F, De Giacomo T, et al. Comparative merits of thoracoscopy, mediastinoscopy and mediastinotomy for mediastinal biopsy. Ann Thorac Surg 1994;57:992 5. 13. Mouroux J, Venissac N, Alifano M. Combined video-assisted mediastinoscopy and video-assisted thoracoscopy in the management of lung cancer. Ann Thorac Surg 2001;72:1698 704. 14. Mountain CF. Revisions in the international system for staging lung cancer. Chest 1997;111:1710 7. 15. The Japan Lung Cancer Society. Classification of Lung Cancer, 1st English edit. Kanehara & Co, Ltd, Tokyo 2000:6 14. 16. Swensen SJ, Jett JR, Sloan JA, et al. Screening for lung cancer with low-dose spiral computed tomography. Am J Resp Crit Care Med 2002;165:508 13. 17. Sobue T, Moriyama N, Kaneko M, et al. Screening for lung cancer with low-dose helical computed tomography: antilung cancer association project. J Clin Oncol 2002;20:911 20. 18. Funatsu T, Matsubara Y, Hatakenaka R, et al. The role of mediastinoscopic biopsy in preoperative assessment of lung cancer. J Thorac Cardiovasc Surg 1992;104:1688 95. 19. Daly BDT, Mueller JD, Faling LJ, et al. N2 lung cancer: outcome in patients with false-negative computed tomographic scans of the chest. J Thorac Cardiovasc Surg 1993; 105:904 11. 20. Takamochi K, Nagai K, Yoshida J, et al. The role of computed tomographic scanning in diagnosing mediastinal node involvement in non-small cell lung cancer. J Thorac Cardiovasc Surg 2000;119:1135 40. 21. Ando S, Kimura H, Iwai N, et al. Optimal combination of seven tumor markers in prediction of advanced stage at first examination of patients with non-small cell lung cancer. Anticancer Res 2001;21:3085 92. 22. Suzuki K, Nagai K, Yoshida J, et al. Clinical predictors of N2 disease in the setting of a negative computed tomographic scan in patients with lung cancer. J Thorac Cardiovasc Surg 1999;117:593 8. 23. Oda M, Watanabe Y, Shimizu J, et al. Extent of mediastinal node metastasis in clinical stage I non-small-cell lung cancer: the role of systematic nodal dissection. Lung Cancer 1998;22:23 30. GENERAL THORACIC