Endoscopic Ultrasound-Guided Fine Needle Aspiration for Staging Patients With Carcinoma of the Lung

Transcription

1 Endoscopic Ultrasound-Guided Fine Needle Aspiration for Staging Patients With Carcinoma of the Lung Michael B. Wallace, MD, MPH, Gerard A. Silvestri, MD, MS, Anand V. Sahai, MD, MS (Epid), Robert H. Hawes, MD, Brenda J. Hoffman, MD, Valerie Durkalski, MPH, Winnie S. Hennesey, RN, MSN, and Carolyn E. Reed, MD Division of Gastroenterology and Hepatology/Digestive Disease Center, Division of Pulmonary and Critical Care Medicine, The Clinical Innovation Group, and Division of Cardiovascular and Thoracic Surgery, Medical University of South Carolina, Charleston, South Carolina Background. Endoscopic ultrasound (EUS)-guided fine needle aspiration is a safe, cost-effective procedure that can confirm the presence of mediastinal lymph node metastases and mediastinal tumor invasion. We studied the accuracy of EUS in a large population of lung cancer patients with and without enlarged mediastinal lymph nodes on computed tomographic (CT) scan. Methods. From 1996 to 2000 all patients referred to our institution with lung tumors and no proven distant metastases were considered for EUS and surgical staging. Patients had endoscopic ultrasound with fine needle aspiration of abnormal appearing mediastinal lymph nodes and evaluation for mediastinal invasion of tumor (stage III or IV disease). Patients without confirmed stage III or IV disease had surgical staging. Results. Two hundred seventy-seven patients met the inclusion criteria, including 121 who had EUS. Endoscopic ultrasound and fine needle aspiration detected stage III or IV disease in 85 of 121 (70%). Among patients with enlarged lymph nodes on CT, 75 of 97 (77%) had stage III or IV disease detected by EUS. Among a small cohort of patients without enlarged mediastinal lymph nodes on CT, 10 of 24 (42%) had stage III or IV disease detected by EUS. For mediastinal lymph nodes only, the sensitivity of endoscopic ultrasound and CT was 87%. The specificity of EUS (100%) was superior to that of CT (32%) (p < 0.001). Conclusions. Endoscopic ultrasound with fine needle aspiration identified and histologically confirmed mediastinal disease in more than two thirds of patients with carcinoma of the lung who have abnormal mediastinal CT scans. Although mediastinal disease was more likely in patients with an abnormal mediastinal CT, EUS also detected mediastinal disease in more than one third of patients with a normal mediastinal CT and deserves further study. Endoscopic ultrasound should be considered a first line method of presurgical evaluation of patients with tumors of the lung. (Ann Thorac Surg 2001;72:1861 7) 2001 by The Society of Thoracic Surgeons Nearly half of patients presenting with carcinoma of the lung will harbor metastases to mediastinal lymph nodes. The accuracy of computed tomography (CT) for documenting malignant involvement, especially for lymph nodes less than 1 cm, is poor [1 4]. The presence of mediastinal lymph node metastases has important prognostic and therapeutic implications. Thus, histologic documentation of lymph node metastases is usually required. Methods for such verification include CT-guided lymph node biopsy, transbronchial biopsy, mediastinoscopy, anterior mediastinotomy (Chamberlain procedure), video-assisted thoracoscopic surgery, and thoracotomy. The latter four procedures require general anesthesia and can be costly. Except for thoracotomy, all of these methods assess only certain nodal levels and have varied results. Endoscopic ultrasound (EUS) with fine-needle aspiration (FNA) has been shown to be highly accurate for diagnosis of mediastinal lymph node involvement in the radiographic aortopulmonary region, the subcarinal region, and the inferior mediastinum. Accuracy in smaller studies has ranged from 94% to 96% [5 7]. Endoscopic ultrasound has the advantage of being noninvasive, safe, and cost effective [8]. We integrated EUS into our staging evaluation of lung tumors and report our results in 121 patients who had mediastinal staging by EUS for lung tumors and an additional 156 patients who had surgical staging only during the same period. The primary aim of this retrospective series was to determine the frequency with which EUS and guided FNA identified advanced (stage III or IV) disease. Accepted for publication Aug 6, Address reprint requests to Dr Wallace, Division of Gastroenterology and Hepatology, Medical University of South Carolina, 96 Jonathan Lucas St, CSB 210, Charleston, SC 29425; wallacem@musc.edu. Material and Methods From January 1996 to December 2000, we retrospectively reviewed records of all patients with lung carcinoma who 2001 by The Society of Thoracic Surgeons /01/$20.00 Published by Elsevier Science Inc PII S (01)

2 1862 WALLACE ET AL Ann Thorac Surg ENDOSCOPIC ULTRASOUND FOR LUNG CANCER 2001;72: had invasive staging (EUS or surgical). Patients who were part of previously reported studies by our group [7] were not included. Patients were excluded if they had been previously treated with radiotherapy or chemotherapy for lung cancer. Patients were also excluded if no malignancy was found or if definitive pathologic staging of the mediastinal lymph node was not complete (ie, negative EUS but patient could not undergo surgical staging). Patients were considered to have definitive pathologic staging of the mediastinum if EUS FNA or biopsy obtained surgically (thoracoscopy, mediastinoscopy, thoracotomy) proved malignant. If EUS FNA was negative for mediastinal spread of tumor, patients were included only if they subsequently had surgical resection of the mediastinal lymph node. The Institutional Review Board of The Medical University of South Carolina approved the study. All patients initially were reviewed by a multidisciplinary tumor board and subsequently referred for staging. During the initial study period 1996 to 1999, patients were selected for EUS if a CT scan showed a suspicious lung mass, no definitive metastatic disease, and enlargement ( 1.0 cm in short axis diameter) of mediastinal lymph nodes in the region of the subcarina (level 7), aortopulmonary window (level 5), or periesophagus (level 8). After interim analysis of our data (showing a high rate of detection by EUS in patients with abnormal CT scans), the criteria for performing EUS were expanded to include all patients with nonmetastatic lung tumors. A pulmonary radiologist at the tumor board meeting interpreted CT scans. Films from referring hospitals were not repeated unless the quality was considered inadequate for accurate interpretation. The location of suspicious nodes and the disease stage were categorized as defined by the cancer staging manual of the American Joint Commission on Cancer Staging [9]. Sensitivity, specificity, and positive and negative predictive values of EUS FNA for correctly identifying the presence or absence of malignancy in the specimen were calculated using FNA (for positive cases) or surgical exploration and pathologic confirmation as the gold standard. Cases were considered negative only if they had complete mediastinal lymph node dissection. After obtaining informed consent, transesophageal EUS was performed on an outpatient basis under conscious sedation. Lymph nodes were selected for sampling by EUS if they had any of the described features of malignancy (size 1 cm, round shape, sharp margin, diffusely hypoechoic) [10]. All suspicious mediastinal lymph node stations were sampled according to the following scheme. Contralateral lymph nodes were sampled first and immediately stained and read by an onsite pathologist. If no malignant cells were seen, any suspicious midline (subcarinal) or ipsilateral lymph nodes were then sampled. All aspirates were stained immediately using a Diff Quik stain for immediate interpretation. Aspirates were obtained until a preliminary cytologic diagnosis of malignancy was given or until the cytologist believed that an adequate quantity of representative tissue had been obtained (usually four to five aspirates Table 1. Characteristics of 277 Patients Staged With Lung Tumors Final Histology n (%) Adenocarcinoma 102 (37) Squamous cell 100 (36) Large cell 18 (6) Non small cell, not specified 39 (14) Small cell 16 (6) Unknown 2 (0.7) Location of primary tumor Left upper lobe 60 (22) Left lower lobe 29 (10) Right upper lobe 92 (33) Right mid lobe 24 (9) Right lower lobe 39 (14) Multiple lung locations 33 (12) per lesion). The definitive cytologic diagnosis was given only after complete staining and interpretation were done by the pathology department later. Lymph nodes were considered malignant only if the cytologic or surgically resected sample showed malignant cells. Lymph nodes were considered benign if both the cytology and all surgically resected lymph nodes from that level (eg, subcarina) were negative for malignancy. Demographic and proportional data were tabulated using Microsoft Access and Excel (Microsoft, Redmond, WA). Univariate and multivariate analyses were performed using SAS version 8.0 (SAS Institute, Cary, NC). The type I error was set at 0.05 for all analyses. Confidence intervals were calculated to 95% using standard formulas [11]. Results From January 1996 to December 2000, 310 patients had invasive staging of the mediastinum (EUS FNA, surgical, or both) for carcinoma of the lung. Eight patients were excluded because of prior radiotherapy or chemotherapy, and 25 were excluded because they were not candidates for surgical staging after a normal EUS. The remaining 277 patients had definitive staging of the mediastinum, including 121 staged by EUS (and surgery if EUS was negative) and 156 staged by surgery (mediastinoscopy, thoracoscopy thoracotomy, of a combination of them). There was no morbidity associated with the EUS, and all patients were discharged after outpatient recovery. Table 1 shows the final histology and location of the primary tumor for patients in that group (n 277). A flow diagram of the patients staging evaluation is shown in Figure 1. Group Staged by Endoscopic Ultrasound One hundred twenty-one patients were selected for EUS. A total of 152 lymph nodes were sampled by FNA (range 0 to 3, mean 1.25). No complications were observed. Ninety-seven patients were selected for EUS based on the presence of enlarged mediastinal lymph nodes on CT

3 Ann Thorac Surg WALLACE ET AL 2001;72: ENDOSCOPIC ULTRASOUND FOR LUNG CANCER 1863 Fig 1. Flow chart of staging outcome. (Chemo/RT chemoradiation therapy; EUS endoscopic ultrasound; FNA fine-needle aspiration.) scan. During the final phase of the study, 24 patients without enlarged mediastinal lymph nodes were staged by EUS. In the 121 patients, EUS detected mediastinal spread of tumor and thus avoided further surgical staging in 70% (85 of 121) (95% confidence interval [CI] 64%, 76%). These included cytologically proven lymph node metastases in 73 patients and mediastinal invasion of the primary tumor in 12 patients. Fourteen patients with lung lesions of unknown histology were diagnosed with small cell lung cancer by EUS FNA, all with mediastinal lymph node metastases. The 36 patients without mediastinal disease evident by EUS had surgical staging. Ten patients had mediastinoscopy (4), thoracoscopy (4), or both (2) to reevaluate enlarged lymph nodes (all at level 5 or 7) on CT scan. Mediastinoscopy or thoracoscopy detected mediastinal lymph node metastasis in 5 patients (all in level 5 or 7). The remaining 31 patients (26 who had no mediastinoscopy or thoracoscopy plus 5 who had negative mediastinoscopy or thoracoscopy) had open thoracotomy with complete mediastinal lymph node resection. Surgical staging found advanced disease (stage III or IV) in 13 of 36 patients (36% false-negative EUS) including 11 with nodal metastases, 1 with mediastinal invasion of tumor, and 1 with T3N1 (American Joint Commission on Cancer Staging stage IIIA but no mediastinal disease). Overall, the sensitivity of EUS to detect advanced (stage III or IV) disease was 87% (85 of 98) (95% CI 84%, 90%). For the detection of mediastinal lymph node metastases alone, EUS and CT had similar sensitivity (87% for both), but EUS had significantly better specificity (100% versus 32%, p 0.001; Table 2). Of the 11 patients in whom EUS missed mediastinal lymph node metastases, 4 were level 5 lymph nodes (all had negative EUS FNA of a level 5 lymph node), 3 were level 7 (2 had negative EUS FNA of a level 7 lymph node, 1 had no level 7 lymph node seen by EUS), 2 were level 4, 1 was level 8 (not seen by EUS), and 1 was level 6 (not seen by EUS). All patients in whom Table 2. Comparison of Endoscopic Ultrasound and Computed Tomographic Staging of Mediastinal (N2 or N3) Lymph Node Metastases to the Final Pathologic Stage Path Stage N2 or N3 Path Stage N0 N1 Sensitivity Specificity Endoscopic ultrasound 73/84 (87%) 37/37 (100%) N2 or N N0 or N Computed tomography 73/84 (87%) 13/37 (32%) N2 or N N0 or N

4 1864 WALLACE ET AL Ann Thorac Surg ENDOSCOPIC ULTRASOUND FOR LUNG CANCER 2001;72: Fig 2. Results of endoscopic ultrasound (EUS) staging of mediastinal lymph nodes according to the computed tomographic (CT) scan results. EUS FNA missed malignant cells, lymph nodes were diffusely infiltrated with tumor, except one which had a microscopic focus of metastasis. In each location however, multiple lymph nodes were sampled during surgery, and both malignant and nonmalignant lymph nodes were found in that location. Computed tomographic scan was a poor predictor of the ability of EUS to detect mediastinal disease (Fig 2, Table 3). Among patients with abnormal mediastinal CT, EUS confirmed mediastinal lymph node metastases or mediastinal invasion of tumor in 77% (75 of 97) (95% CI 71%, 83%). Sixty-six of the 75 patients had pathologically proven mediastinal lymph node metastases, and 9 patients were T4. Among the 24 patients whose CT suggested no mediastinal adenopathy, EUS detected mediastinal lymph node metastases or mediastinal invasion of tumor (T4) in 42% (10 of 24) (95% CI 25%, 59%). Seven of the 10 patients had pathologically proven lymph node metastases, and 3 patients had T4 tumors. The location of enlarged mediastinal lymph node on CT scan was a good predictor of the ability of EUS to confirm mediastinal metastases. In 70 patients, the specific location of the enlarged lymph nodes was stated on the CT report. Among patients where CT showed enlarged lymph nodes at level 5, 7, or 8 (the locations typically accessed by EUS), EUS confirmed mediastinal malignancy at level 5, 7, or 8 in 96% (64 of 67) (95% CI 92%, 100%). Among patients where CT showed enlarged lymph nodes in mediastinal locations other than level 5, Table 3. Relationship Between the Location of the Enlarged Lymph Node on Computed Tomographic Scan and the Location of Proven Lymph Node Metastases by Endoscopic Ultrasound a CT enlarged lymph node level 5, 7, 8 CT enlarged lymph node at level other than 5, 7, 8 EUS FNA level 5, 7, 8 EUS FNA at level other than 5, 7, b 3 0 a Patients whose lymph node location was not specified on CT were excluded. b The three lymph node metastases found by EUS FNA were at levels 2 (1 patient) and 4 (2 patients). EUS endoscopic ultrasound; FNA fine-needle aspiration. 7, or 8, EUS confirmed mediastinal malignancy lymph node metastases to level 5, 7, or 8 in 3 of 3. Most (95%; 72 of 76; 95% CI 91%, 99%) lymph nodes proven to be malignant by EUS FNA were in level 5 or 7, with fewer in level 2 [1], level 4 [2], or level 8 [1]. Non-EUS Staged Group During the same period, 156 patients had direct surgical exploration (thoracoscopy, mediastinoscopy, thoracotomy, or a combination) without EUS staging. Of the 156 patients 29% (95% CI 23%, 35%) were found to have advanced disease (stage III or IV), including lymph node metastases in 27, mediastinal invasion of tumor in 9, distant metastases in 5, and T3N1 (stage IIIA) in 5. Thirty-six of 156 patients (23%) (95% CI 17%, 29%) had CT evidence of enlarged mediastinal lymph nodes but did not undergo EUS for logistic reasons or surgeon preference. Fifteen of those 36 (42% [95% CI 28%, 56%]) were found at surgery to have mediastinal lymph node metastases, all in locations potentially accessible (level 5, 7, or 8) by EUS. Of these 36 patients, additional surgical staging was done by mediastinoscopy in 23, which detected 8 with mediastinal lymph node metastases and 1 with mediastinal invasion of tumor. Thoracoscopy was performed in 4 of 36 patients, and mediastinal lymph node metastases were detected in 2 of 4. The remaining 25 had resection, and pathologic findings showed 4 with mediastinal lymph node metastases. Comment In this series, EUS with FNA of mediastinal lymph nodes identified advanced mediastinal disease in 70% of patients with carcinoma of the lung and enlarged mediastinal lymph nodes on CT scan. Endoscopic ultrasound also identified such metastases in approximately one third of patients without enlarged mediastinal lymph node on CT scan. In addition to detecting advanced nodal disease, EUS and EUS-guided FNA also identified T4 disease in 12 patients; EUS also diagnosed small cell lung cancer in 14 patients, rendering those patients inoperable. More than 164,000 new cases of carcinoma are identified in the United States every year [12]. The therapy for carcinoma of the lung is largely based on the histology

5 Ann Thorac Surg WALLACE ET AL 2001;72: ENDOSCOPIC ULTRASOUND FOR LUNG CANCER 1865 (small cell versus non-small cell) and the presence of mediastinal or distant spread of tumor. The presence of mediastinal lymph node metastases or mediastinal invasion of tumor is considered an advanced stage (IIIA or IIIB) according to the American Joint Commission on Cancer [9]. Patients with stage IIIB disease have a poor prognosis and are generally treated with chemoradiotherapy without surgery. The treatment of stage IIIA disease is more controversial, but many centers including ours, treat IIIA disease with chemoradiotherapy unless surgery is added under investigational protocols. These patients have a poorer prognosis compared with patients with stage I or II disease [13]. There are multiple options for staging the mediastinum. Computed tomography, which had limited accuracy in this report and others [1, 3, 4, 14], can characterize only the size and location of lymph nodes. Computed tomography guided biopsy of the mediastinum is limited by surrounding bony (eg, sternum, spine) and cardiovascular structures. Positron emission tomography is a noninvasive method but has an accuracy of approximately 85% for mediastinal lymph nodes and is particularly limited for lymph nodes less than 1 cm [15, 16]. Because of the implications for therapy, pathologic staging is generally preferred. Bronchoscopy with transbronchial fine needle aspiration or biopsy of the subcarina and hilum is a safe, well-tolerated procedure, with a sensitivity of approximately 53% to 70%. However, it is not capable of accessing the aortopulmonary window (level 5) or inferior mediastinal lymph nodes (level 8) [17]. Mediastinoscopy and thoracoscopy are highly accurate methods of mediastinal staging but are more costly procedures, and each procedure has access to specific regions (levels 2, 4, and 7 for mediastinoscopy and levels 2, 4, 7, 8, and 9 on the right and 5, 6, 8, and 9 on the left for thoracoscopy). Although the cost of mediastinoscopy can be reduced in cases where it is scheduled simultaneously with exploratory open thoracotomy, the savings may be offset by cases where advanced disease was identified by mediastinoscopy, and the time schedule in the operating room for open exploration was unused. Endoscopic ultrasound originally was developed to evaluate gastrointestinal tumors but has ready access to mediastinal structures (especially levels 5, 7, and 8 lymph node stations) through a transesophageal approach. Endoscopic ultrasound can guide FNA with direct visualization of the needle path into the lymph node in real time. Complications of EUS FNA are rare (1 in 327 in one large series) [18]. Several smaller studies have demonstrated the feasibility and safety of EUS and EUS-guided FNA of mediastinal lymph nodes and the left adrenal gland in patients with carcinoma of the lung, lymphoma, and undiagnosed mediastinal masses, as well as benign conditions, such as sarcoidosis and histoplasmosis [5 7, 19 23]. In one series of 35 patients who had nondiagnostic transbronchial biopsy, EUS confirmed malignancy in 25 of 26, and benign disease in 9 of 9 patients, for a sensitivity of 96% and specificity of 100% [5]. The potential cost savings of an EUS-based strategy compared with a surgical staging (mediastinoscopy) has been previously documented by formal cost analyses [8]. Based on those data, we incorporated EUS staging as a routine procedure in patients with lung tumors who presented to our thoracic oncology program. Our study is the largest reported to date and confirms the finding of previous studies, that EUS can be performed safely and can confirm mediastinal lymph node metastases in patients with enlarged mediastinal lymph nodes on CT scan. A major advantage of EUS over CT scan was the improved specificity without sacrificing sensitivity. One in three patients with an enlarged mediastinal lymph node on CT scan was proven by open surgery to have no mediastinal lymph node metastases. The reliance on size criteria alone ( 1 cm in short axis) is inaccurate because of the coexistence of chronic inflammatory disease of the lung from tobacco use in most lung cancer patients. Because EUS-based staging relies on cytologic or histologic proof of malignancy, the specificity is de facto 100%. One limitation of EUS is the imperfect sensitivity (87% in our series). This likely results from both sampling error by FNA methods in a lymph node with micrometastases and the inability of EUS to consistently visualize lymph nodes in the anterior mediastinum. Because ultrasound cannot penetrate through air-filled structures, the region immediately anterior to the trachea (levels 2 and 4) is not well visualized by EUS. In the present series, 7 of 11 missed lymph node metastases were sampled by EUS FNA, but the FNA was interpreted as negative. In each of those patients, however, multiple positive and negative lymph nodes in that specific region (eg, subcarina) were found at surgery. It is not known whether they were missed because of FNA sampling error within a malignant lymph node or because the malignant lymph node was not sampled. Despite these limitations, EUS still identified 87% of all patients with mediastinal spread of tumor. Identification of stage III or IV disease by EUS could avoid further surgical staging in a large proportion of these patients. However some patients might benefit from mediastinoscopy or thoracoscopy after a negative EUS or EUS with only N2 lymph nodes detected when N3 disease is suspected and would change treatment. Because of the imperfect sensitivity of EUS, it is possible that patients with N2 lymph nodes might also have N3 lymph nodes that were not detected by EUS (contralateral level 2 or 4). We minimized this possibility by performing FNA of all suspicious lymph nodes detected by EUS, starting with N3 lymph nodes, and going to N2 lymph nodes if no tumor cells were detected on immediate analysis. In patients with enlarged contralateral level 2 or level 4 lymph nodes (on CT scan) who had only N2 or no lymph nodes detected by EUS, it would be reasonable to perform mediastinoscopy to further evaluate the level 2 or 4 lymph nodes, because this would identify a higher stage (N3, stage IIIB) of disease and alter treatment. That situation did not occur in our study. Other limitations of EUS staging are imperfect accuracy and variability of interpretation of T4 disease. Although the American Joint Commission on Cancer Staging defines T4 disease as invasion into the mediastinum, some

6 1866 WALLACE ET AL Ann Thorac Surg ENDOSCOPIC ULTRASOUND FOR LUNG CANCER 2001;72: patients with only soft tissue invasion but not organ invasion (esophagus, aorta, etc) could be treated by surgical resection. Our results reflect the strict definition according to the American Joint Commission on Cancer Staging stage, but individual surgical practices may vary. In the minority of patients whose advanced disease was based only on T4, our data might overestimate the detection rate of surgically unresectable disease. Last, not all patients who had a negative EUS could undergo definitive surgical staging. We excluded 25 such patients. Although a worst-case scenario is possible (eg, that all EUS-negative cases were false negatives), this is extremely unlikely given the observed results in patients who did have gold standard staging. Our study provides new information about the selection of patients who could benefit from EUS staging. This large series confirms previous studies demonstrating the utility of EUS in patients with enlarged mediastinal lymph node on CT scan [5 7]. The location of the enlarged mediastinal lymph node does not necessarily predict when EUS will be useful. Almost all patients who had enlarged mediastinal lymph nodes on CT had at least one abnormal lymph node in level 5, 7, or 8, the locations most readily accessed by EUS. Three patients in the EUS group had enlarged mediastinal lymph nodes outside the locations typically accessible to EUS (level 5, 7, or 8), but all three were found by EUS to have concomitant metastases to level 5, 7, or 8. Furthermore, EUS could access several lymph nodes in levels 2 and 4 when they were not immediately anterior to the trachea. Our study suggests that patients with mediastinal lymph node metastases will have multiple levels involved, with at least one level accessible to EUS, although the number of such patients is small. Patients with enlarged mediastinal lymph nodes on CT who have a negative EUS-guided FNA should still be considered for mediastinoscopy, thoracoscopy, or both. In this series, one third of patients with a negative EUS had mediastinal lymph node metastases, and half of them were detected by mediastinoscopy or thoracoscopy. Our study provides new information about the utility of EUS in patients without mediastinal lymph node enlargement on CT scan. More than one third of patients whose CT scan showed no mediastinal lymph node enlargement had confirmed mediastinal lymph node metastases. These results, however, are from a small group of patients, and the observed prevalence has wide confidence intervals. This group of patients with negative CT scans needs further study. Among patients without enlarged mediastinal lymph node on CT scan, other criteria could help physicians decide whether EUS should be used. Surgical studies have suggested that the location of the tumor might predict the presence of mediastinal lymph node metastases at certain levels [24]. Lymphatic pathways favor spread to aortopulmonary lymph nodes from left upper lobe tumors and subcarinal nodes from right or left lower lobe lesions [24]. In our study the only clinical factors that predicted the presence of mediastinal lymph node metastases was the presence of enlarged mediastinal lymph node on CT scan. Even that, however, was a weak predictor, with a specificity of 32%. Although the sample of CT-negative patients was small [24], these data suggest that patients without enlarged mediastinal lymph node on CT scan might also benefit from EUS before consideration of surgery. Which subgroups of CT-negative patients most benefit from EUS is the subject of further study. Endoscopic ultrasound can detect evidence of mediastinal spread of tumor in approximately 70% of patients with enlarged mediastinal lymph nodes on CT scan. Surgical staging that is more costly and invasive can thus be avoided. Patients with suspected N3 disease (eg, enlarged N3 lymph node on CT scan), but who have only N2 disease seen at EUS, may still be candidates for further invasive staging. Mediastinal lymph node metastases can be detected by EUS when the CT scan suggests that the enlarged lymph nodes are not in locations typically accessible to EUS (levels 5, 7, and 8) and among patients without enlarged mediastinal lymph nodes. The utility of EUS for staging patients without enlarged mediastinal lymph nodes detected by CT needs further study. References 1. McKenna RJ Jr, Libshitz HI, Mountain CE, McMurtrey MJ. Roentgenographic evaluation of mediastinal nodes for preoperative assessment in lung cancer. Chest 1985;88: McLoud TC, Bourgouin PM, Greenberg RW, et al. Bronchogenic carcinoma: analysis of staging in the mediastinum with CT by correlative lymph node mapping and sampling. Radiology 1992;182: Izbicki JR, Thetter O, Karg O, et al. Accuracy of computed tomographic scan and surgical assessment for staging of bronchial carcinoma. A prospective study. J Thorac Cardiovasc Surg 1992;104: Arita T, Kuramitsu T, Kawamura M, et al. Bronchogenic carcinoma: incidence of metastases to normal sized lymph nodes. Thorax 1995;50: Fritscher-Ravens A, Soehendra N, Schirrow L, et al. Role of transesophageal endosonography-guided fine-needle aspiration in the diagnosis of lung cancer. Chest 2000;117: 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: Silvestri GA, Hoffman BJ, Bhutani MS, et al. Endoscopic ultrasound with fine-needle aspiration in the diagnosis and staging of lung cancer. Ann Thorac Surg 1996;61:1441 5; discussion Aabakken L, Silvestri GA, Hawes R, Reed CE, Marsi V, Hoffman B. Cost-efficacy of endoscopic ultrasonography with fine-needle aspiration vs mediastinotomy in patients with lung cancer and suspected mediastinal adenopathy. Endoscopy 1999;31: AJCC cancer staging manual. Philadelphia: Lippincott- Raven, Bhutani MS, Hawes RH, Hoffman BJ. A comparison of the accuracy of echo features during endoscopic ultrasound (EUS) and EUS-guided fine-needle aspiration for diagnosis of malignant lymph node invasion. Gastrointest Endosc 1997;45: Landis SH, Murray T, Bolden S, Wingo PA. Cancer statistics, CA Cancer J Clin 1998;48: Dillman RO, Herndon J, Seagren SL, Eaton WL Jr, Green MR. Improved survival in stage III non-small-cell lung

7 Ann Thorac Surg WALLACE ET AL 2001;72: ENDOSCOPIC ULTRASOUND FOR LUNG CANCER 1867 cancer: seven-year follow-up of cancer and leukemia group B (CALGB) 8433 trial. J Natl Cancer Inst 1996;88: Marom EM, Erasmus JJ, Patz EF. Lung cancer and positron emission tomography with fluorodeoxyglucose. Lung Cancer 2000;28: Patz EF Jr, Erasmus JJ, McAdams HP, et al. Lung cancer staging and management: comparison of contrast-enhanced and nonenhanced helical CT of the thorax. Radiology 1999; 212: Harrow EM, Abi-Saleh W, Blum J, et al. The utility of transbronchial needle aspiration in the staging of bronchogenic carcinoma. Am J Respir Crit Care Med 2000;161: Williams DB, Sahai AV, Aabakken L, et al. Endoscopic ultrasound guided fine needle aspiration biopsy: a large single centre experience. Gut 1999;44: Serna DL, Aryan HE, Chang KJ, Brenner M, Tran LM, Chen JC. An early comparison between endoscopic ultrasoundguided fine-needle aspiration and mediastinoscopy for diagnosis of mediastinal malignancy. Am Surg 1998;64: Wiersema MJ, Hassig WM, Hawes RH, Wonn MJ. Mediastinal lymph node detection with endosonography. Gastrointest Endosc 1993;39: White P Jr, Ettinger DS. Tissue is the issue: is endoscopic ultrasonography with or without fine-needle aspiration biopsy in the staging of non-small-cell lung cancer an advance? Ann Intern Med 1997;127: Chang KJ, Erickson RA, Nguyen P. Endoscopic ultrasound (EUS) and EUS-guided fine-needle aspiration of the left adrenal gland. Gastrointest Endosc 1996;44: Ikenberry S, Gress F, Savides T, Hawes R. Fine-needle aspiration of posterior mediastinal lesions guided by radial scanning endosonography. Gastrointest Endosc 1996;43: Asamura H, Nakayama H, Kondo H, Tsuchiya R, Naruke T. Lobe-specific extent of systematic lymph node dissection for non-small cell lung carcinomas according to a retrospective study of metastasis and prognosis. J Thorac Cardiovasc Surg 1999;117: