Accurate mediastinal staging is a critical component

Endobronchial Ultrasound for Lung Cancer Staging: How Many Stations Should Be Sampled? Mark I. Block, MD Thoracic Surgery, Memorial Healthcare System, Hollywood, Florida Background. No guidelines exist regarding the number of mediastinal lymph node stations that should be sampled to ensure adequate preoperative staging of lung cancer patients. In recent reports of esophageal endoscopic ultrasound and endobronchial ultrasound (EBUS), fewer than two stations/patient were sampled. An experience with systematic sampling using EBUS to determine how many stations should be sampled to adequately detect mediastinal disease was evaluated. Methods. Records were reviewed for all patients with lung cancer who had EBUS between March 9, 2006, and January 6, 2009. For each station sampled the sequence and result (positive, negative, and nondiagnostic) were recorded. For those with a positive biopsy, the sequence number of the first positive station was determined. The affect of systematic sampling on detection of N3 disease was also evaluated. Results. Ninety-three patients with non small cell lung cancer had EBUS; 271 mediastinal stations were sampled (range, 0 to 6; mean, 2.9 per patient), with N3 and N2 stations sampled in 51 and 90 patients, respectively. Mediastinal disease was found in 25 patients; 15 patients had multistation N2 disease and 6 patients had N3 disease. In 15 patients (60%), mediastinal disease was detected in the first station sampled; three samples were required to detect 90% of disease, and the remaining 3 patients had their disease detected with the fourth sample. Conclusions. Introduction of EBUS as a tool for non small cell lung cancer staging has led to a shift from systematic nodal sampling to targeted nodal sampling. These results indicate that systematic nodal sampling is feasible with EBUS, and that routinely sampling more than two mediastinal stations may improve staging. (Ann Thorac Surg 2010;89:1582 7) 2010 by The Society of Thoracic Surgeons Accepted for publication Jan 25, 2010. Presented at the Forty-fifth Annual Meeting of The Society of Thoracic Surgeons, San Francisco, CA, Jan 26 28, 2009. Address correspondence to Dr Block, 1150 N 35th Ave, Ste 660, Hollywood, FL 33021; e-mail: mblock@mhs.net. Accurate mediastinal staging is a critical component of optimizing care for patients diagnosed with non small cell lung cancer (NSCLC). Mediastinoscopy remains the gold standard for invasive mediastinal staging, but recent introduction of endoscopic ultrasoundguided techniques greatly increases the number and variety of physicians performing invasive mediastinal staging. Transesophageal endoscopic ultrasound (EUS) -guided fine-needle aspiration is often performed by gastroenterologists, and endobronchial ultrasound (EBUS) -guided transbronchial needle aspiration is often performed by interventional pulmonologists. Thus, invasive mediastinal staging is now being performed by physicians with a greater variety of backgrounds, training, and perspectives on lung cancer care. Current American College of Chest Physicians guidelines recommend systematic sampling of mediastinal lymph node stations at the time of resection [1]. However, no clear guidelines exist regarding the number of stations that should be sampled during preoperative invasive staging. Inadequate sampling can lead to either understaging (missing N3 or N2 disease) or incomplete assessment of the extent of disease (missing multistation disease), both of which have the potential to alter recommended initial therapy. Usually, invasive staging with EUS and EBUS is executed with a strategy of sampling N3, then N2, then N1 nodes. This strategy is designed to eliminate the risk of upstaging patients as a result of contamination of specimens with tumor cells from a positive lower-stage lymph node. In addition, rapid on-site cytologic evaluation (ROSE) of needle aspirates is often used to assess sample adequacy and to detect a positive biopsy. Given this strategy, and assuming that most patients do not have N3 disease, most patients should have at least two mediastinal stations sampled. Yet recent studies of EBUS for lung cancer staging consistently report an average of fewer than two lymph nodes sampled per patient (Table 1). Often the average is close to 1.5, suggesting that in these studies most patients had only one mediastinal lymph node station sampled. The purpose of this study was to review an experience with EBUS for NSCLC staging and to determine whether the adequacy of staging was related to the number of lymph node stations sampled. Because pathology support for ROSE is not available at this institution, samples are prepared in the endoscopy suite and sent to pathology for routine processing. This has led to a practice of routinely sampling multiple stations. 2010 by The Society of Thoracic Surgeons 0003-4975/10/$36.00 Published by Elsevier Inc doi:10.1016/j.athoracsur.2010.01.043

Ann Thorac Surg BLOCK 2010;89:1582 7 HOW MANY STATIONS TO SAMPLE WITH EBUS? 1583 Table 1. Recent Publications of Endobronchial Ultrasound for Mediastinal Sampling in Patients With Lung Cancer Citation N Number of N2/3 Sampled Mean per Patient Rintoul et al [2] 109 118 1.1 Herth et al [3] 97 156 a 1.6 Vincent et al [4] 113 167 1.5 Ernst et al [5] 66 120 1.8 Lee et al [6] 102 163 1.6 Herth et al [7] 502 412 0.8 Yasufuku et al [8] 120 147 1.2 a Includes an undisclosed number of N1 lymph nodes. Patients and Methods Institutional Review Board The Institutional Review Board of the Memorial Healthcare System, Hollywood, FL, approved this retrospective chart review. The requirement for informed consent was waived. Endoscopic Ultrasound All endoscopic ultrasound procedures were performed as a part of routine clinical care for patients with known or suspected lung cancer. All patients had chest computed tomography (CT) scans before the procedure, and most had positron emission tomography (PET) or PET/CT studies available. The decision regarding which procedure(s) to recommend was based on procedure availability and on clinical features such as location of the primary lung tumor, location of abnormal mediastinal lymph nodes on CT and PET, and patient characteristics such as prior mediastinoscopy and limited cervical extension. All patients were evaluated by a single surgeon, and all procedures were performed by the same surgeon. All endoscopic ultrasound procedures (EUS and EBUS) were performed in the endoscopy suite on spontaneously breathing patients under monitored anesthesia care (intravenous sedation, typically with propofol or midazolam). Patients undergoing EBUS also received nebulized lidocaine immediately before the procedure and topical lidocaine during the procedure. All elective procedures were done on an outpatient basis. Esophagogastroduodenoscopy was always performed immediately before EUS, and flexible bronchoscopy was always performed immediately before EBUS. If a combined EUS and EBUS were planned for the same patient, EBUS was usually performed first to take advantage of the nebulized lidocaine. Transesophageal endoscopic ultrasound was performed first only if the patient had suspicious findings in the left adrenal gland on imaging studies. Transesophageal endoscopic ultrasound-guided fineneedle aspiration was performed using a flexible esophagoscope equipped with a linear ultrasound array (model GF-UCT140; Olympus Inc, Tokyo, Japan). Endobronchial ultrasound-guided transbronchial needle aspiration was performed using a videobronchoscope equipped with a linear ultrasound array (model BF-UC160F-OL8, Olympus Inc., Tokyo, Japan). An attempt was made to biopsy all visualized lymph nodes. Biopsies were taken first from contralateral (N3) mediastinal lymph nodes, followed by ipsilateral (N2) and then hilar (N1) lymph nodes. Multiple passes were made at each lymph node, and the needle was flushed with normal saline solution between lymph node stations. Specimens were spread onto glass slides and immediately fixed in alcohol for cytology. Occasionally, material from multiple aspirations of the same lymph node was injected into RPMI solution for processing into a cell block. This facilitated performance of special stains. No sample was evaluated by on-site pathology; however, occasional samples were sent immediately to pathology for a STAT reading. Most samples were processed routinely with pathology results available in 1 to 3 days. Biopsy results were categorized as positive if malignant cells were identified in the specimen, negative if there was an abundance of normal lymphocytes without diagnostic malignancy, or nondiagnostic if there were insufficient lymphocytes to indicate reliable lymph node assessment. Data Collection and Analysis Records for all patients undergoing EUS, EBUS, or both were reviewed and abstracted for the sequence number, lymph node station, and result of each biopsy. Data were entered into a Microsoft Access database (Microsoft Corp, Redmond, WA), and appropriate queries were constructed to analyze the data. Only patients with pathologically proven NSCLC who underwent EBUS were included in the final analyses. This was done because EBUS is being performed by both pulmonologists and surgeons, is emerging as the dominant technology for invasive mediastinal staging, and routinely provides access to the majority of involved mediastinal lymph nodes. Results Lymph Node Sampling Between March 9, 2006, and January 6, 2009, endoscopic ultrasound was performed on 200 patients (EUS, 62; EBUS, 103; EUS and EBUS, 35). Of these, 136 (EUS, 43; EBUS, 71; EUS EBUS, 22) had NSCLC proven either before the procedure, as a result of the procedure, or at subsequent resection. As shown in Table 2, a total of 379 mediastinal lymph node stations were sampled in these 136 patients (mean, 2.8 per patient). Ninety-three patients had EBUS with sampling of 271 mediastinal lymph node stations (mean, 2.9 per patient). Figure 1 shows the distribution of the number of mediastinal stations sampled per patient across all 93 patients. Sampling N3 Stations Metastatic disease to level N3 lymph nodes is uncommon. Furthermore, in many patients N3 lymph nodes are small and difficult to find. Therefore, the proportion of patients in whom N3 lymph nodes were sampled is a reflection of the degree to which the EBUS staging strategy was successfully executed. Of the 93 patients with NSCLC who had EBUS, 10 patients had mediastinal

1584 BLOCK Ann Thorac Surg HOW MANY STATIONS TO SAMPLE WITH EBUS? 2010;89:1582 7 Table 2. Number of Lymph Node Stations Sampled in Patients With Non Small Cell Lung Cancer Table 3. Number of Stations Sampled to Detect N2/3 Disease in All 25 Patients ( First Positive Station) Station EUS and EBUS (n 136) a EBUS (n 93) b 1st Positive Station Number of Patients Cumulative Number of Patients N1 46 45 N2 292 201 N3 87 70 N2 N3 379 271 Mean N2/3 per Patient 2.8 2.9 a Four patients, all of whom had only EUS done, had no lymph nodes sampled: 2 had EUS for biopsy of the left adrenal gland, 1 had EUS for biopsy of a mediastinal mass, and 1 had EUS with no biopsies performed. b Two patients had only N1 stations sampled and 1 patient had no lymph nodes sampled by EBUS, but a positive level 7 node sampled by EUS done concurrently. EBUS endobronchial ultrasound; EUS esophageal endoscopic ultrasound. disease only, without a clear primary tumor localized to one lung or the other. In these patients all mediastinal nodes were considered N2. Among the remaining 83 patients, 70 N3 lymph nodes were sampled in 51 patients (61%). Of the stations sampled, 10 were positive, 44 were negative, and 16 were nondiagnostic. On a per patient basis, 6 patients had positive biopsies, 37 patients had negative biopsies, and in the remaining 8 patients all N3 samples were nondiagnostic. Thus, in approximately half of patients (43 of 83), N3 nodes were successfully sampled. Of the 6 patients with EBUS-proven N3 disease, clinical staging suggested N3 disease in 3 patients (2 with CT and PET, 1 with CT only) and N2 disease only in the other 3 (1 with CT and PET, 2 with CT only). Detection of Mediastinal Disease Of the 90 patients who had mediastinal lymph nodes (N2 or N3) sampled, 25 had a positive biopsy. In these 25 patients the number of biopsies required to detect the 1 15 15 (60%) 2 3 18 (72%) 3 4 22 (88%) 4 3 25 (100%) mediastinal disease was determined by identifying the sequence number of the first positive biopsy. Results are shown in Table 3. Although 60% of mediastinal disease was detected with the first station sampled, sampling of four stations was required to detect mediastinal disease in all 25 patients. Sampling in Patients With Negative Mediastinal Nodes Mediastinal sampling was negative in 65 patients. A total of 196 nodal stations were sampled in these 65 patients (mean, 3.0 per patient). The distribution of the number of stations sampled per patient is shown in Figure 2, illustrating that the majority of patients had at least three stations sampled. Of these 65 patients, 47 went on to resection and 6 were found to have mediastinal disease at surgery. Five of these patients had three N2 or N3 stations sampled at EBUS, and the other patient had two stations sampled. Preoperative imaging of these 6 patients with CT or PET/CT suggested mediastinal disease in 3, and were negative in the other 3. Potential Impact of Rapid On-Site Cytologic Evaluation on Number of Stations Sampled Rapid on-site cytologic evaluation is often used not only to confirm that diagnostic material is retrieved but also to identify positive biopsies immediately so that the procedure can be terminated. Therefore, use of ROSE in this Fig 1. Distribution of the number of mediastinal stations sampled per patient. Results for all 93 patients with non small cell lung cancer undergoing endobronchial ultrasound are shown. The number of patients in each group is indicated on top of the corresponding bar. Fig 2. Distribution of the number of mediastinal stations sampled per patient for patients with negative mediastinal sampling. Results for only those 65 patients whose mediastinal sampling was negative are shown.

Ann Thorac Surg BLOCK 2010;89:1582 7 HOW MANY STATIONS TO SAMPLE WITH EBUS? 1585 Table 4. Biopsy Results by Sample Number for All 93 Patients Result manner will limit the number of stations sampled. In this series, 75 stations were sampled in the 25 patients who ultimately proved to have positive mediastinal disease. If sampling had been terminated after the first positive biopsy (as shown in Table 2), only 45 stations would have been sampled (mean, 1.8 per patient). However, ROSE would not have affected the number of stations sampled in the N2- or N3-negative patients. Therefore, had ROSE been used to terminate sampling after the first positive biopsy, the number of stations sampled would have been 241 (45 196), or 2.6 per patient, instead of 271 (2.9 per patient). Reliability of Routine Multistation Sampling A strategy of routine multistation sampling raises two concerns. First, because the initial sampling is targeted to N3 nodes, or less clinically suspicious (smaller) N2 nodes, the incidence of nondiagnostic biopsies may be high. And second, cross-contamination of specimens from a positive biopsy renders results of sampling beyond the first positive biopsy unreliable. Table 4 shows the biopsy results by sample number. As anticipated, the proportion of nondiagnostic biopsies is highest for the first and second samples (21%); only 8% of third samples were nondiagnostic, and all of the subsequent biopsies were diagnostic. Multistation sampling produced positive biopsies in more than one station for 15 of the 25 patients found to have mediastinal disease. Preprocedure imaging with CT or PET/CT suggested multistation disease in 13 of these patients. The remaining 2 patients were seen as inpatients and had only CT scans done before the EBUS. One patient had subcarinal and hilar adenopathy, and the other had only hilar adenopathy. This study is not designed to address the question of cross-contamination, and therefore these results cannot be extrapolated to conclude that 15 patients had true multistation disease. However, of the 25 patients with a positive biopsy, 18 had a subsequent biopsy at a different station; in 6 of these 18 patients (33%), a subsequent biopsy was either negative or nondiagnostic. This indicates that cross-contamination is not inevitable, and suggests that accurate diagnosis of multistation disease with EBUS may be possible without resorting to a new needle for each station. Comment Sample Number 1 2 3 4 5 6 Negative 56 52 42 23 6 1 Positive 15 13 13 7 2 0 Nondiagnostic 19 17 5 0 0 0 Total 90 82 60 30 8 1 % Nondiagnostic 21 21 8 0 0 0 Introduction of endoscopic ultrasound into clinical practice is transforming invasive mediastinal staging of patients with NSCLC. Increasingly gastroenterologists and pulmonologists are performing a function that historically has been performed almost exclusively by thoracic surgeons. In a recent review of practice patterns in lung cancer care, Little and colleagues [9] discovered that there was wide variability in the practice of mediastinoscopy. On the basis of a survey of data on almost 41,000 patients nationwide, they reported that mediastinoscopy was performed in less than 30% of patients undergoing resection, and that when performed no lymph nodes were sampled in more than half the cases. With introduction of EUS and EBUS, the number of patients undergoing invasive mediastinal staging and the number of lymph nodes sampled is likely to increase. This therefore is an opportune moment to establish guidelines for mediastinal sampling that can serve to minimize variability in practice patterns at the same time that a wider variety of physicians are becoming engaged in the process. The European Society of Thoracic Surgery recently published guidelines on preoperative lymph node staging for NSCLC [10]. They concluded that although there are no data to indicate that more systematic nodal sampling is better than less, extrapolation from the surgical data supports that approach. In their review they cited recommendations from the American Thoracic Society [11] and Detterbeck and associates [12] for sampling bilateral upper and lower paratracheal lymph nodes and subcarinal nodes (five stations). The more recent guidelines from the American College of Chest Physicians recommend invasive mediastinal staging for evaluation of mediastinal adenopathy by CT and for assessment of the mediastinum in patients with a normal mediastinum but a central tumor or hilar adenopathy [13]. But the American College of Chest Physicians recommendations stop short of recommending the extent of mediastinal sampling to be performed. In contrast to these recommendations, published studies of endoscopic ultrasound for mediastinal staging generally report that relatively few lymph node stations are sampled. The average number of stations sampled per patient is typically fewer than two, and in most studies is near 1.5. Thus, only one lymph node station is sampled in many if not most patients. If a consistent strategy of sampling N3 and then N2 nodes is used, then even if sampling is terminated after the first positive biopsy by ROSE, at least two stations should be sampled in most patients. Furthermore, the higher the proportion of patients without mediastinal disease, the higher should be the number of stations sampled on average. In the current study, systematic sampling of multiple stations was performed routinely. In the 65 patients without mediastinal disease an average of 3.0 stations were sampled per patient. In the 25 patients with mediastinal disease, even if sampling had been terminated after the first positive biopsy detected by ROSE, the average number of stations sampled would have been 2.7 per patient. Although routine systematic sampling of multiple mediastinal lymph node stations is a logical goal, it is difficult to determine whether it yields any benefit to patients compared with targeted sampling of fewer but more clinically suspicious nodes. The theoretical benefits

1586 BLOCK Ann Thorac Surg HOW MANY STATIONS TO SAMPLE WITH EBUS? 2010;89:1582 7 of more systematic sampling are greater sensitivity for detection of mediastinal disease in general, and N3 and multistation disease in particular. Routine systematic sampling led to biopsy of N3 nodes in half of the patients. In only 6 patients were these biopsies positive, but in 3 the N3 disease was unsuspected. Fifteen of the 25 patients with mediastinal disease (60%) had multistation disease, but this finding is suspect because of the potential for cross-contamination of specimens. The data also show, however, that cross-contamination is not inevitable in 33% of patients who had sampling subsequent to a positive biopsy, a subsequent biopsy did not contain cancer cells. Preprocedure imaging suggested multistation disease in all but 2 of the patients with EBUS evidence of multistation disease. In the remaining 2, PET data were not available and the CT data did not indicate multistation disease. Together these findings suggest that the strategy used in this study was successful in appropriately identifying multistation disease, and that routinely using a new needle for each station may be a significant expense without any clear benefit. The use of a new needle only after a positive biopsy with ROSE would be one way to efficiently target the additional expense to only those patients likely to benefit, but in this institution ROSE is not available. In this study mediastinal disease was missed in 6 patients. Thus, sensitivity for detection of mediastinal disease was only 80% (25 of 31 patients). Most of the false-negative studies were early in the experience, with 5 of the 6 in the first half, and therefore the low sensitivity may reflect a learning curve. Regardless, this finding is disappointingly low compared with existing reports that document a sensitivity of approximately 90% for detection of mediastinal disease with EBUS [3, 8]. The most significant finding of this study is that in many patients, biopsy of multiple stations was required to detect mediastinal disease. Sampling of two stations identified disease in only 72% of the patients ultimately found to have metastases by EBUS, and sampling of three stations was required to approach 90% detection. Although experienced centers may achieve high sensitivity for detection of mediastinal disease with limited mediastinal sampling, these data suggest that with more widespread use of EBUS, limited mediastinal sampling has the potential to understage patients. In general, the recommendations with mediastinoscopy have been to perform systematic nodal sampling. The trend with EBUS seems to be in the opposite direction. As a wider variety of physicians undertakes invasive mediastinal staging, addressing the necessity of systematic nodal sampling assumes greater importance and development of guidelines seems appropriate. References 1. Scott WJ, Howington J, Feigenberg S, Movsas B, Pisters K. Treatment of non-small cell lung cancer stage I and stage II: ACCP evidence-based clinical practice guidelines (2nd edition). Chest 2007;132(Suppl):234S 42S. 2. Rintoul RC, Tournoy KG, El Daly H, et al. EBUS-TBNA for the clarification of PET positive intra-thoracic lymph nodes: an international multi-centre experience. J Thorac Oncol 2009;4:44 8. 3. Herth FJ, Eberhardt R, Krasnik M, Ernst A. Endobronchial ultrasound-guided transbronchial needle aspiration of lymph nodes in the radiologically and positron emission tomography-normal mediastinum in patients with lung cancer. Chest 2008;133:887 91. 4. Vincent BD, El-Bayoumi E, Hoffman B, et al. Real-time endobronchial ultrasound-guided transbronchial lymph node aspiration. Ann Thorac Surg 2008;85:224 30. 5. Ernst A, Anantham D, Eberhardt R, Krasnik M, Herth FJ. Diagnosis of mediastinal adenopathy-real-time endobronchial ultrasound guided needle aspiration versus mediastinoscopy. J Thorac Oncol 2008;3:577 82. 6. Lee HS, Lee GK, Lee HS, et al. Real-time endobronchial ultrasound-guided transbronchial needle aspiration in mediastinal staging of non-small cell lung cancer: how many aspirations per target lymph node station? Chest 2008;134: 368 74. 7. Herth FJ, Ernst A, Eberhardt R, Vilmann P, Dienemann H, Krasnik M. Endobronchial ultrasound-guided transbronchial needle aspiration of lymph nodes in the radiologically normal mediastinum. Eur Respir J 2006;28:910 4. 8. Yasufuku K, Nakajima T, Motoori K, et al. Comparison of endobronchial ultrasound, positron emission tomography, and CT for lymph node staging of lung cancer. Chest 2006;130:710 8. 9. Little AG, Rusch VW, Bonner JA, et al. Patterns of surgical care of lung cancer patients. Ann Thorac Surg 2005;80: 2051 6. 10. De Leyn P, Lardinois D, Van Schil PE, et al. ESTS guidelines for preoperative lymph node staging for non-small cell lung cancer. Eur J Cardiothorac Surg 2007;32:1 8. 11. Mountain CF, Dresler CM. Regional lymph node classification for lung cancer staging. Chest 1997;111:1718 23. 12. Detterbeck FC, Jones DR, Parker LA Jr. Intrathoracic staging. In: Detterbeck FC, Rivera MP, Socinski MA, Roseman JG, eds. Diagnosis and Treatment of Lung Cancer. An Evidence- Based Guide for the Practicing Clinician. Philadelphia: WB Saunders Company; 2001:73 93. 13. Detterbeck FC, Jantz MA, Wallace M, Vansteenkiste J, Silvestri GA. Invasive mediastinal staging of lung cancer: ACCP evidence-based clinical practice guidelines (2nd edition). Chest 2007;132(Suppl):202S 20S. DISCUSSION DR BRYAN FITCH MEYERS (St. Louis, MO): I think a lot has been made of the paper by Alex Little (Little AG, Rusch VW, Bonner JA, et al. Patterns of surgical care of lung cancer patients. Ann Thorac Surg 2005;80:2051 6) a couple of years ago about the poor job in general that we do in mediastinal staging of lung cancer patients. Work like this paper represents a notable exception to that notion that we do not do enough to accurately stage the mediastinum prior to resection. I just want to commend Dr Block. He is in solo practice as a private practitioner in Florida and has now had two presentations in the last 12 months on his results with EBUS (endobronchial ultrasound). He has taken the time in a busy solo private practice to study his results and share them with us. Mark, I wonder if you could just give us an idea of what steps you took to introduce EBUS into your practice and if you have a sense of what impact your careful attention to mediastinal

Ann Thorac Surg BLOCK 2010;89:1582 7 HOW MANY STATIONS TO SAMPLE WITH EBUS? 1587 staging with EBUS has had on your practice. Has there been any measurable change in your referrals as a result of your work with EBUS and the effort that you ve put in presenting it? DR BLOCK: Thanks very much for your kind comments, Bryan. They re very much appreciated. The second first. I have seen a change. Clearly, the community is interested in nonmediastinoscopy ways to stage the mediastinum, and so I do get referrals simply for endoscopic ultrasound. I also get referrals for sampling hilar lymph nodes, which you just can t reach through any other technique. So that s probably been the biggest change in the practice through introduction of endoscopic ultrasound. The way it came about was that in a community setting, many of the specialists who would normally be doing this in an academic setting are not committed to these procedures as an integral part of their practice. The gastroenterologists had no interest in lung cancer staging and so were not committed to esophageal ultrasound for that purpose. I approached the hospital to get the equipment, and after several rounds of discussions they agreed. Once the equipment was available, I was the only one interested in using it for mediastinal sampling. Once I demonstrated success with EUS (transesophageal endoscopic ultrasound) for lung cancer staging, it was much easier to work with the hospital to purchase an ultrasound bronchoscope. Although 1 or 2 pulmonologists have expressed interest in EBUS since the equipment arrived 2 years ago, no one has really taken the time to do it. It has not been a real interest of theirs primarily because of the time commitment involved and the need to focus on lung cancer potentially at the expense of other areas of their practice. So none of them has pursued training. And then a third factor is that referral patterns in private practice tend to favor referral to a surgeon for mediastinal staging. So I think a community setting is probably a very good way to go about establishing this as a part of your practice. And it s been, I think, a pretty successful component of the practice. I rarely do mediastinoscopy for routine non small cell lung cancer staging. DR DAVID C. RICE (Houston, TX): Mark, again, congratulations for a great presentation. All too often I think we re seeing patients who are being staged by EBUS practitioners who may not necessarily be all that familiar with lung cancer staging, and often they will biopsy only the node that s hot on PET (positron emission tomography). So I really commend you for biopsying all stations so that the true nodal stage can be determined. My question is, of the 6 N3 patients, how many of those had an abnormal CT (computed tomography) or PET N3 node preoperatively that perhaps would not have been biopsied by somebody less familiar with lung cancer staging than we thoracic surgeons? DR BLOCK: I don t have the data on the tip of my finger. If my memory serves, it s probably about half of them. Two or three of them probably had a suspicious node by clinical staging, but I did not do that analysis. DR LYNNE A. SKARYAK (Baltimore, MD): Did you look when you did your VATS (video-assisted thoracoscopic surgery) lobectomy or your lobectomies, did you see how many of the negatives were positive by pathology, by doing mediastinal lymph node dissection? DR BLOCK: Yes. Of those 65 patients that had a negative mediastinum with staging, 45 of them went on to surgery. And of those 45, 6 had a positive mediastinum at surgery, occult disease. If you look at the profile of those 6 patients, they had a very similar profile in terms of the number of stations that were sampled by endoscopic ultrasound. DR RAFAEL S. ANDRADE (Minneapolis, MN): I do commend you on doing this because this is one of the few papers that emphasizes the number of stations that should be sampled so we can indeed replicate what we do with mediastinoscopy. Now, I see that you do it in the endoscopy suite so the patient has to first have this procedure done and then be booked for the following procedure. Have you considered taking this a step further and do EBUS intraoperatively like we do mediastinoscopy, and then if indicated go on to lobectomy to help streamline patient care? DR BLOCK: I have not, and there are a couple of reasons for that. The first is that I don t have rapid on-site evaluation, so I don t have a reliable way to accurately assess the results immediately. The second is that it turns out to be a fairly inefficient use of OR (operating room) time if you re doing a lot of positive biopsies and then canceling the subsequent VATS or thoracotomy. And then the third is that I like to have a couple of days to ensure that the pathology report is accurate. We saw yesterday a presentation that showed that rapid on-site evaluation was sometimes a false negative and that the subsequent pathology report came back positive. That s the third consideration. I had a fourth one, but I can t remember it. It has slipped my mind now.