A Standard Endobronchial Ultrasound Image Classification System

Transcription

1 CHEST Original Research INTERVENTIONAL PULMONOLOGY The Utility of Sonographic Features During Endobronchial Ultrasound-Guided Transbronchial Needle Aspiration for Lymph Node Staging in Patients With Lung Cancer A Standard Endobronchial Ultrasound Image Classification System Taiki Fujiwara, MD ; Kazuhiro Yasufuku, MD, PhD, FCCP ; Takahiro Nakajima, MD, PhD ; Masako Chiyo, MD, PhD ; Shigetoshi Yoshida, MD, PhD ; Makoto Suzuki, MD, PhD ; Kiyoshi Shibuya, MD, PhD ; Kenzo Hiroshima, MD, PhD ; Yukio Nakatani, MD, PhD ; and Ichiro Yoshino, MD, PhD Background: Endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) is a minimally invasive procedure with a high yield for lymph node staging of lung cancer. The aim of this study was to assess the utility of sonographic features of lymph nodes during EBUS- TBNA for the prediction of metastasis in patients with lung cancer and to establish a standard endobronchial ultrasound (EBUS) image classification system. Methods: Digital images of lymph nodes obtained during EBUS-TBNA in patients with lung cancer were categorized according to the following characteristics: (1) size (short axis) less or more than 1 cm, (2) shape (oval or round), (3) margin (indistinct or distinct), (4) echogenicity (homogeneous or heterogeneous), (5) presence or absence of central hilar structure, and (6) presence or absence of coagulation necrosis sign. The sonographic findings were compared with the final pathologic results. Results: A total of 1,061 lymph nodes were retrospectively evaluated in 487 patients. The accuracy of predicting metastatic property for each category was as high as 63.8% to 86.0%. A multivariate analysis revealed that round shape, distinct margin, heterogeneous echogenicity, and presence of coagulation necrosis sign were independent predictive factors for metastasis. Two hundred eighty-five of the 664 lymph nodes (42.9%) having at least one metastatic feature of the four categories were pathologically proven metastatic, and 96.0% of lymph nodes (381/397) were proven not metastatic when all four categories were determined as benign. Conclusions: Sonographic features of lymph nodes based on the new EBUS imaging classification may be helpful in the prediction of metastatic lymph nodes during EBUS-TBNA. CHEST 2010; 138(3): Abbreviations : CHS 5 central hilar structure; CP-EBUS 5 convex probe-endobronchial ultrasound CP-EBUS; EBUS 5 endobronchial ultrasound; EBUS-TBNA 5 endobronchial ultrasound-guided transbronchial needle aspiration; EUS 5 endoscopic ultrasound; TBNA 5 transbronchial needle aspiration Lung cancer is the most common cause of cancerrelated death in the Western world. 1 The outcome of the disease depends on staging and therefore proper staging must be performed to determine the treatment plan. 2 Endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) is a minimally invasive procedure performed under local anesthesia that has been shown to have a high sensitivity and diagnostic yield for lymph node staging of lung cancer. 2-7 EBUS-TBNA allows cytologic and histologic examination of mediastinal and hilar lymph nodes. 8 It is performed with the convex probe-endobronchial ultrasound (CP-EBUS), which has a 7.5-MHz ultrasound probe on the tip of a bronchovideoscope. The CP-EBUS is capable of detecting lymph nodes as small as 2 mm, which are sometimes not visible on CHEST / 138 / 3 / SEPTEMBER,

2 CT imaging. Lymph nodes as small as 3 mm in the short axis can be punctured under real-time ultrasound guidance. Well-trained bronchoscopists encounter mediastinal and hilar lymph nodes of various sizes and features. It has been reported that sonographic features are useful imaging tools in the evaluation of cervical lymph node metastasis in head and neck cancers, breast cancers, and thoracic malignancies. 9,10 Furthermore, sonographic features during endoscopic ultrasound (EUS) have been shown to be useful for the prediction of malignant lymph nodes in the mediastinum and the hilum. 11,12 To date, there are no reports on the classification of sonographic features during EBUS-TBNA and the usefulness of these features in the prediction of metastasis in mediastinal and hilar lymph nodes. The aim of this study was to assess the utility of the morphologic features of lymph nodes obtained by endobronchial ultrasound (EBUS) for the prediction of presence or absence of metastasis in mediastinal and/or hilar lymph nodes in patients with lung cancer. Patients Materials and Methods A retrospective chart review was performed in patients who underwent EBUS-TBNA for mediastinal staging of lung cancer at the Department of Thoracic Surgery, Chiba University Hospital, from January 2003 to August EBUS-TBNA was performed in patients with lung cancer or suspected lung cancer without previous treatment, with radiologically defined mediastinal and/or hilar lymph nodes with a short axis of 5 mm on enhanced CT imaging or positive on PET scan (defined as standardized uptake value. 2.5). Chest CT imaging was performed with a single-injection contrast on a multidetector-row CT scan. Lymph node stations and numbers were determined according to the international TNM staging system reported by Mountain and Dresler.13 EBUS-TBNA EBUS-TBNA was performed on an outpatient basis in patients under conscious sedation (midazolam) with local anesthesia. The Manuscript received August 24, 2009; revision accepted March 20, Affiliations: From the Department of Thoracic Surgery (Drs Fujiwara, Yasufuku, Nakajima, Chiyo, Yoshida, Suzuki, Shibuya, and Yoshino) and Diagnostic Pathology (Drs Hiroshima and Nakatani), Graduate School of Medicine, Chiba University, Chiba, Japan; and the Division of Thoracic Surgery (Dr Yasufuku), Toronto General Hospital, University Health Network, Toronto, ON, Canada. Corresponding author : Kazuhiro Yasufuku, MD, PhD, Division of Thoracic Surgery, Toronto General Hospital, University Health Network, 200 Elizabeth S, 9N-957, Toronto, ON, M5G2C4, Canada; kazuhiro.yasufuku@uhn.on.ca. Reproduction of this article is prohibited without written permission from the American College of Chest Physicians ( site/misc/reprints.xhtml ). DOI: /chest CP-EBUS (BF-UC260F-OL8; Olympus; Tokyo, Japan) was used for the examination of the mediastinal and hilar lymph nodes. The CP-EBUS is integrated with a convex transducer (7.5 MHz) that scans parallel to the insertion direction of the bronchoscope. Images can be obtained by contacting the probe directly or by attaching a balloon to the tip. The ultrasound features are processed in a dedicated ultrasound scanner (EU-C2000/EU-C60; Olympus). A dedicated 22-gauge needle was used to perform transbronchial needle aspiration (TBNA) (NA-201SX-4022; Olympus) for lymph node sampling. The needle is equipped with an internal stylet that is withdrawn after passing through the bronchial wall and can be visualized through the optic device and on the ultrasound features. After the initial puncture, the internal stylet is used to clean out the internal lumen, which becomes clogged with bronchial membrane. The internal stylet is then removed and negative pressure is applied with a syringe. After the needle is moved back and forth inside the lymph node, the needle is retrieved and the internal sheath is used once again to push out the histologic core. 3-5 With this method, histologic cores, as well as cytologic specimens, can be obtained. The aspirated material was smeared onto glass slides, air dried, and immediately stained with Diff-Quik (Sysmex Corporation; Kobe, Japan) for immediate interpretation by an on-site cytopathologist to confirm adequate cell material. Furthermore, Papanicolaou and Giemsa staining and light microscopy were performed by an independent cytopathologist who was blinded to the details of the EBUS image characteristics. Histologic cores were fixed with formalin and stained with hematoxylin and eosin. Immunohistochemistry was also performed in some patients. In patients with malignant lymph nodes, the determination was based on malignant cytologic and/or histologic results at EBUS- TBNA or surgical-pathologic confirmation. In patients with benign lymph nodes, this determination was based on surgical-pathologic confirmation of EBUS-TBNA-targeted nodes by lymph node dissection of the lymph node station of interest, or on results of clinical follow-up for at least 6 months demonstrating a lack of clinical or radiologic disease progression. We determined the lymph node location based on the standard lymph node map reported in 1997 by Mountain and Dresler. 13 The lymph nodes were systematically visualized, starting with N1 lymph nodes, followed by N2 nodes, and finally N3 nodes. EBUS-TBNA was then performed first from N3 nodes, followed by N2 nodes, and, if necessary, N1 nodes. If N3 nodes were found to be positive for malignancy on rapid on-site cytologic evaluation, we terminated the procedure. All EBUS-TBNA were performed by three individuals (T. F., T. N., K. Y.) or under their supervision. EBUS Image Characteristics of Lymph Nodes JPEG images and digital video images of all the lymph nodes obtained by CP-EBUS were reviewed by three different individuals (T. F., T. N., K. Y.) blinded to the results of EBUS-TBNA. We avoided the use of special imaging software, which may have affected the evaluation of image characteristics. The lymph nodes were characterized based on EBUS images as follows ( Fig 1 ): (1) short-axis size less or more than 1 cm, (2) shape (oval or round), (3) margin (indistinct or distinct), (4) echogenicity (homogeneous or heterogeneous), (5) presence or absence of central hilar structure (CHS), and (6) presence or absence of coagulation necrosis sign. We measured both the long and the short axis of all lymph nodes. When the ratio of the short to long axis of lymph nodes was, 1.5, we defined the lymph nodes as round. On the other hand, if the ratio was. 1.5, we defined it as oval. The short and long axes were measured as a distance of two perpendicular directions for triangular-shaped lymph nodes. When we observed lymph nodes by CP-EBUS, there existed echogenic differences between the lymph node and the surrounding connective tissue structure. 642 Original Research

3 Figure 1. Different endobronchial ultrasound (EBUS) image characteristics used for the classification of lymph nodes. Size; less or more than 1 cm; shape: oval or round; margin: indistinct or distinct; echogenicity: homogeneous or heterogeneous; central hilar structure (CHS): present or absent; coagulation necrosis sign: present or absent. If the majority of the margin (. 50%) was clearly visualized with a high echoic border, we determined the lymph nodes to be distinct, and if the margin was unclear, we determined them to be indistinct. CHS seen in cervical lymph nodes 9 is defined as a linear, flat, hyperechoic area in the center of the lymph node. The coagulation necrosis sign is a hypoechoic area within the lymph node without blood flow. It is also seen in cervical lymph node features and in mediastinal lymph node features during EUS. 9,14 We defined the intralymphatic lesion, which showed low echoic area with absence of blood flow on Doppler. This sign often correlates with an existence of necrosis within the lymph node. Typical coagulation necrosis signs are found as one low echoic area within the lymph node and they sometimes occupy the majority of the lymph node. Multiple low echoic spots within the lymph node were categorized in some cases as heterogeneous echo features instead of coagulation necrosis signs. The final characteristics of the sonographic findings for each lymph node were based on an agreement of at least two reviewers. Data Analysis All six sonographic characteristics of the lymph nodes were compared with the final pathologic result of the lymph node. The sensitivity, specificity, positive predictive value, negative predictive value, and diagnostic accuracy rate were calculated by standard definitions. We used StatView for Windows, version 5 (SAS Institute; Patient Characteristics Table 1 Patient Characteristics No. Patients 487 Male (female) sex 372 (115) Age, y, mean (range) 68.0 (28-87) Histology Adenocarcinoma 244 Squamous cell carcinoma 144 Large cell carcinoma 19 Small cell carcinoma 32 Other 48 Cary, NC) for statistics analysis. This study was a retrospective chart review and therefore the institutional review board of Chiba University approved the study without the need to obtain informed consent. Patients Results The characteristics of the 487 patients who were enrolled and evaluated in this study are summarized in Table 1. There were 372 men and 115 women, and the average age was 68.0 years. The histologic types of lung cancers were adenocarcinoma in 244, squamous cell carcinoma in 144, small cell carcinoma in 32, large cell carcinoma in 19, and other cancers in 48. A total of 1,061 lymph nodes were analyzed, and the proportion of each station is summarized in Table 2. Radiologically suspected nodes were frequently observed in the station 4R and station 7 lymph nodes (623 in number and 59% in proportion) Table 2 Lymph Node Stations Included in the Study Lymph Node Station No. All 1,061 2R 41 2L 19 4R 335 4L L 75 11R CHEST / 138 / 3 / SEPTEMBER,

4 and were subjected to the analysis. No patients had any complications related to EBUS-TBNA. Morphologic Findings of EBUS Representative morphologic findings of EBUS are shown in Figure 2. The size of evaluated lymph nodes ranged from 2 mm to 33.4 mm, and the number of nodes with a short axis, 1 cm and that of nodes with a short axis of. 1 cm were 642 (60.5%) and 419 (39.5%), respectively. For shape, 612 (57.7%) lymph nodes were characterized as oval and 449 (42.3%) as round. For margin, 430 (40.5%) nodes exhibited indistinct margins and 631 (59.5%) had distinct margins. For echogenicity, 726 (68.4%) nodes were characterized as homogeneous and 335 (31.6%) nodes as heterogeneous. CHS was observed in 334 (31.5%) nodes and the presence of the coagulation necrosis sign in 265 (25.0%) nodes. The presence or absence of metastasis based on each feature is shown in Figure 3. When lymph nodes had the following features: short axis of more than 1 cm, round shape, distinct margin, heterogeneous echogenicity, absence of CHS, or presence of coagulation necrosis sign, they tended to suggest metastatic lymph nodes. Diagnostic yields for each feature are summarized in Table 3. The accuracy of predicting metastatic property was 76.4% for size, 79.3% for shape, 65.7% for margin, 89.9% for echogenicity, 63.8% for CHS, and 86.0% for coagulation necrosis sign. In all the morphologic categories, negative predictive values were higher (88.4% to 96.0%) in comparison with corresponding positive predictive values (43.3% to 78.9%). Logistic regression analysis revealed that shape, margin, echogenicity, and coagulation necrosis sign were independent predictive factors, with respective hazard ratios of 3.1, 3.1, 2.0, and 5.6 ( Table 4 ). Two hundred eighty-five of the 664 lymph nodes (42.9%) having a metastatic feature in at least one of the four categories were pathologically proven metastatic, and 96.0% of lymph nodes (381/397) were pathologically proven not metastatic when all four categories were determined as not metastatic. Discussion Since our first successful EBUS-TBNA in 2002, we have performed. 1,000 EBUS-TBNA procedures without any complications at our department. For mediastinal staging in lung cancer, all the mediastinal lymph nodes and hilar lymph nodes accessible by CP-EBUS are visualized in a systematic way and evaluated. As reported previously, CP-EBUS is capable of detecting even small-sized lymph nodes not visible on CT imaging and lymph nodes as small as 3 mm in the short axis can be punctured under real-time Figure 2. Representative morphologic findings of EBUS characteristics. Size 1 cm ( A ) or more than 1 cm ( B ); shape: oval ( C ) or round ( D ); margin: indistinct ( E ) or distinct ( F ); echogenicity: homogeneous ( G ) or heterogeneous ( H ); central hilar structure: present or absent ( I ); coagulation necrosis sign: present or absent ( J ). See Figure 1 legend for expansion of the abbreviation. ultrasound guidance. 2-7 Although we use preoperative imaging by CT scan and/or PET scan as a reference prior to EBUS-TBNA, we do inspect all lymph nodes based on the site of the primary tumor and on the lobar lymphatic drainage pattern. There are often important lymph nodes that need to be sampled even when preoperative imaging by CT scan and/or PET scan is negative for malignancy. During the process, we have realized that there are some EBUS sonographic features that are suggestive of benign lymph nodes. The aim of the current study was to classify the different sonographic features of mediastinal 644 Original Research

5 Figure 3. The actual results of comparison between EBUS image classification and final pathology. When lymph nodes had small sizes, round shapes, indistinct margins, homogeneous echogenicities, and the presence of CHS, they tended to be benign. On the other hand, when lymph nodes had the presence of CNS, they tended to be malignant. CHS 5 central hilar structure; CNS 5 coagulation necrosis sign. See Figure 1 for expansion of other abbreviation. lymph nodes and to apply this EBUS image classification to the prediction of lymph node metastasis. Our results show that sonographic EBUS features of lymph nodes during EBUS-TBNA are helpful for the prediction of benign lymph nodes in lung cancer patients. Ahuja and Ying 9 reported that sonography is a useful imaging tool in the evaluation of cervical lymph nodes in patients with malignancies of the head, neck, and thorax. As for the evaluation of mediastinal and hilar lymph nodes, EUS was first accepted as a diagnostic modality in the 1990s. 11,12 In both cervical ultrasonography and EUS, the morphologic characters of lymph nodes are analyzed, and size. 10 mm, round shape, distinct margin, heterogeneous echogenicity, absence of central echogenic hilum, and coagulation necrosis are considered signs of lymph node metastasis. 9,11,12,14,15 From our experience, we defined six different morphologic characteristics of mediastinal and hilar lymph nodes during EBUS-TBNA in patients with lung cancer, similar to the features examined in cervix ultrasonography and EUS. Calcification was not included as a significant echo feature in these reports and because the incidence of calcification was rare in our initial experience with EBUS, we did not include it as a criterion. More than 1,000 lymph nodes were investigated morphologically and pathologically by three different individuals (T. F., T. N., K. Y.). When multiple lymph nodes were detected in the same lymph node station, we sampled the largest lymph node by EBUS-TBNA, whereas the pathology of the smaller lymph nodes within the same lymph node station was confirmed by surgery. In this study, we demonstrated the diagnostic accuracies for metastatic property of the four independent predictive characters of EBUS, including shape, margin, echogenicity, and the absence of central necrosis sign. The diagnostic accuracy of predicting metastatic nodes from our new classification system was very high. If we can predict metastatic lymph nodes from EBUS morphology, should we consider just looking Table 3 Diagnostic Yield of Each Endobronchial Ultrasound Image Category for Metastatic Lymph Node Morphologic Category Sensitivity Specificity Positive Predictive Value Negative Predictive Value Diagnosis Accuracy Size:. 10 mm Shape: round Margin: distinct Echogenicity: heterogeneous Central hilar structure: absence Coagulation necrosis sign: presence CHEST / 138 / 3 / SEPTEMBER,

6 Table 4 Logistic Regression Analysis of Endobronchial Ultrasound Image Categories for Prediction of Metastatic Lymph Nodes Morphologic Category Hazard Ratio 95% CI P Value Size (. 10 mm/, 10 mm) Shape (round/oval) ,.0001 Margin (distinct/indistinct) Echogenicity (heterogeneous/homogeneous) Central hilar structure (absence/presence) Coagulation necrosis sign (presence/absence) ,.0001 at the morphology without doing a biopsy if we find such lymph nodes? The answer is no. Studies comparing EUS morphology to EUS-fine-needle aspiration have shown that EUS-fine-needle aspiration is superior to imaging by EUS alone We should always try to get tissue diagnosis from suspicious lymph nodes; thus, EBUS-TBNA should always be performed on lymph nodes that are suspicious for metastasis on EBUS images. On the other hand, one of the potential benefits of this study is the negative predictive value. If an operator performs EBUS-TBNA on a lymph node in a patient with suspected lung cancer and the cytologic specimen reveals only benign lymphocytes, then this is the instance in which the lack of lymph node EBUS malignant features can be more reassuring in confirming the true negativity of that lymph node. This would also be useful during the EBUS-TBNA procedure to avoid unnecessary passes where rapid on-site cytology reveals only benign lymphocytes in lymph nodes with benign ultrasound features. The purpose of this study was to highlight the efficiency and highly precise examination of EBUS-TBNA with the use of echo features. Further prospective studies are recommended to confirm the utility of sonographic features during EBUS-TBNA. The limitation of our study is that we did not include the sonographic features of noncancerous adenopathy. Mediastinal lymphadenopathy can also be observed in patients with noncancerous disease such as sarcoidosis and TB, and autoimmune diseases such as Sjogren syndrome and systemic lupus erythmatosus.9,19 The echo features are quite different from the metastatic findings detected in the present study. There is definitely a role for evaluating the lymph nodes of noncancerous lymphadenopathy. However, we chose not to include this patient population because the findings may have confused the interpretation of lymph nodes in patients with lung cancer. Conclusions In conclusion, by careful examination of the sonographic features of lymph nodes during EBUS-TBNA, we may speculate on the presence of metastatic lymph nodes even in patients with lung cancer. The four sonographically obtained morphologic features, including round shape, distinct margin, heterogeneous echogenicity, and presence of coagulation necrosis sign, are independent predictive factors for nodal metastasis. When at least one such feature is observed during the EBUS procedure, subsequent needle aspiration must be performed. Conversely, when all four categories are sonographically determined not metastatic, we may be able to avoid unnecessary biopsies in such lymph nodes. Our findings and the EBUS image classification system will need to be validated in a prospective study before we can make clinical decisions based on imaging alone during the procedure. Acknowledgments Author contributions: Dr Fujiwara: contributed to the evaluation of the EBUS images, performance of the EBUS-TBNA, and evaluation of the lymph nodes by surgery. Dr Yasufuku: contributed to the evaluation of the EBUS images, performance of the EBUS-TBNA, and evaluation of the lymph nodes by surgery. Dr Nakajima: contributed to the evaluation of the EBUS images, performance of the EBUS-TBNA, and evaluation of the lymph nodes by surgery. Dr Chiyo: contributed to the performance of the EBUS-TBNA and evaluation of the lymph nodes by surgery. Dr Yoshida: contributed to the evaluation of the lymph nodes by surgery. Dr Suzuki: contributed to the evaluation of the lymph nodes by surgery. Dr Shibuya: contributed to the evaluation of the lymph nodes by surgery. Dr Hiroshima: contributed to the pathologic evaluation of the lymph nodes. Dr Nakatani: contributed to the pathologic evaluation of the lymph nodes. Dr Yoshino: contributed to the evaluation of the lymph nodes by surgery. Financial/nonfinancial disclosures: The authors have reported to CHEST the following conflicts of interest: Dr Yasufuku has received unrestricted grants from Olympus Medical Systems for continuing medical education. Drs Fujiwara, Nakajima, Chiyo, Yoshida, Suzuki, Shibuya, Hiroshima, Nakatani, and Yoshino have reported that no potential conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article. References 1. Jemal A, Siegel R, Ward E, et al. Cancer statistics, CA Cancer J Clin ;53: Original Research

7 2. Yasufuku K, Fujisawa T. Staging and diagnosis of non-small cell lung cancer: invasive modalities. Respirology ;12 (2 ): Yasufuku K, Chiyo M, Sekine Y, et al. Real-time endobronchial ultrasound-guided transbronchial needle aspiration of mediastinal and hilar lymph nodes. Chest ;126 (1 ): Yasufuku K, Chiyo M, Koh E, et al. Endobronchial ultrasound guided transbronchial needle aspiration for staging of lung cancer. Lung Cancer ;50 (3 ): 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 ; 130 (3 ): 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 ;28 (5 ): Vincent BD, El-Bayoumi E, Hoffman B, et al. Real-time endobronchial ultrasound-guided transbronchial lymph node aspiration. Ann Thorac Surg ;85 (1 ): Nakajima T, Yasufuku K, Iyoda A, et al. The evaluation of lymph node metastasis by endobronchial ultrasound-guided transbronchial needle aspiration: crucial for selection of surgical candidates with metastatic lung tumors. J Thorac Cardiovasc Surg ;134 (6 ): Ahuja AT, Ying M. Sonographic evaluation of cervical lymph nodes. AJR Am J Roentgenol ;184 (5 ): Kim TH, Kang DK, Kim SY, Lee EJ, Jung YS, Yim H. Sonographic differentiation of benign and malignant papillary lesions of the breast. J Ultrasound Med ;27 (1 ): Lee N, Inoue K, Yamamoto R, Kinoshita H. Patterns of internal echoes in lymph nodes in the diagnosis of lung cancer metastasis. World J Surg ;16 (5 ): 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 ;45 (6 ): Mountain CF, Dresler CM. Regional lymph node classification for lung cancer staging. Chest ;111 (6 ): Roberts SA, Mahon BS, Evans R. Coagulation necrosis in malignant mediastinal nodes on endoscopic ultrasound: a new endosonographic sign. Clin Radiol ;60 (5 ): Catalano MF, Alcocer E, Chak A, et al. Evaluation of metastatic celiac axis lymph nodes in patients with esophageal carcinoma: accuracy of EUS. Gastrointest Endosc ;50 (3 ): Chen VK, Eloubeidi MA. Endoscopic ultrasound-guided fine needle aspiration is superior to lymph node echofeatures: a prospective evaluation of mediastinal and peri-intestinal lymphadenopathy. Am J Gastroenterol ;99 (4 ): Schmulewitz N, Wildi SM, Varadarajulu S, et al. Accuracy of EUS criteria and primary tumor site for identification of mediastinal lymph node metastasis from non-small-cell lung cancer. Gastrointest Endosc ;59 (2 ): Kramer H, Sanders J, Post WJ, Groen HJ, Suurmeijer AJ. Analysis of cytological specimens from mediastinal lesions obtained by endoscopic ultrasound-guided fine-needle aspiration. Cancer ;108 (4 ): Ying M, Ahuja AT, Yuen HY. Grey-scale and power Doppler sonography of unusual cervical lymphadenopathy. Ultrasound Med Biol ;30 (4 ): CHEST / 138 / 3 / SEPTEMBER,