Chest 2003;123:458–62. Herth F, Ernst A, Schulz M, Becker H. Comments: To accurately stage centrally located lung cancer, it is important to determine whether the tumor is only compressing the airway or whether it is actually infiltrating the bronchial or tracheal wall. The primary objective of this article was to study the use of endobronchial ultrasound (EBUS) for this purpose. The authors enrolled 131 consecutive patients with thoracic tumors located close to the large airways. All patients had a chest computed tomography (CT) scan before bronchoscopy. The bronchoscopists could review the hard copies of the CT scan, but the official radiologic interpretation was not available to them before the bronchoscopy. After airway examination, EBUS was performed using a flexible 20-MHz probe passed through the working channel of the bronchoscope. The tumor was classified as infiltrating on the CT scan if no plane could be identified between the mass and the airway. Evidence of invasion of at least the outermost layer of the bronchial wall was needed on EBUS examination to classify the tumor as infiltrating. After the initial workup, 105 of 131 patients underwent surgical resection or invasive staging procedure. The accuracy of CT and EBUS results were then compared using histology as the gold standard. Endobronchial ultrasound had an accuracy of 94%, sensitivity of 89%, and specificity of 100% to identify infiltration by central tumor. Chest CT had an accuracy of 51%, sensitivity of 75%, and specificity of 28%. No patient classified as having tumor invasion by EBUS had false-positive results. On the contrary, the airway walls were free of tumors on histologic examination in 40 of 81 patients who were suspected to have infiltration on chest CT. The EBUS examination added an average of 3.5 minutes to the bronchoscopy time and there were no procedure-related complications. This well-conducted study clearly establishes the superiority of EBUS over chest CT in differentiating between the airway infiltration and extrinsic compression. This information is critical for treatment planning and for selecting patients suitable for surgical treatment of lung cancer. The results of this study indicate that chest CT is not adequate for this purpose. Therefore, the decision against surgery cannot be made on the basis of CT findings alone because it can lead to inappropriate denial of surgical cure to a substantial proportion of patients. On the contrary, no patient in this study would have been inappropriately denied surgery on the basis of EBUS results. It is important to note that EBUS in this study was performed and interpreted by operators highly experienced in this technique. Indeed, the authors of this study had a combined experience performing >5000 EBUS examinations. Every new technique has a learning curve, and it is especially true for ultrasound technology that is highly operator-dependent. Therefore, impressive results noted in this study might not be reproducible elsewhere without adequate training and experience. Nevertheless, the results of this study are very promising and if high specificity is reproduced in other studies, application of EBUS for preoperative staging has the potential to reduce the need for surgical staging and unnecessary thoracotomy for patients with lung cancer located close to large airways. Future studies should directly address this issue. Also, the role of EBUS in determining the extent of surgical resection in central lung cancer needs further clarification.