Results of endobronchial ultrasound-guided transbronchial needle aspiration in lung cancer: Importance of the lymph node involvement prevalence.

Objective. - We studied the non-surgical invasive staging by endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) and we detailed the differences of our series, in order to understand the criteria allowing to achieve a better performance. Methods. - Retrospective observational study conducted between 2007 and 2011, including all patients with proven NSCLC who underwent EBUS-TBNA. Results. - For the 92 EBUS-TBNA performed, we found a sensitivity of 78%, a specificity of 93%, a positive predictive value (PPV) of 98%, a negative predictive value (NPV) of 45%, an accuracy of 80% and a prevalence of lymph node involvement at 84%. A learning curve has been demonstrated and a significant difference was found based on the number of punctures by procedure (P=0.02) or on histological type (P=0.02). By analyzing the data of the literature, we have been able to demonstrate that the accuracy and the negative predictive value are correlated with the prevalence. If we take into account this correlation, we can consider the results of our study close to those of the literature. Conclusion. - We highlighted a number of criteria that will influence the diagnostic yield of EBUS-TBNA. While some have already been described, other criteria such as histological type or patient selection criteria are less discussed. The key point is the correlation between the prevalence and EBUS-TBNA results. Results of the assessment of lymph node involvement techniques should be interpreted according to the prevalence of lymph node involvement. (C) 2015 Elsevier Masson SAS. All rights reserved.