Representativeness of nodal sampling with endobronchial ultrasonography in non-small-cell lung cancer staging.

The objective of our study was to determine the procedure-related requirements of mediastinal node sampling with endobronchial ultrasonography with real-time transbronchial needle aspiration (EBUS-TBNA) that would provide negative predictive value (NPV) for the identification of stage III disease in non-small-cell lung cancer (NSCLC) high enough to consider the technique equivalent to cervical mediastinoscopy. Representative EBUS-TBNA was defined as a sampling procedure obtaining satisfactory samples from normal nodes in regions 4R, 4L and 7 or diagnosing malignancy in mediastinal nodes. NPV was estimated using the results of postsurgical staging in patients who underwent surgery as a reference. Two-hundred ninety-six patients staged with EBUS-TBNA were included. Representative samples from regions 4R, 4L and 7 showing nonmalignant cytology were obtained from 98 patients (33.1%) and EBUS-TBNA detected N2/N3 disease in 150 (50.7%). Accordingly, an EBUS-TBNA procedure accomplishing the representativeness criteria required for sampling was attained in 248 of the participating patients (83.8%). The NPV of the procedure in this setting was 93.6%, with false-negative results only found in 5 patients, four of them with nodal metastasis out of the reach of EBUS-TBNA (regions 5, 8 and 9). In conclusion, representative sampling of regions 4R, 4L and 7 is achieved in more than 80% of patients staged using EBUS-TBNA, and in the procedures that attain this requirement a NPV >90% for mediastinal malignancy is reached, a figure equivalent to cervical mediastinoscopy.