Use and prognostic value of staging mediastinoscopy in non-small-cell lung cancer.

One hundred sixty patients had preoperative mediastinoscopy, resection of the primary tumor, and complete mediastinal lymphadenectomy for non-small-cell carcinoma of the lung. Minimum follow-up was 24 months (mean 40 months). Postoperative staging based on histologic examination of the specimen of the lung and mediastinal lymphadenectomy categorized 59 patients in stage I, 28 in stage II, and 73 in stage III (20 T3N0, 12 T3N1, 29 T1 or T2N2, and 12 T3N2). The sensitivity rate of cervical mediastinoscopy for detection of mediastinal node metastasis was 48.7%. False-negative results of mediastinoscopy occurred in 21 of 41 patients with normal mediastinoscopy: unreachable nodes in eight patients, sampling error of reachable nodes in 11 patients, and error on frozen section in two patients. Eleven of 65 patients with clinical stage I disease and normal mediastinum on chest roentgenogram had mediastinal node involvement; only three were detected by mediastinoscopy, which resulted in a low sensitivity rate (27.3%) and a high rate of unnecessary mediastinoscopy (62/65 patients). The sensitivity of mediastinoscopy increased as the amount of disease present, as measured by the clinical stage of disease or positive gallium 67 scan of mediastinum, increased. Eleven of 29 patients with T1 to T2N2 disease discovered at mediastinoscopy had similar survival rates compared with 18 of 29 patients who had a normal mediastinoscopy examination and mediastinal node involvement discovered at thoracotomy.