[Mediastinoscopy in the diagnosis of mediastinal disease. An analysis of 181 explorations].

To validate our experience with standard cervical mediastinoscopy (SCM) and extended cervical mediastinoscopy (ECM) to diagnose mediastinal nodes and masses, we studied 181 patients between January 1992 and February 2001. SCM and ECM were indicated for diagnostic staging of nodes related to bronchogenic carcinoma (Group I) or of mediastinal masses (Group II). An SCM was performed in all cases to explore the paratracheal region (2R, 2L, 4R, 4L, 7, 10R and 10L); in 21 additional cases, an ECM was performed to explore the aortopulmonary window or the subaortic region (area 5) and the para-aortic region (area 6). In Group I, the sensitivity of SCM was 93.6% and specificity was 100%; the positive predictive value (PPV) was 100%, the negative predictive value (NPV) was 82.8%, and the diagnostic yield was 95.1%. The sensitivity of ECM was 91% and specificity was 100%; PPV was 100%, NPV 93.3% and yield was 96%. In Group II, the sensitivity was 93.3%, specificity 100%, PPV 100%, NPV 81.2% and diagnostic yield 94.8%. The sensitivity of ECM in this group was 80%, specificity was 100%, PPV 100%, NPV 66.7% and yield 85.7%. A 2.7% complication rate was observed, with one case of bleeding after injury to the superior vena cava, one tracheal lesion, one recurring paralysis and two cases of surgical wound infection. The mean postoperative stay was 36 hours and mortality was zero. We conclude that SCM is highly specific for the evaluation of mediastinal node involvement in bronchogenic carcinoma and it is the approach of choice when a diagnosis of lesions located in the mid-mediastinal region has not been reached. ECM is a valid, safe alternative to anterior mediastinotomy for staging nodes and masses occupying para-aortic zones or the aortopulmonary window, with good diagnostic yield, low morbidity and absence of mortality.