Comparative merits of conventional, computed tomographic, and magnetic resonance imaging in assessing mediastinal involvement in surgically confirmed lung carcinoma.

Thirty-four patients with operable malignant tumors of the lung had computed tomography and magnetic resonance imaging of the chest in addition to regular chest roentgenograms and bronchoscopy. The purpose of the study was to assess the extent of tumor involvement in the hilum and the mediastinum by direct invasion and by regional lymph node metastasis. At thoracotomy, 23 tumors were completely resected and 11 were treated by interstitial implantation of radioisotopes. In addition, a mediastinal lymph node dissection or sampling was performed to correlate nodal involvement with the preoperative studies. The tumor was peripheral in 21 patients and central in 13. Histologically, 18 tumors were adenocarcinomas, 14 epidermoid cancers, and two atypical carcinoids. Preoperatively, 18 tumors were classified as N0 disease, nine as N1, and seven as N2. Pathologically, 11 were N0, eight N1, and 15 N2. Plain chest roentgenograms correlated poorly with the nodal findings at operation. Both magnetic resonance and computed tomographic imaging were highly accurate in assessing the hilum and the presence of mediastinal adenopathy, with a sensitivity rate of 87%. Except for identifying contact with the mediastinum, neither method correlated well with mediastinal invasion when present (sensitivity rate 55% for computed tomography and 64% for magnetic resonance) and neither method could differentiate hyperplastic from metastatic nodes. Hence, no advantage of magnetic resonance over computed tomographic scanning was noted in assessing tumor involvement of the mediastinum by direct invasion or by regional lymph node metastasis.