Frozen section diagnosis and surgical biopsy of lymph nodes, tumors and pseudotumors of the mediastinum

Objective: Our experience with the use of frozen section (FS) was reviewed in order to assess its interest and limits, during minimally aggressive mediastinal surgery for staging of lung carcinomas and biopsy of primary lesions. Methods: The retrospective study was based on a series of 420 consecutive mediastinal biopsies with FS performed through cervicotomy (n=12), mediastinoscopy (n=345), mediastinotomy (n=43), manubriotomy (n=13) or videothoracoscopy (n=7), in 417 patients, aged 8–86 years (mean: 52.5±15.6). The FS diagnoses were compared with the final diagnoses and consequences of the FS analyses were analyzed. Results: Biopsies with FS had a 99.4% (351/353) efficiency rate for a precise definitive diagnosis (excluding normal lymph nodes). However, two patients required a second operation: one for typing a lymphoma and the other to correct a FS misdiagnosis of small cell carcinoma on a crushed normal lymph node. The sensitivity for detection of carcinoma lymph node metastases was 99% (200/202). The two false negative FS diagnoses, including one post-chemotherapy, were related to micrometastases. There were 46 correct FS diagnoses of non-metastatic lymph node which were followed by an immediate thoracotomy during the same anesthesia, for benign lung lesions (n=9) or for carcinomas (n=37), including 30 lung carcinomas that were immediately resected and proved to have no mediastinal involvement (n=24), or only a limited involvement in a non-biopsied site (n=6). In the 51 cases of primary mediastinal tumors excluding carcinomas, FS indicated a resectable lesion with a sensitivity of 87.5% (7/8) and a specificity of 97.7% (43/44). Five lesions were immediately resected: one Castleman’s disease, one intrathoracic goiter, two of six thymomas and a Hodgkin’s disease, which was diagnosed as thymoma on FS. An invasive thymoma was resected during a second operation after a FS diagnosis of carcinoma. FS had a sensitivity of 100% in the 62 cases of sarcoidosis and a sensitivity of 90% in the 20 cases of infectious lesions. One of the 18 cases of tuberculosis and an infectious pseudo-tumor of the anterior mediastinum had no microbiologic study because of FS diagnoses of tumor necrosis and lymphoma. Conclusions: FS efficiently secures the adequacy of the samples and guides the surgeon’s decision making for the resection of lung carcinomas, but is less effective for a precise diagnosis of some primary mediastinal lesions, which may have close histologic appearances.