Evidence-based criteria to help distinguish metastatic breast cancer from primary lung adenocarcinoma on thoracic frozen section.

The distinction between primary lung adenocarcinoma and metastatic breast carcinoma in patients with a history of breast cancer is difficult by frozen section (FS) analysis. Our experience with 129 FSs from 121 patients with a pulmonary nodule and a history of breast cancer was reviewed. The pretest odds ratio of primary pulmonary carcinoma/metastatic breast carcinoma was 2.6. The incidence of 12 histopathologic features was assessed in a "training set" composed of 20 FSs, 10 with primary lung adenocarcinoma and 10 with metastatic breast cancer. A differential diagnosis model composed of significant pathologic features that favor the diagnosis of primary lung adenocarcinoma (acini, lepidic growth, nuclear pseudoinclusions, and scar) or metastatic breast carcinoma (comedonecrosis, solid nests, trabecular architecture, and cribriform growth) was identified. The external validity of this model was successfully tested by challenging 19 pathologists and trainees with a test set of 20 unknown FSs, supporting the clinical applicability of the diagnostic model.