Thymic carcinoma. A clinicopathologic study of 60 cases.

The clinicopathologic features of 60 patients with thymic carcinoma were studied. Patients ranged in age from 10 to 76 years (mean, 46), of whom 24 were females and 36 were males. Overall survival at 1, 3, and 5 years was 56.6%, 40%, and 33.3%, respectively. The following morphologic features were correlated with survival: type of tumor margins; presence or absence of a lobular growth pattern; nuclear atypia; necrosis; mitotic activity; and histologic tumor type and grade. Eighty eight percent of patients with poorly circumscribed/infiltrating neoplasms died of their tumors as compared with 16.6% of patients with well-circumscribed neoplasms (P less than 0.0000). Of patients whose tumors had mitotic counts exceeding 10/10 high-power fields (HPF), 84.3% died, as compared with 21.4% of patients with lower mitotic counts (P less than 0.0000). Of patients whose tumors showed lack of lobular growth pattern, 91.6% died, as compared with 29% of those whose tumors had a lobular growth pattern (P less than 0.0000). Finally, 84.6% of patients whose tumors displayed a high-grade histology (lymphoepithelioma-like carcinoma, small cell/neuroendocrine carcinoma, clear cell carcinoma, sarcomatid carcinoma, and anaplastic/undifferentiated carcinoma) died of tumor, as compared with 0% of patients whose tumors were of low-grade histology (well-differentiated squamous carcinoma, mucoepidermoid carcinoma, and basaloid carcinoma) (P less than 0.0000). Evaluation of the various treatment modalities used to treat these patients did not yield any statistically significant correlations with survival. Two clinically distinct groups of patients were identified: one after a relatively favorable clinical course with long survival, and one after a rapidly fatal outcome. The morphologic features of the tumors in these patients correlated well with their clinical behavior; histologic type (and the grade to which it was assigned) constituted the most reliable and important predictor of prognosis.