Clinicopathological features of thymic carcinomas and the impact of histopathological agreement on prognostical studies.

OBJECTIVES Thymic carcinomas have wide ranges of reported survival. Interobserver agreement on diagnosis might affect prognostical studies. Clinicopathological features of thymic carcinomas were compared with thymic epithelial neoplasms in which pathologists disagreed upon. METHODS Patients treated with thymic epithelial neoplasms were reviewed. Three thoracic pathologists independently classified all cases according to the World Health Organization classification. The study group comprised cases in which all pathologists agreed independently on thymic carcinomas. A disagreement group included cases in which pathologists disagreed upon the diagnosis. Tumours were staged according to the modified Masaoka and tumour-node metastasis (TNM) stages. Time-to-death was estimated with the Kaplan-Meier method. Survival outcomes were assessed with the Cox proportional-hazards regression. RESULTS In the study group, 25 of 29 patients presented with symptoms but no autoimmune diseases. Masaoka stage III (18 of 29) and TNM stage III (13 of 24) were most common. Complete tumour resection was achieved in 17 of 29. Four patients had metastasis at diagnosis, and 12 developed metastasis/recurrence post-treatment. The estimated 5-year survival was 35.6%, and recurrence/metastasis-free survival was 34.2%. Overall survival was associated with weight loss (P = 0.02) and metastasis/recurrence with morphology (P = 0.009). In the disagreement group, most disagreements occurred between type B3 thymomas and carcinomas (21 of 29). Twenty-four of the 29 patients presented with symptoms, including autoimmune disorders (12 of 29). Masaoka stage III (10 of 29) and TNM stage IV (10 of 17) were most common. Twenty-one of the 29 underwent complete tumour resection. The estimated 5-year survival was 64.9%. Two patients had metastasis at diagnosis and eight developed metastasis/recurrence post-treatment. The study group had significantly more patients with chest pain and additional treatment than the disagreement group (P = 0.005 and 0.044, respectively). The disagreement group had more patients with myasthenia gravis and a higher TNM stage (P = 0.0003 and 0.025, respectively). The risks of death and recurrence/metastasis were significantly higher in the study group than the disagreement group [P = 0.025, hazard ratio (HR) = 2.44 and P = 0.012, HR = 3.23, respectively]. CONCLUSIONS Thymic carcinomas were diagnosed at high stages and the overall prognosis appeared relatively poor. Autoimmune disease was not a manifestation of thymic carcinomas. Weight loss was associated with survival. The disagreement group in contrast had more patients with autoimmune syndrome and, despite a higher stage, had a better survival, suggesting that interobserver variability in the histopathological classification of thymic carcinomas vs thymomas leads to prognostical variability.