Introduction: In the front of the problems related to the differentiation between benign and malignant thyroid tumors we decided to perform a multicentre study in order to validate diagnoses of malignant thyroid tumors and assess the inter-observer variability.
Material and methods: Material included 690 cases of malignant and benign thyroid lesions with primary histopathology established in 1985-1999. These cases were selected to multicentre study. The studies were sent from centres which agreed to participate in the project and than coded in the independent centre - Department of Nuclear Medicine and Endocrine Oncology. 40 pathologists from 25 centres provided their diagnoses which were compared with the reference ones.
Results: 10 547 diagnoses were evaluated, both on their accuracy of the distinction between malignant and benign lesions and on their accuracy of cancer histotype definition. The reference diagnosis was made by an agreement between four expert pathologists (D.L., S.S., J.S. and A.K.). The participants diagnosed 21% of cases differently than experts. Concerning the diagnosis of cancer histotype, the difference between participants diagnosis and the reference one was even higher. The best concordance was achieved in the diagnosis of papillary thyroid cancer, however, on the cost of cancer overdiagnosis by some participants. Follicular cancer was diagnosed accurately only in 75.4% of cases.
Conclusion: The study documents a high inter-observer variability of thyroid cancer diagnosis and confirms the lesser accuracy of diagnosis of follicular cancer.