[Criteria for histopathologic diagnosis of thyroid cancer in a quick test of accuracy]

A quick test of accuracy of histopathological diagnosis of thyroid carcinoma was performed in May 2000 during the meeting of Polish thyroid cancer group (Committee for Epidemiology, Diagnosis and Treatment of Thyroid Carcinoma). 29 pathologists participated in the test and evaluated 8 cases of thyroid carcinoma and 14 benign thyroid lesions. All cases were chosen from the current material sent for pathologic evaluation to the Institute of Oncology in Gliwice due to diagnostic difficulties. In total, 591 diagnoses were made and were the subject of the presented analysis. They were compared with reference diagnosis in two aspects. First, the accuracy of the distinction between malignant and benign lesions was evaluated. 72.5% of diagnoses were concordant with the reference. The false diagnosis of cancer in a benign lesion was observed 133 times (22.5% of all diagnoses). A reverse error--a false exclusion of cancer--was seen in 29 diagnoses (4.9%). Chi 2 test revealed a statistically significant difference between the participants' diagnoses and reference ones (p < 0.0001). Overdiagnosis of cancer was the most frequent at the diagnosis of follicular or oxyphilic cancer. With reference to the diagnosis of cancer histotype, concordant diagnoses were seen in 40-47% of cases with the lowest accuracy of the diagnosis of oxyphilic (40% of correct diagnoses) and follicular (50%) cancer. The causes of false diagnoses may be divided in two groups: sample-related causes (sampling of surgical specimens, lack of standard description, insufficient number of samples, poor quality of staining) and diagnostic errors: non-compliance with diagnostic criteria and inappropriate setting of diagnoses, which require immunohistochemical confirmation.