Interobserver variation Archives - Revista Brasileira de Ginecologia e Obstetrícia

PURPOSE: to perform a critical evaluation of the histopathological diagnosis of ductal carcinoma in situ (DCIS) of the breast, through the analysis of interobserver variation related to diagnosis, architectural pattern, nuclear grade, and histological grade. METHODS: eighty-five cases with an initial diagnosis of DCIS were reviewed by the same pathologist, specialist in breast pathology, who selected 15 cases for interobserver analysis. The analysis was carried out by five pathologists and an international expert in breast pathology, who received the same slides and a protocol for classifying the lesions as atypical ductal hyperplasia (ADH), DCIS, or ductal carcinoma in situ with microinvasion (DCIS-MIC). If the diagnosis was DCIS, the pathologists should classify it according to the dominant architectural pattern, nuclear grade, and histological grade. The results were analyzed using percent concordance and the kappa test. RESULTS: there was a great interobserver diagnostic variation. In one case we had all diagnoses, from ADH, DCIS to DCIS-MIC. The kappa test for the comparison among the five observers' and the expert's diagnoses showed minimum interobservers' concordance (<0.40). Regarding DCIS classification related to the dominant architectural pattern and the histological grade, the kappa test values were considered poor among the pathologists. The best results were obtained for the nuclear grading, with a kappa index up to 0.80, considered as good concordance. CONCLUSION: the low index of interobserver concordance in diagnosis and classification of DCIS of the breast indicates the difficulty in using the most common diagnostic criteria of the literature and the need for specific training of non-specialist pathologists in breast pathology for the diagnosis of these lesions.