Histopathological Grading of Breast Ductal Carcinoma in Situ: Validation of A Web-Based Survey
Fernando Schuh, Jorge Villanova Biazús, José Antônio Crespo Cavalheiro, Rodrigo Cericatto, Ana Cristina da Costa Bittelbrunn, Alessandra Ventura, Diego de Mendonça Uchôa, Márcia Silveira Graudenz, Maria I.Albano Edelweiss · Journal of the Senologic International Society · 2012
INTRODUCTION: Histopathological grading diagnosis of ductal carcinoma in situ (DCIS) may be very difficult, even if done by experts. The challenge may be due to the inaccurate and/or subjective application of the diagnosis criteria. This study aims to compare interobserver agreement using a web-based survey created to grade DCIS by the Van Nuys Classification. The intraobserver agreement was also studied to find out its reproducibility with the traditional interpretation of digitized microscopy images of DCIS lesions. MATERIAL AND METHODS: The most important criteria to grade DCIS lesions in histopathological bases were selected to compose a point scoring system according to their relevance to DCIS classifications. A software was created in order to be accessed through Internet, in website format. This website offers a questionnaire containing the characteristics and a point scoring system used to compose the DCIS classification systems, and also the digitized microscopy images of 43 DCIS cases selected. Twelve patologists answered this questionnaire and got automatically the histopathological grade of each DCIS case by the point scoring system. After 6 months, one of these pathologists, specialist in breast pathology and with experience in the use of Van Nuys Classification, graded the same set of DCIS cases through Internet by the usual interpretation, without the questionnaire. The results were analyzed using the kappa test. The intraobserver agreement analysis was employed to validate this innovative web-based survey. RESULTS: Diagnostic agreement of these DCIS cases by Van Nuys classification showed a kappa index of 0. 37 + 0. 02. Analyzing the subgroup of experts in breast pathology, this rate raises to 0.58 + 0. 10. Meanwhile, the intraobserver reproducibility of Van Nuys classification resulted in kappa values of 0. 67 + 0. 09. DISCUSSION: The interobserver agreement was acceptable to Van Nuys classification. The diagnostic agreement between the web-based questionnaire and the traditional method, both using digital images, was good for Van Nuys classification. CONCLUSIONS: Our results confirm the difficult interobserver agreement in grading DCIS breast lesions as seemed in the literature. The use of web-based survey does not appear to pose a major risk of presenting large diagnostic disagreements. These findings indicate that the use of this web- based survey to grade DCIS lesions is a promising, useful and a reliable diagnostic tool.