Inter- and intraobserver variability in the histopathological diagnosis of medullary carcinoma of the breast, and its prognostic implications

Abstract

One hundred thirty-one breast carcinomas with medullary features, registered in the Danish Breast Cancer Cooperative Group from 1977-1982, have been histopathologically reviewed by two senior pathologists and classified as typical medullary carcinoma (TMC), atypical medullary carcinoma (AMC), and non-medullary carcinoma (NMC). Diagnostic criteria were based on those put forward by Ridolfi et al. and Fisher et al. The procedure was repeated with an interval of about one year by both pathologists. The diagnostic interobserver agreement was 72% with a Kappa of 0.55. The intraobserver agreement was 77% and 63% with Kappa values of 0.64 and 0.44, respectively. To see whether the observed inter- and intraobserver variability had any prognostic implications, diagnostic subgroups for both pathologists were analyzed with Kaplan Meier plots for recurrence-free survival (RFS) and with log rank tests. In the first evaluation pathologist 1 segregated a group of TMC with a significantly better RFS than for the NMC group, and pathologist 2 segregated a group of TMC with a corresponding strong trend. These findings could not, however, be reproduced in the second evaluation. The study indicates that the criteria of TMC and AMC as proposed by Ridolfi et al. need to be sharpened and simplified in order to reduce inter- and intraobserver variability. Larger studies with a control group of infiltrating ductal carcinomas are mandatory to elucidate the clinical importance of the diagnoses of Typical and Atypical Medullary Carcinoma of the breast.