Interobserver reproducibility of Her-2/neu protein overexpression in invasive breast carcinoma using the DAKO HercepTest

Abstract

Although there are criteria for interpretation of the staining results for Her-2/neu in the DAKO HercepTest, the determination of staining intensity and percentage of complete membrane staining is subjective. We studied 46 cases of invasive breast carcinoma to evaluate interobserver reproducibility among 5 pathologists. Complete agreement was achieved in 22 (48%) of 46 cases. Generalized kappa values indicated substantial agreement (0.80). Discrepancies between negative (0, 1+) and positive (2+, 3+) results occurred in 2 cases (kappa = 0.96). One was because of a tangential cut of the basal part of the tumor that mimicked complete membranous staining, and the other was a borderline case that revealed focal (5%-15%) complete membranous staining. Distinguishing weakly (2+) from strongly (3+) positive results showed agreement in only 13 (59%) of 22 positive cases (kappa = 0.38). If more than 50% of tumor cells revealing strong complete membrane staining were regarded as strongly positive, agreement would be improved (kappa = 0.78). While there was a high percentage (70%-80%) of negative cases during routine evaluation, the good interobserver agreement and high negative predictive value made immunohistochemical analysis an effective screening test to exclude negative cases.