Observer variability in the histopathological reporting of core biopsies of papillary breast lesions is reduced by the use of immunohistochemistry for CK5/6, calponin and p63

Aim: To investigate agreement on core biopsy diagnosis of papillary breast lesions, which is acknowledged as a difficult area, and to determine the effect of the use of immunohistochemistry (IHC) to assist diagnosis. Study design: Haematoxylin and eosin (H&E) sections of 129 core biopsies of papillary breast lesions were circulated to four observers who categorized each case as: B2 (benign), B3a (epithelial proliferation, probably benign but requiring biopsy), B3b (epithelial proliferation with cytological or architectural atypia), B4 (probably malignant but insufficient material or artefact to allow diagnosis), B5 (malignant papillary lesion). In all cases (n = 127) IHC was performed for cytokeratin (CK) 5/6, calponin, p63 (myoepithelial markers), and slides recirculated. Results: There was unanimous agreement in 44% of cases on H&E only which rose to 91% after the use of IHC. Overall, unweighted kappa (Ku; five categories) rose from 0.54 to 0.91. The main effect of IHC was to reduce the use of intermediate categories (B3a, B3b and B4) and allow definitive diagnosis (B2 or B5). Conclusion: Agreement on H&E sections alone in papillary core biopsies of breast is only 44% (Ku = 0.54) but is significantly increased to 91% (Ku = 0.91) by the use of IHC for CK5/6, calponin and p63.