The aim of this study was to investigate in primary breast cancer the congruency of routine clinical predictive biomarker evaluations, including ER, PR and Ki67, obtained using immunocytochemistry (ICC) and immunohistochemistry (IHC).Methods and results:
Clinicopathological data were collected on all women diagnosed with primary breast cancer at Karolinska University Hospital in 2011. A total of 346 patients were included in a retrospective paired comparison of predictive biomarker evaluations on direct smear ICC and IHC. This showed a low congruency between findings with the two methods, especially evident for Ki67 (κ = 0.35–0.42). By suggested adjustments to ICC cut-offs, we managed to improve the inter-rater agreement of Ki67 classification slightly to κ = 0.46.Conclusions:
Our findings suggest that routine clinical ICC and IHC evaluations of predictive biomarkers produce discordant results. Consequently, basing therapeutic decisions on cytology with cut-offs defined for IHC induces a risk that patients will receive suboptimal therapy. However, our analysis shows that local adjustments to biomarker cut-off levels may improve congruency and increase the probability of correct classifications.