Digital image analysis improves the quality of subjective HER-2 expression scoring in breast cancer.

Aim To compare HER-2 scoring reproducibility by subjective and digital image analysis (DIA) scores with each other and with fluorescence in situ hybridization (FISH) assessed HER-2 amplification. Methods Herceptest-stained Tissue Micro Arrays of 219 breast carcinomas were scored (DAKO protocol) by 3 observers (both independent and as consensus), scored by DIA and both scores were compared with FISH amplification results. Results Interobserver subjective scores reproducibility was good (κ 0.82 to 0.86) but therapeutically important 3+/2+discrepancies occurred in 11% to 16% of all 3+ cases. Subjective scores and FISH results differed considerably. Consensus scores by 3 pathologists correlated better with FISH, reducing the number of both Immunohistochemical (IHC) negative/FISH positives and IHC 3+/FISH negatives. DIA scores were well reproducible and correlated better with FISH amplification than did subjective scores. Conclusions DIA scores were comparable with consensus scores between 3 expert pathologists, were very well reproducible and performed better in classifying IHC 3+/FISH+ cases than did subjective scores.