Computer-assisted immunocytochemical determination of breast cancer steroid receptors. Frozen sections vs paraffin sections

Abstract

Previously we have demonstrated that determination of oestrogen (ER) and progesterone (PR) receptors by immunocytochemical assay (ICA) on frozen sections (FS) and cytological smears with image analysis is effective for evaluating steroid receptors. The aim of this study was to determine concordance between ER and PR assessed by ICA on FS and paraffin sections (PS) both evaluated by image analysis. There were 115 breast carcinomas selected. For all cases, ER and PR determination was performed on FS and PS. Computer-assisted image analysis was performed using CAS 200. Results were expressed as percent positive area of neoplastic nuclei compared with total nuclear area of the examined neoplastic cells. Good correlation was demonstrated for both ER (r=0.759; concordance=83.4%) and PR (r=0.800; concordance=87.8%). The unexpected relatively low concordance for ER led to further investigations. We divided the 115 cases in two groups. The first group included specimens from our hospital; the second group specimens from suburban hospitals. In the first group there was better correlation for both ER (r=0.897) and PR (r=0.915) with a concordance of 91.5% and 93.6%, respectively. In the second group, correlation was worse for both ER (r=0.724) and PR (r=0.708), with a concordance of 77.9% and 83.9% respectively. From analysis of discordant cases we conclude that reduction in correlation and concordance with increased false negative cases in group 2 are probably due to delayed fixation. Our data suggest that ICA with automated image analysis is efficient in evaluating ER and PR on paraffin section only when the tumour samples are correctly fixed.