Breast cancer outcomes by steroid hormone receptor status assessed visually and by computer image analysis

Abstract

To compare the assessment of steroid hormone receptor immunohistochemistry by eye and by computer-aided image analysis, and to examine their relationships with survival in breast cancer. Allred scores and weighted histoscores for oestrogen receptor (ER) and progesterone receptor (PR) immunohistochemistry were determined by eye (visual histoscore) for 459 primary invasive ductal breast carcinomas in triplicate tissue microarrays. Histoscores were also determined by computerized image analysis (automated histoscore). ER and PR status determined by these different methods were compared with each other and in their ability to predict survival over at least 142 months of follow-up. Allred and visual histoscore were highly associated for ER and PR (both P < 0.001). By univariate analysis, Allred score and visual histoscore for ER and PR were highly associated with recurrence-free and cancer-specific survival (both P < 0.001) in patients with invasive ductal breast cancer overall, in those who received tamoxifen, and in those with recurrence on tamoxifen. Visual and automated histoscores were in excellent agreement for ER and PR (both P < 0.001), and were equally effective in predicting recurrence and survival for patients with invasive breast cancer who received tamoxifen. Automated histoscore appears to be a valid alternative to visual histoscore or Allred score for determining ER and PR status.