Automated Intra-Operative Analysis of Breast Cancer.

Abstract

Abstract Background: Multiple operations are a major problem for breast conservation therapy. Accurate interpretation of intra-operative tumor margins can limit multiple re-excision procedures in order to obtain negative margins while maintaining a good cosmetic result from BCT. Intra-operative touch preparations have been used in the past but can be difficult to interpret without an experienced cytopathologist.Objective: To examine the reliability of fluorescently labeled intra-operative touch preps in conjunction with automated analysis in correctly identifying cancer cells compared to final pathologic diagnosis.Methods: Touch preps were propctively performed on 38 consecutive cases of patients undergoing breast surgery. The surgical cases included 15 invasive tumors, 9 ductal carcinoma in situ, 5 prophylactic mastectomies, and 9 normal tissues from breast reduction operations. Tumors were grossly located by a surgical pathologist and a cross sectional cut was made at the suspected tumor center. Touch preps with poly-l-lysine coated slides were then taken on the tissue surfaces exposed from the cross sectional cut. Additionally, the location where the touch preps were performed was isolated in a cassette and separated for later comparison. In cases of benign disease, touch preps were performed at the center of the excised tissue. The slides were fixed in 95% ethanol and stained with a Cytokeratin, an epithelial cell marker, and Hoechst, a nuclear stain. The prepared slides were then imaged with an automated fluorescent microscope, and suspected cancer cells were detected through standard image processing and statistical techniques.Results: The overall accuracy of our automated fluorescence analysis was 95% for identifying invasive or pre-invasive cancer compared final pathologic diagnosis. The overall specificity was 100% (there were no false positives). The sensitivity for ductal carcinoma in situ was 75% and invasive cancer was 90%. Overall sensitivity was 87% which is comparable to the best reported results from manual examination of touch preparations.Conclusion: Automated analysis of fluorescently stained touch prep slides is highly accurate in identifying cancer with no false positives. The immunofluorescence stains and automated microscopy may help the pathologist to quickly identify cancer cells in fresh breast tissue during BCT operations, and, thus, limit breast re-excisions in the future. Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 6016.