Identification and characterization of optimal gene expression markers for detection of breast cancer metastasis

Sentinel lymph node (SLN) status is highly predictive of overall axillary lymph node involvement in breast cancer. Historically, SLN-positive patients have undergone axillary lymph node dissection in a second surgery. Intraoperative SLN analysis could reduce the cost and complications of a second surgery; however, existing histopathological methods lack standardization and exhibit poor sensitivity. Rapid molecular methods may lead to improved intraoperative diagnosis of SLN metastasis. In this study, we used a genome-wide gene expression analysis of breast and other tissues to identify seven putative markers for detecting breast cancer metastasis. We assessed the utility of these markers for identifying clinically actionable metastases in lymph nodes through reverse transcriptase-polymerase chain reaction analysis of SLNs from 254 breast cancer patients. Polymerase chain reaction signals were compared to pathology on a per-patient basis. The optimal two-gene combination, mammaglobin and cytokeratin 19, detected clinically actionable metastasis in breast SLNs with 90% sensitivity and 94% specificity. Application of stringent criteria for identifying presumptive hematoxylin- and eosin-positive samples increased sensitivity and specificity to 91 and 97%, respectively. This study represents the first comprehensive demonstration of the utility of gene expression markers for detecting clinically actionable breast metastases. An intraoperative molecular assay using these markers has the potential to significantly reduce second surgeries for patients undergoing SLN dissection.