The utility of an investigational real-time RT-PCR assay system for the detection of micrometastases in sentinel lymph nodes of breast cancer confirmed by 0.2 mm histological interval analysis

Abstract

10560 Background: Comparison between real-time RT-PCR and histological analysis for the detection of lymph node metastasis in breast cancer patients is difficult because of the necessary use of different tissue for the two tests. In order to minimize this challenge this study used histology sections taken every 0.2 mm and all tissue between the sections was analyzed by real-time RT-PCR. Methods: 129 Lymph nodes (LNs) were obtained by Sentinel LN biopsy and/or axillary dissection from 80 breast cancer patients. All LNs were cut in half. One half of each LN was used for the routine intra-operative diagnosis. The other half was used for the 0.2mm histological interval analysis. All 10μm sections cut for histology at 0.2 mm intervals were analyzed using frozen section H&E under a light microscope. All remaining tissue between the sections used for histological analysis was assayed by real-time RT-PCR using the genetic markers mammaglobin and cytokeratin-19 (investigational GeneSearch Breast Lymph Node Assay, Veridex, LLC, Warren, NJ, USA). Cutoffs were pre- set for the real-time RT-PCR to detect only metastasis greater than 0.2 mm. Results: Compared to the histological diagnosis using 0.2 mm interval frozen sections, the real-time RT-PCR results were as follows; sensitivity of 100.0% (34/34), specificity of 93.7% (89/95) and overall accuracy of 95.3% (123/129). Conclusions: These data suggest by utilizing the 0.2 mm histological interval analysis, sampling discrepancies are minimized and a high level of sensitivity and overall accuracy is seen for the real-time RT-PCR assay. Assay specificity calculations may be less than 100% due to interpretation challenges in frozen section histology, especially in the cases of smaller metastases and/or lobular cancer, causing a small percentage of clinically relevant metastasis to be missed. [Table: see text]