Improved detection by ensemble-decision aliquot ranking of circulating tumor cells with low numbers of a targeted surface antigen.

Circulating tumor cells (CTCs) are shed from a solid tumor into the bloodstream and can seed new metastases. CTCs hold promise for cancer diagnosis and prognosis and to increase our understanding of the metastatic process. However, their low numbers in blood and varied phenotypic characteristics make their detection and isolation difficult. One source of heterogeneity among CTCs is molecular: When they leave the primary tumor, these cells must undergo a molecular transition, which increases their mobility and chance of survival in the blood. During this molecular transition, the cells lose some of their epithelial character, which is manifested by the expression of the cell surface antigen known as epithelial cell adhesion molecule (EpCAM). Some tumors shed CTCs that express high levels of EpCAM; others release cells that have a low level of the antigen. Nevertheless, many CTC isolation techniques rely on the detection of EpCAM to discriminate CTCs from other cells in the blood. We previously reported a high-throughput immunofluorescence-based technology that targets EpCAM to rank aliquots of blood for the presence or absence of a CTC. This technology, termed ensemble decision aliquot ranking (eDAR), recovered spiked-in cancer cells (taken from a model EpCAM(high) cell line) from blood at an efficiency of 95%. In this paper, we evaluated eDAR for recovery of cells that have low EpCAM expression and developed an immunofluorescence labeling strategy that significantly enhances the method's performance. Specifically, we used a cocktail of primary antibodies for both epithelial and mesenchymal antigens as well as a dye-linked secondary antibody. The cocktail allowed us to reliably detect a model EpCAM(low) cell line for triple negative breast cancer, MDA-MB-231, with a recovery efficiency of 86%. Most significantly, we observed an average of 6-fold increase in the number of CTCs isolated from blood samples from breast cancer patients. These findings underscore the importance of benchmarking CTC technologies with model cell lines that express both high and low levels of EpCAM.