Adhesion strength of tumor cells predicts metastatic disease in vivo.

Abstract

Although only a fraction of tumor cells contribute to metastatic disease, no prognostic biomarkers currently exist to identify these cells. We show that a physical marker-adhesion strength-predicts metastatic potential in a mouse breast cancer model and that it may stratify human disease. Cells disseminating from murine mammary tumors are weakly adherent, and, when pre-sorted by adhesion, primary tumors created from strongly adherent cells exhibit fewer lung metastases than weakly adherent cells do. We demonstrate that admixed cancer lines can be separated by label-free adhesive signatures. When applied to murine metastatic tumors, adhesion retrospectively predicts metastatic disease with 100% specificity, 85% sensitivity, and area under the curve (AUC) of 0.94. Cells from human reduction mammoplasties have a higher adhesion strength versus resected human tumors, which may also be stratified between invasive and more indolent cancers. Thus, highly metastatic cells may have a distinct physical phenotype that may be a predictive marker of clinical outcomes.