Stromal Cell Adhesion Predicts Severity of Metastatic Disease

Abstract Despite better outcomes with early-stage detection, local invasion significantly reduces patient survival rates for many carcinomas. Heterogeneity within and between tumors has precluded identification of predictive biological markers, but adhesion strength has emerged as a potential biophysical marker. Here we demonstrate that cells disseminating from mammary tumors are weakly adherent, and when presorted by adhesion, primary tumors created from strongly adherent cells exhibit fewer lung metastases than weakly adherent cells or unsorted populations. Migratory ontologies from tumors correlate with freshly sorted cells, suggesting that cell intrinsic differences are maintained in vivo. We further demonstrate that admixed cancer lines can be separated by label-free adhesive signatures using a next-generation flow chamber. When applied to metastatic tumors, the device retrospectively predicted metastatic disease from stromal samples with 100% specificity, 85% sensitivity, and AUC of 0.94. Data from this device suggest that label-free adhesive signatures may effectively predict clinical outcomes in patients.