Atomic force microscopy indentation and inverse analysis for non-linear viscoelastic identification of breast cancer cells.

Breast cancer cells (MCF-7 and MCF-10A) are studied through indentation with spherical borosilicate glass particles in atomic force microscopy (AFM) contact mode in fluid. Their mechanical properties are obtained by analyzing the recorded reaction force-time response. The analysis is based on comparing experimental data with predictions from finite element (FE) simulation. Here, FE modeling is employed to simulate the AFM indentation experiment which is neither a displacement nor a force controlled test. This approach is expected to overcome many underlying problems of the widely used models such as Hertz contact model due to its capability to capture the contact behaviors between the spherical indentor and the cell, account for cell geometry, and incorporate with large strain theory. In this work, a non-linear viscoelastic (NLV) model in which the viscoelastic part is described by Prony series terms is used for the constitutive model of the cells. The time-dependent material parameters are extracted through an inverse analysis with the use of a surrogate model based on a Kriging estimator. The purpose is to automatically extract the NLV properties of the cells with a more efficient process compared to the iterative inverse technique that has been mostly applied in the literature. The method also allows the use of FE modeling in the analysis of a large amount of experimental data. The NLV parameters are compared between MCF-7 and MCF-10A and MCF-10A treated and untreated with the drug Cytochalasin D to examine the possibility of using relaxation properties as biomarkers for distinguishing these types of breast cancer cells. The comparisons indicate that malignant cells (MCF-7) are softer and exhibit more relaxation than benign cells (MCF-10A). Disrupting the cytoskeleton using the drug Cytochalasin D also results in a larger amount of relaxation in the cell's response. In addition, relaxation properties indicate larger differences as compared to the elastic moduli like instantaneous shear modulus. These results may be useful for disease diagnosing purposes.