Characterizing breast cancer cell lines using principal component analysis of high-frequency ultrasonic spectra

Abstract

Current methods for determining the molecular subtypes of breast cancer include DNA microarrays, immunohistochemical staining, and proteomic analysis. These methods are not easily transferable to real-time clinical applications such as the intraoperative assessment of margin or biopsy specimens. The development of a diagnostic method for rapidly determining the molecular subtype of malignant breast tissue in a clinical setting would represent a significant advance in breast cancer detection and treatment. Preliminary results suggest that high-frequency ultrasound may be sensitive to variations between breast cancer cell lines, and thus molecular subtype, due to a mechanism linking subtype mutations to the ultrasonic properties of the cells. This mechanism was explored using an integrated experimental and computational method. Ultrasonic scattering from breast cancer cells in vitro were simulated using a multipole-based approach. Variations between cell lines due to different molecular subtypes were modeled using a range of cell moduli and sizes. Model and experimental spectra were compared using principal component analysis (PCA). The results indicate the properties and thus molecular subtypes of breast cancer cells could potentially be determined by comparing their measured spectra to model spectra using a feature classification program such as PCA. [This work was supported by Utah Valley University.]