MOLECULAR ANALYSIS OF CANCER PROGRESSION WITH LABEL-FREE RAMAN SPECTROSCOPY
Johns Hopkins University
Due to its ability to probe water-containing samples using visible and near-infrared frequencies with high chemical specificity, Raman spectroscopy is an attractive tool for label-free investigation of biological samples. While Raman spectroscopy has been leveraged for exploratory studies in clinical cancer diagnostics, only limited studies have used it to understand the molecular mechanisms driving key characteristics of cancer progression. In this thesis, we present three progressively complex applications of Raman spectroscopy that take advantage of its specificity and synergistic combination with plasmonic nanoparticles and multivariate data analysis for molecular study of cancer. First, we used Au@SiO2 shell-isolated nanoparticle-enhanced Raman spectroscopy (SHINERS) to investigate the roles of microcalcification status and the composition of tumor microenvironment in breast tissue for identification of a range of breast pathologies. We developed a partial least squares-discriminant analysis-based classifier to correlate the spectra with their pathology to obtain high prediction accuracy. A parallel investigation of the genetic drivers of microcalcification formation in breast cancer cells revealed that stable silencing of the Osteopontin gene decreased the formation of hydroxyapatite in breast cancer cells and reduced their migration. Next, we demonstrated the ability to detect premetastatic changes in the lungs of mice bearing breast tumors, in advance of tumor cell seeding, using Raman spectroscopy and multivariate data analysis. Our measurements showed reliable differences in the collagen and proteoglycan features of the premetastatic lungs which uniquely identify the metastatic potential of the primary tumor. Consistent with histological assessment, our results hint at a continuous premetastatic niche formation model dependent on the metastatic potential of primary tumor. Finally, we exploited Raman mapping to elucidate radiation therapy-induced biomolecular changes in murine tumors and uncovered latent microenvironmental differences between treatment-resistant and -sensitive tumors. We used multivariate curve resolution-alternating least squares (MCR-ALS) and support vector machine (SVM) to quantify biomolecular differences in the tumor microenvironment and constructed classification models to predict therapy outcome and resistance. We found significant differences in lipid and collagen content between unirradiated and irradiated tumors. Taken together, these studies pave the way for applications of Raman spectroscopy beyond clinical diagnostics such as metastatic risk assessment and treatment monitoring.
Raman spectroscopy, Label-free, SERS, Multivariate Data Analysis, Support Vector Machine, Partial Least Squares Discriminant Analysis, Breast Cancer, Premetastatic niche, Radiation therapy, Microcalcifications, SHINERS, OPN, Biomedical optics, Cancer