Quantitative Raman spectroscopy of breast cancer malignancy utilizing higher-order principal components: A preliminary study

Abstract

A major challenge in analyses of molecular spectra from biological samples has been the detection of trace biomarkers, which are subtle biochemical alterations (in parts per million (ppm)) caused by disease, buried in pronounced background fluorescence. We report a quantitative chemometrics-assisted Raman study of subtle biochemical alterations associated with breast cancer malignancy using whole blood samples, based on a 785 nm laser excitation. To understand biochemical differences between healthy (control) and diseased samples, spectral analysis was undertaken in the 500–1800 cm−1 region using principal components analysis (PCA), linear discriminant analysis (LDA), and partial least squares discriminant analysis (PLS-DA). The subtle spectral markers at 589, 594, 630, 858, 868, 1005, 1160, 1250, 1347, 1358, 1626, 1630, and 1638 cm−1 differentiated controls from diseased patients and were assigned to proteins, lipids, and nucleic acids. Out of the above, six spectral regions were determined: 589, 594, 630, 1626, 1630 and 1638 cm−1, which can be regarded as new spectral markers for breast cancer. Various pure basic biochemical components were used to develop a partial least squares regression calibration model for quantitative analysis. The relative concentrations of biochemical alterations in healthy and diseased samples were estimated by applying the developed least squares fitting model to the determined trace spectral markers’ measured blood spectrum. The fitting model revealed that the relative concentrations of proteins, lipids, and nucleic acids increased with disease status (p < 0.05). Both PCA-LDA and PLS-DA models yielded sensitivities and specificities > 80%, and overall diagnostic accuracies between 90 and 100%. Considering the limited number of samples involved in this study, preliminary results from this approach are promising, encouraging further investigations.