Raman spectroscopy of x-ray irradiated blood plasma: a pilot study.

Abstract

Biodosimetry is a key technique for retrospective radiation dosimetry that provides individual estimates of the absorbed dose of ionising radiation based on the detection of biological indicators. A critical challenge confronting current biodosimetry methods is the time and technical expertise needed in processing and analysing samples, therefore new high-throughput techniques are required. In this pilot study, we demonstrate a technique based on RS and multivariate analysis of peripheral blood plasma from nine healthy male and female anonymous donors for the classification and biomarker identification ofex vivoirradiated blood exposed to 0 (control), 5 and 20 Gy of x-ray dose. After 4 h post-exposure, the blood was centrifuged, and the blood plasma samples were immediately frozen at -80 °C. Raman spectra were measured from thawed blood plasma using a custom benchtop Raman micro-spectroscopy setup. Data were preprocessed and analysed using partial least squares-discriminant analysis (PLS-DA). We applied a method based on a linear mixed-effects model to compensate for the differences in covariates such as gender, age and complete blood count between donors. After covariate adjustment, the application of PLS-DA to the residual Raman spectral intensities provided improved separation in the binary classification results (0 vs. 5 and 0 vs. 20 Gy). Raman spectral biomarkers responsible for the discrimination were extracted by evaluating the coefficients of the PLS-DA loading vectors. Sparse PLS-DA was demonstrated to be a promising method that offers the potential to further narrow down the regions in the Raman spectra that are dose discriminatory.