The freeze–thaw cycle effect on blood serum autofluorescence, Raman spectroscopy and SERS: implications for sample classification and disease diagnostics†

The issue of variability introduced into blood plasma and serum analysis by preanalytical procedures is the major obstacle to obtaining accurate and reproducible results. While the question of how to overcome this issue has been discussed in biochemical detection of analytes and omics technologies, its relevance to the field of optical spectroscopy remains mostly unexplored. In this work, we evaluated the freeze–thaw cycle (FTC)-induced alternations in blood serum optical properties by means of autofluorescence and Raman spectroscopy, including surface-enhanced Raman spectroscopy (SERS). In the case of regular Raman spectroscopy, FTC-specific spectral variability was estimated to be <1%, being significantly smaller than patient-specific variability, while the t-distributed stochastic neighbor embedding clustering of principal components yielded spectral grouping by patient ID independent of sample freezing. For SERS, FTC-specific and patient-specific spectral variabilities were 15% and >90%, respectively. Finally, parallel factor analysis of autofluorescence excitation–emission matrices revealed that patient-specific variability in the visible spectral range was 13%, whereas FTC-specific variability was 4%. We further evaluated disease-specific variability for two datasets, namely, for colorectal cancer diagnostics with autofluorescence and for chronic kidney disease diagnostics using SERS. Disease-associated variabilities were determined to be 8% and 49%, significantly exceeding the possible FTC-induced variability. Hence, the obtained results suggest that FTC blood serum samples can be used for disease diagnostics by Raman spectroscopy and SERS, as well as through autofluorescence spectroscopy, although the difference in FTC-induced and disease-induced variabilities was lowest in the latter case.