Raman spectroscopy of serum: an exploratory study for detection of oral cancers.

Early diagnosis of oral cancers, one of the major cancers, is of utmost importance as 5-year disease-free survival rates are some of the lowest, despite advances in treatment and surgical modalities. In vivo Raman spectroscopy has shown efficacy in the detection of normal, premalignant and malignant lesions and even of early changes such as cancer-field-effects/malignancy-associated-changes. However, the need for a dedicated instrument and stringent laboratory conditions, at all diagnostic centers, limits wide screening applications of this method. In light of this, it is pertinent to explore ex vivo samples like serum due to its ease of collection, storage, transport and analysis at a centralized facility. Hence, Raman studies were carried out on serum from 14 buccal mucosa and 40 tongue cancers as well as 16 healthy control samples. Spectral features indicate differential contributions of proteins, DNA, and amino acids like Phe, Trp and Tyr and β-carotene in the analyzed groups. Highly intense Raman bands assigned to β-carotene could be due to resonance Raman, and were observed in all sera with the highest relative intensity in normal samples. Higher DNA and protein content were observed in the mean cancer spectra. Principal component-linear discriminant analysis (PC-LDA) followed by cross-validation using leave-one-out cross-validation (LOOCV) were employed for data analysis which was carried out both spectra- and patient-wise. Findings indicate the possibility of classifying normal and oral cancer sera in both these approaches; however, the patient-wise approach could be the preferred mode for prospective studies. Besides, a tendency of classification for buccal mucosa and tongue cancers was also observed. Prospective validation of these results on a large sample size may help in the translation of this methodology to clinics.