Non-invasive Detection Of Nasopharyngeal Carcinoma Using Saliva Surface-enhanced Raman Spectroscopy

Objective: The utility of saliva surface-enhanced Raman spectroscopy(SERS) for non-invasive nasopharyngeal carcinoma(NPC) detection was evaluated in this study.Method: The saliva samples of 32 pathologically proven NPC patients and 30 healthy volunteers were collected, Ag colloids were synthesized by deoxidizing method, the saliva was mixed with Ag colloids in a ratio of 1:1, in Via confocal Raman microscope was applied to test the samples, the raw spectra needed to be preprocessed in order to remove the experimental artifacts, biochemical important Raman bands in the SERS spectra were tentatively assigned. The saliva SERS spectra obtained from healthy subjects and NPC patients were further analyzed by multivariate statistical techniques based on principal component analysis and linear discriminant analysis(PCA-LDA).Results: The saliva SERS spectra of NPC were similar to that of healthy group, The saliva SERS spectrum peaks at 447, 496, 590, 635, 729, 812, 888, 1003, 1052, 1134, 1204, 1270, 1336, 1448, 1619, and 1662 cm-1 of NPC patients also appeared in the saliva SERS spectra of healthy volunteers, especially the strongest peaks at 496, 635, 1003, and 1134 cm-1 were existed in both groups. The key difference between NPC patients and healthy volunteers was the intensity variations of SERS spectra, particularly at 447, 496, 635, 729, 1134, 1270, and 1448 cm-1, which primarily contained signals related to proteins, nucleic acids, fatty acids, glycogen and collagen. Prominent SERS bands assignment showed. The amount of nucleic acid, collagen and fatty acids in the saliva of NPC patients was increased, while the glycogen was lower than which is in healthy volunteers. In addition, as compared to healthy group, the saliva SERS bands of NPC group showed lower intensities at 635cm-1(tyrosine) and 812 cm-1(L-serine), while higher intensities at 1619cm-1(tryptophan).The classification results based on PCA-LDA method provided a relatively high diagnostic sensitivity of 86.7%, specificity of 81.3%, for NPC identification, the ROC curve of PCA-LDA modeling was related to overall diagnostic accuracy of 83.9%.Conclusion: The results from this study demonstrated that saliva SERS analysis in conjunction with PCA-LDA diagnostic algorithms is able to distinguish NPC from normal well, it possessed promising clinical application potential for non-invasive detection of NPC.