Raman Spectroscopy Study For Early Screening And Metastasis Prediction Of Nasopharyngeal Carcinoma

Nasopharyngeal carcinoma(NPC)is a highly aggressive and metastatic malignant tumor in the head and neck,which is highly prevalent in the southern provinces of China,such as Fujian,and is a serious threat to people’s life and health.Early screening and metastasis prediction are two key factors for improving the overall survival rate of NPC.The traditional nasopharyngoscopy combined with the biopsy pathological analysis method has problems such as time-consuming operation,invasive,subjective dependence,and difficulty in detecting early tiny and hidden local lesions.In addition,a standard method for metastasis prediction has not been established clinically.Therefore,it is of great clinical value to explore a new method for rapid,objective and accurate early screening and metastasis prediction of NPC.Raman spectroscopy is a rapid and sensitive optical detection technique that can probe the structural information of molecules and is expected to achieve early cancer detection at the molecular level.According to the urgent need in NPC radiotherapy,this thesis carried out the study of Raman spectroscopy for early screening and metastasis prediction of NPC.The main study work is as follows:1.Surface-enhanced Raman scattering(SERS)spectroscopy was used for the label-free detection of circulating cell-free DNA(cf DNA)in plasma from NPC patients at different clinical stages for the purpose of early screening and staging of NPC at the genetic level.Firstly,cf DNA was extracted from plasma by the "magnetic bead method";then,high-quality and reproducible cf DNA-SERS spectra were obtained by silver nanocolloid and a super-hydrophobic platform;finally,the cf DNA-SERS spectra from 90 stage I,90 stage II,90 stage III,90 stage IV NPC and 90 healthy subjects were analyzed by both principal component analysis-support vector machine(PCA-SVM)and principal component analysis-linear discriminant analysis(PCA-LDA)methods.The results showed that the discrimination accuracies of PCA-SVM method were 98.9%,84.4%,86.7%,90.0% and 93.3% for the normal,stage I,II,III and IV NPC groups,respectively,which were better than those of PCA-LDA(94.4%,73.3%,80.0%,84.4% and 92.2%).This study provides a new optical method based on cf DNA-SERS spectroscopy combined with PCA-SVM for early screening and staging detection of NPC.2.The SERS spectroscopy was applied for the label-free detection of total circulating proteins in plasma from NPC patients at different clinical stages for the purpose of early screening and staging of NPC at the protein level.First,total plasma proteins were rapidly purified by cellulose acetate membrane;then,high-quality and reproducible protein SERS spectra were obtained using silver nanocolloid and a super-hydrophobic SERS platform;finally,the plasma protein SERS spectra from 120 stage I,249 stage II,291 stage III,279 stage IV NPC and 225 healthy subjects were analyzed by convolutional neural network(CNN),PCA-SVM and PCA-LDA statistical methods.The results showed that the discrimination accuracies of CNN-based deep learning for the normal,stage I,II,III and IV NPC groups were 100%,96%,96%,100%and 100%,respectively,which were better than those of PCA-SVM(100.0%,91.7%,96.0%,97.3% and 99.3%)and PCA-LDA(100.0%,70.8%,72.9%,and 77.2% and87.0%),and also better than the discriminant results obtained from cf DNA-SERS method in the previous chapter.This study suggested that blood protein SERS spectroscopy combined with deep learning algorithm has great potential to achieve more accurate early screening and staging detection of NPC.This study provides a new optical method based on plasma protein SERS spectroscopy combined with CNN deep learning for further facilitating early screening and staging detection of NPC.3.The SERS spectroscopy was applied for the label-free detection of circulating proteins in pre-treatment plasma from NPC patients with different prognosis for the purpose of individualized and precise metastasis prediction in locally advanced NPC.First,total plasma proteins were rapidly purified by cellulose acetate membrane;then,non-polarized,right-handed circularly polarized and left-handed circularly polarized laser lights were used to excite plasma protein SERS spectra,respectively;finally,the protein SERS spectra from NPC patients with metastatic and non-metastatic prognosis were analyzed by PCA-LDA method.The results showed that the prediction accuracy of left-handed circularly polarized SERS spectra for predicting the occurrence of metastasis and non-metastasis after NPC treatment was 85%,which was better than that of right circularly polarized(75%)and unpolarized SERS spectra(70%).This may be attributed to the fact that protein molecules and most amino acids are left chiral,and the chiral differences associated with NPC metastasis can be more sensitively detected using the left-handed circularly polarized laser.This study reveals that the left-handed circularly polarized laser excited protein SERS spectra has great potential to be a new method for individualized metastasis prediction in locally advanced NPC.4.A rapid,convenient and label-free method for plasma detection based on shifted excitation Raman difference spectroscopy(SERDS)technique was developed and applied to the early screening and prognosis prediction of NPC.First,high-quality plasma Raman spectra without autofluorescence background interference were obtained using the self-developed SERDS algorithm;then,PCA-LDA statistical method was performed on plasma SERDS spectra between NPC patients and healthy volunteers,between the early and advanced NPC patients,and between the metastatic and non-metastatic NPC patients,respectively.The results showed that the accuracies of the plasma SERDS method were93.3% for the discrimination between NPC and normal group,90.0% for the discrimination between early-stage and advanced NPC group,and 93.3% for the discrimination between metastatic and non-metastatic NPC group.This study reveals that the plasma SERDS method is promising to be a rapid,convenient and objective tool for early screening and prognosis prediction of NPC.5.A SERDS technique-based nasopharyngeal endoscopic detection system was developed and applied to the real-time,non-invasive,and in vivo detection of NPC.First,the high-quality tissue Raman spectra without autofluorescence background were successfully acquired using SERDS technique;then,PCA-LDA statistical analysis was performed on the SERDS spectra of nasopharyngeal tissues from NPC patients and healthy volunteers.The results showed that the discrimination accuracy between NPC and normal group was 100%,which was much higher than that of polynomial fitting Raman spectra(82.1%)and raw Raman spectra(75.0%).Further,the SERDS method was applied to discriminate the early-stage and advanced NPC,obtaining a discrimination accuracy of 81.0%.This study innovatively combined SERDS technique with an endoscopic system for in vivo,non-invasive,and real-time detection of NPC,facilitating the translation of Raman spectroscopy technology into clinical applications.In this paper,both SERS and SERDS techniques were used for label-free detection of blood and tissue samples,and were analyzed by PCA-LDA,PCA-SVM,and CNN based deep learning method for early screening,staging detection and metastasis prediction studies of NPC.The results of this exploratory study showed that our developed new methods of cf DNA-SERS spectroscopy,protein SERS spectroscopy,circularly polarized protein SERS spectroscopy,plasma SERDS spectroscopy,and tissue SERDS spectroscopy,have great potential to facilitate early,rapid,accurate and objective detection of NPC.