Label-free discrimination of different stage nasopharyngeal carcinoma tissue based on Raman spectroscopy

Abstract

Abstract. The present study aimed to evaluate a label-free tissue test for the detection of nasopharyngeal carcinoma (NPC) at early and advanced stages using Raman spectroscopy (RS). RS measurements were performed to acquire high quality Raman spectra on two groups of tissue samples: One group consists of 30 NPC patients at the early stages (I-II), and the other group is 46 NPC patients at the advanced stages (III-IV). Tentative assignment of Raman bands showed specific biomolecular changes associated with cancer development. Furthermore, effective diagnostic algorithms based on principal components analysis (PCA) and linear discriminant analysis (LDA) were applied for distinguishing Raman spectra of nasopharyngeal tissues from different stages, yielding a diagnostic sensitivity of 70% and a specificity of 78%. This exploratory work suggests that RS in conjunction with the PCA-LDA algorithms provides good diagnostic ability for the early and the advanced staged NPC tissues, and RS has enormous potential for the non-invasive detection of early and advanced stage NPC. IntroductionNasopharyngeal carcinoma (NPC) is a malignant neoplasm and retains a high incidence in Southeast Asia and Southern China (1). An accurate tumor-node-metastasis (TNM) staging system is important for estimating the prognosis of malignant carcinoma, constructing treatment plans, and for predicting the 5-year overall survival rate (OS) of NPC (2). The 5-year OS rates of stage I, II, III, and IV NPC are 94, 87, 77 and 65%, respectively (3). It can be difficult to make an early diagnosis of this lesion due to the anatomical position of NPC. At present, the protocols used for NPC staging include flexible fiberoptic nasopharyngoscopy pathological biopsy, blood EB-DNA test, computed tomography (CT) of the chest, magnetic resonance imaging (MRI) of the nasopharynx and neck, bone emission computed tomography (ECT) and position emission tomography. These diagnostic processes are invasive, inconvenient, and time-consuming. Therefore, an improved diagnostic method for NPC is required, that can overcome the above disadvantages.Raman spectroscopy (RS) makes use of the inelastic scattering of light, and provides a spectrographic signature of the structure and conformation of specific molecular species, such as proteins, nucleic acids and lipids (4). At present, the application of RS is used extensively to identify cancer by analyzing excised tissue, blood plasma, saliva, urine, and seminal fluid (5-9). Tissue samples obtained timely and even continually during treatment of previously diagnosed patients, are the ideal material for confirming a precise diagnosis (10). Hitherto, RS has been widely utilized to verify diagnoses using excised samples of cancer tissues, such as esophageal cancer, lung cancer, and gastric cancer (11-13). Furthermore, Raman spectroscopy endoscopic analysis of tissues in vivo has demonstrated the potential for detecting cancers such as laryngeal cancer, esophageal cancer, lung cancer, colorectal cancer and bladder cancer (14-18). T...

The Raman Deuterium Isotope Probing method as a new assay for evaluating the cytotoxicity level of the GSK2334470 to the MCF‐7 cells

“…[ 2 ] Specifically, Raman band intensity from cancerous nasopharyngeal tissues increased at 1,268, 1,579, and 1,660 cm −1 , referring to a higher quantity of proteins, lipids, and nucleic acids in the cancerous cells. [ 22 ] Our results showed that with increase of GSK2334470 concentration, the Raman intensity decreased at 781 cm −1 (cytosine/uracil ring breathing mode of nucleotide), [ 23 ] 1,084 cm −1 (phosphodiester groups of nucleic acids), [ 24 ] 1,263 cm −1 (thymine, adenine; =C‐H bending mode of phospholipids), [ 22,25,26 ] 1,304 cm −1 (CH 2 twisting mode of lipids, adenine, cytosine), [ 27 ] and 1,655 cm −1 (C=O stretching mode of amide I proteins, α‐helix conformation, C=C stretching mode of lipids). [ 27,28 ] Such decrease in the Raman band intensities indicated that the GSK2334470 at certain concentrations could interrupt the synthesis of nucleic acid and protein in the MCF‐7 cells and reduce their intracellular quantity.…”

Real time near-infrared Raman spectroscopy for the diagnosis of nasopharyngeal cancer

“…Hence, the LVs could explain the diagnostic relevant variations rather than the significant differences in the dataset. Many studies have constructed and validated this statistical model to predict the sensitivity and specificity of Raman system in NPC model [ 15 , 23 , 24 ]. Additionally, consistent and reproducible results using this statistical model have been validated in various cancers including, gastric cancer [ 11 , 12 ], esophageal cancer [ 11 , 25 ], colon cancer [ 21 , 26 ] and NPC [ 14 ].…”

Rapid and accurate identification of colon cancer by Raman spectroscopy coupled with convolutional neural networks

“…In addition, Raman spectral differences between pathological and physiological tissues are usually subtle with apparently spectral overlapping. [12][13][14][15][16][17][18][19] Therefore, powerful and robust spectral data processing and diagnostic algorithms are needed to extract the most diagnostically significant Raman spectral features and make them accurately correlated with histopathology. Generally, multivariate statistical analysis methods are used to classify Raman spectra.…”

“…Generally, multivariate statistical analysis methods are used to classify Raman spectra. [14][15][16][17][18][19][20] However, a drawback associated with these methods is that they require lots of spectral preprocessing including smoothing, baseline correction and peak detection, which has the potential risk of smoothing the underlying Raman features and thus compromising their discriminative utility. In addition, traditional classification methods lack a powerful feature learning ability, and the correct identification rate highly depends on valid feature wavelength extraction.…”