To establish a method of serum detection by Raman spectroscopy for the diagnosis of gastric cancer
10.3760/cma.j.cn114452-20220621-00358
- VernacularTitle:建立拉曼光谱技术检测血清的方法用于胃癌诊断的研究
- Author:
Haiyan HE
1
;
Yang ZHANG
;
Yunxia WANG
;
Guorong HUANG
;
Yu XIONG
;
Mengya LI
;
Fengxin XIE
;
Weiling FU
Author Information
1. 陆军军医大学第一附属医院检验科，重庆 400038
- Keywords:
Spectrum analysis, Raman;
Multivariate statistical analysis;
Gastric carcinoma;
Unlabeled detection
- From:
Chinese Journal of Laboratory Medicine
2022;45(8):852-858
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To establish a method of serum detection by Raman spectroscopy for the diagnosis of gastric cancer.Methods:Between April and November 2019, 110 patients with gastric cancer [73 males, 37 females, age (57.4±10.3) years] and 74 patients with colorectal cancer [48 males and 26 females, aged (58.3±12.2) years] were collected at the First Affiliated Hospital of Army Military Medical University, along with 100 healthy subjects [59 males and 41 females, aged (55.6±10.61) years] during the same period. Fasting venous blood serum samples were collected from the subjects. A Raman spectrometer XploRA PLUS was used in this experiment, with an excitation light source of 532 nm, a field of view of 100 times, and a spectrum range of 200-2 000 cm -1, etc. The serum samples were detected by nondestructive and non-contact rapid detection, and the Raman spectra of serum samples were collected. Using the Raman spectrum acquisition and processing supporting software LabSpec6 to smooth, baseline, and normalize the obtained Raman spectrum. Multivariate statistical analysis software SIMCA14.1 were applied to import and analyze the obtained Raman spectrum data by principal component analysis (PCA), orthogonal partial least squares discriminant analysis (OPLS-DA), and other methods for statistical analysis. An operating characteristic curve (ROC) was constructed to evaluate the model analysis effect between serum samples of healthy people and those with gastric cancer. Serum samples from the colorectal cancer group were used to verify the reliability of the model. Results:Six Raman peaks with good repeatability were detected in serum samples in health and gastric cancer group, and peaks were located at 1 001.17, 1 154.63, 1 337.89, 1 446.85, 1 515.33, and 1 658.34 cm -1, respectively. Raman intensities at six Raman peaks were significantly different between healthy and gastric cancer groups. At the displacement of 1 001.17, 1 154.63, and 1 515.33 cm -1, the Raman intensity in the healthy group was higher than that in the gastric cancer group. At 1 337.89, 1 446.85, and 1 658.34 cm -1 displacement, the Raman intensity of the gastric cancer group was higher than that of the healthy group. An OPLS-DA model was constructed to analyze the serum samples of the healthy group and the gastric cancer group. In the model, R 2 is the fitting power, and Q 2 is the predictive ability. The closer the values of R 2 and Q 2 are to 1, the better the performance of the model, and the obtained model's R 2X(cum)=0.809, R 2Y(cum)=0.819, Q 2(cum)=0.758. ROC characteristic curve was drawn based on the OPLS-DA model. The area under the curve (AUC) of the gastric cancer group was 0.998. Six peaks with good repeatability were detected in the serum Raman spectra of gastric cancer stage Ⅰ, Ⅱ, Ⅲ, and Ⅳ, which were located at the displacement of 1 001.85, 1 155.07, 1 338.36, 1 445.75, 1 515.92, and 1 657.68 cm -1, respectively, and at the displacement of 1 155.07 and 1 515.92 cm -1. The Raman intensity of gastric cancer stage Ⅳwas significantly higher than that of gastric cancer stages Ⅰ, Ⅱ, and Ⅲ. Conclusions:According to the model reliability verification, the healthy group, gastric cancer group and colorectal cancer group can also be effectively separated based on OPLS-DA results; it showed a good performance in separating the healthy group from the gastric cancer group. It is possible to detect serum samples from healthy people and gastric cancer patients unlabeled by combining Raman spectroscopy and the OPLS-DA method in multivariate statistics.
