Study on Blood Metabolomics of Early Gastric Cancer Based on Liquid Chromatography-Mass Spectrometry Technology


 Background: Metabolomics is widely used to accurately find the basic characteristics and material basis of life activities. The purpose of this study is to use metabolomics to discover biomarkers for the diagnosis of early gastric cancer.Methods: We collected the blood samples and clinical data of 63 patients with gastric cancer from the First Hospital of Jilin University, including 26 patients with advanced gastric cancer (group A), 37 patients with early gastric cancer (group B), and 18 healthy volunteers (group C). Chromatography-mass spectrometry (LC-MS) is used for detect metabolites and obtain metabolic profile. Support vector machine (SVM) is used to screen the differential metabolites with a weight of 100% from the blood sample. Total ion current diagram, principal component analysis and analysis of variance (ANOVA) are used to identify differential metabolites. PCA and the quadratic discriminant analysis were used to evaluate the similarity between samples. The receiver characteristic curve (ROC) is used to evaluate the diagnostic ability of metabolites. After the nuclear ratio of the selected metabolites is imported into the Human Metabolome Database (HMDB), the structure is identified to determine the corresponding substances, and then the verification group is used to test the accuracy of the metabolites.Results: Through LC-MS, TIC, ANOVA and PCA, differential metabolites were found in different blood samples. Cluster analysis showed similar metabolites in the three groups A, B, and C. ROC curve represented the diagnostic ability of metabolites. The different metabolites between group A and C were spermine, enterostatin, heparin sulfate, and triacylglycerol. The difference metabolites between group A, group B and group C were same as those between group A and C. The cluster analysis and ROC also showed that all four metabolites had high specificity and sensitivity in the verification group. And the results of verification group were consistent with the experimental group.Conclusion: Spermine, enterostatin, heparin sulfate, and triacylglycerol may be potential biomarkers for the diagnosis of early gastric cancer.

acetonitrile. Before the experiment, the mobile phase was ultrasonicated in a numerically controlled ultrasonic cleaner to exhaust the gas.

Data analysis
Analysis of variance (ANOVA) was used to analyze the data of the cation mode and the anion mode respectively. And the metabolites with statistical differences were screened with F value > 1 and P < 0.05 as the standards. Partial least square discriminant analysis (PLS-DA) and support vector machine analysis (SVM) respectively use VIP value > 1 and weight value equal to 100% to screen out highly speci c metabolites. Cluster analysis determined the distribution of metabolites in different groups. ROC curve showed the true positive rate on the abscissa and the false positive rate on the ordinate. Through calculating the area under the curve (AUC value), the accuracy of grouping metabolites is judged. Then the metabolic products were selected and the nucleus ratio is imported into the Human Metabolome Database (HMDB) for structural identi cation to determine the corresponding substance. Finally, the veri cation group is used to verify the differential metabolites and test their reliability and sensitivity.

Different levels of metabolites in blood samples of patients with early gastric cancer
According to clinical data, the samples were divided into 3 groups. Group A included 26 patients with advanced gastric cancer, group B included 37 patients with early gastric cancer, and group C included 18 healthy people. The mass spectrum data of the metabolites were obtained after LC-MS analysis. PCA found signi cant differences in the classi cation of metabolites among there groups ( Fig. 1A-B). Total ion current (TIC) showed signi cant differences in the peak values among the three groups ( Fig. 1C-E). Analysis of variance (ANOVA) used F value > 1 and P < 0.05 were the screening criteria, and 316 statistically signi cant metabolites were obtained (Table 1). Partial least squares discriminant analysis and analysis of variance were used to further analyze the metabolites, and the score map ( Fig. 2A) and VIP value map (Fig. 2B) were obtained. According to the VIP map, there are 5 metabolites with VIP value greater than 1. The support vector machine in the BRB-array software was used for further analysis, and a total of 4 different metabolites with a weight of 100% were found ( Table 2). Sensitivity and speci city were calculated among the groups of overall gastric cancer (group A and B), early gastric cancer (group B), and healthy people (group C) ( Table 3 and Table 4). The cluster analysis of the above four substances showed that there are different metabolites between each group (Fig. 2C). The four metabolites were imported into the human metabolomics database (HMDB), and the mass-to-charge ratio (M/Z) was used to identify their structure to determine the corresponding substances (Table 5).  ROC curve analysis shows that spermine, enterostatin, heparin sulfate, and triacylglycerol had good diagnostic ability in overall gastric cancer (group A and B), healthy people (group C), and early gastric cancer ( Fig. 3A, Table 6, Table 7). Further the bar graph showed the mean values of 4 different metabolites in the three groups, and the results show that: 4 metabolites have signi cant statistical differences in each group (Fig. 3B).  Veri cation of 4 metabolites The samples of the veri cation group were divided into three groups: A (advanced gastric cancer), B (early gastric cancer), and C (healthy volunteers). The blood samples of 24 patients in the three groups were processed by LC-MS to obtain mass spectrum data and then clustered analysis. The analysis results were consistent with the experimental grouping (Fig. 4A). To ensure the reliability of the veri cation experiment results, the experimental results were tested for sensitivity, speci city (Table 8, Table 9) and ROC curve analysis ( Fig. 4B-C, Table 10, Table 11). The results proved that the veri cation group experiment has good reliability.

Discussion
Gastric cancer is one of the common malignant tumors with high morbidity. According to statistics in 2018, there were 1.034 million new cases worldwide, and about 783,000 patients died, accounting for 8.2% of the total deaths 12 . However, the prognosis of advanced gastric cancer and early gastric cancer are completely different. Due to the lack of speci city symptoms of early gastric cancer and insu cient screening methods, more than two-thirds of patients have been at the middle and advanced stages of gastric cancer at the rst diagnosis, resulting in only 30% of 5-year survival rate 13 . However, the 5-year overall survival rate after accurate diagnosis at an early stage and timely radical resection of gastric cancer exceeds 90%. At present, the diagnosis of gastric cancer mainly relies on imaging examination, gastrointestinal endoscopy, and histopathology. However, early gastric cancer has no obvious symptoms or signs. Conventional ultrasound, CT and other imaging detection methods cannot accurately diagnose. Besides, there are no obvious general changes in the early gastric cancer lesions. Therefore, conventional endoscopy is less sensitive to early gastric cancer, with a detection rate of about 7.3% 14 . At the same time, endoscopy is invasive, prone to cross-infection and relatively expensive. Due to its high limitations, it cannot be used as a tool for large-scale gastric cancer screening. Therefore, it is imperative to nd more sensitive early gastric cancer diagnostic Spermine is a more common type of polyamines (PAs). It is widely involved in many cell activities, including cell growth, proliferation, differentiation, amino acid and protein synthesis, and DNA damage 17,18 . Studies have found that changes in the content of polyamines are accompanied by the occurrence and development of tumors, which proved that polyamines are involved in tumor transformation and the regulation of tumor microenvironment 19 . The content of polyamines in proliferating cells is signi cantly higher than that in slow proliferation. Therefore, the content of polyamines is higher in many tumor tissues than that in corresponding normal tissues 20,21 . Besides, the biosynthesis, catabolism, and membrane transport of polyamines in cells are strictly regulated. Therefore, changes in polyamine content also represent abnormal cell membrane functions. Rupesh haturvedi found that Helicobacter pylori strains increase the level of spermine oxidase, which decomposes spermine to increase the level of its metabolite hydrogen peroxide, resulting in DNA damage, and ultimately leads to gastric cancer occur 22 . In his study, we found signi cant different levels of spermine in blood samples between early gastric cancer and healthy people. The results are the same as previous studies, providing a new scienti c basis for biomarker screening and drug targets for early gastric cancer.
Heparan sulfate, also called heparan sulfate, is composed of glucosamine (GLcN) and glucuronic acid (GLcA) or iduronic acid (IdocA). Its function is modi ed by some enzymatic reactions in the synthesis process 23 . Heparan sulfate combines with the core protein on the cell membrane to form the heparan sulfate proteoglycan (HSPG) in the tumor extracellular matrix and basement membrane. It plays an important role in the regulation of in ammation, cell growth, apoptosis, infection, tumor development and other different pathophysiological processes 24 . At the same time, heparin sulfate can also be used as a carrier of cell nutrition and growth factors, which can transport cellular nutrients into cells, and can also transport growth factors without membrane receptors, thereby affecting the signal activation of downstream transduction pathways 25 , Barbara Bartolini systematically described the role of heparin sulfate in the cancer occurrence and development 26 . Other studies have shown that heparin sulfate plays an important role in the embryonic development of the stomach: HS is necessary for the differentiation of the front stomach squamous epithelium and the basal cells. HS activates the FGF in epithelial cells by promoting the combination of FGF and FGFR, which controls the development of glands. HS can also limit the proliferation of gastric interstitial smooth muscle [27][28][29] . This shows that heparin sulfate is involved in the embryonic development of the stomach and has a potential connection with gastric diseases, which supports the results of this study. In this study, it was found that the level of heparin sulfate in patients with gastric cancer rst decreased and then signi cantly increased, which is consistent with previous studies. There is also evidence that heparin sulfate as a carrier can participate in the process of transporting polyamines from outside of the cell to the inside of the cell 30 . Whether it is related to the change in spermine content is worthy of our further study.
Triacylglycerol is a component of lipids, which is a lipid formed by glycerol and 3 fatty acids. Under normal circumstances, the triglycerides in the plasma maintain a dynamic balance. However, the lipid consumption and lipid metabolite levels of gastric cancer patients increase. Therefore, the plasma lipid level of gastric cancer patients decreases and the lipid level can be used as a diagnostic biomarker of gastric cancer 31,32 . In this study, we found that the decrease of lipid metabolites such as triacylglycerol in patients with gastric cancer indicates that the lipid metabolism of patients with early gastric cancer has changed signi cantly. Although there are no reports about the relationship between gastric cancer and triacylglycerol, triacylglycerol serves as a kind of Important ester compounds, may involved in life metabolism, and are likely to have some potential connection with gastric cancer. Therefore, the results of this study provide new insights in this regard. In addition, studies have found that the lipid metabolism process of tumor cells affects the oncogenic signal transduction pathway. And lipids participate in the immune regulation of tumor cells, preventing tumor cells from the destruction of the body's immune system 33 . All the above need to be further studied.
Enterostatin is a pentapeptide released by proenzyme during fat digestion. In addition to the pancreas, intestinal statin immune response cells are also present in the gastric antrum and the proximal small intestine. Enterostatin can selectively reduce fat intake, reduce insulin secretion, and increase energy expenditure by activating brown adipose tissue during high-fat feeding. At present, there are many studies about enterostatin in the treatment of diabetes, but there are few reports related to the relationship between enterostatin and gastric cancer 34,35 . Enterostatin is an important part of the normal metabolism of the human body. In this study, we found that enterostatin in patients with gastric cancer is lower than that in healthy people, indicating that enterostatin may affect gastric cancer occurrence and development, which providing a new way for the diagnosis and therapy of gastric cancer. The mass spectrum data of the metabolites were obtained after LC-MS analysis. PCA found signi cant differences in the classi cation of metabolites among there groups ( Figure 1A-B). Total ion current (TIC) showed signi cant differences in the peak values among the three groups ( Figure 1C-E). Partial least squares discriminant analysis and analysis of variance were used to further analyze the metabolites, and the score map ( Figure 2A) and VIP value map ( Figure 2B) were obtained. According to the VIP map, there are 5 metabolites with VIP value greater than 1. The support vector machine in the BRB-array software was used for further analysis, and a total of 4 different metabolites with a weight of 100% were found ( Table 2). Sensitivity and speci city were calculated among the groups of overall gastric cancer (group A and B), early gastric cancer (group B), and healthy people (group C) ( Table 3 and Table 4). The cluster analysis of the above four substances showed that there are different metabolites between each group ( Figure 2C). ROC curve analysis shows that spermine, enterostatin, heparin sulfate, and triacylglycerol had good diagnostic ability in overall gastric cancer (group A and B), healthy people (group C), and early gastric cancer ( Figure 3A, Table 6, Table 7). Further the bar graph showed the mean values of 4 different metabolites in the three groups, and the results show that: 4 metabolites have signi cant statistical differences in each group ( Figure 3B).

Figure 4
The blood samples of 24 patients in the three groups were processed by LC-MS to obtain mass spectrum data and then clustered analysis. The analysis results were consistent with the experimental grouping ( Figure 4A). To ensure the reliability of the veri cation experiment results, the experimental results were tested for sensitivity, speci city (Table 8, Table 9) and ROC curve analysis ( Figure 4B-C, Table 10, Table 11). The results proved that the veri cation group experiment has good reliability.