Metabolomic profiling on plasma reveals potential biomarkers for screening and early diagnosis of gastric cancer and precancerous stages

Background: Gastric cancer (GC) remains one of the most common cancers all over the world. The greatest challenge for GC is that it is often detected at advanced stages, leading to the loss of optimum time for treatment and giving rise to poor prognosis. Thus, there is a critical need to develop effective and noninvasive strategies for early diagnosis of the disease process. Methods: In total, 82 participants were enrolled in the study, including 50 chronic superficial gastritis (CSG) patients, 7 early gastric cancer (EGC) and 25 advanced gastric cancer (AGC) ones. Metabolites profiling on patient plasma was performed using ultra-high performance liquid chromatography coupled with quadrupole-time-of-flight mass spectrometry ( UPLC-Q-TOF/MS ). Principal components analysis as well as orthogonal partial least squares-discriminant analysis was utilized to evaluate the variation on endogenous metabolites for GC patients and to screen potential biomarkers. Furthermore, the biomarker panels detected above were used to create logistic regression models, which discrimination efficiency and accuracy was ascertained by receiver operating characteristic curve (ROC) analysis. Metabolic pathways were carried out on MetaboAnalyst. Results: Totally 50 metabolites were detected differentially expressed among CSG, EGC and AGC patients. L-carnitine, L-proline, pyruvaldehyde, phosphatidylcholines (PC) (14:0/18:0), lysophosphatidylcholine (14:0) (LysoPC 14:0), lysinoalanine were defined as the potential biomarker panel for the diagnosis among CSG and EGC patients. Compared with EGC patients, 6 significantly changed metabolites, PC(O-18:0/0:0) and LysoPC(20:4(5Z,8Z,11Z,14Z)) were found to be up-regulated, whereas L-proline, L-valine, adrenic acid and pyruvaldehyde to be down-regulated in AGC patients. ROC analysis demonstrated a high diagnostic performance for metabolite panels with area under the curve (AUC) of 0.931 to 1. Moreover, the metabolomic pathway analysis revealed several metabolism pathway disruptions, including amino acid and lipid metabolisms, in GC patients. Conclusions: In this study, a total of six differential metabolites that contributed to GC and precancerous stages were identified, respectively. The biomarker panels further improve diagnostic performance for detecting GC, with AUC values of more than 93.1%. It indicated that the biomarker panels may be sensitive to the early diagnosis of GC disease, which can be used as a promising diagnostic and prognostic tool for disease stratification studies.


Introduction
The International Agency for Research on Cancer estimates that gastric cancer (GC) is the fifth most prevalent type of malignancy and the third mortality of cancer death all over the world [1]. Although surgery is the most preferred treatment, the majority of patients have recurrence after surgery, leading to a poor five year survival rate [2,3]. The growth and proliferation of GC is involved in multi-gene and multi-factor. Correa [4] proposed a theory of human intestinal-type gastric carcinogenesis initiated by normal mucosa, followed by chronic superficial gastritis (CSG), then chronic atrophic gastritis, to intestinal metaplasia, and finally by dysplasia and intestinal-type gastric cancer, which is generally accepted. More specifically, CSG and chronic atrophic gastritis are the critical stage in the development and proliferation of GC, which are accounted as important risk factors for gastric carcinogenesis. So early prevention on gastric precancerous lesions and diseases could reduce the incidence of GC. In recent decades, diagnostic methods based on endoscopic examination, pathological section and barium meal examination, have been widely applied to GC patients [5]. However, these screening methods are limited by various disadvantages, such as time-consuming, invasiveness, laborious and harmful.
Thus, establishment of a sensitive, noninvasive examination method for early detection and prognosis prediction in GC patients is of significant importance.
Metabolomics is an emerging science involving the profiling changes in small-molecular metabolites produced by a biological system under certain conditions [6]. It seems to be a very promising method for biomarker discovery due to the dynamic responses of the metabolome that reflects upstream biological processes in the body. It has several major advantages, such as the readily availability, noninvasive and high sensitivity [7]. For the past few years, metabolomics has been utilized for analysis of metabolic alterations caused by cancers and other diseases, which has led to substantial advances in early diagnosis, mechanism clarification, and discovery of biomarkers [8].
Biomarkers are small-molecular intermediates and end products of active cellular processes, forming a correlation between molecular metabolic changes and phenotype [9,10]. Therefore, they reflect alterations of the physiological state of a biological system (cell, tissue or organism) at a certain point in time. Early gastric cancer (EGC) is asymptomatic. There is no doubt that one of the greatest challenges for biomarker-related is discovering biomarkers that accurately distinguish cancer from precancerous stages, where overlapping signs and symptoms (unintentional weight loss or vague epigastric pain) make differential clinical diagnosis difficult [11]. Till now, few studies on metabolic changes for screening and early diagnosis of GC and precancerous stages have been applied in clinical practice, which are needed to be further explored.
In this work, we developed an ultra-high performance liquid chromatography coupled with quadrupole-time-of-flight mass spectrometry (UPLC-Q-TOF/MS) and biotransformationbased metabolomics profiling approach for determining GC staging and CSG. Briefly, multivariate analyses were utilized to identify differential metabolites that related to GC staging and CSG groups. On the basis of potential biomarkers, the related metabolic pathways and correlation networks were investigated and the global metabolic features were discussed.

Patient information and sample collection
The present study was approved by the ethics Committee of the People's Hospital of Yangzhong City (Yangzhong, China) and all participants provided written informed consent. A total of 86 individual patients who had CSG and GC were recruited at the People's Hospital of Yangzhong City between July and December, 2015. All of tissue specimens were examined by gastroscopic biopsy or pathological examination after surgery. According to the results from pathologic diagnosis, the 86 samples were divided into three groups, including fifty cases of CSG (mean age ± SD, 52.1 ± 7.0 years), seven cases of EGC (mean age ± SD, 66.3 ± 11.9 years) and twenty-five cases of advanced gastric cancer (AGC, mean age ± SD, 67.0 ± 9.7 years). Among them, four cases of patients who had a history of chemotherapy or surgical treatment were excluded from this study. Details of basic information of patients with CSG and GC staging are shown in Table   S1.

Sample preparation
All frozen plasma samples were thawed completely at room temperature for 3 h. Then, 100 µL plasma sample was transferred to an Eppendorf tube, and 300 µL methanol/acetonitrile solution (v/v, 1:1) containing 2-Chloro-L-phenylalanine (5 µg/mL) as internal standard was added. Mixed sample was vortex-mixed for 30 s and placed at 4 °C for 1 h. Vortex-mix again for 30 s and kept at 4 °C for 3 h to fully precipitate the protein in the plasma.

Statistical analysis
The data collected from MS were processed using Progenesis QI, and then analyzed by  in EGC compared with CSG, as shown in Table 1. Similarly, a PCA analysis was used to explore the metabolic profiling differences between the EGC and AGC patients, and the results are presented in Fig. 2C. There were no distinctive differences between EGC and AGC groups. Then, the OPLS-DA model was launched (Fig. 2D). Based on the criteria of OPLS-DA (VIP > 1 and P < 0.05), 16 statistically differentially expressed metabolic molecules in total were screened out and finally 6 metabolic molecules were identified as potential metabolite biomarkers between the two groups. The significantly changed 6 metabolites listed in Table 2. PC(O-18:0/0:0) and

Metabolites detection and identification
LysoPC(20:4(5Z,8Z,11Z,14Z)) were found to be up-regulated, whereas L-proline, L-valine, adrenic acid and pyruvaldehyde to be down-regulated in AGC patients.  The calculated results showed that the proposed biomarker panel model had AUC value of 1 (Fig. 3A), which meant that the multivariate model showed 100% discrimination power to separate EGC patients from CSG patients.
Similarly, to confirm the diagnostic potential for the early detection of GC, we examined the AUC values in stage EGC to AGC patients. As listed in the Table 4 and 100.0% at the best cut-off points (Fig. 3B, Table 3). As indicated by these results, the biomarker combinations presented herein serves not only to discriminate EGC from CSG patients, but is also capable of distinguishing stage I and II GC models with relatively high diagnostic accuracy.

Metabolic Pathway Analysis
On the basis of the detected differential metabolites, pathway analysis was performed by MetaboAnalyst 4.0 to uncover the global metabolic disorders in CSG and GC patients. Figure 4A-B presents the major impacted pathways in the CSG-EGC and EGC-AGC groups, indicated by the red and orange colors (-log(p) > 2 or impact value > 0.1). As shown in Fig. 3, the amino acid metabolism was discovered to be strikingly disturbed, including glycine, serine and threonine metabolism, valine, leucine and isoleucine biosynthesis and so on. The perturbations of central carbon metabolism (e.g., pyruvate metabolism) and lipid metabolism (e.g., glycerophospholipid metabolism, linoleic acid metabolism, alphalinolenic acid metabolism and ether lipid metabolism) were also observed. The changes of detected differential metabolites related to the abnormal metabolic pathways, providing clues for underlying the potential metabolic mechanism in GC.

Discussion
In this study, high-throughput metabolomics couple with UPLC-Q-TOF/MS technology was utilized to investigate GC-related metabolic alterations and elucidate potential diagnostic biomarkers. The present evaluation was performed on patients with CSG and two subgroups of EGC and AGC subjects to search for the correlates between the small molecule metabolites and the disease progression. Most of the metabolites identified were altered on statistically significant level, derived mainly from general biochemical pathways related to amino acid metabolism, energy metabolism and lipid metabolism.
Amino acids, as the substrates for protein synthesis, are crucial for cancer cell migration and proliferation. Previous studies have associated amino acid metabolism aberrations with cancer development [12][13]. It is involved in multiple cancers that regulate several signaling pathways, including protein synthesis, cell growth, lipid biogenesis, autophagy and so on [14] In this study, L-proline was found to be up-regulated in EGC and AGC stage, and L-valine was also found significantly up-regulated in AGC stage. High levels of proline could promote cell proliferation, energy production and resistance to oxidative stress (act as an antioxidant) [15][16][17][18]. L-valine is an essential and important functional amino acid involved in many growth and metabolic processes, and is also a glucogenic amino acid for biosynthesizing macromolecules (e.g., proteins and lipids), which are vital to the growth of cancer cell [19]. The accumulation of amino acids could ascribe to the proliferation by cancer cells, suggesting cancer transformation is linked with adaptive increases in protein synthesis [20].
Except a number of biologic functions, amino acid metabolism is also considered as an essential energy metabolism pathway of cancer cells to meet the high energy requirement [21]. For example, valine could be transformed into pyruvate for energy supply through aerobic glycolysis, resulting in the significant increase in pyruvate [22]. Thus, the upregulated of valine in GC indicates that cancer cell energy metabolism may be significantly increased during cancer progression Another important feature in GC progression was the apparent changes of lipids. It is well known that lipids play an important role at cellular and organismal levels, being the dominant structural components of biomembranes and energy storage entities [23].
Additionally, lipids participate in signal transduction and can be broken down into biologically active lipid mediators, which regulate some carcinogenic processes [24,25].
In the present study, potential biomarkers analyses revealed significant alterations in plasma LysoPC and PC. The down-regulation of them may be mainly due to the increased demand for membrane constituents during malignant transformation, cancer invasion and metastasis. Dysregulation of choline phospholipid metabolism is associated with carcinogenesis and cancer progression, which has been verified in many biomarker studies [26][27][28], including studies of GC [29]. Thus, the abnormal levels of LysoPC and PC may be considered as important biomarkers for GC patients.
A comprehensive understanding of metabolic alterations associated with cancer stages would be helpful in the development of GC. Among the all differential metabolites, there were some metabolites that showed a good potential to differentiate with AUC values more than 0.8, such as L-proline, LysoPC(14:0) (Table 4). However, a question arises as to whether one molecule has sufficient potential in GC detection. Based on the results of previous cancer biomarker studies, it can be assumed that the most efficient sample discrimination will be obtained using metabolite panels. Therefore, we built logistic regression models consisting of multiple metabolites. The AUC values (> 0.931) of multivariate ROC curve were higher than that obtained for single metabolite (Table 3) and the metabolite panels model turned out to be sensitivity enough to distinguish patients correctly. Therefore, the application of a combination strategy allows for better early diagnosis of GC and precancerous stages.
In summary, this study suggests that the biomarker panels possessed the potential value for the diagnosis of GC stages. The identified potential biomarkers and biological pathways might provide new directions for further studies in cancer growth and development. However, the strict inclusion and exclusion criteria decreased the number of recruited patients, and further research should involve the inclusion of multicenter subjects with GC of different stages, to better evaluate the accuracy of the developed models in early diagnosis.

Conclusion
In this study, UPLC-Q-TOF/MS plasma metabolomics has been successfully established for biomarker studies in GC. Differential metabolites signatures were globally depicted, and biomarker panels were defined for the diagnosis of EGC with satisfactory discrimination performance, even for AGC (AUC > 0.931). Metabolic pathway analysis indicated that changes in most potential plasma biomarkers were correlated with general biochemical pathways (amino acid metabolism and lipid metabolism), implying enhanced energy production and cell proliferation. The study highlights the potential advantages of biomarker panels in real clinical diagnostics, which can be used as a promising tool for early-stage GC diagnosis and prognosis.

Availability of data and materials
All the necessary materials can be found in the text or supplementary materials. Due to the privacy policy, the confidential data materials could only be obtained with the permission of the corresponding authors.

Consent for publication
Not applicable.