Clinical samples This study has been approved by the ethics committee of Xijing hospital. Informed consent was obtained from all participants, and the research method complies with Chinese laws and regulations. We collected the Serum samples of 139 gastric adenocarcinoma patients from Xijing hospital of digestive diseases. All the patients had primary gastric adenocarcinoma and were not on any medical treatment prior to sample collection. They were compared to a control group consisting of 156 healthy individuals, whose information was collected from the medical examination center at Xijing hospital. Healthy controls were selected via a health check, including blood tests, endoscopic examinations and diagnostic imaging. We excluded the individuals, whose diagnosis indicated requiring therapy, detailed examination, and/or observations, from the healthy control group. The specific information about the tumor stage and histological differentiation has been shown in Table 1. We have confirmed the diagnosis for all the patients by pathological examinations, and obtained the tissue samples during surgery. The anatomic stage and histological grade were determined according to the American Joint Committee on Cancer (AJCC) TNM staging classification and histological grading for carcinoma of the stomach (Seventh Edition 2010). All serum samples were collected at 6:00 in the morning before breakfast. Standard blood processing for FTICR-MS analysis was performed in the following manner: we collected whole blood by blood collection tubes and stored them at room temperature for up to 2h; we then centrifuged the samples at 4°C, 2000 rpm for 8 min; the corresponding serum was finally aliquoted into micro-centrifuge tubes and quickly frozen and stored in a -80°C freezer for further analysis. To separate the serum from the blood cells, the blood samples for the tumor marker assays were centrifuged at 1000 × g for 10 min. Serum CEA, AFP, CA19-9, and CA12-5 levels were measured using the Roche cobas Elecsys 601 Automatic electrochemiluminescence immunoassay analyzer. The cut-off values for serum CEA, AFP, CA19-9, and CA12-5 were set at 5 ng/ml, 8.1 ng/ml, 20 U/ml and 35 U/ml, respectively.
Table 1
Summary of clinical and demographic characteristics of the gastric adenocarcinoma patients and healthy controls used in this study
characteristics
|
Gastric adenocarcinoma patients
|
Healthy controls
|
Gender
|
|
|
male
|
100
|
113
|
female
|
39
|
43
|
Age(years)
|
|
|
mean ± SD
|
57.4 ± 12.2
|
55.2 ± 9.7
|
Rang
|
24–79
|
31–79
|
AJCC Anatomic stage(2010)
|
|
|
|
23
|
|
Ⅱ
|
49
|
|
Ⅲ
|
64
|
|
Ⅳ
|
3
|
|
AJCC Histologic grade(2010)
well
moderately
poor
|
23
47
69
|
|
Chemicals We have purchased Acetonitrile from Thermo Fisher, formic acid from TEDIA, Reserpine from Sigma (America), Phosphatidylcholine (PC) (34:1) from Avanti polar lipids (Alabaster AL), and Palmitoyl-L-carnitine chloride from Sigma (America). We filtered the distilled water through a Milli-Q system (Millipore, Billerica, MA).
Serum samples preparation. After thawing the serum on ice, we added acetonitrile (400 µl) to the serum (100 µl) in a 2 ml microfuge tube. The sample was then vortex-mixed for 45 seconds and stored at -20°C overnight. The mixtures were centrifuged at 14,000 rpm for 10 min at 4°C, and then 10 µl of the supernatant from each sample was vortex-mixed with 190 µl of the internal standard solution (54 ng/ml of reserpine in water; 50% acetonitrile (v/v); 0.2% formic acid (v/v)). Each sample was stored in an auto-sampler. Quality control (QC)[22] samples were prepared by mixing equal amounts of the serum samples from 30 healthy controls. All the serum samples were randomized along with 20 QC samples. The QC samples were inserted and analyzed in every 15 samples to ensure the stability and repeatability of the MS system.
FTICR-MS analysis Metabolite analysis was performed using a 9.4 T apex-ultra™ hybrid Qh-FTICR mass spectrometer (Bruker Daltonics, Billerica, MA) equipped with an electrospray ionization source. The sample was introduced into the ion source by a syringe pump at a flow rate of 3 µl/min. The mass spectrometer was operated in the positive ion mode with the capillary voltage of 4000 V, the spray shield of 3500 V, the drying gas flow of 5.0 L/min, the drying gas temperature of 180°C, the nebulizing gas flow of 1.0 L/min, the time of flight of 0.0012 s, the source accumulation of 0.05 s and the ion accumulation time of 2.5 s. The mass spectrum was acquired from 4 scans for an averaged spectrum in the m/z range of 100–1000 with 2 M acquisition size, and the max resolution was 260,000 at m/z 400. In the tandem MS experiments, argon was used as the collision gas.
Data preprocessing We obtained the Raw MS data by the Apex Control 3.0.0 software (Bruker Daltonics, Billerica, MA). To detect the peak, we executed the FTMS peaks finder algorithm with the signal-to-noise ratio set at > 5, the relative intensity at 0.2% and the absolute intensity thresholds at 10,000. For peak detection and alignment, we removed the variables having too many missing values according to the 80% principle[23], and only kept the variables with more than 80% non-zero measurement values in the control and gastric adenocarcinoma group on the peak list. The missing values were filled with the mean of the variable, and the peak intensities of all the metabolites were normalized to the total intensity of the sample [24]. After normalization, the data were then exported to the SIMCA-P v 11.5 software (Umetrics AB, Sweden) for multivariate data analysis. Unsupervised principal component analysis (PCA) was initially used to visualize the general separation. At the next step, supervised orthogonal partial least squares-discriminant analysis (OPLS-DA) was performed to model the difference between the GC patients and healthy controls. We selected Potential biomarkers based on the Variable Importance in the Project (VIP) and Z-score [Z-score =\(\frac{{X}_{i}-{\overline{X}}_{control}}{{\text{S}\text{D}}_{\text{c}\text{o}\text{n}\text{t}\text{r}\text{o}\text{l}}}\)] [25]. Furthermore, we performed an independent t-test using SPSS 16.0 for windows to confirm the significant differences of the marker metabolites between the gastric adenocarcinoma and healthy control group, with the level of statistical significance set as p < 0.05. The receiver operating characteristic (ROC) curve was employed to demonstrate the distinguishing ability of the potential marker metabolites. We calculated the area under the curve (AUC) for the ROC curve for the combination of potential marker metabolites. To identify the potential marker metabolites, we used the METLIN database.