| Objective:Most patients of gastric cancer have developed to advanced stages at the time of diagnosis and more than a half have either unresectable tumors or radiographically visible metastases.However,early detection of cancer truly depends on the discovery of specific and sensitive molecular biomarkers.Promising diagnostic patterns have recently been reported using surface enhanced laser desorption/ionization-time of flight-mass spectrometry technology(SELDI-TOF-MS).We performed surface-enhanced desorption ionization Time-of-flight mass spectrometry (SELDI-TOF-MS) using a multi-layer artificial neural network (ANN) to develop and evaluate a proteomic diagnosis approach for gastric cancer. METHODS:Serum samples from 84 gastric cancer patients and 75 controls(included 31 cases of gastric ulcer;20 cases of chronic atrophic gastritis;24 cases of healthy individuals) were randomized into training set (all 106 samples, included 54 gastric cancer patients and 52 controls) and test set(all 53 samples, included 30 gastric cancer patients and 23 controls).The diagnoses of all the gastric cancer patients and controls were identified by endoscopy and biopsy.None of the gastric cancer patients had received any form of cancer treatment before the time of gastroscopy.All serum samples were obtained on an empty stomach and stored at -80℃until used.The mixed serum samples were generated by mixing ten healthy individuals'serum and separately stored at -80℃.At first,we detected the mixed serum samples and the researched samples using SELDI mass spectrometry and CM10 protein chips.Ciphergen Proteinchip software analyzed the proteomic spectra from the mixed serum samples and computed the coefficient variation(CV).The coefficient variation was within 15% and so was satisfactory for the reproducibility of the protein profiling in the study. Biomarker Wizard Software 3.1 analyzed the gastric cancer patients and controls'serum proteomic spectra.To compare the mass peaks of the gastric cancer patients and controls,we used t-test to exam the mass peaks which were collected initially,and p<0.01 was considered statistically significant. Using a multi-layer ANN with a back propagation algorithm,we identified a proteomic pattern that could discriminate cancer from control samples in the training set.The ANN used input and output data (samples for training set) to define (learn) the interrelationships among the data.Once the ANN has been trained, it could then predict outcomes from new sets of input data .The discovered patern was then used to determine the accuracy of the classification system in the test set. RESULTS:1)The proteomic spectra from the mixed serum samples were statistically analyzed,and the CV value was 13.50%. The matric peaks whose mass-to-charge ratio were less than 2000 were excluded to avoid the interference.2)The 722 mass peaks which were found by detecting the gastric cancer patients and controls'serum samples were statistically analyzed.Among the 722 mass peaks,214 mass peaks'p values were less than 0.01. 3)Total 214 differentially expressed proteins between the gastric cancer patients and controls were identified. Among them, nine proteins(M/Z at 2175,2249,2927,3217,3236,3287,3545,6190 and 6450) were chosen to develop ANN based diagnostic model in the training set.The model could correctly discriminate all the gastric cancer patients from controls in the training set.4) The model was blindly tested with the testing set for diagnosing gastric cancer,and the results indicated three false cases found in the 30 gastric cancer patients and two false cases found in the 23 controls. The sensitivity and specificity of the diagnostic model was 90.0% and 91.3% respectively. CONCLUSION:1)SELDI-TOF-MS is very excellent to study gastric cancer.The specific biomarkers to diagnose gastric cancer can be found by using SELDI-TOF-MS.2)The diagnostic model of gastric cancer can be developed by SELDI-TOF-MS in combination with ANN. The model is a useful tool to accurately identify patients with gastric cancer.
|
PDF Full Text Request