| 1 Background and aimsEsophageal cancer (EC) is one of the six most common malignant tumors worldwide. The remarkable geographical distribution is the striking characteristic for EC. The ratio for EC incidence between the high- and low-risk areas could be as high as 500:1. Moreover, EC has a very poor prognosis. The five-year survival rate for the advanced EC is only 10% but for early EC is about 90%. Linzhou City (formerly Linxian County) and its neighboring counties in Henan Province, northern China has been well known as the highest incidence area for EC in the world. Another epidemiological feature for EC is that it is accompanied by the gastric-cardiac adenocarcinoma (GCA) in the same high incidence area (HIA). EC and GCA remain the leading causes of cancer-related deaths in the special area. In clinic, 80% of those patients diagnosed as EC or GCA has been in advanced stage. The reasons for this stable incidence pattern are as follows. Firstly, the exact etiological factors for EC and GCA have not been identified. Secondly, the mechanisms of carcinogenesis for esophageal and gastric cancer are not yet very clear. Thirdly, useful reagents for chemoprevention, sensitive biomarkers for early detection and high-risk subject screening are not available yet.It has been accepted that human esophageal carcinogenesis is a multistage and progressive process. The early indicator for the subject predisposed to EC is epithelial cell hyperproliferation, morphologically manifested as base cell hyperplasia (BCH), esophageal Dysplasia (EDYS), and carcinoma in situ (CIS), which have been considered as precancerous lesions of EC. GCA is a new occurring subject, which requires more research to elucidate its mechanisms of carcinogenesis. The pathogenesis and molecular changes in early stage of carcinogenesis and related risk factors have not been well characterized. Intestinal metaplasia (IM), gastric cardia dysplasia (GDYS) and chronic atrophy gastritis (CAG) were considered as the precancerous lesions for GCA. Alterations of special genes (such as p53 and Rb) in esophageal and gastric cardia carcinogenesis have been documented. However, genetic changes may not reflect the stage and progression of disease directly and objectively. Recently, we have focused on the proteomic changes of EC by using two-dimensional electrophoresis (2-ED) and matrix-assisted laser desorption/ionization time of flight-mass spectrometry (MALDI-TOF-MS) and found three proteins which were associated with stress response. There are several inherent disadvantages (including a low-sensitivity for identifying low-abundance and low-molecular weight proteins, et al) with regard to the above technologies. Recently, a novel proteomics technique, surface-enhanced laser desorption/inionation-time of flight-mass spectrometry (SELDI-TOF-MS), has shown its potency and efficacy overruling the afore-mentioned disadvantages. It can not only offer a high throughput protein profiling sensitively but also provide a protein pattern to discriminate the patients from healthy subjects coupling with bio-information. To develop serum protein patterns for identifying esophageal and gastric cardia precancerous and cancerous lesions and further to explore the mechanisms of esophageal and gastric cardia carcinogenesis, we employed the SELDI-TOF-MS technique coupling with the decision tree classification algorithm to determine the changes of serum proteomics in training and test group of subjects with esophageal and gastric cardia precancerous and cancerous lesions at high- and low-incident areas for EC and GCA.2 Material and Methods2.1 SubjectsThe subjects enrolled in this study were from Linzhou and nearby counties in Henan, the HIA and from Fanxian in Henan, the LIA for EC and GCA. All the subjects were confirmed by endoscopies and pathological morphology. The detailed information described as follows: ?210 symptom-free inhabitants and 30 EC and 15 GCA patients from HIA were selected in the present study. Of these 255 subjects, there are 63 normal inhabitants(male 22, female 41,31 to 69 years of age with a mean 47 years), 40 BCH subjects(male 15, female 25, 38 to 72 years of age with a mean 51 years, grade I 32, grade II 8, gradelll 0 ), 27 EDYS subjects (male 13,female 14,35 to 70 years of age with a mean 49 years, grade I 19, grade II 8, grade III 1), 57 CAG subjects(male 24, female 33, 30 to 69 years of age with a mean 52 years, low-grade 36, middling grade 16, serious grade 5), 23 DYS patients for gastric-cardia (male 13, female 10, 28 to 68 years of age with a mean 51 years, grade I 15, grade II 6, gradelll 2), 30 EC patients (male 21, female 9, 42 to 72 years of age with a mean 59 years, early stage 2, advanced stage 28, including 25 patients for surgical control experiment) and 15 GCA patients (male 10, female 5, 32 to 75 years of age with a mean 56 years, early stage 2, advanced stage 13, including 9 patients for surgical control experiment). (2)125 symptom-free inhabitants from LIA were enrolled in this study. Of these 125 subjects, there are 60 normal inhabitants(male 27, female 33, 30 to 70 years of age with a mean 55 years ), 56 BCH patients(male 27, female 29, 30 to 69 years of age with a mean 53 years, grade I 50, grade II6, gradelll 0 ), 2 EDYS subjects(male 2, female 0, 36 and 41 years, grade I 2, grade II 0, gradelll 0 ), 6 CAG patients(male 5, female 1, 32 to 57 years of age with a mean 48 years, low-grade 5, middling grade 1, serious grade 0), 1 GDYS subject (female, 44 years, grade I ). All subjects were confirmed without acute infection, allergic diseases and autoimmune disease. 5 ml fast blood from each subject was withdrawn and stood at room temperature for 30 min, and then centrifuged at 2000r/min for 20 min. The serum was aliquot into several centrifuge tubes at 100 u 1 and was frozen in liquidnitrogen and followed by storing in -80 °C icebox when the screening was done. The sera of cancer patients were withdrawn at the day of surgery and at the 7th day after surgical operation for surgical control experiment. 2.2 Methods2.2.1 Preparation of Serum Samples and protein chipSerum samples were thawed in ice and centrifuged at 4 °C at 10000 rpm/min for 5 min. A total of 10 u L of each sample was diluted to 20 u L with U9 buffer(9M Urea, 2% CHAPS, 50mM Tris-HCl, PH 9.0) and vortex at 4°C for 30 min. 10 u L of the supernatants was mixed with 110 y- L binding buffer (50mM NaAC at PH3.5), and vortex for 5min. An eight-spot weak cation exchange chip(WCX2) was washed with 150 V L of the binding buffer twice. Then 100u L of the U9-treated sample was applied onto a protein chip array and incubated for 1 h at 4 °C on a shaker. After washing with the same binding buffer twice, followed by a quick water rinse, 1 nL of saturated sinapinic acid (SPA) solution was applied onto each spot and allowed to air-dry.2.2.2 Data collecting and analysis of bioinformaticsChips were subject to Protein Biological System II mass spectrometer reader (PBSII, Ciphergen Biosystems, Inc). Data were collected twice by averaging 65 laser shots with an intensity of 170 and 180, a detector sensitivity 6 and with a highest mass of 50 000 Da and an optimized range of 2 000-20 000 Da. Mass accuracy was calibrated to less than 0.1% using the all-in-one peptide molecular mass standard (Ciphergen Biosystems, Inc). The coefficient of variance for peak height was less than 15%. The differences of the intensity for each labeled peak between the two subjects were analyzed by using T test.A decision tree classification model (DTCM) was built up with Biomarker Pattern's Software (BPS, Ciphergen Biosystems, Inc). Training data set consisted of all samples. The DTCM was set up to divide the data set into two bins based on the intensities of peaks. At each bin a peak intensity threshold was set. If the peak intensity of a sample were lower than or equal to the threshold, this sample would go to the left-side bin. Otherwise, the sample would go to the right-side bin. The process would go on until asample entered a final bin. Peaks selected by the process to form the model were those that yielded the least classification error when they were combined to use. The approach of 10-fold cross validation was used to test the model. Specificity and sensitivity were respectively calculated as the proportion of the number of disease samples correctly identified to the total number of control samples. Positive predictive value gives the probability of disease if a test result is positive. 3 Brief results and conclusions 3.1 Data analysis for subjects at HIA3.1.1 Data analysis for EC and normal (NOR) control samplesTwo proteins with M/Z of M2942.15 and Ml5953.4 were chosen to set up the DTCM and to distinguish EC from NOR control samples with a sensitivity of 60%, a specificity of 85% and a positive predictive value of 64%. It suggests that the protein pattern will probably become a novel screening tool for EC at HIA.3.1.2 Data analysis for GCA and NOR control samplesTwo proteins with M/Z ofM15957.14 and M7944.57 were selected to set up the DTCM and to differentiate the GCA from NOR subject with a sensitivity of 67%, a specificity of 94% and a positive predictive value of 71%. It suggests that the two proteins may be potential screening biomarkers for GCA in high-risk population at HIA.3.1.3 General Data analysis for EC and GCA and NOR control sample 3.1.3.1 General resultsThe serum level of two proteins with respective M/Z of M15953.4 and M15957.14 with two positive charges was the highest in EC patients among all the esophageal precancerous and cancerous subjects and NOR control. Interestingly, these two proteins diminished pronouncedly 7 days after surgical removal of cancer entity. The same phenomena were observed in GCA subjects. The protein with M/Z of M2941.47 is one of the surrogates of the protein with M/Z of M7944.57 during building DTCM. The serum level of proteins with M/Z of M2942.15 and M2941.4 with one positive charge was significantly lower in esophageal and gastric precancerous and cancerous subjects than those in the NOR subject at HIA, respectively. The serum level of theprotein with M/Z of M7944.5 either in EC or in GCA subjects was significantly higher than those in NOR control subjects at HIA, respectively. Interestingly, three proteins, respectively M/Z of with M2773.93, M5350.31, M5918.16, are surrogates of proteins with both M7944.5 and M2942.15 in setting up the DTCM, respectively, and one protein with M/Z of M7988.68 is the surrogate of proteins with both M15953.4 and Ml 5957.14 during building the DTCM. 3.1.3.2 Part conclusionsAccording to the results above, we conclude the following, (T)Two protein patterns for identifying respective EC and GCA may have a similar classification function for screening EC and GCA subjects.(2)Two proteins with M/Z of Ml5953.4 and Ml5957.14 may be same or very similar in bio-function . They may play critical roles in esophageal and gastric cardia carcinogenesis. They are important biomarkers in diagnosis EC and GCA. (3)The change of the protein with M/Z of M2942.15 may be an important early molecular event in esophageal and gastric cardia carcinogenesis. It is probably a key biomarker in esophageal and gastric cardia precancerous lesions.? The change of the protein with M/Z of M7944.57 is well associated with carcinogenesis of EC and GCA. It may be an important biomarker in screening EC or GCA. (5) Some similar proteomic changes were found both in esophageal and gastric cardia carcinogenesis. It suggests that there may be some similar etiologies and pathogeneses in esophageal and gastric cardia carcinogenesis at the same HIA. 3.1.4 Data analysis for EDYS, GDYS and NOR control samples.3.1.4.1 Data analysis for EDYS and NOR control samplesOne protein with M/Z of Ml3765.9 was selected to build the DTCM and to distinguish EDYS subject from NOR control with a sensitivity of 67%, a specificity of 71% and a positive predictive value of 50%. It suggests that the DTCM could not become a screening tool for EDYS in HIA until more information was obtained for it.3.1.4.2 Data analysis for GDYS and NOR control samples.One protein with M/Z of M3894.0 was selected to build the DTCM and to differentiate the GDYS from NOR control subjects with a sensitivity of 87%, a specificity of 86% and a positive predictive value of 69%. It suggests that the proteina sensitivity of 47%, a specificity of 86% and a positive predictive value of 47%. The protein pattern cannot discriminate the two group samples sensitively. It suggests that there may be some similar expressions of proteins in GCA and CAG. It provides another evidence to support the notion that CAG, which has carcinogenic potential, implicates one of the multistage processes of malignancy.3.1.11 Data analysis for GCA and GDYS control samplesTwo proteins with M/Z of M32182.9 and M2955.49 were selected to build the DTCM and to discriminate the GCA from GDYS subject with a sensitivity of 67%, a specificity of 70% and a positive predictive value of 47%, respectively. It suggests that the protein pattern may be a useful tool for following-up the inhabitants with GDYS lesions.3.1.12 Data analysis for GDYS and CAG control samplesFive proteins with M/Z M4101.69, M5000.28, M37834.8, M2050.62 and M8707.03 were selected to build the DTCM and to discriminate the GDYS from CAG subject with a sensitivity of 61%, a specificity of 81%, a positive predictive value of 54%, respectively. It suggests that the protein pattern may be a useful tool for following-up the population with gastric cardia precancerous lesions at HIA.3.1.13 Data analysis for EC subjects surgical control test.One protein with M/Z of M7988.68 was selected to build the optimal DTCM and to discriminate the EC patients before and 7 days after surgical resection with a sensitivity of 28%, a specificity of 80% and a positive predictive value of 58%. The level of the protein was significantly higher in EC than that in NOR subjects and in EC subjects 7 days after surgical treatment. So was the GCA. It suggests that theprotein may play a role in esophageal and gastric cardia carcinogenesis. 3.2 Data analysis for BCH and NOR samples at LIAOne protein with M/Z of M2053.38, with an increase serum level in BCH subjects,was selected to build the DTCM and to discriminate BCH from NOR subject with a sensitivity of 73%, a specificity of 92% and a positive predictive value of 77% at LIA. However, the difference of serum level of the protein was not significant between BCH and NOR subjects from HIA. It suggests that the protein with M/Z of M2053.38appropriate screening tool for CAG at HIA.The serum level of the protein with M/Z of M33316.6 was significantly higher in CAG subjects and was significantly lower in GDYS subject than that in NOR control subjects, respectively. The serum level of the protein was significantly higher in GCA subjects than that in GDYS subjects or in the same GCA patient with surgical treatment 7 day later, respectively. It suggests that the protein may play an important role in GCA carcinogenesis. It may possess predicting potential for evaluating efficacy of surgical regimen for GCA.3.1.7 Data analysis for EDYS and BCH control samples at HIA.One protein with M/Z of M23518.7, with an increase serum level in BCH subjects, was selected to build DTCM and to identify EDYS from BCH subjects with a sensitivity of 81%, a specificity of 88% and a positive predictive value of 81%. It suggests that the protein may be a biomarker for esophageal precancerous lesions and may be a good biomarker for differential diagnosis for esophageal precancerous lesions at HIA.3.1.8 Data analysis for EC and BCH control samplesOne protein with M/Z of M5350.39, with a low serum level in EC subjects, was selected to build a DTCM and to discriminate EC from BCH with a sensitivity of 63%, a specificity of 83% and a positive predictive value of 73%. It suggests that the protein may play a role in the carcinogenesis of BCH lesion. It may be an important biomarker for following-up the subjects with BCH at HIA.3.1.9 Data analysis for EC and EDYS control samplesFour proteins with M/Z of M2749.29, M5443.63, M2667.48 and M2077.83 were selected to set up the DTCM and to identify the EC from EDYS subjects, Both sensitivity and specificity are 67% and a positive predictive value were 69%, respectively. It suggests that these four proteins may be important biomarkers for following-up in the subjects with EDYS at HIA.3.1.10 Data analysis for GCA and CAG control samplesFive proteins with M/Z of M8230.69, M3372.63, M4832.48, M2086.63 and M2021.98 were selected to build the DTCM and to identify the GCA from CAG witha sensitivity of 47%, a specificity of 86% and a positive predictive value of 47%. The protein pattern cannot discriminate the two group samples sensitively. It suggests that there may be some similar expressions of proteins in GCA and CAG. It provides another evidence to support the notion that CAG, which has carcinogenic potential, implicates one of the multistage processes of malignancy.3.1.11 Data analysis for GCA and GDYS control samplesTwo proteins with M/Z of M32182.9 and M2955.49 were selected to build the DTCM and to discriminate the GCA from GDYS subject with a sensitivity of 67%, a specificity of 70% and a positive predictive value of 47%, respectively. It suggests that the protein pattern may be a useful tool for following-up the inhabitants with GDYS lesions.3.1.12 Data analysis for GDYS and CAG control samplesFive proteins with M/Z M4101.69, M5000.28, M37834.8, M2050.62 and M8707.03 were selected to build the DTCM and to discriminate the GDYS from CAG subject with a sensitivity of 61%, a specificity of 81%, a positive predictive value of 54%, respectively. It suggests that the protein pattern may be a useful tool for following-up the population with gastric cardia precancerous lesions at HIA.3.1.13 Data analysis for EC subjects surgical control test.One protein with M/Z of M7988.68 was selected to build the optimal DTCM and to discriminate the EC patients before and 7 days after surgical resection with a sensitivity of 28%, a specificity of 80% and a positive predictive value of 58%. The level of the protein was significantly higher in EC than that in NOR subjects and in EC subjects 7 days after surgical treatment. So was the GCA. It suggests that theprotein may play a role in esophageal and gastric cardia carcinogenesis. 3.2 Data analysis for BCH and NOR samples at LIAOne protein with M/Z of M2053.38, with an increase serum level in BCH subjects,was selected to build the DTCM and to discriminate BCH from NOR subject with a sensitivity of 73%, a specificity of 92% and a positive predictive value of 77% at LIA. However, the difference of serum level of the protein was not significant between BCH and NOR subjects from HIA. It suggests that the protein with M/Z of M2053.38may play a role in BCH genesis at LIA. The etiologies and mechanisms of the BCHdevelopment may be different in the subjects from HIA and LIA.3.3 Comprehensive data analysis for NOR subjects in HIA and LIA.The serum levels of 49 proteins, including the proteins discussed above, viz M15957.14, M15953.4, M7944.57, M9306.61 and M7988.68, were significantly higher or lower in NOR subjects between HIA and LIA. It suggests that there is a significantly difference in the basal levels of these proteins in NOR subjects from the HIA and LIA, resulting in a different susceptibility to EC and to GCA for inhabitants at HIA and LIA. The exact mechanisms need to be further investigated.
|
PDF Full Text Request