Screening Of Tumor-associated Antigens Based On Human Proteomic Chips And Their Autoantibodies Diagnostic Value In Gastric Cancer

Gastric cancer（GC）is one of the most common malignant tumors in the world.Because of the early symptoms of gastric cancer,most patients are in the middle and late stage when diagnosed,resulting in a low five-year survival.Therefore,to find an efficient and economical early diagnosis method for GC is the key to reduce the morbidity and mortality of GC.In recent years,many studies have shown that tumor associated antigen（TAA）is abnormally expressed in tumor patients.Compared with TAAs,anti-TAAs autoantibodies can be more stable and persistent in patients’serum,which has the potential of becoming tumor serum markers.To screen a panel of autoantibodies with high diagnostic value for GC by HuProt^TM human whole proteomic chip is of great significance to improve the early diagnosis rate of GC.PurposeIn this study,we used HuProt^TM human whole proteomic chip to screen potential GC TAAs,and then used indirect ELISA experiments to detect the expression level of anti-TAAs autoantibodies in large samples,and to evaluate the diagnostic value of these autoantibodies for GC.Methods1)Screening candidate TAAs in GC by HuProt^TM human whole proteomic chipBased on HuProt^TM human whole proteomic chip,autoantibodies were detected in the serum of 10 GC patients and 10 healthy controls（HCs）.By nonparametric test（SNR of gastric cancer group>SNR of healthy control group,P<0.05）,Fold change（average SNR of gastric cancer group/average SNR of healthy control group>1.2）,positive rate difference（positive rate of gastric cancer-positive rate of healthy control>80%）and histogram analysis,candidate TAAs with potential diagnostic value were screened.2)The titer of anti-TAAs autoantibodies in serum was detected by ELISAThe ELISA was used to detect the titer of candidate anti-TAAs autoantibodies in serum samples of 80 GCs and 80 HCs.The difference of anti-TAAs autoantibodies titer between GCs and HCs was tested by nonparametric method and the diagnostic value of autoantibodies was evaluated by ROC curve analysis.AUC>0.5 and P<0.05was used to screen autoantibodies as an inclusion criterion.Then,the serum samples of 192 GCs,128 BGDs and 192 HCs were detected by ELISA,the difference of anti-TAAs autoantibodies titer between GCs and HCs was tested by nonparametric method and the diagnostic value of anti-TAAs autoantibodies was evaluated by ROC curve.3)The establishment and validation of GC diagnosis modelBased on the large sample size of cohort 2 as the training set to establish the model,and the small sample size of cohort 1 as the testing set to test the model.On the basis of the titer of anti-TAAs autoantibodies verified above,the GC diagnosis model was constructed by Logistic regression analysis,recursive partitioning approach and support vector machine.Through ROC analysis,the diagnostic value of each model was evaluated according to AUC,sensitivity and specificity,and finally the optimal GC diagnosis model was determined.4)The specificity of four autoantibodies in detecting GCThe titer of autoantibodies against MFGE8,NRAS,PTP4A1,RRAS2 was measured in serum of 80 HCCs,80 ECs,80 LCs,and 80 HCs by ELISA.Nonparametric test was used to compare the differences in the titer of autoantibodies between various cancer groups and HCs,so as to verify the significance of these autoantibodies in gastric cancer Specificity.5)The diagnostic value of optimal model in different clinical subgroups of GC and BGDsAccording to the clinical characteristics,GC cases were divided into different clinical subgroups and substituted into the model to evaluate the diagnostic value of the optimal model for different clinical subgroups of GC,and then the model was substituted into BGDs to evaluate the ability of the model to identify BGDs.Results1)A total of 11 TAAs were screened by Hur Prot^TM human whole proteomic chip,including F8,MFGE8,NRAS,PTP4A1,RRAS2,INPP5A,RHOG,RAC1,TMEM243,SRARP and RGS4.2)The titer of autoantibodies was detected in cohort 1 by ELISA.Finally,the titer of 6 anti-TAAs autoantibodies（anti-F8 autoantibody,anti-MFGE8 autoantibody,anti-NRAS autoantibody,anti-PTP4A1 autoantibody,anti-RRAS2 autoantibody and anti-INPP5A autoantibody）in GCs was significantly higher than those in HCs.The AUC was 0.59-0.75.Then,the titer of 6 anti-TAAs autoantibodies was detected in cohort 2 by ELISA.Finally,the 5 anti-TAAs autoantibodies（anti-F8 autoantibody,anti-MFGE8autoantibody,anti-NRAS autoantibody,anti-PTP4A1 autoantibody and anti-RRAS2autoantibody）were selected as potential diagnostic markers of GC,and their AUC ranged from 0.61-0.80.3)Three methods were used to combine the above five anti-TAAs autoantibodies to construct GC diagnosis model.By comparing the AUC,sensitivity,specificity,accuracy rate and stability of each model,we finally determined that the Logistic regression model had the highest diagnostic value for GC.The model included 4 anti-TAAs autoantibodies（anti-MFGE8 autoantibody,anti-NRAS autoantibody,anti-PTP4A1 autoantibody and anti-RRAS2 autoantibody）.In the training set,AUC was0.87（95%CI:0.83-0.90）,sensitivity was 70.8%,specificity was 85.9%.In the testing set,AUC was 0.83（95%CI:0.76-0.90）,sensitivity was 66.3%,specificity was 85.0%.4)The results of GC specific test of 4 autoantibodies showed that:Only the titer of anti-PTP4A1 autoantibody in HCCs was higher than that in healthy controls.Moreover,the titer of anti-NRAS autoantibody in healthy controls was higher than that in LCs,anti-RRAS2 autoantibody in healthy controls was higher than that in HCCs.However,the decreased anti-TAAs autoantibodies are not suitable as tumor markers.Therefore,the results showed that the 4 autoantibodies had strong GC specificity.5)The Logistic model was used to analyze different clinical subgroups of GC,and the results showed that there was no significant difference in the diagnosis of GCs with different age,gender,clinical stage,T stage,lymph node metastasis and tumor differentiation（P>0.05）.The AUC in the differential diagnosis of GCs and BGDs was0.58（95%CI:0.52-0.65）,and the sensitivity and specificity were 16.4%and 85.4%,respectively.Conclusions1)Five TAAs（F8,MFGE8,NRAS,PTP4A1 and RRAS2）were determined after screening by protein chip technology and verified in large samples by ELISA.The autoantibodies of these five TAAs may become potential markers for the diagnosis of GC.2)Logistic regression model based on autoantibodies of four TAAs（MFGE8,NRAS,PTP4A1 and RRAS2）has high diagnostic value for GC and also has a certain discrimination effect for BGDs.3)The final four autoantibodies have strong GC specificity.