| Background and ObjectiveOvarian cancer(OC)is one of the most lethal gynecological malignancies worldwide.Due to a few specific symptoms in the early stage and a lack of reliable screening methods,the 5-year survival rate for the OC is about 30%.Therefore,it is necessary to find an effective and non-invasive early diagnosis method for OC,which is essential to improve the survival and prognosis of patients with OC.Autoantibodies against tumor associated antigens(TAAs)are more stable and have a relatively higher titer in sera from OC and can be detected in the early stage prior to clinical symptoms.Therefore,these make autoantibody potential as a serum marker for the early detection of OC.The objectives of this study were to discovery potential OC-related TAAs through whole human proteome microarray and detect the corresponding autoantibodies using enzyme-linked immunosorbent assay(ELISA),and further evaluate the diagnostic value of candidate autoantibodies for OC.Based on the results from ELISA,we construct and validate the optimal diagnostic model of multiple anti-TAAs autoantibodies,and evaluate the ability of model to distinguish OC from normal controls or ovarian benign diseases,so as to establish a more effective and non-invasive screening method for diagnosis of OC,and also provide a certain theoretical basis for the serological methods.Methods1.Screening of TAAs by whole proteome microarrayHuman proteome microarray V3.1 which contains 20240 newly sequenced recombinant human proteins was employed to discovery potential TAAs in serum from10 serum pools of 22 OC patients and 10 serum pools of 20 normal individuals.The non-parametric test and difference analysis of positive rate were used to find out TAAs with high-response in serum from OC patients.2.Applicating indirect ELISA to detect candidate anti-TAAs autoantibodies1)Preliminary detecting phase: The indirect ELISA was used to detect anti-TAAs autoantibodies in serum from 60 OC patients and 60 normal controls.The difference of autoantibodies level between the two groups of subjects was compared.The cutoff values were determined by the optical density value corresponding to the maximal Youden index at specificity over 90%.The ROC curve,sensitivity and specificity were performed to assess the diagnostic value of anti-TAAs autoantibodies in OC.2)Validation phase: The differentially expressed anti-TAAs autoantibodies screened in the preliminary detecting phase were further validated by ELISA in 145 OC patients,145 normal controls and 181 ovarian benign diseases patients.Same statistical methods were used to compare the expression levels of anti-TAAs autoantibodies among three groups and to assess the diagnostic performances of meaningful anti-TAAs autoantibodies in OC patients.3.Establishment and evaluation of diagnosis model for OC1)Establishment and validation of diagnosis model for OC: Based on the 290 cases consisting of 145 OC patients and 145 normal controls in the validation phase,the diagnosis models for OC were established through logistic regression analysis,parallel test analysis,and classical decision tree analysis.The cases consisting of 60 OC patients and 60 normal controls in the preliminary detecting phase were applied to evaluate the diagnostic performance of three models,and finally to select the optimal diagnostic model by comparing the AUC and accuracy rate of each model.2)Evaluation of diagnosis model for OC: The AUC,sensitivity and specificity were used to evaluate ability of the optimal diagnostic model for differentiating OC and benign ovarian diseases,and also to further evaluate the diagnostic value of the optimal diagnostic model for OC in different clinical subgroups,including age,FIGO stage,histological type,tumor of family history,lymph node metastasis,and distant metastasis.The positive rate was used to evaluate the diagnostic value of the optimal diagnostic model and CA125 for OC.Results1.A total of 14 candidate TAAs were screened using human proteome microarray technology and multiple analysis methods,including B3GNT7,GTF2 I,ENO2,GALNTL6,SULT1A3,SIGLEC8,ALPP,BASP1,VCL,GGT5,B3GNT3,PRF1,TRIM21 and RAD23 B.2.In the preliminary detecting phase,eight anti-TAAs autoantibodies(GALNTL6,SULT1A3,ALPP,BASP1,VCL,GGT5,B3GNT3 and TRIM21)with high expression level in OC were final identified.The AUC range of eight anti-TAAs autoantibodies to distinguish OC from normal controls was 0.605-0.714,the sensitivity and specificity ranges from 16.67%-38.33% and 90.00%-95.00%,respectively.The validation phase further show that the AUC range of the eight anti-TAAs autoantibodies to distinguish OC from normal controls was 0.604-0.755,the sensitivity and specificity ranges from 20.00%-40.69% and 90.34%-93.79%,respectively,among them,the anti-GGT5 autoantibody has the largest diagnostic AUC,which is 0.755,the sensitivity and specificity were 36.55% and 90.34%,respectively.The AUC to distinguish OC from benign ovarian diseases was range from 0.551-0.703,the sensitivity and specificity ranges from 20.00%-40.69% and 69.61%-91.71%,respectively,among them,the anti-GALNTL6 autoantibody has the largest diagnostic AUC,which is 0.703,the sensitivity and specificity were 31.03% and 91.71%,respectively.3.The diagnostic performance of models for logistic regression,parallel test analysis,and classic decision tree were compared,and the logistic regression model with three anti-TAAs autoantibodies(VCL,GGT5 and TRIM21)was finally selected as the optimal diagnosis model.The diagnostic AUC of the optimal diagnosis model was 0.791(95%CI: 0.740-0.842)in the training set,the sensitivity and specificity were66.20% and 77.24%,respectively.In the validation set,the diagnostic AUC was 0.763(95%CI: 0.679-0.848),the sensitivity and the specificity were 63.33% and 76.67%,respectively.The AUC of this model for distinguishing OC from benign ovarian diseases was 0.649(95%CI: 0.594-0.705),with sensitivity of 65.37% and specificity of 60.22%.The analysis of clinical subgroups demonstrated that there were no significant differences in the diagnostic value of model among ages,FIGO stage,histological type,tumor of family history,lymph node metastasis,and distant metastasis(P>0.05).The positive rate of the optimal diagnostic model combined with CA125 to detect OC patients was 85.61%,which was higher than the positive rate of CA125 alone(61.36%)and the optimal diagnostic model alone(68.18%),and the differences were statistically significant(P<0.001).Conclusions1.Based on human proteome microarray technology and two-phases validation of indirect ELISA,eight anti-TAAs autoantibodies(GALNTL6,SULT1A3,ALPP,BASP1,VCL,GGT5,B3GNT3 and TRIM21)were identified and expected to as potential sera biomarkers for diagnosis of OC.2.The logistic regression model containing 3 anti-TAAs autoantibodies(VCL,GGT5 and TRIM21)showed a certain ability to distinguish OC from normal controls or ovarian benign diseases.3.The positive rate of OC detection was increased for the combination of 3 antiTAAs autoantibodies(VCL,GGT5 and TRIM21)and CA125... |
PDF Full Text Request