| BACKGROUNDLung caner is the most frequently diagnosed cancer worldwide and the leading cause of cancer related death. The incidence of lung caner is correlated with prevalence of cigarette smoking and varied from developed countries to developing countries, while it is much higher in developed countries than that in developing countries. In developed countries, lung cancer incidence has reached the plateau after their government launched interference project which also led death rate decreasing. Lung cancer can be cured while diagnosed with early stage.5-year overall survival of patients with stage la is 60%～80% after complete resection and higher than that ofⅡa andⅡb, which is 40%～50%. But when tumor metastased to the regional, local or distant tissues, the survival decreased to 49%,16% and 2% respectively. Unfortunately, more than 2/3 patients were diagnosed with local or advanced stage, which is incurable, Thus the general overall survival of lung cancer patients is less than 16%. Resent clinical trail for early stage lung cancer screening revealed that screening for early stage lung cancer could improve the overall survival of the population. However current screening methods are unavailable due to the low sensitivity, low specificity and low economic benefit cost ratio.Tumor immunology revealed that certain proteins released by the cancer cells, shedding from cells, or ectopic expression on cell surface could trigger the activation of humeral and cell specific immunity, in which autoantibodies were produced and circulated in whole body. With high abundance of autoantibodies function as immune surveillance, immune defense and immune response factor, its function as diagnostic tumor marker were reported worldwide. Since low sensitivity and low specificity has limited application of a single autoantibody, combinations of multiple autoantibodies could improve the diagnostic power in discriminating lung caner patients and healthy person. Enzyme-Linked Immunosorbent Assay, Phage display, serological proteomics analysis and protein micro array were all applied to profile autoantibodies panel techniquely, which make the methodology variously. Nevertheless, no autoantibodies panels could be used in clinical practice before further validation.Serological proteomics analysis has defined various tumor associated antigens involved in key pathway of tumor survival and development including Cell cycle regulation, proliferation, differentiation, apoptosis inhibition, angiogenesis and metastasis. Relevance of TAAs and their autoantibodies suggested that autoantibodies could be serological prognostic factor.Previously we have successfully expressed 10 TAAs by RTS wheat germ continuous exchanging cell free expression system and established ELISA for autoantibodies detection with these recombinant proteins. Logistic regression analysis showed that 4 of them were independent factor in predicting lung cancer, and a predictive factor from logistic regression diagnosed lung cancer with 93% sensitivity and 92% specificity when adopting an optimums cut-off value. In this study, we use independent sample set to validate the 4 autoantibodies profile and study its relationship with clinical prognosis. OBJECTIVE 4 tumor associated antigens were applied to evaluate autoantibodies in lung cancer and non-tumor control sera. Main object was to analyze correlation between autoantibodies and clinical parameters, access the diagnostic value of each autoantibodies and combination, validate the predictive factor in training set, and explore prognostic value of sera autoantibodies.METHOD116 Lung caner,38 control including healthy volunteers and non-tumor patients sera were subjected to our ELISA system which was established in training set with CECF expressed recombinant proteins of TP53, BIRC7, ANXA1 and ENO1. The mean Optical Density value of control group plus double time or triple time of standard deviations were set as a cut-off value to distinguish tumor and non-tumor. All clinical parameter were collected from Guangdong General Hospital electronical case management system, including age, gender, performance status, smoking behavior, tumor size, TNM staging, treatment and overall survival.Independent-sample T test was used to compare means of two groups, while One-Way ANOVA and Bonferroni test were used to compare means of more than two groups and multiple comparisons respectively. Non-parameter test was used to analyze data that insufficient to use T test, One-Way ANOVA or Bonferroni test. To compare frequency of two independent-samples and that of more than two, Pearsonχ2 test was applied. Receiver-Operation-Characteristics curve and Area Under Curve were used to evaluate diagnostic value. Logistic regression was used to analyze independent diagnostic tumor markers. Correlationship between survival and autoantibodies level and clinical parameters was analyzed by Kaplan-meier. Multivariate analysis was used to confirm independent prognostic factor. All data were analyzed by SPSS 13.0 for windows and statistical significance levels was set at a=0.05.RESULTS1. Autoantibodies level of TP53 (t=7.575,P=0.000), BIRC7 (t=8.094, P=0.000), ANXA1 (t=8.015, P=0.000) and ENO1 (t=10.685, P=0.00) in lung cancer patients were significantly higher than that of control group. No significant differences of autoantibodies level of TP53, BIRC7 and ENO1 in lung cancer patients were observed between different age, gender, smoking behavior, performance status, pathology and clinical stage. No significant differences of ANXA1 autoantibodies level in lung cancer patients were observed between different gender, smoking behavior, performance status, pathology and clinical stage. ANXA1 autoantibodies level in patients younger than 61 years was significantly higher than that in patients older than 61 years (t=2.016,P=0.039).2. Autoantibodies level of TP53 (r=0.444, P=0.005), BIRC7 (r=0.335, P=0.040), ANXA1 (r=0.426, P=0.008) correlate with maximum diameter of primary tumor in adenocarcinoma, the longer diameter, the higher autoantibodies level, while only AXNA1 (r=0.327, P=0.042) antibody level correlate with maximum diameter of primary tumor in squamous carcinoma.3. When set mean OD of training set control+2SD as cut-off value, specificity of TP53 shows significant difference between training set and validation set (x2=30.390, P=0.000), sensitivity(x2=9.834, P=0.002)and specificity(x2=6.569, P=0.010) of BIRC7 changed significantly, sensitivity of ANXA1 changed significantly (x2=9.328, P=0.002). When set mean OD of training set control+3SD as cut-off value, specificity of TP53 shows significant difference between training set and validation set (x2=6.752, P=0.009). No significant sensitivity and specificity changes were observed under both cut-off values.4. AUC of ROC curves of TP53, BIRC7, ANXA1 and ENO1 in training set were 0.83 (95%CI=0.77～0.90),0.84 (95%CI=0.77～0.91),0.86 (95%CI=0.80～0.93),0.90 (95%CI=0.85～0.95) respectively. Significant difference was observed in TP53 autoantibodies between training set and validation set (Z=3.598, P=0.000), While there were no significant AUC changes of BIRC7, ANXA1 and ENO1 autoantibodies between two sets.5.8 diagnostic models were established by logistics regression with TP53, ANXA1, BIRC7 and ENO1 data from training set combined freely. When using validation set data, their AUC of ROC were 0.823,0.828,0.836,0.877,0.847, 0.867,0.889,0.893. Significant AUC changes shows in NO.1, NO.2, NO.3, NO.5 and NO.6 models when comparing training set data with validation set data, however, no significant changes were observed in NO.4, NO.7 and NO.8 models. Maximum yueden index were defined as optimal cut-off value. NO.1, NO.3, NO.6 and NO.8 models showed significant sensitivity changes, NO.1, NO.2, NO.3, NO.5, NO.6, NO.7 and NO.8 models showed significant specificity changes. No significant changes of sensitivity and specificity were in NO.4 model when comparing two data sets. No clinical parameters could influence NO.4 model for lung caner diagnosis.6. Univariate analysis (Log-Rank test) suggested T stage (x2=8.681, P=0.034), N stage (x2=12.894, P=0.005), M stage (x2=11.143, P=0.001), clinical stage (x2=18.071, P=0.000) may be influence factors of overall survival in patient treated in Guangdong General Hospital, while serum ENO1 autoantibodies level (x2=3.775, P=0.052) and weight decreasing (x2=3.572, P=0.059) were associated—with OS. Clinical stage (P=0.000, Hazard Ratio=2.502, 95%Confidential Interval=1.598～3.924) was the only independent prognostic factor revealed by multivariate analysis.CONCLUSION 1. Gender, smoking behavior, performance status, pathology and clinical stage could not influence the sera autoantibodies level of TP53, BIRC7, ANXA1 and ENO1. Age could not influence the sera autoantibodies level of TP53, BIRC7 and ENO1.Age could influence the sera autoantibodies level of ANXA1.2. Tumor size could influence the sera autoantibodies level of TP53, BIRC7 ANXA1 in adenocarcinoma, and may influence ANXA1 antibody level in squamous carcinoma.3. ENO1 autoantibodies diagnostic value and cut-off value were both validated with AUC of ROC 0.917 (0.901 for validation set), sensitivity 70.8%(78.4% for validation set) and specificity 94.8%(92.1% for validation set) when using mean OD of training set control+2SD as cut-off value; sensitivity 51.4%(45.7% for validation set) and specificity 100%(100% for validation set) when using mean OD of training set control+3SD as cut-off value.4. One logistics regression model comprising OD value of ENO1, BIRC7 and ANXA1 (Logit (P)=19.047YENO1+9.444YBIRC7-8.964YANXA1-4.691) established with training set data was validated both in ACU and optimal cut-off value. This model show AUC of 0.929 (0.877 for validation set), sensitivity 81.3% (83.6% for validation set), specificity 92.7% (84.2% for validation set).5. ENO1 could be used in non-small cell lung cancer prognosis. |
PDF Full Text Request