Application Of Logistic Regression And ROC Curve In The Analysis And Evaluation Value Of The Combined Detection Of Of Survivin,γ-IFN And CRP On The Identification Of Malignant And Benign Pleural Effusion

【Object】To investigate the diagnostic value of the pleural effusion survivin,γ-IFN, and CRP and the combined detection survivin,γ-IFN,CRP in lung cancer and tuberculous.【Method】Logistic regression and ROC curve was used to analyze the results about pleural effusion Survivin,γ-IFN,CRP from 42 patients with lung cancer,48 tuberculous.The concentration of survivin,γ-IFN was determined by Enzyme- linked immunosorbent assay.The concentrations of CRPwere determined immuneoturbidimetry by special protein BNⅡautomatic analyzer method.【Result】(1) Logistic regression using multivariate analysis age,gender,whether or not more than 5 years history of smoking(active or passive smoking),with or without fever, with or without abnormal mass(physical examination or imaging),PPD,and pleural effusion in of survivin,γ-IFN,CRP.The establishment of the regression equation P= 1 /[1 + e-(3.218 +1.581 X1 +0.004 X2-0.007X3-0.150X4-2.506X5)](X1 = mass, X2 = Hydrothorax survivin,X3 = pleural fluidγ-IFN,X4 = pleural CRP,X5 = heat ) to produce a predictive value Pre.(2) pleural effusion survivin in lung cancer group was significantly higher than that of tuberculous group(P＜0.05),whereas pleural effusionγ-IFN and CRP in tuberculous group was significantly higher than that of lung cancer group,two group of Comparison is significant different(P＜0.05).(3) Pre,survivin,γ-IFN and CRP in the area under the ROC curve are:0.954,0.873, 0.846 and 0.827.Pre largest area under the curve,CRP smallest area under the curve.(4) According to ROC curve analysis,critical point of Pre,survivinγ-IFN and CRP in lung cancer or tuberculosis clinical diagnosis were 0.383,52.275 pg / ml,141.00 pg / ml,9.50 mg / L.(5) by ROC curve analysis and Youden,s index of the largest cut-off point for clinical diagnosis as a critical point in the Pre and survivin predict pleural effusion in lung cancer sensitivity and specificity were 91.40%,86.80%and 88.60%,84.20%. IFN and CRP in pleural fluid of tuberculous pleural effusion predict the sensitivity and specificity were 73.70%,85.70%and 81.60%,74.30%.(6) pleural effusion in the Pre + Survivin + mass of combined detection of lung cancer predict the diagnosis of pleural effusion maximum performance:sensitivity, specificity,and missed diagnosis rate,misdiagnosis rate,positive predictive value, negative predictive value and accuracy are as follows:95.24%,93.75%,4.76%, 6.25%,93.18%,97.83%,95.56%.Pleural effusion in theγ-IFN + CRP predict the joint detection of tuberculous pleural effusion highest diagnostic performance: sensitivity,specificity,and missed diagnosis rate of misdiagnosis rate,positive predictive value,negative predictive value and accuracy are as follows:92.11%, 94.29%,7.89%,5.71%,94.59%,91.67%,93.15%.Diagnostic value of combined detection of targets is better than a single detection.【Conclusion】(1) detection of pleural effusion Survivin,γ-IFN and CRP concentration on the identification of the nature of pleural effusion have a certain significance.(2) Pre predict detection of lung cancer diagnosis of pleural effusion is higher than the performance of Survivin;γ-IFN detection prediction of tuberculous pleural effusion is higher than the performance of the diagnosis of CRP. (3) Comprehensive evaluation of the combined detection of four indicators of sensitivity,specificity,and accuracy requirements and economics,combined detection pleural effusion of the Pre + Survivin + mass in lung cancer predict the highest diagnosis performance,combined detection pleural fluid ofγ-IFN + CRP in tuberculous predict the highest diagnosis performance.