The Logistic Regression Model’s Predictive Value For Prostate Cancer In PSA "Gray Area"

Objective: Analysis of Prognostic Value of Constructing Prostate Cancer Logistic Regression Model in PSA "Gray Area".Methods: The clinical data of 80 patients with PSA 4.0-10.0ng/ml and initial prostate biopsy in the Bethune First Hospital of Jilin University from January 2013 to January 2018 were analyzed retrospectively,which of 66 patients were listed finally.Single factor analysis was performed respectively among factors such as age,PV,PSA,f PSA,f/t PSA,PSAD and prostate MRI,then,through it and clinical experience,the multivariate Logistic regression was performed and the Logistic regression model was constructed to predict the prostate cancer.Finally,the ROC curve was asked to compare the predictive value for prostate cancer in f/t PSA,PSAD,prostate MRI and Logistic regression model of PSA 4.0-10.0ng/ml.Results: Among the 66 patients,there have fifteen cases(22.73%)of the prostate cancer group and 51 cases(77.27%)of the non-prostate cancer group.The age,PSA,f PSA,and f/t PSA of the two groups are not statistically different(P>0.05),but PV,PSAD and prostate MRI have statistical differences(P<0.05).Depending on the result of single factor analysis and clinical experience,age,PSA,f PSA,PV and prostate MRI were included in the multivariate logistic regression and Logistic regression model,Y=-15.30275+0.17651 age +0.65811 t PSA-1.28375 f PSA-0.083141 PV+3.61812 MRI.The area under the ROC curve of PSA is 0.614(95%CI: 0.486-0.731).The area under the ROC curve of f/t PSA is 0.685(95%CI: 0.559-0.794).The area under the ROC curve of PSAD is 0.817(95%CI: 0.703-0.901).The area under the ROC curve of MRI is 0.689(95%CI: 0.573-0.805).The area under the ROC curve of Logistic regression model is 0.918(95%CI: 0.823-0.971).There are statistical difference of the predictive value between Logistic regression model and other diagnostic indexes(P<0.05),and obviously superior to f/t PSA or PSAD.When the value of this regression model is 0.4011,the sensitivity and specificity for predicting prostate cancer were 86.67% and 92.16%,respectively.Conclusion: The constructed Logistic multivariate regression analysis model is more statistically significant for predicting prostate cancer at PSA "gray area",which can assist the further diagnosis and treatment of the prostate.