Study on the Value of Predicting Prostate Cancer in the Grey Zone of PSA by Use of PI-RADS v2.1 Combined with PSA-related Indicators

Objective: We intend to investigate the diagnostic value of establishing the Logistic regression model based on Prostate Imaging Report and Data System (PI-RADS v2.1) combined with prostate specific antigen (PSA) which is applied in PSA gray area(4~10ng/mL) prostate cancer. Materials and Methods: We retrospectively analyzed the pathologically-certified clinical data of 49 cases with prostate cancer (PCa) and 118 non-cancer cases who underwent prostate biopsy, the data covered age, tPSA, fPSA, PI-RADS v2.1 evaluation, PSAD and fPSA/tPSA. We performed logistic regression analysis on the indicators with statistical difference between the groups, and obtained ascertained independent PCa predictors, furthermore we respectively established regression prediction model by combination with PI-RADS v2.1 evaluation. The diagnostic efficacy of each model was evaluated by the operating characteristic curve (ROC) of subjects. Results: (1) There was no significant statistical difference in age, tPSA and fPSA between the PCa and non-cancer patients. But evaluation of PI-RADS v2.1, fPSA/tPSA, and PSAD showed significant statistical differences. (2) Logistic regression analysis indicated that PI-RADS v2.1evaluation, PSAD, and fPSA/tPSA are independent predictors of PCa; we established prediction models A and B as follows;Model A: Logit (P)=-10.82+2.32×PI-RADS v2.1+11.89×PSAD; Model B: Logit (P)=-6.13+2.19×PI-RADS v2.1-12.02×fPSA/tPSA. The area under the ROC curve was respectively 0.918 and 0.893, both were higher than the PI-RADS v2.1 evaluation independently applied, and we found that the differences were statistically significant. The sensitivity of model A was 0.843 and the specificity was 0.829, which showed better diagnostic efficacy compared with the sensitivity and specificity we got when the PI-RADS v2.1 evaluation was independently used (sensitivity 0.767 and specificity 0.801).Conclusion: The Logistic model established by combing PI-RADS v2.1 evaluation with PSA-related indicators showed better diagnostic efficacy in PSA grey area prostate cancer than PI-RADS v2.1 applied independently, in this way unnecessary needle biopsy can be avoided, and would play a significant instructive role in optimizing clinical treatment strategies.