Analysis Of Related Factors For Prostate Biopsy Results And Establishment Of Prediction Model

Objective: To analyze the relevant influencing factors of prostate biopsy results,and to develop two predictive models for prostate cancer and clinically significant prostate cancer.Methods: The clinical data of 237 patients who underwent ultrasound-guided transrectal prostate biopsy for the initial time in the Department of Urology,Yan’an University Affiliated Hospital from October 2019 to February 2022 were retrospectively analyzed.Analyzing the related factors affecting the results of prostate biopsy,then,two predictive models based on the results of the univariate and multivariate logistic regression analysis for overall prostate cancer and clinically significant prostate cancer were constructed,and the corresponding risk prediction nomograms were further constructed.Prediction models were evaluated by receiver operating characteristic（ROC）,calibration,and decision curves analysis（DCA）.Results: 1.A total of 237 patients were included in this study,of which 144（60.76%）were diagnosed with non-prostate cancer and 93（39.24%）were diagnosed with prostate cancer.Through logistic regression analysis,it was found that age,prostate volume,prostate-specific antigen,and PI-RADS score were independent risk factors for prostate cancer,and their corresponding regression coefficients were 1.096,-0.788,0.869,and0.870,respectively.Based on the above prostate cancer risk factors,a predictive model 1logit P=-3.776+1.096×age-0.788×PV+0.869×PSA+0.870×PI-RADS score was established,and the corresponding prostate cancer risk prediction nomogram was constructed.The AUC for predicting prostate cancer was 0.836（95% CI 0.784-0.887）,which was higher than the AUC for PI-RADS v2 score,PSA and PSAD（0.749,0.631 and0.610,respectively）.Calibration curve and decision curve analysis also showed that the prediction accuracy and clinical application value of the prostate cancer risk prediction nomogram model were favorable.2.All the included patients were further divided into two groups: non-cancer or clinically insignificant prostate cancer and clinically significant prostate cancer.Among the 237 patients,168（70.89%）prostate biopsy results showed non-cancer or clinically insignificant prostate cancer,69 cases（29.11%）were diagnosed with clinically significant prostate cancer.Through logistic regression analysis,it was found that age,digital rectal examination,prostate-specific antigen,and PI-RADS score were independent risk factors for clinically significant prostate cancer,and their corresponding regression coefficients were 0.977,1.797,1.000,and 0.796,respectively.Based on the above clinically significant prostate cancer risk factors,a predictive model 2 logit P=-6.119+0.977×age+1.797×DRE+1.000×PSA+0.796×PI-RADS score was established,and the corresponding risk prediction nomogram was constructed.The AUC for predicting clinically significant prostate cancer was 0.859（95% CI 0.805-0.912）,which was higher than the AUC for PI-RADS v2 score,PSA,and PSAD（0.740,0.679,and 0.615,respectively）.The calibration curve and the decision curve analysis also show that the model has favorable prediction accuracy and clinical application value.Conclusion: We constructed a prostate cancer prediction model based on independent risk factors such as age、PI-RADS score、PSA、PV,and constructed a clinically significant prostate cancer prediction model based on independent risk factors such as age、PI-RADS score、PSA、DRE,and corresponding risk prediction nomograms are further established.By plotting the ROC curve and calibration curve,it is found that the accuracy of the two prediction models is excellent,moreover,through decision curve analysis,it is also found that the two prediction models have favorable clinical application value.Although there is a lack of external validation and further multicenter external validation is required,predictive models can still provide considerable help for clinicians and patients when deciding whether to perform prostate biopsy.