Establishment And Internal Validation Of A Prediction Model For Extraprostatic Extension Of Clinical Localized Prostate Cancer

Objective:To analyze the risk factors associated with extraprostatic extension of localized prostate cancer,screen out the independent risk factors.To develop a clinical risk prediction model and conduct internal cross validation,which can be used for clinical guidance of preoperative and intraoperative decision-making,thus providing reference for clinical diagnosis and treatment,and making a balanced decision between tumor control to ensure negative margin and improvement of urinary control and erectile function which can improve survival and prognosis.Methods:To collect and retrospectively analyze the clinical data of patients who were diagnosed with clinically localized prostate cancer(T1-2N0M0)and underwent radical prostatectomy(RP)in the urology department of the First Affiliated Hospital of Nanchang University from January 2016 to December 2021.According to postoperative pathology,the patients were divided into organ confined(OC)and extraprostatic extension(EPE)groups.The age,body mass index(BMI),total prostate specific antigen(t PSA),free prostate specific antigen(f PSA),%f PSA(f/t PSA),prostate volume(PV),prostate specific antigen density(PSAD),Gleason score,clinical stage,percentage of positive cores,nerve invasion,tumor percentage of the highest biopsy score and EPE grade were compared.Using SPSS to screen out independent risk factors by univariate and multivariate logistic regression analysis.The clinical prediction model was established and internal cross validation was performed by R language,and using the discrimination,calibration and clinical utility to evaluate the model.Results:1.A total of 197 patients with clinically localized prostate cancer were included in this study,including 121 cases of organ confined(p T2)and 76 cases of extraprostatic extension(p T3),with the incidence of EPE as high as 38.6%.2.Univariate analysis showed that t PSA,f PSA,f/t PSA,PV,PSAD,Gleason score,clinical stage,percentage of positive cores,nerve invasion,tumor percentage of the highest biopsy score and EPE grade were significantly correlated with postoperative extraprostatic extension of clinically localized prostate cancer(P < 0.05)),but there were no significant differences in age and body mass index(P > 0.05).3.Multivariate logistic regression analysis showed that PSAD,Gleason score,tumor percentage of the highest biopsy score and EPE grade were independent risk factors for postoperative extraprostatic extension of clinically localized prostate cancer(P < 0.05).4.The area under ROC curve(AUC)of the PSAD,Gleason score,tumor percentage of the highest biopsy score,EPE grade and combined diagnosis were 0.726,0.804,0.802,0.751,0.820 and 0.927.5.The clinical prediction model was visualized as a nomogram,and ROC curve,calibration plot and decision curve analysis were drawn to evaluate the accuracy,calibration and clinical utility.The area under ROC curve was 0.927,indicating that the model has excellent utility and high prediction efficiency.The calibration plot is highly consistent with the standard line,indicating that the predicted risk of the model is highly consistent with the clinical actual risk.The decision curve analysis show that the model curve is keeping away from the two extreme lines significantly,indicating that the overall benefit of the population is good.6.After 200 repetitions of internal 10-fold cross validation,the average AUC calculated by deviation correction is 0.914.Compared with the AUC of 0.927 of the modeling population,indicating that the prediction model is reliable and stable.Conclusions:1.The incidence of EPE in patients with localized prostate cancer after RP in our center was as high as 38.6%.2.PSAD,Gleason score,tumor percentage of the highest biopsy score and EPE grade were independent risk factors for postoperative extraprostatic extension of clinically localized prostate cancer.3.The PSAD based on MRI measurement is better than t PSA in the diagnosis of EPE,and the combined diagnosis of independent risk factors is better than a single clinical indicator.4.In our study,a clinical prediction model was developed to predict the probability of EPE.This model has been internally validated and has excellent performance in discrimination,calibration and clinical utility.