Could Appliacation Of MRI Improve The Clinical Decision Value Of Predictive Nomogram For Initial Prostate Biopsy In PSA Gray Zone Patients?

Purpose.This study construct a predictive model by several clinical parameters such as age,total PSA,and is aimed at exploring the clinical utility of prostate MRI before initial prostate biopsy.Methods.We collect clinical data of 893 inpatients who underwent prostate biopsy from January 2015 to June 2016.All patient were referred for a prostate biopsy due to elevated PSA and/or suspicious prostate nodule found by ultrasound.We excluded the patient with incomplete information and extreme values(PSA?60ng/ml).Finally,376 cases were enrolled for investigation.We first examined the correlation between prostate cancer and each clinical parameters,including age,total PSA level,free PSA to total PSA ratio,prostate volume,suspicious prostate nodule,MRI grade.Then we examined the independence of each correlated parameters.Then we constructed the predictive nomogram by the correlated' and independent clinical parameters.The internal validation is measured by c-statistic and calibration curve,finally clinical decision analysis was made to analysis the clinical evaluation of the prostate MRI grade in predictive nomogram.Results.First we divided the patients into two groups.The patients whose total PSA ranged from 4 to 9.99ng/ml were assigned to Group A,while the patients whose total PSA ranges from 10 to 59.99ng/ml were assigned to Group B.There were 26.6%(49/184)patients confirmed to have prostate cancer in the Group A.And the MRI grade 0,1,2 each attributed to 41.3%(76/184),25.5%(47/184),33.1%(61/184).And there were 55.7%(107/102)patients confirmed to have prostate cancer in the Group B.And the MRI grade 0,1,2 each attributed to 31.8%(61/192),23.4%(45/192),44.8%(86/192).What's surprised was that if MRI grade was 0,prostate cancer were only present 2.2%(4/184)at Group A and all these four patients were classified as low-risk finnaly.After the univariate analysis and multivariate analysis,we constructed the nomogram model 1 for Group A by age,total PAS,suspicious prostate nodule and MRI grade and nomogram model 1 for Group B by age,total PAS,prostate volume,suspicious prostate nodule and MRI grade.And we constructed model 2 for each group which was almost the same as the model 1 for each group except the exclusion of MRI grade.The c-statistic for mode 1 for Group A was 0.845 while the mode 2 was only 0.766.And the c-statistic for mode 1 for Group A was 0.882 while the mode 2 was 0.846.The sensitivity and specificity of model 1 for Group A was higher than model 2 for Group A at every cutoff(p<0.01).The clinical decision analysis shows that in Group A,the mode 1 incorporated with MRI grade performed better than the model 2,while the advantage was not obvious in the Group B.Conclusion.Prostate MRI before biopsy could predict the presence of prostate cancer and the application of MRI in predictive nomograms could increase predictive accuracy and clinical utility,especially in patients whose total PSA ranges from 4 to 9.99ng/ml.