Objective:

To evaluate the predictive value of a nomogram constructed by Prostate Imaging Reporting and Data System version 2 (PI-RADS v2) based on biparametric magnetic resonance imaging (bp-MRI) combined with clinical indicators for the diagnosis of clinically significant prostate cancer (cs-PCa).

Methods:

Clinical and imaging data of 251 patients who underwent prostate bp-MRI and confirmed pathologically by transrectal ultrasound-guided prostate biopsy from Jan. 2015 to Dec. 2019 were retrospectively analyzed

including PI-RADS v2 score

total prostate-specific antigen (t-PSA)

free PSA (f-PSA)

percent free PSA (f/t-PSA)

prostate volume (PV)

and PSA density (PSAD). Multivariate logistic regression analysis was performed to determine independent predictors for the diagnosis of cs-PCa

thereby establishing a nomogram predictive model and internally validated its predictive accuracy and consistency. The receiver operating characteristic (ROC) curve was used to compare diagnostic performance in the predictive model and these independent predictors for cs-PCa.

Results:

PI-RADS v2 score based on bp-MRI

age and PV were independent predictors of cs-PCa (

P

＜0.05). The predictive nomogram based on these independent predictors was developed and proven to have a satisfactory prediction accuracy with C-index of 0.920 for the internal validation. The area under curve (AUC) inthe predictive model was 0.932

significantly greater than those in PI-RADS v2 score (0.864

P

＜0.001)

PV (0.754

P

＜0.001) and age (0.676

P

＜0.001). In addition

diagnostic sensitivity and specificity for the predictive model for cs-PCa were 90.3% and 85.2%

which higher than those for PI-RADS v2 score (85.5%

76.2%)

PV (71.0%

69.3%) and age (85.5%

41.3%).

Conclusion:

The predictive nomogram established by PI-RADS v2 score based on bp-MRI

PV and age shows a satisfactory predictive value for cs-PCa

which can improve the diagnostic perfor

mance and has a preferable clinical practical value.