Patient Selection for Multiparametric Prostate MRI: Identifying Clinical Predictors for Actionable PIRADS Lesions on Imaging
Sources of Funding: None
Introduction
Despite a favorable sensitivity profile for high grade and large volume disease, indiscriminate use of multiparametric prostate MRI (mpMRI) adds prohibitory cost with uncertain benefit. Our study aims to identify clinical predictors of suspicious lesions on imaging to improve patient selection for mpMRI.
Methods
We performed a retrospective review of 839 patients undergoing mpMRI for elevated PSA between March 2012 and October 2014. mpMRI was performed on 3T magnet with pelvic phased array and endorectal coils. Baseline clinical and biochemical patient characteristics were analyzed with univariable and multivariable logistic regression to identify predictors of a positive MRI (PIRADS score 3-5). Using these variables, we constructed a nomogram to predict a positive MRI.
Results
Among 839 patients without prior history of prostate cancer, MRI was positive in 272 (32.4%) patients. Increasing age (P=0.001), abnormal digital rectal exam (P=0.002), prior negative biopsy (P<0.001), increasing pre-MRI PSA (P<0.001), lower prostate volume (P<0.001), and PSA velocity (P=0.044) were significant predictors of a positive MRI in univariable analysis. On multivariable analysis, age (P=0.048), positive digital rectal exam (P=0.017), prior negative biopsy (P=0.061), pre-MRI PSA (P<0.001), and prostate volume (P<0.001) remained independent predictors (Table). A nomogram predicting probability of a positive mpMRI based on these variables demonstrated good calibration and a concordance index of 0.7 (Figure).
Conclusions
Many of the clinical variables traditionally associated with increased PCa risk are also independently associated with a positive mpMRI. A nomogram including these variables can help identify men who are more likely to benefit from a prostate mpMRI while reducing cost by limiting the number of negative studies.
Funding
None
Paras Shah
Daniel Moreira
Arvin George
Geoffrey Gaunay
Jose Vilaro
Manish Vira
Simon Hall