Diagnostic accuracy of biparametric vs multiparametric MRI in clinically significant prostate cancer: Comparison between readers with different experience

Abstract

BackgroundMRI plays a crucial role to identify men with a high likelihood of clinically significant prostate cancer who require immediate biopsy. The added value of DCE MRI in combination with T2-weighted imaging and DWI is controversial (risks related to gadolinium administration, duration of MR exam, financial burden, effects on diagnostic performance). A comparison of a biparametric and a standard multiparametric MR imaging protocol, taking into account the different experience of the readers, may help to choose the best MR approach regarding diagnostic performance. PurposeTo determine the added value of dynamic contrasted-enhanced imaging (DCE) over T2-weighted imaging (T2-WI) and diffusion weighted imaging (DWI) for the detection of clinically significant prostate cancer, and to evaluate how it affects the diagnostic performance of three readers with different grade of experience in prostate imaging. Materials and methodsEighty-five patients underwent prostate MR examination at 1.5 T MR scanner performed because of elevated prostate-specific antigen level and/or suspicion of prostate cancer at digital rectal examination.Two MR images sets (Set 1 = biparametric, Set 2 = multiparametric) were retrospectively and independently scored by three radiologists with 7, 3 and 1 years of experience in prostate MR imaging respectively, according to PI-RADS v2. Sensitivity, specificity, positive predictive value, and negative predictive value were calculated by dichotomizing reader scores. Receiver operating characteristic (ROC) analysis was performed and areas under the curve (AUCs) were calculated for each reader and image set. A comparison of ROC curves was performed to test the difference between the areas under the ROC curves among the three readers. ResultsThere was no significant difference regarding the detection of clinically significant tumor among the three readers between the two image sets. The AUC for the bi-parametric and multi-parametric MR imaging protocol was respectively 0.68–0.72 (Reader 1), 0.72–0.70 (Reader 2) and 0.60–0.54 (Reader 3). ROC curve comparison revealed no statistically significant differences for each protocol among the most experienced (Reader 1) and the other readers (Readers 2–3). ConclusionThe diagnostic accuracy of a bi-parametric MR imaging protocol consisting of T2-weighted imaging and DWI is comparable with that of a standard multi-parametric imaging protocol for the detection of clinically significant prostate cancer. The experience of the reader does not significantly modify the diagnostic performance of both MR protocols.