Peripheral zone lesions of intermediary risk in multiparametric prostate MRI: Frequency and validation of the PI-RADSv2 risk stratification algorithm based on focal contrast enhancement

Purpose: To validate the risk stratification algorithm of the Prostate Imaging Reporting and Data System (PI-RADSv2) for intermediary risk lesions (PI-RADSv2 category 3) in the peripheral zone based on focal contrast enhancement and to compare cancer rates in category 3, upgraded category 4 and category 4 based on markedly low ADC value.

Materials and methods: We retrospectively analyze 172 consecutive patients undergoing prostate MRI with 315 histopathologically verified lesions. We select all lesions either assigned category 3 or category 4 in the peripheral zone for further analysis. We compare cancer rates with the two-sided chi-squared test. To determine inter-observer agreement about contrast enhancement two blinded radiologists evaluate the subset of category 3 lesions based on the diffusion weighted sequence.

Results: The frequency of peripheral PI-RADS 3, upgraded PI-RADS 4 and PI-RADS 4 lesions based on markedly low ADC value is 10.8%, 10.8% and 20.3%, respectively. Cancer rates (significant cancer only) in these subgroups are 8.8% (3/34), 23.5% (8/34) and 40.6% (26/64), P < 0.01. Inter-observer agreement is moderate for evaluation of contrast enhancement with kappa values between 0.46 and 0.5.

Conclusion: We demonstrate a trend of increasing cancer rate from PI-RADSv2 category 3 to upgraded category 4 to category 4 based on markedly low ADC value. Peripheral lesions of intermediary risk in the diffusion weighted sequence account for 21.6% of all prostate lesions encountered. Since it is likely that patient management recommendations will be linked to assessment categories in future versions of PI-RADS, cancer rates in upgraded category 4 and category 4 based on markedly low ADC values should be in a similar range. We conclude that in future studies of PI-RADSv2 upgraded category 4 and category 4 based on markedly low ADC value should be reported separately to generate a database for meta-analysis of cancer rates.