Prospective Evaluation of Prostate Imaging-Reporting and Data System Version 2 Using the International Society of Urological Pathology Prostate Cancer Grade Group System: Beyond the Abstract

The Prostate Imaging-Reporting and Data System version 2 (PI-RADSv2) was announced in 2015 as a joint effort of the American College of Radiology (ACR), the European Society of Urogenital Radiology (ESUR) and the AdMeTech Foundation in an effort to improve standardization in multiparametric magnetic resonance imaging (mpMRI) of the prostate. Besides defining minimal requirements for mpMRI scans, the document introduces a 5-tier assessment system that reports the risk of clinically significant prostate cancer in detected lesions at mpMRI. Categories 1 and 2 define a very low and low likelihood of having clinically significant prostate cancer and biopsying these lesions are discouraged. Category 4 and 5 lesions define a high and very high likelihood for clinically significant prostate cancer and are recommended for biopsy. Category 3 lesions are indeterminate and although biopsy is discouraged the document does not provide explicit recommendations for how to proceed with these lesions.

The first step in understanding the clinical implications of the system would be correlating each assessment category with histopathological data. Our paper introduces the prospective evaluation of the PI-RADSv2 system in 339 patients scanned and biopsied from May 2015 to May 2016 at our institution. Overall 737 lesions were detected and biopsied by transrectal MRI/TRUS fusion-guided biopsy using the UroNav® system. We routinely obtain one core in the axial plane and one core in the sagittal plane. In rare instances, especially when lesions are very large, more cores can be obtained per lesion. The highest Gleason score of any biopsy core was defined to be the lesion’s cancer grade and based on this, an ISUP grade group was assigned additionally. Unlike PI-RADSv2 this five-scale grading group system is validated and correlates with the risk for biochemical recurrence. We calculated cancer detection rates (CDR) for prostate cancer and clinically significant prostate cancer, defined as ≥ ISUP2 (≥ Gleason 3+4), separately. Besides overall PI-RADSv2 risk categories, CDR in the peripheral and transition zone of the prostate were also calculated (Figure 1).
Screen Shot 2017 07 16 at 7.07.33 PM
Figure 1 Cancer detection rates for prostate cancer and clinically significant (CS) prostate cancer in the peripheral and transition zone. The P values refer to the significance of the difference to the next higher category.

As expected, higher PI-RADSv2 categories displayed higher CDR in both anatomical zones of the prostate. We expect selection bias for the CDR of categories 1 and 2 since these lesions are not routinely biopsied at our institution. To biopsy these lesions is an individual decision based on clinical parameters, image-based features from other sequences and patient’s preference. Because of this and the low sample size the data is not representing the true yield of category 1 and 2 lesions and the CDRs are probably higher than in a real clinical scenario. The CDR for category 4 lesions is surprisingly low especially since the definition reads as high likelihood of clinically significant prostate cancer. We believe the main reason for that is a high false positive rate. More data from other centers is needed to determine whether this is a center-specific finding or a systematic limitation of the PI-RADSv2 category system. One of the possible reasons for the high false positive rate can be the lack of quantitative objective criteria in category 4. Biopsy misses due to registration errors during the fusion-guided biopsy procedure is another possible explanation especially since PI-RADSv2 4 lesions are by definition smaller than category 5 lesions. Interestingly, T2 weighted MRI (T2W) category 4 lesions in the peripheral zone display a significantly higher CDR compared the overall PI-RADSv2 category 4 (48% vs. 37%, P=0.001) although the dominant sequence in the peripheral zone is diffusion weighted imaging (DWI). The reason for this still remains unclear. Higher signal-to-noise ratio of T2W and resultant higher spatial resolution could be a potential explanation. Additionally, T2W needs only evaluation of one sequence while DWI has two components (ADC maps and high b value DWI) to be covered by the radiologist. This could potentially lead to lower false positive findings.

The PI-RADSv2 assessment category system is developed to define the risk for clinically significant prostate cancer for detected lesion on mpMRI. However, it was not created to estimate the exact pathological grade. As expected, our analysis revealed a relatively weak correlation between PI-RADSv2 category and ISUP score although most higher ISUP scores were in category 4 and 5 lesions (Figure 2).
Screen Shot 2017 07 16 at 6.57.13 PM
Figure 2 Distribution of ISUP scores among PI-RADSv2 categories. Percentages correspond to proportion of ISUP scores compared to all cancer positive lesions detected for each category.

Our study is a first step in validating the new PI-RADSv2 assessment category system. The more data from other centers become available the more it will be clear which adjustments can possibly improve the performance of the system.

Written By: Sherif Mehralivand, Baris Turkbey

Read the Abstract