Comparison of PIRADS v1.0 and v2.0 for MRI detection of prostate cancer: preliminary findings in patients with whole-mount histological workup
Josephin Otto1, Alexander Schaudinn1, Simone Mucha1, Nicolas Linder1, Nikita Garnov1, Jens-Uwe Stolzenburg2, Lars-Christian Horn3, Thomas Kahn1, Michael Moche1, and Harald Busse1

1Diagnostic and Interventional Radiology Department, Leipzig University Hospital, Leipzig, Germany, 2Urology Department, Leipzig University Hospital, Leipzig, Germany, 3Institute of Pathology, University of Leipzig, Leipzig, Germany

Synopsis

Multiparametric MRI of the prostate has substantially improved the detection of clinically significant prostate cancer (PCa) and the confidence in benign and insignificant findings. The recent v2.0 update of the Prostate Imaging Reporting and Data System (PIRADS) has replaced the sum of individual MRI sequence scores (1-15 or 1-20) with a zone-dependent, dominant MRI sequence (DWI in the peripheral and T2W in the transition zone) with an overall score from 1-5. The aim of this preliminary study was the blinded comparison of both versions for the detection of PCa on the same patients using whole-mount histological workup as gold standard.

Purpose

Multiparametric MRI (mpMRI) has substantially improved the detection of clinically significant prostate cancer (PCa) and the confidence in benign and insignificant findings. In 2012, the first version of the Prostate Imaging Reporting and Data System (PIRADS v1.0) has been issued as guideline; also aiming to reduce the large variability of MRI techniques and parameters. The recent v2.0 update has replaced the sum of individual parameter scores (1-15 or 1-20) with a zone-dependent, dominant MRI sequence (DWI in the peripheral zone PZ and T2W in the transition zone TZ) and overall scores of 1-5 only. The aim of this preliminary study was the blinded comparison of both versions for the detection of PCa on the same patients using whole-mount histological workup as gold standard.

Materials and Methods

This single-center, retrospective analysis included 28 patients with positive transrectal ultrasound-guided biopsy who underwent radical prostatectomy and were followed by whole-mount histological workup of the specimens. Presurgical mpMRI examination at 3 T (Magnetom Tim Trio, Siemens) was performed under institutional IRB approval with written informed consent of the patient. The MRI protocol included T2-weighted (T2W), diffusion-weighted (DWI, b = 50, 500, 800 s/mm2) and dynamic contrast-enhanced (DCE, 15‑20 ml Gd-DTPA) imaging and involved an endorectal coil (eCoil, Medrad). A blinded reader (5 years’ experience in prostate mpMRI) reviewed the data and was allowed to define up to four cancer-suspicious lesions per patient that were scored according to both PIRADS v1 and v2 [1]. The gold standard was provided by whole-mount sections (about 10, 4-5 mm thick) annotated with the Gleason Score GS by a pathologist (15 years’ experience in urogenital pathology). PIRADS scores were analyzed with respect to their sensitivity for the detection of PCa (per patient and per suspected lesion) and their positive predictive value PPV (per lesion). The zone-dependent dominant sequence of v2 (DWI or T2W) was also compared to the one with the highest individual score in v1.

Results

Patients were 48-75 (mean 63.9) years old and had PSA levels of 0.6-56 (mean 12.1) ng/mL. The pathologist marked a total of 39 lesions; 23 of them matched with the PZ (59%), 13 with the TZ (33%) and 3 (8%) with the seminal vesicles. The corresponding mean GS was 6.5. The reader assigned 41 total lesions. A high or very high probability for clinically significant PCa (PIRADS score ≥ 4) was assigned for 34 lesions under v1 and for 26 lesions under v2. Figures 2 and 3 show two sample cases where the strict interpretation of the zone-related dominant sequence resulted in different PIRADS scores under v1 and v2. PCa sensitivity per lesion were 82.1% (v1) and 64.1% (v2), sensitivity per patient were 85.7% and 67.9%, respectively. PPV per lesion was 94.1% (v1) and 96.2% (v2). In 44% of all patients, the prominent sequence (highest score) of v1 agreed with the dominant sequence of v2.

Conclusion

Our preliminary findings indicate that a strict interpretation of the current zone-related PIRADS v2.0 criteria (2015), which reflect the likelihood of clinically significant prostate cancer, may result in a lower sensitivity than the original criteria (v1.0, 2012). In contrast, PPV per lesion was marginally higher with v2.0.

Acknowledgements

No acknowledgement found.

References

[1] Barrett T, Turkbey B, Choyke PL, Clin Radiol. 2015;70(11):1165-76

Figures

Fig 1. Overview of scoring details for PIRADS v2.0 illustrating the role of individual MRI sequences in the peripheral and transition zone of the prostate.

Fig. 2. TZ lesion (segment 3a) with PIRADS rating T2W=2 and DWI=5 in v1 and T2W=2 and DWI=4 in v2. Low T2W score in dominant sequence (T2W, heterogeneous, encapsulated hypointensity) leads to different overall assignment.

Fig. 3. PZ lesion (segment 4p) with PIRADS rating T2W=4 and DWI=4 in v1 and T2W=4 and DWI=2 in v2. Lower v2 score in dominant sequence (DWI, no abnormality on high b-value DWI) leads to different overall assignment.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
2490