Interobserver Agreement and Positivity of PI-RADS Version 2 Among Radiologists with Different Levels of Experience.

RATIONALE AND OBJECTIVES: To evaluate interobserver agreement of Prostate Imaging Reporting and Data System (PI-RADS) v2 category among radiologists with different levels of experience. The secondary objective was to evaluate the positivity for significant cancer among each category (splitting category 4 into two) and among different lesion sizes. MATERIALS AND METHODS: Institutional review board and ethics comitee approved retrospective study. Eight radiologists with different levels of experienced in prostatic magnetic resonance imaging-two more experienced, four with intermediate experience, and two abdominal radiology fellows-interpreted 160 lesions. Reference standard was fusion-targeted biopsy. Percentage agreement, k coefficients, and analysis concordance were used. RESULTS: Coefficient of concordance according to categories was 0.71 considering both zones, 0.72 for peripheral zone (PZ) and 0.44 for peripheral zone (TZ). Agreement for PI-RADS score of 3 or greater was 0.48 in PZ and 0.57 in TZ. Tumor positivity rates were 54.3% and 66.0% for PI-RADS 3 + 1 and 4 for PZ, respectively; and 25.0 and 49.2% for PI-RADS 3 + 1 and 4 for TZ, respectively (p < 0.001 in both analysis). Lesions <10, 10-14, and ≥15 mm had 55.3%, 74.6%, and 93.5% of positivity rates for cancer in PZ (p = 0.002 and <0.001) and 26.7%, 56.5%, and 59.6% in TZ, respectively (p = 0.245 and 0.632). Sensitivities, specificities, and accuracies of magnetic resonance imaging for prostate cancer using PI-RADS v2 were 76%, 72%, and 74% for PZ; and 76%, 69%, and 71% for TZ, respectively. CONCLUSION: This study shows that PI-RADS v2 has overall good interobserver agreement among radiologists with different levels of experience. PI-RADS category 3 + 1 showed lower positivity rates for significant cancer compared to category 4. Lastly, lesions 10-14 mm has similar positivity rates compared to ≥15 mm for TZ lesions.