Prostagram magnetic resonance imaging in a screening population: Prostate Imaging‐Reporting and Data System or Likert?

Abstract

ObjectiveTo compare biopsy recommendation rates and accuracy of the Prostate Imaging‐Reporting and Data System, version 2 (PI‐RADSv2) with the Likert scale for detection of clinically significant and insignificant prostate cancer in men screened within the Imperial Prostate 1 Prostate Cancer Screening Trial Using Imaging (IP1‐PROSTAGRAM).Patients and MethodsMen aged 50–69 years were screened with Prostagram MRI. Scans were prospectively reported using both PI‐RADSv2 (excluding dynamic contrast‐enhanced sequence score) and 5‐point Likert scores by expert uro‐radiologists. Systematic and targeted transperineal biopsy was recommended if the scan was scored ≥ 3, based on either reporting system. The proportion of patients recommended for biopsy and detection rates for Grade Groups (GGs) 1 and ≥ 2 were compared. Receiver operating characteristic (ROC) analysis was performed to compare performance.ResultsA total of 406 men underwent Prostagram MRI. The median (interquartile range) age and prostate‐specific antigen level were 57 (53–61) years and 0.91 (0.56–1.74) ng/mL, respectively. At MRI score ≥ 3, more patients were recommended for biopsy based on Likert criteria (94/406; 23%, 95% confidence interval [CI] 19.2%–27.6%) compared to PI‐RADSv2 (72/406; 18%, 95% CI 14.2%–21.9%; P = 0.03). For MRI scores ≥ 4, PI‐RADSv2 and Likert scales led to 43/406 (11%, 95% CI 7.9%–14.1%) and 35/406 (9%, 95% CI 6.2%–11.9%) men recommended for biopsy (P = 0.40). For GG ≥ 2 detection, PIRADSv2 and Likert detected 22% (95% CI 11.4%–30.8%, 14/72) and 16% (95% CI 9.5%–25.3%, 15/94), respectively (P = 0.56). For GG1 cancers detection these were 11% (95% CI 4.3%–19.6%, seven of 72) vs 11% (95% CI 4.7%–17.8%, nine of 94; P = 1.00). The accuracy of PI‐RADSv2 and Likert scale was similar (area under the ROC curve 0.64 vs 0.65, P = 0.95).ConclusionsIn reporting non‐contrast‐enhanced Prostagram MRI in a screening population, the PI‐RADSv2 and Likert scoring systems were equally accurate; however, Likert scale use led to more men undergoing biopsy without a subsequent increase in significant cancer detection rates. To improve reporting of Prostagram MRI, either the PI‐RADSv2 or a modified Likert scale or a standalone scoring system should be developed.