Presentation Authors: Hari T. Vigneswaran*, Meltem Uyanik, Rolf Reiter, Virgilia Macias, Winnie Mar, Karen L. Xie, Richard L. Magin, Michael R. Abern, Chicago, IL
Introduction: The standard diagnostic evaluation of magnetic resonance imaging (MRI) in which radiologist use apparent diffusion coefficients (ADC) to determine risk of prostate cancer is called the Prostate Imaging Reporting and Data System (PI-RADs). PI-RADs has limitations in terms of accuracy and inter-rater reliability. We sought to evaluate the diagnostic performance of a stretched-exponential model of diffusion-weighted MRI (DW-MRI) for prostate cancer (PCa) detection, which has the potential to assess signal decay in prostate tissue more accurately compared to other validated mono-exponential models.
Methods: In this prospective study, 31 men with a total of 39 individual prostate lesions seen on MRI were consecutively enrolled and underwent MRI/ultrasound fusion prostate biopsies. Men with suspected PCa on multi-parametric MRI at 3-Tesla with b-values of 0, 50, 500, 1000, 1500 and 2000 s/mm2 were included. Clinical parameters were included to determine risk value of prostate cancer based on the Prostate Biopsy Collaborative Group (PBCG) risk calculator. ADC, Î±-values, PI-RADs, and PBCG were calculated using the standard mono-exponential DW-MRI model. ADC and Î±-values were calculated with the stretched-exponential DW-MRI model. The area under the receiver operating characteristic curve (AUC) was calculated to assess diagnostic performance.
Results: Of 39 included individual lesions, PCa was confirmed for 13 lesions (1 lesion with Gleason core (GS) 6 and 12 lesions with GS greater than or equal to 7). Sample cohort demographics and clinical parameters included: a median age of 62 years, a median prostate-specific antigen (PSA) of 7.5 ng/mL, with 28% self-identifying as Black, 23% on active surveillance, and 23% having a positive history of PCa in the family. AUC values were: 0.75 for PI-RADs, 0.64 for the PBCG algorithm, 0.84 for ADC, 0.71 for Î±, and 0.86 using the stretched-exponential model for ADC and Î±.
Conclusions: Stretched-exponential DW-MRI model provided excellent diagnostic performance for the detection of biopsy-proven PCa. Performance was slightly superior to the standard mono-exponential DW-MRI model as well as established clinical models using PI-RADs and PBCG. This motivates further investigation of these models for PCa assessment.
Source of Funding: Support from the Department of Defense: Physician Research Training Award #W81XWH-15-1-0346