1919. Prospective Validation of an Automated Hybrid Multi-dimensional MR Imaging-Based Tool to Identify Areas for Prostate Cancer Biopsy
Authors * Denotes Presenting Author
  1. Grace Lee *; University of Chicago, Department of Radiology
  2. Aritrick Chatterjee; University of Chicago, Department of Radiology
  3. Roger Engelmann; University of Chicago, Department of Radiology
  4. Ambereen Yousuf; University of Chicago, Department of Radiology
  5. Gregory Karczmar; University of Chicago, Department of Radiology
  6. Aytekin Oto; University of Chicago, Department of Radiology
  7. Glenn Gerber; University of Chicago, Department of Urology
The goal of this clinical trial is to validate an automated Hybrid Multidimensional MRI (HM-MRI) based tool to prospectively identify areas for prostate cancer (PCa) biopsy. This study evaluates whether HM-MRI based tool identifies PCa more reliably than random biopsy and/or targets detected by an expert radiologist based on PI-RADS v2 in patients undergoing MR-Ultrasound ( MR-US) fusion biopsy.

Materials and Methods:
In this prospective clinical trial, patients (n=42, mean age=64 years, mean PSA=8.4ng/ml) with known or suspected PCa underwent 3T MRI: multiparametric MRI (mpMRI) (T2W, DWI and dynamic contrast-enhanced (DCE)) and HM-MRI (TE=57,75,150,200ms, b-values=0,150,750,1500s/mm2) before MR-US fusion biopsy. Tissue composition (stroma, epithelium and lumen) was calculated using a three compartment model and suspected PCa regions with elevated epithelium (>40%) and reduced lumen (<20%) were identified using the HM-MRI tool. Patients then underwent 12-core TRUS-guided sextant random biopsy. Additional biopsy targets were selected based on an expert radiologist's mpMRI interpretation (=PI-RADS 3). Up to two additional biopsy targets per patient were selected by the HM-MRI tool, if different from the already selected targets by the radiologist per standard of care and biopsied using Uronav MR-US fusion biopsy device. Analyses based on a per-patient, per-tumor and sextant based analysis were performed, with area under the receiver operating characteristic (ROC) curve as the primary endpoint and sensitivity, specificity, accuracy, positive and negative predictive values (PPV and NPV) as secondary endpoints.

The diagnostic accuracy (AUC) of HM-MRI for clinically significant cancers (=Gleason 3+4), was higher than that of mpMRI on per-patient (0.71 vs 0.71) and sextant analysis (0.76 vs 0.70). HM-MRI had higher accuracy (68-90 vs 50-85%), specificity (59-91 vs 23-87%) and PPV (30-53 vs 20-41%) than mpMRI. HM-MRI had higher sensitivity in per-sector basis (61 vs 52%), but not on per-patient basis (83 vs 100%). HM-MRI had lower NPV on per-patient basis (87 vs 100%) but similar value on sextant basis (97%). On a per-tumor basis, HM-MRI had higher sensitivity (78 vs 71%) and PPV (34 vs 19%) than mpMRI.

This study demonstrates that HM-MRI can improve PCa diagnosis by identifying areas of PCa with improved accuracy compared to mpMRI. Clinical Relevance statement: HM-MRI is an automated, quantitative image interpretation tool that can improve PCa diagnosis and potentially reduce the number of unnecessary biopsies.