For this study, the prospective multi-centre trial (multi-IMPROD, NCT02241122) is a validation study to previous published IMPROD-trial (NCT01864135). A total of 332 men from four different centres in Finland were included (Turku, 129 (39%); Pori, 90 (27%); Tampere, 55 (17%); Helsinki, 58 (17%)). Included men had clinical suspicion of prostate cancer (PSA ≥ 2.5 ng/ml and/or abnormal digital rectal examination). Imaging included anatomical MRI and diffusion weighted imaging (DWI) at 1.5/3 Tesla magnetic field using surface coils. In addition to routine twelve core TRUS biopsies, targeted biopsies based on MRI findings were performed (up to two lesions, two biopsies per lesion) using cognitive fusion. If there were no MRI targets, only systematic biopsies were performed. MRI lesions were classified using Likert scoring system into five categories. Prostate cancer was divided in three risk categories: low risk - biopsy Gleason score 3 + 3; intermediate risk - biopsy Gleason score 3 + 4; high risk - biopsy Gleason score 4 + 3 or higher. Significant prostate cancer was defined as intermediate- and high-risk prostate cancer.
Mean age of men was 64 years, median was PSA 6.8 ng/ml and median prostate volume was 39 cm3. The number of Likert 1 lesions was 32 (10%), Likert 2 43 (13%), Likert 3 66 (20%), Likert 4 55 (17%) and Likert 5 135 (40%), respectively. There were 128 (39%) men had no cancer, 58 (17%) low-risk, 48 (14%) intermediate-risk, and 98 (30%) high-risk cancer.
Comparing Likert 1-2 to Likert 3-5 lesions the negative predictive value (NPV) was 0.93 to rule out significant prostate cancer. The positive predictive value (PPV) was 0.55 and accuracy was 0.64. Comparing Likert 1-3 to Likert 4-5 lesions the corresponding values were NPV 0.91, PPV 0.70 and accuracy 0.79. For the receiver operating characteristic curve (ROC), the area under the curve (AUC) (95%CI) was 0.86 (0.82-0.90) to predict significant prostate cancer on biopsies.
The authors concluded that this validation study confirms the usefulness of the bi-parametric MRI protocol as a tool to aid prostate cancer diagnostics in men with clinical suspicion of prostate cancer. If only men with Likert 3-5 lesions would be biopsied, 23% of men would avoid biopsies missing 2% of significant prostate cancers. By taking biopsies only from Likert 4-5 lesions 141 (42%) men would have been saved from biopsies with 13 (4%) cases of significant prostate cancer missed.
Speaker: Kari Syvanen, Turku University Hospital, Turku, Finland
Co-Authors: Ettala O, Jambor I, Verho J, Knaapila J, Kiviniemi A, Kähkönen E, Kallajoki M, Taimen P, Lamminen T, Seppänen M, Rannikko A, Oksanen O, Riikonen J, Aronen H, Boström P
Written by: Zachary Klaassen, MD, Urologic Oncology Fellow, University of Toronto, Princess Margaret Cancer Centre, Twitter: @zklaassen_md, at the 2018 European Association of Urology Meeting EAU18, 16-20 March, 2018 Copenhagen, Denmark