Can MRI accurately define tumor boundaries to guide focal salvage after radiotherapy?

124 Background: We evaluated the role of MRI (plus/minus biopsy) in delineating tumour boundaries for focal salvage therapy of prostate cancer recurrence after external beam radiotherapy. Methods: Patients with biochemical failure after radiotherapy were enrolled in a prospective clinical trial mapping sites of local recurrence. An integrated diagnostic MRI and interventional mapping biopsy procedure was performed under sedation in a 1.5T scanner. Patients were imaged with a pelvic coil and an endorectal coil attached to a stereotactic transperineal template assembly. Multiparametric MRI images were acquired, followed by targeted radial biopsy of suspicious regions and random sextant sampling of the normal-appearing peripheral zone. Histology maps were generated by delineation and registration of biopsy cores onto diagnostic images using point-based rigid image registration. Two independent blinded observers reviewed images offline and delineated tumours boundaries which were compared against overlaid histology maps. Coverage was considered accurate if all pathologically proven tumour sites were encompassed within delineated boundaries. Results: Of the 18 patients analysed to date, the majority (83%) were found to have local recurrence. Patients with <6 informative cores were excluded, leaving 15 patients for analysis. Observers performed comparably, whereby mean MRI sensitivity, specificity, PPV and NPV for detecting tumor was 0.76, 0.7, 0.7, and 0.75. The MRI tumour boundary was accurate in 5/15 patients, and improved to 8/15 patients with addition of a 5-mm expansion margin. Targeted radial biopsies improved accuracy to 14/15 patients, by excluding false positive regions (n=2), increasing tumor volumes (n=2) or both (n=2). Random sampling biopsy was only relevant in 1 patient by detecting tumor not identified by MRI and targeted biopsy. Conclusions: MRI alone is not sufficiently accurate to define boundaries for tumor-targeted salvage even with addition of an uncertainty margin. Targeted biopsy improved both detection and delineation accuracy for recurrent tumor regions, and changed salvage therapy planning in 40% of patients. No significant financial relationships to disclose.