Iterative intraoperative 3T MRI (iMRI)-guided brachytherapy: A prospective study on enhancing implantation precision and dosimetric gains in advanced gynecologic cancers
To report on primary outcomes and dosimetric results of a prospective clinical trial and protocol for use of iterative intraoperative magnetic resonance imaging (iMRI) in gynecologic brachytherapy. Patients with locally advanced cervical or vaginal cancer (FIGO stages IB2 - IVA, and stage II-IVA, respectively) undergoing pulsed dose rate (PDR) brachytherapy were enrolled in a prospective clinical trial (NCT03634267) using iterative 3T iMRI during brachytherapy implant placement. Applicator and optional interstitial needles were placed under iMRI guidance in a 3T clinical MRI scanner. Imaging, dosimetry and clinical outcomes (local control (LC), recurrence-free survival (RFS), overall survival (OS)), and acute and long-term toxicity were evaluated prospectively and confirmed by chart review. To explore dosimetric gains, an EQD2 estimate comparing iMRI-guidance versus standard of care guidance, as well as a 3-patient analysis of dose changes with iMRI-guided optimization was included. Fourteen patients underwent iMRI-guided brachytherapy. Seventy percent (70%) of patients presented with FIGO stage III disease or higher. Median follow-up was 44 months. . Patients had 2-year median LC, RFS, and OS rates of 83.3%, 76%, and 84.6%, respectively. Acute toxicities were minimal with one (1) case of grade 3 nausea. No grade 3 or higher long-term toxicities were observed. Median operating room (OR) time was 283 minutes (range 174-380 mins). On exploratory analysis, implant placement performed with iMRI guidance demonstrated higher HR-CTV D90 doses (mean difference of +784.7 cGy, p = NS) were achieved compared to US and CT guided implantation in the same patients. iMRI-guidance for gynecologic brachytherapy is safe, associated with minimal high-grade toxicity and excellent clinical outcomes. Future studies to optimize resource use, image acquisition efficiency, and identifying predictive imaging features are warranted.