Treatment Delivery Gating of MRI-Guided Stereotactic Radiotherapy for Prostate Cancer: An Exploratory Analysis of a Phase III Randomized Trial of CT-Vs. MR-Guided Radiotherapy (MIRAGE)

Abstract

Compared to CT-guided radiotherapy, MRI-guided radiotherapy (MRgRT) has been shown to reduce acute physician-scored and patient-reported gastrointestinal and genitourinary (GU) toxicities associated with prostate stereotactic body radiotherapy (SBRT) in the MIRAGE randomized trial (NCT04384770). We hypothesize that real-time intrafraction tracking/gating is important and is a critical enabler of aggressive margin reduction with MRgRT. 79 patients received MRgRT on the MIRAGE trial with a planning margin of 2mm around the prostate and proximal seminal vesicles, which were treated to 40 Gy in five fractions on an MR-Linac. Tracking was performed at 4 frames/second in the sagittal plane during treatment with a gating boundary of 3mm for automatic beam hold. An in-house tool was developed to extract treatment time and beam gating status based on treatment logs and real-time cine images. The ratio of the time that the target was within the gating window/total time of target inside or outside the gating boundary was defined as the duty cycle (DC). Target contours were extracted from each frame of tracking and overlaid to create a motion-convolved target occupancy map. Minimum isotropic expansions of the prostate to cover 85%, 90% and 95% of the intrafraction motion were calculated with and without gating. Median treatment time per fraction including image guidance procedure and beam delivery was 24.3 min (IQR: 22.2-27.7 min). The median time for image guidance 5.4 min (IQR: 4.2-6.7 min). A total of 391 treatment fractions were analyzed and the median DC per fraction was 0.974 (IQR: 0.926 -0.983). 89 (22.8%) and 35 (9.0%) of fractions had DC<90% and <80%, respectively, corresponding to 50/79 (62.3%) and 24/79 (30.4%) of patients having at least one fraction with a DC<90% and <80%, respectively. The minimum duty cycle of all fractions was lower among patients with grade ≥2 GU toxicity compared to those with grade 0-1 GU toxicity (mean 79.8% vs. 85.9%, p = 0.06). The proportion of patients with grade ≥2 GU toxicity was also greater in patients with a minimum gating cycle <80% (37.5% vs. 18.2%, p = 0.06). Gating significantly decreased the minimum isotropic expansion of the prostate to cover 85%, 90% and 95% of the intrafraction motion (p<0.0001 for all). Prostate intrafraction motion tended to be along the bladder-rectum axis secondary to bladder filling, rectal gas and bulk motion. Fractions with large prostate motion were mostly stochastic. A large fraction (30%) of patients had at least of one treatment fraction with DC<80%, which correlated with increased acute GU toxicity. Gating effectively reduces the expansion needed to cover prostate intrafraction motion, and is necessary for real-time motion management given the unpredictable nature of prostate motion.