Stereotactic MR-Guided Adaptive Radiotherapy (SMART) for Ultra-Central Thorax Lesions: Results of a Phase 1 Trial

Abstract

SBRT is an effective treatment for oligometastatic or unresectable primary malignancies, although target proximity to organs-at-risk (OAR) within the ultra-central thorax (UCT) limits safe delivery of ablative dose. SMART may improve the therapeutic ratio using reoptimization to account for daily variation in target/OAR anatomy. This study characterizes dosimetric and clinical outcomes in patients treated with SMART to UCT lesions on a prospective phase I trial. Five patients with oligometastatic (n = 4) or unresectable primary (n = 1) UCT malignancies were treated with SMART as part of a phase I trial (NCT02264886) from 4/2015 to 10/2015, using a clinical MR-IGRT system. Initial SBRT plans (PI) were created at CT/MRI simulation. Prescribed dose was 50 Gy/5 fractions (fx), with goal 95% PTV coverage by 95% of prescription, subject to hard OAR constraints. For fx1, PI was applied to the daily setup MR dataset and dose to PTV/OAR was recalculated and verified by the physician. When application of PI to fx1 anatomy resulted in OAR violation or reduced PTV coverage, daily adaptive plans subject to identical OAR constraints were created (PA). Online re-planning, QA, and delivery of a new plan were performed with the patient on-table. For subsequent fx, the delivered plan from the preceding fx was used as the new PI. When favorable daily anatomy was observed, PTV dose escalation was allowed in PA, provided all OAR constraints were met. PTV/OAR dose resulting from application of PI and PA to anatomy from each fx were compared to characterize the dosimetric impact of online adaptation. Patients were evaluated at three and six months for lesion control by RECIST criteria and monitored for acute grade >=3 toxicity. All PI and PA met hard OAR constraints based on simulation and daily setup MRI anatomy, respectively. Of four patients evaluated for plan adaptation, all received >=1 adapted fx and 10/20 fx were adapted. One patient was not adapted due to physician preference and was treated with MR-localization and cine MR-gating alone. Fifty percent of plan adaptations performed were for improved PTV coverage; 50% were for reversal of >=1 OAR violations of the esophagus (n = 3), trachea (n = 4), and stomach (n = 1) that occurred when PI was applied to the daily anatomy. Median on-table time for all fxs was 69 minutes (range 22-117 min). Local control by RECIST was 100% at three and six months. One locoregional failure was observed at eight months follow-up. Zero Grade >=3 acute toxicities were observed. At median follow-up of 14 months, only one patient treated to a paratracheal node had late toxicity with a benign esophageal stricture at 15 months post-treatment. Overall survival at 6 and 12 months was 100% and 60%. SMART allows PTV coverage improvement and/or OAR sparing compared to non-adaptive SBRT, and may widen the therapeutic index of UCT SBRT. This technique is clinically feasible and offers excellent local control with no observed acute Grade>=3 toxicity.