276 - Design and Implementation of Combined Spatially Fractionated Radiotherapy and Image-Guided On-Treatment Biopsy for Resectable Soft Tissue Sarcoma Patients Enrolled in Prospective Clinical Trial LCCC2250

Purpose/Objective(s): LCCC2250 is a single-arm prospective trial evaluating safety, efficacy, and mechanism of pre-operative spatially fractionated RT (GRID) followed by conventional RT and resection in patients with soft tissue sarcoma. This study required the development of multileaf collimator (MLC)-based grid-therapy combined with an image-guided core biopsy technique to investigate the tumor microenvironment at peak and valley GRID locations.

Materials/Methods: MLC-based GRID was commissioned on a precision radiation medicine company multileaf collimator MLC using a step-and-shoot method, with 1×1 cm² beamlets separated by 2 cm in the MLC (X) direction and 1 cm in the jaw (Y) direction. Commissioning included A16 micro-ionization chamber (µIC) measurements at peak and valley locations and radiochromic film measurements of small (X=3 cm) and large (X=15 cm) GRID fields at depths d_max and d_5cm. Patient-specific quality assurance (QA) was designed to include both µIC measurements and portal images for each MLC control point. Image guidance for GRID delivery included VisionRT surface-guidance (SG) for setup and monitoring, and 6-degrees-of-freedom (6DOF) CBCT for localization. Ultrasound-guided core needle biopsies of trial patients are performed 72-96 hours after GRID, sampling tissue both inside and outside of beamlets. An image-guided biopsy technique was designed in collaboration with interventional radiology. Patients are set up for biopsy following the same immobilization, SG, and CBCT used for GRID treatment. Pre-planned 1×1 cm² MLC fields guide the biopsy needle placement using the light field. Commissioning of the biopsy technique involved training sessions with each interventionalist, using a novel soft tissue-simulating phantom and CBCT to quantify end-to-end needle placement accuracy (sans patient motion).

Results: MLC-based GRID achieved peak-to-valley ratios at d_max between 13.9 to 9.3 for X-field sizes 3 cm to 15 cm wide. Peak and valley µIC measurements agreed with treatment planning system (TPS) on average (±1 standard error) 2.3±1.2% and 1.6±0.1% for the first 3 trial patients, a maximum measured difference of 4.2%. GRID commissioning films analyzed using 3%/2mm gamma passed at 99.3±0.7% when rescaled to µIC results. Interventional radiologists training on the soft-tissue phantom achieved average end-to-end core needle biopsy accuracy of 0.9±0.2 mm (range 0.2-1.4 mm), corresponding to average 2.1% (maximum 4.4%) deviation from TPS dose for peak biopsies and 1.8% (max 2.7%) deviation for valley biopsies. In clinical setting, patient motion may add random error to end-to-end biopsy accuracy.

Conclusion: Commissioning and patient-specific QA results demonstrated that MLC-based GRID delivered using a precision radiation medicine company linear accelerator with a multileaf collimator MLC was accurate and clinically feasible. Tri-modality image guidance enabled end-to-end biopsy accuracy of < 2 mm (< 5% dose difference) on a soft tissue phantom. (NCT06073067)