ESTRO 2020 Abstract book

S923 ESTRO 2020

PO-1604 Quantitative analysis of treatment process time for MRI-LINAC system gating efficiency optimization L. Placidi 1 , V. Pollutri 2 , C. Votta 2 , D. Cusumano 1 , L. Boldrini 3 , G. Chiloiro 3 , L. Azario 1 , M. De Spirito 1 , V. Valentini 3 1 Fondazione Policlinico Universitario A. Gemelli IRCCS - Università Cattolica Sacro Cuore, Medical Physics, Rome, Italy ; 2 Fondazione Policlinico Universitario A. Gemelli IRCCS, Radiotherapy, Rome, Italy ; 3 Fondazione Policlinico Universitario A. Gemelli IRCCS - Università Cattolica Sacro Cuore, Radiotherapy, Rome, Italy Purpose or Objective Real-time gating Magnetic Resonance-guided radiotherapy (RGMRgRT) system allows moving tumours on-line tracking. When treatment is linked to gating system, beam delivery time can be longer than without gating. Quantitative analysis of the treatment process time (TPT) using data obtained from patient treatment logs, has been used to evaluate to following aims: 1. assess if the duration of the slots provided for gating breath hold (BH) inspiration patients (30 min) are sufficient or not, to then optimize the daily treatment room schedule 2. compare how long the TPT has been varied between Viewray MERIDIAN Cobalt-60 three-sources (VMC) and the Viewray MERIDIAN LINAC (VML) RGMRgRT system 3. compute the gating efficiency (GE) for both free breathing (FB) and BH patients, to evaluate how close the estimated TPT provided by the TPS (mechanical and beam- on) is to the actual delivered (mechanical, beam-on and gating) 4. assess possible correlation between the GE and some clinical/physical/mechanical parameters Material and Methods Based on a selected cohort of 210 patients, a preliminary investigation has been carried out on 60 patients, corresponding to 318 fractions delivered in the RGMRgRT system. TPT was computed using data from planning and mechanical logs for each patient and each treatment fraction. Comparison between TPT of the in clinical operation VML (27 patients) and the previously VMC (33 patients) system was also evaluated. Treatment planning data are summarized in Fig. 1. Gating treatment boundary is the margin from CTV to gating the treatment and percent ROI is the allowed percentage of the tracked ROI outside the defined boundary to deliver beam-on. Only 5 plans (8.3%) were prescribed as long course ( >10 fractions with a dose per fraction <3 Gy); the remaining 55 plans (91.7%) were prescribed as SBRT ( <5 fractions, with a dose per fraction in the range of 5-12.5Gy). GE (defined as the percentage difference of the total delivered time and the estimated total time, computed by the TPS) was computed for VML BH and FB dataset.

were regularly taken for tracking every 30-90sec and offsets were compensated by the robotic arm. Residual error was quantified as the difference between corresponding measured translational/rotational shifts of consecutive images. The error distribution for each fraction, patient and the whole population was assessed for each axis/rotation angle. The overall error (OE) [x²+y²+z²]½ (translations along each axis) was also calculated. The OEs for pts treated to T and L spine metastases were estimated and compared, as well as OEs from different sessions for multi-fraction treatments. We verified a possible correlation with pain estimating and comparing the OE of pts with VAS=0 and VAS>0. Significance was evaluated by Mann-Whitney tests. Finally, we simulated the delivered treatments of 4 pts identified as the worst case scenarios (2 pts with the highest single offset and 2 pts with the highest mean error) editing the original beams coordinates, based on the real shifts between consecutive corrections. Resulting “true” dose distributions and planned ones were compared Results The whole population mean error was 0 (SD 0.1mm). Translational (cranio-caudal, lateral, anterior-posterior) shifts >1mm were <2%; rotational (roll, pitch, yaw) shifts >1° were < 1%. No significant differences were found between OE for pts treated to T or L spine (0.08mm, 0.06mm) or between different fractions (0.09mm at fraction 1, 0.07mm at fraction 5). The difference between the OE of a patient group with 0 VAS score (0.05mm) and the others (0.06mm) was not significant; still, the corresponding values of OE in case of no-tracking correction showed a significantly higher fraction of OE≥2mm for pts with VAS>0 compared to VAS=0 (4/14 vs 0/23, p=0.03): Fig.1 summarizes both situations. No relevant differences between the planned and delivered dose distributions (with tracking) were found in the selected worst scenarios: for all pts D0.03cc for OARs differ less than 0.5Gy and V100 for targets differ less than 1%. Fig.2 pictures an example of the worst expected effect on a representative slice

Results TPT (mean±SD) - made up of gantry and MLC mechanical movement, beam-on and gating time - for VML and VMC are: VML_BH=10.8±3.0 min, VLM_FB=11.1±2.9 min, VMC_BH=10.8±8.4 min, VMC_FB=7.7±5.3 min. GE results are depicted in Fig. 2: the red dotted lines represents the mean GE value for VML with FB (6.1 %, range 55% / -5.5%) and BH (39.1%, range 123.4% / -1.3%) dataset. No evidence

Conclusion The high efficiency of spine tracking for non-immobilized pts has been confirmed also in terms of dosimetric impact to the patient. OE seems to be independent from different factors such as target localization, reported pain and fraction number. Without tracking, the rate of OE≥2mm would not be negligible for painful pts