Single- and Multi-CT Robust Optimization Improves Robustness to Anatomic Changes in Head and Neck Cancer Patients Treated with Intensity-Modulated Proton Therapy

Abstract

We hypothesized that robust optimization (RO) can improve dose distribution robustness to not only setup/range uncertainties, but also compensate for anatomic changes (e.g. weight loss and tumor regression) in the delivery of intensity-modulated proton therapy (IMPT) for head and neck (HN) cancer. Furthermore, we theorized that employing multi-CT RO (MCRO) using a digitally modified planning CT (modCT) to simulate patient weight loss would improve robustness. We simulated 10 treatment plans in silico for HN patients who clinically required re-planning due to plan degradation/heterogeneity. All plans included the primary site and bilateral cervical lymph nodes. Five planning approaches were applied to both the simulation CT (simCT) and the rescan CT (reCT) for two representative patients who required re-planning (Pt1 – weight loss, Pt2 – tumor regression). Treatment approaches included nonRO-4-field (4F) (clinical plan) (RAO, LAO, RPO, LPO), RO-4F, RO-3-field (3F) (RAO, LAO, PA), MCRO-4F, and MCRO-3F. RO accounted for 5 mm isocenter position and 3.5% range uncertainty. MCRO was planned to be robust to simCT and modCT, where the superficial 5 mm of tissue on simCT was density overridden to air density to simulate weight loss. Plans were optimized to achieve nearly identical target coverage as the non-RO-4F while similarly optimizing dose to organs at risk (OARs). Dosimetric changes to target coverage and OARs from simCT to reCT were compared. We also evaluated the percentage of V105/V100 as a measure of heterogeneity. Dosimetric outcomes are outlined in the table below. While target coverage and OAR doses remained relatively stable, RO and MCRO techniques dramatically improved plan stability and homogeneity. RO against setup and range uncertainty also compensates for anatomic and tumor changes in HN patients treated with bilateral IMPT. MCRO does not result in inferior dosimetry despite removal of a 5 mm rim of traversed tissue. These results warrant further study in a larger cohort with additional modCT/MCRO techniques. For Pt1, the original plan resulted in a V105/V100 increase from 4.1% to 35.7% (Δ=31.6%) from simCT to reCT, respectively. RO and MCRO plans resulted in more robust plans with less heterogeneity: RO-4F (Δ=24.0%), RO-3F (Δ=33.2%), MCRO-4F (Δ=32.6%), and MCRO-3F (Δ=31.2%). For Pt2, a similar trend was noted: Δ for nonRO-4F, RO-4F, RO-3F, MCRO-4F, MCRO-3F was 15.8, 7.5, 7.9, 11.7, and 10.5%, respectively.Abstract TU_35_3242: Table 1Dosimetric Outcomes for Various Treatment ApproachesNon-RO-4FreCTRO-4FMCRO-4FRO-3FMCRO-3FPt. 1Dmax (%)109.1125.1111.2110.2111.3109.1D1cc (%)106.9119.3108.9107.9107.7107.1V105(cc)23.2223.859.918.412.425.5V105/V100 (%)4.135.711.73.12.54.5Pt 2Dmax (%)106.6123.0111.5110.1112.7112.0D1cc (%)104.8114.4107.7107.3107.9108.0V105(cc)3.533.4534.832.716.121.7V105/V100 (%)0.115.98.48.04.25.4 Open table in a new tab