3091 - Evaluating Liver SBRT Planning with Dynamic Collimator Rotation and Automated Static Angle Modulated Ports Selection

Purpose/Objective(s): To evaluate improvements in external beam radiotherapy plan quality and delivery efficiency for liver treatments from a novel planning solution (VMATp) which inversely-optimizes plans using a new GPU based optimizer and can include dynamic collimator rotation and/or STatic Angle Modulated Ports (STAMPs).

Materials/Methods: 10 liver cases previously treated to 50Gy/5fx were re-planned 4 times using Volumetric Modulated Arc Therapy (VMAT) with or without the new features (VMATp). For each case, the clinically used partial arcs were reoptimized using the new optimizer, rotating collimator, and fully automated selection for STAMP #, location, and weighting. Optimization objectives were fine-tuned on this plan to achieve highest plan quality defined by a standardized set of 18 dose goals. This plan (VMATp_2arc) was then reoptimized with the same objectives but only the first field (VMATp_1arc) to evaluate trade-offs in plan quality and delivery efficiency. To isolate the impact of the novel optimization algorithm, a third plan was optimized with the same objectives using static collimator angles and a single unweighted STAMP (VMATp_OptAlg). The clinical plan (VMAT_Clin) was reoptimized using the same set of objectives to evaluate if metrics could be reached without the novel features (VMAT_New). Plans were compared for changes in plan quality and delivery efficiency. Planners also labeled plans as better/equivalent/worse compared to VMAT_Clin, and which overall was best.

Results: VMATp_2arc used 1-3 STAMPs per arc (average: 4.4 total, VMATp_1arc average: 2.3). Compared to VMAT_Clin, VMATp_2arc and VMATp_OptAlg slightly increased MU (+335.7 MU, +229.4 MU) though impact on Delivery Time was negligible (+21s, +4s); VMATp_1arc negligibly affected both (-26 MU and -10s). All VMATp plans were rated equivalent or better, with all four of the new plans rated better for one patient and VMATp_2arc and VMATp_OptAlg rated better for a second. VMATp plans were preferentially selected as overall best, with no clear preference for any technique: VMATp_2arc (3/10), VMATp_1arc (2/10), VMATp_OptAlg (4/10), VMAT_new (1/10). Of 166 total metrics evaluated (up to 18 per case), VMATp_2arc rescued 8 from out-of-tolerance, while VMATp_1arc, VMATp_OptAlg, and VMAT_New each rescued 6-7. The greatest changes occurred in ITV V110% (+22-30%) and D0.03cc (+2.6-6.2Gy). No changes were statistically significant when evaluated with a paired p-test using Benjamini-Hochberg multiplicity correction, suggesting hotter target doses were achieved without compromising OAR metrics.

Conclusion: A novel planning solution with fully automated parameter selection created Liver SBRT plans of equivalent or better quality to standard VMAT without significantly increasing delivery times. Future studies will explore potential for further dosimetric improvements.