Omitting lateral setup shifts reduces prostate proton optimization time by 22% without compromising robustness

Removing lateral setup-shift scenarios reduced optimization time by 24 minutes across 16 prostate proton plans without significant changes in dosimetry or robust coverage.

KEY POINTS

  • This retrospective technical study evaluated whether lateral isocenter-shift scenarios parallel to the beam direction could be omitted during robust optimization of pencil beam scanning proton plans for prostate cancer. Investigators analyzed 16 clinical plans, including 14 intact-prostate and two prostate-bed cases, with seven plans involving pelvic lymph nodes, seven incorporating a simultaneous integrated boost and two requiring oblique beams because of hip prostheses.
  • The standard robust optimization approach used ±3.5% range uncertainty and ±5-mm setup perturbations along three anatomical axes, resulting in 21 optimization scenarios. Removing the two lateral setup-shift directions reduced this to 15 scenarios, while all other optimization settings were held constant. Importantly, independent robust evaluation retained the full set of setup and range uncertainty scenarios, including those omitted during optimization.
  • The dosimetric rationale is that translating the isocenter along the proton beam axis has minimal influence on water-equivalent path length, unlike shifts perpendicular to the beam. In water-phantom simulations, a clinically representative 5-mm beam-axis displacement changed target D95 by approximately 0.4%, while paired opposed lateral fields provided additional compensation of dose changes within the target.
  • Across the 16 reoptimized clinical plans, omitting lateral shifts reduced mean single-cycle optimization time from approximately 90 to 66 minutes, an average saving of 24 minutes (22%; p=0.004). Individual time reductions ranged from 1 minute (8.3%) to 129 minutes (43.1%), demonstrating substantial variability in computational benefit depending on plan complexity and optimization workload.
  • Dose-volume metrics for normal tissues remained essentially unchanged. Mean differences after omission were 0.14 percentage points for bladder V70, 0.09 points for rectum V70, 5.88 cGy(RBE) for sigmoid D2cc, and 8.92 cGy(RBE) for bowel D2cc. None of the evaluated OAR differences reached statistical significance, although absence of significance in a small cohort does not establish formal dosimetric equivalence.
  • Full-scenario robustness assessment showed negligible changes in worst-case target coverage, with a mean difference in CTV V95 of −0.01 percentage points and a maximum observed deterioration of 0.32 points. Verification-CT recalculations showed a mean difference of +0.01 points. Plans meeting institutional robustness criteria before scenario reduction continued to do so afterward; two plans with intentional pre-existing robustness compromises retained similar limitations after reoptimization.
  • The authors explicitly caution against generalizing the method to nonlateral or oblique beam arrangements, including plans modified for hip prostheses or requiring posterior beams. Although the two oblique-beam examples showed no significant deterioration, their number was insufficient to establish safety. The analysis was also restricted to one institution using RayStation v11a and an IBA Proteus Plus beam model, and the reported time savings describe a single optimization cycle rather than the complete planning workflow.

CLINICAL TAKEAWAY

For prostate proton therapy delivered with opposed lateral fields, omitting beam-axis setup shifts during robust optimization may improve planning efficiency without materially affecting the evaluated target coverage or OAR doses. The approach remains geometry-specific and should be supported by local validation and full-scenario robust evaluation; it should not be automatically extended to oblique fields or other treatment sites.

SOURCE

Journal of Applied Clinical Medical Physics

Browse more research Suggest a correction