Lexicographic Optimization-Based Planning for Single and Multiple Brain Metastasis Radiosurgery with Coplanar Arcs

Abstract

This study validated a not yet commercially available fully-automated lexicographic optimization planning system (LOps) for single and multiple lesions intracranial stereotactic radiosurgery (SRS). Forty-four consecutive SRS treatment plans (21 Gy/1 fx) delivered between November 2019 and August 2022 were retrospectively selected and automatically re-planned by LOps: 25 of them has 1 lesion, 13 had 2 lesions, 4 had 3 lesions, and 2 had 4 lesions. An a-priori assigned priority list, Wish List (WL), was used to define the sequential LO: 4 patient sets (tuning set) were necessary to tune each WL, for single lesion-plans (SLp) and multiple lesion-plans (MLp). While in manual plans (MP), the arc setup is freely chosen, the WL was tuned to use 2 coplanar arcs of 140° and 360° for SLp and MLp, respectively. A 0° and 90° collimator rotation were set for counter-clockwise and clockwise arcs, respectively. The fluence optimization is followed by a Monte Carlo calculation (MCc) with 1 mm-dose grid and 0.5%-statistical uncertainty. A target coverage as high as possible was requested, with at least 80% of the prescription dose covering 99% of the PTV. The main criteria for SLp approval were a brain V12Gy <10 cm3. In MLp this criterion can be overcome to get the minimum target coverage. The remaining 36 SRS plans (21 SL and 15 MLp) were automatically re-planned (testing set). Testing plans were compared in terms of dose-volume constraints, conformality, and monitor units (MUs). Statistical significance was assessed by performing the Wilcoxon test and plan delivery accuracy was verified by pre-treatment QA. WLs-tuning took 3 days. Overall manual and automatic MCc time can be estimated at 8 hours and 3 hours, respectively. Statistically significant increases in SLp and MLp target coverage (GTV_D98%: +3.0% SL, +5.7% ML, PTV_D98%: +4.4% SL, +13.4% ML) and conformity index were registered. Automatic plans showed acceptably higher median brain V12Gy (SL: MP 7.2 cm3, SLp 7.6 cm3; ML: MP 8.7 cm3, MLp 10.3 cm3). The SLp registered a lower median MU number (-4.2%) while MLp were obtained with a higher median number of MU (+9.8%). This not statistically significant difference did not affect gamma passing rates. The novel LOps produced high-quality clinically acceptable SRS SL and ML plans with coplanar arcs according to institutional-specific planning protocols, significantly reducing the overall planning time from about one working day for one MP to about 3 hours for one automatic plan. Together with comparable OAR sparing, the target coverage was significantly increased and the plan deliverability preserved.