Spatial collimation sequencing enhances low-dose sparing and geometric robustness in single-isocenter SBRT for...

Effective stereotactic body radiotherapy (SBRT) for multiple liver metastases (MLM) must balance ablative tumor dosing with minimization of normal liver irradiation to mitigate radiation-induced liver disease. Conventional single-isocenter volumetric modulated arc therapy (VMAT) often leads to excessive low-dose exposure due to the "island blocking" effect and geometric vulnerability to rotational errors. To address these challenges, we propose a novel strategy that spatially clusters liver lesions to reduce inter-target leakage and enhance delivery robustness.

This retrospective study included 20 patients previously treated with SBRT (40 Gy in 5 fractions). For each patient, three RapidPlan-based automated VMAT plans (APs) were generated: conventional single-isocenter (rSiPlan), multi-isocenter (rMiPlan), and the proposed spatial collimation sequencing-based single-isocenter VMAT plan (rSCSPlan). In rSCSPlan, lesions were grouped based on craniocaudal distribution and beam's eye view projections, with each group treated using dedicated arcs and a fixed 90° collimator angle to minimize unnecessary jaw aperture and MLC leakage. Manual counterparts (mSiPlan, mMiPlan, mSCSPlan) were also created. Dosimetric comparisons were conducted using conformity index (CI), gradient index (GI), and mean dose to normal liver (mean Liver-GTVs). To assess geometric robustness, 6-degree-of-freedom (6DoF) baseline shifts and rotational errors were applied to rSiPlan and rSCSPlan. Target coverage degradation (D95 and V95) was evaluated via generalized estimating equation (GEE) regression, considering target volume and isocenter distance.

APs outperformed manual counterparts in low-dose sparing and delivery efficiency. While rSiPlan showed slightly better CI and GI than rSCSPlan, it also exhibited the higher mean Liver-GTVs and susceptibility to rotational setup errors. rSCSPlan significantly reduced normal liver dose compared to rSiPlan (mean Dmean: 10.3 Gy vs. 10.7 Gy, p < 0.001) and showed the lowest normal tissue complication probability (NTCP). Under simulated 6DoF errors, rSCSPlan maintained greater target coverage stability: at 1.5° rotation, V95 loss was markedly lower than in rSiPlan (difference of 10.4%, p < 0.05). GEE modeling confirmed that both target volume and isocenter distance significantly influenced robustness, with rSCSPlan demonstrating reduced sensitivity across all tested perturbations.

The rSCSPlan strategy represents a clinically feasible and dosimetrically advantageous approach for MLM. By mitigating the island blocking effect and enhancing spatial robustness, it achieves superior liver sparing and improved error resilience. These advantages support its clinical implementation as a personalized treatment solution for patients with MLM.