Gradient partial-overlap three-arc VMAT for hippocampal-avoidance whole-brain radiotherapy: a dosimetric and robustness trade-off analysis.

BACKGROUND: Hippocampal-avoidance whole-brain radiotherapy (HA-WBRT) with coplanar VMAT is frequently hindered by the "island-blocking" phenomenon, which limits hippocampal sparing. METHODS: We developed a gradient Three partial-overlap VMAT (ols3-VMAT) strategy using asymmetric superior/inferior arcs and a dedicated medial arc to distribute the modulation burden. In 20 patients, ols3-VMAT was compared against c-VMAT, s-VMAT, and ols2-VMAT regarding target quality, OAR sparing, setup robustness, and delivery efficiency. Additionally, multivariable spatial analyses explored the influence of hippocampal volume, Z-thickness, and bilateral centroid distance on Dmax. RESULTS: ols3-VMAT yielded a median conformity index of 0.88 and homogeneity index of 0.21. Its median hippocampal Dmax was 12.1 Gy, compared with 15.2 Gy for c-VMAT, 13.5 Gy for s-VMAT, and 13.6 Gy for ols2-VMAT. It maintained compliance with the RTOG 0933 Dmax<16 Gy criterion in all evaluated cases under the simulated 1.0-mm translational shift. The median MU was 1323 for ols3-VMAT, representing a 20.7% reduction relative to s-VMAT (1668 MU), but a 57.7% increase relative to c-VMAT (839 MU) and a 30.0% increase relative to ols2-VMAT (1018 MU). Multivariable models (adjusted R2 0.336-0.553) confirmed that hippocampal volume is a key predictor of absolute Dmax, while the relative dosimetric gains (ΔDmax) of the optimized strategies remained stable across anatomical variations. CONCLUSION: ols3-VMAT is a robust, efficient, and purely coplanar HA-WBRT strategy that overcomes the dosimetric ceiling of conventional techniques.