Benchmarking intermediate dose spillage in single target stereotactic radiosurgery for brain metastases

Intermediate-dose spillage is central to normal-brain sparing in stereotactic radiosurgery, yet fixed adverse-event and relative indices do not define achievable plan quality for single-target linear accelerator treatments. The aim was to derive a volume-dependent absolute benchmark for the 50% prescription isodose volume (V_50%; equivalent to V_{10 Gy} for 20 Gy plans) for brain metastases.

In total, 276 single-isocenter single-target plans were analyzed: 178 idealized model plans and 98 clinical plans. Power-law regression in log-log space modeled spillage against planning target volume (PTV, cm³). The volume-scaling exponents were 0.83 for the ideal references and 0.75 for the simplified clinical threshold. The clinical upper threshold was derived using q = 0.90 quantile regression. Threshold-exceeding plans were reviewed for anatomical or technical justification.

In the model cohort V_50% value correlated strongly with PTV (r = 0.994, p < 0.001). The ideal mean model was V_50% = 3.71 × PTV^0.83, the conservative ideal-planning reference was V_50% = 4.10 × PTV^0.83, and the clinical upper threshold was V_50% = 5.00 × PTV^0.75. Ten of 98 clinical plans (10%) exceeded the threshold. Seven lacked anatomical justification and were replanned, reducing V_50% by a median of 14% (range 7-29%) without compromising target coverage.

A volume-based V_50% benchmark provided a complementary quality-assurance reference for single-target linear accelerator radiosurgery for brain metastases. It flagged plans that warranted structured review alongside target coverage, organs of interest, conformity, and gradient metrics.