A Framework for Assessing Plan Quality In Spatially Fractionated Radiation Therapy
Abstract
Purpose
Spatially fractionated radiation therapy (SFRT) aims to deliver high-dose peaks within tumors while maintaining low-dose valleys. However, commonly used peak-to-valley ratios or D5/D95 and D10/D90 metrics derived from gross tumor volume (GTV) DVHs fail to account for spatial dose heterogeneity and geometric constraints. We explored a structure-specific, multi-metric framework to evaluate SFRT planning quality and distinguish acceptable from suboptimal plans.
Methods
SFRT treatment plans were retrospectively generated for 11 patients prescribed 15Gy to the combined vertex volume (PTV_ALL) and 3Gy to the GTV with 95% coverage using uniformly placed lattice vertices. Peak dose was characterized using D10 of PTV_ALL. Valley regions were defined using two tumor substructures: GTV_Control, defined as GTV minus a 6mm expansion of PTV_ALL, and an intermediate-distance valley shell (2CM_Control), defined as the region between 6mm and 2cm from PTV_ALL within the GTV. A local peak-to-valley dose ratio (PVDR_local) was defined as D10(PTV_ALL)/D90(2CM_Control).
Results
The average (range) number of vertices, vertex diameter, minimum inter-vertex separation (edge-to-edge), and vertex density (number of vertices ×100cm³/GTV volume) were 9.9±5.6 (3–20), 1.6±0.4cm (0.9–2.0), 2.4±0.9cm (1.2–4.0), and 1.8±1.5 (0.8–5.3), respectively. Mean D90(2CM_Control) was 3.4±0.3Gy (3.0–3.9). PVDR_local ranged from 4.5 to 6.0 (mean 5.3 ± 0.8). Mean dose to the 2CM_Control and GTV_Control regions was 5.7±0.9Gy and 5.2±0.8Gy, respectively, while the conventional GTV-based D10/D90 ratio was 3.3 ± 0.3(2.7–3.6). PVDR_local demonstrated a stronger inverse relationship with valley dose than the GTV-based ratio, with higher values consistently associated with reduced 2CM_Control dose. Plans with PVDR_local 6.3Gy), a trend not observed using D10/D90 of the GTV.
Conclusion
Conventional PVDR or D10/D90 metrics alone are insufficient to characterize SFRT plan quality. A structure-specific, multi-metric framework incorporating vertex coverage, PVDR_local, and mean valley dose measures could provide a practical and reproducible approach for guiding SFRT planning and evaluating plan quality.