Abstract / Summary
Background: Single-isocenter stereotactic radiosurgery enables efficient treatment of multiple brain metastases (SI-MBM SRS), but demands high geometric accuracy. Plan complexity metrics are increasingly used as indicators for quality assurance (QA) performance, however their applicability to SRS remains uncertain, particularly in the context of multi-institutional variability.
Purpose: This study evaluated the relationship between plan complexity, target geometry, and end-to-end dosimetric QA outcomes for SI-MBM SRS across multiple institutions.
Methods: Forty-two SI-MBM SRS plans from different centers and platforms were delivered to polymer gel phantoms, providing high-resolution 3D dose measurements. Gamma passing rates (GPRs) were calculated under 3%/2 mm, 5%/2 mm and 5%/1 mm criteria and were correlated with eleven established complexity metrics calculated per plan. Geometric factors, including target equivalent diameter and distance-to-isocenter, were analyzed. Receiver-operating-characteristic (ROC) analysis was performed to identify optimal thresholds for predicting QA pass/fail (≥90% GPR).
Results: No statistically significant differences between the different linacs and treatment planning systems were found, nor strong or consistent correlations between complexity metrics and GPRs. In contrast, geometric parameters were more influential: off-axis distance and target size significantly affected QA performance, with the largest differences observed for far-off-axis lesions. ROC analysis identified optimal thresholds of 34.9 mm for distance-to-isocenter and 4.8 mm for equivalent diameter in predicting QA outcomes (AUC∼0.60-0.65), although the predictive performance remained modest.
Conclusions: In this multi-institutional, gel-based end-to-end study of SI-MBM SRS, target geometry was a stronger QA performance predictor than complexity metrics. These findings emphasize the importance of geometry-aware QA strategies and the need for further standardized, multi-institutional evaluations to clarify the interplay between complexity, geometry, and machine performance in SRS.