Purpose: Multiple techniques are available for stereotactic radiosurgery (SRS) treatment of brain metastases on either a linac or Gamma Knife (GK) machine. Aside from the dosimetric differences, these techniques differ from each other in the magnitude of machine uncertainties inherently present and subsequently in the errors those uncertainties may cause. This study was designed to compare the dosimetric robustness of frameless GK, Elements, HyperArc and VMAT techniques as a function of various machine uncertainties.
Methods: 16 patients previously treated for multiple brain metastases were retrospectively identified. 5 SRS plans were generated per patient: frameless GK, Elements, HyperArc, and two VMAT plans by independent planners. All plans met clinical planning standards. PTVs were defined as 1mm isotropic expansion of GTVs. We simulated dosimetric perturbations caused from machine uncertainties by applying a range of translational/rotational displacements to PTVs, up to maximum magnitudes of 0.1 mm/0.3 deg and 2 mm/0.5 deg for frameless GK and linac-based plans, respectively. The impact on V100% (V100), normalized D99% (nD99), and the Paddick Conformity Index (PCI) were assessed and compared for PTVs among plans. The one-way ANOVA and Tukey tests were performed for statistical analysis with a significance level of p<0.05.
Results: The mean V100, nD99, and PCI accounting for machine uncertainties were all significantly different (p<0.001) among the plans. Overall, the GK Frameless plans demonstrated the best robustness across the metrics (V100=0.877; nD99=1.180; PCI=0.962), which could be explained by the much tighter machine tolerance for GK. Among the Linac-based plans, Elements (V100=0.840; nD99=1.120; PCI=0.740) was more robust against machine uncertainties than the HyperArc/VMAT plans. Smaller variances were also observed within the other VMAT plans.
Conclusion: Our results indicate that GK and Elements are more robust against machine uncertainties than the other common linac-based SRS treatment techniques.