PTV Margins in MR-guided and Beam-gated SBRT of Liver Metastases: GTV Dose Escalation Can Reduce the Required PTV.

Abstract

Aims: Determining appropriate PTV margins for SBRT of liver metastases is a non-trivial task, especially with motion management included. The widely used analytical van Herk margin recipe (van Herk et al., 2000) could break down due to (i) a low number of fractions, (ii) non-Gaussian errors, or (iii) non-homogenous dose distributions. We evaluated the validity of the analytical margin recipe in this setting for two very different guidelines for SBRT of liver metastases in three fractions - one with a relatively homogenous dose within the PTV (British) and one allowing much steeper dose gradients within the PTV (Danish).

Materials and methods: We extracted sagittal motion traces for nineteen consecutive MR-guided and beam-gated treatments (57 fractions) on an MR-linac. We used these motion traces to calculate analytical van Herk GTV-to-PTV margins to account for intrafractional motion according to both British and Danish guidelines. We used the same motion traces to validate the analytical margins with motion-compensated dose accumulation in dose distributions obtained from British and Danish plans with varying PTV margins.

Results: Analytical margins for the British guidelines were 2.4 mm superior-inferiorly (SI) and 3.2 mm anterior-posteriorly (AP). For the Danish guidelines, analytical margins were 1.7 mm SI and 2.7 mm AP. Dose accumulation validation showed that a margin of 3 mm SI and 1.5 mm AP would have been sufficient for British plans to ensure 95% of the prescription dose to at least 99% of the GTV in 90% of the treatments (same criterion as used in the analytical calculation) of the patients. No PTV margin was needed to achieve the same with Danish guidelines.

Conclusion: GTV dose escalation can reduce the required motion-related PTV margins in SBRT with motion management. The van Herk margin recipe overestimates PTV margins in SBRT with inhomogeneous target dose distributions and becomes less applicable when the inhomogeneity increases.