KEY POINTS
- This two-institution retrospective planning study included 41 patients with a single brain metastasis treated using CyberKnife stereotactic radiotherapy. Twenty patients received 22 Gy in one fraction and 21 received 30 Gy in three fractions; mean target volume was 1.63 cm³.
- Functional neuroanatomy was automatically segmented from multimodal MRI, identifying 29 cortical, subcortical and brainstem regions within 3 cm of the target. Each clinical plan was then reoptimized with these structures added as avoidance objectives while maintaining the original target and conventional organ-at-risk constraints.
- Function-sparing optimization preserved target coverage, with variation in prescribed-dose target coverage <0.5% and no significant deterioration versus the clinical plans. Conventional organ-at-risk metrics and plan complexity also remained within clinical tolerances.
- Median mean-dose reductions were substantial across several structures: −39.0% for the caudate, −38.8% for the globus pallidus, −30.5% for the thalamus, −27.7% for the hippocampus and −22.8% for the putamen. Maximum-dose reductions followed a similar pattern, including −44.6% for the globus pallidus, −32.0% for the thalamus and −30.1% for the hippocampus.
- Not every structure benefited: dose changes were not significant for the cerebellum and amygdala for either mean or maximum dose, while several other regions showed significance for only one metric. After Holm–Bonferroni correction, reductions remained significant for 14 of 29 structures for maximum dose and 9 of 29 for mean dose.
- Deliverability was preserved, with median patient-specific QA gamma pass rates of approximately 99.2% versus 99.1% using 2%/1 mm criteria. However, no neurocognitive outcomes were assessed, clinically meaningful dose thresholds for many of these structures remain undefined, and segmentation of very small nuclei introduces additional uncertainty.
CLINICAL TAKEAWAY
Functional neuroanatomy can be incorporated into CyberKnife optimization for selected brain metastases and can meaningfully reduce dose to multiple potentially important brain regions without sacrificing target coverage. This is a dosimetric proof-of-concept rather than evidence that such sparing preserves cognition; prospective studies with validated functional endpoints are needed.