Journal: Radiation oncology (London, England)
Study type and population
- Prospective, single‑center, observational cross‑sectional study of 24 patients with unresectable locally advanced colon cancer.
- Simulated 1.5T MR‑linac SBRT workflow; no patients were actually treated per the proposed protocol.
- Feasibility/methods study linked to an ongoing phase I trial (NCT06244537).
Aim
- Quantify intrafraction tumor motion noninvasively using MR‑linac cine‑MRI.
- Evaluate whether motion measurements can be translated into individualized, anisotropic residual‑motion margins for SBRT planning.
Methods
- RBME algorithm development and validation
- Developed and internally validated a rolling‑ball‑based motion extraction (RBME) algorithm on a dynamic phantom on the same 1.5T MR‑linac.
- Phantom validation: Bland–Altman mean bias 0.01 mm, SD 0.49 mm, 95% limits of agreement −0.98 to 0.95 mm (≈±1 mm accuracy).
- Application to patient cine‑MRI
- Applied RBME to patient cine‑MRI to derive axis‑ and patient‑specific motion envelopes (cranio‑caudal, anterior‑posterior, left‑right).
- Analyzed intrafraction motion with repeated‑measures statistics, testing for interactions between motion axis and tumor segment/location.
- Simulated SBRT planning
- Prescription: 25 Gy in 5 fractions.
- Simulated PTV = GTV + RBME‑derived anisotropic residual‑motion margin + 3 mm setup margin.
Key motion findings
- Directional bias
- Motion was directionally biased: larger excursions caudal/right/anterior; smaller cranial/left/posterior.
- Average mean excursions centered around 4–5 mm.
- Displacement metrics across all lesions
- RMS displacement (mean ± SD)
- Cranio‑caudal: 2.64 ± 1.32 mm
- Anterior‑posterior: 2.91 ± 1.05 mm
- Left‑right: 2.75 ± 1.43 mm
- 95th percentile displacement (P95)
- Cranio‑caudal: 4.49 ± 2.43 mm
- Anterior‑posterior: 4.97 ± 1.74 mm
- Left‑right: 4.76 ± 2.62 mm
- RMS displacement (mean ± SD)
- Location‑dependent motion
- Significant axis‑by‑segment interactions for both RMS (χ² 19.93, df 8, P=0.011) and P95 (χ² 20.44, df 8, P=0.008).
- Indicates that motion magnitude and anisotropy depend on tumor location in the colon.
Conclusions and implications
- RBME performance: provides phantom‑validated intrafraction motion quantification on a 1.5T MR‑linac with ~±1 mm accuracy.
- UNLACC motion characteristics: irregular, anisotropic, and location‑dependent intrafraction motion with typical excursions around 4–5 mm and notable directional bias.
- Clinical implications: the derived motion envelopes and the concept of individualized anisotropic residual‑motion margins (plus a small setup margin) are hypothesis‑generating and intended to inform further methodological refinement and protocol design in the ongoing phase I SBRT study, rather than to define current standard margins.