Impact of Online Adaptive Proton Therapy on Linear Energy Transfer In Head-and-Neck Cancer Under Daily Anatomic Variation
Abstract
Purpose
To quantify the effect of interfractional anatomic variation on dose-averaged linear energy transfer (LETd) in head-and-neck intensity-modulated proton therapy (IMPT), and to determine whether daily adaptive proton therapy (APT) further alters LETd compared with non-adaptive treatment. This is clinically relevant because LETd variations may translate into uncertainty in relative biological effectiveness (RBE)-weighted dose.
Methods
Three-field IMPT plans were created retrospectively for 18 head-and-neck patients, with a total of 576 daily cone-beam computed tomography (CBCT) images acquired in treatment position. For each fraction, a synthetic CT was generated using deformable image registration between planning CT and CBCT to represent the daily anatomy. Two treatment scenarios were forward-calculated using Monte Carlo: (i) non-adaptive (NA) delivery by calculating the nominal IMPT plan on the daily anatomy, and (ii) online adaptive (OA) delivery using a validated in-house APT workflow aimed at dose restoration under constant RBE = 1.1. The resulting dose and LETd distributions were compared across anatomical structures by extracting clinically relevant dose-volume metrics, mean LETd, and voxel-wise maximum LETd.
Results
Despite large discrepancies between NA and OA dose metrics, particularly in clinical target volume (CTV) coverage, differences in LETd metrics were minimal. High-risk CTV D98% [min–mean–max] was 77.4–94.4–98.0% (NA) versus 95.4–97.4–98.3% (OA). Low-risk CTV D98% showed even larger differences: 65.6–91.1–96.9% (NA) versus 94.7–96.7–97.7% (OA). Conversely, mean LETd [min–mean–max] in the high-risk CTV was 1.37–1.74–2.08 keV/μm (NA) versus 1.33–1.71–2.02 keV/μm (OA), and maximum LETd was 1.93–2.64–3.55 keV/μm (NA) versus 1.81–2.67–4.11 keV/μm (OA). LETd values in the low-risk CTV were comparable.
Conclusion
LETd remained stable under interfractional anatomic variation and online adaptation, suggesting that for most head-and-neck patients the nominal treatment plan can serve as a reliable predictor of LETd distribution. For outlier cases with large anatomic changes, monitoring LETd of online-adapted plans may still be warranted.