2953 - Does Online Adaptive Dose Restoration In Head and Neck Proton Therapy Alter Linear Energy Transfer Under Anatomic Variation?

Purpose/Objective(s): Interfractional anatomic variation in head and neck intensity-modulated proton therapy (IMPT) can degrade planned dose distributions, motivating online adaptive proton therapy (OAPT). However, plan adaptation modifies beam parameters, potentially altering dose-averaged linear energy transfer (LET_d) and introducing uncertainty in relative biological effectiveness (RBE)-weighted dose. We hypothesized that OAPT performed to restore target coverage under daily anatomic variation would significantly alter LET_d distributions in head and neck IMPT.

Materials/Methods: Eighteen head and neck patients previously treated at one institution were analyzed retrospectively. For each patient, a three-field IMPT plan was created to deliver 70 Gy and 57 Gy to the high- and low-risk clinical target volumes (CTVs), respectively. Daily cone-beam computed tomography (CBCT) images were used to generate a synthetic CT for each fraction to represent the daily anatomy. Treatment delivery was calculated on each daily anatomy using Monte Carlo to simulate both non-adaptive (NA) and online adaptive (OA) treatment. For NA, the nominal IMPT plan was calculated on the daily anatomy without modifying plan parameters. For OA, beam parameters were re-optimized using a validated in-house OAPT workflow aimed at restoring target coverage, assuming RBE = 1.1. Key dose-volume metrics, as well as mean and maximum LET_d, were extracted for CTVs and organs at risk. Fraction-wise median values were compared between NA and OA using Wilcoxon signed-rank tests.

Results: OA significantly improved target coverage (increased D_98% and decreased D_2% for CTVs) and reduced spinal cord D_max compared with NA (all p < 0.01; Table 1). Despite these dose differences, LET_d changes were statistically significant in only 3 of 16 evaluated endpoints, with clinically negligible effect sizes (highest absolute ? of 0.13 keV/µm for left parotid maximum LET_d). All remaining differences in LET_d metrics were non-significant.

Conclusion: Contrary to our hypothesis, OAPT did not meaningfully alter LET_d distributions despite significant improvements in dose endpoints, particularly target coverage. LET_d remained stable under anatomic variation, suggesting that the nominal plan is a reliable predictor of LET_d. These findings reduce concern that OAPT-driven dose restoration introduces RBE-weighted dose uncertainty in head and neck IMPT.