CBCT-based dose tracking in head and neck proton therapy: analysis of dose deviation for adaptive re-simulation assessment

Abstract

Proton therapy has been increasingly utilized in head and neck cancer (HNC) patients due to its unique property of the Bragg Peak. However, proton dose delivery is highly sensitive to positional variations and anatomical changes occurring during the treatment course. These changes can lead to deviations between planned and delivered doses, potentially compromising target coverage and increasing exposure to surrounding organs at risk (OARs). This study aims to evaluate fraction-by-fraction and accumulated target dose deviations using the predefined thresholds, and to assess whether accumulated dose changes warrant adaptive re-simulation. Methods: HNC patients treated with pencil beam scanning (PBS) proton therapy were retrospectively analyzed using daily cone-beam CT (CBCT) imaging. All patients received 33 fractions using a simultaneous integrated boost (SIB) technique, delivering different dose levels to high-risk (CTV70) and low-risk clinical target volumes (CTV60 and CTV54). The delivered dose for each fraction was recalculated on synthetic CBCT images and deformably accumulated to assess cumulative dose differences relative to the treatment plan. Dosimetric parameters, including D98% for target volumes and selected OAR dose metrics, were evaluated. For dosimetric change assessment, a decrease of 3% was used as the threshold for high-risk CTV, and 5% for low-risk CTVs. Results: CBCT-based dose tracking revealed measurable variations in both target coverage and OAR doses. Intermittent reductions in D98 were observed in both high-risk and low-risk target volumes, with greater variability in CTV54. At the fraction level, 30% of fractions in CTV70 exceeded the predefined -3% threshold, with a mean DD98 of -2.71% ± 2.55. In CTV54, 12.7% of fractions exceeded the -5% threshold, with a mean DD98 of -2.69% ± 2.99. Despite these inter-fractional variations, accumulated dose analysis showed that only two patients in the high-risk CTV exhibited deviations beyond the threshold, while low-risk CTVs remained within acceptable limits. Conclusion: CBCT-based dose tracking is a useful tool in head and neck proton therapy. Although inter-fractional dose deviations were observed, the accumulated dose deviations remained limited at the population level. Therefore, not all head and neck patients require adaptive re-simulation. Instead, individualized dose tracking remains important to identify patients who may benefit from adaptive re-planning. This approach enables early detection of clinically significant dose deviations from the planned dose while avoiding unnecessary re-simulation.