Purpose: To compare daily online plan adaptation with offline treatment re-planning for head and neck intensity-modulated proton therapy (IMPT).
Methods: IMPT treatment plans are created for a retrospective cohort of eight head and neck cancer patients with daily cone-beam CT (CBCT) images (median=33 fractions) and a re-scan CT acquired during the course of treatment due to large anatomy changes. Treatment plans are created based on the nominal planning CT. Dose tracking of the IMPT treatment is performed for three different delivery scenarios: re-planning with a 3 mm planning margin (Re-PTV), re-planning without margins (Re-CTV), and online adaptation without margins (OA). Both re-planning scenarios are realized by delivering the initial plan until the re-scan CT was acquired, at which point a completely new plan is created that fulfills the same clinical goals. For OA, the initial CT plan is adapted at each fraction using an in-house developed online adaptation workflow based on Monte Carlo dose calculations on scatter-corrected CBCTs. Plan adaptation is achieved by tuning the weights of a subset of beamlets without changing their energies or positions. Planning contours are propagated from the CT to each daily CBCT using deformable image registration. The same propagated contours are employed for dose tracking and accumulation throughout the treatment.
Results: In terms of target coverage, OA outperformed both re-planning scenarios. For the low-risk CTV, the mean D₉₈ for individual fractions was 91.6%, 94.3%, and 98.4% of the prescribed dose for Re-CTV, Re-PTV, and OA, respectively. For the high-risk CTV, 97.0%, 98.2%, and 98.3% were acquired for the same metrics. In terms of OAR sparing, OA kept the dose comparable to Re-CTV while reducing the dose compared to Re-PTV.
Conclusion: This study demonstrates the superiority of online IMPT plan adaptation over offline treatment re-planning in the context of target coverage and OAR sparing.