Averaged anatomy CT in weekly treatment quality control and offline adaptive intensity modulated proton therapy of head and neck cancers

Adaptive intensity modulated proton therapy (IMPT) for head and neck cancer commonly relies on weekly repeated CT scans (rCTs) to assess anatomical changes and determine whether plan adaptation is required. However, a single rCT often fails to represent the daily anatomical variability observed on daily CBCTs, particularly in soft tissue regions subject to physiologic motion such as the tongue base, and supraglottic airway. This work evaluates an averaged‒anatomy CT (Ave_rCT) generated from deformably registered rCT and multiple preceding daily synthetic CTs (sCTs), following an approach previously validated in photon therapy. Twelve oropharyngeal cancer patients with mid‒treatment adaptation were retrospectively reviewed. Adaptive plans generated using Ave_rCT (AveAP) were compared with clinical adaptive plans (ClinAP) based on a single rCT. Dosimetric evaluation on subsequent rCTs showed comparable robust voxmin CTV D98, voxmax body Dmax and serial OAR D0.03cc, and predicted ΔNTCP between the two methods. The Ave_rCT tended to resemble subsequent weekly anatomy more closely than an individual rCT resulting in relatively smaller inter‒patient variability in target dose coverage and hot spot across post‒adaptation time points. Ave_rCT may improve the robustness of adaptation decision‒making by providing a more representative and stable anatomical reference compared to current practice of relying on single rCT. It appears to be a practical and efficient alternative to single or multi‒CT robust optimization in the offline-adaptive IMPT of head and neck cancers.