KEY POINTS
- This retrospective single-center study from Copenhagen University Hospital–Rigshospitalet examined whether locoregional recurrences after definitive radiotherapy for oropharyngeal squamous cell carcinoma originated in unintentionally underdosed regions. Investigators included 40 patients with 54 biopsy-confirmed recurrence volumes following treatment between 2009 and 2017. Most patients were male (67.5%), 62.5% had p16-negative disease, and 50% received concurrent chemotherapy.
- Definitive treatment delivered 66–68 Gy in 33–34 fractions, generally using IMRT or VMAT with six fractions per week and daily CBCT-based positioning. Eligible patients were offered nimorazole, while patients with more advanced disease could receive weekly cisplatin 40 mg/m². Four of 40 patients (10%) underwent adaptive replanning, including one patient requiring three replans.
- To estimate the delivered dose, the investigators reconstructed approximately 33–34 daily synthetic CT datasets per patient by deformably registering planning CT to each treatment CBCT. Daily dose distributions were recalculated using Acuros XB, mapped back to the planning CT, and summed to obtain accumulated dose. Recurrence volumes were independently delineated on follow-up imaging and mapped to planning anatomy, with the recurrence point of origin determined using the nidus method.
- No systematic underdosage was detected at recurrence points of origin. Accumulated dose was higher than planned dose in 31 of 54 recurrence volumes (57%), while the mean accumulated-minus-planned difference was −0.04 Gy (SD 0.81 Gy; 95% CI −0.26 to +0.18 Gy; p=0.72). Individual differences ranged from −3.17 to +1.58 Gy, with no significant difference between primary-tumor and nodal recurrences (p=0.84).
- Sensitivity analyses using a second deformable registration platform supported the overall finding. The median distance between recurrence points mapped using Velocity and RayStation was 0.28 cm (IQR 0.20–0.52 cm). With RayStation-based recurrence localization, the mean accumulated-minus-planned dose difference remained small at −0.10 Gy (95% CI −0.33 to +0.12 Gy; p=0.36). However, local dose gradients within a 1-cm sphere around recurrence origins ranged from 0.76 to 22.05 Gy/cm, highlighting the potential effect of positional uncertainty in individual high-gradient regions.
- The observed differences were generally smaller than the dose deviations accepted under the relevant DAHANCA target-coverage objectives, which permit dose variation within approximately 95–107% of prescription. The findings therefore suggest that, under daily CBCT guidance, anatomical changes did not usually compromise dose at the central origins of subsequent recurrence. This challenges the assumption that preventing target underdosage is necessarily the principal benefit of adaptive head and neck RT; reducing normal-tissue exposure may be more relevant in some patients.
- Interpretation is limited by the absence of patients without recurrence, retrospective dose reconstruction, deformable registration uncertainty, and analysis of a single estimated recurrence origin rather than the entire recurrent volume. The second software did not independently reproduce the complete accumulation workflow, and intrafraction motion was not modeled. Importantly, the results do not exclude individual marginal misses or demonstrate that adaptive RT is unnecessary, particularly for patients with substantial anatomical changes.
CLINICAL TAKEAWAY
In patients with oropharyngeal cancer who subsequently developed locoregional recurrence, CBCT-based dose accumulation found no evidence that treatment failure was systematically associated with unintended underdosage at the recurrence origin. These findings suggest that biological radioresistance and tumor heterogeneity may be more important explanations for many in-field recurrences, while supporting a more selective role for adaptive RT focused on normal-tissue sparing and anatomically vulnerable cases.