KEY POINTS
- This single-centre retrospective study included 52 patients with stage II–IV nasopharyngeal carcinoma treated with volumetric modulated arc therapy. A total of 312 weekly cone-beam computed tomography scans were converted to synthetic computed tomography for automated segmentation and Monte Carlo dose recalculation.
- A nodal target volume reduction of at least 8% at week 2 predicted the predefined need for replanning with an area under the curve of 0.83, 75% sensitivity, and 94% specificity. The corresponding odds ratio was 2.03 (95% confidence interval, 1.23–3.35; p=0.006).
- For primary targets, a volume reduction of at least 10% at week 4 produced an area under the curve of 0.76, with 54% sensitivity and 98% specificity. A planning target volume reduction of at least 7% at week 4 achieved an area under the curve of 0.84.
- Organ-at-risk triggers were structure-specific: at least 15% contralateral parotid reduction at week 2 yielded an area under the curve of 0.89, while at least 19% left submandibular gland reduction at week 3 yielded an area under the curve of 0.97. A 4% pharyngeal constrictor volume loss at week 2 predicted a subsequent dose increase of at least 0.9 Gy with an area under the curve of 0.93.
- Nine of 52 patients (17%) experienced a reduction exceeding 3 Gy in target D98 at least once during treatment. The study simulated monitoring and replanning indications but did not prospectively implement adaptation or evaluate toxicity, disease control, or patient-reported outcomes.
CLINICAL TAKEAWAY
Weekly synthetic computed tomography could help centres using conventional C-arm linear accelerators identify patients who may benefit from offline adaptive radiotherapy, with different structures requiring assessment at different treatment weeks. However, the thresholds were derived retrospectively from a single-centre cohort using context-dependent endpoint definitions and require prospective external validation before clinical adoption.