KEY POINTS
- The analysis included 69 prostate cancer patients and 1,378 CBCT-based online adaptive fractions delivered on Ethos. Patients received 20 fractions with cumulative doses of 48 Gy, 57.6 Gy and 60 Gy using simultaneous integrated boosts.
- For every fraction, positional change was measured between the initial adaptation CBCT and a second verification CBCT acquired immediately before treatment delivery.
- Mean time between the two scans was 17.5 ± 5.1 minutes. Absolute shifts ranged from 0.2 to 17.5 mm, and 4.6% of fractions exceeded 6 mm, demonstrating that clinically meaningful movement can occur during plan adaptation.
- Motion was strongly patient-dependent. Some patients remained highly stable across treatment, whereas others showed both consistent directional drift and substantially greater fraction-to-fraction variability.
- Derived motion-related margins were asymmetric: 2.3 mm right / 2.2 mm left, 2.9 mm anterior / 5.7 mm posterior, and 2.2 mm superior / 3.2 mm inferior. The posterior direction showed the largest estimated requirement.
- Longer adaptation time was associated with greater movement in the anterior-posterior and superior-inferior directions and in overall vector magnitude. The correlations were modest but statistically significant.
- These margins should not be adopted directly as treatment margins. The CBCT-to-CBCT interval was roughly four times longer than beam-on time, and the measured shifts also include registration uncertainty, making them conservative upper bounds for motion-related uncertainty.
CLINICAL TAKEAWAY
Online adaptation corrects the anatomy present at the first CBCT, but the prostate can move again while the new plan is being created. A second verification CBCT remains an important safety step, particularly as centres reduce margins; the marked patient-to-patient variability also supports future individualized motion strategies.