ESTRO 2024 - Abstract Book

S4006

Physics - Inter-fraction motion management and offline adaptive radiotherapy

ESTRO 2024

Results:

We assessed fifteen PotD approaches from eleven papers (Table 1) and present results alongside eight population margins and seven internal target volumes (ITV) for context (Figure 1). Interfraction motion was substantial; the fixed isotropic margin required to provide 100% coverage of all fractions was 44 mm, with mean PTV size 3316 cc. Mean PTV size with no expansion on the CT-FB CTV LR was 1094 cc. PotD approaches fell into three categories; those that added varying margins to the planning CTV LR shapes, those that modelled the CTV LR against bladder volume, and those that applied incremental margins onto the CTV LR (margin-of the-day). Generally, coverage improved as PTV size increased (Figure 1), but certain strategies punched above their weight. A three-plan library consisting of CTV LR on CT-FB, CT-EB, MRI, each with 10 mm margin (labelled ‘POTD 8’ on Figure 1), provided 96% mean coverage (median 100%, IQR 95-100%) with mean PTV size of 1346 cc. This was one of the simplest PotD methods to implement. Some strategies modelled the CTV LR on CT-EB and CT-FB against bladder volume, and extrapolated the model to cover larger and smaller bladders. This process caused errors in three patients with limited bladder volume variation at planning, creating unnecessarily large and anatomically unfeasible PTVs. Modelling against bladder volume alone is error-prone because interfraction motion is multifactorial. In contrast, the clinician-derived solution involved subjective PTV editing and extrapolation. This resulted in excellent coverage (mean 100%, median 100%, IQR 100-100%, range 96-100%) and small volumes (median 1210 cc) on the 11 patient subset.

The two margin-of-the-day approaches achieved 100% coverage, but with larger PTV sizes (mean 1698 cc and 1691 cc).

Arguably the best non-adaptive strategy was created by linear regression modelling of CT-FB and CT-EB CTV LR shapes against bladder volume, with 10 mm margin (labelled ‘ITV 5’ on Figure 1); this provided 98% mean coverage (median 100%, IQR 97-100%) and mean PTV size 1437 cc. The coverage was statistically superior to POTD 8 (p < 0.001) but mean PTV size was larger.

When deciding which strategy to use, centres should consider the clinical significance of these differences in coverage and PTV size, balanced against the availability of resources for adaptive radiotherapy.

Conclusion:

Published PotD studies are difficult to compare due to differing metrics and small patient numbers (median 14, range 9-23). We have assessed all replicable PotD strategies (15), any non-adaptive comparators assessed in those papers (seven population and six ITV), the EMBRACE-II population and ITV approaches, and the clinician-derived subjective PotD approach used at our centre.

A simple three-plan strategy provides excellent coverage and small PTV sizes compared to more complex strategies involving shape modelling or a greater number of plans. However, non-adaptive centers can also achieve excellent