Probabilistic treatment planning and margin alternatives in cranial radiotherapy
- Programme
- HSST
- Specialty
- Radiotherapy Physics
- Project published
- 29/05/2025
The use of planning target volume (PTV) and planning risk volume (PRV) margins is a long-established and accepted method to account for geometric uncertainty in radiotherapy. Alternatives to margin-based treatment planning, which explicitly account for geometric uncertainties during plan optimisation, include robust optimisation and probabilistic planning. These alternatives have shown promising advantages over traditional margin-based approaches in terms of achieving equivalent confidence of target dose coverage while reducing normal tissue doses, summarised in a review by Unkelbach et al. (2018). However, there are very few clinical studies of these alternatives for cranial radiotherapy, and none considering alternative approaches to organ at risk PRV margins.
This work explores these alternative methods for accounting for geometric uncertainty in plan optimisation for cranial radiotherapy. Geometric uncertainties for cranial radiotherapy were audited and quantified. For a cohort of 13 glioblastoma patients with overlap between PTV and one or more PRVs, probabilistic treatment plans were then produced using an existing probabilistic optimisation code, and compared with standard clinical margin-based plans, as well as plans produced using a commercial robust optimisation package. The performance of each plan under uncertainty was assessed using a probabilistic evaluator, written for this work, which performed Monte Carlo simulations of multiple treatment courses. Both robust and probabilistic plans achieved equivalent clinical target volume (CTV) coverage to margin plans under uncertainty, but with significantly reduced irradiated volumes and mean brain dose. However, probabilistic plans did not always achieve OAR sparing at the required 90% confidence level.
An alternative method for incorporating probabilistic information into treatment planning was proposed, based on spatial probability maps of patient contours derived from simulations of geometric uncertainty performed prior to plan generation. Plans were produced using this alternative (“PreSim”) method and compared to the probabilistic, robust and margin plans. The results show that PreSim method shares similar advantages with robust optimisation and probabilistic planning (achieving equivalent CTV coverage with reduced irradiated volumes) while allowing the use of a conventional optimiser.
The PreSim method was then applied to creating simultaneous integrated boost plans for to a probabilistically defined boost target volume for glioblastoma patients. The boost target volume was chosen to be the GTV within the CTV 95% probability region, avoiding the 0.1% OAR isoprobability volume. Inclusion of a 72 Gy boost to this volume (maintaining 60 Gy in 30 fractions to the remainder of the CTV) was feasible without increasing OAR doses, and also further reduced irradiated volumes and mean brain dose.
Finally, the PreSim method was applied to the evaluation of margins for single isocentre multiple target (SIMT) stereotactic radiosurgery (SRS) treatments. SIMT is a modern technique to deliver SRS using a linear accelerator whereby multiple metastases are irradiated simultaneously. Currently there is no consensus on the optimal PTV and PRV margins to apply, and there is the additional complication of potential impact of rotational errors at off-axis treatment locations which may require differential or larger margins. Geometric uncertainties for SIMT SRS were quantified and simulations of repeat treatments were then performed for 10 multiple metastases patients. The results of this work validated the use of 1 mm PTV and PRV margins in the author’s centre, and shown that a larger PTV margin is not required for GTVs at a larger distance from isocentre for the local equipment and SIMT technique.
Outputs
Thesis
Published abstracts
https://www.sciencedirect.com/science/article/abs/pii/S0167814024022175
https://www.sciencedirect.com/science/article/abs/pii/S016781402500828X