Evaluation of Radiotherapy Plan Robustness: A Systematic Literature Review.

Abstract

Aims: Advances in radiotherapy have led to increasingly conformal and complex treatment plans. The progressive reduction in safety margins around the target volume and the increased use of hypofractionated radiotherapy further heighten their vulnerability to systematic geometric uncertainties, which may compromise target volume coverage and increase doses to normal tissues. Evaluating treatment plan robustness, therefore, is crucial to ensuring the safe and effective delivery of radiotherapy. This systematic literature review provides a comprehensive overview of current practices for assessing treatment plan robustness across radiotherapy modalities.

Materials and methods: A Pubmed search was conducted for studies published up to July 2025 that evaluated plan robustness.

Results: Of 287 publications, 225 met the inclusion criteria. Most studies (173 of 225) focused on proton therapy, with setup (198 studies) and range (184 studies) being the most commonly considered uncertainties. Robustness evaluation methods varied widely and were categorised as dose-volume histogram (DVH)-based, voxel-based and radiobiological metrics. The most commonly used method for evaluating plan robustness involved visualising DVHs by overlapping those from multiple uncertainty scenarios to represent all possible variations. Frequently used dosimetric parameters for clinical target volume (CTV) coverage included variations of CTV D95%, D98% and V95% and the proportion of scenarios in which CTV D98%>95%. Voxel-based metrics, such as Max-Min dose distributions and voxel-wise dose reconstructions, provided spatial information on areas susceptible to uncertainties. Radiobiological metrics assessed robustness through changes in tumour control and normal tissue complication probabilities, highlighting the clinical impact of dose variations arising from uncertainty scenarios.

Conclusion: Currently, there is no international consensus on evaluating plan robustness. We recommend combining DVH-based metrics with spatially informative voxel-based approaches. Establishing a standardised framework for robustness evaluation, along with integrating commercial robust evaluation software tools that enable the generation of these metrics, will be essential for its adoption in clinical practice.