Robustness assessment of radiotherapy treatment plans in Switzerland.

Purpose: Robustness assessment is an essential part of radiotherapy plan quality assessment. However, it is often not evaluated in photon-based radiotherapy. This study aims to conduct a robustness audit to establish a baseline for the role of plan robustness in Switzerland by assessing and comparing robustness across plans from and clinical workflows in multiple institutions.

Materials and methods: A multi-institutional study involving 11 Swiss institutions was conducted. Each institution provided treatment plans for three cases and completed a questionnaire on treatment planning and assessment of robustness in their clinical practice. The plans were planned using the Eclipse treatment planning system and utilized intensity-modulated techniques using a 6 MV flattened photon beam for one brain case, and one unilateral and one bilateral head and neck cases, prescribed 60.0 Gy (one phase), 70.0 Gy (two phases) and 70.0 Gy (three phases) to 95% of the target volume, respectively. Institutions used their standard institutional protocols for the provided CT, structures and prescription. Dose distributions were subsequently recalculated in an in-house Monte Carlo (MC) framework incorporating clinically motivated uncertainties associated to patient setup and multi-leaf collimator (MLC) positions. The uncertainties' impact on the dosimetric plan quality was assessed by evaluating representative target and organ-at-risk (OAR) dose-volume endpoints (e.g. D98% and D2% of the target, mean dose of parallel OARs and near max dose of serial OARs).

Results: Differences in target and OAR dose-volume endpoints in the presence of random patient setup uncertainties (Gaussian distributed with σ = 0.2 cm in the three translational and σ = 0.5° in the three rotational axes) were smaller than ±0.5 Gy. Exceptions were the near max dose-volume endpoints of structures near the target with differences up to ±2.2 Gy for the optic nerve in the brain case. Systematic rotational patient setup uncertainties of ≤3° in either pitch, yaw or roll had similar impact as translational uncertainties ≤0.3 cm in either left-right, superior inferior or anterior-posterior direction with maximal differences in most investigated dose-volume endpoints of 9.0 Gy. Systematic MLC uncertainties of +0.5 mm of all leaves led to an average increase of up to 3.0 Gy in the dose-volume endpoints. The questionnaire revealed diverse practices in terms of planning and assessment for robustness: all institutions use target and OAR margins, 2/11 use robust optimization and 5/11 regularly perform robustness assessments of treatment plans by recalculating the dose distribution including uncertainties. The importance of robustness in treatment planning was rated ≥8 out of 10 (10 as most important) by 6/11 institutions. The need for better commercial tools to assess or integrate robustness into treatment planning was expressed by 9/11 institutions.

Conclusion: This study presents the first multi-institutional inter-comparison of treatment plan robustness in Switzerland, establishing a robustness baseline for intensity-modulated plans. Despite diverse practices to assess plan robustness and to mitigate the impact of uncertainties on dosimetric plan quality, the robustness to the investigated uncertainties was similar across the plans and cases from all institutes. To foster standardization, we recommend to regularly conduct audits focusing on plan robustness to monitor and reduce inter-institutional variability in handling and assessing plan robustness.