Integral multi-scenario robustness evaluation of anatomical robust optimization in head and neck

Purpose/Objective
Classical robust optimization considers uncertainties in patient setup and particle range. Usually plan robustness is evaluated from calculation of perturbed dose distributions based on the planning CT, without considering potential anatomical changes that may occur during the treatment course. Our aim was to compare the overall plan robustness of classical robust optimization (cRO) with the recently proposed anatomical robust optimization (aRO) based on an integral multi-scenario evaluation, considering all types of uncertainties including anatomical variations.

Material/Methods
Datasets for 20 head and neck cancer patients, including a planning CT and weekly control CTs, were analyzed. Two intensity-modulated proton therapy (IMPT) plans were calculated: cRO, using solely the planning CT, and aRO, including additionally the first two control CTs in the plan optimization. For the robustness analysis, perturbed dose distributions with random setup uncertainties and fixed range uncertainty values of -3.5%, 0% and +3.5% were generated, drawing for each fraction n a random number from a Gaussian distribution around 0 mm with a standard deviation of 2.5 mm for the isocenter shift in each cardinal direction (xn, yn, zn). Moreover, in each fraction n the correspondent weekly control CT was used to consider the anatomical changes during therapy. 33 single-fraction perturbed doses were calculated and summed to generate a perturbed whole-treatment dose distribution. The procedure was repeated 10 times for each of the three range uncertainty values, resulting in 30 perturbed dose distributions per plan (Figure 1).

Results
Both nominal plans fulfilled the clinical objective for target coverage (D98% ≥ 95% of the prescribed dose). The median values calculated from the 30 perturbed dose distributions for each patient showed a reduction in the target coverage for the cRO plan, with mean (minimum) values of 94.9% (88.1%) and 95.4% (89.3%) for the low- and high-risk CTV, respectively, in comparison with 96.6% (92.0%) and 96.8% (93.6%), respectively, for aRO (Figure 2a). The variation in CTV coverage between the 30 scenarios, i.e. the width of the perturbed dose distributions, was found to be larger for cRO plans, with median (maximum) values of 1.9 (8.3) and 1.4 (5.6) for low- and high-risk CTV, respectively, in comparison with 1.4 (3.4) and 0.9 (5.2) for aRO plans, respectively. Moreover, the cRO case showed reduced robustness in comparison with aRO for some patients, where certain scenarios violate the clinical objective, as shown in Figure 2b.

Conclusion
Anatomical robust optimization showed superior plan robustness in comparison with the classical approach in a comprehensive multi-scenario evaluation. Anatomical variations play an important role in the overall plan robustness together with setup and range uncertainties, therefore their effect should not be underestimated or neglected.