PURPOSE: To investigate the impact of four-dimensional robust optimization (4DRO) on dose delivered to lung cancer patients in pencil beam scanning proton therapy. METHODS AND MATERIALS: 2 strategies were compared for 20 lung cancer patients, using a different number of breathing phases of the reconstructed 4D computed tomography (CT) included in the plan optimization problem. In the restricted approach combined with gating, only 3 phases close to reference end-exhale were considered instead of the whole breathing cycle. The prescribed dose was 60 Gy(RBE) in 10 fractions. Target coverage (D98%) and dose to healthy tissues were evaluated using Wilcoxon signed-rank test. To assess the robustness against interfractional anatomical and respiratory variations, the optimized plans were recalculated on re-evaluation 4DCTs. To compare the sensitivity of both strategies to interplay effects, we implemented an end-to-end test with a home-made heterogeneous moving phantom and ionization chambers measurements. Robustly optimized plans with prescription doses of 6 Gy(RBE) were delivered in different dynamic conditions. RESULTS: Both 4D robustly optimized plans reached the same target coverage (p = 0.56), while a statistically significant decrease of the homolateral lung dose was observed using the restricted approach (p < 0.0001). Plan recalculations within 15 days from the treatment simulation showed the same robustness of target D98% against interfractional variations (p = 0.48), with an average decrease of approximately 3 Gy(RBE). Phantom measurements confirmed the delivery accuracy of the restricted approach (mean dose deviations <5%). Higher deviations were found for ungated full 4DRO and larger motion amplitude. CONCLUSION: The restricted approach combined with gating improved normal tissue sparing and was shown to be more robust to single fraction deliveries and large motion amplitude.
PURPOSE: To investigate the impact of four-dimensional robust optimization (4DRO) on dose delivered to lung cancerpatients in pencil beam scanning proton therapy. METHODS AND MATERIALS: 2 strategies were compared for 20 lung cancerpatients, using a different number of breathing phases of the reconstructed 4D computed tomography (CT) included in the plan optimization problem. In the restricted approach combined with gating, only 3 phases close to reference end-exhale were considered instead of the whole breathing cycle. The prescribed dose was 60 Gy(RBE) in 10 fractions. Target coverage (D98%) and dose to healthy tissues were evaluated using Wilcoxon signed-rank test. To assess the robustness against interfractional anatomical and respiratory variations, the optimized plans were recalculated on re-evaluation 4DCTs. To compare the sensitivity of both strategies to interplay effects, we implemented an end-to-end test with a home-made heterogeneous moving phantom and ionization chambers measurements. Robustly optimized plans with prescription doses of 6 Gy(RBE) were delivered in different dynamic conditions. RESULTS: Both 4D robustly optimized plans reached the same target coverage (p = 0.56), while a statistically significant decrease of the homolateral lung dose was observed using the restricted approach (p < 0.0001). Plan recalculations within 15 days from the treatment simulation showed the same robustness of target D98% against interfractional variations (p = 0.48), with an average decrease of approximately 3 Gy(RBE). Phantom measurements confirmed the delivery accuracy of the restricted approach (mean dose deviations <5%). Higher deviations were found for ungated full 4DRO and larger motion amplitude. CONCLUSION: The restricted approach combined with gating improved normal tissue sparing and was shown to be more robust to single fraction deliveries and large motion amplitude.