Integrating Multiple Rectal States In Robust Optimization Improves Robustness of Intensity-Modulated Proton Therapy for Cervical Cancer
Abstract
Purpose
This study evaluated the impact of interfractional rectal state changes on plan robustness in cervical cancer intensity-modulated proton therapy (IMPT) and established a robust optimization strategy incorporating multiple rectal conditions.
Methods
Thirty patients were included. For 20 patients, CT scans in filled and simulated air states were used (Group A). For 10 patients, an additional natural empty state CT was available (Group B). Nominal plans were designed on the filled-state CT. These were compared to plans using 4D robust optimization that integrated all rectal states per patient (Fig. 1). Plan robustness was measured by the variation (range) of dose metrics across states.
Results
The benefit of robust optimization in mitigating dose variations across rectal states is summarized in Fig. 2. For Group A (Fig. 2a-b), which considered filled and air states, robust optimization significantly decreased the median variation in CTV D95% and D98% across all three beam strategies compared to nominal planning (p < 0.05). For Group B (Fig. 2e-f), which included an additional natural empty state, the reduction in dose fluctuation was even more pronounced. For example, in Strategy 1 (Fig. 2d), the median variation of CTV D95% decreased from 4.20 Gy to 0.40 Gy (reduction ~90.5%, p < 0.05), and CTV D98% decreased from 12.22 Gy to 1.02 Gy (reduction ~91.6%, p < 0.05). Across both groups and all strategiesthe, robust optimization ensured that nearly all patients met the target dose objective (D95% ≥ 45 Gy) under every rectal state(GroupA: Fig. 2c and GroupB: Fig. 2f), without a significant increase in OAR doses.
Conclusion
Inter-fractional rectal state changes considerably affect the dosimetric robustness of IMPT plans for cervical cancer. Robust optimization integrating multiple rectal states effectively mitigates this uncertainty, ensuring reliable target coverage across anatomical variations.