Dosimetric differences between online adapt-to-position and offline adapt-to-shape plans for adaptive radiotherapy in cervical cancer

Radiation therapy plays a significant role in the treatment of cervical cancer. Additionally, more adaptive workflows using ATP are being implemented in the daily radiotherapy of our organization. Herein, we aimed to investigate the dosimetric differences between online ATP and offline ATS plans for magnetic resonance (MR)-guided adaptive radiotherapy in patients with cervical cancer and determine radiotherapy modalities that address clinical requirements. In total, 25 patients with cervical cancer were enrolled in this study, with 13 in the radical radiotherapy group and 12 in post-operative radiotherapy group. We aimed for the clinical target volume (CTV) to be covered by 95 – 100% of the prescribed dose (50 Gy/25 sessions/5 weeks). MR-Linac was performed daily during treatment, and the images were rigidly aligned with the local computed tomography to generate an online ATP plan. MR images acquired during the first three sessions were selected to recontour the CTV and organs at risk (OAR). Furthermore, an offline ATS plan was generated. In the radical radiotherapy group, the CTV, D98, D95 (5024.65 ± 23.34 vs. 4995.50 ± 14.99 cGy), and Dmean of the ATS were superior compared with those of the ATP. The Dmax was lower in the ATS plan than in the ATP plan. In the post-operative radiotherapy group, the CTV, Dmean, D98, and D95 (5052.61 ± 67.87 vs. 5014.41 ± 24.68 cGy) were better in the ATS plan than in the ATP plan. When evaluating the OAR in the radical radiotherapy group, the minimum doses to the bladder and rectum were greater in the ATS plan than in the ATP plan. In the post-operative radiotherapy group, the V20 of the bladder and rectum were lower in the ATS plan than in the ATP plan. Therefore, ATS is well suited for post-operative radiotherapy, whereas ATP is better suited for radical radiotherapy. Furthermore, ATP can effectively address the clinical requirements of daily workflows.