Evaluation of Differences Between Estimated Delivered Dose and Planned Dose in Nasopharynx Patients Using Deformable Image Registration and Dose Accumulation

To investigate differences in planned dose and estimated delivered dose using deformable image registration (DIR) and dose accumulation for clinical target volumes (CTV) and organs at risk (OAR). Fifty-two nasopharynx patients treated with chemoradiotherapy [70 Gy (high-dose) and 56Gy (low-dose) in 35 fractions] from 2013 and 2015 were included. All clinical plans used 6 MV intensity modulated radiation therapy (IMRT) and daily 3D kV cone beam CT (CBCT) image guidance. Retrospective dose accumulation was performed in RayStation v4.5.2 as follows: DICOM RT data (CT image, plan, dose, and contours) were imported from the clinical planning system. Daily CBCTs were also imported and deformably registered to the planning CT. Daily delivered dose was calculated on the CBCTs and deformed back to and accumulated onto the planning CT to estimate the delivered dose. The planned and delivered doses were compared using representative dose metrics for primary CTVs and OARs including brainstem, mandible, spinal cord, and right and left parotid glands. The impact of daily set-up variation and anatomical changes on target coverage and OAR doses was evaluated by comparing planned dose vs. delivered dose. Relative differences in CTV D98 calculated as percentage of planned dose was used to assess target coverage. The mean relative difference in D98 for 70 Gy CTVs was 0.8% (median 1.1%, range -1.8 to2.6%). The mean relative difference in D98 for 56 Gy CTVs was negligible (median 0.1%, range -2.8 to2.2%). The median (range) maximum dose to 0.1 cc of the brainstem was similar between planned and delivered: 53.0 Gy (40.7-61.5 Gy) and 52.2 Gy (40.0-61.7 Gy). Although some cases exceeded the nominal tolerance dose of 54 Gy in the plan, other cases planned under this tolerance did not exceed 54 Gy in estimated delivered dose. The median maximum dose to 0.1cc of the spinal cord and mandible was 37.8 Gy and 72.0 Gy and delivered dose (range) was 38.8 Gy (30.4-45.8 Gy) and 72.2 Gy (62.9-75.4 Gy), respectively. The D50 planned median dose, delivered median dose (range) for the right and left parotid was 34.4 Gy, 39.3 Gy (23.5-72.3 Gy), and 37.3 Gy, 40.6 Gy (18.6-68.1 Gy). The median planned, delivered dose (range) for the right and left parotid mean dose was 38.4 Gy, 40.3 Gy (30.1-69.8 Gy) and 40.5 Gy, 41.7 Gy (25.3-61.9 Gy). No large differences were seen between planned and delivered OAR doses for the evaluated metrics. Outliers were identified (±2 SD from the average) for the parotids, indicating that variation can be expected for this OAR and warrants further investigation. There was minimal dosimetric change on the serial OARs; the right and left parotids received greater dose than planned. Further studies are required to investigate the impact of this dosimetric change and to identify patients who can benefit from adaptive radiotherapy.