Should Daily Online Adaptive Replanning be Utilized for Every Treatment in MRI-Guided Radiotherapy for Pancreatic Cancer?

Abstract

<h3>Purpose/Objective(s)</h3> Linear accelerators with onboard MRI caused a paradigm shift in radiotherapy by allowing for daily online MRI guided adaptive radiotherapy (MRgART). Customized plans can be generated and delivered using the acquired daily images while the patient is on the treatment table. However, online adaptive radiotherapy re-planning (OLAR) is labor intensive and time consuming. In this study, we investigate whether adapting the radiotherapy plan daily, accounting for interfractional organ motion, is necessary for every fraction during hypofractionated RT for pancreatic cancer. <h3>Materials/Methods</h3> Data from 70 fractions collected for pancreatic cancer patients treated with 30-35 Gy in 5 fractions using OLAR on a 1.5T MR-Linac were analyzed. For each daily MRI scan, a reposition (non-adaptive) plan was generated by re-calculating the initial undelivered reference plan based on the daily anatomy via rigid registration of the planning and daily images. The daily images were re-contoured during the OLAR process. Data for the non-adaptive plans was analyzed using the daily contours. The daily OLAR and non-adaptive plans were then compared to determine whether the non-adaptive plans would have been clinically acceptable and equivalent to the OLAR plan quality. Plan quality was defined by first prioritizing the dose constraints of the organs at risk (OAR) and second by the planning target volume (PTV) coverage. Statistical significance was determined by using a one-tailed t-test (<i>p</i> < 0.05). <h3>Results</h3> Dose constraints for the kidneys, spinal cord, liver, and skin were met for all OLAR and non-adaptive plans. For the OLAR plans, only 5.7% and 8.6% did not meet duodenum and small bowel dose constrains, respectively. For the non-adaptive plans, 45.7%, 5.7%, 28.6%, and 11.4% did not meet duodenum, small bowel, stomach, and colon dose constraints, respectively. The dose difference (17.5-1050 cGy) beyond what's clinically acceptable for the duodenum was the most statistically significant (<i>p</i> < 0.05). In addition, 32 OLAR plans and 12 non-adaptive plans met all planning objectives. From the 12 criteria-met non-adaptive plans, 4 had a significant (<i>p</i> < 0.05) increase in PTV coverage over the OLAR plans. Among all the 70 fractions, the non-adaptive plans were equivalent to the OLAR plans in 25 fractions (36%), i.e., OLAR was necessary in 45 treatment fractions (64%) (<i>p</i> < 0.05). <h3>Conclusion</h3> OLAR can significantly reduce GI OAR dose leading to an improvement in delivered plan quality in MRgART for pancreatic cancer. This study shows that the plan quality differences between the OLAR and non-adaptive plans justify the use of daily online adaptive re-planning for the majority (64%) of the treatments. To increase OLAR efficiency, a focus in online contouring and contour evaluation primarily on GI OAR structures may be warranted. With more automation for OLAR process being developed, the argument for not using OLAR for every fraction may become obsolete.