Intrafractional Changes May Necessitate the Development of Real-time Adaptive Radiation Therapy for Abdominal Tumors

Abstract

MR-Linac enables daily optimization of radiotherapy (RT) via online adaptation to patient daily anatomy. However, intrafractional anatomical changes may occur particularly for targets adjacent to or included in the gastrointestinal (GI) tract. These changes include motions due to respiration, peristalsis and unpredictable air filling of GI organs. In this work, we study the intrafractional variability of abdominal organs and its dosimetric impact for online MR guided adaptive RT (MRgART) and discuss the need for developing real-time ART. Thirteen time-sequenced abdominal MRI dataset pairs (MR1 and MR2), separated by 15-60 minutes (similar to the intrafractional timescale in MRgART), for 9 abdominal cancer patients were analyzed. Eight dataset pairs consisting of T2 and T1 images were collected for 8 patients from a 3T MR simulator, while the remaining 5 pairs were motion averaged MR from 4DMRI acquired on MR-Linac immediately before and after each fraction of a patient five-fraction RT course. For each MR pair, we first created an IMRT plan based on MR1 (mimicking online adaptation plan before the delivery) by optimizing dose to the target (the pancreatic head) and organs-at-risk (OAR) in the presence of a 1.5T magnetic field. We then reconstructed the IMRT dose based on MR2 (i.e., post-treatment plan). The differences of dose volume histogram (DVH) parameters between the two plans were analyzed. The GI air cavities change substantially during fractional planning and delivery (15-60 min duration). Air cavity volumes of the stomach and duodenum were found to vary, on average, by 58.6 and 3.2 cc, respectively. The maximum, minimum and mean differences in selected DVH parameters resulted primarily from the intrafractional air cavity changes and are tabulated below. Here, significant deviations from the planned fractional doses are seen. The change of stomach air volume has a strong correlation to the maximum target dose (|r| = 0.8). Most notably, strong correlations between duodenum air volume and the target maximum, minimum and mean doses (|r| = 0.9, 0.7, and 0.9, respectively) are observed for the longitudinal datasets. Furthermore, maximum dose to OARs is seen to change by up to 50% of the planned limits. GI structures, primarily air cavities, can vary rapidly, causing severe intrafractional anatomical changes during MRgART. Such changes can impact the dose delivery of online ART. These results demonstrate the need to develop real-time adaptation to robustly address the intrafraction variation for abdominal tumors and justify further studies with larger datasets.Abstract 1111; TableDifference between pre- and post-treatment plans: max-min(mean)Max Target Dose (% of Rx)4.8-0.1(1.8)Min Target Dose (% of Rx)51.6-1.0(17.0)Mean Target Dose (% of Rx)6.8-0.2(2.1)Target V (97%) (% of Volume)14.2-0.1(4.3)Max Duodenum Dose (% of Rx)26.7-1.2(6.7)Mean Duodenum Dose (% of Rx)37.6-0.3(8.6) Open table in a new tab