Assessment of Gastrointestinal Luminal Air Effects on Dose Calculation Accuracy for Pancreatic Cancer Stereotactic MR-guided Adaptive Radiation Therapy

Purpose: Clinical outcomes of ablative stereotactic MR-guided adaptive radiation therapy (SMART) for inoperable pancreatic ductal adenocarcinoma (PDAC) have been encouraging. During the on-table adaptive workflow, manual segmentation of the air electron density within gastrointestinal (GI) luminal organs at risk (OARs) is routine at some institutions, but not at others due to lack of published data demonstrating the impact of attenuation and Lorentz force on dosimetric accuracy. We hypothesize that absence of electron density segmentation of GI luminal air would result in a violation of GI luminal OAR constraints.

Methodology: Fifty SMART fractions from inoperable PDAC patients who were treated at our institution were included. All were prescribed 50 Gy in 5 consecutive fractions and treated on a 0.35T MR-Linac in mid-inspiration breath hold. All delivered plan were done accounting for air electron density within GI OARs. Plans created without air segmentation were developed by taking the plans with air segmentation, deleted the air contour, recalculating dose, normalizing to the closest OAR, reinserting the air, and recalculating the dose. Plan comparison of the clinically delivered fraction with and without GI OAR air segmentation was assessed for dosimetric differences to gross, clinical, and planning tumor volumes (i.e., GTV, CTV, and PTV) and neighboring GI OARs (i.e., stomach, duodenum, small bowel, and large bowel) exceeding the prescribed dose constraints (0.5 cc to 35 Gy and 0.03 cc to 40 Gy for stomach, small bowel, and duodenum; 0.5cc to 38 Gy and 0.03cc to 43 Gy to the large bowel). A paired sample t-test was used to investigate plan differences between each clinically delivered fraction with and without air contour.

Results: Dose constraints for all neighboring GI OARs were met using the clinically delivered plan with air contour. Dose constraints for some plans without air contour were not met with a mean ± SD of 0.09±0.10cc for 11 stomach at V35, 0.44±0.72cc and 0.06±0.09cc for 10 and 4 small bowel at V35 and V40, 0.03±0.03cc for 2 large bowel at V38, 0.27±0.38cc and 0.10±0.08cc for 22 and 2 duodenum at V35 and V40 respectively. Plan quality comparison was evaluated for PTV, GTV, and CTV and neighboring GI OARs exceeding the prescribed dose constraint.

Conclusions: Although segmentation of air within GI OARs can be time consuming during an already lengthy on-table adaptive workflow, it appears to be important for dose calculation accuracy.

References: