Purpose: Currently for abdominal treatments, estimates of bowel toxicity are based on pre-treatment dose-volume histogram data. However, the actual dose the bowel receives can depend on intrafraction anatomy changes, such as weight loss and changes in bowel filling. We propose a method to model bowel toxicities, incorporating on-treatment patient information using in vivo transit electronic portal imaging device (EPID) images.
Methods: For 63 patients treated to the lower thorax, abdomen, or pelvis on the Varian Halcyon, chart review was performed to measure incidences of toxicity. Twenty patients presented with acute gastrointestinal (GI) toxicity. For each treatment plan, the absolute volume dose-volume histogram of the bowel was exported and analyzed. The Halcyon collects in vivo EPID images for all treatments, which were downloaded and analyzed with an automated framework. Over the course of treatment, EPID measurements quantifying changes in radiation transmission in vivo were summed after every fraction. Logistic models were used to test correlations between acute GI toxicity and bowel dosimetric parameters as well as in vivo EPID measurements.
Results: The incidence of toxicity versus the volume of 40 Gy fit with a logistic function was superior to an average model (p < 0.0001) and agrees with previously published models. The incidence of toxicity versus the sum of in vivo transmission measurements showed marginal significance after 15 fractions (p=0.10) and a significance of p < 0.05 is seen after 20 fractions when compared to an average model.
Conclusion: This study agrees with other models that reducing bowel dose results in lower acute GI toxicity. Additionally, the study supports the hypothesis that in vivo EPID measurements can be used to monitor and model acute GI toxicity. From this study, it is unknown if the change in the patient's anatomy is a source of toxicity or just correlated.
Funding Support, Disclosures, and Conflict of Interest: Presenting Author receives grant funding from Varian Medical Systems.
In Vivo Dosimetry, Modeling, NTCP
TH- Radiobiology(RBio)/Biology(Bio): Rbio - Outcome models combining dose, imaging, radiomics/radiogenomics and clinical factors: other than machine learning