3747 - Radiation Dose to the Circulating Blood is Associated with Poor Survival and during Treatment Immunosuppression in Patients with Locally Advanced Non-Small Cell Lung Cancer

Purpose/Objective(s): To assess radiation dose to circulating blood in patients with locally advanced non-small cell lung cancer (LA-NSCLC), and to examine its association with overall survival (OS) as well as changes in white blood cell counts (WBCs) during- and post-radiotherapy (RT).

Materials/Methods: A total of 155 LA-NSCLC patients, treated with concurrent chemo-RT, were included in this study. Patients received a median prescription dose of 63 Gy in 1.8-2 Gy fractions using 5-field intensity-modulated RT. The average beam-on time per field was 60s. An in-house artificial intelligence (AI) auto-segmentation algorithm was utilized to delineate 19 blood-bearing organs from the planning CT scans, including the aorta, adrenals, compact bones, coronary arteries, esophagus, gallbladder, heart, inferior-vena-cava, kidneys, lung, liver, pulmonary arteries, pulmonary veins, pancreas, small intestine, superior-vena-cava, spleen, stomach, and spongy bones. Dose-volume histograms of all 19 organs were generated, which were input into the Hematological Dose (HEDOS) framework to estimate dose to circulating blood cells. HEDOS simulates the whole-body blood flow and estimates the whole-body and per organ blood dose. Applying bootstrapping, four blood DVH metrics (mean, standard deviation, 10th, and 90th percentile) were tested for association with OS using Cox regression. Additionally, their correlation with pre-RT normalized WBCs (lymphocytes, neutrophils, and the neutrophil-to-lymphocyte ratio) during-cCRT, 2 and 4 months post-cCRT was assessed using linear regression. The dataset was divided into a training and a set-aside validation set, with a 70% and 30% split. A candidate DVH metric in training (defined by p<0.05) was tested for generalizability in the validation set.

Results: The estimated population mean ± standard deviation dose per fraction to circulating blood was 0.4±0.1 Gy. The corresponding 10th and 90th percentiles of the blood dose were: 0.27±0.08 Gy and 0.5±0.1 Gy. Among the 19 organs, the normal lungs, heart, aorta, and superior-vena-cava were found to be the major contributing organs to blood dose, with a median contribution of 29%, 19%, 15% and 12% respectively. All four DVH metrics were significantly associated with OS (regression coefficient ß: 2.5-18; p-value: 0.003-0.01). The during-cCRT neutrophil-to-lymphocyte ratio was weakly but significantly correlated with all four blood dose metrics (coefficient of determination R²: 0.05-0.08; p-value: 0.001-0.006).

Conclusion: This study demonstrated that radiation dose to circulating blood can be estimated using AI-segmented blood-bearing organs. Higher blood dose was associated with reduced overall survival and was weakly correlated with the during treatment neutrophil-to-lymphocyte ratio. Our findings contributed to understanding the impact of radiation on immunosuppression and may offer valuable insights for developing immune-sparing radiotherapy strategies.