Author: Gabriel Lucas Andrade de Sousa, Einsley-Marie Janowski, Cam Nguyen, Krishni Wijesooriya 👨🔬
Affiliation: Department of Radiation Oncology, University of Virginia, Department of Physics, University of Virginia 🌍
Purpose: Optimizing radiation therapy (RT) to spare the immune system may improve Overall Survival (OS) in cancer patients. This study develops a computational algorithm to identify optimal dose-volume objectives for immune-rich organs that maximize OS. A pancreatic cancer dataset treated with chemo-RT was used for validation.
Methods: For 69 standard fractionation pancreatic cancer patients, DICOM-RT files of CT images, structure contours, dose distributions, and RT plans were collected. Absolute Lymphocyte Counts (ALC) pre-RT, immediately post-RT, and the lowest within six months were used to calculate the fractional ALC drop (estimated RIIS). Spearman correlations identified Organs At Risk (OARs) whose dose-volume metrics (e.g., V10, mean dose) were most associated with RIIS. Kaplan-Meier survival analysis, stratified by medians of each organ-dose pair, identified the largest contributor to OS. Subsequent iterations added dose-volume objectives for immune-rich organs to improve OS, using a maximum threshold p-value of 0.01. The top eight simultaneous objectives that maximize 5-year OS were identified. A manual verification was performed for the automated algorithm selected parameters.
Results: Negative correlations between ALC ratios and dose metrics were strongest for the bowel bag, duodenum, kidneys, liver, spleen, and stomach. Twenty-two dose-volume metrics per organ were analyzed. An optimal set of three constraints improved 5-year survival up to 49% (p=0.0009) and included liver maximum dose, spleen integral dose, and duodenum V40.
Conclusion: Treatment strategies minimizing RIIS and improving OS are critical. Our algorithm guides planning to spare the immune system and enhance patient outcomes. While validated on a pancreatic cancer patient cohort, our model is site-independent and can be utilized in other tumor sites.