Click here to

Session: Imaging for Treatment Assessment and Outcome Modeling [Return to Session]

A Comprehensive Model to Predict Radiation Related Immunosuppression Following Lung SBRT

C Nguyen1*, G Minesinger1, D Lain2, T Showalter2, P Read2, J Larner2, K Wijesooriya1,2, (1)Dept. of Physics, University of Virginia, Charlottesville, VA, (2) Dept. Radiation Oncology, University Of Virginia, Charlottesville, VA


SU-F-206-1 (Sunday, 7/10/2022) 2:00 PM - 3:00 PM [Eastern Time (GMT-4)]

Room 206

Purpose: Lymphocytes are a highly radiosensitive tissue, and post-treatment lymphopenia and immunosuppression constitute two significant side effects of contemporary treatment techniques in radiation oncology. In the era of immunotherapy, it is highly desirable to have a predictive model that predicts the time-dependent immune cell kill from RT.

Methods: This algorithm uses radiation therapy treatment plans, patient CT data sets, dose maps, and organ delineations combined with a blood flow and a lymphatic flow model for lymphocyte recirculation to obtain the absorbed dose for circulating lymphocytes. Primary, secondary lymphoid organs such as bone marrow, lymph nodes, spleen as well as the non-lymphoid organs such as lungs, liver, other organs are included. These absorbed doses are used to predict the fraction of lymphocytes killed by a given treatment plan, using a dose-dependent survival curve. We have developed a comprehensive predictive algorithm for radiation-related immunosuppression using a prospective clinical trial dataset of 16 NSCLC lung cancer patients treated with five fraction SBRT with immune cell levels measured at 3 post-treatment time points: end of treatment, four weeks following treatment, and six months following treatment. An independent retrospective lung SBRT data set of 10 patients was used to validate the model. Both central (11) and peripheral (15) lung tumors were included with a PTV volume ranging from 4.8 cc – 403.1 cc.

Results: The absolute lymphocyte count differences between model prediction and measurement at all time points for the training dataset and for the testing dataset were Mean (STD): 0.22 (0.21) x10^9 cells/L and 0.19 (0.13) x10^9 cells/L respectively.

Conclusion: Our predictive model has an accuracy that will enable treatment plan design and optimization to reduce treatment-related immune suppression.


Not Applicable / None Entered.


Not Applicable / None Entered.

Contact Email