Forecasting Time-Dependent Radiation Induced Lymphocyte Kill In Locally Advanced Non-Small Cell Lung Cancer (NSCLC)
Abstract
Purpose
Radiation-Induced-Lymphocyte-Kill (RILK) varies substantially among patients and the extent to which different factors contribute to RILK remains incompletely understood. We present a model to predict time-dependent RILK for advanced stage Non-Small Cell Lung Cancer (NSCLC) patients treated with chemo-radiation therapy (CRT) with standard fractionation by simulating the lymphocyte circulation dynamics in blood and lymph-rich organs. This is an extension to the model CYTOPREDICT used for early stage lung cancer treated with SBRT on a clinical trial [1,2].
Methods
Data from 78 advanced-stage NSCLC patients treated with RT-only or concurrent CRT were collected, including dosimetric parameters (integral dose, V5 to V50) for immune-rich organs (total lungs-GTV, heart + great vessels (heart+GV), lymph node-GTV (LN-GTV, to exclude draining-LNs), thoracic spine). Treatment fractionations varied between 10 – 37 fractions, with 48 patients treated with chemo+RT, and 30 with RT alone. PTV volume: median (range): 358 (59 – 1624) cc. The Python-based model took radiation therapy DICOM data from patients as input and simulated the real-time coupled dynamics of lymphocyte circulation between blood and lymph-rich compartments to estimate RILK and predict longitudinal peripheral absolute lymphocyte count (ALC) trajectories up to six months post-treatment. The model was optimized using the Metropolis algorithm and a 10-fold cross validation optimization was performed.
Results
In the patient cohort with 10-fold cross validation, the mean (SD) absolute ALC difference between prediction and measurement was mean (SD): 0.177 (0.178) x10^9 cell/L with 95% of predictions within 0.5×10^9 cell/L of observed values and 67% within 0.2x10^9 cell/L.
Conclusion
The model successfully forecasted time-dependent ALC trajectories for advanced-stage NSCLC patients undergoing chemo-RT up to six months post-treatment. This approach enables estimation of individualized RILK from treatment plans, supporting treatment personalization to mitigate immune toxicity while preserving standard-of-care.