Noninvasive Radiation Markers Based on Lung Cancer CT Images and Dosimetry Features to Predict the Immune Response to Radiotherapy in Tumor Microenvironment

Abstract

For lung cancer patients with complex conditions, doctors need to determine treatment plan in combination with changes of immune response during radiotherapy. It was urgent to find a strategy to predict immune response to intervene the treatment at an early stage. The radiotherapy process and tumor microenvironment status of 105 lung cancer patients were tracked and recorded in our hospital from 2019 to 2022, public database TCGA-LUAD/LUSC queue of TCIA image data was downloaded as external verification set. Data was analyzed according to the following methods: First, the 3D slicer software was used to process the CT imageomics data, and the doseomics information was calculated by DVHmetrics package from R software. Then Python was brought in to develop a non-invasive radiation scoring system through machine learning methods (T-test, LASSO, random forest and Xgboost). Finally, combined with the clinical information related to the patient's immune response, a nomogram was made to study the correlation between the score markers and the ratio of CD4 cells/CD8 cells in the tumor microenvironment, the prognosis of lung cancer, as well as the incidence of radiation pneumonia. The radiation markers score formula was obtained with CT imageomics and DVH dosimics features, which achieved an AUC of 0.67-0.75 in predicting CD4/CD8 condition in tumor microenvironment after radiotherapy. It was worth noticing that in predicting the prognosis of lung cancer, it was difficult to distinguish between the radiation score and the patient's survival status (P>0.05). However, we found that the radiation pneumonitis of lung cancer patients in the low score group (42.6%) was significantly higher than that in the high score group (13.4%). Radiation markers score was a non-invasive method to evaluate the immune response of patients with lung cancer, which can be used to predict the incidence of radiation pneumonitis.