The Study Of CD8+T Lymphocyte Infiltration In Stage T1 Non-Small Cell Lung Cancer Based On Radiomics

Objective:To investigate the predictive value of radiomics in CD8+ T lymphocyte infiltration in T1 stage Non-small cell lung cancer.Materials and methods:In this study,we retrospectively analyzed the CT non enhanced scan images and pathological data of 113 patients with stage T1Non-small cell lung cancer(NSCLC)confirmed by surgical pathology in our hospital.The infiltration of CD8+T lymphocytes in T1 stage Non-small cell lung cancer was quantitatively analyzed by immunohistochemical staining.According to the infiltration degree of CD8+T lymphocytes,they were divided into high infiltration group and low infiltration group.All patients were randomly divided into training set and validation set according to the ratio of 8:2.Univariate analysis was used to screen the clinical features(including age,sex,smoking,lymphatic metastasis,Carcinoembryonic antigen,and Cytokeratin fragments 19)and conventional CT morphological features(including greatest tumor diamete,Spiculation sign,Lobulation sign,Pleural indentation,tumor density,Vacuole sign,cavity,Air bronchogram sign Vessel convergence sign)related to the degree of CD8+T lymphocyte infiltration.The tumor regions in each layer of images were segmenting under the lung window on the Rad Cloud cloud platform to obtain three-dimensional images,and 7 categories of imaging radiomics features were extracted,with a total of 1409 radiomics features.Dimensionality reduction and selection of optimal radiomics features through the Variance threshold method,Select KBest and Least absolute shrinkage and selection operator(LASSO)algorithms.And the radiomics prediction model was constructed by Support Vector Machine(SVM)algorithm.Finally,combined with the radiomics model,significant clinical parameters and CT features,a combined prediction model was constructed.Receiver operating characteristic curve(ROC)and Area Under ROC curve(AUC)are used to evaluate the predictive performance of the mode.Results:Among 113 patients,58 cases had high CD8+T lymphocyte infiltration and 55 cases had low CD8+T lymphocyte infiltration.After randomized grouping,91 patients were included in the training set and 22 patients in the validation set.Age was statistically significant between the two groups with high and low CD8+T lymphocyte infiltration.Other clinical parameters and traditional CT morphological characteristics were not statistically significant between high and low CD8+T lymphocyte infiltration group.After a series of dimensionality reduction algorithms,9 optimal features were selected from 1409 radiomics features to establish the radiomics model.The AUC value of the radiomics model in the training set is 0.85,the sensitivity is 0.77,and the specificity is 0.76;the AUC value of the validation set is 0.77,the sensitivity is 0.64,and the specificity is 0.82.Combine age and radiomics model to build clinical-radiomics model,The AUC of the training set of this model is 0.87,the sensitivity is 0.73,and the specificity is 0.72;The AUC,sensitivity and specificity of the validation set of the clinical-radiomics model were 0.71,0.73 and 0.73 respectively.In the training set,the AUC value of the clinical-radiomics model was higher than the radiomics model,but in the validation set,the AUC value of the clinical-radiomics model was significantly lower than the radiomics model.Conclusion:Patients with high CD8+T lymphocyte infiltration in T1 stage NSCLC were older than those with low CD8+T lymphocyte infiltration,and the difference was statistically significant.Traditional CT morphological features have no predictive value on the degree of CD8+T lymphocyte infiltration in T1 stage NSCLC.In the ability to predict the degree of CD8+T lymphocyte infiltration in T1 Non-small cell lung cancer,the clinical-radiomics model is not superior to the single radiomics model.The radiomics model has a good predictive performance for the degree of CD8+ T lymphocyte infiltration in T1Non-small cell lung cancer.Radiomics can provide a noninvasive analysis method for evaluating the prognosis of NSCLC and the efficacy of lung cancer immunotherapy.