Value Of Radiomics In The Assessment Of EGFR Gene Mutation Status In Lung Cancer Liver Metastases

Purpose: The purpose of this study was to investigate the potential value of imaging techniques to identify epidermal growth factor receptor(EGFR)mutant and EGFR wild type in patients with liver metastases from lung cancer.Two radiomics models,one based on computed tomography(CT)images of primary cancer combined with magnetic resonance imaging(MRI)images of liver metastases,and the other based on the tumor-liver interface(TLI)region of multi-sequence MRI,were developed to predict the EGFR mutation status in patients with lung cancer liver metastases.Materials and Methods: This study retrospectively included 174 patients from August2017 to December 2021,including 130 patients from Liaoning Cancer Hospital and 44 patients from Shengjing Hospital of China Medical University,and all patients were pathologically confirmed as patients with lung cancer liver metastases.The third chapter section extracts radiomics features and establishes radiomics models based on the intraand peritumoral regions of CT images of primary tumors and the MRI of liver metastases in patients with lung cancer liver metastases.The fourth chapter builds radiomics models based on contrast enhanced T1-weighted(CET1)and T2-weighted(T2W)MRI sequential images of patients with lung cancer liver metastases,based on TLI regions and the whole tumor regions,respectively.The Least absolute shrinkage and selection operator(LASSO)algorithm was applied to select the extracted features and build a radiomics signature(RS)model for predicting the EGFR mutation status of patients with lung cancer liver metastases.The predictive power of the RS models was evaluated by plotting the receiver operating characteristic(ROC)curves and the blending matrix.The performance of the models was evaluated in terms of area under curve(AUC),sensitivity(SEN)and specificity(SPE).Results: The results in Chapter 3 show that RS models constructed based on patients’ primary lung cancer CT images and liver metastases MRI images and their fusion forms achieved the highest prediction performance in the training set(AUCs,RS-primary lung cancer vs.RS-liver metastases vs.RS-Combined,0.826 vs.0.821 vs.0.908)and test set(AUCs,RS-primary lung cancer vs.RS-liver metastases vs.RS-Combined,0.760 vs.0.791 vs.0.884)in which fusion RS achieved the highest predictive performance through the combination of primary tumor and metastasis.The results of clinical factor analysis showed a significant difference in smoking status between the EGFR mutant and EGFR wild-type groups in the training set(P<0.05).The results in Chapter 4 show that the TLI region(RS-TLI)and the whole tumor region(RS-W)based on multiple sequences of MR images were screened for five and six features highly correlated with EGFR mutation status,respectively.The RS model constructed based on the TLI regions both showed better prediction performance to identify the EGFR mutation status in the training set(AUCs,RS-TLI vs.RS-W,0.842 vs.0.797),internal validation set(AUCs,RS-TLI vs.RS-W,0.771 vs.0.676)and external validation set(AUCs,RS-TLI vs.RS-W,0.733 vs.0.679)than the RS model constructed based on the whole tumor region.Conclusion: The results of this study suggest that MRI of patient of lung cancer liver metastasis-based radiomics model can be used to detect EGFR gene mutation status.The developed multi-organ combined radiomics signature model may help guide individual treatment strategies for patients with lung cancer liver metastatic.TLI region-based medical imaging can improve the predictive performance of radiomics model to identify EGFR mutations status in patients with lung cancer liver metastases.The established multiparametric MRI radiomics model may be used as a new marker for personalized treatment planning.