Diagnostic Model And Immune-related Gene Prognostic Index Of Lung Adenocarcinoma Based On Radiomics And Genomics Analysis

Part 1 Diagnosis model of subsolid nodules(diameter<2cm)based on RadiomicsPurpose:To establish a diagnostic model for subsolid nodules(diameter<2cm)according to CT-based radiomics analysis.Methods:A retrospective analysis was performed on patients who underwent surgery at the First Affiliated Hospital of Soochow University from January 2020 to April 2022.170 patients with pulmonary nodules(diameter≤2 cm,ctr>0.5)were randomly divided into experimental and validation groups according to the ratio of 7:3.First,we used 3D Slicer software to extract the imaging histological features of lung adenocarcinoma by depicting the corresponding areas on the patients’ preoperative CT images based on the postoperative conventional pathology reports.Then,we used the LASSO(Least Absolute Shrinkage and Selection Operator,LASSO)algorithm to extract the Radiomics’ features and obtain the patient’s Radiomics scores(Radscore)to build the Radiomics model.Meanwhile,we performed univariate analysis and multifactor logistic regression analysis on clinical baseline data and imaging features of 170 patients to screen out independent predictors thus building a clinical model.Secondly,a comprehensive model was established by combining Radiomics models with clinical models using multifactorial logistic regression analysis.The three models were evaluated by receiver operating characteristic(ROC)curves and decision curve analysis(DCA).Finally,we performed a correlation analysis between Radscore and PD-L1 expression to assess the predictive capability of Radscore for PD-L1 expression.We analyzed the relationship between Radscore and Ki67 by a similar approach.RESULTS:By extracting the baseline clinical data and imaging features of patients,we established a clinical model with age and burr sign as independent predictor variables.The sensitivity of this model was 66.67%and the specificity was 85%.In the experimental group,by plotting the ROC curve and calculating the area under the curve(AUC).the radiomics model(AUC=0.9256)and the comprehensive model(AUC=0.9331)were better than the clinical model(AUC=0.8081).Meanwhile,the DCA suggested that the radiomics model and the comprehensive model,compared with the clinical model had a better net benefit.In the correlation analysis between Radscore and PD-L1 expression.Radscore was able to serve as an independent predictor with the area under the ROC curve of 0.7168.Radscore also was an independent predictor of Ki67 with AUC=0.8385.Conclusion:The radscores model and the comprehensive model have better diagnostic efficacy in identifying benign and malignant pulmonary nodules,and in predicting the degree of malignancy.Me1anwhile,Radscore could not predict PD-L1 expression very well,so it also couldn’t predict the efficacy and prognosis of immunotherapy.Therefore,in the second part,we establish a model based on genomics to evaluate the efficacy and prognosis.Part 2 Immune-related gene prognostic index for lung adenocarcinoma predicts patients’ prognosis and response to immunotherapyPurpose:Existing biomarkers do not fully reflect the interaction of different factors in the tumor microenvironment(TME),and we aim to identify a prognostic biomarker and predict whether patients will benefit from immune checkpoint blockade(ICB)therapy from multiple perspectives.METHODS:First,we performed weighted gene co-expression network analysis(WGCNA)and univariate COX regression analysis on the lung adenocarcinoma dataset from The Cancer Genome Atlas Program(TCGA)database,thus identifying a total of 55 immune-related pivotal genes.Then,16 genes were screened by multivariate COX regression analysis to construct the Immune-Related Gene Prognostic Ind-ex(IRGPI)model.The IRGPI model was validated by the GSE68465 gene set from Gene Expression Omnibus(GEO)dataset.Next,we divided the samples into high and low subgroups according to the median IRGPI and analyzed the differences between the clinical characteristics,prognosis between the two subgroups.Secondly,the predictive capability of IRGPI was assessed by comparing its AUC with those of existing biomarkers.Subsequently,we analyzed the molecular and immunological properties of the ICB.Finally,the correlation between IRGPI and cytotoxic T lymphocyte-associated antigen-4(CTLA4)expression was analyzed,while T cell dysfunction and exclusion(TIDE)scores were calculated to predict the likelihood of patients benefiting from ICB.RESULTS:Patients with high IRGPI had later clinical stage,more severe disease and poorer prognosis.Patients with low IRGPI had higher immune escape potential and were less effective in receiving immunotherapy.IRGPI is an important indicator that can be used as a predictor of patient prognosis,with AUCs greater than 0.70 at 1-,2and 3-years follow-up in the ROC curve.CONCLUSION:IRGPI is a promising biomarker that can be used to assess patient prognosis and whether they will benefit from ICB treatment.The biomarker can provide a link for imaging to predict tumor microenvironment.