Multi-omics Comprehensive Analysis Predicts Survival Prognosis And Treatment Response In Patients With Non-small Cell Lung Cancer

Objective:Based on the analysis of the expression levels of glycosylation-related genes in patients with non-squamous non-small cell lung cancer(NSCLC)based on tissue transcriptomics in public databases,a risk score model for survival and prognosis of patients and a prediction model for preoperative diagnosis of lymph node metastasis were constructed.Using transcriptomics and genomics to analyze the correlation between glycosylation-related genes and various biological functions,tumor immune microenvironment and tumor mutation load,to explore the impact of abnormal glycosylation levels in tumor tissue and disease progression in patients with non-squamous NSCLC.A prognostic nomogram was developed combining the survival prognostic risk score model of patients with glycosylation-related genes and the clinical characteristics of patients to provide information for patient prognosis and clinical management decisions.Methods:The transcriptomic and genomic detection results of 2383 non-squamous NSCLC cases and 637 normal tissues were collected from the public database to search for differentially expressed glycosylation-related genes.1278 cases were randomly divided into training cohort and test cohort at 1:1,and the remaining 1105 cases were used as an independent validation cohort.Univariate cox regression,LASSO regression and multivariate cox regression were used to establish and verify the prognosis prediction model,and develop a prognostic nomogram Map and make it into a web tool.The transcriptomics data of NSCLC samples in the TCGA database were collected,and the lymph node metastasis was divided into three ordered classifications of NO<N1<N2/3 and constructed Diagnosis prediction model,using 7 data sets to form an external validation cohort to verify the accuracy of the diagnosis model.Use ssGSEA analysis,Esitimate,MCP-Counter,xCELL,Immunophenoscore(IPS)and other algorithms to analyze the association of glycosylation-related genes with various biological functions and tumor immune microenvironment.The predictive effect of the model on the efficacy of immunotherapy was evaluated using TIDE and Submap algorithms,and validated in an independent data set.The Oncopredict algorithm was used to estimate the model’s ability to predict the drug sensitivity of various anticancer drugs in patients.Results:The survival prognostic model established using glycosylation-related genes achieved good prediction results in the training set,test set,and independent validation cohort.Using the model to predict the 5-year overall survival(Overall survival,OS)of patients,the AUC in the training set was 0.745,the AUC in the test set was 0.668,and the AUC in the three external validation data sets were 0.638,0.641,and 0.606,respectively.Further research found that the model also had a good predictive effect on the progressionfree survival(PFS)of patients,and the AUCs for predicting patients'5-year PFS in the three validation datasets were 0.713,0.615,and 0.606,respectively.Combined with clinical characteristics and risk scores,Nomogram was used to develop a comprehensive model.The model predicted the patient’s survival effect well,and the AUC for predicting 5-year OS was 0.763.Using the Shiny tool,the prognostic prediction comprehensive model was made into a web tool and deployed on the Internet for clinical use.For the lymph node metastasis diagnosis prediction model,in the training set,the prediction AUC=0.76 for NO patients,the prediction AUC for N1 patients=0.70,and the prediction AUC for N2/3 patients=0.56;in the external validation cohort,the prediction NO The AUC of N1 stage of lymph node metastasis was 0.646,the AUC of N1 stage of lymph node metastasis was 0.614,and the AUC of N2/3 stage of lymph node metastasis was 0.545.The model has a good diagnostic effect on patients with NO and N1 stages,but it has a poor diagnostic effect on patients with N2/3 stages.It may be that there are few samples of N2/3 stages in the included data,and the model lacks training for these patients.Multi-omics analysis showed that the abnormal expression of glycosylation-related genes was related to various tumorpromoting biological functions.Patients in the high-risk group had worse anti-tumor immune function and higher tumor mutation burden.Further research found that the model also had a good predictive effect on the efficacy of immunotherapy,and the risk score of glycosylation-related genes was significantly correlated with the sensitivity of various antitumor drugs.Conclusions:The glycosylation-associated gene signature identified in this study is a promising set of NSCLC prognostic biomarkers.The model can better predict OS and PFS of patients.At the same time,the expression level of glycosylation-related genes can also be used to diagnose and predict the lymph node metastasis of NSCLC patients.Further studies have found that the expression level of glycosylation-related genes is related to various biological functions and anti-tumor immunity.After verification,it is found that the characteristics of glycosylation-related genes can be used as potential predictors of the efficacy of immunotherapy and chemotherapy.In summary,this study found that glycosylation is involved in many mechanisms related to the progression of non-squamous NSCLC,which deserves further study.