Using Integrated Bioinformatics Methods To Identify Novel Molecular Therapeutic Targets And Potential Prognosis Biomarkers Based On The Concept Of Cytolytic Activity

Background Cutaneous melanoma(CM)is considered to be the most malignant skin cancer due to its characteristics of strong invasion,early metastasis,high recurrence,and ineffective treatment.Globally,the incidence of CM is increasing annually.CM is not only susceptible in elderly persons,but also the third most common cancer in people aged 20 to 39?years.The pathogenesis of CM is complex and its cause still stays unclear.At present,there are scarcely any effective treatment methods for its metastasis and palindromia.Conventional treatments such as surgical excision,radiotherapy,and chemotherapy can’t meet the needs of individualized therapy in clinical.Immunotherapy is one of the most promising immunotherapies.However,not all patients can be benefited from immunotherapy due to limited detection of molecular markers.Therefore,it is of great importance to explore more novel and reliable molecular therapeutic targets,excavate potential biomarkers,in order to achieve the goal of formulating individualized clinical treatment strategy,which shows great clinical value for enhancing the response to immunotherapy and prognosis of CM patients,A formula first devised by American scholars Michael S.Rooney et al.is used to quantify immune cytolytic activity(CYT),defined as the geometric mean of granzyme A(GZMA)and perforin 1(PRF1)expression in transcripts per million(TPM),which can significantly reflect the activation and immune status of CD8+ T cells.CYT correlates with immune responses to immunotherapy and can be used to predict the prognosis of various cancers.Studies have shown that higher CYT expression in colorectal cancer,hepatocellular carcinoma,and endometrial cancer has been confirmed to be associated with higher immunogenicity and a more favorable immune microenvironment,which can lead to better clinical outcomes.Objective The direct use of CYT to predict the prognosis of melanoma is significantly heterogeneous and quite inefficient.The drivers of CYT-related biological functions in melanoma and their effects on immune responses are not fully understood.The definition and description of CYT markers in melanoma are not clear,which complicates the prediction and evaluation of cutaneous melanoma immunotherapy.In this study,the potential roles of CYT-related genes were thoroughly explored by integrating various bioinformatics analysis techniques and consulting public databases such as TCGA,GEO,and HPA,to help improve the clinical prognosis and immunotherapy prediction of CM,and to identify possible CYT molecular therapeutic targets.Methods(1)All datasets containing RNA-seq data and corresponding clinical information from CM patients were obtained from The Cancer Genome Atlas(TCGA)and Gene Expression Omnibus(GEO)databases.The CYT index was calculated in different datasets.The univariate Cox analysis was performed to explore the relationship between CYT and prognosis.Meta-analysis further identified the consistency and heterogeneity of CYT index in three datasets.A total of 864 differentially expressed genes(DEGs)with CYT characteristics were detected via the differential analysis.(2)Non-Negative Matrix Factorization(NMF)algorithm was applied to identify CYTrelated molecular subtypes based on CYT-related genes.The biological uniqueness of different subtypes was verified from the perspectives of prognosis,immune microenvironment and other aspects.(3)The univariate Cox regression analysis was conducted to screen out the prognostic CYT-related genes(CYTRGs).The Lasso-Cox regression analysis was used to further reduce dimension and ultimately establish a CYT-based prognostic signature.(4)Receiver Operating Characteristic(ROC)curves,Kaplan-Meier(KM)survival analysis,Principal Component Analysis(PCA),Calibration Curve Method(CCM),Concordance index(C-index),Restricted Mean Survival(RMS)curves were used to verify the accuracy and reliability of the proposed signature.The multivariate and univariate Cox regression analysis were used to determine the prognostic performance of the signature compared with traditional clinical indicators such as age,sex,and TNM stage.(5)Subsequently,in terms of tumor mutation burden(TMB),copy number variation(CNV),tumor microenvironment(TME),infiltrated immune cells,gene set enrichment analysis(GSEA),immune checkpoint inhibitors(ICIs),the potential mechanisms of tumorigenesis and progression have been elucidated.Referring to the Genotype-Tissue Expression(GTEx)and Human Protein Atlas(HPA)databases,the expression of genes in the signature was evaluated at the transcription and protein level.The random forest algorithm was used to describe the relative prognostic importance of the signature genes.(6)Moreover,the signature was used to predict the differential sensitivity to chemotherapy and immunotherapy in CM patients.In the end,a nomogram was built to better aid clinical application.Results(1)CYT was confirmed to be correlated with prognosis in three independent datasets.Meta-analysis suggested that CYT index alone was inefficient in predicting the prognosis of melanoma patients because of the significant heterogeneity.(2)Up to 864 DEGs were identified.After the screening,553 prognosis-related CYTRGs were enrolled in further analysis.Two molecular subtypes(C1,C2)were identified by Non-Negative Matrix Factorization(NMF)algorithm,and the prognosis of patients in group C1 was better than that in group C2.The number of somatic mutations in group C1 was also higher than that in group C2.There were significant differences between C1 and C2 in GSVA pathway score,ESTIMATE microenvironment score and MCP-counter score.These results mean that the transcriptome-based isoforms identified had significant biological uniqueness.(3)A prognosis signature composed of 14 hub CYTRGs was established using LassoCox regression analysis.Based on the median of the risk score,all CM patients were divided into either the high-or low-risk group.KM survival analysis was conducted on them to find that the disease specific survival(DSS)of the high-risk patients was significantly higher than that of the low-risk patients.ROC curves demonstrated the specificity and sensitivity of the signature.Also,the results of PCA and t-SNE algorithm showed that the signature could be used to well identify the risk degree for CM patients.CYT was proved to be closely related to the prognosis of CM patients,and the 14-CYTRG signature could be served as an independent risk factor.Notably,the prediction of the 14 CYT-related gene signature for melanoma patients was validated in an external independent dataset.The repeatability of this classification was verified in an independent verification dataset,suggesting that our model was highly unlikely to be a false finding due to technical artifacts,chance,or bias in the eligibility criteria for the TCGA sample.(4)Via the findings from the nomogram,ROC curves,DCA curves,C-index,RMS curves,the newly-devised signature was confirmed as an efficient and robust biomarker,with the ability to actually benefit CM patients.Meanwhile,the signature was available as a tool to select appropriate therapy based on the chemotherapy and ICI immunotherapy response analyses.In addition,the low-risk patients exhibited higher immunogenicity,better immune function,and higher expressions of CYT and immune checkpoint genes,which made them more suitable for ICI treatment.(5)The immunohistochemical(IHC)images showed significant differences in gene expression between CM and normal tissues.According to the random forest algorithm,IFITM1,UBA7,CCL8,HAPLN3 and SEMA4 D were identified as key prognostic genes,which provided vital clues for future exploration of the potential roles of CYTRGs in the prevention and cure of cutaneous melanoma.Conclusion With the aid of integrated bioinformatics approaches,this study was the first to innovatively build a 14-hub gene signature based on the novel term CYT.The accuracy and dependability of the signature was then comprehensively verified through multiomics observations,using multi-means.The signature could be used as a practical tool in clinical diagnosis,prognosis analysis and therapy choices.In addition,IFITM1,UBA7,CCL8,HAPLN3 and SEMA4 D were identified by the random forest algorithm as the most critical genes for the prognosis of CM patients,which initially provided insight into exploring the pathogenesis of CM in the future.