Construction And Evaluation Of Prognostic Model For Hepatocellular Carcinoma Based On Pyroptosis-related LncRNA

Objective:In this study,we analyzed the relationship between pyroptosis-related long non-coding RNAs（pylncRNAs）and the prognosis of hepatocellular carcinoma（HCC）by bioinformatics to constrcut a prognostic model.Then we explored the application value of this model in evaluating immune cell infiltration and predicting chemotherapeutic drug sensitivity.Methods:Transcriptome data and clinical data of HCC（374 cases）and adjacent normal tissues（50 cases）were downloaded from The Cancer Genome Atlas（TCGA）database,and 51 pyroptosis genes were retrieved from the Molecular Signatures Database（MSig DB）and previous research.First,we used Perl software to extract pyroptosis genes and lncRNAs.Using R software,correlation analysis was performed on pyroptosis genes and lncRNAs to screen pylncRNAs（selection criteria:|cor|>0.4,P<0.001）.Then,R software was used to screen the differentially expressed pyroptosis-related lncRNAs（DEpylncRNAs）in HCC and adjacent tissues（selection criteria:|log₂FC|>1 and FDR<0.05）.Subsequently,we conbined the expression data of DEpylncRNAs with the survival data,removed the samples with incomplete survival data and screened the prognosis-related pylncRNAs in HCC by Univariate Cox analysis.After that,all TCGA patients were randomly divided into training and validation cohorts with the ratio of 1:1.In the training cohort,based on the results of univariate Cox analysis,LASSO regression analysis was used to identify the pylncRNAs that were most associated with the prognosis of HCC and thus a prognostic model was constructed.Based on the median risk score,the patients in the training cohort were divided into the high-risk group and the low-risk group.Principal component analysis（PCA）,t-distributed stochastic neighbor embedding（t-SNE）analysis,Kaplan-Meier（K-M）analysis and receiver operator characteristic（ROC）curves were used for evaulation.The stability and accuracy of the model were verified in the validation cohort and the entire cohort.In addition,based on the high-or low-risk groups distinguished by the model,ESTIMATE analysis and single sample gene set enrichment analysis（ss GSEA）were used to evaulate the immune cell infiltration;the"p RRophetic"package was used to estimate the sensitivity of chemotherapeutic drugs in different groups;GSEA analysis was used to identify the differentially activated biological pathways in different groups.Finally,we verified the expression of pylncRNAs in the model based on GEPIA database and evaluated the prognostic value of these pylncRNAs by K-M analysis.Results:（1）763 DEpylncRNAs were identified based on correlation analysis.Then,524 DEpylncRNAs were screened,of which 518 DEpylncRNAs were up-regulated and 6 DEpylncRNAs were down-regulated.（2）120 prognosis-related pylncRNAs were identified based on Univariate Cox regression analysis from the DepylncRNAs.（3）The patients with HCC were randomly divided with the ratio of 1:1 into training cohort（n=184）and validation cohort（n=181）.In the training cohort,LASSO regression analysis identified 4 pylncRNAs（PXN-AS1,AL031985.3,MKLN1-AS and AC025178.1）and thus a prognostic model was constructed.The risk score of each patient was calculated according to the model.Then,the patients from training cohort were divided into high-risk and low-risk groups based on the median risk score.PCA and t-SNE analysis showed that the model could accurately distinguish the patients in high-risk and low-risk groups.K-M analysis showed that patients in the high-risk group had worse prognosis than the patients in the low-risk group（P<0.001）.ROC curves revealed that Areas Under Curves（AUCs）with 1～5 years survival rate were0.752,0.741,0.729,0.704 and 0.710.The analysis results of validation cohort and the entire cohort matched well with the results of training cohort.In addition,survival analysis of clinical subgroups showed that the model was suitable for HCC patients with different clinical characteristics.（4）Univariate and multivariate Cox analysis showed that stage III,stage IV and the risk score were independent prognostic factors for the patients with HCC.Nomogram plot was constructed based on TNM stage and the risk score,and it could intuitively and effectively predict the prognosis of the patients with HCC.（5）Based on high-risk and low-risk groups distinguished by the model,ESTIMATE analysis showed that the stromal score of the high-risk group was significantly lower than that of the low-risk group,and the stromal score of HCC patients was negatively correlated with the risk score.The ss GSEA analysis showed there were differences about the abundance of 10 types of immune cells in high-risk and low-risk groups.（6）Drug susceptibility analysis showed that patients in the high-risk group were more sensitive to cisplatin,doxorubicin,gemcitabine,mitomycin C and paclitaxel while patients in the low-risk group were more sensitive erlotinib,gefitinib and sorafenib.（7）GSEA analysis showed that the JAK-STAT signaling pathway,chemokine signaling pathway,MAPK signaling pathway and Toll-like receptor signaling pathway were activated in the high-risk group.（8）The results of GEPIA database showed that PXN-AS1,AL031985.3 and MKLN1-AS were highly expressed in HCC,while AC025178.1 was lower expressed in HCC.Moreover,K-M analysis showed that the high expression of PXN-AS1,AL031985.3 and MKLN1-AS was associated with the poor prognosis of HCC.Conclusion:Based on the 4 pylncRNAs,we successfully constructed a prognostic model of HCC with good predictive ability.This model not only helps to evaluate the immune cell infiltration in the tumor microenvironment among different risk groups,but also provides effective guidance for clinical chemotherapy of HCC patients.In addition,PXN-AS1,AL031985.3 and MKLN1-AS might be biomarkers for poor prognosis of HCC.