Establishment Of Predictive Model For Gastric Cancer Prognosis With LncRNA Based On TCGA And Elastic Net

Objectives Based on the Cancer Genome Atlas(TCGA)database,using Cox proportional hazards regression model and elastic net algorithm to screen the key long non-coding RNA(lncRNA)related to the prognosis of gastric cancer and construct a predictive model.Methods Downloading 407 patients of gastric adenocarcinoma RNA-Seq,clinical and prognosis data in the TCGA database,dividing and processing the data into training and test sets.Using the training set data,univariate Cox proportional hazards regression model was used to screen differentially expressed lncRNAs related to the prognosis of gastric cancer.Based on the screening results,the multivariate Cox proportional hazards regression model and the elastic net algorithm were used to screen the key differentially expressed lncRNAs for gastric cancer prognosis to construct a predictive model.The receiver operating characteristic curve(ROC)was used to analyze the area under the ROC curve(AUC)of the model to evaluate the model’s ability to distinguish the prognosis of gastric cancer.Based on the model,the risk score of each patient was calculated,the median score was used as a cutoff value to divide the subgroups of high and low risk of gastric cancer prognosis.The Kaplan-Meier method was used to compare the survival difference between gastric cancer patients in the two subgroups,and the significance of the difference in survival distribution between the gastric cancer patients in the two subgroups was tested by Log-Rank method.Using the test set,the gastric cancer prognosis lncRNA prediction model constructed in the above steps was verified and evaluated separately.Finally,using starBase database to predict lncRNA markers target gene for gastric cancer prognosis.Results 1 A total of 318 gastric cancer cases were included in this study,and 1369 differentially expressed lncRNAs related to the prognosis of gastric cancer were screened,among which 751 were up-regulated and 618 were down-regulated.2 The AUC of 32-lncRNA predictive model based on the Cox proportional hazards regression model was 0.827.The median risk score was-13.007 after calculating the patient risk score in the training set,it was used as the cutoff value to divide the high and low risk subgroups.The high risk group’s overall survival(OS)of patients was significantly lower than the low risk group(P<0.05),and there was no significant difference between them in the test set(P=0.22).3 Based on the elastic net,when the optimal adjustment parameters α=0.1 and λ=0.12 were determined,the root mean square error was the lowest,and a 19-lncRNA predictive model was constructed by 19 lncRNAs,the model’s AUC was 0.831.In the training and test sets,the OS of patients in the high risk group was significantly reduced compared with the low risk group(P<0.05).4 Using starBase database for target gene prediction analysis found that SLC7A2 may be the target gene of HLX-AS1,and correlation analysis found that HLX-AS1 and SLC7A2 are closely related in gastric cancer(r=0.306,P=1.52e-09).Conclusions The 19-lncRNA marker gastric cancer prognosis predictive model has good stability and differentiation ability,and it may provide a reference for judging the prognosis of gastric cancer patients.Figure 13;Table 5;Reference 147...