Protein-coding genes combined with long noncoding RNA as a novel transcriptome molecular staging model to predict the survival of patients with esophageal squamous cell carcinoma

Abstract

Background: Esophageal squamous cell carcinoma (ESCC) is the predominant subtype of esophageal carcinoma in China. This study was to develop a staging model to predict outcomes of patients with ESCC.

Methods: Using Cox regression analysis, principal component analysis (PCA), partitioning clustering, Kaplan-Meier analysis, receiver operating characteristic (ROC) curve analysis, and classification and regression tree (CART) analysis, we mined the Gene Expression Omnibus database to determine the expression profiles of genes in 179 patients with ESCC from GSE63624 and GSE63622 dataset.

Results: Univariate cox regression analysis of the GSE63624 dataset revealed that 2404 protein-coding genes (PCGs) and 635 long non-coding RNAs (lncRNAs) were associated with the survival of patients with ESCC. PCA categorized these PCGs and lncRNAs into three principal components (PCs), which were used to cluster the patients into three groups. ROC analysis demonstrated that the predictive ability of PCG-lncRNA PCs when applied to new patients was better than that of the tumor-node-metastasis staging (area under ROC curve [AUC]: 0.69 vs. 0.65, P < 0.05). Accordingly, we constructed a molecular disaggregated model comprising one lncRNA and two PCGs, which we designated as the LSB staging model using CART analysis in the GSE63624 dataset. This LSB staging model classified the GSE63622 dataset of patients into three different groups, and its effectiveness was validated by analysis of another cohort of 105 patients.

Conclusions: The LSB staging model has clinical significance for the prognosis prediction of patients with ESCC and may serve as a three-gene staging microarray.

Keywords: Esophageal squamous cell carcinoma; Long non-coding RNA; Overall survival; Protein-coding gene; Staging model; Transcriptome.

Fig. 1

Schedule of the analyses used to develop the transcriptome molecular staging model and validate its predictive efficiency. PCG protein-coding gene, lncRNA long non-coding RNA

Fig. 2

The patients identified from the GSE53624 dataset (n = 119) are grouped with three risk stages. a Univariate Cox proportional hazards regression analysis of the expression profiling data of PCGs and lncRNAs. b Eigenvalues of the principal components show most of the variance in the GSE53624 dataset is contained in the first three principal components. c Clustering of the patients with ESCC identified from the GSE53624 dataset according to the three principal component scores using NbClust (Euclidean distance, complete linkage) indicates that optimal cluster number was three with the largest index. d Principal component analysis of the GSE53624 dataset. Axes are principal components 1, 2, and 3. PCG protein-coding gene, lncRNA long non-coding RNA

Fig. 3

Survival prediction power of PCG-lncRNA grouping versus TNM staging for patients identified from the GSE53624 dataset. a Kaplan–Meier analysis of patient survival when the PCG-lncRNA grouping is applied. b Kaplan–Meier analysis of patient survival when TNM staging is applied. c Comparison of the PCG-lncRNA grouping and the TNM staging systems using ROC analysis. PCG protein-coding gene, lncRNA long non-coding RNA

Fig. 4

The LSB staging model comprising SEMA3A, BEX2, and LINC01800 selected using classification and regression tree (CART) analysis. a SEMA3A, BEX2, and LINC01800 form the classification tree generating using CART analysis. The percentage represents the proportion of patients at every LSB stage in the training set. b Test error result of the classification tree. c Multiclass ROC analysis was performed in the training set, test set, and entire GSE53624 dataset

Fig. 5

Validation of the LSB staging model using the GSE53622 dataset (n = 60) (a, b) and experimental dataset (n = 105) (c, d). Kaplan–Meier analysis and comparison of the LSB staging model and the TNM staging using ROC analysis

Fig. 6

Coexpression network analysis and prediction of the function of SEMA3A, BEX2, and LINC01800. a Coexpression network of SEMA3A, BEX2, and LINC01800 with other genes in the GSE53624 and GSE53622 datasets (Pearson correlation coefficient > 0.5, P < 0.05). Blue or red genes were coexpressed with two or one of the three identified genes in the LSB staging model, respectively. b Functional enrichment of the protein-coding genes which were coexpressed with SEMA3A, BEX2, and LINC01800, using ClueGo