Identification of succinylation related lincRNA in colon tissue samples
The matrix expression of top 30 succinylation genes and 16,877 lncRNAs was screened and then selected succinylation-related lncRNAs by |Pearson R| > 0.4 and p < 0.001. Subsequently, we constructed the network of succinylation-lncRNA coexpression in the Sankey diagram (Figure 2A). The correlation with succinylation genes and 300 succinylation-related lncRNA shown in Figure 2B.
Construction a risk model through succinylation associated lncRNAs in patients of CC
We screened succinylation-related prognostic lncRNAs by univariate Cox regression analysis and nine succinylation-related lncRNAs were significantly correlated with OS in Figure 3A. LASSO Cox regression analysis was used to conduct a novel prognostic target for predicting clinical results, avoiding the occurrence of collinearity of transcriptome data[15]. Therefore, we constructed the lasso cox analysis in the Figure 3B and C and ultimately identified the six lncRNA including FOXD3-AS1, ANK3-DT, EIF3J-DT, MIR210HG, MAFA-AS1, AC024581.1.
We then explore the expression of succinylation-related lncRNA in the CC. The lncRNAs ANK3-DT, EIF3J-DT, MAFA-AS1 and AC024581.1 were demonstrated at high expression in tumor tissues(Figure 3E, F, H and I). However, in the lncRNA FOXD3-AS1 displays the low expression in the tumor tissues(Figure 3D). Besides, lncRNA MIR210HG shows no difference in the CC patients(Figure 3G).
To further examine whether the succinylation-related lncRNAs is related to prognostic capability, we assessed the risk scores in the training set, test set and entire set. Based on the median value of the training set, test set and entire set, six succinylation-related lncRNAs expressed according to the high- and low-risk groups in the heatmap (Figure 4A, B and C). Besides, people were assigned into two group, with the risk score increased, the patients’ death rose and the patients’ survive time decreased in the training set (Figure 4D and G), test set (Figure 4E and H) and entire set (Figure 4F and I). Finally, patients in the high group for colon cancer had a poor OS than those with the low group through the Kaplan-Meier survival analysis. (training set : p<0.001 shown in Figure 4J; test set : p<0.05 shown in Figure 4K; entire set : p<0.001 shown in Figure 4L)
Subsequently, we explored the progression-free survical (PFS), disease-specific survival (DSS) and disease-free survival (DFS) between the high and the low group in CC patients (Figures 5A-C),which shows the high risk of succinylation-related lncRNAs with poor prognosis. Above all, these results implied that succinylation-related lncRNAs signature has an excellent predictive value.
PCA revealed the difference of the succinylation-related lncRNAs signature compared with the low- and high-risk specimens based on the entire gene expression (Figure 6A), succinylation genes (Figure 6B), succinylation-related lncRNAs (Figure 6C) and risk model classified (Figure 6D) based on the succinylation-related lncRNAs, respectively. These results confrmed a risk signature’s classifcation ability between the low- and high-risk groups.
Correlation between the risk score and clinicopathological features
We applied to the discrepancies in OS stratified to study whether the high and low-risk groups according to the universal clinicopathologic characteristics. Moreover, we explored the predictive values of succinylation-related lncRNAs signature, we classified the subgroups by gender, age and tumor stage (Figure 7). In the subgroups other than the M1 stage ( P=0.234, Figure 7F), N0 stage ( P=0.051, Figure 7G), N1 stage ( P=0.194, Figure 7H), Stage I-II (P=0.21, Figure 7J) and T1+T2 stage ( P=0.071, Figure 7L), the results revealed that succinylation-related lncRNAs signature possessed better prognostic values, especially, the OS of patients in the high-risk group discovered an unfavourable prognosis ( P<0.05). Above all in the study, we revealed that succinylation-related lncRNAs signature acts as the novel pivotal indicator to predict the CC patients of prognosis values.
Evaluation of the tumor immune microenvironment and functional enrichment analysis by succinylation-related lncRNAs signature
To explore the asscociation with succinylation-related lncRNAs signature and immune cell infltration, the heatmap displayed the expression of immune indicators such as CCR, T cells co-stimulating and inflammation promoting of the high- and low- risk groups (Figure 8A). To identify the potential biological processes related to the succinylation-related lncRNAs, we screened out Gene Ontology (GO) enrichment analysis (Figure 8B). Based on the effect of predicted variance, we then explored the mutation data using the R package maftools . The top 15 genes with the frequency of highest alteration between the high- and low-risk groups displayed in Figures 8C and 8D and found the high-risk group showed a higher mutation frequency than the low-risk group.
Several studies concluded that higher TMB in patients were related to improve response while receiving immune checkpoint blocking therapy[16]. TMB can be suggested as a predictive biomarker for cancer immunotherapy. Then, we assessed TMB scores, detecting the prognosis correlation with TMB, showing high TMB scores correlated with a worse survival (Figure 8E). Besides, we validated whether the succinylation-related lncRNAs model could be better to predict the OS outcome better than TMB scores. The high TMB scores and low TMB scores with CC patients in the high-risk groups (H-TMB of high risk and L-TMB of high risk) displayed a poorer OS than patients with high- and low- TMB scores in the low-risk groups (Figure 8F).
We constructed the association with succinylation-related lncRNAs signature and clinical features in patients of CC through the univariate in Figure 9A and multivariate in Figure 9B Cox regression analysis. Moreover, with survive time increased, the risk score index was better, implying that the prognosis of patients in CC might be forecasted better by the risk grade we constructed (Figure 9C). The ROC curve demonstrated that succinylation-related lncRNAs signature to predict the 1-, 3-, and 5-year with AUCs of 0.694, 0.729, and 0.802 (Figure 9D). The predictive nomogram showed that the overall survival rates could be predicted relatively well across the entire cohort compared to the ideal model (Figure 9E and F).
Screening Novel Potential Compounds Targeting succinylation-Related lncRNA Models
According to the above analysis, to identify potential drugs targeting via succinylation-related lncRNA model we identified to apply for the CC patients, we evaluated treatments values according to the half-maximal inhibitory concentration (IC50) by the GDSC database. We found that the Cisplatin and Temozolomide compounds show significantly difference between the two groups (Figure 10). Besides, the compounds show positive correlation with the risk score we constructed. Therefore, we constructed the succinylation-related lncRNA more effective for high-risk patients.