A Novel Lipid Metabolism and Endoplasmic Reticulum Stress-Related Risk Model for Predicting Immune Infiltration and Prognosis in Colorectal Cancer

Figures

• 

  Figure 1. Construction of the prognostic signature based on differentially expressed LEGRs. (A) The volcano plot of differently expressed genes in TCGA. (B) The Venn diagrams screening differently expressed LERGs. (C) A total of 28 survival-related genes via Cox regression. (D,E) LASSO coefficient profiles of the 28 survival-related genes. (F) Coefficient profile plot produced against the log (lambda) sequence in the LASSO model. The optimal parameter (lambda) was selected as the first dotted black line indicated.

• 

  Figure 2. Evaluation of the risk model’s efficiency in predicting patient survival. (A–D) Survival curves for the high-risk and low-risk groups are exhibited, as well as the ROC curves and AUCs for predicting 1-year, 3-year, and 5-year OS in the TCGA training dataset, TCGA test dataset, GSE12945 validation dataset, and GSE17536 validation dataset.

• 

  Figure 3. Construction of a predictive nomogram for evaluating patient survival. (A,B) Independent clinical prognostic factors for CRC patients identified via univariate and multivariate Cox regression. (C) A nomogram for predicting OS in CRC patients. (D) ROC curves. (E) Calibration plots for predicting 3-year and 5-year survival for patients. (F) DCA for evaluating clinical utility.

• 

  Figure 4. Correlation between the risk score and multiple clinical indicators. (A–F) Clinical indicators, including age, gender, stage, T, N, and M, were evaluated for their correlation with the risk score in boxplots.

• 

  Figure 5. Schemes follow the same formatting. (A,B) The mutated genes of the high-risk group and the low-risk group from the TCGA cohort. (C) Combined TMB and risk score for predicting CRC patient survival. (D) Combined MSI and risk score for predicting CRC patient survival.

• 

  Figure 6. Drug sensitivity and molecular pathway for the risk model. (A) Therapeutic drugs showing significant IC50 differences in high- and low-risk groups. (B) GSVA results revealing the main divergent pathways. (C) GSEA results showing the top three pathways in two subgroups.

• 

  Figure 7. The different immune landscapes between the low- and high-risk groups. (A) CIBERSORT analysis. (B) Correlation between risk score and immune cell content. (C) Correlation between risk score and immunotherapy response. (D) Correlation between lipid metabolism/ER stress and immune infiltration via ssGSEA. (E) Expression variation in the immune checkpoint. (F) Expression variation of chemokines. * p < 0.05; ** p < 0.01; *** p < 0.001; and **** p < 0.0001. ns p > 0.05.

• 

  Figure 8. Construction of a WGCNA co-expression network and upstream mechanism analysis of hub genes. (A,B) A soft threshold β was set to six. (C) A total of eight modules were obtained. (D) The brown module is most related to the risk sore. (E) The Venn diagrams for screening hub genes. (F) PPI analysis in six hub genes. (G) TF prediction of hub genes.

• 

  Figure 9. Validation of hub genes in the CRC clinical sample. Quantitative real-time polymerase chain reaction (qRT-PCR) analyses of AGRN (A), APOE (B), CAV1 (C), SPP1 (D), MMP3 (E), and TIMP1 (F) expression in 12 pairs of CRC tissues and adjacent non-cancer tissues. Immunohistochemistry showed the protein expressions of six hub genes based on the HPA database in normal and CRC tissues. ns p > 0.05. * p < 0.05, ** p < 0.001, and *** p < 0.001.

• 

  Figure 10. Single-cell analysis for the immune landscape. (A,B) A total of 9 cellular categories were identified from 17 subtypes. (C,D) Interaction patterns exist between these cellular subtypes in the high-risk and low-risk groups. (E,F) The failure scores and cytokine scores obtained were compared between the high- and low-risk groups. (G) The expression of hub genes in nine cell types. **** p < 0.0001.