Identification of a Novel Prognostic Signature Based on N-Linked Glycosylation and Its Correlation with Immunotherapy Response in Hepatocellular Carcinoma

Figures & data

Figure 1 DEGs were identified in normal and tumor tissues to classify HCC patients.

Notes: (A) The heatmap of the N-linked glycosylation related DEGs (blue: low expression level; red: high expression level). *P<0.05; **P<0.01; ***P<0.001. (B) 374 HCC patients were classified into two clusters according to the consensus clustering analysis (k = 2). (C) Kaplan–Meier curves of the survival analysis based on the two clusters. (D) Heatmap based on DEGs associated with clinical features (TNM staging, the degree of tumor differentiation, gender, age). ***P<0.001.

Figure 2 Construction of risk model based on N-linked glycosylation related DEGs in TCGA dataset.

Notes: (A) Univariate cox regression analysis of N-linked glycosylated gene significantly associated with prognosis. (B) Cross-validation for tuning the parameter selection in the LASSO regression. (C) Distribution of HCC patients in TCGA cohort by risk score. (D) Survival status of HCC patients with different risk scores (low-risk population: the left of the dotted line; high-risk population: the right of the dotted line). (E) Principal component analysis plot for HCCs of TCGA based on the risk score. (F) t-Distributed Stochastic Neighbor Embedding (t-SNE) for HCCs of TCGA based on the risk score. (G) Survival curves for the 2 clusters. (H) Univariate analysis in TCGA cohort. (I) Multivariate analysis in TCGA cohort. (J) The AUC curve proves the feasibility of the prediction model.

Figure 3 Validation of prognostic model.

Notes: (A) Distribution of HCC patients in GEO cohort by risk score. (B) Heatmap containing risk scores and clinical features. *P<0.05; **P<0.01; ***P<0.001. (C) Kaplan–Meier curves for comparison of patients between high- and low- risk subgroup. (D) Univariate analysis in GEO cohort. (E) Multivariate analysis in GEO cohort. (F) Time‐dependent ROC analysis for OS prediction in GEO cohort.

Figure 4 Correlation of the HCC subgroups with immune infiltration.

Notes: (A and B) 16 types of immune cells and 13 types of immune function were compared between the high and low risk groups in TCGA cohort. *P<0.05; **P<0.01; ***P<0.001. (C and D) 16 types of immune cells and 13 types of immune function were compared between the high and low risk groups in GEO cohort. *P<0.05; **P<0.01; ***P<0.001. (E) Comparison of 8 different immune checkpoints between high and low risk groups. ***P<0.001.

Figure 5 Prognostic models are effective in predicting responses to immunotherapy.

Notes: (A–C) Representative images of the immunohistochemistry of STT3A, DDOST and TMEM165 in normal tissues and HCC tissues. (D–F) Progression-Free Survival (PFS) between HCC patients with the low and high expression of STT3A, DDOST and TMEM165 in our cohort. (G) Progression-Free Survival (PFS) between HCC patients with the low and high risk score in our cohort. (H) Comparison of the different stage in HCC patients with the low and high risk subgroup. (I) The time-dependent ROC analysis proves the feasibility of the prediction model in our cohort. (J) Comparison of the incidence of the response to immunotherapy in HCC patients with the low and high risk group. (K) The different expression of CD20 between the low and high risk group. **P<0.01. (L) Representative images of the immunohistochemistry of CD20 in HCC patients with the low and high risk score.

Figure 6 Knockdown of the N-linked glycosylation prognostic model related genes suppresses the proliferation of HCC cell lines.

Notes: (A) RT-qPCR analysis of STT3A, DDOST and TMEM165 mRNA expression in hepatocellular carcinoma tissue and paired adjacent liver tissue. *P<0.05; ***P<0.001. (B) The knockdown efficiency of siRNA on STT3A, DDOST and TMEM165 in Huh7 and MHCC-97H cells was detected by RT-qPCR analysis. ***P<0.001. (C) The silencing efficiency of siRNA on STT3A, DDOST and TMEM165 in Huh7 and MHCC-97H cells was detected by Western Blotting analysis (the order of the first blot is Huh7-siNC, Huh7-siSTT3A, 97H-siNC, 97H-siSTT3A; the order of the second blot is Huh7-siNC, Huh7-siDDOST, 97H-siNC, 97H-siDDOST; the order of the third blot is Huh7-siNC, Huh7-siTMEM165, 97H-siNC, 97H-siTMEM165). (D) Cell viability detected by CCK8 assay at 450 nm in Huh7 and MHCC-97H cells. ***P<0.001. (E) Cell viability tested by EdU Kit in Huh7 and MHCC-97H cells. *P<0.05; **P<0.01; ***P<0.001. (F) The effect of silencing STT3A, DDOST and TMEM165 on HCC proliferation was tested by clone formation assay in Huh7 and MHCC-97H cells. *P<0.05; **P<0.01; ***P<0.001.

Figure 7 Flow diagram of this study.

Data Sharing Statement

Seventy N-linked glycosylation related genes were retrieved from the AmiGO2 database (http://amigo.geneontology.org/amigo).

The datasets generated and analyzed during the current study are available (https://portal.gdc.cancer.gov/) on February 12, 2022. A cohort of HCC patients with RNA-sequence and clinical features were download from GEO database on January 22, 2009, and the last follow-up update was on October 6, 2021 (https://www.ncbi.nlm.nih.gov/geo, ID: GSE14520).