3.1 Clinical characteristics of patients
The data (shown in Table 1) were extracted from TCGA in December 2019 and included 374 primary HCC cases with both clinical and gene expression data. Clinical characteristics of HCC patients involving age, gender, BMI, family history of cancer, histologic grade, clinical stage, topography (T), lymph node (N), metastasis (M), residual tumor, tumor status, vascular invasion, Child-pugh, AFP, new tumor event, .These patients included 122 female patients and 255 male patients ,and most of them (91.0%, n=342) were over 40 years old. The tumor status included 236 (67.6%) tumor free and 113 (32.4%) with tumor. In the study cohort, 161 of 341 (47.2%) patients were overweight and had a BMI of more than 25 ,114 of 326 (35.0%) patients had HCC family history, 118 of 358 (33.0%) had history of alcohol consumption. As for tumor grade, there were 235 (63.1%) in G1 and G2, 137 (36.8%) were in G3 and G4. The cancer stage included 262 (74.2%) stage I and stage II, 91 (25.8%) stage III and stage IV. The topography of patients included 280 (74.9%) T1 and T2, 94 (25.1%) T3 and T4. A total of 4 of 261 (1.5%) cases had lymph node metastasis and 4 of 276 (1.4%) cases had distant metastases, 111 of 321 (34.6%) cases had vascular invasion. 223(91.0%) patients were found in Child-pugh A, 22 (9.0%) patients were found in Child-pugh B and C. The distribution of AFP values included 43.6% (n=152) AFP<20, 19.2% (n=67) 20≤AFP<400, 37.2% (n=130) AFP≥400. A total of 8 of 352 (2.3%) patients underwent radiation therapy. After operation, 28 of 353 (7.9%) patients had undergone ablation embolization, 18 of 348 (5.2%) patients had residual tumor, and 98 of 272 (36.0%) had new tumor event.
3.2 E2F2 had a high expression in HCC
In the present study, we applied Wilcoxon rank sum test to compare the expression of E2F2 in 374 HCC tissues and 50 normal tissues. As shown in Fig. 1A, the expression of E2F2 was prominently upregulated in HCC (p=3.428e−25). Besides, we investigated E2F2 expression in 50 HCC tissues and 50 adjacent normal tissues via Wilcoxon signed-rank tests, E2F2 showed significantly higher expression in HCC tissues (p = 1.32e−14) (Fig. 1B). Further, in order to validate the above results, we downloaded RNA-seq and miroRNA raw data from GEO database respectively, namely GSE124535 and GSE54236. The results also showed that E2F2 was prominently overexpressed in HCC samples (Fig.1C-D).
The protein expression level of E2F2 was analyzed using IHC samples from the HPA online database. The protein level of E2F2 was upregulated in HCC tissues in comparison with normal tissues (Fig.1 E-G), indicating that the mRNA and protein expression of E2F2 were similar in different database.
3.3 E2F2-associated PPI network
A E2F2-associated PPI network was constructed based on the STRING database. As it shown in Fig.2G, several genes had close association with E2F2, such as CDK2, CDK4, CDK6, RB1, RBL1, CCNA2 and so on.
3.4 The effects of overexpressed E2F2 on clinicopathological characteristics
As we can see from Fig. 2 (A–F), high level of E2F2 was significantly correlated with histologic grade (p=1.605e−07), clinical stage (p=0.022), T (T1/T2 vs. T3/T4, p=0.011), tumor status (p = 0.006), AFP (p=1.726e−07), family history (p=0.041).
The expression of elevated E2F2 in HCC samples was prominently correlated with histologic grade (OR =2.62 for G3-4 vs. G1-2, p=1.80E-05), clinical stage (OR =1.74 for III-IV vs. I-II, p=0.03), T (OR =1.64 for T3-4 vs.T1-2, p=0.04), tumor status (OR =1.88 for with tumor vs. tumor free, p= 3.79E-03), AFP (OR =3.18 for AFP≥400 vs AFP<20, p= 2.16E-04; OR=2.50 for 20≤AFP<400 vs AFP<20, p=2.56E-03). Univariate analysis using logistic regression uncovered that increased E2F2 expression (based on median expression value) was correlated with poor prognostic clinicopathologic characteristics (Table 2). This uncovered that the HCC patients with high E2F2 are tend to progress to a more advanced stage.
3.4 Survival outcomes
3.4.1 Survival outcomes based on Kaplan-Meier survival analysis
Kaplan-Meier survival analysis showed that increased E2F2 was significantly associated with poor OS (p=7.468e−05), PFI (p=3.183e−05), DFI (p=0.001), DSS (p=4.172e−05), which indicated that HCC patients with high-E2F2 may have a worse prognosis than that with low-E2F2 (Fig.3A-3D).
3.4.2 OS, DFI and PFIoutcomes using univariate and multivariate analysis with Cox regression survival model
Univariate and multivariate analysis with the Cox regression model was employed to uncover the association between clinicopathologic characteristics and HCC patient survival.
At univariate Cox analysis, shorter overall survival (OS) was found in high expression of E2F2 (P =0.002, HR= 2.0 (95% CI [1.3-3.2])), poorer TNM (T: P = 0.033, HR =1.4 (95% CI [1.0-2.0])), worse clinical stage (P =0.012, HR = 1.5 (95% CI [1.1-2.1])) (Table 3). However, worse OS was only significantly associated with high expression of E2F2 in multivariate analysis, with a HR of 2.4 (P = 0.004, 95% CI [1.3-4.2]) (Table 3, figure 4).
At univariate Cox analysis, poorer disease-free interval (DFI) was prominently correlated with high expression of E2F2 (P =0.004, hazard ratio [HR] = 2.2 (95% CI [1.3-3.7])), higher TNM (T: P = 0.001, HR =1.6 (95% CI [1.2-2.1])), advanced clinical stage(P =0.000, HR= 1.7 (95% CI [1.3-2.3])) and postoperative ablation embolization (P =0.001, HR= 3.1 (95% CI [1.6-6.0])) (Table 4). At multivariate analysis, E2F2 (P =0.029, hazard ratio [HR] = 2.0 (95% CI [1.1-3.7])) and postoperative ablation embolization (P =0.000, hazard ratio [HR] = 4.3 (95% CI [2.0-9.4])) were the clinicopathologic characteristics that remained significantly correlated with DFI (Table 4, figure 5).
The univariate Cox analysis indicated that highly expressed E2F2 was significantly associated with worse PFI (P =0.005, hazard ratio [HR] = 1.8 (95% CI [1.2-2.6])). Other clinicopathological characteristics such as higher TNM (T: P = 0.001, HR =1.5 (95% CI [1.2-2.0])), advanced clinical stage(P =0.000, HR= 1.6 (95% CI [1.2-2.1])) and postoperative ablation embolization (P =0.001, HR= 2.8 (95% CI [1.5-5.2])) were also associated with poor survival (Table 5). The multivariate Cox analysis showed that E2F2 (P =0.005, hazard ratio [HR] = 2.2 (95% CI [1.3-3.9])) and postoperative ablation embolization (P =0.001, hazard ratio [HR] = 3.7 (95% CI [1.8-7.9])) were the clinicopathologic characteristics that remained significantly correlated with PFI (Table 5, figure 6).
3.5 Identification of E2F2 related signaling pathway by GSEA
Gene Set Enrichment Analysis (GSEA) was applied to extract prominently activated signaling pathways between low and high E2F2 expression data sets, and validated their significant differences (FDR < 0.05, NOM P-value < 0.05) in enrichment of MSigDB Collection (c2.cp.v6.2.symbols.gmt). 10 signaling pathways with significant differences, including the cell circle, the RNA degradation, the pyrimidine metabolism, the base excision repair, the aminoacyl tRNA biosynthesis, the DNA replication, the p53 signaling pathway, the nucleotide excision repair, the ubiquitin mediated proteolysis, the citrate cycle TCA cycle were filtered out, which were prominently enriched in E2F2 high expression phenotype based on NES, NOM P value, and FDR value (Fig. 4A–4B, Fig.5; Table 5).