Introduction

Liver cancer is a key cause of cancer-related mortality, accounting for 8.2% of total cancer deaths1. Most liver cancer patients are diagnosed at intermediate and late stages and cannot receive curative therapies, which contributes to the low survival rates and high recurrence rates of patients. With surveillance of identifiable high-risk patients and surgical intervention for early-stage patients, the 5-year survival rate has increased to 26%2. However, the recurrence rate of HCC is still over 70% at 5 years even after resection3. Therefore, it is essential to find an early screening strategy for diagnosis and prognosis prediction in liver cancer. Unfortunately, current screening markers are unsatisfactory due to their poor sensitivity and specificity. Reliable biomarkers for to improve diagnosis and to predict prognosis after treatment are urgently required.

SLC25A11 is also known as oxoglutarate carrier (OGC), which regulates electroneutral exchange between 2-oxoglutarate and some dicarboxylates4. It transports glutathione (GSH) from the cytoplasm to the mitochondrial matrix together with a dicarboxylate carrier (DIC; SLC25A10)5. Because of the importance of the GSH system in mitochondria, the regulation of the mitochondrial glutathione (mtGSH) pool can influence the metabolic disorders caused by mitochondrial dysfunction, such as diabetic nephropathy, aging and cancer6,7. Moreover, as a member of the malate-aspartate shuttle (MAS) family, SLC25A11 is essential for ATP generation by NADH transportation from the cytoplasm to mitochondria8. Therefore, these studies suggest that SLC25A11 is a key factor in the transportation of NADH and GSH from the cytoplasm into mitochondria in cancer.

Recent studies suggest that SLC25A11 plays a role in the formation of non-small-cell lung cancer (NSCLC)8. In addition, it promotes liver cancer by maintaining mtGSH under cholesterol loading9. However, the prognostic significance of SLC25A11 expression in liver cancer has not yet been determined. The diagnostic value of SLC25A11 has been evaluated by exploring the correlations between SLC25A11 expression and clinical features from the TCGA. Our results demonstrated that SLC25A11 could be a reliable marker for diagnosis and a predictor of prognosis in liver cancer.

Result

Characteristics of patients in this study

Gene expression data and the clinical data were collected from the TCGA and organized. Clinical characteristics included age, sex, TNM classification, residual tumor status, radiation therapy, overall survival and relapse-free survival (Table 1).

Table 1 Relationship between the clinical features of liver cancer and SLC25A11 expression.
Full size table

Low SLC25A11 expression in liver cancer

The analysis of SLC25A11 was conducted to compare the difference in SLC25A11 expression between primary HCC tissues and normal liver tissue via box plots (Fig. 1). The results showed that SLC25A11 mRNA was downregulated in liver cancer (P = 0.0019), which was also verified by the GSE54236 and HCCDB databases (Fig. 1). In addition, we compared SLC25A11 mRNA expression among different clinical characteristics, including stage, histologic grade, TNM classification, histological type, sample type and vital status, and found that SLC25A11 mRNA expression was significantly different when patients were grouped by stage (P = 0.044) and histologic grade (P = 0.096).

Figure 1
Figure 1
Full size image

SLC25A11 expression in liver cancer. SLC25A11 expression in liver cancer tissues was compared with that in normal tissues. Subgroup analysis of SLC25A11 according to stage, histologic grade, TNM classification, histological type, sample type, and vital status. Verification in the GSE54236 and HCCDB databases.

The diagnostic value of SLC25A11

The diagnostic value of SLC25A11 was evaluated by ROC curve analysis using data from the TCGA. The area below the ROC curve (AUC) was 0.635, suggesting the modest diagnostic value of SLC25A11 (Fig. 2). Subgroup analysis revealed the diagnostic value in different stages of liver cancer with AUC values of 0.593, 0.668, 0.696, and 0.588 for stage I, stage II, stage III, and stage IV, respectively (Fig. 2).

Figure 2
Figure 2
Full size image

The diagnostic value of SLC25A11 mRNA expression in liver cancer. (A) The ROC curve of SLC25A11 expression in liver cancer tissues was compared with that of normal tissues. (B–E) Subgroup analysis for stage I, II, III, and IV liver cancer.

Correlations between SLC25A11 and clinical features with liver cancer

After dividing the patients into low/high SLC25A11 mRNA groups according to the threshold value determined by the ROC curve, the relationships between clinical features and SLC25A11 expression were analyzed (Table 1). Histologic grade (P = 0.0244), clinical stage (P = 0.0428) and vital status (P = 0.0126) were significantly correlated with SLC25A11 expression.

Low SLC25A11 predicts poor prognosis of OS and RFS

Kaplan-Meier survival curves were constructed to explore the prognostic value of SLC25A11 in OS. The results showed that low expression of SLC25A11 was related to worse OS (P = 0.0088; Fig. 3). In addition, a survival curve was constructed to test the prognostic value of SLC25A11 in RFS. The results showed that low SLC25A11 expression was associated with poorer RFS than high SLC25A11 expression (P = 0.002; Fig. 3), which was also confirmed by using the ICGC and GSE54236 databases (Fig. 3).

Figure 3
Figure 3
Full size image

Kaplan-Meier curves for OS and RFS according to SLC25A11 expression in liver cancer. Survival analysis of OS and RFS was performed based on Kaplan-Meier curves in TCGA database, GSE54236 and HCCDB database.

Subgroup analysis revealed the prognostic value of SLC25A11 in terms of OS

Subgroup analysis revealed that the low expression of SLC25A11 was correlated with poor OS in grade G1 and G2 patients (P = 0.0061), older patients (P = 0.026), and male patients (P = 0.042) (Fig. 4).

Figure 4
Figure 4
Full size image

Kaplan-Meier curves for OS according to SLC25A11 expression in liver cancer. Subgroup analysis of OS was performed based on Kaplan-Meier curves according to clinical stage, histological grade, sex and age.

Subgroup analysis revealed the prognostic value of SLC25A11 in terms of RFS

Subgroup analysis revealed that low expression of SLC25A11 was correlated with poor OS in stage I and II patients (P = 0.0091), G1 and G2 patients (P = 0.027), G3 and G4 patients (P = 0.021), male patients (P = 0.0011) and older patients (P = 0.0021) (Fig. 5).

Figure 5
Figure 5
Full size image

Kaplan-Meier curves for RFS according to SLC25A11 expression in liver cancer. Subgroup analysis of RFS was performed based on Kaplan-Meier curves according to clinical stage, histological grade, sex and age.

Low SLC25A11 is an independent predictor of OS

Univariate Cox analysis identified the potential OS-related variables, including the stage, T classification, residual tumor status and SLC25A11. Multivariate Cox analyses showed that SLC25A11 expression (HR value: 1.526, 95% CI: 1.072–2.172, P = 0.019), residual tumor status (HR value: 1.410, 95% CI: 1.103–1.801, P = 0.006), and T classification (HR value: 1.835, 95% CI: 1.458–2.310, P < 0.001) were independent predictors of poor OS (Fig. 6).

Figure 6
Figure 6
Full size image

Cox model of OS among liver cancer patients. Univariate and multivariate analysis of the correlation of SLC25A11 expression with OS among liver cancer patients.

Low SLC25A11 is an independent predictor of RFS

Univariate Cox analysis identified the potential RFS-related variables, including the stage, T classification, residual tumor status and SLC25A11. Multivariate Cox analyses showed that SLC25A11 expression (HR value: 1.604, 95% CI: 1.145–2.246, P = 0.006), residual tumor status (HR value: 1.300, 95% CI: 1.021–1.655, P = 0.034), and T classification (HR value: 1.671, 95% CI: 1.287–2.169, P < 0.001) were independent predictors of poor RFS (Fig. 7).

Figure 7
Figure 7
Full size image

Cox model of RFS among liver cancer patients. Univariate and multivariate analysis of the correlation of SLC25A11 expression with RFS among liver cancer patients.

Discussion

Among cancers worldwide, liver cancer has one of the worst prognoses. Many studies have investigated novel prognostic biomarkers in different types of cancer10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26. In this study, we demonstrated that SLC25A11 was low expressed in liver cancer by analyzing data from the TCGA. Low expression of SLC25A11 is related to poor prognosis. Both univariate and multivariate Cox analyses revealed that SLC25A11 played an important role in OS and RFS, which means that SLC25A11 is a novel independent prognostic factor in liver cancer.

In recent years, high SLC25A11 has been found in NSCLC and melanoma8. However, we found that SLC25A11 was expressed at low levels in liver cancer. ROC analysis suggested that SLC25A11 could be regarded as a marker in the diagnosis and prognosis of liver cancer. Moreover, we observed that histologic grade, clinical stage, and survival status were related to SLC25A11 expression, which confirmed the significance of this molecule.

SLC25A11, a member of the malate-aspartate shuttle (MAS), regulates the shuttling of 2-oxoglutarate by dicarboxylates from different tissues and cells using various methods27,28,29. These studies also provide evidence that SLC25A11 is a GSH carrier. In addition, SLC25A11 induction is a key strategy for maintaining mtGSH levels to limit ROS production in HCC9. Based on the findings that SLC25A11 expression is associated with clinical stage and histologic grade in liver cancer, we postulate that by stabilizing the mitochondrial membrane and maintaining sufficient mtGSH to withstand hypoxia-induced ROS production and cell apoptosis, SLC25A11 improves cell growth and tumor progression.

As one of the characteristics of cancer, fast growing cancer cells requires more ATP than normal cells. Because cancer cells rely mainly on glycolysis for energy, they maintain high mitochondrial oxidative phosphorylation for ATP generation and survival30. SLC25A11 has been proven to be a crucial transporter of NADH from the cytoplasm to mitochondria in the ATP production process, especially in cancer cells. Therefore, the association of the expression of SLC25A11 with poor survival may be due to the effect of SLC25A11 on tumor energy metabolism. This implies that SLC25A11 is necessary for embryogenesis but not for the proliferation of differentiated cells, which suggests that it can be a good marker for cancer.

To our knowledge, our research is the first to show the significant role of SLC25A11 in the prognosis of liver cancer by mining the TCGA database. We found that SLC25A11 could be regarded as a prognostic factor for poor survival in liver cancer. However, the total number of patients evaluated is a limitation of our research. There was no significant correlation between SLC25A11 expression and overall survival in ICGC and the GSE54236 database in Fig. 3. However, the statistical significance does not represent clinical significance because some variables could influence the P-value, such as sample size. For this study, we observed that low SLC25A11 expression was associated with poor prognosis in the ICGC and GSE54236 cohorts, but the P-value did not indicate significance, perhaps because of the sample size. We need to expand the sample size to confirm the prognostic value of SLC25A11 expression levels in the future. Furthermore, we postulate that SLC25A11 plays an essential role in transporting NADH and GSH from the cytoplasm into mitochondria, providing energy for liver cancer cells to withstand harsh environments. However, clinical trials and further experiments are needed to certify these results and clarify the specific molecular mechanism.

Materials and Methods

Data collection and mining

Clinical information and RNA-sequencing expression results were collected from the TCGA dataset, GEO dataset (https://www.ncbi.nlm.nih.gov/gds/), HCCDB dataset (http://lifeome.net/database/hccdb/) and ICGC dataset (https://icgc.org/). All these data were processed in R software (version 3.6.1) (https://www.r-project.org/) and related packages31.

Statistical analysis

Box plots were used to evaluate the expression of SLC25A11 in the TCGA-Liver Hepatocellular Carcinoma (LIHC) dataset. A receiver operating characteristic curve (ROC) was drawn to assess the diagnostic significance of SLC25A11 expression by using the pROC package32. In accordance with the threshold identified by the ROC curve, liver cancer patients were separated into low and high expression groups. OS and RFS were compared between low and high groups via Kaplan-Meier analysis with the package in R33,34. Fisher’s exact and chi-square tests were utilized to evaluate correlations between the expression of SLC25A11 mRNA and clinical information. The potential prognostic factors were selected by Cox analysis. Correlations between SLC25A11 expression and survival and other clinical characteristics of patients were confirmed by using multifactor Cox analysis. *p < 0.05 was considered significant.