RAD50 predicts the prognosis of hepatitis B virus-related hepatocellular carcinoma

RAD50, which is involved in the DNA double-strand break (DSB) repair process, is also involved in cancer outcomes. However, its role in hepatitis B virus (HBV)-associated hepatocellular carcinoma (HCC) remains unclear. RAD50 increased in HCC tissues compared to normal tissues and was correlated with OS in the TCGA cohort. The validation analysis indicated that increased RAD50 was expressed in HCC tissues in the two independent cohorts, 107 and 100 patients from the Affiliated Hospital of Youjiang Medical University of Nationalities (AHYMUN) and the Affiliated Hospital of Nantong University (AHNTU), respectively. In addition, HCC patients with elevated RAD50 expression showed poor OS and DFS in the AHYMUN cohort and decreased OS and DFS in the AHNTU cohort. Furthermore, four datasets obtained from the Oncomine database validated the analysis of the differential expression of RAD50 in HCC tumors and normal tissues. In conclusion, our study reveals that elevated RAD50 expression is correlated with cancer progression and poor survival in HBV-related HCC patients. These data suggest that RAD50 may act as an oncogene and may serve as a promising therapeutic target for HCC patients.


Introduction
In China, the incidence and mortality of primary hepatocellular carcinoma (HCC) is the second highest of all malignant cancers (1). In addition, in China, because the incidence of hepatitis B virus (HBV) infection is high, HBV accounts for at least 80% of HCC cases. In the Guangxi Zhuang Autonomous Region (2) and Qidong in Nantong Prefecture (3), HBV infection is very common, leading to a high incidence of HCC (4, 5). Although alphafetoprotein (AFP) and other markers have been widely used in clinical practice, the accuracy and sensitivity of diagnosis and targeted therapy need to be improved. Therefore, to reduce the cost of treatment for patients with HCC and to improve their quality of life, it is important to understand the underlying molecular mechanisms and apply them to the development of targeted treatments.
The DNA repair component protein Rad50 (RAD50) is a component of the MRN complex, which is a heterotrimer composed of RAD50, meiotic recombination 11 (MRE11), and Nijmegen fracture syndrome 1 (NBS1), that participates in the DNA damage response (DDR). Increased RAD50 expression is associated with multiple tumours, including breast cancer, ovarian cancer, lung cancer and rectal cancer (6-9). For example, RAD50 increases in colorectal cancer patients have been shown to be related to tumour development and prognosis (10). It is speculated that the high expression of RAD50 leads to cell instability, which increases the tendency toward DNA damage accumulation and malignant transformation. Following HBV infection, RAD50 expression in hepatocytes has been shown to be increased (11).
To investigate the expression and prognostic value of RAD50 in HCC, we recruited 207 HCC patients from two centres with confirmed pathological evidence of HCC. The prognostic role of RAD50 in HCC patients was validated and functionally annotated in silico. We speculate that RAD50 is associated with poor prognosis and may serve as a potential therapeutic target for HCC.

Materials And Methods
RAD50 expression profiles in hepatocellular tumour and normal tissue were analysed and displayed using the Oncomine online database (http://www.oncomine.com). A dataset including the known gene expression patterns in human HCC were included to validate the mRNA expression of RAD50 in the two groups. An unpaired t test was used to analyse the significant differences.

Data processing of Gene Set Enrichment Analysis (GSEA)
The Search Tool for the Retrieval of Interacting Genes (STRING; http://string-db.org) (version 10.0) online database was used to predict the PPI network for RAD50. Any interaction with a combined score >0.4 was considered statistically significant. The significant genes were included in functional and pathway enrichment analyses in a bubble chart using DAVID bioinformatics resources, which indicated the most significant signalling pathways according to GO and KEGG analysis. Datasets from the Cancer Genome Atlas (TCGA) database were analysed with the GSEA method using the category version 2.10.1 package in R. For each separate analysis, the Student's t-test statistical score was obtained for each of the consistent pathways, and the mean differential expression of each gene was calculated. A permutation test was performed 1000 times to identify the pathways with significant changes. The adjusted P values (adj. P), which were obtained by default using the Benjamini and Hochberg (BH) false discovery rate (FDR) method, were utilized to correct for the occurrence of false positive results. Significantly related genes were defined by an adj. P less than 0.01 and an FDR less than 0.25.

Results
In this study, the research was conducted in three steps. In the first step, the differential expression of RAD50 in normal and prostate tissues was analysed; in the second step, RAD50 expression in HCC patients was assessed as an indicator of progression and prognosis; in the third step, the differential expression of RAD50 was validated in the Oncomine datasets, and GSEA was used to identify the relationships between significant genes and the involved signalling pathways. Table 1, in the AHYMUM cohort, increased RAD50 expression was significantly correlated with decreased age (p=0.021), advanced AJCC stage (p=0.001), the presence of multiple primary tumour lesions (p=0.001), the presence of microvascular invasion (p=0.028), the presence of capsular invasion (p=0.015), and increased Okuda score (p=0.009). In the AHNTU cohort, increased RAD50 expression was significantly associated with larger tumour size (p<0.001), advanced AJCC stage (p=0.003), the presence of multiple primary tumour lesions (p=0.010), the degree of pathological nuclear differential expression (p=0.005), the presence of microvascular invasion (p=0.001), the presence of capsular invasion (p=0.003), and higher Okuda score (p=0.008). A chi-square test showed that the baseline data were balanced in terms of the distribution of the categorical data, including gender, albumin level, bilirubin level and the presence of liver cirrhosis (p>0.05).

Expression of RAD50 in the AHYMUN and AHNTU cohorts
To analyse the RAD50 expression profile in HCC tissue, IHC was performed to reveal the staining distribution in tumour and normal tissues ( Fig. 2A). The scatter plot of the IHC scores revealed that RAD50 expression was significantly elevated in HCC tissues in the AHYMUM (p<0.001) and AHNTU (p<0.001) cohorts (Fig. 2B).

Cox regression analyses
The univariate Cox regression analysis, as shown in the forest plots, revealed that age, gender, albumin level, bilirubin level and liver cirrhosis were not significantly associated with DFS in both AHYMUN (Fig. 3A) and AHNTU (Fig. 3B) patients. Meanwhile, tumour size was not a significant independent predictor of DFS in the AHYMUN cohort. In the multivariate models, the Okuda score (ref. I) was significantly associated with DFS in HCC patients, indicating the good representativeness of the population in the AHYMUN (HR: 1.914, p=0.045) and AHNTU (HR: 3.084, p=0.001) cohorts in the current study. More importantly, the subgroup analysis of RAD50 expression showed that RAD50 amplification was significantly correlated with DFS in the AHYMUN (HR: 4.612, p<0.001) and AHNTU (HR: 2.660, p=0.006) cohorts. Additionally, in the AHNTU cohort, a number of primary lesions was significantly correlated with poor DFS (HR: 2.563; p=0.007) according to the multivariate model used for the Cox regression analyses ( Table 2).
In Fig. 3C & D, several parameters, including age, gender, albumin level, bilirubin level and the presence of liver cirrhosis, were not significantly associated with OS in the two cohorts. As shown in Table 3 The survival curves suggested that HCC patients in the AHYMUN cohort with elevated RAD50 expression levels showed poor OS (p<0.001) and poor DFS (p<0.001) (Fig. 4A &   B). In addition, in the AHNTU cohort, increased RAD50 expression was significantly associated with decreased OS (p=0.021) and DFS (p<0.001) (Fig. 4C & D).

Discussion
In our study, we demonstrated that RAD50 was positively correlated with poor prognosis in HCC patients in the TCGA cohort. In the two validated cohorts, increased RAD50 expression was observed in cancer tissues compared to that in para-cancer tissues, which was shown be associated with decreased OS and DFS. Four validation cohorts from the Oncomine database also showed elevated RAD50 expression in HCC tissues compared with that in adjacent normal tissues. Additionally, the functional annotation bioinformatics analysis identified significant related hub genes and signalling pathways.
According to the IntAct molecular interaction database, human RAD50 potentially interacts with the transcription factor c-Jun/activator protein 1 (AP-1) (website: https://www.ebi.ac.uk/intact/interactors/id:Q92878*). C-Jun induces HBV-related liver tumourigenesis in mice via upregulating the transcription of its target gene osteopontin (OPN) (15). OPN is overexpressed in several human carcinomas and contributes to inflammation, tumour progression, and metastasis (16,17). Therefore, we speculated that the interaction of RAD50 with c-Jun activated the transcription activity of c-Jun, upregulated OPN expression, facilitated the development of HBV-related HCC, and contributed to poor prognosis.
Our study suggested that increased RAD50 expression in HBV-related HCC is a marker of poor prognosis. Aude Dupré and colleagues designed a chemical forward genetic screen to identify a small-molecule inhibitor of the MRN complex, named Z-5-(4hydroxybenzylidene)-2-imino-1,3-thiazolidin-4-one (mirin) (18). In 2009, other researchers confirmed the structure of mirin (19). However, there have been no reports about the use of mirin in cellular, animal and clinical applications. We questioned whether mirin could be used in an HBV-related HCC animal model to determine the therapeutic effect of the inhibition of the MRN complex.
The predictive power of RAD50 may be as strong as that of the classical factors commonly used in clinical practice. Serum tumour markers are the most commonly used markers for the monitoring and early diagnosis of HBV-related HCC because they are non-invasive, objective, and reproducible. The most common serological test utilizes AFP.
Disappointingly, in a retrospective case-control study, the sensitivity of AFP testing at the most effective cut-off value (10-20 ng/ml) in the diagnosis of HCC was approximately 60%, and its specificity was 80% (20,21). These values are much lower than those required for the treatment and prediction of prognosis of HCC patients. However, RAD50 was shown to be a significant and independent biomarker (22). This means that it may be used to shed further light on the mechanisms underlying HCC progression, to facilitate the discovery of new therapies, and to open up new avenues for the personalized treatment of HCC patients.
In this study, the analysis of the data form the two cohorts supported our hypothesis and clearly demonstrated the high expression of RAD50 in tumour tissues from HCC patients, which results in increases in the HCC recurrence rate and poor overall survival. At the same time, GSEA indicated that RAD50 is involved in some of the most important pathways, including the mitotic spindle, UV response and transforming growth factor beta (TGF-β) signalling pathways, that are enriched in HCC specimens. In addition to participating in the DSB repair pathway, RAD50 also interacts with MMAP, which is expressed by the mitotic-specific MRN complex to maintain optimal genomic stability during mitosis (23).
The strength of our study is that we first attempted to study the role of RAD50 as a prognostic factor for HCC. The relationship between RAD50 and HCC has rarely been investigated (24). However, it is worth noting that RAD50 is a repair factor that has been identified in the DSB reaction and is highly expressed in many cancers. With this in mind, to further explain the role of RAD50 in the invasion and metastasis of HCC, we used GSEA to analyse data from public databases to identify important genes and pathways that may be involved in the relevance of RAD50 to the initiation of carcinogenesis.
The limitations of this study are as follows. First, the retrospective nature of the data set is obvious. Although data was obtained from two institutes at the Nantong University Affiliated Hospital and the Youjiang Nationalities Medical College Affiliated Hospital, the sample size is relatively small, and the patients in the two Chinese centres exhibit economic and geographical differences, which may lead differences in therapeutic strategies and overall prognoses. Second, while our research involved a thorough functional annotation of RAD50 in HCC, we failed to confirm the underlying mechanisms.
In the future, we will conduct multi-center research in Europe and the United States to further explore the role of RAD50 in the development and prognosis of HCC.

Conclusion
In conclusion, our study is the first to reveal that elevated RAD50 expression is significantly correlated with cancer progression and poor survival in HBV-related HCC patients. These data suggest that RAD50, which is a potential therapeutic target, may act as an oncoprotein and could serve as a promising prognostic marker in HCC patients. In this regard, randomized clinical trials and further investigation are required to identify the true value of RAD50 in terms of its clinical application in HCC patients.

Consent for publication: Not applicable.
Availability of data and materials: The datasets and/or analysis results used during the current study are available from the corresponding author upon reasonable request.

Competing interests:
The authors declare no competing interests.            Datasets from the TCGA database were analysed with the GSEA method. For each separate analysis, a Student's t-test was performed to obtained the statistical scores from the consistent pathways, and the means of the differential expression of genes were calculated. A permutation test was performed 1000 times to identify the significantly changed pathways. The adjusted P values (adj. P) were generated using the Benjamini and Hochberg (BH) false discovery rate (FDR) method by default and were used to correct for the occurrence of false positive results. Significantly related genes were defined by an adj. P less than 0.01 and an FDR less than 0.25. The most significant highly involved pathways included the nuclear body, ubiquitin ligase complex, G2M checkpoint and mitotic spindle pathways.