Circular noncoding RNA hsa_circ_0005986 as a prognostic biomarker for hepatocellular carcinoma

Circular RNAs (circRNAs) represent potential biomarkers because of their highly stable structure and robust expression pattern in clinical samples. The aim of this study was to evaluate the expression of a recently identified circRNA, hsa_circ_0005986; determine its clinical significance; and evaluate its potential as a biomarker of hepatocellular carcinoma (HCC). We evaluated hsa_circ_0005986 expression in 123 HCC tissue samples, its clinical significance, and its association with patients’ clinicopathological characteristics and survival. Hsa_circ_0005986 expression was downregulated in HCC tissues. Low hsa_circ_0005986 expression was more common in tumors larger than 5 cm [odds ratio (OR), 3.19; 95% confidence interval (CI), 1.51–6.76; p = 0.002], advanced TNM stage (III/IV; OR, 2.39; 95% CI, 1.16–4.95; p = 0.018), and higher BCLC stage (B/C; OR, 2.71; 95% CI, 1.30–5.65; p = 0.007). High hsa_circ_0005986 expression was associated with improved survival and was an independent prognostic factor for overall [hazard ratio (HR), 0.572; 95% CI, 0.339–0.966; p = 0.037] and progression-free (HR, 0.573; 95% CI, 0.362–0.906; p = 0.017) survival. Moreover, the circRNA–miRNA–mRNA network was constructed using RNA-seq/miRNA-seq data and clinical information from TCGA-LIHC dataset. Our findings indicate a promising role for hsa_circ_0005986 as a prognostic biomarker in patients with HCC.


Materials and methods
Patients and tissue samples. This study included 162 patients with HCC ( Fig. 1) who underwent diagnostic biopsy or surgical resection at Kyungpook National University Hospital, Republic of Korea, between March 2015 and August 2016. Thirty patients who had been previously treated for HCC and nine patients who were lost to follow-up were excluded from the study, resulting in a final sample size of 123 evaluable patients. Tissue samples were obtained by liver biopsy or surgical resection. Liver biopsy was performed to confirm HCC diagnosis and to rule out the presence of other tumors. Patients underwent surgical resection (n = 19) or radiofrequency ablation (n = 47) as curative treatment (n = 66, 53.7%) or transarterial chemoembolization (n = 9), sorafenib (n = 12) or best supportive care (n = 36) as non-curative treatment. This study was conducted according to local ethical guidelines, in accordance with the Declaration of Helsinki.
For post-treatment monitoring, imaging was conducted every 3-6 months using contrast-enhanced dynamic computed tomography (CT) or gadoxetic acid disodium-enhanced liver magnetic resonance imaging (MRI). We defined overall survival as the time between the date of initial HCC diagnosis and either the date of death from any cause or the date of last contact with the patient during follow-up examination. Progression-free survival was defined as the time between the initial date of HCC diagnosis and either the first event of recurrence or progression or until death from any cause. The recurrence of HCC was recognized if a tumor exceeded 1 cm and showed characteristic CT or MRI contrast enhancement in the arterial phase and washout in the venous or delayed phase. Response Evaluation Criteria in Solid Tumors (version 1.1) was used to evaluate tumor response. HCC specimens and adjacent non-tumor tissue specimens were immediately stored at 4 °C for 24 h in RNAlater reagent (Ambion; Life Technologies, Carlsbad, CA, USA) and then stored at − 80 °C. We recorded the patients'

Quantitative real-time polymerase chain reaction (qRT-PCR). We used SYBR Green PCR Master
Mix (Applied Biosystems) to perform qRT-PCR. The expression of hsa_circ_0005986 was normalized to that of glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and then quantified using the 2 −ΔΔCt method. All primers were synthesized by Bionics (Seoul, Korea). In order to amplify only hsa_circ_0005986, but not linear form of RNA, the primers were designed by considering the backsplice junctions of circRNA ( Supplementary Fig. 1).
Statistical analysis. For descriptive statistics, categorical data were expressed as number (%) and numerical data as the mean and standard deviation for normally distributed data and as the median with interquartile range for non-normally distributed data. A paired t-test was used to analyse differences in hsa_circ_0005986 expression between HCC and adjacent non-tumor tissue. We used the chi-square or Fisher's exact probability test to compare clinicopathological characteristics between two groups with different hsa_circ_0005986 expressions. The Kaplan-Meier method was used to generate survival curves, and the log-rank test was conducted to compare survival curves between groups. The prognostic performances 30 of hsa_circ_0005986 were expressed as specificity, sensitivity, and area under the receiver operating curve (AUC). To determine the predictors of survivals, univariate and multivariate analyses based on a Cox proportional hazards model were performed. p-values of < 0.05 were considered statistically significant. Values that were statistically significant in univariate analyses were included in multivariate analyses, with a p-value of < 0.1. We conducted all the analyses using SAS version 9.4 software (SAS Institute Inc., Cary, NC, USA), and GraphPad Prism 6 program for Windows (GraphPad Software, La Jolla, CA, USA) was used to generate figures.
Construction of the circRNA-miRNA-mRNA network 31,32 . Publicly available sequencing data (RNA-seq, miRNA-seq and clinical information data) related to HCC were obtained from The Cancer Genome Atlas (TCGA) using gdc-rnaseq-tool (https:// github. com/ cprei d2/ gdc-rnaseq-tool). Differentially expressed genes and miRNA (|log2-fold change|≥ 1 and adjust p value < 0.05) were analyzed using the DEseq2 33 R package (version 1.30.1) for further analysis. A co-expression network was constructed using a WGCNA package 34 (version 1.70-3) in the R software (version 4.0.3). Potential target miRNAs of hsa_circ_0005986 were predicted via circBank 35 . The overlapping part of the target miRNAs from co-expressed miRNAs from WGCNA were selected. The potential target genes of the selected miRNA were predicted using mirDB 36 . The overlapping part of the

Results
Baseline characteristics of patients with HCC.   Downregulation of hsa_circ_0005986 expression in HCC tissues and advanced HCC. We found that the expression of hsa_circ_0005986 was significantly downregulated in HCC tissues compared with that in adjacent non-tumor tissues (Fig. 2). In addition, we found that the expression of hsa_circ_0005986 in TNM stages III and IV was significantly lower than that in stages I and II. Similarly, lower hsa_circ_0005986 expression was observed in BCLC stages B and C compared with that in stages 0 and A (Fig. 3). Correlation between hsa_circ_0005986 expression and progression-free survival in patients with HCC. Progression-free survival differed significantly between patients depending on the hsa_ circ_0005986 expression level (Fig. 4B). The cumulative 1-, 2-, and 3-year progression-free survival rates were 36.8%, 19.4%, and 14.8%, respectively, in the low-expression group and 61.8%, 43.1%, and 35.8%, respectively, in the high-expression group.
Prediction of hsa_circ_0005986 function. We obtained RNA-seq/miRNA-seq data and clinical information from the TCGA-LIHC dataset, including 425 RNA-seq and 425 miRNA-seq samples (50 normal samples and 374 tumors for RNA-seq; 375 tumors for miRNA-seq), and differentially expressed mRNAs and miRNAs were identified (9004 for mRNA and 310 for miRNA). To further investigate the target of hsa_circ_005986, circBank and WGCNA were performed. Two miRNAs were predicted as target of hsa_circ_005986. The target genes of hsa-mir-3677 and hsa-mir-188 were predicted using mirDB and WGCNA. Finally, we used 2 miRNAs and 52 genes to construct a circRNA-miRNA-mRNA network (Fig. 5A). Figure 5B shows the related biological processes, such as, cell adhesion (p = 0.0015), negative regulation of cell proliferation (p = 0.0044), skeletal system development (p = 0.0062), regulation of inflammatory response (p = 0.013), somatic stem cell maintenance (p = 0.014), and positive regulation of inflammatory response (p = 0.017), all of which were statistically significant.

Discussion
The prognosis of patients with HCC is poor 37 , in part because most HCC cases are diagnosed at an advanced stage, thereby limiting the treatment options available and resulting in poor overall prognosis 38 . HCC exhibits various clinical characteristics because of the diverse etiologies associated with the underlying liver disease, impaired hepatic function, and tumor biology, even between the same disease stage. Because the same HCC stage may result in different clinical outcomes, identifying factors that affect prognosis is of great importance. AFP is a widely used biomarker for HCC; however, it is not specific enough for screening and diagnosing HCC 39 . Limited progress has been made in identifying new candidate markers, such as des-gamma carboxyprothrombin or fucosylated AFP, which have exhibited low accuracy during clinical evaluation 40 . Therefore, discovering novel markers is imperative to facilitate timely and early HCC diagnosis and improve treatment success and survival of patients with HCC.
CircRNAs have recently been shown to be potential biomarkers for many cancers, including HCC [41][42][43] . CircR-NAs are promising biomarkers because of their highly stable structure and robust expression patterns in clinical samples. Owing to the covalently closed structure that prevents their cleavage and degradation by exonucleases, circRNAs are highly stable in blood 44 , saliva 45 , and exosomes 46 . As such, they show enormous potential as cancer biomarkers. A previous study showed that the median half-life of tested circRNAs was 2.5-fold longer than that of their linear counterparts from the same gene and showed that circRNAs were stably expressed, whereas mRNA and microRNA expression levels changed within minutes 47 . Recently, the potential of some circRNAs (circ_005075 and circ_0016788) as HCC diagnostic biomarkers has been suggested 48,49 . For example, circ-CDYL was upregulated in the early stages of HCC but showed a low AUC value, i.e., 0.64 50 , which was lower than those reported for circ_005075 and circ_0016788 (0.94 and 0.85, respectively) 48,49 . Similarly, the development of prognostic biomarkers for HCC has progressed, and some candidates have been identified. Upregulated circ_001569, circ_0008450, and circ_0000267 expressions were associated with poor HCC prognosis suggesting that these circRNAs are independent prognostic markers for HCC [51][52][53] .
By regulating cell proliferation, migration, invasion, apoptosis, metastasis, and EMT, circRNAs play either an oncogenic or a suppressive role in the progression of HCC 25,[54][55][56][57] . The majority of these processes are regulated by circRNAs via miRNA sponging. For instance, to facilitate tumorigenesis, hsa_circ_101280 serves as a sponge for miR-375 and upregulates JAK2 expression, thereby promoting the proliferation of HCC cells as well  54 . In addition, the oncogenic role of circSLC3A2 was shown to be dependent on the regulation of PPM1F expression by sponging miR-490-3p 56 . Another study involving HCC cells revealed higher circASAP1 expression in cells with higher metastatic potential. In addition, circASAP1 was shown to regulate the miR-326/miR-532-5p-MAPK1 pathway, thereby promoting the proliferation of tumor cells in HCC as well as metastasis 43 . Conversely, circMTO1 and hsa_circ_0001445 were found to promote the expression of    www.nature.com/scientificreports/ the tumor suppressor genes p21 25 and TIMP3 57 , respectively. These actions are based on the sponging of miR-9 25 , miR-17-3p, and miR-181b-5p 57 .
Our study revealed a clear relationship between the patients' characteristics and hsa_circ_0005986 expression. We showed that hsa_circ_0005986 exhibited reduced expression in HCC and demonstrated that it was associated with clinical and pathological characteristics of patients with HCC. To our knowledge, this is the first study to validate hsa_circ_0005986 as a prognostic biomarker in a HCC cohort by performing survival and regression analyses. In 2017, Fu et al. showed that hsa_circ_0005986 sponged miR-129-5p to regulate NOTCH1 expression in HCC. Downregulated hsa_circ_0005986 expression led to the liberation of miR-129-5p, leading to lower NOTCH1 expression. This was coupled with an accelerated G0/G1 to S phase transition to promote cell proliferation 58 . The authors also showed an association between hsa_circ_0005986 expression and patient data, which is consistent with our results, except for the correlation of family history with chronic hepatitis B. We did not find an association between hsa_circ_0005986 expression and chronic hepatitis B infection. Moreover, the previous study did not include survival or progression data from 81 patients with HCC. The study by Fu et al. focused on the mechanistic aspect of hsa_circ_0005986 to be considered as a biomarker. On the contrary, our study focused more on the validation of hsa_circ_0005986 as a prognostic biomarker, along with other clinical predictors, based on survival and progression data from a larger number of patients with HCC. Therefore, the difference between this study and the previous study is the validation of a potential prognostic biomarker, hsa_circ_0005986, in a larger HCC cohort, which might be the originality of our study.
We analyzed the whole genome mRNA-miRNA-hsa_circ_0005986 and its interaction network for predicting the role of hsa_circ_0005986 in HCC. On gene ontology analysis, cell adhesion, inhibition of cell proliferation, and skeletal system development were all possibly related to the migration, proliferation, and invasion of HCC. In our study, hsa_circ_0005986 is downregulated in HCC compared with the background liver and also in the higher stages of HCC. These findings were compatible with the potential function of hsa_circ_0005986 as an inhibitor of proliferation. Regulation of the inflammatory response, promotion of the inflammatory response, and somatic stem cell population maintenance might be related to HCC development. Moreover, for predicting the potential mechanism of hsa_circ_0005986, RNA modification or a microbiome should be discussed. Aside from circRNAs relevant to HCC progression and metastasis, RNA modification has also become a popular topic in cancer. Specifically, N4-acetylcytidine modification in other highly stable RNAs, such as circRNA, can be a possible mechanism of aberrant RNA modifications in HCC, which has rarely been studied in the current literature 59 . On the other hand, there might be a regulation network among immunodeficiency, a microbiome, and circRNA. Immunodeficiency can promote adaptive alterations of the host-gut microbiome and affect cancer development and progression 60 .
This study has some limitations. First, we did not examine hsa_circ_0005986 expression at a mechanistic level but rather evaluated the association between its expression and clinical endpoints. Further investigation of the mechanism of action of hsa_circ_0005986 is essential. Second, considering the retrospective nature of this study, there may have been some selection bias, considering that patients with missing medical records were not included. We excluded 39 patients who were either lost to follow-up or were previously treated for HCC, which reduced the total number of patients available for analysis. A larger number of patients are needed to validate the associations with hsa_circ_0005986 expression. It is necessary to improve the performance of hsa_circ_0005986 predicting prognoses, specifically for survival and progression. Third, percutaneous needle biopsy performed to obtain the specimens may not adequately reflect tumor heterogeneity. Some pathological features that affect the survival of patients with HCC cannot be assessed using needle biopsies (histological grade, microvascular invasion, and lymphatic invasion). This also reinforces the need to identify noninvasive biomarkers. Noninvasive diagnostic approaches such as serum or exosome collection are needed to validate whether hsa_circ_0005986 can be used as a prognostic biomarker in patients with HCC.

Conclusions
In conclusion, our results showed the association between hsa_circ_0005986 expression and HCC proliferation and progression. Considering that hsa_circ_0005986 was shown to be a predictor of HCC progression and survival of patients with HCC, we believe that it has potential to become both a prognostic biomarker and a therapeutic target. However, additional studies are needed to clarify the mechanisms underlying the causal role of hsa_circ_0005986 in HCC progression under the Mendelian Randomization framework through integrating multi-omics datasets [61][62][63] . In addition, it is important to develop effective individualized therapeutic strategies to help improve the outcomes of patients with HCC.

Data availability
The datasets used or analysed during the current study are available from the corresponding authors on reasonable request.