Urinary miR-93-5p as A Promising Biomarker for Early HBV-Related Hepatocellular Carcinoma


 Background: The mortality rate of early HBV-related hepatocellular carcinoma (HCC) is increasing annually due to sensitive and readily available diagnostic tools are lacking. This study aims to find such diagnostic biomarkers. Methods: MiR-93-5p was chosen as a candidate biomarker base on the analysis of relevant datasets of Gene Expression Omnibus (GEO). It was subjected to validation using qPCR for the quantification of its expression levels in tissue, plasma and saliva sample sets. Results: miR-93-5p was found to be significantly upregulated in HBV-related HCC tissue. Notably, miR-93-5p in plasma and urine was also significantly increased in patients with early HBV-related HCC. The expression of miR-93-5p was significantly and positively correlated between any two kinds of samples (tissue vs. plasma; tissue vs. urine, plasma vs urine). Moreover, miR-93-5p in plasma and urine reduced significantly over one month after the surgery and returned to normal levels. Finally, ROC analysis showed both plasmatic and urinary miR-39-5p could detect early, advanced and total HBV-relative HCC cases with over 85% of sensitivities and over 93% of specificities.Conclusion: plasma and urine miR-93-5p show great promise as novel and potential biomarkers for early HBV-related HCC.

demonstrate that aberrantly expressed miRNAs in tissue and blood can serves as biomarkers for diagnosing and monitoring various diseases. The kidneys have an extensive blood supply via the renal arteries. Urine is formed in the kidneys through a ltration of blood. The urine is then passed through urinary tract to the outside of the body. Hence, it is reasonably postulated that urinary miRNAs may have the potential to outplace circulating microRNA biomarkers in diagnosis diseases. Dozens of studies reveal that cancers of urinary tract, such as prostate, bladder etc may shed cancer cells into the urine.
Some reports have suggested that a few urinary miRNAs are promising biomarkers for cancers of urinary tract. Additionally, a few studies indicate that miRNAs in urinary supernatant that without effects of exfoliated cells and debris from the urinary tract and are also potential biomarkers for non-urinary diseases, including heart diseases [8], breast, endometrial, ovarian cancer [9], gastric cancer [10]. Due to the easy and non-invasive collection of urine samples, it is highly intriguing to identify more urinary biomarkers for early detection of lethal diseases, such as HCC.
As mentioned above, HBV-related HCC is still increasing in incidence and mortality. Thus far, to the best of our knowledge, there is no study with regard to nding miRNA biomarkers in urine to detect early HBVrelated HCC. This work was undertaken to identify robust biomarker in urine to help diagnose early HBVrelated HCC.

Methods
The selection of candidate microRNA biomarkers In order to seek candidate microRNA biomarkers from relative datasets regard to microarray-based microRNA pro les, and comparing the miRNA expression in HBV-related HCC tissue with normal liver tissue, relevant datasets from the Gene Expression Omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/geo) were collected and further assessed. The following criteria for data inclusion were used: (a) data were miRNA pro ling by microarray; (b) sample size in each study had n > 5; and (c) the study analyzed liver tissue pro les from patients with HBV-related HCC and controls. Two datasets, GSE10694 (including 78 HBV-related HCC cancer tissue, and 10 normal liver tissues) and GSE69580 (including 5 HBV-related HCC tissues and 5 adjacent non-tumor liver tissues), met the aforementioned criteria and were chosen for subsequent analysis. GEO2R tool (http://www.ncbi.nlm.nih.gov/geo/geo2r/) was used to identify aberrantly expressed miRNAs from the GEO datasets [11]. Candidate miRNA biomarkers were de ned as upregulated with P-values < 0.05 in both datasets. Finally, miR-93-5p was selected as a candidate biomarker for HBV-related HCC.

Participants and samples
The study was conducted mainly according to the updated Standards for Reporting of Diagnostic Accuracy (STARD) reporting guideline for diagnostic biomarker studies [12]. Samples from all participants were collected if they met the inclusion criterion mentioned below between July 2014 and July 2020 at The Fifth People's Hospital of Ganzhou and The First A liated Hospital of Gannan Medical University, Ganzhou, China. The 65 healthy controls were de ned as the participants with normal results by health examinations, including chest X-rays, routine blood and urine tests, abdominal ultrasounds, faecal occult-blood testing, blood cancer biomarker assays (AFP, CEA, CA19-9), HBV antigen, HCV, HIV, and syphilis antibodies. Normal human liver tissues were obtained from distal normal liver tissues of liver hemangioma. HBV-related HCC tissues from 64 early HBV-related HCC patients and normal liver tissues were obtained. Some parts of tissue was snap-frozen in liquid nitrogen within ve minutes after surgical removal. Other parts of tissue were preserved by xing them in formalin for pathological examinations.
HCC and normal liver tissues were determined by pathology after hepatectomy. 66 advance HCC patients (Stage IV) was diagnosed based on biopsy of the tumor or CT/MRI. All participants were enrolled without diagnosed concurrent urinary, infectious, autoimmune diseases and diabetes mellitus. All HCC patients were positive with HBsAg and without exposure to other well-recognized causes of HCC, i.e. Hepatitis C virus (HCV), alcohol, a atoxin, fatty liver, chronic biliary disease, genetic and metabolic liver diseases [13].
HBV-related HCC patients with a diagnosis of concurrent other cancers and those receiving chemotherapy and radiotherapy prior to sampling were also excluded. Plasmatic and urinary samples was obtained from all of 64 patients with early HBV-related HCC over 1 week before and over 1 month after curative surgery. Whole blood was collected into commercially available anticoagulant-treated tubes. Blood Cells are removed from plasma by centrifugation for 10 minutes at 2,000 x g using a refrigerated centrifuge. The plasma was transferred into microcentrifuge tubes followed by a second centrifugation at 12,0006 x g for 10 min at 4℃ to completely remove cellular components and debris as nal plasma samples. Up to 5 mL of urine from each participants were collected in a 50-mL centrifuge tube. In order to eliminate the miRNA expression in uences of exfoliated cells and debris from the urinary tract, the urine samples were centrifuged at 3,000 x g for 15 min at 4℃ to spin down exfoliated cells and debris, and the supernatant was transferred into microcentrifuge tubes followed by a second centrifugation at 12,000 x g for 10 min at 4℃ to completely remove cellular components as urinary supernatant samples. All samples were stored in the − 80°C lab freezers until use.
Ethics committees from our hospitals approved the study protocol. All participants provided written informed consent for their information and samples to be stored in the hospital database and used for research.

MiRNAs isolation and quanti cation
The following procedures were nished with seven days after sampling. Total RNAs were isolated from frozen liver tissues using TRIzol (Thermo Fisher Scienti c, USA), and total RNAs in 1mL of plasma or urine were extracted by the mirVana PARIS Kit (Thermo Fisher Scienti c, USA) following manufacturer's protocols. The measurements of miRNAs by quantitative polymerase chain reaction (qPCR) were carried out as previously described [14] Each qPCR reaction contained negative controls included no template control, no reverse transcriptase control, and no ampli cation control. All reactions including controls were perform in triplicate. U6 snRNA is the most frequently used reference gene for miRNA RT-qPCR expression analysis. Several studies indicate that U6 snRNA may be a reliable housekeeping gene for normalization of miRNA RT-qPCR expression analysis for tissue [15], plasma [16] and urine [17]. Hence, the expression levels of miR-93-5p in tissue, plasma and urine were normalized to corresponding expression levels of U6 snRNA. All expression levels were calculated using the 2 −ΔΔCt method [18].

Statistical Analysis
MiR-93-5p expression levels were compared using the Mann Whitney U test or Kruskal-Wallis H test. The differences of the miR-93-5p before and after surgery were analyzed by Wilcoxon signed-rank test.
Receiver operating characteristic (ROC) curves were used to evaluate the diagnostic performance of miR-93-5p for differentiating between two groups. The correlation between two groups was analyzed using the Spearman's correlation test. The statistical analyses above were performed using the SPSS software (ver. 13.0). The differences of discriminatory powers between plasmatic and urinary miR-93-5p were compared with the method of Delong using the MedCalc (version 19.4.1) [19,29]. A two-tailed P value < 0.05 was considered statistically signi cant.

Patient Characteristics
The characteristics of the enrolled participants were presented in Table 1. The patients with HBV-related HCC were divided into two groups: Patients' Group 1 (64 cases) was staged Ia, Ib, IIa with good liver function reserve and eligible for curative hepatectomy according to the Chinese guideline; Patients' Group 2 (66 cases) was staged IV with contraindication to hepatectomy and the presence of distant metastases. Cancer staging is based on China's medical systems [11]. There was no signi cant difference in the distribution of age and sex among the three groups (Healthy controls, early HBV-related HCC, and advanced HBV-related HCC). Serum alpha feto-protein (AFP) level was well balanced between the two patients' groups. 15.6% of early HBV-related HCC had high serum levels of AFP (> 400 ng/mL), and serum AFP concentration of > 400 ng/mL was seen in 16.7% of advanced HBV-related HCC. And over half of patients from the two groups of patients showed normal AFP levels (< 20 ng/mL) with the proportion of 59.4% and 53.0%, respectively. Totally, 56.2% of HBV-related HCC patients had normal AFP concentration regardless of the tumor resectability. Discriminatory Expression of the candidate microRNA biomarker in tissue, plasma and urine In the course of seeking candidate microRNA biomarkers, datasets of microarray-based miRNA pro les related to comparing the miRNA expression in HBV-related HCC tissue with normal liver tissue were searched, downloaded from the Gene Expression Omnibus (GEO) and analyzed. Detailed selection methods were described in the following Section of Materials and methods. Accordingly, two miRNA pro les met the criteria. The miRNA pro le of GSE69580 was assessed in ve HBV-related HCC tissues, and ve normal liver tissues. While GSE10694 revealed the pro le HBV-related HCC tissues, and 10 normal liver tissues. MiR-93-5p was upregulated signi cantly in HCC tissues of both datasets. Accordingly, it is chosen as a candidate microRNA biomarker for HBV-related HCC.
First, the expression levels of miR-93-5p were quanti ed in 64 tissues of HBV-related HCC and 13 normal liver tissues. MiR-93-5p was signi cantly elevated 4.0 times in HCC tissues (P < 0.001). Moreover, miR-93-5p levels in plasma and urine were also measured three groups of participants, including 65 healthy controls, 64 patients with early HBV-related HCC and 66 patients with patients with advanced HBV-related HCC. The results showed that compared with healthy controls, plasma miR-93-5p were signi cantly overexpressed 2.8 times in patients with early HBV-related HCC and 2.9 times in patients with patients with advanced HBV-related HCC (p < 0.001). But the plasma levels of miR-93-5p had no signi cant difference between the two groups of patients regardless of the HCC resectability (P = 0.976). Plasmatic miR-93-5p in all HBV-related patients were also signi cantly elevated 2.8 times compared with normal controls. Finally, urinary miR-93-5p concentration was also tested in 65 healthy controls, 64 patients with early HBV-related HCC and 66 patients with patients with advanced HBV-related HCC. The results displayed that compared with healthy controls, urinary miR-93-5p signi cantly increased 3.7 times in patients with early HBV-related HCC and 3.6 times in patients with advanced HBV-related HCC (p < 0.001). But the two groups of patients exhibited statistically similar miR-93-5p levels in urine (P = 0.963). In addition, regardless of the cancer resectability, urine miR-93-5p in all HBV-related patients also signi cantly increased 3.6 times than healthy controls (Fig. 1).
The miR-93-5p expression correlation between any two kinds of samples of tissue, plasma and urine 64 cancer tissue, plasma and urine samples were collected from the same patients with early HBV-related HCC. The correlation between any two groups of samples was analyzed using the Spearman's correlation test. The results indicated that signi cantly positive correlation of miR-93-5p levels was observed between tissue and plasma, plasma and urine, tissue and urine (p < 0.001) with coe cients of 0.550, 0.431, 0.481, respectively (Fig. 2).
The levels of plasma and urine miR-93-5p changed after curative hepatectomy In order to investigate whether plasma and urine miR-93-5p derived from HCC, 64 paired plasma and 64 urine samples were serially collected from the same group of patients with early HBV-relate HCC one month after primary curative hepatectomy. CT or/and MRI did not indicate cancer recurrence in them after one month of the surgery. The results showed that miR-93-5p in plasma and urine was signi cant reduced 4 times after the surgery (P < 0.001). Meanwhile, the expression levels of plasma and urine miR-93-5p after surgery were compared with the miR-93-5p levels in plasma and urine of healthy controls. The results reveal that both plasma and urine miR-93-5p levels had no signi cant differences between postoperational HCC and healthy controls (P = 0.264, 0.207, respectively) (Fig. 3).
The diagnostic accuracy of plasma and urine miR-93-5p for early HBV-relate HCC Receiver Operating Characteristic (ROC) curves were constructed and used to evaluate the diagnostic power of plasma and urine miR-93-5p for detecting early HBV-relate HCC. Accordingly, plasma miR-93-5p differentiated early HBV-relate HCC from healthy control with sensitivity of 85.9%, and speci city of 95.4%. And, urine miR-93-5p showed 87.5%, and speci city of 97.4% in the detection of early HBV-relate HCC (Fig. 4). There is no signi cant difference between plasma and urine miR-93-5p for detecting early HBV-relate HCC (P = 0.7458), advanced HCC (P = 0.4953), and total HCC cases (P = 0.7629) according to the Medcal software to compare the two diagnostic performances.

Discussion
To the best our knowledge, this is the rst literature that reports plasma and urine miRNA can serve desirable biomarkers for detecting early HBV-relative HCC. miR-93-5p was found to be signi cantly upregulated in cancer tissue of HBV-relative HCC in agreement with the two datasets of GSE69580 and GSE10694 from GEO. Notably, miR-93-5p in plasma and urine was also signi cantly increased in patients with early, advanced and total HBV-relative HCC cases. The expression of miR-93-5p was signi cantly and positively correlated between any two kinds of samples (tissue vs. plasma; tissue vs. urine, plasma vs. urine). Moreover, miR-93-5p in plasma and urine reduced signi cantly after over one month of the surgery and returned to normal levels. Take into consideration the results mentioned above, miR-93-5p was probably derived from HCC, enter the circulation, and secreted into urine at the end. Finally, ROC analysis showed both plasmatic and urinary miR-93-5p could serve as a novel biomarker for early HBVrelative HCC with desirable diagnostic accuracy. And their discriminatory power was similar.
Despite major efforts and advancing science and technology, a large proportion of HCC patients without speci c symptoms are still diagnosed at advanced stage when curative treatments are lacking. Diagnosis of HCC usually depends on imaging abdominal ultrasonography, imagnetic resonance imaging (MRI), and contrast-enhanced computed tomography (CT) and serum AFP. But MIR and CT are costly and not easily available in developing countries. Ultrasonography can detect large tumor but fails to detect small lesions, and because this procedure is operator-dependent, the diagnostic performance varies. AFP is the most frequently used biomarkers for HCC. However, the accuracy of AFP is not desirable with sensitivity ranged 46-59% and speci city ranged 87-93%. HCC at early stage are often missed by AFP analysis, and serum AFP levels are also overexpressed in patients with benign liver diseases, such as hepatitis and cirrhosis [12]. In this study, serum AFP still showed low sensitive in detecting HCC. Around 30% of early and advanced HCC showed high AFP levels (> 400 ng/mL). Accordingly, Asia-Paci c clinical practice guidelines [13] and American Association for the Study of Liver Diseases practice guidelines [14] do not recommend AFP as a diagnostic test for HCC. Hence, robust biomarkers are vital for early detection and diagnosis of HCC. In particular, biomarkers for early HCC is low-cost, non-invasive, and readily available.
Numerous microRNAs in tissue and circulation are demonstrated to be biomarkers with good sensitivities and speci cities for various disease. Dozens of studies prove that microRNAs in tissue [15], circulation [16,17], and several body uids including urine [18] are highly stable and readily detected. They can resist degradation at high and low temperatures, in strong acids and bases, and by RNase digest. Hence any one of microRNAs can be readily detected in any kinds of samples.
The mechanism of how microRNAs exist in urine remains elusive. It is postulated that miRNAs are secreted into the blood both passively after cell apoptosis and necrosis due to tissue damage, and actively through microvesicles [19]. MicroRNAs in the blood enter and pass through the glomerulus and nephrons of kidneys, and then travel via urinary tract to the outside of the body in urine. 1,000-2,000mL of urine are normally produced every day in a human based on uid intake and kidney function.
Both plasmatic and urinary miR-93-5p showed good diagnostic values outperforming AFP for detecting early HBV-related HCC. And miR-93-5p in both kinds of samples exhibited similar diagnostic performance. In addition, urine sampling is non-invasive, and readily performed in a human, while it is relatively expensive, invasive, and cause discomfort or even bloodborne diseases to individuals by blood drawing. Hence urinary miR-93-5p can outplace plasmatic miR-93-5p to server as a more promising biomarker in the help of diagnosing early HBV-related HCC.
Although the functional experiments of miR-93-5p were not performed in our study, some studies have found miR-93-5p increases signi cantly compared with adjacent non-caner tissue and act as an oncomir.  [23]. In contrast, downregulation of miR-93-5p expression results in reducing cell proliferation, migration, and clonogenicity of HCC cells [24]. Taken together, because miRNAs act as key molecule in HCC development, it provides the rationale for its use as also promising target for new HCC therapies.
A few limitations exist in this study. First, miRNA microarray was not conducted in this work. So other potential miRNA biomarkers may not be found. Second, the low number of patient samples of HBVrelated HCC represents a limitation in interpreting our results and evaluating the methods used. But we believe that the study power found in this work will be largely increased when more HCC cases are enrolled. Third, the patients were limited to a Chinese Han population. The ability of miR-93-5p to detect early HCC not associated with HBV and in multi-center worldwide with other ethnicities merits investigation.

Conclusions
In summary, this work demonstrated that plasmatic and urinary miR-93-5p have signi cant diagnostic values for early HBV-related HCC diagnosis with good sensitivity and speci city. Due to simple, noninvasive, and readily sampling, urinary miR-93-5p has high potential as a promising and robust biomarker for detecting early HCC. This study provides new evidence and prospect for the early detection of diseases using a noninvasive screening method of urine sampling, and urinary microRNAs can be noninvasive biomarkers for facilitating the diagnosis and prognosis of various human diseases.

Declarations Acknowledgments
We thank all the participants enrolled in this work.

Funding
This study was self-funded.
Availability of data and materials The datasets generated and/or analysed during the current study are not publicly available but are available from the corresponding author on reasonable request.

Ethics approval and consent to participate
All procedures performed in this study involving human participants were in accordance with the ethical standards of the institutional research ethics committee of The Fifth People's Hospital of Ganzhou and The First A liated Hospital of Gannan Medical University, Ganzhou, China, and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed consent was obtained from all individual participants included in the study.

Consent for publication
Not applicable. Figure 1 The aberrant expression of miR-93-5p in HCC tissue, plasma and urine of HBV-related HCC patients.

Figures
Compared with normal liver tissue, miR-93-5p was signi cantly upregulated in HBV-related HCC tissue (A). Compared with healthy controls, plasma (B) and (C) urine miR-93-5p was signi cantly overexpressed in early, advanced and total HCC patients. And plasma and urine miR-93-5p showed no signi cant difference between early and advanced HCC patients.

Figure 2
The miR-93-5p expression correlation between any two kinds of samples of tissue, plasma and urine. The expression of tissular miR-93-5p was signi cantly and positively correlated with plasmatic miR-93-5p (A).

Figure 3
The expression changes of plasmatic and urinary miR-93-5p after curative hepatectomy. Over one month after the curative hepatectomy, plasmatic (A) and urinary (C) miR-93-5p were signi cantly reduced. And the post-operational levels of plasmatic (B) and urinary (D) miR-93-5p showed no signi cantly differences with healthy controls.