Abstract

Background

Liver cancer is the second most common cause of death from cancer [1]. Hepatocellular carcinoma (HCC) accounts for a large part in all liver cancer cases, and is the fifth most common malignant tumor worldwide [2]. Infections with hepatitis B and C virus are the major risk factors of HCC [3]. Because of the difficulty of clinical diagnosis at the early stage, only 30–40% of cases can undergo curative resection [4]. HCC has an overall 5-year survival rate of 5–9%, and the rate can be raised to 69% in patients receiving curative resection [5]. In the present regimen, a-fetoprotein (AFP) has served as a biomarker to screen HCC. However, AFP only has a sensitivity of 39–56% and a specificity of 76–93%; approximately one-third HCC patients cannot be detected with AFP testing [2].

MicroRNAs (miRNAs) are a class of noncoding, endogenous, single-stranded RNAs with 19–22 nucleotides. They posttranscriptionally regulate gene expression by binding to the untranslated regions (UTRs) of target mRNA and exert their influences on biologic processes, such as the proliferation, differentiation, apoptosis, invasion and metastasis [6]. Over-expression of miR-125b inhibited endometrial cancer cell invasion [7]. Previous researches also indicated many miRNAs were involved in esophageal cancer [8]. Meta-analysis found high expression of miR-21 was associated with poor survival in patients with gastric cancer [9]. The miR-34b/c rs4938723 polymorphism may be associated with the risk of cancers, including nasopharyngeal cancer, osteosarcoma, and renal cell cancer [10]. Dysregulation of miR-15 and miR-16 was observed in chronic lymphocytic leukemia, and investigations have shown that miRNAs were associated with oncogenesis and development of various tumors [11]. miRNAs can exist stably in circulation, and are resistant to RNAase digestion. Moreover, serum miRNAs remained stable even when treated in low (pH=1) or high (pH=13) pH solution for 3 h [12]. Thus, circulating miRNAs have been explored as a biomarker, and some important findings have been reported. Yoshito et al. [13] found miR-21 had an area under the ROC curve (AUC) of 0.953, with a sensitivity of 87.3% and a specificity of 92% for determining HCC from healthy controls; the AUC was 0.773 (61.1% sensitivity and 83.3% specificity) for distinguishing HCC from chronic hepatitis. Liu et al. [5] showed that the combination of miR-15b and miR-130b for HCC detection yielded an AUC of 0.98 (98.2% sensitivity and 91.5% specificity); the sensitivity was 96.7% in a subgroup of HCC with low AFP (<20 ng/ml).

The aim of this study was to determine if plasma miRNAs can be biomarkers for HCC. The potential relationship between circulating miRNAs levels and existing clinicopathological features of HCC, such as the level of AFP, total bilirubin, transaminase, and tumor size, was analyzed.

Material and Methods

Study design and patient sample

This study included four phases

Phase I: Screening candidate miRNAs in HCC. Tumor and adjacent non-tumor tissues were sampled from 6 HCC patients receiving curative surgery, and submitted to miRNA array analysis. miRNAs with an up-regulated expression level were chosen as candidate miRNAs.

Phase II: Determining the target miRNAs. The plasma levels of candidate miRNAs before and after surgery (7–10 days) in 10 patients were detected by RT-qPCR, and those miRNAs that displayed a postoperative decease pattern were selected as the target miRNAs.

Phase III: Determining biomarkers. The plasma levels of target miRNAs in 37 HCC patients, 29 cirrhosis patients, and 31 healthy controls were measured and those miRNAs with up-regulated expressions in HCC patients were selected as biomarkers.

Phase IV: to validate the abilities of potential biomarkers for screening HCC from cirrhosis and healthy cohort, receiver-operator characteristic curves (ROC curve) were constructed and then the relationship between biomarkers and clinicopathologic parameters was validated.

All cases were from the First Affiliated Hospital of Chongqing Medical University. Cancer and adjacent non-tumor tissues from six HCC patients who received curative resection were sampled in June 2010, and were sent to Western Biotechnology, Inc. (Chongqing, China) for miRNA array analysis. The blood samples of 47 HCC patients, 29 cirrhosis patients, and 31 healthy cases were collected from January 2014 to June 2014. The diagnosis of HCC was confirmed by pathological examinations. The clinicopathological variables are summarized in Table 1. This study was approved ethically and scientifically by the local review board. Informed consent was obtained from all patients.

Table 1

Demographic and clinicopathologic parameters of patients from different cohorts.

Parameters	The number of different cohorts			P values
	Patients with HCC (n=47)	Patients with cirrhosis (n-29)	Healthy controls (n=31)	miR-15b-5p	miR-338-5p	miR-764
Sex
Male	44	12	18
Female	3	17	13
Age (years)
≥60	7	5	0
<60	41	24	31
α-fetoprotein (ng/ml)				0.595	0.039	0.102
≥200	19
<200	28
Alanine aminotransferase (U/l)				0.312	0.822	0.875
≥40	31
<40	16
Aspartate aminotransferase (U/l)				0.718	0.555	0.768
≥40	29
<40	18
Total bilirubin(μmol/L)				0.644	0.426	0.54
≥17.1	20
<17.1	27
Child-Pugh grade				0.989	0.7	0.184
A	44
B	3
C	0
Tumor size (cm)				0.346	0.336	0.026
≥5	26
<5	21

Open in a separate window

Results

The expression levels of target miRNAs in 3 different cohorts

To observe the value of plasma target miRNAs as the biomarkers, plasma levels of target miRNAs in 37 HCC patients, 29 cirrhosis patients, and 31 healthy cases were measured by RT-qPCR. In all, we obtained 107 samples, including 10 preoperative plasma samples. As shown in Table 4 and Figure 1, the plasma expression levels of miR-15b-5p, miR-338-5p, and miR-764 in HCC were higher than that in the other 2 cohorts (P<0.05). The plasma levels of miR-15b-5p and miR-338-5p in CIR were higher than in the HC cohort (P<0.0001, P=0.006), but a similar situation was not found in miR-764 (P=0.667). There were outliers for these 3 miRNAs.

Open in a separate window

Figure 1

The plasma levels of miR-15b-5p (A), miR-338-5p (B), and miR-764 (C) in HCC, CIR, and HC. (A) The different plasma expression levels of miR-15b-5p in 3 independent cohorts; (B) The different plasma expression levels of miR-338-5p in 3 independent cohorts; (C) The different plasma expression levels of miR-764 in 3 independent cohorts.“*” for outliers.

Table 4

The plasma expression levels of target miRNAs in HCC, cirrhosis (CIR) and healthy controls (HC).

miRNA P50 (P25, P75)	HCC (n=47)	CIR (n=29)	HC (n=31)	P value
				HCC vs. CIR	HCC vs. HC	CIR vs. HC
miR-15b-5p	1.44607 (1.00000, 3.21384)	1.00000 (1.00000, 1.51870)	0.41094 (0.18895,0.98531)	0.0239	<.0001	<0.001
miR-338-5p	4.16015 (1.56735, 8.79197)	1.00000 (0.77172, 1.51870)	0.79597 (0.09996,1.00000)	<.0001	<.0001	0.006
miR-764	2.49329 (1.03630, 9.01975)	1.00000 (0.55753, 1.15658)	1.00000 (0.25045,1.19243)	0.0002	0.0002	0.667

Open in a separate window

Diagnostic value analysis

The differentiating power of plasma miR-15b-5p, miR-338-5p, and miR-764 expressed in patients with HCC, cirrhosis, and healthy controls was validated. ROC was constructed using 3 models: HCC vs. CIR, HCC vs. HC, and HCC vs. CIR and HC. AUC and the cutoff values (using the highest Youden’s index) were computed. Consequently, we found these 3 miRNAs could differentiate HCC from CIR and HC. Especially, the plasma level of miR-338-5p was helpful for differentiating HCC from CIR and HC, with AUC of 0.799 (74.5% sensitivity, 82.8% specificity), and 0.909 (72.3% sensitivity, 99.68%), respectively. Furthermore, at the cutoff value of 1.784, miR-338-5p yielded an AUC of 0.856 (72.3% sensitivity, 90% specificity) for differentiating HCC samples from a total of 107 samples. Similarly, we found miR-764 and miR-15b-5p yielded an AUC of 0.791(74.5% sensitivity, 77% specificity), 0.765 (68.1% sensitivity, 80% specificity), respectively, for differentiating HCC samples from all samples. Moreover, differentiating power of the 3 plasma miRNAs was better in the HCC vs. HC model than in the HCC vs. CIR model. Details of ROC analysis are summarized in Table 5 and Figure 2.

Open in a separate window

Figure 2

The performance of miR-15b-5p, miR-338-5p, and miR-764 for detecting HCC from different cohorts. (A–C): ROC curve analysis using plasma miR-15b-5p for discriminating HCC from CIR, HC, CIR, and HC; (D–F): ROC curve analysis using plasma miR-338-5p for discriminating HCC from CIR, HC, CIR, and HC; (G–I): ROC curve analysis using plasma miR-764 for discriminating HCC from CIR, HC, CIR, and HC.

Table 5

The differentiating powers of miR-15b-5p, miR-338-5p and miR-764.

miRNA	HCC vs. CIR	HCC vs. HC	HCC vs. CIR and CH
	Cutoff value, AUC (sensitivity, specificity)
miR-15b-5p	1.007, 0.654 (68.1%, 79%)	0.807, 0.871 (87.2%, 74.2%)	1.007, 0.765 (68.1%, 80%)
miR-338-5p	1.743, 0.799 (74.5%, 82.8%)	1.784, 0.909 (72.3%, 99.68%)	1.784, 0.856 (72.3%, 90%)
miR-764	1.181, 0.770 (74.5%, 75.9%)	1.465, 0.811 (70.2%, 83.9%)	1.199, 0.791 (74.5%, 77%)

Open in a separate window

Discussion

Our study was divided into 4 phases according to international classical cases of screening potential biomarkers. Interestingly, we found: (i) the plasma levels of miR-764 in the HCC cohort were higher than that in the other 2 cohorts, while no difference occurred between cirrhosis and healthy cohorts. This phenomenon indicated that miR-764 did not affect the transformation from healthy liver to cirrhosis; instead, it played a role in the transformation from cirrhosis to HCC. Plasma levels of miR-15b-5p and miR-338-5p increased gradually from healthy controls to HCC patients, which indicated that they played roles in the malignant transformation of liver cells. (ii): miR-764 yielded the highest sensitivity for determining HCC cases with AFP<200 ng/ml. This might be useful for determining HCC in early stages when the level of AFP is lower than the diagnostic standard.

Numerous studies have revealed that circulating miRNAs can be potential markers for HCC determination. Liu et al. [5] indicated that the combination of miR-15b and miR-130 was a useful biomarker for screening HCC. Li et al. [16] found circulating miR-18b could be a potential HBV-related HCC marker. For searching more convenient HCC markers, Moemen et al. [17] detected the expression levels of miRNAs in urine samples. MiR-618 and miR-650 were selected out, but further analysis revealed the predictive values were not improved compared to the traditional AFP level-based detection method. In addition to the miRNAs mentioned above, miR-221, miR-885-5p, and miR-122a have been identified to be potential biomarkers for HCC determination [18–20]. We found no published research about circulating miR-338-5p and miR-764 in HCC. Moreover, studies on miR-15b in HCC are limited and the role of miR-15b remains largely unknown. Dai et al. [21] reported that miR-15b promoted HBV replication by targeting hepatocyte nuclear factor 1α. Wu et al. [22] indicated the expression of miR-15b can be repressed by HBX, and then enhanced hepatocellular carcinoma proliferation through fucosyltransferase2-induced Globo H expression.

A growing body of evidence shows that circulating miRNAs are probable markers of various other tumors, including thyroid cancer [23], breast cancer [24], testicular germ cell cancer [25], pancreatic cancer [26], lung cancer [27], and gastric cancer [28].

The main limitation of this study was that the sample size was small. The present findings therefore should be validated in trials with more cases. Because of only 3 female cases were included into our study, the difference between males and females should also be explored in subsequent trials.

Conclusions

Our study found that miR-338-5p and miR-764 hold promise as valuable biomarkers for HCC detection, even in HCC patients with low AFP levels. We also validated the differentiating power of miR-15b-5p. In summary, the present study results suggest that miR-15b-5p, miR-338-5p, and miR-764 might be used as biomarkers for HCC.

Footnotes

References