Inverse relationship between the level of miRNA 148a-3p and both TGF-β1 and FIB-4 in hepatocellular carcinoma.

BACKGROUND AND AIM: Hepatocellular carcinoma (HCC) is a major health burden globally. Dysregulation of miRNA 148a-3p is engaged in carcinogenesis. TGF-β is a profibrogenic cytokine. This study assesses the expression level of miRNA 148a-3p and its relationship with serum TGF-β1 and fibrosis index based on four factors (FIB-4) in Egyptian patients with HCV-associated HCC.
SUBJECTS: and
Methods: The study included 72 HCC patients with HCV, 48 HCV cirrhotic patients, and 47 healthy controls. Serum TGF-β1 was assessed by ELISA and the expression of miRNA 148a-3p was measured by RT-PCR.
RESULTS: Patients with HCC had lower plasma miRNA 148a-3p, higher serum TGF-β1, and higher FIB-4 levels than patients with cirrhosis and controls. miRNA 148a-3p discriminated HCC either from control (AUC: 0.997, 95.83% sensitivity, 85.11% specificity) or from cirrhosis (AUC: 0.943, 91.67% sensitivity, 81.25% specificity). Moreover, it distinguished metastatic from nonmetastatic patients (AUC: 0.800, 88.89% sensitivity, 60.0% specificity). The decreased miRNA 148a-3p and the increased TGF-β1 levels were related to distant metastasis, multinodular lesions, advanced TNM stage, and BCLC score (C). A negative correlation between miRNA 148a-3p and each of FIB-4 and TGF-β1 was detected. The decreased miRNA 148a-3p was associated with poor overall survival and poor progression-free survival.
CONCLUSION: An inverse relationship between miRNA 148a-3p and both TGF-β1 and FIB-4 was observed, which could be involved in HCC pathogenesis. Moreover, this miRNA is a potential diagnostic and prognostic biomarker for HCC.

Introduction

Hepatocellular carcinoma (HCC) is the 6th frequently identified malignancy and the 4th principle cause of malignancy-related mortality worldwide [1]. HCC is terminal, with most patients being diagnosed at advanced stages and an incidence rate very close to the mortality rate [2]. HCC accounts for approximately 75%–85% of primary hepatic malignancies and is considered a health burden worldwide, with Egypt having the highest estimated age-standardized incidence rates in Africa [3]. Hepatitis C virus (HCV) is the most important risk factor for HCC in Egypt [4].

HCV infection is the prime cause of HCC, which could also develop in patients who report sustained virologic response, and hepatic inflammation and fibrosis markers such as FIB-4 index are used as predictors of HCC development [5]. Moreover, HCV infection enhances the development of HCC principally through various pathways such as sustained inflammation, enhancing fibrosis and cellular proliferation and necrosis; the risk of HCC propagation is associated with the fibrosis stage [6].

MicroRNAs (miRNAs) are the main constituents of the noncoding RNA family. miRNAs are about 23 nucleotides in length; they control gene expression posttranscriptionally. They regulate the expression of specific mRNAs mainly by attaching to the 3′-untranslated region of specific mRNA [7]. Moreover, miRNAs are essential factors for controlling the progression of HCC. They act through various mechanisms such as cellular proliferation, cell death, metastasis, and epithelial-mesenchymal transition (EMT) [8].

miR-148a, miR-148b, and miR-152 are members of the miR-148/152 family [9]. miR-148a is located on chromosome 7p15.2 and seems to have a stem-loop structure sequence [10]. miRNA 148a-3p is produced from the 3′ terminal of the pre-miRNAs by Dicer cleavage [11]. Recent data have revealed that miRNA-148a-3p is a tumor-suppressive miRNA that is underexpressed in epithelial ovarian cancer [12], breast cancer [13], and HCC [14,15]. Several growth factors, including transforming growth factor beta (TGF-β), are influenced by miRNA-148a as a tumor suppressor; DNA hypermethylation has been linked to miRNA-148a decreased expression in a variety of metastatic cell lines [16]. TGF-β is a biovital factor that promotes cellular growth and proliferation and is a very effective profibrogenic cytokine in hepatocytes. Moreover, TGF-β is involved in cancer progression by initiating EMT [17]. Based on the association of miRNA-148a-3p and TGF-β1 with fibrosis, this study was designed to evaluate the expression level of miRNA 148a-3p and its relationship with both TGF-β1 level and fibrosis index (FIB-4) in patients with HCV-associated HCC.

Subjects and Methods

Ethical approval

This study was performed in accordance with the Declaration of Helsinki. Written informed consent was signed by all participants and approved by the Ethics Committee of Medical Research, Faculty of Medicine and National Liver Institute, Menoufia University, Egypt (no. BIO-12-2018).

Selection of patients

This study was conducted in the Medical Biochemistry and Molecular Biology Department in collaboration with the Clinical Oncology Department, Faculty of Medicine, and the National Liver Institute, Menoufia, Egypt, from December 2018 to December 2020. The study included 120 individuals divided into three groups: group I, 72 patients with HCC on top of chronic HCV infection; group II, 48 patients with cirrhosis due to chronic HCV infection; group III, 47 age- and gender-matched healthy volunteers as the controls who were negative for HCV antibody, HB surface antigen (HBsAg), and HB core antibody (HBcAb). The selection of patients was based on specific inclusion and exclusion criteria. The inclusion criteria were patients older than 18 years with HCV infection confirmed by anti-HCV detection and positive HCV RNA. The diagnosis of cirrhosis was made based on history, clinical examination, laboratory findings, and imaging criteria such as ultrasonography and computed tomography (CT) [18]. On the other hand, HCC diagnosis was reached by triphasic spiral CT and/or magnetic resonance imaging, showing characteristic HCC features, elevated alpha-fetoprotein (AFP), and/or liver biopsy [19]. HCC staging was defined according to the Barcelona Clinic Liver Cancer staging system (BCLC) [20]. The Child–Pugh score was used to assess the severity of the liver disease [21]. FIB-4 is a convenient and noninvasive index used for assessing and predicting liver fibrosis. It was calculated according to the following formula: aspartate aminotransferase (AST) (U/L) × age (years)/[platelet count ( × 109/L) × alanine aminotransferase (ALT) (U/L) 1/2 ] [22].

The exclusion criteria included patients with positive HBsAg and/or HBcAb, secondary liver cancer, other malignancies, chronic hepatitis, or cirrhosis due to any cause other than HCV infection, those who had previous HCC treatment and history of antiviral treatment, and candidates for localized treatment.

Patients with good liver functions, Child–Pugh A classification, received sorafenib treatment (400 mg twice daily). Patients’ responses to treatment were evaluated every month by clinical examination, CBC test, liver and kidney functions tests and every 3 months by CTs and AFP test. Complete response (CR) is the disappearance of the disease, while partial response (PR) is characterized by a decrease of at least 50% in the sum of the products of the two greatest perpendicular diameters of measurable lesions, for at least four weeks. Stable disease (SD) is defined as a decrease of less than 50% or an increase of less than 25% in the tumor size, while progressive disease (PD) is characterized by an increase of more than 25% in the sum of the products of two perpendicular diameters of at least one tumor or the appearance of a new lesion. Treatment-related toxicities in the group were evaluated in all participants using the NCI-CTC systemV5.0. All patients were followed up for 24 months at least. Furthermore, patients with Child–Pugh B or C classifications were treated with supportive treatment.

Specimen requirements

Seven milliliters of venous blood was drawn using sterile venipuncture and distributed in three different tubes. Four milliliters of blood was placed in a plain tube, allowed to clot, and then centrifuged for 10 min at 4000 rpm. The serum obtained was utilized to measure liver function (ALT, AST, and albumin), hepatitis viral markers, AFP, and TGF-β1 levels. The remaining 3 mL was put in an EDTA tube for complete blood count (CBC), HCV real-time polymerase chain reaction (RT-PCR), and expression analysis of miRNA-148a-3p.

Laboratory techniques

CBC was measured by a Sysmex XN-1000 (Japan (19723); BM Egypt Company). Alanine aminotransferase (ALT) and aspartate aminotransferase (AST) were evaluated using the kinetic UV optimized method IFCC (LTEC Kit, England) [23,24]. Serum albumin was quantified utilizing the technique with enhanced specificity of bromocresol green colorimetric assay (Diamond Diagnostics Kit, Germany) [25]. Electrochemiluminescence immunoassay (ECLA) on COBAS immunoassay analyzer was employed for assessing HCV antibody (anti-HCV) [26], while HBsAg was detected using Sorin Biomedica Co. kit (Italy) [27]. AFP was measured by enzyme-linked immunosorbent assay (ELISA) (IMMULITE 1000 system by a kit provided by Siemens Medical Solutions Diagnostics, USA) [28]. Moreover, the ELISA technique was used to estimate the serum level of TGF-β1 (Human TGF-β ELISA Kit, SunRed, China) [29]. HCV RNA was detected by RT-PCR using COBAS TaqMan HCV quantitative test, version 2.0 (Roche Molecular Systems, Inc., Branchburg, NJ, USA) [30].

Expression assessment of the circulating miRNA 148a-3p

To extract miRNA from plasma, we used the miRNeasy® Mini kit (QIAGEN, Germany). The quantity and the purity of extracted miRNA were estimated by a NanoDrop instrument (Thermo Scientific, USA). Purified miRNA was stored at −80 °C. The extracted miRNA was reverse-transcribed to yield single-stranded cDNA using a miScript II RT kit (QIAGEN, Germany). Reverse transcription reaction was performed on ice in a total volume of 20 μl for every reaction. The mixture consisted of 4 μl of 5x miScript HiSpec RT buffer, 2 μl of 10x miScript Nucleics Mix, and 2 μl of miScript™ reverse transcriptase; 2 μl of nuclease-free water was pipetted into each well; then, 10 μl of extracted miRNA was added. Measurement was completed in a 2720 Applied Biosystems thermal cycler (Bioline, Singapore, USA) for one cycle under the following conditions: 37 °C for 60 min and 95 °C for 5 min. The cDNA was stored at −20 °C. The real-time polymerase chain reaction was achieved using a miScript® SYBR® Green PCR kit (QIAGEN, Germany). Before amplification, cDNA samples were diluted with nuclease-free water at a ratio of 1:5. A total volume of 25 μl was used:12.5 μl of SYBR Green Master Mix, 3.5 μl of nuclease-free water, 4 μl of diluted cDNA, 2.5 μl of miScript universal primer, and 2.5 μl of miScript primer assay. The miRNA RNU6 was used as a reference miRNA. The miScript primer assay (Cat# MS00003556) containing miRNA-specific forward primers was used to detect mature miRNA-148a-3p (UCAGUGCACUACAGAACUUUGU). The amplification was run in an ABI 7500 real-time PCR instrument (software version 2.0.1), guided by the following steps: preliminary activation at 95 °C for 15 min and 40 cycles at 94 °C for 15 s, 55 °C for 30 s, and 70 °C for 30 s. The expression level of miRNA-148a-3p was normalized to that of RNU6, and the relative quantification of the miRNA was estimated using the comparative 2−ΔΔCt method. Furthermore, melting curve analysis confirmed the specificity of the amplification and absence of primer dimers (Fig. 1) [31].

Fig. 1

(A) Amplification plot of the miR-148a-3p. (B) Melting curve analysis.

Statistical analysis

Analysis of data was carried out using IBM SPSS software package version 20.0. (Armonk, NY: IBM Corp). The comparisons between groups for categorical variables were assessed using the chi-square test (Fisher or Monte Carlo). For comparing two groups, Student's t-test was used for normally distributed quantitative variables, whereas Mann–Whitney test was used for abnormally distributed quantitative variables. For comparing the three groups under study, ANOVA was used for normally distributed variables followed by Tukey's post hoc test. In contrast, Kruskal–Wallis test was utilized for abnormally distributed quantitative variables followed by post hoc test (Dunn's for multiple comparisons test). Spearman's coefficient was used to investigate the relationship between quantitative variables. Moreover, the receiver operating characteristic curve (ROC) curve was used to assess the diagnostic performance of the markers. Kaplan–Meier survival and Cox regression analyses were conducted. The significance of the obtained results was judged at the 5% level.

Results

Our data analysis revealed that age was not significantly different when comparing HCC and cirrhosis groups to controls (p2 = 0.069 and p3 = 0.863), whereas patients with HCC were significantly older than patients with cirrhosis (p1 = 0.015). Gender and BMI (p = 0.060 and p = 0.143, respectively) did not show significant variations among the groups under study. However, statistically significant differences in the history of bilharziasis and smoking status were observed in our groups. Differences in serum levels of ALT, AST, albumin, AFP, TGF-β1, and platelets count were statistically significant among groups (p < 0.001) with elevated ALT, AST, AFP, and TGF-β1 and decreased serum albumin and platelets count in HCC and cirrhosis versus controls and in HCC versus cirrhosis. The assessment of the miRNA 148a-3p relative expression level revealed significantly decreased levels in HCC patients compared to both cirrhosis and control groups (p1 < 0.001). Moreover, cirrhosis group showed decreased expression levels compared to controls (p3 = 0.028) (Table 1).

Table (1)

Comparison between the three studied groups according to different parameters.

	HCC (n = 72)	Cirrhosis (n = 48)	Control (n = 47)	Test of sig.	P
Age
Min. – Max.	47–74	46.0–66.0	45.0–68.0	F = 4.705**	0.010**
Mean ± SD.	59.1 ± 5.6	56.2 ± 5.5	56.8 ± 5.1
Sig. bet. grps	p1 = 0.015*,p2 = 0.069,p3 = 0.863
Sex
Male	57 (79.2%)	30 (62.5%)	29 (61.7%)	χ2 = 5.628	0.060
Female	15 (20.8%)	18 (37.5%)	18 (38.3%)
BMI
Min. – Max.	20–33	18.5–35	18–32.5	F = 1.967	0.143
Mean ± SD.	25.7 ± 3.1	26.9 ± 4.1	25.5 ± 4
History of bilhaziasis	44 (61.1%)	19 (39.6%)	0 (0%)	χ2 = 45.304***	<0.001***
Smoking
Non smoker	33 (45.8%)	24 (50%)	39 (83%)	χ2 = 17.599***	<0.001***
Smoker	39 (54.2%)	24 (50%)	8 (17%)
ALT(IU/L)
Min. – Max.	12–77	10–88	6–43	H = 33.792***	<0.001***
Median (IQR)	37 (28–52.5)	31 (22–43)	22 (19–32.5)
Sig. bet. grps	p1 = 0.017*,p2 < 0.001*,p3 = 0.002*
AST(IU/L)
Min. – Max.	34–137	30–114	7–43	H = 93.523***	<0.001***
Median (IQR)	57 (47.5–70.5)	56 (42–69.5)	22 (19–33)
Sig. bet. grps	p1 = 0.359,p2 < 0.001*,p3 < 0.001*
Platelets (103/mm3)
Min. – Max.	46–200	79–200	195–452	F = 191.643***	<0.001***
Mean ± SD.	121.8 ± 39.6	145.4 ± 27.4	277.1 ± 60.4
Sig. bet. grps	p1 = 0.012*,p2 < 0.001*,p3 < 0.001*
Serum albumin (g/dL)
Min. – Max.	1.8–35	2–4.6	3–5.4	H = 28.951***	<0.001***
Median (IQR)	3.4 (3–4.1)	3.7 (3.2–4.2)	4 (3.8–4.8)
Sig. bet. Grps	p1 = 0.165,p2 < 0.001*,p3 < 0.001*
AFP (ng/ml)
Min. – Max.	3–8910	2.1–614	1–8	H = 92.527***	<0.001***
Median (IQR)	152 (31.8–982.5)	17 (8.1–69.8)	6 (5–6)
Sig. bet. Grps	p1 < 0.001*,p2 < 0.001*,p3 < 0.001*
TGF-β1 (ng/ml)
Min. – Max.	1.9–6.1	0.6–3	0.2–2.5	H = 138.508***	<0.001***
Median (IQR)	3.7 (3.5–4)	1.6 (1.2–1.8)	0.3 (0.3–1.7)
Sig. bet. grps	p1 < 0.001*,p2 < 0.001*,p3 = 0.026*
miRNA 148a-3p
Min. – Max.	0.1–4.9	0.4–22.4	1.8–81.3	H = 108.667***	<0.001***
Median (IQR)	0.9 (0.3–1.7)	9.4 (4–17.3)	22 (7.5–36.2)
Sig. bet. grps	p1 < 0.001*,p2 < 0.001*,p3 = 0.028*

χ2: Chi square test.

F: F for ANOVA test, Pairwise comparison bet. each 2 groups was done using Post Hoc Test (Tukey).

H: H for Kruskal Wallis test, Pairwise comparison bet. each 2 groups was done using Post Hoc Test (Dunn's for multiple comparisons test).

p: p value for comparing between the studied groups.

p1: p value for comparing between group I and group II.

p2: p value for comparing between group I and group III.

p3: p value for comparing between group II and group III.

*: Statistically significant at p ≤ 0.05.

**: Statistically significant at p ≤ 0.01.

***: Statistically significant at p ≤ 0.001 IQR: Inter quartile range.

In assessing markers of fibrosis, patients with HCC had elevated FIB-4 compared to patients with cirrhosis (p1 < 0.001), whereas both groups did not show significant variations regarding presence of splenomegaly, ascites, different comorbidities, or Child–Pugh class (Table 2).

Table (2)

Comparison between patients with HCC and cirrhosis according to different parameters.

	HCC (n = 72)	Cirrhosis (n = 48)	Test of sig.	p
Splenomegaly	54 (75%)	31 (64.6%)	χ2 = 1.513	0.219
Ascites
No	53 (73.6%)	41 (85.4%)	χ2 = 2.566	MCp=0.214
Mild	18 (25%)	7 (14.6%)
Moderate	1 (1.4%)	0 (0%)
Comorbidities
DM	27 (37.5%)	18 (37.5%)	χ2 = 0.0	1.000
HTN	15 (22.1%)	17 (35.4%)	χ2 = 2.513	0.113
Heart disease	6 (8.3%)	0 (0%)	χ2 = 4.211	FEp = 0.080
Child Pugh class
A	41 (56.9%)	33 (68.8%)	χ2 = 3.944	MCp=0.124
B	26 (36.1%)	15 (31.3%)
C	5 (6.9%)	0 (0%)
FIB4
Min. – Max.	3–15.3	3.1–6	U = 1022.0***	<0.001***
Median (IQR)	4.5 (3.4–6.3)	3.5 (3.2–3.8)

χ2: Chi square test MC: Monte Carlo FE: Fisher Exact.

t: Student t-test U: Mann Whitney test.

p: p value for comparing between the studied groups.

*: Statistically significant at p ≤ 0.05.

**: Statistically significant at p ≤ 0.01.

***: Statistically significant at p ≤ 0.001 IQR: Inter quartile range.

The detailed clinical characteristics of the patients with HCC are listed in Table 3. From patients with HCC, 27 (37.5%) patients had vascular invasion and a similar percentage represented those with distant metastasis. Regarding imaging findings, 35 (48.6%) patients had large sized tumors (>5 cm), while 37 (51.4%) patients had tumors ≤5 cm. The main site of tumor in the liver was the right lobe (52.8%); tumors in both lobes were found in 26.4% of patients and tumors of the left lobe in 18.1% of patients. The caudate lobe tumor was reported in only 2% of HCC patients. In terms of tumor staging, 22.2%, 5.6%, 31.9%, and 40.3% of the patients were in TNM stage I, II, III, and IV, respectively. Lymph node metastasis was detected in 25% of patients. Regarding BCLC scoring, A, B, and C scores were reported in 12.5%, 43.1%, and 44.4% of the patients, respectively. At the end of follow-up, disease progression was reported in 49 (68.1%) patients and 34 (47.2%) patients died.

Table (3)

Clinical parameters and characteristics of patients with HCC (n = 72).

	No. (%)
Vascular invasion	27 (37.5%)
Distant Metastasis
No	45 (62.5%)
Yes	27 (37.5%)
Tumor
Number
Single	26 (36.1%)
Multiple	46 (63.9%)
Size
≤5 cm	37 (51.4%)
>5 cm	35 (48.6%)
Site
Right Lobe	38 (52.8%)
Left Lobe	13 (18.1%)
Caudate lobe	2 (2.8%)
Both	19 (26.4%)
TNM staging
I	16 (22.2%)
II	4 (5.6%)
III	23 (31.9%)
IV	29 (40.3%)
LN metastasis	18 (25%)
BCLC
A	9 (12.5%)
B	31 (43.1%)
C	32 (44.4%)
Survival
Alive	38 (52.8%)
Died	34 (47.2%)
Progression
No	23 (31.9%)
Yes	49 (68.1%)

ROC curve was performed to assess the potential value of our markers in the diagnosis. ROC results showed that AFP at a cutoff >6 had sensitivity of 98.61% and specificity of 91.49% and miRNA-148a-3p had a sensitivity of 95.83% and specificity of 85.11% at a cutoff ≤3.62 (Table 4; Fig. 2A) for discriminating patients with HCC from controls. However, miRNA 148a-3p had a better performance at a cutoff ≤3.1 with a sensitivity of 91.67% and a specificity of 81.25% and AFP revealed a sensitivity of 90.28% and a specificity of only 54.17% at a cutoff >18 (Table 4; Fig. 2B) for distinguishing patients with HCC from patients with cirrhosis. Additionally, miRNA 148a-3p showed a good performance in predicting metastatic patients from nonmetastatic ones with a sensitivity of 88.89% and specificity of 60% at a cutoff of ≤0.9 (Table 4; Fig. 2C). Combining the values of miRNA-148a-3p and AFP did not improve the results compared to the values of miRNA 148a-3p alone.

Table (4)

Agreement (sensitivity, specificity) for miRNA 148a-3p and AFP to differentiate between different groups.

	AUC	P	95% C.I	Cut off	Sensitivity	Specificity	PPV	NPV
HCC patients (n = 72) from control (n = 47)
miRNA 148a-3p	0.997	<0.001***	0.960–0.998	≤3.62	95.83	85.11	90.8	93.0
AFP (ng/ml)	0.987	<0.001***	0.962–1.012	≥6	98.61	91.49	94.67	97.73
HCC patients (n = 72) from cirrhosis (n = 48)
miRNA 148a-3p	0.943	<0.001***	0.903–0.984	≤3.1	91.67	81.25	88.0	86.7
AFP (ng/ml)	0.777	<0.001***	0.695–0.859	≥18	90.28	54.17	74.7	78.8
metastatic patients (n = 27) from non-metastatic (n = 45) in HCC group
miRNA 148a-3p	0.800	<0.001***	0.697–0.903	≤0.9	88.89	60.0	57.1	90.0

AUC: Area Under a Curve p value: Probability value.

CI: Confidence Intervals.

NPV: Negative predictive value PPV: Positive predictive value.

#Cut off was choose according to Youden index.

*: Statistically significant at p ≤ 0.05.

**: Statistically significant at p ≤ 0.01.

***: Statistically significant at p ≤ 0.001.

Fig. 2

(A) ROC curves of miRNA-148a-3p and AFP to differentiate patients with HCC from controls. (B) ROC curves of miRNA-148a-3p and AFP to differentiate patients with HCC from cirrhosis. (C) ROC curves for miRNA-148a-3p to differentiate metastatic patients from nonmetastatic ones.

The miRNA-148a-3p expression level was significantly decreased in metastatic tumors compared to localized tumors (p < 0.001), multiple tumors compared to single tumors (p = 0.007), and advanced tumor stages (stages III and VI) compared to stage I and II tumors (p < 0.001), indicating aggressive tumor behavior. Moreover, patients with progressive BCLC score (C) had lower expression level than patients with A and B scores (p = 0.011). The Child–Pugh class, tumor site, and size were not associated with the miRNA 148a-3p expression level (Table 5). Regarding the relationship between serum TGF-β1 and different parameters, a significant prevalence of elevated TGF-β1 levels was observed in metastatic tumors compared to that in localized tumors (p < 0.001), tumor represented in both lobes of the liver compared to single lobe tumors (p < 0.033), advanced tumor stage (VI) compared to stage I, II, and III tumors (p < 0.001), and progressive BCLC score (C) compared to patients with A and B score (p = 0.008). Futhermore, no significant variations were observed in terms of the Child–Pugh class of the patients, tumor number, and size (Table 5).

Table (5)

Relation between miRNA 148a-3p expression level and serum TGF-β1 and different parameters in patients with HCC (n = 72).

	N	miRNA 148a-3p	TGF-β1 (ng/ml)
		Median (IQR)	Median (IQR)
Child Pugh class
A	41	0.8 (0.4–1.3)	3.7 (3.5–4)
B	26	0.9 (0.7–2.4)	3.6 (3.4–4)
C	5	0.1 (0.1–0.9)	4 (4–4.1)
H(p)		4.015 (0.134)	2.550 (0.279)
Distant Metastasis
No	45	1.1 (0.8–2.3)	3.6 (3.4–3.9)
Yes	27	0.4 (0.2–0.8)	4 (3.8–4.6)
U(p)		242.50*** (<0.001***)	262.0*** (<0.001***)
Tumor
Number
Single	26	1.3 (0.8–2.1)	3.6 (3–4)
Multiple	46	0.7 (0.3–0.9)	3.9 (3.5–4.2)
U(p)		368.0** (0.007**)	432.0 (0.052)
Size
≤5 cm	37	0.9 (0.4–1.4)	3.6 (3.5–4)
>5 cm	35	0.8 (0.3–1.9)	3.9 (3.5–4.1)
H(p)		554.0 (0.319)	4.32 (0.376)
Site
Rt Lobe	38	0.9 (0.4–2)	3.6 (3.5–3.9)
Lt Lobe	13	0.9 (0.8–1.6)	3.6 (3–4)
Caudate lobe	2	0.6 (0.4–0.8)	3.9 (3.7–4.1)
Both	19	0.4 (0.3–0.9)	4 (3.6–4.4)
H(p)		4.716 (0.194)	8.735* (0.033*)
TNM staging
I	16	1.5 (1.2–3)	3.5 (2.9–3.6)
II	4	2.1 (1.5–2.6)	3.2 (2.8–3.6)
III	23	0.8 (0.4–0.9)	3.7 (3.5–4.1)
IV	29	0.4 (0.2–0.9)	4 (3.8–4.6)
H(p)		20.196*** (<0.001***)	20.282*** (<0.001***)
BCLC
A	9	1.4 (1.3–2)	3 (2.8–3.6)
B	31	0.9 (0.4–1.7)	3.6 (3.5–4.1)
C	32	(0.2–1.1)	3.9 (3.6–4.1)
H(p)		8.954** (0.011*)	9.594** (0.008**)

U: Mann Whitney test H: H for Kruskal Wallis test.

p: p value for comparing between the studied categories.

*: Statistically significant at p ≤ 0.05.

**: Statistically significant at p ≤ 0.01.

***: Statistically significant at p ≤ 0.001.

In patients with HCC, the miRNA 148a-3p expression level was significantly negatively correlated with fibrosis markers, TGF-β1 (r = −0.696, p < 0.001) and FIB-4 (r = −0.615, p < 0.001) (Fig. 3A and B).

Fig. 3

(A) Correlation between the miRNA-148a-3p and TGF-β1 in patients with HCC. (B) Correlation between the miRNA-148a-3p and FIB4 in patients with HCC. (C) Kaplan–Meier survival curve for the overall survival with miRNA-148a-3p in patients with HCC. (D) Kaplan–Meier survival curve for progression-free survival with miRNA-148a-3p in patients with HCC.

In terms of identifying the predictors of survival in HCC patients at the end of the 24-month follow-up period, the univariate analysis revealed that the miRNA 148a-3p expression levels (low expression versus high expression), TGF-β1 serum levels (high levels versus low levels), tumor number (multiple versus single tumors), tumor stage (III + IV versus I + II), and BCLC score (C versus A + B) were significantly associated with decreased mean survival. However, AFP levels, Child–Pugh class, presence of metastasis, and tumor size did not influence patients' survival significantly (Table 6). Moreover, a multivariate Cox regression analysis was performed to identify the independent markers affecting patients’ survival. The miRNA-148a-3p expression level (low expression versus high expression) was the only independent predictor of the decreased overall survival (p = 0.007) (Table 6).

Table (6)

Univariate and multivariate COX regression analysis for predict overall survival in HCC groups.

	Univariate				#Multivariate
	Mean (95%C.I)	SE	Log rank	P	Wald	HR (95%C.I)	P
Child Pugh class
A	16.10 (13.6–18.61)	1.28	0.636	0.728
B	17.27 (14.13–20.41)	1.60			–	–	–
C	15.89 (13.30–18.48)	1.32
Metastatic site
No	17.32 (14.84–19.80)	1.26	1.316	0.251
Yes	14.22 (12.14–16.29)	1.06			–	–	–
Tumor Number
Single	19.89 (17.25–22.54)	1.35	6.106**	0.013**	0.046	0.902 (0.35–2.32)	0.992
Multiple	14.97 (12.61–17.33)	1.20
Tumor Size
≤5 cm	17.33 (14.74–19.92)	1.32	0.467	0.494	–	–	–
>5 cm	14.20 (12.13–16.26)	1.05
TNM staging
I + II	22.30 (20.06–24.54)	1.14	12.656***	<0.001***
III + IV	14.62 (12.45–16.79)	1.11			1.692	3.170 (0.56–18.03)	0.193
BCLC
A + B	18.84 (16.53–21.15)	1.18	6.401**	0.011**
C	12.63 (10.46–14.80)	1.11			2.104	1.713 (0.83–3.54)	0.147
AFP (ng/ml)
Low median	17.36 (14.80–19.93)	1.31	0.319	0.572	–	–	–
High median	16.09 (13.39–18.79)	1.38
TGF-β1 (ng/ml)
Low median	18.737 (16.29–21.18)	1.25	5.530**	0.019**
High median	12.992 (10.93–15.06)	1.06			0.355	0.791 (0.37–1.71)	0.551
miRNA 148a-3p
Low median	13.25 (10.99–15.51)	1.15	20.924***	<0.001***	7.308	5.089 (1.56–16.56)	0.007**
High median	21.67 (19.49–23.84)	1.11

HR: Hazard ratio.

C.I: Confidence interval LL: Lower limit UL: Upper Limit.

#: All variables with p < 0.05 was included in the multivariate.

*: Statistically significant at p ≤ 0.05.

**: Statistically significant at p ≤ 0.01.

***: Statistically significant at p ≤ 0.001.

By applying the log-rank (Mantel–Cox test) of the Kaplan–Meier survival curve analysis to patients with HCC, a low miRNA-148a-3p expression was observed to be significantly linked to the decreased overall survival (p < 0.001) (mean: 13.25; 95% CI: 10.99–15.51) (Fig. 3C). Additionally, the low miRNA 148a-3p expression was significantly associated with the decreased progression-free survival (p < 0.001) (mean: 9.53; 95% CI: 7.47–11.58) (Fig. 3D).

Discussion

Circulating miRNA can mirror cancer cell biology, in addition to being a feasible analytic test for the HCC patients, which gives it an advantage over tissue miRNA [32]. The miRNA-148a-3p and TGF-β1 biomarkers were the focus of this study because of their close association with fibrosis, which is one of the dynamic processes involved in liver cancer, as mentioned by Giannelli et al. (2011) [33].

First, we have observed lower levels of plasma miRNA-148a-3p and higher serum levels of TGF-β1 in HCC patients compared to their levels in cirrhotic patients and controls. Consistently, previous researchers have found decreased miRNA-148a-3p as the patients progressed from cirrhosis to HCC [[34], [35], [36]]. This stepwise decrease of the circulatory miRNA-148a-3p is a warning sign of liver cancer.

Altered homeostasis of some miRNA can direct a normal cell toward forming cancer [37]. As a tumor suppressor, miRNA-148a was found to be decreased in many tumors, including HCC [38]. Sengupta et al. (2018) and Long et al. (2014) have attributed this deficiency to miRNA-148a promoter hypermethylation [16,39]. miRNA-148a could hinder growth, metastasis, EMT, and angiogenesis of hepatoma cells through different regulatory targets such as ubiquitin-specific protease 4 (USP4), Met/Snail, Wnt, and TGFβ-SMAD2 signaling pathways [40].

Using ROC curve analysis, we demonstrated the discriminatory ability of this miRNA in predicting HCC with remarkable sensitivity and specificity. Herein, miRNA-148a-3p was superior to AFP in distinguishing HCC patients from cirrhotic patients. Han et al. (2019) [36] have concluded that plasma miRNA-148a could be used for HCC screening, especially in patients with reduced levels of AFP.

In this study, patients with HCC had higher serum values of TGF-β1 and higher fibrosis scores (FIB-4) than patients with cirrhosis. Consistent with our results, Lee et al. (2012) [41], Shehata et al. (2013) [42], and Kohla et al. (2017) [43] have found higher serum levels of TGF-β1 in HCC patients than those in cirrhotic patients and controls. Giannelli et al. (2011) have stated that TGF-β induces a signaling pathway that initiates hepatic fibrosis and cirrhotic changes ending with HCC [33], meaning that the profibrogenic TGF-β1 has a promoting role in cancer growth and development [17]. Depending on the stage of the tumor, TGF-β can be either beneficial or harmful. In the early stages, it offers a tumor inhibitory role, whereas in the late stages, it promotes genomic instability, EMT, angiogenesis, immune evasion, cell migration, and metastasis [44].

In this study, a higher fibrosis score (FIB-4) was found in patients HCC compared to patients with cirrhosis. Upon HCV infection, hepatocytes secrete TGF-β and then stimulate hepatic stellate cells. Both secrete fibrogenic effectors leading to HCV-advanced fibrosis. Moreover, TGF-β continues to trigger an immune reaction maintaining the inflammation that ultimately leads to progressive fibrosis and liver cancer [33]. Zhu and his colleagues have discovered that the regulatory circuit consisting of H19/miR-148a/USP4 can induce hepatic fibrosis by promoting TGF-β signaling cascade in hepatocytes and hepatic stellate cells [45].

Unrestrained metastasis is a major contributing factor to the bad prognosis of HCC [46]. Therefore, identifying and understanding markers of HCC metastasis could ameliorate tumor outcomes, thus improving the patient's quality of life. In this analysis, we noticed that lower median levels of miRNA-148a-3p and higher levels of TGF-β1 were related to some aggressive tumor phenotypes, namely, distant metastasis, multinodular lesions, advanced TNM stage, and BCLC score (C). Moreover, we reported for the first time a negative correlation between miRNA-148a-3p and FIB-4 score and TGF-β1, indicating their potential in the progression of the carcinogenic process and their possible role in fibrosis. Heo et al. (2014) [47] have found that the underexpression of miRNA-148a concomitant with overexpression of its target USP4 could contribute to the deterioration of HCC by stimulating the TGF-β1 signaling cascades.

The journey of a tumor cell from the site of origin to settle in a distant destination is initiated by EMT [48], which is a fundamental step in tumor invasion and metastasis [49]. TGF-β can induce EMT in cancer by inhibiting genes encoding the epithelial protein, E-cadherin, and stimulation of genes encoding the mesenchymal proteins, N-cadherin, and vimentin [50].

In this study, we discovered that miRNA-148a-3p can be used not only to effectively distinguish patients with HCC from patients with cirrhosis and controls but also to distinguish HCC metastatic patients from nonmetastatic patients.

The results regarding the association of these markers with signs of tumor aggressiveness were supported by several previous studies. Yan et al. (2014) [51] have reported low expression of miRNA-148a in metastatic and poorly differentiated HCC tissues. They have concluded that miRNA-148a suppresses metastasis by preventing EMT and cancer stem cell-like properties by targeting the Wnt regulatory pathway. Heo et al. (2014) [47] have stated that the deficient miRNA-148a was related to the high levels of AFP, advanced TNM stage, poor overall survival, and the recurrence-free survival rate in HCC patients. Similarly, Pan et al. (2014) [34] have confirmed the correlation with TNM stage, metastasis, capsular infiltration, and numbers of tumor nodes.

Survival analysis is a major tool used for the estimation of a patient's prognosis [52]. By applying the Kaplan–Meier survival analysis, this study showed a significant relationship between the reduced median values of miRNA-148a-3p and poor overall survival and progression-free survival among HCC patients. Moreover, the univariate analysis revealed that reduced expression of miRNA-148a-3p, high TGF-β1 level, multiple tumors, advanced TNM stage (III + IV), and BLCC score (C) were correlated with worse survival outcomes. Using Cox regression analysis, we identified that the low level of plasma miRNA-148a-3p was an independent prognostic marker predicting poor overall survival. These results agreed with previous reports [32,35,53]. In terms of the relationship between advanced stage and poorly differentiated HCC tumors, Cheng et al. (2017) [35] have also demonstrated that the low expression of tissue miR-148a3p/miR-148a5p had a significantly worse survival than the high expression, implying its involvement in the pathogenesis of HCC. Futhermore, Song et al. (2019) [53] have concluded that the downregulated miRNA-148a-3p was correlated with gross vascular invasion and poor survival of patients after hepatic resection for HCC.

Conclusion

This study showed a significant inverse relationship between miRNA-148a-3p and both TGF-β1 and FIB4. Moreover, the miRNA-148a-3p exhibited good diagnostic and prognostic power. Taken together, the plasma miRNA-148a-3p and TGF-β1 are valuable biomarkers of potential clinical significance in HCC.

Funding

There was no funding for this study.

Data availability

The data are available from the authors on reasonable request.

Declaration of competing interest

All authors declare no conflicts of interest.