Performance of diagnostic biomarkers in predicting liver fibrosis among hepatitis C virus-infected Egyptian children

The aim of the present study was to identify specific markers that mirror liver fibrosis progression as an alter-native to biopsy when biopsy is contraindicated, especially in children. After liver biopsies were performed, serum samples from 30 hepatitis C virus (HCV) paediatric patients (8-14 years) were analysed and compared with samples from 30 healthy subjects. All subjects were tested for the presence of serum anti-HCV antibodies. Direct biomarkers for liver fibrosis, including transforming growth factor-β1, tissue inhibitor of matrix metalloproteinase-1 (TIMP-1), hyaluronic acid (HA), procollagen type III amino-terminal peptide (PIIINP) and osteopontin (OPN), were measured. The indirect biomarkers aspartate and alanine aminotransferases, albumin and bilirubin were also tested. The results revealed a significant increase in the serum marker levels in HCV-infected children compared with the healthy group, whereas albumin levels exhibited a significant decrease. Significantly higher levels of PIIINP, TIMP-1, OPN and HA were detected in HCV-infected children with moderate to severe fibrosis compared with children with mild fibrosis (p < 0.05). The diagnostic accuracy of these direct biomarkers, represented by sensitivity, specificity and positive predictive value, emphasises the utility of PIIINP, TIMP-1, OPN and HA as indicators of liver fibrosis among HCV-infected children.

These biomarkers have been analysed individually and in combination to assess the severity and progression of hepatic fibrosis and to follow-up changes related to viral treatment (Saitou et al. 2005, Sanvisens et al. 2009, Parkes et al. 2011).
The present work aimed to evaluate certain serum biomarkers in children that contribute to liver fibrosis secondary to HCV infection.The evaluation was performed by measuring the levels of the pro-fibrogenic cytokine TGF-β1, matrix deposition markers [hyaluronic acid (HA), type III procollagen amino-terminal peptide (PIIINP) and tissue inhibitor of matrix metalloproteinase (MMP)-1 (TIMP-1)] and the anti-apoptotic and cell activation factor OPN.The work was extended to measure liver function indices [aspartate (AST) and alanine (ALT) aminotransferases, albumin and bilirubin] that correlate with liver fibrosis in chronic HCV infection.

SUBJECTS, MATERIALS AND METHODS
Ethics -The study protocol was performed according to the guidelines of the Medical Ethical Committee of National Research Centre, Cairo, Egypt.The study was also performed in accordance with the Helsinki Declaration of 1975, as revised in 1983.HCV samples were collected from the National Hepatology and Tropical Medicine Research Institute from March 2011-June 2012, whereas control samples were collected from Abu Elreash Children's Hospital, Cairo University.Written informed consent was obtained from all the parents of the paediatric patients after the nature of the procedure was fully explained.
Patients and samples -Thirty paediatric patients with chronic HCV infection (17 male, 13 female; range of ages at biopsy, 8-14 years; mean age, 10.97 ± 2.11 years) were enrolled in the present study.Thirty healthy children (7-13 years) served as the control group.Patients were selected if they had no other causes of liver disease, autoimmune or metabolic disorders, HCC or co-infection with hepatitis B virus and/or human immunodeficiency virus.The patients were also treatment naïve.Serum samples were obtained at the time of biopsy and stored frozen at -80ºC for further determination of the selected parameters.All experiments were performed in duplicate.
Histopathological examination -Formalin-fixed, paraffin-embedded liver biopsies were used for histological analysis.Histological sections were blindly evaluated by two independent pathologists.Fibrosis staging was semiquantitatively assessed according to the METAVIR system (Theise et al. 2007).Patients with liver fibrosis were classified into two subgroups according to the severity of fibrosis: group 1 (with liver fibrosis stages < 2; nonsignificant fibrosis; 9 patients) and group 2 (with liver fibrosis stages ≥ 2; significant fibrosis; 21 patients).
Biochemical assays -Diagnosis was based on the presence of anti-HCV antibodies in the serum, which were detected by radioimmunoassay (Lumipulse II HBsAg; Fujirebio Co Inc, Tokyo, Japan).The detection threshold was 615 IU (3,200 copies)/mL.
Serum TGF-β1 and TIMP-1 levels were determined by the quantitative sandwich enzyme immunoassay technique (Quantikine, R&D Systems, Inc, MN, USA).The serum concentration of each marker was determined from constructed standard curves.The serum TGF-β1 level was expressed as pg/mL and the TIMP-1 level was expressed as ng/mL.
Serum HA levels were assessed using an ELISA kit.The HA test kit is an enzyme-linked binding protein assay that uses a capture molecule known as HA-binding protein (HABP) and an enzyme-conjugated version of HABP.HA levels in patient and control samples were determined from a constructed reference curve and expressed as ng/mL.
The levels of PIIINP were measured using a competitive radioimmunoassay technique (UNniQ, Orion Diagnostica).Concentrations of PIIINP (µg/L) were obtained from a calibration curve.
Serum concentrations of OPN (ng/mL) were measured by capture ELISA according to the manufacturer's protocol (R&D, San Diego, CA, USA).The optical density was measured at 450 nm using a microplate reader (Thermo-Lab Systems).The I-smart program was used to create a regression curve.
Serum AST and ALT levels (IU/L) were determined by the method of Gella et al. (1985), in which the transfer of an amino group from AST or ALT forms oxaloacetate or pyruvate, respectively.The developed colour was measured at 520 nm.
Serum albumin levels were detected using a human serum EIA kit (Cayman Chemical Co, Ann Arbor, MI, USA).Concentrations of albumin (mg/dL) were obtained from a calibration curve.
Total bilirubin (mg/dL) was measured using the method of Doumas et al. (1973), in which bilirubin reacts with diazotised sulphanilic acid in the presence of caffeine, resulting in an azo-pigment product.The developed colour was measured at 546 nm.
Statistical analysis -Statistical analysis was performed using the Statistical Package for the Social Sciences version 16 for Windows.Numerical data were presented as the mean ± standard deviation.A comparison of variables was performed with a general linear model and Student's t test.The Mann-Whitney U test (a nonparametric test) was used to compare the numerical variables of direct biomarkers between the two groups with liver fibrosis.Statistical significance was assumed at p < 0.05.To assess the ability of the serum biomarkers to differentiate between the stages of liver fibrosis we calculated sensitivity (Se) and specificity (Sp) values for each marker.We determined the cut-off value for each parameter as the maximum value of the sum of the Se and Sp.The diagnostic accuracy was calculated based on Se, Sp, positive predictive value (PPV) and negative predictive value, considering significant liver fibrosis.

RESULTS
The results revealed that 30%, 23.30%, 33.30% and 13.30% of patients showed stages 1, 2, 3 and 4 liver fibrosis, respectively.Bile duct damage was observed in 76.70% of patients.In total, 46.7% of the patients had a history of maternal transmission of infection, whereas 30% had a history of transmission through blood trans-fusion.In total, 70% of the patients showed a significant stage of liver fibrosis (≥ 2) and 30% showed a non-significant stage of liver fibrosis (< 2).
Regarding the levels of direct biomarkers, significant increases in the HA, OPN, TIMP-1 and TGF-β1 levels were detected (p < 0.05) in HCV patients compared with controls, whereas an insignificant difference in PIIINP levels (p > 0.05) was detected (Table I).
As shown in Table II, the levels of indirect biomarkers (serum AST, ALT and total bilirubin) were highly significantly increased in the patient group compared with the control group (p < 0.05).In contrast, serum albumin levels were significantly decreased in the patient group compared with the control group (p < 0.05).
When comparing the stages of liver fibrosis among HCV patients, the levels of the direct biomarkers PII-INP, HA, OPN and TIMP-1 were significantly higher in patients with a fibrosis stage ≥ 2 than in children with a fibrosis stage < 2 (p < 0.05).No significant difference in TGF-β1 was detected between the two groups (p > 0.05) (Table III).
Table IV shows the diagnostic accuracy of the direct serum marker measurements in HCV patients with nonsignificant (F < 2) or significant (F ≥ 2) fibrosis.The cut-off value of PIIINP was 11.21 µg/L, with 48.55% Se, 100% Sp and a PPV of 100%.At the cut-off value of HA (24.69 ng/mL), 42% Se, 93.88% Sp and 92.54% PPV were recorded.The cut-off value of OPN was 101.56 ng/ mL, with a Se, Sp and PPV of 97.67%, 100% and 100%, respectively (Table IV).The cut-off value of TIMP-1 was 102.47 ng/mL, with a Se, Sp and PPV of 90.33%, 45.70% and 75.76%, respectively.The cut-off value of TGF-β1 (6.53 pg/mL) corresponded with 85.70% Se, 42.70% Sp and 71% PPV.
The diagnostic accuracy of the indirect serum markers in HCV patients (2 > F ≥ 2) is shown in Table V.The Se and Sp of AST were 96.90% and 45.12%, respectively, and ALT yielded values of 84.24% and 23.23%, respectively.The cut-off values of AST and ALT were 18 IU/L and 20 IU/L, respectively, and their PPVs were 99.30% and 86.43%, respectively.Therefore, AST may be used as an indicator of liver fibrosis rather than ALT.The Se and Sp of albumin were 95% and 67.40%, respectively, and total bilirubin yielded values of 85.20% and 68.10%, respectively.The PPVs of albumin and total bilirubin were 70.10% and 67.40%, respectively.Therefore, both albumin and bilirubin appeared to be insufficient indicators of liver fibrosis.

DISCUSSION
Liver biopsy is regarded as the gold standard for the assessment of patients with liver fibrosis (Bianchi 2001).However, liver biopsy is an invasive procedure with potential complications and sampling error can result in substantial misdiagnosis and staging inaccuracies.
Recently, biochemical parameters were found to be good indicators of liver fibrosis, especially in children (Valva et al. 2011).Serum markers of liver fibrosis are divided into two categories: direct markers, which reflect ECM turnover, and indirect markers, which reflect alterations in hepatic function, but not ECM metabolism (Grigorescu 2006).
One of these direct markers is PIIINP, which can be used as a measure of matrix deposition (Grigorescu 2006).Walsh et al. (1999a) found a significant correlation between serum PIIINP levels and histological changes such as fibrosis, periportal necrosis and an altered histological activity index.In addition, Fabris et al. (1997) showed a correlation between PIIINP levels and the stages of hepatic fibrosis in alcoholic liver disease, viral hepatitis and primary biliary cirrhosis.In parallel with these observations, our results showed a significant increase in PIIINP levels (by 26.28%) in paediatric HCV patients compared with healthy subjects.PIIINP also showed a statistical association with more severe (highstage) fibrosis.The cut-off value of PIIINP was 11.21 µg/L, with 48.55% Se and 100% Sp.The PPV was 100%, which emphasises the utility of PIIINP as an indicator of liver fibrosis and reduces the need for biopsy for fibrosis detection in HCV-infected children.The second matrix deposition marker is HA.HA is a glycosaminoglycan and a component of the ECM synthesised by hepatic stellate cells (HSCs) (Grigorescu 2006).Under normal circumstances, the endothelial cells of the liver sinusoid are the sites of HA uptake and degradation (Eriksson et al. 1983).Increased levels of HA are due to a decrease in hepatic removal, increased production or both (Grigorescu 2006).Guéchot et al. (1995) postulated high levels of serum HA in patients with liver diseases of different aetiologies and particularly in patients with cirrhosis.Serum levels of HA have been shown to be related not only to the stage of fibrosis (Guéchot et al. 1995), but also to the degree of necroinflammation (Murawaki et al. 1995).Korner et al. (1996) added that the HA concentration has a good prognostic value for complications of liver cirrhosis: hepatic encephalopathy stages III and IV, refractory ascites and portal vein thrombosis.In agreement with these observations, the present study showed a significant increase in HA levels in paediatric HCV patients (56.56% increase) compared with healthy subjects.Additionally, HA was associated with more severe (high-stage) fibrosis.At a cut-off value of 24.69 ng/mL, HA yielded a Se of 42% and a Sp of 93.88%.The PPV reached 92.54%, which confirms the utility of HA for the identification of fibrosis, thus reducing the need for biopsy in these patients.
OPN is an ECM protein that is considered as the third marker for matrix deposition (Pan et al. 2003).Abu El Makarem et al. ( 2011) found a significant elevation in OPN levels in patients with HCV-related HCC.The same authors found that OPN levels are a potential diagnostic marker for HCC, especially among high-risk groups of patients.Shang et al. (2012) monitored serum OPN levels for one year before the diagnosis of HCVrelated HCC patients.Libra et al. (2005) postulated that OPN might play a role in lymphomagenesis, particularly in the context of HCV infection.In agreement with these results, the present study revealed a significant elevation in OPN levels (723.42%increase) that was significantly associated with the developmental stage of fibrosis.The results revealed that OPN is a good indicator of liver fibrosis associated with HCV.
Regarding the markers associated with matrix degradation (metalloproteinases, MMPs and their inhibitors), MMPs show substrate Sp for interstitial collagen, whereas their tissue inhibitors (TIMPs) can irreversibly bind to and inactivate MMPs (Friedman 1999).Although several investigators have discussed the roles of MMPs in liver disease, most focused on their participation in cancer development and metastasis, whereas no study has assessed the role of MMPs in carriers of hepatitis viruses (Youness et al. 2013).Excess production of TIMPs relative to MMPs may be an important factor in the progression of liver fibrosis (Iredal et al. 1992).TIMPs are also observed in late-stage fibrosis, but not in the mild  negatives; e: is the minimal value of each parameter.Diagnostic accuracy test was done in hepatitis C virus (HCV) patients with non-significant (F < 2) and significant (F ≥ 2) stages of liver fibrosis.HA: hyaluronic acid; NPV: negative predictive value; OPN: osteopontin; PIIINP: type III procollagen aminoterminal peptide; PPV: positive predictive value; Se: sensitivity; Sp: specificity; TGF-β1: transforming growth factor-β1; TIMP-1: tissue inhibitor of matrix metalloproteinase.
stage (Walsh et al. 1999b).Additionally, Boeker et al. (2002) found a strong correlation between TIMP-1 levels and histological inflammatory scores and the serum level of TIMP-1 is associated with active inflammatory activity (Ueno et al. 1996).In line with the above observations, our results revealed a significant increase in TIMP-1 levels (82.23% increase) compared with levels healthy children.In addition, TIMP-1 was significantly associated with high-stage fibrosis.The cut-off value of TIMP-1 was 102.47 ng/mL, with Se and Sp reaching 90.33% and 45.70%, respectively.The PPV of TIMP-1 was 75.76%, which indicated that this biomarker is a valuable tool for the detection of liver fibrosis.
In the present study, TGF-β1 levels were significantly increased in paediatric HCV patients (357.23%increase) compared with the control group.TGF-β1 is a cytokine associated with hepatic fibrosis and is a homodimeric polypeptide that is secreted in an inactive form that requires activation.In hepatic pathology, TGF-β1 is the most important stimulus for the production of ECM by HSCs (Sasaki et al. 1992) and is also an inhibitor of hepatocyte growth and proliferation (Nakamura et al. 1985).In liver biopsies from patients with chronic liver disease, TGF-β1 levels are correlated with type I collagen mRNA (Breitkopf et al. 2001).The value of serum TGF-β1 levels has certain limitations, including the potential contamination of samples by TGF-β1 from platelets, interference by plasmin activity in the plasma that increases the amount of TGF-β1, the binding of TGF-β1 to the ECM and vascular endothelium at sites of injury, sequestration by soluble proteins and the complicated clearance of TGF-β1.These factors explain why plasma levels of TGF-β1 are unlikely to be of diagnostic value (Breitkopf et al. 2001).In agreement with these observations and in parallel with our study, serum TGF-β1 levels showed no statistically significant differences among fibrosis stages in paediatric patients.The cut-off value of TGF-β1 was 6.53 pg/mL, with a Se of 85.70%.Due to the low Sp (42.70%), the PPV of 71% and the possibility of contamination of serum TGF-β1 with other TGF-β1 forms, we cannot consider this protein as a marker for the diagnosis of liver fibrosis associated with HCV.Contradictory to our observation, certain studies showed a good correlation between the serum levels of TGF-β1 and the rate of fibrosis progression (Kanzler et al. 2001).
Although serum ALT levels generally reflect liver injury (Grigorescu 2006), the correlation between ALT levels, necroinflammation and the fibrosis score is poor, especially in chronic HCV infection (Grigorescu 2006).AST levels have a stronger correlation with hepatic fibrosis than ALT levels (Gordon et al. 2000).The increase in ALT levels is related to mitochondrial dysfunction and reduced clearance of AST by hepatic sinusoidal cells (Grigorescu 2006).Kruger et al. (2011) confirmed that ALT could not differentiate between the stages of disease or the grades of fibrosis.In agreement with the present study, AST and ALT levels were significantly increased in the HCV group (401.58% and 263.76% increases, respectively).In addition, AST and ALT values showed no statistical correlation with fibrosis stages.The Se and Sp of AST were 96.90% and 45.12%, respectively, and ALT yielded values of 84.24% and 23.23%, respectively.The cut-off values of AST and ALT were 18 IU/L and 20 IU/L, respectively, and their PPVs were 99.30% and 86.43%, respectively.Therefore, AST may be used as an indicator of liver fibrosis rather than ALT.
A significant increase in total bilirubin levels in HCV-infected children (598.76%increase) vs. healthy individuals was recorded.In contrast, albumin levels showed a significant (10.86%) decrease.Albumin and total bilirubin showed no statistical correlation with fibrosis stage.The Se, Sp and PPVs of both albumin and total bilirubin appeared to be insufficient for detecting liver fibrosis.These observations were in accordance with the findings of Hongbo et al. (2007), who stated that routine serum markers, including AST, albumin and total bilirubin, were independent predictors of significant fibrosis.
In conclusion, PIIINP, HA, OPN and TIMP-1 served as clinically useful biomarkers for predicting liver fibrosis in HCV-infected children, whereas AST, ALT, albumin and bilirubin appeared to be insufficient markers for liver damage detection.Due to the relatively limited number of cases in this study, further studies are needed to accurately determine whether to use these biomarkers instead of biopsies, as the small sample size makes it difficult to evaluate the validity of the serum markers under investigation.