Background
The end stage complications of liver disease including cirrhosis, and portal hypertension are related to advanced fibrosis [
1]. Management decisions in patients with chronic liver diseases often depend upon the stage of liver fibrosis. Hence, the accurate estimation of fibrosis is important for the prevention of the subsequent complications. For 60 years, liver biopsy has been regarded as a gold standard diagnostic method for assessing liver fibrosis. Despite its longstanding utility, liver biopsy has some disadvantages which include its invasive nature [
2], expense, as well as inter observer variability. It also has some negative features such as resistance of patients to undergo liver biopsy due to the discomfort [
3], possible complications [
2], and sampling error due to inadequate liver specimen length or fragmentation [
4,
5]. Therefore, alternative and accurate non invasive means to estimate fibrosis are needed.
Multiple hepatic fibrosis markers utilizing either simple blood tests or measurements of components of the extracellular matrix (ECM) have been proposed, and some are in clinical use. Imbert-Bismut and colleagues [
6] proposed an index combining 5 variables (bilirubin, gamma-glutamyltranspeptidase (γ-GT), haptoglobin, alpha 2-macroglobulin, and apoliprotein A1). The index ranges from 0-1 and currently is marketed as the
FibroTest. Subsequently, other hepatic fibrosis markers have been proposed such as the Forns index [
7], the AST/platelet ratio index (APRI) [
8], and the European or Enhanced Liver Fibrosis Index [
9]. These tests generally separate mild from significant fibrosis or mild/moderate from advanced fibrosis. Apart from Fibrotest, they usually have 2 cut offs with an indeterminate zone, where the accuracy is less. They can also be combined in decisional algorithms [
10,
11]. Another non invasive tool is transient elastography, either alone or combined with a hepatic fibrosis marker such as the Fibrotest [
11]. Other investigators have suggested for best accuracy combining both a simple index and an ECM index [
12,
13]. Such approach expands the number of investigations and increases the cost required to stage the degree of fibrosis.
The majority of publications have been concerned with fibrosis among HCV infected patients. Also, few studies have monitored the regression of drug-mediated fibrosis. This triggered our interest to establish a non-invasive index to score the liver fibrosis in mice models induced by injury from a hepatotoxin (thioacetamide, TAA), chronic intake of alcohol or Schistosomal infection. Moreover, we employed the index to monitor the reversibility of fibrosis in mice treated with silymarin and/or praziquantel (PZQ). The index includes four serum biochemical markers (3 liver-specific and one oxidative stress) combined in a logarithmic formula derived from values of both tested and normal control mice.
Discussion
Without diminishing the importance of histological assessment of liver fibrosis, non-invasive indices are moving towards accuracy and reliability. Previously proposed panels have included 2 to 7 markers. Increasing the number of variables may improve the accuracy, however it also increases the complexity and cost of the index. Previously reported indices included one (or more) variables from four categories. These categories include: (i) ECM proteins and enzymes involved in scar production, degradation, or hepatic stellate cells (HSC) activation, such as laminin, collagen, lysyl oxidase, prolyl hydroxylase, lysyl hydroxylase hyalauronic acid, or tissue inhibitors of metalloproteinases-1 (TIMP-1); (ii) Liver specific markers (bilirubin, transaminases or globulin); (iii) Haematological variables (platelets or PT); and (iv) Conjugated or derived lipids (apolipoprotein and cholesterol).
This work adds one more predictive non-invasive index that combines four simple, cheap and informative biochemical serum variables. The panel consists of an average number of variables (four), including three liver-specific, and one oxidative stress marker, simultaneously assessed in both normal (hypothetical standard) and tested samples. The aim was to formulate an index of liver fibrosis, which is experimentally induced with three different methods including: exposure to TAA, chronic alcohol intake and, for the first time, infection with S. mansoni, which predominantly exists in the Nile delta and upper-Egypt. Because the study was performed on mice, and due to the unavailability of animal model for HBV or HCV infection, viral hepatitis-induced liver fibrosis was not included. Also, the experiment was designed to non-invasively score the fibrotic changes during the compensated stages of fibrosis, where animals were maintained in the precirrhotic stages (< F4) to monitor the drug-mediated resolution of fibrosis.
Occupational TAA induced liver fibrosis in man is quite uncommon compared to alcohol, viral and
s. mansoni related chronic liver diseases. It is, however a potent hepatotoxin that is rapidly eliminated and can cause cumulative injury [
26]. In addition to the oxidative stress it generates, TAA decreases the level of some endogenous antioxidants such as α-tocopherol [
27]. Consequently, it is anticipated that supplementation of mice with silymarin (a known anti-oxidant), will compensate the depletion of the vitamin and challenge the oxidant effect of TAA. Treatment of fibrosis in
S. mansoni-infected mice, in contrast, required a specific eradication of the parasite with PZQ [
28], followed by treatment with silymarin. This pattern of progressive and regressive fibrogenesis process was used to score the degree of scar formation (or resolution) using a dual approach of histological and non-invasive biochemical index.
The results revealed that oxidative stress commonly developed in mice with liver fibrosis as indicated by the depletion of serum GSH and elevated hepatic MDA. In such conditions, HSCs are activated and excessively produce ECM proteins (2-fold increase of hepatic collagen was observed). The over production of collagen was accompanied with changes in liver markers and the corresponding histological stage. We believe that the changes in the level (or the activity) of ECM proteins and enzymes, reported in some previous indices [
29], follows the oxidative stress-mediated HSC activation. Novitskiy and his co-workers [
30] found that treatment of HSC with ethanol led to the formation of ROS, activation of HSC and subsequently promotion of fibrogenesis. Thus, both GSH and hepatic MDA are able to monitor the initial steps rather than other ECM variables (collagens, HA, and other ECM proteins). Also, due to the complexity of ECM environment, inclusion of ECM variables may decrease the sensitivity of the index. Moreover, ECM proteins and enzymes are not exclusively liver-specific and may reflect impaired hepatic clearance [
31]. Mizushima and co-workers [
32], for instance, validated prolyl hydroxylase and tissue inhibitor of metalloproteinase (TIMP) as a marker of pancreatic fibrosis. Moreover, ECM molecules are largely affected by the balance between enzymes that control scar formation, stabilization and degradation. In addition certain pathological states, like pulmonary fibrosis, are associated with increased levels of certain fibrosis markers [
33]. Previously we found that transglutaminase (liver derived enzyme known to stabilize the scar formation) did not show significant differences among patients with different grades of cirrhosis and HCC [
34]. Also, the high correlations (
Spearman) we obtained between GSH with both the histological stage and grade (-0.898 and -0.819, respectively) have nominated GSH as an integral member of the proposed index.
The remaining parameters included in the index are known to be directly involved in cellular pathogenesis and the synthetic capability of liver cells. ALT and AST have repeatedly used in hepatic fibrosis indices (individually, as a ratio or combined with other variables such as platelets [
8,
35,
36]. ALT, in particular, was included in some non-invasive panels to indicate the necro-inflammatory activity [
37]. Detection of fibrosis in HCV patients with normal transaminases [
38] may decrease their validity as fibrosis markers. Bilirubin and albumin (the major fraction of plasma proteins) are traditionally among the biochemical parameters used to monitor the severity of liver disease. γ-GT is used as a marker of liver damage by alcohol intake [
32]. This enhanced our predictive usefulness for the proposed index. The high correlations between liver histology (stage and grade) from one side and γ-GT (rather than ALT, AST, and ALP), bilirubin and T. Proteins from the other side (Table
4), have encouraged their combined involvement in the logarithmic formula. Liver dysfunction, leads to a decrease of the level of two of the selected markers (GSH and TP) and an increase the others (bilirubin and γ-GT). This predictive pattern was formulated by the mutual reversing the test/control ratio in the logarithmic formula.
The ground base is achieved in normal liver, where the ratio will be (or near to) "1". Accordingly, the index starts with "0", which corresponds to normal histological pattern. The highest scores (2, 1.5 and 1.8) matched the histological stages F3, F2 and F3 in groups II, V and X. A marked decrease in fibrosis index was observed in treated groups with index values ranging from 0.1 to 0.2.
Histologically, the highest fibrotic stage was observed in mice in groups II, V and X (Figure
2). Different treatments (with silymarin or PZQ), however, led to regression of the fibrotic stages to mild or moderate fibrosis in TAA groups or complete resolution of fibrosis in schistosomiasis and alcohol-induced fibrosis. The index was comparable to histological anaysis both following initial treatments with alcohol, TAA, and
s. mansoni as well as following regression of fibrosis after treatment with silymarin and PZQ. High correlations were observed between the score and both grade and stage. Also, the values of AUROC inhances the potential usefulness of the index (Figure
4 and Tables
6). Within groups the index is correlated well with the corresponding histological stages, where in TAA,
S.mansoni and alcoholism groups (
r = 0.97, 0.84 and 0.97, respectively). Also, the data prospectively obtained from the validation set enhances the index reliability, where it matched the pathological stages and grades, which were independently assessed.
This index resulted in higher AUROC values and better identification of individual stages of fibrosis than the indices that are currently being used clinically in man. This could partly be because compared with these indices, it was studied in a more controlled and predictable environment over a shorter term. Nevertheless, its assessment in chronic liver disease in man may be warranted.
Competing interests
The data presented in this study is not influenced by our personal or financial relationship with other people or organizations. Also there are no financial or non financial competing interests that may cause embarrassment after the publication of the manuscript. 'The authors declare that they have no competing interests.
Authors' contributions
MH: designed the study, generated and analyzed the data of the biochemical investigations performed in the study, put the mathematical formula used as an fibrosis index, prepared the manuscript for publication, supervised the animal care, treatments and he is the corresponding author. IS: carried out the infection of mice with schistosoma, performed the histological analysis and drafted the paragraphs concerned with the methodology of this method. AB: participated in the study design and coordination, participated in drafting the initial version of the manuscript. DB: Carried out the animals care, induced fibrosis in mice with TAA and ethanol, preformed animal treatment, sacrificing, sample collection, validated the proposed index in a separate experiment. NS: He critically revised the manuscript and largely participated in responding to the reviewer's concerns. All authors read and approved the final manuscript.