Liver extracellular matrix in health and disease



Liver fibrosis is the hallmark of every chronic liver disease. It is also the major factor of morbidity and mortality due to the development of cirrhosis and its complications including hepatocellular carcinoma. But even at the beginning of the process of liver fibrosis and due to the strategic position of the extracellular matrix at the interface between blood flow and epithelial compartment, any quantitative or qualitative modification of extracellular matrix will rapidly affect structure and function of the liver. The development of several animal models of liver fibrosis as well as isolation and cultivation of hepatic stellate cells, the major fibrogenic cell type in the liver, led to the gathering of recent knowledge on the mechanism of liver fibrosis. Activation of hepatic stellate cells is a key event in this process and many details on this finely tuned mechanism are now available. In addition to these experimental data, experience from chronic hepatitis C now allows the development of new concepts and perspectives such as liver fibrosis regression and antifibrotic therapies. Copyright © 2003 John Wiley & Sons, Ltd.


Although normal liver is primarily made up of parenchymal cells, and fibrous tissue is quantitatively very limited, extracellular matrix (ECM) is of major importance both in liver physiology and pathology 1–3. Any modification in ECM, whether quantitative, topographic, or qualitative, has a direct effect on liver functions. Despite the diversity of agents which are noxious toward the liver (viruses, drugs, alcohol, autoantibodies, iron, etc), the liver responds via a very standardized pathway which is characterized by liver fibrogenesis. Fibrogenesis, a component of tissue repair, is a dynamic mechanism defined by synthesis of ECM components. When the aggression is arrested, spontaneously or with the assistance of therapy, then the mechanism of tissue repair including tissue remodelling and epithelial regeneration will lead to restitution ad integrum of liver parenchyma. This succession of events is observed, for example, during acute viral hepatitis when immunological defences are able to clear the virus. Under these conditions, although fibrogenesis occurs, fibrosis, defined as the detectable deposit of ECM within the liver, usually is not observed. The tremendous capacity for regeneration of the liver explains why a favourable issue may occur even after more than 80% of liver cells have been destroyed.

In the case of persistence of the noxious agent with prolonged liver injury, liver fibrosis becomes obvious. In humans, fibrosis, with its endpoint cirrhosis, is the hallmark of any chronic liver disease and is the major cause of morbidity and mortality.

Owing to the development of experimental models such as CCl4 intoxication and bile duct ligation, a great deal of knowledge has been gathered in the past 20 years regarding the composition, regulation, and dynamics of liver fibrosis 3, 4. Such progress is now enabling the development of new antifibrotic therapeutic approaches, and is leading to consideration of regression of liver fibrosis, or even cirrhosis, as possible events 5, 6.

Composition, function, and distributionof ECM in normal and fibrotic liver

ECM is a very limited compartment within the normal liver. It comprises less than 3% of the relative area on a normal liver section 7. ECM refers to the array of various macromolecules that comprise the scaffolding of the liver. In addition to Glisson's capsule, ECM is restricted, in normal liver, to portal tracts, sinusoid walls, and central veins. At any site, ECM is part of the frontier between the blood flow and the parenchyma. Such a strategic situation is responsible for the immediate and multiple deleterious consequences of any ECM modifications, whether qualitative or quantitative.

The most frequently found proteins in liver ECM are collagens, with the most abundant being types I, III, IV, and V collagen, although other isoforms including the recently described type XVIII, the precursor of endostatin, are present 8, 9. Each isotype differs in its localization and physical function within the liver. While types I, III, and V, the major constituents of fibrillar collagen, are confined mainly to the portal tract and central vein wall, type IV collagen, in association with laminin and entactin–nidogen, takes part in the formation of a low-density, basement membrane-like material along the sinusoid wall 10. In this position, an electron-dense basement membrane is usually absent although most components can be detected individually 11. The low density of this basement membrane-like structure is critical for allowing easy diffusion between blood and liver cells and for maintaining the differentiated function of neighbouring liver cells such as hepatocytes and the various sinusoidal cells 12. Other major components of the liver ECM are glycoproteins such as laminin, fibronectin, tenascin, nidogen, and SPARC, among others 2. Proteoglycan includes heparan, dermatan, chondroitin sulphate, perlecan, hyaluronic acid, biglycan, and decorin 13. The major function of ECM remains the mechanical coherence and resistance of the liver, but liver ECM also has a role in several major biological functions such as cell proliferation, migration, differentiation, and gene expression.

In fibrotic liver, the ECM components are similar to those present in normal liver but are quantitatively increased (three- to five-fold increase in ECM) 3. In this context, a redistribution of the relative amounts of ECM components occurs. Although fibrosis is a major biological event per se, it is closely associated, in the liver, with other important mechanisms such as architectural distortion, liver cell regeneration, and vascular redistribution. In addition to fibrous deposits, these mechanisms contribute to severe consequences for liver function.

Changes in the perisinusoidal space, the area lying between endothelial cells and the vascular face of hepatocytes (Disse's space), is a common early complication observed during liver fibrosis in both human disease and experimental models 10, 11. It includes different cellular and extracellular modifications such as the accumulation of collagen types I and III and the constitution of a framework of ‘fibrillar collagen’ that replaces the basement membrane material. These extracellular changes are associated with modifications in the phenotype of sinusoidal cells including the loss of pores of endothelial cells 14. These events resume the process termed ‘sinusoid capillarization’ 15. Although, in normal liver, the perisinusoidal space, due to the scarcity of ECM and the porous constitution of endothelial cells, allows free exchanges between liver cells and blood flow, sinusoid capillarization will strongly impair these exchanges. Furthermore, and due to direct connections between cells and their milieu, any quantitative or qualitative modification of the ECM microenvironment will influence cell functions such as hepatocyte function deterioration and changes in the hepatic stellate cell (HSC) phenotype 12, 16. These changes illustrate the major role of ECM in the liver, not only as a scaffold for liver architecture, but also as a continuous network between cells that allows, via cell receptors, the continuous exchange of signals between cells. The signalling function of ECM is also exemplified by the storage, within ECM, of numerous growth factors, hormone, enzymes, and cytokines in their inactive forms that, upon stimulation, will activate and deliver messages that modify the cellular microenvironment 17.

During the process of liver fibrosis, the most obvious change for the pathologist is the expansion of ECM from the portal space or central vein. The preferential site for fibrosis to start is dictated by the mechanism responsible of fibrosis induction and is closely linked to aetiology (central fibrosis for vascular or alcohol fibrosis, portal fibrosis for viral, autoimmune or biliary diseases). In this context, the role of resident fibroblasts located within the central vein wall or portal tracts is also of importance, although it has been less extensively studied 18. The stellate expansion of fibrous tissue around the portal tract or central vein leads to the development of fibrous connections (bridging fibrosis) from one vascular structure to another. A major deleterious consequence of these fibrous septa is not so much the expansion of ECM, but, rather the development of an angiogenic reaction closely associated with fibrogenesis with, as an endpoint, disturbance of the normal vascularization within the liver lobule 19.

At a more advanced stage, when most vascular spaces are interconnected, cirrhosis is constituted. Liver cell regeneration, usually associated with such annular fibrosis, is of varying importance and may even be totally absent. In this new organization, redistribution of incoming liver blood flow is observed with the blood supply derived mainly from branches of the hepatic artery (arterialization) in association with phenotypic modifications of sinusoid endothelial cells associated with capillarization 20.

Hepatic stellate cells: major partnersin liver fibrogenesis

Liver fibrogenesis is the common response of the liver to aggression. In this context, fibrocompetent cells will transcribe genes and produce the components of the liver ECM in a finely tuned, orchestrated pathway. Cells involved in this process are mainly (but not exclusively) HSCs (also known as Ito cells, lipocytes or fat-storing cells) 21, 22. Other cell types of the liver provide only a modest contribution to ECM production 23. Owing to the possibility of isolating and culturing HSCs in vitro 24, much data has been gathered in recent years regarding their function and regulation 25. The major role of other natural fibrogenic cells, and in particular portal and central fibroblasts, in the development of fibrosis and cirrhosis in human liver diseases must be kept in mind. Interestingly, a recent study showed that primary hepatic rat fibroblasts express some specific markers (fibullin) and are more resistant to apoptosis than activated HSCs 26.

HSCs are scarce, isolated cells located within the perisinusoidal space between the vascular face of hepatocytes and the endothelial cells 27. These cells are poorly visible under a light microscope (Figure 1). At the ultrastructural level and in the normal liver, HSCs are characterized by cytoplasmic lipidic vesicles and long cytoplasmic stellate processes that insinuate between the neighbouring liver cell plate and the adjacent subendothelial area 28. In the normal liver, HSCs constitute a small percentage (less than 10%) of total resident liver cells and are located at equal distance from one another (approximately 40 µm from nucleus to nucleus) 29. However, and despite their scarcity, their long cytoplasmic processes can cover the entire perisinusoidal area. Several functional roles have been assigned to HSCs in normal liver, such as metabolism and storage of retinoids 30, synthesis of hepatocyte growth factor and a pericyte-like role in regulation of the sinusoidal blood flow 31. Regarding this issue, it must also be underlined that autonomic nerve endings running within Disse's space enter into contact with HSCs. It has also been demonstrated that HSCs respond to α-adrenergic stimulation with intracytosolic calcium influx 32. Furthermore, several neural markers such as glial fibrillary acid protein (GFAP), nestin, an intermediate filament protein originally identified as a marker of neural stem cells, N-CAM, a central nervous system adhesion molecule, and dystroglycan, a transmembrane receptor typically present in the nerve–muscle junction, have been detected on the HSC cell membrane 33–35. These observations raise the issue of a possible neural origin with shared functions for HSCs. Finally, a certain degree of heterogeneity in HSCs within the normal liver exists according to their phenotype, their vitamin A content and their potential for ECM production 36.

Figure 1.

Hepatic stellate cell at the interface between sinusoid and hepatocyte (✶). Cytoplasm is lipid loaded and nuclei is push aside

Quiescent HSCs have a low proliferation rate, low fibrogenic activity, and no contractile property, but after any liver aggression these cells will progressively activate and shift their lipid storage phenotype towards a myofibroblastic-like phenotype in a reproductive temporal sequence. A similar activation process is observed when primary HSCs of human, rat, or mouse origin are kept in culture on a plastic support for a few days 37, 38. In this context, cells elongate, lose their lipid droplets, and develop myofilaments within their cytoplasm (Figure 2). The myofibroblastic phenotype is exemplified by de novo expression of α-smooth muscle actin and desmin (in rat HSC only), intermediate filaments characteristic of a smooth muscular phenotype 39, 40. Parallel to these morphological changes, HSC will gain new functions such as proliferation, migration, contractility, and protein synthesis. In association with the synthesis of most of the ECM components, activated HSCs produce mediators including growth factors. Most will act on HSC through an autocrine stimulatory loop. Furthermore, activated HSCs are able to produce specialized enzymes such as matrix metalloproteases (MMP), a family of zinc-dependent enzymes that can destroy the normal pericellular ECM before the synthesis of fibrous tissue (ECM remodelling) 41. Therefore, HSCs play a key role in the control of both fibrosis synthesis and destruction.

Figure 2.

Rat hepatic stellate cell. (A) One day after isolation, cells display a stellate shape with short cytoplasmic processes. (B) Six days after isolation. Cells spread on the surface and display a myofibroblastic phenotype with long cell processes

The mechanism of hepatic stellate cell activation has been divided into two successive chronological steps—initiation and perpetuation—although both are tightly associated 21, 25. They are finely tuned successive sequences regulated by several molecules, including chemokines and growth factors. The early step is associated with rapid changes in gene expression that render cells reactive to cytokines and growth factors. Factors that trigger initiation are derived from incoming or resident inflammatory cells (Kupffer cells), adjacent damaged epithelial or endothelial cells, and from subtle changes in the ECM environment. Most of these components act by a paracrine pathway of stimulation. After HSCs engage in the process and gain an activated phenotype, this phenotype will be maintained and amplified due to the perpetuation mechanism involving autocrine and paracrine stimulation by growth factors as well as ECM remodelling. In this setting, HSCs will gain several new functions such as multiplication, migration, contractility, production of several enzymes, growth factors, and receptors. HSCs will also be the major sources of ECM component production and the basis for fibrosis deposition. Reversion of HSC activation is one of the hypotheses associated with fibrosis reversion. In this context, the role of peroxisome proliferator activated receptor (PPAR) has recently been underlined. It has been shown that PPAR-γ and its ligand, but not PPAR-α, is able to induce reversal of HSC activation 42.

Cellular and molecular basis for HSC activation

Activation of HSC is the common pathway to liver fibrogenesis. Many different cytokines, growth factors, and other soluble mediators are involved in this complex mechanism that orchestrates the wound healing response in the liver 43. These mediators are either present locally and stored in an inactive form bound to the ECM, or are produced by different cell types during the acute phase of aggression as part of the inflammatory reaction. The in vitro model of HSC culture allowed evaluation of the effect of many different mediators upon several functional properties of HSC, although this simplistic model gives only a partial insight into integrated mechanisms occurring in vivo during liver fibrogenesis. In any case, some mediators seem to have a major function during HSC activation.

Lipid peroxidation

The role of lipid peroxidation as an initiating factor in HSC activation has been underlined both in vitro and in the context of several human liver diseases [44–46]. Sources of reactive oxygen species (ROS), including intermediate reactive metabolites, free radicals, and nitric oxides, have been detected at the initial phase of liver aggression 47. They are generated both by activated Kupffer cells, polymorphonuclear influx, and damaged hepatocytes. After polyunsatured fatty acid degradation of the cell membrane, ROS lead to the production of different aldehydic end-products such as malondialdehyde and 4-hydroxy-2,3-nonenal. These components have been identified in the course of different human liver diseases including ethanol exposure, iron overload, and chronic viral diseases 45, 46. The deleterious effects of these end-products are many: production of protein and DNA adducts, polymerization of proteins, enzyme inactivation, etc. Of paramount importance is the fact that these byproducts have been shown to stimulate HSC proliferation and procollagen I gene transcription by HSC in vitro 44, 48. Therefore, oxidative stress and lipid peroxidation might be one of the common pathways leading to early HSC initiation after liver injury.

Altered extracellular matrix

Among other factors involved in the early HSC initiation phase, modification of microenvironment such as de novo synthesis of a spliced variant of cellular fibronectin (EIIIA) by altered endothelial cells is of importance since this component is also able to stimulate hepatic stellate cell activation 49.

Transactivation factors

Many efforts have recently been devoted to the investigation of molecular mechanisms associated with regulation of gene transcription during early HSC activation. The roles of several transcription factors such as Sp1, c-myb, nuclear factor κB (NF-κB), STAT-1, and Zf9, a Kruppel-like factor zinc finger gene, have been elucidated 50, 51. Since most of these factors can transactivate several genes involved in ECM synthesis and degradation, a role for them in the early phase of liver fibrogenesis is highly probable.

The pathways of perpetuation of the activated HSC phenotype include the acquisition and maintenance of new functions such as proliferation, release of proinflammatory cytokines and chemokines, matrix degradation, and, of course, fibrogenesis. Most of these new functions are sustained by an autocrine loop characterized by production of several major mediators and enhancement of cell response to these factors through both upregulation of their own membrane receptor and enhancement of intracellular signalling.

Growth factors

Platelet-derived growth factor (PDGF), a dimer of two polypeptide chains, A and B, is the strongest mitogen for cultured HSCs 52. Both PDGF and its cognate receptor are upregulated during HSC activation. The homodimer PDGF-BB is the most potent isoform, in relation to the preferential expression of one of the PDGF receptors by activated HSC, the PDGF-receptor β 53. PDGF receptors belong to the family of tyrosine kinase receptors. Upon binding to its ligand and after dimerization, phosphorylation of its internal tyrosine residue activates several signalling pathways, leading to proliferation and migration of activated HSCs 54. Several other hepatic stellate cell mitogens have been identified, including FGF, VEGF, endothelin-1 (ET-1), epidermal growth factor (EGF), and insulin growth factor. Vasoactive substances also exert a mitogenic effect on activated HSCs, whereas vasodilators have the opposite effect. The most potent vasoconstrictors and mitogenic mediators are thrombin, arginin–vasopressin and angiotensin II 55, 56. Vasodilators with antimitogenic effects include prostaglandin E2 and nitric oxide (NO) 57.

Transforming growth factor beta1 (TGF-β1), a multifunctional growth factor, plays a major role in fibrogenesis, but only a limited role in HSC activation 58. Indeed, in an experimental model of liver fibrosis, TGF-β1 knockout mice display phenotypic modifications of HSC activation in the absence of any collagen synthesis. At the early phase, TGF-β originates both from platelets and resident or incoming inflammatory cells including Kupffer cells. Later on, activated HSC become the major source of TGF-β during the perpetuation phase. In the liver, TGF-β is also present locally in a latent form (LTGF-β) bound to the ECM 59 and can, at any moment, be locally activated toward an active profibrogenic form through modification of the microenvironment, ie local production of proteases such as thrombospondin, tissue plasminogen activator (tPA) and metalloproteinases (MMP) 60. TGF-β1, through TGF-β receptors, activates gene transcription and cell release of most ECM components: types I and III collagens, laminin, and fibronectin 61. Increased collagen mRNA is regulated both at the transcriptional and post-transcriptional level through an increase in mRNA stability. Coordinately, activation of HSCs increases the responsiveness to TGF-β through upregulation or de novo expression of the three TGF-β receptors (types I, II, and III). TGF-β receptors are specific serine–threonine membrane receptors and intracellular signalling involves the Smad molecule family pathway 62. Phosphorylated Smad 2 and 3 recruit Smad 4, translocate within the nucleus, and regulate gene transcription. Other components such as Smad 7 repress TGF-β biological activity. Connective tissue growth factor (CTGF), a cysteine-rich immediate early gene response, is produced by HSCs during liver fibrosis 63. It has been shown that CTGF, in response to TGF-β stimulation, is upregulated, and that this gene stimulates ECM gene transcription 64. Since a TGF-β responsive element has been characterized within the TGF-β promoter, CTGF might be a key intermediate that specifically drives the fibrogenesis pathway of TGF-β 65.

TGF-β can also interact directly with liver fibrosis via other functions. TGF-β can interact with degradation of newly synthesized ECM through synthesis of tissue inhibitor of metalloproteases (TIMP) but can also stimulate MMP synthesis 66. Finally, TGF-β favours the apoptosis of epithelial cells, including hepatocytes, impairing liver regeneration during liver fibrogenesis 67.

Many other mediators are implicated in liver fibrogenesis. Among those recently characterized, leptin, the product of the obese gene, has a direct stimulatory fibrogenic effect on HSCs. It is produced by HSCs that, upon activation, express the long form leptin receptor 68.

Endothelin-1 (ET-1) is the key contractile stimulus for activated HSC. While ET-1 is produced and upregulated in activated HSCs, its two receptors, ET-A and ET-B, are expressed in quiescent cells 69. Therefore ET-1 also acts in an autocrine loop of stimulation. Owing to its perisinusoidal situation, the contraction of HSC has a major impact on liver blood flow and modification of portal resistance during fibrosis and cirrhosis. ET-1 is in balance with nitric oxide (NO), a vasodilator that physiologically antagonizes ET-1 57. During liver fibrosis, both an ET-1 increase and a NO decrease shift the balance toward contractility 70.

HSCs also contribute to the formation of the inflammatory granuloma with synthesis and extracellular release of polymorphonuclear and macrophage–monocyte chemoattractants or differentiation factors such as monocyte chemotactic protein-1 (MCP-1) and macrophage colony-stimulating factor (M-CSF). M-CSF, a member of the CC class of chemokines, actively participates in the influx of peripheral monocytes and their differentiation towards phagocytic cells at the site of injury 71. Moreover, activated human HSC express receptors for a number of chemokines that activate various intracellular pathways involved in the inflammatory reaction 72. These mediators compete with anti-inflammatory cytokines such as Il-10, which are also produced by HSC 73. Il-10 might protect against liver fibrosis since Il-10 knockout mice have more severe hepatic fibrosis after CCl4 administration.

Secretion of cytokines by resident or incoming lymphocytes in the context of the inflammatory response also contributes to the regulation of fibrogenesis. According to their cytokine profiles, CD-4 T cells can promote either cell-mediated immunity, through secretion of IFN-γ, Il-2 or TNF (Th1 cells), or humoral immunity (Th2 cells) through IL4, IL5, or IL13. Experimental evidence suggests that different Th profiles might favour or limit fibrosis 74.

Besides cytokine and growth factors, ECM components can directly act on HSC differentiation. The shift in perisinusoidal ECM composition, characterized by replacement of a basement membrane into a fibrillar matrix, sustains HSC activation 3. Signalling to HSCs occurs through several specific cell membrane receptors such as integrins, discoidin domain receptors, and dystroglycan that connect certain ECM components to various HSC intracellular signalling pathways 34, 75, 76.

Diagnosis of liver fibrosis in human diseases

Complications of fibrosis with its endpoint cirrhosis are the major factors in morbidity and mortality in human chronic liver diseases. Therefore, detection of liver fibrosis is of major importance in the management and follow-up of these patients. Until now, liver biopsy has been the gold standard for fibrosis assessment. However, it is noteworthy that, despite its sensitivity, assessment of liver fibrosis with a biopsy has two major limitations: sampling variability and observer variation. Observer variation relies on differences in fibrosis assessment between pathologists reading the same biopsy and using semi-quantitative scores. Observer variation is highly dependent on the experience of the pathologist and the scoring system used. It is roughly inversely correlated with the complexity of the scoring system. Several works have addressed this point, and the general conclusion is that liver fibrosis is among the most reproducible histopathological features when well-defined semi-quantitative scores such as the METAVIR score or the Knodell index are used (Table 1) 77, 78. These results allow their use in clinical trials. To evaluate liver fibrosis, a more sophisticated and accurate approach is image analysis. This technique is easily applicable to liver fibrosis. It is very sensitive and roughly parallel to the semi-quantitative scores of fibrosis 79. It is note worthy, however, that semi-quantitative scores and image analysis do not assess fibrosis in exactly the same way. Whereas image analysis estimates only the amount of ECM deposited, the scoring system also takes into consideration architectural distortion and topographical distribution.

Table 1. METAVIR, Knodell and Ishak index for staging of liver fibrosis
  1. P-P; portal to portal; P-C, portal to central.

No fibrosis0No fibrosis0No fibrosis0
Portal fibrosis without septa1Fibrous portal expansion1Some portal tract fibrotic ± short fibrous septa1
Most portal tract fibrotic ± short fibrous septa2
Portal fibrosis with few septa2Bridging fibrosis (P-P or P-C)3Most portal tract fibrotic with Occasional P-P bridging3
Septal fibrosis3  Portal tract fibrotic with marked P-P and P-C bridging4
Marked P-P and/or P-C bridging with occasional nodules (incomplete cirrhosis)5
Cirrhosis4Cirrhosis4Cirrhosis (probable or definite)6

Owing to the heterogeneous distribution of fibrosis in the liver tissue, there is a risk of sampling error in fibrosis assessment in core biopsy. The risk increases with the decrease in the length of the biopsy and with the sophistication of the system analysis. Therefore, major efforts are now concentrating on the development of techniques that allow global estimation of the amount of fibrosis, such as imaging techniques and serum markers of fibrosis. As a matter of fact, although liver biopsy can be considered a safe procedure, it is associated with a significant morbidity precluding repeated biopsies in the follow-up of the evolution of a chronic liver disease. Development of non-invasive serum markers will enable repeated and safe fibrosis evaluation. Individual biomarkers derived from ECM molecules (hyaluronan, P-III-P, MMP, TIMP) 80, or composite index made by association of several markers related to liver function now allow a satisfactory prediction of the stage of fibrosis 81, 82. Development of new technological approaches such as proteomic analysis and powerful imaging techniques should lead to further progress.

Dynamics of liver fibrosis: progression and regression


Fibrosis is not static; it increases or decreases. Data has recently been gathered on the dynamics of fibrosis progression and regression. Hepatitis C viral infection, with its high rate of chronic hepatitis development with fibrosis, has provided a huge amount of material for studying such dynamics. Evidence is converging to demonstrate that in a given chronic disease the rate of fibrosis progression is highly variable from patient to patient. In large prospective studies, authors have demonstrated that patients can be divided into several groups according to the fibrosis progression rate. In hepatitis C, groups of rapid, intermediate, and slow fibrosers have been described who develop cirrhosis in an interval varying from 10 to over 50 years 83, 84. In an attempt to explain interindividual differences, epidemiological studies have identified several cofactors mainly related to the host that affect the rate of fibrosis progression. A deleterious effect has been noted for patients of the male sex or for those with an elevated age at the time of infection. Other cofactors related to the environment, such as associated alcohol consumption or HIV co-infection, have also been shown 84, 85. Interestingly, several gene polymorphisms, such those involving TGF-β and angiotensinogen genes promoter, have been associated with a higher risk of fibrosis development 86, 87. Despite intensive efforts, many other cofactors remain to be discovered.


Experimental and human evidence strongly suggests that regression of fibrosis is a possible event 88. In experimental models of fibrosis such as chronic carbon tetrachloride (CCl4) intoxication or bile duct ligation in rat, the arrest of fibrogenic stimulus induces spontaneous reversion of liver fibrosis with virtually normal liver histology as an endpoint (Figure 3) 89. If aggression persists, and when cirrhosis is constituted, regression becomes more uncertain, but also seems possible. Such regression has also been described recently in a subset of patients with different chronic liver diseases 90–92. All these cases of regression have in common the necessary withdrawal of the causative agent either because of natural defence mechanisms or because of treatment.

Figure 3.

Progression and regression of fibrosis in the CCl4 model of cirrhosis in rat (Picrosirius staining). (A) Normal rat liver. (B) Cirrhosis induced after repeated intraperitoneal injection of CCl4. (C) reversal of cirrhosis 3 weeks after toxic withdrawal

From a mechanistic perspective, several biological processes participate in fibrosis reversion.

ECM digestion

Since ECM components in the scarring process are highly stabilized and cross-linked molecules, they are insensitive to most human proteases. Only an expanded specific family of enzymes, the matrix metalloproteinases (MMP), can destroy ECM. This group is composed of different calcium- and zinc-dependent enzymes, each being specific for a group of ECM components either of collagenic or non-collagenic origin. Based on their substrate, MMPs fall into five broad categories: interstitial collagenases (MMP-1, -8), gelatinases (MMP-2, -9), stromelysins (MMP-3, -7, -10, -11), membrane type (MMP-14, -15, -16, -17, -24, -25), and metalloelastases (MMP-12) 93. HSC are the main source of metalloprotease synthesis during liver fibrosis 41, 94. Following HSC activation, these cells express virtually all the key components required for matrix degradation. Metalloprotease activity is regulated mainly at the post-translational level in order to target their activity to a specific region of the microenvironment. This activation depends on maturation from the proform to the active form through proteolytic cleavage under the control of the microenvironment 95. MMP activation is balanced by an inactivation mechanism, ie the binding, in a very specific ratio, to a group of inhibitors known as tissue inhibitors of metalloproteases or TIMPs. During progression of liver fibrosis, there is a marked increase in TIMP-1 and TIMP-2, leading to a net decrease in the catalytic activity of MMPs, so that sustained TIMP-1 production appears to be a keystone for fibrosis progression 96, 97. The substrate specificity of MMPs is of paramount importance. Whereas MMP-1 is the key enzyme for fibrillar collagen destruction, MMP-2 and -9 are of major importance for basement membrane-like material digestion. These enzymes are involved early in the fibrogenesis mechanism, since destruction of the low-density basement membrane-like material of the perisinusoidal space of the normal liver is a prerequisite for fibrillar matrix development (remodelling). As proof of this principle, in the early phase of human fibrogenesis there is upregulation of MMP-2 and -9 and downregulation of MMP-1 96. In this mechanism, the role of a newly discovered subclass of membrane cell receptors, the discoidin domain receptors (DDR), might be of importance 98. These receptors signal in response to fibrillar collagen and are upregulated at the HSC surface after activation 76. One group, DDR-2, upon binding to its collagen ligand, activates MMP-2 transcription in fibroblasts. This could favour the remodelling of ECM. Clearly, metalloproteases are potential tools for liver fibrosis regression and, indeed, this is also an active field of research and development.

Vanishing of HSCs

During liver wound healing, activated stellate cells progressively disappear. Whether these cells reverse to a quiescent phenotype, as observed after culture on basement membrane milieu or stimulation by PPAR-γ ligand, or whether they die by apoptosis, is a matter of debate 42, 99. HSCs, in their myofibroblastic phenotype, express different cell death receptors such as Fas and Fas-ligand and the nerve growth factor receptor 100. Stellate cell apoptosis has been documented during recovery after different experimental liver injury such as bile duct ligation or CCl4 intoxication 89. A link between ECM degradation and HSC apoptosis has recently been suggested 101.

Liver cell regeneration

There is strong evidence that liver regeneration is a crucial condition needed for cirrhosis to regress. A recent morphological study has suggested that the succession from micronodular cirrhosis to macronodular cirrhosis and to incomplete cirrhosis represents the different anatomical steps in cirrhosis regression 88. In this scheme, it is suspected that nodular regeneration of liver cells acts through distension and rupture of fibrous septa leading to a partial recovery of the lobular architecture. However, liver regeneration is not a universal mechanism in cirrhosis. To occur, liver regeneration needs at least two conditions: the internal potential of liver cells to duplicate and the cessation of necroinflammatory processes. Necroinflammation, through production of various cytokines and growth factors, acts as a brake for liver regeneration. As an example, TGF-β, which is produced in large amounts during fibrogenesis, is a potent liver cell apoptosis inducer, so that efficient liver cell regeneration cannot occur in the context of necroinflammation 67. Thus, an efficient treatment in necroinflammation, such as antiviral drug therapy in viral hepatitis, may favour liver regeneration and, as a consequence, cirrhosis regression.

The ability of a liver cell to divide is also related to telomere length. Telomeres are non-coding repeated sequences located at the end of each chromosome arm 102. After each cell division, telomeres will shorten until they reach a critical short size where cells can no longer divide. This mechanism is amplified in the context of chronic hepatitis, where successive episodes of necrosis and regeneration will dramatically shorten telomere ends, as shown by accumulation of replicative senescent cells in some cases of cirrhosis 103. It is of note that reduplication can be induced in senescent cells by the forced expression of telomerase, an enzyme that allows elongation of telomere length. Using gene therapy, this approach has been shown to be successful in experimental models of cirrhosis where forced telomerase expression led to cirrhosis regression and liver cell function restoration in experimental models 104.

Therapeutic approaches to liver fibrosis

Dissecting pathways of activation and perpetuation of HSC led to the development of therapeutic tools acting to block several of these crucial steps (decoy for growth factors or their receptors, mutated receptors, inhibitors of tyrosine kinase receptor signalling, etc) 4, 22. These tools have been shown to be very efficient in stopping fibrogenesis and preventing liver fibrosis in experimental models. One of the main lessons of such approaches is that stopping one of the many pathways involved in the perpetuation step in HSC activation will lead to stopping the others, including fibrogenesis.

Several tools may be potentially effective in the treatment of fibrosis. They may target any of the following biological mechanisms:

  • removal of the initial fibrotic stimulus;

  • inhibition of HSC activation;

  • targeting of liver inflammation;

  • removal of fibrous tissue in excess;

  • stimulation of liver regeneration;

  • removal of initial fibrotic stimulus.

Removing the causal agent is presumably the most efficient way to prevent liver fibrosis progression and stimulate liver fibrosis regression. In chronic hepatitis C, several randomized clinical trials of patients with chronic HCV infection treated either with interferon-α alone or in combination with ribavirin have shown stabilization or even a partial decrease in liver fibrosis in repeated liver biopsies, mainly when there is a virological response 91. The decrease in the inflammatory reaction, which is often observed by antiviral treatment even in the absence of virus eradication, might play a major role, but a direct antifibrotic effect of interferon is also supported by in vitro and experimental evidence 105.

Inhibition of HSC activation

Owing to the new functions acquired by activated HSCs, substances that inhibit either activation or accumulation of HSCs might be of major importance in attenuating the fibrogenic response. Because of the complexity and multiplicity of pathways involved or associated with initial phases of HSC activation and the phase of maintenance, therapeutic targets are multiple.

Different strategies for targeting these pathways have been developed, such as neutralizing anti-TGF-β antibodies, soluble or truncated receptors 106, and inhibitors of the Smad pathway such as halofuginone 107. These strategies have been shown to be efficient in preventing fibrosis in different rat models of liver fibrogenesis. Because of concern regarding the beneficial role of TGF-β in immune response modulation and anticancer defence, human studies have not yet been developed. Such an approach will require the preliminary development of safe, finely tuned vectors for local delivery and control of activity, or the targeting of molecules specifically involved in the fibrogenesis pathways of TGF-β, such as CTGF.

Different therapeutic options targeting the proliferation of activated HSC have also been developed; they act either by blocking PDGF receptors or via intracellular pathways required for cell proliferation. TNP-470, a fumagaline analogue, inhibits HSC proliferation by blocking the G1-S transition in HSCs 108. Similarly, interferon-β and interferon-γ but not interferon-α inhibit rat HSC proliferation 105.

Since the relationship between HSC and ECM is crucial, pharmacological intervention modulating this interaction could have a beneficial effect in preventing liver fibrosis. Soluble Arg-Gly-Asp peptide, by interacting with the integrin binding site, inhibits HSC activation and attenuates liver fibrosis in experimental models of chemically induced liver fibrosis 109. Several substances with antioxidant properties have been tested, with contrasting results.

They might protect or restore the defences against oxidation, or scavenge free radicals such as vitamin E (α-tocopherol), resveratrol, silimarin, and dimethylsulphoxide. Most have shown conclusive effects in various in vitro and experimental models of liver fibrogenesis (CCl4-induced fibrosis or bile duct-ligated models) 110, 111.

Targeting of liver inflammation

The association between liver inflammation and fibrosis is very tight in chronic hepatitis. Promising experimental results suggested that Il-10, an anti-inflammatory and antifibrotic cytokine synthesized by activated HSCs, might be efficient at reversing fibrosis in experimental models mainly by downregulating the Th1 proinflammatory response 73. Direct inhibition of proinflammatory cytokines such as Il-13 has been shown to prevent liver fibrosis in a murine schistosomiasis model 112.

Destruction of excess fibrous tissue

It has recently been suggested that the modulation of MMP activity is an efficient tool for liver fibrosis regression, and there are several potential targets for interrupting or stimulating this pathway, since MMPs are regulated mainly at a post-translational level.

Collagen biosynthesis involves several post-translational modifications leading to the production of highly stable cross-linked molecules. Prolyl-4-hydroxylase is critical in this process since it catalyses the synthesis of the hydroxy-proline residues, critical for stability of the collagen triple helix. Several inhibitors of prolyl-4 hydroxylase (HOE077, S4682) are available. They have shown striking efficacy in animal models by decreasing the amount of collagen accumulation 113. They might also act in concert with other mechanisms such as inhibition of HSC activation.

Simulation of liver cell regeneration

Telomerase is a major reverse polymerase enzyme that allows elongation of chromosome ends, thus allowing infinite replication of somatic cells. Promising results have been obtained with forced expression of hTERT, the catalytic subunit of the telomerase enzyme, and hepatocyte growth factor (HGF), a mitogen for liver cells but also a modulator of HSC activation, collagen formation, and TGF-β expression 104, 114. HGF decreases fibrosis progression in experimental models and might even induce reversion of liver cirrhosis. In view of their proliferative effects on liver cells, these approaches will necessitate further studies to assess the safety of a long-term use specifically focusing on the risk of hepatocarcinogenesis.