The challenge of developing novel pharmacological therapies for non-alcoholic steatohepatitis


Dr Detlef Schuppan, Beth Israel Deaconess Medical Center, Division of Gastroenterology, Harvard Medical School, 330 Brookline Ave; DA-506, Boston, MA 02215, USA
Tel: +1 617 667 2371
Fax: +1 617 667 2767


Non-alcoholic fatty liver disease (NAFLD) is an umbrella term for a series of hepatic pathologies that begin with relatively benign steatosis and can, with appropriate triggers, lead to the serious entity of non-alcoholic steatohepatitis (NASH). This sets the stage for liver fibrosis and finally the development of cirrhosis in up to 20% of patients with NASH. NAFLD, already among the most common diseases in industrialized countries, is increasing in prevalence and roughly affects 30% of US adults and 10% of US children alone. NAFLD is strongly associated with insulin resistance (IR) and represents the hepatic manifestation of the metabolic syndrome. Indeed, treatments aimed at reducing IR are the current mainstay of therapeutic approaches to NAFLD. While lifestyle interventions may produce limited degrees of success, there remains an urgent need for improved pharmacological therapies. Emerging diagnostic and therapeutic opportunities as well as future developments in NAFLD, NASH and liver fibrosis were discussed by a panel of experts and are presented herein. Promising novel therapeutic targets include inhibitors of dipeptidyl peptidase 4 and the renin–angiotensin system. However, improved non-invasive technologies to diagnose and stage NAFLD are needed. Combined with a better understanding of the pathophysiological processes that underlie the mechanisms of hepatic fibrogenesis in NASH, rapid clinical validation of novel therapies is expected.

Non-alcoholic fatty liver disease (NAFLD) encompasses a spectrum of hepatic pathologies, ranging from simple steatosis not to non-alcoholic steatohepatitis (NASH), fibrosis and finally cirrhosis (1). NAFLD and NASH were recognized as distinct entities when Ludwig et al. (2) described liver lesions commonly associated with alcohol abuse (i.e. fatty change, lobular hepatitis and pericellular fibrosis) that were found in patients who did not have a history of excessive drinking.

Today, hepatologists and epidemiologists recognize that NAFLD is the most common liver disease with an estimated prevalence between 20 and 40% in the general population (3). However, epidemiological estimates vary depending on the diagnostic method. Liver biopsy and postmortem series estimate prevalences for NAFLD of between 15 and 39% (4, 5). Estimates derived from unexplained elevations in aspartate aminotransferase (AST) or alanine aminotransferase (ALT) suggest a prevalence of 5.5% (6), but this is an underestimate, because many subjects with NAFLD do not display persistently elevated transaminases. Despite the discordance in epidemiological estimates, it is clear that NAFLD represents a significant reservoir of chronic hepatic disease.

Traditionally, many clinicians regarded NAFLD as relatively benign (7). However, clinico-epidemiological studies contradict this belief, as NASH may account for up to a fifth of NAFLD cases, with an estimated prevalence in Western countries of 2–3% (3). NASH probably causes around 80% of cases of cryptogenic cirrhosis (3) and progresses to advanced fibrosis in 32 to 37% of patients. Obesity, type 2 diabetes mellitus (T2DM) and the associated insulin resistance (IR) increase the risk of fibrosis progression. Thus, between 5 and 20% of non-cirrhotic NASH patients develop cirrhosis during a 10-year follow-up (8, 9) and perhaps one in 200 NASH patients will develop hepatocellular carcinoma (HCC) over a 7-year follow-up (10). HCCs related to NASH are mostly believed to develop from cirrhotic liver, but case reports about HCC arising from non-cirrhotic NASH have been accumulating recently (11). Importantly, liver disease is the second most common cause of mortality in NASH patients and accounts for 12% of deaths within 10 years following the diagnosis of NASH. Clearly, NAFLD is far from being inherently benign, with NASH representing an important and growing public health issue.

The clinical and social burden imposed by NAFLD continues to increase and there are a plethora of factors that raise the risk of development of both NAFLD in general and NASH specifically. Contrary to the early studies, analysis of the Dallas Heart Study indicated that men are up to six times more likely to develop NAFLD compared with women, and at an earlier age (12).

Adolescents and children increasingly present with NAFLD (7). Typically, however, patients at diagnosis tend to be middle-aged and several diseases that rise in prevalence with the advancing age are associated with an increased risk of developing NAFLD. For instance, >80% of NAFLD patients are overweight, with approximately 30% being obese. Furthermore, 80% have concurrent hyperlipidaemia, 30–70% are hypertensive and around 20% have manifest T2DM (3). Thus, NAFLD is part of the metabolic syndrome. The increasing prevalence of IR and obesity in many populations worldwide suggests that the number of cases of NAFLD will rise in coming years (7, 13, 14). In the affluent areas of China, for instance, the number of cases of NAFLD doubled in a decade to affect 15% of the population today, largely driven by the marked increase in average body mass (15).

Currently, weight loss and exercise to reduce IR are the main therapeutic approaches to NAFLD and NASH, as they are for the metabolic syndrome. However, this management paradigm does not work, or only works partially, for most of the patients, even when intensive regimens are applied, because of inability or non-compliance. Equally, current pharmacological treatment, either targeted at weight loss or at improvement of insulin sensitivity, is only modestly effective (16). Therefore, to date NASH is fraught with considerable residual morbidity and mortality and there remains an urgent clinical need for more effective treatments. Despite this unmet need, proposed novel strategies face numerous challenges when moving from the bench to the bedside. For example, liver biopsy as a ‘gold standard’ for grading of disease activity and staging of fibrosis is fraught with high sampling variety (17, 18). Furthermore, it is invasive and risky. Importantly, there are no validated specific non-invasive diagnostic tools to assess NAFLD severity, and as research has yet to elucidate fully the cause(s) of NAFLD and especially its evolution to NASH, there exist only a few potential molecular targets for new agents. Against this background, the review presented here – based on the discussions of a consensus panel of clinical specialists and researchers in NAFLD held in Marbach, Höri, Germany during March 2009 – details progress towards an understanding of the pathology underlying NAFLD and NASH. The review also examines progress towards sensitive and specific non-invasive methods to diagnose, grade and stage this common disease and recent therapeutic developments, such as the emerging role of dipeptidyl peptidase type 4 (DPP-4) and angiotensin receptor inhibitors in NAFLD and NASH.

Non-alcoholic fatty liver disease: the pathogenic enigma partially resolved

Non-alcoholic fatty liver disease potentially manifests as a variety of signs and symptoms including IR, peripheral lipolysis, increased hepatic uptake of fatty acids, hormonal changes (including increased leptin levels), release of pro-inflammatory cytokines and hyperinsulinaemia-induced mitochondrial dysfunction. These signs and symptoms emerge as NAFLD progresses from the mere accumulation of fat in the liver to severe inflammation (NASH), to cirrhosis and eventual death from cirrhosis or liver cancer (Fig. 1).

Figure 1.

 Non-alcoholic fatty liver disease (NAFLD) represents a disease continuum ranging from mere fatty liver to non-alcoholic steatohepatitis (NASH), cirrhosis, primary liver cancer and death.

Numerous factors can contribute to the pathogenesis of NAFLD that represents the underlying clinical presentation and the ‘first hit’. These include drugs and toxins, as well as acute and inherited metabolic diseases. Although a growing body of evidence is beginning to characterize the underlying metabolic disturbances that lead to hepatic steatosis and the pathways that cause liver cell (hepatocyte) injury, inflammation and fibrosis, numerous questions remain unanswered and each advance raises new issues. The enigma that still surrounds many facets of NAFLD hinders attempts to develop innovative, targeted pharmacotherapies.

Traditionally, most pharmacological advances arose from serendipitous observations. However, developing a new pharmacotherapy depends on researchers identifying a pathogenically central molecular target that is amenable to pharmacological manipulation. For example, increased concentrations of glucagon-like peptide-1 (GLP-1) reduce or normalize plasma glucose concentrations in patients with T2DM. DPP-4, mainly expressed in kidney, intestinal brush-border membranes, hepatocytes and vascular endothelium, as well as in a soluble form in plasma, rapidly degrades GLP-1. Therefore, DPP-4 inhibition represented a logical approach to the management of T2DM (19). Identifying an equivalent molecular target for NAFLD and NASH treatment has proved more elusive. Nevertheless, recent advances in our understanding of the underlying pathophysiology of NAFLD and NASH are beginning to yield some potentially valuable therapeutic targets (vide infra).

Some common pathogenic themes emerge from the diversity of suspected causes of NAFLD, most importantly the central role of triglyceride accumulation in the liver. The increased hepatic triglyceride burden, in turn, is thought to promote IR through mechanisms similar to those involved in the development of the metabolic syndrome (20). IR promotes peripheral lipolysis, increased hepatic uptake of fatty acids, leptin levels and pro-inflammatory cytokines (21). Furthermore, increases in triglycerides and free fatty acids (FFA) induce the generation of reactive oxygen species, resulting in oxidative damage to the liver. Increased FFA may also contribute to the development of fibrosis mediated by the accumulation of specialized liver macrophages (Kupffer cells) (22). In particular, transforming growth factor-β (TGF-β1), one of the cytokines produced by a subset of activated Kupffer cells, is a central fibrogenic factor in hepatic fibrosis (23). Finally, predisposition for NAFLD is strongly suggested by studies that identified genetic polymorphisms linked to the metabolic syndrome and T2DM (24), and to hepatic steatosis (25), and by population or family cohort studies (26, 27), while most of the phenotypic expression and pathogenic correlates still require elucidation (Fig. 2). Inflammation coupled with macrophage activation (innate immunity) in the visceral adipose tissue has long been incriminated in the pathogenesis of the metabolic syndrome (and NASH), but recently three tandem papers clearly implicated adaptive immunity and an as-yet unidentified adipose tissue neo-autoantigen in the manifestation of diet-induced obesity (28–30) and likely NASH (31). Thus, CD8+ T cells that are primed in visceral adipose tissue promote, and CD4+ Th2 helper or regulatory T cells suppress, adipose tissue inflammation.

Figure 2.

 The association between adipose tissue inflammation, enhanced circulating free fatty acids, Kupffer cell activation, hepatic lipid accumulation and fibrogenesis (based on (22)).

The current consensus model suggests that the development of NAFLD into NASH requires several ‘hits’ or insults (Fig. 3) (32, 33). According to this model, increased hepatic levels of FFA consequent to impaired insulin sensitivity may serve as the first insult. The increased FFA load would further increase (hepatocyte) IR, steatosis, oxidative stress with lipid peroxidation, endoplasmic reticulum stress, resulting in (hepatic) inflammatory cell accumulation and activation (second hit, Fig. 3). Furthermore, excessive hepatocyte oxidative stress, mitochondrial and microsomal compromise lead to the accumulation of toxic lipid compounds (lipotoxicity), including non-enzymatic products of arachidonic acid and pro-inflammatory metabolites of lipoxygenase, which are also found increased in the plasma of patients with NASH compared with subjects with mere steatosis (34). This finally leads to hepatocyte growth arrest or apoptosis (35), which activates hepatic progenitor cells and associated bile ductular proliferations (36–37), cells that initiate inadequate repair by producing a diverse range and high concentrations of profibrogenic cytokines and growth factors that activate hepatic stellate cells (HSC) and perivascular or portal fibroblasts. The activated HSC themselves can release chemotactic factors that recruit inflammatory cells, creating a ‘positive feedback inflammatory loop’ that leads to fibrogenesis (Fig. 2). Collagen and other extracellular matrix (ECM) components accumulate within the liver, which may result in distortion of the hepatic architecture and finally cirrhosis (23, 35, 38, 39). Cirrhosis and the resultant morbidity and mortality are now regarded as the most relevant ‘hard’ endpoint in clinical studies of NAFLD and NASH. Nevertheless, IR can be considered as the first step on the pathogenic road leading to NASH, fibrosis and cirrhosis. Thus, IR represents the first of the two insults.

Figure 3.

 The multiple hit hypothesis of non-alcoholic fatty liver disease pathogenesis.

Emerging therapeutic options

Currently, most hepatologists attempt to manage NASH using lifestyle changes to reverse the consequences of metabolic disease – such as weight reduction with or without exercise – as well as standard therapeutic interventions to control concomitant associated diseases, such as hyperlipidaemia, hypertension and T2DM. Patients who do not respond adequately to lifestyle changes and pharmacotherapy may benefit from bariatric surgery. However, with the exception of bariatric surgery and rather rare examples of significant lifestyle changes, there is little rigorous direct evidence supporting these treatments in the management of NAFLD.

The large number and mechanistic diversity of pharmacological therapies that researchers have investigated in NASH is testament to the lack of a specific, effective agent. Studies have assessed: insulin-sensitizing agents including thiazolidinediones, which upregulate the activity of the transcription factor peroxisome proliferator-activated receptor (PPAR)-γ; lipid-lowering agents (3-hydroxy-3-methylglutaryl-coenzyme A reductase inhibitors and fibrates); cytoprotective agents (ursodeoxycholic acid); and anti-oxidants (vitamin E, hepatic iron reduction, betaine, S-adenosyl-methionine, N-acetyl cysteine and probucol). As Table 1 indicates, some of these interventions improve certain aspects of the liver damage associated with NASH and NAFLD. Nevertheless, the persistent underlying or residual pathology underscores the remaining need for more effective innovative treatments. Studies into the pathogenesis of NAFLD have yielded several potential targets for novel pharmacotherapies. Among these, DPP-4 and the renin–angiotensin system (RAS) are of great interest.

Table 1.   Outcomes of studies assessing insulin-sensitizing agents and vitamin E in non-alcoholic steatohepatitis
  1. ALT, alanine aminotransferase; GGT, γ-glutamyl transferase; UDCA, ursodeoxycholic acid.

Insulin-sensitizing agents
Caldwell et al. (97)Troglitazone104–6 months↓ inflammation
Ratziu et al. (98)Rosiglitzone6312 months↓ ALT, steatosis
Aithal et al. (99)Pioglitazone7412 months↓ ALT, GGT, ferritin, (fibrosis)
Belfort et al. (100)Pioglitazone556 months↓ ALT, steatosis, ballooning/inflammation, (fibrosis)
Marchesini et al. (101)Metformin144 monthsN/A (↓ IR and ALT)
Lindor et al. (102)UDCA1262 yearsNo change
Neuschwander-Tetri et al. (103)Rosiglitazone2212 months↓ steatosis, inflammation ballooning, fibrosis (Z3)
Promrat et al. (104)Pioglitazone1812 months↓ steatosis, inflammation ballooning, fibrosis
Vitamin E
Dufour et al. (105)UDCA+/− vitamin E402 years↓ ALT, inflammation
Lavine (106)Vitamin E114–10 monthsN/A, ↓ALT
Sanyal et al. (107)Pioglitazone+/− vitamin E216 months↓ steatosis, ballooning, ↓ ALT
Hasegawa et al. (108)A-Tocopherol1212 months↓ 5/9 steatosis, inflammation fibrosis
Kugelmas et al. (109)Exercise+/− Vit E1612 weeksN/A ↓ ALT, no difference
Harrison et al. (110)Vitamin E+C vs. placebo236 monthssimilar ↓ fibrosis in both groups

Dipeptidyl peptidase 4 inhibitors

Epithelial cells and some endothelial cells and lymphocytes express cell surface DPP-4 (40). The DPP-4 gene family includes the cytoplasmic peptidases DPP-8 and DPP-9 as well as DPP-6 and DPP-10, which are expressed on the cell surface but lack enzyme activity (41–43). DPP-4 gene family peptidases have been implicated in several fundamental biological processes, including glycaemic control, fibrosis, tumour growth, immune responses, cell–ECM interactions, vascular physiology and neuronal functions.

In addition to cleaving and inactivating GLP-1, DPP-4 acts on a wide range of substrates that are potential targets in NASH (Table 2). These substrates include several chemokines implicated in the pathogenesis of NASH and its associated complications, and substance P, an important regulator of portal blood flow. Of this wide range of substrates, only substances P, CXCL12, GIP and GLP-2 have been shown to be physiological substrates in vivo. However, additional enzymes attack these substrates and the biological consequences of DPP-4-mediated cleavage of substance P and CXCL12 have not been fully established. Endothelial cells, epithelial cells and lymphocytes in non-diseased and in cirrhotic human livers express DPP-4 and the related fibroblast activation protein (44, 45). Evidence exists supporting DPP-4 as an important player in the pathogenesis of NAFLD. A study involving 31 NASH patients found significantly increased levels of serum DPP-4 relative to control subjects (46). Furthermore, a strong association was found between the intensity of DPP-4 immunostaining in the liver and the histopathological grade and degree of liver steatosis (46). As well as inducing pro-inflammatory immune responses (46), DPP-4 may also control liver fibrogenesis by modulating the interaction between DPP-4-expressing hepatocytes, immune cells and ECM proteins (47, 48). Expression levels of liver DPP-4 increase in diabetic rats fed a high-fat diet (49) while serum levels of DPP-4 are elevated in rodent models of cirrhosis (50) and in patients with hepatitis C virus-related glucose intolerance (51). DPP-4 may influence liver fibrogenesis by modulating the interaction between DPP-4 expressing hepatocytes, cholangiocytes and immune cells and ECM proteins (42, 45). DPP-4 roles in immunity are primarily unrelated to enzyme activity (49) but chemokine inactivation has a potential role (52, 53). The recent discovery that GLP-1 is anti-apoptotic for cholangiocytes (54) suggests a potential indirect pro-inflammatory role for DPP-4 in liver. Thus, a rationale exists for the use of DPP-4 inhibitors to slow the progression of liver steatosis and inflammation, either via their effect on the increasing GLP-1 activity or their potential direct or indirect anti-inflammatory activities in the liver and other tissues.

Table 2.   Substrates for dipeptidyl peptidase type 4
PeptideP2 P1 P1′
  1. Adapted from (44).

Glucagon-like peptide (GLP)-1His Ala Glu-
Glucose-dependent insulinotropic peptide (GIP)Tyr Ala Asp-
Substance PArg Pro Lys-
Neuropeptide Y (NPY)Tyr Pro Ser-
Peptide YYTyr Pro Ile-
Energy homeostasis
GLP-2His Ala Asp-
ProlactinThr Pro Val-
Pituitary adenylate cyclase activating peptide (PACAP) 38His Ser Asp-
Other hormones
PACAP 27His Ser Asp-
Human chorionic gonadotrophin α chainAla Pro Asp-
Growth hormone releasing factor (GHRF)Tyr Ala Glu-
Luteinizing hormone α chainPhe Pro Asn-
Insulin-like growth factor (IGF-1)Gly Pro Glu-
CCL8/eotaxinGly Pro Gly-
CCL22/macrophage-derived chemokineGly Pro Tyr-
CXCL9/interferon-γ-induced monokineThr Pro Val-
CXCL10/interferon-γ-induced protein-10Val Pro Leu-
CXCL11/interferon-inducible T cell α chemoattractantPhe Pro Met-
CCL3L1/macrophage inflammatory protein 1α isoform LD78βAla Pro Leu-
CXCL12/stromal-derived factor 1 α and βLys Pro Val-
Enkephalins, gastrin-releasing peptide, vasostatin-1, peptide histidine methionine, thyrotropin α 

Dipeptidyl peptidase 4 inhibitors are already well-established, effective and well-tolerated treatments for T2DM (55) and have considerable therapeutic potential for chronic liver diseases (56). DPP-4 inhibitors act by increasing the in vivo half-lives of GLP-1 and glucose-dependent insulinotropic peptide (GIP). DPP-4 inhibitors could potentially alleviate NAFLD by increasing bioactive GLP-1 (57) or via both GLP-1 and GIP (58). Notably, reversal of hepatic steatosis in obese (ob/ob) mice has been associated with the augmentation of the GLP-1 signalling pathway (57).

Preclinical studies assessing the novel DPP-4 inhibitor linagliptin offer ‘proof of principle’ that modulating this central enzyme may prove effective in NAFLD and NASH (T. Klein et al., unpublished results). Two studies that used ob/ob and diet-induced obese mice fed a high-fat diet found that linagliptin significantly attenuated hepatic fat accumulation after just 2–3 weeks of therapeutic treatment respectively. Nuclear magnetic resonance (NMR) imaging and analysis of triglyceride levels further confirmed that linagliptin significantly improved liver steatosis. A histological NASH score assessing hepatic steatosis and inflammation in ob/ob mice (Fig. 4a and b) showed that linagliptin was associated with lower scores (i.e. less damage) than untreated animals. In addition, histological scoring correlated well with reduced steatosis as detected via NMR spectroscopy or based on triglyceride content. Clinical data so far suggest a favourable pharmacodynamic and pharmacokinetic profile compared with other DPP-4 inhibitors: linagliptin is rapidly absorbed, has a long duration of action and is primarily excreted via non-renal elimination pathways. These characteristics translate into sustainable reductions in glycosylated haemoglobin levels and a placebo-like tolerability (59, 60). Although the initial clinical development of linagliptin is focusing on its potential as an oral therapy in diabetes, this compound may have future potential in the management of NAFLD and NASH.

Figure 4.

 (a) Representative liver histology from linagliptin- or control-treated animals. Moderate to severe micro and macrovesicular vacuolization, mainly centrolobular and midzonal is observed (controls) and ameliorated following linagliptin treatment. (b) Blinded scoring of liver specimen from animals treated with linagliptin or control for 12 days by severity (severity: 0, none; 1, minimal; 2, mild; 3, moderate; and 4, severe) (95). **P<0.002, t-test, n=9 per group.

Drugs modulating the renin–angiotensin system

Drugs that modulate the RAS offer another rational approach to the treatment of NAFLD and NASH. The ‘classic’ role of the RAS is in the modulation of renal and cardiovascular physiological responses. However, numerous organs and other tissues also produce RAS components, which are regulated independently of circulatory RAS (61).

In the liver, for example, chronic injury upregulates the local tissue RAS, which contributes to the recruitment of inflammatory cells and the development of fibrosis. Specifically, liver injury results in increased concentrations of angiotensin-converting enzyme and angiotensin (Ang) II at the site of tissue damage. Furthermore, in angiotensin 1 (AT1) receptor knockout mice, inflammation and liver fibrosis induced by carbon tetrachloride and bile duct ligation are attenuated (62). This underscores the potential pathogenic significance of the RAS and suggests that drugs modulating local tissue RAS potentially offer a rational approach to the treatment of NASH and liver fibrosis.

Against this background, telmisartan shows the greatest affinity for the AT1 receptor and exhibits the longest elimination half-life of the currently available angiotensin receptor blockers (ARBs). Furthermore, telmisartan's lipophilicity and high liver compartmentalization suggests that this ARB is especially suitable for hepatic indications. Indeed, telmisartan can penetrate interstitial fluid–reflected in an apparent volume of distribution (Vd) that is more than five-fold higher than the ARB with the next highest Vd (63). Furthermore, telmisartan increases the expression of PPAR-γ and PPAR-α target genes in human and murine hepatic cells, in addition to antagonizing the AT1 receptor. These pleiotropic actions resulted in telmisartan showing comparable efficacy to pioglitazone on steatosis and fibrosis in rats with hepatic damage arising from a methionine- and choline-deficient (MCD) diet (64) or in mice fed the MCD in conjunction with a high-fat diet (65, 66). Telmisartan inhibited steatosis, TGF-β1, α-smooth muscle actin and procollagen type I expression. In this model, valsartan (another ARB) produced less marked effects than either pioglitazone or telmisartan (66).

Some other potentially promising interventions are more speculative. For example, stimulating subsets of macrophages could help to dissolve the scar tissue and reverse the activation of HSCs and myofibroblasts resulting in fibrosis reversal (23, 39, 67). The process through which TGF-β modulates the conversion of HSC into myofibroblasts offers another tempting target for antifibrotic disease modulation as this process directly produces scar tissue. Combining stem cell therapy with antifibrotics potentially both induces fibrosis reversal and stimulates hepatocyte regeneration. However, these approaches are currently in the hypothetical or preclinical stage rather than close to clinical development. After all, DPP-4 inhibitors and telmisartan are already widely used clinically, with well-characterized tolerability and safety profiles. They also offer the additional benefit of controlling concurrent associated diseases (T2DM and hypertension respectively).

The need for new animal models

While such molecular leads show prima facie promise, to screen candidate molecules researchers need animal models that mimic NASH in humans. However, developing a single experimental model that has all the characteristics of human NASH has proven difficult. For example, the classic recessive Zucker (fa/fa) obese rat model manifests IR and steatosis, but only develops modest hepatitis and minimal fibrosis. The ideal model would probably undergo a two-hit challenge. The first challenge may take the form of an inherent metabolic defect that is combined with a suitable and plausible environmental insult, such as a high-fat calorie-rich or MCD diet. While developing new NASH models is still an area of intensive research, an increasing variety of small animal models is beginning to more accurately replicate the pathogenesis of NASH than traditional models (Table 3). Furthermore, a spontaneous model of NAFLD has been described that develops in aged bonnet monkeys (68), showing elevations of serum ALT, AST and triglycerides, as well as hepatic micro- and macrovesicular steatosis, and portal, perivenular and perisinusoidal fibrosis However, further studies need to correlate animal models with all aspects of human NASH, to confirm and validate the most appropriate preclinical systems.

Table 3.   Rodent models that show necroinflammation and fibrosis similar to human non-alcoholic steatohepatitis
  1. CAM, cell adhesion molecule; CEA, carcinoembryonic antigen; DEN, diethylnitrosamine; FA, fatty acid; FXR, farnesoid X receptor; HFD, high fat diet; 11β-HSD-1,11β-hydroxysteroid dehydrogenase-1; LDL-R, low-density lipoprotein receptor; MCD, methionine and choline-deficient diet; SREBP-1c, nuclear sterol regulatory element-binding protein 1c.

Animal models that express necroinflammation and fibrosis
Long-Evans Tokushima fatty rats+MCD+HFDOta et al. (111)
SREBP-1c transgenic mice+MCDNakayama et al. (112)
L-SACC1 (CEA-related CAM) knockout mice+HFDLee et al. (113)
LDL-R knock-out+FXR knockout mice+HFDKong et al. (114)
Adipocyte 11β-HSD-1 transgenic mice+HFDMorton and Seckl (115)
Animal models that use diet modification to generate NASH
Atherogenic diet in miceMatsuzawa et al. (116)
Intragastric high polyunsaturated FA (corn oil) in ratsBaumgardner et al. (117)
MCD+HFD in miceCong et al. (118)
MCD+trans-fats+DEN in ratsde Lima et al. (119)

Diagnosis and disease monitoring in preclinical models and clinical studies

Accurate and objective measurements of disease processes and progression are vital to meaningfully assess the potential efficacy of any new treatment for NASH during preclinical and clinical development, as well as to monitor responses to treatment. Currently, liver biopsy represents the gold standard diagnostic procedure. Histological features recorded in liver biopsies allow hepatologists to score the severity of the hepatic disease and to assess prognosis. As NASH is a diffuse liver disease, even a small sample such as a biopsy, representing ∼1:50 000 of the liver, should be sufficient to indicate changes occurring in the whole organ. However, biopsy is fraught with a significant sampling variability, leading to the misclassification of ≥1 fibrosis stage (out of five stages from 0=normal to 4=cirrhosis) in approximately 40% of cases (17, 18, 39). Despite these shortcomings, a decision tree model favoured early liver biopsy over no biopsy for patients with NAFLD, reducing the overall morbidity, mortality and need for liver transplant in the following 5 years (69).

Moreover, liver biopsy is associated with a small risk of complications, is time consuming and is uncomfortable for patients. Therefore, using repeated liver biopsies in trials and to monitor treatment raise ethical and clinical issues. Reliable surrogate biomarkers of disease pathogenesis and dynamic markers for fibrogenesis and fibrolysis would be valuable, both clinically and for drug development. Currently, serum cytokeratin 18 (CK18) fragments are a relatively accurate predictor of hepatocyte apoptosis and appear to correlate well with NASH activity as determined by the NASH activity score. However, CK18 fragments are a general marker for hepatocyte apoptosis and are, therefore, also elevated in other liver diseases, e.g., alcoholic or active viral hepatitis. In addition, they correlate with the fibrosis stage (70–73). This and other markers need further validation in therapeutic studies in patients with NASH (74), ideally combined with changes in histology or imaging.

Progress in imaging techniques

In animal experimental studies, imaging techniques offer new insights into NAFLD and NASH. For example, longitudinal studies using magnetic resonance imaging (MRI) showed a strong correlation with results obtained using ex vivo biochemical methods. NMR also appears to hold promise and a study showing that DPP-4 inhibitors are potentially effective in NAFLD and NASH used this technique (vide supra). MRI offers a rapid (approximately 4.5 min per animal), unbiased, non-invasive analysis of liver lipids, making the technique appropriate for longitudinal studies. However, this method only allows for the quantification of hepatic fat.

Diffusion-weighted MRI (DWI) is a technique that provides image contrast through the measurement of the diffusion properties of water within tissues. Pathological processes within tissue such as fibrosis change the water content and affect local magnetic fields (75, 76). The application of DWI in staging liver fibrosis and diagnosing cirrhosis has been demonstrated (77, 78), but as yet the technique is currently restrained by limited reproducibility and technical issues (76, 78).

A further imaging approach for detecting and quantifying hepatic fibrosis involves the intravenous infusion of superparamagnetic iron oxide (SPIO) particles (79, 80) followed by standard contrast MRI. SPIO particles accumulate within the reticuloendothelial cells of the liver leading to hypo-intensity within the liver parenchyma (80–82). In patients with fibrosis, this leads to fibrotic bands being relatively hyperintense. Thus, alterations in hepatic architecture may be seen on contrast-enhanced MR images. SPIO-based images were highly accurate (85%) for detecting fibrosis compared with histopathological features (82), but it is unclear if it will permit to assess an increase or decrease of scar tissue in clinical studies.

Targeted imaging agents could potentially enhance the acuity of current techniques and allow a quantification of fibrosis or fibrogenesis. The expression of certain cell surface receptors (including several integrins) changes during fibrogenesis. This raises the prospect of using appropriately labelled small molecules that bind to those surface receptors that are upregulated as NASH progresses. For example, as hepatic fibrosis progresses, the expression of integrin αvβ6 in activated cholangiocytes and progenitor cells increases in patients and in rodents (83, 84). A high-affinity, radio-labelled αvβ6 ligand appears to accurately quantify fibrogenesis when imaged using single photon emission computed tomography (SPECT) analysis in spontaneously fibrotic Mdr2 -/- mice (36).

Transient elastography using ultrasound (Fibroscan®, Echosense, Paris, France) or MRI elastography can diagnose cirrhosis with >90% sensitivity and specificity (85–87). Ultrasound elastography provides a rapid and non-invasive measure of hepatic stiffness that correlates with fibrosis. Ultrasound elastography is easy to perform and biologically plausible. However, it lacks the sensitivity and specificity for monitoring the progression or reversal of fibrosis in NAFLD, partly because inflammation may alter the liver's elastic properties. Furthermore, obesity is common in NAFLD patients and increased subcutaneous fat decreases its accuracy (85).

Diagnostic problems and patient recruitment

Despite advances in imaging technologies, the differential diagnosis of NASH can still prove problematic, which further compromises attempts to design studies that evaluate novel treatments. NASH and NAFLD are multifactorial diseases with signs and symptoms that overlap with other conditions, which precludes a clear delineation of the condition in some patients. To complicate matters further, some patients also have risk factors for other hepatic pathologies, such as concomitant alcohol consumption (which is difficult to differentiate from NASH histologically), or infection with the hepatitis C virus (88).

Clinical laboratory findings and histopathology of biopsy samples can aid the differential diagnosis. However, in general, clinicians diagnose NAFLD and NASH by exclusion, especially of significant alcohol consumption (as defined by ≤40 g of daily alcohol in men and ≤20 g in women). For example, unexplained elevations to more than 2.5 times the upper limit of AST, ALT or both, the presence of conditions commonly associated with NAFLD (notably excessive body weight or T2DM) and unexplained hepatomegaly raise the index of clinical suspicion for NASH. Other clues include increases in alkaline phosphastase (present in around 30% of patients), elevated ferritin levels (53–62% of patients) and ALT concentrations exceeding AST levels (in around 65–90% of cases, contrary to many patients with advanced alcoholic liver disease where this is often reversed). Overweight and obese patients frequently show NAFLD and NASH with asymptomatic elevation of serum aminotransferase levels (3). Therefore, increased disease awareness and informed screening of at-risk populations, such as people attending diabetes and obesity clinics, increase the case detection rate.

As the above discussion suggests, the NASH patient population is heterogeneous, with varying patterns of signs and symptoms arising from the multifaceted underlying causes and pathologies. This intrinsic heterogeneity complicates the interpretation and analysis of clinical studies in NAFLD and NASH. However, several predictive factors and biomarkers may allow investigators to enrich studies by selectively enrolling those patients most likely to progress and express the most overt forms of NASH. This approach should improve the homogeneity of clinical trial cohorts and simplify analysis and interpretation. However, the pathology of end-stage disease may differ from that seen earlier in the natural history of NAFLD and respond differently to treatment. This may complicate the extrapolation of the results to a less advanced disease, the patient group that has the greatest chance of regression or remission.

Another approach to enhance the homogeneity of clinical cohorts uses a variety of clinical features to stratify patients. For example, several factors appear to be associated with severe fibrosis in NASH patients, including age>45 years [adjusted odds ratio (OR) 5.6]; body mass index >30 kg/m2 (OR 4.3); ALT over the upper limit of normal (OR 4.3); and the presence of T2DM (OR 3.5) (89). A small study that enrolled patients with different background causes and severity of NASH, from mild disease to cirrhosis, found that measurement of liver stiffness could predict the presence of stage 2 (portal) fibrosis (90, 91). Further studies now need to confirm the validity of these clinical models in larger cohorts during more protracted follow-up.

Selecting the appropriate endpoints

Selecting the appropriate endpoints for clinical studies in NASH presents a clinical trial conundrum. In cirrhosis, improved liver angio-architecture, the restoration of liver function, amelioration of portal hypertension and reduced mortality represent rational endpoints. The delay or cessation of fibrosis would be an appropriate interim endpoint.

However, as alluded above, measuring these outcomes could raise clinical and ethical issues and ‘hard outcome’ studies will potentially require a relatively large cohort followed for a protracted time. The progression of NASH to liver-related death can take up to 20 years. Indeed, NASH progresses in only about 40% of patients over a 5–7-year follow-up. The disease may regress in many of the remaining patients. Up to 30% of patients receiving placebo in clinical studies may experience improvements in their disease, symptoms or both, apparently because of loss of body weight and lifestyle subsequent to diagnosis. In other cases, the steatosis or hepatic fat accumulation may not change. Currently, prospectively identifying which patients will progress is inaccurate, although some clues have emerged (vide supra).

While a baseline biopsy remains relevant for stratification, and when performed early in the disease course improves the outcome and reduces overall costs (69), biopsies may not be sufficiently sensitive to detect subtle but clinically significant changes, such as symptomatic regression or reduced fat accumulation. As mentioned above, most current studies have not validated biomarkers or available imaging techniques in terms of correlating results with histological changes observed in biopsy. Therefore, researchers would benefit from better serum markers and imaging modalities of hepatic matrix deposition and removal (92) (vide supra).

In the future, advanced proteomics and targeted molecular imaging may facilitate the more accurate prospective stratification of patients (Fig. 5) to enhance the accuracy of clinical prognostic predictions and during enrolment of clinical trial cohorts (39, 88, 92). However, the polygenic nature of NAFLD may complicate the implementation of genetic approaches to diagnosis and patient stratification. Indeed, the 25% of patients who develop cirrhosis over 20 years tend to be young males, some of whom may express five or six single nucleotide polymorphisms (93). Nevertheless, even if sensitive and specific surrogate markers of disease progression emerge, future clinical study designs will need to address confounding factors, such as the variable clinical progression.

Figure 5.

 Upregulated and downregulated genes implicated in the pathogenesis of non-alcoholic steatohepatitis (96).

The slow rate at which NAFLD usually progresses mandates a longer follow-up in studies investigating the clinical potential of new treatments for NASH than is typical in similar trials in other diseases. To address the intrinsic heterogeneity, the studies are also likely to need to recruit greater numbers of patients than is typical, with especially rigorous control groups to compensate for the unpredictable aetiology and progression. Alternatively, with the pending advent of better non-invasive biomarkers or imaging techniques of NASH activity or progression to advanced fibrosis/cirrhosis (which will reflect the whole liver), stratification of patients as to risk and speed of progression will be easier, leading to much lower required numbers of patients in clinical studies.

Navigating the regulatory pathway

Although there is a recognized need for new treatments for NASH and NAFLD, the regulatory pathway for achieving approval for any new treatment is not clear. Pharmaceutical and biotechnology companies are currently looking to the Food and Drug Administration (FDA), European Medicines Agency and other regulatory authorities for guidance regarding acceptable methodology for the pivotal studies for license applications. This regulatory confusion could potentially contribute to the death of patients, or at least a delay in approved drugs for NASH and NAFLD.

Until 3 years ago, the FDA did not consider fibrosis or cirrhosis to represent a proven disease state and the authority has not issued a position on appropriate endpoints for studies of either liver fibrosis or NASH. However, in response to questions about what they would view as acceptable clinical and histological endpoints, the FDA suggested prevention of disease progression (raising the prospect of using histological cirrhosis as an endpoint) and either the regression or stabilization of fibrosis (94). The FDA adopts the position that there are not yet any validated markers for antifibrotic efficacy and the organization does not accept clinically used surrogates, such as IR or hepatic inflammation. Therefore, clinical trials also need to validate any putative biomarkers to facilitate future use in regulatory studies.

Against this background, progression of liver fibrosis has become the de facto primary endpoint in future drug trials for the treatment of chronic liver diseases. However, academic researchers and companies should collaborate with regulatory authorities to validate new biomarkers and surrogates. In particular, researchers, companies and regulators need to address the relationship between biomarkers and fibrotic endpoints. Only 3–4% of NASH patients progress to end-stage disease each year and it is difficult to identify these patients prospectively. By the time patients reach end-stage disease, some key pathological processes believed to be involved in progression, such as apoptosis, inflammation and fibrosis may be downregulated (94). Regulators need to address such concerns urgently to ensure that the growing number of patients with these potentially serious diseases can access effective and well-tolerated treatments.


Non-alcoholic fatty liver disease, which goes hand in hand with the metabolic syndrome, is one of the most common diseases in industrialized countries. Demographical trends, such as the expected rise in obesity and the growing number of people with IR or overt T2DM, suggest that the number of NAFLD cases may increase over the next few years. Unfortunately, current approaches to the management of NAFLD and NASH are often suboptimal. There remains a recognized clinical need for an effective treatment, and emerging evidence from animal studies suggests that the liver damage arising from NAFLD may be reversible. Furthermore, a deepening understanding of the underlying pathogenesis is beginning to yield several potential molecular targets, exemplified by DPP-4 and the RAS, which may hit several pathways in the pathogenesis of NASH that could result in a new generation of effective therapies with no or minimal side effects.

Nevertheless, several factors will complicate the development of novel pharmacotherapies including: the multifactorial pathogenesis of NAFLD; the heterogeneity of the patient population; the imprecision of current disease staging techniques; ill-validated surrogate markers; the slowly progressive nature of NASH; and the tendency of a significant proportion of patients to show spontaneous disease regression, likely related to the improvement of metabolic control. To complicate matters further, while phase I/II studies have assessed several compounds, the regulatory pathway into phase III investigation is unclear and awaits definite responses from regulatory authorities. The first steps in addressing these challenges will be to design clinical trials that use conventional histological criteria to assess antifibrotic drug effects. Such studies would probably require at least 200–500 well-stratified homogeneous patients with intermediate-stage fibrosis. Even with optimized histological readouts, studies will have to last at least 1–2 years. Furthermore, the studies will need to concurrently evaluate possible new biomarkers and surrogate endpoints. Meeting the challenges in developing new treatments for NASH and NAFLD will require a considerable investment by any single agency or pharmaceutical company. Despite the challenges, we need to take the unprecedented opportunity offered by the increased understanding of the pathogenesis of NASH and NAFLD to reduce the morbidity and mortality associated with these common diseases, which continue to grow in prevalence.


Boehringer Ingelheim Pharma GmbH & Co. KG supported the meeting and the subsequent development of this review article through an unrestricted educational grant. The authors would like to thank Dr Tim Hardman of PHASE II International for editorial support in preparing this paper.

The participation of the following individuals in the consensus meeting is gratefully acknowledged: Prof. Dr Nezam H. Afdhal, Dr Stephan Glund, Prof. Dr Mark D. Gorrell, Dr Rolf Grempler, Dr Brad Hamilton, Dr Tobias Hildebrandt, Dr Stefan Kauschke, Dr Katalin Kauser, Dr Thomas Klein, Dr Michael Mark, Dr Eric Mayoux, Dr Heiko Niessen, Dr Thomas Rauch, Dr Corinna Schölch, Prof. Dr Detlef Schuppan, Prof. Dr Ulrich Stölzel, Dr Christine Teutsch, Dr Leo Thomas, Dr Amelia Viana and Dr Johannes Zanzinger.

Conflict of interest statements: Dr. Schuppan is a consultant to and has received honoraria from Abbott, Biomerieux, Genentech, Boehringer-Ingelheim, Sanofi-Aventis and Stromedix.

Dr Gorrell has been a consultant to and/or received honoraria or indirect benefits from AstraZeneca, Bristol-Myers Squibb, Merck & Co., Inc. (USA), Merck Sharp & Dohme (Australia) and Boehringer Ingelheim.

Dr Klein and Dr Mark are full-time employees with Boehringer Ingelheim, and have no other conflicts of interest.

Dr Afdhal is a consultant to and has received honoraria from Schering Plough, GSK, Boehringer Ingelheim, Vertex, and Quest. He has received research support from Schering Plough, GSK, EchoSens and Vertex.