Cause, Pathogenesis, and Treatment of Nonalcoholic Steatohepatitis

The new england journal of medicine

Review Article

Dan L. Longo, M.D., Editor Cause, Pathogenesis, and Treatment of Nonalcoholic Steatohepatitis

Anna M. Diehl, M.D., and Christopher Day, M.D.

atty liver disease is the most common liver disease in the world. From the Department of Medicine, Duke About 25% of adults in the United States have fatty livers in the absence of University, Durham, NC (A.M.D); and Newcastle University Medical School, excessive alcohol consumption, a condition termed nonalcoholic fatty liver Newcastle upon Tyne, United Kingdom F (C.D.). Address reprint requests to Dr. disease. More than a quarter of adults with nonalcoholic fatty liver disease are presumed to have nonalcoholic steatohepatitis on the basis of elevated serum Diehl at the Snyderman Bldg., Suite 1073, 1 Duke University, Durham, NC 27710, or aminotransferase levels and an absence of other identifiable causes of liver injury. at annamae.diehl@duke.edu. A definitive diagnosis of nonalcoholic steatohepatitis is currently based on histo- N Engl J Med 2017;377:2063-72. logic evidence not only of fat accumulation (steatosis) in hepatocytes but also of DOI: 10.1056/NEJMra1503519 liver-cell injury and death and accumulation of inflammatory cells (Fig. 1A and Copyright © 2017 Massachusetts Medical Society. 1B). Because livers with nonalcoholic steatohepatitis are more damaged than livers with isolated steatosis, nonalcoholic steatohepatitis is more likely than isolated steatosis to lead to progressive liver fibrosis and eventual liver-related illness and death2-6 (Fig. 1C). This review focuses on our understanding of the epidemiology and pathogenesis of nonalcoholic steatohepatitis, which underpins practice guide- lines and drug development for this life-threatening liver disease.

Epidemiologic Features Nonalcoholic steatohepatitis is strongly associated with overweight or obesity and the metabolic syndrome. A recent analysis of studies involving more than 8.5 mil- lion persons from 22 countries showed that more than 80% of patients with nonalcoholic steatohepatitis are overweight or obese, 72% have dyslipidemia, and 44% have received a diagnosis of type 2 diabetes mellitus.1 This information sup- ports the concept that nonalcoholic steatohepatitis is the hepatic correlate of the metabolic syndrome, a systemic disorder of energy homeostasis that often accom- panies visceral adiposity. Unlike isolated hepatic steatosis, nonalcoholic steatohepatitis is strongly associated with liver fibrosis (scarring), according to liver- biopsy series.4-6 Indeed, these studies show that some level of fibrosis is typical in nonalcoholic steatohepatitis and that liver fibrosis is advanced (defined histologi- cally as fibrosis stage F2 or higher, on a scale ranging from F0 to F4 as follows: no fibrosis [F0], portal fibrosis without septa [F1], portal fibrosis with few septa [F2], bridging septa between central and portal veins [F3], and cirrhosis [F4]) in at least a quarter of patients at diagnosis. Liver fibrosis among patients with nonalcoholic steatohepatitis is to be expected, since wound healing involves fibrosis and patients with nonalcoholic steatohepatitis have greater liver injury than those with isolated steatosis.7 Nonalcoholic steatohepatitis is a dynamic condition that can regress to isolated steatosis, smolder at a relatively constant level of activity, or cause progressive fibrosis that leads to cirrhosis (F4 fibrosis). Natural history studies indicate that the

n engl j med 377;21 nejm.org November 23, 2017 2063 The New England Journal of Medicine Downloaded from nejm.org by LUIS ERNESTO GONZALEZ SANCHEZ on December 7, 2017. For personal use only. No other uses without permission. Copyright © 2017 Massachusetts Medical Society. All rights reserved. The new england journal of medicine

A B

C Prevalence in the United States: 25% of adults 5–6% of adults 1–2% of adults

25% HCC NAFLD, Cirrhosis 25% 25% ?% NASH Type 2 diabetes, Fat plus injury, Fat plus injury, Fat plus injury, Western diet, inflammation, inflammation, inflammation, and other factors and scarring and scarring scarring, and mutation

No NAFLD, 75%

Figure 1. Histologic Features and Prevalence of Nonalcoholic Steatohepatitis (NASH). NASH is a potentially progressive type of nonalcoholic fatty liver disease (NAFLD). Panels A and B show character- istic histologic features of NASH in liver-biopsy specimens: ballooned hepatocytes (arrows), inflammatory infiltrate (arrowheads), and fibrosis. Panel C shows the relative distribution of NASH, cirrhosis, and primary liver cancer in the U.S. adult population. Data in Panel C are from Williams et al.2 and Adams et al.3 HCC denotes hepatocellular carcinoma.

severity of liver fibrosis is the only histologic Progressive fibrosis in patients with nonalco- measure that independently predicts liver-related holic steatohepatitis has critical clinical implica- illness, liver transplantation, and liver-related tions because advanced fibrosis reflects repeated death in patients with nonalcoholic fatty liver futile efforts to regenerate healthy liver architec- disease.5,6 Analyses of sequential liver-biopsy ture, and defective liver regeneration increases specimens in patient cohorts indicate that liver the risk of cirrhosis and primary liver cancer.7 fibrosis progresses at a rate of approximately Cirrhosis related to nonalcoholic steatohepatitis one stage per decade, suggesting that F2 fibrosis will develop in approximately 2% of American will progress to cirrhosis within 20 years. How- adults at some point in their lives. This informa- ever, rates of fibrosis progression (and regression) tion is derived from fibrosis-progression rates vary considerably among individual patients and among patients with nonalcoholic fatty liver dis- may not be linear over time even in a given per- ease, as well as evidence that approximately 6% son. In some cases, for example, fibrosis quickly of the adult population in the United States has diminishes as nonalcoholic steatohepatitis im- nonalcoholic steatohepatitis and that approxi- proves, whereas in others, scarring persists or mately 25% of affected patients have at least F2 worsens even after nonalcoholic steatohepatitis fibrosis at the time of diagnosis.1 Indeed, nonalco- has resolved.8 holic steatohepatitis will most likely be the top

2064 n engl j med 377;21 nejm.org November 23, 2017 The New England Journal of Medicine Downloaded from nejm.org by LUIS ERNESTO GONZALEZ SANCHEZ on December 7, 2017. For personal use only. No other uses without permission. Copyright © 2017 Massachusetts Medical Society. All rights reserved. Nonalcoholic Steatohepatitis

reason for liver transplantation in the United pathogenic mechanisms predominate have be- States by 2020.9,10 come major foci for research because such Nonalcoholic steatohepatitis is also fueling knowledge is necessary for the development of the rising incidence and prevalence of primary effective interventions for preventing and treat- liver cancer in many countries, including the ing nonalcoholic steatohepatitis. United States.11 The incidence of hepatocellular Hepatocyte injury and death are key features carcinoma is at least 1 to 2% per year among that differentiate nonalcoholic steatohepatitis patients with cirrhosis related to nonalcoholic from isolated steatosis.4 Whether hepatocyte in- steatohepatitis12; primary liver cancers can also jury is the primary cause or a secondary conse- develop in patients with noncirrhotic nonalco- quence of liver inflammation is debatable. Both holic steatohepatitis.12,13 are likely to be relevant to the pathogenesis of In addition, hepatic steatosis and nonalcoholic nonalcoholic steatohepatitis because injured he- steatohepatitis occur in children. As with adults, patocytes release factors promoting the accumu- nonalcoholic steatohepatitis in children is strong- lation of immune cells that produce hepatotoxic ly associated with obesity. A study of more than substances and incite further injury and inflam- 250,000 Danish children showed that childhood mation. Conversely, factors that promote inflam- obesity increased the risk of hepatocellular car- mation (e.g., visceral adiposity, the metabolic cinoma in adulthood,14 prompting concern that syndrome, changes in intestinal microbiota, and the childhood obesity epidemic in the United circadian-rhythm disruption) increase hepato- States may spawn an epidemic of chronic liver cyte exposure to cytokines, gut-derived products, disease, cirrhosis, and liver cancer that will and other inflammatory mediators that are haunt the U.S. population for decades to come. hepatotoxic.17,18 The economic burden of nonalcoholic fatty liver Injured hepatocytes induce stress responses, disease in the United States is already enormous, which limit damage to their vital organelles. with a current estimate of more than $100 bil- However, these adaptations may inadvertently lion in annual direct medical costs, much of increase vulnerability to other stressors or may which is attributable to nonalcoholic steatohepa- simply be insufficient to abort signaling cas- titis and its sequelae.15 cades that result in cell death.19,20 Dying hepatocytes, in turn, trigger regenerative responses, Pathogenesis promoting the replacement of dead hepatocytes. For example, dying hepatocytes produce mor- Nonalcoholic steatohepatitis always develops in phogens that enrich injured livers with regenera- the context of hepatic steatosis, but isolated ste- tive cell types that are not abundant in healthy atosis is three or four times as prevalent as non- livers, such as myofibroblasts, immune cells, alcoholic steatohepatitis.1,15 Furthermore, the pres- and liver-cell progenitors.21,22 An orderly pro- ence of nonalcoholic steatohepatitis does not gression of the consequent wound-healing re- correlate with the severity of steatosis as as- sponses results in successful regeneration. How- sessed on the basis of liver biopsy or current ever, the wound-healing process stalls during imaging techniques.6 Since these tests mainly tissue reconstruction when the injury is repetitive capture hepatic triglyceride content and triglyc- or repair becomes dysregulated. Over time, futile erides themselves are not directly hepatotoxic, it regenerative responses promote progressive scar- is presumed that hepatocyte injury is inflicted by ring, which leads to cirrhosis, and perpetuate toxic triglyceride precursors or products of tri- the stimulus for neoplasia, increasing the risk of glyceride metabolism.16 Evidence that multiple liver cancer7 (Fig. 2). Histologic assessment pro- lipid intermediates are cytotoxic, combined with vides a snapshot of liver injury and repair re- the numerous factors that might enhance or responses at the time of biopsy and helps assign duce vulnerability to each of these toxic moieties, patients with nonalcoholic fatty liver disease to suggests that nonalcoholic steatohepatitis may prognostic groups4; patients with progressive be the common manifestation of diverse disease fibrosis are losing the wound-healing battle and processes.8 Characterizing these processes, de- thus may need help to recover. veloping approaches to detect their presence and A number of inherited and environmental intensity, and determining whether one or more factors promote the pathogenesis of nonalco-

n engl j med 377;21 nejm.org November 23, 2017 2065 The New England Journal of Medicine Downloaded from nejm.org by LUIS ERNESTO GONZALEZ SANCHEZ on December 7, 2017. For personal use only. No other uses without permission. Copyright © 2017 Massachusetts Medical Society. All rights reserved. The new england journal of medicine

Normal hepatocyte Fatty hepatocyte Fatty and injured hepatocyte Apoptotic hepatocyte

Adipose Gut and Adipose Gut and Adipose Gut and tissue microbiome tissue microbiome tissue microbiome

Lipotoxicity Metabolic stress

Ballooned

Cell signals

Repair-related cells

Immune Activated sinusoidal Myofibroblastic Reactive cells endothelium stellate cell ductal cells

Wound-healing responses Inflammation, vascular remodeling, fibrogenesis, and accumulation of immature liver epithelial cells

Figure 2. Pathogenesis of Nonalcoholic Steatohepatitis. Various factors, including inflammation, hyperinsulinemia or insulin resistance, and altered lipid homeostasis, can induce metabolic stress, oxidative stress, and endoplasmic reticulum–related stress to develop in fatty hepatocytes (i.e., lipotoxicity). Changes from normal signaling patterns are indicated by thick double-headed arrows. When mechanisms to cope with these stresses become overwhelmed, hepatocytes die. Dying and dead hepatocytes release signals to cells that are necessary for the repair of liver damage, such as immune cells, sinusoidal endothelial cells, hepatic stellate cells, and ductal-type cells. The hepatocyte-derived, damage-associated signals cause the repair-related cells to accumulate and launch wound-healing responses, which include inflammation, vascular remodeling, fibrogenesis, and hepatic accumulation of immature liver epithelial cells. Nonalcoholic steatohepatitis is the sum of injury and repair responses triggered by lipotoxicity.

holic steatohepatitis. Information about some of fatty liver disease and chronic hepatitis C, an the pathogenic factors that are becoming diagnos- infection that promotes hepatic steatosis. The tic or therapeutic targets is summarized below. prevalence of pathologic PNPLA3 polymorphisms differs among ethnic groups, and these differ- Genetic Factors ences generally parallel those for nonalcoholic Genomewide association studies indicate that steatohepatitis and its sequelae: both the poly- polymorphisms in patatin-like phospholipase morphisms and the disorders are most prevalent domain–containing 3 (PNPLA3) and transmem- in Asian and native American populations, are brane 6 superfamily, member 2 (TM6SF2) pro- less common in whites of Northern European mote the development of nonalcoholic steato- ancestry, and are least common in blacks. How- hepatitis and related liver damage (i.e., cirrhosis, ever, PNPLA3 I148M is neither sufficient nor primary liver cancer, or both).23-26 PNPLA3 en- necessary to cause nonalcoholic steatohepatitis, codes adiponutrin, a lipase that regulates both cirrhosis, or liver cancer, and the mechanisms triglyceride and retinoid metabolism. PNPLA3 through which PNPLA3 aberrancy or adiponutrin polymorphisms (e.g., I148M) are strongly associ- dysfunction promotes the pathogenesis of non- ated with hepatic steatosis, steatohepatitis, fibro- alcoholic steatohepatitis are unknown. sis, and cancer. The I148M polymorphism also TM6SF2 also encodes a protein that regulates promotes liver damage caused by alcohol-induced hepatocyte lipid content. A TM6SF2 polymorphism

2066 n engl j med 377;21 nejm.org November 23, 2017 The New England Journal of Medicine Downloaded from nejm.org by LUIS ERNESTO GONZALEZ SANCHEZ on December 7, 2017. For personal use only. No other uses without permission. Copyright © 2017 Massachusetts Medical Society. All rights reserved. Nonalcoholic Steatohepatitis

that enhances hepatocyte secretion of very-low- rodents have identified an epigenetic mechanism density lipoprotein (VLDL) is hepatoprotective that regulates fate decisions in fibrogenic liver but increases the risk of cardiovascular disease, cells, controlling transgenerational susceptibility and another TM6SF2 polymorphism, which re- to cirrhosis.33,34 Similar changes in circulating duces hepatocyte VLDL secretion, is associated DNA correlate with the severity of liver fibrosis with nonalcoholic steatohepatitis and related in patients with nonalcoholic steatohepatitis.35 liver fibrosis, suggesting that proper trafficking of lipids is necessary to prevent lipotoxicity. In- Environmental Factors deed, a deficiency of other proteins required for Modifiable risk factors that challenge systemic VLDL secretion also promotes nonalcoholic ste- and hepatic energy homeostasis have been linked atohepatitis and liver fibrosis, but more research to nonalcoholic steatohepatitis, including shift is needed to determine how lipid retention in- work and alterations in commensal microbiota. cites progressive liver damage. Further research Early studies suggested that intestinal microbiota is also necessary to elucidate the role of other are altered in genetically obese mice with the gene polymorphisms that are associated with metabolic syndrome and fatty livers, and these nonalcoholic steatohepatitis–related liver dam- studies linked the abnormal microbiome with age.23-26 Defining genetic factors that regulate hepatic inflammatory signaling that promotes susceptibility to liver damage is important be- insulin resistance, hepatic steatosis, and nonalco- cause studies in twins indicate that heritable holic steatohepatitis.36-38 Subsequent work showed factors account for about half the interindividual that the gut–liver axis has important bidirec- differences in the prevalence of nonalcoholic tional actions that affect health.39-42 The intes- steatohepatitis with cirrhosis.27 tinal microbiota influence host susceptibility to obesity, hepatic steatosis, nonalcoholic steato- Epigenetic Factors hepatitis, liver fibrosis, and primary liver cancer. Epigenetic events cause heritable changes in gene Conversely, host factors (e.g., diet composition, expression without directly altering the DNA code adiposity, feeding frequency, and sleep–wake itself. For example, epigenetic mechanisms mod- cycles) influence the intestinal microbiota. Early ify chromatin structure, and this change alters models favored abnormal microbiota as a pri- the accessibility of DNA to factors that control mary cause of intestinal permeability defects gene transcription, thereby changing the abun- that expose hosts to noxious gut-derived factors dance of messenger RNA (mRNA). RNA fate is (e.g., bacterial lipopolysaccharide, other toll-like also controlled by other heritable factors (e.g., receptor ligands, and toxic bile acids). However, microRNA and RNA-binding proteins), provid- more recent studies have shown that when host- ing alternative epigenetic strategies for influenc- generated inflammatory signals are insufficient, ing gene expression.28 In mice, both overeating the intestinal microbiota are unable to maintain and undereating during pregnancy induce epi- intestinal barrier functions that are hepatopro- genetic mechanisms that disrupt the insulin-like tective and metabolically favorable.39-42 growth factor axis during fetal development, in- Whether insights gleaned from studies in creasing the susceptibility of offspring to obesity animals can inform efforts to prevent or treat and the metabolic syndrome later in life. Similar nonalcoholic steatohepatitis in humans remains mechanisms have been proposed to explain the an uncertain, but exciting, possibility. Agents that increased incidence of obesity and the metabolic inhibit recruitment of inflammatory cells, block syndrome in children whose mothers were preg- inflammatory signaling, reduce oxidative stress, nant during periods of famine.29-32 Studies of and improve insulin sensitivity in preclinical families with adult-onset obesity have identified models are being tested in patients with non genomewide epigenetic alterations that dysregu- alcoholic steatohepatitis. It remains to be deter- late metabolic pathways controlling adiposity, mined whether more direct manipulation of the insulin sensitivity, and tissue generation or re- intestinal microbiota with antibiotics, prebiotics, generation.31,32 Whether such epigenetic mecha- or probiotics can prevent or treat nonalcoholic nisms influence susceptibility to nonalcoholic steatohepatitis. Profiling nonalcoholic steato- steatohepatitis is being investigated. Studies in hepatitis–related changes in intestinal microbiota

n engl j med 377;21 nejm.org November 23, 2017 2067 The New England Journal of Medicine Downloaded from nejm.org by LUIS ERNESTO GONZALEZ SANCHEZ on December 7, 2017. For personal use only. No other uses without permission. Copyright © 2017 Massachusetts Medical Society. All rights reserved. The new england journal of medicine

is an area of active research that is in its infancy gested that losartan, an adrenergic antagonist, but is benefiting from the recent availability of mitigates nonalcoholic steatohepatitis and fibro- large biobanks of well-characterized human spec- sis in humans.49 imens, as well as advances in RNA sequencing 43 and metagenomics. Diagnostic and Therapeutic Shift work and travel that perturb normal Implications feeding and sleep–wake cycles promote adiposity, the metabolic syndrome, and nonalcoholic fatty Nonalcoholic steatohepatitis and its deadly se- liver disease. Studies in mice have revealed that quelae, cirrhosis and liver cancer, are much less these activities disrupt circadian rhythms at mul- prevalent than isolated hepatic steatosis in the tiple levels (i.e., within the central nervous system, general population. This suggests that most per- liver, and intestinal microbiome), uncovering in- sons with fatty livers are able to avoid, constrain, tegrated mechanisms that normally balance or compensate for stressors that induce nonalco- energy supply and demand.44 Transcriptomic holic steatohepatitis or drive its progression. analysis of circadian clock–regulated liver genes Preclinical research has identified a network of has shown enrichment with transcripts that interacting factors that determine whether non- control lipid metabolism.45 Prolonged disruption alcoholic steatohepatitis develops in fatty livers, of normal circadian rhythms in mice with fatty as well as whether (and if so, how rapidly) fatal livers induces the development of nonalcoholic liver damage ensues. This information is begin- steatohepatitis by dysregulating cross-talk be- ning to be used to stratify persons with fatty tween two nuclear hormone receptors, farnesoid livers into groups that are at higher or lower risk X receptor (FXR) and constitutive androstane for nonalcoholic steatohepatitis. It also provides receptor (CAR), leading to suppression of FXR, a roadmap of potential therapeutic targets, which hepatic accumulation of bile acids, bile acid– can guide the development of pharmacotherapy induced overactivation of CAR, and eventual for nonalcoholic steatohepatitis. CAR-dependent liver injury, fibrosis, and neopla- Epidemiologic data underscore the importance sia.46 Remarkably, the CAR-sensitive liver respons- of differentiating persons with nonalcoholic es are inhibited by blocking β-adrenergic recep- steatohepatitis from those with isolated hepatic tors, which are known mediators of excessive steatosis; only the former group requires resource- sympathetic nervous system activity caused by intensive, liver-targeted interventions to reduce disruption of circadian rhythmicity. the risk of liver-related illness and death. Man- These preclinical findings suggest that liver agement decisions can be refined by staging health depends on an integrated network of meta- liver fibrosis according to its severity. Accurate bolic responses that coordinate energy supply and fibrosis staging is important because biopsy demand. Indeed, dysregulation of FXR, CAR, series show that nonalcoholic steatohepatitis is and adrenergic signaling seems to occur in hu- strongly associated with liver fibrosis, and natu- man nonalcoholic steatohepatitis. FXR is part of ral history studies show that both all-cause and an exquisite bile acid–sensing system that en- liver-specific mortality increase once F2 fibrosis sures the optimal size of the bile acid pool. FXR has developed. The risk of death from liver dis- agonists and certain FXR-regulated factors (e.g., ease increases by a factor of 50 to 80 for patients fibroblast growth factors 19 and 21) are being with nonalcoholic steatohepatitis who have F3 or evaluated as treatments for human nonalcoholic F4 fibrosis, as compared with those who have steatohepatitis. CAR controls hepatic metabolism nonalcoholic steatohepatitis with little or no fi- of xenobiotics and environmental toxins; its ac- brosis.5,6 These caveats do not indicate, however, tivation is required for the carcinogenic actions that isolated hepatic steatosis is entirely benign, of phenobarbital.47 Environmental toxins promote since hepatocyte lipids accumulate in persons both human liver cancer and nonalcoholic steato- who have systemic metabolic stress, and meta- hepatitis48; thus, inhibiting CAR might be benefi- bolic stress increases the risks of cardiovascular cial in patients with nonalcoholic steatohepatitis. disease and cancer — conditions that them- Deleting β-adrenergic receptors abrogates hepatic selves warrant preventive and therapeutic inter- CAR activity in mice,46 and a pilot study has sug- ventions.50-52

2068 n engl j med 377;21 nejm.org November 23, 2017 The New England Journal of Medicine Downloaded from nejm.org by LUIS ERNESTO GONZALEZ SANCHEZ on December 7, 2017. For personal use only. No other uses without permission. Copyright © 2017 Massachusetts Medical Society. All rights reserved. Nonalcoholic Steatohepatitis

Dedicated efforts are necessary to identify Table 1. Lifestyle Modifications to Mitigate Nonalcoholic Steatohepatitis.* persons with nonalcoholic steatohepatitis because neither a diagnosis of the disorder nor the Lose 7% of body weight if overweight or obese severity of associated liver fibrosis is predicted Limit consumption of fructose-enriched beverages simply on the basis of obesity, insulin resistance, Limit consumption of alcohol (≤1 drink/day for women and ≤2 drinks/day or hepatic steatosis.5,6 However, those conditions for men) should trigger a search for factors that increase Drink two or more cups of caffeinated coffee daily the risk of nonalcoholic steatohepatitis, as well as for clues that fibrosis might already be ad- * Fructose increases the odds of the development of nonalcoholic fatty liver in high-risk patients and of nonalcoholic steatohepatitis and more advanced liver vanced. Overweight or obese persons with the fibrosis in patients who already have nonalcoholic fatty liver disease. Caffeinated metabolic syndrome, elevated serum aminotrans- coffee reduces the risk of liver fibrosis in several liver diseases, including non- ferase levels, and a negative noninvasive workup alcoholic fatty liver disease. for other causes of liver disease are likely to have nonalcoholic steatohepatitis. Those who are 45 years of age or older and who have type 2 diabe- inflammatory actions, might also be considered. tes are particularly likely to have advanced fibro- A multicenter clinical trial showed that both sis and an increased risk of bad liver outcomes.53 were superior to placebo for the treatment of Regardless of risk factors for fibrosis, advanced nonalcoholic steatohepatitis in nondiabetic pa- liver fibrosis should be suspected when even tients with mild liver damage and were relatively subtle clinical signs of hepatic dysfunction or safe (although pioglitazone often caused weight portal hypertension are detected. gain that was not easily lost after treatment was Persons at risk for nonalcoholic steatohepatitis stopped).56 That study was not designed to deter- or liver fibrosis should undergo further testing mine the effects on fibrosis progression; hence, to confirm the clinical suspicion of nonalcoholic the efficacy of either agent for preventing or re- steatohepatitis–related liver damage, grade the versing liver fibrosis is unknown. Generally, per- severity of liver injury, and stage the level of fi- sons with nonalcoholic steatohepatitis and F0 or brosis, because this information is essential in F1 fibrosis are evaluated annually for liver dis- formulating plans for disease management. Liver ease progression by means of blood tests and biopsy is currently the most widely accepted ap- physical examinations. More directed screening proach for diagnosing nonalcoholic steatohepati- for fibrosis progression (e.g., with repeat liver tis and staging liver fibrosis. However, less inva- biopsy or newer testing techniques) is often re- sive approaches are being developed that may peated approximately 5 years after diagnosis, but ultimately permit diagnosis and staging of non- the precise timing of such testing is governed by alcoholic steatohepatitis by combining panels of laboratory and examination results because of the serum biomarkers, new imaging tests that can high interindividual variation in rates of fibrosis quantify liver fibrosis, and dynamic tests of liver progression.55 function.54,55 Genetic screening for polymorphisms For patients who have more advanced fibrosis that segregate with a high risk of nonalcoholic (F2 or higher) when nonalcoholic steatohepatitis steatohepatitis, fibrosis, or liver cancer might is diagnosed, management is tailored according further improve diagnostic accuracy.24 to the severity of the fibrosis. Once F3 or F4 fi- Accurate diagnosis and staging of nonalco- brosis occurs, the risk of liver-related illness and holic steatohepatitis are essential for the man- death increases dramatically, justifying much agement of this disorder. Nonalcoholic steato- closer follow-up to detect and treat life-threaten- hepatitis with no fibrosis (F0) or negligible ing complications of portal hypertension. The fibrosis (F1) has an excellent prognosis, and same lifestyle modifications that are advised for thus intensive follow-up and liver-targeted treat- patients who have nonalcoholic steatohepatitis ments are not necessary. Currently, lifestyle mod- with F0 or F1 fibrosis are also recommended for ifications are the main intervention for nonalco- patients with more severe fibrosis. Vigorous ef- holic steatohepatitis without fibrosis (Table 1). forts to determine the most effective methods Adjunctive treatment with vitamin E or piogli- for management of the metabolic syndrome and tazone, an insulin-sensitizing agent with anti- for screening for cancer are important, since

n engl j med 377;21 nejm.org November 23, 2017 2069 The New England Journal of Medicine Downloaded from nejm.org by LUIS ERNESTO GONZALEZ SANCHEZ on December 7, 2017. For personal use only. No other uses without permission. Copyright © 2017 Massachusetts Medical Society. All rights reserved. The new england journal of medicine

Table 2. Pharmacotherapies for Nonalcoholic Steatohepatitis Evaluated fied in patients at lower risk for disease progres- in Phase 2 or 3 Clinical Trials.* sion. Second, therapies that increase the risk of cardiovascular disease or cancer should not be Pharmacologic Agent Therapeutic Target used because these disorders are the leading Metabolic causes of death in patients with nonalcoholic Stress Inflammation Fibrosis steatohepatitis. Third, the heavy societal burden Vitamin E† Yes Yes No imposed by liver damage related to nonalcoholic Pioglitazone (PPAR-γ agonist)† Yes Yes Yes steatohepatitis justifies public health measures aimed at eradicating nonalcoholic steatohepatitis Obeticholic acid (FXR agonist)† Yes Yes Yes and other metabolic syndrome–related diseases Chemokine receptor 2 and 5 No Yes Yes by encouraging healthful eating, facilitating antagonists exercise and fitness, limiting shift work, and PPAR-α and PPAR-δ agonists Yes Yes Yes minimizing exposure to environmental toxins. Lysyl oxidase–like 2 inhibitor No No Yes Galectin 3 No Yes Yes Conclusions Bovine milk colostrum No Yes Yes Stress-activated kinase 1 Yes Yes Yes Nonalcoholic steatohepatitis, a type of liver inhibitor damage that is strongly associated with visceral FGF-21 Yes Yes Yes adiposity and the metabolic syndrome, has be- FGF-19–like agent Yes Yes Yes come a major cause of cirrhosis and liver cancer. The prevalence of nonalcoholic steatohepatitis in * FGF denotes fibroblast growth factor, FXR farnesoid X receptor, and PPAR the United States approaches that of type 2 dia- peroxisome proliferator–activated receptor. betes, and annual medical costs directly attrib- † This agent was superior to placebo in a randomized clinical trial. utable to nonalcoholic fatty liver disease already exceed $100 billion, much of which is attributable to nonalcoholic steatohepatitis, underscor- cardiovascular disease and cancer are the leading the importance of developing interventions ing causes of death in patients with nonalcoholic to prevent and treat this disease. A number of steatohepatitis and cirrhosis. inherited and environmental factors increase the A number of large clinical trials designed to risk of nonalcoholic steatohepatitis and influence identify effective and safe treatments for non- its progression. The pathogenic mechanisms are alcoholic steatohepatitis and liver fibrosis are in being unraveled; metabolic stress, inflammation, progress.57-59 Efforts have focused on ameliorat- and fibrosis have been identified as key pro- ing the three general processes that drive the cesses. Pharmacologic agents that target these pathogenesis and progression of nonalcoholic mechanisms are being investigated in clinical steatohepatitis: metabolic stress, inflammation, trials. and fibrosis. Table 2 lists agents that are in phase 2 or 3 clinical trials. Many of these agents Dr. Diehl reports receiving consulting fees from Pfizer, No- affect more than one of the putative pathogenic vartis, and Lumena, participating in a clinical trial with Gilead Sciences, Conatus, Galmed, NGM Biopharmaceuticals, Bristol- targets; a combination of agents that predomi- Myers Squibb, Madrigal, Galectin Therapeutics, Exalenz Bio- nantly influence distinct targets is also being sciences, Shire, Intercept, and Genfit, receiving grant support considered. Results have been reported for sev- from and participating in a clinical trial with Immuron, receiving grant support for a research collaboration from Metabolon eral studies, but no agent has yet received ap- and Prometheus, receiving consulting fees from and participat- proval by the Food and Drug Administration for ing in a clinical trial with Boehringer Ingelheim, receiving hon- the treatment of nonalcoholic steatohepatitis. oraria from and participating in a clinical trial with Allergan, receiving consulting fees and grant support for a research col- A consensus about general principles of treat- laboration from Celgene, and holding a pending patent applica- ment for nonalcoholic steatohepatitis is emerging. tion for “Development of Novel Therapeutics to Treat Non-Alco- First, relatively risky and expensive treatments holic Steatohepatitis (NASH).” No other potential conflict of interest relevant to this article was reported. merit consideration in patients with a high risk Disclosure forms provided by the authors are available with of bad liver-related outcomes but are not justi- the full text of this article at NEJM.org.

2070 n engl j med 377;21 nejm.org November 23, 2017 The New England Journal of Medicine Downloaded from nejm.org by LUIS ERNESTO GONZALEZ SANCHEZ on December 7, 2017. For personal use only. No other uses without permission. Copyright © 2017 Massachusetts Medical Society. All rights reserved. Nonalcoholic Steatohepatitis

References 1. Younossi ZM, Koenig AB, Abdelatif D, fatty liver, and cirrhosis. Annu Rev Med methylation machinery. Cell Signal 2013; Fazel Y, Henry L, Wymer M. Global epi- 2016;67: 103-17. 25:1837-44. demiology of nonalcoholic fatty liver dis- 14. Berentzen TL, Gamborg M, Holst C, 29. Gabory A, Attig L, Junien C. Develop- ease — meta-analytic assessment of Sørensen TI, Baker JL. Body mass index in mental programming and epigenetics. Am prevalence, incidence, and outcomes. Hep- childhood and adult risk of primary liver J Clin Nutr 2011;94:Suppl:1943S-1952S. atology 2016;64: 73-84. cancer. J Hepatol 2014;60:325-30. 30. Godfrey KM, Costello PM, Lillycrop 2. Williams CD, Stengel J, Asike MI, et al. 15. Younossi ZM, Blissett D, Blissett R, KA. Development, epigenetics and meta- Prevalence of nonalcoholic fatty liver dis- et al. The economic and clinical burden bolic programming. Nestle Nutr Inst ease and nonalcoholic steatohepatitis of nonalcoholic fatty liver disease in the Workshop Ser 2016;85: 71-80. among a largely middle-aged population United States and Europe. Hepatology 31. Lee JH, Friso S, Choi SW. Epigenetic utilizing ultrasound and liver biopsy: 2016;64: 1577-86. mechanisms underlying the link between a prospective study. Gastroenterology 16. Yamaguchi K, Yang L, McCall S, et al. non-alcoholic fatty liver diseases and nu- 2011;140: 124-31. Diacylglycerol acyltranferase 1 anti-sense trition. Nutrients 2014;6: 3303-25. 3. Adams LA, Harmsen S, St Sauver JL, oligonucleotides reduce hepatic fibrosis 32. Lee YS. Genetics of nonsyndromic obe- et al. Nonalcoholic fatty liver disease in- in mice with nonalcoholic steatohepatitis. sity. Curr Opin Pediatr 2013;25:666-73. creases risk of death among patients with Hepatology 2008;47:625-35. 33. Page A, Mann DA, Mann J. The mech- diabetes: a community-based cohort study. 17. Heymann F, Tacke F. Immunology in anisms of HSC activation and epigenetic Am J Gastroenterol 2010;105:1567-73. the liver — from homeostasis to disease. regulation of HSCs phenotypes. Curr 4. Younossi ZM, Stepanova M, Rafiq N, Nat Rev Gastroenterol Hepatol 2016;13: Pathobiol Rep 2014;2:163-70. et al. Pathologic criteria for nonalcoholic 88-110. 34. Mann DA. Epigenetics in liver disease. steatohepatitis: interprotocol agreement 18. Arrese M, Cabrera D, Kalergis AM, Hepatology 2014;60:1418-25. and ability to predict liver-related mortal- Feldstein AE. Innate immunity and in- 35. Hardy T, Zeybel M, Day CP, et al. Plasma ity. Hepatology 2011;53:1874-82. flammation in NAFLD/NASH. Dig Dis Sci DNA methylation: a potential biomarker 5. Ekstedt M, Hagström H, Nasr P, et al. 2016;61: 1294-303. for stratification of liver fibrosis in non- Fibrosis stage is the strongest predictor 19. Ji C, Kaplowitz N, Lau MY, Kao E, alcoholic fatty liver disease. Gut 2017;66: for disease-specific mortality in NAFLD Petrovic LM, Lee AS. Liver-specific loss of 1321-8. after up to 33 years of follow-up. Hepatol- glucose-regulated protein 78 perturbs the 36. Cope K, Risby T, Diehl AM. Increased ogy 2015;61:1547-54. unfolded protein response and exacer- gastrointestinal ethanol production in 6. Angulo P, Kleiner DE, Dam-Larsen S, bates a spectrum of liver diseases in mice. obese mice: implications for fatty liver et al. Liver fibrosis, but no other histo- Hepatology 2011;54:229-39. disease pathogenesis. Gastroenterology logic features, is associated with long- 20. Diehl AM. Lessons from animal mod- 2000;119:1340-7. term outcomes of patients with nonalco- els of NASH. Hepatol Res 2005;33:138-44. 37. Nair S, Cope K, Risby TH, Diehl AM. holic fatty liver disease. Gastroenterology 21. Rangwala F, Guy CD, Lu J, et al. In- Obesity and female gender increase 2015;149(2):389-97.e10. creased production of sonic hedgehog by breath ethanol concentration: potential 7. Angulo P, Machado MV, Diehl AM. ballooned hepatocytes. J Pathol 2011;224: implications for the pathogenesis of non- Fibrosis in nonalcoholic fatty liver dis- 401-10. alcoholic steatohepatitis. Am J Gastroen- ease: mechanisms and clinical implica- 22. Guy CD, Suzuki A, Abdelmalek MF, terol 2001;96: 1200-4. tions. Semin Liver Dis 2015;35:132-45. Burchette JL, Diehl AM. Treatment re- 38. Li Z, Yang S, Lin H, et al. Probiotics 8. Hardy T, Oakley F, Anstee QM, Day sponse in the PIVENS trial is associated and antibodies to TNF inhibit inflamma- CP. Nonalcoholic fatty liver disease: patho- with decreased Hedgehog pathway activ- tory activity and improve nonalcoholic genesis and disease spectrum. Annu Rev ity. Hepatology 2015;61:98-107. fatty liver disease. Hepatology 2003;37: Pathol 2016;11: 451-96. 23. Severson TJ, Besur S, Bonkovsky HL. 343-50. 9. Charlton MR, Burns JM, Pedersen RA, Genetic factors that affect nonalcoholic 39. Machado MV, Cortez-Pinto H. Diet, Watt KD, Heimbach JK, Dierkhising RA. fatty liver disease: a systematic clinical microbiota, obesity, and NAFLD: a dan- Frequency and outcomes of liver trans- review. World J Gastroenterol 2016;22: gerous quartet. Int J Mol Sci 2016;17:481. plantation for nonalcoholic steatohepati- 6742-56. 40. Federico A, Dallio M, Godos J, Logu- tis in the United States. Gastroenterology 24. Anstee QM, Seth D, Day CP. Genetic ercio C, Salomone F. Targeting gut-liver 2011;141:1249-53. factors that affect risk of alcoholic and axis for the treatment of nonalcoholic 10. Wong RJ, Aguilar M, Cheung R, et al. nonalcoholic fatty liver disease. Gastro- steatohepatitis: translational and clinical Nonalcoholic steatohepatitis is the sec- enterology 2016;150(8):1728-44.e7. evidence. Transl Res 2016;167:116-24. ond leading etiology of liver disease 25. Dongiovanni P, Valenti L. Genetics of 41. Kirpich IA, Marsano LS, McClain CJ. among adults awaiting liver transplanta- nonalcoholic fatty liver disease. Metabo- Gut-liver axis, nutrition, and non-alcoholic tion in the United States. Gastroenterology lism 2016;65:1026-37. fatty liver disease. Clin Biochem 2015;48: 2015;148:547-55. 26. Kahali B, Halligan B, Speliotes EK. 923-30. 11. Welzel TM, Graubard BI, Quraishi S, Insights from genome-wide association 42. Vanni E, Bugianesi E. The gut-liver et al. Population-attributable fractions of analyses of nonalcoholic fatty liver dis- axis in nonalcoholic fatty liver disease: risk factors for hepatocellular carcinoma ease. Semin Liver Dis 2015;35:375-91. another pathway to insulin resistance? in the United States. Am J Gastroenterol 27. Loomba R, Schork N, Chen CH, et al. Hepatology 2009;49:1790-2. 2013;108: 1314-21. Heritability of hepatic fibrosis and steato- 43. Boursier J, Mueller O, Barret M, et al. 12. Wong CR, Nguyen MH, Lim JK. Hepato- sis based on a prospective twin study. The severity of nonalcoholic fatty liver cellular carcinoma in patients with non- Gastroenterology 2015;149:1784-93. disease is associated with gut dysbiosis alcoholic fatty liver disease. World J Gas- 28. Bian EB, Zhao B, Huang C, et al. New and shift in the metabolic function of the troenterol 2016;22: 8294-303. advances of DNA methylation in liver fi- gut microbiota. Hepatology 2016;63:764- 13. Marengo A, Rosso C, Bugianesi E. brosis, with special emphasis on the 75. Liver cancer: connections with obesity, crosstalk between microRNAs and DNA 44. Dibner C, Gachon F. Circadian dys-

n engl j med 377;21 nejm.org November 23, 2017 2071 The New England Journal of Medicine Downloaded from nejm.org by LUIS ERNESTO GONZALEZ SANCHEZ on December 7, 2017. For personal use only. No other uses without permission. Copyright © 2017 Massachusetts Medical Society. All rights reserved. Nonalcoholic Steatohepatitis

function and obesity: is leptin the miss- hepatic manifestations of nonalcoholic AJ, Dimick-Santos L. Challenges and op- ing link? Cell Metab 2015;22:359-60. fatty liver disease. Curr Hepatol Rep 2016; portunities in drug and biomarker devel- 45. Fleet T, Stashi E, Zhu B, et al. Genetic 15:75-85. opment for nonalcoholic steatohepatitis: and environmental models of circadian 51. Satapathy SK, Sanyal AJ. Epidemiol- findings and recommendations from an disruption link SRC-2 function to hepatic ogy and natural history of nonalcoholic American Association for the Study of pathology. J Biol Rhythms 2016;31:443-60. fatty liver disease. Semin Liver Dis 2015; Liver Diseases-U.S. Food and Drug Ad- 46. Kettner NM, Voicu H, Finegold MJ, et al. 35:221-35. ministration joint workshop. Hepatology Circadian homeostasis of liver metabo- 52. Lonardo A, Sookoian S, Chonchol M, 2015;61:1392-405. lism suppresses hepatocarcinogenesis. Loria P, Targher G. Cardiovascular and 56. Sanyal AJ, Chalasani N, Kowdley KV, Cancer Cell 2016;30:909-24. systemic risk in nonalcoholic fatty liver et al. Pioglitazone, vitamin E, or placebo 47. Yamamoto Y, Moore R, Goldsworthy disease — atherosclerosis as a major player for nonalcoholic steatohepatitis. N Engl J TL, Negishi M, Maronpot RR. The orphan in the natural course of NAFLD. Curr Med 2010;362:1675-85. nuclear receptor constitutive active/andro- Pharm Des 2013;19:5177-92. 57. Noureddin M, Zhang A, Loomba R. stane receptor is essential for liver tumor 53. Hossain N, Afendy A, Stepanova M, Promising therapies for treatment of non- promotion by phenobarbital in mice. Can- et al. Independent predictors of fibrosis alcoholic steatohepatitis. Expert Opin cer Res 2004;64: 7197-200. in patients with nonalcoholic fatty liver Emerg Drugs 2016;21:343-57. 48. Wahlang B, Beier JI, Clair HB, et al. disease. Clin Gastroenterol Hepatol 2009; 58. Brodosi L, Marchignoli F, Petroni ML, Toxicant-associated steatohepatitis. Toxi- 7:1224-9. Marchesini G. NASH: a glance at the col Pathol 2013;41:343-60. 54. Golabi P, Sayiner M, Fazel Y, Koenig A, landscape of pharmacological treatment. 49. Yokohama S, Yoneda M, Haneda M, Henry L, Younossi ZM. Current complica- Ann Hepatol 2016;15:673-81. et al. Therapeutic efficacy of an angioten- tions and challenges in nonalcoholic ste- 59. Lassailly G, Caiazzo R, Pattou F, sin II receptor antagonist in patients with atohepatitis screening and diagnosis. Ex- Mathurin P. Perspectives on treatment nonalcoholic steatohepatitis. Hepatology pert Rev Gastroenterol Hepatol 2016;10: for nonalcoholic steatohepatitis. Gastro- 2004;40:1222-5. 63-71. enterology 2016;150: 1835-48. 50. VanWagner LB, Rinella ME. Extra 55. Sanyal AJ, Friedman SL, McCullough Copyright © 2017 Massachusetts Medical Society.

images in clinical medicine The Journal welcomes consideration of new submissions for Images in Clinical Medicine. Instructions for authors and procedures for submissions can be found on the Journal’s website at NEJM.org. At the discretion of the editor, images that are accepted for publication may appear in the print version of the Journal, the electronic version, or both.

2072 n engl j med 377;21 nejm.org November 23, 2017 The New England Journal of Medicine Downloaded from nejm.org by LUIS ERNESTO GONZALEZ SANCHEZ on December 7, 2017. For personal use only. No other uses without permission. Copyright © 2017 Massachusetts Medical Society. All rights reserved.