Epidemiology Risk Factors Natural History

Published Date: 02 Nov 2017

Non-alcoholic fatty liver disease (NAFLD) is one of the most common chronic liver diseases all over the world 1. It includes a wide disease spectrum from simple steatosis to non-alcoholic steatohepatitis (NASH), which is the active form of NAFLD. While simple steatosis usually exhibits a benign process, NASH may progress to end-stage liver disease such as cirrhosis, liver failure and hepatocellular carcinoma 2-3.

The gold standard for diagnosing of NAFLD and NASH is liver biopsy. However, it is an invasive test. The risk for major complications and sampling bias cannot be avoided. Moreover, it is not very suitable for repeated evaluation. Several non-invasive approaches have been introduced, including different imaging technologies and biomarkers. Non-invasive imaging methods are mainly used to determine the grade of steatosis. Physical measurements correlated well with fibrosis. Clinical tests, biomarkers and together with several established prediction scores, provided potential options in distinguishing NASH and fibrosis; however, most of them still need to be further validated. With the advances in genome analysis, genetic determinants of NAFLD are also widely studied in the recent years using Genome-wide association study (GWAS). Several genetic determinants of NAFLD, including genetic variants in Patatin-like phospholipase domain-containing protein 3 (PNPLA3), Apolipoprotein C3 (APOC3), Farnesyl diphosphate farnesyl transferase 1 (FDFT1) were identified. Non-invasive evaluation of NAFLD should not only be limited to find surrogates for liver biopsy, but also should be aimed to explore their predicting value in predicting disease prognosis.

The management of patients with NAFLD should not be limited to treating liver disease. Metabolic disorders, which are frequently associated with NAFLD, should be treated at the same time. Multiple modalities including lifestyle intervention, medical treatment as well as bariatric surgery are available for the management of NAFLD 3.

In this Chapter, we will review the epidemiology, natural history, pathogenesis, the progress of non-invasive evaluation and genetic determinants, and management of NAFLD in adult patients.

1.1 Definition, epidemiology, risk factors, natural history and pathogenesis of non-alcoholic fatty liver disease

1.1.1 Definition

Non-alcoholic fatty liver disease (NAFLD) is one of the most common chronic liver diseases all over the world 1-2. It is defined as the presence of hepatic steatosis with the absence of secondary causes for hepatic fat accumulation which are summarized in Table 1.1 3. It includes a wide disease spectrum from simple steatosis to non-alcoholic steatohepatitis (NASH). Patients with simple steatosis have only hepatic steatosis with no evidence of hepatic injury. NASH, which is the active form of NAFLD, is defined as hepatic steatosis with lobular inflammation and hepatocyte injury in the form of ballooning. NAFLD can affect both adults and children. Here we mainly focus on NAFLD in adults.

1.1.2 Epidemiology

The prevalence of NAFLD also differs widely depending on the study population and different diagnostic methods. Liver biopsy, which is the gold standard, is an invasive approach and cannot be adopted in population-based studies. Thus, prevalence of histological confirmed NAFLD can only be assessed from potential liver donors. Marcos et al. reported 11% (14/126) of living liver donors had >30% hepatic steatosis in the United States, which accounted for 20% of the excluded candidates 4; while Lee et al. from Korea reported that 51% (303/589) of living liver donors had NAFLD defined as ≥5% steatosis in biopsy, including 10% (61/589) had >30% steatosis 5.

There are several non-invasive approached can be used to estimate the prevalence of NAFLD in general population. Ultrasound has satisfying sensitivity and specificity in detecting moderate to severe hepatic steatosis; however it is not reliable when the amount of steatosis is less than 30% 6. The prevalence of NAFLD defined by ultrasound is ranged between 17% and 46% 7-13. A population based study in India found 194 NAFLD in 1,168 subjects (17%) 7; while another study from Brooke Army Medical Center, United States, reported 151 NAFLD in 328 subjects recruited from the clinic (46%) 8. Proton-magnetic resonance spectroscopy (1H-MRS) quantifies hepatic steatosis which correlated well with the degree of steatosis by histology, and show superior accuracy than ultrasound 14. In Dallas Heart Study, the prevalence of NAFLD was found to be 31% in 2,287 subjects 15-16. A community-based study in Hong Kong revealed a NAFLD prevalence of 27% in 922 Chinese subjects 17. Elevated aminotransferases is also an indication of suspected NAFLD. However, it can be normal in NAFLD patients and correlated poorly with histological findings 18. The worldwide prevalence of NAFLD estimated by aminotransferases alone ranged from 3% to 21% 3. The estimated prevalence of NAFLD is summarized in Table 1.2.

On the other hand, the definitive diagnosis of NASH relies on liver biopsy. Two studies mention above have reported the prevalence of histological confirmed NASH. Lee et al. from Korea reported 2.2% (13/589) of potential liver donors had NASH 5. In the study from Brook Army Medical Center, a subgroup of 134 ultrasound diagnosed NAFLD patients received liver biopsies, 40 of them was diagnosed as NASH. However, since all subjects in this study were recruited from the clinic, the prevalence of NASH might be overestimated.

Taking together, the estimated worldwide prevalence of NAFLD is ranged from 6.3 to 33%. The estimated worldwide prevalence of NASH is ranged from 3 to 5% 3.

The accurate incidence rate of NAFLD remains unknown. The reported incidence rate of NAFLD ranged widely from 29 cases per 100,000 person-years to 86 cases in 1000 person-years 19-21. The large discrepancy clearly suggests that further studies are required to determine the accurate incidence of NAFLD across different ethic and geographic populations.

1.1.3 Risk factors for NAFLD

Male gender and Hispanic ethnicity are associated with higher prevalence of NAFLD 7-8 10-11 15-17. In a population-based study in which NAFLD was defined by 1H-MRS, Hispanics had an median intrahepatic triglyceride (IHTG) of 4.6%, significantly higher than whites (3.6%) and blacks (3.2%), p < 0.00115. The prevalence of NAFLD in Hispanics, whites and blacks was 45%, 33% and 31%, respectively. White males had significant higher prevalence of NAFLD compared with white females (42% vs. 24%). Moreover, several studies have shown the prevalence of NAFLD was higher in older patients 7 10 22-23. However, older patients also have more other NAFLD risk factors such as metabolic syndrome. It remains unclear whether the higher incidence in older patients is due to the duration of disease or age itself 2.

Metabolic syndrome (MS) is a well-recognized risk factor for NAFLD 3. In a joint interim statement of the International Diabetes Federation Task Force on Epidemiology and Prevention; National Heart, Lung, and Blood Institute; American Heart Association; World Heart Federation; International Atherosclerosis Society; and International Association for the Study of Obesity, metabolic syndrome is defined as the existence of any three of the following: (1) Central obesity; (2) Hypertriglyceridemia; (3) Reduced high-density lipoprotein-cholesterol; (4) Elevated blood pressure; (5) Impaired fasting plasma glucose; or receiving treatment for the above metabolic abnormalities 24. NAFLD is very common in obese individuals. In morbidly obese patients who received bariatric surgeries, the prevalence of NAFLD can be as high as 98% 25-30. Notably, the prevalence of NASH in obese patients ranges from 10 to 56% with a median of 33% 2. 67% of these patients have portal fibrosis 2; 5% of them may even have unsuspected cirrhosis 3. Type 2 diabetes mellitus (T2DM) also has a close relationship with NAFLD. Using both ultrasound and 1H-MRS to evaluate 939 patients from Edinburgh Type 2 Diabetes Study, 42.6% of them were diagnosed as NAFLD 31. Leite and colleagues found the prevalence of NAFLD defined using ultrasound was 69% in 180 T2DM patients in Brazil 32; association of NAFLD with central obesity and hypertriglyceridemia was also observed. In another study from India, Prashanth et al. evaluated 204 T2DM patients and found 127 (62%) had hepatic fatty infiltration on ultrasound. Furthermore, 90 of these 127 patients received liver biopsy, of whom 87% had histologically confirmed NAFLD, 63% had NASH and 37% had fibrosis 33. Hypertriglyceridemia and low serum HDL levels are very common in NAFLD patients. Approximately, 50% of the subjects with dyslipidemia have NAFLD 34.

Sleep apnea and polycystic ovary syndrome (PCOS) are also suggested to be risk factors for NAFLD and NASH 3. Several independent studies reported that sleep apnea was associated with elevated aminotransferases and histological severity in NAFLD patients 35-37. A meta-analysis which pooled 2,183 subjects from 18 cross-sectional studies revealed that sleep apnea at least doubles the risk of NAFLD, NASH and fibrosis 38. NAFLD is found in a great portion of patients with PCOS, and vice versa 39-45. In a study which included 41 patients with PCOS and 31 age- and body-mass index (BMI)-matched control subjects, PCOS patients had significantly higher prevalence of NAFLD (41% vs. 19%) and insulin resistance (63% vs. 35.5%) 43. In another study, Brzozowska et al. screened 14 consecutive female NAFLD patients, 10 (71%) of them had PCOS 42.

1.1.4 Natural history

1.1.4.1 Survival and mortality

While simple steatosis usually exhibits a benign process, its advanced form, NASH, may progress to end-stage liver disease such as cirrhosis, liver failure and hepatocellular carcinoma 2-3 46-47. NAFLD patients have increased overall mortality compared to healthy controls. The leading cause of NAFLD patients are cardiovascular disease, malignancy and liver-related death 2-3. In a study based on the Third National Health and Nutrition Examination Survey (NHANES III) and its Linked Mortality File in the United States, Ong showed the overall mortality was significantly higher in 817 NAFLD patients compared to 10,468 control subjects after adjusting for age, gender, race, education, income, BMI, hypertension disease, and diabetes (hazard ratio [HR]: 1.038; p < 0.001) with a median follow-up of 8.7 years. Liver-related mortality was even higher (HR: 9.32; p < 0.001) 48. In another study which followed up 420 NAFLD patients with a mean duration of 7.6 years, the overall survival was lower than the expected survival for the general population (HR for mortality: 1.34; p = 0.03). Liver disease was the third leading cause of death, as compared with the thirteenth leading cause of death in the general population 49. Furthermore, histology based studies suggest that the increased mortality are attributed to NASH patients alone other than all NAFLD patients. In a study from Sweden, 71 NASH and 58 non-NASH patients were followed up for 13.7 years. The overall survival of NAFLD was significantly lower than control population (78% vs. 84%, p = 0.006). However, non-NASH patients had similar survival compared to the corresponding reference population. In contrast, NASH patients had significant lower survival compared to its reference population (70% vs. 80%, p = 0.01) 50. Also from Sweden, another study reported similar results that increased risk of death was only observed in patients with NASH after a 28-years follow-up 51. Moreover, Matteoni and Rafiq both revealed the increased liver-related mortality in NASH patients compared to non-NASH patients using a same cohort 52-53. All above data suggest that patients with NASH are at risk of increased mortality compared to general population.

1.1.4.2 Disease progression

Patterns and risk factors of disease progression in patients with different stages of NAFLD can be revealed by paired liver biopsy studies. In one study, 12 patients with non-NASH NAFLD received paired liver biopsy after 11 years apart from the initial evaluation due to abnormal results of liver blood tests. None of them developed NASH 47. However, a very different result was reported by showing that after 5 years, inflammation and ballooning were developed in all patients initially diagnosed as simple steatosis 54. In a large scale prospective study, Wong et al. followed-up 52 biopsy-proven NAFLD patients and performed a second liver biopsy 3 years apart from first assessment 46. A semi-quantitative score, NAFLD activity score (NAS) was used to describe the changes in disease status. At baseline, 13 patients had simple steatosis defined as NAS < 3. At month 36, 5 (39%) developed borderline NASH (defined as NAS = 3-4); 3 (23%) developed NASH (defined as NAS ≥ 5). 22 patients had borderline NASH at baseline and 5 (23%) of whom developed NASH at month 36. Interestingly, 2 (15%) simple steatosis patients regressed to normal liver; 4 (18%) borderline NASH patients regressed to simple steatosis; 1 (6%) and 6 (35) NASH patients regression to simple steatosis or borderline NASH, respectively. Reduction in body mass index and waist circumference were identified as independent predicting factors for non-progressive disease. Dysregulation of adipokines (adiponectin, tumour necrosis factor a, interleukin 6 and leptin) were not associated with progression of NAFLD, although it is generally considered to be associated with inflammation.

1.1.4.3 Fibrosis progression

NAFLD, especially NASH, can develop fibrosis. In Western countries, burnt-out NASH is considered to be the leading etiology of cryptogenic cirrhosis 2 55. Advanced fibrosis and its complications contribute to the increased liver-related mortality in NAFLD patients. The presence of fibrosis in NAFLD patients ranges from 38% to 72%; the presence of advanced fibrosis range from 9% to 27% (Table 1.3) 28 56-64. Different risk factors have been identified in cross-sectional studies including older age, obesity, metabolic syndrome, abnormal aminotransferase, inflammation and ballooning degeneration presenting in biopsy (Table 1.3). Risk factors of fibrosis progression are also identified by paired liver biopsy studies. In the study by Wong et al. 46, 26 of 52 (50%) patients had fibrosis at baseline. After 3 years, 14 (27%) patients had fibrosis progression, including 5 (10%) patients had fibrosis progression by at least 2 stages. 25 (48%) had stable disease and 13 (25%) patients had regression of fibrosis. Increase in waist circumference and high baseline low-density lipoprotein-cholesterol were independently associated with fibrosis progression. Adams and colleagues performed serial liver biopsy in 103 patients in a mean follow-up of 3.2 years 65. 38 (37%) patients had fibrosis progression, including 14 (14%) patients had fibrosis progression by at least 2 stages. 35 (34%) had stable disease and 30 (29%) had regression of fibrosis. History of diabetes, earlier fibrosis stage in the initial evaluation and higher BMI were independent risk factors associated with fibrosis progression. Notably, in the study by Adams, 50 patients were treated by ursodiol or clofibrate, although the treatment did not lead to significant change of fibrosis stage compared with placebo 65. Other paired liver studies do not have enough statistical power to identify independent risk factors due to limited case numbers or follow-up duration 50 62 66-72. Argo et al. summarized ten studies comprising 221 patients and performed pooled-analysis 50 62 65-72. Their data showed that age and inflammation on initial biopsy are independent predictors of progression to advanced fibrosis 55. The discrepancy indicates that the risk factors for fibrosis progression in NAFLD patients were still not well recognized. Large-scale prospective study is needed.

1.1.4.4 NAFLD and hepatocellular carcinoma

NAFLD patients are at increased risk for hepatocellular carcinoma (HCC) 73-76. HCC could be a complication of advanced fibrosis and cirrhosis. Up to 27% of patients with NASH-related cirrhosis develop HCC 77. Studies comparing HCC incidence in NAFLD and hepatitis C virus (HCV) patients showed discrepant results. While some studies showed lower risk for HCC in NAFLD patients compared with HCV patients 78-80, other studies showed comparable risk in both groups 76 81. Moreover, indirect evidence suggested NAFLD itself could promote HCC development independent of cirrhosis 82. Obesity and diabetes are closely associated with NAFLD. Experimentally, both genetic and dietary obesity could promote HCC tumorigenesis and growth in an animal model 83. A meta-analysis which pooled 13 cohort studies suggested diabetes can promote HCC before the development of cirrhosis 84. In two studies in which patients with cirrhosis had been excluded, diabetes was still an independent risk factor for HCC 85-86. In a study which compared metabolic syndrome (MS) related HCC patients with other chronic liver disease related HCC patients, the background liver had significantly less advance fibrosis in MS related HCC patients 87. Moreover, some of the MS related malignant tumors were transformed from benign tumors.

1.1.5 Pathogenesis

Hepatic steatosis arises from the abnormal accumulation of triglycerides (TG) in the liver as a result of an imbalance between TG acquisition and removal 77. The factors which initiate NAFLD pathogenesis and promote simple steatosis to NASH are not fully understood. The classic "Two-hit" hypothesis for the pathogenesis of NAFLD is introduced and modified by Day 88-89. The "first hit" is steatosis, which is closely associated with obesity and insulin resistance. It increases the sensitivity of liver to a combination of both environmental and genetic "second hits", which leads to NASH and fibrosis.

1.1.5.1 The pathogenesis of hepatic steatosis

The fatty acids used for hepatic TG formation are derived from diet, de novo synthesis and adipose tissue 77. Accumulation of TG in the liver can arise from disorders in all three ways 90-91. Obese individuals have increased supply of fatty acids compared with lean individuals which may contribute to the NAFLD pathogenesis. Insulin resistance also plays an important role in the development of hepatic steatosis. Insulin promotes lipogenesis in the liver even in the presence of insulin resistance 92. Hyperinsulinemia, as a consequence of insulin resistance, causes hepatic steatosis in different animal models 93. Furthermore, Semple et al. 94 demonstrated that patients with mutations in AKT2, which plays a key role in insulin signaling pathway, result in elevated liver fat content.

Genetic disorders also contribute to the pathogenesis of hepatic steatosis. Inherited disease such as glycogen storage disease type 1a and citrin deficiency can independently lead to severe hepatic steatosis 95-96. A missense mutation in patatin-like phospholipase domain–containing 3 gene (PNPLA3) rs738409 was found to be associated with NAFLD in a genome wide association study and was robustly validated in independent cohorts 97-98. The underlying mechanism of PNPLA3 rs738409 in pathogenesis of steatosis will be further discussed in 1.3.1.

Dietary pattern is also associated with hepatic steatosis 77. For example, the increased consumption of fructose is parallel with increased prevalence of NAFLD. Unlike glucose, fructose cannot be used to synthesize glycogen; instead, it is converted to glyceraldehyde-3-phosphate, providing substrate for de novo lipogenesis. A high fructose diet is used to induce NAFLD in animal models 93. In a study which included 49 biopsy-proven NAFLD patients and 24 age/gender/BMI-matched controls, consumption of fructose in NAFLD patients was over 2-folds compared with control subjects. Hepatic lipogenesis was also increased in NAFLD patients indicated by higher level of hepatic mRNA level of fatty acid synthase 99. Impaired recovery of hepatic adenosine triphosphate (ATP) induced by fructose may also contribute to the development and progression of NAFLD 100.

Emerging evidence has suggested the link between gut microbiota and pathogenesis of hepatic steatosis 101. Transplantation of normal microbiota to germ-free mice leaded to over 2-folds increase in hepatic steatosis and development insulin resistance with reduced food intake 102. Gut microbiota composition is also associated with the presence of obesity in human, which is a major risk factor for NAFLD 101.

1.1.5.2 NAFLD progression

The progression of NAFLD involves the development of inflammation, hepatocyte damage and fibrosis. Dysregulated secretion of cytokines and adipokines is closely associated with the disease progression of NAFLD. Increased pro-inflammatory cytokine such as tumor necrosis factor α (TNF-α) and interleukin 6 (IL-6), and decreased anti-inflammatory cytokine such as adiponectin, is observed in both NASH animal models and human NASH patients compared with those with simple steatosis 77 103. The dysregulation of cytokines may be induced by insulin resistance and hepatic lipotoxicity of free fatty acids 103. Excessive endoplasmic reticulum (ER) stress also contributes to NAFLD progression through activating inflammatory pathways such as nuclear factor κB, c-Jun N-terminal kinase and oxidative stress pathways 104.

Apoptosis is a predominant feature of NASH 105. Feldstein et al. clearly demonstrated that hepatic steatosis increases Fas-mediated hepatocyte apoptosis 106. It is not only associated with the development of NASH but also fibrosis 107. Mitochondrial dysfunction is also activated by Fas-mediated signaling pathway 105. Excessive reactive oxygen species (ROS) generated by mitochondrial dysfunction and apoptosis may further exacerbate inflammation and tissue injury.

The genetic variants in PNPLA3 rs738409 is also associated with disease severity of NAFLD 98. Its role in promoting NAFLD progression will be further discussed in 1.3.1.

Gut microbiota is recently suggested to be an extrahepatic factor which can promote NAFLD progression. It may promote NASH through promoting obesity by improving energy yield; regulating gut permeability; inducing low-grade inflammation through increasing endotoxin production and activation of Toll-like-receptor 4 (TLR-4) signaling; causing immune imbalance; modulating choline and bile acid metabolism; and increasing endogenous ethanol 101.

Table 1.1. Common causes of secondary hepatic steatosis in adults

Nutrition related

Significant alcohol consumption

Parenteral nutrition

Starvation

Viral hepatitis

Hepatitis C

Genetic and metabolic

Wilson's disease

Lipodystrophy

Abetalipoproteinemia

Hypothyroidism

Hypopituitarism

Hypogonadism

Obstetric

Acute fatty liver of pregnancy

HELLP syndrome

Medications

Corticosteroids

Tamoxifen

Amiodarone

Methotrexate

Anti-retroviral drugs

Table 1.2. Population prevalence of NAFLD

Author

Year

Country

Diagnostic method

NAFLD

Total number

Prevalence

Marcos 4

2000

Untied States

Biopsy (>30% steatosis)

14

126

11%

Lee 5

2007

Korea

Biopsy (>5% steatosis)

303

589

51%

Browning 15

2004

United States

1H-MRS (IHTG >5.5%)

708

2,287

31%

Wong 17

2012

China

1H-MRS (IHTG >5%)

252

922

27%

Kojima 12

2003

Japan

Ultrasound

N.A.

N.A.

30%

Bedogni 9

2005

Italy

Ultrasound

135

598

23%

Amarapurkar 7

2007

India

Ultrasound

194

1,168

17%

Caballeria 11

2010

Spain

Ultrasound

198

766

26%

Williams 8

2011

Untied States

Ultrasound

151

328

46%

Hu 10

2012

China

Ultrasound

2,730

7,152

38%

Clark 108

2003

United States

Aminotransferase

N.A.

15,676

7.9%

Ruhl 109

2003

United States

Aminotransferase

N.A.

5,724

2.8%

Patt 110

2003

United States

Aminotransferase

N.A.

1,309

21%

Suzuki 20

2005

Japan

Aminotransferase

143

1,537

9.3%

Ioannou 111

2006

United States

Aminotransferase

N.A.

6,823

7.3%

N.A.: Not available

Table 1.3. Prevalence and risk factors of fibrosis in NAFLD patients

Author

Year

Country

Total number

Fibrosis

Advanced fibrosis

Risk factor identified

Angulo 57

1999

United States

144

107 (74%)

39 (27%)

Age ≥50*

BMI ≥31.1 (Male)/32.5 (Female)*

Diabetes mellitus*

AST/ALT ratio > 1*

Garcı´a-Monzo´n 56

2000

Spain

46

28 (61%)

4 (9%)

Age ≥50

Ratziu 62

2000

France

93

59 (63%)

28 (30%)

Age ≥50*

BMI ≥28*

Necroinflammation present*

Marchesini 61

2003

Italy

163

111 (68%)

34 (21%)

Metabolic syndrome*

Gramlich 59

2004

United States

132

50 (38%)

28 (21%)

Hepatocyte ballooning

Mallory bodies present

Ong 28

2005

United States

197

N.A.

17 (9%)

Waist-hip ratio*

Abnormal AST*

Focal necrosis*

Kleiner 60

2005

United States

576

225 (39%)

121 (21%)

N.A.

Wong 64

2009

China

173

113 (65%)

19 (11%)

Age*

Fasting glucose*

HOMA-IR*

Wong 63

2010

China/France

246

176 (72%)

56 (23%)

N.A.

Brunt 58

2011

United States

934

677 (72%)

215 (23%)

N.A.

N.A.: Not available; BMI, body-mass index; AST, aspartate aminotransferase; ALT, alanine aminotransferase; HOMA-IR: Homeostasis model assessment of insulin resistance. * Risk factors for advanced fibrosis, others for fibrosis

1.2 Evaluation of non-alcoholic fatty liver disease

The gold standard for diagnosing of NAFLD and NASH is liver biopsy. Several histological evaluation systems can be used for NAFLD staging. However, it is an invasive test. The risk for major complications and sampling bias cannot be avoided. Moreover, it is not very suitable for repeated evaluation. Several non-invasive approaches have been introduced, including different imaging technologies and biomarkers.

1.2.1 Liver biopsy

Traditionally, liver biopsy is the gold standard for the diagnosis and assessment of NAFLD 112. It is used to determine the degree of hepatic steatosis and exclude other fatty liver disease such as autoimmune hepatitis. It is also used to assess the presence of necroinflammation and hepatocyte injury, which are the key feature of NASH 113. Fibrosis can also be evaluated and staged 113.

A high quality liver biopsy sample is essential for histological evaluation 112-113. Adequacy of a liver biopsy sample can be assessed grossly by its length and diameter, as well as the number of portal tracts that can be visualized under microscope 113. In a consensus meeting of the American Association for the Study of Liver Diseases (AASLD) 112 on the endpoints and clinical trial design for NASH, a needle core biopsy with a 16 or lower gauge needle is recommended. A tissue core at least 2 cm long and containing at least 10 portal tracts is considered to be of good quality.

Several histological evaluation systems are available. The pathologists’ global assessment, which was original described and modified by Matteoni and colleagues, is widely accepted 53 112. It is well associated with disease severity and long term mortality 52-53. In this assessment, fatty liver is defined by the presence of > 5% steatosis under light microscopic examination of a hematoxylin and eosin stained liver section. A definite NASH is defined by the presence of 1. > 5% steatosis; 2. hepatocellular ballooning of any grade; and 3. lobular inflammation of any grade. A borderline NASH is defined as disease more than simple steatosis but do not meet the criteria for definite NASH.

A semiquantitative scoring system, NAFLD activity score (NAS) is recommended to quantify disease activity 112. NAS is the sum of the scoring of steatosis (0-3), lobular inflammation (0-3), and hepatocellular ballooning (0-2). The detailed scoring method is shown in Table 1.4. It is developed by Nonalcoholic Steatohepatitis Clinical Research Network (NASH CRN) in 2005 60, suggesting that NAS of ≥ 5 correlated with a diagnosis of NASH, and NAS of < 3 correlated with a diagnosis of simple steatosis. However, NASH CRN reevaluated this system in 2011 and found the diagnosis does not correlate well with the cutoffs of NAS 58. As a result, its role is only limited in clinical studies 114. The high sensitivity of NAS to disease change made it a reliable tool in repeated biopsy studies 112.

Fibrosis staging described by NASH CRN is widely accepted to assess fibrosis in NAFLD patients (Table 1.4) 60 112. Stage 3 (bridging fibrosis) and stage 4 (cirrhosis) fibrosis are generally considered as advanced fibrosis.

Liver biopsy is limited by its invasive nature. Although it is generally safe, major complications, such as significant bleeding, biliary peritonitis and pneumothorax, can still occur. There is an estimated morbidity of 0.06% - 0.35% and mortality of 0.01% - 0.1% 113. Some patients may refuse to have the procedure because of perceived pain and potential complications. Bleeding tendency and ascites, which are common situations in advanced liver disease, are relative contraindications for liver biopsy. It required in hospital treatment and observation. The cost is also high. In addition, it is also not very suitable for repeated evaluation due to increasing risk of complication.

Sampling bias is another limitation of liver biopsy. A biopsy sample considered adequate for histological assessment is only 1/50,000 to 1/65,000 of the whole liver mass 113. Merriman et al. reviewed 51 patients who received liver biopsy from both the right and left lobes of liver. Histological results of the two biopsy samples were compared 115. Agreement for steatosis was excellent, however it was only moderate for fibrosis and fair for lobular inflammation and hepatocyte ballooning. Intra-observer and inter-observer variability also existed.

1.2.2 Non-invasive assessment of NAFLD

Because of the limitations of liver biopsy, non-invasive assessment of NAFLD is urgently needed. An ideal non-invasive assessment should be accurate, easy to perform, reproducible and affordable. Its role, however, should not be confined to detecting steatosis. It is more important to distinguish NASH and fibrosis in NAFLD patients. For this purpose, several non-invasive tests are developed. Their advantages and limitations are discussed in this section.

1.2.2.1 Radiological imaging tests for hepatic steatosis

Ultrasonography (US) is the most common test for evaluating hepatic steatosis in daily clinical practice. It is generally available, easy to perform, radiation free with relatively low cost. For patients with hepatic fatty infiltration > 30% of the hepatocytes, US has over 90% sensitivity and specificity in detecting fatty liver 6. Typical ultrasonographic features of fatty liver include diffuse increase in fine echoes in liver parenchyma and impaired visualization of intrahepatic vessels and diaphragm. However, changes in US are not associated with the presence of lobular inflammation, hepatocyte ballooning or Mallory-Denk body, which are features of NASH. In other words, it cannot distinguish NASH patients from patients with simple steatosis 6. Besides, it is insensitive to fibrosis and early cirrhosis 6 116. Since the severity of disease is associated with necroinflammation and fibrosis rather than the degree of steatosis, the role of US in predicting clinical outcome of NAFLD is limited.

Computed tomography (CT) had similar accuracy in detecting hepatic steatosis with US 6 14 116. It cannot distinguish NASH either; fibrosis can only be detected in a cirrhotic stage 6. Furthermore, the radiation exposure to patients limited its wide use for the evaluation of NAFLD.

New modalities of magnetic resonance imaging, including dual gradient echo magnetic resonance imaging (DGE-MRI) and proton magnetic resonance spectroscopy (1H-MRS) largely improve the sensitivity in detecting hepatic steatosis 14. Both methods have around 80% sensitivity and specificity for the detection of hepatic steatosis as low as 5% 14. These tests have the ability to quantify hepatic steatosis with good accuracy and also are radiation free. Thus, they are preferred in recent population based studies for NAFLD screening 15-17. However, like US and CT, these tests cannot distinguish NASH and mild-moderate fibrosis 6. They are expensive, not widely available, which limited their utility in clinical practice.116

1.2.2.2 Physical measurements for NAFLD related fibrosis

Transient elastography (TE) by Fibroscan (Echosens, Paris, France) is a novel and rapid non-invasive measurement of liver stiffness 117-118. TE is equipped with a probe consisting of an ultrasonic transducer mounted on the axis of a vibrator. Once the probe is put at an intercostal space overlying the liver, it transmits a vibration of mild amplitude and low frequency to generate an elastic shear wave which propagates through liver parenchyma. In the meantime, the probe generates ultrasound wave to measure the velocity of the sheer wave. The denser the liver tissue, the faster the shear wave travels. Based on this principle, liver stiffness may be estimated 119. The measurement is quantitative and highly reproducible 120. Since liver stiffness significantly correlated with fibrosis stage, TE can be used as a non-invasive tool to assess fibrosis and cirrhosis in different liver diseases including NAFLD 63 120-123. In NAFLD, the Area under Receiver Operating Characteristic curve (AUROC) ranges from 0.78-0.99 for moderate (≥F2) fibrosis and 0.87-1.0 for advanced (F3-4) fibrosis 116.

The main challenge for TE in NAFLD patients is the high prevalence of obesity in this population. Take Hong Kong Chinese as an example, 86% of biopsy-proven NAFLD patients have central obesity 64. The transmission of shear wave and ultrasound into the liver parenchyma is affected by the thickness of subcutaneous and peri-hepatic fat. As a result, the success rate of liver stiffness measurement (LSM) decreases in obese subjects. The failure rate of TE is reported to be ranging from 3% to 16% in NAFLD patients 116. In the study by Wong et al., the successful rate was over 97% in patients with BMI lower than 30 kg/m2, and dropped dramatically to 75% in patients with BMI of 30 kg/m2 or higher 63. To solve this problem, the manufacturer of Fibroscan has recently developed an XL probe specifically for obese subjects. The XL probe uses lower frequency ultrasound and more sensitive ultrasonic transducer to assess deeper liver parenchyma 35-75 mm from the skin surface 124. It achieves higher success rate of measurement in obese subjects. Although the XL probe generates lower LSM compared with the traditional M probe 125-126, a study which included 193 biopsy-proven NAFLD patients revealed that LSM by XL probe was more likely to be performed successfully in NAFLD patients compared with M probe. AUROCs of XL probe for moderate (≥F2) fibrosis, advanced (≥F3) fibrosis and cirrhosis were 0.80, 0.85, and 0.91, respectively 126.

A real-time tissue elastography or shear wave elastography (SWE) is recently introduced to the non-invasive evaluation of fibrosis in NAFLD 127-128. This system displays the color-coded elastography image over the B-mode image of selected regions of interest in real time. This enables real-time SWE to obtain measurements based on both anatomical and tissue stiffness information. In a pilot study which enrolled 121 biopsy-proven NAFLD patients, AUROC of real-time SWE for moderate fibrosis was 0.92, which was significantly higher than TE (0.84, p = 0.002) 127. However, these results are still needed to be validated in larger populations; the cutoff values for different fibrosis stage also need to be confirmed.

Acoustic radiation force impulse imaging (ARFI, Virtual Touch Tissue Quantification, Siemens ACUSON S2000) is a different method, which explores the elastic properties of a region of interest. Liver tissue is mechanically excited using short-duration acoustic pulses. The displacement tissue generated produces a propagating shear wave whose velocity is calculated and which is proportional to tissue elasticity. Friedrich-Rust and colleagues compared the performance of ARFI with TE in NAFLD patients. No significant difference was found in diagnostic accuracies of TE and ARFI imaging for fibrosis staging 129. In another study, Yoneda et al. also reported similar results 130.

The advantage of ARFI is that it does not have the limitations in obese patients as TE. It is combined with regular ultrasound examination. Non-invasive assessment of fibrosis and surveillance of HCC can be done in the same time. The main limitations of AFRI are that the lack of standardized protocol for measurement; and the lack of precise definitions of ARFI failure 116.

Unlike TE and ARFI, magnetic resonance elastography (MRE) uses magnetic resonance imaging to evaluate the propagation of shear waves in the liver parenchyma. An active driver which is located outside the magnet room generates continuous low frequency vibrations, transmitted via a flexible tube to a transducer placed directly against the anterior right chest wall over the liver to generate shear waves. MRE is not affected by obesity. MRE has the appeal of becoming a one-stop service by combining it with regular magnetic resonance imaging (MRI) for structural examination and proton-magnetic resonance spectroscopy for hepatic steatosis assessment. Interestingly, a study showed that by applying a cut-off value of 2.74kPa, MRE could distinguish histological confirmed NASH form simple steatosis with a sensitivity of 94% and specificity of 73% 131. However, the long examination time, high cost and limited availability of facilities may be prohibitive. Besides, more experience is needed for using this technique in NAFLD patients.

1.2.2.3 Clinical tests and biomarkers

The progression from NAFLD and the development of fibrosis is sometimes accompanied by the changes in certain clinical tests such as elevated aminotransferase. However, the association is weak. In order to improve the accuracy for predicting NAFLD, NASH or fibrosis, several biomarkers and clinical prediction scores combining different clinical parameters have been developed. The utility of these clinical tests, biomarkers and their combines in non-invasive evaluation of NAFLD is discussed in this section.

Clinical tests

Aminotransferase, including alanine aminotransferase (ALT), aspartate aminotransferase (AST) and gamma-glutamyltransferase (GGT) are commonly measured in chronic liver disease patients. They also tend to be elevated in NAFLD patients compared with healthy subjects. However, the association is weak. Take ALT for example, it is commonly used to reflect hepatic inflammation and injury. In population studies, high ALT level is associated with long term increased risk of liver-related mortality and cardiovascular mortality 132-133. Elevated serum ALT sometimes serves as a diagnostic method in population-based studies to estimate NAFLD prevalence 2. However, serum ALT level can be normal in over half of NAFLD patients 134. Its association with histological findings is also poor 18. Even patients with persistently normal ALT may have lobular inflammation and hepatocyte ballooning. The association between ALT level and fibrosis is also poor. In patients with fibrosis progression towards cirrhosis, serum ALT may even decrease 135. However, although the correlation of ALT alone and severity of disease is weak, it could be further improved by be combined with other clinical parameters in different clinical prediction score.

Metabolic syndrome and its components are also associated with increased risk of NAFLD. In a large-scale community-based study, Wong showed that each component of the metabolic syndrome increased the risk of NAFLD in a dose-dependent manner 17. The prevalence of NAFLD in subjects without any component of metabolic syndrome was only 5%. It increased with the number of components and reached to 80% in subjects with all five components. Metabolic syndrome is also associated with the diagnosis of NASH and the presence of fibrosis in NAFLD patients 64. However, since simple steatosis patients may also have metabolic syndrome or its components, the accuracy of using metabolic syndrome to distinguish between patients with and without NASH or fibrosis is likely to be modest.

Other clinical tests, such as mean platelet volume 136, are also suggested as potential non-invasive tests for NAFLD. However, they still need to be validated.

Biomarkers

Several biomarkers have been introduced to overcome the limited accuracy in predicting NASH of clinical tests (Table 1.5). These biomarkers detect different features of NASH development, including cell apoptosis, cell necrosis, dysregulated adipokines, excessive oxidative stress and systemic inflammation.

Hepatocyte apoptosis is a prominent feature of NASH 105. Increase in hepatocyte cell apoptosis is typically present in both animal models of NASH and in human NASH patients. Effector caspases (mainly caspase-3) is activated in apoptotic process. Activated caspase-3 will cleave different intracellular substrate including cytokeratin 18 (CK-18), which is the major intracellular filament protein specific to hepatocytes. As a result, the amount of cleaved CK-18 fragment is increased both in the liver and in blood. The serum or plasma CK-18 level can be captured by specific antibody and measured using an immunoassay such as enzyme-linked immunosorbent assay (ELISA). The test is commercially available (M30-Apoptosense ELISA kit, PEVIVA, Bromma, Sweden), reliable with intra- and inter-test variable less than 10%. It is also rapid and easy-performing, which can be done in one working day. In a multicenter validation study which enrolled 44 simple steatosis, 26 borderline NASH and 69 NASH patients, Feldstein et al. showed that plasma CK-18 fragments M30 levels were significantly increased in patients with NASH 137. CK-18 M30 had high accuracy with an AUROC of 0.83 in diagnosing NASH. By applying different cutoff values from 216 to 287 (U/l), the sensitivity of CK-18 M30 ranges between 65% and 77%, and the specificity ranges between 65% and 92%. The results were validated in different ethnic populations 46 138-142. In a longitudinal paired liver biopsy study, the change of CK-18 M30 was also found to be correlated with disease progression 46. Patients with increased NAS 3 years after initial evaluation had greater increase of CK-18 M30 compared with those have static or decreased NAS.

The plasma level of another biomarker of apoptosis, soluble Fas, is also increased in NASH patients. In a study by Tamimi et al., the AUROC for the diagnosis of NASH was 0.86 for plasma soluble Fas in 95 biopsy-proven NAFLD patients 140. The AUROC can be further increased to 0.93 when combined with M30. In a validation cohort of 82 obese patients received bariatric surgeries, the AUROC for combined application of soluble Fas and M30 was 0.79.

Recent studies suggest that other cell death markers may also be useful in the prediction of NASH. In Addition to apoptosis, necrosis has also been proposed to be responsible for the disease progression in NAFLD patients 143. Unlike CK-18 M30 (M30), the CK-18 M65 (M65) and CK-18 M65ED (M65ED) ELISA (M65 ELISA kit and M65 EpiDeath ELISA kit, PEVIVA, Bromma, Sweden) measures soluble CK-18 released from dying cells and can be used to assess overall cell death due to apoptosis and necrosis 144. Both assays are based on two antibodies, M6 and M5, which are directed against two different epitopes of CK-18 and recognize total CK-18. The difference between M65 and M65ED assays is that M65 assay uses the M5 antibody for detection and M6 for capture; while M65ED assay uses these antibodies inversely. A study in USA suggested that the overall diagnostic accuracy of M65 for NASH was higher than that of M30 (AUROC: 0.81 for M65 vs. 0.71 for M30) 141; in another study in Turkey, the performance was similar (AUROC: 0.81 for M65 vs. 0.78 for M30) 142. Recently, Joka and colleagues suggested that M65/M65ED may have superior performance to M30 in detecting mild fibrosis and steatosis 139. However, that study was limited by the inclusion of different liver diseases and the small number of NAFLD patients 145. As such, the performance of M65 and M65ED in detecting NASH as compared to M30 is still unclear. The clinical significance and test performance of M65/M65ED warrant further validation.

Adipokines are cytokines secreted by adipocytes. They have important roles in regulating metabolism and insulin resistance, contributing to chronic inflammation associated with the metabolic syndrome 146. Most of the recognized adipokines are pro-inflammatory, while some of them are anti-inflammatory.

Adiponectin is one of the few adipokines that has anti-inflammatory effects. In a meta-analysis pooling 27 studies including totally 2243 subjects (698 control subjects and 1545 patients with NAFLD), blood adiponectin level was significantly lower in NASH patients compared with these non-NASH NAFLD patients or control subjects 147. The blood level of adiponectin in control subjects and non-NASH NAFLD patients had no significant difference. Its level is also decreased in obese patients with diabetes and metabolic syndrome 148. In a study including 80 NASH and 29 simple steatosis patients, Hui and colleagues showed adiponectin combined with HOMA-IR had an AUROC of 0.79 in differentiating NASH 149.

On the other hand, pro-inflammatory cytokines such as tumor necrosis factor-alpha and interleukin-6 are increased in NASH patients and yielding modest diagnostic accuracy 64. Adipocyte fatty acid binding protein (AFABP) is involved in the interaction between adipocytes and macrophages, which leads to inflammation and insulin resistance 150. In the study by Milner, serum AFABP was significantly higher in 69 NASH patients than in 31 simple steatosis patients and controls. It also correlated with ballooning, lobular inflammation and fibrosis stage 150. However, the authors did not perform standard c statistics to evaluate the diagnostic accuracy of AFABP.

NASH is characterized by heightened oxidative stress in the liver, which is the basis for the evaluation of anti-oxidant therapy such as vitamin E 151. Fibroblast growth factor 21 (FGF21) is a hormone which regulates lipid oxidation in the liver and stimulates glucose uptake in the adipose tissue 152. It is also termed as a "mitokine" due to its regulation by mitochondrial dysfunction and strong effect on increasing lipid oxidation and browning of white adipose tissue 153. Li et al. found that in human liver, FGF21 mRNA expression level increased with steatosis grade; its serum level was significantly higher in 159 Chinese NAFLD patients compared with 553 healthy controls 154-155. Similar results was reported by Yilmaz et al. 156. However, the correlation was not found in children 157. In a Japanese study of 57 patients, serum thioredoxin, a stress-inducible thiol-containing protein, was significantly increased in NASH patients 158. The test had an AUROC of 0.79 in diagnosing NASH. Other biomarkers of oxidative stress including copper-to-zinc superoxide dismutase, glutathione peroxidase and vitamin E level have also been evaluated with conflicting results.

Other biomarkers

C-reactive protein is a non-specific inflammatory marker that is increased in various conditions such as bacterial infection and coronary artery disease. While it (particularly high specificity C-reactive protein) has been shown to increase in NASH patients (several but not all studies), the diagnostic accuracy is modest and its use is limited by the non-specific nature.

1.2.2.4 Prediction scores

In order to improve the predictive value of single tests, efforts are made to combine these tests and generate an algorithm for a clinical prediction score. Most of these procedures are based on multivariable logistic regression. Independent risk factors for the end point of interest will be used to generate the algorithm following the formula which is estimated by regression model. The main clinical prediction scores are summarized in Table 1.6 159-171.

Among the prediction scores designed for NASH diagnosis, the NAFIC score has the highest overall accuracy 171. It was generated from 177 biopsy-proven NAFLD patients and was further validated in 442 biopsy-proven NAFLD patients in Japan. It is a weighted sum of serum ferritin, fasting insulin and type IV collagen 7S level by gender specific cutoffs. The AUROC for predicting NASH was 0.85 in the training cohort and 0.78 in the validation cohort. However, further validations in other independent cohorts are still needed before the NAFIC score, together with other scores designed for NASH prediction, can be used in clinical practice.

The Fatty Liver Index (FLI) is a simple score comprising BMI, waist circumference, triglycerides, GGT 161. It is designed for the prediction of fatty liver, with an AUROC of 0.84. FLI varies between 0 and 100. A FLI < 30 rules out fatty liver with a negative likelihood ratio of 0.2; while a FLI ≥60 predicts fatty liver with a positive likelihood ratio of 4.3. All parameters are routinely checked in clinical practice. Recently, the accuracy of FLI was validated in a large cohort with 2,652 subjects with an AUROC of 80.1 172. However, both study used ultrasonography as the reference standard. In another study which determined fatty liver by 1H-MRS in 220 diabetes patients, the AUROC of FLI was only 0.65 173. Therefore, its clinical usefulness is limited.

NashTest is a commercially available test comprising 13 parameters: age, gender, height, weight, and serum levels of triglycerides, cholesterol, alpha2-macroglobulin, apolipoprotein A1, haptoglobin, GGT, ALT, AST and total bilirubin. In a validation study led by the manufacturer, NashTest had an AUROC of 0.79 in diagnosing NASH 168.

The NAFLD fibrosis score was developed and validated in 733 NAFLD patients from USA, Europe and Australia, which is the largest cohort for fibrosis prediction so far. It is comprised of 6 simple clinical parameters: age, hyperglycemia, BMI, platelet, albumin, and AST/ALT ratio 160. By applying different cutoffs (low: -1.455; high: 0.676), NAFLD fibrosis score has a negative predictive value (NPV) of 88% or a positive predictive value (PPV) of 82% in excluding or diagnosing advanced fibrosis. The negative predictive value of the low cutoff remains high at 91% when it is applied to the Chinese population; however, few Chinese patients have scores above the high cutoff value 174. This may be because Asians are generally less obese than Caucasians. Further validation in different ethnic groups using ethnic-specific definitions for anthropometry is still needed.

Enhanced Liver Fibrosis panel (ELF) was validated in 196 European NAFLD patients for fibrosis predicting 164. The ELF panel had AUROC of 0.90 for predicting severe (≥F3) fibrosis, 0.82 for moderate (≥F2) fibrosis and 0.76 for any fibrosis. By adding simple clinical parameters, including age, BMI, diabetes or impaired fasting glucose, AST/ALT ratio, platelets, and albumin to the panel, the AUROC could be further improved to 0.98, 0.93, and 0.84 for distinguishing severe fibrosis, moderate fibrosis, and any fibrosis, respectively. Authors suggested at least 82% of liver biopsies for evaluating fibrosis in NAFLD patients could be spared by applying ELF.

FibroTest is a commercially available test panel comprising 5 parameters: α2-macroglobulin, apolipoprotein A1 (ApoA1), haptoglobin, total bilirubin and GGT. It was originally designed to estimate liver fibrosis in patients with chronic hepatitis C. When FibroTest was applied to 170 NAFLD patients, the AUROC for detecting F2-4 disease and F3-4 disease were 0.75 and 0.81, respectively 169. Although these are biomarkers of liver fibrosis, its performance may be affected by other parameters. For example, intravascular hemolysis results in low haptoglobin level and raised bilirubin. GGT is also sensitive to recent alcohol consumption.

In three separate validation studies performing head-to-head comparison in NAFLD subjects from the United States (n=541), United Kingdom (n=145) and France/Hong Kong (n=246), FIB-4 index appeared to have the highest AUROC among all tested clinical prediction scores including AST/ALT ratio, APRI, BARD, FIB-4 and NAFLD fibrosis scores 63 166 175. FIB-4 index is a formula comprised of 4 clinical parameters including age, platelet, AST and ALT. In all studies, the AUROCs of FIB-4 index for F3-4 fibrosis were all ≥0.80.

Among all these prediction scores, AST/ALT ratio is the simplest one first designed to predict fibrosis and cirrhosis in patients with chronic hepatitis C 176. However, the performance of AST/ALT score differed significantly in different validation studies in NAFLD patients. The accuracy in detecting advanced fibrosis was 83%, 74% and 66% in the head-to-head comparison studies mentioned above 63 166 175.

In summary, non-invasive imaging methods are mainly used to determine the grade of steatosis; accuracy in evaluating inflammation and fibrosis is poor. Physical measurements correlated well with fibrosis; however, they are limited by its availability. Clinical tests and biomarkers, together with those prediction scores, provided potential options in distinguishing NASH and fibrosis, however, most of them still need to be validated in large independent cohort. Furthermore, the effort in searching non-invasive tests is merely searching for surrogates for liver biopsy; however, the most important issue is the ability to predict whether a patient will develop hepatic complications in the future. As such, further studies are required to evaluate the performance of these non-invasive tests in predicting clinical events.

Table 1.4. NAFLD activity score and fibrosis staging system by Nonalcoholic Steatohepatitis Clinical Research Network

Item

Definition

Score

Steatosis grade

<5%

0

5%-33%

1

>33%-66%

2

>66%

3

Lobular inflammation

No foci

0

<2 foci per 200X field

1

2-4 foci per 201X field

2

>4 foci per 202X field

3

Ballooning

None

0

Few balloon cells

1

Many cells/prominent ballooning

2

Fibrosis

None

0

Perisinusoidal or periportal

1

Mild, zone 3, perisinusoidal

1A

Moderate, zone 3, perisinusoidal

1B

Portal/periportal

1C

Perisinusoidal and portal/periportal

2

Bridging fibrosis

3

　

Cirrhosis

4

Table 1.5. Biomarkers of NASH

Mechanism

Biomarkers

Apoptosis

Cytokeratin-18 fragments

Soluble Fas

Necrosis

Intact cytokeratin-18

Adipokines

Adiponectin

Tumor necrosis factor-alpha

Interleukin-6

Adipocyte fatty acid-binding protein

Oxidative stress

Fibroblast growth factor 21

Thioredoxin

Copper-to-zinc superoxide dismutase

Glutathione peroxidase

Vitamin E

Systemic inflammation

C-reactive protein

Table 1.6. Clinical prediction scores for NAFLD, NASH diagnosis and fibrosis staging

Purpose

Marker

Parameters

Endpoint

AUROC

NAFLD

Fatty liver index (FLI)

BMI, waist circumference, triglycerides, GGT

Fatty liver

0.84

NASH

Nash Tes

Age, gender, BMI, triglycerides, cholesterol, α-2-macroglobulin, GGT, AST, ALT, haptoglobin, apolipoprotein A1, total bilirubin

NAS ≥5

0.79

Palekar index

Age ≥50 yrs, female gender, AST ≥45 U/L, AST/ALT ratio ≥0.8, BMI ≥30 Kg/m2 , hyaluronate ≥55 ug/l

NASH

0.76

Shimada index

Serum adiponectin, HOMA-IR, serum type IV collagen 7S level

NAS ≥5

N.A.

NAFIC score

Ferritin ≥200 ng/ml (female) or ≥300 ng/ml (male), fasting insulin ≥10.0 l U/ml, and type IV collagen 7S ≥5.0 ng/ml

NAS ≥5

0.85

NASH Clinical score for morbid obesity

Hypertension, diabetes, AST ≥27 IU/L, ALT ≥27 IU/L, obstructive sleep apnea and nonblack race

NAS ≥5

0.8

Fibrosis

FibroTest

α2-macroglobulin, GGT, apolipoprotein A1, haptoglobin, total bilirubin, age, gender

F ≥2

0.81

F ≥3

0.88

NAFLD fibrosis score

Age, fasting glucose , BMI, platelet count, albumin, AST/ALT ratio

F ≥3

0.82

ELF

Age, hyaluronate, MMP-3, TIMP-1

F ≥1

0.76

F ≥2

0.82

F ≥3

0.9

FibroMeter for NAFLD

Age, weight, platelet count, ferritin, glucose, AST, ALT

F ≥2

0.94

BARD score

BMI ≥28, AST/ALT ≥0.82, diabetes

F ≥2

0.81

FIB-4

Age, AST, ALT, platelet count

F ≥2

0.74

F ≥3

0.86

F4

0.86

Hepascore

Age, sex, total bilirubin, GGT, α2-macroglobulin, hyaluronic acid

F ≥2

0.73

F ≥3

0.81

F4

0.91

APRI

AST, platelet count

F ≥2

0.73

F ≥3

0.79

F4

0.75

　

AST/ALT ratio

AST, ALT

F ≥3

0.83

1.3 Genetic determinants of NAFLD

Although obesity and metabolic disorder are the major risk factor for NAFLD, it is not the only mechanism. As mentioned before, ethnicity also affects the prevalence of NAFLD 7-8 10-11 15-17. The high prevalence in Hispanics and Indian Asians of NAFLD cannot be fully explained by ethnic differences in BMI or insulin resistance. Furthermore, NAFLD tends to cluster in families. Struben et al reported 8 index NASH or cryptogenic cirrhosis patients and 10 relatives with similar diseases in 8 kindreds 177. Patterns of afflicted patients included mother-daughter, sister-sister, sister-brother, father-daughter, and male-female cousins, and these disorders were not associated with obesity or diabetes. Moreover, fatty liver was found to be more common in siblings and parents of children with NAFLD compared with overweight children without NAFLD 178. The heritability of NAFLD was estimated to be 39% after adjusting for age, sex, race and BMI. All these data suggest that genetic determinants may play an important role in NAFLD development.

With the advances in genome analysis, genetic determinants of NAFLD are widely studied in the recent years. Genome-wide association study (GWAS) is a high-throughput genotyping technology, which scan genome-wide known single-nucleotide polymorphism (SNP) genetic markers in a case-control setting 179. By comparing the different allele frequency between NAFLD patients and healthy controls, a number of potential genetic determinants have been introduced based on GWAS (Table 1.7) 77 97 180-183.

1.3.1 Patatin-like phospholipase domain containing 3 (PNPLA3) rs738409

The nonsynonymous rs738409 I148M (C/G) variant located in human patatin-like phospholipase domain containing 3 gene (PNPLA3) is the first identified genetic variant associated with higher prevalence of NAFLD in over 2000 participants in the Dallas Heart Study 97. NAFLD was diagnosed by 1H-MRS. PNPLA3 rs738409 was significantly associated with hepatic fat content after adjustment for BMI, diabetes status, alcohol use and ethnicity. Hepatic fat content was more than twofold higher in GG homozygotes than in CC homozygotes. Furthermore, Hispanics had larger proportion of risk allele carriers (49%) than African Americans (17%) and European Americans (23%), which could partly explain the higher risk of NAFLD in Hispanics 15.

The findings were subsequently confirmed by other GWAS 183-184. Speliotes et al. reported PNPLA3 rs738409 was associated with hepatic steatosis evaluated by CT in 7,176 subjects from United States183. Meanwhile, Kawaguchi and colleagues showed similar results in 529 histologically diagnosed NAFLD patients and 932 population controls from Japan 184. It is further validated in many independent cohort studies from different countries and ethnicities 185-194

As the association between PNPLA3 rs738409 and hepatic steatosis is robustly validated and widely accepted, whether it is associated with the histological severity seems to be a little controversial. The Japanese GWAS study which rediscover PNPLA3 rs738409 had a subgroup of histological confirmed NAFLD patients, which allowed us to explore the association of PNPLA3 rs738409 and NASH diagnosis as well as fibrosis 184. In 529 histologically diagnosed NAFLD patients, PNPLA3 rs738409 exhibited the strong association with the histological classifications proposed by Matteoni 53. The distributions of PNPLA3 rs738409 genotype was significantly different between patients with NASH and other NAFLD patients. Moreover, PNPLA3 rs738409 also showed strong association with hyaluronic acid, HbA1c and iron deposition in the liver in 3 independent clinical trials, which may have prognostic effect on NAFLD patients.

However, in another GWAS conducted in 236 biopsy-proven NAFLD patients, PNPLA3 rs738409 did not exhibit any association with histological severity 182. The authors strictly selected non-Hispanic white female NAFLD patients as study subjects. Farnesyl diphosphate farnesyl transferase 1(FDFT1) rs2645424 was found to be associated with NAS, which indicted overall disease severity; 4 other SNPs were associated with lobular inflammation or fibrosis. PNPLA3 rs738409 was not associated with any of them. Authors attributed it to their relative small sample size and highly selected study subjects.

Other than GWAS, more cross-sectional studies which focused on single SNP of PNPLA3 rs738409 have been reported. A number of them included histological confirmed NAFLD patients. Sookoian et al. first demonstrated the significant association of PNPLA3 rs738409 with NASH diagnosis. 12 in 40 (30%) simple steatosis patients were GG homozygotes, while 33 of 63 (52%) NASH patients were GG homozygotes 191. The association of PNPLA3 rs738409 with fibrosis was not reported. In another larger study including 574 NAFLD patients received liver biopsy in Italy and United Kingdom 193, PNPLA3 rs738409 was strongly associated with NASH (Odds ratio [OR]: 1.5, 95% confidence interval [CI]: 1.12-2.04) and moderate fibrosis (OR: 1.5, 95% CI: 1.09-2.12). The GG homozygotes also had significantly higher blood ALT, low-density lipoprotein (LDL), fasting insulin level, HOMA-IR and lower blood high-density lipoprotein (HDL) level compared with CC homozygotes. The study with the largest validation cohort was reported by Rotman et al., who evaluated 894 adult patients from NASH Clinical Research Network (NASH CRN) 190. PNPLA3 rs738409 was found to be strongly associated with steatosis, portal inflammation, lobular inflammation, Mallory-Denk bodies, NAS and fibrosis. GG homozygotes had a mean NAS of 4.5, which was significantly higher than CC homozygotes (4.1). GG homozygotes had a mean fibrosis stage of 4.5, which was also significantly higher than CC homozygotes (4.0). Each G allele had an adjusted OR of 1.5 for predicting advanced fibrosis, translating to 2.3 times