Aliment Pharmacol Ther 2011; 34: 274–285
Background Non-alcoholic fatty liver disease (NAFLD) is a common cause of chronic liver disease, and its worldwide prevalence continues to increase with the growing obesity epidemic. This study assesses the epidemiology of NAFLD in adults based on clinical literature published over the past 30 years.
Aim To review epidemiology and natural history of non-alcoholic fatty liver disease and non-alcoholic steatohepatitis in adults based on clinical literature published over the past 30 years.
Methods An in-depth search of PubMed (1980–2010) was based on five search terms: ‘non-alcoholic fatty liver disease’ OR ‘non-alcoholic steatohepatitis’ OR ‘fatty liver’ OR ‘steatosis’ AND ‘incidence’ [MeSH Terms] OR ‘prevalence’ [MeSH Terms] OR ‘natural history’. Studies of paediatric cohorts were excluded. Articles were categorised by topic and summarised, noting generalisations concerning their content.
Results Four study categories included NAFLD incidence, prevalence, risk factors and natural history. Studies related to NAFLD prevalence and incidence indicate that the diagnosis is heterogeneous and relies on a variety of assessment tools, including liver biopsy, radiological tests such as ultrasonography, and blood testing such as liver enzymes. The prevalence of NAFLD is highest in populations with pre-existing metabolic conditions such as obesity and type II diabetes. Many studies investigating the natural history of NAFLD verify the progression from NASH to advanced fibrosis and hepatocellular carcinoma.
Conclusions Non-alcoholic fatty liver disease is the most common cause of elevated liver enzymes. Within the NAFLD spectrum, only NASH progresses to cirrhosis and hepatocellular carcinoma. With the growing epidemic of obesity, the prevalence and impact of NAFLD continues to increase, making NASH potentially the most common cause of advanced liver disease in coming decades.
Non-Alcoholic Steatohepatitis and NAFLD are common causes of chronic liver disease world wide, and hepatic steatosis is the most common pathological feature.1 Understanding the epidemiology of NAFLD is essential for developing effective treatment and prevention strategies.
This study is based on a PubMed search (1980- August 2010) for ‘non-alcoholic fatty liver disease’ OR ‘non-alcoholic steatohepatitis’ OR ‘fatty liver’ OR ‘steatosis’ AND ‘incidence’ [MeSH Terms] OR ‘prevalence’ [MeSH Terms] OR ‘natural history’. Studies of paediatric cohorts were excluded. Case studies, review articles and studies of paediatric populations were excluded. The remaining articles were categorised by topic and summarised, noting generalisations concerning their content.
Incidence of NAFLD and NASH
The incidence of NAFLD is underreported and varies wildly (Figure 1). A study based on routine health care of Japanese government employees revealed an overall incidence of non-alcoholic hypertransaminasemia of 31 cases per 1000 person-years.2 Another Japanese study reported follow-up data on 3147 individuals without NAFLD at baseline; of these, 308 (10%) developed new cases of NAFLD over 414 days.3 A recent study of an out-patient hepatology clinic in England reported an incidence rate of 29 cases per 100,000 person-years.4 The clear discrepancy among these rates suggests that accurate global incidence rates for NAFLD require further study.
Prevalence of NAFLD and NASH
The prevalence of NAFLD in the general population has been assessed with a variety of diagnostic tools (Figure 2). Liver biopsy, the current gold standard for NASH diagnosis and staging, is invasive and cannot be used in population-based studies. However, liver biopsy findings have been reported for liver transplant donors who are considered to be in good health. A Korean study in which liver biopsies were performed on 589 consecutive potential liver transplant donors reported NAFLD prevalence of 51%.5 In the United States, liver biopsies performed on potential liver donors revealed that 20% of donors were ineligible for organ donation based on the degree of the steatosis (>30%).6
Studies reporting the results of an autopsy series also show mixed results regarding the prevalence of NAFLD. Autopsies of 423 aircrew confirmed 66 cases of fatty liver (16%), 11 (3%) of which were compatible with alcohol abuse.7 Another study from India reported findings on 1230 adult autopsies, 195 (16%) of which showed fatty liver.8 Only 5% of these patients were obese, and high levels of alcohol consumption were associated with fatty liver.8 Therefore, a large proportion of these patients may not have fulfilled the criteria for NAFLD. In an autopsy series of lean individuals from Canada, the prevalence of steatohepatitis and fibrosis was 3% and 7%, respectively.9 Finally, a study from Greece revealed evidence of steatosis in 31% and NASH in 40% of autopsied cases of ischaemic heart disease or traffic accident death after exclusion of hepatitis B seropositivity or known liver disease.10 Importantly, a later study, reporting the prevalence of steatohepatitis in this Greek population, showed that steatohepatitis is much higher compared with other population studies, which the authors attribute to decreased physical activity and alterations in dietary habits in the Greek population over recent decades.10
In addition to liver biopsy-proven fatty liver, several non-invasive diagnostic methods for NAFLD and NASH have been introduced recently. Nevertheless, these methods are somewhat less definitive than histology-based evaluations of biopsies and autopsies. On the other hand, some important conclusions can be drawn from non-invasive population-wide samples studied non-invasively.
Non-invasive radiological modalities used to assess the prevalence of fatty liver include magnetic resonance imaging (MRI) and ultrasonography. In a recent multicentre population study, patients were randomly selected from 25 primary healthcare centres in Spain. After the exclusion of patients with known liver disease and high alcohol intake, the prevalence of NAFLD on ultrasound was 33% in men and 20% in women.11 The DIONYSOS nutrition and liver study based on ultrasonography showed that the prevalence of NAFLD in Italian subjects with and without suspected liver disease was 25% and 20%, respectively.12 In addition, over 10 to 12 years, the prevalence of NAFLD identified with ultrasonography in a cohort of 35 519 Japanese individuals increased from 13% to 30%.13 In India, the use of ultrasound for determining the prevalence of NAFLD estimated a prevalence rate of 17%.14 Another study performed in Brooke Army Medical Center revealed an even higher prevalence of ultrasonographic NAFLD, at 46%, which was confirmed by biopsy in 12% of the total cohort, or 30% of ultrasound positive patients.15 Although less commonly applicable to large scale screening studies of NAFLD than ultrasonography, MRI has been used to determine the prevalence of NAFLD. In the US, a widely cited MRI study from Dallas County, Texas identified NAFLD in 31% of a multi-ethnic, population-based sample.16
Other studies have used elevations in the liver enzymes, alanine aminotransferase (ALT) and aspartate aminotransferase (AST), as non-invasive indicators of NAFLD. Data from the US National Health and Nutrition Examination Survey (1988–1994) reported elevated ALT levels in 3% of the population.17 The 1999–2002 National Health and Nutrition Examination Survey (NHANES) reported elevated ALT (7%), AST (4%), or either ALT or AST (8%) of the participants.18 These liver enzyme elevations are about twice as prevalent as those reported for the NHANES 1988–1994 study,17 although the older study used a different assay. A much higher rate is reported in a retrospective electronic medical record review of patients seen at The Johns Hopkins Hospital Executive Health Program. This study recorded AST levels in all 2294 subjects and ALT levels in 1309 of them.19 Elevated liver enzymes were observed in 14% of non-alcoholic subjects who had at least one test, and 21% of non-alcoholic subjects who had both tests. In addition to these rates, the prevalence of elevated ALT levels observed in other studies ranged from 8% to 9%.20 When considering the results of aminotransferase -based studies as a surrogate for the population prevalence of NAFLD, it is important to bear in mind that the upper normal limit (ULN) of serum alanine- aminotransferase (ALT) normal range has been recently challenged.21, 22 The issue of normal range for aminotransferases and their relationship with gender is described in more detail in following sections. Nevertheless, it is important to emphasise that although elevated ALT is generally associated with histological NASH, a number of patients with normal ALT levels may also have NAFLD and even advanced fibrosis. Therefore, ALT activity alone cannot be used to rule out significant liver disease in patients suspected of having NAFLD, especially those with type II diabetes or hepatomegaly.23
Another interesting approach to population-wide profiling of the prevalence of NAFLD is based on combinations of easy to measure clinical variables. Examples of the NAFLD profiles of this kind include Fatty Liver Index (FLI), [BMI, waist circumference, triglycerides and gamma-glutamyl-transferase (GGT)],24 and Lipid Accumulation Product (LAP) (waist circumference and fasting triglycerides).25 These indices are applicable in community healthcare settings, and could contribute to studies of NAFLD prevalence, but as yet have been applied only to preselected populations.
In summary, most of the US studies report a 10–35% prevalence rate of NAFLD; however, these rates vary with the study population and the modality used to establish the diagnosis (Figures 2 and 3). Because approximately one-third of the US population is considered obese, the prevalence of NAFLD in US population is likely to be about 30% (Figures 2 and 3). On the other hand, liver biopsy is required to make a definitive diagnosis of NASH, and estimates from biopsy series indicate that the prevalence of NASH in the US is between 3% and 5%. It is important to emphasise that NAFLD is not unique to the US. Reports from the rest of the world suggest that the prevalence rate ranges from 6% to 35%, with a median of 20%14, 26–39 (Figure 3). Most studies agree that many features of metabolic syndrome are associated with NAFLD,13, 19–34 with the exception of the studies conducted in Asian countries, which often report NAFLD in non-obese individuals.14, 26–39
Risk factors for NAFLD and NASH: predisposing demographical and clinical factors
The prevalence of NAFLD and NAFLD-related fibrosis increases with age.14, 28, 35, 40–54 Frith and colleagues studied a population of 351 patients with biopsy-proven NAFLD who were divided into older (≥60), middle-aged (≥50 to <60) and younger (<50) groups;55 they showed an association between the prevalence of NAFLD and fibrosis with age. It is important to note that older patients had significantly more NAFLD risk factors, such as hypertension, obesity, diabetes and hyperlipidaemia. Another study assessed the prevalence of NAFLD in a cohort of octogenarians admitted to the rehabilitation departments of a geriatric hospital. This study reported a prevalence rate of 46%, which is greater than that of the general population.54 On the other hand, this study of an aged cohort showed no association between NAFLD and metabolic syndrome, cardiovascular disease, or cirrhosis.54 The authors suggest that NAFLD pathogenesis might be altered in the elderly population.
In addition to the association between age and the prevalence of NAFLD, older patients with NAFLD have a higher likelihood of disease progression or mortality.41–44 Older age also increases the risk of developing related problems such as severe hepatic fibrosis, hepatocellular carcinoma and type II diabetes mellitus.45, 46, 56 In fact, a study of potential living liver donors showed that older age is an independent risk factor for significant hepatic steatosis.5 Cryptogenic cirrhosis, which is now considered ‘burned out NASH’, is more common in older diabetic patients with current or past obesity.51 In a previously mentioned study by Frith et al., older patients also had significantly higher grades of hepatic fibrosis, and cirrhotic patients were significantly older than noncirrhotic patients.55 On the other hand, younger age was associated with higher ALT activity,18, 52 as well as with hepatic steatosis in individuals with unexplained liver enzyme elevations.47 Contrary to these studies, Hui and colleagues note no significant differences in age between individuals with progressive NAFLD and those who did not progress.53
In summary, the relationship between age, NAFLD, and fibrosis remains unsettled. It is of note that that the association between age and the high prevalence of NAFLD, as well as the higher stage of fibrosis and cirrhosis in NAFLD, may be related to the duration of disease rather than to age itself.
Non-alcoholic fatty liver disease was initially thought to be more common in women, but this notion lacks empirical support.37–39, 57, 58 One study of 26 527 Asian subjects receiving medical health checkups showed that the prevalence of NAFLD was 31% in men and 16% in women.59 The clinicopathological profiles of patients from India showed that the majority of NAFLD patients were men.60 Male gender is also associated with elevated aminotransferase levels, the presence of histological NASH, hepatic fibrosis and overall mortality in patients with NAFLD.41, 61, 62 Only a few studies suggest that female gender is associated with NAFLD and fibrosis,42, 63 and one study reports that female gender is an independent risk factor for NASH in patients with metabolic syndrome.60
Given these discrepancies, it is possible that NAFLD behaves differently in men and women. Some experts suggest that the current upper limit of normal ALT activity in women (ULN ≤ 30 U/L) should be reduced to ULN ≤ 19 U/L.21, 22 A recent study by Kunde and colleagues compared the old and the new thresholds for aminotransferases.64 According to the old threshold, the prevalence of elevated ALT was 28% in women with class II/III obesity; this rate increased to 63% with the new threshold. Nevertheless, the prevalence of NAFLD and NASH in patients with normal ALT was similar, regardless of the threshold. Similar reports assessing different thresholds of aminotransferases for male individuals are lacking. We believe that studies using large population based-cohorts are needed to resolve this issue.
Race and ethnicity
Several studies have shown that the prevalence of NAFLD differs with race and ethnicity. Hispanics have the highest prevalence of NAFLD, hepatic steatosis, and elevated aminotransferase levels, followed by non-Hispanic Whites, with the lowest rate reported in African Americans.16, 61, 65 Finally, the prevalence of NAFLD in American Indian and Alaska-Native populations ranges from 0.4% to 2%.66, 67 In the absence of liver biopsies, the true prevalence of NAFLD is probably underestimated.
The explanation for ethnic differences in the prevalence of NAFLD is not clear. Wagenknecht and colleagues report that some correlates of NAFLD, such as the extent of visceral adiposity, are similar across ethnic groups.65 On the other hand, other correlates (e.g., age, TG and PAI-1) are only associated with NAFLD in Hispanics, and serum adiponectin level is only associated with NAFLD in African Americans.65 NAFLD may be affected by important, yet unidentified, genetic or environmental factors.
In the era of genomics medicine, it is important to mention recent evidence connecting genotyping data to the epidemiology of NAFLD. The observed familial clustering of NAFLD68, 69 suggests that it may be influenced by genetic variants. Studies of large family based cohorts indicate that the heritability of hepatic steatosis is approximately 0.27.70 The most important genetic contributor to NAFLD is the I148M allele of the PNPLA3 gene, which encodes adiponutrin.71, 72 In homozygous carriers of this allele, hepatic fat content is more than twofold higher than in noncarriers.71 Notably, this allele is most prevalent in Hispanics (0.49), the group most susceptible to NAFLD, whereas lower frequencies are observed in Caucasians (0.23) and African Americans (0.17). In nonsteatotic subjects, the same allele also predisposes an average increase of 28% in ALT levels.73 In addition, NASH is more frequently observed in I148M homozygotes (OR 3.488).73
Another variant of the same gene, PNPLA3-S453I, that is common in African Americans (0.104) but rare in European Americans (0.003) and Hispanics (0.008), is associated with significantly lower liver fat content. A regression analysis indicates that these two sequence variations account for 72% of the observed ethnic differences in hepatic fat content in the study cohort.71 In addition, resequencing of the PNPLA3 gene in individuals with abnormally high liver fat content revealed three cases of null mutations, which is consistent with the idea that a loss-of-function of PNPLA3 causes hepatic steatosis.71
A recently completed genome-wide association (GWA) study of CT-detected hepatic steatosis in 7176 individuals, and its replication in 592 subjects with biopsy-proven NAFLD, confirms the importance of PNPLA3 variants in the degree of fat accumulation in the liver.70 In addition, genetic variants in NCAN, GCKR and LYPLAL1 are significant NAFLD contributors.70GCKR has also been independently identified as an NAFLD associated gene in Chinese subjects.74 Several other genetic variants have been identified, although with less convincing evidence.
Non-alcoholic fatty liver disease is more prevalent in cohorts of patients with pre-existing metabolic conditions than the general population. Specifically, type II diabetes mellitus and NAFLD have a particularly close relationship. A study of patients with type II diabetes mellitus reported a 69% prevalence of ultrasonographic NAFLD.75 NAFLD was associated with obesity, hypertriglyceridaemia and high-normal ALT levels. However, no relationship was evident between diabetic degenerative complications or glycaemic control and the presence of ultrasonographic NAFLD.75
In addition to a high prevalence of NAFLD, liver disease may be more progressive in patients with type II diabetes. In fact, a recent study showed that 127 of 204 diabetic patients displayed fatty liver on ultrasound. Of those who consented to a liver biopsy and had fatty liver on ultrasound, 87% were diagnosed with biopsy-proven NAFLD.76 This study showed that the prevalence of NASH increases in parallel with components of the metabolic syndrome. Nevertheless, it is important to note that NASH and advanced fibrosis are often observed in diabetic patients without symptoms, signs, or liver enzyme abnormalities.
Recently, polycystic ovarian syndrome (PCOS) was proposed as the ovarian manifestation of metabolic syndrome.77 One study found that 55% (48/88) of PCOS women had both hepatic steatosis and high HOMA-IR scores.78 Another study found that 41% (17/41) of women with PCOS had concomitant NAFLD as diagnosed by hepatic steatosis and abnormal ALT levels, whereas the incidence of NAFLD in the weight and age matched non-PCOS control group was only 19% (P < 0.05).79 Some studies have also shown that for obese PCOS patients the rates of co-diagnoses with NAFLD or NASH are even higher.80 It seems that PCOS patients are at increased risk for developing NAFLD and NASH.80
As previously noted, obesity is associated with NAFLD; the prevalence of simple steatosis in obese individuals ranges from 30% to 37%,81, 82 and NAFLD ranges from 57% of overweight individuals attending out-patient clinics to 98% of nondiabetic obese patients.63, 83–85 It is noteworthy that obesity is additive to other factors predisposing liver steatosis. For example, in heavy drinkers, obesity increases the risk for steatosis by twofold.86
Given the obesity epidemic, bariatric surgery is among the most rapidly growing surgeries in the US Over the past decade, intra-operative liver biopsies have become routine during bariatric surgery, and several resulting studies report a high prevalence of NAFLD and NASH in severely and morbidly obese individuals.56, 63, 81–85, 87–90 For example, according to Boza et al.,81 the prevalence of NAFLD and cirrhosis in a cohort of obese patients undergoing gastric bypass was 63% and 2%, respectively. However, Ong and colleagues showed that over 95% of bariatric surgery patients had fatty liver, 20–30% had NASH, and 10% had advanced fibrosis.63 The median prevalence of NASH in the obese population is 33%, ranging from 10% to 56%.81–83, 85–90 The prevalence of fibrosis in obese patients depends on the type of fibrosis; 67% of gastric bypass surgery patients had portal fibrosis, but only 4% of bariatric surgery patients had perisinusoidal fibrosis.81, 85, 88, 89
Chronic infections and other conditions associated with fatty liver
Although not specifically considered a part of NAFLD, fatty liver can be seen with a number of infections. In fact, hepatitis C virus (HCV) appears to exacerbate metabolic syndrome by eliciting increased insulin resistance (IR).91 Steatosis is present in almost 50% of patients infected by HCV, and occurs especially often in chronic hepatitis associated with HCV genotype 3.92 In addition to patients with HCV, a relatively high prevalence of NAFLD has been reported in HIV-positive patients, especially those treated with Highly Active AntiRetroviral Therapy (HAART), or HIV patients with lipodystrophy.
In addition to chronic infection with HCV and HIV, patients with primary aldosteronism and myotonic dystrophy also can have high prevalence of fatty liver.78–80, 93–96 It is important to emphasise that within these populations, many patients are at risk for metabolic conditions (insulin resistance, obesity, increased BMI, elevated triglycerides and lowered HDL levels), which predispose them to fatty liver.78–80, 93–96
Studies from tertiary care centres, community-based and population-based databases suggest that liver disease is the third leading cause of death among people with NAFLD.41–54 In a community-based report of NAFLD patients from Olmsted County, Minnesota, mortality was 13% after a mean follow-up of 7.6 years.43 A NHANES-based study of 12 822 individuals showed that 817 had NAFLD. Over a median follow-up of 8.7 years, 80 with NAFLD and 1453 without liver disease died.41 Increased mortality was associated with impaired-fasting glucose, cirrhosis, non-Hispanic white race, lower educational level, lower income and risk factors associated with NAFLD.41
In addition to studies assessing long-term outcomes of the entire NAFLD cohort, a few reports dissected NAFLD into its histological subtypes. In terms of potential for progression, NAFLD patients fall into two broad categories: NASH and non-NASH. The non-NASH subtype of NAFLD includes all patients with simple steatosis as well as patients with steatosis and nonspecific changes.97 Although NASH can follow a potentially progressive course for liver disease, the non-NASH subtype does not progress or progresses very slowly. Hepatic injury induced by NASH is similar to that caused by alcohol-induced liver disease; however, NASH seems to progress more slowly and is histologically less severe than steatohepatitis caused by alcohol.1, 97
An assessment of the long-term outcomes of a cohort with simple steatosis included 40 patients, with or without risk factors for progression to NASH.97 After a median of 11 years, 12 patients had abnormal liver tests, although none progressed to NASH or cirrhosis. Of the 14 deceased patients, none died of a liver-related cause. This study suggests that simple steatosis does not follow a progressive course. However, studies of the follow-up biopsies in patients with a histological diagnosis of simple steatosis show that that isolated steatosis may have a very slowly progressive course.98 In one study, after 5 years of follow-up, patients with steatosis developed both inflammation and ballooning, and mild fibrotic changes.98 Wong et al. followed-up on a larger NAFLD cohort of 52 patients that were reevaluated 3 years after initial diagnosis.99 Among 13 patients with simple steatosis at baseline, two (15%) had a normal liver at month 36, three (23%) continued to have simple steatosis, five (39%) developed borderline NASH, and three (23%) developed NASH.99 The discrepancy between clinical and histological data suggests that the very slow and rare progression of liver disease in patients with simple steatosis may lead to their demise from cardiovascular causes rather than advanced liver disease.
In contrast, it is worth noting that modern types of the bariatric intervention (gastric bypass) may lead to the resolution of liver steatosis. In one study of patients biopsied at the time of bariatric surgery and at follow-up, hepatic infiltration with fat was reduced in 65 out of 91 patients, whereas increases in the steatotic score were observed in only three patients.100 Another study of 90 biopsied bariatric patients showed that 16 patients (18%) had the same degree of steatosis, the steatosis pattern improved in 25 (28%), and 49 (54%) had normal hepatic tissue by the second biopsy.101 These observations are in striking contrast with the results of the intestinal bypass procedures formerly performed to correct morbid obesity, which often resulted in worsening steatosis and fibrosis, and occasionally hepatic failure.
Repeat biopsies reveal important information about NAFLD progression. As noted previously, Wong et al. examined 22 patients with borderline NASH at baseline, showing that four (18%) had simple steatosis and 13 (59%) had borderline NASH at month 36, whereas five (23%) developed NASH.99 Among 17 patients with NASH at baseline, 10 (59%) continued to have NASH and six (35%) had borderline NASH at month 36. Only 1 (6%) patient regressed to simple steatosis. In another study, 32% of 22 patients with NASH showed progressive liver fibrosis over a median follow-up of 4.3 years.102 A study based on paired liver biopsies on 22 patients with NASH over a mean of 5.7 years showed that 32% of these patients had histological disease progression, either in the form of increased inflammation or fibrosis score.103 One- third of these patients had rapid progression to advanced fibrosis, and the elevated serum AST was correlated with histological progression. Another repeat biopsy study showed that 44 out of 83 patients with NASH (53%) had fibrosis progression by at least one stage after a median time of 6.4 years.60 A study of Chinese patients with NAFLD reported progressive disease with a worsening fibrosis score in 9 out of 17 (53%) cases followed for a median of 6.1 years.53
Interestingly, some serial liver biopsy studies showed that liver fibrosis in NAFLD might improve. For example, in a cohort of 39 Japanese subjects with NAFLD, liver fibrosis improved in 12 patients (31%), progressed in 11 patients (28%), and remained unchanged in 16 patients (41%) after a median follow-up of 2.4 years.104 In the study by Wong et al., 14 (27%) NAFLD patients had fibrosis progression, 25 (48%) had static disease and 13 (25%) had fibrosis regression.99 Bariatric surgery may also clear perisinusoidal fibrosis at least in some morbidly obese patients.100 The regression of fibrosis is most commonly reported in restrictive bariatric surgeries (Laproscopic Gastric Banding) but progression of fibrosis is reported in some patients undergoing malabsorbtive surgeries (Bilio-pancreatic diversion). Future large scale studies are needed to assess the impact of bariatric surgery on the natural history of NASH.
Beyond population-based studies and those reporting sequential liver biopsies, a study involving tertiary care centres by Mateoni and colleagues assessed liver-related mortality in 132 biopsy-proven NAFLD patients.105 After 8 years of follow-up, 11% of NASH patients developed liver-related mortality, compared to only 2% of the non-NASH cohort. Rafiq and colleagues updated the mortality data for this cohort, reporting that after 18.5 years of follow-up in the same NAFLD cohort, liver-related mortality had risen to 18% for NASH and 3% for non-NASH patients.106 Ekstedt and colleagues reported increased mortality in a NASH cohort after 13.7 years of follow-up but not in patients with simple steatosis.107 Finally, Soderberg and coworkers assessed liver biopsy findings and mortality in 256 patients with elevated liver enzymes who had liver biopsies between 1980 and 1984; 46% of these patients had NAFLD. These investigators showed that mortality was increased in the NASH group but not in those with simple steatosis.108
These studies suggest that a subgroup of NAFLD patients progress to more serious liver disease and have higher liver-related mortality. Studies based on histological data suggest that only patients with NASH show the progressive form of disease.In addition, insulin resistance, diabetes mellitus and/or other components of metabolic syndrome are recognised as risk factors for developing advanced liver disease.
In addition to direct evidence, indirect evidence suggests the progressive nature of disease in some patients with NAFLD. Cryptogenic cirrhosis (CC) is now considered burned out NASH.109, 110 Type II diabetes mellitus and obesity are more common in patients with cryptogenic cirrhosis, and are observed in over 70% of patients with CC.In addition, 74% of liver biopsies performed on a cohort of patients with CC revealed one or more features of NASH.51, 110–112 It is noteworthy that the prevalence of diabetes and obesity in patients with CC is similar to that of patients with NASH. On the other hand, patients with CC are, on average, 10 years older than patients with NASH.51 Both lines of evidence suggest that patients with CC may begin with NASH and over the next decade, some progress to cirrhosis.
Other features related to metabolic syndrome, such as dyslipidemia, insulin resistance, higher BMI and lower high-density lipoprotein, are more prevalent in CC patients than in control patients with hepatitis B- and C-related cirrhosis, and with primary biliary cirrhosis (PBC).111, 113 Additionally, the rate of metabolic conditions in patients with CC is significantly higher than the controls. Specifically, a study of Mexican Mestizo patients with CC shows that the prevalence of the metabolic syndrome was 500% higher than in patients without CC.112
In addition to the indirect link between CC and NASH, the association between NASH and hepatocellular carcinoma (HCC) provides another type of indirect evidence supporting the progressive nature of NASH.114, 115 Advanced fibrosis is an important risk factor for HCC.46, 114 Hashimoto et al.46 found that after a median follow-up of 40.3 months, the 5-year cumulative incidence of HCC was 8% among 118 patients with advanced fibrosis and no HCC at the baseline.
Additional evidence suggests that patients with CC are also at increased risk for HCC. Bugianesi and colleagues identified 46 patients with CC and HCC as well as age- and sex-matched controls with HCC and other types of cirrhosis.111 Their results show a 7% overall prevalence of HCC in patients with CC, which was less than the prevalence of HCC in patients with cirrhosis related to alcohol or viral hepatitis, but greater than that for HCC in patients with cirrhosis related to hemochromatosis or primary biliary cirrhosis. A multivariate analysis shows that hypertriglyceridemia, type 2 diabetes, and normal ALT are independently associated with CC and HCC. This study emphasises that patients with cirrhosis as well as patients with CC are at risk for HCC.111
In addition to CC, patients with NASH-related cirrhosis are also shown to be at risk of developing HCC.42, 46 A study of 247 Japanese patients with biopsy-proven NAFLD reported 46 patients with bridging fibrosis (F3), and 43 patients with cirrhosis (F4). During a median follow-up of 44 months in NAFLD patients with advanced fibrosis (F3 and F4), 10 developed liver-related morbidity and five developed HCC. The 5-year cumulative incidence of HCC was 20% with eight patients dying: two from liver failure, four from HCC, and two from other carcinomas.42
Although risk factors for NASH, particularly obesity and diabetes, may contribute to the development of HCC, the exact pathogenesis of NASH-related HCC remains unclear. In a case-series of 11 patients with NASH-associated HCC, 91% were obese or had type II diabetes, hypertension, or dyslipidemia, suggesting that obesity and type II diabetes may be involved in the pathogenesis of NASH-related HCC.116 Surprisingly, seven of these patients did not have cirrhosis. Although the provocative conclusion of this study suggests that noncirrhotic patients with NASH may also be at risk and should be screened for HCC, these findings are too preliminary to recommend general HCC screening for noncirrhotic patients with NASH.
The spectrum of NAFLD and its subtype of NASH seem to be a final common pathway, in histological terms, of many causes of hepatic metabolic stress.
Given the strong association of NAFLD with metabolic syndrome as well as the worldwide epidemic of obesity, the prevalence of NAFLD and NASH are increasing. In the US alone, an estimated one-third of the population has NAFLD and approximately 2–5% have NASH. Within the NAFLD spectrum, only patients with histologically proven NASH develop progressive liver disease. Progression seems more likely in the setting of diabetes, insulin resistance and other pre-existing conditions.
The majority of the epidemiological studies summarised in this manuscript are based on a relatively small number of patients assessed with different pathological protocols, radiological modalities, or serum biomarkers and are followed for relatively short periods. This circumstance presents several important challenges for assessing the epidemiology of NAFLD. We lack well designed prospective studies profiling large, preferably ethnic-specific cohorts of patients with well characterised NAFLD who are followed for long periods. It is also important to develop consensus on the histological definition of NASH and other subtypes of NAFLD, because they are important for determining study inclusion at baseline. The cost and invasiveness of liver biopsy make it a less attractive modality for assessing long-term outcomes. Although many non-invasive biomarkers for NAFLD assessment have been developed, only some provide strong evidence of external validation.117 Finally, the most important outcomes of mortality, morbidity and resource utilisation are studied only in limited cohorts of NAFLD patients. Therefore, because we expect the future prevalence and impact of NASH to increase with the obesity epidemic, we need to perform larger, longitudinal studies that assess the long-term natural history of NAFLD with validated, non-invasive biomarkers and by integrating morbidity and mortality data. Studies of this kind will provide the type of evidence needed to encourage clinicians, researchers and health policy experts to focus on NAFLD as one of the most common chronic human diseases worldwide.
Declaration of personal and funding interests: None.