Racial differences in prevalence and severity of non-alcoholic fatty liver disease

Maurizio Bonacini, Farah Kassamali, Swathi Kari, Nieves Lopez Barrera, Mohamed Kohla

Maurizio Bonacini, Department of Gastroenterology and Hepatology, California Pacific Medical Center, California Pacific Medical Center, San Francisco, CA 94115, United States

Farah Kassamali, Department of Gastroenterology and Hepatology, California Pacific Medical Center, San Francisco, CA 94115, United States

Swathi Kari, Department of Internal Medicine, St.Mary's medical Center, San Francisco, CA 94117, United States

Nieves Lopez Barrera, Quest Clinical Research, San Francisco, CA 94115, United States

Mohamed Kohla, Department of Hepatology, National Liver Institute, Menoufiya University,Shibin Al Kom 32511, Menoufiya, Egypt

Abstract The aim of this review is to assess the evidence regarding racial differences in the prevalence and severity of nonalcoholic fatty liver disease (NAFLD).We reviewed the published literature that reported prevalence, severity, and genetic associations of NAFLD in different ethnic groups.The metabolic syndrome (MetS) has been associated with NAFLD, but each component of the MetS is present in various races in different percentages and their effect on NAFLD appears to be dissimilar.An elevated triglyceride (TG) level seems to have the strongest association with NAFLD.The latter is more prevalent in Hispanic patients; Blacks have lower TG levels and a lower NAFLD prevalence, compared to Caucasians or Hispanics.The severity of liver fibrosis is lower in some, but not all biopsy-based studies of Black patients.No study has evaluated the severity of liver disease controlling for the individual components of MetS, especially TG.Important racial differences in the prevalence of selected genetic polymorphisms, particularly PNPLA-3 and MBOAT7 have been documented, together with their effects on the prevalence of liver steatosis and fibrosis.Data on overall and liver mortality have found no significant differences according to race/ethnicity, with the possible exception of one paper reporting lower cirrhosis mortality in Black patients.We conclude that NAFLD is more prevalent in Hispanics and less in Blacks.This is supported by differences in key genetic polymorphisms associated with hepatic fat storage.However, there is presently insufficient evidence to firmly conclude that race, per se, plays a role in the development of liver fibrosis and its complications.Further studies, appropriately controlled for diet, exercise, and individual MetS parameters are needed.

Key Words:Race; Ethnicity; Nonalcoholic fatty liver disease; Fatty liver disease;Metabolic syndrome

INTRODUCTION

Nonalcoholic fatty liver disease (NAFLD) is one of the most common diagnoses made in a gastroenterology practice.Several articles suggested differences in the prevalence and severity of liver disease according to patient race/ethnicity.If such differences were proven, this would have an important impact on resource allocation to decrease health disparities.Thus, it is imperative that the available literature be critically reviewed and existing knowledge gaps, if any, identified.

RACE AND NAFLD

Definitions

NAFLD is a condition marked by excess fat storage accounting for > 5% of the liver’s volume in the absence of known alcohol abuse[1].The latter is usually defined as the use of > 20 g alcohol/day for women and > 30 g/d for men[2], although lower limits have been used[3].No study addressing race differences has verified absence of alcohol by testing hair for alcohol or using blood phosphatidylethanol levels[4,5].The diagnosis is usually inferred by imaging studies, typically an ultrasound showing increased hepatic echogenicity[6,7].Elevated alanine aminotransferase (ALT) in the absence of known competing causes has also been accepted as “suspected NAFLD”[7].It is also crucial to differentiate primaryvssecondary causes (medications, genetic or nutritional disorders); however only approximately 12% of studies excluded the latter[8].

We accepted the authors’ race classification, which was typically based upon selfreporting.We recognize that race and ethnicity are “constructs that have no clearcut definition”[9].It is important to keep in mind that Hispanics and Asians include significantly heterogeneous sub-populations[3,7,9].

Since Asians are underrepresented in most United States studies, this review will focus on Blacks (or African-Americans), Hispanics (or Latinos) and Whites (or Caucasians).

For the purpose of this paper, we will accept that the alcohol history is accurate, that a compatible ultrasound and/or elevated transaminases in the appropriate clinical setting are reasonable diagnostic tools, and that all reported cases are primary NAFLD.

Specific aim

To assess the strength of evidence suggesting that race-ethnicity in adults is associated with not only prevalence, but also with severity and prognosis of NAFLD.

Methods

We queried the PubMed English language database using the following keywords in the title or abstract:“fatty liver”, “nonalcoholic fatty liver disease”, “NAFLD”, “l(fā)iver or hepatic steatosis”, “steatohepatitis” AND “race” or “ethnicity”.We eliminated articles including alcoholic liver disease or HIV infected patients.We restricted this narrative review to adult populations.

Prevalence of NAFLD by race/ethnicity

The prevalence of NAFLD is reported to be highest in the Middle East (32%) and South America (31%), followed by Asia (27%), and Europe (23%)[10,11].In Africa and India, the prevalence of NAFLD is approximately 9% of the population[12,13].

The most recent estimate places the United States prevalence of NAFLD at about 32%[14].The United States is unique due to its mix of various races and ethnicities,while maintaining relative homogeneity in terms of geography and alimentary patterns.Therefore, it seems like an optimal population to study to uncover potential racial differences in disease.

A recent meta-analysis[8] shows that in population-based cohorts (i.e., not high-risk patient groups such as diabetics) 23% of Hispanics have NAFLD,vs14% of Caucasians, and 13% of African Americans.These percentages translate into a higher relative risk (RR) for Hispanics being diagnosed with NAFLD (RR = 1.5), and lower for African/Americans (RR = 0.7) compared to Whites[8].If one focuses on patient subgroups that are at high risk for NAFLD, these differences become smaller(Hispanics RR = 1.2 and African-American RR = 0.8) but remain significant[8].Interestingly, a NHANES based study[6], not included in the above meta-analysis, also found that Hispanics have a RR for NAFLD of 1.7 and African-American a RR of 0.8 compared to Whites:however when restricted to ‘never drinkers’, those differences are no longer significant, implying that small amounts of alcohol may have different effects on different races[6].Thus, despite higher rates of HTN and insulin resistance,African-Americans have a lower prevalence of NAFLD[1,6,15-18].

There is relatively little written about Asian patients other that the prevalence may be about 25% in Asia[19], but may be lower in US-residing Asians, where NAFLD is noted in 20%[14].A summary of the estimated prevalence of NAFLD in the United States is shown in Table 1.

Table 1 Estimated prevalence of nonalcoholic fatty liver disease in the general United States population (three main Race-ethnicities)

Prevalence of the metabolic syndrome by race/ethnicity

Contributors to the rising worldwide prevalence of NAFLD include non-modifiable factors like genetics, but also modifiable variables such as diet and lifestyle choices[7,21,22].Identifying, quantifying and controlling for these factors will be useful to establish whether some groups may be at higher risk, and therefore help allocate resources to mitigate those differences[23].

Diet and exercise has been found to be different in different ethnic groups.Asians have better diets (measured with an adapted healthy eating index) than Caucasians who in turn have better diet scores than Latinos and Blacks[21].In Hawaii, however,intake of fruits and vegetables was lowest in Japanese-Americans compared to Filipinos or Native Hawaiians[22].Yet, it is not clear whether a better diet score necessarily translates into a lower NAFLD risk[21,22]; and if so, by how much.

Similarly, exercise habits appear to be different, highest in Caucasians and lowest in Asians[9,22].This is important because exercise decreases intrahepatic fat by MRI,even in the absence of weight loss[24].Unfortunately, in articles focusing on NAFLD,these potentially important variables have not been adjusted for.

Metabolic syndrome (MetS) is accepted as the major association with NAFLD.MetS is defined by the presence of 3 or more out of 5 criteria:Increased fasting glucose,central obesity/waist circumference, low high-density lipoprotein (HDL), elevated triglycerides (TG), and elevated arterial pressure.Meeting this definition is associated with future development of diabetes type 2 (DM) and cardiovascular disease (CVD)[23].There are differences in the prevalence of MetS according to race ethnicity, in non-institutionalized adult individuals living in the United States.A recent assessment shows that the prevalence of MetS was 35% in Whites, 30% in Blacks, followed by Hispanics (termed “Mexican Americans”) (29%)[15].No increased prevalence was noted in the Latino population surveyed[15].A United States military study looked at the incidence of MetS (by ICD-10 codes), and found the highest was in Pacific-Islanders, and the lowest in White personnel[25].

However, there are 3 important problems with MetS as a dichotomous variable.First, individual components of the MetS have a different distribution among races,elevated TG being more common in Latinos and White males and abnormal waist circumference in Blacks and White females[23].In fact, the low TG levels in Blacks have been called “the TG paradox”[26].Thus, African American patients have a higher body mass index (BMI) and similar prevalence of DM, yet they display a better lipid profile and therefore are less likely to have MetS compared to Hispanics (Table 2)[17].The prevalence of DM is lowest in Whites (12%) and similar in Asians (19%), Blacks(20%) and Hispanics (22%)[18].The latter group showed major heterogeneity, South American patients having less DM (12%) compared to other Latino groups[18].

Table 2 Prevalence of metabolic syndrome and its components in African Americans vs Hispanics[17]

Second, a diagnosis of MetS predicts the development of DM or CV disease differently in different races.For example, in patients with MetS, rates of incident DM are highest in Black males and females (17%) and lowest in white women (8%);whereas the rate of development of CVD is highest in White men (25%) and lowest in Black women (6%)[23].Third, the association between individual MetS variables and NAFLD is not the same.In a recent study from China (Asian patients), NAFLD patients had higher levels of each of the 5 MetS parametersvscontrols.However,when a multivariable analysis was run, adjusted for age and sex, the strongest association was with an elevated TG; the prevalence of NAFLD in the highest and lowest TG quartile was 50%vs5 %[27].Therefore az-score, where the MetS is measured on a continuous scale (from -1 to +4) has been developed and shown to predict the development of diabetes and CVD better than the binary MetS[23].When controlled for the z-score, Black individuals have double the rate of DM and higher rates of hypertensionvswhites[16,23].There are no data assessing the prevalence and severity of NAFLD, in patients matched by thez-score.

Fat distribution/obesity

Lean NAFLD (i.e., with normal BMI) is found in as many as 5% of those with NAFLD in the United States[14] and this subgroup has a 65% chance of being metabolically abnormal,i.e., fulfilling criteria for MetS[28].On the other hand, overweight and obese NAFLD patients have a correspondingly higher chance of having MetS, 92% and 95%,respectively.Lean NAFLD seems more common in Asiansvsother ethnic groups[14,20].Elevated TG appears to be the commonality in patients with NAFLD, independent from BMI[17,27,28].

Patterns of visceral and liver fat depositions show ethnic differences and may contribute to the prevalence and severity of NAFLD.Total adiposity, measured by DEXA and MRI to account for visceral, liver and truncal fat was found to be highest in Japanese Americans and lowest in African Americans[17].Interestingly, women had lower visceral fat area than men, except in the Japanese American group[20].African-American adolescents have less visceral fat than either Hispanics or Whites[29].

A study using transient elastography and controlled attenuation parameter estimated hepatic steatosis and fibrosis in 2000 Korean patients.Obese (i.e., BMI 25 or greater) but metabolically healthy (no MetS) individuals had greater liver steatosis and fibrosis than non-obese patients[30].However, in the non-obese group, those with MetS, had higher steatosis estimates but similar fibrosis to those without MetS.BMI rather than MetS was the variable independently associated (P< 0.001) with both steatosis and fibrosis[30].The Dallas heart study quantified visceral fat percentage by MRI in the general population:unfortunately, 3% to 8% of the individuals reported alcohol intake levels exceeding those used to define NAFLD[1].The findings were thatmale Hispanic and White individuals had similar risk (42% to 45%) of having hepatic steatosis greater than 5.5 g TG per 100 g of liver tissue, much higher compared to Black males (23%).Women, both White and Black, had lower rates of abnormal hepatic steatosis (24%) compared to Latinas (45%).The fact that Blacks had higher HTN and Insulin resistance rates, but lower circulating TG levels, suggests racial and genetic differences in intrahepatic TG storage[1,16,20,31].

Genetics

Pathways of lipolysis or lipogenesis (MBOAT7, PNPLA3, TM6SF2,) are some of the genetic polymorphisms that have been linked to NAFLD prevalence and its severity[16,32-34].

In individuals of European descent, a T mutation in the MBOAT7 gene (rs641738)has been associated with severity of NAFLD in those with TT homozygosity[34].Even the presence of one T polymorphism was associated with a small [odds ratio (OR) =1.3] but significant risk of biopsy-proven F2, F3 or F4 fibrosis[34].However, the association between the PNLPA3 G allele and F2-F4 was stronger (OR = 1.6)[34].

The PNPLA3 gene controls hepatic VLDL excretion, likely leading to hepatic TG accumulation; it may also sensitize the liver to environmental stressors, thus contributing to elevated transaminase levels in the presence of obesity[2].The G allele mutation (rs738409), termed I148M (vsCC wild type) is a single-nucleotide polymorphism (SNP), which increases the risk of fat accumulation in the liver and thus NAFLD four-fold[17,32,33].The G allele was found to be more frequent in Hispanics (40%) compared to Africans and Europeans (both 15%).In those with GG alleles, the risk of having NAFLD was similar in Asians and Caucasians (3-fold) and Hispanics (4-fold) but was much higher in Black patients (9-fold) compared to those with wild type genotype[35].

Within the United States population, the PNPLA3-G allele had a significant association with a non-invasive estimate of liver fibrosis, the FIB-4 score[7], but in one study this association was not clear (Table 3)[3].The GG homozygosity has also been associated with a 5-fold increase in HCC risk[33].A recent study from Sicily confirmed that the G allele (either heterozygous or homozygous) was associated with more advanced liver fibrosis[36].In patients with stage 3 and 4 fibrosis, the G allele was associated with more liver decompensation, HCC and liver related death, despite a relatively small total number of patients followed (n= 471)[36].Interestingly, 2/3 patients had the G allele and almost a quarter was homozygous GG[36].

Table 3 Percentage of patients with the PNPLA3 G allele polymorphism and FIB-4 > 2.67[3]

In Hispanics with American ancestry (Mexican-, Central-, and South American), the frequency of PNPLA3-G is higher than in those of European or Afro-Caribbean background[3].A small study in Hmong patients suggests that some Asian subpopulations have high rates of the G SNP and thus may have increased risk for NAFLD[37].These findings underscore the existence of distinct and potentially relevant subpopulations within a traditional race/ethnicity group.

A minor allele (rs58542926) in transmembrane 6 superfamily member 2 (TM6SF2)was associated with hepatic TG content measured by magnetic resonance spectroscopy, in the Dallas Heart Study[1].The C to T polymorphism decreases VLDL excretion, thus increasing TG concentration in the liver[33].In addition, this TM6SF2 polymorphism was noted to increase the risk for hepatic fibrosis independent of age,obesity, diabetes, and PNPLA3 genotype[38].On the other hand, the TM6SF2-T allele mutation E167K had similar low frequencies between Hispanics[3] and those from European ancestry and had a strong association with ALT levels[15].

NAFLD, liver fibrosis and liver complications

Several studies assessed metabolic factors associated with varying histopathological severity of NAFLD.There is agreement that the degree of steatosis is proportional to the number of elements of the MetS[7,16,20,32,39,40].Additionally, one study showed that the MetS was associated with significantly greater risk of liver fibrosis stage 3 or 4(33%vs15% in those without MetS) and necroinflammation (61%vs44%)[39].The same study showed that in patients with NASH, 88% had MetS compared to 67% of those with simple fatty liver[39].

NAFLD is more prevalent in Hispanics[6,15,18,31], but the significance of this finding is debatable, as fibrosis is the only histological variable consistently associated with liver mortality[41].While mortality in NAFLD patients is chiefly associated with cardiovascular events[42,43], it would be useful to tease out whether race independently affects the development of cirrhosis, and therefore liver mortality.

The fact that there is a relationship between elements of MetS and liver steatosis,inflammation and fibrosis[23,39] means that studies comparing liver disease severity between races must be controlled for the 5 MetS variables, keeping in mind that each may be more predictive in specific races.

The multi-ethnic cohort[44] looked at a United States population enriched with Asian minorities.The results showed that NAFLD was the most common cause of chronic liver disease in Japanese Americans (64% of those with liver disease) followed by Hawaiians (58%), Latinos (46%), Whites (41%) and Blacks (39%).When looking at the percentage of patients who had NAFLD-related cirrhosis (by ICD-9 codes) by race,the percentages were 4% (Japanese), 3.1% (Latinos), 1.7% (Whites) and 1.5% (Blacks)[44].

Dulaiet al[42] reviewed 5 studies that assessed baseline liver fibrosis (mostly by biopsy) in patients with NAFLD or NASH.At baseline these 5 studies showed that most (67%) of patients had stage 0/1 fibrosis; 14% had F2; 12% F3 and 7% cirrhosis.Mortality was mainly cardiovascular related (about 40%) followed by cancer (20%)and liver disease (10%)[43].There were no details comparing races within each study.In fact, one study had only Asians[45] and another 88% Whites[34].A Canadian study did not mention race or ethnicity[46].While baseline advanced fibrosis stage (F3/4)varied from 27% in Asians[45] to 12% in Whites[43], the percentages of MetS was also different (63%vs33% respectively).

Within NAFLD, however, NASH on liver biopsy is less common in African-Americans (57%), but not significantly,vsCaucasians (73%)[47].

A recent meta-analysis[8] noted that 11 studies assessed stage of fibrosis (mostly by biopsy) in NAFLD and had data on race.The pooled proportion of patients with NAFLD and significant fibrosis (stages 3 and 4) was 19.5% [95% confidence interval(CI):18.1-20.9].The percentages were numerically highest in Whites (22.3%) and Hispanics (19.6%) and lowest among Blacks (13.1%).However, differences were not statistically significant for WhitesvsHispanics (RR 1.02, 95%CI:0.94-1.11), and borderline significant for WhitesvsBlacks (RR = 1.10, 95%CI:1.00-1.22)[8].A later paper showed that morbidly obese Black patients (mean BMI > 45) had lower % of NASH (4%) and lower % of fibrosis stages 3 and 4 (1.4%)vsWhites (17% and 9%respectively).The 2 groups had similar percentages of DM and hypertension[48].A retrospective but well detailed study based on liver biopsy found advanced fibrosis(F3/F4) in 16% Caucasiansvs2.6% Blacks, despite the fact that the latter had greater BMI and higher DM rates.However, their lipid profile was healthier than Caucasians[49].

The most recent NHANES (1999-2016) evaluation[14] used the US Fatty Liver Index to define NAFLD and two noninvasive marker (FIB-4 and NAFLD Fibrosis Score) to assess advanced liver fibrosis (i.e., stages 3 and 4).The results show that Latinos and Whites had higher likelihood of NAFLD (43% and 33%, respectively),vsAsians (20%)and African Americans (19%).Overall, mortality was associated with DM2 and FIB-4 but not race, and was higher in lean or overweight patientsvsobese[14].

Interestingly, a work by Lomonacoet al[50] found that, when metabolic factors are controlled for, hepatic steatosis, inflammation and fibrosis scores (all by histology)were similar between Caucasians and Hispanics.A study assessing biopsy-confirmed NASH and comparing Latinos and Whites reported that the former were younger, had increased carbohydrate intake, and had a lower prevalence of hypertension[31].However, while there were numerical different rates of F3/F4 (Whites and Blacks 30%,Asians 28% and Latinos 23%) these were not significant.Multivariable analysis identified only age, female gender, HTN and abnormal HOMA-IR as significantly associated with advanced fibrosis, but not race[31].

The preponderance of evidence shows that while Latinos have more NAFLD, they don’t have significantly higher rates of advanced fibrosis.Studies based on liver biopsy, except one[31] have shown Black patients to have less fibrosis[8,49].However,adequate controlling for the variables of the MetS has not been done.

NAFLD is associated with the development of hepatocellular carcinoma (HCC).One report demonstrated that patients with NAFLD have a 10-fold higher chance of developing HCC compared to controls[51].The overall risk of HCC in NAFLD was low (estimated 0.02/100 patient-years), and it was higher in older (> 65 years)Hispanics and lower in Blacks:these subgroups were not matched by MetS risk[51].

Finding racial differences in mortality (especially liver mortality) in patients with fatty liver requires evaluation of a very large database.A NHANES analysis (1988-1994), looked at (mostly) NAFLD patients and found a correlation between high estimated liver fibrosis (by non invasive tests such as FIB-4) with mortality (both all cause and liver-related) up to 2006[52].Unfortunately, liver mortality represented only 3% of the total mortality, so there were too few endpoints to make inferences about racial associations[52].A review of total United States mortality captured in the latest National Vital Statistics (NVSS) database, showed that Hispanics with a diagnosis of NAFLD have lower mortality than Caucasians, although in both groups the trend is towards increased mortality the past 10 years:there was no attempt to adjust the data for underlying metabolic disease[53].In 2016, the NIAAA issued a report on liver cirrhosis mortality.The age-adjusted mortality rates for cirrhosis “without mention of alcohol” were 50% lower in Blacksvswhites, but NAFLD codes were not specifically reported[54].However, a paper looking at hospital charges, length of stay and mortality in non-Federal Community hospitals across the United States, showed that mortality was not statistically different across races in patients admitted with a NAFLD diagnosis[55].

More data on race-specific cirrhosis, HCC and mortality rate in patients with NAFLD are needed.

Response to therapy

There is considerably less data on racial responses to therapy for NAFLD.To date, this includes mainly weight loss strategies, including bariatric surgery.

Vilar-Gomezet al[56] published a small but well-designed study enrolling Cuban patients.They histologically documented decreased liver fibrosis (45% of patients) and resolution of NASH (90% of patients) when a 10% or greater weight loss was achieved[56].The latter endpoint was noted in 10% of patients, all of them Cuban.However,diet and exercise may be beneficial to decrease liver steatosis in the absence of weight loss[24].

Behavioral therapy resulted in a maximum weight loss of 5 kg in Black patients,significantly less than 13 kg in Whites[57].Metformin for one year significantly increased HDL-cholesterol (by 1-2 mg/dL) in White and Black patients:In Hispanics the HDL declined by approximately 1 mg/dL[58].Lorcaserin lead to a placeboadjusted weight loss of 3.2 kg, 2.7 kg and 1.4 kg in Whites, Blacks and Hispanics respectively[59].Semaglutide as an injection for DM control showed minor changes in weight in different races[60].

A study in 3268 patients (1561 Hispanics, 660 Blacks, and 1047 Whites) examined the percentage of excess weight loss (EWL) after Roux-en-Y gastric bypass or adjustable gastric band placement[61].EWL differed by ethnicity (-53% in Hispanics, -50% in Whites and -43% in Blacks), at 6 months post-operatively.These differences persisted at 1 and 2 years after surgery (-69%, -69% and -58%, respectively)[61].A prior meta-analysis, looking at the percentage of EWL (between 12 and 24 mo postoperatively) confirmed an average of 8% lower weight loss in Blacks compared to Whites[62].

In the future, large phase 3 studies using new NASH medications may uncover possible racial differences in baseline histology, and clinical liver outcomes.Those studies will have prospectively collected metabolic data, permitting investigators to assess risk by race, controlled for variables of the MetS[23].

CONCLUSION

In conclusion, there is convincing evidence that the prevalence of NAFLD depends on genetics and the prevalence of the MetS.Its individual components impact fatty liver differently in different populations.Socio-economic, dietary and lifestyle differences may also explain reported racial differences but have not been thoroughly studied in the NAFLD arena.In the United States, NAFLD and NASH seem more prevalent in Hispanics, however most studies have not been controlled for the individual variables of MetS, and this may have overestimated racial differences.African Americans have a lower prevalence of hypertriglyceridemia and this contributes to their lower prevalence of NAFLD despite higher rates of hypertension and DM.Fibrosis scores seem similar in Whites and Latinos:In most biopsy studies, Blacks have shown lower hepatic inflammation and fibrosis levels.There is no evidence that NAFLD mortality is higher in Latinos, and it may be lower in Blacks.We believe that there is presently insufficient evidence to confidently conclude that race, per se, plays a role in the development of the complications of NAFLD.Further studies, appropriately controlled for diet, exercise, and MetS parameters are needed.