NAFLD and HIV: Do Sex, Race, and Ethnicity Explain HIV-Related Risk?

Abstract

Purpose

Here, we review the epidemiology, diagnosis, and management of NAFLD in the general population, discuss HIV-specific differences in NAFLD pathogenesis, and summarize what is known regarding differences in NAFLD by race/ethnicity and sex.

Recent findings

The reported prevalence of NAFLD among people living with HIV varies by age, body mass index, comorbidity, and method of NAFLD diagnosis, but is generally thought to be greater among HIV-infected compared to -uninfected populations. Minorities and women tend to experience poorer HIV treatment outcomes (1–5), and are at the greatest risk for significant weight gain with HIV treatment (6). Thus, woman and minorities living with HIV may be at a higher risk of developing NAFLD and progressive liver disease.

Summary

Disparities in the diagnosis, progression, and prognosis of NAFLD and HIV-associated NAFLD may be, in part, explained by genetic and sex differences however data is limited.

Epidemiology of NAFLD

Non-alcoholic fatty liver disease (NAFLD) is characterized by steatosis (>5% of hepatocytes with lipid droplets), in the absence of other secondary causes of steatosis (7) including alcohol use (8). Worldwide, the prevalence of NAFLD has risen since first described in 1980, paralleling the increase in obesity, diabetes, and the metabolic syndrome (9). Currently, NAFLD is the most common etiology of chronic liver disease, affecting approximately 25% of the population worldwide (10), and between 17–51% of the US population (7, 11–13). A subset of individuals with NAFLD develop non-alcoholic steatohepatitis (NASH), the progressive form of NAFLD. NASH is characterized by the presence of both steatosis and cellular injury, with the latter manifested as hepatocyte ballooning and lobular inflammation. Hepatic fibrosis can also develop with NASH, facilitating progression to cirrhosis, decompensated liver disease and hepatocellular carcinoma. In the United States, NASH is the second leading indication for liver transplantation (14), and the most common indication for those under age 50(15).

Obesity, central adiposity, type 2 diabetes mellitus (T2DM), insulin resistance, and dyslipidemia are strongly associated with, and well-established risk factors for the development of both NAFLD and NASH (Table 1). Indeed, the prevalence of these risk factors is high in NAFLD and higher in NASH with obesity among 51% of NAFLD and 82% of NASH patients, T2D among 23% and 44%, and dyslipidemia among 69% and 72%, respectively (10). NAFLD is also present in a small percentage of lean individuals (body mass index [BMI] < 25 kg/m²): in the National Health and Nutrition Examination Data, 7% of non-obese individuals had NAFLD compared to 28% in obese individuals (16).

Table 1.

Prevalence of NAFLD and its risk factors in studies of patients with HIV.

Reference	n	Age*	M:F	BMI	Race, Nationality, or Ethnicity	Method of Diagnosis	Genotype (%)	NAFLD (%)	Identified Risk Factors	Factors that did not significantly contribute
Rafiq et al., 2013(49)	14,685: 79 HIV+ 14,606 HIV−	HIV+: 38 HIV− 35	3:1, HIV+ 1:1 HIV−	n/a	HIV+: 31% White; 52% Black HIV−: 67% White, 12% Black	Men: ALT>40, AST>37 Women: ALT/AST >31	Unknown	HIV+ 19.3 HIV− 10.9	HIV
Kardashian et al., 2017(50)	228: 121 HIV+ 107 HIV−	35–70	3:2	n/a	US: (race/ethnicity n/a)	Liver Fat Fraction Measured on Magnetic Resonance Imaging and Spectroscopy	Unknown	Women: HIV+ 1.9 HIV− 3.1 Men: HIV+ 4.6 HIV− 4.1	high VAT, insulin resistance (HIV+)	CD4+ cell count, HIV RNA level, ART
Price et al., 2014(51)	719: 465 HIV+ 254 HIV−	53	100:0	25.8	56% White	Liver-spleen attenuation values of <1.1 on Non-contrast Computed Tomography	PNPL A3-60 CC 35 GC 4.4 GG	HIV+ 13 HIV− 19	VAT, insulin resistance, PNPLA3 non-CC genotype (in HIV+), dideoxynucle oside analogs (in HIV+)	GCKR (rs780094), LYPLAL1 (rs12137855) and PPP1R3b (rs4240624) HIV-PNPLA3 (rs738409)
Nishijima et al., 2014 (52)	435 HIV+	>17	n/a	n/a	Japan	Abdominal Ultrasound	Unknown	31	BMI, dyslipidemia, higher ALT:AST ratio	Dideoxynucle oside analogues, duration of ART
Scheiner et al., 2015(53)	177 HIV/HCV	39	3:1	23.1	Austria (race/ethnicity n/a)	Liver biopsy, Transient Elastography with Controlled Attenuation Parameter	PNPL A3-57.8 CC: 37.3 GC: 5.1 GG: IL28B-10.2 TT: 57.1 TC: 32.2 CC	n/a	HCV-GT3, BMI, age	PNPLA3, L28B-genotype
Morse et al., 2015(54)	62 HIV+	n/a	n/a	n/a	US: (race/ethnicity n/a)	Liver Biopsy	Unknown	55 (NASH)	PNPLA3 minor allele, insulin resistance, obesity	Duration of HIV or ART, specific ART, history of OI, immune status, or duration of AST/ALT elevation
Jimenez -Sousa et al., 2016(55)	215 HIV/HCV	n/a	n/a	n/a	Spain	Liver Biopsy	PNPL A3-0.05 CC: 0.38 GC: 0.57G G	NAFLD CC: 70 CG: 64 GG: 57 Fibrosis >3 CC: 0 CG: 19 GG:: 25	n/a	n/a
Crum-Cianflone et al., 2009(56)	216 HIV	40+/−11	9:1	26	48% White, 27% Black, 14% Hispanic, 11% other	Liver Biopsy	Unknown	White: 35 African Americans: 31	BMI, waist circumference, lower HDL levels, higher triglyceride levels, use of lipid- lowering medication, metabolic syndrome	CD4 cell count, HIV-1 RNA, duration of HIV infection and ART
Guaraldi et al., 2008(57)	225 HIV	48	3:1	23.2	Italian	Liver-spleen attenuation values of <1.1 on Non-contrast Computed Tomography	Unknown	Infected: 37	Higher ALT:AST ratio male sex, greater waist circumference, longer NRTI exposure	Coronary artery calcium scores, diabetes
Lui et al., 2016(58)	240: 80 HIV+ 160 HIV−	54+/−11	9:1	n/a	Chinese	Transient Elastography and Proton Magnetic Resonance Spectroscopy	Unknown	HIV+: 28.7 HIV−: 27.5	n/a	n/a
Macias et al., 2015(59)	431: 275 HIV/HCV 156 HIV+	42–50	4:1	23.3	Spain	Transient Elastography	Unknown	NAFLD: 41.5 NASH: 28.3 SAMM50: TT: 60.5 CT/CC: 39.6 rs12743824: AC/AA: 37.8 CC: 49.2	Dominant SAMM50 rs738491-T allele	dominant LPPR4 rs12743824-A allele

^*Mean, median, or range; n/a= not available

NAFLD, non-alcoholic fatty liver disease; NASH, non-alcoholic steatohepatitis; VAT, visceral adipose tissue; BMI, body mass index; NRTI, nucleotide reverse transcriptase inhibitor; ART, antiretroviral therapy; OI, opportunistic infection; AST, aspartate aminotransferase; ALT, alanine aminotransferase

Diagnosis of NAFLD

As NAFLD is often asymptomatic (17), it is most commonly diagnosed incidentally during abdominal imaging performed for other indications, or by elevated liver enzymes tests during routine laboratory testing. The presence of hepatic steatosis can be identified with a variety of imaging techniques including ultrasound, computed tomography (CT) scan, vibration controlled transient elastography (VCTE or Fibroscan®) with controlled attenuation parameter (CAP), magnetic resonance (MR) imaging, MR spectroscopy (MRS) or MR imaging proton density fat fraction (MRI-PDFF) (18–20). However, although several advanced modalities are in development, these currently available imaging tests only identify steatosis and cannot distinguish steatosis alone from NASH.

A number of serum biomarkers for NASH have been tested including alanine aminotransferase (ALT), adiponectin, CK18 M30, leptin, and HOMA-IR, that, although associated with various pathophysiological mechanisms of liver damage, have limited (<80%) sensitivity and specificity for NAFLD versus other causes of liver disease or metabolic disturbance (21). The NAFLD Liver Fat Score predicts the presence of NAFDL with a sensitivity of 86–95% and specificity of 51–71%, depending on the score cut-off value, and includes the presence of metabolic syndrome, T2DM, fasting serum insulin, fasting serum aspartate aminotransferase (AST), and AST/ALT ratio (22).

As no serum biomarkers or imaging studies can currently distinguish steatosis alone from NASH, liver biopsy remains the “gold standard” for the diagnosis of NAFLD, NASH and for fibrosis staging (7). In addition to providing a definitive diagnosis of NAFLD, liver biopsy allows the evaluation of concurrent liver diseases such as hemochromatosis and autoimmune hepatitis and distinguishes between steatosis and NASH. Thus, liver biopsy is vital in ensuring the correct diagnosis, and for ongoing surveillance: patients with stage 3–4 fibrosis have increased risk for hepatocellular carcinoma and should undergo routine screening for hepatocellular carcinoma. Similarly, patients with NASH and/or fibrosis warrant evaluation for intensive medical management, weight reduction surgery, and clinical trial enrollment, where appropriate. Liver biopsy, however, is an invasive procedure and not without risk: serious complications occur in approximately 1% of patients and include infection, injury of other organs, bleeding, and rarely death (23, 24). Additionally, biopsy is expensive and subject to sampling error and observer variability(25). Hence, the development of non-invasive methods to diagnose NASH and fibrosis are of particular interest.

General Issues in the Management of NAFLD

Lifestyle modifications are the cornerstone for management of NAFLD and NASH. In a prospective cohort study of 12 months of lifestyle intervention weight reduction of at least ≥10% total body weight was associated with resolution of NASH in 90% of individuals and improvement in fibrosis by at least one stage in 45% (26). However, the ability to maintain weight loss beyond one year and the impact of weight regain on NASH histology is unknown. Routine physical activity (150–200 minutes per week) and the Mediterranean diet have also been shown to improve hepatic steatosis and are recommended for all individuals with NAFLD (27). However, as diet and exercise goals can be difficult for many patients to achieve and sustain (7), significant interest lies in the development of these pharmacotherapies to prevent and treat disease.

Interventions including vitamin E and pioglitazone have been investigated primarily for the treatment of biopsy-proven NASH, and have demonstrated variable efficacy (28–32). Among non-diabetics, vitamin E improves biopsy-proven NASH in a subset of patients, but has been associated with an increased risk of prostate cancer (33), hemorrhagic stroke (34) and a possible increase in all-cause mortality(32, 35). Pioglitazone has been associated with NASH resolution among both T2DM and non-diabetics (30, 32, 36). Concerns for significant weight gain (36), loss of bone mineral density (37), risk for exacerbation of congestive heart failure (38), and bladder, prostate, or pancreatic cancer (39) however, limit enthusiasm for pioglitazone in clinical practice. Clinical trials of novel therapies (i.e., obeticholic acid(40), tesamorelin [NCT03296831](41)), cenicriviroc [NCT02684591 and NCT03059446](42), OMACOR [NCT00760513](43), and elafibrinor [NCT01694849](44)), are underway.

NAFLD in HIV: Similarities and Differences

Liver disease is a frequent cause of death among HIV-infected persons in several large cohorts (45–48). Although viral hepatitis contributes to the majority of liver-related deaths among HIV-infected adults, the proportion of liver disease attributed to NAFLD is increasing and expected to play an increasing role in liver-related morbidity and mortality. Due in part to differing antiretroviral exposure (i.e., older versus more contemporary therapy), method of NAFLD diagnosis (as in the general population), co-infection with hepatitis C or B virus, and screening of select populations (such as age-specific or sex-specific cohorts), the true prevalence of NAFLD and/or its contribution to progressive liver disease among HIV-infected populations varies considerably (Table 1)(49–59). However, NASH appears to be more common among individuals with NAFLD and HIV when compared to those with NAFLD but without HIV. A case control study compared 33 individuals with biopsy-proven NAFLD and HIV to 33 persons with biopsy-proven NAFLD without HIV. HIV-infected individuals were more likely to have NASH (37% vs. 63%, P = 0.04) and signs of liver injury on biopsy such as lobular inflammation (21% vs. 67%, P < 0.001) or acidophil bodies (15% vs. 64%, P < 0.004)(60).

Many risk factors for NAFLD in HIV-infected patients are reflective of underlying causes of NAFLD in the general population, including obesity, T2DM or insulin resistance, hypertension, and dyslipidemia. However, the prevalence of NASH not only appears to be higher among HIV-infected adults with NAFLD but present among persons without the typical NAFLD phenotype of metabolic syndrome. For example, in a small cohort of biopsy-proven NAFLD, participants with HIV had a lower BMI, lower fat mass, and greater self-reported physical activity than their HIV-uninfected counterparts despite a similar age, caloric intake, liver histology, and degree of insulin resistance (61). Furthermore, the higher ratio of visceral adipose tissue (VAT) to BMI may lead to higher prevalence of NAFLD among lean HIV-infected compared to -uninfected populations (62–64). Thus, additional mechanisms are hypothesized to contribute to NAFLD pathogenesis among those with HIV.

Many early antiretroviral therapy regimens, particularly the nucleoside reverse transcriptase inhibitors, were associated with mitochondrial toxicity through multiple proposed mechanisms, including polymerase-gamma inhibition and reduction in mitochondrial DNA. The resultant mitochondrial toxicity and impaired fatty acid oxidation is thought to play a major role in the ectopic fat deposition in other tissues and organs, including the liver, and in the development of lipodystrophy (65–68). Indeed, therapies such as stavudine and didanosine with the greatest mitochondrial toxicity were associated with the most pronounced effects in regard to lipodystrophy, pancreatitis, and hepatotoxicity. Both Price at al., 2015 and Guaraldi et al., 2008 found associations between nucleoside reverse transcriptase inhibitor use (OR 1.44 per 5 years, P= 0.02 and OR 1.12 per year, P≤0.001, respectively) and hepatic steatosis (51, 57). A recent switch study from efavirenz to raltegravir-based therapy in HIV-infected patients with NAFLD resulted in significant improvement in hepatic steatosis by VCTE with CAP compared to patients who remained on efavirenz, suggesting that other antiretroviral regimens may have more of an effect on hepatic steatosis than previously recognized (69).

Beyond these specific antiretroviral therapies, most studies have not found an association between other HIV-specific characteristics and NAFLD prevalence. As shown in Table 1, the majority of recent studies have failed to find an association between severity of HIV disease (CD4+ T-cell count, HIV-1 RNA level, history of opportunistic infection), duration of HIV infection, or specific antiretroviral exposure and NAFLD presence or progression to NASH. In contrast to epidemiologic studies, a recently published murine model of HIV VPR protein expression suggests that HIV-1 may indeed impact NAFLD. Expression of VPR or infusion of synthetic VPR into transgenic mice expressing the HIV-1 accessory protein VPR resulted in increased de novo lipogenesis, impaired fatty acid oxidation, increased liver triglyceride content, and elevated liver enzymes(70).

Development of NAFLD and progression to NASH is thought to be multifactorial, and while many of these contributors are not specific to HIV, many occur more frequently or to a greater extent among HIV-infected populations. Chronic inflammation and immune activation, for example, are hallmarks of HIV pathogenesis and underlie the development of many comorbidities associated with HIV. Increased adiposity is associated with increased systemic and tissue level inflammation in HIV, and inflammatory cytokine production from adipose tissue and/or intra-hepatic lipid accumulation may contribute to further hepatic fat deposition or progression to NASH and fibrosis (71). Indeed, elevated systemic inflammation (IL-6 and TNF-alpha levels) have been associated with development of fatty liver (72). A recent study within The Multicenter AIDS Cohort Study (MACS), however, found that higher inflammatory cytokine levels (Intercellular Adhesion Molecule [ICAM]-1, C-reactive protein, IL-6, soluble tumor necrosis factor receptor [sTNFR]2) were associated with greater CT-defined hepatic steatosis among HIV-uninfected but not HIV-infected men, despite higher levels of these markers among HIV-infected men. Of note, CT-based diagnosis of hepatic steatosis may underestimate the prevalence of mild-to-moderate NAFLD (73) and limit the ability to compare inflammatory biomarkers by disease severity. Several post hoc analyses of randomized, controlled studies and small biopsy studies in HIV-uninfected populations have suggested a role for the anti-inflammatory effects of statins in NAFLD, as evidenced by improvement in biochemical, histologic, and imaging measures (74). Limited data among HIV-infected populations have failed to find benefit on steatosis (by the liver fat score)(75); studies assessing CT or MRI-based measures may further clarify whether anti-inflammatory agents, such as statins, impact liver steatosis in HIV.

Systemic inflammation in HIV may be facilitated by disruption in the gut-liver axis in this population (76), which may be further affected by differences in the microbiome by NAFLD and by HIV serostatus or antiretroviral therapy (77–79). Disruptions in gut wall permeability and perturbations in the normal microbiome may further exacerbate hepatic inflammation and disrupt glucose metabolism and fatty acid oxidation, resulting in greater hepatic fat deposition or progression of hepatic steatosis (80, 81).

Considerations of Disparities in NAFLD and HIV: Do genetic or sex factors contribute to health disparities in NAFLD and HIV?

Regardless of comorbidities, minorities and women tend to experience later HIV diagnosis and poorer HIV treatment outcomes (1–5). Furthermore, minorities and women living with HIV have a greater prevalence of obesity and the greatest risk for significant weight gain with HIV treatment (6). Although socioeconomic factors may impact these differences, race, ethnicity, and sex further contribute to disparities in NAFLD among HIV-infected populations. Here, we will summarize what is known on the role of genetics and sex on NAFLD risk in the general population, and then detail what is currently known in among persons living with HIV.

Similar to many other disease processes, disparities exist in the prevalence, progression, and prognosis of NAFLD. In the US, studies consistently show that Caucasians and Hispanics have the highest and African Americans have the lowest prevalence of NAFLD. Though the prevalence varies by study population, NAFLD prevalence among Hispanics ranges from 28 to 58% and African Americans from 11 to 35% (13). Globally, NAFLD, as diagnosed by imaging, follows similar racial/ethnic disparities, with South America and the Middle East having the highest prevalence of NAFLD at 30–32%, Asia at 27%, North America and Europe at 24%, and Africa with the lowest prevalence at 13% (10). A recent meta-analysis of 34 studies and nearly 370,000 patients further described a greater prevalence of NASH among Hispanics compared to Caucasians or Blacks, but no significant differences in the proportion of significant fibrosis by race/ethnicity (82). Studies on prognosis were limited and unable to be pooled, but suggested a greater odds of cirrhosis among Hispanics, and all-cause mortality among Blacks (82).

These differences may be partially explained by differences in lifestyle, body fat distribution, and co-morbidities, including T2DM, dyslipidemia, hypertension and the metabolic syndrome, that vary by race and ethnicity. For Hispanics, the higher prevalence of progression from NAFLD to NASH is overwhelmingly attributed to obesity and insulin resistance (13), and studies have shown a similar prevalence of hepatic steatosis between Hispanic and non-Hispanic Caucasians, when adjusted for total body fat (83, 84). In contrast, the low prevalence of NAFLD, NASH, and associated fibrosis in African Americans persisted even after adjusting for differences in BMI (83), insulin resistance, or ethanol ingestion (13). Since the low prevalence in African Americans does not appear to be completely explained by racial differences in risk exposure, this begs the question of the role of genetic differences.

In addition to the difference in prevalence, the rate of heritability of NAFLD also differs based on the race/ethnicity. According to large population studies, Hispanics have high heritability of NAFLD at 33%, Caucasians have 16–27% and African Americans 14% (85). Although some of these differences in prevalence and heritability could be due to cultural, lifestyle, or environmental factors, this again suggests an underlying genetic component in NAFLD and NASH risk.

Genetic Factors in the NAFLD Risk: Data from HIV-Uninfected and HIV-Infected Populations

Genetic differences in the risk for NAFLD or NASH progression in both the general population and among hepatitis C virus (HCV)-infected populations are well described (85–87). Many of these polymorphisms involve genes that are involved in inflammatory responses, metabolism, or mitochondrial haplotypes (87). The strongest and most consistent associations with the presence and progression of NAFLD in these populations has been with the rs738409 single nucleotide polymorphism (SNP) on the PNPLA3 gene(88, 89). The data on the role of rs738409 SNP on PNPLA3 gene among HIV/HCV co-infected or HIV mono-infected populations is more limited, as reviewed below and in Table 2 (51, 53–55, 90).

Table 2.

Role of PNPLA3 in NAFLD in Adults without HIV, with HIV, and with HIV/Hepatitis C Virus Co-infection

	HIV-uninfected	HIV-infected	HIV/Hepatitis C Virus Co-infected
Positive associations	Speliotes et al., 2011(88): PNPLA3 is associated with NAFLD	Price et al., 2014(51): PNPLA3 non-CC genotype (rs738409) is independently associated with increased odds of fatty liver (OR=3.30; 95 % CI=1.66, 6.57; p=0.001) Morse et al., 2015(54): G allele is 3.9 times more likely to be present in NASH than non-specific changes on liver biopsy.	Jimenez-Sousa et al., 2016(55): PNPLA3 minor allele is associated with NASH (CC: 0% CG: 18.5% GG: 25.2%; p=.029)
Negative associations	Price et al., 2014 (51): PNPLA3 (rs738409) is not significantly associated with fatty liver (OR=1.14; 95 % CI=0.51, 2.54; P=0.75).		Schreiner et al., 2015(53): PNPLA3 allele is not an independent risk factor for NAFLD Jimenez-Sousa et al., 2016(55): PNPLA3 minor allele is not associated with NAFLD (CC: 70% CG: 63.4% GG: 57.9%; p=0.26)

NAFLD, non-alcoholic fatty liver disease; NASH, non-alcoholic steatohepatitis

Palatin-like phospholipase domain-containing protein 3 (PNPLA3) gene codes for adiponutrin or calcium independent phospholipase A2-epsilon that enables triglyceride hydrolysis in adipocytes. PNPLA3 has been associated with the presence of NAFLD and with progression to NASH and fibrosis development among HIV-uninfected populations in multiple studies, after accounting for BMI, T2DM, alcohol use, and ancestry (85, 88, 91). The GG genotype is associated with a greater accumulation of steatosis than the CC genotype, with up to 73% increase in percentage of steatosis (88, 89). Prior studies have shown that Hispanics, who have the highest prevalence of NAFLD, also have the highest prevalence of GG genotype (25%) and Blacks, with the lowest prevalence of NAFLD, have the lowest prevalence of GG genotype (2%)(91). A greater percentage of the GG genotype in some patient populations may explain similarities in NAFLD prevalence despite differences in obesity and metabolic syndrome: 13–19% of Asians have GG genotype while only 4% of Caucasians have GG genotype. These genetic differences may, in part, explain the similarity in prevalence of NAFLD between Asians and Caucasians (91).

The rs738409 SNP on the PNPLA3 gene may be most useful in predicting NAFLD prevalence and progression among non-obese populations. For example, among a Japanese cohort, a greater proportion of the rs738409 GG genotype was found among non-obese, NAFLD patients than among individuals with obesity and NAFLD. In a separate cohort in Italy, rs738409 was associated with a greater risk of disease progression among non-obese but not obese NAFLD patients (92).

The data on genetic differences among HIV-infected populations, specifically the role of PNPLA3 and the rs738409 SNP in NAFLD prevalence and severity, are summarized in Table 2. In the MACS, a cohort of men who have sex with men living with or at risk for HIV infection, found a strong association between PNPLA3 non-CC genotype and hepatic steatosis prevalence as defined by a liver-to-spleen attenuation ratio <1.0 on non-contrast CT among HIV-infected men (≤10% with HCV). In this cohort excluding heavy alcohol users, 19% of HIV-uninfected and 13% of HIV-infected men had hepatic steatosis and the proportion of PNPLA3 genotype did not differ by HIV serostatus. The non-CC genotype of PNPLA3 (rs738408) was associated with an increased odds of steatosis in the combined (HIV-infected and uninfected) cohort (OR 2.06; 95% CI 1.24, 3.33 p=0.005) and among HIV-infected men only (OR = 3.30, 95% CI 1.66, 6.57; P= 0.001), but not among HIV-uninfected men only (OR =1.14; 95% CI =0.51, 2.54; p=0.75), with a trend towards an interaction between HIV serostatus and PNPLA3 (51). Of note, CT-based estimates of hepatic steatosis have a lower sensitivity for detection of mild to moderate steatosis (73), thus prevalence of steatosis by CT-based estimates in the MACS is generally lower than prevalence among other cohorts of similar characteristics.

Results from other studies report a variety of findings, likely influenced by differences between studies characteristics including race/ethnicity, sex, co-infection with hepatitis B or C virus, alcohol use, HIV treatment history, and metabolic syndrome parameters. Among 62 HIV-infected adults at the National Institutes of Health clinic with elevated liver function tests and without HCV co-infection (nearly 30% Hispanic), Morse, et al found a significant association between the rs2281135 and rs228115 SNP in PNPLA3 and increased ALT, steatosis grade, frequency of NASH, and fibrosis stage (54). Similar findings were seen in a cohort with HIV/HCV coinfection: Among 215 HIV/HCV co-infected patients with liver biopsy, the PNPLA3 rs738409 SNP was associated higher FIB-4 scores and the presence of fibrosis (55). Similarly, a separate study of 168 HIV/HCV co-infected patients with biopsy-proven NAFLD/NASH (all Caucasian, none with T2DM or insulin resistance) found a strong association between PNPLA3 and severe steatosis, independent of BMI, CD4+ t-cell count, liver enzymes, total cholesterol, antiretroviral regimen, and HCV genotype (93).

In contrast, findings from other HIV study populations do not support this association. Dold, et al. determined the frequency of several SNPs among 117 HIV-infected patients without active hepatitis B or C co-infection: none of the alleles of interest differed by NAFLD prevalence or fibrosis presence (94). Scheiner, et al also failed to find a significant association with PNPLA3 rs738409 and biopsy-proven fibrosis or steatosis among 177 HIV/HCV co-infected patients, and no difference in HCV treatment response by PNPLA3 genotypes (53). Furthermore, among 431 HIV-infected patients in Spain (64% with active HCV co-infection), Macias, et al failed to find a similar association with the PNPLA3 gene, but did find associations between SNPs in the LPPR4 and SAMM50 genes and NAFLD and NASH, respectively (59). Of note, many of these are small studies, and many include patients with HCV co-infection, thus generalizability of the results to HIV mono-infected populations are limited.

LPR4 gene codes for a protein in the Lipid Phosphate Phosphatase (LPP) family. LLP dephosphorylates and inactivates bioactive lipid metabolites of cellular function (95). HIV has been known to infect the hypothalamus and disturb its metabolic functions and LPR4 is expressed in axonal membranes, especially in the hypothalamus. This may explain an association between LPR4 with NAFLD in HIV (90). SAMM50 gene codes for a protein vital to the structure of the mitochondria (59). Several other SNPs have been associated with liver disease progression in HIV/HCV co-infection, without mention of underlying NAFLD/NASH (87). SNPs within the mitochondria, consistent with haplogroups HV and H were associated with slower fibrosis progression and haplotype U with faster progression in study of HIV/HCV co-infection, which may further support racial/ethnic differences linked to haplotype, but again, the role of these mitochondrial haplotypes in NAFLD and NASH without HCV are not established (96).

The Role of Sex in the NAFLD Risk: Data from HIV-Uninfected and HIV-Infected Populations

The impact of sex differences on the prevalence or progression of NAFLD among the general population is conflicting. Both the National Health and Nutrition Examination Survey (NHANES) and the Dallas Heart Study found that men (particularly white men), had nearly double the risk of NAFLD compared to women (9, 13, 57, 97). In contrast, other studies have found that women have greater prevalence of NAFLD in the setting of high visceral fat (98), and an Australian study found that adolescent girls are 1.5 times more likely than adolescent boys to have NAFLD (99). This conflicting data may be related, in part, to the presence of endogenous estrogens within visceral fat, or lower levels of estrogen in adolescent girls. For example, estrogen in pre-menopausal women is associated with hepatoprotective effects including improved lipids and reduced fibrogenesis; these effects may be lost with menopause, and, in conjunction with visceral fat gains in menopause, contribute to NAFLD (100).

Other studies suggest that sexual dimorphism might affect gene variability on NAFLD: a meta-analysis including sixteen studies found a negative correlation between the proportion of men in studies and the observed association between the rs738409 SNP and liver fat content (88), such that as the proportion of men increased, the difference in liver fat content between alleles of rs738409 SNP on PNPLA3 gene decreased. Additionally, proteins such as SREBP-1c and liver X receptor that affect the expression of PNPLA3 gene (101) are modulated by estrogen (102), and a greater association between PNPLA3 rs738409 and NAFLD prevalence in HIV-uninfected women has been observed (26). As many prior studies did not adjust for pre or post-menopausal status, the influence of estrogen could explain conflicting reported data on sex differences. Lastly, the modality of NAFLD diagnosis differed between cohorts, and imaging accuracy may also vary by sex.

Data among HIV-infected populations is limited and similarly conflicting: in a recent analysis of 87 women and 142 men, with and without HIV infection from the San Francisco Women’s Interagency HIV Study and the community, HIV-infected women had less steatosis than HIV-uninfected women after adjusting for demographics, lifestyle, and metabolic factors (50). No differences by HIV serostatus were detected among men in the fully adjusted models. In an Italian cohort with HIV, Guaraldi et al. found that male sex was associated with a greater odds of CT-defined NAFLD (OR 2.5; 95% CI 1.1, 5.8) compared to female sex, after adjusting for waist circumference (57). Multiple studies have suggested that men are at lower risk for progression to fibrosis, though the number of women and low fibrosis stages included in many of these studies limit conclusions (54, 56, 103, 104)

Do Genetic and Sex Differences Impact Clinical Care?

Even though genes such as PNPLA3 are strongly associated with NAFLD, the clinical utility of these genetic underpinnings have yet to be established, and routine screening is not recommended. Currently, genetic association utility is generally limited to use as a covariate in determining associations with NAFLD prevalence or disease severity in research settings, and most studies have not found a marked improvement in predictive accuracy with inclusion of genetic information. Larger studies utilizing genetic data and incorporating other “-omics” based outcomes may help to advance understanding of these underlying genetic factors. As race, ethnicity, and sex underlie some genetic differences, these factors may contribute to an understanding of disease progression or mortality due to NAFLD-associated comorbidities. For example, Blacks with NAFLD in the Multi-Ethnic Study of Atherosclerosis had a greater prevalence of abdominal aortic calcification compared to whites, suggesting that NAFLD-related metabolic disease may also vary by race and ethnicity (105), as NAFLD prevalence does. Similarly, among Chinese adults, the presence of NAFLD was associated with more pronounced differences in lipids and renal function in men than among women (106). Further studies on NAFLD prevalence and prognosis by race, ethnicity, sex, and HIV serostatus can better inform the diagnosis and management of concomitant comorbidities, and understand the role that NAFLD plays in the development of these comorbid disease by race, ethnicity, and sex. Inclusion of adequate numbers of women in studies of NAFLD and HIV may help elucidate sex-based differences in NAFLD pathogenesis that may explain observed differences in prevalence and progression, including hormone-associated pathways. Lastly, very little is known regarding the prevalence and pathogenesis of NAFLD in transgender persons. Although pre-menopausal endogenous estrogen appear hepatoprotective among natal females, exogenous estrogen therapy among transgender persons worsens insulin sensitivity and increase inflammation, triglycerides, and subcutaneous and visceral adipose tissue (107–110).

Conclusion

In conclusion, with the drastic increase in NAFLD prevalence worldwide, there is a tremendous need for accurate scalable screening, prognosis, treatment prediction, and new interventions. Individuals living with HIV have unique risk factors for NAFLD but data on the prevalence of NAFLD and NASH and the impact of gender, race and ethnicity on these risk factors and NAFLD overall remain undefined. Further work is needed to understand the pathogenesis of NAFLD in HIV.