NAFLD and NASH in Postmenopausal Women: Implications for Diagnosis and Treatment

Abstract

Nonalcoholic fatty liver disease (NAFLD) prevalence in women is increasing worldwide. Women of reproductive age have lower rates of NAFLD compared with men; however, this protection is lost following the menopausal transition when NAFLD prevalence in postmenopausal women becomes similar to or surpasses that in age-matched male counterparts. Ongoing epidemiological, clinical, and experimental studies indicate greater NAFLD risk and higher rates of severe hepatic fibrosis in postmenopausal women relative to premenopausal women, and that older women with NAFLD experience greater mortality than men. Investigations involving ovariectomized animal models demonstrate a causal relationship between estrogen deficiency and heightened susceptibility to the development of fatty liver and steatohepatitis, although dietary factors may exacerbate this complex relationship. The accumulated findings suggest that a better understanding of the interplay among menopausal status, metabolic comorbidities, and sex steroids in NAFLD pathogenesis is needed. Further, the mechanisms underlying the difference in NAFLD risk between postmenopausal and premenopausal women remain incompletely understood. The goals of this review are to summarize studies of NAFLD risk in postmenopausal women, discuss results from animal models of estrogen deficiency, and explore the development of NAFD within the context of altered sex hormone profiles resulting from the menopausal transition. Potential implications for the prevention, diagnosis, and treatment of NAFLD in this relatively understudied cohort are also addressed.

Nonalcoholic fatty liver disease (NAFLD) is a chronic disease that develops as a result of excessive accumulation of triglycerides in the liver. Nonalcoholic fatty liver disease is also a common disease, affecting ~24% of the global population (1). NAFLD prevalence in the United States was estimated at 83.1 million cases in 2015 and is predicted to increase 21% (100.9 million cases) within the next 10 years (2). The number of nonalcoholic steatohepatitis (NASH) cases is expected to climb, from 16.5 million cases in 2015 to a projected 27 million cases in 2030, while rates of NASH-related comorbidities, such as decompensated cirrhosis and hepatocellular carcinoma (HCC), will also likely increase (2). Annual direct medical costs for NAFLD are estimated at $103 billion, and societal costs, measured in terms of annual quality-adjusted life-years lost as a result of having NAFLD, were calculated at $188 billion, making the total annual economic burden of NAFLD in the United States $292 billion (3). Even without taking into consideration the indirect costs of the disease, such as lost work-related productivity, it is clear that NAFLD places a substantial burden on the US health care system.

In general, NAFLD prevalence is higher in men compared to women (4). However, recent reports show a trend in increasing NAFLD prevalence among women over the past 10 years (5), and women are experiencing a more steeply rising mortality rate compared to men (6). In addition, NASH is now the leading indication for liver transplantation in women (7). While individuals of any age, including children, may develop NAFLD, postmenopausal women are especially susceptible to the development of both NAFLD and advanced NASH fibrosis compared to premenopausal women. Emerging studies have provided a better understanding of the interplay among menopausal status, metabolic comorbidities, and the role of endogenous hormone deficiency and nutritional patterns in the development of NAFLD in this population; however, the relationships among these factors and the mechanisms underlying the bias in NAFLD risk between postmenopausal and premenopausal remain incompletely understood. The goals of this review are to summarize studies of NAFLD risk in postmenopausal women, discuss results from animal models of estrogen deficiency, and explore the development of NAFD within the context of altered sex hormone profiles resulting from the menopausal transition. Potential implications for the prevention, diagnosis, and treatment of NAFLD in postmenopausal women are also addressed.

Development and Progression of NAFLD

Nonalcoholic fatty liver disease is defined by the presence of hepatic steatosis in the absence of excessive alcohol consumption, certain hereditary disorders, and use of medications that may contribute to secondary hepatic fat accumulation (8). Hepatic steatosis arises from an imbalance of fat accumulation and disposal, which result from increased de novo lipogenesis and/or hepatic fatty acid uptake, and decreased lipid disposal, respectively (9). Hepatic steatosis is defined as liver fat fraction exceeding 5% to 10% by weight (10), and is also estimated as the percentage of hepatocytes containing visible triglycerides (11). The clinical severity of hepatic steatosis is based on the percentage of fat within the hepatocytes: grade 0 (healthy, <5%), grade 1 (mild, 5–33%), grade 2 (moderate, 34–66%), and grade 3 (severe, >66%) (12). Dietary composition and genetic factors are associated with the development of hepatic steatosis, and hepatic steatosis is commonly found in in individuals with obesity, dyslipidemia, and type 2 diabetes (T2D).

Nonalcoholic fatty liver disease spans a histological spectrum that includes steatosis, hepatic inflammation, and ballooning of hepatocytes (ie, NASH), liver fibrosis, and HCC (13, 14). Like steatosis, NASH activity and stage of fibrosis are graded to refine the histological presentation of the disease. In the Brunt grading and staging system (15), which is based on a number of histological features, there are 4 stages of fibrosis: perisinusoidal fibrosis (Stage 1), perisinusoidal with periportal fibrosis (Stage 2), bridging fibrosis (Stage 3), and cirrhosis (Stage 4). The NASH Clinical Research Network scoring system uses a NASH Activity Score (NAS) to grade and stage disease (11). The NAS encompasses the 3 histological features of steatosis (0–3), lobular inflammation (0–3), and ballooning degeneration (0–2), and an expanded Stage 1 fibrosis score that delineates delicate perisinusoidal fibrosis (1A), dense perisinusoidal fibrosis (1B), and portal fibrosis without perisinusoidal fibrosis (1C).

Nonalcoholic fatty liver disease is a progressive disease (16). Approximately 15% to 20% of NAFLD patients progress to NASH (17), 30% to 40% of NASH patients develop fibrosis (18), and 15% to 20% of NASH + fibrosis advance to cirrhosis (19). Nonalcoholic steatohepatitis patients with cirrhosis often develop HCC (20). Nonalcoholic steatohepatitis is recognized as the major cause of chronic liver disease and is emerging as the most common indication for liver transplantation (21). Liver transplantation is the most effective treatment strategy for decompensated liver disease related to NASH fibrosis, although NAFLD recurrence in transplant patients is high (22). Liver-related morbidity and mortality is increased in NASH patients (1) and rises disproportionately with fibrosis stage (23).

Epidemiology of NAFLD in Postmenopausal Women

NAFLD prevalence in postmenopausal women

Recent studies indicate that women are experiencing greater increases in NAFLD prevalence, and are experiencing a rising mortality rate compared with men. In 13 559 women participating in the National Health and Nutrition Examination Surveys, the unadjusted prevalence of NAFLD increased from 18.54% (1988–1994) to 21.36% (1999–2006) to 24.86% (2007–2014) (5). The age-standardized prevalence of NAFLD in this cohort increased from 20.96% (1988–1994) to 26.19% (2007–2014). Women with NAFLD in this cohort had a significantly higher age-adjusted cardiovascular disease (CVD) prevalence, 5-year all-cause mortality rate, and 5-year cardiovascular mortality rate compared to those without NAFLD. A recent analysis of mortality data from the National Vital Statistics System over the 2007–2016 time period found that the age-standardized death rate in individuals with NAFLD increased by 15% (average annual percentage change 1.98%; P < 0.001), with the most significant increases in female decedents (2.99% in women vs 1.16 in men; P = 0.003) (6). The analysis also showed female NAFLD patients experienced 5% more deaths from cirrhosis compared with men, indicating that many women may be experiencing more severe hepatic manifestations of NAFLD.

In general, the prevalence of NAFLD is higher in men than it is in women of reproductive age; however, following menopause, this sex difference is reduced or abolished (24–30). In a cohort of 39 151 Japanese individuals, the average prevalence of fatty liver over a 12-year period was 26% in males and 12.7% in females, and while NAFLD prevalence increased annually in both sexes, only female participants experienced an age-related increase, with the highest prevalence observed in the 60–69 age group (31). In contrast, NAFLD prevalence in males remained consistent (~25–27%) across the 30–60 age group. After the age of 70, women had a higher prevalence of fatty liver than men (19.4% vs 14.9%). In an independent study, gradual age-related increases in NAFLD prevalence were also observed in women (3.9% in the 21–39 age group; 7.6% in 40–49 age group; 14.0% in 50–59 age group; 18.9% in 60–69 age group), but not men (32). The age-dependent disparity between the sexes has largely been attributed to hormonal changes in women occurring as a result of menopause. Indeed, menopausal status is associated with changes in body composition, specifically, increased body mass index (BMI)-adjusted waist circumference (33) and levels of intra-abdominal fat (34, 35), which are associated with greater risk of metabolic diseases, including T2D and insulin resistance (36).

Across a number of studies, the prevalence of NAFLD has been shown to be significantly higher in postmenopausal women relative to premenopausal women (32, 37–42). Key features of these studies are delineated in Table 1. Risk factors for NAFLD also appear to differ according to reproductive status. For example, in Japanese women, metabolic syndrome and weight gain were dependent risk factors for NAFLD in both pre- and postmenopausal women, but age was a risk factor only in premenopausal women (32). Chung et al (38) reported that low HDL-cholesterol, central obesity, and homeostasis model assessment-estimated (HOMA) insulin resistance showed a significant association with increased risk of NAFLD in premenopausal women, while T2D, hypertriglyceridemia, and central obesity increased risk only in postmenopausal women, following adjustment for confounding variables. In postmenopausal, but not premenopausal, women, serum uric acid levels were found to be associated with NAFLD, even after adjusting for factors associated with metabolic syndrome (40). However, an independent study reported that NAFLD risk was associated with increased serum uric acid concentration in all women, independent of reproductive status (37). Univariate analyses showed that postmenopausal status increased the risk of NAFLD, although the statistical significance of this relationship was lost after adjustment for confounding factors in some (32, 38), but not all (41, 43) studies. Interestingly, assessment of independent predisposing factors for fatty liver revealed that the odds ratios for BMI, triglyceride level (TG), and fasting blood glucose (FBS) were much higher in women compared to men (BMI: 8.7 vs 5.5; TG: 5.6 vs 3.3; FBS: 5.9 vs 2.6, respectively) (31).

Table 1.

NAFLD Prevalence in Premenopausal and Postmenopausal Women

Ethnicity	NAFLD Diagnosis	Premenopausal, N (%)	Postmenopausal, N (%)	Ref
Mexican	USa	90 (32.2%)	57 (57.9%)	(39)
Japanese	USa	61 (6.0%)	81 (14.8%)	(32)
Korean	USa	55 (24.9%)	164 (61.7%)	(40)
Korean	US	105 (14.4%)	189 (27.2%)	(38)
Korean	USa	113 (17.4%)	92 (27.0%)	(41)
Chinese	USa	824 (24.2%)	941 (49.1%)	(37)
Chinese	USa	105 (61.0%)	226 (56.9%)	(42)

^aabdominal ultasonography.

NAFLD-related fibrosis and postmenopausal status

Female NAFLD patients were reported to have more deaths from cirrhosis than men, suggesting that women may experience more severe forms of the disease (6). Correspondingly, a number of studies have indicated differences in the prevalence of NAFLD-related fibrosis between men and women. In a pilot analysis of 537 individuals with biopsy-proven NAFLD, the prevalence of advanced fibrosis (stage 3 or 4) was significantly higher in postmenopausal women (36.1%) compared to men (17.7%) and premenopausal women (13.5%) (44). Postmenopausal women with larger waist circumference were more likely to have advanced fibrosis, while in men and premenopausal women, extremity size, but not abdominal size, was associated with stage 3/4 fibrosis. Similarly, in 541 individuals with biopsy-proven NAFLD, the presence of advanced fibrosis was significantly higher in postmenopausal women (27.6%) compared with premenopausal women (14.4%) and men (22.2%) (45). Grade 2 hepatocyte ballooning was also different: 32.8% in postmenopausal women, 22.7% in premenopausal women, and 18.0% in men. After adjusting for site of enrollment, grade of portal inflammation, hepatocyte ballooning, race, diabetes/prediabetes, BMI, and hypertension, and excluding women whose reproductive status was unknown, postmenopausal women had a 1.5-fold increased risk of having greater severity of liver fibrosis than premenopausal women. In addition, age ≥50 years was associated with a significantly increased risk of advanced fibrosis only in women, but not in men (45). In 419 nonobese women with biopsy-proven NAFLD (90 premenopausal and 329 postmenopausal), postmenopausal women had a greater risk of advanced fibrosis compared to premenopausal women (OR = 2.2; [95% CI: 1.1–4.5]; P = 0.03), after adjusting for hepatic inflammation, hepatocyte ballooning, BMI, impaired glucose tolerance or diabetes, and hypertension (46).

In an analysis of 244 Italian women (74 premenopausal and 170 postmenopausal) and 244 age-matched men, a nonsignificant trend was observed between menopausal status and F2-F4 fibrosis, but no association was seen between male sex and fibrosis (47). When restricting the analysis to women, menopause (OR: 2.72; [95% CI: 1.02–7.24]; P = 0.04), NASH (OR: 3.21; [95% CI: 1.74–5.93]; P < 0.001), and HDL-cholesterol (OR: 0.98; [95% CI: 0.96–1.00]; P = 0.04) were independently associated with fibrosis. In a subgroup analysis limited to age- and BMI-matched premenopausal (N = 20) and postmenopausal (N = 20) women, postmenopausal women had a significantly higher prevalence of F2-F4 fibrosis. Interestingly, women who experienced menopause before the age of 40 years (N = 143) had a 90% increased risk of more severe fibrosis (ACOR = 1.9; [95% CI: 1.3-2.7]; P = 0.001) compared with women with an age of menopause ≥40 years (N = 345) (48). When restricting the analysis to individuals with borderline or definite NASH, premature menopause was associated with 60% increased risk of having more severe fibrosis. To address the potential role of duration of estrogen deficiency, the time from menopause, measured as the age at study enrollment minus the self-reported age at menopause, was assessed: a significant association was observed between time from menopause and more severe fibrosis (COR = 1.2; [95% CI: 1.1–1.3]; P < 0.0001), which remained significant even following adjustment for age at enrollment, race, waist circumference standardized by BMI, current smoking, current alcohol use, hypertension, diabetes/impaired fasting glucose, homeostatic model assessment of insulin resistance (HOMA-IR), and hormone replacement therapy (HRT). Restricting the analysis to individuals with histologic diagnosis of borderline or definite NASH did not affect this association. These results suggested that the length of time in which a women experiences estrogen deficiency may exacerbate postmenopausal fibrosis risk.

Estradiol and the Development of NAFLD and NASH

Hepatic steatosis and hepatocellular injury in models of estrogen deficiency

Levels of 17β estradiol (E2), an endocrine hormone produced by the ovaries and circulated in plasma, decline significantly following menopause. The impact of estrogen deficiency on liver lipid metabolism has been investigated extensively in animal models. Aromatase-deficient (ArKO) mice, which lack endogenous estrogen production, accumulated abdominal fat and developed abundant fat droplets in liver cells of the centrilobular and intermediate zones, indicative of hepatic steatosis (49). ArKO mice also showed downregulation of genes encoding enzymes involved in fatty acid β-oxidation and lower basal levels of total fatty acid β-oxidation activity compared to control mice. E2 supplementation prevented the development of steatosis in ArKO mice and restored gene expression and β-oxidation to levels observed in wild-type animals, indicating that estrogen receptor-mediated signaling pathways may play a critical role in basal hepatic lipid metabolism.

Studies with ovariectomized (OVX) mice have been important in understanding estrogen signaling in the development and progression of NAFLD (50–52). Kamada et al (53) investigated the intersection between estrogen deficiency and diet by comparing OVX and sham-operated (SO) mice on normal chow (NC) or high fat, high cholesterol (HFHC) diet. Following 6 weeks of diet, body and liver weights, as well as serum ALT and cholesterol levels, were significantly higher in OVX-HFHC mice compared with SO-HFHC mice. While there were no differences in liver histology between NC-fed OVX and SO mice, OVX-HFHC mice showed greater inflammatory cell infiltration and higher lobular inflammation and hepatocyte ballooning scores compared with SO-HFHC mice. OVX-HFHC mice also experienced greater hepatic fibrosis and higher expression levels of fibrogenic genes relative to SO-HFHC animals. In OVX-HFHC mice implanted with estrogen pellets (OVX-HFHC_E2), body weight (but not liver weight) and serum levels of ALT and cholesterol were significantly reduced compared to OVX-HFHC controls. Importantly, OVX-HFHC_E2 mice had reduced levels of liver inflammation, hepatocyte ballooning, and hepatic fibrosis compared with OVX-HFHC mice. These results have been consistent with reports showing detrimental effects of ovariectomy on hepatocellular injury and liver fibrosis and subsequent amelioration with E2 supplementation (54–56) and support other findings showing mitigation of high-fat diet (HFD) effects on ectopic hepatic lipid content by estradiol in intact female mice (57). In OVX mice fed a diet high in saturated fat (60% fat), hepatic expression of spliced X-box protein-1 (s-XBP1), tumor necrosis factor alpha (Tnfa), phosphorylated eukaryotic initiation factor 2 alpha (p-EIF2a), C/EBP homologous protein (CHOP), and glucose regulated protein 78 (Grp78) increased, as did the activities of cytosolic cytochrome c and caspase 3, compared with SO mice fed normal chow. In these animals, estradiol supplementation restored gene expression and enzyme activity to levels observed in control mice (58). Levels of acetyl CoA carboxylase (ACC), which catalyzes the first step of fatty acid synthesis, were also elevated in OVX mice and restored with E2 (59).

In humans, surgical menopause induced by removal of the ovaries (ie, oophorectomy) is associated with increased risk of NAFLD. In a study of 875 women, the cumulative risk of NAFLD following oophorectomy was 14.1%, 20.5%, and 38.4% at 1, 2, and 5 years post-surgery, respectively (60). The effects of oophorectomy on NAFLD varied by time from surgery and age group. The 2-year cumulative risks of NAFLD for the oophorectomy and nonoophorectomy groups were, respectively, 36.4% and 12.2% for <40 years, 23.9% and 21.2% for age 40–49 years, and 17.1% and 9.8% for ≥50 years. Five-year cumulative risks for the oophorectomy and nonoophorectomy groups followed a different pattern: 49.4% and 49.8% (<40 years), 46.1% and 24.9% (40–49 years), and 34.5% and 22.6% (≥50 years). In this study, oophorectomy was found to be an independent predictor of developing NAFLD after surgical operation (hazard ratio = 1.70; 95% CI: 1.01–2.86; P = 0.047).

Estrogens exert cellular effects largely through estrogen receptor α (ERα) and estrogen receptor β (ERβ). Early studies in mice demonstrated that ERα mediated favorable effects of E2 on HFD-induced glucose intolerance and insulin resistance (61), while liver-specific deletion of ERα led to severe hepatic steatosis and hepatic insulin resistance (59). As expected, E2 was unable to prevent either hepatic lipid accumulation or insulin-mediated suppression of liver triglyceride export in these animals. Silencing of hepatic ERα expression in HFD-fed C57BL/6 female mice led to excessive liver fat accumulation, while ERα overexpression improved steatosis in ob/ob mice (62). Estrogen receptor α regulates gene expression through 2 activation functions (ERα-AF1 and ERα-AF2) located in different domains of the protein. Like ERα knockdown mice, ERα-AF2-deficient animals (ERaAF-2°) develop adiposity, insulin resistance, and glucose intolerance, while the metabolic profile in ERaAF-1-deficient mice was comparable to wildtype mice, suggesting that ERα-AF2, but not ERα-AF1, mediates positive estrogenic effects on metabolic parameters (63). E2 supplementation ameliorated HFD-induced metabolic effects in wildtype and ERaAF-1° OVX mice, but had no impact on ERaAF-2° and ERα ^-/- mice. E2 also affected the expression of lipogenic genes, including Stat3, Srebp1C, Scd1, and Fas, in livers and hepatocytes of wild type and ERaAF-1° mice, but not in ERaAF-2° and ERα ^-/- animals (63).

Aged female zebrafish, with levels of circulating estradiol indicative of ovarian senescence, also developed hepatic steatosis and fibrosis in response to overfeeding and at levels comparable to those observed in male fish (47). In contrast, young female fish developed significantly less steatosis and did not experience appreciable hepatic fibrosis. The hepatic findings in aged female fish are consistent with those observed in OVX mice, suggesting that the intersection between diet and estrogen deficiency may be a potent driver of liver fat accumulation and hepatic fibrosis. Combined, different animal models of ovarian senescence provide consistent evidence supporting a role for estrogen deficiency in the promotion of hepatocellular injury and hepatic fibrosis and suggest potential mechanisms by which estrogen deficiency in postmenopausal women might exacerbate NASH progression.

Serum estradiol levels in postmenopausal women with NAFLD

The relationship between menopause in humans or estrogen deficiency in animals and hepatic fat accumulation suggests that low serum estradiol levels may be associated with NAFLD. In one study, no significant difference in serum estradiol levels was observed between postmenopausal women with (N = 33) and without (N = 33) NAFLD (39). However, when the relationship between the predicted probability of NAFLD and serum estrogen concentration was analyzed, lower serum estradiol levels were associated with a higher probability of developing NAFLD. In 22 postmenopausal women with biopsy-proven NAFLD, including 12 with borderline or confirmed NASH, and 19 age-matched controls, serum estradiol levels were comparable between the NAFLD and control groups and between the NAFLD and NASH groups (64). Further, comparison of NAFLD subgroups according to steatosis grade, fibrosis stage, ballooning, lobular and portal inflammation showed that there were no statistically significant differences in serum estradiol levels, although notably, these comparisons were underpowered. Despite these findings, logistic regression analysis indicated that serum estradiol levels were associated with the presence of NASH (although not NAFLD), independent of age, BMI, and waist circumference. In 2835 women from the Multi-Ethnic Study of Atherosclerosis cohort, 160 of whom had evidence of NAFLD, serum estradiol levels were actually lower in women without NAFLD compared to those with NAFLD (0.07 mmol/L [0.05–0.16] vs 0.10 mmol/L [0.07–0.22]; P < 0.001) (65). In contrast with the other studies, a significantly greater risk of fatty liver prevalence was found in the highest versus the lowest tertile of estradiol (OR = 2.42; [95% CI: 1.37–4.29]), following adjustment for age, race/ethnicity, BMI, hypertension, total and HDL cholesterol, smoking, insulin sensitivity, and hormone replacement therapy use.

Despite declining estradiol levels following menopause and increased prevalence of NAFLD in postmenopausal women compared to premenopausal women, results from the above studies do not indicate that lower estradiol levels are associated with NAFLD in postmenopausal women. It is tempting to speculate that estrogen deficiency by itself is not sufficient to cause NAFLD, but that effects from other factors, for example, genetic predisposition, metabolic risk factors, or dietary composition, may be amplified within the context of estrogen deficiency. This idea is supported by results from the animal models discussed above, in which the combination of dietary manipulation and estrogen depletion was necessary to drive hepatic steatosis and hepatocellular injury.

HRT and protection from NAFLD

A better of understanding of the role of estrogens has been informed from a number of studies that assessed the use of HRT and the presence of NAFLD in postmenopausal women. A double-blind, randomized, placebo-controlled study comprised of 58 postmenopausal women with T2D was conducted to determine whether HRT (1 mg estradiol/0.5 norethisterone [an androgenic progesterone]) could improve levels of liver enzymes. Of the 45 individuals who completed the study, those taking HRT for ≥6 months had significantly lower serum concentrations of alanine aminotransferase (ALT), aspartate aminotransferase (ASP), gamma-glutamyl transferase (GGT), and alkaline phosphatase compared to those receiving placebo (66). In postmenopausal women with NAFLD, those taking HRT for ≥6 months (N = 14) had significantly lower levels of high waist circumference, GGT, ferritin, and insulin resistance compared with women not taking HRT (N = 79) (67). The type of hormone, dose, and route of administration were not specified in this study. A larger cross-sectional study of 251 postmenopausal women, NAFLD prevalence was lower in women taking HRT (14/53; 26.4%) compared with women not taking HRT (79/198; 39.9%). In this study, HRT was based on oral or transdermal use of estrogen in hysterectomized postmenopausal women, a combination of estrogen and progesterone with intact uterus, or tibolone for more than 6 months. While the frequency of NAFLD was lower in postmenopausal women using HRT, the use of HRT and variables related to HRT, including type of HRT, duration of use, and route of administration, were not significantly associated with NAFLD. Levels of GGT, ALT, ferritin, and insulin resistance were higher in NAFLD patients not taking HRT, and women not taking HRT (with or without NAFLD) had higher rates of overweight, obesity, and insulin resistance compared with women taking HRT (68). In a 1-year follow-up of Japanese women with no evidence of NAFLD at baseline, 67 developed NAFLD: 3.5% (28/802) premenopausal women, 7.5% (4/53) menopausal women, 6.1% (24/392) postmenopausal women, and 5.3% (11/206) women who reported taking HRT (type, route of administration, and duration of use not specified) (32). Collectively, these findings suggest that HRT confers partial protection against the development of NAFLD following menopause and attenuates metabolic disturbances associated with NAFLD. The biological pathways and molecular mechanisms by which estrogens exert beneficial effects on metabolic parameters (eg, fat distribution patterns) and regulate hepatic lipid metabolism have been comprehensively reviewed elsewhere (69) and will not be discussed here.

Tamoxifen-induced steatohepatitis

Tamoxifen, a well-tolerated and effective adjuvant therapy in breast cancer treatment (70, 71), is a nonsteroidal triphenylethylene that binds to estrogen receptors and exerts estrogenic and antiestrogenic effects depending on the cellular target (72). In the mid-1990s, a number of case studies reported observations of steatohepatitis (73–75), cirrhosis with steatohepatitis (76), and fatty liver (77) in breast cancer patients being treated with tamoxifen. Lower liver density (36.8 ± 2.9 vs 52.8 ± 1.2 HU; P < 0.001) and higher visceral fat area (135 ± 1.0 vs 81 ± 0.7 cm²; P < 0.001) were observed in tamoxifen users versus controls (78). In a subset of this group (14 tamoxifen users and 10 controls matched for age and BMI), liver density was inversely correlated with BMI and visceral adipose tissue (VAT), but no relationship was observed between the duration of tamoxifen use and liver density or VAT. A large-scale meta-analysis found that high BMI and hypercholesterolemia were the sole risk factors for the development of tamoxifen-related NAFLD (79), consistent with findings from a large randomized controlled trial in which tamoxifen was associated with the development of NASH only in overweight and obese women (80). While tamoxifen might cause increased visceral adiposity and hypertriglyceridemia due to its antiestrogenic effects, no difference in levels of dyslipidemia were found between women with or without NAFLD (79). Instead, women with central VAT distribution, hypertriglyceridemia, and higher levels of sex hormones are at greater risk of developing breast cancer, and therefore, are more likely to receive tamoxifen (78). Interestingly, bezafibrate, a peroxisomal proliferator-activator receptor alpha agonist, improved hepatic steatosis via activation of peroxisomal β-oxidation in 3 tamoxifen-treated breast cancer patients (81).

In some animal studies, tamoxifen has been shown to induce steatosis (82, 83), while in others, it improved both liver (63, 84–86) and lipid metabolism (87–90). Tamoxifen acts as an ERα-AF1 agonist and an ERα-AF2 antagonist, and probably exerts different effects depending on activation function. In HFD-fed OVX mice, tamoxifen significantly improved intrahepatic triglyceride and cholesterol ester content, prevented the development of hepatotoxicity, and reduced plasma ALT levels compared to vehicle-treated animals (84). Notably, other metabolic disturbances, such as weight gain, fat accumulation, glucose tolerance, and insulin resistance were improved by tamoxifen treatment. These effects were abolished in ERα ^-/- mice, indicating a requirement for functional ERα signaling. Likewise, tamoxifen-treated ERα-AF1^-/- mice were not protected from the metabolic consequences of HFD feeding, indicating a requirement for the activation of ERα-AF1. In a set of tamoxifen-regulated genes involved in hepatic lipid and glucose metabolism, all were found to be dependent on ERα-AF1 activation. A subsequent study from the same research laboratory generated hepatocyte-specific ERα-deleted mice (LERKO) and examined the effects of E2 and tamoxifen on metabolic parameters (85). While both E2 and tamoxifen protected HFD-fed OVX mice from hepatic steatosis, these effects were abrogated in LERKO mice. These results not only confirm the role of hepatocyte-specific ERα in mediating protective effects against diet-induced fatty liver, but they also show that both E2 and tamoxifen attenuate hepatocellular injury in response to HFD. Tamoxifen, but not E2, produced an increase of plasma growth differentiation factor 15 (Gdf15), a hepatokine that inhibits food intake, and protective effects of tamoxifen were abrogated in Gdf15^-/-mice. These results suggest that selective modulation of ERα may be an effective treatment for postmenopausal NAFLD.

Hyperandrogenism in women is also associated with hepatic steatosis and fibrosis, independent of insulin resistance and adiposity (91, 92). Increased circulating testosterone levels in middle-aged women were associated with higher VAT (93). Levels of calculated free testosterone, bioavailable testosterone, and free androgen index, but not serum total testosterone, were significantly higher in 22 postmenopausal women with NAFLD compared with 18 age-matched controls (64). Bioavailable testosterone and free androgen index were associated with NAFLD, following adjustment for age, BMI, and waist circumference (64). Higher tertiles of bioavailable testosterone were also associated with higher odds of fatty liver (OR = 1.73; [95% CI: 1.05–2.87]) (65).

Free testosterone levels in premenopausal women may be associated with prevalent NAFLD in mid-life (94). Utilizing data from 1052 women in the Coronary Artery Risk Development in Young Adults cohort, Sarkar et al (94) found increasing quintiles of Year 2 free testosterone presence were associated with 29% increased odds of NAFLD at Year 25 (P = 0.001). Multivariate analyses showed that increasing quintiles of free testosterone remained associated with NAFLD, independent of BMI, waist circumference, HOMA-IR, and serum lipids. The association between free testosterone levels and prevalent NAFLD remained significant in a subset of 955 women without androgen excess (AOR 1.27; [95% CI: 1.05–1.53]; P = 0.016). Visceral adipose tissue volume (mediating effect 41.0%, [95% CI: 22–119%]) and Year 25 triglyceride levels (mediating effect 12.1% [95% CI: 2.8–37.8%]) partially mediated the association of free testosterone with NAFLD; however, there was no significant evidence showing that HOMA-IR mediated the association of free testosterone and NAFLD.

In contrast, a cross-sectional analysis using 2406 women from the National Health and Nutrition Examination Surveys cohort found a lower percentage of women with suspected NAFLD, with increasing quartiles of testosterone: 45.2% in quartile 1 versus 33.4% in quartile 4 (95). When stratified according to reproductive status, this relationship was only observed in postmenopausal women: 52.2% in quartile 1, 45.4% in quartile 2, 43.8% in quartile 3, and 37.2% in quartile 4 (P = 0.043). Multivariate analyses showed that lower levels of serum testosterone remained significantly associated with higher odds of suspected NAFLD (based on ALT values) in models adjusting for a wide range of covariates. The lowest quartile of serum testosterone was also found to be associated with significantly higher odds of suspected NAFLD: lowest quartile versus highest quartile ORs: 2.15 (1.34–3.45), adjusting for age and BMI.

The interaction between dietary factors and estrogens in the development of NAFLD

Results from animal studies indicate that estrogens exert beneficial effects only with HFD diet, but not normal chow (53–56, 61). In HFD-fed mice, E2 supplementation reduced diet-induced insulin resistance and improved insulin signaling in skeletal muscle (61). Likewise, the histological similarities between OVX and SO animals fed normal chow (53) suggest that estrogen deficiency alone is not sufficient to cause hepatocellular injury. High fructose (Fr) in conjunction with HFD exacerbates liver fibrosis in the absence of endogenous estrogens. Both OVX-HFD and SO-HFD-Fr mice developed hepatocyte ballooning, which was significantly amplified and accompanied by infiltration of inflammatory cells in OVX-HFD-Fr mice (96). OVX-HFD and SO-HFD-Fr mice developed a moderate amount of pericellular and periportal fibrosis compared to SO-HFD animals, in whom only minimal pericellular fibrosis appeared. In contrast, in OVX-HFD-Fr mice, pericellular fibrosis was significantly more pronounced and diffused throughout the liver. In addition, the OVX-HFD-Fr mice showed an accumulation of activated hepatic stellate cells, the key fibrogenic cells of the liver, and increased expression of fibrogenic genes. OVX-HFD-Fr mice also had substantially higher ALT levels than any of the other treatment groups. E2 supplementation in this group significantly reduced body weight, serum ALT, and liver triglycerides compared with OVX-HFD-Fr mice. These results suggested that the combination of high dietary fructose intake and estrogen depletion synergistically contribute to liver damage through independent mechanisms involving oxidative stress-induced hepatocyte death and proinflammatory activation, respectively. While similar studies in humans are forthcoming, there is substantial evidence indicating that a high intake of dietary fat and fructose can effectively create a steatogenic and/or fibrogenic environment (97).

Choline, an essential nutrient with important roles in metabolism, has also been associated with NAFLD. Women, in general, were reported to have lower reported choline intake than men (98). In this study, reported choline intake was similar between premenopausal and postmenopausal women, but postmenopausal women with low choline intake had significantly worse fibrosis compared to men or premenopausal women. The disparity in histological liver damage resulting from choline deficiency between pre- and postmenopausal women may be attributable to the presence of estrogen response elements in the promoter of the phosphatidylethanolamine N-methyltransferase gene, which is involved in choline biosynthesis (99, 100). Thus, in postmenopausal women in whom endogenous productions of estrogens is diminished, estrogen-mediated regulation of choline synthesis is abrogated, leading to an increased susceptibility to the development of NAFLD due to chronic states of choline deficiency.

Conclusions

Women are at especially high risk for developing the disease following menopause. It is likely that that the loss of protection conferred by estrogens combined with subclinical disturbances in metabolic parameters prior to menopause (eg, moderate adiposity, mild dyslipidemia, prediabetes) underlie the increased NAFLD risk in postmenopausal women. Results from animal studies indicate that high dietary fat or fructose intake can synergistically enhance the effects of estrogen deficiency, leading to exaggerated effects on hepatocellular injury. Whether this occurs in humans is not yet known; however, it is tempting to speculate that women who have a metabolic phenotype predisposing to NAFLD prior to menopause may develop the disease once the protection conferred by estrogens is lost (Fig. 1). Indeed, normal weight postmenopausal women with lipid and glucose and insulin levels within normal range are at a low risk of developing NAFLD. Therefore, efforts to emphasize healthy diet and regular physical activity should be especially emphasized in middle-aged women as they approach menopause to prevent the development of NAFLD. Likewise, female NAFLD patients should be recognized as a high-risk group for the development of NASH following menopause, and careful disease monitoring should be performed in these women. In addition, recognition of a potential sex bias in the diagnosis of NAFLD is expected to improve detection of the disease in postmenopausal women.

Figure 1.

The relationship between NAFLD risk pre- and post-menopause. Women who have a metabolic phenotype predisposing to NAFLD prior to menopause may develop the disease once the protection conferred by estrogens is lost. Abbreviations: NAFLD, nonalcoholic fatty liver disease; PNPLA3, patatin-like phospholipase domain-containing protein 3.

Postmenopausal women may potentially benefit from a number of treatment options, including HRT or antiandrogens. However, the effects of different hormone combinations, HRT dose, and route of administration, time of HRT initiation, and duration of HRT treatment on NAFLD in postmenopausal women currently represent a critical gap in clinical research. Clinical studies with focused outcomes are necessary to determine if postmenopausal hormonal manipulation or other treatments can prevent or treat NAFLD in these women. Sex-specific strategies to prevent, diagnose, and treat NAFLD in postmenopausal women are critical to improve NAFLD-related morbidity and mortality in this vulnerable population.

Additional Information

Disclosure Summary: The author has nothing to disclose.

Data Availability: Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.