Sex disparity in hepatocellular carcinoma owing to NAFLD and non-NAFLD etiology: epidemiological findings and pathobiological mechanisms

Nonalcoholic fatty liver disease (NAFLD) exhibits sexual dimorphism, with men being more exposed than women to the risk of simple steatosis, nonalcoholic steatohepatitis fibrosis, and hepatocellular carcinoma (HCC), while the protection conferred to women seemingly disappears with aging and reproductive senescence (i.e., menopause). HCC, the most common primary liver cancer, which carries an ominous prognosis, may result from various genetic and non-genetic risk factors. NAFLD is now projected to become the most common cause of HCC. HCC also exhibits a definite sexual dimorphism in as much as it has a worldwide high male-to-female ratio. In this review article, we focus on sex differences in the epidemiological features of HCC. Moreover, we discuss sex differences in the clinical outcome and molecular pathobiology of NAFLD-HCC. By highlighting the research gaps to be filled, the aim of this review is to prompt future research of sex differences in HCC and facilitate developing personalized cancer prevention strategies, detection, and treatments to achieve better patient outcomes in NAFLD-HCC, considering sex differences in HCC pathobiology. Received: First Decision: Revised: Accepted: Published: Science Editor: Copy Editor: Production Editor: Jing Yu Page 2 of 12 Lonardo et al. Hepatoma Res 2020;6:83 I http://dx.doi.org/10.20517/2394-5079.2020.89


METHODS
The PubMed database was extensively searched for articles published as of the 31st of July 2020. The keywords used in our search include, but are not limited to: HCC, liver cancer, sex differences, gender differences, epidemiology, natural course, pathogenesis, risk factors, immune response, genetics, and sex hormones. Additional terms were used to search for articles reporting sex differences and/or the effect of sex hormones in specific mechanisms pertaining to carcinogenesis. Among the retrieved publications, only those that were deemed to be relevant based on consensus among the authors were retained.

EPIDEMIOLOGICAL MODIFIERS OF HCC RISK
Irrespective of its aetiology, HCC affects men more commonly than women owing to complex and multifactorial reasons. This section reviews risk factors of HCC in general and discusses interactions between sex and risk factors.

Geographic area and ethnicity
Eastern Asia, Southeast Asia, and sub-Saharan Africa exhibit a high incidence and prevalence of HCC; Mongolia, China, Japan, Papua New Guinea, and Egypt are top-ranked countries [1] . By contrast, countries with a low incidence and prevalence include India, Russia, northern countries of South America, Argentina, European countries (except for southern countries), USA, and Australia with the rest of the world exhibiting intermediate rates of incidence and prevalence [1] .
Multi-ethnic populations display a clear ethnic gradient. For example, in the United States, Asians/Pacific Islanders have been reported to have the highest incidence rate per 100,000 (11.7), followed by Hispanics (9.5), Blacks (7.5), while Whites had the lowest (4.2) [13] .

Sex
With few exceptions, the male to female (M:F) ratio of the incidence of HCC ranges between 2 to 3 in the most of the countries, irrespective of whether they are high-rate areas or not, and are maximal in middle European countries (M:F ratio up to 5) [13,14] . In contrast, in Costa Rica, Colombia, Ecuador and Uganda, the M:F ratio of the incidence of HCC is smaller, ranging from 1.3 to 1.6 [13,14] .
The biological grounds underlying this sex disparity in the prevalence of HCC are incompletely defined and probably related to multiple behavioural, hormone-metabolic risk factors, and cancer biology. Sex differences in HCC pathogenesis are discussed below under sex disparity in HCC pathobiology.
The difference in the M:F ratio of the incidence of HCC among different countries is intriguing, suggesting potential race/ethnicity-sex interplay in HCC. At this point, sufficient data do not exist to delineate whether the difference is explained by a biological interplay and/or an interplay of gender attributes and culture/ ethnicity.

Age
The overall incidence of HCC consistently peaks at 70 years in various countries worldwide, such as France, Italy, Japan, and USA (whites) and this is approximately 5-15 years before the peak occurrence of cholangiocarcinoma, the second most common PLC after HCC [15] . However, other authors report that the mean ages of diagnosis with HCC are 55-59 years in China and 63-65 years in Europe and North America [14] . In Qidong, China, where the HCC burden is among the world's highest, the age-specific incidence rates increase up to the age of 45 among men and then plateau; while increasing to the age of 60 and then plateauing among women [14] . A surveillance, epidemiology, and end results (SEER) Analysis (from 1988 to 2010) including 39,345 patients with HCC (Men 76%, women 34%) showed that men are diagnosed 4-7 years earlier than women across the race/ethnic groups [16] . These findings suggest that sex and age interact in the occurrence of HCC, implying that consideration of this interaction (as opposed to treating age and sex as independent variables) will be essential in future research.

Severity of liver histology
While cirrhosis is an almost essential pre-requisite for the development of HCC in those with HCV infection, infection with HBV exerts a more direct carcinogenic effect on the liver [1] . Similar to HBV infection and to alcoholic liver disease, NAFLD-HCC may occur in non-cirrhotic livers [14,15] . A Japanese descriptive study reported that men with nonalcoholic steatohepatitis (NASH) developed HCC at earlier liver fibrosis stages than women [17] . The study was too small to confirm the sex difference but provides an intriguing hypothesis pertaining to carcinogenesis. Further larger studies are warranted to investigate this.

Viral hepatitis
With the exceptions of Japan and Egypt (where HCV infection is the chief risk factor of HCC), in most high-risk countries, chronic HBV infection and aflatoxin B1 are the major risk factors for the development of HCC, whereas HCV infection, excessive alcohol consumption, and common metabolic disorders (diabetes, obesity and metabolic syndrome) prevail in low-rate areas [15] . Chronic drinking of alcohol > 80 g/day for over 10 years increases the risk of HCC by a factor of 5; and alcohol consumption enhances the risk of HCC in those with either chronic hepatitis C or NAFLD [10,18] . However, given that these data often combine both sexes, research needs to be conducted urgently to clarify the sex-specific thresholds of alcohol consumption that are associated with a raised HCC risk.

Family history
HCC commonly exhibits familial clustering, and family history of disease is a risk factor for the development of HCC. Interestingly, family history of HCC was identified as a favourable prognostic factor for improved survival particularly in those individuals whose tumours can be radically cured, even in the stage-stratified analysis [19] . In the study, female sex and younger age, non-diabetics, and lifetime nondrinkers were more common among individuals with first-degree family histories of HCC than among those without such histories [19] . The exact mechanisms underlying the above associations remain uncertain.

Genetic risk determinants of NAFLD and inherited metabolic liver diseases
Genetic variants associated with an increased risk of NAFLD, advanced NAFLD, and NAFLD-HCC appear to contribute to the risk of HCC in the general population. A recent study conducted using Danish and UK databases demonstrated that a genetic risk score using three genetic variants [i.e., patatin-like phospholipase domain-containing protein 3 (PNPLA3) p.I148M, transmembrane 6, superfamily member 2 (TM6SF2) p.E167K, and hydroxysteroid 17-beta dehydrogenase 13 (HSD17B13) rs72613567], is associated with an up to 12-fold higher risk of cirrhosis and up to a 29-fold higher risk of HCC in individuals from the general population from these countries [20] .
Certain inheritable metabolic disorders such as hemochromatosis, α-1 antitrypsin deficiency, tyrosinemia, glycogen storage diseases and several porphyrias also increase HCC risk, although they account for a negligible HCC risk globally [21,22] .

Other risk factors of HCC
Smoking and co-infection with HIV also contribute to the development of HCC [1] . Certain environmental factors or occupational factors, such as vinyl chloride, polycyclic aromatic hydrocarbons, aflatoxins, and aristolochic acid, a common ingredient of traditional herbal medicine, have been suggested to play a role in the development of HCC [23,24] . How these factors and underlying mechanisms intersect with sex and sex hormones in the development of HCC has not been fully elucidated.

Interaction between sex/gender and metabolic risk factors
Sexual dimorphisms in metabolism are well-known (recently reviewed elsewhere [9,10,12] ) and likely account for sex differences in HCC risk. A few other risk factors have been suggested for sex/gender-interaction, which are also discussed in this section.
Obesity has been associated with a higher risk of HCC incidence in men than women, especially in non-Asians [25] . A recent study conducted in an Asian population found a different relationship between BMI and HCC risk according to sex, following a U-shaped and a linear curve in men and women, respectively [26] . Studies reported a stronger risk association between pre-diabetes/diabetes and HCC in men than women [26][27][28] .
NAFLD has a definite sexual dimorphism; men are more prone than women to the risk of uncomplicated steatosis, NASH fibrosis, and HCC. However, aging and menopause are associated with the disappearance of protection in women [12,29,30] .
Prospective studies indicate that regular alcohol intake, although within safe thresholds, is a risk factor for the progression to HCC among individuals with NAFLD [31] . Moreover, among those with HCC, alcohol use is more frequent in men than in women [26,32] .
Additionally, men are more prone to acquire HBV and HCV infection, develop chronic hepatitis, cirrhosis and HCC than women [14] .

CLINICAL SEX DIFFERENCES IN HCC
Recent large epidemiological studies conducted in the USA have shown that, compared to men, women with HCC presented with older age, higher frequency of NAFLD, non-cirrhotic HCC, less-advanced tumour stage (by size, local/vascular invasion, metastasis) and lower frequency of alcoholic liver disease [32][33][34] .
Studies regarding sex differences in survival rates have yielded conflicting results so far. Two recent large multi-centre US studies enrolling 5,327 patients with HCC (22.6% women) and 1,110 (23.5% women), respectively, reported higher overall survival rates among women after adjusting for confounding factors [33,34] . Consistently, one of these found that female sex was independently associated with early tumour detection [odds ratio (OR) 1.46] and response to first HCC treatment (OR 1.72) [33] . Conversely, other US studies found no sex-related difference in HCC prognosis [32,35] , while Asian studies did [36][37][38][39] . Another US study showed age, sex, and ethnicity intersection in survival rates; women had higher survival rates from HCC than men before age 55, while after 65 years or among Hispanics, there was no such a survival difference between sexes [38] . Another US study also suggested a similar interplay of age and sex in HCC survival and further possible age-, sex-, and race/ethnicity-interaction in HCC survival [16] . Table 1 provides a synthetic overview of sex differences in risk factors, presentation, and outcome of HCC owing to NAFLD and non-NAFLD aetiologies. Future studies with proper consideration of these interactions are warranted to reconcile some inconsistency in the literature.

SEX DISPARITY IN HCC PATHOBIOLOGY
Chronic persistent injury induces wound-healing responses through the release of pro-inflammatory cytokines (e.g., IL-6, TNF-alpha) and increased oxidative stress in the liver [40] . The wound-healing process, together with persistent liver damage, promotes fibrogenesis and tissue DNA damages and facilitates hepatic carcinogenesis [41] . The liver tissue concentration of 8-OHdG, a marker of oxidative DNA damage, has been associated with epigenetic inactivation of tumour suppressor genes (i.e., methylated tumour suppressor genes) [42,43] . An increased number of methylated tumour suppressor genes was, in its turn, associated with a shorter time-to-HCC development in patients with chronic hepatitis C [44] , demonstrating a mechanistic link of oxidative DNA damage, epigenetic alteration of tumour suppressor genes, and development of HCC. In NAFLD patients, chronic metabolic stress to hepatocytes aggravates oxidative stress, induces cellular protein/DNA damage, and promotes premature senescence of hepatocytes, contributing to an increased risk of HCC among obese individuals [12,45] , even in the absence of cirrhosis.
Sex differences are well documented in cancer mechanisms [46] . Several well studied mechanisms accounting for sex differences are summarized in Table 2 [47][48][49][50][51][52][53][54][55][56] . Compared to females, males are more susceptible to oxidative stress due to a higher NADPH oxidase activity, a lower NFR-2, and lower anti-oxidants [51,52,57] , and have a higher induction of IL-6 by hepatic Kupffer cells under liver injury [54] . In contrast, higher physiological oestrogens protect females from HCC development via the anti-oxidative effects of estrogenic [51,57] , anti-fibrotic effects [12] , and inhibitory effects on IL-6 production by hepatic Kupffer cells [54] . The protective effects of oestrogen are lost after menopause, which may explain the fact that the male predominance observed in HCC incidence decreases with advancing age [58] . Gut microbiota also exhibits sex differences [56,[59][60][61] . In an experimental mouse model, higher hepatic hydrophobic bile acids were observed in males, which was causally associated with a decreased expression of tumour-suppressive microRNA in the liver and increased incidence of HCC [61] . Importantly, similar sex differences in bile acid profile exist in humans [56] , suggesting that sex-differences in gut microbiota and bile acid profile may contribute to male dominance in the development of HCC in man.
Despite the fact that evidence supports oestrogen exerting protective effects on the development of HCC, whether oestrogens have protective effects on the progression of HCC and patients' survival remains  [62] and the better overall survival rates of women compared to men, which, however, disappears in advanced age [16] . However, possible beneficial effects of oestrogen on HCC survival have not been tested in experiments, and the mechanisms, if any, through which oestrogens affect HCC survival, remain uncertain.
HCC tumour tissue expresses oestrogen receptors, although the clinical and pathological significance of these remain controversial [63,64] . The positive expression rates of oestrogen receptors among HCC cases also  [47] Total body fat and abdominal adipose tissue were evaluated in 89 men and 75 women using CT Visceral adiposity After adjusting for total body fat mass, men had significantly higher values of visceral adipose tissue volume and areas, measured by CT, than women. An increase in total fat mass was associated with a significantly greater increase in visceral adipose tissue volume in men than in women Higher FFA release in males induces inflammation, insulin resistance, and lipotoxicity and fosters a tumour-promoting environment in the liver and may contribute to an increased recurrence of HCC Laughlin et al. [48] A cross-sectional study to measure serum leptin, adiponectin and sex hormone levels in 1510 community dwelling men and postmenopausal women aged 50-92 years Visceral adipokines Serum adiponectin and leptin levels were higher in women than in men. In both sexes, adiponectin concentrations were lower, and leptin levels higher, with increasing BMI and waist girth A higher adiponectine level may protect women from developing HCC via the activation of AMPK and p38α [49] Lönnqvist et al. [50] BMI and age matched obese subjects (22 male and 23 female) undergoing elective surgery were evaluated for visceral fat lypolysis

Visceral fat lipolysis
Catecholamine-induced rate of FFA mobilization from visceral fat to the portal venous system is higher in obese men than in obese women, probably due to a larger fat-cell volume but also to a decrease in the function of α 2adrenoceptors, an increase in the function of β 3 -adrenoceptors, and an increased ability of cyclic AMP to activate hormone-sensitive lipase Higher FFA release in males induces inflammation, insulin resistance, and lipotoxicity and fosters a tumour-promoting environment in the liver and may contribute to an increase recurrence of HCC

Oxidative stress/Senescence
Augustine et al. [51] Compared Nqo1 mRNA and protein expression and activity in males and females before and after applying known inducers using SD and August Copenhagen x Irish (ACI) rat strains NAD(P)H: quinone oxidoreducatase 1 (Nqo1) ACI rats showed minimal differences in Nqo1. In SD rats, Nqo1 mRNA, protein, and activity levels were significantly higher in females than in males. Female SD rats showed greater induction than male Higher Nqo1 may lead to greater protection against oxidative stress and thus decreased susceptibility to carcinogens Kratschmar et al. [52] The interaction among corticosteroid, 11b-HSD1, and NFR-2 was evaluated using transfected HEK-293 cells and hepatic H4IIE cells. The hepatic expression levels of 11b-HSD1 and NFR-2 target genes were also compared between male and female Han Wistar rats significantly varies in the literature, probably due to the differences in the methodologies and populations of the studies (e.g., ethnicity, sex), as well as stage and aetiologies of disease [64] . HCC tumour tissues express both oestrogen receptor alpha (ERα) and beta (ERβ), and also a variant form of ERα (vERα), which lacks exon 5 in the hormone-binding domain [65] . Compared to patients with ER-negative HCC, patients with ER-positive HCC have a shorter survival rate after curative resection [66] . Several randomized controlled trials were conducted to ascertain whether blockage of oestrogen signalling in HCC by the anti-oestrogen tamoxifen would improve the survival of patients with HCC. However, the results were consistently negative [67,68] . The presence of the liver vER receptor in the tumour is a strong negative predictor of survival in inoperable HCC patients and is a marker of clinical aggressiveness compared to wild-type ERαn [69,70] . HCC positive for the vER receptor is unresponsive to tamoxifen but responds to megestrol [71] .
Sex differences in HCC biology have extensively been explored in recent years in functional signatures of differentially expressed genes [71,72] , expression quantitative trait loci (eQTL) [73] , and cancer-driver genes [73] . These studies strongly suggest that HCC in males and females are biologically distinct and may respond differently to treatments. This is in agreement with epidemiological data summarized in Table 1. However, no randomized controlled trials have demonstrated sex differences in HCC treatment response or clinical outcomes. The consideration of sex and women's reproductive history in clinical studies designs and Naugler et al. [54] In mice administered with DEN, HCC incidence, and its relationship with hepatic IL-6 induction, Yasuda et al. [55] Using the DEN model, hepatic fibrosis was compared between male and female rats

Stellate cell activation/ fibrogenesis
In male rats the induction of fibrotic response was significantly stronger than in female rats. Estradiol reduced hepatic fibrogenesis in male rats while concomitant administration of a neutralizing antibody against rat estradiol enhanced fibrogenesis. Oophorectomy in the female rats had a fibrogenic effect Higher oestrogen protects premenopausal women from advanced hepatic fibrosis, a major risk factor of HCC

Tumor suppressor genes
Xie et al. [56] The mechanistic link between microbiota and hepatocellular carcinogenesis using a STZ-HFD induced NASH-HCC murine model and compared results for both sexes Bile acid/ microbiota STZ-HFD feeding induced a higher incidence of HCC in male mice, which was associated with increased intrahepatic retention of hydrophobic BAs and decreased hepatic expression of tumorsuppressive microRNAs. Metagenomic analysis showed differences in gut microbiota involved in BA metabolism between male and female mice. Treating STZ-HFD male mice with 2% cholestyramine led to significant improvement of hepatic BA retention, tumor-suppressive microRNA expressions, microbial gut communities, and prevention of HCC Sex differences in microbiota lead to higher intrahepatic retention of hydrophobic BAs, decreased tumor suppressor microRNA in the liver, and an increased incidence of HCC in male mice ALS: alkali labile sites; AMPK: AMP-activated protein kinase; BA: bile acids; CT: computed tomography; DEN: diethylnitrosamine; FFA: free fatty acids; FPG: DNA repair glycosylase; IL-6: Interleukin-6; NASH-HCC: nonalcoholic steatohepatitis-hepatocellular carcinoma; SSB: single strand breaks; SD: sprague dawley; STZ-HFD: streptozotocin-high fat diet analyses is warranted in future trials to better understand whether and how these factors may modify treatment response to specific therapeutic targets and influence clinical outcomes.

CONCLUSIONS AND RESEARCH AGENDA
A robust line of research has shown multifaceted sexual dimorphisms in the NAFLD domain [10][11][12] . It is important to note that the observed sex differences in NAFLD are not linear throughout the course of the disease but rather mechanism-specific. Further studies will eventually contribute to more effectively reducing the NAFLD-HCC incidence by delineating sex differences in individual pathways and therefore allowing the development of a personalized approach in preventing NAFLD progression. Similarly, sex differences in HCC epidemiology have not been fully characterized with proper consideration of women's reproductive status/history. Therefore, we recommend that sex/gender and reproductive history should be considered in future clinical and epidemiological HCC studies. Further mechanistic understanding, together with the epidemiological characterization of sex differences and the impact of reproductive history will predictably help clinicians by allowing more accurate risk stratification and personalized therapeutic approaches in the future. A scoring system combining genetic and non-genetic HCC risk factors while considering biological disparities by sex and reproductive status may improve our future care, although sufficient data to develop such a scoring system is pending future research.