SGLT-2 Inhibitors in NAFLD: Expanding Their Role beyond Diabetes and Cardioprotection

Non-alcoholic fatty liver disease (NAFLD) is an ‘umbrella’ term, comprising a spectrum ranging from benign, liver steatosis to non-alcoholic steatohepatitis, liver fibrosis and eventually cirrhosis and hepatocellular carcinoma. NAFLD has evolved as a major health problem in recent years. Discovering ways to prevent or delay the progression of NAFLD has become a global focus. Lifestyle modifications remain the cornerstone of NAFLD treatment, even though various pharmaceutical interventions are currently under clinical trial. Among them, sodium-glucose co-transporter type-2 inhibitors (SGLT-2i) are emerging as promising agents. Processes regulated by SGLT-2i, such as endoplasmic reticulum (ER) and oxidative stress, low-grade inflammation, autophagy and apoptosis are all implicated in NAFLD pathogenesis. In this review, we summarize the current understanding of the NAFLD pathophysiology, and specifically focus on the potential impact of SGLT-2i in NAFLD development and progression, providing current evidence from in vitro, animal and human studies. Given this evidence, further mechanistic studies would advance our understanding of the exact mechanisms underlying the pathogenesis of NAFLD and the potential beneficial actions of SGLT-2i in the context of NAFLD treatment.


Introduction
Non-alcoholic fatty liver disease (NAFLD) has become a major health problem worldwide with an increasing prevalence ranging from 13% in Africa to 42% in South-East Asia [1,2]. The term NAFLD includes a variety of diseases, ranging from liver fat deposition in more than 5% of hepatocytes (steatosis-non-alcoholic fatty liver (NAFL)) to necroinflammation and fibrosis (non-alcoholic steatohepatitis (NASH)), which can progress into NASH-cirrhosis, and eventually to hepatocellular carcinoma [3,4]. Determining ways to delay the progression of NAFLD has become a global focus.
Lifestyle modifications-namely improved diet, weight management and increased physical activity-play a fundamental role in the treatment in NAFLD, since more than half of the patients with NAFLD have a high body mass index (BMI) [5,6]. Of note, drugs for specifically treating NAFLD are now being developed/approved, thus management of co-morbidities such as obesity, dyslipidemia or type 2 diabetes mellitus (T2DM) remains the cornerstone for the treatment of NAFLD patients [7,8].
Interestingly, since the cardiometabolic disorders associated with NAFLD account for the increased morbidity and mortality in these patients, the term metabolic (dysfunction) associated fatty liver disease (MAFLD) has been recently proposed by a panel of international experts as more appropriate, reflecting the underlying pathogenesis of the NAFLD spectrum [9][10][11]. Masarone et al. revealed that the prevalence of NAFLD was 94.82% among patients with metabolic syndromes (MS) and was presented in all T2DM patients with elevated transaminases when they performed biopsy [12]. Surprisingly, 58.52% of MS and 96.82% of T2DM patients in this study were diagnosed with NASH. As insulin resistance (IR) is crucially linked with both T2DM and NASH pathophysiology [13], they concluded that NASH may be one of the early clinical manifestations of T2DM [12]. Of note, NAFLD, particularly steatohepatitis, has also been associated with an increased risk of cardiovascular-related mortality regardless of age, sex, smoking habit, cholesterolemia and the remaining elements of MS [11]. Several prospective, observational and cross-sectional studies and meta-analyses demonstrated that NAFLD is associated with enhanced preclinical atherosclerotic damage as well as coronary, cerebral and peripheral vascular events with a negative impact on patients' outcome [14]. Furthermore, the severity of liver biopsy-based fibrosis has been independently associated with the worsening of both systolic and diastolic cardiac dysfunction [15,16].
Sodium-glucose co-transporter type-2 (SGLT-2) inhibitors (SGLT-2i) are glucose-lowering agents that improve glucose control, whilst promoting weight loss and lowering serum uric acid levels. These agents have shown great advantages even in patients with no diabetes, gaining approval of use in non-diabetic patients with heart failure and chronic kidney diseases [17,18]. In addition to the anti-hyperglycemic effects and their ability to reduce body weight, SGLT-2i seem to exert potent antioxidant and anti-inflammatory effects making them promising candidates for NAFLD treatment.
Indeed, recent data from animal studies and clinical trials have demonstrated beneficial effects of SGLT-2i on fatty liver accumulation, as judged by improvement of biological markers of NAFLD, as well as by imaging techniques, albeit mainly in T2DM patients.
Herein, we provide current insights into the effects of SGLT-2i on the progression of NAFLD, focusing on the underlying mechanisms of action. Endoplasmic reticulum (ER) stress, oxidative stress, low-grade inflammation, autophagy and apoptosis are among the SGLT-2i-regulated processes that have been shown to mediate the beneficial effects of SGLT-2i on NAFLD. Accordingly, we present the available evidence from in vitro, animal and human studies regarding the potential impact of SGLT-2i on NAFLD development and progression.

Overview of NAFLD Pathogenesis
The main theory concerning the pathophysiology of NAFLD has changed over time, reflecting the advances in our understanding of this multi-factorial disease. For years, the "two-hit" theory was the prevailing one. According to this theory, the pathophysiology of NAFLD consisted of a first "hit" representing the stage of simple steatosis alone (NAFL), which involves hepatocytic lipid accumulation and hepatic insulin resistance [19], as well as of a second "hit" from other factors (e.g., oxidative stress, ER stress, other injury), which was required for the development and progression of hepatic inflammation (NASH) and fibrosis. Replacing this initial theory, the "multiple parallel-hit" model [20] has been more recently proposed to better explain the complex pathogenesis and progression of NAFLD. According to this theory, as its name suggests, different amalgamations of numerous (epi)genetic and environmental factors, representing "hits", dynamically interplay with each other, and can drive the development and progression of NAFLD. These factors include specific genetic polymorphisms and epigenetic modifications [21], features of metabolic syndrome [22][23][24][25] (western diet, lack of physical activity, central obesity, dysregulation of adipoknes and IR), lipotoxicity [26,27], dysbiosis of the gut microbiota [26], dysregulation of autophagy and mitochondrial function [28][29][30], ER stress [31], hepatocyte homeostasis and death [32,33], as well as inflammatory and fibrotic responses [34,35]. Notably, when the hepatic capacity to handle the primary metabolic energy substrates is overwhelmed, toxic lipid species accumulate in the liver, leading to hepatocyte dysfunction and apoptosis, along with metabolically triggered inflammation and subsequent fibrosis [36].
The hallmark of NAFLD pathogenesis seems to be an increased adipocyte-like (dys)function of the hepatocytes, when the capacity of adipose tissue to store excess energy from the diet is diminished [37][38][39][40]. In conditions of energy surplus, increased metabolic substrates in tissues [27] drive hepatic de novo lipogenesis [41][42][43], while enhancing IR leading to a vicious cycle. The accumulation of intrahepatic lipid levels are governed by the balance between lipids synthesis, uptake and lysis [44]. Loss of equilibrium between lipogenesis and lipolysis leads to intracellular accumulation of free fatty acids (FFA) and subsequent hepatocellular damage, IR, worsening of liver function, formation of hepatic steatosis and progression to NASH, cirrhosis, and hepatocellular carcinoma [44].
In most individuals with NAFLD, dysregulation of adipokines (e.g., leptin and adiponectin) and metabolic-induced inflammation impair insulin signaling in adipocytes [45,46]. This impairment, in turn, contributes to reduced FA uptake and accelerated lipolysis in subcutaneous adipose tissue, resulting in exces sive delivery to the liver [45][46][47]. When hepatocytes are overloaded with lipids, physiologically minor pathways of β-oxidation in the peroxisomes and the ER are upregulated, thus increasing the hepatocyte reactive oxygen species (ROS) production [41,48] and generating highly reactive aldehyde by-products [49]. This phenomenon leads to nuclear and mitochondrial DNA damage, phospholipid membrane disruption and cell death. Moreover, due to the mitochondrial dysfunction and the consequent impaired β-oxidation found in patients with NAFLD, FA are alternatively esterified and collected in lipid droplets in the ER [50,51], generating toxic lipid metabolites, such as diacylglycerols [24], ceramides [52,53], and lysophosphatidyl choline species [53], which in turn lead to hepatocyte dysfunction (lipotoxicity) [54,55] and ER stress. The adaptive homeostatic mechanism of ER stress, called the unfolded protein response (UPR) [56], is impaired in NAFLD patients and, thus further cell stress-sensors are activated, whilst inflammatory and apoptotic pathways are upregulated [57,58]. In addition, the gut microbiota seem to also play a critical role in NAFL and NASH pathogenesis. Gut-derived pathogens and damage-associated molecular patterns activate an intrahepatic inflammatory process via Toll-like receptor signaling and NLR family pyrin domain containing-3 (NLRP3) inflammasome activation [35,59,60]. Hepatic innate immune cells, including Kupffer cells and dendritic cells, as well as hepatic stellate cells (HSCs) are then activated and the liver parenchyma is progressively infiltrated by recruited neutrophils, monocytes, T-lymphocytes, and macrophages [34,61]. Subsequently, the secreted cytokines and growth factors intensify the inflammatory process and further contribute to the fibrotic process as an ineffective attempt for tissue regeneration [35]. Overall, NAFLD is characterized by a complex pathogenetic mechanism, where various pathways are implicated, making the discovery of a "wonder drug" a difficult task (Figure 1). Figure 1. SGLT-2 inhibitors as a promising therapeutic agents for treatment of NAFLD/NASH patients. SGLT-2i treatment contributes to alleviation of NAFLD by reduction of hyperglycaemia, improvement of systematic insulin resistance, elevation of caloric loss and reduction of body weight mostly due to glycosuria. Apart from that, SGLT-2i play a hepatoprotective effect through reduction of hepatic de novo lipogenesis, hepatic inflammation, apoptosis, ER-stress, oxidative stress, and increase of hepatic beta-oxidation. Reduced activation of hepatic satellite cells and p53/p21 pathways by SGLT-2i leads to amelioration of hepatic fibrosis and HCC development.

SGLT-2i Overview
Phlorizin, a b-Dglucoside, was the first natural non-selective SGLT-2i isolated from the bark of the apple tree in 1835 and was initially used for fever relief and treatment of infectious diseases [62,63]. Phlorizin contains a glucose moiety and an aglycone in which two aromatic carbocycles are joined by an alkyl spacer. Approximately fifty years later, scientists observed that phlorizin in high dose lowered plasma glucose levels through glucosuria independently of insulin secretion [62][63][64]. However, the mechanism of phlorizin action through the active glucose transport system of the proximal tubule was revealed in the early 1970s [62,65].
Thereafter, researchers began to discover the exact mechanism of phlorizin action and its potential application in the management of hyperglycaemia. Phlorizin, when delivered orally, undergoes hydrolysis of the O-glycosidic bond by intestinal glucosidases, leading to formation of phloretin, which is capable of GLUT transporter inhibition and uncoupling of oxidative phosphorylation. Due to Phlorizin's low SGLT-2 selectivity, rapid degradation, inhibition of the ubiquitous glucose transporter 1 (GLUT1), poor intestinal absorption and its gastrointestinal side effects (e.g., diarrhea and dehydration), possibly due to its higher potency for SGLT-1, it failed to be developed as an antidiabetic drug [62, [66][67][68].
Nevertheless, in order to overcome the aforementioned side effects of phlorizin, a researcher developed a devphlorizin-based analog with improved bioavailability/stability and more selectivity towards SGLT-2 than SGLT-1. Initially, they focused on the development of O-glucoside analogs such as T-1095, sergliflozin and remogliflozin [69,70]. However, due to their poor stability and incomplete SGLT-2 selectivity, pharmaceutical industries turned to another derivative of phlorizin known as C-glucosides [62].
Since then, there have been numerous attempts to synthesize phlorizin C-glucoside analogs with sufficient potency and selectivity for inhibition of SGLT-2. These attempts led to development of dapagliflozin by Meng et al. in 2008 [71]. Dapagliflozin demonstrated more than 1200-fold higher potency for human SGLT-2 versus SGLT-1. Apart from dapagliflozin, during the next years, several other C-glucoside inhibitors have been developed and approved by United States Food and Drug Administration (FDA). Canagliflozin, characterized by a thiophene derivative of C-glucoside, showed over 400-fold difference in inhibitory activities between human SGLT-2 and SGLT-1 [72]. Thereafter, empagliflozin, which has the highest selectivity for SGLT-2 over SGLT-1 (approximately 2700-fold), was the third agent approved by both the European Medicines Agency (EMA) and the FDA, while other SGLT-2i, namely luseogliflozin and topogliflozin, are approved so far for use only in Japan, and ipragliflozin only in Japan and Russia [73][74][75][76][77].
The structure of phlorizin and currently FDA-approved SGLT-2 inhibitors in the United States, including dapagliflozin, canagliflozin, empagliflozin, and ertugliflozin, along with their chemical formulae and brand names are presented in Figure 2 [78]. Diverse structures of other SGLT-2 inhibitors have been previously disclosed in various articles and in a number of patents [79,80].
Studies have indicated that most SGLT-2i exert anti-proliferative activity in several hepatocellular cell lines, through-among other mechanisms-attenuation of glucose uptake [86]. Incubation of both L02 and HepG2 cells with canagliflozin at various concentrations for 24 h resulted in a significant reduction of cell proliferation through an increase in G0/G1 and a decrease in G2/M phase cell population [86]. In addition, an apoptotic effect was exhibited through activation of caspase 3 in HepG2 cells [87]. From a mechanistic point of view, cell growth regulators, cyclins and cyclin-dependent kinases (CDKs) have been indicated as direct targets of SGLT-2i to control proliferation and survival processes. As such, canagliflozin-treated HepG2 cells demonstrated increased expression of the cell growth regulator hepatocyte nuclear factor 4α (HNF4α) [82]. Furthermore, reduced expression of cyclin D1, cyclin D2 and cdk4 leading to cell cycle arrest in a hepatocellular carcinoma (HCC) cell line have been reported after treatment with canagliflozin [82,87]. In another study, incubation of HepG2 cells with canagliflozin resulted in reduced clonogenic cell survival and elevation of anti-carcinogenic potential of γ-irradiation through modulation of ER stress-mediated autophagy and cell apoptosis [88]. On the contrary, trilobatin, a novel SGLT-1/2 inhibitor, increased the HepG2 cell proliferation rate at a dose of 10, 50 and 100 μΜ, while incubation of human HCC cells with tofogliflozin at various concentrations did not alter the HCC cell proliferation rate [89]. Incubation of HepG2 cells with dapagliflozin, canagliflozin and empagliflozin had no effect on cancer cell survival [81,86] and adhesion capacity [81], while the cell sensitivity to dapagliflozin was induced after UDP Glucuronosyltransferase Family 1 Member A9 (UGT1A9) silencing, indicated by an increased number of floating HepG2 cells; of note, UGT1A9 metabolizes and deactivates dapagliflozin [81].
Dapagliflozin treatment remarkably suppressed oleic acid (OA)-induced lipid accumulation and TG content in L02 cells through increased FA β-oxidation, as indicated by elevated proliferator-activated receptor-gamma coactivator-1 alpha (PGC-1α) levels and activation of the AMP-activated protein kinase (AMPK)/mammalian target of rapamycin (mTOR) signaling pathway [84]. Several in vitro studies have shown that AMPK is a key regulator that mediates the various beneficial effects of SGLT-2i related to cholesterol and glucose metabolism in hepatic cells. AMPK activation by dapagliflozin prevented glucose absorption through reduced SGLT-2 expression in OA-stimulated HuS-E/2 cells [90]. This effect was eliminated after incubation of cells with compound C, a potent AMPK inhibitor [84]. Moreover, treatment of HepG2 cells with canagliflozin facilitated hepatic cholesterol efflux via activation of AMPK. In turn, AMPK activation led to increased expression of the liver X receptor (LXR) and its downstream proteins, resulting in subsequent stimulation of cholesterol reverse transport [91]. LXR activation also accelerates fecal cholesterol disposal through regulation of ATP-binding cassette (ABC) transporters ABCG5 and ABCG8 expression [92]. Canagliflozin treatment resulted in reduced expressions of ABCG5 and ABCG8 and LXR, while this effect was inhibited after treatment of cells with compound C [91]. An increased AMP/ADP ratio leads to AMPK activation. It is known that SGLT-2i reduces cellular ATP levels and indirectly activates the AMPK signaling pathway [83,93]. It has been shown that dapagliflozin alleviates the intracellular ATP levels via regulation of glucose metabolism [84].
Stimulation of hepatic cell lines with palmitic acid (PA) mimics hepatocyte situation in NASH [94,95]. Incubation of PA-stimulated L02 and HepG2 cell lines with dapagliflozin resulted in a significant reduction of intracellular lipid accumulation. This effect was attributed to down-regulation of proteins related to lipid synthesis, up-regulation of genes involved in fatty acid oxidation (e.g., peroxisome proliferator-activated receptor alpha (PPARα) and carnitine palmitoyltransferase 1 (CPT1a)), regulation of AMPK/mTOR pathway and autophagy. Interestingly, incubation of cells with compound C abolished the dapagliflozin beneficial effect on reduction of intracellular lipid accumulation [85]. Collectively, these data suggest that SGLT-2i induced improvement of NAFLD is directly dependent on AMPK signaling activation.
The divergent results of the various relevant studies suggest that the effects of SGLT-2i on cell proliferation, survival and apoptosis-which are important processes in hepatocellular carcinoma development and progression-appear to be drug-, dose-and duration-dependent.

Animal Studies
Overall, the animal models of NAFLD/NASH can be divided into two categories. Of these, one comprises genetically modified animal models, such as leptin deficient (ob/ob), leptin-resistant (db/db), Agouti mutation (KK-Ay), and apolipoprotein E knockout (ApoE-/-) mice, as well as the Prague hereditary hypertriglyceridemic (HHTg) rats, Zucker diabetic fatty (ZDF) rats and Otsuka Long-Evans Tokushima Fatty (OLETF) rat. The other category includes dietary/pharmacological manipulation-induced NAFLD/NASH animal models, such as those fed with either a methionine/choline-deficient (MCD) diet, the trans-fat containing AMLN (amylin liver non-alcoholic steatohepatitis, NASH) diet, and the methionine-reduced, choline-deficient and amino acid defined (CDAA) diet, as well as a high-fat diet (HFD), and a HFD + cholesterol diet or a high-fat, high-calorie (HFHC) diet [30, [96][97][98][99][100]. Most of the recent data indicate that both SGLT-1 and SGLT-2 genes are expressed in hepatic tissues of mice and rats [30, 85,87,101]. High levels of SGLT-2 protein have also been detected in hepatic macrophages and T cells [101]. While the exact underlying molecular mechanisms of SGLT-2i-induced beneficial effects on NAFLD have not yet been fully clarified, most of our knowledge regarding the potential hepato-protective mechanism(s) of SGLT-2i actions comes from animal studies.
Although the SGLT-2i-induced weight loss seems to play an important role in hepato-protective effects on NAFLD in humans [102][103][104], data from animal studies suggest that other mechanisms are more likely responsible for the noted SGLT-2i-mediated hepato-protection.
Administration of SGLT-2i leads to net loss of calories and consequent attenuation of body weight gain, while reducing the accumulated white adipose tissue [99,[105][106][107]. In fact, it has been found that SGLT-2i promote weight loss through improvement of systemic IR, increased body temperature and basal metabolism via regulation of the AMP/ATP ratio [19]. Petito da Silva et al. reported that empagliflozin (Table 1) at a dose of 10 mg/kg/day given for five weeks significantly reduces body weight and body mass in HFD-fed male C57Bl/6 mice, despite a slight increase in their appetite [107]. Similar results were observed in HFD-fed C57BL/6J male mice after administration of canagliflozin ( Table 2) at 30 mg/kg dose for four weeks in parallel with improvement of liver function, liver TG content and NAS score [108]. Additionally, short term and low dose administration of dapagliflozin (Table 3) reduced both ALT levels and body weight in both HFD and HFD+MCD diet fed C57BL/6J mice [109]. However, it should be noted that not all studies point towards a beneficial effect of SGLT-2i on body weight. Several studies have shown that administration of SGLT-2i does not exert any significant effect on body weight gain, epididymal fat weight or food intake [30, 105,110,111]. In addition, although several studies reported strong effects of low dose ipragliflozin (Table 4) on weight loss [100,112,113], supplementation of an AMLN diet with 40 mg ipragliflozin/kg for 8 weeks did not significantly alter the body weight of AMLN diet fed mice, while it improved liver function, hepatic fibrosis and NA score [114]. Empagliflozin at the dose of 10 mg/kg/day improved liver function and NAFLD status, while it had no significant effect on the body weight of ZDF, HHTg and Wistar rats [115,116], as well as on the body weight and appetite of ApoE knockout mice [30].
Although there are studies showing that administration of SGLT-2i does not exert any significant effect on body weight gain [30, 100,[110][111][112][113][114], the vast majority of the literature points towards a beneficial effect on weight loss.
However, the protective role of SGLT-2i against NAFLD progression also seems to be mediated by effects beyond those on body weight. Indeed, it has been shown that SGLT-2i treatment improves IR and ameliorates the intracellular FFA, total cholesterol (TC) and TG accumulation through reducing the expression of genes involved in de novo lipogenesis, FA uptake and hepatic TG secretion, whilst it also promotes the expression of key regulatory genes of fatty acid β-oxidation [30, 84,100,107,109,112,117]. Specifically, empagliflozin administration has been shown to reduce hepatic triacylglycerol (TAG) levels and improve IR through reduction of lipotoxic intermediates formation, such as diacylglycerols (DAG) [115], which at elevated levels contribute to IR by activating the protein kinase C (PKC)ε pathway [118]. Of note, a recent study by Hüttl et al. demonstrated that reduction of hepatic lipid accumulation following empagliflozin treatment is mediated via increased nuclear factor erythroid 2-related factor 2 (Nrf2), fibroblast growth factor 21 (FGF21) and altered gene expression of the P450 (CYP) enzyme superfamily of cytochrome [115]. Interestingly, Nrf2 and FGF21 regulate lipid metabolism through inhibiting lipogenesis and improving insulin sensitivity, respectively [119,120].
Sterol regulatory element-binding transcription factor 1 (SREBP1) is one of the key regulators of hepatic lipogenesis and plays a crucial role in the regulation of lipogenic genes, such as fatty acid synthase (Fasn), acetyl-CoAcarboxylase 1 (Acc1), and stearoyl-CoA desaturase1 (Scd-1). Increased SREBP1 expression has been linked to the aggravation of hepatic steatosis [121]. While reduced expression of SREBP1 and Scd-1 by both canagliflozin and empagliflozin has been observed in several studies [30,107,108,115], Kern et al. found that empagliflozin treatment for eight weeks does not markedly affect the Scd-1 levels in db/db mice [122]. Dapagliflozin (1 mg/kg/day) treatment for nine weeks resulted in down-regulation of Scd-1 gene expression, as well as reduced ACC1 phosphorylation in ZDF rats. Taking into account that SREBP1 activity is directly regulated by mTOR signaling [123], reduced expression of SREBP1 and its downstream lipogenic targets could be due to the reduced mTOR expression/activity observed after SGLT-2i administration [84,123].
Peroxisome proliferator-activated receptor gamma (PPAR-γ) upregulation triggers de novo lipogenesis and consequent deposition of lipid droplets into hepatocytes [124]. Eight weeks of dapagliflozin treatment reduced hepatic weight and prevented progression of hepatic steatosis compared to mice treated with insulin glargine [125]. This effect was found to be mediated through decreased expression of PPAR-γ targeted genes involved in fatty acid synthesis, such as Scd-1, monoacylglycerol O-acyltransferase 1 (Mogat1), Cell death inducing DFFA like effector A (Cidea) and cell death inducing DFFA like effector C (Cidec), without affecting insulin sensitivity, liver TC or oleate content [125].
SGLT-2i have been also shown to reverse the HFD-induced down regulation of genes involved in FA β-oxidation and lipolysis in the liver of various mouse models of NAFLD [30, 84,107,108]. Indeed, it has been shown that expression of PPARα is induced by SGLT-2i [84,105,107,114]. PPARα activation is predominant for regulation of genes related to the FA β-oxidation process in mitochondria, such as peroxisomal acyl-CoA oxidase1 (ACOX1) and enoyl-CoA hydratase, CPT1, and cytochrome-mediated (CYP4A1 and CYP4A3) [44]. Ipragliflozin treatment has been shown to result in acceleration of β-oxidation and export of very-low-density lipoprotein (VLDL) through upregulation of PPARα, CPT1A, and Microsomal Triglyceride Transfer Protein (MTTP) gene expression in the liver of AMLN-fed C57BL/6J mice. MTTP plays a crucial role in export of hepatic LDL [114]. Similarly, dapagliflozin attenuated hepatic lipid accumulation in ZDF rats via up-regulation of the FA β-oxidation enzyme ACOX1 [85]. Tofogliflozin administration induced the expression of genes related to FA β-oxidation in non-tumorous liver lesions, while it had no effect on their expression in tumorous liver tissues [126]. However, not all studies show a promotive effect of SGLT-2i on the expression of enzymes and transcription factors involved in the regulation of FA β-oxidation [107,114,126]. Of note, the expression of PPARα remained unaffected in both Wistar and HHTg rats after eight weeks of empagliflozin administration [107,115]. Administration of canagliflozin at doses of 10 and 20 mg/kg/day ameliorated the HFD-induced down-regulation of the expression of the key regulatory molecule of hepatic lipolysis, known as zinc alpha 2 glycoprotein (ZAG). When the same SGLT-2i was given in KK-Ay mice, it reduced the accumulation of lipid droplets in the liver by increasing hepatic prostaglandin E2 (PGE2) protein levels [111]. Furthermore, the combined therapy with canagliflozin and exercise exerted an additive effect on hepatic PPARγ Coactivator 1 Beta (PGC1b) elevation, and reduced expression of hepatic lipogenic genes, such as Scd1 [108]. Notably, PGC1 plays a central role in the control of hepatic gluconeogenesis, while its overexpression has been associated with stimulation of hepatic FA oxidation [127].
The inhibition of de novo lipogenesis, and enhanced lipolysis and β-oxidation by SGLT-2-i lead to a consequent reduction in oxidative stress, hepatic inflammation and apoptosis [128]. Administration of canagliflozin to HFD-fed diabetic Wistar adult male rats resulted in reduced hepatic steatosis through amelioration of oxidative stress, inflammation and apoptosis, as indicated by reduced plasma malondialdehyde (MDA) levels, serum tumor necrosis factor (TNF) and caspase-3 levels, as well as hepatic expression of interleukin-6 (IL-6) [128,129]. Moreover, dapagliflozin treatment reduced the expression of hepatic inflammatory cytokine (TNF-α, IL-1β, IL-18) content and improved hepatic steatosis in HCHF fed male Wistar rats [130]. Tahara et al. have also shown anti-oxidative stress and anti-inflammatory effects of ipragliflozin on HFD-fed and streptozotocin-nicotinamide-induced type 2 diabetic mice [131]. Similarly, dapagliflozin treatment at a dose of 1 mg/kg/day decreased hepatic ROS production of myeloperoxidase (MPO) and F4/80, thus ameliorating serum ALT levels and hepatic fibrosis [132]. MPO is a chlorinating oxidant-generating enzyme that regulates the initiation of an acute inflammatory response and promotes the development of chronic inflammation through oxidant production [133]. Furthermore, canagliflozin treatment of F344 rats for 16 weeks resulted in reduced hepatic ROS production and MDA and 8-hydroxy-2′-deoxyguanosine (8-OHdG) levels [110]. MDA and 8-OHdG are known biomarkers of lipid peroxidation and oxidative DNA damage, respectively. Of note, canagliflozin and teneligliptin combination therapy showed a stronger effect on reduction of hepatic oxidative stress and amelioration of hepatic inflammation [110]. Treatment with both remogliflozin and ipragliflozin also reduced oxidative stress levels, as evaluated by decreased thiobarbituric acid reactive substances (TBARS) levels [100,134]. Instead, administration of dapagliflozin at 1 mg/kg/twice/day for four weeks had no significant effect either on hepatic TBARS and TG levels or on plasma ALT levels [135]. In addition, reduction of oxidative stress by empagliflozin treatment resulted in amelioration of hepatic inflammation [30, 107,116] and steatosis, as judged by down-regulation of inflammatory markers [30, 107,115]. Of interest, decreased macrophage infiltration expression [107,117,134] as well as elevation of autophagy markers [30,117] have also been linked to the anti-inflammatory actions of SGLT-2-i in the liver. Specifically, oxidative stress was reduced after empagliflozin treatment via up regulation of Nrf2, which led to increased antioxidant enzyme activity (SOD) and reduced hepatic inflammation [115]. Empagliflozin exerted anti-inflammatory actions in diet-induced obese mice and NASH mouse models through, at least in part, suppression of hepatic nuclear factor kappa-light-chain-enhancer of activated B cells (NFκΒ), monocyte chemoattractant protein-1 (MCP-1) and TNF-α expression, as well as inhibition of the IL17/IL23 axis [115,117,136]. Reduced hepatic inflammation has also been reported in mice and rats after treatment with low dose (1 mg/kg/day) dapagliflozin and ipragliflozin. Indeed, ipragliflozin administration for 12 weeks alleviated hepatic fibrosis through reduced expression of pro-inflammatory markers Emr1 and Itgax in non-tumoric lesions [126]. On the contrary, a recent study in mice by Hupa-Breier et al. demonstrated that four weeks treatment of diet-induced NASH with empagliflozin alone at a dose of 10 mg/kg/day did not exert any beneficial effect on NASH, while it significantly increased expression of pro-inflammatory and pro-fibrotic genes. In the same study, the combination of dulaglutide (a GLP-1 agonist) and empagliflozin exhibited a hepato-protective effect on diabetic background mice through modulation of the pro-inflammatory immune response and microbiome dysbiosis [137].
Meng et al. demonstrated that induction of AMPK/mTOR activity is dominant pathway in empagliflozin-induced beneficial effects on liver inflammation and ALT levels [117]. Dapagliflozin treatment also induced AMPK/mTOR pathway activation through phosphorylation of liver kinase B1 (LKB1; a ubiquitously expressed master serine/threonine kinase that activates downstream kinases of the AMPK pathway) [105,138]. Previous studies have demonstrated that AMPK activation is required for SGLT-2i induced autophagy-dependent lipid catabolism [30,117]. Both dapagliflozin and empagliflozin promote hepatic autophagy through increased AMPK phosphorylation and BE-CLIN gene expression, as well as reduced P62 levels and mTOR levels and activity [30,85].
Apart from impaired autophagy, ER stress is also implicated in the development of steatosis and the progression of NAFLD/NASH [30,107]. Consistent with the regulatory role of ER stress in the autophagy process, Nasiri-Ansari et al. demonstrated that stimulation of autophagy by empagliflozin leads to amelioration of ER stress and reduced hepatic cell apoptosis [30]. In particular, empaglifilozin administration protects against HFD-induced NAFLD through inhibiting all three branches of ER stress, namely inositol-requiring enzyme 1 (IRE1a), X-box binding protein 1 (Xbp1), activating transcription factor 4 (ATF4), C/EBP homologous protein (CHOP) and activating transcription factor 6 (ATF6) [30,107]. A recent study by Chen et al. also showed that down-regulation of the ATF6 signaling diminished ER stress-induced inflammation and apoptosis of hepatic cells [139]. The synergistic ER stress response and autophagy process regulated hepatic cell apoptosis in HFD-fed ApoE(-/-) mice with NASH, by reducing cleaved caspase 3 levels and elevation of B-cell lymphoma-2 (Bcl-2)/BCL2 Associated X (Bax) ratio [30]. Of note, canagliflozin was previously found to exert an anti-apoptotic effect on HFD-fed mice as revealed by robust Bcl-2 hepatic expression [129].
Chronic liver inflammation leads to transformation of hepatic stellate cells to myofibroblasts, thus contributing to liver fibrosis and progression of NAFLD to NASH [140]. Transforming growth factor beta (TGF-β) is known as the most potent inducer of liver fibrosis. Administration of canagliflozin for 16 and 20 weeks reduced the expression of hepatic TGF-β in both F344 rats and MC4R-KO mice, respectively [101,110]. Interestingly, TGF-β activation leads to increased fibronectin and collagen types I, II, and IV production [141]. Treatment of mice with both empagliflozin and canagliflozin resulted in reduced expression of different types of collagen, implying synergistic action(s) [87,137]. Luseogliflozin also exerts anti-fibrotic effects through reduction of collagen1a1, colla-gen1a2, TGF-β and smooth muscle actin (SMA) expression [142]. Extensive data exist regarding the crucial role of tissue inhibitors of metalloproteinases (TIMPs) in the progression of hepatic fibrosis. Reduced expression of TIMPs and amelioration of hepatic fibrosis have been observed after luseogliflozin and canagliflozin administration [101,126,142]. On the contrary, empagliflozin treatment did not affect the expression of stellate cell functionality markers, such as galectin-3, a-SMA, and collagen1a1 [99].
NASH has been associated with an increased risk of cirrhosis and HCC development. Hepatic tumorogenesis is the result of DNA instability caused by several factors, such as hepatic lipotoxicity, aberrant metabolism and inflammation [141,143]. As a protective effect of SGLT-2i on NAFLD/NASH progression has been shown in the vast majority of the studies, further research aimed to evaluate the effect of SGLT-2i on development and progression of HCC [87,101,142]. Activation of the hepatic P53/P21 signaling pathway is positively correlated with induced fibrosis and hepatocellular carcinogenesis. A 14-week treatment with tofogliflozin has been shown to effectively reduce the P21 expression, alleviating the progression of NASH through, at least in part, mitigating genes related to the hepatocyte cellular senescence-associated secretory phenotype [89]. Daily administration of canagliflozin prevented the occurrence of HCC in both STAM mice-a NASH model-(11 weeks) and HFD-fed MC4R-KO mice (52 weeks), as indicated by a significantly reduced number of hepatic tumorous lesions [87,101]. Moreover, 66 weeks administration of canagliflozin to CDAA-fed F344 rats exerted an anti-carcinogenic effect, as indicated by a reduced number of cells positively stained for placental glutathione S-transferase (a prominent marker of hepatocarcinogenesis) [110]. Finally, canagliflozin administration reduced the development of hepatic preneoplastic lesions in CDAA-fed rats through suppression of hepatic neovascularization markers, namely cluster of differentiation 31 (CD31) and vascular endothelial growth factor (VEGF) [110]. Overall, the key animal studies regarding the impact of SGLT-2i on NAFLD/NASH/HCC are summarized in Tables 1-4.

Human Trials
A number of clinical studies have highlighted the benefit of SGLT-2i in patients with T2DM and NAFLD (Table 5) . In the majority of them, the administration of SGLT-2i has resulted in improvement of serum levels of liver enzymes and hepatic steatosis, as evaluated by magnetic resonance imaging (MRI), ultrasound (U/S), non-invasive biomarkers, such as AST to platelet ratio (APRI) index, NAFLD fibrosis score (NFS) and Fibrosis-4 (FIB-4) score, or even by liver biopsy (LB). In some of these studies, improvement in hepatic fibrosis was found, using transient elastography (TE) or LB, even though this finding was not univocal [146,153,158,159,164,165,168].
Of note, the overall improvement in hepatic steatosis is also found in patients treated with other classes of anti-diabetic drugs, like thiazolidinediones and dipeptidyl peptidase-4 (DPP-4) inhibitors, raising the question of steatosis improvement due to glycemic control [148,160,162,170]. Moreover, the vast majority of the aforementioned studies are limited by the small sample size, heterogenous inclusion criteria, especially regarding the presence of NAFLD, as well as the duration of follow up, thus meta-analyses have been conducted to better assess the true benefit of SGLT-2i in patients with DM and NAFLD [171][172][173]. In the largest one, comprising 9 randomized trials, with 7281 and 4088 patients in the SGLT-2i and control arm, (standard of care (SOC) or placebo), respectively, the use of SGLT-2i resulted in improvement of serum transaminases, body weight, and liver fat as measured by proton density fat fraction. Accordingly, the authors discuss that this improvement derives mainly from the achievement of glycemic control and weight loss. However, in a 2020 study by Kahl et al., including 84 patients with DM and excellent glycemic control, randomly assigned to empagliflozin or placebo, patients on empagliflozin had improved liver fat content, as assessed by MRI [161], strongly suggesting that the good glycemic control and weight loss are not the only mechanisms associated with the beneficial effects of SGLT-2i treatment on hepatic steatosis.
In non-DM patients, only a small single center study exists which studied 12 patients under dapagliflozin and 10 patients under teneligliptin, a DPP4 inhibitor, for a total of 12 weeks, showing that after this intervention period, serum transaminases were decreased in both groups, while in the dapagliflozin group, total body water and body fat decreased, leading to decreased total body weight [174].
Regarding the pathophysiology of NAFLD improvement under SGLT-2i treatment, various mechanisms have been suggested. Treatment with SGLT-2i results in decreases in both glucose and insulin levels (especially in patients with DM), which lead to a large reduction of hepatic de novo lipid synthesis [175]. Moreover, glucagon-secreting alpha pancreatic cells also express SGLT-2, thus the administration of SGLT-2i stimulates glucagon secretion [175][176][177]. In turn, the subsequently elevated plasma glucagon levels stimulate β-oxidation, and this shift from carbohydrate to fatty acid metabolism leads to reduced liver triglyceride content and consequently hepatic steatosis [175,178,179]. Another potential mechanism is mediated by the antioxidant effects of SGLT-2i. Apart from their ability to reduce high glucose-induced oxidative stress, SGLT-2i reduce free radical generation, suppress pro-oxidants, and upregulate antioxidant systems, such as superoxide dismutases (SODs) and glutathione (GSH) peroxidases (Figure 1) [116,[180][181][182][183][184].

Conclusions
There is increasing interest regarding the promising effect(s) of SGLT-2i for the treatment of NAFLD, regardless of the co-existence of T2DM. In addition to weight loss, the beneficial effect(s) of SGLT-2i on NAFLD development and progression appear to be mediated directly through regulation of multiple processes, including ER stress, oxidative stress, low-grade inflammation, autophagy and apoptosis, as revealed by in vitro, animal and clinical studies. Moreover, the observed different effects between members of the SGLT-2i class suggest that there are features specific to individual drugs of this class regarding the underlying mechanism(s) of action and their corresponding effects on NAFLD.