Overview and prospect of NAFLD: Significant roles of nutrients and dietary patterns in its progression or prevention

Abstract

NAFLD is the most common chronic liver disease worldwide, characterized by lipid accumulation in the liver, and usually evolves from steatohepatitis to fibrosis, cirrhosis, or even HCC. Its incidence is rapidly rising in parallel with the increasing prevalence of obesity and metabolic syndrome. Current therapies are limited to lifestyle changes including dietary intervention and exercise, in which dietary modification exerts an important part in losing weight and preventing NAFLD. In this review, we briefly discuss the roles and mechanisms of dietary components including fructose, non-nutritive sweeteners, fat, proteins, and vitamins in the progression or prevention of NAFLD. We also summarize several popular dietary patterns such as calorie-restricted diets, intermittent fasting, ketogenic diets, Mediterranean diets, and dietary approach to stop hypertension diets and compare the effects of low-fat and low-carbohydrate diets in preventing the development of NAFLD. Moreover, we summarize the potential drugs targeting metabolic-related targets in NAFLD.

INTRODUCTION

NAFLD, characterized as a spectrum of pathological stages including NAFL, NASH, and advanced fibrosis, can lead to cirrhosis and finally end-stage liver disease. Driven partially by rising obesity levels, the global prevalence of NAFLD among adults, children, and adolescents has increased in the last few decades,¹ reaching 25% in 2016, with higher rates observed in South America, the Middle East, and Asia.² The overall prevalence of NAFLD has reached 29.1% in Chinese people, and the proportion of young patients is rising, portending a heavier disease burden in the future.³ Though the majority of patients live with NAFL and early NASH without progression to later disease stages, the economic burden of NAFLD is getting heavier due to its huge affected population.¹ NAFLD has been a primary cause of HCC in many countries.⁴ Being able to develop in patients without cirrhosis, NAFLD-related HCC is associated with a later stage at diagnosis and, conceivably, a poorer prognosis.⁵ It is also reported that NAFLD is an independent risk factor for cardiovascular diseases (CVDs), type 2 diabetes mellitus (T2DM), and chronic kidney diseases.⁶

So far, there is no licensed drug for NAFLD treatment. Lifestyle alternation, especially dietary intervention, is recommended to tackle NAFLD and its potential complications, with the primary principle of reducing calories intake and restricting the consumption of lipogenic nutrients. However, specific recommendations for diets are vague and inconsistent among guidelines.⁷ In this review, we briefly generalize the effects of common dietary components on NAFLD and their potential mechanisms, and provide an overview of the dietary patterns found beneficial to NAFLD patients, thereby promoting NAFLD prevention and management in aspects of modern lifestyle.

SWEETENERS AND NAFLD

Key points

High fructose consumption increases hepatic de novo lipogenesis (DNL) and reduces fatty acid β-oxidation (FAO), leading to fatty acid deposition. Increasing oxidative stress and low-grade hepatic inflammation contribute to further NAFLD progression. A series of animal studies suggest that non-nutritive sweeteners (NNSs) have adverse impairment on metabolism as well.

The global change in dietary habits, especially the introduction of sugar-sweetened beverages, has led to increased consumption of carbohydrates, mostly refined sugar, over the last decades. Consumption of added sugar has long been proved a risk factor for the development and progression of NAFLD.⁸ Habitual exposure to sugar-sweetened beverages leads to higher visceral adipose tissue volume, increased liver fat, impaired serum liver enzyme concentrations, and higher NAFLD odds in adults.^9–12 Glucose, fructose, and sucrose are the 3 main carbohydrates consumed in modern diets, among which fructose has in particular gained research attention, not only for the evolution of sweeteners from sucrose to the “cheaper and sweeter” high-fructose corn syrup (HFCS), but also for its unique lipogenic potential.

Fructose and NAFLD

Once used as a sugar substitute for diabetic patients for its smaller influence on plasma glucose and serum insulin levels,¹³ fructose-containing diets have been confirmed in animal experiments as well as human studies to induce DNL and insulin resistance (IR) to a greater extent than normal or glucose-containing ones.^14–17 Researchers have showed an average fructose:glucose ratio of about 60:40 in the 5 most popular HFCS-sweetened sodas in the United States,¹⁸ much higher than that of the traditional HFCS-55, which contains 55% of free fructose, and HFCS-42, containing 42% of free fructose. Thus, it is of significant importance to further demonstrate the mechanisms of fructose-induced NAFLD.

Fructose is mainly metabolized in the liver, since its transporter, glucose transporter type-5 (GLUT5), is poorly expressed in most cells.^19,20 When it comes to hepatocytes, the membrane glycoprotein transmembrane 4 L 6 family member 5 (TM4SF5)–mediated GLUT8 relocalization to the plasma membrane helps transport extracellular fructose.^21,22 The activity of ketohexokinase (KHK), the first enzyme of fructose metabolism, is greater than that of glucokinase, indicating a greater chance of fructose to enter the hepatic lipogenesis compared with glucose.²³ Moreover, fructose metabolism bypasses phosphofructokinase-1, which acts as a rate-limiting enzyme of glycolysis, thereby escaping metabolic regulation.²⁴ Small amounts of fructose are mostly converted by the small intestine to glucose, while large amounts of fructose, especially those consumed in liquid form, can be directly absorbed and access the liver via the portal vein, adding to liver stress and increasing the risk of NAFLD.^19,25,26

High-carbohydrate diets stimulate the expression of carbohydrate-responsive element–binding protein (ChREBP) and sterol regulatory element–binding protein 1C (SREBP-1c) in rats,^27–29 the 2 major transcription factors regulating hepatic lipogenesis.^30,31 ChREBP and SREBP-1c induce lipogenic genes, including acetyl-CoA carboxylase (ACC), fatty acid synthase (FASN), ATP citrate lyase, and Stearoyl-CoA desaturase-1.^30,32 Consistently, all of the abovementioned lipogenic genes have been reported to be significantly upregulated after high carbohydrate exposure,^29,33,34 and several studies have demonstrated a greater change in mice fed with fructose-containing sugar.^15,16

While both SREBP-1c and ChREBP are required for full induction of liver lipogenic genes in response to carbohydrates,³¹ ChREBP is the only one reported to modify monosaccharide transportation and glycolysis. Intestinal ChREBP stimulated by oral fructose ingestion upregulates Slc2a5, which encodes fructose transporter GLUT5, and the intestine-specific ChREBP-KO mice have been proved to be fructose-intolerant.³⁵ Hepatic ChREBP induces liver enzyme KHK and pyruvate kinase (l-PK), both involved in fructose glycolysis.^16,31,36 Increased levels of KHK¹⁶ and l-Pk,³⁷ along with higher pyruvate levels³⁴ after fructose ingestion, indicate thriving glycolysis, while in a cell model of human hepatocyte metabolism, intermediates of glycolysis decrease indeed,³⁸ suggesting an increased consumption of glycolysis intermediates as the carbon source in hepatic DNL. It is worth mentioning that liver X receptors³⁴ and mTORC1³⁰ are the potential upstream regulators of fructose-mediated ChREBP and SREBP-1c activation.

While increased hepatic DNL has been widely accepted as the main cause of fructose-mediated hepatic lipid deposition, there is mounting evidence that the reduction in FAO and increased production of reactive oxygen species (ROS) contribute as well. Hepatic malonyl-CoA, an intermediate from DNL, is increased after fructose ingestion,²⁸ which inhibits carnitine palmitoyl transferase 1 (CPT1), thus disturbing the time-limiting process of FAO of acyl-CoAs transported across the mitochondrial membrane in the form of acylcarnitines.³⁹ Impaired mitochondrial function is another reason for the diminished FAO, since the high-fructose-high-fat diet (HFD) led to a persistent decrease in mitochondrial size and DNA in animal models, which was not observed in high-glucose-high-fat groups.^40,41 The tryptophan pathway, known as an NAD+ synthesis pathway, is significantly decreased in female juvenile rats after fructose consumption, and diminished supply of NAD+ is proved to impair mitochondrial fatty acid oxidation.^42–44 Most HFD-induced NAFLD animal models show increased FAO in response to lipid deposition, indicating that sugar consumption may deteriorate the adverse effects of HFDs.⁴⁵ The rise of antioxidant peptides including ascorbate and glutathione in response to HFCS-sweetened diets suggests increasing ROS production.⁴³ HFCS-fed mice showed a lower diacylglycerol O-acyltransferase 2 expression, which is associated with impaired antioxidant ability and increased fibrosis.^29,46 Furthermore, a sex-specific pattern can be seen in the HFCS-induced hepatic steatosis, with a greater reduction of FAO and higher antioxidant potential in female rats.⁴³

Fructose consumption is linked to low-grade hepatic inflammation, which is a leading cause of NAFLD progression. After consuming 60% fructose solution for 9 weeks, both male and female Wistar rats showed increased hepatic TNF alpha (TNFα), one of the markers of inflammation.^47,48 However, NFκB activation was only detected in male rats, which may be due to the protection effect of the enhanced glucocorticoid signaling restricted to female rats.⁴⁸ Intriguingly, consuming fructose under stressful conditions, as performed by Kovačević et al⁴⁹ in female Wistar rats, counteracted the effects of stress on hepatic lipid accumulation alone. Whether this is also a secondary effect of the elevated glucocorticoids remains unknown.

In addition to the lipogenic nature of the added sugar itself, gluttony induced by sweet food deserves concern as well. Scientists do describe a decreased ability toward moderate energy intake in mice fed with sugar-sweetened diets while provided an additional pre-meal.⁵⁰ Though sugar-fed animals tend to reduce chow intake, they still consume more energy in total.⁵⁰ It has been reported that fructose can induce leptin resistance and reduce satiety in animals by activating central AMP-activated protein kinase, thus inhibiting glucagon-like peptide-1 (GLP-1).^51,52 fMRI studies in humans also showed that fructose ingestion failed to reduce the activity of the hypothalamus and striatum, the 2 brain regions associated with satiety.⁵³ Interestingly, Levy et al⁵⁴ found in mice that the fructose:glucose ratio was negatively correlated to sugar self-administration, resulting in fewer nose pokes and a lower caloric intake in the HFCS-fed group compared with the sucrose-fed one. One reasonable explanation for this conflicting observation is that fructose has a stronger reinforcing nature, which may be associated with reduced self-administration following higher substance concentration as observed in amphetamine and cocaine⁵⁵ (Figure 1).

FIGURE 1.

Sweeteners and NAFLD. Fructose induces ChREBP and SPEBR-1c, resulting in upregulation of glycolytic enzymes and lipogenic genes, both of which contribute to increasing hepatic DNL. The consequent accumulation of malonyl-CoA, an intermediate from DNL, along with the fructose-induced mitochondrial dysfunction, diminishes lipid metabolism and leads to further fatty acid deposition. Increasing oxidative stress and low-grade hepatic inflammation are found after high-fructose consumption, potentially leading to NAFLD progression. It should not be ignored that high-fructose diets can increase food intake by inhibiting satiety and inducing leptin resistance, since excessive energy intake is considered a crucial risk factor of NAFLD. As far as we know, some widely used NNSs such as sucralose and saccharin also exhibit adverse effects including bile acid accumulation, microbiota dysbiosis, and increasing hepatic DNL as well as oxidative stress, which may link them to the development and progression of NAFLD. Abbreviations: AMPK, AMP-activated protein kinase; BA, bile acid; ChREBP, carbohydrate-responsive element-binding protein; CPT, carnitine palmitoyl transferase; Dgat2, diacylglycerol O-aAcyltransferase 2; GLUT, glucose transporter type; KHK, ketohexokinase; l-PK, liver-pyruvate kinase; ROS, reactive oxygen species; SREBP-1c, sterol regulatory element–-binding protein 1C; TNFα, TNF alpha.

Non-nutritive Sweeteners and NAFLD

Although NNSs including sucralose, acesulfame, saccharin, and aspartame are often used as a control in fructose-related experiments to exclude the interference of sweet taste, there have been a series of animal studies showing that NNSs also have adverse impairment on metabolism. Ryuk et al⁵⁶ reported the ability of non-nutritive aspartame and sucralose to induce IR in rats, indicating that the metabolic impairments may start with the sweet taste itself. Sucralose upregulates hepatic expression of Cyp7a1 mRNA, leading to bile acid (BA) accumulation,⁵⁷ which is associated with the development and progression of NAFLD. Altered hepatic protein metabolism is also reported after sucralose consumption.⁵⁸ Aspartame, saccharin, and aspartame may increase hepatic oxidative stress, deplete antioxidant levels, and elevate plasma aminotransferase and alkaline phosphatase activities.^59–61 Saccharin consumption is also reported to potentially increase hepatic enzymes, including phosphofructokinase-1 (PFK-1), l-PK, and Fasn.⁶²

Microbiota dysbiosis is another focused mechanism of NNS-associated hepatic disorders. Sucralose consumption significantly increases the Bacteroides and Clostridium population, resulting in higher deoxycholic acid serum concentration, a potential risk factor for NAFLD.⁵⁷ Both sucralose and saccharin have been found to cause hepatic and systemic inflammation in mice through the alternation of gut microbiomes and their metabolites.^63,64 Moreover, maternal sucralose intake can impair intestinal function and alter the gut microbiota of offspring⁶⁵ (Figure 1).

However, current results are conflicting and sparse about whether NNSs impair hepatic lipogenesis, and whether the detected metabolism effects are sufficient to increase NAFLD risks.^66–68 While some studies have mentioned a weight-loss effect of artificial sweeteners, there is also contrary evidence that aspartame treatment induces abdominal adiposity in rats.⁶¹ When consumed with a HFD, sucralose is observed by Pino-Seguel et al⁶⁹ to prevent weight gain and alleviate glucose intolerance in mice. Contrast results were detected by Wu et al⁷⁰ under a similar protocol wherein sucralose not only had no effect on the body weight of HFD-fed mice but also exacerbated HFD-induced fatty liver. Therefore, more studies are required to precisely evaluate the metabolism effect of NNSs.

LIPIDS AND NAFLD

Key point

Dietary lipids have a dualistic influence on NAFLD. On one hand, excess lipids, including free cholesterol (FCs) and free fatty acid (FFAs), especially saturated fatty acid (SFAs), promote the development of NAFLD. In addition, a higher dietary ratio of n-6 PUFAs to n-3 PUFAs is an independent promoter of NAFLD. On the other hand, increasing the dietary intake of n-3 can alleviate NAFLD.

A strong correlation between HFD and NAFLD is already known. A large amount of dietary lipid intake may cause the hepatocytes to be in an environment with high FFAs and high FCs, which may disrupt the homeostasis of liver lipid metabolism by lipotoxicity and lead to the development of NAFLD. In mouse NASH models induced by HFD or palmitic acid treatment, upregulation of vascular cell adhesion molecule 1 in liver sinusoidal endothelial cells has been observed, which promotes adhesion with liver sinusoidal endothelial cells and activation of macrophages.^71,72 Another study showed that in cell and murine models, the activation of kinase receptor-interacting protein 1 (RIP1) of macrophages mediates inflammatory responses and apoptosis.⁷³ In a high-fat environment, hepatocytes also play an important role in the activation of macrophages. Hepatocytes stimulate activation of macrophages as well as mediate inflammation and oxidative stress by secreting various inflammatory extracellular vesicles.^74–77 Hepatocytes can induce apoptosis and HSC activation through autocrine and paracrine IL-11, respectively.⁷⁸ In addition, the upregulation of lysophosphatidylcholine acyltransferase 2, cytoplasmic phospholipase A2, and 15-lipoxygenase in hepatocytes can cause the increase of arachidonic acid, a proinflammatory factor in the cell membrane,⁷⁹ which results in ROS accumulation, oxidative stress, ER stress, mitochondrial damage, and hepatocyte apoptosis, thus promoting NAFLD. The activated macrophage stimulating 1-AMP-activated protein kinase pathway in macrophages, tank-binding protein kinase 1 (TBK1) in hepatocytes, and the inhibited Sirtuin 3 (SIRT3)-ERK-CREB-Bnip3 pathway all induce hepatocyte apoptosis through mitochondrial autophagy inhibition.^80–82

Oxysterols, the oxidation products of excessive cholesterol, and protein kinase R (PKR)-like endoplasmic reticulum kinase (PERK)-mediated c-Jun N-terminal activation can cause mitochondrial damage, thereby exerting lipotoxicity on hepatocytes in vivo and vitro.^83,84 Caspase-2–mediated hepatocyte apoptosis and SMS1-DAG-PKCδ-NLRC4 axis–mediated pyroptosis are also involved in lipotoxicity damage.^85,86 In addition, Li et al⁸⁷ found that the expression of 3-mercaptopyruvate sulfurtransferase, a key enzyme of hydrogen sulfide biosynthesis, is upregulated by high FFAs, which inhibits CSE/hydrogen sulfide in hepatocytes and leads to NAFLD. High FFAs may also promote NAFLD by BA accumulation.⁸⁸ Besides, IR is one of the crucial factors for NAFLD development. Diabetic mice are more likely to develop NAFLD under HFD conditions than healthy wild-type mice, suggesting that HFD is more harmful to diabetic patients.⁸⁹ In primate experiments conducted by McCurdy et al⁹⁰, it was found that having HFD during pregnancy can increase the risk of developing NAFLD in offspring, whose vulnerability may persist into adulthood, regardless of maternal obesity and IR.

Notably, an equal-calorie high-SFA diet is more harmful to the liver than a high unsaturated fat diet, which may be caused by the enhanced activity of SIRT3 in hepatocytes. SIRT3 is a manganese superoxide dismutase located mainly in the mitochondria. The enhanced activity of SIRT3 leads to the activation of deacetylation-induced manganese superoxide dismutase, inhibition of AMPK, and mammalian target of rapamycin C1 (mTORC1)–related autophagy inhibition.⁹¹

Both omega-3 (n-3) and omega-6 (n-6) PUFAs are essentials for the composition of cell membranes. n-3 is rich in fish oil while n-6 is rich in vegetable oils. The dietary ratio of n-6 to n-3 (n-6/n-3) is another factor that affects lipid deposition in the liver. In mice fed with equal-calorie equal-fat (35%) high-n-3 and high-n-6 diets, respectively, the high-n-6 diet model showed higher levels of liver steatosis, increased cell apoptosis, and decreased hepatocyte proliferation, which may be related to the increased NF-κ–mediated inflammation in the liver.⁹² Bogl team compared the diets of 10 pairs of identical twins with significant differences in liver fat and found that the higher-liver-fat group has an average of 6.6:1 n-6/n-3, while the lower-liver-fat group has an average of 3.2:1 n-6/n-3.⁹³ These studies suggest that the high proportion of n-6 intake may be a promoting factor of NAFLD independently. Similarly, Okada and colleagues found that ER stress and mitochondrial dysfunction caused by lipid deposition and oxidative stress can be significantly reduced after 6 months of treatment with n-3 in NAFLD patients, which is possibly due to the reduction of serum n-6/n-3. They also partially illustrated the mechanism that n-3 competes with n-6 for lipoxygenase and cyclooxygenase during the process of metabolism. Eicosapentaenoic acid and docosahexaenoic acid, metabolites of n-3, have anti-inflammatory effects, while arachidonic acid, the metabolite of n-6, has strong proinflammatory effects. In addition, this process involves the increased expression of protective genes related to lipid metabolism, including fatty acid binding protein – liver type and peroxiredoxin 6 (PRDX6).⁹⁴

Šmíd et al⁹⁵ reported that NAFLD patients treated with n-3 for 12 months undergo weight loss, liver fat reduction, and a significant decrease in serum liver enzymes including γ-glutamyltransferase (GGT), but liver fibrosis is not effectively improved.⁹⁵ In mouse NAFLD models and in vitro experiments, the results suggested that fish oil may play a role in regulating lipid metabolism by inhibiting adipose cells, and inducing adipose-derived stromal cell and 3T3-L1 preadipocyte proliferation, which results in reduction of blood lipids, liver enzymes, and liver fat.⁹⁶ In terms of mechanism, it may be the expression of neuregulin 4, Ucp1, and Pgc1a mRNA in NAFLD model adipose tissue upregulated by n-3 treatment. In vitro cell experiments have shown that n-3 induces browning of white adipose tissue by promoting peroxisome proliferator–activated receptor gamma pathway–dependent neuregulin 4 production.⁹⁷ Besides, in the NAFLD rat model, supplementation of n-3 can regulate lipid metabolism by improving the bile components and regulating the biliary secretion rhythm.⁹⁸

Hong et al⁹⁹ compared the potential of various components of n-3, including α-linoleic acid, eicosapentaenoic acid, and docosahexaenoic acid, to alleviate NAFLD; the results showed that docosahexaenoic acid is the most effective in reducing liver fatty acid synthesis and has the potential to reverse liver fibrosis.

However, a small number of studies have disputed the role of n-3 in NAFLD treatment. Experimental results of fish oil supplementation for 12 weeks in overweight NAFLD patients showed that fish oil has no significant effect in the reduction of liver fat, liver enzymes, and other indicators.¹⁰⁰ Another similar study found that although supplementation of n-3 for 12 weeks in NAFLD patients and overweight or obese people with hypertriglyceridemia can decrease serum TG levels, it also increased serum aspartate transaminase (AST) and alanine transferase (ALT), markers of liver damage, the mechanism of which remains unclear.¹⁰¹ Nevertheless, taking into account previous studies, these results may only suggest that short-term supplementation of n-3 has no significant therapeutic effect on NAFLD.

In addition, different meta-analyses have drawn positive conclusions on the therapeutic effect of n-3 supplementation in patients with NAFLD. Parker et al’s¹⁰² analysis showed that n-3 supplementation can effectively reduce liver fat and AST levels, and cause a downward trend of ALT. Yang et al¹⁰³ found that the therapeutic effect of n-3 on NAFLD may be only effective in patients with early and mild NAFLD, but not in patients with severe NAFLD. According to the latest meta-analysis, n-3 can not only effectively reduce liver fat, but also significantly improve the levels of serum triglyceride, total cholesterol, HDL, body mass index (BMI), and other parameters, so it has a definite therapeutic effect on NAFLD.¹⁰⁴ However, they also suggested that a more comprehensive analysis is needed due to the heterogeneity of the dose and n-3 treatment duration in the studies included (Figure 2).

FIGURE 2.

Lipids and NAFLD. Dietary lipids have a dualistic influence on NAFLD. On one hand, excess lipids, including FCs and FFAs and so forth, especially SFAs, promote the development of NAFLD. Inflammation can be promoted by activating KCs, liver macrophages (Mφ), and stimulating the release of the proinflammatory factor AA and IL-11. Mitochondrial autophagy is also associated with one of the promoting factors of NAFLD through the macrophage stimulating 1/AMPK pathway. Excess FCs increase the occurrence of mitochondrial dysfunction and cause caspase-2–mediated hepatocyte apoptosis and SMS1/DAG/PKC6/NLRC4 axis–mediated pyroptsis. Besides, inhibition of autophagy in hepatocytes may also be one of the factors that promote NAFLD. On the other hand, increasing dietary intake of n-3 PUFAs can alleviate NAFLD mainly by inhibiting proinflammatory factors and AA, and reducing lipid deposition. Besides, n-3 PUFAs contribute to browning of white adipose tissue. Abbreviations: AA, arachidonic acid; AMPK, AMP-activated protein kinase; EV, extracellular vesicle; FC, free cholesterol; FFA, free fatty acid; JNK, c-Jun N-terminal; LOX, lipoxygenase; LPCAT2, lysophosphatidylcholine acyltransferase 2; MPST, mercaptopyruvate sulfurtransferase; mTORC1, mammalian target of rapamycin complex 1; NRG4, neuregulin 4; PERK, PKR-like endoplasmic reticulum kinase; Pgc1α, peroxisome proliferators–-activated receptor γ coactivator lα; PPARG, peroxisome proliferator– activated receptor gamma; RIP1, receptor-interacting protein 1; SIRT3, sirtuin 3; SMS1, sphingomyelin synthaseUcp1, uncoupling protein 1; VCAM-1, vascular cell adhesion molecule 1.

Dietary changes are currently the most important measures to prevent the development of NAFLD. However, current studies on NAFLD mainly focus on animal models, and there is still a lack of research exploring the most appropriate lipid intake for NAFLD patients. Besides, few studies have explored the impact of an extreme low-fat diet (LFD) on liver metabolism and the specific mechanism promoting NAFLD, which are still necessary to find out through larger samples and more accurate experimental programs.

PROTEIN AND NAFLD

Key point

Dietary protein has been reported to exert beneficial effects on NAFLD patients, including reducing the intrahepatic lipid accumulation and improving IR. However, there is a controversy over the application of high-protein diets in NAFLD patients, which may be attributed to the different effects of animal and plant protein on NAFLD. High-protein diets also have some adverse effects.

The impact of protein and amino acids on NAFLD is little known since most studies focused on carbohydrates and fats. It has been reported that a high-protein diet can reduce hepatic lipid accumulation compared with a normal protein diet in a rat model, with an even greater decrease when combined with restriction of calories.¹⁰⁵ Similar to the findings in rats, supplementing proteins without caloric restriction has been shown to lower intrahepatic fat in human studies, despite the stable BMI,^106–108 while dietary protein insufficiency causes lipid accumulation.¹⁰⁹ The additional metabolic advantages of high-protein diets have been displayed in obese and NAFLD patients,^106,110 with improvement in serum lipid levels and hepatic enzymes. Moreover, sarcopenia, which is an independent risk factor of NAFLD, leads to NASH or even advanced fibrosis as well as CVDs, while high-protein diets can prevent muscle loss.^111,112 A high-protein diet is also more suitable for lean individuals, accounting for 25% of the NAFLD patients.¹¹³ Since it is unnecessary and difficult for them to lose weight or be on a diet, changing the relative macronutrient consumption is more practicable. Notably, an increase in the proportion of protein in a diet usually leads to a decrease in carbohydrate intake. Despite patients with a HFD displaying less lipid deposition when increasing their dietary protein, with no differences in carbohydrate intake,¹⁰⁷ whether the reduction of lipid accumulation is attributed to the high protein or low carbohydrate requires further investigation.

Dietary protein can promote more satiety than carbohydrates and fat, which contributes to more weight loss.^114,115 Mechanically, elevated plasma amino acid contents and gut hormones may exert the satiety effects in NAFLD patients. The branched-chain amino acids have been reported to stimulate satiety signaling through receptors in the duodenums and intestines in rat models.^116–118 Anorexigenic hormones such as GLP-1, cholecystokinin, and peptide YY (PYY) are increased in overweight individuals after high protein intake, which can stimulate the vagal activities controlling food intake.^119,120 Glucagon secretion induced by dietary protein also suppresses DNL.^121,122 What’s more, the catabolism of amino acids increases energy expenditure, which causes oxidation of hepatic lipid and depletion of liver fat in rat models consequently.¹²³ A decrease in the expression of acetyl-CoA carboxylase and FAS, which are genes of hepatic lipogenesis, has also been observed in response to high-protein diets, with a lower serum insulin level and reduced hepatic glucose uptake in rat models as well.¹²⁴ In addition, dietary protein deficiency is associated with mitochondrial dysfunction, which contributes to the pathogenesis of NAFLD. Rats with low-protein diets show decreased expression of antioxidant enzymes and peroxisomal membrane protein 70 (PMP70), with a consequent severe oxidative injury and lipid deposition.¹²⁵ Peroxisomal acyl-CoA oxidase 1 and mitochondrial CPT2, which can promote peroxisomal and mitochondrial β-oxidation of fatty acids, are also increased after supplementing protein in diets¹²⁶ (Figure 3).

FIGURE 3.

Protein and NAFLD. Dietary protein can help treat NAFLD by promoting satiety and inhibiting lipid deposition. Elevated branched-chain amino acids can stimulate satiety signaling. Anorexigenic hormones can also promote vagal activities and inhibit DNL. What’s more, dietary protein can increase β-oxidation of lipids in adipocytes, suppress the expression of lipogenesis genes, and decrease serum insulin levels, all of which inhibit hepatic lipid deposition. Proteins can also increase the expression of antioxidant enzymes in the mitochondria and PMP70 in the peroxisomes to decrease oxidation injury. The expression of peroxisomal ACOX1 and mitochondrial CPT2, which can increase β-oxidation of lipids, is also increased after supplementing dietary protein. Abbreviations: ACC, acetyl-CoA carboxylase; ACOX1, acyl-CoA oxidase 1; CPT, carnitine palmitoyl transferase; DNL, de novo lipogenesis; PMP70, peroxisomal membrane protein 70.

However, conflicting results regarding the effects of dietary protein on NAFLD have been exhibited. Individuals who consume high-protein diets show an increased prevalence of NAFLD¹²⁷ and more severe histological manifestations.¹²⁸ Another population study also reported a higher risk of NAFLD development among individuals consuming diets rich in animal protein, especially in overweight patients.¹²⁹ The controversial findings can be attributed to different sources of protein, as plant protein seems to be beneficial to NAFLD while animal protein has an opposite effect.¹³⁰ NAFLD patients consume more red meat than healthy controls,¹³¹ while red meat is reported to be associated with the development of NAFLD, IR, and CVDs,^132,133 even without high intakes,¹³⁴ mainly because red meat contains a high level of homocysteine.¹³⁵ Bergeron et al¹³⁶ reported that either white meat or red meat leads to higher levels of LDL-C and apoB than nonmeat protein sources, suggesting that vegetable-based food is a better choice for health (Figure 3). But Mariya reported that both animal and plant proteins can decrease intrahepatic lipid accumulation to a similar content in patients with NAFLD and T2DM.¹³⁰ Therefore, the effect of different sources of protein on NAFLD requires further investigation. In addition, the results may be attributed to more than the effect of protein, as red meat and vegetables contain other compounds. Red meat contains cholesterol, heme iron, and nitrates, which are related to the increased mortality from chronic liver diseases,^137,138 while vegetables are rich in phytonutrients and fibers.

Although dietary proteins may be beneficial in decreasing hepatic lipid accumulation, their adverse effects should be taken into consideration. High protein intake can lead to glomerular sclerosis as well as renal malfunction in patients vulnerable to kidney diseases,¹³⁹ while it has been reported that the glomerular filtration rate does not decrease when consuming vegetable protein instead of animal protein.^140,141 Furthermore, a higher proportion of protein intake is associated with an increased risk of CVDs as mentioned above.¹⁴² Current evidence supports that more plant protein can decrease the risks of CVD than animal-based protein, while meat that is unprocessed and low in saturated fat is beneficial as well.¹⁴³ The increased levels of nitrosamine and heterocyclic amine in feces resulting from high-protein diets are harmful to the environment of the colon, leading to a higher risk of colorectal cancer.¹⁴⁴ In addition, research in rats showed that a diet high in branched-chain amino acids and fat could lead to the accumulation of succinyl and propionyl-CoA, which can impair insulin sensitivity and is associated with a higher risk of diabetes consequently.¹⁴⁵ A high intake of amino acids can inhibit insulin signaling through the activation of mTOR signaling as well.^146,147 Hence, proper doses of daily protein intake and their sources need to be detected to balance the benefits and risks.

VITAMIN AND NAFLD

Key points

Both vitamin E (VE) and vitamin D (VD) can improve the biochemical and histological manifestations of NAFLD patients. The vitamin D receptor (VDR) is also a promising target for treating NAFLD. However, more research on the effects of vitamins on NAFLD is required to support their application.

Chronic liver diseases are reported to be associated with the deficiency of vitamins,¹⁴⁸ in which context VE and VD are most investigated. VE, which has an antioxidant property, can improve both biochemical and histological manifestations, including aminotransferases, ballooning, inflammation, and steatosis, in NASH patients without T2DM.¹⁴⁹ However, in NAFLD patients with T2DM, the results are conflicting. Bril et al¹⁵⁰ reported that the application of VE alone did not improve liver fibrosis, while a combination of VE and pioglitazone exhibited improvement in histological outcomes. Eduardo reported that transplant-free survival and liver decompensation were improved by the application of VE in NASH patients both with and without T2DM.¹⁵¹ Studies on rats uncovered the potential mechanisms of VE on NAFLD treatment, showing that VE can alleviate oxidative stress and prevent NAFLD-driven HCC by downregulating inducible nitric oxide synthase and NADPH oxidase.¹⁵² Furthermore, VE lowers the expression of TGF-β, which plays a critical part in the progression of liver fibrosis.¹⁵³ VE is also reported to improve lipid and glucose metabolism via activating the nuclear factor erythroid 2-related factor 2/carboxylesterase 1 pathway¹⁵⁴ and enhance insulin sensitivity by inducing the expression of adiponectin via activation of PPAR.¹⁵⁵ Regardless of its effects on NAFLD improvement, VE has been reported to be associated with an increasing risk of prostate cancer¹⁵⁶ and CVDs.¹⁵⁷ Considering the controversial results of VE and its side effects, more investigation is required before its clinical application.

What’s more, there is a deficiency of VD in NAFLD patients compared with healthy controls,¹⁵⁸ associated with the severity of NAFLD’s histological manifestations.¹⁵⁹ Supplementation of VD can improve IR and the levels of serum liver enzymes.¹⁶⁰ In NAFLD rat models, deficiency of VD is associated with upregulation of TLR4-mediated inflammatory signaling in NAFLD rat models, while supplementing VD can reverse the activation of the TLR4 pathway¹⁶¹ and reduce the activation of HSCs as well, exerting an antifibrogenic effect.¹⁶⁰ In addition, VD deficiency can downregulate defensins of intestinal Paneth cells, which damages the intestinal innate immunity and causes IR and NASH consequently.⁷⁵

Although several studies have shown that supplementation of VD yielded no improvement in liver steatosis in NASH patients,^162,163 the VDR is a promising target for treating NAFLD and is widely researched in rat models. Macrophages exhibit the highest expression of VDR mRNA among macrophages, hepatocytes, HSCs, and cholangiocytes.¹⁶⁴ It has been reported that activation of VDRs on hepatic macrophages can improve insulin sensitivity and relieve liver steatosis, therefore exhibiting anti-inflammatory and antidiabetic effects.¹⁶⁵ Activation of VDRs on HSCs can antagonize TGF-β/SMAD signaling, decreasing the expression of several profibrogenic genes consequently.¹⁶⁶ However, although hepatocytes express low levels of VDRs, activation of VDRs on hepatocytes causes lipid accumulation, which suggests that drugs are required to target specific cells.

Furthermore, the impaired metabolism and increased vitamin A accumulation in the liver are associated with the progression of NAFLD, with elevated expression of inflammatory markers such as c-c motif chemokine ligand 2 and IL-6, as well as markers of liver fibrosis such as Timp1 and TGF-β.¹⁶⁷ Supplementation of vitamin C also showed improvement in liver enzymes and glucose metabolism in NAFLD patients.¹⁶⁸ Therefore, the roles of different types of vitamins in NAFLD progression deserve more investigation (Figure 4).

FIGURE 4.

Vitamin and NAFLD. Both vitamin E and vitamin D contribute to the regression of NAFLD. VE can induce the expression of adiponectin by activation of PPAR, enhancing insulin sensitivity consequently. Moreover, VE can inhibit lipogenesis and gluconeogenesis through the Nrf2/CES1 pathway, and alleviate oxidative stress via downregulation of iNOS and NADPH oxidase. It can lower the expression of TGF-β as well, which prevents liver fibrosis. Vitamin D supplementation can upregulate defensins associated with intestinal innate immunity, preventing IR consequently. VD can also suppress TLR4 on KCs and hepatocytes, which can downregulate NF-κB and is associated with inflammation. Activation of the VDRs on KCs and HSCs can improve insulin sensitivity and alleviate liver fibrosis, respectively. Abbreviations: CES1, carboxylesterase 1; iNOS, inducible nitric oxide synthase; Nrf2, nuclear factor erythroid 2-related factor 2; PPAR, peroxisome proliferation–activated receptor; TLR4, Toll-like receptor 4; VDR, vitamin D receptor.

DIET AND NAFLD

Key points

Changes in dietary patterns are promising treatments for NAFLD. The Mediterranean diet (MD) has already been recommended by several guidelines. Effects of other popular diets, including intermittent fasting (IF), calorie-restricted diet (CRD), ketogenic diet (KD), the dietary approach to stop hypertension (DASH) diet, low-carbohydrate diet (LCD), and LFD, are also widely studied in NAFLD patients.

A large number of clinical trials have tested the effects of dietary modifications in the regression of NAFLD. Here we will briefly discuss several popular diets, including IF, CRDs, KD, MDs, and DASH diets, and compare the effects of low-fat and LCDs on NAFLD (detailed in Table 1). Other diets such as plant-based diets and paleolithic diets have also been widely investigated, but until now there is little high-quality evidence to support any diet. We also summarize macronutrients’ distribution in the following diets (Table 2).

TABLE 1.

Clinical trials testing effects of different dietary approaches in NAFLD

References	Type of the study and duration	No. patients and their features	Dietary intervention	Mean weight loss	Changes in liver enzymes	Changes in liver fat, steatosis and stiffness	Other major outcomes
Clinical trials comparing diverse dietary approaches with control (usual diets or exercise)
Oshakbayev et al169	RCT 24 wk	N = 80 Patients with severe NASH and metabolic syndrome including T2DM	Main group: fast weight loss (CRD and walking) vs. drug treatment	In the main group, the weight loss varied from 7–16 kg (p < 0.0001) and achieved in the first 8–10 wk. In the drug group, the weight loss was 2–5 kg and in the end of the treatment.	ALT (p = 0.02) and AST (p < 0.0001) significantly reduced in the main group compared with the drug group. These levels in the control group did not achieve normality.	In the histological scoring system in the main vs. control group there was significant improvement in liver inflammation (p = 0.0019), hepatocellular ballooning (p = 0.0025) and steatosis/fibrosis (p = 0.0056 and 0.0043, respectively).	Glucose metabolism (2-h OGTT: p = 0.012; HbA1c: p < 0.0001), insulin (p < 0.0001) and HOMA-IR (p = 0.002) all improved significantly in the main group compared with the drug group.
Asghari et al170	RCT 12 wk	N = 60 BMI:25–35 kg/m2	CRD vs. control	The weight reduced significantly in the CRD group (−4.06 kg) compared with the control.	ALT and AST significantly reduced in the CRD group compared with the control (p < 0.05).	No significant changes were observed in patients’ liver steatosis grades either within or between the 2 groups.	LDL-C and HDL-C levels did not change significantly between the 2 groups. Antioxidant levels did not improve significantly between the 2 groups.
Johari et al171	RCT 8 wk	N = 43 NAFLD patients	ADF vs. control	There was a significant reduction in mean weight in the intervention group (p  =  0.003) but not within the control group (p = 0.86).	There was a significant reduction of ALT (p = 0.001) and AST (p = 0.004) in the intervention group but not within the control group (p >  0.32). ALT reduced more significantly in the intervention group compared with the control (p  =  0.02). But no difference was reported between the 2 groups with AST level (p  =  0.34).	There was significant reduction of liver steatosis grading (p  =  0.001) and fibrosis (SWE scores) (p  =  0.001) in the intervention group but not within the control group (p  >  0.30). In between-group analysis, a statistically significant reduction was observed in the intervention vs. control group, for liver steatosis and SWE scores (p  =  0.01).	There was no reduction of all lipid parameters (total cholesterol, LDL, HDL and TG; all p  >  0.28) in both groups. Fasting blood sugar reduced significantly in both groups (p = 0.006 vs. 0.08), but no significant changes were observed between the 2 groups (p = 0.34).
Cai et al172	RCT 12 wk	N = 271 NAFLD patients	ADF vs. TRF vs. control	−4.04 kg vs. −3.25 kg vs. −1.85 kg (p-value not specified) The body weight decreased significantly in the ADF and TRF group compared with the control group. No differences between the ADF and CRF groups are observed.	Not evaluated.	Liver stiffness did not improve after the intervention in 3 groups,	TC was significantly decreased in the ADF group compared with the TRF group and control (p-value not specified). Both the ADF and TRF group achieved a significant reduction in serum triglycerides (p <  0.001) after 12 wk of intervention.
Varkaneh et al173	RCT 12 wk	N = 52 Obese or overweight patients with NAFLD	TRF (16 h fasting/8 h feeding daily) plus a low-sugar diet vs. control	There was a significant reduction of weight (p < 0.001) in the intervention group but not within the control group (p  = 0.064). And significant between-group differences in body weight loss were observed (p = 0.029).	Significant decreases in serum levels of ALT (p = 0.013) and AST (p = 0.010) were observed for the intervention group compared with control.	Fibrosis score (p = 0.009) and steatosis score/CAP (p < 0.001) were significantly reduced in the intervention group compared with control.	The intervention group experienced significant reductions in TG (p<0.001), TC (p = 0.001), and LDL-C (p = 0.014) compared with the control. But no changes in HDL-C levels were observed (p = 0.159).
Ezpeleta et al174	RCT 3 mo	N = 80 With obesity and NAFLD	ADF with exercise vs. ADF vs. exercise vs. control	The combination intervention produces superior reductions in body weight vs. exercise alone and control (p < 0.01), but not fasting alone (p = 0.43).	Change in serum ALT in the combination group was significantly different compared with the control group (p = 0.01), but not significantly different compared with the ADF group (p = 0.48) or exercise group (p = 0.09). Change in serum AST did not significantly differ among the 4 groups.	IHTG content was significantly reduced in the combination group compared with the exercise group (p = 0.02) and the control group (p < 0.01). However, the reduction in IHTG content in the combination group was not significantly different compared with the ADF group (p = 0.05).	The combination intervention improved insulin resistance and insulin sensitivity vs. controls, but not vs. exercise alone or fasting alone.
Misciagna et al175	RCT 6 mo	N = 98 With moderate or severe NAFLD	MD vs. control	No significant changes of BMI were observed between the 2 groups.	Lower levels of GGT were observed in the MD group but not in the control.	Fatty liver index (FLI) and NAFLD score measure by liver ultrasonography reduced significantly in the MD group (p-value not specified). A negative interaction between time and MD on the NAFLD score was observed.	—
Zade et al176	RCT 8 wk	N = 60 overweight and obese patients aged with NAFLD diagnosed by ultrasonography scan	DASH diet vs. control	−3.8 kg vs. −2.3 kg (p = 0.006)	ALT (p = 0.02) and ALP (p = 0.001) significantly reduced in the DASH diet group compared with the control.	Both groups showed significant reduction in percentage of NAFLD grades (p < 0.001) and the DASH group had more significant changes (p = 0.003).	Insulin levels (p = 0.01), homeostasis model of assessment-estimated insulin resistance (HOMA-IR) (p = 0.01) significantly decreased and quantitative insulin sensitivity check index (QUICKI) (p = 0.004) significantly increased. Compared with the control diet, the DASH diet has resulted in significant reductions in serum triglycerides (p = 0.04) and total-/HDL-cholesterol ratio (p = 0.01).
Schwimmer et al177	RCT 8 wk	N = 40 adolescent boys with histologically diagnosed NAFLD and evidence of active disease (hepatic steatosis > 10% and ALT level ≥45 U/L)	Diets low in free sugar vs. control	Not evaluated	ALT (p < 0.001), AST (p = 0.005) and GGT (p < 0.001) were significantly lower in the intervention group than the control.	The mean decrease in hepatic steatosis was 8% vs. 1% (p < 0.001)	Adherence to the diet was higher in the intervention diet group. There were no significant differences in glucose, insulin, homeostasis model assessment for insulin resistance, TG, LDL-C, or HDL-C.
Domínguez-Coello et al178	RCT 24 wk	N = 239 BMI: 29-40.99 kg/m2	Low-fructose diet vs. control	Larger decreases in waist circumference, waist circumference/height ratio were seen in the low-fructose diet group compared with the control.	Not evaluated.	Not evaluated.	Larger decreases of fasting blood glucose were seen in the intervention group compared with the control. But the intervention did not reduce insulin resistance.
Cohen et al179	RCT 8 wk	N = 29 adolescent boys with NAFLD	Diets low in free sugar vs. control	−1.4 kg vs. 0.6 kg (p-value not specified)	ALT, AST and GGT all significantly reduced in the intervention group compared with the control (all p < 0.05).	The intervention group experienced a greater decrease in hepatic fat measured by magnetic resonance imaging-proton density fat fraction (MRI-PDFF) compared with the control (p < 0.05). Hepatic DNL was significantly decreased in the intervention group (from 34.6% to 24.1%) vs. the control group (33.9% to 34.6%).	Fasting insulin, fasting glucose, TGs, TC and LDL-C all significantly reduced in the intervention group compared with the control (all p < 0.05). Percentage change in DNL during the intervention correlated significantly with changes in free sugar intake (r = 0.48, p = 0.011), insulin (r = 0.40, p = 0.047), and alanine aminotransferase (ALT) (r = 0.39, p = 0.049), but not hepatic fat (r = 0.13, p = 0.532).
Schmidt et al180	RCT 12 wk	N = 105 Latino youth	Diets low in free sugar intake vs. control	No significant changes in weight were observed in both groups over the intervention period (p = 0;48)	There were no differential changes in AST (p = 0.71) and ALT (p = 0.68) between the 2 groups.	Changes in liver fat fraction (p = 0.50) and liver fibrosis measured by MRE (p = 0.31) did not differ between the intervention group and the control.	Participants with whole-body fat mass (FM) reduction, irrespective of randomization, had significant reductions in liver fat compared with participants without FM reduction (p < 0.001).
Clinical trials comparing MD with LFD
Ryan et al181	RCT 6 wk	N = 12 nondiabetic individuals with biopsy-proven NAFLD	MD vs. LFD	The weight loss in both diets (MD: −1.0 kg vs. LFD: 2.4 kg) was not significant and the difference in weight change between the groups was also not significant (p = 0.1).	ALT and AST did not change appreciably with either diet.	Intrahepatic lipid measured by MRS reduced significantly in MD group compared with the LFD (p = 0.012).	Both the HOMA-IR and the circulating insulin concentration declined significantly (p = 0.008 and p = 0.003) in the MD group but there was no significant improvement in the LFD group.
Properzi et al182	RCT 12 wk	N = 51 With NAFLD	Two isocaloric diets: MD vs. LFD	−2.1 kg vs. −1.6 kg (p = 0.52)	ALT and GGT improved significantly in both groups. But no significant differences were observed between the 2 groups.	Hepatic steatosis measured by MRS had reduced significantly in both groups (p < 0.01), but there was no difference in liver fat reduction between the groups (p = 0.32).	Within-group improvements in Framingham Risk Score, TC, TG, and HbA1c were observed in MD (all p < 0.05), but not in LFD. Adherence was higher for MD compared with LFD (p = 0.048).
Biolato et al183	Crossover clinical trail 48 wk	N = 20 with biopsy-verified NAFLD and increased transaminases.	MD vs. LFD: 16 wk of an MD (W1–W16), 16 wk of a free washout diet (W17–W32) and 16 wk of a LFD (W33–W48).	A significant reduction of weight loss was observed after 16 wk of MD (p = 0.003). There was a significant reduction in mean body weight after 16 wk of MD compared wit LFD (p = 0.004).	A significant reduction of ALT (p = 0.0001) and AST (p = 0.001) was observed after 16 wk of MD. There was a significant reduction in ALT after 16 wk of MD compared with LFD (p = 0.04).	Not evaluated.	No significant changes in intestinal permeability were observed at the end of the MD period or the LFD period.
Gepner et al184	RCT 18 mo	N = 278 With abdominal obesity or dyslipidemia	Mediterranean and low-carbohydrate diet vs. LFD for 6 mo. Physical activity vs. no exercise for 12 mo.	Not evaluated.	Decreased hepatic fat content was associated with reductions in serum GGT and ALT (p  < 0.05). Differences between the 2 groups were not specified.	Mediterranean and low-carbohydrate diet induced a greater hepatic fat content decrease (p  =  0.036) and greater improvements in cardiometabolic risk parameters (p  < 0.05) than LFD.	—
Ristic-Medic et al185	RCT 3 mo	N = 27 BMI:25−35 kg/m2	MD vs. LFD	−9.23 kg vs. −9.71 kg (p = 0.342)	Both diets lowered liver enzymes, but no significant differences were observed between the 2 groups.	Both diets improved FLI and hepatic steatosis index. FLI in MD reduced more significantly (p = 0.022).	Participants on MD had higher levels of HDL-C and monounsaturated and n-3 docosahexaenoic acids in serum phospholipids and lower levels of saturated fatty acids, TG, and TG/HDL ratio when compared with participants on LFD.
Akbulut et al186	RCT 12 wk	N = 70 Children with NAFLD	MD vs. LFD	2.0 kg vs. 2.7 kg (p = 0.006)	ALT and AST levels both decreased significantly in both groups, but no differences between the 2 groups were observed.	Marked decreases in the degree of steatosis and liver stiffness were observed in both groups compared with the baseline. But there were no significant differences between the 2 groups.	Significant reductions in HOMA-IR were obtained in the MD group (18%) compared with the LFD group (12%) (p =  0.010).
George et al2022187	RCT 12 wk	N = 42 With NAFLD	MD vs. LFD	The weight was not significantly different from preintervention to postintervention in both groups.	ALT, AST, and GGT were not statistically significant following the MD (p = 0.32, 0.29 and 0.26, respectively), but significant reductions were noted in the LFD group (p = 0.009, 0.040 and 0.029, respectively). Changes between groups were significant (p-value not specified).	There was a significant improvement of intrahepatic lipids in the low-fat group (p = 0.02) but not in the MD group (p = 0.07). No significant differences for intrahepatic lipids (p = 0.865) and liver stiffness (p = 0.58) measurement were observed between the 2 groups.	No significant differences of HOMA-IR were observed between the 2 groups (p = 0.58).
Yurtdaş et al188	RCT 12 wk	N = 44 Adolescents with NAFLD	MD vs. LFD	The mean body weight of participants significantly decreased in both groups (p < 0.001), but weight loss was not statistically different between groups (p = 0.532).	ALT, AST, and GGT decreased significantly in both groups with no significant differences between groups except for AST, which reduced more in the MD group than in the low-fat diet group (p = 0.039).	A statistically significant decrease was found in the grade of hepatic steatosis after the intervention in both diet groups (p  <  0.05). There was no significant difference between the groups in terms of reduction in the degree of liver fat (p = 0.875).	Total antioxidant status, paraoxanase-1, and GSH-Px levels increased in the MD (p  <  0.05) group but did not change in the LFD group (p  >  0.05) compared with baseline.
Clinical trials comparing LCD with LFD
Ryan et al189	RCT 16 wk	N = 52 With obesity	Two hypocaloric diets: LCD vs. LFD	Both groups experienced a significant decrease in weight (7.0 kg vs. 5.7 kg) (both p < 0.001). But it did not differ between the 2 groups.	ALT significantly reduced in the LCD group (p < 0.04). ALT changes correlated with improvement in insulin sensitivity (p = 0.04). Individuals with ALT concentrations above the proposed upper limits experienced a significant decline in ALT, unlike those with lower ALT levels.	Not evaluated.	SSPG (p < 0.04) and circulating insulin (p < 0.01) significantly reduced in the LCD group but not in the LFD group.
Hernández et al190	RCT 6 mo	N = 59 With NAFLD	LCD vs. LFD	−5.7% vs. −5.5% There was no significant difference in weight loss between the 2 groups.	The LCD group showed more reduction in ALT (41 vs. 33.3%) and AST (31.7 vs. 22.4%) compared with the low-fat group, but there were no significant statistical differences. Changes in weight were positively related with changes in ALT and AST, irrespective of the type of diet.	Not evaluated.	—
Haufe et al191	RCT 6 mo	N = 170 Overweight and obese individuals	LCD vs. LFD	The weight decreased significantly in the LCD group compared with the LFD group (p <0.001).	ALT and AST did not change significantly after the intervention in both groups.	Subjects with high baseline intrahepatic lipids (>5.56%) lost about 7-fold more intrahepatic lipids compared with those with low baseline values (<5.56%) irrespective of diet composition.	TC and LDL-C decreased more in the LFD group compared with the LCD group (p-value not specified).
Goss 2020192	RCT 8 wk	N = 32 Children/adolescents with obesity and NAFLD	LCD vs. LFD	−2.4% vs. −0.4% (p = 0.06)	ALT and AST decreased significantly only in the LCD group (p < 0.05), but no significant differences were observed between groups (p = 0.15 and 0.43, respectively).	Changes in hepatic lipid accumulation did not differ between the 2 groups, but declined significantly (p  <  0.001) only within the LCD group.	Significantly greater decreases in insulin resistance (HOMA-IR, p < 0.05), abdominal fat mass (p  <  0.01), and body fat mass (p  < 0 .01) were found in the LCD group compared with the LFD group.
Hansen et al193	6 mo	N = 165 T2DM	LCD vs. LFD (Both are calorie-unrestricted.)	Participants on the LCD lost more weight than LFD.	Not evaluated.	No statistically significant between-group changes were detected in the assessment of NAFLD by liver biopsies.	Participants on the LCD had greater improvements in HbA1c but less favorable changes in LDL-C. Both groups had higher HDL-C and lower TG at 6 mo but no significant changes were observed between the 2 groups. Changes were not sustained at the 9-month follow-up.
Clinical trials comparing other dietary approaches
Browning et al194	RCT 2 wk	N = 18 BMI: 35 ± 7 kg/m2	CRD vs. LCD	−4.0 kg vs. −4.6 kg (p = 0.0363)	ALT did not change significantly in either group. AST reduced significantly after 2 wk of weight loss (p < 0.001), but no differences were observed between the 2 groups.	Not evaluated.	Liver triglycerides decreased significantly with weight loss (p < 0.001) but decreased significantly more (p = 0.008) in the LCD group than in the CRD group. Dietary fat (p = 0.004), carbohydrate (p = 0.008), post-treatment plasma ketones (p = 0.006), and respiratory quotient (p < 0.001) were related to the reduction in liver triglycerides.
Skytte et al195	RCT 6 wk	N = 30 T2DM	Low-carbohydrate, high-protein diet vs. conventional diabetes diet	−1.4 kg vs. −0.8 kg (p = 0.071)	Not evaluated.	The low-carbohydrate, high-protein diet had more significant reduction in liver fat fraction (p < 0.01) and pancreatic fat fraction (p < 0.05) compared with the conventional diabetes diet.	The low-carbohydrate, high-protein diet had more significant reduction in HbA1c (p < 0.001) and fasting glucose (p <  0.05). TC reduced significantly in the low-carbohydrate, high-protein group compared with the conventional group (p < 0.05).
Thomsen et al196	6 wk	N = 72 T2DM, BMI >25 kg/m2	Low-carbohydrate, high-protein diet vs. conventional diabetes diet	−5.8 kg vs. −5.8 kg (p = 0.83)	Not evaluated.	Hepatic fat content was reduced significantly by 64% and 51% after the low-carbohydrate, high-protein diet, and conventional diabetes diet, respectively, with the difference between groups reaching borderline significance (p =  0.051).	The low-carbohydrate, high-protein diet had more significant reduction in HbA1c (p = 0.018) and diurnal mean glucose (p <  0.001). Fasting glucose, insulin, HOMA-IR, and cholesterol concentrations (total, LDL, and HDL) were reduced significantly and similarly by both diets.
Cunha et al197	RCT 2 mo	N = 39 BMI >30 kg/m2	low-calorie KD vs. CRD	−9.59 kg vs. −1.87 kg (p < 0.001)	AST decreased in the low-calorie KD group (p < 0.05) but not in the CRD group.	Liver fat fraction reduced more significantly in the low-calorie KD group (p < 0.005). No differences in mean liver stiffness values were seen between groups.	Mean reductions in visceral adipose tissue were significantly more pronounced in the low-calorie KD group than CRD group (p < 0.05).
Lin et al198	RCT 12 wk	N = 132 BMI ≥ 30 kg/m2	CRD (450 kcal/d) vs. CRD (800 kcal/d)	−8.37 kg vs. −8.42 kg (p = 0.092)	Not evaluated.	The improvement rate of NAFLD was 41.5% vs. 50.0% (p-value not specified).	Fat mass, blood pressure, triglycerides, and blood glucose were statistically improved from baseline but not between the 2 groups. There is no additional benefit in prescribing the more restrictive diet intervention.
Crabtree et al199	RCT 6 wk	N = 37 Overweight and obese	KD vs. KD + KS (KD with an exogenous ketone salt supplement) vs. KD + PL (KD with placebo)	Mean weight loss at 6 wk was significant (p < 0.001) in all 3 groups. There was no significant difference in weight loss among the KD + KS, KD + PL, and KD groups, representing decreases of 8.1%, 8.5%, and 6.7% of the initial body mass, respectively (p > 0.05).	There were no differences in AST and ALT, from baseline to postintervention.	There was a significant (p = 0.004) decrease in liver fat measured by MRI and hepatic steatosis index postintervention, but no differences among the 3 groups (p > 0.05).	The change in liver fat is highly associated with baseline liver fat rather than weight loss or other changes in anthropometric or circulating markers.
Parry et al200	Crossover clinical trials 15 wk	N = 16 BMI: 25–30 kg/m2	Two eucaloric diets enriched in saturated fat (SFA) vs. free sugars (SUGAR) (4 weeks of SFA or SUGAR, 7 wk of washout period and 4 wk of the alternate diet)	Body weight significantly increased after consumption of the SFA (p < 0.05) but not the SUGAR diet.	No significant changes of ALT were observed in both groups after the intervention.	IHTG significantly increased following the SFA diet (p < 0.05), while it remained unchanged in response to the SUGAR diet.	Significant decreases in plasma TC and HDL-C were observed in the SUGAR group compared with the SFA group (p < 0.05).

Abbreviations: ADF, alternate-day fasting; ALP, alkaline phosphatase; ALT, alanine transferase; AST, aspartate transaminase; BMI, body mass index; CRD, calorie-restricted diet; DASH, dietary approach to stop hypertension; DNL, de novo lipogenesis; GGT, γ-glutamyltransferase; HbA1c, glycosylated hemoglobin type A1c; HOMA-IR, homeostasis model assessment – insulin resistance; IHTG, intrahepatic triglyceride; KD, ketogenic diet; LCD, low-carbohydrate diet; LFD, low-fat diet; MD, Mediterranean diet; MRE, magnetic resonance elastography; MRS, magnetic resonance spectroscopy; OGTT, oral glucose tolerance test; RCT, randomized controlled trial; SFA, saturated fatty acid; SWE, shear-wave elastography; T2DM, type 2 diabetes mellitus; TC, total cholesterol; TG, triglyceride; TRF, time-restricted fasting.

TABLE 2.

Macronutrients of diet patterns

Diet	Carbohydrate (%)	Fat (%)	Protein (%)
KD	<10	>70–90	10–15
MD	35–45	30–40	15–20
DASH diet	52–55	30	16–18
LFD	60	25	15
LCD	40	45	15

Abbreviations: DASH, dietary approach to stop hypertension; KD, ketogenic diet; LCD, low-carbohydrate diet; LFD, low-fat diet; MD, Mediterranean diet.

Calorie-restricted diet

A CRD, regardless of its macronutrient composition, can induce weight loss, which is associated with decreased total body fat and intrahepatic lipid depletion consequently.^191,201 Indeed, Promrat et al²⁰² reported an almost linear dose response between weight loss and resolution of hepatic steatosis, assessed by liver biopsy in NAFLD individuals. Moreover, the negative energy balance is related to reduced serum ALT and AST levels and metabolic benefits, including improved β-cell function and insulin sensitivity.²⁰³ NAFLD patients following CRD have also shown NASH and fibrosis regression histologically.²⁰⁴ However, Asghari et al¹⁷⁰ reported that no changes were observed in the total antioxidant capacity, erythrocyte superoxide dismutase (SOD), and glutathione peroxidase (GSH-Px) activities between the hypocaloric group and the control group after 12 weeks of intervention, indicating that CRD cannot improve the antioxidant abilities of NAFLD patients, which is related to several chronic diseases. Further studies are warranted to target the effects of long-term intervention in NAFLD patients.

Unfortunately, significant weight loss can be difficult to maintain for many NAFLD patients in the long term. The average weight loss is only about 3%–5% from baseline weight after 2–4 years.²⁰⁵ Chronic nutrient deficiency, which is related to liver dysregulation, is another concern for the wide application of CRD in NAFLD patients. Individuals with eating disorders are more susceptible to both acute and chronic liver injury.²⁰⁶ Kwashiorkor, which is a severe consequence of protein deficiency, is also associated with fatty liver and hepatomegaly.²⁰⁷ Therefore, changing mealtimes or the composition of macronutrients in the diet rather than restricting calorie intake alone may be a better alternative.

Intermittent fasting and time-restricted fasting

IF has become increasingly popular in the past few decades. There are several patterns of IF, including alternate-day fasting, every-other-day fasting, and 5:2 fasting, which means consuming extremely low calories on 2 discontinuous days followed by normal eating patterns on 5 unrestricted days. Another widely used pattern of IF is time-restricted fasting (TRF), which refers to feeding within a certain period of time (usually 4–8 h) and fasting for the rest of the day. TRF does not restrict the caloric intake in the feeding time, which seems to be easier than other IF patterns for NAFLD patients to adopt. However, some researchers believe that TRF is not synonymous with IF.²⁰⁸

IF has been shown to be equally effective as traditional caloric restriction diets in weight loss and improving cardiometabolic outcomes.²⁰⁹ An observational cohort study found that the fatty liver index notably reduced after a mean period of 8.5 days of fasting, with a more significant decrease in individuals with diabetes. The improvement of fatty liver index correlated with the duration of fasting and the magnitude of weight loss.²¹⁰ TRF has also been demonstrated to improve insulin sensitivity, blood pressure, and oxidative stress in prediabetes patients without weight loss. However, another study showed that no changes were observed in the fasting insulin level and liver stiffness between the fasting intervention group and the control.¹⁷² Moreover, Ezpeleta et al¹⁷⁴ reported that a combination of IF and exercise can reduce intrahepatic triglyceride (IHTG) content, body weight, and fat mass and improve IR in NAFLD individuals. A combination intervention showed superior improvement versus exercise alone but not fasting alone, indicating that IF is the primary intervening factor benefiting NAFLD patients. Furthermore, an 8-week RCT showed that alternate fasting resulted in a greater improvement in steatosis and fibrosis as measured by ultrasound and elastography.^13–107 However, current evidence concerning the effects of TRE on hepatic fat is still weak.

Research on mice provides more insight into the potential mechanisms of fasting on NAFLD treatment. When glycogen stored in the liver is depleted after fasting for over 12 hours, the metabolic switch, which refers to the shift from utilization of glucose from glycogenolysis to fatty acids and ketones, will occur.²¹¹ Consequently, the increase of adipose tissue lipolysis results in weight loss, improved insulin sensitivity, and decreased risks of CVDs due to lower contents of TC and LDL-C.²¹² Furthermore, IF can elevate the expression of a variety of genes involved in lipolysis and fatty acid oxidation in NAFLD mouse models induced by high-fat, high-sucrose diets (HFHSD), accompanied by decreased accumulation of lipid in the liver.²¹³ Hatori et al²¹⁴ also reported that TRF improved oscillations of the circadian clock, which is related to obesity and diabetes.

However, there are safety concerns about applying IF for NAFLD treatment. IF may aggravate starvation effects and even trigger binge eating in unrestricted periods. Some researchers are also worried that IF may lead to hypoglycemia in patients with diabetes despite only one clinical trial of IF in NAFLD patients reporting it until now.²¹⁵ Although no significant adverse effects were observed in the current research, it is difficult to maintain the daily feeding window because of family and social life, which is another limitation of applying IF.²⁰⁸

The effects of IF in NAFLD prevention and treatment still require further investigation. In most of the clinical trials, the calorie intake of the control group is not the same as the IF group, so we cannot distinguish whether fasting or weight loss contributes to the beneficial effects. The duration of the research is also too short, so the long-term effects and adherence of IF cannot be observed. However, IF provides an easier way for some patients to lose weight than calorie-restricted diets.

Low and very LCD (KD)

KDs are dietary patterns that restrict carbohydrate intake to below 20–50 g per day, accounting for less than 10% of the daily caloric intake. KD has been proven to be effective in losing weight and reducing visceral fat.²¹⁶ A few studies tested KD as a treatment for NAFLD patients. Tendler et al²¹⁷ reported an improvement in hepatic inflammation, steatosis, and fibrosis in 5 NAFLD patients following KDs. Luukkonen et al²¹⁸ also reported that KD could reduce accumulation of IHTGs and improve IR.

Carbohydrate restriction can improve insulin sensitivity, which results in more oxidation of fatty acids and less lipogenesis in overweight and obese individuals.²⁰¹ What’s more, KDs can induce the production of ketone bodies, which can increase hepatic glutathione peroxidase 1 and decrease oxidized glutathione, as well as elevating mitochondrial transcription factor A (TFAM) in NAFLD rat models, which is associated with mitochondrial biogenesis.²¹⁹ Furthermore, NAFLD patients have been reported to display impaired capacity of ketogenesis and elevated acetyl-CoA oxidation in the TCA cycle alternatively, associated with increased oxidative metabolism and impaired glucose homeostasis.²²⁰ Therefore, stimulation of ketogenesis is believed to be a potential target in NAFLD.

However, some researchers argue that it is hard to determine whether weight loss or carbohydrate restriction contributes to the beneficial effects. Pérez-Guisado et al²²¹ reported an improvement in serum liver enzyme levels and a reduction in steatosis degrees in 14 overweight NAFLD individuals who applied KDs, with no restriction in caloric intake. Moreover, patients who apply carbohydrate-restricted diets exhibit a more significant decrease in liver triglyceride compared with calorie-restricted diets.¹⁹⁴ However, in 22 obese individuals who received a calorie-restricted diet, the reduction of hepatic triglyceride was similar in both the low-carbohydrate group and the high-carbohydrate group.²²² A meta-analysis suggested that both macronutrient distribution and calorie restriction are crucial in the regression of NAFLD.²²³ Conflicting results indicate that more large-scale research is required to clarify the roles of carbohydrate restriction in NAFLD.

Furthermore, it is controversial to apply KDs in NAFLD patients since they are also high in fat. TC and LDL-C values change unfavorably in patients with low carbohydrate intake, which is associated with a higher risk of CVDs.²²⁴ It is also reported that KD can lead to hypotension, electrolyte disturbance, and symptomatic gallstones.^225,226 Moreover, individuals following KDs show numerous discomforts such as vomiting, constipation, fatigue, and irritability,²¹² so it is difficult for them to adhere for a long time. Hence, the side effects of KDs should be taken into consideration and more long-term investigation is required before the wide range of applications. However, considering that KD restricts sugary foods in the diet and helps lose weight, it can provide an opportunity for obese individuals to achieve normal BMIs.²²³

Mediterranean diet

The MD, which has been considered as a healthy diet worldwide, is recommended by European Society for Parenteral and Enteral Nutrition guidelines and European Association for the Study of the Liver–European Association for the Study of Diabetes–European Association for the Study of Obesity (EASL–EASD–EASO) Clinical Practice Guidelines to treat NAFLD.^227,228 It is characterized by a high intake of olive oil, which is a major source of fat in the diet, and a low intake of added sugars and refined carbohydrates. Actually, it is also a high-fat, LCD. What’s more, MD contains a large number of vegetables, fruits, and whole grains, as well as moderate wine and dairy food in the diet, while red meat, especially processed meat, intake is low.

MD has been reported to show a more significant reduction of liver steatosis and improvement of insulin sensitivity compared with LFD (39% vs. 7%), although there is no difference in weight loss between the 2 groups.¹⁸¹ Tsaban et al²²⁹ also reported a decrease of serum LDL-C levels and Framingham Risk Score, which is associated with cardiovascular risks, in 293 patients consuming green MD.

Moreover, MD is a dietary pattern rich in antioxidants, mainly because the major source of fat is monounsaturated and polyunsaturated fats instead of saturated fats. The appropriate ratio of n-3 to n-6 fatty acids in MD provides an opportunity to benefit from n-3 polyunsaturated fats,²³⁰ which can prevent intrahepatic lipid accumulation and decrease the risks of steatosis.²³¹ What’s more, patients following MD have decreased serum hypersensitive-C-reactive-protein and inflammatory cytokines such as IL-6, IL-7, and IL-18 compared with the control groups. Endothelial function and IR are also improved in the MD group, associated with decreased rates of metabolic syndromes and risks of CVDs.²³² Another clinical trial, in which patients were divided into 3 groups (MD, MD with supplementation of antioxidants, control), also reported that individuals consuming MD with or without antioxidants showed decreased liver lipid accumulation and fibrosis. However, groups with extra antioxidant intake exhibited more significantly reduced BMI, waist, hip circumference, and improved glucose metabolism compared with patients with MD alone, indicating the role of antioxidants in the prevention of NAFLD.²³³

Furthermore, another characteristic of MD is the consumption of coffee, which is considered to reduce the risk of liver fibrosis in NAFLD.²³⁴ Coffee intake can lower the risks of HCC as well, but interestingly, caffeine may not contribute to this beneficial effect.²³⁵ Paola reported that gut permeability and the gut-liver axis modulated by coffee participate in the prevention of NAFLD in HFD-induced mouse models, with reduced liver lipid accumulation and improved energy metabolism.²³⁶ Coffee is also reported to regulate the expression of long noncoding RNA in pathways of NAFLD progression, such as inhibiting lipogenesis through the long noncoding RNA Gm16551/Srebf1 pathway as well as long noncoding RNA H19 associated with fibrosis.²³⁷ Although the mechanisms of coffee in preventing NAFLD are unclear, coffee consumption provides a hopeful option for patients.

However, MD recommends moderately drinking wine. It is controversial whether NAFLD patients should accept it since current studies in humans show conflicting results. Long et al²³⁸ reported an association between alcohol and hepatic steatosis, indicating that alcohol is a risk factor for NAFLD and should be avoided, while Mitchell et al²³⁹ reported that modest wine consumption(1–70 g/wk) can lower the rates of liver fibrosis. What’s more, it has been reported that any amount of alcohol consumption increases the risk of NASH-induced HCC,²⁴⁰ but specific components of wine such as resveratrol show antioxidant properties and therefore can improve endothelial function and hypertension.²⁴¹ Therefore, NAFLD patients should balance between the risks and benefits. More studies concerning the mechanisms of how wine modulates NAFLD and longitudinal clinical trials are required before clinical recommendation.

DASH diet

The DASH diet, which can decrease cardiovascular risks,²⁴² was originally used to regulate blood pressure, while NAFLD patients have been found to benefit from it as well.¹⁷⁶ The DASH diet is similar to MD but characterized by reduced sodium, added sugars, and saturated fats.

Mohsen et al reported that NAFLD patients following the DASH diet had significantly decreased weight, ALT, alkaline phosphatase, and serum triglyceride levels, as well as improved insulin sensitivity and markers of inflammation.¹⁷⁶ Hekmatdoost et al²⁴³ also reported an inverse relationship between adherence to DASH diets and risks of NAFLD. Moreover, a meta-analysis suggested that the DASH diet is more suitable for obese individuals to lose weight than a low-calorie diet.²⁴⁴ However, currently, studies on the effects of DASH diet in NAFLD patients are limited. More clinical trials concerning the influence and side effects of DASH diets in NAFLD are needed.

Comparison between LFD and LCD

It is controversial whether a LFD or a LCD is a better choice for NAFLD patients, although both of them are effective in losing weight. Ryan et al¹⁸⁹ reported that in 2 groups with hypocaloric diets, individuals following LCDs showed more significantly improved ALT levels and insulin sensitivity than those following LFDs, although body weight loss was similar in both groups. Likewise, there was a more significant reduction of IHTG in the LCD group than in the low-fat, high-carbohydrate group after 48 hours. However, over a long period of 11 weeks, both groups exhibited similar weight loss and no differences in the decrease of IHTG.²²² On the contrary, Jukka et al(245) reported that liver fat decreased in the LFD group but increased in the high-fat, low-carbohydrate group, in parallel with changes in serum insulin levels [245]. A meta-analysis in 2019 also suggested that the LCD and LFD groups had similar improvements in hepatic fat and serum liver enzymes, indicating that current evidence to support either diet is limited.²⁴⁶

There has been no conclusion on which diet pattern is superior to the other for NAFLD patients. The conflicting results may be attributed to the different sources of carbohydrates and fat, as well as adherence of participants to the diets, which should be taken into consideration in the future investigation.

METABOLIC TARGETS IN NAFLD

Key points

Various metabolic drugs have been approved for clinical trials in the treatment of NAFLD/NASH. farnesoid X receptor (FXR), THRβ, and hepatic DNL enzymes have been identified as promising drug targets. Plenty of FDA-approved antidiabetic and lipid-modifying agents are undergoing phase 4 clinical trials for NAFLD/NASH treatment.

Though there has been no licensed drug for NAFLD treatment so far, various metabolic targets have been discovered and plenty of potential drugs have been approved for clinical trials in recent years. Here, we summarized the major metabolic targets and potential monotherapy drugs under development for the treatment of NAFLD/NASH (Table 3).

TABLE 3.

Update of major metabolic targets and potential monotherapy drugs under development for NAFLD/NASH

				Clinical improvements
Target(s)	Agent	Current phase	Condition	Weight loss	ALT	LFC/steatosis	fibrosis	Sponsor	References
FXR	Nidufexor (LMB763)a	Phase 2 (terminated)	NASH	√	√	√	—	Novartis	Chianelli et al247
	TERN-101	Phase 2a	NASH	NA	√	√	√	Terns	Wang et al248
	Cilofexor (GS-9674)a	Phase 2	NASH	—	—	√	—	Gilead	Patel et al249
	MET642a	Phase 2	NASH	NA	NA	√	NA	Metacrine	—
	Px-104	phase 2	NAFLD	—	√	—	TBD	Phenex	Traussnigg et al250
	EDP-297a	Phase 1	NASH	NA	NA	NA	NA	Enanta	—
	EDP-305a	phase 2b	NAFLD	—	√	√	—	Enanta	Ratziu et al251
	HPG1860	Phase 2a	NASH	TBD	√	TBD	TBD	Hepagene	—
	HEC96719	Phase 2	NASH	TBD	TBD	TBD	TBD	HEC Pharm	—
	Tropifexor (LJN452)a	Phase 2 (terminated)	NASH	√	√	√	—	Novartis	Sanyal et al252
	Vonafexor (EYP001a)	Phase 2	NASH	√	√	√	√	Enyo Pharma	Ratziu et al253
	XZP-5610	Phase 1	NASH	TBD	TBD	TBD	TBD	Sihuan pharm	—
	MET409a	Phase 2	NASH	√	—	√	NA	Metacrine	Harrison et al254
	ASC42	Phase 1	NASH	TBD	TBD	TBD	TBD	Ascletis Pharma	—
	Obeticholic acid	Phase 3	NAFLD/NASH with or without T2D	√	√	√	√	Intercept	Younossi et al255
iBAT/ASBT/SLC10A2	Elobixibat (A3309)a	Phase 2	NAFLD NASH	NA	—	—	NA	Albireo	—
	Volixibat (SHP626)a	Phase 2 (terminated)	NASH	NA	—	—	—	Mirum	Newsome et al (2020)256
SGLT2	Empagliflozin (Jardiance)	Phase 4	NAFLD with or without T2D	√	√	√	—	—	Kuchay et al257 Taheri et al258
	Ertugliflozin	Phase 4	NAFLD with T2D	TBD	TBD	TBD	TBD	Getz Pharma	—
	Dapagliflozin	Phase 4	NAFLD with T2D	√	√	√	—	—	Shimizu et al259 Phrueksotsai et al260
	Canagliflozin	Phase 4	NAFLD with T2D	√	√	√	—	—	Inoue et al261
SGLT1&2	Licogliflozin (LIK066)	Phase 2	NASH	√	√	√	√	Novartis	Harrison et al262
PPARα	Pemafibrate (K-877)	Phase 2	NAFLD	—	√	—	—	Kowa Company	Nakajima et al263
	Fenofibrate	Phase 4	NAFLD	√	√	—	—	—	Fernández-Miranda et al264 Oscarsson et al101
PPARγ	Rosiglitazone	Phase 4		—	√	√	—	—	Ratziu V, et al. (2008)265
	Lobeglitazone	Phase 4	—	—	√	√	—	—	Lee et al266
	Pioglitazone	Phase 4	NAFLD/NASH with T2D	—	√	√	√	—	Aithal et al267 Cusi et al268
	PXL065	Phase 2	NASH	—	—	√	—	Poxel	Harrison et al269
PPARδ(PPARβ)	Seladelpar	Phase 2	NASH	NA	√	—	NA	CymaBay	—
PPARα/δ	Elafibranora	Phase 3 (terminated)	NASH	—	√	√	√	Genfit	Ratziu et al270
PPARα/γ	Saroglitazar	Phase 2	NAFLD/NASH	—	√	√	—	Zydus Therapeutics	Siddiqui et al271 Gawrieh et al272
pan-PPAR	Lanifibranor (IVA337)	Phase 3	NAFLD/NASH	—	√	NA	√	—	Sven et al273
	Chiglitazar	Phase 2	NASH	TBD	TBD	TBD	TBD	Chipscreen Biosciences	—
DPP-4	Sitagliptin	Phase 4 (terminated)	NAFLD/NASH	—	—	—	—	—	Cui et al274 Joy et al275
	Evogliptin	Phase 4	NAFLD With T2D	—	√	√	TBD		Han E, et al. (2022)276
GLP-1R	Liraglutide	Phase 4	NAFLD /NASH	√	√	√	TBD	—	Armstrong et al277 Petit et al278
	Semaglutide	Phase 4	NASH	√	√	√	—	—	Flint et al279 Newsome et al280
	Dulaglutide	Phase 4	NAFLD/NASH with T2D	√	—	√	NA		Kuchay et al281
	Ecnoglutide (XW003)	Phase 1	NASH, T2D and obesity	TBD	TBD	TBD	TBD	Sciwind Biosciences	—
	Exenatide	Phase 4	—	√	√	√	√	—	Shao et al282 Liu et al283
	ECC5004	Phase 1	NASH T2D	TBD	TBD	TBD	TBD	Eccogene	—
GLP-1R& GIPR	VK2735	Phase 1	NASH	NA	NA	NA	NA	Viking Therapeutics	—
	Tirzepatide(LY3298176)	Phase 2	NASH	√	√	√	NA	Eli Lilly and Company	—
GLP-1R& GCGR	Efinopegdutide	Phase 2a	NASH	TBD	TBD	TBD	TBD	Merck Sharp & Dohme	—
	Cotadutide (MEDI0382)a	Phase 3	NASH	—	—	√	NA	AstraZeneca	—
	ALT-801 (Pemvidutide)	Phase 1	NAFLD with or without T2D	√	√	√	TBD	Altimmune	—
	BI 456906	Phase 2	NASH	√	TBD	TBD	TBD	Boehringer Ingelheim	Yazawa et al284 Jungnik et al285
	SAR425899 (Bamadutide)	Phase 2 (withdrawn)	NASH	√	NA	NA	NA	Sanofi	—
	DD01	Phase 1	NASH with T2D	TBD	TBD	TBD	TBD	D&D Pharmatech	—
GLP-1R & GIPR & GCGR	Efocipegtrutide (HM15211)	Phase 2b	NASH	TBD	TBD	TBD	TBD	Hanmi	—
FGF21	pegozafermin (BIO89-100)	Phase 2	NASH	—	√	√	TBD	89bio	Loomba et al286
	B1344	Phase 1	NASH	TBD	TBD	TBD	TBD	Tasly Biopharma	—
	NNC0194 0499	Phase 1	NASH	TBD	TBD	TBD	TBD	Novo Nordisk A/S	—
	Pegbelfermin (BMS-986036)	Phase 2	NASH	—	√	√	√	Bristol-Myers Squibb	Sanyal et al287 Brown et al288
	BOS-580	Phase 2	NASH	TBD	TBD	TBD	TBD	Boston Pharmaceuticals	—
	Efruxifermin (EFX)	Phase 2	NASH	√	√	√	√	Akero Therapeutics	Harrison et al289
GLP-1&FGF21	HEC88473	Phase 1	NASH	TBD	TBD	TBD	TBD	HEC Pharm	—
FGFR1&KLB	BFKB8488A	Phase 2	NASH	√	√	√	TBD	Genentech	Baruch et al290 Wong et al291
	MK-3655 (NGM313)a	Phase 2	NASH	NA	NA	NA	NA	Merck Sharp & Dohme LLC	—
FGF19	Aldafermin (NGM282)a	Phase 2	NASH	—	√	√	—	NGM Bio	Harrison et al292
GR	miricorilant	Phase 2 (suspended)	NASH	TBD	TBD	TBD	TBD	Corcept Therapeutics	—
FASN	Denifanstat (TVB-2640)	Phase 2	NASH	—	√	√	√	Sagimet Biosciences	Syed-Abdulet al293 Loomba et al294
	FT-4101a	Phase 2 (terminated)	NASH	—	NA	√	NA	Forma Therapeutics	Beysen et al295
KHK	PF-06835919a	Phase 2	NAFLD	—	—	√	NA	Pfizer	Kazierad et al296 Saxena et al297
	XZP-6019	Phase 1	NAFLD	TBD	TBD	TBD	TBD	Sihuan Pharmaceutical Holdings Group Ltd.	—
DGAT2	PF-07202954	Phase 1	NASH	TBD	TBD	TBD	TBD	Pfizer	—
	Ervogastat (PF-06865571)	Phase 2	NASH	NA	NA	√	NA	Pfizer	—
	Pradigstat (LCQ908)a	Phase 2	NASH	NA	—	√	NA	Novartis	—
	SNP-612a	Phase 2 (terminated)	NASH	NA	NA	NA	NA	Sinew Pharma	—
	SNP-610	Phase 2	NASH	TBD	TBD	TBD	TBD	Sinew Pharma	—
	ION224	Phase 2	NASH	TBD	TBD	TBD	TBD	Ionis Pharmaceuticals	—
SCD-1	Aramchol	Phase 3 (suspended)	NASH	—	√	√	—	Galmed Pharmaceuticals	Ratziu et al298
ACC	Clesacostat (PF-05221304)a	Phase 2	NAFLD NASH	NA	√	√	NA	Pfizer	Bergman et al299 Calle et al300
	Firsocostat (GS-0976)a	Phase 2	NASH	NA	—	√	√	Gilead Sciences	Loomba et al301
	MK-4074a	Phase 1	NAFLD	NA	NA	√	NA	Merck Sharp & Dohme LLC	Kim et al302
THR-β	ALG-055009	Phase 1	NASH	TBD	TBD	TBD	TBD	Aligos Therapeutics	—
	Resmetirom (MGL-3196)	Phase 3	NASH	—	√	√	√	Madrigal Pharmaceuticals	Harrison et al303 Harrison et al304
	ASC41	Phase 2	NASH	TBD	TBD	TBD	TBD	Ascletis Pharma	—
	TERN-501	Phase 2	NASH	TBD	TBD	TBD	TBD	Terns Pharmaceuticals	—
	ECC4703	Phase 1	NASH	TBD	TBD	TBD	TBD	Eccogene	—
	VK2809	Phase 2	NASH	TBD	√	√	TBD	Viking Therapeutics	—
PNPLA3	AMG 609	Phase 1	NASH	TBD	TBD	TBD	TBD	Amgen	—
	ALN-PNP	Phase 1	NASH	TBD	TBD	TBD	TBD	Alnylam	—
	AZD2693	Phase 2	NASH	TBD	TBD	TBD	TBD	AstraZeneca	—
	JNJ-75220795 (ARO-PNPLA3)	Phase 1	NASH	TBD	TBD	TBD	TBD	Arrowhead	—
ANGPTL3	Vupanorsen (ISIS 703802)a	Phase 2	NAFLD with T2D	NA	—	—	NA	Ionis Pharmaceuticals	Gaudet D, et al.305
MPC	MSDC-0602K	Phase 2	NASH	—	√	—	—	Cirius Therapeutics	Harrison et al306
MOTS-c	CB4211	Phase 1	NASH	—	√	—	TBD	CohBar	—
HTR2A	GM-60106	Phase 1	NASH	TBD	TBD	TBD	TBD	JD Bioscience	—
GDF15	NGM395a	Phase 1	NAFLD	NA	NA	NA	NA	NGM Biopharmaceuticals	—
HSD17B13	ARO-HSD (GSK4532990)	Phase 2	NASH	—	√	—	—	Arrowhead Pharmaceuticals (Licensed to GlaxoSmithKline)	Mak et al307
	AZD7503	Phase 1	NASH	TBD	TBD	TBD	TBD	AstraZeneca	—
	ALN-HSD	Phase 2	NASH	TBD	TBD	TBD	TBD	Alnylam	—
HSD11B1	RO5093151a	Phase 1	NASH	√	√	√	NA	Roche	Stefan et al308
miR-103/107	RG-125 (AZD4076)a	Phase 1	NAFLD with T2D	NA	NA	NA	NA	AstraZeneca	—
RORα	TB-840	Phase 1	NASH	TBD	TBD	TBD	TBD	Therasid Bioscience	—
CB1	Namacizumab (RYI-018)a	Phase 1	NASH	TBD	TBD	TBD	TBD	Bird Rock Bio (Johnson & Johnson)	—
	Otenabant (CP-945598)a	Phase 1	NASH	√	NA	NA	NA	Pfizer	Aronne et al309
	Rimonabanta	Phase 3 (terminated)	NAFLD with or without T2D	NA	NA	NA	NA	Sanofi	—
AMPK	PXL770	Phase 2	NASH	—	√	√	TBD	Poxel	Cusi et al310
Androgen	Spironolactone	Phase 4	NAFLD	√	—	√	NA	—	Polyzos et al311

^aFuture development for NAFLD/NASH monotherapy is not pursued.

Abbreviations: ACC, acetyl-CoA carboxylase; ALT, alanine transferase; AMPK, AMP-activated protein kinase; ANGPTL3, angiopoietin like protein 3; CB1, cannabinoid; DPP-4, dipeptidyl peptidase-4; FASN, FGFR1&KLB, β-Klotho/FGFR1c receptor complex; FXR, farnesoid X receptor; GIPR, glucose-dependent insulinotropic polypeptide receptor; GLP-1R, glucagon-like peptide-1 receptor; HSD11B1, 11β-hydroxysteroid dehydrogenase type 1; HSD17B13, hydroxysteroid 17β-dehydrogenase 13; HTR2A, 5-hydroxytryptamine receptor 2A; iBAT, ileal bile acid transporter; LFC, liver fat content; MOTS-c, mitochondrial open-reading frame of the 12 SrRNA type-c; MPC, mitochondrial pyruvate carrier; NA, not available; PNPLA3, patatin-like phospholipase domain–-containing protein 3; PPAR, peroxisome proliferation– activated receptor; PPARG, peroxisome proliferator– activated receptor gamma; SCD-1, stearoyl-CoA desaturase-1; SGLT, sodium-glucose cotransporter; SSPG, steady- state plasma glucose; SWE, shear-wave elastography; TBD, to be determined.

FXR has been one of the most extensively studied targets for NAFLD/NASH since 2015, with the release of the remarkable interim data from a 72-week randomized trial of a FXR agonist, obeticholic acid, in biopsy-proven NASH patients.³¹² FXR is a nuclear receptor expressed in both hepatocytes and intestinal enterocytes. Activation of FXR alleviates hepatic lipid accumulation potentially by 2 mechanisms: (1) the downregulation of hepatic DNL via an FXR-SHP-SREBP-1c pathway, and (2) the reduction of intestinal lipid absorption.³¹³ Improvement in steatosis, inflammation, and fibrosis, as well as a weight-loss effect, was observed in the obeticholic acid therapy.³¹² Despite its efficacy, obeticholic acid was not approved for NASH because of its common adverse events of pruritus and transient increase in LDL-C. Though it remains unclear whether pruritus and LDL-C increases are drug-specific or target-related, they are also observed in most FXR agonists, including Tropifexor, Cilofexor, Vonafexor, EDP-297, EDP-305, and MET-409.^253,314,315 In the recently released data of 2 phase 1 trials of a nonsteroidal FXR agonist, TERN-101, no pruritus and lipid profile change was reported.²⁴⁸ Further studies are required to evaluate its safety and efficacy in patients with longer treatment duration.

As a direct target of FXR, a FGF19 analog, Aldafermin, was also investigated for NAFLD/NASH treatment. However, although improvements in LFC, liver enzymes, and fibrosis markers were observed after the 24-week phase 2b trial, Aldafermin failed to reach the primary end point of fibrosis improvement.²⁹² Another potential target in BA metabolism is the ileal BA transporter, the inhibition of which interrupts enterohepatic recirculation and promotes the hepatic de novo production of BAs. Nevertheless, 2 ileal BA transporter inhibitors, Elobixibat and Volixibat, both failed to improve ALT levels and steatosis in NASH patients, indicating that ileal BA transporter may not be the best choice for NAFLD/NASH therapy.²⁵⁶

Compared with FGF19, FGF21 has attracted much more interest in terms of its multiple effects on obesity and metabolism.³¹⁶ Several FGF21 analogues are going through early clinical trials for NASH patients in various fibrosis stages. Among them, pegozafermin, pegbelfermin, and efruxifermin have shown promising results in phase 2 studies, where all of these drugs significantly improved ALT levels, the hepatic fat fraction, and metabolic biomarkers.^286–289

Resmetirom, a thyroid hormone receptor-β (THRβ) agonist, is perhaps the most promising drug under development for NAFLD/NASH. Significant reduction of hepatic fat measured by MRI-PDFF, reduction of liver enzymes, and improvement of inflammation and fibrosis biomarkers were observed in a multicenter, randomized 36-week phase 2 study.³⁰³ Positive phase 3 results have been released recently, announcing that the dual primary end points of NASH resolution and fibrosis improvement are achieved by resmetirom. Other THRβ agonists under investigation include ALG-055009, ASC41, TERN-501, ECC4703, and VK2809.

Several drugs targeting hepatic DNL have been approved for clinical trials in recent years, including FASN inhibitors (denifanstat, FT-4101), DGAT2 inhibitors (PF-07202954, ervogastat, pradigstat, SNP-610, ION224), the stearoyl-CoA desaturase-1 inhibitor (aramchol), and acetyl-CoA carboxylase inhibitors (clesacostat, firsocostat, MK-4074), most of which are still in early-phase clinical trials and further data are required before a proper review. So far, all 3 acetyl-CoA carboxylase inhibitors have been excluded from monotherapy development because of the side effect of hypertriglyceridemia. KHK, the key enzyme for fructose metabolism, was also investigated as a potential target. The KHK inhibitor PF-06835919 has been proved to be safe and well-tolerated, and reduces liver fat in patients with NAFLD and T2D. Yet, the clinical improvements were not significant enough to support further development.²⁹⁷ The phase 1 clinical trial of another KHK inhibitor, XZP-6019, is currently ongoing.

Since NAFLD/NASH is related to obesity and T2D, plenty of FDA-approved antidiabetic and lipid-modifying agents have been going through phase 4 clinical trials to evaluate the possibility for NAFLD/NASH treatment. Pioglitazone has been recommended by KASL clinical practice guidelines for NAFLD management.³¹⁷ While the side effects of weight gain, edema, and bone loss limit the use of TZDs,³¹⁸ PXL065, a deuterium-stabilized stereoisomer of pioglitazone, is expected to avoid the safety concerns and improve efficacy through non–PPARγ-mediated mechanisms.^269,319

CONCLUSIONS AND FUTURE PERSPECTIVES

There are no drugs approved for the treatment of NAFLD until now, and therapies are limited to weight loss and exercise. The nutrient compositions of diets and dietary patterns have significant impacts on weight loss and progression of NAFLD. In this review, we briefly summarize the effects of 3 macronutrients, vitamins, and several dietary patterns, including CRD, IF, KD, MD, and the DASH diet, in the development of NAFLD. We also compare the effects of LCD and LFD. What’s more, we summarize the potential metabolic targets that seem to be promising in the NAFLD treatments.

However, the current findings are still incomplete. In studies on the effects of sugar on NAFLD, fructose consumption in current mouse models far exceeds the human daily intake, which may overestimate the harm of fructose. There are also significant sex differences in fructose metabolism in the liver, so only using male mouse models in some research may lead to incomplete results. Moreover, the current research is limited to the effects of fructose on NAFLD, but other sugars such as sucrose are widely used in daily cooking as well. The effects of NNSs, which are becoming increasingly popular, on IR, promotion of NAFLD, and microbiota dysbiosis are still uncertain. Furthermore, fructose-induced leptin resistance is likely to increase the intake of other foods. The impacts of fructose on other dietary elements’ metabolic effects require further studies since fructose may exacerbate the adverse effects of glucose, fat, or alcohol³²⁰ on NAFLD. In studies on the effect of fat on NAFLD, research on the effects of low-fat diets is limited to clinical settings, while the mechanism is uncertain. The exact dose of n-3 and its duration are also under investigation before recommending it for the clinical use. In studies on the effects of protein on NAFLD, there is little research concerning the effects of specific amino acids. What’s more, whether the advantages and disadvantages of plant and animal proteins are attributed to the proteins itself or other elements in food and their mechanism requires further research. The side effects of high-protein diets should also be considered before using them on a large scale.

The major problems of studies on dietary patterns are the small sample sizes, short observation time, and insufficiency of means to assess patients’ adherence to the diets. So far, only MD is recommended by some NAFLD guidelines, and a tool to assess adherence to MD—Mediterranean adequacy index—has already been designed.³²¹ However, more clinical trials are needed to standardize the characteristics of MD. More large-scale, long-term studies should be designed to gain further evidence about other dietary patterns, and specific tools should be developed to assess individuals’ adherence.

Notably, there is a lack of research on the most appropriate intake of the 3 macronutrients in the diets. The intake of carbohydrates, fat, and proteins in current clinical trials is inconsistent, and more studies are required to standardize the recommended doses. Another factor that needs to be considered is that although weight loss is traditionally thought to alleviate NAFLD, some individuals with normal BMIs develop NAFLD after losing too much weight in some cases. Therefore, the specific rate of weight loss and whether weight loss is applicable to lean NAFLD patients need further studies.

The development of drugs for the treatment of NAFLD is even more difficult. Most drugs fail to enter the phase III clinical trials, probably due to insufficient sample sizes or the short duration. As reversal of liver histology may be a slow process, it is unlikely that efficacy will be seen in a short period of time. In addition, the mechanisms of these metabolic targets, as well as the pathogenesis of NAFLD, still require more research.

FUNDING INFORMATION

This study was supported by the Project of the Shanghai Municipal Health Commission (20204Y0012), Seed Fund of Renji Hospital (RJZZ18-010), Shenkang 3-year action plan (SHDC2020CR2003A, SHDC2020CR5012), Innovative Research Team of High-Level Local Universities in Shanghai (SSMU-ZDCX20180802), National Natural Science Foundation of China (81972205), and the Project of Shanghai key clinical specialties (shslczdzk05801).

CONFLICTS OF INTEREST

The authors have no conflicts to report.