Review article: omega-3 fatty acids – a promising novel therapy for non-alcoholic fatty liver disease

Abstract

Aliment Pharmacol Ther 31, 679–692

Summary

Background  Non-alcoholic fatty liver disease affects 10–35% of the adult population worldwide; there is no consensus on its treatment. Omega-3 fatty acids have proven benefits for hyperlipidaemia and cardiovascular disease, and have recently been suggested as a treatment for non-alcoholic fatty liver disease.

Aims  To review the evidence base for omega-3 fatty acids in non-alcoholic fatty liver disease and critically appraise the literature relating to human trials.

Methods  A Medline and PubMed search was performed to identify relevant literature using search terms ‘omega-3’, ‘N-3 PUFA’, ‘eicosapentaenoic acid’, ‘docosahexaenoic acid’, ‘non-alcoholic fatty liver disease’ and ‘NAFLD’.

Results  Omega-3 fatty acids are important regulators of hepatic gene transcription. Animal studies demonstrate that they reduce hepatic steatosis, improve insulin sensitivity and reduce markers of inflammation. Clinical trials in human subjects generally confirm these findings, but have significant design inadequacies.

Conclusions  Omega-3 fatty acids are a promising treatment for non-alcoholic fatty liver disease which require to be tested in randomized placebo-controlled trials.

Introduction

Non-alcoholic fatty liver disease (NAFLD) is defined by the pathological accumulation of fat in the liver when no other explanatory disease is present: it encompasses isolated hepatic steatosis, non-alcoholic steatohepatitis (NASH) and cirrhosis. A recent review demonstrated that NAFLD affects 10–35% of the adult population worldwide;¹ it is now the most common cause of liver disease in the US,² and accounts for 11% of referrals to hepatology services.³

Non-alcoholic fatty liver disease may be regarded as the hepatic expression of the metabolic syndrome which consists of hypertension, insulin resistance, obesity and dyslipidaemia.⁴ The basis of this assertion is that the more facets of the metabolic syndrome that are present, the greater the chance of developing NAFLD.^{5, 6} Furthermore, the presence of NAFLD predicts the development of other features of the metabolic syndrome.^{6, 7} NAFLD is independently associated with increased cardiovascular events⁸ and, in the NHANES-III population based study, increased mortality.⁹

The pathogenesis of NAFLD is incompletely understood. Classically, it is considered to be the outcome of ‘two hits’.¹⁰ Steatosis, primarily in the form of triglyderides,¹¹ and insulin resistance are thought to occur first. Mitochondrial dysfunction precipitating oxidative stress follows and this triggers an inflammatory and fibrogenic cascade in the primed liver.¹⁰ Whilst this provides a useful overview and incorporates the major events in the pathogenesis of NAFLD, it has been suggested that this model should be revised and that steatosis is an epiphenomenon of oxidative stress.¹²

Other mechanisms are emerging in the pathogenesis of NASH, notably the role of adipose tissue in the secretion of proinflammatory and prothrombotic adipocytokines, IL-6 and TNF-α^13–15 and the reduced production of the adipocytokine adiponectin, a potent anti-inflammatory, insulin-sensitizing agent.^{16, 17}

At present, there is no consensus on the treatment of NAFLD.¹⁸ Derived from an understanding of the pathogenesis of NAFLD, various treatment strategies can be identified and therapies that have been tested may be grouped accordingly (Table 1).

Table 1.   Treatment strategies in non-alcoholic fatty liver disease

Strategy	Intervention
Weight loss	Lifestyle measures^102–104
	Drugs
	Orlistat^105–107
	Sibutramine^107
	Rimonabant^108
	Bariatric surgery^109–112
Reduce insulin resistance	Metformin^113–117
	Thiazolidinediones^118–122
Antioxidant	Vitamin E^123–125
	Probucol^126
Anti-TNF	Pentoxiphylline^127, 128
Other	Ursodeoxycholic acid^129
	Angiotensin 2 antagonists^130, 131
	Betaine^132
	N-acetylcysteine^133
	Yo Jyo Hen Shi Ko^134

Currently, advice on diet and weight loss and the energetic management of any co-existing features of the metabolic syndrome form the mainstay of treatment. Studies of the dietary habits of NAFLD patients reveal that they consume less oily fish, but double the quantity of soft drinks and 27% more meat compared with the general population, and these dietary differences are associated with an increased risk of NAFLD independent of traditional risk factors.¹⁹

Omega-3 fatty acids have recently been proposed as a potential treatment for NAFLD.²⁰ These fatty acids have proven benefit in lowering serum triglycerides and in the treatment of cardiovascular disease.^21–23 Interest in their potential in the treatment of cancer, mood disorders and cognitive disorders has also emerged.^24–26 There are promising data from both animal and human trials on the use of omega-3s in NAFLD. This study now reviews the potential mechanisms through which omega-3 fatty acids may be of benefit in NAFLD, and the current data supporting its use.

Omega-3 fatty acids: background and metabolism

Omega-3 (N-3) fatty acids are essential, polyunsaturated fatty acids (PUFAs), i.e. they cannot be synthesized in vivo. In diet, large quantities are found naturally in fish oil, flaxseed and some nuts. They derive from α-linolenic acid and mainly occur as eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA), which are both anti-inflammatory.²⁷ These are then converted to active metabolites, in particular, molecules known as resolvins and protectins. These recently discovered lipid products are yet to be fully characterized, but are thought to mediate, at least in part, the anti-inflammatory effect of omega-3 fatty acids.²⁸ The other key group of PUFAs, N-6 fatty acids, are found predominantly in grain. They derive from linolenic acid and their primary metabolite, arachadonic acid, is proinflammatory and prothrombotic.

N-6 and N-3 fatty acids (PUFAs) are competitively metabolized by the same pathway. It has been suggested that the ratio of N-6 to N-3 should be approximately 3:1; however, because the modern diet is rich in N-6 foods, it can be as high as 15:1.^{29, 30} In NAFLD, a biopsy-based study showed that the N-6:N-3 ratio correlated significantly with the quantity of hepatic triglycerides.³⁰ Although some diseases, including breast cancer and asthma, may be associated with a higher N6:N3 ratio, large trials have shown that reduction in cardiovascular risk is linked to the total amount of N-3 fatty acids rather than the N-6:N-3 ratio.³¹ Similarly, we believe this to be the case in NAFLD and it is now established that a low total N-3 level is found in NAFLD, and this is associated with steatosis, increased oxidative stress and NASH.^{11, 32–34}

Omega-3 regulation of hepatic gene expression

Omega-3 fatty acids are key regulators of hepatic gene transcription, with peroxisome proliferator-activated receptor alpha (PPARα) and sterol regulatory element-binding protein-1 (SREBP-1) being best known. These have diverse effects on carbohydrate and lipid metabolism (Figure 1) and may act like hydrophobic hormones: i.e. upon ligand binding and activation, they bind to and alter the function of specific response elements in target genes.^{35, 36}

Figure 1.  Effects of omega-3 fatty acids on hepatic gene transcription.

Peroxisome proliferator-activated receptor alpha.  PPARα is a transcription factor known to reduce plasma lipids (the basis of fibrates)³⁷ and increase mitochondrial beta oxidation.³⁸ Two studies employing a murine model of NASH demonstrated that infusion of a PPARα agonist can prevent steatohepatitis and reverse established disease.^{39, 40} Omega-3s are potent activators of PPARα, which upregulate several genes associated with fatty acid and lipid metabolism that stimulate fatty acid oxidation.^{29, 35, 41–44} Interestingly, in addition to improvements in steatosis, there may be an independent, anti-inflammatory effect via PPARα-mediated suppression of TNF-α and IL-6.^{43, 45}

Sterol regulatory element-binding protein-1.  There are three isoforms of SREBP: 1a,1c and 2. SREBP-1a is a potent activator of all genes under SREBP regulation; SREBP-1c primarily influences genes involved in fatty acids synthesis; SREBP-2 is implicated in cholesterol synthesis.⁴⁶ For simplicity, 1a and 1c will be considered jointly as SREBP-1.

Sterol regulatory element-binding protein-1 plays an important role in insulin resistance and is also a key regulator of fatty acid synthesis.⁴⁷ Levels are elevated in response to high insulin and glucose concentrations by increasing the production of its precursor.^46–48 To form mature SREBP-1, the molecule requires to undergo two post-translational processes – proteolytic processing and proteasomal degradation in the golgi apparatus.³⁸ Following this, the mature SREBP-1 is transported to the nucleus where, by binding to sterol regulatory elements in the promoter areas of genes, it stimulates increased de novo lipogenesis^{48, 49} and glycolysis.³⁵ Hence, over-expression of SREBP-1 results in accumulation of triglycerides in the liver.⁵⁰ Target genes for SREBP-1 include glucokinase, the intracellular enzyme which is the rate limiting step for glycolysis.⁴⁸ Omega-3s reduce the amount of mature SREBP-1 available in the nucleus⁵¹ and thereby inhibit the downstream stimulatory effects of insulin³⁵ so reducing de novo lipogenesis.⁵² This effect may be mediated by reducing the effective half life of SREBP-1 mRNA.⁵³ SREBP-1 has been shown to be suppressed by omega-3 fatty acids in both cell culture and animal models.^{49, 54–56}

Thus, the effect of omega-3 fatty acids on PPARα is to increase hepatic beta oxidation and lipid catabolism, whilst its effect on SREBP-1 is to reduce endogenous lipid production.

Others.  PPARγ, like PPARα, is a nuclear receptor involved in lipid metabolism, being primarily expressed in macrophages and adipose tissue. It is a regulator of adipose tissue metabolism, increases peripheral insulin sensitivity and is a therapeutic target for the thiazolidinediones. Omega-3 fatty acids are also ligands for PPARγ.³⁶ In a study of healthy humans, omega-3 supplementation has been shown to increase fat oxidation and increase insulin sensitivity.⁵⁷ Another study found that omega-3 improved peripheral insulin, but increased hepatic insulin resistance.⁵⁸

In addition to PPARs, there are three other nuclear hormone receptors involved in fat metabolism: liver X receptor (LXR), farnesoid X receptor (FXR) and hepatocyte nuclear factor-4α (HNF-4α). The effect of omega-3 fatty acids on these receptors and their inter-relations are complex and still being elucidated.

Liver X receptors regulate fatty acid and cholesterol transport and metabolism.^{59, 60} These effects are mediated, at least in part, by enhancing genes involved in bile salt synthesis and by inhibiting intestinal cholesterol absorption. Interestingly LXR also increases SREBP-1 expression, thus promoting de novo triglyceride synthesis. Although it initially appeared that LXR expression was suppressed by omega-3s, there is now evidence that its expression remains unaffected.⁶¹

Hepatocyte nuclear factor-4α, which is implicated in the development of juvenile onset type 2 diabetes, plays an important role in lipoprotein production, increases l-pyruvate kinase and glucokinase expression in the liver.⁶² HNF-4α is stimulated by saturated fatty acids and inhibited by omega-3 fatty acids.⁵⁹ Thus, omega-3 fatty acids inhibit glucose flux in the hepatocyte and downregulate hepatic lipogenesis in part through regulation of HNF-4α expression.

Farnesoid X receptor controls bile salt synthesis by inhibiting bile synthesis enzymes and bile salt export pump. It also reduces hepatic triglycerides by inducing PPARα and by inhibiting SREBP-1; the latter effect is mediated by enhancing the transcription of short heterodimer partner.⁶³ Omega-3 fatty acids are known to upregulate FXR.⁶⁴

Another potential mechanism of omega-3s is inhibition of glycolysis by suppressing carbohydrate regulatory element-binding protein (ChREBP) and through this inhibiting l-pyruvate kinase.^{35, 65, 66} ChREBP has been shown to be linked to hepatic steatosis and insulin resistance, and in animal models, blocking this molecule has improved steatosis and increased insulin sensitivity.⁶⁷ Fatty acid synthase (FAS), another suppressor of glucokinase, is similarly inhibited by omega-3 fatty acids.^{68, 69}

Effect of omega-3 on cell membrane composition

Omega-3 fatty acids have important properties as membrane stabilizers and can alter cell membrane fluidity.⁷⁰ Fatty acids are an essential constituent of the cell membrane, where they modulate the action of membrane-bound transporters and enzymes. Dietary intake has been shown to influence hepatocyte membrane phospholipids composition and function.⁷¹ Low levels of omega-3 in skeletal muscle phospholipids are associated with insulin resistance.⁷² They also play an important role in modifying intracellular messengers and altering intracellular functions.⁷³ In a cell-based study, it was shown that EPA increased the oxidation of endogenous fatty acids, and intracellular carnitine palmitoyltransferase-1 (CPT-1) levels, but did not inhibit lipogenesis. It also confirmed that they altered membrane composition and increased the amount of EPA in the adipocyte mitochondrial membrane.⁷⁴

Animal models

The potentially beneficial effects of omega-3 fatty acids in NAFLD are supported by findings from animal studies and summarized in Figure 2. Several animal models have been used in NAFLD studies, although none truly replicates the pathogenesis and metabolic milieu of NAFLD/NASH.⁷⁵ In addition, the quantity of omega-3 fatty acids used in these studies is in excess of that used in human studies: omega-3s generally constituted 5–20% of the animal’s total dietary intake.

Figure 2.  NAFLD pathogenesis with regards to potential omega-3 fatty acids targets.

As would be expected from the cardiovascular trials, omega-3 treatment reduces plasma lipids.^76–78 Omega-3 fatty acids have also consistently been shown to reduce hepatic steatosis in murine models. In a model of parenteral nutrition, which results in universal steatosis and abnormal liver enzymes, omega-3s protected the murine liver against hepatic steatosis, when given prophylactically.⁷⁹ The same group found omega-3 fatty acids also improved established hepatic steatosis in leptin-deficient obese mice.⁸⁰ This finding is confirmed in numerous other studies for both omega-3 mix, and EPA and DHA alone.^{76–78, 81–85}

In addition to improvement in hepatic steatosis, ob/ob mice treated with omega-3s have demonstrated markedly decreased SREBP-1 levels with consequent reduced expression of lipogenic genes such as FAS in the liver.⁸³ The suppression of SREBP-1 and genes involved in lipid metabolism have been confirmed in other publications.^{76, 78, 81, 85}

In a further series of experiments using leptin-deficient ob/ob mice, it has been shown that, in addition to confirming reduced liver triglyceride content with omega-3 supplementation, these animals had lowered plasma alanine aminotransferase (ALT) levels and improved hyperglycaemia and hyperinsulinaemia, in a manner hypothesized to be related to PPARα, although this was not measured.⁸³ The increase in PPARα in response to omega-3 supplementation was later confirmed elsewhere.^{76, 84, 86} In a study in Fischer rats, a group fed fish oil had reduced hepatic triglycerides and blunting of the normal postprandial decline in fatty acid degradation genes, such as PPARα, and the normal postprandial rise in triglyceride synthesis genes such as SREBP-1.⁸⁷

Omega-3s have also been shown to improve insulin sensitivity. In one series of experiments, there was increased expression of insulin-sensitizing genes in adipose tissue and liver (e.g. PPARγ) in mice treated with omega-3. Improved insulin sensitivity was confirmed by insulin tolerance testing. There was also increased expression of resolvins and protectins. These are omega-3-derived mediators which are insulin-sensitizing and anti-steatotic. It was hypothesized that this might contribute to how omega-3 fatty acids exert their effect, as in this study, SREBP-1 expression, TNF-α levels and IL-6 remained unchanged.⁸²

Recently, it has been reported that rats fed a high-fat diet supplemented with fish oil were protected against the severe hepatic steatosis and increased lipid peroxidation seen in a group fed the same diet without omega-3 supplementation.⁸⁸ The oxidative stress and mitochondrial function of rats fed a high-fat diet enriched with omega-3 were similar to those observed in the controls. Furthermore, there is evidence from other studies that omega-3 supplementation reduces reactive oxygen species (ROS).⁷⁶

The anti-inflammatory effects of omega-3 fatty acids are poorly characterized, but thought to be partly mediated via PPARα.⁸⁹ A murine model of NAFLD showed that omega-3 fatty acids reduced leucotriene and prostaglandins.⁹⁰ They also modulate the inflammatory response with reduced TNF-α.^{77, 84, 91, 92} Adiponectin, a powerful insulin-sensitizing agent produced in adipose tissue, is increased in animals treated with omega-3 fatty acids.^{77, 82, 93}

In a rat model, insulin resistance and central obesity were associated with increased TNF-α, decreased PPARα and adiponectin. Animals demonstrated hepatic insulin resistance and this resulted in hepatic steatosis and fibrosis. Omega-3 supplementation restored PPARα and adiponectin levels, reduced TNF-α and ameliorated hepatic steatosis and the degree of liver injury.⁸⁴

Evidence from human trials

To date, three published clinical trials have investigated the therapeutic effects of omega-3s in patients diagnosed with NAFLD. Two other studies have examined the effects of omega-3s in relation to serum adiponectin levels and hepatic steatosis. These trials were of variable duration and used differing preparations of omega-3 fatty acids; the optimal composition, dose and duration of omega-3 therapy have not yet been delineated. All reported good patient compliance with side effects being infrequent. A summary of the trial’s designs, results and methodology is presented in Tables 2 and 3.

Table 2.   Summary of trial design and results

Study	N	Inclusion criteria	Treatment arm	Treatment (n)	Control arm	Control (n)	Duration	Main results of treatment
LFT, liver function tests; ALT, alanine aminotransferase; AST, aspartate transaminase; PUFA, polyunsaturated fatty acids; NASH, non-alcoholic steatohepatitis; EPA, eicosapentaenoic acid; DHA, docosahexaenoic acid; MRS, magnetic resonance spectroscopy. * The composition of the PUFA mix was not specified. † This was a PUFA mixture of 51.4% C20:5, 23.9% C22:6, 2.8% C20:1 and 2.5% C22:0; C16:0 and C16:1 were <2%.
Capanni et al.^94	56	Fatty liver on ultrasound	1 g PUFA daily (EPA:DHA ratio: 0.9:1.5)	42	Observation only	14	12 months	Reduced hepatic steatosis on the US
								PUFA group (baseline vs. after treatment)
								15 Severe fatty liver vs. 6 severe, 6 moderate and 3 mild
								19 Moderate fatty liver vs. 6 moderate, 8 mild and 5 no fatty liver
								8 Mild fatty liver vs. 3 mild, and 5 no fatty liver.
								Control group
								At baseline 6 severe, 5 moderate and 3 mild fatty liver. No change at end of study
								Mean change in serum biochemistry in PUFA vs. Observation group:
								LFTs: ALT: −4 IU/L (s.d. 8) vs. +4 (s.d. 5) (P < 0.002); AST: −2 IU/L (s.d. 4) vs. +4 (s.d. 6) (P < 0.003); γGT: −4 IU/L (s.d. 13) vs. +4 (s.d. 12) (P < 0.03)
								Triglycerides: −46 mg/dL (s.d. 88) vs. +9 (s.d. 27) (P < 0.02)
								Fasting glucose: −6 mg/dL (s.d. 14) vs. +4 (s.d. 9) (P < 0.02)
Spadaro et al.^95	36	Fatty liver on ultrasound	2 g PUFA daily*	18	Dietary advice only	18	6 months	The US findings: steatosis grade (0/1/2/3 (%)). In the PUFA group: 0/0/39/61 at baseline vs. 33/22/44/0 after treatment. In the observation group: 0/0/44/56 at baseline vs. 0/11/50/39 at study end
								Serum biochemistry (baseline vs. after treatment) in the PUFA group
								LFTs: ALT: 56.6 U/L (s.d. 24) vs. 39.5 (s.d. 14) (P < 0.01); AST: 31.5 U/L (s.d. 13) vs. 28 (s.d. 9) (N.S.); γGT: 39.3 U/L (s.d. 26) vs. 28.0 (s.d. 17) (P < 0.05)
								Triglycerides: 147.4 mg/dL (s.d. 41) vs. 110 (s.d. 39) (P < 0.01)
								HOMA-IR: 3.5 (s.d. 2.0) vs. 2.8 (s.d. 1.7) (P < 0.05)
								TNF-α: 3.3 pg/mL (s.d. 0.5) vs. 2.7 (s.d. 0.5) (P < 0.05)
								There was no significant change in serum biochemistry (baseline vs. after treatment) in the observation group
Tanaka et al.^96	23	NASH on liver biopsy	2.7 g EPA daily	23	No control	N/A	12 months	Hepatic steatosis grade on the US changed from 2.1 ± 0.9 at baseline to 1.6 ± 1.1 after treatment (P = 0.004)
								Serum biochemistry (baseline vs. after treatment)
								LFTs: ALT: 79 IU/L (s.d. 36) vs. 50 (s.d. 20) (P = 0.002); AST: 50 IU/L (s.d. 17) vs. 36 (s.d. 15) (P = 0.001)
								Triglycerides: 201 mg/dL (s.d. 90) vs. 183 (s.d. 103) (N.S.)
								Total cholesterol: 219 mg/dL (s.d. 44) vs. 206 (s.d. 38) (P = 0.039)
								HOMA-IR: 4.0 (s.d. 2.0) vs. 3.4 (s.d. 2.4) (P = N.S.)
								Plasma thioredoxin: 30 ng/mL (s.d. 15) vs. 22 (s.d. 6) (P = 0.036)
								Change in histological grade (baseline vs. after treatment):
								Steatosis: 2.4 (s.d. 0.5) vs. 1.7 (s.d. 0.5)
								Fibrosis: 1.7 (s.d. 1.1) vs. 0.7 (s.d. 0.5)
								Lobular inflammation: 2.1 (s.d. 0.7) vs. 1.1 (s.d. 0.7)
								Ballooning: 1.6 (s.d. 0.5) vs. 0.9 (s.d. 0.4)
								NAS: 6.1 (s.d.1.3) vs. 3.7 (s.d. 1.4)
Vega et al.^98	16	Previous elevated liver fat on MRS	9 g PUFA daily†	16	No control	N/A	8 weeks	Plasma triglyceride level pre-treatment 117 mg/dL; after PUFA 74 mg/dL (P < 0.03)
								Liver fat content 7.9% pre-treatment; 8.0% after PUFA
Itoh et al.^77	52	Obesity	1.8 g EPA daily	26	Dietary advice only	26	3 months	Plasma triglycerides (means pre- and post-treatment): 166 mg/dL (S.E. 18.3) and 128 (S.E. 10.9) in the EPA group vs. 159 mg/dL (S.E. 13.3) and 151 (S.E. 12.4) in controls (P < 0.05)
								Serum adiponectin (means pre- and post-treatment): 4.65 μg/mL (S.E. 0.64) and 7.42 μg/mL (S.E. 0.52) in the EPA group vs. 5.51 μg/mL (S.E. 1.04) and 5.86 μg/mL (S.E. 0.71) in controls (P < 0.01)

Table 3.   Summary of trial methodology

Study	Primary diagnosis	Design	Randomized	Sample size calculation	Placebo	Control arm	Primary outcome measures	Histology
NASH, non-alcoholic steatohepatitis; NAFLD, non-alcoholic fatty liver disease, MR, magnetic resonance.
Capanni et al.^94	NAFLD	Open label	No Comparison group self-selected	Yes	No	Observation only	Appearance of the liver on ultrasound	No
Spadaro et al.^95	NAFLD	Open label	Yes Random number sampling	Yes	No	Dietary advice only	Appearance of the liver on ultrasound, transaminases	No
Tanaka et al.^96	NASH	Open label	No	No	No	N/A	Appearance of the liver on ultrasound, transaminases, liver histology	Yes
Vega et al.^98	Fatty liver on MR spectroscopy	Open label	No	No	No	N/A	Plasma and hepatic triglycerides	No
Itoh et al.^77	Obesity	Single blind	Yes	No	No	Dietary advice only	Serum adiponectin	No

The first trial⁹⁴ was an open-label study of 56 patients with fatty liver on ultrasound and a clinical diagnosis of NAFLD. Of these patients, 42 were treated with 1 g of N-3 PUFA daily. The 14 patients who declined treatment acted as a comparison group: they did not receive any other intervention. Groups shared similar baseline clinical and ultrasonic characteristics. The primary outcome measure of the study was the appearance of the liver on B mode ultrasound and duplex Doppler (undertaken by one ultrasonographer).

All participants completed the trial. After 12 months, there was a significant reduction in hepatic steatosis determined on ultrasound in the treatment group, with improvements in 64% of participants (either less steatosis or normal appearances). In contrast, there was no change in the prevalence or severity of steatosis in the comparator group. There were also significant improvements in serum liver function tests, fasting blood glucose and serum triglycerides only in the intervention group, while the N6:N3 ratio was reduced.

In summary, this trial provided encouraging evidence for omega-3s as a potential therapy in NAFLD, but there were design weaknesses, notably, the absence of blinding and randomization, and the use for comparison of a self-selected small group consisting of those patients who had declined entry to the treatment arm.

The second trial⁹⁵ involved 40 patients with a clinical diagnosis of NAFLD. Following recruitment, participants were assigned to two groups on a 1:1 basis using random sampling numbers, but neither participants nor investigators were blinded. Both groups received the same dietary advice. Those in the treatment arm also received 2 g of PUFA daily for 6 months. Outcome measurements included fatty liver as assessed by abdominal ultrasound, liver function tests and insulin resistance assessed by homeostasis model assessment-insulin resistance (HOMA-IR). The ultrasonographer was blinded to the treatment of the patient. The groups were satisfactorily matched for age, gender, body mass index and baseline insulin resistance.

Two patients dropped out in each arm leaving 36 to complete the trial. Outcome measures were compared pre- and postintervention. Results showed improved serum biochemistry with reduction in plasma triglycerides, γGT and ALT in the PUFA group after treatment. There was no significant difference in serum biochemistry after 6 months in the control group. The intervention group also displayed improved insulin sensitivity and decreased TNF-α. Ultrasound grading of liver fat improved in 83% of the intervention group with 33% reverting to normal appearances, whereas in the control group, 72% of participants’ steatosis scores remained unchanged and none reverted to normal.

Although this trial was superior to the first study in that there was randomization with the establishment of a control group, there were still design weaknesses – specifically lack of placebo, and the nonblinding of participants and investigators.

The most recent trial⁹⁶ included 23 patients with NASH confirmed on liver biopsy who received 2.7 g of EPA daily for 12 months. There was no control group and no sample size calculation in what was described as a pilot trial; seven participants agreed to have a repeat biopsy at the end of the treatment period. All patients had previously received dietary advice and other medications remained unchanged throughout the study. Outcome measures were serum liver biochemistry, appearance on ultrasound and liver histology. Histology was graded using the NAFLD activity score.

All patients completed the trial. At the end of the study period, the mean steatosis grade on ultrasound had improved significantly. In 6 of the 7 patients who underwent repeat biopsy, there was reduced steatosis, inflammation and fibrosis. ALT and aspartate aminotransferase significantly improved; cholesterol and free fatty acids were significantly reduced. There was no change in serum triglycerides, high-density lipoproteins, fasting blood sugar, adiponectin levels or insulin resistance. Serum TNF-α improved, but this did not reach statistical significance. Thioredoxin, a molecule associated with hepatic oxidative stress,⁹⁷ also improved.

This was the only human study of omega-3 fatty acids to have histological data, which are generally regarded as the most valid outcome measure. Whilst this trial provides further evidence of the benefits of using omega-3 fatty acids in NAFLD, the absence of randomization, controls and blinding, the small sample and the lack of power rule out reliable conclusions.

A fourth study⁹⁸ involved a 8-week treatment with 9 g of fish oil in 17 patients who had previously been enrolled and demonstrated to have elevated hepatic triglycerides on liver magnetic resonance spectroscopy as part of the Dallas Heart Study.⁹⁹ There was a 4-week run in period with placebo designed to assess compliance. One patient withdrew from the trial leaving 16 patients in the final analyses. Causes of liver disease other than NAFLD were not excluded and alcohol intake was not reported. It remains unclear whether study participants received any other interventions such as diet or lifestyle advice. Primary outcome measures were plasma and hepatic triglyceride levels, the latter assessed on magnetic resonance spectroscopy.

This study demonstrated that omega-3 fatty acids supplementation altered the fatty acid constituent of plasma triglycerides, which were themselves significantly reduced as a result of treatment. Very low-density lipoproteins were also reduced, but hepatic triglycerides were not. The authors could not explain their results with reference to the mechanisms currently believed to regulate hepatic triglyceride levels. It could be hypothesized that this result arose from it being a small, short trial, with an atypical sample (predominantly females and African Americans) whose diagnosis might not have been NAFLD anyway. There was also a skewed baseline hepatic triglyceride content and the dose of PUFA was much higher than that used in other trials.⁹⁸

Finally, a 2007 study examined the effect of EPA on serum adiponectin levels in 52 obese Japanese patients all of whom fulfilled criteria for the metabolic syndrome.⁷⁷ This was a single blinded pilot trial with patients randomized to receive either dietary advice only, or 1.8 g of EPA in addition to dietary advice for 3 months. There was no other change to participant’s medication during the trial period. This study demonstrated that EPA supplementation increased plasma adiponectin in the treatment group, but markers of NAFLD were not reported.

Conclusions

Non-alcoholic fatty liver disease is a common and growing problem worldwide.^{1, 99–100} Although frequently asymptomatic and relatively benign, NAFLD has the potential to progress to cirrhosis. Cirrhosis, when decompensated, has a poor prognosis.¹⁰¹

Currently, the mainstays of treatment are dietary advice, help and encouragement to lose weight, and exercise, and energetic treatment of co-existing disorders, especially, Type 2 diabetes and hypertension. Treatment strategies to date may be grouped into those that address weight loss, improve insulin sensitivity, are antioxidant, anti-TNF or have other mechanisms of action, but none has become an established intervention.

Omega-3 fatty acids have been suggested as a treatment for NAFLD.²⁰ They have several potential mechanisms of action, the most important being to alter hepatic gene expression, thereby switching intracellular metabolism from lipogenesis and storage to fatty acid oxidation and catabolism. There is also evidence that they improve insulin sensitivity, are anti-inflammatory and reduce TNF levels thus offering several potential therapeutic mechanisms.

Animal studies have shown promise with reduction in hepatic steatosis, improved insulin sensitivity, reduced inflammation and oxidative stress consistently reported. In humans, preliminary clinical trials have confirmed this potential with three of the four studies reporting a reduction in hepatic steatosis on imaging, increased insulin sensitivity and improved serum liver function tests. They have also confirmed the tolerability of omega-3 fatty acids, which is an important factor in a disorder which is usually asymptomatic. To date, the trials have all been open-label and none has employed a prospective, randomized, blinded, placebo-controlled, adequately powered trial methodology to submit these promising preliminary findings to proper scientific rigour. Such studies are now urgently required.

Acknowledgements

Declaration of personal interests: None. Declaration of funding interests: There has been no financial support received for this study and the authors have nothing to disclose.