European Journal of Clinical Nutrition (2010) 64, 11791185

& 2010 Macmillan Publishers Limited All rights reserved 0954-3007/10

www.nature.com/ejcn

ORIGINAL ARTICLE

Intake of n-3 polyunsaturated fatty acids and non-alcoholic fatty liver disease: a cross-sectional study in Japanese men and women
J Oya1, T Nakagami1, S Sasaki2, S Jimba1, K Murakami2, T Kasahara3, T Wasada4, H Sekiguchi3, M Hasegawa1, Y Endo3 and Y Iwamoto1
Department of Internal Medicine, Diabetes Center, Tokyo Womens Medical University School of Medicine, Tokyo, Japan; 2Department of Social and Preventive Epidemiology, School of Public Health, The University of Tokyo, Tokyo, Japan; 3Department of Health and Community Medicine, Saitama-ken Saiseikai Kurihashi Hospital, Saitama, Japan and 4Department of Health and Community Medicine, Totsuka Royal Clinic, Tokyo, Japan
1

Background/Objectives: Non-alcoholic fatty liver disease (NAFLD) is a common condition, in which abnormal amounts of triglycerides accumulate in hepatocytes and is closely related to cardiovascular diseases and diabetes. Dietary fats contribute 15% of fat accumulation in the liver and regulate hepatic lipid metabolism. The supplementation of n-3 polyunsaturated fatty acids (n-3 PUFAs) improves NAFLD. The aim of this study is to assess the cross-sectional association between dietary n-3 PUFAs and NAFLD in Japanese men and women. Subjects/Methods: Participants were middle-aged, apparently healthy, 296 men and 496 women, who did not drink alcohol and who participated in a general health check-up program. Dietary information from the previous month was obtained by the brief-type self-administered diet history questionnaire. NAFLD was diagnosed if abdominal ultrasonography revealed the presence of fatty liver. Results: The prevalence of NAFLD was 45.3% in men and 17.5% in women. In comparison with the first tertile, multivariate adjusted odds ratios (95% confidence intervals) for the presence of NAFLD in the second and third tertiles for men taking eicosapentaenoic acid (EPA) were 0.59 (0.311.14) and 0.45 (0.230.90), respectively, (P for linear trend 0.024), and the multivariate adjusted odds ratios (95% confidence intervals) for the presence of NAFLD in the second and third tertiles for men taking EPA docosahexaenoic acid (DHA) were 0.44 (0.230.86) and 0.48 (0.240.95), respectively, (P for linear trend 0.035). However, there was no significant relation between NAFLD and each of these nutrients in women. Conclusions: Dietary EPA and EPA DHA may be independent and preventive nutrients for NAFLD in Japanese men.

European Journal of Clinical Nutrition (2010) 64, 11791185; doi:10.1038/ejcn.2010.139; published online 4 August 2010
Keywords: epidemiology; Japanese; NAFLD; n-3 PUFAs

Introduction
Non-alcoholic fatty liver disease (NAFLD), a condition in which abnormal amounts of triglycerides and their intermediate metabolites (for example, long chain fatty acyl-CoAs, deacylglycerol) accumulate in hepatocytes, is the most common condition seen in a primary health care
Correspondence: Dr J Oya, Diabetes Center, Tokyo Womens Medical University School of Medicine, 8-1, Kawada-cho, Shinjuku-ku, Tokyo, 162-8666, Japan. E-mail: johya@dmc.twmu.ac.jp Received 28 July 2009; revised 14 May 2010; accepted 26 May 2010; published online 4 August 2010

setting in Japan (Kojima et al., 2003). Accumulated research data indicate that NAFLD is closely related to obesity (Chitturi et al., 1990), type 2 diabetes mellitus(Jimba et al., 2005; Goessling et al., 2008), cardiovascular diseases (Donnelly et al., 2005; Targher, 2007; Goessling et al., 2008) and insulin resistance even in non-obese, non-diabetic partcipants (Kim et al., 2004). Asian patients with NAFLD had markedly lower body mass index than Caucasian patients with NAFLD (Weston et al., 2005). Hyperinsulinemia stimulates sterol regulatory elementbinding protein (SREBP)-1, a transcription factor stimulated by insulin that regulates both the rate of de novo fatty acid synthesis and the storage of triglycerides in the liver

n-3 PUFAs and NAFLD J Oya et al

1180
(Browning and Horton, 2004). The pathophysiological mechanism of NAFLD is not fully understood. However, 60% of hepatic triglycerides arise from free fatty acids, which originate in adipose tissue, 25% comes from de novo hepatic lipogenesis and 15% comes from dietary fats (Donnelly et al., 2005). Dietary fats regulate hepatic lipid metabolism by modifying gene transcription (Sessler and Ntambi, 1998). Thus, a distinct role for dietary lipids may exist in people with NAFLD. One Israeli population-based study (Zelber-Sagi et al., 2007) reported a negative correlation between dietary n-3 polyunsaturated fatty acids (n-3 PUFAs) and NAFLD in a subgroup analysis. N-3 PUFAs might inhibit accumulation of triglycerides by modifying hepatic lipid metabolism (Yoshikawa et al., 2002) and could reduce inflammatory and oxidative status (Itoh et al., 2007, Perez-Matute et al., 2007). a-Linolenic acid is a major n-3 PUFA. It is not synthesized in vivo but synthesized from foods, and is converted to docosahexaenoic acid (DHA) through eicosapentaenoic acid (EPA). Typical Japanese diet contains a lot of n-3 PUFAs: edible vegetable oils such as rapeseed oil and soybean oil supply a-linolenic acid, whereas fish oils supply DHA and EPA (Sugano and Hirahara, 2000). However, the relation between n-3 PUFAs and NAFLD has not yet been examined in Japanese participants. A Japanese population may be suitable for investigating the association between NAFLD and dietary PUFAs, especially marine-origin long-chain PUFA such as EPA and DHA following two reasons. The prevalence of NAFLD is dramatically increased for a past decade in Japan (Kojima et al., 2003), though the mean body mass index is still much lower than Caucasians. As Japanese generally eat more fish than Westerners do (Sugano and Hirahara, 2000), whereas some Japanese eat very little fish, it gives us a wide distribution of intakes of fish and marine-origin long-chain PUFAs. Thus, the aim of this study was to examine the crosssectional impact of dietary n-3 PUFAs, especially a-linolenic acid, EPA and DHA, on the presence of NAFLD in Japanese men and women. were later excluded from the study when the brief-type, self-administered diet history questionnaire (BDHQ) indicated that their estimated alcohol consumption was greater than 0 g per day. The remaining 792 participants used for the analysis consisted of 571 participants having a normal liver and 221 participants having NAFLD.

Research design and methods

Study subjects From February 2006 to January 2007, Saitama-ken Saiseikai Kurihashi Hospital conducted a health check-up program in which 4138 participants participated and followed in the Kurihashi Lifestyle Cohort Study; 80.9% of them completed dietary questionnaires. Of the health check-up program participants, 3285 participants underwent abdominal ultrasonography scans. Participants were excluded if they had diabetes, cardiovascular diseases, chronic viral hepatitis (hepatitis B surface antigen, anti-hepatitis C virus antibody), or if they were under pharmacological treatment for hypertension or dyslipidemia. This reduced the study sample to 2624 participants. However, 1832 of these participants European Journal of Clinical Nutrition

Dietary assessment The participants typical dietary habits, mainly the nutrient intake from the previous month, were assessed with the validated BDHQ (Sasaki, 2004). The BDHQ based on the self-administered diet history questionnaire (Sasaki et al., 1998). The BDHQ is a four-page structured questionnaire that consists of questions regarding general dietary behaviors, major cooking methods, the quantity and the frequency of the consumption of five alcoholic beverages (sake, beer, syocyu, whiskey and wine), the quantity and frequency of consuming 58 selected food and non-alcoholic beverage items, use of dietary supplements, the amount of rice and miso (fermented soybean paste) soup consumed daily and eating speed. The food and beverage items and portion size used in the BDHQ were derived primarily from data in the National Nutrition Survey of Japan and several recipe books for Japanese dishes (Sasaki et al., 1998). For fish, the following six questions were used: squid, octopus, shrimp and clam, small fish with bones, dried fish and salted fish, canned tuna, oily fish and lean fish. Sum of EPA and DHA concentrations (g) per portion for each fish group was as follows: 0.13, 0.69, 0.21, 0.96, 1.93 and 0.47 for women, and 0.15, 0.79, 0.23, 1.09, 2.20 and 0.54 for men (Supplementary Table 1). We examined validity of the BDHQ against 16-day weighed dietary records as the gold standard using 92 Japanese men and 92 Japanese women aged 3176 years. Pearsons correlation coefficients in 92 Japanese men and 92 Japanese women aged 3176 years were, respectively, 0.42 and 0.45 for n-3 PUFAs (both Po0.001), 0.30 and 0.32 (both Po0.01) for a-linolenic acid, 0.27 (Po0.01) and 0.37 (Po0.001) for EPA, and 0.26 (Po0.05) and 0.27 (Po0.01) for DHA (Sasaki, 2004). Before this study, the BDHQ was further validated with EPA, DHA and EPA DHA in serum phospholipid concentrations using 91 Japanese men and 91 Japanese women aged 3069 years. Pearsons correlation coefficients between EPA, DHA and EPA DHA intake and the serum phospholipid concentrations were 0.38 (Po0.001) and 0.33 (Po0.001) for EPA, 0.36 (Po0.001) and 0.27 (Po0.01) for DHA, and 0.37 (Po0.001) and 0.31 (Po0.01) for EPA DHA (Sasaki, 2004; Supplementary Table 2).

Diagnosis of NAFLD NAFLD was diagnosed when a participants estimated consumption of alcohol was 0 g per day according to the BDHQ (Sasaki et al., 1998) and if the participant met the ultrasonographic criteria (Gore, 1994) for a fatty liver.

n-3 PUFAs and NAFLD J Oya et al

1181
Other variables The general health check-up examination procedure in Saitama-ken Saiseikai Kurihashi Hospital included biochemical laboratory tests and a self-administered questionnaire regarding a participants smoking status, physical activities during leisure time and work, past medical history and family medical history and, in women, menopause status. Physical activities during leisure time and work were classed as either sedentary or active. Blood samples were collected in the morning after a 10-h fast. The levels of aspartate aminotransferase (obtained by the leucocoloringmatter method), alanine aminotransferase (obtained by the ultraviolet method) and g-glutamyl transpeptidase activity (obtained by the L-g-glutamyl-p-nitroanilide method) were measured in the hospital laboratory. All participants were measured for height, weight and waist circumference. Body height was measured to the nearest 0.1 cm with the participant standing without shoes. Body weight in light indoor clothes was measured to the nearest 0.1 kg. Waist circumference was measured to the nearest 0.1 cm at the navel level at the end of the expiration of a normal breath and with the participant in a standing position. Body mass index was calculated as the body weight (in kg) divided by the body height squared (m). Ultrasonographic examinations were carried out by senior technicians using the Ultrasound Equipment Model SSA-250A (Toshiba Corporation, Tokyo, Japan). The technicians were blinded to the laboratory data.

Results
Demographic, lifestyle and nutritional characteristics in NAFLD The overall prevalence of NAFLD was 45.3% (134 out of 296) in men and 17.5% (87 out of 496) in women. Participants with NAFLD were more obese and had higher levels of liver enzymes than those with a normal liver in both sexes. A high proportion of women participants with NAFLD ate quickly and had reached menopause compared with women with a normal liver (Table 1). As for nutritional components, there was a statistical difference in the dietary intake of n-3 PUFAs and EPA in men diagnosed with NAFLD versus men who had a normal liver (Table 1). However, no statistical difference was found between each of these nutrients and NAFLD in women.

n-3 PUFAs and NAFLD The univariate logistic regression model showed that, in men, n-3 PUFAs, EPA, DHA and EPA DHA had a statistically significant linear negative correlation with NAFLD (all P for trends o0.05) (Table 2). This statistical significance for EPA and EPA DHA remained even after adjusting for potential confounding factors (both P for trends o0.05) (Table 2). The odds ratios associated with the presence of NAFLD decreased as dietary levels of EPA and EPA DHA increased. The odds ratio decreased significantly in the third tertile group in comparison with the first tertile for EPA and EPA DHA by 55 and 52%, respectively (Table 2). By contrast, no significant relation was found between each of these nutrients and NAFLD in women.

Statistical analysis Statistical analyses were carried out and reported separately for men and women. The chi-square test was used to compare proportions and Students t-test was used to compare the means between the two groups. The crude and multivariate adjusted odds ratios and their 95% confidence intervals associated with the presence of NAFLD were calculated for each tertile category of selected nutrients (n-3 PUFAs, a-linolenic acid, EPA and DHA) by using logistic regression models. In the multivariate model, age (continuous), waist circumference (continuous), physical activity during leisure time (categorical), whether or not the participant ate quickly (categorical), whether or not the participant smoked (categorical), and (in women only) menopause status (categorical) were included as confounding factors. The association trend was assessed by a logistic regression model that assigned consecutive integers to the levels of selected nutrients. SPSS for Windows (version 14.0, Chicago, IL, USA) was used for all statistical analyses. All reported P values are twotailed and Po0.05 was considered statistically significant. The study was approved by the Institutional Review Board of Saitama-ken Saiseikai Kurihashi Hospital and informed consent to participate was obtained from the study participants.

Conclusions
This study has shown that dietary n-3 PUFAs and a-linolenic acid, which constitutes high proportion of dietary n-3 PUFAs were not independent risk factors for NAFLD. However, dietary EPA and EPA DHA were identified as linear, independent and preventive nutrients for NAFLD in Japanese men who generally consume more fish than Westerners do (Sugano and Hirahara, 2000). Previous studies have reported that a higher intake of soft drinks, meat, protein and saturated fatty acid and a lower intake of fiber, vitamin C and E were related to an increased risk for NAFLD (Zelber-Sagi et al., 2007, Musso et al., 2003). Increased intake of carbohydrate was also associated with hepatic de novo lipogenesis and hepatic inflammation (Schwarz et al., 1995, Solga et al., 2004). In this study, all these possible foods or nutrients raised above were analyzed, but not identified as risk factors for NAFLD. Allard et al. (2008) reported that not self-reported dietary intake but levels of n-3 PUFAs in hepatocytes was correlated with the degree of liver steatosis in a cross-sectional study. Zelber-Sagi et al. (2007) reported a negative relation between dietary n-3 PUFAs and NAFLD in Israeli participants; European Journal of Clinical Nutrition

n-3 PUFAs and NAFLD J Oya et al

1182
Table 1 The demographic, lifestyle and nutritional characteristics among Japanese with a normal liver and NAFLD and whose estimated alcohol intake is 0 g per day Men Normal liver Number Age (years) Body mass index (kg/m2) Waist circumference (cm) Aspartate aminotransferase (U/l) Alanine aminotransferase (U/l) g-Glutamyl transpeptidase (U/l) Eating quickly (%) Supplement users (%) Physically active during leisure time (%) Physically active during work (%) Current smoker (%) Menopause (%) Dietary intake Energy (kcal per day) Carbohydrate (% energy) Total fat (% energy) Saturated fatty acid (% energy) n-3 Polyunsaturated fatty acids (% energy) a-Linolenic acid (% energy) Ecosapentaenoic acid (% energy) Docosahexaenoic acid (% energy) 162 529 22.22.5 80.67.0 287 239 2517 46.9 24.7 62.3 38.3 42.0 1994493 25.05.0 59.06.4 6.221.61 1.370.41 0.870.26 0.150.09 0.250.13 NAFLD 134 518 25.02.8 88.46.8 3314 3926 3944 49.3 27.6 59.0 34.3 41.0 1952520 24.45.2 60.16.9 6.131.55 1.280.40 0.450.28 0.130.08 0.220.13 P-value Normal liver 409 519 21.72.4 77.08.1 27.09 2011 1817 39.1 36.9 56.0 36.2 6.1 52.0 1832475 26.84.6 57.76.0 6.791.56 1.420.39 0.930.25 0.150.09 0.250.14 Women NAFLD 87 547 25.93.6 88.98.2 30.711 3020 2724 56.3 33.3 47.1 27.6 5.7 68.2 1869525 26.45.1 58.16.8 6.911.71 1.360.39 0.900.27 0.140.08 0.230.12 P-value

0.649 o0.001 o0.001 o0.001 o0.001 0.001 0.727 0.596 0.632 0.545 0.906

0.006 o0.001 o0.001 0.004 o0.001 0.001 0.004 0.624 0.155 0.138 1.000 0.008

0.486 0.316 0.141 0.632 0.047 0.451 0.039 0.072

0.513 0.518 0.515 0.524 0.161 0.219 0.228 0.226

Abbreviation: NAFLD, non-alcoholic fatty liver disease. Continuous variables are presented as the meanthe s.d.s and categorical variables are presented as a percentage of subjects. The values of the continuous and categorical variables were assessed with the independent t-test and the chi-square test, respectively. The normal range for aspartate aminotransferase is 1059 U/l: for alanine aminotransferase, 756 U/l: and g-glutamyl transpeptidase, 1258 U/l.

however, this relation was of borderline significance. In this study, the amount of dietary intake of n-3 PUFAs showed an inverse relation with the presence of NAFLD in men. However, this was no longer statistically significant after adjusting for confounding factors. The EPA and EPA DHA, which are contained in high amounts in fish, had independent, inverse and linear relations to NAFLD in Japanese men. By contrast, no significant relation was found between n-3 PUFAs and NAFLD in Japanese women. The reason for this is not known. However, the proportion of NAFLD in women is less than half of that in men as reported by Weston et al. (2005), and thus these findings in women might be weakened by the sample size. An alternative explanation is that there might be unidentified preventive factors for NAFLD in women, that is, sex hormones. A higher rate of NAFLD was reported in women after menopause than before menopause (Clark et al., 2002). High risk of developing NAFLD was also reported in women treated with potent aromatase inhibitor for breast cancer (Nishino et al., 2003). In this study, however, the results for women in the multivariate model did not change before and after adjustment for menopause. In this study, DHA was identified as an important preventive nutrient for NAFLD, although not statistically significant in the multivariate model. However, EPA changes to DHA in vivo, and thus the independent impact of DHA on NAFLD is inconclusive. European Journal of Clinical Nutrition

Randomized controlled clinical trials have shown the beneficial effect of n-3 PUFAs supplementation on NAFLD (Capanni et al., 2006; Spadaro et al., 2008; Tanaka et al., 2008). Capanni et al. reported that EPA DHA supplementation (at a dose of 1.0 g per day) for 12 months decreased the levels of liver enzymes and reversed liver steatosis to normal or reduced the degree of steatosis in 64% of patients with NAFLD. Similarly, Spadaro et al. showed that the daily supplementation of n-3 PUFAs (taken in a 2.0 g capsule) improved serum aspartate aminotransferase, alanine aminotransferase, tumor necrosis factor-a levels; decreased insulin resistance in patients with NAFLD and completely reversed steatosis in 33.4% of patients with NAFLD. However, Vega et al. (2008) found that 8 weeks supplementation of a high dose of n-3 PUFAs (at a dose of 9 g per day) failed to reduce hepatic triglyceride content in spite of a marked reduction in plasma triglyceride levels. This might be because of the small sample size or the low intake of PUFAs in the study participants. Whereas Tanaka et al. showed that supplementation of a highly purified EPA at a dose of 2.7 g per day for 12 months decreased the serum levels of free fatty acids, alanine aminotransferase and inflammatory cytokines, and improved hepatic steatosis and fibrosis in non-alcoholic steatohepatitis Japanese patients. This result seems to be in accordance with these findings from cross-sectional observational data, although confirmed evidence has not yet been accumulated.

Table 2 Odds ratios with the 95% confidence intervals for NAFLD according to the tertile of selected dietary intake in Japanese men Men (n 296) T1 (n 97) P for trend 1.01 1.42 0.96 52/45 1.00 (ref) 1.00 (ref) 0.033 0.055 0.63 48/49 1.00 (ref) 1.00 (ref) 0.287 0.581 0.07 55/42 1.00 (ref) 1.00 (ref) 0.001 0.024 0.13 55/42 1.00 (ref) 1.00 (ref) 0.007 0.101 0.20 58/39 1.00 (ref) 1.00 (ref) 40/60 0.47 (0.270.83) 0.44 (0.230.86) 0.33 0.57 36/63 0.39 (0.220.70) 0.48 (0.240.95) 0.21 41/59 0.52 (0.290.91) 0.47 (0.240.92) 0.35 38/61 0.46 (0.260.81) 0.57 (0.291.12) 0.12 45/56 0.60 (0.341.05) 0.59 (0.311.14) 0.22 34/65 0.39 (0.220.70) 0.45 (0.230.90) 0.07 32/133 1.00 (ref) 1.00 (ref) 0.13 33/132 1.00 (ref) 1.00 (ref) 0.20 0.81 45/55 0.87 (0.501.52) 0.82 (0.421.57) 1.10 41/58 0.74 (0.421.29) 0.83 (0.431.61) 0.70 37/128 1.00 (ref) 1.00 (ref) 44/56 0.68 (0.391.19) 0.66 (0.341.28) 38/61 0.54 (0.310.95) 0.51 (0.261.01) 33/132 1.00 (ref) 1.00 (ref) 32/134 0.79 (0.461.36) 0.81 (0.411.58) 0.90 24/142 0.59 (0.331.03) 0.77 (0.391.55) 0.13 29/137 0.89 (0.511.56) 0.99 (0.491.98) 0.23 26/140 0.74 (0.421.30) 0.75 (0.371.51) 0.36 1.23 1.70 T2 (n 100) T3 (n 99) T1 (n 165) T2 (n 166) Women (n 496) T3 (n 165) 1.80 22/143 0.65 (0.361.18) 0.66 (0.321.36) 1.15 26/139 0.64 (0.371.12) 0.73 (0.371.46) 0.23 26/139 0.77 (0.431.35) 0.75 (0.371.53) 0.37 28/137 0.81 (0.461.41) 0.81 (0.401.62) 0.59 P for trend

n-3 Polyunsaturated fatty acids (n-3 PUFAs) (% energy, median) Subjects with/without NAFLD Crude OR (95% CI) Multivariate OR (95% CI)

0.152 0.257

a-Linolenic acid (% energy, median) Subjects with/without NAFLD Crude OR (95% CI) Multivariate OR (95% CI)

0.107 0.375

Eicosapentaenoic acid (% energy, median) Subjects with/without NAFLD Crude OR (95% CI) Multivariate OR (95% CI)

n-3 PUFAs and NAFLD J Oya et al

0.360 0.433

Docosahexaenoic acid (% energy, median) Subjects with/without NAFLD Crude OR (95% CI) Multivariate OR (95% CI)

0.439 0.550

Eicosapentaenoic acid Docosahexaenoic acid (% energy, median) Subjects with/without NAFLD Crude OR (95% CI) Multivariate OR (95% CI)

0.001 0.035

34/131 1.00 (ref) 1.00 (ref)

27/139 0.75 (0.431.31) 0.73 (0.371.46)

26/139 0.72 (0.411.27) 0.68 (0.341.38)

0.248 0.282

Abbreviations: CI, confidence interval; NAFLD, non-alcoholic fatty liver disease; OR, odds ratio; T, tertile; ref, reference. Multivariate models included age, waist circumference, smoking status, physical activity level during leisure time, and eating speed in men; and waist circumference, smoking status, physical activity level during leisure time, eating speed, and menopause status in women.

1183

European Journal of Clinical Nutrition

n-3 PUFAs and NAFLD J Oya et al

1184
Recently, it is noted that there are three types of fatty liver related to lifestyle: obesity (or hyperinsulinemia) induced, carbohydrate (or sucrose) induced, high-fat-diet induced. The obesity (or hyperinsulinemia) or carbohydrate (or sucrose) induced fatty liver are mediated by increases in SREBP-1c (Yamazaki et al., 2007). Thus, these types of fatty liver may be effectively prevented by fish oil (Yamazaki et al., 2007). Because n-3 PUFAs, which are contained in fish oil, are the natural ligands of the peroxisome proliferator-activated receptor-a, which modulate lipid metabolism in hepatocytes. N-3 PUFAs stimulate peroxisome proliferator-activated receptor-a activity, which leads to an increase of triglyceride transportation from hepatocytes. N-3 PUFAs suppress hepatic lipogenesis through the downregulation of SREBP-1 expression and/or impairment of SREBP-1 processing, which inhibit the transcription of lipogenic genes (Yoshikawa et al., 2002). An elevated oxidative stress or proinflammatory cytokines may lead to hepatocyte inflammation and fibrosis (Jarrar et al., 2007). However, n-3 PUFAs reduced inflammatory and oxidative status and increased adiponectin and insulin sensitivity (Itoh et al., 2007, Perez-Matute et al., 2007) and EPA may increase adiponectin secretion possibly through the improvement of the inflammatory changes in adipose tissue directly (Itoh et al., 2007). In this study, however, it was unable to measure serum insulin and cytokines or hormones. On the other hand, fish oil did not affect high saturated fat diet (butter)-induced fatty liver but did exacerbate high linoleic acid diet (safflower)-induced fatty liver in an animal study (Yamazaki et al., 2007). The latter may partly be explained by the fact that n-3 PUFAs did not alter expression of SREBP-1c but increased expression of peroxisome proliferator-activated receptor-g in the liver (Yamazaki et al., 2007), accelerating fatty liver. So far, several possible pathophysiological relations between n-3 PUFAs, EPA, DHA and NAFLD have been reported. However, the reduced intake of EPA or DHA in the general population has not been previously identified as a causal risk factor for NAFLD. Further, prospective and intervention studies are needed to confirm this association in different populations. Nevertheless, dietary modification to increase intake of EPA or EPA DHA might be effective for preventing NAFLD in Japanese men, who generally consume more fish than Westerners (Sugano and Hirahara, 2000). There are several limitations to this study. First, the diagnosis of NAFLD was not based on liver biopsy. Although computed tomography is considered a reliable technique to assess steatosis (Saadeh et al., 2002), its use is limited because of radiation exposure. Magnetic resonance imaging is also an accurate technique for diagnosing fatty liver disease, but it is expensive. By contrast, ultrasonography is commonly carried out to detect a fatty liver, as it is sensitive, cheap, non-invasive and easy to carry out. The sensitivity of ultrasonography ranges from 60 to 94% and specificity from 84 to 95% (Foster et al., 1980; Debongnie et al., 1981; European Journal of Clinical Nutrition Saverymuttu et al., 1986; Joseph et al., 1991). However, the diagnosis of fatty liver by ultrasonography was insensitive when the fat content of the liver was o33% (Saadeh et al., 2002). Thus, the use of ultrasonography to diagnose NAFLD may have induced the misclassification of individuals. Second, the intra- and inter-variability of the diagnosis of fatty liver disease was not examined in advance of this study. However, the single-observer reliability and inter-observer reliability were reported as 0.95 and 0.95 (Hamaguchi et al., 2007). Third, we assessed participants for their dietary habits with the BDHQ. The BDHQ does not directly observe the dietary intake of participants. The results should therefore be cautiously interpreted. Fourth, we could not directly assess the relation between hormones including insulin and NAFLD because of the lack of data. Fifth, this cross-sectional study design could not avoid static and susceptible reverse causality and other systematic biases. However, at least, it is hard to imagine that the participants with NAFLD increased their n-3 PUFAs intakes because the guideline for management of NAFLD in Japan (The Japan Society of Hepatology, 2006) does not always recommend high amounts of n-3 PUFAs intake in patients with NAFLD. To avoid systematic bias as much as possible, we were very careful for the data collection. The data collections of dietary habits were done independently without any information contamination. Despite of its limitations, this study has notable strengths. It consists of both men and women participants and nutritional information was obtained by the validated BDHQ, although the validity remained to be examined more carefully, in a well-phenotyped Japanese sample. Men who completed the dietary questionnaire were older than those who did not complete it. However, the metabolic risk factors were not statistically different between men and women who completed the dietary questionnaire and men and women who did not complete the dietary questionnaire in this study. In conclusion, this cross-sectional observational study has shown that dietary EPA and EPA DHA may be independent and preventive nutrients for NAFLD in Japanese men. These results should be further confirmed in prospective and intervention studies.

Conflict of interest
The authors declare no conflict of interest.

Acknowledgements
This analysis has been carried out with the support of a grant from the Japan Medical Womens Association, the Tokyo Womens Medical University Association, the Yayoi Yoshioka Research Fund and the Yazuya Food and Health Research Foundation for TN.

n-3 PUFAs and NAFLD J Oya et al

1185

References
Allard JP, Aghdassi E, Mohammed S, Raman M, Avand G, Arendt BM et al. (2008). Nutritional assessment and hepatic fatty acid composition in non-alcoholic fatty liver disease (NAFLD): a cross-sectional study. J Hepatol 48, 300307. Browning JD, Horton JD (2004). Molecular mediators of hepatic steatosis and liver injury. J Clin Invest 114, 147152. Capanni M, Calella F, Biagini MR, Genise S, Raimondi L, Bedogni G et al. (2006). Prolonged n-3 polyunsaturated fatty acid supplementation ameliorates hepatic steatosis in patients with nonalcoholic fatty liver disease: a pilot study. Aliment Pharmacol Ther 23, 11431151. Chitturi S, Abeygunasekera S, Farrell GC, Holmes-Walker J, Hui JM, Fung C et al. (1990). NASH and insulin resistance: insulin hypersecretion and specific association with the insulin resistance syndrome. Hepatology 12, 11061110. Clark JM, Brancati FL, Diehl AM (2002). Nonalcoholic fatty liver disease. Gastroenterology 122, 16491657. Debongnie JC, Pauls C, Fievez M, Wibin E (1981). Prospective evaluation of the diagnostic accuracy of liver ultrasonography. Gut 22, 130135. Donnelly KL, Smith CI, Schwarzenberg SJ, Jessurun J, Boldt MD, Parks EJ (2005). Sources of fatty acids stored in liver and secreted via lipoproteins in patients with nonalcoholic fatty liver disease. J Clin Invest 115, 13431351. Foster KJ, Dewbury KC, Griffith AH, Wright R (1980). The accuracy of ultrasound in the detection of fatty infiltration of the liver. Br J Radiol 53, 440442. Goessling W, Massaro JM, Vasan RS, DAgostino Sr RB, Ellison RC, Fox CS (2008). Aminotransferase levels and 20-year risk of metabolic syndrome, diabetes, and cardiovascular disease. Gastroenterology 135, 19351944. Gore RM (1994). Diffuse liver disease. In: Gore RM, Levine MS, Laufer I. Textbook of Gastrointestinal Radiology. Saunders: Philadelphia. Hamaguchi M, Kojima T, Itoh Y, Harano Y, Fujii K, Nakajima T et al. (2007). The severity of ultrasonographic findings in nonalcoholic fatty liver disease reflects the metabolic syndrome and visceral fat accumulation. Am J Gastroenterol 102, 27082715. Itoh M, Suganami T, Satoh N, Tanimoto-Koyama K, Yuan X, Tanaka M et al. (2007). Increased adiponectin secretion by highly purified eicosapentaenoic acid in rodent models of obesity and human obese subjects. Arterioscler Thromb Vasc Biol 27, 19181925. Jarrar MH, Baranova A, Collantes R, Ranard B, Stepanova M, Bennett C et al. (2007). Adipokines and cytokines in non-alcoholic fatty liver disease. Aliment Pharmacol Ther 27, 412421. Jimba S, Nakagami T, Takahashi M, Wakamatsu T, Hirota Y, Iwamoto Y et al. (2005). Prevalence of non-alcoholic fatty liver disease and its association with impaired glucose metabolism in Japanese adults. Diabet Med 22, 11411145. Joseph AEA, Saverymuttu SH, Al-Sam S, Cook MG, Maxwell JD (1991). Comparison of liver histology with ultrasonography in assessing diffuse parenchymal liver diseases. Clin Radiol 43, 2631. Kim HJ, Kim HJ, Lee KE, Kim DJ, Kim SK, Ahn CW et al. (2004). Metabolic significance of nonalcoholic fatty liver disease in nonobese, nondiabetic adults. Arch Intern Med 164, 21692175. Kojima S, Watanabe N, Numata M, Ogawa T, Matsuzaki S (2003). Increase in the prevalence of fatty liver in Japan over the past 12 years: analysis of clinical background. J Gastroenterol 38, 954961. Madsen T, Skou HA, Hansen VE, Fog L, Christensen JH, Toft E et al. (2001). C-reactive protein, dietary n-3 fatty acids, and the extent of coronary artery disease. Am J Cardiol 88, 11391142. Nishino M, Hayakawa K, Nakamura Y, Morimoto T, Mukaihara S (2003). Effects of tamoxifen on hepatic fat content and the

development of hepatic steatosis in patients with breast cancer: high frequency of involvement and rapid reversal after completion of tamoxifen therapy. AJR Am J Roentgenol 180, 129134. Perez-Matute P, Perez-Echarri N, Martinez JA, Marti A, Moreno-Aliaga MJ (2007). Eicosapentaenoic acid actions on adiposity and insulin resistance in control and high-fat-fed rats: role of apoptosis, adiponectin and tumour necrosis factor-alpha. Br J Nutr 97, 389398. Saadeh S, Younossi ZM, Remer EM, Gramlich T, Ong JP, Hurley M et al. (2002). The utility of radiological imaging in non-alcoholic fatty liver disease. Gastroenterology 123, 745750. Sasaki S (2004). Development and evaluation of dietary assessment methods using biomarkers and diet history questionnaires for individuals. (head investigator: Tanaka H) Research for Evaluation Methods of Nutrition and Dietary Lifestyle Programs held on Healthy Japan 21. Summary Report of the Ministry of Health, Labour and Warfare. Ministry of Health, Labour and Warfare: Japan, pp 1044 (in Japanese). Sasaki S, Yanagibori R, Amano K (1998). Self-administered diet history questionnaire developed for health education: a relative validation of the test-version by comparison with 3-day diet record in women. J Epidemiol 8, 203215. Saverymuttu SH, Joseph AEA, Maxwell JD (1986). Ultrasound scanning in the detection of hepatic fibrosis and steatosis. BMJ 292, 1315. Schwarz JM, Neese RA, Turner S, Dare D, Hellerstein MK (1995). Short-term alterations in carbohydrate energy intake in humans. J Clin Invest 96, 27352743. Sessler AM, Ntambi JM (1998). Polyunsaturated fatty acid regulation of gene expression. J Nutr 128, 923926. Solga S, Alkhuraishe AR, Clark JM, Torbenson M, Greenwald A, Diehl AM et al. (2004). Dietary composition and nonalcoholic fatty liver disease. Dig Dis Sci 49, 15781583. Spadaro L, Magliocco O, Spampinato D, Piro S, Oliveri C, Alagona C et al. (2008). Effects of n-3 polyunsaturated fatty acids in subjects with nonalcoholic fatty liver disease. Dig Liver Dis 40, 194199. Sugano M, Hirahara F (2000). Polyunsaturated fatty acids in the food chain in Japan. Am J Clin Nutr 71 (Suppl), S189S196. Tanaka N, Sano K, Horiguchi A, Tanaka E, Kiyosawa K, Aoyama T (2008). Highly purified eicosapentaenoic acid treatment improves nonalcoholic steatohepatitis. J Clin Gastroenterol 42, 413418. Targher G (2007). Non-alcoholic fatty liver disease, the metabolic syndrome and the risk of cardiovascular disease: the plot thickens. Diabet Med 24, 16. The Japan Society of Hepatology (2006). Guideline for treatment of NASH/NAFLD. In: Okaue T (ed). Bunkodo: Tokyo. pp 4047 (in Japanese). Vega GL, Chandalia M, Szczepaniak LS, Grundy SM (2008). Effects of n-3 fatty acids on hepatic triglyceride content in humans. J Investig Med 56, 780785. Weston SR, Leyden W, Murphy R, Bass NM, Bell BP, Manos MM et al. (2005). Racial and ethnic distribution of nonalcoholic fatty liver in persons with newly diagnosed chronic liver disease. Hepatology 41, 372379. Yamazaki T, Nakamori A, Sasaki E, Wada S, Ezaki O (2007). Fish oil prevents sucrose-induced fatty liver but exacerbates high-safflower oil-induced fatty liver in ddY mice. Hepatology 46, 17791790. Yoshikawa T, Shimano H, Yahagi N, Ide T, Matsuzaka T, Nakakuki M et al. (2002). Polyunsaturated fatty acids suppress sterol regulatory element-binding protein 1c promoter activity by inhibition of liver X receptor (LXR) binding to LXR response elements. J Biol Chem 277, 17051711. Zelber-Sagi S, Nitzan-Kaluski D, Goldsmith R, Webb M, Blendis L, Halpern Z et al. (2007). Long term nutritional intake and the risk for non-alcoholic fatty liver disease (NAFLD): A population based study. J Hepatol 47, 711717.

Supplementary Information accompanies the paper on European Journal of Clinical Nutrition website (http://www.nature.com/ejcn)

European Journal of Clinical Nutrition

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.