Noora Kanerva, Samuel Sandboge, Niina E Kaartinen, Satu Ma¨nnisto¨, and Johan G Eriksson ABSTRACT Background: High fructose intake has been suggested to be a key factor that induces nonalcoholic fatty liver disease (NAFLD), but the evidence from large epidemiologic studies is lacking. Objective: We examined the cross-sectional association between fructose intake and NAFLD by using the Fatty Liver Index (FLI) and the NAFLD liver fat score. Design: The Helsinki Birth Cohort Study investigated 2003 Finnish men and women born in 1943–1944 in Helsinki who participated in a clinical health examination in the years 2001–2004. Trained study nurses measured weight, height, and waist circumference, and body mass index was calculated. Laboratory staff drew fasting blood for measurements of triglycerides and g-glutamyl-transferase. The FLI and the NAFLD liver fat score were calculated on the basis of these measurements. Habitual fructose and other dietary intake over the past year were assessed by using validated and standardized 131-item food-frequency questionnaires. Data were analyzed in a cross-sectional manner by using logistic regression modeling with statistical software. Results: In a model adjusted for age, sex, and energy intake, participants in the highest fructose intake quartile (range: 29.2–88.0 g/d) had lower risk of NAFLD assessed by using the FLI (OR: 0.56; 95% CI: 0.42, 0.75; P-trend , 0.001) and NAFLD liver fat score (OR: 0.72; 95% CI: 0.53, 0.99; P-trend , 0.001) than that of the lowest intake quartile (range: 2.2–15.2 g/d). This association remained after adjustment for educational attainment, smoking, physical activity, and other dietary variables only for the FLI (OR: 0.68; 95% CI: 0.47, 0.84; P-trend , 0.05). Conclusion: Our cross-sectional results did not support the current hypothesis that high intake of fructose is associated with a higher prevalence of NAFLD as assessed by using the FLI and NAFLD liver fat score. Am J Clin Nutr 2014;100:1133–8.
INTRODUCTION
Nonalcoholic fatty liver disease (NAFLD)5, which is the hepatic manifestation of the metabolic syndrome (MetS) (1, 2), is the most common cause of chronic liver disease in the developed world with a prevalence ranging from 20% to 30% (3, 4). NAFLD is present when the hepatic fat content exceeds 5– 10% of liver mass in the absence of other causes of steatosis such as other liver disorders, excessive alcohol intake, and use of hepatotoxic medication (5). The condition is closely associated with type 2 diabetes and obesity (6, 7) and, in addition, is an independent risk factor for cardiovascular disease (8). A continuum of conditions is included in the NAFLD diagnosis
from simple steatosis via nonalcoholic steatohepatitis to fibrosis and cirrhosis (9). The pathogenesis of NAFLD has been described by a “multihit” hypothesis, the key pathogenic mechanism of which is insulin resistance (1, 9). Hepatic lipid accumulation (ie, steatosis) represents the first hit, whereas several second hits (eg, inflammatory cytokines) are involved in the progression of the disease toward nonalcoholic steatohepatitis and beyond. Several factors have been implicated in the development of NAFLD and its progression, including carbohydrate intake. Fructose intake has received special attention, and several previous studies have linked high fructose intake with NAFLD (10). Fructose stimulates de novo lipogenesis in the liver that leads to the production of triglycerides, which, in turn, can be secreted as VLDLs. If this process is impaired, triglycerides can accumulate intracellularly and led to steatosis (11). There have been some studies that indicated that fructose intake is increased in individuals with NAFLD, but the evidence has been inconsistent (12–14). We have previously studied the prevalence of NAFLD in the Helsinki Birth Cohort Study (HBCS) from a developmental origins of health and disease perspective (15). In that study, we used an algorithm test, the NAFLD liver fat score and equation (16), to define NAFLD. We showed that a small body size during childhood was associated with adult NAFLD, especially in 1
From the Department of Chronic Disease Prevention, National Institute for Health and Welfare, Helsinki, Finland (NK, SS, NEK, SM, and JGE); the Departments of Public Health (NK) and General Practice and Primary Health Care (JGE), University of Helsinki, Helsinki, Finland; the Folkha¨lsan Research Center, Helsinki, Finland (SS and JGE); the Helsinki University Central Hospital, Unit of General Practice, Helsinki, Finland (JGE); and the Vasa Central Hospital, Vasa, Finland (JGE). 2 Funding sources had no role in the design, implementation, analysis, or interpretation of the study 3 Supported by the Academy of Finland [grants 120386 and 125876 (to JGE) and 136895 and 263836 (to SM)]; the Finnish Diabetes Research Society; Liv and Ha¨lsa; the Samfundet Folkha¨lsan; the Finska La¨karesa¨llskapet; and the Signe and Ane Gyllenberg Foundation. 4 Address reprint requests and correspondence to S Sandboge, Folkha¨lsan Research Centre, Samfundet Folkha¨lsan I Svenska Finland rf (Folkha¨lsan), Topeliusgatan 20, 00250 Helsingfors, Finland. E-mail: sandboge@gmail. com. 5 Abbreviations used: AST, aspartate aminotransferase; FFQ, food frequency questionnaire; FLI, Fatty Liver Index; HBCS, Helsinki Birth Cohort Study; ICD, International Classification of Diseases; MetS, metabolic syndrome; NAFLD, nonalcoholic fatty liver disease. Received February 28, 2014. Accepted for publication July 8, 2014. First published online August 6, 2014; doi: 10.3945/ajcn.114.086074.
Am J Clin Nutr 2014;100:1133–8. Printed in USA. Ó 2014 American Society for Nutrition
1133
Downloaded from ajcn.nutrition.org at UNIV OF CALIFORNIA SAN FRANCISCO KALMANOVITZ LIBRARY on December 23, 2014
Higher fructose intake is inversely associated with risk of nonalcoholic fatty liver disease in older Finnish adults1–4
1134
KANERVA ET AL
individuals who were obese as adults. The aim of the current study was to explore the associations between fructose intake and NAFLD. SUBJECTS AND METHODS
RESULTS
According to the FLI, 663 participants (44%) were classified as having NAFLD. Correspondingly, 511 participants (46%) were classified as having NAFLD when assessed by using the NAFLD liver fat score. The agreement between the 2 scores was 76.3% for a negative finding and 78.9% for a positive finding. Participants in the highest quartile of fructose intake were more likely women than men, had more years of education, and were less likely to be current smokers or physically inactive than were those in the lowest fructose intake quartiles (Table 1).
Downloaded from ajcn.nutrition.org at UNIV OF CALIFORNIA SAN FRANCISCO KALMANOVITZ LIBRARY on December 23, 2014
The HBCS included 8760 participants who were born at Helsinki University Central Hospital as singletons between 1934 and 1944 and were still alive and resident in Finland in 1971 when every citizen was given a unique personal identification number. In the year 2000, cohort members were sent questionnaires that resulted in replies from 4515 subjects. Of respondents, a sample of 2902 individuals was derived by using random-number tables and invited to a clinical examination conducted in the years 2001–2004. Eventually, 2003 individuals participated in the health examination. The study was conducted according to the guidelines of the Declaration of Helsinki, and the Ethics Committee of the Hospital District of Helsinki and Uusimaa approved all procedures involving human subjects. Written informed consent was obtained from all participants. At the study clinic, self-administered questionnaires were used to inquire about socioeconomic characteristics (eg, educational attainment), lifestyle factors (eg, smoking status), and current medication. The current use of antidiabetic medication and WHO criteria were used to define type 2 diabetes diagnosis (17). The MetS was defined according to harmonized criteria (18). The Fatty Liver Index (FLI) (19) was calculated for all participants and was used to define the presence of NAFLD. The FLI is an algorithm test that consists of the following 4 variables: BMI (in kg/m2), waist circumference, fasting triglycerides, and fasting g-glutamyl-transferase. In the original publication, an FLI .60 predicted NAFLD with a specificity of 86% and positive likelihood ratio of 4.3 against a diagnosis by using ultrasound. On the basis of this cutoff, a dichotomous variable was constructed with values .60 defined as a positive FLI. In the context of this publication, a positive FLI and the presence of NAFLD were used synonymously. We also calculated the NAFLD liver fat score (16) for all participants for reasons of comparison. This algorithm test is based on the following 5 variables: the presence of type 2 diabetes, presence of MetS, fasting insulin, serum aspartate aminotransferase (AST), and the AST:aminotransferase ratio. At the clinic, trained study nurses measured height to the nearest 0.1 cm and weight to the nearest 0.1 kg. BMI was calculated by dividing weight by the square of height. Fasting blood samples were drawn and stored at 2708C before analysis. Blood samples included a standard 75-g oral-glucose-tolerance test. A hexokinase method (20) was used to measure plasma glucose; a 2-site immunometric assay (21) was used to measure plasma insulin; and photometric International Federation of Clinical Chemistry methods were used to measure aminotransferase, AST, and g-glutamyl-transferase concentrations by using an Architect ci 82000 analysator (Abbott Laboratories). At the study site, participants completed a validated foodfrequency questionnaire (FFQ) (22, 23), which was reviewed by a trained study nurse. The FFQ measured the habitual diet over the previous 12 mo. The consumption of each 128 FFQ item was reported by using 9 frequency categories that ranged from never or seldom to $6 times/d. The predefined sex-specific portion sizes for food items appeared as household and natural units (eg,
glass and slice). A nutritionist entered data into the study database. The average daily intakes of foods, nutrients, and energy intakes were calculated by using the Finnish National Food Composition Database (Fineli) and in-house software (24). From 2003 participants, those who reported ethanol intake $20 g/d (n = 244) were excluded. Because of current or previous liver disease, 80 subjects were excluded. These individuals were identified via cross-referencing with the national hospital discharge register and the national mortality register. We performed a search for International Classification of Diseases (ICD) codes related to liver disease in both registers (ICD 8–9: 070, 155, 291, 303, 570, 571, 572, and 573; ICD 10: B15-19, C22, F10, G62, K70-77, and K86). Of 80 excluded subjects, 71 individuals had an alcohol-related diagnosis, but no cases of explicit nonalcoholic steatosis or steatohepatitis were shown in either register. Furthermore, 54 subjects were excluded because of the use of medication associated with secondary hepatic steatosis according to American Association for the Study of Liver Diseases guidelines on the diagnosis and management of NAFLD (25). In addition, participants in this study were not allowed to have any missing values in main exposure (ie, dietary data) or outcome (NAFLD data) variables. Participants with an implausible energy intake (n = 8) and incomplete measurements for the FLI calculation (n = 6) were excluded. Final data included 1611 participants for the analysis. We performed statistical analyses with the use of R statistical software version 2.15.1 (26). Because the test for the interaction between sex and outcome variables was P . 0.05, we conducted analyses together for men and women. All nutrient variables were adjusted for energy by using the residual method (27). We reported descriptive data by grouping individuals by intake of fructose as the mean (6SE) or percentage. A logistic regression analysis was used to calculate ORs and 2-sided 95% CIs for the FLI and NAFLD liver fat score (main outcomes, dichotomous) by energy-adjusted fructose intake quartiles (main explanatory variable). The selection of confounding variables was based on the current literature. Furthermore, the variable had to be associates with both the outcome and exposure in HBCS data to be included in the modeling (28). Model 1 was adjusted for age (continuous; y), sex (dichotomous; male or female), and intake of energy (continuous; kJ/d). Model 2 was further adjusted for smoking status (categorical; current smoker, former smoker, or never smoker) and leisure-time physical activity (categorical: no exercise, exercise 1–3 times/wk, or exercise $4 times/wk), and model 3 was further adjusted for dietary variables such as fiber (continuous; g/d), total fat intake (continuous; g/d), alcohol (continuous; g/d), and vitamin E (continuous; alphatocopherol equivalents in mg) intakes.
1135
FRUCTOSE INTAKE AND NAFLD TABLE 1 Participant characteristics according to intakes of fructose1 Fructose intake quartiles Characteristic
2
3
4
P
403 2.2–15.2 10.6 6 1.03 41 61.6 6 0.2 11.4 6 0.2 38 16 27.7 6 0.2 95.7 6 0.6 81 32 15 51 17 54 48 35 8970 6 160 22.8 6 0.4 82.2 6 0.5 13.9 6 0.6 12.2 6 0.2
403 15.3–21.8 18.6 6 1.0 55 61.7 6 0.2 12.0 6 0.2 17 9 27.8 6 0.2 95.4 6 0.6 82 32 13 51 14 54 44 32 9230 6 160 26.1 6 0.4 76.9 6 0.5 14.5 6 0.6 13.2 6 0.1
402 21.9–29.1 25.2 6 1.0 67 61.5 6 0.2 12.4 6 0.2 18 9 27.4 6 0.2 93.9 6 0.6 81 26 11 47 12 47 38 29 9420 6 160 28.1 6 0.4 73.6 6 0.5 13.2 6 0.6 13.4 6 0.1
403 29.2–88.0 38.1 6 1.0 76 61.6 6 0.2 12.6 6 0.2 13 10 27.5 6 0.2 94.2 6 0.6 81 26 10 46 13 49 34 28 9110 6 160 32.4 6 0.4 66.3 6 0.5 12.1 6 0.6 13.3 6 0.2
— — — ,0.001 0.91 ,0.001 ,0.001 ,0.01 0.35 ,0.05 0.58 0.06 0.08 0.23 0.36 0.08 ,0.001 0.06 0.07 ,0.001 ,0.001 0.22 ,0.001
1 Values were adjusted for age and sex. All nutrient intakes were also adjusted for intake of energy by using the residual method (27). P values for trends were obtained from linear regression for continuous variables and logistic regression for binary variables. P-values were adjusted for age and sex. For nutrient intakes, P values were additionally adjusted for intake of energy. FLI, fatty liver index; NAFLD, nonalcoholic fatty liver disease. 2 All values are ranges. 3 Mean 6 SEM (all such values). 4 Systolic blood pressure $130 mm Hg or diastolic blood pressure $85 mm Hg. 5 Triglyceride concentration $1.7 mmol/L. 6 HDL concentration ,1.0 mmol/L (men) or ,1.3 mmol/L (women). 7 Fasting glucose concentration $5.55 mmol/L.
Furthermore, waist circumference tended to decrease across the fructose intake quartiles. Participants in the highest quartile of fructose intake had also higher intakes of fiber and vitamin E but lower intake of fat. In the model adjusted for age, sex, and intake of energy (model 1), intake of fructose was negatively associated with NAFLD when assessed by using the FLI or NAFLD liver fat score (Table 2). Subjects in the highest quartile of fructose intake were 28–44% less likely to have NAFLD than were those in the lowest intake quartile. After additional adjustment for lifestyle and nutrient intake variables (model 3), the association attenuated to 32% for the FLI but remained significant (95% CI: 0.47, 0.97; P-trend , 0.05). After further adjustments, the association did not remain for the NAFLD liver fat score (model 3; Table 2).
DISCUSSION
To our knowledge, this study is among the first large-scale studies to report a cross-sectional association between fructose intake and NAFLD assessed by using the FLI and the NAFLD liver fat score. Participants with higher fructose intake were significantly less likely to have a positive FLI score independently of the participants’ age, sex, lifestyle, and various nutrients. When
NAFLD was assessed by using the NAFLD liver fat score, no association between intake of fructose and NAFLD emerged. Much of the evidence on the association of fructose intake and NAFLD has been based on animal studies that showed that high fructose intake increased lipogenesis and triglyceride concentration, lowered HDL cholesterol concentrations, and lead to NAFLD (29). However, the few human trials and small-scale (n , 500) observational studies have provided inconsistent results (12, 14). A meta-analysis of 16 trials did not find any effect on HDL cholesterol when other carbohydrates were exchanged for fructose (13), and higher intake of fructose did not change triglycerides concentrations in healthy individuals (13, 14). In contrast, higher fructose intake has been associated with a lower prevalence of steatosis and a higher fibrosis grade in older subjects (12). Our findings in a cohort of 1611 adult Finns supports the evidence that higher intake of fructose is not associated with a higher prevalence of NAFLD. The conflicting results between animal studies, experimental human trials, and our findings may have stemmed from the different levels of exposure. Animal and human feeding studies have been criticized for providing supraphysiologic doses of fructose that have varied typically between intake of 60–220 g/d. In our cohort, eg, the median fructose intake was 20 g/d, and only 45 participants had fructose intake $60 g/d. Because
Downloaded from ajcn.nutrition.org at UNIV OF CALIFORNIA SAN FRANCISCO KALMANOVITZ LIBRARY on December 23, 2014
n Fructose (g/d)2 Fructose (g/d) Women (%) Age (y) Educational attainment (y) Current smokers (%) Physically inactive participants (%) BMI (kg/m2) Waist circumference (cm) Hypertension (%)4 Hypertriglyceridemia (%)5 Reduced HDL-cholesterol (%)6 Elevated fasting glucose (%)7 Patients with diabetes (%) Metabolic syndrome (%) NAFLD (FLI) (%) NAFLD liver fat score (%) Energy intake (kJ/d) Fiber (g/d) Total fat (g/d) Alcohol (g/d) Vitamin E (a-tocopherol equivalents; mg/d)
1
1136
KANERVA ET AL
TABLE 2 Participant ORs (95% CIs) for NAFLD (FLI and NAFLD liver fat score) by intakes of fructose1 Energy-adjusted fructose intake quartiles Model
204/403 1 1 1 1 157/403 1 1 1 1
2 (15.3–21.8 g/d) 181/403 (0.60, 1.05) (0.65, 1.15) (0.71, 1.26) (0.72, 1.31) 137/403 0.81 (0.61, 1.08) 0.91 (0.68, 1.22) 0.95 (0.70, 1.28) 0.99 (0.72, 1.35)
0.80 0.87 0.94 0.97
3 (21.9–29.1 g/d) 149/402 (0.43, 0.76) (0.50, 0.90) (0.55, 1.00) (0.57, 1.07) 114/402 0.62 (0.46, 0.83) 0.77 (0.57, 1.04) 0.81 (0.59, 1.11) 0.87 (0.62, 1.22)
0.57 0.67 0.74 0.78
4 (29.2–88.0 g/d) 129/403 0.46 (0.34, 0.56 (0.42, 0.62 (0.46, 0.68 (0.47, 103/403 0.54 (0.40, 0.72 (0.53, 0.75 (0.54, 0.88 (0.60,
0.61) 0.75) 0.84) 0.97) 0.73) 0.99) 1.04) 1.29)
P-trend2
P-LRT
— ,0.001 ,0.001 ,0.001 ,0.05 — ,0.001 ,0.001 0.11 0.80
— — ,0.001 ,0.01 0.09 — — 0.15 0.27 0.81
1
Model 1 was adjusted for age, sex, and intake of energy. Model 2 was adjusted as for model 1 and for leisure-time physical activity and smoking. Model 3 was adjusted as for model 2 and for nutrients associated with both the main outcomes (FLI and the NAFLD liver fat score) and the main exposure (intake of fructose): intakes of alcohol, fiber, vitamin E, and total fat. FLI, fatty liver index; LRT, likelihood ratio test; NAFLD, nonalcoholic fatty liver disease. 2 Obtained from logistic regression by using energy-adjusted fructose intake as a continuous variable in the model. 3 Reference. 4 Unadjusted model.
fructose is approximately one-half of the intake of sucrose, these numbers are in line with the sucrose intake in the general Finnish population, which is 54 g/d for men and 43 g/d for women (30). When sugars were compared with other macronutrients, an isocaloric high-fat diet increased liver fat and insulin concentrations more than did an isocaloric high-carbohydrate diet (31). Furthermore, in healthy individuals, overfeeding with fructose increased liver fat less than did overfeeding with fat (32). Overall, it seems that adverse changes related to fructose intake have been more due to excess caloric intake rather than sugar composition (33). This finding has been consistent in clinical trials conducted thus far (34). Our results, which adjusted fructose intake for the intake of energy, also suggest that, in an energy-controlled situation, fructose does not associate with an increased prevalence of NAFLD. Main sources of fructose in the Finnish population are added table sugar, soft drinks, and fruit. Previous studies have established a strong association between soft-drink consumption and an increased prevalence of NAFLD (35, 36). Recently, Petta et al (37) examined 147 biopsy-proven NAFLD patients and showed that intake of industrial but not fruit-originated fructose was independently associated with adverse metabolic features as well as severity of liver fibrosis. These results imply that the source of fructose plays a significant role in inducing NAFLD. In our study, fructose that came from fruit or soft drinks was not separated, which may have confounded the results. In the study population, the proportion of men and women who consumed fruit (excluding fruit juices) was 54–80%, whereas the proportion who consumed soft drinks (excluding artificially sweetened soft drinks) was between 0 and 8%. In subjects who consumed soft drinks, median intake was 42 g/d (nonadjusted), whereas for fruit, the median intake was 230 g/d (nonadjusted). On the basis of this information, it could be assumed that the participants were more likely to receive more fructose from fruit than from soft drinks. Thus, in our analysis, the possible protective effect of fruit on NAFLD may have overcome the possibly harmful effect of soft drinks and NAFLD. Because of the complex nature of the diet,
the effect of fructose source on the development of NAFLD requires additional investigation. Strengths of our study included a relatively large sample size, validated and standardized measurements as well as laboratory analyses, and the careful control of confounding variables. The FFQ has been culturally adapted to the Finnish population (22), and the version of the FFQ used in this study has been validated against 3-d dietary records in the general Finnish population (22, 23). However, there were also some limitations. Our sample was not population-based, which limited the generalizability of our results. Because the study design was cross-sectional, no conclusions could be drawn about the causal relation between fructose intake and prevalence of NAFLD. Thus, we could not rule out the possibility that the prevalence of NAFLD leads to a lower consumption of fructose, eg, due to dietary counseling that NAFLD patients may have received. Furthermore, the FFQ’s validity to measure dietary intake in NAFLD patients has not been separately examined. An overunderestimation or underestimation of fructose-containing foods (eg, fruit and soft drinks) may have affected our results to some extent. Fructose can also be seen as a surrogate for a healthy or unhealthy lifestyle, which is probably associated with NAFLD. We carefully controlled the analysis for lifestyle factors such as physical activity and smoking, but some residual confounding could still have remained because measurements of these factors were based solely on questionnaires. Moreover, we may have failed to take some unknown factor into account (eg, a genetic factor) associated with fructose intake, which also could have confounded the results. The NAFLD diagnosis was based on the FLI algorithm test rather than the gold standard liver biopsy, which limited conclusions of our study. A liver biopsy would not have been a feasible option in our study setting because it is time consuming, costly, and associated with a small but notable procedure-related morbidity and even cases of mortality (38). Alternative methods of diagnosing NAFLD include the use of ultrasound, computer tomography, and magnetic resonance
Downloaded from ajcn.nutrition.org at UNIV OF CALIFORNIA SAN FRANCISCO KALMANOVITZ LIBRARY on December 23, 2014
FLI (cases/total) (n) Crude4 Model 1 Model 2 Model 3 NAFLD liver fat score (cases/total) (n) Crude4 Model 1 Model 2 Model 3
1 (2.2–15.2 g/d)3
FRUCTOSE INTAKE AND NAFLD
The authors’ responsibilities were as follows—JGE: designed the HBCS and was responsible for the original study idea; SM and NEK: were responsible for dietary data collected in the HBCS and critically revised the manuscript; SS and NEK: participated in data handling; NK: performed statistical analyses; NK and SS: wrote the manuscript and had primary responsibility for the final content of the manuscript; and all authors: approved the final manuscript. None of the authors had a conflict of interest.
REFERENCES 1. Vanni E, Bugianesi E, Kotronen A, De Minicis S, Yki-Ja¨rvinen H, Svegliati-Baroni G. From the metabolic syndrome to NAFLD or vice versa? Dig Liver Dis 2010;42:320–30. 2. Marchesini G, Brizi M, Bianchi G, Tomassetti S, Bugianesi E, Lenzi M, McCullough AJ, Natale S, Forlani G, Melchionda N. Nonalcoholic fatty liver disease a feature of the metabolic syndrome. Diabetes 2001; 50:1844–50. 3. Bellentani S, Scaglioni F, Marino M, Bedogni G. Epidemiology of nonalcoholic fatty liver disease. Dig Dis 2010;28:155–61. 4. Loomba R, Sanyal AJ. The global NAFLD epidemic. Nat Rev Gastroenterol Hepatol 2013;10:686–90. 5. Puri P, Sanyal AJ. Nonalcoholic fatty liver disease: definitions, risk factors, and workup. Clin Liver Dis 2012;1:98–102. 6. Musso G, Gambino R, Cassader M, Pagano G. Meta-analysis: natural history of non-alcoholic fatty liver disease (NAFLD) and diagnostic accuracy of non-invasive tests for liver disease severity. Ann Med 2011;43:617–49. 7. Nakao H, Yoneda M. The intertwisted correlations among nonalcoholic fatty liver disease, atherosclerosis, and metabolic syndrome. J Gastroenterol 2009;44:1162–4.
8. Lu H, Liu H, Hu F, Zou L, Luo S, Sun L. Independent association between nonalcoholic fatty liver disease and cardiovascular disease: a systematic review and meta-analysis. Int J Endocrinol 2013;2013: 124958. 9. Lewis JR, Mohanty SR. Nonalcoholic fatty liver disease: a review and update. Dig Dis Sci 2010;55:560–78. 10. Vos MB, Lavine JE. Dietary fructose in nonalcoholic fatty liver disease. Hepatology 2013;57:2525–31. 11. Neuschwander-Tetri BA. Carbohydrate intake and nonalcoholic fatty liver disease. Curr Opin Clin Nutr Metab Care 2013;16:446–52. 12. Abdelmalek MF, Suzuki A, Guy C, Unalp-Arida A, Colvin R, Johnson RJ, Diehl AM; Nonalcoholic Steatohepatitis Clinical Research Network. Increased fructose consumption is associated with fibrosis severity in patients with nonalcoholic fatty liver disease. Hepatology 2010;51:1961–71. 13. Sievenpiper JL, Carleton AJ, Chatha S, Jiang HY, de Souza RJ, Beyene J, Kendall CW, Jenkins DJ. Heterogeneous effects of fructose on blood lipids in individuals with type 2 diabetes systematic review and metaanalysis of experimental trials in humans. Diabetes Care 2009;32: 1930–7. 14. Johnston RD, Stephenson MC, Crossland H, Cordon SM, Palcidi E, Cox EF, Taylor MA, Aithal GP, Macdonald IA. No difference between high-fructose and high-glucose diets on liver triacylglycerol or biochemistry in healthy overweight men. Gastroenterology 2013;145: 1016–25. 15. Sandboge S, Pera¨la¨ M-M, Salonen MK, Blomstedt PA, Osmond C, Kajantie E, Barker DJ, Eriksson JG. Early growth and non-alcoholic fatty liver disease in adulthood-the NAFLD liver fat score and equation applied on the Helsinki Birth Cohort Study. Ann Med 2013;45:430–7. 16. Kotronen A, Peltonen M, Hakkarainen A, Sevastianova K, Bergholm R, Johansson LM, Lundbom M, Rissanen A, Ridderstra˚le M, Groop L, et al. Prediction of non-alcoholic fatty liver disease and liver fat using metabolic and genetic factors. Gastroenterology 2009;137:865–72. 17. World Health Organization. Definition, diagnosis and classification of diabetes mellitus and its complications. Geneva, Switzerland: World Health Organization, 1999:31–3. 18. Alberti KG, Eckel RH, Grundy SM, Zimmet PZ, Cleeman JI, Donato KA, Fruchart JC, James WP, Loria CM, Smith SC Jr, et al. Harmonizing the metabolic syndrome: a joint interim statement of the International Diabetes Federation Task Force on Epidemiology and Prevention; National Heart, Lung, and Blood Institute; American Heart Association; World Heart Federation; International Atherosclerosis Society; and International Association for the Study of Obesity. Circulation 2009;120:1640–5. 19. Bedogni G, Bellentani S, Miglioli L, Masutti F, Passalacqua M, Castiglione A, Tiribelli C. The Fatty Liver Index: a simple and accurate predictor of hepatic steatosis in the general population. BMC Gastroenterol 2006;6:33. 20. Kunst A, Draeger B, Ziegenhorn J. UV-methods with hexokinase and glucose-6-phosphate dehydrogenase. Chapter 6. In: Bergmeyer HU, ed. Methods of enzymatic analysis. 3rd ed. Weinheim, Germany: Verlag Chemie, 1984;163–72. 21. Sobey WJ, Beer S, Carrington CA, Clark P, Frank B, Gray I, Luzio SD, Owens DR, Schneider AE, Siddle K, et al. Sensitive and specific twosite immunoradiometric assays for human insulin, proinsulin, 65-66 split and 32-33 split proinsulins. Biochem J 1989;260:535–41. 22. Ma¨nnisto¨ S, Virtanen M, Mikkonen T, Pietinen P. Reproducibility and validity of a food frequency questionnaire in a case-control study on breast cancer. J Clin Epidemiol 1996;49:401–9. 23. Paalanen L, Ma¨nnisto¨ S, Virtanen MJ, Knekt P, Ra¨sa¨nen L, Montonen J, Pietinen P. Validity of a food frequency questionnaire varied by age and body mass index. J Clin Epidemiol 2006;59:994–1001. 24. Reinivuo H, Hirvonen T, Ovaskainen M-L, Korhonen T, Valsta LM. Dietary survey methodology of FINDIET 2007 with a risk assessment perspective. Public Health Nutr 2010;13:915–9. 25. Chalasani N, Younossi Z, Lavine JE, Diehl AM, Brunt EM, Cusi K, Charlton M, Sanyal AJ. The diagnosis and management of nonalcoholic fatty liver disease: practice guideline by the American Association for the Study of Liver Diseases, American College of Gastroenterology, and the American Gastroenterological Association. Hepatology 2012;55:2005–23. 26. R Development Core Team. R: a language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing, 2013.
Downloaded from ajcn.nutrition.org at UNIV OF CALIFORNIA SAN FRANCISCO KALMANOVITZ LIBRARY on December 23, 2014
imaging. However, these methods were not included in the HBCS protocol. In a previous publication, in which associations between early growth and NAFLD were studied (15), we used the NAFLD liver fat score and equation (16) to define NAFLD. For purposes of the current study, we also used the FLI because it is more widely used. The FLI was validated against an ultrasound in the original publication and has been used in several publications as a surrogate for NAFLD (39–41). An important limitation in the use of an algorithm test, rather than a moreobjective method, was risk of misclassification. We are currently conducting a validation study of both algorithm tests against a diagnosis by using magnetic resonance spectroscopy on a small subset of the study population (n = 41). In the comparison of the 2 scores, the NAFLD liver fat score has had a better balance in sensitivity and specificity (NAFLD liver fat score: sensitivity of 85% and specificity of 71%; FLI: sensitivity of 61% and specificity of 86%). Because of a higher probability of detecting false positives, the FLI score may have incorrectly classified individuals with a healthy liver and their dietary habits to the NAFLD cohort, thereby skewing the true trend between fructose intake and NAFLD cases and leading to inconsistent results between the 2 scores. However, the level of agreement between the 2 scores was considered acceptable (76.3% for a negative finding and a 78.9% for positive finding). In conclusion, high intake of fructose was not associated with higher prevalence of NAFLD assessed by suing the FLI and NAFLD liver fat score. Because calculations of daily nutrient intakes did not separate whether the fructose came from natural or artificial sources, a major question about the possible different effects of these 2 sources on the development of NAFDL remains. Furthermore, because our study was cross-sectional and concerned older adult Finns born in Helsinki, population-based longitudinal studies are needed to confirm the causality of these associations.
1137
1138
KANERVA ET AL 35. Assy N, Nasser G, Kamayse I, Nseir W, Beniashvili Z, Djibre A, Grosovski M. Soft drink consumption linked with fatty liver in the absence of traditional risk factors. Can J Gastroenterol 2008;22:811–6. 36. Abid A, Taha O, Nseir W, Farah R, Grosovski M, Assy N. Soft drink consumption is associated with fatty liver disease independent of metabolic syndrome. J Hepatol 2009;51:918–24. 37. Petta S, Marchesini G, Caracausi L, Macaluso FS, Camma` C, Ciminnisi S, Cabibi D, Porcasi R, Craxı` A, Di Marco V. Industrial, not fruit fructose intake is associated with the severity of liver fibrosis in genotype 1 chronic hepatitis C patients. J Hepatol 2013;59:1169–76. 38. Weigand K, Weigand K. Percutaneous liver biopsy: retrospective study over 15 years comparing 287 inpatients with 428 outpatients. J Gastroenterol Hepatol 2009;24:792–9. 39. Calori G, Lattuada G, Ragogna F, Garancini MP, Crosignani P, Villa M, Bosi E, Ruotolo G, Piemonti L, Perseghin G. Fatty Liver Index and mortality: the Cremona study in the 15th year of follow-up. Hepatology 2011;54:145–52. 40. Gastaldelli A, Kozakova M, Højlund K, Flyvbjerg A, Favuzzi A, Mitrakou A, Balkau B. RISC Investigators. Fatty liver is associated with insulin resistance, risk of coronary heart disease, and early atherosclerosis in a large European population. Hepatology 2009;49: 1537–44. 41. Kozakova M, Palombo C, Paterni Eng M, Dekker J, Flyvbjerg A, Mitrakou A, Gastaldelli A, Ferrannini E; RISC Investigators. Fatty liver index, gamma-glutamyltransferase, and early carotid plaques. Hepatology 2012;55:1406–15.
Downloaded from ajcn.nutrition.org at UNIV OF CALIFORNIA SAN FRANCISCO KALMANOVITZ LIBRARY on December 23, 2014
27. Willett W, Stampfer MJ. Total energy intake: implications for epidemiologic analyses. Am J Epidemiol 1986;124:17–27. 28. Rothman KJ, ed. Modern Epidemiology. 1st ed. Boston, MA: Little, Brown and Company, 1986. 29. de Castro UG, dos Santos RA, Silva ME, de Lima WG, CampagnoleSantos MJ, Alzamora AC. Age-dependent effect of high-fructose and high-fat diets on lipid metabolism and lipid accumulation in liver and kidney of rats. Lipids Health Dis 2013;12:136. 30. Pietinen P, Paturi M, Reinivuo H, Tapanainen H, Valsta LM. FINDIET 2007 Survey: energy and nutrient intakes. Public Health Nutr 2010;13: 920–4. 31. Westerbacka J, Lammi K, Ha¨kkinen A-M, Rissanen A, Salminen I, Aro A, Yki-Ja¨rvinen H. Dietary fat content modifies liver fat in overweight nondiabetic subjects. J Clin Endocrinol Metab 2005;90: 2804–9. 32. Sobrecases H, Leˆ K-A, Bortolotti M, Schneiter P, Ith M, Kreis R, Boesch C, Tappy L. Effects of short-term overfeeding with fructose, fat and fructose plus fat on plasma and hepatic lipids in healthy men. Diabetes Metab 2010;36:244–6. 33. Ngo Sock ET, Leˆ K-A, Ith M, Kreis R, Boesch C, Tappy L. Effects of a short-term overfeeding with fructose or glucose in healthy young males. Br J Nutr 2010;103:939–43. 34. Chiu S, Sievenpiper JL, de Souza RJ, Cozma AI, Mirrahimi A, Carleton AJ, Ha V, Di Buono M, Jenkins AL, Leiter LA, et al. Effect of fructose on markers of non-alcoholic fatty liver disease (NAFLD): a systematic review and meta-analysis of controlled feeding trials. Eur J Clin Nutr 2014;68:416–23.
INTRODUCTION
Nonalcoholic fatty liver disease (NAFLD)5, which is the hepatic manifestation of the metabolic syndrome (MetS) (1, 2), is the most common cause of chronic liver disease in the developed world with a prevalence ranging from 20% to 30% (3, 4). NAFLD is present when the hepatic fat content exceeds 5– 10% of liver mass in the absence of other causes of steatosis such as other liver disorders, excessive alcohol intake, and use of hepatotoxic medication (5). The condition is closely associated with type 2 diabetes and obesity (6, 7) and, in addition, is an independent risk factor for cardiovascular disease (8). A continuum of conditions is included in the NAFLD diagnosis
from simple steatosis via nonalcoholic steatohepatitis to fibrosis and cirrhosis (9). The pathogenesis of NAFLD has been described by a “multihit” hypothesis, the key pathogenic mechanism of which is insulin resistance (1, 9). Hepatic lipid accumulation (ie, steatosis) represents the first hit, whereas several second hits (eg, inflammatory cytokines) are involved in the progression of the disease toward nonalcoholic steatohepatitis and beyond. Several factors have been implicated in the development of NAFLD and its progression, including carbohydrate intake. Fructose intake has received special attention, and several previous studies have linked high fructose intake with NAFLD (10). Fructose stimulates de novo lipogenesis in the liver that leads to the production of triglycerides, which, in turn, can be secreted as VLDLs. If this process is impaired, triglycerides can accumulate intracellularly and led to steatosis (11). There have been some studies that indicated that fructose intake is increased in individuals with NAFLD, but the evidence has been inconsistent (12–14). We have previously studied the prevalence of NAFLD in the Helsinki Birth Cohort Study (HBCS) from a developmental origins of health and disease perspective (15). In that study, we used an algorithm test, the NAFLD liver fat score and equation (16), to define NAFLD. We showed that a small body size during childhood was associated with adult NAFLD, especially in 1
From the Department of Chronic Disease Prevention, National Institute for Health and Welfare, Helsinki, Finland (NK, SS, NEK, SM, and JGE); the Departments of Public Health (NK) and General Practice and Primary Health Care (JGE), University of Helsinki, Helsinki, Finland; the Folkha¨lsan Research Center, Helsinki, Finland (SS and JGE); the Helsinki University Central Hospital, Unit of General Practice, Helsinki, Finland (JGE); and the Vasa Central Hospital, Vasa, Finland (JGE). 2 Funding sources had no role in the design, implementation, analysis, or interpretation of the study 3 Supported by the Academy of Finland [grants 120386 and 125876 (to JGE) and 136895 and 263836 (to SM)]; the Finnish Diabetes Research Society; Liv and Ha¨lsa; the Samfundet Folkha¨lsan; the Finska La¨karesa¨llskapet; and the Signe and Ane Gyllenberg Foundation. 4 Address reprint requests and correspondence to S Sandboge, Folkha¨lsan Research Centre, Samfundet Folkha¨lsan I Svenska Finland rf (Folkha¨lsan), Topeliusgatan 20, 00250 Helsingfors, Finland. E-mail: sandboge@gmail. com. 5 Abbreviations used: AST, aspartate aminotransferase; FFQ, food frequency questionnaire; FLI, Fatty Liver Index; HBCS, Helsinki Birth Cohort Study; ICD, International Classification of Diseases; MetS, metabolic syndrome; NAFLD, nonalcoholic fatty liver disease. Received February 28, 2014. Accepted for publication July 8, 2014. First published online August 6, 2014; doi: 10.3945/ajcn.114.086074.
Am J Clin Nutr 2014;100:1133–8. Printed in USA. Ó 2014 American Society for Nutrition
1133
Downloaded from ajcn.nutrition.org at UNIV OF CALIFORNIA SAN FRANCISCO KALMANOVITZ LIBRARY on December 23, 2014
Higher fructose intake is inversely associated with risk of nonalcoholic fatty liver disease in older Finnish adults1–4
1134
KANERVA ET AL
individuals who were obese as adults. The aim of the current study was to explore the associations between fructose intake and NAFLD. SUBJECTS AND METHODS
RESULTS
According to the FLI, 663 participants (44%) were classified as having NAFLD. Correspondingly, 511 participants (46%) were classified as having NAFLD when assessed by using the NAFLD liver fat score. The agreement between the 2 scores was 76.3% for a negative finding and 78.9% for a positive finding. Participants in the highest quartile of fructose intake were more likely women than men, had more years of education, and were less likely to be current smokers or physically inactive than were those in the lowest fructose intake quartiles (Table 1).
Downloaded from ajcn.nutrition.org at UNIV OF CALIFORNIA SAN FRANCISCO KALMANOVITZ LIBRARY on December 23, 2014
The HBCS included 8760 participants who were born at Helsinki University Central Hospital as singletons between 1934 and 1944 and were still alive and resident in Finland in 1971 when every citizen was given a unique personal identification number. In the year 2000, cohort members were sent questionnaires that resulted in replies from 4515 subjects. Of respondents, a sample of 2902 individuals was derived by using random-number tables and invited to a clinical examination conducted in the years 2001–2004. Eventually, 2003 individuals participated in the health examination. The study was conducted according to the guidelines of the Declaration of Helsinki, and the Ethics Committee of the Hospital District of Helsinki and Uusimaa approved all procedures involving human subjects. Written informed consent was obtained from all participants. At the study clinic, self-administered questionnaires were used to inquire about socioeconomic characteristics (eg, educational attainment), lifestyle factors (eg, smoking status), and current medication. The current use of antidiabetic medication and WHO criteria were used to define type 2 diabetes diagnosis (17). The MetS was defined according to harmonized criteria (18). The Fatty Liver Index (FLI) (19) was calculated for all participants and was used to define the presence of NAFLD. The FLI is an algorithm test that consists of the following 4 variables: BMI (in kg/m2), waist circumference, fasting triglycerides, and fasting g-glutamyl-transferase. In the original publication, an FLI .60 predicted NAFLD with a specificity of 86% and positive likelihood ratio of 4.3 against a diagnosis by using ultrasound. On the basis of this cutoff, a dichotomous variable was constructed with values .60 defined as a positive FLI. In the context of this publication, a positive FLI and the presence of NAFLD were used synonymously. We also calculated the NAFLD liver fat score (16) for all participants for reasons of comparison. This algorithm test is based on the following 5 variables: the presence of type 2 diabetes, presence of MetS, fasting insulin, serum aspartate aminotransferase (AST), and the AST:aminotransferase ratio. At the clinic, trained study nurses measured height to the nearest 0.1 cm and weight to the nearest 0.1 kg. BMI was calculated by dividing weight by the square of height. Fasting blood samples were drawn and stored at 2708C before analysis. Blood samples included a standard 75-g oral-glucose-tolerance test. A hexokinase method (20) was used to measure plasma glucose; a 2-site immunometric assay (21) was used to measure plasma insulin; and photometric International Federation of Clinical Chemistry methods were used to measure aminotransferase, AST, and g-glutamyl-transferase concentrations by using an Architect ci 82000 analysator (Abbott Laboratories). At the study site, participants completed a validated foodfrequency questionnaire (FFQ) (22, 23), which was reviewed by a trained study nurse. The FFQ measured the habitual diet over the previous 12 mo. The consumption of each 128 FFQ item was reported by using 9 frequency categories that ranged from never or seldom to $6 times/d. The predefined sex-specific portion sizes for food items appeared as household and natural units (eg,
glass and slice). A nutritionist entered data into the study database. The average daily intakes of foods, nutrients, and energy intakes were calculated by using the Finnish National Food Composition Database (Fineli) and in-house software (24). From 2003 participants, those who reported ethanol intake $20 g/d (n = 244) were excluded. Because of current or previous liver disease, 80 subjects were excluded. These individuals were identified via cross-referencing with the national hospital discharge register and the national mortality register. We performed a search for International Classification of Diseases (ICD) codes related to liver disease in both registers (ICD 8–9: 070, 155, 291, 303, 570, 571, 572, and 573; ICD 10: B15-19, C22, F10, G62, K70-77, and K86). Of 80 excluded subjects, 71 individuals had an alcohol-related diagnosis, but no cases of explicit nonalcoholic steatosis or steatohepatitis were shown in either register. Furthermore, 54 subjects were excluded because of the use of medication associated with secondary hepatic steatosis according to American Association for the Study of Liver Diseases guidelines on the diagnosis and management of NAFLD (25). In addition, participants in this study were not allowed to have any missing values in main exposure (ie, dietary data) or outcome (NAFLD data) variables. Participants with an implausible energy intake (n = 8) and incomplete measurements for the FLI calculation (n = 6) were excluded. Final data included 1611 participants for the analysis. We performed statistical analyses with the use of R statistical software version 2.15.1 (26). Because the test for the interaction between sex and outcome variables was P . 0.05, we conducted analyses together for men and women. All nutrient variables were adjusted for energy by using the residual method (27). We reported descriptive data by grouping individuals by intake of fructose as the mean (6SE) or percentage. A logistic regression analysis was used to calculate ORs and 2-sided 95% CIs for the FLI and NAFLD liver fat score (main outcomes, dichotomous) by energy-adjusted fructose intake quartiles (main explanatory variable). The selection of confounding variables was based on the current literature. Furthermore, the variable had to be associates with both the outcome and exposure in HBCS data to be included in the modeling (28). Model 1 was adjusted for age (continuous; y), sex (dichotomous; male or female), and intake of energy (continuous; kJ/d). Model 2 was further adjusted for smoking status (categorical; current smoker, former smoker, or never smoker) and leisure-time physical activity (categorical: no exercise, exercise 1–3 times/wk, or exercise $4 times/wk), and model 3 was further adjusted for dietary variables such as fiber (continuous; g/d), total fat intake (continuous; g/d), alcohol (continuous; g/d), and vitamin E (continuous; alphatocopherol equivalents in mg) intakes.
1135
FRUCTOSE INTAKE AND NAFLD TABLE 1 Participant characteristics according to intakes of fructose1 Fructose intake quartiles Characteristic
2
3
4
P
403 2.2–15.2 10.6 6 1.03 41 61.6 6 0.2 11.4 6 0.2 38 16 27.7 6 0.2 95.7 6 0.6 81 32 15 51 17 54 48 35 8970 6 160 22.8 6 0.4 82.2 6 0.5 13.9 6 0.6 12.2 6 0.2
403 15.3–21.8 18.6 6 1.0 55 61.7 6 0.2 12.0 6 0.2 17 9 27.8 6 0.2 95.4 6 0.6 82 32 13 51 14 54 44 32 9230 6 160 26.1 6 0.4 76.9 6 0.5 14.5 6 0.6 13.2 6 0.1
402 21.9–29.1 25.2 6 1.0 67 61.5 6 0.2 12.4 6 0.2 18 9 27.4 6 0.2 93.9 6 0.6 81 26 11 47 12 47 38 29 9420 6 160 28.1 6 0.4 73.6 6 0.5 13.2 6 0.6 13.4 6 0.1
403 29.2–88.0 38.1 6 1.0 76 61.6 6 0.2 12.6 6 0.2 13 10 27.5 6 0.2 94.2 6 0.6 81 26 10 46 13 49 34 28 9110 6 160 32.4 6 0.4 66.3 6 0.5 12.1 6 0.6 13.3 6 0.2
— — — ,0.001 0.91 ,0.001 ,0.001 ,0.01 0.35 ,0.05 0.58 0.06 0.08 0.23 0.36 0.08 ,0.001 0.06 0.07 ,0.001 ,0.001 0.22 ,0.001
1 Values were adjusted for age and sex. All nutrient intakes were also adjusted for intake of energy by using the residual method (27). P values for trends were obtained from linear regression for continuous variables and logistic regression for binary variables. P-values were adjusted for age and sex. For nutrient intakes, P values were additionally adjusted for intake of energy. FLI, fatty liver index; NAFLD, nonalcoholic fatty liver disease. 2 All values are ranges. 3 Mean 6 SEM (all such values). 4 Systolic blood pressure $130 mm Hg or diastolic blood pressure $85 mm Hg. 5 Triglyceride concentration $1.7 mmol/L. 6 HDL concentration ,1.0 mmol/L (men) or ,1.3 mmol/L (women). 7 Fasting glucose concentration $5.55 mmol/L.
Furthermore, waist circumference tended to decrease across the fructose intake quartiles. Participants in the highest quartile of fructose intake had also higher intakes of fiber and vitamin E but lower intake of fat. In the model adjusted for age, sex, and intake of energy (model 1), intake of fructose was negatively associated with NAFLD when assessed by using the FLI or NAFLD liver fat score (Table 2). Subjects in the highest quartile of fructose intake were 28–44% less likely to have NAFLD than were those in the lowest intake quartile. After additional adjustment for lifestyle and nutrient intake variables (model 3), the association attenuated to 32% for the FLI but remained significant (95% CI: 0.47, 0.97; P-trend , 0.05). After further adjustments, the association did not remain for the NAFLD liver fat score (model 3; Table 2).
DISCUSSION
To our knowledge, this study is among the first large-scale studies to report a cross-sectional association between fructose intake and NAFLD assessed by using the FLI and the NAFLD liver fat score. Participants with higher fructose intake were significantly less likely to have a positive FLI score independently of the participants’ age, sex, lifestyle, and various nutrients. When
NAFLD was assessed by using the NAFLD liver fat score, no association between intake of fructose and NAFLD emerged. Much of the evidence on the association of fructose intake and NAFLD has been based on animal studies that showed that high fructose intake increased lipogenesis and triglyceride concentration, lowered HDL cholesterol concentrations, and lead to NAFLD (29). However, the few human trials and small-scale (n , 500) observational studies have provided inconsistent results (12, 14). A meta-analysis of 16 trials did not find any effect on HDL cholesterol when other carbohydrates were exchanged for fructose (13), and higher intake of fructose did not change triglycerides concentrations in healthy individuals (13, 14). In contrast, higher fructose intake has been associated with a lower prevalence of steatosis and a higher fibrosis grade in older subjects (12). Our findings in a cohort of 1611 adult Finns supports the evidence that higher intake of fructose is not associated with a higher prevalence of NAFLD. The conflicting results between animal studies, experimental human trials, and our findings may have stemmed from the different levels of exposure. Animal and human feeding studies have been criticized for providing supraphysiologic doses of fructose that have varied typically between intake of 60–220 g/d. In our cohort, eg, the median fructose intake was 20 g/d, and only 45 participants had fructose intake $60 g/d. Because
Downloaded from ajcn.nutrition.org at UNIV OF CALIFORNIA SAN FRANCISCO KALMANOVITZ LIBRARY on December 23, 2014
n Fructose (g/d)2 Fructose (g/d) Women (%) Age (y) Educational attainment (y) Current smokers (%) Physically inactive participants (%) BMI (kg/m2) Waist circumference (cm) Hypertension (%)4 Hypertriglyceridemia (%)5 Reduced HDL-cholesterol (%)6 Elevated fasting glucose (%)7 Patients with diabetes (%) Metabolic syndrome (%) NAFLD (FLI) (%) NAFLD liver fat score (%) Energy intake (kJ/d) Fiber (g/d) Total fat (g/d) Alcohol (g/d) Vitamin E (a-tocopherol equivalents; mg/d)
1
1136
KANERVA ET AL
TABLE 2 Participant ORs (95% CIs) for NAFLD (FLI and NAFLD liver fat score) by intakes of fructose1 Energy-adjusted fructose intake quartiles Model
204/403 1 1 1 1 157/403 1 1 1 1
2 (15.3–21.8 g/d) 181/403 (0.60, 1.05) (0.65, 1.15) (0.71, 1.26) (0.72, 1.31) 137/403 0.81 (0.61, 1.08) 0.91 (0.68, 1.22) 0.95 (0.70, 1.28) 0.99 (0.72, 1.35)
0.80 0.87 0.94 0.97
3 (21.9–29.1 g/d) 149/402 (0.43, 0.76) (0.50, 0.90) (0.55, 1.00) (0.57, 1.07) 114/402 0.62 (0.46, 0.83) 0.77 (0.57, 1.04) 0.81 (0.59, 1.11) 0.87 (0.62, 1.22)
0.57 0.67 0.74 0.78
4 (29.2–88.0 g/d) 129/403 0.46 (0.34, 0.56 (0.42, 0.62 (0.46, 0.68 (0.47, 103/403 0.54 (0.40, 0.72 (0.53, 0.75 (0.54, 0.88 (0.60,
0.61) 0.75) 0.84) 0.97) 0.73) 0.99) 1.04) 1.29)
P-trend2
P-LRT
— ,0.001 ,0.001 ,0.001 ,0.05 — ,0.001 ,0.001 0.11 0.80
— — ,0.001 ,0.01 0.09 — — 0.15 0.27 0.81
1
Model 1 was adjusted for age, sex, and intake of energy. Model 2 was adjusted as for model 1 and for leisure-time physical activity and smoking. Model 3 was adjusted as for model 2 and for nutrients associated with both the main outcomes (FLI and the NAFLD liver fat score) and the main exposure (intake of fructose): intakes of alcohol, fiber, vitamin E, and total fat. FLI, fatty liver index; LRT, likelihood ratio test; NAFLD, nonalcoholic fatty liver disease. 2 Obtained from logistic regression by using energy-adjusted fructose intake as a continuous variable in the model. 3 Reference. 4 Unadjusted model.
fructose is approximately one-half of the intake of sucrose, these numbers are in line with the sucrose intake in the general Finnish population, which is 54 g/d for men and 43 g/d for women (30). When sugars were compared with other macronutrients, an isocaloric high-fat diet increased liver fat and insulin concentrations more than did an isocaloric high-carbohydrate diet (31). Furthermore, in healthy individuals, overfeeding with fructose increased liver fat less than did overfeeding with fat (32). Overall, it seems that adverse changes related to fructose intake have been more due to excess caloric intake rather than sugar composition (33). This finding has been consistent in clinical trials conducted thus far (34). Our results, which adjusted fructose intake for the intake of energy, also suggest that, in an energy-controlled situation, fructose does not associate with an increased prevalence of NAFLD. Main sources of fructose in the Finnish population are added table sugar, soft drinks, and fruit. Previous studies have established a strong association between soft-drink consumption and an increased prevalence of NAFLD (35, 36). Recently, Petta et al (37) examined 147 biopsy-proven NAFLD patients and showed that intake of industrial but not fruit-originated fructose was independently associated with adverse metabolic features as well as severity of liver fibrosis. These results imply that the source of fructose plays a significant role in inducing NAFLD. In our study, fructose that came from fruit or soft drinks was not separated, which may have confounded the results. In the study population, the proportion of men and women who consumed fruit (excluding fruit juices) was 54–80%, whereas the proportion who consumed soft drinks (excluding artificially sweetened soft drinks) was between 0 and 8%. In subjects who consumed soft drinks, median intake was 42 g/d (nonadjusted), whereas for fruit, the median intake was 230 g/d (nonadjusted). On the basis of this information, it could be assumed that the participants were more likely to receive more fructose from fruit than from soft drinks. Thus, in our analysis, the possible protective effect of fruit on NAFLD may have overcome the possibly harmful effect of soft drinks and NAFLD. Because of the complex nature of the diet,
the effect of fructose source on the development of NAFLD requires additional investigation. Strengths of our study included a relatively large sample size, validated and standardized measurements as well as laboratory analyses, and the careful control of confounding variables. The FFQ has been culturally adapted to the Finnish population (22), and the version of the FFQ used in this study has been validated against 3-d dietary records in the general Finnish population (22, 23). However, there were also some limitations. Our sample was not population-based, which limited the generalizability of our results. Because the study design was cross-sectional, no conclusions could be drawn about the causal relation between fructose intake and prevalence of NAFLD. Thus, we could not rule out the possibility that the prevalence of NAFLD leads to a lower consumption of fructose, eg, due to dietary counseling that NAFLD patients may have received. Furthermore, the FFQ’s validity to measure dietary intake in NAFLD patients has not been separately examined. An overunderestimation or underestimation of fructose-containing foods (eg, fruit and soft drinks) may have affected our results to some extent. Fructose can also be seen as a surrogate for a healthy or unhealthy lifestyle, which is probably associated with NAFLD. We carefully controlled the analysis for lifestyle factors such as physical activity and smoking, but some residual confounding could still have remained because measurements of these factors were based solely on questionnaires. Moreover, we may have failed to take some unknown factor into account (eg, a genetic factor) associated with fructose intake, which also could have confounded the results. The NAFLD diagnosis was based on the FLI algorithm test rather than the gold standard liver biopsy, which limited conclusions of our study. A liver biopsy would not have been a feasible option in our study setting because it is time consuming, costly, and associated with a small but notable procedure-related morbidity and even cases of mortality (38). Alternative methods of diagnosing NAFLD include the use of ultrasound, computer tomography, and magnetic resonance
Downloaded from ajcn.nutrition.org at UNIV OF CALIFORNIA SAN FRANCISCO KALMANOVITZ LIBRARY on December 23, 2014
FLI (cases/total) (n) Crude4 Model 1 Model 2 Model 3 NAFLD liver fat score (cases/total) (n) Crude4 Model 1 Model 2 Model 3
1 (2.2–15.2 g/d)3
FRUCTOSE INTAKE AND NAFLD
The authors’ responsibilities were as follows—JGE: designed the HBCS and was responsible for the original study idea; SM and NEK: were responsible for dietary data collected in the HBCS and critically revised the manuscript; SS and NEK: participated in data handling; NK: performed statistical analyses; NK and SS: wrote the manuscript and had primary responsibility for the final content of the manuscript; and all authors: approved the final manuscript. None of the authors had a conflict of interest.
REFERENCES 1. Vanni E, Bugianesi E, Kotronen A, De Minicis S, Yki-Ja¨rvinen H, Svegliati-Baroni G. From the metabolic syndrome to NAFLD or vice versa? Dig Liver Dis 2010;42:320–30. 2. Marchesini G, Brizi M, Bianchi G, Tomassetti S, Bugianesi E, Lenzi M, McCullough AJ, Natale S, Forlani G, Melchionda N. Nonalcoholic fatty liver disease a feature of the metabolic syndrome. Diabetes 2001; 50:1844–50. 3. Bellentani S, Scaglioni F, Marino M, Bedogni G. Epidemiology of nonalcoholic fatty liver disease. Dig Dis 2010;28:155–61. 4. Loomba R, Sanyal AJ. The global NAFLD epidemic. Nat Rev Gastroenterol Hepatol 2013;10:686–90. 5. Puri P, Sanyal AJ. Nonalcoholic fatty liver disease: definitions, risk factors, and workup. Clin Liver Dis 2012;1:98–102. 6. Musso G, Gambino R, Cassader M, Pagano G. Meta-analysis: natural history of non-alcoholic fatty liver disease (NAFLD) and diagnostic accuracy of non-invasive tests for liver disease severity. Ann Med 2011;43:617–49. 7. Nakao H, Yoneda M. The intertwisted correlations among nonalcoholic fatty liver disease, atherosclerosis, and metabolic syndrome. J Gastroenterol 2009;44:1162–4.
8. Lu H, Liu H, Hu F, Zou L, Luo S, Sun L. Independent association between nonalcoholic fatty liver disease and cardiovascular disease: a systematic review and meta-analysis. Int J Endocrinol 2013;2013: 124958. 9. Lewis JR, Mohanty SR. Nonalcoholic fatty liver disease: a review and update. Dig Dis Sci 2010;55:560–78. 10. Vos MB, Lavine JE. Dietary fructose in nonalcoholic fatty liver disease. Hepatology 2013;57:2525–31. 11. Neuschwander-Tetri BA. Carbohydrate intake and nonalcoholic fatty liver disease. Curr Opin Clin Nutr Metab Care 2013;16:446–52. 12. Abdelmalek MF, Suzuki A, Guy C, Unalp-Arida A, Colvin R, Johnson RJ, Diehl AM; Nonalcoholic Steatohepatitis Clinical Research Network. Increased fructose consumption is associated with fibrosis severity in patients with nonalcoholic fatty liver disease. Hepatology 2010;51:1961–71. 13. Sievenpiper JL, Carleton AJ, Chatha S, Jiang HY, de Souza RJ, Beyene J, Kendall CW, Jenkins DJ. Heterogeneous effects of fructose on blood lipids in individuals with type 2 diabetes systematic review and metaanalysis of experimental trials in humans. Diabetes Care 2009;32: 1930–7. 14. Johnston RD, Stephenson MC, Crossland H, Cordon SM, Palcidi E, Cox EF, Taylor MA, Aithal GP, Macdonald IA. No difference between high-fructose and high-glucose diets on liver triacylglycerol or biochemistry in healthy overweight men. Gastroenterology 2013;145: 1016–25. 15. Sandboge S, Pera¨la¨ M-M, Salonen MK, Blomstedt PA, Osmond C, Kajantie E, Barker DJ, Eriksson JG. Early growth and non-alcoholic fatty liver disease in adulthood-the NAFLD liver fat score and equation applied on the Helsinki Birth Cohort Study. Ann Med 2013;45:430–7. 16. Kotronen A, Peltonen M, Hakkarainen A, Sevastianova K, Bergholm R, Johansson LM, Lundbom M, Rissanen A, Ridderstra˚le M, Groop L, et al. Prediction of non-alcoholic fatty liver disease and liver fat using metabolic and genetic factors. Gastroenterology 2009;137:865–72. 17. World Health Organization. Definition, diagnosis and classification of diabetes mellitus and its complications. Geneva, Switzerland: World Health Organization, 1999:31–3. 18. Alberti KG, Eckel RH, Grundy SM, Zimmet PZ, Cleeman JI, Donato KA, Fruchart JC, James WP, Loria CM, Smith SC Jr, et al. Harmonizing the metabolic syndrome: a joint interim statement of the International Diabetes Federation Task Force on Epidemiology and Prevention; National Heart, Lung, and Blood Institute; American Heart Association; World Heart Federation; International Atherosclerosis Society; and International Association for the Study of Obesity. Circulation 2009;120:1640–5. 19. Bedogni G, Bellentani S, Miglioli L, Masutti F, Passalacqua M, Castiglione A, Tiribelli C. The Fatty Liver Index: a simple and accurate predictor of hepatic steatosis in the general population. BMC Gastroenterol 2006;6:33. 20. Kunst A, Draeger B, Ziegenhorn J. UV-methods with hexokinase and glucose-6-phosphate dehydrogenase. Chapter 6. In: Bergmeyer HU, ed. Methods of enzymatic analysis. 3rd ed. Weinheim, Germany: Verlag Chemie, 1984;163–72. 21. Sobey WJ, Beer S, Carrington CA, Clark P, Frank B, Gray I, Luzio SD, Owens DR, Schneider AE, Siddle K, et al. Sensitive and specific twosite immunoradiometric assays for human insulin, proinsulin, 65-66 split and 32-33 split proinsulins. Biochem J 1989;260:535–41. 22. Ma¨nnisto¨ S, Virtanen M, Mikkonen T, Pietinen P. Reproducibility and validity of a food frequency questionnaire in a case-control study on breast cancer. J Clin Epidemiol 1996;49:401–9. 23. Paalanen L, Ma¨nnisto¨ S, Virtanen MJ, Knekt P, Ra¨sa¨nen L, Montonen J, Pietinen P. Validity of a food frequency questionnaire varied by age and body mass index. J Clin Epidemiol 2006;59:994–1001. 24. Reinivuo H, Hirvonen T, Ovaskainen M-L, Korhonen T, Valsta LM. Dietary survey methodology of FINDIET 2007 with a risk assessment perspective. Public Health Nutr 2010;13:915–9. 25. Chalasani N, Younossi Z, Lavine JE, Diehl AM, Brunt EM, Cusi K, Charlton M, Sanyal AJ. The diagnosis and management of nonalcoholic fatty liver disease: practice guideline by the American Association for the Study of Liver Diseases, American College of Gastroenterology, and the American Gastroenterological Association. Hepatology 2012;55:2005–23. 26. R Development Core Team. R: a language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing, 2013.
Downloaded from ajcn.nutrition.org at UNIV OF CALIFORNIA SAN FRANCISCO KALMANOVITZ LIBRARY on December 23, 2014
imaging. However, these methods were not included in the HBCS protocol. In a previous publication, in which associations between early growth and NAFLD were studied (15), we used the NAFLD liver fat score and equation (16) to define NAFLD. For purposes of the current study, we also used the FLI because it is more widely used. The FLI was validated against an ultrasound in the original publication and has been used in several publications as a surrogate for NAFLD (39–41). An important limitation in the use of an algorithm test, rather than a moreobjective method, was risk of misclassification. We are currently conducting a validation study of both algorithm tests against a diagnosis by using magnetic resonance spectroscopy on a small subset of the study population (n = 41). In the comparison of the 2 scores, the NAFLD liver fat score has had a better balance in sensitivity and specificity (NAFLD liver fat score: sensitivity of 85% and specificity of 71%; FLI: sensitivity of 61% and specificity of 86%). Because of a higher probability of detecting false positives, the FLI score may have incorrectly classified individuals with a healthy liver and their dietary habits to the NAFLD cohort, thereby skewing the true trend between fructose intake and NAFLD cases and leading to inconsistent results between the 2 scores. However, the level of agreement between the 2 scores was considered acceptable (76.3% for a negative finding and a 78.9% for positive finding). In conclusion, high intake of fructose was not associated with higher prevalence of NAFLD assessed by suing the FLI and NAFLD liver fat score. Because calculations of daily nutrient intakes did not separate whether the fructose came from natural or artificial sources, a major question about the possible different effects of these 2 sources on the development of NAFDL remains. Furthermore, because our study was cross-sectional and concerned older adult Finns born in Helsinki, population-based longitudinal studies are needed to confirm the causality of these associations.
1137
1138
KANERVA ET AL 35. Assy N, Nasser G, Kamayse I, Nseir W, Beniashvili Z, Djibre A, Grosovski M. Soft drink consumption linked with fatty liver in the absence of traditional risk factors. Can J Gastroenterol 2008;22:811–6. 36. Abid A, Taha O, Nseir W, Farah R, Grosovski M, Assy N. Soft drink consumption is associated with fatty liver disease independent of metabolic syndrome. J Hepatol 2009;51:918–24. 37. Petta S, Marchesini G, Caracausi L, Macaluso FS, Camma` C, Ciminnisi S, Cabibi D, Porcasi R, Craxı` A, Di Marco V. Industrial, not fruit fructose intake is associated with the severity of liver fibrosis in genotype 1 chronic hepatitis C patients. J Hepatol 2013;59:1169–76. 38. Weigand K, Weigand K. Percutaneous liver biopsy: retrospective study over 15 years comparing 287 inpatients with 428 outpatients. J Gastroenterol Hepatol 2009;24:792–9. 39. Calori G, Lattuada G, Ragogna F, Garancini MP, Crosignani P, Villa M, Bosi E, Ruotolo G, Piemonti L, Perseghin G. Fatty Liver Index and mortality: the Cremona study in the 15th year of follow-up. Hepatology 2011;54:145–52. 40. Gastaldelli A, Kozakova M, Højlund K, Flyvbjerg A, Favuzzi A, Mitrakou A, Balkau B. RISC Investigators. Fatty liver is associated with insulin resistance, risk of coronary heart disease, and early atherosclerosis in a large European population. Hepatology 2009;49: 1537–44. 41. Kozakova M, Palombo C, Paterni Eng M, Dekker J, Flyvbjerg A, Mitrakou A, Gastaldelli A, Ferrannini E; RISC Investigators. Fatty liver index, gamma-glutamyltransferase, and early carotid plaques. Hepatology 2012;55:1406–15.
Downloaded from ajcn.nutrition.org at UNIV OF CALIFORNIA SAN FRANCISCO KALMANOVITZ LIBRARY on December 23, 2014
27. Willett W, Stampfer MJ. Total energy intake: implications for epidemiologic analyses. Am J Epidemiol 1986;124:17–27. 28. Rothman KJ, ed. Modern Epidemiology. 1st ed. Boston, MA: Little, Brown and Company, 1986. 29. de Castro UG, dos Santos RA, Silva ME, de Lima WG, CampagnoleSantos MJ, Alzamora AC. Age-dependent effect of high-fructose and high-fat diets on lipid metabolism and lipid accumulation in liver and kidney of rats. Lipids Health Dis 2013;12:136. 30. Pietinen P, Paturi M, Reinivuo H, Tapanainen H, Valsta LM. FINDIET 2007 Survey: energy and nutrient intakes. Public Health Nutr 2010;13: 920–4. 31. Westerbacka J, Lammi K, Ha¨kkinen A-M, Rissanen A, Salminen I, Aro A, Yki-Ja¨rvinen H. Dietary fat content modifies liver fat in overweight nondiabetic subjects. J Clin Endocrinol Metab 2005;90: 2804–9. 32. Sobrecases H, Leˆ K-A, Bortolotti M, Schneiter P, Ith M, Kreis R, Boesch C, Tappy L. Effects of short-term overfeeding with fructose, fat and fructose plus fat on plasma and hepatic lipids in healthy men. Diabetes Metab 2010;36:244–6. 33. Ngo Sock ET, Leˆ K-A, Ith M, Kreis R, Boesch C, Tappy L. Effects of a short-term overfeeding with fructose or glucose in healthy young males. Br J Nutr 2010;103:939–43. 34. Chiu S, Sievenpiper JL, de Souza RJ, Cozma AI, Mirrahimi A, Carleton AJ, Ha V, Di Buono M, Jenkins AL, Leiter LA, et al. Effect of fructose on markers of non-alcoholic fatty liver disease (NAFLD): a systematic review and meta-analysis of controlled feeding trials. Eur J Clin Nutr 2014;68:416–23.
