REVIEW Fructose: Should We Worry?

GA Bray

Pennington Biomedical Research Center, Baton Rouge, LA, USA

Obesity is a growing problem. In the broadest strokes, it is due to a small positive energy balance that persists over a sufficiently long time. Some forms of obesity develop independent of the type of diet that is eaten, whereas others are dependent on the diet. Among the former are individuals with leptin deficiency or genetic defects in the melanocortin 4 receptor. Most human obesity, however, occurs in the presence of highly palatable foodsFfat and calorically sweetened beverages. The increase in obesity in the last 35 years has paralleled the increasing use of high-fructose corn syrup (HFCS), which first appeared just before 1970. Current soft drinks and many other foods are sweetened with this product because it is inexpensive and has useful manufacturing properties. The fructose in HFCS and sugar makes beverages very sweet, and this sweetness may underlie the relation of obesity to soft drink consumption. Fructose consumption has also been related to the metabolic syndrome and to abnormal lipid patterns. This evidence suggests that we should worry about our current level of fructose consumption, which has been increasing steadily for over 200 years and now represents over 10% of the energy intake of some people. International Journal of Obesity (2008) 32, S127–S131; doi:10.1038/ijo.2008.248

Keywords: HFCS (high fructose corn syrup); soft drinks; calories; food; cardiovascular risk

Introduction titled Pure, White and Deadly, in which he attributed the rising risks of obesity and heart disease to sugar. To quote his Obesity is a major public health problem that is obviously work again: multifactorial, and the foods we eat are clearly part of the When you come to think of it, almost all of the tempting problem. Food provides all of the energy that our bodies get. foods that are taken to satisfy appetite rather than The following quotes from two authors crystallize this issue, hunger contain carbohydrate that is either sugar or and the need to identify what these food elements might be: starch. If then there is reason to be concerned about a dietary These carbohydrate-rich foods have another character- cause of a widespread disease, one should look for some istic; they are all artificial foods that do not exist in constituent of man’s diet that has been introduced nature. recently or has increased considerably, recently.1 Yudkin’s ideas were not taken very seriously, nor were Some factor of diet and/or lifestyle must be driving those of Cleave,3 who blamed weight upward, because human biology and our under- lying genetic code cannot change in such a short time.2 yrefiningyin the carbohydrates [which] deceives the tongue and the appetite, and leads to over-consump- On the basis of reasoning that changes in food patterns tionFand this over-consumption is the sole primary might help to identify a culprit, we examined the changes in cause of obesity. food consumption in the United States from the end of World War II through the end of the 20th century using data With the recognition of the special properties of fructose, from the US Department of Agriculture. The most striking whether in HFCS or as a component of the sucrose molecule change was the increase in the use of high-fructose corn (table sugar), the concepts of Yudkin and of Cleave might be syrup (HFCS), which replaced significant amounts of sugar in reframed as ‘fructoseFpure white and deadly.’ the diet. At about the same time that HFCS was appearing in the US food supply, Professor Yudkin1 had written a book

Positive energy balance produces obesity

Correspondence: Professor GA Bray, Pennington Center, 6400 Perkins Road, Baton Rouge, LA 70708. USA. It is clear that a positive energy balance produces obesity. But E-mail: [email protected] how much surplus of energy do we need on a daily basis to Fructose: should we worry? GA Bray S128 Energy Intake abnormalities in the melanocortin-4 receptor, are clear-cut 4000 examples of genetic disturbances where the drive to eat 3800 is so overwhelming that the composition of the diet is )

-1 3600 irrelevantFthey will eat just about anything. A second 3400 group of obesities in which diet is secondary are those 3200 associated with neuroendocrine diseases, including hypotha- 3000 lamic obesity, Cushing’s syndrome and polycystic ovary syndrome, all of which are also largely independent of diet.4 2800

Energy (kcal day 2600 2400 Obesity that is dependent on diet composition 2200 At the other end of the spectrum are the types of obesity that 2000 are influenced by diet composition. These effects of food can 1900 1920 1940 1960 1980 2000 be related to the cost of foods of different types, to the Year response to portion sizes that are available and to the host response to the pleasurable effects of food.5,6 Lower prices Figure 1 Changes in energy intake during the 20th century based on F consumption data collected by the US Department of Agriculture. The tend to increase food consumption foods rich in fat and corrected data represent an estimate of the difference between available and sugar are considerably less expensive per thousand kilo- actual consumption of available calories. Source: http://www.ers.usda.gov/ calories than are fresh fruits and vegetablesFto take the publications/FoodReview/DEC2002/frvol25i3a.pdf (accessed 16 June 2008). extremes. As most of the rest of this discussion will deal with fructose account for the current rate of weight gain of 1–2 kg per year and ‘sugars,’ let me clarify what is meant by these terms. over a range of 30 years in adult life? A gain of 1 kg in 1 year ‘Sugars’ refers primarily to the small molecules in some foods represents a net storage of about 8000 kcal. As the efficiency and those used in food preparation or at the table. As of storage is around 50%, this translates into a gross intake of a group, carbohydrates consist of one or more ‘simple’ about 16 000 kcal more than the initial energy expenditure 6-carbon molecules. There are three principal 6-carbon units, during the year. If we divide 16 000 by 365 for the number of appropriately called ‘hexoses.’ These are glucose, fructose À1 days in a year, we find that an extra 50 kcal day could and galactose. The simplest combinations are those with two account for a weight gain of 1 kg in 1 year. Figure 1 shows the hexose units, such as lactose (also called milk sugar), which number of calories (energy) available from the food supply is composed of one glucose and one galactose; maltose, during the 20th century. This number was fairly stable until which is composed of two glucoses and sucrose, which is about 1980, after which there was an increase. When common table sugar and contains one glucose and one corrected for ‘plate waste,’ that is, food not eaten, these US fructose. ‘Sugar’ is also widely used to refer to all of the Department of Agriculture data suggest that there was an caloric sweeteners. It is also used in the term ‘blood sugar,’ increase in energy intake between the years 1980 and 2000 of which is more correctly called blood glucose, as there is À1 about 400 kcal day . This is more than enough energy to almost no other hexose that circulates in the blood. When account for the current rate of weight gain. larger numbers of glucose molecules are joined together, they form starches (amylose or amylopectin). It is interesting to note that nature did not select fructose to circulate in the Types of obesity blood or to make into long-chain molecules (except inulin). In general, the ‘bio-reactivity’ of glucose is less than that We have learned a lot about obesity during the 20th of galactose or fructose. Human milk has essentially no century.4 We know that it is not a single entity, and that fructose, nor do the foods that comprise most traditional there are many different ways to produce it. Some depend on diets. diet and others do not.

Calorically sweetened soft drinks Obesity that is not dependent on diet composition At one extreme are the types of obesity that develop One of the consequences of the lower farm prices of the independent of diet composition. At the other extreme are 1970s was a decrease in the price of corn (maize), which those in which diet composition plays a major role. The made it an inexpensive source of starch that could be most obvious examples of obesities that are not dependent converted into HFCS.7 With the development of isomerase on the diet composition are those due to single-gene defects technology in the late 1960s, corn starch could be converted and neuroendocrine disorders.4 The children with leptin into the highly sweet molecule, fructose, which as HFCS deficiency and those who lack leptin receptors, and indivi- (HFCS is a blend of 42 or 55% fructose with glucose) was used duals who fail to produce pro-opiomelanocortin or have to manufacture soft drinks and other highly processed foods.

International Journal of Obesity Fructose: should we worry? GA Bray S129 30 100 compensation. That is, the soft drinks are ‘added’ calories and do not lower the intake of energy in other forms 75 sufficiently to compensate for the beverage energy content. 20 The strengths of these relationships were stronger in women Obesity and in adults. Not surprisingly, the studies funded by the 50 food industry had weaker associations than those funded by 10 independent sources. 25

Percentage obese HFCS Soft drinks are clearly a part of our culture, and their consumption has increased steadily over the past 50 years. A

0 0 HFCS intake (g/capita/day) 20-ounce soft drink made with HFCS has about 250 kcal. 1961 1970 1975 1980 1985 1990 1995 2000 Thus, an extra 20-ounce soft drink each day is more than Year enough to account for the increased body weight over the

Figure 2 Changing intake of fructose and high-fructose corn syrup intake last quarter century. Soft drinks are a prominent part of the plotted against the increasing prevalence of obesity. Adapted from Bray et al.9 fast-food culture. When individuals eat at a fast-food restaurant, compared with a day when they do not, the fast-food day has a larger intake of soft drinks and The rapid increase in HFCS in the food supply is shown in French-fried potatoes, and a smaller intake of cereal, Figure 2.8 From the early 1970s through the mid-1990s, vegetables and milk.12 HFCS rapidly replaced sugar in many manufactured pro- A critique of the relationship of fructose-containing soft ducts, and almost entirely replaced sugar in soft drinks drinks to the development of obesity in children has been manufactured in the United States. In addition to being published by Dietz.13 This is summarized in Table 2. There cheap, HFCS is very sweet. We have argued that this are five studies in children that have shown a positive ‘sweetness’ in liquid form is one factor driving the con- relationship between soft drink consumption and weight sumption of the increased energy that are needed to fuel the gain.14–17 These data make the argument for reducing the current epidemic of obesity.9 consumption of calorie-sweetened beverages by children and Fructose differs in several ways from glucose, the other half adolescents overwhelmingly strong. of the sucrose (table sugar) molecule. Fructose is absorbed from the gastrointestinal tract by a different mechanism than glucose absorption. Glucose directly stimulates insulin Fructose-containing soft drink consumption and release from the pancreatic b-cell, but fructose does not. body weight change Fructose also enters muscle and other cells without depend- ing on insulin, whereas most glucose enters cells in an Several studies on the consumption of calorically sweetened insulin-dependent manner. Finally, once inside the cell, beverages in relation to the epidemic of overweight have fructose can enter the pathways that provide the triglyceride attracted significant attention.8,10 Ludwig and colleagues14 backbone (glycerol) more efficiently than glucose. Thus, reported that the intake of soft drinks was a predictor of high consumption of fructose, as occurs with the increasing initial body mass index (BMI) in children in the Planet consumption of soft drinks and the use of high-fructose Health Study. They went on to show that higher soft drink corn sweeteners, may be a ‘fat equivalent.’10 In spite of consumption also predicted the increase in BMI during these differences, it is the similarities among fructose- nearly 2 years of follow-up. Those subjects who had the containing products (either sucrose or HFCS) that need to highest soft drink consumption at baseline had the highest be emphasized. increase in BMI. In one of the few randomized, well- controlled intervention studies, Danish investigators18 showed that individuals consuming sugar-containing (half Fructose-containing calorically sweetened fructose) calorically sweetened beverages during 10 weeks beverages gained weight, whereas subjects drinking the same amount of artificially sweetened beverages lost weight. Equally The relationship of fructose consumption to energy intake important, drinking sugar-sweetened beverages was asso- and body weight and to the intake of other dietary ciated with a small, but significant, increase in blood components has been examined in both cross-sectional pressure. The Nurses’ Health Study19 also showed that and longitudinal studies.11 These are summarized in Table 1. changes in the consumption of soft drinks predicted changes Of the 12 cross-sectional studies examining the relation in body weight over several years of follow-up. A study in between energy intake and fructose-containing soft drink children, some of whom were advised to reduce the intake of consumption, 10 found a moderately positive association. ‘fizzy’ drinks and replace them with water, showed slower Among the five longitudinal studies, the strength of the weight gain than that in those not advised.20 association was stronger. The authors conclude that when In the Longitudinal Study of Child Development human beings consume soft drinks, there is little energy in Quebec (1998–2002),21 6.9% of children who were

International Journal of Obesity Fructose: should we worry? GA Bray S130 Table 1 Average effect of the size of soft drink consumption by the type of research design

Research design Cross-sectional studies Longitudinal studies Short experimental studies Overall effect size

r (95% CI) Nr(95% CI) Nr(95% CI) Nr(95% CI) N

Energy intake 0.13 (0.12, 0.14) 12 0.24 (0.23, 0.26) 5 0.24 (0.16, 0.31) 9 0.16 (0.15, 0.16) 22 Body weight 0.06 (0.03, 0.08) 12 0.03 (0.00, 0.06) 6 0.24 (0.18, 0.28) 7 0.06 (0.05, 0.08) 25 Milk intake À0.11 (À0.12, À0.10) 15 À0.21 (À0.27, À0.15) 5 F À0.12 (À0.13, À0.11) 20

Adapted from Tables 1–3 of Vartanian.11

Table 2 Relation of soft drink consumption to risk of increasing body weight participants in the Framingham Study.24 Individuals consuming one or more soft drinks per day had a higher Author Number Age range Duration Association of SS prevalence of the metabolic syndrome (OR: 1.48, 95% CI: 1.30–1.69) and an increased risk of developing the metabolic 25 Phillips et al. 132 9–10 Positive syndrome over 4 years of follow-up. Welsh et al.16 10 904 2–3 1 year Positive Berkey et al.15 12 192 9–14 Two 1 year Positive periods Ludwig et al.13 548 11–12 Baseline and Positive 2 years Conclusion Striegel-Moore et al.14 2371 9–10 9–10 years Positive

Abbreviation: SS, subjects. Adapted from Dietz.13 The current epidemic of obesity could be explained by the consumption of one extra 20-ounce soft drink each day. In non-consumers of sugar-sweetened beverages between meals addition to the energy provided from these beverages, they from the age of 2.5–4.5 years were overweight at ages 2.5, 3.5 are a major source of the fructose in our diet. A growing and 4.5 years, compared with 15.4% of children who number of studies suggest that fructose intake, particularly consumed sugar-sweetened soft drinks 4–6 times or more when accompanied by fat, may be unhealthy. The attribu- per week between meals. The overall odds ratio (OR) was tion ‘pure, white and deadly’ to sugar by Professor Yudkin in more than doubled in multivariate analysis but was in- 1972 may yet be proven partly right. It is the fructose part of creased threefold in those with lower income. the sucrose (table sugar) molecule and the fructose from HFCS that best fit the title. HFCS is clearly a marker for highly refined foodsFthe kind of food we want to avoid in Fructose and cardiovascular risk factors our diet. My conclusions will not make the caloric sweetener industry happy. As Yudkin said 25 years ago,1 ‘I suppose it is Fructose consumption, either in beverages or food, may have natural for the vast and powerful sugar interests to seek to an additional detrimental effect. In a study from Switzerland, protect themselves, since in the wealthier countries, sugar dietary fructose was found to predict an increased level of makes a greater contribution to our diets, measured in low-density lipoprotein cholesterol in children.22 Fructose, calories, than does meat or bread or any other single unlike other sugars, increases serum uric acid levels. commodity.’ One needs to evaluate these financial interests Nakagawa and colleagues23 proposed that this takes place in terms of their public health implications. This will not be when fructose is taken into the liver, its major organ for an easy task. metabolism, where adenosine triphosphate is used by the enzyme phosphofructokinase to phosphorylate fructose to fructose-1-phosphate. The adenosine-50-diphosphate can be Conflict of interest further broken down to adenosine-50-monophosphate, then to inosine 50-phosphate and finally to uric acid. Thus, the George A Bray has received consulting fees from Orexigen, metabolism of fructose in the liver leads to the production of Amylin, Theracos and Herbalife as well as lecture fees from uric acid. The authors proposed that the high levels of uric Vindico. acid could set the stage for advancing cardiovascular disease by reducing the availability of nitric oxide, which is crucial for maintaining normal blood pressure and for maintaining References normal function of blood vessel walls (endothelium).23 If this hypothesis is borne out, it will provide another reason 1 Yudkin J. Pure, White and Deadly. Penguin Books: London, 1986. that nature preferred glucose over fructose as a substrate for 2 Taubes G. Good Calories Bad Calories. Challenging the Conventional Wisdom on Diet, Weight Control and Disease metabolism during the evolutionary process. Soft drink . Alfred A Knopf: New York, 2007. consumption has been linked to the development of 3 Cleave TL. Saccharine Disease. The Master Disease of Our Time. cardiometabolic risk factors and the metabolic syndrome in Keats Publishing, Inc: New Canaan, CT, 1974.

International Journal of Obesity Fructose: should we worry? GA Bray S131 4 Bray GA. Metabolic Syndrome and Obesity. Humana Press: Totowa, 17 Welsh JA, Cogswell ME, Rogers S, Rockett H, Mei Z, NJ, 2007. Grummer-Strawn LM. Overweight among low-income preschool 5 Drewnowski A, Darmon N. The economics of obesity: dietary children associated with the consumption of sweet drinks: energy density and energy cost. Am J Clin Nutr 2005; 82 (1 Suppl): Missouri, 1999–2002. Pediatrics 2005; 115: e223–e229. 265S–273S. 18 Raben A, Vasilaras TH, Møller AC, Astrup AA. Sucrose compared 6 Faith MS, Fontaine KR, Baskin ML, Allison DB. Toward the with artificial sweeteners: different effects on ad libitum food reduction of population obesity: macrolevel environmental intake and body weight after 10 wk of supplementation in approaches to the problems of food, eating, and obesity. Psychol overweight subjects. Am J Clin Nutr 2002; 76: 721–729. Bull 2007; 133: 205–226. 19 Schulze MB, Manson JE, Ludwig DS, Colditz GA, Stampfer MJ, 7SchlosserE.Fast Food Nation. Houghton Mifflin Co.: New York, 2001. Willett WC et al. Sugar-sweetened beverages, weight gain, and 8 Nielsen SJ, Popkin BM. Patterns and trends in food portion sizes, incidence of type 2 diabetes in young and middle-aged women. 1977–1998. JAMA 2003; 289: 450–453. JAMA 2004; 292: 927–934. 9 Bray GA, Nielsen SJ, Popkin BM. Consumption of high-fructose 20 James J, Thomas P, Cavan D, Kerr D. Preventing childhood corn syrup in beverages may play a role in the epidemic of obesity by reducing consumption of carbonated drinks: cluster obesity. Am J Clin Nutr 2004; 79: 537–543. randomised controlled trial. BMJ 2004; 328: 1237. 10 Havel P. Dietary fructose: implications for dysregulation of energy 21 Dubois L, Farmer A, Girard M, Peterson K. Regular sugar- homeostasis and lipid/carbohydrate metabolism. Nutr Rev 2005; sweetened beverage consumption between meals increases risk 63: 133–157. of overweight among preschool-aged children. J Am Diet Assoc 11 Vartanian LR, Schwartz MB, Brownell KD. Effects of soft drink 2007; 107: 924–934. consumption on nutrition and health: a systematic review and 22 Aeberli I, Zimmermann MB, Molinari L, Lehmann R, l’Allemand meta-analysis. Am J Public Health 2007; 97: 667–675. D, Spinas GA et al. Fructose intake is a predictor of LDL 12 Paeratakul S, Ferdinand DP, Champagne CM, Ryan DH, Bray GA. particle size in overweight schoolchildren. Am J Clin Nutr 2007; Fast-food consumption among US adults and children: dietary and 86: 1174–1178. nutrient intake profile. JAmDietAssoc2003; 103: 1332–1338. 23 Nakagawa T, Hu H, Zharikov S, Tuttle KR, Short RA, 13 Dietz WH. Sugar-sweetened beverages, milk intake, and obesity in Glushakova O et al. A causal role for uric acid in fructose-induced children and adolescents. J Pediatr 2006; 148: 152–154. metabolic syndrome. Am J Physiol Renal Physiol 2006; 290: 14 Ludwig DS, Peterson KE, Gortmaker SL. Relation between F625–F631. consumption of sugar-sweetened drinks and childhood obesity: 24 Dhingra R, Sullivan L, Jacques RF, Wang TJ, Fox CS, Meigs JB a prospective, observational analysis. Lancet 2001; 357: 505–508. et al. Soft drink consumption and risk of developing cardio- 15 Striegel-Moore RH, Thompson D, Affenito SG, Franko DL, metabolic risk factors and the metabolic syndrome in Obarzanek E, Barton BA et al. Correlates of beverage intake in middle-aged adults in the community. Circulation 2007; 116: adolescent girls: The National Heart, Lung, and Blood Institute 480–488. Growth and Health Study. J Pediatr 2006; 148: 183–187. 25 Phillips SM, Bandini LG, Naumova EN, Cyr H, Colclough S, 16 Berkey CS, Rockett HR, Field AE, Gillman MW, Colditz GA. Sugar- Dietz WH et al. Energy-dense snack food intake in adolescence: added beverages and adolescent weight change. Obes Res 2004; longitudinal relationship to weight and fatness. Obes Res 2004; 12: 778–788. 12: 461–472.

International Journal of Obesity