Effects of excessive doses of fructose

Already in 1993, researchers⁽ Reference Mayes ^{71 )} agreed that excessive fructose consumption (then defined as 7·5 % to 70 % of total energy intake) induces immediate de novo lipogenesis in both animals and humans, because, in different experimental settings, it circumvented substrate inhibition feedback mechanisms that are present for glucose when it enters glycolysis. It was shown that the dietary fructose fraction not converted to lactate in the intestinal epithelium was rapidly taken up by the liver, where it was subsequently converted first into fructose-1-phosphate, and then to triose-phosphate and pyruvate/lactate. These are both potential substrates for liver glycogen synthesis and for fatty acid production, leading to an increased TAG release from the liver into blood. Also, it was found that fructose stimulated key lipogenic enzymes by activating sterol regulatory element-binding protein-1c (SREBP-1c) in the livers of mice⁽ Reference Shimomura, Matsuda and Hammer ^{72 )}.

In addition, it was found that high fructose loads (50 % of total diet) led to an increase in PPARγ co-activator 1α and 1β (PGC-1α and PGC-1β) which promoted insulin resistance and lipogenesis⁽ Reference Rizkalla, Luo and Guilhem ^{73 ,} Reference Havel ^{74 )}, as well as decreased insulin receptor activation and insulin receptor substrate phosphorylation⁽ Reference Puigserver and Spiegelman ^{75 )}. Subsequently, lipogenesis induced by this high fructose load was associated with the formation of larger fat deposits in adipose tissue and muscle, in animal models⁽ Reference Rizkalla, Luo and Guilhem ^{73 ,} Reference Havel ^{74 )}. However, to the best of our knowledge, there are no results of long-term human intervention studies available in which comparable quantities of fructose were investigated. One short-term intervention study (96 h) examined the effects of 50 % excess energy as fructose, sucrose or glucose, and indicated that, even under these drastic conditions, de novo lipogenesis remained a minor pathway for fructose disposal in both lean and obese women⁽ Reference McDevitt, Bott and Harding ^{76 )}.

Hyperuricaemia may occur as a consequence of rapid fructose entry from portal blood into the liver, where fructose will reduce the total adenosine nucleotide (TAN) pool in liver cells. A degradation of hepatic TAN will result in the production of uric acid. In a within-subjects intervention, this was measured in obese men and women where pure fructose intake provided 30 % of total energy intake⁽ Reference Teff, Grudziak and Townsend ^{77 )}. Chronic hyperuricaemia was also proposed to act as a promoter of insulin resistance and type 2 diabetes development⁽ Reference Vuorinen-Markkola and Yki-Järvinen ^{78 )}. Based on recent findings from in vivo research in fructose-fed rats, it was suggested that uric acid may impair the action of insulin by decreasing insulin-mediated muscle vasodilatation⁽ Reference Nakagawa, Hu and Zharikov ^{79 )}. In addition, it may possibly act as an intracellular mediator to enhance hepatic de novo lipogenesis⁽ Reference Lanaspa, Sanchez-Lozada and Cicerchi ^{80 )}. It remains unclear if these metabolic consequences can occur in humans considering moderate fructose intake level and complex dietary composition. We will discuss this in detail below.

In older adults consuming fructose daily through SSB, fructose led to stressful conditions in hepatocytes⁽ Reference Abdelmalek, Suzuki and Guy ^{81 )} resulting in the release of TNF-α, a strong pro-inflammatory messenger involved in insulin resistance development⁽ Reference Togashi, Ura and Higashiura ^{82 )}. Also in rats, excessive fructose intake (>62 % of total energy) induced oxidative stress, mitochondrial and endothelial dysfunction, resulting in hypertension⁽ Reference Bezerra, Ueno and Silva ^{83 )}.

In summary, it appears that excessive fructose intake can have deleterious metabolic effects in both animals and humans.

Disputable interpretations

In contrast to these deleterious effects observed in animal models and in human trials with excessive intakes, the metabolic effects of fructose presented in ordinary human diets remain poorly investigated and highly controversial. The assumption that fructose was directly involved in the occurrence of obesity relied on correlation data between the increase in HFCS consumption and obesity prevalence in the USA. This assumption has been considered as misleading for several reasons.

First, the correlation of HFCS and obesity data only happened in North America. In Europe, there was also an increase in obesity prevalence during the same period, but HFCS was not consumed to any significant amount. Moreover, the term HFCS often led individuals to believe that it had a very high fructose content. In fact, the relative proportion of fructose to glucose in HFCS 55 (55 % fructose; used in most soft drinks) and HFCS 42 (42 % fructose; mostly used in non-beverage applications) is not that different from sucrose (50:50 %)⁽ Reference White ^{46 )}, although absolute levels as analysed in drinks may vary. In this respect, free fructose content in sucrose-sweetened acid-containing beverages, such as colas, was found to be increased during storage due to acid-induced sucrose hydrolysis⁽ Reference Babsky, Toribio and Lozano ^{84 ,} Reference Ventura, Davis and Goran ^{85 )}.

A prospective cohort study⁽ Reference Fung, Malik and Rexrode ^{86 )} indicated that higher consumption of SSB was associated with a higher risk of CHD. Additionally, a cross-sectional study⁽ Reference Brown, Stamler and Van Horn ^{34 )} and two other cohort studies⁽ Reference Chen, Caballero and Mitchell ^{87 ,} Reference Dhingra, Sullivan and Jacques ^{88 )} positively associated a reduction in SSB consumption with a reduction of disease risk factors such as elevated blood pressure or weight gain. It should be mentioned, however, that relevant intervention studies with such risk factors as end-point are lacking. Interestingly, four large cohort studies showed no relationship between moderate sugar intake and type 2 diabetes⁽ Reference Meyer, Kushi and Jacobs ^{89 –} Reference Paynter, Yeh and Voutilainen ^{92 )}. The question of whether the aforementioned effects are really caused by fructose can therefore not be answered by the observational data since these show associations, not causality.

In this respect, it is important to note that through analysis of the same set of data, a positive association between obesity risk and SSB intakes was found without adjustment for total energy intake⁽ Reference Forshee, Anderson and Storey ^{93 )}. These outcomes from the modelling analyses may indicate that SSB consumption was not associated with obesity risk if potential impact of total energy intake was accounted for. In this light, a meta-analysis⁽ Reference Forshee, Anderson and Storey ^{93 )}, a descriptive time-series study⁽ Reference Duffey and Popkin ^{94 )} and a cohort study⁽ Reference de Koning, Malik and Rimm ^{95 )} did report a relationship between sugar intake or SSB intake and diabetes, dyslipidaemia and cardiometabolic risk factors. In all these studies, however, the relationship disappeared when the analysis was adjusted for body weight, strongly suggesting that obesity rather than sugar intake may be the factor associated with the disease status or biomarkers mentioned.

Goran et al. ⁽ Reference Goran, Ulijaszek and Ventura ^{96 )} did find that diabetes prevalence was 20 % higher in European Union (EU) countries with a higher availability of HFCS, as compared with countries with low availability. The authors stated that these differences were retained after adjusting for country-level estimates of BMI, population and gross domestic product. An analysis of the study, however, shed an interesting light on the reliability of these findings. The cited HFCS consumption data for the EU countries were, in fact, not consumption data at all but rather production data. In the EU, HFCS travels freely across EU borders and can thus be consumed anywhere. For instance, the article stressed that Hungary consumed significant amounts of HFCS and also showed a higher prevalence of diabetes⁽ Reference Kmietowicz ^{97 )}. In reality, most HFCS from Hungary, which was one of Europe's leading producers of this ingredient, was exported^{( 98 )}. Consumption and production figures are, as such, two entirely different things. Even if export and import figures were accounted for, food spoilage (which can be up to 30 %⁽ Reference Gustavsson, Cederberg and Sonesson ^{99 )}) makes a serious impact on the above findings. This is also the case in many other epidemiological research papers that have used sugar production or disappearance data as the bases for correlations with obesity, as well as papers that cite such data for building their arguments. More recently, Basu et al. ⁽ Reference Basu, Yoffe and Hills ^{100 )} used food supply data from the UN FAO to capture the market availability of different food items worldwide. From this, the authors concluded that an increase in sugar availability was associated with higher diabetes prevalence after testing for potential selection biases and controlling for other food types, total energy intake, overweight and obesity, period effects, and several socio-economic variables such as ageing, urbanisation and income. As discussed, the market availability of food is a debatable indicator for sugar consumption.

In this respect, a recent New York Times article by Strom⁽ Reference Strom ^{101 )} pointed out that, due to incorrect methodology, as discussed by Muth⁽ Reference Muth ^{102 )}, US sugar consumption in recent years has been overestimated by >20 %. Interestingly, the author implied that sugar consumption has not risen substantially since the 1980 s. This makes many assumptions based on higher production or per capita consumption data unsubstantiated. In addition, data obtained from the US National Health and Nutrition Examination Surveys, in 2005–2010⁽ Reference Ervin and Ogden ^{21 )}, concluded that total energy from added sugars remained rather constant, or even declined in some segments of the population, in recent years. Moreover, the consumption of added sugar through beverages contributed to only one-third of total added sugar intake, indicating that the energy from added sugars mostly came from foods rather than beverages.

Alternative and balancing views

In animal models, excessive-fructose diets lead to hyperphagia, obesity and the development of a metabolic syndrome⁽ Reference Rizkalla ^{27 )}. In human subjects, however, evidence is scarce. Short-term studies that used large amounts of fructose have led to relatively modest changes in metabolic profile (including hypertriacylglycerolaemia) and a moderate decrease in hepatic insulin sensitivity and no change in whole-body/muscle insulin resistance⁽ Reference Lê and Tappy ^{47 ,} Reference Lê, Ith and Kreis ^{103 ,} Reference Faeh, Minehira and Schwarz ^{104 )}. This may suggest that there is a large metabolic plasticity in response to dietary changes and what we observe are minor adjustments of metabolic pathways rather than pathogenic events.

There has been no evidence that relatively high levels of SSB consumption could be associated with obesity, diabetes or cardiometabolic risk in professional athletes who usually consume SSB as energy and dehydration drinks. On the other hand, there is evidence that physical inactivity, even within a few days, causes insulin resistance and dyslipidaemia in normal healthy individuals⁽ Reference Hamburg, McMackin and Huang ^{105 )}. In this regard, two randomised within-subjects studies in healthy males and females showed that higher plasma TAG concentrations, induced by a high-carbohydrate diet, were completely prevented by physical activity⁽ Reference Egli, Lecoultre and Theytaz ^{106 ,} Reference Koutsari, Karpe and Humphreys ^{107 )}. Thus, the metabolic consequences of a high mixed glucose–fructose intake can be significantly modulated by exercise. In a narrative review⁽ Reference Elliott, Keim and Stern ^{108 )}, it was reported that high fructose consumption induces insulin resistance, impaired glucose tolerance, hyperinsulinaemia, hypertriacylglycerolaemia and hypertension in animal models. The data in human subjects, however, were considered less clear. In this respect, fructose consumption, even in large amounts (17 % of total energy), did not result in significant effects in healthy males but did cause these effects in healthy women⁽ Reference Bantle, Raatz and Thomas ^{109 )}. Moreover, such fructose consumption did not stimulate de novo lipogenesis in premenopausal women⁽ Reference Couchepin, Lê and Bortolotti ^{110 ,} Reference Tran, Jacot-Descombes and Lecoultre ^{111 )}. In a review addressing sugars, insulin sensitivity and the postprandial state⁽ Reference Daly ^{112 )}, it was concluded that research on animals, particularly rodents, has shown a clear and consistent effect of high-sucrose and high-fructose diets in decreasing insulin sensitivity. Again, it was underlined that experiments in human subjects have produced very conflicting results, as there is only limited evidence from human consumption data, using fructose levels of higher than 15 % of daily energy intake, for such an effect on insulin sensitivity.

If it is not fructose, is it just added sugars in a solution?

The suggestion that HFCS is causal to obesity⁽ Reference Bray, Nielsen and Popkin ^{32 ,} Reference Lakhan ^{113 )} cannot explain why overweight and diabetes have also increased over the past decades in regions where HFCS is not (or is hardly being) used in soft drinks (for example, Europe and India), or where SSB consumption is limited (Asia and Africa)⁽ Reference Klurfeld, Foreyt and Angelopoulos ^{114 )}.

Several reviews and position papers have proposed that SSB are causally related to obesity because energy-containing liquids do not elicit the same satiety signals as energy-containing solid foods⁽ Reference Hu and Malik ^{17 ,} Reference DiMeglio and Mattes ^{115 –} Reference Soenen and Westerterp-Plantenga ^{118 )}. This hypothesis was partially supported by studies that showed that supplementation with SSB increased body weight, and thus that the intake of energy from other sources was not adequately suppressed⁽ Reference de Ruyter, Olthof and Seidell ^{62 ,} Reference Ebbeling, Feldman and Chomitz ^{63 )}. In such studies, however, the cumulated weight gain observed was substantially lower than expected from added SSB energy, indicating that there was at least partial compensation⁽ Reference Mærsk, Belza and Holst ^{117 ,} Reference Reid, Hammersley and Duffy ^{119 )}. This compensatory effect and other problems in this research area were highlighted recently by Allison⁽ Reference Allison ^{120 )} and quantified by Kaiser et al. ⁽ Reference Kaiser, Shikany and Keating ^{20 )}. In their meta-analysis⁽ Reference Kaiser, Shikany and Keating ^{20 )}, the observed weight gain from six randomised controlled trials in which the effect of SSB on weight gain was tested was compared with the theoretical weight gain in these studies. It was found that the observed data were, on average, 85 % lower than the theoretical weight gain, indicating a high compensation effect. In other words, the effect of added sugars on weight gain was much smaller than the theoretically assumed result. This does not mean that a frequent consumption of SSB does not make an impact on weight gain. It does show, however, that other factors do contribute significantly as well.

In a recent cross-sectional study, it was reported that US adolescents, who consumed high amounts of added sugars (20–30 % of total energy), had higher blood cholesterol and TAG compared with low sugar consumers (10–20 % of total energy)⁽ Reference Welsh, Sharma and Cunningham ^{121 )}. High sugar consumers had similar body weight and total energy intake compared with low sugar consumers, but a lower intake of energy from fat and protein, indicating that sugar intake was at least partially compensated⁽ Reference Welsh, Sharma and Cunningham ^{121 )}. Several smaller studies⁽ Reference Rodin ^{122 ,} Reference Rodin, Reed and Jamner ^{123 )} documented that liquid sugar preloads significantly reduce spontaneous food intake at subsequent buffet meals and that fructose was as efficient as glucose – in some instances even more efficient – in this regard.

Thus, these data provide a basis for arguments against the hypothesis that fructose-containing liquids have a different effect on satiety, since all energy-containing beverages seem to have similar effects⁽ Reference Te Morenga, Mallard and Mann ^{124 )}. This has led Moran⁽ Reference Moran ^{125 )} to conclude that results have been inconsistent and that particular findings concerning the effects of fructose on satiety appear to depend on the timing, eating context and volume of preload relative to the test meal. Another study⁽ Reference Sievenpiper, de Souza and Jenkins ^{126 )} listed the effect of fructose on body weight in controlled feeding trials. Herein, the authors concluded that fructose does not seem to cause weight gain when it is substituted for other carbohydrates in diets providing similar energy content. In this respect, the question arises whether consuming energy through beverages results in fewer satiety signals compared with energy from solid foods. To answer this, the US 2010 Dietary Guidelines Advisory Committee⁽ Reference Slavin ^{127 )} reviewed the literature and concluded: ‘A limited body of evidence shows conflicting results about whether liquid and solid foods differ in their effects on energy intake and body weight, except that liquids in the form of soup may lead to decreased energy intake and body weight.’

Most recently, Page et al. ⁽ Reference Page, Chan and Arora ^{128 )} performed a study on neurophysiological factors that might underlie associations between fructose consumption and weight gain. For this purpose, twenty healthy adult volunteers underwent two MRI sessions at Yale University in conjunction with fructose or glucose drink ingestion in a blinded, random-order, cross-over design. The authors concluded that glucose but not fructose ingestion reduced the activation of the hypothalamus, insula and striatum – brain regions that regulate appetite, motivation and reward processing. Glucose ingestion also increased functional connections between the hypothalamic–striatal network and increased satiety. The disparate responses to fructose were associated with lower systemic levels of the satiety-signalling hormone insulin and were not probably attributable to an inability of fructose to cross the blood–brain barrier into the hypothalamus, or to a lack of hypothalamic expression of genes necessary for fructose metabolism. The authors discussed a number of limitations of this well-designed study, but did not consider the possibility that the observed effects were merely mediated by hyperinsulinaemia present after glucose, but not after fructose, ingestion. They also did not discuss that, in real life, fructose is never consumed as a single carbohydrate source but always together with glucose. Thus, dietary intakes of sucrose and HFCS all raise insulin levels significantly, and should not induce the observed brain responses to feeding fructose alone.

Fructose, uric acid and insulin resistance

In 2009, Johnson et al. ⁽ Reference Johnson, Perez-Pozo and Sautin ^{133 )} hypothesised that excessive fructose intake (>50 g/d) may be one of the underlying factors in the aetiologies of the metabolic syndrome and type 2 diabetes. The authors suggest that this occurs through mechanisms by which rapidly increased fructose phosphorylation in liver cells results in total adenine nucleotide degradation leading to the liberation of elevated uric acid, leading to higher cardiovascular risk⁽ Reference Feig, Kang and Johnson ^{134 )}. In a study of Sánchez-Lozada et al. ⁽ Reference Sánchez-Lozada, Mu and Roncal ^{135 )}, rats were fed either a combination of 30 % fructose and 30 % glucose or 60 % sucrose, while control rats were fed normal rat chow containing 60 % maize starch. Diets containing 30 % of either both free fructose and free glucose, or as the disaccharide sucrose, induced the metabolic syndrome, intra-hepatic accumulation of uric acid and TAG, leading to fatty liver. Relevant for the interpretation of this study is that the level of fructose consumed by the rats was excessive and does not reflect levels consumed by humans.

Another study, by Abdelmalek et al. ⁽ Reference Abdelmalek, Lazo and Horska ^{91 )}, investigated twenty-five diabetic adults receiving an intravenous fructose challenge. Based on their data, the authors concluded that high fructose consumption depletes hepatic ATP and impairs recovery from ATP depletion after an intravenous fructose challenge. This approach, however, relied on the intravenous administration of >25 g of pure fructose (250 mg/kg body weight) within 1 min, resulting in a massive hepatic disposal. Similar ATP depletion has also been observed with large oral fructose load, but led to only small increases in uric acid concentrations in healthy subjects⁽ Reference Oberhaensli, Taylor and Rajagopalan ^{136 )}. However, it has been recently reported that ingestion of even larger amounts of fructose failed to acutely increase uric acid concentration when ingested in split doses throughout several days, suggesting that liver ATP depletion is unlikely to occur with usual patterns of sugar consumption⁽ Reference Lecoultre, Egli and Theytaz ^{137 )}.

Lin et al. ⁽ Reference Lin, Huang and Huang ^{138 )} also observed that fructose consumption resulted in higher serum uric acid levels in individuals with a BMI of >30 kg/m². Interestingly, their study showed that there was no effect of fructose intake in the subjects with a BMI between 25 and 29 kg/m², although serum uric acid showed a trend to be elevated depending on body-weight status. Moreover, the blood samples were drawn after an overnight fast in the morning, ruling out any postprandial effect of fructose ingestion. Accordingly, the effect on serum uric acid was more likely to be secondary to obesity or the metabolic syndrome than to fructose consumption per se. In this study, intake was calculated from food frequency recall, which is known to have a low level of accuracy⁽ Reference Dodd, Guenther and Freedman ^{139 )}. Moreover, food frequency intake data are based on food composition tables that are not controlled for recipe-related changes in food and beverage products on the market. This double chance of error should not be neglected.

Limited data are available on serum uric acid changes after realistic dietary loads of fructose-containing sugars⁽ Reference Angelopoulos, Lowndes and Zukley ^{140 )}. For example, Akhavan & Anderson⁽ Reference Akhavan and Anderson ^{141 )} tested solutions containing different ratios of glucose and fructose. In their study, overnight-fasted men received a standardised breakfast in the morning. At 4 h later, a 300 kcal (1255 kJ) drink was ingested within 3 min. The solutions were sweetened with either HFCS containing 55 % of fructose, sucrose or the monosaccharide forms of glucose and fructose in specific ratios as follows: 80 % glucose: 20 % fructose (G80:F20), sucrose, G50:F50, G35:F65 and G20:F80. At 75 min, uric acid concentrations were highest after G20:F80. The sucrose and F50:G50 solutions each resulted in significantly lower uric acid concentrations than did the G20:F80 solution, but they did not differ significantly from any other solutions. The uric acid AUC did not differ significantly after the G35:F65, G50:G50 and sucrose solutions. In other research⁽ Reference Buemann, Toubro and Holst ^{142 )}, only a weak response of serum uric acid to fructose was found.

Very recently, Wang et al. ⁽ Reference Wang, Sievenpiper and de Souza ^{143 )} conducted a systematic review and meta-analysis of controlled fructose-feeding trials. The authors noted that hyperenergetic supplementation of control diets with excessive fructose (+35 % excess energy, i.e. 213–219 g/d) significantly increased serum uric acid compared with the control diets in non-diabetic participants (mean difference 31·0 (95 % CI 15·4, 46·5) mmol/l). Confounding from excessive energy could not be ruled out in the hyperenergetic trials, because no uric acid-increasing effect of tested fructose, isoenergetically exchanged with other carbohydrates, was noted in either the non-diabetic or diabetic trial.

Zgaga et al. ⁽ Reference Zgaga, Theodoratou and Kyle ^{144 )} recently observed a positive association between plasma uric acid and SSB consumption, but no association with fructose intake, leading the authors to suggest that fructose is not the causal agent underlying the SSB–urate association. In another cross-sectional study⁽ Reference Sun, Flickinger and Williamson-Hughes ^{145 )}, it was also concluded that higher dietary fructose intake was not associated with a higher hyperuricaemia risk in healthy adults. This is in line with the results of a meta-analysis⁽ Reference Livesey and Taylor ^{146 )} and review⁽ Reference Livesey ^{147 )} that refuted the relationship between normal dietary consumption of sugars containing fructose and diabetes.