Health Effects of Fructose and Fructose-Containing Caloric Sweeteners: Where Do We Stand 10 Years After the Initial Whistle Blowings?

A role of FCCS in the pathogenesis of metabolic diseases was originally supported by

Far too many original articles and reviews were published in the scientific literature during this period to review each one in detail here. Evaluating the impact of this large amount of studies can however be facilitated by several recent meta-analyses and critical, systematic reviews that can be grossly classified into (a) epidemiological studies, (b) clinical intervention studies, and (c) controlled mechanistic studies. The major meta-analyses published between 2007 and February 2015 [30‐38, 39•, 40‐42, 43••, 44, 45, 46••, 47, 48••, 49••, 50‐52] are listed in Table 1, together with the type of studies included in the meta-analysis, its major results, and their authors’ main conclusions.

Epidemiological studies are based on analysis of prospective data available from large cohort of subjects with multiple dietary evaluations over time and several years of follow-up on body weight or incidence of metabolic diseases. Such studies allow identifying statistical associations between dietary intake at initial assessment or dietary changes observed between two dietary assessments and subsequent health status. As such, they are highly valuable to assess the plausibility that FCCS may be associated with metabolic diseases in real-life situations [53]. They however do not allow identifying causal relationship between FCCS and health due to the many possible confounders [54].

Epidemiological prospective studies show a strong association between FCCS intake and body weight gain [40, 55, 56••]. Most studies specifically addressed the effects of sugar-sweetened beverage (SSB) consumption, however, and the association between consumption of sugar in solid foods and obesity is relatively understudied. The results indicate that SSB consumption is strongly associated with higher total energy intakes [57], and that either SSB, total fructose, or total FCCS consumption is strongly associated with body weight gain over time. They however also indicate that other dietary food components, mainly fried potatoes, red meat, and processed meat, are also involved in body weight gain [56••]. Furthermore, FCCS consumption is also associated with the incidence of dyslipidemia, insulin resistance, and type 2 diabetes [58], with incidence or risk factors for cardiovascular diseases [59‐62], with cardiovascular mortality [63•], with chronic kidney diseases [39•], and with hyperuricemia and gout [64]. The strength of these associations is generally reduced when data are adjusted for body weight, suggesting that they are, at least in part, secondary to increased body fat mass.

What can be concluded at this stage? The epidemiological data available so far are quite consistent in pointing toward a positive association between FCCS consumption and the development of obesity. The association is particularly robust for SSB. These studies further indicate that the association between SSB consumption and body weight is largely due to SSB being associated with increased total energy intake [31, 48••].

Many clinical intervention studies have assessed the effects of reducing or increasing SSB intake in adults and children. These studies have been further analyzed in very complete meta-analyses [31, 48••]. Increased consumption of SSB has been shown to result in significant weight gain compared to control in both adults and children, and this effect has been attributed essentially to an increased total energy intake. Changes in body weight observed in some of these studies were however much lower than what would have been expected from cumulated total energy from SSB [65••, 66••, 67, 68]. Reduction of SSB consumption was generally associated with significant weight loss and reduced total energy intake. Here again, in many trials, weight loss was far less than what would have been expected based on cumulated SSB energy deficit, suggesting either low compliance to SSB restriction or compensatory increase in total energy intake. Reducing SSB intake was however inconsistently efficient in children, most likely due to low compliance in this class of age [31, 48••].

What have we learned from intervention trials? First, many trials have assessed the effects of adding sugar to the diet. Quite expectedly, increasing sugar consumption was associated with some weight gain. The effects of increasing sugar consumption was not compared to those of increasing starch or fat consumption, and hence, these studies fall short of demonstrating that weight gain is specifically related to sugar. Second, most studies showed a significant reduction of body weight when SSB consumption was reduced, which certainly indicates that the intervention significantly decreased total energy intake. Although the body weights reported in the intervention and control arms were statistically significant, the actual effect on adiposity was often very low: in one study, BMI increased from 30.4 to 30.5 kg/m² in obese adolescent who replaced SSBs with water and from 30.1 to 30.6 kg/m² in the control group [66••]; in another study, body weight non-significantly increased by 1.3 % in adolescent receiving SSBs for 6 months, compared to 0.2 and 0.6 % in those receiving artificially sweetened sodas or water, respectively [69•]. This casts serious doubt on the clinical efficacy of a reduction of SSB consumption alone.

Total energy intake is an important confounding factor in these studies. Studies having compared excess energy intake from FCCS to weight-maintenance, low-FCCS diets consistently demonstrate that excess FCCS energy consumption can, within a few days to a few weeks, increase body weight and body fat stores [70], enhance hepatic glucose production [71], impair hepatic suppression of glucose production by insulin [71‐73], increase fasting and postprandial blood triglyceride concentration [29, 70, 74, 75], increase intrahepatic fat concentration [76•], and cause a rise in blood uric acid concentrations [70, 77]. Two studies suggested that excess fructose may increase visceral fat mass while excess glucose would increase mainly subcutaneous fat [70]. In one of these studies [70], fructose-induced increase in visceral fat was observed only in males, but not in females. Altogether, these mechanistic studies unfortunately cannot sort out the relative roles of FCCS and energy intakes. In addition, many of these studies assessed the effects of pure fructose rather than sucrose or HFCS, and many of them used levels of daily FCCS intake well above consumption observed in the general population.

Hypercaloric FCCS diets consistently caused an increase in blood triglyceride and uric acid concentrations, while hypercaloric high-glucose or high-fat diets did not. One meta-analysis concluded that high-fructose intake was associated with an increased blood triglyceride only when associated with excess energy intake, but had no effects when part of a weight-maintenance diet [34]. In contrast, another study concluded that dietary sugars influence blood lipids independently of body weight [49••]. One meta-analysis specifically addressed the effects of sugar on blood cholesterol and reported that adverse effects of sugars were observed for doses higher than 100 g/day [37].

Many narrative reviews and position papers propose that insulin resistance is a well-recognized effect of dietary fructose [78‐80]. Moderate increases in fasting hepatic glucose production are indeed observed within a few days on a high-fructose diet but are not associated with clinically relevant increases in blood glucose. Surprisingly, all studies having actually assessed muscle insulin sensitivity by hyperinsulinemic-euglycemic clamps failed to report any muscle insulin resistance in healthy individuals after 1–4 weeks on a high-fructose diet [71‐73, 81] or in subjects with type 2 diabetes after 3 months [82]. This strongly suggests that fructose per se, independently of long-term weight changes, does not impair insulin’s actions in muscle. Interestingly, one study observed a significant muscle insulin resistance in middle-aged, overweight offspring of type 2 diabetes subjects after 6 days on a high-fructose diet [83], while another similar study, performed in young, non-overweight offspring of type 2 diabetes, failed to document any changes [81]. This raises the possibility that fructose’s effects may be modulated by both genetic background and additional factors such as age or body fat mass.

Whether FCCS are responsible for an excess food intake through specific effects involving food intake control has been the focus of active research and will not be reviewed comprehensively here. It has been proposed that fructose fails to inhibit the release of the orexigenic gut hormone ghrelin and increases the release of satiating gut hormones such as GLP-1 and PYY to a lesser extent than glucose or starch [29]. This concept has however been challenged by a recent study showing that fructose had substantial effects on these gut hormones in normal-weight adolescent. The same study however reported an impaired suppression of acyl-ghrelin, the active form of ghrelin, in obese, insulin-resistant adolescent [84]. Furthermore, the hypothesis that fructose may exert less suppression of food intake than glucose or complex carbohydrate is based on indirect markers of food intake control, which relates essentially to the so-called homeostatic systems of food intake regulation, located in the brain stem and hypothalamus. Studies with direct measures of food intake and/or satiety feeling failed to document any difference between fructose and other caloric sweeteners in humans [28, 85, 86]. Food intake regulation results from complex interactions between metabolic, hormonal, and neurologic signals [87]. There is growing evidence that the hedonic system of food intake, located in neocortical structures, and which is involved in food preferences and palatability, plays a prominent role in food intake control [88]. It has further been suggested, based on the fact that oral sugar ingestion stimulates dopaminergic neurons of the reward system of the brain stem, hypothalamus, and cortex, that this may confer to FCCS potentially addictive potential, at least in animals [89‐92] and possibly in humans as well [93]. These are early hypotheses, which still need more studies to be fully evaluated. The concept of sugar addiction is highly debated but suffers from lack of clear diagnosis criteria. Finally, there is to date no scientific evidence that in humans, sugars stimulate these neurological pathways to a larger extent than other food substances with similar palatable properties.

The effect of intravenous fructose administration on uric acid production and to cause hyperuricemia has been long known [94]. Interesting novel observations have recently been made to link FCCS-induced uric acid production to metabolic and cardiovascular diseases. According to these novel hypotheses, uric acid may act as a mediator to induce endothelial cell dysfunction, thus preventing insulin-induced muscle vasodilation. This effect of uric acid has been postulated to contribute to fructose-induced insulin resistance and high blood pressure [95]. One isolated clinical study has supported this hypothesis by showing that a short-term hypercaloric high-fructose diet increased blood pressure and that this effect was prevented by administration of the uric acid synthesis inhibitor allopurinol [77]. This observation must however be tempered by the fact that several other studies failed to observe a significant effect of a high-fructose diet on blood pressure [45]. Besides vascular effects, uric acid has also been proposed to enhance lipogenesis and thus to contribute to dyslipidemia and hepatic steatosis [96, 97, 98••]. These novel hypotheses clearly require to be further evaluated, and more human studies are needed to assess their clinical relevance.

What clinically relevant information do these mechanistic studies offer? There is robust evidence that a high-fructose diet can increase blood triglycerides (TG) within a few days, enhance IHCL storage, and increase endogenous glucose production. These changes are highly significant, yet of small magnitudes, and blood triglyceride and glucose concentration, as well as intrahepatic fat content, remain generally within the normal range. They may of course be early markers of insulin resistance and progress with time to the development of metabolic diseases. There is however an alternative explanation which has been widely disregarded until now, i.e., that these changes may reflect mere adaptation to a high-fructose diet.

Unlike fructose, glucose and fatty acids are key energy substrates for animals, and most cells of the human organism synthesize the enzymes required for glycolysis, beta-oxidation, and degradation of glucose- or fat-derived acetyl-CoA in the Krebs cycle, coupled to ATP synthesis in the respiratory chain. In contrast, fructose cannot be used as such by most cells and requires to be first pre-processed into glucose, lactate, or fatty acids [99]. Specific fructose-metabolizing enzymes, fructokinase, aldolase B, and triokinase, are present mainly not only in the liver but also in enterocytes and tubular kidney cells where this pre-processing occurs (Fig. 1). In the liver, fructolysis, unlike glycolysis, is neither regulated by insulin nor inhibited by high concentrations of ATP or citrate. Hepatic production of glucose, lactate, and fatty acids is therefore mainly dependant on the amount of fructose ingested with a meal or a drink. The first adaptation that may occur as a result of fructose ingestion is an increase in hepatic glucose production [71‐73, 100]. As fructose is initially converted to lactate and glucose in the liver, the resulting increased blood lactate and the modest increase in hepatic glucose production observed in healthy subjects consuming a high-fructose diet do not come as a real surprise. The second adaption consists in the well-characterized increase in plasma lipids. However, the metabolic pathways involved markedly differ from those resulting from dietary fat ingestion: fructose conversion into fat occurs essentially in the liver where fructose-derived fat is either temporarily stored as intrahepatic fat and/or secreted into the blood as VLDL triglyceride. In contrast, dietary fats absorbed from the gut are released as chylomicrons in the lymph, thus avoiding first-pass hepatic metabolism to be directly stored in peripheral adipocytes. This difference in the interorgan trafficking involved in the handling and storage of these two lipid sources most likely accounts for increased blood triglyceride concentrations in high-fructose-fed subjects, since chylomicrons-TG have a half-life in circulation considerably faster than VLDL (Fig. 2).

Whether these alterations of glucose and lipid metabolism following fructose ingestion are mere adaptations to a transient change in diet or will in the long term be causally associated with adverse vascular effects remains unknown. If that was the case, it would be important to gather further clinical observation in order to determine whether such deleterious effects of FCCS are restricted to the subgroup of subjects at increased risk due to their genetic background or to specific patterns of FCCS consumption.

The scientific data reported in the literature appears somewhat conflicting at first sight. On one hand, there is no strong evidence that, at similar intake levels, sugar exerts worse deleterious effects than other energy substrates. Whether sugar would specifically impair food intake control, or cause tissue lipotoxicity, remains open questions, but levels of evidence are presently too low to lead to specific recommendations. On the other hand, a huge proportion of the population of an ever-increasing number of countries is overweight. As a nutrient accounting to 15–20 % total energy, sugar is certainly contributing to this global excess energy intake [101‐103]. Furthermore, many sugary products have a low content of essential nutrients and hence low nutritional quality, which makes them a potential target for energy reduction. This is particularly true of SSBs. There is no strong evidence that energy consumed with beverages is more obesogenic than energy consumed with solid foods and no plausible mechanism to account for such an effect. SSB may nonetheless be easily overconsumed, possibly because beverage intake is stimulated by thirst stimuli, such as blood hyperosmolarity and sodium levels in the distal renal tubules, irrespective of energy balance. As such, SSB reduction is certainly a prime target for the prevention of metabolic disorders, assuming that such decrease will not be compensated by the increase in caloric intake from solid food.

However, if sugars contribute to excess energy intake, there is also compelling evidence that sugar-devoid foods, such as potato chips and meat, are also involved [56••]. Furthermore, interventions focused solely on SSB or sugar intake have failed to achieve clinically relevant weight reduction. The excess body weight of the US population corresponds to a ca. 350–500 kcal/day excess energy intake on average [104]. In order to revert obesity, energy intake should be reduced by the same amount; in order to reach this target by specifically decreasing sugar intake, added sugar consumption should be reduced to close to zero, which may be unrealistic in the near future. Reduction of SSBs, and reformulation of sugar-rich industrial foods, may certainly contribute to reduced total sugar and energy intake, but it appears obvious that the consumption of other energy-dense foods and/or multifactorial interventions including consumer education, diet, and physical activity will be needed to achieve these goals.