Review of the nutritional benefits and risks related to intense sweeteners

A meta-analysis covering studies undertaken before 2006 along with around ten randomised experimental studies were identified to address these points. The meta-analysis of 15 randomised experimental studies [9] assessed the effects of aspartame consumed alone or with other (unspecified) IS on food and energy intake during the course of a day in adults. These measurements covered a limited number of subjects (less than 30) and highly variable time periods of a few days to 16 weeks. The main inclusion criterion for studies in this meta-analysis was the measurement of food intakes for at least 24 hours, to assess the full extent of any compensatory effects of the various meals consumed over the day. The authors conclude that consuming aspartame as a sugar substitute results in a decrease of daily energy intake by 220 Kcal on average. Moreover, the authors indicate that this substitution may be more efficient in beverages than in solid foods, since the energy supplied by liquids leads to less satiety than that supplied by solid foods. In fact, the estimated compensation rate is thought to be lower for sugars consumed in liquid form than in solid form [10]. Therefore, according to these authors, the reduction in energy intake due to the replacement of sugar with sweeteners is greater with artificially sweetened beverages than with artificially sweetened solid foods. However, the conclusions of this meta-analysis should be treated with caution, due to several methodological limitations, particularly a lack of essential information on the study selection process, the assessment of their quality and the statistics applied to assess the heterogeneity of the data taken into account. Other experimental studies (that were not included in the meta-analysis since they covered periods of less than 24 hours) analysed the effects of IS on appetite and food intake. These studies used an IS preload approximately one hour before a meal, generally in beverage form (rarely in solid form, i.e. in a food), and measured food intake and calorie intake during the next meal. All of these studies showed that irrespective of the nature of the tested IS, a preload reduced the sensation of hunger and the desire to eat, with a maximum effect immediately after its consumption. However, this effect tended to disappear before the start of the meal, which explains why most studies did not observe reduced food intake during the meal after the preload. Regarding food preferences, several studies assessed the effect of IS on the perception of sweetness (gustatory stimuli) and/or taste preferences for foods. Several studies showed that preference for a sweet food was independent of the sweetening agent (i.e. no difference between an IS and sucrose), but their results differed as to the repercussions of this preference on consumption of this food. However, these studies had extremely variable protocols and objectives, to the extent that it is difficult to compare their results and draw an overall conclusion on the effect of IS on food preferences.

Overall, based on studies dealing with occasional exposure to an IS before a meal, it is not possible to infer the effect of regular IS consumption on sweetness habituation or increased cravings for sweetened products. Most experimental studies show that the occasional consumption of IS before or during a meal has no effect on food intake or energy intake during the next meal. Occasional IS consumption before a meal reduces the sensation of hunger and the desire to eat, just like caloric sweeteners, but this effect is temporary and disappears before the start of the meal. In most cases, the use of IS as sugar substitutes results in a decrease in short-term energy intake due to their low calorie content and the lack of compensation. However, the available data cover insufficient time periods to guarantee the maintenance of this effect over the medium or long term.

Preference for sweetness is innate. It is strong at birth and then tends to decrease. However, it seems to be maintained by the repeated consumption of sweetened foods or beverages during early childhood [11]. A study [12] showed that adding aspartame or sucrose to milk favoured its consumption. Moreover, the work of Birch and their collaborators revealed that children preferred flavours associated with calorie intake, suggesting that sweetness itself is not sufficient to generate food preferences, and that energy density, just as much as (or even more than) sweetness, can determine food preferences [13, 14]. However, there are no data showing whether IS have a specific effect, in relation to caloric sweeteners, on the development of taste and food preferences. A study compared the effects of consuming 250 mL daily of artificially sweetened beverages vs sugar-sweetened beverages on the satiety and desire to eat of children aged seven to 11 years for 18 months [15]. The level of satiety was the same, irrespective of the beverage consumed.

In conclusion, based on the available studies, it is not possible to determine whether IS consumed during early childhood have a specific effect on the development of taste and food preferences or on the short- and medium-term regulation of food intake.

Two meta-analysis (Table 2) [9, 16], a systematic review [17] and several original articles [18‐29]examined the relationship between IS consumption and changes in body composition and weight. A meta-analysis [30] took into account observational studies (with nine articles included) and randomised controlled trials (RCTs, with 15 articles included) in adults and children. The section on observational studies showed no relationship between IS consumption and changes in body weight or fat mass but showed a slight increase in BMI (+0.03 kg/m² on average). The section on RCTs showed that replacing sugars with IS in sweet products resulted in moderate weight loss (with an estimated average effect of 0.8 kg) and a decrease in BMI (-0.24 kg/m² on average) for time periods ranging from three weeks to 18 months. This meta-analysis, of good methodological quality, highlights the extreme variability of results from studies with a similar design (whether RCT or observational) and the differences in results between observational studies and RCTs. In the meta-analysis by De la Hunty [9], eight studies on very heterogeneous populations (people with energy restrictions and unrestricted diets, normal-weight and obese people, in normal living conditions and in metabolic chambers) were included. According to the authors, the effect of IS on weight loss is significant. They extrapolate the theoretical reduction of 220 kcal/day related to the replacement of sugars with aspartame over the long term, and by postulating its maintenance over time, calculate that this reduction could result in weight loss of 0.2 kg per week. However, the methodological weaknesses of this meta-analysis has to be emphasised, particularly the lack of essential information related to the study selection process and the statistics applied to assess heterogeneity. The systematic review by Wiebe et al. (2011) cites two intervention studies comparing the effects of artificially sweetened drinks and sugar-sweetened drinks on BMI [31,32]. These studies, focusing on different populations (normal-weight women in one and overweight women in the other), had different results (no effect in normal-weight women, reduced weight in overweight women). Five other randomised controlled trials [21, 24, 25, 27, 29] were identified. They were all undertaken in overweight subjects and the majority focused on very small populations (between 20 and 50 subjects). Two demonstrated modest weight loss of 1.2 and 1.5 kg on average, but the other three, including the one with the largest study population (n = 318), did not show any effects on weight loss related to the consumption of artificially sweetened beverages compared to the consumption of sugar-sweetened beverages or water. There are also seven prospective observational epidemiological studies with highly heterogeneous results. One study did not show any association between IS consumption and changes in body composition [23]; four studies reported a positive association, i.e. a significantly higher body weight or waist size in IS consumers [18‐20,28]; and two studies reported a negative association [22, 26].

In conclusion, observational and intervention studies report contradictory associations between IS consumption and weight loss. Therefore, no conclusion can be drawn as to the long-term effect of replacing caloric sweeteners with IS on the weight of regular adult consumers of sweet products.

Four RCT studies focusing on the relationship between IS consumption and body composition were identified. In three of these studies, changes in weight and BMI did not differ between IS consumers and non-consumers [33‐35]. These studies focused on overweight or obese children and had methodological limitations. The fourth study, of good methodological quality, examined the effects of consuming 250 mL/day of an artificially sweetened drink, compared to the same amount of sugar-sweetened drink, in 641 normal-weight children (aged four to 12 years), who were regular consumers of sugar-sweetened drinks, for 18 months [36]. This study showed a significant decrease in the BMI z-score (the most relevant criterion to assess changes in corpulence in growing children) in the group that consumed artificially sweetened drinks. The change in body weight between the two groups differed by 1 kg on average. Of the seven prospective epidemiological studies in children, five [37, 38] observed a positive relationship between IS consumption (primarily in beverage form) and weight over time, while two [37] did not find any relationship. To explain these findings, the authors of these studies assumed that subjects 'at risk for weight gain' or with less healthy food profiles were those who consumed the most IS in order to reduce their energy intake.

Most of the prospective observational studies undertaken in children show that IS use is paradoxically associated with weight gain, although the causality of this relationship has not been established. The four available controlled trials showed conflicting results but none reported weight gain. No conclusions can be drawn from all of these studies as to the significance of IS for weight management in children and adolescents.

Regarding the acute effects (i.e. less than 24 hrs.), the available studies did not show any effects related to the consumption of aspartame on an empty stomach [39‐44], saccharine [45] or sucralose [46, 47] on blood glucose and insulin levels. Other studies assessed acute effects of IS on post-prandial glycaemic parameters after a test meal [48‐52]. These studies generally showed that consuming IS before a test meal did not modify post-prandial glycaemic and insulin responses compared to a placebo, and reduced these responses compared to a sucrose preload. These effects were reported irrespective of the tested IS (aspartame, stevia extract, sucralose, beverage containing acesulfame K and sucralose). It should also be noted that the parameters of these studies were highly variable with differences in the composition of test meals, the time between the preload and the meal, the studied subject groups (age, sex, healthy overweight or obese subjects) and the preload form (solid or liquid). Several studies also showed that consuming IS before a meal resulted in increased secretion of GLP1 (Glucagon-like peptide), a gastro-intestinal hormone that usually increases insulin secretion, slows down gastric emptying and reduces glucagon secretion [53]. This increase in GLP1 may be induced by IS activating sweetness receptors, as suggested by data in rats [54].

When considering short- and medium-term effects, several studies assessed the effect of regular IS intake (one to three times per day, for a few days to several weeks), in capsule form or in beverages, on maintaining blood sugar control (glucose and insulin concentrations measured after a night of fasting, glycated haemoglobin HbA1c). For type 2 diabetics, the consumption of sucralose [55] or aspartame [56‐58] for periods of up to 18 weeks did not change fasting glucose levels) compared to sucrose or a placebo. Furthermore, fasting plasma glucose and insulin, and HOMA-IR (insulin sensitivity calculated from the HOMA index) were not modified in non-diabetic obese subjects who had consumed a beverage sweetened with aspartame for six months compared to groups who had consumed sugar-sweetened drinks, water or milk [21]. Other studies covering unspecified IS compared to sucrose in obese or overweight subjects [59] confirm these results. Regarding stevia extracts, the data show either a lack of effect on glucose control in healthy [59] or diabetic [59] subjects or a slight significant decrease in blood glucose levels in healthy subjects [59] or hypertensive subjects [60].

Overall, the vast majority of studies do not show any acute effects of IS intake on blood glucose or insulin concentrations measured on an empty stomach or after a test meal, in healthy subjects or in diabetics. Some studies reported a modest increase in GLP-1 secretion, but with no repercussions on insulin secretion or blood glucose concentrations. IS consumption has no effect on short- and medium-term blood glucose parameters in healthy subjects or in diabetics.

The seven observational studies dealing with IS consumption and the incidence of T2D showed diverging results. Four cohort studies of good quality (three undertaken in North American populations and one in a European population), over periods of nine to 24 years, did not show any relationship between the consumption of artificially sweetened beverages and the risk of developing T2D after adjustment for BMI and energy intake of subjects [61‐63, 26]. Three other cohort studies suggested a positive association between the consumption of artificially sweetened beverages and the incidence of T2D [64‐66]. Among them a French study, [64], showed that the incidence of T2D was significantly higher (HR (95 % CI) 2.21 [1.56-3.14]) in the group of women consuming the largest amounts of artificially sweetened beverages (over 600 mL per week) who were monitored for 14 years, with a linear and dose-dependent relationship. The second study, undertaken in the United States for seven years, reported an increase in the incidence of T2D in subjects consuming more than one artificially sweetened beverage per day in a model with adjustment for the primary confounding factors (HR (95 % CI) 1.67 [1.27–2.20]. The third study, which reported an increased incidence of T2D in subjects consuming more than one artificially sweetened beverage per week (HR (95 % CI) 1.70 [1.13–2.55], focused on a limited-sized Japanese population not representative of the general population, monitored for seven years. It is important to underline the heterogeneity of these data, particularly in terms of the characteristics of the populations and the monitoring periods (from seven to 24 years). Furthermore, in these studies, the consumption of artificially sweetened beverages was recorded when the subjects were first included, often through self-administered frequency questionnaires, with no updating of dietary data over time.

In conclusion, the long-term epidemiological studies on the risk of developing T2D show heterogeneous results, but the most robust studies do not report any effects.

Of the 20 randomised controlled experimental studies analysed, the majority focused on aspartame or stevia extracts. Compared to a placebo, aspartame consumption had no effects on triglycerides or cholesterol concentrations (either total, HDL, LDL or VLDL cholesterol) in various populations (healthy, T2D and overweight subjects) for periods ranging from 13 to 28 weeks. Compared to a caloric sweetener (sucrose, glucose or fructose), of the five identified studies, two showed a modest significant improvement in lipid profile (TG and/or total cholesterol) in the group that received aspartame, still with no differences compared to the placebo. Three studies, two in T2D patients, assessed the effect of stevia extracts on lipid parameters and showed no differences compared to a placebo. Studies using other types of IS (cyclamate, sucralose, IS mixture or unspecified IS) also showed no effects on the assessed lipid parameters. Of the four identified cohort studies, most showed no effects on lipid parameters related to the consumption of artificially sweetened beverages. A single study reported a positive association between the consumption of these beverages and an increase in TG concentrations associated with a lowering of HDLc [67].

In conclusion, the majority of observational studies showed no effects on lipid profile related to IS. Two studies reported that replacing sugars with aspartame reduced plasma triglyceride concentrations but the data are too limited to conclude that IS have a beneficial effect on lipid profile.

Two epidemiological studies are available. In the first [68], a dose-effect relationship was observed, which meant that the risk of pre-term delivery was higher in the heaviest consumers of artificially sweetened beverages (OR (95 % CI) 1.38 [1.15-1.65] for ≥ 1 serving of artificially sweetened carbonated soft drinks/d). In addition to this Danish study, another study, with a similar methodology and including over 60,000 pregnant women, suggested that the consumption of artificially sweetened beverages and sugar-sweetened beverages was associated with an increased risk of spontaneous or induced pre-term delivery (OR (95 % CI) 1.11 [1.00-1.24] for ≥ 1 serving of artificially sweetened soft drinks/d). However, although the association was stronger for sugar-sweetened beverages, the authors concluded that they could not determine whether this risk was caused by the effects of these beverages or by other associated dietary or socio-economic factors [69].

Based on the available data, it is not possible to identify any benefits or draw any conclusions regarding the risk related to the consumption of intense sweeteners during pregnancy, in terms of maternal health, obstetrical parameters or newborn health.

The relationship between IS consumption and cancer in humans was assessed in 55 scientific studies. Thirty-nine of these studies involved the urinary tract and 32 focused exclusively on bladder cancer. The other studies assessed the relationship between IS consumption and the risk of brain cancer (four studies), digestive system cancer (six studies) or other cancers (five studies). Except in the studies focusing on bladder cancer, the IS in question were not identified by the authors. The relationship between saccharine consumption and bladder cancer was the most commonly studied, given that data were available in rodents [70]. The results of studies in humans are conflicting. Based on the analysis of data in humans, it is not possible to determine a relationship (whether for saccharine or for the other studied IS), since the studies did not adjust their results for major confounding factors such as exposure to chemical pollutants. Regarding kidney, brain, digestive system and breast cancers, the data are more limited and do not show any relationship with IS consumption. A recent cohort study examining the risk of lymphoma and leukaemia suggested an increased risk of non-Hodgkin lymphomas and multiple myelomas in males consuming more than one serving (355 mL) per day of artificially sweetened beverages and in the heaviest consumers of aspartame (as a table-top sweetener and in beverages) compared to non-consumers [71]. No significant association was reported in women. The authors specified that due to the differences in the results by sex, the results should be interpreted with caution. Moreover, this study did not take into account exposure to chemical pollutants as a confounding factor. However, it is worth noting that this study attempted to take into account, in its statistical analysis, changes in the individual consumption of artificial sweeteners over time, although little information is available regarding the methodology.

On the whole, the epidemiological studies do not show any effects of IS consumption on cancer risk. Only one recent study suggested a relationship between the consumption of beverages containing IS and the occurrence of non-Hodgkin lymphomas and myelomas, and additional studies are required.

Regarding the potential neurological effects of IS, only aspartame has been studied. There are two studies in healthy adults [72, 73]. No effects of aspartame on the measured parameters (reaction time, headaches, hunger, sedation, electroencephalographic parameters) were observed. The study undertaken in epileptic subjects [74] showed no statistically significant difference between aspartame and the placebo on the incidence of epileptic seizures. The four available studies on migraine subjects [75‐78] show conflicting results. However, no conclusion can be drawn due to their poor methodological quality (no adjustment) and the subjective nature of the measured effects (using non-validated self-questionnaires). Regarding children, there are two studies, one in epileptic children [79] and the other in hyperactive children [80], showing no significant effects of aspartame.

Some studies with significant methodological limitations suggested that aspartame consumption may be involved in triggering epileptic seizures and migraines but no conclusions can be drawn regarding the occurrence of such a risk from the data as a whole.