ATLANTA (March 13, 2015) — The recent study in Nature by Suez et al. (2014) led to another round of warnings about the use of non-caloric sweeteners based on the authors’ conclusions that “artificial sweeteners induce glucose intolerance by altering the gut microbiota” and that their “findings suggest that NAS may have directly contributed to enhancing the exact epidemic (obesity) that they themselves were intended to fight.”
However, a careful reading of the details of this paper illustrates that this conclusion is inappropriate in a number of ways, is not substantiated by the data generated from the authors’ studies, and is in direct conflict with extensive previous human clinical studies on artificial sweeteners.
First, the authors assume that their results based on saccharin alone can be used to make assertions about all types of artificial sweeteners; in both the title of the article and the discussion, the authors make claims about “artificial sweeteners” in general, despite the fact that the study’s findings pertain solely to saccharin. Only one experiment included aspartame and sucralose, as tabletop (not pure) formulations added to drinking water (Figures 1-a and 1-b and extended Figure 3) in doses several hundred times typical human consumption levels. The misrepresentation of data in these figures is discussed below. It is very important to note that there was not one experiment that actually measured the effect of aspartame or sucralose on gut microflora.
Second, the authors did not consider differences in chemistry and metabolism of different artificial sweeteners – another important reason for why making claims about all artificial sweeteners based on primarily on saccharin data is inappropriate. For instance, there is no biological plausibility based on previously published data that aspartame would affect microbiota. Aspartame is completely digested into amino acids and methanol, which are absorbed in the small intestine. Neither aspartame nor its digestion products ever reach the colon; thus aspartame itself cannot affect gut microbiota (EFSA, 2013).
As a large percentage of sucralose is not absorbed and enters the colon, there is biological possibility that sucralose could affect the microbiome. Previous metabolic studies in both rodents and humans reported that all sucralose in feces is unchanged, indicating that sucralose is not a substrate for colonic microflora (Roberts et al., 2000; Sims et al., 2000; Johnet al., 2000). Although this does not rule out the potential for sucralose to alter the composition of the microflora through some other mechanism, the authors have provided no data on microbiome changes in sucralose-treated mice.
In contrast, the observation that saccharin at high doses alters gut microbiota was known in the 80s, and contributed to the establishment of the ADI for saccharin. For example, WHO (1993) summarizes the many studies that followed observations of enlargement of the cecum resulting from feeding high dietary concentrations (5% w/w) of saccharin to rats, accompanied by an increase in the total numbers of microorganisms (Anderson and Kirkland, 1980; Mallett et al. 1985).
Third, there are numerous cases of inappropriate methods and ambiguous, misleading presentation of statistical data.
In the first animal experiment, the authors inappropriately lump all 3 sweetener groups and all 3 different control groups into just two groups in order to show a statistically significant difference in glycemic response; however, a closer examination of the data shows that most of the difference in the sweetener group comes from saccharin alone; sucralose had 3-4 outliers and the range of responses for aspartame-treated mice was within the range of those that drank only water. Stating that “all three NAS-consuming mouse groups developed marked glucose intolerance” is misleading.
When giving a dietary test compound in water, it is critical to know both liquid intake and food consumption over the course of the study. Both are simple to measure, yet in this study food and liquid intake is reported for only 4 of the 20 mice per group and for only 3 days of the 11-week study, found in “Extended data”! Nonetheless, in this short window for this small subset of animals, a dramatic effect on intake of mouse chow (drop of up to 50% in some groups) is obvious from the figures. Mouse chow contains fiber, protein, fat, fermentable carbohydrates and a host of other components that have repeatedly been shown to affect both gut microbiota and glycemic indices. Clearly, these dramatic changes in diet would result in changes in microbiota, and glycemic responses. The authors dismiss these findings rather than addressing them. Dietary factors known to affect human microflora (Fava et al., 2013) were similarly not considered in the human studies.
The authors suggest evidence for a causative relationship between artificial sweetener consumption and blood glucose control in humans based on two experiments. The first is a cohort study and the second is clinical study with saccharin given to 7 participants for 6 days.
The cohort study is missing information on level and type of sweeteners consumed as well as other diet information. The authors did not define what classifies individuals as “high consumers,” explain why they excluded 105 individuals out of the 381 for which they had data, or control for any other factors besides BMI that may be associated with glucose concentration (such as other components of the diet). The range of observations of HbA1C% in the “high consumers” is within the range of the non-consumers – in fact many non-consumers have much higher levels than high consumers. Lastly, there is no explanation as to why a nonparametric rank correlation was used when all measurements were parametric.
Finally, the clinical study provides more examples where the authors inappropriately group data together based on results rather than their experimental design to produce statistically significant outcomes. In presenting data for the 7-person experiment, instead of conducting and comparing a multi-day baseline period without treatment to the treatment period, the authors arbitrarily make comparisons between a grouping of day 1 (without saccharin) plus treatment days 2-4 (with saccharin) versus a second grouping of treatment days 5-7. The only apparent reason seems to be that glycemic response rose on day 5 for 3 individuals, allowing for a visually dramatic comparison. The division of the 7 participants into “responders” and “non-responders” is equally arbitrary. The authors state that 4/7 had poorer glycemic response following saccharin consumption (deemed the “responders”), but there is no reason for this to necessarily be due to saccharin. Considering that the experiment had no control group, had no multi-day period of baseline measurements, and did not control for an other components of the diet, there is no way to know if the variability in glycemic response over the 7 day period is any different than what the control group would have experienced or if it is instead attributable to other components of the participants’ diets.
Most importantly, the authors did not consider extensive data from previously published literature that has addressed sweeteners and glycemic response.
There are numerous studies and consistent evidence in the literature that chronic administration of aspartame and sucralose do not affect blood glucose or response to a glucose load. For aspartame, 4 clinical studies with diabetics have been conducted for periods ranging from 2 to 18 weeks (Colagiuri et al., 1989; Stern et al., 1976; Nehrling et al., 1985; Okumoet al., 1986), consistently showing no effect on glycemic response.
Long-term clinical studies on sucralose have also documented no effect on glycemic response. Grotz et al. (2003) investigated the effect of consumption of sucralose for 3 months in subjects with type 2-diabetes and reported no effect on fasting plasma glucose, serum C-peptide or HbA1c levels. The Scientific Committee on Food (2000) reviewed 5 clinical studies on the effect of sucralose on glucose homeostasis in diabetic and non-diabetic human volunteers during their evaluation. In a 12-week study in normoglycemic subjects, consumption of 1 g sucralose/day had no effect on OGTT response. Seventy-seven participants consumed sucralose for 13 weeks with no significance changes in biochemical parameters (Baird et al., 2000).
The conclusions that can be made regarding saccharin are not as clear. Previous studies that were not cited by Suez et al. (2014) provide evidence that saccharin can alter the gut microbiome in animal studies at high doses (WHO, 1993). In addition, a sweetener consisting of saccharin and neohesperidin dihydrochalcone has been approved for use in animal feed for improvement of animal feed efficiency and performance. The mechanism of improved feed performance may involve both the gut microflora and glucose uptake, although the enhancement of glucose uptake has not been directly linked to the change in microflora. Dalyet al (2014) reported that the addition of the sweetener combination to feed of piglets (0.015% w/w) dramatically increased the caecal population abundance of Lactobacillus,providing a probiotic effect; however, the mechanism of this effect is unknown.
The WHO report supports the safety of saccharin for diabetics, citing evidence of a lack of effect on glycemic response. For example, “No deleterious effects on blood sugar, kidney function, vitamin utilization, blood coagulation or enzyme activity were detected in man (NAS-NRC, 1955). Diabetic patients have received as much as 4.8 g daily for 5 months without adverse effect (Neumann, 1926a,b) and 0.4-0.5 g/day for 15-24 years without any adverse effects (NAS-NRC, 1955). These early clinical studies on saccharin were not available for review of specific details.
The associations between artificial sweetener consumption and glucose intolerance in humans in the Nature paper are based on poorly defined cohorts, and the 7 subject experiment with saccharin is without a control group or baseline data; in both cases, the authors fail to control for important variables such as other components of the individuals’ diets. Thus, there is insufficient data to clearly determine whether saccharin alters the microbiome in humans and if so, at what the dose and in what manner. The current study by Suez et al., (2014) does not provide clear answers to these questions due to the limitations discussed above.
In conclusion, there are no data in the report by Suez et al., to support the claim that the commonly consumed sweeteners, aspartame and sucralose, directly alter the gut microbiome or glycemic response.
In contrast, there is ample evidence from the literature that these non-caloric sweeteners do not have an adverse effect on glycemic response at typical use levels in humans.
Although it has been reported that high doses of saccharin can affect gut microbiota in animals, the hypothesis that saccharin is contributing to the obesity and diabetes epidemic is problematic as saccharin is rarely used in those countries with the highest incidence of these diseases such as the United States.