Artificial sweeteners linked to obesity–poor evidence

artificial-sweetenersThis article was written by Linda Tock, an American living in Denmark, who has an extensive research background in the biomedical sciences. She has a Master’s Degree in Environmental Chemistry and Health, and will be pursuing a Ph.D. Ms. Tock has a fascination for Daphnia, an interesting planktonic crustacean, that is an important organism in studying pollution and environmental stresses. She’s also a passionate Boston Red Sox fan, so she had to endure a difficult year.

So I received a message from a friend of mine, wanting my opinion on this news article, which loudly proclaims that artificial sweeteners are linked to obesity. Because it was a genuine question regarding the science behind the study, and not a ‘concern troll’ about my preference for diet cola, I went and looked at the study itself to see what the fuss was about.

According to the abstract in the article published in the journal Nature,

Non-caloric artificial sweeteners (NAS) are among the most widely used food additives worldwide, regularly consumed by lean and obese individuals alike. NAS consumption is considered safe and beneficial owing to their low caloric content, yet supporting scientific data remain sparse and controversial. Here we demonstrate that consumption of commonly used NAS formulations drives the development of glucose intolerance through induction of compositional and functional alterations to the intestinal microbiota. These NAS-mediated deleterious metabolic effects are abrogated by antibiotic treatment, and are fully transferrable to germ-free mice upon faecal transplantation of microbiota configurations from NAS-consuming mice, or of microbiota anaerobically incubated in the presence of NAS. We identify NAS-altered microbial metabolic pathways that are linked to host susceptibility to metabolic disease, and demonstrate similar NAS-induced dysbiosis and glucose intolerance in healthy human subjects. Collectively, our results link NAS consumption, dysbiosis and metabolic abnormalities, thereby calling for a reassessment of massive NAS usage.

In this study, NAS is saccharin, sucralose and aspartame, and the researchers claimed that consumption of these sweeteners cause GI (glucose intolerance) by inducing a change in the intestinal microbiota. Those changes in biota were canceled by antibiotic treatment (ie: they killed off the gut biota deliberately), and then they performed a fecal transplant of the feces from the NAS fed mice to ‘naive’ (or unexposed) mice, and those mice (with the new biota) had the same symptoms.


This is their first figure. It shows the blood glucose level under a standard 2 hour glucose tolerance test.

All three artificial sweeteners showed a significant (indicated by ***) increase in blood glucose levels initially, and through t = 90 minutes. However, by 120 minutes, the differences between the levels was no longer significant. (As an aside, that’s really a terrible graph, and it’s difficult to read at the size shown. I understand they have limited space though – it’d have been nice to be able to blow out the individual graphs to a larger size so we could accurately determine the distance between the treatment groups by time.)

To determine the effects of NAS on glucose homeostasis, we added commercial formulations of saccharin, sucralose or aspartame to the drinking water of lean 10-week-old C57Bl/6 mice (Extended Data Fig. 1a). Since all three commercial NAS comprise ~5% sweetener and ~95% glucose, we used as controls mice drinking only water or water supplemented with either glucose or sucrose.

Unfortunately, this doesn’t tell us the actual amount of sweeteners administered to the mice, so we don’t know how the dose compares to that found in food or during ‘typical’ use.

Here’s where things get interesting:

As saccharin exerted the most pronounced effect, we further studied its role as a prototypical artificial sweetener. To corroborate the findings in the obesity setup, we fed C57Bl/6 mice a high-fat diet (HFD, 60% kcal from fat) while consuming either commercial saccharin or pure glucose as a control (Extended Data Fig. 1b).

I would agree with their statement that saccharin exerted the most pronounced effect during the glucose tolerance test – it had the greatest AUC (Area under the curve) – however at 2 hours post test, the differences between the groups were shown to not be significant.



Portions C and D are the results of saccharin plus a HFD (high fat diet)–after a 2 hour glucose test, the results are still significantly higher than the controls, an indication that glucose levels are higher in the treatment rats on a HFD. They delved further into the effects of the NAS in the paper..

Here are some of my more troubling concerns about the research

  1. While there was an initial increase blood glucose in the saccharin, sucralose and aspartame groups at t=120 (the standard time frame for a glucose challenge), there was no significant difference between the NAS and controls. Did those mice ‘fail’ the challenge or not? Were their 2h values unacceptably high? Because there’s no * to indicate a significant difference between the groups, I’m left wondering.
  2. Neither sucralose nor aspartame were tested further – not in combination with a HFD – so how can they make any comments regarding their effects when combined with diet?
  3. Images E and F are after fecal transplant from mice fed saccharin – either ‘pure’ or ‘commercial’. Again, at t=2h, there’s no significant difference to be seen between the saccharin mice and the water mice (control group).
  4. Bizarrely enough, they have this figure:


Extended Data Figure 5: Glucose Intolerant NAS-drinking mice display normal insulin levels and tolerance.

Fasting plasma insulin measured after 11 weeks of commercial NAS or controls (N = 10). b, Same as a, but measured after 5 weeks of HFD and pure saccharin or water (N = 20). c, Insulin tolerance test performed after 12 weeks of commercial NAS or controls (N = 10). Horizontal lines (a, b) or symbols (c) represent means; error bars, s.e.m. All measurements were performed on two independent cohorts.

That makes no sense. If the mice have normal insulin levels and tolerance, why are the authors calling them glucose intolerant? The authors then write:

Metabolic profiling of normal-chow- or HFD-fed mice in metabolic cages, including liquids and chow consumption, oxygen consumption, walking distance and energy expenditure, showed similar measures between NAS- and control-drinking mice (Extended Data Fig. 3 and 4). Fasting serum insulin levels and insulin tolerance were also similar in all mouse groups consuming NAS or caloric sweeteners, in both the normal-chow and HFD settings (Extended Data Fig. 5). Taken together, these results suggest that NAS promote metabolic derangements in a range of formulations, doses, mouse strains and diets paralleling human conditions, in both the lean and the obese state.

Given that they only tested saccharin in the ‘normal chow’ and ‘HFD-fed’ mice, I fail to see how they can extrapolate those results to all NAS for both ‘normal chow’ and ‘HFD’ mice. Also, if both groups had similar metabolic profiles, similar fasting serum insulin levels and tolerances–how can they claim that NAS promotes metabolic derangements without saying that conventional sweeteners do as well?

They delve further into the saccharin connection – but then compare it to “High NAS consumers”–which NAS are they consuming? Their own data suggests that, of the three NAS tested, saccharin is the most problematic – but now they’re going to lump them all together? Again, they link to graphs of fasting glucose tests, and yes the NAS groups do have a higher AUC during the test, but their END LEVELS are similar to that of the caloric sweetner.

They do a number of investigations for metabolic pathways – but ONLY for saccharin. There’s no evidence to support their claim that all NAS causes the metabolic issues that they’ve linked to in their paper.

Then we have to ask ourselves – where are the NAS at in our diet? I freely admit to drinking diet soda–which contains aspartame. I have no intake of sucralose nor saccharin. Other than coffee drinkers, I’m trying to imagine who actually uses saccharin daily. I don’t consume ‘low-cal’ foods (jams, etc) – I eat the ‘regular’ ones – but in smaller quantities and less often than I used to.

The authors used the Acceptable Daily Intake (ADI) for these artificial sweeteners in their study–but how many people actually hit that ceiling? For example, what’s the ADI for saccharin?

The following commercially available NAS were dissolved in mice drinking water to obtain a 10% solution: Sucrazit (5% saccharin, 95% glucose), Sucralite (5% Sucralose), Sweet’n Low Gold (4% Aspartame). 10% glucose (J. T. Baker) and 10% sucrose (Sigma Aldrich) solutions were used for controls. The administered doses of 10% commercial NAS dissolved in water were well below their reported toxic dose (6.3 g per kg (body weight)27, 16 g per kg (body weight)28, and 4 g per kg (body weight)29, for saccharin, sucralose and aspartame, respectively). For experiments conducted with pure saccharin (Sigma Aldrich) a 0.1 mg ml−1 solution was used in order to meet with FDA defined ADI for saccharin in humans (5 mg per kg (body weight).

Ok the ADI for aspartame in the US is 50 mg/kg of bodyweight (40 if you’re in the EU). What does that mean?

To put the ADI for aspartame in perspective, this would be 3,750 milligrams per day for a typical adult weighing 75 kilograms (about 165 pounds), far more than most adults take in daily. A 12 ounce can of diet soda usually contains about 192 milligrams of aspartame and a packet of the tabletop sweetener contains about 35mg. An adult weighing 165 pounds would have to drink more than 19 cans of diet soda a day or consume more than 107 packets to go over the recommended level.


If I did the math right previously, an average person have to drink about 8 liters of Coke Zero per day to hit the limit (in the EU). Personally, I don’t drink anywhere NEAR that amount, nor do I know any who does. So what would my risk be from the aspartame compared to the risk from my normal diet and meals.

Just some food for thought.

Editor’s note: based on how quickly stories about this study spread across the internet, one would think that this research was profound, almost Nobel Prize worth. But, once again, one must properly weight the quality of science that makes a new claim, especially one that’s so critical. This is a primary study on mice, the value of which is limited. It needs to be repeated, there has to be some sort of plausibility, and eventually someone has to show that there’s actual clinical significance in humans. Finally, if one were to make a basic criticism of this research is that some of the groups had n=4 population size. There’s almost no way to see small changes with such small numbers.  

Key citations:

The Original Skeptical Raptor
Chief Executive Officer at SkepticalRaptor
Lifetime lover of science, especially biomedical research. Spent years in academics, business development, research, and traveling the world shilling for Big Pharma. I love sports, mostly college basketball and football, hockey, and baseball. I enjoy great food and intelligent conversation. And a delicious morning coffee!