#431: Artificial Sweeteners – Health Impacts and ‘Safe’ Levels

1. Detailed Study Notes
2. Transcript

You are currently not signed-in as a Premium subscriber.

To view our Premium content, please log-in to your account or subscribe to Premium.

Danny’s Key Ideas from this episode are:

1. What does “safe levels” really mean?
2. How to critically address statements of: “But here’s a study showing harm…”

#1: What does “safe levels” really mean?

Prior to approval, several toxicokinetics studies are carried out. Toxicokinetics is essentially looking at the fate of the sweeteners once ingested, including absorption, distribution, metabolism, and excretion (ADME).

Acceptable Daily Intake (ADI) for any food additive set for use in the EU by EFSA and in the U.S. by the FDA, through a process involving submission of both scientific safety evaluation and technical data.

NOAEL (“No Observable Adverse Effect Level”) = highest level at which no negative sides effects have been observed in safety studies/animal toxicology studies.

ADI = established by dividing the NOAEL by an “uncertainty factor” of 100.

From Magnuson et al., 2016:

“The ADI is an important and often misinterpreted value. The ADI is defined as the amount of a food additive, expressed on a body weight basis, that can be consumed daily over a lifetime without appreciable health risk.

The ADI is not a threshold between safe and unsafe; rather, it is a calculated value, derived by dividing the NOAEL observed in toxicology studies by a safety factor. The NOAEL is the daily amount consumed in long-term, repeated-dose studies that was shown to have no adverse effects in the animals; in other words, it is a daily intake level that is too low to cause any biological effects.

The safety factor is established by regulatory agencies and convention to ensure protection of the most susceptible and sensitive individuals in an entire population, including children and pregnant women.
Often, the safety factor used is 100, resulting in the ADI being set at a level 100 times lower than the NOAEL, ensuring a wide margin of safety. For example, if the amount shown in animal studies to have no effect when consumed daily for the majority of the animal’s lifetime was 4000 mg per kilogram of body weight, the NOAEL would be 4000 mg/kg/d and, with a 100-fold safety factor, the ADI would be 40 mg/kg/d. Thus there is a 100-fold reduction from the amount shown to have no effect to the established ADI.

This is a much greater safety factor than exists for most nutrients and naturally occurring food components. Therefore, the ADI is a level of daily intake considered safe for everyone, including those with the highest potential exposure to an ingredient.”

#2: “But here’s a study showing harm…”

A line between accurate scientific appraisal of topic, versus inaccurate or pseudoscience, is not that the latter presents no science, it’s the type of evidence they use to support conclusions. And in nearly all cases of pseudoscience or outright quackery, one of the key features is making a bold claim about an impact on human health, and then supporting that with mechanistic research or animal data.

As we’ve mentioned before, both have their place, but health science is littered with examples of various compounds that have been found to impact a certain pathway, and so therefore it’s biologically plausible that they could affect health outcomes, yet when tested in a human trial, there is no observed impact on the health outcomes we care about. So the compound impacts a certain pathway or enzyme or acute levels of a hormone, etc., yet it is meaningless in the context of actual disease risk or other outcomes of interest.

So when we say there is no good evidence to think that NNS will harm health, that is not saying there are no studies that suggest concern or harms. Because people who advocate for avoiding sweeteners at all costs, can readily point to certain research papers. But when you look at these, they are nearly all mechanistic in nature, or at the very best, pilot trials or hypothesis generating studies.

For example, someone may present an animal toxicology study. But as Alan highlighted in this episode, the intent of toxicology studies in animals is to induce cancer (or other health effects). So if an animal toxicology study shows cancer as a result of exposure, that was likely the INTENT of the study.

So the conclusion that there is no good evidence that NNS harm health, is not to say there may never be any harms found, but what it is to say is that: based on all the current evidence, from safety data, to setting ADIs, to human outcome trials, there is currently no reason to suspect that consuming products with NNS in any reasonable amount, leads to any serious health risk. And claiming that there is harm, is an incredibly strong claim without anywhere near the data to conclude that.

You are currently not signed-in as a Premium subscriber.

What are artificial sweeteners (AS)?

A variety of names are used when discussing different types of sweeteners (e.g. non-nutritive sweeteners; non-caloric sweeteners; non-sugar sweeteners; zero calorie sweeteners; low-calorie sweeteners; artificial sweeteners; high intensity sweeteners)

First, sweeteners can be distinguished as nutritive or non-nutritive:

1. Nutritive – sugars, modified sugars, sugar alcohols, natural caloric sweeteners
   
2. Non-nutritive sweeteners – artificial sweeteners, natural non-caloric sweeteners
   • acesulfame-K, aspartame, saccharin, and sucralose are main NNS currently in use

LNCSs = low- and no-calorie sweeteners

Are all sweeteners made the same?

• There are a number of AS approved by FDA and EFSA
• Each compound is structurally unique, meaning the individual compounds all vary in sweetness potency, duration of sweetness, aftertaste, and mouth feel. Also have different pharmacokinetic profiles, relevant to discussion re: metabolism, etc.

They are commonly labelled by their “E” number; e.g. E 954 for ‘Saccharin’

• E numbers are typically seen as ‘unhealthy’ or ‘harmful’ by consumers.
• Essentially the E number means they have passed safety tests and are approved for use in the EU – E standing for Europe.
• Many are beneficial, E300 is vitamin C.
• The rules were developed to regulate so that dangerous substances could be banned from being added to food.

They are several hundred times sweeter than sugar (200-600 times depending on the specific sweetener):

“… aspartame has 200 times the sweetening potency of sucrose, meaning that when solutions of sucrose and aspartame are compared, the same sweetness associated with a sucrose solution will be associated with an aspartame concentration 200 times lower than the concentration of sucrose. Therefore, very little of an intense sweetener is actually present in the “diet” food or beverage. In most cases, the maximum sweetness levels that can be achieved with intense LNCSs is less than what can be achieved with sucrose due to other “off tastes” from the LNCS, such as bitterness or metallic tastes.”

Magnuson et al., 2016

Why use them? Uses & potential benefits

Using non-nutritive sweeteners allows for products to obtain a sweet taste, without providing sugar and calories (or at least less than usual). Therefore, potentially a simple method to reduce calorie & sugar intake, without much change in dietary choices.

“From a dietetic point of view, having options which allow patients and (or) clients to alter their calorie intake without making significant dietary changes is seen as a favourably viable option. However, a whole diet approach which focuses on overall diet quality rather than specific ingredients and nutrients is important.” – Expert Consensus – Ashwell et al., 2020

So therefore it is plausible that there could be benefits for weight loss, weight loss maintenance, diabetes management, blood sugar control, et cetera. Actual outcome data for each will be discussed later.

An interesting observation is that many people who are successfully maintaining long-term weight loss seem to have a high consumpion of artificially sweetened beverages. For example, Phelan et al. (2009) published a paper in the International Journal of Obesity showing that:

• Compared to an always-normal weight group, weight loss maintainers reported consuming:
  • 3x more daily servings of artificially sweetened soft drinks (0.91 vs 0.37; P=0.003)
  • significantly fewer daily servings of sugar-sweetened soft drinks (0.07 vs 0.16; P=0.03)
  • more daily servings of water (4.72 vs 3.48; P=0.002)

Regulation: How are “safe levels” determined?

Like any food additive, artificial sweeteners have an Acceptable Daily Intake set; by EFSA in the EU and by the FDA in the US.

Magnuson et al., 2016 review: “Concerns about safety often deter the use of low- and no-calorie sweeteners (LNCSs) as a tool in helping control caloric intake, even though the safety of LNCS use has been affirmed by regulatory agencies worldwide. In many cases, an understanding of the biological fate of the different LNSCs can help health professionals to address safety concerns.”

Prior to approval, several toxicokinetics studies are carried out. Toxicokinetics is essentially looking at the fate of the sweeteners once ingested, including:

• absorption
• distribution
• metabolism
• excretion

Acceptable Daily Intake (ADI) for any food additive set for use in the EU by EFSA and in the U.S. by the FDA, through a process involving submission of both scientific safety evaluation and technical data.

Important terms:

• Scientific safety evaluation/data
  • Must include full range of studies on safety, including the anticipated daily intake in the population from all dietary sources, within different ages groups
  • Derived from animal toxicology studies investigating absolute toxicity thresholds and also sub-chronic long-term toxicity for potential effects on reproduction, development, carcinogenicity, genotoxicity, and immunotoxicity.

• Technical data
  • Chemical composition of the compound, its source and manufacturing methods, its stability across a range of food matrices, sensory properties, and anticipated intake by reference to the concentrations of the compound in food and beverages together with the quantity of those foods/drinks typically consumed in the population.

• “Concern level”
  • Dictates evidential criteria required from toxicology studies
  • AS/NNS are considered a “high concern” level duet to:
    1. potential high exposure in the population, and
    2. toxicity potential

• NOAEL (“No Observable Adverse Effect Level”)
  • Highest level at which no negative sides effects have been observed in safety studies/animal toxicology studies.
  • “The NOAEL is the daily amount consumed in long-term, repeated-dose studies that was shown to have no adverse effects in the animals; in other words, it is a daily intake level that is too low to cause any biological effects.” – Magnuson et al., 2016

• ADI (Acceptable Daily Intake)
  • The ADI is defined as the amount of a food additive, expressed on a body weight basis, that can be consumed daily over a lifetime without appreciable health risk.
  • Established by dividing the NOAEL by an “uncertainty factor” or “safety factor”, often set at 100.
  • The safety factor is established by regulatory agencies and convention to ensure protection of the most susceptible and sensitive individuals in an entire population, including children and pregnant women.
  • The average and 95th percentile intakes of acesulfame K, aspartame, and saccharin are below the relevant ADIs (Mortensen, 2006).

Example: As an example of the safety margin that an ADI offers, consider that the ADI in Europe is 40 mg/kg (and 50 mg/kg in the US). And the estimated average intake in the population is about 4 mg/kg. So in other words, the average intake is one-tenth of the ADI and therefore 1/1000th of the NOEL. Remember, the NOAEL is the highest amount studied in long-term, repeat dose studies, at which there are zero side effects. And then the ADI is 100 times lower than this. And the average aspartame intake is 10 times lower than even this ADI!

For further context, imagine that a can of diet soda contained the maxmium allowed amount of aspartame. If a person weighed 60 kg (132 lbs), they would have to drink more than 12 x (330ml) cans every day for the rest of their life to exceed the ADI. And given that aspartame is usually in such drinks at 3-6 times lower than the maximum allowed amount, it would mean that in that situation the person would have to drink more than 36 cans every day.

Do they cause cancer?

One of the primary claims around potential harms is cancer risk. And this is rooted in a combination of observational and animal trials. There are also differences in how detrimental different AS are deemed to be; with aspartame seemingly getting most negative attention, at least in “wellness” and health guru circles.

Cyclamate

• Cyclamate effectively banned 1970 by FDA based on animal data – caused cancer in rats.
  • First, the daily limit of 780 mg lowered to 168 mg.
  • Then in October 1970, the FDA banned cyclamate completely from all food and drug products in the United States.
• In the late 1960s cyclamate was banned in the UK but was approved after being re-evaluated by the European Union in 1996.
• Cyclamate remains banned in the United States and South Korea.

Aspartame

• Aspartame is commonly used in diet soda (diet coke for example)
• 1996: Initial concerns that it might increase risk of brain tumours; based on early 90’s trends in aspartame use in diet drinks and increase incidence of brain tumours in the US.
• Another animal study was conducted in mice who were exposed to aspartame (Soffritti et al., 2010). EFSA largely dismissed this finding (in their 2013 risk assessment), due to it being an animal model, which used mice who were followed over their lifetime.
  • Older animals are more prone to illness. And so if carcinogenicity studies are done in mice more than 2 years old, then age-related changes confound the results.
  • Also the breed of mice used in this study are known to have a high incidence of spontaneous tumours.
• Controversy regarding aspartame stems from 3 studies from the same research group in Europe, all of which purported to show carcinogenicity in rats and mice; EFSA rejected findings as researchers misdiagnosed hyperplasia as malignant tumours and violated OECD testing protocols by administering aspartame during fetal development.
• 2015 meta-analysis of carcinogenic bioassay animal studies concluded there was no significant relationship between various experimental doses of aspartame and occurrence of cancerous tumours.

Saccharin

• Original concerns emerged from early animal toxicology studies in the 1970’s showing bladder cancer developed in rats administered high doses.
• Further research found that the carcinogenic mechanisms identified in rodents were not applicable to humans.
• Subsequent studies have found no associations between saccharin consumption and cancer in humans.
• Defined as safe for human consumption by both the FDA and EFSA

Sucraslose

• “The data support conclusions from regulatory agencies globally that sucralose is safe for its intended uses as a non-caloric sweetener.” – Magnuson et al., 2017

Acesulfame-K

• EFSA concluded Ace-K to be safe, including an extension of its permitted use to children up to 3-years of age
• Ace-K is not a safety concern at habitual levels of consumption, which are a fraction of the established ADI.

Overall Human Data on NNS & Cancer

• Liu et al., 2021 – meta-analysis of case–control studies
  • Consumption of artificial sweeteners was not associated with an increase in cancer when all types of cancers are analyzed comprehensively (OR 0.91, 95% CI 0.75–1.11)

• Both Cancer Research UK and the US National Cancer Institute have said sweeteners don’t cause cancer.
• World Cancer Research Fund: “There is no strong evidence in humans to suggest that artificially sweetened drinks with minimal energy content, such as diet sodas, are a cause of cancer.”

Associations with increased mortality

• Association with AS and mortality in EU – Mullee et al., 2019
  • 26% more likely to die if you regularly drink beverages with AS (n=450,000) – 8% more likely to die prematurely if drank sugar drinks
• Association with AS and mortality in US – Malik et al., 2019
  • Nurses Health Study – mortality and CV risk
  • Consumption of both sugar drinks and high intakes of AS drinks associated with CV risk and mortality
• What might explain this?
  • Do people who consume diet drinks lead more unhealthy lifestyles in general, or do NNS harm health?
  • Are people rationalising unhealthy choices by choosing a seemingly ‘good’ choice?
  • Difficult to currently explain such findings

Impact on Body Weight

1. Does use of NNS lead to weight gain?
   • Hypothesised claims relating to NNS still cause insulin responses or behavioural response to consume more calories (“AS trick the brain”)
2. Does use of NNS help with weight loss?
   • Mechanistically, it seems logical that swapping to NNS as opposed to sugar-sweetened products, will mean less calories ingested and thus a reduction in body weight. But of course, this is a very simplistic line of thinking.

Some mixed observations, for example…

Miller et al. 2014;

• 15 RCT’s
• AS use resulted in lower body weight, BMI and waist circumference
• RCT’s comprised a large sample size, including 4 studies in children, with no study finding that AS led to negative weight outcomes

Tucker et al., 2017 systematic review

• This experiment hasn’t been replicated in humans.
• Diet drink consumers vs non-consumers
• observational evidence that those who drink AS drinks are more likely to gain a small amount of weight.
• Case-control studies or cross over studies – initial weight-loss when participants change from sugar drinks to sugar-free drinks.
• Potentially minimal

Putting the effect on weight change seen in RCTs down to AS/NNS use, assumes that compensations are not made in calorie intake; i.e. substituting in AS but increasing calories from other sources. This would mirror the potential reverse causality observed in some observational studies.

An EFSA 2011 report found no clear cause-and-effect relationship to substantiate the claims that, when artificial sweeteners replace sugars, it causally maintains normal blood sugar levels, or maintains/achieves a normal body weight.

The CHOICE RCT (Tate et al., 2012) found that replacement of caloric beverages with noncaloric beverages as a weight-loss strategy resulted in average weight losses of 2% to 2.5%.

Do AS “trick the brain” into potential overconsumption?

• Frank et al., 2008 RCT comparing sucrose (table sugar) to the artificial sweetener sucralose showed that only sucrose led to brain activation responses.
• Chambers et al., 2009 study investigating the effects of carbohydrate mouth rinsing on performance and using fMRI found that presence of glucose was necessary to elicit a brain response
  • Glucose activated dopaminergic pathways that mediate reward-responses to food, but pathways were unresponsive to saccharin.
  • Carbohydrate mouth rinsing is a method studied as a potential way to improve sport performance. It is defined as “a CHO fluid distribution around the mouth for 5 to 10 seconds with subsequent expulsion by spitting”.
• A systematic review and meta-analyses (Rogers et al., 2015) on the effect of AS consumption on energy intake found that in studies using pre-loads of AS beverages followed by ad libitum test meals, consumption of AS resulted in a reduction in overall calorie intake.

Impact on Glycemia/Glucose Tolerance

We have two opposing claims to investigate:

1. Do NNS actually impair glucose tolerance or cause abnormal glucose/insulin responses?
2. As NNS are replacing sugar, could they be of benefit for glucose control in the long-term?

Some studies have suggested that NNS consumption can alter gut microbiota and therefore induce glucose intolerance. For example, Suez et al. (2014) wrote about this at length in their paper published in Nature.

• One of their studies was a rodent study, where they dosed mice over 11 weeks with water containing saccharin, sucralose or aspartame.
• At week 11, the mice consuming NNS sweetened water developed marked glucose intolerance, relative to those drinking either plain water, or water sweetened with sugar.
• Seemed to be down to changes in gut bacteria; the types of changes observed were “previously associated with type 2 diabetes in humans including over-representation of Bacteroides and under-representation of Clostridiales“
• They then wanted to see the effect in humans: for 1 week they examined seven healthy people who do not normally consume NNS
  • Participants consumed the FDA’s maximum ADI of saccharin (5 mg/kg)
  • 4 out of the 7 developed significantly poorer glycaemic responses 5–7 days after saccharin consumption, compared to their individual glycaemic response on days 1–4
  • Note: Because of the lack of a control group, it is unclear whether some of the worsening response is simply a function of them being exposed to 7 consecutive oral glucose tolerance tests (i.e. daily consumption of 75 grams of glucose). Could this have lead to changes in glucose metabolism in the absence of saccharin in some people?

More recent evidence on this question has been published in Microbiome by Serrano et al., 2021:

• Double-blind RCT investigatingimpact of saccharin on gut microbiota and glucose tolerance.
• Pure saccharin supp didn’t alter microbial diversity or composition of any taxonomic level in humans, no treatment effects for microbial activity – fecal metabolites or SCFAs but they did show some changes in mice.

The majority of interventions to date in this area have demonstrated no significant effect of NNS on glycaemic control (14 out of 18 trials). In contrast, a large number of animal studies have reported the effects on gut microbiota.

Pang et al., 2021 meta-analysis – no effect – The Impact of Artificial Sweeteners on Body Weight Control and Glucose Homeostasis

Alsunni et al., 2020 systematic review – no associations between AS on glucose regulation and its association with type2 DM and obesity.

Similarly, human trials have showed no impact of NNS on insulin response:

• Greyling et al., 2020 systematic review and meta-analysis found that there was no impact on insulin
• Moller, 1991 showed that aspartame did not affect insulin levels
• Wolf-Novak et al., 1990 found that aspartame had no effect on the insulin response in humans, whether alone or combined with carbohydrates
• Ma et al., 2009 found that sucralose caused no change in insulin levels

In the episode, Alan mentioned issues in the NNS glucose metabolism studies, most notably the type of placebo being used. Often water is used as the placebo, where caloric sweeteners (e.g. sugar) would arguably be a better control; if we assume NNS do stimulate glucose uptake, then this would occur in the context of low concentrations of glucose in the digestive tract, while a caloric sweetener like sucrose would result in a greater amount of glucose absorbed. (Renwick & Molinary, 2010 has a good discussion of some issues in the glucose metabolism/NNS studies for those interested).

Who should be cautious over use?

Aspartame should be avoided by people who have the genetic disorder phenylketonuria (PKU). They must avoid aspartame because they can’t process phenylalanine and accumulating high levels of phenylalanine can damage their brains.

ADI have been reviewed at intervals of 3-5 years in Europe and each time they have concluded there is no evidence that the current level is a safety concern. Only exception noted is those diagnosed with PKU.

What about migraine?

I (Danny) have heard anecdotal reports of people suffering with a migraine following consumption of NNS, particularly aspartame. And this is mentioned in the literature as well; with migraine sufferers noted as a group who hypothetically could benefit from avoiding NNS, if indeed they are the cause of the migraine. So is there evidene for this?

One interesting study in the NEJM (Schiffmann et al., 1987) may shed some light on this. In that study, they looked at people who reported having headaches repeatedly after consuming aspartame. When these people knew they were consuming aspartame, 100% of them reported headaches. So the researchers conducted a double blind crossover trial, with people being given either aspartame or a placebo, and then vice versa. They found that 35% had headaches after aspartame, and 45% had headaches after placebo. Therefore, this would suggest that the aspartame was not the actual cause of their headaches.

Conclusions: NNS & Health

Position statements:

1. Diabetes.org, 2018
2. British Dietetic Association (BDA), 2019
3. Academy of Nutrition and Dietetics, 2012

Broad conclusions from current evidence base:

• No human outcome data, particularly in RCTs, that shows negative health effects at the doses they are usually consumed in (even with high regular consumption).
• All hypothesised harms are based on animal and mechanistic data. This data cannot be extrapolated to making conclusions about human health, especially as it is in conflict with human findings.
• If indeed there are risks to health in humans, they have not yet be adequately shown, and therefore advising the avoidance of NNS is not an evidence-based recommendation.

Transcript