The story of sweeteners begins with a simple and reasonable premise. Sugar causes dental decay, contributes to obesity, drives insulin resistance, and feeds the kind of metabolic dysfunction that underlies much of modern chronic disease. If we could provide the sensory experience of sweetness without the caloric and metabolic consequences of sugar, that would be unambiguously beneficial.
This premise drove the development of artificial sweeteners across the twentieth century — saccharin from the 1880s, cyclamates, aspartame approved in 1981, sucralose in 1999 — and more recently the adoption of plant-derived sweeteners like stevia and novel sugar alcohols like erythritol as the "natural" alternatives that would sidestep the concerns about artificial chemistry.
The premise was not unreasonable. But the execution has revealed complications that the regulatory approval process — which assessed toxicity, not long-term metabolic and microbiome effects — was not designed to catch. And the "natural" alternatives, as we will see, are not as straightforwardly benign as their positioning implies.
The mechanisms nobody was looking at
The original sweetener safety assessment asked: is this compound toxic? Does it cause cancer at high doses in rodents? The answer for most approved sweeteners was: not at the doses people consume. That conclusion was and remains broadly accurate as a toxicity assessment.
What the assessment did not ask was: what does this compound do to the gut microbiome? What does it do to sweet taste receptor signalling? What are the long-term metabolic consequences of training the palate to expect sweetness without caloric consequence? These were not the questions being asked in 1981, because the tools to answer them didn't exist.
The Mechanisms Now Under Scrutiny
Gut microbiome disruption
Multiple sweeteners — particularly saccharin and sucralose — have been shown to alter gut microbiome composition in ways that impair glucose tolerance. A landmark 2014 study in Nature (Suez et al.) showed that saccharin consumption induced glucose intolerance in mice and in a subset of human subjects, and that the effect was mediated by microbiome changes — transferable to germ-free mice via faecal transplant. The sweetener itself wasn't directly causing the metabolic problem. The microbiome disruption it induced was. This is the opposite of what they were supposed to do.
Sweet taste receptor signalling
Sweet taste receptors are not confined to the tongue. They are expressed throughout the gut, in the pancreas, in the brain, and in adipose tissue. When these receptors detect sweetness — from any source — they initiate a cascade of signalling that anticipates caloric delivery: cephalic phase insulin release, appetite signalling, gut motility changes. When that caloric delivery doesn't arrive (because the sweetener has no calories), the signalling cascade is initiated but not completed. Over time, this may dysregulate the gut-brain communication about satiety and reward that governs food intake.
Appetite and compensation
Population studies consistently show that people who consume more artificially sweetened beverages do not, on average, consume fewer total calories. Multiple mechanisms may explain this — the sweet taste signalling argument above, reward pathway dysregulation, and the "halo effect" of consuming something perceived as healthy permitting larger overall intake. The epidemiological association between diet drink consumption and obesity, metabolic syndrome, and type 2 diabetes is now well established — though whether this reflects causation or a population that switches to diet drinks because they're already overweight remains contested.
Insulin response
Despite having no calories, some sweeteners produce measurable insulin responses — the cephalic phase insulin release described above. For people managing blood sugar, consuming a sweetened drink and expecting no insulin response may be an incorrect assumption. The degree of response varies by individual (another example of biochemical individuality) and by sweetener, but the idea that zero-calorie sweeteners are completely metabolically inert is not supported by the evidence.
The sweeteners — one by one
Found in: Diet Coke, sugar-free chewing gum, many "diet" products, some medications
Aspartame is a dipeptide of phenylalanine and aspartic acid, with a methanol ester. On digestion, it releases phenylalanine (contraindicated in phenylketonuria), aspartic acid (an excitatory neurotransmitter), and methanol. The methanol concern at normal intake levels is probably overstated — the quantity is small compared to naturally occurring methanol in fruit juices. The more significant development was the WHO's 2023 classification of aspartame as a Group 2B possible carcinogen — based on limited evidence from human studies showing associations with hepatocellular carcinoma at higher consumption levels. Group 2B means "possibly carcinogenic" — not the same as "causes cancer" — but the classification has not been rescinded.
Primary concern
WHO 2023 Group 2B possible carcinogen classification. Phenylalanine content relevant for PKU. Aspartate as excitatory neurotransmitter may be relevant in neurologically sensitive individuals. Gut microbiome effects less studied than sucralose/saccharin but emerging data concerning.
Assessment:Avoid where possible
Found in: Protein powders, protein bars, diet drinks, many "low calorie" products, some baked goods
Sucralose is chlorinated sucrose — sugar with three hydroxyl groups replaced by chlorine atoms. It passes largely unabsorbed through the gut, which is why it has essentially no calories. The chlorine substitution is the source of both its stability (it doesn't break down easily) and its biological effects — organochlorine compounds as a class have a complicated toxicological history. A 2023 study published in the Journal of Toxicology and Environmental Health found that sucralose-6-acetate — a metabolite and contaminant of sucralose — was genotoxic (capable of damaging DNA) at concentrations relevant to typical consumption. The FDA had not evaluated sucralose-6-acetate as a contaminant at the time of sucralose's approval.
Primary concern
Gut microbiome disruption — among the most studied sweeteners for this effect. Sucralose-6-acetate genotoxicity (2023 data). Fat-soluble accumulation in adipose tissue with prolonged use. Widely used in protein supplements specifically — a category consumed daily by health-conscious people who believe they're making a better choice.
Assessment:Avoid — especially in daily-use products like protein powders
Found in: Some soft drinks, tabletop sweeteners (Sweet'N Low), some medicines and toothpastes
The oldest artificial sweetener, discovered in 1879. Removed from the US carcinogen list in 2000 after the original bladder cancer concerns were attributed to mechanisms specific to male rats that don't apply to humans. The more recent and more robust concern is gut microbiome disruption — saccharin was the sweetener used in the landmark Suez et al. Nature 2014 study that demonstrated glucose intolerance induction via microbiome changes in human subjects. Of all the artificial sweeteners, saccharin has the strongest evidence for microbiome-mediated metabolic disruption.
Primary concern
Strongest evidence of any sweetener for gut microbiome disruption leading to impaired glucose tolerance. The mechanism is now fairly well characterised — microbiome changes that alter short-chain fatty acid production and gut-liver metabolic signalling. The metabolic effect is the opposite of what the sweetener was supposed to achieve.
Assessment:Avoid
Found in: Many "natural" products, better-quality protein powders, Seeking Health electrolytes, some soft drinks
Extracted from the leaves of Stevia rebaudiana. Steviol glycosides (rebaudioside A being the most common commercial form) are the sweet compounds. Stevia has a significantly cleaner safety profile than the artificial sweeteners — no genotoxicity concerns, no clear evidence of gut microbiome disruption at normal dietary doses, and some evidence of modest blood pressure-lowering effects. The metabolic effects on insulin signalling appear minimal at normal doses. The main clinical considerations are bitter aftertaste at higher doses, and emerging questions about whether it, like other sweet tastants, activates cephalic phase responses.
Current status
The most defensible sweetener alternative currently available. Particularly well-chosen in electrolyte products where the alternative is artificial sweeteners. Not entirely without question — long-term daily high-dose use hasn't been studied with the depth the artificial sweeteners have received — but the current evidence base supports it as significantly less concerning than the alternatives.
Assessment:Acceptable — best of the current options
Found in: Many "keto" products, Quest bars, some protein products, "natural zero calorie" sweeteners
A sugar alcohol that occurs naturally in small amounts in fruits and fermented foods, but is produced commercially via fermentation of glucose. Unlike other sugar alcohols (xylitol, sorbitol), erythritol is mostly absorbed in the small intestine rather than fermented in the colon — which is why it causes less GI distress than other sugar alcohols and why it doesn't significantly feed gut bacteria. It was considered relatively benign until a 2023 Nature Medicine study (Hazen et al., Cleveland Clinic) found that higher plasma erythritol levels were associated with significantly increased risk of major cardiovascular events — heart attack, stroke, and death — over a 3-year follow-up period. Importantly, the study also showed that consuming a drink sweetened with erythritol produced sustained elevation of blood erythritol levels for days — far longer than previously assumed.
Primary concern
The 2023 Nature Medicine cardiovascular signal is the most concerning recent finding in the sweetener space. The mechanism proposed involves enhanced platelet aggregation — erythritol appears to make platelets more prone to clotting. The study found a dose-response relationship. This is particularly significant because erythritol has been heavily marketed as the clean, natural, gut-friendly sweetener — and is consumed in large amounts in keto and low-carb product categories by people specifically trying to improve their metabolic health.
Assessment:Caution — 2023 cardiovascular signal warrants reduction especially in those with cardiac risk
Found in: "Health food" bars, some organic products, raw food products, wellness cafes
Perhaps the most misleading sweetener in the "natural" category. Agave is marketed as a low-glycaemic sugar alternative because it produces a lower glucose spike than table sugar. The reason it produces a lower glucose spike is that it is 70–90% fructose — significantly higher than table sugar (50% fructose) and high fructose corn syrup (55% fructose). Fructose is metabolised primarily in the liver, not requiring insulin for cellular uptake, which is why the glycaemic index is lower. But hepatic fructose metabolism at high intake drives de novo lipogenesis (liver fat production), raises triglycerides, and contributes to non-alcoholic fatty liver disease through mechanisms that are now well documented. The "low GI" label is technically accurate and clinically misleading simultaneously.
Primary concern
Higher fructose content than table sugar. Marketing as a health food alternative is one of the more egregious examples of appropriating the low-GI concept to sell a product that is worse for liver metabolism than what it replaces. Anyone with fatty liver, elevated triglycerides, or metabolic syndrome should be particularly cautious.
Assessment:Avoid — more metabolically problematic than sugar in liver terms
Found in: High-quality protein powders, Seeking Health products, some better supplements
Extracted from Siraitia grosvenorii, a fruit native to southern China. The sweet compounds are mogrosides — antioxidant compounds that produce intense sweetness without the glycaemic or caloric effects of sugar. Monk fruit has been consumed in China for centuries without documented adverse effects. The modern safety data is limited compared to the extensively studied artificial sweeteners, but no concerning signals have emerged. The mogrosides themselves have antioxidant and anti-inflammatory properties in cell and animal studies. Palatability is good — the sweetness profile is cleaner than stevia without the bitter aftertaste. It is currently the highest-quality sweetener option available in commercial products.
Current status
Best available option. Limited long-term data given relatively recent widespread commercial use, but no concerning signals from available evidence. The antioxidant properties of mogrosides are a genuine additional benefit absent from all artificial sweeteners. Appearance in a product is a positive formulation signal.
Assessment:Best available — clean profile, antioxidant benefit, no concerning signals
The "natural" sweetener problem
One of the more commercially successful moves in the sweetener industry has been the appropriation of the word "natural" — applied to stevia, monk fruit, erythritol, xylitol, and agave in ways that imply a categorical safety advantage over artificial sweeteners.
Nature does not guarantee safety. Ricin is natural. Arsenic is natural. The relevant question is not whether a compound occurs in nature but what it does in the human body at the concentrations being consumed. Erythritol occurs naturally in fruit at very small amounts. The quantities consumed in commercial keto products — 30–50g per serving in some products — are not remotely "natural" in any meaningful sense. The cardiovascular signal in the 2023 data is a dose-response relationship, and it emerges at doses that are common in commercial product consumption, not rare.
Similarly, agave is natural. Its fructose content is not a consequence of artificial chemistry — it is the plant's own carbohydrate profile. That doesn't make high-dose fructose consumption less hepatotoxic.
The clinical question is always the same: what does this compound do, at the doses being consumed, in the biological context of the individual consuming it? "Natural" is a marketing category, not a toxicological assessment.
The sweet taste problem that nobody wants to acknowledge
Beneath the specific concerns about individual sweeteners, there is a more fundamental clinical issue that the sweetener industry has no commercial incentive to address.
The desire for sweetness is a learned preference, maintained by habituation, and modifiable by changing the dietary environment. Population studies consistently show that people who reduce overall sweetener consumption — including artificial sweeteners — find that their threshold for sweetness shifts. Foods that previously tasted insufficiently sweet begin to taste appropriate. Fresh fruit, which tastes moderately sweet to a habituated palate, begins to taste intensely sweet. The preference recalibrates.
Sweeteners — artificial and natural alike — maintain the expectation of sweetness in every drink, every snack, every dessert alternative. They allow the continuation of a dietary pattern organised around sweet reward while modifying the substrate providing it. They do not address the underlying relationship with sweetness that, from a clinical perspective, is one of the primary drivers of the dietary patterns associated with metabolic disease.
The problem was never just the sugar. The problem was the expectation of sweetness at every meal, the dietary pattern built around sweet reward, the palate trained to require it. Sweeteners continue that pattern while changing the chemistry. The better solution is the harder one: recalibrating the palate to need less.
The practical position
This is not an argument that any sugar substitute is automatically worse than sugar. For someone managing type 2 diabetes or significant insulin resistance, reducing sugar intake is important enough that a transition period using better-quality sweeteners (stevia, monk fruit) while recalibrating the palate is a reasonable clinical strategy.
The practical hierarchy is clear from the evidence:
Avoid: Aspartame, sucralose, saccharin, agave syrup. The combination of WHO carcinogen classification (aspartame), genotoxic metabolites (sucralose), microbiome disruption (saccharin), and hepatic fructose load (agave) makes these the worst available options despite being the most widely used.
Use with caution: Erythritol — the 2023 cardiovascular signal is concerning enough to warrant reduction, particularly in anyone with cardiovascular risk factors. Not yet in the avoid category but the picture may develop.
Acceptable where sweetener is genuinely needed: Stevia (steviol glycosides), monk fruit — cleaner profiles, no genotoxicity concerns, no clear microbiome disruption signals at normal doses. Appearance in a product is a positive formulation marker.
The target: Reducing overall sweetener dependence — artificial and natural — and recalibrating palate expectations toward less sweetness overall. This is the clinical endpoint that sweeteners, by design, prevent you from reaching.
Concerned about what's in your daily products?
The DH Clinical Concierge can audit the ingredients in your current supplements and food products — including sweetener assessment — and tell you what to keep and what to replace.
Talk to the Concierge
Related reading