This artificial sweetener study leaves a sour aftertaste

Wanna know what annoys me? (Perhaps a better question is what doesn't annoy me.) Near the top of this very long list is the constant and mostly nonsensical war on all things sweet.

Sugar = bad.

Sugar substitutes = bad.

What is a soda drinker (also bad) supposed to do?

The answer, apparently, is drink nothing other than water, because anything sweet is destined to turn your brain into porridge. I don't like water. Too watery.

And speaking of dripping wet, the American Academy of Neurology recently gave new life to a 2025 study by promoting it in a press release. The predictable result was headlines around the world warning that artificial sweeteners may accelerate brain aging. (Show me something these days that doesn't.)

The press couldn't get enough of it. Nothing attracts attention quite like a story suggesting that a chemical we willingly consume is quietly destroying our health.

But anyone with even a modest background in chemistry or biochemistry should immediately notice a serious flaw in the study's premise. The authors lumped together aspartame, saccharin, acesulfame-K, erythritol, xylitol, sorbitol, and tagatose, then suggested that higher consumption of these sweeteners might accelerate cognitive decline in older adults.

A collection of chemicals with unrelated structures producing the same biological effect is a giant red flag, at least to a chemist. That doesn't mean it's impossible. It means there had better be a convincing common mechanism. There isn't.

So, I call bullsweet. 

Sweet isn't a chemical structure

Biology cares about chemistry—not flavor. One of the first lessons medicinal chemists learn is that biological activity depends on molecular structure. Even tiny changes in a molecule can make a huge difference in its properties. (For example, adding a methyl group to fentanyl, a very small change relative to the size of the drug, can make it 10-100 times more potent.)

Yes, they all taste sweet because they activate the same sweet taste receptor. Once they're swallowed, however, all bets are off. [1]

The rest of the body doesn't classify molecules by flavor. Enzymes, receptors, and transporters recognize three-dimensional shape, charge distribution, hydrogen-bonding patterns, lipophilicity, and a bunch of other structural features. Those are the properties that determine biological activity, not whether something tastes sweet.

Even tiny structural changes can completely change biological activity. Replace one atom, move one functional group, or flip one stereocenter, and a harmless molecule can become toxic—or vice versa. That's why medicinal chemistry exists. It's also why toxicologists generally become nervous when someone proposes that a collection of structurally unrelated compounds all share the same toxic effect. In the absence of a plausible biological mechanism, an observational association deserves a healthy dose of skepticism. 

The paper doesn't really offer a unifying mechanism. Instead, it presents a grab bag of unrelated possibilities:

• Aspartame has been reported to cause neuroinflammation in rodents.
• Artificial sweeteners may alter the gut microbiome.
• Some studies suggest impaired glucose tolerance.
• A "cafeteria diet" containing numerous additives affects the brain.

Those aren't explanations for the study's central observation. They're a series of unrelated hypotheses involving different compounds, different experimental conditions, different organs, and even different species.

Metabolism rules

Aspartame is rapidly broken down into two amino acids and methanol. Saccharin is largely excreted unchanged. Erythritol is mostly absorbed and excreted in the urine. Sorbitol and xylitol are extensively fermented in the colon. These compounds don't share metabolism. They don't share pharmacokinetics. They don't even belong to the same chemical families. Nor do they share molecular targets. Some are synthetic, and others are naturally occurring.

Yet we're supposed to believe they all damage the aging brain? (And while we're at it, guess a formerly common household substance that may have tasted sweet and definitely caused brain damage.) [2]

It simply doesn't make chemical sense. When a conclusion runs counter to basic medicinal chemistry, the statistical evidence had better be exceptionally convincing.

It's not.

There are results, and there are results.

Here's one that bothered me:

Buried in the paper is a sentence that received almost no media attention:

"In complete case analysis, there was no association between tertiles of LNCS [Low- and No-Calorie Sweeteners] consumption and cognitive decline."

Natalia Gomes Gonçalves et al., Neurology (2025)

Translated from statistical jargon into English:

The original analysis included statistical methods to compensate for participants who dropped out or had incomplete follow-up data. When the authors repeated the analysis using only participants with complete data, the association disappeared.

If that's not problematic, I don't know what is. That doesn't make the analysis wrong. But it sure makes it less definitive than the headlines suggest. There's another possibility. Maybe the sweeteners aren't the story.

Opening an umbrella doesn't make it rain

People who consume lots of low-calorie sweeteners aren't randomly selected from the population. Many are overweight. Many have diabetes or prediabetes. Many are actively trying to reduce sugar intake. Many consume large amounts of unhealthy foods.

The authors adjusted for many of these variables, but no observational study can completely eliminate residual confounding. To their credit, they acknowledge this themselves.

Although we adjusted the regression models for several clinical and lifestyle variables, we cannot exclude the possibility of residual confounding, particularly because health behaviors may co-occur and some groups may consume more LNCSs because of their lifestyle and clinical history.

...and also

"...people at higher risk of type 2 diabetes may be more likely to consume LNCSs instead of sugar."

I think the authors underestimate the importance of this point. If six chemically unrelated compounds all produce similar epidemiologic associations despite having different structures, metabolism, and pharmacology, the first explanation I'd consider isn't a shared toxic mechanism. It's a shared pattern of behavior. In other words, the association may say more about who consumes these sweeteners than about the sweeteners themselves.

Correlation isn't chemistry

Observational studies are useful for generating hypotheses. They can reveal patterns worth investigating. But they cannot tell us whether one thing caused another, especially when the proposed mechanism contradicts basic chemistry and biology.

That's what bothers me most about this paper.

If someone told me that aspirin, penicillin, morphine, caffeine, and Lipitor all increased the risk of dementia, my first question wouldn't be, "How much?" It would be, "What on earth do these molecules have in common?"

That's exactly the question that should have been asked here.

Instead, seven chemically unrelated compounds that merely happen to taste sweet were treated as though they formed a meaningful biological category. I just don't buy it.

The study may ultimately turn out to be right. Science has surprised us before. But extraordinary claims require more than modest epidemiologic associations and a handful of speculative mechanisms pulled from unrelated studies.

Until someone demonstrates how these structurally unrelated compounds could plausibly produce the same effect on the aging brain, I'm not buying it.

Not because I have a sweet tooth.

Because I'm a chemist with a sweet tooth.

NOTES:

[1] Strictly speaking, sweet taste receptors (T1R2/T1R3) are found not only on the tongue but also in the gastrointestinal tract and several other tissues, where they may play roles in hormone secretion, nutrient sensing, and metabolism. My point is not that these receptors don't exist. Rather, activation of a common sweet taste receptor says little about what structurally unrelated molecules do elsewhere in the body. Once absorbed, their interactions with enzymes, receptors, transporters, and other biological targets are determined by their individual chemical structures, not by the fact that they all taste sweet.

[2] Some lead compounds, particularly lead acetate ("sugar of lead"), have a sweet taste. This may have contributed to the common belief that lead-based paint tasted sweet, although the principal pigments used in paint were different lead compounds.