The carcinogen your own mouth makes out of ordinary sugar

A carcinogen with a local supply chain

When most of us hear the word carcinogen, we picture something arriving from outside — smoke, a chemical, a bad batch of something. Acetaldehyde is not like that. When it comes from the metabolism of alcohol it sits in Group 1, the top category the international cancer agency uses — the same one as tobacco smoking ². And the point of a new paper is that your own mouth can manufacture it on site, out of the sugar in a biscuit.

The study is by Wakade, Liu, Diep and Daniell at the University of Pennsylvania School of Dental Medicine, published in Molecular Therapy: Oncology in 2026 ¹. Let me walk through what they actually did, because the mechanism is simple enough to picture — and the caveat is big enough that it belongs in the same breath.

What they grew, and what it made

They did not order tidy reference strains from a catalogue. They took saliva from patients with head and neck squamous cell carcinoma and cultured out what was living in it: a yeast of the genus Candida, aerobic streptococci, and two anaerobes long familiar from gum disease — Fusobacterium nucleatum, of the Fusobacteriota, and Porphyromonas gingivalis ¹.

Then they fed them, and watched for a pink halo. The detection trick is charming in its simplicity: acetaldehyde diffusing out of the culture reacts with a dye in the agar and stains a spreading circle. Bigger circle, more carcinogen.

Five separate yeast colonies isolated from patient saliva all made the stuff, in a tight band of 260–321 µM ¹. The colony that made the most, at 321 µM, was carried forward ¹. Feed it more glucose and it made more — rising from 5 mM up to 100 mM, then falling off again at 200–400 mM, probably because that much sugar is itself stressful to a cell ¹. The same shape appeared with alcohol: production peaked around 100 mM ethanol, then declined ¹. The two anaerobes peaked at 170 mM ethanol ¹. The streptococci were the interesting ones — they made acetaldehyde in ethanol-free medium, which the authors put down to glucose metabolism through pyruvate; with ethanol added their output climbed further still, peaking between 500 and 1,100 mM ¹.

That last detail is the one I would underline. This is not only a story about drinking. A mouth carrying the wrong community has a route running from breakfast to acetaldehyde — the molecule the cancer agency puts in Group 1 when it arrives by way of alcohol ², and which the authors of this paper call a Group 1 carcinogen outright ¹.

The lock is the biofilm — the sticky plaque shell, not the bacteria

Here is the part with a mechanism anyone can picture. The authors used protegrin‑1, an antimicrobial peptide, together with an extract of the bean gum, to kill the organisms and stop the output. Against a streptococcal isolate with no real biofilm it worked outright — acetaldehyde fell from 272.06 µM to the edge of detection, and viability collapsed ¹. Against the biofilm-former it went only part of the way: acetaldehyde dropped from 869.7 µM below the detection limit, but just 91.37 % of the colonies died — the rest lived on ¹.

Bacteria in a biofilm live inside a shell of sticky sugar polymer. Pouring an antiseptic over that is like spraying disinfectant on a zipped-up tent and calling the inside clean. So the team added two enzymes that digest exactly that shell — dextranase and mutanase. With the tent open, the peptide finished the job: acetaldehyde and colonies both went to nothing ¹. In the two anaerobes the same combination brought output down from 685.7 µM to negligible levels and from 473.6 µM to undetectable ¹.

The yeast needed a different key. Its wall is chitin, glucans and mannoproteins, so they reached for β-glucanase, chitinase and mannanase. A microbial cocktail took acetaldehyde from 978.8 µM to 38.7 µM — a 96.04 % reduction — with 99.99 % of the cells killed ¹. A plant-made version of the same enzymes, grown inside lettuce cells, brought 873.9 µM down to roughly 70.8 µM over three sequential doses ¹.

The delivery idea is chewing gum: the enzymes made in plant cells and pressed into a tablet, released slowly while you chew — though the peptide used in these experiments was chemically synthesised rather than plant-grown ¹. A related gum from the same laboratory already holds an investigational new drug number, 154,897, for an unrelated antiviral purpose ¹.

Now the catch, out loud

This is a dish. Every result above happened in vitro, in monoculture, and the authors say so themselves — real mouths are mixed communities where organisms feed and shelter one another, and that was not modelled. Mutagenicity was never measured; they quantified acetaldehyde, not DNA damage. There were no healthy-donor yeast isolates to compare against, because Candida had not been found in the healthy saliva analysed earlier. The isolate-level work rests on five yeast colonies, two streptococcal isolates, and three biological replicates per viability assay ¹. No patient was treated. No cancer was prevented. No human trial has been run.

So what survives once the excitement is stripped out? A clean demonstration that the sugar and alcohol sitting in your mouth are raw material, that the biofilm is the shield, and that enzymes which open the shield are a plausible way through. That is a good day's work. It is not a product, and it is not advice.

What it does justify today is unglamorous: fewer sugar exposures spread across the day, less alcohol, and mechanical disruption of the biofilm — brushing and flossing the plaque off — morning and night. Those act on the same supply chain — and unlike the gum, they already exist.