The Gluten Threshold Study: What It Actually Says

A study published in Gastroenterology this year put a number on something coeliacs have argued about for decades: how little gluten it takes to wake the immune system up. The number is small. And within days, a widely shared post in coeliac Facebook groups turned its most quotable line into almost exactly the opposite of what the study says. Here is what it found, and what it does and does not change about an ordinary week.

What did the study actually do?

Researchers at the Wesley Research Institute in Brisbane recruited 51 adults with biopsy-proven coeliac disease who had been on a gluten-free diet for more than two years (median age 52; 69% women). Each took part in three oral gluten challenges, four weeks apart — 153 challenges in total, at doses from 1 mg up to 1000 mg, with placebo mixed in. It was randomised and double-blind: nobody involved knew which dose was which.

The crucial design choice is what they measured. Rather than relying on how people said they felt, they took blood and measured interleukin-2 (IL-2) — a signalling molecule that rises within hours when gluten reaches the immune system of someone with coeliac disease. A challenge counted as a response if IL-2 at least doubled.

What did it find?

A clear dose–response pattern. The more gluten, the more people showed a measurable immune response — and the response did not vanish as neatly as you might expect at the bottom of the ladder.

At the top of the range, five in six responded. At 90 mg, about a third. Below that, roughly one in five — small, but not zero. The lowest dose at which anyone responded was 3 mg; at 2 mg, 1 mg and placebo, nobody did. Modelling the curve, the authors put the dose that would trigger a response in one person in ten at 2.4 mg (with a wide margin of uncertainty, 0–5.3 mg), and the dose for half of people at 111 mg.

One honest wrinkle: the ladder is not perfectly tidy — the 8 mg rung produced more responders than the 13 mg rung above it. With this few challenges per dose, that is the wobble small numbers produce. Read the bottom of the table as a low but real response rate, not as precise percentages.

Why is “no symptom difference” the most misunderstood line in the paper?

Here is the finding that went around the internet: symptom scores rose after the challenges, but not beyond what placebo produced. Below the very top of the range, people could not tell which challenge they had been given. A widely shared post used that line to argue that coeliacs who report reacting to traces are imagining it.

That inversion matters because the two readings lead to opposite behaviour. Read one way, the paper says relax. Read as written, it says the reverse: at the low end, the body was reacting while the person felt nothing. Feeling fine after a meal is evidence about your symptoms — not about your immune system, because in this study the two came apart at exactly the doses daily life serves up.

It cuts the other way too: the placebo challenges also produced symptoms. Guts are noisy narrators in both directions, and a bad afternoon is no more proof that you were glutened than a good one is proof that you were not.

What does this change about your daily life — and what doesn’t it?

It changes nothing about how a well-run gluten-free diet works. The habits stay the habits: read labels, ask about shared fryers and toasters, tell the kitchen it is a medical requirement rather than a preference. Nothing here asks you to eat differently tomorrow.

It strengthens the case for asking about cross-contact. If a few milligrams can register on a blood test, then the questions that feel fussy in the moment — separate board, fresh gloves, own pan, what else goes in that oil — are the ones doing the work. They were always the right questions; this is one more reason they are.

It explains why “I feel fine” isn’t proof. Plenty of coeliacs are told, kindly, that a bit of cross-contact clearly does not bother them because they had no reaction. This is the cleanest answer yet: absence of symptoms is not absence of a response. If you have ever been the person who could not point to a symptom and therefore could not make your case, the data are on your side.

It does not mean gluten-free-labelled food is a problem. Certified and labelled products remain the most predictable food available to us. If any of this makes you want to reconsider your own routine, that is a conversation for your clinician or dietitian rather than for an article — we write guides, not medical advice.

What does it mean for the 20 ppm rule?

Some arithmetic, because it is the part everyone skips. The gluten-free standard used across the EU, the UK, the US, Australia and much of the world is 20 parts per million — 20 mg of gluten per kilogram of food. A 100 g portion sitting exactly at that ceiling would contain 2 mg; a generous 150 g portion, 3 mg. So the lowest rung that produced a response does sit inside the range a labelled product could in principle deliver.

“In principle” is doing real work there. 20 ppm is a ceiling, not a target, and gluten-free food routinely tests well below it — many certified products come in under 5 ppm. The threshold was set as a public-health line: one that works for most people most of the time, and that makes an entire category of bread, pasta and beer exist at all. A world with no threshold is not a world with less gluten in it; it is a world with no labelled food.

What this study does is ask a legitimate question about the margins of that line — whether it leaves enough room for people at the sensitive end. That is a fair question, and one regulators answer over years and across many studies, not on the strength of one trial with 51 participants. Nothing about your shopping changes this week.

What do we still not know?

Quite a lot, and the honest version of this article has to say so.

• It is a small, single-centre study. 51 people, 153 challenges. Spread across nine dose levels, that leaves very few observations per rung — which is why the percentages wobble.
• IL-2 is a marker of immune activation, not a measure of damage. This is the most important limit. A rise in a blood marker is not the same thing as a flare, and certainly not the same thing as villous atrophy. The study did not take biopsies at these doses, and cannot tell us whether these blips matter over years.
• These were one-off challenges. Real life is not a single 3 mg dose four weeks apart; it is small, repeated, unpredictable exposure. Whether that behaves like the single doses tested here is the more relevant question, and this study does not answer it.
• Most people did not respond at the low doses. At the bottom rungs, roughly four in five showed no measurable response. Individual variation is wide, and there is no practical way to know where you personally sit.

Questions we have been asked about it

Does this mean I have been reacting to foods I thought were fine?
It means you might have and would not necessarily have known — and equally that you might not have, since most participants showed nothing at the low doses. What it removes is the assumption that feeling nothing settles the question.

Should I stop buying gluten-free-labelled products?
No. They are the most predictable food we have, they test well below the legal ceiling far more often than not, and nothing here suggests otherwise. Narrowing your diet in response would trade a small, uncertain risk for a definite nutritional one.

Can I get an IL-2 test to find out how sensitive I am?
It is a research assay rather than a routine clinical test, and the study was not designed to grade individuals. If sensitivity is a live worry, your gastroenterologist or dietitian is the person to raise it with.

Is 3 mg a number I should now be tracking?
No — and please do not try. Nobody can count milligrams across a restaurant table, and a diet run on arithmetic anxiety is worse for you than the milligrams. The usable takeaway is behavioural: keep asking about cross-contact, and stop treating “I felt fine” as the final word.

My relative eats the odd crumb with no problem. Does this prove them wrong?
Not quite. It shows that no reaction does not prove no response — but also that most people had no measurable response at the smallest doses. Neither of you can tell from how you feel, which is a better argument for caution than for certainty.

The source, and where to read more

Daveson AJM, Tye-Din JA, Anderson RP, et al. “A Randomized Double-Blind, Placebo-Controlled Dose–Response Study to Assess the Gluten Threshold Dose in Celiac Disease.” Gastroenterology, 2026. DOI 10.1053/j.gastro.2026.03.011 — read the paper.

• What does “gluten free” on a label actually mean? — where the 20 ppm figure comes from and what it promises.
• What does “may contain gluten” mean? — the voluntary warning that is not the same thing at all.
• The new Codex guidance on “may contain” labels — how thresholds actually get set and changed.

Written 29 July 2026. We summarise published findings in our own words; figures above are drawn from the study’s reported results. This is general information for coeliacs, not medical advice — decisions about your own management belong with your clinician or dietitian.