IgG Elimination Diet for Migraine: What a New Trial Found

A 12-week randomized trial asked whether cutting foods flagged by an IgG blood test reduces migraine. The diet group did better — and their inflammation markers dropped too. Here is what the numbers actually say, and why the test behind them is still disputed.

Almost everyone with migraine has been told that food is involved. Skip the cheese. Skip the wine. Skip the chocolate. The advice is everywhere, and it is mostly guesswork — a list of usual suspects handed to people whose actual attacks may have nothing to do with any of them.

A blood test that could tell you your foods would be genuinely useful. That is the promise of food-specific IgG testing, and it is why the test is sold direct to consumers for a few hundred dollars despite a long-standing objection from allergy specialists that it does not measure what people think it measures.

In December 2025, a team at Harbin Medical University in China published a randomized, sham-controlled trial in Frontiers in Nutrition testing the idea head-on in people with migraine. It is one of the better-designed studies on this question, and it found a real difference between groups — including in blood markers of inflammation, and in CGRP, the same molecule that the newest migraine drugs are built to block.

It is also a single study, in 95 people, in one city, over 12 weeks. Both of those things are true at once. Here is what it did.

What Is an IgG Food Test?

IgG is an antibody — one of the proteins your immune system makes. A food-specific IgG test measures how much IgG antibody your blood carries against particular foods.

Here is the part that causes the argument. IgG antibodies against food are normal. Most people have them, and having them mostly means you have eaten the food. That is different from IgE, the antibody behind true food allergy, where a positive result plus a matching history is genuinely diagnostic.

The major allergy societies have said this plainly. A European Academy of Allergy and Clinical Immunology task force concluded that food-specific IgG4 testing should not be used to diagnose food allergy or intolerance, describing the antibodies as a physiological response to food exposure. The American Academy of Allergy, Asthma & Immunology formally endorsed that position, and the Canadian society issued its own statement to the same effect.

So the question this trial is really asking is narrower and more interesting than "does IgG testing work?" It is: if you remove the foods this test flags, do people with migraine get better than if you remove a comparable set of foods it did not flag?

How the Study Was Designed

The design is the strongest thing about this trial, so it is worth walking through.

That sham diet is what makes this trial worth reading. Comparing an elimination diet to "carry on as usual" tells you almost nothing, because being handed a diet plan and asked to follow it carefully is itself a powerful intervention. Here, both groups gave up foods, both groups had to think about what they ate, and both groups were being watched. The only difference was which foods came out.

Compliance was high and, importantly, the same in both groups: 87% of participants said they stuck to their plan 76–100% of the time, with self-reported adherence of 84% in the true diet group and 86% in the sham group.

Participants averaged about 39 years old, roughly two-thirds were women, and they came in with a mean MIDAS disability score around 44–49 — that is the severe range — and about six migraine days in the previous four weeks. The trial was registered in advance (ChiCTR2000039278) and funded by Chinese provincial and hospital research grants, with no commercial or financial relationships declared.

What the Trial Found

Migraine Days: A Small Difference

Over 12 weeks, migraine days in the previous four weeks fell by 1.4 days in the true diet group and 0.4 days in the sham group. The gap between groups was statistically significant (p = 0.028).

Be clear-eyed about the size of that. Net of the sham diet, the true diet bought about one migraine day per month. For someone having six, that is real but not transformative. For comparison, CGRP monoclonal antibodies typically deliver something in the range of one to two days per month better than placebo in their trials — so this is in a similar neighborhood, from a dietary change rather than an injection.

Disability and Pain: A Bigger Difference

The measures of how much migraine interfered with life moved more than the raw attack count did.

A 6-point drop on HIT-6 is not a rounding error; the scale runs 36 to 78 and a change of about 6 points is generally treated as clinically meaningful. The gut symptom result is striking too — the true diet group's digestive complaints improved substantially while the sham group's did not budge at all.

The anxiety, depression, and quality-of-life scales showed nothing. The authors attribute that partly to the timing: enrollment ran through the winter of 2020–21, and a pandemic is a poor backdrop against which to detect a diet's effect on mood.

The Blood Markers: The Most Interesting Part

This is where the study goes beyond previous work. The researchers measured inflammatory and nerve-signaling molecules before and after.

IL-6 and TNF-α are two of the body's main inflammatory signals. CGRP — calcitonin gene-related peptide — is the molecule released by trigeminal nerve endings during a migraine attack, and it is the direct target of Aimovig, Emgality, Ajovy, Ubrelvy, Nurtec, and Qulipta. Seeing a diet push CGRP down is a genuinely notable finding.

The authors went one step further and ran a mediation analysis, which asks whether the inflammation changes explain the symptom changes. When they statistically adjusted for IL-6 and TNF-α, the between-group difference in migraine days weakened from significant (p = 0.0098) to non-significant (p = 0.076). That is consistent with inflammation being part of the causal chain — the diet lowers inflammation, lower inflammation means fewer migraine days. It is suggestive, not proof, and mediation analyses in samples this small should be read gently.

The IgG Levels Themselves

Total positive IgG fell 145.5 U/mL in the true diet group versus 34.3 U/mL in the sham group, and the share of people still testing positive dropped from 100% to 64% in the true group but only to 88% in the sham group.

This confirms people actually followed the diet — stop eating a food and your antibodies to it decline. It is not independent evidence that the antibodies were causing anything. Both would look exactly the same in the data.

Why This Result Is Contested

Three things should temper enthusiasm.

A larger trial of the same idea failed. In 2011, Mitchell and colleagues published a sham-controlled elimination diet trial in Nutrition Journal with 167 participants — considerably more than this one. They found a significant reduction in migraine frequency at week 4, but by week 12 the difference between true and sham diets was no longer statistically significant. The earlier Alpay 2010 crossover trial in Cephalalgia was positive, but had only 30 participants. So the literature is genuinely split, and the largest study to date is the negative one.

The test flags too many people. Of 129 people screened here, 98 — about 74% — had at least one positive food IgG. A test that is positive in three-quarters of everyone it touches is not identifying a rare culprit; it is mostly reporting what people in Harbin eat. Egg, milk, and shrimp topping the list is consistent with exposure, not with a specific pathological reaction.

Some of the benefit may come from eliminating anything. The sham group improved too, on several measures. And there is a plausible mechanism that has nothing to do with IgG: removing egg, milk, or wheat changes far more than one antibody level. It changes fat intake, fermentable carbohydrate load, meal timing, how much processed food you eat, and how much attention you pay to eating at all — and skipped or erratic meals are among the better-established migraine triggers there are.

The Study's Own Limitations

To the authors' credit, they list most of these themselves:

• Small and single-center. 95 completers at one hospital in one Chinese city, on a 14-food panel built around local eating habits. Whether it transfers to other diets and populations is unknown.
• Short. Twelve weeks. Migraine fluctuates on its own over months, and the Mitchell trial is a reminder that a difference visible at week 4 can be gone by week 12.
• Single-blind, with self-reported outcomes. Participants were blinded, but nearly every outcome was a questionnaire. Blinding in diet trials is imperfect — someone told to cut a food they suspected might guess their assignment.
• Nobody was on preventive medication. The trial excluded anyone taking migraine preventives, so it does not tell you whether this adds anything on top of standard treatment.
• Pandemic timing. Enrollment spanned the winter of 2020–21, which the authors say likely flattened the mood and sleep findings.
• No effect on several markers. IL-10, serotonin, and VIP did not move, which the authors read as the study being underpowered for those measures.

What This Means If You Have Migraine

Should I get an IgG food sensitivity test?

Probably not as a first step, and not on the strength of one trial. The major allergy societies advise against these tests, the largest sham-controlled study of this approach was negative at 12 weeks, and the tests typically cost a few hundred dollars out of pocket because insurers do not cover them.

Is it worth trying an elimination diet anyway?

This is the more useful question, and the answer is often yes — but you can run the experiment without the test. The thing this trial demonstrates most solidly is that carefully removing specific foods for several weeks, and tracking what happens, can measurably change migraine burden. You can do the same with a structured trial of one food at a time.

What that requires is honest data. Food effects in migraine are slippery: attacks can lag a trigger by up to 24 hours, and the prodrome phase can create food cravings before the headache — which is how chocolate got its reputation as a trigger when it may often be a symptom. Neither pattern is something you can reliably reconstruct from memory a month later.

What should I do instead, or first?

The interventions with the strongest evidence behind them are still the unglamorous ones: consistent sleep, regular meals, hydration, exercise, and — if you are having frequent attacks — a proper conversation about preventive treatment. We cover what the current guidelines recommend in our plain-English guide to migraine medications. If you are having four or more migraine days a month and are not on a preventive, that conversation will almost certainly do more for you than a food panel.

The Bottom Line

This is a decent trial that found a modest, real-looking benefit and paired it with a plausible biological story — lower IL-6, lower TNF-α, lower CGRP. The sham-controlled design is a genuine strength, and the CGRP finding is the kind of result that deserves a follow-up.

It is not enough to overturn the allergy societies' position on IgG testing, and it does not settle a literature where the largest previous trial came out negative. One study of 95 people over 12 weeks in a single city is a reason to run a bigger, longer, multi-center trial. It is not yet a reason to spend a few hundred dollars on a blood panel.

What it does support is something more modest and more actionable: for some people with migraine, food really is part of the picture, and a careful, tracked elimination trial is a reasonable thing to attempt with a clinician's help.