Autoimmune protocol and paleo: what an elimination diet can do

There is one sentence I hear in my consulting room almost every week. Usually only at the end of the conversation.

It goes: can I eat my way out of this?

What sits behind it is rarely naivety. What sits behind it is exhaustion. An autoimmune disease takes away the feeling that your body is on your side. Food is the one lever you get to pull yourself, every single day.

This is exactly where paleo and the autoimmune protocol come in, AIP for short. Both offer order, both deliver lists. And both have a serious core plus a story that reaches further than the data. I will show you both: what was measured, and where the story grows larger than it can carry.

Can I eat my way out of this? The question behind the question

Many people with an autoimmune diagnosis know this moment. You read online at night, you find lists, you find people who feel better since they dropped tomatoes. That is not a stupid idea. It is an underspecified question.

Autoimmunity means the immune system loses tolerance towards its own tissue. That almost always needs three things: a genetic susceptibility, a trigger from outside, and a mucous membrane that lets it through. Food is the entrance you can steer, which is why it gets all the attention.

One finding has fascinated me for years. A systematic review analysed eleven studies of dietary interventions in inflammatory bowel disease, covering ten different concepts. What is interesting is what sits between them: almost all share the same building blocks. Elimination phase, reintroduction, unprocessed foods. Ten names, one pattern.

And that makes it clear why the answer does not sit in a list of banned foods. A list only answers the first question.

Paleo and AIP are not the same thing, and that matters

Maybe you have compared two eating plans and noticed that they contradict each other. One allows eggs, the other does not. Two different things are sharing one reputation here.

Paleo is a long term way of eating with no expiry date. It drops grains, legumes, dairy, refined sugar and heavily processed products. Its argument is evolutionary, its evidence base is metabolic.

The autoimmune protocol is narrower, time limited, and in form it is not a meal plan but a procedure. On top of paleo it drops eggs, nuts, seeds, nightshades, coffee, alcohol and most additives. Some descriptions also mention painkillers. Anything touching medication belongs in medical hands, not on a diet list.

The concept goes back to the biophysicist Sarah Ballantyne and spread as a book, not as a research programme. That is not a put down, it is context: the idea moved from practice into research.

And now you understand why the question about duration is the most important question about this protocol.

The paper everyone talks about: what it says and what it is

If you research this topic, sooner or later you hit a number that outshines everything else. It is almost never quoted with its source. So let us look at the source.

And now you know why I name this paper instead of leaving it out. You will meet it anyway, and then you should know what you are looking at.

The real AIP studies and their awkward details

They do exist, the proper studies on the autoimmune protocol. Together they cover fewer than a hundred people, and not one of them has a control group. No reason to ignore them. A reason to read them closely.

The same group analysed quality of life. The questionnaire score rose from 46.5 to 62.0, earlier than any measurable change in inflammatory markers. That is two things at once: a hint of benefit in daily life, and the classic place where expectation and attention do their work.

In rheumatoid arthritis there is a feasibility study in nine adults. Seven of nine improved, and at the same time BMI fell from 26.3 to 24.5. Weight loss on its own is already relevant to inflammation.

And there is a contradiction I do not want to smooth over. A 2026 review analysed fifteen studies using a validated quality of life instrument. For several low FODMAP studies it found significant improvements, for the autoimmune protocol it found none. Not a scandal, just ordinary science.

And now you understand why my first question is what you would notice if things were getting better.

What might be happening in the body: barrier, mimicry, brake

Why should leaving things out change anything? The answer has four building blocks. None is proven, all four are well described.

First: the barrier

Your gut lining is not a wall. It is a row of cells with junctions between them that can open and close like doors along a corridor. Zonulin is so far the only known switch for that, and gliadin from wheat can trigger its release. In biopsies from four groups, permeability rose in all of them, most strongly in active coeliac disease. The difference lay in the degree, not in whether it happened. Details under leaky gut, intestinal permeability and zonulin.

There is more to wheat than gluten. Amylase trypsin inhibitors can bind to the TLR4 complex on immune cells in cell experiments and in the mouse model, setting off an innate inflammatory response, in tissue from people with and without coeliac disease. Mechanistically convincing, not yet tested in humans as a treatment approach. More under gluten and gliadin without coeliac disease.

Second: the mix up

Molecular mimicry is one of the most elegant explanations in this field. A foreign protein looks similar enough to one of your own, the immune system attacks it and hits your own tissue along the way.

The major review on this names its own limits: long latency between trigger and disease, too little statistical power, poor transferability from inbred mice. A second paper adds that structural similarity alone often does not explain the activation. Mimicry is recognised as a mechanism and in the individual case it is almost never provable. The story that a food looks like your thyroid is tidier than immunology itself.

Third: the players in your gut

In newly diagnosed rheumatoid arthritis, a sequencing study of 114 stool samples found a strong enrichment of Prevotella copri. In mice the same organism increased susceptibility to gut inflammation. Correlation plus animal model, no proof of causation.

An underrated observation fits here: depending on the bacterial mix, the same wheat protein can be broken into fragments of differing immune reactivity. The same food can generate two different signals in two people. That is the KPNI lens in one sentence.

Fourth: the brake

Your immune system does not only have an accelerator. Regulatory T cells are the brake, and they are partly built in the gut.

And there is the opposite direction. Sodium chloride increased the formation of inflammatory Th17 cells in cell experiments with mouse and human cells. Salt is on no AIP list. It is in almost every convenience product.

And now you know why the mechanism level sounds convincing and still delivers no outcomes. It explains possibilities, not courses of illness.

Lectins: a fact check without picking a side

Few nutrition topics are this fragmented. On one side, tables listing four hundred foods. On the other, a blanket all clear. Both sides usually work without a primary source.

Lectins are proteins that bind to sugar structures. Plants use them for defence. They occur in beans, lentils, grains and nightshades, in very different amounts.

And still, one lectin belongs in the discussion for good reason. Wheat germ agglutinin appears on the candidate list in a review of non coeliac wheat sensitivity, alongside fructans and the amylase trypsin inhibitors. The lectin story simply claims more than these sources can support.

One distinction pays off in daily life. If your symptoms sit mostly in your belly, meaning bloating, cramps, changes in stool, the theme is often not the immune system but fermentation in the small intestine. There is a different and better studied tool for that: how to use a FODMAP diet properly.

And now you know why I get sceptical about lectin lists. They take away something whose benefit is documented, against a harm whose size is not.

What a strict elimination costs

The price of these protocols is rarely written about. Yet it is measurable.

Then there is an effect almost nobody plans for. In the Polish Hashimoto study body weight fell without weight loss being the goal. If you drop thirty food groups you almost inevitably eat less, and an unintended energy deficit can affect thyroid, cycle and exhaustion. More under the calorie myth: quality over quantity.

And then the part missing from almost every guide. The same nine people with rheumatoid arthritis were interviewed a year later. The phase they described as hardest was not the elimination. It was the reintroduction.

There is another price. Eating is social. A protocol that excludes you from every invitation costs you something that shows up in no lab value. And in some people caution tips over into a tense relationship with food, see understanding eating disorders.

And now you know why on this topic I talk more about the exit than about the entry.

What I look at first, before anything gets cut

You now know the weaknesses of elimination. Now the other side. So far this field holds only one large, randomised, blinded trial. It is not about leaving things out. It is about equipment.

How well supplied you are with omega 3 can be measured, see measuring your omega 3 index. Why plant and marine sources do not do the same job is covered under omega 3 from plants and animals.

Then there is a second list of things to drop, better documented than any lectin table.

Two older papers round out the picture. In 1991 the Lancet published the only randomised elimination trial with a control group and a full year of follow up that I know of: 53 people with rheumatoid arthritis, one group with a fasting phase and then a gluten free, initially vegan diet. Joint findings, morning stiffness and CRP improved, and the advantage held across the year. That diet was plant based, so almost the opposite of a meat heavy paleo approach.

And a pooled secondary analysis of two pilot studies on paleo in multiple sclerosis shows the pattern this field has to be measured against: within the paleo group functional scores improved, compared with the control group the difference stayed without significance. Better than before and better than the control group are two different statements.

What I observe in my consulting room goes beyond the study data, and I name it as an observation: people who run a food experiment with a clear start, a clear end and medical supervision come out knowing more about themselves. That is not documented evidence. If you would like to start with support, you will find the option to book an appointment below this article.

And with that you know why my first question is never what you should cut. It is what your immune system is currently getting in equipment and in rest.

Twelve questions I am asked about this most often

Where this topic leads next

Autoimmunity is rarely an isolated event. Inflammatory load, gut lining, mitochondria and hormonal axes hang together.