Understanding autoimmune diseases: why the immune system turns against your own body

First it was the thyroid. A test result with raised antibodies, a tablet in the morning, and at some point you had got used to it.

A few years later, your gut speaks up. Or your joints feel stiff in the morning. Or patches appear on your skin that no longer go away. And a new doctor says a sentence that sticks: this is autoimmune too.

Many people know this pattern. And almost all of them ask the same question: why me, and why again?

That question deserves an honest answer. Not a simple one, but one that shows this: behind very different diagnoses lies a shared pattern, and it has been researched well enough to understand it.

Before we start, there is a box that sits right at the top on purpose.

What autoimmunity is, and how your immune system learns to spare you

Imagine a city where a security team checks millions of faces every day. The vast majority belong to residents, a very few to burglars. The team has to tell the two apart reliably, for a lifetime.

Your immune system does exactly that. It attacks bacteria, viruses and degenerate cells while leaving your thyroid, gut, joints and nerves alone.

This ability is called tolerance. An autoimmune disease develops when this tolerance towards a specific tissue of your own is lost permanently. The immune system is then not simply too strong. It is looking at the wrong thing.

Organ-specific or systemic

In organ-specific diseases, the reaction is directed against one organ or cell type; in systemic diseases, many organs can be affected. The boundary is not sharp, but it makes sorting easier.

How common is all of this together? A very large analysis from the UK delivered a solid figure in 2023.

Many of these diseases share non-specific symptoms: exhaustion that sleep does not make up for, joint pain, skin changes, digestive problems. These symptoms are real, but they also fit iron deficiency, sleep disorders and much more. They are not enough for a self-diagnosis.

The first school: the thymus

Tolerance is not innate like eye colour. It is learned, first in the thymus, a small organ behind the breastbone.

That is where T cells mature, each with a randomly assembled recognition receptor. Cells that would recognise the body's own proteins are sorted out as far as possible. This is called central tolerance.

For this to work, young T cells in the thymus also need to see proteins from the thyroid, pancreas or adrenal gland. A gene called AIRE takes care of that. This is textbook knowledge, and a rare disease shows how important it is.

The second level: mediators throughout the body

No sorting process is perfect. That is why there is a second safeguard, peripheral tolerance. Its most important players are regulatory T cells, Treg for short. Think of them as mediators who calm other immune cells down when those cells react to something that is not a threat.

For discoveries about this second level, the Nobel Assembly at Karolinska Institutet awarded the Nobel Prize in Physiology or Medicine to Brunkow, Ramsdell and Sakaguchi on 6 October 2025. The prize recognised the description of regulatory T cells and the discovery of the Foxp3 gene (press release). According to a review by the group around Sakaguchi, Treg are indispensable for maintaining tolerance to the body's own antigens (Sakaguchi 2008, PMID: 18510923).

One of the brakes that Treg use is called CTLA-4. With this brake, quantity matters. Kuehn and colleagues described people from four unrelated families in whom one of two CTLA4 gene copies was altered. The reduced amount of CTLA-4 disrupted the regulatory T cells and led to immune cells infiltrating organs, while mice with only one intact gene copy remained unremarkable (Kuehn 2014, PMID: 25213377). So animal findings cannot be transferred one to one. The name CTLA-4 is worth remembering.

And now you know why autoimmunity is better understood as the tipping of a balance that your body actively maintains every day, rather than as an error out of nowhere.

Shared risk genes: why one threshold applies to many diseases

Your mother has rheumatism. Your sister has Hashimoto's. Your cousin has had type 1 diabetes since childhood. And you are sitting at the table with a fresh coeliac diagnosis, thinking: this cannot be a coincidence.

Often it is not pure chance. But what is passed on in such families is rarely a specific disease; it is more often a lower threshold.

Imagine a dam that is a little lower in some people. A low dam alone floods nothing. Only when rain comes along, in other words a trigger, does water spill over. Where it spills over, in the thyroid, gut or joints, depends on further factors.

The core evidence: almost half of the variants cross disease boundaries

The three best-known shared genes all have to do with signal strength and brakes in the immune system.

HLA: which fragments are put on display

HLA molecules are shop windows on the cell surface in which small protein fragments are shown to T cells. Which fragments fit well depends on your HLA variant. That is why HLA genes are among the strongest genetic factors in many autoimmune diseases. This is textbook knowledge.

One example: type 1 diabetes and coeliac disease share HLA class II genes. And in rheumatoid arthritis, certain HLA-DR variants, the so-called shared epitope, play a central role, which we will come back to when we look at smoking.

PTPN22: an amplifier that many healthy people carry too

CTLA4: the same brake, three diseases

Psoriasis and inflammatory bowel disease also belong to this network. Ellinghaus and colleagues jointly analysed five inflammatory diseases, including psoriasis, Crohn's disease and ulcerative colitis, in more than 86,000 people and found 244 independent signals relevant to more than one disease. The co-occurrence was best explained by genes that influence several diseases at once (Ellinghaus 2016, PMID: 26974007).

What this means specifically for Hashimoto's, in other words what role HLA-DR, CTLA-4 and PTPN22 play there, is already covered in the Hashimoto's article. Here the focus is on the bigger picture.

An honest assessment belongs here too. Variants such as the one in PTPN22 also occur in many healthy people, and detecting them says little about your personal course. I therefore do not recommend genotyping to predict an autoimmune disease.

Why it clusters in families and within the same person

If genes lower a shared threshold, you would expect to see two things: people with one autoimmune disease develop a second one more often, and other autoimmune diseases also turn up in their families. Both are well documented.

The large UK analysis by Conrad shows how much the strength of this clustering varies. In type 1 diabetes starting in childhood, the incidence rate ratio was 28.4 for coeliac disease, 26.5 for Addison's disease, 13.3 for Hashimoto's and 6.7 for Graves' disease. Multiple sclerosis, by contrast, occurred particularly rarely together with other autoimmune diseases. That fits the genes: not every disease shares equally much with the others.

A systematic review with meta-analysis by Cárdenas-Roldán and colleagues found this familial pattern in all the diseases examined, namely rheumatoid arthritis, lupus, autoimmune thyroid disease, multiple sclerosis and type 1 diabetes. Autoimmune thyroid diseases clustered most frequently in such families (Cárdenas-Roldán 2013, PMID: 23497011).

And now you know why coeliac disease alongside type 1 diabetes, or a second diagnosis alongside Hashimoto's, is often not bad luck in a vacuum but fits a pattern with shared genes.

The environmental side: why genes alone do not explain the changes

May I ask you an uncomfortable question? If genes are so important, why do some autoimmune diseases change measurably within twenty years?

The genetic makeup of a population practically does not change in that time. What does change are infections, smoking habits, diet, daily rhythm, strain and the way diagnoses are made.

First, a sober look at the numbers. In the large UK analysis, the incidence of all 19 diseases together rose only slightly, with an incidence rate ratio of 1.04 (95% CI 1.00 to 1.09) for 2017 to 2019 compared with 2000 to 2002. Individual diseases changed considerably: coeliac disease 2.19, Sjögren's syndrome 2.09 and Graves' disease 2.07. Pernicious anaemia, on the other hand, fell to 0.79 and Hashimoto's to 0.81 (Conrad 2023, PMID: 37156255).

Based on these data, you cannot speak of a general explosion of autoimmune diseases. But there are shifts, and one marker points in a striking direction.

Not every environmental factor is equally well documented. So first, a map.

The Epstein-Barr virus and multiple sclerosis

Most people become infected with the Epstein-Barr virus at some point, often without noticing, sometimes as glandular fever. Only very few develop multiple sclerosis afterwards. For a long time, it was therefore unclear whether the virus plays any role at all.

How could a virus lead to nerve tissue being attacked? One explanation is called molecular mimicry. Imagine a police sketch that happens to match an innocent neighbour as well. The police are looking for the burglar and detain the neighbour along with him.

A third paper builds the bridge to the genes. In computational analyses by Harley and colleagues, almost half of the lupus risk loci were occupied by the EBV protein EBNA2, with similar patterns in MS, rheumatoid arthritis, inflammatory bowel disease, type 1 diabetes and coeliac disease (Harley 2018, PMID: 29662164). These are laboratory and computational data, not disease courses.

For other autoimmune diseases, the role of infections is not established this clearly. The Hashimoto's article, for example, classifies Yersinia and EBV in the thyroid as unproven.

Smoking and rheumatoid arthritis

The German MS guideline also addresses the topic. It sees indications, but no clear evidence, for an increased risk of developing the disease due to smoking. It does, however, state that smoking accelerates progression and recommends informing patients about smoking and passive smoking (DGN guideline MS 2026, recommendation C9).

Vitamin D: four studies, four steps

With vitamin D, each study answers a different question. Only together do they give an honest picture.

Step one is an association. Munger and colleagues compared vitamin D levels of 257 later MS cases from serum samples of more than 7 million US military personnel with controls, measured before symptoms began. In white participants, the odds ratio per 50 nmol/L higher 25-OH vitamin D was 0.59; in black and Hispanic participants, no significant association was found (Munger 2006, PMID: 17179460). This is an observation, not yet proof.

Step two is Mendelian randomisation: gene variants that lower the vitamin D level slightly for life serve as a natural experiment. Mokry and colleagues found, in up to 14,498 MS cases, an odds ratio of 2.0 (95% CI 1.7 to 2.5) for MS per genetically determined decrease of one standard deviation (Mokry 2015, PMID: 26305103). This supports a contributing cause, but does not show that supplements prevent MS.

Gut barrier and microbiome: plausible, but thinly documented

Hardly any topic is linked to autoimmunity in functional medicine as often as the gut. The idea is intuitive: if the intestinal lining becomes more permeable, more foreign substances could come into contact with immune cells. How well is this documented in humans? Mechanistically plausible, human studies thin, that is my honest assessment.

On top of that, there is a measurement problem. Ajamian and colleagues found that commercial zonulin tests mainly detected other proteins such as haptoglobin and complement C3, and advise caution in using serum zonulin as a barrier marker (Ajamian 2019, PMID: 30640940). How to put this into context in detail is covered in the article Leaky gut, intestinal permeability and zonulin, and the question of gluten without coeliac disease in Gluten and gliadin without coeliac disease.

The situation with the microbiome is similar. In the longitudinal TEDDY study with 10,913 stool samples from 783 children, the microbiome of the control children contained more genes for fermentation and short-chain fatty acids, without the differences being consistently tied to specific bacterial species (Vatanen 2018, PMID: 30356183). In new-onset, untreated rheumatoid arthritis, Prevotella copri was strikingly common, and mice colonised with it reacted more sensitively to chemically induced colitis (Scher 2013, PMID: 24192039).

These are associations and animal findings. No bacterium has been shown to be a trigger, and no recommendation for stool tests or probiotics for prevention follows from this.

Strain and trauma

Many people say that their autoimmune disease began after a particularly hard time. After a separation, a death, a year that was barely bearable. Is that imagination? The data say: probably not.

Early strain can leave traces too. In a retrospective cohort of 15,357 adults, 64 percent reported at least one adverse childhood experience. Hospital admissions for autoimmune diseases rose with the number of these experiences. From two adverse experiences onwards, the risk of hospital admission for rheumatic diseases was 100 percent higher (Dube 2009, PMID: 19188532). Childhood was assessed retrospectively, and only hospital admissions were recorded.

Sleep

A Taiwanese register cohort compared 84,996 adults with sleep disorders without sleep apnoea with the same number of controls. The adjusted hazard ratio for autoimmune diseases was 1.47 (1.41 to 1.53), and for lupus 1.81 (Hsiao 2015, PMID: 25669189). The direction may also be reversed, because an emerging disease could disturb sleep. This is thin evidence.

And now you know why genes alone do not tell the whole story. They can lower the threshold, and infections, smoking, strain and probably other factors can help decide whether and where it is crossed.

Why women are affected considerably more often

In the waiting room of a rheumatology practice or a thyroid clinic, a striking number of women are sitting. That is not a coincidence of appointment scheduling.

These figures come from the cohort by Conrad (PMID: 37156255). I deliberately do not use a frequently quoted figure of around 80 percent women, because I could not verify it in the primary literature I checked.

The difference also shows up in the ANA marker: 17.8 percent of women and 9.6 percent of men were positive in a US survey (Satoh 2012, PMID: 22237992).

A more alert immune system

A review by Klein and Flanagan summarises that the immune responses of women and men differ. Some differences persist throughout life, others mainly between puberty and menopause, which points to an interplay of genes and hormones (Klein 2016, PMID: 27546235).

You can picture it as an alarm system set to a more sensitive level. It reports burglars earlier and more reliably. But it also goes off more often when only the wind rattles the window.

The X chromosome counts too

For a long time, the focus was almost entirely on oestrogen. A rare chromosome variant in men shows that this is not enough.

Pregnancy as a natural experiment

How strongly the hormonal situation can influence the immune system becomes visible in pregnancy, when the immune system readjusts.

And now you know why the ratio between women and men probably does not depend on hormones alone, but likely also on the number of X chromosomes.

Autoantibodies: years before symptoms, and why positive does not mean ill

You are holding a test result. At the top it says: ANA positive. And you have spent the evening reading names of diseases that frighten you.

Take a breath. An autoantibody is an important signal, but not a verdict. There are two sides to this coin.

Side one: autoimmunity often begins long before symptoms

A small randomised trial with 76 relatives of people with type 1 diabetes suggests that this preliminary phase could in principle be influenced. Under the antibody teplizumab, the median time to diagnosis was 48.4 months, under placebo 24.4 months (Herold 2019, PMID: 31180194). This is specialised immunotherapy in trials, not self-help.

Side two: positive does not mean ill

For the thyroid, there is a 20-year observation on this. In the Whickham cohort of 2,779 adults, the odds ratio for later hypothyroidism was 8 in women with positive thyroid antibodies alone, and 25 in men. When a raised TSH and antibodies occurred together, it was 38 in women and 173 in men (Vanderpump 1995, PMID: 7641412). Antibodies without a functional disorder are therefore a risk signal that justifies monitoring. On their own, they are not a disease. What can measurably change in Hashimoto's antibodies is covered in Lowering Hashimoto's antibodies.

And now you know why a positive antibody should make you attentive but not panicked, and why good values do not answer every question.

What functional medicine can add, and where the evidence ends

After all the genes and triggers, almost the same question always follows: so what can I do myself? The question is justified. It deserves an answer that neither promises more than the data support nor leaves you with a shrug.

Diet: indications, but no reliable effects

A Cochrane review by Hagen and colleagues included 14 randomised trials and one controlled study with 837 people with rheumatoid arthritis. Overall effects could not be calculated; individual studies showed indications of less pain, but no improvement in physical function or morning stiffness, and more dropouts (Hagen 2009, PMID: 19160281). The randomised crossover trial ADIRA with 50 people with RA missed its primary endpoint: the main analysis showed no significant difference in disease activity, and only an unadjusted secondary analysis came out more favourably (Vadell 2020, PMID: 32055820).

The German MS guideline sees no studies of sufficient quality for specific diets such as intermittent fasting, ketogenic diets, or the Swank or Wahls diet, and recommends a balanced, heart-healthy diet. That is not a rejection of nutrition, but an honest state of knowledge. What paleo and the autoimmune protocol can offer is covered in Autoimmune protocol and paleo, and the evidence for the thyroid in Diet for Hashimoto's.

And once more, because it is so important: if coeliac disease is suspected, gluten is only removed after testing is complete, otherwise the diagnosis can become impossible.

Gut, nutrients, sleep and strain

And now you know why, when it comes to self-help, I distinguish so carefully between documented, plausible and tradition.

Why immunosuppressive therapy is often indispensable

Perhaps the word immunosuppression, methotrexate or cortisone makes you wary. That is understandable, and side effects deserve an honest discussion. But it is worth understanding what these therapies achieve.

There are two different logics. The first is replacement. In type 1 diabetes, the insulin-producing cells are destroyed, which is why insulin is essential for survival. In hypothyroidism due to Hashimoto's, thyroid hormone replaces what the gland no longer produces. The second logic is braking the inflammation. Immunosuppressants and biologics can dampen the attack itself before it permanently damages joints, kidneys or nerves.

What stands out is how much these recommendations aim at sparing the patient: sparing cortisone in lupus, treating early and in a targeted way in RA. Therapy is not the opposite of a holistic view, but often its prerequisite.

And now you know why immunotherapy and the functional view do not stand in opposition. One brakes the attack, the other takes care of the environment in which it takes place.

Frequently asked questions about autoimmune diseases

Where this topic leads next

Autoimmunity does not respect specialty boundaries. So here are twelve paths leading from this overview into the individual organs and questions.