Endometriosis and nutrition: what the studies actually show

It is close to midnight. Your belly has been pulling since the afternoon, the hot water bottle has gone cold. On your phone there is a tab you have already closed three times today. Headline: these twelve foods you should avoid with endometriosis.

Milk is on the list. On the next page it says yoghurt is good. On the page after that it says broccoli is anti inflammatory, and in the forum someone writes that broccoli blows her up like a balloon.

Many women with endometriosis know this moment. And many also know what comes afterwards: the feeling of simply not doing it consistently enough.

I would like to take that thought away from you before the article even starts. Endometriosis does not arise because you eat wrongly, and it does not disappear because you eat correctly. That is not modesty, that is the evidence. And exactly for that reason I can tell you honestly here what nutrition might be able to do and what it cannot.

In endometriosis, nutrition is neither a side issue nor a therapy. It is a lever with a size that can be named. This article names it.

Four routes by which food reaches the abdominal cavity

The first question in the consultation is usually a good one: why on earth should my breakfast have anything to do with tissue that sits in the pelvis?

The answer is anatomy and biochemistry. There are four described routes, and they are supported to very different degrees. That distinction is the point of this section.

Route one: the fluid in which everything floats. In the abdominal cavity there is a thin film of fluid between the organs. In endometriosis this film is measurably different in composition. A review of the immunological findings describes how the scavenger cells of the peritoneum are more numerous and clearly more active than in women without endometriosis, and how they help steer the production of messenger substances, prostaglandins and growth factors there [Mechanism Review]. Inflammation in this condition is therefore not a buzzword but a state you can draw off and measure.

Route two: the lesion that stokes its own fire. Endometriotic cells produce aromatase, the enzyme that converts precursors into oestrogen. Prostaglandin E2 stimulates this aromatase. The oestrogen produced in turn stimulates cyclooxygenase-2, which makes more prostaglandin E2 [In vitro]. Picture a stove that keeps adding its own coal. And because prostaglandins are formed from fatty acids, the idea of intervening there through the fat quality of food is not far fetched.

Route three: the circuit through liver and gut. Oestrogen is packaged in the liver and released into the gut with the bile. There, bacterial enzymes, above all beta glucuronidase, can unwrap that package again. The released oestrogen can be taken up once more. How much really leaves the body therefore also depends on stool volume and on the composition of the gut flora. The details of phase one and phase two in the liver I have written up in a separate article, here it stays at these two sentences.

The most discussed hypothesis in this field also runs through the gut, the estrobolome. I deliberately call it a hypothesis: so far there is no intervention study showing that changing the microbiome improves endometriosis symptoms [Systematic Review]. The current state is written up under prebiotics.

Route four: the pain itself. Pain in endometriosis does not come only from the lesion. It also comes from a nervous system that has learned to pass it on more strongly.

And now you know why nutrition is up for discussion in this condition at all. The next question is the harder one: what of it holds up in a study?

Omega-3: the best signal and the weakest study

If a single nutrient is a star in the endometriosis context, it is this one. And with this one of all things, observation and experiment drift furthest apart.

Let us start with the good news. It comes from the Nurses Health Study II, one of the largest nutrition cohorts in the world. Tens of thousands of nurses in the United States have been filling in a food questionnaire there every few years for decades, and their diagnoses are tracked alongside.

Now the catch, and it is a big one. This cohort measures who receives a diagnosis over time. It does not measure whether a woman who already has the diagnosis has less pain with the same food. Those are two different questions asked of two different groups of people.

For the second question there is exactly one cleanly conducted study. And it turns out differently from what the cohort would lead you to expect.

That leaves the question of the source. And here I deliberately depart from almost all other guides.

And now you know why I sound more cautious about omega-3 than half the internet. The mechanism is plausible, the proof in humans is missing.

Red meat, trans fats, dairy: what the large cohorts show

There is one sentence that comes up in almost every conversation: I have already cut out dairy. And then I have to say that the data point in the other direction.

That is not a comfortable moment, but an important one. It shows how far online advice and the study evidence have drifted apart. Let us go through the big cohort findings one by one.

A second analysis of the same cohort also found a lower later risk for high dairy consumption during the school years [Cohort, n=32,868]. I quote that with the handbrake on: a recall questionnaire spanning decades is the weakest form of nutritional data there is.

The pooled version of these findings comes from a meta-analysis of observational studies.

The 82 percent heterogeneity for meat is a warning sign and belongs said out loud. It means that the individual studies contradict each other considerably, and that the average is correspondingly soft.

And then there are the vegetables, where it gets genuinely uncomfortable.

The blanket advice to eat "plenty of fibre" is therefore not covered in endometriosis. In a further analysis of the same cohort over 24 years, vegetable fibre in the highest fifth went along with RR 1.13 and cruciferous fibre with RR 1.17 [Cohort, n=81,961]. What fibre can do in principle and where the myths start is written up in the fibre article. Here only this one point counts: the amount is not the goal, tolerance is.

One last finding from the same analysis, because it is practically relevant: the highest fifth of the glycaemic index went along with a 12 percent higher risk, while glycaemic load showed no association [Cohort, n=81,961]. That the same number of calories can arrive very differently in the body I have described in more detail using another example.

And now you know why I am not giving you a list. Not out of caution, but because the existing lists contradict what has been measured.

Vitamin D and antioxidants: when cell and mouse convince and the human does not

There is a pattern in medicine you have to see once in order to recognise it everywhere afterwards. Vitamin D in endometriosis is the textbook example: in cell culture it looks excellent, in the animal model too. And then the human arrives, and the numbers become small and uncertain.

Fairness requires saying that there is also a positive randomised trial. It included 60 patients between 18 and 40 years of age, double blind, over twelve weeks. Pelvic pain fell by 1.12 points on the scale, confidence interval minus 2.1 to minus 0.09. High sensitivity CRP fell by 0.64 milligrams per litre, total antioxidant capacity rose. Other symptoms remained unchanged [RCT, n=60].

And the same fairness requires: the SAGE trial from the previous section found no advantage over placebo for its vitamin D arm. Two randomised studies, two results. That is where things stand, and I am not settling it here.

With the antioxidants the case is more interesting. There we have the only study in the whole field that measured an effect exactly where the lesions sit.

And now the two sentences that sort the field best. I take them from two meta-analyses that differ in their assessment.

The authors themselves write that these average effects are highly uncertain and must not be transferred to specific clinical situations [Meta-analysis]. In the subgroup analysis, melatonin showed the most consistent possible reduction in pain.

And the second meta-analysis delivers what may be the most important single sentence of this article: of eleven randomised studies with 716 women, the studies with a low risk of bias of all things showed no significant results against placebo [Meta-analysis, k=11, n=716].

A third meta-analysis turns out more optimistic. It found a positive effect on pelvic pain for vitamin E, with or without vitamin C, but no significant one for vitamin D [Meta-analysis, k=13, n=589]. Reading both assessments side by side is more honest than picking one.

The framing sentence for this whole section, and it matters to me: supplements are not a substitute for gynaecological treatment. They are not a reason to postpone an operation, to stop a hormonal therapy or to reduce pain medication. A systematic review with GRADE assessment rated all included nutrient studies as being of low to very low quality [Systematic Review]. That is the ground we are standing on here.

And now you know why I rarely show enthusiasm about supplements in this field and instead often ask for the measured baseline value.

Gluten free: the most quoted finding and what it does not show

There is one single study behind almost every article on this topic. You have probably already read its number without knowing its name: 75 percent.

Let us look at it closely. Not to tear it down, but to place it correctly.

Now for the interpretation, and for that we need a number from further up. In the SAGE trial, the placebo arm improved by 1.6 points on the pain scale. Those women received a capsule without an active ingredient.

If an empty capsule over six months produces that effect, what does a dietary change produce that is visible every day, effortful and loaded with hope? Without a comparison group this question cannot be answered. With a diet there is no blinding. Every participant knows every day that she is doing something.

And then comes the second study, which practically never appears in the popular coverage.

How do these two findings fit together? My reading, explicitly as a reading: they measure different things. The cohort measures whether gluten has anything to do with the development. The observational study measures whether women feel better after avoiding it. It is quite possible that both are true, and that the reason for the second is not the gluten but the gut.

Because wheat brings more with it than gluten. It brings fructans, meaning a FODMAP group. Anyone eating gluten free almost automatically eats fewer fructans. And anyone with abdominal symptoms notices that. What else is discussed behind this, from zonulin to the amylase trypsin inhibitors, is written up in the article on gluten without coeliac disease.

And now you know why I neither wave gluten free away nor nod enthusiastically. Instead I ask: what exactly has got better? If the answer is bloating and bowel movements, we are at the next section.

The gut as a second construction site: FODMAP, bloating, irritable bowel

There is a belly that hurts, and there is a belly that bloats. Many women with endometriosis have both, and in everyday life the difference blurs.

For the nutrition question, though, it is decisive. Because this is exactly where the best numbers in the whole field lie.

What is remarkable about the same work is which features were linked with endometriosis: pain during sex, referred pain, a cycle dependent worsening of gut symptoms and a family history of endometriosis. So if your gut changes in the rhythm of your menstrual cycle, that is no coincidence and no imagination. Why digestion and cycle are coupled at all is written up in the article on cycle and digestion.

The second study is methodologically the most important in the entire field, because it is the only one that puts a dietary intervention against a real control group.

Two distinctions on that, then I move on.

First: low FODMAP is not a permanent state. It is a three phase procedure with an elimination phase, reintroduction and an individual long term form. Anyone who stops in phase one impoverishes their diet for no reason. The procedure is written up in the FODMAP article.

Second: bloating is not the same as the endometriosis belly. Distinguishing between bowel involvement of endometriosis and an accompanying irritable bowel syndrome is a medical task. It is covered elsewhere. A nutrition trial does not replace it.

And now you know why in the consultation I first separate out which pain we are talking about. The answer decides whether nutrition is the right tool at all.

Inflammatory markers in endometriosis: what is measurable and what it says

A question that deserves an honest answer: can I have my inflammation measured so I can see whether the change is doing anything? The short answer is: measuring yes, drawing conclusions only to a limited degree. The reason is more interesting than it sounds.

That answers the search query "endometriosis inflammatory markers", and differently from how most pages do it. There is no blood value that tells you how your endometriosis is doing right now.

What does exist are markers that move under a change of diet. In the randomised vitamin D study, high sensitivity CRP fell by 0.64 milligrams per litre [RCT, n=60]. In the antioxidant study, RANTES, interleukin-6 and MCP-1 fell in the peritoneal fluid [RCT, n=59]. Those are real changes. They just say nothing about how you are doing.

What silent inflammation actually is, why it shows up in everyday life as fatigue rather than fever and which values are usefully read together I have described in a separate article.

And now you know why I am sparing with laboratory requests on this question and instead ask about the course over time.

What the guidelines say, and why they are cautious

At this point many guides fall silent. The guidelines are either not mentioned at all or dismissed in half a sentence. I do the opposite and write them out in full, because their reasoning is more interesting than their verdict.

Why are both so cautious? Not out of disinterest. But for four understandable reasons, and I find every single one of them justified.

Reason one: the studies are small and short. A systematic review found nine human studies of dietary interventions in endometriosis, and every single one examined a different intervention [Systematic Review]. Nine studies, nine diets. You cannot add them up. That is why there is no endometriosis diet, only nine individual attempts.

Reason two: diets cannot be blinded. With a tablet, nobody knows who is getting the active substance. With a change of diet, every participant knows every day. And we have seen above how large the placebo arm can turn out in this condition.

Reason three: the quality is consistently low. A systematic review with GRADE assessment from 2020 rated all included studies as being of low to very low quality [Systematic Review]. That is not an opinion, it is a methodological judgement against set criteria.

Reason four, and this is the most important: special diets carry their own risks. Undernutrition, social withdrawal and a disturbed relationship with food are real side effects that do not appear on a diet's package insert. When the benefit is unclear, the possible harm weighs more heavily.

Two numbers for context, because they explain why so many women start out on their own in the first place. Endometriosis affects about ten percent of women of childbearing age [Systematic Review]. And in an analysis of German data, five years was identified as the cut off between a short and a long diagnostic delay, with an early symptom onset being the most important risk factor for a long delay [Observational, cross sectional].

Five years. Nobody sits still during that time. You read, you try things, you cut things out. That is not an accusation aimed at gynaecology, that is a finding about the system.

One last point about honesty. In a meta-analysis with 32 studies and 103,204 women, the Mediterranean pattern showed clear associations for pregnancy complications, while the same work reports no robust result for endometriosis [Meta-analysis, k=32, n=103,204]. I do not transfer the one number to the other field.

And now you know why I quote the guidelines instead of passing over them. They are not the opponent of this article, they are its foundation.

Three months, one diary, one variable

If you have read this far, you know the evidence better than most pages that want to sell you a list. That leaves the question you actually brought with you: what do I do on Monday?

My answer is unspectacular, and that is exactly its strength. You run a trial. But one you learn something from afterwards.

Why so strict about the one variable? Because I keep seeing women who after eighteen months have fourteen food groups cut out and not a single answer. Every cut came from an article, none from a trial.

This point matters more to me than all the study numbers above it. A condition that takes control away from you anyway makes controlling tempting. At some point it tips over, and then nutrition is no longer your tool but your supervisor.

The German guideline also names undernutrition explicitly as a risk of special diets without indication. And if you nodded while reading the list above: that is not a character flaw and not a failure, it is a known pattern with good ways out. How disordered eating develops and how to recognise it early I have described in a separate article. It is worth reading that before the list grows long.

Where nutrition ends and the environmental question begins

That leaves the part for which some patients get this link from me. Because the question that often remains after three months of diary is: and what if nothing has changed?

Then I keep looking. Not instead of gynaecology, but alongside it.

The reason for that is mechanistic, and I will name the strength of the evidence precisely. There are substances from the environment that bind to oestrogen receptors. And here the labelling of the levels of evidence matters more than anywhere else. For zearalenone, a mould toxin on grain, oestrogenic activity is well described in cell culture and in the animal model, with observational studies in humans on top. For cadmium, the discussion as a metalloestrogen likewise comes mostly from cell culture and animal model, supplemented by observational data. For endometriosis there is so far no human intervention study on either substance, meaning no study in which exposure was lowered and the course measured afterwards. That level is missing, and that belongs said. I have written about both separately: xenoestrogens in everyday life and zearalenone as a mycoestrogen. Here it stays at these two sentences.

A negative finding explicitly belongs here, because it carries the honesty of this section. The first investigation of dietary pesticide residues in endometriosis found no association. In an analysis with 52,053 women and 956 confirmed cases, the highest compared with the lowest fifth of highly contaminated produce showed a hazard ratio of 0.94, confidence interval 0.73 to 1.23 [Cohort, n=52,053]. No signal, in any subgroup.

Why do I still look, when a course stays stubborn? Because the question of what keeps a body in an inflammatory state over the long run is legitimate in a condition shaped by inflammation. In standard care it is rarely asked, not because it would be nonsensical, but because there are more urgent things to do there.

And now you know why I do not sell a diet on this topic but recommend a diary. If you would like company while putting it into practice: below this article you will find the option to book an appointment.

Frequently asked questions

Where to read on