Dioxins and PCBs: the trail that leads to endometriosis

You are sitting in the car after the appointment, looking at a piece of paper. On it is a word that splits your life into a before and an after. And in your head runs a question that almost nobody says out loud, because it feels accusing: Why me?

I cannot answer that question for you. Nobody can. But I can tell you why, with hormonal complaints, I ask about things that at first glance have nothing to do with the lower abdomen. About the house you live in. About your parents' occupations. About the milk you drank as a child.

These questions come from a story that began in 1993. It is well documented, it is astonishing, and it has a second part that almost nobody tells. I will tell you both parts. Including the one that ruins the nice version.

A colony of monkeys and a finding nobody was looking for

Picture a research group that wanted to study something else entirely. Not the uterus. Not period pain. It was about the general toxicology of a substance that was then considered the most poisonous molecule ever made by humans.

A colony of rhesus monkeys received TCDD in their feed for four years. Two dose levels, 5 ppt and 25 ppt, plus a control group. Then the feeding ended. The animals went on living. Ten years later, researchers looked into the abdominal cavity with a laparoscopy.

What they saw there, nobody had expected.

Pause for a moment at those numbers. Seven animals per group. Seven. If a single animal had come out differently, the percentage would have shifted by 14 points. That is not an argument against the observation. It is an argument for keeping it in mind at the right size.

Two years later the same group followed up and looked at the immune status of the animals. The idea behind it: if a foreign substance lets tissue grow in the abdominal cavity, then perhaps not directly, but through the immune system, which would otherwise clear that tissue away. Cytokines as the link between foreign substance and tissue. That way of thinking is still the field's load-bearing hypothesis today.

There is a sentence from a later review by the same author that explains why the matter made waves. The serum concentrations of the exposed monkeys with endometriosis were similar to, or even lower than, blood levels in the general human population of that time.

With that, the question was no longer academic. It suddenly read: does this concern all of us?

And now you know why I ask about the exposure history. Not because I suspect a cause there. Because this line of research has shown that there is a second entrance into the system.

What dioxins and PCBs actually are

Before we go further, you need a picture of these substances. Otherwise everything that follows stays abstract.

Dioxins are not a product. Nobody wanted to make them. They arise as a side effect, whenever organic material is heated together with chlorine. When waste is burned. In metal processing. In the past as a contaminant in chlorine chemistry. Technically the term covers two families, the polychlorinated dibenzo-p-dioxins and the polychlorinated dibenzofurans, usually abbreviated in reports as PCDD and PCDF.

Polychlorinated biphenyls, PCBs for short, are the opposite. They were made deliberately, over decades, in large amounts. As insulating oil in transformers, as plasticisers, in sealants, lacquers and paints. They were non-flammable, chemically inert and cheap. Exactly the properties that made them attractive back then make them a problem today. In Germany their production has been banned since the 1980s.

A subset of the PCBs, the so-called dioxin-like PCBs, has a flat molecular shape and behaves in the body like a dioxin. The other subset, the non-dioxin-like PCBs, takes different routes. This distinction sounds like a detail. In the studies that follow, it decides the result.

One more term you will meet in every report: TEQ, the toxic equivalent. There are hundreds of individual compounds in these families, and they differ in potency. So that they can be added up, each is given a factor relative to the strongest compound. That strongest compound is TCDD, written out as 2,3,7,8-tetrachlorodibenzo-p-dioxin, with the factor 1. A figure in picograms TEQ is therefore a conversion into a common currency.

And because I know that someone will want to compare figures from different reports: do not do it unchecked. There are TEQ factors from 1998 and from 2005, plus values from a bioassay called CALUX. Those are three different yardsticks. And picograms per kilogram per day is something other than picograms per kilogram per week.

And now you know why these substances turn up in an article about hormones, even though chemically they have nothing to do with hormones.

The glitch in the story that has to be told

Here it gets uncomfortable. And here an honest text parts ways with one that only wants to convince.

Eight years after the first publication, the same working group looked more closely once more. This time not into the abdominal cavity but into the blood. What was actually in these animals, thirteen years after the feeding had ended?

I find this part of the story more important than the first. Not because it devalues the original observation. Because it shows how science actually works. A sensational finding, a decade of thinking, then the realisation that the starting position was more complicated than assumed.

A third paper from the same year looked at immune cells. Thirteen years after the feeding ended, white blood cells from the exposed animals produced more TNF-alpha after stimulation, a signalling substance of inflammation, and their destructive activity against certain target cells was reduced. Other messenger substances were unchanged. Rier SE, Coe CL, Lemieux AM et al. Toxicol Sci. 2001;60(2):327-337. PMID: 11248145 · DOI: 10.1093/toxsci/60.2.327 [In vivo, primate]

Thirteen years. That is the point where I start to ponder in the consulting room. Not because I derive anything from it that the data cannot carry. But because it shakes the idea that an exposure is over when the exposure is over.

There is a second obvious explanation that has been tested and not confirmed. It goes: some people break these substances down less well, and that is why they fall ill. The same author systematically reviewed ten studies in 2006 on gene variants of foreign substance detoxification, among them CYP1A1, CYP2E1, EPHX1, AHR, ARNT, AHRR, NAT1 and NAT2. Almost none of the variants examined showed a link to endometriosis. For one variant of the CYP1A1 gene a roughly 40 percent increased risk was found, without consistent confirmation in other work.

And now you know why I am careful in this field. The nice story has a postscript. Telling it without the postscript means telling it wrong.

The aryl hydrocarbon receptor, a second switch

If you have read this far, you have probably already asked yourself: how is that even supposed to work? A substance that has nothing to do with estrogen intervenes in a hormone-dependent disease. Where is the lever?

It is called the aryl hydrocarbon receptor, in the literature usually AhR. Picture it as the smoke detector of the cell. It hangs inside the cell and waits. If a foreign molecule with the right flat shape comes past, it docks on. The smoke detector travels into the cell nucleus and switches genes on there, above all genes for enzymes meant to convert foreign substances.

At first this is a sensible arrangement. The body recognises something foreign and prepares its breakdown. The problem arises when the smoke detector no longer switches off, because the substance stays. Then a programme designed for short assignments runs permanently.

So what happens when the smoke detector is permanently on? A second piece of work on human endometrial cells from Winnipeg in Canada examined exactly that. Dioxin-like compounds activated the aryl hydrocarbon receptor in a time-, concentration- and substance-dependent way. What was remarkable was what did not happen: the estrogenic processes in these cells were not disturbed by it. Independently of binding to the receptor, the migratory capacity of the cells rose.

That is the actual evidence for the sentence that dioxins are not classic xenoestrogens. They do not need the estrogen receptor.

Aromatase and the progesterone receptor are exactly the settings that are already altered in endometriosis. That makes the observation interesting. It does not make it proof. The rat study used partly very high doses, it dealt with hexachlorobenzene rather than dioxin itself, and rats are not women.

One paper that combined exposure modelling, meta-analysis and gene analysis found five genes influenced jointly by PCBs, bisphenols and phthalates: CYP19A1, EGFR, ESR2, FOS and IGF1. CYP19A1 is the gene for aromatase, the enzyme that produces estrogen in the tissue itself. That is the most direct molecular bridge between an environmental substance and local estrogen production in a lesion. It is bioinformatic, not clinical. Roy D, Morgan M, Yoo C et al. Int J Mol Sci. 2015;16(10):25285-25322. PMID: 26512648 · DOI: 10.3390/ijms161025285 [Bioinformatics, meta-analysis]

And now you know why I do not say that dioxins work like estrogens. They take a different entrance, and that entrance leads through inflammation.

What the human data can show and what it cannot

Now comes the part that really counts. Everything so far was monkeys, mice, rats and cells. Humans are none of those.

The question is: is there a population in which this can be tested? You cannot give women dioxin. So it takes a sad coincidence. And there was one.

On 10 July 1976 a reactor exploded at a chemical plant near Seveso in northern Italy. A cloud containing large amounts of TCDD settled over the surroundings. Years later the Seveso Women's Health Study grew out of this disaster, the best documented cohort with high TCDD exposure that exists.

Before the study proper, a question arose that somebody had to calculate: were the women in Seveso exposed at a level comparable to the monkeys of 1993 at all? A toxicokinetic analysis from Berkeley concluded that the cumulative exposure of the most affected residents was in every case above the exposure of both monkey groups. That made Seveso the decisive test cohort.

So that nobody draws the wrong conclusion: this cohort has indeed delivered clear results, just in a different place. In follow-up to 2008, 66 of 833 women had a cancer diagnosis. Per tenfold rise in serum TCDD, the adjusted hazard ratio for all cancers combined was 1.80, with a 95 percent confidence interval from 1.29 to 2.52. For breast cancer it was 1.44 and not significant.

That matters for context. TCDD is toxicologically not harmless. The open question is only whether endometriosis of all things is its target organ.

So that this figure does not land wrongly, one clear sentence about it: it stands here purely to place the substance group, and not to worry you. It comes from the cohort with the highest TCDD exposure ever documented and cannot be transferred to everyday exposure today. Nothing follows from it, neither an examination I would offer you here nor a need for you to act. Warner M, Mocarelli P, Samuels S, Needham L, Brambilla P, Eskenazi B. Environ Health Perspect. 2011;119(12):1700-1705. PMID: 21810551 · DOI: 10.1289/ehp.1103720 [Cohort, n=833]

The Roman work by Maria Grazia Porpora and colleagues at Sapienza University is the strongest positive human study there is. For the dioxin-like PCB 118 the odds ratio was 3.79 with a 95 percent confidence interval from 1.61 to 8.91, for PCB 153 it was 4.88 with an interval from 2.01 to 11.0, and for the sum of dioxin-like and non-dioxin-like PCBs it was 5.63 with an interval from 2.25 to 14.10.

And in the same data set, of all things the total TEQ from dioxins, furans and dioxin-like PCBs was not linked with endometriosis. So precisely the measure that expresses the actual dioxin burden. This nuance is missing online across the board, and it belongs here.

The counter-check comes from the US state of Washington. Britton Trabert and colleagues examined 20 PCB congeners in 251 surgically confirmed cases and 538 age-matched controls. Individual congeners even fell below the null value, for example PCB 196 with an odds ratio of 0.4. The authors' conclusion: non-coplanar PCBs do not contribute meaningfully to endometriosis risk in the exposure range observed there.

A smaller Roman paper closed the loop back to immune status. In women with endometriosis, the activity of natural killer cells was reduced, as were IL-1-beta and IL-12. Total PCB concentration was 330 versus 160 nanograms per gram of fat in the controls, DDE 770 versus 310. In the test tube, PCBs and DDE lowered the activity of these cells. Quaranta MG, Porpora MG, Mattioli B et al. Life Sci. 2006;79(5):491-498. PMID: 16499933 · DOI: 10.1016/j.lfs.2006.01.026 [Case-control, small sample, plus in vitro]

That is a small study, it is an association, and the caveat from a moment ago applies here too. It is interesting all the same, because it carries the same line of thought as the monkey colony of 1995. Immune status as the link. If you want to know how inflammation and metabolism interlock in women, you will find it in insulin resistance and female hormones.

And now you know why I do not give a short answer to the question about proof. There is none.

Two meta-analyses, two shades

When individual studies contradict each other, you add them up. That is the point of a meta-analysis. It is no magic wand, but it sorts things.

On environmental substances and endometriosis there are two larger pieces of work of this kind. They point in similar directions and arrive at differently strong numbers. Why, I will tell you in a moment.

A word on heterogeneity, because otherwise the term says nothing. I squared describes how strongly the individual results differ from one another, beyond chance. 63 percent means: the studies are not telling the same story. In the subgroup with the strictest design this value fell to 30 percent, and there the effect was 1.51 instead of 1.96.

There are two further review papers that round off the picture. A systematic screening of 27 papers on endocrine disruptors, microRNA and endometriosis placed dioxins, organochlorine pesticides and PCBs as the better documented group, while bisphenol A and phthalates delivered contradictory results. The signalling pathways involved were NF-kappa-B, MAPK and Wnt/beta-catenin. This work does not add up, it sorts, and is therefore weaker than the two meta-analyses. Chandrakanth A, Firdous S, Vasantharekha R et al. Reprod Sci. 2024;31(4):932-941. PMID: 38036864 · DOI: 10.1007/s43032-023-01412-8 [Systematic review, k=27]

The highest available tier is an umbrella review from 2024 that summarised 52 systematic reviews with a total of 759 meta-analyses on plastic-associated chemicals. There, PCBs in the general population are linked among other things with endometriosis, and also with type 2 diabetes, cardiovascular disease and breast cancer. The postscript belongs with it: 78 percent of the included meta-analyses came from reviews of merely moderate methodological quality. Symeonides C, Aromataris E, Mulders Y et al. Ann Glob Health. 2024;90(1):52. PMID: 39183960 · DOI: 10.5334/aogh.4459 [Umbrella review]

And now you know why two serious papers on the same topic deliver different numbers. They selected with different degrees of strictness.

The good news that is rarely told

If you have read this far, you may be feeling uneasy. So now comes the part that environmental articles like to leave out, because it takes away the drama.

Exposure has fallen sharply. Not a little. Sharply. And that is not a claim but documented across several independent measurement series.

The French figure comes from the second national Total Diet Study. 583 ready-to-eat food samples were analysed and combined with national consumption data. Compared with the previous assessment, exposure fell by a factor of 3.2 for PCDDs, PCDFs and dioxin-like PCBs together. Fewer than 4 percent of the population exceeded the health-based guidance value.

From Ireland comes a second independent confirmation. There the mean intake of total TEQ was 0.3 picograms per kilogram of body weight per day, and 1 picogram at the 95th percentile. All values were below the health-based guidance values that applied at the time. Tlustos C, Anderson W, Flynn A, Pratt I. Food Addit Contam Part A. 2014;31(6):1100-1113. PMID: 24645880 · DOI: 10.1080/19440049.2014.905713 [Exposure estimate]

The German figure comes from the Bavarian State Office for Health and Food Safety, which has examined more than 1,000 samples of bulk dairy milk since 1989. The average dioxin content has fallen by around 94 percent. The second part of this statement matters: for about ten years, exposure has been stagnating at this low level. The easy road has been travelled.

Where does exposure come from today in concrete terms? The best breakdown comes from the same Flemish survey with several thousand participants. Fish and seafood contributed 25 to 43 percent of total intake, added fats 22 to 25 percent, dairy products 17 to 20 percent.

Two qualifications on that, because I do not want to make numbers prettier than they are. The data come from 2007 and were collected with a bioassay that gives different numbers than mass spectrometry. They show the distribution of sources, not today's exposure level. The distribution itself has remained plausible, because the physics does not change: these substances sit in fat.

And now you know why I do not leave this section out, even though it takes away the drama. A text about the environment that only points downwards is not a good text.

What follows in practice, and where avoidance runs into limits

Let us come to the question you probably had from the start. What do I do with this now?

The honest answer begins with a limitation. Most of what is in your body got there over decades. Part of it before you were born. What you do differently from tomorrow changes that stock only slowly.

That is no reason to do nothing. It is a reason to keep expectations of your own action realistic.

When you are trying to conceive

For many women, endometriosis and trying to conceive belong together, and this section matters to me in particular. Here a factor plays a part that no environmental measure influences: time.

If you have been trying to conceive without success for some time, that belongs in a gynaecological and, depending on the situation, reproductive medicine assessment, and promptly. Nothing in this article is a reason to wait with that until you have sorted out your surroundings. Both run side by side, not one after the other.

And I want to be very clear at this point: I cannot tell you that anything will change for your fertility if you change your everyday life, and I will not hold it out as a prospect either. There is no intervention study on this question. What I can say is more modest. The questions about housing, occupation and exposure history cost little time, they do not stand in the way of an assessment, and sometimes they bring something to light that can be addressed. They do not replace the assessment.

And what about an ongoing treatment?

A large part of endometriosis treatment runs through prescription medicines. These include combined hormonal contraceptives, colloquially the pill, progestin-only preparations, in Germany above all the substance dienogest for endometriosis, in certain situations GnRH analogues, along with pain treatment and, where indicated, an operation. All of these are prescription medicines. They require a medical examination, an indication and a prescription.

Each of them has a risk side of its own, and that belongs next to the name. Under combined hormonal contraceptives the risk of thrombosis and embolism can be raised, more clearly with smoking, with higher age, with obesity and with certain clotting disorders. Under progestins, breakthrough bleeding, mood changes and a drop in libido can occur. Under GnRH analogues bone density can decrease, which is why they are usually given for a limited period and often together with what is called add-back therapy.

And one point that is rarely mentioned: in Germany the combined pill is mostly prescribed for endometriosis outside its licence, that is, off label. This means separate counselling, a particular duty of medical justification and, as a rule, no reimbursement by statutory health insurance. Dienogest, by contrast, is licensed for this indication.

Which of these might be right for you, which contraindications apply in your case and what dose is appropriate is settled by the doctor treating you. That is why doses deliberately do not appear in this text.

So one clear sentence on that: nothing in this article implies that you should stop, lower the dose of or pause any of these medicines. Every adjustment of a medication belongs in a medical conversation and needs medical supervision, even when it seems small to you. If you are unhappy with what you are taking, that is a good reason for a conversation and not a reason to go it alone. The same applies to thyroid hormones, to metformin and to hormone replacement therapy, if you are receiving any of them alongside.

And one more clarification, because this topic is often presented crookedly online: bioidentical hormones are prescription medicines too. They are not gentle self-help, they need an indication, a medical prescription and follow-up. Doses therefore do not belong in an article, but in a treatment.

Fish: why I draw a distinction here

Fish appears in the exposure data as the largest single source of dioxin-like compounds, with 25 to 43 percent of total intake in the Flemish survey. That is why it appears in this section.

With large predatory fish a second point comes on top, one that has nothing to do with dioxins: mercury content. It concerns above all long-lived predators at the end of the food chain, that is shark, swordfish, marlin, pike and large tuna. From those I advise against. Small, short-lived fish such as sardine, anchovy or herring sit lower down the chain and carry considerably less mercury. Why I look so closely at metals is set out in heavy metals in pregnancy and childhood.

And now the counter-argument that belongs here, because my reticence is not the official recommendation. The German Nutrition Society still advises regular sea fish, in the order of about one portion per week. The German Federal Institute for Risk Assessment recommends avoiding large predatory fish in pregnancy, not avoiding fish altogether. The reason for that carries weight. Sea fish is one of the most important carriers of iodine in the German diet, and iodine matters in pregnancy for the development of the child's brain. On top of that come the long-chain omega-3 fatty acids.

So if you leave fish out, iodine and these fatty acids should be covered another way. Please do not decide that on the basis of this article, and certainly not if you are pregnant or would like to become pregnant. That question belongs with your gynaecologist or your midwife.

Breastfeeding: please read this paragraph to the end

Fat-soluble pollutants pass to the child through breast milk. That is not disputed, and for the mother, breastfeeding is indeed a route of excretion. That is exactly why breastfeeding appears in the half-life formula at all.

And the National Breastfeeding Committee at the German Federal Institute for Risk Assessment, together with the gynaecological and paediatric professional societies, still advises breastfeeding, explicitly in view of this pollutant burden too. Please keep this sentence, if you forget the rest.

A review from the predecessor institute of today's German Federal Institute for Risk Assessment in Berlin frames it like this: the restrictions on production, use and release were successful, visible in a downward trend in breast milk and blood lipids. Transfer before birth is considered more significant for the child's physical development and cognitive function than exposure after birth through breast milk. And the breastfeeding period makes up only about 0.6 percent of the expected lifespan. Przyrembel H, Heinrich-Hirsch B, Vieth B. Adv Exp Med Biol. 2000;478:307-325. PMID: 11065082 · DOI: 10.1007/0-306-46830-1_27 [Review]

For the sake of honesty, the sentence the same paper writes in the same breath belongs here too: intake through breast milk in industrialised countries is ten to a hundred times higher than the tolerable daily intake the WHO derived in 1998. The paper dates from 2000, and exposure has fallen considerably since. The weighing-up still comes out in favour of breastfeeding, because transfer before birth is considered more significant and because the breastfeeding period makes up only this very small part of the lifespan. The authors close explicitly by saying that carefully conducted long-term cohorts have yet to deliver the final answer.

This weighing-up does not belong in an article, and certainly not in a night with a phone in your hand. It belongs in a conversation with your midwife, your paediatrician or your gynaecologist.

Why I usually do not order dioxin blood tests

Technically they are possible. In everyday practice they are rarely useful. The analysis needs high-resolution mass spectrometry and a lot of sample material, it is laborious and hardly available outside specialised laboratories.

The more important reason is a different one. A measurement is useful when it changes a decision. Here it changes none. There is no treatment whose use depends on this value, and no study showing that lowering the value influences the course of endometriosis. What remains is a number that unsettles and opens nothing.

For other groups of substances it looks different, and the distinction matters. What can sensibly be measured with heavy metals and what cannot is covered in measuring heavy metals in blood or urine.

What the guidelines say, and why so little

I am occasionally asked in my consultations why this is rarely mentioned at the gynaecologist's practice. The answer is less spectacular than many expect, and it is not a reproach to anyone.

The current German-language S2k guideline on endometriosis states on lifestyle factors that there are still no clear or robust findings, and that this applies equally to body weight, smoking, caffeine, alcohol, diet and physical activity. It then adds that there are isolated indications of environmental influences that may carry risk, and that chemicals such as polychlorinated biphenyls, bisphenol A and phthalates have been linked with gene changes and inflammatory processes. I did not find the word dioxin in the risk factor section of the long version. [Guideline] S2k guideline endometriosis, AWMF registry number 015/045, long version dated April 2025. The English-language journal version of the same guideline is available under PMID: 41684533.

The European ESHRE guideline gives no recommendation at all on environmental factors. Its only statement on prevention is, in essence, that women can be advised to lead a healthy lifestyle, although there is no direct evidence of a preventive benefit. The strength of that recommendation is weak. [Guideline] ESHRE guideline endometriosis 2022. PMID: 35350465

And now you know why, on the subject of environment and hormones, I neither reassure nor alarm. Either would be easier. Either would be wrong.

Frequently asked questions

Where this topic connects to the rest of your body

Dioxins and PCBs are one section of the picture. They stand alongside other groups of substances, alongside inflammatory status, alongside metabolism and alongside the question of what can sensibly be measured. Here are the connecting points.