Endometriosis has had strong twin evidence for more than 25 years. That makes it different from a topic where twin research is merely proposed: the genetically informative signal is already established.
The unresolved question is what that signal means biologically and clinically. Twin models show substantial inherited variation in population liability, while modern GWAS points to many small-effect loci and shared biology with pain and inflammatory conditions. None of those findings tells one woman what percentage of her disease is genetic or predicts her pain, fertility impact, stage or treatment response.
Quick answer
If twin studies estimate endometriosis heritability around 47–51%, does that mean about half of one woman’s disease is genetic?
No. Those figures are model estimates of population-level variation in underlying liability in particular twin samples. They are not a personal causal percentage. Genetics contributes to susceptibility, but disease expression, pain, lesion subtype, fertility impact and clinical course remain much more complex.
Key takeaways
- The 1999 Australian twin study received responses from 3,096 twins; 215 reported endometriosis, and the fitted model attributed about 51% of latent liability variance to additive genetic influences in that sample.
- A 2015 Swedish twin-registry study of 28,370 women found higher concordance among monozygotic than dizygotic twins and a best-fitting model with 47% additive genetic and 53% unique/nonshared environmental components.
- “Unique environment” is a statistical variance component. It does not mean researchers identified lifestyle, stress, diet or another exposure as the cause of discordance.
- A 2023 GWAS meta-analysis of 60,674 cases and 701,926 controls identified 42 genome-wide significant loci with 49 distinct signals and found stronger effects for stage III/IV disease, driven by ovarian endometriosis.
- A 2026 review traces roughly 25 years of genetic research and reports around 80 mapped risk factors in the reviewed literature, while emphasizing that individual common-variant effects are small and clinical prediction remains limited.
The Australian twin study changed the question
The landmark 1999 Australian study surveyed female monozygotic and dizygotic twin pairs and validated reported diagnoses with medical or pathology information where possible. Monozygotic pair similarity exceeded dizygotic pair similarity.
The authors’ model suggested that about 51% of variance in latent liability to endometriosis in that sample could be attributed to additive genetic influences. The key phrase is latent liability variance: this is a population-model result, not a statement that one woman’s endometriosis is 51% caused by genes.
A much larger Swedish twin sample pointed in the same direction
The 2015 Swedish registry analysis included 28,370 female twins, with 1,228 reporting a history of endometriosis. Probandwise concordance was 0.21 in monozygotic twins and 0.10 in dizygotic twins, and within-pair correlations were also higher in monozygotic pairs.
The best-fitting model attributed 47% of variance to additive genetic factors and 53% to unique or nonshared environmental components. That replication strengthened the evidence for inherited susceptibility while also showing that shared DNA does not make outcomes identical.
What “unique environment” does — and does not — mean
In a classical twin model, the nonshared environmental component captures influences that make co-twins different under the model. Depending on the data and assumptions, measurement error can also contribute to that component.
It is therefore incorrect to translate 53% “unique environment” into 53% lifestyle, stress, diet or toxin exposure. The model tells researchers that there is variation not attributed to the additive genetic component; it does not identify the causal exposure.
Modern genomics shows a polygenic disease with heterogeneous biology
The 2023 Nature Genetics meta-analysis combined 60,674 endometriosis cases and 701,926 controls of European and East Asian ancestry. It identified 42 genome-wide significant loci comprising 49 distinct association signals.
Effects were larger for stage III/IV disease, driven by ovarian endometriosis. The study also reported genetic correlations with several pain and inflammatory conditions, illustrating that endometriosis genetics intersects with biology beyond a single reproductive pathway.
The 2026 view: many risk factors, individually small effects
A 2026 Human Reproduction Update review follows the field from the 1999 twin study through modern GWAS and multi-omic work. It reports that the literature has now mapped around 80 genetic risk factors implicating hormone signalling, reproductive-tract development, cell proliferation and epithelial biology.
The review also emphasizes a familiar complex-trait lesson: individual common genetic effects are small. Genetic discovery can illuminate biology without immediately producing an accurate diagnostic score or a tool that predicts one patient’s symptoms or treatment response.
Pain and lesion burden are not the same phenotype
Endometriosis is clinically heterogeneous. A binary diagnosis does not capture lesion location or subtype, stage, pain pattern, fertility history or longitudinal symptom burden.
That heterogeneity matters for genetically informative research because the most useful question may not be only who has endometriosis. It may be why genetically similar women differ in lesion phenotype, pain sensitization, fertility impact or progression over time.
What we know
Twin studies support a substantial inherited contribution to population liability, and modern genomic studies confirm that endometriosis is polygenic rather than a one-gene disorder.
Genetic findings implicate multiple biological pathways and overlap with pain and inflammatory conditions. Disease severity and subtype can also show different genetic signal strength.
What we still do not know well
We still cannot explain why genetically similar women can differ markedly in whether disease is diagnosed, which lesions develop, how much pain they experience, how fertility is affected or how symptoms change over time.
We also do not yet know which combinations of genetic, developmental, hormonal, immune, treatment and environmental information will become clinically useful rather than merely statistically associated.
The twin study we would like to see
A modern program could recruit monozygotic and dizygotic twin sisters and combine clinically validated diagnosis with lesion subtype and stage, imaging or surgical data where available, detailed pain phenotypes, fertility history, hormonal-treatment history and longitudinal symptom tracking.
Adding genomic, epigenetic and other omic layers could help prioritize mechanisms, especially in discordant monozygotic pairs. Repeated follow-up would be more informative than a single cross-sectional snapshot.
What that study still could not prove
If identical twins were discordant for endometriosis, the contrast could generate useful hypotheses about nonshared biology and exposures. It would not prove that epigenetics, stress, diet, hormones or another observed difference caused the disease.
Observational twin designs strengthen inference by controlling some shared background. They do not replace randomized experiments, mechanistic studies or clinical validation.
TwinPare boundary
TwinPare Research can explain heritability, genomic discovery and research gaps. It must not estimate a woman’s personal endometriosis probability, stage, pain severity, fertility outcome or treatment response from genetics, twin status or app data.
TwinPare Health & Fitness must not diagnose endometriosis from pain, cycle patterns, HRV, sleep, temperature or other wearable signals, and it must not recommend hormonal therapy, surgery, supplements or fertility treatment. Persistent or severe symptoms require appropriate clinical assessment.
Source notes
The sources have been verified and editorially reviewed for this article. The limitations below show which level of conclusion the sources support.
- [treloar-endometriosis-australian-twins-1999] Genetic influences on endometriosis in an Australian twin sample Susan A Treloar; D T O'Connor; V M O'Connor; Nicholas G Martin. Fertility and Sterility, 1999. Evidence type: Australian twin survey with medical/pathology validation where available Limitation: Among 3,096 respondents, 215 reported endometriosis. The twin model suggested that about 51% of latent liability variance in that sample was attributable to additive genetic influences. This is a population-model estimate, not a personal percentage of disease causation. PubMed DOI
- [saha-endometriosis-swedish-twins-2015] Heritability of endometriosis Rama Saha; Hans Järnbert Pettersson; Pia Svedberg; Matts Olovsson; Agneta Bergqvist; Lena Marions; Per Tornvall; Ralf Kuja-Halkola. Fertility and Sterility, 2015. Evidence type: Swedish population-based twin-registry analysis Limitation: The study included 28,370 female twins and 1,228 reported endometriosis histories. The best-fitting model attributed 47% of population-level variance to additive genetic factors and 53% to unique/nonshared environmental components. “Unique environment” is a model component, not a proven list of causal exposures. PubMed DOI
- [rahmioglu-endometriosis-gwas-2023] The genetic basis of endometriosis and comorbidity with other pain and inflammatory conditions Nilufer Rahmioglu et al.. Nature Genetics, 2023. Evidence type: GWAS meta-analysis of 60,674 cases and 701,926 controls of European and East Asian ancestry Limitation: The study identified 42 genome-wide significant loci comprising 49 distinct association signals. Effect sizes were larger for stage III/IV disease, driven by ovarian endometriosis, and genetic correlations were observed with multiple pain and inflammatory conditions. These are population associations, not personal diagnosis or prognosis. PubMed PMC DOI
- [montgomery-endometriosis-genetic-architecture-2026] Genetic architecture of endometriosis: risk factors, comorbidities and clinical implications Grant W Montgomery; Isabelle M McGrath; Michael Toft Overgaard; Mette Nyegaard; Sally Mortlock. Human Reproduction Update, 2026. Evidence type: State-of-the-art review of 25 years of endometriosis genetics through June 2026 Limitation: The review traces the field from the landmark 1999 twin study to modern GWAS and reports around 80 mapped genetic risk factors in the reviewed literature. Common-variant effects are individually small, and current genetics does not provide a stand-alone diagnostic, pain-severity, fertility or treatment-response prediction for an individual. PubMed DOI
Editorial source review
This section shows how the article's key factual claims are linked to the source.
Phrasings that require caution
- Do not state that endometriosis is 47–51% genetic for an individual.
- Describe the 47–51% figures as population-level model estimates of variance in latent liability in specific twin samples.
- Do not equate the nonshared environmental component with proven lifestyle or exposure causes.
- Do not claim monozygotic-twin discordance proves epigenetics caused endometriosis.
- Do not imply a current polygenic score can diagnose endometriosis or predict stage, pain, fertility outcome or treatment response.
- Do not diagnose endometriosis from symptoms, family history, twin status or wearable data.
- Do not recommend hormonal therapy, surgery, supplements or fertility treatment.
- Do not imply TwinPare is conducting the proposed twin study.
| ID | Claim | Source support | Caution |
|---|