Dietary factors and risk for endometriosis: a Mendelian randomization analysis

Overall, we systematically curated genome-wide significant SNPs associated with 18 kinds of food intake exposures to examine the potential causal effects of dietary factors on the risk of endometriosis. These different exposure factors could be categorized into six groups, including vegetable intake (salad/raw vegetable intake and cooked vegetable intake), meat intake (processed meat intake, poultry intake, beef intake, non-oily fish intake, oily fish intake, pork intake, and lamb/mutton intake), staple food intake (bread intake and cereal intake), beverage intake (alcoholic drinks per week, alcohol intake frequency, tea intake, and coffee intake), fruit intake (dried fruit intake and fresh fruit intake), and another food intake (cheese intake). As shown in Table  1 , the number of SNPs chosen as IVs for each diet-related factor ranged from 7 to 92 after a series of quality control steps. Moreover, the F statistics were calculated for each instrument-exposure association and none was less than 10 (range: 13.974 to 29.131), suggesting that all SNPs were strong IVs (Table  1 ). The MR estimates from different methods are presented in Table  2 . In this study, two causal associations from 18 food intakes were observed for endometriosis ( p  < 0.05 by IVW method). As the Figs.  3 and 4 ABshow, the IVW method showed that processed meat intake (OR = 0.550; 95%CI:0.314–0.965; p  = 0.037) was significantly associated with a decreased risk of endometriosis. Similarly, salad / raw vegetable intake (OR = 0.346; 95%CI:0.127–0.943; p  = 0.038) was discovered as a protective factor. Heterogeneity test revealed no significant heterogeneity of these IVs (processed meat intake: p IVW =0.607, p MR−Egger =0.548; salad / raw vegetable intake: p IVW =0.678, p MR−Egger =0.620), so we chose the fixed-effect IVW model for MR analysis. In addition, MR-Egger regression test didn’t support any evidence for horizontal pleiotropy (processed meat intake: p for intercept = 0.865; salad / raw vegetable intake: p for intercept = 0.725), which suggested the findings were stable in the sensitivity analysis. Also, the same conclusion could be drawn based on the symmetry of the funnel plot (Fig.  4 C). Furthermore, the leave-one-out analysis indicated that the causal relationships of the positive findings were highly robust (Fig.  4 D). This study also found that alcoholic drinks per week (OR = 0.599; 95%CI:0.353–2.029; p  = 0.059), alcohol intake frequency (OR = 0.998; 95%CI:0.815–1.223; p  = 0.986), poultry intake (OR = 1.543; 95%CI:0.420–5.664; p  = 0.513), beef intake (OR = 0.799; 95%CI:0.340–1.879; p  = 0.608), non-oily fish intake (OR = 0.815; 95%CI:0.305–2.174; p  = 0.683), oily fish intake (OR = 0.658; 95%CI:0.415–1.045; p  = 0.076), pork intake (OR = 1.611; 95%CI:0.447–5.803; p  = 0.466), lamb/mutton intake (OR = 0.795; 95%CI:0.367–1.721; p  = 0.560), bread intake (OR = 0.816; 95%CI:0.476–1.398; p  = 0.459), cheese intake (OR = 0.775; 95%CI:0.523–1.148; p  = 0.203), cooked vegetable intake (OR = 1.237; 95%CI:0.511–2.994; p  = 0.637), tea intake (OR = 0.839; 95%CI:0.600-1.173; p  = 0.304), fresh fruit intake (OR = 0.818; 95%CI:0.444–1.505; p  = 0.518), cereal intake (OR = 1.048; 95%CI:0.634–1.733; p  = 0.855), coffee intake (OR = 0.675; 95%CI:0.388–1.176; p  = 0.165), dried fruit intake (OR = 0.652; 95%CI:0.355–1.198; p  = 0.168) were not associated with endometriosis. The power to detect associations is present in Table  1 . Table 2 Results of the MR study testing causal association between risk factors and endometriosis Exposure Nsnp Methods OR (95%CI) SE P value Heterogeneity Pleiotropy Q P value Intercept SE P value Alcoholic drinks per week 33 MR Egger 0.680(0.201–2.304) 0.623 0.540 48.699 0.023 -0.002 0.011 0.824 Weighted median 0.510 (0.265–0.984) 0.335 0.045 IVW 0.599 (0.353–1.020) 0.271 0.059 48.778 0.029 Simple mode 0.305(0.088–1.057) 0.635 0.070 Weighted mode 0.420(0.182–0.968) 0.426 0.050 Alcohol intake frequency 92 MR Egger 1.232(0.658–2.304) 0.320 0.516 121.029 0.016 -0.005 0.008 0.489 Weighted median 1.125(0.845–1.497) 0.146 0.420 IVW 0.998 (0.815–1.223) 0.103 0.986 121.679 0.018 Simple mode 1.539 (0.747–3.172) 0.269 0.246 Weighted mode 1.472 (0.902–2.402) 0.250 0.125 Processed meat intake 23 MR Egger 0.700 (0.042–11.646) 1.434 0.806 19.585 0.548 -0.004 0.021 0.865 Weighted median 0.589(0.264–1.313) 0.409 0.196 IVW 0.550 (0.314–0.965) 0.287 0.037 19.614 0.607 Simple mode 0.960(0.234–3.935) 0.720 0.955 Weighted mode 0.873 (0.236–3.228) 0.667 0.841 Poultry intake 7 MR Egger 2.746e + 8(3.135e-9-2.636e + 25) 19.927 0.373 3.501 0.623 -0.206 0.216 0.383 Weighted median 1.030 (1.761e-1-6.030) 0.901 0.973 IVW 1.543(4.202e-1-5.664) 0.664 0.513 4.416 0.621 Simple mode 1.207(1.004e-1-1.451e + 1) 1.269 0.887 Weighted mode 1.258(9.968e-2-1.587e + 1) 1.293 0.865 Beef intake 14 MR Egger 0.099(0.001–17.285) 2.635 0.397 11.122 0.519 0.027 0.033 0.437 Weighted median 0.924 (0.288–2.971) 0.596 0.895 IVW 0.799 (0.340–1.880) 0.436 0.608 11.769 0.547 Simple mode 1.305 (0.141–12.111) 1.137 0.818 Weighted mode 1.171 (0.149–9.184) 1.051 0.883 Non-oily fish intake 11 MR Egger 0.327(0.003–35.398) 2.390 0.651 7.107 0.626 0.011 0.029 0.705 Weighted median 0.627(0.166–2.372) 0.679 0.491 IVW 0.815(0.305–2.174) 0.501 0.683 7.260 0.701 Simple mode 1.174 (0.093–14.852) 1.295 0.904 Weighted mode 0.337 (0.044–2.567) 1.036 0.319 Oily fish intake 60 MR Egger 0.073 (0.011–0.474) 0.954 0.008 90.138 0.004 0.033 0.014 0.021 Weighted median 0.495(0.290–0.843) 0.272 0.010 IVW 0.658 (0.415–1.045) 0.236 0.076 98.880 0.001 Simple mode 0.457 (0.137–1.527) 0.615 0.208 Weighted mode 0.424(0.151–1.187) 0.526 0.108 Pork intake 13 MR Egger 9268.385 (13.182-6.517e + 6) 3.345 0.020 11.088 0.436 0.033 0.014 0.021 Weighted median 2.178 (0.468-1.014e + 1) 0.785 0.321 IVW 1.611(0.447–5.803) 0.654 0.466 18.020 0.115 Simple mode 4.974(0.260-9.523e + 1) 1.506 0.308 Weighted mode 5.471(0.305-9.813e + 1) 1.473 0.271 Lamb/mutton intake 30 MR Egger 3.521(0.135–91.566) 1.662 0.455 34.465 0.186 -0.017 0.018 0.365 Weighted median 0.780 (0.272–2.235) 0.537 0.644 IVW 0.795 (0.367–1.721) 0.394 0.560 35.510 0.188 Simple mode 0.798(0.082–7.745) 1.160 0.847 Weighted mode 0.798(0.086–7.408) 1.137 0.844 Bread intake 25 MR Egger 0.817(0.065–10.265) 1.291 0.877 23.592 0.427 -1.886e-5 0.018 0.999 Weighted median 0.759(0.358–1.610) 0.384 0.472 IVW 0.816(0.476–1.398) 0.275 0.459 23.592 0.485 Simple mode 1.373(0.357–5.276) 0.687 0.649 Weighted mode 0.853 (0.273–2.662) 0.581 0.787 Cheese intake 60 MR Egger 1.472(0.275–7.887) 0.857 0.654 88.564 0.006 -0.011 0.014 0.444 Weighted median 0.722(0.450–1.159) 0.241 0.177 IVW 0.775(0.523–1.148) 0.201 0.203 89.472 0.006 Simple mode 0.582(0.187–1.815) 0.580 0.355 Weighted mode 0.582(0.229–1.478) 0.476 0.260 Cooked vegetable intake 17 MR Egger 0.022(1.303e-6-355.783) 4.955 0.451 9.257 0.864 0.042 0.051 0.425 Weighted median 1.450 (4.426e-1-4.751) 0.605 0.539 IVW 1.237 (5.112e-1-2.994) 0.451 0.637 9.931 0.870 Simple mode 3.557 (3.545e-1-35.695) 1.177 0.297 Weighted mode 3.266(3.415e-1-31.236) 1.152 0.320 Tea intake 39 MR Egger 0.649(0.312–1.349) 0.373 0.254 36.051 0.513 0.006 0.007 0.444 Weighted median 0.804 (0.493–1.3119) 0.250 0.383 IVW 0.839(0.600-1.173) 0.171 0.304 36.651 0.532 Simple mode 0.774(0.336–1.781) 0.426 0.550 Weighted mode 0.805(0.483–1.340) 0.260 0.409 Fresh fruit intake 52 MR Egger 0.315 (0.040–2.512) 1.059 0.281 54.822 0.297 0.009 0.010 0.351 Weighted median 0.669 (0.270–1.659) 0.463 0.386 IVW 0.818(0.444–1.505) 0.311 0.518 55.794 0.299 Simple mode 0.458(0.052–4.033) 1.110 0.485 Weighted mode 0.915(0.192–4.363) 0.797 0.911 Cereal intake 38 MR Egger 1.085 (0.123–9.585) 1.112 0.942 44.818 0.149 -0.001 0.016 0.975 Weighted median 1.086(0.575–2.051) 0.325 0.780 IVW 1.048 (0.634–1.733) 0.257 0.855 44.819 0.177 Simple mode 1.104 (0.263–4.634) 0.732 0.893 Weighted mode 1.079 (0.298–3.904) 0.656 0.908 Salad / raw vegetable intake 18 MR Egger 0.792(0.008–81.540) 2.365 0.923 13.713 0.620 -0.009 0.025 0.724 Weighted median 0.419(0.107–1.644) 0.698 0.212 IVW 0.346 (0.127–0.943) 0.511 0.038 13.842 0.678 Simple mode 0.620(0.067–5.746) 1.136 0.679 Weighted mode 0.707(0.076–6.567) 1.137 0.764 Coffee intake 38 MR Egger 0.795 (0.257–2.460) 0.576 0.693 68.780 0.001 − 0.003 0.009 0.746 Weighted median 0.792 (0.442–1.422) 0.298 0.435 IVW 0.675 (0.388–1.176) 0.283 0.165 68.984 0.001 Simple mode 0.791 (0.217–2.263) 0.598 0.556 Weighted mode 0.744 (0.413–1.342) 0.301 0.333 Dried fruit intake 39 MR Egger 0.048 (0.004–0.646) 1.322 0.028 50.764 0.065 0.033 0.016 0.051 Weighted median 0.702(0.326–1.509) 0.391 0.365 IVW 0.652 (0.355–1.198) 0.310 0.168 Simple mode 1.176(0.209–6.614) 0.881 0.855 56.361 0.028 Weighted mode 1.025(0.210–5.011) 0.810 0.976 Results of the MR study testing causal association between risk factors and endometriosis Fig. 3 The causal effect of 18 dietary factors on endometriosis based on the IVW method. IVW, inverse-variance weighted The causal effect of 18 dietary factors on endometriosis based on the IVW method. IVW, inverse-variance weighted Fig. 4 Scatter plot, forest plot, funnel plot and leave-one-out analysis of the causal effect of dietary factors on endometriosis risk Scatter plot, forest plot, funnel plot and leave-one-out analysis of the causal effect of dietary factors on endometriosis risk

The study design is given in Fig.  1 . The genome-wide association study (GWAS) summary statistics used in this work are available from IEU Open GWAS (( https://gwas.mrcieu.ac.uk/ ), and the samples are all from people of European ancestry in order to mitigate potential bias from population stratification. Also, the ethics vote is not necessary for this study due to IEU Open GWAS is a publicly available database and each study included in it was approved by the local Ethical Review Authority. In MR analysis, eligible single nucleotide polymorphisms (SNPs) used as valid IVs should satisfy three basic assumptions: relevance, independence and the exclusion restriction, as shown in Fig.  2 . This study complied with The Strengthening the Reporting of Observational Studies in Epidemiology Using Mendelian Randomisation (STROBE-MR) standard (Table S1 ). Fig. 1 Flowchart of MR analysis in this study. GWAS, genome-wide association studies; SNPs, single-nucleotide polymorphisms; MAF, minor allele frequency; MR, mendelian Randomization Flowchart of MR analysis in this study. GWAS, genome-wide association studies; SNPs, single-nucleotide polymorphisms; MAF, minor allele frequency; MR, mendelian Randomization Fig. 2 Assumption of the Mendelian randomization study Assumption of the Mendelian randomization study We performed a standard two-sample MR analysis to investigate the potential causal relationship between dietary factors and endometriosis. In total, eighteen diet-related exposure factors were identified: alcoholic drinks per week, alcohol intake frequency, processed meat intake, poultry intake, beef intake, non-oily fish intake, oily fish intake, pork intake, lamb/mutton intake, bread intake, cheese intake, cooked vegetable intake, tea intake, fresh fruit intake, cereal intake, salad/raw vegetable intake, coffee intake and dried fruit intake. Summary GWAS data for exposures were sourced from the UK Biobank (e.g., ukb-b-) and other consortia within the IEU Open GWAS database, while outcome data (endometriosis) were obtained from the FinnGen study (finn-b-N14_ENDOMETRIOSIS). To minimize bias from sample overlap, we confirmed that the exposure and outcome GWAS datasets were derived from independent cohorts with no overlapping participants. More detailed information is presented in Table  1 . While we ensured no explicit sample overlap between exposure (e.g., UK Biobank) and outcome (FinnGen) datasets, residual overlap cannot be entirely ruled out if individual-level data were shared across consortia. However, the two-sample MR framework and strict independence of cohorts mitigate this concern. Table 1 Information of the exposures and outcome datasets IEU GWAS id Exposure and Outcome Population Identified SNPs Participants included in analysis R2 F- statistics MDES(OR) ieu-b-73 Alcoholic drinks per week European 33 335,394 2.11E-03 21.441 ≥ 1.245 or ≤ 0.803 ukb-b-5779 Alcohol intake frequency European 92 462,346 3.69E-03 18.605 ≥ 1.180 or ≤ 0.848 ukb-b-6324 Processed meat intake European 23 461,981 7.53E-04 15.131 ≥ 1.444 or ≤ 0.693 ukb-b-8006 Poultry intake European 7 461,990 2.28E-04 15.044 ≥ 1.949 or ≤ 0.513 ukb-b-2862 Beef intake European 14 461,053 4.85E-04 15.965 ≥ 1.580 or ≤ 0.633 ukb-b-17,627 Non-oily fish intake European 11 460,880 4.30E-04 18.015 ≥ 1.626 or ≤ 0.615 ukb-b-2209 Oily fish intake European 60 460,443 2.28E-03 17.520 ≥ 1.235 or ≤ 0.810 ukb-b-5640 Pork intake European 13 460,162 4.25E-04 15.060 ≥ 1.631 or ≤ 0.613 ukb-b-14,179 Lamb/mutton intake European 30 460,006 9.54E-04 14.649 ≥ 1.386 or ≤ 0.722 ukb-b-11,348 Bread intake European 25 452,236 9.92E-04 17.955 ≥ 1.377 or ≤ 0.726 ukb-b-1489 Cheese intake European 60 451,486 1.85E-03 13.974 ≥ 1.264 or ≤ 0.791 ukb-b-8089 Cooked vegetable intake European 17 448,651 6.37E-04 16.821 ≥ 1.491 or ≤ 0.671 ukb-b-6066 Tea intake European 39 447,485 2.17E-03 24.905 ≥ 1.242 or ≤ 0.805 ukb-b-3881 Fresh fruit intake European 52 446,462 2.20E-03 18.888 ≥ 1.240 or ≤ 0.806 ukb-b-15,926 Cereal intake European 38 441,640 1.50E-03 17.499 ≥ 1.297 or ≤ 0.771 ukb-b-1996 Salad / raw vegetable intake European 18 435,435 5.97E-04 14.459 ≥ 1.511 or ≤ 0.662 ukb-b-5237 Coffee intake European 38 428,860 2.57E-03 29.131 ≥ 1.220 or ≤ 0.820 ukb-b-16,576 Dried fruit intake European 39 421,764 1.58E-03 17.152 ≥ 1.289 or ≤ 0.776 finn-b-N14_ENDOMETRIOSIS Endometriosis European NA 77,257 Information of the exposures and outcome datasets To ensure that the conclusion regarding the causal effect of diet-related factors on endometriosis was accurate, we used a series of quality control criteria to satisfy the three fundamental assumptions of MR analysis. Firstly, we obtained the SNPs using a GWAS p-value < 5 × 10 − 8 and excluded SNPs that were in linkage disequilibrium (LD) (clumping window = 10000 kb; r2 < 0.001). Then, we calculated the F-statistics, using the rigorous mathematical formula: F = R 2 ×(N-K-1)/[K×(1-R 2 )] (R 2 : the proportion of exposure variance explained by each genetic variant, R 2  = 2×MAF×(1-MAF)×(Beta/SD) 2 ; N: the sample size of the GWAS; K: the number of SNP), to eliminate the bias arising from weak instrumental variables in the findings (F statistic > 10) [ 13 ]. Finally, SNPs harmonization was also performed by removing palindrome SNPs with intermediate allele frequencies or SNPs with incompatible alleles. In the current study, we adopted inverse variance weighting (IVW) as the primary MR approach to calculate the causal effect of all SNPs. In addition, we ran MR Egger, weighted median, simple mode, and weighted mode as a complement to test the reliability and stability of the results. When utilizing the IVW method, it is necessary to ensure that all IVs in the analysis are of robust validity [ 14 ]. And finally, the estimated causal effects of individual SNPs were quantified as the odds ratios (ORs) from the Wald ratio method. We calculated the minimum detectable effect size (MDES) using the formula: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:MDES=\frac{z1-\alpha\:/2\text{}+z1-\beta\:}{\sqrt{{N*R}^{2}}}$$\end{document} ,where N is the sample size of the outcome GWAS, R 2 is the variance explained by the genetic instruments, and z -values correspond to α = 0.05 and 80% power. After MR analysis, we performed sensitivity analysis, including heterogeneity and pleiotropy, to further validate the robustness of the results. We used the Cochran’s Q statistic of the IVW method and Rucker’s Q statistic of the MR Egger method to identify heterogeneity, where p   0.05 suggested a lack of horizontal pleiotropy [ 15 ]. More rigorously, we conducted leave-one-out analysis to detect if there was any single SNP disproportionately responsible for the results by removing each instrumental SNP in turn. Finally, we utilized scatter and forest plots to visualize the results of the MR analysis. All statistical analyses were performed in R using the TwoSampleMR package.

Overall, we thoroughly examined the potential causal association between dietary intake and endometriosis. Our MR analysis provides suggestive evidence that higher intake of salad/raw vegetables and processed meat may correlate with a reduced risk of endometriosis. However, these findings should be interpreted as hypothesis-generating, emphasizing the need for triangulation with experimental, epigenetic, and observational studies to reconcile genetic associations with biological mechanisms. Further research is required to elucidate the precise causal relationship and underlying mechanisms linking specific dietary patterns to endometriosis.

The etiology of endometriosis is complex, involving immune imbalance, hormone alteration, and inflammation [ 16 ]. To our knowledge, there is mounting evidence suggesting that certain foods may potentially influence the development of endometriosis in susceptible individuals. However, this is the first MR analysis to evaluate the potential causality between dietary factors and the risk of endometriosis using large-scale summary statistics from food intake GWAS and endometriosis GWAS, which provide unconfounded causal estimates. In our analysis, it has been found that consuming salad / raw vegetable may confer a protective association against endometriosis. Additionally, processed meat intake is associated with a reduced risk factor for the condition, which is inconsistent with findings from previous studies. However, there is little evidence supporting an association between endometriosis risk and alcoholic drinks per week, alcohol intake frequency, poultry intake, beef intake, non-oily fish intake, oily fish intake, pork intake, lamb/mutton intake, bread intake, cheese intake, cooked vegetable intake, tea intake, fresh fruit intake, cereal intake, coffee intake and dried fruit intake. The findings of our study may provide valuable insights for clinicians when discussing dietary modifications as a potential adjunctive strategy for patients with endometriosis, such as increasing salad / raw vegetable intake. However, the association between processed meat intake and endometriosis requires further research. For those at high risk for endometriosis, changing dietary habits can also decrease the likelihood of onset. There have been numerous observational studies on the correlation between vegetable intake and the risk of endometriosis. Most of these have shown that increasing vegetable consumption is linked to a reduction in endometriosis risk. Ashrafi M et al. [ 17 ]found that people who kept higher green vegetables intake took a lower endometriosis risk (OR = 0.39, 95% CI = 0.21–0.74; p  = 0.004) in a retrospective case-control study from Iranian. In another hospital-based case-control study by Parazzini et al., comparing 504 women with endometriosis and 504 women without endometriosis confirmed through laparoscopy, the authors indicated a statistically significant decrease in the consumption of green vegetables among cases (OR = 0.3, 95% CI = 0.2–0.5) [ 7 ]. Likewise, several similar studies conducted in other countries also demonstrated a decreased endometriosis risk for those who increased their vegetables consumption [ 18 , 19 ]. However, not all studies shown the effect of vegetable intake on endometriosis. Based on a population-based case-control study involving 944 participants (284 cases and 660 controls), Trabert et al. reported total vegetable intake was not associated with incident endometriosis [ 20 ]. The authors hypothesized that this finding could be attributed to pesticide exposure, which might generate reactive oxygen species and reduce the antioxidant capacity of vegetables. Some studies [ 21 ], on the other hand, have demonstrated that certain class of pesticides can cause estrogenic effects, thereby promoting the development of endometriosis lesions. Alternatively, a meta-analysis indicated an insignificant correlation between eating vegetable and the risk of developing endometriosis [ 22 ]. Furthermore, Harris et al. [ 5 ] reported that a high intake of some vegetables such as cruciferous vegetables, particularly cauliflower, cabbage was related to an increase in endometriosis risk. Through MR analysis, our study indicated that a high level of vegetables intake might be associated with a decreased risk of endometriosis. Endometriosis is an oestrogen-dependent disease. Typically, populations on a diet rich in green vegetables have higher levels of sex-hormone binding globulin (SHBG), which can attenuate the oestrogenic stimulation of the endometrium and restrict the proliferation of prostaglandin-producing tissues. Additionally, dietary fiber can interrupt enterohepatic circulation of oestrogen conjugates, thereby reducing the risk of endometriosis [ 17 ]. Studies have indicated that a number of nutrients found in vegetables potentially benefit endometriosis. First of all, vitamins, especially vitamin C, are important antioxidants that strongly neutralize free radicals and improve oxidative status to reduce the chances of developing endometriosis [ 23 ]. This finding aligns with a randomized, triple-blind placebo-controlled clinical study that reported a decrease in systemic indicators of oxidative stress in patients with endometriosis after receiving a boost of vitamin C [ 24 ]. Furthermore, the influence of vitamin C on the expression and production of the VEGF gene was investigated in peritoneal macrophages from women diagnosed with endometriosis [ 25 ]. Vitamin A can also play a role in influencing aberrant cytokines production in endometriosis, such as suppressing the transcription and translational processes of IL-6 and VEGF [ 26 ]. Secondly, vegetables are packed with bioactive plant compounds, especially polyphenols (such as curcumin, resveratrol and epigallocatechin gallate). Natural polyphenols have been proven to possess anti-inflammatory and antioxidative properties, making them a cost-effective and easily accessible treatment option for endometriosis [ 27 ]. In addition to these properties, polyphenols can be used as estrogen receptor agonists to combat the condition due to their structural similarity with estradiol [ 28 ]. Thirdly, many vegetables contain phytoestrogens that can be classified into three classes: flavonoids, lignans and stilbenes. These compounds are structural and functional homologies with estrogen and act as weak estrogenic factors by binding to the estrogen receptor and interfering with ER mediated responses [ 29 ]. Furthermore, the mechanism of action of flavonoids is pleiotropic and includes promoting autophagy, down-regulating nuclear factor (NF)-κB activity, reducing interleukin (IL)-6 and tumor necrosis factor α (TNFα), as well as inhibiting oxidative stress, thereby generating proapoptotic, anti-inflammatory, and anti-proliferative effects [ 30 ]. Our MR analysis also indicated a decreased endometriosis risk for those with processed meat intake, which was consistent with the result of a case-control study by Ashrafi M et al. [ 17 ](OR = 0.61, 95% CI = 0.41–0.91, P  = 0.015). However, different results were obtained by other studies. In a large Italian study, endometriosis risk was notably higher among women in the highest intake of red meat, both processed and unprocessed, compared to those in the lowest (OR = 2.0, 95% CI = 1.4–2.8; P  = 0.0004) [ 7 ]. In a Nurses’ Health Study II (NHSII) prospective cohort including 81,908 participants, red meat consumption, especially non-processed rea meat consumption, was correlated with a greater risk of laparoscopically-confirmed endometriosis by approximately 56% (95% CI = 1.22–1.99; P  < 0.0001) [ 8 ]. Likewise, a meta-analysis of observational studies reported that women eating red meat had a 17% higher risk in endometriosis [ 22 ]. In contrast to these findings, a Washington state based case-control study [ 20 ] and a Belgian matched case-control study with prospective recruitment [ 31 ] showed no association between red meat intake and incident endometriosis. Our genetic instrument (ukb-b-6324) was robustly associated with processed meat intake. We identified genetic variants significantly associated with processed meat intake at the genome-wide level ( p  < 5 × 10 − 8 ) from IEU Open GWAS comprising N  = 461,981 individuals (ukb-b-6324). Following clumping (r2 < 0.001), 23 independent SNPs were retained. Collectively, these SNPs explained the variance in processed meat intake, as evidenced by an F-statistic of 15.131, which surpassed the threshold for weak instrument bias (F statistic > 10). Sensitivity analyses were conducted to evaluate the robustness of our findings. Heterogeneity test indicated no significant heterogeneity among these IVs (P IVW =0.607, P MR Egger =0.548), and MR Egger regression test provided no evidence for horizontal pleiotropy (p for intercept = 0.865). These results suggest the findings remained stable under the sensitivity analysis. There are several possible explanations for the discrepancy between our results and those reported in previous studies. First, the genetic instruments for processed meat intake were derived from self-reported dietary data, which may inaccurately reflect true consumption patterns. For example, the GWAS phenotype aggregates broad categories (e.g., ‘processed meat’ includes diverse products like sausages, bacon, and cured meats), potentially obscuring heterogeneity in preparation methods, additives (e.g., nitrates, preservatives), or portion sizes. Misclassification of exposure could bias genetic associations toward the null or spuriously inverse effects. Second, food additives used in processed foods, once ingested into the gastrointestinal system, can significantly impact the composition and functionality of gut microbiota [ 32 ]. Moreover, certain additives or cooking methods may exert unintended anti-inflammatory or antioxidant effects in specific contexts. Alternatively, processed meat consumption might proxy for other unmeasured dietary or lifestyle factors (e.g., higher protein intake, socioeconomic status) that confound observational studies but are disentangled in MR. Third, observational studies may overestimate risks if women with endometriosis consciously reduce processed meat intake due to symptom severity or dietary advice, creating reverse causality. Fourth, another plausible explanation could be the differences in slaughter procedures, such as ritual cutting versus captive bolt stunning. Ritual cutting is associated with reduced pain and lower levels of stress hormones in animals. Additionally, since the heart continues to function for a longer period, more blood and other substances, including hormones, are effectively removed from the animal’s body. Furthermore, there is an association between heme iron intake and the risk of endometriosis [ 17 ]. These hypotheses highlight the complexity of interpreting diet-disease associations and underscore the need for refined dietary phenotyping in GWAS, mechanistic studies, and replication across diverse populations to resolve these contradictions. Beyond genetic predisposition, epigenetic modifications represent a critical interface between diet and disease pathogenesis. Endometriosis is characterized by aberrant DNA methylation patterns in genes regulating inflammation (e.g., NF-κB), hormone response (e.g., ESR1), and cell adhesion (e.g., CDH1) [ 33 ]. Dietary factor, such as acrylamide (present in processed meats), is known to modulate DNA methylation and histone modifications. For example, acrylamide exposure has been associated with hypermethylation of tumor suppressor genes. Although our MR analysis focuses on germline genetic variants, the interplay between diet-associated SNPs and epigenetic regulation remains unexplored. Subsequent research should prioritize multi-omics approaches to unravel these dynamics, combining MR with epigenome-wide association studies (EWAS) to identify diet-modifiable epigenetic signatures in endometriosis. Therefore, our conclusion must be interpreted cautiously. It is crucial to accurately comprehend the correlation between MR analysis and observational studies. MR analysis makes a terrific addition to observational studies as it remains unaffected by common confounding factors or reverse causation bias, yet it cannot replace them entirely. Previous studies provide solid evidence linking red meat consumption to an increased risk of many chronic diseases, including diabetes, hypertension, cardiovascular disease and some cancers [ 22 ]. Although the physiological mechanism of how red meat affects endometriosis remains incompletely understood, it has been postulated to involve several ways. On the one hand, a high intake of animal fat in a meat-based diet such as palmitic acid can further increase endogenous estrogens, which stimulate the formation of proinflammatory PGs. These PGs can also induce the release of aromatase P450, promoting inflammatory conditions in endometriosis [ 8 ]. On the other hand, diets rich in red meat seem to correlate with decreased SHBG and increased estradiol concentrations, that influence pain in women with endometriosis [ 34 ]. Another possible mechanism is iron overload in women with a high intake of red meat, which is related to increased oxidative stress and inflammatory status in endometriosis [ 8 ]. Furthermore, iron overload in the peritoneal fluid of women with endometriosis can decrease GPX4 expression, cause embryotoxicity and induce ferroptosis, which probably participates in endometriosis-associated reproductive failure [ 35 ]. To date, the dietary structure is complex and the contribution of diet to endometriosis has not been investigated. Recently, an increasing body of evidence implicates thar gut microbiota plays a critical role in mediating the link between dietary factors and endometriosis risk. Specifically, gut microbiota, such as Bacteroides, Bifidobacterium, Escherichia coli, and Lactobacillus, can secrete enzymes like β-glucuronidase and β-glucosidase. These enzymes deconjugate estrogens, thereby increasing the reabsorption of free estrogen and elevating systemic estrogen levels, which may create an environment conducive to the progression of endometriosis [ 36 ]. Another potential mechanism by which gut microbiota influences host physiology involves the production of short-chain fatty acids (SCFAs). SCFAs are produced by specific bacteria in the cecum and colon through the fermentation of undigested dietary fiber [ 37 ]. Not only are SCFAs present in the gut, but they can also enter systemic circulation via the bloodstream. The concentrations of SCFAs in blood or tissues can modulate inflammation and immunity by regulating the production of immune mediators, cytokines and chemokines, as well as the differentiation, recruitment and activation of immune cells, including neutrophils, macrophages, dendritic cells (DCs), and T lymphocytes [ 38 ]. Consequently, further observational studies and innovative MR analyses are warranted to elucidate the role of diet in endometriosis. In addition, recent evidence highlights that excessive salt consumption may promote systemic inflammation by activating Th17 cells and upregulating pro-inflammatory cytokines (e.g., IL-17, TNF-α), which are central to endometriosis pathogenesis [ 39 ]. High salt intake has also been linked to gut dysbiosis and impaired immune tolerance [ 40 ], potentially exacerbating endometriosis progression. Future MR analyses should incorporate salt intake as an exposure to clarify its role in endometriosis risk and explore interactions with other dietary factors. Notably, our MR analysis possesses several significant strengths. Firstly, to the best of our knowledge, this MR study is the first to systematically analyze the causality between dietary factors and endometriosis by using genetic variation as IVs, effectively overcoming the reverse causality and confounding bias. Moreover, we utilized European populations as both the exposure and outcome groups to minimize potential biases. Secondly, some of the findings from this study contradict current knowledge, thereby providing valuable insights for future research directions. Thirdly, we took several steps to meet the core assumptions of MR analysis and utilized a large sample size along with SNPs derived from GWAS, thus significantly enhancing the credibility of our findings. Inevitably, this study also has some limitations that warrant acknowledgment. First, the analysis was conducted solely with European participants, which limits the generalizability of these findings to other population groups. Furthermore, we were unable to differentiate the specific effects of different dietary combinations. Second, given that food intake GWAS are still in their early stages, the statistical power of our analysis was constrained by the limited number of SNPs available for certain dietary exposures. For instance, poultry intake was proxied by only 7 SNPs, which may inadequately capture the genetic variance of this exposure and reduce the ability to detect true causal effects. This limitation is reflected in the calculated MDES for such exposures (Table  1 ), where poultry intake exhibited an MDES(OR) ≥ 1.949 (OR = 1.543), indicating insufficient power to reliably identify odds ratios below this threshold. Similarly, exposures like beef intake ( N  = 14 SNPs) and non-oily fish intake ( N  = 11 SNPs) had fewer than 15 SNPs, further limiting precision. While our sensitivity analyses confirmed the robustness of findings, non-significant results for underpowered exposures should be interpreted cautiously, as they may reflect type II errors rather than definitive null effects. Future GWAS with larger sample sizes and improved dietary phenotyping are critical to identify stronger genetic instruments and refine causal estimates for these factors. Third, the lack of summary-level GWAS data across various ages groups precludes further age-stratified analyses. Lastly, it remains unclear whether dose-response relationships exist between dietary factors and the risk of endometriosis. Additionally, the food intake data in our study were obtained through a self-reported questionnaire instead of objective measurements, potentially introducing recall bias. While MR minimizes key biases inherent to observational studies, it does not definitively establish causality. Unmeasured horizontal pleiotropy or residual confounding via gene-environment interactions could still influence results. Additionally, MR estimates reflect lifelong effects of genetic predispositions to dietary habits, which may not fully align with short-term dietary interventions. Thus, our findings should be interpreted as suggestive rather than conclusive evidence of causality.

Endometriosis is an oestrogen-dependent chronic inflammatory process characterized by the presence of endometrial-like tissue outside the uterus, primarily on pelvic tissues [ 1 ]. Common disease symptoms, including severe chronic pelvic pain, secondary dysmenorrhea and infertility, substantially alter the patient’s work productivity, social life and psychological well-being. This condition affects 6-10% of reproductive-age women worldwide and represents a considerable burden on society [ 2 , 3 ]. The core goals of endometriosis treatment proclaimed by the UK Endometriosis Association are to minimize diagnosis time and ensure patients have access to comprehensive treatment and support [ 4 ]. However, the diagnosis is often delayed because symptom severity does not correlate with the extent of endometrial lesions. Moreover, current treatments, including hormonal therapies and surgery, offer symptomatic relief but do not target the underlying etiology, underscoring the critical need to identify modifiable risk factors for prevention and early intervention. Consequently, we must fully understand the etiology of endometriosis to mitigate its health consequences. There is a growing discussion focused on exploring the potential role of dietary patterns and nutritional interventions as both preventive measures and adjunct therapeutic approaches for endometriosis management. A critical knowledge gap lies in understanding whether diet and nutrition can contribute to mitigating symptoms and potentially influencing the progression of endometriosis. Dietary factors may directly contribute to the progression and severity of endometriosis due to their involvement in oxidative stress, muscle contraction, inflammation and steroid hormone metabolism [ 5 ]. Notably, clinical observations from an Australian national online survey revealed that 76% of endometriosis patients employ self-management protocols, frequently combining mindfulness practices, physical activity, and nutritional modifications. Among these women, 44% specifically attempted to manage their condition through dietary adaptations, including gluten-free or vegan diets. Participants rated the effectiveness of dietary interventions at an average of 6.4 out of 10 [ 6 ]. Importantly, a growing body of observational study [ 7 – 10 ] has identified correlations between endometriosis and the consumption of specific food groups, such as green vegetables, fresh fruit, red meat, dairy, and fish. A healthier dietary pattern, characterized by a higher intake of fruits and vegetables and reduced consumption of red meat and trans fats, was associated with a 13% lower risk of endometriosis diagnosis [ 11 ]. These findings suggest that targeted dietary modifications may influence symptom severity and disease progression. However, there is currently no consensus on evidence-based clinical guidelines regarding diet and nutrition in endometriosis. The methodological constraints inherent in observational research, including confounding factors, measurement error, reverse causation, and selection bias, may compromise the reliability of causal inference. Additionally, observational studies often focus on baseline or short-term dietary assessments, failing to reflect long-term dietary patterns. In contrast, mendelian randomization (MR) uses genetic variants as instrumental variables (IVs) to infer potential causal relationships between exposure and disease outcome, reducing susceptibility to confounding and reverse causation [ 12 ]. Moreover, MR reflects the cumulative effect of exposure over the lifespan. As far as we know, there have been few MR studies on the association between dietary factors and endometriosis. Therefore, we performed an MR analysis to explore the causal effect of dietary factors on endometriosis. The 18 dietary factors examined in this MR analysis, such as processed meat, salad/raw vegetables, and others, were selected due to their prominence in previous observational studies and their biological plausibility in endometriosis pathogenesis, particularly their roles in modulating inflammation, oxidative stress, and hormone metabolism.

Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1