Associations between 1400 metabolites and subtypes of endometriosis: a two-sample Mendelian randomisation study

Background: endometriosis is a chronic inflammatory disease with a prevalence of approximately 10% in women of childbearing age. Metabolic pathways have been demonstrated by previous studies to be potential avenues for the development of new therapeutic strategies and may be used for early diagnosis of the disease. This study aimed to investigate the potential causal relationships between 1400 metabolites and various endometriosis subtypes using Mendelian randomisation (MR) analysis.

Data from a genome-wide association study were analysed. MR analysis was performed using the inverse-variance weighted, MR-e gger, and weighted-median methods, accompanied by heterogeneity testing, sensitivity analysis, and pleiotropy analysis. Metabolic-pathway enrichment analysis was conducted on the preliminarily screened differential metabolites, and colocalisation analysis was subsequently performed for exposure–outcome pairs that remained causally associated after multiple-testing correction.

After multiple-testing correction, only the glycerol-to-palmitoylcarnitine ( c16) ratio reduced the risk of stage 1–2 endometriosis ( PFDR = 0.045; odds ratio [ oR], 0.737; 95% confidence interval [ cI], 0.638–0.852) and pelvic peritoneal endometriosis ( PFDR = 0.039; oR, 0.721; 95% cI, 0.619–0.841). c olocalisation analysis revealed that they did not share causal variant loci at the genetic level. no reverse causal associations were found in the reverse Mendelian analysis. Metabolic pathway enrichment analysis identified major metabolic pathways, including caffeine metabolism, glutathione metabolism, arginine biosynthesis, sphingolipid metabolism, pantothenate and c oA biosynthesis, plasmalogen synthesis, and biosynthesis of unsaturated fatty acids.

o ur study suggests potential causal relationships between metabolites and various endometriosis subtypes from an MR perspective. However, the limited number of associations that survived multiple-testing correction indicates that these findings are preliminary and require validation in larger cohorts. This exploratory analysis may contribute to advancing future research on metabolomics-based diagnosis, treatment, and prevention of endometriosis. PLAIN LANGUAGE SUMMARY endometriosis is a condition in which tissues similar to the uterine lining grow outside the uterus. It affects approximately 1 in 10 reproductive-age women and can cause pain and fertility problems. In this study, we investigated whether certain natural substances in the blood, called metabolites, play a role in different forms of endometriosis. We used genetic data to test over 1,000 metabolites and found that only the glycerol-to-palmitoylcarnitine ratio showed a meaningful link. A higher ratio may reduce the risk of mild or pelvic peritoneal endometriosis. This early finding points to a new direction for future research; nevertheless, more studies are required to confirm this and understand how this metabolite balance may be involved in the disease. © 2025 t he a uthor(s). Published by i nforma uK limited, trading as taylor & f rancis Group CONTACT Zongju Huang [email protected] d epartment of a cpuncture and Moxibustion, Jiangbei d istrict Hospital of traditional chinese Medicine, no.35, no.1 Village, Jianxin east r oad, Jiangbei d istrict, chongqing400021, china #f ei yan and Zhouxiang chen contributed equally to this work as co-first authors. supplemental data for this article can be accessed online at https://doi.org/10.1080/01443615.2025.2552402. https://doi.org/10.1080/01443615.2025.2552402 t his is an o pen a ccess article distributed under the terms of the c reative c ommons a ttribution license ( http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. t he terms on which this article has been published allow the posting of the a ccepted Manuscript in a repository by the author(s) or with their consent. ARTICLE HISTORY Received 26 April 2025 Accepted 19 August 2025

Metabolites; endometriosis; glycerol to palmitoylcarnitine ( c16) ratio; Mendelian randomisation 2 F. yAn eT Al.

endometriosis is characterised by the presence of functional endometrial tissues (including glands and stroma) outside the uterine cavity ( olive and Pritts 2001, Burney and Giudice 2012) and affects approxi - mately 10% of women of childbearing age (Horne et  al. 2019), accounting for 25–50% of infertility cases in women ( c ounseller and c renshaw 1951). The primary sites of endometriosis are the ovaries, pelvic peritoneum, and rectovaginal septum, and its common symptoms include dysmenorrhoea, pain during intercourse, chronic pelvic pain, and infertility ( c ounseller and c renshaw 1951, Sinaii et  al. 2008). Women with these chronic symptoms often experience emotional distress, reduced quality of life, and elevated levels of stress, anxiety, and depression. Some patients may not exhibit noticeable symptoms, making the diagnosis of endometriosis challenging (Soliman et  al. 2017), which is often only considered during fertility evaluations or in the presence of other gynaecological symptoms. The pathogenesis of endometriosis is not fully understood. Beyond its local gynaecological manifesta - tions, endometriosis is also recognised as a systemic disease involving complex metabolic dysregulation (Angioni et  al. 2021, Adamyan et  al. 2024). Women with endometriosis exhibit distinct metabolomic pro - files, characterised by significant alterations in oxidative stress pathways, lipid metabolism, and amino acid turnover. These disruptions include enhanced reactive oxygen species (R oS) production, dysregu - lated phospholipid synthesis and fatty acid oxidation, and altered glutamine and tryptophan metabolism, which collectively contribute to chronic inflammation and energy metabolism dysfunction (Angioni et  al. 2021, Adamyan et  al. 2024). These metabolic perturbations highlight the potential of metabolomic bio - markers as diagnostic and therapeutic targets. Metabolomics, an emerging research tool, offers new perspectives for the diagnosis and treatment of various diseases, with existing studies revealing its tremendous potential in endometriosis research. Dutta (Soliman et  al. 2017) utilised nuclear magnetic resonance technology to analyse serum samples from patients with endometriosis and successfully identified multiple metabolite changes closely related to the disease state, including alterations in specific amino acid and organic acid levels. Jana et  al. (Jana et  al. 2013) further explored the roles of blood metabolites in endometriosis. Through a more refined analysis, they discovered significant changes in specific amino and organic acid levels in patients with endome - triosis, deepening our understanding of the value of metabolomics in comprehending the pathophysiol - ogy of endometriosis. Ghazi (negar et  al. 2016) revealed changes in the levels of metabolites such as 2-methoxyestradiol in patients with endometriosis through serum metabolomic analysis, thereby offering new insights into the role of hormone metabolism in the pathogenesis of endometriosis. Vouk (negar et  al. 2016), analysed lipid metabolites in blood and peritoneal fluid and found significant changes in lipid metabolism, especially in the phosphatidylcholine and phosphatidylserine levels, revealing the potential role of lipid metabolism in the pathophysiology of endometriosis. l ee et  al. (l ee et  al. 2014) reported that specific lipid metabolites, particularly glycosphingolipids and lactosylceramides, were ele - vated in patients with endometriosis, providing crucial evidence for lipid metabolic changes during inflammation and cell proliferation processes. Dutta et  al. (Dutta et  al. 2018) conducted a comprehensive analysis of endometrial tissue and serum samples from patients with endometriosis using an integrated metabolomic approach. Their study not only confirmed previously discovered metabolic abnormalities but also revealed new metabolite changes, offering deeper insights into the pathophysiology of endo - metriosis. These studies indicate that metabolic pathways are potential avenues for the development of new therapeutic strategies (li et  al. 2023, l u et  al. 2023) and may be utilised for the early diagnosis of diseases ( ortiz et  al. 2021). nonetheless, the aforementioned investigations are largely observational, employ modest sample sizes, and evaluate limited subsets of metabolites; consequently, they cannot establish whether the observed metabolic alterations are the causes or consequences of endometriosis. A comprehensive causally oriented assessment of the circulating metabolome (> 1,000 metabolites) in relation to disease risk is still lacking. Mendelian randomisation (MR) studies, which utilise genetic markers as instrumental variables (IVs), can effectively infer the causal relationships between exposure and outcomes (Davey Smith and Hemani  2014). This research design overcomes the limitations of traditional observational studies that are susceptible to confounding factors and reverse causality, offering validity similar to that of randomised JouRnAl oF oBSTe TRIcS AnD GynAecoloGy 3 controlled trials, but at a lower cost, serving as foundational research for clinical studies (Burgess and Thompson 2015). In this study, we employed the two-sample MR approach, selecting genome-wide association study (GWAS) data from multiple subtypes of endometriosis, to comprehensively explore the causal relation - ships between 1400 metabolites and endometriosis. We hypothesised that specific circulating metabo - lites exert causal influences, either promotive or protective, on the development of endometriosis. The findings may refine the understanding of disease aetiology and inform early non-invasive diagnostic strategies as well as novel metabolic interventions.

Study design A bidirectional two-sample MR analysis was conducted to evaluate the causal relationships between 1,400 blood metabolite phenotypes and various subtypes of endometriosis. The overall study design and participant flow are summarised in Figure 1 . The MR analysis was based on three core assumptions (Haycock et  al. 2016): (i) the IV is associated with exposure; (ii) the IV is independent of any known or unknown confounding factors that mediate exposure and outcome; and (iii) the outcome is related to the genetic instrument solely through the exposure effect. Ethical considerations This study utilised summary statistics from previously published GWAS that had already received ethical approval from their respective institutions and obtained informed consent from the participants. As our study design was a secondary analysis of publicly available summary data without access to individual-level information, no additional ethical approval, study registration, or informed consent was required. This type of MR analysis using summary statistics is exempt from registration requirements because it does not involve direct interactions with human subjects. o ur research methods complied with all relevant institutional and research governance standards for genetic epidemiology studies that utilise public data. Data source The GWAS data on endometriosis were sourced from a Finnish database ( https://www.finngen.fi/en) com - prising eight distinct datasets. This included two datasets based on disease staging: stage 1–2 endome - triosis and stage 3–4 endometriosis, which were categorised according to the American Society for Reproductive Medicine (ASRM) staging system. Additionally, five datasets were classified according to specific locations (not involving disease staging and infiltration depth): endometriosis of the fallopian tube, intestine, ovary, pelvic peritoneum, rectovaginal septum, and vagina. Another dataset on deep endometriosis referred to cases in which the infiltration depth into the subperitoneal space was equal to or greater than 5 mm, without involving disease staging and location. Detailed information is provided in Table 1. Figure 1. f lowchart of the study design. 4 F. yAn eT Al. Blood metabolite data were obtained from the c anadian l ongitudinal Study on Ageing ( clSA), target - ing 8,299 unrelated european participants for whole-genome genotyping and assessment of circulating plasma metabolites ( chen et  al. 2023). After a series of rigorous quality control steps, including eliminat - ing data that could introduce systematic errors, mismatches, and background noise, 1,091 metabolites (850 known and 241 unknown substances) and 309 metabolite ratios were identified. This study provides a foundational dataset for GWAS. These data were meticulously recorded on the GWAS catalogue website (https://www.ebi.ac.uk/gwas/) (IDs GcST90199621–GcST90201020, detailed in Supplementary Table 1 ). It is important to note that the endometriosis GWAS data were sourced from a Finnish population, whereas the metabolite data were obtained from the c anadian l ongitudinal Study on Ageing ( clSA) cohort, which primarily represents the c anadian population. These two cohorts were from different geo - graphic locations; thus, we can confidently assert that there was no overlap between the participants in the two datasets. This ensured the independence of the exposure (metabolite) and outcome (endome - triosis) datasets, which are critical for the validity of the MR analysis. IV Selection In selecting the IVs for metabolites, we referred to recent studies ( orrù et  al. 2020, yu et  al. 2021), setting the significance level at 5 × 10−6 . The clumping procedure in PlInK software (version 1.90) was used to prune these SnPs (linkage disequilibrium [lD] r 2 threshold < 0.001 within a 10,000 kb distance) (The 1000 Genomes Project c onsortium, 2015). To select IVs for various subtypes of endometriosis, we set the sig - nificance level from a maximum of 5 × 10−6 to a minimum of 5 × 10−8 , depending on the actual situation (detailed in Table 1), using the same clumping procedure used for the metabolites. Furthermore, to assess the reliability and effectiveness of each SnP as a genetic IV in MR analysis, we calculated the F-statistic for each SnP using the formula F = R2(n − 2)/(1 − R 2). Here, R 2 represents the proportion of the variance in the exposure variable explained by IV, and n is the sample size of the original GWAS serving as the outcome variable. Specifically, R 2 is calculated using the formula: 21 21 21 2 2 2 × ×− () × × ×− () ×+ ][ × ×− () ×× ()( EAF EAF EAF EAF EAF EAF N SE β ββ ))    where eAF is the effect allele frequency, β is the SnP effect estimate, and Se(β) is the standard error of the SnP effect estimate. An IV with an F-statistic of less than 10 was considered unreliable and excluded from the subsequent MR analysis. These steps will ensure that our study uses high-quality genetic IVs for reliable analysis. MR analysis The statistical analysis for this study was performed using R software version 4.2.3, and the involved analytical methods were executed using the ‘MendelianRandomization’ package version 0.9.0. The pri - mary method employed was the inverse-variance weighted (IVW) method. Additionally, the MR-e gger and weighted-median methods were utilised to enhance the reliability of the results. To mitigate the impact of multiple testing, the P-values of the IVW method were adjusted using false discovery rate (FDR) correction. A result was considered positive if the corrected P-value was <0.05, even if the P-values Table 1. data sources and demographic profiles. exposures or outcome sample size (cases/controls) a ncestry sex Publication date(year) significance level GWas c atalogue id or url endometriosis asrM stages 1,2 5769/205101 european f emale 2022 5e-8 https://www.finngen.fi/fi endometriosis asrM stages 3,4 7574/203296 european f emale 2022 5e-8 https://www.finngen.fi/fi deep endometriosis 2856 /203296 european f emale 2022 5e-8 https://www.finngen.fi/fi endometriosis of fallopian tube 213/107564 european f emale 2022 5e-6 https://www.finngen.fi/fi endometriosis of intestine 436 /107564 european f emale 2022 5e-7 https://www.finngen.fi/fi endometriosis of ovary 5867 /107564 european f emale 2022 5e-8 https://www.finngen.fi/fi endometriosis of pelvic peritoneum 5628 /107564 european f emale 2022 5e-8 https://www.finngen.fi/fi endometriosis of rectovaginal septum and vagina 2456 /107564 european f emale 2022 5e-8 https://www.finngen.fi/fi metabolites na european f emale and male 2023 5e-6 PubMed id :36635386 JouRnAl oF oBSTe TRIcS AnD GynAecoloGy 5 from the MR-e gger and weighted-median methods were greater than 0.05, provided that the direction of the results for all three methods was consistent. Heterogeneity, pleiotropy, and sensitivity analysis Heterogeneity was tested using the IVW and MR-e gger methods. A P-value less than 0.05 indicated the presence of heterogeneity among SnPs, while a P-value greater than 0.05 indicated no heterogeneity. A sensitivity analysis was conducted using a leave-one-out approach to explore the impact of individual SnPs on causal associations. The horizontal pleiotropy was assessed using the intercept of the MR-e gger method. A significant intercept implied the existence of horizontal pleiotropy. In addition, the MR-PReSSo method was employed to exclude potential horizontal pleiotropic outliers that could significantly affect the MR anal - ysis results. For exposures with horizontal pleiotropy that could not be eliminated, the corresponding MR analysis for that exposure factor was disregarded. Metabolic pathway enrichment analysis Positive metabolites identified from the IVW analysis without correction for multiple testing were sub - jected to enrichment analysis using MetaboAnalyst 5.0 ( https://www.metaboanalyst.ca/) to explore asso - ciated metabolic pathways. The analysis utilised the Small Molecule Pathway Database (SMPDB) and Kyoto encyclopaedia of Genes and Genomes (KeGG) databases. The enrichment method employed was the Hypergeometric Test, with a significance level for metabolic pathway analysis set at 0.01. Colocalisation analysis To evaluate whether exposures and outcomes with a causal relationship might share common causal vari - ants, we conducted Bayesian colocalisation analysis using the ‘coloc’ package in R. The colocalisation region was defined within a 100-kb range up and down from the target SnP (Farh et  al. 2015). Five hypotheses were involved in the colocalisation analysis: (H0) no association with either trait within the selected location; (H1) association with trait 1 but not with trait 2 at the chosen locus; (H2) association with trait 2 but not with trait 1 at the selected site; (H3) association with both but with distinct causal variants; and (H4) associ - ation with both, sharing a common causal variant (Giambartolomei et  al. 2014). The essence of the colocal - isation analysis is to test the posterior probabilities of the five hypotheses. o ur primary focus is on the posterior probability of H4, where 0 indicates a 0% probability, and 1 indicates a 100% probability. A poste - rior probability of H4 greater than 0.8(Giambartolomei et  al. 2014) is considered indicative of colocalisation.

Exploration of the causal effect of metabolites on endometriosis and its subtypes IVW analysis without multiple-testing correction revealed that 59 metabolites were causally associated with stage 1–2 endometriosis, 73 with stage 3–4 endometriosis, 67 with deep endometriosis, 61 with fallopian tube endometriosis, 63 with intestinal endometriosis, 70 with ovarian endometriosis, 65 with pelvic peritoneal endometriosis, and 69 with rectovaginal septum and vaginal endometriosis. Supplementary Figure 1 shows the metabolites associated with three or more endometriosis subtypes. Supplementary Table 2 provides details regarding the metabolites associated with one to two subtypes of endometriosis. After FDR adjustment, only the glycerol-to-palmitoylcarnitine ( c16) ratio reduced the risk of stage 1–2 endometriosis ( PFDR = 0.045; odds ratio [ oR], 0.737; 95% confidence interval [ cI], 0.638–0.852) and pelvic peritoneal endometriosis ( PFDR = 0.039; oR, 0.721; 95% cI, 0.619–0.841). Specifically, each one standard deviation (SD) increase in this ratio was associated with an approximately 26% reduction in the odds of stage 1–2 disease and a 28% reduction in the odds of pelvic endometriosis. The protective effect of the 6 F. yAn eT Al. glycerol-to-palmitoylcarnitine ( c16) ratio against stage 1–2 endometriosis and pelvic peritoneal endome - triosis was supported by both the MR-e gger and weighted-median tests, which provided consistent effect directions and significance levels ( Figure 2 ). Additionally, the MR-e gger intercept and MR-PReSSo global test suggested no horizontal pleiotropy ( Table 2). The robustness of these causal relationships was validated using multiple analytical approaches. For stage 1–2 endometriosis, Supplementary Figure 3A (scatter plot) illustrates the correlation between the effects of SnPs on exposure and outcome. Supplementary Figure 3B (funnel plot) presents the heteroge - neity and potential outliers among the IVs. Supplementary Figure 3c (forest plot) shows consistent effect estimates across individual SnPs. l eave-one-out analysis confirmed that no single SnP disproportionately influenced the overall causal estimate Supplementary Figure 3D ). Similarly, corresponding analyses of pel - vic peritoneal endometriosis ( Supplementary Figure 4A–D ) revealed that all visualisations consistently sup - ported the protective effect of the glycerol-to-palmitoylcarnitine (c16) ratio in both endometriosis subtypes. Exploration of the causal effect of endometriosis and its subtypes on metabolites Reverse Mendelian analysis indicated the causal associations of stage 1–2 endometriosis with 15 metabolites, stage 3–4 endometriosis with 32 metabolites, deep endometriosis with 44 metabolites, fallopian tube endome- triosis with 3 metabolites, intestinal endometriosis with 4 metabolites, ovarian endometriosis with 44 metabo- lites, pelvic peritoneal endometriosis with 16 metabolites, and rectovaginal septum and vaginal endometriosis with 14 metabolites in the unadjusted IVW results(see Supplementary Table 3 for details). Supplementary Figure 2 shows the metabolites associated with three or more endometriosis subtypes. Supplementary Table 4 provides details regarding the metabolites related to one to two subtypes of endometriosis. no causal association was found between endometriosis and its subtypes or metabolites after FDR correction ( PFDR < 0.05).

of metabolic pathway enrichment analysis using the positive metabolites identified from the IVW analysis without correction for multiple testing, we conducted eight metabolic pathway enrichment analyses. With the significance level set at P < 0.01, the results highlighted major metabolic pathways, including caffeine metabolism, glutathione metabo - lism, arginine biosynthesis, sphingolipid metabolism, pantothenate and c oA biosynthesis, plasmalogen synthesis, and biosynthesis of unsaturated fatty acids (see Table 3 for details; complete enrichment anal - ysis results are available in Supplementary Tables 5–12 ). Figure 2. t his figure shows the Mendelian randomisation analysis results of the ratio of Glycerol to Palmitoylcarnitine (c16) with endometriosis stages 1–2 and pelvic peritoneal endometriosis. Table 2. Mr sensitivity analyses of genetically predicted metabolites on endometriosis and its subtypes. o utcome exposure Heterogeneity tests   directional horizontal pleiotropycest

c ochran’sQ(P)   Mr-e gger intercept ( P) Ppleiotropy* endometriosis stages 1–2 Glycerol to palmitoylcarnitine ( c16) ratio Mr e gger, iVW 14.42(0.49),17.01(0.38) −0.04(0.12) 0.43 endometriosis of pelvic peritoneum Glycerol to palmitoylcarnitine ( c16) ratio Mr e gger, iVW 14.74(0.46),18.22(0.31) −0.04(0.08) 0.38 *detect by Mr-Presso Global test; Mr-e gger, Mendelian randomisation-e gger; iVW, inverse-variance weighted. JouRnAl oF oBSTe TRIcS AnD GynAecoloGy 7

of colocalisation analysis c olocalisation analyses were conducted separately for the glycerol-to-palmitoylcarnitine ( c16) ratio in stage 1–2 endometriosis and pelvic peritoneal endometriosis. The posterior probabilities were 0.002 and 0.068, respectively, indicating no support for a shared causal variant between the glycerol-to-palmitoylcarnitine (c16) ratio and stage 1–2 endometriosis or pelvic peritoneal endometriosis ( Supplementary Table 13 and Figure 3A–B ).

In this study, we analysed the potential causal associations between 1,400 metabolites and endometrio - sis and its subtypes using the MR approach. After FDR correction for multiple testing, only the glycerol-to-palmitoylcarnitine ( c16) ratio remained statistically significant, showing potential protective effects against stage 1–2 endometriosis and pelvic peritoneal endometriosis. Sensitivity and pleiotropy analyses supported the robustness of this single significant finding. Reverse Mendelian analysis did not reveal any reverse causal associations. Metabolic pathway enrichment analysis based on uncorrected results suggested the potential involve - ment of pathways, including caffeine metabolism, glutathione metabolism, arginine biosynthesis, sphin - golipid metabolism, pantothenate and c oA biosynthesis, plasmalogen synthesis, and biosynthesis of unsaturated fatty acids. However, these pathway results should be interpreted cautiously given that most individual metabolite associations did not survive multiple-testing correction. c olocalisation analysis revealed low posterior probabilities for shared causal variants (H4 < 0.1) between glycerol-to-palmitoylcar - nitine ( c16) ratio and endometriosis outcomes, indicating limited evidence for genetic colocalisation. The glycerol-to-palmitoylcarnitine ( c16) ratio represents a meaningful metabolic readout that reflects the balance between lipolysis and fatty acid oxidation, two processes that are critically disrupted in the pathophysiology of endometriosis. This ratio provides insight into cellular energy metabolism dysfunc - tion, which is increasingly recognised as a hallmark of endometriosis progression. Palmitoylcarnitine, as an important fatty acid, ( c16) serves as a critical intermediate in mitochondrial fatty acid β-oxidation, facilitating the transport of palmitic acid across the mitochondrial membrane via the carnitine palmito - yltransferase ( cPT) system. emerging evidence suggests that endometriotic lesions exhibit significant mitochondrial dysfunction, characterised by impaired oxidative phosphorylation, reduced ATP produc - tion, and altered fatty acid metabolism (Atkins et  al. 2019) Recent studies have demonstrated that endo - metrium and endometriosis tissue from nonhuman primates with endometriosis showed decreased mitochondrial respiration and reduced complex I and II-mediated oxygen consumption rates compared to normal endometrium (Atkins et  al. 2019). notably, these studies revealed that carnitine levels signifi - cantly decreased in endometriotic tissues, suggesting fundamental disruptions in fatty acid metabolism pathways. The role of altered fatty acid metabolism in endometriosis is further supported by evidence of meta - bolic reprogramming in endometriotic lesions. endometriotic cells have been shown to strategically reduce energy production to avoid excessive mitochondrial R oS production, leading to a metabolic shift from oxidative phosphorylation to glycolysis (Kobayashi et  al. 2021). This metabolic conversion may result in the accumulation of fatty acid metabolites such as palmitoylcarnitine, reflecting the cells’ inability to efficiently utilise fatty acids for energy production. Glycerol, which is released during triglyceride Table 3. Metabolic pathway enrichment analysis results ( P < 0.01). Metabolic pathway trait database P c affeine metabolism endometriosis stages 1–2 sMPdb 0.0073 Glutathione metabolism endometriosis stages 1–2 KeGG 0.0009 a rginine biosynthesis endometriosis stages 1–2 KeGG 0.0046 deep endometriosis KeGG 0.0046 endometriosis of pelvic peritoneum KeGG 0.0046 sphingolipid metabolism deep endometriosis KeGG 0.0014 Pantothenate and c oa biosynthesis endometriosis of fallopian tube KeGG 0.0093 Plasmalogen synthesis endometriosis of intestine sMPdb 0.0062 biosynthesis of unsaturated fatty acids endometriosis of intestine KeGG 0.0001 8 F. yAn eT Al. breakdown, serves as both a gluconeogenic substrate and a marker of lipolytic activity. In endometriosis, dysregulated lipolysis may reflect metabolic reprogramming associated with chronic inflammation and tissue remodelling. Inflammatory cytokines prevalent in endometriosis, such as TnF-α and Il -1β, can sig - nificantly alter lipid metabolism and promote lipolysis (Feingold et  al. 1992, Plomgaard et  al. 2008). TnF-α stimulates lipolysis by activating hormone-sensitive lipase and adipocyte triglyceride lipase, leading to increased free fatty acid release (Plomgaard et  al. 2008). Therefore, the glycerol-to-palmitoylcarnitine ratio Figure 3. a: c o-localization analysis of Glycerol to Palmitoylcarnitine ( c16) ratio and s tage 1–2 endometriosis. b: c o-localization analysis of Glycerol to Palmitoylcarnitine ( c16) ratio and Pelvic Peritoneal endometriosis. JouRnAl oF oBSTe TRIcS AnD GynAecoloGy 9 may capture the imbalance between lipid breakdown (reflected by glycerol) and utilisation (reflected by palmitoylcarnitine), providing a composite biomarker for metabolic dysfunction. The protective effect of a higher glycerol-to-palmitoylcarnitine ratio suggests that enhanced lipolytic capacity relative to fatty acid oxidation demand may be metabolically beneficial in endometriosis. This find - ing aligns with the emerging evidence that interventions targeting fatty acid metabolism may have thera - peutic potential for endometriosis. Palmitoylethanolamide, a derivative of palmitic acid, has demonstrated anti-inflammatory and analgesic effects in endometriosis patients (Indraccolo and Barbieri 2010, Stochino-l oi et  al. 2019). These studies suggest that modulation of fatty acid metabolic pathways may represent a novel therapeutic approach for the management of endometriosis. Furthermore, this ratio may reflect the metabolic flexibility of tissues in adapting to the energy demands of chronic inflammation. Women with higher glycerol-to-palmitoylcarnitine ratio may exhibit greater metabolic resilience, enabling better adaptation to the inflammatory stress characteristic of endo - metriosis and reducing disease susceptibility. While this mechanistic framework provides biological plau - sibility for our findings, direct experimental validation of these proposed mechanisms is essential. Future studies should investigate the expression of key enzymes involved in fatty acid metabolism ( cPT1A, cPT2, and hormone-sensitive lipase) in endometriotic tissues and explore how alterations in the glycerol-to-palmitoylcarnitine ratio correlate with mitochondrial function parameters, inflammatory mark - ers, and clinical outcomes. Additionally, metabolomic studies of endometriotic tissue samples could pro - vide direct evidence for the proposed metabolic dysregulation and validate the systemic metabolic changes reflected in serum measurements. Previous research on the association between metabolites and endometriosis did not involve the glycerol-to-palmitoylcarnitine (c16) ratio. However, previous studies have revealed the significant roles of metabolites derived from lipid metabolism, gut microbiota, environmental exposure, and hormonal path - ways in the development, progression, and symptomatology of endometriosis. Sasamoto et  al. (Sasamoto et  al. 2022) found that the dysregulation of lipid metabolites in younger patients with endometriosis may be associated with persistent pelvic pain after surgery, highlighting the role of lipid metabolism in pain pathophysiology. l e et  al. (l e et  al. 2021) focused on the interaction between the gut microbiota and oestrogen metabolites, noting the association between dysbiosis and changes in urinary oestrogen levels in patients and emphasising the influence of the gut microbiota on disease progression. ermolova et  al. (eRMoloVA et  al. 2020) discussed the role of cytokines, nitric oxide metabolites, and lipid metabolism at different stages of endometriosis and suggested a complex interplay between these factors in the patho - genesis and progression of the disease. li (li et  al. 2020) found that exposure to specific pesticide metab - olites might increase the risk of endometriosis, indicating the potential role of environmental factors. chadchan ( chadchan et  al. 2023) demonstrated the contribution of the gut microbiota and their metab - olites to lesion growth, underlining the importance of the gut microbiome in disease progression. Additionally, emond et  al. (emond et  al. 2022) highlighted the significance of the oestradiol metabolic pathway in ovarian endometriosis and pain-related outcomes. The meta-analysis conducted by c ai et  al. (c ai et  al. 2019) indicated a potential association between certain phthalate metabolites (particularly MeHHP) and endometriosis, suggesting a link between environmental toxins and the disease. c ollectively, these studies demonstrate the potential roles of metabolites from lipid metabolism, gut microbiota, envi - ronmental exposure, and hormonal pathways in the development, progression, and symptomatology of endometriosis. Through enrichment analysis of the differential metabolites selected via Mendelian analysis, we iden - tified several key metabolic pathways. Interactions between these pathways reveal the complex mecha - nisms underlying endometriosis, highlighting the significant role of metabolic regulation in disease progression. Sphingolipid metabolism, which regulates fundamental processes, such as cell proliferation, migration, and apoptosis, plays a pivotal role in the development of endometriosis. Variations in the activity of key enzymes such as sphingomyelin synthase 1 (SMS1), sphingomyelin phosphodiesterase 3 (SMPD3), and glucosylceramide synthase (GcS) may reflect the involvement of sphingolipid metabolism in the pathogenesis of endometriosis, providing opportunities for developing new therapeutic targets (l ee et  al. 2014). Research on glutathione metabolism has revealed a complex relationship between the genetic predisposition to detoxification mechanisms and endometriosis. Specifically, studies on genetic polymorphisms in n-acetyltransferase 2 (nAT2) and glutathione S-transferases (GST) M1 and T1 have 10 F. yAn eT Al. provided strong evidence for the crucial role of detoxification pathways in disease onset (Baranova, 1999). Furthermore, analysis of caffeine metabolism highlights the role of lifestyle and dietary factors in the risk and severity of endometriosis. This necessitates further research to explore how caffeine and its metabolites interact with genetic susceptibility through pathways such as oestrogen regulation, thereby influencing the development of endometriosis. changes in arginine biosynthesis, particularly its upregu - lation in the serum of patients with endometriosis, provide insights into metabolic dysregulation associ - ated with the disease. Broader disruption of amino acid metabolism may be related to immune responses and angiogenesis in endometriosis, which merits further exploration (l etsiou et  al. 2017). Studies on fatty acid metabolism emphasise the potential role of lipid metabolic pathways in endometriosis, with research by Gassler et  al. highlighting the extensive impact of metabolic dysregulation in the disease’s pathology (Gassler et  al. 2005). The link between pantothenate (vitamin B5) and c oA biosynthesis and its associa - tion with endometriosis have not been exhaustively discussed in the existing literature. c oA, a central cofactor in various metabolic pathways, including fatty acid synthesis and oxidation, and the tricarboxylic acid (T c A) cycle, plays a key role in energy metabolism regulation, which could be crucial in the patho - logical environment of endometriosis. Disturbances in pantothenate and c oA biosynthesis, leading to changes in c oA levels, can affect the cellular energy balance, thereby contributing to the disease mech - anism. For instance, pantothenate kinase, which catalyses the initial step of c oA biosynthesis, is tightly regulated and its activity can influence metabolism and cellular homeostasis (l eonardi and Jackowski 2007). In contrast, plasmalogens, a class of ether phospholipids involved in membrane integrity and sig - nalling, have been implicated in oxidative stress and inflammation, which are key components of endo - metriosis. However, direct research linking plasmalogen synthesis with endometriosis is scarce. The significance of lipid metabolism in the pathophysiology of endometriosis suggests that changes in plas - malogen levels or synthesis may affect endometriosis (Zhang et  al. 2004, l eonardi and Jackowski 2007). c olocalisation analysis revealed low posterior probabilities for shared causal variants (H4 < 0.1) between glycerol-to-palmitoylcarnitine (c16) ratio and endometriosis outcomes. This finding presents an important

to our causal inference, as it suggests several alternative explanations for the observed MR associations. First, the low probability of colocalisation may indicate the presence of horizontal pleiotropy, where genetic variants affect both the metabolite ratio and endometriosis through independent bio - logical pathways rather than through a direct causal relationship. Second, it may reflect linkage effects, where the IVs are in linkage disequilibrium with the true causal variants affecting either the exposure or the outcome, but not necessarily both. Third, limited colocalisation evidence suggests that the genetic architectures underlying metabolite levels and endometriosis risk operate through distinct mechanisms. These findings urge caution when interpreting the MR results as definitive evidence of causality. While the MR analysis suggests a potential association, the lack of strong colocalisation evidence indicates that this relationship may be more complex than a simple direct causal pathway. Therefore, the observed association should be considered suggestive rather than confirmatory evidence and requires additional validation through alternative study designs and experimental approaches. o ur study employed several MR approaches to systematically evaluate the potential causal rela - tionships between 1,400 circulating metabolites and multiple endometriosis subtypes. c oncordant

across IVW, MR-e gger, and weighted-median estimators minimised the bias from horizontal pleiotropy and other confounders, thereby reinforcing the robustness of our conclusions. c ompared with previous work, we investigated a broader metabolite repertoire and finer disease classification. Recent pertinent studies are summarised in Supplementary Table 14 . o f particular interest, the sole signal that remained significant after FDR correction—that is, the glycerol-to-palmitoylcarnitine (c16) ratio—can be quantified rapidly through high-throughput lc–MS/MS and therefore has imme - diate translational potential. Pending replication, this ratio could support non-invasive risk stratifi - cation before laparoscopy and serve as a pharmacodynamic marker during treatment. Moreover, its linkage to lipid oxidation pathways provides a tentative basis for personalised management of endometriosis. However, this study has several limitations. First, the fact that only the glycerol/c16 ratio remained significant after multiple-testing correction underscores the exploratory nature of our findings and JouRnAl oF oBSTe TRIcS AnD GynAecoloGy 11 suggests that some initial associations may represent false positives. Second, the low posterior probabil - ity of genetic colocalisation with endometriosis outcomes (H4 < 0.1) weakens causal inference and raises the possibility that the observed association is driven by horizontal pleiotropy or linkage effects. Third, the analysis was confined to individuals of e uropean ancestry, limiting generalisability. Differences in allele frequencies, linkage-disequilibrium patterns, and genetic architecture across populations may have modified effect estimates. To address these gaps, future work should (i) replicate the glycerol/c16 signal across east-Asian, African, South-Asian and admixed populations to evaluate its universality or reveal ancestry-specific effects; (ii) perform longitudinal metabolomic profiling to track temporal changes in the ratio before symptom onset and throughout therapy; (iii) integrate eQTl, pQTl, organoid and animal-model data in multi-omics mediation and functional studies to elucidate underlying mechanisms and exclude pleiot - ropy; and (iv) initiate early-phase intervention trials targeting lipid-oxidation pathways, using the glyc - erol/c16 ratio as a responsive biomarker to explore novel therapeutic strategies.

o ur study provides preliminary evidence for the potential associations between metabolites and endo - metriosis subtypes through bidirectional MR analysis. However, the limited colocalisation evidence (H4 < 0.1) and the fact that only one metabolite ratio remained significant after FDR correction suggest that these relationships may involve complex mechanisms beyond simple causality, including potential horizontal pleiotropy or linkage effects. These findings should be interpreted as hypotheses rather than confirmatory evidence. Future studies with larger sample sizes and experimental validations are essential to confirm these preliminary results. This analysis may advance metabolomics-based research on the diagnosis and treatment of endometriosis. Author contributions cRediT: Fei Yan : c onceptualization, Writing – original draft; Zhouxiang Chen : Resources, Software, Supervision, Validation, Visualization, Writing – original draft; Lingfeng Wu : Formal analysis, Funding acquisition, Investigation, Methodology, Project administration; Zongju Huang : c onceptualization, Writing – original draft. Fei yan and Zhouxiang chen contributed equally to this work as co-first authors. Disclosure statement no potential conflict of interest was reported by the authors. Funding This study received no external funding or financial support from public, commercial, or not-for-profit funding bodies. Data availability statement The datasets analysed in this study were obtained from publicly available genome-wide association studies (GWAS), as described in detail in the Methods section. Summary-level statistics supporting the findings of this study are available from the corresponding author upon reasonable request. Access to individual-level genetic data is restricted in accordance with ethical guidelines to protect participant privacy and confidentiality. This study utilised only sum - mary data from previously published studies, all of which had obtained appropriate ethical approvals for data col - lection and sharing. o ur research protocol adhered to all applicable ethical standards governing human genetic research and data sharing, including considerations for informed consent and privacy protection. The specific data sources for metabolite and endometriosis phenotypes, along with their accession numbers and uRls, are listed in the Supplementary Materials . Due to the volume of supplementary data involved, some files are not directly down - loadable but can be made available upon request from the authors. 12 F. yAn eT Al.

Atkins, H.M., et al. , 2019. endometrium and endometriosis tissue mitochondrial energy metabolism in a nonhuman primate model. Reproductive Biology and Endocrinology: RB&E , 17 (1), 70. Baranova, H., et al. , 1999. Possible involvement of arylamine n-acetyltransferase 2, glutathione S-transferases M1 and T1 genes in the development of endometriosis. Molecular Human Reproduction , 5 (7), 636–641., Burgess, S. and Thompson, S.G., 2015. Mendelian randomization: methods for using genetic variants in causal estimation . cR c Press, Boca Raton, Fl. Burney, R.o . and Giudice, l.c., 2012. Pathogenesis and pathophysiology of endometriosis. Fertility and Sterility , 98 (3), 511–519. c ai, W., et al., 2019. Association between phthalate metabolites and risk of endometriosis: a meta-analysis. International Journal of Environmental Research and Public Health , 16 (19), 3678. chadchan, S.B., et al. , 2023. Gut microbiota and microbiota-derived metabolites promotes endometriosis. Cell Death Discovery, 9 (1), 28. chen, y ., et al., 2023. Genomic atlas of the plasma metabolome prioritizes metabolites implicated in human diseases. Nature Genetics , 55 (1), 44–53. The 1000 Genomes Project c onsortium, 2015. A global reference for human genetic variation. Nature, 526 (7571), 68–74. c ounseller, V.S. and c renshaw, J.l., 1951. A clinical and surgical review of endometriosis. American Journal of Obstetrics and Gynecology , 62 (4), 930–942. Davey Smith, G. and Hemani, G., 2014. Mendelian randomization: genetic anchors for causal inference in epidemio - logical studies. Human Molecular Genetics , 23 (R1), R89–R98. Dutta, M., et al. , 2018. Metabolomics reveals perturbations in endometrium and serum of minimal and mild endo - metriosis. Scientific Reports , 8 (1), 6466. emond, J.-P ., et al. , 2022. c irculating estradiol and its biologically active metabolites in endometriosis and in relation to pain symptoms. Frontiers in Endocrinology , 13, 1034614. ermolova, n.V., et al., 2020. Influence of cytokines production, nitrogen oxide metabolites and lipids exchange on the formation of external genital endometriosis stages in patients of reproductive age. Periódico Tchê Química , 17 (35), 813–824. Farh, K.K.-H., et al. , 2015. Genetic and epigenetic fine mapping of causal autoimmune disease variants. Nature, 518 (7539), 337–343. Feingold, K.R., et al., 1992. Stimulation of lipolysis in cultured fat cells by tumor necrosis factor, interleukin-1, and the interferons is blocked by inhibition of prostaglandin synthesis. Endocrinology, 130 (1), 10–16. Gassler, n., et al. , 2005. expression of acyl- c oA synthetase 5 in human endometrium and in endometrioid adenocar - cinomas. Histopathology, 47 (5), 501–507. Giambartolomei, c., et al. , 2014. Bayesian test for colocalisation between pairs of genetic association studies using summary statistics. PLoS Genetics , 10 (5), e1004383. Haycock, P .c., et al. , 2016. Best (but oft-forgotten) practices: the design, analysis, and interpretation of Mendelian randomization studies. The American Journal of Clinical Nutrition , 103 (4), 965–978. Horne, A.W. and Saunders, P .T.K., 2019. SnapShot: endometriosis. Cell, 179 (7), 1677–1677.e1. Indraccolo, u . and Barbieri, F., 2010. effect of palmitoylethanolamide–polydatin combination on chronic pelvic pain associated with endometriosis: preliminary observations. European Journal of Obstetrics, Gynecology, and Reproductive Biology, 150 (1), 76–79. Jana, S.K., et al. , 2013. 1H nMR based targeted metabolite profiling for understanding the complex relationship con - necting oxidative stress with endometriosis. BioMed Research International , 2013, 329058–329059. Kobayashi, H., Shigetomi, H. and Imanaka, S., 2021. nonhormonal therapy for endometriosis based on energy me - tabolism regulation. Reproduction & Fertility , 2 (4), c42– c57., l e, n., et al. , 2021. Association of microbial dynamics with urinary estrogens and estrogen metabolites in patients with endometriosis. PloS One , 16 (12), e0261362. l ee, y .H., et al. , 2014. Dysregulated sphingolipid metabolism in endometriosis. The Journal of Clinical Endocrinology and Metabolism , 99 (10), e1913–e1921. l eonardi, R. and Jackowski, S., 2007. Biosynthesis of pantothenic acid and coenzyme A. EcoSal Plus , 2 (2), eco - salplus.3.6.3.4. l etsiou, S., et al. , 2017. endometriosis is associated with aberrant metabolite profiles in plasma. Fertility and Sterility , 107 (3), 699–706.e6. li, A.J., et al., 2020. Association of urinary metabolites of organophosphate and pyrethroid insecticides, and phenoxy herbicides with endometriosis. Environment International , 136, 105456. li, B., et al., 2023. Association between endometriosis and metabolic syndrome: a cross-sectional study based on the national Health and nutrition examination Survey data. Gynecological Endocrinology , 39 (1), 2254844. l u, J., et al. , 2023. emerging hallmarks of endometriosis metabolism: a promising target for the treatment of endo - metriosis. Biochimica et Biophysica Acta. Molecular Cell Research , 1870 (1), 119381. JouRnAl oF oBSTe TRIcS AnD GynAecoloGy 13 negar, G., et al., 2016. 1H nMR-based metabolomics approaches as non-invasive tools for diagnosis of endometriosis. International Journal of Reproductive Biomedicine , 14, 1. olive, D.l. and Pritts, e .A., 2001. Treatment of endometriosis. The New England Journal of Medicine , 345 (4), 266–275. orrù, V., et al. , 2020. c omplex genetic signatures in immune cells underlie autoimmunity and inform therapy. Nature Genetics, 52 (10), 1036–1045. ortiz, c. n., Torres-Reverón, A. and Appleyard, c. B., 2021. Metabolomics in endometriosis: challenges and perspec - tives for future studies. Reproduction & Fertility , 2 (2), R35–R50., Plomgaard, P ., et al. , 2008. Tumor necrosis factor-alpha modulates human in vivo lipolysis. The Journal of Clinical Endocrinology and Metabolism , 93 (2), 543–549. Sasamoto, n., et al. , 2022. Presurgical blood metabolites and risk of postsurgical pelvic pain in young patients with endometriosis. Fertility and Sterility , 117 (6), 1235–1245. Sinaii, n., et al. , 2008. Differences in characteristics among 1,000 women with endometriosis based on extent of disease. Fertility and Sterility , 89 (3), 538–545. Soliman, A. M., Fuldeore, M. and Snabes, M. c., 2017. Factors associated with time to endometriosis diagnosis in the united States. Journal of Women’s Health (2002) , 26 (7), 788–797., Stochino-l oi, e ., et al. , 2019. effect of ultramicronized-palmitoylethanolamide and co-micronized palmitoylethanol - amide/polydatin on chronic pelvic pain and quality of life in endometriosis patients: an open-label pilot study. International Journal of Women’s Health , 11, 443–449. Wang, M., et al. , 2021. c arnitine palmitoyltransferase system: a new target for anti-inflammatory and anticancer ther - apy? Frontiers in Pharmacology , 12, 760581. yu, X.-H., et al. , 2021. The causal role of gut microbiota in development of osteoarthritis. Osteoarthritis and Cartilage , 29 (12), 1741–1750. Zhang, y.-M., et al. , 2004. Acyl carrier protein is a cellular target for the antibacterial action of the pantothenamide class of pantothenate antimetabolites. The Journal of Biological Chemistry , 279 (49), 50969–50975. Angioni, S., et al. , 2021. Metabolomic characteristics in endometriosis patients. In: A.R. Genazzani, M. n isolle, F. Petraglia, and R.n. Taylor, eds. Endometriosis pathogenesis, clinical impact and management. ISGe Series. Springer, cham, Switzerland. Adamyan, l., et al., 2024. Metabolomic biomarkers of endometriosis: a systematic review. Journal of Endometriosis and Uterine Disorders , 7, 100077.