Causal relationships between metabolic syndrome, plasma metabolites, and female reproductive diseases: insights from a two-step mendelian randomization approach

In this study, the risk of MetS and the plasma circulating level of each metabolite were treated as exposures, while the risk of female reproductive diseases was considered the outcome. SNPs that were strongly associated with the exposures were employed as IVs [ 31 ]. To ensure validity in the MR analysis, all selected IVs were required to meet MR assumptions: (1) they must be directly linked to the exposures, (2) they should exhibit no correlation with any potential confounding variable of the exposure–outcome association, and (3) they should influence the outcomes exclusively through their effects on the exposures [ 32 ]. For the eight female reproductive diseases, GWAS summary data were procured from the FinnGen research project ( https://www.finngen.fi/en/access_results ) [ 33 ]. FinnGen serves as an excellent resource for GWAS summary data due to its extensive public‒private sample size, diverse dataset, and stringent data quality control standards, which facilitate the identification of disease causes and contribute to population health improvements [ 34 , 35 ]. For ovarian aging, SNP associations were procured from the study by Ruth et al., involving a total of 201,323 individuals [ 36 ]. For MetS, GWAS summary data were retrieved from the Complex Trait Genetics Lab (CTGLAB) and the Walree et al. study, which examined the genetic architecture of MetS among 461,920 participants of European descent [ 37 , 38 ]. Plasma metabolites were analyzed using a GWAS dataset derived from recently published work by Karjalainen et al. [ 39 ]. This dataset included 136,016 participants from 33 cohorts with available NMR metabolic trait measurements and genome-wide SNP data. The analysis revealed associations with 233 plasma metabolic traits, comprising 213 lipid and lipoprotein parameters or fatty acids, and 20 nonlipid traits categorized into amino acids, ketone bodies, glycolysis/gluconeogenesis, fluid balance, and inflammation-related metabolites. The exposure GWAS and outcome GWAS datasets were sourced from distinct cohorts to minimize bias stemming from overlapping samples. Additional details regarding these datasets are depicted in Additional file 1 : Table S1. To increase the precision and robustness of the study outcomes, various quality control measures were implemented during the genetic instrument selection process. For significant SNP selection, a stringent threshold of genome-wide significance ( P  < 5 × 10⁻⁸) was employed [ 40 , 41 ]. To satisfy the second MR hypothesis, the LDtrait ( https://ldlink.nih.gov/?tab=home ) database was utilized to ascertain all eligible SNPs [ 42 ]. Subsequently, to ensure the mutual independence of IVs, SNPs were selected using a clumping procedure, which employed a linkage disequilibrium (LD) threshold of r 2  < 0.001 within a 10,000-kilobase clumping window size [ 43 – 45 ]. The variance in plasma metabolite levels attributable to individual SNPs was quantified using the formula [ 31 ]: R 2  = [2β 2  × EAF × (1-EAF)]/[2β 2  × EAF × (1-EAF) + 2N × EAF × (1-EAF) × SE 2 ], where β and SE denote the effect size and standard error of the SNP-metabolite genetic effect, respectively, EAF denotes the effect allele frequency, and N is the sample size. The IV-exposure association strength was evaluated by computing the F value through the following formula [ 31 ]: F = [R 2  × (N-2)]/(1-R 2 ). The potential for weak IV bias was minimal when the F-statistic exceeded 10, and weak SNPs with F < 10 were excluded as poor genetic instruments [ 45 – 47 ]. The statistical power values for the MR estimates were computed employing an online tool ( https://shiny.cnsgenomics.com/mRnd/ ), with a power > 0.8 considered to indicate statistical significance [ 48 ]. When certain SNPs were missing from the outcome summary data, proxy SNPs were not employed, as the limited proportion of missing SNPs was unlikely to influence the results markedly [ 49 ]. The minor allele frequency threshold was set to less than 0.01. SNPs with inconsistent alleles, such as A/C paired with A/G between the exposure and outcome datasets, or palindromic alleles, such as A/T or G/C, were excluded during harmonization [ 43 , 47 ]. To assess the causative links between female reproductive diseases and MetS or metabolites, odds ratio (OR) coupled with corresponding 95% confidence interval (CI) were selected as the primary metrics for assessing the effects of these diseases. Various methodologies, including the IVW method [ 50 ], the weighted median (WM) method [ 51 ], and the MR-Egger method [ 52 ], have been implemented. The IVW method, recognized as the most reliable estimation approach for valid SNPs and unaffected by pleiotropic effects [ 31 , 50 , 53 ], served as the primary screening technique in two-step MR analyses. Additionally, the WM and MR-Egger approaches were employed as supplementary methods to enhance the IVW by accommodating invalid SNPs and identifying horizontal pleiotropy [ 49 , 53 ]. Unlike the IVW and WM methods, MR-Egger regression maintains the ability to provide consistent causal inferences even in scenarios where all IVs are invalid [ 54 ]. Unlike the IVW method, which assumes no horizontal pleiotropy, the MR-Egger method assumes that all IVs exhibit horizontal pleiotropy [ 55 ], while the WM method allows for up to 50% of IVs to exhibit horizontal pleiotropy [ 54 ]. Hence, in cases where horizontal pleiotropy was detected among SNPs, the outcomes derived from the MR-Egger and WM methods were utilized as references [ 56 ]. We anticipated that the effect estimates would be consistent in both direction and magnitude across all three MR analysis methods (IVW, WM, and MR-Egger methods) [ 31 , 47 , 49 , 57 ]. Ultimately, P  < 0.05 indicated statistical significance according to the IVW [ 58 ]. Moreover, the Benjamini–Hochberg (BH) method was employed to adjust for multiple tests and confirm statistically significant causal links between exposure and outcome [ 59 , 60 ]. To further detect horizontal pleiotropy outliers and refine estimates, the Mendelian randomization pleiotropy residual sum and outlier (MR-PRESSO) method was employed, considering significance at P  < 0.05 [ 61 ]. A range of supplementary and robustness evaluations, such as the Cochran-Q test, MR-PRESSO global test, MR-Egger intercept, leave-one-out (LOO) analysis, MR-Steiger test, and reverse MR analyses, were also conducted to evaluate the validity and robustness of the associations discovered via IVW analysis [ 34 , 47 , 62 ]. Initially, the IVW method, known for its sensitivity to heterogeneity and outliers, was recognized as being susceptible to horizontal pleiotropy [ 58 , 63 ]. During this phase, Cochran’s Q test was applied to detect potential heterogeneity, with heterogeneity indicated when the Cochran-Q-derived P value was < 0.05 [ 47 , 64 ]. Consequently, IVW estimates incorporated fixed-effect and random-effect models, where the fixed-effect model was utilized in the absence of heterogeneity, while the random-effect model was adopted when heterogeneity was present [ 47 ]. Additionally, the MR-Egger method was used to assess directional pleiotropy in IVs by providing an intercept term to indicate its magnitude and direction [ 52 , 65 ]. A P value < 0.05 was interpreted as indicating notable directional pleiotropy [ 44 ]. Additionally, the MR-PRESSO global test was implemented to assess the impact of pleiotropy on causal inference [ 31 , 61 ]. Furthermore, LOO analysis was performed by systematically removing individual IVs [ 66 ] and recalculating MR results for the remaining SNPs to ensure that no single SNP unduly influenced the observed associations [ 52 ]. Finally, to verify the correctness of causal directionality and mitigate the risk of reverse causality [ 56 ], the MR-Steiger test was utilized: female reproductive diseases were treated as the exposure, and MetS and metabolites were treated as the outcomes [ 67 , 68 ]. To ensure the reliability of the findings, a set of rigorous criteria was established to guide the identification of potentially appropriate candidate IVs linked to female reproductive diseases: (1) The MR results were required to confirm the absence of horizontal pleiotropy and heterogeneity, indicated by Egger P for an intercept > 0.05, the MR-PRESSO global test P  > 0.05, the Cochran’s Q test P  > 0.05, and Steiger P  < 0.05; (2) LOO analysis was utilized to ensure that the MR estimates were neither markedly influenced nor biased by any single SNP. To externally validate the associations between plasma metabolites and outcomes reported by Karjalainen et al., a replication analysis was performed. Recently, Chang et al. introduced mGWAS-Explorer 2.0, a tool designed to facilitate the exploration of findings from various mGWASs [ 69 ]. To ensure consistency in metabolite sources, GWAS summary data on blood-derived metabolites from independent samples were retained for replication analysis. Furthermore, the IVW method was employed, with the same threshold criteria applied as in the initial analysis [ 70 ]. Bayesian colocalization analysis was executed using the R package coloc ( https://chr1swallace.github.io/coloc/ ) to explore whether the identified female reproductive disease-related metabolites and corresponding reproductive diseases share a common causal genetic variant. This approach complements MR by addressing its inherent limitations concerning pleiotropy and LD [ 71 , 72 ]. The colocalization analysis relied on GWAS summary statistics encompassing all variants within a ± 250 kb window around each genetic instrument. The default priors of coloc ( p 1  = 1 × 10 ⁻4 , p 2  = 1 × 10 ⁻4 , and p 12  = 1 × 10 ⁻5 ) were applied [ 18 ]. A Bayesian framework was used to calculate posterior probabilities for five mutually exclusive hypotheses regarding causal variance sharing between two traits [ 73 ]: (1) H 0 : neither trait is associated; (2) H 1 : only trait 1 is associated; (3) H 2 : only trait 2 is associated; (4) H 3 : both traits are associated but via distinct SNPs; and (5) H 4 : both traits are associated, sharing a single causal SNP [ 74 ]. A high posterior probability (PP) of H 4 (PPH 4  > 0.80) served as the established threshold for robust evidence of colocalization, whereas a PPH 4  > 0.50 indicated suggestive evidence [ 18 ].

In the initial step of MR analysis, a comprehensive screening process was undertaken to identify genetic instruments for MetS (Additional File 1 : Table S2). A total of 193 SNPs were strongly associated with MetS, while the number of SNPs shared between MetS and the outcomes ranged from 146 to 180. Notably, all the genetic instruments demonstrated statistical power exceeding 0.8 and an F-statistic surpassing 10, signifying a robust link between the IVs and the exposure variables and validating their suitability for inclusion in the MR analysis. Further details regarding the SNPs associated with MetS are depicted in Additional File 1 : Table S2. As presented in Table  1 and Fig.  2 , the primary MR analyses demonstrated that MetS was linked to increased UF risk (OR, 1.14; 95% CI, 1.05–1.25; P  = 0.003, P BH  = 0.017), PCOS (OR, 3.35; 95% CI, 2.37–4.75; P  < 0.001, P BH  < 0.001), GDM (OR, 1.82; 95% CI, 1.57–2.10; P  < 0.001, P BH  < 0.001), eclampsia (OR, 1.90; 95% CI, 1.61–2.24; P  < 0.001, P BH  < 0.001), and miscarriage (OR, 1.11; 95% CI, 1.03–1.19; P  = 0.004, P BH  = 0.021). All three primary analyses yielded consistent estimates in the same direction. Additionally, heterogeneity and pleiotropy tests suggested no evidence of heterogeneity pleiotropy, while the Steiger test indicated significant differences (Additional File 1 : Table S3). For other conditions, the results suggested no causal relationships between MetS and EMs, EP, infertility, or ovarian aging (all P  > 0.05). Sensitivity analyses confirmed the absence of heterogeneity and pleiotropy (Table  1 and Fig.  2 ), and the Steiger test further revealed significant differences (Additional File 1 : Table S3). Collectively, these observations indicate that MetS increases the risk of UF, PCOS, GDM, eclampsia, and miscarriage. Table 1 Mendelian randomization (MR) results for associations between MetS and nine female reproductive diseases Female Reproductive Diseases Exposure Number of IVs IVW Weighted median Egger regression MR-PRESSO OR (95% CI) P P BH OR (95% CI) P OR (95% CI) P OR (95% CI) P EMs MetS 154 1.01(0.89–1.15) 0.883 0.883 0.81(0.64–1.02) 0.070 0.63(0.45–0.88) 0.007 1.01(0.89–1.15) 0.910 UF MetS 174 1.14(1.05–1.25) 0.003 0.017 1.00(0.88–1.15) 0.965 0.92(0.72–1.17) 0.488 1.14(1.05–1.25) 0.002 PCOS MetS 180 3.35(2.37–4.75)  < 0.001  < 0.001 3.09(1.77–5.38)  < 0.001 1.99(0.77–5.18) 0.159 3.35(2.37–4.75)  < 0.001 GDM MetS 164 1.82(1.57–2.10)  < 0.001  < 0.001 2.22(1.75–2.82)  < 0.001 2.50(1.68–3.72)  < 0.001 1.82(1.57–2.10)  < 0.001 Eclampsia MetS 170 1.90(1.61–2.24)  < 0.001  < 0.001 2.02(1.56–2.60)  < 0.001 2.54(1.59–4.07)  < 0.001 1.90(1.61–2.24)  < 0.001 EP MetS 174 1.20(1.00–1.43) 0.053 0.213 1.11(0.82–1.50) 0.486 0.93(0.57–1.52) 0.772 1.20(1.00–1.43) 0.053 Infertility MetS 170 1.01(0.89–1.14) 0.876 0.883 0.84(0.69–1.03) 0.096 0.67(0.48–0.93) 0.019 1.01(0.89–1.14) 0.876 Miscarriage MetS 163 1.11(1.03–1.19) 0.004 0.021 1.09(0.98–1.21) 0.098 0.86(0.69–1.07) 0.182 1.11(1.03–1.19) 0.004 Ovarian aging MetS 146 0.95(0.83–1.08) 0.418 0.883 0.92(0.75–1.12) 0.401 1.07(0.67–1.71) 0.772 0.95(0.83–1.08) 0.418 Abbreviations: EMs endometriosis, UF uterine fibroids, PCOS polycystic ovary syndrome, GDM gestational diabetes mellitus, EP ectopic pregnancy, IVs number of instrumental variables, OR odds ratio per standard deviation (SD), genetically predicted plasma level, CI confidence interval, IVW inverse variance,  MR-PRESSO Mendelian randomization pleiotropy residual sum and outlier, P BH Benjamini‒Hochberg correction P value within each female reproductive disease type Fig. 2 Forest plot for associations between MetS and female reproductive diseases. MetS was significantly associated with the risk of five female reproductive diseases (UF, PCOS, GDM, eclampsia, and miscarriage). Blue ( P  > 0.05 and HR > 1) indicates a nonsignificant association where the hazard ratio suggests a potential increase in risk, but the result is not statistically significant. Green ( P  > 0.05 and HR < 1) indicates a nonsignificant association where the hazard ratio suggests a potential decrease in risk, but the result is not statistically significant. Yellow ( P   1) indicates a statistically significant association with increased risk. Abbreviations: EMs, endometriosis; UF, uterine fibroids; PCOS, polycystic ovary syndrome; GDM, gestational diabetes mellitus; EP, ectopic pregnancy Mendelian randomization (MR) results for associations between MetS and nine female reproductive diseases Abbreviations: EMs endometriosis, UF uterine fibroids, PCOS polycystic ovary syndrome, GDM gestational diabetes mellitus, EP ectopic pregnancy, IVs number of instrumental variables, OR odds ratio per standard deviation (SD), genetically predicted plasma level, CI confidence interval, IVW inverse variance,  MR-PRESSO Mendelian randomization pleiotropy residual sum and outlier, P BH Benjamini‒Hochberg correction P value within each female reproductive disease type Forest plot for associations between MetS and female reproductive diseases. MetS was significantly associated with the risk of five female reproductive diseases (UF, PCOS, GDM, eclampsia, and miscarriage). Blue ( P  > 0.05 and HR > 1) indicates a nonsignificant association where the hazard ratio suggests a potential increase in risk, but the result is not statistically significant. Green ( P  > 0.05 and HR < 1) indicates a nonsignificant association where the hazard ratio suggests a potential decrease in risk, but the result is not statistically significant. Yellow ( P   1) indicates a statistically significant association with increased risk. Abbreviations: EMs, endometriosis; UF, uterine fibroids; PCOS, polycystic ovary syndrome; GDM, gestational diabetes mellitus; EP, ectopic pregnancy Previous research has indicated that MetS is linked to distinctive plasma metabolomic profiles characterized by elevated levels of proinflammatory lysoPCs, various amino acid groups, glycoproteins, urolithin A glucuronide, and phospholipids (PLs), as well as reduced levels of antioxidative ether PCs, hydroxydecanoyl carnitine, methylglutarylcarnitine and acyl-alkyl-phosphatidylcholine in plasma, among other metabolites [ 75 – 79 ]. Based on these findings, it was hypothesized that plasma metabolites contribute to female reproductive diseases through their association with MetS, leading to further MR analysis. In the second-step MR analysis, genetic instruments consisting of 4–52 SNPs for 179 metabolites were derived from 13,389,637 variants associated with 233 metabolites. After Benjamini‒Hochberg (BH) multiple testing correction ( P  < 0.05), a total of 31 significant associations were ascertained for five types of MetS-related female reproductive diseases. These included 19 associations for eclampsia, 10 for PCOS, and 2 for GDM (Fig.  3 , Table  2 , and Additional File 1 : Tables S4–S5). The identified metabolites encompassed 8 distinct types: 4 lipids (lipids in HDL, lipids in VLDL, conjugated linoleic acid [CLA], and estimated degree of unsaturation [UnsatDeg]), 2 ketone bodies (3-hydroxybutyrate [3-HB] and acetone), and 2 glucose metabolism compounds (lactate and glucose). Among these, lipids in high-density lipoprotein (HDL) were linked to the risk of two distinct reproductive diseases (L-HDL-TG for eclampsia and HDL-TG for PCOS). Additionally, the remaining types were associated with specific female reproductive diseases, including 1 for eclampsia, 4 for PCOS, and 2 for GDM. Fig. 3 Forest plot for significant associations between serum metabolites and MetS-related female reproductive diseases. Eight types of serum metabolites were significantly associated with the risk of 3 MetS-related female reproductive diseases (eclampsia, PCOS, and GDM). Yellow ( P   1) indicates a statistically significant association with increased risk. Red ( P  < 0.05 and HR < 1) indicates a statistically significant association with a decreased risk. Abbreviations: PCOS, polycystic ovary syndrome; GDM, gestational diabetes mellitus; HDL-D, mean diameter for HDL particles; S-HDL-P, concentration of small HDL particles; S-HDL-L, total lipids in small HDL; S-HDL-C, total cholesterol in small HDL; S-HDL-PL, phospholipids in small HDL; L-HDL-P, concentration of large HDL particles; L-HDL-L, total lipids in large HDL; L-HDL-C, total cholesterol in large HDL; L-HDL-PL, phospholipids in large HDL; L-HDL-CE, cholesterol ester in large HDL; L-HDL-FC, free cholesterol in large HDL; L-HDL-TG, triglycerides in large HDL; XL-HDL-P, concentration of very large HDL particles; XL-HDL-L, total lipids in very large HDL; XL-HDL-C, total cholesterol in very large HDL; XL-HDL-PL, phospholipids in very large HDL; XL-HDL-CE, cholesterol ester in very large HDL; XL-HDL-FC, free cholesterol in very large HDL; UnsatDeg, estimated degree of unsaturation; VLDL-D, mean diameter for VLDL particles; L-VLDL-P, concentration of large VLDL particles; XL-VLDL-PL, phospholipids in chylomicrons and extremely large VLDL; XL-VLDL-FC, free cholesterol in chylomicrons and extremely large VLDL; XL-VLDL-P, concentration of very large VLDL particles; XL-VLDL-TG, triglycerides in very large VLDL; CLA, conjugated linoleic acid; HDL-TG, triglycerides in HDL; 3-HB, 3-hydroxybutyrate Table 2 MR results for plasma metabolites significantly associated with the risk of female reproductive diseases Metabolite ID Annotation Number of IVs IVW Weighted median Egger regression MR-PRESSO OR (95% CI) P P BH OR (95% CI) P OR (95% CI) P OR (95% CI) P Eclampsia  mgwas-metabolite-2787 HDL-D 44 0.82(0.73,0.92) 0.001 0.012 0.78(0.70,0.87) 0.001 0.77(0.65,0.92) 0.003 0.82(0.73,0.92)  <  0.001  mgwas-metabolite-6353 S-HDL-P 18 1.31(1.07,1.60) 0.009 0.050 1.33(1.09,1.63) 0.005 1.29(0.94,1.78) 0.118 1.31(1.07,1.60)  <  0.001  mgwas-metabolite-6352 S-HDL-L 19 1.32(1.07,1.63) 0.009 0.050 1.34(1.09,1.64) 0.005 1.33(0.96,1.85) 0.091 1.32(1.07,1.63)  <  0.001  mgwas-metabolite-6255 S-HDL-C 21 1.21(1.01,1.44) 0.037 0.050 1.11(0.89,1.39) 0.381 1.2(0.90,1.59) 0.200 1.21(1.01,1.44) 0.101  mgwas-metabolite-6263 S-HDL-PL 27 1.16(1.00,1.34) 0.050 0.050 1.23(1.03,1.47) 0.026 1.07(0.86,1.33) 0.552 1.16(1.00,1.34) 0.027  mgwas-metabolite-3023 L-HDL-P 45 0.87(0.78,0.97) 0.009 0.050 0.99(0.86,1.14) 0.854 0.91(0.78,1.06) 0.242 0.87(0.78,0.97) 0.454  mgwas-metabolite-3020 L-HDL-L 44 0.87(0.78,0.97) 0.011 0.050 0.99(0.86,1.14) 0.852 0.9(0.77,1.05) 0.194 0.87(0.78,0.97) 0.509  mgwas-metabolite-3016 L-HDL-C 47 0.86(0.77,0.96) 0.008 0.050 0.98(0.85,1.13) 0.832 0.91(0.78,1.06) 0.233 0.86(0.77,0.96) 0.493  mgwas-metabolite-3024 L-HDL-PL 45 0.87(0.78,0.97) 0.017 0.050 0.99(0.86,1.15) 0.850 0.91(0.78,1.07) 0.274 0.87(0.78,0.97) 0.248  mgwas-metabolite-3018 L-HDL-CE 46 0.87(0.78,0.97) 0.013 0.050 0.99(0.86,1.14) 0.847 0.89(0.77,1.03) 0.122 0.87(0.78,0.97) 0.660  mgwas-metabolite-3019 L-HDL-FC 46 0.86(0.77,0.97) 0.009 0.050 0.98(0.85,1.13) 0.828 0.91(0.78,1.06) 0.247 0.86(0.77,0.97) 0.508  mgwas-metabolite-2935 L-HDL-TG 21 1.31(1.00,1.71) 0.023 0.050 0.87(0.77,0.98) 0.026 0.83(0.69,1.00) 0.053 0.86(0.75,0.98)  <  0.001  mgwas-metabolite-6983 XL-HDL-P 52 0.81(0.72,0.92) 0.001 0.015 0.83(0.72,0.96) 0.011 0.88(0.76,1.02) 0.085 0.81(0.72,0.92) 0.004  mgwas-metabolite-6979 XL-HDL-L 50 0.80(0.71,0.91)  < 0.001 0.009 0.77(0.67,0.89)  <  0.001 0.8(0.68,0.95) 0.008 0.80(0.71,0.91)  <  0.001  mgwas-metabolite-6976 XL-HDL-C 44 0.80(0.70,0.92) 0.001 0.015 0.75(0.63,0.89) 0.001 0.85(0.72,1.01) 0.064 0.80(0.70,0.92) 0.029  mgwas-metabolite-6984 XL-HDL-PL 50 0.79(0.70,0.89)  < 0.001 0.002 0.80(0.70,0.92) 0.001 0.81(0.69,0.95) 0.008 0.79(0.70,0.89)  <  0.001  mgwas-metabolite-6977 XL-HDL-CE 38 0.80(0.69,0.92) 0.002 0.022 0.75(0.63,0.89) 0.001 0.83(0.69,0.99) 0.043 0.8(0.69,0.92) 0.014  mgwas-metabolite-6978 XL-HDL-FC 51 0.81(0.71,0.92) 0.001 0.015 0.81(0.70,0.94) 0.006 0.86(0.73,1.01) 0.067 0.81(0.71,0.92) 0.028  mgwas-metabolite-2378 UnsatDeg 17 0.88(0.80,0.97) 0.010 0.050 0.87(0.78,0.96) 0.010 0.85(0.75,0.96) 0.014 0.88(0.80,0.97)  <  0.001 PCOS  mgwas-metabolite-6734 VLDL-D 34 1.42(1.13,1.78) 0.002 0.025 1.42(1.05,1.92) 0.024 1.06(0.74,1.52) 0.740 1.42(1.13,1.78)  <  0.001  mgwas-metabolite-2956 L-VLDL-P 38 1.29(1.00,1.66) 0.050 0.050 1.29(0.93,1.78) 0.125 1.11(0.74,1.67) 0.626 1.29(1.00,1.66) 0.048  mgwas-metabolite-7014 XL-VLDL-PL 33 1.31(1.00,1.71) 0.049 0.050 1.36(0.95,1.95) 0.095 1.24(0.77,1.99) 0.370 1.31(1.00,1.71) 0.041  mgwas-metabolite-7010 XL-VLDL-FC 37 1.37(1.05,1.79) 0.020 0.050 1.35(0.94,1.94) 0.102 1.12(0.71,1.78) 0.634 1.37(1.05,1.79) 0.018  mgwas-metabolite-6987 XL-VLDL-P 30 1.31(1.00,1.71) 0.048 0.050 1.32(0.93,1.87) 0.121 1.22(0.77,1.94) 0.408 1.31(1.00,1.71) 0.038  mgwas-metabolite-6989 XL-VLDL-TG 32 1.31(1.01,1.70) 0.041 0.050 1.31(0.93,1.84) 0.125 1.27(0.82,1.97) 0.289 1.31(1.01,1.70) 0.046  mgwas-metabolite-232 CLA 9 1.85(1.12,3.07) 0.017 0.050 2.19(1.13,4.24) 0.020 3.47(0.85,14.21) 0.083 1.85(1.12,3.07) 0.005  mgwas-metabolite-2783 HDL-TG 27 1.31(1.04,1.65) 0.021 0.050 1.23(0.90,1.67) 0.193 1.27(0.86,1.87) 0.222 1.31(1.04,1.65) 0.006  mgwas-metabolite-385 Acetone 9 2.24(1.09,4.60) 0.028 0.028 1.75(0.70,4.36) 0.232 2.93(0.56,15.32) 0.202 2.24(1.09,4.60) 0.017  mgwas-metabolite-1259 3-HB 9 1.86(1.09,3.16) 0.022 0.022 1.40(0.71,2.77) 0.330 2.20(0.55,8.82) 0.265 1.86(1.09,3.16) 0.213 GDM  mgwas-metabolite-88 Lactate 4 0.38(0.15,0.99) 0.044 0.044 0.43(0.22,0.82) 0.011 0.04(0.00,0.56) 0.016 0.38(0.15,0.99) 0.319  mgwas-metabolite-1257 Glucose 16 4.54(2.18,9.47)  < 0.001  <  0.001 2.95(1.87,4.64)  < 0.001 2.74(0.38,19.61) 0.316 4.54(2.18,9.47)  <  0.001 Abbreviations: PCOS polycystic ovary syndrome, GDM gestational diabetes mellitus, IVs number of instrumental variables, OR odds ratio per standard deviation (SD) of genetic prediction plasma level, CI confidence interval, IVW inverse variance, MR-PRESSO Mendelian randomization pleiotropy, residual sum and outlier, P BH Benjamini‒Hochberg correction P value of each female reproductive disease type, HDL-D mean diameter for HDL, S-HDL-P concentration of small HDL particles, S-HDL-L total lipids in small HDL, S-HDL-C total cholesterol in small HDL, S-HDL-PL phospholipids in small HDL, L-HDL-P concentration of large HDL particles, L-HDL-L total lipids in large HDL, L-HDL-C total cholesterol in large HDL, L-HDL- PL phospholipids in large HDL, L-HDL-CE cholesterol ester in large HDL, L-HDL-FC free cholesterol in large HDL, L-HDL-TG triglycerides in large HDL, XL-HDL-P concentration of very large HDL particles, XL-HDL-L total lipids in very large HDL, XL-HDL-C total cholesterol in very large HDL, XL-HDL-PL phospholipids in very large HDL, XL-HDL-CE cholesterol ester in very large HDL, XL-HDL-FC free cholesterol in very large HDL, UnsatDeg estimated degree of unsaturation, VLDL-D mean diameter for VLDL particles, L-VLDL-P concentration of large VLDL particles, XL-VLDL-PL phospholipids in chylomicrons and extremely large VLDL, XL-VLDL-FC free cholesterol in chylomicrons and extremely large VLDL, XL-VLDL-P concentration of very large VLDL particles, XL-VLDL- TG triglycerides in very large VLDL, CLA conjugated linoleic acid, HDL-TG triglycerides in HDL, 3-HB 3-hydroxybutyrate Forest plot for significant associations between serum metabolites and MetS-related female reproductive diseases. Eight types of serum metabolites were significantly associated with the risk of 3 MetS-related female reproductive diseases (eclampsia, PCOS, and GDM). Yellow ( P   1) indicates a statistically significant association with increased risk. Red ( P  < 0.05 and HR < 1) indicates a statistically significant association with a decreased risk. Abbreviations: PCOS, polycystic ovary syndrome; GDM, gestational diabetes mellitus; HDL-D, mean diameter for HDL particles; S-HDL-P, concentration of small HDL particles; S-HDL-L, total lipids in small HDL; S-HDL-C, total cholesterol in small HDL; S-HDL-PL, phospholipids in small HDL; L-HDL-P, concentration of large HDL particles; L-HDL-L, total lipids in large HDL; L-HDL-C, total cholesterol in large HDL; L-HDL-PL, phospholipids in large HDL; L-HDL-CE, cholesterol ester in large HDL; L-HDL-FC, free cholesterol in large HDL; L-HDL-TG, triglycerides in large HDL; XL-HDL-P, concentration of very large HDL particles; XL-HDL-L, total lipids in very large HDL; XL-HDL-C, total cholesterol in very large HDL; XL-HDL-PL, phospholipids in very large HDL; XL-HDL-CE, cholesterol ester in very large HDL; XL-HDL-FC, free cholesterol in very large HDL; UnsatDeg, estimated degree of unsaturation; VLDL-D, mean diameter for VLDL particles; L-VLDL-P, concentration of large VLDL particles; XL-VLDL-PL, phospholipids in chylomicrons and extremely large VLDL; XL-VLDL-FC, free cholesterol in chylomicrons and extremely large VLDL; XL-VLDL-P, concentration of very large VLDL particles; XL-VLDL-TG, triglycerides in very large VLDL; CLA, conjugated linoleic acid; HDL-TG, triglycerides in HDL; 3-HB, 3-hydroxybutyrate MR results for plasma metabolites significantly associated with the risk of female reproductive diseases < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001 Abbreviations: PCOS polycystic ovary syndrome, GDM gestational diabetes mellitus, IVs number of instrumental variables, OR odds ratio per standard deviation (SD) of genetic prediction plasma level, CI confidence interval, IVW inverse variance, MR-PRESSO Mendelian randomization pleiotropy, residual sum and outlier, P BH Benjamini‒Hochberg correction P value of each female reproductive disease type, HDL-D mean diameter for HDL, S-HDL-P concentration of small HDL particles, S-HDL-L total lipids in small HDL, S-HDL-C total cholesterol in small HDL, S-HDL-PL phospholipids in small HDL, L-HDL-P concentration of large HDL particles, L-HDL-L total lipids in large HDL, L-HDL-C total cholesterol in large HDL, L-HDL- PL phospholipids in large HDL, L-HDL-CE cholesterol ester in large HDL, L-HDL-FC free cholesterol in large HDL, L-HDL-TG triglycerides in large HDL, XL-HDL-P concentration of very large HDL particles, XL-HDL-L total lipids in very large HDL, XL-HDL-C total cholesterol in very large HDL, XL-HDL-PL phospholipids in very large HDL, XL-HDL-CE cholesterol ester in very large HDL, XL-HDL-FC free cholesterol in very large HDL, UnsatDeg estimated degree of unsaturation, VLDL-D mean diameter for VLDL particles, L-VLDL-P concentration of large VLDL particles, XL-VLDL-PL phospholipids in chylomicrons and extremely large VLDL, XL-VLDL-FC free cholesterol in chylomicrons and extremely large VLDL, XL-VLDL-P concentration of very large VLDL particles, XL-VLDL- TG triglycerides in very large VLDL, CLA conjugated linoleic acid, HDL-TG triglycerides in HDL, 3-HB 3-hydroxybutyrate For eclampsia, protective effects were observed for several indicators, including the mean diameter of HDL particles (HDL-D), as well as for large HDL particles (L-HDL) and very large HDL particles (XL-HDL). Specifically, these included the concentration, total lipids, total cholesterol, PLs, cholesterol esters, and free cholesterol, as did UnsatDeg (all P  < 0.05). Notably, four associations with small HDL particles (S-HDL), total lipids, concentration, total cholesterol, and PLs, were detected to elevate the risk of eclampsia (all P  < 0.05). Additionally, triglycerides in large HDL particles (L-HDL-TG) were linked to a greater risk of eclampsia ( P  < 0.05). For patients with PCOS, elevated levels of triglycerides in high-density lipoprotein (HDL-TG) were strongly correlated with an elevated risk of PCOS ( P  < 0.05). Additionally, several VLDL-related indicators were identified as contributing risk factors. These included the mean diameter of VLDL particles (VLDL-D), the concentration of large VLDL particles particles (L-VLDL-P), and various components of extremely large VLDL particles (XL-VLDL), such as PLs, particle concentration, triglycerides, and free cholesterol (all P  < 0.05). Moreover, higher levels of acetone, 3-HB, and CLA were also linked to an elevated risk of PCOS (all P  < 0.05). For GDM, glucose was identified as a risk factor among the metabolites ( P  < 0.05). In contrast, lactate was observed to exert a protective effect against the risk of GDM ( P  < 0.05). All 31 significant associations confirmed through the IVM method remained consistent in both direction and relevance when assessed using additional supplementary MR methods (Additional File 1 : Table S5). Notably, more than 51% (16/31) of these links demonstrated P   0.05) and the MR-PRESSO global test (all P  > 0.05) revealed that none of the 31 associations were influenced by horizontal pleiotropy. Notably, while heterogeneity was observed in certain outcomes, the application of random effects IVW tests, despite heterogeneity identified via Cochrane’s Q test ( P  < 0.05), yielded more conservative yet robust estimates. The LOO results further substantiated the robustness of the MR outcomes. Additionally, the Steiger test results (all P  < 0.05) confirmed the absence of significant causal links between the eight types of metabolites and three female reproductive diseases within these 31 associations. The initial screening results indicated that eight types of metabolites were included among the 31 significant associations. The replication study confirmed that these associations and their corresponding metabolites were identifiable using the same available data reported in the study by Chang et al. Furthermore, the IVW method was employed, adhering to the threshold criteria established in the initial analysis. Of the 31 associations, with the exception of VLDL-D, the remaining associations were consistently observed to exhibit the same directional relationship as identified in the initial screening, with 21 of these associations achieving statistical significance ( P  < 0.05) (Additional File 1 : Table S6). Colocalization analyses revealed that HDL and eclampsia share four identical causal variant loci (rs183130, rs2070895, rs6073958, and rs174574), indicating a significant likelihood of shared causal genetic variants between the metabolites and the risk of female reproductive diseases (PPH 4  > 0.50). Additionally, VLDL and CLA share the same genetic locus (rs964184) in PCOS, indicating their concurrent colocalization (PPH 4  > 0.50). Finally, we found that lactate was a common causal variant associated with GDM (PPH 4  = 1.00), as detailed in Additional File 1 : Table S7.

Although prior investigations have offered insights into the associations between MetS and female reproductive diseases, clarifying the causal relationships between them is essential for addressing the potential influence of confounding factors. In this investigation, a two-step MR analysis was utilized to investigate the causal relationships between female reproductive diseases and MetS, along with plasma metabolites. The key findings demonstrated that MetS contributes to an elevated risk of five diseases: UF, PCOS, GDM, eclampsia, and miscarriage (all OR > 1, and P  < 0.05). As a collection of diverse metabolic imbalances, MetS may elevate the risk of these reproductive disorders through significant alterations in plasma metabolites [ 80 ]. Utilizing a recent GWAS that identified SNPs associated with 233 metabolic traits in 136,016 participants from 33 cohorts [ 39 ], the second-step MR analysis suggested causal relationships between the five MetS-related female reproductive diseases and eight distinct classes of plasma metabolites, including lipids and other compounds. These findings enhance the understanding of the etiological connections between metabolic health and reproductive disorders and may guide future strategies for prevention and intervention. PCOS is recognized as a reproductive endocrine and metabolic disorder with significant implications for reproductive health [ 81 ]. Numerous studies have established that patients with PCOS experience dysregulation of glucose and lipid metabolism, in addition to insulin resistance (IR), which contributes to obesity and an increased risk of cardiovascular complications [ 82 – 85 ]. Similarly, these findings indicate that MetS elevates the risk of PCOS by 3.35-fold, with the Steiger test ruling out reverse causality ( P  < 0.05). A recent systematic review comprising 19 studies reported that insulin sensitizers enhance metabolic profiles in these women and slightly improve reproductive outcomes in women with PCOS [ 86 ]. Nevertheless, some evidence suggests that women with PCOS are twice as likely to develop MetS than women without PCOS, while women with MetS frequently exhibit reproductive endocrine traits associated with PCOS [ 87 ]. As IR represents a primary driving mechanism underlying both conditions and serves as the most common pathogenic link [ 88 ], further exploration into the mediating role of IR between MetS and PCOS is necessary. In both the initial and replication studies on metabolites and PCOS, triglycerides in very large VLDL (XL-VLDL-TG), PLs in chylomicrons and extremely large VLDL (XL-VLDL-PL), free cholesterol in chylomicrons and extremely large VLDL (XL-VLDL-FC), and 3-HB consistently exhibited identical risk trends. Previous research has shown that PCOS-related factors are associated with reduced levels of high-density lipoprotein cholesterol (HDL-C) and apolipoprotein A-I (Apo A-I) but elevated levels of total triglycerides (TG), very low-density lipoprotein triglycerides (VLDL-TG), VLDL cholesterol (VLDL-C), large VLDL particle concentrations (L-VLDL-P), and VLDL-D [ 89 – 94 ]. According to Couto et al., lipid and lipoprotein profiles related to serum testosterone levels exhibit significant variations among women with PCOS and abdominal obesity. Testosterone was found to be positively correlated with serum TG, as well as with PL, TG, and total cholesterol, in larger VLDL subclasses. Additionally, total lipids in VLDL subclasses, along with the concentrations and mean diameters of larger VLDL particles (XL-VLDL-P and XL-VLDL-D), were also positively associated with testosterone [ 95 ]. Moreover, hyperandrogenism may result in abdominal fat accumulation, impairing lipid storage in subcutaneous adipocytes and further contributing to MetS [ 96 ]. Consequently, hyperandrogenaemia could induce alterations in lipid metabolism and exacerbate PCOS development through individual or synergistic effects with MetS. Among the nonlipid metabolites associated with PCOS, 3-HB and acetone have been identified as risk factors. These ketone body metabolites participate in lipid metabolism and influence the overall metabolic rate; however, their specific roles in PCOS remain unclear [ 97 ]. CLA, derived from dietary sources such as ruminant meat and milk, has also been recognized as a risk factor for PCOS [ 98 ]. CLA may contribute to inflammation by elevating leptin levels and reducing adiponectin levels, which leads to IR and increased androgen production, further exacerbating PCOS [ 99 – 101 ]. Dysregulation of inflammatory mediators has been connected to immune and metabolic disturbances in PCOS, potentially intensifying IR and androgen production [ 102 , 103 ]. Nevertheless, some studies have reported conflicting effects of CLAs on inflammation prevention [ 104 – 108 ]. This inconsistency may arise from the differing effects of its isomers—c9-t11, which exhibit anti-inflammatory effects, while t10-c12, which demonstrate proinflammatory effects [ 99 , 101 ]. Consequently, additional comprehensive research is needed to elucidate the impact of CLA on PCOS. Eclampsia is characterized as a pregnancy-related hypertensive disorder that typically arises after 20 weeks of gestation. Previous research has indicated that elevated blood pressure, dyslipidemia, and obesity substantially increase the risk of eclampsia [ 109 – 112 ]. In this study, a significant causal relationship between MetS and eclampsia was identified, aligning with prior findings that established a connection between MetS and eclampsia [ 113 ]. Similarly, a recent MR study demonstrated that leptin, fasting insulin, and IR collectively mediated between 20 and 50% of the genetically predicted association of obesity with eclampsia [ 13 ]. Another study corroborated the association between blood pressure (BP) and eclampsia, suggesting that antihypertensive therapy may exert therapeutic effects by modulating critical proteins, such as apolipoproteins, involved in inflammation, immunity, and metabolic regulation [ 114 ]. Additionally, reduced HDL levels and heightened oxidative stress have been identified as significant contributors to eclampsia [ 115 ]. Furthermore, dysregulation of metabolic factors, including adiponectin, leptin, and inhibin, which contribute to overweight and obesity in women, has been shown to exacerbate IR and elevate the risk of eclampsia [ 116 ]. Thus, it is imperative to consider the influence of MetS and associated metabolic factors in the clinical management of eclampsia. HDL particles, classified by size into extra-large (XL-HDL), large (L-HDL), medium, small (S-HDL), and very small (XS-HDL) subclasses, serve essential functions in lipid metabolism [ 117 , 118 ]. Additionally, HDL lipids include PLs and free cholesterol (FC) in the outer layer, while cholesterol esters (CE) and triglycerides (TG) are located in the core, although their specific roles in eclampsia remain poorly understood [ 119 ]. In this study, lipid molecules within large HDL particles, including the concentration of very large HDL particles (XL-HDL-P), the concentration of large HDL (L-HDL-P), cholesterol ester in large HDL (L-HDL-CE), total cholesterol in very large HDL (XL-HDL-C), total cholesterol in large HDL (L-HDL-C), free cholesterol in very large HDL (XL-HDL-FC), free cholesterol in large HDL (L-HDL-FC), phospholipids in very large HDL (XL-HDL-PL), and phospholipids in large HDL (L-HDL-PL), were shown to have protective effects against eclampsia. Conversely, elevated levels of L-HDL-TG and markers related to small HDL particles (S-HDL) were linked to an elevated risk of eclampsia. Furthermore, the mass concentration of large HDL particles (L-HDL-P and XL-HDL-P) has been inversely linked to all-cause and cardiovascular mortality, reinforcing its protective role against eclampsia [ 120 ]. While certain studies emphasize the antioxidant properties of small HDL particles [ 121 ], our findings indicate that the concentration of small HDL particles (S-HDL-P) is positively correlated with eclampsia risk, necessitating further exploration into the mechanisms of HDL subclasses. Clinical studies have demonstrated that patients with eclampsia exhibit higher TG levels and lower HDL-C levels, with elevated TG and reduced HDL-C serving as predictive markers for eclampsia risk [ 122 ]. Elevated TG levels are believed to adversely influence placental function by promoting oxidative stress and lipid peroxidation [ 123 ]. Furthermore, the inhibition of CE transfer protein activity has been shown to decrease HDL-TG levels and increase cholesterol ester levels in HDL (HDL-CE), which may confer protection against eclampsia [ 122 , 124 ]. Higher concentrations of large HDL-C, such as L-HDL-C and XL-HDL-C, have been associated with reduced cardiovascular risk, suggesting analogous protective effects in eclampsia [ 125 , 126 ]. These protective effects are attributed to their potent antioxidant properties and enhanced cholesterol uptake and anti-inflammatory effects facilitated by proteins such as Apo E, CFH, and Pon 1 [ 125 , 127 ]. Previous MR studies have also shown a link between higher levels of XL-HDL-FC and L-HDL-FC and a reduced risk of eclampsia [ 125 ]. Although some research has suggested that elevated HDL-FC levels may impair endothelial cell function and induce inflammation and oxidative stress, thereby contributing to eclampsia development [ 128 ], it is important to note that HDL-FC remains in plasma for only a short duration—potentially a few minutes—and that its concentrations fluctuate markedly. Consequently, HDL-FC levels may not adequately reflect the relationship of these variables with eclampsia risk, warranting further investigation [ 125 , 129 ]. HDL-PLs play a pivotal role in cholesterol efflux, the initial step of reverse cholesterol transport mediated by HDL [ 130 ]. Studies have indicated that higher HDL-PLs enhance cholesterol efflux capacity, while reductions in HDL-PLs may compromise this function [ 131 – 134 ]. These findings support our conclusion that increased levels of L-HDL-PL and XL-HDL-PL are associated with a reduced risk of eclampsia. GDM is characterized by diabetes diagnosed during the second or third trimester of pregnancy that does not present as overt diabetes prior to gestation [ 135 ]. In this investigation, MetS and glucose were identified as significant risk factors for GDM, consistent with earlier findings [ 136 , 137 ]. Maternal MetS during pregnancy has been reported to not only increase the risk of eclampsia but also increase the likelihood of GDM in pregnant women [ 138 – 140 ]. Similarly, a study conducted in Brazil identified high TG levels, WC, fasting plasma glucose (FPG), and BP as independent risk factors for GDM [ 141 ]. Moreover, Nguyen et al. demonstrated that MetS contributes to an increased incidence of GDM during pregnancy and heightens the risk of diabetes development in women with a history of GDM [ 142 ]. Additionally, a prospective cohort study in China established a significant association between the first-trimester triglyceride-glucose index and the risk of incident GDM [ 143 ]. Consequently, addressing MetS components, including obesity, dyslipidemia, hypertension, and IR, may effectively reduce the incidence of GDM and its associated complications. Furthermore, these findings substantiate a notable genetic association and causative link between MetS and the occurrence of UF or pregnancy loss. In alignment with these observations, a comparative analysis demonstrated that UF patients exhibited markedly greater body mass index (BMI), BP, TG, and FPG measurements, while excess weight and elevated BP were identified as contributing factors to fibroid development [ 144 ]. Although our investigation indicated that MetS substantially enhances miscarriage probability, clinical research examining this connection has yielded conflicting findings. A longitudinal investigation revealed a significant correlation between elevated BMI and heightened miscarriage risk [ 145 ]. Conversely, the RaPCo study conducted in Sri Lanka determined that neither MetS nor elevated FPG during early gestation served as a predictive indicator for pregnancy loss, although increased WC was found to markedly increase the likelihood of miscarriage [ 146 ]. These contradictory findings might be attributed to population selection variations across different cohort investigations. The implementation of practical and economically viable approaches—encompassing lifestyle modifications, nutritional adjustments, and systematic evaluation of PG, BP, and lipid parameters—has been demonstrated to enhance metabolic well-being. These therapeutic interventions can potentially diminish the incidence of MetS and subsequently mitigate the risk of female reproductive disorders [ 125 ]. The incorporation of these prophylactic strategies into medical protocols could yield substantial advantages for reproductive wellness in women. Nevertheless, several limitations warrant consideration when translating these research outcomes into clinical applications. Initially, the established correlations between MetS and reproductive disorders in females were derived from GWASs primarily conducted among European descendants, potentially limiting the applicability of these associations across diverse ethnic populations. Subsequent investigations should include participants from various genetic lineages to authenticate and broaden our current understanding of the topic. Additionally, while MR methodology provides robust causality assessment, inherent constraints persist, including pleiotropic effects, IV strength, demographic stratification, and methodological complexities. Further empirical investigations utilizing clinical datasets are essential to substantiate the causative relationships between MetS and various reproductive pathologies. Moreover, the efficacy of dietary modifications may exhibit considerable heterogeneity among different patient cohorts. Thus, individual dietary patterns and preferences merit consideration as potential exclusion parameters during subject selection processes. Finally, we must acknowledge the limitation that the statistics of MetS in this study are based on GWAS data of phenotypic clustering of its components. However, compared with any single factor, a common factor poses a greater risk for MetS [ 38 ]. Furthermore, GWAS data derived from directly measured MetS clinical outcomes are lacking. The experimental findings from this investigation establish a causal relationship between MetS and various reproductive disorders in females, encompassing UFs, PCOS, GDM, eclampsia, and miscarriage. Specific circulating metabolites have been ascertained as causal factors in these MetS-associated reproductive complications. Several protective components against eclampsia, including L-HDL-CE, XL/L-HDL-C, XL/L-HDL-P, XL/L-HDL-FC, and XL/L-HDL-PL, were identified, whereas L-HDL-TG was determined to be detrimental. In the context of PCOS, elevated levels of HDL-TG, XL-VLDL-PL, XL-VLDL-FC, and 3-HB were found to heighten disease susceptibility. Moreover, hyperglycemia was correlated with increased GDM occurrence. These observations substantiate the significant involvement of blood-borne metabolites, particularly those associated with lipid metabolism, in female reproductive pathologies, suggesting their potential utility as diagnostic and prognostic indicators. The results emphasize the clinical significance of maintaining optimal body mass, implementing dietary modifications with reduced fat consumption, and consistently monitoring lipid profiles to enhance metabolic well-being and potentially mitigate these reproductive complications.

Female reproductive diseases, including endometriosis (EMs), uterine fibroids (UFs), polycystic ovary syndrome (PCOS), gestational diabetes mellitus (GDM), eclampsia, ectopic pregnancy (EP), infertility, miscarriage, and ovarian aging, impose a substantial global health burden. These conditions profoundly impact not only women’s health and overall quality of life but also their fertility and pregnancy outcomes [ 1 ]. In an era marked by postponed childbearing and decreasing fertility rates, issues surrounding reproductive health and pregnancy have emerged as pressing global concerns [ 2 ]. The etiology of female reproductive diseases is intricate and influenced by multiple factors, including genetic, hormonal, environmental, and metabolic factors. Although substantial research has been conducted, the root causes of numerous conditions have not been fully elucidated, impeding advancements in prevention and treatment strategies. For example, EMs impact an estimated 10% of women of reproductive age, yet the pathogenesis of EMs has not been fully elucidated [ 3 ]. Similarly, while hormonal and metabolic disruptions are considered pivotal contributors, the precise origins of UF and PCOS have not been fully determined [ 4 , 5 ]. Thus, comprehending the etiology of these diseases is pivotal for enhancing women’s health outcomes and identifying novel diagnostic and therapeutic avenues. Metabolic syndrome (MetS) refers to a combination of physiological irregularities marked by hypertension, hyperglycemia, abdominal obesity, and dyslipidemia, specifically increased triglycerides and reduced high-density lipoprotein cholesterol (HDL-C) levels [ 6 ]. The global occurrence of MetS fluctuates between 20 and 50%, exhibiting a persistent ascending pattern [ 7 ]. Alarmingly, MetS has been increasingly identified in younger demographic groups, with up to 50% of severely obese adolescents being affected [ 8 ]. Evidence suggests that MetS disrupts hormonal homeostasis, altering the levels of gonadotropins, sex steroid hormones, sex hormone-binding globulin, anti-Müllerian hormone (AMH), and inhibin B [ 9 ]. Such hormonal disturbances, including menstrual irregularities, infertility, and various gynecological disorders, are associated with reproductive complications in women. Consequently, establishing the etiological connection between MetS and female reproductive conditions is of critical importance, as it may reveal new approaches for preventive measures and treatment options. Recent studies have underscored robust associations between MetS and female reproductive disorders and between MetS and its components. A population-based investigation demonstrated that women diagnosed with EMs are at an elevated risk of developing MetS, distinguished by increased waist circumference (WC) and decreased HDL-C levels in comparison to women without EMs [ 10 ]. According to a retrospective analysis conducted in China, MetS was found to be strongly correlated with a higher risk of late miscarriage, lower live birth rate, GDM, hypertensive disorders during pregnancy, and preterm birth [ 11 ]. Another study reported that although the serum levels of gonadotropins (FSH, LH), estradiol (E 2 ), progesterone (P), and antral follicle count (AFC) did not notably differ between MetS patients and controls, the total ovarian volume was markedly reduced in MetS patients [ 12 ]. Nevertheless, these surveillance studies face limitations due to possible interfering factors and the likelihood of reverse causation, despite consistent evidence of a connection between MetS and female reproductive conditions. Notably, randomized controlled trials (RCTs), regarded as the gold standard for causal inference, are often unfeasible in this domain due to ethical and practical limitations [ 13 , 14 ]. Consequently, alternative methodologies are urgently needed to clarify the causal relationships between MetS and reproductive health. Emerging evidence from metabolomic studies underscores a robust correlation between MetS and alterations in metabolites [ 15 ]. Furthermore, metabolomics has been used to decipher the regulatory roles of metabolites in MetS-related diseases [ 16 , 17 ]. Circulating metabolites, which represent the dynamic nature of various metabolic processes, hold great potential as biomarkers or therapeutic targets due to their measurable and modifiable characteristics [ 18 ]. Additionally, numerous metabolomics studies have explored the underlying causes of female reproductive diseases. For example, Odibo et al. identified four metabolites—hydroxyhexanoylcarnitine, alanine, phenylalanine, and glutamate—that exhibit elevated levels during eclampsia [ 19 ]. Another investigation revealed strong associations between lysophosphatidylcholine, oleic acid, and 1,25-dihydroxyvitamin D3-26,23-lactone and the development of eclampsia and GDM [ 20 ]. Similarly, multiple studies have examined the metabolomic profiles associated with eclampsia [ 21 , 22 ]. In PCOS, plasma metabolomics analyses have demonstrated disruptions in carbohydrate, lipid, and amino acid metabolism, offering valuable insights into the metabolic basis of PCOS [ 23 ]. Additionally, a study employing UHPLC-MS/MS technology identified six metabolites—arachidonoyl amide, 3-hydroxy-3-methylbutyric acid (3OH3MB), dihexyl nonanedioate, 18-hydroxyeicosatetraenoic acid (18-HETE), cystine, and phosphatidylglycerol (PG 16:0/18:1)—as potential biomarkers for premature ovarian insufficiency, highlighting potential etiological factors [ 24 ]. Although numerous metabolic biomarkers have been proposed for predicting or diagnosing reproductive diseases, observational studies face inherent constraints, such as confounding bias, dietary influences on metabolite levels, and variability in experimental conditions and analytical methods, which often lead to inconsistent findings [ 25 ]. Consequently, innovative strategies are imperative for clarifying the causal relationships between plasma metabolites and MetS-related female reproductive diseases, thereby advancing our understanding of their etiology. As a substitute for RCTs, Mendelian randomization (MR) functions as a powerful method to evaluate causative associations between exposures and outcomes. Grounded in Mendel’s principles of inheritance, MR employs genetic markers, particularly single nucleotide polymorphisms discovered via genome-wide association studies, as instrumental variables (IVs). This technique effectively mitigates residual confounding and minimizes the risk of reverse causality [ 26 ]. By utilizing genetic variants as proxies for exposures, MR offers valuable insights into disease etiological pathways, addressing certain limitations associated with observational studies and RCTs [ 27 ]. For instance, an MR study reported that genetically predicted obesity was linked to a heightened risk of UF, PCOS, heavy menstrual bleeding (HMB), and eclampsia [ 13 ]. Moreover, the same study revealed that genetically determined visceral adipose tissue mass was associated with the onset of HMB, PCOS, and eclampsia. These findings underscore the utility of MR in advancing our understanding of the etiology of female reproductive diseases. Accumulating evidence has shown that plasma metabolites play an important role in MetS-related diseases, including female reproductive diseases [ 28 – 30 ]. Therefore, this study implemented a two-step MR framework to explore the etiological connections between metabolic health and MetS-related reproductive disorders (Fig.  1 ). Initially, two-sample MR analyses were executed to evaluate the causal links between MetS and nine female reproductive diseases: EMs, UF, PCOS, GDM, eclampsia, EP, infertility, miscarriage, and ovarian aging. Subsequently, 233 serum metabolites were examined as exposures to assess their causal associations with the five identified MetS-related reproductive diseases: UF, PCOS, GDM, eclampsia, and miscarriage. SNPs strongly linked to MetS and specific serum metabolites were chosen as IVs per the stringent inclusion and exclusion criteria. In accordance with the MR assumptions, consistent and significant associations were observed using inverse variance weighted (IVW) and other analytical methods. Furthermore, complementary analyses were applied to the significant causal relationships identified, enhancing the robustness and reliability of the findings. This study aimed to elucidate these relationships to discover potential biomarkers for early diagnosis and treatment, ultimately contributing to improved reproductive health outcomes for women globally. Fig. 1 Flowchart of the study design. Abbreviations: GWAS, genome-wide association study; SNPs, single nucleotide polymorphisms; kb, kilobase; MR-PRESSO, Mendelian randomization pleiotropy residual sum and outlier; LOO, leave-one-out Flowchart of the study design. Abbreviations: GWAS, genome-wide association study; SNPs, single nucleotide polymorphisms; kb, kilobase; MR-PRESSO, Mendelian randomization pleiotropy residual sum and outlier; LOO, leave-one-out

Additional file 1: Table S1. Information on genome-wide association studies (GWASs) of female reproductive diseases among individuals of European ancestry. Table S2. Statistical metrics of instrumental variables (IVs) for notable correlations detected through Mendelian randomization (MR) analyses. Table S3. Notable correlations were discovered during the initial MR imaging utilizing the inverse-variance weighted (IVW) approach. Table S4. Statistical parameters of instrumental variables (IVs) concerning 31 notable correlations identified during primary Mendelian randomization (MR) examination. Table S5. Notable correlations were discovered during the initial MR imaging utilizing the inverse-variance weighted (IVW) approach. Table S6. Replication study outcomes employing a separate metabolite GWAS dataset for 31 notable correlations found in initial analyses. Table S7. Colocalization evaluation of 29 notable correlations discovered during initial analyses. Additional file 1: Table S1. Information on genome-wide association studies (GWASs) of female reproductive diseases among individuals of European ancestry. Table S2. Statistical metrics of instrumental variables (IVs) for notable correlations detected through Mendelian randomization (MR) analyses. Table S3. Notable correlations were discovered during the initial MR imaging utilizing the inverse-variance weighted (IVW) approach. Table S4. Statistical parameters of instrumental variables (IVs) concerning 31 notable correlations identified during primary Mendelian randomization (MR) examination. Table S5. Notable correlations were discovered during the initial MR imaging utilizing the inverse-variance weighted (IVW) approach. Table S6. Replication study outcomes employing a separate metabolite GWAS dataset for 31 notable correlations found in initial analyses. Table S7. Colocalization evaluation of 29 notable correlations discovered during initial analyses.