Genetically predicted circulating immune cells and cytokines reveal the causal role of immune factors on female infertility: a two-sample mendelian randomization study.

Each immune trait in the dataset ( n  = 731) was performed mendelian randomization assumption checks with five types of female infertility. Immune cells at nominal significance level were collected in Additional file 5 ( P  < 0.05 before FDR correction). After correcting the P value, we identified that Naive CD8br %CD8br suggestively increases the risk of general female infertility (FEMALEINFERT), and CD80 on monocyte suggestively increases the risk of tubal origin female infertility (FITUB). In addition, HLA DR + NK %NK and CD16 on CD14- CD16 + monocyte were observed significant causal effects on uterine origin female infertility (FIUTERINE) risk, while CD86 + myeloid DC AC is a potential causal protective factor (Fig.  2 ). No causal relationship was observed between peripheral immune cell traits and either anovulation associated female infertility (FIANOV) or cervical, vaginal, other or unspecified origin female infertility (FIOTHNAS) (Fig.  3 ). Additionally, the possibility of horizontal pleiotropy for these associations passes the MR-Egger intercept and the MR-PRESSO examination. Funnel plots and leave-one-out also indicate the reliability of these results (Additional file 6 Fig. S1 , Fig. S3 , Table S1 ). Fig. 2 Scatter plots of Mendelian analyses between immune cell traits and five types of female infertility. All scatter plots verify the robustness of MR results. The x-axis of the scatter plot is the size of the SNP effect of exposure, and the y-axis is the size of the SNP effect of the outcome. Five different color lines correspond to five MR methods. ( A ) Naive CD8br %CD8br on female infertility (FEMALINFERT); ( B ) CD80 on monocyte on tubal origin female infertility (FITUB); ( C ) CD86 + myeloid DC AC on uterine origin female infertility (FIUTERINE); ( D ) HLA DR + NK %NK on uterine origin female infertility (FIUTERINE); ( E ) CD16 on CD14- CD16 +   monocyte on uterine origin female infertility (FIUTERINE) Scatter plots of Mendelian analyses between immune cell traits and five types of female infertility. All scatter plots verify the robustness of MR results. The x-axis of the scatter plot is the size of the SNP effect of exposure, and the y-axis is the size of the SNP effect of the outcome. Five different color lines correspond to five MR methods. ( A ) Naive CD8br %CD8br on female infertility (FEMALINFERT); ( B ) CD80 on monocyte on tubal origin female infertility (FITUB); ( C ) CD86 + myeloid DC AC on uterine origin female infertility (FIUTERINE); ( D ) HLA DR + NK %NK on uterine origin female infertility (FIUTERINE); ( E ) CD16 on CD14- CD16 +   monocyte on uterine origin female infertility (FIUTERINE) Fig. 3 Significant MR results of causal links between immune cell traits and 5 female infertilities Significant MR results of causal links between immune cell traits and 5 female infertilities We evaluated causal effects of 41 circulating cytokines on female infertility subtypes using identical MR protocols as for immune cell traits (Additional File 7 ). Macrophage inflammatory protein 1b (MIP1B) level exhibited protective effects across multiple infertility subtypes including FEMALEINFERT, FIANOV, FITUB and FIOTHNAS. Elevated IL17 levels demonstrated pathogenic potential, increasing risks for both FEMALEINFERT and FIOTHNAS. IL18 elevation showed a positive association with FIOTHNAS risk (Fig.  4 ). Scatter plots give visualized causal effects of these results (Fig.  5 ). The above results passed sensitivity analysis, including MR-Egger intercept, MR-PRESSO, funnel plots, and leave-one-out (Additional file 6 Fig. S2 , Fig. S4 ; Table S1 ). Fig. 4 Significant MR results of causal links between circulating inflammatory cytokines and 5 female infertilities Significant MR results of causal links between circulating inflammatory cytokines and 5 female infertilities Fig. 5 Scatter plots of Mendelian analyses between circulating cytokines and five types of female infertility. All scatter plots verify the robustness of MR results. The x-axis of the scatter plot is the size of the SNP effect of exposure, and the y-axis is the size of the SNP effect of the outcome. Five different color lines correspond to five MR methods. ( A ) MIP1B on female infertility (FEMALINFERT); ( B ) IL17 on female infertility (FEMALINFERT); ( C ) MIP1B on anovulation associated female infertility (FIANOV); ( D ) MIP1B on tubal origin female infertility (FITUB); ( E ) MIP1B on cervical, vaginal, other or unspecified origin female infertility (FIOTHNAS); ( F ) IL18 on cervical, vaginal, other or unspecified origin female infertility (FIOTHNAS); ( G ) IL17 on cervical, vaginal, other or unspecified origin female infertility (FIOTHNAS) Scatter plots of Mendelian analyses between circulating cytokines and five types of female infertility. All scatter plots verify the robustness of MR results. The x-axis of the scatter plot is the size of the SNP effect of exposure, and the y-axis is the size of the SNP effect of the outcome. Five different color lines correspond to five MR methods. ( A ) MIP1B on female infertility (FEMALINFERT); ( B ) IL17 on female infertility (FEMALINFERT); ( C ) MIP1B on anovulation associated female infertility (FIANOV); ( D ) MIP1B on tubal origin female infertility (FITUB); ( E ) MIP1B on cervical, vaginal, other or unspecified origin female infertility (FIOTHNAS); ( F ) IL18 on cervical, vaginal, other or unspecified origin female infertility (FIOTHNAS); ( G ) IL17 on cervical, vaginal, other or unspecified origin female infertility (FIOTHNAS) Prior to multivariable Mendelian randomization (MVMR) analysis, we systematically removed duplicate SNPs across significant exposures for each outcome to mitigate confounding. Among the 731 immune cell traits, only CD16 on CD14- CD16 + monocyte retained significance for FIUTERINE after adjustment (OR:1.289 [95% CI = 1.126–1.477], P  = 2.348 × 10 − 4 , P FDR =0.172), with robustness corroborated by complementary methods (weighted median P  = 0.002; MR-Egger P  = 1.564 × 10⁻⁴; weighted mode P  = 0.001) and sensitivity analyses (Additional file 8 Table S2 ). No overlapping instrumental variables were identified among significant cytokines. In that case, we implemented MVMR on outcomes had multiple positive causal cytokines, including FEMALEINFERT and FIOTHNAS. After adjusting for MIP1B in general infertility, IL17 maintained independent risk effects ( P  = 0.002). Similarly, in FIOTHNAS, IL17 retained significance ( P  = 0.001) whereas IL18 demonstrated protective effects ( P  = 0.008) after adjusting respectively (Table  2 ). Table 2 Multivariable MR results of causal links between inflammatory cytokines and female infertility after adjusting for specific significant cytokines Exposure Outcome Adjustments of cytokines Method Beta SE P Q-Statistic P Egger intercept MIP1B FEMALEINFERT IL17 MVMR-IVW -0.018 0.023 0.451 0.2407 - MVMR-Egger -0.042 0.046 0.361 0.2215 0.053 IL17 FEMALEINFERT MIP1B MVMR-IVW 0.107 0.034 0.002 0.2407 - MVMR-Egger 0.109 0.035 0.002 0.2215 0.053 MIP1B FIOTHNAS IL17, IL18 MVMR-IVW -0.008 0.022 0.720 0.6916 - MVMR-Egger -0.005 0.031 0.861 0.6589 0.916 IL17 FIOTHNAS MIP1B, IL18 MVMR-IVW 0.116 0.034 0.001 0.6916 - MVMR-Egger 0.116 0.034 0.001 0.6589 0.916 IL18 FIOTHNAS MIP1B, IL17 MVMR-IVW -0.062 0.023 0.008 0.6916 - MVMR-Egger -0.062 0.024 0.008 0.6589 0.916 Multivariable MR results of causal links between inflammatory cytokines and female infertility after adjusting for specific significant cytokines

Using two-sample MR, we analyzed 731 immune cell traits and 41 circulating cytokines as separate exposures to investigate their causal relationships with five female infertility subtypes, resulting in 10 distinct MR analyses. For infertility subtypes showing associations with multiple immune features, MVMR was implemented to account for potential mediating effects. The overall analytical workflow is illustrated in Fig.  1 . Fig. 1 Workflow of MR analyses. 10 MR analyses were performed. First, we evaluated the causal links between 731 immune cell traits and five types of infertility. Then, the causal links between 41 circulating cytokines and five types of infertility were evaluated. For the female infertility with multiple significant immune cells or cytokines, MVMR was used respectively to explore direct causal effects Workflow of MR analyses. 10 MR analyses were performed. First, we evaluated the causal links between 731 immune cell traits and five types of infertility. Then, the causal links between 41 circulating cytokines and five types of infertility were evaluated. For the female infertility with multiple significant immune cells or cytokines, MVMR was used respectively to explore direct causal effects All datasets were derived from publicly accessible Genome-Wide Association Study (GWAS) summary statistics [ 16 ], with ethical approvals and informed consent obtained in the original studies. Detailed recruitment criteria and raw sequence data quality controls are provided in the original publications and the finngenR9 website. Table  1 summarizes the details of these studies. The 731 immune cell traits were acquired from the GWAS Catalog (accession numbers from GCST0001391 to GCST0002121). It contains four characteristics of seven circulating immune cell clusters, including Absolute cell count (AC, n  = 118); Relative cell count (RC, n  = 192); Median fluorescence intensities (MFI, n  = 389) which reflect the abundance surface antigen; Morphological parameter (MP, n  = 32) [ 17 ] (Additional file 1 ). Samples from this study were genotyped using Omni-Express, Immuno-Chip, Cardio-MetaboChip, and Exome-Chip, providing 22 million variants for subsequent analysis. 41 Circulating Cytokines datasets originated from Ahola-Olli et al.‘s study which is available from the GWAS Catalog (accession numbers from GCST004420 to GCST004460). Study populations include The Cardiovascular Risk in Young Finns Study (YFS) and FINRISK cohorts. A total of 48 cytokines were measured by Bio-Rad’s premixed Bio-Plex Pro Human Cytokine 27-plex Assay and 21-plex Assay according to the manufacturer’s instructions, and 41 cytokines were available after data analysis and quality control [ 18 ] (Additional file 2 ). Female infertility phenotypes were obtained from FinnGen Consortium R9 release, including general female infertility (FEMALEINFERT), anovulation associated female infertility (FIANOV), tubal origin female infertility (FITUB), uterine origin female infertility (FIUTERINE), cervical, vaginal, other or unspecified origin female infertility (FIOTHNAS). Infertility diagnostic criteria for diseases are based on ICD8, ICD9, and ICD10 [ 19 ]. It is worth mentioning that FIANOV, FITUB, FIUTERINE, and FIOTHNAS are early termination points of the FEMALEINFERT project, and individuals diagnosed with non-inflammatory disorders of the female genital tract were excluded from the control group in this project. Table 1 Details of GWAS data GWAS data Population Source Ncase Ncontrol 731 immune cell traits Sardinians Valeria Orrù et al. ( https://www.ebi.ac.uk/gwas/publications/27989323 ) 3757 - 41 cytokines European Ahola-Olli et al. ( https://www.ebi.ac.uk/gwas/publications/32929287 ) 8293 - finngen_R9_N14_FEMALEINFERT European FinngenR9 ( https://storage.googleapis.com/finngen-public-data-r9/summary_stats/finngen_R9_N14_FEMALEINFERT.gz ) 13,142 107,564 finngen_R9_N14_FITUB European FinngenR9 ( https://storage.googleapis.com/finngen-public-data-r9/summary_stats/finngen_R9_N14_FITUB.gz ) 1483 107,564 finngen_R9_N14_FIANOV European FinngenR9 ( https://storage.googleapis.com/finngen-public-data-r9/summary_stats/finngen_R9_N14_FIANOV.gz ) 2441 107,564 finngen_R9_N14_FIUTERINE European FinngenR9 ( https://storage.googleapis.com/finngen-public-data-r9/summary_stats/finngen_R9_N14_FIUTERINE.gz ) 99 107,564 finngen_R9_N14_FIOTHNAS European FinngenR9 ( https://storage.googleapis.com/finngen-public-data-r9/summary_stats/finngen_R9_N14_FIOTHNAS.gz ) 11,348 107,564 Details of GWAS data To meet the essential requirements for the robustness of Mendelian randomization, we followed a workflow for selecting IVs. The selection of each IV for circulating immune cell traits is based on recent publications [ 20 – 22 ], and involves setting significance levels to 1 × 10 − 5 and removing linkage disequilibrium (LD) between single nucleotide polymorphisms (SNPs) by PLINK software (v1.90) [ 23 ]. [LD] r 2 threshold was set to be < 0.1 within a 500 kb distance, calculated based on European 1000 Genomes Projects as a reference panel [ 24 ]. To avoid weak instrumental bias, IVs with an F statistic  < 10 were eliminated [ 25 ], resulting in a total of 5 to 1,646 independent IVs for immunophenotypes (Additional file 3 ). The same variable (IV) selection protocol was implemented for the 41 circulating cytokines, yielding 9-152 IVs per cytokine (see Additional file 4 ). Five MR methods were employed to assess causal relationships between exposures (731 immune cell traits and 41 cytokines) and outcomes (five female infertility subtypes). Causal effect sizes were mainly evaluated by the inverse variance weighting (IVW) method which assumes that all IVs are valid instrumental variables [ 26 ]. However, in actual analysis, this goal can hardly be fully realized, which means a higher risk of false positives. Therefore, the other four methods were employed to support the robustness of IVW results. MR-Egger was used to correct for directional pleiotropy, while the weighted median, simple mode, and weighted mode methods were employed to account for heterogeneity in the correlation effect estimates. Subsequently, we performed sensitivity analyses [ 27 , 28 ]. To be specific, Cochran’Q based on IVW was utilized to examine the heterogeneity, and funnel plots rendered the robustness of correlation and examinations of the heterogeneity. Horizontal pleiotropic outliers were excluded by MR-Egger intercept significance [ 29 ], leave-one-out, and the MR pleiotropy residual sum, and outlier (MR-PRESSO) method ( NbDistribution  = 10000) [ 30 ]. Considering the accumulation of type I errors, we adjusted P  values in 731 immune cell traits by FDR method ( P FDR <0.2 was suggestive causal associations according to the previous study to enable a more comprehensive assessment of the causal relationship, and P FDR <0.05 was recognized as significant causal effect) [ 21 ], while Bonferroni correction method was employed in 41 cytokines ( P  < 0.0012, Bonferroni correction with 41 tests). All analyses were performed in R 4.2 software and R package "TwoSampleMR" (version 0.5.6). To further evaluate the independent causal effects of immune cell traits or cytokines on female infertility, MVMR was performed when the outcome had multiple positive exposures (immune cells or cytokines). To mitigate potential intermixing among immune cell traits or cytokines, the repeat IVs of each significant immune cell or cytokine were excluded before MVMR respectively. MVMR-IVW and MVMR-Egger were employed in MVMR analysis, and the parameter setting is the same as that of two-sample MR. Furthermore, Q-Statistic and Egger intercept were used for sensitivity analysis, and SNPs unavailable in the outcome summary were substituted with the proxy SNPs from the LDlink website.

Leveraging publicly available GWAS summary data, we analyzed the causal relationship between immune factors and female infertility using two-sample MR and multi-variable MR methods. Our results stress that immune dysregulation contributes significantly to female infertility, highlighting potential therapeutic targets for diagnosis and treatment.

This study establishes causal relationships between peripheral immune signatures and female infertility subtypes through Mendelian randomization, advancing beyond previous correlative observations. Clinical blood analyses in infertility patients revealed elevated plasma IL-17 levels [ 31 ]. Meanwhile, MIP-1B demonstrates protective effects potentially mediated through endometrial receptivity association with improved pregnancy rates [ 32 ], supported by improved IVF outcomes with elevated levels [ 33 ]. Moreover, the abundance of HLA-DR + CD11c + cells in the uterus was found positive correlation with IVF outcomes [ 33 ]. While accumulating evidence implicates immune dysregulation in infertility, observational studies face inherent limitations from confounding and reverse causation [ 34 ]. Nevertheless, female infertility is a highly heterogeneous disease to which different mechanisms contribute. Our approach overcomes these constraints through genetic instruments assessing 731 immune cell traits and 41 cytokines across five infertility subtypes. Our study indicates that Naive CD8br %CD8br is suggestively associated with general female infertility. A change in the proportion of naive CD8 bright cells among total CD8 bright cells may indicate a weak immune response in infertile patients. In addition, results revealed that increased expression of CD80 on monocytes elevates the risk of tubal origin infertility. CD80 plays an important role in T cell activation when located on the surface of monocytes [ 35 ]. This risk may arise from abnormal T-cell activation that interferes with fallopian tube function, consequently impairing fertilization [ 36 ]. Additionally, our results revealed that CD86 + myeloid DC AC (absolute count) is a potential protective factor of uterine infertility. Myeloid DC cells are specialized antigen-presenting cells, responsible for capturing, processing, and presenting antigen to T cells [ 37 ]. CD86 has a similar sequence to CD80, and both bind to CD28 and CTLA-4 on the surface of T cells to regulate T cell activity [ 38 ]. It has been found that local inflammatory response caused by endometrial scratching can increase the success rate of implantation in IVF and embryo transfer, which also seems to suggest that a group of strong APCs like DCs may play an important role in ensuring implantation success [ 39 ]. In addition, DCs are thought to play an important role in maternal recognition of paternal antigens during implantation, by presenting soluble major histocompatibility complex (MHC) of sperm to peripherally derived T cells in the decidua, and inducing amplification of Tregs [ 40 , 41 ]. In our study, HLA DR + NK %NK and CD16 on CD14- CD16 + monocytes increase the risk of uterine infertility. HLA-DR is commonly found on the surface of activated NK cells, associated with APCs, and involved in antigen presentation to CD4 + T cells [ 42 ]. HLA-DR + NK cells induced in vitro show high proliferative activity and high IFN-γ secretion [ 43 ]. In diseases associated with chronic inflammation, the circulating proportion of HLA-DR + NK cells may increase substantially [ 44 , 45 ]. Single-cell transcriptomic analyses reveal an abnormal proportion of NK subgroups in unexplained recurrent pregnancy loss (URPL) patients’ decidua [ 46 ]. This suggests that NK cell dysfunction may be related to the pathogenesis of URPL. CD14- CD16 + monocytes are nonclassical monocytes, with widely recognized anti-inflammatory properties [ 47 ]. Published research shows increased abundance of this monocyte subset in endometriosis, with higher levels in severe versus mild cases [ 48 ]. Furthermore, our study establishes causal roles for cytokines including IL17, IL18, and MIP1B in the development of infertility, which is easier to verify than immune cell traits, providing genetic evidence for future infertility studies. The results derived from our study may provide biomarkers for infertility screening and therapies. Our results are robust, and horizontal pleiotropy was not found. However, some limitations need to be considered. First, the second assumption of Mendelian randomization was not strictly enforced, although we removed the duplicate IVs of positive exposures to try to exclude possible confounding [ 49 ]. Second, potential gender bias arises because exposure GWAS data were not sex-stratified, while outcomes exclusively involved women [ 50 – 53 ]. Third, a looser threshold of FDR-adjusted P value was used, thus suggestive immune cell traits may exist false positives while simultaneously enabling a more comprehensive assessment of the causal relationship between the immune profile and female infertility. Fourth, the sample size of FIUTERINE is relatively small and may cause biases.

Infertility is clinically characterized by the inability to conceive after 12 months of unprotected intercourse [ 1 ]. Approximately 8–12% of couples suffer from infertility worldwide, and female infertility accounts for three-quarters of these cases [ 2 ]. The etiological landscape of female infertility is multifactorial and heterogeneous, with the main factors encompassing ovulatory dysfunction, diminished ovarian reserve, tubal diseases, endometriosis, uterine or cervical factors, and unexplained reasons [ 3 ]. Immune cells are involved in tissue damage and repair via direct cell-cell interactions and cytokine secretion, regulating the dynamics of the tissue microenvironment, and altering tissue functions [ 4 , 5 ]. Accumulating evidence highlights the crucial role of immunity in female infertility pathogenesis. Ovulation relies on paracrine mediators produced by granulosa cells and follicular membrane cells in cooperation with resident and infiltrating immune cells [ 6 , 7 ]. In cases of ovulation dysfunction, the ovary exhibits internal inflammatory disorders [ 8 ]. Polycystic ovary syndrome, frequently linked to ovulatory disorders, demonstrates an altered immune cell profile characterized by increased M1 macrophages and T lymphocytes alongside reduced dendritic cells [ 9 ]. Clinical studies reveal elevated levels of TNF-ɑ, IL-8, IL-6, and TGF-β in tubal infertility patients, suggesting a dysregulated immune milieu [ 10 ]. In addition, the immune system is essential for embryo implantation, which is a critical step in human reproduction. Cross-talk between the embryo and the abundant immune cells in maternal interfaces promotes the maternal adaptation and tolerance of the fetus, which is a prerequisite for endometrial receptivity [ 11 ]. Leukocytes and cytokines involved in this step include uterine natural killer cells, Tregs, macrophages, DCs, IL10, IL15, IFN-γ, and TGF-β [ 12 ]. Notably, a considerable proportion of endometriosis patients suffer from infertility. Compared to healthy individuals, the abundance of M1 macrophage and the reactivity of uterine natural killer cells were increased in these individuals [ 13 ]. Given the essential role of immunity in infertility, it remains to be determined which immune factors are causally associated with female infertility. Mendelian randomization (MR) is a powerful instrument to analyze causal effects by using instrumental variables (IVs) as proxies for exposures [ 14 ]. Confounders and reverse causality in observational studies can be effectively avoided through MR procedure and the exclusion of confounding variables [ 15 ]. In this study, we applied two-sample MR to assess causal relationship between circulating immune cell traits, circulating inflammatory cytokines, female infertility, and its four subtypes. Additionally, multivariable Mendelian randomization (MVMR) was implemented to adjust for mediating effects among significant exposures.

Below is the link to the electronic supplementary material. Additional file 1: Contents of 731 immune cell traits and accession numbers Additional file 1: Contents of 731 immune cell traits and accession numbers Additional file 2: Contents of 41 circulating cytokines and accession numbers Additional file 2: Contents of 41 circulating cytokines and accession numbers Additional file 3: IVs of 731 immune cell traits Additional file 3: IVs of 731 immune cell traits Additional file 4: IVs of 41 cytokines Additional file 4: IVs of 41 cytokines Additional file 5: Results of the causal effects of 731 immune cell traits and five types of female infertility Additional file 5: Results of the causal effects of 731 immune cell traits and five types of female infertility Additional file 6: Sensitivity analysis results of causal effects of significant immune factors on five types of infertility. Fig. S1 : Funnel plots of significant immune cell traits on female infertility. Fig. S2 : Funnel plots of significant cytokines on female infertility. Fig. S3 : Leave-one-out of significant immune cell traits on female infertility. Fig. S4 : Leave-one-out of significant cytokines on female infertility. Table S1. Sensitivity analysis of significant immune factors on female infertility Additional file 6: Sensitivity analysis results of causal effects of significant immune factors on five types of infertility. Fig. S1 : Funnel plots of significant immune cell traits on female infertility. Fig. S2 : Funnel plots of significant cytokines on female infertility. Fig. S3 : Leave-one-out of significant immune cell traits on female infertility. Fig. S4 : Leave-one-out of significant cytokines on female infertility. Table S1. Sensitivity analysis of significant immune factors on female infertility Additional file 7: Results of the causal effects of 41 circulating cytokines and five types of female infertility Additional file 7: Results of the causal effects of 41 circulating cytokines and five types of female infertility Additional file 8: Table S2 : MR results of the causal role between significant immune cell traits and FIUTERINE after removing duplicated IVs Additional file 8: Table S2 : MR results of the causal role between significant immune cell traits and FIUTERINE after removing duplicated IVs