Proteome-wide Mendelian randomization identifies β-nerve growth factor as a target for endometriosis

In summary, our study provides compelling evidence for the causal role of β-NGF in endometriosis and identifies potential therapeutic targets for this debilitating condition. By integrating genetic and pharmacological insights, this research advances our understanding of endometriosis pathogenesis and highlights new opportunities for drug development. Future efforts should focus on translating these findings into clinical applications, with the ultimate goal of improving outcomes for patients with endometriosis.

Endometriosis, a chronic inflammatory disorder affecting millions of women worldwide, is characterized by the growth of endometrial-like tissue outside the uterus. [ 1 , 2 ] Despite its prevalence and significant impact on fertility and quality of life, the underlying mechanisms driving endometriosis remain poorly understood. [ 3 , 4 ] Inflammation has been increasingly recognized as a key contributor to the development and progression of this condition, with numerous studies implicating dysregulated inflammatory pathways in its pathogenesis. [ 5 – 8 ] However, whether these inflammatory proteins play a causal role in endometriosis or are merely bystanders in the disease process remains unclear. To address this question, we utilized a 2-sample Mendelian randomization (MR) framework, a genetic approach that leverages instrumental variables (IVs) to infer causality while minimizing confounding and reverse causation. [ 9 ] We leveraged genetic variants associated with 91 circulating inflammatory proteins, derived from a large-scale, unbiased proteomic study. [ 10 ] This hypothesis-free approach allowed us to systematically assess the causal roles of a broad spectrum of inflammatory mediators, ranging from well-characterized cytokines to less studied proteins, in endometriosis risk. Furthermore, we validated our findings in independent cohorts and explored potential therapeutic targets through drug repurposing analysis. This study provides novel insights into the inflammatory mechanisms underlying endometriosis and highlights potential avenues for therapeutic intervention.

Conceptualization: Chenxi Gao, He Jia. Data curation: He Jia. Formal analysis: Chenxi Gao. Investigation: He Jia. Methodology: Chenxi Gao, He Jia. Software: Chenxi Gao, He Jia. Writing – original draft: Chenxi Gao. Writing – review & editing: He Jia.

We investigated the causal relationship between 91 inflammatory proteins and endometriosis using a 2-sample MR approach. This study is based on 3 fundamental assumptions: the IVs must demonstrate a significant association with the exposure factors [ 11 ] ; the selected IVs should not be related to potential confounding factors [ 12 ] ; and the selected IVs should not be associated with the outcomes, except through the pathways of the exposure factors. The main design of this study is illustrated in Figure 1 . Study design. The genetic instruments for the inflammatory proteins were sourced from a large-scale protein quantitative trait loci (pQTL) study by Zhao et al. [ 10 ] This study measured 91 circulating inflammatory proteins, selected to represent a core set of human inflammatory biomarkers, in 14,824 participants of European ancestry. The selected panel encompasses key players across major immune and inflammatory pathways, providing a comprehensive framework for our hypothesis-free MR investigation. [ 10 ] The pQTLs were classified into 2 categories: cis-pQTLs, which are defined as significant single nucleotide polymorphisms (SNPs) ( P -value < 5 × 10 − 8) located within ± 1 Mb of the gene region, and trans-pQTLs, which encompass all other significant SNPs that fall outside the boundaries of the cis-pQTLs. The genome-wide association study (GWAS) data for endometriosis used in the preliminary analysis were from the Finnish cohort ( https://www.finngen.fi/en ), including 15,088 cases and 107,564 controls. For the validation analysis phase, the endometriosis GWAS data were sourced from the UK bank cohort ( https://www.nealelab.is/uk-biobank ). To minimize population stratification bias and ensure the comparability of genetic instruments, both the discovery (FinnGen) and validation (UK Biobank) GWAS cohorts were restricted to the individuals of European ancestry. This design ensures that major differences in linkage disequilibrium patterns across ancestries do not confound our MR estimates or the validation process. Endometriosis is clinically defined as the presence of ectopically located functional endometrial tissue growing outside the uterine cavity. This definition aligns with the diagnostic criteria established by the Finngen AND UK bank database. This study followed the principles of the Declaration of Helsinki. The GWAS data summaries used for the MR analysis were ethically approved by the appropriate committees. These data could be downloaded publicly. In our study, we performed statistical analysis and created visualizations using R software version 4.2.2. Specifically, we conducted MR analysis by employing the “TwoSampleMR” package. To ensure independence and significance, we carefully selected SNPs with linkage disequilibrium clustering r 2  < 0.001 and a P -value 10 is a widely used empirical threshold indicating a sufficiently strong instrument that minimizes bias. To estimate causal relationships, we used the Wald ratio when only 1 SNP was available for an inflammatory protein and the inverse variance weighting method when multiple SNPs were present. [ 13 ] We employed a false discovery rate (FDR) correction on the MR results. A causal relationship between inflammatory proteins and endometriosis was deemed significant when the FDR was below 0.05. Additionally, the Cochran Q test was employed for heterogeneity assessment. The MR Egger intercept method was utilized to test for horizontal pleiotropy. In the sensitivity analysis, our focus was on conducting a reverse causality analysis, Bayesian colonization analysis, and phenotype scanning. To rule out the possibility that the genetic associations were driven by reverse causation, we performed bidirectional MR. For the reverse analysis, we selected independent (r² < 0.001) and genome-wide significant ( P  < 5 × 10⁻⁸) IVs for endometriosis from the FinnGen cohort. We then extracted their effects on the 3 inflammatory proteins of interest (beta nerve growth factor [β-NGF], C-X-C motif chemokine 11 [CXCL11], and signaling lymphocytic activation molecule [SLAM]) from the Zhao et al pQTL summary statistics. The primary method for causal estimation was the inverse variance weighting method. A significance threshold of P  < .05 was used to evaluate evidence for reverse causality. Bayesian colocalization analysis was performed using the “coloc” package. This analysis was based on 5 hypotheses [ 14 ] : H0: The genetic region is not associated with endometriosis or a specific metabolite. H1: The genetic region is solely associated with endometriosis. H2: The genetic region is solely associated with the specified inflammatory protein. H3: Both endometriosis and the inflammatory protein are associated with the genetic region but are influenced by different causal variants. H4: endometriosis and the inflammatory protein are connected to the genetic region and share the same causal variants. Given the limited power of colocalization analysis, we restricted our interpretation to gene loci where the combined evidence for a shared genetic signal was strong (PPH3 + PPH4 ≥ 0.8). [ 14 – 16 ] We utilized online platforms for phenotype scanning ( http://www.mulinlab.org/vportal/index.html ). The criteria for phenotype screening were as follows: A strong correlation ( P  < 10 –8 ) and GWAS studies focused on individuals of European descent. During the validation phase, we exclusively validated the causal association between inflammatory proteins that remained statistically significant post FDR correction in the primary analysis and endometriosis. The significance threshold for the validation analysis was set at a P -value of .05. We explored the DrugBank database to identify potential drugs associated with the significant inflammatory proteins found in our primary analysis ( https://www.drugbank.ca ).

In accordance with the MR hypothesis, we selected a total of 104 cis-acting SNPs associated with 56 proteins and 151 trans-acting SNPs linked to 58 proteins (see Tables S1 and S2, Supplemental Digital Content, https://links.lww.com/MD/Q622 ). We first assessed the strength of our genetic instruments to rule out potential weak instrument bias. The F-statistics for the instruments of all inflammatory proteins were significantly >10. In the cis-QTL data, the beta nerve growth factor (β-NGF) was the only inflammatory protein that shows a significant association with endometriosis. β-NGF had been identified as a risk factor for endometriosis. As levels of β-NGF rise, the risk of developing endometriosis also increases (OR = 2.23; 95% CI: 1.60–3.09; P -value = 1.75 × 10 −6 ) (Table 1 ). In the trans-QTL data, 2 inflammatory proteins, CXCL11 and SLAM, were found to have a causal association with endometriosis. Elevated levels of CXCL11 (OR = 0.74; 95% CI: 0.62–0.87; P -value = 4.12 × 10 −6 ) and SLAM (OR = 0.74; 95% CI: 0.62–0.89; P -value = 1.28 × 10 −3 ) were associated with an increased risk of endometriosis (Table 1 ). The primary MR analysis revealed no evidence of heterogeneity or pleiotropy (Table S3, Supplemental Digital Content, https://links.lww.com/MD/Q622 ). MR results for cis-QTL and trans-QTL significantly associated with Endometriosis after FDR correction. IVW = inverse variance weighting, FDR = false discovery rate, MR = Mendelian randomization, QTL = quantitative trait loci, SLAM = signaling lymphocytic activation molecule, SNP = single nucleotide polymorphism. Sensitivity analysis outcomes, which substantiate the reliability of our primary results, were systematically presented in Table 2 . To assess the potential for reverse causation, we performed MR using genetic instruments for endometriosis to test its effect on circulating levels of β-NGF, CXCL11, and SLAM. We found no evidence that genetic predisposition to endometriosis had a causal effect on the levels of any of these proteins (β-NGF: OR = 1.01, 95% CI: 0.96–1.07, P  = .68; CXCL11: OR = 1.03, 95% CI: 0.98–1.09, P  = .24; SLAM: OR = 1.03, 95% CI: 0.96–1.09, P  = .43; Table S4, Supplemental Digital Content, https://links.lww.com/MD/Q622 ). The null results from this reverse MR analysis further support the directionality of our primary findings, indicating that changes in β-NGF levels are likely a cause rather than a consequence of endometriosis. In the Bayesian colocalization analysis, the results suggested that β-NGF exhibited genetic variations similar to those associated with endometriosis (PPH3 + PPH4 = 97.22%). However, the PPH3 + PPH4 values for both CXCL11 and SLAM were below 80% (Table 2 ). In the phenotype screening analysis, no statistically significant correlation was observed between β NGF levels and other clinical manifestations (Table S5, Supplemental Digital Content, https://links.lww.com/MD/Q622 ). CXCL11 and SLAM had been identified as significant biomarkers associated with diverse phenotypic manifestations, including blood cells, hypothyroidism, blood pressure, serum lipids, diabetes, cardiovascular disease, physical measurement parameters, cerebrovascular disease, primary sclerosing cholangitis, multiple sclerosis, colorectal cancer, chronic inflammatory diseases, and glaucoma. In addition, CXCL11 had been found to be associated with cognition and education. SLAM was found to be associated with regarding albumin, valine, and menarche. Summary of sensitivity analysis results. IVW = inverse variance weighting, MR = Mendelian randomization, QTL = quantitative trait loci, SLAM = signaling lymphocytic activation molecule, Validation results are detailed in Table 3 . Only β-NGF demonstrated reproducible significance ( P  = .04), while associations for CXCL11 and SLAM were not replicated. DrugBank screening identified 5 potential therapeutic candidates targeting β-NGF: Clenbuterol (β2-adrenergic agonist), RI-624 (selective immunomodulator), Polaprezinc (zinc-L-carnosine complex), Fasinumab (NGF antibody), and Oleandrin (cardiac glycoside) (Table S6, Supplemental Digital Content, https://links.lww.com/MD/Q622 ). MR results of the validation phase. MR = Mendelian randomization, QTL = quantitative trait loci, SLAM = signaling lymphocytic activation molecule,

Our study represents one of the first comprehensive investigations into the causal relationships between inflammatory proteins and endometriosis using an MR approach. The robust association observed between β-NGF and endometriosis risk underscores the potential role of inflammatory pathways in the disease. A meta-analysis revealed significantly elevated levels of NGF in both tissue samples and serum of patients with endometriosis compared with controls. [ 17 ] β - NGF is a key regulatory factor in pain signaling and tissue nerve innervation, previously associated with chronic pain, and recently discovered to be involved in the pain mechanism of endometriosis. [ 18 – 23 ] The strong colocalization evidence (PPH3 + PPH4 = 97.22%) further supports the shared genetic basis between β-NGF and endometriosis, suggesting a potential mechanistic connection that warrants further exploration. The reverse causality analysis further substantiated that increased β-NGF levels are not merely a downstream effect of endometriosis but rather suggest a primary involvement of β-NGF in the initiation and development of the disease. NGF plays a pivotal role in promoting aberrant neuronal outgrowth and hyperinnervation in endometriosis lesions. [ 24 , 25 ] Our MR analysis identifies β-NGF as an upstream causal driver of endometriosis, a finding that is strongly supported by a wealth of existing experimental and clinical evidence delineating its multifaceted role in the disease’s pathogenesis. First, the most mature role of β-NGF in endometriosis is to sensitize pain pathways and promote neurogenesis. Elevated levels of NGF and TrkA, TrkB, and the neurotrophin receptor p75 have been consistently observed in the peritoneal fluid, serum, and endometriosis lesions of patients, and their concentrations are usually correlated with the severity of chronic pelvic pain. [ 26 ] Neurogrowth factors directly sensitize nociceptive neurons and drive the excitability of lesions, which is a hallmark of pain symptoms in this disease. Beyond its neurotrophic functions, β-NGF acts directly on endometriotic cells to enable their survival and growth in an ectopic environment. The engagement of the NGF-TrkA axis activates key downstream pro-survival and proliferative pathways, primarily PI3K/Akt [ 27 , 28 ] and MAPK/ERK, which inhibit apoptosis and promote the proliferation of ectopic endometrial cells. This provides a direct mechanistic link through which genetically elevated β-NGF levels could facilitate the establishment and maintenance of endometriotic lesions. Furthermore, β-NGF contributes to a pro-angiogenic and pro-inflammatory microenvironment. It can stimulate endothelial cell proliferation and new blood vessel formation, supplying the lesions with necessary nutrients. [ 29 ] Concurrently, NGF activates mast cells and macrophages, prompting the release of additional pro-inflammatory cytokines such as TNF-α and IL-1β, thereby creating a chronic inflammatory state that further promotes disease progression. [ 30 ] Therefore, our MR study provides a genetic anchor for this extensive biological narrative. It demonstrates that inter-individual genetic differences leading to elevated circulating β-NGF are a causal risk factor for developing endometriosis, likely through the consolidated mechanisms of enhanced neuronal sensitization, cell survival, and immunovascular remodeling. By binding to its high-affinity receptor TrkA (tropomyosin receptor kinase A), NGF activates PI3K/Akt and MAPK/ERK signaling pathways, thereby inhibiting apoptosis and promoting survival of ectopic endometrial cells in pelvic and other extrauterine environments. Furthermore, NGF may activate immune cells (e.g., mast cells and macrophages), inducing the secretion of pro-inflammatory cytokines including IL-6 and TNF-α, thereby establishing a chronic inflammatory microenvironment that promotes endometrial cell adhesion and proliferation. While extensive research has investigated the involvement of NGF in endometriosis-associated pain, limited attention has been given to its role, specifically that of β-NGF, in the underlying pathogenesis of endometriosis. These findings underscore the need for further studies to elucidate the precise mechanisms by which beta-NGF contributes to endometriosis development and to explore its potential as a therapeutic target. While CXCL11 and SLAM showed suggestive associations in our primary analysis, these findings were not replicated in the independent validation cohort. This discrepancy is not unexpected in MR studies and can be attributed to several factors. First, and most critically, the genetic instruments for CXCL11 and SLAM were trans -pQTLs. Unlike the robust cis -pQTL instrument for β-NGF, trans -pQTLs are notoriously more susceptible to heterogeneity across even closely related populations. Although both our discovery and validation cohorts are of European ancestry, fine-scale population differences between the Finnish (FinnGen) and broader UK/European (UK Biobank) populations could alter the effect size or linkage disequilibrium of these trans -acting variants, compromising their portability. Second, we cannot exclude the possibility that the initial associations represented false positives. The consistent validation of β-NGF across these populations, in contrast, underscores its robust and fundamental role in endometriosis pathogenesis. Nonetheless, the broad phenotypic associations of CXCL11 and SLAM with conditions such as autoimmune diseases, metabolic disorders, and cancer suggest their involvement in systemic inflammatory processes, which may indirectly influence endometriosis risk. Although a study has identified decreased CXCL11 levels in endometriosis patients, [ 31 ] the functional significance of CXCL11 in the initiation and development of endometriosis has not yet been elucidated. Further investigation is required to uncover the molecular mechanisms underlying the protective effects of CXCL11 and SLAM in endometriosis, potentially opening new avenues for targeted interventions. The identification of β-NGF as a validated causal factor opens new possibilities for targeted therapy. Through systematic screening of the DrugBank database, we identified 5 promising drug candidates: Clenbuterol, RI-624, Polaprezinc, Fasinumab, and Oleandrin, that modulate β-NGF signaling. Of particular interest is Fasinumab, a monoclonal antibody targeting NGF, which has shown efficacy in preclinical models of chronic pain and inflammation. [ 32 , 33 ] These findings provide a strong rationale for further investigation of β-NGF inhibitors in the context of endometriosis. Despite these advances, our study has limitations. First, the genetic instruments for the inflammatory proteins and the outcome data for endometriosis were derived exclusively from cohorts of European ancestry. While this design minimizes bias from population stratification, it limits the generalizability of our findings to other ancestral groups, such as Asian or African populations. Genetic variants, their effect sizes, and linkage disequilibrium patterns can differ across populations. Therefore, the causal role of β-NGF identified here may not be transferable, and the specific genetic instruments used may not be valid in other settings. Additionally, while MR reduces confounding, it cannot entirely eliminate the possibility of horizontal pleiotropy. Future studies should incorporate multi-ethnic cohorts and functional experiments to validate these findings and elucidate the biological mechanisms linking β-NGF to endometriosis.

We thank all the researchers and participants from the UKB and FinnGen alliances. We also thank all the researchers and participants from the study on inflammatory proteins.