Integrative Multi-omics Analysis for Prioritization of Candidate Genes in Polycystic Ovary Syndrome

The genetic architecture of polycystic ovary syndrome (PCOS) has not been fully elucidated. Translating genome‐wide association study (GWAS) loci into functional effector genes represents a key challenge for elucidating disease mechanisms and advancing targeted therapeutic strategies. In this study, the FinnGen R12 dataset was utilized in conjunction with the Genotype‐Tissue Expression Project (GTEx) v8 eQTL dataset to conduct cross‐tissue transcriptome‐wide association studies (TWAS). External validation was performed in two independent datasets, comprising a PCOS meta‐analysis of Rotterdam‐ and NIH‐criteria cohorts and UK Biobank summary statistics. To prioritize high‐confidence candidate genes, we integrated a cross‐tissue (UTMOST) TWAS and tissue‐specific (FUSION) TWAS with a gene‐based association test (MAGMA). Furthermore, causal inference analyses including Mendelian randomization (MR), colocalization analysis, summary data‐based MR, and the heterogeneity in dependent instrument (HEIDI) test were carried out on candidate genes. GeneMANIA, gene‐chemical‐disease analysis, and phenome‐wide association study (PheWAS) were further performed on candidate genes. Finally, we explored expression profiles of core candidate genes using transcriptomic datasets from the Gene Expression Omnibus (GEO) database. We prioritized 15 candidate genes potentially associated with PCOS. External analyses across independent datasets showed limited consistency, which may reflect differences in statistical power and cohort heterogeneity. MCM6 was prioritized by all three approaches, whereas the remaining 14 genes were supported by two approaches. Among the 15 candidate genes, six genes were demonstrated to have causal associations with PCOS risk. GeneMANIA network analysis further revealed that these candidate genes were involved in key biological functions, including DNA replication and regulation, DNA repair and recombination, protein‐DNA complex, cell cycle regulation, nucleic acid‐enzyme activities, DNA structure and chromosomal regions, reproductive system development, and sex differentiation. Gene‐chemical‐disease analysis suggested that the prioritized genes may mediate crosstalk between PCOS genetic risk and environmental/metabolic factors. PheWAS analysis highlighted the pleiotropic roles of PCOS candidate genes, which contributed to both PCOS susceptibility and a spectrum of PCOS‐related traits including metabolic and cardiovascular phenotypes. Finally, transcriptomic data showed downregulated MCM6 expression in granulosa cells of PCOS patients. Our study established a multipronged pipeline for PCOS gene prioritization and provided novel insights into the genetic architecture of PCOS.

colocalization, cross‐tissue TWAS, gene prioritization, GWAS, Mendelian randomization, polycystic ovary syndrome This study evaluated the association between genetically regulated gene expression and PCOS risk. The analytical strategy incorporated three computational methods, including cross‐tissue TWAS, single‐tissue TWAS, and MAGMA, which enabled the identification of 15 high‐confidence candidate genes. Six candidate genes, including MCM6, CERS5, LIMA1, MST1R, DNM3, and NR1H3, were further corroborated by MR, colocalization analysis, SMR, and the HEIDI test. Further functional analyses yielded a more comprehensive understanding of the potential roles of the candidate genes in PCOS. 1. Introduction Polycystic ovary syndrome (PCOS) is one of the most prevalent and complex reproductive endocrine disorders among women of reproductive age, with a global prevalence estimated to be 10%–13% [1]. PCOS is a heterogeneous disorder characterized by reproductive abnormalities, including hyperandrogenism, ovulatory dysfunction, and polycystic ovarian morphology [1]. Beyond the classic reproductive symptoms, PCOS has implications for metabolic, cardiovascular, and psychological health [2, 3]. PCOS constitutes a lifelong health risk for women across their lifespan. However, due to the insufficient understanding of its underlying mechanisms, current PCOS treatment remains ad hoc, emphasizing the need for research into clarifying the etiology and pathophysiology of PCOS [1]. Therefore, it is necessary to gain a comprehensive understanding of the underlying mechanism of PCOS. The discovery of novel biomarkers and development of advanced treatment will benefit patients’ lives. Evidence from familial and twin studies underscores the heritability of PCOS. First‐degree relatives, including males and nonreproductive‐age females, exhibit reproductive and metabolic phenotypes consistent with genetic susceptibility to the syndrome [4, 5]. Current evidence supports a high heritability of PCOS. Twin studies reported a heritability estimate of 0.7 [6]. Early genetic studies of PCOS mainly used a candidate gene approach, which involved examining polymorphic sites near selected genes based on their presumed involvement in pathways related to the pathogenesis of PCOS to determine potential associations [5]. In recent years, large genome‐wide association studies (GWAS) have expanded our knowledge of PCOS genetics. The latest GWAS of PCOS identified 94 unique loci associated with PCOS [7]. GWAS analysis was also utilized to identify genetic associations between PCOS and asthma‐related traits [8]. However, elucidating the potential functions of GWAS risk variants is challenging. Most of these variants are noncoding and located in intronic or intergenic regions, with their underlying molecular mechanisms remaining unclear [9]. Complicated linkage disequilibrium (LD) also poses a challenge to distinguishing the causal variants responsible for these associations [10]. The transcriptome‐wide association study (TWAS) integrates eQTL and GWAS datasets to identify new susceptibility genes via genetically predicted gene expression [11]. A cross‐tissue strategy named Unified Test for Molecular Signature (UTMOST) conducts gene‐level association analyses across multiple tissues [12]. Recently, cross‐tissue analysis has been extensively utilized to identify candidate disease‐associated genes for complex multisystem disorders, including migraine [13], obesity [14], and rheumatoid arthritis [15]. However, any individual computational approach carries the potential risk of false‐positive results. Thus, adopting a multimodal method that requires convergent evidence is essential for achieving robust gene prioritization. In this study, we initially implemented a multidimensional computational strategy to prioritize a set of high‐confidence candidate genes for PCOS. In this stage, we conducted a cross‐tissue TWAS analysis, Functional Summary‐based imputation (FUSION), and Multi‐marker Analysis of Genomic Annotation (MAGMA) to curate the panel of candidate genes. Subsequently, Mendelian randomization (MR), colocalization analysis, summary data‐based MR (SMR), and the heterogeneity in dependent instrument (HEIDI) test were carried out on candidate genes, followed by supplementary analyses including GeneMANIA, Gene‐chemical‐disease analysis, and phenome‐wide association study (PheWAS) to further elucidate the biological features. 2. Methods The analytical process is illustrated in Figure 1. 2.1. Data Source 2.1.1. PCOS GWAS Data Source The discovery genome‐wide summary statistics for PCOS were obtained from the FinnGen R12 database (https://www.finngen.fi/en), which include 42,630 cases and 239,434 controls of European ancestry. The first validation summary statistics were obtained from a large meta‐analysis of PCOS (https://www.repository.cam.ac.uk/items/3ccbf35f‐5f69‐4b46‐b4fc‐12cb67af71ae), which contains two distinct cohorts including the Rotterdam criteria cohort (2669 cases and 17,035 controls) and the NIH criteria cohort (2540 cases and 15,020 controls) [16]. The second validation resource comprised UK Biobank (UKB) GWAS summary statistics curated by the Neale Lab (http://www.nealelab.is/uk‐biobank), with a total of 209 cases and 405,386 control individuals. 2.1.2. eQTL Files Source The GTEx V8 dataset [17] provides extensive gene expression profiles derived from 49 distinct tissue types, which were collected from 838 post‐mortem individuals (https://ftp.ebi.ac.uk/pub/databases/spot/eQTL/imported/GTEx_V8). Notably, the sample counts differ across various tissues, with the renal cortex having the smallest sample size (n = 73) and the skeletal muscle possessing the largest (n = 706). 2.2. Cross‐Tissue TWAS Analyses Cross‐tissue analyses were performed using the UTMOST method (https://github.com/Joker‐Jerome/UTMOST?tab=readme‐ov‐file) to quantify gene‐trait associations at the organismal level. This approach facilitates the detection of an expanded set of genes across tissues via enriched trait heritability while enhancing the accuracy of expression imputation concurrently [12]. For the integration of gene‐trait association results, the generalized Berk‐Jones (GBJ) test was applied, which utilizes covariance derived from single‐tissue statistical outputs [18]. After false discovery rate (FDR) adjustment, associations with an FDR < 0.05 were regarded as statistically significant. 2.3. Single Tissue TWAS Analyses The FUSION tool was employed to conduct a TWAS that integrated GWAS data with eQTL data from 49 tissues in the GTEx V8 dataset to evaluate gene‐disease associations [11]. Initially, the LD between the predictive model and SNPs at individual GWAS loci was evaluated using European‐ancestry samples from the 1000 Genomes Project. Subsequently, FUSION employed multiple predictive models, including BLSMM, BLUP, Elastic Net, LASSO, and the top 1 model, to assess the overall impact of SNPs on gene expression weights. These gene weights were subsequently integrated with the genetic effect size of PCOS (represented by the PCOS GWAS Z‐score) to perform the TWAS. Multiple testing correction was performed using the FDR approach, and associations with an FDR < 0.05 were considered statistically significant. 2.4. Gene Analysis of MAGMA Gene analysis was performed using MAGMA software (version 1.10), which utilized data from Phase 3 of the 1000 Genomes Project (Build 37; European only) to aggregate SNP‐level summary statistics to generate gene scores. This procedure enabled the quantification of the associations between each gene and the phenotype [19, 20]. Additional details regarding parameter settings and methodological justifications can be found in the original MAGMA documentation. 2.5. Mendelian Randomization and Colocalization Analysis We performed MR analysis using the “TwoSampleMR” R package [21]. We used cis‐eQTL SNPs as instrumental variables (IVs), gene expression as exposure, and GWAS data on PCOS as outcome. In the first stage, genome‐wide significant SNPs (p < 5E‐08) were selected. Subsequently, LD clumping was applied to identify independent genetic variants (r2 < 0.001). The Wald Ratio method was utilized to calculate the MR effect size, with statistical significance assessed at p < 0.05. Bayesian colocalization analysis was conducted using “coloc” R package to determine whether there is an overlap of causal loci between GWAS and eQTL signals. The analysis focused on the posterior probability (PP) of five potential causal configurations. A PP.H4 value greater than 0.6 indicated the presence of shared causal variants between GWAS and eQTL datasets. 2.6. SMR Analysis and the HEIDI Test SMR analysis and the HEIDI test were completed based on SMR 1.3.1 software. The p value was adjusted using the Benjamini‐Hochberg method to control for the FDR. To exclude confounding effects arising from LD between distinct causal variants, the HEIDI test was applied to distinguish pleiotropy from linkage, using a threshold of p‐HEIDI > 0.01. 2.7. GeneMANIA Analysis The GeneMANIA platform was employed to acquire diverse datasets, including genetic interaction data, pathway information, gene coexpression, physical interactions, and colocalization data for candidate genes. This platform augments the understanding of the potential biological functions of the candidate genes. 2.8. Gene‐Chemical‐Disease Analysis To identify chemical drugs associated with the candidate genes of PCOS, we performed targeted searches for the aforementioned candidate genes in the Comparative Toxicogenomics Database (CTD, https://ctdbase.org/). As a comprehensive public database, CTD integrates extensive datasets related to chemical‐gene/protein interactions, as well as chemical‐disease and gene‐disease associations, which facilitates inferring environmental influence on diseases. 2.9. PheWAS Analysis of Candidate Genes for PCOS To systematically evaluate the associations of genetic variants and diverse human phenotypes, a PheWAS was conducted based on exome sequencing and phenotype data hosted by the AstraZeneca PheWAS Portal [22]. This portal contains comprehensive datasets that facilitate the identification of putative associations between genetic variants and diverse phenotypes. The original study explored associations between shared genes and numerous binary and quantitative phenotypes derived from medical records of approximately 500,000 UK Biobank participants [22]. Based on UK Biobank data, the analysis systematically examined horizontal pleiotropy as well as diverse phenotypic profiles linked to candidate genes for PCOS. 2.10. Expression Profiles of Key Candidate Genes Differential expression analyses were performed on publicly available GEO dataset GSE34526 utilizing the “limma” package, with significance defined by a moderated t‐statistic p < 0.05. Results were visualized using “ggplot2,” and boxes were used to further characterize the expression profiles of core candidate genes. 3. Results 3.1. TWAS Analysis In the cross‐tissue analysis of the discovery dataset, 223 genes associated with PCOS were found with p values less than 0.05 (Table S1), out of which eight genes associated with PCOS exhibited statistically significant after FDR correction (FDR < 0.05) (Table S2), comprising seven protein coding genes (R3HDM1, MCM6, RAB3GAP1, DARS1, ZRANB3, RHOB, KCNK3), and one processed pseudogene (HNRNPKP2). For the single‐tissue analysis, 62 genes linked to PCOS had FDR < 0.05 in at least one tissue (Table S3). Two candidate genes met stringent thresholds in both cross‐tissue analysis and single‐tissue analysis, comprising two protein‐coding genes (MCM6, KCNK3) (Table 1). In independent cross‐tissue and single‐tissue validation analyses, we did not observe consistent evidence supporting the candidate genes identified in the discovery stage (Tables S15–S17). TABLE 1. | Gene | UTMOST | FUSION | MAGMA | Validation count | Validation path | |---|---|---|---|---|---| | MCM6 | √ | √ | √ | 3 | UTMOST&FUSION&MAGMA | | ARL14EP | √ | √ | 2 | FUSION&MAGMA | | | CAMKV | √ | √ | 2 | FUSION&MAGMA | | | CERS5 | √ | √ | 2 | FUSION&MAGMA | | | COX14 | √ | √ | 2 | FUSION&MAGMA | | | DNM3 | √ | √ | 2 | FUSION&MAGMA | | | FAM186A | √ | √ | 2 | FUSION&MAGMA | | | FSHB | √ | √ | 2 | FUSION&MAGMA | | | KCNK3 | √ | √ | 2 | UTMOST&FUSION | | | LHB | √ | √ | 2 | FUSION&MAGMA | | | LIMA1 | √ | √ | 2 | FUSION&MAGMA | | | MST1R | √ | √ | 2 | FUSION&MAGMA | | | NR1H3 | √ | √ | 2 | FUSION&MAGMA | | | R3HDM1 | √ | √ | 2 | UTMOST&MAGMA | | | VPS13B | √ | √ | 2 | FUSION&MAGMA | 3.2. Gene Analysis of MAGMA MAGMA gene analysis identified 36 candidate genes associated with PCOS with significant associations (FDR < 0.05) (Table S4). To enhance the stability of these discoveries, we further integrated the MAGMA results with those from the UTMOST cross‐tissue analysis and the FUSION single‐tissue analysis, respectively. Notably, 13 candidate genes met strict criteria in both MAGMA and single‐tissue analysis. Two candidate genes passed the screening criteria of both MAGMA and cross‐tissue analysis (Table 1). In the validation datasets, MAGMA analysis did not yield significant positive results after FDR correction. 3.3. Convergent Evidence Integration Convergent evidence integration demonstrated that 15 final candidate genes received support from at least two distinct independent methods, achieving robust statistical validation (Table 1). One gene (MCM6) exhibited statistical significance across all three approaches, marking MCM6 as our top‐confidence discovery. The remaining 14 genes (ARL14EP, CAMKV, CERS5, COX14, DNM3, FAM186A, FSHB, KCNK3, LHB, LIMA1, MST1R, NR1H3, R3HDM1, VPS13B) exhibited strong evidence from two approaches, furnishing additional candidates with reliable support. 3.4. MR and Colocalization Results To further establish causal relationships between candidate genes and PCOS, we conducted MR and colocalization analysis. Among the 15 candidate genes identified from the multidimensional computational strategy, convincing MR evidence was observed for six genes, including MCM6, CERS5, LIMA1, MST1R, DNM3, and NR1H3. MR analysis confirmed causal negative associations between MCM6 and PCOS in four tissues (p < 0.05) (Figure 2, Table S5). Colocalization analysis further confirmed this finding, with PPH4 values of 0.68, 0.86, and 0.68 for Adipose_Subcutaneous, Artery_Coronary, and Cells_Cultured_fibroblasts, respectively (Figure 3, Table S6). MR analysis further revealed that both CERS5 and LIMA1 were positively associated with PCOS in 13 tissues and two tissues, respectively (p < 0.05) (Table S5). MST1R was found to have a positive correlation with PCOS in seven tissues through MR analysis (p < 0.05) (Table S5). However, it also identified a negative association between MST1R and PCOS in whole blood (Table S5). MR analysis identified causal positive associations between DNM3 and PCOS in three tissues and negative associations in two tissues (p < 0.05) (Table S5). MR analysis revealed significant associations of NR1H3 with PCOS in two tissues (p < 0.05) (Table S5), with opposite directions of effect across tissues. 3.5. SMR and the HEIDI Test Results We next performed SMR analysis for the 15 candidate genes identified by the integrative prioritization strategy combining UTMOST, FUSION, and MAGMA. Ten of these genes showed nominally significant SMR signals (p 0.01), including ARL14EP, CAMKV, CERS5, COX14, DNM3, LHB, LIMA1, MCM6, MST1R, and NR1H3 (Table S14). These results provided evidence that genetically regulated gene expression of these genes may be involved in PCOS susceptibility and suggested that the observed associations were more likely attributable to a shared underlying variant rather than LD between distinct causal variants. Importantly, the SMR analysis replicated six key candidate genes (MCM6, CERS5, LIMA1, MST1R, DNM3, NR1H3) from MR analysis. This cross‐method consistency strengthened the robustness of the implicated genes and reduced the likelihood that the findings were driven by a single analytic approach. 3.6. GeneMANIA Network Analysis Results To provide a systems‐level view of the broader candidate panel, we first conducted an overall GeneMANIA network for the 15 PCOS candidate genes, including physical interactions, coexpression, predicted interactions, colocalization, genetic interactions, pathways, and shared protein domains. The functional enrichment analysis of the network of the 15 candidate genes revealed enrichment in several major biological themes, including DNA replication and regulation, DNA repair and recombination, protein‐DNA complex, cell cycle regulation, nucleic acid‐enzyme activities, DNA structure and chromosomal regions, reproductive system development, and sex differentiation (Figure 4A, Table S7). These findings suggested partial functional overlap, particularly in processes related to genome maintenance and cell‐cycle control, together with prominent functional heterogeneity among candidate genes. We then focused on the six MR‐supported genes for downstream analyses. Potential gene interaction networks separately constructed for these six candidate genes with putative causal associations with PCOS risk were presented in Figure 4B–G. The MCM6‐related network was most significantly enriched in DNA replication (Figure 4B, Table S8), whereas the LIMA1‐related gene network was enriched in positive regulation of protein‐containing complex assembly, the replisome and the nuclear replisome (Figure 4E, Table S11), suggesting overlap in replication‐associated processes. In contrast, the CERS5‐related gene network was primarily enriched in the sphingolipid biosynthetic process and ceramide metabolic process (Figure 4C, Table S9). The DNM3‐associated gene network showed the most significant enrichment in synaptic vesicle recycling and presynaptic endocytosis (Figure 4D, Table S10). The MST1R‐related gene network showed the most significant enrichment in positive regulation of intracellular protein transport (Figure 4F, Table S12). The most significantly enriched functional term in NR1H3‐related gene network was RNA polymerase II‐specific DNA‐binding transcription factor binding (Figure 4G, Table S13). Overall, these results indicated that although the prioritized genes did not all participate in the same pathway, subsets of genes may converge on shared functional modules relevant to PCOS. 3.7. Gene‐Chemical‐Disease Analysis Findings The CTD was utilized to explore potential chemical‐gene interactions involving the six genes (MCM6, CERS5, DNM3, LIMA1, MST1R, NR1H3), which were shown to have causal associations with PCOS (Figure 5). CTD analysis identified a diverse set of chemicals interacting with MCM6 gene, such as bisphenol A, Phenobarbital, Benzo(a)pyrene, pirinixic acid, Estradiol, Ozone, Tetrachlorodibenzodioxin, Aflatoxin B1, Cadmium Chloride, and Air Pollutants (Figure 5A). The CERS5 gene was predicted to interact with bisphenol A, Benzo(a)pyrene, Tetrachlorodibenzodioxin, titanium dioxide, Acetaminophen, and Aflatoxin B1 (Figure 5B). The DNM3 gene demonstrated the potential for interactions with Valproic Acid, bisphenol A, titanium dioxide, and Air Pollutants (Figure 5C). LIMA1 exhibited potential interactions with bisphenol A, Benzo(a)pyrene, Tetrachlorodibenzodioxin, Estradiol, Particulate Matter, and Ozone (Figure 5D). CTD analysis also indicated that MST1R was associated with Acetaminophen, bisphenol A, Lipopolysaccharides, Benzo(a)pyrene, and nickel sulfate (Figure 5E). Our findings from CTD also indicated that NR1H3 was associated with numerous chemicals, such as T0901317, Fatty Acids, Dietary Fats, Fenofibrate, Cholesterol, and bisphenol A (Figure 5F). 3.8. PheWAS Analysis Results Additionally, a variant‐level PheWAS analysis was performed for the six prioritized PCOS candidate genes (MCM6, CERS5, DNM3, LIMA1, MST1R, NR1H3) using the UK Biobank 470 k (v5) Public dataset from the AstraZeneca PheWAS Portal (Figure 6). Overall, these loci displayed heterogeneous phenome‐wide association patterns across binary and continuous phenotypes, indicating substantial differences in their pleiotropic architectures. Among the six genes, MST1R and NR1H3 showed the broadest and strongest association signals. MST1R showed the strongest overall pleiotropic pattern among the six genes. In binary traits, its signals were most evident in endocrine/metabolic‐related phenotypes, with additional associations involving respiratory, musculoskeletal, and lab‐related categories. In continuous traits, MST1R displayed extensive associations spanning proteomic, cardiovascular, respiratory, musculoskeletal, laboratory, and healthcare‐related domains (Figure 6G–I). A distinct pattern was observed for NR1H3, whose variants were enriched for associations with cardiovascular‐related binary phenotypes and showed particularly strong signals in NMR metabolomics‐related quantitative traits, together with additional associations in proteomics, eye, cardiovascular, musculoskeletal, laboratory, and health service‐related measures (Figure 6K,L). In contrast, DNM3 showed comparatively sparse signals in both binary and continuous phenotypes, suggesting relatively limited phenome‐wide relevance in the current dataset (Figure 6E,F). The remaining three loci showed intermediate patterns. MCM6 displayed only modest associations with binary phenotypes, whereas its variants showed a much richer pattern of association with continuous traits, such as proteomic and NMR metabolomics, implying that its effects may be more readily reflected in quantitative biological measures than in dichotomous disease outcomes (Figure 6A,B). Variants mapped to CERS5 were mainly associated with cardiovascular phenotypes and also demonstrated broader associations with continuous traits, particularly those related to proteomic, cardiovascular, respiratory, laboratory biomarkers, and health service‐related traits (Figure 6C,D). LIMA1 showed a more selective profile, with more evident signals in binary phenotypes and less extensive but still observable associations in continuous traits (Figure 6G,H). Overall, these findings highlighted the pleiotropic roles of PCOS candidate genes, which contributed to both PCOS susceptibility and a spectrum of PCOS‐related traits including metabolic and cardiovascular phenotypes. 3.9. Transcriptomic Validation in PCOS Datasets To further assess the clinical significance of the key candidate genes of PCOS, we analyzed differential expression using the GEO dataset. The results demonstrated that MCM6 was significantly downregulated in granulosa cells of PCOS patients (GSE34526, p < 0.05). However, none of the other 14 candidate genes exhibited significant expression differences in granulosa cells of PCOS patients (Figure S1). 4. Discussion In this study, we utilized PCOS GWAS and GTEx V8 data to implement a multidimensional computational strategy that selected a set of high‐confidence candidate genes for PCOS. Our systematic analyses included an evaluation of the association between genetically regulated gene predisposition and PCOS‐related disease risk. The strategy included three computational methods comprising cross‐tissue TWAS, single‐tissue TWAS, and MAGMA, enabling the identification of 15 high‐confidence candidate genes, which were further corroborated by MR, colocalization analysis, summary data‐based MR, and the HEIDI test. Further functional analyses yielded a more comprehensive understanding of the potential roles of the candidate genes for PCOS. Currently, the integration of multi‐omics association studies has emerged as a prevalent strategy for identifying disease susceptibility genes. To date, the largest GWAS meta‐analysis of PCOS has identified 94 potential PCOS susceptibility genes. Xu et al. identified seven druggable genes associated with PCOS risk using a multi‐omics approach [23]. In another study, imputation models of two TWAS methodologies, S‐PrediXcan and S‐MultiXcan, were applied to prioritize risk loci for PCOS and potential susceptibility genes using eQTL data from GTEx V7 [24]. However, discrepancies in the findings of different studies often stem from variations in sample size, sample origin, and algorithmic differences. Therefore, it is important to implement a multidimensional computational strategy to screen for potential genetic factors involved in PCOS susceptibility. Compared with GTEX v7, the updated GTEX v8 has a 49% expansion in RNA‐seq sample size, with tissue donors increasing by 33%, and it further includes splicing eQTL data [25]. The UTMOST approach elevates statistical power, advances our understanding of the associations between genes and complex traits, and facilitates the detection of otherwise neglected relationships when conducting single‐tissue analyses independently [12]. Using this multipronged framework, we identified 15 candidate genes for PCOS. Notably, six of these candidate genes demonstrated causal associations with the risk of PCOS according to MR and SMR analyses. Among the 15 candidate genes, the extent of prior evidence related to PCOS was variable. ARL14EP [24], FSHB [26], and LHB [27] had comparatively stronger support from previous studies, while MCM6 [28], DNM3 [29], NR1H3 [30], KCNK3 [31], LIMA1 [32], and MST1R [33] were supported primarily by limited or indirect evidence based on the current literature. For CAMKV, CERS5, COX14, FAM186A, R3HDM1, and VPS13B, there was currently little evidence directly linking them to PCOS. MCM6 is a critical protein‐coding gene that regulates DNA replication licensing and exerts an essential function in the progression of the cell cycle [34]. Accumulating evidence indicates that MCM6 modulates multiple biological processes, including cell cycle progression, proliferation, metastasis, immune response, and the maintenance of the DNA replication system [35]. Current research on MCM6 primarily focuses on the pathological mechanism of cancer. MCM6 is a shared susceptibility gene for obesity and metabolic disorders, with causal associations and independent signals. A previous study also suggested that MCM6 may influence the risk of obesity and metabolic disorders by regulating DNA replication, cell proliferation, and interactions with chemical responses [14]. Another study genotyped adult‐type hypolactasia (ATH) based on MCM6 gene variants and found that ATH is associated with PCOS [28]. Importantly, our findings should also be interpreted in the context of previous evidence implicating the MCM gene family in PCOS. A recent study identified a high frequency of potentially pathogenic rare variants in MCM family genes in Chinese Han women with PCOS, suggesting that genetic alterations in this gene family may contribute to disease susceptibility [36]. Such evidence was broadly consistent with our results and further supported the notion that MCM‐associated signaling pathways may exert a critical function in PCOS pathogenesis. While the above study emphasized rare mutation events, our study highlighted MCM6 from the perspective of genetically regulated expression and causal inference. Taken together, these findings suggested that both sequence variation and transcriptional dysregulation of MCM family genes may be relevant to PCOS. To elucidate the causal relationship between MCM6 expression and the risk of PCOS, we conducted MR analysis in this study. MR results suggested that MCM6 exhibited a large negative association with PCOS in four tissues, with the strongest inverse effect and colocalization evidence detected in coronary artery, indicating that reduced MCM6 expression may contribute to PCOS susceptibility, whereas higher expression may exert a protective effect. Accumulating evidence indicated that PCOS exerts profound adverse effects on cardiometabolic health. Early subclinical vascular lesions, such as coronary atherosclerosis and endothelial dysfunction, can emerge in young patients and persist with aging [37]. The inverse association detected in adipose and coronary artery tissue suggested that MCM6 may influence PCOS risk through pathways relevant to metabolic homeostasis and vascular abnormalities. The particularly strong signal in coronary artery may indicate that the genetic regulation of MCM6 in vascular tissue was relevant to the cardiometabolic component of PCOS. In addition, transcriptomic profiling also identified downregulated MCM6 expression in granulosa cells of PCOS patients. Taken together, these findings implied the potential role of MCM6 in the pathogenesis of PCOS. However, the currently available evidence for MCM6 in PCOS was still largely limited to the transcriptomic level. To further explore whether MCM6 has been characterized at the protein level, we examined the PCOSBase database and the related studies cited within this resource [38]. The available evidence identified for MCM6 was currently restricted to gene‐expression analyses, including reports of reduced MCM6 expression in PCOS endometrial tissue at the transcript level [39, 40], whereas evidence at the protein level remained lacking. Because mRNA abundance does not always correlate with protein expression owing to posttranscriptional and translational regulation, future studies are needed to determine whether MCM6 is also dysregulated at the protein level in PCOS‐relevant tissues or cells, such as granulosa cells or endometrial tissues. Bioinformatics analyses supported an association between MCM6 and PCOS, which may be mediated through the participation of MCM6 in DNA replication, cell proliferation and chemical response pathways. MCM6 is a key component of the minichromosome maintenance family and plays an essential role in replication licensing and genome stability [41]. As impaired granulosa cell growth, follicular stasis and ovarian dysfunction are pathological features of PCOS [42], dysregulated MCM6 expression may contribute to PCOS by disrupting follicular cell growth and ovarian homeostasis. MCM6 has been reported to activate the MEK/ERK signaling pathway in hepatocellular carcinoma [43], whereas ERK‐1 and ERK‐2 have been implicated in PCOS pathogenesis [44]. In addition, the PI3K/AKT pathway plays an important role in insulin resistance in PCOS [45], and MCM family genes have been suggested to be associated with PI3K/AKT‐related signaling in cancer [46]. Although these findings do not establish a direct mechanistic link between MCM6 and ERK or PI3K/AKT signaling in PCOS, they suggest a potential direction for future mechanistic studies. Nevertheless, further studies are still required to establish conclusive evidence. In addition, the functional interpretation of the MCM6 locus should not be restricted to a nearest‐gene network. MCM6 is located within a large topologically associated domain (TAD), shows extensive cis‐interaction with neighboring genomic regions, and overlaps a region enriched in regulatory elements and large enhancers. Therefore, genetic variation within or near this locus may influence the expression of distal interacting genes, even if those genes do not themselves harbor PCOS‐associated SNPs. MCM6 has previously been shown to act as a regulatory element of lactase expression, which indicated that MCM6 region may function not only as a coding locus but also as a broader regulatory hub [47]. Accordingly, our stringent candidate‐gene prioritization strategy, also useful for reducing false positives, may not fully capture all biologically relevant effector genes at the locus. Integrating three‐dimensional genome organization data, such as Hi‐C maps from ovary, adipose tissue, or other PCOS‐relevant tissues, may help refine causal gene identification and clarify whether the observed signals act through MCM6 or through long‐range regulation of other target genes. Although such analyses are beyond the scope of the present study, they represent an important research direction for further elucidating the biological significance of the MCM6 locus. CERS5 is a pivotal protein that governs the biosynthesis of C16‐ceramide, a sphingolipid engaged in numerous signaling pathways [48]. Elevated ceramide levels have been observed in women with PCOS relative to healthy individuals [49]. As a bioactive cellular enzyme, CERS5 is involved in pathological metabolic processes and several diseases, including colorectal cancer [49], gastric cancer [50], and nonalcoholic fatty liver disease (NAFLD)‐associated fibrosis [51]. In this study, MR analysis demonstrated a significant association between CERS5 and an increased risk of PCOS in 13 tissues, including 5 brain structural regions. Studies have demonstrated that PCOS is closely associated with structural and functional brain abnormalities, with regional brain alterations linked to glucose metabolic disturbance, disordered eating, and cognitive‐emotional dysfunction [52]. The positive association and strong colocalization evidence detected in brain tissues highlighted the involvement of CERS5 in structural and functional brain abnormalities in PCOS. According to the enriched functional pathways of CERS5‐related gene network, we inferred that CERS5 may modulate sphingolipid biosynthetic process and ceramide metabolic process in PCOS. Subsequent analyses demonstrated notable interactions of CERS5 gene with various chemicals, including bisphenol A, Benzo(a)pyrene, Tetrachlorodibenzodioxin, titanium dioxide, Acetaminophen, Aflatoxin B1, which were documented to impact reproductive health [53], NAFLD [54], and the immune system [55]. Han et al. identified DNM3 as a candidate shared gene between endometriosis and PCOS, and proposed that DNM3 may facilitate the proliferation and invasion of endometrial cells by regulating proteins involved in cell cycle or apoptosis [29]. In this study, MR analysis revealed that DNM3 expression was significantly correlated with PCOS, with different directions in five tissues, indicating the underlying tissue‐specific biological processes of DNM3. The negative associations and strong colocalization evidence of DNM3 were observed in two brain structural regions, further suggesting that DNM3 may contribute to structural and functional brain abnormalities in PCOS. The LIMA1 gene plays an important role in cell growth, cholesterol absorption, and mitochondrial function [56]. Downregulation of LIMA1 is associated with poor prognosis in patients with ovarian cancer, breast cancer, and prostate cancer [56]. MR analysis indicated that LIMA1 exhibited positive associations and strong colocalization evidence with PCOS in two esophageal tissues. Given the conditions of disorder eating in PCOS, we inferred that LIMA1 may be implicated in eating‐related phenotypes. Moreover, women with PCOS frequently exhibit disturbances in lipid metabolism [57]. LIMA1 is likely to influence PCOS through its role in cholesterol metabolism. Recent evidence indicates that MST1R is involved in promoting tumor progression and metastasis. Elevated MST1R expression has been detected in clinical specimens of colon, breast, ovarian, and lung cancers [58]. Studies also show that MST1R plays roles in both tumor progression and the development of multiple immune diseases as well as chronic inflammation [59]. MST1R may act as a new inflammatory regulator that affects systemic metabolism [59]. MR results indicated that MST1R expression was significantly correlated with PCOS in eight tissues. Notably, the correlation in whole blood exhibited the opposite direction compared with other tissues and strong colocalization evidence, yielding new insights into the biological function of MST1R. PCOS patients exhibit dysregulated immune responses across multiple organ systems, and systemic low‐grade chronic inflammation plays an important role in the pathogenesis of PCOS [60]. The inverse association together with robust genetic colocalization evidence observed in whole blood implied that MST1R may reduce PCOS risk by regulating systemic low‐grade chronic inflammation, exerting systemic effects through the circulatory system and thereby alleviating secondary multiorgan dysfunction. Early research indicated that the variation of NR1H3 rs11039155 could be associated with a risk of PCOS among Chinese Han individuals [30]. This study utilized MR to establish that NR1H3 expression was associated with PCOS in two tissues, with strong colocalization evidence in the testis. According to current research, reproductive and metabolic abnormalities show a statistically significant clustering pattern among first‐degree relatives of women with PCOS [4]. The genetic predisposition to PCOS in adulthood has a fetal origin, which may be mediated by the effects of PCOS candidate genes during the development of multiple organs, including the testes [61]. Our findings on NR1H3 enrich the understanding of the role of NR1H3 in PCOS. Nevertheless, further studies are needed to clarify the precise mechanism of NR1H3. Notably, we did not observe strong genetically predicted expression signals in ovary‐ or uterus‐related tissues for several prioritized genes. This finding should be interpreted cautiously rather than as evidence against their relevance to PCOS. PCOS is a complex and heterogeneous disorder with reproductive, metabolic, endocrine, and inflammatory features. Its genetic susceptibility may therefore be mediated through multiple tissues. Moreover, the GTEx V8 reference panel used in this study includes bulk ovary and uterus tissues but does not provide an endometrium‐specific tissue. Furthermore, as a nondiseased bulk‐tissue resource, it may not adequately capture cell‐type‐specific or disease‐specific regulatory effects in reproductive tissues. The findings of PheWAS analysis lent further support to the view that PCOS is a multisystem disorder, and that several prioritized candidate genes may contribute to both PCOS susceptibility and a spectrum of PCOS‐related traits including metabolic and cardiovascular phenotypes. The GeneMANIA analysis suggested that the prioritized genes do not all belong to a single canonical pathway, but instead cluster into several functional modules that may each contribute to PCOS susceptibility. Notably, the full 15‐gene network showed enrichment in processes related to DNA replication, DNA repair, chromosomal organization, and cell cycle regulation, indicating that a subset of candidate genes may converge on pathways involved in genome maintenance and proliferative control. These may be relevant to PCOS, given that tightly regulated cell proliferation and differentiation in follicular development are critical for granulosa‐cell function and endometrial homeostasis. Meanwhile, the networks constructed from six genes with potential causal associations revealed further distinct functional profiles, including lipid/ceramide metabolism (CERS5), vesicle recycling and endocytosis (DNM3), intracellular protein transport (MST1R), and transcription‐related regulation (NR1H3). Moreover, MCM6 and LIMA1 display prominent functional overlap within replication‐ and replisome‐associated processes, indicating that several prioritized genes converge on pathways related to DNA replication and cell cycle dynamics. Overall, these findings support the view that PCOS is a biologically heterogeneous disorder involving multiple molecular processes rather than disruption of a single pathway. The chemical interaction analysis complemented the GeneMANIA analysis results from a different perspective by suggesting that some prioritized genes may lie at the interface between genetic susceptibility and environmental or metabolic exposures relevant to PCOS. Environmental factors serve as one of the key pathogenic factors for PCOS [62]. Recent studies suggested that exposure to toxicants such as bisphenol A and other endocrine‐disrupting chemicals during prenatal development may underlie the pathogenesis of PCOS [62]. Several identified chemicals, including bisphenol A, Benzo(a)pyrene, Tetrachlorodibenzodioxin, Cadmium Chloride, Particulate Matter, Ozone, and air pollutants, have previously been associated with endocrine disruption [63, 64], oxidative stress [65], or ovarian dysfunction [66, 67], all of which were biologically relevant to PCOS. From this perspective, the CTD results suggested that the prioritized genes may represent potential molecular nodes through which environmental or metabolic exposures influence PCOS‐related biology. The chemical interaction pattern of NR1H3 was particularly interesting because it was enriched in lipid‐related molecules and modulators, including fatty acids, dietary fats, cholesterol, fenofibrate, and T0901317. Given the established functions of NR1H3 in cholesterol transport, lipid metabolism, and inflammatory regulation, this finding supported a plausible link between NR1H3 and the metabolic component of PCOS. Likewise, the association of MST1R with lipopolysaccharides suggested a possible connection with inflammatory signaling pathways. These findings suggested that the prioritized genes may lie at the intersection of genetic susceptibility and environmental/metabolic factors in PCOS. In this study, we have identified a set of candidate genes associated with susceptibility to PCOS. This study was distinguished by its multipronged framework for prioritizing candidate genes. The application of multiple computational strategies, encompassing TWAS analyses, causal inference, gene network validation, and gene‐chemical‐disease analysis, has significantly enhanced the reliability of the candidate set in this study and enabled us to postulate their potential functions. These findings fostered advances in early diagnosis, risk evaluation, and the identification of novel therapeutic targets. However, certain limitations of this study must be acknowledged. The genetic analyses were limited to individuals of European ancestry, which restricted the generalizability of our results to other ethnic populations. Besides, an important limitation of the present study was the lack of direct replication of the prioritized genes in the independent validation dataset. Although we performed external validation using the same analytical framework, the genes identified using UTMOST and FUSION in the validation datasets were not consistent with those from the discovery stage, and MAGMA yielded no significant result. This discrepancy may be attributable to several factors. First, the validation dataset may have had insufficient statistical power, especially for detecting modest gene‐level or transcriptome‐mediated effects. Second, differences in PCOS diagnostic criteria, cohort recruitment strategies, phenotype definitions, and population structure across datasets may have introduced heterogeneity, thereby affecting GWAS signals and downstream gene prioritization. Third, UTMOST, FUSION, and MAGMA relied on distinct statistical assumptions and were influenced by the underlying GWAS architecture, LD patterns, and transcriptomic reference models, which may reduce reproducibility across datasets when effect sizes are small. Therefore, our findings should be interpreted with caution. Rather than constituting definitive validated susceptibility genes for PCOS, the genes prioritized in the discovery stage should be regarded as candidate genes that provide useful leads for future functional studies and further validation in larger, better‐powered, and more phenotypically homogeneous cohorts. Further studies are warranted to perform a series of functional experiments to further corroborate the hypothesized pathophysiological mechanism underlying these observed associations. Nevertheless, our study provides novel insights into the pathophysiological mechanism of PCOS. 5. Conclusion In conclusion, we conducted an integrative computational analysis to identify 15 candidate genes for PCOS, with MCM6 being the top‐confidence finding. Six candidate genes (MCM6, CERS5, DNM3, LIMA1, MST1R, NR1H3) were shown to have potential causal associations with PCOS. These findings provide fresh insights into the genetic architecture of PCOS. Downstream functional analyses yielded a more comprehensive understanding of the potential roles of the candidate genes in PCOS. However, further studies are needed to clarify the potential biological functions of these prioritized signals. Author Contributions Ruiqi Guo: conceptualization, formal analysis, methodology, visualization, data curation, writing original draft. Dequan Liao: data curation, writing original draft. Jixian Huang: data curation, writing original draft. Bing Dong: conceptualization, supervision, writing review, and editing. Funding The authors have nothing to report. Ethics Statement The authors have nothing to report. Conflicts of Interest The authors declare no conflicts of interest. Supporting information Data Availability Statement Included in article.

- 1. Stener‐Victorin E., Teede H., Norman R. J., et al., “Polycystic Ovary Syndrome,” Nature Reviews Disease Primers 10, no. 1 (2024): 27, 10.1038/s41572-024-00511-3. [DOI] [PubMed] [Google Scholar] - 2. Dokras A., “Polycystic Ovary Syndrome in 2025‐Insights and Innovations,” Fertility and Sterility 124, no. 5 Pt 2 (2025): 907–909, 10.1016/j.fertnstert.2025.09.025. [DOI] [PubMed] [Google Scholar] - 3. Zhang Y., Chen Z. J., and Zhao H., “Polycystic Ovary Syndrome: A Metabolic Disorder With Therapeutic Opportunities,” Cell Metabolism 37, no. 10 (2025): 1932–1949, 10.1016/j.cmet.2025.08.002. [DOI] [PubMed] [Google Scholar] - 4. Lafci N. G., Yilmaz B., and Yildiz B. O., “PCOS – The Many Faces of a Disorder in Women and Men,” Journal of Endocrinological Investigation 48, no. 4 (2025): 785–798, 10.1007/s40618-024-02512-1. [DOI] [PubMed] [Google Scholar] - 5. Louwers Y. V., Visser J. A., Dunaif A., and Laven J. S. E., “POLYCYSTIC OVARY SYNDROME: ORIGINS AND IMPLICATIONS: Genetics of Polycystic Ovary Syndrome (PCOS),” Reproduction 170, no. 5 (2025): e250126, 10.1530/rep-25-0126. [DOI] [PMC free article] [PubMed] [Google Scholar] - 6. Rosenfield R. L. and Ehrmann D. A., “The Pathogenesis of Polycystic Ovary Syndrome (PCOS): The Hypothesis of PCOS as Functional Ovarian Hyperandrogenism Revisited,” Endocrine Reviews 37, no. 5 (2016): 467–520, 10.1210/er.2015-1104. [DOI] [PMC free article] [PubMed] [Google Scholar] - 7. Zhao H., Xu Y., Xue B., et al., “Multi‐Ancestry Genome‐Wide Association Analyses of Polycystic Ovary Syndrome,” Nature Genetics 57, no. 11 (2025): 2669–2681, 10.1038/s41588-025-02393-x. [DOI] [PubMed] [Google Scholar] - 8. Ji E., Wang Z., Zhong R., et al., “Genomic Correlations, Shared Loci, and Drug Targets Between Polycystic Ovary Syndrome and Asthma: Insights From Genome‐Wide Association Analysis,” Journal of Clinical Endocrinology and Metabolism 110, no. 9 (2025): 2698–2707, 10.1210/clinem/dgaf276. [DOI] [PMC free article] [PubMed] [Google Scholar] - 9. Welt C. K., “Genetics of Polycystic Ovary Syndrome: What Is New?,” Endocrinology and Metabolism Clinics of North America 50, no. 1 (2021): 71–82, 10.1016/j.ecl.2020.10.006. [DOI] [PMC free article] [PubMed] [Google Scholar] - 10. Tam V., Patel N., Turcotte M., Bossé Y., Paré G., and Meyre D., “Benefits and Limitations of Genome‐Wide Association Studies,” Nature Reviews. Genetics 20, no. 8 (2019): 467–484, 10.1038/s41576-019-0127-1. [DOI] [PubMed] [Google Scholar] - 11. Gusev A., Ko A., Shi H., et al., “Integrative Approaches for Large‐Scale Transcriptome‐Wide Association Studies,” Nature Genetics 48, no. 3 (2016): 245–252, 10.1038/ng.3506. [DOI] [PMC free article] [PubMed] [Google Scholar] - 12. Hu Y., Li M., Lu Q., et al., “A Statistical Framework for Cross‐Tissue Transcriptome‐Wide Association Analysis,” Nature Genetics 51, no. 3 (2019): 568–576, 10.1038/s41588-019-0345-7. [DOI] [PMC free article] [PubMed] [Google Scholar] - 13. Gui J., Yang X., Tan C., et al., “A Cross‐Tissue Transcriptome‐Wide Association Study Reveals Novel Susceptibility Genes for Migraine,” Journal of Headache and Pain 25, no. 1 (2024): 94, 10.1186/s10194-024-01802-6. [DOI] [PMC free article] [PubMed] [Google Scholar] - 14. Fu L., Han X., Wang Y., and Hu Y. Q., “Identifying Genetic Overlaps in Obesity and Metabolic Disorders Unlocking Unique and Shared Mechanistic Insights,” Free Radical Biology & Medicine 231 (2025): 80–93, 10.1016/j.freeradbiomed.2025.02.033. [DOI] [PubMed] [Google Scholar] - 15. Ni J., Wang P., Yin K. J., et al., “Novel Insight Into the Aetiology of Rheumatoid Arthritis Gained by a Cross‐Tissue Transcriptome‐Wide Association Study,” RMD Open 8, no. 2 (2022): e002529, 10.1136/rmdopen-2022-002529. [DOI] [PMC free article] [PubMed] [Google Scholar] - 16. Day F., Karaderi T., Jones M. R., et al., “Large‐Scale Genome‐Wide Meta‐Analysis of Polycystic Ovary Syndrome Suggests Shared Genetic Architecture for Different Diagnosis Criteria,” PLoS Genetics 14, no. 12 (2018): e1007813, 10.1371/journal.pgen.1007813. [DOI] [PMC free article] [PubMed] [Google Scholar] - 17. GTEx Consortium , “The Genotype‐Tissue Expression (GTEx) project,” Nature Genetics 45, no. 6 (2013): 580–585, 10.1038/ng.2653. [DOI] [PMC free article] [PubMed] [Google Scholar] - 18. Sun R., Hui S., Bader G. D., Lin X., and Kraft P., “Powerful Gene Set Analysis in GWAS With the Generalized Berk‐Jones Statistic,” PLoS Genetics 15, no. 3 (2019): e1007530, 10.1371/journal.pgen.1007530. [DOI] [PMC free article] [PubMed] [Google Scholar] - 19. de Leeuw C. A., Mooij J. M., Heskes T., and Posthuma D., “MAGMA: Generalized Gene‐Set Analysis of GWAS Data,” PLoS Computational Biology 11, no. 4 (2015): e1004219, 10.1371/journal.pcbi.1004219. [DOI] [PMC free article] [PubMed] [Google Scholar] - 20. de Leeuw C. A., Neale B. M., Heskes T., and Posthuma D., “The Statistical Properties of Gene‐Set Analysis,” Nature Reviews. Genetics 17, no. 6 (2016): 353–364, 10.1038/nrg.2016.29. [DOI] [PubMed] [Google Scholar] - 21. Hemani G., Zheng J., Elsworth B., et al., “The MR‐Base Platform Supports Systematic Causal Inference Across the Human Phenome,” eLife 7 (2018): e34408, 10.7554/eLife.34408. [DOI] [PMC free article] [PubMed] [Google Scholar] - 22. Wang Q., Dhindsa R. S., Carss K., et al., “Rare Variant Contribution to Human Disease in 281,104 UK Biobank Exomes,” Nature 597, no. 7877 (2021): 527–532, 10.1038/s41586-021-03855-y. [DOI] [PMC free article] [PubMed] [Google Scholar] - 23. Xu D., Jia D., Fang X., et al., “Integrative Multi‐Omics Analysis of Druggable Genes for Therapeutic Target Identification in Polycystic Ovary Syndrome,” Journal of Ovarian Research 18, no. 1 (2025): 293, 10.1186/s13048-025-01889-8. [DOI] [PMC free article] [PubMed] [Google Scholar] - 24. Lyle S. M., Ahmed S., Elliott J. E., Stener‐Victorin E., Nachtigal M. W., and Drögemöller B. I., “Transcriptome‐Wide Association Analyses Identify an Association Between ARL14EP and Polycystic Ovary Syndrome,” Journal of Human Genetics 68, no. 5 (2023): 347–353, 10.1038/s10038-023-01120-w. [DOI] [PubMed] [Google Scholar] - 25. GTEx Consortium , “The GTEx Consortium Atlas of Genetic Regulatory Effects Across Human Tissues,” Science 369, no. 6509 (2020): 1318–1330, 10.1126/science.aaz1776. [DOI] [PMC free article] [PubMed] [Google Scholar] - 26. Xiao Z., Zhang X., Zhang Y., et al., “Advancing PCOS Drug Development: Multi‐Omics Discovery of Key Targets and Repurposable Compounds,” Naunyn‐Schmiedeberg's Archives of Pharmacology (2026), 10.1007/s00210-026-05473-9. [DOI] [PubMed] [Google Scholar] - 27. Batista M. C., Duarte E. F., Borba M. D., et al., “Trp28Arg/Ile35Thr LHB Gene Variants Are Associated With Elevated Testosterone Levels in Women With Polycystic Ovary Syndrome,” Gene 550, no. 1 (2014): 68–73, 10.1016/j.gene.2014.08.017. [DOI] [PubMed] [Google Scholar] - 28. Lerchbaum E., Giuliani A., Gruber H. J., Pieber T. R., and Obermayer‐Pietsch B., “Adult‐Type Hypolactasia and Calcium Intake in Polycystic Ovary Syndrome,” Clinical Endocrinology 77, no. 6 (2012): 834–843, 10.1111/j.1365-2265.2012.04334.x. [DOI] [PubMed] [Google Scholar] - 29. Tan H., Long P., and Xiao H., “Dissecting the Shared Genetic Architecture Between Endometriosis and Polycystic Ovary Syndrome,” Frontiers in Endocrinology 15 (2024): 1359236, 10.3389/fendo.2024.1359236. [DOI] [PMC free article] [PubMed] [Google Scholar] - 30. Zhao J., Hu Z., Cai L., et al., “Association Between Single Nucleotide Polymorphisms of Sterol Regulatory Element Binding Protein‐2 and Liver X Receptor α Gene and Risk of Polycystic Ovary Syndrome in a Chinese Han Population,” Cell Biochemistry and Biophysics 70, no. 2 (2014): 1421–1426, 10.1007/s12013-014-0075-5. [DOI] [PubMed] [Google Scholar] - 31. Ahn B., Kim E. J., Kim J. M., et al., “Co‐Expression of TWIK‐Related Acid‐Sensitive K(+) Channel 1 (TASK‐1/KCNK3) and Platelet‐Derived Growth Factor Receptor Alpha (PDGFRα/Pdgfra) in Adult Mouse Ovary,” Biomedicine 13, no. 8 (2025): 1941, 10.3390/biomedicines13081941. [DOI] [PMC free article] [PubMed] [Google Scholar] - 32. Daubney E. R., Flatley C., Hwang L. D., and Evans D. M., “Proteome‐Wide Mendelian Randomisation Study of Adverse Perinatal Outcomes,” Behavior Genetics 55, no. 5 (2025): 360–370, 10.1007/s10519-025-10233-1. [DOI] [PMC free article] [PubMed] [Google Scholar] - 33. Ke H., Tang S., Guo T., et al., “Landscape of Pathogenic Mutations in Premature Ovarian Insufficiency,” Nature Medicine 29, no. 2 (2023): 483–492, 10.1038/s41591-022-02194-3. [DOI] [PMC free article] [PubMed] [Google Scholar] - 34. Forsburg S. L., “Eukaryotic MCM Proteins: Beyond Replication Initiation,” Microbiology and Molecular Biology Reviews 68, no. 1 (2004): 109–131, 10.1128/mmbr.68.1.109-131.2004. [DOI] [PMC free article] [PubMed] [Google Scholar] - 35. Zeng T., Guan Y., Li Y. K., et al., “The DNA Replication Regulator MCM6: An Emerging Cancer Biomarker and Target,” Clinica Chimica Acta 517 (2021): 92–98, 10.1016/j.cca.2021.02.005. [DOI] [PubMed] [Google Scholar] - 36. Zhou J., Liu F., Tian L., et al., “Mutational Analysis of Minichromosome Maintenance Complex Component (MCM) Family Genes in Chinese Han Women With Polycystic Ovarian Syndrome,” Gynecological Endocrinology 39, no. 1 (2023): 2206912, 10.1080/09513590.2023.2206912. [DOI] [PubMed] [Google Scholar] - 37. Ahmad R., Frishman W. H., and Aronow W. S., “Polycystic Ovary Syndrome and Cardiovascular Disease: Pathophysiology, Clinical Evidence, and Preventive Strategies,” Cardiology in Review (2025), 10.1097/crd.0000000000001054. [DOI] [PubMed] [Google Scholar] - 38. Afiqah‐Aleng N., Harun S., AR M. R. A., Nor Muhammad N. A., and Mohamed‐Hussein Z. A., “PCOSBase: A Manually Curated Database of Polycystic Ovarian Syndrome,” Database: The Journal of Biological Databases and Curation 2017 (2017): bax098, 10.1093/database/bax098. [DOI] [PMC free article] [PubMed] [Google Scholar] - 39. Kim J. Y., Song H., Kim H., et al., “Transcriptional Profiling With a Pathway‐Oriented Analysis Identifies Dysregulated Molecular Phenotypes in the Endometrium of Patients With Polycystic Ovary Syndrome,” Journal of Clinical Endocrinology and Metabolism 94, no. 4 (2009): 1416–1426, 10.1210/jc.2008-1612. [DOI] [PMC free article] [PubMed] [Google Scholar] - 40. Pant P., Sircar R., Prasad R., Prasad H. O., and Chitme H. R., “Protein Expression and Bioinformatics Study of Granulosa Cells of Polycystic Ovary Syndrome Expressed Under the Influence of DHEA,” Clinical Medicine Insights: Endocrinology and Diabetes 16 (2023): 11795514231206732, 10.1177/11795514231206732. [DOI] [PMC free article] [PubMed] [Google Scholar] - 41. Yu S., Wang G., Shi Y., Xu H., Zheng Y., and Chen Y., “MCMs in Cancer: Prognostic Potential and Mechanisms,” Analytical Cellular Pathology (Amsterdam) 2020 (2020): 3750294, 10.1155/2020/3750294. [DOI] [PMC free article] [PubMed] [Google Scholar] - 42. Azziz R., Carmina E., Chen Z., et al., “Polycystic Ovary Syndrome,” Nature Reviews. Disease Primers 2 (2016): 16057, 10.1038/nrdp.2016.57. [DOI] [PubMed] [Google Scholar] - 43. Wang Y., Chen H., Zhang J., et al., “MCM Family in Gastrointestinal Cancer and Other Malignancies: From Functional Characterization to Clinical Implication,” Biochimica Et Biophysica Acta. Reviews on Cancer 1874, no. 2 (2020): 188415, 10.1016/j.bbcan.2020.188415. [DOI] [PubMed] [Google Scholar] - 44. Guney G., Taşkın M. I., Sener N., et al., “The Role of ERK‐1 and ERK‐2 Gene Polymorphisms in PCOS Pathogenesis,” Reproductive Biology and Endocrinology 20, no. 1 (2022): 95, 10.1186/s12958-022-00967-6. [DOI] [PMC free article] [PubMed] [Google Scholar] - 45. Kisiała M., Mazepa W., Bauer D., et al., “Molecular Mechanisms of Insulin Resistance and Altered Carbohydrate Metabolism in PCOS: A Scoping Review,” Front Endocrinol (Lausanne) 17 (2026): 1810805, 10.3389/fendo.2026.1810805. [DOI] [PMC free article] [PubMed] [Google Scholar] - 46. Wang L. and Liu X., “Pan‐Cancer Multi‐Omics Analysis of Minichromosome Maintenance Proteins (MCMs) Expression in Human Cancers,” Frontiers in Bioscience (Landmark Edition) 28, no. 9 (2023): 230, 10.31083/j.fbl2809230. [DOI] [PubMed] [Google Scholar] - 47. Morais G. P. D. E., Thé A. P., Dias Nogueira B. M., et al., “Frequency and Diagnostic Utility of the ‐13910C>T MCM6 Gene Polymorphism for Lactose Intolerance in a Brazilian Northeast Population,” In Vivo 40, no. 1 (2026): 178–190, 10.21873/invivo.14183. [DOI] [PMC free article] [PubMed] [Google Scholar] - 48. Tidhar R., Ben‐Dor S., Wang E., Kelly S., A. H. Merrill, Jr. , and Futerman A. H., “Acyl Chain Specificity of Ceramide Synthases Is Determined Within a Region of 150 Residues in the Tram‐Lag‐CLN8 (TLC) Domain,” Journal of Biological Chemistry 287, no. 5 (2012): 3197–3206, 10.1074/jbc.M111.280271. [DOI] [PMC free article] [PubMed] [Google Scholar] - 49. Sharma A., Krick B., Li Y., Summers S. A., Playdon M. C., and Welt C., “The Use of Ceramides to Predict Metabolic Response to Metformin in Women With PCOS,” Journal of the Endocrine Society 6, no. 11 (2022): bvac131, 10.1210/jendso/bvac131. [DOI] [PMC free article] [PubMed] [Google Scholar] - 50. Zhang S., Wang Y., Yuan L., Hu C., Xu Z., and Cheng X., “High Expression of Ceramide Synthase 5 Predicts a Poor Prognosis in Gastric Cancer,” Translational Cancer Research 11, no. 9 (2022): 3209–3221, 10.21037/tcr-22-1348. [DOI] [PMC free article] [PubMed] [Google Scholar] - 51. Mootha V. K., Lindgren C. M., Eriksson K. F., et al., “PGC‐1alpha‐Responsive Genes Involved in Oxidative Phosphorylation Are Coordinately Downregulated in Human Diabetes,” Nature Genetics 34, no. 3 (2003): 267–273, 10.1038/ng1180. [DOI] [PubMed] [Google Scholar] - 52. Jiang N., Deng J., Bao C., et al., “Brain Alteration of PCOS: Neuroimaging and Function,” Frontiers in Neurology 17 (2026): 1825907, 10.3389/fneur.2026.1825907. [DOI] [PMC free article] [PubMed] [Google Scholar] - 53. Hwang S., Kang H. B., Kim D. H., and Park M. H., “Epigenetic Mechanisms of Reproductive Dysfunction Induced by Endocrine‐Disrupting Chemicals: Evidence From Molecular Studies,” Frontiers in Bioscience (Landmark Edition) 31, no. 1 (2026): 42777, 10.31083/fbl42777. [DOI] [PubMed] [Google Scholar] - 54. Li Y., Liang N., Tang T., et al., “Low‐Dose Benzo[a]Pyrene Exposure Induces Hepatic Lipid Deposition Through LCMT1/PP2Ac‐Mediated Autophagy Inhibition,” Food and Chemical Toxicology 179 (2023): 113986, 10.1016/j.fct.2023.113986. [DOI] [PubMed] [Google Scholar] - 55. Chałaśkiewicz K., Kępka‐Borkowska K., Starzyński R. R., et al., “Impact of Aflatoxins on the Digestive, Immune, and Nervous Systems: The Role of Microbiota and Probiotics in Toxicity Protection,” International Journal of Molecular Sciences 26, no. 17 (2025): 8258, 10.3390/ijms26178258. [DOI] [PMC free article] [PubMed] [Google Scholar] - 56. Lindell E. and Zhang X., “Exploring the Enigma: The Role of the Epithelial Protein Lost in Neoplasm in Normal Physiology and Cancer Pathogenesis,” International Journal of Molecular Sciences 25, no. 9 (2024): 4970, 10.3390/ijms25094970. [DOI] [PMC free article] [PubMed] [Google Scholar] - 57. Mishra N., Mishra P., Kapoor V., Kishun J., Kumar A., and Singh U., “Lipid Variations in Different Polycystic Ovary Syndrome Phenotypes: A Systematic Review & Meta‐Analysis,” Indian Journal of Medical Research 161, no. 5 (2025): 491–501, 10.25259/ijmr_1455_2024. [DOI] [PMC free article] [PubMed] [Google Scholar] - 58. Camp E. R., Liu W., Fan F., Yang A., Somcio R., and Ellis L. M., “RON, a Tyrosine Kinase Receptor Involved in Tumor Progression and Metastasis,” Annals of Surgical Oncology 12, no. 4 (2005): 273–281, 10.1245/aso.2005.08.013. [DOI] [PubMed] [Google Scholar] - 59. Huang L., Fang X., Shi D., et al., “MSP‐RON Pathway: Potential Regulator of Inflammation and Innate Immunity,” Frontiers in Immunology 11 (2020): 569082, 10.3389/fimmu.2020.569082. [DOI] [PMC free article] [PubMed] [Google Scholar] - 60. Deng H., Chen Y., Xing J., Zhang N., and Xu L., “Systematic Low‐Grade Chronic Inflammation and Intrinsic Mechanisms in Polycystic Ovary Syndrome,” Frontiers in Immunology 15 (2024): 1470283, 10.3389/fimmu.2024.1470283. [DOI] [PMC free article] [PubMed] [Google Scholar] - 61. Azumah R., Hummitzsch K., Anderson R. A., and Rodgers R. J., “Genes in Loci Genetically Associated With Polycystic Ovary Syndrome Are Dynamically Expressed in Human Fetal Gonadal, Metabolic and Brain Tissues,” Front Endocrinol (Lausanne) 14 (2023): 1149473, 10.3389/fendo.2023.1149473. [DOI] [PMC free article] [PubMed] [Google Scholar] - 62. Siddiqui S., Mateen S., Ahmad R., and Moin S., “A Brief Insight Into the Etiology, Genetics, and Immunology of Polycystic Ovarian Syndrome (PCOS),” Journal of Assisted Reproduction and Genetics 39, no. 11 (2022): 2439–2473, 10.1007/s10815-022-02625-7. [DOI] [PMC free article] [PubMed] [Google Scholar] - 63. Hong Y., Du Z., Li J., et al., “Integrative Causal Inference Illuminates Gene‐Environment Interactions Linking Endocrine Disruptors to Female Infertility,” Ecotoxicology and Environmental Safety 302 (2025): 118679, 10.1016/j.ecoenv.2025.118679. [DOI] [PubMed] [Google Scholar] - 64. Kennedy C. J. and Smyth K. R., “Disruption of the Rainbow Trout Reproductive Endocrine Axis by the Polycyclic Aromatic Hydrocarbon Benzo[a]Pyrene,” General and Comparative Endocrinology 219 (2015): 102–111, 10.1016/j.ygcen.2015.03.013. [DOI] [PubMed] [Google Scholar] - 65. Eko O. W., Eteng M. U., Iwara I. A., et al., “Modulatory Effects of Lannea Welwitschia on Cadmium‐Induced Reproductive and Biochemical Dysregulation via Steroidogenic Enzyme Regulation in Female Rats,” Journal of Ethnopharmacology 357 (2026): 120821, 10.1016/j.jep.2025.120821. [DOI] [PubMed] [Google Scholar] - 66. Chen Y., Zhang J., Zhang T., et al., “Circulating Interleukin‐6 Mediates PM(2.5)‐Induced Ovarian Injury by Suppressing the PPARγ Pathway,” Research (Wash D C) 7 (2024): 0538, 10.34133/research.0538. [DOI] [PMC free article] [PubMed] [Google Scholar] - 67. Liang X., Wu X., Tian Y., et al., “Ozone Exposure at Environmental Level Induces Female Reproductive Impairment via Transcriptomic and Alternative Analysis,” Ecotoxicology and Environmental Safety 306 (2025): 119276, 10.1016/j.ecoenv.2025.119276. [DOI] [PubMed] [Google Scholar] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials Data Availability Statement Included in article.