Association Between Primary Dysmenorrhea and Mental Health Traits: A Study Based on Multi-Phenotype Correlation Network and Mendelian Randomization Analysis in Female College Students

Primary dysmenorrhea (PDM) is a common cyclic menstrual pain that significantly affects the quality of life for women. Several epidemiological studies have suggested a potential association between PDM and mental health traits, including stress, depression, and anxiety. However, there is a lack of systematic investigation into whether a causal relationship exists between PDM and mental health phenotypes compared to other physical phenotypes. In this study, we conducted a large-scale phenome study on a cohort of 7401 young female Chinese college students to explore the association between PDM and various physical and mental health phenotypes. Using a multi-phenotype correlation network model, we discovered that the correlation between the PDM phenotypes and mental health phenotypes was the most dominant among the complex inter-connections across different categories of phenotypes. Furthermore, employing a two-sample Mendelian randomization analysis, we systematically elucidated the genomic-level impact of PDM on the mental health traits of young women. Specifically, we identified an increased risk of depression and anxiety associated with PDM, potentially influenced by several Single-nucleotide polymorphism (SNP) variants such as ZMIZ1, DIO1, GRIK4 and RBFOX1. This study offers valuable insights into the genetic mechanism through which dysmenorrhea impacts mental health, which contributes to a better understanding of the comprehensive management of PDM and its associated psychological challenges. Supplementary Information The online version contains supplementary material available at 10.1007/s43657-024-00213-6.

Primary dysmenorrhea, Mental health traits, Female college students, Multi-phenotype correlation network, Two-sample mendelian randomization

Primary dysmenorrhea (PDM) is a common gynecological condition in women characterized by severe chronic pelvic pain and menstrual cramps, despite normal pelvic anatomy. Studies conducted on a population-based level have revealed that 93% of adolescents experience pain during menstruation. Dysmenorrhea significantly impacts the women's daily lives, leading to restricted activities, school or work absences (Klein and Litt 1981), and an increased risk of postpartum depression (Meng et al. 2019). The most common symptoms of dysmenorrhea reported were irritability, abdominal distension sensation, anxiety and feeling more emotional (Barbosa-Silva et al. 2024). The pathophysiology is closely linked to prostaglandins and leukotrienes, both mediators of inflammation. Patients with primary dysmenorrhea often exhibit a state of micro-inflammation within their bodies (Proctor and Farquhar 2006). Chronic inflammation is also an important factor contributing to the development or exacerbation of certain mental health conditions (Han et al. 2021). Neural cross-talk in the dorsal root ganglia, spinal cord and brain levels might play a role in the development of chronic pain disorders including dysmenorrhea through central sensitization (Furuta et al. 2012). Consequently, it is important to explore the relationship between primary dysmenorrhea, chronic inflammation, and mental health conditions, which is far from fully understood. On the other hand, dysmenorrhea severity is associated with menstrual duration, smoking, obesity, and other phenotypes, and typically improves after delivery (Weissman et al. 2004). Therefore, it is crucial to pay closer attention to the more serious symptoms of dysmenorrhea in young women population. Meanwhile, the mental health of young people, particularly college students, is also a significant concern. Studies have reported high prevalence rates of depression among college students, with rates of 35.2% in Macao, 41% in Hong Kong, and 16.8% in mainland China (Li et al. 2020). Therefore, it is valuable to conduct in-depth research on the relationship between PDM and mental health and the underlying genetic basis. Previous studies have suggested a potential association between PDM and some mental health traits such as depression, anxiety (Dorn et al. 2009; Gagua et al. 2013), and stress (Wang et al. 2004). For instance, Gagua et al. (2013) found that adolescent women with a history of dysmenorrhea (276 samples) were more likely to experience depression or have higher depression scores compared to girls without menstrual pain (148 samples). Kabukçu et al. (2021) reported that 104 adolescents with dysmenorrhea had worse sleep quality, more attention and hyperactivity impulsivity problems, and other psychological symptoms such as anxiety, depression, somatization, negative self-perception, and hostility, compared to 105 adolescents without dysmenorrhea. A review of dysmenorrhea and mental health found that females with dysmenorrhea reported greater levels of depressive and anxiety symptoms than females who do not experience dysmenorrhea (Bajalan et al. 2019). Moreover, a meta-analysis conducted by Pakpour et al. (2020) investigated the relationship between dysmenorrhea and more common mental health conditions. In addition, recent neuroimaging studies have identified abnormal brain function and frequency-specific brain alterations in PDM (Yu et al. 2022). Wu et al. (2021) found altered connectivity between the posterior cerebellar lobule and perigenual anterior cingulate cortex in PDM females. Abnormal brain metabolism, spontaneous activity, and functional network alterations in the default mode network and emotional network have also been observed in PDM females, contributing to higher-level sensory and attention processing related to pain and affect regulation (Farrow et al. 2021; Li et al. 2020, 2017; Meng et al. 2019; Wei et al. 2021; Weissman et al. 2004; Yu et al. 2022). However, the associations between dysmenorrhea and mental health were not quantitatively analyzed, and the identification of important dysmenorrhea variations associated with mental health traits from a genetic perspective remains limited. For a deeper understanding of the impact of dysmenorrhea on mental health, we conducted a large cohort study involving 7401 female college students (aged 17–24) to investigate the association between PDM and mental health traits. We compared the correlation results with other phenotypes by a comprehensive correlation network to establish the significance of this association. Furthermore, we employed Mendelian randomization to provide additional genetic evidence supporting the heightened risk of mental health problems associated with PDM. Figure 1 illustrates the objective, strategy, and analysis pipeline of our study, outlining the key steps and methodologies employed. Our findings contribute to a better understanding of the physiological causes of mental health problems in young women and offer guidance and suggestions for the treatment of psychological issues in this specific population.

Recruitment of the Female College Student Cohort The participants were female college students of East Asian descent, aged between 17 and 24 from Qingdao University in China during 2018–2020. Written informed consent form and mental health questionnaire were approved by the ethics committee with ethics approval number M16033. Prior to the initiation of the research, all eligible participants provided their signed written informed consent forms. We recruited total 8323 eligible participants and measured multi-phenotypes, finally got 7401 female students with complete information on physical measurements, menstruation, and mental health traits by comprehensive data cleaning. The cross-sectional study design follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist as demonstrated in supplementary statement. Measurement of Physical and Mental Health Phenotypes The collected phenotype data comprised physical phenotype data and mental health traits data. Physical phenotype data were collected through physical measurements and questionnaires, while mental health traits were acquired through scales and questionnaires. A total of 98 phenotypes were classified into five categories, including basic information, menstruation phenotypes, past history and family history, body circumference, and mental health traits. Detailed information about these phenotypes can be found in Table S1. To assess the dysmenorrhea phenotype, two scales were utilized. The Visual Analog Scale (VAS) was employed for self-assessment of the degree of lower abdominal pain experienced during menstruation, as shown in Fig. 2a. The scale consists of 10 degrees, ranging from 0 (no pain) to 10 (the most unbearable pain), commonly applied in clinical and epidemiological research (Vural et al. 2016). The Cox Menstrual Symptom Scale (CMSS) consists of 17 dysmenorrhea symptoms, such as abdominal pain, nausea, vomiting, fatigue, dizziness, loss of appetite, headache, back pain, leg pain, stomach pain, general pain, facial color change, insomnia, diarrhea, depression, irritability, and neuroticism (Cox and Meyer 1978). Each symptom is scored via five levels (0–4) in terms of the duration evaluation and severity evaluation respectively (Fig. 2b). Combining VAS and CMSS scales offers a robust assessment of dysmenorrhea symptoms. Among the participants, 3643 samples (49%) exhibited symptoms of dysmenorrhea. For the assessment of mental health traits, three scales were used: the Self-Rating Anxiety Scale (SAS) (Dunstan and Scott 2020), the Self-Rating Depression Scale (SDS) (Dunstan and Scott 2019; Dunstan et al. 2017), and the Personality Diagnostic Questionnaire (PDQ-4) (Chang et al. 2017; Kirchner et al. 2018). The SAS consists of 20 items, and the scores can be interpreted as normal, mild, moderate, or severe anxiety. The SDS also consists of 20 items, and the scores indicate normal, mild, moderate, or severe depression. The PDQ-4 assesses various personality disorders, including Paranoid (PAR), Schizoid (SCH), Schizotypal (SCHT), Histrionic (HIS), Narcissistic (NAR), Antisocial (ANT), Borderline (BOR), Avoidant (AVO), Dependent (DEP), and Obsessive–Compulsive (OBC). By utilizing these scales, we aimed to evaluate and measure mental health traits in the recruited female college students. Among all participants, 1015 individuals displayed some degree of depression (mild: 831, moderate: 153, severe: 31), while 531 subjects exhibited anxiety (mild: 463, moderate: 55, severe: 13). Multi-Phenotype Correlation Network Construction We constructed a multi-phenotype correlation network to explore the associations among various phenotypes of the recruited college students. Building upon the inter-omic correlation network method dealing with genome, proteome, and metabolome data (Price et al. 2017), we developed a network construction framework compatible with heterogeneous data types of diverse phenotypes. This approach comprises three key steps: 1) Data preprocessing: Each phenotype data was mean-centered and scaled by the standard deviation of the measured values. This step eliminated differences caused by varying data ranges and distributions across different categories, and transformed each data set into comparable data vectors for statistical analysis. Since all the participants are female college students aged between 17 and 24, there is no need for gender and age correction. The potential confounder is body mass index (BMI). Each phenotype measurement was BMI corrected if a trimmed mean robust regression identified a significant relationship (p < 0.05) with BMI. If the phenotype measurement was corrected, the residuals of the model were used in place of the original measurements (Price et al. 2017). If no corrections were made, the original mean-centered and scaled measurements were used. 2) Spearman correlation coefficient calculation: For each pairwise combination of phenotype variables (e.g., dysmenorrhea phenotypes versus family history, past history versus mental health phenotypes, etc.), the Spearman correlation coefficient (SCC) of the two vectors (7401 individuals in each vector) were calculated. where represents the difference between the ranks of each paired data point for both variables, denotes the summation of squared rank differences, and n is the total number of data points. 3) Significance assessment: We evaluated whether the SCC is significant by forming a multiple hypothesis test of p-values, the null hypothesis (H0) states that there is no correlation between the two variables of phenotypes, while the alternative hypothesis (H1) states that there is a correlation. Then, the Benjamini–Hochberg method (Benjamini and Hochberg 1995) was employed to adjust the individual p-values for multiple hypothesis testing, which helps control the false discovery rate (FDR). In order to ensure that FDR is less than α, it is necessary to sort the p-values obtained in the previous steps. For N hypothesis test results, we selected the top K of p-values so that all the selected p values can be less than a critical value (). We corrected the original p-values to obtain the adjusted p-values (padj), and ensure that the size order of the adjusted p-values did not change due to the correction. The correction formula is as follows. We chose an adjusted p-value (padj) cutoff of 0.05 as the significance level. By following these steps, we obtained the correlation network across multi-phenotypes, which was visualized by using ggraph package in R (Wickham 2016). Genetic Variants of PDM Genetic variation information of PDM was retrieved from another PDM cohort genome-wide association study (GWAS) study (Li et al. 2017). The study included 2404 PDM cases and 2920 controls of East Asian descent, discovered 16 significant variants of PDM (p < 1 × 10−5), including two significant single-nucleotide polymorphisms (SNPs) rs76518691 () and rs7523831 () identified by meta-analysis of an independent cohort of 1446 individuals (678 cases and 768 controls) (Li et al. 2017). In the current study, five SNPs rs60441301, rs663045, rs72741503, rs7523831, and rs11631919, were excluded as they were not present in the GWAS SNPs associated with various types of mental health traits from UK Biobank used in the Mendelian randomization analysis (Elsworth et al. 2020). Additionally, the SNP rs6509898 was removed due to its significant impact on the result of Leave-one-out sensitivity test. After these exclusions, a total of 10 significant variants were used for the two-sample mendelian randomization (MR) analysis. GWAS Dataset of Mental Health Traits for Two-Sample Mendelian Randomization Analyses For the two-sample MR analyses, the GWAS data for mental health traits were retrieved from the public psychological GWAS summary of East Asian populations. These data were published as part of the pan-ancestry genetic analysis project conducted by the UK Biobank at the Broad Institute in 2020, which is a part of the Integrative Epidemiology Unit (IEU) open GWAS project (Elsworth et al. 2020; Hemani et al. 2018). The dataset included 18 different types of mental health traits, and detailed information about these traits can be found in Table S2. Our research focused on exploring the association between PDM and specific mental health traits. Two-Sample Mendelian Randomization Analyses Since genotyping data for the same sample set measuring both PDM and mental health phenotypes is lacking, we used a two-sample MR analyses to assess the causal association between PDM and mental health traits. PDM was considered the exposure factor, and the 18 mental health traits served as the outcome factors. The inverse variance weighted (IVW) method was employed to estimate the effect of the SNPs. This method combines the individual SNP estimates and their corresponding standard errors to generate an overall estimate of the causal effect. represents the ratio estimate obtained from the jth SNP, is the inverse variance weight of the SNP. The IVW method does not take into account the direction of the intercept and utilizes the reciprocal of the squared standard error (se2) of the GWAS results as the weight for the inverse variance. The correlation was estimated by calculating 95% confidence intervals (CI) and odds ratios (OR). Using this method, we obtained the causal relationship between PDM and the mental health traits of interest. To further validate the statistical results, we also employed additional methods including MR-Egger, weighted median, and weighted mode simple mode (Bowden et al. 2016). These methods provided further analyses to confirm the findings. The Mendelian randomization analysis was conducted using the TwoSampleMR package in R. Functional Annotation of Key SNPs The functional annotation of key SNPs accounting for association between PDM and mental health traits were based on 3DSNP database and VannoPortal database. 3DSNP (Quan et al. 2022) is an integrated database (https://www.omic.tech/3dsnp/) for annotating human noncoding variants, which integrates 3D chromatin interactions, local chromatin characteristics in different cell types, and linkage disequilibrium (LD) information from the 1000 genomes project. The 3DSNP database utilizes several features, such as 3D interacting genes, enhancer state, promoter state, transcription factor binding sites, altered sequence motifs, and conservation scores, to assess the significance of SNP variations (Quan et al. 2022). The importance of an SNP is evaluated by fitting the hit numbers of all SNPs in each chromosome to a Poisson distribution model. The score of an SNP in a specific category is determined using the following formula: A SNP has k hits in a function class F, λ is the fitting parameter corresponding to Poisson model. For the conservation category (F'), the score of the SNP in the conservation category is defined as follows: VannoPortal (Huang et al. 2022) (http://www.mulinlab.org/vportal/) is a variant annotation database that comprehensively integrates genome-wide variant annotations and prediction scores from various biological domains, including allele frequency, linkage disequilibrium, evolutionary signature, disease/trait association, pathogenesis, and tissue/cell type-specific functional profile. We retrieved the information of regulatory score, expression quantitative trait locus (eQTL) and TFBS corresponding to the key SNPs.

The Multi-Phenotype Correlation Network Reveals Significant Correlation Between PDM and Mental Health Traits We constructed a correlation network based on pairwise Spearman correlations between different phenotypic data using the participants' phenotypes as vertices. We selected significant pairs of phenotypes (edges) (padj < 0.05) and obtained a network consisting of 100 vertices and 2694 edges. Among these edges, the majority (2008 edges) involved the menstruation phenotypes, while 798 edges were related to the mental health phenotypes. Within the network, there were 826 edges within the module of menstruation sub-phenotypes and 92 edges within the module of mental health sub-phenotypes. Notably, there were 536 edges connecting the menstruation and mental health modules. For the detailed connections in the correlation network, please refer to Table S3. To further refine the significant phenotype pairs across different categories, we filtered out pairs with higher internal correlation (SCC > 0.3) within the same category of phenotypes, and lower exterior correlation (SCC > 0.1) between different categories of phenotypes. This filtering process resulted in a correlation network across different phenotypes, containing 304 significant phenotype pairs, as visualized in Fig. 3. The correlation between dysmenorrhea phenotypes and mental health phenotypes is particularly prominent in the correlation network, compared to the other categories of nodes. Notably, the total score of the CMSS scale and VAS results for dysmenorrhea showed significant relationships with nearly all mental health phenotypes. Within the mental health phenotype nodes, depression, anxiety, BOR personality disorder, and AVO personality disorder demonstrated significant correlations with nearly all dysmenorrhea phenotypes. The degrees of all mental health phenotype nodes were equal to or greater than five, in contrast, there were very few connections linking other physical phenotype categories with the dysmenorrhea phenotype. Moreover, almost all dysmenorrhea sub-phenotypes demonstrated a significant correlation with depression and anxiety. In particular, the duration of symptoms including dysmenorrhea, depression, neuroticism, irritability, loss of appetite, and the severity of symptoms including dysmenorrhea, depression, irritability, nervousness, and anorexia during menstruation, all exhibited significant correlations with the mental health phenotypes assessed by SDS, SAS, and PDQ. Although there might be possible overlap between depression mood during menstruation and the depressive symptoms by SDS over a longer period, there are still 508 edges connecting the menstruation and mental health traits excluding 28 edges involving dysmenorrhea-depression, which account for majority of inter-module connections in the network. Additionally, there was a significant association between digestive tract diseases and dysmenorrhea. This association may arise from shared risk factors or predisposing factors for both conditions. For instance, hormonal imbalances, genetic factors, and certain lifestyle factors (such as diet, exercise, and stress) can contribute to both conditions. Moreover, stress can exacerbate symptoms of both digestive tract diseases and dysmenorrhea, potentially contributing to the co-occurrence in some individuals. To further investigate the specific correlation between dysmenorrhea and mental health phenotypes, we focused on the 3643 positive samples with PDM. Based on the total scores of the CMSS scale, we divided the samples into three groups with comparable sizes (Fig. 4a): mild (CMSS ≤ 20, 1423 samples), moderate (20 < CMSS < 35, 1151 samples), and severe (CMSS ≥ 35, 1069 samples). After standardization and centralization, we performed hierarchical clustering on the mental health traits data of the three sample groups with different degrees of dysmenorrhea (Fig. 4b). The analysis revealed a trend, as the intensity of dysmenorrhea pain increased, there was a corresponding increase in the scores related to mental health traits. This observation suggested a positive association between the overall severity of dysmenorrhea and the presence of mental health issues, with the severe dysmenorrhea group showing more pronounced distinctions compared to the mild and moderate groups. Specifically, depression and anxiety displayed the most significant variations among the different groups, followed by personality disorders of OBC, PAR, SCHT, NAR, which showed a higher likelihood of mental health issues in the severe dysmenorrhea group. Personality disorders of the AVO and DEP showed similar alteration between different groups. On the other hand, personality disorders of the HIS, BOR and ANT types demonstrated relatively less pronounced differences. These findings indicate a relationship between the intensity of dysmenorrhea and the occurrence of mental health issues, particularly in terms of depression, anxiety, and specific types of personality disorders. The more severe the dysmenorrhea, the greater the likelihood of experiencing these mental health challenges. Causal Relationship Between PDM and Mental Health Traits Revealed by a Two-Sample Mendelian Randomization Analysis Mendelian randomization analysis was utilized to investigate the causal relationship between PDM and mental health phenotypes. In light of the absence of genotyping data for the same sample set measuring both PDM and mental health traits, a two-sample Mendelian randomization approach was undertaken. In this analysis, PDM was considered the exposure variable, while the mental health traits from the UK Biobank dataset served as the outcome variables. We used 16 significant SNP variants associated with PDM discovered by a previous GWAS study on individuals of East Asian descent (Li et al. 2017) as instrumental variables. These SNPs were not directly associated with mental health traits and potential confounding factors such as BMI. We conducted screening to identify significant mental health traits influenced by PDM using the IVW approach. As shown in Fig. 5, The following mental health traits were found to be significant (pIVW < 0.05): nervous feelings (OR [95%CI] = 1.335 [1.05–1.70], pIVW = 0.0178), frequency of depressed mood in the last two weeks (OR [95%CI] = 1.097 [1.01–1.19], pIVW = 0.0200), mood swings (OR [95%CI] = 1.249 [1.01–1.54], pIVW = 0.0365), and frequency of tenseness/restlessness in the last three weeks (OR [95%CI] = 1.088 [1.00–1.18], pIVW = 0.0405). Notably, these depression and anxiety-related phenotypes also displayed significant correlations with the dysmenorrhea phenotype in the multi-phenotype correlated network model. To validate the significant mental health traits identified by the IVW method, we employed additional Mendelian randomization approaches, including MR Egger and weighted median methods, see results in Fig. 5. The outcomes derived from these diverse techniques strengthen the validity of our assertion that PDM notably influences the likelihood of correlated mental health traits. The results of Mendelian randomization for all 18 mental health traits from UK Biobank dataset were listed in Table S4. The forest plot represents the MR estimated value (95% CI) of the causal effect of dysmenorrhea on different mental health traits and the p-value of the results. We also calculated PDM Polygenetic risk score (PRS) for the female samples from GWAS dataset on mental health traits, and then evaluated the Pearson correlation coefficient between PRS and the mental health traits. For the four traits highlighted by MR analysis (restlessness, depression, mood swings and nervous feeling), we observed that PDM PRS has a positive correlation with them (Table S5), which supported that PDM could affect women's mental health state to some extent. The robustness of our two-sample Mendelian randomization findings was assessed through various tests. Firstly, we performed a leave-one-out sensitivity analysis, where each SNP was removed individually from the analysis. Results revealed that the exclusion of any individual SNP did not have a significant impact on our MR findings, enhancing the strength of our analysis (Fig. 6). The black line in the forest map indicates the impact on the results after removing a single SNP, and the red line indicates the overall impact. Next, we conducted a heterogeneity test, which indicated that our results passed the test for heterogeneity (p > 0.05). This suggested that there is no significant heterogeneity among the included SNPs in our MR analysis (Table S6). Furthermore, we conducted a pleiotropy test, specifically examining the Egger intercept. The results showed no statistically significant difference between zero and the Egger intercept (p > 0.05). This finding indicated that there is no evidence of horizontal pleiotropy in our Mendelian randomization experiment (Table S6). Additionally, we inspected the symmetry of the funnel plot to discern potential publication bias. The funnel plot showed overall symmetry, further endorsing the reliability of our Mendelian randomization analysis results (refer to Fig. S1). In summary, the leave-one-out sensitivity analysis, heterogeneity test, pleiotropy test, and funnel plot analysis all supported the robustness and validity of our Mendelian randomization findings. Significant Variants of PDM are Associated with Mental Health Traits The Mendelian randomization study has successfully established that PDM can significantly increase the risk of certain mental health problems through SNP variation. Building upon this, we delved into 10 significant variants obtained from the two-sample MR analysis to elaborate the impact of PDM on mental health traits from a molecular genetic perspective. The closest genes, 3D interactive genes, regulatory score, eQTL and overlapped transcript factor binding peaks measured by ChIP-seq linked to these key SNPs were detailed in Table 1. Of these annotated variants, four SNPs, including rs76518691, rs72829648, rs1475944, and rs2294512, exhibited high scores in terms of their strong interactions with 3D genes, enhancers, promoters, transcription factor binding sites, sequence motif changes, and conservation metrics. These findings suggest that these SNP loci are more likely positioned within promoter or enhancer elements, indicating a heightened potential for regulatory functionality in molecular processes. Particularly, rs1475944 has been identified as mQTL associated with metabolite X-21840 in urine, playing a role in the detoxification and excretion processes in the kidney, as highlighted in a study by Schlosser et al. (2020). On the other hand, rs2294512 has been identified as a mQTL linked to metabolite X-11546 in serum. This association has been instrumental in investigating genetic loci that influence human metabolic traits, as elucidated in research by Shin et al. (2014). These insights shed light on the multifaceted roles of these particular SNPs in biochemical pathways and metabolic functions. Table 1. | Variant ID | Position | Ref/Alt | Score | Linear closest gene | 3D interacting gene | eQTL | Number of eQTLs (± 10 M) | RegulatoryScore | TF Binding Site | |---|---|---|---|---|---|---|---|---|---| | rs76518691 | chr10: 80,918,766 | G/A | 78.46 | ZMIZ1 | ZMIZ1-AS1 | 0.87847 | 16 | || | rs72829648 | chr2: 112,842,807 | A/C | 9.44 | TMEM87B | BMERTK, FBLN7 | ZC3H8 RP11-399B17.1 RGPD8 FBLN7 AC012442.5 | 16 | 0.89690 | 49 | | rs1475944 | chr22: 47,307,286 | C/T | 6.35 | TBC1D22A | TBC1D22A-AS1 | 9 | 0.56165 | 4 | | | rs2294512 | chr1: 54,370,305 | G/A | 4.55 | DIO1 | HSPB11 LDLRAD1 LRRC42 YIPF1 MIR4781 TCEANC2 TMEM59 | DIO1 | 29 | 0.00651 | 1 | | rs4750321 | chr10: 13,272,254 | T/C | 2.71 | UCMA | MCM10 PHYH SEPHS1 | 10 | 0.41288 | 2 | | | rs6551724 | chr4: 63,730,014 | G/A | 2.60 | Intergenic variant EXOC5P1 | 0.00531 | 1 | ||| | rs2027517 | chr6: 136,215,473 | G/A | 1.97 | PDE7B | 12 | 0.35567 | 2 | || | rs79282103 | chr16: 5,928,451 | G/A | 1.16 | RBFOX1 | 0.56725 | 1 | ||| | rs35205792 | chr2: 15,249,410 | T/C | 1.08 | Intergenic variant NBAS | 4 | 0.57183 | 1 | || | rs3133223 | chr11: 120,605,231 | G/A | 0.95 | GRIK4 | GRIK4 | 5 | 0.43621 | 1 | The scores in the columns represent the cumulative scores derived from various sources, including 3D interacting genes, enhancers, promoters, transcription factor binding sites, sequence motif changes, and conservation. A higher score indicates that the SNP locus is more likely to be located in a promoter or enhancer element. The Linear Closest Gene refers to the gene that is located within a range of 2 kB upstream to 2 kB downstream of the SNP. The 3D Interactive Gene refers to the gene that exhibits spatial interaction with the SNP locus. Based on the findings from previous studies and the annotation results of the PDM SNPs, we observed that several SNPs are eQTLs of important genes, moreover the linear closest genes and 3D interactive genes of these SNPs are significantly associated with mental health, psychiatric disorders, neurodevelopment, and other brain-related functions. For instance, SNP rs76518691, its linear closest gene is zinc finger MIZ-type containing 1 (ZMIZ1), which encodes a protein involved in regulating the activity of various transcription factors including Smad3/4 and p53, and plays a role in sumoylation. ZMIZ1 is highly expressed in the ovary, endometrium and brain, and has been implicated in regulating depression-related genes in studies involving depressed mice (Wei et al. 2021). It is also involved in the regulation of postmitotic positioning of pyramidal neurons in the developing cerebral cortex (Carapito et al. 2019). ATAC-Seq identified ZMIZ1 regulates chromatin accessibility near Prox1 genomic loci in lymphatic endothelial cells (Rajan et al. 2023). Moreover, Fusion transcripts involving ZMIZ1/PRR12 have also been associated with female mental retardation and neuropsychiatric diseases (Córdova-Fletes et al. 2015). Another SNP of interest, rs72829648, has been identified as the eQTL of gene transmembrane protein 87B (TMEM87B) and fibulin 7 (FBLN7), which has been associated with protein transport to and from the Golgi, mechanosensitive cation channel activity, and other developmental processes (Yu et al. 2016). SNP rs1475944 is the eQTL of gene TBC1 domain family member 22A(TBC1D22A), and predicted to be involved in activation of GTPase activity and intracellular protein transport. It has been detected hyper-methylation in the subjects with bipolar disorder (Sugawara et al. 2018). SNP rs2294512 is the eQTL of gene iodothyronine deiodinase 1 (DIO1), which catalyzes the activation and inactivation of thyroid hormone by outer and inner ring deiodination, and provides most of the circulating T3, which is essential for growth, differentiation and basal metabolism. Mutations in DIO1 have been associated with altered free thyroxine (FT4) levels in both White and African American subjects, as well as with lifetime major depressive disorder in White female subjects, in particular those from high-risk cohorts (Gałecka et al. 2015). Furthermore, in a recent study, DIO1 has been evaluated as potential biomarker combined with DIO2 and solute carrier organic anion transporter family member 1C1 (SLCO1C1) for bipolar spectrum disorder (Zamar et al. 2023), highlighting the potential clinical relevance of these genes in mood disorders. In addition, the linear closest gene to rs2027517, phosphodiesterase 7B (PDE7B) is one of the enzymes responsible for controlling specifically the intracellular levels of cyclic adenosine 3', 5'-monophosphate in the immune and central nervous systems. It has been reported that PDE7B is involved in dopaminergic signaling in striatal neurons (Sasaki et al. 2004) and plays an important role in brain inflammation and neurodegeneration (Morales-Garcia et al. 2015). There is also eQTL of PDE7B located in ± 10 M region of rs2027517. Another gene glutamate ionotropic receptor kainite type subunit 4 (GRIK4), associated with SNP rs3133223, is also noteworthy, which encodes a protein in the glutamate gated ion channel family and is expressed in the brain. Variants of GRIK4 have been linked to major depression in the Chinese Han population (Pu et al. 2020; Ren et al. 2017). There is also eQTL for rs3133223-GRIK4 association in ± 10 M region. In addition, RNA binding fox-1 homolog 1 (RBFOX1), predominantly expressed in the brain, is an RNA-binding protein that modulates RNA metabolism, influencing neural development and crucial neuronal functions. Dysregulation of the function of RBFOX1 can have profound effects on neuronal transcriptome (Prashad and Gopal 2020). Genetic variants in RBFOX1 have been associated with various psychiatric disorders, including attention-deficit/hyperactivity disorder (ADHD) and autism spectrum disorder (ASD) (Bacchelli et al. 2020; Griswold et al. 2015). RBFOX1 has also been implicated in risk tolerance, severe depression, and schizophrenia based on genetic and whole-genome associations (O’Leary et al. 2022). In summary, these findings suggest that the genes corresponding to PDM SNPs identified in the Mendelian randomization study also have important biological significance in various mental health, psychiatric disorders, neurodevelopment, and brain-related functions.

To the best of our knowledge, this study represents the first systematic large-scale investigation into the association between PDM and mental health traits in the Chinese Han young female population. Previous studies have hinted at a potential correlation between PDM and increased anxiety and depression symptoms (Meng et al. 2019). However, systematic examinations of how PDM affects women's mental health across multiple phenotypic dimensions have been limited, and the causal connections have remained uncertain. This study reveals that experiencing dysmenorrhea not only manifests physically but also carries psychological implications that can influence the development of mental health issues. Through a comprehensive phenome study involving a cohort of young female college students, the research sheds light on the significant psychological impact of PDM on this demographic. We utilized the physical and mental health data of 7401 female students aged 17–24 from Qingdao University to construct a multi-phenotype correlation network. The results revealed that the correlation between the PDM phenotype and mental health traits was significantly stronger compared to other phenotypes. Within the PDM endo-phenotype group, several pairs of correlated phenotypes were identified (SCC > 0.5), with mental health traits exhibiting the strongest correlation. Almost all sub-phenotypes of PDM showed significant correlations with depression and anxiety. Factors such as the duration and severity of dysmenorrhea, depression, neuroticism, irritability, and loss of appetite were found to have significant correlations with mental health traits. Additionally, the severity of PDM was found to be positively correlated with the degree of mental ill health, as indicated by the total score of the CMSS scale and hierarchical clustering of mental health traits data. To establish the causal relationship between the PDM phenotype and female mental health traits at the genomic level, a two-sample Mendelian Randomization analysis using the IVW method was conducted, with PDM as the exposure and 18 mental health traits of East Asian individuals as the outcome. The analysis revealed that PDM significantly increased the risk of various mental health problems, including nervous feelings, frequency of depressed mood in the past two weeks, mood swings, and frequency of tenseness/restlessness in the past two weeks. We identified 10 important PDM-related SNPs accounting for the causal relationship, whose closest genes and 3D interactive genes were found to be closely associated with the risk of ill mental health. For instance, ZMIZ1 has been implicated in the regulation of depression-related genes in depressed mice (Wei et al. 2021), and involved in the regulation of postmitotic positioning of pyramidal neurons in the developing cerebral cortex (Carapito et al. 2019). Also the fusion transcripts of ZMIZ1/PRR12 have been confirmed to be associated with female mental retardation and neuropsychiatric diseases (Córdova-Fletes et al. 2015). Common variants of GRIK4 were linked to major depression in the Chinese Han population, and polymorphisms of GRIK4 were found to influence cognitive dysfunction in patients with major depression. Additionally, RBFOX1 can modulate RNA metabolism, influencing neural development and overseeing crucial neuronal functions. Common variations in RBFOX1 have been identified as risk factors for various psychiatric disorders (Bacchelli et al. 2020; Griswold et al. 2015; Prashad and Gopal 2020). These findings indirectly support the notion that PDM may increase the risk of mental health problems through the regulation of SNP variations. We suggest the potential explanation on the mechanism by which PDM affects mental health. Chronic stress may increase the level of glucocorticoid and impair brain function by causing Hypothalamic–pituitary–adrenal axis disorder, thus aggravating the adverse mental state and emotion-like performance of primary dysmenorrhea. On the other hand, Chronic stress is also an important factor to induce the occurrence and development of anxiety and depression and other adverse emotions. Therefore, the PDM women with pelvic pain had more mental health disorders (depression, anxiety, neuroticism, and somatisation) than pain-free controls. One specific inducing signaling molecule is brain-derived neurotrophic factor (BDNF), Low et al. (2018) found that alterations in brain complexity estimated from resting-state magnetoencephalography signals were associated with the interactions between the BDNF Val66Met polymorphism and menstrual pain experience. Although the neural signaling pathway relevant to PDM has been studied less frequently, we have gained an understanding of endometriosis and its neural signaling transduction. There is a consensus among clinicians that dysmenorrhea is mostly induced by endometriosis. Castro et al. (2023) found that mice with endometriosis developed microgliosis, astrogliosis and enhanced substance P neurokinin-1 receptor immunoreactivity within the spinal cord, suggesting the development of neuroinflammation and sensitization of spinal circuitry in endometriosis. Furuta et al. (2012) found that transient receptor potential V1 (TRPV1) and transient receptor potential A1 (TRPA1) act as nocisensors, mediating not only an afferent signal to the dorsal horn of the spinal cord, but also an efferent signal in the periphery through secretion of inflammatory agents, such as substance P and CGRP. This secretion causes local neurogenic inflammation, making TRPV1 and TRPA1 potential receptor targets for chronic pain disorders. The study by Kato et al. (2019) suggested that heart and neural crest derivatives expressed 2 (HAND2) was less expressed by endometriotic tissues compared to endometrial tissues, resulting in an increase in the expression of fibroblast growth factors (FGFs), and consequently enhanced migration and invasion capacity. Notably, a Reverse Mendelian Randomization analysis was also conducted to investigate the potential impact of mental health status on PDM. However, no evidence of a reverse causal relationship was found. While previous studies had limited evidence on stress affecting PDM, this study, utilizing a larger cohort and conducting two-sample Mendelian Randomization, concluded that PDM increases the risk of depression and anxiety, while mental health do not have a significant impact on PDM. Overall, this study not only demonstrated the significant correlation between the PDM phenotype and mental health traits, surpassing other physical and mental phenotypes using the multi-phenotype correlation network model, but also provided systematic insights into the impact of PDM on the mental health of young women, particularly it highlighted the increased risk of depression and anxiety, at the genomic level through Mendelian Randomization analysis. By exploring relevant SNPs, the study shed light on the biological mechanisms underlying the association between PDM and mental health problems. These findings highlight the significance of early intervention and targeted support for individuals experiencing PDM to minimize the associated psychological challenges. The limitation of this study lies in the disparity in the populations of GWAS for PDM (exposure) and mental health (outcome), as genotyping data for the same sample set measuring both PDM and mental health phenotypes is lacking. To address this challenge and bolster the relationship between PDM and mental health unveiled by the multi-phenotype correlation network, we conducted a two-sample Mendelian randomization analysis. In this analysis, the population included in the mental health traits datasets from the UK Biobank was not restricted to young female students, because the PRS results on limited female samples were not sufficiently significant. Moving forward, we will plan to measure genotypes of the female college student cohort to delve deeper into exploring the association between PDM genotype and mental health traits within the exact same sample dataset.

In conclusion, this study demonstrates a significant association between primary dysmenorrhea and mental health traits among female college students. The multi-phenotype correlation network and Mendelian randomization analysis provide valuable insights into the potential relationship and causal implications. Future research should investigate the underlying mechanisms and identify effective interventions to alleviate the mental health issues experienced by individuals with primary dysmenorrhea. Ethics Approval This study was approved by the Ethics Committee of Bio-X Institute with approval number M16033. We confirmed that all methods of processing human samples and experimental protocols were carried out in accordance with relevant guidelines and regulations in the methods section of the manuscript. Supplementary Information Below is the link to the electronic supplementary material.

The computations in this paper were run on the Siyuan-1 cluster supported by the Center for High Performance Computing at Shanghai Jiao Tong University. Abbreviations - PDM Primary dysmenorrhea - CMSS Cox Menstrual Symptom Scale - VAS Visual Analogue Scale - SDS Self-Rating Depression Scale - SAS Self-Rating Anxiety Scale - PDQ-4 Personality Diagnostic Questionnaire-4 - MR Mendelian randomization - SNP Single-nucleotide polymorphism - IVW Inverse-variance weighted - ADHD Attention-deficit/hyperactivity disorder - ASD Autism spectrum disorder - FDR False discovery rate Authors' Contributions H.J. and H.X. collected participants, conducted phenome analysis and wrote the paper; Y.W. and Z.G. measured and processed the physical and mental health data; Y.Zheng and X.J. prepared the figures; A.F. wrote the paper; J.Z. constructed the multi-phenotype correlation network; Y.Zhao and Q.Z. designed the questionnaire on PDM; W.S. and M.Z. conducted the MR and PRS analysis. J.C. conducted analysis on various types of mental health traits; Z.L. designed the research; Z.W. designed the research and wrote the paper; Y.S. designed and funded the research. Funding This work was supported by the National Key R&D Program of China (2024YFA0916600), Shanghai Municipal Science and Technology Major Project, Natural Science Foundation of China (32070679, U1804284, 32370724, 82102669), Taishan Scholar Program of Shandong Province (tstp20240526) and Natural Science Foundation of Shandong Province (ZR2019YQ14), the fundamental research funds for the central universities (23X010302406, YG2023QNB20, YG2021QN143). Data Availability The GWAS summary statistics used in this study are publicly available from the IEU OPEN GWAS project (https://gwas.mrcieu.ac.uk/). Other materials about the female college student cohort are available at request from the corresponding author. Declarations Conflicts of Interest On behalf of all authors, the corresponding authors state that there is no conflict of interest. Manfei Zhang is the Youth Editorial Board Member of Phenomics, and she was not involved in reviewing this paper. Consent to Participate All participants signed informed written consent (Table S7). Consent for Publication Participants signed informed consents to participate in the study and have their data published. Footnotes Huiting Jiangzhou and Hanpeng Xu are contribute equally to this work. Contributor Information Zhuo Wang, Email: [email protected]. Yongyong Shi, Email: [email protected].

- Bacchelli E, Cameli C, Viggiano M et al (2020) An integrated analysis of rare CNV and exome variation in Autism Spectrum Disorder using the Infinium PsychArray. Sci Rep 10(1):3198. 10.1038/s41598-020-59922-3 [DOI] [PMC free article] [PubMed] [Google Scholar] - Bajalan Z, Moafi F, MoradiBaglooei M et al (2019) Mental health and primary dysmenorrhea: a systematic review. J Psychosomat Obstet Gynecol 40(3):185–194. 10.1080/0167482x.2018.1470619 [DOI] [PubMed] [Google Scholar] - Barbosa-Silva J, Avila MA, de Oliveira RF et al (2024) Prevalence, pain intensity and symptoms associated with primary dysmenorrhea: a cross-sectional study. BMC Womens Health 24(1):92. 10.1186/s12905-023-02878-z [DOI] [PMC free article] [PubMed] [Google Scholar] - Benjamini Y, Hochberg Y (1995) Controlling the false discovery rate: a practical and powerful approach to multiple testing. J Roy Stat Soc: Ser B 57(1):289–300 [Google Scholar] - Bowden J, Del Greco MF, Minelli C et al (2016) Assessing the suitability of summary data for two-sample Mendelian randomization analyses using MR-Egger regression: the role of the I2 statistic. Int J Epidemiol 45(6):1961–1974. 10.1093/ije/dyw220 [DOI] [PMC free article] [PubMed] [Google Scholar] - Carapito R, Ivanova EL, Morlon A et al (2019) ZMIZ1 variants cause a syndromic neurodevelopmental disorder. Am J Hum Genet 104(2):319–330. 10.1016/j.ajhg.2018.12.007 [DOI] [PMC free article] [PubMed] [Google Scholar] - Castro J, Maddern J, Erickson A et al (2023) Peripheral and central neuroplasticity in a mouse model of endometriosis. J Neurochem 24. 10.1111/jnc.15843 [DOI] [PubMed] - Chang H, Yan Q, Tang J et al (2017) Possible association between SIRT1 single nucleotide polymorphisms and predisposition to antisocial personality traits in Chinese adolescents. Sci Rep 7(1):1099. 10.1038/s41598-017-01208-2 [DOI] [PMC free article] [PubMed] [Google Scholar] - Córdova-Fletes C, Domínguez MG, Delint-Ramirez I et al (2015) A de novo t(10;19)(q22.3;q13.33) leads to ZMIZ1/PRR12 reciprocal fusion transcripts in a girl with intellectual disability and neuropsychiatric alterations. Neurogenet 16(4):287–298. 10.1007/s10048-015-0452-2 [DOI] [PubMed] - Cox DJ, Meyer RG (1978) Behavioral treatment parameters with primary dysmenorrhea. J Behav Med 1(3):297–310. 10.1007/bf00846681 [DOI] [PubMed] [Google Scholar] - Dorn LD, Negriff S, Huang B et al (2009) Menstrual symptoms in adolescent girls: association with smoking, depressive symptoms, and anxiety. J Adolesc Health 44(3):237–243. 10.1016/j.jadohealth.2008.07.018 [DOI] [PMC free article] [PubMed] [Google Scholar] - Dunstan DA, Scott N (2019) Clarification of the cut-off score for Zung’s self-rating depression scale. BMC Psychiatry 19(1):177. 10.1186/s12888-019-2161-0 [DOI] [PMC free article] [PubMed] [Google Scholar] - Dunstan DA, Scott N (2020) Norms for Zung’s self-rating anxiety scale. BMC Psychiatry 20(1):90. 10.1186/s12888-019-2427-6 [DOI] [PMC free article] [PubMed] [Google Scholar] - Dunstan DA, Scott N, Todd AK (2017) Screening for anxiety and depression: reassessing the utility of the Zung scales. BMC Psychiatry 17(1):329. 10.1186/s12888-017-1489-6 [DOI] [PMC free article] [PubMed] [Google Scholar] - Elsworth B, Lyon M, Alexander T et al (2020) The MRC IEU OpenGWAS data infrastructure. BioRxiv, pp 2020-08. 10.1101/2020.08.10.244293 - Farrow E, Chiocchetti AG, Rogers JC et al (2021) SLC25A24 gene methylation and gray matter volume in females with and without conduct disorder: an exploratory epigenetic neuroimaging study. Transl Psychiatry 11(1):492. 10.1038/s41398-021-01609-y [DOI] [PMC free article] [PubMed] [Google Scholar] - Furuta A, Suzuki Y, Hayashi N et al (2012) Transient receptor potential A1 receptor-mediated neural cross-talk and afferent sensitization induced by oxidative stress: Implication for the pathogenesis of interstitial cystitis/bladder pain syndrome. Int J Urol 19(5):429–436. 10.1111/j.1442-2042.2012.02966.x [DOI] [PubMed] [Google Scholar] - Gagua T, Tkeshelashvili B, Gagua D et al (2013) Assessment of anxiety and depression in adolescents with primary dysmenorrhea: a case-control study. J Pediatr Adolesc Gynecol 26(6):350–354. 10.1016/j.jpag.2013.06.018 [DOI] [PubMed] [Google Scholar] - Gałecka E, Talarowska M, Orzechowska A et al (2015) Association of the DIO2 gene single nucleotide polymorphisms with recurrent depressive disorder. Acta Biochim Pol 62(2):297–302. 10.18388/abp.2015_1002 [DOI] [PubMed] - Griswold AJ, Dueker ND, Van Booven D et al (2015) Targeted massively parallel sequencing of autism spectrum disorder-associated genes in a case control cohort reveals rare loss-of-function risk variants. Mol Autism 6(1). 10.1186/s13229-015-0034-z [DOI] [PMC free article] [PubMed] - Han VX, Patel S, Jones HF et al (2021) Maternal acute and chronic inflammation in pregnancy is associated with common neurodevelopmental disorders: a systematic review. Transl Psychiatr 11(1):12. 10.1038/s41398-021-01198-w [DOI] [PMC free article] [PubMed] [Google Scholar] - Hemani G, Zheng J, Elsworth B et al (2018) The MR-Base platform supports systematic causal inference across the human phenome. Elife 7. 10.7554/eLife.34408 [DOI] [PMC free article] [PubMed] - Huang DD, Zhou Y, Yi XF et al (2022) VannoPortal: multiscale functional annotation of human genetic variants for interrogating molecular mechanism of traits and diseases. Nucleic Acids Res 50(D1):D1408–D1416. 10.1093/nar/gkab853 [DOI] [PMC free article] [PubMed] [Google Scholar] - Kabukçu C, Basay BK, Basay Ö (2021) Primary dysmenorrhea in adolescents: Association with attention deficit hyperactivity disorder and psychological symptoms. Taiwan J Obstet Gynecol 60(2):311–317. 10.1016/j.tjog.2021.01.033 [DOI] [PubMed] [Google Scholar] - Kato N, Iwase A, Ishida C et al (2019) Upregulation of fibroblast growth factors caused by heart and neural crest derivatives expressed 2 suppression in endometriotic cells: a possible therapeutic target in endometriosis. Reprod Sci 26(7):979–987. 10.1177/1933719118802053 [DOI] [PubMed] [Google Scholar] - Kirchner SK, Roeh A, Nolden J et al (2018) Diagnosis and treatment of schizotypal personality disorder: evidence from a systematic review. NPJ Schizophr 4(1):20. 10.1038/s41537-018-0062-8 [DOI] [PMC free article] [PubMed] [Google Scholar] - Klein JR, Litt IF (1981) Epidemiology of adolescent dysmenorrhea. Pediatrics 68(5):661–664 [PubMed] [Google Scholar] - Li Z, Chen J, Zhao Y et al (2017) Common variants in ZMIZ1 and near NGF confer risk for primary dysmenorrhoea. Nat Commun 8:14900. 10.1038/ncomms14900 [DOI] [PMC free article] [PubMed] [Google Scholar] - Li L, Lok GKI, Mei SL et al (2020) Prevalence of depression and its relationship with quality of life among university students in Macau, Hong Kong and mainland China. Sci Rep 10(1):15798. 10.1038/s41598-020-72458-w [DOI] [PMC free article] [PubMed] [Google Scholar] - Low I, Kuo PC, Tsai CL et al (2018) Interactions of BDNF Val66Met polymorphism and menstrual pain on brain complexity. Front Neurosci 12:17. 10.3389/fnins.2018.00826 [DOI] [PMC free article] [PubMed] [Google Scholar] - Meng L, Li J, Cheng Y et al (2019) Dysmenorrhea increased the risk of postpartum depression in Chinese Han parturients. Sci Rep 9(1):16579. 10.1038/s41598-019-53059-8 [DOI] [PMC free article] [PubMed] [Google Scholar] - Morales-Garcia JA, Aguilar-Morante D, Hernandez-Encinas E et al (2015) Silencing phosphodiesterase 7B gene by lentiviral-shRNA interference attenuates neurodegeneration and motor deficits in hemiparkinsonian mice. Neurobiol Aging 36(2):1160–1173. 10.1016/j.neurobiolaging.2014.10.008 [DOI] [PubMed] [Google Scholar] - O’Leary A, Fernàndez-Castillo N, Gan G et al (2022) Behavioural and functional evidence revealing the role of RBFOX1 variation in multiple psychiatric disorders and traits. Mol Psychiatry 27(11):4464–4473. 10.1038/s41380-022-01722-4 [DOI] [PMC free article] [PubMed] [Google Scholar] - Pakpour AH, Kazemi F, Alimoradi Z et al (2020) Depression, anxiety, stress, and dysmenorrhea: a protocol for a systematic review. Syst Rev 9(1):65. 10.1186/s13643-020-01319-4 [DOI] [PMC free article] [PubMed] [Google Scholar] - Prashad S, Gopal PP (2020) RNA-binding proteins in neurological development and disease. RNA Biol 18(7):972–987. 10.1080/15476286.2020.1809186 [DOI] [PMC free article] [PubMed] [Google Scholar] - Price ND, Magis AT, Earls JC et al (2017) A wellness study of 108 individuals using personal, dense, dynamic data clouds. Nat Biotechnol 35(8):747–756. 10.1038/nbt.3870 [DOI] [PMC free article] [PubMed] [Google Scholar] - Proctor M, Farquhar C (2006) Diagnosis and Management of Dysmenorrhoea Bmj 332(7550):1134–1138. 10.1136/bmj.332.7550.1134 [DOI] [PMC free article] [PubMed] [Google Scholar] - Pu M, Wang X, Zhang J (2020) Impact of GRIK4 gene polymorphisms on cognitive dysfunction in patients with major depression. Rev Neurol 176(10):823–828. 10.1016/j.neurol.2019.11.009 [DOI] [PubMed] [Google Scholar] - Quan C, Ping J, Lu H et al (2022a) 3DSNP 2.0: update and expansion of the noncoding genomic variant annotation database. Nucleic Acids Res 50(D1):D950-D955. 10.1093/nar/gkab1008 [DOI] [PMC free article] [PubMed] - Rajan KC, Patel NR, Shenoy A et al (2023) Zmiz1 is a novel regulator of lymphatic endothelial cell gene expression and function. bioRxiv : the preprint server for biology. 10.1101/2023.07.22.550165 - Ren D, Bi Y, Xu F et al (2017) Common variants in GRIK4 and major depressive disorder: An association study in the Chinese Han population. Neurosci Lett 653:239–243. 10.1016/j.neulet.2017.05.071 [DOI] [PubMed] [Google Scholar] - Sasaki T, Kotera J, Omori K (2004) Transcriptional activation of phosphodiesterase 7B1 by dopamine D1 receptor stimulation through the cyclic AMP/cyclic AMP-dependent protein kinase/cyclic AMP-response element binding protein pathway in primary striatal neurons. J Neurochem 89(2):474–483. 10.1111/j.1471-4159.2004.02354.x [DOI] [PubMed] [Google Scholar] - Schlosser P, Li Y, Sekula P et al (2020) Genetic studies of urinary metabolites illuminate mechanisms of detoxification and excretion in humans. Nature Genet 52(2):167-+. 10.1038/s41588-019-0567-8 [DOI] [PMC free article] [PubMed] - Shin SY, Fauman EB, Petersen AK et al (2014) An atlas of genetic influences on human blood metabolites. Nat Genet 46(6):543–550. 10.1038/ng.2982 [DOI] [PMC free article] [PubMed] [Google Scholar] - Sugawara H, Murata Y, Ikegame T et al (2018) DNA methylation analyses of the candidate genes identified by a methylome-wide association study revealed common epigenetic alterations in schizophrenia and bipolar disorder. Psychiatry Clin Neurosci 72(4):245–254. 10.1111/pcn.12645 [DOI] [PubMed] [Google Scholar] - Vural S, Wang X, Guda C (2016) Classification of breast cancer patients using somatic mutation profiles and machine learning approaches. BMC Syst Biol 10(Suppl 3):62. 10.1186/s12918-016-0306-z [DOI] [PMC free article] [PubMed] - Wang L, Wang X, Wang W et al (2004) Stress and dysmenorrhoea: a population based prospective study. Occup Environ Med 61(12):1021–1026. 10.1136/oem.2003.012302 [DOI] [PMC free article] [PubMed] [Google Scholar] - Wei Y, Qi K, Yu Y et al (2021) Analysis of differentially expressed genes in the dentate gyrus and anterior cingulate cortex in a mouse model of depression. Biomed Res Int 2021:5013565. 10.1155/2021/5013565 [DOI] [PMC free article] [PubMed] [Google Scholar] - Weissman AM, Hartz AJ, Hansen MD et al (2004) The natural history of primary dysmenorrhoea: a longitudinal study. BJOG 111(4):345–352. 10.1111/j.1471-0528.2004.00090.x [DOI] [PubMed] [Google Scholar] - Wickham H (2016) ggplot2: Elegant Graphics for Data Analysis. - Wu X, Yu W, Tian X et al (2021) Altered posterior cerebellar lobule connectivity with perigenual anterior cingulate cortex in women with primary dysmenorrhea. Front Neurol 12:645616. 10.3389/fneur.2021.645616 [DOI] [PMC free article] [PubMed] [Google Scholar] - Yu HC, Coughlin CR, Geiger EA et al (2016) Discovery of a potentially deleterious variant in TMEM87B in a patient with a hemizygous 2q13 microdeletion suggests a recessive condition characterized by congenital heart disease and restrictive cardiomyopathy. Cold Spring Harb Mol Case Stud 2(3):a000844. 10.1101/mcs.a000844 [DOI] [PMC free article] [PubMed] [Google Scholar] - Yu S, Xu J, Shen Z et al (2022) Frequency-specific alterations in brain function in patients with primary dysmenorrhea. Pain Med 23(5):902–911. 10.1093/pm/pnab225 [DOI] [PubMed] [Google Scholar] - Zamar A, Mohamed A, Lulsegged A et al (2023) Towards the first biomarker test for bipolar spectrum disorder: an evaluation of 199 patients in an outpatient setting. J Pers Med 13(7). 10.3390/jpm13071108 [DOI] [PMC free article] [PubMed] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials Data Availability Statement The GWAS summary statistics used in this study are publicly available from the IEU OPEN GWAS project (https://gwas.mrcieu.ac.uk/). Other materials about the female college student cohort are available at request from the corresponding author.