The genetics of primary dysmenorrhoea: a systematic review

This project did not receive external funding. RRM is currently funded by a grant from the Medical Research Foundation as part of the UKRI Strategic Priorities Fund (SPF) Advanced Pain Discovery Platform (APDP), a co-funded initiative by UKRI (MRC, BBSRC, ESRC), Versus Arthritis, the Medical Research Foundation and Eli Lilly and Company Ltd (grant ref: MR/W02697/X1). IWL received a postgraduate studentship from MRC-iCASE and Eli Lilly.

This review was performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement ( Page et al. 2021 ), and the protocol was registered on PROSPERO (registration ID: CRD42023455337) on 17 August 2023. A comprehensive search strategy was designed by EH, MS and KV with advice from a medical librarian, using key terms relating to dysmenorrhoea and genetic factors, for example, ‘dysmenorrhea’, ‘dysmenorrhoea’, ‘menstrual pain’, ‘polymorphisms’ and ‘SNPs’. The search strategy was subsequently trialled and iteratively optimised via multiple pilot searches, which were cross-checked against a list of target articles collated from a preliminary exploration of the literature. Final searches were conducted in October 2023 using three databases (MEDLINE (Ovid), Embase (Ovid) and Web of Science Core Collection). The search strategy used in MEDLINE can be seen in Supplementary Fig. 1 (see section on Supplementary materials given at the end of the article). Auto-alerts were set up following the October 2023 search so that, as new articles were added to the databases, the searches were automatically re-run to identify any additional relevant articles. Any such articles were evaluated and screened until November 2025. Search results were retrieved and duplicates were removed, both automatically in Endnote 21 and manually by EH, when necessary. References were then imported into Rayyan ( Ouzzani et al. 2016 ) for blinded screening. Two authors (EH and MS) independently screened the titles and abstracts (primary screening) and then read the full text of the remaining articles in the second screening stage. Each stage was followed by a conflict resolution meeting to resolve any differences in inclusion/exclusion of articles and establish a consensus. Additional articles identified up to November 2025 were screened by EH and then discussed with MS to validate their inclusion/exclusion. All English language studies investigating specific genes/polymorphisms associated with PDM in females were included. No restriction was placed on year of publication. Non-human studies and studies not published in peer-reviewed journals were excluded. Full inclusion and exclusion criteria are listed in Table 1 . Given that PDM is often poorly defined, with a lack of as yet unidentified pathology assumed to mean the lack of pathology present, we also included studies that did not differentiate between PDM and SDM. However, any articles relating only to SDM were excluded. Inclusion and exclusion criteria. Humans – female only Animal studies PDM as defined by the study SDM – e.g. confirmed diagnosis of endometriosis, adenomyosis or any other identifiable pelvic pathologies ○ If the study does not differentiate between PDM and SDM, it will still be included Age range: menarche to end of life Dysmenorrhoea due to intrauterine contraceptive device Study design of any type Cancer diagnosis Studies presenting original data regarding specific genes/polymorphisms involved in PDM in females Studies not published in peer-reviewed journals – grey literature, conference abstracts, PhD theses, pre-prints, case reports, letters, editorials, reviews, opinion pieces Articles not in English language PDM, primary dysmenorrhoea; SDM, secondary dysmenorrhoea. The relevant data from eligible studies were extracted following the second screening stage. The following data were extracted and tabulated: paper title; first author; year published; ethnic origin of participants; country of study; study design/type of study; how PDM is defined; inclusion criteria for PDM subjects; how SDM cases were excluded; how healthy controls were defined; number of participants; number of controls; mean age of participants; mean age of controls; method of genotyping; statistical analysis software used; genetic associations found (results of the study); other significant variables associated with primary dysmenorrhoea; and author conclusions. The quality of the included studies was also evaluated at this stage to assess the risk of bias. The STREGA guidelines ( Little et al. 2009 ) were used for every study, and the Newcastle–Ottawa Scale (NOS) ( Wells et al. 2000 ), in addition, for case–control studies. The NOS involves three sections, in which stars are awarded for fulfilling certain criteria – selection of cases and controls (max. four stars); comparability of cases and controls (max. two stars); and ascertainment of exposure (max. three stars). Studies were ranked as having a high (0–3 stars), medium (4–6 stars) or low (7–9 stars) risk of bias. All studies were included in the review regardless of the methodological quality assessed by these tools. Data extraction and quality assessment were carried out by one investigator (EH) and, upon completion, checked independently by MS. A narrative synthesis was conducted, summarising the significant genetic variants ( P < 0.05 for candidate gene studies; P < 5 × 10 −8 for GWASs) identified as being associated with PDM, as well as their specific genetic location and known or potential biological function. Meta-analysis was planned for genetic variants, including subgroup analysis based on ethnicity, if sufficient data were available.

The original search (October 2023) yielded 774 articles after duplicates were removed. Publication year ranged from 1950 to 2023. A total of 748 articles (97%) were excluded during the title/abstract screening, meaning that 26 articles progressed to the second screening stage. After a full-text review, a further 11 articles (42%) were excluded, leaving 15 articles to be included in the systematic review. Evaluation of the search auto-alerts until November 2025 identified five potentially relevant articles ( Nacar et al. 2023 , Chen et al. 2024 , Hsu et al. 2024 , Li et al. 2024 , Liu et al. 2024 ); however, none of these met the inclusion criteria following a full-text review. A detailed search and screening history is shown in Fig. 1 . Flow chart detailing identification, screening and inclusion of studies in this review. Of the 15 included studies, seven (46.7%) were case–control studies, four (26.7%) were cross-sectional studies, and four (26.7%) were GWASs. The characteristics of the included case–control and cross-sectional studies are summarised in Table 2 . Five studies involved Turkish participants ( Ozsoy et al. 2015 , Dogru et al. 2016 , Ozsoy et al. 2016 , Esen et al. 2020 , Nacar et al. 2022 ), two studies had a Nigerian population ( Olasore et al. 2022 , Olasore et al. 2023 ), and there was one study with each of the following populations: Taiwanese ( Lee et al. 2014 ), Chinese ( Wu et al. 2000 ), Asian ( Donayeva et al. 2023 ) and South Korean ( Woo et al. 2010 ). It is important to note that Olasore et al . (2022) make reference to over 250 different ethnic groups within Nigeria, so their population is likely to be diverse. The mean age of participants ranged from 15 to 26 years. Two studies were conducted in adolescents (aged ≤ 17 years) ( Woo et al. 2010 , Donayeva et al. 2023 ). One study did not report the mean age of participants ( Wu et al. 2000 ). For case–control studies, sample sizes ranged from 200 ( Lee et al. 2014 ) to 302 ( Ozsoy et al. 2016 ) ( Table 3 ). For cross-sectional studies, sample sizes ranged from 102 ( Olasore et al. 2022 ) to 435 individuals ( Wu et al. 2000 ) ( Table 3 ). Characteristics of case–control and cross-sectional (candidate gene) studies. Values are mean age in years. All participants were aged between 16 and 17 years. Mean age of both cohorts was 20.6 years. Mean age of both cohorts was 21.0 years. Inner Central Black Sea region. Black Sea region. Middle Black Sea region. Kazakhstan Aktobe region. GI, gastrointestinal; NRS, numerical rating scale; NSAIDs, non-steroidal anti-inflammatory drugs; OCP, oral contraceptive pill; PCOS, polycystic ovary syndrome; VAS, visual analogue scale; PDM, primary dysmenorrhoea; SDM, secondary dysmenorrhoea. Findings from case–control and cross-sectional (candidate gene) studies. P values in bold indicate polymorphisms significantly associated with PDM ( P < 0.05). Summary statistic for association with recurrent dysmenorrhoea (no association found for occasional dysmenorrhoea – data not presented here). rs# was not reported in paper and could not be determined after Google search. PDM, primary dysmenorrhoea. It is important to note that five of the seven case–control studies (71%) were conducted by the same research group and recruited patients from the same hospital in the Central Black Sea region of Turkey ( Ozsoy et al. 2015 , Dogru et al. 2016 , Ozsoy et al. 2016 , Esen et al. 2020 , Nacar et al. 2022 ). Therefore, it is possible that the same or a very similar population was used for each, especially given the almost identical mean ages of cases and controls ( Table 2 ). We assessed whether and how these studies differentiated between PDM and SDM in their populations. Among these 11 studies, nine differentiated between PDM and SDM ( Wu et al. 2000 , Lee et al. 2014 , Ozsoy et al. 2015 , 2016 , Dogru et al. 2016 , Esen et al. 2020 , Nacar et al. 2022 , Donayeva et al. 2023 , Olasore et al. 2023 ). Three used ultrasonography ( Lee et al. 2014 , Dogru et al. 2016 , Donayeva et al. 2023 ), while three others reported using a clinical examination only ( Ozsoy et al. 2015 , Esen et al. 2020 , Nacar et al. 2022 ). The remaining three studies used history alone to exclude women with an existing diagnosis of conditions related to SDM, such as endometriosis, PCOS, myoma or ovarian cyst ( Wu et al. 2000 , Ozsoy et al. 2016 , Olasore et al. 2023 ). No studies reported using laparoscopy to exclude SDM. Two studies made no attempt to differentiate between PDM and SDM ( Woo et al. 2010 , Olasore et al. 2022 ); therefore, it is likely that some SDM cases are present in these study cohorts. The results of the 11 case–control and cross-sectional studies are detailed in Table 3 . Twenty different polymorphisms were investigated across the 11 candidate gene studies, selected specifically by studies due to their known or suspected roles in oestrogen metabolism, cytokine, prostaglandin or other inflammatory signalling, circadian regulation of hormone and pain pathways, and pain modulation ( Table 3 ). Three polymorphisms ( ESR1 PvuII, IL4 VNTR and GSTM1 ) were each investigated in two separate studies, but otherwise each study investigated a different polymorphism. Two studies found that the ESR1 -397T>C (PvuII) polymorphism was significantly ( P C (PvuII) polymorphisms were associated with an increased risk of PDM ( Ozsoy et al. 2016 ). Woo et al . (2010) also reported that this polymorphism was associated with PDM under the dominant model (CC + CT:TT) (OR: 3.38, 95% CI: 1.39–8.21, P = 0.007), suggesting that the C allele in the ESR1 polymorphism is associated with dysmenorrhoea. Two studies found that the IL4 intron 3 VNTR polymorphism was not associated with PDM. This was consistent in Ozsoy et al. (2015) using the P2P2 + P1P2:P1P1 genotype (OR: 1.1, 95% CI: 0.2–6.2, P = 0.886), P2P2: P1P2 + P1P1 genotype (OR: 0.8, 95% CI: 0.4–1.3, P = 0.357) or allele frequencies (OR: 0.8, 95% CI: 0.5–1.3, P = 0.440) ( Ozsoy et al. 2015 , Esen et al. 2020 , also using the P1P1 + P1P2:P2P2 genotypes (OR: 0.67, 95% CI: 0.38–1.16, P > 0.05), P1P1:P1P2 + P2P2 genotypes (OR: 0.48, 95% CI: 0.01–6.38, P > 0.05) or allele frequencies (OR: 0.71, 95% CI: 0.43–1.15, P > 0.05) ( Esen et al. 2020 ). The replication of this finding increases confidence in a true lack of effect; although the estimates are imprecise, sample sizes are small ( n = 236–294), and generalisability is limited by both studies being conducted in a Turkish population. With regard to the GSTM1 polymorphism (either the presence or absence of GSTM1 ), two studies produced contrasting results. GSTM1 absence was associated with recurrent PDM in one study ( Wu et al. 2000 ) (adjusted OR: 1.8, CI: 1.0–3.4) ( n = 435), and in the same study, GSTM1 absence also increased the risk of recurrent PDM when combined with the CYP2D6 Aa/aa (adjusted OR: 3.1, 95% CI: 1.2–8.0) or CYP2D6 AA genotype (adjusted OR: 2.3, 95% CI: 1.0–5.1) ( Wu et al. 2000 ). However, Woo et al. (2010) ( n = 202) found that the GSTM1 -absent polymorphism was not associated with dysmenorrhoea (unadjusted OR: 0.6, 95% CI: 0.3–1.4, P = 0.2) ( Woo et al. 2010 ). For the 17 polymorphisms investigated by a single study, six were found to be significantly ( P A ( Dogru et al. 2016 ), PER3 VNTR ( Nacar et al. 2022 ), COX2 -1195G>A ( Olasore et al. 2022 ), VDR TaqI ( Donayeva et al. 2023 ) and eNOS Glu298Asp ( Olasore et al. 2023 ). Two combined genotypes were also associated with PDM: CYP2D6/GSTM1 variant genotypes ( Wu et al. 2000 ) and ESR1 AG-TC (heterozygous at XbaI and PvuII) ( Ozsoy et al. 2016 ). In addition, eight of the investigated polymorphisms and one combined genotype were not found to be associated with dysmenorrhoea: CYP2D6 ( Wu et al. 2000 ); GSTT1 ( Woo et al. 2010 ); GSTP1 ( Woo et al. 2010 ); MTHFR C677T ( Ozsoy et al. 2015 ); MIF -173G>C ( Dogru et al. 2016 ); ESR1 -351A>G (XbaI), IL6 -572G>C and IL6 -597G>A ( Ozsoy et al. 2016 ); and the combined IL6 -572G>C/-597G>A genotype ( Ozsoy et al. 2016 ). The characteristics of GWASs are summarised in Table 4 . Participants in these studies were of European ( Jones et al. 2016 ), Mainland Eastern Chinese ( Li et al. 2017 ), Japanese ( Hirata et al. 2018 ) and Taiwanese Han Chinese ( Lee et al. 2022 ) ancestries. Total cohort sizes ranged from 6,770 ( Li et al. 2017 ) to 15,206 ( Lee et al. 2022 ), while the mean ages of cases and controls ranged from 19.9 and 19.4 years, respectively ( Li et al. 2017 ), to 38.4 and 40.9 years, respectively ( Lee et al. 2022 ). Two studies did not report mean age at group level for cases and controls ( Jones et al. 2016 , Hirata et al. 2018 ); however, Jones et al. (2016) reported that participants were between 18 and 45 years, while Hirata et al. (2018) had mean ages (SD) ranging from 31 (6.4) to 35 (7.0) when presented by pain severity level. Characteristics of genome-wide association studies (GWAS). Study measured dysmenorrhea severity with an ordinal variable: 1,785 women reported no pain; 10,106 women reported at least some pain. Not reported in study. Study reported that participants were between 18 and 45 years old. Study measured dysmenorrhea severity using an ordinal scale: 477 women reported no pain; 10,871 women reported at least some pain. Age was presented for five dysmenorrhea severity levels: mean (SD) for no pain was 34.7(7.0) years; for the highest pain level, mean (SD) was 31.3 (6.4) years. PCOS, polycystic ovary syndrome; VAS, visual analogue scale; PDM, primary dysmenorrhea; SDM, secondary dysmenorrhea. In terms of differentiation between PDM and SDM cases, one study used clinical examination only ( Li et al. 2017 ), while another excluded women with existing diagnoses of conditions related to SDM from their medical history ( Lee et al. 2022 ). Two studies did not differentiate between PDM and SDM ( Jones et al. 2016 , Hirata et al. 2018 ). Seven index SNPs reached genome-wide significance (GWS) ( P < 5 × 10 −8 ) in their respective studies ( Table 5 ). All seven SNPs co-localised with three genes ( NGF , IL1 and ZMIZ1 ), with associations at NGF and IL1 loci being replicated in at least one other study. All four GWASs found an association between SNPs located near the NGF locus and PDM – this represents the most widely validated association based on currently available data in this field. Notably, the associations with NGF were replicated across different ethnic groups ( n = 3 East Asian and n = 1 European), while the findings relating to IL1 were replicated in two East Asian populations ( Hirata et al. 2018 , Lee et al. 2022 ). Hirata et al. (2018) found that the two NGF SNPs that Jones et al. (2016) and Li et al. (2017) had identified in their studies (rs7523086 and rs752381, respectively) were in high linkage disequilibrium (LD, r 2 > 0.8) with their GWS SNP (rs12030576). Similarly, Lee et al. (2022) reported that all three previously identified NGF SNPs were in high LD ( r 2 > 0.8) with the GWS SNP they found at this locus (rs2982742). Significant loci for primary dysmenorrhea from genome-wide association studies (GWAS). Not reported in study. Value is effect size (95% CI) (per unit increase in dysmenorrhea severity on a 4-point ordinal scale). Associations at IL1 were reported by both Hirata et al. (2018) and Lee et al. (2022) – SNPs rs80111889 and rs11676014, respectively – with LD analysis by Lee et al. (2022) replicating Hirata et al. ’s (2018) SNP finding. An association between IL1A gene (SNP rs3783550) and PDM was acknowledged by Li et al. (2017) , but it did not reach GWS, after neither principal component analysis nor gene-based and pathway-based analyses ( P = 2.6 × 10 −5 ). Hirata et al. (2018) , nevertheless, found this SNP (rs3783550) to be in LD with their IL1 variant, suggesting some involvement in dysmenorrhoea. The finding at ZMIZ1 was not replicated, being identified only by Li et al. (2017) . However, it is worth mentioning that all SNPs reported by Li et al. (2017) only reached significance after replication analysis in an independent cohort. Each GWAS also conducted a functional analysis of its GWS findings to identify the putative biological roles of GWS. All GWASs identified index SNPs that were co-localised with the NGF gene, each of which were found to be non-coding (three intergenic ( Jones et al. 2016 , Li et al. 2017 , Hirata et al. 2018 ) and one intronic to NGF ( Lee et al. 2022 )). Each study suggested putative regulatory functions of the identified genetic loci. Li et al. (2017) predicted that most SNPs (including the index SNP and those in LD) resided within promoter and/or enhancer elements, while Jones et al. (2016) also reported that half of SNPs in the cluster overlapped with promoter/enhancer histone marks in gynaecological tissues, suggesting that these variants lie within a regulatory domain that may control the gene expression. Lee et al. (2022) also predicted 21 of 24 SNPs within the NGF -situated cluster to alter regulatory binding motifs. Jones et al. (2016) , Hirata et al. (2018) , and Lee et al. (2022) also reported the relevance of their GWAS hits to RP4-663N10.1 – a conserved antisense long non-coding RNA (lncRNA) that spans the length of the NGF gene. In tissue expression analysis, Jones et al. (2016) reported highest expression of RP4-663N10.1 in adipose tissue and gynaecological tissues, including the cervix, the fallopian tubes and the uterus. Similarly, Hirata et al. (2018) found moderate-to-strong support for the co-localisation of the index and LD SNPs with RP4-663N10.1 eQTLs (expression quantitative trait loci) in adipose, ovarian and uterus tissues. eQTLs are genetic loci that regulate expression levels of mRNAs or proteins and thus explain some variation within a gene expression phenotype ( Nica & Dermitzakis 2013 ). However, these studies found conflicting directions of effect, with Hirata et al. (2018) finding an increased expression of RP4-663N10.1 in gynaecological tissues associated with increased dysmenorrhoea, while Jones et al. (2016) finding an increased dysmenorrhoea pain severity associated with reduced RP4-663N10.1 expression. Fifteen SNPs in Lee et al. (2022) were eQTLs for RP4-663N10.1 but in aortic tissue only. Hirata et al. (2018) found that IL1 SNPs (rs80111889 and 42 high LD variants) were intronic to IL1A or intergenic regions between IL1A and IL1B or IL1A and CKAP2L . Strong support was found for co-localisation with IL1A eQTLs. Fourteen of the 42 high LD SNPs overlapped promoter/enhancer regions for ≥20 tissues. Lee et al. (2022) found that SNPs at the IL1 locus were mostly located in intergenic regions and, similarly to Hirata et al. (2018) , were predicted to have a regulatory function. A novel finding was that two SNPs at the IL1B locus were found to be eQTLs for NT5DC4 , a gene within the NT5DC family that encodes a 5′-nucleotidase that catalyses hydrolysis of intracellular 5′ nucleotides ( Singgih et al. 2021 ). Any relevance to dysmenorrhoea has not yet been explored ( Cros-Perrial & Jordheim 2025 ). Tissue-specific regulatory analysis for the ZMIZ1 index SNP (rs76518691) and its LD SNPs highlighted activity marks in adipose-derived mesenchymal stem cells ( Li et al. 2017 ). All 15 studies were evaluated using the STREGA guidelines. Several checklist items were absent from many, if not all, studies, demonstrating that improvement in the reporting of genetics studies is needed. Surprisingly, the majority of studies did not report participant recruitment periods, with some also not specifying where the population was recruited from. Two further key areas lacking in reporting quality were the laboratory genotyping methods, where many failed to report where genotyping was done and how the DNA was stored prior to analysis, and any discussion of confounders or sources of potential bias by the authors. Some studies also failed to report genetic variants with widely used nomenclature, often lacking the rs#. Supplementary Table 1 provides a comprehensive overview of our STREGA evaluation of all 15 studies. Of the seven case–control studies assessed using the NOS, five had high risk of bias and two had medium risk of bias. No studies adequately matched cases and controls or reported the non-response rate. Other common issues were studies not reporting where controls were recruited from and how PDM cases and controls were defined. The full NOS risk-of-bias assessment can be seen in Supplementary Table 2. There were insufficient data to undertake a meta-analysis as had been planned.

This systematic review has evaluated the current literature on genetics and PDM, showing that both inflammatory and pain pathways appear to be involved. However, the literature is sparse and hampered by poor definitions of PDM and a lack of broad geographic and ethnic coverage, meaning that much remains unknown. Given the limited replicated results in this review, it is clear that further research in this field is urgently required. A better understanding of the pathogenesis of PDM will aid in the development of novel therapies and in identifying at-risk individuals, thus potentially allowing implementation of preventative strategies with the ultimate aim of reducing the burden of this common yet long neglected condition.

This review qualitatively summarises the current state of our understanding of the genetic associations with PDM. We identified evidence for loci at genes involved in inflammatory ( IL1 , TNF ), pain ( NGF , BDNF ) and oestrogen metabolism ( ESR1 ) pathways. Our review also highlights how poorly PDM is defined within studies and the limited data covering populations of mixed ethnicity, with only one transethnic finding ( NGF ). Since several of the gene findings have also been identified as endometriosis loci, this highlights how important proper differentiation between PDM and SDM is in order to understand their disease profiles and inform future treatments but also raises the important question of how distinct PDM and SDM really are. Given that dysmenorrhoea is, by definition, a pain condition, it is perhaps unsurprising that the best replicated finding across studies was an association between NGF SNPs and PDM. NGF is a neurotrophin that stimulates nerve growth and has a well-established role in pain pathophysiology ( Barker et al. 2020 ). Both pre-clinical and human studies have demonstrated a role for NGF in nociceptor sensitisation in both the short term and longer term. Importantly, in the context of dysmenorrhoea (which is related to contractions of the uterine muscle), NGF administration has been shown to be associated with the development of muscle pain ( Lewin et al. 1993 , Petty et al. 1994 ). In addition, NGF can trigger the release of inflammatory mediators and BDNF ( Barker et al. 2020 ). A recent large-scale GWAS for endometriosis ( Rahmioglu et al. 2023 ) also identified a locus at the NGF gene. There is also evidence that NGF-containing peritoneal fluid from endometriosis patients increases neurite outgrowth ( Barcena De Arellano et al. 2011 ). However, other studies are inconsistent in their findings relating NGF levels to endometriosis pain symptoms ( Barcena De Arellano et al. 2013 , Kajitani et al. 2013 ). BDNF has also been shown to relate to symptom severity and is suggested to have an important role in endometriosis-associated pain ( Ding et al. 2018 ). An established role of NGF in PDM could open the door to novel therapeutic opportunities, such as anti-NGF therapies. These have already been explored in other chronic pain conditions, including in women with interstitial cystitis/bladder pain syndrome ( Nickel et al. 2016 ), a visceral pain phenotype with many features in common with dysmenorrhoea. However, significant adverse events, including abnormal peripheral sensation in trials of these therapies in osteoarthritis ( Zhao et al. 2022 ), have led to the recommendation that more research and long-term follow-up are needed to address the current inconsistent, inconclusive safety outcomes. The ESR1 PvuII polymorphism was also identified as a risk factor in two studies ( Woo et al. 2010 , Ozsoy et al. 2016 ); ESR1 is known to be involved in oestrogen signalling. While there is a current lack of research on their role in PDM, ESR1 in particular has been investigated in relation to endometriosis. Studies have demonstrated higher oestrogen receptor expression in endometriosis lesions of women with moderate-to-severe dysmenorrhoea, compared to those with absent or mild pain ( Pluchino et al. 2020 ), potentially demonstrating a pain-specific role of oestrogen in dysmenorrhoea. Given that hormonal anti-oestrogen contraceptives are also used as treatment for PDM, the role of ESR1 should be explored further. Interestingly, however, none of the PDM-focused GWASs identified ESR1 as a significant variant, while endometriosis-focused GWASs have ( Rahmioglu et al. 2023 ). This could be explained by the fact that studies included in this review likely failed to exclude all secondary dysmenorrhoea cases. Inflammatory processes, particularly prostaglandins, have long been considered to be the key pathways involved in dysmenorrhoea, particularly PDM ( Iacovides et al. 2015 ). Studies examining the inflammatory profiles of the menstrual effluent have demonstrated higher levels of both PGE2 (though not a consistent finding) and PGF2α in those with dysmenorrhoea compared to those without, in addition to other inflammatory mediators, including leukotrienes and other epoxy eicosatetraenoic acids ( Powell et al. 1985 , Bieglmayer et al. 1995 ). Interestingly, adolescents with dysmenorrhoea using NSAIDs had higher levels of PGF2α than those who did not use this medication, and this finding needs a more detailed investigation to better inform treatment paradigms ( Kyathanahalli et al. 2025 ). Both IL1 and TNF cytokines stimulate prostaglandin production ( Hertelendy et al. 2001 , Barcikowska et al. 2020 ), as well as other inflammatory mediators and nitric oxide ( Bradley 2008 , Gabay et al. 2010 ). Therefore, a genetic association is biologically plausible. However, inflammation is also a key feature of SDM-related conditions, such as endometriosis and adenomyosis. It is important to note that GWASs identify common genetic variants that do not have high penetrance in the population, meaning that the identified SNPs tend to have small effect sizes. For example, Jones et al. (2016) reported that the likelihood of PDM increased by 0.1 for each level of pain severity on a four-point scale for their NGF SNP. PDM is likely to be polygenic, and much more research is required before targeted interventions can be developed. While this review was comprehensively conducted, several limitations remain. No grey literature and only English language studies were included, which may mean that some literature was missed. Sample sizes were relatively small in candidate gene studies (ranging from 102 to 435 participants), which may have limited the power needed to detect a significant association. Furthermore, five Turkish studies may have been conducted in the same clinical cohort; therefore, arguably, findings should have been subjected to a Bonferroni correction. Few GWASs have been conducted, and it is possible that these also required greater statistical power to detect associations. Studies were generally of low to medium quality. In addition, few ethnic groups were represented. Importantly, there was evident inconsistency in the definition of PDM cases. Without clear exclusion of secondary causes, it is impossible to firmly assert that findings are associated with PDM rather than the presence of underlying pathology. As non-invasive diagnostic procedures for endometriosis improve, this challenge could be reduced for new cohorts, in which previous findings can then be replicated.

Dysmenorrhoea (period pain) is the most common gynaecological condition in women of reproductive age ( Iacovides et al. 2015 ). It is characterised by cyclical menstrual pain, which may be accompanied by headaches, nausea, vomiting, diarrhoea, bloating or fatigue ( Iacovides et al. 2015 ). Prevalence rates for dysmenorrhoea range from 16 to 91% ( Ju et al. 2014 ) or from 2 to 36% for severe pain ( Söderman et al. 2019 ). It has a substantial impact on school or university attendance, social activities and quality of life ( Söderman et al. 2019 ) as well as significant economic consequences ( Ju et al. 2014 ). However, despite its high prevalence and personal and societal impacts, dysmenorrhoea is generally understudied, and the aetiology is not yet fully understood. Risk factors for dysmenorrhoea include smoking, longer bleeding duration, heavy menstrual flow and high stress levels ( Tavallaee et al. 2011 , Muluneh et al. 2018 ). Family history (FHx) has also been recognised as an important risk factor ( Ozerdogan et al. 2009 , Iacovides et al. 2015 , Muluneh et al. 2018 ). This is usually defined as having a first-degree relative (mother, sister) with dysmenorrhoea and can increase the risk of dysmenorrhoea 3.5-fold compared to individuals with no FHx ( Ozerdogan et al. 2009 ). Women with FHx of dysmenorrhoea also tend to experience more severe pain ( Tavallaee et al. 2011 ). This could be due to shared environments and learned behaviours regarding pain ( Hu et al. 2020 ), but a genetic component is also likely, considering twin studies have previously estimated heritability to be 38% for menstrual pain ( Silberg et al. 1987 ). It is important to acknowledge that primary dysmenorrhoea (PDM) (no underlying pelvic pathology; comprising 90% of cases) ( Gutman et al. 2022 ) is likely to have distinct risk factors and pathogenesis to secondary dysmenorrhoea (SDM) (pain associated with an underlying disease, for example, endometriosis, adenomyosis, fibroids or pelvic inflammatory disease) ( Durand et al. 2021 ). While a laparoscopy is the only definitive method for confirming the diagnosis of PDM (by excluding SDM causes not visible on imaging), many studies report a PDM diagnosis on the basis of clinical judgement, e.g. using one or a combination of history taking, clinical examination or ultrasonography, potentially skewing the research focused on PDM. Currently, pharmacological treatments for PDM are limited to hormonal therapies (usually aiming to suppress menstruation or prevent thickening of the uterine lining) or non-steroidal anti-inflammatory drugs (NSAIDs). NSAIDs have been shown to be 4.4 times more effective than placebo in treating PDM ( Marjoribanks et al. 2015 ), supporting the theory that overproduction of prostaglandins, particularly PGF2a and PGE2, contributes to the pathogenesis of dysmenorrhoea ( Iacovides et al. 2015 ). However, current treatment strategies do not provide adequate relief for all who experience dysmenorrhoea, and there is an increasing move away from hormone therapies, especially among adolescents and young adults ( Hellström et al. 2019 ). Thus, novel therapies are urgently needed. A better understanding of the genetic determinants of dysmenorrhoea is key to informing the development of new therapies. Advancing knowledge in this area is likely to have a far-reaching impact given the influence of dysmenorrhoea on all domains of life and the increasing evidence that menstrual pain is associated with adverse long-term health outcomes, such as mental health diagnoses ( Zhao et al. 2021 ) and chronic pain ( Reid-Mccann et al. 2025 ). This systematic review aims to summarise the current evidence for determining the associations between gene polymorphisms and PDM to further inform the field.

EH, IWL, MS, NFST and RRM have no competing interests to declare. KV received research grants from NIHR, NIH and EU IMI-2; received consultancy fees from Gedeon Richter, Bayer Healthcare, Reckitts and Gesynta; and was president of the IASP Special Interest Group on Abdominal and Pelvic Pain (until August 2024). KZ received research grants from NIHR, NIH, Gates Foundation, DoD US, EU Horizon, Aspira Labs Inc., Bayer AG, Chemo Research S.L., Proteomics International Pty Ltd and Roche Diagnostics GmbH and received a personal payment from Bayer AG until June 2022 as royalties associated with scientific collaboration between University of Oxford and Bayer. KZ was also an unpaid board member of the World Endometriosis Society until 2023 and is currently an unpaid board member of the World Endometriosis Research Foundation.

KV and MS were involved in conceptualisation and methodology. EH and MS were involved in screening, data extraction and quality assessment. IWL, MS and RRM carried out supervision. EH wrote the original draft of the manuscript. All authors reviewed and edited the manuscript.