Major depression is one of the most common mental disorders, with an estimated lifetime prevalence of 10%, and affects twice as many women as men. Reference Kessler1–Reference Lim, Tam, Lu, Ho, Zhang and Ho3 Various factors are hypothesised to contribute to this gender difference, which is primarily observed during the reproductive phase. Reference Deecher, Andree, Sloan and Schechter4,Reference Hyde and Mezulis5 Among these, fluctuations in sex steroid hormones, particularly those associated with female reproductive transitions, are considered a major biological contributor. Reference Kundakovic and Rocks6–Reference Halbreich11 One such reproductive transition occurs during the perinatal period, which spans from pregnancy until 1 year after birth and is characterised by very pronounced sex steroid fluctuations. Depression during this period can result in adverse impacts not only on the pregnant person but also on the offspring, the partner and the family. Reference Letourneau, Dennis, Benzies, Duffett-Leger, Stewart and Tryphonopoulos12–Reference Slomian, Honvo, Emonts, Reginster and Bruyère14 Recent data indicate that around 40% of women exhibit depressive symptoms during the perinatal period, with a substantial proportion meeting the diagnostic criteria for depression. Reference Mazza, Avallone, Kotzalidis, Marano, Moccia and Serio15
Perinatal depression
Over the past few years, depression during the perinatal period has become the most frequent complication of childbirth, with an estimated prevalence of 17%. Reference Hahn-Holbrook, Cornwell-Hinrichs and Anaya16–Reference Shorey, Chee, Ng, Chan, Tam and Chong18 Despite a growing body of evidence, it is still debated whether depression during this period is sufficiently distinct from major depression to be defined as a separate diagnosis. Reference Di Florio and Meltzer-Brody19,Reference Johann and Ehlert20 However, this discrepancy can largely be attributed to varying definitions, including the timing of onset, which can range up to 1 year after delivery. Reference Batt, Duffy, Novick, Metcalf and Epperson21 For instance, the DSM-5 uses the specifier ‘with peripartum onset’ for a major depressive episode that is diagnosed either during pregnancy or within the first 4 weeks after delivery. Reference American Psychiatric Association22 In comparison, the definition by the ICD-11 covers the period from pregnancy up to 6 weeks postpartum. 23 Moreover, definitions with even broader time ranges have been used, covering pregnancy and up to 1 year postpartum. Reference O’Hara and Wisner24 To accommodate the range of definitions used in the existing literature, this review refers to perinatal depression as depressive episodes that occur during pregnancy and up to 12 months postpartum.
Aetiology of perinatal depression
Although a conclusive aetiological model for perinatal depression is still lacking, it is widely recognised that biological and psychosocial factors, and interactions thereof, are involved in the pathophysiological mechanisms. Reference Agrati and Lonstein25–Reference Payne and Maguire27 Given the overlap between fluctuations in sex steroid hormones and an increased incidence of depressive symptoms, the involvement of oestrogens and progesterone has been investigated in more detail. So far, empirical evidence has implicated progesterone, particularly its neuroactive metabolite allopregnanolone (a modulator of GABAA receptors), in perinatal depression symptoms. Reference Yu, Liang, Chen, Li, Han and Chen28 Moreover, two neuroactive steroid GABAA modulators, brexanolone and zuranolone, seem to be effective in the treatment of perinatal depression symptoms. Reference Barnes, Vogl and Nelson29,Reference Powell, Garland, Preston and Piszczatoski30 In contrast, the empirical evidence on oestrogens has been inconclusive. Reference Yim, Tanner Stapleton, Guardino, Hahn-Holbrook and Dunkel Schetter31 While findings on the association between perinatal depression symptoms and oestrogen levels are contradictory, Reference Bloch, Daly and Rubinow32,Reference Schiller, Meltzer-Brody and Rubinow33 some studies have identified an increased sensitivity to oestrogen in women with perinatal depression symptoms. Reference Bloch, Schmidt, Danaceau, Murphy, Nieman and Rubinow34–Reference Soares and Zitek36 Therefore, a potential mediating role of oestrogen sensitivity in the aetiology of the disorder has been suggested, which warrants further investigation.
Oestrogen
Oestrogens are known to regulate reproductive functions, but are further involved in numerous other functions in both females and males Reference Nilsson and Gustafsson37 since their synthesis and actions are not restricted to the gonads but also occur in various extragonadal tissues. Reference Vrtačnik, Ostanek, Mencej-Bedrač and Marc38,Reference Cui, Shen and Li39 In the brain, oestrogens are involved in molecular processes across several neurotransmitter systems that affect emotional and cognitive functions implicated in various mental disorders. Reference Hwang, Lee, Kim and Kwon40,Reference Krolick, Zhu and Shi41 Oestrogens can interact with the dopaminergic, serotonergic, GABAergic and glutaminergic pathways by regulating neurotransmitter synthesis, release and receptor interactions. Reference Barth, Villringer and Sacher42,Reference Del Río, Alliende, Molina, Serrano, Molina and Vigil43 However, oestrogens are largely bound to transport proteins, with only around 2–3% being available in the unbound state with the capacity to induce an effect in target tissues. Reference Gruber, Tschugguel, Schneeberger and Huber44
Oestrogen receptors
Oestrogens exert their effects by binding to three oestrogen receptors: oestrogen receptor-α, oestrogen receptor-β and the G protein-coupled oestrogen receptor (GPER). Oestrogen receptors mediate the emotional and cognitive effects of oestrogen, as they are distributed in various brain regions involved in these functions. Reference Osterlund and Hurd45,Reference Hazell, Yao, Roper, Prossnitz, O’Carroll and Lolait46 The receptors consist of distinct protein structures that provide amino acid sequences to encode receptor functions. Reference Ascenzi, Bocedi and Marino47–Reference Enmark and Gustafsson49 Binding to oestrogen receptors initiates cell type-specific transcriptional processes and signalling mechanisms through both genomic and non-genomic pathways, which ultimately regulate gene expression. Reference Fuentes and Silveyra50,Reference Marino, Galluzzo and Ascenzi51 The efficiency of oestrogen-mediated signalling and the associated biological and behavioural outcomes are regulated by the individual sensitivity to oestrogen. Reference Champagne and Curley52 However, oestrogen sensitivity depends on the number and activity of oestrogen receptors, which can be determined at different levels. Reference Hua, Zhang, Kong and Jiang53
Oestrogen receptor genes
The oestrogen receptors are encoded by three distinct genes, consisting of ESR1 encoding oestrogen receptor-α, ESR2 encoding oestrogen receptor-β and GPER encoding GPER, which are important determinants of oestrogen sensitivity. Reference Figtree, Noonan, Bhindi and Collins54 So far, several genetic variations in these genes have been identified, including single nucleotide polymorphisms (SNPs), restriction fragment length polymorphisms (RFLPs), short tandem repeats (STRs) and haplotype combinations. Depending on the sequence location, genetic variations have the potential to differentially modify gene expression levels and the structure and function of the encoded receptors. Reference Maney55 Furthermore, epigenetic modifications, such as DNA methylation, can alter the accessibility of the DNA without changing the DNA sequence itself, thereby mediating the transcriptional activity of the gene. Reference Attwood, Yung and Richardson56 Research suggests that genetic variations, epigenetic modifications and varying gene expression levels of oestrogen receptor genes may lead to inter-individual differences in the receptors’ sensitivity to oestrogen, which may subsequently increase susceptibility to perinatal depression symptoms. Reference McEvoy, Osborne, Nanavati and Payne57–Reference Luo, Zhu, Du, Chen and Cheng59
Scope of review
Previous reviews and meta-analyses have provided evidence that oestrogen receptor gene variants, including rs2234693 and rs9340799, may be involved in the aetiology of major depression in women. Reference Li, Xie, Wang, Liu, Liao and Yin60,Reference Ryan and Ancelin61 However, these approaches did not strictly distinguish between major depression and depression that occurs during reproductive transitions. Although each of the reproductive transition phases is marked by changes in sex steroid hormones, they exhibit distinct patterns of hormonal fluctuations. Reference Miller, Girgis and Gupta7,Reference Payne, Palmer and Joffe9,Reference Serrano and Warnock10,Reference Soares and Zitek36 The most pronounced fluctuations occur during the perinatal period, and there is growing evidence suggesting that depression during this phase is distinguishable from major depression. Reference Kundakovic and Rocks6,Reference Di Florio and Meltzer-Brody19–Reference Batt, Duffy, Novick, Metcalf and Epperson21 Considering that genetic variations, epigenetic modifications and varying gene expression levels of oestrogen receptor genes may lead to inter-individual differences in the sensitivity of the encoded receptor to oestrogen, their involvement in the aetiology of perinatal depression symptoms has been suggested. Reference McEvoy, Osborne, Nanavati and Payne57–Reference Luo, Zhu, Du, Chen and Cheng59 Therefore, this systematic review aims to provide an overview of the research on the association between oestrogen receptor genes and perinatal depression symptoms by including genetic, epigenetic and gene expression studies.
Method
This systematic review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Reference Page, McKenzie, Bossuyt, Boutron, Hoffmann and Mulrow62 The review protocol was registered with the International Prospective Register of Systematic Reviews before data collection (PROSPERO; registration no. CRD42023447446). Ethics approval was not required for this systematic review.
Search strategy
An initial systematic literature search was conducted on 21 August 2023. Records were retrieved from three electronic databases: PubMed, PsycINFO and Web of Science. All data were obtained from published studies, with no restrictions on publication year or language. Potentially eligible studies published in languages other than English were translated into English and assessments were conducted using the translated version. A two-component search string with the Boolean operator ‘OR’ was used: (a) oestrogen receptor synonyms and components, including (epi-)genetics and gene expression, and (b) perinatal depression, including synonyms (see Supplementary Table S1 available at https://doi.org/10.1192/bjp.2025.39). A supplementary manual review of reference lists was performed to identify additional publications. The literature search was updated on 6 December 2024.
Study selection
After removal of duplicates, the remaining records were screened for eligibility using titles and abstracts. To be included, records had to meet the following criteria: (a) original research, (b) human participants and (c) female participants who were (d) pregnant or within the first year postpartum. Notably, the aim of employing a broad period, encompassing both pregnancy and the 12-month period following childbirth, was to accommodate the varying definitions found in the literature and thus to provide a comprehensive overview. The full text of records meeting these criteria was reviewed for final eligibility. The following exclusion criteria were applied: (a) no assessment of depression and/or depressive symptoms, (b) no assessment of genetic variants, epigenetic modifications or gene expression levels of oestrogen receptor genes, (c) focus on child outcomes, (d) focus on gestational/birth factors and (e) same sample. The screening was performed by two independent reviewers and any disagreements were resolved by consensus.
Data extraction
Data extraction was performed according to the guidelines suggested by Sagoo et al. Reference Sagoo, Little and Higgins63 The first data extraction took place on 28 August 2023. Two independent reviewers extracted the data using a predetermined form and categorised them according to genetic, epigenetic and gene expression studies. The genetic studies were further categorised according to two categories: (a) candidate gene studies and (b) genome-wide association studies (GWASs). The candidate gene studies were further classified according to the three oestrogen receptor genes. We extracted information on the first author, year of publication, country of origin, study design, sample characteristics (sample size, mean age, ethnicity/race), method of analysis, diagnostic criteria, time of assessment and main findings. In addition, we extracted information on genotyped SNPs and nucleotides for candidate gene studies. For epigenetic and gene expression studies, we additionally extracted information on the type of epigenetic marker or receptor gene, respectively, as well as on the specimen. Because of the limited number of included studies investigating each predictor variable (most were only investigated once) and the methodological heterogeneity of the studies, encompassing different outcome variables, diverse study designs and different measurement approaches, Reference Little and Higgins64,Reference Valentine, Pigott and Rothstein65 we did not perform a meta-analysis, and data were analysed narratively.
Quality appraisal
To assess risk of bias, we used a checklist developed by Elwood et al Reference Elwood, Murray, Bell, Sinclair, Kernohan and Stockdale58 in accordance with the guidelines of the HuGE Review Handbook. Reference Little and Higgins64 The checklist included the following items: a clearly stated objective and hypothesis, clear eligibility criteria for participants, information bias in genotyping, selection bias, information bias in assessment of environmental factors, information bias in assessment of depression, restrictions on the ethnicity of participants, replicable statistical methods, assessment of Hardy–Weinberg equilibrium (HWE), comprehensive descriptive data and genotype described. Each criterion on the checklist was rated as ‘Yes’ (study satisfied the criteria), ‘No’ (study did not satisfy the criteria) or ‘NA’ (information not available in the study). Only the criterion information bias in genotyping was rated additionally with ‘QC’ (details given of quality control procedures). To assess the quality of the evidence for each included study, the Grading of Recommendation, Assessment, Development and Evaluation (GRADE) criteria were applied. Reference Schünemann, Oxman, Brozek, Glasziou, Jaeschke and Vist66 The initial confidence level was defined according to the study design, with observational studies starting at low confidence. The confidence level was raised by a number of factors, including the use of clinical ratings for perinatal depression symptoms, the inclusion of confounders, the exclusion of psychiatric comorbidities, the use of correction for multiple testing and a large sample size (n ≥ 1000). The confidence level was lowered by factors such as failure to account for confounders, lack of correction for multiple testing, small (n ≤ 100) or very small (n ≤ 50) sample sizes and indirectness of assessment criteria for perinatal depression. Based on these factors, ‘very low’, ‘low’, ‘moderate’ and ‘high’ estimates (−2, −1, 0 and 1, respectively) were assigned. The final confidence level was adjusted by one or two categories if the difference between the number of factors raising or lowering confidence was two or higher or four or higher, respectively. Quality assessment was performed independently by two reviewers, and any disagreements were resolved by consensus.
Results
Study selection
The initial literature search identified a total of 2093 records, from which 561 duplicates were removed. Following an initial screening of titles and abstracts, 1220 articles that did not meet the inclusion criteria were excluded. Reasons for exclusion at this stage encompassed articles that did not cover any original research (e.g. Fernandez et al Reference Fernandez, Grizzell and Wecker67 ), studies that investigated animal samples (e.g. Furuta et al Reference Furuta, Numakawa, Chiba, Ninomiya, Kajiyama and Adachi68 ) and studies with a female sample analysed outside of the perinatal period (e.g. Mehta et al Reference Mehta, Rex-Haffner, Søndergaard, Pinborg, Binder and Frokjaer35 ). Subsequently, a total of 287 articles were evaluated for full-text eligibility. Of these, 258 were removed based on the exclusion criteria. Notably, the reasons for exclusion encompassed lack of assessment of depression and/or depressive symptoms (e.g. Thippeswamy et al Reference Thippeswamy, Paul, Purushottam, Philip, Jain and Chandra69 ), lack of assessment of genetic variants, epigenetic modifications or gene expression levels of oestrogen receptor genes (e.g. Figueira et al Reference Figueira, Malloy-Diniz, Campos, Miranda, Romano-Silva and Marco70 ), a focus on child outcomes (e.g. Wikenius et al Reference Wikenius, Myhre, Page, Moe, Smith and Heiervang71 ), a focus on gestational/birth factors (e.g. Pellicano et al Reference Pellicano, Carola, Bussone, Cecchini, Tambelli and Lai72 ) and the investigation of the same sample (e.g. Nguyen et al Reference Nguyen, Harder, Xiong, Kowalec, Hägg and Cai73 ). Furthermore, GWASs conducting combined analyses of multiple cohorts without reporting individual cohort findings separately were excluded (e.g. Guintivano et al Reference Guintivano, Byrne, Kiewa, Yao, Bauer and Aberg74 ). Although such studies have the potential to increase statistical power, their findings are limited by methodological variability (e.g. heterogeneous definitions of perinatal depression across the cohorts). Therefore, a total of 29 studies were ultimately included in the systematic review. The studies were categorised according to genetic studies (n = 16), epigenetic studies (n = 5) and gene expression studies (n = 8). The genetic studies were further divided into two categories: (a) candidate gene studies (n = 9) and (b) GWASs (n = 7). Since no candidate gene study was found to investigate the association between GPER and perinatal depression symptoms, the studies were further divided into two categories: (a) studies examining genetic variations in ESR1 (n = 8) and (b) studies focusing on genetic variations in ESR2 (n = 6). The study characteristics and results of the included studies were described within these categories (see Supplementary Tables Tables S4–8). Figure 1 summarises the selection process based on the PRISMA guidelines.

Fig. 1 PRISMA flow diagram of the studies included in the review. ESR1, oestrogen receptor-α gene; ESR2, oestrogen receptor-β gene; GPER, G protein-coupled oestrogen receptor gene; GWAS, genome-wide association study.
Study characteristics
The 29 included studies were published between 2010 and 2024. The sample sizes ranged from 14 to 226 707. Seventeen studies conducted a genome-wide approach, while 12 studies employed a targeted approach. Nine genetic studies conducted a candidate gene approach and genotyped and tested a total of 62 genetic variations in oestrogen receptor genes, comprising 61 SNPs, of which four were RFLPs, and one STR (see Fig. 2). In addition, four haplotype combinations were tested for an association with perinatal depression symptoms. Following the recommendations of Takezawa et al Reference Takezawa, Kato, Oota, Caulfield, Fujimoto and Honda75 we avoid broad, arbitrary groupings and therefore report self-reported ethnicities as stated in the original studies to ensure an accurate description of the participants. Three studies examined Asian samples, four examined White samples, ten examined mixed-ethnicity samples, two examined European samples, one examined a Spanish sample, one examined a Japanese sample and eight studies did not provide sufficient information on ethnicity. Twenty-three studies employed a case–control design, and six studies adopted a cohort design. Eleven of the case–control studies collected longitudinal data: six collected data during pregnancy and the postpartum period and the other five collected data only during the postpartum period, spanning from 2–3 days to 12 months postpartum. Reference Costas, Gratacòs, Escaramís, Martín-Santos, Diego and Baca-García76–Reference Osborne, Payne, Sherer and Sabunciyan86 Four of the case–control studies used a cross-sectional design, collecting data within the first 12 weeks postpartum. Reference El-Ibiary, Hamilton, Abel, Erdman, Robertson and Finley87–Reference Pan, Xu, Zhang, Su, Chen and Li90 The remaining eight case–control studies did not report the specific assessment time point during the perinatal period. Reference Jiang, Wei and Xie91–Reference Byrne, Carrillo-Roa, Penninx, Sallis, Viktorin and Chapman98 The six cohort studies collected longitudinal data during pregnancy and postpartum, ranging from 14 weeks of pregnancy to 9 months postpartum. Reference McClain, Farrell, LaSorda, Pan, Peters and Lim99–Reference Pouget, Taylor, Dennis, Grigoriadis, Oberlander and Frey104 The included studies applied various self-report instruments and clinical interviews to assess perinatal depression symptoms. A total of seven different self-report instruments were used; the most frequently administered self-report instrument was the Edinburgh Postnatal Depression Scale (EPDS). Reference Cox, Holden and Sagovsky105 The EPDS was employed in 22 studies, of which 16 reported a cut-off score, ranging from 8 to 20. Reference Costas, Gratacòs, Escaramís, Martín-Santos, Diego and Baca-García76,Reference Zou, Zhang, Xie, Chen and Xia78–Reference Mehta, Newport, Frishman, Kraus, Rex-Haffner and Ritchie80,Reference Guleria, Suneja, Sharma, Banerjee, Bhatia and Chaudhary82,Reference Nakamura, Nakatochi, Kunimoto, Okada, Aleksic and Toyama83,Reference Segman, Goltser-Dubner, Weiner, Canetti, Galili-Weisstub and Milwidsky85–Reference Pan, Xu, Zhang, Su, Chen and Li90,Reference Tan, Lim, Chua, Tan, Lee and Chen93,Reference Tang, Liu, Chen and Wang95–Reference Lapato, Roberson-Nay, Kirkpatrick, Webb, York and Kinser102,Reference Pouget, Taylor, Dennis, Grigoriadis, Oberlander and Frey104

Fig. 2 The genomic structure and approximate locations of the genetic variations in (a) ESR1, (b) ESR2 and (c) GPER investigated in the included candidate gene studies. ESR1, oestrogen receptor-α gene; ESR2, oestrogen receptor-β gene; GPER, G protein-coupled oestrogen receptor gene; SNPs, single nucleotide polymorphisms. Blue color indicates significant findings. *Not significant after correction for multiple testing.
Critical appraisal of included studies
For an overview of the risk of bias assessment, please refer to Supplementary Fig. S1 and Table S2. None of the included studies indicated a high chance of bias for the criteria ‘replicable statistical methods’ and ‘comprehensive descriptive data’. However, most of the studies did not provide information on the assessment of environmental factors, indicating a high chance of bias for this criterion. An overview of the quality of evidence assessment is provided in Supplementary Table S3. In summary, most of the studies were attributed with a low (48.3%) level of confidence, followed by very low (27.6%) and moderate (24.1%). No study achieved a high (0.0%) level of confidence. It should be noted that an initial low confidence level was assigned to observational studies, which affected all of the included studies.
Genetic studies on oestrogen receptor gene 1
Eight candidate gene studies investigated the association between genetic variations in ESR1 and perinatal depression symptoms. Reference Costas, Gratacòs, Escaramís, Martín-Santos, Diego and Baca-García76–Reference Zou, Zhang, Xie, Chen and Xia78,Reference El-Ibiary, Hamilton, Abel, Erdman, Robertson and Finley87,Reference Pinsonneault, Sullivan, Sadee, Soares, Hampson and Steiner88,Reference Tan, Lim, Chua, Tan, Lee and Chen93,Reference Zhang, Chen, Gong and Huang94,Reference Schneider, Pretscher, Goecke, Häberle, Engel and Kornhuber100 These studies genotyped and tested a total of 45 genetic variations, including 44 SNPs, of which two were RFLPs and one was a STR. The most frequently investigated genetic variation was the SNP rs2077647, which was analysed in four studies. The second most frequently investigated variations (each analysed in three different studies) were the SNPs rs9340958 and rs3020434 and the RFLPs rs2234693, also known as the Pvu-ll variant, and rs9340799, also known as the Xba-l variant. The third most frequently investigated variants were the SNPs rs9341052, rs3798577 and rs1801132, and one STR, the thymine-adenine repeat, which were each analysed in two studies. The remaining genetic variations were only tested once. Furthermore, three studies analysed additional haplotype combinations. A significant linkage disequilibrium was found between the haplotype combinations of rs2077647 and rs9340799, rs2077647 and rs2234693 and rs2234693 and rs9340799, but the haplotypes did not show a significant association with perinatal depression symptoms. Reference Tan, Lim, Chua, Tan, Lee and Chen93 Costas et al Reference Costas, Gratacòs, Escaramís, Martín-Santos, Diego and Baca-García76 found that the haplotype combination of rs9340954 and rs3020434 showed the strongest significant association with perinatal depression symptoms, although it did not withstand correction for multiple testing. Another study found that the nine investigated SNPs defined approximately six haplotype combinations, but they were not tested for an association with perinatal depression symptoms. Reference Pinsonneault, Sullivan, Sadee, Soares, Hampson and Steiner88 One study also tested whether the joint genotype distribution of rs2234693 and rs9340799 was associated with perinatal depression symptoms, but no significant association emerged. Reference Zou, Zhang, Xie, Chen and Xia78 Moreover, one study investigated potential gene–gene interactions with oestrogen receptor genes in perinatal depression symptoms, of which three interactions with the thymine-adenine repeat were found to be significant, specifically the dopamine transporter gene (SLC6A3) and two genes involved in serotonergic signalling (SLC6A4 and HTR2A). Reference Pinsonneault, Sullivan, Sadee, Soares, Hampson and Steiner88 The most significant interaction was found for the thymine-adenine repeat and the SERTLPR, although the data were not adjusted for multiple testing. Reference Pinsonneault, Sullivan, Sadee, Soares, Hampson and Steiner88
Genetic studies on oestrogen receptor gene 2
Six candidate gene studies investigated the association between genetic variations in ESR2 and perinatal depression symptoms. Reference Costas, Gratacòs, Escaramís, Martín-Santos, Diego and Baca-García76,Reference He, Hu, Zhou and Li77,Reference El-Ibiary, Hamilton, Abel, Erdman, Robertson and Finley87,Reference Tan, Lim, Chua, Tan, Lee and Chen93,Reference Zhang, Chen, Gong and Huang94,Reference McClain, Farrell, LaSorda, Pan, Peters and Lim99 The studies genotyped and tested a total of 17 SNPs, including two RFLPs. The most frequently genotyped and tested genetic variation was the RFLP rs4986938, also known as the Alu-l variant, which was examined in three studies. The second most frequently analysed genetic variation was the RFLP rs1256049, also known as the Rsa-l variant, which was examined in two studies. The remaining SNPs were genotyped and tested only once, and no haplotype combinations were constructed and tested. Notably, only one study found an association between genetic variations in ESR2 and perinatal depression symptoms. Specifically, the RFLP rs4986938 was associated with perinatal depression symptoms, although the results were not adjusted for multiple testing. Reference He, Hu, Zhou and Li77
Genome-wide association studies
Seven genetic studies conducted a genome-wide approach to investigate associations with perinatal depression symptoms. Reference Tebeka, Gloaguen, Mullaert, He, Boland and Deleuze84,Reference Jiang, Wei and Xie91,Reference Kiewa, Mortlock, Meltzer-Brody, Middeldorp, Wray and Byrne92,Reference Kiewa, Meltzer-Brody, Milgrom, Guintivano, Hickie and Whiteman96–Reference Byrne, Carrillo-Roa, Penninx, Sallis, Viktorin and Chapman98,Reference Pouget, Taylor, Dennis, Grigoriadis, Oberlander and Frey104 The studies examined a total of 12 distinct cohorts. A genome-wide significance for oestrogen receptor genes was reached in one study. Reference Jiang, Wei and Xie91 In detail, the SNP rs2347923 located in Intron 3 of ESR1 was found to be associated with perinatal depression symptoms. Moreover, the SNP rs851981, located in the 5’ untranslated region (UTR) of ESR1, was associated with perinatal depression symptoms after genome-wide significant loci for a latent reproductive factor, including reproductive disorders such as endometriosis, were identified and filtered for nominal significance (p ≤ 0.05) in perinatal depression symptoms. Reference Kiewa, Mortlock, Meltzer-Brody, Middeldorp, Wray and Byrne92
Gene expression studies
Eight studies investigated the association between gene expression levels of oestrogen receptor genes and perinatal depression symptoms. Reference Mehta, Grewen, Pearson, Wani, Wallace and Henders79,Reference Mehta, Newport, Frishman, Kraus, Rex-Haffner and Ritchie80,Reference Guleria, Suneja, Sharma, Banerjee, Bhatia and Chaudhary82,Reference Segman, Goltser-Dubner, Weiner, Canetti, Galili-Weisstub and Milwidsky85,Reference Osborne, Payne, Sherer and Sabunciyan86,Reference Guintivano, Aberg, Clark, Rubinow, Sullivan and Meltzer-Brody89,Reference Pan, Xu, Zhang, Su, Chen and Li90,Reference Redei, Ciolino, Wert, Yang, Kim and Clark103 One of them analysed mRNA levels of ESR1, Reference Guleria, Suneja, Sharma, Banerjee, Bhatia and Chaudhary82 one analysed mRNA levels of ESR2 Reference Redei, Ciolino, Wert, Yang, Kim and Clark103 and five employed a genome-wide approach for mRNA analysis. Reference Mehta, Grewen, Pearson, Wani, Wallace and Henders79,Reference Mehta, Newport, Frishman, Kraus, Rex-Haffner and Ritchie80,Reference Segman, Goltser-Dubner, Weiner, Canetti, Galili-Weisstub and Milwidsky85,Reference Osborne, Payne, Sherer and Sabunciyan86,Reference Guintivano, Aberg, Clark, Rubinow, Sullivan and Meltzer-Brody89,Reference Pan, Xu, Zhang, Su, Chen and Li90 Four studies employed peripheral blood samples to determine mRNA levels, Reference Mehta, Grewen, Pearson, Wani, Wallace and Henders79,Reference Mehta, Newport, Frishman, Kraus, Rex-Haffner and Ritchie80,Reference Guintivano, Aberg, Clark, Rubinow, Sullivan and Meltzer-Brody89,Reference Redei, Ciolino, Wert, Yang, Kim and Clark103 two studies investigated peripheral blood mononuclear cells (PBMCs), Reference Segman, Goltser-Dubner, Weiner, Canetti, Galili-Weisstub and Milwidsky85,Reference Pan, Xu, Zhang, Su, Chen and Li90 one study analysed extracellular vesicles in peripheral blood Reference Osborne, Payne, Sherer and Sabunciyan86 and one study used maternal placental tissue samples. Reference Guleria, Suneja, Sharma, Banerjee, Bhatia and Chaudhary82 Five of the eight studies found significant associations between gene expression levels of oestrogen receptor genes and perinatal depression symptoms. In the case of mRNA levels of ESR1, women with perinatal depression symptoms exhibited a downregulation in maternal placental tissue and PBMCs, as well as an enrichment in peripheral blood, compared to healthy controls. Reference Mehta, Newport, Frishman, Kraus, Rex-Haffner and Ritchie80,Reference Guleria, Suneja, Sharma, Banerjee, Bhatia and Chaudhary82,Reference Pan, Xu, Zhang, Su, Chen and Li90 Furthermore, transcript levels of ESR1 were found to be downregulated in granulocytes and upregulated in B-cells in women with perinatal depression symptoms, but this did not withstand correction for multiple testing. Reference Guintivano, Aberg, Clark, Rubinow, Sullivan and Meltzer-Brody89 For mRNA levels of ESR2, higher levels in peripheral blood were associated with increased perinatal depression symptoms. Reference Redei, Ciolino, Wert, Yang, Kim and Clark103 Moreover, transcript levels of ESR2 were found to be downregulated in peripheral blood and monocytes and upregulated in CD8± T-cells in women with perinatal depression symptoms, but this did not remain significant after correction for multiple testing. Reference Guintivano, Aberg, Clark, Rubinow, Sullivan and Meltzer-Brody89
Epigenetic studies
Five studies investigated the association between epigenetic modifications of oestrogen receptor genes and perinatal depression symptoms. Reference Guintivano, Arad, Gould, Payne and Kaminsky81,Reference Nakamura, Nakatochi, Kunimoto, Okada, Aleksic and Toyama83,Reference Tang, Liu, Chen and Wang95,Reference Lund, Kyläniemi, Pettersson, Kaukonen, Konki and Scheinin101,Reference Lapato, Roberson-Nay, Kirkpatrick, Webb, York and Kinser102 Four of them analysed DNA methylation Reference Guintivano, Arad, Gould, Payne and Kaminsky81,Reference Nakamura, Nakatochi, Kunimoto, Okada, Aleksic and Toyama83,Reference Tang, Liu, Chen and Wang95,Reference Lund, Kyläniemi, Pettersson, Kaukonen, Konki and Scheinin101,Reference Lapato, Roberson-Nay, Kirkpatrick, Webb, York and Kinser102 while one study evaluated microRNAs. Reference Tang, Liu, Chen and Wang95 Three studies investigated peripheral blood samples, Reference Guintivano, Arad, Gould, Payne and Kaminsky81,Reference Nakamura, Nakatochi, Kunimoto, Okada, Aleksic and Toyama83,Reference Lapato, Roberson-Nay, Kirkpatrick, Webb, York and Kinser102 one study examined extracellular vesicles in peripheral blood Reference Tang, Liu, Chen and Wang95 and one study analysed maternal placental tissue samples. Reference Lund, Kyläniemi, Pettersson, Kaukonen, Konki and Scheinin101 All studies investigating DNA methylation employed an epigenome-wide association analysis (EWAS). Reference Guintivano, Arad, Gould, Payne and Kaminsky81,Reference Nakamura, Nakatochi, Kunimoto, Okada, Aleksic and Toyama83,Reference Lund, Kyläniemi, Pettersson, Kaukonen, Konki and Scheinin101,Reference Lapato, Roberson-Nay, Kirkpatrick, Webb, York and Kinser102 In contrast, the study evaluating microRNAs employed a candidate approach. Reference Tang, Liu, Chen and Wang95 Three of the EWAS studies investigated the methylation levels of individual CpG sites, Reference Guintivano, Arad, Gould, Payne and Kaminsky81,Reference Nakamura, Nakatochi, Kunimoto, Okada, Aleksic and Toyama83,Reference Lund, Kyläniemi, Pettersson, Kaukonen, Konki and Scheinin101 whereas one study analysed differentially methylated regions. Reference Lapato, Roberson-Nay, Kirkpatrick, Webb, York and Kinser102 Three of the four EWAS studies found significant associations between enrichment of DNA methylation marks involved in oestrogen signalling and perinatal depression symptoms. Reference Guintivano, Arad, Gould, Payne and Kaminsky81,Reference Lund, Kyläniemi, Pettersson, Kaukonen, Konki and Scheinin101,Reference Lapato, Roberson-Nay, Kirkpatrick, Webb, York and Kinser102 With regard to microRNAs, significantly upregulated levels of miR-211 and miR-744, which target the 3′ UTR of ESR1 and ESR2 and inhibit expression of these genes in plasma exosomes, were identified in women with perinatal depression symptoms. Reference Tang, Liu, Chen and Wang95
Discussion
Previous research has indicated an involvement of oestrogen receptor genes in the aetiology of depression in women. Notably, the strongest evidence was reported for an association of rs2234693 and rs9340799 with severe depressive symptoms and major depression in women. Reference Li, Xie, Wang, Liu, Liao and Yin60,Reference Ryan and Ancelin61 However, these studies failed to distinguish between major depression and depression during the perinatal period. While all of these phases are characterised by changes in sex steroid hormones, they exhibit distinct patterns of hormonal fluctuations. Reference Miller, Girgis and Gupta7,Reference Payne, Palmer and Joffe9,Reference Serrano and Warnock10,Reference Soares and Zitek36 The most pronounced fluctuations are observed during the perinatal period, and there is accumulating evidence suggesting that depression during this phase is distinguishable from major depression. Reference Kundakovic and Rocks6,Reference Di Florio and Meltzer-Brody19–Reference Batt, Duffy, Novick, Metcalf and Epperson21 Therefore, we provided an overview of the published literature on the association between oestrogen receptor genes and perinatal depression symptoms, by including genetic, epigenetic and gene expression studies. Overall, our review revealed that a limited number of genetic variations in ESR1 and ESR2 were associated with perinatal depression symptoms. Most of the associations were found for variations in ESR1, of which the RFLP rs2234693 emerged as the most frequently reported variant to show a significant association with perinatal depression symptoms. Moreover, epigenetic modifications, including DNA methylation marks involved in oestrogen signalling and gene expression levels of ESR1 and ESR2, were found to be associated with perinatal depression symptoms.
When interpreting the associations between oestrogen receptor gene variants and perinatal depression symptoms, it is important to consider some mediating factors. One such factor is the specific genes and the function of their encoded receptor, since three different oestrogen receptor genes were investigated. We only report findings for ESR1 and ESR2, as we found no candidate gene studies investigating the association between GPER and perinatal depression symptoms and no GWASs with genome-wide significance for variations in GPER. Despite the fact that GPERs are membrane-bound oestrogen receptors, which are expressed in various brain areas and are important mediators of oestrogen’s rapid, non-genomic effects through the activation of various signal transduction pathways, their function in the brain and thus their implications for behaviour are still largely understudied. Reference Marino, Galluzzo and Ascenzi51,Reference Hadjimarkou and Vasudevan106–Reference Sundermann, Maki and Bishop108 Consequently, no conclusions can be drawn regarding the impact of genetic variations in GPERs and their associated rapid signalling pathways, which may be implicated in the aetiology of perinatal depression symptoms. However, our review revealed that ESR1 was more frequently associated with perinatal depression symptoms than ESR2. On the one hand, this finding can be partially explained by the higher number of genetic variations studied in ESR1 compared to ESR2, and thus a potential publication bias. On the other hand, the distinct expression patterns of the encoded receptors indicate a more important role in emotional functions and moods for oestrogen receptor-α encoded by ESR1 than for oestrogen receptor-β encoded by ESR2. Reference Osterlund and Hurd45
In addition to the specific gene and the function of the receptor it encodes, the type of genetic variation must also be considered, with a broad spectrum of genetic variation having already been investigated. Most of the reviewed studies analysed SNPs, including RFLPs, and only a few studies examined STRs and haplotype combinations. Overall, 11 SNPs (including two RFLPs) and one STR were found to be associated with perinatal depression symptoms. Reference Costas, Gratacòs, Escaramís, Martín-Santos, Diego and Baca-García76–Reference Zou, Zhang, Xie, Chen and Xia78,Reference El-Ibiary, Hamilton, Abel, Erdman, Robertson and Finley87,Reference Pinsonneault, Sullivan, Sadee, Soares, Hampson and Steiner88,Reference Jiang, Wei and Xie91,Reference Kiewa, Mortlock, Meltzer-Brody, Middeldorp, Wray and Byrne92 Seven of the SNPs did not remain significant after correction for multiple testing, indicating that the initial associations may be susceptible to false positives. Reference Costas, Gratacòs, Escaramís, Martín-Santos, Diego and Baca-García76,Reference El-Ibiary, Hamilton, Abel, Erdman, Robertson and Finley87,Reference Pinsonneault, Sullivan, Sadee, Soares, Hampson and Steiner88 Regarding RFLPs, one of three studies analysing rs4986938 (Alu-l) and two of three studies analysing rs2234693 (Pvu-ll) found a significant association with perinatal depression symptoms Reference Costas, Gratacòs, Escaramís, Martín-Santos, Diego and Baca-García76–Reference Zou, Zhang, Xie, Chen and Xia78,Reference Tan, Lim, Chua, Tan, Lee and Chen93 However, these findings were not adjusted for multiple testing, thus increasing the likelihood of false positive results. Notably, although these variants are, by definition, also SNPs, they are referred to as RFLPs because of their ability to disrupt the recognition sites of specific restriction enzymes, such as Pvu-II or Alu-I, which cut the DNA at these points. Reference Maney55 Previous research linked both of these RFLPs to a risk of depression in women. Reference Li, Xie, Wang, Liu, Liao and Yin60,Reference Ryan and Ancelin61,Reference Keyes, Agnew-Blais, Roberts, Hamilton, Vivo and Ranu109 However, as these investigations did not strictly distinguish between major depression and depression that occurs during reproductive transitions, for the time being, it is not possible to conclude whether rs4986938 and rs2234693 may represent a shared genetic factor for both major depression and perinatal depression or whether they are specific to perinatal depression symptoms. Regarding the STR thymine-adenine repeat, one study found a significant association with perinatal depression symptoms while another detected a trend for an association. Reference Elwood, Murray, Bell, Sinclair, Kernohan and Stockdale58,Reference Ryan and Ancelin61 Notably, a previous study found that the thymine-adenine repeat was in strong linkage disequilibrium with the RFLP rs2234693. Reference Becherini, Gennari, Masi, Mansani, Massart and Morelli110 Considering linkage disequilibrium, single variation analyses might be insufficient, as some alleles occur at different sites more frequently than would be randomly expected. Reference Slatkin111 For instance, while some variants in linkage disequilibrium are highly correlated, one may serve as a marker, whereas another could directly be implicated in gene expression. Analysing a single variant can therefore miss subtle or combined effects that contribute to a phenotype. Therefore, analyses including haplotypes, a combination of frequently co-occurring alleles on a chromosome, provide more information on further genetic variations. Reference Sundermann, Maki and Bishop108,Reference Clark112 This approach improves the robustness and reliability of results and enhances the power of association tests, particularly when multiple alleles are involved. Reference Clark112,Reference Morris and Kaplan113 However, the reviewed haplotype combinations were not found to be significantly associated with perinatal depression symptoms, Reference Costas, Gratacòs, Escaramís, Martín-Santos, Diego and Baca-García76,Reference Tan, Lim, Chua, Tan, Lee and Chen93 although this finding should be interpreted with caution as only four different haplotype combinations were examined, which were each tested only once. Overall, although various types of genetic variations were analysed across the included candidate gene studies, the majority of the specific variants were only examined in a limited number of studies. Given that genetic variations generally predict only small differences in the phenotype, Reference Cannon and Keller114 a careful selection of the genetic variation to be investigated might be crucial to achieve more stable and reliable results. Moreover, genome-wide approaches may provide evidence for novel risk variants without the bias of prior hypotheses, thereby overcoming limitations of candidate gene approaches and leading to a more comprehensive understanding of the genetic architecture of perinatal depression symptoms. Reference Tam, Patel, Turcotte, Bossé, Paré and Meyre115,Reference Duncan, Ostacher and Ballon116 Notably, both SNPs identified by the included GWASs represent novel risk variants, as they have not been investigated by the included candidate gene studies. Reference Jiang, Wei and Xie91,Reference Kiewa, Mortlock, Meltzer-Brody, Middeldorp, Wray and Byrne92 However, GWASs are often underpowered, since a high level of significance is needed to account for multiple testing. Reference Tam, Patel, Turcotte, Bossé, Paré and Meyre115 This limitation was also evident in the included GWASs, where the majority was underpowered. Therefore, conducting meta-analyses combining different cohorts has gained considerable interest in increasing sample size and thus statistical power. However, although 20 distinct cohorts were investigated in a recent meta-analysis of GWASs of perinatal depression, no SNP achieved genome-wide significance. Reference Guintivano, Byrne, Kiewa, Yao, Bauer and Aberg74 To address the power limitation, Kiewa et al Reference Kiewa, Mortlock, Meltzer-Brody, Middeldorp, Wray and Byrne92 employed another approach, using a two-step analysis. First, they identified genome-wide significant loci for a latent reproductive factor, including reproductive disorders such as endometriosis. Second, the loci identified were tested specifically for their association with perinatal depression symptoms, applying a significance threshold of p ≤ 0.05. This approach reduced the multiple testing burden compared to conducting a genome-wide analysis specifically for perinatal depression. In summary, although various genetic variations have been investigated, only a small number of them have been found to be significantly associated with perinatal depression symptoms. Moreover, the majority did not withstand correction for multiple testing or were not adjusted for multiple testing, increasing the risk of false positive results. Initial evidence suggests a more important role for RFLPs such as rs2234693 in perinatal depression symptoms than for other variations in oestrogen receptor genes. However, little is known about their role and the biological pathways that potentially result in perinatal depression symptoms. Reference Sundermann, Maki and Bishop108 Moreover, it is important to note that the observed associations are based on correlations, which precludes any direct causal interpretations regarding the specific biological mechanisms that potentially explain these associations. Consequently, it may also be that these variations do not affect oestrogen signalling directly but are in linkage disequilibrium with another functional variation, which may in turn explain the significant associations with perinatal depression symptoms.
While Mendelian inheritance provides a framework for understanding how single-gene traits are transmitted over generations, the majority of complex disorders, including perinatal depression symptoms, do not follow simple Mendelian patterns and involve multiple factors. Reference Cannon and Keller114 Notably, none of the reviewed studies investigated potential gene–environment interactions with oestrogen receptor genes in perinatal depression symptoms. One study additionally examined the influence of environmental factors and found that a distressed partner relationship, unplanned pregnancy and a self-reported history of depression were predictors of perinatal depression. Reference El-Ibiary, Hamilton, Abel, Erdman, Robertson and Finley87 However, one study examined potential gene–gene interactions with oestrogen receptor genes in perinatal depression symptoms and found that three out of seven potential gene–gene interactions with the thymine-adenine repeat were associated with perinatal depression symptoms, including the dopamine transporter gene (SLC6A3) and two genes involved in serotonergic signalling (HTR2A and SLC6A4). Reference Pinsonneault, Sullivan, Sadee, Soares, Hampson and Steiner88 Although these data were not adjusted for multiple testing, meaning that false positive results cannot be ruled out, the initial evidence suggests that various gene–gene interactions with oestrogen receptor genes may be involved in perinatal depression symptoms.
Besides genetic factors, it is also crucial to consider epigenetic modifications, as they have the potential to regulate gene transcription. Reference Attwood, Yung and Richardson56 Only a limited number of studies (n = 5) were found to examine epigenetic modifications of oestrogen receptor genes in relation to perinatal depression symptoms. The majority investigated these modifications through epigenome-wide DNA methylation analysis. Reference Guintivano, Arad, Gould, Payne and Kaminsky81,Reference Nakamura, Nakatochi, Kunimoto, Okada, Aleksic and Toyama83,Reference Lund, Kyläniemi, Pettersson, Kaukonen, Konki and Scheinin101,Reference Lapato, Roberson-Nay, Kirkpatrick, Webb, York and Kinser102 Notably, none of the reviewed studies investigated methylation levels using a targeted approach. Initial evidence from the reviewed studies suggests that DNA methylation marks in regions involved in oestrogen signalling, rather than in oestrogen receptor genes directly, may be associated with perinatal depression symptoms. However, the ability to directly compare these studies is limited, since there is only a small number of studies, which employed varying methodologies, including the analysis of distinct tissue types. Given the tissue-specific nature of epigenetic modifications, findings from peripheral blood may not be applicable to other tissues. Reference Bakulski, Halladay, Hu, Mill and Fallin117,Reference Smith, Kilaru, Klengel, Mercer, Bradley and Conneely118 Moreover, although one study found an association between microRNAs targeting ESR1 and ESR2 and perinatal depression symptoms, Reference Tang, Liu, Chen and Wang95 the findings were not adjusted for multiple testing, meaning that false positive results cannot be ruled out. Therefore, it is not yet possible to draw definitive conclusions regarding the involvement of epigenetic modifications of oestrogen receptor genes in the aetiology of perinatal depression symptoms. Nevertheless, initial evidence points to an involvement of epigenetic modifications related to oestrogen signalling in perinatal depression symptoms.
Furthermore, gene expression levels also warrant careful consideration, as they may offer crucial information about the functional and biological implications of the genetic variants and epigenetic modifications. While none of the genetic and epigenetic studies investigated whether these (epi-)genetic variations affected gene expression levels, a small number of studies (n = 8) investigated gene expression levels of oestrogen receptor genes in relation to perinatal depression symptoms. Four studies reported an association between mRNA levels of ESR1 and perinatal depression symptoms. Reference Mehta, Newport, Frishman, Kraus, Rex-Haffner and Ritchie80,Reference Guleria, Suneja, Sharma, Banerjee, Bhatia and Chaudhary82,Reference Guintivano, Aberg, Clark, Rubinow, Sullivan and Meltzer-Brody89,Reference Pan, Xu, Zhang, Su, Chen and Li90,Reference Redei, Ciolino, Wert, Yang, Kim and Clark103 However, the findings from one of these studies did not withstand correction for multiple testing, Reference Guintivano, Aberg, Clark, Rubinow, Sullivan and Meltzer-Brody89 and findings from another study were not adjusted for multiple testing, Reference Guleria, Suneja, Sharma, Banerjee, Bhatia and Chaudhary82 indicating a higher risk of false positive results. Similarly for mRNA levels of ESR2, although two studies reported an association with perinatal depression symptoms, one of these did not withstand correction for multiple testing and the other did not adjust for multiple testing. Reference Guintivano, Aberg, Clark, Rubinow, Sullivan and Meltzer-Brody89,Reference Redei, Ciolino, Wert, Yang, Kim and Clark103 Moreover, conflicting results have been reported, with both up- and downregulation of mRNA levels. Given that gene expression levels of oestrogen receptors are known to be tissue specific, Reference Grandien, Berkenstam and Gustafsson119 comparisons between the studies are challenging and interpretations may be limited in terms of generalisability, because of the use of distinct tissue types across the reviewed studies. Furthermore, although blood is suitable for gene expression analysis, variations in collection, storage and extraction may differentially affect transcription profiles. Reference Gautam, Donohue, Hoke, Miller, Srinivasan and Sowe120 Although binding to oestrogen receptors can induce both genomic and non-genomic effects, Reference Björnström and Sjöberg121,Reference Thiebaut, Vlaeminck-Guillem, Trédan, Poulard and Le Romancer122 the studies did not examine the specific biological mechanisms implicated in the manifestation of distinct gene expression levels, meaning that no causal conclusions about the implicated pathways can be drawn. Nevertheless, initial evidence indicates that gene expression levels of ESR1 and ESR2 are associated with perinatal depression symptoms.
In addition, when interpreting the findings of this review, it is essential to take potential confounding factors into account. Notably, 15 studies accounted for confounding variables, most frequently age, followed by genetic ancestry. Although a previous study found no direct association between genetic ancestry and perinatal depression, Reference Guintivano, Sullivan, Stuebe, Penders, Thorp and Rubinow123 it is possible that differences in genetic variations may reflect population-specific differences rather than behavioural differences. This limitation is particularly relevant to the candidate gene approach, since this approach is less effective in accounting for subtle genetic ancestry differences compared to genome-wide approaches. Reference Duncan, Ostacher and Ballon116
The reviewed studies exhibited numerous methodological differences regarding the assessment of perinatal depression symptoms, which may account for some of the inconclusive findings. First, the reviewed studies employed a wide range of assessment time points, spanning from 11–14 weeks of pregnancy to 12 months postpartum. These differences reflect the current inconsistencies regarding the distinction between major depression and perinatal depression, which can largely be attributed to varying definitions of the time of symptom onset and how the postpartum period is defined. Reference Johann and Ehlert20 Research has suggested distinct genetic factors contributing to depression during pregnancy and postpartum, with a greater impact after birth. Reference Figueiredo, Parada, Araujo, Silva and Del-Ben124,Reference Samuelsen, Ystrom, Gjerde and Eilertsen125 However, it was not possible to compare the findings regarding onset during pregnancy and onset postpartum because of the limited number of studies investigating each time period, as well as the heterogeneous study designs. As such, it cannot be ruled out that oestrogen receptor gene variants associated with perinatal depression symptoms differ depending on whether onset occurs during pregnancy or postpartum. Second, the reviewed studies employed a wide range of self-report instruments and clinical interviews to assess perinatal depression symptoms, most of which were only employed in one study and were not specifically designed for use during the perinatal period. Given that some depressive symptoms are associated with common physiological symptoms during the perinatal period, their differentiation may be challenging. Reference Halbreich and Karkun126,Reference Marcus127 However, the most widely used screening tool was the EPDS, which was specifically designed for use during the perinatal period by excluding normative physiological perinatal symptoms associated with depressive symptoms, Reference Cox, Holden and Sagovsky105 although notably, the studies applied various cut-off values for the EPDS, ranging from 8 to 20. Reference Costas, Gratacòs, Escaramís, Martín-Santos, Diego and Baca-García76,Reference Zou, Zhang, Xie, Chen and Xia78,Reference Mehta, Grewen, Pearson, Wani, Wallace and Henders79,Reference Guleria, Suneja, Sharma, Banerjee, Bhatia and Chaudhary82,Reference Nakamura, Nakatochi, Kunimoto, Okada, Aleksic and Toyama83,Reference El-Ibiary, Hamilton, Abel, Erdman, Robertson and Finley87,Reference Pinsonneault, Sullivan, Sadee, Soares, Hampson and Steiner88,Reference Tang, Liu, Chen and Wang95 Although a cut-off value of 11 has been found to maximise both sensitivity and specificity for detecting depression in pregnant and postpartum women, other cut-off values may be used to prioritise either sensitivity or specificity. Reference Levis, Negeri, Sun, Benedetti and Thombs128 Moreover, the majority of the GWASs employed a retrospective assessment of perinatal depression symptoms using the EPDS, which may introduce recall bias and thus affect accuracy of the reported symptoms. Reference Kiewa, Meltzer-Brody, Milgrom, Guintivano, Hickie and Whiteman96–Reference Byrne, Carrillo-Roa, Penninx, Sallis, Viktorin and Chapman98,Reference Pouget, Taylor, Dennis, Grigoriadis, Oberlander and Frey104 Overall, the assessment of perinatal depression symptoms varies widely, which makes it difficult to compare the findings and may explain some of the inconsistencies.
This is the first systematic review to provide an overview of the current literature on the association among genetic variations, epigenetic modifications and gene expression levels of oestrogen receptor genes and perinatal depression symptoms. However, some limitations need to be addressed. Despite 29 studies being included, only a small number of studies investigated each category, with 16 studies investigating genetic variations, five studies examining epigenetic modifications and eight studies analysing gene expression levels of oestrogen receptor genes. Moreover, although 16 genetic studies were identified for inclusion, nine of these employed a candidate gene approach, whereby each genetic variation was only analysed in a limited number of studies, with the majority being tested in only one study. Furthermore, the genetic studies conducting a genome-wide approach were mostly underpowered. In addition, the broad inclusion criteria in terms of study design restricted potential comparisons between the studies. Accordingly, we were unable to compare the findings for subgroups with symptom onset during pregnancy and symptom onset postpartum. Moreover, as all of the included studies employed a correlational design, we cannot draw causal conclusions regarding how the investigated markers may ultimately affect perinatal depression symptoms. Another limitation of this review lies in its primary focus on oestrogen receptor genes. As other sex steroid hormones such as progesterone also fluctuate during the perinatal period and are also associated with mood and behaviour, the sensitivity of their receptors may also be implicated in perinatal depression symptoms. Reference Sundström-Poromaa, Comasco, Sumner and Luders129 Since progesterone and oestrogen can exert both synergistic and antagonistic effects within the same neurotransmitter systems, a complex interplay underlying perinatal depression symptoms cannot be ruled out and warrants further investigation. Reference Del Río, Alliende, Molina, Serrano, Molina and Vigil43
To summarise, initial evidence supports the involvement of oestrogen receptor genes in perinatal depression symptoms, through a potential inter-individual modulation of the oestrogen receptors’ sensitivity to oestrogen. Genetic variations such as rs2234693 in ESR1, along with epigenetic modifications, including DNA methylation marks involved in oestrogen signalling and gene expression levels of ESR1 and ESR2, may serve as future biomarkers for perinatal depression symptoms. However, the current empirical evidence is insufficient to establish these biomarkers for clinical use, and the correlational nature of the results restricts definitive conclusions about the specific biological mechanisms underlying these associations. In view of the present findings, we recommend that future research should not only replicate these findings in larger samples, with an emphasis on reducing methodological inconsistencies and exploring potential interactions, but should also conduct further functional investigations to uncover the specific biological pathways through which oestrogen receptor genes may contribute to perinatal depression symptoms. Future research also needs to explore the potential involvement of GPER in perinatal depression symptoms in more detail. Since non-genomic oestrogen signalling may be crucial in modulating the rapid intracellular responses to oestrogen fluctuations, a consideration of both the genomic and non-genomic effects of oestrogen may provide vital insights into how hormonal changes affect the brain and mood regulation during the perinatal period. Given the complexity of perinatal depression and its aetiology, we recommend that future studies should also employ more comprehensive and integrative approaches, such as multi-omics, to simultaneously analyse data on various genetic, epigenetic, hormonal, environmental and psychosocial factors. Such approaches will be critical in moving towards a better understanding of the aetiology of perinatal depression symptoms and in establishing reliable biomarkers.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1192/bjp.2025.39
Data availability
Data availability is not applicable to this article as no new data were created or analysed in this study.
Acknowledgements
The authors thank Sarah Mannion for proofreading the manuscript and Tara von Grebel University of Zurich, Information Technology, MELS/SIVIC, for creating Fig. 2.
Author contributions
G.Z.: conceptualisation, investigation, visualisation, interpretation of data, writing – original draft. U.E.: conceptualisation, supervision, interpretation of data, writing – reviewing and editing.
Funding
This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
Declaration of interest
None.
Analytic code availability
Analytic code availability is not applicable to this article as no new data were created or analysed in this study.
Research availability
Research materials of this review are available on request from the corresponding author, U.E.
eLetters
No eLetters have been published for this article.