Uterine leiomyoma (UL), commonly known as fibroids, is a benign tumor of smooth muscle in the uterus (Wise and Laughlin-Tommaso, 2016; Stewart et al., 2017). The symptoms include excessive menstrual bleeding, abdominal pain, and pregnancy complications (Zimmermann et al., 2012; Stewart et al., 2017). About 30 percent of reproductive-age women are diagnosed with UL, which is considered as a major factor of gynecologic morbidity (Wise et al., 2004), one of the primary causes of hospitalizations for gynecological disorders, and the most frequent reason for hysterectomy in the United States (Wise and Laughlin-Tommaso, 2016). ULs bear a substantial economic burden: the annual cost of treating this disorder has been estimated to be about $34 billion in the United States alone, which exceeded the combined cost of treating breast and colon cancers (Cardozo et al., 2012).

The biological mechanisms of UL development are not well understood (Rafnar et al., 2018). Family history is an important risk factor for UL (Commandeur et al., 2015; Wise and Laughlin-Tommaso, 2016). A risk to develop the disease is 2.5-fold greater for first-degree relatives of affected women than the population average (Vikhlyaeva et al., 1995), and the concordance among monozygotic twins is almost twice that of dizygotic twins (Luoto et al., 2000). Heritability estimates of UL from twin studies vary from 26 to 69% in European populations (Snieder et al., 1998; Luoto et al., 2000). More estimates come from recent genome-wide association studies (GWASs). Cha et al. (2011) reported candidate loci for UL in chromosome regions 10q24.33 (OBFC1), 11p15.5 (BET1L), and 22q13.1 (TNRC6B) in a cohort of Japanese patients. One more locus, 17q25.3 (FASN, CCDC57, and SLC16A3), was suggested by the genome-wide linkage and follow-up association studies in a meta-analysis of white women (Eggert et al., 2012). The 11p15.5 (BET1L) locus was later replicated in a sample of Caucasians (Edwards et al., 2013). The 22q13.1 (TNRC6B) locus was replicated in Caucasian, American, and Saudi Arabian cohorts (Edwards et al., 2013; Aissani et al., 2015; Bondagji et al., 2017). However, none of the above loci was replicated in African–American women (Wise et al., 2012). A GWAS of a risk for leiomyoma in African–American women suggested a locus on 22q13.1 (CYTH4) (Hellwege et al., 2017). In 2018, two teams presented results of the first GWAS of UL in European populations (Rafnar et al., 2018; Välimäki et al., 2018) reporting 21 and 22 loci associated with the increased risk of UL.

Age at menarche (AAM) has been suggested as a risk factor for UL (Wise et al., 2004; Terry et al., 2010; Velez Edwards et al., 2013; Wise and Laughlin-Tommaso, 2016). Women with an early AAM have, on average, longer period of menstrual cycling in their life and thus a greater lifelong exposure to estrogens, which may promote growth of UL (Wise et al., 2004). Cell division rate in the myometrium is the highest during the luteal phase of the menstrual cycle; therefore, a longer history of cycling may increase a risk of UL (Wise and Laughlin-Tommaso, 2016). On the other hand, some studies reported no significant association between age of menarche and UL (Samadi et al., 1996; Zimmermann et al., 2012).

The significantly different heritability estimates for UL [from 26 to 69% according to twin studies (Snieder et al., 1998; Luoto et al., 2000) and only 13% according to GWAS data single-nucleotide polymorphism (SNP) heritability (Rafnar et al., 2018)] bring about a problem of so-called “missing heritability.” Studies of candidate genes associated with specific risk factors for UL [e.g., AAM and body mass index (BMI)] may help to solve this problem. For example, more than 350 loci have been reported by GWAS for association with AAM (Day et al., 2017). There are data that candidate genes for AAM (e.g., FTO, LIN28B, MAP2K5, TNNI3K, GPRC5B, and FANCL) may also contribute to various anthropometric traits (e.g., BMI, height, and others) (Ong et al., 2009; Elks et al., 2010; Fernandez-Rhodes et al., 2013; Perry et al., 2014), while some of these traits (e.g., an increased BMI) may be risk factors for UL (Luoto et al., 2000; Wise and Laughlin-Tommaso, 2016; McWilliams and Chennathukuzhi, 2017). However, there have been no studies on the possible association of the candidate genes for AAM with a risk of UL.

The purpose of the present study was to analyze 52 candidate loci for menarcheal age for their possible association with UL. Recently, we studied association of these loci with some phenotypic traits (including AAM, height, and BMI) in the same sample of Caucasian females (Ponomarenko et al., 2019). Using the same sample for the analysis of different traits eliminates between-sample heterogeneity and makes comparison and interpretation of the results more meaningful and robust.

The phases of the study and main results are outlined in Figure 1.

This study reports for the first time the associations of 23 candidate loci for AAM with UL. Locus rs7759938 of the LIN28B gene demonstrated the most significant associations: its allele Ñ decreased a risk for UL according to the dominant model (OR = 0.74). This locus is also associated with UL within most significant four-order model of polymorphisms interactions with induced abortions.

According to the GTEx Portal database, rs7759938 and 27 SNPs linked to it have the cis-eQTL significance and may affect expression of LIN28B in the pituitary gland (allele Ñ of rs7759938 increases expression of LIN28B). Association of this SNP with menarcheal age was reported previously (Perry et al., 2009; Elks et al., 2010; Delahanty et al., 2013; Perry et al., 2014; Ponomarenko et al., 2019). Several studies suggested that rs7759938 might play a role in the pubertal development and adult height (Elks et al., 2010; Lango Allen et al., 2010; Widén et al., 2010; Leinonen et al., 2012; Cousminer et al., 2013; Perry et al., 2014). Two polymorphisms, rs9391253 and rs314263, which are linked to rs7759938 (r² ≥ 0.94), were associated with height (Berndt et al., 2013; Wood et al., 2014). Importantly, allele C of rs7759938 was previously determined as associated with later menarche, delayed pubertal growth, and high height (Perry et al., 2009, 2014, Elks et al., 2010; Delahanty et al., 2013). On the other hand, the present study suggested this allele as a protective factor for UL. That means that AAM and a risk for UL may have a shared genetic basis.

Two SNPs of the LIN28B gene (rs314276 and rs4946651) were involved in five out of 11 most significant models of epistatic interactions, and polymorphism rs314280 LIN28B was associated with UL only through interaction with induced abortions and chronic endometritis. All these polymorphisms were reported as associated with AAM (He et al., 2009; Ong et al., 2009; Sulem et al., 2009; Carty et al., 2013). There is evidence that rs314276 may play a role in the pubertal development and height (Ong et al., 2009; Cousminer et al., 2014), BMI, and weight in females (Ong et al., 2011). Previously, we reported the association of the rs314280 and rs4946651 polymorphisms with BMI of adults (Ponomarenko et al., 2019). Two polymorphisms, rs9391253 and rs314263, which are linked to rs314276 (r² ≥ 0.94), were associated with height (Berndt et al., 2013; Wood et al., 2014). The data of GTEx Portal suggest that the UL-associated SNPs of LIN28B have cis-eQTL significance: rs314276 (with 20 SNPs linked to it) and rs4946651 (with 15 SNPs linked to it) are associated with expression level of LINC00577 in the cortex and basal ganglia of the brain, LIN28B in the pituitary gland, and HACE1 in skeletal muscle; rs314276 and 27 are linked to SNPs that may affect expression of the LIN28B gene in the pituitary gland.

The above four UL-associated SNPs of the LIN28B gene are located in region 6q16.3-q21 about 38 kb from each other and are linked to about 40 SNPs with the regulatory potential and cis-eQTL values (affect transcription of the LIN28B, LINC00577, and HACE1 genes).

LIN28B (lin-28 homolog B) is a member of the LIN28 family of proteins and functions as a repressor of the let-7 family of miRNA (Tsialikas and Romer-Seibert, 2015). These miRNAs contribute to developmental processes in invertebrate animals, while their role in vertebrates is not well understood. Some data suggest that the let-7 miRNAs may contribute to cell cycle control and oncogenesis (Tsialikas and Romer-Seibert, 2015). The target genes of the let-7 miRNA, Kras, Myc, Hmga2, and Igf2bp1 are involved in body size control and metabolism in mammals (Zhu et al., 2011). The function of LINC00577 (LIN28B antisense RNA 1) is unknown; according to the information from the Ensembl database, it belongs to the genes encoding long intergenic non-coding RNAs (lincRNAs), which interact with proteins, DNA, and other RNAs, and thus perform important regulatory functions (Ulitsky and Bartel, 2013). A protein product of the HACE1 (HECT domain and ankyrin repeat containing E3 ubiquitin protein ligase 1) gene is a potential tumor suppressor and participates in the specific tagging of target proteins that subsequently results in proteasome degradation²¹.

The rs555621, rs11031010, and rs1782507 polymorphisms of the FSHB gene appeared in the largest number of the most significant models of intergenic interactions (7 out of 11 models); rs11031010 FSHB was presented in 10 out of 16 most significant models gene–environment interactions associated with UL. Recently, we determined association of rs555621 with BMI of adults (Ponomarenko et al., 2019). The above three loci were also associated with AAM (He et al., 2010), and rs11031010 was suggested to be a risk factor of PCOS and affect a level of luteinizing hormone (LH) in patients (Tian et al., 2016).

The FSHB gene encodes a beta-subunit of the follicle-stimulating hormone. This hormone has multiple functions in the female reproductive system, including stimulation proliferation of follicular granulosa cells, rescue of follicles from apoptosis, and synthesis of receptors for the LH on these cells before ovulation.

The recent meta-analysis of 11 GWAS determined the region on chromosome 11 (11p14.1, rs74485684) harboring the FSHB gene as that associated with endometriosis (Sapkota et al., 2017). Importantly, rs11031010 FSHB analyzed in the present study is located just 1.9 kb from the above locus. On the other hand, the first GWAS of UL in European populations (Rafnar et al., 2018) replicated three of the 19 previously reported endometriosis variants (Sapkota et al., 2017). Several loci associated with endometriosis, including rs11031006 located in 11p14.1, were reported for their association with UL in the recent GWAS (Gallagher et al., 2019). Importantly, the rs11031010 polymorphism of the FSHB gene analyzed in the present study is linked to rs11031006 (r² = 0.64).

An epidemiological meta-analysis of 402,868 women suggested at least twofold higher risk for UL in patients with a history of endometriosis (Gallagher et al., 2019). It was hypothesized that both UL and endometriosis were underlain by hormone-related factors (Rafnar et al., 2018) and thus might have a shared genetic basis (Gallagher et al., 2019). This assumption was supported by the recent findings of syntropic genes for these diseases (Rafnar et al., 2018; Gallagher et al., 2019).

There is evidence that some loci linked to rs11031010 may contribute to various reproductive characters, such as concentration of the luteinizing (rs11031002, r² = 0.76) and follicle-stimulating (rs11031005, r² = 0.64) hormones in blood plasma (Ruth et al., 2016), and age at menopause (rs12294104, r² = 0.61) (Stolk et al., 2012). The SNPs upstream of the FSHB transcription start were implicated into a key role in many reproductive processes (Gajbhiye et al., 2018). It is generally acknowledged that women with late menopause or/and early menarche have, on average, an increased lifetime exposure to estrogen due to the larger number of ovulatory cycles. Since mitotic activity of myometrium is the highest during the luteal phase of the cycle, a longer history of menses may theoretically increase a risk for UL. Likewise, this risk may be higher due to the increasing LH level (Wise and Laughlin-Tommaso, 2016).

The present study determined that the rs2164808 polymorphism of the POMC gene was involved in the largest number of the most significant models of gene–gene (four out of 11 models) and gene–environment (five out of 16 models) interactions associated with UL. Previously, we reported the association of the rs2164808 locus with BMI of adults (Ponomarenko et al., 2019). The association of this locus with AAM was first reported by He et al. (2010). Polymorphism rs1561288, which is linked to rs2164808 (r² = 0.25), was associated with BMI (Graff et al., 2013). Locus rs2164808 and linked to it rs4665765 are both associated with the transcription level of DNAJC27 (according to the Blood eQTL browser) and RP11-509E16.1 (according to the GTEx Portal database). The data from the SNP Function Prediction and rSNPBase suggest a significant regulatory effect of rs2164808 (proximal regulation and RNA binding protein mediated regulation effects, regulatory potential score 0.334). The POMC (pro-opiomelanocortin-alpha) gene encodes a preproprotein that undergoes tissue-specific posttranslational modifications resulting in up to 10 biologically active peptides, which participate in various cellular processes, including those related to UL (e.g., lipolysis, mobilization of fatty acids, steroidogenesis, and adrenal development) (see footnote 21). The product of the DNAJC27 [DnaJ heat shock protein family (Hsp40) member C27] gene is a GTPase that is associated with the MEK/ERK pathway and may induce cell transformation (see footnote 21). The function of RP11-509E16.1 is unknown; according to the information from the Ensembl database²², it belongs to the genes encoding lincRNAs, which interact with proteins, DNA, and other RNAs and thus perform important regulatory functions (Ulitsky and Bartel, 2013). The polymorphisms located in the regions of the POMC, DNAJC27 (RBJ), and ADCY3 genes were associated with pubertal development, height, BMI, obesity, and type I diabetes mellitus (Gudbjartsson et al., 2008; Lango Allen et al., 2010; Speliotes et al., 2010; Bradfield et al., 2011; Berndt et al., 2013; Cousminer et al., 2013; He et al., 2015).

One more finding of the present study is that associations with UL were determined not only for candidate SNPs for AAM (Supplementary Table 4) but also for seven polymorphisms manifesting association or tagged with the traits related to menarche (e.g., sex steroid hormone and vitamin D metabolism, and physical characteristics; Supplementary Tables 4, 12). These menarche-related traits are important for pathogenesis of UL (Commandeur et al., 2015; Wise and Laughlin-Tommaso, 2016). The latter are involved in the sex steroid hormone pathway (rs3020394 and rs1884051 ESR1, rs4633 COMT, and rs12324955 FTO and rs7766109 F13A1) and vitamin D metabolism (rs1544410 VDR and rs222020 GC) (Supplementary Table 4, see footnote 21), which are important for UL development (Commandeur et al., 2015; Wise and Laughlin-Tommaso, 2016). For example, rs4633 COMT, which is also associated with AAM, height, and BMI in the studied sample (Ponomarenko et al., 2019) and linked to it rs4680 is associated with the COMT expression level in peripheral blood (according to the Blood eQTL browser). The activity of the COMT enzyme is elevated in human leiomyoma tissue as compared with normal myometrium; since this enzyme is essential for estrogen metabolism, the observed association suggests a possible causal role of COMT in UL formation (Commandeur et al., 2015). In addition, the rs4680 polymorphism encodes a replacement substitution Val158Met in COMT: the Met variant has 40% lower activity as compared with the Val one (Chen et al., 2004). Further support of the possible contribution of rs4680 to UL comes from the meta-analysis by Feng et al. (2013), who reported the association of rs4680 with UL.

We did not find a direct association of AAM with UL but determined that 23 candidate loci for AAM might be associated with UL. Among these, several were also associated with age of menarche, height, and/or BMI of adults in the study sample. This may suggest that menarcheal age per se is not a primary risk factor for UL in the studied sample (the population of Russia) and that other factors may be of greater importance. Specifically, we found that the history of induced abortions significantly increased the risk of UL in the study sample (OR_adj = 2.78). Our results support the recent findings by other researchers (Shen et al., 2017; Song et al., 2017). Induced abortions are common birth control method in Russia (and generally in the former USSR) (Sedgh et al., 2007; David et al., 2007; Douglas et al., 2014), and Russia (along with Cuba) is a leader by this indicator in the world (Sedgh et al., 2007). Induced abortions in the first trimester of pregnancy may cause multiple health complications, including PCOS, hyperprolactinemia, anovulatory menstruation, disorders of the luteal phase of the menstrual cycle, and post-abortion endometritis (Douglas et al., 2014). Induced recurring abortions result in injuries of the uterus. The patient cohort in the present study included 44.99% individuals who had at least two induced abortions. Uterine injuries may induce the expression of the growth factor that speed ups cellular proliferation, suppresses apoptosis, and increases extracellular matrix production (Wise and Laughlin-Tommaso, 2016). Indeed, our results suggest that the risk for UL directly correlates with the number of the abortions in the anamnesis (OR_adj = 1.69): it increases from OR = 3.85 (95% CI 2.79–5.32, p < 0.001) in a case of two abortions up to OR = 13.41 (95% CI 7.22–25.24, p < 0.001) with four and more abortions. Similar results were reported by Song et al. (2017) who showed that the higher number of induced abortions was associated with an increased risk of UL (one induced abortion, OR = 1.32; two induced abortions, OR = 1.45; and ≥ 3 induced abortions, OR = 1.62). PCOS was associated with a 65% increased risk of UL (Wise and Laughlin-Tommaso, 2016). The mechanisms by which PCOS may elevate risk for UL include increased levels of LH or unopposed estrogens and its association with hyperinsulinemia (Wise and Laughlin-Tommaso, 2016). Induced abortions may also be associated with other hormone-related diseases (e.g., breast cancer) (Schneider et al., 2014). Inflammation in the uterus (stress from chronic inflammation) seems a quite possible etiologic factor of UL: studies showed that inflammation increases the production of extracellular matrix and decreases apoptosis in UL (Commandeur et al., 2015; Wise and Laughlin-Tommaso, 2016).

Candidate genes for AAM are associated with UL. Locus rs7759938 LIN28B is associated with UL independently; 20 loci are associated within the 11 most significant intergenic interaction models, and nine loci may contribute to the risk of the disorder through the 16 most significant gene–environment interactions. These associations probably result from multiple effects of more than 510 polymorphisms linked to the above SNPs and implicated in various metabolic processes in organs and tissues that are related to pathogenesis of UL.

The datasets generated for this study can be found in EVA at the following accessions Project: PRJEB40782, Analyses: ERZ1645102.

The studies involving human participants were reviewed and approved by Regional Ethics Committee of Belgorod State University. The patients/participants provided their written informed consent to participate in this study.

IP, MC, and VD made substantial contributions to conception and design, drafted the manuscript, and gave the final approval of the version to be published. ER, AP, IV, IS, and AY analyzed and interpreted of data, drafted the manuscript, and gave the final approval of the version to be published. All the authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.