Abstract
Introduction
Many diseases of adulthood are associated with a woman’s age at menarche. Genetic variation affects age at menarche, but it remains unclear whether in women of African ancestry the timing of menarche is regulated by genetic variants that were identified in predominantly European and East Asian populations.
Methods
We explored the genetic architecture of age at menarche in 3145 women of African ancestry who live in the United States, Barbados, and Nigeria. We undertook a genome-wide association study, and evaluated the performance of previously identified variants.
Results
One variant was associated with age at menarche, a deletion at chromosome 2 (chr2:207216165) (p=1.14×10−8). 349 genotyped variants overlapped with these identified in populations of non-African ancestry; these replicated weakly, with 51.9% having concordant directions of effect. However, collectively, a polygenic score constructed of those previous variants was suggestively associated with age at menarche (beta=0.288 years; p=0.041). Further, this association was strong in women enrolled in the United States and Barbados (beta=0.445 years, p=0.008), but not in Nigerian women (beta=0.052 years; p=0.83).
Discussion
This study suggests that in women of African ancestry the genetic drivers of age at menarche may differ from those identified in populations of non-African ancestry, and that these differences are more pronounced in women living in Nigeria, although some associated trait loci may be shared across populations. This highlights the need for well-powered ancestry-specific genetic studies to fully characterize the genetic influences of age at menarche.
Introduction
For girls, a central event of puberty is the first occurrence of menstruation, which generally occurs between the ages of nine and fifteen. The age at which menarche occurs is robustly associated with health later in life; earlier menarche is associated with increased risk of breast and endometrial cancers (1,2), and lower risk of osteoporosis (3) and cardiometabolic diseases (4,5).
The average age at menarche varies across the globe (6), and its timing is influenced by multiple factors, including nutrition, early childhood stressors, animal protein intake, physical activity, and body mass index (BMI) (7). Inherited genetic variation has also been shown to be associated with age at menarche, and among women of European decent, heritability estimates suggest that more than 45% of the variation in age at menarche may be due to genetic variation (8). Genome-wide association studies (GWASs) and sequencing studies have identified more than four hundred single nucleotide variants (SNVs) that are associated with age at menarche (9). However, these studies are largely comprised of women in high income countries with European or East Asian ancestry. In the few studies that have been undertaken in populations of African ancestry, these variants have generally not continued to be strongly associated with age at menarche (10–12).
This lack of replication may be caused by several factors, each of which would carry differing implications about the genetic architecture of age at menarche in diverse populations. Genotype chip design may tag causal alleles less efficiently in populations of African decent, or the causal alleles identified in European populations may be less frequent in populations of African decent. Further, in some geographic areas, non-genetic factors that are associated with age at menarche may have a comparably stronger influence. Therefore, without additional studies, although gains have been made in understanding the genetic architecture of age at menarche, it remains unclear the extent to which this knowledge will translate into insight into the mechanisms underlying menarche in populations of African Ancestry.
In this study, the genetic architecture of age at menarche is explored in 3145 women of African ancestry who live in the United States (US), Barbados, and Nigeria. A genome-wide association study was undertaken to identify if individual variants were associated with age at menarche. Further, variants identified in past studies were examined for evidence of replication, as well as their combined ability to collectively predict age at menarche as a polygenic score.
Methods
Study participants and phenotyping
The study population is a subset of the participants of the Root GWAS in Breast Cancer in the African Diaspora. The Root consortium is comprised of six epidemiologic studies of breast cancer in women of African ancestry: the Nigerian Breast Cancer Study (Ibadan, Nigeria); the Barbados National Cancer Study (Barbados); the Southern Community Cohort (southern United States); the Baltimore Breast Cancer Study (Baltimore, United States); the Racial Variability in Genotypic Determinants of Breast Cancer Risk Study (Detroit and Pennsylvania, United States); and the Chicago Cancer Prone Study (Chicago, United States). The consortium has been fully described elsewhere (13). A total of 3686 participants were included in the Root consortium (1657 breast cancer cases and 2029 controls). Self-reported age at menarche was collected by questionnaire in each participating study. After excluding women for whom age at menarche was missing, 3145 participants were included in the present analysis.
All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed consent was obtained from all individual participants included in the study. The University of Chicago Institutional Review board approved the current research (IRB 13304B). The genotype data of Root genome-wide association study are posted in dbGaP (phs000383.v1.p1 https://www.ncbi.nlm.nih.gov/projects/gap/cgi-bin/study.cgi?study_id=phs000383.v1.p1).
Genotyping
Genotyping and quality control have been described elsewhere (13). Briefly, genotyping was conducted using the Illumina HumanOmni2.5–8v1 array. Quality control analysis was carried out by checking chromosomal anomalies, participant relativeness, population structure, missing call rates, batch effects, duplicate sample discordance, Mendelian errors, Hardy-Weinberg equilibrium, and duplicate SNV probes. Additionally, an a priori-defined threshold of 90% European ancestry was used to identify 20 subjects (18 African Americans and 2 African Barbadian) with higher-than-expected levels of European ancestry using principal component analyses (the average percent of European ancestry in the remaining sample was 11%). SNV-level quality control excluded SNVs with technical errors, high missingness, discordant calls, Mendelian errors, and Hardy-Weinberg equilibrium test p < 10−4. After these exclusions, 2,116,365 SNV remained.
Genotype imputation was conducted by using the IMPUTE2 software (14), using the 1000 Genomes Project v3 (15) as the reference panel. Variants with imputation score greater than 0.3 were kept, resulting in 35,949,944 variants. These variants were further filtered to include only variants that were common in the study sample (MAF > ), corresponding to 15,815,746 autosomal variants and 597,290 X chromosome variants used in the analyses.
Statistical methods for GWAS
To examine the association between age at menarche and individual genetic variants, we undertook a GWAS assuming a linear additive model of inheritance. The analysis additionally controlled for the top five principal components, study site, and age at study enrollment, to account for the secular increase in age at menarche that has been observed over the last century.(16–18) While the addition of birth year to the analysis would have allowed for a more thorough understanding of age, period, and cohort effects, birth year information was not available for four of the six study sites due to data sharing restrictions. The participants enrolled in the Nigerian study site and those enrolled in the US and Barbados were analyzed separately, and the estimates were combined using a mixed effect meta analysis using PLINK (19,20). The p-value threshold for statistical significance for individual SNVs was set at 5×10−8, consistent with other GWASs (21).
Identifying variants associated with age at menarche in literature
Variants that are associated with age at menarche were identified using three sources: the most recent meta-analysis of women of European ancestry (9), the NHGRI-EBI GWAS catalog (22) studies investigating age at menarche, and multi-ethnic fine mapping work done in the PAGE Study.(23)
The meta-analysis (9) reported 389 variants with p-values below the genome-wide threshold of 5×10−8; each of these variants was retained for replication analyses and the polygenic score analysis (described below).
In the NHGRI-EBI catalog, (22) nineteen studies report on age at menarche. This study includes the nine that investigated age at menarche measured in years. Of those nine studies, six were undertaken in study populations that overlapped with those used in the meta-analysis. Of the three studies that were not included in the meta-analysis, two were undertaken in women of Japanese ancestry (24,25), and one was undertaken in women of African ancestry (26). The studies of Japanese ancestry identified seventeen SNVs with p-values below 5×10−8; each of these variants was retained for replication and the polygenic score analyses. Although only one SNV from the African ancestry population was genome-wide significant, all eleven variants listed were included in the replication analysis and polygenic scores due to the similarity of genetic ancestry between that study population and the present study population. One additional SNV, rs76455660, was identified from the PAGE fine mapping study (23).
In total, 419 unique variants were identified. Of these, 29 were insertions or deletions that were not fully annotated in the literature and were removed. Of the remaining 390 variants, 354 (91%) were genotyped or imputed in the Root study population and present at a MAF > 0.0126 in the pooled sample.
Statistical methods for polygenic score
These 354 SNVs were then further pruned to remove one of every pair that is in close LD, using an R2 of 0.2 as a threshold and retaining the SNV in the pair with the smaller reported p-value in the literature. This pruning removed five SNVs (rs11852771, rs10453225, rs1516883, rs11786868, and rs11767400), and the remaining 349 SNVs were retained for the polygenic score calculation.
Each participant’s polygenic score was calculated as the summation of the product of their number of effect alleles and the published effect size of each variant (recorded in the “beta literature” column in Supplemental Table 1). In addition, we also calculated unweighted polygenic score as the summation of the number of effect alleles, a score that does not have a good biological interpretation but may be robust to transferring polygenic scores across discordant ancestry groups. We used multiple linear regression models to examine the association between polygenic scores and age at menarche after adjusting for age and study site as covariates.
Patient and Public Involvement
Patients were not involved in the study design.
Results
Selected characteristics of the study sample from the time of enrollment are presented in Table 1. The mean age at menarche was 15.2 years for Nigerian women, 13.2 for African Barbadians, 12.3 years in the Chicago site, 12.8 years in the Southern Community Cohort, 12.5 years in the Detroit and Pennsylvania enrollment sites, and 12.5 years in the Baltimore enrollment site. As shown in Supplemental Figure 1, the first principal component can distinguish Nigerian women from women enrolled in North America. The mean estimated African ancestry proportion was 0.80 (standard deviation SD = 0.11) for African Americans, 0.84 (SD = 0.11) for African Barbadians, and 0.99 (SD = 0.005) for Nigerians.
Table 1:
Characteristics of the study sample
| Overall | NBCS | CCPS | SCC | RVGDBCR | BBCS | BNCS | |
|---|---|---|---|---|---|---|---|
|
| |||||||
| Age at Menarche (sd) | 13.7 (2.3) | 15.2 (2.1) | 12.3 (1.8) | 12.8 (1.9) | 12.5 (1.8) | 12.5 (1.8) | 13.2 (1.9) |
| Menopause Status (%) | |||||||
| Pre | 1543 (49.1) | 752 (59.9) | 190 (44.8) | 178 (27.9) | 234 (62.9) | 66 (42.3) | 123 (40.9) |
| Post | 1556 (49.5) | 503 (40.1) | 201 (47.4) | 457 (71.7) | 138 (37.1) | 79 (50.6) | 178 (59.1) |
| Not Specified | 46 (1.5) | 0 (0) | 33 (7.8) | 2 (0.3) | 0 (0) | 11 (7.1) | 0 (0) |
| Family History of BC (%) | |||||||
| Negative | 2570 (81.7) | 1191 (94.9) | 143 (33.7) | 517 (81.2) | 316 (84.9) | 130 (83.3) | 273 (90.7) |
| Positive | 398 (12.7) | 64 (5.1) | 156 (36.8) | 72 (11.3) | 54 (14.5) | 24 (15.4) | 28 (9.3) |
| Not Specified | 177 (5.6) | 0 (0) | 125 (29.5) | 48 (7.5) | 2 (0.5) | 2 (1.3) | 0 (0) |
| Age at Enrollment (%) | |||||||
| under 30 | 177 (5.6) | 92 (7.3) | 33 (7.8) | 0 (0) | 48 (12.9) | 3 (1.9) | 1 (0.3) |
| 30–39 | 573 (18.2) | 287 (22.9) | 106 (25.0) | 0 (0) | 116 (31.2) | 26 (16.7) | 38 (12.6) |
| 40–49 | 967 (30.7) | 414 (33.0) | 162 (38.2) | 157 (24.6) | 114 (30.6) | 43 (27.6) | 77 (25.6) |
| 50–59 | 825 (26.2) | 298 (23.7) | 75 (17.7) | 268 (42.1) | 58 (15.6) | 34 (21.8) | 92 (30.6) |
| 60–69 | 420 (13.4) | 127 (10.1) | 32 (7.5) | 150 (23.5) | 28 (7.5) | 33 (21.2) | 50 (16.6) |
| 70 and older | 183 (5.8) | 37 (2.9) | 16 (3.8) | 62 (9.7) | 8 (2.2) | 17 (10.9) | 43 (14.3) |
| Breast Cancer at Enrollment (%) | |||||||
| Case | 1469 (46.7) | 647 (51.6) | 317 (74.8) | 215 (33.8) | 130 (34.9) | 76 (48.7) | 84 (27.9) |
| Control | 1676 (53.3) | 608 (48.4) | 107 (25.2) | 422 (66.2) | 242 (65.1) | 80 (51.3) | 217 (72.1) |
| BMI at Enrollment (sd) | 29.1 (7.1) | 26.4 (5.6) | 30.3 (7.6) | 32.8 (7.4) | 29.9 (7.1) | 31.9 (7.4) | 29.3 (6.2) |
| Parity at Enrollment (sd) | 3.1 (2.2) | 3.9 (2.2) | 2.0 (1.6) | 3.1 (2.2) | 2.0 (1.5) | 2.6 (2.0) | 2.7 (2.3) |
BMI at Enrollment is missing in 157 participants and parity at Enrollment is missing in 74 participants.
NBCS=Nigerian Breast Cancer Study, CCPS=Chicago Cancer Prone Study, SCC=Southern Community Cohort, RVGDBCR=Racial Variability in Genotypic Determinants of Breast Cancer Risk (Detroit and Pennsylvania), BBCS=Baltimore Breast Cancer Study, BNCS=Barbados National Cancer Study, BC=Breast Cancer, sd=standard deviation
Genome-wide association study
Figure 1 plots the Manhattan (Fig 1A) and QQ plots (Fig 1B) of the single-variant analysis of age at menarche. The QQ plot shows no evidence of genetic inflation, with an overall λgc of 1.008. One variant met the threshold for genome wide significance (beta= −1.25; p=1.14×10−8), a deletion on chromosome 2 (chr2:207216165; a TGATAGATA to T deletion, present in 1.3% of the pooled study sample). This variant was imputed in our study, with an imputation information metric of 0.91. Four other indels were measured at this locus, including rs143651325, which was present in 8% of the study population. No other variants near that locus were also associated with age at menarche (Fig 1C).
Figure 1: Manhattan and QQ plots for single variant analysis of age at menarche.
Manhattan (A) QQ plots (B) of the results of the variant-by-variant analysis of age at menarche (all participants, N=13145). In Manhattan plots, the vertical line marks a p-value threshold of 5·10−8. (C) A subset of the Manhattan plot, focusing on the locus near chr2:207216165, the SNP with the smallest p-value for association. All variants at chr2:207216165 are highlighted in purple.
The relationship between the 2:207216165 deletion and age at menarche was qualitatively similar in the smaller stratified samples (Nigeria: beta=−1.59, p=3.16×10−4, MAF=0.011; US/Barbados: beta=−1.12, p=1.08×105, MAF=0.014), and the pooled study sample (beta= −1.26, p=2.17×10−8) (Supplemental Figure 2 for Manhattan and QQ plots). The results are also substantively similar when controlling for breast cancer status as a covariate (beta= −1.26, p=2.24×10−8).
Replication of previously identified variants
Of the 349 candidate variants examined from the published literature that reached the MAF threshold in all study sites, 181 (51.9%) had directions of effect that were concordant with the effect estimates in our analysis. Of these 181 SNVs with concordant effect directions, 15 (8.3%) had nominally significant (p<0.05) p-values (Table 2, full listing of variants in Supplemental Table 1). When restricting to participants enrolled in the US and Barbados study sites, more variants had directions of effect that were concordant with the published literature (55.9%) than that restricting to the participants enrolled in Nigeria (49.1%). Of note, when examining the 11 SNVs that had been suggestively identified in previous populations of African ancestry (26), the replication rate was similar, with 4 out of 11 SNVs having concordant directions of effect. The results were essentially the same when controlling for breast cancer status as a covariate.
Table 2:
Concordanc of effects between published studies and the current study, by continent of enrollment site
| Variants with Concordant Effect Directions | Concordant variants with nominally significance | ||
|---|---|---|---|
|
| |||
| n (%) | p-value* | n (%) | |
| All Participants (pooled analysis) | 181/349 (51.9) | 0.52 | 15/181 (8.3) |
| Nigerian Enrollment Site | 158/322 (49.1) | 0.78 | 12/158 (7.6) |
| US and Barbados Enrollment Sites | 195/349 (55.9) | 0.032 | 12/195 (6.2) |
| All Participants (meta-analysis) | 165/322 (51.2) | 0.70 | 17/165 (10.3) |
p-values are from binomial test. The number of variants tested differs between study enrollment sites as some variants met the MAF threshold in the overall population, but not the site-specific analyses. In those cases, only variants present at MAF>0.0126 in both the overall population and the site=specific population were examined
Polygenic score analysis
Table 3 summarizes the results of the polygenic score analysis. The association between the weighted polygenic score and age at menarche was only marginally significant (beta=0.288 years, 95% CI: 0.012 to 0.564, p=0.041). In African Americans and African Barbadians, the association was stronger and statistically significant (beta=0.445 years, 95% CI: 0.115 to 0.775, p=0.008). For women enrolled in Nigeria, the effect size shrunk, and was no longer statistically significant (beta=0.052, 95% CI: (−0.434 to 0.539, p=0.83). To understand how sensitive our results were to the effect size published in the literature, we calculated unweighted polygenic score (i.e. sum of the effect allele counts) and we found results were similar, though slightly more variance in age at menarche can be explained.
Table 3:
Polygenic score prediction of age at menarche by continent of enrollment site
| Mean score (sd) | Beta (95% CI)* | p-value | partial R2 | |
|---|---|---|---|---|
|
| ||||
| Weighted Polygenic Score | ||||
| All Participants | −0.27 (0.25) | 0.288 (0.012 to 0.564) | 0.041 | 0.0013 |
| Nigerian Enrollment Site | −0.25 (0.24) | 0.052 (−0.434 to 0.539) | 0.83 | 0.0000 |
| US and Barbados Enrollment Sites | −0.29 (0.25) | 0.445 (0.115 to 0.775) | 0.008 | 0.0037 |
| Unweighted Polygenic Score | ||||
| All Participants | −10.7 (4.9) | 0.090 (0.022 to 0.158) | 0.009 | 0.0022 |
| Nigerian Enrollment Site | −10.5 (4.7) | 0.040 (−0.081 to 0.161) | 0.52 | 0.0003 |
| US and Barbados Enrollment Sites | −10.7 (5.0) | 0.125 (0.045 to 0.206) | 0.002 | 0.0049 |
Multiple linear regression model adjusting for age and study sites. For the unweighted polygenic score, beta coefficients for unit standard deviation was reported
sd, standard deviation; CI, confidence intervals
Discussion
As with previous GWASs of women of African Ancestry(26), this analysis of more than three thousand women of African descent identified few variants that were statistically associated with age at menarche. One imputed deletion at chr2:207216165 surpassed the genome wide association threshold. This deletion is located downstream of the genes ZDBF2 and GPR1-AS, both of which are imprinted and paternally expressed in the placenta (27,28), and upstream of ADAM23, which is expressed in the brain, heart, and testes (29). The expression and methylation of ADAM23 have been implicated in poor survival for breast and ovarian cancer (30,31). However, this deletion was present in our study population with a minor allele frequency that rose only slightly above the threshold that was required for inclusion in the GWAS, and no other variants in this locus were suggestively associated with age at menarche. We also acknowledge that the 5·10−8 p-value threshold may be too lenient in populations with higher levels of admixture, and post hoc power calculations suggest that we had 50% power to detect an effect size of 0.446 for variants with that MAF of 0.1. Further, this variant is not reliably measured across populations in either the NCBI dbSNP browser (32) or UCSC genome browser (33), suggesting that it might not be reliably imputable. While no validation data was available, replication would be required before suggesting that this variant is causally associated with age at menarche.
The outcomes of the replication and polygenic score analyses provided mixed insight. Individual variants replicated weakly. This is consistent with previous work, in which attempts to replicate associations in populations where ancestry was not concordant with the discovery sample were largely ineffective (10–12,23,34). However, although individual variants replicated weakly, when combining the associations of several hundred variants into a polygenic score, this score was able to predict age at menarche in African Americans and African Barbadians. Our study was well-powered to detect such an association, with post hoc power calculations showing that we had 80% power to detect an association of 0.38 years per unit increase in the polygenic score. While the variants were still only able to explain a small fraction of the variation in age at menarche (0.4%), as compared to 7.4% in women of European descent(9), the risk score analysis suggests that the timing of age at menarche may share some genetic influences in populations with some European ancestry.
Future research is needed to determine why the polygenic score was unable to predict age at menarche in Nigerian women. Differential LD between the populations is likely responsible for some of this lack of transportability, given that African Americans and African Barbadians have about 14–22% European ancestry(35). Since women in Nigeria have minimal admixture with European ancestry, this null finding may be a consequence of gene by gene interactions resulting in different causal alleles in Nigerian populations. Further, the women enrolled at the Nigerian site also were likely exposed to differential childhood nutrition (36). Nutrition has shown an epidemiologically consistent association with age at menarche over multiple populations (37–41), and earlier age at menarche is particularly strongly associated with diets that are higher in animal protein and lower in vegetables (42). If nutrition plays a dominant role in determining the age at menarche in Nigerian women, the genetic influences may be more difficult to identify in our sample. This would be consistent with our observed inter-country differences in menarche. All three could possibly obscure the relatively small genetic influences of any given variant, making it difficult to identify its influence in the Nigerian women.
This analysis was limited by sample size, as a study of slightly more than 3000 participants is underpowered to detect novel associations in genome-wide studies. However, our inability to convincingly replicate the variants identified European and East Asian populations indicates that larger studies are required to characterize the genetic architecture of age at menarche in populations of African ancestry. Our analysis was also limited by self-report of age at menarche. However, previous studies have found high agreement between self-reported age at menarche and age at menarche that can be validated through clinical records (kappa >0.8) (43).
In conclusion, this study suggests that in women of African ancestry, age at menarche is influenced by genetic variants that at least partially differ from those that have been identified by previous studies in women of non-African ancestry, but collectively, the loci identified previously may still have predictive power. Given the multitude of pathways through which age at menarche is associated with later health, our findings highlight the need for an increased focus on well-powered ancestry-specific genetic studies, as our understanding of the genetic pathways that drive age at menarche needs to appropriately reflect the likely geographic and ancestral diversity of the driving mechanisms.
Supplementary Material
Key Points.
Age at menarche (AAM), which is strongly associated with later risk of disease, has an established genetic component in women of European and East Asian ancestry
We find only limited evidence that these previously-identified variants continue to be associated in women of African ancestry.
Our findings highlight the need for larger, ancestry-specific studies to enable understanding of the genetic architecture of AAM across diverse populations.
Summary Box.
What is already known on this subject?
The age at which girls experience menarche is strongly associated with health later in life. Research has identified genetic variants that are associated with age at menarche, but most studies to-date have been carried out in women of European or East Asian ancestry
What does this study add?
This study suggests that ancestry-specific studies will be required to understand the genetic architecture of age at menarche in women of African decent, although some associated trait loci may be shared across populations.
Funding
This study was funded by the by the National Institute of Health (grant numbers T32CA057699, 5R25-CA057699, R01CA142996, and R01CA228198) and Breast Cancer Research Foundation (BCRF-21-071).
Footnotes
Conflict of Interest
The authors declare that they have no conflict of interest
Ethical Approval
All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed consent was obtained from all individual participants included in the study. The University of Chicago Institutional Review board approved the current research (IRB 13304B).
Data Availability
[Dataset] Huo D, Feng Y, Haddad S, et al. Data from: Genome-wide association studies in women of African ancestry identified 3q26.21 as a novel susceptibility locus for oestrogen receptor negative breast cancer. Hum Mol Genet. 2016 Nov 1;25(21):4835–46. The genotype data of Root genome-wide association study are posted in dbGaP (https://www.ncbi.nlm.nih.gov/projects/gap/cgi-bin/study.cgi?study_id=phs000383.v1.p1 ).
References
- 1.Collaborative Group on Hormonal Factors in Breast Cancer. Menarche, menopause, and breast cancer risk: individual participant meta-analysis, including 118 964 women with breast cancer from 117 epidemiological studies. Lancet Oncol 2012. Nov;13(11):1141–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Gong T-T, Wang Y-L, Ma X-X. Age at menarche and endometrial cancer risk: a dose-response meta-analysis of prospective studies. Sci Rep. 2015. Sep 11;5:14051. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Zhang Q, Greenbaum J, Zhang W-D, Sun C-Q, Deng H-W. Age at menarche and osteoporosis: A Mendelian randomization study. Bone. 2018. Dec 1;117:91–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Qiu C, Chen H, Wen J, Zhu P, Lin F, Huang B, et al. Associations Between Age at Menarche and Menopause With Cardiovascular Disease, Diabetes, and Osteoporosis in Chinese Women. J Clin Endocrinol Metab. 2013. Apr 1;98(4):1612–21. [DOI] [PubMed] [Google Scholar]
- 5.Plompen MVL, van der Schouw YT, Rutten FH, Verschuren WMM, Boer JMA, Asselbergs FW, et al. Age at menarche and heart failure risk: The EPIC-NL study. Maturitas. 2020. Jan;131:34–9. [DOI] [PubMed] [Google Scholar]
- 6.Lacroix AE, Langaker MD. Physiology, Menarche. In: StatPearls [Internet]. Treasure Island (FL): StatPearls Publishing; 2020. [cited 2020 Apr 15]. Available from: http://www.ncbi.nlm.nih.gov/books/NBK470216/ [Google Scholar]
- 7.Yermachenko A, Dvornyk V. Nongenetic Determinants of Age at Menarche: A Systematic Review [Internet]. BioMed Research International. 2014. [cited 2019 Mar 25]. Available from: https://www.hindawi.com/journals/bmri/2014/371583/abs/ [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Zaitlen N, Kraft P, Patterson N, Pasaniuc B, Bhatia G, Pollack S, et al. Using Extended Genealogy to Estimate Components of Heritability for 23 Quantitative and Dichotomous Traits. PLOS Genet 2013. May 30;9(5):e1003520. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Day FR, Thompson DJ, Helgason H, Chasman DI, Finucane H, Sulem P, et al. Genomic analyses identify hundreds of variants associated with age at menarche and support a role for puberty timing in cancer risk. Nat Genet. 2017. Jun;49(6):834–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Pyun J-A, Kim S, Cho NH, Koh I, Lee J-Y, Shin C, et al. Genome-wide association studies and epistasis analyses of candidate genes related to age at menarche and age at natural menopause in a Korean population. Menopause N Y N. 2014. May;21(5):522–9. [DOI] [PubMed] [Google Scholar]
- 11.Carty CL, Spencer KL, Setiawan VW, Fernandez-Rhodes L, Malinowski J, Buyske S, et al. Replication of genetic loci for ages at menarche and menopause in the multi-ethnic Population Architecture using Genomics and Epidemiology (PAGE) study. Hum Reprod Oxf Engl. 2013. Jun;28(6):1695–706. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Coignet MV, Zirpoli GR, Roberts MR, Khoury T, Bandera EV, Zhu Q, et al. Genetic variations, reproductive aging, and breast cancer risk in African American and European American women: The Women’s Circle of Health Study. PloS One. 2017;12(10):e0187205. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Huo D, Feng Y, Haddad S, Zheng Y, Yao S, Han Y-J, et al. Genome-wide association studies in women of African ancestry identified 3q26.21 as a novel susceptibility locus for oestrogen receptor negative breast cancer. Hum Mol Genet. 2016. Nov 1;25(21):4835–46. (https://www.ncbi.nlm.nih.gov/projects/gap/cgi-bin/study.cgi?study_id=phs000383.v1.p1 ). [Dataset] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Howie BN, Donnelly P, Marchini J. A Flexible and Accurate Genotype Imputation Method for the Next Generation of Genome-Wide Association Studies. PLOS Genet. 2009. Jun 19;5(6):e1000529. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Consortium T1000GP. A global reference for human genetic variation. Nature. 2015. Oct;526(7571):68–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Martinez G. Trends and patterns in menarche in the United States: 1995 through 2013–2017 [Internet]. US Department of Health and Human Services. Centers for Disease Control and Prevention. National Center for Health Statistics; 2020. Sep [cited 2020 Nov 10] p. 12. Available from: http://www.cdc.gov/nchs/data/nhsr/nhsr146-508.pdf [Google Scholar]
- 17.Krieger N, Kiang MV, Kosheleva A, Waterman PD, Chen JT, Beckfield J. Age at menarche: 50-year socioeconomic trends among US-born black and white women. Am J Public Health. 2015. Feb;105(2):388–97. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.McDowell MA, Brody DJ, Hughes JP. Has Age at Menarche Changed? Results from the National Health and Nutrition Examination Survey (NHANES) 1999–2004. J Adolesc Health. 2007. Mar 1;40(3):227–31. [DOI] [PubMed] [Google Scholar]
- 19.Chang CC, Chow CC, Tellier LC, Vattikuti S, Purcell SM, Lee JJ. Second-generation PLINK: rising to the challenge of larger and richer datasets. GigaScience [Internet]. 2015. Dec 1 [cited 2021 Feb 25];4(s13742-015–0047-8). Available from: 10.1186/s13742-015-0047-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Purcell S, Chang CC. PLINK 1.9 www.cog-genomics.org/plink/1.9/.
- 21.Dudbridge F, Gusnanto A. Estimation of significance thresholds for genomewide association scans. Genet Epidemiol. 2008;32(3):227–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.MacArthur J, Bowler E, Cerezo M, Gil L, Hall P, Hastings E, et al. The new NHGRI-EBI Catalog of published genome-wide association studies (GWAS Catalog). Nucleic Acids Res. 2017. Jan 4;45(Database issue):D896–901. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Fernández-Rhodes L, Malinowski JR, Wang Y, Tao R, Pankratz N, Jeff JM, et al. The genetic underpinnings of variation in ages at menarche and natural menopause among women from the multi-ethnic Population Architecture using Genomics and Epidemiology (PAGE) Study: A trans-ethnic meta-analysis. PLOS ONE. 2018. Jul 25;13(7):e0200486. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Tanikawa C, Okada Y, Takahashi A, Oda K, Kamatani N, Kubo M, et al. Genome wide association study of age at menarche in the Japanese population. PloS One. 2013;8(5):e63821. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Horikoshi M, Day FR, Akiyama M, Hirata M, Kamatani Y, Matsuda K, et al. Elucidating the genetic architecture of reproductive ageing in the Japanese population. Nat Commun. 2018. May 17;9(1):1977. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Demerath EW, Liu C-T, Franceschini N, Chen G, Palmer JR, Smith EN, et al. Genome-wide association study of age at menarche in African-American women. Hum Mol Genet. 2013. Aug 15;22(16):3329–46. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.ZDBF2 zinc finger DBF-type containing 2 [Homo sapiens (human)] - Gene - NCBI [Internet]. [cited 2020 Feb 11]. Available from: https://www.ncbi.nlm.nih.gov/gene/57683
- 28.GPR1-AS GPR1 antisense RNA [Homo sapiens (human)] - Gene - NCBI [Internet]. [cited 2020 Feb 11]. Available from: https://www.ncbi.nlm.nih.gov/gene/?term=GPR1%E2%88%92AS
- 29.Fagerberg L, Hallström BM, Oksvold P, Kampf C, Djureinovic D, Odeberg J, et al. Analysis of the human tissue-specific expression by genome-wide integration of transcriptomics and antibody-based proteomics. Mol Cell Proteomics MCP. 2014. Feb;13(2):397–406. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Zmetakova I, Kalinkova L, Smolkova B, Horvathova Kajabova V, Cierna Z, Danihel L, et al. A disintegrin and metalloprotease 23 hypermethylation predicts decreased disease-free survival in low-risk breast cancer patients. Cancer Sci. 2019. May;110(5):1695–704. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Ma R, Tang Z, Sun K, Ye X, Cheng H, Chang X, et al. Low levels of ADAM23 expression in epithelial ovarian cancer are associated with poor survival. Pathol Res Pract. 2018. Aug;214(8):1115–22. [DOI] [PubMed] [Google Scholar]
- 32.Home - SNP - NCBI [Internet]. [cited 2021 Mar 23]. Available from: https://www.ncbi.nlm.nih.gov/snp/
- 33.UCSC Genome Browser Home [Internet]. [cited 2021 Mar 23]. Available from: https://www.genome.ucsc.edu/
- 34.Delahanty RJ, Beeghly-Fadiel A, Long JR, Gao YT, Lu W, Xiang YB, et al. Evaluation of GWAS-identified genetic variants for age at menarche among Chinese women. Hum Reprod Oxf Engl. 2013. Apr;28(4):1135–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Huo D, Zheng Y, Ogundiran TO, Adebamowo C, Nathanson KL, Domchek SM, et al. Evaluation of 19 susceptibility loci of breast cancer in women of African ancestry. Carcinogenesis. 2012. Apr 1;33(4):835–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.National nutrition and health survey (NHHS) 2018. A report on the nutrition and health situation of Nigeria [Internet]. National Bureau of Statistics, National Population Commission, and the Nigeria Federal Ministry of Health; 2018. [cited 2020 Nov 10] p. 161. Available from: https://www.unicef.org/nigeria/media/2181/file/Nigeria-NNHS-2018.pdf [Google Scholar]
- 37.Barros B de S, Kuschnir MCMC, Bloch KV, Silva TLN da. ERICA: age at menarche and its association with nutritional status. J Pediatr (Rio J). 2019. Feb;95(1):106–11. [DOI] [PubMed] [Google Scholar]
- 38.Cooper C, Kuh D, Egger P, Wadsworth M, Barker D. Childhood growth and age at menarche. BJOG Int J Obstet Gynaecol. 1996;103(8):814–7. [DOI] [PubMed] [Google Scholar]
- 39.Rogers IS, Northstone K, Dunger DB, Cooper AR, Ness AR, Emmett PM. Diet throughout childhood and age at menarche in a contemporary cohort of British girls. Public Health Nutr 2010. Dec;13(12):2052–63. [DOI] [PubMed] [Google Scholar]
- 40.Khan AD, Schroeder DG, Martorell R, Haas JD, Rivera J. Early childhood determinants of age at menarche in rural guatemala. Am J Hum Biol. 1996;8(6):717–23. [DOI] [PubMed] [Google Scholar]
- 41.Bosch AM, Willekens FJ, Baqui AH, Ginneken JKSV, Hutter I. ASSOCIATION BETWEEN AGE AT MENARCHE AND EARLY-LIFE NUTRITIONAL STATUS IN RURAL BANGLADESH. J Biosoc Sci 2008. Mar;40(2):223–37. [DOI] [PubMed] [Google Scholar]
- 42.Berkey CS, Gardner JD, Lindsay Frazier A, Colditz GA. Relation of Childhood Diet and Body Size to Menarche and Adolescent Growth in Girls. Am J Epidemiol. 2000. Sep 1;152(5):446–52. [DOI] [PubMed] [Google Scholar]
- 43.Lundblad MW, Jacobsen BK. The reproducibility of self-reported age at menarche: The Tromsø Study. BMC Womens Health [Internet]. 2017. Aug 22 [cited 2020 Sep 25];17. Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5568259/ [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
[Dataset] Huo D, Feng Y, Haddad S, et al. Data from: Genome-wide association studies in women of African ancestry identified 3q26.21 as a novel susceptibility locus for oestrogen receptor negative breast cancer. Hum Mol Genet. 2016 Nov 1;25(21):4835–46. The genotype data of Root genome-wide association study are posted in dbGaP (https://www.ncbi.nlm.nih.gov/projects/gap/cgi-bin/study.cgi?study_id=phs000383.v1.p1 ).