Genetics of Menopause Timing: Findings from the ReproGen Consortium

MaryFran Sowers Memorial LectureApril 3, 2018

Joanne M. Murabito MD ScM

Associate Professor of Medicine

Boston University School of Medicine

Presentation Overview

• Relation of age at menopause to later life health

• Genetic association studies of age at natural menopause

• Follow-up analyses to understand the biologic relevance of the genetic associations & the link to health

http://rinkwestmedical.co.za/wp-content/uploads/2013/08/Womens-health-640x350.jpg

Can Age At Natural Menopause (ANM) as a phenotype uncover important biology?

“the genetic associations…. should provide insight into timing of menopause

but also may yield needed breakthroughs in understanding the genetic

variation conferring risk of breast cancer, cardiovascular disease and other

health events of later adulthood” Patricia Hartge, NCI Nature Genetics 2009

CVD Mortality Trends: US Men and Women

Mehta LS Circ 2016 Acute MI in Women

1 in 3 deaths in women attributed to CVD

Benjamin EJ et al Heart Disease and Stroke Statistics 2018 Update

Ann Intern Med 1976

Women Aged <45 Years at Onset of Menopause vs Women >45 Years at Onset

Muka T et al 2016 JAMA Cardiol

Risk of Mortality According to Menopause Timing

CVD According to Age at Menopause: Nurses’ Health Study

Ley SH et al 2017 JAHA

Coronary Disease Stroke

Wellons M Menopause 2012;19:1081

Menopause Timing & Cardiovascular Disease

Multi-Ethnic Study of Atherosclerosis

Early Menopause Early Menopause

SWAN: Cholesterol by Menopausal Status

Darby CA AJE 2009

Rates of CVD and Breast Cancer in Women by Race/Ethnicity

Mehta LS, et al,2018 Circ AHA Statement: CVD and Breast cancer

Menopause linked to epigenetic age

• DNA methylation age or epigenetic age is biomarker of human aging

• Predicts mortality; potential risk marker CVD

• Age at menopause associated with epigenetic age• Older epigenetic age associated with younger age at menopause

• GWAS of two epigenetic age traits recently published• Age at menopause genetically correlated with epigenetic age

• Each one year earlier age at menopause associated with 0.4 year higher intrinsic epigenetic age (p=3.5 x 10-3)

Levine ME et al 2016 PNAS; Lu AT et al 2018 Nat Comm; Roetker NS, et al 2018 Circ Genom Precis Med

Genetics of Menopause Timing• Identify genetic variants associated with menopause timing

• Provide mechanistic insights into ovarian aging

• Improve understanding of how menopause timing influences later life health

Menopause Age Has High Familial Concordance & Heritability

• Women with mother or sister with early menopause (< 45 years) had 6 fold increase in early menopause

• Women with mother or sister with late menopause (> 54 years) at increased risk late menopause

• Heritability of ANM ~ 50% → 50% of the variation in ANM is due to genetic factors

• Variation in ANM explained by smoking, parity, SES, decade of birth only 5%

Morris DH: 2011 Menopause 18:956

Age At Natural Menopause Trait Definition

• Have your periods stopped for one year or more?

• Age period stop?

• Cause periods stop→ natural

• Menopause normally occurs between 40 to 60 years Reported Age at Natural Menopause:

Framingham Heart Study Women

0

5

10

15

20

25

18 22 26 30 34 38 42 46 50 54 58 62

Age (years)

Perc

ent (%

)

POI

Murabito, et al 2005 JCEM

Self-report and prospective age at menopause are highly concordant

Comparison of self-reported data

• NSHD recruited at birth followed regular intervals

• Million Women Study recruited in middle age

• 541 women in both studies of similar age

• Compared self-reported age at menopause in the two studies

• Mean age 48 years

Cairns BJ et al 2011 BMC Med Res Methodol

ReproGen Consortium: Menopause

• International: US, Europe, Australia

• GWAS of age at natural menopause (ANM)

• Women of European ancestry

• 22 studies in the discovery stage, n=38,968

• 21 studies in the replication stage, n=14,435

• Age at natural menopause between 40 - 60 years• Mean age at natural menopause 48.4 to 50.8 across studies

• Excluded surgical menopause, chemo/radiation, women using HRT or OCP

Stolk, et al Nature Genetics 2012; http://www.reprogen.org/index.html

RHBDL2

EXO1TLK1111 HELQ

TRD3

UIMC1

MPPED2

PRIM1

KPNA3

POLG

NLRP11

Meta-analysis Results for Menopause: Highlight DNA repair & Immune pathways

Stolk L et al NG 2012; Stolk L,et al NG 2009; He C, et al NG 2009

MCM8

TMEM150B, BRSK1

SYCP2L

PRRC2A

ASH2L

FNDC4

In LD with SNPs in FSHBIn LD with SNPs in FSHBIn LD with SNPs in FSHBIn LD with SNPs in FSHB

• MCM8 mutations linked to autosomal recessive disorder: primary amenorrhea, hypothyroidism,

hypergonadotropic hypogonadism

• MCM8 complexes with MCM9 to function in homologous recombination (HR) mediated DNA repair

AlAsiri S et al 2014 J Clin Invest; Bouali N 2017 Fertilty Sterility; Desai S 2017 J Clin Endo Metab

Expanded Menopause GWAS & Exomechip Investigation

• Common and low-frequency protein-coding variation

• 33 studies, ≈69,000 women

• Exomechip genotyping, 22 studies, n=39,000

• Follow-up analyses�Pathway analysis

�Mendelian Randomization

HapMap & Exomechip SNP Associations for Menopause

HELB

SLCO4A1

54 independent SNPs in 44 genomic loci2 rare coding variants

Day FR, et al 2015 Nature Genetics

Common & exome chip variants at SLCO4A1 appeared independent

Day FR, et al. 2015 NG

Multiple Signals at HELB

(DNA helicase B)Top Exomechip signal

maps to an acidic motif & results in

replacement of an aspartate by a

nonpolar glycine residue affects

binding of the helicase.

Biologic Pathway analyses MAGENTA & GRAIL

• We also tested for signal enrichment using pre-identified biologic pathways using all SNPs in the GWAS

• Custom pathways• Primary ovarian insufficiency

• Ovarian function

• Monogenic disorders of puberty

ANM SNPs strongly enriched DNA damage response (DDR) pathways

Adapted from Jackson & Bartek Nature 2009

29/44

regions

contain DDR

gene (s)

STRING Network Analysis:

BRCA1 linked to 15 genes from Menopause GWAS

Common variant in BRCA1 associated with ANM

http:// string-db.org Search Tool for Retrieval of Interacting Genes

Potential Mechanism Linking DDR pathway to Menopause

Oktay K et al 2014 Am Soc Clin OncOktay K et al 2014 Am Soc Clin Onc

ANM SNPs enriched for Primary Ovarian Insufficiency Genes

EIF2B4

MSH5

TDRD3

POLG

MCM8

Day FR, et al 2015 Nature Genetics

Significant enrichment of all ANM SNPs with monogenic puberty gene set (p=0.01)

• TAC3

• CHD7

• KISS1R

• FGFR1

• SOX10

� Animal models show signaling in ovary

(TAC3, KISS1R) in addition to

hypothalamus

� Expression TAC3 and KISS1 in human

female reproductive tissues

� May indicate a neural influence on the

timing of ovarian follicular aging

5 menopause SNPs in/near genes for hypogonadotropic hypogonadism

Is there a link between ANM variants and breast cancer?

• Known epidemiologic link between menopause age and breast cancer risk

• We observed an enrichment with DNA damage response genes

• Would risk be related to this pathway?

• Collaboration with Breast Cancer Association Consortium (BCAC)

• Across the 56 ANM-SNPs there was a positive correlation between effect sizes in ANM and the effect sizes for breast cancer risk in the BCAC

Test causal relation between ANM & breast cancer using Mendelian Randomization

Robinson PC, et al. Nature Reviews Rheum 2016

Use genetic risk scores as the score explains a larger proportion of the

variance in the exposure than any single variant

Mendelian Randomization

• Polygenic Risk Score (PRS):• 56 ANM SNPs (54 HapMap + 2 Exome)

• Weighted by each SNP’s effect size on menopause

• > 40,000 breast cancer cases & >40,000 controls from BCAC

• Tested association between PRS and breast cancer using logistic regression

• 1-unit increase in PRS = 1 year increase in genetically predicted age at natural menopause

PRS and Breast Cancer Risk:Genetically predicted 1-year increase ANM

Odds Ratio

Confidence Interval

P-value

All SNPs 1.06 1.05, 1.08 2.78 x 10-14

OR=1.030 (1.026-1.034)Largest pooled analysis epidemiological studies

Odds Ratio

Confidence Interval

P-value

38 DDR SNPs 1.05 1.03, 1.08 1.06 x 10-7

17 Non-DDR SNPs 1.12 1.06, 1.21 7.84x 10-10

P het= 0.01

Quintiles of menopause PRS

ER positiveOR=1.07 (1.05-1.10)

P=1.73x10-12

ER negativeOR=1.03 (1.00-1.07)

P=0.04

Genetically predicted ANM had a larger effect on ER-positive than ER-negative Breast Cancer

Bre

as

t C

an

ce

r O

R

Potential Mechanisms of SNPs on Breast Cancer Risk

What about an association with CVD?Model earlier menopause when constructing Genetic Risk Score

ANM Genetic Risk Score and First CVD Event

Cohort HR 95% CI p-value

FHS 1.22 1.08, 1.37 0.001

ARIC 1.06 0.93, 1.22 0.36

RSI 1.06 0.99, 1.14 0.11

RSII 1.27 0.95, 1.69 0.11

Meta-analysis 1.10 1.04, 1.16 9.7 x 10-4

Adjusted for age, PCs, family relatedness, study center and CVD

risk factors

1-unit increase in GRS corresponds to a 1 year decrease in ANM

Sarnowski C 2017 Menopause

Genetic Correlation between ANM & GWAS traits

Phenotype Reg. Coef. SE P-value

CAD -0.22 0.09 0.011

BMI -0.13 0.04 0.002

Weight -0.12 0.04 0.003

Waist circumference -0.12 0.05 0.018

Hip circumference -0.12 0.04 0.001

T2DM -0.06 0.07 0.42

Women

Sarnowski C, et al 2017 Menopause

Are the ANM genes associated with other reproductive traits?

FSHB ANM SNPAssociated with Other Reproductive Traits

Trait SNP Direction Study

ANM rs11031006 Earlier ANM Day FR 2015

Menstrual cycle length rs564036233 increase Ruth KS, 2016

PCOS rs11031006 Day FR 2015

LH in PCOS rs11031006 Increase levels Hayes 2015

Dizygotic twinning rs11031006 Mbarek 2016

Endometriosis rs74485684 Sapkota 2017

http://www.ebi.ac.uk/gwas/search?query=FSHB

1000G Meta-analysis for Menopause in progress

• Will include full UKBiobank sample

• Preliminary results: more than doubled the number of signals

• Continue to see signals in DDR pathway

• Gene x environment interactions

• Mendelian Randomization to examine additional disease links such as osteoporosis

https://handluggageonly.co.uk/2015/09/15/so-heres-a-thought/Strong diverse women of the world

Are loci influencing menopause in White women important in women of other ancestries?

• Few GWAS of menopause conducted in women of diverse backgrounds

• Race/ethnic diversity in genetic background

• Risk for many important health outcomes

• Distinct reproductive profiles

• Genetic architecture underlying reproductive aging may differ across ancestry groups

GWAS ANM in African American Women

Collaboration across 11 U.S. studies

N=6510 women

• CARe: ARIC, JHS, MESA,CHS

• WHI

• AA Breast Cancer Consortium

• GENOA, HABC, MESA family

• Mean ANM: 48.8 to 50.2 years

No genome-wide significant associations

Inter-genic SNP on chr 17 lowest p=1.6 x 10-6

Chen C. et al 2014 HMG

Replication of ANM loci identified in EA women in African American Women

• Evaluated SNPs in 30 loci

• Many loci were associated in AA women, directionally consistent

• Of note, the MCM8 SNP in EA women with the largest effect size did not replicate in AA women→ uncovered a second signal

• MAF differences (7% EA women vs. 1.3% in AA women)

• Effect size differences (1.07 EA women vs 0.35 AA women)

What about women of other ancestries?

• Data sparse, small samples, limited number of SNPs/genes tested

• Population Architecture using Genomics & Epidemiology Study

• 5 ANM SNPs genotyped in American Indians, AA, Asians, Hispanics, Native Hawaiians

• Replicated 3 SNPs, most robust for MCM8 SNP:

Carty et al 2013 Hum Reprod

rs16991615 N CAF Beta P-value

EA 10955 0.07 0.82 7.8 x 10-15

AA 2129 0.02 0.57 0.22

Asian 1978 0.0001 1.25 0.46

Hispanic 1143 0.06 1.05 0.0004

Am Indian 934 0.02 2.63 0.001

N Hawaiian 294 0.03 -0.41 0.60

Trans-ethnic meta-analysis in progressAA women: n=5710 menopause, n=14488 menarche

Hispanic women: n=3706 menopause, n=10661 menarche

Asian women: n=1838 menopause, n=4350 menarche

Are the ANM genes associated with biomarkers of ovarian reserve?

Ovarian Aging Study: Markers of Ovarian Reserve

• Community-based cohort of women recruited from eligible KP members

• Evaluated markers of ovarian reserve• FSH

• AMH

• Antral follicle count

• No genome-wide significant results

• MCM8 SNP associated with later ANM, associated with higher follicle count (p=0.018)

Schuh-Huerta SM, et al Hum Reprod 2012; Schuh-Huerta SM et al Hum Genet 2012

Conclusions and Future Directions

• Genetic association studies have identified variants for menopause timing linked to later life health conditions

• Whole Genome Sequencing leveraging TOPMed resource

• Further studies in women of diverse ancestry are needed• New genetic variants in non-European populations, using better

genome coverage of rare variants specific to each population

• Functional studies are needed to identify causal variants and to further our understanding of the biologic mechanisms of menopause timing and links to health outcomes

ReproGen Consortium Investigators

Felix DayKatherine RuthDeborah ThompsonKathryn LunettaNatalia PervjakovaDaniel ChasmanLisette StolkHilary FinucanePatrick SulemBrendan Bulik-SullivanTonu EskoCathy ElksAndrew JohnsonKari StefanssonJenny Visser Ken OngJenny Chang-ClaudeJohn RB PerryAnna Murray

FundingAG029451AG032598

Women Aged 45-49 Years at Onset of

Menopause vs Women >50 Years at Onset Women Aged <45 Years at Onset of

Menopause vs Women >45 Years at Onset

Muka T et al 2016 JAMA Cardiol

Risk of Mortality According to Menopause Timing

Women Aged <45 Years at Onset of Menopause vs Women >45 Years at Onset

Muka T et al 2016 JAMA Cardiol

Risk of CVD Mortality According to Menopause Timing

Menopause & Type 2 Diabetes

• Epic-InterAct study

• Country specific hazard ratios of type 2 diabetes per standard deviation decrease in menopausal age

Brand JS 2012 Diabetes Care

Multiple Signals at HELB

(DNA helicase B) Top Exomechip

signal maps to an acidic motif & results

in replacement of an aspartate by a

nonpolar glycine residue affects

binding of the helicase.