age at menarche and type 2 diabetes risk 2013

8/10/2019 Age at Menarche and Type 2 Diabetes Risk 2013

1/9

Age at Menarche and Type 2 DiabetesRiskThe EPIC-InterAct study

CATHY E. ELKS, PHD1KEN K. ONG, FRCPCH, PHD1

ROBERT A. SCOTT, PHD1

YVONNE T. VAN DERSCHOUW, PHD2

JUDITH S. BRAND, MSC2

PETRAA. WARK, MSC, PHD3

PILARAMIANO, MSC4,5,6

BEVERLEY BALKAU, PHD7,8

AURELIO BARRICARTE, MD, PHD6,9

HEINERBOEING, PHD10

ANA FONSECA-NUNES, MSC11

PAULW. FRANKS, PHD12,13

SARA GRIONI, BSC14

JYTTEHALKJAER, PHD15

RUDOLF KAAKS, PHD16

TIMOTHY J. KEY, DPHIL17KAY TEEKHAW, MBBCHIR, FRCP18

AMALIA MATTIELLO, MD19

PETERM. NILSSON, MD, PHD12

KIMOVERVAD, PHD20,21DOMENICO PALLI, MD, MPH22

J. RAMN QUIRS, MD23

SABINA RINALDI, PHD24

OLOV ROLANDSSON, MD, PHD25

ISABELLE ROMIEU, PHD24

CARLOTTA SACERDOTE, MD, PHD26,27

MARIA-JOS S ANCHEZ, MD, PHD6,28

ANNEMIEKE M.W. SPIJKERMAN, PHD29

ANNE TJONNELAND, DRMEDSCI15

MARIA-JOSE TORMO, MD, PHD6,30,31

ROSARIO TUMINO, MD, MSC, DLSHTM32,33

DAPHNE L. VAN DERA, PHD29

NITAG. FOROUHI, PHD, MRCP, FFPH1

STEPHEN J. SHARP, MSC1

CLAUDIA LANGENBERG, MD, PHD1ELIO RIBOLI, MD, MPH, SCM3

NICHOLAS J. WAREHAM, MD, PHD1

THEINTERACTCONSORTIUM

OBJECTIVEdYounger age at menarche, a marker of pubertal timing in girls, is associated withhigher riskof later type 2 diabetes. We aimed to conrm this association andto examine whetherit is explained by adiposity.

RESEARCH DESIGN AND METHODSdThe prospective European Prospective Investi-gation into Cancer and Nutrition (EPIC)-InterAct case-cohort study consists of 12,403 incident type2 diabetescasesand a stratied subcohortof 16,154 individuals from26 research centers across eightEuropeancountries. We tested the association between age at menarche and incident type 2 diabetes

using Prentice-weighted Cox regression in 15,168 women (n= 5,995 cases). Models were adjustedin a sequential manner for potential confounding and mediating factors, including adult BMI.

RESULTSdMean menarcheal age ranged from 12.6 to 13.6 years across InterAct countries.Each year later menarche wasassociatedwith 0.32 kg/m2 loweradultBMI. Women in theearliestmenarche quintile(811 years, n = 2,418) had70% higherincidence of type 2 diabetes comparedwith those in the middle quintile (13 years,n= 3,634), adjusting for age at recruitment, researchcenter, and a range of lifestyle and reproductive factors (hazard ratio [HR], 1.70; 95% CI, 1.491.94;P, 0.001). Adjustment for BMI partially attenuated this association (HR, 1.42; 95% CI,1.181.71; P, 0.001). Later menarche beyond the median age wasnot protective against type 2diabetes.

CONCLUSIONSdWomen with history of early menarche have higher risk of type 2 diabetesin adulthood. Less than half of this association appears to be mediated by higher adult BMI,suggesting that early pubertal development also may directly increase type 2 diabetes risk.

Diabetes Care36:3526

3534, 2013

The inuence of etiological factorsunderlying type 2 diabetes risk maybegin many years before the mani-

festation of disturbed glycemic control.Biological processes implicated in puber-tal development may be one example ofearly life determinants of later diseaserisk. Age at menarche, the age at onsetofrst menstruation in girls, is frequentlyused as a marker of puberty timing in epi-demiological studies to explore the asso-ciation between developmental tempoand later disease risk. Age at menarche

represents a distinct event in puberty, isusually well recalled into adulthood, andtherefore is a convenient noninvasivemeasure of pubertal timing (1). Earliermenarche has been associated with ad-verse outcomes, including obesity, car-diovascular disease, some types of cancer,and mortality, demonstrating potentiallong-lasting effects of the timing of pubertaldevelopment on health (2).

Increasing rates of type 2 diabetes inrecent decades (3) have occurred in par-allel with a secular decline in the aver-age age at menarche (4). Although themarked increases in diabetes incidenceare undoubtedly driven by dramaticchanges in environment and lifestyle,these concurrent trends may, to some ex-tent, reect a biological link between pu-berty timing and diabetes risk. In supportof a direct link between puberty timingand glucose regulation, a recent studyshowed that the insulin-sensitizing agentmetformin delays menarche in girls withprecocious pubarche (5). A number ofstudies have reported adverse metabolicconsequences of early sexual maturation

(6

9). However, some smaller studieshave not found this association, possiblybecause of insufcient power (10,11).

c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c

From the 1Medical Research Council EpidemiologyUnit, Institute of Metabolic Science,AddenbrookesHospital, Cambridge, U.K.; the 2Julius Center forHealth Sciences and Primary Care, UniversityMedical Center Utrecht, Utrecht, the Nether-lands; the 3Department of Epidemiology andBiostatistics, School of Public Health, ImperialCollege London, St. Marys Campus, NorfolkPlace, London,U.K.; the 4PublicHealth Division of

Gipuzkoa, Basque Government, San Sebastian,Spain; the 5Instituto BIO-Donostia, Basque Gov-ernment, San Sebastian, Spain; the 6Consortiumfor Biomedical Research in Epidemiology andPublic Health (CIBER Epidemiologa y SaludPblica), Madrid, Spain; 7INSERM, CESP Centrefor Research in Epidemiology and PopulationHealth, U1018: Epidemiology of Diabetes, Obe-sity, and Chronic Kidney Disease Over the Life

Course, Villejuif Cedex, France; the 8UniversitParis-Sud, UMRS 1018, Villejuif, France; the9Navarre Public Health Institute, Pamplona,Navarra, Spain; the 10Department of Epidemi-ology, German Institute of Human Nutrition,Potsdam-Rehbruecke, Nuthetal, Germany; the11Department of Epidemiology, Nutrition, En-vironment, and Cancer Unit, Catalan Institute ofOncology,LHospitalet de Lolgbregat, Barcelona,

3526 DIABETESCARE, VOLUME36, NOVEMBER2013 care.diabetesjournals.org

E p i d e m i o l o g y / H e a l t h S e r v i c e s R e s e a r c h

O R I G I N A L A R T I C L E


2/9

Others have reported associations be-tween earlier menarche timing and in-termediate quantitative traits such asincreased blood glucose levels (11,12),impaired glucose tolerance (13), and insu-lin resistance (14,15), further supporting alink between early pubertal developmentand an adverse metabolic prole.

It is conceivable that part of the as-sociation between menarche timing andtype 2 diabetes risk is explained by in-creased adiposity. It is well establishedthat younger age at menarche is associ-ated with higher BMI in later life (16), andelevatedBMI itselfis a majorrisk factorfortype 2 diabetes(17). One study reported acomplete attenuation of the associationbetween age at menarche and diabetes af-ter accounting for BMI (18); however, no-tably, that study relied on self-reportedBMI. In the smaller cross-sectionalKORA study, the association betweenyounger age at menarche and risk oftype 2 diabetes remained after adjustmentfor measured BMI (9). In addition to theuncertainty about the degree to which ad-iposity explains the relationship betweenmenarche and diabetes risk, there is fur-ther uncertainty regarding the shape ofthe association. Although earlier menar-che is associatedwith higher risk of type 2diabetes, it is unclear whether older age atmenarche beyond the average is associ-ated with lower risk. Nonlinear asso-ciations between age at menarche and

cardiovascular disease and mortalityhave been observed previously (19). Wesought to investigate the association be-tween age at menarche and type 2 diabetesincidence in a large European prospectivestudy, and to explore whether observed as-sociations were mediated by adult BMI.

RESEARCH DESIGN ANDMETHODS

Participants and study designInterAct is a large, prospective, case-cohort study of 27,779 individuals fromeight European countries (Denmark,France, Germany, Italy, the Netherlands,

Spain, Sweden, and U.K.) nested withinthe European Prospective Investigationinto Cancer and Nutrition (EPIC) (20).The objective of the study was to investi-gate the interplay between genetic andlifestyle factors and the effect on risk oftype 2 diabetes (21). The sample wastaken from a total cohort of 340,234individuals comprising 3.99 millionperson-years of follow-up time; 12,403individuals (6,238 women) withclinicallyincident type 2 diabetes were selectedfor InterAct together with a representa-tive subcohort of 16,835 individuals(10,378 women). After exclusion ofthose with prevalent diabetes (n= 548),those without information regarding di-abetes status (n = 129), and individualswith postcensoring diabetes (n = 4),16,154 subcohort members were avail-able for analysis (10,043 women), 778 ofwhom were also incident cases (394women). All EPIC centers obtained writ-ten informed consent from participantsand obtained approval from local ethicscommittees.

Type 2 diabetes case ascertainmentAs previously described (21), cases of in-cident type 2 diabetes were ascertainedusing multiple sources of evidence, in-cluding self-report (self-reported or doctor-diagnosed history of type 2 diabetes ordiabetes drug use), linkage to primary or

secondary care registers, medication use re-ported from drug registers, hospital ad-missions, and mortality data. In mostcenters, evidence of type 2 diabetes wassought from at least two independentsources from those listed or by reviewingindividual medical records to verify inci-dent cases. The exceptions were Danish

and Swedish centers where cases wereascertained through local and nationaldiabetes and pharmaceutical registersrather than self-reported; thus, all caseswere considered to be veried. Informa-tion regarding diabetes status was usedfrom any follow-up visit or external evi-dence with a date later than that of baselinedata collection. Follow-up was censored atthe date of diagnosis, 31 December 2007,or the date of death, whichever occurredrst. If a diagnosis date could not be iden-tied from any of the sources described,then the midpoint between recruitmentand censoring was used.

Ascertainment of age at menarcheInformation regarding age at menarche incompleted whole years was obtained byquestionnaire in each of the contribut-ing centers. The normal physiological agerange at menarche was considered tobe between 8 and 18 years. A total of34 women were excluded because theyreported menarche after age 18 years,which may reect an underlying patho-logical cause. Another 685 women were

excluded because they did not provideinformation about age at menarche. Thisleft 15,168 women, of whom 5,955 wereconsidered incident type 2 diabetes ca-ses and 9,590 were represented in thesubcohort (377 of whom became type2 diabetes cases). Early menarche was

c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c

Spain; the 12Department of Clinical Sciences,Clinical Research Center, Skne UniversityHospital, Lund University, Malm, Sweden; the13Department of Public Health and ClinicalMedicine, Ume University, Ume, Sweden; the14Fondazione IRCCS Istituto Nazionale TumoriMilan, Via Venezian, Milan, Italy; the 15DanishCancer Society Research Center, Copenhagen,Denmark; the 16Division of Cancer Epidemiol-ogy, German Cancer Research Centre (DKFZ),Heidelberg, Germany; the 17Cancer Epidemiol-ogy Unit, Nufeld Department of Clinical Med-icine, University of Oxford, Oxford, U.K.; the18Department of Public Health and Primary Care,University of Cambridge, Addenbrookes Hospi-tal, Cambridge, U.K.; the 19Dipartimento diMedicina Clinica e Chirurgia, Federico II Uni-versity, Naples, Italy; the 20Department of PublicHealth, Section for Epidemiology, Aarhus

University, Aarhus, Denmark; the 21Departmentof Cardiology, Aalborg Hospital, Aarhus Uni-versity Hospital, Aalborg, Denmark; the 22Molec-

ular and Nutritional Epidemiology Unit, ISPO PonteNuovo, Florence, Italy; the 23Consejera de Sanidad,Public Health Directorate, Oviedo-Asturias, Spain;the 24International Agency for Research on Cancer,Lyon Cedex, France; the 25Department of PublicHealth and Clinical Medicine, Ume University,Family Medicine, Ume University, Ume, Sweden;the26Center for Cancer Prevention (CPO-Piemonte),Turin, Italy; the 27Human Genetics Foundation,Turin, Italy; the 28Andalusian School of PublicHealth, Granada, Spain; the 29National Institute forPublic Health and the Environment (RIVM), Bilt-hoven, the Netherlands; the 30Department of Epi-demiology, Murcia Regional Health Council, Rondade Levante, Murcia, Spain; the 31Department ofHealth and Social Sciences, Universidad de Murcia,

Campus de Espinardo, Spain; the 32Cancer Registryand Histopathology Unit, CivileM.P. ArezzoHospital, Ragusa, Italy; and the 33Associazone

Iblea per la Ricerca EpidemiologicaOnlus, Ragusa,Italy.

Corresponding author: Cathy E. Elks, [email protected].

Received 22 February 2013 and accepted 23 May2013.

DOI: 10.2337/dc13-0446This article contains Supplementary Data online at

http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc13-0446/-/DC1.

2013 by the American Diabetes Association.Readers may use this article as long as the work isproperly cited, the use is educational and not forprot, and the work is not altered. See http://creativecommons.org/licenses/by-nc-nd/3.0/ fordetails.

care.diabetesjournals.org DIABETESCARE, VOLUME36, NOVEMBER2013 3527

Elks and Associates
mailto:[email protected]:[email protected]://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc13-0446/-/DC1http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc13-0446/-/DC1http://creativecommons.org/licenses/by-nc-nd/3.0/http://creativecommons.org/licenses/by-nc-nd/3.0/http://creativecommons.org/licenses/by-nc-nd/3.0/http://creativecommons.org/licenses/by-nc-nd/3.0/http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc13-0446/-/DC1http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc13-0446/-/DC1mailto:[email protected]:[email protected]


3/9

dened as onset of menstruation occur-ring between ages 8 and 11 years.

Other measurementsBaseline information regarding reproduc-tive factors was collected using self-administered questionnaires, whichasked about age at rst full-term preg-

nancy, number of still births and livebirths, menopausal status, and use of theoral contraceptives and hormone replace-ment therapy. Paritywas derived from thenumber of reported live births and stillbirths. In the Bilthoven center (the Neth-erlands), this information was unavailableand number of children was used. Life-style factors including smoking status,alcohol consumption, and educationallevel also were assessed at baseline by self-report. Levels of physical activity wereassessed using a questionnaire previouslyvalidated against objective methods andcoded as inactive, moderately inactive,moderately active, and active (22). Height,weight, and waist circumference of partic-ipants were measured by trained healthprofessionals, except in the French centersand the U.K. Oxford center, where heightand weight were self-reported. Waist cir-cumference was unavailable in the SwedishUme center. BMI was calculated as weight(in kg) divided by height (in m) squared.Recalled body weight at age 20 years wasavailable in ve countries (Italy, U.K., Ger-many, Sweden, and Denmark) and was

used to calculate estimated BMI in youngadulthood (recalled body weight at age20 [kg] divided by height [m] squared).BMI cutoffs of 25 kg/m2 and 30 kg/m2

were used to dene overweight andobesity, respectively, in accordancewith guidelines from the World HealthOrganization (23).

Statistical analysisBaseline characteristics of InterAct studyparticipants were explored by categoriesof age at menarche. The association be-tween age at menarche and adult BMI wastested in the InterAct subcohort usinglinear regression with adjustment for ageat recruitment and center in each countrybefore using random-effects meta-analysisto obtain a pooled estimate.

To estimate the association betweenage at menarche and type 2 diabetes,Prentice-weighted Cox regression mod-els, with age as the underlying time scale,were tted within each country (24).Country-specic hazard ratios (HRs) fortype 2 diabetes per year later age at men-arche were subsequently combined using

random-effects meta-analysis. Heteroge-neity between countries was assessed us-ing the I2 statistic. Because girls previouslyhave been shown to be at increased risk forlater disease according to age at menarchein a nonlinear manner (6,19), we also mod-eled menarche as a categorical variable(811, 12, 13, 14, and 1518 years, which

approximately equate to quintiles) with themedian age (13 years) as the reference cate-gory. A quadratic age at menarche termadded to the existing linear model also wastested for signicance.

Basic models were adjusted for cen-ter, age at recruitment, and date of birthto account for potential cohort effects.

A structured approach then was used toadjust for additional potential confounders.First, lifestyle factors including smokingstatus (never, former, current), alcohol con-sumption (,0.5 g/day, 0.510 g/day, 10.530 g/day, $30.5 g/day), education level(none, primary school, technical schoolor professional school, secondary school,longer education), and physical activitylevel were included. Next, we includedthe following reproductive factors: ageat rst full-term pregnancy (modeledcontinuously), parity (no children, oneor two children, three or more children),menopausal status (premenopausal, per-imenopausal, and postmenopausal), useof the oral contraceptive pill (ever ornever), and use of hormone replacementtherapy (ever or never).

Subsequent models were performedto examine potential mediation by BMI.First, adult BMI was added to the modelthat included all these potential con-founders, and a subsequent model addi-tionally included waist circumference.Finally, in the centerswith this informationavailable, estimated BMIat age 20 years wasincluded in place of adult BMI and waistcircumference. Because information re-garding all covariates was not available inthe full sample of women with menarchedata, for each model we repeated the basicmodel including only individuals with allrelevant covariate information. In addition,because information regarding reproduc-tive and lifestyle factors was missing formany individuals in the InterAct sample,we tested the association between menar-che timing and incident type 2 diabetes in amaximum samplebasic model with andwithout adjustment for adult BMI.

To calculate the proportion of theassociation between early menarche andtype 2 diabetes risk that is mediated byadult BMI, we performed mediation ana-lyses in the InterAct subcohort using

methods previously described for dichot-omous outcomes (25,26). This approachallows estimation of the direct effect ofearly menarche on diabetes risk (i.e.,that driven directly by menarche timingand independently of the mediator) andan indirect effect (i.e., via the mediator) ofearly menarche using logistic regression.

Tests for interaction between early(811 years) versus later (1218 years)menarche and menopausal status, BMI(continuous), obesity (BMI .30 kg/m2),overweight(BMI.25 kg/m2), smoking sta-tus, and physical activity level and the ef-fects on the risk of diabetes were performedin Prentice-weighted Cox regression mod-els adjusted for center, age at recruitment,lifestyle and reproductive factors, andadult BMI. Analyses were performed sep-arately by country and were pooled usingrandom-effects meta-analysis.

To explore whether the associationbetween menarche timing and type 2diabetes differed according to the age ofonset of diabetes, we dened the follow-ing two new outcomes: type 2 diabetesdiagnosed before age 60 years and type 2diabetes diagnosed at age 60 years or older.The association between age at menarcheand each outcome was then estimatedusing the methods described, with follow-upcensored at age 60 years for the analysisof younger-onset diabetes. All analyseswere conducted using Stata version 11.2(StataCorp, College Station, TX).

RESULTS

Study participant characteristicsAge at menarche was normally distrib-uted in both the InterAct incident type 2diabetes cases (mean, 13.08 years; SD,1.73) and subcohort (mean, 13.14 years;SD, 1.58). Women not reporting age atmenarche were slightly older at recruitmentand weighed more than those included inour analyses. Characteristics of the subco-hort by age at menarche are described inTable 1. Those with younger age at menar-che tended to be younger at baseline (likelyreecting the secular decline in menarchealage), heavier, shorter, less active, more ed-ucated, and more likely to have higherparity. Overall, the average age at recruit-ment was 52 years and the mean BMI was25.7 kg/m2. Age at diagnosis of type 2 di-abetes ranged from 32 to 85 years.

Age at menarche and adult BMIin the InterAct subcohortMean ages at recruitment and ages at men-arche of the InterAct subcohort stratied


Age at menarche and risk of type 2 diabetes


4/9

Table 1dBaseline characteristics of the InterAct subcohort by age at menarche

Age at menarche, years

811 12 13 14 1518

Variables 1,392 (14.5) 1,995 (20.8) 2,396 (25.0) 2,166 (22.6) 1,641 (17.1) P Overall subcohort

Age at recruitment,

years 50.1 (8.5) 50.4 (9.1) 51.9 (9.0) 53.5 (8.9) 55.9 (8.7) ,0.0019,590

52.4 (9.1)

Anthropometry

BMI, kg/m2 26.8 (4.8) 25.8 (4.5) 25.6 (4.5) 25.4 (4.5) 25.3 (4.2) ,0.0019,526

25.7 (4.5)

BMI at age 20, kg/m2 22.0 (2.8) 21.7 (2.7) 21.3 (2.7) 21.1 (2.5) 20.8 (2.5) ,0.0013,665

21.3 (2.7)

Waist circumference, cm 82.9 (12.1) 81.1 (11.2) 81.0 (11.0) 80.9 (11.2) 81.0 (11.0) ,0.0019,303

81.3 (11.3)

Height, cm 159.8 (6.5) 161.0 (6.6) 161.4 (6.7) 161.9 (6.9) 162.1 (6.6) ,0.0019,558

161.3 (6.7)

Lifestyle factors,n (%)

Smoking status

Never 758 (54.5) 1,138 (57.0) 1,309 (54.6) 1,200 (55.4) 937 (57.1) 5,342 (56.0)

Former 297 (21.3) 415 (20.8) 533 (22.3) 466 (21.5) 352 (21.5) 2,063 (21.6)

Current 333 (23.9) 430 (21.6) 542 (22.6) 494 (22.8) 337 (20.5) 0.22 2,136 (22.4)Physical activity

Inactive 454 (32.8) 514 (26.0) 660 (27.7) 554 (25.7) 397 (24.4) 2,579 (27.1)

Moderately inactive 487 (35.2) 705 (35.7) 855 (35.9) 795 (36.9) 529 (32.5) 3,371 (35.4)

Moderately active 269 (19.4) 429 (21.7) 474 (19.9) 438 (20.3) 349 (21.4) 1,959 (20.6)

Active 174 (12.6) 327 (16.6) 394 (16.5) 368 (17.1) 355 (21.8) 0.006 1,618 (17.0)Alcohol intake

0 g/day 291 (24.3) 354 (21.5) 422 (22.5) 354 (21.2) 251 (20.1) 1,672 (21.9)

0.510 g/day 686 (57.3) 925 (56.3) 1,033 (55.0) 975 (58.3) 728 (58.3) 4,347 (56.9)

10.530 g/day 201 (16.8) 332 (20.2) 389 (20.7) 313 (18.7) 255 (20.4) 1,490 (19.5)

$30.5 g/day 20 (1.7) 33 (2.0) 35 (1.9) 31 (1.9) 15 (1.2) 0.86 134 (1.8)

Educational level

None 154 (11.2) 158 (8.0) 213 (8.9) 200 (9.3) 148 (9.1) 873 (9.2)

Primary school 423 (30.7) 591 (29.9) 763 (31.9) 754 (35.0) 585 (35.9) 3,116 (33.0)

Technical/professional school 277 (20.1) 445 (22.5) 546 (22.9) 509 (23.6) 450 (27.6) 2,227 (23.6)

Secondary school 246 (17.8) 354 (17.9) 382 (16.0) 315 (14.6) 218 (13.4) 1,515 (16.0)

Longer education 262 (19.0) 415 (21.0) 463 (19.4) 362 (16.8) 219 (13.4) ,0.001 1,721 (18.2)Reproductive factors

Age at rst full-term

pregnancy, years 24.9 (4.5) 25.0 (4.3) 24.9 (4.3) 24.8 (4.3) 25.0 (4.2) 0.79

8,402

24.5 (4.2)

Parity,n (%)0 180 (13.1) 286 (14.6) 312 (13.3) 250 (11.7) 191 (11.8) 1,219 (12.9)

12 802 (58.5) 1,140 (58.2) 1,360 (57.8) 1,238 (57.9) 887 (54.8) 5,427 (57.5)

$3 388 (28.3) 534 (27.2) 683 (29.0) 649 (30.4) 540 (33.4) 0.01 2,794 (29.6)Menopausal status

Premenopausal 575 (42.9) 792 (41.3) 773 (33.4) 615 (29.3) 289 (18.4) 3,044 (32.9)

Perimenopausal 246 (18.3) 351 (18.3) 412 (17.8) 342 (16.3) 235 (15.0) 1,586 (17.2)Postm enopausal 521 (38.8) 776 (40.4) 1,128 (48.8) 1,139 (54.3) 1,048 (66.7) 0.54 4,612 (49.9)

Use of contraceptive pill

Ever 803 (57.9) 1,171 (58.9) 1,358 (56.8) 961 (44.5) 772 (47.3) 5,394 (56.4)

Never 583 (42.1) 818 (41.1) 1,033 (43.2) 1,201 (55.6) 861 (52.7) 0.80 4,167 (43.6)

Use of HRT

Ever 278 (21.5) 425 (23.3) 538 (25.6) 537 (28.1) 446 (31.0) 2,224 (26.0)

Never 1,014 (78.5) 1,398 (76.7) 1,564 (74.4) 1,376 (71.9) 992 (69.0) 0.38 6,344 (74.0)

Data presented as mean (SD) or n (%). Overall cohort,N= 9,590. Pfor trend across categories of age at menarche was calculated from linear regression analysesadjusted for age at baseline (except for analyses for age at recruitment), center, and country. Results considered statistically signicant (P, 0.05) are presented inboldface. HRT, hormone replacement therapy.


Elks and Associates


5/9

by country are presented in Table 2. Meanmenarcheal age ranged from 12.6 years inItaly to 13.6 years in Sweden and Den-mark. Older age at menarche was consis-tently associated with decreased adult BMIacross all countries in the InterAct subco-hort (Table 2). Overall, menarche at eachyear later was associated with 0.32 kg/m2

lower BMI in adulthood (95% CI, 0.270.32;P, 0.001). The magnitude of asso-ciation with estimated BMI at age 20 years(in the ve countries that had this infor-mation available, Italy, U.K., Germany,Sweden, and Denmark; overall n =3,665) was smaller, with menarche ateach year later conferring0.21 kg/m2 lowerBMI (95% CI, 0.150.26; P, 0.001). Het-erogeneity across countries was not statis-tically signicant in either analysis.

Age at menarche and type 2 diabetesThe incidence of type 2 diabetes inwomen with history of early menarchein the InterAct subcohort was 4.3 casesper 1,000 person-years, compared with2.9 cases per 1,000 person-years in thosewith the median age at menarche. Men-arche at each year later was associatedwith a 9% lower risk of development oftype 2 diabetesin a linear model adjustingfor age at recruitment, date of birth, andcenter (HR, 0.91;95% CI,0.880.93; P,0.001) (Table 3, model 1). This associa-tion was largely unchanged after adjust-ing for lifestyle and reproductive factors

(Table 3, models 2 and 3). Quadratic ageat menarche terms were signicant (P,0.001). In models using categorical age atmenarche, compared with women in themiddle category of age at menarche (13years), the risk of development of diabetesin women in the earliest category (811years) was 70% higher (HR, 1.70; 95%CI, 1.481.94;P, 0.001) after account-ing for potential confounding factors.There also was a less pronounced associ-ation with diabetes in women who hadmenarche at age 12 years compared withthe median age (HR, 1.26; 95% CI, 1.101.44; P= 0.001). In contrast, theredid notappear to be a protective effect of olderage at menarche in women with menarchebetween 15 and 18 years (HR, 0.97; 95%CI, 0.841.11 in women; P= 0.63), sug-gesting a nonlinear association betweenmenarche timing and diabetes (Fig. 1).

Mediation of the early menarcheand type 2 diabetes associationby BMI

Adjusting for adult BMI partially reducedthe HR for type 2 diabetes in girls with

early menarche (811 years) comparedwith a median menarcheal age from1.70 to 1.42 (95% CI, 1.181.71; P,0.001) in models adjusted for all mea-sured potential confounders (Table 3,model 4). Additional inclusion of waistcircumference into the model had a neg-ligible effect on these results (model 5). Ad-

justingfor BMI at the age of 20 years, ratherthan later adult BMI, only modestly at-tenuated the association between earlymenarche and diabetes from 1.63 (95%CI, 1.302.04) to 1.58 (95% CI, 1.252.01) (model 6). In a maximum sampleanalysis, the addition of adult BMI to thebasic model resulted in a similar attenua-tion in the HR(Table 3, model 7). Bycoun-try,the association between early menarche(811 years compared with all others) andtype 2 diabetes was directionally consistentboth before and after adjustment for adultBMI, with no detectable heterogeneity(Supplementary Fig. 1).

In mediation analyses in the InterActsubcohort, the estimated HR for type 2diabetes in girls with early menarche (811 years) compared with menarche atolder ages (total effect, 1.68; 95% CI,1.272.22) comprised a direct effect ofearly menarche (HR, 1.33; 95% CI,1.001.78) and an indirect effect medi-ated by adult BMI (HR, 1.23; 95% CI,1.181.30). Overall, 41% (95% CI, 3150%) of the higher risk of type 2 diabetesrelated to early menarche in the InterAct

subcohort was attributable to higher adultBMI (Supplementary Table 1).

Interactions with age at menarcheThere was no statistically signicant in-teraction between menarcheal age andmenopausal status, BMI, obesity status,smoking status, or physical activity leveland the effect on the risk of developmentof type 2 diabetes (data not shown).

Age at menarche association withtype 2 diabetes diagnosed beforeor after age 60 yearsThe association between early menarcheand type 2 diabetes diagnosed at age60 years or older (HR, 1.67; 95% CI,1.431.95;P, 0.001) was comparablewith the association with type 2 diabetesdiagnosed before age 60 years (HR, 1.64;95% CI, 1.431.89; P , 0.001) in amodel adjusting for potential confound-ing factors. For both end points, the as-sociation between early menarche andincident diabetes was partially attenuatedby adjustment for adult BMI (Supple-mentary Table 2).

CONCLUSIONSdIna largecase-cohortstudy nested with a large pan-Europeancohort study, we have demonstrated anassociation between early menarche andincreased risk of type 2 diabetes that washighly consistent across eight Europeancountries. Girls in the youngest category ofage at menarche had the highest risk, with

women who were between 8 and 11 yearsat the time of menarche having a 70%higher incidence of future diabetes thanwomen who had menarche at the averageage of 13 years. Although these ndingsare generally consistent with previousreports (69,18), the large size of ourstudy allowed us to demonstrate thenonlinear nature of the association andto formally quantify the extent of media-tion by higher adult BMI, which explainedless than half of the increased diabetes riskassociated with early menarche. Thesendings suggest that early puberty hasan effect on metabolic disease risk, whichis partly mediated by increased BMI, butalso has some direct effect through otherbiological pathways that act independentlyof adiposity.

Overall, early menarche conferred a42% increase in the risk of developmentof diabetes independently of adult BMI. Incontrast to our ndings, results from theNursesHealth Study (NHS) showed thataccounting for self-reported BMI col-lected during follow-up in adulthoodcompletely attenuated the association be-

tween early menarche and diabetes risk(18). However, among younger womenin NHS-II, the increased risk of type 2 di-abetes persisted after adjustment for BMI(18). In the Atherosclerosis Risk in Com-munities (ARIC) study, adulthood adi-posity partially attenuated the associationbetween early menarche and prevalent di-abetes and completely attenuated theassociation with incident diabetes (6). InEPIC-Norfolk, the effect of later menarcheon diabetes risk appeared to be completelyattenuated after adjustment for adult BMI(7). Such contrasting ndings may be at-tributable to the lower power of thesestudies to detect the more modest directassociation between age at menarche andtype 2 diabetes (7), or because of error in-troduced by using self-reported ratherthan measured BMI in some of the studies(18). Consistent with our ndings, in theKORA F4 study of 1,503 German women,the association between older age at men-arche and lower diabetes risk remainedsignicant after adjustment for BMI (9).

The causality of the association be-tween menarche timing and diabetes is


http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc13-0446/-/DC1http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc13-0446/-/DC1http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc13-0446/-/DC1http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc13-0446/-/DC1http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc13-0446/-/DC1http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc13-0446/-/DC1http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc13-0446/-/DC1http://care.diabetesjournals.org/lookup/suppl/doi:10.2337/dc13-0446/-/DC1


6/9

unclear. The association between menar-che and diabetes may exist because earlymenarche may be a marker of higherprepubertal BMI, with prolonged effectsof increased adiposity being the main riskfactor for diabetes. However, girls withearlier puberty have higher weight,height, BMI, and waist circumference mea-surements in young childhood, andit is highly unlikely that early maturationis associated with adiposity or lean massalone (27,28). Our results suggest an as-sociation between menarche and diabetesrisk independent of adult BMI. However,the true nature of this relationship is dif-

cult to determine without informationabout body size during childhood, whichwas unavailable in our study. BMI at age20 years was available in a subset of In-terAct participants, which is likely to be abetter marker of prepubertal BMI thanwhen assessed in later adulthood. Consis-tent with some previous reports (8,18),the association between age at menarcheand diabetes risk was attenuated to alarger extent when accounting for lateradult BMI compared with BMI measuredin childhood, adolescence, or early adult-hood. This suggests that excess adiposityin later adulthood, a more likelyindicatorof long-term exposure to being over-weight, is more important in the media-tion of increased diabetes risk associatedwith early menarche than BMI earlier inlife. In further support of this, one studyhas shown that the association betweenmenarche and increased adult BMI is notexplained by higher prepubertal BMI(29).

Using genetic variation at LIN28B,which is associated with age at menarchebut not prepubertal BMI, we previously

provided evidence for a possible directeffect of puberty timing on adiposity inyoung adult life in women (30). Our nd-ings of a consistent association betweenearly menarche and diabetes in womendiagnosed at ages older than and youngerthan 60 years support thending that pu-berty timing has a long-lasting effect onmetabolic health. Findings from the NHSshowed that there was a stronger associa-tion between early menarche and diabe-tes in younger women (younger than45 years) compared with older women(45 years or older). Because our partici-pants were older at recruitment, we did

not have sufcient cases to explore theassociation between menarche timingand diabetes diagnosed at this particularlyyoung age.

The strengths of this study lie in thelarge scale and the prospective nature ofInterAct, together with the heterogeneityin age at menarche across eight Europeancountries, the accurate assessment ofBMI, and comprehensive considerationof a range of confounding factors. Thehigh number of veried incident diabetescases incorporated using the case-cohortdesign make this study a powerful re-source to address questions relating todiabetes etiology. Our prospective analy-sis of incident cases of type 2 diabetesavoids the problem of reverse causality,whereby insulin resistance during earlylife may increase the tempo of develop-ment (31).

Limitations of this study include thefact that age at menarche was recalled inadulthood and was only recorded to thenearest year. However, age at menarcheassessed by recall during adulthood hasshown high correlations with prospectively

assessed childhood data (1,32). Further-more, any misclassication would be ex-pected to be nondifferential with respectto type 2 diabetes status and thereforecause an underestimation of the true asso-ciation. It is likely that BMI may not ade-quately account for adiposity, because itdoes not differentiate between fat andfat-free mass. However, accounting forwaist circumference, which is a better in-dicator of central adiposity, made little dif-ference in our ndings. Furthermore,results from the Framingham Heart Studyhave shown that BMI appears to be a goodmarker of differences in body composition

(33). Body weight at age 20 years was as-sessed by recall and is likely to introduceerror rather than bias, therefore leading toan underestimation of the true association.However, in the NHS recalled body weightat age 18 years was highly correlated withmeasured weight (Pearson coefcientr=0.96) (34), so the degree of error may besmall. Finally, although we considered awide range of potential confounders, it ispossible that imprecision in measurementcould lead to residual confounding, andthere remains the possibility of confound-ing by other unconsidered factors.

Mechanisms explaining the direct re-lationship between pubertal timing andlater metabolic risk arenot well understood.

We and others identied LIN28B as the rstgenetic locus to be robustly associatedwith age at menarche (35). Mice that over-expressLin28, a homolog of this gene, ex-hibit both later pubertal maturation andincreased glucose uptake (36). In addi-tion, they are resistant to obesity andtype 2 diabetes when placed on a high-fat diet (36). This shows a direct link be-tween genes controlling puberty timing

Table 2dMean age at menarche by country and its association with adult BMI in the InterAct subcohort

Subcohort characteristics Association between age at menarche and adult BMI

NMean age at

recruitment SD

Mean age

at menarche SD

Adjusted age

at menarche*

Beta (kg/m2 per +1 year

of age at menarche) 95% CI P

Sweden 1,372 56.2 8.2 13.6 1.4 13.5 20.31 (20.47 to 20.14) ,0.001

Denmark 946 56.5 4.3 13.6 1.6 13.5 20.37 (20.54 to 20.20) ,0.001

Germany 1,194 48.9 8.9 13.2 1.6 13.1 20.42 (20.58 to 20.26) ,0.001UK 747 55.8 11.2 12.9 1.6 13.3 20.34 (20.52 to 20.16) ,0.001

Netherlands 1,199 54.0 9.9 13.4 1.6 13.3 20.34 (20.47 to 20.20) ,0.001

France 585 56.3 6.5 12.9 1.4 13.2 20.30 (20.51 to 20.09) 0.006

Italy 1,339 50.4 8.0 12.6 1.5 12.9 20.25 (20.41 to 20.09) 0.003

Spain 2,218 48.2 8.2 12.9 1.6 12.7 20.27 (20.39 to 20.15) ,0.001

Total subcohort 9,590 52.4 9.1 13.1 1.6 13.1 20.32 (20.38 to 20.27) ,0.001

*Mean age at menarcheadjusted forage at recruitment.Resultsare fromlinearregressionwith adjustment forage at recruitment and center.Froma meta-analysis ofcountry-specic results.


Elks and Associates


7/9


8/9

and glucose homeostasis and provides apossible mechanistic link between pu-berty and diabetes risk. Metabolic diseasemay be linked to differential exposure tosex hormones; early menarche is associatedwith higher level of sex hormones (37) anddecreased levels of sex hormonebindingglobulin (38), which may have a role inthe pathogenesis of type 2 diabetes (39).

In conclusion, women with youngerthan average age at menarche are morelikely to develop type 2 diabetes, and thisassociation is only partially explained byhigher adult BMI. In contrast, womenwith older than average age at menarchehave no reduction in diabetes risk. Al-though avoidance of adult overweightand obesity may attenuate the risk oftype 2 diabetes in women with earlymenarche, ourndings suggest that strat-egies to prevent early menarche may be

important in their own right.

AcknowledgmentsdThe InterAct study re-ceived funding from the European Union(Integrated Project LSHM-CT-2006-037197in the Framework Programme 6 of the Euro-pean Community). Verication of diabetescases was additionally funded by NL Agencygrant IGE05012 and an incentive grant fromthe Board of the UMC Utrecht (to Y.T.v.d.S.).

P.W.F. acknowledges the support of theSwedish Research Council, Novo Nordisk,Swedish Diabetes Association, and SwedishHeart-Lung Foundation. J.H. acknowledges the

support of the Danish Cancer Society. R.K.acknowledges the support of the GermanCancer Aid, German Ministry of Research(BMBF). T.J.K. acknowledges the support ofCancer Research UK. K.T.K. acknowledges thesupport of Medical Research Council UK,Cancer Research UK.P.M.N.acknowledges thesupport of the Swedish Research Council. K.O.

acknowledges the support of the Danish Can-cer Society. J.R.Q. acknowledges thesupport ofthe Asturias Regional Government. O.R. ac-knowledges the support of The VsterbotenCounty Council. A.M.W.S. acknowledges thesupportof the Dutch Ministry of Public Health,Welfare, and Sports (VWS);NetherlandsCancerRegistry (NKR); LK Research Funds; DutchPrevention Funds; Dutch ZON (Zorg Onder-zoek Nederland); World Cancer Research Fund(WCRF); and Statistics Netherlands. A.T. ac-knowledges the support of the Danish CancerSociety. M.-J.T. acknowledges the support ofthe Health Research Fund (FIS) of the SpanishMinistry of Health, the CIBER en Epidemiologa

y Salud Pblica (CIBERESP) Spain, and MurciaRegional Government (6236). R.T. acknowl-edges the support of AIRE-ONLUS Ragusa,AVIS-Ragusa, and Sicilian Regional Govern-ment. D.L.v.d.A. acknowledges the support ofVWS, NKR, LK Resea rch Funds, Dutch Pre-vention Funds, Dutch ZON (Zorg OnderzoekNederland), WCRF, and Statistics Nether-lands. E.R. was supported in this work by theImperial College Biomedical Research Centre.No other potential conicts of interest relevantto this article were reported.

The funders had no role in study design,data collection and analysis, decision to pub-lish, or preparation of the manuscript.

C.E.E.analyzed thedata andwrote the rstdraft of the manuscript. K.K.O. wrote therstdraft of the manuscript. R.A.S. and S.J.S.analyzed the data. J.S.B., N.G.F., C.L., Y.T.v.d.S.,P.A.W., and N.J.W. contributed to discussionand revised the manuscript. P.A., B.B., A.B., H.B.,A.F.-N., P.W.F., S.G., J.H., R.K., T.J.K., K.T.K.,A.M., P.M.N., K.O., D.P., J.R.Q., S.R., O.R.,I.R., C.S., M.-J.S., A.M.W.S., A.T., M.-J.T.,

R.T., D.L.v.d.A., N.G.F., S.J.S., C.L., E.R., andN.J.W. contributed to either EPIC or InterActdata collection, study management, or co-ordination. All authors reviewed the manu-script. C.E.E. is the guarantor of this workand, as such, had full access to all the data inthe study and takes responsibility for the in-tegrity of the data and the accuracy of the dataanalysis.

The authors thank participants and stafffrom all EPIC centers for their contribution tothe study and Nicola Kerrison (MRC Epide-miology Unit, Cambridge) for her assistancewith managing the EPIC-InterAct data.

References1. Must A, Phillips SM, Naumova EN, et al.

Recall of early menstrual history and men-archeal body size: after 30 years, how welldo women remember? Am J Epidemiol2002;155:672679

2. Walvoord EC. The timing of puberty: isit changing? Does it matter? J AdolescHealth 2010;47:433439

3. Colagiuri S, Borch-Johnsen K, Glumer C,Vistisen D. There really is an epidemicof type 2 diabetes. Diabetologia 2005;48:14591463

4. Ong KK, Ahmed ML, Dunger DB. Les-sons from large population studies ontiming and tempo of puberty (secular trendsand relation to body size): the Europeantrend. Mol Cell Endocrinol 2006;254-255:812

5. Ibaez L, Lopez-Bermejo A, Diaz M,MarcosMV, de Zegher F. Early metformintherapy to delay menarche and augmentheight in girls with precocious pubarche.Fertil Steril 2011;95:727730

6. Dreyfus JG,Lutsey PL, Huxley R, et al. Ageat menarche and risk of type 2 diabetesamong African-American and white womenin the Atherosclerosis Risk in Communi-

ties (ARIC) study. Diabetologia 2012;55:237123807. Lakshman R, Forouhi N, Luben R, et al.

Association between age at menarche andrisk of diabetes in adults: results from theEPIC-Norfolk cohort study. Diabetologia2008;51:781786

8. Pierce MB, Kuh D, Hardy R. The role ofBMI across the life course in the relation-ship between age at menarche and dia-betes, in a British Birth Cohort. DiabetMed 2012;29:600603

9. Stckl D, Dring A, Peters A, et al. Age atmenarche is associated with prediabetesand diabetes in women (aged 32-81

Figure 1dHRs for type 2 diabetes by category of age at menarche in the EPIC-InterAct study.Results presented are HRs for diabetes in each group compared with the reference category

(13 years, representing the median age at menarche). Results are from Prentice-weighted Coxregression models performed within each country participating in InterAct and pooled using therandom-effects meta-analysis. Models are adjusted for age at recruitment, date of birth, center,and lifestyle and reproductive factors (C), and with additional adjustment for adult BMI ().


Elks and Associates


9/9

years) from the general population: theKORA F4 Study. Diabetologia 2012;55:681688

10. Cooper GS, Ephross SA, Sandler DP. Men-strual patterns and risk of adult-onset di-abetes mellitus. J Clin Epidemiol 2000;53:11701173

11. Saquib N, Kritz-Silverstein D, Barrett-Connor E. Age at menarche, abnormal

glucose tolerance and type 2 diabetesmellitus: The Rancho Bernardo Study. Cli-macteric 2005;8:7682

12. Heys M, Schooling CM, Jiang C, et al. Ageof menarche and the metabolic syn-drome in China. Epidemiology 2007;18:740746

13. Gambineri A, Pelusi C, Manicardi E, et al.Glucose intolerance in a large cohort ofmediterranean women with polycysticovary syndrome: phenotype and associ-ated factors. Diabetes 2004;53:23532358

14. Kivimki M, Lawlor DA, Smith GD, et al.Association of age at menarche with car-diovascular risk factors, vascular structure,

and function in adulthood: the Cardiovas-cular Risk in Young Finns study. Am J ClinNutr 2008;87:18761882

15. Chen L, Zhang C, Yeung E, et al. Age atmenarche and metabolic markers for type2 diabetes in premenopausal women: theBioCycle Study. J Clin Endocrinol Metab2011;96:E1007E1012

16. Adair LS, Gordon-Larsen P. Maturationaltiming and overweight prevalence in USadolescent girls.Am J Public Health 2001;91:642644

17. Narayan KM, Boyle JP, Thompson TJ,Gregg EW, Williamson DF. Effect of BMIon lifetime risk for diabetes in the U.S.Diabetes Care 2007;30:15621566

18. He C, Zhang C, Hunter DJ, et al. Age atmenarche and risk of type 2 diabetes: re-sults from 2 large prospective cohort stud-ies. Am J Epidemiol 2010;171:334344

19. Lakshman R, Forouhi NG, Sharp SJ, et al.Early age at menarche associated with car-diovascular disease and mortality. J ClinEndocrinol Metab 2009;94:49534960

20. Bingham S, Riboli E. Diet andcancerdtheEuropean Prospective Investigation intoCancer and Nutrition. Nat Rev Cancer2004;4:206215

21. Langenberg C, Sharp S, Forouhi NG,et al.; InterAct Consortium. Design andcohort description of the InterAct Project:an examination of the interaction of ge-netic and lifestyle factors on the incidence

of type 2 diabetes in the EPIC Study. Di-abetologia 2011;54:22722282

22. InterAct Consortium. Validity of a shortquestionnaire to assess physical activity in10 European countries. Eur J Epidemiol2012;27:1525

23. World Health Organization.Physical Sta-tus: the Use and Interpretation of Anthro-pometry. Geneva, WHO, 1995

24. Prentice RL. A case-cohort design for ep-idemiological cohort studies and diseaseprevention trials. Biometrika1986;73:111

25. Vanderweele TJ, Vansteelandt S. Oddsratios for mediation analysis for a dichot-omous outcome. Am J Epidemiol 2010;

172:1339134826. The InterAct Consortum. The link be-

tween family history and risk of type 2diabetes is not explained by anthropo-metric, lifestyle or genetic risk factors: theEPIC-InterAct study. Diabetologia 2013;56:6069

27. Kaplowitz PB, Slora EJ, Wasserman RC,Pedlow SE, Herman-Giddens ME. Earlieronset of puberty in girls: relation to in-creased body mass index and race. Pedi-atrics 2001;108:347353

28. Thankamony A, Ong KK, Ahmed ML,Ness AR, Holly JM, Dunger DB. Higherlevels of IGF-I and adrenal androgens atage 8 years are associated with earlier ageat menarche in girls. J Clin EndocrinolMetab 2012;97:E786E790

29. Pierce MB, Leon DA.Age at menarche andadult BMI in the Aberdeen children of the1950s cohortstudy. Am J Clin Nutr 2005;82:733739

30. Ong KK, Elks CE, Wills AK, et al. Associa-tions between the pubertal timing-related

variant in LIN28B and BMI vary across thelife course. J Clin Endocrinol Metab 2011;96:E125E129

31. Ekelund U, Ong KK, Linn Y, et al. As-sociation of weight gain in infancy andearly childhood with metabolic risk inyoung adults. J Clin Endocrinol Metab2007;92:98103

32. Casey VA, Dwyer JT, Coleman KA,

Krall EA, Gardner J, Valadian I. Accu-racy of recall by middle-aged partic-ipants in a longitudinal study of theirbody size and indices of maturation ear-lier in life. Ann Hum Biol 1991;18:155166

33. Trikudanathan S, Pedley A, Massaro JM,et al. Association of female reproductivefactors with body composition: the Fra-mingham Heart Study. J Clin EndocrinolMetab 2013;98:236244

34. Troy LM, Hunter DJ, Manson JE, ColditzGA, Stampfer MJ, Willett WC. The val-idity of recalled weight among youngerwomen. Int J Obes Relat Metab Disord

1995;19:57057235. Elks CE, Ong KK. Whole genome associ-

ated studies for age at menarche. BriefFunct Genomics 2011;10:9197

36. Zhu H, Shyh-Chang N, Segr AV, et al.;DIAGRAM Consortium; MAGIC Inves-tigators. The Lin28/let-7 axis regulatesglucose metabolism. Cell 2011;147:8194

37. Apter D, Reinil M, Vihko R. Some en-docrine characteristics of early menarche, arisk factor for breast cancer, are preservedinto adulthood. Int J Cancer 1989;44:783787

38. Vikan T, Schirmer H, Njlstad I, Svartberg J.Low testosterone and sex hormone-bindingglobulin levels and high estradiol levelsare independent predictors of type 2 di-abetes in men. Eur J Endocrinol 2010;162:747754

39. Ding EL, Song Y, Manson JE, et al. Sexhormone-binding globulin and risk oftype 2 diabetes in women and men. NEngl J Med 2009;361:11521163



age at menarche and type 2 diabetes risk 2013

Documents