Association Between Physical Activity and BMD in Young Men IsModulated by Catechol-O-Methyltransferase (COMT) Genotype: The

GOOD Study

Mattias Lorentzon,1 Anna L Eriksson,1 Staffan Nilsson,2 Dan Mellström,1 and Claes Ohlsson1

ABSTRACT: In this large population-based study in young men, we show that the COMT val158met poly-morphism modulates the association between physical activity, aBMD (DXA), and trabecular vBMD (pQCT).

Introduction: Peak BMD is an important predictor of future risk of osteoporosis and is largely determined bygenetic factors but also by environmental factors, among which physical activity (PA) is a strong contributor.Estrogens are believed to influence the mechanical strain signal generated by bones subjected to mechanicalloading. Catechol-O-methyltransferase (COMT) is involved in the degradation of estrogens. A functionalpolymorphism in the COMT gene (val158met), results in a 60–75% difference in enzyme activity between theval (high activity � H) and met (low activity � L) variants. The aim of this study was to determine if theCOMT val158met polymorphism modulates the association between PA and BMD in young men.Materials and Methods: The Gothenburg Osteoporosis and Obesity Determinants (GOOD) study consists of1068 men (age, 18.9 ± 0.6 yr). Areal BMD (aBMD) was measured by DXA, whereas cortical and trabecularvolumetric BMD (vBMD) were measured by pQCT. Study subjects were genotyped and classified asCOMTLL, COMTHL, or COMTHH. The amount (h/wk) of PA was determined through questionnaires.Results: Using a linear regression model (including age, height, weight, smoking, and calcium intake ascovariates), significant interactions between the COMT genotype and PA were seen for aBMD at all sites andfor trabecular vBMD in both the radius and the tibia. The difference in adjusted aBMD and trabecular vBMDbetween high (�4 h/wk) and low PA (<4 h/wk) was greater in COMTLL subjects than in subjects homozygousfor the COMTHH (total body aBMD: COMTLL 4.2% versus COMTHH 1.5%, p � 0.02; lumbar spine aBMD:COMTLL 7.8% versus COMTHH 3.9%, p � 0.04; tibia trabecular vBMD: COMTLL 7.1% versus COMTHH

1.0%, p < 0.01). The COMT polymorphism was associated with aBMD, at all sites and with trabecular vBMDin the low-PA subjects, but not in their high-PA counterparts.Conclusions: We show that the COMT val158met polymorphism modulates the association between PA,aBMD, and trabecular vBMD, suggesting that this polymorphism is of importance for BMD in subjects witha low level of PA.J Bone Miner Res 2007;22:1165–1172. Published online on April 23, 2007; doi: 10.1359/JBMR.070416

Key words: physical activity, BMD, trabecular, men, polymorphism, estrogen

INTRODUCTION

A PLETHORA OF EVIDENCE indicates that the highest at-tained bone mass in life (i.e., peak bone mass) has a

strong impact on bone mass at old age and affects the riskof developing osteoporosis.(1,2) Low BMD has been shownto be a strong independent predictor of fractures, both inmen and in women.(3,4) Twin and family studies have shownthat as much as 60–80% of the variation in the age-specificBMD can be attributed to genetic factors.(5–7) Several en-vironmental factors also contribute considerably to thelevel of attained peak bone mass. Among these factors,

physical activity (PA) and calcium intake are believed tohave a stimulatory role on bone mass accretion, whereassmoking has been suggested to reduce bone acquisitionearly in life.(8–10)

A large number of studies have investigated genetic poly-morphisms in several candidate genes for osteoporosis andtheir association with bone mass,(11) but only a few reportsare available on how genetic polymorphisms in these can-didate genes interact with environmental factors. In a small(n � 140) controlled and randomized study of middle-agedFinnish men, the PvuII polymorphism of estrogen receptor� (ER�) was associated with a BMD increase in subjectsenrolled in a 4-yr aerobic exercise program, whereas noassociation was seen between this polymorphism and BMDThe authors state that they have no conflicts of interest.

1Center for Bone Research at the Sahlgrenska Academy (CBS), Department of Internal Medicine, Gothenburg University, Gothen-burg, Sweden; 2Mathematical Statistics, Chalmers University of Technology, Gothenburg, Sweden.

JOURNAL OF BONE AND MINERAL RESEARCHVolume 22, Number 8, 2007Published online on April 23, 2007; doi: 10.1359/JBMR.070416© 2007 American Society for Bone and Mineral Research

1165

JO611756 1165 1172 August

in the reference group.(12) Suuriniemi et al.(13) showed thatthe association between exercise and pubertal BMD in girlsis modulated by ER� genotype. Increasing evidence impli-cates estrogen status as an important determinant of malebone mass.(14–18) Several reports indicate that bone cellsrespond to mechanical strain and estrogen through thesame signaling pathways.(19–22)

We have previously shown that both PA and a polymor-phism in the estrogen degrading enzyme catechol-O-methyltransferase (COMT) are associated with areal BMD(aBMD) and trabecular volumetric BMD (vBMD) inyoung men.(23,24) The investigated COMT polymorphismconstitutes a functional G to A polymorphism, which re-sults in a valine to methionine substitution at codon 158.This amino acid replacement leads to a 60–75% differencein enzyme activity between the Val (high activity) and Met(low activity) variants.(25) The alleles are therefore termedCOMTL (L � low activity) and COMTH (H � high activ-ity), respectively.

We hypothesized that the COMT polymorphism couldalter estrogen status, in the bone microenvironment and/orin the circulation, and therefore modulate the associationbetween PA and bone parameters. The aim of this studywas to determine if the COMT polymorphism modulatesthe association between PA and BMD in young adult menin the Gothenburg Osteoporosis and Obesity Determinants(GOOD) study.

MATERIALS AND METHODS

Subjects

The GOOD study was initiated with the aim to deter-mine both environmental and genetic factors involved inthe regulation of bone and fat mass. Study subjects wererandomly identified using national population registers,contacted by telephone, and asked to participate in thisstudy. A total of 1068 men (age, 18.9 ± 0.6 yr) from thegreater Gothenburg area were included. To be included inthe GOOD study, subjects had to be >18 and <20 yr of ageand willing to participate in the study. There were no otherexclusion criteria; 48.6% of the contacted study subject can-didates agreed to participate and were included in thisstudy. A standardized questionnaire was used to collect in-formation about amount of present PA (h/wk), nutritionalintake (dairy products), and smoking. Calcium intake wasestimated from dairy product intake and semiquantitatedinto quintiles.

The GOOD study was approved by the ethics committeeat Gothenburg University. Written and oral informed con-sent was obtained from all study participants. We have pre-viously presented baseline data on aBMD and bone param-eters measured with pQCT in this cohort.(26)

Anthropometrical measurements

Height and weight were measured using standardizedequipment. The CV values were <1% for these measure-ments.

DXAaBMD (g/cm2) of the whole body, total femur, trochan-

ter, femoral neck (of the left leg), and the lumbar spine

(L2–L4) were assessed using the Lunar Prodigy DXA (GELunar Corp., Madison, WI, USA). The CVs for the aBMDmeasurements ranged from 0.5% to 3%, depending on ap-plication.

pQCT

A pQCT device (XCT-2000; Stratec Medizintechnik,Pforzheim, Germany) was used to scan the distal leg (tibia)and the distal arm (radius) of the nondominant leg and arm,respectively. A 2-mm-thick single tomographic slice wasscanned with a voxel size of 0.50 mm. The cortical vBMD(not including the bone marrow; mg/cm3) and corticalcross-sectional area (CSA, mm2) were measured using ascan through the diaphysis (at 25% of the bone length in theproximal direction of the distal end of the bone) of theradius and tibia. Trabecular vBMD (mg/cm3) was measuredusing a scan through the metaphysis (at 4% of the bonelength in the proximal direction of the distal end of thebone) of these bones. Tibia length was measured from themedial malleolus to the medial condyle of the tibia, andlength of the forearm was defined as the distance from theolecranon to the ulna styloid process. The CVs were <1%for all pQCT measurements.

DNA isolation and genotyping

Genomic DNA was isolated from EDTA stabilizedwhole blood using the DNA Purigene Kit (Gentra Sys-tems). The val158met polymorphism (rs 4680) of theCOMT gene was successfully genotyped in 1050 subjects, aspreviously described.(23) Using this assay, we failed to geno-type 18 of the 1068 subjects. The genotypes were classifiedas COMTLL (AA:Met), COMTHL (AG:Met/Val), orCOMTHH (GG:Val).

Serum analyses of sex hormone binding globulinand sex steroids

Total estradiol was measured using an ultrasensitive ra-dioimmunoassay (RIA; Orion Diagnostics, Esboo, Finland;intra-assay CV 3%, interassay CV 6%). Total testosteronewas measured using RIA (Orion Diagnostics; intra-assayCV 6%, interassay CV 6%). Sex hormone binding globulin(SHBG) was measured using IRMA (Orion Diagnostics;intra-assay CV 3%, interassay CV 7%).

Free testosterone and free estradiol were calculated ac-cording to the method described by Vermeulen et al.(27)

and Van den Beld et al.,(28) taking the concentrations oftotal testosterone, total estradiol, and SHBG into accountand assuming a fixed albumin concentration of 43 g/liter.

Statistical analysis

Values are given as mean ± SD, unless otherwise indi-cated. All calculations were performed with the SPSS Sta-tistical Software (version 14.0; SPSS, Chicago, IL, USA).Hardy-Weinberg equilibrium for the COMT polymorphismwas calculated using a �2 test.

�2 tests were used to determine whether the distributionof smokers differed between the COMT genotype groups.

Continuous variables were compared between individu-

LORENTZON ET AL.1166

als with different COMT genotypes using ANOVA, fol-lowed by Bonferronıs correction for multiple comparisons.

The independent predictors of the various bone param-eters were tested using multiple linear regression analysis,including the investigated genetic polymorphism, height,weight, age, calcium intake (quintiles), smoking (yes/no),and PA (h/wk).

The general linear model (GLM) was used to performlinear regression analysis and to calculate interactions be-tween the COMT polymorphism and PA. The percentageof the variation of each bone parameter (r2) explained bythe COMT gene polymorphisms was calculated, using lin-ear regression models. To illustrate the magnitude of thediscrepancies in bone parameters between the high- andlow-PA groups in subjects homozygous for COMTHH ver-sus the corresponding discrepancies in subjects homozy-gous for COMTLL, we performed two-way analysis of co-variance (ANCOVA), including an interaction termbetween PA groups and homozygous genotypes, with thesame covariates as above.

A p value <0.05 was considered significant.

RESULTS

Anthropometrics, genotype distribution, and sexhormone levels

Anthropometrics, sex hormone levels, and bone vari-ables of the whole cohort are presented in Table 1. Threehundred fourteen subjects (29.9%) were homozygous forthe low activity COMT allele (COMTLL), 527 subjects

(50.2%) were heterozygous (COMTHL), and 209 (19.9%)subjects were homozygous for the high activity COMT al-lele (COMTHH). The genotype distribution was in Hardy-Weinberg equilibrium (p � 0.88).

There were no differences in age (LL, 18.9 ± 0.6 [SD] yr;HL, 18.9 ± 0.5 yr; HH, 18.9 ± 0.6 yr; p � 0.64), height (LL,181.8 ± 6.7 cm; HL, 181.8 ± 6.7 cm; HH, 181.5 ± 6.8 cm;p � 0.25), weight (LL, 75.0 ± 13.5 kg; HL, 73.2 ± 10.7 kg;HH, 74.0 ± 10.0 kg; p � 0.09), frequency of smokers (LL,8.9%; HL, 8.2%; HH, 10.5%; p � 0.59), daily calcium in-take (LL, 1124 ± 728 mg/d; HL, 1104 ± 669 mg/d; HH, 1031± 849 mg/d; p � 0.34), or weekly PA (LL, 4.1 ± 5.0 h/wk;HL, 4.5 ± 5.3 h/wk; HH, 4.2 ± 5.6 h/wk; p � 0.49) betweenthe genotype groups. Furthermore, the COMT polymor-phism was not associated with total or free levels of eitherestradiol (free estradiol [pM]: LL, 1.39 ± 0.65; HL, 1.38 ±0.66; HH, 1.51 ± 0.74; p � 0.30; data not shown) or testos-terone (free testosterone [nM]: LL, 0.46 ± 0.17; HL, 0.46 ±0.17; HH, 0.48 ± 0.16; p � 0.10; data not shown).

PA and COMT polymorphism as independentpredictors of aBMD and trabecular vBMD

The independent predictive role of PA and COMT geno-type for aBMD (DXA), as well as cortical and trabecularbone parameters (pQCT), were studied using linear regres-sion, including age, weight, height, smoking, calcium intake,present amount of PA, and COMT polymorphism (Table2). Both amount of PA and the COMT val158met polymor-phism were independent predictors of aBMD of the totalfemur, trochanter, and femoral neck. Amount of PA alsoindependently predicted aBMD of the total body and lum-bar spine, whereas the COMT val158met polymorphismdid not predict aBMD of the lumbar spine but showed astrong tendency toward being associated with aBMD of thetotal body (p � 0.05; Table 2). pQCT analyses showed thatamount of PA and COMT val158met polymorphism wereboth independent predictors of trabecular vBMD, but onlyamount of present PA predicted cortical bone size (Table2). Amount of present PA was a negative independent pre-dictor of cortical vBMD in the radius but not in the tibia.The COMT val158met polymorphism was not associatedwith cortical vBMD (Table 2).

COMT genotype modulates the association betweenPA and BMD

Using a linear regression model (including age, height,weight, smoking, and calcium intake as covariates), signifi-cant interactions between COMT genotype and PA wereseen for aBMD at all sites and for trabecular vBMD in boththe radius and the tibia (Table 2), whereas no interactionswere seen for parameters of cortical bone size or corticalvBMD (Table 2).

We have previously shown that the lowest amount of PAassociated with bone parameters in this cohort was 4h/wk.(24) Based on this knowledge, subjects were dividedinto a low (<4 h/wk, n � 554) and high (�4 h/wk, n � 514)PA group. To further characterize the interaction betweenPA and the COMT polymorphism for bone parameters, the

TABLE 1. ANTHROPOMETRICS, SEX HORMONE LEVELS, AND

BONE VARIABLES

Variables Mean ± SD

Age (yr) 18.9 ± 0.6Height (cm) 181.4 ± 6.8Weight (kg) 73.8 ± 11.9PA (h/wk) 4.4 ± 5.3Smoking (%) 8.7Calcium intake (mg/d) 1095 ± 724Estradiol (pM) 70.6 ± 29.2Free estradiol (pM) 1.50 ± 0.66Testosterone (nM) 17.0 ± 5.8Free testosterone (nM) 0.46 ± 0.17SHBG (nM) 20.4 ± 7.4DXA

Total body aBMD (g/cm2) 1.25 ± 0.10Lumbar spine aBMD (g/cm2) 1.24 ± 0.15Total femur aBMD (g/cm2) 1.17 ± 0.16Femur neck aBMD (g/cm2) 1.17 ± 0.16Trochanter aBMD (g/cm2) 0.97 ± 0.15

pQCTTibia cortical CSA (mm2) 270 ± 34Tibia cortical vBMD (mg/cm3) 1156 ± 20Tibia trabecular vBMD (mg/cm3) 266 ± 34Radius cortical CSA (mm2) 96 ± 12Radius cortical vBMD (mg/cm3) 1165 ± 23Radius trabecular vBMD (mg/cm3) 219 ± 41

Values are given as mean ± SD, n � 1068.

COMT POLYMORPHISM, PHYSICAL ACTIVITY, AND BMD 1167

high- and low-PA groups were further divided according toCOMT genotype into a total of six subgroups (Fig. 1). An-thropometrics and sex hormone levels for each respective

group are presented in Table 3. Because there were dis-crepancies between the different PA COMT genotypegroups in anthropometric characteristics (Table 3), aBMD

FIG. 1. The association between PAamount, aBMD (A–D), and trabecularvBMD (E and F) is dependent on COMTgenotype. +PA, physical activity � 4h/wk;−PA, physical activity < 4 h/wk; HH,COMTHH; HL, COMTHL; LL, COMTLL.BMD values are adjusted for age, height,weight, calcium intake, and smoking.ANOVA was used to calculate p values fortrend. Mean ± SE are presented.

TABLE 2. AMOUNT OF PRESENT PA AND COMT GENOTYPE AS INDEPENDENT PREDICTORS OF aBMD AND TRABECULAR vBMD

Linear regression not including the interaction termLinear regression including the interaction term

(interaction of PA and COMT: p)PA � PA p COMT � COMT p

DXAaBMD

Total body 0.20 <0.001 0.05 0.05 <0.001Femur neck 0.26 <0.001 0.06 0.02 0.02Trochanter 0.26 <0.001 0.06 0.02 0.01Femur total 0.26 <0.001 0.08 0.007 0.005Lumbar Spine 0.23 <0.001 0.04 0.19 0.003

pQCTTrabecular vBMD tibia 0.18 <0.001 0.06 0.03 0.02Trabecular vBMD radius 0.15 <0.001 0.06 0.04 0.002Cortical vBMD radius −0.07 0.01 0.00 0.93 0.64Cortical vBMD tibia −0.03 0.41 0.04 0.17 0.99Cortical CSA tibia 0.22 <0.001 0.03 0.30 0.08Cortical CSA radius 0.15 <0.001 0.03 0.31 0.13

Linear regression, including amount of present PA (h/wk), COMT genotype (coded as HH > HL > LL), age, height, weight, calcium intake, and smoking.Standardized � and p values are presented. n � 1050.

LORENTZON ET AL.1168

and trabecular vBMD values were adjusted for smokingstatus, calcium intake, age, height, and weight. After adjust-ment for these covariates, aBMD of the lumbar spine andtotal body, as well as trabecular vBMD of both the radiusand tibia, were not higher in the high PA than in the low-PA COMTHH subjects (Figs. 1A, 1B, 1E, and 1F). In con-trast, aBMD at all sites, as well as trabecular vBMD of boththe radius and tibia, were higher in the high-PA groups thanin the low-PA groups of both COMTHL and COMTHH

subjects (Figs. 1A–1F). The difference in aBMD and tra-becular vBMD between high and low PA was in generalgreater in COMTLL subjects than in subjects homozygousfor the COMTHH (total body aBMD: COMTLL 4.2% ver-sus COMTHH 1.5%, p � 0.02; lumbar spine aBMD:COMTLL 7.8% versus COMTHH 3.9%, p � 0.04; total fe-mur aBMD: COMTLL 9.2% versus COMTHH 5.3%, p �0.07; radius trabecular vBMD: COMTLL 9.9% versusCOMTHH 2.3%, p � 0.02; tibia trabecular vBMD:COMTLL 7.1% versus COMTHH 1.0%, p < 0.01).

The COMT polymorphism was a clear predictor(COMTHH > COMTHL > COMTLL) of aBMD of the totalbody, lumbar spine, total femur, and trochanter (Figs. 1A–1D), as well as for trabecular vBMD (Figs. 1E–1F) in thesubjects with low but not in the subjects with a high amountof PA.

To determine the magnitude of the variation of the vari-ous bone parameters explained in the low-PA subjects (<4h/wk), by the COMT polymorphism, we used a multiplelinear regression model, including age, height, weight,smoking, calcium intake, amount of PA as covariates. Ad-dition of COMT genotypes to this regression model in-creased the percentage (aBMD total femur: 1.8%; trabecu-lar vBMD tibia: 2.2%; data not shown) of the variation inaBMD and trabecular vBMD explained (r2) by the model.In contrast, in the subjects with high PA (�4 h/wk), theinclusion of COMT genotypes in the model only marginallyincreased the percentage of the variation in bone param-eters explained (aBMD total femur: 0.2%, not significant;trabecular vBMD tibia: 0.1%, not significant; data notshown).

DISCUSSION

In this study, we confirmed our previous findings(23) inthat the investigated COMT val158met polymorphism wasassociated with aBMD at several sites, as well as with tra-becular vBMD, in an enlarged cohort of young men (theGOOD cohort). To the best of our knowledge, no otherreport studying the COMT val158met polymorphism in re-lation to BMD in men is available. A recent study foundthat the COMT val158met polymorphism was associatedwith bone loss of the radius but not to baseline aBMD atany site in Japanese postmenopausal women.(29) However,another report did not find any association between COMTval158met polymorphism and aBMD or bone loss in post-menopausal women.(30) In these reports on postmeno-pausal women, the possible interaction between the COMTval158met polymorphism and PA was not studied.

We recently reported that amount of PA was stronglyindependently associated with aBMD and trabecular

TA

BL

E3.

AN

TH

RO

PO

ME

TR

ICC

HA

RA

CT

ER

IST

ICS

AN

DSE

XH

OR

MO

NE

LE

VE

LS

AC

CO

RD

ING

TO

PA

LE

VE

LA

ND

CO

MT

GE

NO

TY

PE

Gro

up

12

34

56

Hig

hP

A,

HH

(n=

96)

Low

PA

,H

H(n

=11

3)p

1vs

.2

Hig

hP

A,

HL

(n=

264)

Low

PA

,H

L(n

=26

3)p

3vs

.4

Hig

hP

A,

LL

(n=

144)

Low

PA

,L

L(n

=17

0)p

5vs

.6

Age

(yr)

18.9

±0.

618

.9±

0.6

NS

18.8

±0.

519

.0±

0.5

0.01

18.8

±0.

519

.0±

0.6

NS

Hei

ght

(cm

)18

0.6

±7.

018

0.9

±6.

6N

S18

1.3

±6.

618

1.7

±6.

9N

S18

1.6

±6.

918

1.9

±6.

5N

SW

eigh

t(k

g)74

.4±

11.1

73.8

±12

.9N

S74

.5±

9.9

71.8

±11

.3N

S77

.0±

13.0

73.4

±13

.8N

SC

alci

umin

take

(mg/

d)11

97±

1040

890

±61

20.

0312

13±

717

995

±60

00.

0112

22±

739

1041

±71

0N

SSm

okin

g(%

)6.

214

.2N

S4.

511

.8<

0.01

6.3

11.2

NS

SHB

G(n

M)

21.4

±7.

219

.6±

7.3

NS

21.0

±7.

220

.2±

7.5

NS

20.1

±7.

120

.3±

7.9

NS

Fre

ees

trad

iol

(pM

)1.

49±

0.7

1.53

±0.

8N

S1.

38±

0.6

1.39

±0.

7N

S1.

43±

0.6

1.37

±0.

7N

SF

ree

test

oste

rone

(nM

)0.

48±

0.2

0.48

±0.

2N

S0.

45±

0.2

0.47

±0.

2N

S0.

47±

0.1

0.45

±0.

1N

S

Mea

ns±

SDar

egi

ven.

pva

lues

are

pres

ente

dfo

rco

mpa

riso

nof

the

grou

psby

one-

way

AN

OV

A,f

ollo

wed

byB

onfe

rron

i’sco

rrec

tion

for

mul

tipl

eco

mpa

riso

ns.D

iffe

renc

esin

smok

ing

perc

enta

gebe

twee

nth

egr

oups

wer

ete

sted

usin

g�

2te

sts.

n�

1050

.L

owP

A,<

4h/

wk;

high

PA

,�4

h/w

k;C

OM

TH

H,h

omoz

ygou

sfo

rth

ehi

ghac

tivi

tyva

rian

t;C

OM

TH

L,h

eter

ozyg

ous;

CO

MT

LL

,hom

ozyg

ous

for

the

low

acti

vity

CO

MT

vari

ant;

SHB

G,s

exho

rmon

ebi

ndin

ggl

obul

in;

NS,

not

sign

ific

ant.

COMT POLYMORPHISM, PHYSICAL ACTIVITY, AND BMD 1169

vBMD in the GOOD cohort.(24) Given that both amount ofPA and the COMT polymorphism were associated withboth aBMD and trabecular vBMD, in addition to the factthat both PA and the estrogen system affect bone cellsthrough common pathways,(19–21,31) we hypothesized thatthe COMT val158met polymorphism, possessing the abilityto alter estrogen degradation, could interact with the asso-ciation between PA and these bone parameters. The resultsof this study showed that COMT genotype modulates theassociation between PA and bone parameters, includingaBMD at several locations and trabecular vBMD. The gen-eral pattern was that the COMT genotype predicts boneparameters in subjects with low amounts of PA but not insubjects with high amounts of PA. Our findings showed thatthe COMTHH genotype is associated with higher aBMDand trabecular vBMD than the COMTLL genotype,whereas the latter genotype displays a stronger association,than the former, with PA. These findings indicate that theCOMTLL subjects have more to gain by PA, in terms ofincreased BMD, than their COMTHH counterparts. An al-ternative explanation is that subjects with a high amount ofPA have a maximal response to mechanical loading thatcan not be further modulated by the COMT genotype,whereas subjects with a low amount of PA and submaximalresponse to mechanical loading are clearly affected by theCOMT polymorphism.

We observed several-fold higher percentage differencesbetween the high- and low-PA groups in the COMTLL sub-jects than in the COMTHH subjects (e.g., total body aBMD4.2% versus 1.5% and tibia trabecular vBMD 7.1% versus1.0%), suggesting that the COMT genotype might modu-late the BMD response to mechanical loading. Our dataindicate that early identification of COMTLL subjects andPA intervention could aid in maximizing peak BMD inyoung adult men. However, our results are based on youngmen and can therefore not be generalized and interpretedin regard to possible risk prevention in the population pri-marily subjected to osteoporosis (i.e., elderly women andmen). The studied population in this study is primarily ofwhite origin. Hence, the results from this study may not betransferable to populations of other ethnicity.

A mechanism, termed the mechanostat, has been pro-posed to sense the mechanical strain signal (which is pro-duced when bones are subjected to loading), and as a result,regulate bone formation and resorption, to adapt the skel-eton to the forces applied to it. Estrogen has been proposedto modulate the mechanostat.(32,33) Using intact and ovari-ectomized rats on earth and on the orbiting space station,Westerlind et al.(31) showed that estrogen governs the rateof bone turnover, but the greatest impact on the balancebetween bone formation and resorption is exerted by me-chanical loading. It was recently shown that mice with agene deletion of the ER� are unable to respond to PA witha periosteal bone expansion as seen in wildtype mice,(34)

supporting the notion that estrogen signaling is involved inthe skeletal response to mechanical loading induced by PA.

The roles of gene–environment interactions in the sexhormone system, and in general, have been inadequatelystudied. However, two recent studies report that the asso-

ciation between exercise and BMD in girls and in middle-aged men was modulated by ER� genotype.(12,13)

We have previously found that the functional COMTval158met polymorphism was associated with serum estra-diol levels in middle-aged men,(35) but we failed to detectsuch a relationship in this study. Lack of this association wasalso reported from two separate cohorts of postmenopausalwomen.(29,36) It is not yet known whether the COMT geneis expressed in bone cells. 2-hydroxylation and 16�-hydroxylation constitute the two major pathways for inac-tivation of circulating estrogen.(33) COMT inactivates 2-and 4-hydroxyestradiol, suggesting that a polymorphismthat reduces the activity of COMT could result in a shifttoward a higher proportion of 2-and 4-hydroxy metabolites,leading to a shift toward the 16�-pathway.(25) Thesechanges could possibly occur in the local bone environment,without affecting the serum levels of these metabolites. 16-hydroxy-estradiol retains estrogenic activity, whereas the2-hydroxy-metabolites are devoid of estrogenic activity andmay also possess anti-estrogenic activities.(33,37) Because ofthe lack of association between the COMT val158met poly-morphism and serum sex hormone levels in this study, themechanisms by which the COMT val158met polymorphismmodulates the association between PA and BMD remainsunclear. One could, however, speculate that the strongerassociation between PA and BMD seen in the COMTLL

than in the COMTHH subjects can be explained by higherlevels of bioactive estrogens in the bone microenvironment,giving rise to a favorable mechanical strain signal in re-sponse to PA in the COMTLL subjects. The COMT enzymeinduced degradation of 2-hydroxyestradiol results in themetabolite 2-methoxyestradiol. This metabolite has beenshown to be protective against bone loss after ovariectomyin rats.(38) Hence, an alternative explanation of how theCOMT val158met polymorphism affects bone metabolismcan be that the COMTLL subjects, in this study, could havelower levels of 2-methoxyestradiol, leading to lower BMDthan the COMTHH subjects.

It is also possible that the COMT polymorphism affectspeak bone mass through mechanisms not involving sex hor-mones and their metabolites. COMT is also involved in thedegradation of catecholamines, and results from animalstudies indicate that the sympathetic nervous system has acatabolic effect on bone.(39)

In conclusion, in this large population-based study, weshowed that the COMT val158met polymorphism modu-lates the association between PA, aBMD, and trabecularvBMD in young men, suggesting that gene–environmentinteractions are of importance and can exert a substantialimpact on the attainment of peak BMD in men.

ACKNOWLEDGMENTS

This study was supported by the Swedish ResearchCouncil, the Swedish Centre for Research in Sports Medi-cine (CIF), the Swedish Foundation for Strategic Research,European Commission Grant QLK4-CT-2002-02528, theLundberg Foundation, the Torsten and Ragnar SöderbergsFoundation, Petrus and Augusta Hedlunds Foundation, the

LORENTZON ET AL.1170

ALF/LUA grant from the Sahlgrenska University Hospital,and the Novo Nordisk Foundation.

REFERENCES

1. Hui SL, Slemenda CW, Johnston CC Jr 1990 The contributionof bone loss to postmenopausal osteoporosis. Osteoporos Int1:30–34.

2. Kelly PJ, Morrison NA, Sambrook PN, Nguyen TV, EismanJA 1995 Genetic influences on bone turnover, bone densityand fracture. Eur J Endocrinol 133:265–271.

3. Johnell O, Kanis JA, Oden A, Johansson H, De Laet C, Del-mas P, Eisman JA, Fujiwara S, Kroger H, Mellstrom D, Meu-nier PJ, Melton LJ III, O’Neill T, Pols H, Reeve J, Silman A,Tenenhouse A 2005 Predictive value of BMD for hip and otherfractures. J Bone Miner Res 20:1185–1194.

4. Cummings SR, Nevitt MC, Browner WS, Stone K, Fox KM,Ensrud KE, Cauley J, Black D, Vogt TM 1995 Risk factors forhip fracture in white women. Study of Osteoporotic FracturesResearch Group. N Engl J Med 332:767–773.

5. Seeman E, Hopper JL, Bach LA, Cooper ME, Parkinson E,McKay J, Jerums G 1989 Reduced bone mass in daughters ofwomen with osteoporosis. N Engl J Med 320:554–558.

6. Pocock NA, Eisman JA, Hopper JL, Yeates MG, SambrookPN, Eberl S 1987 Genetic determinants of bone mass in adults.A twin study. J Clin Invest 80:706–710.

7. Howard GM, Nguyen TV, Harris M, Kelly PJ, Eisman JA 1998Genetic and environmental contributions to the association be-tween quantitative ultrasound and bone mineral density mea-surements: A twin study. J Bone Miner Res 13:1318–1327.

8. Rizzoli R, Bonjour JP 1999 Determinants of peak bone massand mechanisms of bone loss. Osteoporos Int 9(Suppl 2):S17–S23.

9. Heaney RP, Abrams S, Dawson-Hughes B, Looker A, MarcusR, Matkovic V, Weaver C 2000 Peak bone mass. OsteoporosInt 11:985–1009.

10. Lorentzon M, Mellstrom D, Haug E, Ohlsson C 2007 Smokingin young men is associated with lower bone mineral densityand reduced cortical thickness. J Clin Endocrinol Metab. J ClinEndocrinol Metab 92:497–503.

11. Ralston SH 2002 Genetic control of susceptibility to osteopo-rosis. J Clin Endocrinol Metab 87:2460–2466.

12. Remes T, Vaisanen SB, Mahonen A, Huuskonen J, Kroger H,Jurvelin JS, Penttila IM, Rauramaa R 2003 Aerobic exerciseand bone mineral density in middle-aged finnish men: A con-trolled randomized trial with reference to androgen receptor,aromatase, and estrogen receptor alpha gene polymorphismssmall star, filled. Bone 32:412–420.

13. Suuriniemi M, Mahonen A, Kovanen V, Alen M, LyytikainenA, Wang Q, Kroger H, Cheng S 2004 Association betweenexercise and pubertal BMD is modulated by estrogen receptoralpha genotype. J Bone Miner Res 19:1758–1765.

14. Orwoll ES 2003 Men, bone and estrogen: Unresolved issues.Osteoporos Int 14:93–98.

15. Riggs BL, Khosla S, Melton LJ III 1998 A unitary model forinvolutional osteoporosis: Estrogen deficiency causes bothtype I and type II osteoporosis in postmenopausal women andcontributes to bone loss in aging men. J Bone Miner Res13:763–773.

16. Bouillon R, Bex M, Vanderschueren D, Boonen S 2004 Estro-gens are essential for male pubertal periosteal bone expansion.J Clin Endocrinol Metab 89:6025–6029.

17. Vanderschueren D, Vandenput L, Boonen S, Lindberg MK,Bouillon R, Ohlsson C 2004 Androgens and bone. Endocr Rev25:389–425.

18. Vidal O, Lindberg MK, Hollberg K, Baylink DJ, Andersson G,Lubahn DB, Mohan S, Gustafsson JA, Ohlsson C 2000 Estro-gen receptor specificity in the regulation of skeletal growth andmaturation in male mice. Proc Natl Acad Sci USA 97:5474–5479.

19. Damien E, Price JS, Lanyon LE 1998 The estrogen receptor’s

involvement in osteoblasts’ adaptive response to mechanicalstrain. J Bone Miner Res 13:1275–1282.

20. Damien E, Price JS, Lanyon LE 2000 Mechanical strain stimu-lates osteoblast proliferation through the estrogen receptor inmales as well as females. J Bone Miner Res 15:2169–2177.

21. Lanyon LE 1996 Using functional loading to influence bonemass and architecture: Objectives, mechanisms, and relation-ship with estrogen of the mechanically adaptive process inbone. Bone 18(1 Suppl):37S–43S.

22. Jessop HL, Sjoberg M, Cheng MZ, Zaman G, Wheeler-JonesCP, Lanyon LE 2001 Mechanical strain and estrogen activateestrogen receptor alpha in bone cells. J Bone Miner Res16:1045–1055.

23. Lorentzon M, Eriksson AL, Mellstrom D, Ohlsson C 2004 TheCOMT val158met polymorphism is associated with peak BMDin men. J Bone Miner Res 19:2005–2011.

24. Lorentzon M, Mellstrom D, Ohlsson C 2005 Association ofamount of physical activity with cortical bone size and trabecu-lar volumetric BMD in young adult men: The GOOD Study. JBone Miner Res 20:1936–1943.

25. Lachman HM, Papolos DF, Saito T, Yu YM, Szumlanski CL,Weinshilboum RM 1996 Human catechol-O-methyltransferasepharmacogenetics: Description of a functional polymorphismand its potential application to neuropsychiatric disorders.Pharmacogenetics 6:243–250.

26. Lorentzon M, Swanson C, Andersson N, Mellstrom D, Ohls-son C 2005 Free testosterone is a positive, whereas free estra-diol is a negative, predictor of cortical bone size in young Swed-ish men:The GOOD Study. J Bone Miner Res 20:1334–1341.

27. Vermeulen A, Verdonck L, Kaufman JM 1999 A critical evalu-ation of simple methods for the estimation of free testosteronein serum. J Clin Endocrinol Metab 84:3666–3672.

28. van den Beld AW, de Jong FH, Grobbee DE, Pols HA, Lam-berts SW 2000 Measures of bioavailable serum testosteroneand estradiol and their relationships with muscle strength,bone density, and body composition in elderly men. J ClinEndocrinol Metab 85:3276–3282.

29. Gorai I, Inada M, Morinaga H, Uchiyama Y, Yamauchi H,Hirahara F, Chaki O 2007 CYP17 and COMT gene polymor-phisms can influence bone directly, or indirectly through theireffects on endogenous sex steroids, in postmenopausal Japa-nese women. Bone. Bone 40:28–36.

30. Tofteng CL, Abrahamsen B, Jensen JE, Petersen S, TeilmannJ, Kindmark A, Vestergaard P, Gram J, Langdahl BL,Mosekilde L 2004 Two single nucleotide polymorphisms in theCYP17 and COMT Genes–relation to bone mass and longitu-dinal bone changes in postmenopausal women with or withouthormone replacement therapy. The Danish Osteoporosis Pre-vention Study. Calcif Tissue Int 75:123–132.

31. Westerlind KC, Wronski TJ, Ritman EL, Luo ZP, An KN, BellNH, Turner RT 1997 Estrogen regulates the rate of bone turn-over but bone balance in ovariectomized rats is modulated byprevailing mechanical strain. Proc Natl Acad Sci USA94:4199–4204.

32. Frost HM 1999 On the estrogen-bone relationship and post-menopausal bone loss: A new model. J Bone Miner Res14:1473–1477.

33. Riggs BL, Khosla S, Melton LJ III 2002 Sex steroids and theconstruction and conservation of the adult skeleton. EndocrRev 23:279–302.

34. Lee K, Jessop H, Suswillo R, Zaman G, Lanyon L 2003 En-docrinology: Bone adaptation requires oestrogen receptor-alpha. Nature 424:389.

35. Eriksson AL, Skrtic S, Niklason A, Hulten LM, Wiklund O,Hedner T, Ohlsson C 2004 Association between the low activ-ity genotype of catechol-O-methyltransferase and myocardialinfarction in a hypertensive population. Eur Heart J 25:386–391.

36. Worda C, Sator MO, Schneeberger C, Jantschev T, Ferlitsch K,Huber JC 2003 Influence of the catechol-O-methyltransferase(COMT) codon 158 polymorphism on estrogen levels inwomen. Hum Reprod 18:262–266.

37. Westerlind KC, Gibson KJ, Malone P, Evans GL, Turner RT

COMT POLYMORPHISM, PHYSICAL ACTIVITY, AND BMD 1171

1998 Differential effects of estrogen metabolites on bone andreproductive tissues of ovariectomized rats. J Bone Miner Res13:1023–1031.

38. Sibonga JD, Lotinun S, Evans GL, Pribluda VS, Green SJ,Turner RT 2003 Dose-response effects of 2-methoxyestradiolon estrogen target tissues in the ovariectomized rat. Endocri-nology 144:785–792.

39. Takeda S, Elefteriou F, Levasseur R, Liu X, Zhao L, ParkerKL, Armstrong D, Ducy P, Karsenty G 2002 Leptin regulatesbone formation via the sympathetic nervous system. Cell111:305–317.

Address reprint requests to:Mattias Lorentzon, MD, PhD

Division of Endocrinology,Department of Internal Medicine

Gröna Stråket 8, Sahlgrenska University Hospital413 45 Gothenburg, Sweden

E-mail: Mattias.Lorentzon@medic.gu.se

Received in original form December 5, 2006; revised form March19, 2007; accepted April 17, 2007.

LORENTZON ET AL.1172