JOURNAL OF BONE AND MINERAL RESEARCH

Volume 1 I, Number 2, 1996

Hlackwell Science, Inc.

Exercise Effects on Bone Mass in Postmenopausal Women

Are Site-Specific and Load-Dependent

D E B O R A H KERR,' ALAN MORTON,' IAN DICK,' and R I C H A R D PRINCE'

ABSTRACT
It is considered that skeletal mass in humans may respond to loading or the number of loading cycles. The aim
of this study was to examine the effect of a 1year progressive resistance training program on the bone mass of 56
postmenopausal women. Assignment was by block randomization to one of two resistance training groups: a
strength trained group (3 X 8 repetition maximum) or an endurance group (3 x 20 repetition maximum). The
resistance exercises were selected to stress the ipsilateral forearm and hip region. The exercising side was
randomly assigned with one side exercised while the alternate side acted as the nonexercise control. Bone mineral
density (BMD) was measured every 3 months at the radial forearm and four hip sites using the Hologic QDR 2000
bone densitometer. A linear regression function was fitted for each individual's bone density results, and the slope
was compared for the exercise and control side using paired t-tests. The bone mass increase with the strength
regimen was significantly greater at the trochanteric hip site (control -0.6 f 2.2%, exercise 1.7 f 4.l%,p < 0.01),
at the intertrochanteric hip site (control -0.1 +- 2.1%, exercise 1.5 -C 3.0%,p < 0.05), Ward's triangle (control 0.8
f 5.2%,exercise 2 3 f 4.0%,p < 0.05), and at the ultradistal radial site (control - 1.4 f 2.3%, exercise 2.4 -t 4.3%.
p < 0.01). There was no significant increase in BMD with the endurance regimen except at the radius midsite
(control -1.0 f 23%, exercise 0.1 & 1.4%, p < 0.01). In both the endurance and the strength group, muscle
strength, tested by a one-repetition maximum (1RM) test, increased significantly for all 10 exercises ( p < 0.01)
and to a similar degree in the two groups. In the strength group but not the endurance group there were significant
correlations between the slope of the change in BMD and the percentage increase in strength as follows: trochanter
with leg press; intertrochanter with leg press ( p < 0.05); and Ward's triangle with hip extension and hip adduction
( p < 0.05). Thus these results support the notion of a site-specific response of bone to maximal loading from
resistance exercise in that although the trochanter and intertrochanteric bone density was elevated by the
resistance exercises undertaken, there was no effect on the femoral neck value. Postmenopausal bone mass can be
significantly increased by a strength regimen that uses high-load low repetitions but not by an endurance regimen
that uses low-load high repetitions. We conclude that the peak load is more important than the number of loading
cycles in increasing bone mass in early postmenopausal women. (J Bone Miner Res 1996;11:21f&225)

INTRODUCTION exercise enhancing bone mass. Many studies have not

shown an exercise effect due in part to a lack o f true
~ I K WI IOYI IS wtwiri. I ~ ~ A K I N Gis frequently recom-
randomization, low subject to control for
E m c n d c d for the prevention and treatment of osteopo- hormone replacement therapy, and the bone density not
rosis. There is, however, evidence for' ' ") and again st"^"'
being measured at the site of loading. In addition, the
~

choice of exercise has emphasized aerobic training rather

*This paper was in part a t the lhth ~~~~~~l ASBMR, than skeletal loading. The way in which exercise is thought
Scptcmber 1994, Kansas City Missouri, U.S.A. to act on the skeleton is through gravitational forces 01-

'Dcpartnicnt of Medicine, University o f Western Australia. Australia.

'Dcpartnicnt o f Human Movcmcnt, University of Western Australia, Australia.

218
EXERCISE EFFECTS O N BONE MASS IN WOMEN 219

niusclc pull producing strain within the skeleton which ;ire sions. Subjects were also excluded if they had taken more
perceived hy bone cells ;IS ostcogcnic. If ii load is detected than 500 mg o f calcium tablet supplementation per day for
;is being greater than the load threshold, there will be an more than I month in the previous year. Eligible subjects
iiicrc;isc i n internal strain and bone formation will occur.("') completed a n informed consent document before entering
Studies i n lifctimc tennis players that have shown a greater the study. The study was approved by the Human Rights
bone density" I ) o r increased humcral hypertrophy"" in Committee o f the University o f Western Australia. Fifty-six
the playing a r n i compared with the nonplaying arm and subjects entered the study and were randomized to 28 in
studies in hcniiplegia in which there was ii reduction in bone each exercise group.
nims i n the paralyzed iirm suggest that the action is site- Bone density was measured at the hip and radial forearm
'
specific.' " sites o f both limbs at three monthly intervals using dual-
Animal studies have shown that the load applied t o the enerby X-ray absorptionictry technology (DXA) on B O D R
skeleton procluccs 21 greater ostcogcnic response than the 2000 machine (Hologic. Waltham, MA, U S A . ) . A wholc-
numbcr of loading cycles.' " I J ) Application t o a human body scan using the pencil beam mode was measured at
model implies that iiii exercise that applies high loads but baseline. The hip site was measured using the array mode
few repetitions t o the skeleton would result in the greatest and included the area of the femoral neck, trochanter, and
increase in hone imss. Thus, 11 progressive resistance weight intcrtrochantcr site. The neck site w;is defined ;IS ;I rcctan-
tr;iining program that ensures the greatest muscular glc 6 nim wide traversing the femoral neck placed against
strength gains should result in the greatest skeletal load. the greater trochanter with ;I coefficient o f variation ( C V )
I,ittlc is known. however. about the effects of resistance o f 1.5%. The trochanter site was a triangular region whose
weight training on the bone inass o f postmenopausal wom- boundaries were defined as the lateral edge o f the femoral
c l l ( s . I ~ l Ithough there have been several studies in pre- neck area t o the point where the edge of the femur changes
mcnop;\us;ll WOlncIl.~ I " )
''I curvature helow the trochanter (CV 1.3%)).The intertro-
'1'0 cxaminc the effect o f ;I 12-month progressive resis- chanteric region was the remainder of the femur extending
tance training program on hone density in postmenopausal 10 mm helow the lesser trochanter (CV 1.3%)). The Ward's
women. two established training protocols were used which triangle site was ;I machine-determined site in the center o f
differed i n the load applied and the number of repetitions. the femoral neck (CV 3.3%)).
Genetic and cnvironment;il differences in bone density be- The radial forearm site WBS defined as the iirea o f the
tween subjects were controlled for by each subject acting a s radius at the ultraclistal site (UD), the midsite, and the one
their own control and exercising one side of their body only. third site. The forearm was in a horizontal position with the
~ I ' l i u athe study was designed t o address two questions. First, elbow resting o n the table and the hand held in a loose fist.
is the loxi more important thim the numbcr of loading The UD site was defined as the area from 2 pixels proximal
cycles i n increasing hone ma Second. is the action spe- to the base o f the articular surface. at the base o f the
cific to certain skclet:il sites'? ulnar-styloid process to 10 cm proximally. The midsite was
defined as the area extending from the proximal edge o f the
ultradistal area t o the distal edge o f the one third site. The
MATERIALS AND METHODS one third site was defined as the area extending from the
edge o f the midsite. The CV for the radius UD site was
Suhjects wci-e recruited via media articles inviting them I.h%, for the radius midsite 1.4%, and for the radius one
to participate in ;I resistance exercise program. The respon- third site 1.3%.
dents were initially screened using ;I short questionnaire
iidministercd by telephone interview. Women were eligible
Study c1e.r;gtz
for the study if they were between I and 15 years postnieno-
paiisal ;ind aged hetwecn 30 and 70 years. The date of Subjects were allocated by ;I block randomization design
ineiiop:iuse was defined a s the date o f the last menstrual to one o f two progressive resistance training groups that
period. For hysterectomized women, the onset o f hot flushes differed in the amount of resistance o r load and the number
was used to date menopause. Three hysterectomized women of repetitions. The exercising limb was allocated by ran-
were included in the study. All subjects were physically domization to either the left side or the right side with the
cqxihlc o f entering the exercise groups but were not already contralateral side acting as the nonexcrcising control. The
exercising inore than 3 hiweck at ii high intensity. Women high load and low repetition (strength) groups were re-
who were playing o r who had played racquet sports o r who quired to complete three sets o f eight repetition maximum
were engaged in resistance training in the previous 5 years (XRM, the maximum load that can be lifted eight times but
were excluded. Other exclusion criteria were a femoral neck not nine) for each exercise. The low load and high repcti-
bonc density outside the norm;il reference range for aged- tion (endurance) completed three sets o f 20RM (the max-
matched controls (more than 2 SD from the mean) and imum amount that can be lifted 20 times hut not 21).
significant chronic diseases or medications known to affect Subjects in both groups were continually encouraged t o
hone density including estrogen. steroid hormones, or thi- attempt a n additional repetition so that there was a contin-
aziclc diuretics. Cardioviisciiliir disease, a history o f high ual overload of the muscle or miisclc groups and a progres-
hlood pressure (ahove 1OO/%), or any physical or orthope- sive increase in res nee throughout the study. When the
dic disabilities that placed the suhject at risk or limited their strength group participants were able t o complete three sets
ability to perform weight training exercises were also exclu- o f 10 repetitions they were advised t o attempt the next
220 KERR ET AL.

available weight (usually in 2.5 t o 5 kg increments). Simi- TABLE I . WHOLE-BODY BONLDENSITY, BODYCOMIWSI IION,
larly when the endurance group participants were able to D A ONEREPE.III.ION
A N D WE1cti.r L I ~ EFOR M A X I M UTM IN
complete three sets of 25 repetitions the subject attempted ( K G ) OF TI1E E N D U R AAN
N DCSTRENGTII
~ GROUI’S A T BASf:I.INt
the next weight increment. The choice of repetitions was
Endururice Strength
based o n standard resistance training protocols to deter-
mine if loading was more important than repetition in Characteristic group group
preventing bone loss. Traditionally these two protocols have No. of subjects 28 28
been referred t o as strength and endurance strength proto- Age (years) 55.7 -t 4.7 58.4 !3.1*
cols. Both groups exercised to maximum repetitions and Years since menopause 6.3 -t 4.0 7.8 !3.5
were required to attend three sessions per week with one Body mass (kg) 70.8 -t 10.0 69.4 !1 1.4
rest day in between. All sessions were supervised by the Stature (cm) 165.2 !6.1 165.2 !7.0
same instructor and took place in the Human Movement whole body bone density 1 .OO 2 0.12 0.98 2 0.10
Department Weight Training Laboratory with the excep- body fat (kg) 28.2 !8.6 26.5 5 8.4
tion of four subjects who exercised their upper limb only at percent body fat (%) 40.4 t 7.1 38.1 t 7.7
home. The resistance training was designed to stress one lean body mass (kg) 40.5 t 4.0 41.5 t 5.7
upper limb and one lower limb, with the alternate limb I-repetition maximum
acting as the control for each subject. (wt lifted in kg)
Changes in muscle strength were assessed by a one rep- wrist curl 10 !2 10 5 2
etition maximum test ( I R M ) for each lift.‘”’) Subjects reverse wrist curl 6? 1 6 t2
warmed up with 5-10 repetitions at 40-60% of the per- wrist pronationisupination 9 2 2 9?2
ceived maximum. After stretching and a 1 minute rest, three biceps curl 6 2 2 6 5 1
to five repetitions were performed at 60-80% of the per- triceps push down 15t3 15 5 3
ceived maximum. The 1 RM was determined as the weight hip extension 53 t 13 56 ? 14
that could be lifted once but not twice. Subjects rested for hip flexion 45 t 10 45 ? 12
2-3 minutes between each attempt. The 1RM testing was hip abduction 44 t 13 4 2 2 12
done at the commencement and completion of the study. hip adduction 4 7 t 11 50 i- 15
Both free weights and resistance machines were used. leg press 48 t 15 42 i- 14
The exercises used on the exercising limb were (1) upper
limb-biceps curl, wrist curl, reverse wrist curl, triceps ex- * p < 0.05 for age between endurance and \trength group.
tension (with pulley), and forearm pronation and supina- Results are mean 5 SD.
tion (with dumbbell), and (2) lower limb-leg press, hip
abduction and adduction, hamstring curl, hip flexion, and
hip extension. The strength group started with 60% of IRM sion line used reflected the change in bone density in that
for the leg and 40% of 1 RM for the arm, and the endurance individual. To test if the slope of the BMD for the exercis-
group started with 20% of 1RM for the leg and 10% of ing limb was significantly different from zero change, a
IRM for the arm.(”) In the first 3 weeks, subjects were paired sample t-test was performed. The summary statistic
taught correct lifting techniques, after which time the was pooled for the control and exercise limb for each group
strength group exercised at a level to reach a maximum of and used to calculate the percent of bone loss per year. All
eight repetitions for three sets (i.e., 3 x 8RM), and the results are reported as the mean and standard deviation. All
endurance group exercised at a level to complete 20 rcpe- p values are two-tailed.
titions for three sets (i.e., 3 X 20RM). Subjects were in-
structed t o rest for approximately 2-3 minutes between sets.
All subjects completed stretching of the involved muscles RESULTS
prior t o the session, three sets of each exercise, and stretch-
ing at the end of each session. To be compliant, subjects There were no significant differences at baseline in the
were required to attend more than 75%’ of sessions for the whole body bone density, body mass, stature, or percentage
study duration. For each subject, the number of sets eom- of body fat between the two groups (Table 1). The strength
pleted, the weight lifted, and attendance at each session group was older ( p < 0.05), but there was no significant
were recorded. Subjects were instructed to maintain their difference in years since menopause. Table 2 shows the
usual diet throughout the study. Adherence was calculated baseline BMD for the hip and forearm sites comparing the
a s the percentage of subjects who completed 12 months of exercise with the control side for the strength and cndur-
testing and the resistance exercise program. anee groups. There was a significant difference hetween the
Statistical analysis was conducted using SPSS for Win- exercise and the control side for the endurance group at the
dows. A linear regression function was fitted for each indi- radius one third site.
vidual completing three or more bone density estimations. Ten subjects did not complete the study (3 strength
The slope was compared for the exercise and control side group, 7 endurance group). Of these 10,7 withdrew because
using a paired t-test for the hip and forearm bone mineral of the time commitment. The strength group took between
density. All p values were calculated from the mean slope 20 and 30 minutes to complete the session, whereas thc
difference; however, the bone density data are presented as endurance group took 45 to 60 minutes. No serious injuries
the percentage change from baseline for clarity. The regres- occurred as a result of the program. However, several sub-
EXERCISE EFFEC'I'S O N BONE MASS IN WOMEN 22 I

Gidiiruticc grorip

Si/c C'otitrol siclc E.wrcisc sirlr Control side Excrcise sidc

I lip hone mineral density

trochantcric 0.h5 t 0.10 0.05 t 0.I0 0.64 t 0.I I 0.05 t 0. I I
intcrt rochan tcric I .02 t 0.15 I .02 t 0.IS 0.w i 0.15 0.w t 0 . 15
neck 0.7s t 0. I 1 0.75 t 0.12 0.72 I t 0.10 0.72 t 0 . I 0
Ward's triangle 0.(10 t 0. I5 0.58 t 0.10 0.56 ' I 0.12 0.55 t 0.10
Fo re ar ni bone m i ne r;i I de nsi t y
rxlius ultra distal 0.36 ? 0.00 0.37 t 0.06 0.37 I+ 0.05 0.36 t 0.06
r;idiiis mid 0.52 -c 0.07 0.53 t 0.07 0.50 t 0.06 0.50 t 0.07
radius l/3 0.61 t 0.08 0.60 2 0.08* 0.59 I+ 0.07 0.56 ? 0.08
*/I . 0.05. tlic cxerciw side is significantly different from the control aide.

Sircwgili group

Sitc CotI tro1 sirlc Excrcisc, .side C'otitrol .sidc fixc~rcrtrcidc

-
I lip (tz = 19) (ti = 23)
trochanter + 1.2 t 3.6 +0.3 t 2.8 -0.0 t 2.2 +1.7 f 4.1*
intcrtroch;intcr +o.s t 2.5 +0.3 5 2.4 -0.1 -t 2.1 t 1.5 ? 3.01
neck o f femur - 1.0 i 3.1 +0.2 t 4.2 -0.4 t 2.4 +o.o 2 3.1
W;ird's triangle +0.9 ? 0.6 + I.9 t 8.Y +0.8 t 5.2 +2.3 -c 4.0"
Forearm (ti = 21) (ti = 25)
radius ultra distal -0.3 t 3.5 -0.2 t 5.9 -1.4 t 2.3 +2.4 t 4.3*
rxliiis mid - 1.0 i 2.3 +O.l ? 1.4* -0.7 t I .Y +0.5 -c 3.2
r;idius 113 -0.0 t 2.4 +0.4 t 2.8 -0.2 t 2.2 +O.Y t 3.5

* p . 0.01 arid ' / I I 0.05 for the comp;criaon with control side.
.

11 viilucs iirc c;ilcul;itcd trom the slope.; of the regression lines fitted to the bone density data for each auhject with three or more data
poiiits.

jccts developed mild tendonitis mid worked at a reduced comparing the exercise with the control limb for the same
load for approximately 3 weeks until the symptoms sub- subject. There was a significant increase in bone mass for
sided. One subject broke her ankle after 9 months in the the strength group for the exercising limh compared with
study while walking. She continued t o exercise her upper the nonexercising limb at the trochanteric hip site (Fig. la),
limh at Iiome but her lower limb data was only included for the intertrochanteric hip (Fig. 1 b) site, and Ward's triangle
the first 0 months. Three other subjects were unable to (Fig. Ic). However, for the endurance group there were n o
attend the gymnasium but were supplied with weights so significant difference in BMI) between the exercise and
that they could continue t o train their upper body at home. control side at any o f the hip sites. At the forearm ultradis-
Only the upper limb was included unless they had com- tal site, the strength group increased in bone mass (Fig. 221).
pleted 0 months o f the study in the gymnasium in which the The endurance regimen had a small effect at the radius
upper and lower limb was trained. The overnll adherence to midsite (Fig. 2h). In addition, for the strength group, the
the program. including these four subjects was 8 2 % . The positive slope o f the BMD for the exercising limb wiis
overall compliance to the exercise regimen (attendance) for significantly different from zero change at the forearm
all subjects was 87% (range 07-100%~).The compliance for ultradistal site ( p < O.OS), the intertrochanteric hip site ( p
the strength and endurance groups was 87% (range 67- < 0.05), and the Ward's triangle site ( p < 0.05). Furthcr-
0tif'; ) and SOc;' (range W - I O O % ~ ) , respectively. There was more, the negative slope of the BMD for the control limb at
no significant diffcrcncc in exercise compliance between the forearm ultradistal site was significantly different from
the two groups. zero change ( p <: 0.05).
_I :hie
. 3 shows the mean percentage changes from base- Muscle strength, tested by a IRM test, increased signif-
line i n the hip and forexni bone mineral density (BMD) icantly for all 10 exercises but did not differ between the two
222 KERR ET AL.

A 2.50 A 1.50

2.00 1.00
nu
2 1.50

P 0.50
2u 05 0.00
s
4.50 -1.50

-1.00 -2.00
0 3 6 9 12

TIME (MONTHS)

B
2.00 -
2.50
1.50 - 2.00

0.50
;es 0.00
-0.50
-1.00
,

0 3 5 S 12
-(Mom)

c 5.00
4.50
4.00
3 1
FIG. 2. The percentage change ( 5 SEM) from baseline in
bone mineral density over the 12 months of the study at the:
(A) forearm mid site for the endurance group, the exercise
side was significantly different ( p < 0.01) from the control
side; (B) forearm ultra distal site for the strength group, the
exercise side was significantly different ( p < 0.01) from the
control side. All p values calculated from the mean slope
difference for individuals with three o r more time points.
0 Exercise side. 0 control side.
0.50
0.00
4 0 ’ 1
By comparing the slope difference between the exercise
0 3 6 0 12 and the control side for the strength and endurance group
TIMEW-) by an unpaired t-test, at the ultradistal forearm site the
strength group had a significantly greater increase ( p <
FIG. 1. The percentage change ( ? SEM) from baseline in
BMD ovcr the 12 months of the study in the strength group 0.05) in bone mass than the endurance group. At the tro-
only at: (A) the intertrochanter hip site, the exercise side chanteric hip site the slope difference in the strength group
was significantly different ( p < 0.05) from the control side; was greater ( p < 0.01) than in the endurance group.
(B) the trochanter hip site, the exercise side was signifi-
cantly different ( p < 0.01) from the control side; and (C)
the Ward’s triangle hip site, the exercise side was signifi- DISCUSSION
cantly different ( p < 0.01) from the control side. All p
values calculated from the mean slope difference for indi- In this study we have produced data t o support extensive
viduals with three or more time points. 0 Exercise side, animal studies to demonstrate that the skeleton adapts t o
0 control side. the increasing load applied by progressive resistance training
in postmenopausal women by increasing in bone density.
Furthermore, this effect appears to be specific for certain
groups (Table 4). There were significant correlations be- skeletal sites.
tween the percentage change in 1RM and the slope of the Although animal studies would suggest that resistance o r
bone density at various hip sites for the strength group weight training may have a beneficial effect o n bone mass.
(Table 5 ) . There were no correlations between the change few prospective studies have been undertaken in postmeno-
in IRM and the bone density in the endurance group at the pausal women. A 12 month study in premenopausal women
hip sites or in either group at the forearm sites. failed to shown an effect o f weight lifting on bone density at
EXERCISE EFFECTS ON BONE MASS IN WOMEN 223

I-KM for lower body (n = 19) (I1 = 23)

hip extcnsion 5 1 5 12 98 t 18 102 t 57 54 t 14 99 t 22 94 t 83
hip flexion 43 t 9 87 2 18 l l O ? 49 45 t 13 86 t 26 106 5 97
hip ahduction 4 2 5 13 82 t 22 128 +- 139 4 2 5 13 85 t 27 I I4 5 YO
hip adduction 45 i- 8 86 t 18 96546 49 t 14 8 4 ? 17 86 5 8')
Icg press 46 -+ 15 69 t 20 505 44 41 i- 13 73 t 26 95 2 I06
I-KM for upper body ( I ? = 21) (n = 25)
wrist curl 10 +- 2 12 t 2 21 +- 20 1052 13 ? 2 38 t 40
reverse wrist curl 6 2 1 8t2 38 +- 42 6tl 8 5 2 40 5 42
wrist pron~itionisupination 9t2 11 t 3 25 +- 22 9t2 12 ? 2 33 t 26
biceps curl 6 5 2 9 5 2 52 ? 36 6 5 I 10 r 2 67 +- 43
triceps push down 15 5 3 2s 5 6 68 2 36 15 +- 3 26 t 6 71 2 26
I<esults ;ire niciiii 5 SD. All exercises were significantly different from baseline. There were n o significant differences hctwccn the
strength and endurance groupa.

Endururzce group Strcwgth group

~ ~~

~;..\-l~rcisl~ Hip site cow. cot$? p vulul~ cow. coqfl: p vulul~

Hip cxtension t roc han t e r 0.1 1 NS 0.43 0.05

Leg press trochanter 0.02 NS 0.60 0.0 I
Leg p1ess inter t roc han tcr -0.22 NS 0.44 0.05
Hip adduction Ward's triangle 0.32 NS 0.45 0.05
ltip extension Ward's triangle 0.44 NS 0.5I 0.0 I

Results arc iiieiin ? SD.

the spine or o s calcis despite there being a significant enough to provide an osteogenic stimulus equivalent t o
increase in strength in the weight lifting group.('") I n a 9 weight training.
month resistance training study, in premenopausal women A recent study which examined the bone mass o f gym-
there was ;I 4% decrease in BMD at the lumbar spine in the nasts and runners with a similar prcvalencc o f menstrual
exercise group. and the authors questioned the safety of disturbances found gymnasts had a higher femoral neck
weight training."S' liowever. subjects were not random- BMD than the runners or controls.'"' Higher impact forces
ized. there were only two scssionsiwcck with a total of 17 from the gymnastics training were thought to account for
suljccts. A lack of exercise cffcct o n BMD was also found the differences. These findings support the results of the
in ;I I year home-based weight training program in a group current study that the magnitude of loading is more impor-
o f middle aged women; despite ii significant increase in tant than the number of cyles or repetitions.
strength there was n o increase in BMD.@) Similarly, a There are several factors that may account for the signif-
uniliitcral strength training study in young women showed icant increase in BMD from resistance exercise that wc
an increase in strength but n o increase in hone density over have demonstrated that have not been uniformly demon-
;I ycx.' I") However. a n 8 month randomized resistance strated in previous studies. First, 82%) o f the 56 subjects
training and running intervention in younger women did completed the study, and the compliance t o the exercise
show ;I significant increase in BMD in both groups at the program was high. In addition, subjects were randomized t o
lumbar spine. suggesting that either running or weight both the exercise protocol and the exercising side. By using
training may he heneficial i n young woman.('') This study each subject as their own control, genetic and environmen-
appears t o conflict with the results o f the current study. tal determinants of bone density were controlled for so that
However. ;is suggested by the authors themselves, the num- the question of site specificity could be addressed. A limi-
her of load rcpetitions from running may have been high tation o f this study was that the resistance training was for
224 KERR ET AL.

I year only. but in contrast to most previous studies we due to forces transmitted t o the control leg while the excr-
examined thc effects of exercise o n the postmenopausal cise leg was working as four of the five lower body exercises
estrogen-deficient skeleton where there is ongoing bone were undertaken while standing o n the control leg.
loss. In the upper limb there was a significant change in BMD
Since increasing muscle strength was correlated with the in the strength group at the forearm ultradistal site and at
increase in bone m i i s at several hip sites for the strength the midsite in the endurance group. Overall in the strength
group, it would suggest that the mechanism by which the group the percentage change in bone density at the radial
osteogenesis occurred is by muscle pull. This action of ultradistal site was greater than at the trochanter and inter-
muscle pull is mediated through the force o f the muscular trochanter sites. However, the changes in strength in the
contraction at the site o f attachment of the tendon onto the upper limb were a lot smaller than in the lower limb. This
bone, thus the hone may respond locally t o rearrange the suggests that the load threshold for osteogenesis is much
forces generated from the muscle at the site o f loading. lower for the upper limb. Thus it is possible that different
There wiis n o significant increase in BMD at the neck of sites in the skeleton may have different osteogenic thresh-
femur site, ii site that does not have muscle attachments. olds for loading. For example, the non-weight bearing
However, the trochanter and the intertrochanter sites do bones of the forearm may not require as great ii load to
have sites for attachments o f muscle groups. The psoas and induce ostcogenesis as the weight-bearing bones o f the
ilixus muscles, which are responsible for hip flexion for the femur. The percentage change in muscle strength was n o t ;is
Icg press exercise, both insert o n the lesser trochanter. It is great for the upper body exercises as it was for the lower
therefore interesting t o note that in the strength group the body exercises, suggesting that the change in force required
increase in strength of the muscle associated with hip flex- to stimulate osteogenesis via muscle pull was much lower at
ion correlated with the change in bone density at the inter- the forearm site.
trochanteric site, an area that includes the lesser trochanter Studies where the upper body is used as the site o f
but not other areas. Hip extension is primarily due to the loading generally have shown a positive effect o f exercise.
contraction o f the gluteus maximus which inserts on the For example, lifetime tennis players have shown ;I greater
greater trochanter. In this case in the strength group the bone mass in the playing arm compared with the nonplaying
increase in strength of the hip extensors was correlated with arm,(ll.12,25’ In a study by Beverly(3) in which elderly suh-
the increase in bone density at the trochantcr site, an area jects were required to squeeze a tennis ball for 30 s daily for
that includes the greater trochanter. The changes in bone 6 weeks, there wiis a significant increase in the bone mineral
density in Ward’s triangle, found in the center o f the fem- content at the forearm (3.4%) and grip strength at the wrist
oral neck, also correlated with changes in the strength of (14.5%). The effect of isometric exercises for the distal
muscles controlling hip adduction and extension. forearm on bone mass was examined in a group o f post-
I t is important t o recognize that both groups in this study menopausal women over 5 months.‘” Bone density in-
gained similar amounts of muscular strength, but the gain in creased by 3.8% in the exercise group and dccreased by
bone density was only significantly increased in the strength 1.9% in the control group. Both of these studies were o f
group, who were lifting heavier weights but with fewer relatively short duration and did not place as large a load on
repetitions compared with the endurance group. Since the skeleton because weight lifting exercises were not used
there was n o difference in strength gained between the but nevertheless showed an increase in bone density.
strength and endurance groups, this suggests that the os- In conclusion, postmenopausal bone density at clinically
teogenic response is related to the peak instantaneous load important sites can be significantly increased by a low rep-
o r the magnitude o f the change in load applied to the bone etition high-load regimen but not by a high repetition low-
rather than the ability for muscle hypertrophy. This conclu- load regimen. Furthermore, in the strength group but not
sion agrees with the findings in animal models which the endurance group the greater the increase in muscle
showed that osteogenesis wiis induced by dynamic rather strength the greater the increase in bone mass. Thus these
than static loading and that the magnitude of the load was results support the notion of site-specific response o f bone
more important than the number o f loading cycle^.("'^'^' to loading from re nce exercise in that although the
Importantly these results support the notion of a site- trochanter and intertrochanteric bone mass was elevated by
specific response of bone to exercise in that there was a the resistance exercises undertaken there was n o effect on
significant increase in bone density at some sites in the the femoral neck value. It is concluded that the load is more
exercising limb but not at others when compared with the important than the number of loading cycles in increasing
control limb. This supports our previous finding of a site- bone mass in early postmenopausal women. A progressive
specific effect of walking at the femoral neck site but not resistance program that uses a high-load low-repetition pro-
other sites.“” Walking was found to be effective in slowing tocol can be recommended as a practical and time-effective
bone loss from the legs as well as the trunk in a group of approach for the treatment and prevention o f osteoporosis
healthy postmenopausal women.(74)The findings were con- in postmenopausal women.
sistent with ;I local effect of walking on bone density. This
a n d the previous study suggest that some areas of the
skeleton arc sensitive to gravitational forces while others ACKNOWLEDGMENTS
arc sensitive t o muscle pull. The increase in bone density in
some area o f the control side hip o f both the endurance and This work was supported by Hcalthway, the Western
strength group wiis somewhat surprising and may have been Australian Health Promotion Foundation.
EXERCISE EFFECTS ON BONE MASS IN WOMEN 225

REFERENCES 17. Snow-Harter C, E3ouxsein ML, Lewis BT, Carter DR, Marcus
R 1992 Effects o f rcsistance and endurance exercise on hone
I. I>;ilsky GP. Stocke KS. Ehsani AA. Sliitopolsky E, Lee WC, mineral status of young women: A randomized exercise intcr-
I3irgc SJ 19XX Weight-hearing exercise training and lumbar vention trial. J Bone Miner Res 7:761-709.
hone miiicriil content in postmenopausal women. Ann Intern 18. Rockwell JC, Sorcnson AM, Bakcr S,Leahey D, Stock J L 1000
MeJ lOX:X24-S28. Weight training decreases vertehral hone density in premeno-
7. Sinikiii A. A y h n J. Leichter I 1987 Increased trabccular bone pausal womcn: a prospective study. J Clin Endocrinol Mctah
density due t o hone-loading exercises in postmenopausal OS- 71~988-993.
teoporotic women. Calcif Tissue Int 4059-63. 19. Vuori I, Heinonen A, Sievanen H. Kannu\ P, Pasancn M, Oja
3. t3cvcrly M('. Rider TA. Evans MJ. Smith R lY89 Local bone P 1994 Effects o f unilateral strength training and detraining o n
mineral response to brief exercise that stresses the skeleton. Br bone mineral density and content in young women: a study o f
Mcd J 299233-235. mechanical loading and deloading on human bones. Calcif
4, Smith El-. Gilligan C. McAdam M. Ensign CP, Smith PE 1989 'Tissue Int 55:59-67.
Deterring bone loss by exercise intervention in premenopausal
20. Kracmer WJ, Gordon SE, Fleck SJ, Marchitelli 11,Mello R,
and postmcnopausnl women. Calcif Tissue Int 4 4 5 12-321.
Ilziados JE, Fried1 K, Harman E 1991 Endogenous anabolic
5 , (;rove KA. Londeree BR 1902 Bone density in postmeno-
hormonal and growth factor responses t o heavy resistance
p;ius;tl women: high impact vs low impact exercise. Med Sci
exercise in males and females. Int J Sports Med 12:228-235.
Sports Excrc 24( I1):IlO(l-l 104.
(1. ('how R. Harrison JE, Notarius C I987 Effect of two random-
21. Pollock ML, Wilmorc J H 1990 Exercise in Health and Disease:
programmes on hone mass of healthy postmeno- Evaluation and Prescription for Prevention and Rehabilitation.
p;iusal women. Br Med J 2951441-1444. 2nd Ed. WB Saunders Company, Phil;idelphia, Pennsylvania.
7. Cnv;tn;iugh DJ. Cnnn CE IOXX Brisk walking does not stop 22. Robinson TL, Snow-Harter <', Taaffc DR, Gillis D, Shaw J,
hone loss in postmenopausal women. Bone 9:201-204. Marcus R 1995 Gymnasts exhibit higher hone m
8. Peterson SE. Peterson MD, Raymond G, Gilligan C, Checo ncrs despite similar prevalence o f amenorrhea and oligomcn-
vich MM. Smith EL 1991 Muscular strength and bone density orrhea. J Bone Miner Res 10:2h-35.
with weight training in middle-aged women. Med Sci Sports 23. Prince RL, Devine A, Dick I,Criddlc A, Kcrr D, Kent N, Price
11.tcrc 23:409 -504. R, Randell A 1995 The effects of calcium supplementation
0 . Martin I>. Notclovitz M I003 Effects of aerobic training on (milk powder or tablets) and exercise on bone density in post-
hone mineral density of postmenopausal women. J Bone Miner menopausal women. J Bone Miner Rcs lOlOhX-107S.
Res x:o3 I -03h. 24. Krall EA, Dawson-Hughes B 1994 Walking is related to hone
10. 1,;inyon I , € , Ruhin CT I084 Static vs dynamic loads a s an density and rates o f bone loss. Am J Mcd 9620-26.
intlucncc o n hone remodelling. J Biomech 17:X97-905. 25. Pirnay F, Bodeux M, Crielaard JM, Franchimont P 1987 Bone
I I . tluddlcston AL. Rockwell D, Kulund DN, Harrison R B 1980 mineral content and physical activity. Int J Sports Med 8331-
I3one mass in lifetime tennis athletes. JAMA 2441 107-1 109. 335.
12. Jones HH, Priest JD. Hayes WC. Tichenor CC, Nagel DA 1977
I lumcr;il hypertrophy in response to cxercixe. J Bone Joint
Surg 59A304-20X. Address reprint requcsts to:
13. Prince RL. Price RI, H o S 19x8 Forearm bone loss in hemi- D. A . Kew
plcgix ;I model for the study o f immohilization osteoporosis. J
U t i i v c w i ~D i p r t m c n t of Medicine
13onc Miner Res 3:305-3 10.
14. Ruliin CT. Lainyon LE I985 Regulation o f bone mass hy me-
4th Floor, G block
chanical strain magnitude. Calcif Tissue Int 3741 1-417. Queen Elizabeth I1 Medical Centre
15. Notclovitz M, Martin D. T e s u R, Khan FY, Probart C, Fields Sir Charles Gairdner Hospital
C', McKcnric 1- 1991 Estrogcn therapy and variable-resistance Nedlatids, P u t h
weight training increase hone mineral in surgically menopausal 6009 Westerri Australia
women. J Bone Miner Res 6(6):.5X3-590.
Ih. <;Iceson PB. Protas EJ, LeBlanc AD, Schneider VS, Evans HJ
lOO(l Effects of weight lifting on bone mineral density in pre- Received in original form December 5, 1094: in revised form May
niciiopiiusiil women. J I3one Miner Res 5(2):153-158. 22, 1995; accepted December 1 I, IOOS.