European Journal of Clinical Nutrition (1999) 53, 211±215
ß 1999 Stockton Press. All rights reserved 0954±3007/99 $12.00
http://www.stockton-press.co.uk/ejcn
Longitudinal changes in radial bone density in older men
TR Overton1 and TK Basu1*,
1
Department of Biomedical Engineering and Department of Agricultural, Food & Nutritional Science, University of Alberta, Edmonton,
Alberta, Canada
Objective: To examine changes in radial bone density and biochemical status, with particular reference to
calcium, over 18 months in a group of older men.
Subjects: Thirty-six healthy men (aged 66 ± 76 y) were recruited to the study during July and August, 1993.
These men were free-living residents of Edmonton who were recruited through local organizations for the retired
and semi-retired. Data for the younger group of men (age 29 ± 60 y) were taken from a previous study conducted
in our laboratory.
Design: Using special-purpose computed tomography (gCT), trabecular (TBD), cortical (CBD) and integral
(IBD) bone densities (gm=cm3) were measured in the ultra-distal radius at 6-month intervals over 18 months. At
baseline, and at each subsequent study visit, serum was obtained from each subject for determinations of
calcium, phosphate, 25-hydroxyvitamin D, alkaline phosphatase, and immunoreactive parathyroid hormone. A
24-h urine sample was also obtained at each study visit for determination of urinary calcium, phosphate and
creatinine.
Results: In repeated measures analysis of variance of the data for the older men serum 25-hydroxyvitamin D was
signi®cantly decreased (P < 0.001) over time, while TBD was increased ( ‡ 0.60% per year, P < 0.01).
Longitudinal rates of change for TBD, CBD and IBD were: 7 0.94%, 0.92% and 0.74% per year respectively
when bone density data at baseline for the older men and the historical data for younger men were combined.
However, separate analyses of the data for the younger and the older men indicated no signi®cant age-related
changes in bone density for men aged 29 ± 60 y, or for men aged 66 ± 76 y. However, differences in TBD, CBD
and IBD between the younger and older groups of men were signi®cant (P < 0.001).
Conclusions: In a group (n ˆ 36) of older men (mean age 71.7 y) studied longitudinally over 18 months, bone
density in the distal radius did not decrease over time. Mean bone density in this group of men was, however,
signi®cantly (P < 0.001) lower than in a group of younger men (n ˆ 17, mean age 46.7 y). Regression analysis
using cross-sectional bone density data at baseline for the older male group, and historical data for the younger
male group, indicates that bone loss occurs with increasing age at a rate of about 1% per year averaged over ages
29 ± 76 y. Bone density variables were not correlated with either height or weight, or with any biochemical or
hormonal variable measured in this study.
Sponsor: The work was funded in part by the Dairy Bureau of Canada.
Descriptors: bone density; computed tomography; 25-hydroxyvitamin D; older men
Introduction
Decreased bone mass is a well-known consequence of
increased age for both men and women. The cause of
age-related bone loss, however, has been more carefully
investigated for women than for men. Using radiographic
photodensitometry (Garn, et al, 1967; Newton-John &
Morgan, 1970), photon absorptiometry (Riggs et al, 1981;
Mazess, 1982; Yano et al, 1984) and quantitative computed
tomography (Meier et al, 1984), cross-sectional studies in
men have provided evidence for a progressive decline in
bone mass. Histomorphometric studies have also demonstrated age-related loss of both cortical and trabecular bone
volume in men, and further that this bone loss appears to be
the result of decreased bone formation with little or no
change in bone resorption (Francis et al, 1989). A longitudinal study involving men aged 30 ± 70 y (Orwoll et al,
1990) has shown that bone mass decreases at both axial and
*Guarantor, correspondence: TK Basu, Department of Agricultural, Food
and Nutritional Science, University of Alberta, Edmonton, Alberta,
Canada T6G 2P5.
Received 23 July 1998; revised 16 October 1998;
accepted 30 October 1998
peripheral sites with a rate of vertebral loss of about twice
that of radial bone loss (2.3% vs 1.0% per year).
While osteoporosis is still primarily considered a
women's disease, the clinical approach to aging in men
must also consider the consequences of skeletal degeneration as life expectancy increases. In order to improve the
clinical management of age-related bone loss in men, a
better understanding of causal mechanisms is necessary,
and for this purpose more longitudinal studies are required.
The present study was undertaken to monitor changes in
radial bone density and bone mineral metabolism in healthy
older caucasian men over an 18-month period.
Subjects and methods
Subjects
Through advertising in local organizations for the retired
and semi-retired, 36 healthy men aged 66 ± 76 y (mean 71.7
y) were recruited to this study during the months of July
and August, 1993. Subjects entered into the study were free
living in the community, and had no evidence of medical
conditions or took medications known to affect bone
metabolism. All subjects were non-smokers who consumed
Radial bone density changes in older men
TR Overton and TK Basu
212
only social quantities of alcohol. No subject exercised
regularly or had a history of minimal trauma fracture. All
subjects provided written informed consent to participate in
a manner approved by an institutional Ethics Review
Board.
The younger group of men (age 29 ± 60 y, mean 47 y)
had participated in an earlier study conducted in our
laboratory to validate the precision of the gCT method
(Overton & Wheeler, 1992). All subjects in this younger
group were also healthy, had no medical condition, and
took no medication known to affect bone metabolism; they
also had given written informed consent to participate in
that particular study.
Method
Trabecular (TBD), cortical (CBD) and integral (IBD) bone
densities (g=cm3) were measured in the ultradistal radius on
four occasions, at 6-month intervals over 18 months, using
special-purpose computed tomography (gCT) (Hangartner
& Overton, 1982; Hangartner et al, 1987). The distal limit
of the measurement site was about 2 mm proximal to the
lowest part of the end-plate and extended 14 to 18 mm (7 ±
9 CT slices) proximal from that plane. The measurement
volume is usually de®ned by two parallel planes, separated
by about 16 mm, which are approximately perpendicular to
the long axis of the radius (Figure 1). For each CT slice,
TBD is evaluated as the average density of the inner 45%
of the slice area; IBD and CBD are de®ned over 100% and
75 ± 90% of the slice area, respectively. Averaging these
values over several CT slices within this measurement
volume provides TBD, CBD and IBD for a well-de®ned
and accurately reproducible bone volume over repeated
measurements. At each time point, duplicate measurements
were made with the subject repositioned between measurements, and the average of these two measurements was
used in the regression analysis for bone density on age.
At baseline, and at each study visit, serum samples were
taken for determinations of ionized calcium, parathyroid
hormone (intact PTH), 25-hydroxyvitamin D, and alkaline
phosphatase; a 24-h urine sample in a container containing
6 mol=l HCl was also obtained at each study visit for
determinations of urinary calcium, phosphate and creatinine. Serum ionized calcium was measured by ion-selective
electrode (Ciba-Corning 634), serum immunoreactive parathyroid hormone was measured using the IRMA assay
(Allegro Intact PTH, Nichols Institute, San Juan, Capistrano, CA, USA). Urine calcium was measured by ¯ame
absorption spectroscopy (Perkin-Elmer Z5000), urine phosphorus and creatinine were measured by colorimetric
methods (Dupont ACA 3 and Beckman CX3, respectively).
Statistical analysis
Each study variable was characterized by its mean, standard deviation (s.d.) and standard error of the mean
(s.e.m.). To determine longitudinal rate of change in bone
density, a linear least squares regression of each bone
density variable on observation time (y) was calculated
with standard error of the estimate (s.e.e.) for each subject.
Average rate of change in bone density, expressed as a
percentage of the mid-study value predicted from the
regression line was then calculated from the individual
subject values. A one-sample t-test was used to determine
the signi®cance of the difference in mean rate of change
from zero. A multivariate, general linear model was used to
test the signi®cance of changes over time for other study
variables to determine interrelationships between variables,
and to assess their in¯uence on rate of change in bone
density. P-values < 0.05 (two-tailed) were considered to be
signi®cant. For the cross-sectional analysis, a linear regression on age was calculated for each bone density variable;
coef®cients from these regressions were then used to
estimate rates of changes in bone density with age.
Data analysis was performed using Systat (SYSTAT
Inc., Evanston, IL, USA) on a personal computer.
Results
In repeated measures over time, all continuous variables
were found to be distributed normally, and for each variable there was homogeneity of variance. General characteristics of the two study groups at baseline are shown in
Table 1. There were no signi®cant differences in height or
weight between the older (mean age 71.7 y) or younger
(mean age 47 y) groups of men, but mean values for the
bone density variables (TBD, CBD and IBD) were signi®cantly higher for the younger group (P < 0.001).
Regression analysis parameters for TBD, CBD and IBD
on age for the combined data for the younger and older
groups of men, and for the older and younger groups
separately, are shown in Table 2. Scatterplots of these
data for the combined data with the regression lines are
shown in Figure 2. For men above age 66 years, the crosssectional variability in bone densities, indicated by the
s.e.e. is much greater than for men below age 60 years.
Annual rates of change for TBD, CBD and IBD in the ultradistal radius, determined from the longitudinal data, are
Table 1 Anthropomorphic characteristics of 36 healthy, older men at
baseline, and historical data for 17 healthy, younger men.
Variable
Figure 1 Measurement site in the ultra-distal radius using g-CT. Bone
volumes used for averaging in the determination of trabecular (TBD),
cortical (CBD), and integral (IBD) bone densities are shown.
Age (y)
Height (cm)
Weight (kg)
TBD (g=cm3)
CBD (g=cm3)
IBD (g=cm3)
Older men (n ˆ 36)
Mean (s.d.)
71.7
173.0
79.0
0.297
0.618
0.473
(2.7)
(5.0)
(9.0)
(0.069)
(0.101)
(0.080)
Younger men (n ˆ 17)
Mean (s.d.)
46.7
176
74.6
0.374
0.778
0.596
(9.8)
(6.0)
(10.6)
(0.058)
(0.087)
(0.066)
Radial bone density changes in older men
TR Overton and TK Basu
213
Table 2 Coef®cients of linear regression for radial bone density variables on age for all men (ages 29 ± 76 y, mean
63.7 y), older men (ages 66 ± 76 y, mean 71.7 y), and younger men (ages 29 ± 60 y, mean 46.7 y)
All men (n ˆ 53)
Intercept (s.e.m.)
Slope (s.e.m.)
s.e.e.
P
Rate of change (% per year)
Older men (n ˆ 36)
Intercept (s.e.m.)
Slope (s.e.m.)
s.e.e.
P
Rate of change (% per year)
Younger men (n ˆ 17)
Intercept (s.e.m.)
Slope (s.e.m.)
s.e.e.
P
Rate of change (% per year)
TBD (g=cm3)
CBD (g=cm3)
IBD (g=cm3)
0.509 (0.044)
7 0.003 (0.001)
0.065
< 0.001
7 0.94
1.033 (0.076)
7 0.006 (0.001)
0.098
< 0.001
7 0.92
0.799 (0.052)
7 0.004 (0.001)
0.076
< 0.001
7 0.74
0.217 (0.315)
0.001 (0.004)
0.070
n.s.
‡ 0.35
1.175 (0.449)
7 0.008 (0.006)
0.100
n.s.
7 1.3
0.747 (0.361)
7 0.004 (0.005)
0.080
n.s.
7 0.86
0.488 (0.056)
7 0.002 (0.002)
0.053
n.s.
7 0.53
0.871 (0.093)
7 0.002 (0.002)
0.087
n.s.
7 0.30
0.701 (0.067)
7 0.002 (0.001)
0.063
n.s.
7 0.30
n.s., not signi®cant; s.e.m., standard error of mean; s.e.e., standard error of estimate.
Figure 2 Trabecular bone density (TBD), Cortical bone density (CBD) and integral bone density (IBD) versus age for all men at baseline.
shown in Table 3. TBD was increased signi®cantly
( ‡ 0.6% per year, P < 0.01); CBD and IBD were
unchanged over this time. In a multivariate analysis of
covariance, no variable measured at baseline had any
signi®cant effect on rate of change in TBD, CBD or IBD.
Biochemical results relating to bone metabolism are
summarized in Table 4. In repeated measures analysis of
variance, 25-hydroxyvitamin D was found to be signi®cantly decreased (P < 0.001) over the study period. No
signi®cant change in any other biochemical variable,
including calcium, phosphate, PTH, alkaline phosphatase
or creatinine was detected.
Discussion
Using special-purpose computed tomography, we made
longitudinal measurements of trabecular, cortical and integral bone densities in the ultra-distal radius in a group of
healthy men aged 66 ± 76 y (mean 71.1 years at entry).
Measured over an 18-month period, a statistically signi®cant increase ( ‡ 0.06% per year, P < 0.01) in trabecular
bone density was observed while CBD and IBD were
unchanged. Regressions of the bone density variables on
Table 3 Longitudinal rate of change in bone density predicted by the
regression for 36 healthy, older men.
Variable
3
TBD (g=cm )
CBD (g=cm3)
IBD (g=cm3)
Change per
year % (s.d.)
95% CI
P
‡ 0.600 (1.12)
‡ 0.040 (1.89)
‡ 0.212 (1.45)
7 1.25 to ‡ 2.43
±
±
0.007
n.s.
n.s.
Value given is average of individual rates of change with 95% con®dence
interval (CI). Signi®cance (P) of the change per year is for a 1-sample
t-test for a difference from zero; n.s., not signi®cant.
age were also calculated after combining the baseline data
for the older group of men with historical data from a group
of younger men (age 29 ± 60 y, mean 46.7 y). Crosssectional rates of change of TBD, CBD and IBD, derived
from this regression analysis, were 7 0.94%, 7 0.92%
and 7 0.74% respectively (Table 2, Figure 2). Using data
for the older group of men only or for the younger group
only, regression analysis indicated no signi®cant changes in
the bone density variable with age. However, the variability
Radial bone density changes in older men
TR Overton and TK Basu
214
Table 4 Mean values (s.d.) for selected biochemical variables at four times during study for 35 healthy, older men. Signi®cance of change in mean values
was determined in a multivariate repeated measures analysis of variance
Variable
25(OH)D3 (nmol=l)
iPTH (ng=l)
sCa2‡ (mmol=l)
sTotal Ca (mmol=L)
sAlk. Phos. (IU=l)
sPhos. mmol=l
uCa (mmol=d)
uPhos. (mol=d)
uCreat. (mmol=d)
Baseline
122
36
1.29
2.25
212
0.96
4.7
30
12.7
(48)
(16)
(0.04)
(0.08)
(54)
(0.16)
(2.1)
(8)
(3.2)
6 Months
96
36
1.30
2.30
215
0.93
5.0
33
13.7
(47)
(12)
(0.04)
(0.09)
(53)
(0.14)
(2.5)
(9)
(3.1)
12 Months
18 Months
72
35
1.30
2.30
220
0.93
5.0
31
13.5
62
36
1.31
2.31
211.
0.93
4.6
30
13.2
(25)
(15)
(0.04)
(0.08)
(58)
(0.12)
(3.0)
(8)
(2.7)
(12)
(14)
(0.05)
(0.09)
(57)
(0.14)
(2.3)
(8)
(2.6)
Change=year %
P
7 35
0
1
1.5
0
0
0
0
0
< 0.001
n.s.
P < 0.05
P < 0.01
n.s.
n.s.
n.s.
n.s.
n.s.
25(OH)D3, 25-hydroxyvitamin D; iPTH, intact parathyroid hormone; Alk. Phos., alkaline phosphatase; Phos., phosphate; Creat., creatinine; S, serium;
u, urine; n.s., not signi®cant.
in the bone density data for the older group of men was
much greater than that in the younger group.
Age-related appendicular bone loss in men has been
reported to begin between the early thirties and mid ®fties
and then to continue throughout life at a slower rate (Riggs
et al, 1981; Mazess, 1982; Meier et al, 1984). In the present
study, analysis of the combined cross-sectional data for the
older and younger men indicates that both trabecular and
cortical bone densities are decreased at a rate of about 1%
per year over the age range 29 ± 76 y. This rate of loss is
only slightly larger than that reported previously in crosssectional studies (Riggs et al, 1981; Meier et al, 1984;
Yano et al, 1984), and the same as that reported in a
longitudinal study (Orwoll et al., 1990). When our data for
the older and younger men are analyzed separately, bone
loss with age is not signi®cant for either group; however,
this ®nding could be the result of the large variability in
bone density indices in the older group of men, and of the
small number of men in the younger group.
In a longitudinal study of bone mass in men, Orwoll et al
(1990) found loss rates of 1% per year at both the distal and
proximal radial sites; bone loss was reported to commence
at about age 35 y and to continue throughout life. Further,
in men above age 65 y the rate of bone loss in the distal
radius (trabecular bone) was found to be slightly accelerated with respect to earlier bone loss, while rate of loss for
the proximal radius (cortical bone) was unchanged. Our
results for the combined cross-sectional data for men also
indicated bone loss rates of about 1% per year for both
trabecular and cortical bone at the distal radius. However,
separate analyses of the cross-sectional data for the older
and younger men, and our longitudinal data for the older
men, suggests that men in the late seventh and early eighth
decades do not continue to lose bone at the distal radius
site. Our ®ndings, for both longitudinal and cross-sectional
data, of no change with age in CBD or IBD in the distal
radius in men aged above 66 years, agree with the crosssectional data reported by Riggs et al (1981) for this same
age group. Riggs and his associates further reported no
signi®cant changes in radial cortical bone mass throughout
life. However, our combined cross-sectional data indicate a
decrease in TBD, CBD and IBD of between 10% and 15%
for men with a mean age of 71.7 y relative to men with a
mean age of 47 y. Our result also con¯icts with the ®ndings
of Yano et al (1984) and Orwoll et al (1990).
The older men in our study were healthy, well-nourished, and ambulatory. Average dietary calcium intake for
these subjects was 884 mg=d. At baseline, mean 25-hydroxyvitamin D (122 nmol=l) for the group was slightly above the
normal range for our laboratory (10 ± 120 nmol=l), while
the mean intact PTH (36 ng=l) was in the low normal range
(10 ± 65 ng=l). However, neither calcium intake nor baseline levels of 25-hydroxyvitamin D or PTH had any
in¯uence on the measured rates of change of any bone
density variable. Further, PTH was unchanged in the group
over the 18-month study period, while 25-hydroxyvitamin
D decreased signi®cantly (P < 0.001) but remained within
the normal range for this assay.
This is the ®rst report of a longitudinal study in older
men in which trabecular and cortical bone have been
separately evaluated using high-precision gCT. Our longitudinal data substantially agree with previous crosssectional reports of little or no bone loss in older men
(Riggs et al, 1981), but do not agree with previous longitudinal study in men in which a continuing loss of bone is
reported after age 65 years (Orwoll et al, 1990). However,
our cross-sectional data, including results for both older and
younger men, are in general agreement with Orwoll's
longitudinal results. To resolve the discrepancies between
these several different investigations, prospective studies of
bone loss in men throughout life are warranted.
Acknowledgements ÐContributors: T.R.O.: Recruitment of the study subjects, follow-up of the study, measurements of radial bone density, data
analysis, and preparation of the manuscript. T.K.B.: Assistance with
recruitment of the study subjects, biochemical analysis, data analysis,
and manuscript preparation
References
Francis RM, Peacock M, Marshall DH, Horsman A & Aaron JE (1989):
Spinal osteoporosis in men. Bone Mineral 5, 347 ± 357.
Garn SM, Rohmann CG & Wagner B (1967): Bone loss as a general
phenomenon in man. Fed. Proc. 26, 1729 ± 1736.
Hangartner TN, Battista JJ & Overton TR (1987): Performance evaluation
of density measurements of axial and peripheral bone with x-ray and
gamma-ray computed tomography. Phys. Med. Biol. 332, 139 ± 146.
Hangartner TN & Overton TR (1982): Quantitative measurement of bone
density using gamma-ray computed tomography. J. Comput. Assist.
Tomogr. 6, 1156 ± 1162.
Mazess RB (1982): On aging bone loss. Clin. Orthop. Rel. Res. 165, 239 ±
252.
Meier AE, Orwoll ES and Jones JM (1984): Marked disparity between
trabecular and cortical bone loss with age in healthy men. Ann. Intern.
Med. 101, 605 ± 612.
Newton-John HF & Morgan DB (1970): The loss of bone with age,
osteoporosis, and fracture. Clin. Orthop. Rel. Res. 71, 229 ± 252.
Radial bone density changes in older men
TR Overton and TK Basu
Orwoll ES, Oviatt SK, McClung MR & Deftos LJ (1990): The rate of bone
mineral loss in normal men and the effects of calcium and cholecalciferol supplementation. Ann. Intern. Med. 112, 29 ± 34.
Overton TR & Wheeler GD (1992): Bone mass measurements in the distal
forearm using dual-energy X-ray absorptiometry and g-ray computed
tomography: a longitudinal, in-vivo comparative study. J. Bone Mineral
Res. 7 (4), 375 ± 381.
Riggs BL, Wahner HW, Dunn WL, Mazess RB, Offord KP, Melton
LJ III (1981): Differential changes in bone mineral density of the
appendicular and axial skeleton with aging. J. Clin. Invest. 67, 328 ±
335.
Yano K, Wasnich RD, Vogel JM and Heilbrun LK (1984): Bone mineral
measurements among middle-aged and elderly Japansese residents in
Hawaii. Am. J. Epidemiol. 119, 751 ± 761.
215