JOURNAL OF BONE AND MINERAL RESEARCH Volume 5, Number 6, 1990 Mary Ann Liebert, Inc., Publishers

Influence of Age and Body Weight on Spine and Femur Bone Mineral Density in U.S. White Men RICHARD B. MAZESS,’ HOWARD S. BARDEN,’ PAUL J. DRINKA,’ STEVEN F. BAUWENS,3 ERIC S. ORWOLL,4 and NORMAN H. BELL’

ABSTRACT Bone mineral density (BMD) was measured in normal white males using 153 Gd dual-photon absorptiometry. Measurements were made on the lumbar spine ( n = 315) and on the proximal femur ( n = 282) utilizing three regions of interest. There was a small but significant age-related decrease in spinal BMD ( r = -0.11; -0.001 g/cm2 per year) and trochanteric BMD ( r = 0.27; -0.002 g/cm’ per year). The BMD of the other femoral sites decreased more rapidly; the femoral neck ( r = -0.58; -0.005 g/cm2 per year) and Ward’s triangle ( r = -0.69; -0.007 g/cm2 per year) declined by about 21 and 34%, respectively, from age 20 to age 70. These femoral BMD decreases were three to four times greater than those usually seen in the peripheral skeleton in males but less than the decreases of 25-30 and 40% in the femoral neck and Ward’s triangle of white females. This pattern of aging bone loss may partially explain the paucity of spine fractures and the lower incidence of hip fractures in males versus females.

INTRODUCTION TUDIES OVER THE PAST TWO DECADES have

demonstrated that the increase in fractures occurring in both sexes with aging is associated with a decrease in bone strength due in large part to a diminution of bone mineral density (BMD). Most studies have evaluated the bone status of women because of the very high fracture rate in postmenopausal women, but there is also bone loss in men and an increase in fracures, albeit of lesser magnitude. Largescale studies of normal values have been done in white populations, both male and female, at peripheral skeletal sites. These cross-sectional studies have demonstrated that appendicular bone loss in males has its onset at age 50-60 and proceeds at a rate of about 0.3%/year, about one-third that in females [see review by Mazess(*)].Studies using dual-photon absorptionmetry (DPA) of the common fracture sites in the axial skeleton, the spine and proximal femur, have concentrated on the changes in women. l 9 - l 7 )

S

In general these studies have shown that the initial spinal bone loss at or close to the menopause (10-15% over 5 years) tapers off, at least in the normal women. In the femur(9.11,17) there is some premenopausal bone loss followed by a menopausal loss of lesser magnitude than that in the By age 70 normal women exhibit about a 20% diminutiion in spinal BMD compared to young normal, a 25% diminution in femoral neck BMD, and a 40% decrease in BMD of the Ward’s triangle.‘”) Preliminary studies in men demonstrated a small aging decrease in spinal BMD and an intermediate aging decrease in femoral BMD.(’O) It would be understandable that men would show an advantage with regard to both spine fracture and hip fracture if they had slight spinal osteopenia and a lesser degree of femoral osteopenia than women. We examined the BMD values for men over a wide age range to better outline the pattern of bone diminution of the spine and proximal femur.

ILunar Radiation Corporation and Medical Physics Department, University of Wisconsin, Madison, WI. ’Department of Veterans Affairs, Wisconsin Veterans Home, King, WI 54946. 3Pharmacy Department, University of Wisconsin, Madison, WI 53792. 4Portland VA Medical Center and Oregon Health Sciences University, Portland, OR 97201. ’VA Medical Center and Medical University of South Carolina, Charleston, SC 29403.

645

646

MAZESS ET AL.

METHODS

64 outpatients and the other cases, so they were pooled with data from the other groups. The only significant difference among the three male populations analyzed for each decade of life was a lower trochanter BMD in South Carolina males aged 20-29 years. Similarly, when larger groupings were compared (20-39, 40-69, and 70+ years) there were no geographic differences in mean BMD or in the regressions of BMD on age. The assumption of homogeneity of variance for BMD at each bone site was tested by a one-way analysis of variance of the residuals of BMD on age for ages 20-39, 40-69, and 70-89. There was no significant heterogeneity of variances among these subgroupings. The results in men were compared to those seen in a large series of normal U.S. white women.'") The spine and femur BMD values in these women and the changes with age were identical to those seen in similar groups from Australia(14' and England.'12) Statistical analysis was done with Minitab (State College, PA) on a personal computer. Data distributions were expressed as mean and standard deviation. Body mass index (BMI) was calculated as kg/m2. Linear regressions were used to characterize the interrelationships of BMD, age, body weight, and BMI. Student's t-test was used to evaluate differences between mean BMD values and between regression slopes. All p values are two-tailed.

BMD was measured using 153 Gd dual-photon absorptiometry (LUNAR Radiation model DP3). Determinations were made on ambulatory white males (aged 20-89 years) in three different locations (Madison, WI, Charleston, SC, and Portland, OR). The origin of the data is shown in Table 1. The three bone scanners used were properly calibrated according to the three-chamber standard provided with the instrument. Measurements on this standard phantom at each location gave values within 0.5% of those expected. Measurements were made on the lumbar spine using the L2-4 region of interest. Measurements were made on the proximal femur, including the three conventional regions of interest (femoral neck, Ward's triangle, and trochanter). The pattern of aging changes and the mean values for results from the different centers did not differ significantly, so the data were pooled. The criteria for inclusion in the sample differed in the three centers. In South Carolina the subjects were all normal, ambulatory, and free of metabolic bone disease or drugs that affect skeletal metabolism. The subjects from Oregon were more rigorously defined. They were not only free of disease and medication, but lateral spine radiographs were done to exclude cases with detectable radiographic abnormalities, but not osteophytes. Many of the volunteers ( n = 102) in Wisconsin also were normal subjects who were free of disease and medications. Another group ( n = 112) were outpatient volunteers at the VA Hospital; all had lateral spine radiographs. About RESULTS 40% (n = 48) had conditions that may have influenced the skeleton and were excluded. Among these patients, there The average height, weight, and BMI were 179 f 7 cm, were 16 who had osteophytes in the L2-4 region; their 79 f 11 kg, and 25 f 3 kg/m2 (see Table 2). There was spine BMD averaged 1.54 g/cm2, a 20% increase above little change in weight or BMI with aging. The mean BMD normal. In addition, we excluded 32 with medical condi- values for each decade group are shown in Table 3. The tions affecting bone, including alcoholism ( n = 7), corti- standard deviations of the decade values were larger than costeroids ( n = 9), and absorptive defects ( n = 5 ) . We did usually seen in studies of females, particularly in the older not detect a significant difference between the remaining populations. This may reflect heterogeneity among males, that is, a possible subset of those who lose bone. The regresions of BMD with age are shown in Fig. 1 and Table 4; a comparison of regression slopes in males and females with f-test values is given in Table 5 . The sex differTABLE1. DISTRIBUTION OF NORMAL SPINEAND FEMUR ences were also tested using a Z test, but the results were BMD DATA(NUMBER OF CASES) BY AGEGROUP almost identical because of the large sample sizes. There AND GEOGRAPHIC LOCATION was only a small nonsignificant decrease in spine BMD values with age. In males of all age periods there was no sig20-39 40-69 70+ AN nificant change in spine BMD. In females this was true in young adulthood, but between 40 and 69 years the regresSpine Wisconsin 78 51 37 166 sion slope with age was fourfold higher than in males (p < Oregon 3 36 23 62 0.001). After age 70 females, like males, showed no significant age decrease. Largely as a consequence of the menoSouth Carolina 52 26 9 87 pausal loss, the regression slope of spine BMD was seven Total 133 113 69 315 times higher in women than in men. ~

Femur Wisconsin Oregon South Carolina Total

56 3 53 112

41 36 26 103

~~

35 23 9 67

132 62 88 282

In contrast to the spine, there was a linear decline of BMD in the proximal femur from young adulthood onward. The correlation of age and BMD at the femoral neck site was -0.58 (p < 0.001); the annual loss rate was about 0.5% from young adulthood. An even greater rate of loss (about 0.7%/year) and higher correlation ( r = -0.69, p

BONE DENSITY IN MEN

647

TABLE 2. MEANAND STANDARD DEVIATION OF HEIGHT (cm), WEIGHT(kg), AND BMI (kg/m’) IN NORMAL MEN BY DECADES Height

BMI

Weight

Age

n

X

SD

X

SD

X

SD

20-29

83

179.5

6.9

76.1

9.7

23.6

2.3

30-39

50

179.6

7.2

79.6

9.9

24.7

2.6

24.6

2.4

40-49

28

179.8

7.1

79.7

11.5

50-59

20

179.1

5.4

81.0

6.8

25.3

2.1

60-69

65

177.8

7.2

81.2

12.4

25.7

3.4

70-79

60

177.3

7.4

79.0

10.7

25.1

3.3

80-89

9

173.1

9.4

73.7

17.7

24.3

3.5

TABLE3. BMD (g/Cm’)

FOR THE

LUMBARSPINE AND THREEFEMORAL SITES IN NORMAL MEN BY DECADES Femur

Spine (L2-L4) Age

n

~

X

SD

Neck

n

X

Ward‘s SD

X

Trochanter

SD

X

SD

~

20-29

83

1.26

0.14

71

1.09

0.14

1.00

0.15

0.93

0.13

30-39

50

1.26

0.13

41

0.98

0.12

0.86

0.13

0.86

0.11

40-49

28

1.31

0.18

24

0.97

0.16

0.84

0.16

0.89

0.14

50-59

20

1.21

0.14

18

0.92

0.11

0.74

0.14

0.84

0.13

60-69

65

1.25

0.18

61

0.89

0.13

0.70

0.14

0.84

0.15

70-79

60

1.21

0.17

58

0.85

0.13

0.65

0.14

0.83

0.14

80-89

9

1.24

0.22

9

0.76

0.17

0.58

0.19

0.79

0.26

< 0.001) was evident for the Ward‘s triangle area. Trochanter BMD, like spine BMD, was poorly associated with age and showed only a small aging decrease. There was no significant age change in the trochanter for the two older groups, whereas decreases at the femoral neck and Ward’s triangle sites were evident in all three age groups (Table 5). Females showed significant bone decreases in femoral neck BMD in all three age groups. There was no significant sex difference of femur regression slopes in subjects over 40 years, but there was in the groups as a whole (20-89 years). Males had higher rates of decrease at both femoral sites from age 20 to 39 years, but the slope from age 20 to 89 years was slightly but significantly lower in males than in females for both the femoral neck (0.5 versus 0.7%) and the Ward’s triangle (0.7 versus 0.9%). The stability of L2-4 BMD with age in males was manifest in the virtual constancy of both BMC and area (67 g and 51 cm’, respectively). Osteophytes, a relatively common problem in older males, and other anomalies were ex-

cluded radiographically from over half the male sample over 70 years of age. Hence, the observed stability is probably not unduly influenced by artificial elevation of spine BMC. The area of the 1.5 cm path across the femoral neck was also invariant with age (5.5 cm’). However, the BMC declined from 6 g in young adults to about 4.5 g in the oldest decade. There was a low but significant correlation of spine BMD and body weight, much as has been seen in studies in women (Table 4). The correlations of weight and femoral densities were lower (Table 4). The regressions indicate that each 10 kg increase in body weight is associated with about a 4% increase in spine BMD and a 3% increase in femoral BMD.

DISCUSSION Our study in white males is not the first documentation of the aging pattern for this group at the two common

MAZESS ET AL.

648

aI.fls)has also observed almost identical results in a study of 182 males in New Zealand. Earlier results from the Mayo C l i n i ~ ‘ showed ~ ~ ’ ~ ~that there was only slight spinal loss and intermediate femoral loss in small samples of men (n = 82 and 52 for spine and femur, respectively). The apparent bone diminution with age in the Mayo Clinic men was greater than that we observed (0.007 versus 0.004 g/ cm’ per year for the femoral neck and 0.002 versus 0.001 g/cm’ per year for the spine). However, all studies, includ14/-------1 ing this one, are cross-sectional and the apparent rates of loss may be influenced by cohort effects. It is apparent that men differ from women in showing little aging diminution of spinal or trochanteric BMD. The E apparent loss of 1 and 2% per decade for these two sites \ M differs significantly (p < 0.001) from that observed in a n B large series of women‘”’ in whom the average rate of postm menopausal loss was 10 and 4% per decade, respectively. In contrast, in males the BMD of femoral neck and Ward’s CI TIIII~IIANTI~II triangle declined from young adulthood onward, with a ,J 0 4 7 marked rate of loss (and high correlation). Males and fe20 40 GO 80 males in young adulthood have similar BMD values for the 4Cr ( w a r \ ) spine and proximal femur, although BMC and area values FIG. 1. The mean BMD values are plotted at 5 year in- are 20% lower in females. In both sexes there were no tervals. Linear regressions are plotted for the three femoral aging changes in the area of the spine or the femur, so the sites (neck, Ward‘s triangle, and trochanter) and a second BMD differences observed with age almost entirely reflect degree of polynomial for the lumbar spine (spine) based on changes in BMC. By age 70 the average spinal BMD in the 5 year averages. The linear regressions of the un- “normal” women is reduced by 20 versus 3% in males, and grouped data are given in Tables 4 and 5. BMD in the femoral neck is reduced by 25-30% versus

fracture sites for osteoporosis, but it is the first large-scale study using DPA. Our observed BMD values are almost identical to those from a study of men (500 spine and 200 femur cases) in the Baltimore area (Chris Plato, National Institute of Aging, personal communication). Gilchrist et

-

p.l

-

I

1

7

T

7---

,

OF BMD OF THE SPINE( n = 315) AND THREEFEMORAL SITES( n = 282) TABLE4. REGRESSIONS ON AGE, BODYWEIGHT,AND BMIa

Site Spine Neck Ward‘s Trochanter Spine Neck Ward’s Trochanter Spine Neck Ward’s Trochanter Spine Neck Ward’s Trochanter Spine Neck Ward’s Trochanter

- 0.0009 x age - 0.0046 x age = 1.14 - 0.0068 x age

=

1.29

= 1.18 = 0.96

- 0.0019 x age

= 0.88

+ 0.0047 x + 0.0027 x + 0.0021 x

= 0.74 = 0.64

0.66 = 0.88 = 0.82 = 0.75 =

=

0.66

+ 0.0027 x

+ 0.0150 x + 0.0053 x + 0.0020 x

+ 0.0082 x

- 0.0011 x 0.94 - 0.0047 x 0.94 - 0.0069 x 0.74 - 0.0020 x 0.90 - 0.0014 x

= 0.92 = = =

=

= 0.92 =

- 0.0049 x

0.90 - 0.0071 x - 0.0022 x

= 0.71

weight weight weight weight BMI BMI BMI BMI age + 0.0049 age + 0.0032 age + 0.0026 age + 0.0028 age + 0.0171 age + 0.0114 age + 0.0105 age + 0.0109

x weight x weight x weight x weight x BMI x BMI x BMI x BMI

SEE

r

0.160 0.133 0.144 0.139 0.153 0.160

-0.11 - 0.58

0.198 0.141 0.155 0.162 0.199 0.142 0.151 0.129 0.141 0.135 0.153 0.129 0.141 0.135

0.11 0.20 0.27 0.09 0.03 0.17 0.35 0.62 0.71 0.35 0.32 0.61 0.71 0.35

- 0.69

- 0.27

0.32 0.18

aThe SEE is given in g/cm’. The correlation coefficient and its significance level are shown.

P NS