Effects of Ethnicity and Age or Menopause on the Remodeling and Turnover of Iliac Bone: Implications for Mechanisms of Bone Loss



We measured histologic indices of bone remodeling and turnover separately on the cancellous, endocortical, and intracortical subdivisions of the endosteal envelope, and on the combined total surface, in transiliac bone biopsies obtained after double tetracycline labeling in 142 healthy women, aged 20–74 years, 34 black and 108 white, 61 premenopausal and 81 postmenopausal. The data were analyzed by two-way analysis of variance of the four groups defined by age/menopause and ethnicity and by linear regression of the major variables on age. None of the interaction terms was significant and none of the regression slopes on age differed between blacks and whites, indicating that, as for the previously reported structural indices, the effects of ethnicity and of age/menopause are independent. Accordingly, the data were also analyzed separately for the effect of ethnicity (pre- and postmenopausal combined) and age/menopause (blacks and whites combined). The analyses led to the following conclusions. (1) The geometric mean bone formation rate on the combined total surface was 25% lower in blacks than in whites; other histologic differences between ethnic groups were inconsistent between surfaces. (2) Serum osteocalcin (OC) but not bone-specific alkaline phosphatase (BSAP) was lower by about 15% in blacks than in whites. (3) The lower bone turnover in blacks is most likely in the directed rather than in the stochastic component because of a higher bone mass and consequent reduced susceptibility to fatigue damage. (4) All Class 1 bone formation variables and the three resorption indices were significantly higher in the postmenopausal compared with the premenopausal subjects, reflecting a 33% increase in activation frequency. (5) BSAP, but not OC, was increased relatively more (66%) than the bone formation rate (BFR). Consequently, BSAP is more sensitive to the effects of menopause than OC, but OC is more sensitive to the effects of ethnicity than BSAP. (6) There were highly significant differences between the three subdivisions of the endosteal envelope for every non–cell-related variable. All Class 1 formation variables were highest on the endocortical surface, but the magnitude and pattern of the differences otherwise was inconsistent between variables. The contributions of the different subdivisions to the total bone formation rate were cancellous 54%, endocortical 13%, and intracortical 33%. (7) The previously reported changes in bone surface location, together with the presently reported changes in activation frequency and wall thickness indicated that there was no significant effect of age/menopause on erosion depth on the cancellous and intracortical surfaces but a large increase in erosion depth on the endocortical surface. (8) The increase in bone turnover that results from hormonal changes is most likely in the stochastic rather than in the directed component because it serves no purpose but has harmful effects on skeletal integrity.


It is well known that menopausal estrogen deficiency is followed by a generalized increase in bone turnover, as well as accelerated bone loss. The increase in turnover has been demonstrated prospectively by radiokinetic,(1) biochemical,(2) and histologic(3) indices, supported by cross-sectional studies using each of the three types of methods.(4-7) Only white subjects were included in these studies; the effect of menopause on bone turnover in black women has not been examined prospectively, although similar cross-sectional increases in biochemical indices with age were noted in blacks and whites.(8) Biochemical indices of bone turnover are generally lower in black than in white subjects,(9-11) although there are inconsistencies in the data.(8) When studied histologically without regard to the effects of menopause, bone turnover in black women has been reported to be lower than,(12) the same as,(13) or higher than(14) in white women.

We recently reported that the structural differences in iliac bone between healthy pre- and postmenopausal women and the regression slopes on age for several structural indices were essentially identical in black and white women, and that structural differences between the groups were probably present at skeletal maturity and reflected differences in bone acquisition during growth.(15) We now report our data on histologic indices of bone remodeling and turnover in the same subjects. Previous reports from our laboratory of bone remodeling in healthy women(5-7) did not include all subdivisions of the endosteal envelope and included only a smaller number of white subjects. Women classified according to menopausal status necessarily differ also in age. These factors cannot be distinguished in a cross-sectional study, but increases in indices of remodeling are most likely due to menopause and decreases most likely due age.(5,7) We provide the first comprehensive description of the effects of ethnicity and age/menopause on the remodeling and turnover of iliac bone in healthy women, including all three subdivisions of the endosteal envelope, and discuss the significance of the results, not only for clinical medicine but also for human biology.


One hundred forty-two normal women were recruited in one of two ways, as previously described.(15) The subjects were the same as in the previous paper,(15) except for the exclusion of two subjects with recent onset of an endocrine disorder. They included 95 institutional employees and 47 members of Health Alliance Plan, the institutional health maintenance organization (HMO). There were no differences in ethnic proportions, height, or weight among subjects recruited in different ways. In total, there were 34 black and 108 white subjects, a proportion that reflected the ethnic composition of the sampled populations. Sixty-one subjects were premenopausal and 81 were postmenopausal; this classification was based only on self-reported menstrual status. All subjects were skeletally healthy according to previously defined criteria.(11,16) Selected calciotropic hormones and biochemical indices of bone turnover were measured in some subjects, according to methods previously cited or described.(11) Each subject underwent in vivo double tetracycline labeling with an interlabel time of 14 days; the schedule was oxytetracycline, 250 mg every 8 h for 3 days, followed by an 11-day interval, then demethylchlortetracycline, 150 mg every 8 h for 3 days, finishing 4 days before the biopsy.(17) To keep the latter interval constant, the biopsy was invariably scheduled before the dates of label administration were determined.

Cylindrical transiliac biopsies were obtained using a trephine with an internal diameter of 7.5 mm,(18) placed immediately in 70% ethanol, stained en bloc by the Villanueva method,(19) and embedded, sectioned, stained, and mounted as previously described.(20) All measurements based on tetracycline labels were made on sections 5 μm thick taken from the stained block without further treatment; the methods of fluorescence microscopy have been described previously in detail.(17) The length of the first (oxytetracycline) label is systematically shorter than the length of the second (demethylchlortetracycline) label,(17) and was multiplied by 1.18 before calculation of the length of the mineralizing surface (MS) as the mean of the two labels.(21) The surface-based bone formation rate (BFR/BS) was calculated as MS/BS∗ mineral apposition rate (MAR), with the latter corrected for section obliquity by multiplying by π/4.(21) If only one label was present, the minimum value for MAR of 0.236 μm/day (0.3 × π/4) was used.(22) This occurred three times on the cancellous, six times on the intracortical, and seven times on the endocortical surface. If neither label was present, MAR was treated as a missing value, and MS/BS and BFR/BS were reported as zero.(22) This occurred once each on the cancellous and intracortical surfaces and 10 times on the endocortical surface, which has much shorter profiles than the other two surfaces. At least one label was present on at least one surface in every case.

The lengths of osteoid surface (OS/BS) and osteoblast surface (Ob.S/BS) as previously defined(7) were measured in sections stained by a modified toluidine blue method.(23) These variables, together with MS/BS and BFR/BS, constitute class 1 bone formation indices, which depend mainly on activation frequency (Ac.f),(7) calculated as (BFR/BS)/wall thickness (W.Th).(21) MAR is a class 2 bone formation index, which depends mainly on osteoblast team performance, but is included here because it is a component of the calculation of BFR. Wall thickness is also a class 2 bone formation index, but is included here because it is a component of the calculation of Ac.f. Also included were three indices of bone resorption, eroded surface as a fraction of bone surface (ES/BS), and osteoclast surface as a fraction of bone surface (Oc.S/BS) and of mineralized (nonosteoid) surface (Oc.S/Md.S). Because the values are much smaller, indices of bone resorption have greater sampling variation and lower precision than indices of bone formation.(19,21) All these variables were measured separately on the cancellous (Cn), intracortical (Ct), and endocortical (Ec) subdivisions of the endosteal envelope, demarcated as previously described.(16) BFR was also expressed as the percent per year in relation to bone volume (BV), tissue volume (TV), and core volume (CV) referents, using the appropriate surface-to-volume ratios previously reported.(15) Mean bone age (years) was calculated as the reciprocal of BFR/BV. Since the core volume referent is the same for each surface subdivision, the individual values for BFR/CV can be summed to obtain the total bone formation rate for the entire biopsy core.(21) This is the most representative expression available from a biopsy of the bone formation rate for the whole iliac bone and hence for the entire axial skeleton.(24)

The data were analyzed by two-way analysis of variance (ANOVA) of the four groups classified according to age/menopause and ethnicity, and of the six data sets defined by age/menopause and surface, and by ethnicity and surface. Differences between means for pooled data were tested by one-way ANOVA and/or Student's t-tests. Many variables are reported, but they are functionally related to a small number of physiologic mechanisms, as previously indicated. Consequently, p values are given without a Bonferroni correction. Many histomorphometric variables do not conform to a Gaussian distribution, but to allow comparison with previous reports the data are presented as mean (standard deviation). For some variables, geometric means and multiplicative standard deviations were calculated. In addition, for some variables, the upper 95% confidence limit was calculated using the appropriate t-value for the premenopausal groups, and the numbers of individual values outside these limits in the postmenopausal groups were determined. Regressions on age were calculated, and, where appropriate, differences between slopes and adjusted mean values were tested by analysis of covariance (ANCOVA). The calculations were performed using the Sigma Stat Software package (Jandel Scientific) in accordance with accepted principles.(25,26)


After adjusting for the difference in age, none of the results differed between the two subject sources, so that the data were pooled. Mean values in the four demographic subgroups are given in Tables 1-5. In the premenopausal subjects, the only significant difference between blacks and whites was that osteoclast indices were higher in blacks on the cancellous surface (Table 1) and on the combined total surface (Table 4). Osteoid and osteoblast surfaces were slightly higher, and mineralizing surface and surface-based bone formation rate slightly lower in blacks, but these differences were inconsistent between surfaces and were not statistically significant. Bone age was higher in blacks, but this is a consequence of the lower surface-to-volume ratios(15) and not of differences in surface remodeling. In the postmenopausal subjects, similar trends were evident, but most of the differences were not significant. However, wall thickness (Table 5) was significantly lower in blacks on the cancellous and combined total surface.

Table Table 1. Resorption and Formation (Class I) Variables on Cancellous Surface in the Four Groups
original image
Table Table 2. Resorption and Formation (Class 1) Variables on the Endocortical Surface in the Four Groups
original image
Table Table 3. Resorption and Formation (Class 1) Variables on Intracortical Surfaces in the Four Groups
original image
Table Table 4. Resorption and Formation (Class 1) Variables in Combined Total Surface in the Four Groups
original image
Table Table 5. Class 2 Formation Variables on Each Surface in the Four Groups
original image

The most consistent and significant effects of age/menopause were increases in the osteoid surface and osteoblast surface of about 50%, evident on all three subdivisions of the endosteal envelope (Tables 1-4) and of similar magnitude in blacks and whites. There was about a 30% increase in the mineralizing surface of similar magnitude in blacks and whites, but this was inconsistent between surfaces and significant only on the endocortical and combined total surfaces. Because of a slight, although not significant, age-related decline in the mineral apposition rate (Table 5), bone formation rates were increased only by about 25%, and the difference in mean values was not, or only marginally, significant. However, a much higher than expected proportion of postmenopausal subjects had values above the upper 95% confidence limit based on the premenopausal data (Fig. 1). Because of the decline in wall thickness (Table 5), the activation frequency was increased on the cancellous and combined total surfaces. Logarithmic transformation improved the significance value only for bone age. None of the interaction terms in the two-way ANOVAs was significant for any variable on any surface.

Figure FIG. 1.

Values for indicated variables in individual postmenopausal subjects (W, white; B, black) compared with premenopausal confidence limits, calculated as ± t (SD). Expected numbers above 97.5 percentile are W 1.5, B 0.4, and above 99.5 percentile are W 0.3, B 0.1.

Linear regressions for selected variables on age for the combined total surface are shown in Table 6. As expected from the two-way ANOVAs, there were highly significant increases with age, but the correlations were relatively weak and no r value exceeded 0.4. The correlations were generally weaker for the combined group than for the white subjects alone. None of the regressions on age was significant in black subjects alone, but this was a consequence of smaller sample size rather than a biological difference between the groups, because comparison of slopes by ANCOVA showed no significant differences between them.

Table Table 6. Regressions of Selected Variables on Age (Combined Total Surface)
original image

The three subdivisions of the endosteal envelope were compared first by two-way ANOVAs using age/menopause or ethnicity as second variables of classification. None of the interaction terms was significant, so that the data for all subjects were pooled, and the surface subdivisions are compared by one-way ANOVA in Table 7. There were highly significant differences between subdivisions for every non–cell-related variable. For all variables that tended to increase with age, the values were higher on the endocortical surface. For MAR, which tended to fall with age, the value was lower on the endocortical surface, and for W.Th, which also tended to fall with age, the value was lower on the cancellous surface. For osteoid and eroded surface, the values were higher on the cancellous than on the intracortical surface, but for the tetracycline-based indices, the values were higher on the intracortical than on the cancellous surface. For each subdivision, the values were substantially higher than found on the periosteal surface in a subset of the present series.(6) The contributions of the three subdivisions of the endosteal envelope to the total core formation rate are shown in Table 8. The cancellous surface contributed 54% and total cortical (intracortical + endocortical) 46%. The surfaces adjacent to the bone marrow (cancellous + endocortical) contributed 67%, and the intracortical surface not in contact with the bone marrow contributed 33%.

Table Table 7. Comparison of Different Subdivisions of the Endosteal Envelope in the Entire Group
original image
Table Table 8. Contributions of Different Subdivisions of Endosteal Envelope to Bone Formation
original image

Selected biochemical data are shown in Table 9. Serum osteocalcin was significantly lower in blacks, but bone-specific alkaline phosphatase (BSAP) was not different. Serum 25-hydroxy D was more than 30% lower in blacks, and 1,25-dihydroxy D and intact parathyroid hormone (PTH) were higher; the latter differences were not significant, but collectively the results were similar to those previously reported.(9) Serum osteocalcin was also higher in postmenopausal than in premenopausal subjects, but BSAP showed a greater proportional increase that was more highly significant. Serum 25-hydroxy D was slightly higher, but this difference was a consequence of the higher proportion of white subjects in the postmenopausal than in the premenopausal groups (81 vs. 69%). The significantly higher postmenopausal value for serum 1,25-dihydroxy D could not be accounted for by the difference in ethnic proportions. The difference has been previously noted and appears to be due both to increased production and to decreased metabolic clearance of calcitriol.(27) Intact PTH levels showed no significant effects of ethnicity or age/menopause.

Table Table 9. Biochemical Results
original image


Subject to the limitation of sample size, as discussed previously,(15) our data permit some general conclusions concerning influences of ethnicity on bone remodeling. We carried out 52 two-way ANOVAs of the histologic data, and not a single interaction term was significant. This means that, as for the structural and architectural indices previously reported,(15) the effects of age/menopause are the same in blacks as in whites, and the differences between blacks and whites are the same in younger premenopausal women as in older postmenopausal women. Consequently, it seems reasonable to compare the two ethnic groups without regard to age or menopausal status; for both practical and statistical reasons, the comparison is restricted to the combined total surface (Table 10). Only for bone formation rates were there significant differences. The activation frequency was not different, but BFR/BS was significantly lower in blacks after but not before logarithmic transformation by about 25%. Because of the lower bone surface–to-bone volume ratio in blacks, the relative difference was somewhat greater for BFR/BV, with a corresponding increase in mean bone age. BFR/CV, which is unaffected by surface-to-volume ratio but depends on the total area of surface in the biopsy, was about 20% lower in blacks. This is somewhat greater than the differences reported in biochemical indices of bone turnover, which are in the range 5–20%.(10,11) We found a significant difference of 15% for serum osteocalcin and a nonsignificant difference of 4% for BSAP, a discrepancy that we are unable to explain. The histomorphometric differences between ethnic groups were smaller than previously reported for the cancellous surface alone,(12) because we found lower values in white subjects and similar values for black subjects, but the earlier study had fewer subjects recruited from different sources and included men as well as women.

Table Table 10. Comparisons of All White and Black, and All Pre- and Postmenopausal Subjects
original image

Since the effects of age/menopause did not differ between blacks and whites, the ethnic groups were pooled to permit a more powerful analysis, which, as for the black/white comparison, is restricted to the combined total surface (Table 10). All class 1 bone formation variables and the three resorption indices were significantly higher in the postmenopausal compared with the premenopausal subjects, reflecting a 33% increase in activation frequency. The increase in BFR/BS was slightly smaller, because of the decline in wall thickness, and the increase in BFR/BV was slightly greater, because of the decline in BS/BV, especially in cortical bone, with a corresponding significant reduction in mean bone age. However, the difference in BFR/CV, which reflects the age-related loss of bone surface in the entire core,(15) did not quite reach statistical significance. The proportional increase was about the same for osteocalcin as the increase in BFR, but was much greater for BSAP than the increase in BFR. Why BSAP should be less sensitive than osteocalcin to ethnic differences but more sensitive than osteocalcin to the effects of menopause is unclear, although other discrepancies between these usually concordant indices have been reported.(28) We previously found that bone formation on the intracortical surface contributed more to the serum total alkaline phosphatase than bone formation on the other surface subdivisions,(24) but in the present study, the intracortical surface did not show any greater effect of age/menopause than the other surfaces.

In earlier cross-sectional studies,(4,29) somewhat larger differences between pre- and postmenopausal subjects were reported than we observed. But a detailed description of changes throughout the postmenopausal period is still lacking. The most popular view is that the effects of estrogen deficiency on bone turnover continue throughout life(4); the bone formation rate measured histologically increases progressively with age in women but not in men,(30) and estrogen replacement may significantly reduce bone turnover even after age 80.(31) However, the rate of bone loss attributable to estrogen deficiency appears to slow down progressively, with an exponential approach to an asymptomatic value.(32,33) This is partly due to early expansion of the remodeling space toward a new steady-state value(34) and partly to the previously reported loss of cancellous surface.(15) But the decline in rate of loss applies also to cortical bone in the forearm,(32) in which the remodeling space is much smaller(35) and in which the bone surface is more likely to increase rather than decrease with age.(15)

The conjunction of persistent increases in bone turnover and declining rates of bone loss suggests that the effects of estrogen deficiency on bone turnover and on bone loss are dissociated. A possible explanation could be that increased turnover is the result of cytokine-mediated increases in cell recruitment(36) that are nonspecific but sustained, whereas increased bone loss is the result also of a delay in osteoclast apoptosis(37) that is specific to estrogen deficiency, but unsustained. An alternative view is that the direct effects of estrogen deficiency to accelerate bone loss and increase bone turnover both subside within 10 years of menopause(38); beyond that, biochemical indices remain stable or even fall toward premenopausal levels.(39,40) Increased bone turnover after age 65 is more likely the result of other age-related changes, such as secondary hyperparathyroidism,(38) due to a decline in renal function or impaired vitamin D nutrition.(41) Secondary hyperparathyroidism may also be an indirect effect of estrogen deficiency, mediated by a decreased tubular reabsorption of calcium.(42,43) If PTH increases cell recruitment via the same cytokines as estrogen deficiency,(44) estrogen replacement could correct high bone turnover regardless of its mechanism.

We found a slight nonsignificant decline in the mineral apposition rate, and a small (6%) but significant decline in wall thickness. Although this decline was significant only on the cancellous and not on the other two surfaces, the proportional decline did not differ significantly between the three subdivisions of the endosteal envelope. Our previous report that wall thickness declines only on the cancellous surface(45) was based on a smaller sample size. The contribution of reduced wall thickness to bone loss can now be examined in more detail. In terms of standard remodeling theory, focal imbalance leading to bone loss from a surface is the result of some combination of decreased W.Th and increased erosion depth (E.De). The latter was not measured in the present study, but it can be estimated indirectly(46) from the relationship:

equation image

As explained elsewhere,(45) the first term in this equation can be estimated from previously reported structural data,(15) and the second and third terms are included in the present report, so that the fourth term can be calculated.

The results of the calculation for each surface subdivision are given in Table 11. There was no significant change in erosion depth on the cancellous surface, indicating that continued thinning of residual trabeculae is due entirely to reduced wall thickness.(7,46) Since these calculations can be applied only to the trabeculae that remain and not to those that have been removed, this is entirely consistent with increased erosion depth leading to perforation and loss of trabeculae at some earlier time.(33) There was also no significant change in erosion depth on the intracortical surface. Although the increase in canal radius was not significant in the earlier study,(15) a modest increase in canal radius due to a modest decline in wall thickness has previously been reported.(47) However, on the endocortical surface, the situation was entirely different. As previously estimated for cortical bone loss in the metacarpal,(48) the major cellular mechanism leading to thinning of cortical bone in the ilium was a substantial increase in erosion depth. The data indicate that the early response to menopause is an increase in erosion depth on the cancellous, as well as on the endocortical surface, but the former abnormality subsides after a few years, whereas the latter continues, possibly throughout life.(49) By contrast, there is no increase in erosion depth at any time on the intracortical surface. Clearly, the mechanism of cortical bone loss cannot be inferred from histologic measurements restricted to cancellous bone.

Table Table 11. Indirect Calculation of Change in Erosion Depth with Age or Menopause
original image

Are the cellular mechanisms of bone loss from the three surface subdivisions related to the differences in indices of bone remodeling (Table 7)? For unknown reasons, BFR/BS was higher on the endocortical than on the cancellous bone surface, as reported by others,(50) but did not differ from the intracortical surface, from which bone loss is trivial. Although differences in the mean values for osteoclast surface extent were not significant, in a few subjects the postmenopausal values on the endocortical surface were much higher than any values on the other bone surfaces (Fig. 1). If erosion depth is determined by the timing of osteoclast apoptosis,(37) then delayed osteoclast apoptosis due to estrogen deficiency must persist for much longer on the endocortical surface than on the cancellous surface. If the osteoclast life span is increased because of delayed apoptosis, the surface extent and number of osteoclasts would be increased out of proportion to osteoclast birth rate. Such an abnormality could be present in all postmenopausal women but detectable only in those with extreme values because of the imprecision in osteoclast measurements. A persistent delay in osteoclast apoptosis on the endocortical surface would account for both the greater bone loss and the inferred increase in erosion depth (Table 11). Since PTH excess also increases endocortical erosion depth to a greater extent than on the cancellous surface,(51) a surface-specific increase in sensitivity to PTH(52) could underlie the effects of estrogen deficiency.

There were large between-subject differences in bone turnover (Fig. 1); the coefficient of variation (CV = SD/mean × 100) for activation frequency was about 55% in both pre- and postmenopausal subjects (Table 10). The CV for biochemical indices of turnover is usually in the range 25–45%(53,54); this is higher than the within-subject CV,(55) indicating true individual differences.(40) The higher CV for histologic than for biochemical indices is the result of regional differences within the skeleton(56) and site-to-site variation within the same bone.(57) The main purpose of bone remodeling is to maintain the mechanical competence of bone as a structural material by preventing excessive increases in bone age and so preventing fatigue microdamage accumulation.(56) In cancellous bone adjacent to hematopoietic marrow, and in the corresponding cortical bone, the rate of remodeling is higher than is needed to maintain mechanical competence, probably because of additional functions related to calcium homeostasis and support of hematopoiesis. Remodeling is directed to specific targets, which are regions of bone in need of replacement, but these also include a stochastic component that provides a margin of safety, by continuing after the targeted bone has been replaced(56); whether individual differences are mainly in the directed or the stochastic component is unknown. The lower values in blacks than in whites are probably in the directed component, because higher values for bone mass would be expected to reduce the susceptibility to fatigue microdamage.(58) Although, in general, slower bone turnover is associated with slower bone loss,(56) cross-sectionally determined rates of loss do not differ between blacks and whites.(15) The increases in bone remodeling that result from hormonal changes are most likely in the stochastic component,(59) because they serve no purpose, but accelerate bone loss and increase the likelihood of mechanical failure in vertical trabeculae.(60)


We thank Michael Kleerekoper for making it possible to recruit subjects from HAP, Paulette Wilson and Nayana Parikh for the biochemical measurements, Paula Dillon for preparation of histologic sections, and Constance Mott for preparation of the manuscript. This work was supported by National Institutes of Health grants numbers AG10381 and AG/AR13918.