Effects of Ethnicity and Age or Menopause on Osteoblast Function, Bone Mineralization, and Osteoid Accumulation in Iliac Bone



We measured histologic indices of osteoblast function, bone mineralization, and osteoid accumulation separately on the cancellous, endocortical, and intracortical subdivisions of the endosteal envelope and on the combined total surface in transiliac biopsies obtained after double tetracycline labeling in 142 healthy women, aged 20–74 years, 34 who were black (19 pre- and 15 postmenopausal) and 108 white (42 pre- and 66 postmenopausal). The data were subjected to two-way analysis of variance of the four groups defined by age/menopause and ethnicity. Also, linear regressions of selected variables on age and between functionally related but independently measured variables were examined. 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 and remodeling indices, the effects of ethnicity and of age/menopause are independent. Accordingly, the data were analyzed separately for the effects of ethnicity (pre- and postmenopausal combined) and age/menopause (blacks and whites combined). The analyses led to the following conclusions (1) Osteoid surface and volume were higher and adjusted apposition rate and osteoid mineralization rate lower in postmenopausal than in premenopausal subjects, but none of the indices of osteoid accumulation differed between blacks and whites. (2) Each index of osteoid accumulation was significantly correlated with its primary independently measured kinetic determinant (osteoid thickness with adjusted apposition rate, osteoid surface/bone surface with activation frequency, and osteoid volume/bone volume with bone formation rate/bone volume). None of the regression parameters differed significantly between blacks and whites. (3) The ratio of mineralizing surface to osteoid surface (MS/OS) was substantially lower in all demographic groups than could be accounted for by the later onset of mineralization than of matrix apposition at individual bone forming sites. (4) The low values for MS/OS can be explained by terminal mineralization being too slow to trap enough tetracycline molecules to produce detectable fluorescence, and do not require that mineralization be interrupted. (5) MS/OS was about 25% lower in blacks than in whites on all surfaces with corresponding differences in derived indices based on MS/OS, including adjusted apposition rate, mineralization lag time, and formation period. (6) The lower values for MS/OS in blacks are most likely due to slower terminal mineralization. This could not be accounted for by a lower serum level of calcidiol, but is consistent with the reported effect of reduced bone blood flow. (7) All differences in bone cell function between blacks and whites that we have observed could be the result of the ethnic, and presumably genetic, difference in bone accumulation during growth. Higher bone mass would result in less fatigue microdamage, less need for repair by directed bone remodeling, lower bone turnover, lower bone blood flow, and slower terminal mineralization. (8) If this explanation is correct, there are no fundamental differences in the biology of bone remodeling between ethnic groups.


DURING NORMAL BONE REMODELING, the replacement of resorbed bone by new bone is carried out by successive teams of osteoblasts, each team comprising cells that arrive at the same location at the same time to carry out a common purpose.1 The total rate of bone formation is the product of two independent quantities. First is the number of new teams recruited per unit time, of which the best estimate is the activation frequency of bone remodeling.2 Histomorphometric variables of bone formation that reflect activation frequency, which generally increase with age or menopause, have been designated as class 1 formation indices.3 Second is the average volume of bone matrix made by each team, of which the best estimate is mean wall thickness.3 This reflects the number of osteoblasts that assemble on the cement line at the beginning of each episode of bone formation, and the average volume of bone matrix made by each osteoblast during its active life span, or capacity.1 Histomorphometric variables of bone formation that reflect the function of individual teams of osteoblasts, which generally decline with age or menopause, have been designated as class 2 formation indices.3

After bone matrix is deposited by osteoblasts, cross-linking of its collagen fibers and other changes occur that prepare the matrix for mineralization.4 Because of the need for these changes, collectively referred to as maturation, there is both spatial and temporal separation between matrix and mineral apposition that is manifested by the existence of osteoid tissue. The thickness, surface extent, and volume of osteoid are each regulated by different aspects of osteoblast function1; all three osteoid indices are increased in osteomalacia but for different reasons.4 There are three surface-based histologic indices of bone formation which tend to vary together, but differ in magnitude. The surface extent of osteoid is invariably greater than the surface extent of mineralization. The reason is still incompletely understood, but the relative difference between them, expressed as mineralizing surface/osteoid surface (MS/OS), is relevant to both osteoblast function and mineralization.1 The surface extent of mineralization is invariably greater than the surface extent of osteoblasts, because of changes in osteoblast morphology during the life history of the osteoid seam.1

We have previously reported independent effects of ethnicity, and of age or menopause, on a variety of histologic measurements in iliac biopsies. In both black and white subjects, bone structure deteriorates5 and bone turnover increases6 with age or menopause. In both pre- and postmenopausal subjects, blacks have substantially thicker cortices and trabeculae5 and modestly lower bone turnover6 than white subjects. We now report our data on osteoblast function, bone mineralization, and osteoid accumulation in the same subjects, including all three subdivisions of the endosteal envelope, and examine the relationships between the various histologic indices. We discuss the significance of our results, not only for clinical medicine but for human biology.


One hundred forty-two normal women were recruited in one of two ways, as previously described.5 The subjects were the same as in a previous paper6 and included 34 blacks and 108 whites; based on self-reported menstrual status, 61 subjects were premenopausal and 81 were postmenopausal (Table 1). They were not selected on the basis of factors likely to affect socioeconomic status. Most were members of the same health maintenance organization and represented most of the communities around Detroit.7 Recruitment procedures were the same for blacks and whites. All subjects were skeletally healthy according to previously defined criteria.8,9 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 demethyl-chlortetracycline 150 mg every 8 h for 3 days, finishing 4 days before the biopsy.10 To maintain the latter interval constant, the biopsy was invariably scheduled before the dates of label administration were determined.

original image

Cylindrical transiliac biopsies were obtained using a trephine with internal diameter 7.5 mm,11 placed immediately in 70% ethanol, stained en bloc by the Villanueva method,12 and embedded, sectioned, stained, and mounted as previously described.13 All measurements based on tetracycline labels were made on sections 5 μm thick, taken from the stained block without further treatment. The length of the first (oxytetracycline) label is systematically shorter than the length of the second (demethylchlortetracycline) label,10 and was multiplied by 1.18 before calculation of the length of mineralizing surface (MS) as the mean of the two labels.14 Missing labels were dealt with as previously described.15 At least one label was present on at least one surface in every case.

Osteoid indices were measured in sections stained by a modified toluidine blue method.16 Osteoid thickness (O.Th) was measured directly at multiple locations about 100 μm apart, including only values ≥ 2 μm.3 Wall thickness (W.Th) and label separation were measured in a similar manner, and all three were corrected for section obliquity using π/4.14 Osteoid volume/bone volume (OV/BV) was calculated indirectly using the osteoid thickness and surface measurements.12 Adjusted apposition rate (Aj.AR) was calculated as mineral apposition rate (MAR)∗MS/OS, osteoid maturation time (Omt) as O.Th/MAR, mineralization lag time (Mlt) as O.Th/Aj.AR, osteoid mineralization rate (OMR) as the reciprocal of Mlt, formation period (FP) as W.Th/Aj.AR, and osteoblast vigor (Ob.Vg) as the reciprocal of FP.15 OMR and Ob.Vg provide the same information as Mlt and FP, but the values have a lower limit of zero instead of an upper limit of infinity. All these variables were measured separately on the cancellous (Cn), intracortical (Ct), and endocortical (Ec) subdivisions of the endosteal envelope, demarcated as previously described,17 and reported also for the combined total (Tt) surface.

It follows from the method of calculating Mlt that the following relationship holds:

equation image(1)

It can also be demonstrated1,4,17 that the following relationships hold:

equation image(2)
equation image(3)

where BFR is the bone formation rate. In each of these equations, the first two terms are independently measured and do not contain any common factor, whereas the first and third terms are not independent but contain a common factor. Consequently, it is appropriate to plot the first term as a dependent variable against the second term as an independent variable and to calculate the regression parameters. When this is done, the third term represents the slope of a family of lines radiating from the origin.1,4,18

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 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. 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 (Jandell Scientific, Corte Madera, CA, U.S.A.) in accordance with accepted principles.19,20


After adjusting for the differences in age and ethnic proportions, there were small differences in MS/OS and O.Th between the two subject sources, but correcting for these differences did not alter any of the analyses, so that the primary data were pooled. Mean values in the four demographic subgroups are given in Tables 1, 2, 3, 4. Data for OS/BS, MAR, and W.Th have been previously reported6 but are included here for different reasons. In both pre- and postmenopausal subjects, MS/OS and indices influenced in the same direction by MS/OS, including Aj.AR, OMR, and Ob.Vg, were all lower in blacks than in whites on the cancellous and combined total surfaces. Some of these differences were also present on the intracortical and endocortical surfaces but were less consistent. The differences in MS/OS were the result of slightly lower values for MS/BS and slightly higher values for OS/BS in blacks, although these differences were not significant.6 The most striking effects of age/menopause were increases in osteoid volume in both cancellous and cortical bone, due entirely to the increases in osteoid surface, and a decline in Aj.AR and OMR that was significant only on the cancellous and combined total surfaces. There were no consistent effects of ethnicity or age/menopause on O.Th or Omt. In the two-way ANOVA, only 1 of 47 interaction terms was significant at p < 0.05, consistent with chance expectation. In the total group and in the white subjects alone, there was a significant decline of Aj.AR with age on the cancellous surface (p < 0.01) and on the combined total surface (p < 0.05).

original image
original image
original image

In Table 5 and Figs. 1, 2, 3 are shown the highly significant regressions of each osteoid index on its primary determinant (second terms in Eqs. 1–3) for the pooled data. For O.Th, the correlation with MAR was not significant on the cancellous surface and was weaker than the correlation with Aj.AR on the other surfaces (data not shown). None of the regressions on secondary determinants (third terms in Eqs. 1–3) was significant. The regressions on primary determinants were also significant in each demographic subgroup except that O.Th was not significantly related to Aj.AR in blacks on any surface. The regressions on secondary determinants were also not significant in the subgroups, except that in blacks O.Th was significantly related to Mlt on the cancellous surface (r = 0.415, p < 0.05), and in whites O.Th was inversely related to Mlt on the cancellous, intracortical, and combined total surfaces. However, there were no significant differences between blacks and whites for any regression parameter (slope, intercept, or adjusted mean) by ANCOVA. For the regression of O.Th on Aj.AR, there were no differences in regression parameters between pre- and postmenopausal subjects. But for the other two regressions, although there was no difference in slope, both the intercepts and adjusted means were higher in postmenopausal than in premenopausal subjects on each surface, although not on the combined total surface (Table 6).

Figure FIG. 1..

Relationship between osteoid thickness (O.Th, μm) and adjusted apposition rate (Aj.AR, μm/day) on cancellous surface in 61 normal premenopausal women (see Eq. 1). Solid line is linear regression line (y = 7.66 + 5.14x; r = 0.426, p < 0.001). Interrupted lines connect equal values for mineralization lag time (Mlt, days) as indicated.

Figure FIG. 2..

Relationship between osteoid surface/bone surface (OS/BS, %), and activation frequency (Ac.f, /year) on cancellous surface in 61 normal premenopausal women (see Eq. 2). Solid line is linear regression line (y = 5.07 + 27.3x; r = 0.656, p < 0.0001). Interrupted lines connect equal values for formation period (FP, years) as indicated.

Figure FIG. 3..

Relationship between osteoid volume/bone volume (OV/BV, %) and bone formation rate/bone volume (BFR/BV, %/year) on cancellous surface in 61 normal premenopausal women (see Eq. 3). Solid line is linear regression line (y = 0.384 + 0.106x; r = 0.763, p < 0.0001). Interrupted lines connect equal values for mineralization lag time (Mlt, years) as indicated.

original image
original image

There were no significant interactions between surface differences and either ethnicity or age/menopause, so that the data were pooled, and the three surfaces are compared in Table 7. There were significant differences between them for every variable except osteoid thickness. MS/OS was higher on the intracortical surface than on the two surfaces which are in contact with bone marrow, with corresponding differences in the derived indices. MS/OS was higher on the endocortical than on the cancellous surface, but the variables derived from MS/OS did not differ between these surfaces.

original image


Our study population had many fewer blacks than whites because of our method of recruitment.5,7 Subject to this limitation, the absence of significant interaction in the two-way ANOVAs means that, as for the structural and remodeling indices previously reported,5,6 the differences between blacks and whites are the same in younger premenopausal women as in older postmenopausal women, and the effects of age/menopause are the same in blacks as in whites. Consequently, it seems reasonable to compare the two ethnic groups without regard to age or menopausal status, and the two menopausal groups without regard to ethnic composition. For both practical and statistical reasons, the comparisons are restricted to the combined total surface (Table 8). Comparing black and white subjects, none of the osteoid indices differed significantly but MS/OS was about 25% lower in blacks, with corresponding differences in all of the indices derived from MS/OS. Comparing pre- and postmenopausal subjects, osteoid surface and volume were significantly higher, and Aj.AR significantly lower, in the postmenopausal group. The decline in Aj.AR resulted from modest declines in both MAR and MS/OS, although neither of these differences was significant. Also, there were no significant differences in any of the other indices derived from MS/OS.

original image

The significant relationship between each index of osteoid accumulation and its primary kinetic determinant validates the use and interpretation of tetracycline labeling and supports the theoretical analyses expressed in Eqs. 1–3.1,4,18 In Eq. 1, the tighter correlation of O.Th with Aj.AR than with MAR confirms the former as a more accurate index of the mean rate of matrix apposition.1 Equations 2 and 3 are analogous to the epidemiologic principle that the size of a population depends on both its birth rate and its mean life span.18 The lack of relationship between O.Th and Aj.AR in black subjects is probably a reflection of limited sample size rather than a true biologic difference, because the regression parameters did not differ significantly by ANCOVA between ethnic groups. The general lack of relationship between the osteoid indices and their secondary kinetic relationships, and the inconsistent relationships between O.Th and Mlt in subgroups, are consequences of the complex functional interdependence between them. For example, an increase in Mlt will, other things being equal, lead to an increase in O.Th. But in the absence of osteomalacia, an increase in Mlt is an inevitable consequence of a reduction in Aj.AR, which will decrease O.Th.1,18 In osteomalacia, the relationship between Aj.AR and O.Th is reversed, and Mlt becomes the primary determinant of O.Th.4

The disparity between the extent of labeled surface and the extent of osteoid surface in healthy subjects in this study was somewhat greater than in previous reports, in whom values for MS/OS ranged from approximately 40 to 60%, with somewhat lower values in patients with osteoporosis.21 The value for MS/OS will be influenced by the criteria for identifying osteoid; the minimum acceptable osteoid thickness (2.0 μm in our study) has usually not been stated, but the same criterion was used in a study that did not include normal subjects.21 At locations where mineralization begins or ends during the 14-day time interval between the midpoints of the two labeling periods, there can be only one label, a phenomenon referred to as label escape.22 But in a steady state, the mean extent of mineralizing surface at any time will be unaffected by this circumstance.18 At each bone-forming site, matrix apposition begins before mineralization. The extent of unlabeled surface due to this mechanism depends on the duration of the initial mineralization lag time; this is approximated by the osteoid maturation time,1,4,7 which was about 16 days in our study, but this could account for no more than a 10–20% disparity between the osteoid and mineralizing surfaces.1,21 Unlabeled osteoid has been attributed to one or more interruptions of mineralization during the life span of the osteoid seam, referred to as “off time”,23 but a careful search for this phenomenon in human cancellous bone remodeling using multiple labels found no evidence for it.21

The systematic difference in extent of labeled surface between different tetracyclines10 and the influence of tetracycline dose on the extent of fluorescence24 suggest that unlabeled osteoid could result from the retention of too few molecules of tetracycline to give rise to detectable fluorescence with the particular microscopic system in use.10,18,21,25 The trapping of tetracycline depends entirely on the deposition of a sufficient thickness of mineral to prevent the escape of tetracycline when it is no longer present in the circulation.18 During the life span of the osteoid seam, after its thickness reaches a maximum and begins to decline, the rates of matrix and mineral apposition progressively fall and the osteoblasts become progressively flatter.1,18,26 Lack of label occurs preferentially at locations where the osteoid is thin and distant from the cement line, and covered by flat cells that have probably stopped making bone matrix and have almost completed their morphologic transformation to lining cells.1,10,26 This suggests that during terminal mineralization, the layer of new mineral added each day may be too thin to prevent escape of tetracycline during the 4-day interval between the end of the second labeling period and the biopsy.10,25 The correlations we demonstrated indicate that the underestimation of labeled surface by this mechanism is systematic and not random, but as discussed later, it probably leads to modest underestimation of the BFR.10

The existence of interrupted mineralization, or off time, has been inferred from label escape theory.27,28 If, at each bone-forming site, mineralization proceeded to completion without interruption, then the relationship SL/DL = 2 MI/FP(Act) will hold, where SL/DL is the ratio of single- and double-labeled surface extents, MI is the marker interval (14 days in the present study), and FP(Act) is the active formation period, calculated as W.Th/MAR. In both normal subjects and in patients with osteoporosis, SL/DL is consistently higher than predicted by this equation,27,28 as it was in this study by 2- to 3-fold (data not shown). Since every on/off switch would increase the extent of SL due to label escape,29 the higher than predicted value for SL/DL is taken as evidence for off time, but failure to capture terminal mineralization will also increase SL/DL. Depending on the timing of the label sequence in relation to events at a particular surface, three different situations must be considered. If only the first label is lost, then a single label will be converted to no label; if only the second label is lost, then a double label will be converted to a single label. Finally, if both labels are lost, then a double label will be converted to no label. The net result of these three outcomes will be reduction in the extent of double label and no change in the extent of single label. Consequently, the observed increase in SL/DL can be explained by slow terminal mineralization, as well as by interrupted mineralization.

If excess unlabeled osteoid is due not to interruption of mineralization but to uncaptured terminal mineralization, the significance of several derived indices must be reexamined.21 The adjusted apposition rate is analogous to the radial closure rate in osteonal remodeling, calculated on the assumption that unlabeled osteons are in a temporary resting state.30 Both quantities are believed to be estimates of the rate of matrix apposition averaged over the life span of each osteoid seam, since in the absence of osteomalacia the total volumes of bone matrix and mineralized bone formed are the same.18,30 According to this reasoning, the mean life span of the seam (FP or sigma) including “off” time as well as “on” time, is given by W.Th/Aj.AR. But this calculation can be justified even in the absence of off time, since the rate of terminal mineral apposition will in most cases be less than 0.1 μm/day. This is sufficiently close to zero that Aj.AR and FP, calculated as at present, would usually not be in error by more than 15%. For example, if MS/OS = 50% and MAR = 0.6 μm/day, the true value for Aj.AR would be no more than 0.5(0.6 + 0.1) = 0.35 instead of 0.3 μm/day as presently calculated, and if W.Th = 42 μm, FP would be 120 instead of 140 days as presently calculated.

Mlt is also dependent on the calculation of Aj.AR, a procedure which is supported by the stronger correlation with O.Th than for MAR, as previously indicated. During the life span of a typical osteoid seam, the instantaneous Mlt increases progressively, and the true mean value lies between an initial value of about 10–15 days (approximated by Omt, calculated as O.Th/MAR) and a terminal value of about 50–60 days.1,18 Whether the increase in Mlt reflects a slowing of osteoid maturation or a decline in the supply of mineral is unknown,4 but in either case the mean value calculated as O.Th/Aj.AR will be a reasonable approximation to the true mean value.18 Thus, the conceptual basis for calculating Aj.AR, FP, and Mlt remains valid, even if the derived values are modestly in error.1 By making several reasonable assumptions, it is possible to calculate more accurate values from the same primary data, but it seems more sensible to maintain the current practice and to keep in mind that the penumbra of uncertainty that inevitably surrounds bone histomorphometric data is somewhat wider than was previously thought.

If slow terminal mineralization accounts for unexpectedly low values for MS/OS, it is likely that slower terminal mineralization in blacks than in whites is responsible for their lower values for MS/OS and the consequent differences in Aj.AR, FP and Mlt (Tables 1, 2, 3, 4, and 8). The difference in MS/OS was previously observed in a study confined to premenopausal women, although attributed to a difference in off time rather than in label retention.31 In postmenopausal women, the difference in MS/OS is similar but of somewhat smaller magnitude—25% versus 30% on the combined total surface (Table 4). Slower terminal mineralization does not account for the lower BFR in blacks that we recently reported.6 The extent to which BFR is underestimated depends on the thickness of osteoid at the time that mineral apposition falls below the critical rate needed to prevent tetracycline escape and is unaffected by further changes in the rate. For example, if osteoid thickness is 4 μm at that time point, which is a reasonable estimate,1,10,18 and wall thickness is 40 μm, the underestimation of bone formation will be 10%. There is no reason to believe that this relationship is influenced by ethnicity, since neither MAR, which reflects only mineralization before the terminal phase, nor W.Th, which represents the endpoint of mineralized bone formation at any location, differs significantly between blacks and whites6 (Table 8).

The slower terminal mineralization in blacks is unlikely to be due to the lower serum level of calcidiol6,32 for several reasons. First, even in the mildest cases of osteomalacia due to vitamin D deficiency, the onset of mineralization is delayed and its termination premature, so that indices of osteoid accumulation are increased and osteoid maturation time prolonged.4 None of these measurements differed between blacks and whites. Second, a positive rather than a negative correlation between plasma calcidiol and the extent of unlabeled osteoid was found in osteoporotic patients.25 We found a similar positive correlation in normal white subjects and a stronger negative correlation between serum calcidiol and MS/OS (Table 9). There were no significant correlations with serum calcidiol in black subjects, but the correlations remained significant when the groups were combined. It is unclear why an index of vitamin D nutrition should be associated with slower terminal mineralization, but evidently the difference between blacks and whites cannot be accounted for by this relationship.

original image

A more likely explanation for the difference in terminal mineralization is a difference in vascular perfusion. In patients with osteoporosis, there was a significant correlation between Aj.AR and total skeletal blood flow.33 Although MS/OS was not reported, most of the variability in Aj.AR is due to variability in MS/OS, rather than MAR. The data suggest that the rate of terminal mineralization is related to the rate of perfusion. From the mean values for Aj.AR in blacks and whites, which differed by 30% (Table 8), the predicted mean values for skeletal blood flow, expressed as percent of blood volume/minute33 were 4.55 and 5.17, a difference of only 12%. Bone blood flow correlates with bone turnover, both between subjects33 and between different sites in the same subject.2 If the lower turnover in black than in white subjects6 was accompanied by proportionately lower bone blood flow, it would completely account for slower terminal mineralization. We have argued elsewhere that the lower bone turnover in black subjects is a consequence of their higher bone mass6; whether differences in bone mass within ethnic groups are also accompanied by differences in bone turnover and in the rates of terminal mineralization remains to be determined. But if so, all the differences in bone cell function that we have demonstrated could be consequences of the genetic difference in bone accumulation during growth, and the biological basis of bone remodeling would be fundamentally the same in all ethnic groups.


This work was supported in part by National Institutes of Health grants AG10381 and PO1 AG/AR13918.