Bone Remodeling Increases Substantially in the Years After Menopause and Remains Increased in Older Osteoporosis Patients


  • The authors have no conflict of interest


Bone remodeling rates (Ac.f) were measured in transilial biopsy specimens from 50 healthy premenopausal women before and 1 year after menopause, in 34 healthy women 13 years past menopause, and in 89 women with untreated osteoporosis. Ac.f nearly doubled 1 year after menopause, tripled 13 years after menopause, and remained elevated in women with osteoporosis.

Introduction: Increased bone remodeling rates are associated with increased skeletal fragility independent of bone mass, partially accounting for the age-related increase in fracture risk in women that is independent of bone loss. We examined bone remodeling rates before and after menopause and in women with osteoporosis by measurements of activation frequency (Ac.f, #/year) in transilial bone biopsy specimens.

Materials and Methods: We recruited 75 women, >46 years old, who had premenopausal estradiol and gonadotropin levels and regular menses. During 9.5 years of observation, 50 women experienced normal menopause and had 2 transilial bone biopsy specimens after tetracycline labeling, one at the beginning of observation and the second 12 months after the last menses, when serum follicle-stimulating hormone (FSH) was >75 mIU/ml and serum estradiol was <20 pg/ml. Ac.f was also computed for a group of older healthy postmenopausal women and a group of women with untreated osteoporosis studied earlier by the same biopsy (Bx) and labeling protocol.

Results: Median Ac.f rose from 0.13/year to 0.24/year (p < 0.001) across menopause and was greater still in the older normals (p < 0.008) than in the second Bx. Ac.f was not significantly greater in the osteoporosis patients than in the older postmenopausal normals.

Conclusion: Bone remodeling rates double at menopause, triple 13 years later, and remain elevated in osteoporosis. This change contributes to increases in age-related skeletal fragility in women.


Menopause plays a major role in bone loss and subsequent fracture risk in adult women.(1–5) However, loss of bone mass (usually expressed as BMD obtained from DXA) is not the only basis for increased fracture risk after menopause and may not even be the most important one.(6) The bone loss paradigm of skeletal fragility has come under scrutiny because of anomalies emerging from its application. For example, fracture risk increases with age independent of change in BMD(7), (8); prior fracture predicts risk of future fracture independent of BMD(9), (10); reduction in fracture risk with antiresorptive treatment is not quantitatively concordant with change in BMD(11), (12); and increase in bone remodeling rates is associated with increase in fracture risk independent of change in BMD.(13) Heaney(6) has recently pointed out that increase in remodeling rate is a prominent factor underlying these anomalies in the BMD paradigm listed and that increased bone remodeling rates may be the most important element underlying the increased fracture risk with age, rather than loss of BMD.

Thus, measurement of bone remodeling rates in healthy women before, during, and after normal menopause is of interest in examining the mechanisms behind the age-related increase in fracture risk that is independent of BMD loss. Furthermore, the opportunity of studying tissue-level remodeling rates bypasses the problems encountered in assessing bone remodeling rates using urinary markers of bone remodeling. The transilial biopsy approach provides unambiguous tissue-level information on remodeling rates.

This report describes activation frequency (Ac.f) in three groups of women: (1) normal healthy women just before menopause and again 12 months after their last menses, when serum estradiol (E2) and follicle-stimulating hormone (FSH) are clearly at menopausal levels; (2) healthy postmenopausal women evenly distributed between ages 45 and 75; and (3) a group of untreated osteoporosis patients. Histomorphometric results from the latter two groups have been partially described in two earlier papers.(14), (15)


Human subjects

Data from three groups of subjects are included in this report. The Creighton Institutional Review Board reviewed and approved all protocols, and all subjects provided signed written consent. Details of recruitment for all three groups have been presented elsewhere.(14–16)

For group 1, in 1988–1989, we enrolled a convenience sample of 75 healthy white women into a study to determine bone tissue- and cell-level changes in bone remodeling that occur at the time of rapid bone loss at menopause.(16) The study design called for transilial bone biopsy specimens after tetracycline labeling on entry into study, and again 12 months after the last menses. Entry criteria included ≥46 years of age and at least six menses during each of the prior 2 years, even if somewhat irregular. Entry also required serum levels of E2 >50 pg/ml and FSH <25 mIU/ml on samples obtained between the 17th and 25th day of the menstrual cycle. We excluded anyone with a diagnosis or treatment that would affect bone health(16) and documented good health by physical and clinical laboratory examination. The Creighton University Institutional Review Board approved the study, and all participants provided written consent before entry.

Study visits occurred at 6-month intervals until the end of the period of observation 9.5 years later. The date of menopause was defined as the date of the last menses that was followed by 12 months without menses. The definition also required a serum E2 <20 pg/ml and FSH >75 mIU/ml at 12 months of absent menses. Bone densitometry by DXA, serum and urine biochemical measurements, and health histories were obtained at each visit. A previous report from this study(16) included densitometry and biochemistry data, but no description of the transilial biopsy data. This report focuses on the Ac.f data from the biopsy specimens because it is the most direct measurement of tissue-level bone remodeling, the focus of this communication.

Sixty paired biopsy specimens were obtained. Fifty-one members of this group of 60 underwent natural menopause during observation without hormone replacement therapy. One subject did not have visible tetracycline labeling on the first biopsy, leaving 50 subjects that form the basis of this report. The other nine subjects who underwent a second biopsy were either placed on hormone replacement therapy before menopause by their private physicians, usually to treat menopausal symptoms, or did not pass through menopause during the 9.5 years of observation. These subjects are not included in this report.

Group 2 included 34 healthy postmenopausal individuals recruited for a single transilial biopsy to establish a normal postmenopausal biopsy reference database.(14) Their ages were evenly distributed between 45 and 75 years, and they were on average 13 years after menopause. Good health was documented by clinical and laboratory exam.

Group 3 consisted of untreated patients with postmenopausal osteoporosis,(15) defined as having at least two or more vertebral deformities (i.e., 25% reduction in anterior vertebral height as judged by inspection of radiographs by an experienced observer). In this study, we recruited 57 subjects from the Omaha area and 33 from the New York area to compare with the normals.

Ac.f, the subject of this study, had not been reported for either of these groups at the time of their respective publication. One of the osteoporotic women did not have visible tetracycline labels in their biopsy specimens. Because we could not verify that she complied with the tetracycline regimen, she was not included in the analysis. Three of the osteoporotic women did not have visible tetracycline label in the trabecular bone but did so in the cortex. Their values for Ac.f are each included as zero in the analysis of trabecular Ac.f.



Initially, for the perimenopausal women, we obtained serum specimens for E2 and gonadotropin levels between the 17th and 25th day of the menstrual cycle. Later, when irregular bleeding intervals supervened, the serum samples were obtained at times unrelated to menstrual cycles. Details of the assay methods used in this study have been previously published.(16)

Transilial biopsy

In all three groups of women, each subject received in vivo double tetracycline labeling as follows: oral tetracycline hydrochloride (250 mg, qid) for 3 days (label 1), followed by a 14-day drug-free interval, and then 3 days of oral tetracycline hydrochloride (250 mg, qid; label 2). Five to 10 days after the end of label 2, a transilial biopsy was performed using a trephine with inner diameter of 7.5 mm as described previously.(14) The biopsy specimen was fixed, embedded, sectioned, and evaluated as described previously.(17) Two sections were read from the central part of each biopsy, taken 250 mm apart. The histomorphometric variable, Ac.f, is the annual probability of activation of a new remodeling site at any given locus on the trabecular surface and is expressed as #/year. It is calculated as follows: Ac.f = (BFR/BS)/W.Th, where BFR/BS is the bone formation rate, surface referent, and W.Th is the wall thickness. BFR/BS is calculated as: MAR × MS/BS, where MAR is the mineral apposition rate (μm/day), corrected for obliquity (ρ/4), and MS/BS is the mineralizing surface divided by the total trabecular surface (a dimensionless ratio).(18) W.Th is directly measured on toluidine blue-stained sections as the average distance between cement lines and quiescent trabecular surfaces corrected for obliquity.(14) Measurements are chosen at sites where cement lines are the product of a completed remodeling site. Measurements are made at the midpoint of the packets.

Measured and calculated variables in trabecular bone histomorphometry used here are described elsewhere.(18) Histomorphometric methods for all three studies were identical.

Data analysis

The analysis of within-subject change in variables in the 50 women from the perimenopausal women (Table 1) was done by paired t-test. Comparisons between the three groups of women were done using Kruskal-Wallis and Mann-Whitney tests (SPSS for Windows; SPSS, Chicago, IL, USA).

Table Table 1.. Age, Anthropometrics, and Biopsy Data8
original image


Subject characterization

Table 1 contains relevant data on the subjects whose biopsy specimens form the basis for this report. In the perimenopausal group, the average age on entry was 49.4 ± 1.9 years, and the average age at the second biopsy was 54.6 ± 2.2 years. The average height at the time of the two biopsy specimens was 1.646 ± 0.051 and 1.649 ± 0.049 m, respectively (not significant). The average weights at the time of each biopsy were 68.14 ± 12.17 and 71.4 ± 13.28 kg, respectively (p < 0.01), exhibiting the usual mid-life weight gain. The following data for these 50 subjects are reported in previous publications(14), (16) and are not shown in Table 1: the average age at last menses was 53.0 ± 2.1 years, with a range of 48–58 years; the mean T score for spine BMD was −0.454 ± 1.064 and for femoral neck BMD was 0.014 ± 1.043. The average age for the postmenopausal group was 60.0 ± 7.6 years, and the mean height was 1.624 ± 0.052 m. For the osteoporotic women, the average age was 67 ± 7.2, the average height was 1.56 ± 0.06 m, and the average weight was 58.3 ± 11.7 kg. The average time since menopause was 13 ± 8 years.

Hormone data

Hormone data were obtained only in the perimenopausal group. Their E2 levels, previously reported,(16) were above 50 pg/ml on entry in every subject, began to decline about 1.5 years before the last menses, and were unmeasureable in every subject (<20 μg/ml) at the time of the second biopsy 1 year after the last menses. In every case, the date assigned for menopause was the date of the last menses. At the time of the first biopsy, serum FSH levels were <25 mIU/ml in every case, and at the time of the second biopsy were >75 mIU/ml in every case.


The median Ac.f was 0.13/year (95% CI, 0.02, 0.63) in the first biopsy of the perimenopausal group and 0.24/year (95% CI, 0.01, 0.77) in the second biopsy, an 85% increase (p < 0.001, paired t-test; see Table 1). The median Ac.f in the healthy postmenopausal women at an average age of 60.0 ± 7.6 years was 0.37/year (95% CI, 0.06, 0.94). This value is nearly triple the premenopausal value and is 54% higher than the value for early postmenopausal women at an average age of 54.6 ± 2.2 years (p < 0.008, Mann-Whitney test; see Table 1 and Figure 1). In the patients with postmenopausal osteoporosis, median Ac.f was 0.42/year (95% CI, 0.00, 1.40), which was higher still than in the older postmenopausal normal subjects (not significant).

Figure FIG. 1..

Frequency distributions of the Ac.f values in all four data sets. The median values for 12 months after menopause, for 13 years after menopause, and for osteoporotics were all significantly greater than the premenopausal median. The median for the osteoporotics, although higher than in the other groups, was not significantly different from the normals 13 years past menopause. Note, especially, the progressive widening of the distributions with age and disease (Reprinted with permission from Robert R. Recker, 2003).20

Thus, Ac.f increased substantially immediately after menopause with loss of endogenous estrogen, increased further an average of 13 years later, and was marginally higher still in individuals with osteoporosis. The changes in Ac.f are shown in Fig. 1, where we plot frequency distributions of the values for all four sets of data. The values for the healthy premenopausal women are distributed at the low end of the range. The distributions begin to spread to higher values 12 months after the last menses and spread still more in the older menopausal women and individuals with osteoporosis. As is visibly evident on inspection, the principal change is a marked increase in the dispersion of the Ac.f values, with some members of each group exhibiting low values, whereas others increased markedly. The women with osteoporosis exhibited several values ≥1/year, whereas none of the values in the other distributions were >0.94/year.

Table 1 also contains values for BFR/BS and W.Th, the variables used in the calculation of Ac.f. The median value for BFR/BS increased from 0.005 to 0.009 mm3/mm2/year (p < 001) in the perimenopausal normals and to 0.012 mm3/mm2/year (not significant versus the second biopsy in the perimenopausal normals) in both the postmenopausal normals and osteoporotics (not significant). W.Th values did not show a significant transmenopausal change in the perimenopausal normals. The value was 32.2 mm in the postmenopausal normals (p > 0.001 versus the perimenopausal normals) and 28.3 mm in the osteoporotics (p < 0.001 versus the postmenopausal normals).


Activation frequency is calculated from microscopic measures of fluorochrome tissue-time markers from transilial bone biopsy specimens. As such, it is a direct, tissue-level measure of bone remodeling rate. Its major limitation is the small sample size of bone from a single skeletal site, but because it is a direct, tissue-level measurement, it does not suffer the major limitations of the biochemical markers of remodeling. However, while Ac.f is a reliable measure of the remodeling rate, other factors also affect bone, such as the volume-based formation rate expressed as BFR/BV, the annual fractional rate of trabecular bone tissue renewal. While Ac.f is, strictly speaking, an expression of the intensity of those signals that lead to the initiation of remodeling, it is also a measure of bone remodeling rate because calculations of both Ac.f and BFR/BV contain the product of MAR times MS/BS in their numerators.

The value for Ac.f in our premenopausal subjects was lower than the value reported by Han(19) (0.13 versus 0.36), and the value in our osteoporotics was higher than that reported in the same center.(20) There is no ready explanation for these differences. They are likely caused by differences in the populations studied rather than differences in histomorphometry technique, but this cannot be tested. The increases in Ac.f at menopause and in osteoporosis patients are greater than the increases in biochemical markers of bone remodeling reported in similar subjects.(21) This may be largely because of the fact that, because they are markers of global remodeling, their values will be affected by the size of a subject's skeleton, whereas the Ac.f will not be so affected. There are surely other reasons as well.

One potential problem in analysis of Ac.f in groups is the occasional absence of fluorochrome in a given biopsy. This may happen because of poor specimen quality or very low remodeling rates or because subjects failed to take the tetracycline. With low remodeling, absent labels may be because of the fact that labels were so infrequent that, by chance, they did not appear in the particular sections that were prepared for microscopy. Fortunately, no biopsy specimens exhibited missing fluorochromes in our data set from older normals,(14) and there was only one missing in the transilial biopsy specimens in 90 untreated patients with postmenopausal osteoporosis.(15) This subject's specimen exhibited no fluorochrome label on trabecular or cortical surfaces after an extended search of multiple sections. Because we could not verify that this woman complied with the tetracycline regimen, she is not included in this analysis. Three others in the osteoporosis group exhibited no fluorochrome label on trabecular surfaces but did so in the cortex. The values for Ac.f each are included as zero in the analysis.

Variation in bone remodeling rates has become recognized as a powerful determinant of fracture risk.(11) In fact, it may now be displacing BMD as the best explanation for fracture risk in humans.(6) The importance of increases or decreases in bone remodeling rates is seen in population studies,(22), (23) and in osteoporosis treatment trials using antiresorptive agents.(11), (24–26) Every successful antiresorptive trial has shown reduction in fracture risk out of proportion to the increase in BMD.(11) Furthermore, all have shown substantial reductions in remodeling, as measured by bone formation or resorption markers. In these reports, fracture risk was inversely related to remodeling even in the placebo groups.

A number of observations seen in population studies in the past several decades have shown anomalies in the paradigm of low bone mass as the determinant of fracture risk. These anomalies may be resolved, at least to some extent, if bone remodeling rates are recognized as determinants of fracture risk that are as strong as, or stronger, than BMD. Perhaps the earliest hint of this was the demonstration by Hui et al.(7) that age was a strong determinant of fracture risk independent of BMD measured at the forearm. This was confirmed recently at other skeletal sites using DXA measurements.(8) In the latter report, the risk of hip fracture in the presence of a hip T score of −2.5 was about 2-fold greater at the age of 75 than at the age of 65.

There are several mechanisms one can hypothesize to explain the increased fragility resulting from excessive remodeling. These are (but not limited to) (1) increase in rate of loss of trabecular elements, (2) loss of connectivity, (3) weakened trabeculae because of increased numbers of unfilled resorption lacunae, (4) erosion of the cortex on the endosteal surface causing reduction in the moment of inertia, and (5) overall increase in bone loss because of negative bone balance at each remodeling site. One cannot choose among these, and perhaps all may play a role.

Heaney has postulated that bone remodeling activity is higher than required for repair of microdamage.(6) These observations confirm the presence of greatly increased remodeling activity in the population most at risk of osteoporotic fractures. Tissue-level remodeling rates increase at menopause and reach nearly 3-fold premenopausal levels by 13 years after menopause. Moreover, the high rates are maintained, or even increased, in untreated postmenopausal women with osteoporosis characterized by the presence of fragility fractures. Thus, the observations we report here offer a mechanism explaining at least part of the age-related increase in fracture risk that is independent of measured bone mass.

How can it be that a process designed by evolution to strengthen bone by removing aged and damaged tissue should turn out to be a source of weakness for contemporary humans? A possible answer is that, while remodeling strengthens bone, it produces transient weakness, because removal of old bony material further weakens bone locally. If most remodeling is driven by mechanical need, the transient weakness is more than offset by the improved strength. However, any remodeling in excess of mechanical need contributes only weakness. These data suggest that the remodeling rate, as measured by Ac.f in healthy premenopausal women, is the rate that satisfies mechanical need. Thus, the very much higher rates seen immediately after menopause, and sustained in older postmenopausal and osteoporotic women, are largely not driven by mechanical need and thus weaken the skeleton.

Perhaps the most striking feature of the observations reported here is not that Ac.f increases at midlife, but that it does so only in some women. The increase in the dispersion of Ac.f values is even more drastic than the increase in the median values from premenopausal levels (Fig. 1). Our data do not permit an answer to the question of why some do and some do not and of whether the two groups have differing fracture risk. The relatively low remodeling in those who retain a premenopausal level of remodeling may mean that they are at low risk of fracture or that the fractures they have are not osteoporotic fragility fractures. On the other hand, they may be subjects with osteoporosis at risk of low trauma fracture, not related to high remodeling. If so, they may be members of a subpopulation who do not respond to antiresorptive therapy and thus remain at risk fracture while on treatment. The answers cannot be determined from these data.

The perimenopausal changes in Ac.f that we report here result largely from transmenopausal increases in values for BFR/BS, a term appearing in the numerator of the Ac.f calculation. However, the further increase in Ac.f in the postmenopausal normals and osteoporotics in this report results largely from reduced values for W.Th, a term appearing in the denominator of the Ac.f calculation. One interpretation of these findings is that a decline in W.Th signals a mechanism (unknown) that causes BFR/BS to be maintained. Alternatively, the sustained increase in Ac.f and reduction in W.Th may be independent and unrelated consequences of the aging process that coincidentally result in maintenance of BFR/BS. Other mechanisms may also be operating to explain these phenomena.

The observations we report here in transilial bone biopsy specimens in normals and untreated osteoporotics confirm the presence of marked increases in trabecular bone remodeling at the tissue level. These observations offer support for a shift in the paradigm of BMD as the major determinant of bone strength, focusing instead on bone remodeling rate as a determinant of bone strength, perhaps of more importance than BMD.


The authors thank June Bierman and Jan Leist. This work was supported by NIH Grants AR39221 and AG04275.