Transmenopausal Changes in Trabecular Bone Quality



Bone strength depends on its amount and quality. Bone quality includes its structural and material properties. Bone material properties are dependent on bone turnover rates. Remodeling rates are significantly increased immediately after menopause. In the present study, we used Raman microspectroscopic analysis of double iliac crest biopsies with a spatial resolution of 1 µm obtained before and immediately after menopause (1 year after cessation of menses) in healthy females to investigate changes in material properties attributable to menopause. In particular, the mineral/matrix ratio, the relative proteoglycan and lipid content, the mineral maturity/crystallinity, and the relative pyridinoline collagen cross-link content were determined in trabecular bone as a function of surface metabolic activity and tissue age. The results indicate that significant changes (specifically in mineral/matrix ratio) were evident at active bone forming surfaces, whereas the relative proteoglycan content was altered at resorbing surfaces. These changes were not accompanied by altered mineral content or quality as monitored by Raman microspectroscopic analysis. © 2014 American Society for Bone and Mineral Research.


Hormonal changes attributable to menopause are considered to be among the culprits of postmenopausal osteoporosis.[1-4] These changes are believed to result in bone tissue loss and decrease in bone mineral density (BMD), mainly through alterations in bone remodeling rates. Specifically, bone remodeling rates (based on activation frequency of new remodeling sites) in women 1 year after cessation of menses are roughly twice those of healthy premenopausal women.[5, 6]

Loss of bone mass, measured clinically as change in BMD, is considered an important risk factor for bone fragility. However, BMD is not the sole predictor of whether an individual will experience a fracture,[7, 8] and there is considerable overlap in BMD between populations that do and do not develop fractures.[9-11] It has been demonstrated that for a given bone mass an individual's risk of fracture increases with age.[12] Consistent with these findings, numerous investigators have shown that mechanical variables directly related to fracture risk are either independent of[13] or not totally accounted for by bone mass itself.[14-18] Epidemiological evidence also shows considerable overlap of bone density values between fracture and nonfracture groups, suggesting that low bone quantity alone is an insufficient explanation for fragility fractures.[19-21] It is becoming evident then, that in addition to BMD, bone quality should also be considered when assessing bone strength and fracture risk. Bone quality is a broad term[22] encompassing a plethora of factors such as geometry and bone mass distribution, trabecular bone microarchitecture, microdamage, and increased remodeling activity, along with genetics, body size, environmental factors, and changes in bone mineral and matrix tissue properties.[10, 11]

One of the obstacles to be circumvented when assessing mineral and matrix tissue properties is tissue heterogeneity at the microscopic level. In normal humans, cortical bone constitutes approximately 80% of the human skeletal mass and trabecular bone approximately 20%.[23] Bone surfaces may be undergoing formation or resorption, or they may be inactive. These processes occur throughout life in both cortical and trabecular bone.[23] The rate of cortical bone remodeling, which may be as high as 50% per year in the midshaft of the femur during the first 2 years of life, eventually declines to a rate of 2% to 5% per year in the elderly. Rates of remodeling in trabecular bone are proportionately higher throughout life and may normally be 5 to 10 times higher than cortical bone remodeling rates in the adult.[23] This information is critical when evaluating bone at the microscopic level.

Raman microspectroscopic analysis is a useful research tool that provides information on mineral and matrix properties in undecalcified bone with a spatial resolution of ∼ 0.6 to 1 µm. Variables that can be established with this analysis include mineral/matrix ratio, mineral maturity/crystallinity, relative pyridinoline (Pyd) collagen cross-link content, and relative proteoglycan and lipids content. Moreover, because of Raman's superb spatial resolution, these variables may be reported as a function of tissue age, using histologic landmarks such as fluorescent double labels, thus minimizing the effect of bone turnover changes.

In the present study, we utilized Raman microspectroscopic analysis to establish the above-mentioned bone tissue properties in double iliac crest biopsies obtained from healthy women before (PreM) and after menopause (PostM), as a function of tissue age. The results indicate that all monitored outcomes were dependent on tissue age, whereas the mineral/matrix ratio at active bone forming surfaces was also dependent on menopause status, as was the relative proteoglycan content at resorbing surfaces.

Materials and Methods

Bone specimens

The biopsies used in the present study were a randomly selected subset of nine double biopsies from a previously published study[6] in which iliac crest biopsies were obtained from healthy premenopausal women (aged >46 years) before and 1 year after cessation of menses. First and second biopsies were taken at the average ages of 49.0 ± 1.9 (standard deviation [SD]) and 54.6 ± 2.2 (SD) years. Inclusion in the study required serum levels of E2 >50 pg/mlL and FSH <25 mIU/mL on samples obtained between the 17th and 25th day of the menstrual cycle. Baseline characteristics are provided in Table 1.

Table 1. Patient Baseline Characteristics
Patient IDAge (years)Body mass indexFollicle-stimulating hormone (u/mL)Estradiol (pg/mL)Hydroxyproline (µm\L\mMol\L)Bone-specific alkaline phosphatase (u/mL)

Raman microspectroscopic analysis

For Raman microspectroscopic analyses, a Senterra (Bruker Optik GmbH, Ettlingen, Germany) instrument was employed. A continuous laser beam was focused onto the sample through a Raman fluorescence microscope (Olympus BX51, objective 50 ×) with an excitation of 785 nm (100 mW) and a lateral resolution of ∼0.6 µm.[24] The Raman spectra were acquired from the biopsy block polished surface, using a thermo-electric–cooled charge-coupled device (CCD) (Bruker Optik GmbH). All data analysis was done with the Opus Ident software package (OPUS 6.5, Bruker Optik GmbH). Once acquired, the Raman spectra were baseline corrected (rubber band, five iterations) to account for fluorescence, and the following Raman variables were calculated as published elsewhere.[24] The mineral/matrix ratio was expressed as the ratio of the integrated areas of the v2PO4 (410 to 460 cm−1) to the amide III (1215 to 1300 cm−1) bands. In cases where there were differences between the two groups, the individual mineral and organic matrix values were calculated as the ratio of v2PO4/PMMA (494 to 509 cm−1) and amide III/PMMA, respectively. The relative proteoglycan (PG) content was defined as the PG/matrix ratio, which was calculated from the ratio of the integrated areas of the proteoglycan/CH3 (1365 to 1390 cm−1) band representative of glycosaminoglycans,[25, 26] to the amide III (1215 to 1300 cm−1) bands, respectively. The relative lipids content was expressed as the ratio of the integrated area of the lipids band ∼1298 cm−1/amide III.[27] The maturity/crystallinity of the bone mineral apatite crystallites was approximated from the inverse of the full width at half height (FWHH) of the v1PO4 (930 to 980 cm−1) band.[28, 29] Finally, the relative Pyd content (a major trivalent collagen cross-link) was calculated as the absorbance height at 1660 cm−1/area of the amide I (1620 to 1700 cm−1).[29-33]

Fluorescent labeling

All women received 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 labeling, an iliac crest biopsy was obtained under local anesthesia using a trephine with inner diameter of 7.5 mm.

Anatomical area selection criteria

For each biopsy, trabeculae were analyzed at three distinct tissue areas corresponding to different tissue ages: actively forming surfaces (based on the presence of fluorescent double labels), trabeculae centers (indicative of quiescent, heavily mineralized anatomical areas), and resorbing surfaces (based on the presence of resorption pits).

In the case of actively forming bone trabecular surfaces, three trabeculae with clearly discernible double tetracycline labels were analyzed in the following regions (Fig. 1): between second label and mineralizing front corresponding to a mineralized tissue age of about 3 days designated as “youngest,” between the two labels corresponding to an age of 12 days designated as “label,” and right behind the first label corresponding to an age of 20 days designated as “older.” In each of these regions, three measurements were obtained, for a total of nine measurements per patient per anatomical area. For each patient, these nine values, representative of every tissue age-specific region, were averaged and the resultant value was treated as a single statistical unit.

Figure 1.

(A) A trabecula with double fluorescent labels present on one of its surfaces and resorption pits on another are evident. (B) An actively forming trabecular surface with the anatomical locations corresponding to “youngest,” “label,” and “older” locations are appropriately labeled. (C) A picture showing the anatomical location where measurements labeled as “center” were obtained. (D) A typical Raman spectrum with the appropriate peaks labeled. In the inset, the spectral range showing the peaks utilized for proteoglycan and lipid content is isolated.

Additionally, three measurements in the geometrical center of each trabecula utilized were also acquired, representative of older, more mineralized tissue with prolonged secondary mineralization designated as “center.”

Finally, three trabecular surfaces with resorption pits evident were also analyzed for each biopsy examined, in the middle of the pit and right at the mineral edge.

Statistical analysis

In the case of actively forming trabecular surfaces, the data were evaluated by two-way ANOVA to determine the effects of menopause status and tissue age (because the presence of double labels allows the exact determination of tissue age), with Bonferroni post hoc tests.

In the case of trabecular centers and resorbing surfaces, if the data were normally distributed they were compared by paired t tests; otherwise they were compared by Wilcoxon matched-pairs test. Statistical significance was assigned to p < 0.05. Data are presented as mean ± SEM.

Finally, to check whether the outcomes of the present study were indeed independent of changes in bone remodeling rates, the results of the Raman analysis as a function of tissue age at forming and resorbing surfaces were correlated with appropriate histomorphometric outcomes in the same patients, and specifically, mineral apposition rate (MAR), bone formation rate (BFR/BS), osteoblast surfaces (ObS/BS), activation frequency (AcF), and osteoclast surfaces (OcS/BS), using either Pearson's or Spearman's correlation tests depending on whether the data were normally distributed.


Typical photomicrographs showing the anatomical areas considered, along with typical Raman spectra, are provided in Fig. 1. Histomorphometric variables are shown in Fig. 2, where MAR (A), AcF (B), BFR/BS (C), ObS/BS (D), and OcS/BS (E) are compared in premenopausal versus postmenopausal biopsies. All values were significantly elevated in the postmenopausal group of bone specimens compared with premenopausal, with the exception of MAR, signifying that the postmenopausal subjects had a higher bone turnover state.

Figure 2.

Comparison of bone turnover histomorphometric parameters between the PreM and PostM. Although there were no differences in MAR (A), the AcF (B), BFR/BS (C), ObS/BS (D), and OcS/BS (E) values were significantly elevated in the PostM group.

At actively forming bone surfaces (based on the presence of fluorescent double labels), all monitored variables were significantly dependent on tissue age (Table 2). The mineral/matrix ratio (Fig. 3A), the relative proteoglycan content (Fig. 3D), the mineral maturity/crystallinity (Fig. 3F), and the relative Pyd content (Fig. 3G) increased, whereas the relative lipids content decreased as a function of tissue age (Fig. 3E). Significant differences within the two subgroups as a function of tissue age are denoted by bars (solid in the case of PreM and dotted in the case of PostM), whereas p values are indicated by open circles. Moreover, two-way ANOVA (Table 2) indicated that the mineral/matrix ratio was also dependent on menopause status, exhibiting higher values postmenopause at the oldest of the three tissue ages considered at active bone forming trabecular surfaces, attributable to decreased organic matrix content (Fig. 3C), whereas the mineral content remained unaltered (Fig. 3B).

Table 2. Results of Two-Way ANOVA Analysis at Active Bone Forming Trabecular Surfaces for the Various Variables Calculateda
 InteractionMenopause statusTissue age
  1. a The p values are listed. Statistical significance is denoted in bold typeface.
Relative proteoglycan content (PG/matrix)0.2100.24740.0028
Relative lipids content (lipids/matrix)0.87490.7559<0.0001
Mineral maturity/crystallinity (1/FWHH)0.06580.1696<0.0001
Relative pyridinoline collagen cross-link content (Pyd/matrix)0.10230.0887<0.0001
Figure 3.

At active bone forming trabecular surfaces, the mineral/matrix ratio increased both as a function of tissue age and in postmenopausal (PostM) specimens (A; Table 1). This increase was not the result of a change in mineral content (B) but rather a decrease in the organic matrix content (C). The relative proteoglycan content increased as a function of tissue age only (D; Table 1), as was the case with the relative lipids content (E; Table 1). Similarly, the mineral maturity/crystallinity increased as a function of tissue age but was independent of menopause status (F; Table 1). Finally, the relative Pyd content increased as a function of tissue age (G; Table 1). Significant differences between the various tissue ages within the PreM group are denoted by solid bars and within the PostM group by dotted bars.

At the center of trabecular bone, none of the monitored variables differed between pre- and postmenopause (Fig. 4).

Figure 4.

In trabecular centers, the values of the mineral/matrix (A), the relative proteoglycan content (B), the relative lipids content (C), the mineral maturity/crystallinity (D), and the relative Pyd content (E) were similar between premenopausal (PreM) and postmenopausal (PostM) bone specimens.

At resorbing trabecular surfaces, none of the monitored outcomes differed between pre- and postmenopause (Fig. 5), with the exception of relative proteoglycan content (Fig. 5B), which was significantly higher postmenopause.

Figure 5.

At resorbing surfaces, there were no significant differences between premenopausal (PreM) and postmenopausal (PostM) specimens in mineral/matrix (A), relative lipids content (C), mineral maturity/crystallinity (D), and relative Pyd collagen cross-link content values (E), whereas the relative proteoglycan content (B) was higher in the postmenopausal specimens.

Correlations between the various Raman analysis outcomes as a function of tissue age and appropriate histomorphometric data on the same patients are listed in Table 3. There were no significant correlations between Raman and histomorphometry parameters in the PreM group, with the exception of the relative lipids content, which was directly correlated with osteoblast surface at active bone forming trabecular surfaces. On the other hand, several significant correlations with histomorphometry parameters exist in the PostM group. The mineral apposition rate was significantly and inversely correlated with the relative proteoglycan content at active bone forming trabecular surfaces. Moreover, bone formation rate was inversely correlated with the mineral/matrix ratio, mineral maturity/crystallinity, and relative Pyd content at active bone forming surfaces. Osteoblast surface directly correlated with relative lipids content in the PreM group and inversely with relative Pyd content at active bone forming trabecular surfaces in the PostM group. Activation frequency was directly correlated with the relative proteoglycan content at resorbing surfaces and inversely with mineral maturity/crystallinity at active bone forming ones in the PostM group. Finally, the osteoclasts surface was inversely correlated with the relative proteoglycan content at resorbing surfaces.

Table 3. Correlations Between the Raman Analysis Parameters and Histomorphometrya
Raman parametersTissue ageMineral apposition rate (MAR)Bone formation rate (BFR/BS)Osteoblast surfaces (ObS/BS)Activation frequency (AcF)Osteoclast surface (OcS/BS)
  1. aThe p values are listed. Significant correlations are listed in bold typeface, and in these cases, the second entry is the correlation coefficient.
 Resorbing      0.7040.0540.5920.438
 Resorbing      0.6300.0430.5880.023
         0.681 −0.737
 Resorbing      0.3580.8940.9150.613
Mineral maturity/crystallinityYoungest0.4400.8340.2560.3610.8960.2770.2840.631  
     −0.815   −0.903  
 Resorbing      0.8640.5000.9990.746
Relative Pyd contentYoungest0.3180.1670.7780.9980.6600.5360.7080.883  
     −0.660 −0.699    
 Resorbing      0.2520.4020.2310.744


The results of the present study indicate that in addition to bone remodeling rates, menopause affects bone organic matrix at trabecular surfaces exclusively and differentially (forming versus quiescent versus resorbing surfaces) and in particular mineral/matrix ratio (increased) at actively forming trabecular surfaces and relative proteoglycan content (increased) at resorbing trabecular surfaces. No differences were evident at the centers of the trabecular struts analyzed. These changes were not accompanied by changes in the quantity and quality of the bone mineral component at the same anatomical locations.

Mineral/matrix ratio is an expression of bone density that, unlike BMD by dual-energy X-ray absorptiometry (DXA), accounts for both the amount of mineral and organic matrix in the bone volume analyzed, and has been reported to be directly proportional to bending stiffness and failure moment.[34] On the other hand, one has to be aware that a change in this ratio may be the result of several scenarios involving the amounts of mineral and organic matrix components. For example, an increased ratio may even be attributable to a disproportionate decrease in both mineral and organic matrix content (greater reduction in the amount of organic matrix compared with the reduction of mineral content). Although several scenarios may result in an increased mineral/matrix ratio, the effects on bone strength are not expected to be the same, as the mineral and organic matrix content have been associated with the plastic and elastic mechanical properties of bone, respectively.[35] In the present study, although no differences in this ratio were observed between PreM and PostM bone biopsies either at trabecular centers or resorbing surfaces, at active bone forming surfaces it significantly increased as a function of tissue age and was also significantly higher in the PostM ones, owing to a decrease in the organic matrix content, whereas the mineral content remained unaltered. The anatomical site (forming surfaces exclusively) and the observed change (decrease) of the organic matrix content are in agreement with published reports showing that estrogen deprivation influences osteoblasts and attenuates collagen synthesis.[36-39] Moreover, the increase in this ratio is because of a decrease in the organic matrix content, unlike what has been previously reported for osteoporotic patients treated with either estrogen-replacement therapy or bisphosphonates,[40-42] where the increased ratio was mostly the result of elevated mineral content. Finally, this ratio was inversely correlated with bone formation rate at active bone forming surfaces in the PostM specimens.

Proteoglycans have been suggested in the literature to play a multifaceted role involving the modulation of both organic matrix mineralization and remodeling rates. Moreover, different proteoglycans are present at different microanatomical locations fulfilling different roles. In in vitro studies, proteoglycans have been shown to regulate hydroxyapatite crystal formation.[43] In osteoblastic cell cultures, decorin has been shown to modulate matrix mineralization.[44, 45] In animal models, biglycan deficiency has been reported to increase osteoblast activity and affect the osteoclasts as well, resulting in an osteoporosis-like phenotype.[46-48] Proteoglycans have also been identified in perilacunar matrix around the osteocyte lacunae and around the canaliculi[49] in compact lamellar rat and human bone, and it has been proposed that a plausible role of these (and in particular perlecan/Hspg2 [PLN]) was to prevent mineralization so as to ensure uninhibited interstitial fluid movement.[50] Proteoglycans have been implicated in osteoclastogenesis and remodeling regulation as well.[46, 51, 52] In the present study, at actively forming surfaces, the relative proteoglycan content was independent of menopause status, whereas it increased as a function of tissue age. One plausible explanation for the tissue age-dependent changes may be differences in canalicular network density between active bone forming surfaces and older, more densely mineralized bone. Similar values between PreM and PostM specimens were also observed at trabecular centers, whereas increased values were measured at resorbing surfaces in the PostM group. Relative proteoglycan content correlated with histomorphometry parameters only in the PostM group at metabolically active surfaces. Specifically, it was inversely correlated with MAR at actively forming surfaces, in agreement with both the reported inhibitory effect on hydroxyapatite crystal nucleation and subsequent growth,[43] as well as the reported increase in MAR in the biglycan knockout mice.[46] At resorbing surfaces, it was directly correlated with AcF and inversely correlated with osteoclast surface (consistent with their previously mentioned role in osteoclastogenesis and remodeling regulation[52] and relationship between biglycan deficiency and osteoclast differentiation.[47] Unfortunately, Raman microspectroscopic analysis to date cannot discriminate between different proteoglycan species. On the other hand, it should be kept in mind that the Raman spectral signature of proteoglycans is the result of the glycosaminoglycan (GAG) chains,[25, 26] and in bone, chondroitin 4-sulfate constitutes ∼90% of the total GAG content and is found predominantly in biglycan and decorin.[53]

The relative lipid content was also independent of menopause status and dependent on tissue age (decreasing as a function of tissue age) at active bone forming trabecular surfaces, whereas no differences were apparent at either trabecular centers or resorbing surfaces. Lipids have been reported to be nucleators of collagen fibers.[54-56] Interestingly, the relative lipid content directly correlated with osteoblast surface at active bone forming trabecular surfaces exclusively in the PreM group, in agreement with the mineral nucleators' point of view, and potentially indicating an altered mineralization mechanism postmenopause.

At forming trabecular surfaces, consideration of the relative proteoglycan and lipids content (normalized to organic matrix content) within groups showed some significant differences as a function of tissue age. On the other hand, two-way ANOVA analysis showed that there was no dependency on menopause status, something that was confirmed by individual t tests, as no differences between the PreM and PostM groups were evident at any of the specific tissue ages. Because mineralization is expected to be dependent on free calcium and phosphate as well as the ratio of mineralization inhibitors and promoters amongst others, the lack of dependency of the relative proteoglycan and lipids content (normalized to organic matrix content) on menopause status may contribute to the lack of differences in mineral content between the pre- and postmenopausal biopsies in the present study.

The spectroscopically determined mineral maturity/crystallinity is a metric describing the chemical makeup of the apatite crystallites and by extrapolation their size/shape.[28, 29, 57, 58] Normal bone crystallites exhibit a range of sizes, and deviations from this range have been encountered in cases of fragile bone.[22, 59, 60] In the present study, mineral maturity/crystallinity was independent of menopause status at any of the three distinct anatomical areas considered, whereas it was significantly dependent on tissue age at active bone forming trabecular surfaces. It was also inversely correlated in the PostM group, exclusively, with bone formation rate and activation frequency at active bone forming trabecular surfaces, suggesting that the kinetics of mineral crystallite maturation is altered postmenopause. Crystals of higher maturity/crystallinity have been associated with compromised bone properties and osteoporosis,[22, 59, 61-63] yet it should be kept in mind that the subjects of the present study were healthy with no increased fracture incidence, implying that this bone quality metric, although important, may not be sufficient by itself to govern bone's resistance to fracture.

A distinct feature of type I collagen in mineralized tissues is its cross-linking chemistry and molecular packing structure,[64] which provides the fibrillar matrices with mechanical properties such as tensile strength and viscoelasticity. The importance of collagen intermolecular cross-links on the mechanical performance of bone is apparent in the pyridoxine-deficient chick animal model,[65] as well as in lathyrism.[32, 66] Pyridinoline is a mature, nonreducible, trivalent collagen cross-link abundant in mineralizing type I collagen.[64] Published reports based on biochemical analysis of homogenized tissue indicate that normal bone has a range of pyridinoline content that reaches a plateau at the age of 10 to 15 years, deviations from which are associated with fragile bone.[67-69] In the present study, the relative pyridinoline content at actively forming trabecular surfaces significantly increased as a function of tissue age, whereas no differences between PreM and PostM specimens were evident at any of the anatomical locations investigated. It was also inversely correlated with bone formation rate and osteoblast surface at actively forming surfaces in the PostM group, in agreement with the expected dependency on elevated remodeling rates.[69-71]

In the present study, anatomical areas for analysis were chosen based on tissue age, so as to minimize the effects of bone turnover on the monitored parameters. With the exception of relative lipids content at active bone forming surfaces, this was the case in the PreM group, as suggested by the lack of correlations between Raman-derived and histomorphometry parameters. On the other hand, significant correlations were evident in the PostM group, suggesting that menopause exerts effects on both cell numbers and productivity/function, providing clues as to how the changes in remodeling rates postmenopause affect bone quality, independent of BMD.

In summary, previous studies of a larger cohort from which the biopsies of the present study were randomly selected indicated that bone remodeling rates double at menopause,[6] accompanied by deterioration of trabecular bone structure,[5] yet no significant differences in intrinsic properties determined by nanoindentation were observed between healthy PreM and PostM subjects.[3] The results of the present study indicate that menopause affects the organic matrix at bone active (forming or resorbing but not quiescent) surfaces. There were no differences in either mineral content or mineral maturity/crystallinity at any of the micro-anatomical areas analyzed, suggesting that organic matrix is exclusively affected shortly after menopause. Consideration of these changes in combination with changes in bone turnover rates may offer clues concerning the mechanisms underlying these alterations. Understanding changes attributable to aging milestones such as menopause will enable better discrimination between changes in bone quality owing to normal aging and heterogeneous diseases such as osteoporosis.


All authors state that they have no conflicts of interest.


This study was supported by the Allgemeine Unfallversicherungsanstalt (AUVA), research funds of the Austrian workers compensation board, and the Wiener Gebietskrankenkasse (WGKK), Viennese sickness insurance funds.

Authors' roles: SG and WB were responsible for the performance of the Raman analysis. WB and EPP were responsible for the statistical analysis of data. RR was responsible for the design of the clinical aspects of the study (identification of subjects, obtaining the biopsies, obtaining clinical and histomorphometry data), and drafting of the manuscript. EPP and KK were responsible for the design of the spectroscopic analysis study, evaluation of data, and drafting of the manuscript.