Quantity and Quality of Trabecular Bone in the Femur Are Enhanced by a Strongly Anabolic, Noninvasive Mechanical Intervention


  • Clinton Rubin Ph.D.,

    Corresponding author
    1. Musculo-Skeletal Research Laboratory, Department of Biomedical Engineering, State University of New York, Stony Brook, New York, USA
    • Department of Biomedical Engineering Psychology-A, Third Floor State University of New York Stony Brook, NY 11794-2580, USA
    Search for more papers by this author
    • Dr. Clinton Rubin and Dr. Kenneth McLeod serve as consultants. Dr. A. Simon Turner receives some salary support from Exogen. Dr. Ralph Müller, Dr. Wei Lin, Dr. Yi-Xian Qin, and Dr. Erik Mittra do not have any conflict of interest

  • A. Simon Turner,

    1. Department of Clinical Sciences, Colorado State University, Ft. Collins, Colorado, USA
    Search for more papers by this author
    • Dr. Clinton Rubin and Dr. Kenneth McLeod serve as consultants. Dr. A. Simon Turner receives some salary support from Exogen. Dr. Ralph Müller, Dr. Wei Lin, Dr. Yi-Xian Qin, and Dr. Erik Mittra do not have any conflict of interest

  • Ralph Müller,

    1. Orthopedic Biomechanics Laboratory, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts, USA
    Search for more papers by this author
    • Dr. Clinton Rubin and Dr. Kenneth McLeod serve as consultants. Dr. A. Simon Turner receives some salary support from Exogen. Dr. Ralph Müller, Dr. Wei Lin, Dr. Yi-Xian Qin, and Dr. Erik Mittra do not have any conflict of interest

  • Erik Mittra,

    1. Musculo-Skeletal Research Laboratory, Department of Biomedical Engineering, State University of New York, Stony Brook, New York, USA
    Search for more papers by this author
    • Dr. Clinton Rubin and Dr. Kenneth McLeod serve as consultants. Dr. A. Simon Turner receives some salary support from Exogen. Dr. Ralph Müller, Dr. Wei Lin, Dr. Yi-Xian Qin, and Dr. Erik Mittra do not have any conflict of interest

  • Kenneth McLeod,

    1. Musculo-Skeletal Research Laboratory, Department of Biomedical Engineering, State University of New York, Stony Brook, New York, USA
    Search for more papers by this author
    • Dr. Clinton Rubin and Dr. Kenneth McLeod serve as consultants. Dr. A. Simon Turner receives some salary support from Exogen. Dr. Ralph Müller, Dr. Wei Lin, Dr. Yi-Xian Qin, and Dr. Erik Mittra do not have any conflict of interest

  • Wei Lin,

    1. Musculo-Skeletal Research Laboratory, Department of Biomedical Engineering, State University of New York, Stony Brook, New York, USA
    Search for more papers by this author
    • Dr. Clinton Rubin and Dr. Kenneth McLeod serve as consultants. Dr. A. Simon Turner receives some salary support from Exogen. Dr. Ralph Müller, Dr. Wei Lin, Dr. Yi-Xian Qin, and Dr. Erik Mittra do not have any conflict of interest

  • Yi-Xian Qin

    1. Musculo-Skeletal Research Laboratory, Department of Biomedical Engineering, State University of New York, Stony Brook, New York, USA
    Search for more papers by this author
    • Dr. Clinton Rubin and Dr. Kenneth McLeod serve as consultants. Dr. A. Simon Turner receives some salary support from Exogen. Dr. Ralph Müller, Dr. Wei Lin, Dr. Yi-Xian Qin, and Dr. Erik Mittra do not have any conflict of interest


The skeleton's sensitivity to mechanical stimuli represents a critical determinant of bone mass and morphology. We have proposed that the extremely low level (<10 microstrain), high frequency (20-50 Hz) mechanical strains, continually present during even subtle activities such as standing are as important to defining the skeleton as the larger strains typically associated with vigorous activity (>2000 microstrain). If these low-level strains are indeed anabolic, then this sensitivity could serve as the basis for a biomechanically based intervention for osteoporosis. To evaluate this hypothesis, the hindlimbs of adult female sheep were stimulated for 20 minutes/day using a noninvasive 0.3g vertical oscillation sufficient to induce approximately 5 microstrain on the cortex of the tibia. After 1 year of stimulation, the physical properties of 10-mm cubes of trabecular bone from the distal femoral condyle of experimental animals (n = 8) were compared with controls (n = 9), as evaluated using microcomputed tomography (μCT) scanning and materials testing. Bone mineral content (BMC) was 10.6% greater (p < 0.05), and the trabecular number (Tb.N) was 8.3% higher in the experimental animals (p < 0.01), and trabecular spacing decreased by 11.3% (p < 0.01), indicating that bone quantity was increased both by the creation of new trabeculae and the thickening of existing trabeculae. The trabecular bone pattern factor (TBPf) decreased 24.2% (p < 0.03), indicating trabecular morphology adapting from rod shape to plate shape. Significant increases in stiffness and strength were observed in the longitudinal direction (12.1% and 26.7%, respectively; both, p < 0.05), indicating that the adaptation occurred primarily in the plane of weightbearing. These results show that extremely low level mechanical stimuli improve both the quantity and the quality of trabecular bone. That these deformations are several orders of magnitude below those peak strains which arise during vigorous activity indicates that this biomechanically based signal may serve as an effective intervention for osteoporosis.


Osteoporosis, a disease characterized by the progressive loss of bone tissue, particularly in the weight-supporting areas of the skeleton, is one of the most common complications of aging.(1) Although the bone tissue that remains in osteoporotic individuals is normal and capable of repair, the effective strength of the skeleton is compromised by the overall reduction in bone mass.(2) Although pharmaceutical treatment protocols are systemic by nature, manifestations of the disease (fractures) are focal (primarily wrist, hip, and spine). Further, each osteoporosis therapy approved by the Food and Drug Administration (FDA) works by inhibiting bone resorption, and, thus, retention of bone density is achieved by suppressing erosion to the same extent that aging or menopause has suppressed formation.(3) In such circumstances, although bone density may be maintained (retained quantity), the structural attributes of the bone may be jeopardized (reduced quality).(4) Clearly, therapies that increase bone formation are highly desirable, particularly if they concurrently improve bone quality, but unfortunately such interventions are not nearly as well developed. In this study, the potential of a unique mechanically based intervention, shown to be strongly anabolic,(5–9) is evaluated in terms of its ability to influence both the quantity and the quality of bone.

The ability of bone to rapidly accommodate changes in its mechanical environment ensures that sufficient skeletal mass is appropriately placed to withstand the rigors of functional activity.(10) The tissue's sensitivity to functional loading suggests that mechanical stimuli should be capable of providing a site-specific, nonpharmacologic intervention for the inhibition and/or reversal of bone loss. Consistent with the anabolic potential of mechanical stimuli, extensive, long-term exercise has been shown to increase skeletal mass.(11, 12) However, the efficacy of lengthy bouts of strenuous exercise as a countermeasure for osteopenia remain uncertain,(13) are complicated by compliance considerations, and could potentiate the very fractures the intervention is intended to inhibit. If the anabolic constituents of load bearing could be distilled, the attributes of a mechanically based intervention to build bone mass could exploit the “form-follows-function” interdependence of the skeleton:

  • Mechanical strains arise in bone tissue from functional load bearing(14)

  • Removal of load bearing suppresses anabolic activity in the skeleton(15)

  • If a mechanical signal is anabolic, only brief exposure maximizes the response(16)

  • The stimulus is self-localizing (strains are greatest where loads are high and bone mass is low)

  • The stimulus is self-optimizing (strains diminish as bone mass is increased)(17)

  • Osteogenic mechanical stimuli ultimately produce lamellar bone(18)

Given the structural responsibilities of the skeleton, it has been presumed that the bone tissue is primarily responsive to the large strains it must withstand.(19) In an effort to fully characterize the load history of bone, recent work has shown that the spectrum of strain within functionally loaded bones can be characterized as having a power-law relationship between the magnitude of strain events and the frequency with which these events occur.(20) If the functional strain environment of bone has a power-law (1/f) distribution, then it is reasonable to assume that the bone tissue depends as much on the persistent, barrage of tens of thousands of low magnitude strain events (<10 microstrain) that arise through postural muscle activity as it does on the relatively large (>1000 microstrain), rarely occurring strain events.(21) From this perspective, it can be argued that the bone wasting that occurs in aging, bedrest, or space flight may result not only from diminished load-bearing responsibility, but the fiber-specific sarcopenia that parallels it.(22, 23) In support of this hypothesis, as the frequency of the mechanical stimulus increases from 1 to 60 Hz, the strain necessary to maintain cortical bone decreases over an order of magnitude, from 1000 microstrain to <100 microstrain.(24, 25)

Ultimately, if bone's anabolic response to mechanical stimuli can contribute to combating skeletal disorders such as osteoporosis, it is essential to show that the bone structure is also enhanced by the intervention, as reflected by improvements in both quantity and quality of bone. Here, high-resolution microcomputed tomography (μCT) and materials testing were used to determine the quantity and quality of trabecular bone in the distal femur of the sheep after 1 year of brief daily exposure to a noninvasive mechanical stimulus.


Experimental design

All procedures were reviewed and approved by both the Stony Brook University and Colorado State University animal use committees and met or exceeded all federal guidelines for consideration of animal welfare. Eighteen adult female sheep (Warhill, intact ewes, 60-80 kg) 6-8 years of age, were randomized into two groups: (1) experimental and (2) untreated controls. For 20 minutes/day, 5 days/week, the experimental sheep stood constrained in a chute such that only the hindlimbs were subject to a vertical ground-based vibration, induced by a small plate oscillating at 30 Hz, to create peak-peak accelerations of 2.9 m/s2, referred to as a fraction of earth's gravitational field, 0.3g (1g = 9.8 m/s2). The resonant-based platform was controlled by an accelerometer such that the control signal would modulate the drive output—regardless of shifting by the animal—to provide the appropriate acceleration (Fig. 1).(26) When the animals were not being treated, they joined the control animals to freely roam a pasture area. After 1 year of stimulation, the animals were killed by a saturated barbiturate and the femora and radii removed. With careful consideration given to the anatomic orientation of the femur (Fig. 2), 1-cm cubes were harvested from the medial condyle of the left femur using a water-cooled diamond wafer saw. The ultra-distal radius and proximal femur were evaluated using histomorphometry, the results of which are reported elsewhere.(7, 8)

Figure FIG. 1..

Five days per week, the sheep lined up for their 20-minute mechanical session (left). As each animal stood constrained in a chute, only the hindlimbs were subject to the oscillatory stimulus (right). The front limbs remained untreated and served as an intra-animal control. A resonant-based platform generated a 30-Hz, 0.3g acceleration, engendered peaks of approximately 5 microstrain on the diaphyseal shaft of the tibia, as measured by strain gauges in calibration animals.

Figure FIG. 2..

The position and orientation of each cube of trabecular bone harvested from the medial condyle was carefully registered relative to the anatomy of the femur (left). Then, each 10-mm cube was measured using a microtomographic imaging system (μCT 20; Scanco Medical). A total of 300 microtomographic slices (512 × 512 pixels) were acquired for each control (middle) and experimental (right) specimen. Nine of the 300 slices for each cube are shown (8.3 × 8.3 mm2), spanning the height of the cube. Data were stored in 3D image arrays with a 34-μm3 voxel size. A 3D Gaussian filter was used to partly suppress the noise in the volumes. All samples were binarized using the same parameters for the filter width (1.2 voxels), the filter support (2 voxels), and the threshold (10.2% of the maximal attenuation). Some indication of differences in density and Tb.Th are visible, even in these CT slices.

Bone strain measurements

At the beginning of the protocol, the oscillating device was calibrated by measuring strain from the tibia of two calibration animals that were the same age, gender, and breed as the animals used in the long-term study. Under a general halothane anesthesia using aseptic conditions, three three-element stacked rosette strain gauges (2-mm-gauge length, 120 Ω, FRA-2-11; TML Gauges, Kenkyujo, Japan) were attached to the left tibial midshaft, a site selected because of the ease in which the three surfaces of the tibia can be exposed with only a very minimum of musculoskeletal disruption.(27) Exposure of the tibial surface is prepared by removing a small (50 mm2) area of the periosteum, drying the exposed surface with anhydrous diethyl ether, and gluing the gauge to the surface with isobutyl 2-cyanoacrylate monomer (Ethicon, Ltd., Somerville, NJ, USA). Within 3 h of surgery, the animals were awake and fully ambulatory and were immediately capable of unrestricted activity.

Each 120-Ω gauge was conditioned with a 3-V bridge excitation voltage and amplified with the Vishay Measurements group model 2100 system (Raleigh, NC, USA). A bridge amplifier gain of 1300× was used, and the amplified analog strain signals were filtered through low-pass anti-aliasing filters (Keithley Metrabyte AAF-8; Keithley Inc., Cleveland, OH, USA) with a cut-off frequency of 50 Hz and digitized with 16-bit resolution (National Instruments AT-MIO-16X; National Instruments, Austin, TX, USA) at a sampling rate of 102.4 Hz. In this configuration, resolution of the system is better than 0.1 microstrain (με), yet strains in the range of ±2500 με also can be recorded. The filters and A/D board occupied three I/O slots of a 486 DX2 (66 MHz) PC. Data were stored in binary files on the PC's hard disk, using 37 kilobytes for each of the nine channels of strain signal. The strain caused by the oscillation of the mechanical device was digitized in real time and transferred to workstation storage (IBM RISC 6000) for analysis.

Microtomographic imaging

To generate three-dimensional (3D) models of the trabecular samples from the femur, the cubes were measured using a compact, fan-beam-type desktop μCT system (μCT 20; Scanco Medical, Zurich, Switzerland), also referred to as desktop μCT.(28) For each bone cube, a total of 300 microtomographic slices (512 × 512 pixels) were acquired. Data were stored in 3D image arrays with a 34-μm3 voxel size. A constrained 3D Gaussian filter was used to partly suppress the noise in the volumes. Bone tissue was segmented from marrow and soft tissue using a global thresholding procedure.(29) All samples were binarized using the same parameters for the filter width (1.2 voxels), the filter support,(2) and the threshold (10.2% of the maximal attenuation). To ensure that the data were not influenced by edge artifacts due to cube preparation, only a centered 8.3-mm cube, automatically positioned within the 10-mm cube, was considered. The operator who measured the samples was blinded with respect to the grouping. Samples were numbered consecutively and the analysis was performed automatically without any interference by the operator.

In addition to the visual assessment of structural images, morphometric indices were determined from the microtomographic data sets using direct 3D morphometry(30) rather than the traditional plate model approach. Previous studies have shown structural metrics measured using 3D morphometry and μCT to correlate closely with those measured using standard histomorphometry.(31, 32) The volume of the trabeculae (bone volume [BV]) was calculated using tetrahedrons corresponding to the enclosed volume of the triangulated surface used for surface area calculation. From this measure, BV density (BV/TV) was calculated, in which the total volume (TV) is the volume of the whole sample. Additionally, bone mineral content (BMC) was estimated from BV assuming a constant tissue density.(33)

Mean trabecular thickness (Tb.Th) then was calculated directly by determining a local thickness at each voxel representing bone. The same method was used to calculate the mean trabecular separation (Tb.Sp) by applying the thickness calculation to the nonbone parts of the 3D image. Thus, the separation was the thickness of the marrow cavities. Trabecular number (Tb.N) was calculated as the inverse distance between the midsection of the plates. For this purpose, the inverse of the mean distance between the midaxes of the structure was used to calculate the mean number of elements per unit length.

In addition to the computation of direct metric parameters, a number of nonmetric parameters were calculated to describe the 3D nature of the trabecular bone samples. Trabecular bone pattern factor (TBPf) in its 3D implementation was calculated, representing a measure of the amount of concave and convex structures and therefore a means of approximating how the structural elements of a composite structure are designed.(34) The geometrical degree of anisotropy (DA) of a structure, typically defined as the ratio between the maximal and minimal radii of the mean intercept length (MIL) ellipsoid, that is, the ratio of the maximal and minimal eigen value of the MIL tensor, was calculated by superimposing parallel test lines in different directions on the 3D image. The directional MIL was the total length of the test lines in one direction divided by the number of intersections with the bone-marrow interface of the test lines in the same direction. The MIL ellipsoid was calculated by fitting the directional MIL to a directed ellipsoid using a least square fit.(30, 34)

Materials testing

On completion of the tomographic analysis, the stiffness and strength was measured for each cube of bone. Effective modulus testing was performed in three orthogonal directions, testing the bone in the longitudinal, anteroposterior (AP), and mediolateral (ML) planes (during weight bearing, the femur of sheep are approximately 40° off the vertical axis; evaluation of the cubes was done such that the longitudinal plane was considered perpendicular to the ground, Fig. 2). To reduce any end-artifact, which might otherwise be introduced by platen loading, the compressive force was applied through a single point onto an articulating platen, thus allowing any imperfections in the parallel-planed cube to be accounted for.(35, 36)

All loading was performed at room temperature with a closed-loop servohydraulics test system (MTS System w/STAR-II controller; MTS, Eden Prairie, MN, USA). The load train was centered to within 0.025 mm,(37) and stiffness tests were performed under closed-loop feedback load control (N), and strength was determined under displacement control. Deformations were measured using an linear variable differential transformer (LVDT), using a strain rate for all tests controlled at 0.005/s. Displacement and load data were digitally sampled at 1000 Hz and filtered at 50 Hz. Using cube samples from the same age and gender of sheep (calibration animals), the compressive failure strength of the 10-mm cubes was determined to occur with loads between 2400N and 4500N, depending on the orientation. To remain well away from the yield and ultimate strength of the bone during determination of stiffness, the load limit for the stiffness measurements was set at 600N. Compressive loads were performed in each of the three anatomic directions and repeated three times. After the stiffness data were collected, the cube was loaded to failure in the longitudinal direction. Data are reported as failure strength or the maximum stress the cube supported before load was reduced. As with the μCT scanning, the stiffness and strength tests were performed without knowledge of whether the specimen was harvested from an experimental or control animal.


Comparison of groups was performed using an unpaired t-test with statistical significance considered at p < 0.05.


In the 1-year protocol, one experimental animal was lost for reasons unrelated to the intervention, leaving nine control and eight experimental animals. Over the course of the study, no significant changes in animal mass occurred in either the experimental animals or controls, and there were no significant differences measured between groups (experimental animals, 71.1 ± 7.1 kg; controls, 70.3 ± 9.4 kg). Strain gauges attached to the tibia of animals used to calibrate the device showed that the 0.3g, 30-Hz signal generated peak-to-peak strains of approximately 5 microstrain, one-thousandth of the strain magnitude necessary to cause yield failure in bone.(38) Although these strain signals were small relative to peak strains generated during activities such as postural sway, they were readily detectable (Fig. 3). There were no significant differences between control and experimental animals in any parameter of density (dual-energy X-ray absorptiometry [DXA]) or static and dynamic histomorphometry of the bone examined in the radius (DXA and histomorphometry data(8)).

Figure FIG. 3..

Strain gauge readings from the cortical surface of a sheep tibia show that the strains induced by the 30 Hz, 0.3g stimulation are on the order of 5 με, well below the levels induced even by postural sway (broad amplitude and slower domain shifts in the gauge recording). The 5 microstrain peak-to-peak strain signals are several orders of magnitude below those strains that initiate yield damage in the bone tissue. Although transmissibility of a ground-based signal at 30 Hz is very high,(39) it is still necessary to extrapolate from the cortical surface of a bone to estimate the strains on the trabeculae. Although we do not know the strain environment on these trabeculae subject to 30-Hz, 0.3g conditions, we are confident that they are much lower than those strain magnitudes generated during walking, which exceed 1000 με on the cortical surface.

BV for the cubes of trabecular bone harvested from the medial femoral condyle of the experimental animals was 359.3 mm3 (±29.6 mm3), 10.2% higher than that seen in the controls (326.1 ± 46.7 mm3; p < 0.05). Accordingly, BMC, calculated directly from BV, also was 10.2% greater in animals exposed to the mechanical stimulus (Table 1), with the experimental animals at 0.61 mm3 ± 0.05 g, and the controls at 0.55 mm3 ± 0.08 g (p < 0.05). Because the TVs for each of the cubes were identical, the volume fraction or BV/TV increased in an identical fashion, with the experimental specimens at 54.5 ± 4.5%, 10.2% greater than the controls (49.1 ± 7.1%; p < 0.05). These numbers serve to illustrate that the quantity of bone was greater in animals exposed to this stimulus (Fig. 4).

Table Table 1. Morphometric Property Measurements of the Medial Condyle of the Femur After 1 Year of 30 Hz Mechanical Stimulation at 0.3g, for 20 Minutes/Day
original image
Figure FIG. 4..

Three-dimensional reconstructions are possible by stacking the 300 slices, permitting visualization of the effective structural properties of both the control (left) and experimental (right) specimens. These images are supportive of the quantifiable changes in both density and architecture.

Tb.N in the experimental animals was 8.3% higher (p < 0.01) in the experimental animals, whereas trabecular spacing was decreased by 11.3% (p < 0.01). The TBPf decreased 24.2% (p < 0.03), and the eigen values of the MIL tensor were significantly different in the longitudinal direction (−7.8%; p < 0.01). Stiffness was significantly greater in the longitudinal direction (12.1%; p < 0.05), and failure strength in the longitudinal direction was 26.7% greater in the experimental animals (p < 0.05; Fig. 5). Control versus experimental comparisons of strength and stiffness in the ML or AP directions showed no significant differences. These data indicate that the quality of bone, preferentially oriented in a given direction, was greater in the experimental animals than the controls.

Figure FIG. 5..

Physical property measurements of the bone cubes showed significant differences between the control and experimental animals in both stiffness (12.1% increase, left) and failure strength (26.7% increase, right); p < 0.05 in both cases. These data indicate that the strongly anabolic nature of the mechanical signal, as shown by histomorphometry, which leads to the increases in Tb.N and BV as shown by tomography, ultimately serve to enhance both the stiffness and the strength of the bone, and thus improves both the quantity and quality of bone.


One-year exposure to extremely low level mechanical signals increased the quantity of bone in the distal femur. Considering the relatively little change in DXA values in control animals measured over the course of the year and the strong anabolic nature of these signals as measured in the experimental animals,(7, 8) we are confident that the observed differences between control and experimental animals are achieved primarily through augmentation of trabecular architecture over the course of the year rather than the presence of the signal serving to inhibit bone loss (which was not occurring, even in these older animals). That the directional nature of the response was observed only in the longitudinal direction also implies that the adaptive response works primarily through stimulating new bone formation rather than inhibiting bone loss, although the antiresorptive attributes of this low-level stimulation also are relevant to the prevention of osteoporosis.(9) This interpretation is supported further by the absence of any significant differences in the radius of the experimental animals, which was not subject to the mechanical stimulus(8) and indicates that the mechanical intervention influences only that bone tissue subject to the signal rather than serving as a systemically based therapy.

The mechanical signal was induced in the vertical direction via a ground-based vibration. Although it is difficult to determine what strains were engendered in the trabeculae in the distal femoral condyle, strains measured at the tibial diaphysis were approximately 5 microstrain or more than two orders of magnitude below those strains generated in the tibia during walking.(14, 16, 27) Certainly, that we do not know the actual strain in the trabeculae of the distal femur and that we must assume some level of transmissibility from the ground-based signal to the femur and from the cortical shell to the trabecular bone are limitations of this study. In studies performed on humans, accelerometers attached to Steinmann pins screwed into the hip and spine of volunteers show that approximately 80% of a 30-Hz ground-based vibration is transmitted to the femur and spine of the standing human.(39) Therefore, we are confident that a great deal of the mechanical information successfully reached the trabeculae within the femur, and it is this mechanical signal (or a byproduct thereof, such as fluid flow or electric potentials) that the bone is adapting to. Currently, we are working on finite element models of the bone cubes to better approximate the strain environment that is generated with these high-frequency loads.

Considering that these signals are so small relative to the peak strains that normally arise during locomotion,(20) it is important to consider the physical means by which such low-level mechanical signals might stimulate the skeletal system to adapt. One means by which dynamic strain could influence the cell population would be through a frequency-dependent perturbation of intramedullary pressure(25) and the resultant intracortical fluid flow within the bone.(40, 41) Using an avian ulna model of disuse osteopenia, it was shown that when loaded to the same degree, the intramedullary pressure rose monotonically with increasing loading frequency in the range from 0.1 to 100 Hz. Indeed, in the axially loaded mode, the load generated pressure at 30 Hz was 10× higher than that measured at 0.1 Hz.(42, 43) This is indicative of a physical mechanism whereby the extremely low level signals generated by muscle may have a disproportionate influence on bone mass and morphology. In addition, a frequency-selective responsiveness of the cell's cytoskeleton, with actin polymerization being more responsive to nanometer level deformations induced at 15 Hz than those induced at 1.5 Hz or 150 Hz,(44) also implies an enhanced sensitivity of musculoskeletal tissues (as well as other systems) to these higher frequency domains. Together, these data indicate that the osteogenic potential of mechanical signals is derived, at least in part, by byproducts of matrix deformation, rather than the strain of the tissue directly.

Our data show that the increase in bone quantity was achieved by increasing the number of trabeculae (indications that trabeculae were created de novo) as well as by increasing the thickness of existing trabeculae. To our knowledge, this is the first intervention (biomechanical or biochemical) that results in the formation of new trabeculae in the adult animal. This interpretation is supported by histomorphometric evaluation of the proximal femur,(7, 8) which shows both thickening of existing trabeculae as well as an increase in the number of trabeculae. Although it is remarkable that new trabeculae may be stimulated in adult bone by this intervention, considering that trabeculae are formed through enchondral ossification during processes such as fracture healing,(45) it may well be an attribute of the mechanical loading that makes this observation so unusual. These data also show that the new trabeculae serve to enhance the stiffness and strength of the trabecular bone and, thus, imply an improvement in the quality of bone. Importantly, changes in stiffness and strength were significant only in the direction in which the loads were applied and the eigen values of the MIL tensor also were different primarily in this direction, which indicates adaptation to a specific, focal stimulus. In this sense, these data suggest that quantity and quality of bone are enhanced by mechanical intervention, while the directionality of the site-specific adaptation indicates that the bone is being selectively deposited to augment specific mechanical parameters.

TBPf was used in this study to indicate how the morphology of the trabecular elements are adapting in response to the low-level, high-frequency loads. TBPf is measured as the ratio of the change in surface and the change in volume resulting from the dilatation of the structure.(46) In that sense, TBPf indicates the number of concave structures (plates) versus the number of convex structures (rods). This ratio is used also to calculate the structure model index (SMI), which describes how plate- or rodlike a structure is. Although BV density (i.e., bone quantity) correlated well with bone stiffness and strength, strong correlations also were evident with TBPf and Tb.Th (i.e., bone quality), implying that improved structure is achieved not only by the amount of bone, but how the bone is placed (Table 2).

Table Table 2. Coefficient of Correlations (r Values) Between Architectural and Structural Properties in the Longitudinal Direction
original image

The data presented here make it clear that extremely low level mechanical stimuli, when induced at frequencies similar to the contractibility spectra of muscle, are strongly anabolic. If these low-level, high-frequency stimuli are critical determinants of bone mass and morphology, it becomes important to determine the origin of these dynamic strains and whether age somehow alters these strain dynamics, and, thus, the etiology of the bone loss. Muscle mass is well known to decrease with age,(47) paralleled by a dramatic change in the dynamics of the muscle action.(48) With age, the spectral content of muscle dynamics also changes, as shown by recording surface accelerations associated with postural muscle activity of the soleus,(23) a principal muscle associated with standing.(49) Spectra obtained from these recordings show muscle activity in the frequency range above 20 Hz decreases by a factor of 3 in the elderly as compared with that seen in young adults, a response consistent with loss of fast oxidative-type fibers. Although speculative, it appears that the high-frequency components seen in bone strain arise through muscle activity; thus, the deterioration of the postural muscle contraction spectra with age would therefore contribute to a decrease in the strain spectral content above 20 Hz.

From this perspective, it can be argued that the sarcopenia that parallels aging is a principal etiologic factor in osteoporosis, because this portion of the strain spectra is demonstrably osteogenic. Although the association between loss of muscle dynamics and bone mass does not show a causal relationship, it provides support for the concept that the chronic demands that gravity places on postural muscle activity may be a dominant force in the positive control of bone mass.(50) This hypothesis is consistent with observations made on skeletal muscle in rats flown in space. For example, in one study, 7 days exposure to weightlessness resulted in a 23% decrease in soleus muscle mass as compared with age and weight-matched controls remaining on earth, versus an 11% decrease in the mass of the extensor digitorum longus.(22) Because the soleus is predominantly a postural muscle (with a characteristic firing frequency of 20-25 Hz),(51) there appears to be general congruence between microgravity and ground-based observations. This leads to a highly speculative hypothesis; if aging also leads to loss of specific muscle fibers with firing rates in the range of >20 Hz and this frequency domain is critical to the regulation of the musculoskeletal system, osteoporosis could be inhibited by providing a “surrogate” such as this low-level mechanical stimulus for the lost spectral strain history. In support of this perspective, the high-frequency, ground-based vibration, which doubled bone formation rates in the tibia of the adult rat, also served to normalize bone formation activity in animals subject to 23 hours and 50 minutes per day of disuse; something that 10 minutes of normal weight bearing failed to achieve.(9)

In summary, low-level mechanical stimuli, several orders of magnitude below that which arises during vigorous functional activity, are strongly anabolic. Over a 1-year period, this anabolic signal stimulates an increase in both the quantity and the quality of bone. The fact that this signal is native to the bone tissue and the adaptive response is not only site specific but also directional supports the development of a nonpharmacologically based therapy for prevention and/or reversal of bone loss. The fact that a noninvasive stimulus generated via a small ground-based oscillation effectively increases BV fraction, Tb.N, thickness, stiffness, and strength, all in a sight well removed from where the ground-based stimuli are introduced, suggests that this intervention may provide a unique strategy for the treatment of metabolic bone disorders such as osteoporosis. Preliminary trials evaluating the efficacy of this intervention to inhibit bone loss in a postmenopausal population have begun, and the early results are encouraging.(5)


The authors are grateful to Jack Ryaby and Roger Talish for their help, and to Exogen, Inc. for supplying the mechanical devices. They also are grateful to the many veterinary students who spent countless hours with the sheep during the daily loading protocols. This study was funded by the National Institutes of Health (NIH) AR 43498.