Abstract

Significant effort continues to be made to understand whether differences exist in the structural, compositional, and mechanical properties of cortical bone subjected to different strain modes or magnitudes. We evaluated juvenile sheep femora (age = 4 months) from the anterior and posterior quadrants at three points along the diaphysis as a model system for variability in loading. Micro-CT scans (50 micron) were used to measure cortical thickness and mineral density. Three point bending tests were performed to measure the flexural modulus, strength, and post-yield displacement. There was no difference in cortical thickness or density between anterior or posterior quadrants; however, density was consistently higher in the middle diaphysis. Interestingly, bending modulus and strength were higher in anterior quadrants compared to posterior quadrants. Together, our results suggest that there is a differential spatial response of bone in terms of elastic bending modulus and mechanical strength. The origins of this difference may lie within the variation in ongoing mineralization, in combination with the collagen-rich plexiform structure, and whether this is related to strain mode remains to be explored. These data suggest that in young ovine cortical bone, modulation of strength occurs via potentially complex interactions of both mineral and collagen-components that may be different in regions of bone exposed to variable amounts of strain. Further work is needed to confirm the physiological load state of bone during growth to better elucidate the degree to which these variations are a function of the local mechanical environment.

1 Introduction

Osteoporosis and osteopenia are skeletal conditions characterized by a systematic loss in bone quality and increase in bone fragility. One option for improving bone health is via exercise in which the adaptation to mechanical stimuli can be exploited to encourage bone formation [13]. The development of exercise interventions for bone health would benefit from a better understanding of how bone adapts to specific strain modes.

Mechanical loading incurred from movement is known to be critical for the development of robust weight-bearing bones. For example, strains in the femoral neck during walking are maximum inferiorly and smaller superiorly [4]. Accordingly, the inferior cortex is thicker than the superior cortex [5], suggesting a correspondence between mechanical strain and bone development. Numerous in vivo studies in a range of mammalian species have investigated the magnitude and frequency of loading that results in bone adaptation under physiological and experimental conditions [2,611].

While the magnitude of mechanical loading has been shown to affect bone adaptation, the specific mode of loading (e.g., tensile or compressive) is also important to the process of bone growth. Some studies report increased porosity, mineral content, and new remodeling events in regions of bone exposed to compressive strains compared to tensile strains [1215]. However, others report higher remodeling, yield strain, ultimate strain, and Young's modulus in tensile regions compared to compressive [13,14,16]. Still others found no difference in the mechanical, structural, or compositional properties of cortical bone in regions subjected to tensile and compressive loads [17,18]. Therefore, whether bone growth occurs differently in regions of tension or compression remains unclear.

One limitation of previous work is the lack of characterization of both structural and material properties. Further, in those studies that performed mechanical testing, either tensile or compressive testing was performed, but not both. Strain mode (tension versus compression) is an important factor in the mechanical evaluation of bone because cortical bone fails by different mechanisms when loaded in tension versus compression [19]. Moreover, most studies chose a single cross section of bone for analysis, thus, the degree of spatial heterogeneity in bone properties along the diaphysis—and its sensitivity to mechanical loading—remains unclear.

Consequently, our aim was to provide a more comprehensive understanding of how cortical bone growth is affected by the local strain environment using sheep femur as a model system. The orientation of the femur is a result of the hip and knee flexion angles, which are more flexed in large quadrupeds than in bipedal mammals. Biomechanical models suggest that the combination of joint and muscle forces results in a state of bending with the anterior region in tension and the posterior region in compression [15,20,21]. While this remains to be definitely confirmed, it is likely that sheep femora are under a state of complex loading including both compression and bending resulting in a variable strain state in different anatomical quadrants. Thus, our aims were to (1) evaluate whether bone properties differ between two distinct anatomical regions (anterior versus posterior) and (2) evaluate if bone growth is heterogeneous along the diaphysis. The overarching goal was to elucidate differences in bone growth in response to local mechanical loading and thus inform the development of targeted exercise interventions designed to reduce fracture risk.

2 Materials and Methods

Right femoral bone samples (n = 5) from juvenile sheep were used for all analyses. To summarize, intact samples were first micro-CT scanned to analyze cortical thickness and density along the diaphysis. All image processing was performed semi-automatically using custom matlab code (Mathworks). Next, three subregions along the diaphysis were defined from which parallelopiped samples were machined and mechanically tested in three-point bending. Details of all analyses are described in the remaining sections.

2.1 Subjects.

Five juvenile Hampshire sheep were housed in a large indoor pen at the University of Illinois Sheep and Beef Cattle facility. All sheep were able to move freely while in the pen and access to food and water was not limited. Subjects were humanely euthanized at the age of 120 days, limbs were dissected, muscles and ligaments were removed, and the remaining bone segments were frozen at −20 °C.

2.2 Bone Imaging and Image Processing.

Intact sheep femora were thawed, and microcomputed tomography (micro-CT) data were acquired (Inveon, Siemens, Princeton, NJ) at an isotropic 50 μm voxel size (80 kVp, 20 cm field of view, 1472 × 1472 matrix size). Three hydroxyapatite phantoms (density = 25, 100, 500 mg/cm3, model 092, CIRS, Norfolk, VA) were included in the scans to convert Hounsield units to hydroxyapatite density.

The diaphysis was defined as the middle 40% of total bone length and ten equally spaced transverse slices were then identified along the diaphysis (Fig. 1(b)). The diaphysis was sub-divided into three regions by thirds resulting in a proximal, middle, and distal diaphyseal regions per sample with corresponding transverse slices used to quantify thickness and density within each region. The choice of ten transverse slices was motivated by our observation that the variability within a given region was low (see Supplemental Figure S1A available in the Supplemental Materials on the ASME Digital Collection). When we quantified the median thickness using all slices within a longitudinal cross section of a representative sample (see Supplemental Figure S1B available in the Supplemental Materials) and compared it to our calculations of median thickness using three slices we found that the error was an order of magnitude lower than our sample-sample variability. Therefore, the ten transverse slices were assigned to one of the three diaphyseal regions: the proximal three slices for the proximal diaphysis, four slices in the middle, and the remaining three slices were in the distal diaphysis.

Fig. 1
(a) Micro-CT images at ten locations along the length of the diaphysis were used to quantify cortical thickness and density in the anterior and posterior quadrants (lower image). (b) For each femur, the diaphysis was separated into three regions: proximal, middle, and distal. (c) Two parallelopiped samples were machined from each quadrant and region of the diaphysis. (d) Transverse image of a sample scanned at 5 μm voxel size with evidence of circumferential bone along the periosteal border. (e) Test configurations used for each anatomical quadrant in which the periosteal border was either in compression or tension.
Fig. 1
(a) Micro-CT images at ten locations along the length of the diaphysis were used to quantify cortical thickness and density in the anterior and posterior quadrants (lower image). (b) For each femur, the diaphysis was separated into three regions: proximal, middle, and distal. (c) Two parallelopiped samples were machined from each quadrant and region of the diaphysis. (d) Transverse image of a sample scanned at 5 μm voxel size with evidence of circumferential bone along the periosteal border. (e) Test configurations used for each anatomical quadrant in which the periosteal border was either in compression or tension.
Close modal

Each CT image was individually binarized using a global threshold based on Otsu's method (Fig. 1(a)). Next, the centroid of the transverse slice was calculated, and voxels along the periosteal border were identified. The border was discretized into 360 segments radiating outward from the centroid of the image with the origin of a polar coordinate system set at the centroid and 0 deg positioned posteriorly with increasing angles counterclockwise. The posterior quadrant was defined as the region of the cross section  315 deg and  45 deg. The anterior quadrant was defined as the region of the cross section  135 deg and  225 deg. (Fig. 1(a), bottom).

2.3 Thickness and Density Measurements.

Cortical thickness was calculated using the binarized images and defined as the distance between the periosteal and endosteal border within each anatomical quadrant of interest and in each longitudinal region of the diaphysis. For example, within the slices corresponding to proximal diaphysis (the proximal three slices), the median of all thickness measurements in the anterior quadrant was calculated. This process was repeated for the posterior quadrant of the proximal diaphysis, and similarly for the middle and distal regions of the diaphysis where the middle four slices were used for the mid-diaphysis and distal three slices used for the distal diaphysis.

The binarized images were used to segment the micro-CT data for measurement of mineral density within each quadrant and region along the diaphysis. Subject specific hydroxyapatite calibration curves were used to convert micro-CT Hounsfield units to tissue mineral density. The average density within the slices corresponding to the anatomical quadrant and region of interest was calculated.

2.4 Sample Preparation.

The location of the central 40% of the femoral diaphysis was measured on each bone sample and marked. The regions proximal (femoral head and epiphysis) and distal (epiphysis and condyles) to the diaphysis were removed using a diamond band saw (Model C-40, Gryphon, Sylmar, CA). Next, the diaphysis was potted within custom three-dimensional-printed fixtures using Bondo and then secured within the bed of a 5 Axis CNC machine (Model HY6040, ChinaCNCzone, Shenzhen, China). Two parallelopiped full-thickness samples (20 mm × 5 mm × cortical thickness) were milled from each anatomical quadrant (Fig. 1(c)) using titanium coated 1-mm end mills (12,000 rpm, 50% feed rate).

2.5 Mechanical Testing.

All samples were tested in three-point bending. In a small imaging pilot study, a subset of the parallelopiped samples were scanned at 5 μm voxel size (10 W and 50 kV, Model Micro-XCT-400, Zeiss Xradia, Oberkochen, Germany). We observed a consistent ring of circumferential cortical bone on the periosteal surface of all samples (Fig. 1(d)). Two configurations of testing were therefore used to account for whether the circumferential bone would influence our mechanical test results: strain mode specific (SMS) and nonstrain mode specific (xSMS).

The purpose of the SMS condition was to test samples in the expected physiological loading condition. For the anterior quadrant, this corresponds to the periosteal side in tension while samples from the posterior quadrant would have the periosteal side in compression (Fig. 1(e)). The nonstrain mode specific condition is the reverse: the periosteal side would be in compression for anterior samples and tension for posterior samples.

Each bone sample was submerged in PBS for at least 15 min prior to testing to ensure it was hydrated. The sample was placed on support pins (diameter = 4 mm) of a three-point-bend test fixture (span = 15 mm) for use in a mechanical test machine (MTS criterion 43, Instron, Norwood, MA). Each sample was loaded to failure at a rate of 0.01 mm/s in displacement control and force was measured with a 1 kN load cell. Displacement was measured using the cross-head displacement of the test machine. The test was ended if either the sample failed or when the displacement limit of 2 mm was reached.

Post-processing of mechanical test data were performed in matlab (Mathworks R2019b, Natick, MA). The linear region of the load–displacement curve was identified using an iterative correlation in which the start and end points (up to maximum force) were selected such that the strength of the linear regression between these points was maximized (threshold: r = 0.999). For each sample, the bending modulus (EB) and strength (SB) was calculated using linear elastic beam theory
EB=m*Ls34*w*Th3
(1)
SB=3*F*Ls2*w*Th2
(2)

where m is the slope from the linear region of the load–displacement curve, F is the maximum load during mechanical testing, Ls refers to the span length (15 mm), and w and Th are the width (5 mm) and the thickness of the specimen, respectively. Post-yield displacement was defined as the displacement occurring between yield (end of linear region) and maximum force.

2.6 Statistical Analysis.

Normality of data was tested using a Shapiro–Wilk test with a significance level of 0.05. Since some data were not normally distributed, summary data are presented as medians and median absolute deviations. The effect of fixed variables (region and quadrant) was evaluated using a linear mixed model with subject included as a random variable. Posthoc analyses were then performed using a paired samples Wilcoxon Test (significance level = 0.05) accounting for samples from the same bone and a Bonferonni correction was applied to account for multiple-comparisons. All analyses were performed in R (v3.6.2).

3 Results

3.1 Cortical Thickness and Density.

We successfully machined pairs of parallelopipeds from the anterior and posterior quadrants along the three regions of the diaphysis (n = 12 total/bone). Of the parallelopipeds machined from the same anatomical quadrant, there was no difference in either cortical thickness or density between the pairs based on the micro-CT analyses (see Supplemental Table S1 available in the Supplemental Materials on the ASME Digital Collection). We therefore pooled the samples within each anatomical quadrant to evaluate cortical thickness and density between quadrants and in regions along the diaphysis.

Table 1

Cortical thickness and density in each region along the diaphysis

Thickness (mm)
RegionQuadrantMedian (MAD)Regions (p-value)Quadrant (p-value)
Proximal (P)Anterior2.37 (0.48)0.25AntP v M0.69
Posterior2.51 (0.24)P v D0.35
Middle (M)Anterior2.42 (0.25)0.25M v D0.02
Posterior2.57 (0.28)PostP v M0.69
Distal (D)Anterior2.15 (0.29)0.69P v D0.08
Posterior2.35 (0.29)M v D0.08
Density (mg HA/cm3)
RegionQuadrantMedian (MAD)Regions (p-value)Quadrant (p-value)
Proximal (P)Anterior902.37 (41.20)0.07AntP v M0.02
Posterior879.89 (52.50)P v D0.77
Middle (M)Anterior979.32 (12.98)0.07M v D0.02
Posterior921.45 (49.76)PostP v M0.02
Distal (D)Anterior897.92 (21.07)0.70P v D0.72
Posterior868.30 (64.75)M v D0.02
Thickness (mm)
RegionQuadrantMedian (MAD)Regions (p-value)Quadrant (p-value)
Proximal (P)Anterior2.37 (0.48)0.25AntP v M0.69
Posterior2.51 (0.24)P v D0.35
Middle (M)Anterior2.42 (0.25)0.25M v D0.02
Posterior2.57 (0.28)PostP v M0.69
Distal (D)Anterior2.15 (0.29)0.69P v D0.08
Posterior2.35 (0.29)M v D0.08
Density (mg HA/cm3)
RegionQuadrantMedian (MAD)Regions (p-value)Quadrant (p-value)
Proximal (P)Anterior902.37 (41.20)0.07AntP v M0.02
Posterior879.89 (52.50)P v D0.77
Middle (M)Anterior979.32 (12.98)0.07M v D0.02
Posterior921.45 (49.76)PostP v M0.02
Distal (D)Anterior897.92 (21.07)0.70P v D0.72
Posterior868.30 (64.75)M v D0.02

Within each region, thickness and density were compared between anterior and posterior quadrants (n = 10/quadrant). Differences along the diaphyseal regions, within the same quadrant (e.g., proximal anterior versus middle anterior) were also compared. Significant p-values (p < 0.05) are indicated in bold font. Density in the middle anterior diaphysis was higher than the proximal and distal diaphysis. MAD = median absolute deviation.

Cortical thickness within the anterior diaphysis was 2.39 ± 0.41 mm and 2.48 ± 0.26 mm in the posterior diaphysis (Fig. 2(a)). The linear mixed models revealed that thickness was sensitive to region (p = 0.019) but not quadrant (p = 0.78). Accordingly, there were no significant differences in anterior versus posterior cortical thickness (p-range= 0.25–0.69, Table 1) within any given region of the diaphysis. Within the anterior quadrant, cortical thickness in the distal diaphysis was significantly thinner than the mid-diaphysis (−11%, p = 0.02).

Fig. 2
Cortical thickness (a) and density (b) in the anterior (blue) and posterior (orange) anatomical quadrants along different regions of the diaphysis. Circles and squares indicate samples from SMS and xSMS conditions, respectively. Cortical thickness was uniform while density in the anterior quadrant was highest in the mid-diaphysis. Within each row (proximal, middle, distal) that spans both subfigures, posterior data is printed above anterior data. ∗p < 0.05.
Fig. 2
Cortical thickness (a) and density (b) in the anterior (blue) and posterior (orange) anatomical quadrants along different regions of the diaphysis. Circles and squares indicate samples from SMS and xSMS conditions, respectively. Cortical thickness was uniform while density in the anterior quadrant was highest in the mid-diaphysis. Within each row (proximal, middle, distal) that spans both subfigures, posterior data is printed above anterior data. ∗p < 0.05.
Close modal

In contrast to thickness, quadrant was a strong predictor of density (p = 0.00014) though when analyzed within a given region there were only moderate differences in density with the anterior quadrant more dense than the posterior quadrant in the proximal (2.5%, p = 0.07) and middle diaphysis (6.3%, p = 0.07, Fig. 2(b)). Mineral density was consistently higher in the mid-diaphysis compared to the proximal and distal diaphysis (p = 0.02, Table 1) with the greatest increases occurring in the anterior quadrant (8.8%) compared to the posterior quadrant (5.4%).

3.2 Bending Modulus, Strength, and Plastic Deformation.

The bending modulus within each anatomical quadrant was insensitive to whether the periosteal side was in compression or tension (SMS versus xSMS, see Supplemental Table S2 available in the Supplemental Materials on the ASME Digital Collection). Both quadrant (p< 0.0001) and the distal region (p = 0.009) were significant predictors of bending modulus based on the linear regression model. Overall, the bending modulus in the anterior quadrants were higher than the posterior quadrants (Fig. 3(d), Table 2) with significant increases in the proximal (47%, p = 0.04) diaphysis. Within a given quadrant, there were no significant differences along the diaphysis.

Fig. 3
Force-displacement curves (a)–(c), bending modulus (d), bending strength (e), and post-yield displacement (f) in the anterior (blue) and posterior (orange) anatomical quadrants along different regions of the diaphysis. Circles and squares indicate samples from SMS and xSMS conditions, respectively. Post-yield displacement was uniform whereas there was heterogeneity in modulus and strength. In subfigures (a)–(c), data for the anterior quadrant is represented by a dashed line, while data for the posterior quadrant is shown in a solid line. Within each row (proximal, middle, distal) that spans subfigures (d)–(f), posterior data is printed above anterior data. ∗p < 0.05, ∗∗p < 0.01.
Fig. 3
Force-displacement curves (a)–(c), bending modulus (d), bending strength (e), and post-yield displacement (f) in the anterior (blue) and posterior (orange) anatomical quadrants along different regions of the diaphysis. Circles and squares indicate samples from SMS and xSMS conditions, respectively. Post-yield displacement was uniform whereas there was heterogeneity in modulus and strength. In subfigures (a)–(c), data for the anterior quadrant is represented by a dashed line, while data for the posterior quadrant is shown in a solid line. Within each row (proximal, middle, distal) that spans subfigures (d)–(f), posterior data is printed above anterior data. ∗p < 0.05, ∗∗p < 0.01.
Close modal
Table 2

Comparisons of mechanical test results: modulus, strength, and post-yield displacement between quadrants and along different regions of the diaphysis

Modulus (GPa)
RegionQuadrantMedian (MAD)Quadrant (p-value)Regions (p-value)
Proximal (P)Anterior4.89 (0.63)0.04AnteriorP v M0.29
Posterior3.33 (0.85)P v D0.19
Middle (M)Anterior3.94 (1.11)0.19M v D0.48
Posterior2.79 (0.76)PosteriorP v M0.48
Distal (D)Anterior3.60 (0.87)0.19P v D0.49
Posterior2.95 (0.52)M v D0.48
Strength (MPa)
RegionQuadrantMedian (MAD)Quadrant (p-value)Regions (p-value)
Proximal (P)Anterior153.73 (17.81)0.01AnteriorP v M0.87
Posterior102.67 (4.77)P v D0.07
Middle (M)Anterior151.32 (12.55)0.01M v D0.04
Posterior93.61 (8.37)PosteriorP v M0.07
Distal (D)Anterior124.85 (16.28)0.07P v D0.93
Posterior103.05 (15.21)M v D0.11
Post-yield Displacement (mm)
RegionQuadrantMedian (MAD)Quadrant (p-value)Regions (p-value)
Proximal (P)Anterior0.21 (0.08)0.48AnteriorP v M0.39
Posterior0.35 (0.29)P v D0.48
Middle (M)Anterior0.14 (1.5 × 10−4)0.29M v D0.17
Posterior0.27 (0.19)PosteriorP v M0.55
Distal (D)Anterior0.23 (0.11)0.55P v D0.99
Posterior0.30 (0.11)M v D0.48
Modulus (GPa)
RegionQuadrantMedian (MAD)Quadrant (p-value)Regions (p-value)
Proximal (P)Anterior4.89 (0.63)0.04AnteriorP v M0.29
Posterior3.33 (0.85)P v D0.19
Middle (M)Anterior3.94 (1.11)0.19M v D0.48
Posterior2.79 (0.76)PosteriorP v M0.48
Distal (D)Anterior3.60 (0.87)0.19P v D0.49
Posterior2.95 (0.52)M v D0.48
Strength (MPa)
RegionQuadrantMedian (MAD)Quadrant (p-value)Regions (p-value)
Proximal (P)Anterior153.73 (17.81)0.01AnteriorP v M0.87
Posterior102.67 (4.77)P v D0.07
Middle (M)Anterior151.32 (12.55)0.01M v D0.04
Posterior93.61 (8.37)PosteriorP v M0.07
Distal (D)Anterior124.85 (16.28)0.07P v D0.93
Posterior103.05 (15.21)M v D0.11
Post-yield Displacement (mm)
RegionQuadrantMedian (MAD)Quadrant (p-value)Regions (p-value)
Proximal (P)Anterior0.21 (0.08)0.48AnteriorP v M0.39
Posterior0.35 (0.29)P v D0.48
Middle (M)Anterior0.14 (1.5 × 10−4)0.29M v D0.17
Posterior0.27 (0.19)PosteriorP v M0.55
Distal (D)Anterior0.23 (0.11)0.55P v D0.99
Posterior0.30 (0.11)M v D0.48

Significant p-values (p < 0.05) are indicated in bold font. MAD, median absolute deviation.

Differences in bending strength of the anterior and posterior quadrant were more pronounced (Fig. 3(e)) and were reflected in the linear model prediction of strength (p< 0.0001). Strength in the anterior quadrant of was 62% higher (p = 0.01) than the posterior quadrant in the mid-diaphysis and 50% higher in the proximal diaphysis (p = 0.01). When comparing strength along the longitudinal axis of the diaphysis, the largest differences were in the anterior quadrant with strength in middle region 21% higher than the distal region (p = 0.04). Within the posterior quadrant, there were no significant differences in bending strength across regions.

Finally, the amount of post-yield displacement were greater in the posterior than anterior quadrant though there were no statistically significant differences (p-range = 0.29–0.55). There were no significant differences along the diaphysis within either the anterior or posterior quadrant.

4 Discussion

In this study, we have used high-resolution imaging and mechanical test data to evaluate the degree of spatial heterogeneity in young cortical bone in different anatomical compartments: anterior versus posterior. Our data showed no difference in cortical thickness in regions of the diaphysis that likely experience variability in strain magnitudes and potentially strain mode, suggesting relatively uniform bone formation around the femoral diaphysis during early growth in sheep. Thicker cortex has been reported in regions of compression (compared to tension) in the deer calcaneus [19] and rat tibia [22], while others have shown thicker cortex in the tensile regions of sheep radii [23] and equine metacarpus [13]. One explanation for the difference may be due to age. The studies summarized here were performed in skeletally mature bone, in contrast to our analysis of bone in early stages of postnatal development. It may be that during ontogeny, adaptations to increasing mass or speed (or a combination thereof) lead to regional variations. A study of juvenile (4–5 month old) sheep radius found disproportionately thicker cortex in tensile regions compared to newborns [24]. While we do not have data at birth to confirm if results at the radius would translate to the femur, our data suggests that in early growth, sheep femora begin with uniform cortical thickness at the cross-sectional level though we did observe thinner cortices at the distal diaphysis.

In contrast to thickness, our comparison of mineral density revealed increased mineralization in bone in the anterior versus posterior compartment (which approached significant levels) with the greatest difference occurring in the mid-diaphysis. Again, our data is in contrast to mineral content reported in deer calcaneus [19] and human tibia [25] but in agreement with data from the equine metacarpus [13]. When compared longitudinally, the anterior mid-diaphysis was at least 8% more mineralized than the proximal or distal diaphysis, and similar but lower differences (5.4% more mineralized) were found in the posterior quadrant.

Our observations of spatial variation in bone properties begs the question of what might be the underlying stimulus leading to this variability. During movement, long bone deformation causes pressure gradients to drive fluid flow within bone and has been hypothesized to flow from regions of compression to regions of tension [26]. The resulting strains sensed by the osteocytes tethered to the bone matrix are 1–2 orders of magnitude greater than the tissue level strains generated by muscles [2629]. Fluid flow velocity can also vary between the opposing cortices of the same bone and has been shown to affect bone formation with new bone associated with low velocities that occurred in tensile regions [22].

However, whether osteocytes in tensile regions are more activated than in compressive regions continues to be debated: Skedros and colleagues [30] reported significantly more remodeling events in the tensile cortex compared to the compressive cortex while the use of the tibia loading model in mice has shown decreased levels of sclerostin and correspondingly increased bone area in tensile regions [31]. Nguyen and Barak reported a larger Haversian system in the tensile cortex suggesting increased remodeling, but there were no difference in mineral content and compressive stiffness between opposing cortices [32]. While the underlying mechanism is unclear, our data indicate that for young ovine bone, mineral density was higher in the anterior region which we hypothesize to be under tension (though statistical significance was limited by our sample size) and was greatest overall in the mid-diaphysis. Several studies have suggested that the ovine femur is in a state of bending [15,20,21], but whether this is true or if the strains are in fact maximum in the mid-diaphysis is not known. It is likely that the femur is under a set of combined loads including bending, axial loading, and potentially other modes.

Indeed, it has been proposed that bone under habitual bending would result in regional adaptations whereas bone under torsion (resulting in shear strains) would not. In reality, the complex set of time-varying muscle and joint forces that bone is subjected to results in a wide range of strain magnitudes and modes. The distributions of tension, compression, and shear based on neutral axis rotations during the midportion of the stance phase has been used to categorize bone based on anticipated load complexity with the femur falling within the high complexity category and supported by several human and animal studies (summarized in Ref. [33] and also discussed in Refs. [34,35]). Unfortunately, this has not been confirmed in the ovine femur and, to our knowledge, no in vivo or in vitro strain data of the ovine femur is available in the literature. However, a musculoskeletal model of the ovine hind limb has been developed by others that would be useful to better understand the loading conditions of femur and is the subject of future work by our own group. Finally, it is also worth noting that the mid-diaphysis is the location of the primary ossification center, where mineralization first occurs, and thus may also explain why the mid-diaphysis is more mineralized than other regions.

We next turned to investigating how these changes in thickness and density manifested in differences in mechanical properties—a novel component of our study. We found that the elastic bending modulus was higher in the anterior compartment versus the posterior compartment. Moreover, these differences were amplified with respect to strength where bending strength was up to 62% higher in the tensile quadrant and is likely attributable to the increase in mineralization. To our knowledge, two other studies have tested cortical bone samples from both tension and compression regions of a bone [16,17]). While Skedros et al. reported higher flexural modulus, yield and ultimate strain in samples from tensile regions of the equine third metacarpal (MC3) bone, Cuppone et al. reported no difference between the tensile and compressive quadrants of the midshaft of the human femur. These contrasting results may be due to the variation in load complexity between the two bones.

Interestingly, while mineralization at the mid-diaphysis was greater than the proximal diaphysis, there was no difference in proximal or middle levels of strength in the tensile quadrants. Moreover, the posterior quadrant was moderately lower in strength at the mid-diaphysis. Thus, while the role of mineralization corresponded to differences in strength in the opposing cortices, the degree of spatial heterogeneity within a given quadrant along the length of the diaphysis was not fully explained by variation in mineralization levels. One reason may be related to the presence of plexiform bone which is known to occur in rapidly growing animals, including sheep. In a study of goat femur, tensile regions were shown to have plexiform bone whereas compressive regions had more Haversian bone [15].

Plexiform bone serves as an initial collagen-rich template for subsequent modeling and remodeling activity in which the plexiform bone is eventually replaced by Haversian bone. Hence, the role of collagen in young bone may be another important determinant of strength. The collagen fibers in the equine third metacarpal cortex exposed to tensile loading were more longitudinal compared to the compressive lateral cortex [13]. Similarly, two separate studies on equine radius also demonstrated longitudinal collagen fiber orientation in the tensile cortex and oblique/transverse collagen in the compression cortex [36,37]. While the relationship between collagen fiber orientation and bone strength seems to be undisputed, the mechanism of how it affects bone strength is not fully understood.

Our study has several noteworthy contributions. To our knowledge, this work is the first to investigate structure, composition, and material properties of cortical bone in both physiological (SMS) and nonphysiological loading (xSMS) conditions. Our findings demonstrated that bone in the anterior quadrant with convex curvature is associated with increased strength regardless of testing mode (SMS versus xSMS). Moreover, we have demonstrated the degree of spatial heterogeneity along the diaphysis highlighting the potential for young growing bone to serve as a model for understanding natural modeling events in response to varying strain environments. Our data suggests that there is an increased response of bone in the anterior cortex of young sheep via changes in mineralization and this may therefore serve as a biomechanical target for exercise studies aiming to improve bone properties when young. However, our work is not without limitations. Our sample size is somewhat small and limited to a single time point of age. It remains to be demonstrated the degree to which these trends would be sustained during ontogeny. The lack of complete correspondence between mineralization levels and strength indicate a need to evaluate other metrics that may affect bone strength, namely, structural variations in plexiform versus Haversian bone and the role of collagen alignment. Moreover, further work is required to confirm the strains on the ovine femur during locomotion to better establish the load state.

In summary, we have shown that the structural, compositional and material properties of cortical bone are affected differently in relation to the local mechanical environment. Furthermore, bone growth is spatially heterogeneous as seen from results along the longitudinal axis of the diaphysis and in comparing opposing cortices. Hence, this study will aid in developing a comprehensive understanding of young cortical bone growth in response to the local mechanical strain environment. Based on our experimental findings, cortical ovine bone is stronger in the anterior cortex than the opposing posterior region, and that this difference in strength is likely a consequence of the combined mineral density and collagen interactions associated with variability in the local mechanical environments. These results suggest that bone's modeling and remodeling processes can alter local bone strength which could be useful in exercise studies designed to diminish fracture risk.

Conflict of Interest

None.

Acknowledgment

This study was funded by the National Science Foundation (NSF-1638756). We are grateful for the technical support from David Ehrhardt from Advanced Materials Testing and Evaluation Laboratory at UIUC, and Dr. Leilei Yin and Travis Ross for their help with micro-CT scanning and post-processing technical support at the Beckman Institute, UIUC. Thanks also to Mark Pinson, Gary Sedberry and Kyle Cheek from the Mechanical Science and Engineering machine shop for their assistance. Portions of figures were generated using Biorender.

Funding Data

  • National Science Foundation (Grant No. NSF-1638756; Funder ID: 10.13039/100000001).

References

1.
Hammer
,
A.
,
2015
, “
The Paradox of Wolff's Theories
,”
Irish J. Med. Sci.
,
184
(
1
), pp.
13
22
.10.1007/s11845-014-1070-y
2.
Robling
,
A. G.
,
Duijvelaar
,
K. M.
,
Geevers
,
J. V.
,
Ohashi
,
N.
, and
Turner
,
C. H.
,
2001
, “
Modulation of Appositional and Longitudinal Bone Growth in the Rat Ulna by Applied Static and Dynamic Force
,”
Bone
,
29
(
2
), pp.
105
113
.10.1016/S8756-3282(01)00488-4
3.
Frost
,
H. M.
,
1987
, “
Bone “Mass” and the “Mechanostat”: A Proposal
,”
Anat. Rec.
,
219
(
1
), pp.
1
9
.10.1002/ar.1092190104
4.
Kersh
,
M. E.
,
Martelli
,
S.
,
Zebaze
,
R.
,
Seeman
,
E.
, and
Pandy
,
M. G.
,
2018
, “
Mechanical Loading of the Femoral Neck in Human Locomotion
,”
J. Bone Miner. Res.
,
33
(
11
), pp.
1999
2006
.10.1002/jbmr.3529
5.
Kersh
,
M. E.
,
Pandy
,
M. G.
,
Bui
,
Q. M.
,
Jones
,
A. C.
,
Arns
,
C. H.
,
Knackstedt
,
M. A.
,
Seeman
,
E.
, and
Zebaze
,
R. M.
,
2013
, “
The Heterogeneity in Femoral Neck Structure and Strength
,”
J. Bone Miner. Res.
,
28
(
5
), pp.
1022
1028
.10.1002/jbmr.1827
6.
Warden
,
S. J.
,
Mantila Roosa
,
S. M.
,
Kersh
,
M. E.
,
Hurd
,
A. L.
,
Fleisig
,
G. S.
,
Pandy
,
M. G.
, and
Fuchs
,
R. K.
,
2014
, “
Physical Activity When Young Provides Lifelong Benefits to Cortical Bone Size and Strength in Men
,”
Proc. Natl. Acad. Sci. USA
,
111
(
14
), pp.
5337
5342
.10.1073/pnas.1321605111
7.
Turner
,
C. H.
, and
Burr
,
D. B.
,
1993
, “
Basic Biomechanical Measurements of Bone: A Tutorial
,”
Bone
,
14
(
4
), pp.
595
608
.10.1016/8756-3282(93)90081-K
8.
Torrance
,
A. G.
,
Mosley
,
J. R.
,
Suswillo
,
R. F.
, and
Lanyon
,
L. E.
,
1994
, “
Noninvasive Loading of the Rat Ulna In Vivo Induces a Strain-Related Modeling Response Uncomplicated by Trauma or Periostal Pressure
,”
Calcif. Tissue Int.
,
54
(
3
), pp.
241
247
.10.1007/BF00301686
9.
Weatherholt
,
A. M.
,
Fuchs
,
R. K.
, and
Warden
,
S. J.
,
2013
, “
Cortical and Trabecular Bone Adaptation to Incremental Load Magnitudes Using the Mouse Tibial Axial Compression Loading Model
,”
Bone
,
52
(
1
), pp.
372
379
.10.1016/j.bone.2012.10.026
10.
Lieberman
,
D. E.
,
Pearson
,
O. M.
,
Polk
,
J. D.
,
Demes
,
B.
, and
Crompton
,
A. W.
,
2003
, “
Optimization of Bone Growth and Remodeling in Response to Loading in Tapered Mammalian Limbs
,”
J. Exp. Biol.
,
206
(
18
), pp.
3125
3138
.10.1242/jeb.00514
11.
Main
,
R. P.
,
2007
, “
Ontogenetic Relationships Between In Vivo Strain Environment, Bone His- Tomorphometry and Growth in the Goat Radius
,”
J. Anatomy
,
210
(
3
), pp.
272
293
.10.1111/j.1469-7580.2007.00696.x
12.
Lanyon
,
L. E.
, and
Baggott
,
D. G.
,
1976
, “
Mechanical Function as an Influence on the Structure and Form of Bone
,”
J. Bone Jt. Surg., Ser. B
,
58-B
(
4
), pp.
436
443
.10.1302/0301-620X.58B4.1018029
13.
Skedros
,
J. G.
,
Mason
,
M. W.
,
Nelson
,
M. C.
, and
Bloebaum
,
R. D.
,
1996
, “
Evidence of Structural and Material Adaptation to Specific Strain Features in Cortical Bone
,”
Anat. Rec.
,
246
(
1
), pp.
47
63
.10.1002/(SICI)1097-0185(199609)246:1<47::AID-AR6>3.0.CO;2-C
14.
Skedros
,
J. G.
,
Su
,
S. C.
, and
Bloebaum
,
R. D.
,
1997
, “
Biomechanical Implications of Mineral Con- Tent and Microstructural Variations in Cortical Bone of Horse, Elk, and Sheep Calcanei
,”
Anat. Rec.
,
249
(
3
), pp.
297
316
.10.1002/(SICI)1097-0185(199711)249:3<297::AID-AR1>3.0.CO;2-S
15.
Mayya
,
A.
,
Banerjee
,
A.
, and
Rajesh
,
R.
,
2013
, “
Mammalian Cortical Bone in Tension is Non-Haversian
,”
Sci. Rep.
,
3
(
1
), pp.
1
6
.10.1038/srep02533
16.
Skedros
,
J. G.
,
Dayton
,
M. R.
,
Sybrowsky
,
C. L.
,
Bloebaum
,
R. D.
, and
Bachus
,
K. N.
,
2006
, “
The Influence of Collagen Fiber Orientation and Other Histocompositional Characteristics on the Mechanical Properties of Equine Cortical Bone
,”
J. Exp. Biol.
,
209
(
15
), pp.
3025
3042
.10.1242/jeb.02304
17.
Cuppone
,
M.
,
Seedhom
,
B. B.
,
Berry
,
E.
, and
Ostell
,
A. E.
,
2004
, “
The Longitudinal Young's Modulus of Cortical Bone in the Midshaft of Human Femur and Its Correlation With CT Scanning Data
,”
Calcif. Tissue Int.
,
74
(
3
), pp.
302
309
.10.1007/s00223-002-2123-1
18.
Wallace
,
I. J.
,
Demes
,
B.
,
Mongle
,
C.
,
Pearson
,
O. M.
,
Polk
,
J. D.
, and
Lieberman
,
D. E.
,
2014
, “
Exercise-Induced Bone Formation is Poorly Linked to Local Strain Magnitude in the Sheep Tibia
,”
PLoS ONE
,
9
(
6
), pp.
e99108
5
.10.1371/journal.pone.0099108
19.
Jg
,
S.
,
Rd
,
B.
,
Mw
,
M.
, and
Dm
,
B.
,
1994
, “
Analysis of a Tension/Compression Skeletal System: Pos- Sible Strain-Specific Differences in the Hierarchical Organization of Bone
,”
Anat. Record
,
239
(
4
), pp.
396
404
.10.1002/ar.1092390406
20.
Rubin
,
C. T.
,
1984
, “
Skeletal Strain and the Functional Significance of Bone Architecture
,”
Calcif. Tissue Int.
,
36
(
S1
), pp.
S11
S18
.10.1007/BF02406128
21.
Shahar
,
R.
,
Banks-Sills
,
L.
, and
Eliasy
,
R.
,
2003
, “
Stress and Strain Distribution in the Intact Canine Femur: Finite Element Analysis
,”
Med. Eng. Phys.
,
25
(
5
), pp.
387
395
.10.1016/S1350-4533(03)00002-X
22.
Knothe Tate
,
M. L.
,
Steck
,
R.
,
Forwood
,
M. R.
, and
Niederer
,
P.
,
2000
, “
In Vivo Demonstration of Load- Induced Fluid Flow in the Rat Tibia and Its Potential Implications for Processes Associated With Functional Adaptation
,”
J. Exp. Biol.
,
203
(
18
), pp.
2737
2745
.10.1242/jeb.203.18.2737
23.
Lanyon
,
L. E.
,
Magee
,
P. T.
, and
Baggott
,
D. G.
,
1979
, “
The Relationship of Functional Stress and Strain to the Processes of Bone Remodelling. An Experimental Study on the Sheep Radius
,”
J. Biomech.
,
12
(
8
), pp.
593
600
.10.1016/0021-9290(79)90079-4
24.
Doutre
,
M. S.
,
Skedros
,
J. G.
,
Brown
,
E. B.
,
Skedros
,
G. A.
,
Mears
,
C. S.
, and
Bloebaum
,
R. D.
,
2016
, “ORS 2016 Annual Meeting Poster No. 1550: Temporal Changes in Sagittal Curvature and the Emergence of Load Predictability in the Sheep Radius,”
ORS 2016 Annual Meeting
, Vol.
225
, Orlando, FL, Mar. 5–8.https://teambone.com/wp-content/uploads/2021/05/119.-Doutre-et-al.-2016-Temporal-changes-in-sagittal-curvature-and-the-emergence-of-loadpredictability.pdf
25.
Lai
,
Y. M.
,
Qin
,
L.
,
Hung
,
V. W.
, and
Chan
,
K. M.
,
2005
, “
Regional Differences in Cortical Bone Mineral Density in the Weight-Bearing Long Bone Shaft—a pQCT Study
,”
Bone
,
36
(
3
), pp.
465
471
.10.1016/j.bone.2004.11.005
26.
Han
,
Y.
,
Cowin
,
S. C.
,
Schaffler
,
M. B.
, and
Weinbaum
,
S.
,
2004
, “
Mechanotransduction and Strain Amplification in Osteocyte Cell Processes
,”
Proc. Natl. Acad. Sci.
,
101
(
47
), pp.
16689
16694
.10.1073/pnas.0407429101
27.
Robling
,
A. G.
, and
Bonewald
,
L. F.
,
2020
, “
The Osteocyte: New Insights
,”
Annu. Rev. Phys.
,
82
(
1
), pp.
485
506
.10.1146/annurev-physiol-021119-034332
28.
Smalt
,
R.
,
Mitchell
,
F. T.
,
Howard
,
R. L.
, and
Chambers
,
T. J.
,
1997
, “
Induction of NO and Prostaglandin E2 in Osteoblasts by Wall-Shear Stress but Not Mechanical Strain
,”
Am. Can. J. Physiol. Endocrinol. Metab.
,
273
(
4
), pp.
E751
E758
.10.1152/ajpendo.1997.273.4.E751
29.
Thi
,
M. M.
,
Suadicani
,
S. O.
,
Schaffler
,
M. B.
,
Weinbaum
,
S.
, and
Spray
,
D. C.
,
2013
, “
Mechanosen- Sory Responses of Osteocytes to Physiological Forces Occur Along Processes and Not Cell Body and Require αvβ3 Integrin
,”
Proc. Natl. Acad. Sci. USA
,
110
(
52
), pp.
21012
21017
.10.1073/pnas.1321210110
30.
Skedros
,
J. G.
,
Mason
,
M. W.
, and
Bloebaum
,
R. D.
,
2001
, “
Modeling and Remodeling in a Developing Artiodactyl Calcaneus: A Model for Evaluating Frost's Mechanostat Hypothesis and Its Corollaries
,”
Anat. Rec.
,
263
(
2
), pp.
167
185
.10.1002/ar.1094
31.
Moustafa
,
A.
,
Sugiyama
,
T.
,
Prasad
,
J.
,
Zaman
,
G.
,
Gross
,
T.
,
Lanyon
,
L. E.
, and
Price
,
J.
,
2012
, “
Mechanical Loading-Related Changes in Osteocyte Sclerostin Expression in Mice Are More Closely Associated With the Subsequent Osteogenic Response Than the Peak Strains Engen- Dered
,”
Osteoporosis Int.
,
23
(
4
), pp.
1225
1234
.10.1007/s00198-011-1656-4
32.
Nguyen
,
J. T.
, and
Barak
,
M. M.
,
2020
, “
Secondary Osteon Structural Heterogeneity Between the Cranial and Caudal Cortices of the Proximal Humerus in White-Tailed Deer
,”
J. Exp. Biol.
,
223
(
11
), p.
jeb225482
.10.1242/jeb.225482
33.
Skedros
,
J. G.
,
2011
, “Chapter 7:
Interpreting Load History in Limb-Bone Diaphyses: Important Considerations and Their Biomechanical Foundations
,”
Bone Histology: An Anthropological Perspect.
, 1st ed., CRC Press, Boca Raton, FL, pp.
153
220
.10.1201/b11393
34.
Daegling
,
D. J.
,
Bhramdat
,
H. D.
, and
Toro-Ibacache
,
V.
,
2021
, “
Efficacy of Shear Strain Gradients as an Osteogenic Stimulus
,”
J. Theor. Biol.
,
524
, p.
110730
.10.1016/j.jtbi.2021.110730
35.
Keenan
,
K. E.
,
Mears
,
C. S.
, and
Skedros
,
J. G.
,
2017
, “
Utility of Osteon Circularity for Determining Species and Interpreting Load History in Primates and Nonprimates
,”
Am. J. Phys. Anthropol.
,
162
(
4
), pp.
657
681
.10.1002/ajpa.23154
36.
Riggs
,
C. M.
,
Lanyon
,
L. E.
, and
Boyde
,
A.
,
1993
, “
Functional Associations Between Collagen Fibre Orientation and Locomotor Strain Direction in Cortical Bone of the Equine Radius
,”
Anat. Embryol.
,
187
(
3
), pp.
231
238
.10.1007/BF00195760
37.
Mason
,
M. W.
,
Skedros
,
J. G.
, and
Bloebaum
,
R. D.
,
1995
, “
Evidence of Strain-Mode-Related Cortical Adaptation in the Diaphysis of the Horse Radius
,”
Bone
,
17
(
3
), pp.
229
237
.10.1016/8756-3282(95)00213-W

Supplementary data