Regional Variations of HR-pQCT Morphological and Biomechanical Measurements of Bone Segments and Their Associations With Whole Distal Radius and Tibia Mechanical Properties

Osteoporosis is a major bone disease characterized by reduced bone mass, compromised bone strength, and increased risk of fracture. In clinical practice, measurement of areal bone mineral density by dual energy X-ray absorptiometry is used to diagnose osteoporosis and predict fracture risk. However, other measurements such as bone three-dimensional (3D) geometry, trabecular and cortical bone microstructure, which cannot be assessed by dual energy X-ray absorptiometry, are also important contributors to bone strength [1–3]. High resolution peripheral quantitative computed tomography (HR-pQCT) provides in vivo 3D visualization of the human distal radius and tibia and permits reliable evaluation of trabecular and cortical bone microstructural properties [4–6]. Based on HR-pQCT images, finite element analysis (FEA) that incorporates subject-specific geometrical and microstructural characteristics provides an integrative measure of bone quality [6–8]. Thus, HR-pQCT demonstrates great potential to improve fracture risk assessment [9–13].

Standard HR-pQCT scans a ∼9 mm section that begins 9.5 mm and 22.5 mm proximal to a reference line placed manually at the endplate of the radius and tibia. The selection of this region of interest (ROI) permits appropriate morphological evaluations of both trabecular and cortical bone. But it is not clear how sensitive the morphological measurement could be to region selections. Using a prototype of the HR-pQCT system with 89 μm in-plane and 93 μm out-plane voxel size, marked trabecular microstructural variations were observed among five consecutive sections in radius [14], suggesting that measurements of trabecular microstructure may be very sensitive to the selection of ROI. In addition, it also remains unclear how much one HR-pQCT scan of radius and tibia segment can quantify mechanical properties of the whole distal radius and tibia, respectively. Whole bone mechanical properties include whole bone stiffness and whole bone strength. Whole bone strength is the ultimate load that a bone can sustain before fracture and is correlated to fracture risk [15]. Whole bone stiffness is highly correlated to whole bone strength and has been often used as a surrogate for whole bone strength [8]. The advanced imaging analysis technique, individual trabecula segmentation (ITS), which automatically segment trabecular microstructure into collections of trabecular plates and rods, has also been applied to HR-pQCT images of patients. It has been shown that ITS analyses of HR-pQCT provide independent quantitative measures for delineating pathological changes in trabecular microstructure and better predictions of fracture outcomes [16,17]. Previous studies showed that microstructural and biomechanical parameters of segments from the most distal end of radius were correlated better with whole radius or whole distal radius strength than the standard HR-pQCT region [14,18]. But the relations between the trabecular plate/rod characterized bone microstructural parameters from segments at different regions and whole distal bone stiffness still need to be further quantified to better understand the representative power of HR-pQCT scans.

We hypothesized that finite element predicted bone stiffness in the standard segment is a better predictor of whole distal bone stiffness and microstructural parameters from the region closest to the distal loading surface correlated higher with whole distal bone stiffness than the regions far away. The aims of this study were to (1) characterize the trabecular and cortical microstructural variations through different regions within the distal radius and tibia using standard HR-pQCT and HR-pQCT-based ITS analyses; (2) examine the correlations between the microstructural/mechanical properties of segment from the standard HR-pQCT region and whole distal radius and tibia bone stiffness, measured by mechanical testing; and (3) determine if the correlations with whole distal bone stiffness can be improved for microstructural and mechanical measurements from other regions other than standard HR-pQCT region.

Twenty-six sets of fresh-frozen tibia and radius bones from the same donors (72±11-yr-old, 13 male and 13 female) were obtained from Life Legacy Foundation (Tucson, AZ). The specimens were screened and contact X-ray analyses were performed to exclude any samples with bone pathology or fracture (two were excluded). The radial and tibial bones were cut 110 mm away from the distal end.

After thawing, HR-pQCT scans were performed to image the distal radius and tibia at 82 μm isotropic voxel size (XtremeCT I, Scanco Medical, Brüttisellen, Switzerland). The distal radius and tibia bones were wrapped with pork skin to simulate in situ conditions. The setting for the scan was 60 kVp peak energy, 1000 μA current, and 100 ms integration time. The HR-pQCT scan of the radius and tibia segments included a total of 1341 slices, corresponding to a physical length of 110 mm along the specimen.

The image was divided into three adjacent segment regions (Fig. 1). The standard segment was selected based on the standard operating procedure in clinical HR-pQCT scans. The distal segment was defined as the region between the distal surface of the standard segment and the subchondral plate of radius and tibia. The proximal segment began at the proximal surface of the standard segment and extended 24 mm toward the proximal direction, to include all trabecular compartments. The standard clinical patient evaluation protocol was performed to examine bone microstructural properties for all three regions at the radius and tibia. The periosteal surfaces of the radius and tibia HR-pQCT images were contoured using a semi-automatic method and then segmented by an automatic threshold-based algorithm to separate the cortical and trabecular compartments. The mineralized phase was thresholded automatically by a Laplace–Hamming filter, followed by a global threshold value assuming 40% of the maximum possible gray value of the image. Total, cortical, trabecular bone mineral density (BMD), bone volume fraction (BV/TV^d), trabecular number (Tb.N), trabecular thickness (Tb.Th), trabecular separation (Tb.Sp), and cortical thickness (Ct.Th) were calculated.

An automatic segmentation algorithm implemented in Image Processing Language (IPL V5, Scanco Medical) was used to segment the trabecular and cortical regions for images from all three regions at both radius and tibia [19,20]. Based on the segmented trabecular region, ITS was applied to quantify plate and rod microstructural parameters [21]. The entire trabecular network was decomposed into individual trabecular plates or rods using a complete volumetric decomposition technique. Briefly, digital topological analysis-based skeletonization was applied first to transform each trabecular bone image into a reduced structure composed of 1-voxel-thick surface and curve skeletons while preserving the topology [22–24]. Then digital topological classification was performed and each skeletal voxel was uniquely classified as either a surface or a curve type. Using an iterative reconstruction method, each voxel of the original image was classified as belonging to either a plate (surface) or a rod (curve). Based on the 3D evaluations of the trabecular bone network, the following parameters were evaluated: bone volume fraction using direct voxel count (BV/TV), plate and rod bone volume fraction (pBV/TV and rBV/TV), axial bone volume fraction (aBV/TV), plate to rod ratio (P–R ratio), plate and rod number density (pTb.N and rTb.N, 1/mm), plate and rod trabecular thickness (pTb.Th and rTb.Th, mm), plate trabecular surface (pTb.S, mm²), and rod trabecular length (rTb.ℓ, mm); trabecular connection densities between plate–plate, plate–rod, and rod–rod (P–P Junc.D, P–R Junc.D and R–R Junc.D, 1/mm³).

Based on the HR-pQCT images of the distal radius and tibia segments from the three adjacent regions, μFE models were generated through a conversion of each voxel to eight-node brick element with element size of 82 × 82 × 82 μm³. Linear μFE analysis was performed to determine bone stiffness. Bone tissue was assumed to have an isotropic tissue elastic modulus of 15 GPa and Poisson's ratio of 0.3 [25,26]. The μFE simulated uniaxial compression on each segment from the radius and tibia to displacement corresponding to 1% apparent strain. Bone stiffness was defined as the load divided by the corresponding displacement.

The whole distal radius and distal tibia bones were cleaned to remove soft tissues. Gauze was used to remove any additional moisture on bone surface before embedding. Both ends of the distal radius and tibia were embedded in poly methyl methacrylate (PMMA) to facilitate mechanical testing with well-controlled boundary conditions. The embedding procedure for the radius was similar to that reported elsewhere in a configuration simulating the outstretch hand during the falling process, namely Colles' fractures [27]. The tibia was embedded to simulate single leg stance with the PMMA cylinder aligned with the axial direction of tibia. The embedded distal radius and tibia bones were shown in Fig. 2. Nondestructive mechanical testing was performed to examine the elastic properties of radius and tibia because the specimens were later physically cut into segments for destructive testing in other studies. The specimen was loaded using rigid platens that could not rotate to a compressive displacement corresponding to 0.35% apparent strain with one cycle of preconditioning. The displacement was chosen to make sure the sample was still in the elastic region. The load was applied using a displacement controlled quasi-static loading profile. A representative mechanical testing load–displacement curve was shown in Fig. 3. Each specimen was repositioned and mechanically tested three times to make sure the force–displacement curve was repeatable. Whole distal bone stiffness was calculated based on the linear force–displacement curve (Fig. 3).

Mean and standard deviation of the standard HR-pQCT parameters of the three segments were calculated. Microstructural differences among the three adjacent regions were tested for significance using analysis of variance with repeated measures. Pearson correlation coefficients were used to characterize the linear relations between the microstructural and biomechanical parameters from the distal, proximal, and standard segments. The correlation coefficients between the microstructural and biomechanical parameters of the three segments and whole distal bone stiffness, determined by mechanical testing, were also determined. The differences of correlation coefficient between microstructural/mechanical parameters of segments and whole distal bone stiffness were calculated using Fisher transformation.

The standard HR-pQCT parameters from the three regions for both radius and tibia are shown in Table 1. At the tibia, total BMD, cortical BMD, Tb.Th, Tb.Sp, and Ct.Th were significantly higher, whereas trabecular BMD, Tb.N, and bone area were significantly lower at proximal than the standard segment (p < 0.05). On the contrary, total BMD, cortical BMD, Tb.Th, Tb.Sp, and Ct.Th were significantly lower while Tb.N and bone area were significantly higher at the distal than the standard segment (p < 0.05). Morphological differences between the proximal, distal, and standard segments at the radius resembled those at the tibia except that Tb.N and Tb.Sp did not differ between the distal and standard segments.

Correlations of morphological parameters and FE-predicted bone stiffness between the proximal or distal segments and the standard segment are shown in Table 2. Total BMD, cortical BMD, Ct.Th, bone area, and stiffness of distal and proximal segments were correlated strongly with those from the standard segment (r = 0.81–0.95), except for cortical parameters at the distal radius (r = 0.65–0.69).

Trabecular plate and rod parameters analyzed by ITS were shown in Table 3. At the tibia, BV/TV, rBV/TV, aBV/TV, pTb.N, rTb.N, R–R, P–R, and P–P Junc.D were significantly lower while P–R ratio, pTb.Th, pTb.S, and rTb.ℓ were significantly higher at the proximal than the standard segment (p < 0.05). Interestingly, pBV/TV and rTb.Th did not differ between the proximal and standard segments. Distal segment ITS measurements of BV/TV, rBV/TV, rTb.N, R–R, P–R, and P–P Junc.D were significantly higher and aBV/TV, P–R ratio, pTb.Th, and pTb.S were significantly lower than the standard segment. pBV/TV, pTb.N, and rTb.ℓ were not different between distal and standard segments at the tibia. At the radius, the differences between the ITS measurements of the proximal and standard segments were similar to those at the tibia. However, aBV/TV of the proximal segment was not different from the standard segment. For the distal segment, most ITS parameters were similar to those from the standard segment, except rTb.N, pTb.Th, rTb.Th, pTb.S, and R–R Junc.D.

The correlations between the morphological parameters from the three adjacent regions and whole distal bone stiffness as measured from experiment for both radius and tibia were shown in Table 4. For the standard segment, HR-pQCT measurements of total BMD, cortical BMD, Tb.N, and Ct.Th were highly correlated with whole distal bone stiffness (r = 0.65–0.75). Moreover, HR-pQCT-based μFE predictions of stiffness of the standard segment correlated higher with whole distal bone stiffness than microstructural parameters at both radius and tibia (r = 0.94). At the tibia, correlations between HR-pQCT parameters from proximal segment and whole distal bone stiffness were similar to those of the standard segment. Trabecular BMD, Tb.N, and Tb.Th at the proximal segment correlated less strongly with whole distal bone stiffness as the same parameters at the standard segment (p < 0.05). In addition, total BMD, trabecular BMD, Tb.Th from distal segment had higher correlations with whole distal bone stiffness than those from the standard segment, at both radius and tibia (p < 0.05). Interestingly, the correlations between HR-pQCT-based μFE stiffness predictions from three regions and whole distal bone stiffness were not significantly different.

The correlations between the trabecular plate and rod microstructural properties and whole distal bone stiffness were further characterized (Table 5). At the standard segment, BV/TV, pTb.N, rTb.ℓ, P–R, and P–P Junc.D were all highly correlated with whole distal bone stiffness at radius and tibia (r = 0.63–0.76). Compared to the standard segment, radius proximal segment BV/TV, aBV.TV, pTb.N, rTb.N, rTb.ℓ, P–R, and P–P Junc.D measurements had lower correlations with whole distal bone stiffness. At the tibia, the correlations between the BV/TV, pBV.TV, pTb.Th, rTb.Th, and whole distal bone stiffness were significantly higher at the distal segment than the standard. However, at the radius, correlations between ITS parameters and whole distal bone stiffness were not different between the distal and standard segments except for pTb.S and pBV/TV. The linear relations between three segments stiffness and whole distal bone stiffness were also compared between male and female at both radius and tibia and no difference was observed (Table 6).

In this study, we examined the microstructural and biomechanical variations of trabecular and cortical bone from different regions of distal radius and tibia using HR-pQCT, HR-pQCT-based ITS and μFE analyses. The associations between whole distal bone stiffness and HR-pQCT and ITS parameters from different regions of radius and tibia were also quantified. We found that correlations between HR-pQCT microstructural parameters from the distal segment and whole distal bone stiffness were significantly higher compared with those parameters from the proximal and standard segments at both radius and tibia. To our knowledge, this is the first comprehensive study of microstructural and biomechanical variations and their associations with whole distal bone stiffness at the distal tibia.

Our observations about the microstructural changes of BV/TV, Tb.Sp, and Ct.Th at the distal radius as well as the correlations between microstructural parameters and whole distal radius strength are in agreement with previous observations [14]. In addition, our study also examined the relations between the mechanical properties of segments and whole distal bone and demonstrated the mechanical integrity between three segments. However, the microstructural variations reported previously were analyzed using a prototype HR-pQCT scanner with a nominal resolution of 89 μm in-plane and 93 μm slice thickness. Thus, our study provides more accurate observations based on images at a higher resolution. Our observations of the better prediction power of HR-pQCT measurements from the distal section of radius are consistent with previous studies, which concluded that the most distal section of the radius demonstrated higher correlation power than the standard segments [18]. This indicates that with smaller segment region and thus less scanning time and radiation exposure, HR-pQCT measurements of distal radius segment provide equivalent or better whole distal radius bone quality representations. Furthermore, we also found that whole distal tibia stiffness, similar to the radius, can be better predicted by HR-pQCT measurements of the distal segment, suggesting that bone mechanical competence is perhaps best assessed in the most distal segment.

In this study, we found the standard HR-pQCT parameters such as total BMD, cortical BMD, Tb.N, Tb.Th, Tb.Sp, and Ct.Th in the distal and proximal segment were significantly different from those in standard segment, indicating those microstructural measurements are sensitive to regional changes. This is important because the ROI of clinical HR-pQCT scans is selected in a way that does not account for individual differences in radius and tibia length. Therefore, HR-pQCT measurements could be affected by the ROI selected for different individuals. The ITS measurements further indicated the importance of selecting the proper ROI, with parameters such as P–R ratio significantly differing between regions in both radius and tibia. A previous study has suggested that Chinese-American women had significantly thicker cortical bone and a higher P–R ratio than Caucasian women [28–31]. A Chinese-American woman is about 2% shorter than a Caucasian American. Therefore, the ROI in Chinese women could be about 3.5 mm more proximal than that in Caucasian American women. According to the results in this study, it is possible that the previous observations were confounded by the standard ROI selection. However, the major difference observed between American-Chinese and Caucasian women was pBV/TV, which is relatively constant among regions. Previous studies have shown that the combination of ITS analysis and μFE simulations based on clinical HR-qCT scans improve prediction of fracture risk [16,32]. The findings from this study indicate that diagnostic capability of image-based microstructural and mechanical analyses could be further enhanced.

There are also several limitations in this study. First, the distal end of radius and tibia were embedded in PMMA to generate well-controlled boundary conditions in the mechanical testing procedure. Although this embedding technique at the radius has been shown to be capable of reproducing Colles' fracture, it does not fully represent the real loading condition the radius experiences during a fall. The falling load is transferred from hand to radius through lunate and scaphoid, resulting in nonuniform force distribution on the surface of the distal radius. In addition, the configuration of the distal tibia represents only the standing condition. Therefore, the results at tibia cannot be interpolated to other loading conditions in the occurrences of shear. Second, the samples used in this study were from an older population. Therefore, the results from this study may not be translatable to younger groups. Future studies should consider including more specimens from a younger population. In addition, we only measured whole distal bone stiffness using nondestructive mechanical testing because we needed bone segments for other mechanical testing purposes. Future studies should examine the associations of bone strength and toughness with segment microstructural and mechanical properties, as both parameters are important determinates of bone mechanics. This study only demonstrated that the measurements from the distal segment of radius and tibia correlated higher with whole distal bone stiffness, further efforts should be paid to examine whether measurements of this region could improve fracture risks predictions with regards to bone strength and bone toughness.

In conclusion, we observed significant regional microstructural and biomechanical variations in human distal radius and tibia. Compared with the correlations between the HR-pQCT and ITS parameters from the standard segment, the distal segment, which is closer to the location where force is exerted, had higher prediction power of whole distal bone stiffness. This study indicates that HR-pQCT scans of the radius and tibia are sensitive to regional differences and should move toward the distal direction to better represent the whole distal bone quality.

• National Institutes of Health (Grant Nos. AR051376 and AR058004, Funder ID: 10.13039/100000002).