Teriparatide Increases Strength of the Peripheral Skeleton in Premenopausal Women With Idiopathic Osteoporosis: A Pilot HR-pQCT Study

Abstract

Context:

In premenopausal women with idiopathic osteoporosis (IOP), treatment with teriparatide leads to substantial improvement in bone density and quality at central skeletal sites. The effects of teriparatide may differ on cortical and trabecular bone and also at the central and the peripheral skeleton.

Objective:

The objective of the study was to determine whether teriparatide was associated with improvements in compartmental volumetric bone mineral density (BMD), bone microarchitecture, and estimated bone strength of the distal radius and tibia as assessed by high-resolution peripheral quantitative computed tomography.

Design, Setting, and Participants:

Premenopausal women (n = 20, age 41 ± 5 y) with IOP (low trauma fractures and/or Z-scores ≤ −2.0) were scanned with high-resolution peripheral quantitative computed tomography at baseline and after 18 months of teriparatide treatment. Cortical and trabecular volumetric BMD and microarchitecture were measured by both standard and advanced techniques, including individual trabecula segmentation, and bone strength was estimated by finite element analysis.

Main Outcome Measures:

The total volumetric BMD and homogeneous bone stiffness were measured.

Results:

Trabecular volumetric BMD increased significantly by 2.6% (1.8, 6.2) [median (interquartile range)] at the radius and 2.5% (1.1, 3.6) at the tibia. In addition, trabecular plate bone volume fraction increased by 9.1% (2.1, 17.1) at the radius and 7.6% (1.0, 9.7) at the tibia. Cortical thickness and volumetric density did not change; however, cortical porosity increased at the radius but not at the tibia. Despite these changes, whole-bone stiffness and failure load estimated by finite element analysis increased at both the radius and tibia.

Conclusions:

In premenopausal women with IOP, 18 months of teriparatide was associated with increases in trabecular volumetric BMD, improved trabecular microarchitecture, and estimated bone strength at both the radius and tibia.

Idiopathic osteoporosis (IOP) is defined as osteoporosis that affects young, otherwise healthy individuals with intact gonadal function and no secondary cause of bone loss or fragility (1). Transiliac bone biopsies indicate that compared with normal healthy women, premenopausal women with IOP have thinner cortices in addition to thinner, fewer, and more widely separated trabeculae, which contribute to lower estimated bone stiffness (2). In addition, high-resolution peripheral quantitative computed tomography (HR-pQCT) has shown that women with IOP have abnormal bone microstructure at the distal radius and tibia, with deterioration of both trabecular and cortical bone (3).

Teriparatide is a recombinant form of PTH (1–34) that increases areal bone mineral density (aBMD) by dual-energy x-ray absorptiometry (DXA) and decreases fractures in postmenopausal women (4), glucocorticoid-induced osteoporosis (5, 6), and men with senile osteoporosis or IOP (7, 8). In a recent pilot study, we reported that teriparatide administration also increased bone mass in 21 premenopausal women with IOP (9). By 24 months, teriparatide significantly increased aBMD by DXA at the spine by 10.8% ± 6.4% and at the total hip and femoral neck by 6.2% ± 5.7% and 7.6% ± 3.4%, respectively, with no significant change at the one third radius. Paired transiliac bone biopsies performed before and after treatment revealed increased trabecular bone volume (34%) and number (18%) as well as a 71% increase in trabecular stiffness. In addition, there was a 22% increase in cortical thickness; however, cortical porosity also increased by 46% (9).

DXA is a two-dimensional imaging modality that cannot distinguish between cortical and trabecular bone compartments. HR-pQCT is a noninvasive technique that can distinguish the cortical and trabecular regions in vivo and measure volumetric bone mineral density (vBMD) and bone microstructure (10). The effects of teriparatide on bone microstructure in postmenopausal women with osteoporosis have been assessed by HR-pQCT (11, 12). Although trabecular microstructure improved in one study (11), there was evidence for deterioration in the other (12). In addition, both studies reported a trend toward (12) or significant increases (11) in cortical porosity. However, both of these studies also applied finite element analysis (FEA) to estimate bone strength and found that, despite the cortical bone changes, bone strength was maintained, albeit not improved (11, 12).

It is unknown whether teriparatide would have similar effects on bone microstructure and estimated bone strength in premenopausal women with IOP. In addition, it is unclear whether mineralization changes due to teriparatide treatment may confound estimates of bone strength in typical FEA models that assign the same material properties to all bone tissue. We hypothesized that in premenopausal women with IOP, teriparatide would be associated with improved quality of trabecular bone structure but an increase in cortical porosity at the distal radius and tibia. We also hypothesized that, despite the increase in cortical porosity, estimated bone strength would increase with teriparatide treatment.

Materials and Methods

Participants and treatment regimen

Premenopausal women (n = 21) with a history of low-trauma fracture and/or low spine or hip aBMD by DXA (Z-score ≤ −2.0) were recruited as previously described (9) at Columbia University Medical Center (New York, New York) and Creighton University (Omaha, Nebraska) by advertisement, self-referral, or physician referral. We excluded women with secondary osteoporosis related to estrogen deficiency, eating disorders, endocrinopathies, celiac or gastrointestinal diseases, hyperparathyroidism, marked hypercalciuria (>300 mg/g creatinine), serum 25-hydroxyvitamin D levels less than 20 ng/ML, and exposure to drugs that cause osteoporosis (9).

All participants received 20 μg of teriparatide daily in the morning or evening, according to their preference, for 18–24 months. They also received 630 mg of calcium and 800 IU of vitamin D. The institutional review boards of both institutions approved this study. All women used contraception during the study and provided written informed consent.

Imaging assessments

aBMD was measured by DXA (Discovery; Hologic Inc) at baseline and at 6,12, 18, and 24 months at the spine, femur, and radius, respectively, as previously described (2, 9).

All participants were scanned with the same HR-pQCT (XtremeCT; Scanco Medical) scanner at Columbia University Medical Center. The nondominant distal radius and left distal tibia were scanned unless there was a previous fracture at the desired site, in which case we scanned the opposite limb. Highly trained operators acquired and analyzed all scans according to the manufacturer's standard in vivo protocol. We scanned all participants using 60 kVp effective energy, 900 μA current, and 100 milliseconds integration time to acquire 110 slices at an 82-μm nominal isotropic resolution. Scans were manually scored for motion on a scale of 1 (no motion) to 5 (significant blurring of the periosteal surface, discontinuities in the cortical shell, or streaking in the soft tissue). No images had a score of 4 or 5; thus, all scans were included. We used the manufacturer's standard method to filter and binarize the HR-pQCT images (13). To segment the cortical and trabecular regions, we used a previously validated, automatic segmentation algorithm independently on all scans (14). We assessed all standard HR-pQCT morphological microstructure outcomes, including trabecular and cortical vBMD (milligrams hydroxylapatite per cubic centimeter) and trabecular bone volume fraction (BV/TV; percentage), number (1 per millimeter), thickness (millimeters), separation (millimeters), and heterogeneity (trabecular separation SD; millimeters) (10). These measurements were previously validated against microcomputed tomography (15, 16) and, in adult populations, have in vivo short-term reproducibility of coefficient of variation (CV) less than 4.5% (17). Short-term reproducibility in our laboratory is a CV less than 1.06% for the BMD measurements and CV less than 5.2% for the structural parameters at the radius and tibia.

Using the automatic segmentation method we also calculated total (Tt.Ar, square millimeters), cortical (square millimeters), and trabecular (square millimeters) cross-sectional areas. In addition to the standard cortical morphological outcomes, we assessed cortical porosity (percentage) calculated as the number of void voxels within the cortex (18) and directly measured cortical thickness (millimeters), and cortical bone mineral density (milligrams hydroxylapatite per cubic centimeter). We also performed a more detailed analysis of the trabecular bone microstructure by individual trabecula segmentation (ITS), as previously applied to premenopausal women with IOP and other patient populations (19–21) and as described in detail elsewhere (20, 22, 23). Briefly, this method creates a skeleton of the trabecular region using digital topologic skeletonization and classification, in which elements are classified as a surface or a curve. These elements are then mapped back to the voxels of the original image to determine whether individual voxels belong to a plate or rod through iterative volumetric reconstruction (22). Based on this segmentation, we obtained trabecular structural measurements including plate (pBV/TV) and rod BV/TV (rBV/TV), axial BV/TV (aBV/TV), plate (pTb.N) and rod number (rTb.Th; 1 per millimeter), and the average thickness of plates and rods (plate trabecular number thickness and rod trabecular number thickness; millimeters).

Bone strength estimates

We estimated bone strength from the HR-pQCT images using FEA based on the voxel conversion approach (24, 25). We simulated uniaxial compression on each radius and tibia model up to 1% strain using a homogeneous Young's modulus of 6829 MPa and Poisson's ratio of 0.3 (26). Homogeneous models that assign the same material properties to all of the bone tissue are commonly used; however, these do not account for differences in mineral content that may have an effect on bone stiffness. Thus, to account for the fact that teriparatide therapy may result in formation of large amounts of bone that is not yet fully mineralized at the time of the scan acquisition, we also scaled the tissue moduli to estimate bone strength. Briefly, the computed tomography attenuation values are converted to tissue modulus values based on a previously determined relationship that accounts for varying bone strength with density (26, 27). This method uses the gray-scale values directly without thresholding and assigns an appropriate modulus value. These modulus values are then incorporated into the finite element model. We used a custom FEA solver (FAIM, version 6.0; Numerics88) on a desktop workstation (Linux Ubuntu 12.10, 2 × 6-core Intel Xenon, 64 GB RAM) to solve the models as previously described (18, 28). We estimated whole-bone strength (ultimate stress, MPa), and failure load (N) for both the homogeneous and scaled finite element models. To estimate the risk of forearm fracture, we calculated the load to strength ratio (Φ) as the expected fall load (equation 1) divided by the scaled estimated radius failure load (29–31).

$$\text{fall}\text{load}=670\text{Ns}/\text{m}\times \sqrt{2\times 9.81\times \left(\text{height}/2\right)}$$ (equation 1)

Statistical analysis

We used Wilcoxon's signed rank test to compare the HR-pQCT and FEA measurements at baseline and 18 months. The main outcome variables were total vBMD and FEA estimated, homogeneous bone stiffness. All statistical analysis was performed with R (version 2.15.2).

Results

Participant characteristics

Participant characteristics are shown in Table 1. On average, the women were approximately 40 years old at baseline. aBMD Z-scores were at the low end of the expected range for age. Sixteen women (80%) had a history of low-trauma fracture with an average of 1.7 fractures each. All had normal calciotropic hormones and indices of mineral metabolism. HR-pQCT data at baseline and 18 months were available for 20 of the 21 participants. We excluded the radius data for a participant who sustained a distal radius fracture during the study and was scanned at the opposite limb. We also excluded the radius data for one participant who did not have a matching 18-month scan. Thus, the final analysis included baseline and follow-up measurements of 18 radius and 20 tibia scans.

Table 1.

Participant Characteristics

Subject Characteristics (n = 20)	Mean ± SD
Age, y	40.3 ± 4.6
Height, cm	163.5 ± 6.5
Weight, kg	57.3 ± 9.5
BMI, kg/m2	21.5 ± 3.6
Baseline BMD (Z-score)
Lumbar spine	−1.9 ± 0.8
Total hip	−1.7 ± 0.5
Femoral neck	−2.0 ± 0.7
Distal radius	−1.8 ± 0.7
Adult low-trauma fractures, n, %	16 (80%)
Number of adult low trauma fractures	1.7 ± 1.7
Serum 25-OHD, ng/mL	41.2 ± 11.4
Serum PTH, pg/mL	34.0 ± 9.8
Serum calcium, albumin adjusted, mg/dL	9.0 ± 0.3

Abbreviations: BMD, bone mineral density; BMI, body mass index; 25-OHD, 25-hydroxyvitamin D.

aBMD by DXA

By 18 months, there were significant increases in lumbar spine (9.8% ± 5.2%, P < .001), total hip (2.9% ± 3.6%, P < .001), and femoral neck (3.5% ± 5.3%, P < .01) aBMD but no significant changes in ultradistal radius or one third radius aBMD. At 24 months there was a similar pattern of increase at the lumbar spine (10.8% ± 6.4%, P < .001), total hip (6.2% ± 5.7%, P < .001), and femoral neck (7.6% ± 3.4%, P < .01) and no changes at the radius.

Bone microarchitecture

Mean values of the baseline and 18-month follow-up measurements are shown in Table 2 for both the radius and tibia. Significant changes are shown as median percentage changes for the radius and tibia in Figures 1 and 2, respectively. There was an average overlapping region of 96% between the baseline and follow-up scans at both the radius and tibia. There was an increase in trabecular vBMD at both the radius (2.6%, P < .001) and the tibia (2.5%, P = .002) and also an increase in total vBMD at the radius (0.6%, P = .048) and the tibia (1.5%, P = .007). There were no changes in cortical vBMD at either site. Cortical porosity increased by 17.8% (P = .009) at the radius but remained unchanged at the tibia. Although the percentage increase was large at the radius, the absolute change in cortical porosity at the radius was very small, from 0.6% to 0.7%. There were no significant changes in directly measured cortical thickness at either site. In the trabecular compartment, we observed increased BV/TV at both the radius (3.9%, P = .002) and tibia (2.1%, P = .003) and a decrease in trabecular separation at the tibia (−1.3%, P = .036). With the exception of cortical porosity (P = .009), there were no significant differences between the changes at the radius and the tibia.

Table 2.

Distal Radius and Tibia Mean ± SD Values at Baseline and at 18 Months for Volumetric Density and Microstructure as Measured by HR-pQCT

	Radius			Tibia
	Baseline	18 Months	P Value	Baseline	18 Months	P Value
Total vBMD, mg HA/cm3	291.4 ± 45.2	294.6 ± 45.5	.048	233.4 ± 38.1	237.1 ± 38.8	.007
Tb vBMD, mg HA/cm3	99.7 ± 34.1	103.0 ± 33.8	<.001	109.6 ± 28.4	112.6 ± 29.3	.002
Ct vBMD, mg HA/cm3	910.4 ± 46.3	910.3 ± 45.9	1.000	861.5 ± 58.7	861.0 ± 65.7	.475
Total area, mm2	223.8 ± 35.4	223.6 ± 35.3	.130	638.6 ± 104.7	638.3 ± 104.1	.216
Tb area, mm2	172.9 ± 31.9	172.5 ± 31.7	.349	538.9 ± 101.8	537.9 ± 101.9	.035
Ct area, mm2	50.9 ± 6.7	51.0 ± 6.9	.896	99.7 ± 15.2	100.4 ± 15.4	.076
Tb BV/TV	0.083 ± 0.028	0.086 ± 0.028	.002	0.091 ± 0.024	0.094 ± 0.025	.003
Tb number, 1/mm	1.53 ± 0.30	1.54 ± 0.23	.193	1.50 ± 0.28	1.51 ± 0.28	.073
Tb thickness, mm	0.053 ± 0.009	0.055 ± 0.012	.180	0.061 ± 0.010	0.062 ± 0.011	.133
Tb separation, mm	0.623 ± 0.140	0.609 ± 0.108	.150	0.632 ± 0.150	0.623 ± 0.148	.036
Tb separation SD, mm	0.292 ± 0.090	0.289 ± 0.088	.338	0.320 ± 0.123	0.314 ± 0.117	.341
Ct porosity, %	0.6 ± 0.3	0.7 ± 0.4	.009	3.8 ± 2.6	3.9 ± 2.6	.654
Ct thickness, mm	0.92 ± 0.10	0.92 ± 0.10	.734	1.16 ± 0.18	1.17 ± 0.19	.114
Plate BV/TV	0.048 ± 0.023	0.053 ± 0.027	<.001	0.083 ± 0.026	0.088 ± 0.029	.001
Rod BV/TV	0.147 ± 0.030	0.148 ± 0.025	.038	0.135 ± 0.028	0.135 ± 0.029	.812
Axial BV/TV	0.063 ± 0.020	0.069 ± 0.026	.022	0.092 ± 0.018	0.097 ± 0.020	.006
Plate Tb number, 1/mm	1.14 ± 0.17	1.17 ± 0.17	.003	1.31 ± 0.11	1.33 ± 0.11	<.001
Rod Tb number, 1/mm	1.79 ± 0.14	1.79 ± 0.11	.325	1.72 ± 0.14	1.72 ± 0.14	.452
Plate Tb thickness, mm	0.203 ± 0.006	0.206 ± 0.006	.021	0.218 ± 0.009	0.219 ± 0.009	.004
Rod Tb thickness (mm)	0.207 ± 0.005	0.209 ± 0.008	.393	0.214 ± 0.006	0.214 ± 0.005	.294

Abbreviations: Ct, cortical; HA, hemagglutinin; Tb, trabecular. P values indicate significant differences between baseline and 8-months based on a Wilcoxon's signed rank test.

Figure 1.

Median percentage changes and interquartile range (box) for HR-pQCT and FEA parameters measured at the distal tibia. Only parameters with significant changes between baseline and follow-up are depicted (P < .05). aBV/TV, axial BV/TV; Ct.Po, cortical porosity; pBV/TV, plate BV/TV; pTb.N, plate trabecular number; pTb.Th, average thickness of plates; rBV/TV, rod BV/TV;

Figure 2.

Median percentage changes and interquartile range (box) for HR-pQCT and FEA parameters measured at the distal radius. Only parameters with significant changes between baseline and follow-up are depicted (P < .05). aBV/TV, axial BV/TV; pBV/TV, plate BV/TV; pTb.Th, average thickness of plates; Tb.Area, trabecular cross-sectional area; Tb.Sp, trabecular separation;

ITS analysis revealed increases in plate BV/TV at both the radius (9.1%, P < .001) and tibia (7.6%, P = .001). rBV/TV also increased at the radius (3.4%, P = .038) and aBV/TV, a measure of the axial distribution of the trabecular network that is positively associated with bone stiffness, increased at both sites (radius: 3.9% P = .022; tibia: 5.0%, P = .006). Only rBV/TV (P = .034) changes were significantly different between the radius and tibia. After Bonferroni correction, results remained significant for rBV/TV and tibial pBV/TV, radial trabecular vBMD, and tibial plate trabecular number.

Estimated bone strength by FEA

Estimates of bone strength are presented in Table 3. At the radius, using the homogeneous models, there was a trend for increased stiffness (P = .060) but no change in the estimated failure load. In contrast, the model using scaled tissue properties revealed a 1.3% (P = .001) increase in stiffness and a 1.1% (P = .021) increase in failure load (Figure 2). The load to strength ratio decreased significantly by 1.1% (P = .018). At the tibia, both the homogeneous and scaled tissue property models revealed significant increases in estimated strength. With the homogeneous models, we found a 3.6% (P = .002) increase in stiffness and a 3.3% (P = .001) increase in failure load. Using the scaled models, we found a 2.0% (P < .001) increase in stiffness and a 1.6% (P < .001) increase in failure load (Figure 2). After Bonferroni correction, results remained significant for radial and tibial scaled stiffness as well as tibial homogeneous and scaled failure load.

Table 3.

Distal Radius and Tibia Mean ± SD Values at Baseline and 18 Months for Estimated Bone Strength as Measured by HR-pQCT and FEA

	Radius			Tibia
	Baseline	18 Months	P Value	Baseline	18 Months	P Value
Homogeneous stiffness, N/mm	29247 ± 3590	30283 ± 3968	.060	82977 ± 13185	85244 ± 12866	.002
Homogeneous failure load, N	1354 ± 173	1386 ± 169	.117	3721 ± 477	3821 ± 460	.001
Scaled stiffness, N/mm	22716 ± 2375	22997 ± 2404	.001	57556 ± 6426	58395 ± 6424	<.001
Scaled failure load, N	780 ± 75	790 ± 71	.021	1994 ± 187	2031 ± 184	<.001
Load to strength ratio	1.36 ± 0.12	1.34 ± 0.11	.018	N/A	N/A	N/A

Abbreviation: N/A, not available. P values indicate significant differences between baseline and 18 months based on a Wilcoxon's signed rank test.

Discussion

In this pilot study, we found that 18 months of teriparatide was associated with increased areal and volumetric bone density, improved trabecular microarchitecture, and increased estimated bone strength of the peripheral skeleton in premenopausal women with IOP. In particular, total vBMD, trabecular vBMD, and trabecular pBV/TV increased at both the radius and tibia. Although there was also a small increase in cortical porosity at the radius, whole-bone stiffness and failure load estimated by scaled finite element models increased at both sites.

This is the first study to use HR-pQCT to examine the effects of teriparatide in premenopausal women with IOP. These noninvasively obtained results are consistent with studies of paired transiliac biopsy samples in the same population and show that teriparatide is associated with significant improvements in trabecular microstructure and estimated bone strength at the peripheral as well as the central skeleton (9). Prior to this study, HR-pQCT had been used only to examine the effects of teriparatide in postmenopausal women with osteoporosis. Hansen et al (11) found no changes in trabecular microarchitecture at the radius but an increase in trabecular number and BV/TV at the tibia. In a small 18-month study, also in postmenopausal women treated with teriparatide, Macdonald et al (12) reported contrasting results, namely decreased vBMD, trabecular thickness, and BV/TV at both the distal radius and distal tibia. In both studies, the women had received prior bisphosphonate treatment, a possible explanation for the lack of improvement in the trabecular structure.

HR-pQCT allows us to examine the cortical and trabecular regions separately and discern the differing compartmental effects of teriparatide on the peripheral skeleton. Hansen et al (11) reported that teriparatide treatment increased cortical thickness at the radius and tibia in postmenopausal women. Increased cortical thickness in response to teriparatide has also been demonstrated in animal studies in rabbits (32) and monkeys (33). These increases in cortical thickness have been attributed to increased bone formation at the endocortical and periosteal surface (34). We have previously reported increases in cortical thickness at the iliac crest on paired transiliac bone biopsies in these subjects with IOP. In contrast to data in postmenopausal women and in the same premenopausal cohort with IOP, neither cortical thickness nor cortical vBMD changed at the radius or tibia in response to teriparatide. However, there was a significant decrease in the trabecular area (P = .035), and a trend toward increased cortical area (P = .076) at the tibia. The reasons for the discrepancies between pre- and postmenopausal women and between the iliac crest and peripheral sites are unclear.

Within the trabecular compartment, there were significant improvements in both trabecular vBMD as well as trabecular microarchitecture. Using the standard HR-pQCT analysis, there was an increase in BV/TV at both the radius and tibia. At the tibia there was also a decrease in trabecular separation, although trabecular number and thickness did not change. However, using ITS, with its ability to classify individual trabeculae as plates and rods, we were able to appreciate significant improvements in trabecular plate microarchitecture, specifically increases in trabecular pBV/TV, thickness, and number at both radius and tibia. This is noteworthy because improvement in trabecular plate architecture has been shown to be associated with greater bone strength (22). We did not find as many significant changes in the rod trabecular structure, which may be due to the smaller size of the rods and the inability to resolve them fully using HR-pQCT (23).

Similar to our results, Hansen et al (11) reported small but significant increases in cortical porosity at the radius as well as at the tibia with teriparatide treatment. Although not significant, possibly due to the small sample size, cortical porosity also tended to increase in the study by Macdonald et al (12). In both of these studies of postmenopausal women, this increase in porosity was thought to be due to the expected acceleration of intracortical remodeling (35). On the endocortical surface, PTH is also known to stimulate bone formation more than resorption (36). Because we classified cortical and trabecular bone independently in the baseline and follow-up images, it is possible that with endocortical formation, regions that were classified as dense trabecular bone in the baseline scans were classified as porous cortical bone in the 18-month scans. This may explain the small absolute increase in porosity at the radius; however, we did not find any significant increases in porosity at the tibia, possibly due to the protective benefit of weight bearing (37). We did not find any significant changes in the cortical vBMD, possibly because the cortical porosity changes were not large enough to influence the overall density.

In general, there were trends toward greater percentage improvements in trabecular bone structure and estimated bone strength at the tibia than the radius. There were significant differences in the changes between the radius and tibia only in cortical porosity and rod bone volume fraction. This discrepancy between radius and tibia may be due to the greater precision of HR-pQCT at the tibia (17), a site with less motion artifact than the radius. It is also possible that the load-bearing nature of the tibia complements the effect of teriparatide treatment to improve bone structure. Our findings are consistent with our prior DXA results, in which there were site-specific differences in response to teriparatide in premenopausal women with IOP (9). At the load-bearing spine, total hip, and femoral neck, there were large and significant increases. In contrast, at the unloaded forearm, there was a transient decrease in areal BMD at 6 months, with return to baseline by 12 months (9). Similar observations have also been reported in postmenopausal women (11). The increases in vBMD detected by HR-pQCT at the radius were not detected by DXA, possibly because DXA is a two-dimensional image and matching of the scan regions may be less reproducible.

To account for tissue mineral changes due to teriparatide, we used both homogeneous and scaled finite element models to estimate bone strength. The scaled models assign material properties based on the various tissue densities within the bone rather than on homogeneous material properties for all bone tissue based on a fixed threshold. This would account for bone that is newly formed due to teriparatide treatment and not yet fully mineralized by 18 months. At the tibia, both types of models showed significant increases in the estimated stiffness and failure load with teriparatide treatment. At the radius, only the scaled models showed a significant increase in stiffness and failure load, although there was a trend (P = .06) for increased stiffness with the homogeneous models. Thus, our data suggest that scaled models may be a more sensitive means of detecting the changes in bone strength (26), given that greater heterogeneity in mineralization would be expected with teriparatide treatment. At the radius, the newly formed bone may not have mineralized fully; thus, the changes were detected only by the scaled FEA models. When the images are binarized and homogeneous material properties are used, subtle mineralization changes will not be accounted for. It is possible that radius aBMD by DXA does not detect these mineralization changes. In postmenopausal women treated with teriparatide, bone strength estimated by FEA has been shown to increase by 28% at the vertebrae (38), a much greater increase than we observed at the radius and tibia, and to be preserved at the total hip (39). In contrast to our findings, no significant changes in stiffness were found at the peripheral skeleton in postmenopausal women (11, 12). Our results are similar to the increases in bone stiffness estimated by HR-pQCT-based finite element models in postmenopausal women treated with strontium ranelate (40).

Our study has limitations. Our sample size was relatively small and we did not have an untreated control group. Larger randomized studies should be completed to verify these results. Although HR-pQCT has been validated with higher-resolution microcomputed tomography for standard measurements (15), advanced cortical bone parameters (18), and ITS parameters (23), the nominal isotropic resolution is limited to 82 μm, and thus, changes in small cortical pores or small, rod-like trabeculae may not be detected. Finally, it is possible that teriparatide forms new bone that is not yet fully mineralized; this may result in an underestimation of the changes in any standard parameters based on the binarized images. In addition, although the marrow composition changes at the peripheral skeleton are unclear, we have previously found decreases in adipocyte area at the iliac crest in patients treated with teriparatide. It is possible these changes may have an effect on the attenuation measurements and thus warrants further investigation. Future work should focus on developing methods to better quantify tissue mineralization changes.

In summary, teriparatide treatment in premenopausal women with IOP was associated with increased total and trabecular bone density as well as improved trabecular microstructure at both the radius and tibia. Despite small absolute increases in cortical porosity at the radius, bone strength estimates increased at both sites. Although larger, randomized controlled studies are needed to confirm these results, the measured changes are consistent with the expected actions of teriparatide. We conclude that teriparatide has beneficial effects on both the central and the peripheral skeleton in premenopausal women with idiopathic osteoporosis.