• finite element analysis;
  • teriparatide;
  • PTH(1-34);
  • alendronate;
  • bone strength;
  • bone quality


  1. Top of page
  2. Abstract
  7. Acknowledgements

FE modeling was used to estimate the biomechanical effects of teriparatide and alendronate on lumbar vertebrae. Both treatments enhanced predicted vertebral strength by increasing average density. This effect was more pronounced for teriparatide, which further increased predicted vertebral strength by altering the distribution of density within the vertebra, preferentially increasing the strength of the trabecular compartment.

Introduction: Teriparatide 20 μg/day (TPTD) and alendronate 10 mg/day (ALN) increase areal, measured by DXA, and volumetric, measured by QCT, lumbar spine BMD through opposite effects on bone remodeling. Using finite element (FE) modeling of QCT scans, we sought to compare the vertebral strength characteristics in TPTD- and ALN-treated patients.

Materials and Methods: A subset of patients (N = 28 TPTD; N = 25 ALN) from the Forteo Alendronate Comparator Trial who had QCT scans of the spine at baseline and postbaseline were analyzed. The QCT scans were analyzed for compressive strength of the L3 vertebra using FE modeling. In addition, using controlled parameter studies of the FE models, the effects of changes in density, density distribution, and geometry on strength were calculated, a strength:density ratio was determined, and a response to bending was also quantified.

Results: Both treatments had positive effects on predicted vertebral strength characteristics. At least 75% of the patients in each treatment group had increased strength of the vertebra at 6 months compared with baseline. Patients in both treatment groups had increased average volumetric density and increased strength in the trabecular bone, but the median percentage increases for these parameters were 5- to 12-fold greater for TPTD. Larger increases in the strength:density ratio were also observed for TPTD, and these were primarily attributed to preferential increases in trabecular strength.

Conclusions: These results provide new insight into the effects of these treatments on estimated biomechanical properties of the vertebra. Both treatments positively affected predicted vertebral strength through their effects on average BMD, but the magnitudes of the effects were quite different. Teriparatide also affected vertebral strength by altering the distribution of density within the vertebra, so that overall, teriparatide had a 5-fold greater percentage increase in the strength:density ratio.


  1. Top of page
  2. Abstract
  7. Acknowledgements

Osteoporosis is a disease characterized by decreased bone strength predisposing an individual to an increased susceptibility for fracture.(1) Finite element analysis (FEA)—a well-established engineering computational method of strength analysis for complex structures(2)—has been applied to QCT scans of the spine(3–6) to provide a noninvasive measure of bone strength. Conceptually, FEA of QCT scans biomechanically integrates the material and geometric information in the QCT scan to produce an estimate of whole bone strength that is a better predictor of actual strength than estimators derived from QCT or DXA.(3,7) Moreover, by averaging bone mass over the voxels of the FE mesh, an integral measurement of volumetric density for the vertebra can be obtained. This, in turn, enables calculation of a vertebral strength:density ratio, which can be considered a quantitative measure of structural efficiency.(8) FE modeling has been used in many osteoporosis biomechanics studies over the past 15 years(4–6,9–12) and most recently has been applied clinically to the assessment of glucocorticoid treatment at the hip.(13) In addition to providing noninvasive measures of strength and density, controlled parameter studies of the FE models of a patient's vertebra can be performed to assess the independent effects on vertebral strength of such parameters as bone geometry, average density, density distribution, and the relative contribution of the trabecular bone.

The purpose of this analysis was to compare the effects of teriparatide and alendronate on vertebral strength using FEA of the QCT scans obtained during the Forteo Alendronate Comparator Trial.(14) Moreover, we sought to gain insight into how the respective mechanisms of action of these two established therapies for osteoporosis translate into the biomechanical properties of the vertebra. This is the first study that compares two approved treatments for osteoporosis on FE-derived biomechanical measurements of the spine in a clinical setting.


  1. Top of page
  2. Abstract
  7. Acknowledgements

Study design and participants

The Forteo Alendronate Comparator Trial was a randomized, double-blinded, double-dummy 18-month active comparator study of teriparatide injection 20 μg/day (N = 102) versus oral alendronate 10 mg/day (N = 101).(14) Study participants were women between the ages of 45 and 85, who were ambulatory, at least 5 years postmenopausal, with a BMD T score between −2.5 and −4.0 at the lumbar spine or femoral neck. All participants were provided daily supplements of 1000 mg calcium and 400–800 IU vitamin D. A detailed description of the study design and methods is published elsewhere.(14) Institutional review board approval was obtained from each of the 19 clinical trial sites, and written informed consent was obtained from all participants. All study methods and procedures were conducted according to the ethical principles of the Declaration of Helsinki.


QCT scans of the lumbar spine (L1–L3) were taken at seven study sites at baseline and 6 and 18 months; study endpoint scans were collected from randomized patients who discontinued early from the study. A total of 53 patients, randomized to either the teriparatide (N = 28) or alendronate (N = 25) groups, provided scans at baseline and postbaseline. All lumbar spine scans were acquired at 120 kVp using a CT slice thickness of 3 mm with an original in-plane voxel size between 0.6 and 0.8 mm that varied across subjects. The same model liquid K2HPO4 calibration phantom was used at all imaging centers to correct for drift and to convert QCT numbers (in Hounsfield units) to equivalent BMD (in mg/cm3). Analysis and quality assurance (QA) of the bone density scans acquired by QCT were performed by a central imaging facility (Bio-Imaging Technology, Newtown, PA, USA) using commercial software (QCT PRO; Mindways Software, San Francisco, CA, USA). Measurement of volumetric BMD within the central portion of the trabecular compartment (vtBMD)(15) was performed by a technician blinded to treatment assignment. System characterization scans (N = 10) were performed on each scanner before the start of the study. The purpose of these scans was to establish a baseline of each individual CT system's performance. These scans included the simultaneous imaging of the QA phantom and a calibration phantom. In addition, system monitor scans, identical to the system characterization scan, were done weekly (N = 3) and at the time of patient scans. The same CT scanner thickness and spacing setting were used for QA and patient scans. These scans were intended to monitor system performance over the course of the study. QA scans were interpreted in the context of a previous characterization scan and were used to identify changes in the operational characteristics of the CT scanner. Monthly QA scan data were centrally analyzed to assure CT scanners operational characteristics were consistent throughout the study. Gradual changes in operational characteristics may occur over time as a scanner ages; however, no changes were significant enough to warrant re-characterization for QCT analyses.

FEA-assessed strength

FE models of the L3 vertebrae for each patient were generated from digitized scans using custom software. If the L3 image was not available (e.g., previous fracture, imaging artifact, technician error) for a patient at all time-points, L2 was analyzed. The general process of converting the QCT scan into the FE model is similar to what has been described elsewhere(3,16) but with some minor differences. Each vertebra image (less posterior elements) was rotated into a standard coordinate system, thresholded, and converted into a 1 × 1 × 1.5-mm3 voxel-type FE mesh (Fig. 1), using 8-noded brick elements. The QCT mineral density of each bone voxel was obtained, and the mean across all voxels in the FE mesh was used as the integral measure of volumetric density (viBMD). For each voxel the axial elastic modulus (Ez) was calculated using the QCT mineral density of the voxel and the empirical correlation between elastic modulus and calibrated BMD for human vertebral trabecular bone.(17) Elastic anisotropy of the bone was accounted for by assuming fixed ratios of the various elastic constants with respect to Ez, as described elsewhere.(3) Material failure of the bone was modeled by assigning an elastic-perfectly-plastic von Mises failure criterion to the bone. A relation between yield stress and calibrated BMD(17) was modified to convert it to an ultimate stress relation by assuming the ultimate stress was 1.2 times the yield stress.(3) The elastic modulus and ultimate stress values were all multiplied by a constant factor of 1.28 to account for the so-called “side-artifact” errors. This corrects for the fact that excised specimens of trabecular bone that are used in biomechanical test experiments have lower mechanical properties than in situ because of the interruption of connectivity at their free edges.(18) Preliminary work has shown that correction of such artifacts improves the absolute accuracy of vertebral strength predictions.(16)

thumbnail image

Figure Figure 1. Voxel QCT-based FE models of the same lumbar vertebral body in a teriparatide-treated patient at baseline and after 18 months of treatment (21.1% increase from baseline in vertebral strength). Shown also is a typical transverse slice from the QCT scan and the assumed relation between ultimate stress and QCT-derived BMD for the bone within each finite element.

Download figure to PowerPoint

A thin layer of polymethylmethacrylate (PMMA) was virtually placed at the ends of each vertebral model to simulate test conditions commonly used in compression strength testing of cadaver vertebrae.(19) Uniform compressive displacement boundary conditions were applied to the external surfaces of these PMMA layers. The compressive strength of the vertebra was computed as the total reaction force generated at an imposed displacement equivalent to an overall bone compressive strain of 2% (applied displacement divided by bone height). Using twice-replicated analyses on 30 randomly chosen scans and an accepted precision formula,(20) the CV for the measurement of compressive strength was 1.08%. Cadaver studies using this approach have shown excellent predictions of measured vertebral strength.(16)

FEA-controlled parameter studies

A series of parameter studies was conducted in which each model was varied in a controlled fashion and the compressive strength simulation was rerun on the altered models to compute a number of other outcome variables. The results of these studies helped to gain insight into the biomechanical mechanisms of action of the treatments.

First, inter- and intravertebral BMD variations were removed by applying an arbitrary reference density (100 mg/cm3) uniformly across all voxels in the FE mesh for all vertebrae, and the resulting geometric strength (GEOM) values were computed. Changes in GEOM over time isolate the effect of changes in geometry on changes in strength. Second, intravertebral BMD variations were removed from each vertebra by applying the viBMD uniformly across all voxels of the FE mesh, and the resulting homogenized-density strength (HOM) values were computed. Changes in HOM over time reflect changes in strength caused by the combination of changes in viBMD and changes in geometry. Third, intervertebral average BMD effects were removed from each vertebra by multiplying each voxel's density by 100/viBMD. Each resulting normalized model had the same value of viBMD, 100 mg/cm3, but the relational distribution of density within the model remained unchanged. Therefore, changes in normalized-density strength (NORM) values over time reflect changes in strength caused by the combination of changes in the distribution of density within the vertebra and changes in geometry. Finally, trabecular strength (TRAB) values were calculated by removing the outer 2 mm of bone from the model and recomputing strength for the remaining trabecular compartment. Because the QCT scans only provide a blurred representation of the very thin cortical shell,(21) the strength associated with the cortical bone was not derived.

The FE model was also used to provide an assessment of the effect of treatment on the response to anterior-posterior (AP) type bending loads. For this analysis, a pure bending rotation of 1° was applied to the top endplate using linearly elastic analysis. A bending stiffness was computed, and this value was compared with the measurement for compressive stiffness.(22)

Statistical analysis

For the baseline characteristics, two-sample t-tests were used to compare age, years postmenopausal, body mass index, and BMD measurements between the treatment groups. Wilcoxon rank sum tests were used for comparisons of the baseline bone markers and baseline FEA measurements of vertebral strength, and a Fisher's exact test was used to compare racial origin between the treatment groups. Two-sample t-tests were used to compare percent changes in BMD from baseline between the treatment groups at each time-point, and nonparametric rank sum tests were used for comparisons of percent changes from baseline in the bone markers and most of the FEA measurements of vertebral strength at each time-point. For each t-test and rank sum test, exact p values were obtained using Monte Carlo simulation with 25,000 samples. Estimates of the SE for the median were obtained by bootstrapping.

The relationship between change in predicted vertebral compressive strength and the three measurements of change in BMD—areal BMD (aBMD) by DXA, vtBMD by QCT alone, and viBMD by FEA—were examined by calculating Pearson correlation coefficients for each treatment using the percentage changes at study endpoint. For each density measure, the percentage change in vertebral strength was also modeled using an ANCOVA model with treatment as the main effect and percentage change in density as the covariate and including also their interaction. These regression models were used to determine if the linear relationship between changes in vertebral strength and changes in bone mass were different between the treatments.

No adjustments were made for the multiple comparisons. All analyses were performed using SAS statistical software (version 8; SAS Institute, Cary, NC, USA).


  1. Top of page
  2. Abstract
  7. Acknowledgements

The patients in this substudy were slightly younger with a higher percentage of Hispanic patients than the overall study cohort,(14) but the baseline demographics of the patients included in the FE analyses were similar between the treatment groups (Table 1). Moreover, average percent change for the primary and secondary endpoints in this substudy (Table 2) were similar to those observed in the overall study cohort.

Table Table 1. Baseline Characteristics of Participants in the Biomechanical CT Analysis
Thumbnail image of
Table Table 2. Median Percent Change (25th, 75th Interquartile Range) in Primary and Secondary Endpoints of the Forteo Alendronate Comparison Trial in the Subset of Patients Analyzed With Biomechanical CT
Thumbnail image of

Both treatments had positive effects on vertebral compressive strength (Fig. 2). Vertebral compressive strength was increased from baseline for at least 75% of the patients in each treatment group at month 6, but the median percentage increase for teriparatide was significantly greater than for alendronate, 13.0% versus 4.9%, respectively. From months 6 to 18, vertebral strength in the alendronate group remained relatively stable, whereas the median percentage increase for the teriparatide group doubled. At 18 months, the median percentage increase for teriparatide was 5.7-fold greater from baseline compared with alendronate, 21.1% versus 3.7%. The median percentage increase in viBMD for the alendronate group was 2.5–3.4% at the different time-points, and this increase was more pronounced for the teriparatide group at months 6 and 18, 7.9% and 9.9%, respectively (Fig. 2). Both treatments also increased the strength:density ratio from baseline, but the median percentage increases were three to five times larger for the teriparatide group (6.01% versus 2.15%, p = 0.002 at month 6; 9.31% versus 1.86%, p < 0.001 at month 18; Fig. 2).

thumbnail image

Figure Figure 2. Median percent change in biomechanical CT-predicted whole vertebral compressive strength, average vertebral density, and the ratio of whole vertebral compressive strength to average vertebral density in teriparatide- and alendronate-treated women. In each box, the line represents the median, upper end of the box is the 75th interquartile range, and lower end of box is the 25th interquartile range. *p < 0.001 and **p < 0.05 within group from baseline; †p < 0.001, ‡p < 0.01 between group. NS, nonsignificant; ALN, alendronate; TPTD, teriparatide.

Download figure to PowerPoint

Increases in vertebral strength caused by geometric changes, as measured by changes in GEOM, were statistically significant within each treatment group but small. Median percentage increases at study endpoint were 0.86% (p < 0.05) and 0.74% (p < 0.001) for alendronate and teriparatide, respectively. Because of the negligible nature of these changes, changes in HOM were dominated by changes in viBMD, and changes in NORM were dominated by changes in the distribution of density within the vertebrae. Consistent with the trends seen for viBMD (Fig. 2), the median percentage increases in HOM were greater for teriparatide than for alendronate (10.63% versus 4.57%, p = 0.066 at month 6; 15.71% versus 3.73%, p < 0.001 at month 18; Fig. 3). The increases in NORM were smaller than the increases in HOM for both treatments (both p < 0.05), but the median percentage changes in NORM were significantly larger for teriparatide than for alendronate (4.71% versus 0.45%, p = 0.005 at month 6; 7.62% versus 0.67%, p < 0.001 at month 18; Fig. 3). The effect of alendronate on TRAB was a consistent 3–6% median percent increase from baseline that was observed at month 6 and maintained through month 18 (Fig. 3). The changes in TRAB were much greater and continued to increase over time for the teriparatide group (Fig. 3).

thumbnail image

Figure Figure 3. Median percent change (25th, 75th interquartile range) in whole vertebral compressive strength (VERT), homogenous strength (HOM), normalized strength (NORM), and trabecular strength (TRAB). In each box, the line represents the median, upper end of the box is the 75th interquartile range, and lower end of box is the 25th interquartile range. *p < 0.001 and **p < 0.05 within group from baseline; †p < 0.001, ‡p < 0.01 between group. NS, nonsignificant; ALN, alendronate; TPTD, teriparatide.

Download figure to PowerPoint

Median percentage increases in AP bending stiffness (Fig. 4) were also greater for the teriparatide group compared with the alendronate group (11.17% versus 4.95%, p = 0.025 at month 6; 16.67% versus 2.7%, p < 0.001 at month 18). The increases in AP bending stiffness for teriparatide were, on average, smaller than the increases in compressive stiffness (p = 0.002). The results for compressive stiffness (Fig. 4) were similar to those for compressive strength, because endpoint compressive stiffness was highly correlated with endpoint compressive strength (r = 0.96, p < 0.0001).

thumbnail image

Figure Figure 4. Median percent change in compressive stiffness (COMP), anterior-posterior bending stiffness (BEND), and their ratio (BEND/COMP) in teriparatide- and alendronate-treated women. In each box, the line represents the median, upper end of the box is the 75th interquartile range, and lower end of box is the 25th interquartile range. *p < 0.001, **p < 0.01, and ***p < 0.05 within group from baseline; †p < 0.001, ‡p < 0.01, and † †p < 0.05 between group. NS, nonsignificant; ALN, alendronate; TPTD, teriparatide.

Download figure to PowerPoint

Comparison of change in the FEA vertebral compressive strength at endpoint versus the various endpoint changes in aBMD, vtBMD, and viBMD indicated that DXA did not capture well the changes in biomechanical properties that occurred with treatment, and that in general, the relation between BMD changes and FEA strength changes depended on treatment. The correlation with endpoint aBMD (Fig. 5A) for the teriparatide treatment group was moderate (r2 = 0.58, p < 0.001), but no relationship was observed in the alendronate treatment group (p = 0.895). Stronger, statistically significant (p < 0.001) correlations existed with the two volumetric BMD measures for both treatment groups (Figs. 5B and 5C, r2 = 0.76–0.92). For all three BMD measures, the regression lines versus FEA strength for the teriparatide and alendronate groups were statistically different (p < 0.003 for aBMD and vtBMD; p = 0.048 for viBMD). For the volumetric integral BMD measure (viBMD), the regression lines for the two treatments differed only in the intercept parameter; for the other two BMD measures, both slopes and intercepts differed between treatments.

thumbnail image

Figure Figure 5. The correlation between percent change from baseline to endpoint in FE vertebral compressive strength and (A) DXA aBMD, (B) QCT VtBMD, and (C) ViBMD in teriparatide- and alendronate-treated patients. Solid and dashed lines show relations for teriparatide and alendronate, respectively.

Download figure to PowerPoint


  1. Top of page
  2. Abstract
  7. Acknowledgements

The goal of this analysis was to use FEA of QCT scans to provide a noninvasive biomechanical description and comparison of the effects of up to 18-month treatment with a bone-forming agent (teriparatide) and an antiresorptive treatment (alendronate) in postmenopausal women with osteoporosis. This analysis technique has been shown to provide estimates of vertebral compressive strength that correspond closely with direct ex vivo biomechanical measurements for cadaveric vertebrae.(16) To our knowledge, this is the first application of the FE technique to compare two different osteoporosis treatments in a clinical study. The results indicated that both treatments increased predicted vertebral compressive strength, but the effects for teriparatide were much larger, particularly as time progressed. Both treatments increased average density and increased strength in greater proportion than average density, as evidenced by the strength-to-density ratio. Whereas this latter effect was caused in part by the slightly nonlinear relation assumed between bone tissue strength and density (see Fig. 1), comparison of the trends seen with the controlled parameter studies—specifically the variables that measured strength changes caused by geometry (GEOM), combined average density and geometry (HOM), and combined intravertebral variations in density and geometry (NORM), respectively—indicated that there were also biomechanical effects beyond those of average density, particularly for teriparatide. Given the negligible strength changes associated with any geometry changes, the statically greater increase in normalized strength (NORM), which was only observed for teriparatide, was indicative of strength gains caused by alterations in the distribution of BMD within the vertebra. Larger increases in trabecular strength relative to the increases in whole vertebral strength were observed for the teriparatide group, consistent with effects seen in the spines of Cynomolgus monkeys treated with teriparatide for 18 months, in which the main effect was a conversion of low BMD bone into medium BMD bone, with little change to the high BMD bone.(23) Taken together, these findings provide new insight into the biomechanical effects of two biologically different treatments. In particular, both treatments enhanced vertebral strength by increasing average density with a more pronounced effect for teriparatide, and teriparatide further increased vertebral strength by altering the distribution of density within the vertebra, in particular, by preferentially increasing the strength of the trabecular compartment.

It is currently unknown how well vertebral fracture incidence is correlated with these reported changes in vertebral strength (i.e., how much of an increase in vertebral strength is required to prevent a vertebral fracture). Biomechanically, a treatment-induced increase in strength protects from fracture by moving the bone over or away from its fracture threshold (i.e., the point at which the in vivo forces applied to the bone during a strenuous activity or fall exceed the strength of the bone).(24–26) In contrast, age-related decreases in strength increase the risk of fracture by moving the bone closer to or below its fracture threshold. Mosekilde et al.(27,28) estimated that ˜10% of vertebral strength is lost per decade, and this loss is exacerbated further in individuals with osteoporosis. Considering treatment with alendronate is known to decrease vertebral fracture incidence by about 50% at 12 months in postmenopausal women having osteoporosis,(29) the results of this analysis suggest that even modest treatment-induced increases in vertebral strength, such as those observed for a majority of alendronate-treated patients in this study, may prevent osteoporotic vertebral fractures by primarily halting age- and osteoporosis-related declines in vertebral strength.

A related issue is the possible fracture risk reduction benefits conferred by larger treatment-induced increases in vertebral strength in individuals, such as those observed for a majority of teriparatide-treated patients and a few alendronate patients in this study. Biomechanically, any fracture that is prevented for a bone having a pretreated strength below its biomechanical fracture threshold would only be prevented if the increase in vertebral strength was sufficiently large to protect against the applied in vivo loads.(24–26,30) Patients having numerous and/or severe prevalent vertebral fractures are at increased risk of subsequent fracture,(31,32) presumably because they have reduced strength of multiple vertebrae from the biomechanical deterioration associated with systemic bone loss. Hence, it would be expected that a large percentage of patients with prevalent fractures have other vertebrae that are near or below their biomechanical fracture threshold. Clinically, treatment with alendronate has shown roughly the same relative risk reduction (with respect to placebo) in patients with prevalent fractures as in patients without prevalent fractures.(29,33) In contrast, for treatment with teriparatide, the relative risk reduction seems to increase with both the number and severity of prevalent fractures.(34) The results from this study provide a biomechanical explanation for these clinical results, namely that larger increases in vertebral strength could provide a clinical benefit for those patients whose pretreatment level of bone strength is further below its fracture threshold. Additional longitudinal clinical studies of different treatments and/or a placebo group that include FEA-predicted vertebral strength from QCT scans and fracture as outcomes would provide substantial further insight into these issues.

Some limitations to the analysis should be noted. First, the FE models provide only predictors of strength and were derived from a compressive strength test of isolated vertebral bodies, which is one standard in the measurement of vertebral strength in cadaver studies.(19,35–39) Second, it was assumed that neither treatment altered the input voxel-level strength-density relation that was derived from the cadaver studies (see Fig. 1). This assumption is supported by a series of preclinical bisphosphonate studies. High doses of risedronate in dogs did not appreciably alter the strength-density relations of the trabecular bone or the tissue level mechanical properties.(40,41) Similarly, in ovariectomized baboons, 2 years of alendronate treatment did not alter (cortical) tissue mechanical properties, nor did it alter the nonlinear strength-density relation for trabecular bone.(42) Studies in cynomolgus monkeys have shown that ibandronate preserves trabecular architecture and biomechanical strength.(43) In a clinical study of risedronate, μCT analysis of biopsies have shown that risedronate preserves trabecular architecture in postmenopausal women with osteoporosis.(44) Taken together, these studies suggest that antiresorptives preserve both microarchitecture, tissue material properties, and biomechanical strength and do not fundamentally alter any structure-function relations of the bone. Similarly, preclinical studies of teriparatide have not shown any alterations of the strength-density relation of bone,(45,46) nor are we aware of any other preclinical studies that have shown the contrary for contemporary treatments.(8) Indeed, biopsy studies indicate that teriparatide improves trabecular microarchitecture by increasing connectivity.(47) However, given that some theoretical studies have suggested that microcavities from the resorption space might alter the strength-volume relation under certain conditions,(48) additional preclinical studies are needed to address more definitively such possible treatment-induced alterations to the voxel-level density-mechanical property relations.

In addition, our analysis focused mostly on compressive strength. Some aspects of bending behavior were addressed, but the bending loads studied do not mimic the complex anterior bending in vivo boundary conditions through a disc because these conditions are not currently well understood and may depend on disc quality.(49) Bending stiffness increased preferentially for teriparatide but did not increase as much as compressive stiffness or strength. This is consistent with the observed preferential increases in trabecular strength, because vertebral bending stiffness, like bending structural parameters for beam type structures, is most sensitive to changes in the outer bone.(22)

There are several implications of the results from the correlation and regression analyses between the FEA measure of strength and the various measures of BMD. First, the statistically significant treatment effect on the correlations between changes in FEA vertebral strength and changes in trabecular volumetric BMD, vtBMD, provide further evidence that teriparatide achieves strength gains beyond those expected from increases in trabecular density alone. Specifically, the data in Fig. 5B indicate that, for a given increase in trabecular volumetric BMD, teriparatide achieves larger vertebral strength increases than alendronate. Related, this treatment effect shows a limitation associated with use of just trabecular volumetric BMD or any volumetric subregion because it does not capture the integrative biomechanical effects of treatment. As shown in Fig. 5C, use of a more integrative measure of BMD can help overcome this limitation, but subtle treatment effects can still linger. This limitation on using subregions for densitometric analysis is consistent with results from recent micromechanical analyses of the human vertebral body that quantified load sharing between the cortical and trabecular bone and showed that the trabecular contribution to load sharing was least in the midvertebral region.(50) Being able to attribute the additional increase in strength for teriparatide to the alteration of density distribution within the vertebra could only be identified by altering the FE models in a controlled fashion. Another noteworthy finding was that changes in aBMD from DXA were poorly correlated with the changes in FE strength. This shows that DXA failed to capture the treatment-induced biomechanical changes, particularly for alendronate. Consistent with concerns raised in the literature about limitations of DXA,(51) this finding might help explain why DXA fails to adequately explain the efficacy of antiresorptive treatments.(52,53)

In conclusion, FE modeling of QCT scans showed that teriparatide and alendronate both increased vertebral strength by increasing average density, and this biomechanical effect was much greater for teriparatide. Furthermore, teriparatide also altered the distribution of density within the bone, preferentially increasing trabecular strength and the strength:density ratio. These effects were not evident from the DXA or QCT density data. Taken together, these findings may aid clinicians in choosing the most appropriate therapy for an individual patient and highlight the potential of FE analysis of QCT scans to provide unique clinical insight into the biomechanical effects of osteoporosis therapies at the spine.


  1. Top of page
  2. Abstract
  7. Acknowledgements

The authors thank Mindy Rance and Mary Ellen Perron, BS, for assistance with the figures and Kathryn Engstrom, BS, for statistical programming. Dr Keaveny has a financial interest in O.N. Diagnostics and both he and the company may benefit from the results of this research. Funding was provided by Lilly Research Laboratories.


  1. Top of page
  2. Abstract
  7. Acknowledgements
  • 1
    NIH Consensus Development Panel on Osteoporosis Prevention, Diagnosis and Therapy 2001 Osteoporosis prevention, diagnosis, and therapy. JAMA 285: 785795.
  • 2
    Reddy JN 1993 An Introduction to the Finite Element Method, 2nd ed. McGraw-Hill, New York, NY, USA.
  • 3
    Crawford RP, Cann CE, Keaveny TM 2003 Finite element models predict in vitro vertebral body compressive strength better than quantitative computed tomography. Bone 33: 744750.
  • 4
    Faulkner KG, Cann CE, Hasegawa BH 1991 Effect of bone distribution on vertebral strength: Assessment with patient-specific nonlinear finite element analysis. Radiology 179: 669674.
  • 5
    Liebschner MA, Kopperdahl DL, Rosenberg WS, Keaveny TM 2003 Finite element modeling of the human thoracolumbar spine. Spine 28: 559565.
  • 6
    Homminga J, Weinans H, Gowin W, Felsenberg D, Huiskes R 2001 Osteoporosis changes the amount of vertebral trabecular bone at risk of fracture but not the vertebral load distribution. Spine 26: 15551561.
  • 7
    Cody DD, Gross GJ, Hou FJ, Spencer HJ, Goldstein SA, Fyhrie DP 1999 Femoral strength is better predicted by finite element models than QCT and DXA. J Biomech 32: 10131020.
  • 8
    Hernandez CJ, Keaveny TM 2006 A biomechanical perspective on bone quality. Bone (in press).
  • 9
    Silva MJ, Keaveny TM, Hayes WC 1998 Computed tomography-based finite element analysis predicts failure loads and fracture patterns for vertebral sections. J Orthop Res 16: 300308.
  • 10
    Keyak JH, Rossi SA, Jones KA, Skinner HB 1998 Prediction of femoral fracture load using automated finite element modeling. J Biomech 31: 125133.
  • 11
    Ford CM, Keaveny TM, Hayes WC 1996 The effect of impact direction on the structural capacity of the proximal femur during falls. J Bone Miner Res 11: 377383.
  • 12
    Keyak JH, Kaneko TS, Tehranzadeh J, Skinner HB 2005 Predicting proximal femoral strength using structural engineering models. Clin Orthop Relat Res 437: 219228.
  • 13
    Lian KC, Lang TF, Keyak JH, Modin GW, Rehman Q, Do L, Lane NE 2005 Differences in hip quantitative computed tomography (QCT) measurements of bone mineral density and bone strength between glucocorticoid-treated and glucocorticoid-naive postmenopausal women. Osteoporos Int 16: 642650.
  • 14
    McClung MR, San Martin J, Miller PD, Civitelli R, Bandeira F, Omizo M, Donley DW, Dalsky GP, Eriksen EF 2005 Opposite bone remodeling effects of teriparatide and alendronate in increasing bone mass. Arch Intern Med 165: 17621768.
  • 15
    Richardson ML, Genant HK, Cann CE, Kolb FO, Ettinger B, Gordan GS 1985 Noninvasive assessment of skeletal mass. J Comput Assist Tomogr 9: 636638.
  • 16
    Crawford RP, Brouwers JEM, Keaveny TM 2004 Accurate prediction of vertebral strength using voxel-based non-linear finite element models. 50th Annual Orthopaedic Research Society Meeting, March 7–10, 2004, San Francisco, CA, USA.
  • 17
    Kopperdahl DL, Morgan EF, Keaveny TM 2002 Quantitative computed tomography estimates of the mechanical properties of human vertebral trabecular bone. J Orthop Res 20: 801805.
  • 18
    Ün K, Bevill G, Keaveny TM 2006 The effects of side-artifacts on the elastic modulus of trabecular bone. J Biomech 39: 19551963.
  • 19
    Kopperdahl DL, Pearlman JL, Keaveny TM 2000 Biomechanical consequences of an isolated overload on the human vertebral body. J Orthop Res 18: 685690.
  • 20
    Pearson D, Miller CG 2002 Clinical Trials in Osteoporosis. Springer, New York, NY, USA.
  • 21
    Silva MJ, Wang C, Keaveny TM, Hayes WC 1994 Direct and computed-tomography thickness measurements of the human, lumbar vertebral shell and end-plate. Bone 15: 409414.
  • 22
    Crawford RP, Keaveny TM 2004 Relationship between axial and bending behaviors of the human thoracolumbar vertebra. Spine 29: 22482255.
  • 23
    Sato M, Westmore M, Clendenon J, Smith S, Hannum B, Zeng GQ, Brommage B, Turner CH 2000 Three-dimensional modeling of the effects of parathyroid hormone on bone distribution in lumbar vertebrae of ovariectomized Cynomologus macaques. Osteoporos Int 11: 871888.
  • 24
    Hayes W 1991 Biomechanics of cortical and trabecular bone: Implications for assessment of fracture risk. MowV, HayesW (eds.) Basic Orthopaedic Biomechanics, Raven Press, New York, NY, USA, 93142.
  • 25
    Myers ER, Wilson SE 1997 Biomechanics of osteoporosis and vertebral fracture. Spine 22(Suppl 24): 25S31S.
  • 26
    Bouxsein ML, Black DM, Muller J, Bilezikian J, Greenspan S, Ensrud K, Palermo L, Crawford PM, Rosen CJ, Keaveny TM 2005 Combining biomechanical measures of vertebral strength with estimates of spinal loading to assess the effects of osteoporosis therapies on vertebral fracture: A pilot study in PaTH. J Bone Miner Res 20: S1; 411S.
  • 27
    Mosekilde L, Mosekilde L, Danielsen CC 1987 Biomechanical competence of vertebral trabecular bone in relation to ash density and age in normal individuals. Bone 8: 7985.
  • 28
    Mosekilde L, Mosekilde L 1986 Normal vertebral body size and compressive strength: Relations to age and to vertebral and iliac trabecular bone compressive strength. Bone 7: 207212.
  • 29
    Black DM, Thompson DE, Bauer DC, Ensrud K, Musliner T, Hochberg MC, Nevitt MC, Suryawanshi S, Cummings SR 2000 Fracture risk reduction with alendronate in women with osteoporosis: The Fracture Intervention Trial. FIT Research Group. J Clin Endocrinol Metab 85: 41184124.
  • 30
    Bouxsein ML 2005 Determinants of skeletal fragility. Best Pract Res Clin Rheumatol 19: 897911.
  • 31
    Klotzbuecher CM, Ross PD, Landsman PB, Abbott TA III, Berger M 2000 Patients with prior fractures have an increased risk of future fractures: A summary of the literature and statistical synthesis. J Bone Miner Res 15: 721739.
  • 32
    Delmas PD, Genant HK, Crans GG, Stock JL, Wong M, Siris E, Adachi JD 2003 Severity of prevalent vertebral fractures and the risk of subsequent vertebral and nonvertebral fractures: Results from the MORE trial. Bone 33: 522532.
  • 33
    Nevitt MC, Ross PD, Palermo L, Musliner T, Genant HK, Thompson DE 1999 Association of prevalent vertebral fractures, bone density, and alendronate treatment with incident vertebral fractures: Effect of number and spinal location of fractures. The Fracture Intervention Trial Research Group. Bone 25: 613619.
  • 34
    Gallagher JC, Genant HK, Crans GG, Vargas SJ, Krege JH 2005 Teriparatide reduces the fracture risk associated with increasing number and severity of osteoporotic fractures. J Clin Endocrinol Metab 90: 15831587.
  • 35
    McBroom RJ, Hayes WC, Edwards WT, Goldberg RP, White AA 1985 Prediction of vertebral body compressive fracture using quantitative computed tomography. J Bone Joint Surg Am 67: 12061214.
  • 36
    Cheng XG, Nicholson PHF, Boonen S, Lowet G, Brys P, Aerssens J, VanderPerre G, Dequeker J 1997 Prediction of vertebral strength in vitro by spinal bone densitometry and calcaneal ultrasound. J Bone Miner Res 12: 17211728.
  • 37
    Ebbesen EN, Thomsen JS, Beck-Nielsen H, Nepper-Rasmussen HJ, Mosekilde L 1999 Lumbar vertebral body compressive strength evaluated by dual-energy X-ray absorptiometry, quantitative computed tomography, and ashing. Bone 25: 713724.
  • 38
    Eriksson SA, Isberg BO, Lindgren JU 1989 Prediction of vertebral strength by dual photon absorptiometry and quantitative computed tomography. Calcif Tissue Int 44: 243250.
  • 39
    Edmondston SJ, Singer KP, Day RE, Price RI, Breidahl PD 1997 Ex vivo estimation of thoracolumbar vertebral body compressive strength: The relative contributions of bone densitometry and vertebral morphometry. Osteoporos Int 7: 142148.
  • 40
    Eswaran SK, Allen MR, Burr D, Keaveny TM 2006 Effect of risedronate treatment-induced changes in microarchitecture on trabecular bone strength-density characteristics. 52nd Annual Orthopaedic Research Society Meeting, March 19–22, 2006, Chicago, IL, USA.
  • 41
    Day JS, Ding M, Bednarz P, van der Linden JC, Mashiba T, Hirano T, Johnston CC, Burr DB, Hvid I, Sumner DR, Weinans H 2004 Bisphosphonate treatment affects trabecular bone apparent modulus through micro-architecture rather than matrix properties. J Orthop Res 22: 465471.
  • 42
    Balena R, Toolan BC, Shea M, Markatos A, Myers ER, Lee SC, Opas EE, Seedor JG, Klein H, Frankenfield D, Quartuccio H, Florauanti C, Clair J, Brown E, Hayes WC, Rodan GA 1993 The effects of 2-year treatment with the aminobisphosphonate alendronate on bone metabolism, bone histomorphometry, and bone strength in ovariectomized nonhuman primates. J Clin Invest 92: 25772586.
  • 43
    Muller R, Hannan M, Smith SY, Bauss F 2004 Intermittent ibandronate preserves bone quality and bone strength in the lumbar spine after 16 months of treatment in the ovariectomized cynomolgus monkey. J Bone Miner Res 19: 17871796.
  • 44
    Borah B, Dufresne TE, Chmielewski PA, Johnson TD, Chines A, Manhart MD 2004 Risedronate preserves bone architecture in postmenopausal women with osteoporosis as measured by three-dimensional microcomputed tomography. Bone 34: 736746.
  • 45
    Sato M, Westmore M, Ma YL, Schmidt A, Zeng QQ, Glass EV, Vahle J, Brommage R, Jerome CP, Turner CH 2004 Teriparatide [PTH (1-34)] strengthens the proximal femur of ovariectomized nonhuman primates despite increasing porosity. J Bone Miner Res 19: 623629.
  • 46
    Burr DB, Hirano T, Turner CH, Hotchkiss C, Brommage R, Hock JM 2001 Intermittently administered human parathyroid hormone (1-34) treatment increases intracortical bone turnover and porosity without reducing bone strength in the humerus of ovariectomized cynomolgus monkeys. J Bone Miner Res 16: 157165.
  • 47
    Dempster DW, Cosman F, Kurland ES, Zhou H, Nieves J, Woelfert L, Shane E, Plavetic K, Muller R, Bilezikian J, Lindsay R 2001 Effects of daily treatment with parathyroid hormone on bone microarchitecture and turnover in patients with osteoporosis: A paired biopsy study. J Bone Miner Res 16: 18461853.
  • 48
    Hernandez CJ, Gupta A, Keaveny TM 2006 A biomechanical analysis of the effects of resorption cavities on cancellous bone strength. J Bone Miner Res 21: 12481255.
  • 49
    Adams MA, Green TP, Dolan P 1994 The strength in anterior bending of lumbar intervertebral discs. Spine 19: 21972203.
  • 50
    Eswaran SK, Gupta A, Adams MF, Keaveny TM 2006 Cortical and trabecular load sharing in the human vertebral body. J Bone Miner Res 21: 307314.
  • 51
    Faulkner KG 2000 Bone matters: Are density increases necessary to reduce fracture risk? J Bone Miner Res 15: 183187.
  • 52
    Delmas PD, Seeman E 2004 Changes in bone mineral density explain little of the reduction in vertebral or nonvertebral fracture risk with anti-resorptive therapy. Bone 34: 599604.
  • 53
    Cummings SR, Karpf DB, Harris F, Genant HK, Ensrud K, LaCroix AZ, Black DM 2002 Improvement in spine bone density and reduction in risk of vertebral fractures during treatment with antiresorptive drugs. Am J Med 112: 281289.