Dr Keaveny has served as a speaker/consultant to Merck, Amgen, Pfizer, Lilly, and Novartis; has received financial support for research projects from Merck, NPS, Lilly, Pfizer, P&G, Amgen, and GSK; and has ownership/equity interests in O.N. Diagnostics, where he serves as part-time Chief Scientist. Mr Hoffmann has equity interests in O.N. Diagnostics and is a full-time employee there. Dr Bilezikian has served as a consultant to Merck, Lilly, Amgen, NPS, Radius, and Alliance for Better Bone Health. Dr Greenspan has served as a speaker/consultant to Merck and NPS and has received research support from Merck and NPS. Dr Black has served as a consultant to Roche, Merck, and NPS and has received research support from Novartis. All other authors state that they have no conflicts of interest.
Femoral Bone Strength and Its Relation to Cortical and Trabecular Changes After Treatment With PTH, Alendronate, and Their Combination as Assessed by Finite Element Analysis of Quantitative CT Scans
Article first published online: 4 AUG 2008
Copyright © 2008 ASBMR
Journal of Bone and Mineral Research
Volume 23, Issue 12, pages 1974–1982, December 2008
How to Cite
Keaveny, T. M., Hoffmann, P. F., Singh, M., Palermo, L., Bilezikian, J. P., Greenspan, S. L. and Black, D. M. (2008), Femoral Bone Strength and Its Relation to Cortical and Trabecular Changes After Treatment With PTH, Alendronate, and Their Combination as Assessed by Finite Element Analysis of Quantitative CT Scans. J Bone Miner Res, 23: 1974–1982. doi: 10.1359/jbmr.080805
- Issue published online: 4 DEC 2009
- Article first published online: 4 AUG 2008
- Manuscript Accepted: 31 JUL 2008
- Manuscript Revised: 16 JUN 2008
- Manuscript Received: 11 JUL 2007
- bone strength;
- drug therapy;
- clinical trials;
The “PTH and Alendronate” or “PaTH” study compared the effects of PTH(1-84) and/or alendronate (ALN) in 238 postmenopausal, osteoporotic women. We performed finite element analysis on the QCT scans of 162 of these subjects to provide insight into femoral strength changes associated with these treatments and the relative roles of changes in the cortical and trabecular compartments on such strength changes. Patients were assigned to either PTH, ALN, or their combination (CMB) in year 1 and were switched to either ALN or placebo (PLB) treatment in year 2: PTH-PLB, PTH-ALN, CMB-ALN, and ALN-ALN (year 1-year 2) treatments. Femoral strength was simulated for a sideways fall using nonlinear finite element analysis of the quantitative CT exams. At year 1, the strength change from baseline was statistically significant for PTH (mean, 2.08%) and ALN (3.60%), and at year 2, significant changes were seen for the PTH-ALN (7.74%), CMB-ALN (4.18%), and ALN-ALN (4.83%) treatment groups but not for PTH-PLB (1.17%). Strength increases were primarily caused by changes in the trabecular density regardless of treatment group, but changes in cortical density and mass also played a significant role, the degree of which depended on treatment. For PTH treatment at year 1 and for ALN-ALN treatment at year 2, there were significant negative and positive strength effects, respectively, associated with a change in external bone geometry. Average changes in strength per treatment group were somewhat consistent with average changes in total hip areal BMD as measured by DXA, except for the PTH group at year 1. The relation between change in femoral strength and change in areal BMD was weak (r2 = 0.14, pooled, year 2). We conclude that femoral strength changes with these various treatments were dominated by trabecular changes, and although changes in the cortical bone and overall bone geometry did contribute to femoral strength changes, the extent of these latter effects depended on the type of treatment.
The PTH and Alendronate (aka “PaTH”) study investigated the effects of PTH(1-84), alendronate, and their combination on BMD in postmenopausal, osteoporotic women,[1, 2] in which the main outcomes were measures of BMD and mass by both DXA and QCT. One finding from the QCT data was a differential effect of treatment on density and mass changes within the cortical versus trabecular bone. However, it is difficult to interpret these differential effects in terms of biomechanical significance and changes in bone strength because of the structural complexity of the proximal femur and poorly understood nature of the load sharing between the cortical and trabecular bone.[3-5] Because bone fracture is ultimately a biomechanical event, providing an integrative measure of proximal femoral strength for these various treatments and relating such treatment-induced changes in femoral strength to the underlying alterations in the density, mass, and geometric properties of the cortical and trabecular bone may provide insight into differentiation of these various treatments. Clinically, this would be important because head-to-head fracture outcome studies of multiple treatments will likely never be performed because of the great expense of performing such studies.
Whereas biomechanical properties of the hip can be addressed in clinical studies by specialized processing of DXA[6-10] and QCT[3, 4, 11] scans, finite element analysis of hip QCT scans[12-16] represents the most rigorous technique currently available in a research clinical setting for noninvasive strength analysis of the proximal femur. Finite element analysis is a well-established engineering method for analysis of complex structures and has been used in orthopedic biomechanics research for decades. For the hip and spine, cadaver studies have consistently shown such finite element analysis of QCT scans provides mechanistic, robust estimates of whole bone strength[12, 13, 19-22] that are better than those provided empirically by BMD as measured by either DXA or QCT.[12, 19, 22] More recently, this analysis technique has been applied in clinical studies to provide unique insight into vertebral and hip fracture etiology,[23, 24] vertebral strength changes in response to teriparatide and alendronate,[25, 26] and femoral strength changes in response to gluccocorticoids.
The goal of this study was to apply finite element analysis of the hip QCT scans in the PaTH study to provide insight into the strength changes associated with these various treatments and the relative roles of changes in the cortical and trabecular compartments and femoral geometry on such strength changes. Specifically, we sought to (1) compare average strength changes at the hip for the different treatments in the PaTH study at 1 and 2 yr, using finite element analysis of QCT scans and (2) determine the contribution of alterations in the trabecular and cortical bone and femoral geometry to these femoral strength changes. This study is unique by using finite element analysis in a clinical study to address how four different treatments for osteoporosis affect femoral strength and by relating such treatment-induced changes in femoral strength to the underlying alterations in the densitometric properties of the cortical and trabecular bone.
MATERIALS AND METHODS
Described in more detail elsewhere,[1, 2] the PaTH study was a multicenter, randomized, double-blind trial comparing the effects of PTH(1-84) versus alendronate versus a combination of the two in 238 postmenopausal, osteoporotic women (55–85 yr of age). The entry criteria were a T-score from DXA of less than –2.5 at the femoral neck, total hip, or spine or a T-score of less than –2.0 at any of these sites and either age >65 yr, a previous postmenopausal fracture, or a maternal hip fracture. Women were excluded if they had any condition known to affect bone metabolism or had been on bisphosphonates for either 12 mo total in the past or >4 wk over the previous 12 mo before enrollment. Treatments during year 1 were either PTH plus a placebo that matched the alendronate (PTH), alendronate plus a placebo that matched the PTH (ALN), or a combination of PTH plus alendronate (CMB). All participants received daily doses of calcium and vitamin D. After 1 yr of these treatments, the PTH group was randomly split into two groups for a second year of treatment, with one group given alendronate and the second given placebo (PLB); women in the other two groups received alendronate for the second year. There were ∼60 subjects in each of the four groups, and the study lasted 2 yr. Changes in BMD and mass, as measured by both DXA and QCT, have been reported elsewhere for the 1- and 2-yr intervals.
Whereas DXA scans were taken (Hologic QDA-4000A or Delphi Densitometer) for areal BMD measurement for all participants, QCT scans were taken for only 178 of the participants from three of the four sites. QCT scans (helical or contiguous slice, depending on the scanner used) were taken with the subject's hip overlying a solid hydroxyapatite phantom (Image Analysis, Columbia, KY, USA) for subsequent image calibration and using nominal acquisition settings of 80 kVp and 280 mAs. Images were reconstructed at 3.0-mm sections with in-plane pixel sizes ranging from 0.781 to 0.938 mm, depending on field of view. Full sets of finite element data, QCT data, and DXA data for each subject were available for 162 subjects after appropriate quality control analysis.
Using custom code (O.N. Diagnostics, Berkeley, CA, USA), the QCT images were processed and converted into finite element models using 1.5-mm-sided, cube-shaped, eight-noded brick elements (Fig. 1). A uniform threshold value was used to segment all images and any discontinuous edges were semiautomatically filled by the analyst for masking purposes. In the remaining images, cortical bone and trabecular bone were defined on the basis of their calibrated density values (see below). Isotropic material properties were assigned to all elements by converting the calibrated density to material properties using empirical relations,[27-29] with different assignments made to cortical versus trabecular bone. All bone material was assigned a higher strength in compression than tension[30, 31] and was modeled using a von Mises type elastic-perfectly-plastic material with tension-compression strength asymmetry. Trabecular elastic and strength properties were adjusted by a factor of 1.28 to account for the inevitable side artifact errors that occur during the biomechanical testing of cadaveric trabecular bone cores used to establish the empirical relations between density and mechanical properties.[33, 34] Similar to what has been used in a number of previous cadaver studies,[3, 12, 35-38] boundary conditions were applied to simulate a fall to the side of the hip, the diaphysis angled at 10° with respect to the ground with 15° of internal rotation. This represents a severe, unprotected fall to the side of the hip, which is known to be associated with a high risk of fracture.[39, 40] The models for each of the two follow-up times (years 1 and 2) were registered to the baseline model using a semiautomated routine. After creation of the finite element models, the nonlinear stress analyses were performed using custom finite element code. Strength was calculated from the resulting nonlinear force-deformation curve as the force at 4% deformation of the femoral head with respect to the greater trochanter (Fig. 1). The measurement precision error in this process, as assessed by repeat analysis of 24 randomly chosen baseline scans, was 1.51%. All image processing and finite element analysis was performed blinded to treatment by two analysts (PFH and MS), and results were sent to UCSF for statistical analysis (LP).
To gain insight into the relative role of changes in the cortical and trabecular bone on overall femoral strength changes, we performed a series of controlled parameter studies on each patient-specific finite element model. Cortical bone was defined as any bone appearing in the QCT scan having an apparent density >1.0 g/cm3 (equivalent to a QCT-calibrated mineral density of >693 mg/cm3), and trabecular bone was defined as any bone with an apparent density <1.0 g/cm3. We defined the peripheral compartment as all cortical bone plus any trabecular bone within 3 mm of the periosteal surface and the trabecular compartment as all remaining trabecular bone. To assess femoral strength changes caused only by alterations in the peripheral compartment, we held the density of the trabecular compartment fixed in all bones and over all time points. To do this, we assigned uniform material properties to the trabecular compartment that were the same across all bones and over time, in effect creating a surrogate trabecular compartment but without altering the peripheral compartment in any way. The strength of the femur containing this surrogate trabecular compartment was computed (PERIPH) at each time point. We note that geometry changes could still occur over time so this strategy did not separate out density and geometry effects. In another variation, we created a surrogate peripheral compartment, but did not constrain the trabecular compartment in any way, and computed a TRAB strength. In this case, separate uniform material properties were assigned to the cortical and trabecular bone within the surrogate peripheral compartment. To address changes in strength only associated with changes in bone geometry, we also created models with a single uniform material property assigned throughout the entire femur and over time, such that strength changes of those models would only be caused by changes in external geometry. This model was referred to as the GEOM model. A somewhat similar strategy was used previously in an analysis of vertebral strength in which the cortical bone was peeled off the vertebra to compute a trabecular strength. However, analysis for the femur is more complex because it is not possible to simply peel away the cortical bone without substantially altering load transfer patterns, and thus this surrogate compartment approach was used instead. In addition, we measured the BMD and mass (BMC) of the cortical bone, the trabecular bone, the cortical and trablecular bone combined (“integral” density and mass), the peripheral compartment, and the trabecular compartment, sampling all bone to a distance of one head diameter distal to the femoral head center. On average, the cortical bone, trabecular bone, peripheral compartment, and trabecular compartment in this volume of interest made up ∼10%, 90%, 58%, and 42%, respectively, of the total femoral bone mass.
The main outcome parameter was the strength of the proximal femur (denoted, femoral strength), and from this, percent changes versus baseline were calculated at the year 1 and year 2 follow-ups. Secondary outcomes were the strength changes for the peripheral and trabecular compartments, the geometric strength, the various QCT density and mass measures, and areal BMD (total hip) as measured by DXA. Percent changes with respect to baseline were compared across treatment groups using ANOVA and were compared against zero and between each other using the general least squares means procedure. Linear regression analysis was used to determine relations between the primary and various secondary variables. For brevity, the latter correlation analysis was restricted to the year 2 data. All significance levels are reported at the p < 0.05 level.
Baseline femoral strength values, on average, varied between 2494 and 2592 N (p = 0.91) across the four (year 2) treatment groups, with individual values ranging from 1164 to 4187 N. Overall, the average percent changes in femoral strength from baseline for any of the treatment groups at year 1 were in the range 2.08–3.60% depending on treatment and at year 2 were in the range 1.17–7.74% (Fig. 2). At year 1, the percent changes across treatments were not statistically different from each other (p = 0.67), although the changes for PTH (2.08%, p < 0.05) and ALN (3.60%, p < 0.01) were statistically different than zero, and the change for the CMB group was numerically larger than for PTH and nearly reached statistical significance (2.50%, p = 0.09). At 2 yr compared with baseline, although effects across treatments did not reach statistical significance (p = 0.06), the changes compared with baseline were negligible for PTH-PLB (1.17%, p = 0.51) but were highly significant for the other three treatment groups (PTH-ALN, 7.74%, p < 0.0001; CMB-ALN, 4.18%, p < 0.05; ALN-ALN, 4.83%, p < 0.01). In all cases, there was much variability in the percent changes on a patient-specific basis, as indicated by the appreciable 95% CIs. Statistical analysis for all treatments pooled indicated significant but weak negative correlations between baseline femoral strength and percent changes in femoral strength (versus baseline) at year 1 (r = −0.19, p < 0.05) and year 2 (r = −0.21, p < 0.01).
At year 1, there was a significant decrease in geometry-associated strength for PTH treatment and loss of cortical mass (Table 1). Treatment with PTH resulted in decreased density (−1.32%) and mass (−6.24%) of the cortical bone and increased density of the trabecular compartment (2.57%), but no significant changes in integral bone density or mass as measured by QCT or in total hip areal BMD as measured by DXA. Thus, the significant increase in strength at year 1 for PTH was attributable to the increase in density of the trabecular bone, despite the unique loss in mass and density of the cortical bone that occurred for this treatment. In contrast, treatment with ALN resulted in a small but significant increase in strength of the peripheral compartment, which was similar in magnitude to the nonsignificant increase in strength of the trabecular compartment, both of which contributed approximately equally to the overall increase in femoral strength. CMB therapy fell between the PTH and ALN treatment for all measures and showed no trends for any decrease in cortical density or mass, and, as with ALN treatment but unlike PTH treatment, changes in total hip BMD by DXA were similar to the overall change in femoral strength.
At year 2 (compared with baseline), there were significant changes in strength of the trabecular compartment for the three treatments that showed increases in overall femoral strength (PTH-ALN, CMB-ALN, and ALN-ALN), no significant strength changes associated with the peripheral compartment, and a significant geometric effect for the ALN-ALN group (Table 2). Densitometric analysis indicated no significant changes in cortical density for any treatment group. There was a significant decrease in density of the peripheral compartment for the PTH-PLB group, which is indicative of a decrease in density of the trabecular bone close to the periosteal surface because cortical density did not decrease. This had no significant biomechanical effect, as indicated by the small and nonsignificant reduction in strength of the peripheral compartment for this group. For the other three groups, increases in density of the trabecular compartment were greater than those of the overall trabecular bone, indicating the greater changes seen on the more central trabecular bone. For the ALN-ALN group, increases in mass were generally greater than increases in density, indicative of an increase in bone volume. This trend was not seen in the other groups and is consistent with the significant effect in the geometric strength (GEOM) seen only for the ALN-ALN group. In contrast to year 1, changes in areal BMD by DXA at year 2 tended to be lower than the strength changes and some trends moved in opposite directions. For example, for continuous ALN treatment, areal BMD indicated changes of 3.23% and 3.12% at years 1 and 2, respectively, compared with changes in femoral strength of 3.60% and 4.83%.
At year 2, despite a consistency seen in the trends for average changes in femoral strength and average changes in areal BMD by DXA across treatment groups at this time point (Table 2), the amount of variation in femoral strength explained by changes in areal BMD was very low (1–37% across all groups; 14% pooled), and the correlation between changes in areal BMD and changes in femoral strength on a patient-specific basis was not statistically significant for two of the groups (Table 3). Changes in integral bone mass (as measured by QCT) were strongly related to changes in femoral strength (67–81% across all groups; 75% pooled) but were only weakly correlated with changes in total hip BMC as measured by DXA (r2 = 0.10 pooled, p < 0.0001).
Regarding the role of cortical and trabecular changes on overall femoral strength changes, the 2-yr changes in femoral strength were more highly correlated with changes in trabecular bone than in cortical bone, and the degree of association differed between treatments (Table 3). At year 2, changes in strength of the trabecular and peripheral compartments explained between 76–90% and 53–69%, respectively, of changes in femoral strength depending on treatment group. There were no significant relations between changes in femoral strength and changes in external geometry for the PTH-PLB and PTH-ALN groups, but there was a weak but significant correlation for the other two groups. Changes in trabecular density and/or mass had a greater association with changes in femoral strength than did changes in cortical density or mass. Changes in cortical density had no role in femoral strength changes for three groups, but did have a significant role for the PTH-PLB group, explaining 27% of the observed variance in femoral strength for that group. In contrast, changes in cortical mass were related to femoral strength changes in all four groups, explaining 22–59% of the femoral strength variations.
We have shown previously using both DXA and QCT that the various treatment groups in the PaTH study were associated with complex patterns of density changes for the cortical and trabecular bone.[1, 2] It is difficult to interpret such densitometric outcomes in terms of whole bone femoral strength because of the biomechanical complexity of the proximal femur—particularly for sideways fall loading conditions—and also because of the poorly understood relative biomechanical role of the cortical and trabecular bone. With this in mind, our primary goal in this study was to use finite element analysis of the hip QCT scans to provide insight into the femoral strength changes associated with these various treatments and the relative roles of changes in the cortical and trabecular bone and geometry on such femoral strength changes. At the 2-yr follow-up, we found changes in average femoral strength ranging from 0.78% to 7.62% across the various treatments, although some individual subjects displayed much greater gains and losses. The large difference between the femoral strength at year 2 between the PTH-PLB and PTH-ALN groups reinforces our prior conclusions regarding the importance of following treatment with PTH by an antiresorptive agent to better preserve the gains of the PTH treatment. At year 2, average changes in areal BMD by DXA across treatment groups were consistent with the femoral strength average changes but the correlations between these various outcomes on a patient-specific basis were generally quite weak, and such relations were not even statistically significant for two of the groups. Furthermore, at year 1, areal BMD by DXA did not show a significant treatment effect for PTH treatment, although the finite element–based biomechanical analysis did. This is because the strength gains associated with changes in the trabecular bone outweighed the strength losses associated with changes in the cortical bone, whereas the opposing densitometric effects of the cortical and trabecular bone were simply averaged with both the DXA and QCT densitometric analyses. Overall, femoral strength changes with these various treatments at both time points were dominated by trabecular changes, and although changes in the cortical bone and overall bone geometry did contribute to femoral strength changes, the extent of these latter effects depended on the type of treatment.
Whereas no noninvasive method can provide exact measures of bone strength, these types of finite element–based biomechanical analyses provide a noninvasive clinical assessment of femoral strength that is based on an integrative mechanical analysis of the 3D density and structure information in the QCT scan in combination with the relevant biomechanics. Conceptually, finite element analysis of QCT scans can be thought of as a method to optimize the information content of the QCT scan from a biomechanical perspective. Originally applied clinically to the spine well over a decade ago, the technique is evolving, as is the underlying CT technology. The range of our baseline values for femoral strength for a sideways fall (∼1000–4000 N) is consistent with measured cadaver values for similar types of loading conditions in a similar demographic having low BMD.[3, 35-38] Although there are no published clinical data yet available to validate their ability to prospectively predict fractures in surveillance or drug studies, finite element analysis of QCT scans has shown success in discriminating fracture subjects in case-control studies for the spine[23, 41] and has provided unique insight for understanding treatment effects at the spine[25, 26] and the hip. A unique feature of our analysis was the use of controlled variations on each patient-specific model to provide metrics of strength changes caused only by changes in the trabecular and peripheral compartments and to identify any overall geometric effects. This controlled variation technique has been used to provide insight into treatment effects of teriparatide and/or alendronate at the spine.[25, 26] Coupled with analysis of density and mass changes in the trabecular bone, cortical bone, trabecular compartment, and peripheral compartment and as applied to six different treatments—three in year 1 and three variations of these treatments in year 2—these analyses provided unique insight into local treatment effects. For example, the results showed the complex interaction between trabecular and cortical changes for PTH treatment and identified a geometric strengthening effect for ALN. Given that our estimates of femoral strength have inherent uncertainties and that this technology is still evolving, our approach is most appropriately considered as a best-available technique using noninvasive imaging methods for the clinical assessment of treatment effects on whole bone strength behavior.
Despite the sophistication of these finite element models, a paucity of clinical data has prevented the opportunity for validation in any placebo-controlled drug treatment studies. The major caveat for this analysis technique when applied to a drug study—which is also applicable to any DXA or QCT analysis—is that the models contain no effects of possible submillimeter changes in the bone microarchitecture, remodeling space and “stress risers”, microdamage, or collagen and/or mineralization properties. As discussed in more detail elsewhere, there is little evidence that either PTH or ALN cause such effects to any appreciable biomechanical extent. One iliac crest biopsy study on postmenopausal women treated with risedronate showed no fundamental alteration in trabecular microarchitecture, and another biopsy study on teriparatide showed increased connectivity, plate-like structures, and cortical thickness. Monkey studies have shown effects for PTH(1-34) such as increased cortical porosity and added trabecular bone mass—effects that were detected in this study using clinical CT scans—and a recent study using non-osteoporotic dogs showed a subtle difference in the vertebral strength-to-areal BMD ratio for raloxifene treatment but no such effect for alendronate or risedronate. One human study has shown evidence of osteoporosis-related changes in the microarchitectural anisotropy ratio and transverse mechanical properties of femoral head trabecular bone,[47, 48] a subtle effect that was not included in our models but that likely would not affect comparisons between treatments. Possible treatment effects on changes in marrow content could bias density measurements,[49, 50] but any such possible effects are poorly understood and were therefore not included in the analysis. In general, at this juncture, there is little evidence suggesting any appreciable bone quality effects with the types of treatments investigated in this study, although further research is required to directly validate the ability of finite element analysis of QCT scans to correctly describe treatment-induced changes in femoral strength.
A more technical caveat of our approach is that our image processing and registration methods were not fully automated and thus introduced some random error in estimation of changes in strength between time points. This contribution to imprecision in our estimates of the average responses decreased our ability to detect statistically significant differences between treatments at both follow-ups. However, such precision errors are expected to decrease as the image-processing techniques continue to evolve[51-54] and are unlikely to affect our estimates of average changes in strength across treatments because any such errors would be unbiased and random in nature. The magnitude of the heterogeneity in responses that was observed between subjects is too large to be dominated by precision errors or acquisition limitations and suggests that some subjects responded very well to the various treatments, but some much less so. The negative correlations found between the percent changes in femoral strength and baseline femoral strength are consistent with clinical findings showing trends for a greater efficacy of alendronate for hip fracture for patients having lower areal BMD scores. However, the weak nature of these correlations indicates that baseline femoral strength alone is not useful in predicting the future response of any individual patient. Finally, the scan acquisition protocol, with 3-mm slice reconstruction, was relatively coarse, particularly for detection of subtle cortical changes, and the images generally were noisy. It is not clear at this juncture how such limitations with the scan acquisition might affect our reported outcomes, particularly regarding measures of cortical behavior, and this is an area of ongoing research. Realizing this, our “cortical” strength analysis included neighboring trabecular bone by our use of the peripheral surrogate compartment approach and was not intended to represent the true cortical bone per se.
Despite reasonable agreement between changes in areal BMD and changes in femoral strength at year 2, overall, there were some notable discrepancies. First, areal BMD failed to show a treatment effect for PTH at year 1, an effect explained by its averaging of cortical and trabecular effects, whereas the finite element analysis indicated a preferential strength effect associated with the trabecular changes. Second, whereas the finite element analysis showed numerical increases in femoral strength with continuous ALN treatment between years 1 and 2—because of the geometric strength effect—areal BMD showed a slight numerical decrease between these time points. Third, the correlations between changes in areal BMD and changes in femoral strength on a patient-specific basis were very weak and were even nonsignificant for two of the treatment groups. Such findings are consistent with our prior study of alendronate effects at the spine. Taken together, these findings raise questions as to the optimal methodology for individual patient monitoring of treatment effects in a clinical setting, particularly for any treatments that might have differential effects on the cortical and trabecular bone or that might affect femoral geometry.
One unique result from this study is the finding that changes in femoral strength with treatment were more sensitive to changes in the trabecular bone than cortical bone, although there was a significant influence of the cortical bone that depended on the type of treatment. At year 2, for example, changes in cortical density played a significant role in femoral strength changes only for the PTH-PLB group. In contrast, changes in cortical mass played a significant role for all groups, although the strength of the correlation depended on treatment. Consistent with these results, the previous QCT densitometric analysis of this cohort noted more significant treatment effects on average for cortical mass than for cortical density. This study provides unique insight into such treatment effects by relating these changes in density and mass of the various compartments to changes in femoral strength. It is interesting to note, for example, that changes in cortical density did play a significant role in femoral strength changes for one of the treatments, even though the cortical bone only comprised about 10% of the integral bone mass. We also note the distinction between a treatment effect on average values of cortical density—which we did not observe in this analysis at year 2—versus a treatment effect on the degree of association (r2 value) between changes in femoral strength and changes in cortical density—an effect we did observe. The latter is not possible to determine without the use of finite element analysis, whereas the former is difficult to interpret without finite element analysis. These findings show how finite element analysis of QCT scans can provide unique insight beyond densitometric analysis alone of the QCT images.
Another important issue is the potential of a treatment to cause biomechanically significant changes in geometry. In this regard, one notable observation was our finding of a small but statistically significant geometric strength effect observed at year 2, an effect observed only for the ALN-ALN treatment group. On average, this geometric effect was about one half the magnitude of the effect seen for overall femoral strength for the ALN-ALN group, but on a specimen-specific basis (Table 3), the changes in geometric strength were associated with only 11% of changes in femoral strength. These results indicate the complexity of the associated biomechanics and highlight the interaction between the biomechanical effects of geometry and density changes. The potential role of geometry changes on strength changes is poorly understood, as is the actual existence of any geometry changes with treatment. For example, one study that used structural analysis of DXA scans reported a significant increase (versus baseline) in the outer diameter of the femoral neck and intertrochanteric regions after 3 yr of treatment either with placebo (calcium and vitamin D), hormone replacement therapy (HRT), alendronate, or combined HRT + alendronate. Those findings suggest that any external geometric changes in the proximal femur do not depend on treatment. However, other derived geometric indices from that analysis, such as cross-sectional area and moment of inertia, did depend on treatment, in apparent contradiction to the results on outer diameter. DXA scans are clearly limited for analysis of 3D structures, particularly in treatment studies in which changes are typically small, and QCT studies are just now beginning to be reported for such treatment effects. The exact nature of the significant geometric strength effect seen for continuous alendronate treatment was not explored in our study and represents an interesting topic for further investigation. Likewise, much new insight into treatment effects should be forthcoming by combining the type of finite element analyses presented here with geometric and densitometric analysis of QCT and DXA scans.
Funding was provided for the original PaTH study by NIAMS N01-AR92245 and for this finite element analysis by Merck and NPS, with partial support also from NIH AR49828 and AR053986. We also acknowledge all the other PaTH Study Investigators who made this study possible. All QCT imaging was supervised by Dr Thomas F. Lang. TMK has a financial interest in O.N. Diagnostics and both he and the company may benefit from the results of this analysis.
- 172006 An Introduction To The Finite Element Method, 3rd ed. McGraw-Hill Higher Education, New York, NY, USA.
- 262008 Vertebral strength changes in rheumatoid arthritic patients treated with alendronate as assessed by finite element analysis of clinical CT scans. Arthritis Rheum (in press)., , , , , , , , , ,
- 321990 Plasticity Theory. Macmillan, New York, NY, USA.
- 432002 Risedronate preserves trabecular architecture and increases bone strength in vertebra of ovariectomized minipigs as measured by three-dimensional microcomputed tomography. J Bone Miner Res 17: 1139–1147., , , , ,