By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
Dr. Schwarz has received consulting fees, speaking fees, and/or honoraria from MedImmune, Regeneron, Codevax, and LAGeT (less than $10,000 each) and from DePuy and Johnson & Johnson (more than $10,000), and he has served as an expert witness on behalf of Smith & Nephew.
Glucocorticoid (GC) therapy is associated with increased risk of fracture in patients with rheumatoid arthritis (RA). To elucidate the cause of this increased risk, we examined the effects of chronic erosive inflammatory arthritis and GC treatment on bone quality, structure, and biomechanical properties in a murine model.
Mice with established arthritis and expressing human tumor necrosis factor α (TNFα) transgene (Tg) and their wild-type (WT) littermates were continually treated with GC (prednisolone 5 mg/kg/day via subcutaneous controlled-release pellet) or placebo for 14, 28, or 42 days. Microstructure, biomechanical properties, chemical composition, and morphology of the tibiae and lumbar vertebral bodies were assessed by micro–computed tomography, biomechanical testing, Raman spectroscopy, and histology, respectively. Serum markers of bone turnover were also determined.
TNF-Tg and GC treatment additively decreased mechanical strength and stiffness in both the tibiae and the vertebral bodies. GC treatment in the TNF-Tg mice increased the ductility of tibiae under torsional loading. These changes were associated with significant alterations in the biochemical and structural composition of the mineral and organic components of the bone matrix, a decrease in osteoblast activity and bone formation, and an increase in osteoclast activity.
Our findings indicate that the concomitant decrease in bone strength and increase in bone ductility associated with chronic inflammation and GC therapy, coupled with the significant changes in the bone quality and structure, may increase the susceptibility of the bone to failure under low-energy loading. This may explain the mechanism of symptomatic insufficiency fractures in patients with RA receiving GC therapy who do not have radiographic manifestations of fracture.
Nearly 1.3 million Americans are affected by the debilitating autoimmune inflammatory disease rheumatoid arthritis (RA). To ease the pain of inflammation, RA patients are routinely prescribed glucocorticoid (GC) therapy, which unfortunately, leads to bone loss and increases the incidence of symptomatic insufficiency fractures, with rates reaching 30–50% (1). RA has been demonstrated to be an independent risk factor for fractures, which in 70% of the cases, could be classified as insufficiency fractures (2, 3). Changes in systemic bone metabolism associated with RA could be attributed to proinflammatory cytokines, such as tumor necrosis factor α (TNFα) (4–7), which are released from local inflammatory lesions into the systemic circulation. Previous studies have demonstrated that TNFα overexpression significantly increases osteoclast activity and decreases osteocalcin levels in mice (8). Furthermore, Diarra et al (9) demonstrated a significant increase in Dkk-1 level in the TNF-Tg mice, which implicates a deficiency in bone formation in this model. Collectively, these observations indicate that systemic inflammation involving overexpression of TNFα leads to a decoupling of osteoblast-mediated bone formation and osteoclast-mediated bone resorption, effecting an imbalance in bone metabolism toward bone loss.
Administration of GCs has also been identified as a substantial risk factor for fragility fracture (2, 10). Numerous studies have demonstrated that GCs have a strong negative effect on bone metabolism, leading to deterioration in bone quantity and quality. The pathogenesis of GC-induced osteoporosis has also been attributed to an imbalance in bone turnover; however, there are multiple factors influencing bone metabolism that confound our ability to identify the underlying mechanisms that lead to bone fragility. These factors include age, sex, severity and duration of disease, level of physical activity, and the concomitant use of GCs and other immunosuppressive agents (11). These factors are not independently presented in clinical cases of RA or GC-induced osteoporosis, however, which confounds our ability to clinically assess fracture risk in these patients.
In this study, we sought to investigate how chronic inflammation leads to fracture risk and how GC therapy increases this risk in the setting of erosive inflammatory arthritis. We hypothesized that the mechanism of the additive effect of GCs on fracture risk in patients with RA encompasses reductions in bone strength due to decreasing bone quantity (e.g., due to osteoclastic erosion) as a result of the inflammatory condition, and deterioration in bone quality at the microstructural and compositional levels as a consequence of the GC therapy. We also sought to understand why RA patients receiving long-term GC therapy have a higher risk of insufficiency fracture at various sites of the skeleton that are less common in primary osteoporosis.
To address these questions, an established transgenic mouse model that expresses human TNFα transgene (TNF-Tg) was chosen as a model of erosive inflammatory arthritis for this study based on the observation that these mice develop generalized osteoporosis, severe local erosion of cartilage and bone, and periarticular osteopenia similar to the clinical presentation of RA (5, 6, 8, 12, 13). Specifically, we investigated the effects of prolonged overexpression of TNFα, GC treatment, and the combination of these 2 factors on bone quality in polyarthritic TNF-Tg mice and healthy wild-type (WT) mice by means of histology, serum analysis, Raman spectroscopy, micro–computed tomography (micro-CT) imaging, and detailed biomechanical analysis of the long bone diaphysis (midshaft of the tibiae) and the flat bones (vertebral bodies).
MATERIALS AND METHODS
Animals and experimental design.
All animal studies were performed in accordance with protocols approved by the Committee on Animal Resources at the University of Rochester. Experiments were performed with male TNF-Tg mice and WT littermate controls. Based on our previous study on the natural history of joint inflammation in this transgenic mouse (14), we initiated GC administration at the age of 5 months, when the mice invariably develop severe joint arthritis.
Five-month-old male mice received a subcutaneous implant of either placebo or prednisolone pellets (Innovative Research of America), which release GCs at a dosage of 5 mg/kg/day, as previously described (15). The mice were divided into 4 groups (n = 12 per group). Groups 1 and 2 were WT littermates that received placebo and prednisolone, respectively. Groups 3 and 4 were TNF-Tg mice that received placebo and prednisolone, respectively. The mice were euthanized at 14, 28, or 42 days post pellet implantation, respectively (n = 4 per time point per treatment group), and blood was immediately collected from the vena cava during autopsy. Serum was stored at −80°C until biochemical markers analysis. The left tibia was examined by Raman spectroscopy and then stored at –80°C until micro-CT analysis and subsequent biomechanical testing. To determine the baseline (day 0) values of all study parameters, a separate set of 8 TNF-Tg mice and 8 WT littermates were euthanized at the age of 5 months.
Raman spectroscopic analysis of bone chemical composition.
Raman spectroscopy was used to measure the biochemical composition of each tibia, as previously described (16, 17). Briefly, spectra were acquired from the medial side of the proximal, distal, and middiaphyseal regions of the excised tibiae, with an exposure time of 300 seconds per region. A locally constructed Raman spectroscopy system delivered approximately 80 mW of 830-nm excitation light to a 1.5-mm diameter spot on the surface of the bone. Following acquisition, the spectra were background corrected, smoothed, and normalized to the area under the amide I peak near 1660 cm−1. A number of metrics related to bone biochemistry were calculated, including the mineral-to-matrix ratio (MTMR; PO43–/amide I, 960 cm−1/1660 cm−1 peak area ratio), which describes the degree of phosphate mineralization, the carbonate-to-phosphate ratio (CTPR; CO32−/PO43−, 1070 cm−1/960 cm−1 peak area ratio), which describes the amount of carbonate substitution in the apatite crystal lattice, and the 1660 cm−1/1690 cm−1 intensity ratio, which indicates the ratio of mature to immature collagen crosslinks. The PO43–, CO32–, and amide I peak areas were calculated by summing the Raman intensity between 900 cm−1 and 990 cm−1, 1040 cm−1 and 1120 cm−1, and 1630 cm−1 and 1730 cm−1, respectively.
Micro-CT bone structure analysis.
Left tibiae and L2 vertebral bodies were scanned and measured individually by micro-CT (VivaCT 40; Scanco Medical) at a 10.5μ isotropic resolution, using an integration time of 300 msec, energy of 55 kVp, and intensity of 145 μA. Thresholds for segmentation of trabecular bone and cortical bone as determined a priori were 375 mg and 423 mg of hydroxyapatite (HA)/cm3, respectively, based on comparing the grayscale images with the binary images per published standards (18).
The whole bone mineral density (BMD; mg of HA/cm3) of each tibia was measured in the proximal half of the bone. A volume of interest of 502.5 μm (50 slices) encompassing the region of the proximal metaphysis distal to the growth plate was used to assess trabecular bone morphology. For L2 vertebral bodies, the centrum of each specimen, from cephalad to caudal end plates and inside the endosteal margin, was used to assess trabecular bone. Cortical bone was assessed over 200 μm–thick regions (19 slices) of the tibial middiaphysis and the midlevel of the L2 vertebral body, and the results for the tibia and the L2 vertebral body were averaged over these slices.
Trabecular bone parameters of the tibial metaphysis and vertebral body, including the trabecular bone volume fraction (BV/TV), connectivity density (ConnD), structural model index (SMI), trabecular number (TbN), trabecular thickness (TbTh), and trabecular separation (TbSp), were determined using 3-dimensional analysis tools (direct model) from Scanco Medical (18). For cortical bone analysis of the tibiae, the tissue mineral density (TMD; calibrated to hydroxyapatite, as mg of HA/cm3), cortical thickness (CtTh), and polar moment of inertia (J) were determined using the manufacturer's analysis tools.
The biomechanical properties of the proximal one-third of the tibial diaphysis were measured by torsional testing, as previously described (19). The ends of the tibiae were cemented into square 6.35-mm2 aluminum tube holders using polymethylmethacrylate (PMMA) and were then hydrated and mounted on an EnduraTEC TestBench system (200 Nmm torque cell; Bose-EnduraTEC) and tested in torsion at a rate of 1°/second. The data were used to determine the yield and maximum torques, torsional rigidity, energy to yield and maximum torques, and postyield rotation and energy. The postyield rotation, a measure of ductility, was defined as the normalized rotational deformation (radians of rotation divided by specimen gauge length) beyond the yield point up to the point of maximum load. The yield point was determined by the intersection of the torque-normalized rotation curve with a line parallel to the linear elastic region, offset by 0.0007 rad/mm (∼0.065% strain), based on standard recommendations (20).
The L2 vertebral bodies were tested in compression according to a modification of published methods (5). Briefly, the end plates of the vertebral body were cemented onto small plastic plates using PMMA to ensure axial alignment of the bodies and even load distribution. The specimens were hydrated and then tested in compression, with a 1N preload, at a rate of 1 mm/minute until failure, using a DynaMight 8841 servohydraulic materials testing machine (Instron). Compression data were analyzed in a manner similar to that of the torsion data, with displacements normalized by the height of each vertebral body, to determine the corresponding mechanical parameters in compression.
Biochemical markers of bone turnover.
Sandwich enzyme-linked immunosorbent assays were performed using commercially available kits to determine serum levels of osteocalcin (Biomedical Technologies), Dkk-1 (R&D Systems), and tartrate-resistant acid phosphatase type 5b (TRAP-5b; Immunodiagnostic Systems). All samples were assayed in duplicate. A standard curve was generated for each protein, and the absolute concentrations were extrapolated from the standard curve.
Histology and histomorphometry.
The tibiae were fixed in 10% neutral buffered formalin, and decalcified in 10% EDTA. For each specimen, at least 2 nonconsecutive 3-μm paraffin-embedded midsagittal sections were stained for TRAP activity. Two nonconsecutive parasagittal sections that were at least 100 μm apart were measured in each animal, and the average values were used. Therefore, a total of 14 fields (0.875 mm2) per animal were analyzed. Histomorphometric analysis of osteoclast activity was performed using OsteoMetrics software. The number of osteoclasts per millimeter of trabecular bone surface (TbOcN/BS) and the number of osteoclasts on the endosteal cortical bone surface (ECtOcN/BS) were measured at the proximal metaphyses of the tibiae and the proximal tibial diaphyses, respectively.
TUNEL staining was performed on paraffin sections of the tibiae using an ApopTag peroxidase in situ apoptosis detection kit (Millipore) according to the manufacturer's instructions. To calculate the percentage of osteocyte apoptosis, an average of 1,000 osteocytes were counted from the cortical bone of the tibial diaphysis.
For dynamic bone formation analysis, calcein (10 mg/kg) and alizarin red (10 mg/kg) were injected subcutaneously 7 days and 12 days, respectively, before the mice were euthanized. Distal femurs were fixed in 10% neutral buffered formalin and subjected to undecalcified tissue processing. The specimens were dehydrated in a graded ethanol series, embedded in glycolmethacrylate, and sectioned at 5 μm in the coronal plane. Trabecular bone at the distal metaphysis of each femur was examined by fluorescence microscopy (AxioImager M1m; Carl Zeiss) to evaluate dynamic parameters of bone formation. Interlabel width, single-labeled surface (sLS) and double-labeled surface (dLS) perimeters, and total bone surface (BS) were measured. The mineral apposition rate (MAR), mineralized surface/bone surface (MS/BS), and bone formation rate/bone surface (BFR/BS) were calculated from the sLS, dLS, and BS according to the method of Parfitt et al (21).
Two-way analysis of variance tests were used to determine the statistical significance of the treatment (GC), the genotype (TNF-Tg), and the treatment–genotype interaction effect on each measured parameter. Student's t-tests were used to compare baseline values in the WT and TNF-Tg mice. P values less than 0.05 were considered significant.
Effects of prolonged overexpression of TNFα and GC treatment on bone biomechanical properties.
The torsional biomechanical properties of the proximal tibiae were evaluated at 0, 14, 28, and 42 days postimplantation of the pellet. In general, the tibiae from placebo-treated WT and TNF-Tg mice did not exhibit any significant changes over time, and the treatment-related differences were qualitatively similar at each time point. Thus, only day 42 data are discussed here in detail (Figure 1).
On day 42, the tibiae from GC-treated TNF-Tg mice were found to be significantly weaker, more compliant, and more ductile than those from the other 3 groups (Figure 1A). Compared to the placebo-treated WT mice, tibiae from GC-treated WT mice exhibited a significant decrease in yield and maximum torques (P < 0.05) (Figures 1B and C). Prolonged overexpression of TNFα independently decreased the yield and maximum torque as compared to tibiae from placebo-treated WT mice (P < 0.05). Treatment of the TNF-Tg mice with GCs further decreased the yield and maximum torque of the tibiae as compared to bones from placebo-treated WT mice (P < 0.05). However the interaction of prolonged TNFα overexpression and GC treatment were not significant (Figure 1H), indicating additive, but not synergistic, effects of these risk factors on long bone fragility. GC treatment had no significant effects on torsional rigidity (Figure 1D). However, prolonged overexpression of TNFα resulted in significant decreases in torsional rigidity of the tibiae compared to those from WT mice (P < 0.05) (Figure 1H), with no significant interaction effects. The energy to yield (Figure 1E), however, was significantly decreased by both prolonged overexpression of TNFα and GC treatment, with no significant interaction. The energy to maximum torque or toughness (Figure 1F) was decreased by GC treatment in WT mice, but was increased by GC treatment in TNF-Tg mice and showed a significant interactive effect. Postyield rotation was increased by GC treatment (Figure 1G), with no significant interaction effects.
Biomechanical properties derived from compression tests of L2 vertebral bodies on day 42 of treatment are shown in Figure 2. The maximum compressive force (Figure 2A) was decreased by GC treatment (P < 0.05), regardless of the genotype. Prolonged overexpression of TNFα also reduced the compressive strength (P < 0.05) compared to specimens from WT mice. However, there were no significant interaction effects on the maximum compressive force between prolonged overexpression of TNFα and GC treatment. Similar trends were observed for the yield force (Figure 2B), albeit significant reductions were observed only because of prolonged overexpression of TNFα (P < 0.05). Despite trends suggesting decreased compressive stiffness due to prolonged overexpression of TNFα, there were no significant effects of genotype or treatment (Figure 2C). Moreover, despite trends suggesting decreases in the energy to maximum force (Figure 2D) and postyield strain (Figure 2E) with both prolonged overexpression of TNFα and GC treatment, these effects were not statistically significant.
Effects of prolonged overexpression of TNFα and GC treatment on cortical and trabecular bone structure and volume.
The BMD of the proximal half of the entire tibia was significantly decreased by overexpression of TNFα, while the effects of GC treatment were not statistically significant (Table 1). The main effect of prolonged overexpression of TNFα had a statistically significant impact on the ConnD, TbN, TbTh, TbSp, cortical TMD, CtTh, and J values. GC treatment had a statistically significant effect on only the cortical TMD. None of the interaction effects were statistically significant.
Table 1. Micro–computed tomography parameters of the tibiae and L2 vertebral bodies in WT and TNF-Tg mice*
Values are the mean ± SD. WT = wild-type; TNF-Tg = tumor necrosis factor–transgenic; GCs = glucocorticoids; BMD = bone mineral density; HA = hydroxyapatite; BV/TV = trabecular bone volume fraction; ConnD = connectivity density; SMI = structural model index; TbN = trabecular number; TbTh = trabecular thickness; TbSp = trabecular separation; TMD = tissue mineral density; CtTh = cortical thickness; J = polar moment of inertia.
P values less than 0.05 were considered significant, as determined by two-way analysis of variance.
Tibia, proximal half
BMD, mg of HA/cm3
1,121 ± 31
1,099 ± 48
1,089 ± 18
1,072 ± 25
1,048 ± 16
1,038 ± 7
Tibial metaphysis, trabecular region
19.3 ± 4
11.1 ± 5.7
13.9 ± 2.6
14 ± 7
11.5 ± 6.2
9.9 ± 6.7
110 ± 36
81 ± 49
90 ± 16
78 ± 60
30 ± 9
44 ± 22
1.9 ± 0.5
2.4 ± 0.5
2.1 ± 0.3
2.3 ± 0.5
2.1 ± 0.7
2.6 ± 0.6
5.0 ± 0.4
4.6 ± 0.9
4.1 ± 0.2
4.4 ± 0.9
3.3 ± 0.1
3.9 ± 0.4
0.057 ± 0.004
0.043 ± 0.006
0.052 ± 0.006
0.052 ± 0.009
0.042 ± 0.006
0.037 ± 0.002
0.19 ± 0.02
0.22 ± 0.04
0.24 ± 0.02
0.23 ± 0.05
0.3 ± 0.01
0.25 ± 0.03
Tibial diaphysis, cortical region
TMD, mg of HA/cm3
1,240 ± 26
1,210 ± 23
1,270 ± 25
1,240 ± 34
1,234 ± 12
1,204 ± 23
0.22 ± 0.01
0.17 ± 0.01
0.21 ± 0.03
0.19 ± 0.02
0.16 ± 0.01
0.15 ± 0.02
0.26 ± 0.06
0.2 ± 0.04
0.33 ± 0.11
0.26 ± 0.08
0.16 ± 0.05
0.15 ± 0.03
L2 trabecular bone
22.9 ± 2.8
17.1 ± 2.8
21.3 ± 2.1
20.4 ± 4.5
12.3 ± 3.6
15.0 ± 1.2
243 ± 44
209 ± 45
173 ± 36
143 ± 25
126 ± 21
141 ± 27
1.0 ± 0.2
1.5 ± 0.2
0.9 ± 0.1
1.2 ± 0.3
1.3 ± 0.5
1.6 ± 0.2
5.3 ± 0.4
5.1 ± 0.4
4.7 ± 0.5
4.4 ± 0.3
4.2 ± 0.2
4.4 ± 0.3
0.048 ± 0.003
0.042 ± 0.002
0.050 ± 0.006
0.052 ± 0.006
0.044 ± 0.006
0.042 ± 0.002
0.18 ± 0.01
0.19 ± 0.02
0.21 ± 0.02
0.22 ± 0.02
0.23 ± 0.01
0.22 ± 0.02
L2 cortical bone
0.069 ± 0.007
0.062 ± 0.005
0.073 ± 0.009
0.066 ± 0.004
0.058 ± 0.008
0.054 ± 0.005
In L2 vertebral bone, prolonged overexpression of TNFα significantly decreased BV/TV and TbTh values and increased the SMI, but had no effect on the ConnD, TbN, or TbSp values (Table 1). GC treatment did not significantly influence any of these microstructural parameters. As with the tibiae, the CtTh of the L2 vertebral bodies was significantly decreased by overexpression of TNFα, while GC treatment induced no statistically significant effects despite trends toward decrease.
Effects of prolonged overexpression of TNFα and GC treatment on Raman spectroscopy parameters.
Overexpression of TNFα significantly, and GC treatment independently, decreased the MTMR of the proximal tibia, which indicates a reduction in the mineralization of the bone, and significantly increased the CTPR, which indicates the amount of carbonate substitution for phosphate or hydroxide in the mineral crystals (Table 2). These changes are consistent with the findings of other studies showing an increase in the CTPR in rodent bones due to aging (22). Overexpression of TNFα did not affect the 1660 cm−1/1690 cm−1 intensity ratio; however, GC treatment significantly increased this parameter. There were no interactive effects of prolonged overexpression of TNFα and GC treatment on any of the Raman spectroscopy parameters.
Table 2. Raman spectroscopic biochemical parameters of the tibiae*
The Raman parameters were normalized by the average value calculated among mice in the wild-type (WT) placebo-treated group. Normalization values were 3.1 for the mineral-to-matrix ratio (MTMR), 0.3 for the carbonate-to-phosphate ratio (CTPR), and 1.7 for the 1660 cm−1:1690 cm−1 intensity ratio. Values are the mean ± SD. TNF-Tg = tumor necrosis factor–transgenic; GCs = glucocorticoids.
P values less than 0.05 were considered significant, as determined by two-way analysis of variance.
1.02 ± 0.08
0.85 ± 0.22
1.00 ± 0.08
0.87 ± 0.07
0.87 ± 0.17
0.72 ± 0.04
0.99 ± 0.06
1.06 ± 0.12
1.00 ± 0.03
1.05 ± 0.03
1.04 ± 0.09
1.12 ± 0.02
1660 cm−1: 1690 cm−1 ratio
0.98 ± 0.11
0.99 ± 0.24
1.00 ± 0.02
1.10 ± 0.06
0.99 ± 0.14
1.13 ± 0.06
Effects of prolonged overexpression of TNFα and GC treatment on systemic bone metabolism.
Serum TRAP-5b levels were significantly increased in TNF-Tg mice as compared to WT mice at baseline (Figure 3). Prolonged overexpression of TNFα and GC treatment also increased TRAP-5b levels in both WT mice and TNF-Tg mice at 14 days of treatment. While the effect of overexpression of TNFα on TRAP-5b levels was also statistically significant on days 28 and 42, the GC treatment effects were not significant at these time points, suggesting that the GC treatment–induced osteoclast activation is transient. Joint synovitis and bone destructive changes were seen in all TNF-Tg mice (Figures 3B–M); however, these lesions were limited to the joint region and did not progress beyond the epiphysis (Figures 3H–M). The number of osteoclasts was slightly increased in TNF-Tg mice in both the trabecular bone and the cortical bone. GC treatment dramatically increased the number of osteoclasts on both trabecular bone surfaces and endocortical surfaces at 14 days of treatment, but these effects were lessened at 28 days of treatment (Figures 3N–Q).
Serum osteocalcin levels showed a trend, albeit insignificant, toward a decrease in TNF-Tg mice as compared to WT mice (Figure 4A), consistent with our previously published observations (8). On day 14, GC treatment dramatically decreased osteocalcin levels in both WT mice and TNF-Tg mice. This trend persisted in GC-treated animals throughout the observation period of 42 days. Serum Dkk-1 levels showed a trend opposite of that for osteocalcin. Dkk-1 was significantly elevated in TNF-Tg mice as compared to WT mice on day 0. GC treatment also significantly increased Dkk-1 levels in both WT mice and TNF-Tg mice (Figure 4B).
Analysis of bone formation rates from day 21 to day 26 in specimens obtained on day 28 revealed that double-labeled surfaces decreased in specimens from GC-treated WT and TNF-Tg mice (Figures 4C, D, E, and F). The distance between double labeling was shorter in GC-treated mice (Figures 4G, H, I, and J). The MAR (Figure 4K) was significantly decreased by GC treatment and the MS/BS (Figure 4L) was significantly decreased by both GC treatment and the TNFα transgene. Consequently, the BFR/BS, which is calculated from the MAR and the MS/BS, was suppressed in placebo-treated TNF-Tg mice as compared to placebo-treated WT mice, and it was also suppressed by GC treatment in both WT mice and TNF-Tg mice (Figure 4M). The percentage of apoptotic osteocytes was significantly larger in the GC-treated mice. However, the percentage of apoptotic osteocytes in the GC-treated TNF-Tg mice was only 2% on average (Figure 4N).
The results of this study show that prolonged overexpression of TNFα significantly decreases bone strength and that GC treatment has an additive negative impact on bone strength in the TNF-Tg mouse model of RA. These results are consistent with the findings in clinical studies showing that RA patients who are not taking GCs have an increased risk of fracture and that GC treatment independently increases fracture risk in RA patients (2, 23). The results of this study support our hypothesis by demonstrating that GCs mainly induce biochemical changes in bone; however, the mechanism of increased fracture risk caused by the erosive inflammatory arthritis appears to be more complicated than has been hypothesized. Prolonged overexpression of TNFα significantly decreased values of bone quality parameters, including trabecular and cortical bone architecture and the composition of the bone matrix, including the TMD, suggesting that prolonged inflammation has pleiotropic effects on bone metabolism, as reported previously (4, 7).
We observed different results from biomechanical testing of long bone diaphyses (midshaft tibiae) and flat bones (vertebral bodies) under different modes of loading (in torsion and compression, respectively). It should be noted that torsional testing of the tibiae was performed because it represents a mode of loading that commonly leads to long bone fractures in humans. However, qualitatively similar results would be expected from bending tests, as has been reported during skeletal growth (24) and in a model of skeletal fragility (25). In contrast to the increased ductility observed in the tibiae, the vertebral bones exhibited decreased ductility under compressive loading. In both compressive and torsional loading, the mineral phase primarily contributes to the stiffness (elastic modulus) and yield strength (26–28).
The mechanisms of plastic deformation (ductility) in bone are different between torsional and compressive loading, however (27, 29–31). Under tensile and shear stresses, as the bone yields and failure initiates in the form of microcracks through the mineral phase, the collagen matrix can act as a significant toughening mechanism against failure (30). Ductility in the compression of vertebral bodies largely arises from structural plasticity, which refers to the geometric displacement of composite materials in the cortical bone (i.e., slip planes) (31) as well as buckling of individual trabeculae (32). Thus, we propose that the increased yielding of the more ductile long bone without fracture may lead to microfractures and diffuse damage that are undetectable by radiography but can become symptomatic with pain and edema, which are the hallmarks of insufficiency fractures. Furthermore, deterioration in the bone matrix biochemical composition (MTMR and CTPR) as well as in the macrostructure, including cortical thickness, polar moment of inertia, and mineral density, likely contribute to the increased risk of fragility fracture in cortical bone (33).
In addition to changes in the degree of mineralization (MTMR) and the relative carbonate content (CTPR), we observed alterations in the 1660 cm−1/1690 cm−1 intensity ratio in GC-treated mice, as measured by Raman spectroscopy. Infrared spectroscopy studies have shown that this parameter is indicative of the ratio of mature (pyridinoline) to immature (dehydrodihydroxylysinonorleucine, or deH-DHLNL) collagen crosslinks (34). Therefore, the GC-induced increase in the 1660 cm−1/1690 cm−1 intensity ratio suggests that collagen maturity may be increased by GC treatment. Given that newly formed “young bone” contains less mature pyridinoline crosslinks, a GC-induced increase in collagen maturity is considered to reflect strong inhibition of new bone formation (35). This change in the collagen matrix may be associated with alterations in the biomechanical properties of bone that were observed in this study, consistent with reports that collagen crosslinking is an important determinant of bone strength, especially in postyield mechanical properties (36–39).
One limitation in this study is that the TNF-Tg animals lost body mass over time as compared to the WT mice. In addition, all animals lost body mass with GC treatment (data not shown). While normalization of structural properties by body mass could theoretically correct for size differences, it may not be appropriate in this study, as GC treatment and inflammatory arthritis might also effect changes in physical activity and food intake due to pain, joint dysfunction, and systemic inflammation. These are difficult to quantify and account for. Therefore, we believe that the effects of GC treatment and joint arthritis on bone biomechanics encompass the changes in body mass, physical activity, food intake, and other symptoms associated with bone metabolism.
As for the biologic mechanisms responsible for the increased fracture risk due to erosive arthritis and GC treatment, most of our findings are consistent with current concepts of the pathophysiology of GC-induced and RA-induced osteoporosis (4, 10, 11, 40). Both GC therapy and RA induce a slower bone turnover, with a disproportionate increase in bone resorption over a reduction in bone formation. This phenomenon can be explained by the direct negative effects of GCs and proinflammatory cytokines on osteoblasts and osteoblastic precursor cells (4–6, 8, 41). Our data suggest that the Wnt signaling inhibitor Dkk-1 contributes to both GC- and RA-induced suppression of systemic bone formation. This idea is not novel and has been reported in recent studies (9, 42, 43). Other studies using mice show that GCs induce osteocyte apoptosis, resulting in an enlargement of their lacunar space and the generation of a surrounding sphere of hypomineralized bone (15, 44–47). In this study, however, the percentage of apoptotic osteocytes was slightly increased upon GC treatment in TNF-Tg mice, with a maximum ratio of 2%, but was unchanged by GC treatment in WT mice.
Although the increased serum TRAP-5b levels may be largely due to increased osteoclast activity in the inflamed joints of TNF-Tg mice, the increased osteoclast activity at the tibial diaphysis (Figures 3N and P) indicates that enhanced bone resorption may also occur at sites outside of the joints. This is not surprising, since there are increased circulating levels of proinflammatory cytokines such as TNFα and osteoclast precursors (48). Interestingly, we found that while increased osteoclast activity is persistent in TNF-Tg mice, GC-induced osteoclastogenesis appears to be transient. Similar observations have been reported both in clinical studies in humans and in animal studies (15, 49). Reduction of osteoprotegerin levels, which antagonizes RANKL stimulation of osteoclast differentiation, during the early phase of GC treatment may be associated with the transient increase in osteoclast activity (50).
Another limitation of this study is the small sample size examined at each time point. To assess the sample size effect on the validity of the conclusions from the multiple comparisons, the positive false discovery rate, or the expected proportion of significant parameters that are false positive, was calculated using the distribution of P values from all multiple comparisons (51). Specifically, this estimate was 0.03 (95% confidence interval [95% CI] 0.01–0.06), 0.05 (95% CI 0.02–0.09), and 0.20 (95% CI 0.09–0.52) for the genotype, treatment, and interaction effects, respectively, suggesting that the majority of the statistically significant findings described herein were not false discoveries, despite the small sample size.
In conclusion, prolonged overexpression of TNFα and GC treatment additively decrease bone strength and increase its ductility, with a concomitant deterioration in bone microstructure and biochemical composition. Changes in tissue mineralization, microstructure, and organic matrix crosslinking may be associated with the high incidence of insufficiency fractures at various sites in patients with RA receiving GC therapy. These observations have important implications for the development of therapeutic strategies for RA.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Awad had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Takahata, Maher, Xing, Schwarz, Awad.
Acquisition of data. Takahata, Maher, Juneja, Inzana, Berger.
Analysis and interpretation of data. Takahata, Maher, Juneja, Inzana, Xing, Schwarz, Berger, Awad.
We would like to thank Ryan Tierney and the histology core for their excellent technical assistance. We also thank Michael Thullen for assistance with the micro-CT. The 3647 line of human TNF-Tg mice, which develops inflammation and erosive arthritis, was originally acquired from Dr. George Kollias (Institute of Immunology, Alexander Fleming Biomedical Sciences Research Center, Vari, Greece).