Decreased Quantity and Quality of the Periarticular and Nonperiarticular Bone in Patients With Rheumatoid Arthritis: A Cross-Sectional HR-pQCT Study

Authors

  • Roland Kocijan,

    1. Department of Internal Medicine 3 and Institute of Clinical Immunology, University of Erlangen-Nuremberg, Erlangen, Germany
    2. St. Vincent Hospital–Medical Department II, The VINFORCE Study Group, Academic Teaching Hospital of Medical University of Vienna, Vienna, Austria
    Search for more papers by this author
  • Stephanie Finzel,

    1. Department of Internal Medicine 3 and Institute of Clinical Immunology, University of Erlangen-Nuremberg, Erlangen, Germany
    Search for more papers by this author
  • Matthias Englbrecht,

    1. Department of Internal Medicine 3 and Institute of Clinical Immunology, University of Erlangen-Nuremberg, Erlangen, Germany
    Search for more papers by this author
  • Klaus Engelke,

    1. Institute of Medical Physics, University of Erlangen, Erlangen, Germany
    Search for more papers by this author
  • Jürgen Rech,

    1. Department of Internal Medicine 3 and Institute of Clinical Immunology, University of Erlangen-Nuremberg, Erlangen, Germany
    Search for more papers by this author
  • Georg Schett

    Corresponding author
    1. Department of Internal Medicine 3 and Institute of Clinical Immunology, University of Erlangen-Nuremberg, Erlangen, Germany
    • Address correspondence to: Georg Schett, MD, Department of Internal Medicine 3, Rheumatology and Immunology, University of Erlangen-Nuremberg, Ulmenweg 18, Erlangen, D-91054, Germany. E-mail: georg.schett@uk-erlangen.de

    Search for more papers by this author

ABSTRACT

Rheumatoid arthritis (RA) is a highly bone destructive disease. Although it is well established that RA leads to bone loss and increased fracture risk, current knowledge on the microstructural changes of bone in RA is still limited. The purpose of this study was to assess the microstructure of periarticular and nonperiarticular bone in female and male RA patients and compare it with respective healthy controls. We performed two high-resolution peripheral quantitative computed tomography (HR-pQCT; Xtreme-CT) scans, one of the distal radius and one of the ultradistal radius in 90 patients with RA (60 females, 30 males) and 70 healthy controls (40 females, 30 males) matched for sex, age, and body mass index. Volumetric bone mineral density (vBMD), bone geometry, and bone microstructure including trabecular bone volume fraction (BV/TV), trabecular number (Tb.N), trabecular thickness (Tb.Th), cortical thickness (Ct.Th) and cortical porosity (Ct.Po) were assessed. At the distal and ultradistal radius, trabecular (p = 0.005 and p < 0.001) and cortical BMD (p < 0.001 and p < 0.001) were significantly decreased in male and female patients with RA, respectively. BV/TV was also decreased at both sites, based on lower Tb.N in female RA (p < 0.001 for both sites) and lower Tb.Th (p = 0.034 and p = 0.005) in male RA patients compared with respective healthy controls. Cortical thinning (p = 0.018 and p = 0.002) but not Ct.Po (p = 0.070 and p = 0.275) was pronounced in male and female RA patients at the distal radius. Cortical perimeter was increased in male and female RA patients at both sites. Multiple regression models showed that bone geometry (cortical perimeter) is predominantly influenced by age of the RA patient, cortical thickness by both age and disease duration, and trabecular microstructure predominantly by the disease duration. In summary, these data show profound deterioration of bone microstructure in the appendicular skeleton of RA patients at both periarticular and nonperiarticular sites. © 2014 American Society for Bone and Mineral Research.

Introduction

Rheumatoid arthritis (RA) is a chronic inflammatory disorder characterized by local and systemic bone loss caused by increased bone resorption.[1] Proinflammatory cytokines and antibodies directed against citrullinated proteins, which are found in two-thirds of RA patients, are considered as central triggers for enhanced bone resorption in RA patients by inducing receptor activator of NF-κB ligand (RANKL) expression, enhancing the number of osteoclast precursors, and stimulating osteoclast differentiation.[2-4] In addition, general risk factors, such as low vitamin D level, postmenopausal state, and immobilization further exacerbate bone disease in patients with RA.[5] Together, these factors result in rapid loss of bone mass at periarticular[6] and nonperiarticular sites including the femoral head and the spine.[7, 8] In addition, RA is considered an independent risk factor for secondary osteoporosis and osteoporosis-related vertebral and nonvertebral fractures,[9-12] which is also reflected by its inclusion into the fracture risk assessment tool (FRAX).[13]

Less is known about micro-architectural changes of cortical and trabecular bone in RA. High-resolution peripheral quantitative computed tomography (HR-pQCT) has been previously validated for the assessment of volumetric bone mineral density (vBMD), bone microstructure, and bone geometry.[14] Moreover, HR-pQCT allows the detection of inflammation-caused periarticular bone changes such as bone erosion and bony proliferation at different skeletal sites.[15, 16] Differences in vBMD and microstructure between Chinese women with RA and healthy control subjects were recently introduced.[17] However, bone density, geometry, trabecular and cortical microstructure, as well as fracture incidence in the Asian population is not comparable to the white population.[18, 19]

Not only inflammation but also treatment modalities influence bone quality in RA. On one hand, glucocorticoids are well known to have adverse effects on the bone,[20] and effect of glucocorticoid treatment on bone fragility has been recently demonstrated in experimental arthritis models.[21] On the other hand, their potent anti-inflammatory potential partly counteracts the negative effects on bone metabolism.[7, 22] Furthermore, disease-modifying antirheumatic drugs (DMARDs) and biologic agents are established therapies for reducing RA joint damage and are widely used in daily clinical practice. Some of these were reported to arrest bone loss[6, 23] or even to increase bone mineral density.[24] However, their influence on bone microstructure and geometry in humans remains unclear.

Therefore, the aim of the present study was to investigate the impact of chronic inflammation and antirheumatic drugs including glucocorticoids, DMARDs, and biologics on bone structure, density, and geometry—three main components of bone strength—in male and female patients with RA.

Materials and Methods

Patients and study design

A total of 90 patients with RA (60 females, 30 males) were recruited at the Department of Internal Medicine 3 of the University of Erlangen-Nuremberg. All RA patients fulfilled the 2010 American College of Rheumatology/European League Against Rheumatism classification criteria for rheumatoid arthritis.[25] Data were compared with healthy, age-matched controls (CTRL, n = 70, 40 females, 30 males). All RA patients and healthy controls were white. Healthy controls balanced for age and sex were recruited from the resident population. The study was approved by the local ethic committee and the National Radiation Safety Agency (Bundesamt fuer Strahlenschutz). Subjects were enrolled in the study after agreeing to participate and signing an informed consent form. The study was performed in accordance with the Declaration of Helsinki.

Demographic (age, sex, weight, height, body mass index) and disease-specific variables (disease duration, disease activity score 28 [DAS-28], C-reactive protein level, rheumatoid factor [RF] positivity, and anticyclic citrullinated protein antibody [ACPA] positivity) were determined in all RA patients. DAS-28 was calculated using C-reactive protein (mg/L), tender joint count (0–28), swollen joint count (0–28), and visual analog scale (VAS, mm; formular: 0.56*sqrt[TJC28] + 0.28*sqrt[SJC28] + 0.36*ln[CRP + 1] + 0.014*VAS + 0.96). The functional status of all RA patients was documented by the Health Assessment Questionnaire (HAQ). Current and previous treatment with high-dose oral glucocorticoids (≥7.5 mg equivalent to prednisolone daily for ≥3 months), low-dose oral glucocorticoids (≤5 mg equivalent to prednisolone daily for ≥3 months), conventional DMARDs (methotrexate, leflunomide, hydroxychloroquine, sulfasalazine), and biologic DMARDs (TNF-alpha inhibitors, abatacept, rituximab, and tocilizumab) as well as antiresorptive drugs (oral and intravenous bisphosphonates and denosumab) was recorded. In addition, fracture history was recorded, wherein nontraumatic fracture was defined as a fracture acquired without any identifiable trauma or after a minor injury that would not typically lead to a bone fracture in an individual.[26]

HR-pQCT assessment

To assess bone geometry, volumetric bone mineral density (vBMD), and bone microstructure, HR-pQCT measurements were performed by an Xtreme-CT scanner (Scanco, Bruettisellen, Switzerland) using the manufacturer's standard in vivo protocol. Measurements were performed with the latest software (version 6.0) by two well-trained physicians (RK and SF). Measurements at the distal radius were carried out 9.5 mm proximal to the reference line. The reference line was set manually in the mid-joint. For scanning, the hand was immobilized in a carbon fiber cast. An anteroposterior scout view was used to determine the volume of interest. One hundred ten slices (82 µm nominal isotropic resolution, 60 kVp effective energy, 900 µA) were carried out. Daily cross-calibrations with standardized control phantoms (Moehrendorf, Germany) were conducted to standardize measurements. Reproducibility of measurements was not systematically assessed in this study. However, multicenter precision errors for the standard in vivo protocol of the Xtreme-CT scanner is less than 5%, according to recent studies.[27]

Because the volume of interest at the distal radius is not covered by synovia, a second measurement at the periarticular bone at the ultradistal radius of the same arm was performed. The reference line was set through the edge of the medial part of the radio-carpal joint. One hundred ten slices were scanned proximal to this reference line. Fig. 1 shows the two volumes of interest, distal radius and ultradistal radius. Longitudinal stability of the Xtreme-CT measurements was ascertained using the standard phantoms delivered with the scanner, which is in accordance with the procedures described in the reference manual (Xtreme-CT, User's Guide, Revision 5.05). We used the standard analysis software (version 6.0) implemented on the Xtreme-CT scanner to determine volumetric bone mineral density, microstructure, and geometric parameters, as previously described.[28]

Figure 1.

Anatomical regions for assessing periarticular and nonperiarticular bone in rheumatoid arthritis. High-resolution peripheral quantitative computed tomography (HR-pQCT) scan of the radius, transversal view. Volumes of interest of the distal radius (DR, green) and the ultradistal radius (UDR, blue) are shown. The green dotted line is the reference line for the distal radius, whereas the blue dotted line is the reference line for the ultradistal radius.

Bone geometric parameters including cortical perimeter (Ct.Pm, mm), cortical area (Ct.Area, mm2), and trabecular area (Trab.Area, mm2) were evaluated using the standard analysis protocol. Volumetric bone mineral densities (mgHA/cm3) including average BMD (D100, mgHA/cm3), trabecular BMD (Dtrab, mgHA/cm3), and cortical BMD (Dcomp, mgHA/cm3) were determined. Furthermore, bone microstructure including trabecular bone volume fraction (BV/TV), trabecular number (Tb.N, 1/mm), inhomogeneity of the network (Tb.1/N.SD, mm), cortical thickness (Ct.Th, mm), and cortical porosity (Ct.Po, %), using the advanced cortical analysis protocol, were analyzed. Because cortical bone width at the ultradistal radius is too small, only geometric parameters, trabecular BMD, and trabecular microstructure were analyzed at the ultradistal site. Moreover, bone erosions were documented in the wrist, carpal, metacarpal, and proximal interphalangeal joints in all patients with RA. All measurements in RA patients and controls were taken from the dominant arm, except in case of previous arm fracture or surgery at the dominant side, when the contralateral arm was used to avoid artefacts.

Statistical analysis

Statistical analyses were performed using SPSS software, version 21.0 (SPSS Inc., Chicago, IL, USA). Differences in bone geometry, bone mineral density, and microstructure between RA and CTRL were determined by Student's t test. The variance analysis presented in this article comprises a number of multivariate, linear regression models in order to predict bone geometry (Ct.Pm), cortical bone (Ct.Th, at distal region only), and trabecular bone (BV/TV and Tb.N) as dependent variables at the distal and ultradistal regions. In RA patients, we investigated the relations of independent to the aforementioned dependent variables within a two-step procedure in each regression model: the first step comprised a dummy coding of the multicategorical variable for steroid treatment comparing the effect of high- and low-dose glucocorticoid therapy compared with no glucocorticoid therapy as a reference. We favored this intermediate step to get a first idea of the contribution of this dummy coded variable. In the second step, the remaining independent variables (ie, age, disease duration, height, weight, sex, and the prescription of DMARD or biologic therapy) were included into regression analysis. To meet the requirement of Gaussian distributed residuals, Ct.Pm as a dependent variable in the RA regression models had to be logarithmized. To account for the inflammatory impact of RA, another set of linear regression models investigated the relation of Ct.Pm, Ct.Th, BV/TV, and Tb.N at the distal region to age, height, weight, sex, and the presence of RA diagnosis. These regression analyses merely comprised a single step, including all variables at once. All regression models used an enter method (returning results for all significant and nonsignificant independent variables) and included the individual case only if information and all regression variables were completed. Additionally, a binary logistic regression for the occurrence of fractures was calculated. For this model, we included Ct.Pm, BV/TV, Tb.N, and Ct.Th from the distal radius as well as the bone structure-related demographic variables age and sex. All independent variables were included in one step (enter method) with an additional intercept term. The maximum number of model iterations was set to 20.

Results

Demographic characteristics

Patients in the RA group were comparable to healthy controls regarding age (53.6 ± 12.8 versus 52.3 ± 12.6 years, p = 0.529), height (169.5 ± 10.2 versus 169.7 ± 10.0, p = 0.900), weight (75.9 ± 19.6 versus 75.8 ± 15.5 kg, p = 0.971) and BMI (26.2 ± 5.2 versus 26.3 ± 5.0, p = 0.856; Table 1). Mean duration of disease was 9.5 ± 8.0 years for RA. Positivity for ACPA and rheumatoid factor was found in 68% and 66% of RA patients, respectively, which is standard in an RA population. In RA, disease activity according to the DAS28 score was 3.3 ± 1.5 resembling low to moderately active disease, and mean C-reactive protein value was 5.8 ± 7.5 mg/L (normal range less than 5 mg/L). Bone erosions were detected in 72% of patients by HR-pQCT. In RA, 12% of patients had a history of nontraumatic fracture, whereas no nontraumatic fractures were recorded in the controls. The majority of RA patients received conventional (68%) and biological (69%) DMARD treatment, respectively. The total number of patients receiving glucocorticoids (≥7.5 mg equivalent to prednisolone daily for ≥3 months) was small (10%), whereas 34% of patients received low-dose oral glucocorticoids (≤5 mg equivalent to prednisolone daily for ≥3 months). None of the controls received glucocorticoids. Ten patients (11%) in the RA group and none in the control group received antiresorptive drugs. All demographic and disease-specific characteristics are shown in Table 1.

Table 1. Descriptive Characteristics of the Study Population
 RACTRL
  1. RA = rheumatoid arthritis; CTRL = healthy controls; BMI = body mass index; RF = rheumatoid factor; ACPA = anticitrullinated protein antibodies; DMARDs = disease-modifying antirheumatic drugs.
  2. Results are mean ± SD or absolute values and percentage.
  3. aDefined as presence of at least one erosion at the metacarpal or carpal joints.
  4. b≥7.5 mg prednisolone for at least 3 months.
  5. cLow-dose (LD) glucocorticoids: ≤5 mg prednisolone for at least 3 months.
Demographic characteristics
No. of patients9070
Sex (male/female)60/3040/30
Age (years)53.6 ± 12.852.3 ± 12.6
Height (cm)169.5 ± 10.2169.7 ± 10.0
Weight (kg)75.9 ± 19.675.8 ± 15.5
BMI26.2 ± 5.226.3 ± 5.0
Disease-specific characteristics
Duration of disease (years)9.5 ± 8.0
ACPA positive, n (%)61 (68)
RF positive, n (%)59 (66)
C-reactive protein (mg/L)5.8 ± 7.5
Disease Activity Score in 28 joints (units)3.3 ± 1.5
Health assessment questionnaire (units)1.02 ± 0.77
Erosive disease, n (%)a65 (72)
Nontraumatic fractures, n (%)11 (12)0
Treatment modalities
Current DMARDs, n (%)61 (68)
No. of current and previous DMARDs2.4 ± 1.5
Current biologics, n (%)62 (69)
No. of current and previous biologics1.8 ± 1.7
Current glucocorticoids, n (%)b9 (10)0
Current LD glucocorticoids, n (%)c31 (34)0
Current antiresorptive drugs, n (%)10 (11)0

Comparison of bone geometry in RA patients and controls

At the distal radius, cortical perimeter was significantly higher in male RA patients (Δ8.4%, p = 0.013) and by trend in female RA patients (Δ3.9%, p = 0.054) compared with the respective controls. Similar results were found at the ultradistal radius, where cortical perimeter was significantly higher in male (Δ7.9%, p = 0.017) and female RA patients (Δ8.5%, p = 0.001) than in the respective controls (Table 2, Fig. 2).

Table 2. Bone Geometry, Volumetric BMD, and Microstructure by HR-pQCT in RA and Controls
 RA maleCTRL malep ValueRA femaleCTRL femalep Value
  1. Results are mean ± SD.
  2. Bold indicates statistically significant findings.
  3. RA = rheumatoid arthritis; CTRL = healthy controls Ct. = cortical; Tb. = trabecular; vBMD = volumetric bone mineral density; BV/TV = bone volume fraction; Tb.N = trabecular number; Ct.Th = cortical thickness.
Distal radius
Bone geometry
Ct. perimeter (mm)91.6 ± 12.384.5 ± 8.80.01371.9 ± 6.769.2 ± 6.80.054
Volumetric bone mineral density
Average BMD (HA/cm3)294.6 ± 78.9345.8 ± 56.10.006280.5 ± 68.9336.2 ± 47.1<0.001
Tb. BMD (HA/cm3)166.3 ± 48.0200.2 ± 41.80.005128.4 ± 43.5168.8 ± 32.3<0.001
Ct. BMD (HA/cm3)754.2 ± 123.8854.2 ± 46.6<0.001789.5 ± 96.6860.0 ± 59.0<0.001
Bone microstructure
BV/TV0.139 ± 0.040.167 ± 0.030.0050.106 ± 0.040.141 ± 0.03<0.001
Tb. number (mm−1)2.01 ± 0.512.19 ± 0.270.1071.74 ± 0.502.07 ± 0.25<0.001
Tb. thickness (mm)0.070 ± 0.010.076 ± 0.010.0340.062 ± 0.010.068 ± 0.010.010
Inhomogeneity (mm)0.22 ± 0.150.17 ± 0.040.0700.33 ± 0.350.18 ± 0.070.002
Ct. thickness (mm)0.74 ± 0.300.89 ± 0.180.0180.70 ± 0.210.81 ± 0.140.002
Ct. porosity (%)4.4 ± 2.52.9 ± 1.30.0362.44 ± 1.702.04 ± 1.000.275
Ultradistal radius
Bone geometry
Ct. perimeter (mm)111.1 ± 14.4103.0 ± 10.20.01793.1 ± 11.485.8 ± 7.30.001
Volumetric bone mineral density
Tb. vBMD (HA/cm3)175.3 ± 45.4205.6 ± 33.80.006139.4 ± 37.9169.0 ± 27.1<0.001
Bone microstructure
BV/TV0.146 ± 0.040.171 ± 0.030.0060.116 ± 0.030.141 ± 0.02<0.001
Tb. number (mm−1)2.16 ± 0.422.30 ± 0.240.1251.88 ± 0.422.17 ± 0.23<0.001
Tb. thickness (mm)0.067 ± 0.010.074 ± 0.010.0050.062 ± 0.010.065 ± 0.010.105
Inhomogeneity (mm)0.23 ± 0.200.15 ± 0.030.0580.28 ± 0.160.17 ± 0.05<0.001
Figure 2.

Comparative analysis of bone microstructure in rheumatoid arthritis and controls. Sex- and disease-specific differences of bone microstructure and geometry at the distal radius[34] and the ultradistal radius (UDR) in male and female patients with rheumatoid arthritis (RA) and respective healthy controls (CTRL) are shown. BV/TV = bone volume fraction; Tb.N = trabecular number; Tb.Th = trabecular thickness; Ct.Pm = cortical perimeter; Ct.Th = cortical thickness. Asterisks indicate significant changes between RA and controls (p < 0.05).

Comparison of vBMD in RA patients and controls

In male RA, average BMD (Δ14.8%, p = 0.006), trabecular BMD (Δ16.9%, p = 0.005), and cortical BMD (Δ11.7%, p < 0.001) were significantly lower than in healthy male controls at the distal radius. Moreover, trabecular BMD was significantly lower at the ultradistal radius in male RA patients than in controls (Δ14.7%, p = 0.006) (Table 2). Similar differences were found for female RA compared with female controls: At the distal radius, average BMD (Δ16.6%, p < 0.001), trabecular BMD (Δ24.0%, p < 0.001), and cortical BMD (Δ8.2%, p < 0.001) were significantly lower in female RA patients than in controls (Table 2). Additionally, lower trabecular BMD (Δ17.5%, p < 0.001) was found at the ultradistal radius in female RA compared with respective controls.

Comparison of bone microstructure in RA patients and controls at the distal radius

At the distal radius, bone volume fraction (Δ16.8%, p = 0.005 and Δ24.8%, p < 0.001), trabecular number (Δ9.0%, p = 0.107 and Δ15.9%, p < 0.001), trabecular thickness (Δ8.4%, p = 0.034 and Δ9.4%, p = 0.010), and cortical thickness (Δ17.3%, p = 0.018 and Δ14.0%, p = 0.002) were decreased in male and female RA patients, when compared with healthy male and female controls, respectively (Table 2, Fig. 2). Moreover, inhomogeneity of the network was higher in RA patients than in controls (Δ31.8%, p = 0.070 and Δ82.4%, p = 0.002). Cortical porosity was significantly higher in male RA patients compared with male controls (Δ51.7%, p = 0.036). However, the high mean value was pushed by two outliers (9.6% and 10.3%, respectively). When these were excluded from analysis, the statistical significance disappeared, although we still found a trend to higher cortical porosity in RA patients compared with controls (3.7 ± 1.5 versus 2.9 ± 1.3%, Δ27.6%, p = 0.070). No significant differences were found between female RA patients and female controls regarding cortical porosity (Δ19.6%, p = 0.275). Fig. 3 shows representative HR-pQCT scans of RA patients and healthy controls.

Figure 3.

Cortical bone changes in patients with rheumatoid arthritis. Three-dimensional reconstruction of high-resolution peripheral quantitative computed tomography scans show the cortex of the distal radius of patients with rheumatoid arthritis (RA) and healthy, age-related healthy controls (CTRL). (A) Cortical thinning in female and male RA compared with controls. (B) Comparable cortical porosity between RA and controls.

With regard to the ultradistal radius, trabecular thickness (Δ9.5%, p = 0.005) but not trabecular number was significantly decreased in male RA patients compared with respective controls (Table 2). In contrast, female RA patients showed significantly lower trabecular number (Δ12.9%, p < 0.001) but not trabecular thickness than controls. Moreover, BV/TV (Δ14.6%, p = 0.006 and Δ17.7%, p < 0.001) and the inhomogeneity of the network (Δ53.3%, p = 0.058 and Δ64.7%, p < 0.001) at the ultradistal radius was significantly different among RA patients and controls in both sexes.

Regression analysis

The first regression model, including all subjects (RA patients and healthy controls) showed that “the diagnosis of rheumatoid arthritis” is independently and significantly related to increased cortical perimeter (β = −0.13, t = −2.80, p = 0.006) as well as decreased BV/TV (β = 0.40, t = 5.89, p < 0.001), Tb.N (β = 0.32, t = 4.41, p < 0.001), and Ct.Th (β = 0.27, t = 3.62, p < 0.001) (Table 3, Model 1). Similar results were found for vBMD (data not shown).

Table 3. Regression Models of Selected HR-pQCT Parameters
 Ct.PmBV/TVTb.NCT.Th
BetaTSig.BetaTSig.BetaTSig.BetaTSig.
Model 1, RA and healthy CTRL, distal radius
(Intercept) −1.031.304 −.313.755 .067.947 3.651.000
Age.1964.000.000−.002−.024.981−.023−.316.753−.2012.623.010
Height.4726.284.000.1801.691.093.1491.306.194−.2552.169.032
Weight.1081.837.068.055.663.508.2162.424.017.1151.257.211
Sex−.285−3.999.000−.188−1.861.065−.007−.062.951−.2902.598.010
Diagnosis RA−.134−2.803.006.3985.893.000.3194.408.000.2703.621.000
r2 adjusted .647  .293  .188  .138 
Model 2, RA, distal radius
(Intercept) 10.815.000 .700.486 .556.580 1.462.148
Age.2042.911.005.−033−.298.767−.011−.100.921−.334−3.002.004
Duration of disease−.037−.499.619−.135−1.149.254−.240−2.039.045.2211.874.065
Sex−.382−3.981.000−.184−1.218.227−.015−.102.919−.200−1.316.192
Height.4023.700.000.072.422.674.111.647.520−.102−.594.554
Weight.075.838.405.2001.423.159.2281.617.110.053.373.710
DMARD (y/n)−.153−2.417.018−.109−1.094.277.−148−1.476.144.2282.268.026
Biologics (y/n).1261.765.082−.008−.073.942.1321.162.249.037.329.743
No GC versus GC current GC−.003−.037.971−.118−1.092.278−.104−.966.337−.192−1.777.079
No GC versus LD−.009−.135.893.082.753.454.105.962.339.083.760.449
Erosive disease (y/n).0781.185.240.1121.080.283.057.544.588−.013−.124.902
r2 adjusted .662  .158  .153  .152 
  1. Model 1 was performed for all patients with rheumatoid arthritis and healthy controls at the distal radius. Model 2 was performed for RA at the distal radius and model 3 for RA at the ultradistal radius.
  2. Bold indicates statistically significant findings.
  3. Ct.Pm = cortical perimeter; BV/TV = bone volume fraction; Tb.N = trabecular number; Ct.Th = cortical thickness; current GC = glucocorticoids (≥7.5 mg prednisolone for at least 3 months); current LD GC = low-dose glucocorticoids (≤5 mg prednisolone for at least 3 months).
Model 3, RA, ultradistal radius
(Intercept) 11.403.000 1.146.256 1.136.260
Age−.004−044.965−.212−1.875.078−.128−1.100.275
Duration of disease.2562.608.011.1411.199.691−.119−.981.330
Sex−.603−4.781.000−.249−1.642.105−.131−.840.404
Height.118.797.428.102.570.570.122.663.509
Weight−.126−1.058.294.142.987.327.139.947.347
DMARDs (y/n)−.016−.193.847−.150−1.492.140−.149−1.440.154
Biologics (y/n).043.459.647−.050−.444.658.105.905.368
No GC versus GC.044.480.633−.195−1.784.078−.198−1.759.083
No GC versus LD GC.1231.318.191.045.399.691.043.377.707
Erosive disease (y/n).2332.666.009.072.686.495.037.347.730
r2 adjusted .443  .193  .149 

We also performed a second regression model addressing the factors determining bone architecture in the distal radius in patients with RA. In this model, cortical perimeter was independently influenced by age (β = 0.20, t = 2.91, p = 0.005), sex (β =−0.38, t = −3.98, p < 0.000), height (β = 0.40, t = 3.70, p = 0.000) as well as DMARD treatment (β = −0.15, t = −2.42, p = 0.018; all independent variables). Also, cortical thickness significantly related to age (β = −0.33, t = −3.00, p = 0.004) and DMARDs intake (β = 0.23, t = 2.27, p = 0.026) as well as disease duration (β = 0.22, t = 1.87, p = 0.065) and glucocorticoid treatment (β = −0.19, t = −1.78, p = 0.79) in RA (Table 3, Model 2). Trabecular number was inversely related to disease duration (β = −0.24, t = −2.04, p = 0.045) but not age, indicating a loss of trabecular bone during disease.

The respective regression model at the ultradistal radius showed significant associations between cortical perimeter and disease duration (β = 0.26, t = 2.60, p = 0.011), sex (β = −0.60, t = −4.78, p < 0.001), and erosive disease (β = 0.23, t = 2.67, p = 0.009), indicating that disease duration becomes a more dominant factor for determining bone geometry at more periarticular localized skeletal sites. BV/TV and Tb.N were inversely related to high-dose glucocorticoid intake by trend (β= −0.20, t = 1.78, p = 0.078 and β= −0.20, t = 1.76, p = 0.083, respectively, Table 3, Model 3).

Furthermore, we set up a binary logistic regression model for assessing whether specific parameters of bone architecture are different between RA patients with and without nontraumatic fractures. Although this model showed a high predictive value (88.9%), no specific parameters of bone architecture were identified to discriminate between RA patients with and without fracture (Supplemental Table S1).

Discussion

In the present study, we demonstrate differences in volumetric bone mineral density, microstructure, and geometry at the periarticular and nonperiarticular radius in RA patients when compared with healthy, age- and sex-related controls. Volumetric BMD was decreased at both trabecular and cortical sites in patients with RA. Low trabecular bone volume was mainly caused by decreased number of trabeculae in female and by trabecular thinning in male RA patients. These changes were reported to be typical sex- and age-related features of trabecular bone loss in males and females, respectively,[29] indicating an accelerated bone aging in patients with RA. The reduction in trabecular number, in particular, has a higher impact on bone strength compared with trabecular thickness.[30] Association between low trabecular number at the distal radius, assessed by HR-pQCT, and prevalent fractures were reported in both men and women.[31, 32] Moreover, in the study of Bréban and colleagues, trabecular bone score (TBS), measured at the lumbar spine, was significantly decreased in RA patients with vertebral fractures compared with those without fractures.[33] Nonetheless, the relation of microstructural changes to increased fracture risk remains to be determined in RA patients. In our study, individual micro-architectural parameters were unable to discriminate between RA patients with and without fractures, which may suggest that more complex deteriorations of bone microstructure, bone mineral density, and bone geometry are responsible for determining fracture risk.

Not only trabecular but also geometric and microstructural properties of cortical bone contribute to bone strength. Thus, cortical bone is a major component to resist axial load.[34] The association between cortical thickness at the radius and prevalent fractures was published previously.[31, 32] In the present study, cortical thickness and cortical volumetric BMD were significantly decreased in male and female RA patients when compared with healthy controls. Aside from age, height, and sex, DMARD intake was also associated with changes in cortical thickness in patients with RA. In addition, cortical porosity was reported to influence bone strength.[35] Indeed, cortical porosity was significantly increased in male RA patients compared with controls in this study. However, after excluding two outliers, the significance disappeared. In addition, cortical porosity was comparable in female RA patients and female controls. In contrast, Zhu and colleagues reported that cortical porosity is the most dramatic deterioration in microstructure measured by HR-pQCT in Chinese women with RA.[17] However, they also concede that the higher cortical porosity in RA was linked to a few patients with exaggerated periosteal bone apposition, which could result from secondary osteoarthritis in these patients. When those patients were excluded, values were comparable between RA and controls. It has to be considered that interpretation of cortical porosity in case of exaggerated periosteal bone apposition is challenging and prone to bias because bone quality of hypertrophic periosteal bone is still unknown. Moreover, using a different analysis algorithm average porosity in the cortex as measured by HR-pQCT was recently reported to be much higher[36] than in studies such as ours using the manufacturer-provided analysis algorithm. Additionally, the resolution of HR-pQCT is insufficient to detect smaller pores in cortex, and precision errors resulting from motional artefacts, especially at the radius, are higher for cortical porosity than for BMD.[37] Therefore, the overall porosity may be at least partly reflected by decreased cortical volumetric BMD, and data on cortical porosity obtained with HR-pQCT should be interpreted with caution.

We confirm the findings of Zhu and colleagues regarding the changes of the volumetric BMD, bone volume, and inhomogeneity of the trabecular network in RA. Conversely, cortical thickness was strongly reduced in our white RA patients but apparently similar in Chinese women with RA compared with respective healthy controls.[17] Marked ethnic differences in trabecular and especially cortical bone between Asian and white people were reported. Chinese-American women have thicker, denser cortices but reduced total cross-sectional area compared with white women, suggesting a compensatory mechanism to counteract fewer trabeculae and smaller bone diameters in the Asian population.[38] Also, thicker trabeculae have been reported for young Chinese and Chinese-American women compared with white women.[18, 19] Moreover, less porosity has been detected in Asian compared with white bone.[39] Chevalley and colleagues recently reported an association between reduction in cortical thickness and cortical vBMD at the distal radius with fractures in young healthy women.[40] In their study, cortical porosity was low and did not differ between women with fracture and without fracture,[40] indicating a more important role of cortical density and thickness than cortical porosity concerning bone strength.

It is known from investigations in juvenile arthritis patients that inflammation can change bone geometry.[41] In the present study, cortical perimeter was enlarged at the distal and ultradistal radius in males and females with RA compared with respective controls. Cortical thinning with periosteal apposition reflected by increased cortical perimeter, as found in our study, is a physiological process to restore bone strength.[42] This mechanism seems to be accelerated by inflammation-related endosteal bone resorption in RA.[42] Thus, an increase in outer circumference by periosteal bone formation could be explained as a compensatory mechanism to counteract cortical thinning and to improve mechanical properties.[41, 43] This hypothesis is also supported by the findings of Aberli and colleagues.[43] In their pQCT study in female RA patients, comparable results to ours regarding bone geometry, including decreased cortical thickness with compensatory increase of the outer bone diameter and cross-sectional area at the radius, were found when compared with healthy controls.[43] Apart from cortical thinning and the increase in cortical perimeter, deformities of bone shape were also observed in the RA patients in our study. However, these deformities do not occur in all patients with RA, and currently there is no validated technique available for quantifying these deformities.

Biological agents such as TNF-alpha-inhibitors were reported to arrest bone loss in RA.[6, 23] However, a specific benefit on prevention of osteoporosis or fractures has not yet been shown.[44, 45] Furthermore, little is known about the influence of conventional DMARDs. Data on MTX are conflicting, and the influence of leflunomide on bone remains unclear.[46] In the present study, current DMARDs intake was associated with an increased cortical perimeter and smaller cortical thickness at the distal radius. In contrast, biological agents were associated with decreased cortical perimeter. Neither conventional DMARDs nor biological agents influenced bone structure and geometry at the periarticular bone. It has to be considered that modification of treatment modalities in RA is common. However, no association was found between number of conventional DMARDs and biological agents ever used and microstructure parameters (data not shown). This suggests an even more important role of inflammation, age, and sex than conventional and biological DMARDs on bone microstructure. However, no negative effects were found for DMARDs on bone. A valuable explanation for the lack of association between DMARDs and microstructure could also be the limited resolution of HR-pQCT.

Glucocorticoids are well known to have adverse effects on bone.[20] An association between glucocorticoids and nonvertebral fracture risk in patients with RA was reported recently.[45] In our multiregression model, high-dose glucocorticoids were associated with cortical thinning, which was in accordance with previous findings, suggesting an increased endosteal bone resorption.[42] Moreover, an association between glucocorticoids and low bone volume fraction and trabecular number was found in the ultradistal radius, although these differences did not reach statistical significance. No effects were found for low-dose glucocorticoids on bone, suggesting that the anti-inflammatory effect of low-dose glucocorticoids counteracts its negative effects on skeleton.[8]

One of the strengths of this study is the assessment of bone composition at two different anatomical sites in both healthy and diseased individuals. Stunningly, the differences between the measurements obtained at the distal radius and the ultradistal radius were similar in RA and in healthy controls and characterized by increased cortical thinning, decreased total bone density, and increased cortical perimeter at the more distal site. Apart from this study, dual-energy X-ray absorptiometry (DXA) and quantitative ultrasound (QUS) have been used to evaluate bone characteristics of the distal radius in patients with RA. In the study of Madsen and colleagues, significant differences in QUS measures were found between RA and healthy controls.[47] Correlations between DXA results of the spine and the hip and bone erosions as well as volumetric BMD in the HR-pQCT were also reported.[48] Interestingly, this latter study showed no association between the bone changes and the actual disease activity and functional state of the RA patients, measured by DAS28 and HAQ score, respectively, which was also confirmed by our study (data not shown). This observation could be based on the limitations of momentary compared with long-term recording of disease activity and function in these studies.

In summary, both trabecular and cortical bone are severely affected in RA (Fig. 4). We found a decreased trabecular bone volume caused by decrease in number and width of the bony trabeculae in female and male RA patients, respectively. Moreover, cortical thinning but not cortical porosity was common in RA. However, micropores in RA, not detected by HR-pQCT, could be the cause for the low cortical vBMD, as found in the present study. The increase in cortical perimeter in RA may reflect a compensatory mechanism to counteract cortical thinning and to restore bone strength. A strong influence of conventional and biological DMARDs on the bone could not be demonstrated. Thus, our data suggest that bone quantity and quality are significantly decreased in the appendicular skeleton of RA patients at both periarticular and nonperiarticular sites.

Figure 4.

Cortical and trabecular bone changes in rheumatoid arthritis. Schematic drawing of bone changes in patients with rheumatoid arthritis over time. Factors determining the key changes in bone geometry, volumetric bone mineral density, and bone microstructure are indicated.

Disclosures

All authors state that they have no conflicts of interest.

Acknowledgments

RK was supported by a postdoctoral research fellowship from the Bayerische Forschungsstiftung. This study was supported by the Deutsche Forschungsgemeinschaft (SPP1468-IMMUNOBONE), the Bundesministerium für Bildung und Forschung (BMBF; project ANCYLOSS), the Marie Curie project OSTEOIMMUNE, the TEAM and MASTERSWITCH projects of the European Union, and the IMI-funded project BTCure. The authors acknowledge Rudolf Hueber for recruiting the control collective as well as Tommy Vacca and Annemarie Kocijan for proofreading.

Authors' roles: RK and SF collected the data. RK, SF, and ME analyzed the data. RK, KE, JR, and GS designed the study. RK and GS wrote the manuscript. KE, JR, SF, and ME revised the manuscript. RK takes responsibility for the integrity of the data analysis.

Ancillary