Original Article
Identification of Rheumatoid Arthritis Patients With Vertebral
Fractures Using Bone Mineral Density and Trabecular Bone Score
Sophie Br
eban,
*
Karine Briot, Sami Kolta, Simon Paternotte, Mirieme Ghazi,
Jacques Fechtenbaum, and Christian Roux
Department of Rheumatology, Paris Descartes University, Cochin Hospital, Paris, France
Abstract
The aim of this study was to test bone mineral density (BMD), trabecular bone score (TBS), and their combination, for detection of rheumatoid arthritis (RA) patients with vertebral fractures (VFs). One hundred eighty-five women aged 56.013.5 yr, with RA since 15.59.9 yr were studied. Lumbar spine, total hip, and femoral neck BMD were assessed by dual-energy X-ray absorptiometry (DXA). TBS was calculated from anteroposterior image of lum-bar spine BMD. VFs from T4 to L4 were evaluated using Vertebral Fracture Assessment software on DXA device. The proportions of patients with VF and T-scores2.5 were only 24.2%, 21.2%, and 33.3% at lumbar spine, total hip, and femoral neck, respectively. T-scores were significantly lower in patients with VF than in patients without VF, the largest difference being observed at femoral neck (p50.0001). TBS was significantly lower in patients with VF vs without VF (p50.0001). The areas under the curves were 0.621, 0.704, 0.703, 0.719, and 0.727 for lumbar spine BMD, TBS, lumbar spine BMDþTBS, total hip BMD, and femoral neck BMD, respectively. The threshold of 1.173 for TBS had the best sensitivity (63%) and specificity (74%). TBS measured at the lumbar spine has a better discrimination value than lumbar spine BMD, and similar to femoral neck BMD, for prediction of presence of VF in patients with RA. In RA subjects with osteopenia, the proportion of patients with VF was higher in the lowest tertile of TBS when compared with the highest tertile. In this population, at low risk according to BMD, TBS could help to detect patients with VF.
Key Words:DXA; glucocorticoids; rheumatoid arthritis; trabecular bone score; vertebral fractures.
Introduction
Bone involvement is the main extra articular complication of rheumatoid arthritis (RA). Patients with RA have a greater risk of osteoporosis and fracture than the general population
(1). The most important factor involved in the pathogenesis of osteoporosis is the inflammation because of disease activ-ity through the effect of inflammatory cytokines. Other well-known risk factors are glucocorticoids (GCs) use, menopausal status, low body mass index (BMI, kg/m2), and reduced
physical activity (1). Population-based controlled studies have shown that the relative risk of having at least 1 vertebral fracture (VF) is 1.7e2.3(2e4) and up to 6.2(5) in RA pa-tients. Consequences of VFs such as chronic back pain, tho-racic kyphosis, functional impairment, and back-related disability, are added to the disability of the RA itself(1).
A low bone mineral density (BMD) is also a determinant of VF risk. Prevalence of osteoporosis in RA is 20e30% at the spine and 7e26% at the hip (6e10). However, there is a discrepancy between low BMD and fracture risk, and a num-ber of fractures are observed in patients with T-scores, which are not in the osteoporotic range. T-scores in RA patients with VFs are between1.2 and 2.7 at the spine(2,11), and be-tween1.4 and 1.7 at total hip(2,11), but the risk of VFs has been described to be higher in RA compared with patients with postmenopausal osteoporosis or controls(2,4,5,11). This
Received 05/12/11; Revised 01/17/12; Accepted 01/23/12. *Address correspondence to: Sophie Breban, PhD, Department of Rheumatology, Paris Descartes University, Cochin Hospital, 27, rue du Faubourg Saint Jacques, Paris Cedex 75014, France. E-mail:
sophie_breban@yahoo.fr DOI:10.1016/j.jocd.2012.01.007
discrepancy may be related to alterations of bone, which are not captured by BMD measurements, that is, changes in bone quality. Such decreases in bone quality (including parameters of mineralization, bone matrix, microarchitecture) have been described to be related to both inflammation (1) and long-term GCs treatment (12,13). A challenge in clinical practice is thus to have a tool able to detect patients with a risk of having fractures although their BMD is not in the osteopo-rotic range.
The trabecular bone score (TBS) is a texture parameter as-sessing the pixel gray-level variations in dual-energy X-ray absorptiometry (DXA) images. The method was initially based on micro-computed tomography (mCT, 3 dimensional [3D]) images, then adapted for 2-dimensional projections ob-tained by DXA. The method builds an experimental vario-gram from the gray-scale variations in pixels in multiple random directions, and TBS is the slope at the origin of this variogram (on a log-log representation). There is no di-rect relation of TBS with microarchitectural parameters, nor trabecular network. On trabecular bone specimens, TBS mea-sured with an experimental tool is correlated with the main 3D microarchitectural parameters, measured bymCT(14,15). The software for TBS computation can be installed on DXA machines, and TBS is automatically calculated consec-utively on BMD measurement. A low TBS value indicates few gray-level variations of large amplitude and is intuitively interpreted as a low quality of bone texture.
In postmenopausal women, TBS is lower in patients with osteoporotic fractures compared with BMD-matched women without fracture(16). In a retrospective analysis of the Man-itoba Study, TBS predicts osteoporotic fractures independent of bone density(17).
The aim of this study was to test TBS, BMD, and their combination in the detection of RA patients with VF.
Patients and Methods
Study Subjects Selection
Participants were 185 women with RA who fulfilled the American College of Rheumatology criteria(18). They were consulting in the tertiary Department of Rheumatology of Cochin Hospital, Paris, France between February 2009 and July 2010 for a BMD measurement as part of the routine pro-cedure in RA.
Clinical assessment included demographic data: age, height, weight, and BMI (kg/m2). Disease duration was defined as the time elapsed between the onset of first disease-related symptoms and enrollment. Detailed information on history of all low trauma fractures (site, date and number of fractures, parental hip fracture), use of oral GCs (daily current dose, cumulative dose of prednisone equivalent), menopause status (age, duration), smoking, and alcohol consumption were col-lected using a questionnaire filled in by the physician.
The clinical activity of RA was quantified by the Disease Activity Score (DAS) 28, and the severity by the Health Assessment Questionnaire (HAQ) score. Disease-modifying
antirheumatic drugs (DMARDs), biological agents, and antios-teoporotic treatments were collected using patients’ data report.
Vertebral Fractures Assessment
Vertebral fractures from T4 to L4 were evaluated using Vertebral Fracture Assessment software on the DXA device. They were classified with the Genant semiquantitative ap-proach (17). The severity of the fracture was quantified from grades 1e3 for a reduction in anterior, middle, and/or posterior vertebral height of 20e25%, 25e40%, and over 40%, respectively. Patients with at least 1 grade 1 fracture were considered as fractured. The diagnosis was directly as-sessed on the screen, by 1 single reader, expert in this field and blinded of the patients’ characteristics.
BMD Measurements
Bone mineral density (g/cm2) was assessed by DXA (Hologic, QDR 4500A, Bedford, MA, software version 12.6) at lumbar spine (L2eL4) and left hip (total hip, femoral neck). A single device was used for the whole study. The World Health Organization classification was used to define osteoporosis as T-score 2.5 at lumbar spine, total hip, or femoral neck. The quality control protocol for the DXA de-vice included daily scanning of a phantom.
Trabecular Bone Score
The TBS (unitless) was obtained after reanalysis of DXA lumbar spine (L2eL4) scans with the TBS iNsight software version 1.8.1.0 (Med-Imaps, Pessac, France). TBS was calcu-lated as the mean value of the individual measurements for vertebrae L2eL4.
Statistical Analysis
Differences between patient groups were analyzed by non-parametric Wilcoxon tests. Prevalence of osteoporosis, VF, DMARDs, biological agents, hormonal and osteoporosis treatments, GCs use, and calcium and vitamin D supplemen-tation was assessed with a Fisher test.
Correlations between total hip, femoral neck, lumbar spine BMDs, HAQ, and DAS28 with TBS were calculated with the Spearman’s correlation coefficient.
The discriminative value of BMD at all bone sites and TBS (L2eL4) was assessed by determining the area under the re-ceiving operator characteristic (ROC) curve. A threshold value of TBS was calculated as the point that corresponds to the best sensitivity and specificity according to the VF pre-diction by ROC curve. Specificity, sensitivity, and negative and positive predictive values were calculated in patients with a T-score 2.5 at at least 1 bone site (osteoporotic), and in patients with a T-scoreO2.5 at the 3 bone sites (non-osteoporotic). The proportion of patients with VFs was com-puted between TBS tertiles also stratified by BMD T-scores. The short-term reproducibility values of TBS and BMD were assessed on 2 repeated measurements in 60 female sub-jects, aged 62.112.3 yr with an average normal BMI
(26.05.4 kg/m2). All measurements were performed on the same day after repositioning of the patient. The short-term re-producibility was estimated by calculating the coefficient of variation (CV, %), which is based on the calculation of the average root mean square (RMS) according to the following formula(19): RMS CVð%Þ5 ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 n Xn j51 SD2 j x2 i v u u t 100
All the analysis was performed on SAS 9.1 statistical software (SAS Institute Inc., Cary, NC).
Results
Patients’ Characteristics
Between February 2009 and July 2010, 185 women with RA (mean age of 56.013.5 yr) were included in the study. Their main characteristics are summarized inTable 1. Among them, 133 (71.9%) women were rheumatoid factor positive.
The mean disease duration of RA was 15.59.9 yr. In our population, 162 women (88.1%) were treated with DMARDs, 130 (70.3%) with biological agents, and 112 (60.5%) were currently treated with GCs at a mean daily dose of 6.44.3 mg per day equivalent prednisone.
Thirty-three patients (17.8%) had at least 1 VF; 16 patients had thoracic fractures (48.5%), 11 had lumbar fractures (33.3%), and 6 at both levels (18.2%). RA patients with VFs were significantly older and had a higher HAQ than RA patients without VFs. Bone characteristics of the patients are in Table 2. RA women with VFs had more frequently a previous history of non-VF and a lower BMD at all sites. The prevalence of osteoporosis was significantly higher in RA patients with VFs than in patients without VFs (48.5% vs 27.0%). One hundred twenty-eight patients were nonosteo-porotic, that is, had T-scores higher than2.5 at all the 3 sites (lumbar spine, femoral neck, and total hip), and 16 of them had VFs: thus, 50% of the fractured population has a T-score that did not reach the osteoporotic threshold. There were no significant differences in BMDs and T-scores between nonos-teoporotic patients with VFs and nonosnonos-teoporotic patients
Table 1 RA Patients’ Characteristics RA Patients’ Characteristics RA women with VF (n533) RA women without VF (n5152) Whole RA population (n5185) p
Age (yr) (meanSD) 63.114.9 54.512.8 56.013.5 0.0004
Weight (kg) (meanSD) 61.512.7 62.213.0 62.113.0 0.88 Height (m) (meanSD) 1.570.95 1.600.66 1.600.72 0.10 BMI (kg/m2) (meanSD) 24.74.1 24.35.1 24.34.9 0.67 RA disease duration (yr)
(meanSD)
17.912.9 14.99.2 15.59.9 0.40 RF positive (n, %) 21 (63.6%) 112 (73.7%) 133 (71.9%) 0.42 DAS28 (meanSD) 4.12.6 3.61.4 3.61.6 0.55 HAQ (meanSD) 1.91.0 1.40. 8 1.50.9 0.01
Current use of DMARDs (n, %) 21 (70.0%) 141 (93.4%) 162 (88.1%) 0.001
Current use of biological agents (n, %)
19 (63.3%) 111 (73.5%) 130 (70.3%) 0.27 Current use of glucocorticoids (n, %) 17 (56.7%) 95 (63.3%) 112 (60.5%) 0.54 Mean current dose of prednisone
(mg/d) (meanSD)
6.32.3 6.54.7 6.44.3 0.41 Mean cumulative dose of
prednisone (g) (meanSD)
37.336.1 32.641.8 33.640.7 0.44 Prevalence of menopause 24 (72.7%) 102 (67.1%) 126 (68.1%) 0.39 Duration of menopause 21.410.4 11.88.9 13.59.9 0.0002
Current smoking (n, %) 7 (21.2%) 17 (11.2%) 24 (13.0%) 1.00 Excessive alcohol consumption
(O3 glasses of wine per units) (n, %)
6 (18.2%) 29 (19.1%) 35 (18.9%) 0.08
Note:Resultsstandard deviation.
Abbr:BMI, body mass index (weight/height2); RA, rheumatoid arthritis; VF, vertebral fracture; DMARD, disease-modifying antirheumatic drug; DAS, Disease Activity Score; HAQ, Health Assessment Questionnaire; RF, rheumatoid factor; SD, standard deviation.
without VFs (0.10!p!0.57). Patients with VFs reported more frequently the use of antiosteoporotic treatment (al-though only 51.5% of them received such a treatment).
Patients treated with GCs had a significantly lower total hip BMD when compared with patients not treated with GCs (p50.03), but there was no significant difference for spine and femoral neck BMDs (data not shown).
Trabecular Bone Score
The short-term reproducibility (RMS CV) calculated after repositioning in 60 patients was 1.44% and 1.18%, for TBS and lumbar spine BMD, respectively. TBS was significantly lower in RA patients with VFs compared with RA patients without VFs (1.1310.195 vs 1.2450.106, p50.0001), and in patients receiving GCs treatment (1.2110.116) com-pared with patients without a current treatment (1.2650.098) (p!0.0001).
TBS was significantly correlated with HAQ (r50.23,
p50.008) but not with DAS28, in the whole population. BMDs were not correlated to HAQ or DAS28, except for total hip BMD, which was correlated to HAQ (r50.18,
p50.02). Correlations of TBS and BMDs are reported in
Table 3. All the correlations were statistically significant but weak. The coefficient of correlation of TBS (measured at lum-bar spine) and BMD was similar for lumlum-bar spine (r50.58,
p!0.0001), femoral neck (r50.53, p!0.0001), and total hip (r50.58,p!0.0001).
Among patients with VFs (N533), 12 patients out of 33 had VFs at the lumbar spine (36.4%). There was no signifi-cant difference between mean TBS calculated with and with-out excluding fractured vertebrae, (p50.71) and between mean TBS calculated by excluding fractured vertebrae and mean TBS of the fractured vertebrae (p50.28). Similar re-sults were observed with lumbar spine BMD (0.24!p!
0.37).
VFs Discrimination Using TBS
The areas under the curves (AUCs) were 0.704, 0.621, 0.727, and 0.719 for TBS, lumbar spine BMD, femoral neck BMD, and total hip BMD, respectively (Fig. 1). The AUCs for the combination of lumbar spine BMD and TBS, femoral neck BMD and TBS, and total hip BMD and TBS were 0.703, 0.730, and 0.724, respectively.
In the whole population, the AUC for TBS tended to be higher than the AUC for lumbar spine BMD (p50.08), but there was no difference between AUC for TBS vs AUC for femoral neck BMD (p50.70). The AUC for femoral neck BMD was significantly higher when compared with the AUC for lumbar spine BMD (p50.04). In the nonosteopor-otic population, there was no statistically significant differ-ence for AUC TBS vs AUC lumbar spine, femoral neck, or total hip BMD (0.22!p!0.53). There were no significant differences in AUC values after adjustment for current GC’s treatment and for current antiosteoporotic treatment (data
Table 2
Bone Parameters’ Characteristics of RA Patients
Bone Parameters’ Characteristics
RA patients with VF (n533) RA patients without VF (n5152) RA whole population (n5185) p Lumbar spine BMD (g/cm2) (meanSD) 0.8811.162 0.9560.156 0.9430.159 0.01
Lumbar spine T-score (meanSD) 1.51.5 0.81.5 0.91.5 0.01
Total hip BMD (g/cm2) (meanSD)
0.7040.149 0.8210.150 0.8020.156 0.0001
Total hip T-score (meanSD) 1.81.1 0.91.0 1.11.1 0.0001
Femoral neck BMD (g/cm2) (meanSD)
0.5960.137 0.6950.119 0.6780.127 0.00008
Femoral neck T-score (meanSD) 2.31.1 1.41.0 1.61.1 0.0001
TBS (meanSD) 1.1310.195 1.2450.106 1.2320.112 0.0001
Prevalence of osteoporosis (n, %) 16 (48.5%) 41 (27.0%) 57 (30.8%) 0.04
Previous non-VFs (n, %) 21 (63.6%) 37 (24.3%) 58 (31.3%) !0.0001
Current use of antiosteoporotic treatments (n, %)
17 (51.5%) 47 (30.9%) 64 (34.6%) 0.01
Current use of calcium (n, %) 20 (60.6%) 70 (40.0%) 90 (48.7%) 0.11 Current use of vitamin D (n, %) 27 (81.8%) 99 (65.1%) 126 (68.1%) 0.02
Current use of hormonal replacement treatment (n, %)
7 (21.2%) 37 (24.3%) 44 (23.8%) 1.00
Note:Resultsstandard deviation.
Abbr:RA, rheumatoid arthritis; VF, vertebral fracture; BMD, bone mineral density; TBS, trabecular bone score; SD, standard deviation.
not shown). The AUCs were similar according to VF’s grades (data not shown).
The discriminative ability of TBS to reach the best sensi-tivity and specificity was determined using the ROC curve (Fig. 1). The ROC curve showed an inflexion point for a sensitivity value of 63% and a specificity of 74% for TBS51.173.
Vertebral fractures distribution among the nonosteoporotic patients (n5128), according to this threshold of TBS is shown in the Table 4: 16 had VFs and 8 of them had
a TBS value lower than 1.173. Among the 16 nonosteoporotic patients with VFs, 11 were osteopenic and 5 had a normal BMD. The mean T-scores of the 8 nonosteoporotic patients who had both VFs and a TBS below the threshold of 1.173 were 1.10.8, 2.00.4, and 1.70.5 at the lumbar spine, femoral neck, and total hip, respectively. None of the 5 patients with normal BMD had a TBS below the threshold. Results of sensibility, specificity, positive predictive value, and negative predictive value of BMD and the combination of BMD and TBS are presented inTable 5. Patients were strat-ified according to tertiles of TBS and BMD status (normal, osteopenia, and osteoporosis) (Table 6). The largest popula-tion was nonosteoporotic patients, that is, T-scoreO2.5 at all measured sites. In this population, there was a statistically significant difference in the proportions of patients with frac-tures, a lower TBS indicating a higher risk of having such fractures.
Discussion
This study is the first assessing the value of TBS in patients with RA. A high proportion of these patients have VFs al-though their bone density is above the osteoporotic threshold, and a low TBS is associated with a higher risk of having such fractures.
Our results confirm that a generalized osteoporosis is ob-served in RA: 30% of our patients with a mean age of 56 yr had a T-score below2.5, and 18% had at least 1 VF. Both hip BMD and lumbar spine TBS were correlated to HAQ, which assesses the severity of the disease. However, among patients with VFs, more than half of them had a T-score that did not reach the osteoporotic threshold at any site. RA can induce both skeletal and extra skeletal alterations. Inflamma-tion and RA treatments can affect quality parameters of bone, which are not measured by BMD but apparently by TBS. Confounding factors affect the interpretation of BMD in RA patients, as the risk of fracture is also related to the se-verity of the disease, the radiological damages, the disability, and the risk of falls, which is increased by changes in body
0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1 0 0,2 0,4 0,6 0,8 1 Specificity Sensitivity TBS (AUC=0.704) LS BMD (AUC=0.621) FN BMD (AUC=0.727) TH BMD (AUC=0.719)
Fig. 1.Receiving operator characteristic curves of lumbar spine trabecular bone score (TBS), lumbar spine (LS) bone mineral density (BMD), femoral neck (FN) BMD, and total hip (TH) BMD models, in vertebral fracture prediction, in the whole population of patients with rheumatoid arthritis (n5185).
Table 3
Correlation Analysis of TBS
RA patients’ characteristics
Spearman’s correlation coefficient (vs TBS) RA patients with VF (n533) RA patients without VF (n5152) RA whole population (n5185) HAQ 0.09 (p50.63) 0.32(p50.0001) 0.23(p50.008) DAS28 0.10 (p50.62) 0.21(p50.01) 0.15 (p50.09) Lumbar spine BMD 0.52(p50.003) 0.41(p!0.0001) 0.58(p!0.0001) Total hip BMD 0.38(p50.05) 0.45(p!0.0001) 0.58(p!0.0001) Femoral neck BMD 0.32 (p50.10) 0.42(p!0.0001) 0.53(p!0.0001)
Note:Results are expressed inrvalue and the correlation was significant whenp0.05.
Abbr:RA, rheumatoid arthritis; VF, vertebral fracture; BMD, bone mineral density; TBS, trabecular bone score, HAQ: health assessment questionnaire; DAS, disease activity score.
composition and joints destructions. In addition, it has been shown that even in normal postmenopausal women, the high-est number of fractures occurs in those with osteopenia(20). This does not preclude the value of BMD in the assessment of bone fragility, and is related to both the higher number of subjects with osteopenia when compared with those with osteoporosis, and to nonquantitative alterations of bone strength in some individuals.
Our data show that a low femoral neck BMD is a risk fac-tor for the presence of VFs and is in this matter better than lumbar spine BMD. Our patients with VF had a mean age of 63 yr old, that is, 9 yr older than patients without VF, and thus may be concerned by cortical bone loss, which is best assessed at the femoral neck. Moreover, artifactual changes in spine BMD can be related to degenerative changes at this site (21). Half of the patients with VFs did not have osteoporosis, even at the femoral neck.
The TBS assessment has been described to be well fitted to texture measurement on small and/or irregular surfaces of analysis, such as DXA regions of interest. This method has been validated in vitro in 2 sets of human bone reconstruc-tions from different anatomical sites. Using an experimental tool with high resolution, it has been described as highly sig-nificant correlation between TBS values and the standard 3D trabecular microarchitecture characteristics. In vivo, 4 studies have been conducted in postmenopausal women with or with-out VFs. TBS was able to discriminate VFs compared with all
fracture types combined, independently of lumbar spine BMD, and was able to identify patients with osteopenia and VF. In all studies, the AUC for the combination of spine BMD and TBS was significantly higher than the AUC for spine BMD alone (14,22). In a retrospective analysis of data from 29,407 women aged older than 50 yr, TBS and spine BMD predicted fractures equally well, and the combi-nation was superior to either measurement alone(17).
Our study is the first to assess the potential of TBS to dis-criminate VF in RA patients, a population in which BMD does not assess the whole fracture risk. TBS measured at the spine had a better discriminative value than spine BMD alone, confirming data obtained in the general population
(22) and suggesting that TBS result is not totally explained by BMD. This is also suggested by the results of correlations between TBS and spine BMD (r50.58). Spine BMD calcu-lation is affected by the presence of osteoarthritis-related degenerative changes at the lumbar spine(21), but it is un-known how this artifact can change the TBS results. In our patients with VFs at the lumbar spine, there was no difference in TBS values between fractured and nonfractured vertebrae, which have completely different structures; we recognize that the low sample size precludes any definitive conclusion on that point.
The discriminative value of TBS was similar to one of the femoral neck BMD in our population with RA, as assessed by AUCs. However, our aim was to assess the added value of TBS above BMD, specially in this population with a high pro-portion of patients with osteopenia or even normal BMD. In-deed, in patients with osteoporosis, that is, a T-score!2.5 at at least 1 site, there was no difference in proportion of patients with a VF according to the tertile of TBS value, sug-gesting that TBS has no added value in these patients already detected by BMD. In contrast, in patients with osteopenia the proportion was higher in lower values of TBS than in higher values of TBS, indicating that in this subpopulation TBS added to BMD could help to detect at risk patients.
This study has several limitations; it was performed in RA female patients consulting in a tertiary Department of Rheu-matology, with severe and long-lasting disease; the current conclusions may not be applicable to ambulatory patients with different procedure of the use of GCs, DMARDs, and/ or biological agents, or younger women or men with RA. Moreover, a cross-sectional study cannot conclude about cau-sality between low TBS and occurrence of fractures.
Table 5
Specificity, Sensibility, and Positive and Negative Predictive Values for the 2 Models: Model 1, Patients With T-score2.5 at All Bone Sites; Model 2, Patients With T-scoreO2.5 at All Bone Sites With a TBS Below the
Calculated Threshold Value
Model 1 Model 2 p
Sensibility 0.47 0.73 0.005
Specificity 0.73 0.61 !0.001
Positive predictive value 0.25 0.28 NA Negative predictive value 0.87 0.92 NA
Abbr:NA, nonavailable; TBS, trabecular bone score.
pexpressed the statistical difference between Model 1 and Model 2.
pwas significant when0.05.
Table 4
Vertebral Fracture Discrimination by TBS in Nonosteoporotic Women
Population of women without osteoporosis (T-scoreO2.5 at all sites) (N5128); 5 missing data in the analysis
TBSO1.173 TBS1.173 p-Value of the exact Fisher test
N (%) N (%) 0.001
Vertebral fracture5no (N5107) 91 (85.0%) 16 (14.9%) Vertebral fracture5yes (N516) 8 (50.0%) 8 (50.0%)
RA patients are at increased risk of VFs, and most of them occur in patients with bone density above the osteoporotic threshold. In patients with RA and osteopenia, TBS is an easily applied parameter to detect patients with VF.
References
1. Roux C. 2011 Osteoporosis in inflammatory joint diseases. Osteoporos Int 22:421e433.
2. Orstavik RE, Haugeberg G, Mowinckel, et al. 2004 Vertebral de-formities in rheumatoid arthritis: a comparison with population-based controls. Arch Intern Med 164:420e425.
3. de Nijs RN, Jacobs JW, Bijlsma JW, et al. 2001 Prevalence of vertebral deformities and symptomatic vertebral fractures in cor-ticosteroid treated patients with rheumatoid arthritis. Rheuma-tology (Oxford) 40:1375e1383.
4. Spector TD, Hall GM, McCloskey EV, Kanis JA. 1993 Risk of vertebral fracture in women with rheumatoid arthritis. BMJ 306:558.
5. Peel NF, Moore DJ, Barrington NA, et al. 1995 Risk of vertebral fracture and relationship to bone mineral density in steroid treated rheumatoid arthritis. Ann Rheum Dis 54:801e806. 6. Lodder MC, de Jong Z, Kostense PJ, et al. 2004 Bone mineral
density in patients with rheumatoid arthritis: relation between disease severity and low bone mineral density. Ann Rheum Dis 63:1576e1580.
7. Lodder MC, Haugeberg G, Lems WF, et al. 2003 Radiographic damage associated with low bone mineral density and vertebral deformities in rheumatoid arthritis: the Oslo-Truro-Amsterdam (OSTRA) collaborative study. Arthritis Rheum 49:209e215. 8. Sinigaglia L, Nervetti A, Mela Q, et al. 2000 A multicenter cross
sectional study on bone mineral density in rheumatoid arthritis. Italian Study Group on Bone Mass in Rheumatoid Arthritis. J Rheumatol 27:2582e2589.
9. Haugeberg G, Uhlig T, Falch JA, et al. 2000 Bone mineral den-sity and frequency of osteoporosis in female patients with rheumatoid arthritis: results from 394 patients in the Oslo County Rheumatoid Arthritis register. Arthritis Rheum 43: 522e530.
10. Gough AK, Lilley J, Eyre S, et al. 1994 Generalised bone loss in patients with early rheumatoid arthritis. Lancet 344:23e27. 11. El Maghraoui A, Rezqi A, Mounach A, et al. 2010 Prevalence
and risk factors of vertebral fractures in women with rheumatoid
arthritis using vertebral fracture assessment. Rheumatology (Oxford) 49:1303e1310.
12. Chappard D, Legrand E, Basle MF, et al. 1996 Altered trabecu-lar architecture induced by corticosteroids: a bone histomorpho-metric study. J Bone Miner Res 11:676e685.
13. Lespessailles E, Siroux V, Poupon S, et al. 2000 Long-term cor-ticosteroid therapy induces mild changes in trabecular bone texture. J Bone Miner Res 15:747e753.
14. Winzenrieth R, Dufour R, Pothuaud L, Hans D. 2010 A retro-spective case-control study assessing the role of trabecular bone score in postmenopausal Caucasian women with osteope-nia: analyzing the odds of vertebral fracture. Calcif Tissue Int 86:104e109.
15. Pothuaud L, Carceller P, Hans D. 2008 Correlations between grey-level variations in 2D projection images (TBS) and 3D microarchitecture: applications in the study of human trabecular bone microarchitecture. Bone 42:775e787.
16. Pothuaud L, Barthe N, Krieg MA, et al. 2009 Evaluation of the potential use of trabecular bone score to complement bone min-eral density in the diagnosis of osteoporosis: a preliminary spine BMD-matched, case-control study. J Clin Densitom 12: 170e176.
17. Hans D, Goertzen AL, Krieg MA, Leslie WD. 2011 Bone micro-architecture assessed by TBS predicts osteoporotic fractures independent of bone density: the Manitoba study. J Bone Miner Res 26:2762e2769.
18. Arnett FC, Edworthy SM, Bloch DA, et al. 1988 The American Rheumatism Association 1987 revised criteria for the classifica-tion of rheumatoid arthritis. Arthritis Rheum 31:315e324. 19. Gluer CC, Blake G, Lu Y, et al. 1995 Accurate assessment of
precision errors: how to measure the reproducibility of bone den-sitometry techniques. Osteoporos Int 5:262e270.
20. Pasco JA, Seeman E, Henry MJ, et al. 2006 The population burden of fractures originates in women with osteopenia, not osteoporosis. Osteoporos Int 17:1404e1409.
21. O’Gradaigh D, Debiram I, Love S, et al. 2003 A prospective study of discordance in diagnosis of osteoporosis using spine and proximal femur bone densitometry. Osteoporos Int 14: 13e18.
22. Rabier B, Heraud A, Grand-Lenoir C, et al. 2010 A multicentre, retrospective case-control study assessing the role of trabecular bone score (TBS) in menopausal Caucasian women with low ar-eal bone mineral density (BMDa): analysing the odds of verte-bral fracture. Bone 46:176e181.
Table 6
Number and Proportion of RA Patients With Vertebral Fractures According to TBS Tertiles and T-Scores Classification T-score Low!1.182 Middle (1.182e1.293) High1.293 p-Value T-scoreO1 at all sites N54 N511 N519 0.455
0/4 (0.0%) 1/11 (9.1%) 0/18 (0.0%)
T-scoreO2.5 at all sites N522 N537 N531 0.013 8/22 (36.4%%) 5/37 (13.5%) 2.31 (6.5%)
T-score!2.5 at at least 1 site N532 N511 N56 0.528 9/31 (29.0%) 1/11 (9.1%) 1.6 (16.7%)
