Background
Spondyloarthritis (SpA) comprises a group of diseases with shared genetic and pathophysiologic backgrounds affecting the axial and peripheral skeleton. Axial disease comprises nonradiographic axial SpA (nr-axSpA) and ankylosing spondylitis (AS). Axial inflammation leads to local bone formation with progressive ankylosis of sacroiliac joints as well as syndesmophyte formation. Nonetheless, systemic bone loss and increased vertebral and nonvertebral fracture risk have been described in AS [
1].
Authors of a recent review of the literature on bone mass in AS reported a prevalence of osteoporosis varying from 3% to 47% according to different measurement techniques and patient selection criteria, whereas osteopenia has been reported in up to 88% of patients [
2]. These variations may be based on the fact that bone analysis in nr-axSpA and AS is challenging, particularly in the axial skeleton, where it is confounded by local new bone formation. Especially in patients with syndesmophytes, dual-energy X-ray absorptiometry (DXA) is unreliable because it sums up new bone formation with bone loss owing to its two-dimensional nature [
3,
4]. However, bone loss has also been found at the hip of patients with AS [
5‐
8]. Furthermore, quantitative computed tomography (QCT), which separately measures trabecular and cortical volumetric bone density, has supported the occurrence of bone loss in AS, showing reduced trabecular volumetric bone mineral density (vBMD) in the lumbar spine in severe AS with syndesmophyte formation [
9].
Systemic inflammation is at least partly responsible for systemic bone loss in patients owing to proinflammatory cytokines leading to a direct activation of osteoclastogenesis [
10]. In prior studies, patients with AS showed significant bone loss as measured by DXA at the lumbar spine compared with patients with mechanical back pain [
11]. Further, in the DESIR (DEvenir des Spondyloarthrites Indifférenciées Récentes) cohort, 71.4% of patients with inflammatory back pain fulfilled Assessment of Spondyloarthritis international Society (ASAS) classification criteria for axial spondyloarthritis (axSpA), and in multiple logistic regression analyses, bone marrow edema seen on magnetic resonance imaging scans, markers of inflammation such as C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR), and male sex were associated with lower bone mineral density (BMD) at any site [
12]. Inhibition of tumor necrosis factor (TNF)-α has been reported to show a beneficial effect on BMD and bone turnover markers of patients with AS, thereby supporting the role of systemic inflammation on bone metabolism [
13‐
15]. In accordance with this, vertebral fractures in patients with AS are associated with the longer disease duration independent of age [
16].
Bone strength depends not only on BMD but also on bone geometry and microarchitectural aspects of cancellous and cortical bone [
17,
18]. High-resolution quantitative computed tomography (HR-pQCT) permits a noninvasive, three-dimensional assessment of bone geometry, volumetric BMD, and microarchitecture, which resembles a virtual bone biopsy of peripheral bone [
19,
20]. HR-pQCT data have been shown to correlate with BMD results obtained by DXA and with incident fracture risk in the radius, hip, and spine in postmenopausal women [
21,
22].
The aim of the present study was to investigate bone microstructure, geometry, and vBMD using HR-pQCT in a large cohort of patients with nr-axSpA in early stages of disease and to search for potential risk factors for deterioration of bone microstructure. In this context, it is important to mention that patients with AS with long-standing disease are characterized by reduction of total and cortical vBMD and cortical thickness, as well as increased cortical porosity [
23,
24]. Data on early disease, however, which may reflect the initial changes of bone in nr-axSpA, are missing to date.
Methods
Patients and control subjects
A total of 107 Caucasian patients with a diagnosis of nr-axSpA were recruited at the Department of Internal Medicine 3 of the University of Erlangen-Nuremberg, Germany. All patients fulfilled the ASAS classification criteria. Patients with AS were not included in this study because we were specifically seeking to study a population with early bone changes. Recruitment of healthy control subjects has been described elsewhere, and these individuals were matched by age and sex [
25]. Briefly, exclusion criteria for being a healthy control subject were (1) presence or history of chronic joint pain/swelling, (2) presence of systemic diseases, (3) documented osteopenia/osteoporosis, (4) present or past use of bisphosphonates or prednisolone, and (5) positivity for anticitrullinated protein antibodies (ACPA) or rheumatoid factor. This study was approved by the local ethics committee and the national radiation safety agency (Bundesamt fur Strahlenschutz). Informed consent was obtained from each patient, and the study was performed in accordance with the Declaration of Helsinki.
Demographic and disease-specific characteristics
Demographic characteristics, disease duration, features of SpA (psoriasis, inflammatory bowel disease [IBD], uveitis), human leukocyte antigen (HLA)-B27 status, ESR, and serum CRP levels were recorded in all patients. Current disease activity was determined by Ankylosing Spondylitis Disease Activity Score (ASDAS-CRP; inactive disease ASDAS-CRP < 1.3, 1.3–2.0 moderate disease activity, 2.1–3.5 high disease activity, > 3.5 very high disease activity). Spinal mobility was assessed by Bath Ankylosing Spondylitis Metrology Index (BASMI). For the assessment of present enthesitis, the Maastricht Ankylosing Spondylitis Enthesitis Score was used. Peripheral arthritis was assessed by 78 tender and 76 swollen joint counts. The Spondylitis Disease Activity Index (BASDAI) was assessed as a secondary disease activity measure, and the Bath Ankylosing Spondylitis Functional Index (BASFI) was used for patient-reported outcomes.
The use of conventional disease-modifying antirheumatic drugs (methotrexate, sulfasalazine, azathioprine, leflunomide, chloroquine, gold) and biologic agents (TNF inhibitors [TNFi]) was recorded. Further, the current or previous use of systemic glucocorticoid (GC) treatment exceeding continuous treatment with 5 mg of prednisolone-equivalent daily for more than 3 months was assessed. Treatment with nonsteroidal anti-inflammatory drugs was also recorded (on demand, daily). History of nontraumatic fractures, diagnosis of osteoporosis, and previous or ongoing antiresorptive treatment or supplementation of 25(OH)vitamin D3 and calcium was taken.
High-resolution peripheral quantitative computed tomography
HR-pQCT imaging was performed in all patients and 50 healthy age- and sex-matched control subjects using the XtremeCT scanner (SCANCO Medical, Brüttisellen, Switzerland). The scan region was selected according to the manufacturer’s standard in vivo protocol of the ultradistal radius of the dominant hand. For measurement, the patient’s hand was immobilized using a carbon fiber shell to reduce movement. Standardization of measurements was ensured by daily cross-calibrations with a standardized control phantom (QRM, Moehrendorf, Germany). The ROI was determined with an anteroposterior scout view and was fixed 9.5 mm proximal from the reference line. The effective dose for each scan was < 3 μSv. The reference line was set manually. The scan ROI was examined in 110 parallel slices (82-μm voxel size) with a total measurement time of 2.8 minutes. All measurements and evaluations were performed using the manufacturer’s standard software. Motion grading (1–5) of each scan was performed using SCANCO Medical standard operating procedure scale, and scans graded > 3 were excluded from analysis.
vBMD, bone microstructure, and geometry were measured with HR-pQCT. Three-dimensional vBMD of the total radius (total BMD in mg of hydroxyapatite [HA]/cm
3), the cortical shell (Ct. BMD, mg HA/cm
3), and the trabecular compartment (Tb.BMD, mg HA/cm
3) were extracted. Additional, distinctive results of trabecular BMD adjacent to bone cortex (mg HA/cm
3) and central medullary trabecular BMD (mg HA/cm
3) were expressed. Results of bone microstructure included bone volume fraction (BV/TV, %), trabecular number (mm
−1), trabecular thickness (μm), trabecular separation (Tb.Sp, μm), inhomogeneity of the trabecular network (μm), cortical thickness (Ct.Th, μm), and cortical porosity (Ct.Po, %). Furthermore, bone geometry was represented by total, cortical, and trabecular bone area (mm
2). All these parameters were calculated by using automated software. Reliability of the automated contouring method of the Xtreme CT scanner software has recently been shown [
26].
Statistical analysis
Data were collected, organized, and analyzed using IBM SPSS Statistics software (IBM, Armonk, NY, USA). If not stated otherwise, categorical variables are presented as number and percent, and continuous variables are provided as median (IQR). Inferential comparisons comprised chi-square tests for categorical variables (indicated as number and percent in the tables) to check for observed deviations from expected frequencies, as well as the Kruskal-Wallis and Mann-Whitney U tests to compare data derived from interval scales. To investigate potential relationships of total, cortical, and trabecular vBMD with disease-related or demographic parameters, multiple linear regression models were computed with an enter procedure including all predictors at a single step. The first model incorporated sex, age, BMI, and smoking status. The second model included sex, age, BMI, remission status, disease duration, treatment with TNFi, prior GC treatment, HLA-B27 status, and peripheral arthritis. A p value less than 0.05 was considered significant.
Discussion
Bone loss is a well-known phenomenon in axSpA, especially in long-standing disease. In the present study, we performed a detailed analysis of bone microstructure in a large cohort of patients with axSpA, with disease duration of less than 2 years in nearly 50% of patients. Our analysis shows that patients with nr-axSpA are characterized by a virtually exclusive cortical but not trabecular bone pathology, which is remarkable. Cortical bone changes characterized by changed geometry, BMD, and microstructure were found early in the disease course of axSpA.
To date, different structural and compartmental changes in bone have been identified using HR-pQCT in systemic inflammatory diseases. Whereas in rheumatoid arthritis a significant deterioration of cortical and trabecular bone has been described, especially in ACPA-positive patients, patients with psoriatic arthritis show predominantly changes of trabecular bone [
27,
28]. In contrast, in patients with IBD, primarily a loss of cortical bone has been found [
29]. The described changes of cortical bone in nr-axSpA reflect bone structural changes found in patients with IBD.
Our study shows that cortical bone loss as evidenced by cortical thinning happens early in axSpA and can already be found within the first 2 years of disease. In accordance with this, a previous HR-pQCT study of male patients with established AS showed a reduction of cortical vBMD and increased cortical porosity [
24]. In this study, however, patients had long-standing disease, with a median duration of symptoms longer than 20 years. Cortical thinning and low cross-sectional area in the peripheral skeleton in patients with AS were strongly associated with the presence of vertebral fractures [
24]. These findings are in accordance with the previously described association of cortical thinning and loss of cortical BMD at the radius and the tibia with the occurrence of vertebral fractures in men in the general population [
30]. The median disease duration in our study was much shorter (6.5 years) than in the aforementioned study. Nonetheless, about half of the patients with nr-axSpA did not have early disease (< 2 years) anymore and were treated with TNFi. The fact that some patients had a disease duration longer than 2 years may also explain the observation that 7% of the patients with axSPA already had a history of low traumatic fracture. Overall, however, our data show systemic deterioration of cortical bone microstructure even at an early stage of disease in patients with nr-axSpA.
Data are so far limited regarding the factors that influence bone microarchitecture in patients with nr-axSpA and patients with AS. Multiple logistic regression identified male sex to be associated with lower cortical vBMD. Interestingly, loss of cortical vBMD was independent of standard disease-related features such as age, BMI, remission status, anti-TNF treatment, HLA-B27 status, and peripheral arthritis. In contrast to cortical vBMD, trabecular vBMD was lower in women than in men. These findings are in accordance with previous population-based studies showing higher trabecular vBMD in men, whereas cortical vBMD is higher in women [
31]. Disease duration was associated positively with higher trabecular vBMD in the present study, which can be explained by longer disease duration in men than in women. These results confirm previous findings of Haroon et al., who investigated sex differences among patients with AS and showed decreased cortical vBMD in men, whereas trabecular vBMD was worse in women [
23].
Although nr-axSpA specifically affects cortical bone, an evaluation of those patients treated with GCs > 5 mg daily for > 3 months in the past showed specific loss of trabecular bone associated with reduction of trabecular vBMD, lower total bone volume, and higher trabecular separation and inhomogeneity. Sutter et al. investigated postmenopausal women treated with oral GCs for > 3 months and age-/race-matched control subjects using HR-pQCT and DXA. In their study, despite no difference in areal BMD, GC-treated women showed an impairment of cortical and trabecular vBMD and bone microarchitecture [
32]. Moreover, later stages of the disease, particularly in patients with severe AS with syndesmophyte formation, also trabecular vBMD appears to decrease, which is reflected by a study from Devogelaer and colleagues [
9].
The strength of the present study is a detailed assessment of bone macro- and microarchitecture using HR-pQCT in patients with axSpA. To date, this is the largest nr-axSpA cohort with detailed bone analysis by HR-pQCT. A further strength and novelty of this work is the assessment of patients with short disease duration. A limitation is, of course, that HR-pQCT cannot analyze the spine but is confined to peripheral sites. However, several studies have previously shown that structural changes in the peripheral bones reflect bone changes as well as fracture risk in the axial skeleton [
22,
23].
The reason why nr-axSpA virtually exclusively affects cortical bone is not entirely clear. It can be speculated that such changes reflect increased cortical bone remodeling. For instance, it has long been known that the remodeling of cortical bone is highly dependent on strain [
33], and later concepts even suggested that cortical bone remodeling may even be entirely dependent on microcracks [
34]. Hence, altered bone responses to mechanical forces in nr-axSpA may be triggers for increased cortical bone remodeling and bone loss. In accordance with this, prostaglandin E
2, a prototype inflammatory mediator released upon injury and having a central role in nr-axSpA, has been shown to enhance cortical bone remodeling [
35]. The disbalance in cortical bone in patients with nr-axSpA, however, may stem from cytokines such as interleukin-17, which effectively suppress bone formation but at the same time increase bone resorption [
36,
37].
Acknowledgements
This study was supported by the Bundesministerium fuer Bildung und Forschung (BMBF project METARTHROS), the Deutsche Forschungsgemeinschaft (CRC1181), the Marie Curie project Osteoimmune and the Innovative Medicines Initiative-funded project Be The Cure (BTCure). CPF was supported by Cienciassem Fronteiras from Conselho Nacional de Desenvolvimento Cientifico e Tecnologico (CNPq), Brazil.