Effect of rheumatoid arthritis and age on metacarpal bone shaft geometry and density: A longitudinal pQCT study in postmenopausal women

ARTICLE IN PRESS Seminars in Arthritis and Rheumatism 000 (2019) 1 8

Contents lists available at ScienceDirect

Seminars in Arthritis and Rheumatism journal homepage: www.elsevier.com/locate/semarthrit

Effect of rheumatoid arthritis and age on metacarpal bone shaft geometry and density: A longitudinal pQCT study in postmenopausal women €llera, P.M. Villigera D. Aeberlia,*, N. Fankhauserb, R. Zebazec,d, H. Bonele, B. Mo a

Department of Rheumatology, Immunology and Allergology, University Hospital and University of Bern, 3010 Bern Switzerland Clinical Trial Unit (CTU), University of Bern, 3012 Bern, Switzerland c Department of Endocrinology, Austin Health, University of Melbourne, Melbourne, Australia d Department of Medicine, School of Clinical Sciences, Monash Health, Monash University, Melbourne, Australia e Department of Radiology, University Hospital and University of Bern, Switzerland b

A R T I C L E

I N F O

Keywords: Rheumatoid Arthritis Metacarpal Bone Geometry pQCT

A B S T R A C T

Objective: This study aimed to elucidate the effects of changes in the geometry and density of the metacarpal bone of patients with rheumatoid arthritis (RA). Methods: This prospective study included consecutive postmenopausal RA patients who met the American College of Rheumatology Criteria and healthy controls (HC). Peripheral quantitative computed tomography scans at 50% of the total metacarpal shaft (third metacarpal bone) were obtained at baseline and follow-ups. Use of bisphosphonates (BP), glucocorticoids (GC), biologics, and disease-modifying anti-rheumatic drugs (DMARD) was monitored (baseline to follow-up). Total cross-sectional area (CSA), cortical-transitional zone and compact zone CSA, cortical volumetric bone mineral density, and compact cortex porosity were measured. A linear mixed-effects model was used to determine significant differences in the rate of change in the RA and control groups and in RA patient subgroups. Results: Thirty-nine RA patients and 42 consecutive postmenopausal HC were followed for 63 months. RA and HC depicted a time-dependent increase of medullary CSA (+0.41 mm2/year, P < 0.0001), while total CSA remained stable (P = 0.2). RA status was associated with a loss of cortical bone mineral density (interaction: 3.08 mg/mm3; P = 0.014). In RA subgroup analysis, GC use 5 mg/day was positively correlated with a fourfold increase of medullary CSA (0.67 mm2/year P = 0.009), which resulted in a three- to fourfold loss of cortical density ( 6.6 mg/mm3/year; P = 0.002) and cortical CSA ( 0.57 mm2/year, P = 0.004). Patients with high disease activity and high GC dose at baseline demonstrated an increase in the total CSA (0.29 mm2/y; P = 0.049) and a loss of cortical BMD ( 5.73 mg/mm3/y; P = 0.05) despite good clinical response. Conclusion: Increase in medullary metacarpal CSA and thinning of the cortical CSA are physiological and time dependent. RA status is associated with loss in cortical density. Even upon biological therapy, low glucocorticoid dose affects metacarpal bone shaft geometry and density over time. © 2019 Elsevier Inc. All rights reserved.

Introduction Rheumatoid arthritis (RA) is associated with articular bone erosion, juxta-articular osteopenia, and systemic bone loss [1,2]. Reduced bone mass at the metacarpal shaft in RA, detected using digital X-ray radiogrammetry, or at the hand, determined using dual Xray absorptiometry, has been documented in a number of studies [3 7]. A previous study that used high-resolution peripheral Abbreviations: ACPA, anti-cyclic citrullinated peptide antibody; BMD, bone mineral density; CSA, cross-sectional area; DAS28, disease activity composite score; DMARD, disease-modifying anti-rheumatic drugs; FUP, follow-up; HA, hydroxyapatite; pQCT, peripheral quantitative computed tomography; RA, rheumatoid arthritis; RF, rheumatoid factor; SSI, stress strain index * Corresponding author. E-mail address: [email protected] (D. Aeberli). https://doi.org/10.1016/j.semarthrit.2019.08.003 0049-0172/© 2019 Elsevier Inc. All rights reserved.

quantitative computed tomography (pQCT) also reported decreased cortical bone quantity at the distal radius in RA patients [8]. Of importance, loss of metacarpal shaft density on plain radiographs has been found to be predictive of subsequent joint damage in patients with active RA [9] and a reliable key finding in the diagnosis of and a surrogate marker for RA progression [3]. For bone geometry, cross-sectional studies showed that RA patients have increased cross-sectional area (CSA) at the midshaft and reduced cortical thickness compared with healthy controls (HC) [10], and cortical thickness and metacarpal index have been found to be inversely correlated with bone erosiveness [10,11]. Similar results were found in juvenile idiopathic arthritis patients: articular and periarticular inflammation was associated with bone loss and changes in bone geometry, particularly reduced cortical thickness and increased bone CSA [12 15]. The specificity of these RA-induced inflammatory geometric changes was

ARTICLE IN PRESS 2

D. Aeberli et al. / Seminars in Arthritis and Rheumatism 00 (2019) 1 8

determined by neutralization of driving forces, such as receptor activator of nuclear factor kappa B ligand (RANKL), by treatment with denosumab, which in turn resulted in attenuation of the decreased cortical thickness in RA patients [16] and the effect of anti-tumor necrosis factor (TNF) therapy [6]. Besides, the inflammatory effect of RA and its counter-regulation by glucocorticoids, bone geometry, and density were found also to be dependent on age [17] and muscle force [18 21]. The latter is of particular interest, since muscle force and loading were shown to have a high impact on bone adaptation and geometry [19]. Therefore, it is critical to interpret changes in bone geometry and density in patients with RA and HC as a product of disease, therapy, age, and muscle force. Bone is a three-dimensional (3D) structure. Thus, accurate assessment of its properties in clinical and research settings requires a technology that can capture the 3D spatial distribution of the mineralized bone matrix. Dual X-ray absorptiometry is inexpensive and has a relative low radiation dose (effective dose of less than 0.01 mSv) [22], but it is a two-dimensional (2D) projection measurement of a 3D object, which limits the geometric and structural information that can be derived from this imaging modality and thus its ability to assess bone material and structural composition, and therefore fracture risk. The same limitation applies to other 2D-based imaging such as Digital X-ray radiogrammetry (DXR) which estimates hand BMD from standard hand 2D X-ray images. To overcome the limitations of these 2D imaging technologies, quantitative computed tomography (QCT) is used to assess bone properties. It provides 3D measurement properties. However, QCT assesses bone at moderately high resolution (voxels  1 mm in all dimensions), which is a limitation. To compensate for the low radiation of QCT, highresolution peripheral quantitative computed tomography (HR-pQCT) was introduced in the assessment of the bone. HRpQCT uses a dedicated gantry for small unmoved volumes of peripheral body parts, which enables the system a much higher precision compared to a diagnostic CT scanner [23]. HRpQCT is a non-invasive research tool and not reimbursable. This low-radiation method is particularly used for assessing bone microarchitecture and volumetric bone mineral density (BMD) in cortical and trabecular compartments of the distal radius and distal tibia. This is a technological advance as it allows the assessment of bone at a much higher spatial resolution and smaller slice thickness with an isotropic voxel size of 82 mm. Furthermore, the radiation dose is low. The HR-pQCT single-scan effective dose is estimated to be 3 mSv [24]. Altogether, it provides a detailed spatial distribution of mineralized bone matrix in vivo settings. These advantages allowed us in this paper to assess metacarpal bone geometry and cortical density using StrAx software analysis for further segmentation of the cortical bone into its compact-appearing, outer, and inner transitional zones [25]. Data on longitudinal changes in the geometry and density of the metacarpal bone are scarce. Currently, no study has examined whether changes in metacarpal bone geometry and density are physiological and therefore dependent on time, or are primarily an effect of RA. The aims of this prospective, longitudinal study were therefore to determine these factors by comparing an RA group and a HC group over approximately 5 years.

of Rheumatology, Immunology and Allergology, Inselspital Bern, were included. For the control group, we recruited healthy female volunteers by locally distributed flyers and advertisement on the hospital internal web. The inclusion criterion for both groups was age greater than 45 years. The exclusion criteria for both groups were bone metabolic diseases, hyper-/hypoparathyroidism, hyper-/hypothyroidism, chronic renal insufficiency, anabolic bone therapy, RANKL-Ab (denosumab), cancer, pregnancy, and drug addiction, which were determined based on medical history and questionnaires for osteoporosis risk factors. For the control group, but not the RA group, exclusion criteria also included established osteoporosis and previous or present bisphosphonate therapy. All patients and volunteers provided written informed consent. For the follow-up analysis, RA patients and HC were contacted by phone and by mail. Assessment of disease characteristics Erosiveness was assessed using total Ratingen score [27] for the non-dominant hand by a blinded radiologist. Data on rheumatoid factor (RF), anti-cyclic citrullinated peptide antibody (ACPA), disease duration, disease activity composite score (DAS28) [28], therapy with biological or conventional DMARD (b or cDMARD), and bisphosphonate and glucocorticoid use were obtained from longitudinal clinical examinations and medical records. Bone measurements Bone measurements were performed with a Stratec XCT 3000 scanner (Stratec Medical, Pforzheim, Germany), which is a pQCT apparatus that measures attenuation of X-rays linearly transformed into hydroxyapatite (HA) densities. Unlike other pQCT scanners, Stratec XCT 3000 is calibrated with respect to water and is set at 60 mg of HA; thus, fat results in 0 mg of HA [29]. HA equivalent densities are automatically calculated from attenuation coefficients by employing the manufacturer’s phantom, which is calibrated with respect to the European Forearm Phantom (QRM, Erlangen, Germany) [29]. pQCT measurements of the metacarpal bone were performed on the non-dominant hand. Metacarpal measurements Measurements of the metacarpal bone were performed, as described previously [10]. The length of the third metacarpal bone of the non-dominant hand was palpated and measured from the base to the head (to the nearest 5 mm) using a measuring tape. A scout view of the head of the third ossa metacarpalia was obtained, and the reference line was placed at the distal end of the bone. Scans were performed at 50% of the total bone length measured from the distal bone end. Slice thickness was 2.2 mm, voxel size was set at 0.3 mm edge length, and scanning speed was set at 15 mm/s. Follow-up measurements were considered valid when circumference and bone shape corresponded to baseline measurements. Parameter measurement

Materials and methods We conducted a prospective, longitudinal observational study that compared postmenopausal women with RA with an HC group. The study protocol was approved by the Ethics Committee of the Canton of Bern (Number 168/07). The study was conducted in compliance with the Declaration of Helsinki. Subjects Consecutive female patients with RA who met the American College of Rheumatology criteria [26] and were seen in the Department

The threshold for the periosteal surface was set at 280 mg/cm3 [10], and from this, the total CSA was measured. The cortical bone was selected with the threshold set at 710 mg/cm3 [10], and from this, cortical CSA (excluding bone marrow space) and cortical BMD were measured. Cortical thickness and periosteal and endosteal circumferences were calculated under the assumption that the bone shaft is cylindrical based on the total CSA, which included the bone marrow and cortical CSA from the diaphyseal scans (50% for metacarpals). Polar stress strain index (SSI) is a measure of diaphyseal bone resistance to bending and torsion [30] and can be used as a surrogate for bone strength [31]. For the differentiation of compact and transitional

ARTICLE IN PRESS D. Aeberli et al. / Seminars in Arthritis and Rheumatism 00 (2019) 1 8

3

Fig. 1. Study participant flow chart.

cortex and measurement of the porosity of the compact cortex, baseline and last follow-up measurements were obtained and analyzed using StrAx software (StraxCorp, Melbourne, Australia). The cortex was further segmented into its compact-appearing, outer, and inner transitional zones; thus, the porosity and cortical fragments produced by unbalanced intracortical remodeling are confined to the transitional zone [25]. Porosity in the total cortical compartment was also analyzed.

average, 2.9 follow-up scans were measured. The two groups were comparable with regard to age, height, weight, and total observation time. However, RA patients had a 20% lower handgrip force (P = 0.01). Of the patients, 79% of patients (n = 31) were on glucocorticoid therapy, but none in the control group. The subject characteristics are presented in Table 1.

Handgrip force

Baseline bone characteristics in RA group and HC group

Because handgrip force correlates with metacarpal bone geometry [11], we measured grip strength direct either by a Leonardo GF sensor (Leonardo GF Typ 8N600591A-SNovotec Medical GmbH, Pforzheim, Germany) or by a soft-tissue dynamometer.

The total and medullar CSA of the metacarpal bone were 7% and 17%, respectively, greater in the RA group than in the HC group (P = 0.013 and P = 0.004, respectively). Cortical thickness was 7% thinner in RA patients (P = 0.041) (for representative scans, please see Fig 2). The porosity of compact cortical bone was increased by 12% (P = 0.026), the area of transitional zone of the cortical bone was wider by 21% (P = 0.005), and the compact cortical zone was lowered by 10% (P = 0.041) in the RA group. No difference was found in cortical BMD at baseline. The baseline bone parameters are shown in Table 2.

Statistical analysis Missing values for handgrip force (16 of 81 participants) were imputed using multivariate imputation by chained equations using the mice R package on the variables age, radius 66% muscle area, and RA status. Linear mixed-effects models from the “lme4” package were used to predict outcome progression in patient subgroups. RA status or further subgroup variables as well as observation time were used as fixed effects. A random intercept was used for individual participants, and a random slope was used for observation time. The presence of significant interactions between the grouping variable and observation time was determined based on the p-value computed by the “lmerTest” package. Confounding variables were not added, as all relevant parameters (age, weight, height) were balanced in this study. All analyses were performed using R version 3.4.1 for Windows (2017-06-30). Results

Clinical parameters The RA group had the following disease characteristics (median [IQR]) at baseline: disease duration (months) (198.0 [174.0, 282.0]); DAS28 BSR (3.4 [2.5, 4.4]), Ratingen score (0.5 [0.0, 2.3]); treatment with bisphosphonates (n = 16; 41.03%), cDMARDs (n = 16, 41.03%), bDMARDs (n = 27; 69.2%), prednisone dose (mg/day) throughout the observational period (2.6 [0.2, 5.0]); and titer RF (34.0 [9.0, 87.0]). RF positivity was noted in 62% and ACPA positivity in 75% of patients.

Table 1 Subjects’ anthropometric data (median and [IQR])

Subject parameters Initially, 86 HC and 68 RA patients were enrolled in this study. For the follow-up measurements, 42 HC and 39 RA patients were eligible; dropouts were due to death (n = 2), fracture of the forearm (n = 2), joint prosthesis (n = 1), loss to follow-up (n = 35), medication use (anabolic bone therapy, denosumab, estrogen replacement) (n = 16), metabolic diseases (n = 9), and tumor (n = 8). The exact patient flow is presented in Fig. 1. The median [interquartile range (IQR)] of the total observation time was 63.50 months [44.3, 74.0] in HC and 55.00 months [36.0, 71.5] in RA patients (P = 0.33). On

Age (median [IQR]) Age first scan (median [IQR]) Observation time (mo) (median [IQR]) Height (median [IQR]) Weight (median [IQR]) Handgrip_(mm Hg) (median [IQR]) Prednisone Therapy (%)

Ref (n = 42)

RA (n = 39)

72.87 [65.49, 81.15] 62.13 [55.00, 72.42] 63.50 [44.25, 74.00]

68.68 [64.41, 73.70] 58.92 [54.84, 63.77] 55.00 [36.00, 71.50]

164.00 [158.00, 168.00] 159.25 [156.75, 166.25] 66.00 [59.00, 75.00] 70.00 [62.50, 76.25] 200.00 [160.00, 260.00] 160.00 [117.50, 200.00] 0 (0.00)

31 (79.49)

P value 0.071 0.133 0.332 0.099 0.290 0.010 <0.001

IQR indicates the lower (25%) and upper (75%) quartiles. P values were obtained using the Kruskal Wallis test. IQR, interquartile range. P < 0.05 is indicated in bold.

ARTICLE IN PRESS 4

D. Aeberli et al. / Seminars in Arthritis and Rheumatism 00 (2019) 1 8

cortical BMD (Fig. 3 and Supplementary Fig. 1). RA and HC groups demonstrated an age-dependent loss of cortical CSA of 0.35 mm2/ year [95% confidence interval (CI): 0.46 to 0.23 mm2/y, P = 1.32e07) and an increase of medullar CSA of 0.41 mm2/year (95% CI: 0.26 0.56 mm2/year, P = 4.91e-07), whereas total CSA remained stable (+0.06 mm2/y) (95% CI: 0.04 to 0.16 mm2/y; p-value = 0.212). RA status was associated with a significantly higher loss of cortical BMD (P = 0.014; interaction 3.08 mg/mm3), which holds true after correction for muscle area and handgrip force. Relationship between disease characteristics and bone parameters at baseline When comparing ACPA/RF +/+ with -/-, we found higher compact cortex porosity (P = 0.05), higher transitional zone CSA (P = 0.044), lower cortical BMD (P = 0.048), and lower compact cortical CSA (p = 0.038) in ACPA/RF +/+. Patients with a Ratingen score greater than 0.5 at baseline had lower total and compact cortical CSA (P = 0.044 and P = 0.024, respectively) and lower cortical thickness (P = 0.032). With regard to disease activity at baseline and disease duration, no difference was found for in-bone parameters. Association of changes in bone geometry and densitometry with disease characteristics Fig. 2. pQCT scans were performed at 50% of the total bone length of the third metacarpal bone (MCP 3) of the non-dominant hand. The left side is a cross-section of the total midhand with the arrow on MCP 3. The right side is a magnification of MCP 3. Top panel depicts a pQCT scan of a 70-year-old RA patient. Compared to healthy control (bottom panel), cortical thickness is reduced and total CSA and medullar CSA are increased. Bottom panel shows a pQCT scan of a 75-year-old healthy control. CSA, cross-sectional area; pQCT, peripheral quantitative computed tomography.

Change in bone characteristics in RA and HC groups The linear model predictions superimposed on the raw measurements are depicted for total CSA, cortical CSA, medullary CSA and

Table 2 Baseline bone parameters at the third metacarpal bone in RA patients and healthy controls

Polar SSI, mm3 (median [IQR]) Endosteal circumference, mm (median [IQR]) Periosteal circumference, mm (median [IQR]) CSA medulla, mm2 (median [IQR]) Total CSA, mm2 (median [IQR]) Cortical CSA, mm2 (median [IQR]) Cortical thickness, mm (median [IQR]) Cortical BMD, mg/mm3 (median [IQR]) Compact cortex CSA, mm2 (median [IQR]) Transitional zone CSA, mm2 (median [IQR]) Transitional zone thickness, mm (median [IQR]) Compact cortex porosity, % (median [IQR])

Ref (n = 42)

RA (n = 39)

P value

73.09 [67.55, 83.00]

76.36 [62.19, 87.02]

0.865

16.87 [15.12, 17.88]

18.57 [15.65, 20.58]

0.004

26.79 [25.72, 27.63]

27.75 [26.39, 28.92]

0.013

22.63 [18.18, 25.45]

27.45 [19.49, 33.70]

0.004

57.11 [52.63, 60.75]

61.29 [55.44, 66.56]

0.013

34.97 [32.72, 37.55]

34.11 [29.97, 39.42]

0.502

1.64 [1.47, 1.79]

1.52 [1.26, 1.66]

0.041

1179.40 [1151.72, 1204.00] 28.24 [25.49, 30.90]

1157.70 [1115.30, 1195.25] 25.46 [21.44, 29.76]

0.069 0.041

11.85 [8.58, 15.02]

15.01 [11.94, 18.97]

0.005

0.85 [0.74, 0.95]

0.88 [0.81, 1.01]

0.105

19.33 [17.86, 21.98]

22.01 [18.68, 26.75]

0.026

IQR indicates the lower (25%) and upper (75%) quartiles. P values were obtained using the Kruskal Wallis test. BMD, bone mineral density; CSA, cross-sectional area; IQR, interquartile range; SSI, stress strain index. P < 0.05 is indicated in bold.

The changes in bone geometry and densitometry were analyzed based on ACPA/RF positivity, disease activity at baseline, improvement of disease activity, erosion score, and disease duration. Analysis of bone parameters in relation to DAS28 improvement from baseline to the last follow-up showed that DAS28 improvement of greater than 1.2 was significantly correlated with increased total CSA (interaction +0.29 mm2; P = 0.049) and a loss of cortical BMD ( 5.73 mg/ mm3; P = 0.05) (Tables 3 and 4). Patients with a DAS improvement greater than 1.2 showed higher disease activity at baseline than those with an improvement less than 1.2 (median IQR DAS28 = 4.61 vs. 2.87; P < 0.001) and were more frequently on glucocorticoid therapy (100% vs. 78%) (results not shown). No correlation with changes in metacarpal bone parameters, ACPA/RF positivity, nor Ratingen score was found. Nevertheless, disease duration tended to have an effect on cortical and medullary CSA (Tables 3 and 4). A forest plot of the change per year and 95% CI of geometric and densitometric bone parameters are shown in Fig. 4. Influence on DMARD, bisphosphonate therapy, and prednisone on bone parameters We analyzed the effect of bDMARDs (rituximab, abatacept, infliximab, etanercept adalimumab, golimumab, or tocilizumab) at any time on bone parameters. No significant influence was found in group difference testing and mixed model analysis. The same was found for cDMARDs (methotrexate, hydroxychloroquine, leflunomide, or sulfasalazine). Those with bisphosphonate therapy tended to have increased compact cortex porosity (P = 0.052), lower cortical CSA (P = 0.003), greater medullary CSA (P = 0.06), and lower polar SSI (P = 0.034) at baseline. However, no difference in change per year was found. Patients on prednisone dose of 5 mg/day or greater exhibited no significant difference in bone parameters at baseline. However, in longitudinal comparisons, glucocorticoid use of 5 mg/day or greater was positively correlated with increased medullary CSA (interaction +0.67 mm2; P = 0.009) and resulted in a loss of cortical density ( 6.6 mg/mm3; P = 0.002) and cortical CSA ( 0.57 mm2; P = 0.004) compared with the group who took glucocorticoid in a dose less than 5 mg/day (Tables 3 and 4 and Fig. 4).

ARTICLE IN PRESS D. Aeberli et al. / Seminars in Arthritis and Rheumatism 00 (2019) 1 8

5

Fig. 3. Model prediction of raw data of geometric and densitometric bone changes. RA (solid line, dots), healthy controls (dashed line, triangles). RA, rheumatoid arthritis.

Table 3 RA subgroup analysis with change per year (95% CI) Total CSA [mm2]

Control RA ACPA-/RFACPA+/RF+ Prednisone <5 mg/day Prednisone 5 mg/day DAS28 BL—FUP  1.2 DAS28 BL—FUP > 1.2 DAS28 BL <3.2 DAS28 BL >3.2 Ratingen BL <0.5 Ratingen BL >0.5 Disease > 15 y Disease <15 y

0.09 ( 0.05 0.23) 0.04 ( 0.10 0.18) 0.16 ( 0.13 0.45) 0.15 ( 0.02 0.32) 0.02 ( 0.19 0.22) 0.09 ( 0.20 0.37) 0.06 ( 0.24 0.13) 0.32 (0.00 0.64) 0.06 ( 0.28 0.17) 0.14 ( 0.09 0.37) 0.14 ( 0.14 0.42) 0.01 ( 0.22 0.20) 0.04 ( 0.15 0.24) 0.03 ( 0.27 0.32)

Medulla CSA [mm2]

0.35 (0.14 0.55) 0.49 (0.27 0.70) 0.88 (0.31 1.44) 0.69 (0.37 1.02) 0.22 ( 0.09 0.53) 0.96 (0.55 1.38) 0.38 (0.06 0.69) 0.84 (0.30 1.39) 0.28 ( 0.10 0.65) 0.71 (0.33 1.10) 0.49 (0.00 0.98) 0.49 (0.13 0.86) 0.65 (0.33 0.96) 0.15 ( 0.32 0.61)

Cortical CSA [mm2]

0.28 ( 0.43 ( 0.67 ( 0.50 ( 0.19 ( 0.81 ( 0.41 ( 0.47 ( 0.33 ( 0.53 ( 0.35 ( 0.45 ( 0.57 ( 0.14 (

0.43 0.60 1.21 0.81 0.43 1.12 0.66 0.89 0.62 0.83 0.73 0.72 0.81 0.49

0.12) 0.26) 0.13) 0.18) 0.05) 0.50) 0.17) 0.04) 0.03) 0.23) 0.02) 0.18) 0.32) 0.22)

Porosity of the compact cortex [percent] 0.04 ( 0.30 ( 0.45 ( 0.18 ( 0.01 ( 0.83 ( 0.15 ( 0.82 ( 0.12 ( 0.52 ( 5.78 ( 5.99 ( 6.09 ( 4.77 (

0.40 0.16 0.93 0.62 0.73 0.20 0.54 0.34 0.74 0.34 9.44 8.70 8.61 8.44

0.47) 0.77) 1.83) 0.98) 0.76) 1.86) 0.84) 1.98) 0.97) 1.37) 2.13) 3.28) 3.57) 1.09)

Cortical BMD [mg/mm3]

2.54 ( 5.61 ( 7.68 ( 4.55 ( 3.25 ( 9.52 ( 4.44 ( 9.33 ( 5.13 ( 6.25 ( 0.14 ( 0.01 ( 0.04 ( 0.03 (

4.17 0.90) 7.36 3.87) 11.57 3.78) 6.80 2.30) 5.43 1.07) 12.43 6.60) 6.76 2.11) 13.38 5.29) 8.04 2.21) 9.21 3.28) 0.14 0.42) 0.22 0.20) 0.15 0.24) 0.27 0.32)

Outcome progression as predicted by the linear mixed-effects model. The first two rows show the progression in the control compared with the RA group. The following pairs of rows are further subdivisions of the RA group. The numbers in parentheses are the lower and upper 95% confidence intervals computed by the model. ACPA, anti-cyclic citrullinated peptide antibody; BL, baseline; FUP, follow-up; RA, rheumatoid arthritis; RF, rheumatoid factor; DAS28, disease activity composite score.

ARTICLE IN PRESS 6

D. Aeberli et al. / Seminars in Arthritis and Rheumatism 00 (2019) 1 8 Table 4 Associations of Interaction by factors of RA disease characteristics (95% CI) Total CSA [mm2] RA ACPA+/RF+ Prednisone >5 mg/d DAS28 BL FUP >1.2 DAS28 BL Ratingen >0.5 Disease Duration >15 y

0.05 ( 0.01 ( 0.03 ( 0.29 ( 0.09 ( 0.15 ( 0.03 (

0.25 0.34 0.24 0.01 0.14 0.50 0.41

0.15) 0.33) 0.30) 0.59) 0.33) 0.20) 0.34)

P= 0.619 0.964 0.684 0.049 0.246 0.406 0.930

Medulla CSA [mm2] 0.14 ( 0.16 0.43) 0.18 ( 0.85 0.47) 0.67 (0.30 1.03) 0.43 ( 0.02 0.89) 0.38 (0.04 0.72) 0.00 ( 0.61 0.61) 0.61 ( 1.20 0.02)

P= 0.361 0.591 0.009 0.158 0.121 0.993 0.091

Cortical CSA [mm2] 0.15 ( 0.38 0.08) 0.17 ( 0.45 0.80) 0.57 ( 0.85 0.28) 0.13 ( 0.49 0.24) 0.24 ( 0.51 0.03) 0.10 ( 0.56 0.37) 0.55 (0.09 1.00)

P= 0.205 0.591 0.004 0.843 0.345 0.685 0.058

Cortical BMD [mg/mm3] 3.08 ( 3.12 ( 6.63 ( 5.73 ( 2.77 ( 0.21 ( 2.72 (

5.51 1.07 9.55 9.56 5.66 4.79 1.99

0.70) 7.51) 3.75) 1.98) 0.10) 4.31) 7.32)

P= 0.014 0.231 0.002 0.050 0.602 0.930 0.565

P-values and interactions (lower and upper 95% confidence intervals) between the grouping factors and observation time were computed using a linear mixed-effects model. ACPA, anti-cyclic citrullinated peptide antibody; BMD, bone mineral density; CSA, cross-sectional area; DAS28 BL, DAS 28 baseline; RA, rheumatoid arthritis; RF, rheumatoid factor. P < 0.05 is indicated in bold.

Fig. 4. Forest plot of change per year and 95% confidence interval of total CSA, cortical CSA, medullary CSA, and cortical BMD in relation with ACPA/RF positivity, prednisone dose/ day, disease activity, Ratingen score, and disease duration. Outcome progression as predicted by the linear mixed-effects model. Each panel plots the progression of one outcome. The first two items in each panel show the progression in the control compared with the RA group. The following items are further subdivisions of the RA group. The progression as change per year is indicated by the position of the black square relative to the x-axis labels. The 95% confidence intervals are represented by a gray line around the square. ACPA, anti-cyclic citrullinated peptide antibody; BMD, bone mineral density; CSA, cross-sectional area; RA, rheumatoid arthritis; RF, rheumatoid factor.

ARTICLE IN PRESS D. Aeberli et al. / Seminars in Arthritis and Rheumatism 00 (2019) 1 8

Discussion To the best of our knowledge, this is the first prospective longitudinal pQCT study to elucidate whether changes in geometry and volumetric density of the metacarpal bone shaft are physiological and show an aging pattern over time or are driven by disease characteristics and therapy of RA disease. In this study, we found changes in bone geometry such as total CSA, medullary CSA, and cortical CSA to be comparable in the HC and RA groups; however, significant loss of cortical density was found only in the RA group. After controlling for disease-specific parameters, RA status and glucocorticoid dose were associated with loss of cortical density; low-dose glucocorticoid dependency such as 5 mg of prednisolone per day was associated with widening of the medullary CSA and thinning of the cortical area at a comparable increase in outer bone shaft circumference. In addition, patients with high disease activity and high glucocorticoid dose at baseline demonstrated an increase in total CSA and a loss of cortical BMD despite good clinical response. RA causes systemic skeletal breakdown and increases the risk of non-traumatic fractures [32,33]. By densitometry, RA patients have lower total BMD at the distal metacarpal bone [34], distal radius [8], and hip [35 37]. At the metacarpal shaft, cortices are thinner, and the outer diameter tends to increase in RA patients [10,12]. In line with these findings, Kocijan et al. showed that cortical density and thickness of the distal radius are lower, and cortical perimeter at the ultra-distal radius is increased in RA patients. In our study, baseline data showed an identical pattern as in Kocijan et al. In addition, we found an increased transitional-cortical zone and higher cortical porosity in RA patients. These results are most likely signs of an enhanced trabecularization of the inner part of the cortex and increased remodeling and enlargement of the Haversian canals [25]. These results are line with histomorphometric studies from biopsies collected at the femoral neck, which show an increased number of osteoclasts in RA patients [38]. Metacarpal hand bone loss was found to be a sign of persistent disease activity and an independent predictor of radiological progression of RA [39 41]. Eser et al. assessed factors associated with differences in bone geometry. Based on linear models with exploratory variables (muscle CSA, age, RA status, and sex), they found that RA patients have a greater age-related decrease in cortical thickness associated with an increase in outer bone circumference [11]. Kocijan confirmed such an enhanced aging pattern in RA after calculating the regression in the RA group compared with the HC group [8]. In our longitudinal study, changes in cortical loss and increase in medullary CSA were found to be age-dependent, both in RA patients and HC. However, for cortical bone mineral density, positive RA status was found to be associated with a significantly higher loss. This is of special interest since cortical hand bone loss is a surrogate of generalized bone loss. For example, porosity—a proxy of cortical bone loss—measured using DXR radiogrammetry at the hand distinguished women with a distal radius fracture from HC, and this is even after adjustment for confounders such as age, body mass index (BMI), and smoking. Furthermore, it remained significant when adjusted for BMD measurements such as femoral neck BMD [42]. Although an inverse relationship between the development of erosions at the wrist and fingers and loss of areal BMD at the metacarpal bone [1,43] or at the hand [44] has been described, we found only a group difference in Ratingen score severity and ACPA/RF positivity. No association was found in our study between bone geometry and bDMARDs or a neutralizing antibody, despite the fact that bDMARDs, such as anti-TNF, are potential inhibitors of bone resorption [45]. This finding may point to the fact that glucocorticoid use and inflammation may be more influencing factors on bone geometry in this longitudinal study, as earlier postulated by Roth et al. [12].

7

Glucocorticoids were demonstrated to have a strong negative effect on bone metabolism [46], which in turn leads to deterioration of bone structure, bone quality, and bone quantity and to a reduction in biomechanical properties of vertebral and tibial bones [47]. The latter is observed particularly in inflammatory diseases, such as RA [48]. Glucocorticoid use, quantified either by daily or cumulative dose, is strongly related to increased fracture risk [49,50]. Prednisone use greater than 5 mg/day is associated with vertebral fractures in postmenopausal women within the first 6 months of treatment [51]. In a recent cross-sectional analysis of the distal radius by Kocijan et al., high-dose glucocorticoids were associated with cortical thinning; however, no effects on the bones were found for glucocorticoid use less than 7.5 mg/day [8]. A study by Roldan et al. showed a negative effect of glucocorticoids and systemic inflammation on combined cortical thickness over 2 years [52]. So far, the effect of RA and glucocorticoids on volumetric density and bone geometry were not studied. In our study, we demonstrated the effects of dosages of glucocorticoids of 5 mg/day or higher, such as widening of the medullary CSA, thinning of the cortical area, and loss of cortical density. In this study, we showed that even in patients mostly on biological treatment and a low-dose glucocorticoid, the influence on bone geometry is present and may lead to bone damage and increased fracture risk. The limitations of our study are the number of subjects who drop out over 5 years and the low number of patients. To estimate the bias involved by the loss of participants, we compared baseline data of eligible RA patients and HC to excluded individuals. With regard to age, height, weight, and handgrip, eligible and excluded RA patients were comparable. However, bone measurements at baseline of excluded RA patients showed lower polar SSI and lower cortical thickness. In HC, excluded patients were older and weighed less. With regard to bone parameter at baseline, excluded HC were comparable in polar SSI and cortical thickness (data not shown). We therefore estimate the bias of excluded individuals as alleviated; particularly, RA patients with distinct geometric pattern were excluded, and analysis was adjusted to muscle force.

Conclusions An increase in medullary CSA and loss of cortical CSA at comparable total CSA are physiological changes on metacarpal bone geometry in RA patients and HC. RA status had a negative effect on metacarpal bone density only, whereas geometry was not affected over 5 years. In a sub-analysis, low-dose glucocorticoids correlated with an increase in medullary CSA, loss of cortical area, and loss of cortical BMD. This study highlights that even low doses of glucocorticoids affect metacarpal bone shaft geometry and density over time.

Funding This study was funded by the Novartis Foundation for MedicalBiological Research to DA.

Declaration of Competing Interest The authors declare that they have no competing interests.

Acknowledgments We thank all the study subjects for their time and effort. We appreciate the careful work of Ms. Inna Galli Lysack, who helped with the pQCT measurements.

ARTICLE IN PRESS 8

D. Aeberli et al. / Seminars in Arthritis and Rheumatism 00 (2019) 1 8

Supplementary materials Supplementary material associated with this article can be found in the online version at doi:10.1016/j.semarthrit.2019.08.003. References [1] Stewart A, Mackenzie LM, Black AJ, Reid DM. Predicting erosive disease in rheumatoid arthritis. A longitudinal study of changes in bone density using digital X-ray radiogrammetry: a pilot study. Rheumatol (Oxford) 2004;43(12):1561–4. [2] Peel NF, Spittlehouse AJ, Bax DE, Eastell R. Bone mineral density of the hand in rheumatoid arthritis. Arthritis Rheum 1994;37(7):983–91. [3] Pfeil A, Haugeberg G, Renz DM, Reinhardt L, Jung C, Franz M, et al. Digital X-ray radiogrammetry and its sensitivity and specificity for the identification of rheumatoid arthritis-related cortical hand bone loss. J Bone Min Metab 2017;35(2):192–8. [4] Boyesen P, Hoff M, Odegard S, Haugeberg G, Syversen SW, Gaarder PI, et al. Antibodies to cyclic citrullinated protein and erythrocyte sedimentation rate predict hand bone loss in patients with rheumatoid arthritis of short duration: a longitudinal study. Arthritis Res Ther 2009;11(4):R103. [5] Hoff M, Haugeberg G, Kvien TK. Hand bone loss as an outcome measure in established rheumatoid arthritis: 2-year observational study comparing cortical and total bone loss. Arthritis Res Ther 2007;9(4):R81. [6] Hoff M, Haugeberg G, Odegard S, Syversen S, Landewe R, van der Heijde D, et al. Cortical hand bone loss after 1 year in early rheumatoid arthritis predicts radiographic hand joint damage at 5-year and 10-year follow-up. Ann Rheum Dis 2009;68(3):324–9. [7] Hoff M, Kvien TK, Kalvesten J, Elden A, Haugeberg G. Adalimumab therapy reduces hand bone loss in early rheumatoid arthritis: explorative analyses from the premier study. Ann Rheum Dis 2008. [8] Kocijan R, Finzel S, Englbrecht M, Engelke K, Rech J, Schett G. Decreased quantity and quality of the periarticular and nonperiarticular bone in patients with rheumatoid arthritis: a cross-sectional HR-pQCT study. J Bone Min Res 2014;29 (4):1005–14. [9] Wevers-de Boer KV, Heimans L, Visser K, Kalvesten J, Goekoop RJ, van Oosterhout M, et al. Four-month metacarpal bone mineral density loss predicts radiological joint damage progression after 1 year in patients with early rheumatoid arthritis: exploratory analyses from the improved study. Ann Rheum Dis 2015;74(2):341–6. [10] Aeberli D, Eser P, Bonel H, Widmer J, Caliezi G, Varisco PA, et al. Reduced trabecular bone mineral density and cortical thickness accompanied by increased outer bone circumference in metacarpal bone of rheumatoid arthritis patients: a crosssectional study. Arthritis Res Ther 2010;12(3):R119. [11] Eser P, Aeberli D, Widmer J, Moller B, Villiger PM. Abnormal bone geometry at the metacarpal bone shaft of rheumatoid arthritis patients with maintained musclebone relationship. Arthritis Care Res (Hoboken) 2011;63(3):383–9. [12] Roth J, Linge M, Tzaribachev N, Schweizer R, Kuemmerle-Deschner J. Musculoskeletal abnormalities in juvenile idiopathic arthritis a 4-year longitudinal study. Rheumatol (Oxford) 2007;46(7):1180–4. [13] Roth J, Palm C, Scheunemann I, Ranke MB, Schweizer R, Dannecker GE. Musculoskeletal abnormalities of the forearm in patients with juvenile idiopathic arthritis relate mainly to bone geometry. Arthritis Rheum 2004;50(4):1277–85. [14] Burnham JM, Shults J, Dubner SE, Sembhi H, Zemel BS, Leonard MB. Bone density, structure, and strength in juvenile idiopathic arthritis: importance of disease severity and muscle deficits. Arthritis Rheum 2008;58(8):2518–27. [15] Bechtold S, Ripperger P, Dalla Pozza R, Schmidt H, Hafner R, Schwarz HP. Musculoskeletal and functional muscle-bone analysis in children with rheumatic disease using peripheral quantitative computed tomography. Osteoporos Int 2005;16(7):757–63. [16] Sharp JT, Tsuji W, Ory P, Harper-Barek C, Wang H, Newmark R. Denosumab prevents metacarpal shaft cortical bone loss in patients with erosive rheumatoid arthritis. Arthritis Care Res (Hoboken) 2010;62(4):537–44. [17] Milovanovic P, Adamu U, Simon MJ, Rolvien T, Djuric M, Amling M, et al. Age- and sex-specific bone structure patterns portend bone fragility in radii and tibiae in relation to osteodensitometry: a high-resolution peripheral quantitative computed tomography study in 385 individuals. The J Gerontol Ser A, Biol Sci Med Sci 2015;70(10):1269–75. [18] Frost HM, Ferretti JL, Jee WS. Perspectives: some roles of mechanical usage, muscle strength, and the mechanostat in skeletal physiology, disease, and research. Calcif Tissue Int 1998;62(1):1–7. [19] Frost HM. In: Frost HM, editor. The Utah paradigm of skeletal physiology, 1. ISMNI; 2004. [20] Frost HM. From Wolff's law to the mechanostat: a new ''face'' of physiology. J Orthop Sci 1998;3(5):282–6. [21] Frost HM. The role of changes in mechanical usage set points in the pathogenesis of osteoporosis. J Bone Min Res 1992;7(3):253–61. [22] Damilakis J, Adams JE, Guglielmi G, Link TM. Radiation exposure in X-ray-based imaging techniques used in osteoporosis. Eur Radiol 2010;20(11):2707–14. [23] Laib A, Hauselmann HJ, Ruegsegger P. In vivo high resolution 3D-QCT of the human forearm. Technol Health Care 1998;6(5 6):329–37. [24] MacNeil JA, Boyd SK. Improved reproducibility of high-resolution peripheral quantitative computed tomography for measurement of bone quality. Med Eng Phys 2008;30(6):792–9. [25] Zebaze R, Seeman E. Cortical bone: a challenging geography. J Bone Min Res 2015;30(1):24–9.

[26] Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF, Cooper NS, et al. The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum 1988;31(3):315–24. [27] Rau R, Wassenberg S, Herborn G, Stucki G, Gebler A. A new method of scoring radiographic change in rheumatoid arthritis. J Rheumatol 1998;25(11):2094–107. [28] Prevoo ML, van’t Hof MA, Kuper HH, van Leeuwen MA, van de Putte LB, van Riel PL. Modified disease activity scores that include twenty-eight-joint counts. Development and validation in a prospective longitudinal study of patients with rheumatoid arthritis. Arthritis Rheum 1995;38(1):44–8. [29] Augat P, Gordon CL, Lang TF, Iida H, Genant HK. Accuracy of cortical and trabecular bone measurements with peripheral quantitative computed tomography (pQCT). Phys Med Biol 1998;43(10):2873–83. [30] Schiessl H, Ferretti J, Tysarczyk-Niemeyer G, Willnecker J. Noninvasive bone strength index as analyzed by peripheral quantitative computed tomography €nau E, editor. Paediatric osteology: new developments in diag(pQCT). In: Scho nostics and therapy. Amsterdam: Elsevier Science; 1996. p. 141–6. [31] Lochmuller EM, Lill CA, Kuhn V, Schneider E, Eckstein F. Radius bone strength in bending, compression, and falling and its correlation with clinical densitometry at multiple sites. J Bone Min Res 2002;17(9):1629–38. [32] Spector TD, Hall GM, McCloskey EV, Kanis JA. Risk of vertebral fracture in women with rheumatoid arthritis. BMJ 1993;306(6877):558. [33] Bernstein CN, Blanchard JF, Leslie W, Wajda A, Yu BN. The incidence of fracture among patients with inflammatory bowel disease. A population-based cohort study. Ann Intern Med. 2000;133(10):795–9. [34] Alenfeld FE, Diessel E, Brezger M, Sieper J, Felsenberg D, Braun J. Detailed analyses of periarticular osteoporosis in rheumatoid arthritis. Osteoporos Int 2000;11 (5):400–7. [35] Haugeberg G, Orstavik RE, Uhlig T, Falch JA, Halse JI, Kvien TK. Comparison of ultrasound and X-ray absorptiometry bone measurements in a case control study of female rheumatoid arthritis patients and randomly selected subjects in the population. Osteoporos Int 2003;14(4):312–9. [36] Martin JC, Munro R, Campbell MK, Reid DM. Effects of disease and corticosteroids on appendicular bone mass in postmenopausal women with rheumatoid arthritis: comparison with axial measurements. Br J Rheumatol 1997;36(1):43–9. [37] Gough AK, Lilley J, Eyre S, Holder RL, Emery P. Generalised bone loss in patients with early rheumatoid arthritis. Lancet 1994;344(8914):23–7. [38] Takashima T, Kawai K, Hirohata K, Miki A, Mizoguti H, Cooke TD. Inflammatory cell changes in haversian canals. A possible cause of osteoporosis in rheumatoid arthritis. The J Bone Jt Surg Br Vol 1989;71(4):671–6. [39] Szentpetery A, Heffernan E, Haroon M, Kilbane M, Gallagher P, McKenna MJ, et al. Striking difference of periarticular bone density change in early psoriatic arthritis and rheumatoid arthritis following anti-rheumatic treatment as measured by digital X-ray radiogrammetry. Rheumatol (Oxford) 2016;55(5):891–6. [40] Forslind K, Boonen A, Albertsson K, Hafstrom I, Svensson B, Barfot Study G. Hand bone loss measured by digital X-ray radiogrammetry is a predictor of joint damage in early rheumatoid arthritis. Scand J Rheumatol 2009;38(6):431–8. [41] Dirven L, Guler-Yuksel M, de Beus WM, Ronday HK, Speyer I, Huizinga TW, et al. Changes in hand bone mineral density and the association with the level of disease activity in patients with rheumatoid arthritis: bone mineral density measurements in a multicenter randomized clinical trial. Arthritis Care Res (Hoboken) 2011;63(12):1691–9. [42] Dhainaut A, Hoff M, Syversen U, Haugeberg G. Cortical hand bone porosity and its association with distal radius fracture in middle aged and elderly women. PLoS ONE 2013;8(7):e68405. [43] Hoff M, Haugeberg G, Odegard S, Syversen SW, Landewe R, van der Heijde D, et al. Cortical hand bone loss after one year in early rheumatoid arthritis predicts radiographic hand joint damage at 5 and 10 year follow-up. Ann Rheum Dis 2009;68 (3):324–9. [44] Guler-Yuksel M, Allaart CF, Goekoop-Ruiterman YP, de Vries-Bouwstra JK, van Groenendael JH, Mallee C, et al. Changes in hand and generalised bone mineral density in patients with recent-onset rheumatoid arthritis. Ann Rheum Dis 2009;68(3):330–6. [45] Smolen JS, Van Der Heijde DM, St Clair EW, Emery P, Bathon JM, Keystone E, et al. Predictors of joint damage in patients with early rheumatoid arthritis treated with high-dose methotrexate with or without concomitant infliximab: results from the aspire trial. Arthritis Rheum 2006;54(3):702–10. [46] Compston J. Management of glucocorticoid-induced osteoporosis. Nat Rev Rheumatol 2010;6(2):82–8. [47] Takahata M, Maher JR, Juneja SC, Inzana J, Xing L, Schwarz EM, et al. Mechanisms of bone fragility in a mouse model of glucocorticoid-treated rheumatoid arthritis: implications for insufficiency fracture risk. Arthritis Rheum 2012;64(11):3649–59. [48] van Staa TP, Geusens P, Bijlsma JW, Leufkens HG, Cooper C. Clinical assessment of the long-term risk of fracture in patients with rheumatoid arthritis. Arthritis Rheum 2006;54(10):3104–12. [49] Van Staa TP, Leufkens HG, Abenhaim L, Zhang B, Cooper C. Use of oral corticosteroids and risk of fractures. J Bone Min Res 2000;15(6):993–1000. [50] van Staa TP, Leufkens HG, Abenhaim L, Zhang B, Cooper C. Oral corticosteroids and fracture risk: relationship to daily and cumulative doses. Rheumatol (Oxford) 2000;39(12):1383–9. [51] van Staa TP, Leufkens HG, Cooper C. The epidemiology of corticosteroid-induced osteoporosis: a meta-analysis. Osteoporos Int 2002;13(10):777–87. [52] Roldan JF, Del Rincon I, Escalante A. Loss of cortical bone from the metacarpal diaphysis in patients with rheumatoid arthritis: independent effects of systemic inflammation and glucocorticoids. J Rheumatol 2006;33(3):508–16.