Anneke Spoorenberg has received research grants from Abbott, Pfizer, and Wyeth and consulting fees from Abbvie, Pfizer, and UCB. Elisabeth Brouwer and Suzanne Arends have received research grants from Abbott, Pfizer, and Wyeth. Reinhard Bos has received research grants and consulting fees from Pfizer. This does not alter the authors' adherence to PLOS ONE policies on sharing data and materials. The other authors have declared that they have no competing interests.
Conceived and designed the experiments: AS EB HB EV SA. Performed the experiments: FM AS EB RB ME RC PO RW. Analyzed the data: FM. Contributed reagents/materials/analysis tools: FM AS NV EV SA. Wrote the paper: FM AS SA. Critically revised the manuscript: AS EB RB ME RC NV PO RW HB EV SA.
To evaluate spinal radiographic damage over time and to explore the associations of radiographic progression with patient characteristics and clinical assessments including disease activity in ankylosing spondylitis (AS) patients treated with tumor necrosis factor-alpha (TNF-α) blocking therapy in daily clinical practice.
Consecutive outpatients from the Groningen Leeuwarden AS (GLAS) cohort were included based on the availability of cervical and lumbar radiographs before start of TNF-α blocking therapy and after 2, 4, and/or 6 years of follow-up. Clinical data were assessed at the same time points. Radiographs were scored by two independent readers using the modified Stoke AS Spine Score (mSASSS). Spinal radiographic progression in relation to clinical assessments was analyzed using generalized estimating equations.
176 AS patients were included, 58% had syndesmophytes at baseline. Median mSASSS increased significantly from 10.7 (IQR: 4.6–24.0) at baseline to 14.8 (IQR: 7.9–32.8) at 6 years. At the group level, spinal radiographic progression was linear with a mean progression rate of 1.3 mSASSS units per 2 years. Both spinal radiographic damage at baseline and radiographic progression were highly variable between AS patients. Male gender, older age, longer disease duration, higher BMI, longer smoking duration, high CRP, and high ASDAS were significantly associated with syndesmophytes at baseline. Significantly more radiographic progression was seen in patients with versus without syndesmophytes (2.0 vs. 0.5 mSASSS units per 2 years) and in patients >40 versus ≤40 years of age (1.8 vs. 0.7 mSASSS units per 2 years). No longitudinal associations between radiographic progression and clinical assessments were found.
This prospective longitudinal observational cohort study in daily clinical practice shows overall slow and linear spinal radiographic progression in AS patients treated with TNF-α blocking therapy. At the individual level, progression was highly variable. Patients with syndesmophytes at baseline showed a 4-fold higher radiographic progression rate than patients without syndesmophytes.
Ankylosing spondylitis (AS) is a chronic rheumatic inflammatory disorder which usually begins before the fourth decade of life. AS is characterized by inflammation in combination with new bone formation and bone loss. The disease mainly affects the axial skeleton and causes pain, stiffness, and impaired functioning of the spine. The disease course is found to be highly variable between AS patients. Excessive bone formation is an important disease outcome of AS. In the spine, this comprises the formation of syndesmophytes which may lead to complete fusion of the spine, resulting in a so-called ‘bamboo spine’. In most AS patients, it takes years from the first disease symptoms to manifestations of bone formation on radiographs [
Tumor necrosis factor-alpha (TNF-α) blocking therapy leads to a clear improvement in disease activity, functional outcome measures, and quality of life in the majority of AS patients who do not respond to conventional treatment [
These findings triggered the debate about the effect of TNF-α blocking therapy and the relationship between disease activity and spinal radiographic progression in AS. In previous cross-sectional and longitudinal studies in AS patients with a large variability in disease duration, disease activity, and treatment regimens, disease activity at baseline and over time were associated with spinal radiographic damage and progression [
The aim of this prospective longitudinal cohort study was to evaluate spinal radiographic damage over time and to explore the associations of radiographic progression with patient characteristics and clinical assessments including disease activity in AS patients treated with TNF-α blocking therapy in daily clinical practice.
The present analysis was based on data from the Groningen Leeuwarden Ankylosing Spondylitis (GLAS) cohort. GLAS is an ongoing prospective longitudinal observational cohort study in the northern part of the Netherlands. Since November 2004, this cohort included consecutive AS outpatients who started TNF-α blocking therapy at the University Medical Center Groningen (UMCG) or the Medical Center Leeuwarden (MCL) because of active disease [
The choice of the TNF-α blocking agent (infliximab, etanercept, or adalimumab) was based on the judgment of the treating rheumatologist and/or the specific preference of the patient. As described previously, the standard regimen for infliximab was 5 mg/kg intravenously at 0, 2, 6 weeks and then every 8 weeks, for etanercept 50 mg (once) or 25 mg (twice) subcutaneous injection every week, and for adalimumab 40 mg subcutaneous injection every two weeks [
Patients were clinically evaluated at baseline, after 3 months, and then every 6 months according to a fixed protocol. Disease activity was measured at each follow-up visit and treatment continuation was based on BASDAI improvement (≥50% or two units compared with baseline) and/or expert opinion. Patients were allowed to switch between different TNF-α blocking agents and to receive concomitant medication as usual in daily clinical practice. Type, dose, and frequency of TNF-α blocking therapy were recorded at all follow-up visits. Temporary stop was registered and the total duration of exposure to TNF-α blocking therapy was expressed as the percentage of follow-up time.
Patients included in the analysis started with TNF-α blocking therapy between 2004 and October 2011 and had lateral radiographs of the cervical and lumbar spine available at baseline and after at least 1 follow-up visit at 2, 4 and/or 6 years.
The GLAS cohort was approved by the local ethics committees of the MCL and the UMCG. All patients provided written informed consent according to the Declaration of Helsinki.
Baseline characteristics included: gender, age, symptom duration, time since diagnosis, HLA-B27 status, history of smoking (duration in years), and history of extra-articular manifestations.
At baseline and at each follow-up visit, disease activity was assessed with BASDAI [
Lateral radiographs of the cervical and lumbar spine were independently scored by two trained readers (FM and RC). In order to blind readers for patient characteristics and time sequence, all identifying information including exam dates were removed from the radiographs. Radiographs were scored using the modified Stoke AS Spine Score (mSASSS). The anterior corners of lower C2 until upper Th1 and lower Th12 until upper S1 were scored for the presence of erosions, sclerosis, and/or squaring (1 point per vertebral site), non-bridging syndesmophytes (2 points per site), and bridging syndesmophytes (complete bridging of vertebrae; 3 points per site). The mSASSS was calculated as the sum of the scores of all individual sites (range 0–72). Patients with complete spinal ankylosis (mSASSS of 72) at baseline were excluded since no radiographic progression could occur in these patients. If ≤3 scores of vertebral sites were missing, the scores of these sites were substituted by the mean score of the vertebrae of the corresponding spinal segment, as proposed by Wanders
Inter-observer reliability between readers for baseline mSASSS was very good, with an intraclass correlation coefficient (ICC; two-way mixed effects model, single measures, absolute agreement; before reassessment) of 0.987 (95% confidence interval (CI): 0.982–0.991). Inter-observer reliability for mSASSS change scores was moderate to good with ICC’s of 0.690 (95% CI: 0.596–0.765) for 0–2 year interval, 0.690 (95% CI: 0.545–0.794) for 2–4 year interval, and 0.400 (95% CI: 0.110–0.626) for 4–6 years interval. Bland-Altman plots revealed no systematic error. The mean difference in progression scores between the two readers was 0.1 (95% CI -3.2−3.4) for all 2-years intervals (
Presence of syndesmophytes at baseline was defined when both readers scored a non-bridging or bridging syndesmophyte (≥2 points) at one or more vertebral sites. Inter-observer reliability for presence of syndesmophytes was very good with Cohen’s kappa of 0.89 (95% CI 0.83−0.96) and absolute agreement of 95%.
Definitions of spinal radiographic progression according to Baraliakos
Results were expressed as mean ± SD or median (interquartile range (IQR)) for normally distributed and non-normally distributed data, respectively. Independent samples T-test, Mann-Whitney U test, Chi-Square test, and Fisher Exact test were used to compare differences in baseline characteristics between groups.
Generalized estimating equations (GEE) was used to analyze spinal radiographic progression over time within subjects and to calculate mean radiographic progression rate at the group level. Because correlations of spinal radiographic damage were approximately equal at different time points, the exchangeable correlation structure was used. Different models of time (linear, quadratic, cubic, square, logarithmic, and exponential) were used to investigate whether time was linear or non-linear associated with radiographic progression. In case residuals were non-normally distributed, parameters were transformed (log or square root) before entered into the equation.
In the baseline analysis, interactions between time and the following patient characteristics and baseline clinical assessments were tested: gender, age, symptom duration, time since diagnosis, HLA-B27 status, BMI, duration of smoking, NSAID use, disease activity (BASDAI, ASDAS, physician’s and patient’s GDA, CRP, ESR), and presence of syndesmophytes. If interaction effects with time were found (p-values ≤0.05), the mean radiographic progression rate was calculated after stratification into subgroups based on clinically relevant or median values.
In the longitudinal analysis, the relationship between radiographic progression and disease activity, BMI, and NSAID use over time was investigated with an autoregressive marginal time-lag model. This model investigates the influence of disease activity at the start of a 2-year interval (eg. BASDAIt), BMI at the start of a 2-year interval (BMIt), or mean cumulative NSAID use during a 2-year period (ASAS-NSAIDt−t+1) on the radiographic score at the end of a 2-year interval (mSASSSt+1), adjusted for the radiographic score at the start of this interval (mSASSSt) so radiographic progression was modeled. The following models were tested: mSASSSt+1 modeled by mSASSSt and BASDAIt, ASDASt, physician’s GDAt, patient’s GDAt, CRPt, ESRt. BMIt, and ASAS-NSAIDt−t+1,
Statistical analysis was performed with IBM SPSS Statistics 22 (SPSS, Chicago, IL, USA). P values ≤0.05 were considered statistically significant.
In total, 176 of the 267 AS patients who started with TNF-α blocking therapy between November 2004 and October 2011 were included in the analysis (
Of the 176 included patients, 69% were male, mean age was 42 ± 11 years, median symptom duration 14 years (IQR: 7–24), and 77% were HLA-B27 positive (
Baseline syndemophytes | ||||
---|---|---|---|---|
All patients (n = 176) | Present (n = 102) | Absent (n = 74) | p-value | |
121 (69) | 78 (77) | 43 (58) | ||
42.3 ± 11.1 | 46.8 ± 10.1 | 36.2 ± 9.4 | ||
14 (7–23) | 18 (10–25) | 10 (5–17) | ||
5 (1–14) | 8 (1–19) | 3 (1–11) | ||
134 (77) | 77 (76) | 57 (77) | 0.903 | |
26.4 ± 4.1 | 27.2 ± 4.1 | 25.2 ± 3.7 | ||
12 (0–22) | 16 (0–25) | 7 (0–16) | ||
130 (74) | 79 (78) | 51 (69) | 0.203 | |
50 (0–100) | 50 (0–100) | 40 (0–100) | 0.602 | |
6.1 ± 1.6 | 6.1 ± 1.5 | 6.0 ± 1.7 | 0.929 | |
BASDAI >6 | 84 (48) | 52 (51) | 32 (43) | 0.310 |
3.7 ± 0.8 | 3.8 ± 0.7 | 3.6 ± 0.8 | 0.150 | |
ASDAS >3.5 | 104 (60) | 68 (68) | 36 (49) | |
4 (3–6) | 4 (3–6) | 4 (3–6) | 0.701 | |
Physician’s GDA >6 | 34 (20) | 23 (23) | 11 (16) | 0.238 |
7 (5–8) | 7 (5–8) | 7 (6–8) | 0.280 | |
Patient’s GDA >6 | 105 (60) | 56 (55) | 49 (66) | 0.151 |
12 (4–22) | 14 (6–22) | 9 (4–21) | 0.141 | |
CRP >10 mg/L | 97 (56) | 63 (62) | 34 (47) | |
21 (10–34) | 20 (10–34) | 21 (9–34) | 0.832 | |
ESR >20 mm/hr | 87 (50) | 49 (49) | 38 (52) | 0.691 |
11 (5–24) | 21 (12–38) | 4 (2–7) |
Values are presented as number of patients (%), mean ± SD, or median (IQR).AS: ankylosing spondylitis; HLA: human leukocyte antigen; BMI: body mass index; NSAID: non-steroidal anti-inflammatory drug; ASAS: Assessment of SpondyloArthritis international Society; BASDAI: Bath AS Disease Activity Index; ASDAS: AS Disease Activity Score; GDA: global disease activity; CRP: C-reactive protein; ESR: erythrocyte sedimentation rate; mSASSS: modified Stoke AS Spine Score.
All patients had high disease activity at baseline (91% BASDAI ≥4, 99% ASDAS ≥2.1, and 68% CRP ≥6mg/L). Twenty-seven (15%) patients started with infliximab, 110 (63%) with etanercept, and 39 (22%) with adalimumab. During follow-up, 45 (26%) switched to another TNF-α blocking agent. Patients were exposed to TNF-α blocking therapy for 97% of the follow-up time (IQR: 83%- 100%).
At baseline, median mSASSS was 11 (IQR: 5–24) and 102 (58%) patients had at least one syndesmophyte according to both readers. Patients with syndesmophytes at baseline were more frequently male, older, had longer symptom and diagnosis duration, higher BMI, longer duration of smoking, and had more often very high disease activity based on ASDAS (>3.5) and CRP (>10 mg/L) (
Mean clinical follow-up time was 3.8 ± 1.8 years (range 1–7). During this period, 176, 151, 98, and 50 patients had mSASSS data available at baseline and after 2, 4, and 6 years of follow-up, respectively. Baseline characteristics were comparable in all these groups, only a significantly longer symptom duration and higher ASAS-NSAID index were seen in patients with 6 years data (
Median mSASSS increased significantly from 10.7 (IQR: 4.6–24.0) at baseline to 14.8 (IQR: 7.9–32.8) at 6 years (
Status scores | Baseline (n = 176) | 2 year (n = 151) | 4 year (n = 98) | 6 year (n = 50) |
---|---|---|---|---|
Mean mSASSS | 16.9 ± 16.7 | 17.5 ± 17.2 | 21.4 ± 18.5 | 21.4 ± 18.7 |
Median mSASSS | 10.7 (4.6–24.0) | 10.9 (4.6–25.5) | 15.0 (5.6–32.1) | 14.8 (7.9–32.8) |
Mean change mSASSS | 1.3 ± 3.2 | 1.4 ± 2.8 | 1.3 ± 2.0 | |
Median change mSASSS | 0.8 (-0.5–2.1) | 1.2 (-0.6–2.8) | 1.0 (-0.1–2.8) | |
<2 mSASSS units/2 years (slow) | 106 (70) | 44 (59) | 27 (64) | |
2–5 mSASSS units/2 years (moderate) | 27 (18) | 25 (33) | 13 (31) | |
>5 mSASSS units/2 years (fast) | 18 (12) | 6 (8) | 2 (5) |
Values are presented as number of patients (%), mean ± SD or median (IQR).
mSASSS: modified stoke AS spine score.
At the individual level, both spinal radiographic damage at baseline and radiographic progression over time were highly variable between AS patients (
The presence of syndesmophytes and older age were significantly associated with spinal radiographic progression. Patients with syndesmophytes at baseline had a 4-fold higher radiographic progression rate than patients without syndesmophytes (2.0 vs. 0.5 mSASSS units per 2 years). This increased progression rate also applies for patients with only 1 syndesmophyte compared to patients without syndesmophytes (1.8 vs. 0.5 mSASSS units per 2 years). Patients >40 years of age showed a 2.5-fold higher radiographic progression rate than patients ≤40 years (1.8 vs. 0.7 mSASSS units per 2 years) (
β (95% CI) | p-value | Intervals | n | |
---|---|---|---|---|
1.25 (1.10–1.40) | 475 | 176 | ||
0.104 | 475 | 176 | ||
475 | 176 | |||
Age ≤40 years | 0.68 (0.49–0.90) | 206 | 75 | |
Age >40 years | 1.78 (1.64–1.93) | 269 | 101 | |
0.565 | 448 | 166 | ||
0.516 | 465 | 171 | ||
0.855 | 473 | 175 | ||
0.221 | 360 | 141 | ||
0.117 | 362 | 133 | ||
0.246 | 475 | 176 | ||
0.668 | 381 | 130 | ||
0.526 | 475 | 176 | ||
0.764 | 467 | 173 | ||
0.123 | 466 | 172 | ||
0.506 | 473 | 175 | ||
0.375 | 469 | 174 | ||
0.339 | 466 | 173 | ||
475 | 176 | |||
Without syndesmophytes | 0.52 (0.39–0.66) | 203 | 74 | |
With syndesmophytes | 2.02 (1.88–2.17) | 272 | 102 |
See
Disease activity improved significantly during TNF-α blocking therapy. From baseline to 3 months, mean BASDAI improved from 6.1 to 3.2, mean ASDAS from 3.7 to 2.1, median physician’s GDA from 4 to 2, median patient’s GDA from 7 to 3, median CRP from 12 to 3 mg/L, and median ESR from 21 to 6 mm/hr (all p<0.001). These improvements remained stable during long-term follow-up (data not shown). Mean BMI showed a small increase during follow-up, from 26.4 at baseline to 26.6 at 6 years (p<0.05). NSAID use decreased significantly from 74% at baseline to 41%, 37%, and 25% at 2, 4, and 6 years, respectively (p<0.001). Mean cumulative NSAID intake according to the ASAS-NSAID index decreased from 24.3 during the first 2 years to 14.4 and 9.2 during the 2–4 and 4–6 years time intervals, respectively (p<0.001).
During TNF-α blocking therapy, no significant longitudinal associations were found between spinal radiographic progression and disease activity, BMI, or NSAID use over time (
β (95% CI) | p-value | Intervals | n | |
---|---|---|---|---|
1.02 (1.00–1.04) | 268 | 159 | ||
-0.02 (-0.15–0.10) | 0.737 | 259 | 159 | |
0.24 (-0.08–0.55) | 0.143 | 255 | 158 | |
-0.05 (-0.19–0.09) | 0.497 | 264 | 156 | |
-0.03 (-0.16–0.10) | 0.637 | 267 | 159 | |
0.03 (0.00–0.06) | 0.071 | 255 | 158 | |
0.03 (0.00–0.06) | 0.056 | 252 | 158 | |
-0.02 (-0.09–0.05) | 0.596 | 266 | 181 | |
0.00 (-0.01–0.01) | 0.905 | 196 | 87 |
See
This observational longitudinal cohort study prospectively investigated spinal radiographic damage over time and the associations of radiographic progression with patient characteristics and clinical assessments including disease activity in 176 AS patients treated with TNF-α blocking therapy in daily clinical practice. Spinal radiographic progression was linear at the group level with a mean progression rate of 1.3 mSASSS units per 2 years. This indicates that, on average, AS patients treated with TNF-α blocking therapy showed slow radiographic progression according to the definitions of Baraliakos
Before start of TNF-α blocking therapy, all AS patients had high disease activity and more than half (58%) had syndesmophytes, which is comparable to the results of previous studies. In these previous studies, the proportion of patients with syndesmophytes at baseline varied from 30% in ‘early’ AS (≤10 years symptom duration) [
At the individual level, spinal radiographic progression was highly variable. The mean mSASSS progression rate was 4-fold higher in patients with syndesmophytes at baseline and 2.5-fold higher in patients >40 years of age. These findings indicate that radiographic progression during the treatment of TNF-α blocking therapy mainly occurs in older AS patients and in patients with more advanced disease. Previous studies with different treatment regimens also identified the presence of syndesmophytes at baseline as the most important predictor for the development of more radiographic damage in both ‘early’ axial SpA [
In our analyses, none of the disease activity assessments at baseline and over time were significantly associated with spinal radiographic progression. This is probably due to the low variability in disease activity since all patients had high disease activity before start of TNF-α blocking therapy and stable low disease activity during treatment. Moreover, the mean change in mSASSS at the group level was small, which makes it difficult to observe significant differences. This was also confirmed by the low observed regression coefficients of the time-lagged autoregressive GEE models. Historical longitudinal observational cohort studies in AS patients that have found significant relationships between disease activity and radiographic progression included patients with a high variability in disease activity status and treatment regimens [
Previous studies in AS reported a positive effect of continuous use of NSAIDs on the reduction of radiographic progression [
In the present study the reading of the radiographs was done without known time sequence which may lead to negative and smaller progression rates than when the reading was done in chronological time order [
This large prospective observational cohort study in AS patients treated with TNF-α blocking therapy in daily clinical practice showed that spinal radiographic progression was overall slow and linear at the group level. At the individual level, radiographic progression was highly variable. Patients with syndesmophytes at baseline had a 4-fold increased radiographic progression rate and patients >40 years of age had a 2.5-fold increased radiographic progression rate. No longitudinal associations between radiographic progression and clinical assessments were found. A direct longitudinal comparison between cohorts with long-term follow-up and large study populations of AS patients treated with and without long-term TNF-α blocking therapy is required to evaluate the effect of this treatment and to further investigate the relationship between clinical assessment (e.g. disease activity) and spinal radiographic progression.
(TIF)
(PDF)
The authors would like to thank all patients who participated in the GLAS cohort. Furthermore, the authors wish to acknowledge Dr. P.M. Houtman, Mrs. W. Gerlofs, Mrs. S. Katerbarg, Mrs. A. Krol, and Mrs. R. Rumph for their contribution to clinical data collection.