Objectives To investigate construct validity and responsiveness of the novel ankylosing spondylitis (AS) disease activity score (ASDAS) in patients with spondyloarthritis (SpA).
Methods In a 46-week prospective longitudinal multicentre study of 60 patients with SpA (80% men, median age 40 years (range 21–62)) treated with tumour necrosis factor α (TNFα) inhibitors (infliximab, n=41; etanercept, n=13; adalimumab, n=6), the responsiveness of ASDAS, conventional clinical measures of disease activity and treatment response and the Berlin MRI sacroiliac joint (SIJ) and lumbar spine inflammation scores were compared.
Results After 22 weeks, 58.3% of the patients were clinical responders (50% or 20 mm reduction in the Bath AS Disease Activity Index (BASDAI)). At baseline, clinical responders had significantly higher median (range) ASDAS than non-responders (4.15 (1.98–6.04) vs 2.99 (2.05–6.19), p=0.008). Changes in ASDAS correlated with changes in clinical measures of disease activity (including BASDAI (ρ=0.76) and C-reactive protein (CRP) (0.79)), MRI SIJ inflammation (0.46) and MRI total inflammation scores (0.34). Patients with higher BASDAI or Assessment of SpondyloArthritis International Society (ASAS) responses obtained more profound reductions in ASDAS. ASDAS had the highest responsiveness with an effect size of 2.04 and a standardised response mean of 1.45, whereas BASDAI (effect size 1.86; standardised response mean 1.36) and CRP (effect size 0.63; standardised response mean 0.70) were less responsive. Linear regression showed that a change in BASDAI of 20 mm or 50% corresponded to a change in ASDAS of 1.38 and 1.95, respectively.
Conclusion ASDAS demonstrates construct validity and high responsiveness during treatment with TNFα inhibitors in patients with SpA. The proposed thresholds for disease activity and treatment response need further validation.
Trial registration number NCT00133315.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
In patients with spondyloarthritis (SpA) including ankylosing spondylitis (AS), it is difficult to monitor changes in disease activity owing to the lack of sensitive and specific objective clinical measures. In clinical practice, subjective measures of disease activity such as patient's global assessment, pain, morning stiffness and the Bath AS Disease Activity Index (BASDAI)1 are used, but these measures have the disadvantage of covering the patient perspective exclusively and are not all sensitive to change (responsive).2 3 Laboratory tests and imaging are other methods to assess disease activity. Serum C-reactive protein (CRP) is often used but it has low sensitivity.4 Conventional radiography can only reliably detect disease progression over years.5 MRI is a promising new tool as it allows visualisation of inflammatory changes in the sacroiliac joints (SIJ) and spine, but feasibility issues reduce the possibility for routine monitoring.6 7 Consequently, there is a need for improved and feasible measures of disease activity and treatment response in patients with SpA. Recently, a Working Group of the SpondyloArthritis International Society (ASAS) has proposed a composite disease activity score, the AS Disease Activity Score (ASDAS), for patients with AS.8 9
The aim of this study was to investigate construct validity and responsiveness of ASDAS in comparison with conventional clinical measures of disease activity and treatment response and experimental imaging measures of disease activity in a cohort of patients with axial SpA treated with tumour necrosis factor α (TNFα) inhibitors.
The BIOSPA study was a prospective investigator-initiated longitudinal observational multicentre study that included 60 TNFα inhibitor-naive patients with SpA (48 men and 12 women aged 21–62 years) starting TNFα inhibitor therapy. Patients were included from nine Danish departments of rheumatology from September 2004 to March 2006 and were followed for 46 weeks. MRIs and x-rays were performed at three departments of radiology.
Patients were eligible for the study if they (1) fulfilled the European Spondyloarthropathy Study Group (ESSG) classification criteria for SpA10; (2) had sacroiliitis (on x-rays according to the modified New York criteria11) or MRI (inflammatory or chronic changes12) evaluated by the same radiologist (AGJ); (3) had a BASDAI >30 mm despite treatment with non-steroidal anti-inflammatory drugs; and (4) had clinical indication for TNFα inhibitor therapy. Prednisolone and disease-modifying antirheumatic drugs other than methotrexate were not allowed during the study.
Clinical and biochemical examinations
The patients were assessed clinically and biochemically at inclusion and after 22 weeks of treatment by the use of different instruments, as specified in table 1.
Measures of disease activity
Treatment response was defined as a reduction in BASDAI of 50% or 20 mm at week 22.13 In addition, ASAS20, ASAS40 and ASAS5/6 responders were identified.14 15 ASDAS is a score of disease activity comprising three items from BASDAI: (1) back pain (question 2); (2) peripheral pain/swelling (question 3); and (3) duration of morning stiffness (question 6), as well as patient's global assessment and CRP. ASDAS was calculated using the formula: 0.121 × back pain + 0.058 × duration of morning stiffness + 0.110 × patient's global assessment + 0.073 × peripheral pain/swelling + 0.579 × Ln(CRP+1).8 9 Before calculation of ASDAS, clinical data were re-scaled from 0–100 mm to 0–10 cm. ASDAS <1.9 has been proposed as low disease activity, ASDAS ≥1.90 but ≤4.50 as moderate disease activity and ASDAS >4.5 as high disease activity, whereas a change in ASDAS (ΔASDAS) of <0.40 has been proposed as no improvement, ΔASDAS ≥0.40 but ≤1.85 as moderate improvement and ΔASDAS >1.85 as considerable improvement.9
MRIs of the SIJ and lower spine (Th9–S1) were performed on 1.5T systems using appropriate coils. Spine images included sagittal T1-weighted turbo spin echo (TSE) and short τ inversion recovery (STIR) sequences. SIJ images included semicoronal T1-weighted TSE sequences before and after intravenous injection of 0.1 mmol/kg body weight of the gadolinium-containing contrast agent gadodiamide (Omniscan; GE Healthcare, UK) and, from March 2006, gadoteric acid (Dotarem; Guerbet, France). Post-contrast images were obtained with fat saturation. MRIs were anonymised and evaluated blind to chronology and clinical data. A musculoskeletal radiologist experienced in MRI evaluation of SpA (KGH) evaluated the MRIs according to the Berlin MRI spine and SIJ scoring methods (see below).
Berlin MRI spine and SIJ scoring method
The Berlin MRI spine inflammation score graduates the volume of bone marrow oedema (BME) in each discovertebral unit (DVU) into no inflammation (grade 0), >0% but ≤25% (grade 1), >25% but ≤50% (grade 2), or >50% (grade 3).16 The spine score is defined for MRI images of the whole spine (23 DVUs) with a range of 0–69. Thus, in this study a limited version of the spine score was applied to L1–S1 (5 DVUs; score range 0–15). In the Berlin MRI SIJ inflammation score, each SIJ is divided into four quadrants which are scored individually into no inflammation (grade 0), enhancement in joint space, joint capsule or in erosions, or volume of BME <10% (grade 1), BME ≥10% but <33% (grade 2), BME ≥33% but <66% (grade 3) and BME ≥66% (grade 4) (total score range 0–32).17 A total score for inflammation was calculated by adding up SIJ and lumbar spine inflammation scores. In the Berlin MRI SIJ chronicity score, chronic changes of each SIJ are graded semiquantitatively (grade 0–4) based on the overall impression of the changes in each joint (total score range 0–8).17
Comparisons between groups were made with the χ2 test and Fisher exact test for dichotomous variables and the Mann–Whitney test for continuous variables. Associations were evaluated with Spearman rank correlation coefficients. Linear regression modelling was used to estimate the reduction in ASDAS corresponding to a reduction in BASDAI of 20 mm or 50% for patients with baseline BASDAI ≥40 mm. Graphically, residual plots were screened for linearity, variance homogeneity and normal distribution. Effect size (ES) was calculated as the difference between the mean baseline scores and follow-up scores at week 22 divided by the SD of the baseline scores, and standardised response mean (SRM) as the difference between the mean baseline scores and follow-up scores at week 22 divided by the SD of the change scores.18 Values of <0.20, ≥0.20 to <0.50, ≥0.50 to <0.80 and ≥0.80 were considered to represent no, small, moderate and large responsiveness, respectively.18 Before the analysis of responsiveness, histograms of baseline data were examined for the presence of floor effects. p<0.05 was considered as statistically significant. The statistical analyses were done using SPSS 12.0 (SPSS Inc, Chicago, Illinois, USA).
Eighty-two patients were screened for the study and 60 met the inclusion criteria (figure 1). All patients fulfilled the new ASAS criteria for axial SpA.19 20 Patients were treated with infliximab (3 mg/kg (n=30) or 5 mg/kg (n=11) intravenously at baseline, weeks 2 and 6 and then every sixth to eighth week), etanercept (25 mg subcutaneously ×2/week (n=12) or 50 mg weekly (n=1)) or adalimumab (40 mg subcutaneously every other week (n=6)).
ASDAS at baseline
Stratified according to ASDAS disease activity, no patients had low disease activity and 15 patients (25%) had high ASDAS disease activity (table 1). Patients with high ASDAS disease activity had significantly higher baseline BASDAI, BASFI, pain, patient's and physician's global assessment and CRP than those with moderate ASDAS, except for Berlin MRI lumbar spine inflammation score and the total inflammation score.
At baseline, patients with BASDAI ≥40 mm or elevated CRP (>8 mg/l) had statistically significant higher ASDAS scores than patients with BASDAI <40 mm or normal CRP, respectively (table 2). ASDAS did not differ between patients with/without peripheral arthritis or with/without psoriatic arthritis. However, patients with elevated CRP or peripheral arthritis had significantly more profound reductions in ASDAS after 22 weeks of treatment with a TNFα inhibitor. Patients without inflammation on MRI (total score) had significantly higher ASDAS scores than patients with inflammation, but they did not differ in disease duration (p=0.2) or in the presence/absence of swollen joints (p=0.6).
At baseline, ASDAS correlated with BASDAI, CRP, pain and patient's and physician's global assessment (ρ=0.63–0.75), BASFI and fatigue (ρ=0.41–0.46) and with lumbar spine inflammation scores (ρ=−0.30) (table 3). Changes in ASDAS from baseline to week 22 correlated with changes in the abovementioned conventional clinical and biochemical measures (ρ=0.61−0.79) and BASMI (ρ=0.48) and with changes in the MRI SIJ and total inflammation scores (ρ=0.46 and 0.34).
ASDAS and treatment response
After 22 weeks of treatment with TNFα inhibitors, 35 patients (58.3%) achieved the definition of a clinical response (ie, a BASDAI reduction of 50% or 20 mm). Of the 18 non-responders, five patients changed dose, five changed TNFα inhibitor and eight continued treatment without any changes.
Clinical responders had significantly higher ASDAS scores at baseline and more profound reductions in ASDAS than non-responders (table 2). Of the responders, 7% had no improvement in ASDAS and 52% improved considerably. Patients achieving higher BASDAI or ASAS responses at weeks 22 and 46 more often showed considerable improvement in ASDAS and they had more profound reductions in ASDAS (table 4).
At week 22, responder rates according to different response criteria in the intention-to-treat (ITT) population were BASDAI20 80.0%, moderate/considerable improvement in ASDAS 73.3%, ASAS20 68.3%, ASAS5/6 58.3%, BASDAI50 51.7%, ASAS40 48.3%, considerable improvement in ASDAS 40.0% and BASDAI70 35.0%.
Linear regression showed that a change in BASDAI of 20 mm corresponded to a change in ASDAS of 1.38, p<0.001 (y=0.38+0.048x, R2=0.75) and a 50% change in BASDAI corresponded to a change in ASDAS of 1.95, p<0.001 (y=0.35+0.032x, R2=0.67). With a threshold of response/non-response at ASDAS 1.38 in this study, 29 patients (48.3%) of the ITT population would be classified as responders and 18 patients (30%) as non-responders.
ASDAS was the most responsive clinical measure of disease activity from baseline to week 22 with an ES of 2.04 and SRM of 1.45 (table 5). BASDAI had an ES of 1.86 and a SRM of 1.36, and CRP had an ES of 0.63 and a SRM of 0.70. All conventional subjective measures of disease activity showed large responsiveness. Objective clinical measures of disease activity showed low to moderate responsiveness.
In this study we have investigated the responsiveness of the novel ASDAS. Compared with conventional clinical measures of disease activity, ASDAS showed the highest responsiveness during treatment with anti-TNFα therapy (ES 2.04, SRM 1.45). CRP was less responsive than ASDAS (ES 0.63, SRM 0.70). The conventional clinical measures also demonstrated high responsiveness, and in the same ranges as previously reported.3
Re-calculation of data from one of the ASDAS validation studies9 revealed lower SRMs for ASDAS (1.46) and fatigue (0.89) in patients treated for 3 months with TNFα inhibitors, whereas the SRMs for physician's global assessment (1.48), BASDAI (1.29) and CRP (0.72) were at the same level as in our study.
This longitudinal study validating ASDAS was performed independently from the ASDAS Working Group. In accordance with the expectations of markers of disease activity exposed to effective anti-inflammatory treatment, ASDAS decreased when TNFα inhibitor was given, was reduced more in responders than in non-responders, and changes in ASDAS correlated with changes in conventional measures of disease activity. This supports the construct validity of ASDAS and the notion that ASDAS in patients with SpA reflects different levels of and changes in disease activity.
If patients are considered candidates for TNFα inhibitor therapy based on the proposed levels of high ASDAS disease activity (ASDAS >4.50), our study shows that the target group will be reduced considerably compared with the BASDAI >40 mm criterion commonly used.25 Furthermore, the proposed level of a considerable ASDAS response (ΔASDAS >1.85) reduces the percentage of responders (40.0% compared with 58.3% being BASDAI responders (50% or 20 mm response)). The proportion of ASAS20 and BASDAI50 responders in our study did not differ from previous reports.26,–,29 This suggests that the threshold for a considerable ASDAS response may be set too high and may benefit from adjustment. Further studies are needed to establish which ASDAS level is appropriate as a threshold for initiation of TNFα inhibitor therapy and treatment response.
The correlations between ASDAS, BASDAI and CRP were similar to the correlations of ASDAS with data from the ISSAS and OASIS study.8 However, the correlation between ASDAS and physician's global assessment was higher in our study and the correlations with BASFI and swollen joint counts were lower, which may be due to differences in the patient cohorts. Interestingly, BASDAI correlated much better with BASFI than did ASDAS, possibly reflecting the fact that both scores are entirely patient-reported outcomes.
The changes in ASDAS correlated, although relatively weakly, with changes in the Berlin MRI SIJ and total inflammation scores. This is in agreement with a previous study which reported low correlations between the Berlin MRI SIJ and total spine inflammation scores and conventional clinical measures of disease activity.30 Spinal inflammation on MRI is known to be most frequent in the thoracic spine.31 In this study, MRI scans of the cervical and thoracic spine were not performed, and patients without inflammation in SIJs and lumbar spine may have had inflammation above these spinal levels which is often seen in patients with longstanding disease. This may contribute to the inverse relation between MRI inflammation and ASDAS disease activity. The low responsiveness of the Berlin MRI inflammation scores may be explained by the lack of thoracic and cervical MRI and a slower resolution of BME than 22 weeks. The latter is supported by an increased responsiveness at 46 weeks of all MRI inflammation scores and especially the SIJ score (ES/SMR: SIJ score, 0.58/0.64; lumbar score, 0.53/0.69; total score, 0.67/0.72).
The Outcome Measures in Rheumatoid Arthritis Clinical Trials (OMERACT) group requires clinical measures of disease activity to be true, discriminative and feasible in clinical practice.32 In our study ASDAS was more responsive to changes in disease activity than conventional measures, and it demonstrated construct validity.33 It is more complex to calculate ASDAS than BASDAI, and CRP has to be available. Nevertheless, the algorithm is no more complex than for DAS28, which is widely accepted and is the most used disease activity score for patients with rheumatoid arthritis.34
In conclusion, this longitudinal study of patients with SpA starting TNFα inhibitor therapy shows that the recently developed composite disease activity score ASDAS is more responsive than the conventional clinical, biochemical and experimental imaging measures of disease activity. The thresholds for disease activity and treatment response proposed by the ASAS working group may benefit from further validation in larger studies and potential adjustments. Overall, the present study indicates that ASDAS may be a very useful tool for monitoring SpA in clinical trials and clinical practice.
The authors thank Abbott Denmark for assistance with printing the case report forms used in the study.
Funding SJP's salary during her position as a PhD student was financed by the Faculty of Health Sciences, University of Copenhagen, Denmark. Abbott Denmark printed and distributed the case report forms used in the study.
Competing interests None.
Ethics approval The study was approved by the regional scientific ethical committee (reference number H-KF-02–050/04) and conducted in accordance with the Helsinki II declaration.
Provenance and peer review Not commissioned; externally peer reviewed.