Disease activity is associated with spinal radiographic progression in axial spondyloarthritis independently of exposure to tumour necrosis factor inhibitors

ASDAS is associated with spinal radiographic pro- gression in axial spondyloarthritis, independently of treatment with TNF inhibitors. Sir, The ASDAS is the clinical outcome measure that best captures disease activity in axial SpA (axSpA) [1]. The relationship between disease activity, interventions used to control it, and damage accrual in axSpA is not yet fully understood [2]. Available evidence stemming from the Outcome in Ankylosing Spondylitis International Study (OASIS) supports a longitudinal association between ASDAS and spinal radiographic progression [3]. However, patients with axSpA included in OASIS do not represent current clinical practice, as only a minority was treated (and only late in the study follow-up) with the now widely prescribed TNF inhibitors (TNFis). We aimed to test the predictive association between disease activity and the subsequent 2-year spinal radiographic progression in axSpA in a contemporary real-life cohort, including patients treated with TNFi. Patients with the clinical diagnosis of axSpA fulfilling the modified New York classification criteria (i.e. with radiographic axSpA; r-axSpA) were included from 2008 onwards in this prospective, observational cohort study (ALBERTA FORCAST). Clinical and imaging data were collected at baseline and every 2 years for up to 10 years of follow-up. Radiographs of the spine were independently scored by two trained central readers with known chronology but blinded to clinical data, using the modified Stoke Ankylosing Spondylitis Spine Score (mSASSS). One adjudicator scored all films of each patient with a discrepancy in the change of mSASSS of ≥5 units between the two readers in ≥1 interval. To be included, patients had to have ≥1 interval with data on mSASSS from ≥1 reader and complete data on ASDAS and TNFi exposure at the start of the interval. First, we tested whether ASDAS at baseline was associated with mSASSS progression (time and ASDAS interaction) in a multilevel generalized estimating equations model with individual mSASSS scores per reader as outcome and adjusting for correlation within reader. If the interaction was significant (P < 0.15), a model with time as the independent variable and mSASSS as outcome was fit per each ASDAS disease activity state [4]. Second, we tested the association between ASDAS at the start of the interval and mSASSS 2 years later in two types of longitudinal generalized estimating equation models: In ‘model 1’, we used individual mSASSS scores per reader as outcome, as explained above; in ‘model 2’, we used as outcome the average of the scores either between the two readers or between the adjudicator’s score and the closest score of the two readers. Both models were adjusted for mSASSS at the start of the interval (autoregression), which allows for a truly longitudinal interpretation of the data, and also for potential confounders selected on clinical grounds (Fig. 1). In addition, treatment with conventional synthetic DMARDs, treatment with NSAIDs, smoking, number of conventional synthetic DMARDs and NSAIDs before inclusion were tested in univariable models; if significant (P < 0.20), the variable was added to the multivariable model and finally selected if proved significant (P<0.05) or to confound the association of interest. Longitudinal effect of ASDAS at the start of the interval on mSASSS 2 years later Multilevel, multivariable linear GEE model (N = 313), also adjusted for mSASSS at the start of the interval (autoregression), number of TNFi before baseline and treatment with TNFi at the start of the interval (yes vs no). Treatment with csDMARDs or NSAIDs, smoking (all binary and time-varying) and previous treatment with csDMARDs or NSAIDs (both continuous and time-fixed) not selected in the final multivariable model. ASDAS: ankylosing spondylitis disease activity score; mSASSS: modified Stoke Ankylosing Spondylitis Spine Score; GEE: generalized estimating equations; TNFi: TNF inhibitors; csDMARDs: conventional synthetic DMARDs. In total, 314 patients (442 2-year intervals) were included. The majority were male (74%), had long symptom duration [mean 18 (2 s.d.) years] and were HLA-B27 positive (83%). At baseline, the mean ASDAS was 2.7 (1.3 s.d.), the mean mSASSS was 13.8 (18.9 s.d.) and 7% had previously been treated with ≥1 TNFi. During follow-up, 213 (68%) patients received a TNFi in ≥1 visit. The average progression was 1.33 (2.68 s.d.) mSASSS-units per 2-year interval. The interaction between ASDAS and time was significant (P = 0.006): the 2-year progression of mSASSS over time increased with higher levels of baseline ASDAS: inactive disease [β (95% CI): 0.35 (0.21, 0.50)]; low disease activity [β (95% CI): 0.45 (0.34, 0.56)]; high disease activity [β (95% CI): 0.56 (0.44, 0.68)]; very high disease activity [β (95% CI): 0.72 (0.52, 0.93)]. We also found a longitudinal association between ASDAS and mSASSS. In model 1, one ASDAS-unit increase at the start of the interval was associated with an increase of 0.25 mSASSS units 2 years later (Fig. 1). Results were similar with model 2 [β (95% CI): 0.25 (0.08, 0.43)]. These data add to the literature by showing that higher ASDAS was longitudinally associated with higher spinal radiographic progression in patients with r-axSpA independently of treatment with TNFi. In OASIS, the strength of the association was somewhat higher: on average, one unit increase in ASDAS was associated with 0.72 units increase in mSASSS 2 years later. Whether or not such difference is (partially) explained by differences in treatment between cohorts cannot be ascertained. Differences in patient characteristics, scoring methods and intangible residual confounding likely play a role. Similar findings were reported in a more recent study, from the Swiss Clinical Quality Management cohort: one unit increase in ASDAS increased by 41% the likelihood of progression of ≥2 mSASSS units 2 years later in axSpA patients treated with TNFi [5]. Enough evidence has now accumulated to convincingly tell us that the association between ASDAS and spinal radiographic progression in axSpA is real. However, it is also clear that not all radiographic progression is explained by clinical measures of disease activity. Here we found significant progression of mSASSS even in patients with inactive disease, which is similar to what has been previously seen in OASIS. This could reflect inflammation that may not be captured by the ASDAS, such as that observed on MRI, or development of new bone unrelated to inflammation. The authors sincerely thank the staff and all patients of the ALBERTA FORCAST study. All authors participated in designing the study. A.S. performed the statistical analyses. A.S. wrote the first draft of the manuscript. All authors critically interpreted the results, reviewed the draft version and approved the final manuscript. A.S. is supported by a doctoral grant from ‘Fundação para a Ciência e Tecnologia’ (SFRH/BD/108246/2015). Data may be obtained from a third party and are not publicly available. Funding: FORCAST is financially supported by an unrestricted grant from Abbvie. Disclosure statement: The authors have declared no conflicts of interest.