Patients
Patients with axSpA from the GErman SPondyloarthritis Inception Cohort (GESPIC) were included in the current analysis based on the availability of clinical data and radiographs at baseline and year 2. A detailed description of the entire cohort as well as of the radiographic subset has been reported previously [10,11,12]. Briefly, patients included in GESPIC had a definite clinical diagnosis of axSpA according to the treating rheumatologist with symptom duration of up to 5 years for the non-radiographic form and up to 10 years for the radiographic form of axSpA. The final classification as nr-axSpA or r-axSpA was performed based on the central reading of the radiographs as previously reported [10]. Treatment of the patients was defined by local rheumatologists and followed national and international guidelines; however, since the cohort had started prior to introduction of tumor necrosis factor inhibitors (TNFi) in the clinical practice, the vast majority of the patients were off TNFi therapy at inclusion and during first 2 years of observation.
Radiographic assessments
Radiographs of cervical (lateral view) and lumbar spine (lateral and AP views) were obtained at baseline and after 2 years. Images were centrally collected, digitized, anonymized, and scored independently by two trained readers (DP and HH), who were blinded to the chronological order of the images and to all clinical data. According to the mSASSS, anterior corners of the vertebral bodies from lower C2 to upper T1 (cervical spine) and from lower T12 to upper S1 (lumbar spine) were scored as follows: 0 = normal, 1 = erosion, sclerosis, and/or squaring, 2 = non-bridging syndesmophyte, and 3 = bridging syndesmophyte, giving a range for the entire score from 0 to 72 [5]. No adjudication was performed. The final mSASSS score included in the analysis was calculated as a mean of the mSASSS scores of both readers. In addition, left and right, upper and lower vertebral corners of vertebral bodies visible on lumbar AP radiographs from lower Th12 to upper S1 were scored according the same grading system. The combination of the scores from both lateral cervical and lumbar views and lumbar AP views composed the “extended mSASSS” with a total range from 0 to 144. The final extended mSASSS was calculated as a mean of the scores of both readers. Syndesmophytes were considered to be present if both readers gave a score of ≥ 2 to a vertebral corner.
Missing scores of single vertebral corners were substituted with scores for the respective vertebral corners at other time point, if available, or with a score of zero if scores for both time points were missing.
Data analysis
To assess the applicability of the extended mSASSS as a tool for the measurement of radiographic spinal progression, it was compared to the conventional mSASSS according to the three aspects of the Outcome Measures in Rheumatology (OMERACT) filter: feasibility, discrimination, and truth [13].
Feasibility
The feasibility aspect of the OMERACT filter refers to the pragmatic reality on applying or using a specific measure; which is decisive on its success: “Can the measure be applied easily, given constraints of time, money, and interpretability?” [13]. The feasibility was assessed in the context of time, radiation exposure, and cost excess associated with extended mSASSS compared to the conventional one.
Discrimination
The discrimination aspect addresses the question: “Does the measure discriminate between situations of interest?” This aspect of the OMERACT filter focuses on reliability and sensitivity to change [13].
We used the intraclass correlation coefficient (ICC) and Cohen’s kappa coefficient to measure the inter-observer reliability for continuous and categorical variables, respectively. The analysis was performed for the overall scores as well as separately for the components of the scores (cervical spine lateral view, lumbar spine lateral view, lumbar spine AP view). Additionally, a Bland and Altman plot was drawn and the smallest detectable change (SDC) was determined to assess the sensitivity to change of both methods (mSASSS and extended mSASSS). The SDC expresses the smallest change in scores that can be detected without measurement error, and it was calculated as follows:
SDC = 1.96 × SEM,
where SEM denotes the standard error of measurement of the change score obtained from a two-way analysis of variance by taking the square root of the error variance [14].
To measure the sensitivity to change, we used the following definitions of progression: (1) change of the absolute score (calculated as means of the scores of both readers), (2) change of the score by ≥ 2 points after 2 years, (3) development of at least one new syndesmophyte (score of 0 or 1 at baseline and score of 2 or 3 after 2 years at the same vertebral corner in the opinion of both readers), and (4) development of new syndesmophytes (as described above) or growth of existing syndesmophytes (score of 2 at baseline and 3 after 2 years at the same vertebral corner) in the opinion of both readers. The sensitivity to change was evaluated in the entire group and in r-axSpA and nr-axSpA subgroups. We also analyzed both status and progression scores in the components of the mSASSS and the extended mSASSS (cervical spine lateral view, lumbar spine lateral view, lumbar spine AP view). A variance component analysis was performed.
Truth
The truth aspect addresses the question: “Is the measure truthful, does it measure what is intended? Is the result unbiased and relevant?” [13]. Earlier studies showed an association between the mSASSS and spinal mobility/functional status in axSpA [2,3,4]. With a linear regression analysis, we investigated the relationship between both scores and spinal mobility (assessed by the Bath Ankylosing Spondylitis Metrology Index—BASMI, an original two-step definition) and function (assessed by the Bath Ankylosing Spondylitis Functional Index—BASFI).
Statistical analysis was performed with IBM SPSS Statistics version 24 (IBM, Armonk, NY, USA) and SAS version 9.4 (SAS Institute Inc., Cary, NC, USA).