Abstract
Objective To investigate the association between spinal damage and functional capacity in patients with axial spondyloarthritis (axSpA) and to compare the performance of 2 radiographic scores (modified Stoke Ankylosing Spondylitis Spine Score [mSASSS] and Combined Ankylosing Spondylitis Spine Score [CASSS]).
Methods Radiographs from 101 patients with axSpA were scored for cervical facet joints (CFJ) and mSASSS for vertebral bodies. CASSS was calculated as the sum of both scores. Physical function was assessed by Bath Ankylosing Spondylitis Functional Index (BASFI); disease activity by Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) and Ankylosing Spondylitis Disease Activity Score (ASDAS); mobility by Bath Ankylosing Spondylitis Metrology Index (BASMI); and quality of life by Ankylosing Spondylitis Quality of Life (ASQOL). Univariate and multivariate analyses were performed to investigate the association between possible explanatory variables and outcomes.
Results BASFI correlated strongly with ASQOL (Spearman ρ 0.66) and BASDAI (ρ 0.70), moderately with BASMI (ρ 0.46) and ASDAS (ρ 0.59), and weakly with mSASSS (ρ 0.29) and CASSS (ρ 0.28). A best-fit multivariate model for BASFI, adjusted for symptom duration, age, sex, and smoking status, included BASDAI (B 0.76, P < 0.001), BASMI (B 0.62, P < 0.001), and history of total hip arthroplasty (B 1.22, P = 0.05). Radiographic scores were predictors of BASFI only when BASMI was removed from the model (mSASSS: B 0.03, P = 0.01; CASSS: B 0.02, P = 0.01).
Conclusion Spinal damage was independently associated with physical function in axSpA, but to a lesser extent than disease activity and mobility. Moreover, incorporating CFJ assessment in the mSASSS did not improve the ability to predict function.
Axial spondyloarthritis (axSpA) is characterized by inflammatory involvement of the sacroiliac joints and spine, which can result in new bone formation with syndesmophytes and spinal ankylosis. The modified Stoke Ankylosing Spondylitis Spine Score (mSASSS) is the most validated and widely used method for assessing radiographic damage in axSpA. However, the association between structural damage of the spine on mSASSS and clinical outcomes, such as physical function, remains controversial. Previous studies have demonstrated a weak-to-moderate correlation between structural damage and functional capacity,1-5 but others have been unable to confirm this association.6,7
In addition to radiographic damage seen at the anterior elements of the vertebrae as graded by the mSASSS, cervical facet joints (CFJ) are often affected in axSpA with subsequent loss of spinal mobility, especially cervical rotation. Further, facet joint damage often does not correlate well with vertebral body lesions, such as syndesmophyte formation.4,8,9 Maas et al10 proposed a new radiographic score, the Combined Ankylosing Spondylitis Spine Score (CASSS), which combines mSASSS with the examination of CFJ. They reported that CASSS detected more patients with definite spinal damage and definite progression during follow-up than mSASSS alone, and that the increase of mSASSS correlated weakly with the worsening of the CFJ score.10
It is unclear if the heterogeneous and usually slow formation of syndesmophytes is a good predictor of clinical outcomes in axSpA. Moreover, the usefulness of the CASSS in the assessment of structural damage and progression in axSpA has not been properly addressed. The present study aimed to investigate the association between structural damage of the spine (assessed by 2 different radiographic scores [mSASSS and CASSS]) and clinical health outcomes (mainly functional capacity), and to evaluate how much CASSS outperforms mSASSS in identifying spinal damage.
METHODS
Patients. This was a cross-sectional study that included consecutive patients with axSpA attending the Rheumatology Unit of Hospital das Clínicas da Universidade Federal de Minas Gerais (UFMG) in Brazil, from January 2019 to December 2021. All participants were aged ≥ 18 years and fulfilled the Assessment of SpondyloArthritis international Society (ASAS) classification criteria for axSpA.11 Patients were excluded from the study if they had another inflammatory rheumatic disease. Female patients were excluded if they were pregnant or lactating. The study protocol was approved by the local ethics committee (CAAE: 43536921.6.0000.5149) and written informed consent was obtained from all patients.
Clinical and laboratory assessment. Disease-related and demographic variables, including sex, age, HLA-B27 status, symptom duration, time since diagnosis, smoking status (absent or current/past smoking), current treatments, and history of arthritis, uveitis, inflammatory bowel disease, psoriasis, and total hip arthroplasty (THA) were collected from medical records. Disease activity was assessed by the Bath Ankylosing Spondylitis Disease Activity Index (BASDAI; 0-10 numerical rating scale [NRS])12 and by the Ankylosing Spondylitis Disease Activity Score (ASDAS)13 based on C-reactive protein. The Bath Ankylosing Spondylitis Functional Index (BASFI; 0-10 NRS)14 was used to evaluate functional capacity. Quality of life (QOL) was assessed by the Ankylosing Spondylitis Quality of Life (ASQOL; 18-item questionnaire).15 Finally, spinal mobility was assessed by the Bath Ankylosing Spondylitis Metrology Index (BASMI; 0-10 points) using the 11-step calculation method.16 For all scores, higher results represent worse clinical outcomes.
Radiographic assessment. Lateral radiographs of the cervical and lumbar spine of all patients were obtained. The images were independently scored by 3 trained readers, which included 2 rheumatologists (GGR and RdCL) and 1 radiologist (WCTJ), all blinded to the patient’s characteristics. Anterior corners of the vertebrae were scored using the mSASSS (range 0-72).17 Additionally, CFJ (C2-C3 to C6-C7) were scored (range 0-15) according to the de Vlam et al method9 as normal = 0, joint space narrowing or erosion = 1, partial blurring or ankylosis = 2, and complete blurring or ankylosis = 3. The composite score, CASSS, was achieved by summing the total mSASSS and the total CFJ score.10 Changes in each vertebral corner and CFJ were registered separately, allowing the calculation of both scores as well as the analysis of each spinal segment. The mean of the 3 readers’ scores was used in the analysis.
For the mSASSS, patients were excluded if > 3 anterior vertebral corners were missing. If ≤ 3 sites were missing, the mean score of the vertebrae from the same spinal segment was used as a substitute for the missing sites, as previously proposed.5,7 Regarding the CFJ score, if a facet joint could not be assessed, the missing score was substituted by the mean score of the remaining facet joints.8 Definite damage for the mSASSS was defined as ≥ 1 nonbridging or bridging syndesmophyte. Syndesmophytes were considered present if identified by at least 2 of the 3 readers at the same vertebral corner. For the CASSS, the presence of ankylosis (partial or complete) in ≥ 1 facet joint or at least 1 nonbridging or bridging syndesmophyte was considered definite damage.10
Statistical analysis. Descriptive analysis was performed using absolute and relative frequencies for categorical variables and quantitative measures (mean [SD] and median [range]) for numerical variables. The normality of numerical variables was evaluated using the Kolmogorov-Smirnov test. The independent samples t test and the Mann-Whitney U test were used as appropriate to compare subgroups. Correlations between clinical and radiographic scores were investigated using Spearman correlation coefficient (ρ) and were interpreted as follows: 0.00-0.20 as very weak, 0.20-0.40 as weak, 0.40-0.60 as moderate, 0.60-0.80 as strong, and 0.80-1.00 as very strong.10
Simple and multiple linear regression analyses were used to assess the independent associations between the variables of interest, using BASFI, BASMI, and ASQOL as dependent variables. Possible interactions between independent variables were tested. Factors with a P value < 0.20 in the univariate analyses were retested in multivariate analyses. Covariates potentially influencing the relationship between clinical and radiographic outcomes, such as sex, age, symptom duration, smoking status, and history of THA, were also investigated in the multivariate linear regression models. Models whose residuals were not normally distributed had their dependent variables transformed by Box-Cox.
Reliability between readers concerning the mSASSS and the CFJ score was estimated by the intraclass correlation coefficient (ICC). ICC values of 0.00-0.20, 0.20-0.40, 0.40-0.60, 0.60-0.80, and 0.80-1.00 were interpreted as poor, fair, moderate, good, and excellent, respectively.10
Analyses were performed using IBM SPSS Statistics 22 (IBM Corp.) and R version 3.2.3 (R Development Core Team). P values < 0.05 were considered significant.
RESULTS
Baseline characteristics. Results were reported following Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. A total of 101 patients with axSpA were included (96 with radiographic axSpA [r-axSpA] and 5 with nonradiographic axSpA [nr-axSpA]), of which 67 (66.3%) were male and the mean age was 47.7 (SD 12.5, range 18-74) years. HLA-B27 was available for 76/101 (75.2%) patients and was positive in 65/76 (85.5%). The mean symptom duration was 23.7 (SD 11.2, range 4-54) years, and it had been > 10 years since the first symptom for 87.1% of the sample. A large proportion of the study population had high disease activity: 32.7% had BASDAI values ≥ 4 and 54.5% had ASDAS values ≥ 2.1. Table 1 summarizes the characteristics of the included patients.
Clinical and demographic characteristics of the study population.
Radiographic damage of the spine was detected in all patients (mSASSS, range 1.3-71.3), including patients with nr-axSpA (mSASSS, range 9.7-27.4). Of the total number of patients, 48.5% met the criteria for definite damage according to the mSASSS, whereas 54.5% had definite damage according to the CASSS. Spinal damage scores are shown in Table 2.
Spinal damage according to the mSASSS and CASSS.
Reliability and agreement. Interobserver reliability was excellent for mSASSS, with an ICC of 0.90 (95% CI 0.84-0.93), and good to excellent for CFJ score, with an ICC of 0.88 (95% CI 0.76-0.93). Percentages of agreement between the 3 readers varied from 65% to 94% for sclerosis, 64% to 88% for squaring, 89% to 95% for erosion, 83% to 89% for syndesmophytes, and 88% to 93% for bridging syndesmophytes. Regarding the CFJ score, agreement varied from 81% to 86% for joint space narrowing or erosion, 78% to 92% for partial blurring or ankylosis, and 95% to 96% for complete blurring or ankylosis. The mSASSS was calculated in 97% (98/101) of patients. Three patients were excluded because > 3 vertebral corners were missing in the lumbar or cervical spine. No CFJ was missing.
Subgroup analyses. Subgroup analyses were performed according to sex, HLA-B27 status, disease activity, smoking status, and history of peripheral arthritis and uveitis. The last 2 variables were chosen because of their high prevalence in our sample. Men had less disease activity than women (BASDAI 2.6 vs 3.8, P = 0.01 and ASDAS 2.1 vs 2.5, P = 0.05), but worse mobility (BASMI 3.8 vs 2.8, P = 0.003) and more structural spinal changes (mSASSS 21.1 vs 9.2, P < 0.001 and CFJ score 5.2 vs 2.0, P < 0.001), including a worse cervical mSASSS (12.0 vs 4.5, P < 0.001). QOL impairment was similar between male and female patients (ASQOL 6.1 vs 7.1, P = 0.36). There was no difference between subgroups according to the HLA-B27 status, probably due to the small number of HLA-B27–negative patients (n = 11). Patients with high disease activity (ASDAS ≥ 2.1) showed higher scores in BASFI (4.9 vs 2.3, P < 0.001) and ASQOL (9.0 vs 3.4, P < 0.001), but no difference was observed for radiographic damage of the spine (mSASSS 18.9 vs 15.1, P = 0.30 and CFJ score 4.5 vs 3.7, P = 0.39). Current and past smokers had worse mobility (BASMI 4.0 vs 3.1, P = 0.03) and more structural damage in the spine (mSASSS 23.7 vs 12.7, P = 0.004 and CASSS 28.6 vs 16.3, P = 0.01) than patients who never smoked. They showed no difference in physical function (BASFI 4.1 vs 3.4, P = 0.16), QOL (ASQOL 7.0 vs 6.1, P = 0.38), and CFJ scores (5.0 vs 3.6, P = 0.20). Patients with a history of peripheral arthritis were younger (45 vs 52 yrs, P = 0.003) and had fewer structural spinal changes (mSASSS 13.8 vs 22.2, P = 0.03) than those without arthritis. The mean age and mean symptom duration were higher in patients with a history of uveitis than in those without uveitis (50 vs 45 yrs, P = 0.03, and 26 vs 20 yrs, P = 0.01, respectively). Considering the remaining health outcomes, no significant differences were observed comparing patients with and without uveitis.
Correlation matrix. Table 3 presents a correlation matrix for health outcomes in our population. BASFI correlated strongly with ASQOL (ρ 0.66, P < 0.001) and BASDAI (ρ 0.70, P < 0.001), moderately with BASMI (ρ 0.46, P < 0.001) and ASDAS (ρ 0.59, P < 0.001), and weakly with mSASSS (ρ 0.29, P = 0.004) and CASSS (ρ 0.28, P = 0.01). BASMI correlated strongly with mSASSS (ρ 0.70, P < 0.001) and CASSS (ρ 0.73, P < 0.001) and moderately with CFJ score (ρ 0.58, P < 0.001). There was a moderate correlation between mSASSS and CFJ score (ρ 0.56, P < 0.001).
Correlation coefficients (π) between health outcomes.
Multivariate linear regression analysis considering BASFI, BASMI, and ASQOL as the dependent variables. Multivariate linear regression analysis showed that disease activity (BASDAI), mobility (BASMI), and history of THA were independently associated with BASFI. Other models (Table 4; models 3-7) showed that disease activity assessed by ASDAS and spinal damage (CFJ score, CASSS, mSASSS, and definite damage according to the mSASSS) were also predictors of physical function. A best-fit model for BASFI, adjusted for symptom duration, age, sex, and smoking, included BASDAI (B 0.76, 95% CI 0.61-0.91, P < 0.001), BASMI (B 0.62, 95% CI 0.41-0.83, P < 0.001) and THA (B 1.22, 95% CI 0.001-2.43, P = 0.05). Disease activity had a greater effect on function (BASDAI, standardized β 0.64) than mobility (BASMI, β 0.43) and history of THA (β 0.13). The model fit was improved when BASDAI (adjusted R2 0.65) was used instead of ASDAS (adjusted R2 0.48).
Multivariate linear regression models for BASFI.
We further investigated the contribution of fatigue and peripheral arthritis to functional impairment by testing BASDAI without question 1 (fatigue) and, subsequently, BASDAI without question 3 (peripheral arthritis); there was the same number of variables for each model. These modifications in BASDAI still resulted in a good fit in the models (adjusted R2 0.63 and 0.68, respectively), suggesting that, in this sample, fatigue and arthritis had no significant effect on BASFI. Also, analyses of questions 1 (fatigue) and 3 (peripheral arthritis) as predictors of functional capacity (BASFI) did not reach statistical significance.
Radiographic scores were significantly associated with BASFI only when BASMI was removed from the model because of collinearity. Similarly, they were tested in multivariate analyses individually because of multicollinearity among them; those tested included mSASSS (B 0.03, 95% CI 0.01-0.05, P = 0.01), CASSS (B 0.02, 95% CI 0.01-0.04, P = 0.01), CFJ score (B 0.12, 95% CI 0.05-0.19, P = 0.001), and definite damage according to the mSASSS (B 0.03, 95% CI 0.01-0.04, P = 0.01). Based on these estimates, it would take 38.5 points in mSASSS or 41.7 points in CASSS to increase 1 point in BASFI. Considering definite damage, at least 20 nonbridging syndesmophytes would be necessary to increase 1 point in BASFI. Table 4 presents the main multivariate linear regression models for BASFI.
Moreover, BASMI was independently associated with structural damage according to CASSS (B 0.06, 95% CI 0.05-0.07, P < 0.001), mSASSS (B 0.07, 95% CI 0.05-0.09, P < 0.001), definite damage according to the mSASSS (B 0.06, 95% CI 0.05-0.08, P < 0.001), and CFJ score (B 0.21, 95% CI 0.15-0.27, P < 0.001). Symptom duration was statistically significant only in model 5. Neither disease activity (BASDAI or ASDAS) nor sex or age reached statistical significance (Table 5). Regarding the ASQOL, the best-fit model (adjusted R2 0.51) included BASDAI (B 0.92, 95% CI 0.50-1.33, P < 0.001) and BASFI (B 0.68, 95% CI 0.34-1.03, P < 0.001). No other variable showed statistical significance.
Multivariate linear regression models for BASMI.
DISCUSSION
In the present study, structural damage of the spine contributed to predicting physical function in patients with axSpA with long disease duration. However, both radiographic scores (mSASSS and CASSS) showed a weak correlation with BASFI. Deterioration of functional capacity was better explained by higher clinical disease activity, worse spinal mobility, and a history of THA.
Overall, clinical variables, such as disease activity (BASDAI and ASDAS) and spinal mobility (BASMI), showed a stronger association with function (BASFI) than irreversible damage of the spine (mSASSS and CASSS). BASDAI had the strongest association with BASFI (based on the higher standardized β), indicating that inflammation control plays a central role in improving functional capacity. In the same way, Poddubnyy et al6 demonstrated that BASFI remained stable in patients with r-axSpA treated with tumor necrosis factor inhibitors for up to 10 years despite radiographic progression.
Interestingly, BASDAI showed a better fit in the multivariate model than ASDAS, which we hypothesized could be explained by fatigue (a question present in BASDAI but not ASDAS). However, subsequent analyses did not demonstrate a significant association between BASDAI question 1 and BASFI.
As previously reported, in the subgroup analyses, we observed more structural damage in men.18-20 Conversely, Resnick et al21 described more cervical radiograph abnormalities in women than in men, but more recent studies did not confirm this finding.22,23 In our study, men had worse lumbar and cervical spine damage. Women had worse BASDAI, a more subjective measure of disease activity than ASDAS, for which no difference was observed. In contrast to other publications,18,19 women did not show a worse ASQOL. As expected, patients with high disease activity reported poorer physical function and QOL. However, there was no difference concerning structural damage in the spine between patients with high and low disease activity, which should be attributable to the cross-sectional study design. Longitudinal data from the Outcome in Ankylosing Spondylitis International Study (OASIS) cohort demonstrated that disease activity, mainly ASDAS, was associated with radiographic progression in r-axSpA.24 In line with other reports,25-28 smoking was related to more damage to the spine and worse mobility. Nevertheless, although some studies have also demonstrated an association between smoking and impairment of function27-29 and QOL,27,28 this was not observed in our study.
The role of imaging in monitoring axSpA is still a matter of debate. There is currently no recommendation to repeat spine radiographs at regular intervals as a standard approach.30 In our study, structural damage showed a modest effect on physical function: it would take approximately 40 points in mSASSS or CASSS to increase 1 point in BASFI. In contrast, Landewé et al1 demonstrated in the OASIS cohort, which also involved patients with long symptom duration and significant radiographic damage in the spine, a higher impact of structural damage on functionality, most evident in patients in the highest mSASSS category (range 37-72): it would take 17.5 points in mSASSS to increase 1 point in BASFI (variable estimate 0.057). A possible explanation for this difference is that patients with longstanding SpA could adapt to their physical limitations, and coping might be distinct among culturally and economically different populations. Hence, the BASFI, a patient-reported measure, would be underestimated. In line with this argument, Ward et al4 found that the strength of the association between mSASSS and BASFI decreased in late r-axSpA. More recent data from a metaanalysis comparing clinical aspects in patients with r-axSpA and nr-axSpA showed that, despite increased spinal structural damage (mSASSS) in the r-axSpA group, no relevant difference was found in function (BASFI) and QOL (ASQOL).31 It is possible that for the subgroup of patients with rapid radiographic progression, structural damage may have a greater and earlier effect on their mobility, physical function, and, consequently, their QOL. For those patients, a more frequent radiographic evaluation might be important for planning treatment goals.
Compared to the mSASSS, more patients with definite damage were identified according to the CASSS, though the correlation with functional capacity (BASFI) was not superior to that presented by mSASSS. The model with mSASSS explained 53% of the BASFI variance, whereas the model with CASSS explained 54% (both models included BASDAI and history of THA). Maas et al10 demonstrated an increase of 4% of patients with definite damage from baseline when using CASSS instead of mSASSS, a percentage comparable to our data (6%). A weak correlation between the radiographic scores and BASFI and no correlation with ASQOL were also reported.10 Despite detecting a few more patients with definite damage, the CASSS did not have a higher impact on the change in BASFI.
Regarding BASMI, structural damage in the spine was the primary determinant of mobility. This finding is consistent with a previous study reporting that both radiographic damage of the spine and spinal inflammation as assessed by magnetic resonance imaging contribute to mobility impairment.32 Wanders et al33 also reported a strong correlation between BASMI and mSASSS, with a Spearman correlation coefficient of 0.76, which was similar to what we found.
The worsening of QOL was significantly related to disease activity and physical function decline, confirming the results of earlier studies.19,34,35 Neither spinal structural damage nor mobility contributed to predicting QOL, sustaining the hypothesis that patients with axSpA could habituate to some disability, as previously mentioned.
A limitation of our study is its cross-sectional design. Thus, the performance of mSASSS and CASSS in assessing the progression of radiographic damage has not been investigated, nor have the effect of treatment on disease activity been measured. Another limiting factor may be the potential lack of generalizability. Our sample consisted mostly of patients with long symptom duration, and results may not apply to cohorts of patients with shorter disease duration. Moreover, patients’ BMI and physical activity level, which are other factors that may contribute to functionality and mobility, were not evaluated.
In summary, our data show that hip involvement, mobility impairment, and mainly disease activity play significant roles in physical disability in patients with axSpA. These results may indicate that, even in patients with long disease duration and high structural damage, prioritizing the control of inflammatory activity should be the main goal to improve function and QOL, although this observation must be further explored in prospective studies. Despite the focus on controlling inflammation, measuring structural damage of the spine remains relevant in the assessment of axSpA and should not be neglected, particularly in patients with rapid radiographic progression. The addition of CFJ assessment to the mSASSS (ie, the CASSS) does not seem to improve its ability to predict physical function.
Footnotes
The authors declare no conflicts of interest relevant to this article.
- Accepted for publication May 6, 2024.
- Copyright © 2024 by the Journal of Rheumatology
