Abstract
Objective. To investigate the demographic and clinical characteristics associated with early, extensive radiographic changes in ankylosing spondylitis (AS).
Methods. Radiographic severity was assessed cross-sectionally in 235 patients with AS using the Bath AS Radiological Index spine score (BASRI-s). Patients with extensive radiographic changes on the lumbar portion of BASRI-s were defined as the early axial ankylosis (EAA) Group. ANCOVA and logistic regression analyses were used to identify factors affecting EAA.
Results. Most study patients were men (139/235, 59.0%). Mean disease duration was 12.4 ± 9.3 years. Fifteen percent of women and 34.8% of men with AS were in the EAA group. HLA-B27-positive men with AS had significantly higher BASRI-lumbar scores, while HLA-B27 had no effect on radiographic progression of axial disease in women with AS. Peripheral joint involvement was associated with slow radiographic progression. Hip involvement had no effect on axial progression but uveitis was more frequent in the male EAA group. The odds for an HLA-B27-positive male patient with AS who did not have peripheral arthritis of having a BASRI-lumbar score of 3 or higher were 3.4 (77% chance to have axially progressive disease). Presence of uveitis increased these odds to 93%. Only 15% of female patients with AS had EAA, and the absence of peripheral arthritis was the only clinical measure associated with EAA in this group.
Conclusion. EAA was more frequent in men with AS than in women. Absence of peripheral arthritis, HLA-B27 positivity, and uveitis were associated with multiple syndesmophytes or fusion of multiple vertebrae of the lumbar vertebrae.
Ankylosing spondylitis (AS) is a longstanding, chronic inflammatory disease with a relatively slow rate of progression. Future restriction of spinal mobility is the major concern of patients with AS at the time of diagnosis; about 40% of patients have severe functional loss several decades after the onset1,2,3,4. The rate of radiographic progression differs substantially among patients with similar durations of AS2,5,6. At present, knowledge about prognostic factors affecting radiographic severity for individual patients is sparse. Previously, peripheral arthritis has been suggested to be a predictor of disease progression1 and loss of function7,8. Similarly, male sex, uveitis, and hip involvement were found to be associated with radiographically progressive axial disease9,10,11,12,13,14. However, several studies assessing the role of HLA-B27 on radiographic severity reported no association with progression13,15,16.
Previous data on the natural course of AS suggest that up to 15% of patients have an accelerated disease course within the first decade after disease onset. This accelerated functional loss seems to predict the disease severity in longstanding AS1.
We therefore aimed to determine the association of demographic and clinical characteristics affecting disease severity in patients with AS with short disease duration. We used the Bath Ankylosing Spondylitis Radiology Index for the spine (BASRI-s) to determine the severity17.
MATERIALS AND METHODS
Patients and data collection
Three hundred sixty-nine consecutive patients with AS diagnosed according to the modified New York criteria at 4 different medical centers in Istanbul, Turkey, were registered between January 2003 and September 2010 for a prospective followup after a final evaluation at the Marmara University Hospital Rheumatology outpatient clinic18. At each medical center, patients completed questionnaires on personal and medical history and underwent clinical evaluation by an experienced physician. Patients meeting the modified New York criteria were then referred to the rheumatology department of Marmara Medical School. Final enrollment in the study was done by 2 rheumatologists. Patients with a history or current diagnosis of inflammatory bowel disease or psoriasis were excluded.
Clinical and laboratory variables
Clinical and demographic data on age at onset, sex, HLA-B27 status, delay in diagnosis, erythrocyte sedimentation rate (ESR), and C-reactive protein (CRP) levels at the time of diagnosis were collected. Data were also collected on family history for AS, uveitis, and hip or peripheral joint involvement along axial disease. Disease duration was defined as the time in years starting from the first low back pain episode until the radiographic analysis.
Radiographs of pelvis, lumbar spine on anteroposterior and lateral planes, and cervical spine on lateral plane were scored using the Bath Ankylosing Spondylitis Radiology Index for the spine (BASRI-s) to determine radiographic severity17. BASRI-s is a reliable and valid measure of radiographic damage in AS19,20. BASRI-s is based on scoring of the lateral view of the cervical spine (C1 to C7), and the frontal or lateral views of the lumbar spine (Th12 to S1) and the sacroiliac joints. Grading is done as follows: Grade 0: normal findings; Grade 1: dubious changes; Grade 2: slight changes, including erosion, vertebral squaring or sclerosis, with or without syndesmophytes, on fewer than 2 vertebrae; Grade 3: syndesmophytes on more than 3 vertebrae, with or without fusion involving 2 vertebrae; Grade 4: fusion involving more than 3 vertebrae.
Radiographs were scored by a single rheumatologist. The intrareader reliability of the BASRI-s, based on 2 readings of the radiographs of 100 patients performed at least 3 months apart, was 0.975 (95% CI 0.979–0.989).
HLA-B27 was determined by standard microcytotoxicity assays, and whenever the first test result was negative it was repeated to minimize the possibility of an erroneous negative result. Scoring of radiographs of the lumbar vertebral column on anteroposterior and lateral planes was used to investigate the effect of clinical and laboratory factors on early axial progression. Scoring of the radiographs of sacroiliac joints and cervical vertebral column was omitted from the evaluation and BASRI-s score for the lumbar portion was used for statistical analysis.
The University of Marmara institutional review board approved our study, and study patients gave informed consent.
Statistical analysis
Descriptive statistics were used for the data by the calculation of means with SD and range for continuous data. For dichotomous and ordinal data, frequencies were calculated. Differences between the groups were tested by computing OR and 95% CI. Associated p values for dichotomous data were calculated by chi-squared statistics, and p values for continuous data were calculated by Mann-Whitney U tests. Since radiographic progression appears to be a function of time, ANCOVA was used to explore the effect of individual variables on radiographic progression between groups.
The effect of multiple variables [sex, age at onset, disease duration, HLA-B27, hip arthritis (BASRI-hip score ≥ 2), peripheral arthritis, enthesitis, or uveitis] on severe axial radiographic damage was assessed with logistic regression analysis. Interactions between the sets of variables were explored in additional analyses that included the main effect variables and the interaction term for each interaction studied. The exponents of the estimates of those logistic regression analyses were used to calculate OR for each category of radiographic damage. All analyses were performed in SPSS 13.0.
RESULTS
Patients
Three hundred sixty-five patients were registered for followup between January 2003 and September 2010. At the time of analysis, 102 patients were receiving anti-tumor necrosis factor-α therapy, and radiographs of 28 patients were not suitable for evaluation. These patients were excluded from the study and the final analysis was done on 235 patients. All patients fulfilled the modified New York criteria for AS. Demographic features of the study patients and disease characteristics are given in Table 1. Mean age of study patients was 39.1 ± 11.2 years and mean disease duration since the onset of low back pain was 12.4 ± 9.3 years. Most study patients were men (139/235, 59.1%) and 70.2% were HLA-B27-positive (men AS = 75.5%, women AS = 63.3%).
Demographic and clinical features of study patients with AS.
Characteristics of patients according to radiographic progression
Comparison of clinical characteristics and demographic data of patients on the basis of radiographic progression is given in Tables 1 and 2. Sixty-two patients (26.4%) had radiographic evidence of extensive lumbar spine involvement (with a BASRI-lumbar score of Grade 3 or 4). One hundred seventy-three patients (73.6%) had no or early changes of lumbar spine involvement (BASRI-lumbar score of Grade 0, 1, or 2). A history of peripheral arthritis was recorded in 99 (42.1%) patients, enthesitis in 119 (50.6%), hip involvement in 120 (51%), and uveitis by patient report in 55 patients (23.4%). Radiographic change of progression of the lumbar spine was more common in men (mean BASRI-lumbar in male vs female patients: 1.77 ± 0.11 vs 0.97 ± 0.14, respectively; p < 0.0001, 95% CI −1.16 to −0.39; Table 3).
Demographic and clinical features of AS patients according to radiographic progression.
Factors associated with axial ankylosis in patients with AS.
As shown, disease duration is associated with radiographic progression15. Therefore, univariate ANOVA was used to eliminate the effect of disease duration on radiographic progression. This approach revealed that radiographic progression in the lumbar spine was more extensive in HLA-B27-positive patients with AS (mean BASRI-lumbar in HLA-B27-positive vs negative patients with AS: 1.58 ± 0.12 vs 0.85 ± 0.18, respectively; p = 0.002, 95% CI −1.14 to −0.18; Table 3). Disease duration was comparable in HLA-B27-positive and negative patients with AS of both sexes. But when analyzed within gender groups, radiographic progression of axial disease was not affected by the presence of HLA-B27 in women with AS (mean disease duration in HLA-B27-positive vs negative patients with AS 12.24 ± 9.44 vs 12.20 ± 8.80 years; p = 0.97, 95% CI −2.45 to 2.55; mean BASRI-lumbar score in HLA-B27-positive vs negative women with AS 1.10 ± 0.18 vs 0.90 ± 0.20; p = 0.47, 95% CI −0.81 to 0.36).
HLA-B27-positive men with AS were 3.37 times more likely than HLA-B27-negative men with AS to have an early radiographic progression (mean BASRI-lumbar in HLA-B27-positive vs negative men with AS 1.83 ± 0.16 vs 0.81 ± 0.32, respectively; p = 0.006, 95% CI −0.62 to 0.45; Tables 3 and 4).
Logistic regression analysis of early axial ankylosis.
Radiographic progression of axial involvement was milder in patients with AS who had peripheral joint involvement (mean BASRI-lumbar score in women with AS who had peripheral arthritis vs women who did not have peripheral arthritis 0.68 ± 0.16 vs 1.30 ± 0.16, respectively; p = 0.011, 95% CI 0.27 to 1.07; mean BASRI-lumbar score in men with AS who had peripheral arthritis vs men who did not have peripheral arthritis 1.21 ± 0.22 vs 2.04 ± 0.15; p = 0.003, 95% CI 0.09 to 0.91; Table 3). Family history for AS had no effect on axial radiographic changes (mean BASRI-lumbar score in patients with family history vs those without family history 1.49 ± 1.45 vs 1.50 ± 1.48; p = 0.85, 95% CI −0.58 to 0.49). Mean delay in diagnosis was 6.13 ± 6.2 years and it did not differ between the sexes (mean delay in diagnosis in women vs men with AS 6.11 ± 6.05 vs 6.11 ± 38 years; p = 0.99, 95% CI −2.01 to 2.00).
A history of enthesitis was protective for lumbar extension of the disease in the whole group, but when analyzed within gender groups the protective effect was lost (Table 3). A history of uveitis was associated with extensive radiographic progression in male patients only. Radiographic progression in the lumbar portion of the vertebral column was not different in patients with and those without hip involvement (mean BASRI-lumbar in patients with vs without hip involvement 1.36 ± 0.13 vs 1.55 ± 0.13; p = 0.33, 95% CI −0.29 to 0.51).
It has been shown that radiographic progression in AS is more severe in men, and that it is a function of disease duration2. Therefore, we applied logistic regression models separately for men and women when analyzing the effect of multiple clinical and laboratory measurements on severe radiographic progression (Table 4).
HLA-B27-positive men with AS who have uveitis and no history of peripheral arthritis were 13.0 times more likely to have extensive radiographic progression (OR e(−1.16 + 1.69 + 1.04 + 1.01) = 2.71(−1.16 + 1.69 + 1.04 + 1.01) = 2.71(2.58) = 13.0; Table 4).
Among the women patients, a Grade 3 or 4 BASRI-lumbar score was seen in only 15% and women with AS had only a 37% chance to have radiographic changes of early axial progression (Table 1). Absence of peripheral arthritis was the only clinical feature associated with a worse radiographic outcome in women with AS (Table 4).
The mean BASRI-lumbar scores of patients treated with biologic agents did not differ from those treated with conventional disease-modifying antirheumatic drugs (1.37 ± 1.13 vs 1.41 ± 1.01, respectively; p = 0.76, 95% CI −0.91 to 1.21). Disease duration and HLA-B27 prevalence did not differ between these groups.
DISCUSSION
Prognostic information about the future course of AS is helpful for consulting patients and can be used to select patients most at risk for developing severe radiographic damage and hence functional loss. At present, knowledge about accurate prognostic information for individual patients is poorly defined. It was suggested that the rate of natural radiographic progression in patients with AS is rather linear2,6 and a predictable disease pattern emerges within the first 10 years after the onset of low back pain1. In a longitudinal prospective study, Carette, et al reported that 74% of patients with mild spinal restriction within the first 10 years did not progress to a more severe category of functional loss after a mean disease duration of 38 years.
We aimed to evaluate the rate of axial radiographic progression in patients with AS who had relatively short disease duration. After a mean disease duration of 12 years, 52.3% of the patients studied had no axial radiographic changes. However, extensive radiographic changes of the lumbar vertebral column were already present in 26.4%, suggesting that a substantial proportion of patients are at risk for developing extensive axial ankylosis and may have severe functional restriction eventually.
In our study, a small set of clinical variables and HLA-B27 testing differentiated patients with extensive axial progression. Radiographic progression was more severe in men with AS and this finding is in accord with previous studies9,12,21,22. Disease onset was earlier in male patients, but disease duration did not differ with sex, and age at disease onset did not differ in the more progressive group, as supported by previous studies2,9. Interestingly, HLA-B27 was more frequent in the male AS group, and still its overall prevalence of 70.2% was much lower than previously reported23. In contrast to previous reports addressing the role of HLA-B27 in disease severity24,25 and radiographic progression26, HLA-B27 was found to be associated with severe radiographic progression in men. However, other studies either did not assess radiographic changes as an outcome measure or had different sets of patients with longer mean disease durations and high HLA-B27 prevalences. The association of HLA-B27 presence with the progression of ankylosis in AS is difficult to address in cohorts where the prevalence of HLA-B27 is as high as 90%. To overcome this difficulty, family studies and studies in siblings were undertaken, and a possible contribution of HLA-B27 was supported by the presence of high heritability of radiographic severity and concordance among siblings27,28. In contrast with reports from Europe and North America, our patient group had a lower HLA-B27 prevalence. Interestingly, this low prevalence is in accord with a previous report from the same region29.
One other difficulty of defining radiographic disease progression toward the vertebral column may be the use of total BASRI-lumbar score, which is the sum of scores of 3 different levels of the vertebral column. Extensive involvement of the sacroiliac joints, increasing the total score, may blunt the limited differences in the total score detected by plain radiographs. As shown, the best predictor for future ankylosis is the presence of syndesmophytes or ankylosis of the vertebral column itself at the time of analysis30. We therefore used only the BASRI score of the lumbar portion of the vertebral column31. This approach allowed us to choose patients with the highest risk of progression according to current knowledge in the AS literature.
Clinical and genetic factors contributing to early radiographic progression differed between the sexes. This finding was in accord with previous reports2,9,11,13. In contrast to previous studies, HLA-B27 was found to be associated with early axial progression in men only. The absence of peripheral arthritis or patient-reported history of uveitis was also associated with multiple syndesmophytes or fusion of multiple vertebrae.
In our patient population, almost 53% of patients with AS had no radiographic changes of the vertebral column except sacroiliitis. Studies addressing radiographic severity in patient groups with almost twice the disease duration are in accord with this slow progression rate15. A man with AS who was HLA-B27-positive and without a history of peripheral arthritis was 3.37 times more likely to have axially progressive disease (77% chance to have an axially progressive disease). Adding uveitis to the model when present increased the OR to 13.0 (93% chance to have an axially progressive disease). In contrast, axial progression in women with AS was associated only with the absence of a cumulative history of peripheral arthritis.
A limitation of our data was their cross-sectional nature, and variables shown to be associated with spinal changes may not be “prognostic” factors. On the other hand, median disease duration of 10 years and the strict criteria to discriminate axial progression of ankylosis in the presence of previous research reporting the prognostic value of severity in the first 10 years on future functional loss, one can assume that clinical and genetic factors reported here suggest the presence of possible prognostic factors. To assess the prognostic value of the measures reported here, further studies, prospective in design, are needed.
- Accepted for publication July 12, 2010.
