Methods

The patients were drawn from the database of the Spondylitis Clinic in Toronto Western Hospital. This database is composed of all consecutive patients seen between July 2003 and March 2007. All patients attending this clinic are ⩾17 years of age. The referral sources for the Spondylitis Clinic include The Hospital for Sick Children in Toronto, as well as community rheumatologists, family physicians, gastroenterologists and ophthalmologists. Patients with AS are evaluated according to a standard protocol in which full demographic, clinical, laboratory and radiographic details are recorded at regular intervals. All patients who met the modified New York criteria for AS2 or axial spondyloarthropathy criteria18 were extracted from the database. Those with an onset of musculoskeletal symptoms at ⩽16 years were compared with patients with a symptom onset at ⩾17 years.

Clinical features recorded systematically in the database include the following: age, gender, ethnicity, age at symptom onset, age at diagnosis, arthritis pattern at diagnosis, family history and demographics (including education, employment status, alcohol and smoking history). Extra-articular manifestations—specifically uveitis, inflammatory bowel disease (IBD) and psoriasis—are documented. The location and character of joint symptoms are also recorded in detail. A comprehensive list of medications is included in this assessment. The physical examination includes full spinal clinimetrics, a swollen and tender joint count and an assessment of enthesitis. The general physical examination is also recorded.

The Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) was used to assess disease activity. The BASDAI is a six-question patient survey which uses a visual analogue scale (VAS) to assess fatigue, axial and peripheral joint pain, tenderness, and severity and duration of morning stiffness. It is a reliable and validated measure limited to assessment of symptoms.19 The Bath Ankylosing Spondylitis Functional Index (BASFI) was used to assess the functional status. It comprises 10 VAS questions measuring the functional ability in the previous week. Scores on each item range from 0 (easy) to 10 (impossible).20 The BASFI has been shown to be sensitive in demonstrating an improvement in functional ability in patients with AS.

The Short Form 36 (SF-36) was used to assess health-related quality of life. The SF-36 consists of eight dimensions: physical functioning, social functioning, physical role, emotional role, mental health, vitality, bodily pain and general health perceptions. Scores range from 0 (worst) to 100 (best), with higher scores indicating better health status.21 Other self-recorded questionnaires performed by the patient at each assessment include the AS quality of life (QoL) index and the health assessment questionnaire (HAQ), both of which are used frequently in the assessment of quality of life and function, respectively, in patients with AS.22 Patient global disease activity, nocturnal and total back pain score and the fatigue severity index (FACIT) are also recorded. Laboratory studies include complete blood count, creatinine, liver function tests, erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP). The HLA-B27 status is determined at the initial visit.

Each patient attending the Spondylitis Clinic has the following set of radiographs performed at 2-year intervals: AP pelvis, cervical and lumbar spine. For the purposes of this analysis, if a patient had a set of radiographs performed within 1 year of the clinical assessment, they were included in the dataset. The AP radiograph of the pelvis was used to grade the severity of sacroiliitis according to the modified New York criteria2 and to document the presence of hip disease. The lateral cervical and lumbar spine radiographs were scored by the modified Stoke Ankylosing Spondylitis Spine Score (mSASSS).23 All radiographs are scored by one reader (FOS) experienced in reading and scoring them.

Statistical analysis

Comparisons between the JoAS and AoAS cohorts were performed using descriptive statistics and parametric and non-parametric tests as appropriate. In all cases the most recent clinical assessment was used. Descriptive statistics were computed for the demographic data, family history and personal history. The odds ratio (OR) and 95% confidence intervals (95% CI) were calculated for each characteristic adjusted by either disease duration or current age to determine the odds of AoAS versus JoAS. We conducted two multivariable logistic regressions to determine what characteristics were associated with AoAS, with backwards stepping with p to remove at 0.10. The following characteristics were entered into the model: axial pattern, alcohol, family history of arthritis, family history of AS, iritis, neck pain, neck stiffness, back pain, back stiffness, sacroiliac pain, buttock pain, duration >3 months, physical component scale, mental component scale and the HAQ. Current age was forced into the first model and disease duration was forced into the second model.

Results

Demographic data

A total of 267 patients with AS were identified from the database; 84 met the criteria for JoAS and 183 met the criteria for AoAS. These two groups were then compared. The mean age at symptom onset for patients with JoAS and those with AoAS was 12.8 years and 26.0 years, respectively; the mean age at diagnosis for the two groups was 17.2 years and 32.4 years, respectively, and mean disease duration was 14.7 years and 16.7 years for patients with JoAS and AoAS, respectively. Between the cohorts there were no differences in gender ratio (81% male for the JoAS group and 79% for the AoAS group), HLA-B27 profile (75% positive in the JoAS group and 81% positive in the AoAS group) or ethnicity (80.5% Caucasian in the JoAS group and 86.8% in the AoAS group).

Extra-articular manifestations

There was no difference in the frequency of a family history of AS, psoriasis, psoriatic arthritis, uveitis or IBD between the two groups (table 1), nor in a personal history of psoriasis or IBD. Uveitis was more common in JoAS than in AoAS (27.7% vs 24.5%). In fact, when adjusted for current age, patients with JoAS were more than twice as likely to report at least one episode of uveitis at some stage during their illness (OR 2.22 (95% CI 1.05 to 4.76)).

Pattern of disease

The axial/peripheral pattern of disease at onset differed significantly: axial involvement (with or without peripheral involvement) was seen in 95% of patients with AoAS and in 74% of those with JoAS (OR 6.93 (95% CI 2.80 to 17.20) when adjusted for disease duration; table 2). This difference in axial/peripheral pattern continued throughout the course of the disease. Adjusted for disease duration, axial features were more prominent in AoAS than in JoAS, as reflected by neck stiffness, back stiffness and lumbar pain. Morning stiffness was also more common in AoAS than in JoAS (table 2).

With regard to last clinical assessment, there was no difference between the tender or swollen joint count between the two groups. Patients with JoAS had a mean tender joint count of 1.3 (range 0–10) compared with 1.1 (range 0–22) in those with AoAS. Patients with JoAS had a mean swollen joint count of 0.2 (range 0–10) while those with AoAS had a mean tender joint count of 0.2 (range 0–12). The number of patients in remission (defined as no back pain and no active joints) at the most recent clinical assessment differed between the JoAS and AoAS groups (11% and 2%, respectively). There was no difference between the cohorts with respect to current enthesitis (JoAS 10%, AoAS 9.6%) or previous joint replacement and/or history of spinal surgery (JoAS 7.2%, AoAS 7.1%).

Clinimetric data

Consistent with the higher frequency of back symptoms in patients with AoAS, this group had greater impairment of spinal mobility. Chest expansion, occiput-to-wall, Schober test, intermalleolar distance, cervical rotation and lateral spinal bending were all significantly more impaired in the AoAS group than in the JoAS group when adjusted for disease duration (table 3). The absolute percentage of patients with an abnormal measurement for each of these variables was also significantly higher in the AoAS group than in the JoAS group, even when adjusted for disease duration. For example, 80.5% of patients with AoAS had an abnormal Schober test copared with 53.3% of the JoAS group. When adjusted for disease duration, the AoAS cohort was more than three times more likely to have an abnormal Schober test (OR 3.27 (95% CI 1.79 to 5.99)).

Disease activity

There were no differences in disease activity at the time of assessment in the clinic. Although there was a trend for higher BASDAI, ESR and CRP in the AoAS group, these differences were not significant when corrected for disease duration and current age. Mean BASDAI scores for the JoAS and AoAs groups were 3.9 and 4.7, respectively; mean ESR levels were 12.6 and 15.9, respectively, and mean CRP levels were 9.2 and 13.9, respectively. Complete blood count, creatinine and liver function tests were similar in the two cohorts. There was also no difference in the use of medications between the two groups. Current use of non-steroidal anti-inflammatory drugs was 43% in the JoAS group and 47% in the AoAS group; current use of disease-modifying antirheumatic drugs/biological agents was 38% in the JoAS group and 33% in the AoAS group.

Functional impairment and self-reported questionnaires

AoAS was associated with higher BASFI scores than JoAS (3.9 vs 2.6). When adjusted for disease duration, the AoAS cohort had 20% more severe functional impairment (OR 1.20 (95% CI 1.06 to 1.35)). This finding was further supported by the other instruments which showed that the AS quality of life measure, fatigue severity scores, total back pain scores and HAQ scores, when adjusted for disease duration, were significantly worse in the AoAS group (table 4).

Radiographic features

Significant differences were identified between the radiographic severity of the two cohorts as judged by the mSASSS. Mean (SD) mSASSS scores were 6.9 (14.6) for the JoAS group and 23.9 (25.9) for the AoAS group; these values were significant when adjusted for disease duration (OR 1.05 (95% CI 1.03 to 1.08)). Of the AoAS cohort, 53.6% had an mSASSS score of ⩾10 units compared with 17.3% of the JoAS group (OR 9.18 (95% CI 3.53 to 23.90) in favour of the AoAS group when adjusted for disease duration). Radiographic evidence of hip disease was seen in 17% of the JoAS group and 30% of the AoAS group, but these values were not significantly different when adjusted for disease duration and current age.

Multivariate analysis

In the multivariable logistic regression analysis it was found that, after adjusting for disease duration, individuals who had axial symptoms were more likely to have had AoAS (OR 6.32 (95% CI 1.87 to 21.31)). After adjusting for current age, it was found that individuals who had axial symptoms were more likely to have AoAS (OR 7.61 (95% CI 1.37 to 42.20)). However, individuals who had a history of uveitis (OR 0.17 (95% CI 0.05 to 0.58)) or a family history of uveitis (OR 0.06 (95% CI 0.01 to 0.94)) were less likely to have AoAS.

Discussion

We have found significant differences between two cohorts of patients with AS based on the age of symptom onset, and these differences were independent of gender and HLA-B27 status. Patients with JoAS were more likely to present with peripheral disease than those with AoAS. Subsequently, the patients with AoAS were more likely to have greater axial symptoms throughout the course of their disease, even when adjusted for disease duration. Continuing with this trend, the AoAS cohort had more severely impaired clinimetric data and radiographic features. This, in turn, was reflected in a higher functional impairment in the AoAS group.

A unique feature of our study in comparison with previously published studies is the fact that all 267 patients with AS were seen in the one institution using a uniform protocol. Other distinctive aspects include the comprehensive scope of the clinical assessment including demographic details, clinical features, laboratory and radiographic variables, as well as detailed self-reported questionnaires of severity, activity and functional impairment. This is the first study to systematically address all these variables in one single cohort.

To date, studies have been contradictory regarding the impact of age of onset on the clinical course of AS. The functional impact of AS has been variably reported to be worse in JoAS than in AoAS,16 or the same.17,24 Some older studies have even suggested a trend towards better functional outcomes in JoAS than in AoAS, as well as less disability and impact of disease.25 Differing study design may contribute to these discrepancies. One study was based on a postal survey analysing BASFI scores of patients on a mailing list of an AS patient support group with no formal assessment or examination of the cohort.16 A recent study used a cross-sectional component of a larger cohort recruited from a number of different centres across North America.17 Our study is the first to show that functional outcomes are actually better in JoAS than in AoAS. This conclusion is based not only on the BASFI, the most widely used instrument for functional assessment in AS, but also other well established indices such as the HAQ and SF-36. The tendency for a better functional outcome in JoAS was present even when corrected for disease duration, current age and gender. Older studies that examined functional outcomes in JoAS used much smaller numbers of patients, making any formal comparisons difficult.12,14

Studies of disease activity in AS have traditionally used the BASDAI as the primary outcome measure. Our standard protocol allowed us to analyse other aspects of disease activity such as pain intensity, fatigue levels, patient global assessment and inflammatory markers. We have shown that JoAS is not associated with higher disease activity than AsAS at a random cross-sectional observation in the disease course. Other clinical variables such as extra-articular features and hip disease were in agreement with the BASDAI trends. There was a trend towards higher disease activity in patients with AoAS that has not been demonstrated previously, but this did not reach statistical significance.

With regard to radiographic severity, our findings using mSASSS are consistent with previous studies using BASRI.10,17 The mSASSS has been shown to be an accurate measure of radiographic damage in AS, and is more sensitive to change over time than the BASRI. In general, the mSASSS has superseded the BASRI in most studies of radiographic damage in AS. The advantage of using the mSASSS is that patients can be grouped into various different categories depending on the severity of their radiographs. In our cohort, 83% of the patients with JoAS had mSASSS scores <10, indicating a low burden of radiographic damage. In contrast, only 46% of the patients with AoAS had mSASSS scores <10. We did not have the necessary data to examine the radiographic findings in the peripheral joints of the entire cohort. Old data have demonstrated more severe peripheral radiographic findings in JoAS than in AoAS.26

One previous study that examined the clinical and radiographic differences between JoAS and AoAS used a cohort of patients with long disease duration (⩾20 years).17 We chose to include all patients with JoAS in our clinic in the analysis and to adjust for disease duration and current age. AS is a condition with a spectrum of severity, even in its early stages, and our cohort reflects general experience. Another difference between the previous study and ours is the definition of the age of onset of symptoms. In the previous study, age of onset referred to onset of symptoms of inflammatory back pain only. In our cohort, a quarter of the patients with JoAS started with exclusively peripheral disease and yet went on to develop radiographic sacroiliitis sufficient to meet the criteria for AS. These patients constitute an important part of the JoAS spectrum. Another study, using a smaller cohort than in our study, examined the relationship between clinical and radiographic data in an AS cohort depending on the age of symptom onset.8 That study also found that, clinically and radiographically, JoAS had a less severe spinal disease course than AoAS. Their cohort differed from ours in that it included non-Caucasian subjects, had very few female patients and excluded all patients with psoriasis and IBD. It did not include an assessment of functional impairment or disease activity.

From the perspective of the paediatric rheumatologist, the concept of juvenile spondyloarthropathy (JSpA) refers to a group of rheumatic diseases with onset ⩽16 years of age characterised by arthritis, enthesitis, inflammatory back pain and an association with HLA-B27.27,28 JoAS is the best characterised member of this family, which falls under the unifying term enthesitis-related arthritis (ERA), one of the subtypes of chronic arthritis in childhood.29,30 These different conditions share common features of axial and peripheral arthritis, enthesitis, male preponderance, positive family history, association with HLA-B27 and distinctive extra-articular features such as uveitis. Any of the undifferentiated forms of JSpA can progress to JoAS over time; however, the exact percentage has been a source of much debate in the literature. One follow-up study of children with seronegative enthesopathy and arthropathy (SEA) syndrome found that 50% had developed into a definite AS by 11 years.31 Another study reported that 75% had developed into AS within a shorter time frame of 5 years.32 It has also recently been suggested that patients with ERA had poorer physical outcomes than those with other forms of juvenile idiopathic arthritis.33

To obtain a more accurate comparison, we limited our study to the juvenile onset cohort that had subsequently developed AS by the time of assessment. Contrary to previous work in this area, we found that JoAS had less functional impact than AoAS, as well as less severe spinal disease both clinically and radiographically. Further work is needed into the biology of this fundamental difference, as it will undoubtedly give insights into the pathogenesis of AS across the entire age spectrum. Our studies suggest that there is significant heterogeneity in the adult AS population, and that age of onset may represent one contributor to this heterogeneity. This has implications for genetic studies in AS, which have regarded patients meeting New York criteria for this disease as a homogeneous entity. It also has implications for defining the expected trajectory of radiographic progression in AS. It is clear that age of onset influences this trajectory. Defining the time course of structural damage has proved to be an important variable with which to interpret the impact of biological agents on the natural history of AS.34,35,36 Heterogeneity in the AS patient population by such variables as age of onset may complicate this analysis.

In conclusion, if JoAS represents the same disease process as AoAS but presenting at an earlier age, it would be expected that the pattern of disease expression would be comparable between the two if corrected for disease duration. This was not the case—JoAS follows a clinical course which is distinctive in its axial/peripheral pattern. Furthermore, these distinctive clinical features must be dictated by factors other than male gender, ethnicity, family history and HLA-B27. This raises the notion that JoAS and AoAS are fundamentally different conditions, and the phenotypic differences persist over time. Further studies into other prognostic variables in AS will be invaluable for interpreting findings in pathogenesis and treatment.