Although oligoarticular-onset juvenile rheumatoid arthritis (oligo-JRA) has been described as the benign-onset form of JRA (1–4), some studies suggest that the severity and chronicity of this disease have been underestimated (5, 6). In particular, those whose disease follows an extended (i.e., polyarticular) course are likely to have a worse prognosis (2, 7, 8). An extended disease course occurs in 10–50% of patients (5, 9, 10). Disease extension, when it occurs, does so within the first 2 years in only 20–30% of patients (5, 11); therefore, a polyarticular disease course in itself is not useful as an early predictor of disease outcome.
Identification of early predictors of poor outcome for patients with JRA should help determine who should receive early aggressive therapy, enable more appropriate counseling, and hopefully, improve outcome. Factors known to be associated with poor outcome have largely been identified coincidentally in studies that have been primarily concerned with measuring outcome. The interpretation of these studies has been limited by small sample sizes or univariate analysis of data. Two studies have used multivariate analysis either to identify early predictors of outcome for all onset types of JRA (3) or to examine disease factor determinants of disability in patients with juvenile chronic arthritis (JCA) (2). Although these studies examined arthritis onset types as factors that might predict poor outcome, they did not identify predictive factors associated with poor outcome specifically for oligo-JRA. Knowledge of early predictors of disease course for oligo-JRA using multivariate analysis is limited to one recent report, which describes outcome and predictors of severity for patients with oligoarticular-onset juvenile idiopathic arthritis (JIA) (5).
Extension to a polyarticular course is one measure of outcome that is easily evaluated. Not only is a polyarticular disease course inherently more severe than persistent oligoarticular disease, it may also be a marker of poor outcome measured in other ways (2, 7). Using polyarticular extension of disease as the principal measure of poor outcome, we examined clinical features in the first 6 months that might be associated with this outcome.
PATIENTS AND METHODS
- Top of page
- PATIENTS AND METHODS
We retrospectively reviewed the medical records of 365 patients with oligo-JRA who were first seen at the Pediatric Rheumatology Clinic at the British Columbia Children's Hospital and at the Mary Pack Arthritis Center between February 1965 and January 1994. The 205 patients eligible for study fulfilled the American College of Rheumatology (formerly, the American Rheumatism Association) criteria for a diagnosis of oligo-JRA (12), were followed up for at least 5 years, were not rheumatoid factor positive, and during the first 6 months of their disease course, did not fulfill criteria for juvenile psoriatic arthritis (13), seronegative enthesopathy and arthropathy (SEA) syndrome (14), or juvenile ankylosing spondylitis (juvenile AS) (15). Persistent oligoarthritis was defined as arthritis involving ≤4 joints at any time during the disease course; extended oligoarticular arthritis was defined as arthritis of oligoarticular onset with an accumulation of >4 affected joints after the first 6 months of disease.
Application of the International League of Associations for Rheumatology (ILAR) criteria for oligoarthritis in JIA (16) would have excluded 28 of the 205 patients for the following reasons: 2 male patients were HLA–B27 positive and had disease onset after the age of 8 years, 6 patients had a family history of HLA–B27–associated disease, and 22 patients had a family history of psoriasis. Many patients (n = 80) were not tested for rheumatoid factor within the first 6 months of disease, and only a few were tested twice for rheumatoid factor within the first 6 months of disease. A diagnosis of psoriasis was not verified by a dermatologist in many patients. For these reasons, the ILAR criteria could not be used for this study. Since rheumatoid factor is a well-described risk factor for poor outcome, 5 rheumatoid factor–positive patients were excluded to improve the homogeneity of the patient population.
Several clinical and demographic features identifiable in the first 6 months were considered as potential predictors of an extended disease course (Table 1). Some of these factors have previously been identified as predictors of outcome for all onset subtypes of JRA (3) or for oligo-JRA (5). Symmetry of joint involvement was defined as bilateral involvement of any joint in the first 6 months of disease; this definition was used in the study by Ruperto et al (3). Asymmetric disease was therefore defined as arthritis that in no instance affected the same joint of both sides of the body. Hand disease (3) and upper limb disease (5) have been described as predictors of poor outcome. In the present study, we defined hand involvement as arthritis of the small joints of the hand. We did not use upper limb disease as a possible predictor, but examined the wrist and the hand separately. Erythrocyte sedimentation rates (ESRs) were determined and tests for antinuclear antibodies (ANAs) were performed routinely in hospital laboratories by conventional methods, although neither of these tests was performed routinely in the first 6 months of disease. When several tests for ESR had been performed, the highest value in the first 6 months was recorded. HLA testing was performed using standard typing sera and microcytotoxicity assays.
Table 1. Independent variables and their respective possible values assessed within the first 6 months of illness*
|Sex (female, male)|
|Age at onset (<6 years, ≥6 years)|
|Disease duration before diagnosis (<1 year, ≥1 year)|
|Symmetric disease (yes, no)|
|Subtype onset (1 joint, 2–4 joints)|
|Ankle disease (yes, no)|
|Wrist disease (yes, no)|
|Ankle and/or wrist disease (yes, no)|
|Hand disease (yes, no)|
|ESR (<20 mm/hour, ≥20 mm/hour)|
|ANAs (positive, negative)|
|HLA (present or absent) for each of the following: A1, A2, B27, B35, C3, DR1, DR4, DR5, DR6, DR7, DR8|
The principal outcome measure was disease extension with a cumulative joint count beyond 4. For individual patients, a change in joint count with extension after 6 months to 5 or 6 involved joints may have had limited clinical significance in the patient who already had 4 involved joints. Clinically meaningful extension was therefore arbitrarily defined, for the purposes of this study, as extension to ≥10 joints; this joint count was said to define more severe disease in at least one previous report (17), and at a joint count of 10, all patients would have at least doubled their joint count. We estimated that about one-half of the patients whose disease extended beyond a count of 4 joints would fall into this category, and therefore, this number would also enable a meaningful analysis.
Other recorded measures of course and outcome were also defined categorically and dichotomized as shown in Table 2. Disease-modifying antirheumatic drugs (DMARDs) used at any time during the disease course included gold, methotrexate, hydroxychloroquine, and sulfasalazine. Remission was defined as the absence of subjective or objective evidence of active arthritis (warmth, swelling, erythema, or pain at the end of range of motion) in patients who were receiving no medication for at least 2 years. The physician's global assessment (PGA) of disease activity was recorded at the last clinic visit using a scale of 0–4, where 0 indicates no disease activity and 4 indicates very severe disease activity. Radiographic examinations of affected joints had been performed when clinically indicated at any time during routine followup at the discretion of the examining physician. If radiographs were obtained, the presence of erosions was noted.
Table 2. Outcome variables at last followup visit and their respective possible values*
|Disease extension to polyarticular course|
| >4 joints (yes, no)|
| ≥10 joints (yes, no)|
|Any DMARD use (yes, no)|
|Any remission occurred (yes, no)|
|Erosive changes on radiographs (yes, no)|
|Physician's global assessment of disease activity at last visit (0, ≥1)|
|C-HAQ/HAQ score (≤0.12, >0.12)|
The Childhood Health Assessment Questionnaire (C-HAQ) (18) and the HAQ (19) for patients older than 18 years were used to measure health status at followup. Disability was scored from 0 to 3, where 0 indicates no difficulty and 3 indicates the inability to perform activities of daily living. Median cutoff scores for “no disability” versus mild, moderate, and severe disability have also been described, and mild disability only becomes apparent at scores of >0.12 (18). Relatively few patients completed the C-HAQ or HAQ, since these questionnaires were only administered routinely during one 2-year period, which ended 1 year prior to data collection for this study. Therefore, for this dependent variable alone, the time for outcome assessment was 0–3 years earlier in each patient, depending on the date of their last followup assessment.
Continuous variables were categorized, and both Fisher's exact test and odds ratios (ORs) were calculated in univariate analysis for statistical significance. Variables identified as being associated with clinically significant disease extension (extension to ≥10 joints) by univariate analysis were then entered into a forward stepwise multiple logistic regression model to predict extension. Disease duration, although not a predictor variable identifiable in the first 6 months, was included as an independent variable in the multivariate predictive model, since this was a likely confounding factor. These same variables were then examined in separate models to examine other measures of outcome. Analysis was performed with SPSS software version 10.1 (SPSS, Chicago, IL).
- Top of page
- PATIENTS AND METHODS
The median age at onset of arthritis in the 205 study patients (160 females, 45 males) was 4.9 years (range 0.9–15.7 years). Median followup from diagnosis was 10.8 years (range 5–26.6 years). Eighty-one patients (39.5%) developed extended oligoarthritis, and 36 patients (17.6%) developed arthritis in ≥10 joints. In the first 6 months, 67 patients (33%) had ankle disease, 19 (9%) had wrist disease, 83 (40%) had ankle and/or wrist disease, 3 (1.5%) had both ankle and wrist disease, 14 (7%) had hand disease, and 34 (17%) had upper limb disease (shoulder, elbow, wrist, or hand). Only 4 patients with upper limb disease had neither wrist nor hand disease. Four patients (2%) had small joint disease of the feet (toes). Of the 18 patients with small joint disease of the hands or feet, only 1 patient had monarthritis and only 2 patients had >1 small joint affected. Knee arthritis was present in 174 patients (85%); 78 had monarthritis of the knee (86% of all 91 patients with monarthritis), 20 had arthritis in 1 knee in addition to other joints, 28 had arthritis limited to both knees, and 48 had both knees involved in addition to other joints. Fifty-nine patients (29%) had a symmetric pattern of arthritis in the first 6 months. Data were complete for ESRs in 142 patients, ANAs in 175 patients, and HLA typing for class I and II antigens in 141 patients. Joint radiographs were available for 120 patients, PGAs of disease activity for 197 patients, and C-HAQ/HAQ scores for 87 patients.
In 19 patients (9%) who initially fulfilled diagnostic criteria for oligo-JRA, the disease evolved into other forms of arthritis at a mean of 5.3 years (range 2–16 years) after onset. These patients' new diagnoses were SEA syndrome (n = 1), juvenile AS (n = 3), and psoriatic arthritis (n = 15; 2 possible, 3 probable, and 10 definite). In 4 of these 19 patients, the disease followed a persistent oligoarticular course and in 15, an extended course (the disease in 6 of these 15 patients extended to involve ≥10 joints). To be clinically useful, predictors within the first 6 months should include all patients with JRA, regardless of any subsequent change of diagnosis. Nevertheless, statistical analysis was performed initially including, and subsequently excluding, these 19 patients.
Clinical features in first 6 months of oligo-JRA associated with an extended disease course (univariate analysis) (Table 3).
Ankle disease, wrist disease, ankle and/or wrist disease, hand disease, symmetric disease, disease of 2–4 joints (versus monarthritis), an elevated ESR, and the presence of ANAs were all significantly associated with an extended disease course with >4 joints involved when examined by univariate analysis (Table 3). Sex, age at onset, and duration of disease before diagnosis were not associated with extension (data not shown). Except for the presence of ANAs and hand disease, all factors significantly associated with extension beyond 4 joints were also associated with extension to ≥10 joints (Table 3).
Table 3. Clinical features in first 6 months of oligoarticular-onset juvenile rheumatoid arthritis that are significantly associated with disease extension (univariate analysis)*
|Predictor variable (no. of patients)||Disease extension|
|To >4 joints||To ≥10 joints|
|Yes||No||OR (95% CI)||Yes||No||OR (95% CI)|
|Ankle disease (205)|| || || || || || |
| Yes||38||29||2.89 (1.58–5.3)†||18||49||2.45 (1.18–5.09)‡|
| No||43||95|| ||18||120|| |
|Wrist disease (205)|| || || || || || |
| Yes||16||3||9.92 (2.79–35.3)§||10||9||6.8 (2.5–18.4)†|
| No||65||121|| ||26||160|| |
|Ankle and/or wrist disease (205)|| || || || || || |
| Yes||52||31||5.38 (2.9–9.9)§||26||57||5.11 (2.3–11.33)§|
| No||29||93|| ||10||112|| |
|Hand disease (205)|| || || || || || |
| Yes||11||3||6.34 (1.71–23.5)‡||3||11||1.31 (0.35–4.9)|
| No||70||121|| ||33||158|| |
|Symmetric disease (205)|| || || || || || |
| Yes||36||23||3.51 (1.87–6.6)§||26||33||10.72 (4.71–24.1)§|
| No||45||101|| ||10||136|| |
|No. of active joints (205)|| || || || || || |
| 2–4||63||51||5.01 (2.65–9.45)§||32||82||8.49 (2.87–25.06)§|
| 1||18||73|| ||4||87|| |
|ESR, mm/hour (142)|| || || || || || |
| ≥20||41||31||4.89 (2.32–10.1)§||20||52||4.10 (1.54–10.9)‡|
| <20||15||55|| ||6||64|| |
|ANAs (175)|| || || || || || |
| +||47||59||1.61 (1.04–2.49)‡||22||84||1.79 (0.85–3.79)|
| −||19||50|| ||8||61|| |
No specific HLA was associated with higher risk of progressing to involvement of >4 joints. Data for HLA–DP and HLA–DQ were incomplete and were not examined. Table 4 illustrates the frequencies of certain antigens (selected because of their previous reported associations with oligo-JRA or polyarticular JRA) among all patients compared with the frequencies among subgroups of patients with either an extended articular disease course or a persistent oligoarticular disease course. The frequency of B27 was 10% in both the persistent oligoarticular group and the extended group and was not different from that in the general population (8%). Similarly, the frequencies of DR1 and DR4, which are associated with polyarticular JRA, were not increased in the group as a whole (12% and 10%, respectively) and did not differ significantly between patients who had extension of arthritis and those who did not.
Table 4. Frequency (percentages) of common HLA types in all 141 patients with oligoarticular-onset juvenile rheumatoid arthritis (oligo-JRA), as well as in patients subgrouped as having an extended articular disease course (>4 joints) and persistent oligoarticular disease
|Total (n = 141)||Extended (n = 59)||Persistent (n = 82)|
Clinical features in the first 6 months of oligo-JRA associated with a polyarticular disease course involving ≥10 joints (multivariate analysis).
Factors significant for extension to ≥10 joints that were identified by univariate analysis were examined together in a forward stepwise logistic regression model (Table 5). Although disease duration was not an early predictor, it was included as a confounding independent variable and found to be associated with this outcome. The variables of symmetric disease, ankle and/or wrist disease, and an elevated ESR remained significantly predictive of a polyarticular disease course with ≥10 joints involved. Because the incomplete data for ESR greatly limited the number for analysis in the multivariate model, the calculation was performed first including ESR as an independent variable (analysis 1) and then excluding ESR (analysis 2). The results of this second analysis were similar to those of the first, except that wrist disease alone, which was eliminated in the first analysis, also became a significant predictor.
Table 5. Predictive clinical features in first 6 months of oligoarticular-onset juvenile rheumatoid arthritis that contribute to different measures of outcome using a logistic regression model for each outcome*
|Outcome||Predictor variables (confounder)†||Analysis 1‡||Analysis 2§|
|n||OR (95% CI)||P||n||OR (95% CI)||P|
|Polyarticular course (≥10 joints)||Symmetry||142||19.2 (5.46–67.8)||0.000||205||17.1 (5.6–49.5)||0.000|
| ||Ankle and/or wrist disease|| ||6.61 (1.97–22.1)||0.002|| ||5.03 (1.70–14.9)||0.004|
| ||ESR ≥20 mm/hour|| ||3.76 (1.09–12.9)||0.036|| || || |
| ||Wrist disease|| ||E|| || ||3.91 (1.05–14.6)||0.043|
| ||(Disease duration)|| ||1.18 (1.03–1.34)||0.018|| ||1.26 (1.13–1.4)||0.000|
|DMARD use||Symmetry||142||11.5 (4.22–31.3)||0.000||205||5.98 (2.87–12.5)||0.000|
| ||Wrist disease|| ||5.87 (1.51–22.8)||0.010|| ||E|| |
| ||ESR ≥20 mm/hour|| ||6.47 (2.2–18.9)||0.001|| || || |
| ||Ankle and/or wrist disease|| ||E|| || ||2.62 (1.26–5.43)||0.010|
| ||(Disease duration)|| ||E|| || ||1.09 (1.01–1.18)||0.022|
|Erosive disease||Symmetry||83||4.73 (1.47–15.2)||0.009||120||2.61 (1.03–6.62)||0.043|
| ||Ankle and/or wrist disease|| ||3.59 (1.15–11.2)||0.027|| ||2.55 (1.0–6.51)||0.050|
| ||(Disease duration)|| ||1.19 (1.05–1.37)||0.007|| ||1.19 (1.08–1.31)||0.000|
|PGA ≥1 at last followup visit||Symmetry||135||3.23 (1.45–7.2)||0.028||197||3.10 (1.59–6.1)||0.001|
| ||Wrist disease|| ||4.01 (1.16–13.8)||0.004|| ||2.73 (1.0–7.41)||0.049|
|No remissions||Symmetry||142||4.73 (2.15–10.4)||0.000||205||3.70 (1.95–7.1)||0.000|
| ||ESR ≥20 mm/hour|| ||2.30 (1.04–5.08)||0.039|| || || |
|Disability (C-HAQ score >0.12)||Symmetry||72||2.95 (1.01–8.6)||0.048||87||E|| |
Clinical features in first 6 months of oligo-JRA predictive of additional measures of poor outcome (multivariate analysis).
As for polyarticular-disease-course outcome, the predictor variables for other outcomes were analyzed by multiple logistic regression, first including ESR as an independent variable (analysis 1) and then excluding ESR as an independent variable (analysis 2) (Table 5). Symmetric disease was predictive of all measures of poor outcome: extension ≥10 joints, the need to use DMARDs, radiographic demonstration of erosive disease, inflammatory activity at last followup visit, no remissions, and disability as measured by C-HAQ score >0.12. Ankle and/or wrist disease was predictive of extension and erosions. Wrist disease was predictive of DMARD use (OR 5.87) and of inflammatory disease activity at last followup visit. An elevated ESR was predictive of extension, DMARD use, and no remissions. Disease duration was a confounding variable for extension and erosive disease. The results of the second analysis were similar, with the following exceptions: for DMARD use, wrist disease was eliminated from the predictive model and ankle and/or wrist disease and disease duration were included in the predictive model; for disability, there were no longer any predictors. When the 19 patients whose diagnosis changed during the disease course were excluded from the multivariate analysis, there was almost no change in the predictor variables as summarized in Table 5, except that there were no longer any predictors of disability (data not shown).
- Top of page
- PATIENTS AND METHODS
This study demonstrates that in children with oligo-JRA followed up for >5 years, progression to an extended course with >4 involved joints occurred in 39.5% of children, and extension to ≥10 joints occurred in 17.6%. There are many measures of outcome other than “extension,” and multiple clinical, demographic, therapeutic, and genetic factors may influence or predict these outcomes. In this study, using multivariate analysis, we determined that for children with oligo-JRA, the early patterns of joint involvement, either symmetric or ankle/wrist involvement, were strongly predictive of extension and other deleterious outcomes, including the need to use DMARDs, the presence of erosive joint disease, and disease activity at followup.
There are several inherent limitations of this retrospective study. It is impossible to assign current classification criteria retrospectively when data are incomplete. Nonetheless, in an attempt to allow better comparison with future studies, we have defined our study population to resemble the oligoarthritis subset of patients with JIA. This more homogeneous group excludes patients with juvenile psoriatic arthritis, spondylarthropathy-like diseases, and rheumatoid factor positivity. During the 30 years in which these patients have been gathered, there have been many therapeutic advances, including new DMARDs and increasing use of intraarticular injection with triamcinolone hexacetonide. It is not known specifically how these medications influence the prognosis with respect to extension in oligo-JRA, but it is likely that they would be confounding factors. We did not attempt to include this multifaceted shift of therapy as a confounder.
Other reports describe extension in 10–50% of children with oligoarthritis (JRA, JCA, or JIA) (2, 5, 9, 10, 20), but these studies included patients followed up for <5 years. Patients with disease resolution within 5 years are more likely to have been lost to longer followup, and not including this patient population may have been a bias in the present study. Disease duration is not a predictor variable, but it was a significant confounding independent factor associated with an increased likelihood for the arthritis to extend. This continuing accumulation of affected joints with time, however, is contrary to the notion that if extension to a polyarticular disease course is to occur, it will occur in the first 2–3 years after onset. Guillaume et al (5) reported that only 30% of patients who extended did so in the first 2 years. Time to extension was not documented for this study; however, 140 of the study patients were included as part of a larger Canadian Multicenter Study of outcome for JRA (11). For the oligoarthritis patients in the Canadian Multicenter Study, the median time to accumulate 5 joints was 3.9 years after onset (range 0.7–15.7 years), and extension occurred within 2 years of onset in only 21%.
Symmetry may easily be overlooked in the first 6 months of disease if <5 joints are involved, but, as in the present study, it can be simply defined and identified. We observed that early symmetric joint arthritis was not only predictive of polyarticular disease extension, but also associated with a poor outcome as manifest by erosive disease, the need to use DMARDs, and the development of disability as measured by the C-HAQ. Ruperto et al (3) also demonstrated that symmetric arthritis had a strong ability to predict future disability for all patients with JRA, but those investigators did not distinguish the oligo-JRA population. In the study by Guillaume et al, symmetry (undefined) occurred in only 13% of patients, and no associations with outcome were described (5). A symmetric pattern of joint involvement characterizes the course of polyarticular JRA, and we surmise that for oligo-JRA, a symmetric pattern, rather than the total number of joints in the first 6 months, may better characterize the likely long-term disease course.
In this study, ankle and/or wrist disease was highly associated with polyarticular extension and erosive disease. Wrist disease in particular was also predictive of the need to use DMARDs and the presence of active disease at the last followup visit. There are limited studies that indirectly support these findings. Gare and Fasth (21) showed that the presence of ankle joint disease was associated with more active disease, although the issue of disease extension was not addressed. Ravelli and colleagues (22) showed that among patients with oligo-JRA, those without wrist joint disease were more likely to respond to treatment with methotrexate. Cassidy et al (23) clustered lack of wrist involvement with lack of progression to polyarthritis in a subset of oligo-JRA patients. Both the wrist and ankle represent joint complexes, rather than single joints, and it could be argued that disease of either of these joints already represents “polyarticular” disease. Unlike the study by Ruperto et al (3), which included all types of JRA, the present study did not find that hand involvement was associated with a poor outcome. Our study findings support those of Guillaume et al (5), who describe early upper limb disease as being predictive of a poor outcome. Eighty-eight percent of our patients with upper limb disease at onset had either wrist or hand disease. Hand disease was not associated with disease extension; therefore, we suspect that the wrist was the involved upper limb joint that was primarily associated with poor outcome.
The only other variable predictive of poor outcome was an elevated ESR; this was predictive of polyarticular extension, the need to use DMARDs, and the failure to achieve remission of disease. Early elevation of ESR has not been uniformly associated with poor outcome; the association has been described for a JRA population with all types of onset (3) and for an oligoarticular-onset JIA population (5), but not for a JCA population (21). A companion report to the study published in reference 21 by Gare and Fasth (2) was unique among those already discussed here in demonstrating that female sex was predictive of poor outcome with continuing disease activity.
In the present study, HLA data did not differentiate patients with oligo-JRA who were likely to have a polyarticular course of disease. In their study of JRA of all onset types, Ruperto et al (3) did not find any HLA associated with any poor outcome. They did find an association of HLA–DR5 with the symptom of pain; they also described HLA–B5, HLA–DR3, and HLA–C3 as being protective against pain. We did not consider pain as an outcome measure, and none of these antigens were associated with a polyarticular disease course. HLA–DR1 and HLA–DR4 have been associated with polyarticular-onset disease (more particularly with rheumatoid factor–positive disease) (24), and HLA–DR5 and HLA–DR8 have been associated with oligo-JRA (25); none of these antigens were associated with extension, and rheumatoid factor–positive patients were excluded from our group. In our population, the lack of HLA association with polyarticular outcome suggests that other nongenetic factors or genetic factors other than those defined by HLA are more important determinants of a polyarticular disease course.
In summary, we have shown that oligo-JRA is not an invariably benign disease, since polyarticular extended disease occurred in more than one-third of our patients. The early pattern of joint disease and an elevated ESR are statistically associated with a poor outcome. In practical terms, the early presence of ankle and/or wrist disease, symmetric involvement of joints, and an elevated ESR in a child with oligo-JRA indicates the likelihood of disease progression and should prompt early introduction of DMARDs. Prospective studies should evaluate homogeneous groups as defined by the ILAR classification and examine outcome with the new instruments that measure not only functional disability, but also quality of life.