To analyze inflammatory parameters as possible predictors for the development of uveitis in juvenile idiopathic arthritis (JIA) patients. Further, to analyze the predictive value of demographic and clinical factors at the onset of arthritis.
Retrospective cohort study.
In 358 children with oligoarthritis and rheumatoid factor–negative polyarthritis, erythrocyte sedimentation rate (ESR), C-reactive protein, leukocyte count, presence of antinuclear antibodies (ANA), presence of human leukocyte antigen (HLA-)B27, age of onset of JIA, and sex were analyzed for their predictive value for the onset of uveitis.
One hundred forty-seven patients (41%) were diagnosed with chronic anterior uveitis. Young age of onset, presence of ANA, and elevated ESR appeared to be predictive factors according to univariate analyses ( P = .029, P = .007, and P = 5E −4 , respectively). According to multivariate analysis, young age of onset and elevated ESR appeared to be predictive after adjusting for the other relevant factors ( P = .004 and P = .001, respectively). A prediction model was developed.
Elevated ESR appears to be a predictor for the occurrence of uveitis in patients with JIA. Since ESR is already routinely tested in patients with recently diagnosed arthritis, its use as a biomarker can easily be implemented in daily practice.
Juvenile idiopathic arthritis (JIA) is the most common rheumatic disease in childhood and is defined as arthritis without a known etiology that begins prior to the age of 16 and persists for at least 6 weeks. Uveitis, typically chronic anterior uveitis, is the most common extra-articular manifestation in patients with JIA and JIA is the most common systemic association of uveitis in children. The JIA subtype with the highest association with uveitis is oligoarthritis, followed by rheumatoid factor (RF)-negative polyarthritis, with an incidence of 13%–45% and 10%, respectively.
The severity of uveitis at presentation is known to predict a severe course and worse visual outcome. Since the course of chronic anterior uveitis is asymptomatic, routine ophthalmologic examination is required in patients with JIA and early treatment is critical to prevent visual loss. Despite early detection, aggressive autoimmune disease can cause harmful uveitis with a worse outcome as well. Complications associated with poorly controlled or untreated uveitis include posterior synechiae, cataract, glaucoma, cystoid macular edema, and band keratopathy.
Among patients with JIA, the oligoarticular subtype, antinuclear antibodies (ANA) positivity, young age of onset, and female sex in early-onset arthritis are predictive factors for the development of uveitis. Identification of more predictors can help to improve screening protocols for routine examination in order to focus even more on those with the highest risk and to protect them from visual loss.
The aim of this study is to analyze inflammatory parameters and demographic and clinical factors at the onset of arthritis as possible predictors for the development of uveitis among patients with JIA.
Patients and Methods
A retrospective cohort study of 358 patients with JIA visiting the ophthalmologist or the pediatric rheumatologist at the University Medical Center of Utrecht, Leiden and Groningen in the Netherlands was performed. Only patients with the JIA subtypes oligoarthritis and RF-negative polyarthritis were included. The JIA diagnosis based on the criteria of the International League of Associations for Rheumatology was confirmed by a pediatric rheumatologist and all patients with JIA were screened by an ophthalmologist at least as many times as recommended by the guidelines of the American Academy of Pediatrics. Patients with onset of uveitis before arthritis were excluded. Also, patients with uveitis entities other than chronic anterior uveitis were excluded. Patients were divided into 2 groups: JIA patients with (Group 1) and without (Group 2) uveitis. The patients in the second group had an ophthalmologic follow-up of at least 4 years without signs of uveitis. The collection of data from patients’ medical charts for the research goals as described in this article was approved by the Institutional Review Board of the Utrecht University Medical Center and is in compliance with the Helsinki principles.
The values of the inflammatory parameters erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), and leukocyte count at the time of diagnosis of JIA were collected from patients’ medical charts. Only results of patients with inflammatory parameters tested 12 months or less prior to the diagnosis of JIA, 12 months or less after diagnosis of JIA, and before the start of systemic immunosuppressive therapy were included in the analyses. Information about the presence of ANA at the time of diagnosis of JIA was collected. Additional patient characteristics including date of birth, sex, date of onset of JIA, date of onset of uveitis, and presence of human leukocyte antigen (HLA-)B27 were collected from the medical charts. The age of onset of JIA was calculated from the date of birth and date of onset of JIA. The date of onset of JIA and the date of onset of uveitis were used to calculate the time interval between JIA and uveitis.
In some medical charts it was unclear whether the patient had ever been diagnosed with uveitis. Therefore, a questionnaire was sent to 42 patients with JIA, who were screened by an ophthalmologist outside the University Medical Centers. Three patients confirmed the uveitis diagnosis; these patients or their parents were asked to sign an informed consent in order to make a copy of the ophthalmologist’s patient’s medical chart available. This chart was additionally checked for the presence of uveitis, to be sure all patients would be placed in the correct group (uveitis vs non-uveitis). The collected additional information of all these 42 patients was obtained from the medical charts of the University Medical Centers.
Statistical analyses were performed with IBM SPSS Statistics version 20 for Windows and the R rms package. Univariate analyses were performed to find independent predictors that might enter the multivariate analysis. The Pearson χ 2 test or Fisher exact test was applied for univariate analysis of categorical variables. Since there were no normally distributed variables, confirmed by the Kolmogorov-Smirnov-test, the Mann-Whitney U test was used for continuous, abnormally distributed variables. Subgroup analyses were performed for patients with RF-negative polyarthritis and oligoarthritis. Statistically significant variables in univariate analysis as well as previously known predictors were selected for multivariate analysis by logistic regression. The continuous variable age of onset of JIA was dichotomized according to clinical standards based on prior literature with an age of 6 years as a cutoff point. P values ≤.05 were regarded as statistically significant. For presentation, medians were used for the abnormally distributed variables. Based on the results of the multivariate analysis and the current screening guidelines for uveitis in patients with JIA (which includes ANA and age of onset of JIA), a prediction model was developed. To test the ability to discriminate between patients with and without uveitis, the area under the receiver operating characteristic curve was determined. When prediction models are derived from multivariate regression analyses, overestimation of regression coefficients is a known phenomenon that results in too extreme predictions in new patients. Therefore, internal validation with bootstrapping techniques was applied, which resulted in a shrinkage factor for the regression coefficients. Also, the value for the area under the receiver operating characteristic curve was corrected for optimism using the bootstrap procedure.
General Characteristics of Study Population
From a total of 358 patients with oligoarthritis and RF-negative polyarthritis entering the study, 147 (41%) were diagnosed with chronic anterior uveitis. All of these patients had at least 1+ cells in the anterior chamber at consecutive visits that needed treatment with topical steroids or immunomodulating medication. The median time between JIA onset and uveitis onset was 1.0 year (range 0.0–24.3 years). Fifty percent of the patients developed uveitis within the first year after onset of JIA, 66% did so within the first 2 years, and in 85% uveitis occurred within the first 4 years. In the majority of the patients the inflammatory parameters were tested at the moment of JIA diagnosis (median time between diagnosis of JIA and laboratory records: 0 months, interquartile range: 0–0 months). The proportion of subtypes of JIA in the group of patients with uveitis could be compared to that in the group of patients without uveitis, with oligoarthritis being the most common subtype. Likewise, the female-to-male ratio was similar in both groups ( Table 1 ).
|Sex, no. (%)|
|No. patients||147||211||.394 a|
|Female||112 (76)||151 (72)|
|Male||35 (24)||60 (28)|
|JIA subtype, no. (%)|
|No. patients||147||211||.282 b|
|Oligoarthritis||108 (74)||141 (67)|
|Polyarthritis||39 (26)||70 (33)|
|Age (y) of onset JIA|
|Median (range)||2.7 (0.9–9.5)||3.1 (0.7–15.0)||.029 c , d|
|ANA, no. (%)|
|Positive||113 (77)||135 (64)||.007 a , d|
|Negative||33 (23)||76 (36)|
|HLA-B27, no. (%)|
|Positive||9 (17)||11 (15)||.811 a|
|Negative||45 (83)||62 (85)|
|Median (range)||32 (2–150)||23 (2–115)||5E −4, d|
|Median (range)||11.0 (0–136)||6.0 (0–303)||.053 c|
|Leukocyte count × 10 9 /L|
|Median (range)||9.4 (4.5–109)||9.32 (3.8–18.9)||.749 c|
Age of Onset
The median age of onset of JIA appeared to be statistically different between the 2 groups. The median age was 2.7 years in patients with uveitis and 3.1 years in patients without uveitis ( P = .029).
Antinuclear Antibodies and Human Leukocyte Antigen B27
Patients with uveitis were significantly more often ANA-positive compared to patients without uveitis ( P = .007). Presence of HLA-B27 did not entail a significant difference between the 2 groups.
Statistically, ESR at the time of diagnosis of JIA was significantly more elevated in the uveitis group (median 32 mm) compared to the non-uveitis group (median 23 mm; P = 5E −4 ; Table 1 ). Also, in the oligoarticular and polyarticular subgroups, ESR was significantly more elevated in patients with uveitis. In patients with oligoarthritis, the median value was 30 mm in patients with uveitis and 20 mm in patients without uveitis ( P = .008). In patients with polyarthritis, the median value was 40 mm in patients with uveitis and 27 mm in patients without uveitis ( P = .010). After analyzing ESR exclusively in patients who developed uveitis in the first year after onset of JIA, there appeared to be a strong statistical difference between the ESR in the 2 groups with 39 mm in the uveitis group and 23 mm in the non-uveitis group ( P = 1.7E −5 , uveitis n = 55, non-uveitis n = 197). In the second (n = 18, ESR = 30 mm), third (n = 10, ESR = 26 mm), and fourth (n = 11, ESR = 19 mm) year after JIA diagnosis, ESR was slightly elevated in the second and third year in the uveitis group. In these 3 groups there was no statistically significant difference compared to the non-uveitis (n = 197, ESR = 23 mm) group ( P = .318, P = .839, and P = .493, respectively). The inflammatory parameters CRP and leukocyte count at the time of diagnosis of JIA did not statistically differ between the 2 groups ( Table 1 ).
Multivariate Analysis and Prediction Model
Because of the statistically significant outcomes in univariate analysis, ANA, age of onset of JIA, and ESR were selected for multivariate analysis. Additionally, the factors sex and JIA subtype were selected. Adjusted for age of onset of JIA, ANA, sex, and JIA subtype, ESR appeared to be a statistically significant predictor for the occurrence of uveitis in patients with JIA with an odds ratio (OR) of 1.016 (95% confidence interval [CI] 1.006–1.026, P = .001), which means that for each elevation of 1 mm ESR, the odds for the occurrence of uveitis increase by 0.016 ( Table 2 ). Onset of JIA before the age of 7 years was, adjusted for ANA, ESR, sex, and JIA subtype, a statistically significant predictive factor for the occurrence of uveitis in patients with JIA with an OR of 3.167 (95% CI 1.432–7.006, P = .004; Table 2 ). The predictors ANA, age of onset, and ESR were included in the prediction model ( Figure 1 ). The area under the receiver operating characteristic curve of the model was 0.644 (95% CI: 0.582–0.706; Figure 2 ).