Incidence and predictors of Uveitis in juvenile idiopathic arthritis in a Nordic long-term cohort study

In this longitudinal study, children with long-term ophthalmologic follow-up were selected from a prospective multicentre Nordic JIA cohort. The original multicentre cohort consisted of consecutive children with newly diagnosed JIA and disease onset between January 1, 1997 to June 30, 2000 from defined geographical areas of Denmark, Finland, Norway and Sweden aiming for the cohort to be population-based [20]. Clinical information including family history, assessment of disease activity, self-reported questionnaires and information from ophthalmology visits was registered per protocol at baseline and at 1 to 3 year intervals. Detailed description of the methods of patient accrual and data collection has been published previously [21].

At the final study visit, at least 7.0 years (median 98 months, range 84–147 months) after disease onset, extended data including information from the most recent ophthalmologic visit was collected. The interval of ophthalmologic visits was every 2–3 months the first 2 years, then longer intervals according to time after disease onset and JIA category. The intervals were in line with the guidelines for uveitis screening at each center, based on international uveitis screening guidelines. Uveitis was registered as present and active, when the ophthalmologist prescribed treatment for inflammation in the uvea. Uveitis was classified according to the recommendations by the Standardization of Uveitis Nomenclature (SUN) working group [22]. Uveitis was classified as “acute”; symptomatic with acute onset, sometimes recurrent course but of limited duration, or “chronic”; insidious onset, mainly asymptomatic with chronic recurrent or persistent course. Collected data also included locally analyzed HLA-B27, antinuclear antibodies (ANA) and rheumatoid factor (RF), and a positive result of the latter two had to be confirmed, at least 3 months apart. Each physician interpreted the results of ANA and RF analysis as positive or negative according to the reference values of their local laboratory; the cut-off value for a positive test was ≥ 1/80 in Tromsø, Trondheim and parts of Sweden, ≥ 1/320 in Finland, and ≥ 1/160 in Copenhagen, Århus and some parts of Sweden. ANA was measured using immunofluorescence on HEp-2 cells. Serum was taken at a median of 4 (1st quartile of 2, 3rd quartile of 7) months after the disease was diagnosed, frozen and stored at −70 °C in aliquots to avoid repeated freeze-thaw cycles. AHA were detected with Varelisa Histone Antibodies Enzyme Induced Assay (EIA) kit for antihistone IgM/IgG antibodies using a combined IgM/IgG conjugate (Phadia Diagnostics, Freiburg, Germany) at UIT The Arctic University of Norway, Tromsø, Norway. Tests were run according to the manufacturer’s instructions [23]. According to the kit instructions a level ≥ 30 U/ml is considered positive, 15–29 U/ml borderline, and < 15 U/ml negative, based on a test population of 432 healthy adults, normative values are not given for children. In the present study, we investigated ≥ 15 U/ml as a positive threshold based on our previous study reporting a mean concentration of 4.3 U/ml in 58 healthy children utilizing the same Varelisa Histone EIA-kit [11]. The presence of AHA in a Norwegian JIA cohort, also including the Norwegian subsets of this Nordic JIA cohort, has previously been reported and was therefore not included here [11]. No sera were available from the Finnish participants, and subsequently analyses of the AHA were performed only in the Swedish and Danish sera.

The study report followed the STROBE guidelines [24]. Statistical analyses were performed using the STATA version 14 software (STATA Corp., College Station, TX, USA). Descriptive statistics were used to summarize clinical characteristics of the population and disease activity measures. The chi-square test was used as appropriate for comparison of categorical variables. The Student’s t-test was used to compare means in continuous outcomes between groups, and Mann-Whitney rank sum test for comparison of medians for skewed data. A logistic regression analysis of baseline characteristics was performed to calculate odds ratio for developing uveitis. The p-value <0.05 was considered significant.

Uveitis developed in 89 (20.5%) of the 435 children. Clinical characteristics at baseline according to the presence of acute and chronic uveitis is shown in Table 1. Among the 9 children with acute uveitis, 8 had enthesitis-related arthritis (ERA), and 7 were HLA-B27 positive. Chronic insidious uveitis developed in 80 (18.4%) children. The cumulative incidence of uveitis according to the International League of Associations for Rheumatology (ILAR) categories is shown in Table 2. There were no uveitis in the systemic nor in the polyarticular RF positive categories, and highest incidence was noted in the juvenile psoriatic arthritis (35.7%) and enthesitis-related arthritis (25.0%) categories. The incidence of uveitis was significantly higher in Finland than in the other Nordic countries (Table 3). Finland also had the highest proportion of children with young age at onset of JIA, but neither gender, oligoarticular onset nor the prevalence of HLA-B27, ANA, or AHA showed any corresponding difference between countries.

Time from onset of arthritis to onset of uveitis is shown in Fig. 1, including 5 children in which uveitis was diagnosed before the arthritis, one of these had acute uveitis. Uveitis developed within the first year after onset of arthritis in 48 (53.9%), within 4 years in 73 (82.0%) of the children, and 4 children (4.5%) had onset of uveitis more than 8 years after onset of arthritis, 2 of these had chronic asymptomatic uveitis. The number of children developing uveitis within the first year after onset of arthritis is 6 times the number developing uveitis during the second year, and 16 times the number during the fifth year. All together uveitis developed with an interval of median 0.8 (range − 4.7 to 9.4) years after onset of arthritis, while chronic uveitis developed with a median interval of 0.7 (range − 3.0 to 8.6) years after onset of arthritis.

Young age at onset of JIA and ANA were significant predictors of uveitis (Tables 4 and 5). AHA ≥ 15 U/ml was found in 23 (17.2%) and AHA ≥ 30 U/ml in 10 (7.5%) of the 134 participants with AHA analyses. AHA ≥ 30 U/ml was found in 3 of the 5 females with uveitis and AHA analyses. The presence of AHA ≥ 15 U/ml was significantly higher in children with uveitis than in children without (OR 4.8 (1.8 to 13.4)) (Table 4). Serum concentration of AHA IgM/IgG antibodies was significantly higher in the group with uveitis (mean 19.2 U/ml) than in children without uveitis (mean 10.2 U/ml) in the Swedish and Danish cohort with available sera for antihistone analyses (p = 0.002; chi-square, data not shown).

Neither gender, oligoarticular onset nor other ILAR categories predicted uveitis. Presence of HLA-B27 was borderline significant as a predictor of uveitis, primarily because there was a high presence of HLA-B27 among children with acute uveitis (Tables 1 and 4). There was no significantly reduced risk of developing uveitis in children that had started disease-modyfing antirheumatic drug (DMARD) treatment, including biologics, neither at the first (data not shown) nor at the second study visit (Tables 1 and 4), taking into account the start date of DMARDs and the date of diagnosis of uveitis. At the final study visit the percentage of children ever using DMARDs was higher in the uveitis group (78.7%) compared to the group without uveitis (53.2%) (data not shown), but the number of children receiving DMARDs to treat primarily their uveitis is not known. These analyses on DMARDs were also performed for methotrexate, showing the same result. There were no significant difference in incidence of uveitis in children that had used etanercept (18.5%) during the disease course versus those that never used etanercept (20.7%), p = 0.706. Also the incidence of uveitis was similar in children that had used sulfasalazine (16.1%) versus those that never used sulfasalazine (21.2%), p = 0.361. During the disease course, 14.3% of the cohort used sulfasalazine, while 12.4% used etanercept. The main predictors of uveitis in Table 4 were stratified for gender (Table 5). This analysis showed that girls with uveitis had a lower median age at onset of arthritis of 3.4 versus 6.8 in girls without uveitis (p < 0.001). The difference in age at onset was highly significant in girls, but not in boys (p = 0.390) (Tables 4 and 5). ANA was a significant predictor, both in girls and in boys (Table 4). Analysis stratified for gender showed that median median AHA was significantly higher in girls (p = 0.003), but not in boys (p = 0.794) (Table 5).