Disease-modifying Antirheumatic Drug Use in the Treatment of Juvenile Idiopathic Arthritis: A Cross-sectional Analysis of the CARRA Registry

DISCUSSION

Using cross-sectional data for 2748 children with prevalent JIA enrolled in the CARRA Registry at 51 different clinical sites throughout the US, we observed that 74% of all patients had ever received nonbiologic DMARD and 45% had ever received biologic DMARD in clinical practice. The use of systemic glucocorticoids (38% ever use) and NSAID (51% current daily use) was also common. In addition, we identified several patient factors that were strongly and independently associated with particular medication usage.

We found that a considerable proportion of children with JIA are treated with biologic agents by pediatric rheumatologists in the US. Among children with the systemic arthritis and RF+ polyarthritis categories of JIA, about two-thirds of patients had ever received biologic DMARD. Even among children with the persistent oligoarthritis category and without uveitis, 12% had received biologic DMARD, a practice that has been recommended for refractory disease⁸ but has not been the subject of any controlled studies. To our knowledge, there are not similar published reports of the use of biologic DMARD among all children with JIA from other countries with which to compare our results.

For children without systemic arthritis, the current ACR Recommendations generally specify a variable trial of nonbiologic DMARD prior to initiation of TNF inhibitors⁸. Correspondingly, we observed that the vast majority of children (92%) without systemic arthritis who received TNF inhibitors had also received nonbiologic DMARD. Children who received TNF inhibitors in the absence of nonbiologic DMARD use were significantly more likely to have ERA, suggesting that some pediatric rheumatologists may believe that nonbiologic DMARD are less effective in the treatment of ERA. This opinion may be based, in part, on the fact that nonbiologic DMARD have not been shown to be efficacious in the treatment of adults with ankylosing spondylitis¹¹. Accordingly, the ACR Recommendations specify a lower threshold for the initiation of TNF inhibitors for children with active SI arthritis compared to children without SI arthritis⁸. Nevertheless, when we restricted our analyses of the ever use of biologic DMARD to children with ERA, we did not find a significant association with SI tenderness. The reason for this result is unclear, but it is possible that not all patients with reported SI tenderness had clinically important SI arthritis.

Our results support the importance of the number of affected joints (rather than the ILAR category) in clinical decision-making, as presented in the ACR Recommendations⁸. A history of arthritis of ≥ 5 joints remained strongly and independently associated with biologic DMARD use when controlling for other patient factors. It was also strongly and independently associated with biologic DMARD use among children with ERA and PsA, the ILAR categories that may be associated with more or less than 4 affected joints. There was not a marked difference in the proportion of patients who received biologic DMARD in the extended oligoarthritis versus RF– polyarthritis categories (46% vs 54%; p = 0.053). Also consistent with the ACR Recommendations, the presence of radiographic damage or ACPA was associated with biologic DMARD use. We were unable to assess other prognostic features reported in the ACR Recommendations (e.g., hip or cervical spine arthritis).

TNF inhibitors are not always completely effective or universally tolerated, which may lead to switching among agents for individual patients. In our study, 28% of TNF inhibitor users had received > 1 anti-TNF agent during their disease course. This proportion is higher than about 10% reported from biologics registries in the United Kingdom¹² and the Netherlands¹³, but is lower than the about 35% reported from Finland¹⁴. These differences likely reflect, in part, the relative availability of different biologic agents in the respective countries and the time periods of the studies.

Etanercept was the most commonly received TNF inhibitor, most likely because it was the first TNF inhibitor studied and approved for the treatment of JIA by the FDA². However, infliximab was received by a significant proportion of children with JIA, including those without uveitis or IBD, and has not received an FDA-approved label for this indication. In a randomized clinical trial in JIA, infliximab failed to demonstrate efficacy for the primary endpoint versus placebo⁴, despite convincing evidence of clinical effectiveness during open-label use^15,16,17. We observed that the monoclonal antibody TNF inhibitors are used more among children with uveitis. This medication usage pattern is supported by numerous observational studies^18,19,20, although no randomized studies have been reported. These TNF inhibitor usage patterns suggest that pediatric rheumatologists do not rely solely on the results of controlled clinical trials or FDA-approved labeling when making treatment decisions for children with JIA.

MTX represented the vast majority of nonbiologic DMARD use, and we identified several patient factors that were independently associated with its use. The strongest associations with MTX use were uveitis, ACPA, RF+ polyarthritis, and extended oligoarthritis. Most users of MTX received it through the subcutaneous route at some time in their disease course. This is not surprising because subcutaneous administration of higher doses of MTX has been suggested to be more efficacious than doses typically administered through the oral route in children with JIA²¹. In adults with rheumatoid arthritis, one study found that subcutaneous administration was more efficacious than identical doses of orally administered MTX²². It cannot be known from these data how many children initiated oral MTX and subsequently failed to respond. Based on the results of one survey published in 2007, most pediatric rheumatologists in the US and Canada would have recommended the oral route of administration for children with oligoarthritis in whom they were initiating therapy with MTX²³. In contrast, results from a recent clinical trial suggested that using subcutaneous MTX at 0.5 mg/kg/week (maximum 40 mg) at initiation of therapy for polyarthritis may be a superior approach²⁴. The most appropriate dose and route of administration for the initiation of MTX therapy remains uncertain.

SSZ was used by a minority of patients, most of whom had ERA or concurrent IBD. In the ACR Recommendations, SSZ use was recommended under some circumstances for children with ERA, but was uncertain for children without ERA⁸. LEF has been shown to be efficacious in the treatment of JIA^8,25. Nevertheless, we found that LEF was used sparingly in the treatment of JIA and very infrequently in the absence of prior therapy with MTX. This suggests that LEF was likely reserved for instances of MTX intolerance or failure.

The use of DMARD for children with systemic arthritis demonstrated some associations with patient factors, but we were unable to examine most of the poor prognostic features found in the ACR Recommendations⁸, such as hip arthritis or a 6-month duration of significant active systemic disease. A history of polyarthritis was associated with MTX use, but was not associated with cyclosporine, TNF inhibitor, or IL-1 inhibitor use. Radiographic damage was strongly associated with all DMARD except MTX and likely represents a marker of severe refractory disease.

Despite the widespread use of DMARD, the use of systemic glucocorticoids was common. More than one-third of patients with JIA received systemic glucocorticoids during their disease course and > 20% of children with RF+ polyarthritis or systemic arthritis were current users at the time of enrollment. Nevertheless, there are almost no published studies of systemic glucocorticoids in the treatment of JIA, and consequently the ACR Recommendations remained silent on the appropriateness of their use⁸. Clearly, rigorous studies of the safety and effectiveness of systemic glucocorticoids in the treatment of JIA are needed²⁶.

Uveitis may occur in the context of any of the ILAR categories of JIA, although it is most common among children with oligoarthritis²⁷. In multivariable models, uveitis was strongly and independently associated with nonbiologic and biologic DMARD use. Uveitis disease activity is commonly independent of arthritis disease activity²⁸. This implies that uveitis may frequently be the determining factor in the systemic treatment of children with JIA. Nevertheless, there are no published sizable randomized studies of the systemic treatment of uveitis in children²⁹; clearly, more research about the most appropriate treatment for uveitis is needed.

Our study had limitations. Patients enrolled in the CARRA Registry represent a convenience sample of children with prevalent JIA cared for at pediatric rheumatology centers. It is not known whether children who were not enrolled in the CARRA Registry had different disease severity or received different treatment than children who were enrolled at the same clinical site. However, selection bias in patient enrollment was likely to be idiosyncratic and center-specific and therefore minimized by the large number of contributing centers. The distribution of JIA categories in the CARRA Registry is similar to those found in recently published JIA inception cohorts^30,31, with the notable exception of fewer patients with persistent oligoarthritis. Children with less severe disease (e.g., oligoarthritis) are likely clinically evaluated less frequently, and it is possible that they may have fewer opportunities to be recruited to the registry. In addition, it is likely that children who receive care at pediatric rheumatology centers may have more severe disease than children who receive care elsewhere. Medication histories were not systematically obtained, but were recorded by the local study investigators from several sources, including family report, physician recollection, and limited medical record review. It is not known how this nonsystematic data collection may have influenced our results, including the potential for recall bias. Laboratory and radiographic studies were performed at the discretion of the treating physicians as part of routine clinical care. We accepted the JIA ILAR category as assigned by the treating pediatric rheumatologist and did not attempt to reclassify patients based on the data collected in the registry, although there are recognized difficulties in implementing the ILAR categorization system in the routine clinical setting³². Our cross-sectional study design prevented us from making any causal inferences. For example, it cannot be known from the data whether radiographic damage occurred before or after the initiation of a biologic DMARD. The data did not contain some important clinical factors potentially associated with medication usage, such as the specific joints involved and historical disease activity and severity measures.

We found that nonbiologic and biologic DMARD were frequently used in the treatment of JIA and were associated with several specific patient factors. These associated factors were largely in agreement with published ACR Recommendations for the treatment of JIA. Our study results also highlighted several areas in significant need of further clinical investigation, in particular the appropriate management of uveitis with systemic immunosuppression and the best use of systemic glucocorticoids for the treatment of JIA.