Clinical Prediction Models regarding Methotrexate Treatment Response in children with Juvenile Idiopathic Arthritis: A systematic review and internal validation study

Abstract

Juvenile Idiopathic Arthritis (JIA) is the most common rheumatic disease of childhood, affecting 16-150 children per 100,000 worldwide, and approximately 60,000 in Europe alone. The disease is an autoinflammatory disease (i.e., the body’s immune system induces inflammatory responses against itself), characterized by joint inflammation of an unknown cause, persisting for at least 6 weeks in at least 1 joint, with an onset before the age of 16 years. If left untreated, joint inflammation can lead to joint destruction and subsequent permanent disability. Therefore, it is important to properly treat JIA with anti-inflammatory drugs. These drugs are designed to suppress inflammatory responses, limiting or completely alleviating joint inflammation.

Current treatment recommendations for JIA consist of a step-up approach. This approach recommends starting treatment with the most “simple” form of disease modifying anti-rheumatic drugs (DMARDs), namely non-steroidal anti-inflammatory drugs (NSAIDs) and intra-articular corticosteroid injections. Children who do not respond to either of these treatment options are escalated to more intensive DMARDs, the conventional synthetic DMARDs (csDMARDs). The most frequently prescribed csDMARD is methotrexate (MTX), specifically as oral tablet. If inactive disease with csDMARDs is not achieved within 6 months, treatment is intensified with concomitant biologic DMARDs (bDMARDs). These are more effective, more expensive, and can almost exclusively be prescribed in the form of a subcutaneous injection.

Recent studies on long-term disease outcomes and JIA have shown that patients with early attainment of inactive disease exhibit better long-term outcomes. Identification and prediction of which patients will not respond to specific treatment options would be highly valuable for clinical practice. Because MTX is the most frequently prescribed treatment option, this research project focussed on the prediction of MTX nonresponse in children with JIA.

First, a systematic review was conducted to review and critically appraise available prediction models on MTX response in JIA. The review identified methodological concerns in all included studies that developed and/or validated clinical prediction models. Therefore, a second study was conducted to develop new clinical prediction models regarding MTX nonresponse in children with JIA, simultaneously addressing the methodological limitations identified by the systematic review.

The newly developed clinical prediction models demonstrated moderate predictive performance and calibration, comparable to previously developed models based on routinely collected clinical variables. Although the sample size was not sufficiently large to further validate the developed models, the development study provides a methodological foundation for future prognostic research in JIA. In the future, larger sample sizes and added predictive value of potential biomarkers could further improve predictive performance and model validity, and hopefully result in the adoption of clinical prediction models to tailor treatment strategies in JIA.