Machine Learning Assists in Targeting Effective Arthritis Treatment for Children
Researchers from the University of Manchester have found that machine learning techniques can aid doctors in identifying children and young people with arthritis who are most likely to benefit from methotrexate, the first-line treatment for Juvenile idiopathic arthritis (JIA). The study, published in the journal eBioMedicine, revealed that only half of the children and young people who receive methotrexate experience its benefits or can tolerate it. This leaves the other half waiting longer for second-line therapies, prolonging their joint pain and other debilitating symptoms.
The application of machine learning enables more precise research into identifying response predictors to methotrexate, including biomarkers. This breakthrough can significantly enhance the ability to forecast the outcomes of drug initiation. The study confirms that approximately one in eight children and young people who start methotrexate show improvements in inflammatory features of the disease but may still have some lingering symptoms. Furthermore, 16% of children taking methotrexate experience slower improvements in disease activity compared to others over time.
Lead author Dr. Stephanie Shoop-Worrall highlighted the importance of machine learning in optimizing treatment decisions for children with arthritis. As of now, giving methotrexate to children who will not be helped by it wastes valuable time, money, effort, and exposes them to potential side effects unnecessarily. However, machine learning allows experts to predict which aspects of a child’s disease will respond positively to the drug, enabling them to start alternative therapies either alongside or instead of methotrexate right away.
The study also sheds light on how clinical trials for childhood-onset arthritis often fail to capture the complexity of the disease. By oversimplifying drug response as either effective or ineffective, key symptoms such as pain may be overlooked, or significant improvements in one aspect of the disease may be disregarded. The research team accessed data from four nationwide cohorts of children and young people who started their treatment before January 2018. Machine learning techniques were used to identify distinct disease patterns following methotrexate treatment, predict clusters, and compare them to existing treatment response measures.
The study identified five clusters of patients with different disease patterns: Fast improvers (11%), Slow Improvers (16%), Improve-Relapse (7%), Persistent Disease (44%), Persistent physician global assessment (8%), and Persistent parental global assessment (13%). These clusters offer a basis for stratified treatment decisions, allowing doctors to tailor therapies to individual patients based on their specific disease features.
Dr. Shoop-Worrall emphasized the need for further investigation into the longer-term impact of slower disease control. The study’s findings demonstrate the utility of machine learning methods in uncovering clusters of children for targeted treatment decisions. This research builds upon previous studies on methotrexate treatment response, highlighting the importance of considering response variability across different disease features within individuals.
The results of this study have the potential to transform the treatment approach for children and young people with arthritis by enabling personalized and effective therapies from the outset. By harnessing the power of machine learning, doctors can make informed decisions that optimize treatment outcomes, alleviate pain, and improve the quality of life for children living with this debilitating condition.
Funding for this research was provided by the Medical Research Council, Versus Arthritis, Great Ormond Street Hospital Children’s Charity, Olivia’s Vision, and the National Institute for Health Research through the CLUSTER consortium.
