Genetic Risk Scores May Facilitate Clinical Diagnosis of Juvenile Idiopathic Arthritis

boy pain sitting on the couch in the knee
Researchers evaluated the use and performance of genomic risk scores for the clinical diagnosis of juvenile idiopathic arthritis.

Genomic risk scores (GRSs) may help in the diagnosis of juvenile idiopathic arthritis (JIA) and its subtypes, according to study results published in Annals of the Rheumatic Diseases. Developed and validated using data from 3 nationwide cohort studies, researchers noted that GRSs for JIA were able to reliably distinguish between patients with JIA from those without.

While previous literature has supported the heritability of JIA, GRSs are not yet being used for the diagnosis of JIA. Instead, JIA is identified purely through clinical presentation, which can vary between patients.

To assess the predictive capacity of GRSs for JIA, the researchers extracted data from 3 case control cohorts conducted in Australia, the United Kingdom, and the United States. All cases had received a clinical diagnosis of JIA from a rheumatologist; control participants were selected from the general population. Single nucleotide polymorphism genotyping had been performed for all 3 cohorts. Lasso-penalized linear models were used to create the GRS for JIA and its subtypes. Researchers trained GRSs in the UK cohort and validated them in the Australian and US cohorts. The area under the receiver operating characteristic curve (AUC) was used to assess discriminatory capacity. Logistic regression models were used to calculate the odds ratio (OR) of correctly distinguishing a case from a control participant for each standard deviation of GRS.

The JIA GRS had an AUC of 0.670 in the UK cohort, indicating an acceptable ability to distinguish between cases and control participants. Externally-validated AUCs were 0.657 and 0.671 in the US and Australian cohorts, respectively. In logistic regression models, the ORs of having JIA per increasing standard deviation of GRS were 1.831 (95% CI, 1.685-1.991) and 1.838 (95% CI, 1.686-2.007) in the US and Australian cohorts, respectively. For each standard deviation increase in GRS, the odds of being a case increased 2-fold. These estimates were not attenuated by adjustments for sex or the top 10 genetic principal components.

The strongest predictive capacity was observed in the GRS for enthesis-related JIA, which had AUC values of 0.82, 0.84, and 0.70 in the UK, Australia, and US cohorts, respectively. In addition, the GRS for the oligoarthritis subtype also outperformed that of the overall GRS (AUC range, 0.72-0.77).

These data suggest that GRSs may have the ability to facilitate diagnosis of JIA. However, its performance needs to be further assessed in other cohorts before implementation. In addition, the majority of all cohort participants were of European descent, limiting data generalizability.

“[A] diagnostic algorithm based on a JIA-GRS may provide more timely, accessible and reliable means of assessing children with musculoskeletal symptoms who may be JIA cases, thus enabling appropriate triage and referral, facilitating early access to appropriate care, and reducing the pain, complications of the disease and poor long-term health outcomes, due to delayed diagnosis and treatment,” the researchers concluded.


Cánovas R, Cobb J, Brozynska M, et al. Genomic risk scores for juvenile idiopathic arthritis and its subtypes. Published online September 4, 2020. Ann Rheum Dis. doi:10.1136/annrheumdis-2020-217421