Ultrasonographic phenotypes of different juvenile idiopathic arthritis subtypes and differentiated strategies for disease activity assessment: a cross-sectional study

Juvenile idiopathic arthritis (JIA) encompasses highly heterogeneous subtypes, and conventional clinical assessment often fails to detect subclinical synovitis. Musculoskeletal ultrasound can sensitively depict synovial hypertrophy, joint effusion and abnormal blood flow signals, yet the ultrasound phenotypes of different subtypes have not been systematically compared. This study aims to clarify the characteristic joint involvement patterns of different JIA subtypes, providing evidence for individualized ultrasonographic assessment in JIA.

A cross-sectional study was conducted on 261 children with JIA (135 girls, 51.7%; median age 9.0 years; median disease duration 21.5 months) admitted to the Children's Hospital of Chongqing Medical University between November 2023 and October 2024, encompassing 5 subtypes: polyarticular JIA (pJIA, n=77), systemic JIA (sJIA, n=67), oligoarticular JIA (oJIA, n=48), enthesitis-related arthritis (ERA, n=36), and undifferentiated JIA (uJIA, n=33). Ultrasonography was performed on 68 peripheral joints according to a standardized protocol. Grayscale (GS) synovitis and power Doppler (PD) signals were assessed using the Outcome Measures in Rheumatology definitions and the semiquantitative scoring system proposed by the Childhood Arthritis and Rheumatology Research Alliance. Examinations and scoring were independently completed by 2 musculoskeletal ultrasound physicians blinded to clinical data. Additionally, 75 GS and 75 PD static images were randomly selected from 40 children for independent scoring by 2 physicians to evaluate interobserver agreement. Disease activity was evaluated using the JADAS27 score (active disease defined as JADAS27 >1). In the main cross-sectional analysis, JIA subtype and ultrasound scores served as grouping/predictor variables, while joint-level ultrasound positivity rate, GS/PD scores, and disease activity state defined by JADAS27 served as outcome measures. Hierarchical clustering analysis was carried out to analyze involvement patterns, Chi-square test to identify joints with significant differences across subtypes, Spearman correlation analysis to evaluate the correlations between ultrasound scores and JADAS27, and ROC curve analysis to assess diagnostic performance. A small-sample longitudinal follow-up subset (square test to identify joints with significant differences across subtypes, Spearman correlation analysis to evaluate the correlations between ultrasound scores and JADAS27, and ROC curve analysis to assess diagnostic performance. A small-sample longitudinal follown=37) was included to evaluate the changes before and after treatment.

①The interobserver agreement of the scoring system was excellent, with intraclass correlation coefficients of 0.87 (95%CI: 0.78 to 0.95) for GS scores and 0.85 (95%CI: 0.78 to 0.91) for PD scores. A total of 17748 joints were evaluated, with synovitis detected in 829 joints (4.7%) and PD positivity in 525 joints (3.0%). ② Distinct patterns of joint involvement were observed across subtypes: pJIA showed extensive polyarticular involvement (6.2±10.0 joints per case), with prominent small-joint involvement in the hands (proximal interphalangeal 32.5%, metacarpophalangeal 19.5%); sJIA predominantly involved large joints (3.1±4.3 joints per case), with shoulder involvement in 19.4% of patients (81.3% of those with shoulder involvement had sJIA); oJIA had the fewest involved joints (0.9±1.4 per case), predominantly the knee (18.8%); ERA had a mean of 1.5±1.3 involved joints per case. Significant differences across subtypes were found in the shoulder, elbow, wrist, metacarpophalangeal, proximal interphalangeal, and distal interphalangeal joint (adjusted P<0.05). The hand-wrist joint feature (positivity in wrist/MCP/PIP) yielded an AUC value of 0.61 for identifying pJIA; shoulder joint positivity showed a specificity of 0.99 for sJIA. PD scores were correlated more strongly with JADAS27 score (r=0.54, P<0.001) than GS scores (r=0.48, P<0.001), with the highest correlation coefficient observed in pJIA (r=0.71, P<0.001). ③ PD scores demonstrated moderate diagnostic performance for discriminating active disease (AUC=0.67, 95%CI: 0.61 to 0.75, P<0.001), with an optimal cutoff of ≥1.0, a sensitivity of 51% and a specificity of 82%; the AUC value was 0.73 in pJIA, 0.71 in both sJIA and oJIA, and not statistically significant in ERA (P=0.33). ④ Longitudinal follow-up showed that PD scores could reflect treatment response. During a mean follow-up of 4.4 months in 32 patients with active disease at baseline, the JADAS27 score was decreased from 8.0 (4.0, 15.3) to 3.5 (1.0, 8.1) (P<0.001, SRM=−0.74); PD scores were decreased from 1.0 (0, 3.0) to 0 (0, 1.0) (P=0.003, SRM=−0.25), whereas GS scores showed no significant change (P=0.67).

Distinct ultrasonographic phenotypes are identified across JIA subtypes: shoulder positivity strongly points to sJIA, wrist/MCP/PIP positivity tends to pJIA, while oJIA and ERA rarely involve the shoulder or small joints of the hands. The PD score can sensitively reflect disease activity and treatment response, but its diagnostic performance varies among subtypes: strongest correlation with disease activity in pJIA, whereas relatively weak in sJIA and ERA, requiring comprehensive assessment in considering systemic inflammatory indicators.