Background/Purpose

The use of ultrasonography (US) in rheumatoid arthritis (RA) is rapidly increasing. Currently, there is no consensus regarding which joints and tendons should be systematically assessed. Validity, including comprehensiveness, and responsiveness must be weighted against feasibility. Our objectives were to develop candidate sets for assessment of US joint inflammation through a data driven approach using data from early RA patients, and then perform initial validation in an established RA cohort.

Methods

Between January 2010 and June 2013 patients (pts) were included in one of two cohorts: Early RA (DMARD-naïve pts with RA of <2 yrs symptom duration fulfilling 2010 ACR/EULAR classification criteria), and established RA (pts starting or switching biologic DMARDs). An extensive US examination was performed by experienced sonographers using a validated grey-scale (GSUS) and power Doppler (PDUS) semi-quantitative scoring system with scores 0-3 for GSUS and PDUS in each of the following 36 joints and 4 tendons: MCP 1-5, PIP 2-3, radiocarpal, distal radioulnar, intercarpal, elbow, knee, talocrural, MTP 1-5, extensor carpi ulnaris and tibialis posterior tendons, bilaterally. An US atlas was used as reference¹. We performed principal component factor analyses (PCA) in the early RA US data to identify joint groups with high internal correlation, and selected candidate joint/tendon sets based on these analyses. We assessed the loss of information compared to the full score by R² from linear regression analysis. Finally, the candidate sets were validated in the established RA cohort.

Results

A total of 439 patients were included, 227 with early and 212 with established RA; 62% vs. 77% anti-CCP pos, mean(SD) age 51(14) vs. 52(13) yrs, DAS28 4.7(1.2) vs. 4.7(1.4), median(25-75 percentile) 28-SJC 6(3-11) vs. 5(2-10), disease duration 0.5(0.2-0.9) vs. 8(3-15) yrs, mean(95% CI) 36-joint GSUS score 23(21-25) vs. 28(25-30) (p=0.003), 36-joint PDUS score 11(10-12) vs. 13(11-15) (p=0.20). Nearly 17,000 individual joints/tendons were assessed. We identified 9 groups based on PCA in the early RA data, presented in table 1. Comparisons between the candidate sets and the total GSUS and PDUS scores in the early RA cohort as well as validation in the established RA cohort are presented in table 2.

Conclusion

We used a data driven approach to develop candidate sets of joints/tendons to be assessed by GS and PD US, and the resulting reduced scores retained most of the information from the total score of 40 joints/tendons. Unilateral reduced scores explained 78% to 85% of the total score, while bilateral reduced scores explained 89% to 93% of the total score. The candidate scores performed equally well in a validation cohort of established RA. Our results show that a reduced US assessment may efficiently contribute to disease assessment in RA. Further validation in longitudinal RA cohorts and data on responsiveness are needed.

¹Hammer HB et al. ARD 2011