Updated Risk Matrix Developed to Predict Rapid Radiographic Progression in Early Rheumatoid Arthritis

An updated matrix to predict rapid radiographic progression (RRP) in early rheumatoid arthritis (RA) displayed high precision and moderate sensitivity and specificity, according to study results published in Rheumatology. Researchers note that the matrix may have utility in developing treatment plans for patients with early RA, which could ultimately aid physicians in making treatment decisions in daily clinical practice.

Investigators conducted a post-hoc analysis of pooled data from 2 cohorts (Leuven and ESPOIR) and 3 randomized clinical trials (SWEFOT, BeST, and ASPIRE). The Leuven cohort assessed the efficacy of various therapeutic modalities for early RA; the ESPOIR cohort followed the prognosis of patients with early RA from community practice. Each of the 3 randomized trials was designed test the efficacy of methotrexate (MTX) in combination with other medications used in the treatment of early and active RA. From each cohort, investigators gathered data on patients with active early RA whose were treated with MTX or leflunomide after study enrollment. Patients with prior exposure to disease-modifying antirheumatic drugs (DMARDs) were excluded from analyses. The main outcome of the study was the presence of RRP after 1 year of follow-up. Logistic model parameters were used to create a risk matrix for RRP probability. Predictors for RRP were selected such that the area under the receiver operating characteristic curve (AUC) was maximized. Model calibration and discriminatory power were estimated.

Data from 1306 patients were pooled. Of the patients, 20.6% exhibited RRP within 1 year of follow-up. Mean disease duration at baseline was 16.1 (±21.6) weeks. Four predictors of RRP were retained in the risk matrix: (1) rheumatoid factor positivity; (2) the presence of at least 1 RA erosion on radiographs (3) C-reactive protein (CRP) level >30 mg/L; and (4) number of swollen joints. The final prediction matrix estimated 1-year RRP risk for 36 possible combinations of baseline predictors. For example, RRP risk was 1.5-fold greater than average among patients with rheumatoid factor positivity, >1 RA erosion, CRP levels >30 mg/L, and a swollen joint count ≥ 10. Model calibration was high, per a Hosmer-Lemeshow goodness-of-fit test (P =.79). Model precision exceeded that of prior published matrices (95% CI, ±0.02 to ±0.08). The matrix also displayed moderate sensitivity and specificity (AUC, 0.68; 95% CI, 0.643-0.716).

An updated risk matrix for RRP in early active RA displayed excellent calibration and high precision. However, the researchers note that data missing from the included studies may have influenced the predictive capacity of the model. They also noted that patients lost to follow-up may have had increased risk for RRP. Even so, the updated matrix presents a robust means of assessing RRP risk in early RA.

Disclosure: Two study authors declared affiliations with the pharmaceutical industry. Please see the original reference for a full list of authors’ disclosures.

Reference

Vanier A, Smolen JS, Allaart CF, et al. An updated matrix to predict rapid radiographic progression of early rheumatoid arthritis patients: pooled analyses from several databases [published online November 13, 2019]. Rheumatology (Oxford). doi: 10.1093/rheumatology/kez542