Predicting risk for radiographic damage in rheumatoid arthritis: comparative analysis of the multi-biomarker disease activity score and conventional measures of disease activity in multiple studies

Abstract

Objective: To compare the multi-biomarker disease activity (MBDA) score with the DAS28-CRP and CRP for predicting risk of radiographic progression in patients with rheumatoid arthritis. Methods: Published studies of the MBDA score and radiographic progression with ≥100 patients per cohort were evaluated. Rates of radiographic progression over 1 year were determined across the low/moderate/high categories for MBDA score (low/moderate/high: <30, 30–44, >44), DAS28-CRP (low/moderate/high: ≤2.67, >2.67–4.09, >4.09) and CRP (low/moderate/high: ≤10, >10–30, >30 mg/L), with positive and negative predictive value (PPV, NPV) and relative risk (RR) determined for high vs. not-high (i.e. low and moderate combined) categories. Patient-level data from studies having all three measures was pooled to: (1) determine a combined RR for radiographic progression in the high vs. not-high categories for each measure; and (2) compare the predictive ability of MBDA score vs. DAS28-CRP by comparing the rates of radiographic progression observed in subgroups created by cross-classifying the high and not-high categories of each measure. Results: Five cohorts were identified for inclusion (total N=929). In each, radiographic progression was more frequent with increasing MBDA scores. Among the three cohorts with requisite data, PPVs were generally similar using categories of MBDA score, DAS28-CRP or CRP but NPVs were greater for MBDA score (93–97%) than DAS28-CRP or CRP (77–87%). RRs for radiographic progression were greater when based on categories of MBDA score than DAS28-CRP or CRP and the combined RR was greater for MBDA score (4.6, p