Background/Purpose: To compare the abilities of MBDA score, patient global assessment (PGA) and evaluator global assessment (EGA) to assess risk for radiographic progression (RP), and to assess the ability of MBDA score to predict RP among patients with concordant or discordant PGA and EGA.

Methods: Patients were pooled from two RCTs of patients with recent onset RA treated with conventional and biologic DMARDs (OPERA and SWEFOT, N=386) and from a registry of patients with predominantly established RA and diverse treatments (BRASS, N=380). Pearson correlations were determined between MBDA scores (adjusted for the effects of age, sex and adiposity) (scale 1-100), PGA and EGA (each on a scale of 1-10) at baseline. PGA and EGA were considered discordant when they differed by >2.5. Univariate logistic regression assessed ability to predict RP (ΔTSS >5 over 1 year) for MBDA score, PGA and EGA as continuous variables; and for discordance of PGA and EGA as 2-level (concordant vs. discordant) or 3-level (PGA >EGA, concordant, EGA >PGA) categorical variables. Multivariable regression considered the main effect and interaction terms of the MBDA score, as a continuous variable, paired with each other variable, to test the ability of each pair to assess risk of RP. All models included a random effect on cohort. Odds ratios were reported for every 10-unit increase in MBDA score. Frequency of RP was determined in subgroups with MBDA score low (< 30), moderate (30-44) or high ( >44) for patient groups based on PGA/EGA concordance or discordance.

Results: The 766 patients studied were 76% female, 76% positive for RF and/or anti-CCP Ab, with mean age 55 years, DAS28-CRP 4.7, CRP 22 mg/L, CDAI 26, SJC 9.1, PGA 4.4, EGA 3.4, MBDA score 53. No interaction was seen between MBDA score and type of cohort (early vs established RA). PGA and EGA were discordant in 294 of 766 (38%) patients and were weakly to moderately correlated (r=0.38). Among discordant patients, PGA was >EGA in 227 cases and EGA was >PGA in 67 cases. Correlations between MBDA score and PGA or EGA were r=0.41 and r=0.34, respectively. In univariate analyses, MBDA score was a statistically significant predictor of radiographic progression (OR=1.53, p=6.3×10^-8) whereas PGA, EGA, 2-level discordance and 3-level discordance were not (p=0.38, 0.47, 0.74, 0.83, respectively). In multivariable analyses, significant interactions were observed between MBDA score and discordance (2-level, p=0.0029; 3-level, p=0.0087). The interaction analysis demonstrated, in PGA/EGA-concordant patients, low risk of radiographic progression when MBDA score was low and elevated risk when it was high (OR=1.33 [1.1, 1.59]). A relationship between MBDA score and RP risk was also demonstrated, with heightened trend, among discordant patients with PGA >EGA (OR=2.04 [1.53, 2.81]) and EGA >PGA (OR=3.43 [1.37, 13.8]) (Figure 1).

Conclusion: MBDA score was a significant predictor of radiographic progression, whereas PGA and EGA were not. MBDA score predicted progression whether PGA and EGA were concordant or discordant. These results suggest that MBDA score detects joint-damaging disease activity more accurately than PGA and EGA and it does so whether or not PGA and EGA are in agreement.

Figure 1. Rate of radiographic progression (RP) by MBDA category for patient groups based on concordance or discordance of Patient Global Assessment (PGA) and Evaluator Global Assessment (EGA).

« Back to ACR Convergence 2020

ACR Meeting Abstracts - https://acrabstracts.org/abstract/comparison-of-mbda-score-patient-global-assessment-and-evaluator-global-assessment-for-predicting-risk-of-radiographic-progression/