Background The multi-biomarker disease activity (MBDA) score measures 12 serum biomarkers to assess disease activity in patients with rheumatoid arthritis (RA) on a scale of 1–100. The MBDA score was validated in several different cohorts but has not been evaluated in a cohort consisting only of patients initiating TNF inhibitor therapy in clinical practice. We utilized patients enrolled in the Corrona-CERTAIN study to evaluate MBDA scores for patients initiating adalimumab (ADA) in several clinical practices in the US.
Objectives Evaluate the ability of the MBDA score to assess response to treatment with ADA.
Methods We studied 106 biologic-naïve RA patients who had been treated with ADA for at least 12 months, had initiated ADA in CERTAIN while in moderate or high disease activity by CDAI, and for whom sera were available at baseline (BL) and Months 3 and 6. Changes (Δ) in MBDA score and DAS28-CRP were evaluated from BL to Months 3 and 6 by the one-sample paired t-test. ΔMBDA scores were evaluated for patients grouped by EULAR response categories, using the Cochran-Armitage test for trend. Receiver Operating Characteristic (ROC) analysis with bootstrap sampling (20,000 iterations) was used to evaluate MBDA score ability to discriminate ΔDAS28-CRP improvement >1.2 units at Month 3, and to determine the optimal MBDA threshold by maximizing the sum of sensitivity and specificity (Youden's index criterion).
Results At BL, median values were age 54.5 years, disease duration 2 years, BMI 28.2, DAS28-CRP 4.7, CDAI 24, SDAI 25.4; 74.5%/65.4% were RF+/ACPA+. Median MBDA score was 49 with 17 (16%) patients in low (<30), 23 (22%) patients in moderate (30–44), and 66 (62%) patients in high (>44) MBDA categories. The relative magnitude and the direction of median change from BL to Months 3 and 6 were similar for MBDA score (−8, −9) and DAS28-CRP (−1.4, −1.8), with statistically significant changes from BL for each (line graphs in Figure). Similar results were observed for SDAI and CDAI. Pearson's correlations with ΔMBDA score at Months 3 and 6 were 0.58, 0.59, respectively, for ΔDAS28-CRP; 0.48, 0.47 for ΔSDAI; and 0.42, 0.42 for ΔCDAI (all p<0.0001). Median reductions in MBDA score were significantly greater for patients with concurrent DAS28-CRP improvement >1.2 units (n=67) vs. ≤1.2 units (n=39) at Month 3 (15 vs. 2) and Month 6 (13 vs. 1.5) (both p<0.0001); and for patients with EULAR Good vs. Moderate vs. Non-responses (p=0.0002 at Month 3, p<0.0001 at Month 6) (bar graphs in Figure). Area under the ROC curve (AUROC) for the ability of ΔMBDA score from BL to Month 3 to discriminate DAS28-CRP improvement >1.2 units at Month 3 was 0.82 (95% CI, 0.74–0.90). The optimal threshold for this discrimination was a reduction in MBDA score >9 units, with sensitivity/specificity=0.63/0.82 and PPV/NPV =0.86/0.56.
Conclusions This study expands the previous validation of the MBDA score by demonstrating its ability to assess response to treatment with adalimumab in a US clinical practice cohort. The AUROC value of 0.82 for discriminating improvements in DAS28-CRP >1.2 units indicates a significant association between change in MBDA score and clinical improvement.
Disclosure of Interest D. Pappas Grant/research support from: AbbVie, Consultant for: AbbVie, Employee of: CORRONA, LLC, E. Sasso Shareholder of: Myriad Genetics, Inc., Employee of: Crescendo Bioscience Inc., R. Bolce Shareholder of: Myriad Genetics, Inc., Employee of: Crescendo Bioscience Inc., X. Liu Shareholder of: Myriad Genetics, Inc., Employee of: Crescendo Bioscience Inc., C. Etzel Consultant for: Merck, Employee of: CORRONA, LLC