Model-Based Meta-Analysis Compares DAS28 Rheumatoid Arthritis Treatment Effects and Suggests an Expedited Trial Design for Early Clinical Development

Abstract

A nonlinear mixed effects modeling approach was used to conduct a model-based meta-analysis (MBMA) of longitudinal, summary-level, baseline-corrected 28-joint Disease Activity Score (ΔDAS28) clinical trial data from seven approved rheumatoid arthritis (RA) drugs (abatacept, adalimumab, certolizumab, etanercept, rituximab, tocilizumab, and tofacitinib), representing 130 randomized clinical trials in 27,355 patients. All of the drugs except tocilizumab were found to have relatively similar ΔDAS28 time courses and efficacy (baseline-corrected and placebo-corrected) at 24 weeks and beyond of approximately 0.87-1.3 units in the typical RA patient population. Tocilizumab was estimated to have a differentially greater response of 1.99 at 24 weeks, likely due to its disproportionate effect on the acute-phase cytokine interleukin-6. Baseline DAS28, disease duration, percentage of male participants, and the year of conduct of the trial were found to have statistically significant effects on the timing and/or magnitude of ΔDAS28 in the control arms. Clinical trial simulations using the present MBMA indicated that abatacept, certolizumab, etanercept, tocilizumab, and tofacitinib would be expected to have a greater than 70% probability of showing a statistically significant difference vs. control at Week 6 with a sample size of ~ 30 patients per arm. In future RA clinical trials, an interim analysis conducted as early as 6 weeks after treatment initiation, with relatively small sample sizes, should be sufficient to detect the ΔDAS28 treatment effect vs. placebo.

Figure 1

DAS28‐ESR and DAS28‐CRP relationship (a) by mechanism of action and (b) by duration of treatment. DAS28‐CRP = −0.194 ± 0.0484 × DAS28‐ESR + 0.899 ± 0.00568. APE, absolute percentage error; CD, cluster of differentiation; DAS28, 28‐joint Disease Activity Score; CRP, C‐reactive protein; ESR, erythrocyte sedimentation rate; IL, interleukin; JAK, Janus kinase; MAPE, mean APE; TNF, tumor necrosis factor.

Figure 2

Mean observed and predicted DAS28 change from baseline vs. time. Solid lines: median predicted ΔDAS28; dotted lines: 95% prediction interval for ΔDAS28; shaded region: 90% prediction interval for ΔDAS28; circles represent observed trial mean ΔDAS28 with size proportional to sample size, and a separate color for each trial. Data above the prediction interval for abatacept were from an arm with a small sample size (n = 43) in Manders et al. that had a relatively low baseline (DAS28‐ESR ~ 4.7; DAS28‐CRP ~ 4.0). ⁴⁸ ΔDAS28, change from baseline in DAS28; DAS28, 28‐joint Disease Activity Score.

Figure 3

Change from baseline in (a) mean DAS28 and (b) placebo‐corrected. Treatment regimens: abatacept: 10 mg/kg IV q.4wk or 125 mg SC q.wk; adalimumab: 40 mg SC q.2wk; certolizumab: 200 or 400 mg SC q.2wk; etanercept: 50 mg SC q.wk; rituximab: 1,000 mg INF × 2/year; tocilizumab: 8 mg/kg SC q.4wk; tofacitinib: 5 mg PO b.i.d. DAS28 response in a population that failed methotrexate, is on a background of DMARD and that is 19% male, with a mean age of 53 years, a disease duration of 8.2 years, and a mean baseline DAS28 of 6.2. b.i.d., twice daily; DAS28, 28‐joint Disease Activity Score; DMARD, disease‐modifying anti‐rheumatic drug; INF, infusion; IV, intravenous; PO, oral; q.wk, weekly; q.2wk, every 2 weeks; q.4wk, every 4 weeks; SC, subcutaneous.

Figure 4

Clinical trial simulation to evaluate the effect of timing and sample size on the utility of DAS28 to demonstrate a treatment effect for different RA therapies. (a) Probability of detecting a treatment effect vs. placebo as a function of time with a constant sample size of N = 50 per arm. (b) Probability of detecting a treatment effect vs. placebo as a function of sample size (per arm) after 6 weeks of treatment. Treatment regimens: abatacept: 10 mg/kg IV or 125 mg SC; adalimumab: 40 mg SC q.2wk; certolizumab: 200 or 400 mg SC q.2wk; etanercept: 50 mg SC q.wk; rituximab: 1,000 mg INF × 2/year; tocilizumab: 8 mg/kg SC q.4wk; tofacitinib: 5 mg PO b.i.d. Dotted lines represent the interim analysis time of (a) 6 weeks or (b) the sample size of N = 50. ΔDAS28, change from baseline in 28‐joint Disease Activity Score; b.i.d., twice daily; DAS28, 28‐joint Disease Activity Score; INF, infusion; IV, intravenous; PO, oral; q.wk, weekly; q.2wk, every 2 weeks; q.4wk, every 4 weeks; RA, rheumatoid arthritis; SC, subcutaneous.