Comparative effectiveness of interferons in relapsing–remitting multiple sclerosis: a meta-analysis of real-world studies

Abstract

Background: Differences between interferons have been evaluated for over 20 years. While randomized controlled trial (RCT) data is mainly used for assessments and strong data for causal inferences, it does not necessarily reflect everyday practice. Real-world data may provide additional information.Purpose: To assess the results, quality, and representativeness of observational studies directly comparing interferons (IFNs) in RRMS.Methods: Medline and Embase were searched for observational studies comparing IFN-beta-1a 30 mcg IM (Avonex1), IFN-beta-1a 44 mcg SC (Rebif2) and/or IFN-beta-1b 250 mcg SC (Betaseron3). Outcomes included annualized relapse rate (ARR), proportions relapse free, confirmed progression free, treatment persistence, and neutralizing antibodies rates (NABs) measured up to 5 years of treatment. Data was combined using random effects meta-analyses. Categorical values were analyzed using chi-squared and Mann–Whitney tests.Results: Thirty-six studies examining 32,026 patients (72.5% females, age = 39.2 ± 3.7 years, disease duration = 5.6 ± 2.0 years) were identified. Thirty-three studies investigated IFN-beta-1a IM (N = 11,925), 30 IFN-beta-1a SC (N = 10,684) and 34 IFN-beta-1b SC (N = 9417). Baseline ARRs were similar (1.37 ± 0.35, 1.51 ± 0.27 and 1.55 ± 0.23, respectively; P = .101) as were EDSS scores (2.24 ± 0.39, 2.33 ± 0.30, 2.55 ± 0.38; P = .070) and >75% were naïve to IFNs. On treatment, ARRs were comparable (IFN-beta-1a IM 0.52 ± 0.27, IFN-beta-1a SC 0.51 ± 0.24, IFN-beta-1b SC 0.55 ± 0.23; P = .595). Proportions of relapse-free patients were similar between drugs (P > .05 for all data points), except that IFN-beta-1a SC was superior to IFN-beta-1b SC in years 3–5 (all P ≤ .001). After 1 year, EDSS scores were comparable; after 2 years, IFN-beta-1a IM and IFN-beta-1a SC incurred less disease progression than IFN-beta-1b SC (P < .02). Confirmed progression-free rates and persistence were similar over 5 years. Fewer patients developed NABs with IFN-beta-1a IM (4.7 ± 1.5%) versus IFN-beta-1a SC (21.4 ± 2.8%) (P < 0.001) or IFN-beta-1b SC (32.2% ± 3.3%) (P < .001).Conclusions: In this comprehensive meta-analysis of real-world studies in RRMS, IFN-beta-1a IM, IFN-beta-1a SC and IFN-beta-1b SC had similar clinical profiles. When selecting an IFN, practitioners should consider observational data in their decision making process.