A randomized controlled trial comparing PF-06438179/GP1111 (an infliximab biosimilar) and infliximab reference product for treatment of moderate to severe active rheumatoid arthritis despite methotrexate therapy

Background This double-blind, active-controlled, randomized, multinational study evaluated the efficacy, safety, pharmacokinetics (PK), and immunogenicity of PF-06438179/GP1111 (IxifiTM/Zessly®), an infliximab biosimilar, vs infliximab (Remicade®) reference product sourced from the European Union (infliximab-EU) in biologic-naïve patients with moderate to severe active rheumatoid arthritis (RA) despite methotrexate therapy. This paper reports results from the initial 30-week treatment period. Methods Patients (N = 650) were stratified by geographic region and randomized 1:1 to PF-06438179/GP1111 or infliximab-EU (3 mg/kg intravenous at weeks 0, 2, and 6, then every 8 weeks). Dose escalation to 5 mg/kg was allowed starting at week 14 for patients with inadequate RA response. The primary endpoint was American College of Rheumatology criteria for ≥ 20% clinical improvement (ACR20) response at week 14. Therapeutic equivalence was declared if the two-sided 95% CI for the treatment difference was within the symmetric equivalence margin of ± 13.5%. Statistical analysis was also performed with a two-sided 90% CI using an asymmetric equivalence margin (− 12.0%, 15.0%). Results Patients (80.3% female; 79.4% seropositive) had a mean RA duration of 6.9 years, and mean baseline Disease Activity Score in 28 joints, four components based on C-reactive protein was 6.0 in both arms. Week 14 ACR20 in the intention-to-treat population was 62.7% for PF-06438179/GP1111 and 64.1% for infliximab-EU. Week 14 ACR20 using nonresponder imputation was 61.1% for PF-06438179/GP1111 and 63.5% for infliximab-EU, and the 95% (− 9.92%, 5.11%) and 90% (− 8.75%, 4.02%) CIs for the treatment difference (− 2.39%) were entirely contained within the prespecified symmetric and asymmetric equivalence margins, respectively. No differences were observed between arms for secondary efficacy endpoints. Overall postdose antidrug antibody (ADA) rates through week 30 were 48.6% and 51.2% for PF-06438179/GP1111 and infliximab-EU, respectively. Efficacy and immunogenicity were similar between treatments for patients with dose escalation (at or after week 14), as well as between treatments for patients without dose escalation. Safety profiles of PF-06438179/GP1111 and infliximab-EU were similar, with no clinically meaningful differences observed between arms, including after ADA development. Serum drug concentrations were similar between arms at each time point during the initial 30-week treatment period. Conclusion PF-06438179/GP1111 and infliximab-EU demonstrated similar efficacy, safety, immunogenicity, and PK with or without dose escalation in patients with moderate to severe active RA on background methotrexate. Trial registration ClinicalTrials.gov, NCT02222493. Registered on 21 August 2014. EudraCT, 2013-004148-49. Registered on 14 July 2014. Electronic supplementary material The online version of this article (10.1186/s13075-018-1646-4) contains supplementary material, which is available to authorized users.


Supplementary Statistical Methods
The tipping point analysis for ACR20 at week 14 was performed for the asymmetric margin using a sequential logistic regression imputation model that included terms for region, ACR20 at weeks 6 and 12 (both potentially imputed), and treatment. This analysis varied the assumed probabilities of response among patients with missing data via a series of delta adjustments over the range of -1 to +1 that increased the treatment difference relative to logistic regression imputation without adjustment (delta=0). Tipping points were defined by deltas associated with a 90% CI that was not entirely contained within the asymmetric equivalence margin (-12%, 15%).
In the analysis using observed data (without NRI), the 95% and 90% CIs for the treatment difference in ACR20 response at week 14 were all contained within the prespecified equivalence margins. Likewise, the 95% and 90% CIs for the treatment difference in ACR20 response at week 14 were contained within the pre-specified equivalence margins in the analysis adjusting for geographic region (Supplementary   Table S1). Furthermore, a tipping point analysis based on multiple imputation demonstrated the asymmetric equivalence criterion was uniformly met over the range of adjustments (delta of -1 to +1) for treatment difference that were explored.

ACR20 responses over time
CR20 response rates across all study visits up to week 30 were analyzed using a generalized linear model for repeated-measure data, adjusting for geographic region.
Although the lower bounds of the 95% and 90% CIs for the week 4 ACR20 response rate treatment difference in the PP population were slightly outside of the symmetric and asymmetric margins, respectively, which were only defined for week 14 for the primary endpoint, results in Supplementary Table S2 show that the equivalence criteria were achieved at all other study visits (i.e., at 6 of the 7 post-treatment visits). This finding in and of itself provided clear evidence for similarity between the two treatment groups. 4

Supplementary Tables
Supplementary