key: cord-1040847-cwsmesl9
authors: Hanrahan, Conor; Lee, Todd
title: Network meta-analysis of infliximab biosimilars for the treatment of rheumatoid arthritis
date: 2021-02-18
journal: Am J Health Syst Pharm
DOI: 10.1093/ajhp/zxab042
sha: 1079f86346f7fef252385116fa760f0657feed03
doc_id: 1040847
cord_uid: cwsmesl9

DISCLAIMER: In an effort to expedite the publication of articles related to the COVID-19 pandemic, AJHP is posting these manuscripts online as soon as possible after acceptance. Accepted manuscripts have been peer-reviewed and copyedited, but are posted online before technical formatting and author proofing. These manuscripts are not the final version of record and will be replaced with the final article (formatted per AJHP style and proofed by the authors) at a later time. PURPOSE: This article assesses the relative efficacy and safety of infliximab biosimilars in treatment of patients with rheumatoid arthritis (RA). METHODS: A frequentist, random-effects network meta-analysis was performed to evaluate evidence from randomized controlled trials that examined the use of infliximab biosimilars for treatment of patients with RA. PubMed and MEDLINE and other sources were searched for reports evaluating rates of response to treatment with the reference product (infliximab) vs an infliximab biosimilar. The primary efficacy outcome of interest was the rate of attainment of ACR20 (ie, 20% improvement in American College of Rheumatology core measures). The primary safety outcome was the rate of treatment-related serious adverse events (SAEs). Data were extracted by the primary author, and an assessment for risks of methodological bias was performed for each evaluated study. RESULTS: Five studies that enrolled a total of 2,499 patients were included. Overall comparisons using odds ratios and 95% confidence intervals (CIs) did not indicate statistically significant differences in response to treatment with biosimilar agents relative to each other or the infliximab reference product. ORs for ACR20 response for biosimilars vs infliximab were as follows: 1.475 (95% CI, 0.940-2.315) for infliximab-axxq, 1.259 (95% CI, 0.854-1.855) for infliximab-dyyb, 0.865 (95% CI, 0.5511.358) for infliximab-qbtx, and 0.832 (95% CI, 0.506-1.367) for infliximab-abda. Similar findings were observed in reported SAE rates among patients treated with the various biosimilars. CONCLUSION: ACR20 response appears to be comparable and nonsignificantly different between infliximab biosimilars. In the absence of any meaningful differences in safety or efficacy, biosimilar cost may be the deciding factor in choosing a treatment or agent for formulary inclusion.

A c c e p t e d M a n u s c r i p t Am J Health-Syst Pharm. 2021;78:xxx-xxx Tumor necrosis factor (TNF) is a proinflammatory endogenous cytokine that plays a role in many inflammatory conditions, including rheumatoid arthritis (RA), ankylosing spondylitis, Crohn's disease, and ulcerative colitis. [1] [2] [3] [4] TNF inhibitors may decrease symptoms and slow the progression of disease in these patients. 3, 4 Current guidelines from the American College of Rheumatology (ACR) recommend using TNF inhibitors in patients who do not respond to monotherapy with first-line disease-modifying antirheumatic drugs, such as methotrexate. 4 Numerous TNF inhibitors have been approved by the US Food and Drug Administration (FDA), with the oldest being infliximab (Remicade, Janssen Biotech) which came to market in 1998. 5 In recent years, several biosimilars to infliximab have been developed, including infliximab-dyyb (Inflectra, Pfizer), infliximab-abda (Renflexis, Merck), infliximab-qbtx (Ixifi, Pfizer), and infliximabaxxq (Avsola, Amgen). 6 Biosimilars are highly similar to their respective originator reference products, with only minimal clinical differences in safety, purity, or potency. 7, 8 Although noninferiority or equivalency studies of biosimilars and their originator products are conducted, biosimilars are not often directly compared with each other. Thus, there may be unrecognized differences in efficacy or safety between biosimilar agents themselves. This lack of evidence makes it challenging for clinicians, payers, and healthcare organizations to make decisions about drug formularies and clinical care when choosing between biosimilars.

To date, no indirect or head-to-head studies comparing all infliximab biosimilar agents against one another have been published. Therefore, a network meta-analysis may be useful because it allows for multiple comparisons across a range of interventions, even in the absence of direct evidence. 9, 10 The objective of the study described here was to evaluate the comparative efficacy and safety of FDA-approved infliximab biosimilars for the treatment of RA using a network meta-analysis framework. M a n u s c r i p t

The study was designed and is reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) extension guidelines for network metaanalyses. 10 Study selection and eligibility criteria. Studies were included if they met the following parameters: a randomized controlled trial (RCT) study design; inclusion of patients diagnosed with RA; evaluation of an FDA-approved infliximab biosimilar in patients who had an incomplete response to methotrexate; and reporting of efficacy in terms of ACR20 response rate (ie, 20% improvement in American College of Rheumatology core measures). There was no minimum sample size requirement for inclusion of a study in the meta-analysis. Studies were excluded if they had a quasi-RCT design, included duplicate data (ie, were repeat publications), or involved patients who were not naïve to infliximab therapy (eg, studies that evaluated switching from Remicade to an infliximab biosimilar). These switching studies were excluded because most were extensions of the original published trials; therefore, it was believed that including them could bias the analysis due to counting of the same patient cohorts multiple times within the network.

Outcome measures. The primary efficacy endpoint was the ACR20 response rate. The ACR20 response rate was chosen because it is often the primary measure of RA treatment efficacy across published studies 11 and is recognized as a core measure of disease activity by ACR. 12 ACR20 is a dichotomous endpoint whose achievement requires 20% or greater improvement in tender and swollen joint counts (per an assessment of prespecified joints), as well as an improvement of ≥20% in 3 of 5 other areas: inflammation (as evidenced by an elevated erythrocyte sedimentation rate or C-reactive protein level), patient assessment of disease activity (based on Likert response scale or specific questions within broader self-assessment instruments), physician assessment of disease activity (based on horizontal visual analog scale or Likert scale assessment), patient assessment of pain (based on horizontal visual analog scale or Likert scale assessment), and patient assessment of A c c e p t e d M a n u s c r i p t physical function (using any physical function scale, such as the Arthritis Impact Measurement Scale [AIMS] or Health Assessment Questionnaire-Disability Index [HAQ-DI]). [11] [12] [13] The primary safety endpoint was the rate of occurrence of any serious adverse event (SAE).

Although SAEs are not always well defined in clinical trials, reporting of SAEs generally follows the definition outlined by FDA, which includes a life-threatening event, death, hospitalization, disability, permanent damage, and other outcomes that may jeopardize the patient or require medical intervention. 14,15 A comparison of rates of specific adverse reactions was not deemed to be feasible due to differences in how they were generally codified and reported in the articles. independent evaluation was needed to validate questionable article inclusions. After initial screening, full reports for all titles that appeared to meet the inclusion criteria, as well as those whose appropriateness for inclusion was uncertain, were obtained. The full-text reports were then screened against the inclusion criteria. The following data were extracted using a standardized collection form: study design parameters, sample size, key patient demographics (age, markers of disease severity, etc), information regarding interventions (drug dose, frequency, etc), details related to study efficacy endpoints (primary efficacy outcome, follow-up period, prespecified equivalence A c c e p t e d M a n u s c r i p t margin, intention-to-treat vs per-protocol methodology, results for primary outcome, etc), and key information related to safety endpoints (primary safety outcome, frequencies and types of of adverse effects [AEs], type of adverse effect, etc). Of note, the abstracted data from the intentionto-treat population in each study were used in the primary efficacy and safety evaluation.

Validity assessment. The Cochrane Risk of Bias Tool (version 2) was used to assess the quality of studies based on the following criteria: randomization process, deviations from intended interventions, missing outcome data, measurement of the outcome, and selection of reported results. 16 Quality of evidence in each domain was then rated as low, high, or unclear. A qualitative synthesis of clinical and methodological heterogeneity was also conducted, as well as an assessment of transitivity. 17, 18 Statistical analysis. Stata version 16.1 (StataCorp LLC, College Station, TX) was used to perform the network meta-analysis using the "mvmeta" and "network" commands, a method that follows a frequentist approach. [17] [18] [19] The "network meta inconsistency" command within Stata was used to assess model consistency. A pairwise odds ratio (OR) and 95% confidence interval (CI) was determined to measure the association between a treatment and its efficacy. Results were considered statistically significant if the 95% CI did not contain the value 1. The Confidence in Network Meta-Analysis (CINeMA) Tool (version 1.9.1), which also uses a frequentist approach, was used to generate a corresponding league table comparing the relative effects between agents, including indirect pairwise comparisons between biosimilars. 20 A random-effects model was used for the CINeMA analysis. This same approach was used to perform prespecified sensitivity analyses to evaluate the impact of follow-up period differences on ACR20 response.

The efficacy and safety of each biosimilar agent in the treatment of RA was then arranged in the order of the probability of being ranked as the best-performing agent and displayed via a rankogram. Information on relative effects was converted to a probability that a treatment would be the best, second best, third best, fourth best, or worst by ranking each therapy according to the surface under the cumulative ranking curve (SUCRA). The SUCRA represents a numeric summary of A c c e p t e d M a n u s c r i p t the overall rank distribution associated with each treatment and supplements graphical representations for a given outcome. The SUCRA value is 1 when a treatment is certain to be the best and 0 when a treatment is certain to be the worst. [21] [22] [23] Results Study characteristics. The initial literature search yielded 294 results. After removing duplicates, 192 article titles and abstracts were reviewed, which resulted in 181 exclusions. The most common reason for exclusion was a non-RCT article type. The remaining 11 articles underwent fulltext review and, of these, 6 were excluded ( Figure 1 ). Thus, a total of 5 studies that cumulatively enrolled 2,499 patients met inclusion criteria and were included in the analysis. A summary of these studies is provided in Table 1 . Figure 2 A c c e p t e d M a n u s c r i p t Risk of bias. All 5 of the included studies were determined to have a low overall risk of bias (eFigure 1). Although all studies were randomized, authors of 3 of the 5 studies failed to provide specific details on how the random allocation sequence was generated; however, based on reviewer judgment, this was not deemed to have resulted in a high risk of bias given that these were large, multicenter trials run by experienced clinical trial teams. This decision was acceptable under the Cochrane Risk of Bias Tool (version 2) methodology. 16 Comparative efficacy. In general, individual study results for efficacy demonstrated equivalence of each biosimilar to the infliximab reference product in terms of ACR20 response (Table   1 ). Infliximab-axxq 24 and infliximab-dyyb 27,28 were the only products associated with a higher ACR20 response rate than the reference product, whereas infliximab-qbtx 25 and infliximab-abda 26 both were associated with lower response rates (albeit still within the prespecified equivalence margins).

Notably, the initial risk difference for infliximab-axxq exceeded the upper bound of the prespecified equivalence margin, such that superiority might be considered; however, this study was not specifically designed to test for superiority, and a post hoc analysis reduced the CI to within the equivalence range. 24 As described in Table 2 , results of the network meta-analysis did not indicate any significant differences in ACR20 response between biosimilar agents relative to each other or the infliximab reference product. Although point estimates exceeded 1 in several comparisons, all 95% CIs crossed 1 and there was wide CI overlap between treatments. Relative to the infliximab reference product, the OR for ACR20 achievement was 1.47 (95% CI, 0.94-2.32) with use of infliximab-axxq;, 1.259 (95% CI, 0.854-1.855) for infliximab-dyyb, 0.865 (95% CI, 0.551-1.358) for infliximab-qbtx, and 0.832 (95% CI, 0.506-1.367) for infliximab-abda. These findings were consistent with results of a predefined sensitivity analysis to evaluate the impact of follow-up period differences on ACR20 response. A c c e p t e d M a n u s c r i p t Results of the cumulative ranking probability analysis suggested that infliximab-axxq has a 63.4% chance of being the best-performing agent, while the cumulative probability of it being at least second best is 92% (Table 3 and eFigure 2). The numerical summary of the rank distribution for each treatment (ie, SUCRA ranking probabilities) was 0.9 for infliximab-axxq, 0.7 for infliximab-dyyb, 0.4 for the infliximab reference product, and 0.2 for both infliximab-qbtx and infliximab-abda.

Comparative safety. In general, the rates of AEs differed across individual studies. Overall AE rates were slightly lower in the evaluation of infliximab-axxq by Genovese and colleagues 24 and substantially higher in the evaluation of infliximab-dyyb by Takeuchi and colleagues, 28 in which over 80% of patients in both the biosimilar and reference product arms experienced an AE. The rate of SAEs followed a similar trend, with lower rates reported in the article by Genovese and colleagues 24 and higher rates in the evaluation by Takeuchi and colleagues. 28 Additional details are provided in Table 1 . Unfortunately, differences in data reporting made it difficult to compare rates of specific AEs; however, the most commonly reported AEs across the studies were infusion reactions, hypersensitivity, infection, and elevated liver enzymes. Results of our meta-analytic analysis do not suggest any statistically significant differences in SAE rates between biosimilars (Table 4 ).

The study was conducted to compare the relative clinical effectiveness of infliximab and its FDA-approved biosimilars. A network meta-analysis design was chosen to perform an indirect treatment comparison given the lack of head-to-head comparative data on the 5 agents. Overall, the point estimates and 95% CIs of our ACR20 efficacy results suggest that there are no clinically meaningful differences between agents. Although infliximab-axxq was found to have the highest probability of being the best at achieving ACR20 response based on SUCRA ranking probabilities, this result must be interpreted with caution. Specifically, SUCRA rankings may exaggerate small differences in relative effects, especially when based on data from a limited network 23 ; this is because SUCRA analysis does not consider the magnitude of the difference in effects. Therefore, it is A c c e p t e d M a n u s c r i p t possible that the first-ranked treatment may be either slightly better or vastly better than the second-ranked treatment. Additionally, SUCRA values do not account for the possibility of random chance in the model. 23 Similar to the primary efficacy results, the results of analysis of SAEs showed comparable and nonsignificant differences between products. However, these safety results should also be interpreted with caution given that the follow-up period varied between studies from 22 to 54 weeks. [24] [25] [26] [27] [28] Likewise, definitions of SAEs were not uniformly defined across studies.

Overall, our findings further support the use of biosimilars in practice and add additional credence to the clinical similarity of the evaluated products. Biosimilars must undergo a rigorous comparison against the reference product in order to gain regulatory approval. Specifically, manufacturers must demonstrate biosimilarity through a totality-of-the-evidence approach, which includes structural and functional analytical studies, pharmacokinetic and pharmacodynamic evaluations, and, if necessary, immunogenicity studies, animal studies, and other comparative clinical trials. 7 However, individual biosimilars are not directly compared with one another. Thus, although a biosimilar may be determined to be equivalent or noninferior to a reference product (depending on the study design), this does not preclude comparative differences between biosimilar agents. Thus, the results of this evaluation may help clinicians, payers, and healthcare organizations make decisions about the choice of biosimilar agent. With all other things being equal, product cost and contractual opportunities may become the primary deciding factor when selecting an infliximab biosimilar for formulary inclusion.

To our knowledge, this is the only published study using a network meta-analytic technique to compare all 4 infliximab biosimilars with one another for the treatment of RA. Previous literature has mostly reported on direct comparisons of biosimilars with the reference product in RCTs. Aside from these head-to-head trials, 2 recent meta-analyses comparing biosimilars to the infliximab reference product have been published. In a study by Bae and colleagues, 29 pooled outcomes data for infliximab-abda and infliximab-dyyb plus methotrexate were compared against data on both use of infliximab plus methotrexate and use of a placebo plus methotrexate. Although individual A c c e p t e d M a n u s c r i p t biosimilars were not compared in that study, results indicated no major difference in ACR20 response between the pooled biosimilars and infliximab. 29 Similarly, a study by Graudal and colleagues 30 found that the individual treatment effects of infliximab-abda and infliximab-dyyb were comparable to the reference product's in terms of RA progression; however, these biosimilars were not directly compared.

The results of our meta-analysis should be considered in the context of several limitations. Importantly, the network of trials was sparse, with only one study being identified for each of the biosimilars except for infliximab-dyyb, for which 2 studies were identified; this limited the robustness of our evaluation and the indirect comparisons performed. Additionally, while the included studies were all highly similar with regard to methodology and design, follow-up periods differed significantly. Long-term data were available only for infliximab-dyyb and infliximab-abda, which could have impacted the applicability of our findings. To address this limitation, a sensitivity analysis was conducted to evaluate ACR20 response at 30 weeks instead of 54 weeks, which was the follow-up period reported in the 2 longer-term studies. The results of the sensitivity analysis were consistent with findings of the primary efficacy analysis, indicating that follow-up period duration did not greatly impact the results (see eTable 2). Lastly, the primary efficacy analysis focused on only one outcome, ACR20, and assessed only efficacy data from RCTs. Investigators who conduct future network meta-analyses may consider evaluating additional outcomes or incorporating real-world effectiveness data from observational studies.

Results of a network meta-analysis suggest that infliximab biosimilars are generally comparable with regard to ACR20 response and SAEs. In the absence of any meaningful differences in safety or efficacy, cost may be the deciding factor when choosing a biosimilar agent for formulary inclusion.

The authors have declared no potential conflicts of interest.

Lexicomp Online

Wolters Kluwer Health

IBM Corporation; 2020

AGA clinical practice guidelines on the management of moderate to severe ulcerative colitis

American College of Rheumatology Guidelines for the Treatment of Rheumatoid arthritis

US Food and Drug Administration. Drugs@FDA. Accessed

IPD Analytics

Biosimilar development, review, and approval

Biosimilars in rheumatology: understanding the rigor of their development

Interpreting indirect treatment comparisons and network meta-analysis for health-care decision making: report of the ISPOR Task

Force on Indirect Treatment Comparisons Good Research Practices: part 1. Value Health

The PRISMA extension statement for reporting of systematic reviews incorporating network meta-analyses of health care interventions: checklist and explanations

Guidance for industry: rheumatoid arthritis: developing drug products for treatment

The American College of Rheumatology preliminary core set of disease activity measures for rheumatoid arthritis clinical trials. The Committee on Outcome Measures in Rheumatoid Arthritis Clinical Trials

American College of Rheumatology Committee to Reevaluate Improvement Criteria. A proposed revision to the ACR20: the hybrid measure of American College of Rheumatology response

What is a serious adverse event?

American College of Rheumatology position statement: pharmacovigilance

Revised Cochrane risk-of-bias tool for randomized trials

Network meta-analysis: an introduction for clinicians

Network meta-analysis: application and practice using Stata

Network meta-analysis

CINeMA: An approach for assessing confidence in the results of a network meta-analysis

Indirect treatment comparison

Analyzing data and undertaking meta-analyses

Approaches to interpreting and choosing the best treatments in network meta-analyses

Comparative clinical efficacy and safety of the proposed biosimilar ABP 710 with infliximab reference product in patients with rheumatoid arthritis

A randomized controlled trial comparing PF-06438179/GP1111 (an infliximab biosimilar) and infliximab reference product for

A c c e p t e d M a n u s c r i p t A c c e p t e d M a n u s c r i p t A c c e p t e d M a n u s c r i p t A c c e p t e d M a n u s c r i p t A c c e p t e d M a n u s c r i p t a All data are odds ratios with 95% confidence intervals. Odds ratios less than 1 favor the intervention listed in the corresponding row.