The totality-of-the-evidence approach to the development and assessment of GP2015, a proposed etanercept biosimilar

Abstract

Objective: The aim of this review is to describe the inherent variability that is natural to biologics and, using the proposed etanercept biosimilar (GP2015) as an example, provide details on the “totality-of-the-evidence” concept, whereby all physicochemical, biologic, preclinical, and clinical data for a biosimilar and reference medicine are evaluated in an iterative, stepwise manner and shown to be highly similar.

Methods: This review was carried out by a search of published articles, reviews, abstracts and patents in PubMed/Medline and Google Scholar up to November 2016.

Results: Analytical, functional, preclinical, and clinical data provide a comprehensive understanding of both GP2015 and reference etanercept, and demonstrate a high level of similarity between the two products in accordance with regulatory requirements. The totality of the evidence from all analyses and performed trials provides a robust scientific bridge between the biosimilar and clinical experience with the reference medicine, and is used to justify the use of the biosimilar in all indications for which the reference medicine is approved.

Conclusion: Biologic therapies have revolutionized the treatment of immune-mediated inflammatory diseases. The availability of biosimilars has the potential to improve patient access to biologic medicines and stimulate innovation. Physicians may be unfamiliar with the totality-of-the-evidence concept; therefore education and information on this unique approach to developing biosimilars is required to facilitate the use of biosimilars in clinical practice and allow physicians to make informed treatment decisions.

Keywords:

• Biologic
• biosimilar
• etanercept
• GP2015
• tumor necrosis factor alpha inhibitor

Introduction

Chronic immune-mediated inflammatory diseases (IMIDs) such as rheumatoid arthritis (RA), ankylosing spondylitis (AS), psoriatic arthritis (PsA), psoriasis, and inflammatory bowel disease significantly impact quality of life, morbidity, and mortality. The discovery that imbalances in inflammatory cytokines, such as tumor necrosis factor alpha (TNFα) and interleukins (ILs), are central to the pathogenesis of IMIDs led to the use of systemic, mechanism‐based treatment with biologic therapies that target inflammatory pathways¹.

Biologic therapies are developed using sophisticated, recombinant deoxyribonucleic acid (DNA) technology, whereby a gene encoding a specific protein is inserted into a host cell line. Cells are cultured in such a way that they express the desired protein or glycoprotein, which is then isolated and formulated for human use². In its final dosage form, the protein has a single, defined amino acid sequence; however, proteins often have a number of post-translational modifications, such as disulfide formation and glycosylation. The distribution of these post-translational modifications varies slightly from batch to batch, but must be contained within predefined limits to ensure consistent quality and clinical performance³. This variability has presented a challenge in determining the terminology that should be used for defining new versions of existing biologic medicines, including antibodies, soluble receptors, and fusion proteins⁴. Because the existing medicine and new version share inherent variability, they cannot be shown to be identical⁴; therefore, regulatory agencies in the US and EU decided on the term “biosimilar”, while in Canada they are called Subsequent Entry Biologics (SEBs)^4,5. Neither term is ideal, as they could imply that the function of these new versions differs from the existing biologic, which is not the case. Rather, biosimilars are biologics with identical structure, function and quality, comparable safety and equivalent efficacy to an already licensed reference medicine. The development and approval of biosimilars is based on the “totality-of-the-evidence” concept, whereby all physicochemical, functional, preclinical, and clinical data for a biosimilar and reference medicine are evaluated, compared, and shown to be highly similar (Figure 1)⁶. The term biosimilar or SEB, therefore, is a regulatory term reserved to describe products that are approved following a stringent regulatory pathway in which this complete data package is evaluated⁷. Biosimilars or SEBs are therefore not the same as “biomimics” or “biocopies” since these medicines, which are available in countries such as Bolivia, China, India, and some parts of Latin America, have not been directly compared against a licensed reference medicine according to the same stringent regulatory pathway^4,8,9.

Figure 1. Totality-of-the-evidence approach for demonstrating biosimilarity of GP2015, a proposed etanercept biosimilar. PD: pharmacodynamics; PK: pharmacokinetics.

Notably, the same approach used to demonstrate biosimilarity has been used since the early eighties to evaluate manufacturing changes of biologics in general⁴. When such manufacturing changes are needed, they will only be approved by regulators if the pre- and post-change medicines are highly similar and there are no clinically meaningful changes in structure or function¹⁰.

A number of biologics are currently licensed for the treatment of IMIDs, including TNFα inhibitors (TNFis) such as etanercept, infliximab, adalimumab, golimumab, and certolizumab pegol; agents that target ILs, such as secukinumab (IL-17A), tocilizumab (IL-6) and ustekinumab (IL-23), and agents that target cell surface antigens, such as rituximab (CD20)¹¹. These drugs have revolutionized the treatment of IMIDs due to their efficacy, speed of onset, and tolerability¹.

Although vitally important, biologics are expensive, which results in restricted access for many patients. This not only adversely affects patients’ health, but limits physicians’ options as there is a need to control public health expenses¹². The development of biosimilars builds upon the expensive early discovery and development trials of the reference medicine. As a result, biosimilars are less costly to develop and their introduction drives economic competition. Resulting price reductions have increased patient access to more affordable medicines¹³. Furthermore, potential cost savings, as well as competition from biosimilars, will ultimately promote innovation and thus benefit healthcare systems around the world¹⁴.

Biosimilars, therefore, have become an increasing focus of development¹⁵. The first biosimilar (Omnitrope¹) was approved by the European Medicines Agency (EMA) in 2006¹⁶. A decade later, more than 20 biosimilars are available in Europe, comprising hormones, erythropoietins, granulocyte colony-stimulating factors (G-CSFs), insulins, and TNFis, including biosimilars to infliximab (Inflectra²/Remsima³) and etanercept (Benepali⁴), which were approved by the EMA in 2013 and 2015, respectively^17,18. Furthermore, in 2015, a biosimilar G-CSF (Zarxio⁵) became the first biosimilar approved by the US Food and Drug Administration (FDA)¹⁹, paving the way for the US approval of biosimilar infliximab (Inflectra²) in 2016.

The aim of this review is to describe the inherent variability that is natural to biologics and, using the proposed etanercept biosimilar (GP2015) as an example, provide details on the type and extent of analytical, preclinical, and clinical data that are considered by regulatory agencies with respect to the approval of biosimilars. Relevant English-language articles, reviews, abstracts, and patents were identified through a search of PubMed/Medline and Google Scholar articles published between 1999 and November 2016. Broad search terms were used, including “biological products”, “biosimilar”, “etanercept”, “extrapolation”, “GP2015”, “immune-mediated inflammatory disease”, “psoriasis”, “rheumatoid arthritis” and “totality of the evidence”. Results of the literature review were supplemented with authors’ own data gathered during the target-directed development of GP2015.

Biosimilar development

The development of biosimilars utilizes a stepwise approach divided into several stages^17,20. The first stage is understanding and characterizing the reference medicine, without detailed knowledge of the respective cell bank or manufacturing process. During production, the manufacturer has to deliver consistent product quality to guarantee reproducible clinical performance. This means the variability of molecular and functional attributes that are important to clinical properties (i.e. critical quality attributes, [CQAs]) must be kept within acceptable ranges, as agreed by appropriate regulatory authorities. Current state-of-the-art analytical tools can detect even small changes in CQAs between batches of a product (or between products undergoing manufacturing changes). The manufacturer must scientifically justify that these differences are not clinically relevant. If they can’t, the post-manufacturing change medicine cannot be approved by regulatory agencies for use in humans. For example, following a change in manufacturing facilities of efalizumab, a formerly available antibody that was designed to treat autoimmune diseases, variations in the pharmacokinetic (PK) properties of the biologic were noticed. The FDA mandated that all efficacy and safety trials were repeated, thereby delaying FDA approval by 2 years²¹. Since state-of-the-art analytical methods can detect very minor structural and functional differences, which may have no clinical relevance, these laboratory tools are substantially more sensitive for identifying differences between two medicines than randomized clinical trials (RCTs). When commercial batches of etanercept sourced from the EU and US between 2007 and 2010 were analyzed using glycan mapping and cation exchange chromatography, the data revealed a highly consistent quality profile for batches having expiry dates until the end of 2009. However, the glycosylation profile of batches changed after this time, with a 20% decrease in the amount of variants containing the N-glycan G2F and a 25–30% increase in the amount of the basic variants²². More recently, the use of peptide mapping and mass spectroscopy allowed for the detection of incorrect disulphide bond variants that contribute to variability in the bioactivity of etanercept²³. Since a relatively large patient population has been exposed to a biologic containing these minor variations in quality attributes, it is reasonable to conclude that they do not have a detrimental effect on the product’s clinical safety and efficacy²².

During the first stage of biosimilar development, therefore, multiple batches of the reference medicine are characterized to understand structural and functional attributes, and determine the variability of post-translational modifications over time and from batch to batch. These data are then used to set the development target and boundaries of acceptable variability within which a biosimilar should fall. Differences beyond these boundaries may be acceptable if it can be shown that they will not change clinical properties. Otherwise, the product would not qualify as a biosimilar.

The second stage involves target-directed development of a manufacturing process for the biosimilar molecule. By using a comprehensive, state-of-the-art panel of analytical methods, each step of the manufacturing process is optimized to deliver a product that has the same structural and functional properties as the reference medicine.

The final stage is confirmation of high similarity, starting with comparison at the structural and functional level to determine whether the biosimilar molecule is essentially the same as the reference molecule. These analytical data serve as the foundation of the overall comparability exercise and totality-of-the-evidence concept²⁴. Next, preclinical properties (i.e. toxicology, tolerability and non-human pharmacology) are compared using relevant in vitro and in vivo models. Phase I PK and pharmacodynamic (PD) studies are then conducted, often initially in healthy volunteers (i.e. a population with a fully competent immune system) to demonstrate bioequivalent profiles. The final phase III confirmatory clinical efficacy and safety trial is then conducted in a relevant and sensitive enough population to allow detection of clinically meaningful differences between the biosimilar and reference medicine, should such differences exist.

If the biosimilar is proven to be highly similar to the reference medicine, as per regulatory requirements²⁵, then healthcare providers and patients can expect the same clinical outcomes with the biosimilar as the reference medicine across the same indications^26,27. The totality-of-the-evidence from all analyses and performed trials is used to approve the product and justify use of the biosimilar in all indications for which the reference medicine is approved, without the need for clinical safety and efficacy data from every approved indication or condition of use. This extrapolation is why a single confirmatory study can be conducted to address any residual uncertainty regarding clinical performance, rather than undertaking RCTs to demonstrate equivalence in every indication for which the reference medicine is licensed. It is important to note, however, that extrapolation is not automatically granted; it requires thorough scientific justification and is evaluated by regulators on a case-by-case basis, taking into account the totality-of-the-evidence and existing knowledge of the mechanism of action (MoA) of the medicine in each specific clinical indication^6,25,28. As with reference medicines, ongoing pharmacovigilance and post-marketing authorization studies are also required to identify rare adverse events (AEs) and monitor the long-term efficacy, safety and immunogenicity of biosimilars in every indication of use¹⁸.

Etanercept

Etanercept (Enbrel⁶) was among the first TNFis to be approved for use in rheumatic diseases²⁹. It is a dimeric human TNFα receptor (TNFR) fusion protein made of two extracellular regions of the human 75 KD (p75) TNFR, linked by the constant Fc portion of human immunoglobulin 1 (IgG1)³⁰. It contains IgG1-specific N-glycosylation sites in the Fc part and multiple O-glycans in the receptor part³¹.

Etanercept is licensed for treatment of RA, certain forms of juvenile idiopathic arthritis, AS, PsA and plaque psoriasis²⁹, having been extensively studied in RCTs involving almost 5000 patients with these diseases^32–43.

The primary MoA of etanercept across all therapeutic indications is competitive binding to soluble and/or membrane-bound TNFα, which is overexpressed in the aforementioned conditions⁴⁴. Etanercept competitively inhibits binding of TNFα to its receptors, thereby inhibiting TNFα-induced downstream signal transduction pathways that mediate inflammation.

Overall, used as monotherapy or in combination with methotrexate, etanercept offers rapid and sustained clinical treatment responses, reducing the signs and symptoms of disease. The most common AEs comprise injection site reactions and infections. The safety and efficacy of etanercept therapy has also been confirmed with long-term data derived from a number of real-world observational and registry studies^45,46. In addition, immunogenicity against etanercept is generally low; therefore, it may be a good option for patients who have developed antidrug antibodies (ADAs) against other biologics in the past⁴⁷.

Robust evidence supporting the sustained efficacy and safety of etanercept makes it an attractive molecule to target for biosimilar development.

Development of GP2015

Analytical characterization

As the manufacturer of a proposed biosimilar has no direct knowledge of the manufacturing process for the reference medicine, differences in the raw materials, equipment, processes, and process controls may exist. To define the development target for GP2015, multiple batches of reference etanercept were sourced at regular intervals from the US and EU, and characterized using state-of-the-art analytical methods. CQAs were ranked according to their potential importance (i.e. criticality) to PK/PD, efficacy, safety, and immunogenicity properties based on knowledge from the literature and clinical experience with the reference medicine. Based on this extensive knowledge and understanding of the reference medicine, the biosimilar manufacturing process for GP2015 was iteratively designed and optimized to develop a product with CQAs and biologic functions within the observed variability ranges of the reference medicine over time.

Reference etanercept is produced by recombinant DNA technology in a Chinese hamster ovary (CHO) mammalian cell expression system. The expression system used can affect the types of process- and product-related substances, impurities and variants that may be present in the protein product, therefore minimizing possible differences between the chosen expression system (i.e. host cell and the expression vector) of the proposed biosimilar and that of the reference medicine will enhance the likelihood of producing a highly similar medicine.

A CHO expression vector was therefore redesigned to allow expression of GP2015 from a single gene encoding the same amino acid sequence as reference etanercept. After cloning in different CHO cell types and evaluation of hundreds of subclones, a final subclone was selected based on product quality, titer, and genetic integrity and stability. Master and working cell banks of the final clone were established and a manufacturing process typical for monoclonal antibodies and fusion proteins was developed, finalized, and validated.

Two strengths of GP2015 were developed, consistent with the reference medicine: 25 mg/0.5 mL and 50 mg/1.0 mL. Due to intellectual property reasons, the GP2015 formulation contained citrate and lysine buffer components instead of phosphate and arginine, which are used in the reference formulation.

With an understanding of which quality attributes are clinically relevant, a comprehensive, three-way analytical similarity assessment was performed to gain a comprehensive understanding of the characteristics of GP2015 and reference etanercept sourced from the EU and US. The quality attributes of GP2015 and reference etanercept were compared based on physicochemical and in vitro functional assays.

Biochemical attributes such as primary structure, higher-order structure (secondary and tertiary), carbohydrate structure, size, charge, hydrophobicity, and other attributes, including product-related variants and impurities, were evaluated using various orthogonal methods. First, reducing peptide mapping was performed using the digestion enzyme, endoproteinase Lys-C. The resulting peptide fragments were separated by reversed-phase high-performance liquid chromatography (HPLC) on an Acquity⁷ ultra-performance liquid chromatography BEH300 column employing a gradient of acetonitrile containing 0.1% trifluoroacetic acid (TFA). Ultraviolet (UV) absorption at 215 nm was recorded by a UV detector. The results of the peptide mapping showed the amino acid sequences of GP2015 and reference etanercept to be identical; a prerequisite for the determination of biosimilarity (Figure 2(A)).

Figure 2. Structural analysis of GP2015 and reference etanercept. (A) Comparison of UV chromatograms of Lys-C digested GP2015 and reference etanercept sourced from the EU and US. (B) Overlay of one GP2015 (blue) and reference etanercept (grey) batch bound to TNFα, as determined by X-ray crystallography.

Correct folding of a protein, including secondary, tertiary, and quaternary protein structure, is crucial to triggering the same biological functions. To compare the protein structures of GP2015 and reference etanercept, Fab fragments were generated by papain digestion of the full-length dimeric fusion proteins. Fc fragments were removed by affinity chromatography using a Protein A column and the receptor component incubated with TNFα. The yielded complex was further purified on a Superdex 200 gel filtration column (GE Healthcare Life Sciences, Chicago, IL, USA), after which fractions containing the TNFα/etanercept receptor complex were additionally concentrated using an ultrafiltration device with a molecular weight cutoff of 50 kilodalton (Vivascience, Hannover, Germany). Obtained crystals were flash frozen and measured at a temperature of 100 Kelvin. X-ray diffraction data sets (space group H 3 2) were collected at the Swiss Light Source (Villigen, Switzerland) using cryogenic conditions. Data processed using X-ray Detector Software and XSCALE programs showed that the higher-order structure of GP2015 and reference etanercept were indistinguishable (Figure 2(B)).

Protein content, which is essential for the clinical efficacy of a biologic, was accurately determined under native conditions at a wavelength of 279 nm using a UV/visible spectral photometer. Quantitation was performed according to the Lambert–Beers law, using an experimentally determined extinction coefficient of ɛ = 1.15 cm²/mg. Results showed that the protein content of GP2015 was within the range for reference etanercept sourced from the EU and overlapped with the range for reference etanercept sourced from the US (Figure 3(A)).

Figure 3. Level of (A) protein content, (B) alpha-galactosylated N-glycan, (C) incorrect disulphide bond, (D) high molecular weight, and (E) low molecular weight variants in GP2015 and reference etanercept sourced from the EU and US.

Because product-related impurities can impact the efficacy, safety and/or immunogenicity of a biologic, they should be controlled and maintained at as low a level as possible. Etanercept molecules that contain alpha-galactosylated N-glycans are considered potential immunogenic impurities, which must be closely monitored. A normal-phase, HPLC analysis confirmed that the alpha-galactosylated N-glycans of GP2015 were within the range of reference etanercept and below the upper limit defined by the maximum amount of impurities measured in batches of reference etanercept (Figure 3(B)).

Incorrect disulphide bond variants, also termed “wrongly bridged variants”, and size variants also represent product-related impurities of etanercept. The amount of incorrect disulphide bond variants in GP2015 was determined by non-reducing peptide mapping employing a tryptic digestion step (Figure 3(C)). Separation was performed using an Ascentis⁸ Express Peptide ES-C18 HPLC column with a gradient of 0.1% TFA in water, as well as 0.1% TFA in 90% acetonitrile/10% water as mobile phases. UV absorption at 215 nm was recorded by a UV detector. Size variants, comprising degradation and aggregation products, were detected and quantified using size-exclusion chromatography under native conditions on a TSKgel⁹ G3000SWXL column using 150 mM K-phosphate pH 6.5 for elution, coupled with UV detection (Figure 3(D,E)). In all cases, the level of impurities in GP2015 was always well below the maximum amount found in batches of reference etanercept.

Overall, GP2015 contained essentially the same active ingredient as reference etanercept in terms of primary structure, secondary, and tertiary structure, molecular mass/size, charge, protein content, glycosylation, size, and amino acid modifications.

Furthermore, each of the three products demonstrated similar binding affinity for TNFα, Fc receptors (FcγRI, FcγRII, FcγRIII, and FcRn), and C1q, and had similar bioactivity in terms of apoptosis inhibition⁴⁸. Binding and functional neutralization of TNFα and TNFβ, which reflects the primary mechanism of action of etanercept^44,49,50, was measured using HEK293 cells stably transfected with an NFκB-luciferase reporter gene (Figure 4). In a microtiter plate, 20,000 cells per well were stimulated with 4 ng/mL TNFα or TNFβ in the presence of graded amounts of GP2015 or reference etanercept. After overnight incubation, cells were lyzed and luciferase activity quantified using a luminogenic substrate. The relative potency of GP2015 and reference etanercept was calculated by comparison with a standard.

Figure 4. TNF neutralization activity of GP2015 and reference etanercept sourced from the EU and US.

These analytical studies not only provide evidence for a high level of comparability between GP2015 and reference etanercept, but also confirm that the characteristics of reference etanercept sourced from the EU and US are indistinguishable.

Preclinical program

The next level contributing to the totality-of-the-evidence is preclinical comparison using relevant in vitro and animal models (Figure 1). The preclinical program of GP2015 consisted of five studies⁴⁸: a dose-finding PD study in mice and four comparative studies of PD, PK, and toxicity, including local tolerability and immunogenicity, between GP2015 and reference etanercept (Table 1). As reported by Hofmann et al.⁴⁸, the preclinical comparability exercise confirmed that GP2015 and reference etanercept were similar with regard to their PK and toxicological profiles.

Table 1. GP2015 preclinical development program.

	Species/Strain	Toxicokinetics/pharmacokinetics	Pharmacodynamics/efficacy	Local tolerability	Toxicity	Immunogenicity
Noncomparative study to select therapeutically sensitive dose	Human TNFα transgenic mouse model (Tg197)		X
GP2015 vs. reference etanercept at therapeutically sensitive dose	Human TNFα transgenic mouse model (Tg197)		X
Comparative pilot study to support selection of formulation, single dose	Rabbits/Himalayan	X		X
Comparative study, single dose	Rabbits/Himalayan	X		X
Comparative pivotal toxicity, 4 week treatment, multiple dose	Monkeys/cynomolgus	X		X	X	X

Results from the PK studies in rabbits showed almost complete overlap of the serum concentration–time curves for GP2015 and reference etanercept from the EU and US⁴⁸. In a transgenic mouse model of polyarthritis, the effects of GP2015 and reference etanercept on age-related arthritis progression (as assessed by morphological and functional changes on both ankle joints) were indistinguishable. During the course of the comparative study, GP2015 and reference etanercept (both administered intraperitoneally at 10 mg/kg) had similar activity in terms of inhibiting arthritic pathology and the underlying histopathology⁴⁸.

A study in cynomolgus monkeys showed comparable toxicological profiles, with injection-site reactions presenting as the main toxicity for both GP2015 and reference etanercept. Histopathological changes at the injection site generally correlated with those animals demonstrating decreased exposure and the production of ADAs, and was therefore considered to be related to immunogenicity. Additional evidence of immunogenicity was observed around Day 28 in one GP2015-treated and two reference-treated monkeys; the symptoms resolved within 3 days⁴⁸.

The results from the preclinical studies reduce any residual uncertainties and support the totality-of-the-evidence demonstrating similarity between GP2015 and reference etanercept.

Clinical confirmation of biosimilarity

For a biosimilar, the aim of clinical studies is not to establish safety and efficacy de novo, as this has already been done with the reference molecule, but to confirm the “sameness” of the biosimilar to the reference medicine, which has already been established analytically and in preclinical models^20,51. The clinical development program for GP2015, which is the final component in the totality-of-the-evidence approach to biosimilar development (Figure 1), includes data from four, phase I PK studies in healthy volunteers, as well as data from a phase III confirmatory efficacy and safety study in patients with moderate-to-severe chronic plaque-type psoriasis (Table 2).

Table 2. Overview of the GP2015 clinical development program.

Study design	Population	Duration	Treatment
Randomized, double-blind, two-way crossover study	Healthy volunteers (n = 57)	3 months	Two single 50 mg subcutaneous injections of GP2015 or reference etanercept sourced from US
Randomized, open label, two-way crossover study	Healthy male volunteers (n = 51)	3 months	Two single 50 mg subcutaneous injections of GP2015 by autoinjector or prefilled syringe
Randomized, double-blind, two-way crossover study	Healthy volunteers (n = 54)	3 months	Two single 50 mg subcutaneous injections of GP2015 or reference etanercept sourced from EU
Randomized, double-blind, two-way crossover study	Healthy male volunteers (n = 54)	3 months	Two single 50 mg subcutaneous injections of GP2015 or reference etanercept sourced from EU

PK studies

PK studies are pivotal to confirming the similarity of the biosimilar to the reference medicine. For GP2015, the primary PK study was a randomized, double-blind, cross-over, single-center study, designed to compare the PK, safety, and immunogenicity profiles of GP2015 and reference etanercept in healthy adults^52,53. Treatment was administered as two single doses of 50 mg during two treatment periods, separated by a wash-out period of at least 35 days. Fifty-seven subjects were randomly assigned in a 1:1 ratio to receive GP2015 50 mg during period 1 and reference etanercept 50 mg during period 2, or vice versa.

Bioequivalence criteria for the primary endpoints were met, with area under the curve up to the last measurable concentration and maximum concentration of the drug after administration falling within the prespecified equivalence margin of 80 to 125%^52,53. A supportive PK study confirmed bioequivalence between GP2015 and reference etanercept, irrespective of whether treatment was administered by autoinjector or prefilled syringe^53,54.

Across all PK studies, GP2015 was generally well tolerated. The incidence of injection site reactions was low and similar in healthy volunteers after administration of GP2015 or reference etanercept⁵³.

Immunogenicity, as determined by the formation of anti-etanercept antibodies, was assessed using a three-step procedure comprising a validated electrochemiluminescence assay (screening and confirmatory assays) for binding antibodies and a validated competitive ligand-binding neutralization antibody assay. In line with recent guidance from the EMA and FDA, this approach used a single, highly-specific and sensitive assay to detect antibodies against both the biosimilar and the reference molecule, even in the presence of circulating drug product. Among all healthy volunteers (n = 216), anti-etanercept antibodies were detected in just three subjects on Day 65, after the second treatment period. All three subjects (enrolled in the same study) had received GP2015 in the first treatment period and reference etanercept in the second treatment period. Titers were near the detection limit and none of the detected binding anti-etanercept antibodies were neutralizing^52,53.

The assessment of immunogenicity and PK in normal, healthy volunteers further supports the demonstration of no clinically meaningful differences between the two medicines and completes the third stage of the totality-of-the-evidence pyramid (Figure 1).

Confirmatory study in patients with moderate-to-severe plaque-type psoriasis

The design of confirmatory studies to eliminate or reduce any residual uncertainty that the biosimilar will have the same predicted clinical performance as the reference medicine may be quite different from those used in pivotal studies of the reference medicine itself. The goal of the confirmatory study is to use a clinical indication, primary endpoint, and study duration that has the best chance of detecting differences in efficacy or safety differences between the biosimilar and reference product, should they exist. In addition, because it is challenging to evaluate human immunogenicity of molecules in vitro or in a preclinical setting, the confirmatory study should be in the patient population most susceptible to developing an immune response and immune-related AEs.

Plaque-type psoriasis was considered to represent a highly sensitive indication to detect potential differences between GP2015 and reference etanercept for a number of reasons. Firstly, the Psoriasis Area and Severity Index (PASI), which measures the redness, thickness, and scaliness of lesions, is a reliable and valid measure that is relatively easy to quantify, and consistent from investigator to investigator⁵⁵. As the most commonly used measure in previous studies of reference etanercept in psoriasis, the PASI not only allows comparison of the results of this study with those in the literature, but also has an adequately large effect size⁵⁶. The advantage of a population with a large effect size is that small differences in the administered molecules are more easily detected. Secondly, etanercept is used as monotherapy in psoriasis, which increases the likelihood of detecting any potential differences in immunogenicity between the two products. Administering a treatment as monotherapy reduces confounding factors and the risk of immunosuppression resulting from concomitant medications such as methotrexate, which are often used in the treatment of arthritic conditions such as RA. Finally, in psoriasis, a dose of 50 mg falls into the linear phase of the dose–response curve, in which differences in dose can be seen as a difference in efficacy⁴¹.

Therefore, a confirmatory study was conducted to demonstrate equivalent efficacy and similar safety and immunogenicity between GP2015 and reference etanercept in patients with moderate-to-severe chronic plaque-type psoriasis (ClinicalTrials.gov Identifier: NCT01891864)⁵⁷. Patients (n = 531) were randomized 1:1 to receive either GP2015 50 mg or reference etanercept 50 mg twice weekly for 12 weeks. At Week 12, patients who achieved at least a PASI 50 response were re-randomized to either continue the same treatment on a once-weekly dosing schedule, or to undergo a sequence of treatment switches between GP2015 and reference etanercept until Week 30 (Figure 5). In the extension phase of the study, patients continued to receive the last treatment that they had received at Week 30 for the remainder of the study until Week 52⁵⁷.

Figure 5. Design of confirmatory study in patients with moderate-to-severe plaque-type psoriasis.

The primary efficacy endpoint was the PASI 75 response rate at Week 12, since this measure is well understood and considered clinically meaningful⁵⁵. The key secondary efficacy endpoint was the percentage change from baseline in PASI score up to Week 12. Additional secondary endpoints included PASI 50, 75, and 90 response rates over time, and the Investigator’s Global Assessment of psoriasis, among others, which allows regulators to evaluate multiple endpoints taking the totality-of-the-evidence into consideration⁵⁷.

The PASI 75 response rate after 12 weeks of treatment was 73.4% with GP2015 and 75.7% with reference etanercept. As the 95% confidence interval (CI) for the treatment difference at Week 12 (-2.3 [95% CI -9.85 to 5.30]) was contained within the prespecified margin range (-18% to 18%), therapeutic equivalence was demonstrated⁵⁷. These data were corroborated by other efficacy outcomes that established both equivalent efficacy and no difference between long-term treatment with GP2015 and reference etanercept. The incidence of treatment-emergent AEs up to Week 52 was generally comparable between the treatment groups and no new or unexpected safety issues were reported⁵⁷. Furthermore, the incidence of ADAs during the study was low and consistent with that reported in other large-scale trials of etanercept in patients with psoriasis^40,41,57.

Data from this confirmatory efficacy and safety study, conducted in a sensitive, clinical population, confirms biosimilarity and contributes to the overall totality-of-the-evidence. By showing that GP2015 and reference etanercept are highly similar, the extensive body of analytical, preclinical, and clinical data provides a robust scientific bridge between GP2015 and clinical experience with reference etanercept, and supports its extrapolated use across all licensed indications. In August 2016, after considering the totality-of-the-evidence, the US FDA approved GP2015 (Erelzi¹⁰, etanercept-szzs) for all indications included in the US-licensed reference medicine’s label.

Shortly after, in September 2016, the US FDA also approved ABP 501 (Amjevita¹¹; adalimumab-atto), a proposed biosimilar to adalimumab, based on the totality-of-the-evidence submitted by its manufacturer. The safety and efficacy of ABP 501 was also investigated in patients with moderate-to-severe plaque psoriasis. Data from this phase III confirmatory trial showed that overall efficacy, safety, and immunogenicity were comparable among patients treated with either ABP 501 or the reference medicine (Humira¹²). Furthermore, there were no differences in safety and immunogenicity between patients who transitioned from the reference medicine to the biosimilar at Week 16 compared with those who continued treatment with the reference medicine for the duration of the 52 week study⁵⁸.

Conclusions

Historically, physicians have been trained to primarily look at clinical data when making choices about which drug to use. For biosimilars, however, physicians need to be aware that the complete data package (analytical, preclinical, and clinical data) is considered by regulatory agencies when concluding whether a proposed biosimilar is approvable as a biosimilar or not. In an environment where so many other treatment options, including newer biologics and small molecule inhibitors, are available or in development, the uptake of biosimilars will encounter substantial barriers if this totality-of-the-evidence concept is not understood or explained.

Biosimilars have the potential to allow more patients to benefit from treatment in all approved indications. Their availability could also free up funds that could be used to support innovative research into potential targets for new drugs, the clinical effects of treating patients earlier in the development of their disease, and improving the treatment of chronic diseases through the investigation of flexible dosing schedules. Biosimilars also offer an opportunity to learn more about biologic medicines in general; for example, how changes in the CQAs of a biologic correlate with clinically relevant outcomes.

The extent to which biosimilar products are used in daily clinical practice will largely depend on the confidence placed in them by physicians, nurses, patients, payers, pharmacists, and regulators. The decision to use a biosimilar or reference medicine should be multifactorial and shared between all stakeholders; i.e. not based solely on financial grounds. This is particularly true of countries where biomimics or biocopies are marketed without totality-of-the-evidence available to conclude bioequivalence. For the benefits of biosimilars to be fully realized, high-quality, comprehensive data on the real-world effectiveness, safety, immunogenicity, and value for money of biosimilars and reference medicines are needed. To assist with this, an intensive pharmacovigilance monitoring system must be established. In addition, clear education is needed to ensure that prescribing decisions are based on a sound understanding of the science supporting the development and approval of biosimilars, rather than perception.

Transparency

Declaration of funding

Medical writing assistance for this manuscript was paid for by Hexal AG, a Sandoz company. None of the authors received financial compensation for authoring the manuscript.

Author contributions: The authors alone are responsible for the content and writing of the paper and agree to be accountable for all aspects of the work. Employees of Sandoz who did not meet authorship criteria were responsible for reviewing the manuscript for scientific accuracy only. All authors participated in the collection, analysis and interpretation of the data, and in the writing, reviewing and approval of the final version.

Declaration of financial/other relationships

V.S. has disclosed that she has received honoraria from AbbVie, Alder, Amgen, BMS, Celgene, Genentech, Janssen, Novartis, Pfizer and UCB. G.G. has disclosed that he has received honoraria from Abbvie, Abiogen, Almirall, Amgen, Biogen, Boehringer Ingelheim, Celgene, Galderma, Janssen, Eli Lilly, Hospira, Leo Pharma, Merck, Mundipharma, MSD, Novartis, Pierre Fabre, Pfizer Regeneron, Sandoz and Sun Pharma. M.S., R.E.M. and H.F.-Q. have disclosed that they are paid employees of Hexal AG, a Sandoz company. M.M. has disclosed that he was a paid employee of Hexal AG, a Sandoz company, at the time of manuscript preparation.

CMRO peer reviewers on this manuscript have received an honorarium from CMRO for their review work, but have no relevant financial or other relationships to disclose.

Acknowledgments

Medical writing support was provided by Fiona Bolland, Spirit Medical Communications, funded by Hexal AG, a Sandoz company.

Notes

Notes

1 Omnitrope is a registered trade name of Sandoz GmBH, Holzkirchen, Germany

2 Inflectra is a registered trademark of Hospira, a Pfizer Company, Lake Forest, IL, USA

3 Remsima is a registered trademark of Celltrion Healthcare Co. Ltd., Incheon, South Korea

4 Benepali is a registered trade name of Biogen, Cambridge, MA, USA

5 Zarxio is a registered trade name of Sandoz Inc., Princeton, NJ, USA

6 Enbrel is a registered trade name of Amgen Inc., Thousand Oaks, CA, USA

7 Acquity is a registered trade name of Waters Corporation, MA, USA

8 Ascentis is a registered trade name of Sigma-Aldrich Company, MO, USA

9 TSKgel is a registered trade name of Tosoh Bioscience GmBH, Griesheim, Germany

10 Erelzi is a registered trademark of Sandoz GmBH, Holzkirchen, Germany

11 Amjevita is a registered trademark of Amgen Inc., Thousand Oaks, CA, USA

12 Humira is a registered trademark of AbbVie Inc., Chicago, IL, USA