FormalPara Key Summary Points

ABP 938 is a biosimilar candidate of aflibercept reference product (RP).

This comparative similarity assessment employed orthogonal analytical techniques to detect potential structural and functional differences between ABP 938 and aflibercept RP, including primary structure, higher-order structure and biophysical properties, product-related substances and impurities, thermal stability and forced degradation, general properties, and biological activities.

The results demonstrated that ABP 938 is structurally and functionally similar to aflibercept RP.

Introduction

Retinal diseases such as neovascular (wet) age-related macular degeneration (nAMD) are leading causes of severe vision impairment around the world [1,2,3]. The discovery that vascular endothelial growth factor (VEGF) is a key mediator involved in the pathogenesis of angiogenic eye disorders [4] led to the introduction of anti-VEGF treatments. Commonly used anti-VEGF agents such as aflibercept have revolutionized the management of various retinal diseases and enhanced patient quality of life [5,6,7,8].

Aflibercept reference product (RP) is a recombinant fusion protein consisting of extracellular Domain 2 of human VEGF receptor-1 (VEGFR-1) and Domain 3 of VEGFR-2 fused to the fragment crystallizable (Fc) portion of human immunoglobulin G1. It acts as a soluble decoy receptor that binds VEGF-A and placental growth factor (PlGF), inhibiting the binding and activation of their cognate VEGF receptors and blocking the key signaling pathway responsible for angiogenesis and vascular leakage [9, 10]. ABP 938 is being developed as a biosimilar to aflibercept RP. To gain approval for a biosimilar, the United States (US) and European Union (EU) regulatory guidelines recommend a stepwise approach in developing the evidence needed to demonstrate biosimilarity, including structural and functional characterization, nonclinical evaluation, human pharmacokinetic and pharmacodynamic data, clinical immunogenicity, and comparative clinical efficacy and safety results [11,12,13,14]. The availability of biosimilars can potentially provide healthcare providers with additional treatment options, increase patient access to lower cost medications, and facilitate more equitable health outcomes [15].

This analytical similarity assessment was designed to characterize the structural and functional attributes of ABP 938 and aflibercept RP and to evaluate whether any differences have the potential to meaningfully impact clinical efficacy and safety.

Methods

Test Materials

ABP 938 was manufactured by Amgen Inc. Aflibercept RP was sourced from the US (Eylea®, Regeneron Pharmaceuticals, Inc.) and the EU (Eylea®, Bayer AG) as recommended in the regulatory guidance documents for biosimilar development [12, 14]. The RPs were stored and handled according to the manufacturer’s instructions and tested as part of the analytical similarity assessment plan.

Methods Summary

The methods used for the analytical similarity assessments were selected on the basis of knowledge of the structure, function, and expected structural heterogeneity of aflibercept RP and ABP 938, including characteristics critical to biological activity, safety, and stability. The testing plan (Table 1) was designed to assess primary structure, higher-order structure (HOS) and biophysical properties, product-related substances and impurities, general properties, thermal stability and forced degradation, and biological activity. Details of experimental methods are included in the Supplementary Materials.

Table 1 Testing plan and analytical methods for the physicochemical and functional similarity assessment of ABP 938 and aflibercept RP
Full size table

ABP 938 lots used in this study were acquired from development, clinical, and process validation production runs. Aflibercept RP lots acquired from the US and EU represented the product profile over multiple years to evaluate process variability during the course of ABP 938 development. Testing of aflibercept RP lots was performed shortly after receipt of the materials. For attributes expected to change as a function of time, aflibercept RP lots were tested again at or near the end of its shelf-life.

This article does not describe any studies with human participants or animals. All laboratory health and safety procedures have been complied with in the course of conducting this experimental work.

Results

Physicochemical similarity assessment results for ABP 938 and aflibercept RP are presented in Table 2. Functional similarity assessment results are shown in Table 3.

Table 2 Summary of physicochemical similarity assessment results on selected attributes for ABP 938 and aflibercept RP
Full size table
Table 3 Summary of biological activities of ABP 938 and aflibercept RP
Full size table

Primary Structure

The similarity between ABP 938 and aflibercept RP primary structures was assessed using a number of state-of-the-art methods, including intact molecular mass by matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF-MS), reduced and deglycosylated molecular mass by electrospray ionization time-of-flight mass spectrometry (ESI-TOF-MS), and reduced peptide mapping with liquid chromatography–mass spectrometry (LC–MS). Aflibercept is a dimeric glycoprotein with a molecular mass of 115 kilodaltons (kDa) [13]. The intact mass analysis results demonstrated that ABP 938 and aflibercept RP had similar intact molecular masses (Table 2). Reduced and deglycosylated molecular mass analyses confirmed that ABP 938 and aflibercept RP had similar molecular masses that were within ± 100 parts per million (ppm) of the theoretical mass (Table 2). In addition, the deconvoluted mass profiles were similar and there were no new species observed for the reduced deglycosylated ABP 938 as compared to aflibercept RP (Fig. 1a). The major peaks corresponded to the expected theoretical masses, which suggested that the C-terminal lysine residues were processed and the conversion of five asparagine residues to aspartic acid residues as a result of removal of the five N-glycans by peptide:N-glycosidase F (PNGase F). There were no detectable levels of N-terminal variants observed in ABP 938; however, partially processed N-terminal signal peptides were detected in aflibercept RP (Table 2). This minor difference is not expected to impact clinical performance because both ABP 938 and aflibercept RP had similar biological activities (see “Biological Activity” section). Finally, the primary amino acid sequence was determined by overlapping peptide map with tandem MS. The results confirmed that ABP 938 and aflibercept RP have the same amino acid sequence. Moreover, the peptide map profiles were visually similar, and no new peaks above the integrity limit were detected in ABP 938 as compared to aflibercept RP (Fig. 1b).

Fig. 1
figure 1

Primary structure of aflibercept (EU), ABP 938, and aflibercept (US). a Reduced and deglycosylated mass profile by ESI-TOF-MS. *Adducts are artifacts of ionization. b Reduced peptide map by LC–MS. ESI-TOF-MS electrospray ionization time-of-flight mass spectrometry, LC-MS liquid chromatography–mass spectrometry

Full size image

ABP 938 and aflibercept RP displayed similar extinction coefficient values as determined by amino acid analysis using ultraviolet (UV) spectroscopy, and a similar identity as evaluated by a dual anti-idiotype enzyme-linked immunosorbent assay (ELISA) (data not shown).

Higher-Order Structure and Biophysical Properties

The HOS and biophysical properties of ABP 938 and aflibercept RP were assessed using multiple techniques. The secondary structure was characterized by Fourier transform infrared (FTIR) spectroscopy and showed strong bands in the spectra at 1639 cm−1 and 1690 cm−1, indicative of a β-sheet secondary structure (Fig. 2a). The tertiary structure was characterized by near-UV circular dichroism (CD) spectra which contained signals from tryptophan, tyrosine, and phenylalanine superimposed on the broad disulfide signal from 240 to 350 nm (Fig. 2b). The spectral similarity value of ABP 938, aflibercept (US), and aflibercept (EU) met the pre-defined similarity assessment criteria of ≥ 95% for both FTIR and near-UV CD spectroscopy (Table 2).

Fig. 2
figure 2

Higher-order structure (HOS) and biophysical properties of aflibercept (EU), ABP 938, and aflibercept (US). a Secondary structure FTIR spectra. b Near-UV CD spectra. c DSC thermograms. d SE-HPLC-LS profiles. AU absorbance units, FTIR Fourier transform infrared spectroscopy, UVCD ultraviolet circular dichroism, DSC differential scanning calorimetry, SE-HPLC-LS size exclusion high-performance liquid chromatography-light scattering

Full size image

Thermostability was characterized by differential scanning calorimetry (Fig. 2c). ABP 938 and aflibercept RP displayed similar thermal melting transition temperatures at approximately 68.1 °C for Tm1 and 85.0 °C for Tm2 (Table 2). An absence of submicron particles in ABP 938 and aflibercept RP was confirmed by dynamic light scattering (Table 2). In addition, ABP 938 and aflibercept RP had similar levels of monomer, with no detectable aggregates assessed by sedimentation velocity analytical ultracentrifugation (SV-AUC) (Table 2). Finally, ABP 938 and aflibercept RP had similar molar masses for the monomer and high molecular weight (HMW) species as evaluated by size exclusion high-performance liquid chromatography (SE-HPLC) coupled with light scattering detector (Fig. 2d). ABP 938 had a lower level of HMW species compared to aflibercept RP (Fig. 2d). The theoretical masses of aflibercept RP monomer and dimer were approximately 115 kDa and 229 kDa, respectively, and the molar masses for monomer and HMW of ABP 938 and aflibercept RP were within the theoretical molar masses (Table 2). The molar masses of HMW observed in ABP 938 and aflibercept RP were consistent with that of dimers.

Product-Related Substances and Impurities

ABP 938 and aflibercept RP were evaluated for product-related substances and impurities. Size and charge variants are indicative of the stability of the products, which may change over time at the recommended storage conditions. ABP 938 and aflibercept RP had visually similar HMW and main peak profiles with no new or missing peaks as assessed by size exclusion ultrahigh-performance liquid chromatography (SE-UHPLC) (Fig. 3a). The stability data demonstrated that under the recommended storage conditions, ABP 938 maintained a lower level of HMW species compared to aflibercept RP for up to 36 months (Fig. 3a inset).

Fig. 3
figure 3

Product-related substances and impurities of aflibercept (EU), ABP 938, and aflibercept (US). a SE-UHPLC profiles; SE-UHPLC HMW stability at recommended storage condition (inset). b rCE-SDS profiles; rCE-SDS main peak stability at recommended storage condition (inset). c nrCE-SDS profile; nrCE-SDS main peak stability at recommended storage condition (inset). AU absorbance units, HMW high molecular weight, kDa kilodaltons, nrCE-SDS non-reduced capillary electrophoresis–sodium dodecyl sulfate, rCE-SDS reduced capillary electrophoresis–sodium dodecyl sulfate, SE-UHPLC size exclusion ultrahigh-performance liquid chromatography

Full size image

Size variants were assessed by capillary electrophoresis–sodium dodecyl sulfate (CE-SDS) in reducing and denaturing conditions. This method is suitable for quantifying fragments and the purity of the samples. ABP 938 and aflibercept RP had visually similar total pre-peak (fragmentation) and main peak (purity) profiles, with no new or missing peaks as assessed by reduced CE-SDS (rCE-SDS) (Fig. 3b). The stability data demonstrated that the level of rCE-SDS main peaks for ABP 938 remained higher than aflibercept RP for up to 36 months (Fig. 3b inset). Size variants assessed by non-reduced CE-SDS (nrCE-SDS) showed that ABP 938 and aflibercept RP had visually similar profiles with no new or missing peaks (Fig. 3c). Size variants included fragments and partial molecules; however, partial molecule species were not observed in either ABP 938 or aflibercept RP samples. The pre-peaks observed are fragmented species, which were present at low levels. The stability data demonstrated that the levels of main peaks were similar for ABP 938 and aflibercept RP (Fig. 3c inset).

General Properties

Aflibercept RP is supplied in a single-dose vial and a single-dose pre-filled syringe (PFS) designed to deliver 0.05 mL at 40 mg/mL [16]. ABP 938 and aflibercept RP demonstrated similar product strength as assessed by protein concentration and volume in both vial and PFS presentations (Table 2).

Biological Activity

Aflibercept RP acts as a soluble decoy receptor that binds VEGF-A and PlGF, thereby inhibiting the binding and activation of its cognate VEGF receptors [10, 16]. The VEGFR domain-mediated binding and cell-based functional studies demonstrated that ABP 938 was similar to aflibercept RP with respect to inhibition of VEGF-A165-mediated proliferation in human umbilical vein endothelial cells (HUVEC), VEGF-A165/VEGFR-2 signaling inhibition in HUVEC, VEGF-A165/VEGFR-2 signaling inhibition in HEK293 cells, and VEGF-A165/VEGFR-2 inhibition of binding (Fig. 4a–d). Additionally, the data suggested that ABP 938 and aflibercept RP were similar in PlGF-1/VEGFR-1 dimerization inhibition in human osteosarcoma U2OS cells, PlGF-2/VEGFR-1 dimerization inhibition in U2OS cells, PlGF-1 and PlGF-2 binding, and PlGF-1/VEGFR-1 inhibition of binding (Fig. 5a–e).

Fig. 4
figure 4

Inhibition of VEGF-A-mediated biological activities of aflibercept (EU), ABP 938, and aflibercept (US). a Relative inhibition of VEGF-A165-mediated HUVEC proliferation. b Relative VEGF-A165/VEGFR-2 signaling inhibition in HUVEC. c Relative VEGF-A165/VEGFR-2 signaling inhibition in HEK293 cells. d Relative inhibition of VEGF-A165/VEGFR-2 binding. HEK293 human embryonic kidney 293 cells, HUVEC human umbilical vein endothelial cells, VEGF-A vascular endothelial growth factor A, VEGFA165 vascular endothelial growth factor A165 isoform, VEGFR-2 vascular endothelial growth factor receptor 2

Full size image
Fig. 5
figure 5

Inhibition of PlGF-mediated biological activities of aflibercept (EU), ABP 938, and aflibercept (US). a Relative PlGF-1/VEGFR-1 dimerization inhibition in U2OS cells. b Relative PlGF-2/VEGFR-1 dimerization inhibition in U2OS cells. c Relative PlGF-1 binding. d Relative PlGF-2 binding. e Relative inhibition of PlGF-1/VEGFR-1 of binding. PlGF-1 and -2 placental growth factor 1 and 2, VEGFR-1 vascular endothelial growth factor receptor 1

Full size image

Aflibercept RP also binds to galectin-1 and VEGF-B but not to VEGF-C and VEGF-D [17, 18]. ABP 938 was similar to aflibercept RP with respect to galectin-1 and VEGF-B binding activity (Fig. 6a, b), and the absence of VEGF-C and VEGF-D binding activity (Table 3). Similar to aflibercept RP, ABP 938 showed a lack of antibody-dependent cell-mediated cytotoxicity (ADCC), antibody-dependent cellular phagocytosis (ADCP), and complement-dependent cytotoxicity (CDC) activities (Table 3).

Fig. 6
figure 6

Galectin-1 and VEGF-B binding activities of aflibercept (EU), ABP 938, and aflibercept (US). a Relative galectin-1 binding. b Relative VEGF-B167 binding. VEGF-B167 vascular endothelial growth factor B167 isoform

Full size image

The studies evaluating Fc domain-mediated binding activity demonstrated that ABP 938 and aflibercept RP had similar levels of neonatal fragment crystallizable receptor (FcRn) and Fc-gamma receptor 1 (FcγRI) binding. While ABP 938 had lower levels of complement component 1q (C1q) and other FcγR binding compared to aflibercept RP (Table 3), the differences are not considered critical for clinical performance because both ABP 938 and aflibercept lacked ADCC, ADCP, and CDC activities (Table 3).

Discussion

Comparative analytical similarity data from structural and functional studies of a proposed biosimilar and its RP provide a critical foundation for the assessment of biosimilarity [19]. This comparative similarity assessment employed orthogonal analytical techniques to evaluate physicochemical properties and functional activities to detect potential differences between ABP 938 and aflibercept RP.

The combined results of both intact glycosylated molecular mass analysis and reduced and deglycosylated molecular mass analysis provided evidence that ABP 938 and aflibercept RP are dimeric glycoproteins with a protein molecular weight of 97 kDa and contain glycosylation, resulting in a total molecular mass of 115 kDa. The polypeptide compositions were similar and consistent with the theoretical masses of ABP 938 and aflibercept RP. The reduced peptide map demonstrated that ABP 938 and aflibercept RP had the same primary amino acid sequence and similarly low levels of post-translational modifications, indicating that they have a similar primary structure. The HOS and biophysical properties are important for confirming structural integrity. The analysis of the HOS confirmed that ABP 938 and aflibercept RP have similar secondary and tertiary structures. Additionally, ABP 938 and aflibercept RP demonstrated similar thermal stability profiles. Furthermore, ABP 938 and aflibercept RP were similar for the molar masses of the monomer and HMW species.

Product-related substances and impurities of ABP 938 and aflibercept RP were assessed using an array of separation methods that evaluate size and charge variants. The SE-UHPLC profiles were similar with no new or missing peaks observed for ABP 938 and aflibercept RP. ABP 938 had a lower level of HMW species compared to aflibercept RP, and the levels of HMW for ABP 938 remained below those of aflibercept RP throughout the study. HMW species can have a potential impact on clinical safety and immunogenicity; however, the HMW observed in ABP 938 and aflibercept RP were consistent with that of dimers, and a lower level of HMW in ABP 938 would not be expected to impact clinical safety. The rCE-SDS profiles were similar between ABP 938 and aflibercept RP. ABP 938 showed a higher level of main peaks (i.e., purity), and conversely, a lower level of pre-peaks (i.e., fragmentation), when compared to aflibercept RP throughout the study. The nrCE-SDS profiles were visually similar. A low level of pre-peaks observed were likely fragmented species. ABP 938 and aflibercept RP showed similar levels of main peak over time under the recommended storage conditions.

VEGF family members, including VEGF-A, VEGF-B, VEGF-C, VEGF-D, and PlGF, are key contributors to vasculogenesis and angiogenesis and therefore crucial for embryonic development and vessel repair [20, 21]. Aflibercept is known to bind to VEGF-A, PlGF, and VEGF-B, but not to the VEGF-C or VEGF-D family members [22]. Among the VEGF family, VEGF-A is the most potent mitogen of angiogenesis [20]. Proliferation of vascular endothelial cells and angiogenesis is induced predominantly through VEGF-A-mediated VEGFR-2 activation [23]. Endothelial cell response to VEGF-A leads to neovascularization and vascular permeability that contributes to disease pathology [10]. At least nine VEGF-A isoforms have been identified. The VEGF-A111, VEGF-A121, VEGF-A145, VEGF-A165, VEGF-A189, and VEGF-A206 isoforms are the most commonly expressed transcripts [24, 25]. All VEGF-A isoforms bind both VEGFR-1 and VEGFR-2 through conserved VEGFR binding domains [26,27,28,29]. Because VEGF-A165 is the most extensively investigated and representative VEGF-A isoform, it was used to assess the functional activities of ABP 938 and aflibercept RP. ABP 938 and aflibercept RP demonstrated similar biological activities related to the mechanism of action in various assays (Fig. 4a–d).

PlGF acts as a paracrine factor upon activation of VEGFR-1, leading to proliferation, migration, and survival responses in endothelial cells [30, 31]. Although the contribution of PlGF in pathological ocular neovascularization in humans has not been fully elucidated [32], its expression is elevated in patients with angiogenic eye disorders [33]. In humans, PlGF exists as four splice isoforms that bind VEGFR-1 through conserved receptor-binding domains [34, 35]. The most relevant isoforms involved in ocular diseases are PlGF-1 and PlGF-2 [32, 36], whereas expression of PlGF-3 and PlGF-4 is primarily in the placenta and umbilical vein endothelial cells [37]. Therefore, PlGF-1 and PlGF-2 were used in orthogonal assays to demonstrate that ABP 938 and aflibercept RP have similar biological activities (Fig. 5a–e).

Galectin-1 directly binds to N-glycans on VEGFR-2 [16] and has been identified in neovascular eye tissues surgically excised from patients with angiogenic retinal disorders. Aflibercept blocks galectin-1-induced VEGFR-2 phosphorylation in human retinal microvascular endothelial cells [33]; however, the biological role of galectin-1 in the pathophysiology of eye disease remains to be determined. ABP 938 and aflibercept RP demonstrated similar galectin-1 binding activity.

VEGF-B binding is a known property of aflibercept RP [14]; however, VEGF-B has not been implicated in the pathogenesis of angiogenic eye disorders. It exists as two isoforms which bind to VEGFR-1 through conserved receptor binding domains [38]. VEGF-B167 is the predominant isoform expressed in most tissues and organs, accounting for more than 80% of the total VEGF-B transcripts, while VEGF-B186 is expressed at lower levels and in a limited number of tissues [39]. Therefore, VEGF-B167 was used to demonstrate that ABP 938 and aflibercept RP were similar in VEGF-B binding activity.

ABP 938 showed similar binding of FcRn and FcγRI and lower binding of C1q, FcγRIIa, FcγRIIb, FcγRIIIa, and FcγRIIIb as compared to aflibercept RP. However, the lower binding activities are not considered clinically meaningful because both ABP 938 and aflibercept had no Fc domain-mediated effector functions. The biological functions of aflibercept RP are primarily mediated through its VEGFR domains. The Fc domain of the molecule does not induce effector functions, as members of the VEGF protein family exist as predominantly soluble proteins [22]. Overall, the functional activity results indicate that ABP 938 is highly similar to aflibercept RP. It should be noted that the exhaustive structural and functional comparisons of ABP 938 with aflibercept RP that have been presented here are a first step in the development of ABP 938 as a biosimilar to aflibercept and provide the scientific basis for its further development in the clinical setting. Indeed, a comparative clinical study (NCT04270747) of ABP 938 and aflibercept RP in patients with nAMD has recently been completed. Publication of the results of this study is awaited and should provide evidence related to the clinical efficacy and safety of ABP 938. Taken together, these results should provide the totality of evidence for biosimilarity and form the basis for clinical decisions.

Conclusions

Biosimilars have the potential to provide patients with access to more affordable treatments for retinal diseases. Structural and functional characterization of a candidate biosimilar is an important step in generating the totality of evidence necessary for regulatory approval. In this comparative analytical similarity assessment, we investigated critical attributes of ABP 938 and aflibercept RP, including the primary structure, HOS, product-related substances and impurities, general properties, thermal stability and forced degradation, and biological activities. The results demonstrated that ABP 938 is structurally and functionally similar to aflibercept RP. Since aflibercept RP sourced from the US and EU are analytically similar, this allows for the development of a scientific bridge such that a single-source RP can be used in nonclinical and clinical studies.