Nonclinical Similarity of the Biosimilar Candidate ABP 938 with Aflibercept Reference Product

Abstract

Introduction

ABP 938 is being developed as a biosimilar to Eylea^® (aflibercept reference product [RP]), an anti-vascular endothelial growth factor (VEGF) drug used in the management of retinal diseases. Previously, a comparative analytical similarity assessment demonstrated that ABP 938 and aflibercept RP have the same amino acid sequence and exhibit similar higher-order structure and biological activity. The nonclinical studies described here were designed to assess the in vitro pharmacology and the in vivo pharmacokinetics (PK), toxicokinetics (TK), and safety profiles of ABP 938 compared to aflibercept RP.

Methods

In vitro target-binding kinetics and affinity for VEGF-A and placental growth factor (PIGF) isoforms were evaluated using surface plasmon resonance (SPR). Effector functions were assessed by cell-based assays. PK was evaluated in a nonterminal intravitreal (IVT) ocular distribution study in rabbits. Safety was assessed in a 1-month IVT study in cynomolgus monkeys.

Results

SPR results demonstrated that ABP 938 is similar to aflibercept RP in binding kinetics and affinity for VEGF-A₁₁₁, VEGF-A₁₂₁, VEGF-A₁₆₅, VEGF-A₁₈₉, PlGF-1, and PlGF-2 isoforms. No antibody-dependent cellular cytotoxicity, antibody-dependent cellular phagocytosis, or complement-dependent cytotoxicity was observed with ABP 938 and aflibercept RP. Results from the nonterminal ocular distribution study in rabbits indicated that there were no meaningful differences in the distribution kinetics between intravitreally injected ABP 938 and aflibercept RP. Additionally, there was no evidence of ocular or systemic toxicity associated with IVT administration of ABP 938 in a repeat-dose, 1-month toxicology study in cynomolgus monkeys; toxicokinetic and toxicology profiles were similar to aflibercept RP.

Conclusions

This integrated assessment of results from the in vitro pharmacology assessment and in vivo PK and TK/toxicology profiles formed the nonclinical portion of the totality of evidence demonstrating ABP 938 is a biosimilar to aflibercept RP.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40123-024-01043-5.

Plain Language Summary

Biological drugs, commonly referred to as biologics, have transformed the lives of millions of patients. They are made using highly complex manufacturing processes that involve living cells. Biosimilars are comparable to approved “originator” biologics (also known as reference products) in terms of potency, safety, and efficacy. ABP 938 is currently being developed as a biosimilar to aflibercept, an originator biologic used to treat certain eye diseases. We have conducted studies to compare the characteristics of ABP 938 and aflibercept in laboratory experiments and in animal models. The results from these studies indicate that there are no meaningful differences between ABP 938 and aflibercept reference product. Since then, a study of ABP 938 in humans with age-related macular degeneration has recently been completed, adding to the totality of evidence supporting ABP 938 as a treatment option for healthcare providers and patients in the future.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40123-024-01043-5.

Key Summary Points

Why carry out this study?

ABP 938 is being developed as biosimilar to aflibercept reference product (RP), an effective anti-vascular endothelial growth factor (VEGF) biologic agent used in the management of various retinal diseases.

The goal of the study was to assess the in vitro binding kinetics and affinity for certain VEGF and placental growth factor (PlGF) isoforms, and in vivo pharmacokinetics/toxicokinetics (TK) and toxicology profiles of ABP 938 compared to aflibercept RP.

What was learned from the study?

The results from binding kinetics and affinity assays, an ocular distribution study in rabbits, and a 1-month intravitreal (IVT) toxicology study in cynomolgus monkeys indicated that there were no meaningful differences between ABP 938 and aflibercept RP. There was no ocular or systemic toxicity associated with IVT administration of ABP 938.

These results provided a scientific basis for further development of ABP 938 in the clinical setting, with the overall goal of providing patients with an additional treatment option for retinal disease in the future.

Introduction

Aflibercept is a biologic agent that belongs to the pharmacologic class of vascular endothelial growth factor (VEGF) inhibitors and is approved for the intravitreal (IVT) treatment of ophthalmic diseases associated with retinal neovascularization [1–3]. It is a recombinant fusion protein consisting of portions of the extracellular domains of human VEGF receptor (VEGFR)-1 and -2 fused to the fragment crystallizable (Fc) portion of human immunoglobulin isotype class G subclass 1 (IgG1) [4–6]. Aflibercept acts as a soluble decoy receptor that binds to VEGF type A (VEGF-A) and placental growth factor (PlGF) and thereby inhibits the binding and activation of VEGFR-1 and VEGFR-2, two closely related receptor tyrosine kinases implicated in pathological neovascularization and excessive vascular permeability [1, 2, 7, 8].

A biosimilar is a biologic agent that is highly similar to and has no clinically meaningful differences in safety, purity, and potency from an approved biologic, often referred to as the reference product (RP) [8–12]. The demonstration of biosimilarity is based on a stepwise totality of evidence (TOE) approach, which includes as its foundation, analytical comparisons of structural and functional similarity by in vitro methodologies. When warranted, nonclinical in vivo studies including pharmacokinetics (PK), toxicokinetics (TK), and/or toxicology assessments may be required in addition to clinical studies comparing the efficacy, safety, immunogenicity, PK, and/or pharmacodynamics of the biosimilar to the RP.

ABP 938 (aflibercept-ayyh; Pavblu^™) is a biosimilar of aflibercept which received FDA approval in August 2024 [13]. In the past year, a number of other aflibercept biosimilars have also been approved for the treatment of retinal diseases, including aflibercept-yszy (Opuviz^™), aflibercept-jbvf (Yesafili^™), aflibercept-mrbb (Ahzantive^®), and aflibercept-abzv (Enzeevu^™) [14–17]. The nonclinical studies described here were conducted to evaluate the pharmacology, PK, TK, and toxicology profiles of ABP 938 in comparison to aflibercept RP.

Methods

Materials

ABP 938 was manufactured by Amgen Inc. Aflibercept RP was sourced from the United States (aflibercept [US]; Regeneron Pharmaceuticals, Inc.) and the European Union (aflibercept [EU]; Bayer AG). The reference products were stored and handled according to the manufacturer’s instructions. Aflibercept US and aflibercept EU were used as comparators for the in vitro studies. Aflibercept EU was used as the RP in the in vivo studies.

Animals

All animals were housed in an Association for Assessment and Accreditation of Laboratory Animal Care, International-accredited facility and cared for in accordance with the Guide for the Care and Use of Laboratory Animals, 8th Edition [18]. All research protocols (study numbers 8392437 and 8395840) were reviewed and approved by Labcorp (formerly Covance) and Amgen Institutional Animal Care and Use Committees.

Ten male New Zealand White rabbits (Covance Research Products, Inc., Alice, TX, USA) at least 4 months of age and weighing between 2.6 and 3.0 kg were individually housed in stainless steel caging and acclimated to study conditions for at least 6 days prior to dosing. Rabbits were fed Certified High-Fiber Rabbit Diet #5325 (PMI Nutrition International LLC., St Louis, MS, USA), with supplemental treats and fresh water provided ad libitum. Animals were provided various environmental and psychological enrichments. The animal room was set to maintain a temperature of 16 to 22 °C, a relative humidity of 50 ± 20%, and a 12-h light/12-h dark cycle. Upon completion of the in-life portion of the study, animals were returned to the existing colony. A board-certified veterinary ophthalmologist conducted ophthalmic examinations prior to dosing. All animal procedures followed Animal Welfare Act Regulations (9 CFR 3).

Twelve female cynomolgus monkeys (Covance Research Products, Inc., Alice, TX, USA) were acclimated to the test facility for 26 days. At initiation of dosing, animals were 27 to 35 months old, and body weights ranged from 2.1 to 2.4 kg. Animals were group-housed (two animals/cage) in stainless steel cages. Animals were offered Certified Primate Diet #5L4L (PMI Nutrition International LLC, St Louis, MS, USA) one to two times daily, unless fasted for study procedures, and had continuous access to clean water. Environmental controls were set to maintain a temperature range of 20 to 26 °C, a relative humidity range of 30 to 70%, eight or greater air changes/hour, and a 12-h light/12-h dark cycle. Animals were provided various cage-enrichment devices (e.g., a toy) and fruit, vegetable, or dietary enrichment per the CRO’s standard operating procedures. All animal procedures followed applicable animal welfare acts and were approved by the local Institutional Animal Care and Use Committee.

In Vitro Pharmacology Studies

The binding kinetics and affinity for VEGF A₁₁₁, VEGF A₁₂₁, VEGF A₁₆₅, VEGF A₁₈₉, PIGF -1, and PlGF-2 were evaluated by surface plasmon resonance (SPR). Details are available in the Supplementary Materials.

In Vivo Distribution Study in Rabbit Eye After IVT Injection

ABP 938 and aflibercept EU were conjugated to amine N-hydroxysuccinimide (NHS)-reactive Alexa Fluor^™ 488 (AF488; Invitrogen, Cat # 20100). Each molecule of ABP 938 and aflibercept was conjugated to 1 to 2 molecules of AF488 to minimize disruption of the structure and function of the conjugated molecules.

New Zealand White male rabbits were administered intramuscular (IM) flunixin meglumine prior to sedation with ketamine, dexmedetomidine, and atropine in preparation for IVT injection and ophthalmic procedures. Following application of a topical anesthetic, eyes were rinsed with an iodine solution followed by a saline rinse. Rabbits were administered IM buprenorphine upon recovery 5 to 7 h later, and again at least 16 h after the first administration. Bland ophthalmic ointment was administered to dosed eyes in conjunction with the second administration of buprenorphine.

On Study Day 1, anesthetized rabbits (n = 3/group) were administered by IVT injection in each eye, a single 1 mg/eye (25 µl/eye) dose of AF488-ABP 938 combined with ABP 938 or AF488-aflibercept combined with aflibercept EU in a 1:3 ratio. After dosing, fluorescence of AF488-conjugated proteins in the aqueous and vitreous humors were measured using a Fluorotron^™ Master Ocular Fluorophotometer (Ocumetrics Technology Corp., Calgary, Canada) [19]. A non-compartmental PK analysis was performed on the corrected concentrations from fluorescence scan signals from both eyes. Further details regarding the labeling of ABP 938 and aflibercept with AF488 is included in the Supplementary Materials.

Pharmacokinetic analysis was conducted using Phoenix^® WinNonlin^® Version 8.1 software (Certara USA, Inc., Princeton, NJ, USA). A non-compartmental approach, consistent with the IVT route of administration (IVT bolus for the vitreous humor and extravascular for the aqueous humor), was used for parameter estimation [20]. Statistical analyses included descriptive statistics such as mean and standard deviation. Negative ABP 938 and aflibercept concentrations were treated as zero for descriptive statistics and PK analysis. Details regarding statistical tests and criteria for biosimilarity were described previously [24]. The mean of all left eye (oculus sinister), right eye (oculus dexter), and both eyes’ (oculus uterque) ABP 938 and aflibercept concentrations for each animal, expressed as maximum observed concentration (C_max) of the fluorescence scan signal (Peak Method) and area under the fluorescence curve scan signal (Area Method) were used for analysis at each nominal sampling time point. At least three scans of each eye were performed. The intermittent Peak and Area Method data provided the concentrations and areas for each specific tissue at each nominal sampling time point. Non-compartmental analysis was subsequently applied to the resulting Peak and Area Method data for each animal across the nominal sampling time points to generate PK parameters for each tissue during the entire study. No terminal necropsies were performed, and rabbits were returned to the stock colony upon completion of the study.

IVT Toxicology Assessment in Cynomolgus Monkeys

A Good Laboratory Practice-compliant toxicology study was conducted in female cynomolgus monkeys at Covance Laboratories Inc. (Madison, WI, USA). Sterile 0.3-cc insulin syringes with attached 30-gauge × 1/2 -inch needles were prefilled using aseptic techniques with filtered and pooled 0.9% sodium chloride for injection as a negative control, ABP 938 (40 mg/ml) as the test article, or aflibercept EU (40 mg/ml per supplier specifications) as a comparator in the event of unexpected toxicity. Animals (n = 4/group) were administered IM flunixin meglumine and buprenorphine prior to sedation with ketamine and/or dexmedetomidine. A topical ocular anesthetic (0.5% proparacaine) was instilled and both eyes cleaned with a dilute povidone iodine solution followed by a sterile saline wash prior to IVT administration. Each dosing time was recorded as the time at the completion of the injection. Ophthalmic examinations, including indirect ophthalmoscope, slit-lamp biomicroscope and intraocular pressure (IOP) measurements, were conducted by a board-certified veterinary ophthalmologist on dilated eyes (1% tropicamide) once during the pre-dose phase and on study days 3, 15, 29 (pre-dose), 31, and 36. Full-field electroretinography (ERG) including scotopic, photopic, and visual evoked potential tests were conducted by trained personnel once during the pre-dose phase and once during week 5. Animals were closely monitored for clinical observations, abnormalities, and/or signs of pain or distress. Body weight and food consumption were monitored throughout the study. Blood was collected for clinical pathology assessment (hematology, clinical chemistry, coagulation, urinalysis) twice during the pre-dose phase and prior to necropsy. One week after the second dose, animals were anesthetized with sodium pentobarbital, exsanguinated, and necropsied. Terminal procedures included collection of select organ weights, macroscopic observations of external features and tissues, and collection of a standard full tissue list for histopathologic evaluation. After euthanasia and immediately following enucleation of the eye, a sample of vitreous fluid (target volume of 100 μl) was collected from each eye and frozen for future bioanalytical analysis. The eye was then injected with modified Davidson’s fixative, and the globe and optic nerves collected into modified Davidson’s fixative for 48 to 96 h and subsequently stored in 10% neutral-buffered formalin. Eye tissues were embedded and sectioned to facilitate examination of the fovea, injection sites, macula, optic disc, and optic nerve. Remaining tissues were embedded in paraffin and appropriately sectioned. All slides were stained with hematoxylin and eosin. Following completion of the primary microscopic evaluation by a board-certified veterinary pathologist, an independent peer review was conducted by a board-certified veterinary ophthalmic pathologist.

Bioanalytical samples were quantified using a validated electrochemiluminescent method capable of measuring ABP 938 and aflibercept. Toxicokinetic analysis was performed on the individual serum concentration–nominal time data using Phoenix^® WinNonlin^® Version 6.4.0.768 (Pharsight Corporation, Mountain View, CA, USA) to estimate the C_max, the time at which this was attained (T_max), and the area under the concentration–time curve (AUC) from time zero to the time of the last measurable concentration (AUC_168hr and AUC_672hr, as applicable). Details regarding the statistical analysis of the IVT toxicology assessment in cynomolgus monkeys can be found in the Supplementary Materials.

Results

Binding Kinetics and Affinity of ABP 938 and Aflibercept RP for VEGF-A and PlGF Isoforms

The VEGFR domain-mediated binding kinetics and affinity of ABP 938 and aflibercept (US and EU) were assessed using SPR. The binding kinetics and affinity results for ABP 938 and aflibercept RP are provided in Table 1. The data showed that ABP 938 and aflibercept RP have similar association rate constants (k_a) and minor differences in their dissociation constants (k_d) for VEGF-A₁₁₁, VEGF-A₁₂₁, VEGF-A₁₆₅, and VEGF-A₁₈₉. Representative SPR sensorgrams and model fits for each VEGF-A isoform are provided in Fig. 1.

Table 1.

Binding kinetics and affinity of VEGF-A isoforms for aflibercept EU, ABP 938, and aflibercept US

VEGF-A isoforms	Sample (n = 3)	ka (1/ms) 1 × 106	kd (1/s) 1 × 10–5	KD (pM)	Steady-state KD (nM)
VEGF-A111	Aflibercept EU	18.7 ± 0.1	12.0 ± 0.9	6.5 ± 0.5	0.222 ± 0.002
	ABP 938	19.5 ± 0.7	8.7 ± 1.2	4.5 ± 0.7	0.207 ± 0.010
	Aflibercept US	19.2 ± 0.8	11.2 ± 1.0	5.9 ± 0.7	0.221 ± 0.002
VEGF-A121	Aflibercept EU	6.0 ± 0.3	9.4 ± 0.4	16.1 ± 1.0	1.406 ± 0.445
	ABP 938	6.9 ± 0.4	7.4 ± 0.8	10.8 ± 1.5	1.069 ± 0.068
	Aflibercept US	6.6 ± 0.6	9.4 ± 0.5	14.3 ± 2.0	1.166 ± 0.095
VEGF-A165	Aflibercept EU	9.1 ± 0.1	9.5 ± 0.2	10.5 ± 0.2	0.640 ± 0.017
	ABP 938	9.4 ± 0.0	6.9 ± 0.5	7,4 ± 0.5	0.589 ± 0.037
	Aflibercept US	9.1 ± 0.1	9.4 ± 0.3	10.3 ± 0.2	0.648 ± 0.032
VEGF-A189	Aflibercept EU	15.0 ± 0.4	10.8 ± 0.3	7.2 ± 0.4	0.321 ± 0.005
	ABP 938	15.9 ± 0.3	8.5 ± 0.2	5.3 ± 0.1	0.305 ± 0.005
	Aflibercept US	15.0 ± 0.3	10.5 ± 0.4	7.0 ± 0.4	0.323 ± 0.002

Values are the mean ± SD (n = 3)

EU European Union, k_a association rate constant, k_d dissociation rate constant, K_D equilibrium dissociation constant, ms millisecond, n number, nM nanomolar, pM picomolar, s second, SD standard deviation, US United States, VEGF vascular endothelial growth factor

Fig. 1.

Representative SPR sensorgrams and model fit for binding of VEGF-A isoforms to aflibercept EU, ABP 938, aflibercept US. Representative SPR sensorgrams (black lines) and the results from non-linear least squares regression analysis of the data (red lines). Association and dissociation phase data were globally fit to a 1:1 binding model to determine the association rate coefficient (k_a), dissociation rate coefficient (k_d), and the maximal response (R_max) value for each interaction. Y-axis: response units, x-axis: time given in seconds (s). EU European Union, SPR surface plasmon resonance, US United States, VEGF vascular endothelial growth factor

Based on published studies, the inherent affinity of the interacting complex, instrument limitations, and method precision [21–23], kinetic rate and equilibrium binding constants within two-fold are considered similar within the context of SPR. The off rates for these interactions are considered too slow for the measurements to yield reliably accurate results [22]. To address this limitation, steady-state affinity measurements are provided in Table 1. ABP 938, aflibercept US, and aflibercept EU had comparable binding kinetics and affinity for all VEGF-A isoforms, including heparin-binding (VEGF-A₁₆₅ and VEGF-A₁₈₉) and non-heparin-binding (VEGF-A₁₁₁ and VEGF-A₁₂₁) isoforms. In addition, binding kinetics and affinity of ABP 938, aflibercept US, and aflibercept EU for PlGF-1 and PlGF-2 are similar and fit the 1:1 binding model (Table 2). Representative sensorgrams and fits for each PlGF isoform binding to ABP 938 and aflibercept RP are shown in Fig. 2. These VEGF-A and PlGF binding kinetics and affinity results support the conclusion that ABP 938 and aflibercept RP display similar ligand-receptor-binding activities associated with the inhibition of VEGF signaling.

Table 2.

Binding kinetics and affinity of PlGF isoforms for aflibercept EU, ABP 938, and aflibercept US

PlGF isoforms	Sample (n = 3)	ka (1/ms) 1 × 106	kd (1/s) 1 × 10–3	KD (nM)
PlGF-1	Aflibercept EU	4.09 ± 0.18	1.61 ± 0.05	0.39 ± 0.01
	ABP 938	4.25 ± 0.18	1.58 ± 0.02	0.37 ± 0.01
	Aflibercept US	4.20 ± 0.15	1.64 ± 0.04	0.39 ± 0.00
PlGF-2	Aflibercept EU	2.77 ± 0.13	1.17 ± 0.06	0.42 ± 0.00
	ABP 938	2.89 ± 0.03	1.21 ± 0.03	0.42 ± 0.01
	Aflibercept US	2.84 ± 0.13	1.16 ± 0.05	0.41 ± 0.00

Values are the global fits of 3 replicates

EU European Union, k_a association rate constant, k_d dissociation rate constant, K_D equilibrium dissociation constant, ms millisecond, n number, nM nanomolar, PlGF placental growth factor, s second, US United States

Fig. 2.

Representative SPR sensorgrams and model fit for binding of PlGF isoforms to aflibercept EU, ABP 938, and aflibercept US. SPR sensorgrams (black lines) and the results from non-linear least squares regression analysis of the data (red lines). Association and dissociation data were global fits to a 1:1 binding model to determine the association rate coefficient (k_a), dissociation rate coefficient (k_d), and the maximal response (R_max) value for each interaction. Y-axis: response units, x-axis: Time given in seconds (s). EU European Union, PlGF placental growth factor, SPR surface plasmon resonance, US United States

Lack of Effector Function for ABP 938 and Aflibercept RP

Functional assessment of ABP 938 included evaluation of antibody-dependent cellular cytotoxicity (ADCC), antibody-dependent cellular phagocytosis (ADCP), and complement-dependent cytotoxicity (CDC) activities with cell-based assays described previously [24]. As shown in Fig. 3a, ABP 938, aflibercept (US and EU), and the negative control induced a similar level of basal cell death (target and effector cells only), while the positive control showed the expected ADCC activity in peripheral blood mononuclear cells (PBMCs) at a 100-fold lower concentration. Similarly, ABP 938, aflibercept (US and EU), and the negative control showed a comparable level of phagocytosis with basal ADCP activity (target and effector cells), while the positive control displayed the expected ADCP activity in PBMCs at the 100-fold lower concentration (Fig. 3b). In addition, ABP 938, aflibercept (US and EU), and the negative control exhibited a similar number of viable cells, while the positive control showed the expected CDC activity (Fig. 3c). These results confirmed that ABP 938 and aflibercept RP have no Fc domain-mediated effector functions. Taken together, the results of the in vitro binding and cell-based assays demonstrated that ABP 938 and aflibercept RP have similar target-binding activity and lack of effector function.

Fig. 3.

Lack of a ADCC, b ADCP, and c CDC activities of aflibercept EU, ABP 938, and aflibercept US. a Graph presents luminescence assay signal as the ADCC activity measured at 100 μg/ml for aflibercept EU, ABP 938, and aflibercept US. Error bars indicate standard deviation. Mean ± SD of 4 replicates is plotted. PBMCs from an FcγRIIIα 158 V/F allotype donor (left panel) and FcγRIIIα 158 V/V allotype donor (right panel) were used. b Graph presents assay signal as the ADCP activity at 0.87 µM for aflibercept EU, ABP 938, and aflibercept US. Mean ± SD of four replicates for test samples, three replicates for control samples, and two replicates for effector cells alone are plotted. Error bars indicate standard deviation. c Graph presents the lack of CDC activity as measured by luminescence by live cells at 100 µg/ml for aflibercept EU, ABP 938, and aflibercept US tested with 20% rabbit serum. Mean ± SD of four replicates is plotted. Error bars indicate standard deviations. µg microgram, ADCC antibody-dependent cell-mediated cytotoxicity, ADCP antibody-dependent cellular phagocytosis, CD14 +   cluster of differentiation 14, CDC complement-dependent cytotoxicity, EU European Union, FcγR fragment crystallizable gamma receptor, ml milliliter, PBMCs peripheral blood mononuclear cells, SD standard deviation, US United States

Distribution Kinetics of ABP 938 and Aflibercept RP in Rabbit Eyes After IVT Injection

A study in rabbits was performed to compare the ocular distribution kinetics of AF488-labeled ABP 938 and AF488-labeled aflibercept EU administered after a single IVT injection. Both AF488-conjugated and unconjugated ABP 938 and aflibercept EU were comparable with respect to biological activity, higher-order structure, product impurities, and protein concentration, suggesting a low number of dye molecules were conjugated to ABP 938 and aflibercept EU (Supplementary Materials Table S1). To mitigate concerns regarding potential ocular inflammation during the distribution study, endotoxin levels were also monitored in the final labeled products (Supplementary Materials Table S1).

Following IVT injection, a non-compartmental PK analysis was performed on the corrected drug concentrations. ABP 938 and aflibercept EU reached mean (standard deviation [SD]) T_max values of 26.0 (21.1) and 34.0 (24.2) hours, respectively, in the vitreous fluid, and 42.0 (47.6) and 72.0 (41.6) hours, respectively, in the aqueous fluid (Table 3). After reaching T_max, concentrations of ABP 938 and aflibercept declined in the vitreous and aqueous fluids, with levels measurable through 552 h in all animals (Fig. 4). Given the small number of animals and PK variability, the assessment criteria for similarity of PK parameters in the distribution study was within twofold. Exposure, as assessed by mean C_max and AUC from time zero to infinity (AUC_0-inf), was similar (within twofold) in the vitreous and aqueous fluid of ABP 938- and aflibercept EU-treated animals (Table 3). A single IVT administration of AF488-conjugated ABP 938 or aflibercept EU was well tolerated in rabbits, as determined by indirect and slit-lamp ophthalmic examinations conducted on multiple days post-IVT dosing. These results indicate that ABP 938 and aflibercept EU administered by IVT injection have similar exposures in the vitreous and aqueous fluid of the rabbit eye.

Table 3.

Mean (SD) pharmacokinetic parameters of ABP 938 and aflibercept EU following a single IVT dose in rabbit eyes (peak method)

Test article	Dose level (mg)a	Cmax (μg/ml)	Tmax (h)	AUC0-552 (h*μg/ml)	AUC0-inf (h*μg/ml)	t1/2 (h)	Vss (ml)	CL (ml/h)
Vitreous humor (peak method)
AF488-ABP 938	1.03	678 (31.7)	26.0 (21.1)	140 000 (7670)	146 000 (8070)	120 (6.26)	1.24 (0.0656)	0.00706 (0.000398)
AF488-Aflibercept EU	1.01	824 (255)	34.0 (24.2)	159 000 (3610)	169 000 (5590)	134 (16.3)	1.15 (0.137)	0.00597 (0.000201)

Test article	Dose level (mg)a	Cmax (μg/ml)	Tmax (h)	AUC0-552 (h*μg/ml)	AUC0-inf (h*μg/ml)	t1/2 (h)
Aqueous humor (peak method)
AF488-ABP 938	1.03	40.9 (6.49)	42.0 (47.6)	6960 (682)	7030 (680)	76.8 (1.09)
AF488-aflibercept EU	1.01	47.7 (4.57)	72.0 (41.6)	8480 (875)	8600 (874)	82.7 (1.45)

n = 3 rabbits, OU

Values are mean (± standard deviation)

The peak method of analysis used the C_max of the fluorescence scan signal

μg microgram, AF488 Alexa Fluor™ 488, AUC area under the curve, AUC_0-552 AUC from time zero to 552 h, AUC_0-inf AUC from time zero to infinity, CL clearance, C_max maximum observed concentration, EU European Union, h hours, IVT intravitreal, mg milligram, ml milliliter, OU oculus uterque (both eyes), SD standard deviation, t_1/2 elimination half-life, T_max time to reach C_max, V_ss volume of distribution at steady state

^aActual dose. Nominal dose was 1 mg

Fig. 4.

Mean OU (both eyes) concentration vs. time in a vitreous humor and b aqueous humor (peak method) in rabbits following IVT administration of ABP 938 and aflibercept RP. h hours, IVT intravitreal, OU oculus uterque (both eyes), RP reference product

Repeat-Dose IVT Toxicology Study of ABP 938 and Aflibercept RP in Cynomolgus Monkeys

To eliminate any residual uncertainty about the ocular safety of ABP 938 in comparison to aflibercept EU, female cynomolgus monkeys (n = 4/group) were administered saline control, 1 mg/eye ABP 938, or 1 mg/eye aflibercept EU on study days 1 and 29 (two total doses given 4 weeks apart) in both eyes via IVT injection in a 25-µl dose volume. The 1 mg/ml/eye dose was selected to provide a human equivalent dose (HED) of 2 mg/dose/eye adjusting for vitreous volume differences between monkeys (2.0 ml) and humans (4.0 ml) [25]. Necropsy occurred 1 week after the second dose (Day 36).

The systemic exposure of ABP 938 and aflibercept EU was comparable (within a twofold range) as measured by AUC and C_max after the first and second doses of IVT injections (Table 4). The median T_max across the ABP 938 and aflibercept EU groups occurred at 3 days following dosing on days 1 and 29. The mean vitreous fluid concentration of ABP 938 or aflibercept EU obtained at necropsy were 135 µg/ml and 107 µg/ml, respectively, in the left eye samples and 146 µg/ml and 70 µg/ml, respectively, in the right eye samples. ABP 938 and aflibercept EU concentrations in samples from animals receiving saline control were below quantifiable limits.

Table 4.

A summary of results from repeat-dose toxicity studies (cynomolgus monkeys)

Dose (1 mg/eye intravitreal injection)	Control, no intravitreal injection	ABP 938	Aflibercept EU
Number of animals (female)	4	4	4
Toxicokinetics (serum)
AUC0–168h (day*μg/ml) mean
Day 1	BQL	16.5	23.1
Day 29	BQL	19.7	20.6
AUC0–672h (day*μg/ml) mean
Day 1	BQL	27.1	39.1
Cmax (μg/ml) mean
Day 1	BQL	3.09	4.22
Day 29	BQL	3.70	3.59
Tmax (day) median
Day 1	NA	3.0	3.0
Day 29	NA	3.0	3.0
Drug concentration in vitreous fluid prior to necropsy (μg/ml) mean
Left eye	BQL	135	107
Right eye	BQL	146	70
Died or sacrificed moribund (number of animals)	–	–	–
Clinical observations	–	–	–
Body weight (kg)	–	–	–
Ophthalmic observations-slit-lamp biomicroscope	–	–	–
Ophthalmic observations—indirect ophthalmoscope	–	–	–
Intraocular pressure measurements	–	–	–
Full-field electroretinography	–	–	–
Hematology	–	–	–
Clinical chemistry	–	–	–
Organ weights and organ weight ratios (g)	–	–	–
Histopathology	–	–	–

μg microgram, AUC_0–168h area under the serum concentration–time curve from time 0 to 168 h, AUC_0–672h area under the serum concentration–time curve from time 0 to 672 h, BQL below quantification limit (< 3.00 ng/ml), C_max maximum observed drug concentration during a dosing interval, EU European Union, F female, g grams, IVT intravitreal, kg kilogram, mg milligram, ml milliliter, NA not applicable, T_max time to reach C_max

–  no noteworthy findings

Intravitreal injections of ABP 938 and aflibercept EU were well tolerated in monkeys. There were no ABP 938-related effects on ophthalmic endpoints (i.e., ophthalmic examinations [indirect ophthalmoscopy and slit-lamp biomicroscopy], IOP measurements, full-field ERG) and no histopathologic observations of ocular toxicity in any treated animal. Transient, sporadic, and minor ocular findings/observations in saline, ABP 938, or aflibercept EU-dosed eyes were considered procedural-related and/or consistent with silicone droplets that originated from the syringe. There was no evidence of systemic toxicity associated with IVT administration of ABP 938; no treatment-related clinical observations or changes in body weight, qualitative food consumption, or clinical and anatomic pathology endpoints were reported in any animals. The no-adverse-effect-level for ABP 938 was considered to be greater than 1 mg/eye, the highest dose tested, based on the absences of ocular or systemic toxicity in female cynomolgus monkeys. The TK of ABP 938 were similar to those of aflibercept EU when administered by IVT injection.

Discussion

We present here nonclinical data demonstrating that ABP 938 is similar to aflibercept RP in binding kinetics and affinity for VEGF-A and PlGF isoforms, biodistribution kinetics in rabbit eyes, and in TK and toxicology profiles in non-human primates.

Among VEGF family members, VEGF-A is the most potent mitogen of angiogenesis [26, 27]. Aflibercept exhibits similar high-affinity binding to VEGF-A in humans, mice, rats, and rabbits [8]. Cynomolgus monkey and human VEGF-A proteins are 99% identical in amino acid sequences [28]. All VEGF-A isoforms bind both VEGFR-1 and VEGFR-2 through conserved VEGFR-binding domains [29–31]. VEGF-A expression analysis in primary retinal pigment epithelium cells in mice indicated that VEGF-A₁₆₅ is the most abundant isoform (75%). In contrast, the VEGF-A₁₂₁ isoform represents 24% of all transcripts, while the VEGF-A₁₈₉ isoform was almost undetectable [32]. Thus, VEGF-A₁₆₅ and VEGF-A₁₈₉ as the long heparin-binding isoforms and VEGF-A₁₁₁ and VEGF-A₁₂₁ as the short non-heparin-binding isoforms, were used to evaluate binding kinetics and affinity for ABP 938, aflibercept US, and aflibercept EU by SPR. Kinetic rate and equilibrium binding constants within two-fold are considered similar within the context of SPR [21–23]. ABP 938, aflibercept US, and aflibercept EU showed comparable binding kinetics and affinity to all VEGF-A isoforms (Tables 1 and Fig. 1).

Although the role of PlGF in the pathogenesis of neovascular (wet) age-related macular degeneration (AMD) is not fully understood, it is well established that PlGF acts as a paracrine factor upon activation of VEGFR-1, leading to proliferation, migration, and survival responses in endothelial cells [33, 34]. Additionally, high levels of PlGF have been found in the aqueous and vitreous humors and/or retina of patients with wet AMD [35], branch retinal vein occlusion and central retinal vein occlusion [36, 37], and proliferative diabetic retinopathy [38], suggesting that PlGF may have a pathogenic role in ocular diseases. All PlGF isoforms bind VEGFR-1 through conserved receptor-binding domains [39]. Therefore, PlGF-1 and/or PlGF-2 were used to assess binding kinetics and affinity by SPR. The results showed that ABP 938, aflibercept US, and aflibercept EU have similar PlGF-1 and PlGF-2 binding kinetics and affinity (Table 2 and Fig. 2).

Immune effector functions that involve the Fc portion of the antibody and C1q may result in CDC or the recruitment of Fc receptor-expressing cytotoxic effector cells that mediate target cell lysis via ADCC or phagocytosis [40, 41]. However, the biological functions of aflibercept RP are primarily mediated through its VEGFR-binding domains; the Fc domain does not induce effector functions as VEGF family members exist as predominantly soluble proteins [8]. The lack of effector functions for ABP 938 and aflibercept RP were confirmed using ADCC, ADCP, and CDC cell-based assays (Fig. 3) [24].

Previously, we performed a comprehensive analytical similarity assessment comparing ABP 938, aflibercept US, and aflibercept EU with respect to VEGFR domain- and Fc domain-mediated binding and activity [24]. The results across multiple orthogonal assays demonstrated that ABP 938 is structurally and functionally similar to aflibercept US and aflibercept EU. These results support a scientific bridge between aflibercept US and aflibercept EU and may be used to justify the relevance of the data generated with aflibercept EU in the ocular distribution study in rabbits and the IVT toxicology study in monkeys across the 2 RPs.

Nonclinical in vivo studies were designed to evaluate the similarity of ABP 938 compared to aflibercept RP with respect to ocular biodistribution, serum TK, and toxicology. An ocular biodistribution study was conducted in rabbits to compare ABP 938 and aflibercept RP after a single dose by IVT injection. The rabbit has been shown to provide comparable single-dose eye PK translation to that of the human eye [42, 43]. Additionally, the rabbit eye is sufficiently large to allow for IVT drug delivery [25, 44]. Ocular distribution kinetics were monitored using fluorophore-conjugated ABP 938 and aflibercept RP (Table 3). PK results were similar for each eye; therefore, combined values from right and left eyes were used for interpretation. The PK results, using concentration data from the peak method, were consistent with those from the AUC method, hence the summary of PK analysis using data from the peak method are presented (Table 3). Mean concentration–time profiles showed that, after IVT injection, ABP 938 and aflibercept RP concentrations generally declined in the vitreous and aqueous fluid after reaching T_max, with concentrations measurable through 552 h in all animals (Fig. 4). Exposure, based on mean C_max and AUC_0-inf, were also similar in the vitreous and aqueous fluid for ABP 938 and aflibercept RP. The results from the IVT ocular distribution study demonstrated that ABP 938 and aflibercept RP exposures were similar in the vitreous and aqueous fluid of the rabbit eye.

Animal toxicology studies are useful in biosimilar development when uncertainties remain about the safety of the proposed product [9, 12]. A 1-month repeat-dose IVT toxicology study was conducted in female cynomolgus monkeys to evaluate the ocular safety of the ABP 938 with 0.9% sodium chloride used as the control; aflibercept EU was included as a basis of comparison for the qualitative toxicologic profiles. The study was conducted in a single sex to reduce animal use and justified by the similar ocular profiles in male and female monkeys administered aflibercept EU [8]. The 1 mg/dose/eye is a HED of 2 mg/dose/eye, adjusting for vitreous volume differences between non-human primates and humans, and was considered a toxicologically appropriate dose level to meet the study objectives as recommended by a health authority and in consultation with subject matter experts. Saline was used as a negative control to differentiate any subtle ABP 938-related ocular findings from procedure-related findings associated with IVT injections, background spontaneous findings observed in the control group of cynomolgus monkeys, or artifacts associated with processing ocular tissue [25, 45]. As anticipated, no evidence of ocular toxicity or functional ocular deficits were identified in monkeys administered two IVT doses of ABP 938 or aflibercept RP at 1 mg/dose/eye in a 1-month study (Table 4). Bioanalytical assays of vitreous fluid at necropsy from ABP 938- or aflibercept EU-dosed eyes were conducted to demonstrate the presence of drug in the intended location (i.e., IVT space) in the event that no ocular toxicity or unexpected ocular toxicity was identified. ABP 938 or aflibercept RP was detected in the vitreous fluid of all dosed animals. The approximately twofold difference in ABP 938 or aflibercept EU vitreous fluid concentrations was considered comparable; variability was expected given the high viscosity of the non-standard sample matrix (i.e., vitreous fluid), non-standard route of administration (IVT), and low numbers of animals/group (n = 4/group), which were tested at a single timepoint. Although systemic exposure was expected to be low following IVT administration [46], a full clinical pathology assessment and comprehensive histopathology evaluation of tissues was conducted to provide a comprehensive toxicology study. There were no differences in organ weights and clinical pathologic and histopathologic findings were unremarkable in ABP 938- and aflibercept-dosed animals. A similar serum exposure, as estimated by AUC, was reported in the ABP 938- and aflibercept EU-dosed animals. The results of this study concluded that no ocular or systemic toxicities were observed with ABP 938 and there were no apparent differences in these toxicities between saline-, ABP 938- and aflibercept EU-dosed animals.

The findings of this study should be interpreted in light of certain limitations. SPR may not accurately measure slow dissociation rates characteristic of high-affinity interactions, potentially leading to an underestimation of affinity. Future studies could address this limitation by employing alternative techniques such as a solution-based kinetic exclusion assay. In addition, the clearance and distribution of biologics may differ between rabbit and human eyes due to differences in ocular blood flow and metabolism. Similarly, subtle anatomical differences in retinal vasculature of cynomolgus monkey eyes could affect drug distribution.

Conclusions

This integrated assessment of the in vitro pharmacology and in vivo PK, TK, and toxicology profiles demonstrated the safety of ABP 938 and contributes to the TOE indicating that ABP 938 is similar to aflibercept RP.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

Neungseon Seo, Scott Kuhns, Dina A. Andrews, Alexander Colbert, Vincent Chow, and Jennifer Liu were responsible for and/or involved in study conception and design. All authors were responsible for and/or involved in data collection and interpretation. Neungseon Seo was responsible for drafting the manuscript. All authors read and approved the final manuscript.

Funding

This study and the journal’s Rapid Service Fees were funded by Amgen Inc., Thousand Oaks, CA, USA.

Data Availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Conflict of Interest

Neungseon Seo, Scott Kuhns, Dina A. Andrews, Alexander Colbert, Vincent Chow, and Jennifer Liu are employees and stockholders of Amgen Inc.

Ethical Approval

All laboratory health and safety procedures were followed in the course of conducting this experimental work. All animals were housed in an Association for Assessment and Accreditation of Laboratory Animal Care, International-accredited facility and cared for in accordance with the Guide for the Care and Use of Laboratory Animals, 8th Edition. All research protocols (study numbers 8392437 and 8395840) were reviewed and approved by Labcorp (formerly Covance) and Amgen Institutional Animal Care and Use Committees. All rabbit procedures followed Animal Welfare Act Regulations (9 CFR 3). All primate procedures followed applicable animal welfare acts and were approved by the local Institutional Animal Care and Use Committee.