Pharmacokinetics and pharmacodynamics of the proposed biosimilar denosumab GP2411 and reference denosumab in healthy males

ABSTRACT

Background

This Phase I study compared the pharmacokinetic (PK) and pharmacodynamic (PD) similarity of GP2411 proposed denosumab biosimilar to reference denosumab (a monoclonal antibody for specific pro-resorptive conditions).

Research design and methods

Healthy males (28–65 years, 50–90 kg) were randomized to a single sub-therapeutic subcutaneous injection of 35 mg GP2411, EU-Xgeva® or US-Xgeva®, and followed for 39 weeks. The primary endpoints were AUC_inf, AUC_last, and C_max.

Results

Four hundred ninety-two participants completed treatment. The 90% confidence intervals (CIs) (AUC_inf, AUC_last, and C_max) and 95% CI of the geometric mean ratios of AUEC of % change from baseline in serum CTX were fully contained within the prespecified equivalence margins (0.80, 1.25), demonstrating similarity. The occurrence of treatment-emergent adverse events (TEAEs) with GP2411, EU-Xgeva® and US-Xgeva® was similar (72.9%, 76.0%, and 71.0% of participants, respectively). Most were Grade 1 or 2, <30% were treatment-related, and there was only one TEAE-related study discontinuation. Rates of positive anti-drug antibodies (ADAs) were similar (57.8%, 64.9%, and 69.1% of participants respectively), but immunogenicity was only borderline detectable and of very low magnitude. Ninety-nine percent of positive ADAs were transient.

Conclusion

GP2411 demonstrated similarity with EU-Xgeva® and US-Xgeva® in PK, PD, safety, and immunogenicity in this population.

Clinical trial registration

EudraCT 2019-001651-39

Plain Language Summary

Denosumab is a biological treatment that inhibits bone degradation. It is very effective in conditions characterized by elevated bone degradation, such as osteoporosis in women who have gone through the menopause, and in the treatment of specific bone cancers. However, the cost of the original patented denosumab (‘reference denosumab’) treatment may result in fewer eligible patients receiving denosumab treatment. A biosimilar is highly similar to the original treatment but at a lower price, enabling more patients to benefit.

GP2411 is being developed as a proposed biosimilar to denosumab. This Phase I clinical trial was the first clinical trial to compare GP2411 to the EU and US versions of the reference denosumab (EU-Xgeva® and US-Xgeva®). All three products were given at a dose of 35 mg to 502 healthy males. The dose was lower than the dose that would be used in clinical practice to provide a more sensitive evaluation of similarity. Healthy males were chosen because they have fewer hormonal fluctuations than females, and are considered the most appropriate population for detecting differences in pharmacological effects of denosumab.

The results demonstrate that GP2411 proposed denosumab biosimilar is highly similar to the reference products in absorption, distribution, and elimination, and other outcomes, including bone turnover. The incidence of adverse events was also comparable, most adverse events were very mild, and GP2411 was not associated with a higher rate of immune reactions.

These results support its continued development and GP2411 may, in future, enable more patients to benefit from denosumab treatment.

KEYWORDS:

• Biosimilar
• denosumab
• GP2411
• immunogenicity
• pharmacodynamics
• pharmacokinetics
• phase I
• safety

1. Introduction

Denosumab is a fully human monoclonal antibody (IgG2) that targets and binds to receptor activator of nuclear factor kappa-b (RANK) ligand (RANKL) [1,2]. RANKL is a protein that promotes osteoclast formation and inhibits osteoclast apoptosis (when transmembrane-bound on the surface of osteoblasts) and also plays a role in osteoclast maturation and functioning (via proteolytic cleavage when in soluble form) [1,2]. By binding to RANKL, denosumab inhibits osteoclast formation, activity, and survival [2] resulting in a rapid decrease in bone resorption and, subsequently, a decrease in levels of bone resorption markers such as collagen C-terminal telopeptide (CTX) [2,3]. In addition to being reflective of denosumab’s mechanism of action, measurement of CTX also provides an early indication of denosumab’s action on osteoclasts [3], making it a suitable biomarker for a similarity assessment.

Osteoporosis is caused by an imbalance favoring bone resorption over formation, that results from excessive osteoclastic activation by RANKL. This is also the cause of bone destruction in secondary bone malignancies [4,5]. Denosumab is approved by the U.S. Food and Drug Administration (FDA) and the European Medicines Agency (EMA) for treating osteoporosis in postmenopausal women, and men at increased risk of fractures (60 mg once every 6 months, Prolia®; Amgen), and for preventing skeletal-related events in patients with bone metastases from malignancies, and for the treatment of giant cell tumor of bone that is either unresectable or where resection is likely to result in severe morbidity (120 mg every 4 weeks, with an additional 120 mg doses on Days 8 and 15 of the first month of therapy, Xgeva®; Amgen) [6–9]. However, there is a need for a biosimilar to provide a more affordable option for patients with osteoporosis [10].

Biosimilars are biological medicinal products that contain a version of the active substance of an authorized original biologic medicinal product (reference medicine), with no clinically meaningful differences in purity, biologic activity, quality, efficacy, and safety [11,12]. Biosimilarity between a proposed biosimilar medicine and the reference medicine is determined with a stepwise approach that includes structural and functional characterization and clinical evaluation [12]. Regulatory authorities consider the overall ‘totality of evidence’ when evaluating the demonstration of biosimilarity [12]. Biosimilars have the potential to promote market competition and innovation, reduce healthcare costs, and enable more patients to access treatment [13].

GP2411 is a proposed biosimilar to Amgen’s denosumab that is currently in the final stages of clinical development. It has already demonstrated structural and analytical similarity to reference denosumab, including similarity in binding affinity to RANKL (data on file). A Phase III study conducted in parallel to this study has demonstrated similar efficacy, safety, and immunogenicity of GP2411 to reference denosumab (Prolia®) in women with postmenopausal osteoporosis [14,15]. Here, we present the results of a Phase I study that evaluated similarity in PK, PD, safety, and immunogenicity between GP2411, EU-authorized Xgeva® (EU-Xgeva®), and US-licensed Xgeva® (US-Xgeva®) in healthy male volunteers.

2. Patients and methods

2.1. Design

This Phase I, double-blind, randomized, three-arm parallel group study was conducted at two sites in Germany. Participants were randomized 1:1:1 to receive a single sub-therapeutic 35-mg subcutaneous injection of GP2411, EU-Xgeva®, or US-Xgeva®, and were followed-up for 39 weeks. This dose was selected based on the published denosumab population PK model in healthy subjects and postmenopausal women with osteoporosis [16] in combination with the semi-mechanistic PD model of bone cycling [17]. In house simulations using the published population PK and PD model showed that this dose covers both elimination pathways (the linear saturable targeted mediated drug disposition pathway and the linear IgG clearance pathway via FcRn binding), and is positioned in the middle of the steep part of the dose–response curve for AUEC of %CfB in serum CTX, thus providing an optimal dose for measuring PD similarity [16,17]. In addition, the 35 mg dose captures the entire PK profile over 39 weeks, and allows the return of CTX to baseline in most subjects [16,17]. Randomization was stratified by weight category (<70 kg, ≥70 kg and <80 kg, and ≥80 kg) and vitamin D replenishment status (yes/no). A single vitamin D replenishment was attempted in participants with a serum level of 25-hydroxy vitamin D that was <20 ng/mL during screening, and vitamin D level was reevaluated before randomization. Participants with a continued level of <20 ng/mL were excluded. Participants also received supplementary calcium (1,000 mg/day) and vitamin D (400 IU/day) from Day −7 (prior to randomization) to Day 28 to reduce the risk of hypocalcemia, which is highest in the weeks immediately after the denosumab dose. More frequent monitoring was performed during this time. All participants provided written informed consent before entering the study. The study was conducted in accordance with the ethical principles of the Declaration of Helsinki and with the ICH E6 Guideline for Good Clinical Practice. It was also reviewed and approved by the Ethics Committee of the Land Berlin, Berlin, Germany. The Phase I clinical trial was registered at EudraCT (Nr: 2019–001651–39).

2.2. Participants

Eligible participants were healthy males aged 28–65 years, with body weight 50–90 kg, and body mass index (BMI) 18.5–30.0 kg/m². Key exclusion criteria included previous exposure to denosumab, or any medication for treating osteoporosis (at any time point) or modifying bone metabolism (within 1 year prior to the first dose of study drug); current hypocalcemia or hypercalcemia; oral or dental conditions such as osteonecrosis of the jaw (ONJ) or risk factors for ONJ (e.g. periodontal disease, poorly fitting dentures, or invasive dental procedures such as tooth extractions in the 6 months before screening); vitamin D deficiency (see Section 2.1); and current or previous hypo-/hyper-thyroidism or parathyroidism.

2.3. Endpoints

The primary endpoints were the three following serum PK endpoints: area under the serum concentration–time curve measured from the time of dosing and extrapolated to infinity (AUC_inf), area under the serum concentration–time curve measured from the time of dosing to the last measurable concentration (AUC_last), and maximum observed serum concentration of study drug (C_max). Secondary PD endpoints included the area under the effect–time curve (AUEC) of the percentage change from baseline in serum CTX, and the serum concentrations of CTX and PINP (procollagen type I N-propeptide per visit. AUEC was only calculated for CTX because its large dynamic range (>90% maximal reduction), rapid onset, and relatively low inter-subject variability provide a high level of sensitivity that exceeds that of PINP, and other bone turnover markers like urinary cross-linked N-telopeptides of type 1 collagen and bone alkaline phosphatase [3,18]. Secondary safety endpoints included the incidence of adverse events, serious adverse events, and treatment-related adverse events (including injection site reactions, vital signs, electrocardiogram abnormalities, and laboratory safety endpoints). Immunogenicity was evaluated based on the proportion of participants testing positive for anti-drug antibodies (ADAs), assessed with titers and neutralizing capacity.

2.4. Sampling and analysis

For PK and PD assessments, blood samples were collected 16 times over 39 weeks (pre-dose and 6 hours post-dose on Day 1, then on Days 4, 8, 11, 15, 29, 57, 85, 113, 141, 169, 197, 225, 254, and 275). Samples for PD assessments were collected in the morning (7:30–10:00 AM) in a fasting state. Blood was collected, clotted, and centrifuged to separate serum from the remaining components. The serum samples obtained were stored frozen (at −70°C) until analysis. Serum drug concentrations were quantified using a validated enzyme-linked immunosorbent assay with a lower limit of quantification of 5 ng/mL (the suitability of the assay to quantify GP2411, EU-Xgeva®, and US-Xgeva® was demonstrated). This assay comprises a bridging format of immunocomplexes by bivalent binding to two molecules of RANKL. The drug was captured by RANKL and detected by biotinylated RANKL which bound streptavidin-HRP (horseradish peroxidase). The HRP-catalyzed oxidation of corresponding substrate resulted in a color change, the intensity of which correlated with the serum denosumab concentration.

CTX and PINP concentrations were quantified using chemiluminescence-based assays (IDS-iSYS semi-automated immunoassay platform), with lower limits of quantification of 0.33 ng/mL and 2.0 ng/mL, respectively. The assays comprised a sandwich format using biotinylated monoclonal antibodies (as capture) and acridinium labeled monoclonal antibodies (as detection agents). Serum drug, CTX and PINP method development and validation was performed by the Sponsor’s own bioanalytical laboratory in compliance with current international guidelines on bioanalytical method validation [19–21].

ADAs were detected using a drug-tolerant electrochemiluminescence assay with a high sensitivity (in 100% serum) of 6 ng/mL. Samples were initially evaluated in a screening assay and, in case of a positive result (a value above the screening cut point of 6 ng/mL), a confirmatory (specificity) assay was performed. Confirmed positive ADAs were further evaluated in a titer assay to measure their magnitude, and their neutralizing capacity was assessed in a dedicated qualitative neutralizing antibody (NAb) assay. ADAs with a concentration below the titer cut point (20 ng/mL) were reported as negative and no titer value could be reported; ADAs with a concentration above the titer cut point (≥20 ng/mL) were positive, and a corresponding titer value was reported. ADAs were classified as persistent if a participant had at least two consecutive positive ADA results and a positive ADA result at the final visit, regardless of missing assessments between them. Positive ADAs not qualifying as persistent were considered transient.

Adverse events (AEs), including serious adverse events (SAEs), were recorded throughout the study. They were coded with Medical Dictionary for Regulatory Activities and their severity was evaluated with Common Terminology Criteria for Adverse Events (CTCAE) v5.0.

2.5. Statistical analysis

A sample size calculation was performed for PK and PD endpoints. The coefficients of variation of the AUC_inf, C_max and AUEC of baseline corrected serum CTX (% change from baseline), and correlation between AUC_inf and C_max, were derived from 10,000 simulations using a published denosumab population PK/PD model [16,17]. Assuming a 15% dropout rate, a sample size of 501 participants was required to achieve a statistical power of 90%. Similarity in PK primary endpoints and the PD endpoint was assessed using an analysis of covariance model of the log-transformed endpoints. For PK endpoints (AUC_inf, AUC_last, and C_max), the model included treatment as a fixed factor and weight as a continuous covariate and, for PD endpoints (AUEC of percentage change from baseline in CTX), treatment and vitamin D replenishment were included as fixed factors, and the log of baseline CTX was included as a covariate. Statistical analyses were conducted using SAS version 9.4.

PK and PD outcomes were derived by non-compartmental analysis using Phoenix WinNonlin (Certara, Version 8.3). Similarity in PK and PD was concluded if the 90% CIs and 95% CIs of the geometric mean ratios respectively, for all three pairwise comparisons were contained entirely within the prespecified acceptance margin of 0.8–1.25. The similarity comparison is consistent with the two one-sided tests procedure for bioequivalence at the 5% (2.5%) significance level [22]. The PK and PD analysis sets (PKS and PDS) included all participants who received study treatment, had at least one evaluable primary PK or PD value respectively, and completed the study without a protocol deviation that had a relevant impact on respective endpoints. Requirement for any other medication which may have impacted bone resorption (excluding vitamin D and calcium intake) was classified as a relevant protocol deviation. In PK analyses, values below the lower limit of quantification (LLOQ) were treated as missing values (except for the pre-dose sample, which was treated as zero) to enable arithmetic means to be calculated. In PD analyses, in the calculation of AUEC of percentage change from baseline in CTX, values below the LLOQ were imputed with the actual value for the LLOQ unless they were missing in which case they were handled as such. Missing serum concentrations of CTX were not imputed. Descriptive analyses were performed for the serum concentrations of CTX, safety, and for the following additional PK endpoints: percentage of AUC_inf due to extrapolation from the time of the last observed concentration to infinity (AUC_%extrap), time to reach maximum serum concentration (T_max), terminal elimination rate constant [day⁻¹] (Lambda_z), and apparent terminal half-life [day] (T1/2). The safety analysis set (SAF) included all participants who received at least one dose of study treatment.

3. Results

3.1. Participant characteristics and disposition

A total of 502 participants were randomized to GP2411 (n = 166), EU-Xgeva® (n = 171), or US-Xgeva® (n = 165). Participant baseline characteristics were balanced between treatment groups; median age ranged from 44.5 to 47.0 years and 98.4% of participants were White (Table 1). Three participants discontinued the study before receiving treatment (US-Xgeva®, [1.8%]), and seven participants discontinued the study during the 39-week post-treatment observation period (EU-Xgeva®: 6 participants [3.6%]; US-Xgeva®: 1 participant [0.6%]). Overall, 492 participants completed treatment (Figure 1). Nine participants were excluded from the PKS and 14 from the PDS; reasons for exclusion were protocol deviations that may impact results (3 and 4 participants, respectively), not receiving treatment on Day 1 (3 participants in each analysis set), or missing PK or PD values (6 and 10 participants, respectively).

Figure 1. Participant disposition.

PDS, pharmacodynamics analysis set; PKS, pharmacokinetics analysis set; SAS, safety analysis set.

*Three participants from the US-denosumab group discontinued study participation before receiving treatment due to randomization by mistake, participant decision, and physician decision.

Table 1. Participant demographics and baseline characteristics.

	GP2411 n = 166	EU-Xgeva® n = 171	US-Xgeva® n = 162
Age, years, mean (SD)	47.0 (9.13)	44.5 (9.52)	46.0 (9.44)
Ethnicity, n (%)
Hispanic or Latino	1 (0.6)	1 (0.6)	0 (0.0)
Not Hispanic or Latino	162 (97.6)	170 (99.4)	162 (100)
Not reported	3 (1.8)	0 (0.0)	0 (0.0)
Race, n (%)
Asian	1 (0.6)	2 (1.2)	0 (0.0)
Black or African American	1 (0.6)	1 (0.6)	1 (0.6)
Multiple	1 (0.6)	1 (0.6)	0 (0.0)
White	163 (98.2)	167 (97.7)	161 (99.4)
Weight, kg, mean (SD)	79.9 (6.82)	80.1 (7.24)	79.3 (7.29)
Height, cm, mean (SD)	179.5 (5.99)	179.2 (6.12)	178.5 (6.13)
Vitamin D replenishment, n (%)
No	106 (63.9)	108 (63.2)	102 (63.0)
Yes	60 (36.1)	63 (36.8)	60 (37.0)

Weight category and vitamin D replenishment were stratification factors.SD, standard deviation.

3.2. PK and PD endpoints

The 90% CIs of the geometric mean ratios for the primary PK endpoints were all fully contained within the prespecified equivalence margins (0.80, 1.25), demonstrating PK similarity (Figure 2). The serum concentration–time profile of GP2411 was also similar to that of EU-Xgeva® and US-Xgeva® over the full 39 weeks (Figure 3). Additional PK endpoints (AUC%_extrap, T_max, Lambda_z, and T_1/2) measured in the GP2411 arm were also similar to values obtained with EU-Xgeva®, and US-Xgeva® (Table 2).

Figure 2. Geometric mean ratios and 90% CI of primary PK parameters.

AUC_inf, area under the serum concentration–time curve measured from the time of dosing and extrapolated to infinity; AUC_last, area under the serum concentration–time curve measured from the time of dosing and extrapolated to the last measurable concentration; CI, confidence interval; C_max, maximum observed serum concentration.

Figure 3. Mean drug serum concentration–time profiles. (a) Mean (SD) linear scale. (b) Mean, semi-logarithmic scale.

SD, standard deviation.

Linear scale: no lower SD whisker displayed if mean SD < 0, given it is biologically implausible.

Semi-logarithmic scale: values of 0 for which no log transformation is defined are presented as 1.

Table 2. Pharmacokinetic endpoints.

Mean (SD)	GP2411 n = 165	EU-Xgeva® n = 167	US-Xgeva® n = 161
AUCinf (day*ng/mL)	135,000 (396,00)	124,000 (40,000)	130,000 (448,00)
AUClast (day*ng/mL)	132,000 (39,900)	123,000 (39,900)	128,000 (44,900)
Cmax (ng/mL)	3,050 (892)	2,960 (817)	3,070 (966)
AUC%extrap (%)	1.60 (4.24)	1.23 (1.30)	1.36 (2.26)
Lambda_z (1/day)	0.05 (0.02)	0.05 (0.02)	0.05 (0.01)
T1/2 (day)	16.3 (5.54)	15.5 (4.65)	15.8 (4.46)
Tmax (day); median (range)	9.99 (2.96–32.0)	9.96 (2.96–31.1)	8.04 (2.96–31.1)

AUC_%extrap, percentage of AUC_inf due to extrapolation from the time of the last observed concentration to infinity; AUC_inf, area under the serum concentration–time curve measured from the time of dosing and extrapolated to infinity; AUC_last, area under the serum concentration–time curve measured from the time of dosing to the last measurable concentration; Lambda_z, terminal elimination rate constant; SD, standard deviation; T_1/2, apparent terminal half-life; T_max, time to reach maximum serum concentration.

Regarding the PD outcome, the 95% CIs of the geometric mean ratios of AUEC of percentage change from baseline in serum CTX were also fully contained within the prespecified equivalence margins (Figure 4). In addition, the mean serum concentrations of CTX and PINP for GP2411, and the associated percentage changes from baseline, was similar to that of EU-Xgeva®, and US-Xgeva® throughout the study (Figure 5). Missing data for CTX and PINP was minimal. The proportion of participants with missing values for CTX was 1.9% (n = 3/161) for GP2411, 1.8% (n = 3/166) for EU-Xgeva®, and 1.2% (n = 2/161) for US-Xgeva®. For PINP, it was 1.2% (n = 2/161) for GP2411, 1.2% (n = 2/166) for EU-Xgeva® and 1.2% (n = 2/161) for US-Xgeva®.

Figure 4. Geometric mean ratio and 95% CIs of AUEC of percentage change from baseline in serum CTX.

AUEC, area under the effect–time curve; CI, confidence interval; CTX, carboxy-terminal crosslinked telopeptide of type I collagen; ref, reference product.

Figure 5. Change from baseline in PD endpoints. (a) serum CTX levels. (b) serum PINP levels.

CTX, carboxy-terminal crosslinked telopeptide of type I collagen; PD, pharmacodynamics; PINP, procollagen type I N-propeptide; SD, standard deviation.

A lower SD whisker is not displayed if mean SD <-100%, given it is biologically implausible.

3.3. Safety endpoints

Safety profiles were similar across treatment groups. For GP2411, EU-Xgeva® and US-Xgeva®, TEAEs affected 121 (72.9%), 130 (76.0%), and 115 (71.0%) participants respectively. Treatment-related AEs affected 49 (29.5%), 47 (27.5%) and 40 (24.7%) participants respectively. Most TEAEs were of Grade 1 or 2 severity, and the most common any-grade TEAEs across treatment groups were headache, nasopharyngitis, COVID-19, and hypocalcemia (Table 3). Seven participants experienced SAEs, which included 3 participants in the GP2411 arm (1.8%), 3 in the EU-Xgeva® arm (1.8%), and 1 in the US-Xgeva® arm (0.6%). The SAEs associated with GP2411 were cerebral hemorrhage, inguinal hernia and pulmonary contusion; for US-Xgeva®, an intervertebral disc protrusion and; for EU-Xgeva®, they were alcohol abuse, appendicitis, and mantle cell lymphoma.

Table 3. Treatment-emergent adverse events reported in ≥2% of participants in any treatment group.

	GP2411 n = 166		EU-Xgeva® n = 171		US-Xgeva® n = 162
	All grades n (%)	Grades 3–4 n (%)	All grades n (%)	Grades 3–4 n (%)	All grades n (%)	Grades 3–4 n (%)
Participants with at least one treatment-emergent adverse event	121 (72.9)	2 (1.2)	130 (76.0)	3 (1.8)	115 (71.0)	2 (1.2)
Infections and infestations	57 (34.3)	0 (0.0)	59 (34.5)	1 (0.6)	40 (24.7)	0 (0.0)
COVID-19	18 (10.8)	0 (0.0)	17 (9.9)	0 (0.0)	20 (12.3)	0 (0.0)
Nasopharyngitis	17 (10.2)	0 (0.0)	30 (17.5)	0 (0.0)	12 (7.4)	0 (0.0)
Rhinitis	7 (4.2)	0 (0.0)	6 (3.5)	0 (0.0)	5 (3.1)	0 (0.0)
Suspected COVID-19	6 (3.6)	0 (0.0)	8 (4.7)	0 (0.0)	2 (1.2)	0 (0.0)
Nervous system disorders	44 (26.5)	1 (0.6)	39 (22.8)	1 (0.6)	26 (16.0)	0 (0.0)
Headache	38 (22.9)	0 (0.0)	38 (22.2)	1 (0.6)	19 (11.7)	0 (0.0)
Musculoskeletal and connective tissue disorders	33 (19.9)	0 (0.0)	36 (21.1)	0 (0.0)	39 (24.1)	1 (0.6)
Back pain	13 (7.8)	0 (0.0)	14 (8.2)	0 (0.0)	13 (8.0)	0 (0.0)
Arthralgia	11 (6.6)	0 (0.0)	10 (5.8)	0 (0.0)	6 (3.7)	0 (0.0)
Pain in extremity	5 (3.0)	0 (0.0)	4 (2.3)	0 (0.0)	3 (1.9)	0 (0.0)
Gastrointestinal disorders	28 (16.9)	1 (0.6)	28 (16.4)	0 (0.0)	28 (17.3)	0 (0.0)
Diarrhea	10 (6.0)	0 (0.0)	6 (3.5)	0 (0.0)	8 (4.9)	0 (0.0)
Toothache	5 (3.0)	0 (0.0)	3 (1.8)	0 (0.0)	4 (2.5)	0 (0.0)
Abdominal pain	2 (1.2)	0 (0.0)	4 (2.3)	0 (0.0)	2 (1.2)	0 (0.0)
Abdominal discomfort	1 (0.6)	0 (0.0)	2 (1.2)	0 (0.0)	4 (2.5)	0 (0.0)
General disorders and administration site conditions	19 (11.4)	0 (0.0)	17 (9.9)	0 (0.0)	10 (6.2)	0 (0.0)
Fatigue	6 (3.6)	0 (0.0)	6 (3.5)	0 (0.0)	2 (1.2)	0 (0.0)
Chills	4 (2.4)	0 (0.0)	1 (0.6)	0 (0.0)	1 (0.6)	0 (0.0)
Respiratory, thoracic and mediastinal disorders	15 (9.0)	0 (0.0)	15 (8.8)	0 (0.0)	6 (3.7)	0 (0.0)
Cough	7 (4.2)	0 (0.0)	5 (2.9)	0 (0.0)	1 (0.6)	0 (0.0)
Oropharyngeal pain	5 (3.0)	0 (0.0)	6 (3.5)	0 (0.0)	4 (2.5)	0 (0.0)
Injury, poisoning and procedural complications	14 (8.4)	0 (0.0)	17 (9.9)	0 (0.0)	21 (13.0)	0 (0.0)
Ligament sprain	4 (2.4)	0 (0.0)	1 (0.6)	0 (0.0)	4 (2.5)	0 (0.0)
Metabolism and nutrition disorders	14 (8.4)	0 (0.0)	13 (7.6)	0 (0.0)	18 (11.1)	0 (0.0)
Hypocalcemia	14 (8.4)	0 (0.0)	11 (6.4)	0 (0.0)	17 (10.5)	0 (0.0)
Skin and subcutaneous tissue disorders	11 (6.6)	0 (0.0)	11 (6.4)	0 (0.0)	5 (3.1)	0 (0.0)
Erythema	0	0 (0.0)	4 (2.3)	0 (0.0)	1 (0.6)	0 (0.0)
Immune system disorders	8 (4.8)	0 (0.0)	13 (7.6)	0 (0.0)	15 (9.3)	0 (0.0)
Immunization reaction	8 (4.8)	0 (0.0)	11 (6.4)	0 (0.0)	15 (9.3)	0 (0.0)
Psychiatric disorders	3 (1.8)	0 (0.0)	4 (2.3)	1 (0.6)	1 (0.6)	0 (0.0)

n represents number of participants. A participant with multiple severity grades for an adverse event was only counted under the maximum grade.

Overall, seven participants experienced TEAEs of Grade 3 severity, which included 2 in the GP2411 arm and the US-Xgeva® arm (1.2% in both) and 3 in the EU-Xgeva® arm (1.8%). These included a cerebral hemorrhage and an inguinal hernia in the GP2411 arm; alcohol abuse, appendicitis, and headache in the EU-Xgeva® arm; and aspartate aminotransferase increased and an intervertebral disc protrusion in the US-Xgeva arm. None of the serious or Grade 3 TEAEs were considered treatment-related. Only one participant discontinued the study because of a TEAE, which was a cerebral hemorrhage of Grade 3 severity in the GP2411 arm that was considered treatment-unrelated. There were no deaths in this study.

3.4. Immunogenicity endpoints

Similarity in the incidence of ADAs over 39 weeks was also observed. Positive ADAs were observed in 96 participants in the GP2411 arm (57.8%), 111 participants in the EU-Xgeva arm (64.9%) and 112 participants in the US-Xgeva arm (69.1%) (Table 4). All confirmed positive ADAs (including NAbs) in this study were on the borderline of the detection limit of the highly sensitive assay (irrespective of treatment group), and between the screening cut-point (6 ng/mL) and titer cut-point (20 ng/mL). Hence, none of the participants had a measurable (positive) ADA titer and the vast majority of ADAs (99%) were transient, indicating that GP2411, EU-Xgeva® and US-Xgeva® have low immunogenic capacity. The presence of NAbs was rare, only identified in 2 participants in the GP2411 arm (1.2%) and 1 in the EU-Xgeva® arm (0.6%), and cases were transient.

Table 4. Incidence of ADAs and immunogenicity outcomes.

	GP2411 n = 166	EU-Xgeva® n = 171	US-Xgeva® n = 162
Baseline, n (%)
ADA-positive	1 (0.6)	0 (0.0)	0 (0.0)
ADA titer-positive	0 (0.0)	0 (0.0)	0 (0.0)
ADA titer-negative	1 (0.6)	0 (0.0)	0 (0.0)
NAb-positive	0 (0.0)	0 (0.0)	0 (0.0)
NAb-negative	1 (0.6)	0 (0.0)	0 (0.0)
ADA-negative	165 (99.4)	171 (100)	162 (100)
Overall, n (%)
ADA-positive	96 (57.8)	111 (64.9)	112 (69.1)
ADA titer-positive	0 (0.0)	0 (0.0)	0 (0.0)
ADA titer-negative	96 (57.8)	111 (64.9)	112 (69.1)
Transient	95 (57.2)	110 (64.3)	111 (68.5)
Persistent	1 (0.6)	1 (0.6)	1 (0.6)
NAb-positive	2 (1.2)	1 (0.6)	0 (0.0)
NAb-negative	94 (56.6)	110 (64.3)	112 (69.1)
ADA negative	70 (42.2)	60 (35.1)	50 (30.9)

ADA, antibody-drug antibody; NAb, neutralizing antibodies.

‘Transient’ indicates a participant with a positive ADA result but not qualifying as ‘Persistent.’ ‘Persistent’ indicates a participant with a positive ADA result at the final visit and with at least two consecutive positive ADA results, irrespective of missing results. ‘ADA titer-positive’ indicates an ADA level ≥20 ng/mL; ‘ADA titer-negative’ indicates an ADA level ranging 6–20 ng/mL.

4. Discussion

This study demonstrated similarity in PK and PD between the proposed denosumab biosimilar GP2411, EU-Xgeva®, and US-Xgeva® in a large cohort of healthy male volunteers who received a single dose of 35 mg instead of the therapeutic dose of 60 mg (Prolia®) or 120 mg (Xgeva®) [6–9], and were followed-up for 39 weeks after treatment. This is based on the geometric mean ratios of all assessed PK and PD endpoints being contained within the prespecified equivalence margins. No two biologics are absolutely identical, but the differences in AUC were not clinically meaningful based on the prespecified equivalence margins. A high level of PK similarity is further supported by the consistent outcomes observed with additional PK endpoints (AUC%_extrap, T_max, Lambda_z, and T_1/2).

In addition, GP2411 was well tolerated, no new or unexpected safety issues were reported, and the number of subjects with at least one TEAE was similar to EU-Xgeva® and US-Xgeva® arms. Most TEAEs were of Grade 1 or 2 severity, and the minority of participants who experienced serious TEAEs or TEAEs of Grade 3 severity was also similar to EU-Xgeva®, and US-Xgeva® arms. Adverse events (e.g. hypocalcemia) were also in line with the known safety profile of denosumab, as reported in the Package Inserts and Summaries of Product Characteristics [6–9].

GP2411 was not associated with clinically relevant immunogenic responses, and immunogenicity profiles were similar between GP2411 and the reference medicines. However, the incidence of ADAs in all groups was higher than has been reported historically (<1% of patients with osseous metastases who were treated with 30–180 mg of Xgeva® every 4 or 12 weeks for up to 3 years and evaluated with an electrochemiluminescent bridging immunoassay) [23]. Indeed, the observed incidence of ADAs is highly dependent on the sensitivity and specificity of the assay [24]. The likely explanation for the difference is the high sensitivity (6 ng/mL) of the ADA assay used in this study, combined with high testing frequency, which enabled detection of transient ADA responses.

Notably, serum samples pertaining to all positive ADAs reported in this study had very low and non-quantifiable titers (<20 ng/mL), and immune responses were generally transient. This demonstrates the low immunogenic capacity of GP2411, EU-Xgeva® and US-Xgeva®, and the similarity between GP2411 and the two reference medicines. Guidance released by the FDA in 2019 for testing the immunogenicity of therapeutic protein products states that ADAs with concentrations of 100 ng/mL or more may be associated with clinical events [25]. Therefore, the applied method to detect ADAs with a sensitivity of 6 ng/mL is ~17 times more sensitive than that required to detect clinically meaningful immunogenicity.

This study used an appropriate design to demonstrate similarity while minimizing confounding and other potential biases. The randomized design ensured that factors that could affect PK (e.g. weight) or PD (e.g. vitamin D replenishment) were balanced between treatment groups. The use of stratified randomization also ensured treatment arms were comparable. Males were chosen because they have fewer hormonal fluctuations than females [26,27], and therefore represent a more homogenous study group for evaluating similarity in the PD of denosumab. Restricting eligibility for weight to 50–90 kg, and age to 28–65 years, in this study limited PK variability, and ensured all participants were skeletally mature, respectively. Replenishment of vitamin D further ensured participants had healthy bone metabolism prior to administration of the study drug.

The key strengths of this study were the use of a subtherapeutic dose of Xgeva®, i.e. 35 mg rather than the therapeutic dose of 120 mg every 4 weeks [8,23], and the evaluation of CTX concentration–time profiles over 39 weeks. The use of a subtherapeutic dose allowed for a more sensitive detection of any PK differences as it utilizes both elimination pathways of denosumab. These are the non-linear elimination pathway – i.e. target mediated drug disposition (TMDD) via non-linear and saturable target binding to RANKL, which is saturated at therapeutic doses, and linear IgG clearance (via neonatal Fc receptor binding) [16]. The 39-week timeframe aimed to ensure that CTX levels returned to baseline in a significant portion of subjects, which optimized the sensitivity of the PD assessment for detecting any potential differences.

Serum CTX levels after administration of denosumab 35 mg were almost undetectable for ~6 months in this study of healthy males aged 28–65 years, whereas a higher dose of denosumab 60 mg is required to inhibit CTX over 6 months in postmenopausal women with osteoporosis [14,15]. The most likely explanation for this difference is the effect of gender and age on serum CTX [28]. In men, serum CTX levels decrease with increasing age between 25 and 55 years then plateau, explaining why almost undetectable levels were sustained for ~6 months. This is different in premenopausal women, in which CTX concentrations are stable between 30 and 54 years of age, and then markedly increase post menopause between the ages of 50 and 55 years due to decreased estrogen levels and increased osteoclastic activity [29].

With biosimilars, similarity established at a subtherapeutic dose can be extrapolated to the therapeutic dose. Regulatory authorities allow scientifically justified extrapolation of evidence between doses and indications, without the need for additional clinical studies, as part of the totality of evidence approach to biosimilarity [30]. Extrapolation of PK outcomes with the subtherapeutic dose in this study to the therapeutic dose is particularly well justified based on the already-established constant absorption and distribution of reference denosumab, saturation of the non-linear elimination pathway at therapeutic doses, and linear elimination within a dose range of 6–210 mg [16], which includes the 35 mg dose evaluated in this study. PD similarity established with the subtherapeutic dose can also be extrapolated to the therapeutic dose, as the PD profile of serum CTX concentrations is mainly dose-independent, except for a later return to baseline at higher doses [31]. Evidence from this study on the similarity in PK, PD and safety, evaluated for a single subtherapeutic dose of GP2411, is complemented by established PK, PD and safety similarity in a separate Phase 3 study of GP2411 in postmenopausal women with osteoporosis that used a therapeutic dose of 60 mg [14,15].

While there is currently no approved denosumab biosimilar in the US or Europe, several are in clinical development and are also being evaluated in clinical trials that aim to demonstrate PK and PD similarity with Xgeva® at a therapeutic dose [32,33]. Use of a therapeutic dose (i.e. 120 mg versus the subtherapeutic 35 mg in this study) and the shorter follow-up durations (22–23 weeks versus 39 weeks in this study) do not allow serum CTX levels to return to baseline. Resultantly, the CTX assessments in those studies have less sensitivity for detecting potential differences between the biosimilar and the reference medicine (data on file). The sample size of this study was also much larger (n = 502 versus n = 120 and n = 154), improving the representativeness of the results [32,33]. Overall, this was a robust study that thoroughly characterized similarity between GP2411 and the reference denosumab medicines, and had a high level of sensitivity for detecting differences in PK, PD, and immunogenic responses. The outcomes are particularly promising because availability of an approved denosumab biosimilar is likely to enhance treatment access and potentially reduce treatment costs, as has been observed with other biosimilars [34].

5. Conclusions

In summary, this Phase I study demonstrated that the proposed biosimilar denosumab GP2411 has similarity to EU-Xgeva® and US-Xgeva® in healthy male participants across PK, PD, safety, and immunogenicity outcomes. The findings strengthen the totality of evidence for GP2411 as a promising denosumab biosimilar candidate.

Declaration of interest

B Vogg, J Poetzl, R El Galta, A Schwebig and S Sekhar are employees of Hexal AG (a Sandoz company). R Fuhr is employee of Parexel International GmbH. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Reviewer disclosures

Peer reviewers on this manuscript have received an honorarium from Expert Opinion on Biological Therapy for their review work. The reviewers have no other relevant financial relationships to disclose.

Ethics statement

All participants provided written informed consent before entering the study. The study was conducted in accordance with the ethical principles of the Declaration of Helsinki and with the ICH E6 guideline for good clinical practice. It was also reviewed and approved by the ethics committee of the Land Berlin, Berlin, Germany.

Author contributions

B Vogg contributed to writing, review and editing, conceptualization, software, data curation, methodology, formal analysis and validation. J Poetzl contributed to review and editing, conceptualization, data curation, methodology and validation. R El Galta contributed to review and editing, conceptualization, data curation, methodology, formal analysis and validation. R Fuhr contributed to writing, review and editing and investigation. A Schwebig contributed to writing, review and editing and conceptualization. S Sekhar contributed to review and editing, conceptualization, data curation and methodology. All authors contributed to the article and approved the submitted version.

Acknowledgments

We thank the people who participated in this study. Medical writing assistance was provided by Syneos Health Medical Communications, and funded by Hexal AG, Holzkirchen, Germany. Data from this study was previously presented in summary form at the 2023 ASCO Annual Meeting I, held from 2 to 6 June in Chicago. The corresponding abstract was published in Journal of Clinical Oncology: Vogg B, Poetzl J, El Galta R, Fuhr R, Schwebig A, and Sekhar S. Pharmacokinetics and pharmacodynamics of proposed denosumab biosimilar and reference denosumab in healthy male subjects. DOI: 10.1200/JCO.2023.41.16_suppl.e14500 Journal of Clinical Oncology 41, no. 16_suppl (June 01, 2023) e14500-e14500. Published online May 31, 2023.

Data availability statement

Data from this study are available from the corresponding author on request.

Additional information

Funding

This study was funded by Hexal AG, Holzkirchen, Germany.