Clinical immunogenicity and pharmacokinetic assessments of E3112, a recombinant human hepatocyte growth factor

Abstract

Aims

E3112 is a recombinant human hepatocyte growth factor (HGF) intended for the treatment of acute liver failure. As the presence of anti-drug antibody (ADA) against E3112 could pose a significant risk in clinical settings if it cross-reacts with the body’s natural HGF, we have developed assays for E3112 and its ADA in human serum.

Methods

Assays of E3112 and its ADA developed by ligand binding assays were validated in accordance with bioanalytical guidelines and applied to clinical pharmacokinetic (PK) and immunogenicity assessments.

Results

These assays demonstrated the ability to detect E3112 at a concentration as low as 0.156 ng/mL, whereas the sensitivity of ADA was determined to be 42.3 ng/mL. The validation studies, incorporating quality control for the PK assay and positive control of ADA, substantiated the reproducibility of the assays. The ADA and PK assays were applied to the real sample assays supporting a clinical trial of E3112. Following the intravenous administration of E3112, serum E3112 levels declined with a half-life of 19.3 h. No ADA was detected in predose or postdose samples.

Conclusion

These findings collectively indicate that E3112 exhibited a favorable PK profile with minimal immunogenicity within the clinical context.

PLAIN LANGUAGE SUMMARY

E3112 is a new protein-based medicine being developed to treat acute liver injury. To understand how this drug works in the body, it is important to measure both the drug levels in the blood and whether the body produces anti-drug antibodies (ADA), which may affect its safety or efficacy. In this study, we developed simple and reliable assays to measure E3112 and its ADA in human serum. These methods were fully validated according to regulatory guidelines and were successfully used in a clinical study to evaluate how E3112 behaves in the body (pharmacokinetics) and whether it triggers immune responses (immunogenicity).

Using these methods, we showed that E3112 can be accurately measured in human serum, and no ADA was detected after administration in the clinical study. These results suggest that E3112 has a favorable pharmacokinetic profile and a low risk of immune reactions in humans. Overall, this study provides practical tools and important information that help clinicians and researchers better understand how E3112 behaves in the body.

ARTICLE HIGHLIGHTS

1. E3112 is a recombinant human hepatocyte growth factor under development.

2. A ligand-binding assay was developed to quantify E3112 and detect anti-drug antibodies (ADAs).

3. The assays for E3112 and its ADAs were validated in accordance with bioanalytical guidelines.

4. The pharmacokinetics and immunogenicity of E3112 were evaluated in humans.

1. Introduction

Hepatocyte growth factor (HGF) is an endogenous growth factor mainly generated by mesenchymal cells such as fibroblasts, and can be also generated by macrophages and neutrophils [1,2]. It has been demonstrated to promote the growth and repair of endothelial cells by way of its angiogenic effects [3]. Recombinant HGF has been developed as a potential treatment for a number of diseases, including hepatic failure, amyotrophic lateral sclerosis, and damaged spinal cord [4,5]. E3112, a novel recombinant human HGF, was developed by Eisai Co., Ltd for the treatment of acute liver failure. The immunogenicity and pharmacokinetic (PK) assessments of E3112 have been reported in nonclinical settings in rats and monkeys [6,7]. However, repeated intravenous administration of E3112 has been associated with the production of anti-drug antibody (ADA) in rats and monkeys [6]. This finding warrants further investigation, as the presence of ADA against E3112 could pose a significant risk in clinical settings, particularly if it cross-reacts with the body’s natural HGF. It is hypothesized that the immunogenicity observed in nonclinical settings does not necessarily translate to clinical settings. Consequently, it is imperative to assess the potential for the immunogenicity of E3112 in clinical trials. A similar discrepancy has been observed in the PK, where the translatability from nonclinical to clinical settings might also be questionable. It is widely accepted that the PK of therapeutic antibodies in humans can be reliably predicted based on the PK observed in non-human primates, such as cynomolgus monkeys [8]. However, the reliability of PK prediction of recombinant growth factors in nonclinical animals, including monkeys, as a predictor of human PK remains to be fully elucidated. The PK of E3112 in nonclinical animals has been shown to exhibit nonlinear PK [7], which complicates the prediction of PK in humans compared to drugs with linear PK. These uncertainties in the prediction of PK and immunogenicity in humans from nonclinical animals highlight the necessity for further investigation into the assessment of PK and immunogenicity in human subjects. To assess the PK and immunogenicity of E3112 in humans, we have developed simple and reproducible assays for E3112 and its ADA in human serum using the ligand binding assay (LBA). The LBA platform is a well-established method for the analysis of biotherapeutics, including recombinant growth factors. The established assays were then subjected to assay validation in accordance with the relevant bioanalytical guidance from health authorities. These assays were subsequently applied to the PK and ADA assays in the clinical trial of E3112.

Recently, analytical methods other than LBAs have been developed for the quantification of biotherapeutics such as E3112 and the detection of ADA. In particular, liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS) has been increasingly applied for the analysis of biotherapeutics and, in some cases, ADA [9]. In the present study, LBAs were employed for the quantification of E3112 and the detection of its ADA in human serum. These assays were established based on the LBA methods previously developed and validated in nonclinical studies [6,7].

The use of LBAs offers several advantages. First, LBAs generally provide superior sensitivity compared with LC-MS/MS, enabling the detection of very low analyte concentrations. This is particularly important in clinical studies where the administered dose is low and circulating drug levels are expected to be minimal. Second, LBAs allow high-throughput sample analysis using microplate-based formats (e.g., 96-well plates), whereas LC-MS/MS typically requires more labor-intensive sample preparation procedures. On the other hand, LBAs also have several limitations. One major disadvantage is their relatively lower specificity compared with LC-MS/MS. LC-MS/MS can distinguish intact drug molecules from their catabolites or degradation products based on molecular mass, whereas LBAs generally cannot differentiate these forms if they retain binding activity. In addition, LBA performance is highly dependent on the quality of critical reagents, such as capture and detection antibodies, the development of which can be time-consuming and costly. Considering these factors, LBAs were selected in this study because high analytical sensitivity was essential for accurately quantifying E3112 at the expected low concentration levels in human samples.

Bridging electrochemiluminescence (ECL)-based assays are widely used for ADA detection, but circulating drug may interfere with ADA detection and cause false-negative results. Therefore, improving drug tolerance, defined as the ability of an assay to detect ADA in the presence of circulating drug, is important to reduce the risk of false-negative ADA results. Several approaches have been reported to improve drug tolerance. Acid dissociation pretreatment dissociates drug-ADA immune complexes under low-pH conditions [10]. Solid-phase extraction with acid dissociation (SPEAD) removes residual drug before ADA detection [11]. Biotin-drug extraction with acid dissociation (BEAD) extracts total ADA (free and drug bound ADA) using two-step acid dissociation, biotinylated drug, and streptavidin-coated magnetic beads [12]. Precipitation and acid dissociation (PandA) detects total ADA by polyethylene glycol precipitation of drug–ADA complexes followed by acid dissociation [13].

In the present study, the conventional bridging ECL assay approach was employed for the ADA detection because the developed assay demonstrated sufficient drug tolerance under the clinical conditions of this study, where serum concentrations of E3112 were relatively low. Specifically, the assay showed drug tolerance up to at least 1000 ng/mL, which is substantially higher than the observed C_max of E3112 in this study (13.0 ng/mL). Therefore, interference by circulating drug was considered negligible.

The primary motivation of this study is to evaluate the PK and immunogenicity of E3112, a novel recombinant human HGF, in humans. Understanding these properties is essential for assessing the clinical performance of E3112, as PK and immunogenicity can significantly influence both efficacy and safety.

The data generated in this study are expected to contribute to a better understanding of the dose-exposure relationship, as well as the potential impact of ADAs on drug disposition. Such information is critical for optimizing dosing strategies and for predicting therapeutic outcomes and potential adverse effects in clinical settings.

To our knowledge, this is the first study to report the PK and immunogenicity of this novel recombinant HGF in humans. Therefore, the findings provide important translational insights from nonclinical to clinical settings. In addition, this study may be informative for clinicians involved in the treatment of acute liver failure, where recombinant HGF is considered a promising therapeutic approach.

2. Experimental

2.1. Materials

E3112 was prepared at Eisai Co., Ltd (Ibaraki, Japan). A rabbit anti-E3112 antibody was prepared as a surrogate positive control of ADA (PC) by Shin Nippon Biomedical Laboratories, Ltd. (Kagoshima, Japan). Recombinant human HGF receptor/c-Met was procured from R&D systems, Inc. (Minneapolis, MN, USA). The HGF Otsuka ELISA kit was procured from Otsuka Pharmaceutical Co., Ltd (Tokyo, Japan). Human serum was obtained from Tennessee Blood Services (Memphis, TN, USA). Phosphate-buffered saline (PBS) and EZ-link Sulfo-NHS-biotin were procured from Thermo Fisher Scientific (Waltham, MA, USA). MSD GOLD SULFO-TAG NHS-ester was procured from Meso Scale Diagnostics (MSD; Rockville, MD, USA). Tris buffered saline with Tween 20 and ProClin 150 were obtained from Sigma-Aldrich (St. Louis, MO, USA). Block Ace powder was procured from DS Pharma Bio Medical (Osaka, Japan).

2.2. Preparation of critical reagents

Biotin-labeled E3112 and SULFO-TAG labeled E3112 were prepared by the procedures previously reported [6]. The Sulfo-NHS-biotin was mixed with E3112 at a molar ratio of 20:1, where the molecular weight of E3112 is 90,000, and then incubated at room temperature. The resulting mixture was then applied to a Zeba column (Thermo Fisher Scientific), and the eluate was obtained by centrifugation of the column. The protein concentration of the eluate was determined by the BCA Protein Assay kit (Thermo Fisher Scientific). The storage solution (10% BSA in PBS containing 0.05% ProClin 150) was added to achieve a concentration of 0.51 mg/mL of biotinylated E3112, which was subsequently stored at −80 °C. The SULFO-TAG labeled E3112 was prepared by mixing SULFO-TAG solution with E3112 at the molar ratio of 12:1 and then the mixture was incubated at room temperature. The resulting solution was subsequently applied to the Zeba column, and the eluate (SULFO-TAG labeled E3112) was obtained as previously described. The storage solution was then added to the eluate, yielding a concentration of 0.471 mg/mL. The solution was subsequently stored at −80 °C. The master mix solution was prepared by combining the biotinylated E3112 solution at concentration of 1.0 μg/mL diluted with assay buffer (mixture of equal volume of blocking buffer and wash buffer) and 2.0 μg/mL of SULFO-TAG labeled E3112 solution (1:1, v/v).

2.3. Preparation of calibration and PC and quality control (QC) samples

2.3.1. ADA assay

The original PC solution (1.69 mg/mL) was added to human serum to prepare PC samples at five concentration levels. A negative control (NC) sample was also prepared by mixing serum from 50 individuals. Calibration samples were prepared at the PC levels of 100, 200, 600, 2000, 6000, 20,000, and 60,000 ng/mL in human serum. The QC concentrations of PC were 100, 300, 3000, 48,000, and 60,000 ng/mL for the lower limit of quantification (LLOQ), low concentration PC (LPC), middle concentration PC (MPC), high concentration PC (HPC), and upper limit of quantification (ULOQ), respectively. In the drug tolerance test, E3112 was spiked in human serum at 0, 1, 10, 100, 1000, 10,000, 100,000, and 1,000,000 ng/mL.

2.3.2. PK assay

Calibration samples at the concentrations of 0.156, 0.313, 0.625, 1.25, 2.50, 5.00, and 10.0 ng/mL were prepared in phosphate buffer. The QC samples were prepared at 0.156, 0.450, 1.50, 6.00, and 10.0 ng/mL for the LLOQ, low concentration QC (LQC), middle concentration QC (MQC), high concentration QC (HQC), and the ULOQ, respectively. QC samples were prepared in phosphate buffer in preparing the LLOQ and ULOQ samples, while in human serum in preparing the other concentrations (LQC, MQC, and HQC).

2.4. Assay procedure

2.4.1. ADA assay

A 50 µL sample diluted 10-fold with the assay buffer was mixed with the master mix (100 µL) in duplicates in a 96-well polypropylene plate. The mixture was then subjected to overnight incubation at 5 °C, resulting in the formation of a reaction solution. The blocking buffer (150 µL) was added to each well of MULTI-ARRAY 96-well Streptavidin Gold Plate (ECL Plate) and incubated at 5 °C overnight. The plates were then washed three times with wash buffer (0.05 mol/L Tris, 0.138 mol/L NaCl, 0.0027 mol/L KCl, 0.05% Tween 20, pH 8.0), followed by the addition of 50 µL of the reaction solution in duplicates. The plates were incubated at room temperature for 1 h. Subsequently, the plates were washed three times, and 150 µL of MSD Read Buffer T (4×) with Surfactant diluted two-fold with purified water was added. The ECL signal was subsequently measured using the Sector Imager 6000 or MESO QuickPlex SQ 120 (MSD).

2.4.2. PK assay

The enzyme linked immunosorbet assay (ELISA) procedure of the PK assay previously reported was used in this study [7]. The concentrations of E3112 in human serum were determined using an ELISA kit (Otsuka Pharmaceutical Co., Ltd). The 96-well plates, which had been pre-coated with anti-HGF monoclonal antibody, were washed with a wash buffer three times. Then, 50 µL of phosphate buffer was added to each well. A 50 µL aliquot of the samples was added to the wells in duplicates, resulting in a minimum required dilution of 2, and the plates were shaken at room temperature for 1 h. The wells were then washed five times. The anti-HGF polyclonal antibody solution (100 µL) was added to each well, and the plates were shaken at room temperature for 1 h. The wells were washed five times, and then the peroxidase conjugated anti-rabbit IgG antibody solution (100 µL) was added. After the plates were shaken at room temperature for 1 h, the wells were then washed five times. The substrate solution of o-phenylenediamine dihydrochloride (100 µL) was added to each well and incubated at room temperature for 20 min. Subsequently, the stop solution (100 µL) was added to each well, and the absorbances at 492 and 620 nm were determined by a microplate reader.

2.5. Method validation

The method validation study was conducted based on the bioanalytical guidance from the United States of America Food and Drug Administration (US-FDA) for the PK assay [14], and the immunogenicity guidance from the US-FDA for the ADA assay [15].

2.5.1. ADA assay

The screening- and confirmatory-cutoff were statistically determined with the outliers excluded in the statistical analysis. The screening cutoff was determined using blank serum (n = 50) across six batches. The screening cutoff in each batch was calculated by dividing the upper limit of the 95th percentile of the ECL signals of the blank serum by that of the negative control. The mean screening cutoff across these six batches was then defined as the cutoff for the screening process. The same blank sera from individuals were used to estimate the confirmatory cutoff. The confirmatory cutoff was defined by the upper limit of the 99.9 percentile of the inhibition percentage in individual blank sera by the addition of excess E3112. The evaluation process of the assay encompassed the assessment of linearity, prozone effect, intra- and inter-batch reproducibility, selectivity, drug tolerance, and stability. The concentrations of the calibration samples of PC ranging from 100 to 60,000 ng/mL and their corresponding ECL signals were plotted, and the plot was subjected to 4-parameters regression with a 1/signal² weighting using the following equation: signal = b1 + (b2 − b1)/[1 + (concentration/b3)^b4], where b1, b2, b3, and b4 represent the ECL signal at the infinite concentration, the y-intercept, the concentration at the midpoint between b1 and b2, and the slope of the calibration curve, respectively. The relative error (RE) of calibration samples should be within ± 20% (± 25% at the LLOQ). The potential of the prozone effect was evaluated by assaying PC sample at concentration greater than the ULOQ. The prozone effect is regarded as significant when the signal of the PC sample (320,000 ng/mL) is lower than that of the ULOQ (60,000 ng/mL). The intra- and inter-batch reproducibility of the PC concentrations was evaluated at the five levels (LLOQ, LPC, MPC, HPC, and ULOQ). The intra-batch test involved the analysis of six replicates per concentration, while the inter-batch test involved the analysis of a single replicate across six batches. The RE and relative standard deviation (RSD) at each PC level were calculated. The RE and RSD should be within ± 25% and 25%, respectively, at the LLOQ, whereas ± 20% and 20%, respectively, at the other concentrations. The selectivity of the assay was evaluated using blank sera from 10 individuals. Non-spiked sera and sera spiked with PC at 100 ng/mL were assayed. The RE of the spiked sera should be within ± 25% in at least eight of the ten individuals. The dilution integrity was investigated by diluting PC by 10- and 100-fold. The RE and RSD of the diluted samples were expected to be within ± 20% and 20%, respectively. The sensitivity was determined by plotting the response index (RI), calculated by dividing the ECL signals of the samples by that of a negative control, at PC concentrations ranging from 15.6 to 2000 ng/mL against the corresponding nominal PC levels. The interpolated concentration by linear regression, which yielded the screening cutoff, was considered as the sensitivity. The assessment of drug tolerance was conducted by evaluating the combination of two levels of the drug (E3112) and PC. The PC concentrations were set at 100 and 300 ng/mL, while the E3112 levels were set at 1, 10, 100, 1000, 10,000, 100,000, and 1,000,000 ng/mL. At each PC level, the ECL signals were determined in the presence of E3112 at the varying levels. The highest E3112 level that yielded an ECL signal greater than the screening cutoff was designated as the drug tolerance level at each PC level. The stability of PC in human serum was evaluated at low and high PC levels (n = 3/concentration) for the following parameters: bench-top stability, freeze-thaw stability, and frozen stability. The bench-top stability was evaluated at room temperature for 24 h. The freeze-thaw stability was assessed after six freeze-thaw cycles between either −20 °C or −80 °C and room temperature. The frozen stability was evaluated at either −20 °C or −80 °C for 735 days. The RE of the stored PC samples should be within ± 20% when deemed to be stable. While the PC utilized in the stability assessment may not accurately reflect the ADA in actual human sera obtained following E3112 doses, the outcomes of the stability assessment are regarded as valuable in demonstrating the general stability of the ADA under standard storage conditions.

2.5.2. PK assay

Linearity, prozone effect, intra- and inter-batch reproducibility, selectivity, dilution integrity, effect of the target ligand (c-Met), and stability were evaluated in the method validation study for the determination of E3112 in human serum. The linearity of the assay was evaluated using calibration samples containing E3112 concentrations ranging from 0.156 to 10.0 ng/mL. The nominal concentrations of E3112 and their corresponding signals were plotted and subjected to 4-parameter nonlinear regression, as previously described, with no weighting. The linearity was ensured when the RE of calibration samples was within ± 25% at the LLOQ and ULOQ and within ± 20% at the other concentrations. The potential for the prozone effect was evaluated using samples spiked with E3112 at 2.00 and 2000 µg/mL. It was determined that the signals of the samples should exceed those of the ULOQ to ensure the absence of prozone effects. The intra- and inter-batch reproducibility of the samples was evaluated using QC samples at the five concentrations (LLOQ, LQC, MQC, HQC, and ULOQ). Within each batch, six replicates were assessed for intra-batch reproducibility, while a single replicate was evaluated across six batches for inter-batch reproducibility. The RE and RSD should be within ± 25% and 25%, respectively, at the LLOQ and ULOQ, while at the other concentrations, they should be within ± 20% and 20%, respectively.

The RE and RSD were calculated based on the following equation. The RE should be calculated by taking endogenous HGF levels into consideration in blank human serum for QC samples at LQC, MQC, and HQC, while endogenous HGF level is assumed to be zero for the LLOQ and ULOQ since these two levels QC samples were prepared in phosphate buffer.

$$\begin{array}{c}\begin{array}{c}\text{RE\hspace{0.17em}}\left(\%\right)\hspace{0.17em}=\hspace{0.17em}\left[\text{observed\hspace{0.33em}concentration\hspace{0.17em}}-\hspace{0.17em}\left(\text{endogenous\hspace{0.33em}HGF\hspace{0.33em}level\hspace{0.17em}\hspace{0.17em}}+\text{\hspace{0.17em}\hspace{0.17em}nominal\hspace{0.33em}added\hspace{0.33em}concentration}\right)\right]\hspace{0.17em}/\\ \left(\text{endogenous\hspace{0.33em}HGF\hspace{0.33em}level\hspace{0.17em}\hspace{0.17em}}+\text{\hspace{0.17em}\hspace{0.17em}nominal\hspace{0.33em}added\hspace{0.33em}concentration}\right)\times 100.\end{array}\\ \text{RSD\hspace{0.17em}}\left(\%\right)=\text{\hspace{0.17em}\hspace{0.17em}Standard\hspace{0.33em}deviation\hspace{0.17em}}/\text{\hspace{0.17em}Mean}\times 100.\end{array}$$

The selectivity of the assay was assessed in human serum from 10 individuals. The individual sera were spiked with E3112 at a concentration of 0.450 ng/mL, and the RE of the samples was determined. The RE is expected to vary within a range of ± 20% across eight of the ten subjects. The dilution integrity was assessed by multiple dilution folds, including 10-, 2000-, and 500,000-fold. The dilution integrity samples were prepared at 40, 4000, and 1,000,000 ng/mL and diluted by 10-, 2000-, and 500,000-fold, respectively. The RE and RSD of the diluted samples should be within ± 20% and 20%, respectively. The potential impact of c-Met on the assay of E3112 was assessed at c-Met concentrations of 250 and 500 ng/mL against the E3112 levels of 0.450 and 6.00 ng/mL. The RE of E3112 in the presence of c-Met was determined and should be within ± 20% to confirm no effect of c-Met on the assay of E3112. The stability of E3112 in human serum was evaluated for bench-top stability, freeze-thaw stability, and frozen stability. E3112 samples at low (0.450 ng/mL) and high (6.00 ng/mL) concentrations were examined, and the RE of the stored samples was subsequently determined. The mean RE of triplicate determinations should be within ± 20%. The bench-top stability was tested when samples were stored at room temperature for 24 h under protection from light. The freeze-thaw stability was tested after five cycles between the room temperature and either −20 °C or −80 °C. The frozen stability was assessed when stored for 733 days at either −20 °C or −80 °C.

2.6. Application to a clinical study

A clinical study was conducted at Houeikai Sekino Clinical Pharmacology Clinic in Japan from 11 January 2017 to 12 June 2017. This study was performed according to the principles of the Declaration of Helsinki, the International Conference on Harmonization guidelines for Good Clinical Practice, and local regulations in Japan. This study was reviewed and approved by Houeikai Institutional Review Board with the approval number: 56001 on 8 December 2016, and registered at www.ClinicalTrials.gov. The registration number is NCT03014895 (first registered on 1 January 2017, URL: https://www.clinicaltrials.gov/study/NCT03014895). All participants provided written informed consent to participate in the study and for their data to be published.

E3112 was administered at a dose of 1 mg to six healthy volunteers by intravenous infusion for 1 h. For PK assessment, blood samples were collected in the predose, at the end of infusion (EOI, 1 h), 2, 3, 4, 6, 8, 10, 12, 24, 48, 72, 168, 336, and 672 h. For the immunogenicity assessment, blood samples were collected predose (Day 1), Day 8, Day 14, and Day 28. Serum samples were prepared by first collecting blood samples and then allowing them to keep at room temperature for approximately 1 h. Following this, the samples were subjected to centrifugation (2500 rpm, 4 °C, 15 min). The concentrations of E3112 were determined by the validated ELISA, and ADA levels were semi-quantitatively estimated by the validated ECL assay. Endogenous HGF subtraction was performed for all study samples. Specifically, serum E3112 concentrations were calculated by subtracting individual endogenous HGF concentrations from the observed total HGF (E3112 + endogenous HGF) at each sampling point. Endogenous HGF levels were assumed to be constant over time within each individual and were determined using predose samples collected from each subject. Therefore, each subject’s predose HGF concentration was used for correction at all postdose timepoints. This subject-specific correction minimizes the impact of inter-individual variability in endogenous HGF levels on the estimation of serum E3112 concentrations. To ensure the reproducibility of the PK assay, incurred sample reanalysis (ISR) was conducted by repeat assay of 44 samples. The bias between the original and repeat assays was calculated by the following equation and should be within ± 30% for at least two-thirds of the tested samples: bias (%) = (reassay data − original data)/mean of the original and reassay data × 100.

The PK parameters of E3112 were estimated by non-compartmental analysis using WinNonlin software (Certara, Radnor, PA, USA). The maximum concentration in serum (C_max), the area under the curve of serum concentration-time profile from the time = 0 to infinite time (AUC_0-inf), the elimination half-life (T_1/2), the total clearance (CL_tot), and the volume of distribution at the steady state (V_ss) were estimated.

In the ADA assay, a tiered approach was taken. Initially, the samples were assessed using a screening assay to ascertain whether the RI equaled or exceeded the screening cutoff. Subsequently, samples with the RI equal to or greater than the screening cutoff were subjected to a confirmatory assay. The final determination of ADA positivity was based on the inhibition percentage in the presence of E3112 at 10.0 μg/mL, which had to exceed the confirmatory cutoff. A schematic diagram for the ADA assay workflow is shown in Figure 1.

Figure 1.

A schematic diagram for the ADA assay workflow. During assay validation for anti-E3112 antibodies (ADA), a screening cutoff was established using blank human serum samples from multiple lots, and a confirmatory cutoff was determined in the presence of excess E3112 by statistical analysis. In the sample analysis, ADA signals were measured for each sample. Samples with signals equal to or exceeding the screening cutoff were considered screening-positive and were subsequently subjected to the confirmatory assay. Samples were defined as ADA-positive if the inhibition percentage in the presence of E3112 exceeded the confirmatory cutoff. As minimal ADA responses were observed in this study, no further characterization (e.g., titer determination) was performed.

3. Results and discussion

3.1. Method validation

3.1.1. ADA assay

The ECL signals of blank human serum from 50 individuals were determined in six batches (Figure 2A). The screening cutoff across the six batches ranged from 1.023 to 1.086, and the mean cutoff was determined to be 1.050, which was defined as the screening cutoff. The inhibition percentages of the ECL signals in blank serum from the same 50 individuals in the presence of the excess E3112 were determined in the six batches (Figure 2B). The inhibition percentage ranged from 37.1% to 39.7%, with an average of 39.0% across the six batches, which was designated as the confirmatory cutoff.

Figure 2.

Cutoff determination of anti-drug antibody (ADA) of E3112 in human serum. (A) screening cutoff (B) confirmatory cutoff. A total of 50 naïve human serum samples were assayed in six assay batches, both in the absence (A) and in the presence (B) of 10.0 µg/mL E3112. The ratios of signals from the samples to the negative control (NC) are represented in (A). The inhibition percentages of the signals in blank sera in the presence of E3112 are represented (B).

The calibration curves of PC in human serum exhibited a fit to a 4-parameter regression with a coefficient of determination of 0.999 or higher. The RE of calibration PC samples was within ± 25% at the LLOQ and ULOQ, and within ± 20% at the other concentrations (Figure 3). When prozone test samples at a PC level of 320,000 ng/mL were assayed, ECL signals were lower than that of the ULOQ, indicating a prozone effect (Figure 3). To assess the reproducibility within and across batches, PC samples were utilized at five levels: LLOQ, LPC, MPC, HPC, and ULOQ. The RE and RSD of PC samples are summarized in Table 1. The RE and RSD were within ± 20% and 20%, respectively, at all the levels that were tested. The selectivity of the assay was evaluated in serum from 10 individuals. The ECL signals in unspiked samples were lower than that of the LLOQ, and the RE of spiked samples (100 ng/mL PC) was within ± 20% in nine of ten individuals (Table 2). These findings ensure the assay’s selectivity. The dilution integrity was assessed by 10- and 100-fold dilution of samples. The RE and RSD were 10.9% and 1.9%, respectively, by 10-fold dilution. By 100-fold dilution, the RE and RSD were 5.5% and 2.0%, respectively. These findings confirm the assay’s dilution integrity. The sensitivity was estimated by assaying calibration samples in six batches. The mean sensitivity in the six batches was 42.3 ng/mL, which was sufficiently low given the recommendation by the immunogenicity guidance from the US-FDA [15]. It is imperative to assess drug tolerance on the PC assay within the context of the immunogenicity assessment. In this study, drug tolerance was evaluated at the PC levels of 100 and 300 ng/mL in the presence of varying concentrations of E3112 and the results are shown in Figure 4. At a PC level of 100 ng/mL, ECL signals of PC were higher than the screening cutoff in the presence of E3112 at 1000 ng/mL, while lower than the screening cutoff in the presence of 10,000 ng/mL E3112. At a PC level of 300 ng/mL, ECL signals were higher than the screening cutoff in the presence of 10,000 ng/mL E3112. However, when E3112 levels were increased to 100,000 ng/mL, ECL signals demonstrated a lower magnitude. These findings suggest that the minimum drug tolerance levels were 1000 and 10,000 ng/mL, respectively, at PC levels of 100 and 300 ng/mL. Stability assessments, including bench-top stability, freeze-thaw cycles, and frozen stability, were conducted at 300 and 48,000 ng/mL, and the results are represented in Supplementary Table S1. The stability of the drug at room temperature was ensured for 24 h. The freeze-thaw stability was confirmed even after six cycles between thawed at room temperature and frozen at either −20 °C or −80 °C. The stability of PC in serum under frozen conditions was ensured for 735 days at either −20 °C or −80 °C. Although the RE values of the stability samples under various conditions were within the acceptance criteria, some values were close to the predefined limits. Therefore, variability among individual replicates was further evaluated. For all stability tests, three replicates per concentration were analyzed. The RSDs of the triplicate determinations were ≤2.9% for freeze–thaw stability, ≤3.3% for bench-top stability, ≤4.8% for frozen stability at −80 °C, and ≤10.7% for frozen stability at −20 °C, indicating low variability among replicates. Some individual replicates did exceed the acceptance criteria, however, these deviations were marginal (i.e., 20.3%, 23.0%, and 23.3% against ± 20%), indicating that the stability of E3112 in human serum was not compromised.

Figure 3.

Representative calibration curves of anti-E3112 antibody in human serum. Calibration samples prepared by spiking positive control of anti-E3112 antibody (PC) in human blank serum at 100–60,000 ng/mL were assayed and the corresponding signals were determined and plotted at x- and y-axis, respectively. The four-parameter regression was employed to yield a calibration curve. In the prozone effect test, a sample at higher than the upper limit of quantification (320,000 ng/mL) was assayed and the signal was plotted in open circle.

Table 1.

Accuracy and precision of anti-drug antibody positive control of E3112 in human serum.

Intra-batch	Concentration (ng/mL)	RE (%)	RSD (%)
LLOQ	100	2.9	5.1
LPC	300	1.3	3.4
MPC	3000	3.7	1.2
HPC	48,000	−0.4	2.0
ULOQ	60,000	−3.7	2.1
Inter-batch	Concentration (ng/mL)	RE (%)	RSD (%)
LLOQ	100	8.3	4.9
LPC	300	0.8	3.6
MPC	3000	1.4	4.6
HPC	48,000	−2.1	1.5
ULOQ	60,000	−1.1	2.9

Data represent the mean relative error (RE) and relative standard deviation (RSD) of six replicate of positive control of anti-E3112 antibody (PC) per concertation in the intra- and inter-batch assays. HPC, high concentration PC; LLOQ, lower limit of quantification; LPC, low concentration PC; MPC, middle concentration PC; ULOQ, upper limit of quantification.

Table 2.

Selectivity of anti-drug antibody positive control of E3112 in human serum.

Individuals	Unspiked	PC (100 ng/mL)
	Concentration (ng/mL)	Concentration (ng/mL)	Recovery (%)
Male-1	5.99	85.8	79.8
Male-2	0.492	84.6	84.1
Male-3	0.00	83.1	83.1
Male-4	0.00	83.0	83.0
Male-5	0.00	87.1	87.1
Female-1	3.76	85.7	81.9
Female-2	0.00	90.3	90.3
Female-3	0.00	95.3	95.3
Female-4	4.38	90.9	86.5
Female-5	0.00	85.5	85.5

The determination of positive control of anti-E3112 antibody (PC) concentrations in unspiked sera and spiked sera is represented. The recovery percentage was determined using the following equation: Recovery (%) = (determined concentrations in spiked samples − determined concentrations in unspiked samples) / nominal concentration × 100.

Figure 4.

Drug tolerance for the assay of anti-E3112 antibody in human serum. Positive control of anti-E3112 antibody (PC) at 100 (A) and 300 (B) ng/mL in human serum was assayed in the presence of varying concentrations of E3112 (1–1,000,000 ng/mL). The highest E3112 concentrations which yielded signals greater than the screening cutoff (dotted line) were designated as the drug tolerance.

3.1.2. PK assay

Calibration curves were fitted by 4-parameters nonlinear regression curves with the concentrations ranging from 0.156 to 10.0 ng/mL. The coefficient of determination for these calibration curves was 1.00. The potential prozone effect was evaluated by assaying the sample at concentrations greater than the ULOQ (2.00 and 2000 µg/mL). The signal of the sample exceeded that of the ULOQ, thereby indicating an absence of prozone effect for the assay. The outcomes of the intra- and inter-batch reproducibility assessments are represented in Table 3. The RE and RSD in the intra- and inter-batch tests were not more than ± 20% and 20%, respectively. These findings collectively suggest that the assay’s accuracy and precision have been adequately validated. The selectivity test results are presented in Table 4. The RE in all the samples from ten individuals was within ± 20%, indicating that the selectivity of the assay was confirmed. The dilution integrity of the samples was assessed by conducting 10-, 2000-, and 500,000-fold dilutions. The RE and RSD by 10-fold dilution were −5.3% and 1.3%, respectively, 4.4% and 1.1%, respectively, by 2000-fold dilution, and 14.8% and 0.8%, respectively, by 500,000-fold. These findings indicate that the dilution integrity was ensured up to 500,000-fold. The effect of c-Met, a soluble ligand of E3112, on E3112 assay was evaluated at two levels of E3112. At concentrations of 0.450 and 6.00 ng/mL, the RE of E3112 was −3.7% and 5.9%, respectively, at c-Met levels of 500 ng/mL. This finding suggests that c-Met does not exert a significant influence on the quantification of E3112 at concentrations up to 500 ng/mL. Given that the soluble c-Met levels in human serum were lower than 500 ng/mL [16], it is unlikely that c-Met interferes with the E3112 assay. The stability of E3112 was assessed in various conditions, including bench-top stability, freeze-thaw stability, and frozen stability (Supplementary Table S2). The stability at room temperature was ensured for 24 h, while the freeze-thaw stability was tested at between either −20 °C or −80 °C and room temperature. Notably, the stability was confirmed to persist even after five successive freeze-thaw cycles at either temperature. The frozen stability was evaluated at both −20 °C and −80 °C. The frozen stability was ensured up to 733 days at both temperatures. The findings from these method validation studies indicated that the developed assay was both robust and reproducible, thereby ensuring its applicability to real human serum samples for the determination of concentrations of E3112.

Table 3.

Accuracy and precision of E3112 assay in human serum.

Intra-batch	Concentration (ng/mL)	RE (%)	RSD (%)
LLOQ	0.156	−6.1	4.7
LQC	0.450	−1.6	3.2
MQC	1.50	−2.6	2.1
HQC	6.00	−3.9	3.7
ULOQ	10.0	−7.0	3.1
Inter-batch	Concentration (ng/mL)	RE (%)	RSD (%)
LLOQ	0.156	−0.9	9.3
LQC	0.450	1.3	6.7
MQC	1.50	0.9	5.7
HQC	6.00	−0.7	3.3
ULOQ	10.0	−5.9	2.6

Data represent the relative error (RE) and relative standard deviation (RSD) of six replicates of E3112 quality control (QC) samples per concentration. HQC, high concentration QC; LLOQ, lower limit of quantification; LQC, low concentration QC; MQC, middle concentration QC; ULOQ, upper limit of quantification.

Table 4.

Selectivity of E3112 in human serum.

Individuals	Unspiked	E3112 (0.450 ng/mL)
	Concentration (ng/mL)	Concentration (ng/mL)	RE (%)
Male-1	0.241	0.675	−2.3
Male-2	0.206	0.592	−9.7
Male-3	0.195	0.555	−13.9
Male-4	0.177	0.548	−12.6
Male-5	0.136	0.481	−17.8
Female-1	0.114	0.476	−15.7
Female-2	0.121	0.478	−16.4
Female-3	0.151	0.519	−13.6
Female-4	0.131	0.510	−12.2
Female-5	0.168	0.572	−7.5

E3112 concentrations in both naïve unspiked sera and spiked sera at the low concentration quality control level (LQC, 0.450 ng/mL) from 10 individuals are represented. The relative error (RE) was calculated by the following equation: RE (%) = [determined concentration in spiked samples − (determined concentrations in unspiked samples + nominal concentration)] / (determined concentration in unspiked samples + nominal concentration) × 100.

3.2. Application to a clinical study

3.2.1. Immunogenicity

The ADA against E3112 was evaluated in samples obtained in predose, Day 8, Day 14, and Day 28 post the EOI of E3112 at 1 mg by the validated ECL assay. In the screening assay, three samples from two subjects were determined to be positive with an RI of 1.050 or higher. Subsequent confirmatory assay of the potential ADA positive samples showed that the inhibition percentage of all samples was below the confirmatory cutoff of 39.0%. The findings indicated the absence of ADA in samples following the single intravenous administration of E3112 to humans.

3.2.2. PK

E3112 concentrations in serum were determined using serum samples obtained from subjects dosed at 1 mg by the validated ELISA. In this study, predose endogenous HGF levels in subjects receiving 1 mg intravenous administration ranged from BLQ (< 0.156 ng/mL) to 0.338 ng/mL. Because calibration standards for the E3112 assay were prepared in serum-free buffer without endogenous HGF, the assay itself does not incorporate endogenous HGF into the calibration curve. As a result, endogenous HGF does not directly affect assay calibration or analytical sensitivity.

The serum E3112 levels exhibited a maximum at the EOI, subsequently declining with a T_1/2 of 19.3 h (Figure 5). The PK parameters of E3112 in humans are represented in Table 5. The C_max of E3112 was 13.0 ng/mL, and the AUC_0-inf was 38.6 ng × h/mL. The CL_tot and V_ss were 43.2 L/h (approximately 720 mL/h/kg assuming a body weight of 60 kg) and 1720 L (approximately 29 L/kg), respectively. The inter-individual variability in PK parameters was relatively high, as indicated by the RSDs. The RSDs of C_max, AUC_0–inf, T_1/2, CL, and V_ss were 33.3%, 84.8%, 117%, 62.1%, and 104%, respectively. Two main factors may explain the large inter-individual differences observed in the PK of E3112.

Figure 5.

Time profiles of E3112 concentrations in serum after intravenous administration of E3112 to humans. E3112 was administered by intravenous infusion for 1 h to six healthy volunteers at 1 mg. Blood samples were collected up to 168 h after the initial dose, and serum concentrations of E3112 were determined by a validated enzyme-linked immunosorbent assay. Data represent mean ± standard deviation at each time point in six subjects.

Table 5.

Pharmacokinetic parameters of E3112 in humans after intravenous administration.

Parameters	Units	Estimate
C max	ng/mL	13.0 ± 4.32
AUC0-inf	ng × h/mL	38.6 ± 32.7
T 1/2	h	19.3 ± 22.7
CLtot	L/h	43.2 ± 26.8
V ss	L	1720 ± 1790

Data represent the arithmetic mean ± standard deviation of six subjects who received 1 mg of E3112 by the intravenous infusion for 1 h. Blood samples were serially collected up to 672 h postdose, and the resulting serum samples were used for the pharmacokinetic (PK) assay. The PK parameters were estimated by the non-compartmental analysis. AUC_0-inf, area under the concentration-time curve of serum from zero time to infinity; CL_tot, total clearance; C_max, maximum concentration in serum; T_1/2, elimination half-life; V_ss, volume of distribution at the steady state.

The first factor is the impact of endogenous HGF levels in each subject. Because E3112 concentrations in serum were calculated by subtracting individual endogenous HGF levels from the observed concentrations, the influence of endogenous HGF is relatively greater when E3112 concentrations are low. Endogenous HGF levels were determined using predose samples from each subject; however, potential intra-individual variability due to time-dependent fluctuations in endogenous HGF levels may have affected the PK parameter estimation. The second factor is variability in the PK characteristics of E3112 among individuals. HGF is primarily taken up by the liver, and hepatic uptake is considered a major clearance pathway. Therefore, inter-individual differences in hepatic uptake of E3112 may also contribute to the observed variability in PK parameters.

The ISR was conducted using 44 samples. The bias between the original assay and reassay was within ± 30% in 39 samples, which met the acceptance criteria of the ISR.

The CL_tot and V_ss in male cynomolgus monkeys at 0.5 mg/kg were 470 mL/h/kg and 1450 mL/kg, respectively, indicating comparable CL_tot with humans but significantly divergent V_ss for reasons that remain unclear. Given the nonlinear PK of E3112 in monkeys and the considerably lower dose in humans (approximately 0.04 mg/kg), it is conceivable that the V_ss in monkeys may be elevated at lower doses relative to those at 0.5 mg/kg, potentially attributable to the saturation of E3112 uptake into the liver.

Safety, efficacy, and PK profiles of another recombinant HGF biotherapeutics have been reported [4,5,17]. KP-100 has been administered intrathecally in humans and the other recombinant HGF was administered intravenously. In a Phase I/II clinical study, recombinant HGF was intravenously administered to patients with acute liver failure [17]. A complex dosing regimen was employed, consisting of an intravenous dose of 0.6 mg/m² combined with stepwise dose escalation from 0.06 mg/m² to 0.36 mg/m². PK evaluations were performed on Days 1, 5, and 11. Serum HGF levels increased in a dose-dependent manner, and the C_max ranged from 18.8 to 22.3 ng/mL. The T_1/2 in serum ranged from 633 to 844 minutes. The AUC_0-inf at 0.6 mg/m² (approximately 1 mg) was 1994 ng × min/mL (33 ng × h/mL), which was similar to E3112 at the dose of 1 mg. It should be noted that recombinant HGF may not have been fully distinguishable from endogenous HGF in these measurements. Therefore, direct comparisons of PK parameters between E3112 and the biotherapeutic should be interpreted with caution. In a phase I study, KP-100 was intrathecally administered to patients with amyotrophic lateral sclerosis [4]. The mean T_1/2 of KP-100 in cerebrospinal fluid (CSF) was 1.2–1.4 days, and dose-dependent PK in CSF were observed. Systemic exposure was minimal following intrathecal administration, with only negligible levels detected in plasma. Furthermore, a Phase II study was conducted in patients with acute spinal cord injury [5]. Although PK was not evaluated in that study, KP-100 demonstrated potential therapeutic benefit during the acute phase, as indicated by improved motor function.

This study has several limitations. As a limitation of the ADA assay, target tolerance was not evaluated. High concentrations of the target antigen may potentially interfere with ADA detection and lead to false-positive results. Serum concentrations of c-MET, the target of E3112, have been reported to be 189 ng/mL (range 124–397 ng/mL) in 26 healthy volunteers [16]. However, as minimal ADA responses were observed in the present clinical study, it is unlikely that such interference had a meaningful impact on the results. Therefore, false-positive ADA findings due to target interference are considered to be of limited relevance in this study. In the PK assay, E3112 and endogenous HGF could not be distinguished; therefore, E3112 concentrations were calculated by subtracting individual predose endogenous HGF levels. Intra-subject variability in endogenous HGF levels may impact the accuracy of E3112 PK assessment. To minimize the influence of endogenous HGF, PK evaluation at higher doses may be more appropriate. In addition, this study was a single-dose study conducted in healthy subjects. Although initial PK characterization is essential, further studies are required to evaluate potential changes in PK and immunogenicity in patients, as well as ADA development following multiple doses of E3112.

4. Conclusion

Assays for the assessment of the immunogenicity and PK of E3112 were developed and successfully validated in human serum using LBAs. The established ECL-based ADA assay demonstrated acceptable reproducibility and sufficient drug tolerance for detecting ADA positivity in human samples following intravenous administration of E3112. In addition, the ELISA assay utilizing a commercially available kit was simple, sensitive, and reproducible, with an LLOQ of 0.156 ng/mL, which was low enough to quantify E3112 concentrations in human serum. This assay was subsequently applied to PK characterization in a clinical study of E3112.

The PK profile of E3112 was evaluated in humans for the first time and showed a T_1/2 of 19.3 h following intravenous infusion at a dose of 1 mg. The ISR results demonstrated good reproducibility compared with the original assay results, further supporting the robustness of the PK assay. Overall, the established ADA and PK assays were useful for characterizing the immunogenicity and PK of E3112 in clinical settings.

Funding Statement

This research was supported by the Medical Research and Development Programs Focused on Technology Transfer: Adaptable and Seamless Technology Transfer Program through Target-Driven Research and Development (A-STEP) of the Japan Agency for Medical Research and Development (AMED).

Authors contributions

MA: conceptualization, data curation, formal analysis, investigation, manuscript reviewing. TT: conceptualization, data curation, investigation, manuscript reviewing. FO: conceptualization, data curation, investigation, manuscript reviewing. YM: conceptualization, supervision, manuscript writing, and reviewing.

Disclosure statement

The authors have no 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. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Data availability statement

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