Accelerating the speed of innovative anti-tumor drugs to first-in-human trials incorporating key de-risk strategies

ABSTRACT

Pharmaceutical companies have recently focused on accelerating the timeline for initiating first-in-human (FIH) trials to allow quick assessment of biologic drugs. For example, a stable cell pool can be used to produce materials for the toxicology (Tox) study, reducing time to the clinic by 4–5 months. During the coronavirus disease 2019 (COVID-19) pandemic, the anti-COVID drugs timeline from DNA transfection to the clinical stage was decreased to 6 months using a stable pool to generate a clinical drug substrate (DS) with limited stability, virus clearance, and Tox study package. However, a lean chemistry, manufacturing, and controls (CMC) package raises safety and comparability risks and may leave extra work in the late-stage development and commercialization phase. In addition, whether these accelerated COVID-19 drug development strategies can be applied to non-COVID projects and established as a standard practice in biologics development is uncertain. Here, we present a case study of a novel anti-tumor drug in which application of “fast-to-FIH” approaches in combination with BeiGene’s de-risk strategy achieved successful delivery of a complete CMC package within 10 months. A comprehensive comparability study demonstrated that the DS generated from a stable pool and a single-cell-derived master cell bank were highly comparable with regards to process performance, product quality, and potency. This accomplishment can be a blueprint for non-COVID drug programs that approach the pace of drug development during the pandemic, with no adverse impact on the safety, quality, and late-stage development of biologics.

KEYWORDS:

• Acceleration
• characterization
• CMC
• first-in-human
• monoclonal antibody
• risk mitigation

Introduction

In the traditional pharmaceutical life cycle, the pace from sequence discovery to the clinical stage is a critical rate-limiting factor in quickly searching for promising therapeutics. Chemistry, manufacturing, and controls (CMC), an essential element in drug supplements, is a tedious and time-consuming process that restricts the speed of first-in-human (FIH) trials. Amgen, Pfizer, Biogen, and other large pharmaceutical companies have explored different strategies to accelerate CMC and non-clinical development before entering the clinical stage.^1–4 Among these strategies, using a stable cell pool instead of a single-cell clone (SCC) to generate materials for toxicology (Tox) studies is one of the top choices. With technological advancements, such as the introduction of the transposon system to cell line development,^5–7 a more homogenous cell pool with improved productivity and a representative quality profile is feasible. A comparable titer, purity, charge, glycan profile, and potency between the pool and SCC have been demonstrated for multiple candidates that have entered the clinical phase.⁸ These new strategies shorten the preclinical development timeline from approximately 2 years to less than 18 months because of the use of cell pool materials in the Tox study.²

Since the 2019 pandemic outbreak, the exploration of new approaches to enable rapid entry into Phase 1 clinical studies has gained increasing attention.^9–12 The adjustment of the regulatory guidelines in response to coronavirus disease 2019 (COVID-19) triggered faster CMC timelines from DNA transfection to FIH trials and commercial use.^13,14 WuXi Biologics successfully achieved a 6-month investigational new drug (IND) timeline in America using proteins generated from a stable cell pool for Phase 1 clinical trials. However, this shortened timeline was achieved by sacrificing several steps during early CMC development: (1) using clonal-derived drug products for phase I clinical trials, (2) complete Master cell bank (MCB) stability study to ensure phenotypic stability during the production process, and (3) a comprehensive viral clearance and Tox study to demonstrate biosafety before submission to a regulatory agency. Some companies have even used transient cell lines to generate clinical materials to accelerate drug development during the COVID-19 pandemic.^12,14,15

Currently, with the waning global pandemic, it is unclear whether these accelerated strategies implemented for COVID-19 drug products can also be applied to non-COVID projects and become a common practice for biologics drug development.¹⁶ In contrast to COVID-19 programs, the complexity of the molecules, strict requirements for complete CMC data packages, and varying requirements of health/regulatory agencies in each country result in a longer time to FIH studies and need to be taken into consideration for traditional drug development.

In this study, we introduce the CMC workflow of a novel mAb project in which we applied accelerated actions taken from past COVID project experiences and BeiGene-specific strategies that mitigate safety and comparability risks, i.e., BeiGene’s “Fast-to-FIH” approach. This mAb is an innovative biological developed to treat advanced and metastatic solid tumors. A Chinese hamster ovary (CHO) α-1,6-fucosyltransferase gene knockout (FUT8-/-) cell line was used in this project to express an afucosylated (AF) antibody with enhanced antibody-dependent cell-mediated cytotoxicity (ADCC). The study aims to prove the feasibility of this “Fast-to-FIH” approach in substantially reducing CMC time for non-COVID drug development while also reducing extra efforts post-FIH and meeting global regulatory requirements. This case study of a successful mAb project demonstrates the importance of streamlining a rapid CMC workflow for a de-risk design and illustrates opportunities to accelerate future projects involving biologics with more complex structures and mechanisms.

Material and methods

CHO AF cell line creation for cell pool and SCC generation

The BeiGene CHO FUT8-/- cell line was created by deleting both alleles of FUT8 from the parental CHO glutamine synthetase gene knockout (GS-/-) cell line (Catalogue number HD-BIOP3), which was purchased from Horizon Discovery Ltd. Deletion was achieved using BeiGene’s proprietary zinc finger nuclease (ZFN) technology. The antibody produced in the AF cell line is a full-length, humanized IgG1 monoclonal antibody, which lacked core fucosylation but displayed a normal glycosylation pattern.

Based on the amino acid sequences, the heavy and light chains were synthesized in pUC57. The two expression fragments were inserted into one expression vector by ligating the corresponding restriction enzyme-digested fragments. The vector contained a GS-selectable marker from hamsters under the control of the SV40 promoter. The AF cell line underwent electroporation to introduce synthetic DNA encoding heavy and light chains for mAb expression. After electroporation, the cells were recovered in growth media for 5 weeks to obtain high viability and target growth rates. The stable cell pool was assessed for productivity, growth performance, and product quality using fed-batch cultures. The surface human Fc expression of the stable cell pool was analyzed using the guava flow cytometers (Cytek).

The stable pool was subjected to several passages, and single-cell cloning was performed using a WOLF cell sorter (Nanocellect) and limiting dilution. Plates were seeded at 1 and 0.6 cell/well in cloning medium. All plates were analyzed using Cell Metric Imager to assess clonality (Cell Metric), and plates were imaged on days 0, 1, 3, 8, and 12. Only the wells containing a single cell on day 0 were selected for scaling. Cells were passaged in spin tubes and subjected to a fed-batch assay. In total, 37 SCCs were identified for a side-by-side comparison of productivity and quality with the stable pool. These 37 clones were further narrowed down to four clones leveraging viable cell density (VCD), viability, titer, purity, charge variants, glycan profiles, and potency. The four clones were banked (60 vials per clone) to create research cell banks (RCBs), which were tested by WuXi Biologics Biosafety Testing (Suzhou) Co., Ltd. For sterility, mycoplasma detection, and adventitious virus contaminants (in vitro assay).

Cell line stability study

Banks of the stable cell pool and the four SCCs were passaged repeatedly every three days over a period of 60 population doubling levels (PDLs) using the same culture conditions. Intermediate banks were frozen at PDL0, PDL30, and PDL60; a 14-day fed-batch assay was performed on these intermediate banks. The supernatant was subjected to productivity measurements using Cedex Bio HT and a one-step purification with Protein A. The purified samples were analyzed for purity, charge variants, and glycan profiles. The lead SCC was determined using the PDL30 stability data for MCB preparation to accelerate the release of the MCB into clinical manufacturing. Genetic stability tests were applied to the MCB stability study to demonstrate genetic stability through 60 PDLs.

Tox and clinical batch production

The platform process used in both the Tox and clinical batches was evaluated in 3 (STR3, Sartorius) and 50 L bioreactors (STR50, Sartorius).

For Tox material generation, a frozen vial of stable cell pool banks was thawed and expanded from a 125 mL shake flask (N-5 stage) to a 20 L WAVE bag (N-1 stage). The cells were then inoculated into STR50 for a 14-day fed-batch culture. The supernatant was harvested, filtered through depth filtration, and purified using affinity chromatography and anion exchange and cation exchange chromatography. Low pH and viral filtration were used to remove the virus, and the drug substance (DS) was formulated after ultrafiltration and diafiltration.

The lead SCC selected from multiple clones generated from the stable cell pool was used for clinical batch manufacturing. The clinical batch was scaled up to a 500 L bioreactor with the same media, feeding strategy, target seeding density, temperature, and pH control. The downstream steps and materials were identical to those of the Tox batch, including larger filter sizes and column diameters.

Quality analysis

The supernatant titer was measured using high-performance liquid chromatography (HPLC) with step-wise elution and a protein A column (POROS™ A Column, Thermo Fisher Scientific). Purity, including monomer and high molecular weight (HMW) and low molecular weight (LMW) components, was determined using size exclusion chromatography-HPLC (SEC-HPLC) with a BioResolve mAb column (Waters). Purity was also evaluated using non-reduced and reduced capillary gel electrophoresis (NR/R-CGE) using a PA800 Plus system (AB SCIEX/Beckman Coulter). Charge variants were analyzed using imaged capillary isoelectric focusing (icIEF) with a Maurice system (Protein Simple). The released glycans were analyzed using a hydrophilic interaction liquid chromatography-ultra-performance liquid chromatography (HILIC-UPLC) system (Waters) after incubation with PNGase F and labeling with 2-AB.

The relative potency was determined using the ADCC assay. Hut78 and JK-Luc/CD16 cells were used, and the results were analyzed using a Bio-Glo luciferase assay system (Promega). Relative activity was determined as a percentage of test samples compared with the control. Binding kinetics and affinities of neonatal Fc receptor (FcRn) were measured using the surface plasmon resonance (SPR) technology (Biacore T200, Cytiva).

The primary structure and post-translational modifications (PTMs) were evaluated using liquid chromatography-mass spectrometry (LC-MS). Circular dichroism (CD) spectroscopy, differential scanning calorimetry (DSC), dynamic light scattering (DLS), and analytical ultracentrifugation (AUC) were used for higher-order structure analyses. The free thiol content was measured using Ellman’s method.

Reference standard was included in all relevant tests and the method variability was calculated to ensure the comparability conclusion was not impacted by run-to-run variability. The release test methods are qualified to ensure their specificity, precision, accuracy, linearity, range, and robustness.

Results

Overview of “fast-to-FIH” CMC timeline and risk mitigation strategies

The CMC workflow of this mAb product candidate was completed in 10 months, as shown in Figure 1, including cell line development (CLD), platform process fit, formulation screening, Tox and clinical batch production, as well as comparability and characterization studies. The CLD work spanned 6.5 months to sequentially perform stable pool generation, top SCC screening, RCB stability analysis, and MCB generation. Traditionally, the MCB is used in Tox batch manufacturing, supplying a representative DS for Tox studies. This means the Tox study can start only 8 months after the molecule sequence is determined. In this “fast-to-FIH” path, the Tox batch started after a quick platform process check and formulation screening using a stable cell pool, saving approximately 4 months for Tox material production. The DS from the Tox batch was also used for stability studies, drug filling, and lyophilization. Two months were required to perform a comprehensive comparability study on Tox and clinical batch DSs.

Figure 1. Fast-to-FIH timeline and related risk mitigation strategy. A stable cell pool is used for Tox material generation, thus accelerating the development of novel non-COVID therapy in clinical trials. Several risk-mitigation strategies have been implemented to ensure rapid development and comparability between Tox and clinical batches.

A 10-month CMC timeline using the “Fast-to-FIH” approach is shown, including cell line and process development, manufacturing, comparability, and Tox study. Risk mitigation strategies are listed next to the CMC timeline.

In contrast to the 3–6 months CMC timeline using a transient cell or cell pool directly for clinical studies, this mAb program still used the MCB to produce a clinical batch. Thus, the comparability between the Tox and clinical batches was the gating factor. A unique cell-line platform combining a glutamine synthetase variant and a transposon system was used to build a stable cell pool with high homogeneity. This cell line platform ensured sufficient cell line stability to support the production of pilot-scale Tox materials and a high possibility of selecting monoclonal cell lines that are highly consistent with the stable cell pool in terms of process performance, product quality, and efficacy. A transient pool material was used for method development and pre-evaluation of any potential key attributes critical for the product. The platform process and methods were verified over several runs. The SCC, which was most similar to the cell pool, was selected for the production of clinical materials. Limited changes in process parameters were applied to the clinical batch to avoid potential variations. Finally, a comparability evaluation covering release and stability data, side-by-side characterization, and forced degradation was performed on the Tox and clinical materials to demonstrate comparability.

Assessment of fast-to-FIH feasibility

Other than general developability assessment performed on molecular candidates before entering the CMC development stage, the stability and PTM risks were also evaluated to determine the “fast-to-FIH” feasibility. The mAbs produced from the transient cell line were prepared in four different buffers at pH 4.0, 5.8, 6.0, and 7.5 at 10 mg/mL.

The product was freeze-thawed for three cycles, and its purity was measured using SEC-HPLC at each cycle (Figure 2a). The average diameters were analyzed using DLS (data not shown). The monomer content remained at approximately 99%, and there was no significant difference between each freeze-thaw cycle. The product under four pH conditions were stored at 4, 25, and 40°C for 4 weeks. Purity (SEC-HPLC, NR/R-CGE), charge variants (icIEF), and PTMs were analyzed at the beginning (T0), 2- (2w) and 4-week (4w) timepoints. There were no significant changes in product qualities at 4 (data not shown) and 25°C (Figure 2b,c). The SEC monomer of all four pH conditions initially ranged from 99.3% to 99.8% (pH 7.5) and then decreased to 98.9% to 98.0% (pH 7.5) after 4-weeks storage at 25°C. The main peak (MP) remained at approximately 60% with a minor decrease (5%) at pH 7.5 after 2 weeks. The stability data indicate that the molecule should have a low risk during purification and intermediate holding, which usually lasts less than 3 days at room temperature. The slight differences in the measured values at T0 may be due to variations in the methods. No significant PTM changes were observed at pH 4.5 for the following: succinimide at N322 and N391, oxidation at M259, M365, and W52, and deamidation at N22 and N216 (Figure 2d). A 5% increase in oxidation of M259 after 4 weeks at 40°C was observed. This mAb can be considered to have no specific PTM hotspots because the individual values of PTMs were under 7% after stress testing.

Figure 2. Pre-evaluation of drug substance stability and modification risk. The purity (determined using SEC-HPLC) of the DS from the transient cell pool remained stable under four different pH conditions during freeze-thaw cycles (a). No significant decreases in purity (b) and charge variants (c) were established for four pH values under 25°C. The PTM analysis under 4, 25, and 40°C at pH 4.5 demonstrated low risk of modifications during stability (d).

Read the detailed description of this figure

The DS from transient cell pool was freeze-thawed up to 3 cycles in 4 pH conditions (pH 6.0, 4.5, 5.8, and 7.5), and the line graph shows the purity (determined using SEC-HPLC) trends from the 1st to 3rd freeze-thaw cycle. The line graph shows the purity (determined using SEC-HPLC) trends for 4 weeks in 4 pH conditions (pH 6.0, 4.5, 5.8, and 7.5) under 25 ℃. The line graph shows the charge variants (determined using icIEF) trends for 4 weeks in 4 pH conditions (pH 6.0, 4.5, 5.8, and 7.5) under 25 ℃. The stacked column chart shows the PTM trends for 4 weeks at pH 4.5 under 4, 25, and 40 ℃. The modifications include succinimide, oxidation, and deamidation.

Stable pool evaluation and platform process fit

After electroporation, the transfected cells were grown in growth medium. After recovery, the stable pool was evaluated for cell culture performance (VCD and viability) using a 14-day fed-batch assay. The supernatant was further purified using a one-step affinity column and tested for product titer and quality (SEC-HPLC, NR-CGE, glycan profile, and potency).

The stable pool reached 4.1 g/L titer, approximately 98% purity in both SEC-HPLC and NR-CGE tests, 2.7% Man5, 62.8% G0, and 1.2% G0-GN after 14-day fed-batch production (Figure 3a). The titer and quality met the expected levels for producing sufficient Tox materials. Intact and reduced mass spectrometry was used to confirm the primary structure of the product from the stable pool, including deglycosylated intact protein molecular weight and reduced protein molecular weight (Figure 3b). All results were consistent with the proposed structure without mutations. Fluorescence-activated cell sorting analysis (FACS) of surface human Fc expression also revealed a homogeneous stable pool with a majority of high-expressing cells (Figure 3c).

Figure 3. Stable cell pool evaluation and platform fit. The stable pool achieved good productivity and purity (a), with a targeted intact and reduced mass profile (b). FACS analysis revealed the homogeneity of the stable pool (c). Verification runs with platform processes at different scales (3 vs. 50 L) showed consistent culture performance (d) and product quality (e), indicating a robust platform process.

The table summarizes quality data of a stable cell pool, and the product quality attributes include purity (SEC, NR-CGE), charge variants (icIEF), glycans, and titer. Intact and reduced mass profiled of the stable cell pool were plotted indicating the proposed structure was met without mutations. The FACS plot shows a homogeneous stable cell pool with high expression. VCD and viability of a stable cell pool were compared between 3 and 50 L scale fed-batch productions, and the line graph shows good consistency during scale-up. The bar chart shows consistent purity (SEC, NR/R-CGE) and glycan profile (G0, Man5) between 3 and 50 L production scales.

The platform process was confirmed using a stable cell pool at both 3 L and 50 L bioreactor scales. A comparable VCD and viability behavior was observed between the 3 and 50 L verification runs (Figure 3d). The product titers of 3 and 50 L runs reached 4.7 and 4.9 g/L, respectively, and over 98% purity was achieved after purification (Figure 3e). The glycan profiles including G0 and Man5 were consistent during the scale-up from 3 to 50 L. These two verification runs demonstrated that the platform process could fit the mAb product and could be successfully scaled up.

Figure 4. Screening of single-cell clones using the locked process. The 37 single cell clones were analyzed to assess product quality, titer (a), VCD (b, only four SCCs), and metabolic profiles (c, only four SCCs). The top four SCCs were selected for further ADCC testing (d). The data were compared side-by-side using pooled samples. The lead clone (green dot) was selected for the clinical batch because of its best-fit match with the pool (red dot) used for the Tox batch.

Read the detailed description of this figure

Titer, purity (SEC monomer), charge variants (icIEF MP) and glycans (Man5) were analyzed for 37 SCCs and are shown in a scatter plot. VCD and viability were compared between a stable cell pool and 37 SCCs, 4 of which are shown in the line graph with most comparable performance. Metabolism profiles including lactate, glutamate, NH4+, K+, and Na+ were compared between a stable cell pool and 37 SCCs, four of which are shown in the line graph with most comparable performance. The cell sources evaluated in Figures 4b,c and the relevant ADCC results are summarized in a table. The benchmark (stable cell pool) used in the Tox batch and the lead clone selected for the clinical batch are highlighted in bold. The symbols of a stable cell pool and lead clone are red and green dots, respectively.

SCC selection and phenotypical stability

A stable cell pool was recovered and scaled up for single-cell cloning. Plates were seeded with cells in the cloning medium and analyzed for clonality. Wells containing single cells on the first day were selected for scale-up and inoculation in a fed-batch assay using the platform culture process and the selected production media.

Product quality, such as purity (SEC-HPLC), charge variants (icIEF), and glycan profiles, as well as cell growth performance (titer, VCD, viability, and lactate) were evaluated on 37 SCCs and compared with stable cell pools side-by-side (Figure 4a-c). The stable pool achieved a 4.0 g/L titer, 98% monomer content, 59% charge MP, and 3.8% Man5. The product quality of most SCCs closely resembled that of the stable pool, except for a few differences in charge and the N-glycan content. Because quality variations are mainly observed in the charge and N-glycans, these two factors are essential for the final SCC selection. Based on this analysis, four SCCs were selected for further potency assay (ADCC) using the stable pool as the reference. The relative variations (RV) in EC50 were 107%, 105%, 103%, and 93% for SCCs 1–4, respectively (Figure 4d). The comparison of VCD, viability, and cell metabolism (glutamate, lactate, NH⁴⁺, Na⁺, and K⁺) between these four SCCs in the 14-day production also indicated the most similar trends between SCC3 and the stable pool.

The phenotypic stability of the stable pool and the four SCCs (only SCC3 data shown) was studied to assess the cell growth, product titer, and product quality for differently aged cells over a period of 60 PDLs in regular passage after vial thawing. A 14-day fed-batch assay of the stable pool and SCCs was performed at PDL0, PDL30, and PDL60 to evaluate productivity and product quality (Figure 5a,b). The purified supernatants were analyzed using SEC-HPLC, icIEF, NR-CGE, and N-glycan analysis. The titer and product quality attributes of both the pool and SCCs remained similar among the different PDLs, with only a minor charge decrease of 4%. Based on the integral analysis of cell culture performance, product quality, potency, and phenotypic stability, SCC3 was selected as the best match for the stable pool to generate the MCB.

Figure 5. Cell line stability of pool and single clone. The cell line stability within PDL 60 was comparable between the pool and RCB in terms of titer (a) and product quality (b).

Comparability study between tox and clinical batch DS

One 50 L Tox batch was produced using a stable cell pool in a pilot lab, and three 500 L clinical batches were manufactured using the MCB under good manufacturing practice (GMP) conditions. To avoid potential differences caused by process adjustments, limited improved process conditions were introduced in clinical batches only if deemed necessary. The key process attributes, including scale, critical raw materials, process parameters, analytical methods, and formulation, are summarized in Table 1. In addition to scaling up from 50 to 500 L, only the sparger for gas control and the buffer in the formulation were changed. The gas sparger was changed owing to equipment fit, whereas the buffer in the formulation was changed because of the lyophilization performed during the drug product process to ensure long-term stability.

Table 1. Changes in process attributes from Tox to clinical batches.

Attributes		Tox Batch	Clinical Batch
Scale (L)		50	500
Raw materials	Media	No change
	Resin
	Filter
Process Parameter	Feeding strategy	No change
	Gas control	0.8 mm sparger	1.0 mm sparger
	Capacity	No change
	Flow rate	No change
	pH/conductivity	No change
Formulation	Buffer	Acetic acid	HCl
	Stabilizer/surfactant	No change
	Concentration	No change
Analytical Method		No change

A comprehensive list of quality attributes was covered in the comparability study. Release tests were performed on these batches per the specifications (Table 2). A side-by-side characterization study of key physicochemical, structural, and biological functions was conducted to evaluate the pharmaceutical comparability of the Tox and clinical DS batches (Table 3). The release test results of the Tox and clinical DS were similar within the determined DS specifications. Minor differences in the glycan content and the proportion of charge variants were observed. However, according to the ADCC test results, the presence of minor charge variants and glycan differences between Tox and clinical DS had no negative impact on the product efficacy.

Table 2. Comparison of product quality attributes between Tox and two clinical batches.

Product Quality Attributes		Tox Batch	Clinical Batch 1	Clinical Batch 2
SEC	MP%	97.9	98.9	98.9
	LMW%	0.9	0.6	0.5
	HMW%	1.2	0.5	0.6
NR-CGE	MP%	96.9	97.7	97.6
R-CGE	HC+LC%	97.3	98.0	97.9
icIEF	MP%	58.0	65.6	66.1
	APG%	41.1	33.2	32.6
	BPG%	0.9	1.2	1.3
Glycans	G0-GN%	1.3	1.4	1.5
	G0%	58.9	72.1	68.8
	G1%	24.8	18.1	19.9
	G2%	4.4	2.7	3.1
	Man5%	2.8	1.5	1.9
ADCC	RV%	91	115	112

Table 3. Side-by-side characterization of Tox and clinical DS.

Product Quality Attributes			Tox Batch	Clinical Batch
Primary structure	Molecular weight of intact protein G0/G0 (Da)		149981	149980
	Sequence coverage (%)		100	100
	Reduced protein mass (Da)		LC (glycation variant): 24484; HC (G0): 50683	LC (glycation variant): 24484; HC (G0): 50683
Higher-order structure	CD		Far-UV: 217.0–201.5 nm Near-UV: 261.0–293.0 nm	Far-UV: 2176.0–201.5 nm Near-UV: 261.5–292.0 nm
	DSC		Tm1: 73.4°C; Tm2: 81.8°C	Tm1: 73.6°C; Tm2: 82.1°C
	DLS		Z-average diameter: 6.4 nm Average PDI: 0.09	Z-average diameter: 5.8 nm Average PDI: 0.11
	AUC		Comparable
Molecular microheterogeneity	Free thiols (mol/mol)		0.18	0.18
Post-translational modification	C-Terminal lysine (%)		98.5	98.5
	N-terminal pyroglutamate (%)		99.9	100.0
	Oxidation (%)	M365	0.7	0.6
		W52	0.1	0.2
		M259	2.7	2.0
		M4	0.4	0.5
		W56	0.5	0.5
	Deamidation (%)		<0.1	<0.1
	Succinimide (%)	N322	0.9	0.9
		N391	0.3	0.3
	Isomerization (%)		ND	ND
Biological function	FcRn-binding by SPR (RV%)		99	102

Meanwhile, the evaluation of ADCC activity of Tox DS with a high galactose ratio (G1%, G2%) treated by galactosidase (β-gA) demonstrated that minor differences in galactose ratio had no effect on ADCC activity (Table S1). A previous study on the N-glycosylation of IgG and IgG-like recombinant therapeutic proteins also indicated that this minor difference was noncritical, with no potential effect on product efficacy or safety.¹⁷ In a side-by-side characterization study, Tox and clinical DSs showed consistent primary and higher-order structural characteristics, physicochemical properties, and PTMs. In addition, Tox and clinical DS displayed constant bioactivity in functional biological assays (FcRn binding characterization using SPR).

Forced degradation study at high temperature (40°C) was also performed as a part of the comparability study for a better comparison of degradation products and degradation trends within a short time. SEC-HPLC, NR/R-CGE, icIEF, ADCC, and FcRn binding were used for the analyses at T0 and 2, 4, and 8 weeks timepoint. All test items met the specifications of the stability study, and there was no difference in the stability trend between Tox and clinical DS which increased by 3.5% and 1.1%, respectively, over 8 weeks in the LMW and HMW groups (Figure 6a). An increase of approximately 39% in the acidic peak group (APG) was observed for both Tox and clinical DS (Figure 6b). Similar declines in ADCC activity and FcRn binding affinity were also observed in both Tox and clinical DS (Figure 6c,d). Long-term and accelerated stability was also evaluated for Tox and clinical DS and the current data indicated that Tox and clinical DS had a comparable long-term stability trend over a 6-month period (Figure 6e,f). The PTMs of both Tox and clinical DSs also showed a highly comparable trend over an 8-week period.

Figure 6. Forced degradation and long-term stability assessments. The stability trends of purity (a), charge variants (b), ADCC (c), and FcRn binding (d) under the forced degradation condition (40°C) and purity (e) and charge variants (f) under the long-term (−40°C) condition are comparable between the DS from tox batch and clinical batch.

Read the detailed description of this figure

The stacked column chart shows the DS trends of purity (SEC monomer, HMW, and LMW) of Tox and clinical batches over 8 weeks at 40°C. The stacked column chart shows the DS trends of charge variants (icIEF MP, BPG, and APG) of Tox and clinical batches over 8 weeks at 40°C. The line graph shows the DS trends of ADCC of Tox and clinical batches over 8 weeks at 40°C. The line graph shows the DS trends of FcRn binding of Tox and clinical batches over 8 weeks at 40°C. The stacked column chart shows the DS trends of purity (SEC monomer, HMW, and LMW) of Tox and clinical batches over 6 months at −40°C. The stacked column chart shows the DS trends of charge variants (icIEF MP, BPG, and APG) of Tox and clinical batches over 6 months at −40°C. In all assays, similar DS stability trends were observed between the Tox and clinical batches under forced degradation and long-term conditions.

In summary, the executed pharmaceutical comparability study and assessment, including analytical results and evaluations of the (1) manufacturing process; (2) release test results; (3) trends of long-term, accelerated, and stressed stability data; (4) structure, physicochemical attributes, and biological activities; and (5) degradation characteristics and trends under forced degradation conditions, support the conclusion of the overall comparability between Tox and clinical DS.

Discussion

The use of a stable cell pool to produce Tox materials is a proven way to significantly accelerate the transformation of DNA to FIH. Using this novel idea, the CMC workflow can be significantly shortened to 12–18 months. During the COVID pandemic outbreaks, more CMC strategies have been explored to further accelerate the entire CMC timeline to 3–6 months for COVID-19 drugs.¹⁸ In this paper, we report a new non-COVID drug case using a stable pool for the Tox batch and a lead SCC for clinical use. Several CMC acceleration strategies used in previous COVID cases were also used in this “fast-to-FIH” approach including: (1) selecting the lead clone based on a pre-PDL30 stability; (2) using simple in-house cell bank biosafety tests on the RCB instead of full biosafety tests to ensure GMP-compliant site entry; (3) using the platform process for both Tox and clinical batches with limited changes to save process optimization time. The entire CMC package was completed within 10 months based on BeiGene’s robust integral AF cell line, process, and analytical platform.

The improved cell pool productivity and quality ensured that sufficient materials were generated in the small-scale Tox batch. The Tox batch titer was 4.7 g/L, and the DS purity reached 98% as shown using SEC-HPLC. Comparability between Tox and clinical batches was critical to leverage the data from the Tox study in the clinical setting. BeiGene’s optimized cell line supports the generation of a homogeneous cell pool with similar performance and product quality as those of SCCs. The comparison between the stable pool and selected SCC in terms of culture productivity, metabolic activity, quality attributes, and cell line stability demonstrated high consistency. The comparability between the final 50 L Tox and 500 L clinical batches was comprehensively evaluated through release test results, stability data, structural and physicochemical attributes, biological activity, and forced degradation characteristics. Only minor differences were observed in the glycan and charge profiles, which may be due to variations in scale and gas control.^19,20 The clinical batch had higher purity than the Tox batch, which had no impact on molecular efficacy.

Significant time savings during CMC development have been achieved by adopting risk-based approaches, as exemplified by COVID-19 drug development. WuXi biologics achieved shorter IND timelines by submitting without the following: cell line stability data, a complete Tox study, a viral clearance study, long-term stability data, and a comparability study,¹¹ which added pressure regarding biosafety concerns and moved additional work to the post-IND phase. Considering a potential higher requirement of non-COVID drugs, to further lower the risk, BeiGene applied its own “fast-to-FIH” strategies. The data in this report show that BeiGene could accomplish a 10-month timeline (already the fastest for non-COVID drugs) incorporating cell line stability, complete viral clearance and Tox study, release and stability data, and comprehensive comparability study. Pre-evaluation of drug candidates was performed to gain an in-depth understanding of the molecules, including thermal stability, modification hotspots, and strength, and several runs at different scales verified a robust process and analysis platform. This pre-evaluation is limited by the time and amount of material provided by product transient expression. Thus, the transient expression system needs to be in place to provide enough protein at the earliest possible time point.

In the future, higher product quality, better comparability, and faster CMC timelines can be achieved by optimizing site-specific integration techniques,^21,22 eliminating verification runs, and simplifying comparability studies. Currently, most “fast-to-FIH” strategies are performed on standard (e.g., full-length, monospecific) mAbs owing to better understanding and control on purity. However, with deeper investigations on control of impurities and high-resolution analytical methods,^23–25 opportunities to optimize bispecific antibody or antibody-drug conjugate development with this “fast-to-FIH” approach are possible.

For further widespread application, selecting a molecule appropriate for the “fast-to-FIH” approach is the first hurdle to success. Several factors must be considered when molecules become increasingly complex. First, a molecule with significant toxicity concern is not recommended for this fast approach because it is difficult to justify whether any minor change between the Tox and clinical batches would lead to sensitive Tox issues.²⁶ Product quality is another critical issue to decide whether a candidate is feasible for “fast-to-FIH”. Although in the preclinical stage, the product critical quality attributes (CQA) are always unclear, some typical attributes that might affect efficacy and safety, such as aggregates, fragments, charge heterogeneity, and binding activity, need to be considered. General quality analyses, including purity, charge, glycan profiles, identity, potency, process performance (titer, viability, purification yield, and so on), process-related impurities, and stability, are essential for evaluating the behavior of molecules. As the first case trying the “fast-to-FIH” strategy, in addition to general release test items, characterization items covering structural, molecular and bioactivity contents were all checked in pre-evaluation to ensure a well-behaved candidate was allowed in this approach. In future applications, candidates should be reconsidered if there are any mechanisms of action affecting PTM hotspots or special quality liabilities. To what extent the minor difference between the cell source and process will affect these quality liabilities is key to the “fast-to-FIH” success rate. Finally, regulatory positions are very important, especially for new strategies; thus, close communication with regulatory agencies is necessary for bringing novel biologic drugs to clinic through faster and safer CMC approaches.

Abbreviations

AF	=	Afucosylated
ADCC	=	Antibody-dependent cell-mediated cytotoxicity
APG	=	Acidic peak group
BPG	=	Basic peak group
CLD	=	Cell line development
CMC	=	Chemistry manufacturing and control
CHO	=	Chinese hamster ovary
DS	=	Drug substrate
FIH	=	First-in-human
FACS	=	Fluorescent-activated cell sorting
GMP	=	Good manufacturing practice
HMW	=	High molecular weight
HPLC	=	High-performance liquid chromatography
icIEF	=	Imaged capillary isoelectric focusing
IND	=	Investigational new drug
LMW	=	Low molecular weight
MCB	=	Master cell bank
MP	=	Main peak
NR/R-CGE	=	Non-reduced and reduced capillary gel electrophoresis
PDL	=	Population doubling level
PTM	=	Post-translational modification
RV	=	Relative variations
RCB	=	Research cell bank
SCC	=	Single cell clone
SEC-HPLC	=	Size exclusion chromatography-HPLC
SPR	=	Surface plasmon resonance
Tox	=	Toxicology
VCD	=	Viable cell density
ZFN	=	Zinc finger nuclease

Author contributions

The manuscript was written by Yuqi Wang. All the authors approved the final version of the manuscript.

Acknowledgments

The authors thank Xuening Li, Xin Pei, Xiaoqing Jin, Yu Ji, and Leo Chan for supporting technology transfer; Oliver Shen and Richard Lu for manufacturing support; Jinsong Feng and Xiaojing Chu for quality assurance; Wenjing Shi, Bo Zhou, Meng Jiang, Chunlong Li, Gaofeng Chen, Yaomin Li, and Shiqi Xu for DP manufacturing support; and Jun-Hsiang Lin for leading the Tox study.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/19420862.2023.2292305

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.