2.2 Methods
Statistical comparisons of adalimumab-aqvh with the reference product were made for quantitative results to establish whether acceptance criteria were met. Statistical analyses were carried out using two approaches based on risk ranking of the product quality attributes. The most critical attributes with highest risk to clinical outcome were compared using equivalence tests based on standard deviation and confidence intervals derived from adalimumab lots [
10,
11]. Equivalence is shown if the 90% two-sided confidence interval of the difference between means for adalimumab-aqvh and adalimumab is within the equivalence acceptance criterion of ± 1.5
σ based on the standard deviation (
σ) calculated from tested adalimumab lots. The remaining quantitative quality attributes were compared using a quality range. The quality range is defined as the mean ±
k × standard deviation of adalimumab results, with
k = 3. Similarity is shown if at least 90% of adalimumab-aqvh results are within the one-sided (mean ± 3 standard deviation) or two-sided (mean ± 3 standard deviations) limits, as appropriate. The value of
k is chosen as 3 consistent with a normal distribution where 99.73% of the population is included within three standard deviations of the population mean.
The methods used for the comparative analytical assessment are described in the following sections. These methods were qualified and shown to be fit for purpose (data not shown). The adalimumab-aqvh primary reference standard lot was used as the reference standard for all functional testing for which relative activity was used to assess similarity. Adalimumab-aqvh samples were analyzed together with reference product samples wherever possible.
2.2.1 Reduced Peptide Mapping
Reduced peptide mapping with trypsin digestion and liquid chromatography with tandem mass spectrometry (LC–MS/MS ) was performed on a ThermoFisher Scientific Orbitrap Q Exactive coupled to an ultra performance liquid chromatography (UPLC) system. Peptides were separated on a ThermoFisher Scientific Acclaim Vanquish C18 column (2.2 µm, 2.1 mm × 250 mm) at 25 °C using a gradient of acetonitrile/water containing formic acid, 0.1%, at a flow rate of 0.3 mL with UV absorbance detection at 215 nm. MS was performed in positive mode at an Orbitrap resolution of 60,000 over a range of 200–2000 mass to charge ratio (m/z). MS/MS was achieved by 28% higher-energy collisional dissociation energy and detected by Orbitrap by auto-scan range. Data were processed using ThermoFisher Scientific BioPharma Finder software, version 3.2, to identify peptide fragments.
2.2.2 Intact Mass
Intact, reduced, and deglycosylated protein samples were injected onto a Waters Mass PREP Micro desalting column (2.1 × 5 mm, 20 μm particle). A gradient of acetonitrile/water containing formic acid, 0.1%, was used. Mass spectra were acquired in positive ion polarity mode using an electrospray ion source kept at 125 °C (desolvation temperature of 450 °C); capillary voltage was set at 2.5 kV.
2.2.3 Non-Reduced Peptide Mapping
For the non-reduced peptide mapping, samples were treated with trypsin under non-reducing conditions. LC–MS/MS was performed using a ThermoFisher Scientific Orbitrap Fusion Lumos Tribrid high-resolution mass spectrometer coupled to a UPLC system. The disulfide-linked peptides were separated using a ThermoFisher Scientific Acclaim Vanquish C18 UPLC reversed-phase column (2.1 × 250 mm, 2.2 μm particle) and a gradient of acetonitrile/water containing formic acid, 0.1%, at a flow rate of 0.3 mL with UV absorbance detection at 215 nm. MS was performed from m/z 197 to 2000 Da under the positive polarity mode by Orbitrap resolution of 60,000. MS/MS was performed by 28% higher-energy collisional dissociation energy and detected by Ion Trap from m/z 120. Data were processed using ThermoFisher Scientific BioPharma Finder software, version 3.2, to identify peptide fragments.
2.2.4 FTIR Spectroscopy
An FTIR spectrometer equipped with a room temperature deuterated triglycine sulfate detector/liquid nitrogen cooled mercury cadmium telluride detector was used. Protein samples were concentrated to ~ 90 mg/mL and placed between two CaF2 windows with a 6- to 7-µm spacer. Spectra were collected at 4 cm−1 resolution, with a data average of 256 scans. A buffer (filtrate from concentration of protein samples) spectrum and residual moisture peaks were subtracted. Second-derivative spectra were calculated using the Savitzky–Golay method, with a second order of polynomial function.
2.2.5 Far and Near UV CD
Samples were diluted to 1 mg/mL in a common formulation buffer for far and near UV circular dichroism (CD) analysis using a CD spectrophotometer. Measurements were carried out at room temperature using 1 cm/cell and 0.02 cm/cell for near and far UV CD, respectively. After subtracting the buffer spectrum, the CD spectrum of each sample was converted to the mean residue ellipticity using the mean residue molecular weight of 109.35 and the path length of the cell.
2.2.6 Intrinsic Fluorescence
Each sample was diluted to 0.1–0.3 mg/mL in a common formulation buffer. Excitation was at 280 nm, and intrinsic fluorescence data was collected from 280 to 450 nm using a fluorescence spectrophotometer. The excitation and emission slits were both at 5 nm, and the scan rate was 300 nm/min.
2.2.7 Differential Scanning Calorimetry
Differential scanning calorimetry analysis was conducted using a Malvern VP-Capillary DSC. The samples were diluted to 0.5 mg/mL in a common formulation buffer, loaded, and scanned from 10–95 °C at 1 °C/min with a 10-second data averaging period or 10–110 °C at 1 °C/min with an 8 second data averaging period with the formulation buffer in the reference cell. Data analysis was done using Origin software, version 7.0 (OriginLab). The heat capacity profiles were normalized to protein concentration.
2.2.8 Glycan Analysis
The protein was first bound to a protein A cartridge, washed, and digested with N-glycanase. The released glycans were centrifuged and labeled with 2-aminobenzamide (InstantAB; Prozyme). The labeled N-glycans were analyzed by hydrophilic interaction chromatography-high performance liquid Chromatography (HPLC) on a Waters X-bridge Amide 3.5 µm 2.1 × 150 mm column at 45 °C and a flow rate of 0.5 mL/min with fluorescence detection. A trifluoroacetic acid/ acetonitrile gradient was used in the mobile phase.
2.2.9 Size-Exclusion Chromatography with Multi-Angle Light Scattering
Size-exclusion chromatography with multi-angle light scattering (SEC–MALS) analysis was done using an HPLC system with a multi-angle light scattering and a refractive index detector and a Sepax SRT SEC-300 7.8 × 300 mm, 5-µm particle column at 25 °C and a flow rate of 0.7 mL/min. The running buffer was sodium phosphate 50 mM, sodium chloride 250 mM (pH 6.8). The protein load was 100–200 µg, and the molecular weight of the species was determined using a refractive index increment value of 0.185 mL/g.
2.2.10 Size-Exclusion–Ultra-Performance Liquid Chromatography
Ten micrograms of sample were injected using a UPLC system and a Waters Acquity BEH-200 SEC, 1.7 µm, 4.6 × 300 mm column running at a flow rate of 0.1 mL/min and a column temperature of 25 °C. The mobile phase was sodium phosphate 100 mM, sodium sulfate 150 mM, and 1-propanol, 5% (pH 6.3).
2.2.11 Reduced and Non-Reduced Capillary Electrophoresis in Sodium Dodecyl Sulfate
Samples were analyzed using a Beckman PA800 Plus CE instrument with UV absorbance detection. Samples at 2 mg/mL were denatured under heat with sodium dodecyl sulfate (SDS). Reducing conditions were achieved by including β-mercaptoethanol, ~ 5%. Under non-reducing conditions, ~ 12 mM iodoacetamide was included to block disulfide interchange. The denatured proteins were separated electrophoretically through a bare fused-silica capillary with a 20 cm effective length and a photodiode array detector set to 220 nm.
2.2.12 Sedimentation Velocity Analytical Ultracentrifugation
Sedimentation velocity analytical ultracentrifugation analysis was carried out at 20 °C, 40,000 rpm for 6 h or 35,000 rpm for 7.3 h monitored at 280 nm using an analytical ultracentrifuge. The samples were diluted to 0.5 mg/mL in a common formulation buffer and loaded into cells with two-channel charcoal-epon centerpieces with a 12-mm optical path length. The dilution buffer was run in the reference channel. Data analysis was done using the NIH SEDFIT program (version 11.3/version 16.2b) to derive the distribution of sedimentation coefficient. The fluorescence intensity ratios and meniscus position were fitted to find the best overall fit of the data for each sample. A maximum entropy regularization probability of 0.683 (1σ) was used, and time-independent noise was removed. The size distribution peaks were integrated using the OriginLab Origin program, version 9.0.0.
2.2.13 Cation-Exchange Chromatography
Cation-exchange chromatography was carried out on samples before and/or after carboxypeptidase B (CPB) treatment to eliminate the influence of variable processing of the C-terminal lysine and to allow for better comparison of stability-indicating charge variants. Proteins were separated using an HPLC system and a SCX-NP5 column (4.6 × 250 mm, 5 µm), a pH gradient, and a column temperature of 30 °C at a flow rate of 0.8 mL/min. At 5 mg/mL, 20 µL of each sample was injected and detected at 280 nm. Resolved peaks were integrated and reported as acidic, main, and basic peak groups.
2.2.14 Imaged Capillary Isoelectric Focusing
Samples were prepared at 0.3 mg/mL in urea 3.6 M; methylcellulose, 0.35%; and Pharmalyte 8-10.5, 4% (Millipore Sigma), and spiked with bracketing isoelectric point (pI) markers. The pI of the main peak was determined by linear regression between the two included pI markers.
2.2.15 Reversed-Phase Ultra-Performance Liquid Chromatography
Separation was achieved using a 2.7 µm Halo column (2.1 × 150 mm); a UPLC system; a trifluoroacetic acid, water, and acetonitrile gradient; and a flow rate of 0.25 mL/min at 75 °C. Absorbance was detected at 215 nm. Samples were treated with IdeS protease, which specifically targets IgG to generate F(ab)2 and Fc fragments, and with CPB to remove heavy chain C-terminal lysine.
2.2.16 Post Translational Modifications
For quantification of post-translational modifications, reduced peptide mapping data were processed using ThermoFisher Scientific BioPharmaFinder software, version 3.2, for which unmodified and modified peptides were identified using MS/MS. The MS peak area of each m/z for the peptide containing the attribute of interest was calculated by use of the software. The percentage of modification was obtained by dividing the peak area of the peptide with the modification over the total peak area of both the modified and the unmodified peptides.
2.2.17 Free Thiol
Levels of free thiol were determined using 5,5′-dithiobis-(2-nitrobenzoic acid) (DTNB; Ellman’s reagent). Samples were mixed with a guanidine hydrochloride solution containing ethylenediaminetetraacetic acid 2 mM and DTNB 0.1 mM (pH 8.0). Released 2-nitro-5-thiobenzoic acid, which is proportional to the free thiols in the protein, was quantified by absorbance at 412 nm using a UV-vis spectrophotometer.
2.2.18 Host-Cell Protein
Host-cell protein content was analyzed using a commercial enzyme-linked immunosorbent assays (ELISA) kit according to the manufacturer’s instructions (CHO protein ELISA 3G kit #F550, Cygnus Technologies, Inc.).
2.2.19 Protein Concentration
Protein concentrations were determined using absorbance at 280 nm and measured using a SoloVPE variable pathlength spectrophotometer. A minimum of 20 µL of sample was loaded directly without dilution and analyzed using an experimentally determined extinction coefficient of 1.45 mg/mL·cm.
2.2.20 Subvisible Particle Analysis
Microflow imaging was carried out using a ProteinSimple MFI 5200 instrument. Drug product samples were expelled from syringes into 5-mL clean glass vials and degassed under light vacuum before analysis. A morphologic filter was applied to report particles with an aspect ratio < 0.85 to reduce interference from silicone oil droplets.
2.2.21 Soluble Tumor Necrosis Factor α-Binding Enzyme-Linked Immunosorbent Assay
A qualified sTNF-α–indirect ELISA was used to compare binding of adalimumab-aqvh and adalimumab. Plates were coated with TNF-α, followed by addition of a dilution series of adalimumab. Anti-human IgG conjugated to horseradish peroxidase was then added to detect bound adalimumab. Plates were washed, followed by the addition of substrate (tetramethylbenzidine) to the plates. Stop solution (sulfuric acid 1 M) was added, followed by data collection. Curve fit parameters were analyzed using SoftMax Pro (Molecular Devices). Relative binding was calculated against the primary reference standard adalimumab-aqvh. The adalimumab sTNF-α ELISA design enabled evaluation of the mean relative binding result from three independently prepared assay plates for each test material.
2.2.22 Soluble Tumor Necrosis Factor α Binding Affinity by Surface Plasmon Resonance
The binding affinity was characterized by surface plasmon resonance (SPR) using a Biacore T200 SPR instrument. A biosensor chip pre-immobilized with protein A was used to capture the sample via the Fc portion of the molecules on flow cells 2, 3, and 4, with flow cell 1 used as the reference. The running buffer was 0.01 M HEPES pH 7.4, 0.15 M NaCl, 3 mM EDTA, 0.005% v/v Polysorbate 20 (HBS-EP+). Soluble TNF-α at 0.30–9.53 nM, along with buffer blanks, were injected over the surface of the protein A-bound adalimumab. The sensorgrams were fit to a 1:1 kinetic interaction model using the Biacore T200 evaluation software (version 3.2). The affinity was calculated by taking the ratio of kd/ka.
2.2.23 Apoptosis Assay
Soluble TNF-α neutralization potency was determined using a validated cell-based assay that is used to measure the ability of adalimumab to bind to TNF-α and inhibit sTNF-α–induced apoptosis in the TNF-α-sensitive human monocytic cell line. Apoptosis was detected and quantified by measuring caspase activation using a Caspase-Glo 3/7 luciferase kit (Promega), and the relative potency was measured against the adalimumab-aqvh primary reference standard.
2.2.24 Antibody-Dependent Cell-Mediated Cytotoxicity Activity
Cells overexpressing mTNF-α were seeded on plates. To analyze samples, mTNF-α target cells were co-incubated with peripheral blood mononuclear cell (PBMC) effector cells at a ratio of 1:50. Samples were added at various dilutions and incubated for 18–22 h. Cell cytotoxicity was measured by adding CytoTox-Glo reagent (Promega) 30 min before reading the luminescence. Samples were tested in triplicate on three independently prepared assay plates. The ADCC dose-response data were imported into PLA 2.0 software and were analyzed for parallelism, regression, and linearity of the test material response compared with the reference standard. The mean relative potency [half-maximal effective concentration (EC50)] obtained from the dose-response curve and mean relative cell death were calculated.
2.2.25 Complement-Dependent Cytotoxicity Activity
The reduction in viability was measured using mTNF-α-overexpressing cells to determine CDC activities in the presence of rabbit complement. The reduction in cell viability was detected by CellTiter-Glo (Promega) as an assay readout. Samples were tested in triplicate on two independently prepared assay plates. The relative potency was analyzed using PLA 2.0 software by comparing the EC50 from the dose-response curve of the reference standard and test samples using a restricted fit and four-parameter logistic curve model.
2.2.26 Cell-Based Membrane-Associated Tumor Necrosis Factor α Binding by Meso Scale Discovery
A qualified meso-scale discovery (MSD) electrochemiluminescence assay was used to compare the binding of adalimumab-aqvh and adalimumab in CHO cells engineered to express TNF-α on the cell surface. Cells were plated, followed by blocking before the addition of reference standard and test materials. After 1 hour of incubation, MSD-conjugated detection reagent and then read buffer were added. Plates were read on the MSD Sector 2400 reader (Meso Scale Diagnostics). Data analysis was performed using SoftMax Pro and PLA 2.0 (Stegmann Systems GmbH) software. Relative mTNF-α binding was calculated against the adalimumab-aqvh primary reference standard.
2.2.27 Mixed Lymphocyte Reaction Assay
Mixed lymphocyte reaction assays were performed using PBMCs from patients with Crohn’s disease as either stimulator or responder cells with a subset of adalimumab and adalimumab-aqvh lots. PBMCs from responders were diluted to 2 × 106 cells/mL and added at 50 μL/well. The stimulators from PBMCs from healthy individuals or patients with Crohn’s disease or the monocytes were irradiated at 3000 Rad, diluted in medium to 2 × 106 cells/mL or 1 × 106 cells/mL (depending on the responder:stimulator ratio of 1:2 or 1:1), and added at 50 μL/well. After incubation for 1 h, diluted test samples were added at 50 μL/well (40 μg/mL) and incubated for 6 days. Experiments were performed in 96-well round-bottom plates in quadruplicate. Mean values for fluorescence were used to derive the percentage of inhibition of proliferation activity. For flow cytometric analysis of mTNF-α expression and macrophage phenotype, MLR reactions were set up as just described but scaled up to 12-well plates with 8 × 105 responders/well and 8 × 105 stimulators/well. After incubation for 7 days, cells were transferred into 96-well round-bottom plates for staining.
2.2.28 C1q Binding by Enzyme-Linked Immunosorbent Assay
A validated C1q direct ELISA was used to evaluate the relative binding of adalimumab-aqvh and adalimumab. Plates were coated with adalimumab dilution series and controls. After incubation, washing, and blocking, C1q was added to plates for incubation at room temperature. Plates were washed and substrate (tetramethylbenzidine) was added to the plates. Stop solution (sulfuric acid 1 M) was added and data were collected. Curve fit parameters were analyzed using SoftMax Pro. Relative binding was calculated against the primary reference standard.
2.2.29 Surface Plasmon Resonance Binding Affinity Assays for Fcγ Receptors
The affinities of Fcγ receptors (FcγRs) were measured on a Biacore 8K instrument with HBS-EP+ running buffer at pH 7.4. The method consisted of amine coupling of sTNF-α as a capture reagent for adalimumab-aqvh on all flow cells of all channels, with one flow cell being used as a reference. After capture of the reference standard or adalimumab-aqvh on flow cell 2, FcγRIIIa, FcγRIIIb, FcγRIIa, FcγRIIbc, and FcγRI were injected at 15 °C over all channels. Binding affinities were determined using either a steady state model or a fit to a 1:1 kinetic interaction model (FcγRI only).
2.2.30 Surface Plasmon Resonance Binding Affinity Assay for Neonatal Fc Receptor
The binding of FcRn to adalimumab-aqvh was determined using a Biacore T200 instrument with a running buffer of sodium phosphate 10 mM; sodium chloride 150 mM; Tween 20, 0.05%; and glycerol, 20% (pH 5.8). The method consisted of amine coupling of neutravidin in sodium acetate buffer 10 mM (pH 5.0) as a capture reagent for biotinylated FcRn on flow cells 2, 3, and 4, with flow cell 1 being the reference flow cell. After capture of biotinylated FcRn, adalimumab or reference standard was injected at 15 °C over all flow cells, and binding affinities (KD) were determined using a 1:1 kinetic interaction model. The FcRn surface was regenerated using a sodium phosphate buffer (pH 7.4).
2.2.31 Forced Degradation Under Heat Stress
Adalimumab-aqvh and adalimumab samples were subjected to dialysis into adalimumab formulation buffer to assess degradation independent of formulation differences and to equalize any dialysis effects. Samples were heated at 40 °C for up to 3 months. Several stability-indicating analytical methods [e.g., SE-UPLC and cation-exchange chromatography (CEX), same as described previously] were used to provide a comprehensive assessment of potential degradation pathways.