Determination of Anti-drug Antibody Affinity in Clinical Study Samples Provides a Tool for Evaluation of Immune Response Maturation

Reagents (Singulex Erenna Assay)

PF-06480605 (unlabeled, biotinylated, and Alexa Fluor 647–labeled), drug-coated paramagnetic beads, and mouse anti-idiotype monoclonal antibody against PF-06480605 as PC, were all developed by Pfizer (Andover, MA). The following buffers were all supplied by Pfizer: storage buffer (phosphate-buffered saline, calcium, and magnesium–free (PBS-CMF) with 0.05% Tween®20 and 1% bovine serum albumin, pH 7.2 (PBST/1% BSA), Tris High Salt Wash Buffer (THST) (50 mM Tris, 1 mM glycine, 500 mM NaCl, 0.05% Tween®20), acid dissociation buffer (100 mM glycine, pH 2.0), and neutralization buffer (1 M Tris, pH 9.0). Pooled normal human serum was obtained from BioIVT (Westbury, NY). SMC™ Capture Antibody Labeling Kit (cat #03–0077-02) and SMC™ Bead–Based Assay Development Kit (cat #03–0178-00) were purchased from Millipore Sigma (Burlington, MA). A total of 384-well polypropylene microtiter plates (cat #264573) and Axygen™ polypropylene V-bottom 96-well microplates (cat #P-96-450 V-C) were both purchased from Thermo Fisher Scientific (Waltham, MA). The instrument used for ligand-binding affinity assay analysis in patient samples was the Erenna® Immunoassay System using SMC™ technology from Millipore Sigma (Burlington, MA).

Reagents (Gyrolab Assay)

PF-06480605 (unlabeled, biotinylated, and Alexa Fluor 647–labeled), drug-coated paramagnetic beads, and mouse anti-idiotype monoclonal antibody against PF-06480605 as PC, were all developed by Pfizer (Andover, MA). The following buffers were all supplied by Pfizer: storage buffer (phosphate-buffered saline, calcium, and magnesium–free (PBS) with 0.05% Tween®20 and 1% bovine serum albumin, pH 7.2 (PBST/1% BSA). Tris High Salt Wash Buffer (THST) (50 mM Tris, 1 mM glycine, 500 mM NaCl, 0.05% Tween®20), acid dissociation buffer (100 mM glycine, pH 2.0), and neutralization buffer (1 M Tris, pH 9.0). Pooled normal human serum was obtained from BioIVT (Westbury, NY). Wash station 1 buffer (PBS + 0.01% Tween®20) was prepared by Pfizer. Wash station 2 buffer (Gyrolab wash buffer pH 11, cat #P0020096), Gyrolab Bioaffy 1000 nanoliter CDs (cat #P0004253), Rexxip F buffer (cat #P0004825), 96-well PCR plates (cat #P004861), and microplate foil seals (cat #P003313) were all purchased from Gyros Protein Technologies (Warren, NJ). The instrument used for ADA assay development was the Gyrolab xPand with Gyrolab affinity software module (cat# P0020413).

Clinical Samples for Analysis

Samples were obtained from a subset of patients participating in the TUSCANY trial, a phase 2a study which evaluated safety, tolerability, efficacy, PK, and immunogenicity of PF-06480605 in patients with moderate to severe UC (10). Samples selected for affinity analysis were limited, based on availability of sufficient sample volumes and demonstration of PF-06480605 ADA log(10) titer ≥ 2.0. Using this criterion, a total of 7 patients had evaluable samples. Samples generally spanned multiple time points ranging from 15 to 183 days after drug administration, thereby enabling characterization of KD over time. Three patients had available samples spanning 5–7 time points within this timeframe, and one patient had available samples from 3 time points over a more limited timeframe (141–183 days). Three additional patient samples from 1 or 2 time points each were included in measurements.

Bead Labeling with PF-06480605 for ACE Procedure and ADA LBA Capture Reagent

PF-06480605-coupled beads for the ACE procedure were produced at Pfizer (Andover, MA). Briefly, sulfo N-hydroxysuccinimide (NHS) activated carboxyl-coated paramagnetic beads (2 × 109 beads/mL) were directly conjugated with 0.45 mg PF-06480605 using EDC (1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride) chemistry through primary amines. For the Singulex Erenna ADA LBA, biotinylated PF-06480605 was coated on streptavidin paramagnetic beads using the SMC™ Capture Antibody Labeling Kit at 6.25 µg IgG/mg beads. The bead solution was incubated on an inversion rotator for 60 min at room temperature. Coated beads were washed 5 times, and beads stored in kit coating bead buffer at 2–8°C until use.

Assay Procedure

The ADA affinity assay procedure is comprised of 3 steps as listed below and depicted in Fig. 1. Steps 1 and 2 are common to both Singulex Erenna and Gyrolab ADA assays (sample pre-treatment steps). Step 3 has LBA platform-specific assay procedure descriptions. Both platforms use the same biotinylated and Alexa Fluor 647–labeled PF-06480605 as capture and detection reagents, respectively, and the same MAb PC.

Fig. 1figure 1

ADA affinity assay workflow. (A) An affinity capture elution (ACE) method was used to enrich ADA from serum matrix and improve drug tolerance. (B) Enriched ADA at one dilution or PC was incubated with a range of PF-06480605 concentrations to reach equilibrium. (C) Solution-phase equilibrium samples were assayed using a bridging ADA LBA to assess free ADA levels against PF-06480605. Data were subsequently analyzed using a customized R Shiny application for apparent KD determination

Step 1: ACE Procedure for ADA and PC Enrichment

The PC was prepared by spiking an idiotype specific anti-PF-06480605 mouse MAb into neat (100%) pooled normal human serum for a final concentration of 160 µg/mL and incubating for 30 min at room temperature (RT) on a shaker. The PC preparation was tested with and without PF-06480605 (25 µg/mL). The ACE procedure was performed on the PC samples and clinical study samples to enrich ADA concentrations and improve drug tolerance by reducing residual PF-06480605. One milliliter of PF-06480605 cross-linked paramagnetic beads was placed on a magnetic stand and the storage buffer was removed. Beads were acid washed to reduce drug leaching during the ACE procedure. Beads were resuspended in 500 µL of 100 mM glycine pH 2.0 and rotated end-over-end for 30 min at RT. Following incubation, the beads were placed on a magnetic stand and glycine buffer was removed, followed by resuspension in PBST/1% BSA and letting stand for 1 min. This wash procedure was repeated 2 more times and beads were stored in PBST/1% BSA at 2–8°C until use. To ACE-treat the study samples and PC preparation, 50 µL of sample was added to 25 µL Assay Buffer from the SMC™ Bead–Based Assay Development Kit in a polypropylene V-bottom plate. Fifty microliters of 100 mM glycine pH 2.0 was subsequently added and the solution in the plate was incubated for 30 min at RT on shaker. Following the incubation, 60 µL of the drug-coated beads was added to the plate wells containing the treated study samples and PC preparation, followed by addition of 14 µL of 1 M Tris pH 9.0 to neutralize the mixture. The plate was then sealed and placed on a shaker overnight at RT. After the incubation, the plate was placed on a BioTek ELx405™ plate washer with magnetic plate adapter and washed 6 times with 300 µL 1X THST buffer. Elution of the affinity captured ADA or PC was performed by adding 50 µL of 100 mM glycine pH 2.0 to the pelleted beads in each well followed by a 10-min incubation at RT on a plate shaker. The plate was placed on a magnetic adapter and beads were pelleted for 1 min. Fifty microliters of eluted material was removed and neutralized with 7 µL of 1 M Tris pH 9.0 in separate tubes. Storage buffer (PBST/1% BSA) was added in the amount of 25 µL and tubes were stored frozen at −70°C until use.

Step 2: Preparation of ADA-Enriched Solution-Based Drug Equilibrium Samples for Affinity Determination

A solution-based equilibrium approach was used to measure the apparent affinity of ADA binding to drug. Briefly, enriched ADA from patient samples or PC-spiked serum was kept at a constant dilution (fixed interactant) and incubated with a broad concentration range of unlabeled PF-06480605 (variable interactant) until equilibrium was reached. Equilibrium time was determined to be approximately 24 h, based on the MAb PC (data not shown). Initially, the drug stock was diluted serially 5-fold for a total of eight dilutions ranging from 1000 to 0.01 ng/mL using Standard Diluent from the SMC™ Bead–Based Assay Development Kit and served as the variable affinity interactant. The ACE-processed samples and PC were fixed to one dilution as follows: processed samples were diluted 20-fold and processed PC diluted to 1:15,000 in Standard Diluent. Affinity solution samples were prepared by mixing 50 µL of each of the dilutions of the PF-06480605 to 50 µL of the ACE-processed patient samples or PC and incubated on a plate shaker at RT until equilibrium was reached (at least 24 h).

Step 3(A): ADA Sample Analysis on the Singulex Erenna Free ADA Ligand-Binding Assay

The equilibrated affinity solution sample set was analyzed with a bead-based free ADA bridging LBA using Singulex Erenna. The bridging ADA format used PF-06480605 to capture and Alexa fluorophore labeled PF-06480605 to detect and measure free (unbound) ADA in the affinity solution samples (Fig. 1C). A reference curve using the ACE-processed MAb PC was also included in each run to evaluate apparent ADA concentration in the processed affinity samples. The concentration of the processed MAb PC was based on an approximated 30% recovery post-ACE procedure, which was determined during ACE method optimization (data not shown). The % recovery of the MAb PC post-ACE procedure was determined by interpolating the ACE-treated MAb PC concentration from a MAb PC reference curve (not ACE processed) of known concentration. The apparent (estimated) ADA concentration of the processed affinity samples was determined using the same approach. A 1:20 dilution of the processed affinity samples was assayed and signal interpolated off the MAb PC reference curve to calculate apparent concentration. After the 24 h incubation step, 100 µL of the solution affinity samples, MAb PC, and the 1:20 dilution of ACE-enriched ADA samples were added to a polypropylene V-bottom plate and 100 µL of drug capture beads added to the samples and the plate was shaken at room temperature for 2 h. The plate was washed with 1× wash buffer from the SMC™ Bead–Based Assay Development Kit using a BioTek plate washer ELx 405 with magnetic bead plate adapter. The PF-06480605 Alexa Fluor 647 detector was centrifuged at 14,000 × g for 5 min and diluted in kit Assay Buffer to 10 ng/mL and filtered through a 0.2-µm syringe filter. After plate wash completion, 20 µL/well of the filtered detector was added to each well and incubated for 1 h at room temperature on a plate shaker. Post incubation, the plate was washed 2 times with magnetic plate adaptor using 1× wash buffer. Following the final aspiration on the plate washer, 11 µL/well of kit Elution Buffer B was added to the plate and placed on a plate shaker for 10 min. Ten microliters/well of kit Neutralization Buffer D was added to appropriate wells of a polypropylene 384-well plate. After 10 min of incubation, the Elution Buffer plate was placed on the magnetic plate adaptor for 2 min and 10 µL of eluted sample was transferred into the 384-well reading plate with neutralization buffer. The 384-well reading plate was centrifuged at 1100 × g for 5 min and heat-sealed with foil using FluidX thermosealer (Brooks Life Sciences, cat#4ti-0655). Finally, the sealed plate was read on the Erenna instrument.

Fig. 2figure 2

Anti-PF-06480605 PC was spiked into neat serum at a range of concentrations and processed through the ACE workflow. Resulting enriched PC preps were diluted serially and assayed on Gyrolab and Singulex Erenna using the ADA bridging assay format. Left Y-axis denotes Erenna assay signal as detected events, and right Y-axis denotes Gyrolab Response Units signal. X-axis denotes PC on-plate concentrations for the dilution curves in pg/mL 

Step 3(B): ADA Analysis on the Gyrolab-Free ADA Ligand-Binding Assay

The equilibrated affinity solution sample set was analyzed with a bead-based free ADA bridging LBA using Gyrolab. The bridging ADA format used biotinylated PF-06480605 to capture and Alexa fluorophore–labeled PF-06480605 to detect and measure free (unbound) ADA in the affinity solution samples (Fig. 1C). A reference curve using the ACE-processed MAb PC was also included in each run to evaluate apparent ADA concentration in the processed affinity samples. The concentration of the processed MAb PC was estimated based on an approximated 30% recovery post-ACE procedure, which was determined during ACE method optimization (data not shown). The % recovery of the MAb PC post-ACE procedure was determined by interpolating the ACE-treated MAb PC concentration from a MAb PC reference curve (not ACE processed) of known concentration. After the 24 h incubation step, biotinylated PF-06480605 drug was diluted to 3 µg/mL in Wash Station 1 buffer, and Alexa Fluor 647–labeled PF-06480605 was centrifuged at 14,000 × g for 5 min and diluted to a final concentration of 2 µg/mL in Rexxip F buffer. Reagents and PC samples were added to the 96-well PCR plate(s) which were sealed with specific Gyrolab foil seal(s) and loaded onto the instrument with the run executed with appropriate number of Gyrolab 1000 nL Bioaffy CDs.

Data Analysis: Custom R Shiny Application for Calculating Apparent KD

Because there is no affinity module associated with the Erenna instrument, a custom application using R Shiny was developed to calculate apparent KD values using the principles of binding interactants in solution at equilibrium. In addition, the polyclonal nature of the response was considered by using an approach by Stevens (19), where the response may be simplified as biclonal with the assumption that two populations (here, low and high affinity) exist in the sample, and a geometric mean of both populations could give one KD readout.

The following two model equations were used for fitting data to determine apparent KD and are described in detail elsewhere (19) where Ab is the maximum signal for the corresponding phase, Dose is the amount of PF-06480605 added to the equilibrium affinity solution, KA is the equilibrium association constant and KD is the apparent equilibrium dissociation constant:

Monophasic Approach

$$Response= \frac_)}_)}^}$$

Biphasic Approach

$$Response= \frac_\, (1+2\, Dose\, _)}_)}^}+ \frac_\, (1+2\, Dose\, _)}_)}^}$$

Using KD = 1/KA, we can also write the above models substituting KA for KD as follows:

$$Response=Ab\, (1- \frac_}+1\right)}^})$$

$$Response=_ \left(1- \frac_}+1\right)}^}\right)+ _\, (1- \frac_}+1\right)}^})$$

Affinity curves were fit with both monophasic and biphasic models followed by applying adjusted Akaike information criterion (AIC) to determine which model was best suited for each curve (20). The adjusted AIC considers both the number of parameters and the sample size (number of different concentrations of the variable interactant). The probability of each model is calculated to determine the best fit, with probability > 50% designated as the model fit acceptance criteria.

The monophasic model fit result is relatively straightforward and has been described in detail elsewhere (21). Briefly, the biphasic model has two separate outputs described below:

(1)

Proportion: The biphasic model assumed there were two affinity populations present (high and low KD) and had two distinct Ab values, one for each phase. Each phase represented a portion of the overall response. The two Ab values were normalized, and the output was the corresponding % proportion attributed to phase 1 (high affinity) and phase 2 (low affinity).

(2)

Geometric mean: The biphasic model had two KD's, KD1 and KD2, for high and low affinity populations, respectively. For simplicity, one KD measurement to describe the overall affinity profile was also included. To that end, the geometric mean of KD1 and KD2 was computed (denoted KD(geo)) and used as one practical affinity measurement which accounted for both high and low affinity populations. The KD(geo) fell between KD1 and KD2 and was generally close to the monophasic KD.

The quality of the affinity curve was dependent on the free ADA LBA signal window achieved. The signal window was defined as the ratio of the lowest signal to the highest signal as the drug concentrations were varied during solution equilibrium incubations. The free ADA LBA signal window must be high enough to demonstrate a sufficient level of inhibition over the range of drug concentrations so KD may be accurately calculated. The accuracy of the KD value would be negatively impacted if the affinity inhibition curve was too flat. A practical acceptance test, referred to as a “flatness test,” was applied to determine if the affinity curves (both monophasic and biphasic) had an appropriate inhibition profile and signal window to generate an accurate KD. First, the affinity curves were fit for both monophasic and biphasic models. At every drug concentration, based on the fitted curve, there would be a predicted response value with a confidence bound limit. The upper and lower confidence bound of each prediction was computed with confidence level set at 99% across the entire drug concentration range, from the highest level to the lowest. If the lower bound of the prediction at the lowest drug concentration point (highest assay signal) overlapped with the higher bound limit at the highest drug concentration (lowest assay signal), then the responses at either end of the curve are not statistically differentiated, implying signal and/or inhibition was not obtained and the KD measurement was not acceptably accurate. For the best quality KD value, both monophasic and biphasic curve fits must pass the acceptance criteria.

留言 (0)

沒有登入
gif