10th Annual EUR BEBPA Bioassay Conference

September 27-29, 2017
St. Julian’s, Malta


Abstracts (Alphabetical by Speaker/Presenter)

Alternatives to the 5pl for Asymmetric Data

  • Speaker: Francis Bursa, Quantics BioStatistics
  • Abstract: The five-parameter logistic (5pl) model is often used to fit asymmetric bioassay data. However, it has several disadvantages in practice. In particular, it can only accommodate small amounts of asymmetry, and there can be mathematical problems in fitting it. In this presentation we will discuss three alternative functions which are also asymmetric but avoid some of the problems with the 5pl. These are (1) a mixture of four-parameter Gompertz models; (2) a function inspired by a mathematical model of bacterial growth; (3) a function which modifies the four-parameter logistic using a hyperbolic dependence on the dose. We will discuss the advantages and disadvantages of each of these alternatives, and we will show the results of a simulation study comparing their behavior when fitting realistic bioassay data.
  • (Day 2: Nuts & Bolts of Assay Lifecycle Management Session)

Development of Reporter Based Bioassays Capable of Selecting Optimal Human Immunostimulatory Antibodies with Varying Fc Gamma Receptor Dependence

  • Speaker: Mei Cong, Promega
  • Abstract: (Pending)
  • (Day 1: Bioassays for Antibody Products Session)

What’s the Best Strategy for Collecting Data to Estimate a Four-Parameter Logistic Curve? A Simulation Study Using Nonlinear Designs in JMP

  • Speaker: Ian Cox, SAS Institute
  • Abstract: The Four Parameter Logistic Regression or 4PL nonlinear regression model is commonly used for curve-fitting analysis in bioassays or immunoassays such as ELISA, RIA, IRMA or dose-response curves. It is characterized by its classic “S” or sigmoidal shape with top and bottom plateaus, and the focus is usually on getting the best estimates of IC50, EC50 or EC50. Statistically designed experiments, or DOE, provide the best way to determine data collection plans to meet practical objectives, and optimal designs have some specific advantage in many cases. Applying DOE to the 4PL case is simple because only one factor or independent variable is involved, x, and complicated because the nonlinearity of the underlying model necessitates an iterative approach. This paper consists of a simulation study that compares the efficacy of data collected at x values defined via a nonlinear optimal design with data collected at equally-spaced x values. Different sample sizes and random noise are simulated, and we show that the nonlinear DOE approach usually provides better estimates of the parameters of interest.
  • Contributing Authors: Massimo Martucci, JMP Division, SAS Institute
    Ian Cox, JMP Division, SAS Institute
  • (Day 1: Bioassays for Antibody Products Session)

Using Affinity Chromatography to Characterize Fc-Receptor Interactions

  • Speaker: Florian Cymer, Roche
  • Abstract: Receptors that bind the Fc-portion of IgG type antibodies are critical mediators of effector functions and can also critically influence the serum half-life of therapeutic antibodies. Characterizing the interaction of IgG-type antibodies with different Fc-receptors can be a challenging task. Current binding assays often characterize interaction of a monomeric therapeutic antibody with an Fc-binding protein, whereas e.g. in the case of complement component C1q involved in complement dependent cytotoxicity (CDC), hexameric and other oligomeric IgG-complexes clustered on the target cell surface are thought to represent the relevant binding partner(s). Another example is the interaction of the neonatal Fc-receptor (FcRn) with the Fc-portion of IgG-type antibodies, which occurs in a pH-dependent manner, and both binding at acidic pH and dissociation at neutral pH appear to be critical in terms of FcRn-dependent serum half-life. Nevertheless, binding experiments are often performed at acidic pH only.
    We report the development of affinity chromatographic methods to characterize the highly complex interaction of C1q and FcRn with Fc-portions of IgG-type antibodies. We use a wide range of samples including IgG-subtypes, well described mutants, enriched glycoforms as well as degraded samples to investigate the suitability of the affinity chromatographic methods and compare these to other methods commonly used during extended characterization of therapeutic antibodies. The advantages and limitations of these methods as well as their integration into a comprehensive assay strategy for extended biological characterization will be discussed.
  • Contributing Authors: Florian Cymer1, Michael Marshall2, and Tilman Schlothauer2
    1 Pharma Technical Development Analytics Biologics, F. Hoffmann-La Roche Ltd., 4070 Basel, Switzerland; 2 Biochemical and Analytical Research, Large Molecule Research, Roche Pharma Research and Early Development (pRED), Roche Innovation Center, 82377 Penzberg, Germany
  • (Day 2: Product Specific Bioassays Session)

Development, Optimization and Validation of a TNFα Neutralizing Assay Using a Quality by Design Approach

  • Presenter: Arnaud Delobel, Quality Assistance S.A.
  • Abstract: Development and optimization steps are of major importance to ensure the delivery of reliable results in the routine use of a bioassay. Quality by Design (QbD) approach requires a combination of thorough knowledge of the scientific background of the assay and the use of proper statistical tools to achieve best results. This is a stepwise process including analytical target profile (ATP) definition, critical quality attributes (CQA) setting, candidate method analysis, risk assessment of any parameter that may influence assay results, screening experiments to identify the most important parameters, optimization experiments on the major parameters. In the end, the “design space” of the assay is obtained defining the range of operating conditions that ensures, with a given probability of success, that results will meet CQA. In this study, we developed and optimized a TNFα neutralizing assay on L929 cells for Adalimumab testing using a QbD approach and validated it according to USP <1033> guideline.
  • Contributing Authors: Fabian Vandermeers, Capucine Lepers, Florence Dusserre-Bresson, Christelle Plennevaux, Stéphanie Pennincx, Julie Nicaise, Estelle Lara and Arnaud Delobel
  • (Poster)

Does it or Doesn’t it? Only the Assay Control Samples Knows for Sure

  • Speaker: Stan Deming, Statistical Designs and Mike Sadick, Catalent
  • Abstract: (Pending)
  • (Day 2: Nuts & Bolts of Assay Lifecycle Management Session)

Harnessing BLI Technology for Overcoming the Effects of Protein Aggregates on Fc Receptor Binding

  • Speaker: Marina Feschenko, Biogen
  • Abstract: Aggregates of antibodies and Fc-fusions are known to affect protein–protein binding. The magnitude of these effects varies for different products and assays. We demonstrated that presence of aggregates in samples significantly increased binding potency values in AlphaScreen-based FcRn and Fcγ receptor binding assays, sometimes masking the loss of potency in stressed samples. Biolayer interferometry technology was found to be less sensitive to aggregates and presented fast and reliable method for measuring Fc receptor binding.
  • (Day 2: Product Specific Bioassays Session)

Binding and Bioactivity Assays for Biosimilarity Assessment of Adalimumab and Bevacizumab

  • Speaker: Christoph Giese, ProBioGen AG
  • Abstract: The first biopharmaceuticals are running off patents, numerous biosimilar candidates are under development and a few biosimilar products are entering the market. According to the guidelines of EMA, FDA, WHO and others the candidates have to be extensively assessed for similarity against the marketed originator products. During the different stages of pre-clinical and clinical development the relevant in vitro and in vivo methods have to be performed in campaigns on a reasonable number of originator’s batches. The assay methods mature over time from robust lab methods to qualified and later validated methods prior to Phase III clinical testing.We will present binding and bioactivity assays data from early stages of biosimilarity assessment for – new Adalimumab (TNFa neutralizing mAb) and Bevacizumab (VEGF-neutralizing mAb) candidates including:- Cell-based neutralization assays- ADCC assays against soluble and/or membrane bound forms (e.g. memTNFa)
    – Target molecule and target cell binding (ELISA/BLI, flowcytometry)
    – C1q-binding (ELISA/BLI)
    – CDC assay
    – Fc receptor binding assays (FcgRIIIa V/F, FcgRIIa, FcgRI, FcRn; BLI)
    We will also present our current assays developments using new functional NK cell lines and recombinant cell lines stably expressing e.g. memTNFa and Fc receptors.
  • (Day 1: Bioassays for Antibody Products Session)

In-Vitro Functional Bioassays of Candidate Therapeutics in Immuno-Oncology

  • Speaker: Severine Giltaire, ImmunXperts SA
  • Abstract: During the last years, significant advancement has been made in the clinical application of cancer immunotherapies. Molecules directed against immune checkpoints and other agonists show great promise for treatment of a variety of malignancies. Next to CTLA-4 and PD-1 blockade, a wide range of therapeutics with the potential to reverse the tumor-induced suppression are under development.This increasing interest in the tumour microenvironment leads to focus on the development and the standardization of in vitro bioassays using human immune cells in order to evaluate the effectiveness of candidate therapeutics and combination therapies.Among these in vitro bioassays, the mixed lymphocyte Reaction (MLR), the antigen-specific recall activation assay or CD3/CD28 activation with human T cells are models that mimic a physiological T cell response. These kinds of assays can be used to screen new candidate therapeutics such as immune checkpoint blocking antibodies.An important factor for sensitive assays and consistent results is the quality of the primary immune cells. PBMC are isolated and cryopreserved shortly after blood redrawn. All donor preparations are quality controlled and HLA typed and optimized procedures are used to generate functional dendritic cells which are co-cultured with allogenic T cells. Response levels can be evaluated by the assessment of proliferation or measurement of cytokine production.
  • (Day 2: Product Specific Bioassays Session)

Potency Testing: A Landscape Overview

  • Speaker: Marie Gottar-Guillier, Novartis
  • Abstract: (Pending)
  • (Day 1: Assay Development Session)

Insulin Bioassay Development to Replace In-Vivo Rabbit Test

  • Presenter: Christine Graf, Sanofi
  • Abstract: (Pending)
  • (Poster)

Potency Testing for Hematopoietic Stem Cell Therapies

  • Speaker: Dirk Haubert, Novartis
  • Abstract: (Pending)
  • (Day 2: Product Specific Bioassays Session)

Mechanism of Action Assays to Determine the Fc Effector Function of Palivizumab

  • Speaker: Ulrike Herbrand, Charles River Biopharmaceuticals Services GmbH
  • Abstract: The Fc (crystallizable fragment) region of therapeutic antibodies can have an important role in their safety and efficacy. Palivizumab is a humanized monoclonal IgG1 antibody directed against an epitope in the A antigenic site of the F protein of respiratory syncytial virus (RSV) and is used for the prevention of RSV infection in children. Even though the Fc effector function potential was assessed to be low for Palivizumab there is a requirement to evaluate these capabilities, especially with regards to glyco-optimized follow-on biologics with potentially altered Fc effector function potential.
    The presented model for mechanism of action (MoA) testing of Palivizumab is based on a human lung carcinoma cell line with epithelial-like morphology. The cells are infected with RSV A Long, a prototype strain for subtype A. The infected cells are then used as target cells in reporter surrogate approaches to assess the antibody-dependent cellular cytotoxicity (ADCC) and antibody-dependent cellular phagocytosis (ADCP) capabilities of Palivizumab. The reporter assays are reliable, reproducible and precise, which is reflected by the rising use and acceptance of surrogate approaches instead of the tedious, time consuming and sometimes highly variable primary MoA assays.
  • (Day 2: Product Specific Bioassays Session)

Method Comparison Revisited: R Package for Bivariate Least Squares Regression

  • Speaker: Walter Hoyer, GSK
  • Abstract: We have developed an in-house statistical tool for general method comparison studies. After uploading data and setting a few options, the user receives a complete pdf report within seconds. In addition, one of our team has just released the R-package “BivRegBLS” on CRAN, which provides a generalization of Deming regression and will be presented in the talk. We plan to integrate this methodology into the existing automated report.
  • (Day 2: Nuts & Bolts of Assay Lifecycle Management Session)

Design-of-Experiment Based Optimization of a Potency Assay for an Enzyme-Inhibiting Nanobody®-Based Therapeutic

  • Speaker: Jasper Jacobs, Ablynx NV
  • Abstract: Ablynx is a biopharmaceutical company engaged in the development of Nanobodies®, proprietary therapeutic proteins based on the smallest functional fragments of heavy chain antibodies, naturally occurring in Camelidae. To support batch release, stability studies and biological characterization of a novel enzyme-inhibiting Nanobody compound, a fluorescence resonance energy transfer (FRET)-based potency assay was developed. In this assay, enzyme activity and inhibition thereof are assessed through kinetic measurement of the cleavage of a fluorogenic peptide substrate by the target enzyme. The assay is therefore considered to be reflective of the mechanism of action (MoA) of the Nanobody.
    Following initial development, the FRET-based assay failed the accuracy and linearity criteria, particularly in the range of 50-80% potency, which prompted further assay optimization. We describe a custom multifactorial design-of-experiment (DOE) based approach to improve the assay performance. Using JMP statistical software, a DOE was set up to investigate the effects of selected assay parameters, particularly on the accuracy of a test sample with 50% potency. The tested factors included the concentrations of target enzyme [E] and fluorogenic substrate [S], as well as pre-incubation time of enzyme with the Nanobody. Additionally, different operators were included and analyzed as a random variable in order to find an optimum that is robust to inter-operator variation.
    The DOE results exposed major influence of [E] and [S], but not pre-incubation time, on the accuracy response. An optimum for the tested assay parameters was predicted by the DOE model, applied in the assay and subjected to qualification. The accuracy was well within the acceptance criteria for test samples with 50% potency, but accuracy and linearity now failed at other levels across the range, specifically those samples with 80% and 150% potency. This suggested that although essentially the DOE optimized the accuracy of the 50% test sample, not all crucial factors and/or responses were included in the DOE design. Therefore the ,DOE dataset was augmented, including additional factors (reaction temperature and reaction buffer pH and reaction buffer salt concentration) and additional responses (accuracy of 80% and 150% test samples). Analysis of the augmented DOE revealed that these extra factors all significantly impacted the accuracy of at least one of the test samples and predicted new optimal assay conditions. Next, the optimized assay was successfully qualified, with excellent performance well within the acceptance criteria for accuracy (-3.5 to 6.2%), precision (3.4 to 9.4%) and linearity across the range of 50% to 150%, as well as for whole-plate uniformity, specificity and robustness to reagent (enzyme and substrate) lot changes and freeze-thaw cycles.
    In conclusion, our case study shows that DOE is a powerful tool for the optimization of your bioassay, while at the same time its success depends strongly on the selected factors, their levels and the relevancy of the included responses. Careful consideration and evaluation of the DOE design is strongly recommended. In addition, this case study highlights the advantage of a two-step approach, allowing adaptation of the included factors and responses and of the selected factor levels in the augmented experiment.
  • (Day 1: Bioassays for Antibody Products Session)

Ready-To-Use Cryopreserved Cells in a GMP Bioassay for an ADC

  • Presenter: Stephanie Katzenbach, Abbvie Deutschland GmbH & Co. KG
  • Abstract: (The use of cells as a cryopreserved, ready-to-use reagent is a very common practice in high-throughput screening (HTS) of compounds by cell-based assays. This process is considerably less labor intensive, more consistent and allows more flexibility than traditional continuous cell culture which is generally employed in GMP bioassays. The purpose of the present study was to determine if the use of cryopreserved ready-to-use cells could also be adapted to a bioassay intended for lot release and stability testing. The cytotoxicity assay of ABBV-ADC was chosen as one of our pilot projects.)
  • (Poster)

Functional Characterization Strategy for T-cell Bispecific Antibodies

  • Speaker: Julia Krueger, Roche
  • Abstract: T cell bispecific antibodies (TCBs) that recruit and engage T cells for tumor cell killing have gained strong interest. A profound assay strategy for extensive functional characterization is required to study TCB properties and to control efficacy and safety.
    We have developed a two cell-line bridging reporter assay covering tumor target-dependent T cell activation as potency assay for TCBs. In addition, a tumor target-independent reporter cell assay and SPR binding assays for structure-function correlations will be presented.
  • (Day 1: Bioassays for Antibody Products Session)

Single Donor KILR®CD16 Effector Cells to Drive Robust and Reproducible ADCC and T-cell Redirection

  • Speaker: Jane Lamerdin, DiscoverX Corporation
  • Abstract: Class I therapeutic antibodies achieve their clinical efficacy not only by binding to their target antigen, but also through Fc domain-mediated recruitment of immune cell effectors to attack and kill the target cell. Therefore, developers of therapeutic antibodies must assess all possible mechanisms of action (MOA) of their molecules, including antibody-dependent cell-mediated cytotoxicity (ADCC). Success of ADCC assays is highly dependent on the quality of effectors used. However, primary human cells (such as PBMCs or NK cells) suffer from inter-individual variability, while NK cell lines overexpressing CD16 often show high background lysis in susceptible cell models and functional variability under different culture conditions.Here, we introduce the KILR CD16 Effector Cells, primary cytotoxic T lymphocytes from a single donor stably expressing CD16. These uniformly manufactured primary human cells maintain their T cell phenotype (TCR, CD3+ and CD8+), and have the added ability to drive ADCC through the overexpression of CD16. The single-donor KILR CD16 Effector cells demonstrate consistent ability to drive ADCC and T-cell redirection (TCR) over time by eliminating donor variability. In addition, multiple batches of manufactured cells demonstrate consistent CD16 expression and killing capacity further reducing long term variability. These cells are optimized as frozen ready-to-use cells, and generate robust assay windows, with excellent repeatability and precision, enabling their use for lot release and characterization studies. Data will be presented comparing the performance of these KILR CD16 Effector Cells and primary PBMCs, for applications such as ADCC and TCR.
  • (Day 1: Assay Development Session)

How do we Calculate a Reportable Values for Multiplate (Multirun) Potency Assays

  • Speaker: Laureen Little, BEBPA
  • (Day 1: Interactive Session)

Make Your Bioassay Great…the First Time

  • Speaker: Mike Merges and Mike Sadick, Catalent
  • Abstract: (Pending)
  • (Workshop 1)

Development of Combinatorial Potency Assays to Analyze Bi-Specific Protein Therapeutics

  • Speaker: Matthias Naumer, AbbVie Deutschland
  • Abstract: Bi-specific antibodies represent an emerging class of multi-specific protein therapeutics that open up a wide range of applications, e.g., the simultaneous targeting of two disease-mediating entities. The growing complexity associated with combining different targeting moieties in a single molecule relates not only to a more complex in vivo mode-of-action, but also requires potency assays reflecting this activity. We applied Reporter Gene Assay technology for the development of state-of-the-art potency assays that simultaneously detect the binding of different target molecules in one assay, instead of two independent assays, and at the same time address agency demands for bi-specific molecules.All authors are employees of AbbVie. The design, study conduct, and financial support for this research was provided by AbbVie. AbbVie participated in the interpretation of data, review, and approval of the publication.)
  • (Day 2: Product Specific Bioassays Session)

Case Study: A Two-Tiered Approach for Outlier Identification in a Cell-Based Potency Assay

  • Presenters: Merle Nebel and Sonja Klingelhoefer, Richter-Helm-BioLogics
  • Abstract: The potency assay described in this case study is based on stimulation of a G-protein-coupled (hormone) receptor in a murine cell line, which subsequently leads to the intracellular release of the second messenger cAMP. The accumulated cAMP is finally quantified as response for potency determination via end-point measurement of a fluorescence signal generated by enzyme fragment complementation technique.
    Since this assay requires several small volume additions in a short time frame, it is prone to outliers. According to USP removing observations only based on statistical considerations should be used rarely. Therefore, a two-tiered approach for identifying outliers was evaluated with the aim to reduce the risk of excessive outlier identification in assays with low variability in raw data. The finally implemented procedure consists of two steps. In the first step, a visual assessment utilizing statistical characteristics like studentized residuals as a guidance for statistical outlier testing is performed. If further testing is indicated, the raw data are subsequently analyzed with respect to outliers by an algorythm established in SAS that uses the ROUT method (Motulsky et al., BMC Bioinformatics, 7:123 (2006)). The method combines robust regression with outlier removal in bioassay raw data.
    By using the two-tiered approach which is based on empirical data an efficient tool for reasonable outlier identification for the cell-based cAMP release assay could be established.
  • (Poster)

Introduction to Statistics for Potency Assays

  • Speaker: Nancy Niemuth, Battelle
  • Abstract: (Pending)
  • (Workshop 2)

Automated, Generic Direct Binding ELISA for Potency Measurements

  • Speaker: Petr Obrdlik, Novartis
  • Abstract: Selection of potency bioassay depends on the mode of action of the bio-therapeutic drug. If applicable, direct target-binding ELISA is often a preferred potency assay for pre-clinical and early clinical trials because of its low variability and easy execution. Nevertheless, the establishment of such potency ELISA still requires target-specific development and method qualification.
    To simplify the process, we first established a manual, generic direct binding ELISA method for measuring potency. This manual ELISA protocol is successfully used for several drugs in pre-clinical and clinical trials. As the next step, the method protocol was successfully adapted for fully automated potency measurement for development as well as QC. Here, we describe the critical protocol parameters and changes which had to be taken to achieve this goal. The lessons learned from this exercise will help with the automation of other potency QC methods in future.
  • (Day 2: Nuts & Bolts of Assay Lifecycle Management Session)

Complex and Novel Biological Therapeutics: Represent Challenges in Bioassay Analytical Development

  • Presenter: Ciara O’Farrell, Janssen Sciences
  • Abstract: Antibodies manufactured for therapeutic use have traditionally mimicked the action of the human IgG1’s whereby they bind specific target antigens. They are as such, monospecific, in that all molecules manufactured target the same antigen, are bivalent and have two binding sites per molecule capable of binding the target antigen. However, much research in recent decades has been focused on developing next generation and antibodies and antibody-like molecules. These new scaffolds include Bi-Specific Antibodies (e.g. Duobodies) and antibody fragment molecules such as BiTE (bispecific T-cell-engaging) and DART (Dual-affinity retargeting) molecules. Complex mechanisms of action (MOA) may also be engaged including T-cell redirection, ADCC, CDC, ADCP and apoptosis.These developments in biopharmaceuticals provide potential new therapeutic advances but also present new challenges for bioassay method development. This poster will outline the obstacles encountered when developing Bioassays for this new generation of biotherapeutics. Obstacles include molecules with multiple MOA for which more than one Bioassay may be required. Other considerations include controlling the extensive array of customized reagents, reference material qualification in multiple assays, the related management of normalization factors and the lack of experience and knowledge for developing assays with these MOA’s. This poster outlines the challenges faced from complex products arising from the Janssen pipeline and how they are being overcome to allow for commercial approval.
  • (Poster)

VAC2VAC – Vaccine Batch to Vaccine Batch Comparison by Consistency Testing

  • Speaker: Dieter Pullirsch, Austrian Medicines and Medical Devices Agency
  • Abstract: (Pending)
  • (Day 2: Nuts & Bolts of Assay Lifecycle Management Session)

2-in-1: Increasing Efficiency of Potency Assay Optimization via DoE

  • Speaker: Johannes Solzin, Boehringer Ingelheim
  • Abstract: Especially for complex potency assays, often needed for multi MoA-molecules, more labs are switching for assay optimization from classical OFaT (One Factor-at-a-Time)-approaches to DoE-approaches which generate more information of the assay. Therefore, typically a ≥2 tiered DoE approach is chosen, where at first, a screening DoE is performed to identify and exclude irrelevant factors, which is then followed by an optimization DoE, to find the optimal assay settings.
    Here we show view case studies, where screening and optimization DoE were directly combined in a single response surface model DoE, allowing more statistical power, and thus, more information about the assay. The DoE can be performed within the same time as OFaTs, or even faster than the classical 2-tiered DoE-approach.
    In order to determine the assay optimum via DoE, it is necessary to assess the assay quality with significant parameters. Here we considered parameters, like Signal-to-Noise, Signal-to-Background, Mean Weighted Standard Deviation, Z-factor, R2, etc., and discuss their advantages and limits for different assay types.
  • (Day 1: Assay Development Session)

Challenges in Development & Validation of Early Stage Assays Where Reagents Including Reference & QCs Are Very Limited

  • Speaker: Kelly Thomas, Public Health England
  • Abstract: (Pending)
  • (Day 1: Assay Development Session)

Can We Consign “The Edge Effect” to History?

  • Speaker: Ann Yellowlees, Quantics Biostatistics
  • Abstract: We can waste about 40% of multi-well plates to avoid biasing the relative potency (RP) due to “the edge effect”. Quantics recently used a Latin square design to support a batch release assay designed to eliminate variation across the whole assay plate. This design allowed row and column effects to be identified, and correctly accounted for in the estimation of reportable values, avoiding any assay bias. The assay data clearly demonstrated that row and column effects are sometimes seen and most importantly can be managed. In this presentation we will describe the Latin square and explain how it can be used in a typical multi-well plate assay to overcome these issues and maximize the efficiency of the assay. We will present data sets showing these effects. We will show the results of a simulation study demonstrating the extent of estimation bias which can be avoided using this technique.
  • (Day 1: Assay Development Session)