We performed a comprehensive analysis of 93 human or humanized therapeutic antibodies to investigate the relationship between immunogenicity risk and V/D/J gene usage. We found that V gene usage is significantly associated with clinical immunogenicity of human-origin antibodies. The result suggests that certain V genes should preferably be used in antibody engineering to reduce immunogenicity risk.

Immunogenicity risk assessment of biotherapeutics is routinely performed through various in vitro techniques for candidate selection. In vitro assays using PBMCs can mimic protein uptake, immune cell maturation, pro-inflammatory cytokines secretion and T cell proliferation, to assess immunogenicity potential for candidate selection process. Simultaneously, critical quality attributes of formulations can be evaluated for their impact on immune response by using cell lines and human PBMC cytokine secretion assays. We have successfully developed toolbox for assessing the immunogenicity risks.

Predicting the risk of unwanted immunogenicity against proteins and engineering to directly address these risks is a significant challenge when developing biotherapeutic drugs. Utilizing machine learning and computational protein design, we can tackle the complexity of this multiparameter task. Here we will review fundamental concepts involving computational prediction and mitigation of unwanted immunogenicity. We will utilize proof-of-concept examples to illustrate how we incorporate these techniques into the drug development process.

We have developed a new state of the art graph neural network model for MHC class ii peptide presentation, graph-pMHC. We show a simple strategy to accurately determine the anti-drug antibody rate of antibody drugs using the model. The work represents a step forward in accuracy for both MHC class ii peptide presentation modeling, and the prediction of immunogenicity of antibody drugs.

Immune tolerance induction is necessary to optimize patient outcome in diverse clinical settings and treatments. In most such settings, a single magic bullet approach will be inadequate and thus, strategies to diminish extant effector mechanisms and minimize innate immune pathways will be essential to prime the immune milieu to allow induction/boosting of regulatory cells and mechanisms. This seminar will present such considerations in the context of protein, gene, and cell therapies.

The presence of pre-existing anti-AAV neutralizing antibodies (AAV NAb) may limit the efficacy of AAV-gene therapy. Moreover, a first administration of an AAV vector induces high titers of treatment emergent AAV NAb, which may compromise repeat dose administration with the same vector. Depletion of AAV NAb by immunoadsorption plasmapheresis (IAP) is a strategy that could allow successful vector administration in recipients with either pre-existing or treatment-emergent antibodies. The objective of this study was to evaluate the effectiveness of IAP to remove AAV serotype 5 (AAV5) NAb from animals sensitized by an initial gene therapy dose to enable a successful second dose administration. Results showed multiple rounds of IAP over multiple days enabled up to 99% reduction of AAV-Induced TAb titers in NHPs and enabled a successful second dose administration.

Adeno-associated viruses (AAV) are potent vectors used for gene therapy, but recent clinical findings have revealed immunogenicity-related challenges, including formation of a cytotoxic immune response against AAV capsid protein in transfected cells. We report that deamidation, a spontaneous post-translational modification that occurs in multiple AAV serotypes, can result in formation of CD4 T cell epitopes with a Th1 cytokine pattern in donor samples with specific HLA backgrounds.

Immune responses to AAV vectors are a key limitation to successful gene therapy. However, the lack of fully predictive in vitro assays and animal models has limited the speed of progress in this key area of research. The development of an end-to-end approach to immunogenicity assessment and management is critical to define an ideal vector design, manufacturing process and targeted immunosuppressive regimens.

In vitro tools are widely used during the development of biologics to assess immunogenicity liabilities and select candidates. However, in the absence of assay standardization and harmonization, the interpretation and translation into an overall immunogenicity risk of these preclinical data remain challenging. Here we will discuss the cross-industry work led by HESI and AAPS to develop common reference biologics for immunogenicity risk assessment, a critical first step toward assay harmonization.

Immunogenicity is monitored at the patient level and reported at the study level, whereas aggregate program data may provide a broader understanding of humoral immunogenicity. We report data from a mAb cocktail (REGEN-COV) against the SARS-CoV-2 spike protein, across multiple studies that include a range of doses, administration routes, and durations of follow-up. We highlight the relationship between immunogenicity, exposure, and administration, as well as important implications for development of therapeutic proteins.

This presentation will cover ADA assessment of multi-domain therapeutics with a flow chart on Boehringer Ingelheim’s risk-based approach on clinical ADA characterization of multi-domain therapeutics. Case studies of multi-domain biotherapeutics containing scFv will be provided that describe the reagents and tools used to identify the location of pre-existing ADA and domain specificity characterization in clinical ADA samples. This presentation will also share our strategy for the development of an orthogonal clinical ADA method to differentiate pre-existing ADA from treatment emergent ADA.

We will discuss and describe the IEDB & CEDAR initiatives, which represent two sibling databases and analysis resources, focused on infectious diseases, allergy and autoimmunity epitopes, and cancer epitopes, respectively. We will start by reviewing the database structures and the epitopes derived from scientific literature and direct submissions that are hosted in the freely accessible database. Next, we will review how the data can be queried for a number of exemplary user applications. In addition, we will illustrate the various bioinformatic tools that are hosted in the databases. These tools range from tools to predict MHC binding, processing and immunogenicity, to tools to facilitate the analysis of epitope related data.