Biologic drugs have the potential to induce adaptive immune responses, resulting in the formation of anti-drug antibodies (ADA). Though they can also be benign, the negative consequences of ADA include rapid drug clearance, loss of efficacy, and anaphylaxis. According to a 2016 review of prescribing information, upwards of 90% of approved biologics induce ADAs in a subset of the populations in which they have been studied (Wang et al., 2016). While the likelihood of ADA induction is increased for products that are less similar to native human proteins (Harding et al., 2010), other factors also have potential to influence the formation of ADAs, such as formulation, patient genetics, patient history, and treatment regimen (Andlauer et al., 2020; van Brummelen et al., 2016; Chung et al., 2008). Ultimately, the purpose of ADA measurement is three-fold: assessment permits understanding as to whether drug efficacy, pharmacokinetics, or adverse events are related to the presence or absence of ADA. For example, understanding if ADAs reduce the circulating concentration or plasma halflife of the product is an important consideration as these types of ADAs will likely impair the utility of a therapeutic and may form large complexes that influence drug safety. Lastly, understanding the role of ADAs in treatment outcomes provides clinicians and investigators the opportunity to design or select alternative products which may improve the likelihood of successful treatment.

ADA responses can roughly be categorized into two classes: neutralizing ADA, which bind to the drug product and inhibit pharmacological function, and non-neutralizing ADA, which bind to but do not directly inhibit pharmacological function of the drug (Gunn 3rd et al., 2016). Both types of ADA can modify efficacy by inducing rapid clearance (Rojas et al., 2005), preventing clearance and increasing bioavailability (Gomez-Mantilla et al., 2014), or inducing other immune-mediated outcomes (e.g. cytokine release syndrome, anaphylaxis, or target depletion). While the incidence rate of ADA responses varies greatly across drug products and populations, improvements in assay sensitivities have resulted in increased ADA detection, and varying observations about their clinical relevance (Zhou et al., 2013; Plotkin, 2010; Song et al., 2016).

Methods for ADA detection typically follow a tiered approach. An initial, high sensitivity screening assay is used to determine if ADA are present in a sample and is followed by a confirmatory assay in order to determine if the positive result in the screening assay is due to a specific (competable) interaction. Finally, titering and/or functional assays (e.g. neutralization, isotype, epitope) are carried out to further characterize the ADA response and measure ADA response magnitude. For the initial detection assay, various methodologies that measure protein-protein interactions have been implemented. These include enzyme linked immunosorbent assays (ELISA), electrochemiluminescent (MesoScale Diagnostics), microarray (SQI Diagnostics), solid phase (ImmunoCap), and bead-based (Gyros, Luminex) assays. Other less widely-used methods include surface plasmon resonance, radioimmunoprecipitation, and biolayer interferometry (Wadhwa et al., 2015). Regardless of methodology, factors such as sensitivity, tolerance to matrix, and the impact of circulating therapeutic or therapeutic target are all important factors to consider. As such, the FDA and other government agencies have drafted guidance regarding how to design, qualify, and/or validate these assays (Immunogenicity Testing of Therapeutic Protein Products — Developing and Validating Assays for Anti-Drug Antibody Detection, 2019; Guideline on Immunogenicity Assessment of Therapeutic Proteins European Medicines Agency, 2017).

While the FDA recommends evaluating and optimizing ADA assay sensitivity, the method by which sensitivity is defined is not fully agreed upon. One challenge, as noted by Song et al. (Song et al., 2016), is that positive controls (PC), typically monoclonal antibody-idiotypes (anti-id) isolated from hyper-immunized animals, may or may not reflect typical characteristics of a human polyclonal antibody response (Gunn 3rd et al., 2016). While PCs are not typically used to directly assign the Tier 1 positivity threshold, or cut-point, their role in optimizing and characterizing assay performance underscores the importance of understanding how different PCs behave and what impacts they may have on the detection of ADAs and discovery of their relationships with study outcomes. Differences in PC antibody clonality, affinity, and target epitope may all impact sensitivity (Ishii-Watabe et al., 2018), and are further confounded by how the PC functions in the presence of circulating drug. In fact, increasing clinical use of antibodies with extended plasma half-life, novel genetic delivery methods that lead to persistent expression, and other new depot technologies are being used to prolong the presence of therapeutic antibody in circulation. Such extension has implications for ADA testing. and must be considered during assay development as signal in the most widely used ADA assay formats can be reduced or abolished in the presence of drug, resulting in false negative observations. Methodologies have been developed to detect ADAs after dissociation from the therapeutic, but these approaches often incur a cost to assay sensitivity, precision, and practicality (Song et al., 2016).

In this study, we sought to compare different methodologies for ADA detection and better understand how the choices made during assay development and qualification impact the way that ADAs are defined during clinical studies. With a particular emphasis on the impact of PC selection, we aimed to inform interpretation of ADA assay results as the field continues to work to enhance the ability to detect and characterize the ADAs that are most likely to be clinically relevant.