Identifying Safety Thresholds for Immunosuppressive Drugs: Applying Insights from Primary Antibody Deficiencies to Mitigate Adverse Events in Secondary Antibody Deficiencies Using Mathematical Modeling of Preclinical and Early Clinical Data

Abstract

Immunosuppressive drugs can alleviate debilitating symptoms of autoimmune diseases, but, by the same token, excessive immune suppression can result in an increased risk of infection. Despite the dangers of a compromised immune system, clear definitions of what constitutes excessive suppression remain elusive. Here we review the most common infections associated with primary antibody deficiencies (PADs), such as agammaglobulinemia, common variable immunodeficiency (CVID), and IgA deficiency, as well as infections that are associated with drug-induced or secondary antibody immunodeficiencies (SADs). We identify a number of bacterial, viral, and fungal infections (e.g., Listeria monocytogenes, Staphylococcus sp., Salmonella spp., Escherichia coli, influenza, varicella zoster virus, and herpes simplex virus) associated with both PADs and SADs, and suggest that diagnostic criteria for PADs could be used as a first-line measure to identify potentially unsafe levels of immune suppression in SADs. Specifically, we suggest that, based on PAD diagnostic criteria, IgG levels should remain above 2-3 g/L, IgA levels should not fall below 0.07 g/L, and IgM levels should remain above 0.4 g/L to prevent immunosuppressive drugs from inducing mimicking PAD-like effects. We suggest that these criteria could be used in the early stages of drug development, and that pharmacokinetic and pharmacodynamic modeling could help guide patient selection to potentially improve drug safety. We illustrate the proposed approach using atacicept as an example and conclude with a discussion of the applicability of this approach for other drugs that may induce excessive immune suppression.

Fig. 1

Infections that are commonly associated with secondary antibody deficiencies [SADs] (data adapted from [31]), and infections that are also common in primary antibody deficiencies [PADs] (data adapted from [25]). Bacterial infections are colored red, viral inflection are in blue, fungal infections are in brown, and parasitic infections are in green. The occurrence of similar infections in patients with PADs and SADs suggests there may be value to using diagnostic criteria from PADs to inform safety thresholds for SADs

Fig. 2

Summary of criteria that correlate with increased infections in patients with primary antibody deficiencies, as described in Sect. 2; it is proposed that these criteria could be used as safety thresholds for secondary antibody deficiencies

Fig. 3

A schematic diagram of the sample model described in system (1). Parameter descriptions and units are given in Table 1

Fig. 4

A schematic representation of the effect of a drug on a biomarker, such as a generic immunoglobulin IgX, as captured by indirect response models (note that the graph does not depict a specific compound; it is used for illustrative purposes)

Fig. 5

Using modeling and safety thresholds to guide initial patient selection to minimize the risk of adverse events associated with immunosuppressive drugs. PK pharmacokinetic(s), PD pharmacodynamic(s)

Fig. 6

Schematic representation of the IgG indirect response model used to model the effect of atacicept on IgG production

Fig. 7

Scenario analysis of the variation in the drug concentration and IgG concentration with the administration of 150 mg of atacicept weekly for 24 weeks. Pharmacokinetic parameters are defined in Table 1. The treatment schedule and the maximum IgG reduction of up to 40% were obtained from [35]. A Simulated  pharmacokinetics (PK) of 150 mg of atacicept administered weekly for 24 weeks (to mimic the design of the ADDRESS II clinical trial of atacicept). B Simulated dynamics of the safety biomarker IgG for various baseline levels

Fig. 8

Simulated IgG profiles following 150 mg weekly (QW) administration of atacicept in systemic lupus erythematosus (SLE) subjects for 24 weeks. The baseline IgG was taken to be between 5 and 7 g/L, with a median concentration of 6 g/L. The proportion of subjects for whom IgG is predicted to drop below the safety threshold of 3 g/L at any time point is 3.8%