Regulatory Forum Commentary*

Abstract

One of the biggest challenges in drug development is interpretation of findings suggestive of immunostimulation, particularly when the findings are unexpected based on the known mechanism of action of the drug. Findings suggestive of immunostimulation do not necessarily preclude development of the drug. By understanding mechanism, the patient population, and species differences and by careful clinical monitoring, many of the drugs that produce immunostimulation can be successfully developed. This opinion piece describes various types of immunostimulation and a risk assessment strategy for management of unexpected immunostimulation in toxicity studies.

One of the biggest challenges in drug development is interpretation of findings suggestive of immunostimulation, particularly when the findings are unexpected based on the known mechanism of action of the drug. Findings suggestive of immunostimulation do not preclude development of the drug. By understanding mechanism, the patient population, and species differences and by careful clinical monitoring, many of the drugs that produce immunostimulation can be successfully developed. For the purpose of this opinion piece, “immunostimulation” is defined as stimulation of cells involved in innate or acquired immunity as a result of drug administration. Immune-mediated, drug-induced hypersensitivity reactions are beyond the intended scope. In this piece, I will start with a description of various types of immunostimulation, which will build the foundation for my proposed risk assessment strategy for management of unexpected immunostimulation in toxicity studies.

Background

The immune system is very complex and is not fully understood, despite active research in identifying targets for pharmaceutical intervention. Immunity involves a stunning variety of different cell types, including cells traditionally associated with acquired immunity (i.e., the various T and B lymphocytes and dendritic cells), cells typically associated with innate immunity such as natural killer (NK) cells and neutrophils, and cells such as monocytes/macrophages, which are associated with both innate immunity and driving recall responses in acquired immunity to a particular antigen. In fact, the very separation of immune processes into “innate” versus “acquired” categories is somewhat misleading because of the considerable interplay between cell types and soluble factors, including cytokines. In addition, newly discovered functions and phenotypes within specific cell types, such as NK cells, suggest that a given cell type previously labeled as belonging to a particular immune response may play roles in both types of immune responses. A growing number of cell types not previously considered part of the “immune system,” such as epithelial cells, platelets, and endothelial cells, have been identified as having immune functions. To guard against the cell-damaging effects of a stimulatory immune response, pathways that temper the response are activated almost simultaneously. In addition, a molecule intended to be inhibitory on one target may be stimulatory on unintended cell types or may interact with unintended targets. Furthermore, the activity of a drug on a given target may be more potent at lower concentrations than at higher concentrations, challenging traditional approaches to risk assessment (Cooper, Colonna, and Yokoyama 2009; Cooper and Yokoyama 2010; Stebbings et al. 2007; Murphy, Travers, and Walport 2008).

Because of the complexity of the immune system and the involvement of multiple, intertwined pathways in generating an immune response, it is not surprising that the immune system may respond in unexpected ways to pharmaceutical administration, especially if a drug is intended to alter an aspect of the immune response in some way, whether to dampen or stimulate a particular response. Also, because “immune cells” are located throughout the body and can be recruited at a moment’s notice or produced locally, there is potential for immunostimulation findings to be seen in virtually any tissue in the body. Cells of the immune system can be stimulated by various ways, including direct stimulation, challenge by pathogens or other antigenic proteins, necrosis of other cells and tissues, and immune complex formation and deposition. Intended or unintended immunostimulation warrants caution but need not preclude further drug development. Because the findings in toxicology studies are usually nonspecific, an understanding of mechanism of immunostimulation is important in order to understand human risk potential and develop a mitigation strategy, and this understanding usually requires further investigative work. Below, I have described various examples of immunostimulation which might be seen in toxicology studies.

Examples of Immunostimulation

Cytokine Release

One of the most devastating manifestations of immunostimulation is “cytokine storm,” a severe form of cytokine release syndrome. Cytokines released by activated cells trigger a cascade of events resulting in a systemic inflammatory response that resembles severe infection, with fever, hypotension, and potentially multiorgan failure. Most companies are currently conducting in vitro cytokine release assays (CRA) using human cells as part of their de-risking strategies for drugs, particularly large molecules (i.e., biologics such as antibodies and proteins) that interact with lymphocytes. However, the current state-of-the-science precludes making specific recommendations as to the conduct of these assays (Vidal et al. 2010). There is very little clinical experience paired with in vitro data to allow a clear extrapolation from in vitro data to cytokine release potential in the patient, and no clear threshold defining how much cytokine release is required in vivo to cause an adverse event. Furthermore, the target cell that might be responsible for cytokine release may not be present in peripheral blood, the typical donor material used to conduct the assay. Finally, interpretation of in vitro data can be challenging due to high variability in donors’ responses even to negative controls. CRAs are primarily useful for hazard identification, especially in identifying products that may produce severe, life-threatening cytokine release (cytokine storm) such as that produced by the anti-CD28 superagonist TGN1412 in 2006 (Suntharalingam et al. 2006), and are less useful for determining relative risk, exposure margins, or clinical consequences of in vitro cytokine release. If it is determined that a drug produces cytokine release, the product may still move forward, provided an appropriate risk mitigation strategy is in place (Vidal et al. 2010). In the wake of the TGN1412 incident, the Committee for Medicinal Products for Human Use developed the Guideline on Strategies to Identify and Mitigate Risks for First-in-Human Clinical Trials with Investigational Medicinal Products (Muller et al. 2009; European Medicines Agency [EMEA] 2007). Essentially, this guidance outlines an approach for first-in-human dose selection based on integrating in vivo and in vitro preclinical data to arrive at a minimal anticipated biologic effect level (MABEL), careful clinical monitoring, and gradual dose escalation.

Pseudoallergy

In addition to their role in IgE-induced hypersensitivity reactions, mast cells can be direct targets of IgE-independent stimulation by pharmaceuticals due to the chemical properties of the drug or drug solution in an IV infusion, or interactions with surface receptors on mast cells, resulting in degranulation and release of granule contents, including histamine. Also, drug, antidrug antibodies, or immune complexes formed by drug/antidrug antibody conjugates may activate complement, resulting in the formation of complement split products that bind to mast cell receptors and stimulate degranulation (Walker et al. 2010; de Weck 1984; Tawara et al. 2008). The clinical presentation in both of these scenarios mimics an IgE-mediated (Gell-Coombs Type I) allergic response because the effector molecules are the same. The term “pseudoallergy” is used because, in these instances, the degranulation is not mediated by IgE. The reaction typically occurs after the first dose within minutes of an IV infusion or around the time of maximum plasma concentration (C_max) with oral administration, and tends to diminish with subsequent administrations of the drug. In vitro mast cell assays, which measure histamine release upon exposure of mast cells to the drug, are valuable diagnostic and predictive tools; however, these assays are not readily available. If samples are collected around the time the reaction is seen, blood histamine concentrations can be determined to aid in the diagnosis. In vitro assays with human mast cells may also be useful in determining relative human risk. Pseudoallergy need not preclude development of a drug; the reaction is usually self-limiting and may be manageable with prior or concurrent antihistamine administration along with a slow infusion rate (to decrease C_max; Sylvia 2010).

FcR and Complement Binding

Some monoclonal antibodies (mAbs) bound to their intended target on cell surfaces interact with Fc receptors (FcRs) on NK cells, triggering the NK cells to kill the target cell (antibody-dependent cell-mediated cytotoxicity [ADCC]), or the mAbs can interact with FcRs on other effector cells, resulting in death of the targeted cell, cytokine release, and/or inflammatory responses. In complement-dependent cytotoxicity (CDC), complement is activated when mAbs bind with the c1q component of complement, activating the complement cascade, which activates immune cells. ADCC or CDC may be the desired mechanism of action, that is, when the target is a tumor cell or effector cell in other disease processes. However, if the target antigen is also expressed on normal cells, there is a high potential for undesired effects due to misdirected or excessive immunostimulation. Another situation in which undesirable ADCC or CDC occurs is when recognition of FcR by effector cells or complement binding, and subsequent killing of target cells, was unintentional. The immunoglobulin isotype selected as the framework has a great deal to do with the likelihood of ADCC and/or CDC, and selection should be based on promoting desired effects and avoiding undesirable effects. IgG1 has high binding affinity for FcRs as well as complement, and may be most appropriate for oncology indications or immunomodulatory indications aimed at reducing the numbers of effector cells. IgG2 has much less binding affinity for both FcRs and complement. Although IgG3 has strong binding affinity for most FcRs as well as complement, it is rarely used as a framework for therapeutic mAbs due to its short half-life. Finally, IgG4 binds weakly to FcRs and does not fix complement, but instabilities in its hinge region pose liabilities due to half-antibody exchanges in vivo, where therapeutic antibodies recombine with endogenous IgG4 (Brennan et al. 2010; Shapiro et al. 2011; Labrijn et al. 2009). Consequently, most pharmaceutical mAbs not intended to be cytotoxic are IgG2 or IgG1, which are preselected or modified to minimize Fc binding or activity as a result of Fc binding (Brennan et al. 2010). As part of a preclinical development program, preferably prior to selection of a candidate to go into animal toxicity studies and further drug development, FcR and C1q binding assays should be conducted, with positive results to be followed by ADCC and CDC assays, respectively. If the molecule demonstrates ADCC or CDC activity, and such activities are not desired effects, using a different Ig class or modification of the Fc portion of the molecule should be considered to minimize the potential for binding and cytotoxic consequences. Selection of a target antigen that is not widely expressed on cells not involved in the disease process will also minimize toxicity.

Immune Complexes

Administration of large molecules intended to be nonantigenic or minimally antigenic in humans often results in an immune response in nonclinical species within 7 to 14 days of dosing. The resulting anti-drug antibodies (ADAs) themselves represent a normal stimulation of the animal’s immune system, but the formation of immune complexes by ADAs binding to drug can have consequences that then stimulate the immune system further. Immune complexes may deposit in tissues, trigger complement activation, and result in inflammation and tissue damage. However, it is important to note that they may have no adverse consequence, so the finding of immune complexes in the absence of inflammation or tissue damage does not constitute “immune complex disease.” Findings resulting from ADA formation in animal studies are not likely to suggest human risk potential unless ADAs are also seen in humans. Immune complexes may also be formed by aggregated drug, which may have human risk potential if the drug is prone to aggregation under circumstances similar to those in animals.

Infusion Reactions

The term “infusion reaction” is a clinical term for reactions that happen within minutes to hours of administering a drug intravenously. The clinical signs include lethargy, respiratory distress, vomiting, hypothermia, hypotension, and fever; they can be transient and self-limiting, or can result in death. The term is nonspecific in that it does not implicate a particular mechanism; however, more specific terms should be avoided unless the mechanism is evident from a weight of evidence data review. Some of the mechanisms for infusion reactions include pseudoallergy or immediate (IgE-related) hypersensitivity, activation of complement, and cytokine release. Infusion reactions are usually manageable in the clinic, depending on the underlying mechanism. Slow infusion, scaling up the dose, pretreatment, and managing symptoms with corticosteroids and/or anti-histamine have been used to allow dosing with drugs known to cause infusion reactions (Dillman 1999; Chung 2008; Hong et al. 2012; Tawara et al. 2008; Vogel 2010).

Immunosuppression as a Cause of Immunostimulation

It may seem paradoxical to include immunosuppression as a cause of immunostimulation, but drugs which suppress immune functions may trigger an excessive response in the spared pathways by either responding to pathogens they might not otherwise encounter or compensating for the impacted pathways in dealing with common everyday exposure to microbes. When the apparent stimulation is secondary to immunosuppression of one or more immune functions, the biggest concern is increased susceptibility to infection or virus-related tumors. There are numerous tools to assess the potential for immunosuppression (weight of evidence review of standard toxicity study data, T-cell dependent antibody response, cell-based assays that assess specific immune functions, host resistance assays, etc.). However, the risk of carcinogenicity due to immunosuppression is more difficult to predict. A 2010 workshop conducted by the Health and Environmental Sciences Institute (HESI) Immunotoxicity Technical Committee (ITC) explored some of the tools available to identify immunosuppression and latent virus recrudescence as well as challenges in predicting virus-related carcinogenicity using nonclinical models (Kawabata et al. 2012).

Findings in Toxicity Studies That Suggest Immunostimulation

In addition to the findings previously discussed under “infusion reactions,” immunostimulation may be evident from clinical signs, hematology, or histopathology. Interpretation requires the integration of all of the data generated in a study; individual parameters are generally meaningless alone. Clinical signs, including fever, anorexia, rash, or other signs consistent with abnormalities in organ systems, may be seen with immunostimulation. The most common hematologic evidence of immunostimulation is leukocytosis involving one or more cell lineages; the cells involved may be helpful in discerning the mechanism. Histopathology findings include inflammatory infiltrates in tissues. When there is evidence of cell damage along with the infiltrates, it may be challenging to differentiate immunostimulation that is causing the tissue damage from direct cell or organ toxicity that stimulates an inflammatory response. Determining cell types, including lymphocyte subsets, present in the infiltrate may help determine the mechanism or correlate the finding with known pharmacology. Identifying complement or immune complexes in tissues where an “inflammatory” or “immune” response is suspected may also be useful in determining mechanism. A search for organisms may reveal that immunosuppression is the real concern and the infiltrates are the response to infection. Drug substance accumulation in macrophages or increases in macrophages in the presence of drug (such as alveolar macrophage accumulation in inhalation toxicology studies) may simply be a normal response and not a direct “immunostimulation,” and is unlikely to be adverse unless accompanied by necrosis or other evidence of inflammation. However, these findings may prompt the need for macrophage function studies to ensure there is no evidence of functional impairment or enhancement. Additionally, increased bone marrow cellularity can occur in a number of situations, including immunostimulation. This finding must be interpreted in conjunction with peripheral blood hematology data, and unexplained hypercellularity should prompt a cytologic evaluation of the bone marrow to determine the lineage(s) involved and provide clues as to mechanism (Evans 2008).

Lymphoid hyperplasia in spleen, bone marrow, or lymph nodes is suggestive of immunostimulation. In some cases, there may be direct stimulation by the drug, triggering proliferation or release of cytokines. Lymphoid hyperplasia may be secondary to certain infections or changes in neutrophil function or trafficking, which result in increased exposure to aberrant antigens such as oral or intestinal commensal or pathogenic organisms, which are normally kept in check by innate immune mechanisms. When lymphoid hyperplasia is seen in monkeys, the potential for lymphocryptovirus activation due to immunosuppression should be strongly considered and investigated.

Risk Assessment Strategy

Immediate, severe immunostimulation can be life-threatening. Even less serious, transient events such as pseudoallergy and milder forms of cytokine release syndrome can be uncomfortable for patients and disconcerting for health care providers. Long-term direct immunostimulation carries at least theoretical concerns (not all of which are supported by reports in the literature) such as accelerated destruction of normal red blood cells by overactive macrophages, autoimmune disorders, an increase in allergic responses, myeloproliferative disorders and neoplasia, and exhaustion of constantly stimulated cells (Brennan et al. 2010; Wherry et al. 2007; EMEA 2006). It is not generally clear whether immunostimulation per se will progress to these conditions, but in conjunction with other factors, immunostimulation could potentially exacerbate a preexisting condition, particularly in the case of autoimmune diseases. While some information can be obtained from chronic and carcinogenicity toxicology studies related to long-term effects on bone marrow and hematopoiesis, there are no adequate tools to predict autoimmunity or other theoretical or multifactorial effects in humans.

As with any finding in a nonclinical study, certain fundamental questions need to be answered whenever an observation suggestive of immunostimulation is observed. Is the finding drug-related? If so, is it part of intended pharmacology? Is it a direct effect, secondary to another effect, or an adaptive response? In other words, does the drug directly stimulate the immune system, or is it actually suppressing an immune response, and are other pathways responding to infection or being activated to compensate for the blocked pathway? Is it adverse? Is it reversible? Is it relevant to the intended patient population? If the target is not fully understood, or downstream or alternative pathways have not been fully characterized, it may be difficult to answer these questions.

Early in a program, determining the relevance to human risk of any findings of immunostimulation within nonclinical studies may not be simple since the consequences of long-term dosing may not be known, and determining if a finding is “adverse” may be challenging. For example, in the case of chronically stimulated neutrophil production, will a new steady state be reached with no progression and no adverse consequences, will chronic stimulation lead to bone marrow “exhaustion,” or is there a potential for hematopoietic neoplasia or myeloproliferative disorders when normal mechanisms that keep proliferation in check are circumvented or overridden? It may be difficult to determine what is truly adverse until chronic and carcinogenicity studies are completed. While the finding of increased neutrophil production, lymphocytosis, and accumulation of granulocytes or mononuclear cells in tissues without any evidence of necrosis, infection, or clinical signs may not seem to have adverse consequences to the animal in the context of a toxicology study, this call is difficult to make without understanding the mechanisms involved. Further, animal data may not be predictive of human risk due to differences in immune responses. Because of considerable variation between species in the mechanics of immune responses, and the increasing specificity of targets for pharmaceutical intervention, it is extremely important to understand the differences between animal efficacy and toxicology models and humans. Of particular note was the failure of monkey studies to predict adverse cytokine release in humans resulting from administration of the CD28 agonist TGN1412; current hypotheses suggest that monkey T cells may have inhibitory mechanisms not present in humans or that the effector cells are CD28+ in humans but CD28− in monkeys (Brennan et al. 2010; Pallardy and Hunig 2010; Nguyen et al. 2006; Eastwood et al. 2010).

Summary

Immunostimulation per se is not a reason to terminate a drug development program. It can be a beneficial or intended effect, for example in a vaccine program or immune response directed against tumor cells where the goal should be control, contain, or direct the response to the target. Also, unintended immunostimulation does not necessarily rule out further development. To allow better understanding and management of immunostimulation, findings in toxicity studies suggesting immunostimulation should be investigated and integrated with other study data, along with what is known about the biology of the drug and drug target and any previous studies or work in other species. Additional diagnostic or functional studies may be required to understand the mechanism, species specificity, and relevance to human risk, and to provide means of monitoring safety in the clinic. Finally, if the immunostimulation is related to the structure or immunoglobulin class of the molecule, modifications may be made to create a molecule with less immunostimulation liability.