Abstract

The regulation of drug-metabolizing enzymes and transporters by cytokines has been extensively studied in vitro and in clinic. Cytokine-mediated suppression of cytochrome P450 (CYP) or drug transporters may increase or decrease the systemic clearance of drug substrates that are primarily cleared via these pathways; neutralization of cytokines by therapeutic proteins may thereby alter systemic exposures of such drug substrates. The Food and Drug Administration recommends evaluating such clinical drug interactions during clinical development and has provided labeling recommendations for therapeutic proteins. To determine the clinical relevance of these drug interactions to dose adjustments, trends in steady-state exposures of CYP-sensitive substrates coadministered with cytokine modulators as reported in the University of Washington Drug Interaction Database were extracted and examined for each of the CYPs. Coadministration of cytochrome P450 family 3 subfamily A (CYP3A) (midazolam/simvastatin), cytochrome P450 subfamily 2C19 (omeprazole), or cytochrome P450 subfamily 1A2 (caffeine/tizanidine) substrates with anti-interleukin-6 and with anti-interleukin-23 therapeutics led to changes in systemic exposures of CYP substrates ranging from ∼ –58% to ∼35%; no significant trends were observed for cytochrome P450 subfamily 2D6 (dextromethorphan) and cytochrome P450 subfamily 2C9 (warfarin) substrates. Although none of these changes in systemic exposures have been reported as clinically meaningful, dose adjustment of midazolam for optimal sedation in acute care settings has been reported. Simulated concentration-time profiles of midazolam under conditions of elevated cytokine levels when coadministered with tocilizumab, suggest a ∼six- to sevenfold increase in midazolam clearance, suggesting potential implications of cytokine–CYP drug interactions on dose adjustments of sensitive CYP3A substrates in acute care settings. Additionally, this article also provides a brief overview of nonclinical and clinical assessments of cytokine–CYP drug interactions in drug discovery and development.

SIGNIFICANCE STATEMENT There has been significant progress in understanding cytokine-mediated drug interactions for CYP-sensitive substrates. This article provides an overview of the progress in this field, including a trend analysis of systemic exposures of CYP-sensitive substrates coadministered with anti-interleukin therapeutics. In addition, the review also provides a perspective of current methods used to assess these drug interactions during drug development and a focus on individualized medicine, particularly in acute care settings.

Introduction

Cytokines are a group of glycoproteins predominantly produced by T-cells, macrophages, and B-cells. In some instances, mast cells, fibroblasts, and endothelial cells may also produce cytokines. These are key mediators of inflammation and may be triggered by pathogens, cancers, autoimmune conditions, or in some cases drug therapies. Cytokines have been classified into tumor necrosis factors (TNFs), interleukins (ILs), lymphokines, interferons (IFNs), colony-stimulating factors (CSFs), and transforming growth factors and are numbered based on the cell type producing them: type 1 cytokine produced by complement-dependent T-helper cells subtype 1 (e.g., IL-2, IL-12, IL-23, IFN-γ, TNF-α, and TNF-β), type 2 cytokine produced by complement-dependent T-helper cells subtype 2 (e.g., IL-4, IL-5, IL-6, IL-10, and IL-13), IL-9 and IL-10 produced by Th9, and IL-17 subtypes produced by Th17. Cytokines may be proinflammatory (IL-6, IL-12, IL-17, IL-23, TNF-α, IFN-γ, etc.) or anti-inflammatory (e.g., IL-1, IL-9, IL-10) (Tang et al., 2012; Sallusto, 2016; Fajgenbaum and June, 2020; Liu et al., 2021). Endogenous cytokine release has been reported under several disease conditions including a broad range of inflammatory disorders (Megha et al., 2021), infections (Kim et al., 2021; Tang et al., 2021), cancer (Grivennikov et al., 2010), and organ impairment (Albillos et al., 2014; Chen et al., 2022). Exogenous cytokine release caused by administration of novel investigational drugs gained significant attention after healthy volunteers (N = 6) dosed with CD28 superagonist TGN1412 showed a systemic inflammatory response and elevation of proinflammatory cytokines accompanied by clinical manifestations and hospitalization (Suntharalingam et al., 2006). A fatty acid amide hydroxylase inhibitor, BIA10-2474, caused a similar cadence of events during first-in-human (FIH)/dose escalation studies that led to neurologic events and the death of one subject in the trial (Rocha et al., 2022). Under disease conditions, the production of proinflammatory cytokines may result in persistent inflammation and tissue damage (Floege et al., 2012). In addition, uncontrolled release of proinflammatory cytokines can be life-threatening (Fajgenbaum and June, 2020) and can cause organ infiltration followed by tissue damage, multiorgan failure, and eventually sepsis. To minimize the likelihood of such events, cytokine release is often monitored during early stages of development of immunomodulators (Frey and Porter, 2019; Cosenza et al., 2021). Apart from safety concerns associated with cytokine release, regulation of drug-metabolizing enzymes and transporters has received significant attention based on the seminal work conducted by (Morgan, 1993); early reports by the authors have shown suppression of multiple cytochrome P450 (CYP) mRNAs in female rats treated with endotoxins. This initial work led to an extensive study of the mechanisms underlying CYP regulation and clinical impact on molecules that were primarily cleared by CYPs, particularly under conditions of inflammation and disease (Morgan, 1997). During recent years, rapid advances have been made in the development of immunotherapeutics to treat inflammation and cancer (McCune, 2018). This has led to an increased interest in understanding the clinical relevance of cytokine-mediated drug interactions and their implications on the safety and efficacy of CYP substrates. During the COVID-19 pandemic, these drug–disease interactions, and their impact on the pharmacokinetics (PK) of COVID-19 therapies, gained significant attention. This was especially important in critical care settings where patients may be at a risk of developing a cytokine storm, and such interactions could impact the effectiveness or safety of concomitantly administered therapeutics (Deb and Arrighi, 2021; Pilla Reddy et al., 2023). This mini-review summarizes the historical perspectives from the work in this field conducted by Morgan and colleagues and advances made by consortia and working groups that have culminated into guidance for industry and sponsors. In addition, a trend analysis conducted using the University of Washington Drug Interaction Database (UW DIDB) provides a snapshot of the changes in systemic exposures because of such interactions reported to date. Clinical relevance of the observed trends has been contextualized to critical care settings using recent case reports, to assess the potential impact of such drug interactions on dose adjustments of CYP substrates. Finally, we discuss the potential drug interactions that may occur when CYP substrates are coadministered alongside cell therapies that may carry a risk of cytokine release. In addition, research provides insights into how cytokine-mediated PK drug interactions can be assessed during drug development using an integrated approach of clinical studies and model-informed approaches.

Historical Perspectives

In human hepatocytes, proinflammatory cytokines including IL-6, TNF-α, and IFN-γ have been found to downregulate cytochrome P450 family 3 subfamily A (CYP3A), CYP2C, and CYP2B6 mRNA levels (Morgan, 1997; Aitken and Morgan, 2007); similarly, tumor growth factor-β has been found to downregulate CYP2C and CYP2B6 while no effect of cytokines has been reported on the mRNA or protein levels of cytochrome P450 subfamily 2D6 (CYP2D6). At a transcriptional level, this regulation occurs via nuclear and hormone receptors (Wu and Lin, 2019). The clinical relevance of this phenomenon on the PK of CYP-sensitive substrates has been discussed previously (Reiss and Piscitelli, 1998; Stipp and Acco, 2021). Changes in systemic exposures under different levels of inflammation have been illustrated by Harvey and Morgan (2014) using computational approaches. These simulations predicted an increase in systemic exposures of CYP substrates under disease conditions, with a further increase in exposures following treatment with immunomodulators when compared with healthy individuals. Additionally, these simulations predicted a subsequent reduction in exposure of CYP substrates due to neutralized or reduced levels of proinflammatory cytokines. These in silico predictions were eventually validated in the clinic by interactions reported between tocilizumab, an anti-IL-6 monoclonal antibody (mAb), and simvastatin, a sensitive CYP3A substrate (Schmitt et al., 2011). Similarly, Coutant and Hall (2018) have extensively reviewed the disease states that have led to nonstationary kinetics of CYP-sensitive substrates under disease conditions. With significant progress made in immunotherapies and immune checkpoint inhibitors, much attention has been drawn toward the release or neutralization of cytokines during such drug therapies and its clinical relevance on the PK and pharmacodynamics of CYP substrates due to changing inflammatory tone. During drug development, such complex interactions may also impact the design of combination regimens, where an immunomodulator may be combined with a CYP substrate to maximize efficacy or minimize safety concerns of the combination partners or both. To assess the likelihood of such interactions, recommendations for best practices are in place (Kraynov et al., 2011; Evers et al., 2013; Yu et al., 2023). In 2013, the International Consortium of Quality and Innovation and Food and Drug Administration (FDA) developed a risk assessment to determine if a dedicated clinical drug interaction study was needed for investigational therapeutic proteins during drug development (Kenny et al., 2013). The FDA released its first draft guidance on considerations for conducting clinical drug–drug interaction (DDI) studies to assess therapeutic protein–drug interactions and labeling recommendations (Huang et al., 2010; Kenny et al., 2013). Over the years, through clinical trial observations or using population-based approaches, an understanding of cytokine-mediated PK DDIs has evolved (Machavaram et al., 2013; Khatri et al., 2019; Sathe et al., 2021).

Recent Advances

Most recently, an International Consortium of Quality and Innovation white paper has summarized various mechanisms of therapeutic protein–drug interactions of clinical concern and has suggested that clinical drug interaction studies are not required for all therapeutic proteins or in patient populations with low inflammatory burden (Coutant et al., 2023). In the current landscape of drug development, cell therapies such as bispecifics and chimeric antigen receptor T-cell therapies are emerging. These therapies have the potential to release proinflammatory cytokines, the levels of which may vary depending on disease state, especially in critical care settings. Predicting the changes in the PK and thereby the optimal dose of concomitant medications or supportive therapies that are sensitive CYP substrates (e.g., corticosteroids, pain agents, sedatives, etc.) could be challenging under such settings.

To investigate the trends in cytokine-mediated CYP drug interactions, we queried the UW DIDB for sensitive substrates (FDA, 2020) of CYP3A (simvastatin, midazolam), CYP2D6 (dextromethorphan), cytochrome P450 subfamily 2C19 (omeprazole), cytochrome P450 subfamily 2C9 (warfarin), and cytochrome P450 subfamily 1A2 (caffeine, tizanidine) when coadministered with broad class of immunomodulators. The sensitive CYP substrates were queried as “objects” and the therapeutic classes of “anti-inflammatory” and “immunomodulators” were queried as “precipitants,” “non-precipitants,” and “other” mechanisms within the UW DIDB (Fig. 1). All precipitants, non-precipitants, and other therapeutic drug classes that were noncytokine modulators were filtered out. The remaining pairs were merged. The percent change in area under the curve (AUC) (Fig. 2) for each of the pairs was plotted by CYP substrates. Overall, percent change in AUC ranged from approximately –58% to up to 35%. Minimal changes were observed for CY1A2 substrates with the largest observed percent change in AUC of caffeine (Fig. 2, Panel 2A) in the presence of the anti-IL-23 antibody risankizumab (Aitken and Morgan, 2007). Similarly, coadministration of omeprazole, 20 mg, single dose with sirukumab, 300 mg, single dose (Fig. 2, Panel 2B) resulted in up to 48% decrease in AUC of omeprazole (Zhuang et al., 2015). As observed in Fig. 2, Panel 2C, coadministration of anti-IL-6 mAbs (tocilizumab, sirukumab, sarilumab), anti-IL-17A mAb (brodalumumab), and anti-IL-23 mAb (risakizumab) with sensitive CYP substrate(s) resulted in significant changes in AUC of these substrates; the highest magnitude of percent change in AUC was reported for simvastatin, 40 mg, single dose (decrease in AUC by 58%) when coadministered with intravenous tocilizumab, 800 mg (Schmitt et al., 2011); in contrast, the anti-IL-17A antibody broadalumab, 210 mg single dose s.c., increased the AUC of midazolam, 2 mg, administered orally, by ∼30%. This small magnitude of change in systemic exposures of midazolam with brodalumumab is not anticipated to cause sedation in these participants; interestingly, it is directionally opposite to the in vitro observations (Aitken and Morgan, 2007) and is attributed to increased serum IL-17A levels due to administration of broadalumab (Roman and Chiu, 2017). There are no known drug interactions of cytokine modulators with CYP2D6 sensitive substrates in the UW DIDB, which is consistent with in vitro assessments made by Morgan and colleagues. Based on the collective analysis, clinically meaningful changes in systemic exposures of sensitive CYP substrates in the presence of cytokine modulators appear to be of low incidence; no exposure response or significant changes to the overall safety and efficacy of these CYP substrates have been reported to date.

Fig. 1.

Flow diagram to extract PK drug interaction dataset of sensitive CYP substrates coadministered with cytokine modulators using the UW DIDB.

Fig. 2.

Percent change in AUC of substrates of cytochrome P450 subfamily 1A2 [caffeine (CAFF)] in Panel 2A, cytochrome P450 subfamily 2C19 [omeprazole (OMP)] in Panel 2B, and CYP3A [simvastatin (SIMVA) or midazolam (MDZ)] in Panel 2C, when coadministered with cytokine modulators.

While the current analysis suggests no clinically meaningful drug interactions of cytokines with sensitive CYP substrates, recent case reports suggest that potential interactions and implications to dose adjustments of sensitive CYP substrates cannot be ruled out. Mefford et al. (2022) have recently reported that five critically ill COVID-19 patients with acute respiratory distress syndrome required higher doses of intravenous midazolam to maintain sedation goals. These patients were administered tocilizumab 800 mg s.c. (one to two doses administered over 3 to 15 days) to manage cytokine release. The optimal sedation dose range of intravenous midazolam in healthy volunteers is 0.02 to 0.1 mg/kg/h (Barr et al., 2013). However, in this case study, patients required a dose of 0.15 to 0.68 mg/kg/h of intravenous midazolam for optimal sedation. Assuming that the higher dose of midazolam required for optimal sedation was due to increased systemic clearance of midazolam resulting from restored CYP3A activity due to neutralization of IL-6 by tocilizumab, we investigated the magnitude of drug interaction using the simulation approach. To assess this drug interaction between tocilizumab and midazolam under conditions of cytokine release, we simulated concentration-time profiles of intravenous midazolam at varying systemic clearance and at fixed volume of distribution (USPI). Changes in midazolam clearance in the presence of tocilizumab, 800 mg s.c., were assessed as summarized in Fig. 3 and Table 1, using midazolam, i.v. 0.1 mg/kg/h PK parameters (USPI) in healthy, adult populations [body weight = 65 kg, one compartment (1C) PK model, clearance = 0.49 L/kg/h; volume of distribution = 1.44 L]. Assuming no change in volume of distribution, the 1C PK model predicted an increase in systemic clearance of midazolam to account for the dose of 0.68 mg/kg/h (average concentration achieved in plasma of ∼204 ng/mL) that was required to achieve optimal sedation in these patients. Overall, these simulations suggested a ∼six- to sevenfold increase in systemic clearance of midazolam to explain the ∼sixfold dose adjustment reported in these patients to achieve sedation (Table 1). Similarly, Günes and colleagues (Güneş et al., 2020) have recently reported a potential for drug interaction between warfarin and tocilizumab, with no changes in coagulation parameters or any increases in bleeding risks. Neutralization of IL-6 by tocilizumab may restore cytochrome P450 subfamily 2C9 levels, increasing the systemic clearance of warfarin. Such interactions may be consequential to dosing decisions for warfarin due to the narrow therapeutic range required for maintenance of the international normalized ratio. As no pharmacokinetic data on warfarin was available in this case report, we are unable to assess the magnitude of drug interaction.

Fig. 3.

Simulations of concentration-time profile of midazolam, intravenous infusion with or without tocilizumab, to predict systemic clearance of midazolam at different dose levels. Midazolam (MDZ) (IVCL, 0.1 mg/kg/h, red): predicted concentration-time profile of MDZ at 0.1 mg/kg/h using 1C PK model, and intravenous clearance (CL) of 0.49 L/kg/h; MDZ (IVCL, 0.68 mg/kg/h, blue): predicted concentration-time profile of midazolam at 0.68 mg/kg/h using 1C PK model, and intravenous CL of 0.49 L/kg/h; tocilizumab (TOCI) + MDZ (IVCLadj, 0.68 mg/kg/h, orange): predicted concentration-time profile of MDZ at 0.68 mg/kg/h, by simulating its systemic clearance to achieve exposures comparable to those at 0.1 mg/kg/h; a 6.8X higher systemic clearance of MDZ was required to achieve this exposure equivalence.

TABLE 1

Summary of PK parameters of IV midazolam to predict steady-state plasma concentration as shown in Fig. 3, at 0.1 mg/kg/h and 0.68 mg/kg/h, assuming body weight = 65 kg of an average individual

Perspectives on Future Directions and Key Challenges

As there is emerging interest in the development of cell therapies like bispecifics and chimeric antigen receptor T-cells, which have the potential to release cytokines during drug therapy (Supplemental Table 1), cytokine-mediated drug interactions and their impact on safety, efficacy, and dose adjustment of concomitantly administered sensitive CYP substrates will continue to be of interest, especially in critical care settings. Currently, in vitro assays in whole blood or peripheral blood mononuclear cells provide a limited, qualitative assessment of an investigational agent to release cytokines (Finco et al., 2014); for cell therapies, prior to the start of FIH studies, assessment of cytokine release is conducted in nonhuman primates, as they closely mimic the human innate immune system (Taraseviciute et al., 2016). The initial indication of cytokine release in these nonclinical models could be used as a guide to generate a preliminary assessment of cytokine-mediated drug interactions using human hepatocyte models that have been previously developed (Aitken and Morgan, 2007). However, these in vitro and nonclinical models are not designed to quantitatively predict clinical outcomes or to advise dose adjustments in the clinic. Therefore, this nonclinical assessment of cytokine–CYP drug interactions requires exploration during early clinical development. During FIH dose escalation studies of immunomodulators and cell therapies, the time of onset, magnitude, and duration of cytokine release can be measured; doses may be optimized using step-up dosing (Hosseini et al., 2020) as exemplified by mosentuzumab (Budde et al., 2022) and glofitamab (Carlo-Stella et al., 2021). Cytokine release following these therapeutic modalities is often transient, occurring during the first 24 to 72 hours post-first dose of the drug; thereby, it is important to collect the duration and magnitude of cytokine release for purposes of subsequent in vitro or computational assessments. Once an optimal dose or dose range of the therapeutic modality has been identified, all available information on the dose range/exposures and cytokine release profile (magnitude and duration) over the intended dose range should be collected. Using this information, the cytokine-mediated drug interaction with concomitantly administered standard-of-care supportive therapies that may be sensitive CYP substrates could be evaluated using a physiologically based PK approach (e.g., mosentuzumab) (Chen et al., 2023). This may enable the prediction of cytokine-mediated drug interactions with CYP substrates/inhibitors/inducers typically administered in intended patient populations, especially during later stages of drug development. The recently updated FDA guidance provides a risk-based approach to assess the need for clinical drug–drug interaction studies of therapeutic proteins under investigation (FDA, 2023). While population-based variability associated with cytokine release may not allow for an accurate prediction of cytokine-mediated drug interactions, recently such approaches have been used successfully to design clinical DDI studies (Xu et al., 2015; Jiang et al., 2016; Sathe et al., 2021) and could aid in labeling recommendations. Most recently, a systems-based approach has been described for midazolam, where a population-based PK model of midazolam in critically ill patients (neonates, infants, children, and adults) has been developed using baseline C-reactive protein levels and using organ failure as covariates on systemic clearance (Vet et al., 2016; Brussee et al., 2018). In acute care settings, a similar systems-based approach could be used to evaluate dose adjustments of CYP substrates and minimize any impact on efficacy and safety toward personalized dosing (Mefford et al. (2022)).

In conclusion, this analysis suggests that the overall incidences of PK drug interactions under clinical conditions of cytokine release are relatively low; while the magnitude of anti-IL-6 and anti-IL-23 with CYP3A and cytochrome P450 subfamily 2C19 appears to be ∼40% to 60% (decrease in AUC of sensitive CYP substrates), the clinical relevance to overall safety, efficacy, and dose adjustments appears low. In clinical settings where high levels of IL-6 or IL-23 or high interindividual variability in these cytokines may be prevalent, leveraging model-informed approaches to support a precision dosing approach may be required for optimal dosing.

Acknowledgments

The authors would like to sincerely thank Eddie Morgan for his expertise and discussions during the development of the work.

Data Availability

The data extracted for this study are openly available in the University of Washington Drug Interaction Database or UW DIDB. All other data presented are contained within the manuscript/Supplemental Material.

Authorship Contributions

Participated in research design: Sawant-Basak, Olabode.

Conducted experiments and performed data analysis: Sawant-Basak, Olabode.

Wrote or contributed to the writing of the manuscript: Sawant-Basak, Olabode, Dai, Vishwanathan, Phipps.

FootnotesReceived September 2, 2023.Accepted February 8, 2024.

This work received no external funding.

No author has an actual or perceived conflict of interest with the contents of this article.

dx.doi.org/10.1124/dmd.123.001499.

↵This article has supplemental material available at dmd.aspetjournals.org.

Abbreviations1Cone compartmentAUCarea under the curveCYPcytochrome P450CYP3Acytochrome P450 family 3 subfamily ACYP2D6cytochrome P450 subfamily 2D6DDIdrug-drug interactionFDAFood and Drug AdministrationFIHfirst-in-humanIFNinterferonILinterleukinmAbmonoclonal antibodyPKpharmacokineticsTNFtumor necrosis factorUW DIDBUniversity of Washington Drug Interaction DatabaseCopyright © 2024 by The Author(s)