Evaluation of the potential impact on pharmacokinetics of various cytochrome P450 substrates of increasing IL‐6 levels following administration of the T‐cell bispecific engager glofitamab

Abstract

Glofitamab is a novel T cell bispecific antibody developed for treatment of relapsed‐refractory diffuse large B cell lymphoma and other non‐Hodgkin's lymphoma indications. By simultaneously binding human CD20‐expressing tumor cells and CD3 on T cells, glofitamab induces tumor cell lysis, in addition to T‐cell activation, proliferation, and cytokine release. Here, we describe physiologically‐based pharmacokinetic (PBPK) modeling performed to assess the impact of glofitamab‐associated transient increases in interleukin 6 (IL‐6) on the pharmacokinetics of several cytochrome P450 (CYP) substrates. By refinement of a previously described IL‐6 model and inclusion of in vitro CYP suppression data for CYP3A4, CYP1A2, and 2C9, a PBPK model was established in Simcyp to capture the induced IL‐6 levels seen when glofitamab is administered at the intended dose and dosing regimen. Following model qualification, the PBPK model was used to predict the potential impact of CYP suppression on exposures of various CYP probe substrates. PBPK analysis predicted that, in the worst‐case, the transient elevation of IL‐6 would increase exposures of CYP3A4, CYP2C9, and CYP1A2 substrates by less than or equal to twofold. Increases for CYP3A4, CYP2C9, and CYP1A2 substrates were projected to be 1.75, 1.19, and 1.09‐fold following the first administration and 2.08, 1.28, and 1.49‐fold following repeated administrations. It is recommended that there are no restrictions on concomitant treatment with any other drugs. Consideration may be given for potential drug–drug interaction during the first cycle in patients who are receiving concomitant CYP substrates with a narrow therapeutic index via monitoring for toxicity or for drug concentrations.

Study Highlights.

• WHAT IS THE CURRENT KNOWLEDGE ON THE TOPIC?

Elevated IL‐6 concentrations can lead to reduced CYP metabolism in vitro and in vivo. Physiologically‐based pharmacokinetic (PBPK) modeling offers the possibility to simulate this effect and predict the clinical outcome.

• WHAT QUESTION DID THIS STUDY ADDRESS?

This study outlines use of PBPK modeling to assess the impact of glofitamab‐associated transient increases in IL‐6 on selected CYP substrates.

• WHAT DOES THIS STUDY ADD TO OUR KNOWLEDGE?

PBPK analysis predicts that, in the worst case, glofitamab administration may lead to exposure increases of less than twofold for CYP3A4, CYP2C9, and CYP1A2 substrates.

• HOW MIGHT THIS CHANGE DRUG DISCOVERY, DEVELOPMENT, AND/OR THERAPEUTICS?

Development of therapeutic proteins can benefit from PBPK modeling to project the impact IL‐6 changes on exposures of co‐administered CYP substrates.

INTRODUCTION

The use of physiologically‐based pharmacokinetic (PBPK) modeling to predict drug concentrations in plasma and tissue has demonstrated utility for accelerating pharmaceutical development, and now is an integral part of many drug development programs. ¹ , ² , ³ , ⁴ , ⁵ Advancement in the discipline has been followed by increasing acceptance by regulatory authorities, ³ , ⁵ , ⁶ , ⁷ , ⁸ , ⁹ and there are now numerous examples of drug approvals supported by PBPK modeling in lieu of in vivo clinical studies. ¹⁰ , ¹¹ , ¹² , ¹³ One area where PBPK modeling is particularly widely used is the prediction of drug–drug interactions (DDIs), in part because it allows quantitative predictions in complex scenarios, for example, simultaneous induction and inhibition, multiple perpetrators, etc. ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ Understanding the clinical consequences of such interactions has been further facilitated by the development of models which simultaneously predict effects on multiple pharmacologically active species. ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² , ²³

Cytokines, such as IL‐6, have been shown to suppress the activity of CYP enzymes in vitro. ²⁴ , ²⁵ It has been shown that IL‐6 concentrations are raised in inflammatory disorders, such as rheumatoid arthritis (RA), post‐surgery, and during infection, leading to reductions in CYP‐mediated drug elimination. ²⁶ , ²⁷ This has also been reported in patients with neuromyelitis optica (NMO). ²⁸ Similarly, it has been shown that in patients with raised IL‐6 concentrations, administration of an anti‐IL‐6 therapy can decrease the exposure of co‐administered CYP substrates due to the removal of cytokine mediated CYP suppression. ²⁹ , ³⁰ Modeling and simulation can be especially valuable to quantify these effects. For example, PBPK modeling was recently used to assess the potential disease‐drug interactions when anti‐IL‐6 therapy Tocilizumab reverses CYP suppression caused by raised IL‐6 concentrations in patients with NMO and NMO spectrum disorders. ³¹

Glofitamab is a novel T cell bispecific (TCB) antibody being developed for the treatment of relapsed‐refractory diffuse large B cell lymphoma and other non‐Hodgkin's lymphoma (NHL) indications. Glofitamab has a “2:1” molecular format comprising two fragment antigen binding regions that bind CD20 (on the surface of B cells) and CD3 (on the surface of T cells). It is based on the human IgG1 isotype, but contains an Fc part without cγR and C1q binding, thereby preventing FcγR‐mediated co‐activation of innate immune effector cells, NK cells, monocytes/macrophages, and neutrophils. By simultaneously binding to human CD20‐expressing tumor cells and to the CD3 on T cells, glofitamab induces tumor cell lysis, in addition to T cell activation, proliferation, and cytokine release.

More generally, TCBs are an important class of antibody therapeutics in immuno‐oncology which activate T cells and redirect their cytotoxicity against tumor cells. Cytokine release syndrome (CRS), a systemic inflammatory response driven by cytokine release, is a common dose‐limiting adverse event for TCBs. CRS is mainly a first‐administration adverse event, observed to a lesser extent in following administrations. A “priming” dose strategy (i.e., a lower initial dose followed by a higher maintenance dose, also called step‐up dosing) has been implemented in the clinic to mitigate CRS and to achieve efficacious doses with TCBs, including glofitamab. ³² The transient elevation of cytokines, mainly IL‐6, IL‐10, and IFN‐c, has been observed within the first 48 h of glofitamab administration, as previously described for other constructs, such as blinatumomab. ³³

PBPK modeling can be utilized to simulate cytokine profiles following various dosing regimens and may assist the design of clinical dosing strategies for TCBs. The suppression of CYP enzymes by cytokines at physiologically relevant concentrations has been studied in vitro using hepatocytes. ²⁵ CYP3A4, CYP1A2, and CYP2C9 are reported as the enzymes most sensitive to cytokine suppression and IL‐6 is the most potent CYP suppressor among the cytokines tested. ²⁵ , ³³ In vitro CYP suppression data has been utilized in PBPK models to predict the impact of raised IL‐6 concentrations on CYP activity in vivo and to study the effect of administration of anti‐IL‐6 therapies in patients concomitantly exposed to small molecules. ³³ , ³⁴

Here, we describe a PBPK model building and qualification process aiming to assess the DDI impact of transiently increased IL‐6 levels on the pharmacokinetics (PKs) of several CYP probe substrates, such as simvastatin and midazolam (CYP3A4 substrates), caffeine, and theophylline (CYP1A2 substrates), and S‐warfarin (CYP2C9 substrate) in a virtual population with demographic characteristics matching those of patients included in the glofitamab NHL population from the NP30179 pivotal study. ³⁵

METHODS

Model development

The PBPK model development, qualification, and simulation strategy used to assess the DDI potential of glofitamab is depicted in Figure 1.

FIGURE 1.

Glofitamab‐induced IL‐6 PBPK model building steps. DDI, drug–drug interaction; PBPK, physiologically‐based pharmacokinetic.

First, the in vitro CYP suppression data for CYP3A4, CYP1A2, and 2C9 from Dickmann et al. ²⁵ were included into an IL‐6 compound file previously developed for a different patient population. ³¹ , ³⁶ Then, the preliminary PBPK model was refined via a sensitivity analysis on the PK parameters and their variability in order to capture the IL‐6 kinetics (i.e., a transient IL‐6 increase) observed in the patients with NHL of study NP30179 following the glofitamab 2.5/10/30 mg step‐up dosing regimen. A second model refinement focused on a sensitivity analysis of the IL‐6 input parameters (i.e., infused dose and infusion duration) to capture the “high,” “median,” and “low” IL‐6 peaks observed in the 2.5/10/30 mg step‐up dosing patients from the NP30179 study. The next step aimed at assessing the kinetics of intrinsic clearance via the CYP3A4, CYP1A2, and CYP2C9 enzymes following the increase in IL‐6 levels. Finally, the ultimate goal was to predict the potential impact of CYP suppression on the exposures (i.e., maximum plasma concentration [C _max] and area under the curve [AUC]) of CYP3A4, CYP1A2, and CYP2C9 probe substrates.

Model verification

First, IL‐6 plasma concentrations simulated in virtual North European White populations following an i.v. infusion at different dose levels were compared to observed data in patients with “low,” “medium,” or “high” IL‐6 concentrations. ³⁰ , ³⁶ The approach of providing three scenarios to cover IL‐6 levels was followed due to the large between‐subject variability (Figure 2) and the sparse sampling of this cytokine in patients. We therefore did not estimate exposure metrics derived from individual observed profiles, such as C _max or AUC, but chose to explore different scenarios with the “high” scenario being considered a worst case.

FIGURE 2.

Overlay of spaghetti plots of observations and predicted plasma concentration‐time profiles of IL‐6. Overlay of spaghetti plots of observations in the patients with glofitamab step‐up dosing regimen and median (5th–95th) predicted plasma concentration‐time profiles of IL‐6 (bold solid red line is median and red area is 5th–95th percentiles) during the first 17 days of glofitamab treatment representing high (a), medium (b) and low (c) increased IL‐6 profiles. Curves in blue represent observations for patients not treated with tocilizumab as tocilizumab reverses the CYP‐suppressing effect of IL‐6 as described in Schmitt et al. ³⁰

The impact of IL‐6 levels on CYP suppression and the effect of this suppression on CYP substrates has been verified previously ³¹ , ³³ and will not be covered in the present analysis.

Model application

The prediction of the impact of “high,” “medium,” and “low” increased IL‐6 concentrations in virtual patients was assessed with simvastatin (40 mg), midazolam (5 mg), caffeine (150 mg), theophylline (125 mg), or S‐warfarin (10 mg) following single and repeated oral dosing.

The virtual patients were generated to match the age, body weight, height, and gender demographics of the NHL 2.5/10/30 mg step‐up dosing patients of study NP30179.

Physiologically‐based pharmacokinetic platform

The Simcyp Simulator (version 20) was used (www.simcyp.com). A minimal PBPK model, which considers liver and intestinal metabolism, is incorporated into the software and was applied in all simulations of plasma concentrations of IL‐6.

The CYP probe substrate models used were standard library models as provided in the Simcyp Simulator (version 20) and were used without any modifications. These Simcyp library files are verified by SimCYP and documented in reports available to licensed users of the software (see Tables S1–S6 for details).

Enzyme dynamics and inhibition

Previous in vitro experiments with a CD19/CD3 bispecific construct ³³ showed cytokine suppression mainly on CYP1A2 (>50% of suppression from all donors) and on CYP3A4/5 and CYP2C9 (>50% of suppression in 2/3 donors). Negligible suppression was observed on CYP2C19 and CYP2D6. In the present analysis, the CYP enzyme suppression was incorporated into a semi‐mechanistic dynamic model to simulate the effects of IL‐6 on CYP3A4/5, 1A2, and 2C9‐mediated hepatic metabolism of probe substrates. ¹⁵ , ³¹

Unbound IL‐6 concentrations in the liver are used as the driving force for suppression and the changes in active enzyme levels (ENZ_act) are described using the following equation:

$$\frac{d{\mathrm{ENZ}}_{\mathrm{act}}}{\mathit{dt}}=k_{{\mathrm{deg}}_{\mathrm{enz}.\mathrm{act}}}*{\mathrm{ENZ}}_0\left(1+\frac{\left(E_{\min }-1\right)*{\left[I\right]}_t}{{\mathrm{EC}}_{50}+{\left[I\right]}_t}\right)-k_{{\mathrm{deg}}_{\mathrm{enz}.\mathrm{act}}}*{\mathrm{ENZ}}_{\mathrm{act}}$$ (2)

Where ENZ_act is the amount of active enzyme at any given time in the liver; ENZ₀ is the basal amount of the CYP isozyme in question; E _min is the minimum amount of active enzyme observed in the in vitro system (i.e., the maximum amount of suppression) expressed as a fraction of the vehicle control value; EC₅₀ is the concentration that supports half E _min (i.e., half of the maximal suppressive effect); [I]t is the free concentration of IL‐6 in the liver at time t; and k _deg is the enzyme degradation rate constant.

Intrinsic turnover (k _deg) of hepatic CYP3A4, 1A2, and 2C9 used in the simulations were 0.0193, 0.0183, and 0.0067 h⁻¹, respectively. ³⁷ , ³⁸ A key assumption in the model is that the unbound liver IL‐6 concentration is in rapid equilibrium with the unbound plasma IL‐6 concentration and that these concentrations result in the changes in the enzyme level via the equation outlined above.

To utilize the in vitro data for IL‐6 suppression within the Simcyp simulator, the minimum amount of active enzyme observed in the in vitro system (E _min) is entered on the inducer interaction as Ind_max for the relevant enzyme and the concentration that supports half E _min (i.e., half of the maximal suppressive effect; EC₅₀) is entered into the concentration of inducer that supports the half maximal induction/suppression (IndC₅₀).

Population data

Simulations in Simcyp use a population of virtual individuals generated with values describing demographic, anatomic, and physiological variables. In this study, the Simcyp version 20 default parameter values for a North European White population ³⁹ , ⁴⁰ , ⁴¹ were used except that the demographic characteristics (i.e., age and body weight) were matching those of the 163 patients receiving 2.5/10/30 mg step‐up dosing regimen in the NP30179 study (Table S1).

IL‐6 pharmacokinetic simulation plan

Physicochemical data, plasma protein binding, distribution, and clearance

The IL‐6 model was based on a previously published model ³¹ , ³⁶ but with some modifications made so that simulations captured the observed transient elevation of IL‐6 exposures following a 2.5/10/30 mg step‐up dosing regimen of glofitamab. Observed data were for 163 patients of NP30179 study (cohorts D2‐2 and D2‐4, D3, and D5) and simulations were run to achieve three different IL‐6 levels (high, median, and low) which encompassed the observations. The input value for IL‐6 volume of distribution at steady state (V _ss) of 0.43 L/kg was taken from Machavaram et al. ³¹ , ³⁶ and the clearance following an intravenous infusion dose (CL_i.v.) was set to 0.5 L/h, which is close to the value described by Xu et al. ³³ for blinatumomab, a CD19/CD3 bispecific in a cancer population. The associated percent coefficient of variation for both V _ss and CL_i.v. were increased from 30% to 100% to capture the high variability of IL‐6 profiles observed in the NP30179 study. The observed profiles in glofitamab step‐up dosing patients showed an IL‐6 peak at ~12 h post‐glofitamab dose, thus IL‐6 was introduced as a 12‐h i.v. infusion at a dose of 0.0002, 0.004, and 0.04 mg to achieve the low, medium, and high IL‐6 increase profiles, respectively. It was observed that the transient increase of IL‐6 concentrations in the 2.5/10/30 mg step‐up dosing patients was in the same range following 2.5 mg dose of glofitamab (first administration), 10 mg (second administration at day 8), and 30 mg (third administration at day 15). This immune tolerance or priming effect is expected with the step‐up dosing. Consequently, for the simulated level of IL‐6 increase (i.e., low, medium, and high), the same dose of IL‐6 was administered at each dosing event even though the observed IL‐6 levels after the second and third doses tended to be slightly lower, to be more conservative from a DDI point of view.

CYP suppression by IL‐6

Suppression of CYP3A4, 1A2, and 2C9 by IL‐6 in vitro was studied in hepatocytes. ²⁵ The E _min and EC₅₀ concentrations from this study were used in the simulations and are summarized in Table 1.

TABLE 1.

Input parameter values used for IL‐6.

Parameter	Value	Method/reference
Molecular weight (g/mol)	21,000	www.invivogen.com
log P	0.01	Assumed
Compound type	Neutral
B/P	1	Assumed
Fu	1	Assumed
Main plasma binding protein	Human serum albumin
Distribution model	Minimal PBPK model
V SS (L/kg)	0.43	Machavaram et al. 36 Machavaram et al. 31
CLi.v. (L/h)	0.5	Xu et al. 33
CLR (L/h)	0	Assumed
Enzyme	CYP1A2
E min	0.23	Dickman et al. 25
IndC50 (μM)	5.96E‐05	Dickman et al. 25
Enzyme	CYP2C9
E min	0.053	Dickman et al. 25
IndC50 (μM)	5.76E‐06	Dickman et al. 25
Enzyme	CYP3A4
E min	0.24	Dickman et al. 25
IndC50 (μM)	3.48E‐06	Dickman et al. 25
Enzyme	CYP3A5
E min	0.24	Same values used as CYP3A4; Dickman et al. 25
IndC50 (μM)	3.48E‐06	Same values used as CYP3A4; Dickman et al. 25

Abbreviations: CL_i.v., clearance following an intravenous infusion dose; CL_R, renal clearance; CYP, cytochrome; E _min, minimum amount of active enzyme observed in the in vitro system (i.e. the maximum amount of suppression) expressed as a fraction of the vehicle control value; IL‐6, interleukin 6; IndC₅₀, concentration of inducer that supports the half maximal induction/suppression; V _ss, volume of distribution at steady‐state.

PBPK model verification of IL‐6 concentrations

The model performance was verified by comparing the predicted IL‐6 profiles following the three dose levels (i.e., 0.0002, 0.004, and 0.04 mg, respectively) to those observed in the study NP30179 patients who did not receive tocilizumab (n = 124).

RESULTS

Input parameters for the IL‐6 PBPK model

The final parameters used for the IL‐6 compound file for simulation of IL‐6 PKs are shown in Table 1.

Model performance

Simulated to observed IL‐6 plasma concentration‐time profiles in 2.5/10/30 mg step‐up dosing patients

The median (5th–95th percentiles) predicted plasma concentration‐time profiles for IL‐6 following administration of intravenous infusion doses of IL‐6 at 0.0002, 0.004, and 0.04 mg weekly were overlaid with the observed IL‐6 concentrations in patients who did not receive tocilizumab (Figure 2a,b,c, respectively). These figures suggest that the three levels of exposures to IL‐6 generated in the 2.5/10/30 mg step‐up dosing patients represent well the clinical setting and can be used to assess the potential DDI risk depending on the associated increased IL‐6 level.

A different visualization of how the model captures the increased IL‐6 concentrations at the low, medium, and high dose levels is presented in Table 2. This table compares predicted IL‐6 concentrations achieved over the 24 h following the first, second, and third administrations to observed IL‐6 concentrations. The observations include only those patients not treated with tocilizumab (n = 124) as tocilizumab treatment leads to recovery of the CYP activity. ³⁰

TABLE 2.

Predicted and observed IL‐6 concentrations in pg/mL in patients treated with the 2.5/10/30 mg step‐up dosing regimen.

Administration	Predictions (pg/mL)			Observations (pg/mL)
	Median [min–max]			Median [min–max]
	Low dose (n = 163)	Medium dose (n = 163)	High dose (n = 163)	Without Toci (n = 124)
First administration (day 1)	5.03 [0.00–42.9]	101 [0.00–858]	1010 [0.00–8580]	30.2 [1.45–2670]
Second administration (day 8)	6.00 [0.00–42.9]	120 [0.00–857]	1200 [0.00–8570]	16 [1.62–2620]
Third administration (day 15)	6.32 [0.00–43.0]	126 [0.00–860]	1260 [0.00–8600]	7.32 [1.45–1740]

Note: Median [min–max] of predicted IL‐6 concentrations (pg/mL) achieved over the 24 h following the first, second, and third administration of 0.0002 (low), 0.004 (medium), and 0.04 mg (high) of IL‐6 and observed IL‐6 concentrations (pg/mL) after the first (2.5 mg), second (10 mg), and third (30 mg) glofitamab administration in the patients with 2.5/10/30 mg step‐up dosing regimen.

Abbreviations: IL‐6, interleukin‐6; Toci, tocilizumab.

As seen in Figure 2 and Table 2, most of the patients are within the range predicted by the low or the medium dose levels of IL‐6. However, some extreme values from a limited number of patients are better captured with the high IL‐6 dose level (Figure 2a).

Simulated in vivo suppression of CYP isozymes following administration of IL‐6

The change in CYP enzyme levels simulated in subjects administered intravenous infusion doses of IL‐6 at low, medium, and high dose levels are shown in Figure 3.

FIGURE 3.

Changes in mean CYP levels following administration of IL‐6 (0.0002, 0.004, and 0.04 mg) over the simulation period. Changes in mean CYP level following administration of IL‐6. (a) CYP3A4, (b) CYP1A2, and (c) CYP2C9. Data are for simulations without IL‐6 (solid black line), and with low 0.0002 mg (dashed green line), medium 0.004 mg (dashed orange line), and high 0.04 mg (dashed red line) IL‐6 doses.

These figures suggest that depending on the IL‐6 levels (i.e. low, medium, or high) and on the CYP isoform (i.e., CYP3A4, CYP1A2, or CYP2C9), the maximum CYP suppression occurs at times varying from 2 to 6 days post‐administration. Therefore, to maximize the predicted DDI ratio, simulations of CYP probe substrate were started at 3.5 days (84 h) following the first administration of IL‐6.

Following the first administration, the maximum mean suppression varied from 3.07% to 46.2% for CYP3A4, from 0.215% to 19.9% for CYP1A2, and from 1.43% to 37.3% for CYP2C9 in the virtual North European White population (Table 3).

TABLE 3.

Changes in mean hepatic CYP levels following administration of low, medium and high IL‐6 in virtual North European White subjects following the first administration.

IL‐6 level	Maximum mean suppression of CYP3A4 (%)
Low (0.0002 mg)	3.07
Medium (0.004 mg)	24.2
High (0.04 mg)	46.2

	Maximum mean suppression of CYP1A2 (%)
Low (0.0002 mg)	0.215
Medium (0.004 mg)	3.76
High (0.04 mg)	19.9

	Maximum mean suppression of CYP2C9 (%)
Low (0.0002 mg)	1.43
Medium (0.004 mg)	15.1
High (0.04 mg)	37.3

Abbreviations: CYP, cytochrome; IL‐6, interleukin 6.

Model application

Simulations with simvastatin

Predicted plasma concentration‐time profiles of simvastatin following repeated oral dosing of 40 mg in the absence and presence of low (0.0002 mg dose) to high (0.04 mg dose) levels of IL‐6 in the 163 virtual patients are shown in Figure 4a. For each simulation, mean concentration‐time profiles representative of the total virtual population (N = 163) are shown. The predicted C _max and AUC values and associated ratios in exposure for simvastatin (CYP3A4 substrate) at various IL‐6 levels (i.e., low, medium, and high) are shown in Table S2. The susceptibility of simvastatin to a DDI with IL‐6 increases with increasing IL‐6 level, that is, the mean predicted AUC ratio following the first administration increases from 1.03 to 1.73, as the IL‐6 concentration is increased from low (0.0002 mg dose) to high (0.04 mg dose); and during the whole simulation period from 1.03 to 1.94 from low to high IL‐6 levels.

FIGURE 4.

Mean predicted plasma concentration‐time profiles of (a) simvastatin (40 mg), (b) midazolam (5 mg), (c) caffeine (150 mg), (d) theophylline (125 mg), and (e) S‐warfarin (10 mg) following the first dose in the absence and presence of low, medium, and high levels of IL‐6. Mean predicted plasma concentration‐time profiles of simvastatin, midazolam, caffeine, theophylline, and S‐warfarin following the first dose on day 3.5 in the absence (solid black line) and presence (dashed lines) of low (green), medium (orange) and high (red) of IL‐6 in North European Whites.

Simulations with midazolam

Predicted plasma concentration‐time profiles of midazolam following repeated oral daily dosing of 5 mg in the absence and presence of low (0.0002 mg dose) to high (0.04 mg dose) of IL‐6 in the 163 virtual patients are shown in Figure 4b. For each simulation, mean concentration‐time profiles representative of the total virtual population (N = 163) are shown. The predicted C _max and AUC values and associated ratios in exposure for midazolam (CYP3A4 substrate) at various IL‐6 levels (i.e., low, medium, and high) are shown in Table S3. The susceptibility of midazolam to a DDI with IL‐6 increases with increasing IL‐6 level, that is, the mean predicted AUC ratio following the first administration increases from 1.03 to 1.75, as the IL‐6 concentration is increased from low (0.0002 mg dose) to high (0.04 mg dose); and during the whole simulation period from 1.03 to 2.08 from low to high IL‐6 levels.

Simulations with caffeine

Predicted plasma concentration‐time profiles of caffeine following repeated oral daily dosing of 150 mg in the absence and presence of low (0.0002 mg dose) to high (0.04 mg dose) of IL‐6 in the 163 virtual patients are shown in Figure 4c. For each simulation, mean concentration‐time profiles representative of the total virtual population (N = 163) are shown. The predicted C _max and AUC values and associated ratios in exposure for caffeine (CYP1A2 substrate) at various IL‐6 levels (i.e., low, medium, and high) are shown in Table S4. The susceptibility of caffeine to a DDI with IL‐6 increases with increasing IL‐6 level, that is, the mean predicted AUC ratio following the first administration increases from 1.00 to 1.19, as the IL‐6 concentration is increased from low (0.0002 mg dose) to high (0.04 mg dose); and during the whole simulation period from 1.00 to 1.28 from low to high IL‐6 levels.

Simulations with theophylline

Predicted plasma concentration‐time profiles of theophylline following repeated oral t.i.d. dosing of 125 mg in the absence and presence of low (0.0002 mg dose) to high (0.04 mg dose) of IL‐6 in the 163 virtual patients are shown in Figure 4d. For each simulation, mean concentration‐time profiles representative of the total virtual population (N = 163) are shown. The predicted C _max and AUC values and associated ratios in exposure for theophylline (CYP1A2 substrate) at various IL‐6 levels (i.e., low, medium, and high) are shown in Table S5. The susceptibility of theophylline to a DDI with IL‐6 increases with increasing IL‐6 level, that is, the mean predicted AUC ratio following the first administration increases from 1.00 to 1.05, as the IL‐6 concentration is increased from low (0.0002 mg dose) to high (0.04 mg dose); and during the whole simulation period from 1.00 to 1.19 from low to high IL‐6 levels.

Simulations with S‐warfarin

Predicted plasma concentration‐time profiles of S‐warfarin following repeated oral t.i.d. dosing of 10 mg in the absence and presence of low (0.0002 mg dose) to high (0.04 mg dose) of IL‐6 in the 163 virtual patients are shown in Figure 4e. For each simulation, mean concentration‐time profiles representative of the total virtual population (N = 163) are shown. The predicted C _max and AUC values and associated ratios in exposure for S‐warfarin (CYP2C9 substrate) at various IL‐6 levels (i.e., low, medium, and high) are shown in Table S6. The susceptibility of S‐warfarin to a DDI with IL‐6 increases with increasing IL‐6 level, that is, the mean predicted AUC ratio following the first administration increases from 1.00 to 1.09, as the IL‐6 concentration is increased from low (0.0002 mg dose) to high (0.04 mg dose); and during the whole simulation period from 1.01 to 1.49 from low to high IL‐6 levels.

DISCUSSION

Treatment of patients with NHL results in transient alterations in IL‐6 levels following the first dose that can potentially result in the suppression of CYP activity. In order to mitigate CRS, which is observed mainly after first dosing, and within the first treatment cycle, a step‐up dosing regimen has been adopted, comprising pretreatment with 1000 mg obinutuzumab on C1 D1, 2.5 mg glofitamab on C1 D8, 10 mg glofitamab on C1 D15, and 30 mg glofitamab on C2 D1, with single 30 mg doses on day 1 of each subsequent cycle, for a total of 12 cycles.

The aim of the current PBPK modeling was to simulate the IL‐6 levels observed in cycle 1 of the clinical study, and to conduct further simulations using specific probe substrates for CYPs 3A4, 1A2, and 2C9 to enable a quantitative estimation of the magnitude of potential drug interactions.

In the current investigation, in vitro data describing the suppression effect of IL‐6 on levels of CYP isozymes ²⁵ was incorporated into PBPK models and used to predict quantitatively the magnitude of clinical DDIs between IL‐6 and known isozyme selective CYP substrates (simvastatin, midazolam, caffeine, theophylline, and S‐warfarin) in a virtual North European White population with age and body weight demographics matching those of the 2.5/10/30 mg step‐up dosing patients from the NP30179 study. This approach to prediction of IL‐6 mediated DDIs for small molecules has been successful in disease conditions, such as RA, bone marrow transplantation, post‐surgical trauma, leukemia, and coronavirus disease 2019 (COVID‐19). ³¹ , ³³ , ³⁶ , ⁴²

Transient increase of IL‐6 concentrations following treatment with cancer immunotherapy treatments, such as CARTs and other TCBs has been previously reported. ⁴³ , ⁴⁴ , ⁴⁵

Simulations in the present analysis with the “low” IL‐6 dose level achieved IL‐6 concentrations toward the highest levels of those reported for healthy subjects (up to 10 pg/mL) and corresponded to the lowest levels in glofitamab treated patients (up to 20–30 pg/mL for C _max). At these IL‐6 levels, the exposure of CYP substrates following the first administration (simvastatin, midazolam, caffeine, theophylline, and S‐warfarin) increased only marginally (up to 1.03‐fold for AUC ratio), independently of the CYP isoform. This suggests no impact on CYP activity at low systemic levels of IL‐6. The predicted interaction ratio was not higher when looking at the whole simulation period (i.e., AUC from 3.5 to 17 days).

The IL‐6 simulations for the medium scenario cover the observed IL‐6 measurements following the first administration for the majority of patients. Thus, the most probable suppression level may be associated with the medium scenario. When the transient increased IL‐6 concentrations were increased to the medium level (corresponding to the 0.004 mg IL‐6 dose; associated with a mean C _max around 200 pg/mL), there was a trend to a mild increase in systemic exposure of simvastatin and midazolam (CYP3A4 substrates) following the first administration of IL‐6. The AUC ratio was 1.28 and 1.30, respectively. A negligible impact was predicted for S‐warfarin (CYP2C9 substrate), caffeine and theophylline (CYP1A2 substrates) exposure ratios of 1.04, 1.03, and 1.01, respectively. A trend to slightly higher ratio was observed when looking at the whole simulation period (i.e., AUC from 3.5 to 17 days) for CYP3A substrates (ratios of 1.33 and 1.36 for simvastatin and midazolam, respectively), followed by the CYP2C9 substrate S‐warfarin (with a ratio of 1.17). However, no difference in the interaction ratio between the first administration and the whole simulation period was seen for CYP1A2 substrates (with a ratio of up to 1.04).

The high observed IL‐6 concentrations from a very limited number of patients were better covered by the high simulated scenario and so this may be considered as a representing a worst case for suppression. There was a greater degree of interaction following the first administration of the “high” IL‐6 dose level (i.e., 0.04 mg associated with a mean C _max around 2000 pg/mL) with mean predicted AUC ratio of up to 1.73 and 1.75‐fold for simvastatin and midazolam as CYP3A4 substrates. This was followed by CYP1A2 substrates caffeine and theophylline with mean predicted AUC ratio up to 1.19 and 1.05‐fold, respectively, and by CYP2C9 substrate S‐warfarin with mean predicted AUC ratio up to 1.09‐fold. This suggests that drugs predominantly metabolized by CYP3A4 enzyme will have the highest susceptibility for IL‐6 suppression mediated DDIs induced by the transiently increased IL‐6 levels above 1000 pg/mL. In contrast, there was no or only a mild difference in systemic exposure predicted for CYP2C9 and CYP1A2 substrates (S‐warfarin, caffeine, and theophylline) with high IL‐6 dose levels, following the first administration. A trend to slightly higher interaction ratio was observed when looking at the whole simulation period (i.e., AUC from 3.5 to 17 days) for CYP3A substrates (with a ratio of 1.94 and 2.08 for simvastatin and midazolam, respectively), followed by CYP2C9 substrate S‐warfarin (with a ratio of 1.49), and finally by CYP1A2 substrates caffeine and theophylline (with a ratio of 1.28 and 1.19, respectively). Due to the negligible suppression observed in vitro (Xu et al. ³³ ), the effect of IL‐6 elevation on substrates of CYP2C19 and CYP2D6 was not evaluated by PBPK in the present analysis. However, the effect is expected to be lower compared to other CYP isoforms assessed in the present analysis.

The present PBPK analysis predicted the effect of transient elevation of IL‐6 level on exposures of CYP3A4, CYP2C9, and CYP1A2 substrates to be less or equal to twofold in “high” IL‐6 dose level scenario. Among the substrates tested, the CYP3A4 substrates (i.e., simvastatin and midazolam appeared to be the most sensitive to the suppression effect of IL‐6, partially because of the higher susceptibility of CYP3A4 to IL‐6 suppression in vitro (Dickmann et al. ²⁵ ). Because simvastatin and midazolam are sensitive substrates of CYP3A4, the effect on less‐sensitive substrates should be lower than the effect projected for simvastatin and midazolam.

In summary, our investigation suggests that the magnitude of the suppressive effect of transient IL‐6 increase on hepatic CYP enzyme activities is less than 50%. In addition, the changes in exposures to substrates of CYP3A4, CYP1A2, and CYP2C9 are expected to be lower than or equal to twofold in the worst‐case scenario and the magnitude of CYP suppression is dependent on the duration of cytokine elevation. Based on this analysis, it is recommended that there are no restrictions on concomitant treatment with any other drugs. Consideration may be given for potential DDI during the first cycle in patients who are receiving concomitant CYP substrates with a narrow therapeutic index. In these patients, monitoring for toxicity (e.g., warfarin) or for drug concentrations (e.g., cyclosporine) can be warranted. This modeling work was reviewed by the health authorities during the filing of glofitamab and no clinical DDI studies with CYP substrates were performed (https://www.fda.gov/media/140909/download). The drug label (https://www.accessdata.fda.gov/drugsatfda_docs/label/2023/761309s000lbl.pdf) states that “For certain CYP substrates where minimal concentration changes may lead to serious adverse reactions, monitor for toxicities or drug concentrations of such CYP substrates when coadministered with COLUMVI.” Similarly the summary of product characteristics states “On initiation of Columvi therapy, patients being treated with CYP450 substrates with a narrow therapeutic index should be monitored…” (https://www.ema.europa.eu/en/documents/product‐information/columvi‐epar‐product‐information_en.pdf).

AUTHOR CONTRIBUTIONS

N.D. wrote the manuscript. N.D. designed the research. N.D. performed the research. N.D., N.P., E.G., A.O., and D.C. analyzed the data. F.J. contributed new reagents/analytical tools.

CONFLICT OF INTEREST STATEMENT

N.P., F.J., A.O., E.G. and D.C. are employees of F. Hoffmann‐La Roche. N.D., N.P., F.J., A.O., E.G. and D.C. hold Roche shares. N.D. was an employee of F. Hofmann‐La Roche.

Supporting information

Tables S1–S6

Djebli N, Parrott N, Jaminion F, O’Jeanson A, Guerini E, Carlile D. Evaluation of the potential impact on pharmacokinetics of various cytochrome P450 substrates of increasing IL‐6 levels following administration of the T‐cell bispecific engager glofitamab. CPT Pharmacometrics Syst Pharmacol. 2024;13:396‐409. doi: 10.1002/psp4.13091