Pharmacokinetic-Interactions of BI 425809, a Novel Glycine Transporter 1 Inhibitor, With Cytochrome P450 and P-Glycoprotein Substrates

Abstract

Purpose/Background

Glycine transporter-1 inhibitors may ameliorate cognitive deficits in schizophrenia. This study evaluated potential drug-drug interactions with the glycine transporter-1 inhibitor BI 425809.

Methods/Procedures

Interactions with cytochromes P450 (CYP) and P-glycoprotein (P-gp) were assessed in in vitro assays using human hepatocytes and Caco-2 cells, respectively. Pharmacokinetic characteristics of probe drugs were subsequently assessed in a Phase I, open-label, single-sequence crossover study in healthy male participants. Participants received a probe-drug cocktail containing midazolam (CYP3A4), warfarin (CYP2C9), and omeprazole (CYP2C19) and a separate dose of digoxin (P-gp), alone and on a background of steady-state BI 425809 25 mg once daily in 2 treatment periods. Adverse events were monitored.

Findings/Results

In vitro assays revealed concentration-dependent induction of CYP3A4 and inhibition of P-gp by BI 425809. In the clinical study, 12 of 13 participants completed both periods. With BI 425809, area under the plasma concentration curve from administration to the last measurement (AUC_0–tz) and maximum plasma concentration (C_max) for midazolam were lower than when administered alone. Adjusted geometric mean ratios (90% confidence interval) were 70.6% (63.9%–78.1%) for AUC_0–tz and 77.6% (67.3%–89.4%) for C_max. For warfarin and digoxin, AUC_0–tz and C_max were similar with and without BI 425809. For omeprazole, BI 425809 slightly reduced AUC_0–tz but not C_max versus omeprazole alone. No new safety signals were identified.

Implications/Conclusions

These findings indicate induction of CYP3A4 by once-daily BI 425809 25 mg (the assumed highest therapeutic dose) and no meaningful effects on CYP2C9, CYP2C19, or P-gp in vivo.

Cognitive impairment in schizophrenia is thought to involve impaired glutamatergic signaling mediated by N-methyl-d-aspartate (NMDA) receptors, and as such, promotion of NMDA receptor signaling is a promising target for novel therapies to ameliorate these symptoms.^1–3 Glycine is an obligatory co-agonist at glutamatergic NMDA receptors, and inhibition of the glycine uptake transporter can lead to an elevated synaptic concentration of glycine that may enhance NMDA receptor signaling.⁴ BI 425809, a potent and selective inhibitor of the glycine uptake transporter-1, is currently under development as a treatment for cognitive impairment associated with schizophrenia. The molecular structure of BI 425809 is shown in Figure 1. Phase I studies have indicated that once-daily oral administration of BI 425809 up to 75 mg is generally well tolerated, and doses of 10 mg and 25 mg can increase cerebrospinal fluid glycine levels by an average of 50% in healthy volunteers.^5–8 Furthermore, BI 425809 has shown procognitive effects over placebo at doses of 10 mg and 25 mg in a Phase II trial in patients with schizophrenia, with Phase III trials currently under way.

FIGURE 1.

Structure of BI 425809. Figure reused with permission from Ref. 5. BI 425809 molecular weight, 512.43 nM.

Cytochrome P450 (CYP) 3A4 is the most highly expressed CYP isoform in the intestine and liver and so is particularly important in determining exposure to orally administered medications.⁹ Previous studies have demonstrated that the major metabolic pathway for BI 425809 is mediated by CYP3A4.^10,11 However, it is unknown whether BI 425809 can influence CYP3A4 activity, which is regulated by the pregnane X receptor (PXR).¹² Induction of CYP2C9¹³ and CYP2C19¹⁴ is also regulated by PXR, and so any effect of BI 425809 on PXR could potentially lead to clinically relevant changes in the exposure of a wide range of drugs that are substrates of these three CYP isoforms.⁹ Induction of P-glycoprotein (P-gp), a common pathway for limiting the oral bioavailability and efflux of therapeutic compounds from tissues, is also regulated by PXR,¹⁵ and many P-gp substrates are also substrates for CYP3A enzymes.¹⁶ Drug-drug interactions caused by induction of CYP enzymes or transporters by BI 425809 could affect the bioavailability and exposure, and, consequently, the efficacy and safety profile, of concomitant medications that are also substrates of the induced enzymes or transporters. It is therefore important to understand how BI 425809 could interact with other medications. This article reports the findings of preclinical and clinical studies that assessed potential drug-drug interactions for BI 425809. The objective of the preclinical study was to assess the in vitro effects of BI 425809 on CYP3A4 and P-gp activity. Based on these findings, a clinical study was conducted to determine the effect of multiple doses of BI 425809 25 mg once daily, the assumed highest therapeutic dose, on the pharmacokinetics (PK) of sensitive substrates for CYP3A4 (midazolam), CYP2C9 (warfarin), CYP2C19 (omeprazole), and P-gp (digoxin). Midazolam, warfarin, and omeprazole are components of a previously validated drug cocktail and can be coadministered because of a lack of PK interactions,^17,18 whereas digoxin was administered separately.

MATERIALS AND METHODS

In Vitro Analysis of CYP3A4 Induction by BI 425809

The potential of BI 425809 to induce CYP3A4 was assessed by measuring increases in enzyme activity and messenger RNA (mRNA) expression in sandwich-cultured primary human hepatocytes from three donors (Discovery Labware). Hepatocytes were treated with either BI 425809 (0.1, 1, 3, 10, 30, or 100 μM), vehicle control (0.1% dimethyl sulfoxide), or the CYP3A4 inducer rifampicin (25 μM) for 48 hours. CYP3A4 enzyme activity was then assessed in situ by incubation of the hepatocytes with testosterone (200 μM for 45 minutes) and subsequent quantification of the metabolite, 6β-hydroxytestosterone, via liquid chromatography–tandem mass spectrometry. Messenger RNA was isolated from sandwich-cultured hepatocytes using the RNeasy 96 Kit (Qiagen Inc, Germantown, MD) as previously described.¹⁹ Levels of CYP3A4 mRNA expression in hepatocytes treated with BI 425809 or vehicle control were measured using TaqMan real-time reverse transcriptase polymerase chain reaction. The assay was performed using the 2-step protocol involving reverse transcriptase reaction to generate cDNA, followed by TaqMan real-time PCR. Real-time reverse transcriptase polymerase chain reaction assay plates were prepared by combining 20 μL of reaction mixture with 5 μL of each cDNA sample in triplicate. The PCR assays were performed using the ViiA 7 Real-Time PCR System (Applied Biosystems, Waltham, MA).

Fold CYP3A4 induction relative to control was determined using CYP3A4 mRNA response normalized to β-actin mRNA response and the ΔΔCt method for calculating relative changes in gene expression.²⁰ The maximum fold induction response (E_max) and the concentration of BI 425809 resulting in half-maximal enzyme induction response (EC₅₀) were derived by plotting the mRNA induction response against the concentration of BI 425809 and fitting the data to a 3-parameter sigmoidal model described by the following equation (where Y is the fold induction, X is the concentration of BI 425809, and b is a factor describing the steepness of the curve).

$$Y=\frac{E_{\max }}{1+e^{-\left(\frac{X-{\mathit{EC}}_{50}}{b}\right)}}$$

The potential of BI 425809 to cause clinical induction was assessed using a basic induction model documented in the 2020 Food and Drug Administration (FDA) guidance on CYP enzyme- and transporter-mediated drug interaction studies (docket number, FDA-2017-D-5961), which is described by the following equation (where R₃ is the estimated change in hepatic clearance of a sensitive CYP3A4 substrate and I_max,u is the maximum unbound plasma concentration of BI 425809 [approximately 121 nM]).

$$R_3=\frac{1}{1+\left[\frac{E_{\max }\times 10\times I_{max,u}}{{\mathit{EC}}_{50}+\left(10\times I_{max,u}\right)}\right]}$$

In Vitro Analysis of Inhibition of Principal Human CYPs by BI 425809

To investigate the potential of BI 425809 to reversibly inhibit the major human CYPs, CYP-selective substrates (phenacetin 60 μM [CYP1A2], bupropion 80 μM [CYP2B6], amodiaquine 2 μM [CYP2C8], diclofenac 5 μM [CYP2C9], S-mephenytoin 80 μM [CYP2C19], dextromethorphan 5 μM [CYP2D6], midazolam 2 μM, and testosterone 50 μM [CYP3A4/5]) were incubated with human liver microsomes and BI 425809 (0.015, 0.046, 0.137, 0.411, 1.23, 3.70, 11.1, 33.3, and 100 μM). For positive control reactions, BI 425809 was replaced with a CYP-selective inhibitor (α-naphthoflavone [CYP1A2], ticlopidine [CYP2B6], montelukast [CYP2C8], sulfaphenazole [CYP2C9], benzylnirvanol [CYP2C19], quinidine [CYP2D6], and itraconazole [CYP3A4/5]). Substrate metabolites were quantified with liquid chromatography–tandem mass spectrometry using gradient elution (mobile phase for amodiaquine metabolite—A, 5 mM ammonium formate in water/formic acid [100:0.1, v/v]; B, acetonitrile/formic acid [100:0.1, v/v]; mobile phase for all other substrate metabolites—A, water/formic acid [100:0.1, v/v]; B, acetonitrile/formic acid [100:0.1, v/v]) on a Synergi Hydro RP column (50 × 2.0 mm, 4 μm; Phenomenex) with positive electrospray ionization.

IC₅₀ values were obtained using a 3-parameter dose-response, 4-parameter dose-response, or normalized dose-response model; model comparisons were performed in Prism 6 (GraphPad Inc) to determine the optimal model for each data set. A least-squares fitting approach was used, and the Hill slope was not constrained for the 4-parameter model.

In Vitro Analysis of P-Gp Inhibition by BI 425809

Digoxin transport was investigated using a bidirectional transcellular transport assay across a monolayer of Caco-2 cells. Cells were equilibrated in assay buffer (Hanks' Balanced Salt Solution with HEPES 15 mM adjusted to pH 7.4 with NaOH) for 30 minutes before [³H]digoxin (1 μM) was added to the donor compartment (apical side of the monolayer in apical-to-basal transport experiments, or the basal side in basal-to-apical experiments). BI 425809 (1, 3, 10, 30, or 100 μM) or zosuquidar (1 μM) was added to both the donor compartment and the receiver compartment on the opposite side of the monolayer. The assay was initiated after 30 minutes of preincubation with digoxin. Samples were collected from the donor compartment at −30, 0, and 90 minutes and from the receiver compartment at 0, 30, 60, and 90 minutes. Sample radioactivity was measured using a liquid scintillation counter.

The permeability coefficient (P_app) value was calculated using the transport rate and the initial concentration of radioactivity in the donor compartment using the following equation, where P_app is the permeability coefficient (cm/s), C_t0 is the initial radioactivity concentration in the donor compartment at time t₀ (dpm/mL), A is the area of the filter (cm²), V_R is the volume of buffer in the receiver compartment (mL), and ΔC_R/Δt is the change in radioactivity concentration over time in the receiver compartment (dpm/[mL·s]):

$$P_{\mathrm{app}}=\frac{V_{\mathrm{R}}}{A\cdot C_{t0}}\cdot \frac{∆C_{\mathrm{R}}}{∆t}$$

The transport rate ∆C_R/∆t was calculated based on the linear part of the compound concentration in the receiver compartment over time curve.

The efflux ratio (ER) for digoxin was calculated as the ratio of the permeability coefficients for basal-to-apical (BtoA) and apical-to-basal (AtoB) transport, using the following equation:

$$\mathrm{ER}=\frac{P_{\mathrm{app}}\text{BtoA}}{P_{\mathrm{app}}\text{AtoB}}$$

The concentration of inhibitor resulting in 50% inhibition of P-gp was calculated based on iterative nonlinear regression analysis of the dose-response relationship, which was performed using XLfit (version 5.3.1.3; IDBS, Guildford, United Kingdom). IC₅₀ values were calculated, assuming standard (hyperbolic) Michaelis-Menten kinetics, using the following equation (where ER is the observed ER, h is the slope factor, I is the concentration of inhibitor [μM], ER_max is the ER at I = 0, and ER_min is the ER at I = infinity):

$${{\mathrm{IC}}_{50}}^h=\frac{\left(\mathrm{ER}-{\mathrm{ER}}_{\min }\right)\cdot I^h}{{\mathrm{ER}}_{\max }-\mathrm{ER}}$$

Clinical Study Design

This was a Phase I, open-label, single-sequence crossover study conducted in healthy male participants at the Boehringer Ingelheim Human Pharmacology Center, Biberach, Germany (Clinicaltrials.gov identifier, NCT02783040). The study consisted of 2 treatment periods in a fixed sequence. In period 1, participants received a single dose of a probe-drug cocktail containing midazolam (2 mg), warfarin (10 mg), and omeprazole (20 mg) on day 1 and a single dose of digoxin (0.25 mg) on day 2. Period 1 was followed by a wash-out period of ≥5 days. In period 2, participants received BI 425809 25 mg once daily on days 1 to 14, a single dose of the probe-drug cocktail on day 10, and a single dose of digoxin on day 11. On days 1 and 2 of period 1 and days 10 and 11 of period 2, participants fasted overnight for at least 10 hours before dosing and for 4 hours afterward. All participants attended an end-of-trial visit 11 to 16 days after the last administration of BI 425809.

Ethical Approval and Informed Consent

This study was carried out in compliance with the protocol and in accordance with the principles of the Declaration of Helsinki, the International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH) Harmonized Tripartite Guideline for Good Clinical Practice, applicable regulatory requirements and Boehringer Ingelheim standard operating procedures. Before admission into the study, all participants provided signed and dated informed consent in writing after the experimental procedures and risks were explained, in accordance with ICH Harmonized Tripartite Guideline for Good Clinical Practice and local legislation. The study protocol was reviewed and approved by the local independent ethics committee (Ethikkommission der Landesärztekammer Baden-Württemberg, Stuttgart, Germany) and the relevant local authorities.

Participants

Participants were included in the study if they were healthy according to the investigator's assessment, were 18 to 55 years of age, and had a body mass index of 18.5 to 29.9 kg/m². Participants were excluded if they had any finding in the medical examination deviating from normal or laboratory value outside the reference range or evidence of concomitant disease, which was judged to be clinically relevant by the investigator. Other exclusion criteria included use of drugs that might influence the results of the study, smoking (>10 cigarettes per day), or alcohol or drug abuse. There were also several exclusion criteria related to risks associated with the probe drugs; patients were excluded if they had an increased risk of bleeding, any use of anticoagulant medications within 10 days before administration of the trial medication, an atrioventricular block or a PQ interval of >200 ms detected by electrocardiogram (ECG) at screening. Full inclusion and exclusion criteria are listed in Supplementary Table S1, http://links.lww.com/JCP/A836.

Pharmacokinetic Endpoints and Assessments

The PK endpoints included the PK characteristics of each of the probe drugs administered, either as a component of the probe-drug cocktail (midazolam, S-warfarin, R-warfarin, omeprazole) or alone (digoxin), in the absence of BI 425809 in period 1 and on a background of steady-state BI 425809 in period 2. The PK characteristics of interest were area under the concentration-time curve (AUC) of the probe drug in plasma from administration (time t = 0) to the last quantifiable data point (AUC_0–tz), the maximum measured concentration of the probe drug in plasma (C_max), time from dosing to maximum measured concentration of the probe drug in plasma (t_max), and terminal half-life of the probe drug in plasma (t_1/2). The attainment of BI 425809 steady state was explored by assessing trough plasma concentration immediately before the next dose (C_pre) on days 8, 9, and 10 in period 2. For BI 425809 at steady state, AUC over a uniform dosing interval (AUC_τ,ss), C_max over a uniform dosing interval (C_max,ss), and t_max (t_max,ss) were also described. Plasma samples for probe drugs were collected on days 1 to 6 of period 1 and on days 10 to 15 of period 2. Plasma samples for BI 425809 were collected on days 1 to 4, 6, and 8 to 15 of period 2 and at the end-of-trial visit.

Plasma concentrations of BI 425809 and probe drugs were assessed using validated liquid chromatography–tandem mass spectrometry methods (details are shown in Supplementary Table S2, http://links.lww.com/JCP/A836).

Pharmacokinetic Modeling

Mechanistic static modeling was conducted using DDI Predict software (2012 Edition; Aureus Sciences, Paris, France), which uses the change in AUC for probe drugs in plasma in the presence of a candidate compound to predict potential drug-drug interactions with a large panel of comedications.²¹ In this study, mechanistic static modeling was used to predict the effects of BI 425809 on exposure to comedications taken by >5% of patients with schizophrenia. The ratio of AUC in the presence and absence of BI 425809 (AUC_i/AUC) was calculated using the equation hereinafter, as previously described,²² where AUC_i/AUC is the predicted ratio of AUC in the presence and absence of BI 425809, C_max,u is the maximal unbound drug concentration in plasma, F_G is the fraction escaping intestinal extraction, f_m is the fraction metabolized, k_deg,h is the apparent first-order degradation rate constant for CYP3A in the liver, k_deg,g is the apparent first-order degradation rate constant for CYP3A in the intestine, K_I is the concentration of inactivator at which the rate of inactivation is half maximal, k_inact is the maximal inactivation rate constant, and [I]_ent is the concentration of BI 425809 in the enterocyte (calculated as the product of the fraction absorbed [1], the absorption rate constant [0.0033 min⁻¹], and the dose divided by the intestinal blood flow rate [0.3 L/min]).

$${\mathrm{AUC}}_i/\mathrm{AUC}=\begin{array}{l}\frac{1}{\left[f_{\mathrm{m}}\times \left[\left(\frac{k_{deg,\mathrm{h}}}{k_{deg,h+\frac{C_{max,\mathrm{u}}\times k_{\text{inact}}}{C_{max,\mathrm{u}}+K_I}}}\right)\times \left(1+\frac{E_{\max }\times C_{max,\mathrm{u}}}{C_{max,\mathrm{u}}+{\mathrm{EC}}_{50}}\right)\right]+\left(1-f_{\mathrm{m}}\right)\right]}\\ \times \frac{1}{\left[\left[\left(\frac{k_{deg,\mathrm{g}}}{k_{deg,\mathrm{g}+\frac{{\left[I\right]}_{\mathrm{ent}}\times k_{\text{inact}}}{{\left[I\right]}_{\mathrm{ent}}+K_{\mathrm{I}}}}}\right)\times \left(1+\frac{{\left[I\right]}_{\mathrm{ent}}\times E_{\max }}{{\left[I\right]}_{\mathrm{ent}}+{\mathrm{EC}}_{50}}\right)\right]\times \left(1-F_{\mathrm{G}}\right)+F_{\mathrm{G}}\right]}\end{array}$$

The parameter values used in the model are defined in Supplementary Table S3, http://links.lww.com/JCP/A836. The ratio of AUC in the presence of BI 425809 to AUC in the absence of BI 425809 (AUC_i/AUC) was calculated, with E_max adjusted from 18.7 to 3.9 to match the observed in vivo AUC_i/AUC for midazolam. For comedications not listed in the DDI Predict software, the AUC_i/AUC values were calculated manually.

Safety Endpoints and Assessments

No specific safety endpoints were defined, but safety and tolerability were assessed based on the number of participants with treatment-emergent adverse events (AEs) including clinically relevant findings from physical and neurological examinations, drug-related AEs, serious AEs, AEs of special interest, and other significant AEs. Other safety and tolerability assessments included safety laboratory parameters, 12-lead ECG, vital signs (systolic and diastolic blood pressure and pulse rate), visual tests (Ishihara Color Plates, Jaeger Eye Chart, and Amsler Grid Test), and suicidality monitoring using the Columbia-Suicide Severity Rating Scale.

Statistical Analysis

Determination of the sample size was not based on a formal power calculation. A sample size of 13 participants was considered sufficient to achieve the aims of this study.

Of the trial populations included in this analysis, the treated set (TS) included all participants who entered the study and received ≥1 dose of study medication. The TS was used to analyze all demographic and safety data. The PK set was used for PK analyses and included all participants in the TS who provided ≥1 value for a PK endpoint.

The impact of BI 425809 on probe drug PK endpoints was assessed using the relative bioavailability method, comparing the adjusted geometric mean (gMean) ratios for test treatments (ie, the probe-drug cocktail or digoxin administered in combination with BI 425809) against reference treatments (ie, the probe drugs administered alone). For the probe drugs and for R-warfarin, AUC_0–tz and C_max data were analyzed using an analysis of variance model on the logarithmic scale including “participant” as a random effect and “treatment” as a fixed effect. The effects of BI 425809 on t_max and t_1/2 of probe drugs were assessed descriptively. To assess the attainment of BI 425809 steady state after multiple dosing, a repeated-measures linear model on the logarithmic scale was applied, with “participant” as a random effect and “time” as a repeated effect. A lack of effect on time indicates that BI 425809 has reached steady state. Safety and tolerability measures were analyzed descriptively, and no statistical tests were performed. All analyses were conducted with SAS software, version 9.4 (SAS Institute, Cary, NC).

RESULTS

In Vitro Induction of CYP3A4 by BI 425809

Concentration-dependent induction of CYP3A4 activity was observed in human hepatocytes incubated with BI 425809 or rifampicin, a prototypical PXR activator. E_max for induction of CYP3A4 activity at BI 425809 concentrations of 10 to 30 μM in hepatocytes from donors 1, 2, and 3 was 10-fold (68% of positive control effect), 3.4-fold (48% of positive control effect), and 6.6-fold (37% of positive control effect), respectively (Fig. 2A). Similarly, concentration-dependent induction of CYP3A4 mRNA levels was observed in hepatocytes from donors 1, 2, and 3, with E_max values of 14-fold (65% of positive control effect), 13-fold (91% of positive control effect), and 37-fold (22% of positive control effect), respectively, at 10 to 30 μM BI 425809 (Fig. 2B). No decrease in cell viability was observed relative to vehicle control in hepatocytes from any donor with BI 425809 up to 30 μM. When the observed mRNA levels from each of the donor hepatocytes were fitted to a 3-parameter sigmoidal model, EC₅₀ and E_max values for induction of CYP3A4 by BI 425809 were 1.96 μM and 18.7-fold, respectively (Fig. 2C).

FIGURE 2.

Relative levels of (A) CYP3A4 activity and (B) CYP3A4 mRNA expression and (C) dose-response curve for induction of CYP3A4 mRNA levels in sandwich-cultured human hepatocytes after treatment with BI 425809 or the CYP3A4 inducer rifampicin. Error bars denote SD. RIF, rifampicin.

In Vitro Inhibition of Principal Human CYPs by BI 425809

An IC₅₀ value of 54.9 μM was calculated for inhibition of CYP2C9 by BI 425809. For CYP1A2, CYP2B6, CYP2C8, CYP2C19, CYP2D6, and CYP3A4/5, the calculated IC₅₀ values were >100 μM (the highest tested concentration of BI 425809). Mean IC₅₀ values for positive control inhibitors were consistent with previous reports (CYP1A2, 0.01 μM; CYP2B6, 0.22 μM; CYP2C8, 0.08 μM; CYP2C9, 0.15 μM; CYP2C19, 0.05 μM; CYP2D6, 0.03 μM; CYP3A4/5, 0.15 μM [midazolam substrate] and 0.05 μM [testosterone substrate]).^23–27

In Vitro Inhibition of P-gp by BI 425809

Under the experimental conditions used in this study, the selective P-gp inhibitor zosuquidar^28,29 inhibited P-gp–mediated transcellular transport of digoxin across a Caco-2 cell monolayer. In addition, concentration-dependent inhibition on P-gp–mediated transport was observed in the presence of BI 425809 (Fig. 3A). The IC₅₀ for inhibition of P-gp by BI 425809 was 14 μM (95% confidence interval, 0.69–27.5) (Fig. 3B), which was used to calculate an I₂/IC₅₀ ratio of approximately 13 (where I₂ = dose[mol]/250 mL).

FIGURE 3.

In vitro inhibition of P-gp by BI 425809 as shown by (A) ER of digoxin in cultured Caco-2 cell monolayers in the presence of increasing concentrations of BI 425809 and in the presence of the selective P-gp inhibitor zosuquidar and (B) nonlinear regression analysis of the ER of digoxin in the presence of increasing concentrations of BI 425809. Error bars denote SD. ER_max, efflux ratio maximum; ER_min, efflux ratio minimum; Hill, Hill coefficient; IC₅₀, inhibitor concentration at 50% of maximal inhibition; ZSQ, zosuquidar.

Participant Disposition and Baseline Characteristics

A total of 13 participants entered the study, and 12 participants completed both treatment periods. All participants were healthy White males. The mean (SD) age was 37.1 (13.1) years, and the mean (SD) body mass index was 24.5 (2.3) kg/m². At the time of the study, 8 participants (61.5%) had never smoked, 2 (15.4%) were former smokers, and 3 (23.1%) were current smokers. None of the participants consumed alcohol to the extent of interfering with the study, and 1 participant did not drink alcohol at all. No participants received any concomitant medication at baseline.

Assessment of BI 425809 Steady State

Individual and gMean BI 425809 trough plasma concentrations over 0 to 196 hours are shown in Figure 4. Pairwise comparisons of C_pre on days 8, 9, and 10 in period 2 are summarized in Supplementary Table S4, http://links.lww.com/JCP/A836. The adjusted gMean ratios of the compared sampling time points were all close to 100%. This suggests that BI 425809 reached steady state by day 8, before administration of the probe-drug cocktail and digoxin. Further PK parameters of BI 425809 at steady state are described in Supplementary Table S5, http://links.lww.com/JCP/A836.

FIGURE 4.

Individual and geometric mean BI 425809 plasma concentration after multiple oral doses of BI 425809 25 mg.

Effect of Steady-State BI 425809 on Probe Drugs

The presence of steady-state BI 425809 affected the PK characteristics of midazolam, reducing AUC_0–tz and C_max relative to when the probe-drug cocktail was administered alone, in the absence of BI 425809 (Fig. 5, Supplementary Fig. 1, http://links.lww.com/JCP/A836). The adjusted gMean ratio (90% confidence interval) was 70.6% (63.9%–78.1%) for AUC_0–tz and 77.6% (67.3%–89.4%) for C_max (Table 1). On a background of steady-state BI 425809, gMean (geometric coefficient of variation) t_1/2 of midazolam was also slightly reduced compared with midazolam alone (2.8 [48.4] hours vs 3.3 [22.8] hours). The PK characteristics of S-warfarin, R-warfarin, and digoxin were similar in the presence and absence of steady-state BI 425809, with adjusted gMean ratios within ±4% of 100% for AUC_0–tz and C_max (Table 1). For omeprazole, AUC_0–tz was slightly lower when the probe-drug cocktail was administered with BI 425809 at steady state than when the cocktail was administered alone. However, C_max was similar in both the presence and absence of BI 425809 (Table 1). The values of t_1/2 and t_max for S-warfarin, R-warfarin, omeprazole, and digoxin were similar with and without steady-state BI 425809 (data not shown). The arithmetic mean values for C_max, AUC_0–tz, AUC_0–∞, and elimination half-life of the probe drugs are detailed in Table 2.

FIGURE 5.

Arithmetic mean plasma concentration-time profiles of midazolam 2 mg administered in cocktail with and without BI 425809 25 mg at steady state (PK set) Data are represented by mean values, and error bars show standard deviation.

TABLE 1.

Inferential Analysis of the Effects of BI 425809 at Steady State on AUC_0–tz and C_max of Probe Drugs (PK Set)

	AUC0–tz				C max
	Probe Drug + BI 425809		Probe Drug Only		Adjusted gMean Ratio (90% CI), %	Probe Drug + BI 425809		Probe Drug Only		Adjusted gMean Ratio, % (90% CI)
	n	gMean, ng⋅h/mL	n	gMean, ng⋅h/mL		n	gMean, ng/mL	n	gMean, ng/mL
Midazolam	12	16.6	13	23.5	70.6 (63.9–78.1)	12	6.6	13	8.5	77.6 (67.3–89.4)
S-warfarin	12	15,205.1	12	15,107.9	100.6 (97.2–104.3)	12	553.7	13	534.5	103.6 (95.1–112.8)
R-warfarin	12	22,457.1	12	22,615.1	99.3 (96.8–101.9)	12	541.8	13	525.0	103.2 (95.5–111.6)
Omeprazole	12	429.8	13	483.3	88.9 (81.0–97.6)	12	225.1	13	220.3	102.2 (93.6–111.7)
Digoxin	12	9.5	12	9.7	97.9 (86.6–110.6)	12	1.1	12	1.1	100.6 (86.9–116.5)

AUC_0–tz, area under the concentration-time curve of the probe drug in plasma from administration (time t = 0) to the last quantifiable data point; CI, confidence interval; C_max, maximum measured concentration of the probe drug in plasma; n, number of participants analyzed.

TABLE 2.

Arithmetic Mean (SD) PK Parameters for the Probe Drugs

	Cmax, ng/mL	AUC0–tz, ng⋅h/mL	AUC0–∞, ng⋅h/mL	t1/2, h
Midazolam 2 mg in cocktail	9.03 (3.3)*	24.7 (8.6)*	26.1 (9.5)*	3.4 (0.8)*
+BI 425809 25 mg	6.9 (3.2)	16.7 (6.1)	17.4 (6.3)	3.1 (1.5)
Omeprazole 20 mg in cocktail	287.0 (219.0)*	782.0 (893.0)*	837.0 (110.0)	1.0 (0.7)
+BI 425809 25 mg	257.0 (187.0)	604.0 (733.0)	609.0 (733.0)	1.0 (0.6)
R-warfarin 10 mg in cocktail	537.0 (121.0)*	23,000.0 (4860.0)	27,900.0 (6680.0)	46.5 (7.7)
+BI 425809 25 mg	558.0 (166.0)	23,000.0 (5360.0)	27,500.0 (7200.0)	44.5 (10.0)
S-warfarin 10 mg in cocktail	547.0 (126.0)*	15,400.0 (3480.0)	16,900.0 (3870.0)	35.8 (2.9)
+BI 425809 25 mg	572.0 (174.0)	15,600.0 (3590.0)	17,000.0 (3880.0)	35.7 (4.3)
Digoxin 0.25 mg	1.1 (0.3)	10.3 (3.1)	21.4 (5.6)	59.1 (15.2)
+BI 425809 25 mg	1.1 (0.3)	10.0 (3.1)	20.1 (5.6)	54.2 (18.9)

n = 12 unless otherwise noted.

AUC_0–tz, area under the concentration-time curve of the probe drug in plasma from administration (time t=0) to the last quantifiable data point; AUC0_-∞, area under the concentration-time curve of the probe drug in plasma from administration (time t=0) extrapolated to infinity; C_max, maximum measured concentration of the probe drug in plasma; n, number of participants analyzed; P, pharmacokinetics; SD, standard deviation; t_1/2, half-life.

*n = 13.

The predictions from mechanistic static modeling regarding the effects of BI 425809 on exposure to commonly prescribed comedications in patients with schizophrenia are listed in Supplementary Table S6, http://links.lww.com/JCP/A836. Of 15 commonly prescribed comedications, 13 showed predicted AUC_i/AUC ratios of 0.80 to 1.00; only atorvastatin (0.64) and simvastatin (0.69) were predicted to show slightly reduced exposure in the presence of BI 425809.

Adverse Events

All 13 participants experienced at least 1 AE during the study (Table 3). In total, 4 of 13 participants (30.8%) experienced AEs during period 1, and 11 of 12 (91.7%) experienced AEs during period 2, including the 11-day period after last dosing with BI 425809. Adverse events considered related to the study medication were observed in 8 of 12 participants (66.7%) after administration of the probe drugs on a background of steady-state BI 425809, whereas no participant experienced drug-related AEs after administration of the probe drugs alone.

TABLE 3.

Overview of Number (%) of Participants With AEs and Number (%) of Participants With AEs by Preferred Terms (TS)

Overview of Participants With AEs	Probe Drugs, n (%) (N = 13)	Probe Drugs + BI 425809, n (%) (N = 12)	Total on Treatment, n (%) (N = 13)
Participants with any AE(s)	4 (30.8)	11 (91.7)	13 (100)
Participants with severe AE(s)	1 (7.7)	0	1 (7.7)
Participants with drug-related AE(s)	0	8 (66.7)	8 (61.5)
Participants with AE(s) leading to discontinuation of study drug	0	0	0
Participants with AE(s) of special interest	0	0	0
Participants with serious AE(s)	0	0	0
Participants with other significant AE(s)	0	0	0

	Probe Drugs (N = 13)		Probe Drugs + BI 425809 (N = 12)		Total on Treatment (N = 13)
Participants with AE(s) by preferred term	Total, n (%)	Drug Related, n (%)	Total, n (%)	Drug Related, n (%)	Total, n (%)	Drug Related, n (%)
Fatigue	0	0	4 (33.3)	4 (33.3)	4 (30.8)	4 (30.8)
Headache	1 (7.7)	0	3 (25.0)	3 (25.0)	4 (30.8)	3 (23.1)
Nasopharyngitis	0	0	2 (16.7)	0	2 (15.4)	0
Oropharyngeal pain	0	0	2 (16.7)	0	2 (15.4)	0
Dry eye	0	0	1 (8.3)	1 (8.3)	1 (7.7)	1 (7.7)
Pollakiuria	0	0	1 (8.3)	1 (8.3)	1 (7.7)	1 (7.7)
Rhinitis	0	0	1 (8.3)	0	1 (7.7)	0
Conjunctivitis	1 (7.7)	0	0	0	1 (7.7)	0
Oral herpes	1 (7.7)	0	0	0	1 (7.7)	0
Upper respiratory tract infection	1 (7.7)	0	0	0	1 (7.7)	0
Arthropod bite	1 (7.7)	0	0	0	1 (7.7)	0

N indicates the number of participants analyzed; n, number of participants with AE.

One severe AE was observed following administration of the probe-drug cocktail alone but was not considered related to the study medication; this participant reported a severe headache and subsequently withdrew from the study. All AEs were reported as resolved at the end-of-trial visit. No other serious AEs, AEs of special interest, or other significant AEs, as defined by the 1996 ICH E3 guidelines were observed. No abnormal findings from visual tests or Columbia-Suicide Severity Rating Scale signals were reported as AEs. Laboratory tests and the evaluation of vital signs and ECGs revealed no clinically relevant findings.

DISCUSSION

In this study, in vitro assays demonstrated that BI 425809 can induce CYP3A4 in human hepatocytes. According to the basic model for the prediction of clinical induction outlined in the 2020 FDA guidance for in vitro drug-drug interaction studies (docket number, FDA-2017-D-5961), a predicted ratio of victim drug AUC in the presence and absence of the inducer (R₃) ≤0.8 necessitates further investigation of the inducer in a clinical drug-drug interaction study. In our preclinical study, the determined R₃ (change in hepatic metabolic clearance of a sensitive CYP3A4 substrate) for induction of CYP3A4 by BI 425809 was 0.12, calculated based on an unbound plasma fraction of 0.196 derived from other preclinical studies (data not shown) according to the formula provided in the 2020 FDA guidance. This R₃ value indicates that BI 425809 could potentially cause clinically relevant induction of CYP3A4. Furthermore, induction of CYP3A4 is regulated by the transcriptional factor PXR,¹² and as such, BI 425809 treatment may also affect the expression of CYP2C9, CYP2C19, and P-gp, which are also regulated by PXR.^13–15 However, additional preclinical findings in this study suggested that BI 425809 may also inhibit P-gp in vivo, making the net effect of BI 425809 treatment on P-gp activity difficult to predict based on preclinical evidence alone. Potential inhibition of CYP3A4/5 and other major human CYP isoforms by BI 425809 was also evaluated; the calculated IC₅₀ values were in the range of 50 to 100 times greater than the reported therapeutic steady-state BI 425809 C_max in humans,^5–7 suggesting that drug-drug interactions as a result of reversible CYP inhibition are unlikely to occur at clinically relevant BI 425809 doses.

Based on these preclinical findings, a Phase I clinical study was subsequently conducted to evaluate the effects of steady-state BI 425809 25 mg once daily on probe drugs for CYP3A4, CYP2C9, CYP2C19, and P-gp in healthy participants. Midazolam, warfarin, and omeprazole are components of a validated cocktail approach for drug-drug interaction studies,¹⁷ which requires exposure of fewer participants to the developmental compound compared with individual administration of each probe drug in separate studies. Digoxin is not included in the validated cocktail and was administered separately after a 1-day washout of the probe-drug cocktail but was not expected to interact with the cocktail based on the PK characteristics of the probe drugs.

Statistical evaluation of trough BI 425809 plasma concentration indicated that steady state was attained before administration of the probe drugs in period 2; this is consistent with a previous study, in which steady state was reached after 6 days.⁵ Because BI 425809 was maintained at steady state for several days before the probe drugs were administered on days 10 and 11, it is likely that maximal activation of PXR and therefore maximal induction of CYPs was reached by the time the probe drugs were administered. Under these conditions, AUC_0–tz and C_max for midazolam were lower than when the probe-drug cocktail was administered alone, indicating induction of CYP3A4 by BI 425809. There was no observed change in AUC_0–tz and C_max for S-warfarin, R-warfarin, or digoxin in the presence of BI 425809 25 mg once daily at steady state, indicating that BI 425809 did not influence CYP2C9 or P-gp activity. Although our findings suggest that BI 425809 can activate PXR, the lack of any net induction of P-gp in healthy participants may be due to a direct inhibitory effect of BI 425809 on P-gp that was observed in the in vitro transcellular transport assay. Alternatively, it may be that there is no significant induction or inhibition of P-gp by BI 425809 in vivo. Regulatory agencies recommend that a ratio of nominal gastrointestinal concentration estimated as the highest dose per 250 mL ([I₂])/IC₅₀ >10 is taken to indicate a potentially clinically relevant drug-drug interaction with P-gp inhibitors.³⁰ The [I₂]/IC₅₀ ratio for inhibition of P-gp by BI 425809 in this study was approximately 13 and therefore close to the cutoff value; however, the [I₂]/IC₅₀ >10 cutoff has been described as resulting in a high rate of false positives,³⁰ supporting that the possibility the inhibition of P-gp by BI 425809 does not result in clinically relevant inhibition in vivo. In the presence of steady-state BI 425809 25 mg once daily, AUC_0–tz for omeprazole was slightly reduced; however, because the change in AUC_0–tz was small and the adjusted gMean ratio for C_max was close to 100%, these results were not considered to demonstrate a meaningful influence of BI 425809 on CYP2C19 activity.

At a dose of 25 mg once daily, BI 425809 affected the PK characteristics of midazolam, decreasing the adjusted gMean ratio by 29.4%. The findings of the present study therefore suggest that BI 425809 treatment at a dose of 25 mg once daily leads to induction of CYP3A4. The mild induction observed in the clinical study was less extensive than the moderate-to-strong induction that was predicted using a net-effect model (data not shown). This discrepancy may be partly explained by the high level of variability in drug-drug interaction predictions using in vitro induction data, which can arise due to differing responses to the inducer in cells from different donors. In our in vitro induction assay, greater induction was observed in hepatocytes from donor 3 than in hepatocytes from donors 1 and 2. This could lead to overprediction of induction potential if the high response in hepatocytes from donor 3 is not representative of the normal physiological response to BI 425809. Furthermore, the net-effect calculation relies on conservative estimates of portal vein and gastrointestinal concentrations of the inducer, which are assumed to remain constant over the entire dosing interval. As such, the static model used in the net-effect calculation does not account for normal fluctuations in the in vivo concentrations of BI 425809, which are a normal result of dosing at regular intervals.

For concomitant administration with BI 425809 25 mg once daily, AUC_i/AUC ratios lower than 0.8 were predicted only for sensitive CYP3A4 substrates, whereas other medications including antipsychotic drugs were predicted to have no relevant change in plasma exposure. It should also be noted that BI 425809 25 mg is the highest dose currently tested in Phase II studies, representing the most extreme potential for drug-drug interactions, and it is possible that lower doses of BI 425809 will not lead to CYP3A4 induction. However, at the 25-mg dose level, the mild induction of CYP3A4 by BI 425809 might affect exposure to comedications that are sensitive CYP3A4 substrates, which could lead to a clinically relevant drug interaction, particularly for drugs with a narrow therapeutic index.

Because BI 425809 is principally metabolized by CYP3A4,¹⁰ induction of the enzyme may lead to autoinduction of BI 425809 metabolism, limiting exposure of BI 425809. However, in a Phase I study (NCT02337283) in healthy volunteers, less than linear exposure accumulation, which may be linked to autoinduction, was only observed at the highest dose group of BI 425809 75 mg twice daily.⁷ This dose is considerably higher than the anticipated highest therapeutic dose of BI 425809 25 mg once daily. This is in contrast to the reduction in exposure of midazolam in the present study, which was mild but observable after repeated dosing with BI 425809 25 mg once daily. These findings suggest that BI 425809 is a less sensitive CYP3A4 substrate than midazolam, and as such, autoinduction is not considered to influence BI 425809 PK at clinically relevant doses, although it may limit exposure in events such as overdosing.

In this study, the proportion of patients reporting AEs considered related to BI 425809 treatment, and the types of reported AEs were consistent with previous studies.^6–8 BI 425809 and the probe drugs were well tolerated, and no new safety signals were identified.

CONCLUSIONS

Once-daily BI 425809 at 25 mg is a mild inducer of CYP3A4 in vivo, as revealed by a decrease in exposure to midazolam administered on a background of steady-state BI 425809. Activity of CYP2C9, CYP2C19, and P-gp was not affected by steady-state BI 425809 25 mg. No new safety signals were identified.