Drug–Drug Interaction Potential of Mavacamten with Midazolam: Combined Results from Clinical and Model‐Based Studies

Abstract

Mavacamten is a potential inducer of cytochrome P450 (CYP) 3A4 and could reduce the effectiveness of concomitant drugs that are metabolized by CYP3A4, such as midazolam. This study aimed to determine if repeat doses of mavacamten achieving clinically relevant exposures affected midazolam exposure. This was a single‐center, open‐label study in healthy participants. Participants received: on day 1, midazolam 5 mg; on days 2‐3, mavacamten 25 mg; on days 4‐16, mavacamten 15 mg; and on day 17, mavacamten 15 mg and midazolam 5 mg. Plasma concentrations of mavacamten, midazolam, and the midazolam metabolite 1′‐hydroxymidazolam were measured. A physiologically based pharmacokinetic (PBPK) model was used to simulate the effect of mavacamten‐mediated CYP3A4 induction on midazolam exposure by CYP2C19 phenotype. Thirteen adult participants were enrolled (46.2% were female; mean [SD] age: 34.0 [9.0] years). Compared with midazolam alone, midazolam coadministered with mavacamten decreased the maximum observed plasma concentration (C_max), area under the drug concentration‐time curve (AUC) from time zero to infinity (AUC_0‐inf), and AUC from time zero to last measurable concentration (AUC_0‐last) for midazolam by 7%, 13%, and 24%, respectively; for 1′‐hydroxymidazolam, AUC_0‐inf and AUC_0Ȁlast increased by 20% and 11%, respectively. Ten participants experienced adverse events and the majority were mild in severity. The PBPK model predicted the clinical trial data well. The PBPK simulation assessed that the overall impact of mavacamten on midazolam C_max and AUC was predicted to be weak regardless of CYP2C19 phenotype. At clinically relevant exposures, mavacamten had a negligible effect on midazolam exposure.

Introduction

Hypertrophic cardiomyopathy (HCM) is a chronic myocardial disease that is characterized by left ventricular (LV) hypertrophy that cannot be explained by any other condition. ¹ , ² Approximately 70% of patients with HCM have obstructive HCM, which is characterized by LV outflow tract (LVOT) obstruction, defined as peak LVOT gradients of at least 30 mm Hg at rest or when provoked. ¹ , ² , ³ Mavacamten is a cardiac myosin inhibitor that has been approved in five continents for the treatment of adults with symptomatic New York Heart Association class II‐III obstructive HCM. ⁴ , ⁵ , ⁶ , ⁷ , ⁸ , ⁹ The estimated oral bioavailability of mavacamten is at least 85%, and mavacamten has a median time to maximum plasma concentration (t_max) of 1 to 2 h. ⁴ , ¹⁰ Mavacamten is primarily metabolized in the liver by cytochrome P450 (CYP) enzymes, including CYP2C19 (74%), CYP3A4 (18%), and CYP2C9 (8%). ⁴ CYP2C19 is polymorphic, and its drug‐metabolizing phenotypes can be classified according to gene variants and the corresponding level of metabolic activity: poor metabolizers (PMs; *2/*2, *3/*3, and *2/*3; little to no CYP2C19 activity), intermediate metabolizers (IMs; *1/*2, *1/*3, *2/*17, and *3/*17; intermediate CYP2C19 activity), normal metabolizers (NMs; *1/*1; normal CYP2C19 activity), rapid metabolizers (RMs; *1/*17; higher than normal CYP2C19 activity), and ultrarapid metabolizers (UMs; *17/*17; greatly higher than normal CYP2C19 activity). ¹¹ , ¹² Therefore, mavacamten exposure can vary depending on an individual's CYP2C19 status, resulting in a half‐life (t_1/2) of 6‐9 days in NMs and 23 days in PMs. ⁴

Preclinical in vitro studies in cryopreserved human hepatocytes indicated that at high concentrations of free drug (approximately 137 times higher than what should be encountered in clinical practice), mavacamten has the potential to be an inducer of the CYP3A4 enzyme. ¹³ Therefore, drug–drug interactions (DDIs) between mavacamten and CYP3A4 substrates may be possible. ¹³ As such, mavacamten could potentially reduce the effectiveness of concomitant drugs that are metabolized by CYP3A4. When investigating the potential of a drug as a CYP enzyme inducer, the US Food and Drug Administration (FDA) recommends use of index substrates that have well‐characterized CYP pathways of elimination, dosing regimens, safety profiles, and anticipated interaction effects with index inhibitors and inducers. ¹⁴ The benzodiazepine midazolam is recommended by regulatory authorities as a probe substrate for testing CYP3A DDIs clinically. ¹⁵ , ¹⁶ It is almost completely metabolized by the CYP3A4 and CYP3A5 enzymes, resulting in the formation of the primary metabolite 1′‐hydroxymidazolam. ¹⁷ Interaction studies of midazolam with CYP3A inhibitors (ketoconazole, voriconazole, and ritonavir) resulted in increases in midazolam exposure. ¹⁸ , ¹⁹ For example, administration with ketoconazole increased midazolam exposure by approximately 10‐14‐fold. ¹⁸ , ¹⁹ Studies with CYP3A inducers (carbamazepine, phenytoin, and rifampicin) resulted in reductions in midazolam exposure to below 10% of baseline exposure. ²⁰ , ²¹ Substantial changes in the exposure of the CYP3A substrate from DDI with CYP3A inducers or inhibitors can result in safety concerns and alter the effectiveness of the substrate. ¹⁴

This study was designed to investigate the effect of clinically relevant mavacamten exposures on the exposure to the CYP3A4 index substrate (midazolam) and to evaluate the safety and tolerability of repeated doses of mavacamten in conjunction with a single dose of midazolam in healthy participants. Using the results from this clinical DDI study, a previously developed and updated physiologically based pharmacokinetic (PBPK) model (Simcyp Simulator version 21) was validated and then used to simulate the effect of mavacamten‐mediated CYP3A4 induction on midazolam exposure by CYP2C19 phenotypes.

Methods

Ethics

This study was approved by IntegReview institutional review board (Austin, TX) before initiation. All participants provided informed consent before taking part in any study‐related procedures. The study was conducted under a US Investigational New Drug application in accordance with the ethical principles from the Declaration of Helsinki, the International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use Good Clinical Practice Guidelines, US Title 21 Code of Federal Regulations, and all applicable laws and regulations.

Design

This was a single‐center, open‐label, three‐period, fixed‐sequence DDI study in healthy participants who received a single dose of midazolam before and after a 16‐day course of mavacamten (Figure 1). Dosing in each period occurred at the same time every day. In period 1, eligible participants received a single oral dose of midazolam 5 mg on day 1. Blood samples were obtained to determine midazolam and 1′‐hydroxymidazolam plasma concentrations on day 1 before dosing and at 0.25, 0.5, 1, 2, 3, 4, 6, 8, 12, and 24 h after midazolam administration. In period 2, participants received mavacamten 25 mg orally on days 2 and 3, and mavacamten 15 mg orally on days 4‐16. Blood samples were obtained to determine mavacamten plasma concentrations before dosing and on days 2, 3, 7, 10, 13, 15, and 16. In period 3, participants received midazolam 5 mg orally and mavacamten 15 mg orally at the same time on day 17. Blood samples were obtained to determine midazolam and 1′‐hydroxymidazolam plasma concentrations on day 17 before dosing and at 0.25, 0.5, 1, 2, 3, 4, 6, 8, 12, and 24 h after midazolam administration; and at the end of study (EOS) or early termination (ET) visit. Blood samples were obtained to determine mavacamten plasma concentrations on day 17 before dosing and at 0.5, 1, 1.5, 2, 3, 4, 8, 12, and 24 h after the last mavacamten administration.

Figure 1.

Study design. ^aBlood samples were obtained for PK profile analysis of midazolam and 1′‐hydroxymidazolam before dosing and at 0.25, 0.5, 1, 2, 3, 4, 6, 8, 12, and 24 h after midazolam administration on days 1 and 17, and at the EOS or ET visit. CYP, cytochrome P450; EOS, end of study; ET, early termination; NM, normal metabolizer; PK, pharmacokinetic; RM, rapid metabolizer; UM, ultrarapid metabolizer.

Treatments

Mavacamten immediate‐release oral capsules were supplied as 10‐ and 15‐mg capsules. Midazolam syrup (2 mg/mL) was available commercially. All participants fasted overnight before administration of midazolam or mavacamten. A minimum of 10 and 8 h of fasting was required before administration of midazolam and mavacamten, respectively. Participants were also required to fast for at least 4 and 2 h after administration of midazolam and mavacamten, respectively. Water was allowed ad libitum except during the period 1 h before and 1 h after midazolam or mavacamten administration. Mavacamten and midazolam were administered orally with at least 240 mL of water. All prescription medications were prohibited from 28 days before screening through to the termination visit. All over‐the‐counter medications (including herbal preparations and supplements, but not including acetaminophen at a dosage of ≤1.5 g/day) were prohibited from 14 days before screening through to the termination visit.

Participants

Healthy volunteers aged 18‐60 years were recruited for the study. In total, up to 14 participants were to be enrolled to account for dropouts. A sample size of 12 participants provided 90% probability for a two‐sided 90% CI for the geometric mean ratio (GMR) of area under the drug concentration‐time curve (AUC) and maximum observed plasma concentration (C_max) to be within the bounds of 70%‐143%, assuming a true GMR of 1.0 and an intrasubject coefficient of variation of 23%, based on a similar study. ²² Participants had to be NMs, RMs, or UMs, based on CYP2C19 genotyping. Key exclusion criteria included a history of clinically significant arrhythmia, LV systolic dysfunction, or coronary artery disease. A full list of inclusion and exclusion criteria is presented in Table S1.

Outcomes

The primary pharmacokinetic (PK) endpoints included treatment ratios for midazolam and 1′‐hydroxymidazolam after administration of midazolam with and without coadministration of mavacamten for AUC from time zero to infinity (AUC_0‐inf) and C_max. Secondary PK endpoints included treatment ratios for midazolam and 1′‐hydroxymidazolam after administration of midazolam alone and with coadministration of mavacamten for AUC from time zero to last measurable concentration (AUC_0‐last), percentage of AUC_0‐inf obtained by extrapolation (%AUC_ex), t_1/2, and t_max. CYP2C19 genotyping was performed at prescreening in all participants who consented. Safety was assessed from the time the participant provided informed consent to the end of the study.

Safety Assessments

Safety assessments included medical history, physical examinations, vital signs, electrocardiograms (ECGs), observed and participant‐reported adverse events (AEs), and safety laboratory results. Twelve‐lead ECG evaluations were performed supine after at least 5 min of rest at screening, on days −1, 1, 2, 7, 10, 13, 16, 17, and 18, and at the EOS or ET visit. ECG was performed before dosing at the same time each day. Additionally, on day 2, an ECG was performed at 3 h ± 30 min after dosing, and on day 17, an ECG was performed at 3 h ± 30 min and 12 h ± 2 h after dosing. A complete physical examination was conducted at screening only, and abbreviated physical examinations were conducted on days −1 and 18, and at the EOS or ET visit.

All safety data were reported descriptively. AEs were coded using the Medical Dictionary for Regulatory Activities (MedDRA) Version 22. Treatment emergent AEs (TEAEs) were defined as AEs with onset on or after the first dose of study drug, or with an onset before the first dose of study drug that increased in severity on or after the first dose of study drug. AEs were summarized by period.

Simulations Using a PBPK Model

A PBPK model was previously developed using Simcyp Simulator version 19 (Certara, Princeton, NJ) to investigate the impact of CYP inducers and inhibitors, and CYP2C19 phenotype on mavacamten exposure. ²³ In December 2021, an updated version of the PBPK software was released, Simcyp Simulator version 21 (Certara, Princeton, NJ), which implemented updates to the CYP2C19 characteristics in the adult population, including the values for hepatic and intestinal CYP2C19 enzyme abundance, changes to the existing UM population, and the addition of IM and RM phenotypes. ²⁴ The mavacamten model initially developed in Simcyp Simulator version 19 was updated in Simcyp Simulator version 21 by incorporating additional mechanistic absorption and distribution components (Simcyp Simulator M‐ADAM), and refining enzymatic clearance pathways and DDI‐related parameters. These changes to the PBPK model improved the predictions of mavacamten t_1/2 across all healthy participants, particularly CYP2C19 PMs. Mavacamten induction parameters used in the PBPK model were assessed in human hepatocyte incubations and based on observed changes in CYP mRNA. CYP3A4 EC₅₀ and E_max for mavacamten and reference inducer rifampin were determined. Final mavacamten induction parameters were based on scaling to the rifampin control (Table S2). Final parameters used in the PBPK model are reported in Table S3. Simulations with midazolam were conducted using the midazolam compound file in Simcyp Simulator version 21. The PBPK model for 1′‐hydroxymidazolam was informed by the previous published literature. ²⁵ The updated mavacamten and 1′‐hydroxymidazolam PBPK models (Simcyp Simulator version 21) were verified by comparing simulated and observed concentration‐time profiles for mavacamten and for midazolam with and without concomitant administration of mavacamten. The acceptance criteria for model performance included prediction of observed AUC_last and C_max parameters within 2‐fold and DDI ratios according to Guest et al. ²⁶

The PBPK model was first verified by comparing the PBPK parameters and the simulated concentration‐time profiles to the observed values from the clinical DDI study. The simulations for verification consisted of 10 trials each with 13 healthy participants (n = 130) with similar demographics to the clinical study. The study design and dosing regimen of the clinical study were used for the simulation. The verified PBPK model was then used to predict DDI between mavacamten and midazolam across different CYP2C19 phenotypes. In the EXPLORER‐HCM trial, the therapeutic upper threshold of mavacamten plasma concentration was set at 1000 ng/mL and treatment was interrupted if this threshold was exceeded. ²⁷ Therefore, simulations were run with a C_max of 1000 ng/mL to simulate the “worst‐case scenario” of mavacamten plasma concentration for all the phenotypes. In the simulations, a mavacamten C_max of 1000 ng/mL was achieved with the following theoretical simulated doses: PM, 15 mg; IM, 30 mg; NM, 35 mg; and UM, 100 mg (Table S4). The simulation times for IMs, NMs, and UMs were reduced compared with those for PMs owing to the variation in t_1/2 by phenotype. Simulation times were as follows: mavacamten, days 1‐81 for non‐PMs and days 1‐241 for PMs; midazolam 5 mg once daily, days 41‐81 for non‐PMs and days 121‐241 for PMs.

Statistical Analyses

Midazolam data were summarized by administration with or without mavacamten. Descriptive summary statistics were reported for continuous variables, including the mean, SD, median, minimum, and maximum. Categorical variables were summarized using counts and percentages. The PK analysis population was defined as all participants who received at least one dose of midazolam and for whom PK parameters were available and interpretable. Safety analyses were performed on the safety analysis population, which was defined as all participants who received at least one dose of midazolam/mavacamten, regardless of whether PK parameters were available. Bioanalytical methods, including determination of plasma concentrations have been previously described in detail. ¹³ PK parameters were calculated using Phoenix WinNonlin version 8.0 or later (Certara, Princeton, NJ). Midazolam and 1′‐hydroxymidazolam PK parameters were summarized using descriptive statistics for midazolam alone and for midazolam with mavacamten. The GMRs for midazolam alone, for midazolam with mavacamten, and for 1′‐hydroxymidazolam after administration of midazolam alone and with coadministration of mavacamten for the log‐transformed PK parameters (C_max, AUC_0‐inf, and AUC_0‐last) were calculated together with the 90% CI using a mixed‐effect analysis of variance model with treatment as a fixed effect and participant as a random effect.

Results

Study Population

In total, 13 healthy adult participants (female, 46.2%; mean [SD] age, 34.0 [9.0] years) enrolled in the study and were included in the PK and safety populations. Twelve participants completed the study, with one participant withdrawing on day 9 owing to personal reasons. The majority of participants were CYP2C19 NMs (61.5%) and the remaining were CYP2C19 RMs (38.5%) (Table 1).

Table 1.

Demographics and Baseline Characteristics

Parameter	Study Participants (n = 13)
Age, mean (SD), years	34.0 (9.0)
Female, n (%)	6 (46.2)
Ethnicity, n (%)
Hispanic/Latino	3 (23.1)
Non‐Hispanic/Latino	10 (76.9)
BMI, mean (SD), kg/m2	24.5 (2.8)
CYP2C19 metabolizer phenotype, n (%)
Normal (*1/*1)	8 (61.5)
Rapid (*1/*17)	5 (38.5)

BMI, body mass index; CYP, cytochrome P450.

PK Analyses

Mean plasma concentration‐time profiles for midazolam and its metabolite 1′‐hydroxymidazolam after administration of midazolam alone or with mavacamten are presented in Figure 2. Midazolam was rapidly absorbed, with a t_max of less than 1 h, and absorption was not affected by mavacamten coadministration (Table 2). The geometric means for AUC_0‐last, AUC_0‐inf, C_max, and t_1/2 for midazolam and 1′‐hydroxymidazolam are presented in Table 2. The GMR (90% CI) of midazolam alone compared with midazolam coadministered with mavacamten was 0.76 (0.61‐0.95) for AUC_0‐last, 0.87 (0.68‐1.10) for AUC_0‐inf, and 0.93 (0.77‐1.13) for C_max, corresponding to decreases of 24%, 13%, and 7%, respectively (Figure 3a). The GMR (90% CI) of 1′‐hydroxymidazolam after administration of midazolam alone compared with after coadministration of midazolam with mavacamten was 1.11 (0.96‐1.29) for AUC_0‐last, 1.20 (1.04‐1.39) for AUC_0‐inf, and 1.28 (1.00‐1.65) for C_max, corresponding to increases of 11%, 20%, and 28%, respectively (Figure 3b). AUC_0‐inf could not be determined in all PK population participants.

Figure 2.

Plasma concentration‐time profiles for (a, b) midazolam and (c, d) 1′‐hydroxymidazolam after administration of midazolam alone or with coadministration of mavacamten (PK population). Data are arithmetic mean and error bars are SD. PK, pharmacokinetic.

Table 2.

Summary of PK Parameters for Midazolam and 1′‐Hydroxymidazolam After Administration of Midazolam Alone or with Coadministration of Mavacamten (PK Population)

Parameter	Midazolam Alone (n = 13)	Midazolam + Mavacamten (n = 12)
Midazolam
Cmax, ng/mL	16.59 (21.9)	15.45 (43.3)
tmax, median (range), h	0.53 (0.48‐2.1)	0.51 (0.50‐1.0)
AUC0‐inf, ng∙h/mL	55.28 (62.4)	48.84 (47.1) (n = 9) a
AUC0‐last, ng∙h/mL	51.00 (55.7)	39.51 (55.1)
t1/2, h	5.28 (39.3)	4.84 (34.4) (n = 9) a
1′‐Hydroxymidazolam
Cmax, ng/mL	12.66 (47.3)	16.18 (37.2)
tmax, median (range), h	1.00 (0.50‐2.08)	0.51 (0.50‐1.0)
AUC0‐inf ng∙h/mL	41.47 (24.1) (n = 10) b	47.76 (36.3) (n = 7) b
AUC0‐last, ng∙h/mL	38.63 (22.2)	43.25 (38.6)
t1/2, h	6.48 (24.7) (n = 10) b	5.22 (29.9) (n = 7) b

AUC, area under the concentration‐time curve; AUC_0‐inf, AUC from time zero to infinity; AUC_0‐last, AUC from time zero to last measurable concentration; %AUC_ex, percentage of AUC_0‐inf obtained by extrapolation; C_max, maximum observed plasma concentration; CV%, coefficient of variation percentage; PK, pharmacokinetic; t_1/2, half‐life; t_max, time to peak plasma concentration.

Data are geometric mean (geometric CV%) unless otherwise stated.

^a Two participants were excluded from the analysis for AUC_0‐inf (ng∙h/mL) because “Adjusted R² < 0.8.” A third participant was excluded because the elimination phase could not be characterized for midazolam + mavacamten treatment. For the same reasons, t_1/2 could not be calculated for these participants.

^b Three participants in the midazolam alone treatment group and five participants in the midazolam + mavacamten treatment group were excluded from the analysis for AUC_0‐inf (ng∙h/mL) because “Adjusted R² < 0.8” and/or “%AUC_ex was >30%.” For the same reasons, t_1/2 could not be calculated for these participants.

Figure 3.

PK parameters for (a) midazolam and (b) 1′‐hydroxymidazolam after administration of midazolam alone or with coadministration of mavacamten (PK population). AUC, area under the concentration‐time curve; AUC_0‐inf, AUC from time zero to infinity; AUC_0‐last, AUC from time zero to last measurable concentration; C_max, maximum observed plasma concentration; GMR, geometric mean ratio; PK, pharmacokinetic.

Safety Analyses

In total, 10 participants (76.9%) experienced 14 AEs during the study, with three participants (23.1%) experiencing an AE in period 1 before mavacamten administration. The majority of AEs were considered mild in severity, and one participant (7.7%) experienced an AE of moderate severity during period 2 (Table 3). There were no serious or severe AEs reported, and no AEs resulted in discontinuation or death. There were no clinically significant abnormal findings reported from clinical laboratory tests, physical examinations, vital sign assessments, or ECGs. The most common system organ classes of reported AEs were gastrointestinal disorders (four participants reported AEs in period 2 and one participant reported AE in period 3) and nervous system disorders (two participants in period 1 and one participant in period 3).

Table 3.

Summary of Participants Who Experienced AEs (Safety Population) ^a

	Period 1 Midazolam Alone (n = 13)	Period 2 Mavacamten Alone (n = 13)	Period 3 Midazolam + Mavacamten (n = 12)
AEs, n (%)	3 (23.1)	9 (69.2)	2 (16.7)
Severity of AE, n (%)
Mild	3 (23.1)	9 (69.2)	2 (16.7)
Moderate	0	1 (7.7)	0
Severe	0	0	0
Serious AEs, n (%)	0	0	0
Related AEs, n (%)	2 (15.4)	5 (38.5)	1 (8.3)
AEs leading to treatment discontinuation, n (%)	0	0	0
AEs leading to death, n (%)	0	0	0

AE, adverse event.

^a The numbers reported in the table are the numbers of unique participants who experienced AEs.

Simulation Results

All simulated midazolam PK parameters (AUC_0‐last and C_max) and drug‐interaction ratios were within 1.3‐fold of observed values (Table S5; Figure S1). For midazolam alone, the simulated and observed values of AUC_0‐last were 53.54 and 51.00 ng∙h/mL, respectively, resulting in a simulated/observed ratio of 1.05. The simulated and observed values of AUC_0‐last for midazolam with mavacamten were 45.47 and 39.51 ng∙h/mL, respectively, resulting in a ratio of 1.15. For midazolam alone, the simulated and observed values of C_max were 19.06 and 16.59 ng/mL, respectively, giving a ratio of 1.15. For midazolam with mavacamten, the simulated and observed values of C_max were 17.14 and 15.45 ng/mL, respectively, giving a ratio of 1.11. The ratios between simulated and observed values for the GMR of midazolam with mavacamten compared with midazolam alone were 1.12 for AUC_0‐last and 0.97 for C_max.

For the metabolite 1′‐hydroxymidazolam, simulated C_max values were underpredicted; however, AUC_0‐last values were captured well. The simulated and observed values of AUC_0‐last for 1′‐hydroxymidazolam after administration of midazolam alone were 40.37 and 38.63 ng h/mL, respectively, resulting in a ratio of 1.05. The simulated and observed values of AUC_0‐last for 1′‐hydroxymidazolam after midazolam and mavacamten coadministration were 38.68 and 43.25 ng∙h/mL, respectively, giving a ratio of 0.89. For 1′‐hydroxymidazolam after administration of midazolam alone, the simulated and observed values of C_max were 8.22 and 12.66 ng/mL, respectively, giving a ratio of 0.65. For 1′‐hydroxymidazolam after midazolam and mavacamten coadministration, the simulated and observed values of C_max were 8.30 and 16.18 ng/mL, respectively, giving a ratio of 0.51. The ratios between simulated and observed values for the GMR of 1′‐hydroxymidazolam after midazolam and mavacamten coadministration compared with 1′‐hydroxymidazolam after administration of midazolam alone were 0.86 for AUC_0‐last and 0.79 for C_max.

The PBPK model predicted the determined PK parameters for midazolam alone and with mavacamten adequately; however, for the metabolite 1′‐hydroxymidazolam after administration of midazolam alone and with mavacamten, the simulations underpredicted C_max compared with the observed values. Overall, the PBPK model captured the changes in midazolam and 1′‐hydroxymidazolam exposures after midazolam and mavacamten coadministration.

To predict DDIs between mavacamten and midazolam across CYP2C19 phenotypes, a worst‐case scenario C_max concentration of 1000 ng/mL was simulated across all CYP2C19 phenotypes. A summary of doses simulated in the PBPK mavacamten perpetrator DDI simulations can be found in Table S4. The PBPK model predicted the DDI effects of mavacamten on the AUC and C_max of midazolam, in a worst‐case scenario, to be similar across all CYP2C19 phenotypes, with PMs having a numerically larger decrease in midazolam exposure (35%) than other CYP2C19 metabolizer phenotypes (29%‐31%) (Table 4). Similarly, the PBPK model predicted the effects of mavacamten on the AUC and C_max of 1′‐hydroxymidazolam to be similar across all CYP2C19 phenotypes.

Table 4.

Changes in Midazolam and 1′‐Hydroxymidazolam PK Parameters After Administration of Midazolam Alone or with Coadministration of Mavacamten in a Virtual Healthy Volunteer Population by CYP2C19 Phenotype

	Midazolam		1′‐Hydroxymidazolam
CYP2C19 Phenotype	AUC GMR a	Cmax GMR a	AUC GMR a	Cmax GMR a
PM	0.65	0.79	0.98	1.09
IM	0.69	0.82	0.96	1.05
NM	0.71	0.84	0.96	1.04
UM	0.69	0.81	0.96	1.05

AUC, area under the concentration‐time curve; C_max, maximum observed plasma concentration; CYP, cytochrome P450; GMR, geometric mean ratio; IM, intermediate metabolizer; NM, normal metabolizer; PK, pharmacokinetic; PM, poor metabolizer; UM, ultrarapid metabolizer.

^a The 90% CIs could not be calculated in Simcyp Simulator version 21.

Discussion

This study in healthy adult volunteers was conducted to assess the effect of coadministration of mavacamten with midazolam on the exposure to midazolam, a CYP34A‐sensitive substrate. Midazolam was administered the day before mavacamten dosing and again on the last day of a 16‐day course of mavacamten. In this study, clinically relevant mavacamten exposures similar to the targeted mavacamten exposures of the phase 3 EXPLORER‐HCM study were achieved. ²⁸ The US FDA recommends that if an investigational drug decreases the AUC of an index substrate by ≥80% it should be classed as a strong inducer, by ≥50% to <80% as a moderate inducer, by ≥20% to <50% as a weak inducer, or by <20% as not being an inducer. ¹⁴ Although midazolam exposures (C_max, AUC_0‐inf, and AUC_0‐last) decreased and 1′‐hydroxymidazolam exposures increased after midazolam and mavacamten coadministration, the effect of mavacamten on midazolam, an index CYP substrate, did not meet the criteria for induction. Therefore, mavacamten did not induce CYP3A4 to a clinically meaningful extent and is not likely to affect the effectiveness of drugs metabolized by CYP3A4. This is largely in agreement with the results reported in a DDI study of mavacamten with oral contraceptives (ethinyl estradiol and norethindrone). ²²

A population PK study of mavacamten demonstrated that the CYP2C19 phenotype of healthy individuals and patients with HCM affects the exposure of mavacamten. ²⁹ Individuals who are CYP2C19 PMs experience a greater level of exposure to mavacamten than CYP2C19 NMs. ⁴ The potential for mavacamten CYP3A4 induction is dependent on the level of exposure to mavacamten. ¹³ To account for CYP2C19 polymorphism, the mavacamten dosing regimen involves a titration scheme personalized to individual patient clinical responses. ⁴ , ²⁸ The aim is to achieve the same level of mavacamten exposure for all patients regardless of CYP2C19 phenotype. ²⁸ To reflect that mavacamten dosing is titrated in the clinical setting, the PBPK simulations used the same mavacamten exposure across all CYP2C19 phenotypes. The upper threshold of the therapeutic range of mavacamten 1000 ng/mL was used to reflect the worst‐case scenario assessment of mavacamten as a CYP3A4 inducer. ²⁸ The DDI simulations were completed in CYP2C19 PM, IM, NM, and UM populations with a simulation time of up to 241 days to ensure steady‐state mavacamten exposure was achieved. Using worst‐case scenario levels of exposure, the PBPK model simulations predicted minimal effects of mavacamten on the exposure of midazolam and 1′‐hydroxymidazolam regardless of CYP2C19 phenotype. In a similar study, assessing the effects of mavacamten on the exposure of oral contraceptives, the overall impact of mavacamten was predicted to be minimal across all CYP2C19 phenotypes. ²²

Overall, the PBPK model simulation results were similar to the data reported in the present DDI study; however, the model predicted a small decrease in AUC for 1′‐hydroxymidazolam, as opposed to the increase observed in the present DDI study. This suggested that the PBPK model overestimated the impact of mavacamten on the metabolite 1′‐hydroxymidazolam, and, therefore, that the performance of the metabolite PBPK model needs further improvement.

Mavacamten treatment was well tolerated during the study. The majority of AEs were mild in severity and only one participant experienced an AE of moderate severity. There were no reported serious or severe AEs, AEs leading to treatment discontinuation, or deaths during the study. The most common system organ classes of reported AEs were gastrointestinal and nervous system disorders. There were no clinically significant abnormal vital signs, physical examination findings, laboratory test results, or ECG findings reported during the study.

The results of this study should be interpreted in the context of several limitations. The PBPK model did not consider the induction of any CYP2C enzymes, which are regulated by the same nuclear receptor as CYP3A4; however, the effects of mavacamten on these enzymes are expected to be small. ¹⁴ Additionally, verifying the modeling data for CYP2C19 PMs would be challenging owing to the rarity of PMs in the general population. ¹²

Conclusions

This DDI study demonstrated that therapeutically relevant exposures of mavacamten with coadministration of the index substrate (midazolam) did not reduce midazolam exposure to a clinically significant level that may result in a decrease in effectiveness. Additionally, coadministration of midazolam with mavacamten was well tolerated, with the majority of AEs being mild in severity. The PBPK modeling also confirmed that mavacamten had a negligible impact on midazolam exposure, regardless of CYP2C19 phenotype.

Conflicts of Interest

Samira Merali, Caroline Sychterz, Lu Gaohua, Victoria Florea, and Bindu Murthy are employees of Bristol Myers Squibb and own stocks of Bristol Myers Squibb. Vidya Perera is a former employee of Bristol Myers Squibb.

Funding

The study was funded by MyoKardia, Inc., a wholly owned subsidiary of Bristol Myers Squibb.

Supporting information

Supporting Information

Data Availability Statement

Bristol Myers Squibb's policy on data sharing may be found at https://www.bms.com/researchers‐and‐partners/independent‐research/data‐sharing‐request‐process.html