The myeloperoxidase inhibitor mitiperstat (AZD4831) does not prolong the QT interval at expected therapeutic doses

Abstract

Mitiperstat is a myeloperoxidase inhibitor in clinical development for treatment of patients with heart failure and preserved or mildly reduced ejection fraction, non‐alcoholic steatohepatits and chronic obstructive pulmonary disease. We aimed to assess the risk of QT‐interval prolongation with mitiperstat using concentration–QT (C‐QT) modeling. Healthy male volunteers were randomized to receive single oral doses of mitiperstat 5, 15, 45, 135, or 405 mg (n = 6 per dose) or matching placebo (n = 10) in a phase 1 study (NCT02712372). Time‐matched pharmacokinetic and digital electrocardiogram data were collected at the baseline (pre‐dose) and at 11 time‐points up to 48 h post‐dose. C‐QT analysis was prespecified as an exploratory objective. The prespecified linear mixed effects model used baseline‐adjusted QT interval corrected for the heart rate by Fridericia's formula (ΔQTcF) as a dependent variable and plasma mitiperstat concentration as an independent variable. Initial exploratory analyses indicated that all model assumptions were met (no effect on heart rate; appropriate use of QTcF; no hysteresis; linear concentration–response relationship). Model‐predicted mean baseline‐corrected and placebo‐adjusted ΔΔQTcF was +0.73 ms (90% confidence interval [CI]: −1.73, +3.19) at the highest anticipated clinical exposure (0.093 μmol/L) during treatment with mitiperstat 5 mg once daily. The upper 90% CI was below the established threshold of regulatory concern. The 16‐fold margin to the highest observed exposure was high enough to mean that a positive control was not needed. Mitiperstat is not associated with risk of QT‐interval prolongation at expected therapeutic concentrations.

Modelled relationship between baseline‐corrected placebo‐adjusted QTcF and maximum plasma mitiperstat concentration.

Abbreviations

Δ

mean baseline‐adjusted

ΔΔ

mean baseline‐adjusted placebo‐corrected

C‐QT

concentration–QT

dECG

digital electrocardiogram

HFmrEF

heart failure with mildly reduced ejection fraction

HFpEF

heart failure with preserved ejection fraction

HR

heart rate

LOESS

locally estimated scatterplot smoothing

TQT

thorough QT

1. INTRODUCTION

Heart failure with preserved ejection fraction (HFpEF) occurs when stiffness of the left ventricle prevents adequate filling of the heart with blood. HFpEF affects approximately half of all patients with chronic heart failure. ¹ It is associated with decreased quality of life and premature mortality, and represents a significant clinical and economic burden. Despite the high prevalence of HFpEF, pharmacological advances have been generally slow and disappointing. ²

Myeloperoxidase has recently emerged as a promising therapeutic target for treatment of HFpEF. Myeloperoxidase is stored within the primary granules of neutrophils and released upon activation and degranulation. When released from cells, myeloperoxidase promotes inflammation by catalyzing the formation of reactive oxygen species. High plasma myeloperoxidase levels are associated with advanced heart failure and poor clinical outcomes. ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ , ⁹

Mitiperstat (AZD4831) is a covalent myeloperoxidase inhibitor with high selectivity at low oral doses, and is in clinical development for treatment of HFpEF, as well as non‐alcoholic steatohepatitis and chronic obstructive pulmonary disease. ¹⁰ , ¹¹ , ¹² The initial single‐dose escalation study in healthy volunteers evaluating mitiperstat doses from 5 to 405 mg showed that the exposures, AUC and C _max, increased in a dose proportional manner. ¹¹ Also, and in agreement with the long half‐life of mitiperstat of ~40–70 h, a 2 to 4‐fold accumulation was observed at steady state following once daily dosing of mitiperstat. ¹² SATELLITE was a phase 2a trial of mitiperstat safety and target engagement in 41 patients with HFpEF or heart failure with mildly reduced ejection fraction (HFmrEF). ¹³ Daily oral doses of mitiperstat 2.5 mg for 10 days followed by 5 mg for 80 days reduced myeloperoxidase specific activity by more than 50% from the baseline and by 75% versus placebo, with few treatment‐related adverse events. ¹³ The highest expected therapeutic dose of mitiperstat is 5 mg once daily, and this is the highest dose being tested in the ongoing phase 2b dose‐finding study in patients with HFpEF/HFmrEF (ENDEAVOR, NCT04986202). ¹⁴

Major regulatory bodies worldwide have published guidelines requiring assessment of the risk of QT interval prolongation by new drugs with systemic bioavailability. Concentration–QT (C‐QT) modeling is currently accepted as the primary method for analysis of this risk. ¹⁵ , ¹⁶ This approach requires time‐matched pharmacokinetic and digital electrocardiogram (dECG) data for input into a prespecified linear mixed effect model. ¹⁶ Data for C‐QT modeling can be collected from an early clinical study (e.g., a first‐time‐in‐human study), provided the dECG data are of sufficiently high quality. Because C‐QT modeling utilizes data from multiple doses and time points, it requires relatively small number of participants. It has therefore become an attractive alternative to the more conventional intersection–union test analysis, which requires a larger number of participants and which was historically used in dedicated “thorough” QT (TQT) studies. ¹⁷ , ¹⁸ , ¹⁹ TQT studies require a positive control, can be challenging to perform, and are larger, less time‐efficient and less cost‐effective than C‐QT modeling assessment using the existing data. ¹⁸

The aim of the present analysis was to assess QT prolongation risk associated with mitiperstat exposure using C‐QT modeling. To assess this risk, baseline‐adjusted and placebo‐corrected QT interval was derived for the highest expected clinical mitiperstat exposure level, using data from a published phase 1 study in healthy male volunteers. ¹¹

2. METHODS

2.1. Study design

This was an analysis of data from the single ascending‐dose part of a phase 1, randomized, single‐blind, placebo‐controlled, parallel group study of the safety, tolerability, pharmacokinetics, and pharmacodynamics of mitiperstat in healthy male volunteers. The study was prospectively designed as a TQT substitute, and the present C‐QT analysis was specified as an exploratory study objective.

The study was completed in 2016 and complete methods and results have been previously published. ¹¹ Briefly, healthy volunteers were randomized to receive single oral doses of mitiperstat 5, 15, 45, 135 or 405 mg (n = 6 per dose) or matching placebo (n = 10). The main inclusion criteria were male sex, age of 18–50 years, bodyweight of 50–100 kg, and a body mass index of 18.0–29.9 kg/m².

The study was performed in accordance with the principles of the Declaration of Helsinki, the International Council for Harmonisation Good Clinical Practice, and the AstraZeneca policy on Bioethics and Human Biological Samples. All participants provided written informed consent before starting the study. The study was registered on ClinicalTrials.gov (NCT02712372).

2.1.1. Pharmacokinetics and electrocardiography

Twelve time‐matched pharmacokinetic blood samples and dECG recordings were collected from all 40 participants, at the baseline (pre‐dose; nominal t = 0) and 0.5, 1, 2, 3, 4, 6, 8, 12, 24, 36, and 48 h after dosing. All dECG recordings were collected immediately before pharmacokinetic sampling.

Twelve‐lead continuous dECG were recorded with a Schiller Cardiovit CS‐200 recorder (Schiller AG, Baar, Switzerland) for at least 10 min at the baseline and at least 5 min at post‐dose time points. The same recorder was used for each participant at all time points, where possible. The AstraZeneca ECG core laboratory performed dECG analysis according to standard procedures for recording and data transmission, using EClysis© version 3.2, or higher. Ten‐second dECG recordings were extracted twice per minute and were initially analyzed automatically.

2.2. Statistical analyses

2.2.1. Initial analyses

Exploratory graphical analyses were performed before the prespecified C‐QT analysis to assess whether the following four key model assumptions were met ¹⁶ : first, that there was no effect of drug on heart rate; secondly, that correction of QT interval using Fridericia's method was appropriate; thirdly, that there was no delay between change in corrected QT interval and mitiperstat concentration (hysteresis); and finally that the relationship between mitiperstat concentration and corrected QT interval was linear.

The effect of mitiperstat on heart rate was assessed by comparing mitiperstat concentration–time profiles with mean baseline‐adjusted heart rate (ΔHR) and mean baseline‐adjusted placebo‐corrected heart rate (ΔΔHR). Assessment of QT correction method was performed by visual inspection of scatterplots of interval durations for RR versus uncorrected QT interval, QT interval corrected for heart rate by Bazett's formula and QT interval corrected for heart rate by Fridericia's formula (QTcF). The time delay between plasma mitiperstat distribution and QTcF was evaluated using a hysteresis plot comparing plasma mitiperstat concentration with baseline‐adjusted placebo‐corrected QTcF (ΔΔQTcF), stratified by dose. Linearity of the C‐QT relationship was assessed using linear regression and locally estimated scatterplot smoothing (LOESS).

QT, PR, and QRS intervals were also analyzed categorically to assess outlier data points (QT >450 ms, QT >480 ms, QT >500 ms, ΔQT >30 ms, ΔQT >60 ms, QT >450 ms and ΔQT >30 ms, QT >500 ms and ΔQT >60 ms; the same thresholds for QTcF; PR >200 ms; and QRS >110 ms).

2.2.2. C‐QT model

The prespecified C‐QT analysis was a linear mixed effects model using the baseline‐adjusted QTcF (ΔQTcF) as a dependent variable, the time‐matched plasma mitiperstat concentration as an independent variable, and nominal time, treatment group and baseline QTcF as additional factors (Equation 1). A normal distribution was assumed for random effects, with mean [0, 0] and an unstructured covariance matrix.

$$\begin{array}{c}∆{\text{QTcF}}_{\mathit{ijl}}=\left({\theta }_0+h_{0,i}\right)+{\theta }_1\mathrm{TR}{\mathrm{T}}_j+\left({\theta }_2+h_{2,i}\right)C_{\mathit{ijl}}\hfill \\ +{\theta }_3\mathrm{TIM}{\mathrm{E}}_l+{\theta }_4\left({\text{QTcF}}_{\mathit{ijl}=0}-\overline{{\text{QTcF}}_0}\right)\hfill \end{array}$$ (1)

In Equation 1, $∆{\text{QTcF}}_{\mathit{ijl}}$ is the change from the baseline in QTcF for participant i in treatment group j (0, placebo; 1, mitiperstat) at time l; ${\theta }_0$ is the population‐mean intercept in the absence of a treatment effect; ${\theta }_1$ is the fixed effect associated with treatment; ${\theta }_2$ is the population‐mean slope of the assumed linear association between concentration and$∆{\text{QTcF}}_{\mathit{ijl}}$; $h_{2,i}$ is the random effect associated with the slope ${\theta }_2$; $C_{\mathit{ijl}}$ is the concentration for i in j at l; ${\theta }_3$ is the fixed effect associated with time; ${\theta }_4$ is the fixed effect associated with baseline ${\text{QTcF}}_{\mathit{ijl}=0}$; and $\overline{{\text{QTcF}}_0}$ is the mean of all baseline line QTcF values.

ΔΔQTcF was derived according to Equation 2 as the difference between model‐derived at a given mitiperstat concentration versus placebo (i.e., zero mitiperstat concentration).

$$\begin{array}{c}\text{Mean}∆∆\text{QTcF}=\text{Mean}\left(∆{\text{QTcF}}_{\mathit{ij}}|j=1,C_{\mathit{ij}}=C\right)\hfill \\ -\text{Mean}\left(∆{\text{QTcF}}_{\mathit{ij}}|j=0,C_{\mathit{ij}}=0\right)\hfill \end{array}$$ (2)

All analyses were performed using R (version 3.5.1) and lme4 (version 1.1‐7).

2.2.3. High clinical exposure scenario

Model‐derived ΔΔQTcF was estimated for “high clinical exposure” to mitiperstat. According to current guidance from the International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use (ICH), “high clinical exposure” is the C _max at steady state, achieved when the maximum therapeutic dose is administered in the presence of the intrinsic or extrinsic factor that has the largest effect on increasing the C _max. ¹⁵ The expected therapeutic dose of mitiperstat is no higher than 5 mg once daily, administered as oral tablets, and this is the highest dose being tested in the ongoing phase 2b dose‐finding study (ENDEAVOR; NCT04986202). ¹⁴ The “high clinical exposure” scenario for this dose at steady state, as defined in the ICH E14/S7B guidance questions and answers document (FDA 2022) is 0.093 μmol/L. This scenario considers the impact of renal impairment on mitiperstat exposure. Renal elimination is the major route of mitiperstat excretion, with 32%–44% of the dose excreted unchanged in urine. ²⁰ In patients with moderate to severe renal impairment, exposure is expected to increase by up to two‐fold compared with healthy volunteers, therefore 5 mg dose in patients is expected to have exposure similar to that after 10 mg in healthy volunteers. In the phase 1 study NCT03136991, ¹² C _max at the steady state was 0.093 μmol/L after 14 days treatment with mitiperstat 10 mg in healthy volunteers. This value was therefore chosen for the “high clinical exposure” scenario.

2.3. Nomenclature of targets and ligands

Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY, ²¹ and are permanently archived in the Concise Guide to PHARMACOLOGY 2023/24. ²²

3. RESULTS

3.1. Patients

Forty healthy white male volunteers aged 19–50 years were enrolled and randomized, all of whom received one dose of mitiperstat (n = 30) or placebo (n = 10) and completed the single ascending dose part of the study. Baseline demographics have been previously reported. ¹¹ The present analysis included data from all 40 participants.

3.2. Outlier evaluation

Categorical evaluation of QTcF, PR and QRS interval measurements in the 40 participants did not reveal significant outliers in the data. QT and QTcF intervals did not exceed 450 ms in any participant at any time point and change from the baseline in these measurements exceeded 30 ms in only one participant at one time point. One participant in the mitiperstat 135 mg had a PR interval above 200 ms. Three participants had QRS intervals above 100 ms, one each in the mitiperstat 5, 45 and 405 mg groups.

3.3. C‐QT model assumptions

All four assumptions of the prespecified linear mixed effects model were met, based on initial exploratory analysis. First, no significant impact of mitiperstat on heart rate was observed when comparing plasma mitiperstat concentration–time profiles with ΔHR and ΔΔHR (Figure 1; Figure S1), with mean ΔHR and ΔΔHR remaining below 10 bpm. Secondly, scatterplot analyses supported use of the Fridericia correction for QT interval (Figure S2). Thirdly, no evidence of hysteresis was found in plots of potential time delay between mitiperstat concentration and ΔΔQTcF (Figure S3), supporting the assumption of a direct relationship. Finally, graphical assessment of mitiperstat concentration versus ΔΔQTcF showed that the LOESS line overlaid the linear regression line, indicating a linear relationship (Figure S4).

FIGURE 1.

Time courses of baseline‐adjusted QTcF, heart rate, and plasma mitiperstat concentration. Points connected with lines denote mean values within one dosing arm; treatment and placebo arms are marked with different colors as described in the figure legend; error bars denote 90% CI for heart rate measurements and mean ± SD for plasma drug concentration. ΔHR: baseline‐adjusted heart rate; ΔQTcF, baseline‐adjusted QT interval corrected by Fridericia; CI, confidence interval.

3.4. Concentration—QT model

The model estimated a statistically nonsignificant relationship between mitiperstat concentration and ΔQTcF, with a slope of 1.684 ms/μmol/L (95% confidence interval [CI]: −0.393, 3.760). Model parameters, parameter estimates and their precision are reported in Table S1. Treatment group and slope for drug concentration were not identified as significant factors affecting ΔQTcF. Diagnostic plots evaluating the quality and robustness of the final model showed no indication of bias across concentration, baseline QTcF, nominal time, or study treatment (mitiperstat or placebo), and no major abnormalities were detected in the distribution of the residuals (Figure S5). This suggests that the final model was robust and adequately described the data.

3.5. Model predictions

Model‐derived mean ΔΔQTcF was +0.73 ms (90% CI: −1.73, +3.19) for the “high clinical exposure” scenario in patients with HFpEF (Table 1; Figure 2), based on a therapeutic dose of mitiperstat 5 mg/day and a C _max of 0.093 μmol/L at steady state (exposure in severe renal impairment). The upper bound of the two‐sided 90% confidence interval did not exceed the 10 ms threshold of regulatory concern. ¹⁶ Model‐derived ΔΔQTcF across the mitiperstat exposure range tested is shown in Figure 2. Model‐derived ΔΔQTcF under the “high clinical exposure” scenario is shown in Table 1.

TABLE 1.

Model‐derived estimates of the baseline‐adjusted QTcF and baseline‐adjusted placebo‐corrected QTcF.

Exposure	Mitiperstat dose	C max (μmol/L)	ΔQTcF, mean (90% CI)	ΔΔQTcF, mean (90% CI)
Daily therapeutic dose a	5 mg	0.093	−0.41 (−1.65, +0.84)	+0.73 (−1.73, +3.19)
Single supra‐therapeutic dose	405 mg	3.040	+4.55 (−0.24, +9.34)	+5.69 (+1.44, +9.94)

Abbreviations: CI, confidence interval; C _max, maximal plasma concentration at the steady state; HFpEF, heart failure with preserved injection fraction; ΔQTcF, baseline‐corrected QTcF; ΔQTcF, baseline‐corrected placebo‐adjusted QTcF; QTcF, QT interval corrected by Fridericia.

^a “High clinical exposure” scenario: maximum anticipated clinically relevant dose, accounting for renal impairment in patients with HFpEF.

FIGURE 2.

Modeled relationship between the baseline‐corrected placebo‐adjusted QTcF and maximum plasma mitiperstat concentration. Blue line with shaded area denotes mean model estimates with 90% CI; blue arrow represents the upper 90% CI of the estimated baseline and placebo‐corrected QT effect under the “high clinical exposure” scenario (maximum anticipated clinically relevant dose, accounting for renal impairment in patients with HFpEF). ΔΔQTcF: baseline‐adjusted and placebo corrected QT interval corrected by Fridericia; CI, confidence interval; QD, once daily.

A single dose of 405 mg in healthy volunteers resulted in an observed geometric mean C _max of 3.037 μmol/L. This is approximately 16 times higher than the steady‐state C _max under the “high clinical exposure” scenario above. Model‐derived mean ΔΔQTcF for the supra‐therapeutic dose was +5.69 (90% CI: +1.44, +9.94) (Table 1). ¹⁶

4. DISCUSSION

This prespecified C‐QT analysis of matched pharmacokinetic and dECG data from a phase 1 study revealed that once‐daily oral mitiperstat treatment does not prolong QT interval at clinically relevant exposure levels. Estimates of the effect of mitiperstat on ΔΔQTcF fell well within regulatory thresholds of concern.

The prespecified linear mixed‐effects model of the relationship between plasma mitiperstat concentration and ΔQTcF adequately described the observed data and met the published criteria for quality and robustness, according to the goodness‐of‐fit plots. ¹⁶ Furthermore, exploratory analyses indicated that all four key model assumptions were met (no effect on heart rate; QTcF appropriate method; no hysteresis; linear relationship). We therefore consider that the model was appropriate and fit for purpose.

We based “high clinical exposure” to mitiperstat on a maximum daily dose of 5 mg in patients with HFpEF/HFmrEF. In the ongoing phase 2b dose‐finding study (ENDEAVOR [NCT04986202], in which 2.5 and 5 mg doses are being tested), mitiperstat 5 mg once daily is the highest dose under evaluation. ¹⁴ Results from this study will be used to guide phase 3 dose selection. In the previous phase 2a study (SATELLITE), mitiperstat 5 mg once daily reduced myeloperoxidase specific activity by approximately 50% versus the baseline and approximately 70% versus placebo after 90 days. ¹³ The present “high clinical exposure” scenario assumes that renal impairment will increase mitiperstat C _max by up to two‐fold. Based on in vitro data, the risk for drug–drug interaction through CYP inhibition or transporter inhibition by mitiperstat is considered low (data on file, AstraZeneca, Gothenburg, Sweden). A clinical drug–drug interaction study with a co‐administered strong CYP3A4 inhibitor (itraconazole) showed an increase of approximately 30% in mitiperstat C _max. ²¹ Based on the high renal clearance and the absorption, distribution, metabolism, and elimination characteristics of mitiperstat, ²⁰ hepatic impairment is expected to have less impact than renal impairment (supported by predictions from a PBPK model; data on file, AstraZeneca, Gothenburg, Sweden). Given that no other factors are currently expected to increase the exposure by a greater factor than two‐fold, it can be concluded that “high clinical exposure” was adequately defined.

The upper bound of the two‐sided 90% confidence interval for model‐predicted ΔΔQTcF at the high clinical exposure was below the threshold for regulatory concern (10 ms). The highest dose tested in the present single‐dose study and included in the C‐QT model was mitiperstat 405 mg, which resulted in observed C _max values approximately 16 times higher than in the “high clinical exposure” scenario. The highest exposure included in the model is therefore considered to be a sufficiently high multiple of the clinically relevant exposure to justify the absence of a positive control. ¹⁹ It can therefore be concluded that mitiperstat does not prolong QT interval at clinically relevant exposure levels. Results from the C‐QT modeling presented here were shared and approved by the regulatory authorities (the US Food and Drug Administration and the European Medicines Agency) as a part of the TQT waiver for mitiperstat. The results of the present C‐QT analysis agree with pre‐clinical results on inhibition of Kv11.1 (hERG)‐containing potassium channels (encoded by human Ether‐à‐go‐go‐Related Gene [KCNH2]), which are essential for normal electrical activity of the heart. Mitiperstat did not inhibit hERG, except at concentrations corresponding to more than 700 times the predicted maximum clinically relevant systemic exposure (AstraZeneca data on file).

Strengths of the present analysis include pre‐specification of the C‐QT model as an exploratory efficacy objective of the study, conduct of the C‐QT modeling according to published guidelines, the design of the study to include time‐matching of the pharmacokinetic sampling and dECG recording, and the use of a broad dose range. Although the analysis presented here did not include steady‐state data, the exposures achieved in the single‐dose study were sufficiently high to provide the required exposure margin. Overall, this C‐QT analysis provides a robust assessment of QT prolongation risk without the need for a dedicated TQT study, in accordance with regulatory guidelines.

AUTHOR CONTRIBUTIONS

J.P, D.R., K.N., H.E., A.E., C.D., and M.S. participated in the research design. J.S. and M.S. performed the data analysis. J.P., J.S., D.R, K.N., H.E., A.E., C.D., and M.S. wrote or contributed to the writing of the manuscript.

CONFLICT OF INTEREST STATEMENT

JP, DR, KN, HE, AE, CD, and MS are the employees of AstraZeneca and own stock or stock options.

ETHICS STATEMENT

The study took place at the PAREXEL Early Phase Clinical Unit in Berlin, Germany. It was conducted in accordance with the principles of the Declaration of Helsinki and the International Conference on Harmonization and Good Clinical Practice. An independent ethics committee (Landesamt für Gesundheit und Soziales Berlin Geschäftsstelle der Ethikkommission des Landes Berlin) and institutional review board reviewed and approved the study protocol and its amendments. All participants freely gave their written informed consent before starting the study.

Supporting information

Appendix S1:

Parkinson J, Sundell J, Rekić D, et al. The myeloperoxidase inhibitor mitiperstat (AZD4831) does not prolong the QT interval at expected therapeutic doses. Pharmacol Res Perspect. 2024;12:e1184. doi: 10.1002/prp2.1184

DATA AVAILABILITY STATEMENT

Data underlying the findings described in this manuscript may be obtained in accordance with AstraZeneca's data sharing policy described at https://astrazenecagrouptrials.pharmacm.com/ST/Submission/Disclosure.