Effect of axitinib on the QT interval in healthy volunteers - Springer Link

2 downloads 32 Views 391KB Size Report
Jan 15, 2015 - Cancer Chemother Pharmacol (2015) 75:619–628. DOI 10.1007/s00280-015-2677-z. ORIGINAL ARTICLE. Effect of axitinib on the QT interval ...
Cancer Chemother Pharmacol (2015) 75:619–628 DOI 10.1007/s00280-015-2677-z

ORIGINAL ARTICLE

Effect of axitinib on the QT interval in healthy volunteers Ana Ruiz‑Garcia · Brett E. Houk · Yazdi K. Pithavala · Melvin Toh · Nenad Sarapa · Michael A. Tortorici 

Received: 2 December 2014 / Accepted: 4 January 2015 / Published online: 15 January 2015 © Springer-Verlag Berlin Heidelberg 2015

Abstract  Purpose  Axitinib is a potent and selective inhibitor of vascular endothelial growth factor receptors 1–3, approved for second-line treatment of advanced renal cell carcinoma (RCC). Preclinical studies did not indicate potential for axitinib-induced delayed cardiac repolarization. Methods  The effect of axitinib on corrected QT (QTc) prolongation was evaluated with one-stage concentration– QTc response modeling using data from a definitive randomized crossover QT phase I study in healthy volunteers administered one single 5-mg axitinib dose alone or in the presence of steady-state ketoconazole (400 mg once daily). Results  Axitinib and ketoconazole had opposite effects on heart rate: Axitinib lowered it, ketoconazole raised it. The final analysis showed a flat relationship between QTc and axitinib concentration (slope −0.0314 ms·mL/ng) for axitinib

alone. Mean highest placebo-matched change from baseline in QTc was −3.0 [90 % confidence interval (CI) −5.4, −0.6] ms. At supratherapeutic axitinib exposures achieved with potent cytochrome P450 3A4/5 inhibition by ketoconazole, the model predicted mean QTc change of 6.5 (90 % CI 4.4–8.5) ms. The slope population mean estimate was −0.331 (95 % CI −0.860, 0.198) ms·mL/µg for ketoconazole alone and 0.0725 (0.0445–0.1005) ms·mL/ng for axitinib in the presence of ketoconazole. The results were then compared with those obtained based on more widely used Fridericia’s, Bazett’s, and study-specific correction methods. Conclusions  Since axitinib plasma concentrations observed in this study exceeded the range of concentrations observed in patients with RCC at the highest approved clinical dose (10 mg twice daily), axitinib was not associated with clinically significant QTc prolongation in target populations.

M. A. Tortorici was employed at Pfizer Inc during the time of this study and development of the manuscript.

Keywords  Axitinib · QTc interval · Concentration–QTc response modeling · Pharmacokinetics · QT correction for heart rate

Electronic supplementary material  The online version of this article (doi:10.1007/s00280-015-2677-z) contains supplementary material, which is available to authorized users. A. Ruiz‑Garcia · B. E. Houk · Y. K. Pithavala · M. A. Tortorici  Clinical Pharmacology, Pfizer Inc, San Diego, CA, USA M. Toh  CK Life Sciences Int’l (Holdings) Inc, Hong Kong, China N. Sarapa  Clinical Pharmacology‑Oncology, Bayer Healthcare Pharmaceuticals, Montville, NJ, USA M. A. Tortorici (*)  Clinical Pharmacology and Pharmacometrics, CSL Behring Biotherapies for Life™, 1020 First Avenue, King of Prussia, PA 19406, USA e-mail: [email protected]

Introduction Some pharmacological agents are known to cause a delay in ventricular repolarization, which is associated with a risk of torsade de pointes [1]. The International Conference on Harmonization (ICH) E14 guidance [2] recommends that all new drugs be evaluated for effects on cardiac repolarization in a well-controlled clinical study in healthy volunteers early in the clinical development. Although previously exempted, collection of robust corrected QT (QTc) interval data from a dedicated QTc study [so-called hybrid thorough QT/QTc (TQT) study in cancer patients] is generally required in the registration dossier for most nonadjuvant anticancer agents (NAAs).

13

620

Axitinib, a potent and selective inhibitor of vascular endothelial growth factor receptors 1, 2, and 3 [3], is approved for second-line treatment of advanced renal cell carcinoma (RCC) [4]. At the recommended starting dose of 5 mg twice daily, which is the maximum tolerated dose [5], axitinib is generally well tolerated. The common adverse events reported in the pivotal phase III study in patients with metastatic RCC (mRCC) were diarrhea (55 %), hypertension (40 %), and fatigue (39 %) [6], which are observed class effects of anti-angiogenic tyrosine kinase inhibitors [7]. Axitinib is metabolized primarily in the liver by cytochrome P450 (CYP) 3A4/5 and, to a lesser extent, by CYP1A2, CYP2C19, and uridine diphosphate glucuronosyltransferase 1A1 [8]. The two major circulating metabolites in humans, N-glucuronide and sulfoxide, are considered pharmacologically inactive [8]. Preclinical studies did not indicate a potential for axitinib-induced QT prolongation. In the in vitro human ethera-go-go-related gene (hERG) assay, axitinib resulted in 7,500-fold higher than the unbound peak plasma concentration (0.155 ng/mL, derived from the estimated median steady-state peak plasma concentration of 30.9 ng/ mL) for the highest approved dose of 10 mg twice daily in patients (data on file, Pfizer Inc, San Diego, CA, USA). In a telemeterized dog cardiovascular safety study, dysrhythmia, change in waveform morphology, or QTc prolongation was not observed following administration of single oral axitinib doses up to 30 mg/kg (unbound mean plasma concentration 4.5 ng/mL 2-h post-dose), which provides an approximate 29-fold margin over the free plasma concentration observed in patients (data on file, Pfizer Inc, San Diego, CA, USA). The effect of axitinib on QT/QTc prolongation was evaluated in a definitive randomized crossover phase I study. A single 5-mg oral dose of axitinib was administered alone or in the presence of steady-state ketoconazole, a strong inhibitor of metabolism via CYP3A4/5, in healthy volunteers [9]. The study demonstrated a twofold increase (from geometric mean 196.7 to 404.8 ng·h/mL) in axitinib area under the plasma concentration–time curve from 0 to infinity and a 1.5-fold increase (from geometric mean 51.0 to 76.7 ng/ mL) in the axitinib maximum observed plasma concentrations (Cmax) in the presence of ketoconazole. In this report, the effect of axitinib on QT intervals corrected for heart rate using a linear mixed-effects model with on-treatment data [10, 11] are presented. In addition, the results are compared with those obtained using either fixed correction methods or baseline-generated QT correction methods, extensively described in the literature [12–14], to assess the impact of different mathematical approaches in characterizing QTc prolongation.

13

Cancer Chemother Pharmacol (2015) 75:619–628

Methods Subjects and study design This was a randomized single-blind, two-way crossover phase I study conducted in healthy volunteers. The details of study design and subjects have previously been provided [9]. The study was approved by the Institutional Review Board (Research Consultants’ Review Committee, Austin, TX, USA) and was carried out in compliance with the Declaration of Helsinki, the ICH Good Clinical Practice guidelines, and applicable federal and local regulatory requirements. Written informed consent was obtained from each subject before study entry. Treatment Study treatments were previously described in detail [9] and outlined in the study design (Online Resource 1). In brief, a total of 35 healthy volunteers were randomly assigned to one of the two different treatment sequences: [A → B] sequence (n = 20) or [B → A] sequence (n = 15), with at least a 14-day wash-out period between treatments. A computer-generated randomization schedule was used to assign subjects to treatment sequences in order to exclude time-dependent changes over the study days. All subjects received placebo on day −1 in the absence of study medication. In treatment A, subjects were administered a single 5-mg oral dose of axitinib on day 1. In treatment B, subjects were administered ketoconazole 400 mg orally once daily on days 1–7 with a single 5-mg oral dose of axitinib on day 4. Axitinib or placebo was administered in the morning after overnight fasting (≥8 h) in a single-blinded manner (blinded only to subjects). Ketoconazole was administered with breakfast in the morning, except on day 4 of treatment B, when it was administered simultaneously with axitinib after overnight fasting. In order to standardize conditions, all subjects were required to refrain from lying down [except for taking blood pressure (BP), pulse rate, and electrocardiogram (ECG) measurements], eating, and drinking beverages other than water during the first 4 h after dosing. Water could be consumed ad libitum. Subjects were to abstain from using prescription or nonprescription drugs, vitamins, and dietary supplement within 14 days before the first dose of study medication and throughout the study. All concomitant medication taken during the study was to be recorded. Strict experimental conditions were implemented and controlled for during each day of the study when QT assessments were made. For example, the QT measurements were controlled for the time of day on each study day. Additionally, standardized ECG machines were provided to the single clinical site where this study was performed. The QT data were transmitted on a daily

Cancer Chemother Pharmacol (2015) 75:619–628

basis from the clinical site to the central ECG laboratory [Bio Medical Systems (BMS), Maryland Heights, MO, USA] commissioned for this study. The raw ECG tracings from the study were also uploaded to the ECG warehouse for the US Food and Drug Administration (FDA). Volunteers were scheduled for QT collections, BP measurements as well as pharmacokinetic assessments in a controlled fashion. Assessments ECGs were collected after subjects had rested in a supine position for 5 min. Time-matched 12-lead ECGs in triplicate (2 min apart) were obtained at day −2 (baseline) and 1, 2, and 3 h after a single dose of placebo at day −1 (placebo), after a single dose of axitinib at day 1 of treatment A (axitinib alone), and after multiple doses of ketoconazole at day 3 (ketoconazole alone) and after multiple doses of ketoconazole plus a single dose of axitinib at day 4 (axitinib plus ketoconazole) of treatment B. For each subject at each time point, QTc interval values were calculated as the average of triplicate measurements. A centralized ECG collection system provided by BMS was utilized for the study. Standardized ECG machines (GE Marquette MAC 1200, GE Healthcare, Wauwatosa, WI, USA) with consistent algorithms and software were supplied to the single clinical site by BMS, and the digital ECG recordings were transmitted electronically to BMS on a daily basis for measurement of RR and QT intervals. However, the principal investigator of the study was required to review every ECG reading for machine errors to minimize the likelihood of false results. Machine-read ECGs were used for the analysis. For determination of axitinib plasma concentrations, blood samples were collected from subjects in a controlled fashion and assayed as described previously using the validated protocol [9]. Data and statistical analyses All subjects who received at least one dose of study medication and had, at a minimum, one set each of triplicate baseline and post-baseline ECG measurements and pharmacokinetic data were included in the analysis. Because QT interval is heart rate dependent, correction methods used to remove the influence of heart rate will impact the results of drug concentration–QTc response analyses. In this study, the following approach was taken to evaluate the effect of axitinib on QTc prolongation: (1) assess the effect of axitinib and ketoconazole on heart rate by analyzing the relationship between drug plasma concentrations and RR interval, (2) if a correlation exists, characterize the relationship between drug plasma concentration and change in QT interval corrected for heart rate using a

621

one-stage QT correction method (QTcOS), and (3) assess the adequacy of the final model and compare the results with those obtained with a two-stage approach in which QT interval was corrected for heart rate using commonly used fixed (i.e., Fridericia’s and Bazett’s) or baseline-generated (i.e., study-specific) correction methods. Another common correction method calculated in each individual in the study (QTcI) was considered unsuitable and not used here since the current study had too few ECG data in each subject over a relatively short period of time (3 h) following the administration of drugs, leading to an insufficiently broad range of observed heart rate. Triplicate time-matched ECG measurements were collected on day 3 of administering ketoconazole alone (at steady state following 400 mg once daily dosing), but ketoconazole plasma concentrations were not measured in this study. Therefore, in order to characterize the influence of ketoconazole on RR and QT intervals, ketoconazole steady-state plasma concentrations were simulated using a nonlinear mixed-effect modeling approach (NONMEM version 7.1.2; ICON Development Solutions, Hanover, MD, USA). Population pharmacokinetic parameters for the predictions were obtained from a published population analysis, where data were pooled from five clinical studies with ketoconazole administered orally in single and multiple doses ranging from 100 to 800 mg [15]. The predicted ketoconazole concentrations corresponded to a typical individual. The relevant information needed for evaluating the potential ketoconazole QT effect is the slope of the QTc versus concentration, which could be estimated using population-predicted ketoconazole plasma concentrations using a well-established pharmacokinetic model if the dose and time of administration are known. Diagnostic plots for the analyses were generated and examined for assessment of model adequacy and possible lack of fit including, but not limited to: (1) observed versus population-predicted and individual-predicted values, (2) weighted residuals (WRES) versus population predictions, (3) weighted residuals versus plasma concentration, and (4) individual conditional weighted residuals (CWRESI) versus plasma concentration. In addition, the performance of the final model was evaluated by simulating data using final parameter estimates as well as interindividual and residual variability estimates from the final model (fixed and random effects) and conducting a visual predictive check (VPC). Effect of axitinib and ketoconazole on heart rate For evaluation of the effect of a single drug on heart rate, a linear relationship was applied:

RRij −baseline RRij = (placebo RRij −baseline RRij ) + (ΘSlope · CONCij ) + εij

(1)

13

622

Cancer Chemother Pharmacol (2015) 75:619–628

where j indexes the measurement time for the ith individual; RR represents the mean RR interval on drug (axitinib or ketoconazole); placebo RR and baseline RR are the mean placebo and baseline RR interval, respectively; ΘSlope represents the association between drug plasma concentration and RR interval; CONC represents drug concentration and the εij parameter represents the jth residual error for the ith individual. For evaluation of the effect of axitinib in the presence of ketoconazole on heart rate following concomitant administration, the following model was used:

  RRij −baseline RRij = placebo RRij −baseline RRij + (ΘSlope · AGCONCij ) + εij

(2)

where RR now represents the mean RR interval on axitinib in the presence of ketoconazole and AGCONC represents axitinib plasma concentration in the presence of ketoconazole. Only axitinib concentration was used to characterize the relationship. A sex effect was tested on ΘSlope in the model as follows, with the variable equal to 0 for male and 1 for female:

Slope = Θslope · (1 + Θsex )

(3)

Concentration–QTc response modeling The one-stage QT interval correction method was employed as follows:  β QTij / RRij /1000 −baseline QTcOSij

= (placebo QTcOSij − baseline QTcOSij )

(4)

+ (ΘSlope · CONCij ) + εij where j indexes the measurement time for the ith individual; QT represents the mean of triplicate QT interval; placebo QTcOS and baseline QTcOS are the mean placebo and baseline study population correction QT intervals, respectively; ΘSlope represents the association between drug plasma concentrations and QT interval; CONC represents either axitinib or ketoconazole plasma concentrations; RR is mean triplicate RR interval; and β represents the population correction factor. The εij parameter represents the jth residual error for the ith individual. When axitinib was administered in the presence of ketoconazole, only axitinib concentrations were used to characterize the relationship since plasma concentrations of ketoconazole were not measured in this study (only predicted). A sex effect also was tested on ΘSlope in the model using Eq. 3. The model development and simulations were performed using NONMEM version 7.1.2, and VPCs were conducted using Perl-speaks-NONMEM (PsN). Pre- and post-processing of data was carried out using S-PLUS 7.0 (TIBCO Software Inc., Palo Alto, CA, USA).

13

Fig. 1  Median changes from baseline in heart rate observed in subjects 1, 2, and 3 h after administration of placebo, axitinib alone, ketoconazole alone, or axitinib in the presence of ketoconazole. BPM beats per minute

Results Subject disposition and baseline characteristics Twenty (18 male and 2 female) healthy subjects were randomized to the sequence A → B and 15 (14 male and 1 female) to the sequence B → A. The treatment was terminated in six subjects in the A → B sequence group and 1 subject in the B → A sequence group, at investigator discretion (n  = 4), due to withdrawal of consent (n  = 2) or treatment-unrelated adverse event (n  = 1). Median age and the proportion of whites were similar between the two groups (33.5 vs. 32.0 years and 50 vs. 60 % in the A → B vs. B → A sequences, respectively), as well as median systolic/diastolic BP and heart rate at baseline (114.0/67.5 vs. 112.0/72.0 mm Hg and 59.0 vs. 59.0 beats per minute [bpm], in the A → B vs. B → A sequence, respectively) [9]. Since demographics and baseline clinical characteristics were comparable between the two sequence groups, data were pooled from both sequence groups and used for the QT/QTc interval analyses. Effects of axitinib and ketoconazole on heart rate After the administration of axitinib alone, heart rate was decreased modestly from baseline, with the largest median change of −4 bpm. However, after administration of ketoconazole alone, there was a notable increase in heart rate (largest median change of 11 bpm; Fig. 1). When axitinib was co-administered with ketoconazole, the largest median change from baseline was −4 bpm. The relationship between RR interval and drug concentration was further analyzed using a linear model based on Eq.  1. Mean steady-state plasma concentrations of ketoconazole, based on simulations, were predicted to be 6.50,

Cancer Chemother Pharmacol (2015) 75:619–628

623

Table 1  Parameter estimates for the RR interval versus drug plasma concentrations Parameter Slope Mean estimatea 95 % CIb Sex effect Mean estimate 95 % CIb

Axitinib alone

Ketoconazole alone

Axitinib +  ketoconazole

2.28 1.46, 3.10

−16.2 −19.6, −12.8

1.14 0.715, 1.57

−0.704

−1.01, −0.396

NS

NS





CI confidence interval, NS not significant a

  Units: ms·mL/ng for axitinib alone and in the presence of ketoconazole and ms·mL/µg for ketoconazole alone b 

  Calculated as estimate ±1.96 × standard error

7.53, and 7.13 µg/mL at 1, 2, and 3 h, respectively. Parameter estimates from the RR interval–drug plasma concentration models are summarized in Table 1. The results indicated a positive correlation for axitinib (i.e., an increase in the RR interval or decreased heart rate) with increasing axitinib plasma concentrations and a negative correlation for ketoconazole (i.e., a decrease in the RR interval or increased heart rate) with increasing ketoconazole plasma concentrations. In the presence of both axitinib and ketoconazole, the net effect on RR interval was a positive relationship. Three females in the dataset had evaluable QT interval and pharmacokinetic data and were included in the analysis. The mean effect of sex on the slope of RR interval versus plasma concentration for axitinib (but not for ketoconazole or axitinib in the presence of ketoconazole) was observed as significant and was retained in the final model (Table 1). Diagnostic plots for each RR interval–drug concentration linear model demonstrated that there was good agreement between the predicted and observed concentrations and a lack of bias in the residuals as well as the predicted concentrations over time (data not shown). CWRESI plots showed that no value was greater than 4, indicating no outliers. Concentration–QTc response analysis The concentration–QTc response analysis was initially conducted using fixed correction factors based on baseline data [i.e., using Fridericia’s, Bazett’s, and a study-specific correction factor, to correct QT for heart rate (QTcF, QTcB, and QTcS, respectively)]. However, QTc values obtained using these corrections retained a dependency on heart rate due to study drug. Therefore, the relationship was analyzed using a one-stage approach with QTcOS, in which the effect of drug on both RR and QT intervals was

simultaneously ascertained (QTcS or QTcOS vs. RR interval: Online Resource 2). The relationships between changes from baseline in placebo-corrected QTc (ΔΔQTcOS) and drug plasma concentrations are depicted for axitinib alone, ketoconazole alone, and axitinib in the presence of ketoconazole (Fig. 2). The population parameter estimates from the one-stage analysis final model are presented in Table 2. The estimates for the population correction factor (β value) were 0.472, 0.401, and 0.425 for axitinib alone, ketoconazole alone, and axitinib in the presence of ketoconazole, respectively. The population mean estimate for the slope for axitinib alone was −0.0314 ms·mL/ng, with the 95 % confidence interval (CI) −0.0675, 0.0047 encompassing 0, which implied an apparent lack of relationship between ΔΔQTcOS and axitinib plasma concentration. Diagnostic plots of eta on slope in males and females indicated no difference between the two groups and, consequently, sex effect was not included in the final model. The slope population mean estimate (95 % CI) for ketoconazole was −0.331 (−0.860, 0.198) ms·mL/µg. The slope population mean estimate (95 % CI) for axitinib in the presence of ketoconazole was 0.0725 (0.0445, 0.1005) ms·mL/ng. Using the estimated β value for QT correction from the one-stage approach, the highest mean placebo-corrected change from baseline in QT interval was −3.0 ms for axitinib alone and −2.8 ms for ketoconazole alone, with the upper bound of the 90 % CI being −0.6 and −0.5 ms, respectively (Table 3). In the presence of ketoconazole (i.e., at the supratherapeutic doses of axitinib), the predicted highest mean change was 6.5 ms, with the upper bound of the 90 % CI (8.5 ms) being below 10 ms, which is the threshold of concern for potential effect of drug on QT prolongation per ICH E14 guidance. The lower bound of 90 % CI was greater than 0 (4.4 ms). Diagnostic plots for drug concentration–QTc response models indicated that the predicted and observed values were in good agreement, and no bias in the residuals was detected (Online Resource 3). Both the WRES and CWRESI were evenly distributed across the range of observations and displayed no systematic deviation over time. CWRESI plots did not show any values above 4, indicating absence of substantial outliers, and demonstrated adequacy of the model. The predictability of the axitinib concentration– ΔΔQTcOS models was assessed by conducting simulations (n = 1,000 each for axitinib alone or in the presence of ketoconazole) using the final model population parameter estimates and interindividual and residual variability with the available dataset. The VPC showed that simulated concentrations generally agreed well with observed concentrations, with no systematic bias as the majority of the observations fell within the 95th percentile of the predicted

13

624

a

Cancer Chemother Pharmacol (2015) 75:619–628 Table 2  Parameter estimates for the ΔΔQTc interval versus drug plasma concentration, based on the one-stage QT correction method Parameter

Axitinib alone

Ketoconazole alone

Slope Mean estimatea −0.0314 −0.331 −0.0675, 0.0047 −0.860, 0.198 95 % CIb

Axitinib + ketoconazole

0.0725 0.0445, 0.1005

β (population correction factor) Mean estimate 0.472

0.401

0.425

95 % CIb

0.362, 0.440

0.385, 0.465

0.421, 0.523

CI confidence interval

b

a

  Units: ms·mL/ng for axitinib alone and in the presence of ketoconazole; and ms·mL/µg for ketoconazole alone b

c

  Calculated as estimate ±1.96 × standard error

The median axitinib Cmax observed in the absence (51.2 ng/mL) or presence (84.6 ng/mL) of ketoconazole in this study was higher than the median (range) Cmax of 30.9 (13.3–96.6) ng/mL in patients for whom axitinib doses were increased to the highest permitted dose of 10 mg twice daily (Online Resource 4). Therefore, the results of QT prolongation obtained in this study are applicable to the intended patient population. Furthermore, Cmax obtained in this study was higher than those obtained in subjects with mild and moderate hepatic impairment receiving the axitinib 5-mg dose (Online Resource 4) [16]; hence, this study provided adequate coverage over the commonly encountered situations in clinical studies that may result in higher than typical plasma concentrations of axitinib. One‑stage QT correction versus other QT correction methods

Fig. 2  ΔΔQTcOS versus drug plasma concentration from the model based on one-stage QT correction method: a axitinib 5 mg alone, b ketoconazole 400 mg alone, and c axitinib 5 mg in the presence of ketoconazole 400 mg. Circles indicate individual data points. Dotted lines represent lines of regression. CI confidence interval

values (data not shown), supporting validity of the final model. In order to predict QT intervals at the supratherapeutic doses of axitinib, additional simulations (n = 1,000) were performed at the highest mean Cmax of 79.5 ng/mL, which was observed after the administration of a 5-mg oral dose of axitinib in the presence of ketoconazole in this study. The predicted mean (90 % CI) ΔΔQTcOS was 5.7 (3.8–7.6) ms (Fig. 3a), compared with 6.5 ms for the observed mean for all subjects in the study.

13

An initial graphical representation of the baseline QT interval, corrected using the three widely used correction methods, versus RR interval displayed the most horizontal pattern for QTcS, indicating that the study-specific correction method performed better than Fridericia’s or Bazett’s when assessing the off-treatment data (data not shown). Owing to the effect of axitinib on heart rate, an assessment of the QT–RR correlation was necessary using on-treatment data. Online Resource 2, in which on-drug QTc interval versus RR is displayed, demonstrates that QTcS was unable to account for the heart rate effect when drug was on board; however, one-stage correction was able to successfully correct the overall effect on heart rate. The highest mean placebo-corrected changes from baseline in QTc estimated using these three correction methods (QTcF, QTcB, and QTcS, respectively) were 6.1, −0.7, and 2.2 ms for axitinib, −5.7, 4.2, and 0.7 ms for ketoconazole, and 9.1, 4.4, and 6.1 ms for axitinib in the presence

Cancer Chemother Pharmacol (2015) 75:619–628

625

Table 3  Highest mean placebo-corrected changes from baseline (ΔΔQTc) using various heart rate correction methods Correction method

Two-stage approach Fridericia’s (ΔΔQTcF) Bazett’s (ΔΔQTcB) Study-specific (ΔΔQTcS)

Correction factor/β value

Mean change from baselinea (90 % CIb), ms Axitinib alone

Ketoconazole alone

Axitinib + ketoconazole

0.33 0.50 0.44

6.1 (1.8, 10.5) −0.7 (−4.2, 2.8) 2.2 (−1.9, 5.4)

−5.7 (−8.7, −2.7) 4.2 (1.3, 7.1) 0.7 (−2.2, 3.5)

9.1 (6.7, 11.5) 4.4 (1.9, 6.9) 6.1 (3.7, 8.4)

0.472 0.401

−3.0 (−5.4, −0.6) –

– –

0.425



– −2.8 (−5.1, −0.5)

One-stage approach (ΔΔQTcOS)



6.5 (4.4, 8.5)

CI confidence interval a

  Highest value reported at 1-, 2- or 3-h post-dose

b

  Calculated as estimate ±1.645 × standard error

of ketoconazole, respectively (Table 3). The upper bound of the 90 % CI was below 10 ms, except for ΔΔQTcF for axitinib in the absence or presence of ketoconazole (10.5 and 11.5, respectively). Comparisons between fixed correction factors and the one-stage approach were further made by performing simulations (n = 1,000) of the ΔΔQTc at the highest mean Cmax of 79.5 ng/mL using the final model parameters for QTcB, QTcF, QTcS (Fig. 3b–d). The mean (90 % CI) ΔΔQTcB was 3.1 (1.0–5.1) ms; ΔΔQTcF 8.2 (6.4–10.3) ms; and ΔΔQTcS 5.0 (3.4–6.7) ms, compared with 5.7 (3.8– 7.6) ms for ΔΔQTcOS (Fig. 3a).

Discussion This study investigated the effect of axitinib on cardiac repolarization by analyzing the relationship between drug plasma concentration and changes in QT interval using data from a phase I definitive QT study of axitinib in healthy volunteers. The modeling and simulation analysis showed three important findings: (1) axitinib and ketoconazole, respectively, had a statistically significant effect on decreasing and increasing heart rate; (2) standard fixed correction factors did not remove the correlation between QT and RR; and (3) axitinib at supratherapeutic concentrations did not have a clinically meaningful effect on QT interval. The ICH E14 guidance recommends that a TQT study be conducted during clinical development of a drug to determine whether it has an effect on cardiac repolarization [2]. The recommended design is a randomized, doubleblind, placebo- and positive-controlled study, typically conducted in healthy volunteers [2], with the maximum mean time-matched difference in baseline-adjusted QTc between the experimental and placebo arms as the primary endpoint,

using a statistical analysis known as the intersection union test. However, the role of concentration–QT response analysis has been addressed in a number of publications to support planning and interpretation of TQT studies [17– 21]. As per the ICH E14 guidance, such a relationship is established using sufficiently high plasma concentrations, including those higher than observed following the anticipated therapeutic doses, which can be achieved through the use of a high dose of study drug or metabolic inhibition as part of a drug–drug interaction study [22, 23]. Challenges with conducting a TQT study for NAAs include ethical concerns for administering these agents, which are often genotoxic, to healthy volunteers and difficulty obtaining supratherapeutic plasma concentrations since they are often dosed at maximum tolerated doses. Hence, alternative designs have been the subject of many publications [14, 24, 25]. In this study, supratherapeutic concentrations of axitinib were achieved by metabolic inhibition using ketoconazole. As highlighted in the FDA publication, utilization of a drug–drug interaction study as a dedicated QT study includes the underlying assumption that the inhibitor has no effect on heart rate or QT interval [23]. In the current study, we clearly demonstrated that ketoconazole significantly increased heart rate (median 11 bpm), which is consistent with some prior reports [23]. Additionally, axitinib alone was shown to decrease heart rate (median −4 bpm), making the analysis and interpretation of data difficult when axitinib was administered in the presence of ketoconazole. Many methods have been reported to correct for heart rate when a drug inherently increases heart rate, including Holter-bin analysis, dynamic beat-to-beat analysis, and model-based methods that use a one-stage approach to analyze QT, RR, and drug concentration data simultaneously [11]. Some have suggested that when the effect of a drug on heart rate exceeds

13

626

a

b

c

d

around 5 bpm, standard fixed correction methods may be inadequate [11]. The impact of using fixed corrections for drugs that cause changes in heart rate was shown in a TQT study of tolterodine [26], which produces a mean heart rate increase of 6.3 bpm. With Bazett’s correction method, the upper confidence limit for mean QTc change was 15.8 and

13

Cancer Chemother Pharmacol (2015) 75:619–628

◂Fig. 3  Simulated distributions and observed mean maximum change

in QT interval, using various correction methods for heart rate: a one-stage (QTcOS), b Bazett’s (QTcB), c Fridericia’s (QTcF), and d study-specific (QTcS), following administration of axitinib at the highest mean maximum observed plasma concentration of 79.5 ng/ mL (i.e., at a 5-mg axitinib dose in the presence of ketoconazole) using the final model parameters for QTcOS, QTcB, QTcF, QTcS, respectively. The vertical dark solid lines and vertical dot-dashed lines represent the mean and the 5th and 95th percentile, respectively, of the simulated largest change in QTc interval. The vertical dotted line represents the mean of the observed value

9.8 ms with Fridericia’s correction method. The choice of correction factor here produces very different conclusions. This was also shown with sibenadet that increases heart rate ~10 bpm [27]. When fixed correction factors (QTcF, QTcB, QTcS, and QTcI) were applied, the maximum upper bound of the 95 % CI exceeded 10 ms in all cases. On the other hand, when applying the one-stage modeling approach, there was no effect of sibenadet on the QT interval. Ketoconazole has been shown to increase the heart rate by 6 bpm [28]. In the paper by Chaikin et al. [29], individualized corrections were employed using baseline (off-drug) data. This analysis, which did not account for the increased heart rate caused by ketoconazole, noted a QTc increase of 6.96 ms, which is in contrast to other studies that have shown a decrease in QTc interval; one study by Kosoglou et al. [28] demonstrated a −1.65 ms change in QTcB and another study by Robert et al. [30] showed a range of −2.5 to −1.1 ms. The results presented here demonstrate that when applying the one-stage modeling approach to the ketoconazole alone data, the highest mean placebo-corrected change from baseline was −2.76 ms. The results from the one-stage approach here indicated the concentration–QTc response relationship to be flat for axitinib, with an insignificant slope estimate where the 90 % CI included 0. For axitinib in the presence of ketoconazole, the value for the QTc prolongation per increase by one concentration unit (ng/mL) was 0.0725 ms. Simulations with the final model parameters provided important insights into the clinical implications of this study. At the supratherapeutic plasma concentrations of axitinib obtained with potent CYP3A4/5 inhibition, which reflects the worstcase scenario for the potential increase in plasma exposure in patients with mRCC or those with hepatic impairment, the highest mean (90 % CI) ΔΔQTcOS was estimated as 6.5 (4.4–8.5) ms. The ICH E14 guidance states that a TQT study indicates lack of a substantial QT prolongation if the upper bound of the 95 % CI around the mean excludes 10 ms at all time points; hence, the current results suggested a lack of a clinically significant association between the QTc interval and axitinib plasma concentrations in target patient populations. In addition, ΔΔQTcF, ΔΔQTcB, and ΔΔQTcS were estimated, all of which had the upper bound of the 90 % CI below 10 ms, except ΔΔQTcF for

Cancer Chemother Pharmacol (2015) 75:619–628

axitinib in the absence or presence of ketoconazole (10.5 and 11.5 ms, respectively). The latter is likely due to the fact that the standard two-stage approach used for the estimation of ΔΔQTcF assumes lack of a drug effect on heart rate. However, even these values would not translate to a clinically significant shift in the risk-benefit ratio for axitinib as a proven efficacious treatment for mRCC. For NAAs, the FDA has used 20 ms as the upper bound because of the grave prognosis in advanced cancer and the need for more efficacious agents to fulfill unmet needs [31]. This study, however, has limitations. First, there were no ECG data collected past 3-h post-dosing (e.g., time points such as 24-h post-dosing to explore a possible contribution by ion channel trafficking to the late onset of the druginduced QTc prolongation). It is noteworthy that the ICH E14 guidance recommends that the timing of ECG collection should be based on the known pharmacokinetics of the drug. The pharmacokinetic samples were collected up to 96 h post-dosing, which allowed adequate characterization of axitinib pharmacokinetics. The maximum plasma concentration of axitinib in this study was reached 1.50 h (range 1.00–3.00) and 2.00 h (1.00–4.13) after dosing in the absence or presence of ketoconazole, respectively. Hence, ECG readings collected in this study are considered to have been adequately conducted at the time around the Cmax. In addition, axitinib has a relatively short effective plasma half-life (2.5–6.1 h), with minimum accumulation after multiple doses, and axitinib metabolites in plasma are considered pharmacologically inactive [8]. Therefore, additional collection of ECG readings at later time periods was not critical in this case. A preliminary assessment of time dependence on QT interval by graphical displays of the highest mean ΔΔQTcOS and axitinib plasma concentration over time (1-, 2-, and 3-h post-dosing) suggested that the highest change occurred at 2-h post-dosing; hence, potential delayed effect of axitinib on QTc interval is not anticipated. The second limitation was the absence of a positive control, such as moxifloxacin, which is frequently used to establish assay sensitivity in TQT study. Inclusion of a positive control to show that the assay was sufficiently sensitive to detect threshold of QTc changes of regulatory concern was important in light of the lack of clinically meaningful QT effect observed for axitinib. However, it should be emphasized that the analysis was based on robust data obtained in the randomized, single-blind, twoway crossover study conducted in a single study site, with strictly controlled experimental conditions and utilization of standardized and validated methods for ECG and pharmacokinetic data collection and processing, and was performed using the established concentration–QT response modeling analysis with appropriate one-stage QT interval correction method. Taken together with preclinical studies

627

that did not suggest any potential proarrhythmic risk of axitinib, the current assessment of axitinib-induced QTc effect based on the early phase study would likely provide valid quantification. In conclusion, simulations with a model-based concentration–QTc response analysis provided important insights. Based on the concentration–QTc response analysis conducted with data obtained in a drug–drug interaction study in healthy subjects indicative of the worst-case increase in plasma exposure in the target population, axitinib is not expected to have clinically significant QTc prolongation in patients with cancer administered recommended doses of axitinib. Acknowledgments  This study was sponsored by Pfizer Inc. Medical writing support was funded by Pfizer Inc and provided by Mariko Nagashima, PhD, of Engage Scientific Solutions (Southport, CT, USA). Conflict of interest  Ana Ruiz-Garcia, Brett E. Houk, and Yazdi K. Pithavala are employees of and own stock in Pfizer Inc. Michael A. Tortorici, who was employed by and owned stock in Pfizer Inc during the time of this study and development of the manuscript, is currently an employee of CSL Behring Biotherapies for Life™. Melvin Toh, who was employed by and owned stock/options in Pfizer Inc at the time of this study, is currently an employee of CK Life Sciences Int’l. (Holdings) Inc and no longer owns stock/options in Pfizer Inc. Nenad Sarapa, who was employed by and owned stock/options in Pfizer Inc at the time of this study, is currently employed by Bayer Healthcare Pharmaceuticals.

References 1. Roden DM (2008) Cellular basis of drug-induced torsades de pointes. Br J Pharmacol 154:1502–1507. doi:10.1038/ bjp.2008.238 2. U.S. Department of Health and Human Services Food and Drug Administration (2005) Guidance for industry, E14 clinical evaluation of QT/QTc interval prolongation and proarrhythmic potential for non-antiarrhythmic drugs. http://www.fda.gov/downloads/ RegulatoryInformation/Guidances/ucm129357.pdf. Accessed 4 Feb 2013 3. Hu-Lowe DD, Zou HY, Grazzini ML, Hallin ME, Wickman GR, Amundson K, Chen JH, Rewolinski DA, Yamazaki S, Wu EY, McTigue MA, Murray BW, Kania RS, O’Connor P, Shalinsky DR, Bender SL (2008) Nonclinical antiangiogenesis and antitumor activities of axitinib (AG-013736), an oral, potent, and selective inhibitor of vascular endothelial growth factor receptor tyrosine kinases 1, 2, 3. Clin Cancer Res 14:7272–7283. doi:10.1158/1078-0432.CCR-08-0652 4. Inlyta® (axitinib) prescribing information (2012). http://labeling. pfizer.com/ShowLabeling.aspx?id=759. Accessed 22 Jan 2013 5. Rugo HS, Herbst RS, Liu G, Park JW, Kies MS, Steinfeldt HM, Pithavala YK, Reich SD, Freddo JL, Wilding G (2005) Phase I trial of the oral antiangiogenesis agent AG-013736 in patients with advanced solid tumors: pharmacokinetic and clinical results. J Clin Oncol 23:5474–5483. doi:10.1200/JCO.2005.04.192 6. Rini BI, Escudier B, Tomczak P, Kaprin A, Szczylik C, Hutson TE, Michaelson MD, Gorbunova VA, Gore ME, Rusakov IG, Negrier S, Ou Y-C, Castellano D, Lim HY, Uemura H, Tarazi J,

13

628 Cella D, Chen C, Rosbrook B, Kim S, Motzer RJ (2011) Comparative effectiveness of axitinib versus sorafenib in advanced renal cell carcinoma (AXIS): a randomised phase 3 trial. Lancet 378:1931–1939. doi:10.1016/S0140-6736(11)61613-9 7. Di Lorenzo G, Porta C, Bellmunt J, Sternberg C, Kirkali Z, Staehler M, Joniau S, Montorsi F, Buonerba C (2011) Toxicities of targeted therapy and their management in kidney cancer. Eur Urol 59:526–540. doi:10.1016/j.eururo.2011.01.002 8. Chen Y, Tortorici MA, Garrett M, Hee B, Klamerus KJ, Pithavala YK (2013) Clinical pharmacology of axitinib. Clin Pharmacokinet 52:713–725. doi:10.1007/s40262-013-0068-3 9. Pithavala YK, Tong W, Mount J, Rahavendran SV, Garrett M, Hee B, Selaru P, Sarapa N, Klamerus KJ (2012) Effect of ketoconazole on the pharmacokinetics of axitinib in healthy volunteers. Invest New Drugs 30:273–281. doi:10.1007/s10637-010-9511-6 10. Tornoe CW, Garnett CE, Wang Y, Florian J, Li M, Gob buru JV (2011) Creation of a knowledge management system for QT analyses. J Clin Pharmacol 51:1035–1042. doi:10.1177/0091270010378408 11. Garnett CE, Zhu H, Malik M, Fossa AA, Zhang J, Badilini F, Li J, Darpo B, Sager P, Rodriguez I (2012) Methodologies to characterize the QT/corrected QT interval in the presence of druginduced heart rate changes or other autonomic effects. Am Heart J 163:912–930. doi:10.1016/j.ahj.2012.02.023 12. Sweeney KR, Gastonguay MR, Benincosa L, Cronenberger CL, Glue P, Malhotra BK (2010) Exposure–response modeling and clinical trial simulation of the effect of tolterodine on QT intervals in healthy volunteers. Drug Discov Ther 4:44–53 13. Desai M, Li L, Desta Z, Malik M, Flockhart D (2003) Variability of heart rate correction methods for the QT interval. Br J Clin Pharmacol 55:511–517. doi:10.1046/j.1365-2125.2003.01791.x 14. Bello CL, Mulay M, Huang X, Patyna S, Dinolfo M, Lev ine S, Van Vugt A, Toh M, Baum C, Rosen L (2009) Electrocardiographic characterization of the QTc interval in patients with advanced solid tumors: pharmacokinetic–pharmacodynamic evaluation of sunitinib. Clin Cancer Res 15:7045–7052. doi:10.1158/1078-0432.CCR-09-1521 15. Chien JY, Lucksiri A, Ernest CS 2nd, Gorski JC, Wrighton SA, Hall SD (2006) Stochastic prediction of CYP3A-mediated inhibition of midazolam clearance by ketoconazole. Drug Metab Dispos 34:1208–1219. doi:10.1124/dmd.105.008730 16. Tortorici MA, Toh M, Rahavendran SV, Labadie RR, Alvey CW, Marbury T, Fuentes E, Green M, Ni G, Hee B, Pithavala YK (2011) Influence of mild and moderate hepatic impairment on axitinib pharmacokinetics. Invest New Drugs 29:1370–1380. doi:10.1007/s10637-010-9477-4 17. Garnett CE, Beasley N, Bhattaram VA, Jadhav PR, Madabushi R, Stockbridge N, Tornoe CW, Wang Y, Zhu H, Gobburu JV (2008) Concentration–QT relationships play a key role in the evaluation of proarrhythmic risk during regulatory review. J Clin Pharmacol 48:13–18. doi:10.1177/0091270007307881 18. Russell T, Riley SP, Cook JA, Lalonde RL (2008) A perspective on the use of concentration–QT modeling in drug development. J Clin Pharmacol 48:9–12. doi:10.1177/0091270007311115 19. Chapel S, Hutmacher MM, Bockbrader H, de Greef R, Lalonde RL (2011) Comparison of QTc data analysis methods recommended by the ICH E14 guidance and exposure–response analysis: case study of a thorough QT study of asenapine. Clin Pharmacol Ther 89:75–80. doi:10.1038/clpt.2010.220 20. Darpo B, Garnett C, Benson CT, Keirns J, Leishman D, Malik M, Mehrotra N, Prasad K, Riley S, Rodriguez I, Sager P, Sarapa N, Wallis R (2014) Cardiac Safety Research Consortium: Can the

13

Cancer Chemother Pharmacol (2015) 75:619–628 thorough QT/QTc study be replaced by early QT assessment in routine clinical pharmacology studies? Scientific update and a research proposal for a path forward. Am Heart J 168:262–272. doi:10.1016/j.ahj.2014.06.003 21. Darpo B, Sarapa N, Garnett C, Benson C, Dota C, Ferber G, Jarugula V, Johannesen L, Keirns J, Krudys K, Ortemann-Renon C, Riley S, Rogers-Subramaniam D, Stockbridge N (2014) The IQCSRC prospective clinical phase 1 study: “Can early QT assessment using exposure response analysis replace the thorough QT study?”. Ann Noninvasive Electrocardiol 19:70–81. doi:10.1111/ anec.12128 22. Darpo B, Nebout T, Sager PT (2006) Clinical evaluation of QT/ QTc prolongation and proarrhythmic potential for nonantiarrhythmic drugs: the international conference on harmonization of technical requirements for registration of pharmaceuticals for human use E14 guideline. J Clin Pharmacol 46:498–507. doi:10.1177/0091270006286436 23. Zhu H, Wang Y, Gobburu JV, Garnett CE (2010) Considerations for clinical trial design and data analyses of thorough QT studies using drug–drug interaction. J Clin Pharmacol 50:1106–1111. doi:10.1177/0091270009358710 24. Rock EP, Finkle J, Fingert HJ, Booth BP, Garnett CE, Grant S, Justice RL, Kovacs RJ, Kowey PR, Rodriguez I, Sanhai WR, Strnadova C, Targum SL, Tsong Y, Uhl K, Stockbridge N (2009) Assessing proarrhythmic potential of drugs when optimal studies are infeasible. Am Heart J 157:827–836. doi:10.1016/j. ahj.2009.02.020 25. Curigliano G, Spitaleri G, Fingert HJ, de Braud F, Sessa C, Loh E, Cipolla C, De Pas T, Goldhirsch A, Shah R (2008) Druginduced QTc interval prolongation: a proposal towards an efficient and safe anticancer drug development. Eur J Cancer 44:494–500. doi:10.1016/j.ejca.2007.10.001 26. Malhotra BK, Glue P, Sweeney K, Anziano R, Mancuso J, Wicker P (2007) Thorough QT study with recommended and supratherapeutic doses of tolterodine. Clin Pharmacol Ther 81:377–385. doi:10.1038/sj.clpt.6100089 27. Li J, Birmingham B, Mosqueda-Garcia R, Sander N, Newbold P, Sager P (2008) Evaluation of drug-induced QT/QTc prolongation in the presence of drug inducted changes in heart rate using population PK/PD modeling approach: sibenadet experience. In: American conference on pharmacometrics (ACoP) annual meeting, Tucson, AZ., Mar 9–12, 2008 28. Kosoglou T, Salfi M, Lim JM, Batra VK, Cayen MN, Affrime MB (2000) Evaluation of the pharmacokinetics and electrocardiographic pharmacodynamics of loratadine with concomitant administration of ketoconazole or cimetidine. Br J Clin Pharmacol 50:581–589. doi:10.1046/j.1365-2125.2000.00290.x 29. Chaikin P, Gillen MS, Malik M, Pentikis H, Rhodes GR, Roberts DJ (2005) Co-administration of ketoconazole with H1-antagonists ebastine and loratadine in healthy subjects: pharmacokinetic and pharmacodynamic effects. Br J Clin Pharmacol 59:346–354. doi:10.1111/j.1365-2125.2005.02348.x 30. Robert M, Salva M, Segarra R, Pavesi M, Esbri R, Roberts D, Golor G (2007) The prokinetic cinitapride has no clinically relevant pharmacokinetic interaction and effect on QT during coadministration with ketoconazole. Drug Metab Dispos 35:1149– 1156. doi:10.1124/dmd.106.010835 31. Sarapa N, Britto MR (2008) Challenges of characterizing proarrhythmic risk due to QTc prolongation induced by nonadjuvant anticancer agents. Expert Opin Drug Saf 7:305–318. doi:10.1517/14740338.7.3.305