A Single-Dose, Crossover, Placebo- and Moxifloxacin-Controlled Study to Assess the Effects of Neratinib (HKI-272) on Cardiac Repolarization in Healthy Adult Subjects

Regulatory agencies have increasingly emphasized the importance of characterizing the effect of new chemical entities on the QT interval. The QT interval assessment provides a measure of the effect of a drug on cardiac repolarization and is a surrogate marker for risk of ventricular tachycardia, most notably for torsade de pointes (1, 2). The International Conference for Harmonization (ICH) E14 guidelines (3) for detecting drug-induced QT prolongation outline the ideal thorough QT (TQT) study as one that includes the evaluation of both therapeutic and supratherapeutic doses of an investigational drug to account for the extremes of plasma drug concentrations that may be encountered in the target patient population. In addition, the TQT study design incorporates recommendations for placebo and positive control assessments to provide scientific rigor. TQT studies are most often done in healthy subjects given the logistical complexities and nontherapeutic nature of the assessment. Unfortunately, some investigational drugs cannot be evaluated using the E14-recommended TQT study design because safety and tolerability limit the doses that can be evaluated (4).

The ability to execute a healthy subject TQT study according to ICH E14 guidelines is particularly challenging for oncology compounds because many agents can only be evaluated in cancer patients due to potential for irreversible adverse sequelae. A cancer patient population, however, does not readily accommodate many aspects of the TQT design, including cumbersome logistics, rigorous standards for ECG collection and interval measurements, and administration of placebo and positive controls, such as moxifloxacin. Even those anticancer agents that can be administered as single doses to healthy subjects, such as some signal transduction inhibitors, face challenges in TQT study execution. In particular, achieving multiples of the steady-state plasma concentrations observed in cancer patients may not be realistic in healthy subjects. Even in cancer patients, supratherapeutic plasma concentrations are difficult to achieve without causing significant dropout due to toxicities (5, 6).

Neratinib (HKI-272) is an orally administered, small-molecule, irreversible pan ErBb tyrosine kinase receptor inhibitor that blocks signal transduction through inhibition of ErbB-1, ErbB-2, and ErbB-4 (7–9), and is currently in phase 3 clinical development for the treatment of patients with ErbB-2 positive breast cancer. Neratinib is generally acceptably tolerated by cancer patients, although gastrointestinal adverse events are common and limit the investigational dose to 240 mg daily in ongoing phase 3 trials (10, 11).

Investigation of neratinib in single-dose clinical trials in healthy subjects provided information suggesting that supratherapeutic plasma concentrations supporting a TQT study could be achievable. In a single ascending dose study to evaluate neratinib from 120 to 800 mg, neratinib exposure [peak plasma concentration (C_max) and area under the concentration-time curve (AUC)] reached a plateau at doses >400 mg when administered to fasted subjects (12). Nevertheless, gastrointestinal tolerability continued to decrease with doses >400 mg. The separation of tolerability from systemic plasma concentrations suggested that the gastrointestinal adverse events that limit dose escalation are determined,a at least in part, by local effects as opposed to systemic effects.

The inferred mechanism of gastrointestinal toxicities suggested that supratherapeutic plasma concentrations of neratinib might be achievable if local gastrointestinal effects could be bypassed. We exploited the knowledge that neratinib is a CYP3A4 substrate and showed, in a drug interaction study, that supratherapeutic neratinib plasma concentrations were well tolerated if achieved by coadministration with ketoconazole, an inhibitor of neratinib metabolism (13). Supratherapeutic neratinib concentrations achieved through CYP3A4 inhibition would not provide information on effects of neratinib metabolites nor would it provide information on steady-state effects of the parent drug. Nevertheless, the available clinical data provided an opportunity to conduct a TQT study of neratinib in healthy subjects closely approximating the ICH E14 guidelines.

Healthy male (ages 18–50 years inclusive at screening) and female (ages 18–60 years) subjects of non-childbearing potential who met all the qualifying criteria were eligible for enrollment in this study. Body mass index and body weight were required to be in the range of 18 to 30 kg/m² and ≥50 kg, respectively. Subjects were ineligible to participate in this study if they had a history of long QT syndrome, syncope, or seizure or a family history of unexplained, sudden, cardiac-related death,. Likewise, subjects were excluded if their Ca²⁺, Mg²⁺, and K⁺ levels were below the lower limit of normal or if their QTc duration was ≥450 milliseconds based on the machine-read tracing at screening or on study day −1. This study was conducted in accordance with the ICH guideline for Good Clinical Practice and the ethical principles that have their origins in the Declaration of Helsinki. Written informed consent was obtained from all subjects before enrollment in the study. The study protocol, amendments, and informed consent form were approved by an independent ethics committee or an institutional review board. Finally, the protocol underwent Special Protocol Assessment by the Food and Drug Administration before enrollment.

This was a single-center, two-part, five-period, randomized, single-dose, double-blind (with respect to neratinib), crossover, placebo- and open-label moxifloxacin-controlled study in healthy subjects. In part 1 (three periods), subjects were randomly administered single doses of 240 mg neratinib, 400 mg moxifloxacin, and placebo, one in each period, with a high-fat meal. In part 2 (two periods), subjects were administered singles dose of 240 mg neratinib and placebo, one in each period, in combination with 400 mg ketoconazole in a fasting state. Thus, parts 1 and 2 provide assessments of therapeutic and supratherapeutic neratinib plasma concentrations, respectively. Ketoconazole (400 mg) was administered 12 hours before neratinib on day −1, and was coadministered with neratinib on day 1, as previously reported (13). Treatments were administered in a crossover pattern. Subjects were randomly assigned to 1 of 12 dosage administration sequences, which consisted of a combination of each of the five treatment arms. Each period was separated by a 5-day washout. In addition, parts 1 and 2 were separated by a 9-day outpatient washout to ensure that each neratinib dose was separated by a minimum 14-day period.

Venous blood samples (5 mL each) for analysis were collected on study day 1 for each of the five periods at 2 hours predose and at 1.5, 3, 4, 5, 6, 8, 12, 24, and 48 hours following dose administration. The concentrations of neratinib and its metabolites, ketoconazole and moxifloxacin, in the plasma samples were measured by using a validated liquid chromatography tandem mass spectrometry assay. The lower limit of quantitation for neratinib and its metabolites, ketoconazole and moxifloxacin, was 3, 20, and 25 ng/mL, respectively. The plasma neratinib, neratinib metabolites, ketoconazole and moxifloxacin concentration data for each subject were analyzed by using a noncompartmental method (14) with WinNonlin Enterprise application version 4.1 (Pharsight Corporation).

Triplicate 12-lead electrocardiogram (ECG) recordings for each subject were obtained on study day 1 for each of the five periods at −1, −0.5, and 0 hours (immediately before dose administration), and at 1.5, 3, 4, 5, 6, 8, 12, 24, and 48 hours following dose administration in all periods. ECG results, including rhythm, heart rate (HR), PR, QRS, QT, standard Bazett correction (QTcB), and standard Fridericia correction (QTcF) intervals were interpreted and measured by eRT. The QT intervals were measured manually, and the superimposed median beat and threshold methods were used. A population correction (QTcN) and individual correction (QTcI) were computed using the estimated slope from a linear regression model of log QT versus log RR on all predose measures in part 1 for the study population and each individual, respectively.

Statistical analyses were conducted on the baseline-adjusted data for QTc, QT interval, and HR. The average of each triplicate ECG tracings at each postdose time point for each subject was used in the statistical analyses. The baseline-adjusted values were calculated by subtracting the average of the −1, −0.5, and 0 hours averaged triplicate data on day 1 from the averaged triplicate data collected after dose administration.

The primary objective of the analysis was the comparison of baseline-adjusted QTc for neratinib versus placebo (part 1, therapeutic plasma concentrations) and for neratinib plus ketoconazole versus placebo plus ketoconazole (part 2, supratherapeutic plasma concentrations). The primary method for correcting the QTc interval for HR was QTcN. A mixed analysis of covariance model was used to assess the treatment differences in change from baseline QTc. This model had fixed effects for sequence, treatment, period, and time as well as a treatment by time interaction term. Baseline was included as a covariate and subject as a random effect. A two-sided 90% confidence interval (90% CI) was computed for the baseline-adjusted difference in QTc at each postdose time point between neratinib versus placebo, and between neratinib plus ketoconazole versus placebo plus ketoconazole.

In our study, the difference in model-based baseline-adjusted QTcN between moxifloxacin and placebo (ΔΔQTcN) was statistically compared using two-sided 90% CIs. If moxifloxacin showed a >5-millisecond increase over placebo (lower bound of the 90% CI of the QTcN difference was >5) for at least one time point, the study would be deemed acceptable to evaluate the neratinib effect on QTc.

The relationship between the placebo- and baseline-adjusted QTcN (ΔΔQTcN) and the log-transformed concentrations of neratinib and moxifloxacin were evaluated separately by using linear regression models. The plasma concentrations were log transformed to minimize the influence of high concentrations on the analysis. The pharmacokinetic/pharmacodynamic (PK/PD) relationship of ketoconazole was not explored because determining its QT effect was not an objective of the study. The following model was used:

where β₀ is the intercept, β₁ is the coefficient for the log concentration, and β₂ is the coefficient for the baseline QTcN.

For this study, data for ≥44 subjects were required to achieve 83% power for obtaining 90% CIs for the difference in the baseline-adjusted QTc, between neratinib and placebo, and between neratinib plus ketoconazole and placebo plus ketoconazole, fully below 10 milliseconds at each time point. The assumptions made in the sample size calculations were as follows: (a) the intrasubject SD was 8 milliseconds; (b) the true maximum QTc difference in change from baseline versus placebo was ≤3 milliseconds and was done using the option for two one-sided equivalence tests for a crossover design in nQuery Advisor (Statistical Solutions).

Sixty healthy subjects were enrolled in this study [47 (78.0%) men and 13 (22.0%) women]. The median age of enrolled subjects was 36 years (range, 18–57 years). Forty-two (70.0%) subjects were white. In part 1 of the study, 56 subjects received therapeutic doses of 240 mg neratinib, 58 subjects received placebo, and 59 subjects received 400 mg moxifloxacin. In part 2 of the study, 54 subjects received the supratherapeutic dose of 240 mg neratinib plus ketoconazole, and 53 subjects received placebo plus ketoconazole.

Data for all 60 enrolled subjects were included in the safety analysis. The most commonly reported adverse events (AE) were gastrointestinal disorders (57% subjects), including diarrhea (42%), nausea (28%), and vomiting (13%), consistent with the previously reported tolerability profile of neratinib in healthy subjects (10, 11). The frequency of gastrointestinal AEs in subjects who were administered 240 mg neratinib and 240 mg neratinib plus ketoconazole was 34% and 48%, respectively. No cardiovascular AEs related to administration of neratinib were reported. No torsade de pointes, sudden death, ventricular tachycardia, ventricular fibrillation, ventricular flutter, syncope, or seizure was observed. No QTc intervals were >450 milliseconds or increased from baseline by 30 milliseconds following administration of neratinib or neratinib plus ketoconazole. Other interval data were within limits typical for studies in healthy subjects. All AEs were considered by the principal investigator to be mild in severity, with the exception of moderate ventricular extrasystoles (one subject), which occurred after administration of moxifloxacin. No serious AEs or deaths were reported during this study.

Of the 60 subjects who enrolled, 52 (87%) completed the study and 8 (13%) discontinued participation before completing all five periods of the study. Only one subject discontinued due to an AE (ventricular extrasystoles following the administration of moxifloxacin).

Blood samples from 59 subjects were available for PK analysis. Following administration of single-dose (240 mg) neratinib with food, the mean (%CV) for C_max and AUC was 68 ng/mL (40%) and 1,236 ng·h/mL (39%), respectively, consistent with published data for both healthy subjects and cancer patients (10, 12). Following coadministration of 240 mg neratinib with ketoconazole, the mean (%CV) for C_max was 163 ng/mL (46%) and AUC was 3,801 ng·h/mL (49%), a 2.4-fold and 3-fold increase, respectively, compared with single-dose neratinib alone. Neratinib data are consistent with previous observations (13).

In this study, two major circulating metabolites of neratinib, neratinib pyridine N-oxide (M3) and neratinib dimethylamine N-oxide (M7), were analyzed (data not shown). Following administration of 240 mg neratinib alone, the C_max and AUC_t of M3 were 17% and 7% of the neratinib, respectively, and the C_max and AUC_t of M7 were 16% and 9% of the neratinib, respectively. Following neratinib coadministration with ketoconazole, M3 exposure (AUC_t) was decreased by ∼10-fold, from 7% to 0.7%, due to ketoconazole inhibition of the CYP3A4 metabolic pathway, which is responsible for the generation of M3. However, there was no significant change in M7 exposure (AUC_t; 13%) following coadministration of neratinib with ketoconazole because M7 is mainly produced by flavin-containing monooxygenases.

Plasma neratinib concentrations peaked at 5 and 6 hours for the therapeutic and supratherapeutic dose conditions, respectively (Fig. 1), and the elevated plasma neratinib concentrations represented meaningful extremes compared with those observed in cancer patients [2.2-fold increase over the mean observed in cancer patients who received 240 mg neratinib daily at steady-state with food (mean C_max,74 ng/mL)]. Furthermore, 13 subjects achieved neratinib mean C_max >219 ng/mL, a 3-fold increase compared with the mean C_max observed in cancer patients (10). The highest C_max achieved in an individual subject in the supratherapeutic dose treatment was 327 ng/mL. By comparison, the highest neratinib plasma concentration observed to date in any individual cancer patient is 247 ng/mL.¹

Because ketoconazole can independently increase the QT interval (15), the pharmacokinetics of ketoconazole were evaluated when neratinib was coadministered with ketoconazole and when placebo was coadministered with ketoconazole. Following oral administration of multiple doses of ketoconazole in combination with a single dose of 240 mg neratinib or placebo, the ketoconazole C_max was 9,140 ng/mL (CV = 35%) and 9,446 ng/mL (CV = 42%), respectively, and the median t_max was 3 hours for both treatment groups. For the ketoconazole plus neratinib and ketoconazole plus placebo groups, ketoconazole AUC was 77,292 ng·h/mL (CV = 41%) and 72,967 ng·h/mL (CV = 48%), respectively. The least-square geometric mean ratios (and the 90% CI around the ratios) of ketoconazole plus neratinib versus ketoconazole plus placebo and for C_max and AUC were 101.33% (90% CI, 91.75–111.92) and 112.20% (90% CI, 100.77–124.93), respectively. Overall, the data suggest that ketoconazole exposures following multiple oral doses of ketoconazole concomitantly administered with a single oral dose of neratinib (period 4) were equivalent to exposures following multiple oral doses of ketoconazole with a placebo (period 5). Therefore, the relative effect on the QT interval of supratherapeutic neratinib concentrations compared with placebo can be evaluated by comparisons between periods 4 and 5.

After a single oral dose of 400 mg moxifloxacin, the moxifloxacin C_max was reached with a median t_max of 3 hours (range 1.5–6 hours). The mean moxifloxacin Cmax was 2,012 ng/mL (CV = 41%). The moxifloxacin C_max observed in this study was consistent with the reported C_max of this drug (16).

The QT interval, highly dependent on HR, was corrected to produce measures not related to HR. Four different QT corrections, QTcB, QTcF, QTcI, and QTcN, were applied. The correction factor for QTcN was 0.204 with a 95% CI of 0.186 to 0.221. The average correction factor for QTcI was 0.201 with a 95% CI of 0.171 to 0.231. Correlation coefficients for QTcB, QTcF, QTcN, and QTcI were calculated to assess the relationship of each correction to HR. The Pearson correlations for QTcB, QTcF, QTcN, and QTcI were 0.528, 0.170, −0.173, and −0.173, respectively. QTcN, the prespecified primary end point of the study, was not related to HR. It should be noted that all HR readings were between 45 and 120 bpm throughout the study, and there were no changes from baseline HR >15 bpm at any time during the study. In addition, change in baseline-adjusted HR following neratinib administration was not significantly different from that following placebo administration at any of the postdose time points, thus suggesting that neratinib does not affect HR.

To verify that the assay was adequately sensitive to detect the effects of neratinib on QT prolongation, moxifloxacin was used as positive control. The statistical analysis results of the change from baseline QTcN for moxifloxacin compared with placebo are summarized in Table 1. The lower bound of the 90% CIs for ΔΔQTcN exceeded 5 milliseconds at 4 and 8 hours following dose administration, thus suggesting that the assay was acceptably sensitive for assessing the effect of neratinib on the QTc interval. In addition, the upper bound exceeded 10 milliseconds at these time points. Similar results were observed for QTcB, QTcF, and QTcI (data not shown).

The primary end point was the effect of therapeutic and supratherapeutic plasma neratinib concentrations on QTcN at all postdose time points. The effect of therapeutic concentrations was calculated as the difference in baseline-adjusted QTcN for neratinib compared with placebo (reference) and the effect of supratherapeutic concentrations was calculated as the difference in baseline-adjusted QTcN for neratinib plus ketoconazole compared with placebo plus ketoconazole (reference). The upper bounds of the 90% CI for ΔΔQTcN were <10 milliseconds greater than the respective reference, at all postdose time points, for both the therapeutic and the supratherapeutic dose comparisons (Tables 2 and 3). This finding confirms that neratinib does not produce an effect on the QT interval. In addition, at no time point was the lower bound of the 90% CI greater than 0, thus indicating that the effect of neratinib was, statistically, not different from placebo. Similar results were obtained for QTcB, QTcF, and QTcI (data not shown).

The PK/PD relationship between QTcN versus neratinib and moxifloxacin, separately, were examined graphically and statistically. Scatter plots of the individual QTcN change from baseline values versus neratinib plasma concentrations following administration of 240 mg neratinib, and coadministration of 240 mg neratinib with ketoconazole revealed that the QTcN change was randomly distributed with no clear trend relative to neratinib plasma concentrations (Fig. 2). Multiple linear regression models on placebo- and baseline-adjusted QTcN (ΔΔQTcN) versus log-transformed concentrations were fitted with postdose data for all subjects included in the statistical analysis for each analyte separately. A summary of the results for the models for each analyte is presented in Table 4. The slope coefficient on log-transformed neratinib concentrations of 0.33 was not significantly different from 0 (P = 0.3029), and the 95% CI (−0.29 to 0.94) contained 0, thus suggesting no relationship between plasma neratinib concentrations and change from baseline QTcN.

The conduct of a healthy subject TQT trial circumvents many of the obstacles to cardiac repolarization evaluations in cancer patients. Importantly, the collection of robust ECG data standardized for operator, interval measurements, and critical variables, including time of day, is possible. Administration of placebo and positive controls to healthy subjects is routine. Finally, recruitment of subjects for participation in a multiperiod, inpatient, procedure-intensive, nontherapeutic clinical trial is more readily accomplished for healthy populations than cancer patient populations. Because 240 mg neratinib has an acceptable safety and tolerability profile in healthy subjects when administered as a single dose, alone or in combination with ketoconazole, it was feasible to conduct the study in a noncancer patient population (13).

In this study, we evaluated the effects of neratinib on cardiac repolarization in healthy subjects while adhering to the recommended study design of the ICH E14. Our study fulfilled the requirements for a TQT study, including evaluations of both therapeutic and supratherapeutic neratinib concentrations and the appropriate negative and positive controls. Although many drugs in the oncology therapeutic area are not suitable for such an analysis, we were able to exploit the knowledge that supratherapeutic plasma concentrations of neratinib could be achieved by coadministration of well-tolerated doses of neratinib with ketoconazole.

A neratinib dose of 240 mg has shown activity in phase 2 trials of patients with ErbB-2 positive breast cancer (11, 17) and is currently in phase 3 development for breast cancer. As anticipated from previous experience with single 240 mg doses of neratinib in healthy subjects, the tolerability profile was only modestly diminished when plasma neratinib exposures (C_max and AUC) were increased ∼2.4- and 3-fold by coadministration of an inhibitor of neratinib metabolism. The plasma neratinib concentrations achieved exceeded those observed in cancer patients, and were well above those that could be tolerated by healthy subjects following administration of single neratinib doses without ketoconazole. All AEs observed during the period of neratinib administration were mild in severity and resolved without the need for medical intervention.

The evaluation of both therapeutic and supratherapeutic doses of neratinib produced no detectable change in baseline-adjusted QTc interval compared with the relevant comparator. QTcN was prespecified as the primary end point and method of correction, and no relevant differences were observed for QTcN or other corrections, including QTcB, QTcF, or QTcI. Likewise, an evaluation of the concentration-effect relationship between neratinib and QTcN failed to reveal any effect of neratinib on cardiac repolarization. Importantly, the positive control, moxifloxacin, produced a significant effect on QTcN, thus verifying that the study conditions were appropriately sensitive for detecting the effects on cardiac repolarization. The absence of an effect of neratinib on the QT interval is noteworthy given that other ErbB inhibitors, such as lapatinib and XL647, have been found to prolong the QT interval (18, 19).

Although this study provides valuable information related to the effect of neratinib on cardiac repolarization, it has several limitations. First, the use of the CYP3A4 inhibitor ketoconazole to achieve supratherapeutic plasma neratinib concentrations eliminates the utility of the study for assessments of effects of neratinib metabolites. Indeed, several active neratinib metabolites are produced by CYP3A4.²

The cardiac repolarization effect of investigational drugs is a component of risk-benefit assessment by patients, physicians, and regulatory agencies. Many approved anticancer pharmaceutical agents can prolong the QT interval, and the knowledge is valuable for the individualized selection of treatments and safe management of patients. Our study provides an approach to optimally characterize the cardiac repolarization liability of an investigational signal transduction inhibitor.

All authors were employees of Wyeth Research at the time of the study and may have held stock in Wyeth, which was acquired by Pfizer Inc., in October 2009.

We thank all subjects who participated in this study; the principal investigator, Robert S. Kahn, M.D., of Charles River Clinical Services, Tacoma, WA; and Joseph Ramcharan, Ph.D., for writing assistance that was funded by Pfizer Inc.

Grant Support: Wyeth Research, which was acquired by Pfizer Inc., in October 2009.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.