Evaluation of moxifloxacin in canine and non-human primate telemetry assays: Comparison of QTc interval prolongation by timepoint and concentration-QTc analysis

Results from a Joined Prospective Study to Evaluate the Sensitivity of the In Vivo Dog QT Assay in Line with the ICH E14/S7B Q&A Best Practices

The ICH E14/S7B Q&As highlighted the need for best practices concerning the design, execution, analysis, interpretation, and reporting of the in vivo non-rodent QT assay as a component of the integrated risk assessment to potentially support a TQT waiver or substitute. We conducted a dog telemetry study to assess the effects on QTc of six reference compounds (five positive and one negative) previously evaluated by Darpo et al. (2015) in humans. The sensitivity of the assay to detect QTc increases was determined, and exposure-response analysis was performed, as done in clinical practice. By-timepoint analysis showed QTc prolongation induced by moxifloxacin, dofetilide, dolasetron, ondansetron, and quinine within human relevant plasma exposures ranges. Moreover, a hysteresis was observed for quinine. As expected, levocetirizine showed no statistically significant effect on QTc across a range of exposure, well exceeding the therapeutic Cmax. Power analyses confirmed the study ability to detect statistically significant QTc changes of less than 10 milliseconds with 80% probability, even with a sample size as low as n = 4 animals. Finally, concentration-QTc modeling enabled to predict the minimal plasma concentration needed to detect a 10 milliseconds QTc prolongation, including for quinine. The comparison with clinical available data supported the relevance of dogs under these experimental conditions as a robust translational predictor of drug-induced QTc prolongation in humans as a key pillar of the integrated risk assessment.

Characterization of ascending dose canine telemetry model supports its use in E14/S7B QT integrated risk assessments

INTRODUCTION

Nonclinical evaluation of the cardiovascular effects of novel chemical or biological entities (NCE, NBEs) is crucial for supporting first-in-human clinical trials. One important aspect of these evaluations is the assessment of potential QT/QTc prolongation risk, as drug-induced QT prolongation can have catastrophic effects. The recent publication of E14/S7B Q&As allows for the situational incorporation of nonclinical QTc data as part of an integrated risk assessment for a Thorough QT (TQT) waiver application provided certain best practice criteria are met. Recent publications provided detailed characterization of nonclinical QTc telemetry data collected from the commonly used Latin square study design.

METHODS

To understand whether data from alternate telemetry study designs were sufficient to serve as part of the E14/S7B integrated risk assessment, we report the performance and translational sensitivity to identify clinical risk of QTc prolongation risk for an ascending dose telemetry design.

RESULTS

The data demonstrated low variability in QTci interval within animals from day to day, indicating a well-controlled study environment and limited concern for uncontrolled effects across dosing days. Historical study variances of the ascending dose design with n = 4 subjects, measured by least significant difference (LSD) and root mean square error (RMSE) values, were low enough to detect a + 10 ms QTci interval change, and the median minimum detectable difference (MDD) for QTci interval changes was <10 ms. Furthermore, concentration-QTci (C-QTci) assessments to determine +10 ms QTci increases for known hERG inhibitors were comparable to clinical CC values listed in the E14/S7B training materials, supporting the use of the ascending dose design in an E14/S7B integrated risk assessment.

DISCUSSION

These findings suggest that the ascending dose design can be a valuable tool in nonclinical evaluation of QT/QTc prolongation risk and the support of TQT waiver applications.

A simple accurate method for concentration-QTc analysis in preclinical animal models

INTRODUCTION

In preclinical cardiovascular safety pharmacology studies, statistical analysis of the rate corrected QT interval (QTc) is the focus for predicting QTc interval changes in the clinic. Modeling of a concentration/QTc relationship, common clinically, is limited due to minimal pharmacokinetic (PK) data in nonclinical testing. It is possible, however, to relate the average drug plasma concentration from sparse PK samples over specific times to the mean corrected QTc. We hypothesize that averaging drug plasma concentration and the QTc-rate relationship over time provides a simple, accurate concentration-QTc relationship bridging statistical and concentration/QTc modeling.

METHODS

Cardiovascular telemetry studies were conducted in non-human primates (NHP; n = 48) and canines (n = 8). Pharmacokinetic samples were collected on separate study days in both species. Average plasma concentrations for specific intervals (CAverage0-X) were calculated for moxifloxacin in canines and NHP using times corresponding to super-intervals for the QTc data statistical analysis. The QTc effect was calculated for each super-interval using a linear regression correction incorporating QT and HR data from the whole super-interval. The concentration QTc effects were then modeled.

RESULTS

In NHP, a 10.9 ± 0.06 ms (mean ± 95% CI) change in QTc was detected at approximately 1.5× the moxifloxacin plasma concentration that causes a 10 ms QTc change in humans, based on a 0-24 h super-interval. When simulating a drug without QT effects, mock, no effect on QTc was detected at up to 3× the clinical concentration. Similarly, in canines, a 16.6 ± 0.1 ms change was detected at 1.7× critical clinical moxifloxacin concentration, and a 0.04 ± 0.1 ms change was seen for mock.

CONCLUSIONS

While simultaneous PK and QTc data points are preferred, practical constraints and the need for QTc averaging did not prevent concentration-QTc analyses. Utilizing a 0-24 h super-interval method illustrates a simple and effective method to address cardiovascular questions when preclinical drug exposures exceed clinical concentrations.

Comparison of electrocardiogram and blood pressure recording methods in non-rodent toxicology studies: A retrospective analysis

Our study retrospectively examines 51 non-rodent general toxicology studies conducted over the past 8 years to ascertain the influence of recording methodologies on baseline cardiovascular (CV) parameters and statistical sensitivity. Specifically, our work aims to evaluate the frequency of cardiovascular parameter recording categorized by therapeutic modality and study type, to assess the variability in these parameters based on measurement techniques, and to determine the sample sizes needed for detecting relevant changes in heart rate (HR), blood pressure (BP), and QTc interval in non-human primate (NHP) studies. Results indicate that electrocardiogram (ECG) measurements in dogs and NHP were recorded in 63% of studies, combined with BP recording in 18% of studies, while BP was never recorded alone. Trend analysis reveals a decline in the utilisation of restraint-based methods for ECG measurements post-2017, to the benefit of telemetry-based recordings, particularly Jacketed External Telemetry (JET). There was a marked difference in baseline values, with restraint-based methods showing significantly higher HR and QTc values compared to JET, likely linked to animal stress. Further analysis suggests an unrealistic and unethical sample size requirement in NHP studies for detecting biologically meaningful CV parameter changes using restraint-based methods, while JET methods necessitate significantly smaller sample sizes. This retrospective study indicates a notable shift from snapshots short-duration, restraint-based methods towards telemetry approaches over the recent years, especially with an increased usage of implanted telemetry. The transition contributes to potential consensus within industry or regulatory frameworks for optimal practices in assessing ECG, HR, and BP in general toxicology studies.

No Influence of Asundexian on Cardiac Repolarization

Inhibition of activated factor XI reduces thrombogenesis while maintaining physiological hemostasis, with the expectation of reduced bleeding risk compared with standard of care in the clinical setting. Asundexian (BAY 2433334), an activated factor XI inhibitor, is in clinical development for the prevention of thromboembolic events. The effect of asundexian and its plasma metabolite M10 on cardiac repolarization and potential interactions with the hNav1.5 sodium, hCav1.2 calcium, and human ether-à-go-go-related gene (hERG) potassium channels was investigated in vitro. Additionally, asundexian effects on cardiac parameters and electrocardiogram were examined in telemetered beagle dogs. A randomized, placebo-controlled, 4-way crossover, thorough QT study in healthy adults evaluated the influence of 50 and 150 mg of asundexian on the corrected QT interval, including 400 mg of moxifloxacin as positive control. Across all studies, asundexian and M10 were not associated with any effects on cardiac repolarization. The largest in vitro effects of asundexian (approximately 20% inhibition) were seen for hCav1.2 and hERG. Throughout the thorough QT study, the upper limits of the one-sided 95% confidence interval of placebo-corrected mean changes from baseline in Fridericia corrected QT for 50 and 150 mg of asundexian were below Δ = 10 milliseconds. Asundexian demonstrated favorable safety and tolerability profiles.

The New S7B/E14 Q&A Document Provides Additional Opportunities to Replace the Thorough QT Study

Since 2015, concentration-QTc (C-QTc) analysis has been used to exclude the possibility that a drug has a concerning effect on the QTc interval. This has enabled the replacement of the designated thorough QT (TQT) study with serial electrocardiograms (ECGs) in routine clinical pharmacology studies, such as the first-in-human (FIH) study. The E14 revision has led to an increased proportion of FIH studies with the added objective of QT evaluation, with the intention of replacing the TQT study. With the more recent revision of the S7B/E14 Q&A document in February 2022, nonclinical assays/studies can be brought into the process of regulatory decisions at the time of marketing application. If the hERG (human ether-a-go-go-related gene) and the non-rodent in vivo study are conducted according to the described best practices and are negative, the previous requirement that a QTc effect of >10 milliseconds must be excluded in healthy subjects at plasma concentrations 2-fold above what can be seen in patients can be reduced to covering the concentrations seen in patients. For drugs that cannot be safely given in high doses to healthy subjects, ECG evaluation is often performed at the therapeutic dose in patients. If a QTc effect of >10 milliseconds can be excluded, an argument can be made that the drug should be considered as having a low likelihood of proarrhythmic effects due to delayedrepolarization, if supported by negative best practices hERG and in vivo studies. In this article, we describe what clinicians involved in early clinical development need to understand in terms of the hERG and in vivo studies to determine whether these meet best practices and therefore can be used in an integrated clinical/nonclinical QT/QTc risk assessment.

Safety pharmacology 2023 and implementation of the ICH E14/S7B Q&A guidance document

This editorial prefaces the annual themed issue on safety pharmacology (SP) methods published since 2004 in the Journal of Pharmacological and Toxicological Methods (JPTM). We highlight here the content derived from the recent 2022 Safety Pharmacology Society (SPS) and Canadian Society of Pharmacology and Therapeutics (CSPT) joint meeting held in Montreal, Quebec, Canada. The meeting also generated 179 abstracts (reproduced in the current volume of JPTM). As in previous years the manuscripts reflect various areas of innovation in SP including a comparison of the sensitivity of cross-over and parallel study designs for QTc assessment, use of human-induced pluripotent stem cell (hi-PSC) neuronal cell preparations for use in neuropharmacological safety screening, and hiPSC derived cardiac myocytes in assessing inotropic adversity. With respect to the latter, we anticipate the emergence of a large data set of positive and negative controls that will test whether the imperative to miniaturize, humanize and create a high throughput process is offset by any loss of precision and accuracy.

Best practice considerations for nonclinical in vivo cardiovascular telemetry studies in non-rodent species: Delivering high quality QTc data to support ICH E14/S7B Q&As

The ICH E14/S7B Questions and Answers (Q&As) guideline introduces the concept of a "double negative" nonclinical scenario (negative hERG assay and negative in vivo QTc study) to demonstrate that a drug does not produce a clinically relevant QT prolongation (i.e., no QT liability). This nonclinical "double negative" data package, along with negative Phase 1 clinical QTc data, may be sufficient to substitute for a clinical Thorough QT (TQT) study in some specific cases. While standalone GLP in vivo cardiovascular studies in non-rodent species are standard practice during nonclinical drug development for small molecule programs, a variety of approaches to the design, conduct, analysis and interpretation are utilized across pharmaceutical companies and contract research organizations (CROs) that may, in some cases, negatively impact the stringent sensitivity needed to fulfill the new Q&As. Subject matter experts from both Pharma and CROs have collaborated to recommend best practices for more robust nonclinical cardiovascular telemetry studies in non-rodent species, with input from clinical and regulatory experts. The aim was to increase consistency and harmonization across the industry and to ensure delivery of high quality nonclinical QTc data to meet the proposed sensitivities defined within the revised ICH E14/S7B Q&As guideline (Q&As 5.1 and 6.1). The detailed best practice recommendations presented here cover the design and execution of the safety pharmacology cardiovascular study, including optimal methods for acquiring, analyzing, reporting, and interpreting the resulting QTc and pharmacokinetic data to allow for direct comparison to clinical exposures and assessment of safety margin for QTc prolongation.

Comparing the sensitivity of cross-over and parallel study designs for QTc assessment: An analysis based on a single large study of moxifloxacin in 48 nonhuman primates

The cardiovascular safety pharmacology (SP) study conducted to satisfy ICH S7A and S7B has commonly used a cross-over study design where each animal receives all treatments. In an increasing number of cases, cross-over designs are not possible and parallel studies have to be used. These can seldom be as large as 8 animals/treatment to match an n = 8 cross-over. Animals in parallel designs receive only one treatment. Parallel studies will have a different sensitivity to detect changes. This sensitivity is a critical question in using nonclinical QTc evaluations to support an integrated proarrhythmic risk assessment under the newly released ICH E14/S7B Q&As. The current analysis used a study large enough (n = 48) to be analyzed both as a parallel and as a cross-over design to directly compare the performance of the two experimental designs coupled to different statistical models, while all other study conduct aspects were the same. A total of 48 nonhuman primates (NHP) received 2 different treatments twice: vehicle, moxifloxacin (80 mg/kg), vehicle, moxifloxacin (80 mg/kg). Post-dose QTc interval data were recorded for 48 h for each treatment. Data were analyzed using 12 animals randomly selected for each treatment in a parallel design or as an n = 48 animal cross-over study. Different statistical models were used. The primary endpoint was the residual deviation (sigma) from the models applied to hourly time intervals. The sigma was used to determine the minimal detectable difference (MDD) for the study design-statistical model combination. Two statistical models were applicable to either study design. They gave similar sigma and resulting MDD values. In cross-over designs, the individual animal identification (ID) can be used in the statistical model. This enabled the smallest MDD value. Simple statistical models for analysis were identified: Treatment + Baseline for parallel designs and Treatment + ID for cross-over designs. The statistical sensitivity of NHP parallel study designs is reasonable (MDD for n = 6 of 12.7 ms), and in combination with testing exposures higher than likely to be necessary in man could be used in an integrated risk assessment. Where sensitivity of the NHP in vivo QTc assessment is critical, the cross-over design enabled a higher sensitivity (MDD 12.2 ms for n = 4; 8 ms for n = 8).

The incidence of spontaneous arrhythmias in telemetered beagle dogs, Göttingen Minipigs and Cynomolgus non-human primates: A HESI consortium retrospective analysis

INTRODUCTION

Characterization of the incidence of spontaneous arrhythmias to identify possible drug-related effects is often an important part of the analysis in safety pharmacology studies using telemetry.

METHODS

A retrospective analysis in non-clinical species with and without telemetry transmitters was conducted. Electrocardiograms (24 h) from male and female beagle dogs (n = 131), Göttingen minipigs (n = 108) and cynomolgus non-human primates (NHP; n = 78) were analyzed.

RESULTS

Ventricular tachycardia (VT) was observed in 3% of the dogs but was absent in minipigs and NHPs. Ventricular fibrillation (VF) was not observed in the 3 species. Ventricular premature beats (VPBs) were more frequent during daytime and atrioventricular blocks (AVBs) were more frequent at night in all species. A limited number of animals exhibited a high arrhythmia frequency and there was no correlation between animals with higher frequency of an arrhythmia type and the frequency of other arrythmias in the same animals. Clinical chemistry or hematology parameters were not different with or without telemetry devices. NHP with a transmural left ventricular pressure (LVP) catheter exhibited a greater incidence of VPBs and PJCs compared to telemetry animals without LVP.

DISCUSSION

All species were similar with regards to the frequency of ventricular ectopic beats (26-46%) while the dog seemed to have more frequent junctional complexes and AVB compared to NHP and minipigs. Arrhythmia screening may be considered during pre-study evaluations, to exclude animals with abnormally high arrhythmia incidence.

Evaluation of levocetirizine in beagle dog and cynomolgus monkey telemetry assays: Defining the no QTc effect profile by timepoint and concentration-QTc analysis

In prior clinical studies, levocetirizine (LEVO) has demonstrated no effect on ventricular repolarization (QTc intervals), therefore it is a relevant negative control to assess in nonclinical assays to define low proarrhythmic risk. LEVO was tested in beagle dog and cynomolgus monkey (nonhuman primate [NHP]) telemetry models to understand the nonclinical-clinical translation of this negative control. One oral dose of vehicle, LEVO (10 mg/kg/species) or moxifloxacin (MOXI; 30 mg/kg/dog; 80 mg/kg/NHP) was administered to instrumented animals (N = 8/species) using a cross-over dosing design; MOXI was the in-study positive control. Corrected QT interval values (QTcI) were calculated using an individual animal correction factor. Blood samples were taken for drug exposure during telemetry and for pharmacokinetic (PK) analysis (same animals; different day) for exposure-response (C-QTc) modeling. Statistical analysis of QTc-by-timepoint data showed that LEVO treatment was consistent with vehicle, thus no effect on ventricular repolarization was observed over 24 h in both species. PK analysis indicated that LEVO-maximum concentration levels in dogs (range: 12,300-20,100 ng/ml) and NHPs (range: 4090-12,700 ng/ml) were ≥4-fold higher than supratherapeutic drug levels in clinical QTc studies. Slope analysis values in dogs (0.00019 ms/ng/ml) and NHPs (0.00016 ms/ng/ml) were similar to the human C-QTc relationship and indicated no relationship between QTc intervals and plasma levels of LEVO. MOXI treatment caused QTc interval prolongation (dog: 18 ms; NHP: 29 ms). The characterization of LEVO in these non-rodent telemetry studies further demonstrates the value and impact of the in vivo QTc assay to define a "no QTc effect" profile and support clinical safety assessment.

Comparison of validity of standard nonclinical group size selection versus standard clinical group sizes for nonhuman primate QTc prolongation evaluation

The number of animals used in a nonhuman primate (NHP) in vivo QTc assessment conducted as part of the safety pharmacology (SP) studies on a potential new drug is relatively small (4-8 subjects). The number is much smaller than the number of healthy volunteers in a conventional thorough QT (TQT) study (40-60 volunteers). How is it possible that such small studies could offer an equivalent sensitivity in an integrated nonclinical and clinical cardiac repolarization risk assessment? This study provided the opportunity to empirically demonstrate in a large number of NHPs the performance of a nonclinical evaluation at a similar size to a TQT study. By contrasting an analysis mimicking the sampling and aggregation of QTc interval data in a manner which is TQT-like with a more conventional SP-like analysis it was demonstrated that the SP-like analysis was more sensitive. In prospective power calculations 80% power at p = 0.05 can be achieved for a 5 ms QTc change with only n = 8 NHPs using the SP-like analysis and in a group of only 4 NHPs 80% power to detect 10 ms could be achieved. By contrast groups of 24 NHPs would be required to achieve 80% power to detect 5 ms using the TQT-like sampling and aggregation approach. Overall, this study has demonstrated that smaller safety pharmacology in vivo QTc assessments using all the available data in larger data aggregates can achieve sensitivity comparable to a human TQT study.

Improving the in Vivo QTc assay: The value of implementing best practices to support an integrated nonclinical-clinical QTc risk assessment and TQT substitute

Recent updates and modifications to the clinical ICH E14 and nonclinical ICH S7B guidelines, which both relate to the evaluation of drug-induced delayed repolarization risk, provide an opportunity for nonclinical in vivo electrocardiographic (ECG) data to directly influence clinical strategies, interpretation, regulatory decision-making and product labeling. This opportunity can be leveraged with more robust nonclinical in vivo QTc datasets based upon consensus standardized protocols and experimental best practices that reduce variability and optimize QTc signal detection, i.e., demonstrate assay sensitivity. The immediate opportunity for such nonclinical studies is when adequate clinical exposures (e.g., supratherapeutic) cannot be safely achieved, or other factors limit the robustness of the clinical QTc evaluation, e.g., the ICH E14 Q5.1 and Q6.1 scenarios. This position paper discusses the regulatory historical evolution and processes leading to this opportunity and details the expectations of future nonclinical in vivo QTc studies of new drug candidates. The conduct of in vivo QTc assays that are consistently designed, executed and analyzed will lead to confident interpretation, and increase their value for clinical QTc risk assessment. Lastly, this paper provides the rationale and basis for our companion article which describes technical details on in vivo QTc best practices and recommendations to achieve the goals of the new ICH E14/S7B Q&As, see Rossman et al., 2023 (this journal).

The Challenges of Predicting Drug-Induced QTc Prolongation in Humans

The content of this article derives from a Health and Environmental Sciences Institute (HESI) consortium with a focus to improve cardiac safety during drug development. A detailed literature review was conducted to evaluate the concordance between nonclinical repolarization assays and the clinical thorough QT (TQT) study. Food and Drug Administration and HESI developed a joint database of nonclinical and clinical data, and a retrospective analysis of 150 anonymized drug candidates was reviewed to compare the performance of 3 standard nonclinical assays with clinical TQT study findings as well as investigate mechanism(s) potentially responsible for apparent discrepancies identified. The nonclinical assays were functional (IKr) current block (Human ether-a-go-go related gene), action potential duration, and corrected QT interval in animals (in vivo corrected QT). Although these nonclinical assays demonstrated good specificity for predicting negative clinical QT prolongation, they had relatively poor sensitivity for predicting positive clinical QT prolongation. After review, 28 discordant TQT-positive drugs were identified. This article provides an overview of direct and indirect mechanisms responsible for QT prolongation and theoretical reasons for lack of concordance between clinical TQT studies and nonclinical assays. We examine 6 specific and discordant TQT-positive drugs as case examples. These were derived from the unique HESI/Food and Drug Administration database. We would like to emphasize some reasons for discordant data including, insufficient or inadequate nonclinical data, effects of the drug on other cardiac ion channels, and indirect and/or nonelectrophysiological effects of drugs, including altered heart rate. We also outline best practices that were developed based upon our evaluation.