Implementing ICH E14 to define cardiac safety in new drug development
In May 2005, concepts on how cardiac safety of a new drug should be determined during its clinical development were formalized with completion of Step 4 of an intensive International Conference on Harmonisation (ICH) process, resulting in the document: ICH Harmonised Tripartite Guideline: The Clinical Evaluation of QT/QTc Interval Prolongation and Proarrhythmic Potential for Non-Antiarrhythmic Drugs (ICH E14)[http://www.ich.org/cache/compo/276-254-1.html]. This guidance has now been transmitted for implementation by the regulatory bodies of the three ICH regions—European Union, Japan, and the United States. The purpose of this article is to provide a summary of this guidance document with commentary on areas requiring more definition/details that should be of assistance in developing strategies for its practical implementation to ensure compliance with best practices.
ICH E14 represents a significant landmark. It is the first globally harmonized regulatory guidance for assessment of cardiac safety during clinical drug development. Concurrently with ICH E14, a second ICH guideline also reached Step 4 status for implementation by the three ICH regions: The Non-Clinical Evaluation of the Potential for Delayed Ventricular Polarization (QT Interval Prolongation) by Human Pharmaceuticals (ICH S7B) [http://www.ich.org/cache/compo/276-254-1.html]. The objective of ICH S7B is to describe the types of studies that can be carried out during nonclinical testing to identify the potential for a drug candidate to delay ventricular repolarization and relate this effect to the concentrations of the drug candidate and its metabolites.1 The results of these nonclinical studies are intended to be considered in conjunction with other clinical studies of the sort that will be discussed in this article to provide an integrated assessment of cardiac proarrhythmic risk.
Cardiac safety of drugs is the most common cause of drug withdrawal from the market or delay in or denial of regulatory approval for marketing. Examples include astemizole, cisapride, grepafloxacin, levacetylmethadol, lidoflazine, prenylamine, sertindole, sparfloxacin, terfenadine, and terodiline. In addition, it is important to note that numerous biologics have been associated with a QTc prolongation effect and/or proarrhythmias. These substances include oxytocin, vasopressin, and gonadotropin-releasing hormone agonists and antagonists. Although this association is rare and the causality uncertain, the E14 guidance should also provide an aid in formulating clinical development programs for biologics as well.2-5
The global clinical research process should anticipate that regulators worldwide will require cardiac safety data in accordance with the principles set forth in ICH E14. This will result in much more careful and uniform evaluation of the ECG profiles of new drugs. Over time, significant data and information will be aggregated in a standard format to facilitate effective assessment of compounds within a therapeutic area and enable sophisticated data mining to detect patterns within and across drug classes.
A primary focus of ICH E14 is the call for a Thorough QT/QTc Study or, stated more appropriately, a Thorough ECG Trial (TET).6 Although ICH E14 focuses on QTc interval, the more expansive TET designation points to the need that the trial should address the full description of all ECG changes observed. The TET is a robust trial to determine the cardiac (ECG) effects of a new drug on healthy subjects and, with exceptions that are detailed later in this article, is likely to be required for every new bioactive compound.7 In addition, a TET is also likely to be required when a drug already approved is being considered for development of a new dose, route of administration, or in a new patient population, especially when increased exposure and/or concentration are likely to occur.
The TET should be conducted early in the clinical development program as soon as relevant pharmacologic and early clinical efficacy data are available. It is advisable to conduct the TET prior to exposing large patient populations with co-morbidities in later Phase IIb or III studies. Depending on compound specifics, a TET is usually conducted in early Phase II. This sequencing offers a benefit to sponsors, who reach an important milestone of understanding the cardiac safety profile of their product, to make a go/no go decision and to design proper ECG collection schedules in the target population based on the results of the TET.
TETs may utilize a parallel or crossover design. Crossover studies are thought to provide an advantage in that classically they require a smaller number of subjects than parallel designs. However, the endpoint is not a difference between treatment and placebo on-treatment but a placebo-corrected change from baseline, and as such there is minimal advantage in the use of a crossover design for a reduced sample size. In contrast, parallel study designs are preferable for drugs with a long half-life for the parent drug or its pharmacologically active metabolites, or if multiple doses or treatment groups are being compared.8 As a practical matter, crossover studies take longer to complete and run higher risks associated with subject drop-out9 and potential period effects that may invalidate the trial. Experience of a leading ECG core laboratory across about 45 TETs indicates that sponsors have chosen parallel design in about 70% of trials, with multiple-dose studies running at a ratio of about 2:1 over single-dose studies.10 Single-dose studies are appropriate when the pharmacokinetic parameters of a single dose are identical to those parameters after multiple dosing to steady state.
A typical TET design includes at least four treatment arms: placebo, positive control, therapeutic dose (single or multiple), and a supratherapeutic dose of the test agent. ICH E14 specifies that consideration should be given to addition of arms using a drug from the same class if the compound being studied belongs to a class that has been associated with QT/QTc prolongation.11 Sponsors have elected to add therapeutic comparator arms to TETs during the most recent two years for a range of scientific and competitive purposes. The use of a positive control with a well-defined QT/QTc effect is intended to detect a change that represents the threshold of regulatory concern—typically about 5 milliseconds with an upper confidence bound of 10 milliseconds.12 This technique confirms the sensitivity of the assay used for a specific trial. Oral moxifloxacin has been overwhelmingly the preferred drug, but not exclusively so, used for this purpose. Moxifloxacin may be used as a single dose even in studies with multiple doses to steady state designs. In these cases, placebo is administered in the positive control arm for the days just prior to a single dose of moxifloxacin so that the positive control arm is treated identically to all other treatment arms in the trial. This treatment arm need not be blinded; if feasible, the active agent and placebo should be double-blinded. The positive control should be employed within the randomization schedule applied to the placebo and new agent, and thus not used either first or last in a crossover design or before randomization in a parallel design.
The goal of the supratherapeutic dose is to expose healthy volunteers to the highest plasma levels and systemic exposure of the parent drug and active metabolites that could occur in the patient population, including those associated with maximal inhibition of metabolism of the drug and impairment of liver and kidney functions and target disease-related susceptibility. Obviously, if a maximum tolerated dose of the new agent has not been properly established in Phase I, it may require a separate study to determine whether the ideal supratherapeutic dose is tolerable and safe for the volunteer population enrolled in the TET.
Collection of ECGs at multiple timepoints at baseline for each treatment group and again at the same timepoints for each treatment is an important design element of the TET. This technique is the most reliable way to determine if the trial conducted has minimized the large degree of spontaneous QTc variability that can be observed within a single subject. Variability results from many factors, such as diurnal variation, effects of food, and other autonomic influences. Typically, three baseline ECGs are collected at each of many timepoints that match the collection of all on-treatment ECG timepoints. Further, meal composition and schedules, activity level, and environmental stress should be equalized between baseline and on-treatment periods. Conditions of ECG recording should always be consistent, e.g., supine, immediately prior to blood sampling. These special attributes for conducting a TET study require a well-equipped clinical pharmacology unit (CPU) with trained and experienced staff.
Perhaps one of the most significant clarifications that evolved in the process of finalizing ICH E14 was the move from "time-averaged" to "time-matched" analysis. Earlier ECG trials usually employed the time-averaged approach. The "time-averaged" approach requires that all baseline ECGs for each subject are averaged and that all on-treatment ECGs for each subject are averaged. Subjects on active treatment are corrected for placebo and the mean change from baseline for each subject is calculated. The procedure is then repeated for all individuals in a treatment group. The obvious limitation of this approach is that the maximum effects on QTc intervals are "diluted," and therefore, in absence of a separate analysis of maximum mean central tendency, a false negative conclusion may be drawn. To avoid this, the "time-matched" method indicated in the guidance was developed. In the "time-matched" method, each timepoint on-treatment (active, placebo or positive control) is compared with the baseline value for the corresponding timepoint. The value at each timepoint is achieved by averaging the ECGs taken at that timepoint for each individual. This enables one to calculate an increase in QTc interval from baseline (ΔQTc) at each timepoint for each treatment arm. The parameter of interest is the largest difference (ΔΔQTc) between the change on the active treatment and the placebo (the two ΔQTc at the same timepoint). This approach requires that the timepoints at baseline and on-treatment be identical. This allows for change from placebo analysis to help ensure that impact of spontaneous QTc changes is minimized.
In addition, an outlier analysis should be conducted. The focus of this analysis should be to identify the percentage of subjects (as opposed to the number of ECG observations) that exhibit a pre-determined QTc effect (QTc change from baseline of 30–60 milliseconds or ≥60 milliseconds and a new QTc value of >500 milliseconds). While other thresholds are frequently recommended (such as new QTc values of >450 milliseconds and >480 milliseconds), these are often redundant or overlapping to the above more generally accepted criteria. Any new instances of morphological changes such as abnormal T or U waves should be noted.
The "time-matched" method of analysis has significant implications on sample size when examined in conjunction with the threshold of regulatory concern of around 5 milliseconds as evidenced by an upper bound of the 95% confidence interval around the mean effect on QTc of 10 milliseconds.13 The sample size for parallel or crossover studies results from defining the variance in QTc duration and results in about 60–80 subjects (typically 65 subjects) exposed per treatment. Thus for a parallel study with four treatment arms, 260 (4 times 65) subjects are recruited, whereas for a crossover just over 65 volunteers (allowing for potential dropouts) are exposed to each treatment and washout period.
If the upper bound of the 95% confidence interval of the "time-matched" effect on QTc interval exceeds 10 milliseconds, the guidance designates the study as positive. The impact of a positive TET on subsequent trials will be discussed later in this article. However, it is important to note that a positive TET does not of itself imply that a drug is proarrhythmic. There is little doubt that drugs prolonging the mean QTc by >20 milliseconds have a substantial likelihood of being proarrhythmic and that proarrhythmias could be observed during drug development. In contrast, the 5-millisecond threshold can seem overly cautious; yet, the potentially fatal torsade de pointes (TdP) arrhythmia can occur in the absence of QTc prolongation. In practice, the risk associated with this scenario is too remote to be of clinical or regulatory concern; however, this fact does point out that even rigorous TET standards may fail to detect all drugs with a very low proarrhythmic potential.14
Careful consideration should be given to the degree of QTc prolongation in conjunction with the overall risk/benefit ratio of the compound. A range for basic risk assessment can be found in Table 1. Also, it is important to note that QTc is an imperfect surrogate for the risk of TdP. While there is good evidence that the extent of QTc prolongation relates to the degree of TdP risk, there could be other properties that might mitigate or enhance risk following prolongation of the QTc interval.15
Table 1. Likely prognostic significance of mean placebo-corrected peak effects on QTc interval (clinical dose x 1)
A drug developer faces the potential, albeit not very likely, scenario of executing an appropriate TET study that yields a single timepoint on a single subject showing a categorical response discussed earlier while the trial otherwise appears negative. In this case, the use of PK/PD relationships will prove important. If the slope of the plasma concentration of parent and possibly metabolites versus change in QTci (QTc interval following individual correction) is relatively flat, then an abnormal single timepoint is likely to represent a false positive signal.
QT interval has an inverse relationship to heart rate. This means that it is important to correct the measured QT intervals for heart rate to determine if they are prolonged compared with those calculated at baseline. Because a TET is designed to detect prolongation of a relatively subtle threshold of 5 milliseconds, the impact of QT correction formulae is important. The most commonly used formulae are Bazett's and Fridericia's, and data corrected by these formulae are required for at least historical comparative purposes. For TET studies, data using both formulae should be submitted, along with uncorrected QT, RR, and heart rate data.16 In recent years, techniques that correct using QT/RR relationships using individual subject data have been developed. This approach is considered most appropriate for a TET because each individual's exponent for correction is defined as the single exponent that gives a heart rate to QTci slope relationship as close to one as possible.17
Beyond ICH E14, regional guidance may be implemented in incremental areas. For example, the FDA is currently implementing new cardiac safety data submission, review, and warehousing requirements that include support of an HL7 XML annotated ECG waveform standard. The FDA has entered into a Cooperative Research and Development Agreement (CRADA) with Mortara Instrument to develop a digital ECG warehouse, which will provide tools for FDA reviewers and the industry to manage, review, and analyze digitally submitted cardiac safety data and waveforms more efficiently.18
The method of determining interval duration measurements (IDMs) is among the most critical, yet the most widely misunderstood, aspects of the ECG core laboratory industry.19 Manual IDM determination has long been acknowledged as the best method to determine the ECG IDMs, especially QT intervals. In this method, a highly trained and skilled professional marks IDMs on an ECG waveform without the aid of or visibility to any measurements determined by ECG collection devices or software algorithms. Today, the blinded analysis is typically accomplished using on-screen calipers to mark or "annotate" IDMs on a digital ECG waveform. The waveform along with annotations can be saved in a standard format and recalled for viewing later in the analysis process or during regulatory review by the FDA. This truly manual methodology has been generally accepted to be more precise and has been preferred by sponsors for use in greater than 95% of the TETs contracted out to a leading ECG core laboratory.20 The manual standard process has been considered more precise because it addresses nuances in ECG waveforms that cannot be recognized and interpreted effectively by a software algorithm.
Interim drafts of the E14 guidance made reference to "semi-automatic" IDM methodologies, in which software algorithms execute initial caliper placement and one or more individuals subsequently review already annotated waveforms on-screen and make determinations on which caliper placements require modification or "adjudication." This methodology may be appropriate for routine early and later phase studies where cardiac safety is not targeted as a primary or secondary endpoint and where immediate patient safety is the primary use for the ECG data.
The final guidance addresses the significance of methodology and states, "the method chosen will depend on the level of precision appropriate for a given trial." The TET is cited as an example of a trial requiring particular attention to IDM methodology. The guidance as ratified makes no reference to the acceptability of semi-automated methodologies. It simply specifies "measurement by a few skilled readers (whether or not assisted by a computer) operating from a centralized ECG laboratory."21 Semantics of the parenthetical expression have already sparked a debate, with some claiming that the reference to computer assistance implies semi-automated measurements. In fact, one member of the ICH E14 Expert Working Group has stated unequivocally his understanding that this reference to computer assistance is not intended to include semi-automated measurements whereby trained professional readers operating from a central laboratory review the machine-read ECGs and recompute any intervals for which the automated placement of the fiduciary points is considered to be inappropriate.22
Another important principle concerning methodology is the concept of consistency of a clinical development program with respect to data that are pooled for analysis. The TET stands alone in this regard, which means that a program manager may safely choose the E14 specified true manual measurement standard even if earlier trials employed semi-automated methods. Later phase development for the same compound can leverage semi-automated methodologies if desired, provided that consistency at the level of data pooling is respected. Sponsors should insist on on-site, visual inspection of ECG core laboratory processing to confirm that the methodology to be deployed is the one specified by them.
Continuous 12-lead Holter recording technologies have emerged as a just-in-time advancement critical to successful conduct of TETs. These platforms should not be confused with traditional two- or three-channel analog Holter devices used primarily for heart rate variability and arrhythmia analysis studies. Continuous 12-lead Holter solutions have been used extensively in TETs performed by at least one leading ECG core laboratory. The platform has been chosen for about 75% of the approximately four-dozen studies contracted to this laboratory. This methodology has gained regulatory acceptance, and multiple compounds have obtained approval, while others are in the approval process.23
These platforms enable continuous 12-lead ECG collection throughout an entire 24-hour period. ECG waveforms are collected and stored on solid-state memory cards that are forwarded to the ECG core laboratory. Continuous 12-lead ECG platforms enhance precision, as the core laboratory can extract multiple ECGs at specific timepoints without having to rely on CPU staff for exactly timed collection. If an ECG at a timepoint is of poor quality, the core laboratory can extract a quality ECG closest to the specified timepoint. In addition, the technology is highly cost-effective in two important ways. First, CPU costs for collection of 15,000 to 20,000 ECGs are displaced by once-daily subject connection to the continuous recording device and occasional safety ECGs collected with traditional 12-lead equipment that features local printing capabilities. Second, the ability to extract additional timepoints for analysis from the continuous 12-lead recording platform provides an easy mechanism for adding additional timepoints to a study retroactively to address unexpected issues, such as the behavior of a metabolite.
ICH E14 recognizes that TETs may not be possible with some compounds due to safety or tolerability concerns, e.g., cytotoxic oncology agents. It may be possible that a TET or a variation can be conducted in patients. When this is not possible either, other means of focusing on the potential for QT/QTc prolongation should be explored, such as collection of ECGs at multiple timepoints across multiple doses in a tightly controlled setting during early clinical pharmacology studies.
If a drug is administered under a continuous cardiac monitoring setting, then the need for a TET study may be reduced. Also, there may be a reduced need if a drug is not absorbed system-wide, such as with certain topical products and therapies administered locally.
TET results are key to defining cardiac safety data collection and analysis in later phase development. In almost all cases, a negative TET will allow for routine collection of ECGs in later phase trials in a manner consistent with current practice for the relevant therapeutic area. However, if the TET is negative and non-clinical results were strongly positive and the discrepancy cannot be explained by data, then expanded ECG collection and analysis in later phase development should be considered.
More intensive ECG collection and analysis is required subsequent to a positive TET. The focus should be on fully characterizing the QT/QTc effects of the new drug candidate on the target patient population, with emphasis on dose- and concentration-related effects. Analysis should include both mean QT/QTc change and outliers. In addition, particular interest should be placed on patient subgroups such as those with:
ICH E14 establishes a clear mandate for the important role of the TET in clinical development programs. However, it is important to consider other data and issues to derive a comprehensive "cardiac safety package." Chief among these are key nonclinical studies and verification of cardiac effect through routine monitoring following a negative TET. Also important are relative risk/benefit and comparison to other available in-class therapies. Negative nonclinical data and a positive TET need not result in a "black box." Even in the face of a negative TET, a strong nonclinical signal may still require intense monitoring throughout the entire development program.
A consistent high-level study design framework and efforts to standardize the format for data analysis by the FDA and others will expedite the aggregation of data and information that will expand our understanding of the nuances and complexities of the effect of compounds on cardiac function. The role of ECG core laboratories will continue to grow in scope and importance, providing an important trusted third-party function in analysis of data, along with the science, process, and technology required to meet the rigorous demands of both high-intensity TETs, requiring close collaboration with sponsors and CPUs, and massively distributed later phase trials.
Scott Grisanti is senior vice president and chief marketing officer with eResearchTechnology, Inc., 1150 U.S. Highway 22 East, Bridgewater, NJ 08807, (908) 203-6430, fax (908) 947-5122, email: email@example.com Joel Morganroth, MD, is clinical professor of medicine at the University of Pennsylvania School of Medicine, Philadelphia, PA, and chief scientist with eResearchTechnology, Inc., (215) 972-0420, fax (215) 972-8775, email: firstname.lastname@example.org Rashmi R. Shah, BSc, MD, FRCP, is a pharmaceutical consultant. He was previously a senior clinical assessor with the Medicines and Healthcare products Regulatory Agency, London, UK, and EU topic leader and member of ICH E14 Expert Working Group, email: email@example.com
1. ICH Harmonised Tripartite Guideline: The Non-Clinical Evaluation of the Potential for Delayed Ventricular Polarization (QT Interval Prolongation) by Human Pharmaceuticals, S7B, Step 4, 12 May 2005, Page 2, Section 2.1.
2. V.F. Mauro, J.F. Bingle, S.M. Ginn, F.M. Jafri, "Torsade de pointes in a patient receiving intravenous vasopressin," Crit Care Med., 16, 200–201 (1988).
3. B. Charbit, C. Funck-Brentano, E. Samain, V. Jannier-Guillou, P. Albaladejo, J. Marty, "QT Interval Prolongation After Oxytocin Bolus During Surgical Induced Abortion," Clin Pharmacol Ther., 76, 359–364 (2004).
4. R. Marfella, F. Nappo, L. De Angelis, M. Siniscalchi, F. Rossi, D. Giugliano, "The Effect of Acute Hyperglycaemia on QTc Duration in Healthy Man," Diabetologia, 43, 571–575 (2000).
5. S.C. Liou, C. Chen, S.Y. Wong, K.M. Wong, "Ventricular Tachycardia After Oxytocin Injection in Patients with Prolonged Q-T Interval Syndrome—Report of Two Cases," Acta Anaesthesiol Sin., 36, 49–52 (1998).
6. ICH Harmonised Tripartite Guideline: The Clinical Evaluation of QT/QTc Interval Prolongation and Proarrhythmic Potential for Non-Antiarrhythmic Drugs, E14, Step 4, 12 May 2005, Page 3, Section 2.2.
7. ICH Harmonised Tripartite Guideline: The Clinical Evaluation of QT/QTc Interval Prolongation and Proarrhythmic Potential for Non-Antiarrhythmic Drugs, E14, Step 4, 12 May 2005, Page 2, Section 1.3.
8. ICH Harmonised Tripartite Guideline: The Clinical Evaluation of QT/QTc Interval Prolongation and Proarrhythmic Potential for Non-Antiarrhythmic Drugs, E14, Step 4, 12 May 2005, Page 4, Section 2.2.1.
9. Dr. L. James, MBChB, BSc, "QT Interval Prolongation: Implications for Drug Safety and Development," European Pharmaceutical Contractor (Summer 2005).
10. Source: eResearchTechnology, June 2005.
11. ICH Harmonised Tripartite Guideline: The Clinical Evaluation of QT/QTc Interval Prolongation and Proarrhythmic Potential for Non-Antiarrhythmic Drugs, E14, Step 4, 12 May 2005, Pages 3–4, Section 2.2.1.
12. ICH Harmonised Tripartite Guideline: The Clinical Evaluation of QT/QTc Interval Prolongation and Proarrhythmic Potential for Non-Antiarrhythmic Drugs, E14, Step 4, 12 May 2005, Page 5, Section 2.2.4.
13. ICH Harmonised Tripartite Guideline: The Clinical Evaluation of QT/QTc Interval Prolongation and Proarrhythmic Potential for Non-Antiarrhythmic Drugs, E14, Step 4, 12 May 2005, Page 5, Section 2.2.4.
14. R.R. Shah, "Drugs, QT Interval Prolongation and ICH E14: The Need to Get It Right," Drug Safety, 28 (2) 115–125 (2005).
15. Dr. R. Temple, FDA, January 2003, Shady Grove Meeting.
16. ICH Harmonised Tripartite Guideline: The Clinical Evaluation of QT/QTc Interval Prolongation and Proarrhythmic Potential for Non-Antiarrhythmic Drugs, E14, Step 4, 12 May 2005, Page 8, Section 3.1.
17. ICH Harmonised Tripartite Guideline: The Clinical Evaluation of QT/QTc Interval Prolongation and Proarrhythmic Potential for Non-Antiarrhythmic Drugs, E14, Step 4, 12 May 2005, Page 9, Section 3.1.2.
18. "Mortara Instrument Announces Collaboration with US Food and Drug Administration (FDA)," www.mortara.com, 8 June 2004.
19. S. Grisanti, "The Thorough Phase I ECG," Applied Clinical Trials (October 2004).
20. Source: eResearchTechnology, June 2005.
21. ICH Harmonised Tripartite Guideline: The Clinical Evaluation of QT/QTc Interval Prolongation and Proarrhythmic Potential for Non-Antiarrhythmic Drugs, E14, Step 4, 12 May 2005, Page 7, Section 2.5.1.
22. R.R. Shah, "Drugs, QT Interval Prolongation and Final ICH E14 Guidance: An Important Milestone with Challenges Ahead," Drug Safety [Manuscript submitted].
23. Source: eResearchTechnology, June 2005.
24. ICH Harmonised Tripartite Guideline: The Clinical Evaluation of QT/QTc Interval Prolongation and Proarrhythmic Potential for Non-Antiarrhythmic Drugs, E14, Step 4, 12 May 2005, Page 6, Section 2.3.