Early Accuracy in New Drug Safety Analysis

October 2, 2007

Applied Clinical Trials Supplements

Supplements-10-02-2007, Volume 0, Issue 0

A look at ECG evaluation for drug-related alterations in repolarization and Torsade de Pointes.

A major challenge faced by the pharmaceutical industry in the development of new noncardiovascular drugs is evaluating the potential risk for life-threatening ventricular arrhythmias—in particular, Torsade de Pointes (TdP). In the past two decades, several noncardiovascular drugs, after a long and expensive approval process, were found to induce TdP, resulting in their withdrawal from the market and profound negative financial consequences to sponsors.

A potentially "torsadogenic" drug may initially appear safe when administered to large numbers of patients in a trial if patients at risk, by chance or by design, are excluded from the study population. The proarrhythmic risk may only become manifest after approval, when administered to individuals who have a latent genetic risk or an acquired predisposition due to underlying heart disease, electrolyte abnormalities or concurrent medications adversely affecting the drug's metabolism. On the other hand, a potentially useful and safe drug may be inappropriately withdrawn during early drug testing because of the erroneous impression that the ECG data indicates a potential proarrhythmic risk.

The many issues concerning drug safety are extremely complex, and the costs of evaluating new drugs are staggering (over $800 million per drug). This article will consider current methodologies for evaluating the risk of new noncardiovascular drugs for TdP and other life-threatening arrhythmias. Limitations in current guidelines will be considered, and recommendations for further study will be reviewed.

TdP under the microscope

Torsade de Pointes is an unusual but eye-catching and life-threatening polymorphic ventricular tachycardia that has its origin in patients with the hereditary or acquired long QT syndrome (LQTS).1 The onset of TdP is often precipitated by one or more premature ventricular complexes occurring during the prolonged repolarization phase of the preceding beat. Typically, the arrhythmia is short lived, but it can unexpectedly degenerate into ventricular fibrillation leading to syncope and sudden death.

The characteristic QRS morphology during TdP involves recurring cycles of gradually changing QRS direction around the ECG baseline (see Figure 1). While the mechanism of the arrhythmia is not well understood, it is clear that abnormal dispersion of ventricular repolarization plays an important role in its initiation. Currently, the electrocardiogram is the basic tool for assessing TdP risk, and a prolonged QT interval is almost a necessary prerequisite for making the diagnosis.

Figure 1. Onset of Torsade de Pointes in patient with LQTS.

The prevalence of TdP in the general population is exceedingly rare, somewhere between 1/100,000 and 1/1,000,000. Because it is so uncommon, yet potentially life threatening, the FDA and other regulatory agencies in Europe and Japan have struggled to develop appropriate guidelines for analyzing the ECG data obtained during the evaluation of new drugs. The most recent consensus guidelines were approved by the International Committee on Harmonization (ICH) and published in 2005.2

The European Union, Japan, and the United States have adopted these guidelines (ICH E-14), which rely almost entirely on one ECG parameter, the QT interval, and its correction for changes in heart rate. Other reliable and reproducible ECG markers of abnormal repolarization and TdP are also being investigated, but a more definitive marker for abnormal repolarization has yet to be defined.

Based on the ICH E-14 guidelines, Morganroth discussed the "thorough ECG trial," also known as the "thorough QT/QTc trial" (TQT), for assessing the TdP risk of new drugs undergoing early Phase II testing.3 The TQT is designed to detect small but important drug-induced changes in QT intervals that would identify a drug as potentially dangerous and preclude it from further large and expensive Phase III trials.

By focusing on QT interval changes, the TQT demands a large number of digitally acquired, 12-lead ECGs to compare manually measured QT/QTc intervals from subjects on placebo with those on active drug treatment. Multiple ECGs are required at baseline and at many time periods after administering either the drug or placebo. ECGs are also obtained during a supra-therapeutic dose of the drug. Depending on the pharmacokinetics of the drug under investigation, a crossover or parallel design study protocol is chosen. Specific criteria are defined for the increase in rate-corrected QT (usually >20ms) from which to conclude that TdP risk is quite high.

While TdP has been primarily defined as the prototypic arrhythmia in patients with LQTS, it is clear that prolonged QT and QTc have not always predicted the risk of TdP. Drugs such as amiodarone, phenobarbital, and ranolazine lengthen the QT and QTc but do not usually induce TdP, and amiodarone may actually protect patients from dangerous ventricular tachyarrhythmias. It has also been reported recently that abnormally short QT/QTc can have arrhythmogenic risk of its own.4 These factors must be considered when attempting to use prolonged QT and QTc as unequivocal predictors of risk for TdP.


Although the ICH E-14 guidelines and, in particular the TQT design, are currently being required by various regulatory agencies, there are a number of conceptual and practical implications of this approach.

Current ECG equipment manufacturers are primarily driven by health care delivery needs including hospitals, clinics, and physician offices, but not by the clinical research environment, which, by comparison, is a minor component of the overall ECG equipment market. Generally, equipment manufacturers follow published ECG signal and processing standards,5 but the overall quality of clinical ECG data in health care delivery can be quite variable and not always reliable for the purpose of making careful, quantitative measurements necessary for clinical drug trials. Issues related to skin preparation, lead placement, and data quality are often overlooked or not corrected immediately, and there is no compelling need for getting it right the first time.

Acronym Reference Guide

Sophisticated signal processing methods are only as good as the signals acquired. What is needed in the drug research environment are specially designed clinical trial ECG recorders, Holter monitors, and software that address common quality problems of poor electrode position and contact, baseline drift, sample speed, and recording duration.

QT interval measurement issues

While imperfect, the QT and heart-rate corrected QT (QTc) continue to be the primary ECG focus in drug development for identifying repolarization abnormalities predictive of TdP risk. In fact, the focus on QT has been so entrenched that a new clinical entity has been defined: Drug-induced Long QT Syndrome.

There are, however, many reasons for QT interval measurements to be imprecise and inaccurate. These include difficulties in identifying the end of the T wave, choice of optimal lead(s) for measuring the QT, U-wave related issues, adjustments for heart rate changes, and intersubject or intrasubject variability. In addition, some drug study protocols require ECG lead II to be used for all QT measurements even when the end of the T wave is better seen in alternative leads, usually V3-5. QT intervals measured in a single lead may not be as accurate as those made from a global composite of all 12 leads.

The choice of the best formula for heart rate correction (e.g., Bazett, Fridericia, Framingham) is, however, very controversial, and different formulae may lead to different conclusions regarding a drug's TdP risk. Rautaharju provides compelling arguments against adjusting QT for heart rate using power functions of the RR intervals (e.g., Bazett and Fridericia's) because of nonlinear distortions of the normal QT upper limits at heart rate ranges above and below 60 bpm.6 He recommends adjusting QT using a simple linear function of RR interval. In some studies, the heart rate corrected QT is based on individualized QT interval ranges obtained at different heart rate—RR interval bins prior to administering the study drug.

All the different methods proposed for QT/RR adjustments assume that there is a common physiologic relationship between QT and heart rate that is similar for all individuals. This is clearly not the case.

Ventricular repolarization

While measures of QT and QTc prolongation have been utilized as surrogate indices of disordered repolarization, numerous studies have shown that this is an overly simplistic assumption. Even ignoring the measurement problems stated above, the QT interval is but one of many possible parameters related to ventricular repolarization, and prolonged QT intervals don't necessarily always reflect disorganized repolarization.

Evidence for electrical heterogeneity in ventricular myocardium has been published by Antzelevitch and colleagues.7 Using arterially perfused animal wedge preparations of ventricular myocardium, these investigators have shown that transmural dispersion of repolarization (TDR) results from differences in action potential duration (APD) in three electrophysiologically distinct cell types found in the epicardium, the endocardium, and the "M" cells of the middle layers. In particular, the M cells have the longest APD and the epicardial cells have the shortest APD. Furthermore, the M cells are more likely to be affected by QT-prolonging drugs, as well as by slow heart rates and other proarrhythmic perturbations.

In the wedge preparation, voltage gradients between epicardium and M cells define the upslope of the T wave; the T wave peak reflects the end of epicardial cell repolarization; voltage gradients between M cells and endocardial cells define the downslope of the T wave; the end of the T wave occurs when M cell repolarization is complete. The interval between T peak and T end (TpTe) is a measure of TDR in wedge preparation.

Both animal and clinical studies have concluded that TdP and other ventricular tachyarrhythmias are initiated by local potential gradients within the three cell layers leading to triggered activity from early after-depolarization or re-entrant excitation. Several studies have recently suggested that prolonged TpTe corrected for heart rate might be a better marker for TdP risk than QTc,8,9 but this measurement has not yet been implemented in the guidelines for studying new drugs. It is also unlikely that TpTe measured on the 12-lead ECG reflects the same information about TDR as found in the simpler animal wedge preparation.

Other ECG measures of disordered repolarization have been investigated with conflicting results. One of these, QT dispersion, is obtained by subtracting the QT interval from the ECG lead with the shortest QT from the longest QT interval in the 12-lead ECG. Although this parameter has no physiologic basis, it has been studied intensely, and has finally proven to be of minimal value in the evaluation of TdP risk. Rautaharju has been one of the strongest critics of QT dispersion and has essentially signed its death certificate.10

T wave alternans (TWA) is a promising technique that has been used in patients with heart disease to stratify their risks for ventricular tachyarrhythmias and sudden death.11 TWA is a measure of microvolt variations in beat-to-beat changes in T wave amplitude using digital signal processing of the 12-lead ECG at rest or during submaximal exercise. Studies have shown that a "negative" TWA test in these patients is associated with a very low risk of life-threatening arrhythmias. Although TWA is useful in selected patient populations, it has not been carefully studied in the setting of new drug evaluations, and it is unlikely to be abnormal in the normal population.

Finally, detailed analysis of T wave morphology and complexity has been considered potentially useful for studying disorganized repolarization. Specific T wave phenotypes have been reported for the three common types of hereditary LQTS,12 and similar T wave morphologies are also described in the acquired LQTS.8 In both conditions, prolonged QTc and TpTe intervals are associated with unusual T wave morphologies and increased TdP risk. It is not certain how these various T wave parameters can be used to improve the sensitivity and specificity for TdP risk in the study of new drugs.


The search for valid ECG parameters to assess the risk of TdP and other life-threatening arrhythmias is ongoing. QT and QTc continue to be the most studied ECG parameters in the evaluation of new drug safety, and current guidelines attempt to standardize the process of analyzing and reporting the ECG data.

Improvements in ECG equipment, computer analyses, patient preparation, and standardized protocols are needed to reduce the enormous costs related to drug development. The unwarranted and premature demise of potentially useful new drugs because of false positive conclusions of adverse repolarization changes continues to be a problem that needs attention.

Frank G. Yanowitz, MD, is professor of medicine at the University of Utah School of Medicine, medical director, Intermountain Healthcare ECG services, at LDS Hospital in Salt Lake City, UT, and a member of the VIASYS Clinical Services ECG CorLab Board of Expert Over Readers, email: fyanow@mac.com


1. N. El-Sherif and G. Turitto. "Torsade de Pointes," Current Opinions in Cardiology, 18, 6–13 (2003).

2. ICH E14 Guidelines, "The Clinical Evaluation of QT/QTc Interval Prolongation and Proarrhythmic Potential for Non-Antiarrhythmic Drugs (online)," http://www.ich.org//cache/compo/276-254-1.html.

3. J. Morganroth, "Cardiac Repolarization and the Safety of New Drugs Defined by Electrocardiography," Clinical Pharmacology & Therapeutics, 81, 108–113 (2007).

4. E. Schulze-Bahr and G. Breithardt, "Short QT Interval and Short QT Syndromes," Journal of Cardiovascular Electrophysiology, 16, 397–398 (2005).

5. P. Kligfield, L.S. Gettes, J.J. Bailey, R. Childers et al., "Recommendations for the Standardization and Interpretation of the Electrocardiogram. Part I: The Electrocardiogram and Its Technology," Circulation, 115, 1306–1324 (2007).

6. P.M. Rautaharju and Z. Zhang, "Linearly Scaled, Rate-Invariant Normal Limits for QT Interval: Eight Decades of Incorrect Application of Power Functions," Journal of Cardiovascular Electrophysiology, 13, 1211–1218 (2002).

7. C. Antzelevitch, "Cellular Basis for the Repolarization Waves of the ECG," Annals of the New York Academy of Sciences, 1080, 268–281 (2006).

8. I. Topilski, O. Rogowski, R. Rosso, D. Justo et al., "The Morphology of the QT Interval Predicts Torsade de Pointes During Acquired Bradyarrhythmias," Journal of the American College of Cardiology, 49, 320–328 (2007).

9. M. Yamaguchi, M. Shimizu, H. Ino, H. Terai et al., "T wave peak-to-end and QT Dispersion in Acquired Long QT Syndrome: A New Index for Arrhythmogenicity," Clinical Science, 105, 671–676 (2003).

10. P.M. Rautaharju, "A Farewell to QT Dispersion. Are the Alternatives Any Better?" Journal of Electrocardiology, 38, 7–9 (2005).

11. A.F. Osman and M.R. Gold, "T Wave Alternans for Ventricular Arrhythmia Risk Stratification," Current Opinions in Cardiology, 18, 1–5 (2002).

12. L. Zhang, K.W. Timothy, G.M. Vincent, M.H. Lehmann et al., "Spectrum of ST-T Wave Patterns and Repolarization Parameters in Congenital Long-QT Syndrome. ECG Findings Identify Genotypes," Circulation, 102, 2849–2855 (2000).

Related Content: