Robert R. Fenichel


 Drugs that Alter Ventricular Repolarization

Some antiarrhythmic drugs achieve their desired effects by altering ventricular polarization, but with most drugs any effect on repolarization is an incidental side effect.  Prolongation of repolarization (usually manifested as QT prolongation on the surface electrocardiogram) is sometimes benign or even antiarrhythmic (as with amiodarone, diltiazem, pentobarbital, ranolazine, and verapamil), but sometimes it is associated with arrhythmias, syncope, and death.

The normal quasiperiodic electrical activity of the heart is the result of the flow of ions through channels in the membranes of myocardial cells.  Drugs affect ventricular repolarization by interfering with the opening and closing of these channels.  The accompanying pages contain a table of various drugs and the concentrations at which they have been found to block some of these channels.  Where available, the table also includes data as to the concentrations at which drugs bind to other receptors (e.g., histamine receptors) of interest.  The table began as a joint project with John Koerner at FDA, but we were unsuccessful in obtaining permission to post these public-domain data on the FDA Web site.  Additional data are added as they are brought to my attention. 

For several years, I had  here also posted a paper that arose out of my collaboration with John Koerner.  That paper was published in 2002 in the American Journal of Therapeutics (9(2): 127-139), but except for correcting some reference details it had not been revised since mid-2000, and the long delay since its composition had left it more than a little dated.  That paper has now been removed from the site.  It has been entirely superseded by a paper in the April 2004 issue of Journal of Cardiovascular Electrophysiology (15(4): 475-495).   I posted some recommendations for electrophysiological workup early in 2002, but I never kept them up to date, so now they, too, have been removed.  News in this area will be posted on this page from time to time, with no commitment to thoroughness.  A page of Frequently-Asked Questions goes over some of the same ground as the recommendations did, and (because the FAQ genre imposes no requirements of completeness or even consistency) it seemed more likely to be kept up to date, but now it too is falling behind.  The paper is the best source.

In late April 2002, I participated in Phase5 Sciences' conference on Changes in Ventricular Repolarization: Implications for Drug Development.  This was an exciting meeting; the other speakers and panelists were Charles Antzelevitch, Arthur Brown, Timothy Callahan,  Louis Cantilena, William Crumb, Marek Malik, Jean-Loup Parier, and Silvia Priori.  To give a flavor of the discussion, I here summarize only one of the sessions.

To detect and characterize a drug-induced change in ventricular repolarization, several participants utilize 24-hour electrocardiograms, at least one at each subject's baseline and at least one during the subject's exposure to the suspect drug.  Each such electrocardiogram captures about 105 beats, and different subsets of the beats are used in different participants' analyses.  To facilitate description of the different approaches, I use a taxonomy of ECG complexes that is in some ways a logical union of the taxonomies used by the speakers: 

  • Some beats (Group A) are so distorted by electrical artifact that they are not amenable to measurement.

  • Consider the beats not in Group A. After a change in heart rate, the QT interval is commonly observed to take about 2 minutes to achieve a new steady state.  For this reason, one should for each beat examine the range of heart-rate variation observed during the previous few minutes.  Beats preceded by more-than-minimal heart-rate variation are segregated into Group B.

  • Consider the beats not in Group A or Group B.  Even though each such beat was preceded by a period of stable heart rate, some beats might have been preceded by a few minutes of more-than-minimal QT variation.  Such beats constitute Group C, while the remaining beats constitute Group D.

To be able to detect drug-induced changes in repolarization, Malik first characterizes each subject's personal QT-RR relationship, using the beats of Groups C and D in a recording made at baseline.  Reasoning by analogy from in vitro practice, Brown suggested that a plausibly more stable estimate might be derived if Malik's analysis were restricted to beats from Group D.

Rather than attempt to describe a subject's QT-RR relationship in isolation, Parier and Callahan look at the frequency distributions of rate-corrected QT duration , comparing the baseline distribution of beats in Groups B-C-D to the distribution seen when beats from the same groups are collected during drug treatment. 

Antzelevitch noted that the compounds that in wedge preparations induce harmless prolongation of repolarization can be distinguished from those that induce arrhythmias by experiments in which the basic cycle length is abruptly changed.  During studies of a proarrhythmic compound, sometimes the preparation exhibits unremarkable behavior at either of the steady-state cycle lengths, but early afterdepolarizations are seen for a minute or two after the change in rate.   Returning to the 24-hour human electrocardiograms, some participants speculated on the basis of Antzelevitch's observations that although beats of Group B might be uninterpretable in attempts to define the QT-RR relationship, these beats might turn out to hold the information distinguishing harmless QT prolongation from proarrhythmia. 

In May 2003, the annual meeting of the North American Society for Pacing and Electrophysiology included a special session on QT-Prolonging drugs, organized by Antzelevitch.  The primary speakers (Antzelevitch, Malik, Priori, Dan Roden, Michael Rosen, Jeremy Ruskin, and Michael Sanguinetti) continue to improve their slides and examples, but their main messages will be familiar to those who have followed this area in recent years:

  • The chemical requirements for potent IKr blockade are increasingly well understood, as is the chemical basis for the HERG channel's susceptibility to mild blockade by all manner of small molecules;

  • Different tissues vary in their ion-channel mix, so the question "What does <drug> do to the action potential?" is not a well-formed question;

  • Some ion channels important in human electrophysiology are minimally present in some animal tissues;

  • Torsade emerges from transmural dispersion of repolarization (TDR); 

  • The T wave also reflects TDR, but in a slightly different way;  

  • Measurements of TDR in canine tissues are more predictive of human torsade than QT measurements in intact humans;

  • The most predictive animal studies other than TDR measurements seem to be proarrhythmia studies in variously stressed dogs and rabbits;

  • Animal QT measurements are of uncertain value;

  • Human QT measurements are confounded by population-independent "corrections" like those of Bazett and Fridericia; 

  • Human QT measurements are poor predictors of torsade; and

  • Human torsade is vanishingly rare in people with QT intervals < 500 ms.

  The last segment of the meeting was given over to presentations of preclinical and clinical guidelines now making their way through the International Harmonization process.   Rosen described these presentations as capturing the flavor of the science of 5 or 10 years ago; other participants expressed similar sentiments.  The guidelines now moving toward the status of supranational law include suggestions

  • to answer the question "What does <drug> do to the action potential?" by studying the effect of the drug on guinea-pig Purkinje fibers, a tissue in which IKs is essentially inactive; and

  • to direct the greater part of the preclinical electrophysiological workup toward animal QT-prolongation studies; and

  • that preclinical findings of QT prolongation without increase of TDR were interesting, but would always need to be confirmed in human studies; but

  • that they could not be confirmed in feasible human studies;

  • that QT intervals measured in human studies should always be accompanied by Bazett-corrected figures; and

  • that changes larger than 5-10 ms in intensive early studies would lead to stringent requirements for later studies.

  The FDA has committed itself to silly guidelines before, and it has managed to reverse itself without too much difficulty.  In May of 1988, for example, the FDA released a draft guideline (subsequently neither withdrawn nor promoted from draft status) stating that

based on its presumption that an antihypertensive drug should retain most of its peak effects at trough, the Division [of Cardio-Renal Drug Products] has determined that the drug effect at trough (measured as the difference from the placebo effect) should be no less than one-half to two-thirds of the peak effect. . . .

Six months later, the Cardio-Renal Advisory Committee was asked to apply this abstractly-conceived rule to the concrete case of nicardipine, whose trough/peak ratio was less than 30%.  Because excessive effect was not observed at peak, the Committee decided to ignore the guideline (the story is told slightly more completely in a paper I wrote with Ray Lipicky).  No portion of the guideline was ever formally rescinded, so later sponsors collected large quantities of irrelevant data, but no application was ever rejected on the basis of the trough/peak guideline. 

One might hope that the FDA will behave similarly here, but the new consideration is the fact that the repolarization guidelines are moving rapidly toward international ratification.  Once the guidelines are officially laminated into plastic, FDA may find it difficult to modernize its criteria without going through the ponderous ratification process all over again.  The times are perilous. 

On the other hand, one can derive some comfort from developments before and during the meeting of the FDA's Cardio-Renal Advisory Committee on 29 May 2003.  In its review of (at least) the alfuzosin application [disclosure: I consulted for Sanofi], and in its briefing book for the Committee, the reviewing division made no reference whatsoever to the not-quite-guidelines that other parts of FDA are pushing toward ICH.  The ICH-bound suggestions might not have been mentioned at all, if they had not been the organizing principle behind Bayer's slick but coolly-received presentation of the vardenafil application.  The Committee voted to encourage the continued presentation of Bazett and Fridericia computations, but they were much more enthusiastic about the promise of methods that used each subject as his own control , as in the bin methods used by Sanofi-Synthélabo in the alfuzosin application.  With one modern example similar to atropine, I would expect this Committee to give up the Bazett formula forever.

The QT effects of alfuzosin were about 1 ms at therapeutic doses, and about 4 ms at maximal achievable exposure; the Committee's unanimous opinion (and mine, for what it's worth) was that these changes are of no clinical significance, presumably meaning that they should not affect clinicians' behavior.  The QT effects of the other a blockers have never been reported, but Sanofi's studies showed that the respective potencies as IKr blockers were

drug IKr IC50 (µM)
doxazosin 2.5
prazosin 3.4
terazosin 21.4
alfuzosin 83.3
tamsulosin 104.8

These results mean that rational labeling for alfuzosin will be nontrivial to construct.  Almost any mention of QT prolongation frightens the ignorant, so almost any mention of the QT effects of alfuzosin in its label will have the effect of steering clinicians away from alfuzosin, toward the alternative a blockers.  This result would be counter to the Committee's implicit recommendation.

  Worse yet, such a result would be logically incoherent.  Someone who believes that the QT effects of alfuzosin might be hazardous must without actual QT data on the other a blockers expect the various QT effects to be proportional to the drugs' potency at IKr blockade.  That is, the QT effects probably line up as doxazosin > prazosin > terazosin > alfuzosin > tamsulosin.  If the QT effects of alfuzosin are worth worrying about, then one must logically worry even more about the unknown but probably greater effects of doxazosin, prazosin, and terazosin.  

  Still, incoherence hasn't stopped them before.

  The previous few paragraphs were added just before alfuzosin was actually approved, and now (22 June 2003) it turns out that the result was the one I most feared.  I haven't seen alfuzosin's final approved U.S. labeling, but Sanofi's press release of 16 June 2003 gives a pretty clear idea of what the label will be like. 

   As of mid-2004, the would-be drug developer has a bizarre range of potential standards to consider.  The spectrum includes the positions of

  • Most academics in this area, who agree that transmyocardial dispersion of repolarization, measured in vitro, is the best predictor of human torsade de pointes.  They regard the surface electrocardiogram as a means of exhibiting, perhaps ambiguously, the underlying physiology.

  • Some academics, who retain greater or lesser interest in the electrocardiogram, if only because the in vitro methods are of marginal relevance to clinical practice.  Because of the intra-individual variation of the RR-QT relationship, they favor various means of using each patient as his or her own control.

  • The FDA (e.g., the 2003 Advisory Committee), which is still focused on the electrocardiogram, but dissatisfied with it, in part because of the cases (amiodarone, pentobarbital, ranolazine, etc.) in which QT measurement appears to be intrinsically misleading, and in part because of the difficulty of assigning risk, even when risk is undeniably present, to any moderate level of QT prolongation.  It is receptive to individualized QT assessment and uneasy about population-independent methods (Bazett, etc.), but it is not quite ready to abandon the latter methods altogether. 

  • The FDA (e.g., those pushing certain documents toward ICH endorsement), which does not acknowledge the possibility of substantial QT prolongation without risk of torsade.  Also, it seems to believe that it would be unfair to discriminate against traditional QT-assessment methods just because they have been invalidated.  On the other hand, this sect does recognize the possibility that some degrees of QT prolongation must be considered inconsequential.

  • The FDA (e.g., those making recent regulatory decisions, not only about alfuzosin; the dark side of the Agency, using "dark" in the sense of "benighted"), which remains fond of Bazett correction, and unfamiliar with the notion of de minimis physiological changes. 

  The foregoing portions of this page have not been changed since late 2004.  As of mid-2014, recent messages from the FDA have been mixed.  On the one hand, a late-2013 presentation by Norman Stockbridge suggested that the FDA's bond to QT measurement was finally fraying.  On the other hand, in June of 2014 FDA invited me to join an (unpaid) extramural task force* whose first task was to use AERS data to "rank a series of compounds** with respect to their clinically manifested ability to cause torsade de pointes."  I opted out, because the exercise seemed to resemble a sportswriters' ranking of football teams.  I did provide the task force with a copy of the slides I used for a presentation in May of 2003 [sic], covering much of the same ground, and even including a ranking of some of the same compounds.

  Moreover, the suggested dependence on AERS data may be revealing.  FDA admits that AERS data "cannot be used to calculate the incidence of an adverse event or medication error in the U.S. population," and this disclaimer is accepted by most people other than tort lawyers.  AERS still has a constituency within FDA, however, and I now believe that the task force has been mustered by Stockbridge and others as a means to strengthen their position against retrogressive factions within the Agency.


  * The other invited members were John Camm, Borje Darpo, Peter Kowey, Pierre Maisonblanche, Eric Michelson, Gerald Naccarelli, Ignacio Ropdriguez, Michael Rosen, Paul Volders, and Ray Woosley.

  ** The compounds were astemizole, azimilide, bepridil, chlorpromazine, cisapride, clarithromycin, clozapine, diltiazem, dofetilide, domperidone, droperidol, flecainide, ibutilide, loratidine, methadone, metoprolol, mexiletine, nifedipine, nitrendipine, ondansetron, pimozide, quinidine, ranolazine, risperidone, tamoxifen, vandetanib, and verapamil.


Page revised: 08/31/2014 14:36