FQA3

A small but strangely heated literature addresses this question. Here, it's torsade.

Torsade de pointes (from the French for “twisting of the points”) is a re-entrant arrhythmia with a distinctive electrocardiographic appearance, sometimes described as a sine wave within a sine wave. This example is from Braunwald E, Zipes DP, and Libby P (eds.), Heart Disease [6^th edition] (Philadelphia: W.B. Saunders, 2001), page 868:

The pattern is called "twisting" because when the peaks are at their smallest in one lead, they are generally at their largest in another, as if a vibrating string were having its plane of vibration slowly rotated about the long axis of the string.

Many episodes of torsade de pointes are asymptomatic and stop by themselves, but the mechanical activity associated with this electrical pattern is generally dysfunctional, and torsade can degenerate into other arrhythmias that are both dysfunctional and irreversible.

Torsade de pointes is seen in various congenital defects known collectively as the long-QT syndromes. In addition, torsade sometimes happens to otherwise-normal people as an adverse effect of drugs. Other things being equal, the risk of torsade is increased in women, in people who are hypokalemic or hypocalcemic, and in patients with congestive heart failure.

8. Congenital Long-QT Syndromes

Mutations in the genes whose products make up the major ion channels give rise to syndromes known as the long-QT syndromes (LQT syndromes). These syndromes are congenital, but not necessarily hereditary, since many of the cases turn out to be new mutations.

The most frequent of the long-QT syndromes are known as LQT1 (impaired opening of the (repolarizing) I_Ks channel), LQT2 (impaired opening of the (repolarizing) I_Kr channel), and LQT3 (impaired closing of the (depolarizing) I_Na channel).

A few thousand patients have been identified worldwide, but they are thought to be much more common.

About half of the affected people sustain major arrhythmias before the age of 40. In addition, an unknown but nonzero fraction of the deaths attributed to the catchall “sudden infant death syndrome” (SIDS) are caused by LQT-syndrome-related arrhythmias.

8.5. Do people with the same hereditary long-QT syndrome all have the same mutation?

No. In fact, a new kindred with one of the established syndromes is likely to have a family-specific mutation. As of June 2002, cases of the LQT1, LQT2, and LQT3 syndromes had been found as the results of about 300 different mutations.

8.6. Do people with the same ion-channel mutation all have equally prolonged QT intervals?

No. When a family carrying a LQTS mutation is investigated, usually about a third of the family members with the mutation have normal electrocardiograms, and the others have varying degrees of QT prolongation.

8.7. Do people with the same ion-channel mutation all have an equal likelihood of arrhythmic events?

No. Those with longer QT intervals have a higher incidence of events, but even those with normal electrocardiograms have an incidence of events that is higher than that of people who do not have LQTS.

They are at high risk of malignant tachyarrhythmias, similar to those seen in people with congenitally long intervals. For the shortest possible demonstration that risk does not increase monotonically with QT interval, look to the short-QT syndrome.

9. Drug-Induced Changes in Ion-Channel Function

9.1. How is a drug's propensity to cause changes in ion-channel function tested?

Drug-induced interference with channel opening is described in terms of the concentration of drug needed to impair ion flux by 50%. This measurement is denoted by IC_50.For example, some extreme values for blockade of the I_Kr channel are that of astemizole, a very potent blocker with an IC₅₀of 0.9 nM, and that of ofloxacin, an extraordinarily weak blocker with an IC₅₀of 1.4 mM.

Yes. Because arrhythmias are usually the result of ion-channel dysfunction, a drug with no discernible effect on ion channels would be unlikely to be developed as an anti-arrhythmic.

9.4. Are some ion channels more susceptible to drug-induced interference than others?

Yes. The I_Kr channel turns out to have a wide, funnel-shaped vestibule, so many different kinds of small molecules can approach the channel and partially block it. Other channels are generally more selective. Even so, most drugs known to interfere with one channel are known to interfere – at least to some degree – with others.

9.6. Do ion-channel studies provide close analogies to distortions of human electrocardiograms?

9.7. Do ion-channel studies provide close analogies to human torsade de pointes?

9.8. Has in vitro ion-channel testing been validated as a means of predicting drug-induced torsade de pointes in humans?

Yes and no. A drug could not induce torsade de pointes if it did not in some way interfere with the function of ion channels, but the converse is not true. For example, verapamil is a moderately potent I_Kr blocker (its IC₅₀ is about 0.14 µM), but it never induces torsade or any other tachyarrhythmia.

10. Drug-Induced Changes in the Action Potential

Changes in action-potential duration are usually quantified by measuring the time required for the repolarization process to reach 50% (APD₅₀) or 90% (APD₉₀) completion.

10.3. Do action-potential studies provide close analogies to distortions of human electrocardiograms?

No, except insofar as a drug that never prolongs the action potential could not prolong the QT interval.

10.4. Do action-potential studies provide close analogies to human torsade de pointes?

10.5. Does prolongation of action potentials in vitro reliably predict drug-induced torsade de pointes in humans?

No. Most drugs known to induce torsade are known to prolong the action potential in at least some types of cells. Other drugs, however (e.g., pentobarbital), by prolonging the action potential in some cells more than others are known to reduce the risk of torsade.

11. The Wedge Preparation

Ion-channel and action-potential studies are generally performed in single cells (or in small multicellular specimens chosen for their homogeneity), maintained by immersion in an appropriate bath. The specimen taken for wedge-preparation studies is a full-thickness slab of the canine ventricular wall, several cubic centimeters in total volume, maintained by direct perfusion of its native coronary vessels.

The voltage measured between the endocardial and epicardial surfaces of the slab integrates the action potentials of all of the cells of the slab. It resembles a surface electrocardiogram not only in its means of generation, but also in its overall appearance.

Electrodes floating on the raw intra-myocardial surfaces of the slab record action potentials typical of the local cells.

When drugs are added to the perfusate, the various electrodes provide simultaneous views of different cell types' action potentials and of the integrated quasi-electrocardiogram.

The quasi-electrocardiogram and action-potential recordings are described as ECGs and APs are usually described, with the pivotal added note that they were simultaneously recorded.

11.3. Do wedge-preparation studies provide close analogies to distortions of human electrocardiograms?

11.4. Do wedge-preparation studies provide close analogies to human torsade de pointes?

11.5. Have wedge-preparation studies been validated as a means of predicting drug-induced torsade de pointes in humans?

I'm not sure what "validated" means here. In such experiments as have been done, wedge-preparation studies to evaluate transmural dispersion of repolarization have provided the best available predictions of human proarrhythmia. Compared to QT prolongation as a predictor, the results aren't even close:

	effect in canine LV (wedge preparation)		effect in humans
drug	EADs	ÝTDR	ÝQT	*torsade*
amiodarone	–	–	+	±
azimilide	+	+	+	+
cisapride	+	+	+	+
erythromycin	+	+	+	+
ßI_Ks* with βstim*	–	+	+	+
ßI_Ks* with βblock*	–	–	+	–
pentobarbital	–	–	+	–
quinidine, low conc	+	+	+	+
quinidine, high conc	–	–	+	–
sotalol	+	+	+	+
terfenadine	+	+	+	+
verapamil	–	–	±	–
* Epi = epicardial cells; ßI_Ks = LQT1 (I_Ks defect) in humans, drug-induced I_Ks blockade in dogs; βblock= beta-adrenergic blockade; βstim= beta-adrenergic stimulation; M = M cells; PF = Purkinje fibers

12. Drug-Induced QT Prolongation in Animals

12.3. Do QT prolongation and other ECG abnormalities in animals closely resemble these phenomena in humans?

12.4. Have studies of drug-induced QT prolongation in animals been validated as a means of predicting drug-induced torsade de pointes in humans?

13. Drug-Induced Torsade de Pointes in Animals

13.1. How is a drug's propensity to cause torsade de pointes in animals tested?

13.3. Have studies of drug-induced torsade de pointes in animals been validated as a means of predicting drug-induced torsade de pointes in humans?

14. Drug-Induced QT Prolongation in Humans

Lists of drugs that prolong the QT interval are widely available on the Web and elsewhere, but most of the data underlying these tables were obtained using Bazett-corrected electrocardiograms, so many drugs are probably listed in error. Conversely, many older drugs were never tested for this property, so they may well have greater effects than some of the drugs listed.

14.3. If a given dose of a drug prolongs the QT interval by a certain amount, will a higher dose of the drug prolong the QT interval by the same amount or greater?

Not necessarily. If the higher dose alters the function of depolarizing channels, in addition to the repolarizing channels affected at the lower dose, then the two effects may balance each other out at the higher dose. This fanciful-sounding possibility is actually observed with some drugs, notably quinidine.

No. QT prolongation is evidence that the repolarization of some cardiac cells has been delayed, and this delay can lead to torsade if adjacent cells are repolarizing much sooner. On the other hand, the AP prolongation that is evident as QT prolongation may – as is seen with amiodarone, diltiazem, pentobarbital, ranolazine, and verapamil – be concurrent with AP prolongation in adjacent cells, so that local heterogeneity of repolarization has actually been reduced, together with the likelihood of torsade de pointes.

15. Drug-Induced Torsade de Pointes

Lists of suspect drugs are widely available on the Web and elsewhere, but these lists should not be accepted uncritically. Some of the data come from FDA case reports of patients taking multiple drugs, sometimes with underlying conditions that could by themselves have been responsible for the arrhythmias reported.

15.2. Do all of the true culprit drugs interfere with the function of ion channels?

Even though only one (LQT2) of the three most common long-QT syndromes involves I_Kr, and even though various non-therapeutic drugs can reliably induce torsade de pointes in animals via interference with other channels, drug-induced torsade de pointes in humans has almost always been attributable to interference with I_Kr. In some cases, this attribution is difficult to understand, and it may reflect lack of data with respect to other channels. For example, the capacity of grepafloxacin to cause torsade is attributed to its ability to block I_Kr, even though the IC₅₀ of grepafloxacin for this blockade is about 50 µM. Grepafloxacin's activity at other channels has not been publicly reported.

15.4. How often will a patient with drug-induced torsade be found to have been unusually susceptible because of a subclinical congenital channel dysfunction?

Not very often. A few patients have sustained torsade after what should have been low-risk drug exposure, and have later been found to have subtle ion-channel dysfunctions, but this pattern seems to be present in only 3-5% of the incident cases of drug-induced torsade.

15.5. Do all drugs that block I_Kr cause torsade de pointes if given in sufficient dose?

No. For example, diltiazem and (especially) verapamil block I_Kr, but they do not cause torsade.

15.6. Do all of the drugs known to cause torsade de pointes prolong the QT interval?

15.7. Conversely, will any drug that prolongs the QT interval cause torsade de pointes if it is given in sufficient dose?

No. Diltiazem and verapamil are again counterexamples. At least in models in which animals are pretreated so as to make them especially susceptible to torsade, these drugs (and others) prolong the QT interval but actually reduce the likelihood of torsade.

15.8. Are there drugs that cause torsade de pointes even though they prolong the QT interval only slightly?

Inasmuch as some patients with long-QT syndromes are at heightened risk of torsade even though their QT intervals are only slightly prolonged or even normal, there probably are drugs that can induce torsade while lengthening the QT interval only minimally (or not at all).

On the other hand, no such drugs are known. Certainly there are drugs that can induce torsade under some conditions but that under other conditions cause only minimal changes to the QT interval, but this is a non sequitur. For example, when a normal dose of terfenadine is given to a metabolically-competent person, the average amount of QT prolongation is said to be only 6 ms. In the circumstances in which terfenadine is known to have induced torsade, however (higher doses, metabolic inhibition, etc.), the amount of QT prolongation is much greater.

16. Drug Development and Regulation

16.1. How important are metabolic effects during human trials of drugs that might affect repolarization?

Metabolism is no more important in thinking about torsade than it is in thinking about other adverse drug effects that might be clinically concentration-related. That is to say, it is very important indeed.

All drug effects could be said to be concentration-related, inasmuch as they are all related to molecular interactions that follow the rules of mass action. In the clinical context, an effect is said to be concentration-related only if it runs through its range of incidence or severity within a range of concentrations that is of clinical interest. For example, ACE-inhibitor-induced cough is said not to be concentration-related, because its incidence no longer increases once drug concentrations have reached levels well below those that are clinically useful. Morphine-induced sedation, on the other hand, is increasing throughout (and beyond) the range of therapeutic concentrations, so this effect is said to be concentration-related.

Impaired drug metabolism can result in concentrations greatly exceeding those attained by timid dose-ranging, especially when a drug is dependent upon a unique metabolic pathway. For example, concentrations of simvastatin are raised by more than an order of magnitude when that drug is co-administered with ketoconazole. Returning to the repolarization arena, while such populations as women and heart-failure patients tend to have more pronounced responses to repolarization-altering drugs, these differences are small and predictable, almost negligible when compared to the differences that are sometimes produced by metabolic interference.

Another set of considerations under the metabolic heading comes from the fact that the molecule directly responsible for an adverse drug effect may be different from the molecule administered. Although most metabolites are more polar than their parent compounds, and therefore less active, some exceptions are important. For example, the CNS toxicity of meperidine is caused by normeperidine, a long-lived metabolite. This toxicity could not be detected in short-term trials.

As regards repolarization changes (or other adverse drug reactions), the goal of Phase 1 trials is to see how much harm the drug can be made to do. Dosing (and metabolic inhibition, if relevant) should be increased until it's not feasible to increase any more. This is the time to discover unpleasant facts that will be much more unpleasant if they are not discovered until later phases of development or – even worse – until after marketing.

Because the human metabolism may not be completely understood until later, some early Phase 1 trials may need to be repeated with revised protocols. For example, if some of a drug's toxicity is attributable to a slowly-accumulating metabolite, then it may be necessary to do new, longer trials, or to do new trials in which the metabolite is administered directly.

Repolarization changes are special in two ways. First, the associated risk (death or infarction as the result of malignant tachyarrhythmias) is one from which Phase 1 subjects can be totally protected. Similar arrhythmias, after all, are deliberately induced in electrophysiology laboratories as a matter of routine. When an appropriately monitored Phase 1 subject has a suspicious electrocardiogram, the investigator might reasonably be more willing to continue than he would be if the subject had had an equally-suspicious panel of hepatocellular enzymes.

Second, a subject with an unusual repolarization response to a new drug is reasonably likely to have similar responses to others, and electrophysiological evaluation of the subject is a negligible-risk option that is therefore in the interest of both the drug developer and the subject. No similar statements can be made when a subject's response to a studied drug suggests hematologic or hepatic toxicity.

16.3. How important is the non-occurrence of torsade de pointes during human drug trials?

Because torsade is so rare (even in LQT patients at high risk, and even in patients taking the drugs known to cause torsade most frequently), the non-occurrence of torsade in clinical trials provides little reassurance.

About 10 years. All that is new is that there have been personnel changes at FDA.

16.6 How does my company get reimbursed for the millions of dollars wasted on Thorough QT Studies?

Frequently Questioned Answers (3)

7. Torsade de Pointes