Abstract
This is a brief review of properties of cardiovascular function that should be considered for interrogation in studies of toxicology and/or safety pharmacology for non-cardiologists and non-physiologists. Since concern over the rarely occurring, unusual, and drug-induced tachycardia, Torsade de pointes, is a leading cause for cessation of development of potential drugs and for removal of drugs from the market, therefore, the toxic manifestation of drugs will be emphasized. The putative origin of torsade de pointes, and the origin of the electrocardiogram and electrocardiographic features of ventricular arrhythmias will be discussed.
Introduction
A search to identify potential cardiovascular toxicity of drugs or potential drugs has been a part of toxicology since its inception, however relatively recently a new manifestation of cardiovascular toxicity has emerged—Torsade de pointes. In 1966, Dessertenne (1966) reported on a woman with complete heart block who developed a rapid ventricular tachycardia. Configurations of her QRS complexes during the tachycardia varied between being predominantly positive to predominantly negative (Figure 1). He termed the arrhythmia torsade de pointes (TdP) because the changing configuration of her QRS complexes reminded him of a ballet dancer on her toes twisting (torsade) around a point (pointes). Although drug-induced arrhythmia was reported initially in 1918 (Frey, 1918), in 1993 sudden deaths were reported in persons taking high doses of the non-sedating antihistamine (terfenadine) for urticaria. The deaths were attributed to ventricular fibrillation evolving from TdP (Honig et al., 1993). Since the persons were taking this drug for a rather trivial non-cardiac indication, federal regulatory agencies, the pharmaceutical industry, and academia became very interested in the prevalence of drug-induced TdP, its potential mechanisms, and how to predict which drugs might be torsadogenic (Salle et al., 1985; EMEA, 1997; FDA, 2000; Woosley, 2000; Temple and Himmel, 2002; Redfern et al., 2003; Roden, 2004; Shah, 2004). Torsade de pointes appears to develop only in the presence of other risk factors (Table 1), some of which are modifiable* and others of which cannot be modified#.
Ventricular fibrillation leading to sudden death appears to develop in fewer than 20% of patients with TdP (Sanguinetti et al., 1995). Subsequently, it was discovered that retardation of ventricular repolarization—indicated by lengthening of the QT of the electrocardiogram (ECG)—was caused principally by a reduced rate of exit of cytosolic potassium ions through a specific sarcolemmel ion channel (the hERG channel) (Sanguinetti et al., 1995; Yang and Roden, 1996; Liu et al., 1998; Cheng et al., 1999; Watanabe et al., 1999; Chachin and Kurachi, 2002; Fossa et al., 2004). This channel is inhibited by drugs used to treat heart disease, but more importantly by drugs indicated for many noncardiac disorders (e.g., antihistamines, antibiotics, prokinetics, antipsychotics). In particular, it occurred in patients who manifested genetic peculiarities (polymorphisms) in those channels (Jervell and Lange-Nielsen, 1957; Napolitano et al., 2000; Moss, 2002; Priori et al., 2003, 2005). Persons may develop TdP without therapeutic provocation if they possess these polymorphisms. (This phenomenon was highly instrumental in introducing the discipline of pharmacogenomics, which tailors drugs for persons with certain polymorphisms, and for producing the sub-specialty of safety pharmacology, designed to test for toxic manifestations of potential therapeutic agents existing at or above the estimated therapeutic concentrations.)
Identifying a toxic potential is a daunting task: important, difficult/“tricky,” time-consuming and expensive. It has affected the pharmaceutical industry, academia, and regulatory agencies profoundly, and has added years and millions of dollars to the process of approval of a drug for human use. Toxicity may be identified by alterations in (1) structure (macroscopic, microscopic, ultramicroscopic), (2) biochemistry (e.g., release of troponins, altered myocardial oxygen consumption and ATP deprivation), (3) physicochemistry (e.g., decreased density or conduction through specific ion channels and altered binding of calcium to troponin-C), or (4) physiology (e.g., rate of discharge of SA node, conduction through cardiac structures, irritability, inotropy, lusitropy, baroreceptor sensitivity).
There are literally hundreds of ways to search for a toxic liability in studies of toxicology and/or safety pharmacology. In designing these studies, as in ordering diagnostic tests for patients, it is important to consider advice proffered by the great toxicologist (arguably the father of modern toxicology) Gerhard Zbinden (Table 2A) to which I have added 3 corollaries (Table 2B).
The most challenging aspect of drug-induced TdP is that it occurs so rarely. Of 300,000 sudden cardiac deaths/year in the United States, it is estimated that only 15,000 result from TdP (Fung et al., 2000). Whereas TdP may occur in 1 in 100 patients receiving dofetilide or 5 in 100 receiving quinidine, it may occur in no more than 1 in 10,000 to 100,000 patients receiving drugs like fexofenadine for noncardiac indications. Thus, to identify a torsadogenic potential from clinical trials in which patients are given therapeutic doses, a large number of subjects must be interrogated. If no instances of TdP are observed in 5,000 subjects (a very large clinical experience, indeed), the upper limit of the 95% confidence interval for the risk of TdP in the general population is given by 3/5,000, or 1 in 1667. Therefore, it would be unlikely to observe a positive signal in a clinical trial of a drug that produces TdP in 1 patient in 10,000 or 1 in 100,000. For this reason, it appears prudent to conduct preclinical trials in which supratherapeutic doses may be given to amplify the possibility of identifying a liability to prolong ventricular repolarization, and to minimize the potential risk for subjects in clinical trials by giving clinicians hints on what form toxicity may be manifest. Still, both the pharmaceutical industry and federal regulatory agencies appreciate that there is never assurance that a drug may not manifest toxicity in a given patient.
We must recognize the fact that potential therapeutic agents may manifest toxicity—and even produce death—by affecting any or all of the properties of the cardiovascular system (Table 3). For example, a lengthening of ventricular repolarization may translate into, at most, 15,000 deaths per year from TdP. By comparison, it is estimated that systemic arterial hypertension, an important risk factor for stroke and a coronary events, occurs in approximately 60,000,000 persons in the United States, and it is known that a doubling of morbidity or mortality occurs for a 10 mm Hg increase in systolic pressure (Kannel, 2000). Thus, a drug that produces a relatively small (10 mmHg) increase in systolic blood pressure may translate into stroke or heart attack in manifold more patients than a drug that produces even a monumental lengthening of ventricular repolarization.
Another challenge in identifying drug toxicity is that some toxicities are manifested only after lengthy patient exposure (Kuryshev et al., 2005) or after rapid administration (Carlsson et al., 1993; Ficker et al., 2004), or in the presence of compounds that alter their metabolism (Abernethy and Flockhart, 2000). Whereas, some compounds produce toxicity within seconds or minutes after administration (e.g., dofetilide, cis-apride, erythromycin), others (e.g., fenfluramine, rofecoxib, and doxorubicin) may require years of exposure before toxicity is manifested.
It goes without saying that drugs typically are not given to normal persons, and that persons afflicted with any number of diseases that occur in large numbers of humans (e.g., heart failure, ventricular hypertrophy, diabetes, hypokalemia and/or hypomagnesemia) constitute particular risk for developing toxicity. Therefore, unless the predictive value (i.e., sensitivity and specificity) of studies on normal animals is perfect, it might be prudent to conduct studies on animals afflicted with the diseases with which humans targeted to receive the drugs are afflicted, particularly if it can be demonstrated the studies conducted on models with disease are more predictive.
In the case of TdP, heart failure is a known risk (Torp-Pedersen et al., 1999). Normal rabbits given most test articles known to lengthen QTc in man, also develop lengthening of QT (Figure 2, top), but not TdP, when exposed to escalating doses of dofetilide. However, rabbits with heart failure produced by ligation of coronary arteries develop not only greater lengthening of QT than normal rabbits, but also TdP (Figure 2, bottom). Thus, this animal model of a disease that occurs prevalently in man and that amplifies the risk to TdP may be useful for predicting torsadogencity in man. Other preparations are more permissive for developing TdP, not merely prolongation of QTc, but other surrogates (e.g., complete heart block (Vos et al., 1995)) veratridine, methoxamine (Ben-David and Zipes, 1990; Carlsson et al., 1993) do not mimic the physiological state of the vast numbers of persons who develop drug-induced TdP. Again, until information on specificity and specificity for preclinical studies is known, it is not known if studies on subjects with disease are necessary.
Because the arrhythmia TdP is so important to drug development (a leading cause for “death” of a drug in development or for removal from the market) and it is an electrocardiographic phenomenon, it is appropriate to discuss some fundamentals of electrocardiography (A, B, C, D). The ECG (Figure 3) is a recording of voltages from the surface of the torso and is produced by waves of depolarization and waves repolarization (Figure 4) traversing the heart. The P wave is produced when waves of depolarization, emanating from the SA node at the juncture of the cranial vena cava and right atrium, traverse the atria. Not only do these waves produce the P wave of the ECG, but they “shock” the atria into contraction. The wave of depolarization travels across the atrioventricular (AV) conduction system, after tarrying in the head of the AV node located just above the leaflets of the tricuspid valve. After the wave traverses the head of the AV node, it travels very rapidly over the His-Purkinje fibers to the subendocardial layers of both ventricles. In primates and carnivores, it then travels first through the apical-third of the interventricular septum predominantly from left ventricle toward right ventricle, and finally in a general endocardial to epicardial direction through the ventricular free-walls (but not in most other species). Finally, the wave traverses the bases of both ventricles and of the interventricular septum in a general apico-basilar direction. Depolarization of the ventricular myocardium (as the intracellular milieu goes from negative to positive) produces the QRS complex of the ECG. As the ventricles repolarize (as the intracellular milieu goes back to negative), the ST-T portion of the ECG is produced. Atrial re-polarization occurs immediately after the P wave and, for the most part, is obfuscated by the QRS complex. Atrial repolarization may produce a small “hammock-like” deflection—the Ta wave—between the P wave and QRS complex.
Torsade de pointes is a form of ventricular tachycardia (a period of rapid heart action), which occurs when regions of the ventricle do not wait for a wave of depolarization to come to them from the SA node through the AV conduction system and via the His-Purkinje system; rather regions from within the ventricular myocardium discharge by “self-stimulation”. This “self-discharge” in ventricular myocardium and/or Purkinje fibers may take many forms (Figure 5): (A) single premature depolarizations; (B) paroxysms (bursts) of ventricular premature depolarizations from a single focus; (C) a sustained discharge (ventricular tachycardia); (D) a discharge exchanging its origin between the ventricles (torsade de pointes). Most seriously, multiple foci, somewhat isolated from one another, may discharge and produce hundreds of wavelets meandering around the ventricles so that many regions of the ventricles are stimulated and therefore contract independently of the others. This is known as ventricular fibrillation (Figure 6, and may evolve from torsade de pointes or arise from other causes. In fact, a ventricular arrhythmia arising from any mechanism (Figure 7) may evolve (deteriorate) in to either ventricular torsade de pointes or ventricular fibrillation.
Torsade de pointes usually occurs when a mid-myocardial and/or Purkinje fiber undergoes an after-depolarization (Figure 8). An after-depolarization is a depolarization triggered by a previous relatively normal depolarization (Cranefield and Aronson, 1988). As the cell begins to repolarize due to potassium ions exiting the cytosol (through numerous ion-specific channels like the hERG channel) during the normal action potential, abnormal calcium fluxes occur inexplicably and the cell redepolarizes by itself. This depolarization then travels through the ventricles in a circuitous path, returns to the site of initiation, and may keep circulating, which produces a ventricular tachycardia originating from a single site. However, with TdP, the site rotates (twists) around a site in the ventricles, and oscillates between sites in the left ventricular and sites in the right ventricular myocardium. Thus, a single early after-depolarization initiates TdP, but the rapid ventricular tachycardia is sustained by so-called reentrant pathways that twist around a point (Figure 2). Torsade de pointes possesses many features which distinguish it from other arrhythmias (Table 4).
The early after-depolarization is caused probably by abnormal calcium homeostasis (January and Riddle, 1989; Nuss et al., 1999). Prolongation of the QT of the ECG is a surrogate for that abnormal homeostasis that may be caused by hyperphosphorylation of ryanadine channels in the sarcoplasmic reticulum (Marks, 2000), the phosphorylation probably caused by binding of calcium to calmodulin, which activates protein kinase A. If that binding is prevented with compound W-7, the QT remains prolonged but the tendency for development of TdP is diminished greatly (Gbadebo et al., 2002). Therefore, lengthening of QT is a surrogate—and a rather imperfect one at that—for TdP.
Preclinical and clinical trials, together, never will assure the absence of toxicity of a drug in some persons. However, we seek to minimize that risk, and we judge our success by the sensitivity and specificity of the paradigm used for prediction. To identify sensitivity and specificity, it is essential to know the percentage of times toxic effects are predicted accurately (sensitivity), and the percentage of times the absence of toxic effects are predicted accurately (specificity). Of course, if development of a potential therapeutic agent is terminated in a preclinical trial because it manifested toxicity and therefore the compound never reached the clinic, it would be impossible to estimate specificity. Currently, too little information is available to determine, precisely, the predictive values of preclinical, clinical, or preclinical and clinical studies together. Not only must we evaluate predictive value to identify toxicity manifested on physiological parameters (e.g., rhythm, force of contraction, myocardial stiffness) but also toxicity manifested on morphological parameters (e.g., infarction/fibrosis, myofibrillar degeneration, mitochondrial swelling). Therefore, safety pharmacology and toxicology have evolved into disciplines that search for potential toxic manifestation in both structure and function. A person who dies from Torsade de pointes evolving into ventricular fibrillation may not have any morphological or biochemical finding. A person who has a drug-induced elevation of alkaline phosphatase or troponin, or who has myofibrillar degeneration or mitochondrial swelling, may have neither morbidity nor mortality. For these reasons, studies to detect and/or to predict arrhythmia must include analyses of structural, biochemical, physicochemical, and physiological properties (all potential targets for toxicity) of usual and elevated concentrations of test articles for rather long durations.