Psychotropic Drugs, Cardiac Arrhythmia, and Sudden Death

20 Pages·2003·1.04 MB·English

by  Power Mac I

REVIEW ARTICLE
Psychotropic Drugs, Cardiac Arrhythmia,
and Sudden Death
Harry J. Witchel, PHD*, Jules C. Hancox, PHD*, and David J. Nutt, DM, MRCP, FRCPSYCH†
a cardiac arrhythmia occurring within an hour of onset of
Abstract: A variety of drugs targeted towards the central nerv-
symptoms.5 In many cases psychotropic drugs, particu-
ous system are associated with cardiac side effects, some of which
larly tricyclic antidepressants (TCAs) and antipsychotic
are linked with reports of arrhythmia and sudden death. Some
agents, are correlated with iatrogenic prolongation of the
psychotropic drugs, particularly tricyclic antidepressants (TCAs)
QT interval of the ECG,6which is associated with the dan-
and antipsychotic agents, are correlated with iatrogenic prolonga-
gerous polymorphic ventricular tachyarrhythmia, TdP.7
tion of the QT interval of the electrocardiogram (ECG). In turn,
Furthermore, of all the drugs that are taken in lethal over-
this is associated with the arrhythmia torsades de pointes(TdP). This
dose, prescribed antidepressants are among the most com-
review discusses the association between psychotropic agents,
mon; when considering cardiotoxicity, psychotropic drugs
arrhythmia and sudden death and, focusing on TCAs and anti-
must be considered in a completely different light from
psychotics, considers their range of cellular actions on the heart;
other drugs such as antiarrhythmics or nonsedating anti-
potentially pro-arrhythmic interactions between psychotropic and
histamines because of the suicide risk.
other medications are also considered. At the cellular level TCAs,
Although it is thought that the savings in lives from
such as imipramine and amitriptyline, and antipsychotics, such as
the use of these psychotropic agents (due to their therapeu-
thioridazine, are associated with inhibition of potassium channels
tic actions) outweighs the risks of their being abused for self-
encoded by HERG. In many cases this cellular action correlates
poisoning and overdose (e.g.8), when considered within an
with ECG changes and a risk of TdP. However, not all psycho-
epidemiological context, there are marked differences in
tropic agents that inhibit HERG at the cellular level are associated
overdose toxicity between drug classes and, in some cases,
equally with QT prolongation in patients, and the potential for QT
between individual drugs within a class.9This review is con-
prolongation is not always equally correlated with TdP. Differ-
cerned with the connection between psychotropic drugs
ences in risk between classes of psychotropic drugs, and between
(particularly TCAs and antipsychotic agents), arrhythmia
individual drugs within a class, may result from additional cellular
and sudden cardiac death, at normal therapeutic concentra-
effects of particular agents, which may influence the consequent
tions, during overdose, and in patients with established car-
effects of inhibition of repolarizing potassium current.
diovascular disease. The different classes of psychotropic
(J Clin Psychopharmacol2003;23: 58–77) drugs, and even different drugs within the same class, are
associated with differential risks, and this will be described
in detail. In order to provide an overview of the issues and
It has long been recognized that some psychotropic drugs, evidence for this connection at the clinical and cellular
including neuroleptics, antipsychotics, antidepressants, level, the proposed mechanisms will be described at the
stimulants and antianxiety agents, can be associated with level of cardiac electrophysiology and propositions for fu-
risks of cardiac arrhythmia1,2,3and sudden death. In addi- ture research will be presented.
tion, antiepileptics and drugs used for movement disorders
may also result in cardiac effects in some patients (e.g. pi- CLINICAL ISSUES AND EVIDENCE
mozide4). Sudden cardiac death is generally defined as
Prolongation of the QT interval
death from unexplained circulatory arrest, usually due to
Given the link between the risk of cardiotoxicity and
the prolongation of the QT interval by psychotropic drugs,
*Department of Physiology and Cardiovascular Research Laborato-
it is beneficial to consider first the issue of QT prolongation.
ries; †Psychopharmacology Unit, School of Medical Sciences, University
of Bristol, Bristol, United Kingdom In addition to ventricular conduction time, the QT interval
Received June 18, 2001; accepted March 11, 2002. of the ECG reflects the duration of the ventricular action po-
Address reprint requests to: Harry J. Witchel, PhD, Department of tential (AP) at the cellular level (Fig. 1). Thus, the prolonga-
Physiology and Cardiovascular Research Laboratories, School of Med-
tion of the QT interval may reflect effects on ion channels
ical Sciences, University of Bristol, University Walk, Bristol, BS8 1TD,
involved in ventricular AP generation. However, the QT
United Kingdom. Address e-mail to: [email protected]
10.1097/01.jcp.0000057188.35767.1d interval also depends on the heart rate, and therefore the
58 Journal of Clinical Psychopharmacology • Volume 23, Number 1, February2003
Journal of Clinical Psychopharmacology • Volume 23, Number 1, February 2003 Psychotropic Drugs and Arrhythmias
scope of this problem has made it an issue for drug devel-
opers and health authorities,10as well as being of critical im-
portance for attending physicians. The non-cardiac agents
associated with QT prolongation belong to many different
pharmacological classes including psychotropic drugs, pro-
kinetic medicines, antimalarial medicines, antibiotics be-
longing to several different chemical classes, antifungal
agents and non-sedating antihistamines.14,15 Among the
psychotropic drugs associated with QT prolongation and
the risk of sudden cardiac death under some circumstances
(Table 1) are TCAs (e.g. imipramine and amitriptyline
among others), tetracyclic antidepressants (e.g. maproti-
line), phenothiazine derivative antipsychotics (e.g. thiorida-
zine), potentially some new atypical antipsychotics (e.g.
sertindole), and under limited circumstances drugs indicated
for movement disorders including neuroleptics, antiepilep-
tics and antiparkinsonian agents. Of these agents, the most
well described risks are associated with TCAs.
FIG. 1. QT interval correlates with ventricular action potential
duration. The absolute QT interval is the QRS time course (rep-
resenting propagation time of AP depolarization through ven- TABLE 1. Selected drugs associated with arrhythmia risk
tricular tissue) plus the JT interval (representing the overall time factors
course for the plateau phase plus ventricular repolarization to
Clinical HERG
occur). Schematic diagrams of ventricular action potentials are
CNS drug reference reference
shown, A being prolonged with respect to B. B is labeled with
the phases of the AP: phase 0, rapid upstroke; 1, early repolar- Amitriptyline 16 17,18,19
ization; 2, plateau phase; 3, late repolarization; and 4, post- Amoxapine 20,9
repolarization (slow depolarization in pacemaker tissues). Be- Carbamazepine With other factors, 21,22
low are the respective QT intervals and a schematized ECG Chloral hydrate Rare, 23
showing their relative timing. The depolarization (or rapid up- Chlorpromazine 24 19
stroke) representing inward flux of Na(cid:2)ions, and the repolar- Citalopram In overdose, 25 26
ization, representing primarily outward flux of K(cid:2)ions, are in- Clomipramine 27
dicated in the upper ventricular action potential. Clozapine with other drugs, 28
Desipramine 27
determination of QT prolongation in patients is usually Doxepin 29
Droperidol 30 31
measured using the heart rate corrected QT interval (QT ).
c Fluoxetine with other drugs, 32 26
QT can be calculated in several ways, including the Frid-
c Haloperidol 33 34
ericia formula, QT (cid:1)QT /3(cid:1)RR, or the more prevalent cor- Imipramine 35 17
c
rection formula of Bazett, QT (cid:1)QT / (cid:1)RR, the latter being Lithium wither other drugs, 36
c
relatively inaccurate because it under- or over-estimates the Maprotiline 37
Nortriptyline 38
true duration of repolarization at low and high rates, re-
Olanzapine Not significant, 39 40
spectively.10Prolongation of the QTcinterval (in excess of Pergolide (cid:2)CAD or levodopa, 41
440–450 milliseconds (ms) in men and 460–470 ms in adult Pimozide 4 42
women11) is associated with an increased cardiovascular Prochlorperazine In overdose, 43
risk,12and QT prolongation is considered a marker of risk Quetiapine In overdose, 44,45
c Risperidone Rare, 46 40
for TdP or sudden death by drug regulating authorities.11
Sertindole 47 48
TdP is characterized by an ECG pattern of continuously
Sultopride 49
changing morphology of the QRS complexes that seem Trazodone In overdose, 50
to twist around an imaginary baseline.13 In acquired long Thioridazine 6,51 52
QT syndrome (LQTS) the QT interval is prolonged, and Venlafaxine In overdose, 53
c Zimeldine With hypokalaemia, 54
the surface ECG often shows accentuated T- and U-wave
Ziprasidone 55 40
abnormalities.
The putative arrhythmogenic effects of psychotropic and other se-
Vulnerability to iatrogenic QT prolongation lected CNS drugs. Clinical references refer to either evidence of TdP,
QT prolongation, or sudden cardiac death. In some examples the cita-
A wide range of non-cardiovascular therapeutic agents c
tions refer to cases of overdose and poisoning.
have been shown to prolong the QT interval, and the large
©2003 Lippincott Williams & Wilkins 59
Witchel and Associates Journal of Clinical Psychopharmacology • Volume 23, Number 1, February 2003
Given this potentially large number of drugs associ- with pre-existing conduction disease, particularly bundle
ated with risk, it is important to note that among patients branch block, are at a higher risk to develop symptomatic
taking these drugs, significant QT prolongation (QT (cid:3) AV block than depressed patients that are free of conduc-
c c
500 ms) is a relatively uncommon occurrence, and that tion anomalies;62 the severity of the patient’s pre-existing
TdP is much more rare. For a given medication, arrhyth- condition is a greater determinant of vulnerability to such
mogenic risk must be understood to be partially depend- complications than the choice of which agent within a par-
ent mechanistically on vulnerability that is intrinsic to the ticular drug family is used.63
individual patient with regard to the particular drug. Ac- The cardiac actions of TCAs have been suggested to
quired LQTS is typically seen during bradycardia, irregu- result from “quinidine-like” activity.64 Quinidine, a class I
lar rhythm, hypomagnesia and/or hypokalemia,11 and it antiarrhythmic, is known to cause slowing of phase 0 de-
appears to be more prevalent in women than in men. Vul- polarization of the AP (Fig. 1) resulting in slowing of con-
nerability to medication-induced prolongation of the QT duction through the His-Purkinje system and myocardium.
interval may also reside in the patient with respect to the It is well established that, in addition to its antiarrhythmic
underlying psychiatric diagnosis, any underlying medical effects, quinidine can have pro-arrhythmic activity.65Pre-
illness (especially cardiovascular disease), genetic predis- viously, TCAs were known to have class I antiarrhythmic
position to drug-induced QT prolongation, and the pa- activity and were recommended for patients with ventric-
tient’s age, as described below. ular arrhythmias.66Due to more recent evidence showing
There are cardiovascular risks associated with the pa- that after myocardial infarction (MI) some class I antiar-
tients’ underlying psychiatric diagnosis, which has been rhythmics are associated with significantly increased mor-
well demonstrated in the case of depression. For example, tality due to arrhythmia,67 the cardiac actions of TCAs
the propensity for TCA-induced orthostatic hypotension is have led to the recommendation that alternatives to TCAs
increased in the depressed state,56and the risk for cardio- be considered under those circumstances.68
vascular mortality in post-MI patients is increased in the These cardiac effects of antidepressant drugs after
presence of depressive symptoms. Similarly, in schizo- MI are important because depression and MI are often co-
phrenic patients there may be issues concerning self-care morbid, with the prevalence of symptoms of major de-
or overall medical compliance that may create risk. In pression in patients with MI being up to 18%,69as opposed
many cases these risk factors can increase the vulnerabil- to 5% in some cross-sectional studies of the adult popula-
ity of the patient to other risk factors and possibly act in tion.70For some time, depression has been examined as an
concert with the undesired pharmacological actions of additional independent risk factor for mortality in cardiac
some therapeutic agents to increase the risk of QT prolon- patients, particularly with respect to sudden arrhythmic
gation, TdP and sudden death. With psychiatric patients deaths.71–75Given the significant levels of depression asso-
having known or putative risks, inspection of each patient’s ciated with MI, antidepressants are considered to be un-
ECG may be important in assessing risk associated with derprescribed in this population, leading to morbidity and
drugs that can have arrhythmogenic properties. decreased quality of life.76,8,71Considering the overall data
currently available regarding the relative merits of TCAs
and SSRIs, it is now thought to be safer to treat depressed
Cardiac effects of TCAs versus Selective
cardiac patients who also have established ischemic heart
Serotonin Reuptake Inhibitors (SSRIs)
disease (IHD) with SSRIs than with TCAs, although treat-
Cardiac effects of TCAs ment decisions must be individualized for each patient.77
TCAs are used for treatment of depression and other
conditions including nocturnal enuresis, panic disorder, ob- TCAs in psychiatric patients without
sessive compulsive disorder and bulimia.57,58,59TCAs have established coronary artery disease
cardiovascular side effects including orthostatic hypoten- In psychiatric patients without established coronary
sion, atrioventricular conduction delay, reduced heart rate artery disease (CAD), there have been several studies asso-
variability, tachycardia, syncope, and lengthening of the ciating the therapeutic use of psychotropic drugs with inci-
QT interval, particularly in cases with high dosages and in dent MI, raising the possibility that psychotropic-induced
patients with concurrent cardiovascular disease.60 TCAs sudden death could be mediated by CAD; however, the
are associated with a significant clinical risk of arrhythmia majority of sudden cardiac death patients have no evidence
in patients with pre-existing heart disease or conduction of recent MI, although many have active lesions.78,79In one
abnormalities60,61 and in overdose.9 In the case of TCA case-control study of cardiovascular mortality in women
overdose, delayed conduction may lead to a complete aged 16–39 not designed to test any hypothesis about psy-
heart block or ventricular reentry arrhythmias. At thera- chotropic drugs, one incidental finding was an unexpected
peutic plasma concentrations of TCAs, depressed patients 17-fold increase in the risk of fatal MI associated with the
60 ©2003 Lippincott Williams & Wilkins
Journal of Clinical Psychopharmacology • Volume 23, Number 1, February 2003 Psychotropic Drugs and Arrhythmias
current use of psychotropic medications.80Another cohort sion.87The importance of these studies is that they may rep-
study of risk found associations between fatal MI and both resent the natural course of affective disorder, as there were
benzodiazepine and antidepressant use.81Neither of these no somatic treatments for depression available at the time.
studies controlled for depression or other mental illnesses, Given the complexity of separating the putative car-
but the Baltimore ‘Epidemiologic Catchment Area’ (ECA) diotoxic effects of therapeutic TCA use from those of the
survey follow-up did. After adjustment for major depres- underlying psychiatric indications, more recent work has
sive episode and dysphoria, the Baltimore ECA follow-up focused mechanistically on whether therapeutic TCA use
found that the two most commonly prescribed classes of is a significant risk factor for QT interval lengthening. In
c
agents for depression at the time of their baseline inter- a study by Reilly et al. of 495 psychiatric patients in vari-
view, TCAs and benzodiazepines, were each associated ous inpatient and community settings who were compared
with non-significant increased relative risks (RR) of (cid:2)1.3 to normal controls, therapeutic use of TCAs was associated
for self-reported incident MI;82 nevertheless, it is contro- with QT prolongation;6 in that study stepwise regression
c
versial as to whether sudden cardiac death may be sec- analysis showed that the only other significant independ-
ondary to acute ischemia in patients who have not devel- ent risk factors besides the use of specific psychotropic
oped enzymatic or ECG evidence of acute MI.5 drugs were ‘age (cid:3) 65’ and female gender. Furthermore,
Large epidemiological studies of the association be- studies of “healthy” pediatric psychiatric patients do not
tween TCA use and total mortality are difficult to cite as support a differential risk between age groups, and they
proof of a causal connection. In a case-control study suggest that aggregate pediatric ECG findings are consis-
Weeke et al. compared the mortality rate for first hospital- tent with those found in adults, although desipramine,
ized manic-depressive patients in the era before TCAs be- which can bind to noradrenaline reuptake sites in cardiac
came available in Scandinavia to that after the introduction sympathetic neurons in man in vivo,88has been singled out
of TCAs, and they found that the rate in the post-TCA era as the cause of four pediatric nonsuicide deaths.89–91TCA-
was lower.83 Similarly, Avery and Winokur found that induced QT prolongation may result from direct effects of
c
mortality from cardiac causes over a three year follow-up TCAs on ion channels within the myocardium, and as
was more prevalent among depressed patients who had shall be explained below, there is evidence that TCAs can
received “inadequate” treatment for depression rather inhibit repolarizing ionic current in cardiac ventricular
than those who had received adequate treatment;84 unfor- myocytes, an observation that would account for drug-
tunately, this data is not prospective and is open to criticisms induced lengthening of the QT interval of the ECG.
c
of poor health care overall in the “inadequate” population,73 An advantage of SSRIs over TCAs is that they are
and a similar argument can be made for comparison of the rarely associated with QT prolongation,92 and they have
c
pre- and post-TCA era mortality rate. Overall, the epidemi- fewer cardiotoxic side effects93 and a higher margin of
ological data, especially data for non-suicide mortality, must safety than TCAs.94 In chronic use, some SSRIs cause a
be interpreted carefully because the causal links often lack mild bradycardia of a few beats per minute.38There have
pathological evidence, and the pathology of sudden death is been occasional reports of syncope and arrhythmia asso-
often not definitive except in rare cases where the patient is ciated with fluoxetine in some patients.95,96However, clin-
hospitalized or being Holter monitored.85 ical doses of the SSRIs fluoxetine, sertraline and paroxe-
tine generally have been found thus far to be safe in
Mechanistic issues in the association of patients with those forms of pre-existing heart disease that
antidepressants and sudden death have been tested (e.g.97,98,38), although all of these studies
The statistical connection between psychotropic drugs were small, and the SSRIs are known to have effects on the
and cardiac death can be further confounded by observa- cytochrome P450 system, raising the possibility of drug-
tions that the psychiatric disorders that indicate the use of drug interactions in patients with multiple medications.
these drugs are also associated with high rates of cardiovas- Data from national registries of suicides support a view that
cular mortality. In 1937 Malzberg compared the mortality SSRIs, particularly fluoxetine, are in general dramatically
rate for patients hospitalized with involutional melancholia safer (10–100 fold) than TCAs,9particularly with regard to
to that of the general population, and he reported that the cardiovascular effects and mortality.
non-suicide death rate was six times greater in the melan-
cholia patients; this was consistent across the age span of 40 Neuroleptic Agents
to 75 years.86 In that study cardiac disease accounted for It has long been known that some antipsychotic
nearly 40% of all mortality among the melancholia patients, agents, particularly phenothiazines, are associated with
and the rate of cardiac death among the melancholia pa- ECG changes representing repolarization anomalies in-
tients was eight times greater than in the general population. cluding QT prolongation,99,100 and the first report of sud-
Other early evidence has supported Malzberg’s conclu- den arrhythmic death due to thioridazine was published in
©2003 Lippincott Williams & Wilkins 61
Witchel and Associates Journal of Clinical Psychopharmacology • Volume 23, Number 1, February 2003
1963.2The difference in toxicity of the different neurolep- In the absence of long-term epidemiological data on
tics and antipsychotics was demonstrated by an Australian standardized mortality ratios, data on QT prolongation,
study of 299 consecutive patients admitted with neurolep- heart rate variability or self-poisoning suggests that the
tic poisoning; thioridazine was significantly more likely to new atypical antipsychotic drugs have somewhat different
cause tachycardia, prolonged QT (cid:3)450 ms, a widened QRS profiles. Clozapine, which carries a risk of agranulocytosis
c
((cid:3) 100 ms) and arrhythmias (p (cid:1) 0.004) than chlorpro- and may show evidence of being associated with my-
mazine, trifluoperazine, pericyazine, haloperidol, prochlor- ocarditis and cardiomyopathy,112–115 also significantly re-
perazine, fluphenazine, or other neuroleptics.43Likwise, a duces measures of heart rate variability associated with
Finnish study of sudden unexplained deaths associated parasympathetic control, and it significantly increases the
with psychotropic drugs found phenothiazines to be over- LF/HF ratio, which is a measure of sympathovagal bal-
represented in the sample, and thioridazine was involved ance.39,116 In a study of overdoses monitored by the Na-
in a disproportionate number of cases.51TdP has been ob- tional Poisons Information Service in London, clozapine
served after thioridazine101,102and haloperidol overdose.103 overdose was associated with sinus tachycardia (as well as a
Even at dosages used for therapy, Reilly et al. found number of non-cardiac symptoms) in over 66% of the single-
that antipsychotic drugs, particularly thioridazine and dro- agent overdoses reported;117in the same study over 66% of
peridol, were associated with QT lengthening in a dose the patients reported with an overdose of risperidone were
c
dependent manner;6this study was the first evidence men- asymptomatic. Given the issues associated with the agranu-
tioned in the announcement by the Committee on Safety locytosis risk and the arrhythmogenesis risk, it is not surpris-
of Medicines (UK) that indications for thioridazine were to ing that one survey of ordinary clinical practice in South
be restricted and that droperidol was to be voluntarily dis- London concluded that the new atypical antipsychotics ap-
continued by the manufacturer.104 In the USA, therapeu- pear to be replacing older neuroleptics as the first-line treat-
tic levels of traditional antipsychotic agents have also been ment of schizophrenia, while clozapine is mostly reserved for
associated with risk in a recently published epidemiologi- poor responders.118
cal retrospective study; in that cohort of 481,744 persons In a recent clinical trial comparing mean QT (Bazett’s)
c
aged 15–84 years, those patients currently (during the at maximal therapeutic dosages of ziprasidone, risperidone,
study period of 1988 to 1993) prescribed moderate doses olanzapine, quetiapine, thioridazine and haloperidol, as ex-
of antipsychotics were found to be at a significantly in- pected thioridazine increased QT the most (35.6 ms), fol-
c
creased risk of sudden cardiac death (multivariate risk ra- lowed by ziprasidone (20.3 ms), quetiapine (14.5 ms), risperi-
tio 2.39) compared to patients currently prescribed low done (9.1 ms), olanzapine (6.8 ms) and haloperidol (4.7
doses of antipsychotics, formerly prescribed antipsychot- ms).55 Mechanistically, these figures cannot be assumed to
ics, or non-use, and the risks were higher when comparing correlate directly with arrhythmogenic risk, given the wide
patients with severe cardiovascular disease.105 range of possible correction formulae for QT. An alterna-
c
tive investigation of risk is to characterize drugs for those cel-
New Atypical Antipsychotic Agents lular mechanisms associated with TdP. Indeed, in recent pre-
The new atypical antipsychotics, promising both clinical studies these drugs, as well as some drugs indicated
greater efficacy and reduced side effects, have not yet been for movement disorders that have been associated with ar-
studied in as much detail as traditional antipsychotics, but, rhythmogenic risk (e.g. pimozide),4have also been shown to
with the exceptions of sertindole and ziprasidone, they share the cellular mechanism believed to be responsible for
have not been consistently statistically proven to be asso- acquired LQTS attributed to many traditional antipsychotics
ciated with either QT prolongation or sudden cardiac and to TCAs.42,55
death at therapeutic concentrations.106,107Sertindole is as-
sociated with an increase in the QT interval at therapeu- CELLULAR EFFECTS
c
tic dosages,108,39and after release in the UK, accumulating The cellular mechanism of psychotropic-mediated
evidence of unexplained deaths and serious but not fatal effects on the AP can be examined using a broad range of
arrhythmias led the manufacturer to voluntarily withdraw in vitromodels including heterologous expression systems
the drug pending further study.106Overall, there is a diffi- (involving transference of the relevant genes into mam-
culty in ascribing sudden deaths to particular antipsy- malian cell lines or into Xenopus oocytes), disaggregated
chotics at therapeutic doses because of the known associa- cells (studied acutely or in culture), isolated tissues, and the
tion between schizophrenia and increased mortality due to isolated perfused heart.119 Control of the duration of the
cardiovascular causes,109,110and even in the case of sertin- AP in the ventricular myocyte is mediated by the equilib-
dole in the UK, a statistically significant differential in sud- rium between inward and outward currents across the cell
den death compared to other new atypical antipsychotics membrane120(Fig. 2). The longer action potential duration
was not shown.111 (APD) implicit in QT prolongation (Fig. 1) results from a
62 ©2003 Lippincott Williams & Wilkins
Journal of Clinical Psychopharmacology • Volume 23, Number 1, February 2003 Psychotropic Drugs and Arrhythmias
ionic currents) differs between species and according to re-
gional origin of the preparation (e.g. atrium versus ventri-
cle). Conclusions concerning APD from dissociated cell
preparations should also be considered in the context that
intrinsic variation in APD in those systems can occur.119
Given these caveats, it is perhaps not surprising that the re-
ported effects of psychotropic drugs on APD vary across
the different experimental preparations studied. For ex-
ample, imipramine causes shortening of the APD in iso-
lated bovine122 and guinea pig123 ventricular myocytes,
whereas in rabbit atrial myocytes124and rat atrial fibers,125
it lengthens the APD. The differential effects of the psy-
chotropics on AP repolarization will depend upon the bal-
ance between the inhibition of the various inward currents
versus the inhibition of outward currents.
Several distinct K(cid:2)channels play prominent roles in
the initiation and completion of AP repolarization. The rate
of net K(cid:2)efflux, and by extension the rate of repolarization,
is determined by the density and gating properties of dif-
ferent K(cid:2) channels. Of particular relevance to regulating
the duration of the AP plateau is the delayed rectifier cur-
rent (I ) comprised of rapid (I ) and slow (I ) components
K Kr Ks
mediated by distinct channel subtypes with distinct kinetic
properties126,127 (Fig. 2). Composite I develops progres-
K
sively during the plateau phase, opposing the inward cur-
rents underlying plateau depolarization. As the net balance
of outward current exceeds inward current, repolarization
occurs. At a cellular level, pharmacologically-induced slow-
ing of AP repolarization can lead to spontaneous depolar-
FIG. 2. Ionic activity during the cardiac ventricular action po-
tential. Above is a schematic of the dominant ionic currents izations generated on the falling phase of the AP plateau.
moving across the myocyte membrane over time (abbrevia- Unlike the AP itself, these events (termed ‘Early After De-
tions: K (cid:1)potassium, Na (cid:1)sodium, Ca (cid:1)calcium; I (cid:1)tran- polarizations’ or ‘EADs’) are not synchronized with an exci-
to
sient outward current, which is a compound current). Below
tatory stimulus (in vivoan incoming wave of excitation, ex-
that is shown the resultant membrane voltage profile over time
perimentally an applied injection of current), and so can
during an idealized action potential. Downward is polarized
(negative voltages) and upward is depolarized (positive volt- give rise to asynchronous tissue excitation and, thereby, ar-
ages). The arrow indicating the time occurs at 0 mV. The lower rhythmogenesis.
7 traces show the main ionic currents contributing to the ac-
tion potential over time; downward represents inward (depo- Cloned channels responsible for delayed
larizing) currents, upward represents outward (repolarizing)
repolarization and LQTS
currents. For the sake of simplicity, some currents have been
omitted (e.g. Na/Ca exchange), and the current records only Molecular genetic studies have played important
demonstrate relative polarities and time courses–relative am- roles not only in establishing links between particular
plitudes are not to scale. channelopathies and LQTS, but also in producing impor-
tant insights into the function of individual ion channel
disturbance of the normal balance between inward (depo- types. Therefore, before considering the pharmacological
larizing) and outward (repolarizing) currents during the AP modulation of ion channel currents involved in pharma-
plateau phase. Prolongation of the AP could theoretically cologically induced LQTS, it is first useful to consider the
arise from an increase in depolarizing current, or alterna- basis for ‘inherited’ or ‘congenital’ LQTS.
tively, from a decrease in repolarizing current carried by Thus far five genes have been identified as sites of
potassium (K(cid:2)) ions. Both of these conditions have been mutation for the congenital disorder (Table 2). Genetic ap-
found to exist in different forms of inherited LQTS.121 proaches have shown that all five of these genes are re-
In assessing the effects of psychotropic (or other) sponsible for ion channel proteins that are present in the
drugs on APD in experimental systems, it is worth noting myocardium, providing insight into the electrophysiologi-
that the AP profile (and therefore underlying balance of cal causes of LQTS. A major breakthrough occurred when
©2003 Lippincott Williams & Wilkins 63
Witchel and Associates Journal of Clinical Psychopharmacology • Volume 23, Number 1, February 2003
with clarithromycin-induced arrhythmias, and mutations
TABLE 2. Channels associated with LQTS
in this gene are thought to be clinically silent until com-
Original Channel type Chromosomal Alternative bined with additional stressors.134 Finally, mutations in a
name (subunit type) location nomenclature sodium (Na(cid:2)) channel gene SCN5A135 (locus 3p21–p23)
HERG K(cid:2)((cid:4))–I 7q35–36 KCNH2 can also cause congenital LQTS; rather than blocking an
Kr
KvLQT1 K(cid:2)((cid:4))–IKs 11p15.5 KCNQ1 outward current of myocyte repolarization (as with K(cid:2)mu-
SCN5A Na(cid:2)((cid:4)) 3p21–p24 hH1
tations), this mutation interferes with channel inactivation,
minK K(cid:2)((cid:5))–I 21q21–p23 KCNE1
hMiRP1 K(cid:2)((cid:5))–IKs 21q21–p23 KCNE2 causing the channel to conduct a sustained inward current
Kr
during the plateau phase of the AP, thereby delaying re-
polarization.121
HERG (Human Ether-a-go-go Related Gene; 7q35–q36) was
identified as responsible for one form of LQTS;128,129this Psychotropic Effects on Potassium Channels
gene encodes the alpha ((cid:4)) subunit (the main body of the Whilst inhibition of either I or I might result in
Kr Ks
ion channel) of a K(cid:2) channel which mediates I .130 The acquired LQTS, it is now well established that a diverse
Kr
discovery that HERG, a gene that codes for a cardiac ion range of psychotropic drugs associated with LQTS also in-
channel subunit, was responsible for congenital LQTS, has hibit HERG/I 14,136 (Table 1), making HERG a likely
Kr
supported the notion that pharmacological inhibition of primary mediator of acquired LQTS. Indeed, the TCAs
cardiac K(cid:2) channels was a possible mechanism for ac- imipramine137,17 and amitriptyline17–19 have been re-
quired LQTS.130 Another K(cid:2) channel protein associated ported to inhibit HERG/I at clinical dosages (Fig. 3), and
Kr
with LQTS named KvLQT1131 (locus 11p15.5) is the (cid:4)- to produce a profound channel block at higher dosages,
subunit responsible for I .132The gene KCNE1(21q22.1– although they can result in paradoxical AP shortening
Ks
p22) encodes a K(cid:2) channel (cid:5)-subunit (an accessory sub- because of calcium (Ca2(cid:2)) channel blockade.123 The ami-
unit) called minK that associates with KvLQT1 to form triptyline block of HERG is use-dependent and voltage-
functional channels mediating I .132,133 Another gene, dependent.18 The imipramine block shows weak voltage
Ks
KCNE2(21q11.1), encoding a K(cid:2)channel (cid:5)-subunit called dependence, and some dependence on the extent of chan-
MiRP1, which is part of the channel complex with HERG nel activation, as demonstrated by an ‘envelope of tails’
mediating I , has recently been shown to be associated protocol.17 Imipramine also delays the activation of the
Kr
FIG. 3. TCAs inhibit current mediated by heterologous HERG. (A) Shows the currents elicited from a CHO cell transfected with
HERG and from an untransfected cell. Membrane potential was held at (cid:6)80 mV, depolarized to (cid:2)20mV for 400 ms, before the tail
current was observed at (cid:6)40mV. The HERG current of a typical cell in extracellular solution is shown in (B) in the absence of drug
(control), in the presence of 3 (cid:7)M imipramine, and after washout. The holding potential and voltage steps are identical to those
in Figure 3a. Interpulse interval was 15 seconds. (C) Shows a typical cell superfused with 3 (cid:7)M amitriptyline under the same con-
ditions as in (B). The dotted line in (B) and (C) indicates zero current level. In (D), the concentration response curve shows mean
((cid:8)SEM) inhibition of tail currents measured with the same protocol as above. Reproduced with permission, from reference17.
64 ©2003 Lippincott Williams & Wilkins
Journal of Clinical Psychopharmacology • Volume 23, Number 1, February 2003 Psychotropic Drugs and Arrhythmias
slowly activating K(cid:2)current (I ) in guinea pig ventricular cluding prolongation of the myocyte AP, EADs, and QT
Ks
myocytes,137 which might also affect repolarization, but prolongation on the ECG.143 In preparations in which I
K
the block of I is more potent than that of I . Such data is small and I dominates repolarization, the block by
Kr Ks to1
are in accord with the proposition that, of these two potas- imipramine of I is likely to explain the prolongation of
to1
sium channels, the pathological effects of imipramine are APD in those cell types, such as rat,125 rabbit,124 and hu-
more likely to be mediated by blockade of I . man144 atrial and rat ventricular myocytes.145 However,
Kr
The associations between QT prolongation, the as- the human ventricular AP has a strong I component,146
c K
sociated risk of mortality and some antipsychotics (e.g. so the contribution of drug effects on I to the net drug ef-
to1
thioridazine and droperidol) are stronger than those for fects on the AP are likely to be complex. Overall, the ex-
TCAs;6 likewise the potency of thioridazine for blocking isting data suggest that inhibition of I (e.g. by TCAs) may
to1
HERG is higher than that of TCAs.19 Whilst TCAs have change the electrical heterogeneity of the myocardium,147
effects on APD that vary between experimental prepara- while TCA inhibition of HERG/I is likely to underlie or
Kr
tions (see section on “Cellular Effects”), antipsychotics are contribute very substantially directly to AP prolongation
generally seen to lengthen APD. The antipsychotic drug and to the pro-arrhythmia reported for this class of drugs.
thioridazine prolongs the APD in guinea pig ventricular Finally, an inwardly rectifying K(cid:2)current (I ) regu-
K1
myocytes by inhibiting I ,52and the effects of thioridazine lates ventricular AP repolarization over the final rapid re-
K
on I are over ten times more potent than its effects on I . polarization phase, and is also important in maintaining
Kr Ks
Sertindole, which has been shown to produce QT prolon- the normal resting potential.120In bovine ventricular my-
gation in humans at therapeutic doses,108 has also been ocytes, imipramine has been reported to enhance I tran-
K1
shown to block HERG.48 Data showing that the newer siently, before then reducing it (by only 19% at 3.6 (cid:7)M).122
atypical antipsychotics ziprasidone, olanzapine and ris- To summarise: many neuroleptics and TCAs can
peridone block HERG and I in a similar concentration block HERG/I , and this potentially mediates the repo-
Kr Kr
range to haloperidol has been presented in abstract form.40 larization anomalies associated with the risk of TdP. This
Even amoxapine (which is classified as a TCA, but also mechanism is more easily illustrated for neuroleptics; the
possesses neuroleptic properties138) prolongs the APD in connection between TCAs and QT prolongation is more
c
guinea pig isolated papillary muscles.139 This correlates complex.
with British toxicology data from the early 1990s, which
showed amoxapine to be associated with far more over- Psychotropic blockade of Na+ or Ca2(cid:1)
dose-attributed deaths per million prescriptions (the fatal inward currents
toxicity index – FTI) than any other antidepressant;9 In addition to effects on outward K(cid:2) currents, vari-
nevertheless, FTI figures would not be enough to label ous psychotropic drugs have been shown to inhibit inward
amoxapine as more dangerous than other TCAs, because ionic currents (especially those mediated by Na(cid:2) and
the FTI figures do not control for the effects of the under- Ca2(cid:2)). In myocytes where I dominates the transition from
K
lying indications. plateau phase to repolarization, those psychotropic drugs
TCAs have additional effects on other K(cid:2)channels. that inhibit I will tend to lengthen the AP, but if there is
K
Another K(cid:2) current, I , a transient outward current (I ) also a concomitant inhibition of the inward Ca2(cid:2)currents
to1 to
component (I can also include a chloride mediated cur- during the plateau phase, the APD may be shortened. This
to
rent component, I ) is important in repolarization occur- may in part explain why some psychotropics that block
to2
ring before the plateau phase. Antidepressant drugs with HERG/I do not prolong the ventricular APD in some
Kr
different chemical structures (including imipramine, ami- cellular models; for example, imipramine, which blocks
triptyline, mianserin, maprotiline, and trazodone) have HERG/I , shortens the APD in guinea-pig myocytes,
Kr
also been shown to block the transient outward potassium which can be ascribed to imipramine’s block of the L-type
current,140and the consequences of I blockade are a sub- (dihydropyridine-sensitive) calcium current (I ) and of
to1 CaL
ject of intense research. Recordings obtained from canine the late sodium current flowing during the plateau phase.123
cardiac tissue from different locations within the wall of the Pimozide, which is associated with inhibition of HERG
ventricular myocardium have shown a correlation be- and risk of TdP,42also blocks I , and in rat papillary mus-
CaL
tween the density of I and the brief repolarization that cles and ventricular myocytes it reduces both stimulation-
to1
occurs before the plateau phase of the AP (the phase 1 induced twitch tension and intracellular Ca2(cid:2) transients,
“notch” – see figure 2),141,142suggesting that it is the reason respectively.148 Several psychotropic drugs including chlor-
that APs in some species in limited parts of the my- promazine, haloperidol and imipramine are associated with
ocardium have a “spike and dome” morphology. In the block of I in cardiac myocytes,149showing use-dependence
Na
mouse, the use of genetic methods to knock out K(cid:2)chan- caused by a higher affinity of the drugs for the inactivated
nel subunits mediating I results in profound changes in- than for the resting state of sodium channels, and by a very
to1
©2003 Lippincott Williams & Wilkins 65
Witchel and Associates Journal of Clinical Psychopharmacology • Volume 23, Number 1, February 2003
slow repriming of the drug-bound sodium channels from in- means of mitigating QT prolongation associated with
c
activation. Such effects could be antiarrhythmic or cardio- TCAs or antipsychotic drugs.152
toxic, dependent on the health (e.g. CAD, or post MI) of the While current screening at the cellular level for risk
myocardium. of repolarization anomalies is focused on HERG or I ,
Kr
The net effect on APD of psychotropics that affect much work needs to be done to characterise this system for
overlapping inward and outward currents will therefore in vitro in vivo scaling, and further understanding of the
depend on the overall balance between such currents dur- complex relationship between cellular electrophysiology
ing the AP plateau (Table 3), and their relative sensitivity and QT prolongation is likely to involve an entire battery
to the particular agent in question; for many psychotropic of electrophysiological measurements, rather than of just
drugs the overall effects of the altered balance of currents HERG. The potency of drug-induced block of I does not
Kr
has not yet been determined. show a clear correlation with the risk of sudden death in
patients with acquired LQTS.19 In isolated feline hearts
Psychotropics, QT prolongation, and haloperidol lengthens the QT interval significantly more
arrhythmogenesis: future directions than sertindole,153yet (albeit in different heterologous sys-
The use of cellular electrophysiology has been a crit- tems) sertindole is 100–300 times more potent as a blocker
ical adjunct to measurements of the ECG from healthy an- of HERG48 than haloperidol.34 In addition, thioridazine
imals for the assessment of mechanistic risk during drug and chlorpromazine have similar potencies for blockade of
design. Furthermore, cellular electrophysiology has re- HERG,19 yet in stepwise regression analysis of 495 psy-
vealed the mechanism for an important intervention in ac- chiatric patients, thioridazine and TCAs were independ-
quired LQTS. HERG is highly sensitive to extracellular ently associated with QT prolongation, but chlorpro-
c
K(cid:2) concentrations,130 and repolarization abnormalities, mazine was not.6 Assuming that these differences are not
caused by both HERG-linked inherited LQTS150,151 and attributable to systematic errors,19 these data would sug-
quinidine-induced acquired LQTS have been corrected by gest that there exists one or more drug-induced mecha-
raising serum K(cid:2). If the potency of particular psychotropic nisms that makes the heart vulnerable to (or protected
drugs as HERG blockers can be reduced by raising exter- from) QT prolongation by blockade of the HERG potas-
c
nal K(cid:2) (as observed in vitro with amitriptyline18), then sium channel. If this is the case, the most dangerous mod-
modulating serum K(cid:2) concentrations may also offer a ulating factors for QT prolongation would appear not to
c
TABLE 3. Selected sites of drug actions and associated IC ’s
50
Drug Class Action IC ((cid:7)M) Model system
50
Imipramine TCA HERG blockade - (I 3.4 CHO cells*
out
Imipramine TCA I blockade - (I 4 Isolated rat ventricular myocytes
CaL in
Imipramine TCA 5-HT uptake blockade 0.6 Crude rat brain synaptosomes
Imipramine TCA Metabolised (aromatic 2-hydroxylation) K (cid:1)25 Human liver microsomes
m
Imipramine TCA I blockade - (I 3 Isolated guinea pig ventricular myocytes
Na in
Fluoxetine SSRI HERG blockade - (I 1.5 HEK cells*
out
Fluoxetine SSRI I blockade - (I 2.8 Isolated rat ventricular myocytes
CaL in
Fluoxetine SSRI 5-HT uptake blockade 0.075 Rat spinal cord synaptosomes
Fluoxetine SSRI Metabolised to norfluoxetine K (cid:1)33 Human liver microsomes
m
Fluoxetine SSRI Inhibits CYP 2D6 K (cid:1)3 Human liver microsomes†
i
Thioridazine Pheno. Inhibits HERG - (I 1 tsA201 cells
out
Thioridazine Pheno. Inhibits CYP 2D6 K (cid:1)1.4 Human liver microsomes‡
i
Thioridazine Pheno. Dopamine D2 receptor K (cid:1)0.01¶ Pig anterior pituitary
i
Haloperidol Butyro. Inhibits HERG - (I 1 Xenopus oocytes*
out
Haloperidol Butyro. I blockade - (I EC (cid:1)20 PC12 cells
CaL in 50
Haloperidol Butyro. I blockade - (I 5 Isolated guinea pig ventricular myocytes
Na in–
Haloperidol Butyro. Dopamine D2 receptor K (cid:1)0.0004 Cloned human receptor
i
Haloperidol Butyro. Inhibits CYP 2D6 K (cid:1)7.2 Human liver microsomes‡
i
Comparison of IC ’s for actions of selected drugs. Note that the in vitromodel systems tested in the literature are diverse and may not be directly
comparable due to acc50ess of drug site of action, intracellular solutions, etc. (I (cid:1)decreases outward repolarising current. (I (cid:1)decreases inward
depolarising current. Pheno. (cid:1)phenothiazine antipsychotic. Butyro. (cid:1)butyropouhtenone antipsychotic. in
*Heterologous system in which channel was added using molecular techniques.
†Using desipramine as a substrate.
‡Using O-demethylation of dextromethorphan as a substrate.
¶Using 3H-spiperone as a ligand.
66 ©2003 Lippincott Williams & Wilkins
Journal of Clinical Psychopharmacology • Volume 23, Number 1, February 2003 Psychotropic Drugs and Arrhythmias
be associated with chlorpromazine but to be positively as-
sociated with TCAs and thioridazine.
QT interval prolongation is not an invariable pre-
cursor to TdP,7and there is no direct evidence that the ex-
tent of drug-induced QT lengthening is associated directly
c
with an increased risk of TdP or sudden death.154–157In an
analysis of over 1000 family members in LQTS families
enrolled in the International Registry who had a normal
((cid:9)440 ms) QT interval, 6% of them experienced syncope
or cardiac arrest.158 In contrast, the HERG mutation
A561V, a mutational hot spot for congenital LQTS origi-
nally associated with high penetrance, has been found to
be associated in some families with a penetrance as low as
25%,159 such that some family members have the genetic
mutation but lack a prolonged QT or any of the other clin-
c
ical signs. This suggests that the variation in clinical phe-
notype is not simply a result of variation in known ion
channel mutations, but may also depend on other putative
genetic factors, which can vary from family to family. It is
feasible that environmental triggers might also play a role.
It seems likely that some asymptomatic gene carriers (most
of whom are over 40 years of age) may be predisposed to
the possible occurrence of drug-induced TdP.159 Con-
versely, acquired LQTS may be a forme frusteof LQTS or
a similar genetic repolarization abnormality,121appearing
as an atypical form of the genetic disorder in which the full
symptoms (i.e. QTc prolongation) do not appear until a FIG. 4. Factors influencing arrhythmogenic risk. This dia-
QT-prolonging drug is administered.134,160 gram illustrates the influence of heart rate, HERG blockade,
beta adrenergic activity, and calcium channel blockade within
Sympathetic modulation of AP prolongation by K(cid:2) the myocardium upon the pathway from AP prolongation to
arrhythmogenic risk. Small upward arrows ()) indicate an in-
channel inhibition crease in levels, whereas small downward arrows (() indicate
One further potential mechanism that may affect vul- a decrease. Larger arrows suggest the pathway indicated by
nerability to I -inhibition-induced arrhythmias may be a lo- the arrow is increased, whilst blunt (T shaped) arrows indicate
Kr
cal imbalance in cardiac autonomic activity.161,162In patients that the pathway is attenuated (e.g. an increased heart rate
shortensthe cardiac ventricular action potential).
with CAD, increased sympathetic tone can be a causal fac-
tor in the generation of ventricular arrhythmias163 (Fig. 4).
Both depression164and schizophrenia165,166are known to in- tonin, which has been associated with a reduction in sympa-
crease plasma norepinephrine (NE) levels, and it is possible thetic activity,169leads to a protective increase in the thresh-
that psychotropic-induced arrhythmogenic risk may be aug- old for both extrasystoles and ventricular fibrillation in
mented by psychotropic-induced blockade of NE reup- whole animal models.170–173 This finding may help to ex-
take.167 For example, desipramine (a TCA associated pri- plain the much higher fatal toxicity index of desipramine
marily with NE reuptake blockade, and that does not lower compared with clomipramine (a TCA associated primarily
NE when administered chronically164) had almost twice the with serotonin reuptake blockade),9 and may also explain
incidence of QT prolongation compared to other TCAs in the qualitatively higher incidence of QT prolongation in pe-
c c
one meta-analysis; in that meta-analysis the averageQT in- diatric patients associated with desipramine compared to
c
terval for desipramine was similar to the other TCAs,89sug- other TCAs.89
gesting that there may exist a subgroup of particularly sus- However, adrenergic stimulation is mechanistically
ceptible patients. In anesthetized dogs, desipramine has been associated with opposing effects on arrhythmogenic risk
shown to potentiate the pressor, chronotropic, inotropic and (Fig. 4), in that the resulting increased heart rate will tend
coronary sinus blood flow responses to NE infusions,168sug- to shorten the APD and lower arrhythmogenic risk, while
gesting that psychotropics that inhibit the reuptake of NE the increased Ca2(cid:2) mobilization will tend to potentiate
may increase the influence of spontaneous physiological in- EADs and increase arrhythmogenic risk. The resultant ef-
creases in sympathetic activity. By contrast, increasing sero- fect on arrhythmogenic risk of these two opposing mecha-
©2003 Lippincott Williams & Wilkins 67