Pharmacology & Therapeutics 119 (2008) 199–209

Contents lists available at ScienceDirect

Pharmacology & Therapeutics

j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / p h a r m t h e r a

Preclinical Torsades-de-Pointes Screens: Advantages and limitations of surrogate and

direct approaches in evaluating proarrhythmic risk
Gary A. Gintant ⁎
Department of Integrative Pharmacology, Abbott Laboratories (Dept. R46R, Bldg AP-9), 100 Abbott Park Road, Abbott Park, IL 60064-6119, United States

A R T I C L E I N F O A B S T R A C T

Keywords: The successful development of novel drugs requires the ability to detect (and avoid) compounds that may
Torsades-de-Pointes provoke Torsades-de-Pointes (TdeP) arrhythmia while endorsing those compounds with minimal torsado-
QT genic risk. As TdeP is a rare arrhythmia not readily observed during clinical or post-marketing studies,
Preclinical studies numerous preclinical models are employed to assess delayed or altered ventricular repolarization (surrogate
hERG
markers linked to enhanced proarrhythmic risk). This review evaluates the advantages and limitations of
Proarrhythmia model
Delayed repolarization
selected preclinical models (ranging from the simplest cellular hERG current assay to the more complex in vitro
perfused ventricular wedge and Langendorff heart preparations and in vivo chronic atrio-ventricular (AV)-
node block model). Specific attention is paid to the utility of concentration–response relationships and
“risk signatures” derived from these studies, with the intention of moving beyond predicting clinical QT
prolongation and towards prediction of TdeP risk. While the more complex proarrhythmia models may be
suited to addressing questionable or conflicting proarrhythmic signals obtained with simpler preclinical
assays, further benchmarking of proarrhythmia models is required for their use in the robust evaluation of
safety margins. In the future, these models may be able to reduce unwarranted attrition of evolving compounds
while becoming pivotal in the balanced integrated risk assessment of advancing compounds.
© 2008 Elsevier Inc. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
2. General considerations (and limitations) of using surrogate markers to evaluate proarrhythmic risk . . . . 200
3. hERG assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
4. In vitro arterially-perfused left ventricular wedge preparations . . . . . . . . . . . . . . . . . . . . . 202
5. In vitro Langendorff perfused isolated heart preparations . . . . . . . . . . . . . . . . . . . . . . . . 203
6. A comparison of in vitro proarrhythmia assays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
6.1. Effects of experimental conditions on the sensitivity of the Langendorff rabbit heart model . . . . . 204
7. In vivo chronic AV-blocked heart models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
8. Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207

1. Introduction
The pharmaceutical industry and regulators are under increasing
pressure to quickly develop novel, effective, and safe therapeutics. One
vexing problem that plagues drug development is that of avoiding
compounds with a propensity of eliciting Torsades-de-Pointes (TdeP),
⁎ Tel.: 847 935 1688; fax: 847 938 5286.
E-mail address: Gary.Gintant@Abbott.com.
a rare arrhythmia that can lead to ventricular fibrillation and sudden
cardiac death. Since becoming a prominent safety issue with such
drugs as the antianginal agent prenylamine and the antihistamine
terfenadine in the late 1980's and early 1990's, at least ten drugs have
been removed from the market, with warnings added to the product
labels of others due to issues of cardiac proarrhythmia. Such actions
demonstrate the urgent need to avoid this safety hurdle while pro-
ducing novel therapeutic agents.
TdeP is a rare arrhythmia not readily observed during clinical trials
or post-marketing surveillance with most drugs. For example, the

0163-7258/$ – see front matter © 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.pharmthera.2008.04.010
200 G.A. Gintant / Pharmacology & Therapeutics 119 (2008) 199–209
estimated incidence of adverse cardiac events with cisapride is about 1
in 111,000 prescriptions (Malik & Camm, 2001). Most drugs linked to
TdeP are associated with delayed or altered repolarization, manifest
clinically as prolongation of the QT interval on the surface ECG.
Consequently, regulatory guidances (such as the E14-Guidance, (E-14
Guidance for Industry (2005a))) expect a rigorous clinical study to
evaluate the ability of drug candidates to prolong the QT interval in
humans (a so-called thorough QT study). Such studies are designed to
provide robust and precise estimates of QT prolongation with
therapeutic (and supratherapeutic) drug exposures. These studies are
also expensive in terms of resources and additional development time.
Despite the key role they have assumed in drug development, it is
recognized that QT prolongation remains a surrogate marker for
proarrhythmia. Preclinical and clinical data suggest that there is no
fixed relation between the extent of QT prolongation and the risk of TdeP
(see Belardinelli et al., 2003; Roden, 2004; Shah, 2005a, 2005b, 2005c).
The purpose of this article is to provide perspectives on the
strengths and limitations of selected preclinical proarrhythmia models
used to evaluate torsadogenic risk. For comparison purposes, we start
first with a discussion of the simplest in vitro surrogate marker of
delayed repolarization, namely the hERG current assay. Subsequently,
increasingly complex in vitro models are introduced, including the in
vitro ventricular wedge and Langendorff-perfused rabbit heart
preparations, and the in vivo chronic AV-node blocked model. Specific
attention is provided to discerning the utility of quantitative
(concentration–response) relationships to guide drug discovery and
avoid TdeP risk. This review is limited to studies of acute electro-
physiologic drug effects, and does not consider drugs that may have
more long-term effects on repolarization, such as those that alter hERG
channel trafficking and channel density (see Kuryshev et al., 2005;
Dennis et al., 2007). Available data suggests that the more complex
proarrhythmia models are presently most useful to resolve conflicting
or questionable proarrhythmic signals generated in simpler assays.
Further benchmarking of proarrhythmic models (required for more
precise estimates of model sensitivity and specificity) may support
their role in reducing attrition of advancing compounds and a place in
the integrated risk assessment for cardiac safety.
2. General considerations (and limitations) of
using surrogate markers to evaluate proarrhythmic risk
It is instructive to consider screening assays for TdeP based on the
endpoints acquired with the different models, namely, either surrogate
(or indirect) markers of TdeP or direct measures of the incidence of
TdeP (see Table 1). Examples of surrogate endpoints used to evaluate
drug effects range from the deceptively simple IC50 value for hERG
current block to characterizing changes in the duration and shape of
the cardiac action potential (from tissues or multiple cardiac regions)
to changes in the configuration of T waves from ECG (or ECG-like)
recordings. Only more complex models can be used to directly measure
the incidence of TdeP, as one cannot elicit TdeP using a single channel
Table 1
Summary of parameters typically evaluated in preclinical “QT” models
Model Parameters measured or emphasized
Preparation Curr. Repolariz. EADs Triang.
hERG Channel X
APD Repolarization Tissue APD X X
QT measures Animals
Wedge Tissue slice APD X X
Langendorff Organ (Hearts) MAPD X X
Chronic AV-block Animals MAPD X X
Models are arranged according to increasing complexity. Measures of the incidence of TdeP (rightmost column) is only one of eleven parameters evaluated in the six selected assays.
Repolariz. = repolarization; EADs = early afterdepolarizations, Triang. = triangulation, QT Morph. = QT morphology (Tpeak–Tend), MAPD = monophasic action potential duration,
Disper. = APD dispersion, ADP = transmembrane action potential, TdeP = Torsades-de-Pointes.
expression system, myocytes, or (arguably) a wedge of ventricular
tissue. As would be expected, more complex assays provide for greater
numbers of surrogate markers for evaluation (ion channel b myocytes b
tissue b organ b animal). Multiple surrogate endpoints complicate the
evaluation of torsadogenic risk, as weighting factors must be assigned
to the different markers in order to provide “risk signatures” that
match preclinical and clinical perceptions. The relative clinical risk of
TdeP is also difficult to discern, due to the rare incidence of TdeP. In
addition, multiple additional factors (for example, drug–drug interac-
tions that may prominently increase drug concentrations) may
contribute to the perceived clinical risk independent of direct effects
on myocardium. Finally, different definitions of TdeP are often used
with comparing TdeP in preclinical models with the clinical experi-
ence. For example, a clinical definition of TdeP cited by Napolitano et al.
(1994) (“progressive twisting of QRS axis around an imaginary base-
line, complete 180 degree twist, and markedly prolonged QT interval in
the last sinus beat preceding the onset of arrhythmia” (Drugs)) is very
different than that cited for ventricular wedge preparations by Shimizu
and Antzelevitch (1998) (“programmed electrical stimulation-induced
polymorphic ventricular tachycardia displaying characteristics of
TdeP”(Circ )). Together, these above considerations make quantitative
comparisons of concentration–response relationships across preclini-
cal and clinical studies difficult, and highlight the need for wide safety
margins derived from preclinical studies.
It is generally accepted that TdeP results from the culmination of
multiple factors acting synergistically to initiate (and sustain) this
arrhythmia. TdeP was initially considered an idiosyncratic drug reac-
tion (that is, an adverse drug reaction that does not occur in most
people at doses used clinically), and undoubtedly some cases of drug-
induced Torsades are likely idiosyncratic (for example, those involving
genetic mutations affecting cardiac potassium channels). However,
most cases can be attributed to the contributions of multiple risk
factors that include a drug (pharmacodynamic) effect (Roden, 1998;
Zeltser et al., 2003; DeBruin et al., 2005). Consequently, it would be
expected that the most useful surrogate marker(s) to interrogate
would be those most frequently linked to TdeP and that also play a key
role in the initiation of TdeP, with other markers assuming lesser (but
still important) roles. Given that multiple risk factors are recognized to
play a role in the initiation of TdeP (female gender, bradycardia,
hypokalemia, hypomagnesia, and structural heart disease), it is not
surprising that simple surrogate markers do not provide 100%
sensitivity or specificity with regards to predictability of clinical
outcome (TdeP). Thus, in order to provide a robust quantitative
(concentration–response relationship) of the TdeP risk, it is imperative
to understand 1) the role and limitations of each surrogate marker, and
2) the influence (which may be dynamic and nonlinear) of contributing
risk factors on the predictability of the surrogate marker under study.
The identification of discrepancies between preclinical “signals” and
clinical findings, while highlighting less than perfect assay sensitivity
or specificity, represents an important first step towards under-
standing the limitations of any surrogate marker.
Reverse use-dep. Instability/variability Disper. Pseudo-ECG-QT ECGQT QTMorph. TdeP(like)
X
X
X X X
X X X X X
X X X
 G.A. Gintant / Pharmacology & Therapeutics 119 (2008) 199–209
3. hERG assay
IKr is a voltage- and time-dependent outward (repolarizing)
current that is activated upon depolarization and plays a prominent
role in triggering final repolarization in ventricular tissue (Sanguinetti
et al., 1995). In humans, iKr is encoded by hERG, and is typically
referred to as hERG current. Numerous studies have demonstrated
that hERG block may delay repolarization and prolong the QT interval
on the ECG. For most drugs, block of hERG current results from drug
binding within the channel pore after gaining access from the cyto-
plasmic pool (for a review, see Sanguinetti & Tristani-Firouzi, 2006).
Virtually all drugs associated with TdeP block iKr/hERG current
(Roden, 2004).
It is now relatively easy to study drug effects on hERG current using
voltage clamp techniques and either manual patch or automated
planar patch techniques applied to heterologous expression systems.
In its simplest form, block is evaluated based on changes in the am-
plitude of the deactivating tail current measured upon repolarization
(preceded by a depolarizing square pulse typically applied at slow
frequencies). Block potency is typically compared based on IC50 values
derived from fitting concentration–response curves; Hill coefficients
for derived fits and confidence intervals should also be provided.
Testing an accepted standard to benchmark assay sensitivity is essen-
tial, as is measuring bath concentrations; the later is crucial to prevent
underestimating block potency resulting from reduced bath concen-
trations (resulting, for example, from drug absorption to tubing).
Experimental conditions (for example, temperature) as well as
different voltage clamp protocols (different amplitude and duration of
depolarizing square pulses, square vs. repolarizing ramp clamp pulses
[to approximate ventricular repolarization]) may affect IC50 determi-
nations. While comparable values IC50 values are attained using such
varied conditions with most drugs, a few compounds provide sub-
stantially different values (Kirsch et al., 2004; Yao et al., 2005). More
recent studies have highlighted the effect of stimulation rate on hERG
current block (Stork et al., 2007). Modulation of block by stimulation
rate likely relates to unique kinetics of drug binding and unbinding to
the channel. These considerations argue for a wide therapeutic win-
dow if calculations are based solely on hERG current measures.
For insoluble compounds, the hERG assay is typically conducted
with concentrations to the limit of solubility. The hERG assay should
be conducted in the absence of plasma proteins, since drug binding
within the myocardial cell near the hERG channel is unknown. This
practice also avoids generating parallel datasets of IC50 values in the
absence and presence of plasma proteins. In the case of highly plasma
protein bound compounds, “corrected” IC50 reflecting calculated
(equivalent) total plasma drug concentrations are sometimes pro-
vided. However, it should be recognized that plasma protein binding
varies with species, is not necessarily linear across drug concentra-
tions, and represents only one factor that determines drug distribu-
tion to the myocardium. Controversy exists in regards to “correcting”
for plasma protein binding with in vitro studies. In a hERG study with
a series of five antipsychotic drugs with high plasma protein binding
(83–99%), the ratio of hERG IC50 values to total plasma concentration
corresponded well with QT prolongation, while free (unbound) drug
plasma levels did not provide a reliable correlation (Kongsamut et al.,
2002). In contrast, the ratio of hERG IC50 values to free plasma con-
centrations was predictive of QT prolongation for seven fluoroquino-
lone antibiotics with weak plasma protein binding (range 20–30%,
Kang et al., 2001). In an in vitro repolarization study using canine
Purkinje fibers, plasma protein binding did reduce APD prolongation
seen with bound drugs, but the extent of the effect was not well
correlated with the calculated clinical free fraction (Limberis et al.,
2007). Some drugs have been shown to accumulate in cardiac tissues;
cardiac tissue/plasma ratios have been reported for clozapine, (2.2),
olanzapine (2.7), sertindole (3.1), risperidone (4.5), and haloperidol
(6.4) (Titier et al., 2004). These authors suggested that with ratios
201
higher than 4, arrhythmias may be expected. For drugs shown to
appreciably accumulate in myocardium, it may be difficult to translate
hERG block measured in simple CHO or HEK cell expression systems to
electrophysiologic effects in vivo.
IC50 values for hERG block are typically compared to predicted (or
actual) clinical Cmax values in order to characterize safety margins for
delayed repolarization. Redfern et al. (2003) have proposed that an
IC50 value for hERG block 30-fold greater than the maximal calculated
unbound effective therapeutic plasma concentration (ETPCunbound)
provides a reasonable, balanced compromise for delineating torsado-
genic risk and calculating safety margins. These authors noted excep-
tions to this rule (see below for some examples) and also commented
that this margin should be increased for future drug discovery pro-
grams and for drugs aimed at non-debilitating diseases. Safety
margins based on IC20 values, while providing for greater sensitivity,
also are likely more variable due to the reduced slope in that range of
the concentration–response relationship.
Not all drugs that block hERG current result in delayed repolariza-
tion. For example, verapamil is not associated with either QT prolon-
gation or TdeP despite blocking hERG at concentrations comparable to
therapeutic (total) plasma concentrations. This result is most easily
attributed to concomitant block of L-type (inward) calcium current
mitigating hERG current (Zhang et al., 1999). Fluoxetine and citalopram
are other examples of HERG blocking drugs expected to reduce hERG
current at (or moderately above) therapeutic concentrations that do
not typically affect QT prolongation or TdeP clinically (Witchel et al.,
2002; Fossa et al., 2007). It has been suggested that fluoxetine would
not have been developed in the 1970's if the hERG current assay were
in place (Fermini & Fossa, 2003). In general, these exceptions may be
attributed (at least in part) to drug effects on non-hERG cardiac
channels that may mitigate or obscure hERG block (Bril et al., 1996, see
also Gintant et al., 2006). Such effects, termed multi-channel block
(Martin et al., 2004), are readily apparent when comparing hERG block
and delayed repolarization with high content cellular assays or tissue
repolarization studies. Fig. 1 compares the effects of 10 drugs on hERG
current block and prolongation of canine Purkinje fiber action
potential duration. In these experiments, identical concentrations of
each drug were tested in both assays. With drugs such as E-4031 and
Fig. 1. A comparison of APD prolongation and hERG current inhibition with ten
structurally diverse drugs. Plotted are changes in APD obtained from (abscissa) versus
hERG current block (ordinate). Matching concentrations of each drug representing
therapeutic values and higher multiples (see Martin et al., 2004) were tested in each
assay. Nine of 10 drugs demonstrated prominent concentration-dependent hERG block
at supratherapeutic concentrations; however, only 4 demonstrated convincing
monotonic concentration-dependent APD prolongation with increasing hERG block
(E-4031, erythromycin, moxifloxacin, and telithromycin). Comparable levels of hERG
block did not predict the extent of APD prolongation. Values represent means of a
minimum of 6 determinations per drug for each assay.
202 G.A. Gintant / Pharmacology & Therapeutics 119 (2008) 199–209
moxifloxacin, there is a clear increase in APD with increased hERG
current block. We call these drugs overt hERG blockers. In contrast,
increasing hERG block is not associated with increasing APD prolonga-
tion for such drugs as haloperidol and fluoxetine. We call these drugs
covert hERG blocking agents. Cisapride represents a special case in that
increasing drug concentrations initially prolong and then shorten the
APD (a bell-shaped concentration–response curve). This effect is likely
due to block of L-type calcium current only at higher drug concentra-
tions. Such studies emphasize that hERG block represents only one
possible ion channel (off-target) effect. Clearly, there is a need to
mechanistically define the level of torsadogenic risk with covert hERG
blocking agents. As hERG is presently positioned in lead optimization to
differentiate (and eliminate) compounds early in drug discovery, the
number of compounds wrongly excluded based solely on the early
application of hERG screening assays is difficult to determine.
Given results from a well-conducted hERG study one would expect
it possible to estimate concentration-dependent effects on QTc
clinically (and hence, therapeutic margins for delayed repolarization)
based upon comparisons between IC50 values for block and therapeutic
drug concentrations. For some specific hERG blocking drugs this may
be the case. Dofetilide is a drug used to treat atrial arrhythmias that is
also a potent iKr blocker (IC50 values approximately 10–20 ng/mL) and
that delays ventricular repolarization. A modeling study comparing
dofetilide's effect on hERG current with results from four recent clinical
thorough QT studies demonstrates striking sensitivity of QT prolonga-
tion to hERG current block: 7% block of hERG current affects 5 msec ms
mean prolongation of QT, and 13% block affecting 20 ms prolongation
(Jonker et al., 2005). This exquisite sensitivity may be attributed (at
least in part) to specificity of iKr block by dofetilide. In contrast, the QT
prolonging response with cisapride is not as sensitive as that of
dofetilide (Wallis, 2008), likely due to multi-channel block. Obviously,
one must also consider the likelihood of additional off-target drug
effects in vivo that may indirectly affect cardiac repolarization (for
example, extracardiac alpha1-adrenergic stimulation, see Farkas et al.,
2006).
4. In vitro arterially-perfused left ventricular wedge preparations
A more complex preclinical assay used to interrogate delayed
repolarization and torsadogenic risk is the arterially-perfused ventricular
wedge preparation (Wu et al., 2005; Yan et al., 1998a, 1998b; 2001). This
model has proven useful in delineating mechanisms of proarrhythmia,
including spatial and temporal dispersion of repolarization, drug-induced
enhancement of repolarization heterogeneity (leading to transmural
dispersion), and triggered activity (early afterdepolarizations). Electro-
physiological recordings (obtained with loosely-held (“floating”) micro-
electrodes) allow for comparisons of action potentials recorded from
different myocardial layers (epi-, mid-, and endocardial regions) as well
as recordings of pseudo-ECG (derived either from either microelectrode
recordings or extracellular field electrodes for transmural signals). While
TdeP-like recordings have been published from these wedge prepara-
tions, typical surrogate markers include: delayed repolarization (QT or
APD prolongation), dispersion of repolarization (Tpeak–end/QT ratio or
TDR), and incidence of EADs. The rabbit wedge preparation is claimed
advantageous compared to the Langendorff rabbit heart (see below) in
regards to electrical stability (more than 4 h for wedge compared to lesser
times for the Langendorff perfused rabbit heart), allowing for four 40 min
drug equilibration periods (Chen et al., 2006). One challenge of this model
(or any model evaluating multiple endpoints) is to correctly align the
measured parameters to optimize sensitivity and specificity when
defining proarrhythmic risk.
With wedge preparations, particular attention is often focused on
greater drug-induced prolongation within the ventricular myocardium
(“M cells”, see for example Yan et al., 1998; Di Diego et al., 2003) that
is exaggerated at excessively slow stimulation rates. It should be noted,
however, that some reports find lesser evidence of repolarization het-
erogeneity in the in situ heart compared to wedge preparations (see
Anyukhovsky et al., 1999; Taggart et al., 2003; Ueda et al., 2004, also
Conrath & Opthof, 2006 for a review). Changes in the configuration of
the “T waves” from pseudo-ECG recordings (measured as the interval
between the peak and end of the T wave [Tpeak–Tend]) have been
proposed as an index of transmural dispersion of repolarization in this
model (Yan & Antzelevitch, 1998; Antzelevitch & Shimizu, 2002).
However, the validity of this marker has been questioned, at least
when applied to in situ preparations (Opthof et al., 2007). In the canine
wedge, transmural dispersion of repolarization is sensitive to the
stimulation site, with dispersion increasing with epicardial vs. endo-
cardial stimulation (Di Diego et al., 2003; Circ.). This effect modifies the
sensitivity of preparations to a Torsades-like arrhythmia with low doses
of cisapride, and needs to be considered when evaluating results with
this model.
In a blinded validation of the isolated arterially-perfused rabbit
ventricular wedge preparation, Liu and colleagues (2006) compared the
effects of 13 (8 positive, 5 negative controls) drugs with varying pro-
pensity for causing QT prolongation or TdeP. Electrophysiologic effects
were studied over a wide range of drug concentrations including those
corresponding to calculated free clinical therapeutic plasma concentra-
tions (see Fig. 2 from Liu et al., below). These investigators were also
careful to validate bath concentrations to ensure drug exposures, a factor
not considered in similar studies. Relative TdP risk assessment was based
on a scoring system evaluating QT interval prolongation, Tpeak–end/QT
ratio, and development of early afterdepolarizations with and without a
closely coupled extrasystole; the development of EAD-induced extra-
systoles received the greatest weight in calculation of TdeP risk scores.
The authors reported a high sensitivity and specificity for identifying
drugs with a potential for causing TdP. As shown in Fig. 2 (below), all
five “negative” compounds maintain low TdeP risk scores throughout the
0.1–100× range of free therapeutic plasma Cmax concentrations.
Liu et al. did note differences in the TdeP risk scores for the 8
“offender” compounds associated with either QT prolongation or TdeP
in humans, especially near concentrations 10-fold free therapeutic
plasma Cmax values. For example, in the approximate range of 7–9-
fold free therapeutic plasma Cmax concentrations, the composite risk
score ranking was terfenadine b moxifloxacin b sparfloxacin; moxiflox-
acin demonstrated greater TdeP risk scores than did terfenadine over
the entire 0.1–100+-fold range of free therapeutic plasma Cmax
concentration multiples. It is debatable as to whether this ranking
reflects the intrinsic ability of these drugs to elicit TdeP clinically
Fig. 2. Effect of thirteen compounds in a rabbit ventricular wedge preparation. Eight
compounds chosen on the putative basis of ability to prolong QTc and/or induce TdeP in
humans (closed symbols) increase TdP risk score, while all five drugs associated with
cardiac safety in clinical usage (open symbols) do not affect the TdP risk score. From Liu
et al. (2006).
 G.A. Gintant / Pharmacology & Therapeutics 119 (2008) 199–209
based solely on direct cardiac electrophysiologic activity (pharmaco-
dynamic factors), or whether the clinical experience is biased by the
possible range of high exposures achieved clinically with both com-
pounds (pharmacokinetic factors).
For each of the five tested drugs recognized clinically for cardiac
safety (verapamil, clozapine, captopril, fexofenadine, and desipra-
mine), there was a clear lack of TdeP risk (open symbols, TdP risk
scores consistently below zero at up to 100× the free therapeutic
plasma Cmax). Interestingly, of the 5 “safe” compounds tested, only
fexofenadine had a hERG ratio (defined as hERG IC50/free therapeutic
plasma concentration and calculated based on values from Table 2 of
Liu et al.) greater than 30 (respective values of 20, 5, and 1.75 are
calculated for desipramine, clozapine, and verapamil). Despite these
low margins, all five did not demonstrate enhanced TdeP risk scores.
These results suggest that this model provides evidence of safety of
hERG blocking drugs with relatively low hERG safety margins.
In a separate study also published in 2006, Chen and colleagues
independently validated the arterially-perfused rabbit left ventricular
wedge model and examined its use in predicting proarrhythmic
risk. This work described results obtained with a set of six drugs
(dofetilide, DL-sotalol, cisapride, risperidone, and moxifloxacin as posi-
tive controls, fluoxetine as a negative control), along with a fairly detailed
methodological description of the preparation. In this study, the scoring
scheme was based on QT interval prolongation, Tpeak–Tend, and the
appearance of phase 2 EAD's (comparable to a scheme employed by Liu
et al., 2006, above). Overall, this model could distinguish between the
negative and positive control compounds across a minimum 30-fold
range of concentrations encompassing calculated free Cmax values
(see Fig. 3, below). Concentration-dependent increases in arrhyth-
mogenic scores was noted for each of the five positive controls.
Dofetilide, DL-sotalol, cisapride, risperidone and moxifloxacin in-
creased endo- and epicardial APD90, QT interval and TP-E (peak-to
end time of the T wave) in a reverse use-dependent manner within
clinically relevant concentration ranges. Phase 2 early afterdepolar-
izations (EADs) were observed at 1.6, 2.3, 16.7, 37.5 and 7.9 fold,
respectively, their corresponding unbound therapeutic concentra-
tions. Based on multiples of free Cmax values, ranking of TdeP risk was
sotalol = dofetilide N moxifloxacin Nrisperidone = cisapride. Somewhat
surprisingly, at approximately four-fold free Cmax concentrations,
moxifloxacin demonstrated a 2-fold greater proarrhythmic score com-
pared to cisapride (though both fall below the indicated “threshold”
that would indicate a high risk for proarrhythmia).
Fig. 3. Arrhythmogenic scores of drugs in relation to their clinically relevant (calculated
free) concentrations in a rabbit ventricular wedge preparation. Drugs tested included (from
left to right);DL-sotalol (squares), dofetilide (circles), moxifloxacin (left triangles), cisapride
(up triangles) = risperidone (down triangles), and fluoxetine (diamonds). The dashed box
indicated threshold for alarm that was obtained by averaging the scores of preparations
demonstrating EADs; the lower line reflects the greater sensitivity of cisapride to initiate
EADs (observed in 2 of 4 preparations with low (2%) QT prolongation). It should also be
noted that near the threshold scoring point substantial increases in average QTc values
(89%) and Tpeak–end (211%) were observed. Adopted from Chen et al. (2006).
203
Moxifloxacin and cisapride are the only two compounds tested by
both Lu et al. and Chen et al. in wedge preparations. The Chen study
differentiates moxifloxacin as having higher proarrhythmic risk
compared to cisapride over the entire range of Cmax multiples tested.
In contrast, the Liu study ranks the two drugs equal at multiples up to
9-fold Cmax; above this multiple, risk with cisapride increase 5-fold,
while scores with moxifloxacin remain essentially constant. It should
also be noted that cisapride produced EAD's in the rabbit model at
concentrations near 100 nM, while no EAD's were observed at higher
concentrations (5 μM) in the dog wedge preparation (Di Diego et al.,
2003). These comparisons demonstrate the possible variability across
laboratories and species when attempting to quantify proarrhythmic
risk, and highlights the need for benchmarking of each specific model
with known standards to demonstrate assay sensitivity.
5. In vitro Langendorff perfused isolated heart preparations
The Langendorff perfused isolated rabbit heart is perhaps the most
widely used in vitro preparation used to evaluate drug-induced
delayed repolarization and proarrhythmia. This model extends studies
pioneered by Carlson et al. with the in vivo methoxamine-sensitized
rabbit heart proarrhythmia assay used to assess proarrhythmic effects
of Class III antiarrhythmic agents (Carlsson et al., 1993). Developed on
an automated platform by Hondeghem, the SCREENIT Langendorff
assay measures ventricular monophasic action potential recordings
from endocardial and epicardial surfaces (as well as pseudo-ECG's)
obtained from methoxamine-pretreated female rabbit hearts subject
to escalating drug concentrations (see Hondeghem et al., 2001;
Hondeghem & Hoffmann, 2003 and earlier references contained
therein). Dispersion of repolarization (measured as difference in
monophasic action potential duration recorded from ventricular
endocardial and epicardial surface) may also be measured as an
index of proarrhythmia. Complex pacing protocols (encompassing
rapid and slow stimulation periods) are also employed, and transient
changes in APD in response to rate changes are monitored. Such
measures provide for an evaluation of repolarization instability,
defined as difference between upper and lower quartile estimates of
APD values of MAP recordings from a sequence of 60 action potentials.
Instability is one of the components of “TRIaD”, an acronym used to
describe multiple drug-induced electrophysiologic effects; T refers to
triangulation of the action potential configuration, R refers to reverse
use-dependence, I refers to instability of repolarization and D refers to
dispersion of repolarization (added in later versions of SCREENIT). Each
of the elements of TRIaD have been implicated in the genesis of
proarrhythmia (Shah & Hongehem, 2005, see also Millberg et al.,
2004), and may act synergistically. As with other assays utilizing
multiple parameters as endpoints, it was unclear as to how to
optimally rank factors towards a composite score when assessing
proarrhythmic risk.
In an early blinded validation of this assay using 14 drugs (8 asso-
ciated with QT prolongation clinically, 3 linked to QT shortening, the
remaining 3 with no QT duration effects), SCREENIT identified
potentially proarrhythmic drugs at selected concentrations based on
measures of repolarization instability, triangulation, and/or reverse
use-dependence (Hondeghem & Hoffmann, 2003); none of the drugs
considered as safe were labeled as proarrhythmic. In some cases,
however, concentrations necessary to identify proarrhythmic risk
greatly exceeded therapeutic concentrations. For example, terfena-
dine demonstrated effects only at high free drug concentrations
(5 mcg/mL), much higher than those encountered clinically either in
the absence or presence of metabolic inhibition in humans (approx
13 ng/mL total [0.4 ng/mL calculated free concentration], associated
with a mean QTc prolongation of 42.4 msecms, Abernethy et al., 2001).
In a separate validation study, 55 blinded compounds (grouped
according to the five torsadogenic categories following the outline of
Redfern et al., 2003) were subjected to a rigorous quantitative analysis
204 G.A. Gintant / Pharmacology & Therapeutics 119 (2008) 199–209
using SCREENIT to evaluate it's predictive value for TdeP liability
(Lawrence et al., 2006). Compounds were evaluated using four or five
escalating concentrations spaced at either full or half-log concentra-
tions at multiples of 1–300 the highest effective free therapeutic
clinical plasma concentration (EFTPCmax). Of eight drugs generally
associated with high TdeP risk (Category 1), two false negative
compounds (azimilide and disopyramide) were detected. Of 21 drugs
generally considered free of TdeP risk, 3 false positives (moxifloxacin,
vardenafil, and risperidone) were determined. For drugs comprising
the middle three categories (representing graded levels of putative
proarrhythmic risk), the model was described as “less precise”.
Within this Lawrence study, a subset analysis was performed for
seven drugs that spanned the five TdeP risk categories described by
Redfern yet demonstrated an approximate 10-fold hERG IC50/
EFTPCmax ratio. Using hERG potency as a filter, a clear separation
between Category 1 (procainamide, most risk example) and Category 5
(ketoconazole, least risk example) risk scores was apparent across the
1–100 multiple range of EFTPCmax. In contrast, no clear separation
between drugs in the intermediate risk categories (2, 3, and 4) was
apparent; domperidone (Category 4) had a higher proarrhythmic score
than grepafloxacin or sertindole (Categories 2 and 3) at a 10-fold
multiple of EFTPCmax. In conclusion, this validation study by Lawrence
et al. demonstrated that with drugs taken as groups, the Langendorff
SCREENIT model could predict clinical outcome, particularly for drugs
associated with either very high or very low torsadogenic risk.
However, the model was less precise in predicting risk for drugs at
other than the extreme margins of perceived TdeP liabilities. Following
the lead of Redfern et al., the authors proposed that a 30 fold margin
between risk scores and EFTPCmax would provide confidence that a
new chemical entity should proceed through development without
incurring substantial risk of eliminating potentially beneficial drugs.
A recent study by Steidl-Nichols et al. (2008) provided further
insights into the utility and limitations of the Langendorff perfused
rabbit heart proarrhythmia model. In their investigations, a blinded set
of nine compounds of varying proarrhythmic risk (along with a
negative control) was independently evaluated in the SCREENIT model
using a series of experimental protocols that explored a), effects
measured across overlapping concentration ranges, b) increased
number of concentration ranges, c) extended perfusion times, and d)
increasing the number of experiments. Proarrhythmia scores were
derived based on those described by Lawrence et al. (2006). While the
model correctly identified proarrhythmic agents qualitatively, the
study design was shown to significantly impact the proarrhythmic risk
assessment. Of note was the finding that the concentration at which
action potential prolongation, triangulation, instability, reverse use-
dependence and ectopic beats occurred varied based on the over-
lapping concentration ranges used in the different protocols. This
result, which is likely due to inadequate equilibration time with cardiac
tissue, demonstrated the critical role of study protocol in defining
preclinical safety margins in this model. It should also be noted that
some limitations demonstrated by Steidl-Nichols et al. were previously
recognized by Hondeghem as model limitations (for example, duration
of equilibration period for steady-state effect and use of predefined,
standardized concentrations, (Hondeghem & Hoffmann, 2003). Col-
lectively, the above data suggests that a strength of the Langendorff
perfused rabbit heart assay lies in delineating proarrhythmic actions of
drugs in a more complex, integrated system (though not necessarily in
providing precise concentration–response relationships to define
proarrhythmic risk), and to aid in demonstrating (or refuting) potential
proarrhythmic effects of compounds showing prominent effects in
simpler assays (i.e., potent hERG current block).
6. A comparison of in vitro proarrhythmia assays
A study directly comparing the electrophysiologic effects of four
antibiotics (two fluoroquinolones and two macrolides) in four in vitro
assays (hERG current, Purkinje fiber repolarization, left ventricular
wedge and arterially-perfused Langendorff heart proarrhythmia
models) was recently published by Lu et al. (2007); with the exception
of the hERG assay, preparations studied were all derived from the
rabbit. The antibiotics were tested at concentrations based on total
therapeutic Cmax concentrations, and included sparfloxacin (removed
from the market for TdeP). Different rankings were found for the
antibiotics dependent on the assay employed (see Fig. 4, below). For
example, moxifloxacin and erythromycin had comparable rankings in
the hERG assay, while erythromycin demonstrated lowest score and
moxifloxacin the highest score in the Langendorff-perfused heart. The
authors suggested that the hERG current, Purkinje fiber, and isolated
Langendorff-perfused rabbit hearts could be used to assess the ability
of antibiotic compounds to delay ventricular repolarization, while the
wedge preparation appeared more predictive and suitable for detect-
ing torsadogenic effects of antibiotics. These conclusions are critically
dependent on the correct ranking of clinical (pharmacodynamic) risk.
Lu et al. also demonstrated that grepafloxacin elicited QT prolongation
in the arterially-perfused rabbit left ventricular wedge model, but
failed to do so in the Langendorff-perfused heart, emphasizing the
need to benchmark assay sensitivity across different models.
6.1. Effects of experimental conditions
on the sensitivity of the Langendorff rabbit heart model
An understanding of the sensitivity of any biological assay is
essential to understanding the utility of that model. This concept is
exemplified in two studies utilizing the in vitro female Langendorff
rabbit heart model in the absence and presence of the sea anemone
toxin (ATX-II, an agent that increases cardiac late sodium current). In
the first study, the proarrhythmic effects of moxifloxacin were
evaluated (Wu et al., 2006). ATX-II alone caused neither EAD's nor
VT, while ATX-II (3 nM) and moxifloxacin (400 ng/mL(1 μM)) together
caused spontaneous and pause-triggered EAD's, ectopic ventricular
beats, and polymorphic VT in each of seven hearts studied. This
moxifloxacin concentration is considered subtherapeutic and falls far
below the range of typical Cmax values (4 mcg/mL) measured during
“safe” thorough QT studies or following oral administration of a 400 mg
therapeutic dose (Moxifloxacin product label) and is considered
subtherapeutic. In the second study with the Langendorff rabbit
heart model, Wu et al. (2008) demonstrated that acutely administered
escalating doses of a wide range of amiodarone concentrations (1 nM–
10 μM) caused an insignificant increase in the monophasic action
Fig. 4. Summary of rank order of proarrhythmic scores of four antibiotics in four different
preclinical assays. Test concentrations of each antibiotic were chosen based on
therapeutic Cmax plasma concentrations. For drugs with approximately equal scores,
drug abbreviations were placed between rank levels (1 least liability, 4 greatest liability).
Spar = sparfloxacin, Moxi = moxifloxacin, Ery = erythromycin, Teli = telithromycin. Data
derived from Lu et al. (2007), (with clinical rankings based on Torsades.org).
 G.A. Gintant / Pharmacology & Therapeutics 119 (2008) 199–209
potential duration and did not elicit TdeP. In contrast, in the presence of
3 nM ATX-II, low amiodarone concentrations (1–30 nM) increased
transmural dispersion of repolarization, beat-to-beat variability, and
prolonged MAP duration. Higher concentration ranges (30–300 nM
(0.02–0.2 mcg/mL)) induced TdP in 16/17 hearts, while clearly
excessive concentrations (1–10 μM) abbreviated APD, decreased
beat-to-beat variability, and suppressed TdeP. It should be noted that
the higher concentrations tested (0.02–0.2 mcg/mL) represent values
12-fold lower than the high therapeutic concentration range for
amiodarone (1–2.5 mcg/mL, Jurgens et al., 2003); the demonstrated
excessive sensitivity is even greater if one considers plasma protein
binding of amiodarone (approx 95%, Jurgens et al., 2003). Amiodarone
is a drug that is recognized as prolonging QT interval but essentially
devoid of TdeP liability in man (Lazzara, 1989; Auer et al., 2002; Singh,
2006) and following acute administration in the chronic AV-blocked
dog model (van Opstal et al., 2001a; Thomsen et al., 2004). These in
vitro studies emphasize the importance of experimental conditions in
determining assay sensitivity and possibly defining false-positive
compounds when evaluating new chemical entities.
7. In vivo chronic AV-blocked heart models
Perhaps the most extensively studied in vivo large animal
proarrhythmia model is that of the chronic atrio-ventricular blocked
(CAVB) dog. In this model, the AV node is ablated, and animals
(typically dogs or rabbits) are studied after sufficient time is allowed
for bradycardia-induced electrophysiologic ventricular remodeling
leading to decreased repolarization reserve (Volders et al., 1998, 1999;
Van Opstal et al., 2001b, Tsuji et al., 2002; Gross, 2006). Conscious or
anesthetized animals may be studied, with more invasive instrumen-
tation possible with anesthetized animals. Parameters evaluated in
this model can include QT prolongation, ectopic beats, and mono-
phasic action potentials typically recorded with catheters positioned
on the left and/or right ventricular endocardium. Early afterdepolar-
izations (EADs) can be recorded from monophasic action potentials,
and interventricular dispersion calculated by comparing MAPs from
left and right ventricles. Additional stressors that have been applied
include female gender, hypokalemia (as with furosemide), high drug
concentrations (by IV or oral administration), catecholamine chal-
lenge (epinephrine), and additional potassium channel blockers to
reduce repolarization reserve (for example, iKs blockade). Being an
in vivo model, potential additional off-target drug effects (heart rate,
blood pressure changes) that may contribute to proarrhythmia may be
manifest; while off-target effects may render mechanistic interpreta-
tions more difficult, such effects may provide a more realistic
assessment of proarrhythmic risk.
The CAVB model has been useful in exploring mechanisms of TdeP,
defining novel surrogate markers, as well as providing direct measures
of the incidence of TdeP. Dose–response relationships are not typically
studied due (at least in part) to the time required for achieving steady-
state plasma (or tissue) concentrations in vivo with anesthetized,
instrumented animals. With this model, drug-induced QT prolonga-
tion has been observed without TdeP. For example, despite QT
lengthening with chronic amiodarone (21%) and augmented inter-
ventricular dispersion (difference between left and right endocardial
MAP recordings, 3 of 7 anesthetized dogs) no TdeP was observed
(consistent with clinical experience) (van Opstal et al., 2001a). The
absence of TdeP with amiodarone (despite QT prolongation) was
attributed to homogeneous lengthening of APD in most animals as
well as the lack of EAD's and ventricular ectopy. In contrast, the
noniodinated congener dronedarone elicited QT prolongation (31%),
increased interventricular dispersion in 7 of 8 dogs, and elicited TdeP
in 4 of 8 dogs with the highest interventricular dispersion. Subsequent
studies with the antipsychotic drug sertindole in CAVB anesthetized
dogs demonstrated no TdeP with a low dose (but not high dose) of i.v.
administered sertindole despite comparable increases in QTc (Thom-
205
sen et al., 2003; van Opstal et al., 2001c); ectopic beats were observed
in the high dose, as was a tendency for increased interventricular
dispersion. Three interventions to promote repolarization (infusion of
KCl, pharmacologic activation of iKATP with levcromakalim, and
steady ventricular pacing (60 bpm)) abolished TdeP without reducing
drug-induced QT prolongation. Thus, QT prolongation does not always
correlate with the risk of TdeP in this diseased-based model.
Studies with the anesthetized CAVB dog model (albeit with a
somewhat limited number of drugs and drug concentrations) have
demonstrated proarrhythmia consistent with the clinical perception of
TdeP risk. Thomsen et al. (2006) compared QT prolongation and the
incidence of TdeP with two doses of either moxifloxacin or azithro-
mycin administered i.v. using a randomized cross-over experimental
design (n = 5–6 dogs); dofetilide was used as a positive control. Mea-
sured total plasma concentrations of moxifloxacin and azithromycin
were approximately 1–6 fold and 7–30 fold the expected mean clinical
Cmax concentrations (low and high dose infusions, respectively).
Results demonstrated greater prolongation of QTc with dofetilide
(29%) compared to b) low and high exposures of moxifloxacin (7 and
21% prolongation); azithromycin demonstrated essentially no QTc
prolongation. While dofetilide and moxifloxacin elicited QT prolonga-
tion, dofetilide elicited TdeP in each of five dogs studied, while no TdeP
was observed with either dose of moxifloxacin (consistent with the
perceived clinical risk). The CAVB model has also been used to
demonstrate minimal torsadogenic risk with AVE-0118, a novel atrial-
active drug (Oros et al., 2006); despite effects on atrial repolarization,
this agent does not affect ventricular repolarization and does not
induce TdeP.
More recent studies with the anesthetized CAVB dog model have
suggested that beat-to-beat variability of repolarization (BVR), a
measure related to temporal dispersion of repolarization, is a novel
surrogate marker of proarrhythmia. BVR can be explored graphically
using Poincare diagrams that plot consecutive pairs of x,y values of
parameters (typically APD values) from the nth and nth + 1 beat (Fig. 5,
above). As used previously with MAP recordings from Langendorff
perfused hearts (Hondeghem et al., 2001), the polygon formed by
consecutive x–y points widens around the line of identity with greater
beat-to-beat variability. BVR can be characterized based on measures
of short term variability (STV), representing the width of the Poincare
diagram from the line of identity (see Thomsen et al., 2004; van der
Linde et al., 2005 and references therein).
In anesthetized CAVB dogs, MAP measures of left ventricular STV
are preferred to quantify repolarization liability, as MAP measures of
right ventricular STV are a poor predictor of drug-induced TdeP
(Thomsen et al., 2004, 2005). The increase in baseline STV in the
anesthetized, drug-free CAVB dog model has been attributed to long-
term electrical remodeling and blunted beta-adrenergic receptor
signaling (Stengl et al., 2006), and not to changes in heart rate or
average measures of variability in RR interval (Thomsen et al., 2007).
However, in remodeled hearts, the additional contribution of heart rate
variability to drug-induced increases in STV (and TdeP) is uncertain. It
should be noted that CAVB dogs can be divided into two phenotypes;
those susceptible and resistant to TdeP with infusion of dofetilide
(Thomsen et al., 2003; van Opstal et al., 2001c). As proarrhythmia
resistance may be related to the extent of ventricular remodeling, dogs
must be prescreened to reduce baseline variability when evaluating
proarrhythmic risks of novel compounds.
Studies with D-sotalol in the anesthetized CAVB model demonstrated
that increases in STV (but not QT prolongation) were predictive of drug-
induced TdeP (see Fig. 5A and B, above; Thomsen et al., 2004). A recent
paper by Oosterhoff et al. (2007) summarized studies with the CAVB
model from the Vos lab regarding the utility of increased BVR (based
on endocardial MAP recordings) in screening for drug-induced TdeP.
A graphic summary of these studies (derived from reviewed data) is
shown in Fig. 5C. More prominent increases in STV (abscissa) with
dofetilide, NS-7,D-sotalol are associated with increasing incidence of
206 G.A. Gintant / Pharmacology & Therapeutics 119 (2008) 199–209

Fig. 5. A. Representative Poincare plots of left ventricular monophasic action potential duration recorded from anesthetized CAVB dog under control conditions (open arrows)
and during infusion ofD-sotalol (2 mg/kg, left, 4 mg/kg right). Closed arrows indicate time of first extrasystole. B. Changes in short term variability (shown in A) plotted on time
axis. Whereas no TdeP was observed at the 2 mg/kg dose, TdeP was observed with the 4 mg/kg dose at 19 min after start ofD-sotalol infusion. Results obtained from same dog
(adopted from Thomsen et al. Circ, 2004). C. Summary of effects of six compounds on short term variability (STV) and incidence of Torsades (TdeP). For drugs that do not elicit
TdeP, STV does not increase (amiodarone, moxifloxacin, azithromycin). In contrast, STV increases withD-sotalol, dofetilide, and NS-7 are associated with increased incidence of
TdeP. However, the absolute STV value does not correlate with the TdeP incidence (74% incidence of TdeP with dofetilide associated with a relatively low STV (4.3), vs. 25%
incidence of TdeP with D-sotalol associated with a relatively high (5.5) msec ms STV value. STV measured from LV MAPsMAP's in anesthetized dogs. Panel C derived from review
data provided by Oosterhoff et al. (2007).

TdeP (ordinate), while smaller STV increases elicited with amiodarone,
moxifloxacin, and azithromycin did not result in TdeP. In most cases,
TdeP was observed only when statistically significant increases were
observed. However, absolute STV values were not good predictors of
TdeP incidence; D-sotalol (5 ms increase in STV) affected a 25% inci-
dence of TdeP, while dofetilide (4 msec ms increase in STV) affected
a 74% TdeP incidence. Thus, STV qualitatively (but not necessarily
quantitatively) demonstrated a strong predictive value in the screening
for proarrhythmia.
It is recognized that the above summary spans a period of multiple
years; a blinded, prospective comparative study with select drugs
would be necessary to rigorously assess the precision of STV measures
in predicting TdeP with multiple compounds.
Collectively, the above studies suggest that the CAVB dog model
represents a stringent (yet not overly sensitive) model for predicting risk
of TdeP that does not rely on measures of QT prolongation. The utility of
this model is likely due to electrical remodeling reducing repolarization
reserve to a level comparable to that found in humans, as well as a
comparable range of heart rates to those of humans. Additional drug
studies with this model are limited by the low throughput, expense, and
technical expertise required to successfully complete such studies. On
the issue of in vivo model dependency, Chiba et al. (2004a,b) compared
the sensitivity of surrogate markers of drug-induced TdeP in halothane-
anesthetized dogs without chronic AV block. Using sparfloxacin (a non-
specific IKr channel blocker that elicits TdeP in experimental and clinical
studies), they demonstrated prolongation of the QT and QTc intervals,
effective refractory period, and endocardial MAP90 values from both
ventricles. In spite of significant prolongation of the QT interval and MAP
duration, no early afterdepolarizations, ectopic beats, or dispersion of
ventricular repolarization (measured as interventricular dispersion of
endocardial MAPs) was observed; delayed repolarization of phase 3
repolarization (without triangulation) was observed. This study suggests
that the sensitivity of surrogate markers may vary dramatically for
bradycardia-remodeled and “non-remodeled” normal hearts. An in vivo
rabbit heart failure model to characterize torsadogenic risk has recently
been described (Kijtawornrat et al., 2006); further studies with this
model are essential to evaluate it's sensitivity and specificity with
therapeutic and supratherapeutic drug concentrations.
Finally, the extent to which STV measures are predictive of TdeP
may also depend on the model under study. A recent study with
anesthetized, open-chest, phenylephrine-stimulated “normal” rabbits
demonstrated that TdeP elicited by a variety of single or combination
agents (E-4031, HMR1556, ATX-II) was not preceded by an increase in
STV values derived from left ventricular epicardial MAP's (Michael et
al., 2007). These results contrast with those of the anesthetized CAVB
dog model (where STV increases measured from left ventricular
endocardial MAP's predict TdeP); it is uncertain whether these
disparate results relate to species or experimental details. The value
of increased STV measures derived from QT intervals (rather than MAP
recordings) require further study; a recent pilot clinical study suggests
that in the absence of QTc prolongation, baseline BVR (based on QT
interval measures) is of potential value for identifying patients at risk
for proarrhythmia with QT prolonging drugs (Hinterseer et al., 2008);
block of multiple cardiac potassium channels has been shown to
increase short term QT interval variability and provoke TdeP in
conscious dogs (Lengyel et al., 2007).
8. Conclusions
In 1997, the Points to Consider Document suggested additional
preclinical in vitro repolarization studies to assess the proarrhythmic
risk of TdeP (CPMP, 1997). Specifically, this document suggested
cellular electrophysiologic investigations to evaluate the potential to
evaluated delayed ventricular repolarization of drugs based on action
potentials and repolarization abnormalities such (as early after-
depolarizations). Since that time, further progress has been made
elucidating the mechanisms responsible for the initiation (and
perpetuation) of TdeP. Such studies have focused on the “permissive”
role of hERG block (the most prevalent cause of delayed repolariza-
tion), demonstrated that multi-channel block may affect repolariza-
tion and effectively modulate the untoward effects expected of hERG
block alone, and provided an appreciation of the role of repolarization
 G.A. Gintant / Pharmacology & Therapeutics 119 (2008) 199–209
heterogeneity (arising from spatial and temporal dispersion) in setting
the stage for TdeP. Additional models established to evaluate
proarrhythmic risk include more complex preparations such as the
in vitro ventricular wedge and Langendorff heart preparations, and in
vivo chronic AV-node block models. One measured parameter shared
between these more popular models is that of cellular repolarization
(evaluated either using transmembrane or monophasic action
potentials). From action potentials, measures of spatial dispersion of
repolarization are derived by comparing repolarization recorded from
multiple cardiac regions, while measures of temporal dispersion are
provided by following dynamic changes in action potentials evoked
with irregular rhythms or pacing protocols. A second measured
parameter shared between these models is that of either standard or
“pseudo” ECG recordings to evaluate QT prolongation. Simple changes
in the morphology of the QT interval may provide an index of spatial
changes in repolarization heterogeneity. Together, these markers (in
various formats) provide a more comprehensive “risk signature” to
define the potential of drug-induced TdeP.
It is appreciated that defects in cardiac channels for iKr (hERG), IKs,
and INa account for the majority of cases of congenital long QT
syndromes (LQT2, LQT1, and LQT3 respectively) and congenital
proarrhythmia. Similarly, drug effects on these three channels (as
well as the inward rectifier iK1) can lead to acquired long QT
syndromes and proarrhythmia. While most drugs that delay repolar-
ization block hERG block, and potency of block can be measured with
great precision, hERG block is not always predictive of either delayed
repolarization or torsadogenic risk. This may be due (at least in part) to
the fact that Torsades results from “multiple hits” from recognized risk
factors superimposed on an altered or diseased electrophysiologic
substrate (i.e., cardiac hypertrophy, bradycardia-induced electrical
remodeling) and dynamic conditions (such as the short–long–short
rhythm sequence) that may act synergistically to initiate TdeP. Based
on this complex interplay of factors leading to TdeP, it is unrealistic to
expect that any simple single assay or surrogate marker would
provide either 100% sensitivity or specificity to detect proarrhythmic
risk or provide a precise dose–response relationship in regards to
drug-induced proarrhythmia. Similar arguments can be made regard-
ing the utility of QT interval prolongation to quantitatively predict
proarrhythmic risk. As most drugs that have warranted removal from
the market have demonstrated hERG liability, this assay remains
invaluable in early screening of proarrhythmic risk. However, evolving
drugs (especially those displaying a mixture of modest or conflicting
preclinical QT signals) must be evaluated beyond simpler hERG and QT
prolongation studies in order to avoid discarding efficacious and
potentially safe. This will be best accomplished using high content
cellular assays and/or evolving proarrhythmia models.
Proarrhythmia models of increasing complexity (by inclusion of
known risk factors for TdeP) have the potential to more realistically
evaluate drug-induced proarrhythmic risk. However, such models
require considerable expertise and thus are not easily amenable to
routine screening efforts. It is presently difficult to determine how to
meaningfully rank various proarrhythmic risk indices derived from
these complex models to provide a realistic, multifactorial measure of
proarrhythmic liability (and hence an accurate therapeutic margin).
Furthermore, rankings of indices may vary with experimental con-
ditions or drug concentrations. Furthermore, additional “stressors”
may increase sensitivity at the expense of specificity (for example,
addition of ATXII to in vitro Langendorff models). Despite such issues,
the more complex in vivo proarrhythmia models may prove advanta-
geous in that plasma protein binding and metabolites (an integral part
of the experiment) may be comparable to clinical studies. Benchmark-
ing the appropriate experimental conditions to mimic clinical risk is
essential to establish the utility of proarrhythmia models. This exercise
is presently hampered by the fact that the perceived clinical risk of
TdeP is related to both direct cardiac effects as well as indirect effects
(for example, “metabolic multiplication” resulting from drug–drug
207
interactions that may result in excessive drug concentrations and
corresponding electrophysiologic effects). Thus, the perceived risk of
TdeP may not be reflected in the risk signature of compounds mea-
sured directly in a complex proarrhythmia model.
Based on the studies reviewed above, the more complex proar-
rhythmia models are presently most useful for qualitatively clarifying
the torsadogenic risk of drugs, especially those linked to proarrhyth-
mia based on simpler surrogate markers. This approach is exemplified
by published in vitro studies (with rabbit ventricular wedge prepara-
tions (demonstrating low TdP risk scores for compounds with narrow
hERG margins) and in vivo studies with the anesthetized CAVB dog
model (evaluating the effects of amiodarone, van Opstal et al., 2001a;
selected antibiotics, Thomsen et al., 2006; and dose-dependent effects
of sertindole, Thomsen et al., 2003; van Opstal et al., 2001c). Indeed,
one might consider proarrhythmia models in the category of “follow-
up” studies as part of the integrated risk assessment in the ICH S7B
guidelines (Guidance for Industry, 2005b, S7B). Proarrhythmia models
could prove useful in reducing attrition of drugs based primarily on
hERG assay results, as well as clarifying “weak signals” obtained from
extremely precise (and expensive) thorough clinical QT studies.
However, clarification of both the sensitivity and specificity of the
preclinical models must be provided in support of this role. PK/PD
comparisons drawn from preclinical and early clinical studies (see for
example, Garnett et al., 2008) will also be necessary to provide the
necessary contextual link and guide the common interpretation of
study results.
A drug classification system based on drug effects on surrogate
markers would be useful in providing a more global perspective of
safety profiles and evolving risk signatures. This system could
categorize simpler concentration-dependent effects on multiple
cardiac ion channels as well as more complex parameters derived
from increasingly integrated assay systems (including cellular repo-
larization changes (action potentials, spatial and temporal dispersion)
and global (QT) repolarization changes (such as Tpeak–Tend).
Concordance of risk signatures with preclinical or perceived clinical
TdeP risk would be based on comparable drug concentrations.
Additional modulating “off-target” effects would also have to be
considered (for example, alpha-adrenergic agonism/antagonism). This
necessitates a detailed accounting of experimental conditions (tissues
studied, K+ levels, measured drug concentrations, acute vs. chronic
exposures, etc.) to provide context. A simple starting question for such
a classification could be delineating characteristics of hERG blocking
drugs that do not delay repolarization or prolong the QT interval
(covert hERG blocking agents). Our future challenge is to devise and
implement a reliable, robust testing strategy that extends beyond
hERG block and QT prolongation surrogate markers to provide a more
comprehensive integrative risk assessment of torsadogenic risk.
References
Abernethy, D. R., Barbey, J. T., Franc, J., Brown, K. S., Feirrera, I., Ford, N., et al. (2001).
Loratadine and terfenadine interaction with nefazodone: both antihistamines are
associated with QTc prolongation. Clin Pharmacol Ther 69(3), 96−103.
Antzelevitch, C., & Shimizu, W. (2002). Cellular mechanisms underlying the long QT
syndrome. Curr Opin Cardiol 17, 43−51.
Anyukhovsky, E. P., Susunov, E. A., Gainullin, R. Z., & Rosen, M. R. (1999). The controver-
sial M cell. J Cardiovasc Electrophysiol 10, 244−260.
Auer, J., Berent, R., & Eber, B. (2002). Amiodarone in the prevention and treatment of
arrhythmia. Curr Opin Investig Drugs 3(7), 1037−1044.
Belardinelli, L., Antzelevitch, C., & Vos, M. A. (2003). Assessing predictors of drug-
induced Torsade de Pointes. Trends Pharmacol Sci 24(12), 619−625.
Bril, A., Gout, B., Bonhomme, M., Landais, L., Faivre, J. F., Linee, P., et al. (1996). Combined
potassium and calcium channel blocking activities as a basis for antiarrhythmic
efficacy with low proarrhythmic risk: experimental profile of BRL-32872.
J Pharmacol Exp Ther 276, 637−646.
Carlsson, L., Abrahamsson, C., Andersson, B., Duker, G., & Schiller-Linhardt, G. (1993).
Proarrhythmic effects of the class III agent almokalant: importance of infusion rate,
QT dispersion, and early afterdepolarizations. Cardiovasc Res 27(12), 2186−2193.
Chen, X., Cordes, J. S., Bradley, J. A., Sun, Z., & Zhou, J. (2006). Use of arterially perfused
rabbit ventricular wedge in predicting arrhythmogenic potentials of drugs.
J Pharmacol Toxicol Methods 54(3), 261−272.
208 G.A. Gintant / Pharmacology & Therapeutics 119 (2008) 199–209
Chiba, K., Sugiyama, A., Hagiwara, T., Takahashi, S. -I., Takasuna, K., & Hashimoto, K.
(2004a). In-vivo experimental approach for the risk assessment of fluoroquinolone
antibacterial agents-induced long QT syndrome. Eur J Pharmacol 486, 189−200.
Chiba, K., Sugiyama, A., Takasuna, K., Hashimoto, K. (2004b). Comparison of sensitivity
of surrogate markers of drug-induced Torsades de Pointes in canine hearts. Eur J
Pharmacol. 11;502(1–2):117–22.
Committee for Proprietary Medicinal Products (CPMP) (1997). Points to consider: the
assessment of the potential for QT interval prolongation by non-cardiovascular
medicinal products. European Agency for the Evaluation of Medicinal Products. Human
Medicines Evaluation Unit Accessed on April 4, 2008 at CPMP/986/96http://www.
tga.health.gov.au/docs/pdf/euguide/swp/098696en.pdf
Conrath, C. E., & Opthof, T. (2006). Ventricular repolarization. An overview of (patho)
physiology, sympathetic effects, and genetic aspects. Prog Biophys Mol Biol 92,
269−307.
DeBruin, M. L., Pettersson, M., Meyboom, R. H., Hoes, A. S. W., & Leufkens, H. G. (2005).
Anti-HERG activity and the risk of drug-induced arrhythmias and sudden death. Eur
Heart J 26(6), 590−597.
Dennis, A., Wang, L., Wan, X., & Ficker, E. (2007). HERG channel trafficking: novel targets
in drug-induced long QT syndrome. Biochem Soc Trans 35, 1060−1063.
Di Diego, J. M., Belardinelli, L., & Antzelevitch, C. (2003). Cisapride-induced transmural
dispersion of repolarization and Torsade de Pointes in the canine left ventricular
wedge preparation during epicardial stimulation. Circulation 108(8), 1027−1033.
Farkas, A.S., Acsai, K., Tóth, A., Dézsi, L., Orosz, S., Forster, T.et al. (2006). Importance of
extracardiac alpha1-adrenoceptor stimulation in assisting dofetilide to induce
Torsade de Pointes in rabbit hearts. Eur J Pharmacol. 10;537(1–3):118–25.
Fermini, B., & Fossa, A. A. (2003). The impact of drug-induced QT interval prolongation
on drug discovery and development. Nat Rev Drug Discov 2(6), 439−447.
Fossa, A. A., Gorczyca, W., Wisialowski, T., Yasgar, A., Wang, E., Crimin, K., et al. (2007).
Electrical alternans and hemodynamics in the anesthetized guinea pig can
discriminate the cardiac safety of antidepressants. J Pharmacol Toxicol Methods 55(1),
78−85.
Garnett, C. E., Beasley, N., Bhattaram, V. A., Jadhav, P. R., Madabushi, R., Stockbridge, N., et al.
(2008). Concentration–QT relationships play a key role in the evaluation of
proarrhythmic risk during regulatory review. J Clin Pharmacol 48(1), 13−18.
Gintant, G. A., Su, Z., Martin, R. L., & Cox, B. F. (2006). Utility of hERG assays as surrogate
markers of delayed cardiac repolarization and QT safety. Toxicol Pathol 34(1), 81−90.
Gross, G. J. (2006). Bradycardia-mediated ventricular electrical remodeling. J Cardiovasc
Electrophysiol 17(Suppl 1), S178−S182.
Guidance for Industry (2005a, Oct). E14 Clinical Evaluation of QT/QTc Interval
Prolongation and Proarrhythmic Potential for Non-antiarrhythmic Drugs. ICH.
accessed at http://www.fda.gov/cber/gdlns/iche14qtc.pdf on March 29, 2008.
Guidance for Industry (2005b, Oct). S7B Nonclinical Evaluation of the Potential for
Delayed Ventricular Repolarization (QT Interval Prolongation) by Human Pharmaceu-
ticals. ICH. Oct accessed at http://www.fda.gov/cder/guidance/6885fnl.pdf on
March 29, 2008.
Hinterseer, M., Thomsen, M. B., Beckmann, B. M., Pfeufer, A., Schimpf, R., Wichmann, H. E.,
et al. (2008). Beat-to-beat variability of QT intervals is increased in patients with drug-
induced long-QT syndrome: a case control pilot study. Eur Heart J 29(2), 185−190.
Hondeghem, L. M., & Hoffman, P. (2003). Blinded test in isolated female rabbit heart
reliably identifies action potential duration prolongation and proarrhythmic drugs:
Importance of triangulation reverse-use dependence and instability. J Cardiovasc
Pharmacol 41, 14−24.
Hondeghem, L.M., Carlsson, L., Duker, G. (2001). Instability and triangulation of the
action potential predict serious proarrhythmia, but action potential duration
prolongation is antiarrhythmic. Circulation 17;103(15), 2004–2013.
Jonker, D. M., Kenna, L. A., Leishmann, D., Wallis, R., Milligan, P. A., & Jonsson, E. N.
(2005). A pharmacokinetic-pharmacodynamic model for the quantitative predic-
tion of dofetilide clinical QT prolongation from human ether-a-go-go-related gene
current inhibition data. Clin Pharmacol Ther 77(6), 572−582.
Jurgens, G., Graudal, N. A., & Kampmann, J. P. (2003). Therapeutic drug monitoring of
antiarrhythmic drugs. Clin Pharmacokinet 42(7), 647−663.
Kang, J., Wang, L., Chen, X. L., Triggle, D. J., & Rampe, D. (2001). Interactions of a series of
fluoroquinolone antibacterial drugs with the human cardiac K+ channel HERG. Mol
Pharmacol 59(1), 122−126.
Kijtawornrat, A., Nishijima, Y., Roche, B. M., Keene, B. W., & Hamlin, R. L. (2006). Use of a
failing rabbit heart as a model to predict torsadogenicity. Toxicol Sci 93(1), 205−212.
Kirsch, G. E., Trepakova, E. S., Brimecombe, J. C., Sidach, S. S., Erickson, H. D., Kochan, M. C.,
et al. (2004). Variability in the measurement of hERG potassium channel inhibition:
effects of temperature and stimulus pattern. J Pharmacol Toxicol Methods 50(2),
93−101.
Kongsamut, S., Kang, J., Chen, X.L., Roehr, J., Rampe, D. (2002). A comparison of the
receptor binding and HERG channel affinities for a series of antipsychotic drugs. Eur
J Pharmacol. 16;450(1):37–41.
Kuryshev, Y. A., Ficker, E., Wang, L., Hawryluk, P., Dennis, A. T., Wible, B. A., et al. (2005).
Pentamidine-induced long QT syndrome and block of hERG trafficking. J Pharmacol
Exp Ther 312(1), 316−323.
Lawrence, C. L., Bridgland-Taylor, M. H., Pollard, C. E., Hammond, T. G., & Valentin, J. -P.
(2006). A rabbit Langendorff heart proarrhythmia model: predictive value for
clinical identification of Torsades de Pointes. Br J Pharmacol 149, 845−860.
Lazzara, R. (1989). Amiodarone and Torsade de Pointes. Ann Intern Med 111(7), 549−551.
Lengyel, C., Varró, A., Tábori, K., Papp, J. G., & Baczkó, I. (2007). Combined pharma-
cological block of I(Kr) and I(Ks) increases short-term QT interval variability and
provokes Torsades de Pointes. Br J Pharmacol 151(7), 941−951.
Limberis, J. T., McDermott, J. S., Salmen, H. J., Su, Z., Mikhail, A., Green, J. R., et al. (2007).
The effects of plasma proteins on delayed repolarization in vitro with cisapride,
risperidone, andDL, -sotalol. J Pharmacol Toxicol Methods 56(1), 11−17.
Liu, T., Brown, B. S., Wu, Y., Antzelevitch, C., Kowey, P. R., & Yan, G. -X. (2006). Blinded
validation of the isolated arterially perfused rabbit ventricular wedge in preclinical
assessment of drug-induced proarrhythmias. Heart Rhythm 3, 948−956.
Lu, H. R., Vlaminckx, E., Van de Water, A., Rohrbacher, J., Hermans, A., & Gallacher, D. J.
(2007). In-vitro experimental models for the risk assessment of antibiotic-induced
QT prolongation. Eur J Pharmacol 577(1–3), 222−232.
Malik, M., & Camm, A. J. (2001). Evaluation of drug-induced QT interval prolongation:
Implications for drug approval and labeling. Drug Safety 24, 323−351.
Martin, R. L., McDermott, J. S., Salmen, H. J., Palmatier, J., Cox, B. F., & Gintant, G. A.
(2004). The utility of hERG and repolarization assays in evaluating delayed cardiac
repolarization: influence of multi-channel block. J Cardiovasc Pharmacol 43(3),
369−379.
Michael, G., Dempster, J., Kane, K. A., & Coker, S. J. (2007). Potentiation of E-4031-
induced Torsade de Pointes by HMR1556 or ATX-II is not predicted by action
potential short-term variability or triangulation. Br J Pharmacol 152(8), 1215−1227.
Millberg, P., Ramtin, S., Monnig, G., Osada, N., Wasmer, K., Berithardt, G., et al. (2004).
Comparison of the in vitro electrophysiology and proarrhythmic effects of
amiodarone and sotalol in a rabbit model of acute atrioventricular block. J Cardiovasc
Pharm 44, 278−286.
Moxifloxacin product label. (2001). Accessed at http://www.fda.gov/cder/foi/nda/2001/
21-085S010_Avelox_prntlbl.pdf on 2008, March 29.
Napolitano, C., Priori, S. G., & Schwartz, P. J. (1994). Torsade de Pointes. Mechanisms and
Management. Drugs 47(1), 51−65.
Oosterhoff, P., Oros, A., & Vos, M. A. (2007). Beat-to-beat variability of repolarization: a
new parameter to determine arrhythmic risk of an individual or identify
proarrhythmic drugs. Anadolu Kardiyol Derg 7(Suppl 1), 73−78.
Opthof, T., Coronel, R., Wilms-Schopman, F. J., Plotnikov, A. N., Shlapakova, I. N., Danilo,
P., Jr., et al. (2007). Dispersion of repolarization in canine ventricle and the
electrocardiographic T wave: Tp-e interval does not reflect transmural dispersion.
Heart Rhythm 4(3), 341−348.
Oros, A., Volders, P. G. A., Beekman, J. D. M., van der Nagel, T., & Vos, M. A. (2006). Atrial-
specific drug AVE-0118 is free of Torsades de Pointes in anesthetized dogs with
chronic complete atrioventricular block. Heart Rhythm 3(11), 1339−1345.
Redfern, W. S., Carlsson, L., Davis, A. S., Lynch, W. G., MacKenzie, Il, Palethorpe, S., et al.
(2003). Relationships between preclinical cardiac electrophysiology, clinical QT
interval prolongation and Torsade de Pointes for a broad range of drugs: evidence
for a provisional safety margin in drug development. Cardiovasc Res 58(1), 32−45.
Roden, D. M. (1998). Taking the “idio” out of “idiosyncratic”. Pacing Clin Electrophysiol 21(5),
1029−1034.
Roden, D. M. (2004). Drug-induced prolongation of the QT interval. N Engl J Med 350
(10), 1013−1022.
Sanguinetti, M.C., Tristani-Firouzi, M. (2006). hERG potassium channels and cardiac
arrhythmia. Nature. 23;440(7083):463–9.
Sanguinetti, M. C., Jiang, C., Curran, M. E., & Keating, M. T. (1995). A mechanistic link
between an inherited and an acquired cardiac arrhythmia: HERG encodes the IKr
potassium channel. Cell 81, 299−307.
Shah, R. R. (2005a). Drugs, QT interval prolongation and ICH E14. The need to get it right.
Drug Safety 28(2), 115−125.
Shah, R. R. (2005b). Drug-induced QT interval prolongation—regulatory guidance and
perspectives on hERG channel studies. Novartis Found Symp 266, 251−280.
Shah, R. R. (2005c). Drugs, QTc interval prolongation and final ICH E14 guideline: an
important milestone with challenges ahead. Drug Safety 28(11), 1009−1028.
Shah, R. R., & Hongehem, L. M. (2005). Refining detection of drug-induced proarrhythmia:
QT interval and TRIaD. Heart Rhythm 2, 758−772.
Shimizu, W., & Antzelevitch, C. (1998). Sodium channel blockers on transmural
dispersion of repolarization and Torsade de Syndrome: effects of β-adrenergic
agonists and antagonists and sodium cellular basis for the ECG features of the LQT1
form of the long-QT. Circulation 98, 2314−2322.
Singh, B. N. (2006). Amiodarone: a multifaceted antiarrhythmic drug. Curr Cardiol Rep 8(5),
349−355.
Steidl-Nichols, J. V., Hanton, G., Leaney, J., Liu, R. -C., Leishman, D., McHarg, A. S., et al.
(2008). Impact of Study Design on proarrhythmia prediction in the SCREENIT rabbit
isolated heart model. J Pharmacol Toxicol Methods 57(1), 9−22.
Stengl, M., Ramakers, C., Donker, D. W., Nabar, A., Rybin, A. V., Spätjens, R. L., et al. (2006).
Temporal patterns of electrical remodeling in canine ventricular hypertrophy: focus
on IKs downregulation and blunted beta-adrenergic activation. Cardiovasc Res 72(1),
90−100.
Stork, D., Timin, E. N., Berjukow, S., Huber, C., Hohaus, A., Auer, M., et al. (2007). State
dependent dissociation of HERG channel inhibitors. Br J Pharmacol 151(8), 1368−1376.
Taggart, P., Sutton, P., Opthof, T., Coronel, R., & Kallis, P. (2003). Electrotonic cancellation
of transmural electrical gradients in the left ventricle in man. Prog Biophys Mol Biol
82, 243−254.
Thomsen, M. B., Volders, P. A., Stengl, M., Spätjens, R. L., Beekman, J. D. M., Bischoff, U., et al.
(2003). Electrophysiological safety of sertindole in dogs with normal and remodeled
hearts. J Pharmacol Exp Ther 307, 776−784.
Thomsen, M.B., Verduyn, S.C., Stengl, M., Beekman, J.D., de Pater, G., van Opstal, J.et al.
(2004). Increased short-term variability of repolarization predictsD-sotalol-induced
Torsades de Pointes in dogs. Circulation. 19;110(16):2453–9.
Thomsen, M. B., Truin, M., van Opstal, J. M., Beekman, J. D., Volders, P. G., Stengl, M.,
et al. (2005). Sudden cardiac death in dogs with remodeled hearts is associated
with larger beat-to-beat variability of repolarization. Basic Res Cardiol 100,
279−287.
Thomsen, M. B., Beekman, J. D. M., Attevelt, N. J. M., Takahara, A., Sugiyama, A., Chibba, K.,
et al. (2006). No proarrhythmic properties of the antibiotics moxifloxacin or
azithromycin in anesthetized dogs with chronic AV block. Br J Pharmacol 149,
1039−1048.
 G.A. Gintant / Pharmacology & Therapeutics 119 (2008) 199–209
Thomsen, M. B., Oros, A., Schoenmakers, M., van Opstal, J. M., Maas, J. N., Beekman, J. D.,
et al. (2007). Proarrhythmic electrical remodeling is associated with increased beat-
to-beat variability of repolarization. Cardiovasc Res 73, 521−530.
Titier, K., Canal, M.., Déridet, E., Abouelfath, A., Gromb, S., Molimard, M., & Moore, N.
(2004). Determination of myocardium to plasma concentration ratios of five
antipsychotic drugs: Comparison with their ability to induce arrhythmia and
sudden death in clinical practice. Toxicol Appl Pharmacol 199(1), 52−60.
Tsuji, Y., Opthof, T., Yasui, K., Inden, Y., Takemura, H., Niwa, N., et al. (2002). Ionic
mechanisms of acquired QT prolongation and Torsades de Pointes in rabbits with
chronic complete atrioventricular block. Circulation 106(15), 2012−2018.
Ueda, N., Zipes, D. P., & Wu, J. (2004). Functional and transmural modulation of M cell
behavior in canine ventricular wall. Am J Physiol Heart Circ Physiol 287, H2569−2574.
van der Linde, H., van de Water, A., Loots, W., van Deuren, B., Lu, H. R., van Ammel, K., et al.
(2005). A new method to calculate the beat-to-beat instability of QT duration in
drug-induced long QT in anesthetized dogs. J Pharmacol Toxicol Methods 52(1),
168−177.
Van Opstal, J. M., Leunissen, J. D., Wellens, H. J., Vos, M. A., et al. (2001a). Azimilide and
dofetilide produce similar electrophysiologic and proarrhythmic effects in a canine
model of Torsade de Pointes arrhythmias. Eur J Pharmacol 412, 67−76.
van Opstal, J. M., Schoenmakers, M., Verduyn, S. C., de Groot, S. H., Leunissen, J. D., van
Der Hulst, F. F., et al. (2001b). Chronic amiodarone evokes no Torsade de Pointes
arrhythmias despite QT lengthening in an animal model of acquired long-QT
syndrome. Circulation 104, 2722−2727.
van Opstal, J. M., Verduyn, S. C., Leunissen, H. D. M., de Groot, S. H. M., Wellens, H. J. J., &
Vos, M. A (2001c). Electrophysiological parameters indicative of sudden cardiac
death in the dog with chronic complete AV-block. Cardiovasc Res 50(2), 354−361.
Volders, P. G., Sipido, K. R., Vos, M. A., Kulcsár, A., Verduyn, S. C., & Wellens, H. J. (1998).
Cellular basis of biventricular hypertrophy and arrhythmogenesis in dogs with
chronic complete atrioventricular block and acquired Torsade de Pointes. Circulation
98(11), 1136−1147.
Volders, P. G., Sipido, K. R., Vos, M. A., Spätjens, R. L., Leunissen, J. D., Carmeliet, E., et al.
(1999). Downregulation of delayed rectifier K(+) currents in dogs with chronic
complete atrioventricular block and acquired Torsades de Pointes. Circulation 100(24),
2455−2461.
Wallis, Robert M. (2008). Preclinical Testing vs. Clinical Results; Implications for Internal
Decision Making and Regulatory Actions. Drug Information Association Meeting;
QT and Arrhythmia Issues in Drug Development. Bethesda MD (April 7, 2008).
209
Witchel, H.J., Pabbathi, V.K., Hofmann, G., Paul, A.A., Hancox, J.C. (2002). Inhibitory
actions of the selective serotonin re-uptake inhibitor citalopram on HERG and
ventricular L-type calcium currents. FEBS Lett 13;512(1–3):59–66.
Wu, Y., Carlsson, L., Liu, T., Kowey, P. R., & Yan, G. X. (2005). Assessment of the
proarrhythmic potential of the novel antiarrhythmic agent AZD7009 and dofetilide
in experimental models of Torsades de Pointes. J Cardiovasc Electrophysiol 16(8),
898−904.
Wu, L., Shryock, J. C., Song, Y., & Belardinelli, L. (2006). An increase in late sodium
current potentiates the proarrhythmic activities of low risk QT prolonging drugs in
female rabbit hearts. J Pharmacol Exp Ther 316, 718−726.
Wu, L., Rajamani, S., Shryock, J. C., Li, H., Ruskin, J., Antzelevitch, C., et al. (2008).
Augmentation of late sodium current unmasks the proarrhythmic effects of
amiodarone. Cardiovasc Res 77, 1−8.
Yan, G. X., & Antzelevitch, C. (1998). Cellular basis for the normal T wave and the
electrocardiographic manifestations of the Long QT syndrome. Circulation 98,
1928−1936.
Yan, G. X., Shimizu, W., & Antzelevitch, C. (1998). Characteristics and distribution of M
cells in arterially perfused canine left ventricular wedge preparations. Circulation
98(18), 1921−1927.
Yan, G. X., Su, Y., Liu, T., Wang, J., Marinchak, R. A., & Kowey, P. R. (2001). Phase 2 early
afterdepolarization as a trigger of polymorphic ventricular tachycardia in acquired
long-QT syndrome: direct evidence from intracellular recording in the intact left
ventricular wall. Circulation 103, 2851−2856.
Yao, J. A., Du, X., Lu, D., Baker, R. L., Daharsh, E., & Atterson, P. (2005). Estimation of
potency of HERG channel blockers: impact of voltage protocol and temperature.
J Pharmacol Toxicol Methods 52(1), 146−153.
Zeltser, D., Justo, D., Halkin, A., Prokhorov, V., Heller, K., & Viskin, S. (2003). Torsade de
Pointes due to noncardiac drugs. Most patients have easily identifiable risk factors.
Medicine (Baltimore) 82, 282−290.
Zhang, S., Zhou, Z., Gong, Q., Makielski, J. C., & January, C. T. (1999). Mechanism of block
and identification of the verapamil binding domain to hERG potassium channels.
Circ Res 84(9), 989−998.