The Relationship of Clinical QT Prolongation to Outcome in the Conscious Dog Using a Beat-to-Beat QT-RR Interval Assessment
- Dr. Anthony A. Fossa, Pfizer Global Research and Development, Eastern Point Rd., Building 118, MS 4036, Groton, CT 06340. E-mail:anthony_a_fossa{at}groton.pfizer.com
Abstract
QT interval prolongation of the electrocardiogram has been associated with the occurrence of life-threatening fatal ventricular arrhythmias. To understand the relationship between preclinical cardiac conduction assessment to clinical outcome, comparisons of free (unbound)-plasma drug concentrations and their associated effects in the conscious mongrel dog were made to the free plasma concentrations in humans reported to produce QT prolongation. E-4031 (an experimental class III antiarrhythmic), cisapride, terfenadine, terodiline, and verapamil all affect cardiac repolarization and can produce QT prolongation in humans. In the conscious dog, the QT interval was assessed on a beat-to-beat basis in relation to each preceding RR interval at concentrations approximating the same unbound human concentrations. E-4031, cisapride and terodiline statistically increased the QTRR1000 interval [the QT interval at a 60 beats/min (bpm) heart rate] 23, 8, and 9 ms, respectively, at concentrations 0.3 to 15.8 times their relevant clinical level. Increases were not observed for terfenadine or verapamil (p > 0.05 at all doses). Inspection of individual dog QT versus RR interval relationships showed clear QT interval responses specific to each treatment but not readily apparent when data are averaged at a heart rate of 60 bpm. For specific rectifier K+ current (IKr) blockers, robust effects on mean QT prolongation can be detected. However, for drugs that affect repolarization through multiple channels, the effect on the mean QT interval may be more difficult to detect. Inspection of the beat-to-beat QT-RR interval relationship in an individual animal can increase the sensitivity for more accurate clinical prediction.
In the past four years, several drugs such as sertindole, cisapride, and terfenadine have been withdrawn from the marketplace due to the rare occurrence of the fatal ventricular arrhythmia, Torsades de Pointes (TdP). Retrospective evaluation of the European database on these compounds indicated an association of QT prolongation in patients on these medications (Haverkamp et al., 2000). As a result, the Committee for Proprietary Medicinal Products (1997) issued a Points to Consider document on the conduct of studies for the development of noncardiovascular drugs. This and further impending guidance from the International Committee for Harmonization have focused a great deal of attention on the preclinical assessments used to predict the liability for patient populations to experience QT prolongation. Despite the ongoing controversy concerning whether there is a definitive cause and effect relationship, QT prolongation has by default become a surrogate marker for risk of developing TdP.
This study is one of a series of assessments from our laboratories examining the utility of a variety of both in vitro and in vivo preclinical assessments of drugs that are known to produce clinically detectable QT prolongation in humans. Although the magnitude of the QT interval prolongation and the risk associated with TdP are still unclear for many of the drugs, plasma concentrations of the parent drug that produce increases in the QT interval are discernible. E-4031 (an experimental class III antiarrhythmic), cisapride (gastrointestinal prokinetic agent), terfenadine (antihistamine), terodiline (antimuscarinic), and verapamil (antianginal) were all evaluated at their clinically relevant unbound plasma concentrations for QT prolongation in the conscious mongrel dog. Given the dearth of data in the literature from conscious dogs and their relationship to positive clinical outcome with these standards, a careful evaluation was necessary. The purpose of this study was to provide clarity as to the utility of the conscious canine model, in general, and a proposed beat-to-beat technique for predicting the magnitude of QT prolongation with newly discovered or even marketed drugs.
Materials and Methods
Electrocardiographic and Hemodynamic Assessments in the Conscious Dog
Experimental Animals.
Four mongrel dogs (two males and two females) weighing between 8 to 13 kg were instrumented with a Lead II configuration ECG and arterial blood pressure telemetry device (Data Sciences International, Inc., St. Paul, MN). All studies were conducted in accordance with an Animal Care and Usage Protocol approved by the Institutional Animal Care and Use Committee. Dogs were sling trained, and a resting heart rate below 70 beats/min was the criterion for inclusion in the study. Percutaneous jugular and cephalic vein catheters were used for blood sampling and intravenous infusion of vehicle or test compound, respectively. The cardiovascular waveforms were sampled and saved to disk at 1000 Hz using a Po-Ne-Mah data acquisition and analysis system (Gould, Inc., Valley View, OH).
Drugs.
E-4031, cisapride, and terodiline were synthesized for use by the Department of Chemistry at Pfizer, Inc. (Sandwich, Kent, UK). Terfenadine and (±)-verapamil hydrochloride were purchased from Sigma-Aldrich (St. Louis, MO).
Dosing Protocol.
Each dog received vehicle or drug in a randomized fashion using the dose escalation protocol detailed in Table1 to ascertain a dose-response relationship for changes in the QT interval. Each dose level was achieved using an initial loading dose over 5 min followed by an infusion for 15 min to maintain a steady-state drug level during the period of cardiovascular measurements and plasma drug sampling. The volume of the i.v. loading dose was 0.1 ml/kg/min and was adjusted appropriately to deliver the desired amount of drug during the maintenance infusion using the same dosing solution. Between doses, the intravenous dosing syringe was exchanged with the appropriate concentration for the next dose level.
Dose escalation protocol for the various drugs administered intravenously in the conscious mongrel dog
Plasma Sample Analysis.
Plasma concentrations of cisapride, E-4031, terfenadine, terodiline, and verapamil were determined by high-pressure liquid chromatography/tandem mass spectrometry after isolation by cation exchange 96-well solid-phase extraction. Samples were mixed with acidic buffer and internal standard (disopyramide, Sigma d-6035) before extraction. After separation by high-pressure liquid chromatography, analytes and internal standard were detected on a model API3000 mass spectrometer (Applied Biosystems/MDS SCIEX, Foster City, CA). The dynamic range of the method was from 0.2 to 200 ng/ml.
Plasma Protein Binding.
Free unbound drug concentrations were calculated from totals described by the following: E-4031 [31% free (Webster et al., 2001)], terfenadine [3 and 2% free in humans and dogs, respectively (McTavish et al., 1990; Webster et al., 2001)], cisapride [2.5 and 5% free in humans and dogs, respectively (Veereman-Wauters et al., 1991)], terodiline [14% free (Hallen et al., 1990)], verapamil [10% free in humans (Manitpisitkul and Chiou, 1993) and 15% free in dog (Belpaire et al., 1990)].
Expression of Results and Statistical Analyses
QT Analysis on Individual Cardiac Cycles.
The ECG wave form was replayed using the Po-Ne-Mah system data acquisition and analysis system (Gould, Inc.). Approximately
250 to 300 consecutive cardiac cycles (3 to 5 min of continuous data) at the time of steady-state drug exposure for each of
the four dose levels were analyzed interactively by the technician to ensure correct detection of the end of the T wave. This
yielded a confluence of data across an RR interval range of 200 to 2200 ms, representative of the dog's irregular sinus rhythm.
QT was then analyzed as a function of the previous RR interval for each cardiac cycle of a selected time period. An asymptotic
decaying exponential growth curve fit was used to describe the relationship between QT and RR (Raunig et al., 2001).
The equation has also been shown by Matsunaga (1997) to best fit QT-RR data from the dog. The coefficients represent different aspects of the QT-RR relationship. The coefficient
A represents the behavior of QT at very large values of RR. The coefficient Brepresents the behavior of QT at very low values of RR. The coefficientC represents the relationship of the intermediate points and the steepness of the curve at low- and high-RR values.
From the curve fit, the QT at an RR interval of 1000 ms (QTRR1000) was calculated at baseline, and changes in QTRR1000 from the baseline values were calculated for each dose level for all agents. The differences from baseline were analyzed for significance using a one-tailed ttest for each dose level with no multiple comparison corrections. Corrections for multiple comparisons were not done because the multiple comparison adjustment protects against false positives and is not an appropriate adjustment in studies of safety in which the emphasis is to minimize false negatives. The estimates of the least-squares mean of the change in QTRR1000 were calculated using the Mixed procedure in SAS version 6.12 (SAS Institute, Inc., Cary, NC). The least-squares standard errors were used because they are a more powerful method of estimating the error characteristics of the data rather than individual S.E.M. calculations that may be biased by the small sample size.
Results
QT Assessment in the Conscious Dog
E-4031.
E-4031 produced an increase in the QT interval in the conscious dog at all infusion levels ranging from total doses of 8.7 to 430.1 μg/kg (Table 2). The unbound plasma concentrations that were achieved bracketed the human therapeutic unbound levels (Fujiki et al., 1994) of 3 to 4 nM. The magnitude of the QT prolongation was independent of heart rate indicated by the parallel increase in the QT interval across the range of RR intervals of approximately 400 to 1700 ms in individual dogs (Fig. 1). When the QT interval change from baseline is assessed at an RR interval of 1000 ms, dose-related increases in QTRR1000 intervals of 7, 23, 28 and 33 ms were obtained (n = 4). These changes were statistically significant and qualitatively similar in all dogs at the three highest free drug concentrations of 1.7, 8.8, and 41.6 nM of E-4031, respectively (Table 2).
Change in the QT interval following intravenous dose escalation with vehicle or drug compared to predose measurements in the conscious mongrel dog
The QT-RR interval relationship on a beat-to-beat basis in an individual dog (identification no. 7-00318) treated with E-4031 by intravenous infusion. E-4031 at a plasma free drug concentration close to its therapeutic use level (3.5 nM) caused an upward shift in the QT-RR relationship. Best-fit lines were estimated using a multicoefficient exponential model. The individual 95% confidence intervals of the QTRR1000 measurement are included for comparison between vehicle and drug treatment.
Cisapride.
The cisapride infusion protocols resulted in all four free drug concentrations in the dogs being above the clinical use concentrations (Gladziwa et al., 1991) of 2 nM. However, a statistically significant increase of 8 ms in the mean QTRR1000 interval was only observed at 31.6 nM or about 15 times the clinically relevant level. The mean QTRR1000 interval returned toward baseline (increase of 2 ms) at the highest concentrations studied (97 nM). This response is consistent with a bell-shaped dose response reported for effects on action potential duration by cisapride in canine Purkinje fibers (Gintant et al., 2001). In that study, a peak effect on action potential occurred at 1 μM, a considerably higher concentration than found presently in vivo when adjusted for protein binding.
Terfenadine.
No statistically significant effect on the mean QTRR1000 interval was observed at concentrations up to 10.17 nM or 17 times the normal therapeutic concentration (McTavish et al., 1990; Honig et al., 1993) of 0.6 nM.
Terodiline.
Terodiline concentrations ranging from 32.9 nM up to the clinically relevant level (Thomas et al., 1995) of 120 nM caused statistically significant increases in the mean QTRR1000 of 9 to 12 ms. As in the case of cisapride, the dose-response curve was bell-shaped, with the greatest mean QT prolongation occurring at the third free drug concentration level achieved of 32.9 nM.
Verapamil.
Verapamil produced no statistically significant increases in QTRR1000 compared with baseline responses on the initial dosing protocol of steps 1 to 4 (Tables 1 and2). Mean free plasma concentrations achieved ranged from 0.12 to 7.3 nM and are considerably lower than human therapeutic concentrations of 34 nM free drug or 100 ng/ml total (Muller et al., 1986). Initially, evaluation of the QT interval responses was not performed at higher concentrations because profound hypotension and PR interval increases of 25 to 40% were observed in preliminary anesthetized dog studies (data not reported). However, a subsequent group conducted at higher infusion concentrations achieved levels up to 65 nM (Table 2). The absence of QT interval prolongation at levels encompassing the therapeutic concentration was consistent with observations at lower doses.
Discussion
Detecting the liability of medications to cause QT prolongation has become of paramount importance in pharmaceutical development. Regulatory concern over the validity and/or sensitivity of the assays used has been questioned because of a lack of data in preclinical models that can be compared at concentrations that produce discernible QT prolongation in humans. In the present study, pharmacokinetic differences in protein binding and between humans and dog were controlled by consistently using free drug concentration comparisons across all data sets. The results show that the conscious dog is a viable model to detect drugs that specifically produce a robust effect on the QT interval in humans by IKr inhibition. The dog may lack sensitivity with drugs that have subtle effects unless more dogs are used in the study or more robust measures of outlier analyses on individual beats of the QT-RR interval relationship are used (Raunig et al., 2001).
In humans, the IKr specific inhibitor E-4031 produced a mean 12% or 60-ms increase in the QTc interval at clinical concentrations (Fujiki et al., 1994). This magnitude of QT change is consistent for other class III antiarrhythmics (Lande et al., 1998) but is still large compared with the changes observed with noncardiovascular agents that have been removed from the marketplace (Pratt et al., 1996). In the conscious dog, if the QT interval from each heart beat is plotted against the previous RR interval for approximately 300 consecutive beats, a pattern of points specific for each individual animal becomes apparent. After an infusion of E-4031 to achieve clinically relevant plasma concentrations, an upward parallel shift in the pattern occurs across all RR intervals, clearly identifying QT prolongation (refer to Fig. 1). The magnitude of the shift is so profound that almost all of the individual beats will fall outside of the upper 95% confidence limits of the control baseline for each dog. For agents not designed specifically to prolong cardiac repolarization, however, more subtle effects occur, and simple analyses of mean changes in the QT interval do not seem to suffice.
Cisapride consistently shows dose-related prolongation in humans at therapeutic concentrations starting at approximately 2 nM free drug, causing a 6 ms increase in QTc (van Haarst et al., 1998). In the dog, however, these changes are more difficult to detect and were only observed at approximately 32 nM, when the mean QTRR1000 was evaluated from 3 nM up to 100 nM. We used 1000 ms or a 60-beat/min interval for this evaluation of shift in the QT interval magnitude because this is where commonly used correction factors, such as Bazett (Bazett, 1920) or Fridericia (Fridericia, 1920), attempt to normalize QT for changes in heart rate. In addition, examination of the beat-to-beat plots of the QT versus RR interval relationship in all four individual dogs revealed that the upward shift and generation of aberrant outliers seems to predominate at higher heart rates or shorter RR intervals (Fig.2). It can be speculated that these aberrant outliers may be indicative of the inherent susceptibility toward arrhythmogenesis and are usually not detected by more traditional averaging of the data. Although an outlier analyses has been reported by this laboratory to increase the sensitivity for detection of QT prolongation in an individual dog (Raunig et al., 2001), this technique could not be applied to the treatment population because of the variability in the small sample size.
The QT-RR interval relationship on a beat-to-beat basis in an individual dog (identification no. 7-30352) treated with cisapride by intravenous infusion. Cisapride at a plasma free drug concentration of 144 nM caused an upward shift and increase in number of aberrant outliers at shorter RR intervals. Best-fit lines were estimated using a multicoefficient exponential model. The individual 95% confidence intervals of the QTRR1000 measurement are included for comparison between vehicle and drug treatment.
Evaluation of terfenadine and terodiline in the conscious dog produced a similar outcome to cisapride, with no consistent dose-related effect being observed in the mean QT, unlike the findings in humans. These drugs have been reported to have effects on multiple cardiac ion channels, such as sodium and calcium in addition to potassium, that can both shorten and prolong action potential duration leading to variable effects on the QT interval (Larsson-Backstrom et al., 1985; Ming and Nordin, 1995; Jones et al., 1998). Differences in the contribution of these channels can exist due to such things as autonomic tone and disease states (Kaab et al., 1998; Balke and Shorofsky, 1998). It seems that in the normal conscious dog there may be more sensitivity to the calcium channel-blocking activity for these compounds, thus translating to smaller changes in the QT interval compared with humans. Due to the presence of a profound sinus arrhythmia in most dogs, however, a wide range of RR intervals occurs from beat to beat (Hariman et al., 1980), allowing for an easier detection of outlier beats upon individual QT-RR analyses that may be more reflective of QT prolongation liability. The nature of where these outliers occur in relation to heart rate seems to vary from drug to drug. As opposed to cisapride, the outliers observed with higher concentrations of terfenadine and terodiline seem to predominate at lower heart rates (higher RR intervals; Figs. 3 and 4) in approximately half the dogs. The impact of bradycardia has also been reported in cases of arrhythmia with these agents in humans and dogs (Connolly et al., 1991; Salata et al., 1995; Delpon et al., 1999).
The QT-RR interval relationship on a beat-to-beat basis in an individual dog (identification no. 7-30349) treated with terfenadine by intravenous infusion. Terfenadine at a plasma free drug concentration of 10 nM (17-fold its therapeutic use level of 0.6 nM) caused a small upward shift and increase in number of aberrant outliers at higher RR intervals. Best-fit lines were estimated using a multicoefficient exponential model. The individual 95% confidence intervals of the QTRR1000 measurement are included for comparison between vehicle and drug treatment.
The QT-RR interval relationship on a beat-to-beat basis in an individual dog (identification no. 7-00317) treated with terodiline by intravenous infusion. Terodiline at a plasma free drug concentration of 27 nM (4-fold below its therapeutic use level of 120 nM) caused an upward shift in the QT-RR relationship at higher RR intervals. Best-fit lines were estimated using a multicoefficient exponential model. The individual 95% confidence intervals of the QTRR1000 measurement are included for comparison between vehicle and drug treatment.
Verapamil is a potent calcium channel blocker (IC50 = 70 nM) and an inhibitor of the IKr current (IC50 = 143 nM; Zhang et al., 1999). At normal therapeutic use concentrations in humans, verapamil does not prolong the QT interval. Although verapamil is well recognized for its cardiac safety, at concentrations approximately 5-fold higher, QTc can be increased 30 ms (De Cicco et al., 1999). Overdose concentrations of verapamil have only rarely been associated with death due to Torsades de Pointes; so verapamil remains an enigma as to its lack of fatal arrhythmogenicity. In our studies with verapamil at free plasma concentrations close to the clinical use level of 34 nM, we recorded no increases in the mean QTRR1000 interval. In addition, unlike terfenadine and terodiline, beat-to-beat analyses of individual dogs showed no rate-dependent upward shifts in the QT versus RR interval relationships but did show clear reductions (Fig.5). In fact, verapamil in a subsequent experiment, when given to a plasma level of 65 nM, which caused severe hemodynamic effects (10–20% decrease in mean arterial pressure and atrioventricular block), only showed an increase in the QT interval in one of three dogs (data not shown). This latter point may give some insight to the lack of ventricular arrhythmic potential with verapamil compared with other arrhythmogenic calcium channel blockers, such as terodiline. Therefore, the more potent calcium channel-blocking activity with verapamil compared with terodiline may obfuscate the effects of QT prolongation at concentrations that have been reported to affect IKr conduction.
The QT-RR interval relationship on a beat-to-beat basis in an individual dog (identification no. 7-00317) treated with verapamil by intravenous infusion. Verapamil at a plasma free drug concentration of 16 nM caused a downward shift in the QT-RR relationship at all RR intervals. Best-fit lines were estimated using a multicoefficient exponential model. The individual 95% confidence intervals of the QTRR1000 measurement are included for comparison between vehicle and drug treatment.
In summary, five drugs that are known to cause QT prolongation in humans were tested in the conscious dog at their clinically relevant free drug concentrations. For specific IKr blockers, robust effects on mean QT prolongation can be detected. For drugs that affect repolarization through multiple channels, however, the effect on the mean QT interval may be more difficult to detect. Individual responses to the QT-RR interval relationship can increase the sensitivity for more accurate clinical prediction.
Acknowledgments
It is with much gratitude that we acknowledge the contributions of Roxanne L. Winslow and Chin-hu Huang in conducting the experimentation using the conscious dog. The contributions of Robert Walsky and Jessica-lyn Gremminger in generating the plasma concentrations are gratefully acknowledged.
Footnotes
-
DOI: 10.1124/jpet.102.035220
- Abbreviations:
- TdP
- Torsades de Pointes
- QTRR1000
- QT at an RR interval of 1000 ms
- IKr
- rectifier K+ current
- E-4031
- an experimental class III antiarrhythmic
-
- Received February 20, 2002.
- Accepted April 24, 2002.
- The American Society for Pharmacology and Experimental Therapeutics
