- Email: [email protected]
Electrocardiogram, hemodynamics, and core body temperatures of the normal freely moving laboratory beagle dog by remote radiotelemetry
Journal of Pharmacological and Toxicological Methods 53 (2006) 128 – 139 www.elsevier.com/locate/jpharmtox
Original article
Electrocardiogram, hemodynamics, and core body temperatures of the normal freely moving laboratory beagle dog by remote radiotelemetryB David V. Gauvin a,*, Larry P. Tilley b, Francis W.K. Smith Jr. c, Theodore J. Baird a a
Safety Pharmacology MPI Research, Inc., 54943 North Main St., Mattawan, MI 49071-9399, USA b VetMed Consultants, Sante Fe, NM, USA c VetMed Consultants, Lexington, MA, USA Received 14 November 2005; accepted 15 November 2005
Abstract Introduction: The objectives of this study were to provide baseline normative values for circadian changes in the time-series data collected over the course of a normal day in laboratory-housed dogs and to assess the relative efficiency of standard correction formulas to correct for the variations in QT intervals and heart rate functions. Methods: One hundred and twenty-three beagle dogs (65 M, 58 F) were equipped with radiotelemetry transmitters and continuously monitored, while freely moving in their home cages. Electrocardiograms (ECGs), hemodynamic parameters (diastolic, systolic, and mean arterial pressures) as well as core body temperatures were recorded for 22 h. Results and discussion: Blood pressures and core body temperatures demonstrated only very slight variations in their respective values over the 22-h monitoring period. ECGs were measured by a computerized waveform analysis program and quantitative elements reported as RR, PR, QRS, and QT intervals. Little circadian rhythmicity was demonstrated in the ECG intervals. Standard study-specific correction formulas appeared to satisfactorily normalize (i.e., compensate for) the relationship between heart rate and QT intervals in these beagle dogs but elevated the values of the QTc as compared to the uncorrected QT intervals. In sharp contrast, a subject-specific correction method based on analysis of covariance produced a more linear function between heart rates and QT intervals and, more importantly, provided QTc values within the normal range of actual, recorded QT interval data. D 2005 Elsevier Inc. All rights reserved. Keywords: Blood pressure; Body temperature; Beagle; Canine; ECG; Electrocardiogram; Hemodynamics; Canis familiaris; Methods; Radiotelemetry; QT prolongation
1. Introduction A growing number of preclinical pharmacology studies are being conducted with beagle dogs. Many studies using this strain of dog are drug trials involving safety profiling, as promulgated by the International Commission on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH, 2004), and particularly specific i
All authors of the current study claim no conflict of interest, real or imagined, financial or economic with respect to the equipment, companies, or institutions related to the conduct of this study. No portion of this study has been submitted or published elsewhere. * Corresponding author. Tel.: +1 269 668 3336x1613; fax: +1 269 668 4151. E-mail address: [email protected] (D.V. Gauvin). 1056-8719/$ - see front matter D 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.vascn.2005.11.004
safety pharmacology evaluations. Studies conducted under the guidance of the ICH are designed to discover and characterize the potential adverse cardiovascular effects of biologically active new chemical entities (NCE) that may present as an unintended consequence of exposure (ICH S7A, B). Remote monitoring of the ECG by radiotelemetry has provided an alternative means of obtaining physiological measurements from awake and freely moving laboratory animals, without introducing either physically or chemically mediated stress or restraint artifacts in the data. It is generally assumed in contemporary pharmacology practice employing in vivo test systems that the quality of physiological measurements, when collected from conscious, freely moving animals, is superior to those obtained under physical or chemical restraint situations that are known to induce large-scale
D.V. Gauvin et al. / Journal of Pharmacological and Toxicological Methods 53 (2006) 128 – 139
deviations from normal homeostasis. Long-term continuous sampling of the physiological endpoint of interest using telemetry protocols that minimize distortions due to stressful or noxious environments is now commonplace in many laboratories. Moreover, there has been explicit acknowledgement in ICH and other regional regulatory drug safety testing guidelines that conditions present during routine telemetric monitoring may most closely approximate the normal physiological state of the animal, and therefore safety endpoints evaluated under such conditions may consequently demonstrate the greatest predictive validity to outcomes of similar testing in human beings. There are a number of advantages to the use of remote radio-monitoring in preclinical safety pharmacology: (1) telemetry-enabled observation provides the most humane method for monitoring physiologic endpoints in conscious, freely moving laboratory animals, eliminating the stress related to the use of restraint; (2) in the intervening decades since its inception, remote radiotelemetry has become more affordable and reliable, and now easy-to-use commercial products are readily available for monitoring a variety of physiological signals; (3) indwelling sensors for monitoring flow or pressure are usually more accurate and reliable than alternate noninvasive methods, such as transcutaneous laser-Doppler devices, tail cuff blood pressure monitors, etc.; and (4) biotelemetry allows for automated, high temporal resolution, continuous, and long-term data collection via computer, for days, weeks, or months, without any special animal care or maintenance. When used in tandem with remote video monitoring, radiotelemetry can be used to the exclusion of any direct human contact for prolonged periods of time so as to minimize this additional source of interference with physiologically normal baselines. The minimization of human contact during recording intervals can greatly augment the readability, as well as the reliability, of the data derived from emotional animals, whose physiological parameters demonstrate significant synchronous changes with such contact. One of the most important features of radiotelemetry, however, is the reduction of animal use by 60% to 70% in single dose studies (van Acker et al., 1996) and by more than 90% in multiple dose or repeat studies (Kinter, 1996; Kramer & Kinter, 2003). The ECG in beagle dogs has been previously described as demonstrating normal sinus arrhythmias. The irregularity in the ECG appears to be secondary to fluctuations in vagal tone associated with the respiratory cycle (Tilley & Goodwin, 2001). Significant differences are found in the ECG waveform measurements when the dog is assessed in the sitting position when compared to sternal recumbency or lateral recumbency. Coleman and Robson (2005) suggest that reference range values for right lateral recumbency are not valid for ECGs obtained in the sitting position or sternal recumbency (Coleman & Robson, 2005). The present study was a one-year retrospective examination of 123 kennel-raised beagle dogs (65 males, 58 females) surgically implanted with remote cardiovascular and hemodynamic radiotelemetry physiologic transducers. Remote monitoring of each dog’s baseline ECG, blood pressures (diastolic,
129
systolic, and mean arterial), and body temperature values were collected for 22 consecutive hours prior to any test initiation. The purposes of this study were four-fold: (1) to expand and supplement current knowledge on circadian stability of the serially and continuously sampled baseline ECG, blood pressures, and core body temperatures for clinically normal, kennel-born and reared, freely moving dogs within a laboratory setting; (2) to define the circadian characteristics of the ECG waveforms (RR, PR, QRS, QT, and QTc intervals) collected in individually housed laboratory beagle dogs; (3) to examine the relative usefulness of a number of correction formulas applied to normalize QT interval and heart rate functions in dogs; and (4) to compare these findings with a recent and similar report from this group using cynomolgus monkeys (Gauvin, Tilley, Smith, & Baird, 2005). 2. Methods 2.1. Regulatory guidelines All procedures, data review, collection, and analysis were conducted on equipment and software validated in accordance with the Standard Operating Procedures of MPI Research, Inc. (Mattawan, MI), and/or the United States Food and Drug Administration (FDA) Good Laboratory Practice Regulations, 21 CFR Part 58. Data presented were reviewed by the Quality Assurance Unit of MPI Research. 2.2. Subjects Sixty-five male and 58 female kennel-bred beagle dogs (Canis familiaris), approximately 5 –8 months of age, were purchased from Covance Research Products Inc. (Portage, MI) or Marshall Farms (North Rose, NY) and placed in stainless steel double-wide cages according to MPI Research Standard Operating Procedures. Following the testing facility’s standard operating procedures, complete physical, immunological, electrocardiographic, and hematological screening was used to verify the general well-being and good health of each dog. In full accordance and strict adherence to the National Institutes of Health, U.S. Department of Agriculture, U.S. Food and Drug Administration, and AAALAC guidelines, the dogs were approved to undergo surgical procedures to implant the radiotelemetry device. All experimental protocols and procedures were performed following approval of the MPI Research Institutional Animal Care and Use Committee. Baseline recordings occurred approximately 1 month following arrival (2-week acclimation, 2-week post-surgery recovery). Dogs were approximately 6 to 9 months of age when the baseline parameters reported in this study were monitored. The body weights were approximately 5.5 to 8 kg for male dogs and 5.0 to 7.5 kg for female dogs during the monitoring period. 2.3. Surgical implantations Both the surgical placement of the radiotelemetry devices and post-surgical care were conducted by the veterinary team
130
D.V. Gauvin et al. / Journal of Pharmacological and Toxicological Methods 53 (2006) 128 – 139
of the Clinical Medicine Department of MPI Research in a dedicated surgical suite using routine aseptic techniques in accordance with standard operating procedures. Briefly, dogs were fasted overnight prior to surgery and premedicated with carprofen 25 mg orally the morning of surgery. Just prior to surgery, atropine 0.05 mg/kg (IM), acepromazine 0.1 mg/kg (IM), and buprenorphine 0.1 mg/kg (SC) were administered. Anesthesia was induced by pentothal 14 mg/kg (i.v.) to effect. An endotracheal tube was inserted and general anesthesia was maintained with isoflurane 0.5 –2.5% delivered in oxygen via precision vaporizer and non-rebreathing anesthetic circuit. The animal also received cefazolin 25 mg/kg (i.v.) during the initial stages of surgery. The radio transmitter was placed on the dorsal lumbarsacral area in the subcutaneous space. The negative ECG lead was passed subcutaneously from the DSI transmitter to the right upper quadrant under the clavicle. The positive ECG lead was passed subcutaneously from the DSI transmitter to the left lower quadrant approximately 5 cm below the heart. The ECG leads were secured with nonabsorbable suture to the underlying musculature though the small skin incisions. The fluid-filled catheter articulated to the pressure transducer was placed in the femoral artery and advanced to the abdominal aorta. The transducer was anchored to the artery with nonabsorbable suture, all subcutaneous tissues were closed with absorbable suture, and the skin incisions were closed with skin staples and/or surgical glue. Each dog received cefazolin 20 mg/kg (i.m.) immediately post-operatively and cephalexin 250 mg (p.o.) post-operatively as needed and BID the day after surgery. Carprofen 25 mg (p.o.) or buprenorphine 0.01 mg/kg (s.c.) BID was given for 3 days post-surgery for pain control. Animals and their incisions were observed at least once daily during surgical recovery until staples were removed. At least a 2-week recovery period was allowed prior to the initiation of telemetry recordings. 2.4. Radiotelemetry recording A DataQuesti OpenARTi (Version 2.2, Transoma Medical, Inc., St. Paul, MN) biotelemetry acquisition system was linked to a Life Science Suitei, Gould/PoNeMahR (P3, version 3.322) data acquisition and analysis system (now LDS Test and Measurement, LLC, Valley View, OH). The Gould/ PoNeMahR system includes an embedded analog-to-digital conversion matrix linked to IBM-based personal computer systems. Each experimentally and drug-naı¨ve dog was equipped with a triple channel transmitter (Data Sciences, Inc., Model TL11M2-D70-PCT, St. Paul, MN). ECG waveforms were radio-transmitted along with the converted temperature and blood pressure signals through a Data Exchange Matrix (Model 20CH, DataSciences, Inc., St. Paul, MN). ECG waveforms were coded by the frequency of digital pulse trains from the Lead II spatially configured electrodes. These data epochs were examined through the short-term frequency changes from the input device. Each ECG
channel’s frequency was continuously collected at a rate of 500 cycles/s (Hz) and blood pressure signals are collected at a rate of 250 cycles/s. The Gould/PoNeMahR Physiology Platform was utilized to integrate the three independent signals, record and analyze the signals, and to plot the data on the computer screen for Freal-time_ review during data acquisition. A thermal array strip chart-type recorder/printer (AR200, GSI Lumonics, Billerica, MA) was interfaced with the computer-based monitoring system that enabled the printing of selected ECG waveforms for further analysis. 2.5. Radiotelemetry monitoring Each experimentally naı¨ve dog was monitored for 22 consecutive hours in its own stainless steel home cage (92 cm 82 cm 82 cm) under normal 12-h light/dark cycle. Temperature and humidity were monitored and recorded daily and maintained to the maximum extent possible between 18 and 29 -C and 30% to 70%, respectively. Each dog had ad libitum access to drinking water and certified dog chow. Environmental enrichments (KongR toys) were present during acclimation, during surgical recovery, and during telemetry data recording intervals. 2.6. ECG review The PoNeMahR software system utilizes pre-programmed, validated algorithms, constructed as independent modules based on accepted formulas for derived parameters extracted from the published scientific literature. Validation of the algorithms is done on a beat-to-beat basis while the system is acquiring and analyzing the data in real time. Digitized waveforms and raw data are saved in binary file format. The ‘‘Good Wave’’ counter in the software system counts the total number of complete complexes detected during the logging period. A complex is considered to be complete or ‘‘good’’ when the ‘‘Q’’, ‘‘P’’, and ‘‘T’’ waves are detected for the waveform. Once the 22 h of data collection had been completed, each animal’s ECG was visually inspected during a replay session in which the electronic image analyzer of the PoNeMah’sR waveform capture software was reviewed. Individual channel attributes for each waveform, in conjunction with low pass (3 Hz) and high pass filter settings (usually 50 Hz or 70 Hz) that attenuate extraneous frequencies associated with electrical signals correlated with respiration rates and external 60 Hz electrical noise, were reviewed by trained telemetry technicians during these replay sessions. The individually colored validation marks for each attribute (P wave, QRS interval, QT interval, etc.) for the replayed waveforms were reviewed and set to ensure the correct identification of each desired waveform component and, therefore, proper quantification of standard ECG intervals. For the current version of PoNeMah used for this study, the PR interval algorithm measures the distance from the beginning of the P wave to the peak of the Q wave interval as the PR interval. The data generated from each individual
D.V. Gauvin et al. / Journal of Pharmacological and Toxicological Methods 53 (2006) 128 – 139
replay session were then combined with those of other subjects on study for graphic representation of the data, as well as statistical comparisons and QT interval corrections. 2.7. QT interval corrections Regulatory authorities presently require an accurate and dependable measurement of the QT interval that is uncompromised by its inherent inverse relationship to heart rate. There is a current debate regarding the meaningful precision of heart rate correction of the QT interval and what specific correction method should be used (Batchvarov et al., 2002; Malik & Camm, 2001; Malik, Fa¨rbom, Batchvarov, Hnatkova, & Camm, 2002, Malik, Hnatkova, & Batchvarov, 2004). For comparative purposes, the present study used four computational methods to correct the QT intervals with respect to heartrate functions: Three ‘‘study-specific’’ methods were used to calculate QTc: ffiffiffiffiffi Bazett’sformula(1920): QTcB ¼ pQT RR ffiffiffiffiffi Fridericia’sformula (1920): QTcF ¼ pQT 3 RR Van de Water et al.’s (1989) formula: QTcV = QT 87(60 / HR 1)
131
seconds) is then determined from the equation: logðQTcÞ ¼ logðQTÞ bT½logðHRÞ logðHRmÞ QTc is obtained by the inverse log function:analysis of covariance: logðQTcÞ ¼ logðQTÞ bT½logðHRÞ logðHRmÞ b ¼ log regression coefficient HRm ¼ reference heart rate Each dog’s heart rate-corrected QT interval was then summarized to calculate and plot the group mean QTc. 2.8. Statistical analyses Main group (gender) effects were assessed by conducting repeated measures ANOVAs for each dependent measure recorded in this study. A desk-top windows-driven version of SAS (SASR v8.2 for Windows [SAS Institute Inc., Cary, NC, USA]) was used for the analyses. Statistical significance was set a p < 0.05.3. 3. Results
In addition, a fourth, modified, ‘‘subject-specific’’ method first described by Spence, Soper, Hoe, and Cole (1998) and modified by Miyazaki and Tagawa (2002) was used. The subject specific rate-correction utilizing linear model analysis of covariance (ANCOVA) was performed for individual subjects according to a modification of the original Spence et al. method (1998). In the Spence et al.’s correction, pooled group data are used to calculate a linear regression. In the Miyazaki and Tagawa (2002) method, the association or relation between QT and heart rate for each experimental subject is analyzed by linear regression to estimate the coefficient b for each subject. Having estimated b for each dog, the heart rate adjusted QT interval (in
Fig. 1 shows the core body temperature for male (left panel) and female (right panel) beagle dogs over the 22-h monitoring period. A low amplitude circadian rhythm can be seen in the telemetry recorded core body temperatures with the lowest body temperatures for both sexes being logged during the lights-out period (trend seen between 8 and 18 h on the abscissa). The group mean core body temperatures were maintained within a 1.5 -C bandwidth over the entire 22h monitoring period. Fig. 2 shows a well-controlled hemodynamic range of pressures over the 22 h of monitoring. Both systolic (SBP, red)
Fig. 1. Group mean core body temperatures (-C) continuously recorded by remote telemetry from freely moving male and female beagle dogs in their home cages over a 22-h period. Each dog’s body temperature was summarized into 5-min bins, grouped, and plotted over the monitoring period. The red lines indicate the upper and lower 95% confidence limits for the grand mean scores. Time ‘‘0’’ on the abscissa represents an approximate time of 10:30 a.m.
132
D.V. Gauvin et al. / Journal of Pharmacological and Toxicological Methods 53 (2006) 128 – 139
Fig. 2. Group mean systolic blood pressure (SPB: mm Hg; red), diastolic blood pressure (DBP: mm Hg; blue), and mean arterial pressure (MAP: mm Hg; green) continuously recorded by an implanted pressure transducer and transmitted by remote telemetry from freely moving male and female beagle dogs in their home cages over a 22-h period. Each dog’s pressures were summarized into 5-min bins, grouped, and plotted over the monitoring period. The colored lines indicate the upper and lower 95% confidence limits for the grand mean scores. Time ‘‘0’’ on the abscissa represents an approximate time of 10:30 a.m.
and diastolic (DBP, blue) blood pressures demonstrated an extremely stable pattern across the 22-h monitoring period. Mean arterial pressures (MAP, green) also remained within a limited normal operating window in both male and female dogs. There were statistically significant gender differences in systolic blood pressure ( F[1,121] = 9.2, p = 0.003), but not for diastolic or mean arterial pressures. Similar to core body temperatures, group mean heart rate (Fig. 3) for both gender of dogs demonstrated very little circadian rhythmicity across the day-long monitoring period. There was a main group (gender) effect found for heart rates ( F[1,121] = 123.86, p < 0.0001). The heart rates were quantified from the transduced blood pressure pulse signal. While the lowest heart rates were recorded at night (8 to 20 h on the
abscissa), daytime heart rates hardly differed from those recorded during nighttime hours. By comparing the heart rates (Fig. 3) with the quantified Rto-R intervals of the resulting measurements of the ECG waveforms (Fig. 4), the predicted inverse relationship can be seen. A slight circadian rhythm is demonstrated for the R-to-R intervals, more so in the male dogs when compared to their female cohorts. As would be suspected, with the significant main gender effect found for heart rate, there was a coincident statistically significant main group (gender) effect found for the R-to-R intervals as well ( F[1,121] = 202.25, p < 0.0001). In general, the nighttime interval, showing the lowest heart rates (Fig. 3), is associated with the longest R-to-R intervals in Fig. 4 (8 to 20 h on the abscissa).
Fig. 3. Group mean heart rate derived from the blood pressure pulse (beats per minute, bpm) continuously recorded by an implanted pressure transducer and transmitted by remote telemetry from freely moving male and female beagle dogs in their home cages over a 22-h period. Each dog’s heart rate was summarized into 5-min bins, grouped, and plotted over the monitoring period. The red lines indicate the upper and lower 95% confidence limits for the grand mean scores. Time ‘‘0’’ on the abscissa represents an approximate time of 10:30 a.m.
D.V. Gauvin et al. / Journal of Pharmacological and Toxicological Methods 53 (2006) 128 – 139
133
Fig. 4. Group mean R – R intervals (ms) derived from the ECG waveforms continuously recorded by a set of surgically implanted Lead II configured wires secured to the chest muscles of freely moving male and female beagle dogs. The recorded waveforms from each dog were transmitted by remote telemetry from in their home cages over a 22-h period. Each waveform was isolated, analyzed, and integrated into distinct morphological components by a computer-based waveform analysis program. Each dog’s R – R interval was summarized into 5-min bins, grouped, and plotted over the monitoring period. The red lines indicate the upper and lower 95% confidence limits for the grand mean scores. Time ‘‘0’’ on the abscissa represents an approximate time of 10:30 a.m.
The measurements of the PR intervals are shown in Fig. 5. The PR interval, as a measure of atrial and atrioventricular conduction, demonstrated similar stability over the 22h monitoring periods as body temperature and heart rate functions in both male and female dogs. There was no statistically significant difference between male and female dogs for PR intervals. Similar to the stable PR intervals in these beagles, Fig. 6 shows the stability of QRS durations for male and female dogs. Very little circadian rhythmicity can be detected over the monitoring interval for this aspect of ventricular function. There were no statistically significant differences between male and female dogs with respect to QRS durations. Fig. 7 shows the QT intervals. There were only limited, changes in QT intervals over the entire 22-h monitoring period.
There were no grouptime interaction effects found for this parameter, but there were main group (gender) effects ( F[1,121] = 7.02, p < 0.05). The results of the QT intervals corrected as a function of heart-rate using three common study-specific correction procedures (Bazett, 1920; Fridericia, 1920; Van de Water, Verheyen, Xhonneux, & Reneman, 1989) and a subjectspecific correction procedure (Miyazaki & Tagawa, 2002 [see Methods section for details]) are shown for both male (Fig. 8) and female (Fig. 9) beagle dogs over the 22-h monitoring period. While all four normalization procedures provided generally well fitted linear (straight-line) functions for heartrate correction of the QT intervals, only the subject-specific Miyazaki and Tagawa (2002) correction formula resulted in normalized QT intervals in the range of actual QT intervals
Fig. 5. Group mean PR (ms) derived from freely moving male and female beagle dogs (for details, see Fig. 4).
134
D.V. Gauvin et al. / Journal of Pharmacological and Toxicological Methods 53 (2006) 128 – 139
Fig. 6. Group mean QRS (ms) derived from freely moving male and female beagle dogs (for details, see Fig. 4).
recorded during the monitoring periods in the present study. All three study-specific correction formulas (Bazett, Fridericia, and Van de Water) resulted in normalized or corrected QT intervals that were elevated or, in general, greater than the actual averaged QT intervals recorded during the monitoring periods. Fig. 10 shows the number of good-waveforms collected over the 22-h monitoring period expressed as a percentage of total waveforms received from the beagle dogs. Overall, the current monitoring system was able to achieve a greater than 0.80 reliability measure of ECG waveforms over the entire monitoring period. During evening hours, when ‘‘systemic noise’’ is minimized, the group mean reliability of the radiowaveforms increased to almost 90%. The grand mean, standard deviation, and ranges for each of the cardiovascular parameters are summarized in Tables 1 –3. These values represent the normative range of radio telemetry-
recorded cardiovascular parameters in freely moving laboratory-bred beagle dogs. 4. Discussion There is a normal circadian rhythm to core body temperature in beagle dogs ranging from approximately 35 to 37 -C (Lawson, 1999). Beagles in the present study demonstrated a normal circadian range of core body temperatures over the 22 h of monitoring. The relatively stable hemodynamic parameters (diastolic and systolic pressures, and the mean arterial pressures) across the 22-h monitoring period in the present study reflect a group of normotensive, well-acclimated, freely moving, male and female laboratory-bred and -housed beagle dogs. The recorded values of pressures will vary with the breed, age, bodyweights, and health status of the dog. Anderson, Talan, and Engel
Fig. 7. Group mean QT (ms) derived from freely moving male and female beagle dogs (for details, see Fig. 4).
D.V. Gauvin et al. / Journal of Pharmacological and Toxicological Methods 53 (2006) 128 – 139
135
this study. Blood pressure is known to vary with age, body condition, gender, and breed in dogs (Littman & Fox, 1999). While statistically significant differences between the male and female dogs were found for systolic blood pressures and the heart rates, R-to-R intervals as well as the PR and QRS intervals, it is well known that serial dependency is a typical characteristic of time-series data (Gottman & Glass, 1978; Jones, Vaught, & Weinrott, 1977). Time-series studies using large sample sizes have often addressed the issue of statistical significance and meaningful change (Levin, 1975). Unfortunately, statistical and biological significance are not necessarily the same, particularly in the study of electrophysiological signals (Bramblett, 1994) as complex as cardiac function. A difference between two populations can always be established if the observer refines and increases the measurements sufficiently (Simpson, Roe, & Lewontin, 1960). Since variability in nature ensures that no two sets of observations are
Fig. 8. A comparative representation of the ability to dissociate the effect of heart rate on the QT interval between a subject-specific analysis of covariance formula (Miyazaki and Tagawa) and three other standard study-specific formulas (Bazett, Fridericia, and Van de Water et al.; see text). Mean QTs are plotted for all male dogs on study (top panel). For comparison, four different correction formulas were applied to these QT intervals and plotted separately for the same group of male dogs (bottom panel) across the entire 22h monitoring period.
(1990) have previously reported that dogs tend to be awake, eat, and drink intermittently throughout the day and night and that no progressive overnight changes in cardiac output or total peripheral resistance occurred in the laboratory dog. Hemodynamic function in laboratory-bred, and maintained, dogs is related to both environmental and species differences in sleep and ingestive behaviors. The dogs in the present study demonstrated hemodynamic values well within a restricted but well-controlled operating range of blood pressures that are similar to those reported in the previously published literature for laboratory maintained beagle dogs (Anderson et al., 1990; Matsunaga et al., 2001; Mishina et al., 1999; Miyazaki, Yoshida, Samura, Matsumoto, Ikemoto & Tagawa, 2002). While statistically significant differences were found between male and female dogs with respect to the systolic blood pressure parameter, these results were believed to be the result of the extraordinarily high number of datapoints analyzed in
Fig. 9. A comparative representation of the ability to dissociate the effect of heart rate on the QT interval between a subject-specific analysis of covariance formula (Miyazaki and Tagawa) and three other standard study-specific formulas (Bazett, Fridericia, and Van de Water et al.; see text). Mean QTs are plotted for all female dogs on study (top panel). For comparison, four different correction formulas were applied to these QT intervals and plotted separately for the same group of female dogs (bottom panel) across the entire 22-h monitoring period.
136
D.V. Gauvin et al. / Journal of Pharmacological and Toxicological Methods 53 (2006) 128 – 139
Fig. 10. The group mean number of good waves collected (expressed as a percentage of total waves) over the 22-h monitoring period for male and female beagle dogs. Each ECG waveform is transmitted to the Gould/PoNeMahR waveform analysis software package. Systemic, electrical, and/or random noise is censored by the software system and excluded from the analysis. The number of good waves is a general measure of system reliability and represents the proportion of cardiac contractile cycles where all relevant waveform features were detected and integrated relative to the total number of contractile cycles recorded at each interval. The red lines indicate the upper and lower 95% confidence limits of the group mean percentage of good waves collected. Time ‘‘0’’ on the abscissa represents an approximate time of 10:30 a.m. The greatest degree of collection of reliable waveforms appears to occur during the evening hours when the presence of personnel, use of electrical equipment, and institutional activity is minimal (approximately 85% good wave).
exactly the same, the investigator must be concerned with the biological significance of the statistically verified difference. Some rare events may be profoundly important while some significant variability may have little biological significance (Bramblett, 1994). The analysis of the electrocardiogram represents the advantages of visually analyzed charts over statistically finer-grained analyses. When examining for ‘‘between-group’’ differences in complex time-series experiments, reliable changes in the plotted rhythms are correlated with the known and timed introduction and removal of the independent variable. Thus, in the functional analysis of cardiac activity, as well as other complex events, it is this analytic function of the graph that has become of particular significance. This analysis is essentially a visual process; determination of change is dependent on the change being of sufficient magnitude to be apparent to the eye. Compared with the complex algebraic sophistication of statistical tests of significance, the visual inspection of data may appear simplistic, but it remains an important and valuable tool in the analysis of ongoing activity in physiological systems (Baer, 1977; Baer, Wolf, & Risley, 1968; Parsonson & Baer, 1978). While a number of gender-dependent differences were found in the present study, none of these particular parametric differences reflect physiologically meaningful events. There is a general circadian rhythm to heart rate variability in most species (Cornelissen, Haus, & Halberg, 1994; Hossmann, Fitzgerald, & Dollery, 1980; Lemmer, 1994; Lemmer, Scheidel, Blume, & Becker, 1991; Matsunaga et al., 2001). Normal resting heart rates in dogs can range from 70 to 220 beats/min, depending on the age of the dog. Barking or emotionality usually elicits higher heart rates that characterize the heart rhythm in dogs as ‘‘normal sinus tachycardia’’. In the beagle dog, daily variations in heart rate are related to short-
term changes caused by excitement and exercise (Miyazaki et al., 2002). It has been previously reported that blood pressures and heart rate increased at feeding and that meals tend to cycle across the full 24 h in laboratory dogs (Anderson et al., 1990; Mishina et al., 1999). These authors reported that heart rates generally decreased progressively after feeding. This variability can be seen in the ‘‘saw tooth’’ pattern of heart rates across the 22-h monitoring period in the present study. The stable pattern of heart rate changes recorded from freely moving beagle dogs in the present study and the very limited circadian rhythmicity in heart rate variability are similar to those reported by Miyazaki et al. (2002) and Anderson et al. (1990). The R-to-R interval reflects the time between the most intense phases of ventricular depolarization of one heart contraction to the peak of the next contraction. The peak of Table 1 Summarized ECG intervals for male and female beagle dogs Endpoint
Meana (ms)
S.D.
Min/max 95% confidence limits
ECG interval RR PR QRS QT QTcb
values in male beagle dogs 580.42 32.41 96.42 1.6 41.66 0.89 220.10 3.37 217.98 1.77
416.94 – 700.86 90.91 – 103.80 36.12 – 45.88 200.71 – 232.96 207.52 – 226.92
ECG interval RR PR QRS QT QTcb
values in female beagle dogs 518.82 23.30 90.82 1.85 41.33 0.76 217.60 3.56 215.54 1.87
407.50 – 610.70 85.99 – 99.07 22.20 – 36.82 198.34 – 231.90 205.00 – 226.85
a Grand mean values (ms) of all 5-min bin intervals averaged over the full 22h monitoring period. b Refers to Miyazaki and Tagawa (2002) correction values only.
D.V. Gauvin et al. / Journal of Pharmacological and Toxicological Methods 53 (2006) 128 – 139 Table 2 Summarized hemodynamic parameters for male and female beagle dogs Meana (mm Hg)
S.D.
Min/max 95% confidence limits
Blood pressure values in male beagle dogs Systolic blood pressure Diastolic blood pressure Mean arterial blood pressure
139 83 105
2 2 2
128.98 – 148.14 75.13 – 90.62 96.33 – 113.75
Blood pressure values in female beagle dogs Systolic blood pressure Diastolic blood pressure Mean arterial blood pressure
137 82 105
3 2 2
127.15 – 148.71 74.55 – 92.27 96.22 – 115.58
Endpoint
a Mean values represent the grand means of data collected in each individual animal across a 22-h interval.
the R wave is used clinically as an indicator of whether or not ventricular depolarization is proceeding over normal pathways, and to calculate heart rate. Comparative analyses of Figs. 3 and 4 reveal the inverse relationship between heart rates and R-to-R intervals. The dogs demonstrated a prolongation of the R-to-R intervals during the time-frame of the monitoring period in which heart rate decreases were recorded, and vice-versa. As would be predicted, the circadian variability is of approximately equivalent magnitude for the R-to-R interval when compared to the heart rates derived from the blood pressure pulse in these freely moving beagle dogs. The period of time from the initiation of the P wave to the beginning of the QRS complex is designated as the PR interval and indicates the time it takes for an action potential to spread through the atria to the atrioventricular (AV) node. The PR interval, therefore, largely reflects the time of atrial depolarization. The PR interval is usually reported to be within 60 to 130 ms in duration (Martin 2000; Miller, Tilley, Smith, & Fox, 1999; Tilley & Goodwin, 2001) in beagles. The present study found a more restricted circadian variability and range of approximately 10 ms (¨ 90 to 100 ms) for PR intervals in freely moving laboratory-bred, raised, and maintained beagle dogs. The QRS complex reflects the time required for ventricular depolarization. The normal range for QRS duration for the beagle dogs reported in the published literature is 40 to 50 ms (Martin, 2000; Miller et al., 1999; Tilley, 1992; Tilley & Goodwin, 2001). The QRS durations reported in the present study, using freely moving dogs, are consistent with the range of durations of these intervals documented in the literature. The normal range of uncorrected QT intervals in the dog reported in the published literature is 150 to 250 ms at normal heart rates (Martin, 2000; Miller et al., 1999; Tilley, 1992; Tilley & Goodwin, 2001). QT variability is a non-invasive marker of cardiac repolarization lability and a higher QT variability has been associated with sudden death (Nolan, Girand, Bailie, & Yeragani, 2004). The range of the 95% confidence intervals for the QT interval in the present study was within a more restricted range for both male and female dogs than previously reported. Nolan et al. (2004) have recently reported a diurnal pattern of QT intervals similar to healthy humans. Similar to almost all cardiac parameters
137
recorded in the present study there was limited QT interval variability in the 123 dogs used in this study. Heart rate-corrected intervals were obtained using standard study-specific correction formulas used in a number of laboratories and first described by Bazett (1920), Fridericia (1920), and Van de Water et al. (1989). In contrast, we also used a subject-specific correction formula derived from a modification of a method first described by Spence et al. (1998) (cf. Miyazaki & Tagawa, 2002). The QT interval correction for heart rate (QTc) is believed to reflect sympathovagal balance (Lemarre-Cliche´, Lacourciere, de Champlain, Poirier, & Larochelle, 2003), and it is the traditional method of assessing the duration of ventricular repolarization (Crow, Hannan, & Folsom, 2003). Although many mathematical models have been proposed to describe the relationship between the QT interval and heart rate, none of the models (and corresponding heart rate correction formulas) has been accepted as universally applicable (Aytemir et al., 1999; FunckBrentano & Jaillon, 1993; Hodges, 1997; Malik, 2001; Rautaharju, Warren, & Calhoun, 1990). As many authors have pointed out, any formula describing the relationship between the QT interval and the preceding heart rate is limited by physiological factors, such as the extra-heart rate influences on myocardial repolarization (Browne, Prystowsky, Heger, & Zipes, 1983; Cappato, Alboni, Pedroni, Gilli, & Antonioli, 1991; Rickards & Norman, 1981) and the ‘‘lag hysteresis’’ of the QT –heart rate relationship (Franz, Swerdlow, Liem, & Schaefer, 1988; Lau et al., 1988). Recent approaches to the issue of heart rate correction of the QT interval have largely abandoned the use of fixed rate correction formulas (also referred to as ‘‘study-specific corrections’’ such as Bazett, 1920; Fridericia, 1920; Van de Water et al., 1989, etc.) and have adopted an individual rate correction derived in each experimental subject from multiple observations of the QT – heart rate data in a baseline, drug-free state (also referred to as ‘‘subject-specific corrections’’). The QT – heart rate relationship has been shown to have substantial between-subject variability as well as high within-subject stability (Batchvarov et al., 2002). Relative to the above, a growing number of authors have recently shifted away from formulas relying upon heterogeneous, pooled estimates of the QT/ HR relationship in favor of formulas individualized for each experimental subject (Batchvarov et al., 2002; Malik & Camm, 2001; Malik et al., 2002; Smetana, Batchvarov, Hnatkova, Camm, & Malik, 2003). According to the Food and Drug Administration’s Center for Drug Evaluation and Table 3 Summarized heart rate values for male and female beagle dogs Endpoint
Meana (bpm)
S.D.
Heart rate values in male beagle dogs Heart rate 111 6 Heart rate values in female beagle dogs Heart rate 120 5 a
Min/max 95% confidence limits 89.77 – 149.65
102.35 – 150.88
Mean values (beats/min) represent the grand means of data collected in each individual animal across a 22-h interval.
138
D.V. Gauvin et al. / Journal of Pharmacological and Toxicological Methods 53 (2006) 128 – 139
Research (CDER), Office of Drug Evaluation (ODE1), there is wide agreement that Bazett’s correction is not the best correction formula to use in many cases and most researchers now believe that the regression approach is better (Huml, Barker, Tonkens, & Hebert, 2004). Our data support this conclusion. A recent study by Malik et al. (2004) demonstrated that the subject-specific corrections (Spence-like) led to maximum errors in single millisecond range (2.4, 5.7, and 2.6 ms with linear, log/log linear, and exponential models, respectively), while the study-specific corrections (Bazett and Fridericia’s) led to substantially greater errors (error range of 17.8, 19.4, and 16.9 ms with linear, log/log linear, and exponential models, respectively). These authors found that both Bazett and Fridericia’s corrections led not only to substantial errors (error range of 28.3 and 16.9 ms), but also to regular bias with systematically false negative and false positive conclusions dependent on modeled heart rate acceleration and deceleration. The present study clearly found inflated QTc values above any recorded actual uncorrected QT interval recorded during the study. All three study specific correction formulas over-estimated the QTc values and, therefore, misrepresented the actual data. Similar to our recent report on the cardiovascular function of cynomolgus monkeys, the subjectspecific Miyazaki and Tagawa correction provided the best heart-rate correction for QT interval in beagle dogs in the present study. 5. Conclusion The present study has provided normal operating range values for core body temperature, hemodynamic, and ECG parameters for well-acclimated, freely moving laboratoryhoused beagle dogs using radiotelemetry recording methodologies. With a collection speed of 500 cycles/s for 22 h of continuous monitoring of a complete set of endpoints, the current system achieved a minimum signal integration accuracy of greater than 80%. In a companion paper examining the cardiovascular, hemodynamic, and body temperature changes in cynomolgus monkeys, there were clear circadian variations in almost all parameters across the 22-h monitoring period. In the present study, there was little evidence of large amplitude circadian rhythms in any measured parameter in these beagle dogs. While the extremely stable, normotensive data from the present study may not represent the ‘‘normal’’ variations recorded in canine pets seen within a veterinary clinic, they are believed to represent the ‘‘normal’’ laboratory-bred and raised dog that is well-acclimated, not stressed by the environment, and well adjusted to the daily patterns of activity found within a Contract Research Organization. The differences in the degree of human contact, meal availability, and environmental changes that the ‘‘normal’’ canine pet is exposed to may influence the cardiovascular and hemodynamic parameters recorded within a standard veterinary clinic and may underlie the differences between the current data and those that may be derived by most primary care veterinarians in the clinic.
Radiotelemetry substantially reduces animal use and associated research costs by increasing the number of parameters that can be collected from a group of animals and thereby eliminating the need for satellite groups or separate studies to collect the same data, by permitting the use of blocked study designs, and by increasing the data precision in both individual animals and study groups. Radiotelemetry also offers refinement alternatives by reducing animal stress and discomfort, improving the quality and precision of experimental data, and reduces the need for isolation from other animals of the same species while reducing the amount of direct hands-on contact with trained laboratory staff (Kramer & Kinter, 2003). We believe these procedures represent contemporary industry preferred practice for measuring such parameters under the ICH guidelines and are amenable to routine use in a variety of other relevant safety/efficacy studies. Acknowledgement The work described in this manuscript was paid for by MPI Research, Inc. References Anderson, D. E., Talan, M. I., & Engel, B. T. (1990). Nocturnal hemodynamic patterns in dogs. Physiology and Behavior, 48, 485 – 487. Aytemir, K., Maarouf, N., Gallagher, M. M., Yap, Y. G., Waktare, J. F., & Malik, M. (1999). Comparison of formulae for heart rate correction of QT interval in exercise electrocardiograms. Pacing and Clinical Electrophysiology, 22, 1397 – 1401. Baer, D. M. (1977). Perhaps it would be better not to know everything. Journal of Applied Behavioral Analysis, 10, 167 – 172. Baer, D. M., Wolf, M. M., & Risley, T. R. (1968). Some current dimensions of applied behavior analysis. Journal of Applied Behavioral Analysis, 1, 91 – 97. Batchvarov, V. N., Ghuran, A., Smetana, P., Hnatkova, K., Harries, M., Dilaveris, P., et al. (2002). QT – RR relationship in healthy subjects exhibits substantial intersubject variability and high intrasubject stability. American Journal of Physiology, 282, H2356 – H2363. Bazett, H. C. (1920). An analysis of the time-relations of electrocardiograms. Heart, 7, 258 – 353. Bramblett, C. A. (1994). Patterns of primate behavior. Prospect heights. IL’ Waveland Press. Browne, K. F., Prystowsky, E., Heger, J. J., & Zipes, D. P. (1983). Modulation of the Q – T interval by the autonomic nervous system. Pacing and Clinical Electrophysiology, 6, 1050 – 1056. Cappato, R., Alboni, P., Pedroni, P., Gilli, G., & Antonioli, G. E. (1991). Sympathetic and vagal influences on rate-dependent changes of QT interval in healthy subjects. American Journal of Cardiology, 68, 1188 – 1193. Coleman, M. G., & Robson, M. C. (2005). Evaluation of six-lead electrocardiograms obtained from dogs in a sitting position or sternal recumbancy. American Journal of Veterinary Research, 66, 233 – 237. Cornelissen, G., Haus, E., & Halberg, F. (1994). Chronobiologic blood pressure assessment from womb to tomb. In Y. Touitou, & E. Haus (Eds.), Biological rhythms in clinical and laboratory medicine (pp. 428 – 452). New York’ Springer Verlag. Crow, R. S., Hannan, P. J., & Folsom, A. R. (2003). Prognostic significances of corrected QT and corrected JT interval for incident coronary heart disease in a general population sample stratified by presence or absence of wide QRS complex: The ARIC Study with 13 years of follow-up. Circulation, 108, 1985 – 1989. Franz, M. R., Swerdlow, C. D., Liem, L. B., & Schaefer, J. (1988). Cycle length dependence of human action potential duration in vivo. Effects of single
D.V. Gauvin et al. / Journal of Pharmacological and Toxicological Methods 53 (2006) 128 – 139 extrastimuli, sudden sustained rate acceleration and deceleration, and different steady-state frequencies. Journal of Clinical Investigation, 82, 972 – 979. Fridericia, L. S. (1920). Die systolendauer im electrokardiogramm bei normalen menschen und bei herzkranken. Acta Medica Scandinavica, 53, 469 – 486. Funck-Brentano, C., & Jaillon, P. (1993). Rate-corrected QT interval: Techniques and limitations. American Journal of Cardiology, 72, 17B – 22B. Gauvin, D.V., Tilley, L.P., Smith, F.W.K. Jr., Baird, T.J. (2005). Electrocardiogram, hemodynamics, and core body temperatures of the normal freelymoving cynomolgus monkey by remote telemetry. Journal of Pharmacology and Toxicological Methods 53, 87 – 98. Gottman, J. M., & Glass, G. V. (1978). Analysis of interrupted time-series experiments. In T. R. Kratochwill (Ed.), Single subject research: Strategies for evaluating change (pp. 197 – 237). New York’ Academic Press. Hodges, M. (1997). Rate-correction of the QT interval. Cardiac Electrophysiology Review, 3, 360 – 363. Hossmann, V., Fitzgerald, G. A., & Dollery, C. T. (1980). Circadian rhythm of baroreflex reactivity and adrenergic vascular response. Cardiovascular Research, 14, 125 – 129. Huml, R. A., Barker, G., Tonkens, R. M., & Hebert, D. A. (2004, May). The clinical and regulatory implications of QT interval prolongation. Regulatory Affairs Focus, , 40 – 43. International Conference on Harmonisation of Technical Requirements for Registration of Pharmacetuticals for Human Use (10 June, 2004). The nonclinical evaluation of the potential for delayed ventricular repolarization (QT interval prolongation) by human pharmaceuticals, S7B. . Jones, R. R., Vaught, R. S., & Weinrott, M. (1977). Time-series analysis in operant research. Journal of Applied Behavioral Analysis, 10, 151 – 166. Kinter, L. B. (1996). Cardiovascular telemetry and laboratory animal welfare: New reduction and refinement alternatives (Abstract). General Pharmacology/Safety Pharmacology Meeting, Philadelphia, PA. , . Kramer, K., & Kinter, L. B. (2003). Evaluation and applications of radiotelemetry in small laboratory animals. Physiological Genomics, 13, 197 – 205. Lau, C. P., Freedman, A. R., Fleming, S., Malik, M., Camm, A. J., & Ward, D. E. (1988). Hysteresis of the ventricular paced QT interval in response to abrupt changes in pacing rate. Cardiovascular Research, 22, 67 – 72. Lawson, P. T. (Ed.). (1999). American association for laboratory animal science assistant laboratory animal technician training manual. Chelsea (MI)’ Sheridan Books, Inc. Lemarre-Cliche´, M., Lacourciere, Y., de Champlain, J., Poirier, L., & Larochelle, P. (2003). Does QTc interval predict the response to betablockers and calcium channel blockers in hypertensives. Heart Disease, 5, 244 – 252. Lemmer, B. (1994). Cardiovascular chronobiology and chronopharmacology. In Y. Touitou, & E. Haus (Eds.), Biological rhythms in clinical and laboratory medicine (pp. 418 – 427). New York’ Springer Verlag. Lemmer, B., Scheidel, B., Blume, H., & Becker, H. -J. (1991). Clinical chronopharmacology of oral sustained-release isosorbide-5-mononitrate in healthy subjects. European Journal of Clinical Pharmacology, 40, 70 – 75. Levin, J. R. (1975). Determining the sample size for planned and post hoc analysis of variance comparisons. Journal of Educational Measurement, 12, 99 – 108. Littman, M. P., & Fox, P. R. (1999). Systemic hypertension: Recognition and treatment. In P. R. Fox, D. Sisson, & N. S. Moise (Eds.), Textbook of canine and feline cardiology: Principles and clinical practice (2nd edR). Philadelphia’ Saunders. Malik, M. (2001). Problems of heart rate correction in assessment of druginduced QT interval prolongation. Journal of Cardiovascular Electrophysiology, 12, 411 – 420.
139
Malik, M., & Camm, A. J. (2001). Evaluation of drug-induced QT interval prolongation: Implications for drug approval and labeling. Drug Safety, 24, 323 – 351. Malik, M., Fa¨rbom, P., Batchvarov, V., Hnatkova, K., & Camm, A. J. (2002). Relation between QT and RR intervals is highly individual among healthy subjects: Implications for heart rate correction of the QT interval. Heart, 87, 20 – 228. Malik, M., Hnatkova, K., & Batchvarov, V. (2004). Differences between studyspecific and subject-specific heart rate corrections of the QT interval in investigations of drug induced QTc prolongation. Pacing and Clinical Electrophysiology, 27(1), 791 – 800. Martin, M. (2000). Small animal ECG’s: An introductory guide. Oxford, UK’ Blackwell Science. Matsunaga, T., Harada, T., Mutsui, T., Inokuma, M., Hashimoto, M., Miyauchi, M., et al. (2001). Spectral analysis of circadian rhythms in heart rate variability of dogs. American Journal of Veterinary Research, 62, 37 – 42. Miller, M. S., Tilley, L. P., Smith, F. W. K., Jr., & Fox, P. R. (1999). Electrocardiography. In P. R. Fox, D. Sisson, & S. Moise (Eds.), Textbook of canine and feline cardiology: Principles and clinical practice (2nd ed.). Philadelphia’ Saunders. Mishina, M., Watanabe, T., Matsuoka, S., Shibata, K., Fukii, K., Maeda, H., et al. (1999). Diurnal variations of blood pressure in dogs. Journal of Veterinary Medical Science, 61, 643 – 647. Miyazaki, H., & Tagawa, M. (2002). Rate-correction technique for QT interval in long-term telemetry ECG recording in beagle dogs. Experimental Animals, 51(5), 465 – 475. Miyazaki, H., Yoshida, M., Samura, K., Matsumoto, H., Ikemoto, F., & Tagawa, M. (2002). Ranges in diurnal variation and the pattern of body temperature, blood pressure and heart rate in laboratory beagle dogs. Experimental Animal, 51, 95 – 98. Nolan, E. R., Girand, M., Bailie, M., & Yeragani, V. K. (2004). Circadian changes in the QT variability index in the beagle dog. Clinical and Experimental Pharmacology and Physiology, 31, 783 – 785. Parsonson, B. S., & Baer, D. M. (1978). The analysis and presentation of graphic data. In T. R. Kratochwill (Ed.), Single subject research: Strategies for evaluating change (pp. 101 – 166). New York’ Academic Press. Rautaharju, P. M., Warren, J. W., & Calhoun, H. P. (1990). Estimation of QT prolongation. A persistent, avoidable error in computer electrocardiography. Journal of Electrocardiology, 23, 111 – 117. Rickards, A. F., & Norman, J. (1981). Relation between QT interval and heart rate. New design of physiologically adaptive cardiac pacemaker. British Heart Journal, 45, 56 – 61. Simpson, G. G., Roe, A., & Lewontin, R. C. (1960). Quantitative zoology. NY’ Harcourt, Brace and World. Smetana, P., Batchvarov, V., Hnatkova, K., Camm, A. J., & Malik, M. (2003). Circadian rhythm of the corrected QT interval: Impact of different heart rate correction models. Pacing and Clinical Electrophysiology, 26, 383 – 386. Spence, S., Soper, K., Hoe, C. -M., & Coleman, J. (1998). The heart ratecorrected QT interval of conscious beagle dogs: A formula based on analysis of covariance. Toxicological Science, 45, 247 – 258. Tilley, L. (1992). Essentials of canine and feline electrocardiography (3rd edR). England’ Blackwell Publishing. Tilley, L. P., & Goodwin, J. -K. (2001). Manual of canine and feline cardiology (3rd edR). Philadelphia’ Saunders. van Acker, S.A.B.E, Kramer, K., Voest, E. E., Grimbergen, J. A., Zhang, J., van der Vijgh, W. J. F., et al. (1996). Doxorubicin-induced cardiotoxicity monitored by ECG in freely moving mice. A new model to test potential protectors. Cancer Chemotherapy and Pharmacology, 38, 95 – 101. Van de Water, A., Verheyen, J., Xhonneux, R., & Reneman, R. S. (1989). An improved method to correct the QT interval of the electrocardiogram for changes in heart rate. Journal of Pharmacological Methods, 22, 207 – 217.
Original article
Electrocardiogram, hemodynamics, and core body temperatures of the normal freely moving laboratory beagle dog by remote radiotelemetryB David V. Gauvin a,*, Larry P. Tilley b, Francis W.K. Smith Jr. c, Theodore J. Baird a a
Safety Pharmacology MPI Research, Inc., 54943 North Main St., Mattawan, MI 49071-9399, USA b VetMed Consultants, Sante Fe, NM, USA c VetMed Consultants, Lexington, MA, USA Received 14 November 2005; accepted 15 November 2005
Abstract Introduction: The objectives of this study were to provide baseline normative values for circadian changes in the time-series data collected over the course of a normal day in laboratory-housed dogs and to assess the relative efficiency of standard correction formulas to correct for the variations in QT intervals and heart rate functions. Methods: One hundred and twenty-three beagle dogs (65 M, 58 F) were equipped with radiotelemetry transmitters and continuously monitored, while freely moving in their home cages. Electrocardiograms (ECGs), hemodynamic parameters (diastolic, systolic, and mean arterial pressures) as well as core body temperatures were recorded for 22 h. Results and discussion: Blood pressures and core body temperatures demonstrated only very slight variations in their respective values over the 22-h monitoring period. ECGs were measured by a computerized waveform analysis program and quantitative elements reported as RR, PR, QRS, and QT intervals. Little circadian rhythmicity was demonstrated in the ECG intervals. Standard study-specific correction formulas appeared to satisfactorily normalize (i.e., compensate for) the relationship between heart rate and QT intervals in these beagle dogs but elevated the values of the QTc as compared to the uncorrected QT intervals. In sharp contrast, a subject-specific correction method based on analysis of covariance produced a more linear function between heart rates and QT intervals and, more importantly, provided QTc values within the normal range of actual, recorded QT interval data. D 2005 Elsevier Inc. All rights reserved. Keywords: Blood pressure; Body temperature; Beagle; Canine; ECG; Electrocardiogram; Hemodynamics; Canis familiaris; Methods; Radiotelemetry; QT prolongation
1. Introduction A growing number of preclinical pharmacology studies are being conducted with beagle dogs. Many studies using this strain of dog are drug trials involving safety profiling, as promulgated by the International Commission on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH, 2004), and particularly specific i
All authors of the current study claim no conflict of interest, real or imagined, financial or economic with respect to the equipment, companies, or institutions related to the conduct of this study. No portion of this study has been submitted or published elsewhere. * Corresponding author. Tel.: +1 269 668 3336x1613; fax: +1 269 668 4151. E-mail address: [email protected] (D.V. Gauvin). 1056-8719/$ - see front matter D 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.vascn.2005.11.004
safety pharmacology evaluations. Studies conducted under the guidance of the ICH are designed to discover and characterize the potential adverse cardiovascular effects of biologically active new chemical entities (NCE) that may present as an unintended consequence of exposure (ICH S7A, B). Remote monitoring of the ECG by radiotelemetry has provided an alternative means of obtaining physiological measurements from awake and freely moving laboratory animals, without introducing either physically or chemically mediated stress or restraint artifacts in the data. It is generally assumed in contemporary pharmacology practice employing in vivo test systems that the quality of physiological measurements, when collected from conscious, freely moving animals, is superior to those obtained under physical or chemical restraint situations that are known to induce large-scale
D.V. Gauvin et al. / Journal of Pharmacological and Toxicological Methods 53 (2006) 128 – 139
deviations from normal homeostasis. Long-term continuous sampling of the physiological endpoint of interest using telemetry protocols that minimize distortions due to stressful or noxious environments is now commonplace in many laboratories. Moreover, there has been explicit acknowledgement in ICH and other regional regulatory drug safety testing guidelines that conditions present during routine telemetric monitoring may most closely approximate the normal physiological state of the animal, and therefore safety endpoints evaluated under such conditions may consequently demonstrate the greatest predictive validity to outcomes of similar testing in human beings. There are a number of advantages to the use of remote radio-monitoring in preclinical safety pharmacology: (1) telemetry-enabled observation provides the most humane method for monitoring physiologic endpoints in conscious, freely moving laboratory animals, eliminating the stress related to the use of restraint; (2) in the intervening decades since its inception, remote radiotelemetry has become more affordable and reliable, and now easy-to-use commercial products are readily available for monitoring a variety of physiological signals; (3) indwelling sensors for monitoring flow or pressure are usually more accurate and reliable than alternate noninvasive methods, such as transcutaneous laser-Doppler devices, tail cuff blood pressure monitors, etc.; and (4) biotelemetry allows for automated, high temporal resolution, continuous, and long-term data collection via computer, for days, weeks, or months, without any special animal care or maintenance. When used in tandem with remote video monitoring, radiotelemetry can be used to the exclusion of any direct human contact for prolonged periods of time so as to minimize this additional source of interference with physiologically normal baselines. The minimization of human contact during recording intervals can greatly augment the readability, as well as the reliability, of the data derived from emotional animals, whose physiological parameters demonstrate significant synchronous changes with such contact. One of the most important features of radiotelemetry, however, is the reduction of animal use by 60% to 70% in single dose studies (van Acker et al., 1996) and by more than 90% in multiple dose or repeat studies (Kinter, 1996; Kramer & Kinter, 2003). The ECG in beagle dogs has been previously described as demonstrating normal sinus arrhythmias. The irregularity in the ECG appears to be secondary to fluctuations in vagal tone associated with the respiratory cycle (Tilley & Goodwin, 2001). Significant differences are found in the ECG waveform measurements when the dog is assessed in the sitting position when compared to sternal recumbency or lateral recumbency. Coleman and Robson (2005) suggest that reference range values for right lateral recumbency are not valid for ECGs obtained in the sitting position or sternal recumbency (Coleman & Robson, 2005). The present study was a one-year retrospective examination of 123 kennel-raised beagle dogs (65 males, 58 females) surgically implanted with remote cardiovascular and hemodynamic radiotelemetry physiologic transducers. Remote monitoring of each dog’s baseline ECG, blood pressures (diastolic,
129
systolic, and mean arterial), and body temperature values were collected for 22 consecutive hours prior to any test initiation. The purposes of this study were four-fold: (1) to expand and supplement current knowledge on circadian stability of the serially and continuously sampled baseline ECG, blood pressures, and core body temperatures for clinically normal, kennel-born and reared, freely moving dogs within a laboratory setting; (2) to define the circadian characteristics of the ECG waveforms (RR, PR, QRS, QT, and QTc intervals) collected in individually housed laboratory beagle dogs; (3) to examine the relative usefulness of a number of correction formulas applied to normalize QT interval and heart rate functions in dogs; and (4) to compare these findings with a recent and similar report from this group using cynomolgus monkeys (Gauvin, Tilley, Smith, & Baird, 2005). 2. Methods 2.1. Regulatory guidelines All procedures, data review, collection, and analysis were conducted on equipment and software validated in accordance with the Standard Operating Procedures of MPI Research, Inc. (Mattawan, MI), and/or the United States Food and Drug Administration (FDA) Good Laboratory Practice Regulations, 21 CFR Part 58. Data presented were reviewed by the Quality Assurance Unit of MPI Research. 2.2. Subjects Sixty-five male and 58 female kennel-bred beagle dogs (Canis familiaris), approximately 5 –8 months of age, were purchased from Covance Research Products Inc. (Portage, MI) or Marshall Farms (North Rose, NY) and placed in stainless steel double-wide cages according to MPI Research Standard Operating Procedures. Following the testing facility’s standard operating procedures, complete physical, immunological, electrocardiographic, and hematological screening was used to verify the general well-being and good health of each dog. In full accordance and strict adherence to the National Institutes of Health, U.S. Department of Agriculture, U.S. Food and Drug Administration, and AAALAC guidelines, the dogs were approved to undergo surgical procedures to implant the radiotelemetry device. All experimental protocols and procedures were performed following approval of the MPI Research Institutional Animal Care and Use Committee. Baseline recordings occurred approximately 1 month following arrival (2-week acclimation, 2-week post-surgery recovery). Dogs were approximately 6 to 9 months of age when the baseline parameters reported in this study were monitored. The body weights were approximately 5.5 to 8 kg for male dogs and 5.0 to 7.5 kg for female dogs during the monitoring period. 2.3. Surgical implantations Both the surgical placement of the radiotelemetry devices and post-surgical care were conducted by the veterinary team
130
D.V. Gauvin et al. / Journal of Pharmacological and Toxicological Methods 53 (2006) 128 – 139
of the Clinical Medicine Department of MPI Research in a dedicated surgical suite using routine aseptic techniques in accordance with standard operating procedures. Briefly, dogs were fasted overnight prior to surgery and premedicated with carprofen 25 mg orally the morning of surgery. Just prior to surgery, atropine 0.05 mg/kg (IM), acepromazine 0.1 mg/kg (IM), and buprenorphine 0.1 mg/kg (SC) were administered. Anesthesia was induced by pentothal 14 mg/kg (i.v.) to effect. An endotracheal tube was inserted and general anesthesia was maintained with isoflurane 0.5 –2.5% delivered in oxygen via precision vaporizer and non-rebreathing anesthetic circuit. The animal also received cefazolin 25 mg/kg (i.v.) during the initial stages of surgery. The radio transmitter was placed on the dorsal lumbarsacral area in the subcutaneous space. The negative ECG lead was passed subcutaneously from the DSI transmitter to the right upper quadrant under the clavicle. The positive ECG lead was passed subcutaneously from the DSI transmitter to the left lower quadrant approximately 5 cm below the heart. The ECG leads were secured with nonabsorbable suture to the underlying musculature though the small skin incisions. The fluid-filled catheter articulated to the pressure transducer was placed in the femoral artery and advanced to the abdominal aorta. The transducer was anchored to the artery with nonabsorbable suture, all subcutaneous tissues were closed with absorbable suture, and the skin incisions were closed with skin staples and/or surgical glue. Each dog received cefazolin 20 mg/kg (i.m.) immediately post-operatively and cephalexin 250 mg (p.o.) post-operatively as needed and BID the day after surgery. Carprofen 25 mg (p.o.) or buprenorphine 0.01 mg/kg (s.c.) BID was given for 3 days post-surgery for pain control. Animals and their incisions were observed at least once daily during surgical recovery until staples were removed. At least a 2-week recovery period was allowed prior to the initiation of telemetry recordings. 2.4. Radiotelemetry recording A DataQuesti OpenARTi (Version 2.2, Transoma Medical, Inc., St. Paul, MN) biotelemetry acquisition system was linked to a Life Science Suitei, Gould/PoNeMahR (P3, version 3.322) data acquisition and analysis system (now LDS Test and Measurement, LLC, Valley View, OH). The Gould/ PoNeMahR system includes an embedded analog-to-digital conversion matrix linked to IBM-based personal computer systems. Each experimentally and drug-naı¨ve dog was equipped with a triple channel transmitter (Data Sciences, Inc., Model TL11M2-D70-PCT, St. Paul, MN). ECG waveforms were radio-transmitted along with the converted temperature and blood pressure signals through a Data Exchange Matrix (Model 20CH, DataSciences, Inc., St. Paul, MN). ECG waveforms were coded by the frequency of digital pulse trains from the Lead II spatially configured electrodes. These data epochs were examined through the short-term frequency changes from the input device. Each ECG
channel’s frequency was continuously collected at a rate of 500 cycles/s (Hz) and blood pressure signals are collected at a rate of 250 cycles/s. The Gould/PoNeMahR Physiology Platform was utilized to integrate the three independent signals, record and analyze the signals, and to plot the data on the computer screen for Freal-time_ review during data acquisition. A thermal array strip chart-type recorder/printer (AR200, GSI Lumonics, Billerica, MA) was interfaced with the computer-based monitoring system that enabled the printing of selected ECG waveforms for further analysis. 2.5. Radiotelemetry monitoring Each experimentally naı¨ve dog was monitored for 22 consecutive hours in its own stainless steel home cage (92 cm 82 cm 82 cm) under normal 12-h light/dark cycle. Temperature and humidity were monitored and recorded daily and maintained to the maximum extent possible between 18 and 29 -C and 30% to 70%, respectively. Each dog had ad libitum access to drinking water and certified dog chow. Environmental enrichments (KongR toys) were present during acclimation, during surgical recovery, and during telemetry data recording intervals. 2.6. ECG review The PoNeMahR software system utilizes pre-programmed, validated algorithms, constructed as independent modules based on accepted formulas for derived parameters extracted from the published scientific literature. Validation of the algorithms is done on a beat-to-beat basis while the system is acquiring and analyzing the data in real time. Digitized waveforms and raw data are saved in binary file format. The ‘‘Good Wave’’ counter in the software system counts the total number of complete complexes detected during the logging period. A complex is considered to be complete or ‘‘good’’ when the ‘‘Q’’, ‘‘P’’, and ‘‘T’’ waves are detected for the waveform. Once the 22 h of data collection had been completed, each animal’s ECG was visually inspected during a replay session in which the electronic image analyzer of the PoNeMah’sR waveform capture software was reviewed. Individual channel attributes for each waveform, in conjunction with low pass (3 Hz) and high pass filter settings (usually 50 Hz or 70 Hz) that attenuate extraneous frequencies associated with electrical signals correlated with respiration rates and external 60 Hz electrical noise, were reviewed by trained telemetry technicians during these replay sessions. The individually colored validation marks for each attribute (P wave, QRS interval, QT interval, etc.) for the replayed waveforms were reviewed and set to ensure the correct identification of each desired waveform component and, therefore, proper quantification of standard ECG intervals. For the current version of PoNeMah used for this study, the PR interval algorithm measures the distance from the beginning of the P wave to the peak of the Q wave interval as the PR interval. The data generated from each individual
D.V. Gauvin et al. / Journal of Pharmacological and Toxicological Methods 53 (2006) 128 – 139
replay session were then combined with those of other subjects on study for graphic representation of the data, as well as statistical comparisons and QT interval corrections. 2.7. QT interval corrections Regulatory authorities presently require an accurate and dependable measurement of the QT interval that is uncompromised by its inherent inverse relationship to heart rate. There is a current debate regarding the meaningful precision of heart rate correction of the QT interval and what specific correction method should be used (Batchvarov et al., 2002; Malik & Camm, 2001; Malik, Fa¨rbom, Batchvarov, Hnatkova, & Camm, 2002, Malik, Hnatkova, & Batchvarov, 2004). For comparative purposes, the present study used four computational methods to correct the QT intervals with respect to heartrate functions: Three ‘‘study-specific’’ methods were used to calculate QTc: ffiffiffiffiffi Bazett’sformula(1920): QTcB ¼ pQT RR ffiffiffiffiffi Fridericia’sformula (1920): QTcF ¼ pQT 3 RR Van de Water et al.’s (1989) formula: QTcV = QT 87(60 / HR 1)
131
seconds) is then determined from the equation: logðQTcÞ ¼ logðQTÞ bT½logðHRÞ logðHRmÞ QTc is obtained by the inverse log function:analysis of covariance: logðQTcÞ ¼ logðQTÞ bT½logðHRÞ logðHRmÞ b ¼ log regression coefficient HRm ¼ reference heart rate Each dog’s heart rate-corrected QT interval was then summarized to calculate and plot the group mean QTc. 2.8. Statistical analyses Main group (gender) effects were assessed by conducting repeated measures ANOVAs for each dependent measure recorded in this study. A desk-top windows-driven version of SAS (SASR v8.2 for Windows [SAS Institute Inc., Cary, NC, USA]) was used for the analyses. Statistical significance was set a p < 0.05.3. 3. Results
In addition, a fourth, modified, ‘‘subject-specific’’ method first described by Spence, Soper, Hoe, and Cole (1998) and modified by Miyazaki and Tagawa (2002) was used. The subject specific rate-correction utilizing linear model analysis of covariance (ANCOVA) was performed for individual subjects according to a modification of the original Spence et al. method (1998). In the Spence et al.’s correction, pooled group data are used to calculate a linear regression. In the Miyazaki and Tagawa (2002) method, the association or relation between QT and heart rate for each experimental subject is analyzed by linear regression to estimate the coefficient b for each subject. Having estimated b for each dog, the heart rate adjusted QT interval (in
Fig. 1 shows the core body temperature for male (left panel) and female (right panel) beagle dogs over the 22-h monitoring period. A low amplitude circadian rhythm can be seen in the telemetry recorded core body temperatures with the lowest body temperatures for both sexes being logged during the lights-out period (trend seen between 8 and 18 h on the abscissa). The group mean core body temperatures were maintained within a 1.5 -C bandwidth over the entire 22h monitoring period. Fig. 2 shows a well-controlled hemodynamic range of pressures over the 22 h of monitoring. Both systolic (SBP, red)
Fig. 1. Group mean core body temperatures (-C) continuously recorded by remote telemetry from freely moving male and female beagle dogs in their home cages over a 22-h period. Each dog’s body temperature was summarized into 5-min bins, grouped, and plotted over the monitoring period. The red lines indicate the upper and lower 95% confidence limits for the grand mean scores. Time ‘‘0’’ on the abscissa represents an approximate time of 10:30 a.m.
132
D.V. Gauvin et al. / Journal of Pharmacological and Toxicological Methods 53 (2006) 128 – 139
Fig. 2. Group mean systolic blood pressure (SPB: mm Hg; red), diastolic blood pressure (DBP: mm Hg; blue), and mean arterial pressure (MAP: mm Hg; green) continuously recorded by an implanted pressure transducer and transmitted by remote telemetry from freely moving male and female beagle dogs in their home cages over a 22-h period. Each dog’s pressures were summarized into 5-min bins, grouped, and plotted over the monitoring period. The colored lines indicate the upper and lower 95% confidence limits for the grand mean scores. Time ‘‘0’’ on the abscissa represents an approximate time of 10:30 a.m.
and diastolic (DBP, blue) blood pressures demonstrated an extremely stable pattern across the 22-h monitoring period. Mean arterial pressures (MAP, green) also remained within a limited normal operating window in both male and female dogs. There were statistically significant gender differences in systolic blood pressure ( F[1,121] = 9.2, p = 0.003), but not for diastolic or mean arterial pressures. Similar to core body temperatures, group mean heart rate (Fig. 3) for both gender of dogs demonstrated very little circadian rhythmicity across the day-long monitoring period. There was a main group (gender) effect found for heart rates ( F[1,121] = 123.86, p < 0.0001). The heart rates were quantified from the transduced blood pressure pulse signal. While the lowest heart rates were recorded at night (8 to 20 h on the
abscissa), daytime heart rates hardly differed from those recorded during nighttime hours. By comparing the heart rates (Fig. 3) with the quantified Rto-R intervals of the resulting measurements of the ECG waveforms (Fig. 4), the predicted inverse relationship can be seen. A slight circadian rhythm is demonstrated for the R-to-R intervals, more so in the male dogs when compared to their female cohorts. As would be suspected, with the significant main gender effect found for heart rate, there was a coincident statistically significant main group (gender) effect found for the R-to-R intervals as well ( F[1,121] = 202.25, p < 0.0001). In general, the nighttime interval, showing the lowest heart rates (Fig. 3), is associated with the longest R-to-R intervals in Fig. 4 (8 to 20 h on the abscissa).
Fig. 3. Group mean heart rate derived from the blood pressure pulse (beats per minute, bpm) continuously recorded by an implanted pressure transducer and transmitted by remote telemetry from freely moving male and female beagle dogs in their home cages over a 22-h period. Each dog’s heart rate was summarized into 5-min bins, grouped, and plotted over the monitoring period. The red lines indicate the upper and lower 95% confidence limits for the grand mean scores. Time ‘‘0’’ on the abscissa represents an approximate time of 10:30 a.m.
D.V. Gauvin et al. / Journal of Pharmacological and Toxicological Methods 53 (2006) 128 – 139
133
Fig. 4. Group mean R – R intervals (ms) derived from the ECG waveforms continuously recorded by a set of surgically implanted Lead II configured wires secured to the chest muscles of freely moving male and female beagle dogs. The recorded waveforms from each dog were transmitted by remote telemetry from in their home cages over a 22-h period. Each waveform was isolated, analyzed, and integrated into distinct morphological components by a computer-based waveform analysis program. Each dog’s R – R interval was summarized into 5-min bins, grouped, and plotted over the monitoring period. The red lines indicate the upper and lower 95% confidence limits for the grand mean scores. Time ‘‘0’’ on the abscissa represents an approximate time of 10:30 a.m.
The measurements of the PR intervals are shown in Fig. 5. The PR interval, as a measure of atrial and atrioventricular conduction, demonstrated similar stability over the 22h monitoring periods as body temperature and heart rate functions in both male and female dogs. There was no statistically significant difference between male and female dogs for PR intervals. Similar to the stable PR intervals in these beagles, Fig. 6 shows the stability of QRS durations for male and female dogs. Very little circadian rhythmicity can be detected over the monitoring interval for this aspect of ventricular function. There were no statistically significant differences between male and female dogs with respect to QRS durations. Fig. 7 shows the QT intervals. There were only limited, changes in QT intervals over the entire 22-h monitoring period.
There were no grouptime interaction effects found for this parameter, but there were main group (gender) effects ( F[1,121] = 7.02, p < 0.05). The results of the QT intervals corrected as a function of heart-rate using three common study-specific correction procedures (Bazett, 1920; Fridericia, 1920; Van de Water, Verheyen, Xhonneux, & Reneman, 1989) and a subjectspecific correction procedure (Miyazaki & Tagawa, 2002 [see Methods section for details]) are shown for both male (Fig. 8) and female (Fig. 9) beagle dogs over the 22-h monitoring period. While all four normalization procedures provided generally well fitted linear (straight-line) functions for heartrate correction of the QT intervals, only the subject-specific Miyazaki and Tagawa (2002) correction formula resulted in normalized QT intervals in the range of actual QT intervals
Fig. 5. Group mean PR (ms) derived from freely moving male and female beagle dogs (for details, see Fig. 4).
134
D.V. Gauvin et al. / Journal of Pharmacological and Toxicological Methods 53 (2006) 128 – 139
Fig. 6. Group mean QRS (ms) derived from freely moving male and female beagle dogs (for details, see Fig. 4).
recorded during the monitoring periods in the present study. All three study-specific correction formulas (Bazett, Fridericia, and Van de Water) resulted in normalized or corrected QT intervals that were elevated or, in general, greater than the actual averaged QT intervals recorded during the monitoring periods. Fig. 10 shows the number of good-waveforms collected over the 22-h monitoring period expressed as a percentage of total waveforms received from the beagle dogs. Overall, the current monitoring system was able to achieve a greater than 0.80 reliability measure of ECG waveforms over the entire monitoring period. During evening hours, when ‘‘systemic noise’’ is minimized, the group mean reliability of the radiowaveforms increased to almost 90%. The grand mean, standard deviation, and ranges for each of the cardiovascular parameters are summarized in Tables 1 –3. These values represent the normative range of radio telemetry-
recorded cardiovascular parameters in freely moving laboratory-bred beagle dogs. 4. Discussion There is a normal circadian rhythm to core body temperature in beagle dogs ranging from approximately 35 to 37 -C (Lawson, 1999). Beagles in the present study demonstrated a normal circadian range of core body temperatures over the 22 h of monitoring. The relatively stable hemodynamic parameters (diastolic and systolic pressures, and the mean arterial pressures) across the 22-h monitoring period in the present study reflect a group of normotensive, well-acclimated, freely moving, male and female laboratory-bred and -housed beagle dogs. The recorded values of pressures will vary with the breed, age, bodyweights, and health status of the dog. Anderson, Talan, and Engel
Fig. 7. Group mean QT (ms) derived from freely moving male and female beagle dogs (for details, see Fig. 4).
D.V. Gauvin et al. / Journal of Pharmacological and Toxicological Methods 53 (2006) 128 – 139
135
this study. Blood pressure is known to vary with age, body condition, gender, and breed in dogs (Littman & Fox, 1999). While statistically significant differences between the male and female dogs were found for systolic blood pressures and the heart rates, R-to-R intervals as well as the PR and QRS intervals, it is well known that serial dependency is a typical characteristic of time-series data (Gottman & Glass, 1978; Jones, Vaught, & Weinrott, 1977). Time-series studies using large sample sizes have often addressed the issue of statistical significance and meaningful change (Levin, 1975). Unfortunately, statistical and biological significance are not necessarily the same, particularly in the study of electrophysiological signals (Bramblett, 1994) as complex as cardiac function. A difference between two populations can always be established if the observer refines and increases the measurements sufficiently (Simpson, Roe, & Lewontin, 1960). Since variability in nature ensures that no two sets of observations are
Fig. 8. A comparative representation of the ability to dissociate the effect of heart rate on the QT interval between a subject-specific analysis of covariance formula (Miyazaki and Tagawa) and three other standard study-specific formulas (Bazett, Fridericia, and Van de Water et al.; see text). Mean QTs are plotted for all male dogs on study (top panel). For comparison, four different correction formulas were applied to these QT intervals and plotted separately for the same group of male dogs (bottom panel) across the entire 22h monitoring period.
(1990) have previously reported that dogs tend to be awake, eat, and drink intermittently throughout the day and night and that no progressive overnight changes in cardiac output or total peripheral resistance occurred in the laboratory dog. Hemodynamic function in laboratory-bred, and maintained, dogs is related to both environmental and species differences in sleep and ingestive behaviors. The dogs in the present study demonstrated hemodynamic values well within a restricted but well-controlled operating range of blood pressures that are similar to those reported in the previously published literature for laboratory maintained beagle dogs (Anderson et al., 1990; Matsunaga et al., 2001; Mishina et al., 1999; Miyazaki, Yoshida, Samura, Matsumoto, Ikemoto & Tagawa, 2002). While statistically significant differences were found between male and female dogs with respect to the systolic blood pressure parameter, these results were believed to be the result of the extraordinarily high number of datapoints analyzed in
Fig. 9. A comparative representation of the ability to dissociate the effect of heart rate on the QT interval between a subject-specific analysis of covariance formula (Miyazaki and Tagawa) and three other standard study-specific formulas (Bazett, Fridericia, and Van de Water et al.; see text). Mean QTs are plotted for all female dogs on study (top panel). For comparison, four different correction formulas were applied to these QT intervals and plotted separately for the same group of female dogs (bottom panel) across the entire 22-h monitoring period.
136
D.V. Gauvin et al. / Journal of Pharmacological and Toxicological Methods 53 (2006) 128 – 139
Fig. 10. The group mean number of good waves collected (expressed as a percentage of total waves) over the 22-h monitoring period for male and female beagle dogs. Each ECG waveform is transmitted to the Gould/PoNeMahR waveform analysis software package. Systemic, electrical, and/or random noise is censored by the software system and excluded from the analysis. The number of good waves is a general measure of system reliability and represents the proportion of cardiac contractile cycles where all relevant waveform features were detected and integrated relative to the total number of contractile cycles recorded at each interval. The red lines indicate the upper and lower 95% confidence limits of the group mean percentage of good waves collected. Time ‘‘0’’ on the abscissa represents an approximate time of 10:30 a.m. The greatest degree of collection of reliable waveforms appears to occur during the evening hours when the presence of personnel, use of electrical equipment, and institutional activity is minimal (approximately 85% good wave).
exactly the same, the investigator must be concerned with the biological significance of the statistically verified difference. Some rare events may be profoundly important while some significant variability may have little biological significance (Bramblett, 1994). The analysis of the electrocardiogram represents the advantages of visually analyzed charts over statistically finer-grained analyses. When examining for ‘‘between-group’’ differences in complex time-series experiments, reliable changes in the plotted rhythms are correlated with the known and timed introduction and removal of the independent variable. Thus, in the functional analysis of cardiac activity, as well as other complex events, it is this analytic function of the graph that has become of particular significance. This analysis is essentially a visual process; determination of change is dependent on the change being of sufficient magnitude to be apparent to the eye. Compared with the complex algebraic sophistication of statistical tests of significance, the visual inspection of data may appear simplistic, but it remains an important and valuable tool in the analysis of ongoing activity in physiological systems (Baer, 1977; Baer, Wolf, & Risley, 1968; Parsonson & Baer, 1978). While a number of gender-dependent differences were found in the present study, none of these particular parametric differences reflect physiologically meaningful events. There is a general circadian rhythm to heart rate variability in most species (Cornelissen, Haus, & Halberg, 1994; Hossmann, Fitzgerald, & Dollery, 1980; Lemmer, 1994; Lemmer, Scheidel, Blume, & Becker, 1991; Matsunaga et al., 2001). Normal resting heart rates in dogs can range from 70 to 220 beats/min, depending on the age of the dog. Barking or emotionality usually elicits higher heart rates that characterize the heart rhythm in dogs as ‘‘normal sinus tachycardia’’. In the beagle dog, daily variations in heart rate are related to short-
term changes caused by excitement and exercise (Miyazaki et al., 2002). It has been previously reported that blood pressures and heart rate increased at feeding and that meals tend to cycle across the full 24 h in laboratory dogs (Anderson et al., 1990; Mishina et al., 1999). These authors reported that heart rates generally decreased progressively after feeding. This variability can be seen in the ‘‘saw tooth’’ pattern of heart rates across the 22-h monitoring period in the present study. The stable pattern of heart rate changes recorded from freely moving beagle dogs in the present study and the very limited circadian rhythmicity in heart rate variability are similar to those reported by Miyazaki et al. (2002) and Anderson et al. (1990). The R-to-R interval reflects the time between the most intense phases of ventricular depolarization of one heart contraction to the peak of the next contraction. The peak of Table 1 Summarized ECG intervals for male and female beagle dogs Endpoint
Meana (ms)
S.D.
Min/max 95% confidence limits
ECG interval RR PR QRS QT QTcb
values in male beagle dogs 580.42 32.41 96.42 1.6 41.66 0.89 220.10 3.37 217.98 1.77
416.94 – 700.86 90.91 – 103.80 36.12 – 45.88 200.71 – 232.96 207.52 – 226.92
ECG interval RR PR QRS QT QTcb
values in female beagle dogs 518.82 23.30 90.82 1.85 41.33 0.76 217.60 3.56 215.54 1.87
407.50 – 610.70 85.99 – 99.07 22.20 – 36.82 198.34 – 231.90 205.00 – 226.85
a Grand mean values (ms) of all 5-min bin intervals averaged over the full 22h monitoring period. b Refers to Miyazaki and Tagawa (2002) correction values only.
D.V. Gauvin et al. / Journal of Pharmacological and Toxicological Methods 53 (2006) 128 – 139 Table 2 Summarized hemodynamic parameters for male and female beagle dogs Meana (mm Hg)
S.D.
Min/max 95% confidence limits
Blood pressure values in male beagle dogs Systolic blood pressure Diastolic blood pressure Mean arterial blood pressure
139 83 105
2 2 2
128.98 – 148.14 75.13 – 90.62 96.33 – 113.75
Blood pressure values in female beagle dogs Systolic blood pressure Diastolic blood pressure Mean arterial blood pressure
137 82 105
3 2 2
127.15 – 148.71 74.55 – 92.27 96.22 – 115.58
Endpoint
a Mean values represent the grand means of data collected in each individual animal across a 22-h interval.
the R wave is used clinically as an indicator of whether or not ventricular depolarization is proceeding over normal pathways, and to calculate heart rate. Comparative analyses of Figs. 3 and 4 reveal the inverse relationship between heart rates and R-to-R intervals. The dogs demonstrated a prolongation of the R-to-R intervals during the time-frame of the monitoring period in which heart rate decreases were recorded, and vice-versa. As would be predicted, the circadian variability is of approximately equivalent magnitude for the R-to-R interval when compared to the heart rates derived from the blood pressure pulse in these freely moving beagle dogs. The period of time from the initiation of the P wave to the beginning of the QRS complex is designated as the PR interval and indicates the time it takes for an action potential to spread through the atria to the atrioventricular (AV) node. The PR interval, therefore, largely reflects the time of atrial depolarization. The PR interval is usually reported to be within 60 to 130 ms in duration (Martin 2000; Miller, Tilley, Smith, & Fox, 1999; Tilley & Goodwin, 2001) in beagles. The present study found a more restricted circadian variability and range of approximately 10 ms (¨ 90 to 100 ms) for PR intervals in freely moving laboratory-bred, raised, and maintained beagle dogs. The QRS complex reflects the time required for ventricular depolarization. The normal range for QRS duration for the beagle dogs reported in the published literature is 40 to 50 ms (Martin, 2000; Miller et al., 1999; Tilley, 1992; Tilley & Goodwin, 2001). The QRS durations reported in the present study, using freely moving dogs, are consistent with the range of durations of these intervals documented in the literature. The normal range of uncorrected QT intervals in the dog reported in the published literature is 150 to 250 ms at normal heart rates (Martin, 2000; Miller et al., 1999; Tilley, 1992; Tilley & Goodwin, 2001). QT variability is a non-invasive marker of cardiac repolarization lability and a higher QT variability has been associated with sudden death (Nolan, Girand, Bailie, & Yeragani, 2004). The range of the 95% confidence intervals for the QT interval in the present study was within a more restricted range for both male and female dogs than previously reported. Nolan et al. (2004) have recently reported a diurnal pattern of QT intervals similar to healthy humans. Similar to almost all cardiac parameters
137
recorded in the present study there was limited QT interval variability in the 123 dogs used in this study. Heart rate-corrected intervals were obtained using standard study-specific correction formulas used in a number of laboratories and first described by Bazett (1920), Fridericia (1920), and Van de Water et al. (1989). In contrast, we also used a subject-specific correction formula derived from a modification of a method first described by Spence et al. (1998) (cf. Miyazaki & Tagawa, 2002). The QT interval correction for heart rate (QTc) is believed to reflect sympathovagal balance (Lemarre-Cliche´, Lacourciere, de Champlain, Poirier, & Larochelle, 2003), and it is the traditional method of assessing the duration of ventricular repolarization (Crow, Hannan, & Folsom, 2003). Although many mathematical models have been proposed to describe the relationship between the QT interval and heart rate, none of the models (and corresponding heart rate correction formulas) has been accepted as universally applicable (Aytemir et al., 1999; FunckBrentano & Jaillon, 1993; Hodges, 1997; Malik, 2001; Rautaharju, Warren, & Calhoun, 1990). As many authors have pointed out, any formula describing the relationship between the QT interval and the preceding heart rate is limited by physiological factors, such as the extra-heart rate influences on myocardial repolarization (Browne, Prystowsky, Heger, & Zipes, 1983; Cappato, Alboni, Pedroni, Gilli, & Antonioli, 1991; Rickards & Norman, 1981) and the ‘‘lag hysteresis’’ of the QT –heart rate relationship (Franz, Swerdlow, Liem, & Schaefer, 1988; Lau et al., 1988). Recent approaches to the issue of heart rate correction of the QT interval have largely abandoned the use of fixed rate correction formulas (also referred to as ‘‘study-specific corrections’’ such as Bazett, 1920; Fridericia, 1920; Van de Water et al., 1989, etc.) and have adopted an individual rate correction derived in each experimental subject from multiple observations of the QT – heart rate data in a baseline, drug-free state (also referred to as ‘‘subject-specific corrections’’). The QT – heart rate relationship has been shown to have substantial between-subject variability as well as high within-subject stability (Batchvarov et al., 2002). Relative to the above, a growing number of authors have recently shifted away from formulas relying upon heterogeneous, pooled estimates of the QT/ HR relationship in favor of formulas individualized for each experimental subject (Batchvarov et al., 2002; Malik & Camm, 2001; Malik et al., 2002; Smetana, Batchvarov, Hnatkova, Camm, & Malik, 2003). According to the Food and Drug Administration’s Center for Drug Evaluation and Table 3 Summarized heart rate values for male and female beagle dogs Endpoint
Meana (bpm)
S.D.
Heart rate values in male beagle dogs Heart rate 111 6 Heart rate values in female beagle dogs Heart rate 120 5 a
Min/max 95% confidence limits 89.77 – 149.65
102.35 – 150.88
Mean values (beats/min) represent the grand means of data collected in each individual animal across a 22-h interval.
138
D.V. Gauvin et al. / Journal of Pharmacological and Toxicological Methods 53 (2006) 128 – 139
Research (CDER), Office of Drug Evaluation (ODE1), there is wide agreement that Bazett’s correction is not the best correction formula to use in many cases and most researchers now believe that the regression approach is better (Huml, Barker, Tonkens, & Hebert, 2004). Our data support this conclusion. A recent study by Malik et al. (2004) demonstrated that the subject-specific corrections (Spence-like) led to maximum errors in single millisecond range (2.4, 5.7, and 2.6 ms with linear, log/log linear, and exponential models, respectively), while the study-specific corrections (Bazett and Fridericia’s) led to substantially greater errors (error range of 17.8, 19.4, and 16.9 ms with linear, log/log linear, and exponential models, respectively). These authors found that both Bazett and Fridericia’s corrections led not only to substantial errors (error range of 28.3 and 16.9 ms), but also to regular bias with systematically false negative and false positive conclusions dependent on modeled heart rate acceleration and deceleration. The present study clearly found inflated QTc values above any recorded actual uncorrected QT interval recorded during the study. All three study specific correction formulas over-estimated the QTc values and, therefore, misrepresented the actual data. Similar to our recent report on the cardiovascular function of cynomolgus monkeys, the subjectspecific Miyazaki and Tagawa correction provided the best heart-rate correction for QT interval in beagle dogs in the present study. 5. Conclusion The present study has provided normal operating range values for core body temperature, hemodynamic, and ECG parameters for well-acclimated, freely moving laboratoryhoused beagle dogs using radiotelemetry recording methodologies. With a collection speed of 500 cycles/s for 22 h of continuous monitoring of a complete set of endpoints, the current system achieved a minimum signal integration accuracy of greater than 80%. In a companion paper examining the cardiovascular, hemodynamic, and body temperature changes in cynomolgus monkeys, there were clear circadian variations in almost all parameters across the 22-h monitoring period. In the present study, there was little evidence of large amplitude circadian rhythms in any measured parameter in these beagle dogs. While the extremely stable, normotensive data from the present study may not represent the ‘‘normal’’ variations recorded in canine pets seen within a veterinary clinic, they are believed to represent the ‘‘normal’’ laboratory-bred and raised dog that is well-acclimated, not stressed by the environment, and well adjusted to the daily patterns of activity found within a Contract Research Organization. The differences in the degree of human contact, meal availability, and environmental changes that the ‘‘normal’’ canine pet is exposed to may influence the cardiovascular and hemodynamic parameters recorded within a standard veterinary clinic and may underlie the differences between the current data and those that may be derived by most primary care veterinarians in the clinic.
Radiotelemetry substantially reduces animal use and associated research costs by increasing the number of parameters that can be collected from a group of animals and thereby eliminating the need for satellite groups or separate studies to collect the same data, by permitting the use of blocked study designs, and by increasing the data precision in both individual animals and study groups. Radiotelemetry also offers refinement alternatives by reducing animal stress and discomfort, improving the quality and precision of experimental data, and reduces the need for isolation from other animals of the same species while reducing the amount of direct hands-on contact with trained laboratory staff (Kramer & Kinter, 2003). We believe these procedures represent contemporary industry preferred practice for measuring such parameters under the ICH guidelines and are amenable to routine use in a variety of other relevant safety/efficacy studies. Acknowledgement The work described in this manuscript was paid for by MPI Research, Inc. References Anderson, D. E., Talan, M. I., & Engel, B. T. (1990). Nocturnal hemodynamic patterns in dogs. Physiology and Behavior, 48, 485 – 487. Aytemir, K., Maarouf, N., Gallagher, M. M., Yap, Y. G., Waktare, J. F., & Malik, M. (1999). Comparison of formulae for heart rate correction of QT interval in exercise electrocardiograms. Pacing and Clinical Electrophysiology, 22, 1397 – 1401. Baer, D. M. (1977). Perhaps it would be better not to know everything. Journal of Applied Behavioral Analysis, 10, 167 – 172. Baer, D. M., Wolf, M. M., & Risley, T. R. (1968). Some current dimensions of applied behavior analysis. Journal of Applied Behavioral Analysis, 1, 91 – 97. Batchvarov, V. N., Ghuran, A., Smetana, P., Hnatkova, K., Harries, M., Dilaveris, P., et al. (2002). QT – RR relationship in healthy subjects exhibits substantial intersubject variability and high intrasubject stability. American Journal of Physiology, 282, H2356 – H2363. Bazett, H. C. (1920). An analysis of the time-relations of electrocardiograms. Heart, 7, 258 – 353. Bramblett, C. A. (1994). Patterns of primate behavior. Prospect heights. IL’ Waveland Press. Browne, K. F., Prystowsky, E., Heger, J. J., & Zipes, D. P. (1983). Modulation of the Q – T interval by the autonomic nervous system. Pacing and Clinical Electrophysiology, 6, 1050 – 1056. Cappato, R., Alboni, P., Pedroni, P., Gilli, G., & Antonioli, G. E. (1991). Sympathetic and vagal influences on rate-dependent changes of QT interval in healthy subjects. American Journal of Cardiology, 68, 1188 – 1193. Coleman, M. G., & Robson, M. C. (2005). Evaluation of six-lead electrocardiograms obtained from dogs in a sitting position or sternal recumbancy. American Journal of Veterinary Research, 66, 233 – 237. Cornelissen, G., Haus, E., & Halberg, F. (1994). Chronobiologic blood pressure assessment from womb to tomb. In Y. Touitou, & E. Haus (Eds.), Biological rhythms in clinical and laboratory medicine (pp. 428 – 452). New York’ Springer Verlag. Crow, R. S., Hannan, P. J., & Folsom, A. R. (2003). Prognostic significances of corrected QT and corrected JT interval for incident coronary heart disease in a general population sample stratified by presence or absence of wide QRS complex: The ARIC Study with 13 years of follow-up. Circulation, 108, 1985 – 1989. Franz, M. R., Swerdlow, C. D., Liem, L. B., & Schaefer, J. (1988). Cycle length dependence of human action potential duration in vivo. Effects of single
D.V. Gauvin et al. / Journal of Pharmacological and Toxicological Methods 53 (2006) 128 – 139 extrastimuli, sudden sustained rate acceleration and deceleration, and different steady-state frequencies. Journal of Clinical Investigation, 82, 972 – 979. Fridericia, L. S. (1920). Die systolendauer im electrokardiogramm bei normalen menschen und bei herzkranken. Acta Medica Scandinavica, 53, 469 – 486. Funck-Brentano, C., & Jaillon, P. (1993). Rate-corrected QT interval: Techniques and limitations. American Journal of Cardiology, 72, 17B – 22B. Gauvin, D.V., Tilley, L.P., Smith, F.W.K. Jr., Baird, T.J. (2005). Electrocardiogram, hemodynamics, and core body temperatures of the normal freelymoving cynomolgus monkey by remote telemetry. Journal of Pharmacology and Toxicological Methods 53, 87 – 98. Gottman, J. M., & Glass, G. V. (1978). Analysis of interrupted time-series experiments. In T. R. Kratochwill (Ed.), Single subject research: Strategies for evaluating change (pp. 197 – 237). New York’ Academic Press. Hodges, M. (1997). Rate-correction of the QT interval. Cardiac Electrophysiology Review, 3, 360 – 363. Hossmann, V., Fitzgerald, G. A., & Dollery, C. T. (1980). Circadian rhythm of baroreflex reactivity and adrenergic vascular response. Cardiovascular Research, 14, 125 – 129. Huml, R. A., Barker, G., Tonkens, R. M., & Hebert, D. A. (2004, May). The clinical and regulatory implications of QT interval prolongation. Regulatory Affairs Focus, , 40 – 43. International Conference on Harmonisation of Technical Requirements for Registration of Pharmacetuticals for Human Use (10 June, 2004). The nonclinical evaluation of the potential for delayed ventricular repolarization (QT interval prolongation) by human pharmaceuticals, S7B. . Jones, R. R., Vaught, R. S., & Weinrott, M. (1977). Time-series analysis in operant research. Journal of Applied Behavioral Analysis, 10, 151 – 166. Kinter, L. B. (1996). Cardiovascular telemetry and laboratory animal welfare: New reduction and refinement alternatives (Abstract). General Pharmacology/Safety Pharmacology Meeting, Philadelphia, PA. , . Kramer, K., & Kinter, L. B. (2003). Evaluation and applications of radiotelemetry in small laboratory animals. Physiological Genomics, 13, 197 – 205. Lau, C. P., Freedman, A. R., Fleming, S., Malik, M., Camm, A. J., & Ward, D. E. (1988). Hysteresis of the ventricular paced QT interval in response to abrupt changes in pacing rate. Cardiovascular Research, 22, 67 – 72. Lawson, P. T. (Ed.). (1999). American association for laboratory animal science assistant laboratory animal technician training manual. Chelsea (MI)’ Sheridan Books, Inc. Lemarre-Cliche´, M., Lacourciere, Y., de Champlain, J., Poirier, L., & Larochelle, P. (2003). Does QTc interval predict the response to betablockers and calcium channel blockers in hypertensives. Heart Disease, 5, 244 – 252. Lemmer, B. (1994). Cardiovascular chronobiology and chronopharmacology. In Y. Touitou, & E. Haus (Eds.), Biological rhythms in clinical and laboratory medicine (pp. 418 – 427). New York’ Springer Verlag. Lemmer, B., Scheidel, B., Blume, H., & Becker, H. -J. (1991). Clinical chronopharmacology of oral sustained-release isosorbide-5-mononitrate in healthy subjects. European Journal of Clinical Pharmacology, 40, 70 – 75. Levin, J. R. (1975). Determining the sample size for planned and post hoc analysis of variance comparisons. Journal of Educational Measurement, 12, 99 – 108. Littman, M. P., & Fox, P. R. (1999). Systemic hypertension: Recognition and treatment. In P. R. Fox, D. Sisson, & N. S. Moise (Eds.), Textbook of canine and feline cardiology: Principles and clinical practice (2nd edR). Philadelphia’ Saunders. Malik, M. (2001). Problems of heart rate correction in assessment of druginduced QT interval prolongation. Journal of Cardiovascular Electrophysiology, 12, 411 – 420.
139
Malik, M., & Camm, A. J. (2001). Evaluation of drug-induced QT interval prolongation: Implications for drug approval and labeling. Drug Safety, 24, 323 – 351. Malik, M., Fa¨rbom, P., Batchvarov, V., Hnatkova, K., & Camm, A. J. (2002). Relation between QT and RR intervals is highly individual among healthy subjects: Implications for heart rate correction of the QT interval. Heart, 87, 20 – 228. Malik, M., Hnatkova, K., & Batchvarov, V. (2004). Differences between studyspecific and subject-specific heart rate corrections of the QT interval in investigations of drug induced QTc prolongation. Pacing and Clinical Electrophysiology, 27(1), 791 – 800. Martin, M. (2000). Small animal ECG’s: An introductory guide. Oxford, UK’ Blackwell Science. Matsunaga, T., Harada, T., Mutsui, T., Inokuma, M., Hashimoto, M., Miyauchi, M., et al. (2001). Spectral analysis of circadian rhythms in heart rate variability of dogs. American Journal of Veterinary Research, 62, 37 – 42. Miller, M. S., Tilley, L. P., Smith, F. W. K., Jr., & Fox, P. R. (1999). Electrocardiography. In P. R. Fox, D. Sisson, & S. Moise (Eds.), Textbook of canine and feline cardiology: Principles and clinical practice (2nd ed.). Philadelphia’ Saunders. Mishina, M., Watanabe, T., Matsuoka, S., Shibata, K., Fukii, K., Maeda, H., et al. (1999). Diurnal variations of blood pressure in dogs. Journal of Veterinary Medical Science, 61, 643 – 647. Miyazaki, H., & Tagawa, M. (2002). Rate-correction technique for QT interval in long-term telemetry ECG recording in beagle dogs. Experimental Animals, 51(5), 465 – 475. Miyazaki, H., Yoshida, M., Samura, K., Matsumoto, H., Ikemoto, F., & Tagawa, M. (2002). Ranges in diurnal variation and the pattern of body temperature, blood pressure and heart rate in laboratory beagle dogs. Experimental Animal, 51, 95 – 98. Nolan, E. R., Girand, M., Bailie, M., & Yeragani, V. K. (2004). Circadian changes in the QT variability index in the beagle dog. Clinical and Experimental Pharmacology and Physiology, 31, 783 – 785. Parsonson, B. S., & Baer, D. M. (1978). The analysis and presentation of graphic data. In T. R. Kratochwill (Ed.), Single subject research: Strategies for evaluating change (pp. 101 – 166). New York’ Academic Press. Rautaharju, P. M., Warren, J. W., & Calhoun, H. P. (1990). Estimation of QT prolongation. A persistent, avoidable error in computer electrocardiography. Journal of Electrocardiology, 23, 111 – 117. Rickards, A. F., & Norman, J. (1981). Relation between QT interval and heart rate. New design of physiologically adaptive cardiac pacemaker. British Heart Journal, 45, 56 – 61. Simpson, G. G., Roe, A., & Lewontin, R. C. (1960). Quantitative zoology. NY’ Harcourt, Brace and World. Smetana, P., Batchvarov, V., Hnatkova, K., Camm, A. J., & Malik, M. (2003). Circadian rhythm of the corrected QT interval: Impact of different heart rate correction models. Pacing and Clinical Electrophysiology, 26, 383 – 386. Spence, S., Soper, K., Hoe, C. -M., & Coleman, J. (1998). The heart ratecorrected QT interval of conscious beagle dogs: A formula based on analysis of covariance. Toxicological Science, 45, 247 – 258. Tilley, L. (1992). Essentials of canine and feline electrocardiography (3rd edR). England’ Blackwell Publishing. Tilley, L. P., & Goodwin, J. -K. (2001). Manual of canine and feline cardiology (3rd edR). Philadelphia’ Saunders. van Acker, S.A.B.E, Kramer, K., Voest, E. E., Grimbergen, J. A., Zhang, J., van der Vijgh, W. J. F., et al. (1996). Doxorubicin-induced cardiotoxicity monitored by ECG in freely moving mice. A new model to test potential protectors. Cancer Chemotherapy and Pharmacology, 38, 95 – 101. Van de Water, A., Verheyen, J., Xhonneux, R., & Reneman, R. S. (1989). An improved method to correct the QT interval of the electrocardiogram for changes in heart rate. Journal of Pharmacological Methods, 22, 207 – 217.