- Email: [email protected]
Higher-order polynomial modeling of rabbit epicardial ventricular electrograms
Poster 4 Introduction: The effects of APD restitution (APD-R) on wave propagation are usually modelled assuming uniform restitution properties or a smooth uniform gradient as seen in some animal models. Spatial heterogeneity of APD-R has been shown to enhance electrical instability and be pro-arrhythmic. The spatial properties of APD-R in humans are unknown. Methods: We studied 14 patients undergoing cardiac surgery for coronary artery disease (CAD; n⫽8) or aortic valve disease (AVD; no significant CAD; n⫽6). Prior to the commencement of cardiopulmonary bypass, activation-recovery intervals, as a conventional surrogate for APD, were recorded using an epicardial sock (256 electrodes) connected to a UnEmap recording system. Using an S1-S2 pacing protocol, standard restitution curves were constructed using a monoexponential fit to each set of electrode data (biphasic restitution data were observed for ⬍2% of the recordings). Results: Smooth uniform spatial gradients of APD-R slope were absent for all 14 subjects. Instead, multiple gradients between areas of steep and shallow slope were observed. Illustrated is an example of one of several such regions for an AVD patient. Overall slopes for all electrode sites were steeper for AVD patients particularly at flatter slopes. Overall mean slopes for AVD patients were (mean⫾SEM: 1.23⫾0.03) versus CAD patients (1.00⫾0.02, p⬍0.01); and steeper over the LV than the RV (AVD: 1.35⫾0.04 vs 1.13⫾0.03, p⬍0.01; CAD: 1.06⫾0.03 vs 0.94⫾0.03, p⬍0.01). Conclusions: In humans APD restitution is flatter in CAD patients and steeper in AVD patients particularly in the left ventricle which may have important implications for the behaviour and interpretation of wave propogation and arrhythmogenesis.
P4-8 SEGMENTAL VERSUS TRANSMURAL REMODELING AS ELECTROPHYSIOLOGICAL BASIS FOR T-WAVE MEMORY Darwin D. Jeyaraj, MD, Lance D. Wilson, MD, Steven Poelzing, PhD, Xiaoping Wan, PhD and David S. Rosenbaum, MD. Metrohealth Campus, Case Western Reserve University, Cleveland, OH. Although the T-wave is thought to arise from transmural repolarization gradients (TRG), the electrophysiological basis for T-wave memory remains unclear. Recently, we demonstrated that short term pacing diminished TRG, suggesting an alternate mechanism for T-wave memory. Therefore, we hypothesize that segmental repolarization gradients (SRG), i.e. action potential duration (APD) differences between different segments of left ventricle (LV), form the electrophysiological basis for T-wave memory. Dogs (n⫽5) were paced (VDD mode) from base of the anterior LV epicardium for 3-4 weeks. T-wave memory was characterized by change in T-wave vector angle (35 degrees), vector magnitude (0.9 mV), and QT prolongation (by 43⫾21 ms, p⬍0.001) compared to baseline. High-resolution optical mapping was used to record action potentials from transmural surface of arterially perfused canine wedge preparations harvested separately from LV segments proximal (anterior base) and distal (lateral and posterior base) to the site of pacing. TRG was defined as the APD difference between epicardial and M cells within a segment, whereas SRG was defined as difference in mean APD between any 2 segments. Identical measurements were made in 4 unpaced dogs (controls). In controls, TRG was significantly greater than SRG (38⫾10 ms vs. 10⫾2 ms, p⬍0.001), supporting the hypothesis that TRG underlies the normal T wave. Conversely, SRG were significantly larger than TRG in T-wave memory (73⫾21 ms vs. 38⫾10 ms, p⬍0.004). Memory-induced SRG was
S217 attributable to preferential prolongation of APD in posterior or late activated segments of myocardium (297⫾21 ms) compared to proximal segments (249⫾11 ms, p⬍0.03). Conclusions: These data provide direct experimental evidence that TRG and not SRG underlies the normal T wave. By contrast, the T wave changes (including QT prolongation) underlying T-wave memory were caused by SRG arising from preferential electrophysiological remodeling of LV segments distal to the site of pacing. These data suggest an intriguing hypothesis that altered mechanical stresses known to occur at distal sites are responsible for T-wave memory. P4-9 LOVASTATIN IS ANTIARRHYTHMIC IN ISCHEMIC HEART TISSUE BY BLOCKING TRIGGERED ACTIVITY Dezhi Xing, MD, Raymond Hohl, MD, PhD and James Martins, MD. University of Iowa College of Medicine and VAMC, Iowa City, IA. Background: AVID and MADIT-II trials showed that patients with coronary artery disease taking HMG-CoA reductase inhibitors have fewer ICD shocks suggesting anti-arrhythmic effects. Our aim was to analyze the effects of acutely administered lovastatin (LOVA): 1) on ICD defibrillation threshold (DFT) since anti-arrhythmics can worsen DFT, and 2) on triggered activity (TA) recorded from ischemic endocardium excised from the anterior wall of anesthetized dogs with anterior descending coronary occlusion. Methods and Results: Dogs received LOVA dissolved in DMSO. DMSO alone produced no effect on DFT. LOVA 10mg IV did not alter effective refractory period and tended to improve defibrillation success (from 20⫾3 to 17⫾3 J, P⫽0.09 vs control). Ischemic endocardium was studied using standard microelectrode techniques with normal Tyrode’s solution. There were no delayed afterdepolarizations (DADs) or TA at baseline in this depolarized tissue (-71⫾ 4mV). Isoproterenol (5x10-7M) superfusion, produced sustained TA with DAD amplitudes of 4.2 ⫾1 mV. LOVA (10-7 M) produced no change in resting membrane potential (-70⫾4 mV), action potential duration at 90% (from 231⫾11 to 229⫾8 msec, ns), action potential duration at 50% (168⫾7 to 166⫾6 msec, ns), action potential amplitude (89⫾5 to 85⫾5 mV, ns), overshoot (11⫾1 to 10⫾1 mV) or DAD amplitude (4.0⫾1 mV, ns). Interestingly, TA was attenuated to 1-5 complexes with LOVA (p⬍0.05) and not with DMSO alone. Reversibility to sustained TA also occurred after LOVA washout (20 min) as well as with co-superfusion with mevalonic acid (10-7 M, n⫽4), which also did not change action potentials or DADs. Conclusions: These results are consistent with the possibility that a prenylated protein down-stream from the mevalonic acid pathway may underlie this effect. Thus LOVA in concentrations achievable in human plasma may be anti-arrhythmic for TA without a deleterious effect on DFT. P4-10 HIGHER-ORDER POLYNOMIAL MODELING OF RABBIT EPICARDIAL VENTRICULAR ELECTROGRAMS Jeffrey L. Williams, MD, *Leonard I. Ganz, MD, Vladimir Shusterman, MD, PhD, Barry London, MD, PhD and Samir Saba, MD. University of Pittsburgh, Pittsburgh, PA. Objective: To construct a higher-order model of the rabbit’s epicardial ventricular electrogram (EGM). Background: A higher-order representation of the ventricular EGM allows the capture of depolarization and repolarization phenomena using a limited number of coefficients. Methods: New Zealand White rabbits underwent chronic implantation of pacemakers through a left thoracotomy approach. Unipolar ventricular EGMs sampled at a frequency of 1 kHz were stored digitally in one-minute segments before and after intravenous injection of isoproterenol or procainamide. Each cardiac cycle was divided into a QR and an RST segment which were modeled separately using a 6th-order polynomial equation. A
S218 6th-order equation was the smallest order model that faithfully captured the morphology of the raw EGM (norm of residuals ⬃ 1.0). Results: The 14 coefficients of each cardiac cycle were stable and reproducible throughout the baseline recordings (r ⬎⫽ 0.94, p ⬍ 0.002). Isoproterenol caused no changes in the coefficients of the QR-segment but significantly altered all but one of the 7 coefficients of the RST-segment (p6⫽0.0039, p5⫽0.017, p4⫽0.00007, p3⫽0.112, p2⫽0.00016, p1⫽0.0086, pa⫽0.00003). Procainamide caused statistically significant changes in both QR-segment (p6⫽0.018, p5⫽0.287, p4⫽0.019, p3⫽0.176, p2⫽0.016, p1⫽0.362, pa⫽0.000044) and RST-segment (p6⫽0.0028, p5⫽0.036, p4⫽0.002, p3⫽0.058, p2⫽0.022, p1⫽0.718, pa⫽0.0018) coefficients. Conclusion: Our data demonstrate the feasibility of a higher-order polynomial equation that models the rabbit ventricular EGM. This model is stable, reproducible, and can predict the changes expected with antiarrhythmic drug administration. If reproduced in humans, these findings can have wide applications in patients with implantable devices, ranging from morphologic discrimination of arrhythmias to early detection of metabolic derangements or drug effects. P4-11 ANGIOTENSIN II TYPE 1 RECEPTOR (AT1R) BLOCKADE HAS NO EFFECT ON AF SELF-PERPETUATION OVER 4 WEEKS IN BURST PACED GOATS Mark C. S. Hall, MRCP, Senthil Kirubakaran, MRCP, Clifford J. Garratt, MD and Nicholas S. Peters, MD, FRCP. Manchester Heart Centre, Manchester, United Kingdom and Imperial College, London, United Kingdom. Background: Previous studies suggest that AT1R blockade prevents shortterm electrical remodeling and longer-term (4 weeks) structural remodeling in a rapidly-paced dog model of AF. We examined the effects of AT1R blockade with candesartan on the self-perpetuation of AF by 4 weeks of atrial burst pacing in a goat model. Method: 8 goats underwent pacemaker implantation. Repeated bursts of atrial stimuli (2s, 50Hz, 4x diastolic threshold) were applied as in the original Allessie protocol. 4 goats received oral candesartan (0.5mg/kg/ day) for 1 week prior to and during 4 weeks of burst pacing, whilst 4 received placebo. AERP, AF duration, AF cycle length (AFCL), and sinus cycle length (SCL) were recorded where appropriate twice daily for 5 days, then daily. In 6 separate goats, biological activity of candesartan (0.5mg/ kg/day) was determined by measuring its effect on the pressor response to infused angiotensin II (up to 40ng/kg/min). Results: Candesartan compared to burst pacing alone had no effect on SCL (595⫾9ms vs 581⫾16ms, p⫽ns), the AERP at baseline (155⫾21.8ms vs 155⫾19.4ms, p⫽ns), the fall in AERP with atrial pacing (after 3 days 77.5⫾2.5ms vs 85⫾15.5ms, p⫽ns) or AFCL (103⫾2.9ms vs 103⫾1.4ms, p⫽ns). Candesartan had no demonstrable effect on the increasing stability of AF with time (figure). There was significant suppression of the pressor response to angiotensin II (40ng/kg/min angiotensin II produced a 8.1⫾3.2mmHg systolic BP rise in candesartan treated goats and 34⫾1.9mmHg in controls, p⬍0.002). Conclusions: Candesartan is absorbed orally in the goat and has activity at the AT1 receptor. Blockade of the AT1 receptor has no effect on atrial refractory period or induced AF duration during a 4-week protocol of atrial burst pacing in the conscious goat model.
Heart Rhythm, Vol 2, No 5, May Supplement 2005 P4-12 EXPERIMENTAL MODEL FOR APNEA INDUCED ATRIAL FIBRILLATION Muhammad Ghias, MD, *Benjamin J. Scherlag, PhD, William S. Yamanashi, PhD, Sunny S. Po, MD, PhD, *Hiroshi Nakagawa, MD, PhD, Ralph Lazzara, MD and *Warren M. Jackman, MD. Cardiac Arrhythmia Research Institute at the University of Oklahoma Health Science Center, Oklahoma City, OK and Cardiac Arrhythmia ResCardiac ArrhythCardiac Arrhythmia Research Institute at the University of Oklahoma Health Science Center, Oklahoma City, OK. Recent clinical reports have shown a close association between sleep apnea and atrial fibrillation (AF). Methods: In 10 Na-pentobarbital anesthetized dogs, under positive pressure ventilation, a right thoracotomy allowed placement of electrode catheters at the right superior pulmonary vein-atrial (RSPV-A) junction and right atrial appendage (RAA). Programmed electrical stimulation at a cycle length of 330ms (2X-20X threshold) was performed at both sites to determine atrial refractory periods (ARP) and AF inducibility. A insulated metal electrode except for the tip was inserted and stabilized in a cluster of autonomic ganglia adjacent to the RSPV to continuously record neural activity during control and apnea periods. Respiration was stopped during end expiration and intermittent PES was instituted at an intensity of stimulation at which no AF occurred during the control state and at 5-10ms shorter than the ARP at that level. Standard ECG leads, blood pressure (BP) and core temperature were continuously monitored. Results: During a 2 minute period of induced apnea, S1-S2 which initially failed to activate the atrium began to induce single then multiple responses leading to atrial tachycardia (AT) and AF. AF occurred at an average time of 85⫾38 sec after the onset of apnea. The entire sequence was associated with a significant increase in systolic BP compared to control (control, 149⫾26mmHg; apnea/AF, 193⫾38mmHg, p⫽0.05) but not diastolic BP (control, 108⫾14mmHg; apnea/AF, 122⫾20mmHg, p⫽0.08). Neural potentials recorded from AGs increased in frequency with “bursting” which preceded the onset of any electrophysiologic effect. Autonomic blockade with intravenous atropine, 2mg, and esmolol or propranolol, 1mg/kg, prevented apnea induced AF and was associated with a marked fall in BP. Conclusions: Induced apnea in the anesthetized dog consistently resulted in AT/AF associated with both increased systolic BP and neural firing in local autonomic ganglia. The response to autonomic blockade which suppressed the arrhythmias suggests a strong autonomic basis for apnea induced AF. P4-13 MAPPING THE ONSET OF ATRIAL FIBRILLATION ARISING FROM PULMONARY VEIN AND NON-PULMONARY VEIN SITES IN THE DOG HEART *Benjamin J. Scherlag, PhD, William S. Yamanashi, PhD, *Jeff Edwards, RN, Neng Geng, MD, Sunny S. Po, MD, PhD, Ralph Lazzara, MD and *Warren M. Jackman, MD. Cardiac Arrhythmia Research Institute at the University of Oklahoma Health Science Center, Oklahoma City, OK. High frequency stimulation (HFS) of autonomic ganglia (AG) has been shown to facilitate initiation of atrial fibrillation (AF) in man and animal models. The purpose of this study was to identify the onset of AF with and without HFS of AG. Methods: In 11 dogs anesthetized with Na-pentobarbital, a right thoracotomy was performed. An octapolar electrode catheter was apposed to the right superior pulmonary vein (RSPV) close to a cluster of AG found within the fat pad (FP) on the adjacent atrium. A decapolar catheter was attached to the right atrium (RA), 2-3cm from the FP extending to the right atrial appendage (RAA). Induction of AF was tested by programmed stimulation with single extrastimuli (S1-S2) from each of the electrode pairs at the RSPV and RA sites. The window of vulnerability (WOV), i.e., the longest S1-S2 minus the shortest S1-S2 at which AF was induced, was
P4-8 SEGMENTAL VERSUS TRANSMURAL REMODELING AS ELECTROPHYSIOLOGICAL BASIS FOR T-WAVE MEMORY Darwin D. Jeyaraj, MD, Lance D. Wilson, MD, Steven Poelzing, PhD, Xiaoping Wan, PhD and David S. Rosenbaum, MD. Metrohealth Campus, Case Western Reserve University, Cleveland, OH. Although the T-wave is thought to arise from transmural repolarization gradients (TRG), the electrophysiological basis for T-wave memory remains unclear. Recently, we demonstrated that short term pacing diminished TRG, suggesting an alternate mechanism for T-wave memory. Therefore, we hypothesize that segmental repolarization gradients (SRG), i.e. action potential duration (APD) differences between different segments of left ventricle (LV), form the electrophysiological basis for T-wave memory. Dogs (n⫽5) were paced (VDD mode) from base of the anterior LV epicardium for 3-4 weeks. T-wave memory was characterized by change in T-wave vector angle (35 degrees), vector magnitude (0.9 mV), and QT prolongation (by 43⫾21 ms, p⬍0.001) compared to baseline. High-resolution optical mapping was used to record action potentials from transmural surface of arterially perfused canine wedge preparations harvested separately from LV segments proximal (anterior base) and distal (lateral and posterior base) to the site of pacing. TRG was defined as the APD difference between epicardial and M cells within a segment, whereas SRG was defined as difference in mean APD between any 2 segments. Identical measurements were made in 4 unpaced dogs (controls). In controls, TRG was significantly greater than SRG (38⫾10 ms vs. 10⫾2 ms, p⬍0.001), supporting the hypothesis that TRG underlies the normal T wave. Conversely, SRG were significantly larger than TRG in T-wave memory (73⫾21 ms vs. 38⫾10 ms, p⬍0.004). Memory-induced SRG was
S217 attributable to preferential prolongation of APD in posterior or late activated segments of myocardium (297⫾21 ms) compared to proximal segments (249⫾11 ms, p⬍0.03). Conclusions: These data provide direct experimental evidence that TRG and not SRG underlies the normal T wave. By contrast, the T wave changes (including QT prolongation) underlying T-wave memory were caused by SRG arising from preferential electrophysiological remodeling of LV segments distal to the site of pacing. These data suggest an intriguing hypothesis that altered mechanical stresses known to occur at distal sites are responsible for T-wave memory. P4-9 LOVASTATIN IS ANTIARRHYTHMIC IN ISCHEMIC HEART TISSUE BY BLOCKING TRIGGERED ACTIVITY Dezhi Xing, MD, Raymond Hohl, MD, PhD and James Martins, MD. University of Iowa College of Medicine and VAMC, Iowa City, IA. Background: AVID and MADIT-II trials showed that patients with coronary artery disease taking HMG-CoA reductase inhibitors have fewer ICD shocks suggesting anti-arrhythmic effects. Our aim was to analyze the effects of acutely administered lovastatin (LOVA): 1) on ICD defibrillation threshold (DFT) since anti-arrhythmics can worsen DFT, and 2) on triggered activity (TA) recorded from ischemic endocardium excised from the anterior wall of anesthetized dogs with anterior descending coronary occlusion. Methods and Results: Dogs received LOVA dissolved in DMSO. DMSO alone produced no effect on DFT. LOVA 10mg IV did not alter effective refractory period and tended to improve defibrillation success (from 20⫾3 to 17⫾3 J, P⫽0.09 vs control). Ischemic endocardium was studied using standard microelectrode techniques with normal Tyrode’s solution. There were no delayed afterdepolarizations (DADs) or TA at baseline in this depolarized tissue (-71⫾ 4mV). Isoproterenol (5x10-7M) superfusion, produced sustained TA with DAD amplitudes of 4.2 ⫾1 mV. LOVA (10-7 M) produced no change in resting membrane potential (-70⫾4 mV), action potential duration at 90% (from 231⫾11 to 229⫾8 msec, ns), action potential duration at 50% (168⫾7 to 166⫾6 msec, ns), action potential amplitude (89⫾5 to 85⫾5 mV, ns), overshoot (11⫾1 to 10⫾1 mV) or DAD amplitude (4.0⫾1 mV, ns). Interestingly, TA was attenuated to 1-5 complexes with LOVA (p⬍0.05) and not with DMSO alone. Reversibility to sustained TA also occurred after LOVA washout (20 min) as well as with co-superfusion with mevalonic acid (10-7 M, n⫽4), which also did not change action potentials or DADs. Conclusions: These results are consistent with the possibility that a prenylated protein down-stream from the mevalonic acid pathway may underlie this effect. Thus LOVA in concentrations achievable in human plasma may be anti-arrhythmic for TA without a deleterious effect on DFT. P4-10 HIGHER-ORDER POLYNOMIAL MODELING OF RABBIT EPICARDIAL VENTRICULAR ELECTROGRAMS Jeffrey L. Williams, MD, *Leonard I. Ganz, MD, Vladimir Shusterman, MD, PhD, Barry London, MD, PhD and Samir Saba, MD. University of Pittsburgh, Pittsburgh, PA. Objective: To construct a higher-order model of the rabbit’s epicardial ventricular electrogram (EGM). Background: A higher-order representation of the ventricular EGM allows the capture of depolarization and repolarization phenomena using a limited number of coefficients. Methods: New Zealand White rabbits underwent chronic implantation of pacemakers through a left thoracotomy approach. Unipolar ventricular EGMs sampled at a frequency of 1 kHz were stored digitally in one-minute segments before and after intravenous injection of isoproterenol or procainamide. Each cardiac cycle was divided into a QR and an RST segment which were modeled separately using a 6th-order polynomial equation. A
S218 6th-order equation was the smallest order model that faithfully captured the morphology of the raw EGM (norm of residuals ⬃ 1.0). Results: The 14 coefficients of each cardiac cycle were stable and reproducible throughout the baseline recordings (r ⬎⫽ 0.94, p ⬍ 0.002). Isoproterenol caused no changes in the coefficients of the QR-segment but significantly altered all but one of the 7 coefficients of the RST-segment (p6⫽0.0039, p5⫽0.017, p4⫽0.00007, p3⫽0.112, p2⫽0.00016, p1⫽0.0086, pa⫽0.00003). Procainamide caused statistically significant changes in both QR-segment (p6⫽0.018, p5⫽0.287, p4⫽0.019, p3⫽0.176, p2⫽0.016, p1⫽0.362, pa⫽0.000044) and RST-segment (p6⫽0.0028, p5⫽0.036, p4⫽0.002, p3⫽0.058, p2⫽0.022, p1⫽0.718, pa⫽0.0018) coefficients. Conclusion: Our data demonstrate the feasibility of a higher-order polynomial equation that models the rabbit ventricular EGM. This model is stable, reproducible, and can predict the changes expected with antiarrhythmic drug administration. If reproduced in humans, these findings can have wide applications in patients with implantable devices, ranging from morphologic discrimination of arrhythmias to early detection of metabolic derangements or drug effects. P4-11 ANGIOTENSIN II TYPE 1 RECEPTOR (AT1R) BLOCKADE HAS NO EFFECT ON AF SELF-PERPETUATION OVER 4 WEEKS IN BURST PACED GOATS Mark C. S. Hall, MRCP, Senthil Kirubakaran, MRCP, Clifford J. Garratt, MD and Nicholas S. Peters, MD, FRCP. Manchester Heart Centre, Manchester, United Kingdom and Imperial College, London, United Kingdom. Background: Previous studies suggest that AT1R blockade prevents shortterm electrical remodeling and longer-term (4 weeks) structural remodeling in a rapidly-paced dog model of AF. We examined the effects of AT1R blockade with candesartan on the self-perpetuation of AF by 4 weeks of atrial burst pacing in a goat model. Method: 8 goats underwent pacemaker implantation. Repeated bursts of atrial stimuli (2s, 50Hz, 4x diastolic threshold) were applied as in the original Allessie protocol. 4 goats received oral candesartan (0.5mg/kg/ day) for 1 week prior to and during 4 weeks of burst pacing, whilst 4 received placebo. AERP, AF duration, AF cycle length (AFCL), and sinus cycle length (SCL) were recorded where appropriate twice daily for 5 days, then daily. In 6 separate goats, biological activity of candesartan (0.5mg/ kg/day) was determined by measuring its effect on the pressor response to infused angiotensin II (up to 40ng/kg/min). Results: Candesartan compared to burst pacing alone had no effect on SCL (595⫾9ms vs 581⫾16ms, p⫽ns), the AERP at baseline (155⫾21.8ms vs 155⫾19.4ms, p⫽ns), the fall in AERP with atrial pacing (after 3 days 77.5⫾2.5ms vs 85⫾15.5ms, p⫽ns) or AFCL (103⫾2.9ms vs 103⫾1.4ms, p⫽ns). Candesartan had no demonstrable effect on the increasing stability of AF with time (figure). There was significant suppression of the pressor response to angiotensin II (40ng/kg/min angiotensin II produced a 8.1⫾3.2mmHg systolic BP rise in candesartan treated goats and 34⫾1.9mmHg in controls, p⬍0.002). Conclusions: Candesartan is absorbed orally in the goat and has activity at the AT1 receptor. Blockade of the AT1 receptor has no effect on atrial refractory period or induced AF duration during a 4-week protocol of atrial burst pacing in the conscious goat model.
Heart Rhythm, Vol 2, No 5, May Supplement 2005 P4-12 EXPERIMENTAL MODEL FOR APNEA INDUCED ATRIAL FIBRILLATION Muhammad Ghias, MD, *Benjamin J. Scherlag, PhD, William S. Yamanashi, PhD, Sunny S. Po, MD, PhD, *Hiroshi Nakagawa, MD, PhD, Ralph Lazzara, MD and *Warren M. Jackman, MD. Cardiac Arrhythmia Research Institute at the University of Oklahoma Health Science Center, Oklahoma City, OK and Cardiac Arrhythmia ResCardiac ArrhythCardiac Arrhythmia Research Institute at the University of Oklahoma Health Science Center, Oklahoma City, OK. Recent clinical reports have shown a close association between sleep apnea and atrial fibrillation (AF). Methods: In 10 Na-pentobarbital anesthetized dogs, under positive pressure ventilation, a right thoracotomy allowed placement of electrode catheters at the right superior pulmonary vein-atrial (RSPV-A) junction and right atrial appendage (RAA). Programmed electrical stimulation at a cycle length of 330ms (2X-20X threshold) was performed at both sites to determine atrial refractory periods (ARP) and AF inducibility. A insulated metal electrode except for the tip was inserted and stabilized in a cluster of autonomic ganglia adjacent to the RSPV to continuously record neural activity during control and apnea periods. Respiration was stopped during end expiration and intermittent PES was instituted at an intensity of stimulation at which no AF occurred during the control state and at 5-10ms shorter than the ARP at that level. Standard ECG leads, blood pressure (BP) and core temperature were continuously monitored. Results: During a 2 minute period of induced apnea, S1-S2 which initially failed to activate the atrium began to induce single then multiple responses leading to atrial tachycardia (AT) and AF. AF occurred at an average time of 85⫾38 sec after the onset of apnea. The entire sequence was associated with a significant increase in systolic BP compared to control (control, 149⫾26mmHg; apnea/AF, 193⫾38mmHg, p⫽0.05) but not diastolic BP (control, 108⫾14mmHg; apnea/AF, 122⫾20mmHg, p⫽0.08). Neural potentials recorded from AGs increased in frequency with “bursting” which preceded the onset of any electrophysiologic effect. Autonomic blockade with intravenous atropine, 2mg, and esmolol or propranolol, 1mg/kg, prevented apnea induced AF and was associated with a marked fall in BP. Conclusions: Induced apnea in the anesthetized dog consistently resulted in AT/AF associated with both increased systolic BP and neural firing in local autonomic ganglia. The response to autonomic blockade which suppressed the arrhythmias suggests a strong autonomic basis for apnea induced AF. P4-13 MAPPING THE ONSET OF ATRIAL FIBRILLATION ARISING FROM PULMONARY VEIN AND NON-PULMONARY VEIN SITES IN THE DOG HEART *Benjamin J. Scherlag, PhD, William S. Yamanashi, PhD, *Jeff Edwards, RN, Neng Geng, MD, Sunny S. Po, MD, PhD, Ralph Lazzara, MD and *Warren M. Jackman, MD. Cardiac Arrhythmia Research Institute at the University of Oklahoma Health Science Center, Oklahoma City, OK. High frequency stimulation (HFS) of autonomic ganglia (AG) has been shown to facilitate initiation of atrial fibrillation (AF) in man and animal models. The purpose of this study was to identify the onset of AF with and without HFS of AG. Methods: In 11 dogs anesthetized with Na-pentobarbital, a right thoracotomy was performed. An octapolar electrode catheter was apposed to the right superior pulmonary vein (RSPV) close to a cluster of AG found within the fat pad (FP) on the adjacent atrium. A decapolar catheter was attached to the right atrium (RA), 2-3cm from the FP extending to the right atrial appendage (RAA). Induction of AF was tested by programmed stimulation with single extrastimuli (S1-S2) from each of the electrode pairs at the RSPV and RA sites. The window of vulnerability (WOV), i.e., the longest S1-S2 minus the shortest S1-S2 at which AF was induced, was