The purposes of surgical treatment for atrial fibrillation

Transcription

1 Radial Approach: A New Concept in Surgical Treatment for Atrial Fibrillation. II. Electrophysiologic Effects and Atrial Contribution to Ventricular Filling Takashi Nitta, MD, Richard Lee, MD, Hiroshi Watanabe, MD, Kevin M. Harris, MD, John M. Erikson, MD, Richard B. Schuessler, PhD, John P. Boineau, MD, and James L. Cox, MD Divisions of Cardiothoracic Surgery and Cardiology, Washington University School of Medicine, St. Louis, Missouri Background. In a previous study the atrial incisions that follow the concept of the radial approach were designed according to the activation sequence during sinus rhythm and the atrial coronary artery anatomy in normal dogs. The purpose of the present study was to determine whether the radial approach provides a more physiologic activation sequence and atrial transport function than the maze procedure. Methods. Ten dogs that had undergone the radial approach (n 5) or the maze procedure (n 5) were studied 6 weeks postoperatively. Sinus node function and inducibility of atrial fibrillation were examined before and after operation. The atria were mapped endocardially with 212 electrodes, and atrial activation sequences during sinus rhythm and right atrial pacing were examined. Atrial transport function was assessed by transepicardial Doppler echocardiography. Results. No dogs developed sinus node dysfunction postoperatively. Both the radial approach and the maze procedure equally prevented sustained atrial fibrillation. The atrial activation sequence was more synchronous after the radial approach than after the maze procedure. There was no electrically isolated region after the radial approach. The total activation time of the left atrium was significantly shorter after the radial approach than after the maze procedure ( versus ms, p < 0.05). The ratio of peak flow velocity of the E wave to the A wave (peak E/A) of the transmitral Doppler flow was significantly smaller after the radial approach than after the maze procedure ( versus , p < 0.05). The atrial filling fraction of the transmitral Doppler flow was significantly larger after the radial approach than after the maze procedure (29.9% 7.3% versus 14.8% 5.0%, p < 0.01). There was no significant difference in peak E/A and atrial filling fraction of the transtricuspid Doppler flow between the two procedures. Conclusions. The radial approach provides a more synchronous activation sequence and atrial transport function, and thus may represent a more physiologic alternative to the maze procedure as a surgical treatment for atrial fibrillation. (Ann Thorac Surg 1999;67:36 50) 1999 by The Society of Thoracic Surgeons The purposes of surgical treatment for atrial fibrillation (AF) include restoration of sinus rhythm and atrial transport function and prevention of thromboembolism [1, 2]. However, previous studies have revealed that left atrial transport function is smaller in patients after the maze procedure than in normal control subjects, whereas right atrial transport function is comparable [3 6]. We hypothesized that the isolation of the posterior left atrium, discordant activation in certain adjacent left atrial segments, nonphysiologic atrioventricular contraction coupling caused by delayed atrial activation, and the potential atrial ischemia beyond the incision are the mechanisms underlying insufficient left atrial transport Accepted for publication Oct 12, Address reprint requests to Dr Boineau, Division of Cardiothoracic Surgery, Washington University School of Medicine, 660 S Euclid Ave, Box CSRB, St. Louis, MO function after the maze procedure [7]. We therefore developed a new concept of surgical treatment for AF the radial approach in which atrial incisions radiate from the sinus node toward the atrioventricular annular margins to allow a more physiologic atrial activation sequence and to parallel the atrial branches of coronary arteries to preserve the blood supply to most atrial segments. On the basis of the activation sequence during sinus rhythm and the atrial coronary artery anatomy in normal dogs, the atrial incisions that follow the concept of the radial approach were designed. Animal studies were performed to determine whether the radial approach represents a more physiologic alternative to the maze procedure as a surgical procedure for AF. Specifically, the goals of the present study were to examine (1) inducibility of AF, (2) sinus node function, (3) atrial activation sequence, and (4) atrial transport function after the two procedures by The Society of Thoracic Surgeons /99/$20.00 Published by Elsevier Science Inc PII S (98)

2 Ann Thorac Surg NITTA ET AL 1999;67:36 50 RADIAL APPROACH FOR AF: PART 2 37 Material and Methods Surgical Technique In 12 adult mongrel dogs of either sex (age, 6 to 13 months; weight, 20 to 26 kg), anesthesia was induced with intravenous sodium thiopental (20 mg/kg). The animals were intubated and ventilated using a volume-limited ventilator. The animals were maintained with inhaled isoflurane (1% to 3%) at an appropriate depth of anesthesia for the duration of the procedure. A femoral artery catheter was inserted to monitor systemic arterial pressure continuously. Arterial blood samples were drawn every 30 minutes to determine arterial oxygen tension, acid base balance, and electrolyte levels. Ringer s lactate solution was continuously infused, and sodium bicarbonate, potassium chloride, and calcium chloride were supplemented as indicated to maintain ph and electrolytes within normal values. The chest was exposed through a right thoracotomy at the fourth intercostal space with sterile surgical technique. The pericardium was opened, and the heart was suspended in a pericardial cradle. The azygos vein was divided. After systemic heparinization (3 mg/kg), a 14F arterial cannula was inserted into the right femoral artery, and 24F and 28F right-angled venous cannulas were inserted into the superior and inferior vena cavas. Total cardiopulmonary bypass was established, and the animals were cooled to an esophageal temperature of 30 to 32 C. The dogs were randomly divided into two groups and underwent either the radial approach (n 5) or the maze procedure (n 5). Two animals underwent hypothermic cardiopulmonary bypass for 120 minutes with cardioplegic arrest for 60 minutes without atrial incisions (sham). The surgical technique for the maze procedure was exactly the same as described previously [8]. The atrial incisions of the radial approach have been illustrated in a previous report [7]. The surgical technique is as described herein. Before cannulation, the extracardiac tissue around the superior vena cava and inferior vena cava was dissected. Then, the interatrial groove was dissected. Because there was no incision around the right upper pulmonary vein, the extent of dissection was minimized at this region. Because exposure of the left atrium between the right lower pulmonary vein and inferior vena cava is extremely important in the radial approach, the extracardiac tissue was extensively dissected in this region. The procedures for the right atrium in the radial approach were the same as those for the maze procedure, except for the right atrial appendage incision. In the radial approach, the right atrial appendage was not excised but was incised with at least 2 cm of atrial tissue left between the incision and the sinus node region. The incision was extended anteromedially down to the tricuspid valve annulus 1 cm above the anteroseptal commissure of the tricuspid valve, and in the opposite direction, extended inferiorly toward the lower right atrium with 1 cm of atrial tissue between the intermediate right atrial artery and the incision. Trabeculae in the right atrial appendage were incised and divided to eliminate reentry using the bridging muscles. There were two incisions on the left atrium and interatrial septum in the radial approach. One incision started at the anterior limbus of the fossa ovalis, extended inferoposteriorly, and merged with a lower right atrial incision at the lower posterior septum beneath the right lower pulmonary vein orifice. This incision was further extended toward the inferior left atrium, passing near the right and left lower pulmonary vein orifices, and down to the mitral valve annulus at the commissure between the middle and posteromedial scallops. The coronary sinus was dissected, skeletonized, and cryoablated using a 3-mm cryoprobe (Frigitronics, Inc, Coopersurgical, Shelton, CT) for 2 minutes at a temperature of 60 C. The left atrial appendage was resected. The other incision started at the superior left atrium between the right and left upper pulmonary veins, with 2 cm of atrial tissue left between the right upper pulmonary vein orifice and the incision. This incision was connected to the left atrial appendage excision and extended anteromedially further down to the mitral valve annulus at the anterolateral commissure. Great care was taken to avoid injury to the left circumflex coronary artery during the dissection of the atrioventricular fat pad at this region. To prevent reentry around the right upper pulmonary vein, the atrial tissue between the upper and lower right pulmonary veins was cryoablated. In addition, the atrial tissue between the left atrial incision and each pulmonary vein orifice was also cryoablated to eliminate potential reentry around the pulmonary vein orifice, and conduction block over the incision was ensured by cryoablation at each atrioventricular annulus. The right atrial incisions were performed while the heart was beating. The aorta was then cross-clamped, and 500 ml of hyperpotassium (26 meq/l) crystalloid cardioplegic solution (Plegisol; Abbott Laboratories, North Chicago, IL) was infused antegradely through an aortic root cannula. Additional cardioplegic solution (300 ml) was infused 30 minutes later. The left atrial and septal incisions were performed under cardioplegic arrest. A 4-0 polypropylene suture was used to close all atriotomies. After closure of all the left-side incisions, air was removed from the heart, and the heart was reperfused. Ventricular fibrillation was defibrillated by epicardial shocks, if necessary. Dogs were rewarmed and weaned from cardiopulmonary bypass. Protamine sulfate was administered to neutralize heparinization. The chest was closed in layers over a chest tube. The animal was allowed to recover. After no evidence of air leak was confirmed, the chest tube was removed. Postoperatively, supplemental oxygen was given for 6 hours, and buprenorphine (0.01 mg/kg) was also given for pain relief. The dogs received ceftiofur sodium (2 mg/kg) subcutaneously once a day for 3 days postoperatively. Six weeks after the initial surgical procedure, the dog was transferred from the farm for assessment. The animal was anesthetized, intubated, and ventilated as described earlier. A femoral arterial catheter was inserted to monitor systemic arterial pressure continuously. Arterial blood samples were drawn every 30 minutes to deter-

3 38 NITTA ET AL Ann Thorac Surg RADIAL APPROACH FOR AF: PART ;67:36 50 Fig 1. Electrode molds used in the present study. The atria were mapped endocardially with 212 unipolar electrodes mounted on a molded sponge designed to fit the postoperative atria. One hundred four electrodes were mounted on the right atrial mold and 108 electrodes on the left atrial mold. The electrode molds were inserted into each atrium across the atrioventricular valve retrogradely through ventriculotomy procedures during cardiopulmonary bypass in study animals. The electrode molds are shown as if the atrial endocardial surfaces are observed from the back. (IVC inferior vena cava; LA left atrium; LLPV left lower pulmonary vein; LUPV left upper pulmonary vein; MV mitral valve; RA right atrium; RAA right atrial appendage; SVC superior vena cava; TV tricuspid valve.) mine arterial oxygen tension, acid base balance, and electrolyte levels. Ringer s lactate solution was continuously infused, and sodium bicarbonate, potassium chloride, and calcium chloride were supplemented as indicated to maintain ph and electrolyte levels within normal values. The heart was exposed through a midline thoracotomy and was suspended in a pericardial cradle. After systemic heparinization (3 mg/kg), the inferior vena cava, superior vena cava, and left femoral artery were cannulated, and normothermic cardiopulmonary bypass was instituted. Bilateral ventriculotomies were performed, and the mitral and tricuspid valve leaflets were excised. Two electrode forms carrying 212 unipolar electrodes were inserted into both atria through ventriculotomies across the atrioventricular annuli retrogradely. The electrode forms were made of urethane sponge and were designed to fit the postoperative shape of the canine atrial cavities (Fig 1). The electrodes were custommanufactured unipolar electrodes, as described previously [9]. The electrode was silver plated and had a diameter of 2 mm. A total of 104 and 108 electrodes were distributed on the right and left atrial electrode forms, respectively. The interelectrode distance ranged from 2 to 5 mm. An indifferent electrode was placed on the chest wall for unipolar reference. After the study, the animal was killed, and the heart was excised with the electrodes in place. The anatomic correlation between each atrial incision and the electrodes was verified in each dog. Electrophysiologic Study Sinus node function and the duration of induced AF were examined before and 6 weeks after the operation. These measurements were performed before initiation of cardiopulmonary bypass. Sinus node function was assessed by the sinus cycle length and the maximal sinus node recovery time (SNRT). Right atrial pacing was performed at cycle lengths of 500, 400, 300, and 200 ms for 30 seconds, and SNRT was measured for each pacing cycle length. To induce AF, burst pacing was performed at the right atrial appendage at a cycle length of 50 and 100 ms. Burst pacing was performed three times for each pacing cycle length. The maximal duration of induced AF was recorded. Sustained AF was defined in the present study as an irregular atrial tachyarrhythmia with an average atrial cycle length of less than 160 ms perpetuated for 10 seconds or more. In dogs in which no atrial repetitive response was induced with burst pacing postoperatively, acetylcholine (200 g/kg) was administered intravenously, and the atrial burst pacing was repeated while the animals were on cardiopulmonary bypass. All stimulation was performed using a programmable electrical stimulator (DTU-101; Bloom Associates Ltd, Reading, PA) through the bipolar pacing electrodes placed on the right atrial appendage preoperatively and on the top of the anterior right atrium adjacent to the atrial appendage incision postoperatively. The pacing threshold was determined, and all stimulations were performed at a pulse width of 2 ms and at twice the diastolic threshold. The electrocardiogram, pacing artifacts, and reference atrial electrograms were recorded on an ink jet chart recorder (Recorder 2800S; Gould Inc, Cleveland, OH). The SNRT was measured at a paper speed of 100 mm/s, and the duration of AF was measured at a paper speed of 25 mm/s. Mapping Data Acquisition and Analysis A 256-channel mapping system that allowed simultaneous recording of up to 256 signals was used for data acquisition and analysis. The system is based on a Vax/Station II/GPX graphics workstation (Digital Equipment Corp, Maynard, MA) connected to two 128-channel PDP 11/23 Plus-based data acquisition subsystems. The

4 Ann Thorac Surg NITTA ET AL 1999;67:36 50 RADIAL APPROACH FOR AF: PART 2 39 system is run with in-house developed software (GLAS). Unipolar electrograms were directed to the differential amplifiers with an input impedance of ohms, and the frequency response was set at 50 to 500 Hz. The input range of the analog-to-digital converter was 10 V, and an amplifier gain of 250 or 500 was used to generate an input signal range of 20 mv. Unipolar electrograms as well as the electrocardiogram, pacing artifacts, and reference bipolar right atrial electrograms were digitized at 1,000 Hz with a 12-bit resolution. Data processing was performed at a graphic workstation (IRIS Crimson; Silicon Graphics Inc, Mountain View, CA). Local activation times were determined as the maximal negative derivative of the unipolar electrogram. Activation maps were displayed as dynamic images on a three-dimensional surface model of the canine atria. Location of the atriotomies on the maps was determined by the anatomic correlation between the atriotomies and the electrodes, which was verified in each animal after the experiment. The location was confirmed by the configuration of the electrograms; the electrograms from the electrodes adjacent to the atriotomies frequently showed double or polyphasic potentials. The location and extent of isolation of the posterior left atrium in the maze procedure were also verified by the anatomic correlation between the region and the electrodes. The electrograms recorded at the isolated atrium were usually of low voltage, and the activation times were frequently discordant with the activation times of the surrounding electrodes, suggesting remote activation [10]. To characterize the temporal distribution of atrial activation, the number of electrodes activated during each 10-ms period was represented as a histogram in all dogs. Furthermore, to quantify the atrial activation, total activation time was determined in each atrium from the activation maps during pacing at the right atrium at a paced cycle length of 400 ms. To eliminate the nonspecific delayed activation, the time between the earliest activation and the 95th percentile activation was defined as the total activation time. Intracardiac Pressures and Cardiac Output Intracardiac pressures and cardiac output were determined before the initiation of cardiopulmonary bypass. The intracardiac pressure was measured using a fluidfilled catheter (Edwards Swan-Ganz catheter 93A-931H- 7.5F; Baxter, Irvine, CA) inserted through the superior vena cava. A 4-0 polypropylene pursestring suture was placed on the superior vena cava 1 cm caudal to the bifurcation of the innominate vein for insertion of the catheter. Systemic blood pressure was obtained from a 16-gauge plastic catheter inserted into the femoral artery. All pressure waves were digitized, recorded, and processed using the previously described system. Systolic, diastolic, end-diastolic, and mean pressures were calculated from the average of five consecutive cardiac cycles. Cardiac output was measured by the thermodilution method using the same catheter described earlier and a cardiac output computer (COM-1; American Edwards Laboratories, Irvine, CA). A 5-mL bolus of cold (0 to 5 C) saline was injected. The average of three measurements was recorded. Cardiac output was measured during sinus rhythm and during pacing in the right atrium or the right ventricle at the same pacing cycle length (500 ms). All pacing was performed at twice the diastolic pacing threshold. There was a 5-minute interval between the measurements during pacing at different sites. The atrial contribution to cardiac output was determined by the difference between the cardiac output during atrial and ventricular pacing divided by the cardiac output during atrial pacing. Doppler Echocardiography Transepicardial pulsed Doppler echocardiography (Sonos 1500; Hewlett-Packard, Andover, MA) was performed before the initiation of cardiopulmonary bypass using a 2.5- or 5-MHz imaging transducer. The study was carried out from the apical four-chamber view with the sample volume positioned at the level of the tip of the mitral or the tricuspid valvular leaflets. Subsequently, the sample volume was placed at the entrance of the right superior pulmonary vein to the left atrium. Flow velocity spectra were recorded on videotape at a display speed of 100 mm/s. To characterize the transmitral and transtricuspid flow, the peak velocities of the early filling (E) and atrial filling (A) waves and their ratio (peak E/A) were determined. The total diastolic time velocity integral (TDi) and the time velocity integral of early ventricular filling (Ei) were measured directly by planimetry of the diastolic mitral and tricuspid flow velocity spectra. The time velocity integral of atrial filling (Ai) was calculated as the difference between the integrals (Ai TDi Ei). Each measurement was obtained as an average of three to five cardiac cycles. The atrial filling fraction (AFF), as an expression of atrial transport flow, was then derived as the percentage of the time velocity integral of atrial filling in relation to the total diastolic time velocity integral (AFF 100 Ai/TDi) [12]. The mitral deceleration time of the E wave was also measured from the transmitral Doppler flow spectra. From the pulmonary venous flow velocity spectra, the peak velocities of the systolic (S) and diastolic (D) flow components were obtained. The time velocity integrals of the systolic (Si) and the diastolic (Di) flow components were also measured by planimetry of the flow velocity spectra waves. The ratio of the systolic to diastolic peak velocity (S/D) as well as the ratio of the systolic to diastolic time velocity integral (Si/Di) were determined. To obtain the average values for these variables in normal dogs, transepicardial Doppler echocardiography was performed in 5 animals through a right thoracotomy preoperatively. Because intraoperative echocardiography during the initial operation significantly increased the incidence of infection and mortality, we did not take the baseline echocardiographic data from all the animals preoperatively.

5 40 NITTA ET AL Ann Thorac Surg RADIAL APPROACH FOR AF: PART ;67:36 50 Weighing of Isolated Posterior Left Atrium After the study, the animal was killed, and the heart was excised. The ventricular myocardium, the great arteries, and the extracardiac tissues were removed from the atria. In the post maze procedure heart, the electrically isolated part of the posterior left atrium was separated from the remainder of the atria. Both parts of the atria were weighed (Sartorius; Brinkmann Instruments, Inc, Westbury, NY). The percentage of the isolated portion in relation to the entire atria was calculated. Statistical Analysis Continuous variables are expressed as mean standard deviation. Statistical significance of data between the two groups was determined by the Student s unpaired t test. Comparison of the paired data from the same animal was analyzed by the Student s paired t test. A value of p 0.05 was considered statistically significant. Multiple analysis of variance was used to determine the effects of the type of operation and the pacing site on cardiac output. All animals received humane care in compliance with the Principles of Laboratory Animal Care formulated by the National Society of Medical Research and the Guide for the Care and Use of Laboratory Animals prepared by the National Academy of Science and published by the National Institutes of Health (NIH publication No , revised 1985). In addition, the study protocol was approved by the Washington University Animal Studies Committee. Results Cardiopulmonary Bypass and Aortic Cross-Clamp Time Cardiopulmonary bypass time was minutes for the radial approach and minutes for the maze procedure. Aortic cross-clamp time was 51 7 minutes for the radial approach and 57 4 minutes for the maze procedures. There were no significant differences between the two procedures in cardiopulmonary bypass time and aortic cross-clamp time. Cryothermia was used in five animals in the maze procedure and in nine in the radial approach. During aortic cross-clamp time, cryothermia was used in three animals in the maze procedure and in seven in the radial approach. Sinus node function Preoperative sinus cycle length was ms after the radial approach and ms after the maze procedure. Postoperative sinus cycle length was ms after the radial approach and ms after the maze procedure. There were no significant differences between the preoperative and postoperative sinus cycle lengths in the radial approach or the maze procedure. No dogs exhibited sinus bradycardia or spontaneous atrial arrhythmias postoperatively. Preoperative maximal SNRT was ms in the radial approach and ms in the maze procedure. Postoperative maximal SNRT was Fig 2. Duration of AF induced before (Pre Op) and 6 weeks after (Post Op) the radial approach (solid circles) and the maze procedure (open circles) ms after the radial approach and ms after the maze procedure. No animals showed prolongation of the maximal SNRT preoperatively or postoperatively. Inducibility of Atrial Fibrillation Preoperatively, sustained AF of at least 10 seconds in duration was inducible in all dogs. The average duration of AF was and seconds in animals undergoing the radial approach and the maze procedure, respectively. There was no statistical difference in the preoperative duration of AF between the two procedures. Both procedures markedly suppressed the duration of AF (Fig 2). Postoperatively, sustained AF was not inducible in any animal. Two animals after the radial approach and 4 after the maze procedure showed brief periods of repetitive atrial responses (up to 10 cycles). The average duration of AF (repetitive atrial responses) was seconds after the radial approach and seconds after the maze procedure. Examples of induction of AF by atrial burst pacing preoperatively and postoperatively are shown in Figure 3. The duration of AF was significantly shorter after than before operation, and there was no statistical difference between the two procedures. Acetylcholine was administered in 3 dogs after the radial approach and in 1 after the maze procedure when no repetitive atrial responses were induced by burst pacing. Acetylcholine provoked the repetitive atrial responses. The induced repetitive responses lasted for 0.2 to 4.8 seconds. Sustained AF was never inducible in any animal. Atrial Activation Sequence Endocardial activation maps of the atria during sinus rhythm after the radial approach are shown in Figure 4, and those for the maze procedure are shown in Figure 5. The differences in activation sequence were determined by the atrial incisions of each procedure. The differences between the procedures were remarkable at (1) the right atrial appendage, (2) the interatrial septum, (3) the posterior left atrium, and (4) the lateral left atrium. The atrial incisions in the lateral right atrium were similar between the two procedures except for the right atrial appendage excision. Although the appendage was not excised and

6 Ann Thorac Surg NITTA ET AL 1999;67:36 50 RADIAL APPROACH FOR AF: PART 2 41 Fig 3. Atrial burst pacing to induce AF. Preoperative (PRE-OP) burst pacing at a cycle length of 100 ms induced AF, which was sustained for 18 seconds. Six weeks after the radial approach, burst pacing at the same cycle length did not induce sustained AF and resulted in only a brief period ( 1 second) of atrial repetitive responses. (ECG electrocardiogram; POST-OP after surgical treatment; RA-EGM right atrial electrogram; RR repetitive responses; SR sinus rhythm.) was preserved in the radial approach, the activation sequence around the right atrial appendage did not differ between the procedures. The activation time at the anterosuperior right atrium between the atrial appendage incision and the right ventricle was 50 to 60 ms after sinus node activation in dogs after both the radial approach and the maze procedure. One component of the activation wavefront from the sinus node propagated inferiorly along the crista terminalis and spread over the lateral right atrium. This wavefront was blocked by the intercaval longitudinal incision at the posterior right atrium and by the transverse incision at the lower right atrium. The other component of this wavefront propagated superiorly then medially toward the upper atrial septum through atrial tissue between the superior vena cava and the right atrial appendage incision. The interatrial septum was activated from the top of the posterior septum at the junction with the superior vena cava. The activation sequence at the septum was different between the two procedures. In the radial approach, the pattern of activation was the same as that in normal atria, propagating from the top of the posterior septum toward the inferior left atrium. In the maze procedure, the wavefront detoured around the septal incision to activate the posterior septum, which was encompassed by the septal incision and the intercaval longitudinal incision. In addition, the superior part of the posterior septum was scarred. The scarring of the posterior septum was present in all animals after the maze procedure, whereas no animals showed septal scarring after the radial approach (Fig 6). The activation sequence was similar between the procedures in the remainder of the right atrium. The wavefront at the septum propagated toward the lower right atrium, passing through the area between the inferior vena cava and the tricuspid valve annulus. This wavefront was finally blocked by the intercaval longitudinal incision and the transverse incision at the lower right atrium. The left atrial activation pattern was considerably different between the two procedures. The earliest activation of the left atrial endocardium occurred at the posterior septum ms after the sinus node activation in the radial approach. After the maze procedure, the earliest activation occurred ms after the sinus node activation, and the earliest activation site was shifted anteriorly because the septal incision was located at the posterior septum, where the earliest activation occurred. From the earliest activation site of the left atrial endocardium, the activation wavefront spread out in three directions in the radial approach: inferiorly at the septum, posteriorly between the pulmonary veins, and laterally at the superior left atrium. The first component of the wavefront at the septum was further split by the septal incision. One wavefront propagated toward the posterior septum between the right pulmonary veins and the septal incision. This pathway was a dead end, and the wavefront was blocked by the septal incision, cryolesions, and right pulmonary vein orifices. The other wavefront propagated inferiorly in the anterior septum, then laterally in the inferior left atrium, beneath the left lower pulmonary veins. The second component of the wavefront propagated between the right and left upper pulmonary veins, then inferiorly, leftward between the left upper and lower pulmonary veins into the lateral left atrium toward the mitral valve annulus. The third component of the wavefront propagated leftward in the superior left atrium between the upper left atrial incision and mitral valve annulus through Bachmann s bundle and was blocked by the upper left atrial incision that extended down to the mitral valve annulus. The latest atrial activation occurred in the lateral left atrium approximately 80 to 90 ms after the sinus node activation in the radial approach. In the maze procedure, the activation wavefront spread out in two directions from the earliest activation site of the left atrial endocardium. One component of the wavefront propagated the anterior septum inferiorly then laterally at the inferior left atrium, blocking at the mitral annular incision. The activation sequence was detoured at the posterior septum between the septal incision and the pulmonary isolation incision. The wavefront from the anterior septum turned around

7 42 NITTA ET AL Ann Thorac Surg RADIAL APPROACH FOR AF: PART ;67:36 50 Fig 4. Atrial endocardial maps during sinus rhythm 6 weeks after the radial approach. The boxed area on the electrocardiogram (ECG) is the data window analyzed to construct the activation map. The wide QRS configuration in the electrocardiogram was the consequence of the ventriculotomy procedures. The two middle maps represent the lateral (LAT) and septal (SEPT) surfaces of the right atrial (RA) endocardium. The three lower maps represent the left lateral, inferior (INF), and septal aspects of the left atrium (LA). The sinus node is indicated as an oval on the right atrium at the superior vena caval (SVC) junction. The border of the interatrial septum is denoted as dashed lines. The activation sequence is indicated by arrows. The asterisk in the left atrial septum indicates the earliest activation site of the left atrial endocardium. Atrial incisions are shown, and cryolesions are denoted as small dark circles. Numbers represent the activation times associated with each wavefront (wavy lines) and together depict the activation sequence. (CS, coronary sinus; FO, fossa ovalis; LAA left atrial appendage; L.LAT left lateral; RPVs, right pulmonary veins; other abbreviations are as in Fig 1.)

8 Ann Thorac Surg NITTA ET AL 1999;67:36 50 RADIAL APPROACH FOR AF: PART 2 43 Fig 5. Atrial endocardial maps during sinus rhythm 6 weeks after the maze procedure. The shaded area denotes the electrically isolated region. (Symbols and abbreviations are as in Fig 4.) the edge of the septal incision to activate the posterior septum. As in the right endocardium of the right septum, the superior part of this region was also determined to be scarred electrophysiologically and pathologically (Fig 6). The other component of the wavefront traversed laterally at the superior left atrium, turned around the left atrial appendage excision line, then propagated inferiorly in the lateral left atrium, blocking at the mitral annular

9 44 NITTA ET AL Ann Thorac Surg RADIAL APPROACH FOR AF: PART ;67:36 50 Fig 6. Microscopic photographs of a cross section of the posterior atrial septum after the maze procedure (left panel) and the radial approach (right panel). Note that a large part of the atrial septum was scarred after the maze procedure, whereas most of the septum was preserved after the radial approach. (LA left atrium; RA right atrium.) incision. The latest atrial activation occurred at the lateral left atrium more than 90 ms after the sinus node activation in the maze procedure. Because the procedure isolated the posterior left atrium, there was no activation in this region. Total Activation Time Examples of temporal distribution of atrial activation during sinus rhythm in a sham animal and in animals after the procedures are shown in Figure 7 as histograms of the number of electrodes activated during each 10-ms period. In the sham animal, both the right and left atrial activation were distributed over a narrow time range. The right and left atrial-activation completed within 50 and 70 ms after the sinus node activation, respectively. In animals after the radial approach or the maze procedure, the distribution of the right atrial activation was slightly widened and shifted rightward, and the total activation time was prolonged. As confirmed in the activation maps (Figs 4, 5), the left atrial activation started later in the maze procedure than in the radial approach. The distribution of the left atrial activation was widened and the total activation time was slightly prolonged after the radial approach. However, activation at more than 95% of the electrodes in the left atrium was completed within 70 ms (median, 60 ms) after sinus node activation. After the maze procedure, these changes were significantly different. The distribution of the left atrial activation shifted rightward (median, 97 ms), and as a result, the total activation time was prolonged. These patterns of atrial activation distribution were specific to each procedure. The total activation time during right atrial pacing for the two procedures is shown in Figure 8. There was no significant difference between the procedures in total activation time of the right atrium ( ms for the radial approach versus ms for the maze procedure). The total activation time of the left atrium in the maze procedure was significantly longer than that in the radial approach ( versus ms, p 0.05). Cardiac Hemodynamic Variables Hemodynamic variables during sinus rhythm after both procedures are shown in Table 1. There were no significant differences in cardiac output and intracardiac pressures during sinus rhythm between the two procedures. No animals showed abnormal hemodynamic variables, such as elevated right atrial or pulmonary capillary wedge pressure or low cardiac output. Cardiac output during pacing at a cycle length of 500 ms was compared between the two procedures and between the pacing sites, as shown in Table 2. Cardiac output during right atrial pacing was significantly greater ( p 0.05) than during right ventricular pacing both after the radial approach and the maze procedure. There were no significant differences between the two procedures in cardiac output during pacing at the right atrium ( p 0.11) or the right ventricle ( p 0.27). The atrial contribution to cardiac output was 14.0% 7.9% after the radial approach and 6.8% 2.5% after the maze procedure; nevertheless, the difference between the procedures was not significant ( p 0.09). Transmitral Flow Velocity The A wave of the transmitral flow was detected in all 5 dogs after the radial approach and in 4 of 5 after the maze procedure. Examples of flow velocity spectra across the mitral valve in the animals after both procedures are shown in Figure 9. The peak velocities of the E and A waves across the mitral valve and peak E/A are shown in Table 3 and Figure 10. There were no significant differences between the procedures in the peak velocities of the E or A wave. Peak E/A after the maze procedure was

10 Ann Thorac Surg NITTA ET AL 1999;67:36 50 RADIAL APPROACH FOR AF: PART 2 45 Table 1. Hemodynamic Variables After the Maze Procedure and the Radial Approach a Variable Radial Approach Maze Procedure HR (beats/mm) CO (L/min) SV (ml beat 1 mm 1 ) AP (mm Hg) Systolic Diastolic Mean PCWP (mm Hg) PAP (mm Hg) Systolic Diastolic Mean RVP (mm Hg) Systolic End-diastolic RAP (mm Hg) a Data presented are mean standard deviation; no significant differences for all comparisons. AP arterial pressure; CO cardiac output; HR heart rate; PAP pulmonary artery pressure; PCWP pulmonary capillary wedge pressure; RAP right atrial pressure; RVP right ventricular pressure; SV stroke volume. wave. The AFF after the maze procedure was significantly smaller than the AFF preoperatively and after the radial approach. The mitral deceleration time was ms preoperatively, ms after the radial approach, and ms after the maze procedure. The difference was not significant between the procedures. Fig 7. Temporal distribution of atrial activation during sinus rhythm in a sham dog and in dogs after the radial approach and the maze procedure. Histograms represent the number of electrodes activated during each 10-ms period. Open bars indicate right atrial activation; solid bars indicate left atrial activation. See text for explanation. significantly larger than the peak E/A preoperatively and after the radial approach. The time velocity integrals of the E and A waves across the mitral valve and the AFF are shown in Table 3 and Figure 11. There were no significant differences between the procedures in time velocity integrals of the E or A Transtricuspid Flow Velocity The A wave of the transtricuspid flow was detected in all dogs after both the radial approach and the maze procedure. The peak velocity and the time velocity integral of the E and A waves across the tricuspid valve and the peak E/A and AFF are shown in Table 4. There was no significant difference between the procedures in any of the variables compared. Table 2. Cardiac Output During Right Atrial and Right Ventricular Pacing Pacing Site Radial Approach (mean SD) Maze Procedure (mean SD) RV a RA b a,b a Not significant, radial approach versus maze procedure. b p 0.05, (RV) right ventricle versus right atrium (RA). SD standard deviation. Fig 8. Total activation times of the right and left atria during right atrial pacing after the radial approach and the maze procedure. (NS not significant.)

11 46 NITTA ET AL Ann Thorac Surg RADIAL APPROACH FOR AF: PART ;67:36 50 Fig 9. Doppler flow tracing across the mitral valve after the radial approach (upper panel) and the maze procedure (lower panel). Note that the peak velocity and the area under the curve of the A wave are larger in the radial approach than in the maze procedure. (E and A Doppler flow during early and atrial filling of the left ventricle, respectively.) Pulmonary Venous Flow The peak velocities and the time velocity integrals of the S and D waves, as well as their ratios (S/D and Si/Di), are shown in Table 5. There were no significant differences between the procedures in peak S/D and in Si/Di. However, S/D after the maze procedure and Si/Di after both the radial approach and the maze procedure were significantly smaller than preoperative values. Weight of Isolated Posterior Left Atrium The weight of the isolated posterior left atrium in dogs after the maze procedure was g, which represents 18.3% 1.2% of the whole atria. There was no isolated atrial region in animals after the radial approach. Comment Atrial Activation Sequences An important finding in the present study was that the atrial activation sequence was more synchronized after the radial approach than after the maze procedure and prevented AF and preserved sinus node function equally as well as the maze procedure. In the radial approach, the posterior left atrium between the pulmonary veins was used as a contractile atrial component, whereas the maze procedure isolated the region. This technique also allowed the lateral left atrium, beneath the appendage, to be activated earlier in the radial approach than in the maze procedure, resulting in a shorter left atrial activation time in the radial approach than in the maze procedure. Detouring the activation sequence around the left atrial appendage resulted in delayed activation of this region in the maze procedure. The activation sequence of the interatrial septum was also normal after the radial approach, whereas it was delayed in the maze procedure. Moreover, the present data demonstrated scarring in the posterior atrial septum of animals after the maze procedure, caused by interruption of the blood supply to the posterior septum at two different sites [6]. The septal incision interrupts the left anterior atrial artery, and the intercaval longitudinal incision interrupts the blood supply from the right intermediate atrial artery through the crista terminalis. In humans, the atrial septum is supplied by the arteria anastomotica auricularis magna (Kugel s artery) originating from the proximal segment of either the right coronary artery or the left circumflex coronary artery [11] and by the atrioventricular node artery from the distal right coronary artery. Because the septal incision of the maze procedure intersects the anterior limbus

12 Ann Thorac Surg NITTA ET AL 1999;67:36 50 RADIAL APPROACH FOR AF: PART 2 47 Table 3. Peak Velocity and Time Velocity Integral of Transmitral Doppler Flow a Variable Preoperative Radial Approach Maze Procedure Peak E (cm/s) Peak A (cm/s) Peak E/A b Ei (cm) Ai (cm) AFF (%) c,d a Data presented are mean standard deviation. b p 0.05, maze procedure versus before operation and radial approach. c p 0.005, maze procedure versus before operation. d p 0.01, maze procedure versus radial approach. AFF atrial filling fraction; Ai time-velocity integral of A wave; Ei time velocity integral of E wave; Peak A peak velocity of A wave; Peak E peak velocity of E wave; Peak E/A ratio of peak velocity of E wave to A wave. at its center, the incision may interrupt the blood supply to the posterior septum beyond the incision in humans. Elimination of Atrial Fibrillation The radial approach prevented the induction of sustained AF in canine atria equally as well as the maze procedure. The rationale for operation in AF is that a line of conduction block interrupts the macroreentrant pathway and blocks the fibrillatory wavelets by narrowing the atrial tissue [1]. It is theoretically impossible for the fibrillatory wavelet to reciprocate within atrial tissue in which the circumference is smaller than the wavelength of the wavelet. The radial approach uses the posterior left atrium as a contractile atrial component, resulting in a larger area of atrial tissue in the posterior and superior left atrium to be activated than in the maze procedure. We extended the upper left atrial incision toward the right upper pulmonary vein orifice with approximately Fig 11. The AFF across the mitral valve after the maze procedure and the radial approach. The AFF for the radial approach is within normal limits for a day. The AFF after the radial approach was significantly larger than that after the maze procedure (p 0.01). (NS not significant.) 2 cm of atrial tissue left between the end of the incision and the right upper pulmonary vein orifice to minimize the potential for reentry in this region and to allow the activation wavefront to enter the posterior left atrium. Moreover, the tissue between the left atrial incision and the pulmonary vein orifice was cryoablated to eliminate the potential reentrant pathway around one or more pulmonary vein orifices. More recently, microreentry or automatic activity arising from one pulmonary vein orifice has been shown to be a source for AF in certain patients [12]. Therefore, it is strongly recommended that not only the tissue between the incision and the pulmonary vein orifice be cryoablated, but also each pulmonary vein orifice be cryoablated circumferentially to block the activation propagating from the pulmonary vein orifices in patients. In the present study, the radial approach was tested only in AF induced in otherwise normal canine atria. Further study is necessary to confirm that the procedure also prevents AF in enlarged or spontaneously fibrillating atria. Table 4. Peak Velocity and Time Velocity Integral of Transtricuspid Doppler Flow a Variable Preoperative Radial Approach Maze Procedure Fig 10. Ratio of the peak velocity of the E wave to A wave (peak E/A) across the mitral valve after the maze procedure and the radial approach. The horizontal line and the shaded zone represent the average and the range (within 1 standard deviation) of peak E/A in normal dogs. Peak E/A after the radial approach was significantly smaller than after the maze procedure (p 0.05). (NS not significant.) Peak E (cm/s) Peak A (cm/s) Peak E/A Ei (cm) Ai (cm) AFF (%) a Data presented are mean standard deviation. AFF atrial filling fraction; Ai time velocity integral of A wave; Ei time velocity integral of E wave; Peak A peak velocity of A wave; Peak E peak velocity of E wave; Peak E/A ratio of peak velocity of E wave to A wave.

13 48 NITTA ET AL Ann Thorac Surg RADIAL APPROACH FOR AF: PART ;67:36 50 Table 5. Pulmonary Venous Flow Variable Preoperative Radial Approach Maze Procedure Peak S (cm/s) Peak D (cm/s) S/D b Si (cm) Di (cm) b Si/Di b c a Data presented are mean standard deviation. b P 0.05, radial approach and maze procedure versus before operation. c p 0.001, maze procedure versus before operation. Di time velocity integral of D wave; Peak D peak velocity of D wave; Peak S peak velocity of S wave; Si time velocity integral of S wave; Si/Di ratio ratio of time integral velocity of S wave to D wave; S/D ratio ratio of peak velocity of S wave to D wave. Atrial Contribution to Ventricular Filling The most important finding in the present study was that the left atrial contribution to ventricular filling was larger in animals after the radial approach than after the maze procedure. The peak E/A was smaller and the AFF was larger in animals after the radial approach than after the maze procedure. The mechanism for this superior atrial mechanical function of the radial approach may be explained by the more physiologic atrial activation sequence. Because the atrial incisions parallel the major directions of activation wavefronts and atrial coronary arteries, the procedure preserves a more synchronized left atrial activation sequence and blood supply to most atrial segments. Moreover, there was no electrically and mechanically isolated or scarred region in animals after the radial approach. We measured the AFF, as determined by the Doppler flow spectra, as an index of atrial contribution to ventricular filling. Although it is theoretically possible to quantify the relative amount of flow occurring at a given phase in diastole by measuring the flow velocity integral at the level of the atrioventricular annulus, precise assessment of the flow volume requires accurate measurement of annular area and the assumption that the annular size remains constant, which is not the case. Therefore, we expressed the relative ratio of flow (AFF) rather than actual flow by sampling ventricular inflow at the level of the leaflet tips in this study. The AFF has been shown to quantify the atrial contribution to ventricular filling in normal and diseased hearts [13]. Feinberg and colleagues [3] used Doppler echocardiography to examine atrial function in patients after the maze procedure, 72% of whom were without organic heart disease. They revealed that an A wave was detectable across the mitral valve in 61% of patients, and the mean AFF was 20% in patients with a detectable A wave. Other investigators [4 6] showed that the AFF ranged from 18% to 20% in patients after the modified maze procedure concomitant with mitral valve operation. Atrial function in dogs has not been studied using Doppler echocardiography. The AFF in normal dogs was 33.1% in the present study, a value similar to that in normal human control subjects [3, 13]. The A wave was detectable in 4 of 5 animals after the maze procedure, and the AFF was 14.8% in these animals. Although the AFF after the radial approach (29.9%) was significantly greater than the AFF after the maze procedure in normal dogs, further studies in a long-term model of AF are required. As shown in Table 3, the significant differences in the peak E/A and the AFF of transmitral flow between the two procedures were brought about by the relatively smaller A waves and larger E waves after the maze procedure than after the radial approach. Similar results have been observed in patients after the maze procedure [4]. The time velocity integral of the E wave in patients after the maze procedure was larger than that in normal subjects, whereas the A wave was identical, resulting in a smaller AFF in patients after the maze procedure than in normal subjects. Increased flow velocity and volume during early diastole may be the compensatory process in ventricular filling. Because of a small volume transported into the ventricle during the late filling phase as a result of lesser atrial contraction, the amount of volume transported during early diastole is increased to maintain sufficient ventricular filling, which is the preload for the ventricle to generate cardiac output. The left atrial contribution to ventricular filling is not only determined by the mechanical left atrial contraction, but is also affected by diastolic function of the left ventricle. In a heart with increased ventricular stiffness, diastolic filling of the ventricle is impaired. Elevated left ventricular end-diastolic pressure diminishes the pressure gradient between the left ventricle and the left atrium, thus impairing passive filling of the ventricle, which occurs in early diastole. Ventricular filling is compensated by atrial contraction in late diastole, leading to a larger A wave, and thus a smaller peak E/A and a larger AFF. Ventricular stiffness can be caused by abnormal myocardial elements with malaligned muscle fibers, increased extracellular components, or interstitial fibrosis [14]. In the present study, the differing ages of the animals or myocardial ischemia during cardioplegic arrest, or both, could have affected the ventricular stiffness. However, the age of the animals ranged from 6 to 13 months in the present study, which excludes the possibility of age-related causes of ventricular stiffness [13]. In addition, the cardiopulmonary bypass and the cardioplegic arrest times were identical between the two procedures. In fact, the mitral deceleration time, which is the best single measurement of chamber stiffness [15], was within the normal range in animals after both procedures. Therefore, a difference in ventricular stiffness may not be the cause of the difference in the peak E/A and AFF between the procedures, suggesting that left atrial mechanical function was better preserved in animals after the radial approach than in animals after the maze procedure. Although there was a substantial atrial contribution in cardiac output both after the radial approach and the maze procedure, the difference in percent atrial contribution between the two procedures did not reach statis-