- Email: [email protected]
Pulmonary arteriovenous malformations: Options for management
CORRESPONDENCE
Ann Thorac Surg
519
1994;57:51&24
Pulmonary Arteriovenous Malformations: Options for Management To the Editor: Puskas and colleagues recently published an article [ l ] on therapeutic options for pulmonary arteriovenous malformations (PAVMs). Nine of 21 patients were treated surgically and 5 of 21 patients were treated by balloon or coil embolization over a 28-year period. Puskas and colleagues made some interesting observations and, in their discussions and the comments that followed, reference was made to work published by our colleagues and ourselves. I appreciate having the opportunity to respond. We began treating PAVMs 15 years ago using detachable balloons and coils. Results of the first 76 patients treated by us were published in 1988 [2]. As of December 1992, we had treated an additional 84 patients, bringing our experience up to 160 consecutive patients treated by embolotherapy. In Puskas and colleagues’ series, they have 2 patients with late balloon deflations, both with PAVMs that were thought to have “recanalized.” Patient 5 had documented balloon inflation for 4 months, but not at 5 years, and patient 6 had balloon deflation discovered 4 years after placement. Both PAVMs were “recanalized.” Experimental work reported from our laboratories in 1979 demonstrated in animals that as long as balloons remained inflated 21 days, there was antegrade and retrograde organized thrombus formation to the next branch point and continued occlusion 6 months after deflation of the balloon [3, 41. The deflated shell of the balloon was always incorporated into organized thrombus. In patient 5 of Puskas and colleagues’ series, the balloon was inflated for 4 months. Why was this vessel recanalized? Doctor Jacques Remy and colleagues have demonstrated with high-resolution computed tomography a “reperfusion” phenomenon in 10% of PAVMs occluded with coils [5]. In most instances, this is due to an accessory feeding artery to the PAVM that was not recognized initially or enlarged with time. Less commonly, there may be continued patency through coils placed in the afferent artery, and additional coils placed at a second study are needed to produce permanent occlusion. We have also observed this phenomenon since our 1988 study and we agree with Remy and colleagues’ explanation. Twenty percent of PAVMs are complex with multiple feeding arteries, so ligating or occluding radiologically the principal afferent feeder is not enough [6]. If arterial oxygen tension decreases 30 days after occlusion or the aneurysm does not regress on the chest radiograph, then a reperfusion phenomenon has occurred and requires a simple overnight admission for occluding the remaining feeders or the still-patent primary afferent artery. What about stroke and brain abscess after embolotherapy or operation? We have not observed stroke or brain abscess as a complication of embolotherapy in our patients, although migration of balloons or coils at the time of deployment is well documented by others in the learning phase of this technique. We have seen stroke occur years after treatment when large PAVMs were left unoccluded [7]. Recently, we documented that PAVMs with 3 mm or larger afferent arteries are responsible for paradoxical embolization and stroke and, therefore, we now recommend occlusion of all PAVMs with 3 mm or larger arteries [ 8 ] . Patients with treated PAVMs need follow-up every 3 to 5 years to assess growth of smaller PAVMs not occluded initially. It is our practice to place these patients on a prophylactic antibiotic regimen, as per the American Heart Association guidelines, before dental work for the rest of their lives. Brain abscess
has been documented after dental work; there are always small unoccluded PAVMs present in patients after treatment. As Puskas and colleagues pointed out, we had only restudied by pulmonary angiography 16 of 76 patients through 1987 [2]. We requested these patients return for additional treatment when we realized how common paradoxical embolization was from large PAVMs that were left unoccluded. Up to 1984, our goal had been to increase the arterial oxygen tension to 60 mm Hg or greater, not to block all the large PAVMs. In the 16 patients we restudied between 1978 and 1987, all arteries originally occluded were still occluded. Since 1989, our practice has been to obtain sitting arterial oxygen tension values on room air and on 100% oxygen before and 24 hours after embolotherapy. We also obtain a chest radiograph (Fig 1)and repeat arterial blood gas measurements 30 days after occlusion and every 3 to 5 years thereafter, as growth of small unoccluded PAVMs is well documented with time [9, 101. Finally, Dr McLaughlin queried why the results of embolotherapy of 5 patients with PAVMs in Puskas and colleagues’ series was different from the results of 76 patients occluded at Johns Hopkins Hospital between 1978 and 1988, and 84 patients occluded at Yale-New Haven Hospital between 1988 and 1992. As with all procedures, particularly challenging ones in patients with rare disorders, one has to perform an as yet undefined number per year to minimize complications and maximize results. This is certainly true of embolotherapy of PAVMs, and we believe experience by the same operator and colleagues makes a difference. We agree with Puskas and colleagues that additional long-term assessment of PAVM recurrences is necessary after embolotherapy or operation. We believe the risk of balloon or coil migration is small and is only a factor at the time of the procedure. These risks are minimal when radiologists are experienced in treating PAVMs, and we have not observed stroke or brain abscess occur as a complication of embolotherapy in our patients [2]. We still believe conservative balloon or coil embolization remains the treatment of choice for PAVM.
Robert I . White, Jr, M D Iefrey 5. Polluk, M D Department of Diagnostic Radiology Yule University, School of Medicine 333 Cedar S t PO Box 208042 New Haven, CT 06520-8042
References 1. Puskas JD, Allen MS, Moncure AC, et al. Pulmonary arteriovenous malformations: therapeutic options. Ann Thorac Surg 1993;56:253-8. 2. White RI Jr, Lynch-Nyhan A, Terry P, et al. Pulmonary arteriovenous malformations: techniques and long-term outcome of embolotherapy. Radiology 1988;169:66>9. 3. Barth KH, White RI Jr, Kaufman SL, Strandberg JD. Metrizamide, the ideal radiopaque filling material for detachable silicone balloon embolization. Invest Radiol 1979;14:3!540. 4. Kaufman SL, Strandberg JD, Barth KH, Gross GS, White RI Jr. Therapeutic embolization with detachable silicone balloons: long term effects in swine. Invest Radiol 1979;14:15661. 5. Remy J, Remy-Jardin M, Wattinne L, Deffontaines C. Pulmonary arteriovenous malformations: evaluation with CT of the chest before and after treatment. Radiology 1992;182:809-16. 6. White RI Jr, Mitchell SE, Barth KH, et al. Angioarchitecture of pulmonary arteriovenous malformations: an important consideration before embolotherapy. AJR 1983;140:681-6.
E
D F Fig 1. Chest radiographs (A, B) and selective right and left pulmonary angiograms (C-F) before embolotherapy (A, C, E ) and 7 years after embolotherapy (B, 0, F ) in a 54-year-old man with hereditary hemorrhagic telangiectasia and a large right and left lower lobe pulmonary arteriovenous malformation (PAVM). This patient was initally seen at age 47 years with dyspnea and fatigue and a standing room air arterial oxygen tension of 39 mm Hg. His selective right and left pulmonary angiograms demonstrated a large right and left lobe PAVM with aferent (simple) artery to each PAVM. After occlusion (two balloons in the right PAVM and one in the left PAVM) his arterial oxygen tension increased to 72 mm Hg. He returned to Yale 7 years later and in the interim had done well without recurrence of dyspnea and fatigue. The balloons had deflated between 2 and 3 years after placement. Selective pulmonary angiogram performed 7 years after treatment ( F ) demonstrated continued occlusion of the PAVMs. His sitting arterial oxygen tension in room air was 66 mrn Hg.
B
A
C
N 0
ul
CORRESPONDENCE
Ann Thorac Surg
521
1994;5751&24
7. Hewes RC, Auster M, White RI Jr. Cerebral embolism: first manifestation of pulmonary arteriovenous malformation in patients with hereditary hemorrhagic telangiectasia. Cardiovasc Intervent Radio1 1985;8:151-5. 8. Rosenblatt M, Pollak JS, Fayad PB, Egglin TE, White RI Jr. Pulmonary arteriovenous malformations: what size should be treated to prevent embolic stroke? Radiology 1993;186:937. 9. White RI Jr. Pulmonary arteriovenous malformations: how do we diagnose them and why is it important to do so? Radiology 1992;182:633-5. 10. Vase P, Holm M, Arendrup H. Pulmonary arteriovenous fistulas in hereditary hemorrhagic telangiectasia. Acta Med Scand 1985;218:105-9.
Reply
To the Editor:
In brief, Drs White and Pollak argue that (1) animal models have shown permanent organized thrombus in surgically created systemic arteriovenous fistulas after balloon embolotherapy if the balloons remained inflated 21 days or longer; (2) the “reperfusion phenomenon” occurs in approximately 10% of PAVMs occluded with coils, and that this is usually due to accessory feeding arteries to the PAVM that were not recognized and embolized initially (technical error); and (3) operator experience is critical to successful embolotherapy of PAVMs, and few institutions have the experience to perform this uncommon angiographic procedure effectively. In response, let us restate that in our 2 failed cases of embolotherapy we were able to document inflation of the balloons for greater than 21 days, and that after 4 and 5 years, respectively, these 2 patients had angiographically documented recanalization of their same feeding vessels. The repeat angiograms did not show new or additional accessory vessels. It is reasonable to conclude that in these 2 patients the PAVMs responded to balloon occlusion differently than did the systemic arteriovenous fistulas cited in the animal literature. Pulmonary arteries occluded by pulmonary emboli are known to recanalize after a period of time. This is the most likely scenario to explain the 2 patients in our series who “recanalized’ their PAVMs. The surgical specimens did not reveal any remnants of the deflated balloon-it presumably either dissolved, migrated, or simply could not be identified in the specimen. Indeed, the skill of the operator is crucial to the successful outcome of any procedure. We thought that these 2 patients represented a failure of the balloon and the inherent liability of the procedure rather than any lack of experience in the performance of the embolization. In fact, the PAVMs were ”successfully” occluded for 4 and 5 years. Once the balloons were no longer inflated in these 2 patients, the remaining thrombus presumably recanalized. Balloon embolotherapy clearly has a role to play in the management of patients with numerous PAVMs not amenable to resection, patients who refuse operation, and those with other contraindications to operation. If embolotherapy is used, patients should be followed up at yearly intervals for signs of recurrence of their PAVMs. We believe that conservative resection remains the treatment of choice for large or solitary PAVMs whenever feasible and for selected patients with multiple PAVMs.
Iohn D. Puskas, M D Douglas 1. Mathisen, M D Thoracic Surgical Service Massachusetts General Hospital Warren 11th Floor Boston, M A 02114
~ ~ To the Editor:
~Approach ~ to the~ Mitralivalve~
~
We read with interest the article by Dr Kon and associates [I] about the superior-septa1 exposure of the left atrium. We also have had some experience with this approach, which really gives a superb view of the mitral valve. However, we would like to warn everybody against the indiscriminate use of this approach, as it may be associated with severe, sometimes life-threatening arrhythmias. We have used the superior-septa1 approach in 15 patients in the last 2 years. Included were 9 patients undergoing reoperations for mitral xenograft failure, 5 patients with double-valve replacements (aortic and mitral), and a single patient with open mitral valve repair who had a small left atrium. Cardioplegia was applied for myocardial protection in 6 patients (retrograde continuous blood cardioplegia in 4 and intermittent cold crystalloid cardioplegia in 2 patients), and in 9 patients the operations were done without aortic cross-clamping, the myocardium being continuously perfused from the aorta. Preoperatively 11 patients were in sinus rhythm and 4 had chronic atrial fibrillation. At the end of the operation 11 patients were in junctional rhythm, 1 was in sinus rhythm, 1 in atrial fibrillation, and 2 in atrioventricular block, which lasted 6 hours in 1 and 24 hours in the other. On the first, second, and third postoperative days the following rhythms were recorded: junctional rhythm in 8, 6, and 6 patients, respectively; sinus rhythm in 3, 4, and 2 patients; and atrial fibrillation in 4, 5, and 7 patients. At the end of 6 weeks 10 patients were back to their original sinus rhythm (2 of them with prolonged PQ times of 0.27 and 0.24 seconds), 4 were in atrial fibrillation (the same persons as preoperatively), and 1 patient had junctional rhythm, with a pacemaker implanted at 18 days postoperatively. During the postoperative course 8 patients needed transient pacing from 24 hours to 6 days; 2 other patients received a permanent pacemaker on the 18th postoperative day. Only 5 patients were not paced postoperatively. There was no difference in the frequency of development of junctional rhythm postoperatively depending on the type of myocardial protection, the operation performed, or the preoperative rhythm. Junctional rhythm developed postoperatively in all 4 patients who were in atrial fibrillation preoperatively. We have lost 1 patient: a 62-year-old woman in chronic atrial fibrillation who came to operation in failure owing to mitral xenograft insufficiency. Her graft was replaced with a St Jude valve, using the superior-septa1 approach, without aortic crossclamping. Postoperatively junctional bradycardia developed, and she was paced through a temporary myocardial electrode. On the eleventh postoperative day, however, her pacing wire was inadvertantly removed, which was followed by six episodes of ventricular fibrillation. She was defibrillated repeatedly, and an intravenous electrode was passed into the right ventricle, which stabilized her condition. On the 18th postoperative day she received a permanent pacemaker. Five weeks postoperatively her atrial fibrillation returned. Unfortunately, she never regained consciousness, and in spite of good heart function she died on the 12th postoperative week. Although the number of our patients is small, the high incidence of early postoperative arrhythmias points to the causal role of the superior-septa1 incision, very likely by transecting the sinus node artery. Similarly, damage of this artery in Mustard or Senning operations usually results in the development of junctional rhythm (21. In contrast, in another group of 300 patients,
-
Ann Thorac Surg
519
1994;57:51&24
Pulmonary Arteriovenous Malformations: Options for Management To the Editor: Puskas and colleagues recently published an article [ l ] on therapeutic options for pulmonary arteriovenous malformations (PAVMs). Nine of 21 patients were treated surgically and 5 of 21 patients were treated by balloon or coil embolization over a 28-year period. Puskas and colleagues made some interesting observations and, in their discussions and the comments that followed, reference was made to work published by our colleagues and ourselves. I appreciate having the opportunity to respond. We began treating PAVMs 15 years ago using detachable balloons and coils. Results of the first 76 patients treated by us were published in 1988 [2]. As of December 1992, we had treated an additional 84 patients, bringing our experience up to 160 consecutive patients treated by embolotherapy. In Puskas and colleagues’ series, they have 2 patients with late balloon deflations, both with PAVMs that were thought to have “recanalized.” Patient 5 had documented balloon inflation for 4 months, but not at 5 years, and patient 6 had balloon deflation discovered 4 years after placement. Both PAVMs were “recanalized.” Experimental work reported from our laboratories in 1979 demonstrated in animals that as long as balloons remained inflated 21 days, there was antegrade and retrograde organized thrombus formation to the next branch point and continued occlusion 6 months after deflation of the balloon [3, 41. The deflated shell of the balloon was always incorporated into organized thrombus. In patient 5 of Puskas and colleagues’ series, the balloon was inflated for 4 months. Why was this vessel recanalized? Doctor Jacques Remy and colleagues have demonstrated with high-resolution computed tomography a “reperfusion” phenomenon in 10% of PAVMs occluded with coils [5]. In most instances, this is due to an accessory feeding artery to the PAVM that was not recognized initially or enlarged with time. Less commonly, there may be continued patency through coils placed in the afferent artery, and additional coils placed at a second study are needed to produce permanent occlusion. We have also observed this phenomenon since our 1988 study and we agree with Remy and colleagues’ explanation. Twenty percent of PAVMs are complex with multiple feeding arteries, so ligating or occluding radiologically the principal afferent feeder is not enough [6]. If arterial oxygen tension decreases 30 days after occlusion or the aneurysm does not regress on the chest radiograph, then a reperfusion phenomenon has occurred and requires a simple overnight admission for occluding the remaining feeders or the still-patent primary afferent artery. What about stroke and brain abscess after embolotherapy or operation? We have not observed stroke or brain abscess as a complication of embolotherapy in our patients, although migration of balloons or coils at the time of deployment is well documented by others in the learning phase of this technique. We have seen stroke occur years after treatment when large PAVMs were left unoccluded [7]. Recently, we documented that PAVMs with 3 mm or larger afferent arteries are responsible for paradoxical embolization and stroke and, therefore, we now recommend occlusion of all PAVMs with 3 mm or larger arteries [ 8 ] . Patients with treated PAVMs need follow-up every 3 to 5 years to assess growth of smaller PAVMs not occluded initially. It is our practice to place these patients on a prophylactic antibiotic regimen, as per the American Heart Association guidelines, before dental work for the rest of their lives. Brain abscess
has been documented after dental work; there are always small unoccluded PAVMs present in patients after treatment. As Puskas and colleagues pointed out, we had only restudied by pulmonary angiography 16 of 76 patients through 1987 [2]. We requested these patients return for additional treatment when we realized how common paradoxical embolization was from large PAVMs that were left unoccluded. Up to 1984, our goal had been to increase the arterial oxygen tension to 60 mm Hg or greater, not to block all the large PAVMs. In the 16 patients we restudied between 1978 and 1987, all arteries originally occluded were still occluded. Since 1989, our practice has been to obtain sitting arterial oxygen tension values on room air and on 100% oxygen before and 24 hours after embolotherapy. We also obtain a chest radiograph (Fig 1)and repeat arterial blood gas measurements 30 days after occlusion and every 3 to 5 years thereafter, as growth of small unoccluded PAVMs is well documented with time [9, 101. Finally, Dr McLaughlin queried why the results of embolotherapy of 5 patients with PAVMs in Puskas and colleagues’ series was different from the results of 76 patients occluded at Johns Hopkins Hospital between 1978 and 1988, and 84 patients occluded at Yale-New Haven Hospital between 1988 and 1992. As with all procedures, particularly challenging ones in patients with rare disorders, one has to perform an as yet undefined number per year to minimize complications and maximize results. This is certainly true of embolotherapy of PAVMs, and we believe experience by the same operator and colleagues makes a difference. We agree with Puskas and colleagues that additional long-term assessment of PAVM recurrences is necessary after embolotherapy or operation. We believe the risk of balloon or coil migration is small and is only a factor at the time of the procedure. These risks are minimal when radiologists are experienced in treating PAVMs, and we have not observed stroke or brain abscess occur as a complication of embolotherapy in our patients [2]. We still believe conservative balloon or coil embolization remains the treatment of choice for PAVM.
Robert I . White, Jr, M D Iefrey 5. Polluk, M D Department of Diagnostic Radiology Yule University, School of Medicine 333 Cedar S t PO Box 208042 New Haven, CT 06520-8042
References 1. Puskas JD, Allen MS, Moncure AC, et al. Pulmonary arteriovenous malformations: therapeutic options. Ann Thorac Surg 1993;56:253-8. 2. White RI Jr, Lynch-Nyhan A, Terry P, et al. Pulmonary arteriovenous malformations: techniques and long-term outcome of embolotherapy. Radiology 1988;169:66>9. 3. Barth KH, White RI Jr, Kaufman SL, Strandberg JD. Metrizamide, the ideal radiopaque filling material for detachable silicone balloon embolization. Invest Radiol 1979;14:3!540. 4. Kaufman SL, Strandberg JD, Barth KH, Gross GS, White RI Jr. Therapeutic embolization with detachable silicone balloons: long term effects in swine. Invest Radiol 1979;14:15661. 5. Remy J, Remy-Jardin M, Wattinne L, Deffontaines C. Pulmonary arteriovenous malformations: evaluation with CT of the chest before and after treatment. Radiology 1992;182:809-16. 6. White RI Jr, Mitchell SE, Barth KH, et al. Angioarchitecture of pulmonary arteriovenous malformations: an important consideration before embolotherapy. AJR 1983;140:681-6.
E
D F Fig 1. Chest radiographs (A, B) and selective right and left pulmonary angiograms (C-F) before embolotherapy (A, C, E ) and 7 years after embolotherapy (B, 0, F ) in a 54-year-old man with hereditary hemorrhagic telangiectasia and a large right and left lower lobe pulmonary arteriovenous malformation (PAVM). This patient was initally seen at age 47 years with dyspnea and fatigue and a standing room air arterial oxygen tension of 39 mm Hg. His selective right and left pulmonary angiograms demonstrated a large right and left lobe PAVM with aferent (simple) artery to each PAVM. After occlusion (two balloons in the right PAVM and one in the left PAVM) his arterial oxygen tension increased to 72 mm Hg. He returned to Yale 7 years later and in the interim had done well without recurrence of dyspnea and fatigue. The balloons had deflated between 2 and 3 years after placement. Selective pulmonary angiogram performed 7 years after treatment ( F ) demonstrated continued occlusion of the PAVMs. His sitting arterial oxygen tension in room air was 66 mrn Hg.
B
A
C
N 0
ul
CORRESPONDENCE
Ann Thorac Surg
521
1994;5751&24
7. Hewes RC, Auster M, White RI Jr. Cerebral embolism: first manifestation of pulmonary arteriovenous malformation in patients with hereditary hemorrhagic telangiectasia. Cardiovasc Intervent Radio1 1985;8:151-5. 8. Rosenblatt M, Pollak JS, Fayad PB, Egglin TE, White RI Jr. Pulmonary arteriovenous malformations: what size should be treated to prevent embolic stroke? Radiology 1993;186:937. 9. White RI Jr. Pulmonary arteriovenous malformations: how do we diagnose them and why is it important to do so? Radiology 1992;182:633-5. 10. Vase P, Holm M, Arendrup H. Pulmonary arteriovenous fistulas in hereditary hemorrhagic telangiectasia. Acta Med Scand 1985;218:105-9.
Reply
To the Editor:
In brief, Drs White and Pollak argue that (1) animal models have shown permanent organized thrombus in surgically created systemic arteriovenous fistulas after balloon embolotherapy if the balloons remained inflated 21 days or longer; (2) the “reperfusion phenomenon” occurs in approximately 10% of PAVMs occluded with coils, and that this is usually due to accessory feeding arteries to the PAVM that were not recognized and embolized initially (technical error); and (3) operator experience is critical to successful embolotherapy of PAVMs, and few institutions have the experience to perform this uncommon angiographic procedure effectively. In response, let us restate that in our 2 failed cases of embolotherapy we were able to document inflation of the balloons for greater than 21 days, and that after 4 and 5 years, respectively, these 2 patients had angiographically documented recanalization of their same feeding vessels. The repeat angiograms did not show new or additional accessory vessels. It is reasonable to conclude that in these 2 patients the PAVMs responded to balloon occlusion differently than did the systemic arteriovenous fistulas cited in the animal literature. Pulmonary arteries occluded by pulmonary emboli are known to recanalize after a period of time. This is the most likely scenario to explain the 2 patients in our series who “recanalized’ their PAVMs. The surgical specimens did not reveal any remnants of the deflated balloon-it presumably either dissolved, migrated, or simply could not be identified in the specimen. Indeed, the skill of the operator is crucial to the successful outcome of any procedure. We thought that these 2 patients represented a failure of the balloon and the inherent liability of the procedure rather than any lack of experience in the performance of the embolization. In fact, the PAVMs were ”successfully” occluded for 4 and 5 years. Once the balloons were no longer inflated in these 2 patients, the remaining thrombus presumably recanalized. Balloon embolotherapy clearly has a role to play in the management of patients with numerous PAVMs not amenable to resection, patients who refuse operation, and those with other contraindications to operation. If embolotherapy is used, patients should be followed up at yearly intervals for signs of recurrence of their PAVMs. We believe that conservative resection remains the treatment of choice for large or solitary PAVMs whenever feasible and for selected patients with multiple PAVMs.
Iohn D. Puskas, M D Douglas 1. Mathisen, M D Thoracic Surgical Service Massachusetts General Hospital Warren 11th Floor Boston, M A 02114
~ ~ To the Editor:
~Approach ~ to the~ Mitralivalve~
~
We read with interest the article by Dr Kon and associates [I] about the superior-septa1 exposure of the left atrium. We also have had some experience with this approach, which really gives a superb view of the mitral valve. However, we would like to warn everybody against the indiscriminate use of this approach, as it may be associated with severe, sometimes life-threatening arrhythmias. We have used the superior-septa1 approach in 15 patients in the last 2 years. Included were 9 patients undergoing reoperations for mitral xenograft failure, 5 patients with double-valve replacements (aortic and mitral), and a single patient with open mitral valve repair who had a small left atrium. Cardioplegia was applied for myocardial protection in 6 patients (retrograde continuous blood cardioplegia in 4 and intermittent cold crystalloid cardioplegia in 2 patients), and in 9 patients the operations were done without aortic cross-clamping, the myocardium being continuously perfused from the aorta. Preoperatively 11 patients were in sinus rhythm and 4 had chronic atrial fibrillation. At the end of the operation 11 patients were in junctional rhythm, 1 was in sinus rhythm, 1 in atrial fibrillation, and 2 in atrioventricular block, which lasted 6 hours in 1 and 24 hours in the other. On the first, second, and third postoperative days the following rhythms were recorded: junctional rhythm in 8, 6, and 6 patients, respectively; sinus rhythm in 3, 4, and 2 patients; and atrial fibrillation in 4, 5, and 7 patients. At the end of 6 weeks 10 patients were back to their original sinus rhythm (2 of them with prolonged PQ times of 0.27 and 0.24 seconds), 4 were in atrial fibrillation (the same persons as preoperatively), and 1 patient had junctional rhythm, with a pacemaker implanted at 18 days postoperatively. During the postoperative course 8 patients needed transient pacing from 24 hours to 6 days; 2 other patients received a permanent pacemaker on the 18th postoperative day. Only 5 patients were not paced postoperatively. There was no difference in the frequency of development of junctional rhythm postoperatively depending on the type of myocardial protection, the operation performed, or the preoperative rhythm. Junctional rhythm developed postoperatively in all 4 patients who were in atrial fibrillation preoperatively. We have lost 1 patient: a 62-year-old woman in chronic atrial fibrillation who came to operation in failure owing to mitral xenograft insufficiency. Her graft was replaced with a St Jude valve, using the superior-septa1 approach, without aortic crossclamping. Postoperatively junctional bradycardia developed, and she was paced through a temporary myocardial electrode. On the eleventh postoperative day, however, her pacing wire was inadvertantly removed, which was followed by six episodes of ventricular fibrillation. She was defibrillated repeatedly, and an intravenous electrode was passed into the right ventricle, which stabilized her condition. On the 18th postoperative day she received a permanent pacemaker. Five weeks postoperatively her atrial fibrillation returned. Unfortunately, she never regained consciousness, and in spite of good heart function she died on the 12th postoperative week. Although the number of our patients is small, the high incidence of early postoperative arrhythmias points to the causal role of the superior-septa1 incision, very likely by transecting the sinus node artery. Similarly, damage of this artery in Mustard or Senning operations usually results in the development of junctional rhythm (21. In contrast, in another group of 300 patients,
-