Stratified application of the automatic implantable cardioverter defibrillator

Stratified application of the automatic implantable cardioverter defibrillator

J THORAC CARDIOVASC SURG 1988;96:141-9 Stratified application of the automatic implantable cardioverter defibrillator Since June 1983 we have devel...

757KB Sizes 0 Downloads 2 Views




Stratified application of the automatic implantable cardioverter defibrillator Since June 1983 we have developed a stratified regimen for staged implantation of the automatic implantable cardioverter defibrillator system. The protocol for management in patients who have fully recovered from sudden cardiac death is initiated with the use of standard electrophysiologic evaluation. Treatment in order of application has consisted of drugs followed by implantation of the device for patients with drug-refractory arrhythmias in whom direct cardiac surgical intervention for anatomic substratesfor sudden death are absent. In surgical candidates, combinations of coronary revascularization and ablative therapy have been used to mitigrate the potential for lethal arrhythmia. Sensing and defibrillator lead systems have been placed at corrective operations to be followed later by implantation of the cardioverter defibrillator generator for either inducible or spontaneous tachyarrhythmia. This staged application has been effective in markedly reducing actual sudden cardiac death while at the same time saving on unnecessary device implantation. Morbidity of lead implantation alone remains a concern, particularly for infective complications. Additional follow-up is required to assess the validity of this approach.

Gordon N. Olinger, MO; Peter O. Chapman, MO,b Paul J. Troup, MO,b and G. Hossein Almassi, MO; Milwaukee, Wis.

h e automatic implantable cardioverter defibrillator (AICO) is a proved, effective device for detection and immediate treatment of lethal ventricular tachyarrhythmia.!' Clinical experience is still relatively small, * however, and strategies for selective use of the AICO to balance clinical efficacy against cost and potential morbidity are still evolving. Through analysis of our experience beginning with clinical trials in June 1983, we have developed a stratified regimen for staged application of the device that specifically addresses selected use. Materials and methods Patient population. From June 1983 to December 1986, we evaluated 172 patients with lethal ventricular arrhythmias From the Department of Cardiothoracic Surgery; Section of Cardiology," The Medical College of Wisconsin, Milwaukee, Wis. Read at the Thirteenth Annual Meeting of The Western Thoracic Surgical Association, Colorado Springs, Colorado, June 24-27, 1987. Address for reprints: Gordon N. Olinger, MD, Department of CardiothoracicSurgery, The Medical Collegeof Wisconsin, 8700 West Wisconsin Ave., Milwaukee, WI 53226 "According to a personal communication from Cardiac Pacemakers, Inc., St. Paul, Minnesota (May 20, 1987), 1577 patients have AICD implants.

unassociated with acute myocardial infarction. Seventy-five patients were managed with drugs alone (drugs only). In patients who were candidates for the device and in whom device availability early in the experience permitted implantation, the decision not to employ the AICD was based on freedom from spontaneous or inducible ventricular tachyarrhythmia during maintenance of a tolerable regimen of antiarrhythmic agents. Of the other 97 patients, 43 were unresponsive to medical therapy, had no critical coronary artery disease (2:70% cross-sectional stenosis), and had no ventricular scar deemed amenable to surgical ablation. These 43 were treated with the AICD combined as appropriate with drug therapy (drugs/ AICD). Three of the remaining 54 patients were believed to have had arrhythmia mediated by ischemia. They had critical coronary stenoses and no ventricular scar and were treated with coronary revascularization alone (surgery). Sixteen patients had the complete AICD system, leads and generator, placed at the time of cardiac operation (surgery/ AICD). This group was a mix of 12 patients with isolated coronary revascularization and four patients with ablative procedures, three of which were combined with revascularization. The indication for the AICD generator in three of the patients undergoing ablation was concern about reliability of ablation; in the fourth it was the established policy in our first 3 months of experience when staging was not being considered. None of the 12 patients having coronary revascularization had scar amenable to ablation and, although ischemia was a part of their presentation, the AICD was believed to be a necessary backup. This left 35 patients who had implantation of the AICD lead system only in the context of attempted corrective

14 I

The Journal of Thoracic and Cardiovascular Surgery

142 Olinger et al.

Table I. Stratified application of A/CD: Study population M


Ave. age (yr)

75 43 3 16 17

54 36 3 14 14





21 7 0 2 3 -.Q 33

61 57 68 62 57 62 60

Patients Drugs only Drugs/AICD Surgery only Surgery/ AICD Surgery/leads Surgery/leads + AICD Total

Etiology * Cardiomyopathy



3 16 17

4 2 0 0 1

10 I I







26 18 0 0



-.Q 45

0 0 -.Q 12

'Some patients had more than one diagnosis. tPrimary = primary electrical disease without identifiablestructural heart disease

Table II. Stratified application of A/CD: Study population Rhythm

Drugs only Drugs/AICD Surgery only Surgery/ AICD Surgery/leads Surgery/leads + AICD Total



53 13 2 7

22 8









0 21 0 2



0 0 0

37 36 41 39 35




9-75 8-85 39-42 17-82 21-66 16-73

19 2 18 12 17


7 6

LVEF(%j Unknown






VF. Ventricular fibrillation; VT. ventricular tachycardia; LVEF, left ventricular ejection fraction; SO. standard deviation.

coronary revascularization or surgical ablation, or both, These patients all were eligible for delayed generator implantation to be based on postoperative electrophysiologic evaluation and clinical course, Seventeen were managed without the generator (surgery/Ieads) and 18 subsequently had the generator implanted (surgery/leads + AICD), The treatment groups as delineated and their demographic and clinical characteristics are detailed in Tables I to IV. Electrophysiologic evaluation. Preoperative studies were performed in which single, double, and triple premature stimuli were used during three basic pacing rates, Preoperative mapping of the left ventricle was performed selectively when clinically tolerable monomorphic sustained ventricular tachycardia was induced: Anesthesia. For all AICD implants and for procedures necessitating cardiopulmonary bypass, anesthesia was induced with thiopental sodium and with low-dose fentanyl citrate and was maintained with isoflurane and 100% oxygen, Cardiopulmonary bypass. Bypass was maintained at 38 ° C during preablative left ventricular mapping and during postrepair electrophysiologic evaluation, AICD lead system evaluation, and determination of defibrillation threshold, Moderate hypothermia at 30° C was used during revascularization and ablation except in the few cases in which myocardial protection entailed cold hyperkalemic cardioplegia, For these, systemic temperature was reduced to 28° C. Pulsatile perfusion was used throughout cardiopulmonary bypass, Coronary revascularization. In most patients, coronary revascularization was done with intermittent ischemic arrest.' This technique has proved more favorable in terms of postop-

erative ventricular function, particularly in patients with severely compromised ventricles, Ventricular mapping. An attempt was made to target all ablative procedures done for monomorphic ventricular tachycardia." Preoperative electrophysiologic mapping had efficacy in guiding intraoperative ablation in four patients. Intraoperative mapping was attempted in 27 patients with probable mappable monomorphic ventricular tachycardia. Mapping was done through ventriculotomies placed within known scar, anterior in 22 patients and inferior in six patients. A complete endocardial map was achievable in 21 patients. Ablation. Subendocardial resection was performed in nine patients, cryothermal ablation (-60 0 C for 2 minutes) in five patients, and combinations of these techniques in 14 patients, Cryothermia was used when resection could not encompass targeted areas of suspected early activation, Although three of the nine subendocardial resections could be targeted, eight of the nine consisted of circumferential resection of the marginal zone of the subendocardial scar. We have found this more extensive, essentially blind resection to be simpler and perhaps more reliable, AICD lead implantation. Leads were implanted via a sternotomy in all patients having other concomitant cardiac procedures and via a left anterolateral thoracotomy in patients having Al CD implantation alone, On the basis of our evaluation of defibrillation threshold with various lead systems,' we placed anterior and posterior patch leads in most patients, Defibrillation threshold was assessed with a standard protocol for which an external cardioverter defibrillator was used (Intec Systems Inc" Pittsburgh, Pa.). All leads were tunneled

Volume 96 Number 1

A/CD 143

July 1988


m. Antiarrhythmic drugs

at final treatment Antiarrhythmic drug class I

Drugs only Drugs/ACID Surgery only Surgery/ AICD Surgery/leads Surgery/leads + AICD

43 24 3 7




Total patients*


36 23 2 8 6 5


75 43 3 16 17 18

8 0 2 2 5



7 0 I

2 2

Antiarrhythmic drug class according to Vaughan Williams." 'Patients may have been taking more than one drug.

Table IV. Stratified application of A/CD: Cardiac operations

Surgery only

Surgery/A! CD Surgery/leads Surgery/leads Total





3 16 17

3 15 14

0 0











0 5 28

CAB. Coronary artery bypass; MVR. mitral valve replacement. •Average 2.5 grafts; II internal mammary artery grafts.

through the rectus abdominis fascia into a subcostal subcutaneous pocket to be the site of possible generator implantation. AICD generator implantation. Staged generator implantation was done during general anesthesia in the surgery / leads + AICD group. A left subcostal subcutaneous pocket was used in all but one patient who had a preexisting colostomy in the left upper quadrant. In this individual the leadsand later the generator were placed on the right. AICDs were placed within 3 weeks in five patients, from 3 to 6 weeks in seven, at 9 weeks in one, at 8 months in one, in the second year in two, in the third year in one, and in the fourth year in one. Patient follow-up. We have regularly followed up all of the patients who received an AICD. Surgical patients without AICDs have been followed up by personal visit or telephone or through referring physicians. Drugs-only patients were often managed only during drug trials early in the experience and 15 have been lost to follow-up. Definitions. The following standard definitions were used uniformly during the study. Electrophysiologic results

1. Noninducible-fewer than 5 repetitive responses produced by the stimulation protocol 2. Nonsustained ventricular tachycardia-between 5 complexes and 30 seconds of ventricular tachycardia and no requirement for cardioversion for hemodynamic embarrassment 3. Sustained ventricular tachycardia-more than 30 seconds' duration of ventricular tachycardia or requirement for cardioversion for hemodynamic embarrassment 4. Ventricular fibrillation-a ventricular tachyarrhythmia without discernible QRS complexes on the surface electrocardiographic leads monitored

5. Monomorphic ventricular tachycardia-uniform QRS configuration in the monitored electrocardiographic leads (typically I, F, VI) 6. Polymorphic ventricular tachycardia-nonuniform QRS morphology A/CD shock propriety I. Appropriate-the ventricular tachycardia was documented by continuous electrocardiographic recording at the time of shock, or the patient had typical recurrent symptoms or ventricular tachycardia before receiving the shock, or the patient had witnessed syncope with witnessed discharge of the device 2. Inappropriate-the shock was delivered during other than ventricular tachyarrhythmias as documented by telemetry or ambulatory electrocardiographic monitoring 3. Uncertain-all other shocks

Results Mortality rates and causes of death in the various treatment groups are detailed in Table V. There were 11 sudden cardiac deaths in the drugs-only population for an observed mortality rate of 11 per patient-year (patients lost to follow-upare excluded). Three of the six noncardiac deaths were from arniodarone pulmonary toxicity. Two of the seven deaths in the drugs/AICD group were from sudden cardiac death for an observed sudden death mortality of three per patient-year. There were no other sudden cardiac deaths. AICD discharges were documented in 41 of 62 patients at risk for a shock. One hundred twelve of these

14 4

The Journal of Thoracic and Cardiovascular Surgery

Olinger et al.

Table V. Stratified application of A/CD: Mortality Deaths Periop.


CHF. low output


SCD/ pt-yr

5 3 0 1 5 0


0 0 6 0

11 2 0 0 0 0

11 3 0 0 0 0

F/U Drugs only Drugs/AICD Surgery only Surgery/ AICD Surgery/Ieads Surgery/leads + AICD




75 43 3 16

19.7* 16.9 37.9 25.7 13.3

22 7 1 2 7 0

17 18


2 1 1 2 0

F/U. Average length of follow-up; No .. number of deaths from all causes; Periop .. death during hospitalization or within 30 days; SCD. sudden cardiac deaths: unexpected death within 2 hours of onset of symptoms; CHF. congestive heart failure; low output, cardiac death in the setting of low cardiac output. hypotension, and decreased vital organ perfusion; SCD/pt-yr. (No. SCD)/(patient-years of follow-up). • Fifteen drugs-only patients were lost to follow-up

Table VI. Surgery/leads + A/CD: Electrophysiologic indications and results Patients with A/CD shock POSlOp. rhythm Noninducible/spontaneous/V'T/YF Polymorphic nonsustained Polymorphic sustained Monomorphic nonsustained Monomorphic sustained Induced YF Total

2 3 1 7 4

.I 18




0 0 0 2 2 Q 4

0 1 0 3 0

2 2 1 2 2 Q 9

1 5

VT. Ventricular tachycardia; VF. ventricular fibrillation.

were believed to be appropriate and 166 were of uncertain propriety. There were 32 inappropriate shocks documented in 11 patients. Nineteen patients in the drugs/AICD groups sustained shocks of appropriate or uncertain propriety. The hypothetical expected mortality (including actual deaths) from sudden cardiac death in this group was therefore 35 per patient-year. There were two observed deaths in the surgery / AICD patients, one of pneumonia at 11 months and another of heart failure at 8 months. Six patients had appropriate or uncertain AICD discharges for a hypothetical expected mortality from sudden cardiac death in this group of 18 per patient-year. Of the 35 patients with initial AICD leads only and a concomitant other cardiac procedure, six died in the postoperative period, two of infection and four of ventricular dysfunction. This left 29 patients for whom the decision could be made regarding whether or not implantation of the AICD generator was indicated. As summarized in Table VI, 18 patients had either spontaneous recurrent ventricular arrhythmia (one perioperatively and a second in the fourth year of follow-up) or inducible ventricular tachycardia at postoperative electrophysiologic study (16 patients) and therefore underwent AICD generator implantation. Nonsustained

tachycardia was the electrophysiologic result in 10 of these patients and sustained tachycardia in six. Eleven patients had neither spontaneous nor inducible ventricular tachycardia postoperatively and were followed up without generator implantation. One of these patients died 3 months postoperatively in end-stage low-output heart failure terminating in ventricular fibrillation. There were no observed deaths in the surgery/ leads + AICD group. There were documented appropriate or uncertain AICD discharges in nine patients for a hypothetical expected mortality from sudden cardiac death of 22 per patient-year. Although a variety of complications occurred in each of the treatment groups, most germane to this study are those complications specific to the use of the AICD generator as seen in Table VII. There were four significant patient-related complications specific to the generator alone. There has been no documented generator erosion or generator drift. Early generator depletion has been the cause of premature explantation and replacement in 18 patients. The average generator life span (to point of prolonged charge time exceeding the manufacturer's elective replacement indicator) has been 15.3 months (range 1 to 25, standard deviation 6.1). Complications specifically implicating the lead sys-

Volume 96 Number 1 July 1988

tems included documented constrictive pericarditis in one patient, lead fractures in one patient, endocardial lead migration in one patient, coronary erosion in one, and infections of lead systems in six patients (6% of the total of 94 patients who had AICD hardware placed). Details of lead infections are given in Table VIII. Three of these infections occurred fairly early postoperatively in the setting of concomitant or prior infection at other sites in the body. One of these infections in a renal transplant recipient was due to superficial breakdown of the generator implant incision. Three infections occurred late with no antecedent or identifiable concomitant contributory cause. Two of the infected lead systems included a generator. Five of the leads were placed at sternotomy, and one was placed at thoracotomy without other concomitant cardiac operation, the latter in the patient receiving a transplant. Discussion Convincing data from several centers suggest that the AICD can impact significantly on mortality in patients identified at risk for sudden cardiac death. Follow-up of our 35 patients who have undergone coronary revascularization or surgical ablation with implantation initially of AICD leads only provides information that may assist developmentof a protocol for management to optimally select patients in whom AICD system implantation can be done to maximize the ratios of cost and risk against the ultimate benefit of rescue from sudden cardiac death. We have chosen criteria for implantation of the leads based principally on convenience and flexibility with additional consideration given to minimizing morbidity of subsequent AICD generator implantation. Criteria for implantation of the AICD generator in these patients have included considerations regarding the ability of electrophysiologic and clinical evaluation for predicting postoperative sudden cardiac death on the one hand and the considerable cost along with morbidity of repeated generator replacement for unnecessary generator implantation on the other. Although our length of follow-up is still on the order of months to a few years and the numbers of patients in each treatment group are still relatively small, the trends in morbidity and mortality cited permit us to propose a protocol for stratified application of the device that seems reasonable at this time. Patients are screened at the outset according to standard electrophysiologic criteria. Viable survivors without surgically treatable disease are treated with drugs. Now that the AICD is more readily available, implantation for refractoriness to usual or to possibly toxic drug therapy is recommended. Early in our experience during AICD trials, criteria for implantation



Table VII. Stratified application of A/CD: A/CD complications No. Lead specific Fracture Infection Migration Coronary erosion Constrictive pericarditis Generator specific Early depletion Infection Miscellaneous electric


6 1 I 1 18 I


*TWQ subsequent stimulating fractures and one sensing fracture.

were more stringent and at least two of our drugs-only patients died suddenly during extended trials. Additionally, three died of amiodarone pulmonary toxicity. Ideally these deaths would be avoided today. In contrast, six of the drugs-only patients who had sudden cardiac death during follow-up were not candidates for the AICD because of malignant disease, central nervous system depression, or clinical instability. We are concerned that pressure to prevent all sudden cardiac deaths could result in implantation in such patients. Indeed, one of our drugs/AICD deaths occurred on postoperative day 1 from persistent ventricular tachycardia that could not be effectively contained medically preoperatively and, in our judgment now, was treated inappropriately with an AICD. Our approach for implantation of the AICD and leads in patients not requiring another cardiac operation is left anterolateral thoracotomy. This incision with intra pericardia I placement offers good exposure for patch electrode positioning to optimize defibrillation threshold. Sometimes subtle alterations in patch-patch geometry will resolve what otherwise may be an excessive threshold. We have used intrapericardial placement in all but a few patients who had particularly tenacious pericarditis wherein the posterior patch was placed extrapericardially. In patients who have had previous cardiac operations, the anterior patch is positioned posterior to the sternum in whatever anatomic plane is available. This may be intra pericardia I or extrapericardial. Of the 43 drugs/AICD patients, 18 (42%) had had a previous cardiac operation elsewhere, and another seven (16%) had significant pericarditis of drug, viral, or postinfarction origin. Sternotomy is certainly another excellent option for good exposure, but it is an incision with more potential serious morbidity and is not as ideal for repeat operations, since mobilization of the heart for posterior patch electrode placement would be more involved. Even a

The Journal of Thoracic and Cardiovascular

I 4 6 Olinger et al.


Table VIII. A/CD infective complications Age (yrj

Implant incision

Other infection


Yes No No Yes

Sternal Sternal Sternal Thoracic

None None None Subcutaneous pocket







Phlebitis, sternum Sternum

72 70 75


Interval to infection (mol 18 7 30 3


Cardiac operation

A/CD reimplant

Candida Staph. epidermidis Staph. aureus

CAB CAB, ablation CAB None

Yes No No Yes

Staph. aureus



Alive and well Alive and well Alive and well Dead at 3 mo of Serratia sepsis Alive and well

Staph. aureus



Alive and well

Organism Gram-positive


CAB, Coronary artery bypass.

well-constructed thoracotomy can still be potentially hazardous for placement of leads in patients who have had an operation, particularly during the more vigorous phases of healing in the first 3 to 6 weeks. Although we have had no mishaps to date, WI: are ever aware of the threat of violating a coronary artery, vein graft, or mammary graft, or of tearing the anterior wall of the right ventricle. Because of the challenge of our fairly extensive experience with thoracotomy in implanting leads in patients originally operated on elsewhere, we prefer in our patients who undergo primary cardiac operations for sudden death syndrome to place leads at sternotomy and, providing generator implant is not necessarily indicated at that operation (as in our surgery/ AICD group), to access the leads later in the subcutaneous subcostal pocket if it becomes necessary to implant an AICD. This delayed generator implantation is a simple procedure analogous to a generator replacement. Whether this empirically derived policy is appropriate, however, can be questioned. We have concern that the morbidity of the leads alone may be sufficient to outweigh the risk of delayed implantation of the full system, should it be required postoperatively for failure of revascularization or ablation to eliminate the potential for sudden death. Lead morbidity is a significant issue." In our hands it has included constrictive pericardial reaction necessitating a difficult explantation and pericardiectomy, coronary erosion that necessitated reexploration for bleeding, lead fractures that could obviate later usefulness, and, most important, infection. Four of six infected systems were in leads-only patients. Two occurred early postoperatively in association with infection focused elsewhere, which stresses the need for extraordinary precautions to prevent infections in these patients. More disturbing is that three infections, two in leads-only patients, occurred late postoperatively with-

out intercurrent or preceding other infection elsewhere. We do not know whether there was late hematogenous seeding of these devices or whether organisms entrapped at the original procedure remained indolent for the 7,18, and 30 months before the infections appeared. Mucusproducing Staphylococcus epidermidis is an apparent culprit in this process. Nevertheless, the potential for infection is clear and seems to exist for the life of the lead. At present, six of the original surviving surgery/ leads patients are wearing leads that have no apparent utility and remain a small but definite risk to them. The other side of this question is the morbidity of full system implantation after another cardiac operation. Twelve of the 18 delayed AICD implants were inserted at between I and 6 weeks after operation. This is the most risky time for lysis of adhesions, and an additional thoracotomy could be a major stress to someone with a recent sternotomy. Furthermore, lead placement may be less than optimal for obtaining satisfactory defibrillation threshold. Our present opinion is that these latter features outweigh in risk those of infection, but we have no data to support this notion since we have taken the staged approach in the vast majority of AICD implants early after other cardiac operations. Should the prevalence of infection (6% overall) that we have seen persist or increase, we would be persuaded to eliminate the leads-only concept. Another factor that clouds precise delayed postoperative application of the AICD is the selection of criteria that reliably predict persistem risk of sudden cardiac death. Placement of a generator in every patient would maximize efficacy but at a significant recurring cost and risk of potential morbidity. The pulse generator costs about $13,000 and, in our hands, according to the original replacement criteria from the manufacturer, has had an average efficacious life span of 15.3 months. This is consistent with the experience of others? but

Volume 96 Number 1 July 1988

should improve with recently revised replacement criteria and more energy-efficient generators. Generator replacement for premature depletion requires hospitalization and, at our institution, a general anesthetic to reestablish system viability. An additional generator-related problem is that the device can discharge without the capability to retrieve electrographic data that could indicate propriety of the shock. Witnessed shocks generally allow the physician to determine that the discharge was either appropriate or inappropriate, but many (over 50% in our experience) will be of uncertain propriety. Uncertain events still require evaluation and force the physician to prove they are not significant. The psychologic impact of these and inappropriate shocks on the patient and family is undeniable. We have assumed a liberal electrophysiologic definition of failure of surgical therapy in eradicating potential for sudden cardiac death. Any repetitive rhythm, monomorphic or polymorphic, of five or more beats responsive to programmed stimulation of up to three premature beats is deemed inducibility. Using these criteria, we stratified 29 patients. Seventeen patients received generators (16 because of inducible ventricular tachycardia and one because of spontaneous postoperative ventricular tachycardia). During follow-up, four of these patients received appropriate defibrillator shocks and an additional five patients received shocks of uncertain propriety. Therefore nine of 17 or 53% of these patients in whom arrhythmia was predicted to recur by our algorithm were shocked by the device. Hypothetical expected sudden death mortality in patients with AICDs was considered as actual sudden death plus the first AICD shock deemed appropriate or of unknown propriety. This definition is based on the presumption that these discharges occurred in the setting of recurrent lethal arrhythmias and that the patients would have died without the device. Since not all recurrent sustained arrhythmias are fatal and some shocks of uncertain propriety may in fact have been inappropriate, this calculation overestimates the sudden death rate. Until interpretable telemetry becomes incorporated into the implanted defibrillator, determining the propriety of shocks occurring in unmonitored asymptomatic patients will remain a problem. Of the 12 patients without inducible or spontaneous postoperative tachycardia in whom generators were not implanted, one patient had recurrent sustained ventricular tachycardia in the fourth year of follow-up and subsequently had a generator implant. Therefore 11 of 12 patients or 92% predicted to be free of recurrence remained so throughout the follow-up period.

A/CD 147

Ideally we would like to be able to improve our ability to identify patients at low risk of recurrent arrhythmia to minimize the cost and potential morbidity of the device. At the same time any stratification algorithm must maintain a high predictive accuracy in identifying patients who remain at high risk for lethal arrhythmias and who therefore should be protected by the AICD. Other authors) have recommended routine implantation of the pulse generator in patients undergoing endocardial resection for ventricular arrhythmias. We believe that a group of low-risk patients can be identified with our approach, but clearly further study of larger numbers of patients will be necessary to define the optimal application of the pulse generator. It is tempting not to replace a depleted generator in apparently lower risk patients if they have no documented discharge events during AICD follow-up. The problem with this approach is that risk does not necessarily seen to be decremental with time. Griffith and colleagues'? have recently confirmed this notion, showing that first AICD discharge did not occur in some of the AICD-treated patients followed up by the Johns Hopkins group until the second, third, or even fourth generator had been implanted. We are being forced to accept the dictum that "once an AICD, always an AICD." Initial selection, therefore, is critical if unnecessary cost and morbidity are to be avoided. A strategy similar to ours has been suggested by Bolooki and colleagues. 11 They voice many of our concerns. Now that the device is beyond clinical trials and use is widely dispersed, only systematic long-term follow-up and evaluation of sizable groups of patients by programs equipped to do so will sort out the most effective, fiscally and medically responsible strategies for use of the AICD system. We encourage engineering developments that will increase generator longevity and improve assessment of discharge propriety, and we applaud physician education that is imperative to assure that this lifesaving but very expensive technology will be used responsibly in the future.

Addendum In 1987, 12 additional patients entered the surgery/leads group. This brings to 29 the number being followed up in this category. There have been no additional infections. Our policy for stratified application remains unchanged. REFERENCES 1. Reid PR, Mirowski M, Mower MM, et al. Clinical evaluation of the internal automatic cardioverter-defibrillator in survivors of sudden cardiac death. Am J Cardiol 1982;51:1608-13.

1 4 8 Olinger et al.

2. Echt OS, Armstrong K, Schmidt P, Oyer PE, Stinson EB, Winkle RA. Clinical experience, complications, and survival in 70 patients with the automatic implantable cardioverter defibrillator. Circulation 1985;71:289-96. 3. Platia EV, Griffith LSC, Watkins L Jr, et al. Treatment of malignant ventricular arrhythmias with endocardial resection and implantation of the automatic cardioverter defibrillator. N Engl J Med 1986;314:213-6. 4. Josephson ME, Horowitz LN, Spielman SR, Waxman HL, Greenspan AM. Role of catheter mapping in preoperative evaluation of ventricular tachycardia. Am J Cardiol 1982;49:207-20. 5. Olinger GN. Intermittent ischemic arrest. In: Roberts A, ed. Myocardial protection in cardiac surgery. 1st ed. New York: Marcel Dekker, 1986:207-19. 6. Horowitz LN, Josephson ME, Harken AH. Epicardial and endocardial activation during sustained ventricular tachycardia in man. Circulation 1980;61:1227-38. 7. Troup PJ, Chapman PO, Olinger GN, Kleinman LH. The implanted defibrillator: relation of defibrillating lead configuration and clinical variables to defibrillation threshold. J Am Coli Cardiol 1985;6:1315-21. 8. Gabry MD, Brodman R, Johnston 0, et al. Automatic implantable cardioverter-defibrillator: patient survival, battery longevity and shock delivery analysis. J Am Coli Cardiol 1987;9:1349-59. 9. Marchlinski FE, Flores BT, Buxton AE, et al. The automatic implantable cardioverter defibrillator: efficacy, complications, and device failures. Ann Intern Med 1986;104:481-8. 10. Griffith L, Guarnieri T, Mower M, et al. Does an AID generator need replacement when the battery is depleted? [Abstract]. J Am Coli Cardiol 1987;9:168A. 11. Bolooki H, Palatianos GM, Zamaw L, Thurer RJ, Luceri RM, Myerburg RJ. Surgical management of postmyocardial infarction ventricular tachyarrhythmia by myocardial debulking, septal isolation, and myocardial revascularization. J THORAC CARDIOVASC SURG 1986; 92:716-25. 12. Vaughan Williams EM. A classification of antiarrhythmic actions-reassessed after a decade of new drugs. J C1in Pharmacol 1984;24:129-47.

Discussion Dr. Alden Harken (Denver, Colo.). Dr. Olinger, you appear to be a conscientious investigator who is wrestling with a major problem. It is impressive that you have presented these kinds of data, because this is a big problem and the technology has not really caught up yet. Can you walk us through your algorithm as to when this kind of technology is appropriate, because I am not yet sure I understand that. A patient is admitted with "sudden death." You conduct an electrophysiologic test. If he has inducible ventricular tachycardia, meaning a monomorphic kind of rhythm, what do you do? If he has only ventricular fibrillation, what do you do? My impression is that you prescribe a

The Journal of Thoracic and Cardiovascular Surgery

number of drugs. If that fails, then the AICD does make some sense. What do you do with amiodarone, recognizing that it is not as "testable" as some of the other drugs? What do you tell patients about myocardial revascularization, and is that going to make them more or less susceptible to cardiac arrhythmias? One question about the frontiers: Obviously you have wrestled through many of the current problems and historical problems with implantation of these kinds of devices. What is the status currently of an antitachycardia pacer with AICD backup? Dr. Olinger. The algorithm has changed. One reason is that we have more devices now that Cardiac Pacemakers Inc. has taken over, is using more hybrid circuitry than Intec Systems used, and probably has better quality control. Nevertheless, we have more patients who need them than we have devices, as I believe is generally true in this field. If a patient comes in with lethal cardiac arrhythmia and is a viable survivor, not someone with neurologic problems or cancer or another terminal disease, then the patient is potentially a candidate for the device. He does go through drug trials, but not nearly as extensive as they used to be. One reason is that there are not as many experimental drugs. Amiodarone still is probably the drug most commonly used in most of the patients and, as I mentioned, is usually combined with a class I agent. If a patient reacts well to the drug, we prefer to use this treatment, providing the drug regimen is tolerable and we are not pushing the drugs to toxicity. If the drugs are intolerable or if we are pushing them to toxicity, then we will implant the device. Patients should be well controlled by drugs, that is, they should not have incessant tachycardia, because this device is merely "suspenders." It is not going to save them if ventricular tachycardia or ventricular fibrillation is occurring every other day. If there is a surgical substrate for this kind of problem, that is, coronary disease rather than just cardiomyopathy, or a primary electrical problem, then we look carefully at a surgical approach. If the disease appears to be correctable, that is, the patient has coronary disease with a tight lesion (it is not 50% stenosis, it is 90% stenosis), and if there is scar in a patient who has monomorphic tachycardia, something we think we can map and on which we can do an ablative procedure, then we will generally recommend that that patient have some sort of surgical correction. On the other hand, if the patient has a polymorphic tachycardia that cannot be mapped, really cannot be evaluated in the catheterization laboratory, then generally we will go to the AICD. We operated on some patients early in the experience who were very sick, and we may have done a little bit more operatively than we should have. This was commissive both on my part, as a surgeon, and on the part of our electrophysiologists, who aggressively evaluated these patients in the operating room by inducing recurrent fibrillation and ventricular tachycardia, trying to get as much information as possible to evaluate their rhythm problems. Even though we use pulsatile bypass in all these patients and try to maintain decent perfusion pressures, we are dealing with hearts that have been on bypass for 2\!2 or 3 hours already, and at normothermia these hearts do not respond well to excessive stimulation: Their condition visibly deteriorates as electrical stimulation is applied. Today we are very conservative, doing the minimum that is necessary to get the information to put in the device and make sure that we do not have a rhythm that

Volume 96 Number 1 July 1988

will be a problem early postoperatively. If we are satisfied that we have done a good job, we leave just the leads in and do not put in the generator. I have not yet seen the patient whose condition was made worse by revascularization. A number of patients are better because of revascularization, but we still like to use the leads. As far as frontiers go, the people at Cardiac Pacemakers Inc. have some big ideas. We will be entering a trial this fall



with their next generation device. A still more future generation, which is still in the diagrammatic phase, will have everything one could possibly ask for, including multiprogrammability and pacing. Again the question will be how many systems failures will occur with those devices, how much they will cost, and whether they will be worthwhile for every patient.