Long-Term Prognosis of Individuals With Right Precordial ST-Segment–Elevation Brugada Syndrome
Background— Brugada syndrome is an arrhythmogenic disease characterized by an ECG pattern of ST-segment elevation in the right precordial leads and an increased risk of sudden cardiac death as a result of ventricular fibrillation. Controversy exists with regard to risk stratification and therapeutic management, particularly in asymptomatic individuals.
Methods and Results— A total of 212 individuals (mean age, 45±6 years) with a type 1 Brugada ECG pattern were studied. Of these, 123 (58%) were asymptomatic, 65 (31%) had ≥1 syncope of unknown origin, and 24 (11%) had to be resuscitated because of ventricular fibrillation. In 125 individuals (59%), a spontaneous type 1 ECG was recorded. In the remaining, drug challenge with a class I antiarrhythmic agent unmasked a Brugada ECG. The mean ST elevation was 2.3±1.2 mm in symptomatic patients and 1.9±1.5 mm in asymptomatic individuals (P=0.04). During a mean follow-up of 40±50 months, 4 of the 24 patients (17%) with aborted sudden cardiac death and 4 of 65 (6%) with a prior syncope had a recurrent arrhythmic event, whereas only 1 of 123 asymptomatic individuals (0.8%) had a first arrhythmic event. Four of 9 patients with arrhythmic events during follow-up were not inducible during programmed electrical stimulation. A previous history of aborted sudden death or syncope and the presence of a spontaneous type 1 ECG were predictors of adverse outcome.
Conclusions— The present study reports data on a large population of individuals with a type 1 Brugada ECG pattern with the longest follow-up reported so far. A very low incidence of severe arrhythmic events, particularly in asymptomatic individuals, was found during follow-up. In the presence of very few arrhythmic events on follow-up, programmed electrical stimulation showed very little accuracy in predicting outcome.
Received June 17, 2004; revision received September 29, 2004; accepted October 11, 2004.
Brugada syndrome is an arrhythmogenic disease characterized by an ECG pattern of ST-segment elevation in the right precordial leads and an increased risk of sudden cardiac death as a result of polymorphic ventricular tachyarrhythmias or ventricular fibrillation.1 Controversy exists on risk stratification and therapeutic management, particularly in asymptomatic individuals with such an ECG pattern. Brugada et al2 very recently reported on a large number of individuals with an ECG diagnostic of Brugada syndrome and no previous cardiac arrest. During a mean follow-up of 2 years, 8% of these patients suffered sudden cardiac death or had documented ventricular fibrillation. In contrast, Priori et al3,4 demonstrated that asymptomatic individuals and in particular individuals with only transient ECG abnormalities are at low risk of sudden cardiac death. Therefore, our goal was to verify these 2 opposite standpoints and to present long-term follow-up data on clinical and electrophysiological parameters in a large number of individuals with a so-called type 1 ECG5 compatible with Brugada syndrome (Figure 1).
Patients diagnosed with an ECG characteristic of Brugada syndrome at the following 4 university hospitals were included: Academic Medical Center, Amsterdam (the Netherlands); University of Münster (Germany); Medical Clinic of Mannheim, University of Heidelberg (Germany); and University of Nantes (France). Three groups were classified according to the circumstances under which the abnormal ECG was documented. The diagnosis was made after an episode of aborted cardiac arrest (group A), during diagnostic evaluation of syncope of unknown origin (group B), or during routine examination, or as a consequence of family screening after the diagnosis of Brugada syndrome was made in a family member (group C). An “index” patient was defined as the first individual with a clinical diagnosis within a family.
Clinical Data, ECG, and Electrophysiological Testing
Clinical data—including age at diagnosis, gender, family history of sudden cardiac death, and results of drug testing for unmasking the characteristic coved-type ECG pattern and invasive electrophysiological testing—were collected in the 4 centers. Patients were included in this study only if they had a type 1 ECG at baseline or after provocation with a class I antiarrhythmic drug. A type 1 ECG (Figure 1) was defined as a prominent coved ST-segment elevation displaying J-wave amplitude or ST-segment elevation ≥2 mm or 0.2 mV at its peak followed by a negative T wave.5 The choice of drug was determined by the availability of the drug in the participating hospitals concerned. Either intravenous ajmaline (1 mg/kg body weight at a rate of 10 mg/min), flecainide (2 mg/kg body weight in 10 minutes with a maximum of 150 mg), or procainamide (10 mg/kg at a rate of 100 mg/min) was used. Echocardiography, cardiac catheterization, chest x-ray, and exercise testing excluded underlying structural heart disease. In addition, laboratory tests excluded acute ischemia and metabolic or electrolyte disturbances. In total, 212 individuals were studied. Most (n=183, 86%) were genetically tested for the presence of an SCN5A mutation, which has been linked to Brugada syndrome.6
Baseline electrophysiological studies were recommended in all individuals and were finally performed in 186 (88%). A maximum of 3 ventricular extrastimuli were delivered from 2 ventricular sites unless ventricular fibrillation or a sustained ventricular tachyarrhythmia (lasting >30 seconds, causing syncope, or requiring intervention to be terminated) was elicited at a previous step. Premature beats were started in late diastole; coupling intervals were then reduced in 10-ms decrements until refractoriness was reached. Patients with inducible nonsustained ventricular arrhythmias (>6 beats that terminated spontaneously within 30 seconds) were classified as noninducible. Patient treatment was based on clinical judgment of the participating center. No patient received antiarrhythmic drugs, and 113 (53%) received an implantable cardioverter-defibrillator (ICD). During follow-up, patients were considered to have an arrhythmic event if sudden death occurred or ventricular fibrillation was documented.
Mutation Analysis of SCN5A
In the 183 patients (136 index) studied, all 28 exons of SCN5A were amplified by polymerase chain reaction from DNA isolated from peripheral leukocytes using intronic primers. Polymerase chain reaction products were subjected to direct sequencing of all coding regions.7 To increase the probability of detecting the presence of any sequence change, SSCP was carried out at 2 different temperatures for each exon, and the size of fragments for SSCP was kept at ≈300 bp.8
Data were analyzed with the SPSS package for paired and unpaired data and for survival curves. Survival curves were plotted by use of the Kaplan-Meier method and analyzed by the log-rank test. Student’s t test or the Mann-Whitney test was performed when appropriate to test for statistical differences. A value of P<0.05 was considered statistically significant. When applicable, data are presented as mean±SD.
Patient Population and Genetic Analysis
The population consisted of 212 patients (152 male) with a mean age of 45±6 years at diagnosis (median, 45 years; range, 6 to 87 years) (Table 1). At diagnosis, 123 (58%) were asymptomatic, 65 (31%) had had ≥1 syncope of unknown origin, and 24 (11%) had been resuscitated because of ventricular fibrillation. One hundred sixty-five individuals (78%) were index patients, and 47 (22%) were identified during family screening. The asymptomatic individuals were identified mainly during a routine ECG (n=63; 51%), because of a family history of Brugada syndrome (n=40, 32%), or from complaints such as presyncope or palpitations (n=20; 16%). Age at diagnosis was similar between symptomatic and asymptomatic individuals (46±14 versus 44±14 years; P=NS). A predominance of male patients was observed (152 compared with 60 female patients). All but 2 patients with a history of ventricular fibrillation were male (92%). However, gender did not significantly differ between symptomatic and asymptomatic patients (76% male versus 68% female; P=NS).
The study included 57 individuals (27%) with an identified SCN5A mutation. In the group of index patients, 34 patients (21%) had a family history of sudden cardiac death, and 32 of 136 screened index patients (24%; 22 men, 10 women) were found to have an SCN5A mutation. Mutation carriers with a normal ECG both at baseline and after drug provocation were not included in this study.
ECG Parameters, Sodium Channel–Blocking Drugs, and Electrophysiological Investigations
In 125 individuals (59%), a spontaneous coved-type ECG (type 1 ECG) was found that was defined as an ECG pattern at baseline as described above (Figure 1). In the remaining individuals, class I antiarrhythmic drug administration unmasked the diagnostic type 1 ECG. The mean ST elevation was 2.3±1.2 mm in symptomatic patients (2.6±1.4 mm in group A; 2.2±1.2 mm in group B) and 1.9±1.5 mm in asymptomatic individuals (P=0.04). In the presence of class I administration, ST elevation significantly increased (group A, 3.6±1.3 mm; group B, 3.8±1.5; group C, 3.9±1.6; P<0.05). The pattern of ST elevation (spontaneous type 1 ECG) was distributed equally between symptomatic patients (60%) and asymptomatic individuals (55%; P=NS).
Data on the results of programmed electrical stimulation are presented in Table 1. In 93 of 186 individuals investigated (50%), sustained ventricular tachyarrhythmias or fibrillation was inducible. The rate of induction was higher in symptomatic patients compared with asymptomatic individuals (63% versus 39%; P<0.001).
Follow-Up and Predictors of Outcome
The mean follow-up period for the entire study population was 40±50 months. Follow-up was significantly longer in patients with aborted sudden cardiac death (group A, 83±66 months; median, 74 months; range, 12 to 344 months) compared with patients with syncope (group B, 39±37 months; median, 33 months; range, 5 to 226 months) and asymptomatic individuals (group C, 34±52 months; median, 26 months; range, 3 to 178 months) (P=0.0001). This difference is explained by the fact that individuals identified early in the participating centers were all survivors of sudden cardiac death. During follow-up, 4 of the 24 patients (17%) with prior cardiac arrest had a new arrhythmic event (ie, ICD intervention as a result of a sustained ventricular tachyarrhythmia or ventricular fibrillation). In patients identified after syncope, 4 of 65 patients (6%) had an arrhythmic event. In asymptomatic individuals, only 1 of 123 individuals (1%) had a first arrhythmic event. The difference in outcome in the 3 groups is shown in Figure 2 (for statistical analysis, see Study Limitations). In the presence of low statistical power because of the limited number of events, a previous history of aborted sudden cardiac death and syncope were predictors of adverse outcome (P=0.006 compared with asymptomatic individuals). The period from diagnosis to arrhythmic event ranged from 2 to 36 months (mean, 10±11 months). The only asymptomatic individual who became symptomatic during follow-up was a 45-year-old male (age at diagnosis) who was found to have a characteristic type 1 ECG during a routine examination. No SCN5A mutation was found. During programmed electrical stimulation, ventricular fibrillation was inducible with 2 extrastimuli (basic cycle length, 600 ms; S1, 300; S2, 240 ms). Two months after ICD implantation, ventricular fibrillation occurred that was terminated by the device. All patients except one (89%) who experienced an appropriate ICD intervention during follow-up had a spontaneous type 1 ECG. All 9 patients who had experienced an arrhythmic event during follow-up had undergone programmed electrical stimulation, and 5 patients were inducible (56%; 2 with 1 and 3 with 2 extrastimuli, respectively). In the presence of 9 arrhythmic events during follow-up, the predictive values of programmed electrical stimulation were low. Using up to 3 ventricular extrastimuli, we found a positive predictive value of only 5.4% (negative predictive value, 95.7%, with a low sensitivity and specificity of 55.6% and 50.3%, respectively). If programmed ventricular stimulation had been stopped at a maximum of 2 ventricular extrastimuli, 17 individuals would not have been inducible. This would have resulted in positive and negative predictive values of 6.6% and 96.4%, respectively (sensitivity, 55.6%; specificity, 59.9%; see Study Limitations). In 3 of the 9 patients with arrhythmic events during follow-up, the coupling interval of the extrastimulus required to induce a sustained ventricular tachycardia or ventricular fibrillation was <200 ms. Only 1 patient had a family history of sudden cardiac death, but 5 patients (55%) had other family members with Brugada syndrome. In only 1 patient was a SCN5A mutation found.
A difference in the cardiac arrest–free survival curves (Figure 3) was observed when patients with ST-segment elevation at baseline (regardless of clinical status) were compared with patients without a spontaneous type 1 ECG (log-rank test, P=0.046; see Study Limitations). All asymptomatic individuals (n=53) and sudden death patients (n=9) with a transient type 1 ECG pattern were free of events during a mean follow-up of 32±25 and 78±26 months, respectively. Only 1 of 40 patients with a prior syncope and a type 1 ECG unmasked by class I drug administration demonstrated ventricular fibrillation during follow-up. In the presence of only 9 arrhythmic events during follow-up, gender, the presence of an SCN5A mutation, induction of sustained ventricular tachyarrhythmias, and a combination of any of these variables were not predictors of outcome.
The aim of this study was to evaluate the long-term prognosis and to assess the value of clinical and electrophysiological parameters for risk stratification in individuals with a type 1 Brugada ECG. The study included 212 patients from 165 families. It should be discussed in view of the recent reports by Brugada et al2,9 and Priori et al3 (Table 2), who presented conflicting data with regard to prognosis and risk stratification in patients with Brugada syndrome.
Predictors of Outcome
During follow-up, we identified only 9 of 212 patients who developed a new or first arrhythmic event. Despite longer follow-up, we found a much lower incidence compared with the data reported by Brugada et al.9 In our study, only 4 of 24 patients (17%) compared with 44 of 71 (62%) identified after prior cardiac arrest had another arrhythmic event during follow-up. Brugada et al9 reported subsequent arrhythmic events in 19% of patients with prior syncope (14 of 73) and 8% of the asymptomatic individuals (16 of 190) compared with 6% (4 of 65) and 0.8% (1 of 123), respectively, in the present cohort (Table 2). This difference is most likely related to baseline differences in the 2 patient populations. The worse outcome of patients in the “Brugada” registry compared with our data or the “Priori” registry most likely indicates that in particular families with a very severe form of the disease are reported to the “Brugada” registry, whereas the participating clinics of the present study included all patients who presented with a type 1 ECG over several years. The number of asymptomatic individuals with a family history of sudden cardiac death reported by Brugada et al9 was much higher than in our database [131 (72%) versus 39 (32%)] or in the report by Priori et al3 (20%). Thus, a selection bias of more severely affected families may have significantly contributed to the fact that the outcome of patients reported by Priori et al3 and our own patients was much better than that of the Brugada registry.
We found a clear trend that previous syncope and a spontaneous type 1 ECG indicate a worse prognosis compared with only a transient abnormal ECG. This is in line with observations by Priori et al,3 who identified a spontaneous pattern of Brugada ECG combined with a history of syncope as a predictor of sudden cardiac death. Because of the ongoing discussion of the characteristics of a typical Brugada ECG, only patients with the so-called type 1 ECG pattern5 occurring either spontaneously or after class I antiarrhythmic drug provocation were included in the present study. We excluded all other patients, eg, those with (only) saddle-type ECG changes not changing to a type 1 pattern after drug testing with a class I agent. Of note, both Priori et al3 and our registry included fewer cases with a spontaneous type 1 ECG pattern (51% and 59%, respectively) compared with Brugada et al (71%).9
Our duration of follow-up was longer compared with the data reported by Brugada et al9 (Table 2). Because cumulative risk increases with longer follow-up periods, one might expect even more marked differences between the 2 patient populations during identical follow-up periods. More recently, Brugada et al2 reported on 547 individuals with the ECG pattern of Brugada syndrome and no previous cardiac arrest. Because all syncope and asymptomatic patients had been included in the previous report,9 it is not surprising that most of these patients also had a family history of sudden cardiac death (55%), which may again support the hypothesis of a selection bias toward more severely affected families. During a mean follow-up of 24±33 months compared with 34±52 months in the present study (Table 2), 45 individuals (8.2%) experienced sudden cardiac death or ventricular fibrillation. The probability of a serious arrhythmic event (sudden death or documented ventricular fibrillation) during a 2-year follow-up ranged between 0.5% for asymptomatic individuals with transient ECG abnormalities and 27% in patients with syncope and a spontaneously abnormal ECG. These very high event rates are again in marked contrast to the low event rates reported by Priori et al3 and in our database during a longer mean follow-up period.
Treatment Strategy in Individuals With A Brugada ECG
Consensus exists on the indication of an ICD in cardiac arrest survivors.10 This consensus is supported by our observation that a previous cardiac arrest is a strong indicator of adverse outcome, yet treatment of most asymptomatic individuals with a Brugada ECG or patients with syncope remains less clear. Following the recommendations of Brugada et al,2 implantation of a defibrillator in patients with a previous syncope and a type 1 ECG “seems to be as mandatory as it is in patients with resuscitated sudden arrhythmic death.” Very recently, Belhassen et al11 reported that quinidine suppresses spontaneous arrhythmias in patients with Brugada syndrome and thus may be a safe alternative to ICD therapy in a substantial portion of patients. In patients with syncope, event rates during a relatively long follow-up period were remarkably low, although we included only individuals with a type 1 Brugada ECG. However, in the presence of low statistical power (see Study Limitations), our data also showed a clear trend that patients with syncope had more arrhythmic events than asymptomatic individuals did during a mean follow-up of ≈3 years. Hence, we recommend, on the basis of currently available data, that symptomatic patients (aborted sudden cardiac death or syncope) with the Brugada pattern have an ICD. In asymptomatic individuals, the occurrence of arrhythmic events was strikingly different compared with the Brugada registry data (0.8% versus 8%).9 Therefore, our data question the widespread use of defibrillators in patients with a Brugada ECG but no previous cardiac arrest or syncope. It remains unclear what risk of sudden death is acceptable in a young person. Hence, aggressive management of these individuals may be justifiable on the basis of the lethality of the events but has to be weighed against the economic costs and the costs in terms of quality of life, including the risk for inappropriate shocks.
Conflicting data exist on the prognostic role of programmed electrical stimulation in Brugada syndrome.4,12,13 The present data support the view that electrophysiological testing is not helpful in identifying individuals at risk of life-threatening arrhythmias. This may not be surprising in a disease with intermittent changes of repolarization and thus electrical instability.4 We assessed the value of inducibility and confirmed previous reports of limited accuracy.3,4 We observed a clear trend that programmed electrical stimulation has a low positive predictive value but a high negative predictive value. Remarkably, all asymptomatic, noninducible individuals remained asymptomatic during follow-up. On the other hand, programmed electrical stimulation with a maximum of 3 extrastimuli “missed” 4 of 9 patients with arrhythmic events during follow-up. Thus, the difficult choice to implant or not to implant an ICD in an asymptomatic individual has to be made independently of the results of programmed electrical stimulation. Because some patients with a Brugada ECG pattern complain about palpitations and several groups have reported on supraventricular tachyarrhythmias in Brugada syndrome,14,15 invasive electrophysiological testing still plays a role in selected patients. However, according to our data, all asymptomatic individuals, particularly those with only transient type 1 ECG abnormalities, are at very low risk of sudden cardiac arrest. Further investigation, including programmed electrical stimulation or drug testing to unmask the characteristic type 1 ECG in those with transient ECG abnormalities, does not seem to be of additional value for risk stratification. These individuals, their families, and their physicians should be taught about the conflicting data on outcome in the existing registries. In addition, data on the various worldwide registries should be combined. Longer follow-up periods are necessary before definite conclusions can be drawn on the long-term prognosis of individuals with a Brugada ECG pattern. As Brugada et al2 pointed out, it may be of particular value to monitor low-risk individuals and carefully to avoid potential triggers of ventricular tachyarrhythmias such as fever or proarrhythmic drug effects.
The present study has limitations that should be acknowledged. Although the mean follow-up of 40±50 months was longer than that in recent reports by Brugada et al2,9 and Priori et al,3 it clearly is still too short for final conclusions regarding the management of individuals with a Brugada ECG. In addition, our follow-up durations were variable, with many patients having a relatively short follow-up period. It is conceivable that prognosis might worsen as follow-up prolongs. Only with a larger number of patients and longer follow-up we will be able to clarify the long-term prognosis of individuals with assumed Brugada syndrome. In the meantime, however, it is very important to collect data that indicate that at least the “short-term” prognoses in the existing registries differ markedly.
Because of the low number of events during follow-up, the statistical analysis of the survival times suffers from low power. The given probability values correspond to the individual comparison. Because of the low number of patients with arrhythmic events during follow-up, no correction for multiple testing was performed. To detect a survival difference of ≈5% at a baseline of 90%, we would have had to include >700 patients with ≈50 events during follow-up (power, 80%; α=5%). Thus, conclusions on predictors of outcome cannot be drawn from the present data, but the observed differences indicate important trends that are of clinical relevance. In addition, the low number of events clearly indicates that asymptomatic individuals are at very low risk for sudden death.
Programmed electrical stimulation was performed in only 89% of the included individuals. Because of the low number of arrhythmic events in this large population, we cannot conclude that the predictive role of programmed electrical stimulation remains that low during a longer follow-up period.
This study reports data on the second largest population of individuals with a Brugada ECG pattern with the longest follow-up reported so far. A low incidence of severe arrhythmic events was found during the first years after the ECG abnormality was diagnosed and/or the first clinical event. In the presence of low statistical power, previous histories of aborted sudden cardiac death and syncope were predictors for adverse outcome. Programmed electrical stimulation was found to have limited accuracy in predicting arrhythmic events.
This work was supported in part by grants from Deutsche Forschungsgemeinschaft (Bonn, Germany), Sonderforschungsbereich 556 (projects C1 and C4), and Schu 1082/3-1 Fondation Leducq (Paris, France). This work was supported by grant 902-16-193 from ZonMW (the Netherlands Organization for Scientific Research), The Hague, the Netherlands, PHRC 2001 no du CHU de Nantes, France.
↵*The first 3 authors contributed equally to this study.
Brugada J, Brugada R, Brugada P. Determinants of sudden cardiac death in individuals with the electrocardiographic pattern of Brugada syndrome and no previous cardiac arrest. Circulation. 2003; 108: 3092–3096.
Priori SG, Napolitano C, Gasparini M, Pappone C, Della Bella P, Giordano U, Bloise R, Giustetto C, De Nardis R, Grillo M, Ronchetti E, Faggiano G, Nastoli J. Natural history of Brugada syndrome: insights for risk stratification and management. Circulation. 2002; 105: 1342–1347.
Priori SG, Napolitano C, Gasparini M, Pappone C, Della Bella P, Brignole M, Giordano U, Giovannini T, Menozzi C, Bloise R, Crotti L, Terreni L, Schwartz PJ. Clinical and genetic heterogeneity of right bundle branch block and ST-segment elevation syndrome: a prospective evaluation of 52 families. Circulation. 2000; 102: 2509–2515.
Wilde AA, Antzelevitch C, Borggrefe M, Brugada J, Brugada R, Brugada P, Corrado D, Hauer RN, Kass RS, Nademanee K, Priori SG, Towbin JA, for the Study Group on the Molecular Basis of Arrhythmias of the European Society of Cardiology. Proposed diagnostic criteria for the Brugada syndrome: consensus report. Circulation. 2002; 106: 2514–2519.
Chen Q, Kirsch GE, Zhang D, Brugada R, Brugada J, Brugada P, Potenza D, Moya A, Borggrefe M, Breithardt G, Ortiz-Lopez R, Wang Z, Antzelevitch C, O’Brien RE, Schulze-Bahr E, Keating MT, Towbin JA, Wang Q. Genetic basis and molecular mechanism for idiopathic ventricular fibrillation. Nature. 1998; 392: 293–296.
Smits J, Eckardt L, Le Marec H, Bezzina CR, Schott JJ, Remme CA, Haverkamp W, Breithardt G, Escande D, Schulze-Bahr E, LeMarec H, Wilde AA. Genotype-phenotype relationship in Brugada syndrome: electrocardiographic features differentiate SCN5A-related patients from non SCN5A-related patients. J Am Coll Cardiol. 2002; 40: 350–356.
Brugada J, Brugada R, Antzelevitch C, Towbin J, Nademanee K, Brugada P. Long-term follow-up of individuals with the electrocardiographic pattern of right bundle-branch block and ST-segment elevation in precordial leads V1 to V3. Circulation. 2002; 105: 73–78.
Remme CA, Wever EF, Wilde AA, Derksen R, Hauer RN. Diagnosis and long-term follow-up of the Brugada syndrome in patients with idiopathic ventricular fibrillation. Eur Heart J. 2001; 22: 400–409.
Belhassen B, Glick A, Viskin S. Efficacy of quinidine in high-risk patients with Brugada syndrome. Circulation. 2004; 110: 1731–1737.
Eckardt L, Kirchhof P, Schulze-Bahr E, Rolf S, Ribbing M, Loh P, Bruns HJ, Witte A, Milberg P, Borggrefe M, Breithardt G, Wichter T, Haverkamp W. Electrophysiologic investigation in Brugada syndrome: yield of programmed ventricular stimulation at two ventricular sites with up to three premature beats. Eur Heart J. 2002; 23: 1394–1401.
Morita H, Kusano-Fukushima K, Nagase S, Fujimoto Y, Hisamatsu K, Fujio H, Haraoka K, Kobayashi M, Morita ST, Nakamura K, Emori T, Matsubara H, Hina K, Kita T, Fukatani M, Ohe T. Atrial fibrillation and atrial vulnerability in patients with Brugada syndrome. J Am Coll Cardiol. 2002; 40: 1437–1444.