DISCUSSION
In a preliminary Brazilian long-term follow-up cohort of 138 Brugada syndrome patients, we did not find any variable with high sensitivity, specificity and accuracy to predict severe symptoms. Association with life threatening event might be underpowered due to the low risk profile of our population. This was the first Brazilian registry of long-term follow-up of patients with either a spontaneous or drug-induced type-1 Brugada ECG pattern. The overall characteristics of our middle-aged men population were similar to the usual presentation of BrS in the literature.17, 18 It is expected that proband, the first family member to look for medical assistance, might be more prevalent in the symptomatic group, which was demonstrated in our study and by other authors.19
Precise definition of spontaneous type-1 ECG pattern is somewhat dubious and has changed overtime. Curiously, in our study we demonstrated that symptoms were associated with the spontaneous pattern only if expressed in standard leads (p=0.004). In our registry, there is a high frequency of spontaneous type-1 Brugada ECG pattern detected only when using HPL ECG technique (~70%), which is a maneuver to improve sensitivity, widely used since 2005. Despite a previous clinical comparison between with type-1 BrS detection in conventionalversus HPL ECG showed similar risk profile,20Curcio et al observed a decreased risk of arrhythmic events when type-1 Brugada pattern is defined in the HPL.21 An important limitation to this analysis is defining the real high-risk group, as we intended to refine in our methods. Traditional international registries have pointed out the symptomatic group as at higher risk, but syncope interpretation is challenging as well.22 Therefore, grouping these patients only with the category of “severe syncope” might support the high-risk profile of the group.
Many efforts have been conducted to define non-invasive and invasive markers of risk, and there is a default of reproducibility in large scale. There is some evidence that RVOTcd is the main mechanism underlying ventricular arrhythmia in BrS and this conduction abnormality can be expressed in the 12-lead ECG as AVR sign, deep S wave in lead I and QRS-f. Several reports showed its association with arrhythmic events,9,11,23 however when we analyzed these ECG parameters we found no association with symptoms.
Early repolarization in inferior leads and QRS-f had a low prevalence in our population, precluding association analysis. Even in the multicenter Prelude Study, the prevalence of QRS-f was low, although a possible marker of risk (~8%).10 Differently, Morita et al pointed out QRS fragmentation as a risk factor for VF in both symptomatic and asymptomatic patients, while early repolarization was an important marker of VF risk recurrence. In this cohort they found a high prevalence of this ECG finding (~50%).24
Of note, the natural history of patients with BrS has changed over time.4 Active familiar screening and improvements in electrocardiographic recognition may contribute to the growing number of low risk patients. These findings could partially explain the low annual event rate of our long-term registry as compared to other groups.10,18
Patients in our study who had SCD as the first event totalized 6.5% of the overall population. Even in this group, recurrence of LAE during follow-up was not as remarkable (22.2%) as previously described in long-term follow-up (nearly 50%).22 The increased medical diagnosis, and the progressive improvement in survival might also be consequence of the recently introduced lifestyle change support, regularly standardized in our Institution (awareness of drugs to be avoided, exercise orientation, prompt treatment of fever, and regular visits to reinforce educational measures).1 Comparison of our profile with larger series is summarized in Table 5.
Few symptomatic patients underwent EPS at our institution and nevertheless, our rate of inducible VF/VT was similar to the Finger Brugada registry findings, when comparing symptomatic and asymptomatic patients (46 vs 37%, p = 0.02).18 Heterogeneity in the EPS protocol among registries, use of anesthetic drugs or sodium channel blockers during procedure could be a confounding factor to a powerful analysis. Lack of inducibility at EPS could be considered a predictor of low risk in asymptomatic subjects. In our cohort, negative predictive value in asymptomatic patients achieved 100%, similar to the observations described by Sieira et al (98.3%).25However, if we combine the groups with and without symptoms, the negative predictive value drops to 91.8% in our casuistic and is as high as 97.5% in the FINGER registry.18 On the other hand, the positive predictive value was low (11.5%), in agreement with other larger series.25 PRELUDE was a prospective multicenter study with homogenous enrolment criteria and has shown no predictive accuracy of sustained VT/VF inducibility by the EPS.10 The low event rate in this elegant publication also limited the accuracy and predictive value.
We also analyzed some aspects of the genotype-phenotype association. The majority of pathogenic variants reported in BrS were located in SCN5A locus, a gene that encodes the α subunit of the cardiac sodium channel, which accounts for less than 30% of clinically diagnosed BrS patients. Although we understand that this prevalence might be reviewed after the emerging concepts toward variants classification,26 we found pathogenic or likely pathogenic SCN5A variants in 22% of our patients, according to the ACMG criteria.16 Despite recent controversies, SCN5A is still the main gene associated with BrS.26 Studies were not elucidative to identify new variants in minor genes.27
SCN5A mutations in BrS cause loss of function of the sodium channel, resulting in delayed conduction. Slowed and discontinuous conduction has been found in computer models simulating decreased sodium current.28 The aVR sign, deep S-wave in lead 1 and QRS-f might be signals of conduction delay in the right ventricular anterior wall and these aspects were more frequent in SCN5A carriers in our registry.13, 29
Besides, patients with SCN5A variants showed a longer HV interval; PR and QRS also tended to be longer, but they were not statistically significant. These findings are also in line with data from other previous reports.29 We could not establish the association of SCN5A variants and symptoms.
Classical registries did not explore the association of SCN5A variants with these conduction delay markers, somehow related to risk. One group showed a higher aVR sign in subjects with H558R polymorphism in SCN5A.30 However, H558R polymorphism was once related to better outcomes.31