A score model to predict risk of events in patients with Brugada Syndrome

Abstract

Aims

Risk stratification in Brugada Syndrome (BS) remains challenging. Arrhythmic events can occur life-long and studies with long follow-ups are sparse. The aim of our study was to investigate long-term prognosis and risk stratification of BS patients.

Methods and results

A single centre consecutive cohort of 400 BS patients was included and analysed. Mean age was 41.1 years, 78 patients (19.5%) had a spontaneous type I electrocardiogram (ECG). Clinical presentation was aborted sudden cardiac death (SCD) in 20 patients (5.0%), syncope in 111 (27.8%) and asymptomatic in 269 (67.3%). Familial antecedents of SCD were found in 184 individuals (46.0%), in 31 (7.8%) occurred in first-degree relatives younger than 35 years. An implantable cardioverter defibrillator (ICD) was placed in 176 (44.0%). During a mean follow-up of 80.7 months, 34 arrhythmic events occurred (event rate: 1.4% year). Variables significantly associated to events were: presentation as aborted SCD (Hazard risk [HR] 20.0), syncope (HR 3.7), spontaneous type I (HR 2.7), male gender (HR 2.7), early SCD in first-degree relatives (HR 2.9), SND (HR 5.0), inducible VA (HR 4.7) and proband status (HR 2.1). A score including ECG pattern, early familial SCD antecedents, inducible electrophysiological study, presentation as syncope or as aborted SCD and SND had a predictive performance of 0.82. A score greater than 2 conferred a 5-year event probability of 9.2%.

Conclusions

BS patients remain at risk many years after diagnosis. Early SCD in first-degree relatives and SND are risk factors for arrhythmic events. A simple risk score might help in the stratification and management of BS patients.

 Brugada syndrome , Clinical characterization , Risk stratification , Prognosis , Sudden cardiac death

Introduction

Brugada syndrome (BS) is an inherited disease characterized by coved-type ST-segment elevation in the right precordial leads and increased risk of sudden cardiac death (SCD) in the absence of structural heart disease.¹

Since the first description of the syndrome, more than 20 years ago, the search of predictors to identify those patients at risk of arrhythmic events has been ongoing. As arrhythmic risk persists life long, studies with a very long follow-up are of upmost importance. Nevertheless, mean follow-up in the available literature is usually shorter than 3 years.^{2  ,  3}

Many risk factors have been described, some consistently reported, such as type I electrocardiogram (ECG) pattern or symptomatic patients,^{2  ,  4} whilst others remain under deep controversy, specifically inducible ventricular arrhythmias (VA) during electrophysiological study (EPS).^{3  ,  5  ,  6} Furthermore, implantable cardioverter defibrillator (ICD) is the most accepted therapy for patients at high risk.^{4  ,  7} Implantable cardioverter defibrillator implantation is recommended in those BS patients that have suffered an aborted SCD or have a spontaneous sustained ventricular tachycardia (VT), when presenting syncope and spontaneous type I ECG and it might be considered when VA are induced during EPS.⁸ As patients are otherwise healthy individuals at their forties, risk of device related complications and inappropriate shocks are relevant.⁹ Due to all these considerations, decision making in BS syndrome is still challenging.

The purpose of this study was to analyse our single-centre 20 years’ experience in BS patients, focusing on the search of arrhythmic events predictors. We sought to construct a comprehensive and simple risk score to help in the management of patients.

Methods

Study population

Since 1992, all consecutive patients diagnosed with BS in the Universitair Ziekenhuis Brussel have been included in a registry and followed in a prospective fashion. Patients were included until end 2013. The ethics committee approved the study protocol. Inclusion criteria were: spontaneous or drug-induced Brugada type I ECG and follow-up longer than 1 year. Patients followed at out centre were used for the analysis and creation of a score and those patients not followed in our centre were used as an external validation cohort. Medical history, physical examination and baseline ECG were obtained and underlying structural cardiac abnormalities were excluded. Electrocardiograms were classified as Brugada type I, type II, or normal. An ECG was considered diagnostic if a coved type ST elevation ≥2 mm was documented in ≥1 lead from V1 to V3 in the presence or absence of a sodium-channel blocker. Electrocardiograms were performed with the leads positioned at the sternal margin of the third and fourth intercostal space. Two independent electrophysiologists analysed all ECGs; in case of disagreement a third was consulted. A ‘proband’ was defined as the first Brugada patient diagnosed in a family based on a type 1 Brugada ECG pattern. Patients presenting an aborted SCD or a syncope were considered as symptomatic. Syncope was considered if suspected to have arrhythmic origin. Syncope workup included transthoracic echocardiography and Holter monitoring. It was considered arrhythmic in the absence of prodromes and triggering circumstances and in the presence of brief loss of consciousness or severe trauma. Specific attention was paid to exclude vasovagal syncope. Diagnosis of sinus node dysfunction (SND) was based on the correlation of symptoms with the presence of documented arrhythmias such as sinus bradycardia, sinus arrest, paroxysmal supraventricular tachycardia alternating with periods of bradycardia, or even asystole. Chronotropic incompetence, characterized by an impaired heart rate response to exercise, was considered as an additional manifestation of SND and was defined as failure to achieve 85% of the age-predicted maximum heart rate. Conventional 24- or 48-h Holter monitoring was performed in any case of suspected SND. Moreover, electrophysiological evaluation of SND was performed in all patients who underwent an EPS. Electrophysiological evaluation of SND included measurement of sinus node recovery time (SNRT) (considered abnormal if more than 1500 ms), corrected SNRT (more than 550 ms), SNRT/sinus cycle length 100% (more than 160%), and sinoatrial conduction time (more than 125 ms). Supraventricular arrhythmia as a cause of syncope was also tested during EPS if appropriate.

Drug challenge

Ajmaline (1 mg/kg) was administered intravenously over a 5 min period to unmask the diagnostic pattern in case of non-diagnostic baseline ECG. Flecainide (2 mg/kg) or procainamide (10 mg/kg) were administered over 10 min as an alternative. The test was considered positive if type I ECG was documented in ≥1 right precordial leads. The drug infusion was discontinued if QRS prolongation exceeded 30%, frequent premature ventricular beats or type I Brugada ECG occurred or development of high-degree artrio-ventricular block.

Electrophysiological study

Electrophysiological study protocol used a single site of stimulation (right ventricular apex), three basic pacing cycles (600, 500, and 430 ms), and introduction of up to three ventricular premature beats down to a minimum of 200 ms. It was considered inducible if a ventricular fibrillation (VF) or ventricular tachycardia (VT) lasting at least 30 s was induced.

Implantable cardioverter defibrillator implantation

Beginning from 2005, the indication to ICD therapy was determined using the recommendations of the second Brugada consensus conference.¹⁰ Implantable cardioverter defibrillator programming at the time of implantation changed over the time. Ventricular fibrillation detection rate was increased from 180 to more than 200 b.p.m. for primary prevention implants,¹¹ long-detection intervals were adopted in order to avoid unnecessary therapies.¹² Conversely, in patients receiving ICD therapy for secondary prevention, a monitor zone (> 150 b.p.m.) and fast VT zone (180–200 b.p.m.) with ATP and shocks were programmed and supraventricular discriminators activated if available.

Follow-up

Clinical follow-up of patients was performed at least every 6 months in case of device therapy patients and every 2 years otherwise. All patients were contacted via telephone and post mail in 2007 and 2012. We keep a specific database managed by a dedicated research nurse (G.P.). Arrhythmic events refer to SCD and appropriate ICD shocks. Deaths secondary to other causes were censored. Implantable cardioverter defibrillator implantation not due to a resuscitated SCD was not considered as an event. All patients with an event within the first year had a follow-up longer than 1 year. All available electrograms of shocks were analysed by at least two investigators independently. Electrical storm was defined by three or more sustained VAs or ICD appropriate shocks within 24 h.

Statistical analysis

All data and analysis presented in this study, were performed in the training cohort. Data are presented as mean ± standard deviation or as absolute values and percentages where appropriate. Comparisons between continuous variables were performed using the unpaired Student’s t-test or Mann-Whitney U test. The χ² test or the Fisher’s exact test were used to compare categorical variables. Event-free survival was estimated by Kaplan–Meier method and compared by log-rank test. Time to first event was modelled. Hazard ratios were calculated using Cox proportional hazards models. Variables selected for the analysis were those that had previously demonstrated a predictive value in the BS literature and those that we thought that had biological plausibility to predict events. A predictive model containing all variables related with events in the univariable analysis was constructed. Predictive performance was evaluated by means of Heagarty’s C.¹³ Among the reduced models, the simplest one with a predictive performance loss of less than 10% was selected as the final model. The selected model was validated in the external cohort by means of Heagarty’s C. A P-value less than 0.05 was considered statistically significant. Statistical analyses were conducted using the SPSS software (v22, Chicago, IL, USA).

Results

Study population

Four hundred consecutive patients with BS fulfilled the inclusion criteria (mean age 41.1 ± 17.8 years, range: 6 months to 80 years). Table  1 shows baseline characteristics and evolution along the years.

Table 1

Baseline clinical characteristics of study population and according to diagnosis period

	Overall  (n = 400)	Group 1  (n= 193)	Group 2  (n = 207)	P-value
Age, years	41.1  ± 17.8	41.2  ± 16.6	40.6  ± 18.9	0.58
Age < 18 years  (%)	50  (12.5)	21  (10.9)	29  (14.0)	0.19
Male sex, n  (%)	233  (58.3)	117  (60.6)	116  (56.0)	0.35
Spontaneous type I, n  (%)	78  (19.5)	57   (29.5)	21  (10.1)	<0.01
SCD, n  (%)	20  (5.0)	11  (5.7)	9  (4.3)	0.54
Syncope, n  (%)	111  (27.8)	41  (21.2)	70  (33.8)	<0.01
Asymptomatic, n  (%)	269  (67.3)	141  (73.1)	128  (61.8)	0.02
Proband, n  (%)	122  (30.5)	32  (16.6)	90  (43.5)	<0.01
Family history of SCD, n  (%)	184  (46.0)	102  (52.8)	82  (39.6)	<0.01
Family history of SCD <35 year, n  (%)	31  (7.8)	13  (6.7)	18  (8.7)	0.46
Previous AF, n  (%)	34  (8.5)	23  (11.9)	11  (5.3)	0.02
Previous SND, n  (%)	8  (2.0)	7  (3.6)	1  (0.5)	0.03
PR, ms	173.6  ± 32.6	173.6  ± 33.0	173.4  ± 31.5	0.97
QRS, ms	99.9  ± 20.5	102.6  ± 18.4	90.8  ± 24.6	<0.01
HV, ms	45.9  ± 9.8	46.1  ± 9.6	44.9  ± 10.9	0.53
PES induciblea	72  (19.7)	51  (28.8)	21  (11.2)	<0.01
ICD implantation, n  (%)	176  (44.0)	80  (41.5)	96  (46.4)	0.32

Group I refers to patients with diagnosis made before 2005 and Group II refers to patients with diagnosis during or after 2005.

HV, HV interval; PR, PR interval; PES, programmed electrical stimulation; SCD, sudden cardiac death; AF, atrial fibrillation; SND, sinus node dysfunction; ICD, implantable cardioverter defibrillator.

a

Percentages calculated only among patients who underwent the test.

Seventy-eight (19.5%) patients presented a spontaneous type I ECG, the remaining 322 (80.5%) were diagnosed by means of drug challenge (303 (91.1%) ajmaline, 6 (1.9%) procainamide and 13 (4.0%) flecainide).

Aborted SCD was the clinical presentation in 20 (5.0%) patients, syncope in 111 (27.8%) and the remaining 269 (67.3%) were asymptomatic. Thirty-four (8.5%) patients had history of atrial fibrillation (AF) and 8 (2.0%) presented SND. Table  2 shows baseline characteristics according to symptoms at presentation.

Table 2

Baseline clinical characteristics of study population and according to symptoms at diagnosis

	SCD  (n = 20)	Syncope  (n = 111)	Asymptomatic  (n = 269)	P-value
Age, years	37.9  ± 18.3	42.7  ± 17.6	40.7  ± 17.8	0.47
Male sex, n  (%)	14  (70.0)	68  (61.3)	151  (56.1)	0.36
Spontaneous type I, n  (%)	8  (40.0)	26  (23.4)	44  (16.4)	0.02
Proband, n  (%)	10  (50.0)	56  (50.5)	56  (20.8)	<0.01
Family history of SCD, n  (%)	6  (30.0)	48  (43.2)	130  (39.6)	0.23
Family history of SCD <35 year, n  (%)	4  (20.0)	6  (5.4)	21  (7.8)	0.08
Previous AF, n  (%)	3  (15.0)	11  (9.9)	20  (7.4)	0.41
Previous SND, n  (%)	1  (5.0)	4  (3.6)	3  (1.1)	0.18
PR, ms	186.1  ± 44.3	181.9   ± 30.1	169.8  ± 31.8	0.02
QRS, ms	111.4  ± 32.8	104.3  ± 21.1	97.5  ± 18.6	<0.01
HV, ms	48.0  ± 9.8	46.6  ± 11.5	45.5  ± 9.3	0.72
PES induciblea	4  (20.0)	36  (32.4)	32  (11.9)	<0.01
ICD implantation, n  (%)	19  (95.0)	99  (89.2)	58  (21.6)	<0.01

SCD, sudden cardiac death; PES, programmed electrical stimulation; AF, atrial fibrillation; SND, sinus node dysfunction; ICD, implantable cardioverter defibrillator.

a

Percentages calculated only among patients who underwent the test.

 

Table 3

Clinical characteristics of patients with sinus node dysfunction

	Gender	Agea	Proband	Early Family History SCDb	Spontaneous type I	Symptoms at presentation	Event
1	Male	35	No	No	No	Asymptomatic	No
2	Female	2	No	Yes	Yes	Syncope	Yes
3	Male	2	Yes	Yes	Yes	SCD	Yes
4	Female	26	Yes	No	Yes	Asymptomatic	No
5	Female	71	Yes	No	No	Syncope	No
6	Male	16	No	No	Yes	Syncope	No
7	Male	42	No	No	No	Syncope	No
8	Male	57	No	Yes	Yes	Asymptomatic	Yes

SCD, sudden cardiac death.

a

Age indicates age at diagnosis.

b

Indicates history of SCD in first degree relatives younger than 35 years.

Familial antecedents of SCD were present in 184 (46.0%), in 31 (7.8%) in a first-degree relative younger than 35 years. Electrophysiological study was performed in 365 (91.3%) patients, in 72 (19.7%) a sustained VA was induced. One hundred twenty-two (30.5%) were probands. A total of 215 genetic tests (53.8%) were obtained and 53 (25.5%) were positive for a SCN5A gene mutation.

Management

An ICD was implanted in 176 (44.0%) patients. It was placed in 95.0% (19) of patients presenting an aborted SCD, 89.2% (99) of those with syncope and in 21.6% (58) asymptomatic. Patients receiving an ICD were significantly older (43.5 ± 17.0 vs. 39.5 ± 18.1, P = 0.03), more frequently male (50.6% vs. 34.7%, P < 0.01) and more commonly presented a spontaneous type I pattern (64.1% vs. 39.1%, P < 0.01). Longer PR (179.9 ± 31.3 ms vs. 169.3 ± 32.9 ms, P = 0.01) and broader QRS (106.2 ± 19.0 ms vs. 20.5 ms, P < 0.01) was registered in ICD patients. Probands underwent more frequently ICD implantation (71.3% vs. 32.0%, P < 0.01).

Follow-up

After a mean follow-up of 80.7 ± 57.2 months (range 1.4–298.0), 34 (8.5%) arrhythmic events occurred, accounting for a rate of 1.4% year. Follow-up was available for all patients. Kaplan–Meier event free survival was 96.5% at 1 year, 91.4% at 5 years and 89.3% at 10 and 15 years. Events consisted in 4 (1%) SCD (one successfully reanimated) and 30 (7.5%) VA, appropriately treated by ICD therapies.

Four (1.0%) patients presented an electrical storm during follow-up. All initially presented as aborted SCD, 3 (75.0%) were men, 2 (50.0%) had a spontaneous type I ECG, mean age was 41.0± 10.3 years and 2 (50.0%) had already received an appropriate ICD therapy.

Twelve (3%) patients died during follow. Aetiology was SCD in three patients, ischaemic heart disease in one, septic shock after device revision in one and the remaining seven suffered a non-cardiac death.

Eight patients presented a supraventricular tachycardia during the follow-up: six atrioventricular nodal re-entry tachycardia, one flutter, and one focal atrial tachycardia. Thirty-four patients presented AF. In four of whom coexisted with SND. Eight patients received quinidine during follow-up.

Risk factors associated to events

Symptoms at presentation

Patients initially presenting symptoms had a worse prognosis (HR 5.67, 95% confidence interval [CI] 2.6–12.7, P < 0.01) (Figure  1). Initial presentation as aborted SCD conferred a HR of 20.0 (95% CI 8.1–49.4, P < 0.01) and syncope of 3.7 (95% CI 1.6–8.6, P < 0.01). Arrhythmic event incidence rate was 11.1% year for patients presenting with aborted SCD, 2.2% year for syncope and 0.6% year for those asymptomatic.

Figure 1

Event probability according to Kaplan-Meier method for different risk factors. Number of patients at risk is shown under each graph.

Electrocardiogram pattern

Patients with spontaneous type I ECG were at a higher arrhythmic risk (HR 2.7, 95% CI 1.3–5.4, P < 0.01). Event incidence rate was 2.3% year for the spontaneous type I patients and 1.1% year for those drug-induced.

Gender

Male patients had worse prognosis (HR 2.7, 95% CI 1.2–6.2, P = 0.02). Incidence rate in women was 0.7% year vs. 2.0% year in men.

Family history of SCD

Family history of SCD lacked a statistical relationship with events (HR 0.6, 95% CI 0.3–1.3, P = 0.20). Conversely, when considering SCD in 1st-degree relatives younger than 35 years, it resulted significant, with a HR of 2.9 (95% CI 1.2–7.0, P = 0.02) (Figure  1). Patients with early SCD in first-degree relatives had an event incidence rate of 3.1% year as compared with 1.3% year if it was not present.

Sinus node dysfunction

Those patients presenting concomitant SND had a significantly worse prognosis (HR 5.0, 95% CI 1.5–16.3, P < 0.01) (Figure 1). Incidence rate was 4.3% year in patients with SND and 1.3% year in those without. Table 3 shows clinical characteristics and events in patients presenting SND.

Electrophysiological study

Inducible VA during EPS conferred a significant higher risk for events (HR 4.7, 95% CI 2.2–10.2, P < 0.01) (Figure  1). Patients with inducible arrhythmias presented an incidence rate of 3.3% year as compared with 0.7% year in non-inducible. Amongst those patients diagnosed after the beginning of 2005, inducible VA was significantly related to events (adjusted HR 4.7, 95% CI 2.1–10.4, P < 0.01).

Other risk factors

Proband status and QRS duration (per ms) showed a significant relation with events (HR 2.1, 95% CI 1.0–4.2, P = 0.04 and HR 1.03, 95% CI 1.01–1.04, P < 0.01, respectively). History of AF presented a non-significant borderline association (HR 2.3, 95% CI 0.9–5.6, P = 0.07).

Risk score model

A model with all variables that showed statistical relationship with events was created. The variables included were gender, presentation as syncope, as aborted SCD, SND, proband status, early familial antecedents of SCD in first-degree relatives, spontaneous type I pattern, QRS duration and VA inducibility. This model had a high predictive performance: 0.90.

Amongst the reduced models, the selected one included presentation as syncope, as aborted SCD, spontaneous type I pattern, SND, early familial antecedents of SCD in 1st-degree relatives and inducible VA. Predictive ability of this model was high: 0.82. When applied to asymptomatic patients, it was 0.81, for syncope patients: 0.63 and for patients presenting after and aborted SCD it was 0.71. If applied to asymptomatic or syncope patients it was 0.79.

A risk score for this model was constructed by rounding the regression coefficients. Performance of the score was the same as the model (0.82). Figure  2 shows the score and its performance. Examining the data, we found that patients with a score greater than 2 presented a significant higher event probability as compared with lower scores (P = 0.02). A score of two points conferred an event free survival of 97.4% at 1 year and 90.8 at 5 and 10 years with a sensitivity of 79.4% and a specificity of 72.2%. Table  4 shows clinical characteristics of patients according to the score and Table  5 the event free survival in each category.

Figure 2

Selected risk model to predict events. (A) Score corresponding to each item. SCD refers to presentation as aborted sudden cardiac death. (B) Event probability according to Kaplan–Meier method in each score category.

Table 4

Baseline clinical characteristics and events of study population and according to score punctuation

	0  (192)	1  (40)	2  (78)	3  (25)	4  (n = 36)	≥5  (n = 29)	P-value
Age, years	40.2  ± 18.7	41.0  ± 16.6	41.7  ± 18.1	45.2  ± 13.7	43.2  ± 16.3	40.4  ± 17.4	0.82
Male sex, n  (%)	91  (47.4)	29  (72.5)	47  (60.3)	18  (72.0)	23  (63.9)	25  (86.2)	<0.01
Spontaneous type I, n  (%)	0  (0)	25  (62.5)	3  (3.8)	21  (84.0)	5  (13.9)	24  (82.8)	<0.01
SCD, n  (%)	0  (0)	0  (0)	0  (0)	0  (0)	9  (25.0)	11  (37.9)	<0.01
Syncope, n  (%)	0  (0)	0  (0)	59  (75.6)	10  (40.0)	25  (69.4)	17  (58.6)	<0.01
Asymptomatic, n  (%)	192  (100)	40  (100)	19  (24.4)	15  (60.0)	2  (5.6)	1  (3.4)	<0.01
Proband, n  (%)	36  (18.8)	9  (22.5)	47  (60.3)	9  (36.0)	12  (33.3)	9  (31.0)	<0.01
Family history of SCD, n  (%)	91  (47.4)	19  (47.5)	27  (34.6)	15  (60.0)	17  (47.2)	15  (51.7)	0.25
Family history of SCD <35y, n  (%)	0  (0)	15  (37.5)	3  (3.8)	3  (12.0)	4  (11.1)	6  (20.7)	<0.01
Previous AF, n  (%)	14  (7.3)	1  (2.5)	8  (10.3)	2  (8.0)	4  (11.1)	5  (17.2)	0.34
Previous SND, n  (%)	0  (0)	0  (0)	0  (0)	1  (4.0)	1  (2.8)	6  (20.7)	<0.01
PES induciblea	0  (0)	0  (0)	16  (20.5)	14  (56.0)	23  (63.9)	19  (65.5)	<0.01
ICD implantation, n  (%)	18  (9.4)	7  (17.5)	68  (87.2)	21  (84.0)	36  (100.0)	26  (89.7)	<0.01
Events	3	1	6	4	10	10	<0.01

PES, programmed electrical stimulation; SCD, sudden cardiac death; AF, atrial fibrillation; SND, sinus node dysfunction; ICD, implantable cardioverter defibrillator.

a

Percentages calculated only among patients who underwent the test.

 

Table 5

Event free survival at 1, 5, and 10 years, in each score category

Score	1 year	5 years	10 years	P-value
0	100%	98.4%	97.2%
1	100%	96.4%	96.4%	0.79
2	97.4%	90.8%	90.8%	0.02
3	88.7%	83.4%	83.4%	<0.01
4	91.2%	75.2%	70.1%	<0.01
≥5	79.3%	68.2%	61.4%	<0.01

P-value refers to the log-rank comparison with the reference category.

External validation of the proposed score

A cohort of 150 BS patients was assessed for external validation of the model. The validation cohort presents some differences as compared with the training cohort. Table  6 shows baseline clinical characteristics and event rates of both cohorts. Patients in the validating cohort were older with a mean age of 45.9 ± 13.2 years, male sex was more frequent (83.3%) and patients displayed more the spontaneous type I ECG pattern (52.7%). Symptoms at presentation and VA inducibility were distributed similarly. Consequently, the mean score punctuation was higher (1.89 ± 1.84 vs. 1.44 ± 1.79 points). Event incidence rate was also higher (1.9% per year vs 1.4% per year). The score performance in the validating cohort was 0.81.

Table 6

Clinical characteristics of study population and validation cohort

	Training  (n = 400)	Validation  (n = 150)	P-value
Age, years	41.1  ± 17.8	45.9  ± 13.2	<0.01
Male sex, n  (%)	233  (58.3)	125  (83.3)	<0.01
Spontaneous type I, n  (%)	78  (19.5)	79  (52.7)	<0.01
SCD, n  (%)	20  (5.0)	10  (6.7)	0.53
Syncope, n  (%)	111  (27.8)	32  (21.3)	0.16
Asymptomatic, n  (%)	269  (67.3)	108  (72.0)	0.30
Proband, n  (%)	122  (30.5)	63  (42)	0.01
Family history of SCD, n  (%)	184  (46.0)	78  (52)	0.21
Family history of SCD <35 year, n  (%)	31  (7.8)	19  (12.7)	0.09
Previous AF, n  (%)	34  (8.5)	9  (6.0)	<0.01
Previous SND, n  (%)	8  (2.0)	4  (2.7%)	0.74
PES induciblea	72  (19.7)	36  (25.9)	0.29
ICD implantation, n  (%)	176  (44.0)	54  (36.0)	0.10
Events incidence rate	1.40	1.90	0.42
Score points	1.44  ± 1.79	1.89  ± 1.84	0.01

PES, programmed electrical stimulation; SCD, refers to presentation as aborted sudden cardiac death; AF, atrial fibrillation; SND, sinus node dysfunction; ICD, implantable cardioverter defibrillator.

a

Percentages calculated only among patients who underwent the test.

Discussion

The most challenging issue in the management of BS patients is identifying those at risk of SCD. Some risk factors, such as symptoms or spontaneous type I pattern have been consistently associated to a worse prognosis,^{2  ,  4  ,  7} whilst controversy still exists around others, as the role of EPS.⁵ Furthermore, most studies in BS have a mean follow-up not longer than 3 years and, as arrhythmic risk persists lifelong, studies with very long follow-up are needed.

We present a big cohort of BS patients followed for a mean of 80 months. This constitutes the longest follow-up published heretofore. Amongst the main BS registries, only FINGER involved more patients but our follow-up is almost three times longer.²

The aim of our study was to offer a simple tool to stablish the arrhythmic risk in BS. We confirm the value of some widely accepted risk factors such as spontaneous type I pattern and symptomatic presentation. We present further evidence supporting others, as inducible VAs during EPS, sex or SND. A relation between arrhythmic events and familial history of early SCD was found.¹⁴ A risk score model has been constructed provide a comprehensive approach that might help in the management of BS patients.

Clinical profile

Recent evidence shows a shift in the clinical profile of BS patients.¹⁵ Higher risk patients were described in the early years after the description of the syndrome.

In our study patients had a mean age of 41 years. Male sex was predominant but more than 40% were females, a fact already highlighted in a recent publication.¹⁶ This contrasts with the main BS registries,^{2  ,  3} where women constituted only 20–30% of the population. We believe that this might be due to meticulous familial screening and proactive search in patients with suspicion of BS.

Twenty percent of patients presented spontaneous type I ECG and most patients were asymptomatic. Interestingly 46% of patients had a history of SCD in their family and 8% in a 1st-degree relative younger than 35 years. In the FINGER, registry age of diagnosis and symptomatic status at presentation are similar but around half of the patients presented a spontaneous type I pattern.² The PRELUDE registry presented a similar profile.³ Inducible VA during EPS were found in 20% of our patients. Contrast with the main BS registries is notorious, as inducibility rates are much higher in the former studies. This could be explained by differences in the stimulation protocol, less aggressive in our cohort.

Therefore, clinical profile of the patients present in this study is more favourable than those in the main BS registries, furthermore it has evolved over time. This could be of importance as it might represent better the profile of BS patients diagnosed nowadays.

Long term follow-up and risk factors for arrhythmic events

Arrhythmic event rate is crucial to provide robust management recommendations. Event rate in our study is 1.4% year, similar to FINGER and PRELUDE (1.6% and 1.5%, respectively).^{2  ,  3} Amongst risk factors, familial history of SCD, EPS inducibility and SND merit special consideration.

Relation between familial history of SCD and prognosis has not been consistently reported. Our group published a specific study addressing this issue, finding that multiple SCD in 1st-degree relatives younger than 35-years-old were predictive of future events.¹⁴ In the present study, almost 50% of patients have a family history of SCD but in only 10% in a young first-degree relative. The genetic background of the syndrome makes reasonable to expect a high incidence of SCD in those families with a more severe form of the disease. Nevertheless, all major registries have failed in finding this association.^{2  ,  3} In the present study, early familial antecedents of SCD confers a HR of 2.9 (adjusted of 4.0), a similar magnitude as spontaneous type I.

Another interesting finding is that SND is related to a more severe form of the disease. Sodium channelopathies have already been associated to SND.¹⁷ Patients presenting SND tend to be young, specifically 40% of them have less than 18 years. Despite SND is present in only 2% of patients, it confers a high risk for events (HR 5.0), even greater than syncope.

Ventricular arrhythmia inducibility remains still under deep controversy as a stratifying tool, with contradictory findings.^{3  ,  18} Some recent evidence is highlighting the value of EPS in risk stratification.^19–21 In our experience, VA induced during EPS is associated with a worse prognosis.

Risk model

Risk stratification in BS is complex and many factors should be taken in consideration. Implantable cardioverter defibrillator implantation entails potential long-term complications, therefore risks and benefits should be carefully evaluated. We sought to develop a risk model with enough predictive performance to help in the management of patients. Based on a maximum model we simplified it without a substantial power loss. The selected model has six items including novel markers such as SND and early family history of SCD. This model had a good predictive performance in the training cohort

To add further support of the value of our score, it was tested in a validating cohort of 150 BS patients. Of note, the validation cohort presented some differences as compared with the training one. Briefly, patients in the validating cohort had a higher arrhythmic risk profile: spontaneous type I pattern was more frequent, they were more frequently male and probands. As a reflect of this, arrhythmic events were more frequent. However, the score performed well in the validating cohort, with a similar performance.

Establishing a threshold that classifies patients as high risk is difficult and open to a variety of opinions. An estimated 5-year event risk of 6% seems reasonable as it is used to recommend ICD implantation in patients with hypertrophic cardiomyopathy.⁸ This corresponds to a score of two points in our model. Nevertheless, the decision for an aggressive management should be based in objective data but should be done individually, taking into consideration patients’ preferences and clinicians’ experience.

Limitations

Patients with BS present a lifelong risk of arrhythmias and even longer follow-up is advisable. The overall number of events might limit the search of predictors. This limited number of events might also over fit the score model. Competing risks might be of relevance; nevertheless, their effect is probably small as the population is composed by healthy individuals. Implantable cardioverter defibrillator shocks as a surrogate of SCD in BS population overestimates arrhythmic events,²² however, is a common endpoint in BS registries.^{2  ,  3  ,  6} Validation of the score system in another large cohort would add more support to its utility and confirm the value of the selected risk factors.

Conclusions

The present study, with the longest follow-up published to date, demonstrates that BS patients have an overall event rate of 1.4% year. We add more clinical evidence to widely accepted risk factors, as presentation symptoms and ECG pattern. Inducible VA during EPS and other novel factors such as SND and early familial history of SCD predict events. The proposed risk model has a good predictive value and can help in the management of BS patients.

Conflcit of interest: C.dA. receives compensation for teaching purposes and proctoring from AF solutions, Medtronic, member steering committee ETNA-AF-Europe Daiichi Sankyo Europe. G.B.C. receives compensation for teaching purposes and proctoring from AF solutions, Medtronic. P.B. receives research grants on behalf of the centre from Biotronik, Medtronic, St Jude Medical, Sorin, Boston Scientific and speakers’ fees from Biosense-Webster, Biotronik, Medtronic and Boston Scientific.