Feasibility of Improving Risk Stratification in Brugada Syndrome

It is estimated that approximately 600 apparently fit and healthy individuals aged under 35 years die suddenly in the UK every year Many of these sudden cardiac deaths (SCD) in the young are the result of inherited cardiac conditions (ICC), the majority of which are the direct consequence of single mutations in sarcolemmal ion channels (e.g. Brugada syndrome - BrS, Long QT syndrome - LQTS), intercalated disc proteins (arrhythmogenic right ventricular cardiomyopathy - ARVC) or the cardiac sarcomere (hypertrophic cardiomyopathy - HCM) Poor genotype-phenotype correlations, due in part to incomplete penetrance, variable expressivity, role of gene modifiers and external environmental factors, limits the use of an individual's genetic makeup in the core task of predicting risk of death. As a result, identifying high risk individuals who should receive an implantable cardioverter defibrillators (ICD) is still a significant challenge.

In Brugada Syndrome, follow up data from the largest registries reveal a spontaneous Type I BrS pattern (ST elevation in the anterior leads) on electrocardiogram (ECG) and/or history of syncope to be independent predictors for SCD events. These are employed in conventional risk stratification to identify those with high risk. By this approach, the low risk have an annual SCD rate below 0.5% and the high risk have an annual SCD rate ~1% 3,8,9. Therefore, not only are SCD events in the unselected BrS population relatively low, the differences in event rates between those perceived to be at high and low risk are small, highlighting the limitations of current risk stratification. Thus, calculations of sensitivity and specificity to predict SCD risk using syncope is 61%/52% and for spontaneous type I BrS ECG pattern is 86%/36% 10.

Amongst individuals presenting with SCD, a significant portion of individuals have neither spontaneous Type I BrS pattern or previous syncope, and would have been considered low-risk. Only 50% of participants in the FINGER registry with previous SCD events had spontaneous Type I BrS pattern. Similarly in a study of 50 SCD probands with a familial diagnosis of BrS, only 20% had a history of prior syncope and in those with ante-mortem ECGs, only 20% had a spontaneous type I pattern. This reflects our own experience with a cohort of 149 BrS patients; we have 10 individuals with SCD events, of whom only 1 has a spontaneous Type I BrS pattern and/or a history of syncope.

All these data seem to consistently suggest that the majority of SCD events occur in the larger 'low-risk' cohort, further underlining the limitations of current risk stratification.

The decision to implant an ICD must be weighed against the risk of complications. At present appropriate therapy occurs in around 12%, which is higher than the SCD event rate; this is consistent with normal ICD function 'over-treating' ventricular arrhythmias which would have been non-sustained. Inappropriate shock rates have been reported between 5-37% depending on follow-up time, and death has been reported from inappropriate therapy shock. Lead failure also occurred in 29% of the 378 BrS patients over as 10 year follow up. Implanting more ICDs to compensate for the limitations of risk stratification may only increase morbidity and even mortality from inappropriate therapy and complications.

Understanding the electrophysiological mechanisms behind fatal arrhythmias in Brugada Syndrome may help in developing more objective means of identifying those patients most at risk of SCD. There are currently two hypotheses. The "repolarisation hypothesis" was formulated from explanted canine ventricular experiments. Exposure to sodium channel blockers caused loss of action potential (AP) dome and shortening of AP duration in the right ventricular epicardium but not in endocardium, creating a transmural voltage gradient that would hypothetically form the substrate for re-entry. In contrast, proponents of the "depolarisation hypothesis" postulate that slowed conduction caused by reduced inward sodium current function underlies the arrhythmogenic tendency in BrS. Mapping studies in SCN5A-knockout models have shown that conduction abnormalities contribute towards ventricular arrhythmogenesis. There is also human data to support this hypothesis with regional conduction delay and fractionated electrograms in the RVOT. The major problem with the arrhythmogenic mechanisms proposed above for both BrS is that they do not address the critically important clinical question: why are some individuals with ICCs are more predisposed to SCDs than others with the same condition?

An additional compounding factor is the observation that in ICC patients, SCD is often triggered by sudden changes in autonomic or metabolic status. SCD in Brugada syndrome often occurs at rest when vagal tone is predominant and during episodes of fever. Typical Brugada ECG changes may be unmasked or intensified by vagal stimulation, parasympathomimetic drugs, anti-adrenergic drugs, or α-adrenergic receptor stimulators, and diminished by exercise or isoproterenol infusion. This suggests that studying the response of the electrophysiological substrate to extrinsic stressors may identify individuals at risk of SCD. Indeed, labile or exaggerated electrophysiological response to extrinsic stressors may represent the final common pathway predisposing to SCD in the ICCs, regardless of the specific genotype, syndrome or diagnosis.

To investigate this hypothesis, a BHF project grant was awarded to use non-invasive electrocardiographic imaging (ECGi) to understand the effect of external stressors on ventricular conduction. During PG/15/20/31339, we demonstrated changes in activation recovery interval (ARI) in patients with aborted SCD, but these were only evident at peak exercise. These findings were consistent with mathematical models predicting that heterogeneity of repolarisation and the resulting conduction abnormalities would predispose to fibrillatory activation. This raised the possibility that ARI abnormalities with exercise could be the basis of a risk stratification tool. However ARI measurements are labour intensive, so we developed a novel technique to rapidly identify conduction abnormalities. This system uses a 252-electrode vest and calculates the appearance of the epicardial unipolar electrogram from body surface electrograms and low-resolution CT. Data from this study has shown that patients with previous VT/VF develop heterogeneities in conduction following exertion. The example used non-invasive ECGi to develop a new risk stratification tool called 'Ventricular Conduction Stability Test'. The figure below shows how data from the electrode vest (A) generates body surface potentials (B), reconstructs epicardial electrograms onto a CT-generated torso model (C) then projects these onto a 3D cardiac surface (D). Comparing beats during rest and exercise produces an assessment of how stable conduction remains at peak exercise (E).

The lower panel in (E) shows multiple abnormal areas coloured red/blue at peak exercise in a patient with previous aborted SCD - no cardiac investigations were able to detect an abnormality. In the upper panel (control), conduction remains stable, denoted by white areas. These changes can be used to mathematically generate a V-CoS score. The graphs below are results from 62 patients comparing V-CoS scores between survivors of SCD (idiopathic (iVF), Brugada (BrS), Hypertrophic Cardiomyopathy (HCM) (30pts)) against controls (normal hearts or low risk BrS/HCM by current risk stratification (32 patients).

The Ventricular Conduction Stability Test assigns a V-CoS score of '100' when conduction patterns at peak exercise is the same as at rest, and the median score falls below '95' in survivors of SCD due to >5% of the activation being abnormal at peak exercise. If risk stratification were based on V-CoS score following exercise, we can achieve a sensitivity and specificity of 90% and 73% respectively, which is far better than current risk stratification methods as shown in adjacent table.

The ideal next step would be a prospective study to validate V-CoS Score as a reliable risk stratification tool. Unfortunately, the low incidence of Brugada syndrome in the general population would be challenging for recruitment in a UK only study.

However, the extent of the limitations of current risk models is apparent on the adjacent figure. Patients with Brugada syndrome and aborted SCD were categorised as high/medium/low risk by current techniques and most SCD events occur in patients who would have been considered low risk. The V-CoS score appears to differentiate these groups very well at a cut-off of 95%.

RATIONALE FOR CURRENT STUDY In the absence of any other method to support risk stratification in Brugada syndrome, V-CoS score would be appealing to clinicians. A larger series to confirm a low incidence of 'false negatives' and more detailed exploration of other cardiac pathology that may lead to a 'false positives' is required to support the use of V-CoS scores as a clinical tool.

ORIGINAL HYPOTHESIS V-CoS score is a reproducible and specific method for identifying marked heterogeneities in activation that predispose to VT/VF in Brugada Syndrome.

This will be a retrospective cohort study. It is expected to last for 3 years. There will be a total of 100 subjects.

The hypothesis will be addressed by a series of sub-studies:

i) Patients considered to be at low risk of SCD (controls) will be recruited to confirm that these patients have a V-CoS score >95

In the preliminary study described earlier, the control group were relatively homogenous with SCD being the only differentiating parameter. However, during clinical practice there may be a range of other cardiac abnormalities. Our primary hypothesis would predict that only groups of patients known to be at risk of SCD would have V-CoS score <95. In order to test this, we will study groups of patients expected to have low SCD risk, but may have other cardiac abnormalities that could cause false positives. All patients recruited will be from groups who would normally benefit from investigation to risk stratify for sudden cardiac death, or ECGi.

1. Patients undergoing ablation with ECGi system for other arrhythmias (n=10) -these patients will be similar to our original controls and provide a repeat set of controls.

2. Patients with AF undergoing ablation with ECGi system (n=10). These patients will be older and have varying RR intervals which may cause a falsely low V-CoS.

3. Relatives of Brugada patients with confirmation of no pathology (n=10) - these patients will be closest to 'true normal' and can be a surrogate control group for idiopathic VF families.

4. Out-of-hospital cardiac arrest primary PCI with full recovery of left ventricular function and full revascularisation (n=10) - the purpose of this group is to confirm that the changes detected in our SCD group are not secondary to the SCD event. These are patients who have had a cardiac arrest secondary to coronary occlusion, but have made a full recovery with normal LGE-MRI and no indication for ICD.

5. Athletic Hypertrophy (n=10) - Elite athletes often have physiological LVH and abnormal ECGs at rest. It is unclear if these variations in activation will lead to a decrease in V-CoS.

ii) Patients at high risk of SCD will be recruited to confirm that these patients have a V-CoS score <95

Our primary hypothesis was that abnormalities of conduction should occur in all patients at risk of SCD. Therefore, the SCD group included patients with SCD caused by a range of underlying pathologies. If V-CoS score are to be used for Brugada risk stratification without a prospective study, then we need to confirm that the ROC data holds true for a larger patient series. We will aim to recruit 50 Brugada patients with previous SCD or ICD therapy for VT/VT to validate the preliminary findings. The control group for these will be (a) and (c) from part (i). This increase in the number of patients will increase the power of the ROC data.

iii) Alternative techniques for identifying lowest V-CoS score

In the preliminary study we used ETT and Tilt testing as the external trigger for producing 'arrhythmogenic' changes in the ventricular substrate. We found that the ETT caused more profound changes and so we have focused on this in our study. However, most arrhythmias are triggered by ventricular ectopics. We have not tested the relationship between V-CoS score and prematurity of ventricular ectopy. If lower scores can be generated by ectopics, it could be used as the preferred method of identifying a patient's lowest V-CoS score. In addition, it may also provide another way of risk stratifying individuals who have physical difficulties undertaking an exercise treadmill test.

Subject enrolment: Participants will be identified from Imperial College Healthcare.

Cardiology out-patient clinics and elective catheter scheduling office:- Patients scheduled for electrophysiological catheter studies +/- ablation for atrial and ventricular ectopy, atrio-ventricular nodal re-entry tachycardia and atrial fibrillation. ICHNT currently undertakes roughly 650 procedures for these conditions per year.

Cardiology in-patient wards:- Individuals admitted with out-of-hospital cardiac arrest who have undergone full revascularisation with primary PCI and full recovery of left ventricular function. Hammersmith Hospital at ICHNT provides a tertiary primary PCI service and performs an estimated 40 PCI for OOHVF arrest per year.

Inherited Cardiac Conditions Service:- The service currently has approximately 150 confirmed cases of BrS, 32 with ICD implants, under surveillance and provides a comprehensive screening service for relatives of affected individuals. Unaffected relatives of BrS patients and BrS patients surviving an out-hospital cardiac arrest or who have previously received appropriate ICD therapy will be recruited. We will also be looking to recruit BrS patients with previous SCD events at St George's Healthcare NHS Trust and Barts Health NHS Trust.

Sports teams:- elite athletes will be recruited from sports teams.

Study protocol: As part of a half-day visit to the cardiac investigations unit, patients will undergo non-invasive studies with the ECGi vest. The 252 electrode vest will be applied onto the individual and then undergo a non-contrast thoracic CT scan. This involves a low dose of radiation equivalent to 15 standard chest x-rays or 6 months of natural background radiation.

i) Non-invasive programmed stimulation will be performed in those with ICDs and have been specifically counselled and consented for this procedure.

ii) All patients will undergo Exercise Treadmill Test protocol: A resting baseline recording in the supine position will be performed before undertaking the Bruce protocol with the aim of getting the individual to achieve maximal exercise capacity or reach 100% of their maximum target heart rate for age. On achievement of either of these targets, participants will be placed back in the supine position to minimise artefact noise whilst ECGi recordings are performed post peak exertion and for a 10 minute recovery period.

iii) For patients undergoing EP studies, pacing from RV apex will be performed at twice the diastolic threshold with sensed extras and incremental ventricular pacing to measure Ventricular CoS scores prior to performing ablation