Abstract
Background/Introduction
The founder mutation p.Arg14del in the gene encoding phospholamban (PLN) is a known cause of arrhythmogenic cardiomyopathy (ACM) with distinct clinical features, such as microvoltages on the ECG and right or left ventricular dysfunction or both. At present, risk stratification for ICD implantation in carriers of this mutation is based on left ventricular ejection fraction and previous ventricular arrhythmia's, which are not specific risk factors for the PLN p.Arg14del cardiomyopathy.
Purpose
Our goal is to develop a mutation specific prediction model for incident malignant ventricular arrhythmia to guide ICD implantation.
Methods
Data were collected from p.Arg14del carriers with no history of malignant VA at baseline, identified between September 2009 and June 2018 in three Dutch university hospitals. Genetic analysis of PLN was performed in a clinical setting in index patients with clinical signs of DCM/ACM, or in family members of p.Arg14del carriers. We collected clinical data from the first cardiac evaluation and follow-ups. Malignant VAs were defined as sustained VA, appropriate ICD intervention or (aborted) sudden cardiac death. A prediction model was developed using Cox Proportional Hazard regression. Candidate baseline predictors were pre specified based on literature and clinical expertise. Age, sex, proband status, sudden cardiac death (SCD) in 1st degree relative, repolarization abnormalities, microvoltages, premature ventricular complexes (PVC) burden on 24hrs Holter monitoring, LVEF and non-sustained ventricular arrhythmias (NSVT) were considered. The multivariable model was fitted using stepwise backward selection based on Akaike's Information Criterion.
Results
We included 440 p.Arg14del carriers with a mean age of 41±18 years and 41% males. During a median follow-up of 4.7 years (IQR 1.7–7.3), 44 incident malignant VA occurred, 20 sustained VAs and 24 appropriate ICD therapies. The multivariable HR's of selected predictors were: 2.2 for minor and 4.5 for major repolarization abnormalities vs no abnormalities (p value respectively 0.06 and 0.01), 2.2 for LVEF <45% (p value 0,1) and 7.7 for >500PVC/24hrs (p value 0.003). Carriers with and without events could be accurately distinguished with this model, with an optimism corrected C-statistic of 0.81. The model calibrated well, with agreement of observed and predicted 5 year malignant VA risk. Risk groups were split into quintiles of predicted risk, figure 1 shows the Kaplan Meier curve of incident malignant VA per risk group. The 5 year risk of incident malignant VA in the lowest 3 quintiles was 0%, 9% in the 4th quintile and 25.2% in the highest quintile.
Conclusion
We created a PLN p.Arg14del mutation specific prediction model to estimate risk of incident malignant VA. With this model a clear distinction between high risk and low risk patients can be made and can be used to guide ICD implantation for primary prevention.