Identifying patients at risk of slow recurrent VT: a predictive model for optimizing ICD programming
I Szakal, F Komlosi, B Arnoth, G Y Bohus, Z S Varga, P Toth, H Santa, I Osztheimer, N Szegedi, P Perge, Z Sallo, E Tanai, B Merkely, L Geller, K V NagyAbstract
Background
ICD programming in secondary prevention of VT remains largely empirical, particularly regarding the lowest detection threshold and therapy zones. Although current guidelines recommend setting the VT zone at least 10 bpm below the documented VT rate, a subset of patients experience recurrent VTs with a slower rate than the detection threshold, resulting in undetected arrhythmias.
Purpose
To identify those clinical, echocardiographic and arrhythmia-related factors, which are associated with substantially slower recurrent VT in structural heart disease in order to develop a predictive model for risk stratification.
Methods
We retrospectively analyzed consecutive patients with structural heart disease hospitalized for sustained monomorphic VT. Patients were followed for 12 months and categorized based on VT recurrence characteristics. Group 1 included patients who experienced a recurrent VT with a rate below 185 bpm and a frequency decrease of at least 10 bpm, while Group 2 included patients a) without recurrence or b) with a recurrent VT above 185 bpm or c) with less than 10 bpm rate change compared to the index VT. A multivariate logistic regression model was constructed using a backward conditional stepwise approach to determine the optimal predictor combination.
Results
A total of 249 patients were included in our study (Group 1: n=40, Group 2: n=209). In the multivariate analysis electrical storm or incessant VT (OR 4.18, p < 0.01), left ventricular end-systolic diameter greater than 50 mm (OR 3.10, p = 0.02), and moderate to severe mitral regurgitation (OR 2.37, p = 0.04) were identified as independent predictors of slower recurrent VT, while amiodarone therapy showed a trendwise association (OR 3.59, p = 0.1). The model demonstrated good discrimination with an AUC of 0.80 and an overall accuracy of 72.7% with 82.5% sensitivity and 70.8% specificity.
Conclusion
We successfully identified structural and arrhythmic factors associated with a markedly slower recurrent VT in patients with structural heart disease. The proposed predictive model reliably identifies patients at risk of slow recurrent VT, a subgroup in whom conventional ICD programming may fail to detect malignant arrhythmias. These findings underscore the importance of individualized ICD programming strategies in optimizing arrhythmia detection and therapy for secondary prevention of VT.