Multiple premature ventricular contraction morphologies is associated with increased incidence of ventricular tachycardia and fibrillation in real-world patients with insertable cardiac monitors
E W Aabel, J W Dukes, D L Lustgarten, S A Saha, N V K Pothineni, G Rajagopal, P Zimmerman, S Sarkar, Y Cho, D R LakkireddyAbstract
Background
Insertable cardiac monitors (ICM) have the capability to continuously monitor premature ventricular contraction (PVC) burden over time and also detect ventricular tachycardia and fibrillation (VT/VF). Occurrence of multifocal PVCs, rather than unifocal, are associated with structural heart disease that can lead to elevated risk of VT/VF.
Objective
We investigated the association of VT/VF risk with the number of distinct PVC morphologies present in a large real-world cohort of patients implanted with ICMs using a unique PVC morphology classification algorithm developed for ICMs.
Methods
Patients implanted with an ICM for arrhythmia monitoring with PVC detection turned on were included from a de-identified real world data warehouse. Patients were included if they had at least 90 days of PVC burden follow-up. An Artificial Intelligence (AI) model was used for identifying episodes that were likely VT/VF episodes which were then manually adjudicated for true incidence of VT/VF. All the stored ECGs were first run through a PVC detection algorithm to detect PVC beats. All the identified PVC beats were then run through a single lead PVC morphology classification algorithm to classify different PVC morphologies for each patient. The algorithm compares the QRST complexes of the PVCs to each other to iteratively identify morphologies that were significantly different than each other. The incidence rate of true VT/VF during a moving 30-day window was analyzed based on if 0,1,2, and >=3 distinct PVC morphologies were detected during a 90-day lookback period using Negative Binomial Regression.
Results
A total of 6,991 patients (avg. age: 66.5±14.9 years; 50% male) were included in the analysis, with an average follow-up length of 326 days. There was a total of 22,954 tachycardia episodes from 2,131 patients that were detected. After AI model probability-based adjudications, 795 true VT/VF episodes were identified from 301 patients. 30-day windows with 1, 2, and >=3 morphologies detected during the 90-day lookback period all had significantly higher rates of 0.03, 0.04, and 0.11 VT/VFs per 30 days, respectively, than windows with no morphologies detected during the lookback, which had a rate <0.01 (p<0.001 for all) (Figure). In addition, 30-day windows with 1, 2, and >=3 morphologies had statistically significant incidence rate ratios (IRRs) of 4.1, 9.4, and 13.5, respectively, than windows with no morphologies, even after adjusting for age, gender, and device-detected PVC burden (p<0.001 for all).
Conclusion
The number of distinct PVC morphologies as identified by an ICM during a 90-day lookback period was associated with increased risk of VT/VF within the next 30 days, with the highest risk being when 3 or more distinct morphologies are detected. The number of distinct PVC morphologies detected by ICMs may be used as a risk stratification tool to determine when patients are at increased risk of VT/VF.