Predicting electrical storms by remote monitoring of implantable cardioverter-defibrillator patients using machine learning.

AIMS: Electrical storm (ES) is a serious arrhythmic syndrome that is characterized by recurrent episodes of ventricular arrhythmias. Electrical storm is associated with increased mortality and morbidity despite the use of implantable cardioverter-defibrillators (ICDs). Predicting ES could be essential; however, models for predicting this event have never been developed. The goal of this study was to construct and validate machine learning models to predict ES based on daily ICD remote monitoring summaries. METHODS AND RESULTS: Daily ICD summaries from 19 935 patients were used to construct and evaluate two models [logistic regression (LR) and random forest (RF)] for predicting the short-term risk of ES. The models were evaluated on the parts of the data not used for model development. Random forest performed significantly better than LR (P < 0.01), achieving a test accuracy of 0.96 and an area under the curve (AUC) of 0.80 (vs. an accuracy of 0.96 and an AUC of 0.75). The percentage of ventricular pacing and the daytime activity were the most relevant variables in the RF model. CONCLUSION: The use of large-scale machine learning showed that daily summaries of ICD measurements in the absence of clinical information can predict the short-term risk of ES.

The use of guideline recommended beta-blocker therapy in primary prevention implantable cardioverter defibrillator patients: insight from Danish nationwide registers.

Aims: We aimed to examine the use of guideline recommended beta-blocker therapy prior to and after primary prevention implantable cardioverter defibrillator (ICD) implantation in a 'real-life' setting. Methods and results: From the Danish Pacemaker and ICD Registry we identified all 1st-time primary prevention ICD and cardiac resynchronization therapy defibrillator (CRT-D) implantations in Denmark from 2007-12 (n = 2935). Use of beta-blocker, type and dose was acquired through the Danish Prescription Registry. According to guideline recommendations, we defined target daily doses as ≥50 mg carvedilol and ≥200 mg metoprolol. Prior to implantation 2427 of 2935 (83%) patients received beta-blocker therapy, with 2166 patients (89%) having initiated treatment 3 months or more prior to implantation. The majority of patients was prescribed carvedilol (52%) or metoprolol (41%). Patients on carvedilol reached target dosages more frequently than patients on metoprolol, with 39% of patients on carvedilol and 26% of patients on metoprolol at the time of implantation (P < 0.001 for all time-points). Increase in proportion of patients reaching target daily doses was observed for both carvedilol and metoprolol after ICD implantation. Carvedilol treatment was a strong predictor for being on target dose of BB at time of implant, as was treatment with angiotensin-converting enzyme inhibitors and/or spironolactone, no history of myocardial infarction, younger age and less pronounced heart failure symptoms. Conclusion: In a real-life setting of primary prevention ICD patients, 39% and 26% of patients were titrated to optimal target dose of carvedilol or metoprolol prior to implantation. A higher proportion of patients on carvedilol reached target dose, as compared with metoprolol.

The impact of co-morbidity burden on appropriate implantable cardioverter defibrillator therapy and all-cause mortality: insight from Danish nationwide clinical registers.

AIMS: In a nationwide cohort of primary (PP-ICD) and secondary prevention (SP-ICD) implantable cardioverter defibrillator (ICD) patients, we aimed to investigate the association between co-morbidity burden and risk of appropriate ICD therapy and mortality. METHODS AND RESULTS: We identified all patients >18 years, implanted with first-time PP-ICD (n = 1873) or SP-ICD (n = 2461) in Denmark from 2007 to 2012. Co-morbidity was identified in administrative registers of hospitalization and drug prescription from pharmacies. Co-morbidity burden was defined as the number of pre-existing non-ICD indication-related co-morbidities including atrial fibrillation, diabetes, chronic obstructive pulmonary disease, chronic renal disease, liver disease, cancer, chronic psychiatric disease, and peripheral and/or cerebrovascular disease, and divided into four groups (co-morbidity burden 0, 1, 2, and ≥3). Through Cox models, we assessed the impact of co-morbidity burden on appropriate ICD therapy and mortality. Increasing co-morbidity burden was not associated with increased risk of appropriate therapy, irrespective of implant indication [all hazard ratios (HRs) 1.0-1.4, P = NS]. Using no co-morbidities as reference, increasing co-morbidity burden was associated with increased mortality risk in PP-ICD patients (co-morbidity burden 1, HR 2.1; comorbidity burden 2, HR 3.7; co-morbidity burden ≥3, HR 6.6) (all P < 0.001) and SP-ICD patients (co-morbidity burden 1, HR 2.2; co-morbidity burden 2, HR 3.8; co-morbidity burden ≥3, HR 5.8). With increasing co-morbidity burden, an increasing frequency of patients died without having utilized their device, with 72% PP-ICD and 45% SP-ICD patients with co-morbidity burden ≥3 dying without prior appropriate ICD therapy. CONCLUSION: Increasing co-morbidity burden was not associated with increased risk of appropriate ICD therapy. With increasing co-morbidity burden, mortality increased, and a higher proportion of patients died, without ever having utilized their device.

Validation of defibrillator lead performance registry data: insight from the Danish Pacemaker and ICD Register.

AIMS: The validity of registry data on defibrillator lead performance is described only sparsely, despite its clinical importance. This study investigated the validity of defibrillator lead performance registry data in a nationwide and population-based registry. METHODS AND RESULTS: We identified all reported surgical interventions due to defibrillator lead events in the Danish Pacemaker and ICD Register (DPIR) from 2000 to 2013. Medical records of all patients (n = 753) were examined blinded for 5 predefined intervention types and 18 reasons for lead intervention. The overall level of agreement for the types of lead intervention had a positive predictive value (PPV) of 89.4% [95% confidence interval (CI): 87.0-91.5%] and an adjusted agreement (κ value) of 0.81 (95% CI: 0.77-0.85) representing an almost perfect match. Regarding the reasons for lead intervention, the overall PPV was 63.0% (95% CI: 54.8-61.7%) with a κ value of 0.60 (95% CI: 0.55-0.64) representing a moderate match with only few subcategories having low PPV. By redistribution of the specific reasons for lead interventions into three categories commonly used to report lead performance, the overall PPV improved to 87.9% (95% CI: 85.2-90.2%) with a κ value of 0.82 (95% CI:0.78-0.86) representing an almost perfect match. CONCLUSION: The validity of data on defibrillator lead performance recorded in the DPIR is excellent for the specific types of lead intervention and good for the specific reasons for defibrillator lead intervention. The validity of the less detailed overall reasons for lead interventions commonly used to report lead performance is also excellent. These findings indicate high registry data quality appropriate for scientific analysis and industry-independent post-marketing surveillance.

Safety of magnetic resonance scanning without monitoring of patients with pacemakers.

AIMS: The objective of this study was to investigate whether it is safe to perform 1.5-Tesla magnetic resonance imaging (MRI) scans in pacemaker (PM) patients without pulse oximetry or electrocardiogram monitoring and with no special specific absorption rate (SAR) or time limits, provided that the PMs are interrogated and programmed to asynchronous mode prior to the scan. METHODS AND RESULTS: This study reports the outcome of 207 MRI scans on PM patients at Rigshospitalet, Copenhagen University Hospital from June 2010 to September 2013. All MRIs were performed with the PMs in asynchronous mode and without additional monitoring. There were no adverse events registered among the PM patients during the study period. The only statistically significant change after MRI scans was a small, but clinically insignificant increase in atrial sense. No occurrences of reprogramming to power-on-reset were registered. CONCLUSION: It is possible to perform MRI scans relatively safely in PM patients without additional monitoring or change in the normal MRI protocol, given that the PM has been assessed and reprogrammed prior to MRI. This is especially important to remember in the acute setting where MRI scans may be delayed when monitoring facilities are unavailable.

Patient satisfaction and suggestions for improvement of remote ICD monitoring.

PURPOSE: The study aim was to evaluate patient acceptance and content with remote follow-up (FU) of their implantable cardioverter defibrillator (ICD) and to estimate patients' wish for changes in remote follow-up routines. METHODS: Four hundred seventy-four ICD patients at the device follow-up clinic at Rigshospitalet using CareLink® (Medtronic) remote follow-up, who had made ≥2 transmissions, received a questionnaire. RESULTS: Three hundred eighty-five patients (81.2%) answered. Mean time with ICD was 56 ± 45 months and mean age was 62 ± 13 years; 80% was male. Diagnosis related to ICD implant was: ischemic heart disease in 56% and dilated cardiomyopathy in 21%. Twenty-six percent had primary prophylactic indication. Mean time on remote FU was 16.4 ± 6.9 months. Mean time spent on in-clinic FU (two-way transport and FU) was 4 h and 36 min ± 7 h and 50 min, excluding 12 patients from Greenland and Faroe Islands. Ninety-five percent of the patients was very content or content with remote FU compared to in-clinic FU; 3% was less content and 2% was not content. For scheduled transmissions, 21% of the patients wished for a faster reply (sms or e-mail) compared to current practice with a letter. Eighty-four percent preferred more detailed information concerning ICD leads, battery status, and ICD therapies. A total of 96 patients (25%) had performed extra unscheduled remote transmissions: 20 due to shock, 20 due to alarm, 35 due to palpitations, and 18 for other or combined reasons. CONCLUSION: Ninety-five percent of the patients were content with the remote FU. Only 25% had unscheduled transmissions and most unscheduled transmissions were for appropriate reasons. Eighty-four percent of the patients wished for a more detailed response and 21% wished for a faster reply after routine transmissions.

[Safety of magnetic resonance imaging in patients with pacemaker and implantable defibrillator]. / Sikkerhed af magnetisk resonans-skanning hos patienter med pacemaker og implanterbar defibrillator.

INTRODUCTION: The presence of a cardiac implantable device is ICD considered an absolute contraindication to magnetic resonance imaging (MRI). The purpose of this study was to evaluate the safety of performing MRI in patients with cardiac pacemakers and ICDs that had a compelling clinical need for MRI examination. MATERIAL AND METHODS: During a period of nine years we have included 65 patients with cardiac devices (60 pacemakers and five ICDs) who underwent a total of 73 MRI examinations at 1.5 T. All pacemakers were reprogrammed before MRI to asynchronous mode to avoid MRI-induced inhibition or to sense only mode to avoid MRI-induced competitive pacing and potential pro-arrhythmia. All devices were interrogated immediately before and after the MRI examination, which included measurement of sensitivity, pacing capture threshold (PCT) and lead impedance. RESULTS: MRI examinations were completed safely in 63 patients. Inhibition of pacemaker output was observed in one patient and induction of ventricular fibrillation was observed in another with ICD. A significant increase in PCT was rare and only detected in 1% of all electrodes. CONCLUSION: MRI can be performed safely in patients with pacemakers with an acceptable risk-benefit ratio, while MRI of patients with ICDs must still be considered an experimental procedure.

Indication of the radiofrequency induced lesion size by pre-ablation measurements.

BACKGROUND: During radiofrequency ablation of arrhythmias tissue heating and hence lesion size depend on electrode-tissue contact and cooling of the electrode tip caused by cavitary blood flow. These factors are unique and unknown for each catheter placement in the beating heart. A tool for assessing these factors prior to ablation may indicate the lesion size which will be obtained for any given catheter position. METHODS AND RESULTS: Radiofrequency ablation was performed in vitro on strips of left ventricular porcine myocardium during two different levels of convective cooling (0 or 0.1 m/s), two different contact pressures (10 or 30 g) and parallel or perpendicular electrode-tissue orientation using 7F 4 mm tip catheters. Prior to ablation the impedance rise (DeltaIMP) caused by the obtained contact and the temperature rise with a 0.6 W 5 s test pulse (DeltaT) were measured. Subsequently, during unchanged conditions, radiofrequency ablation was performed as either temperature-controlled, power-controlled or irrigated tip ablation and lesion size was determined. DeltaIMP increased significantly (P < 0.05) by improved contact, whereas it was not affected by convective cooling. DeltaT was significantly increased by increasing contact pressure (P < 0.05) and significantly decreased by increased cooling (P < 0.001). DeltaT was not systematically affected by electrode orientation. The product of DeltaT and DeltaIMP showed a significant correlation between the obtained lesion size and power output for temperature-controlled and between lesion size and tip temperature for power-controlled ablation (P < 0.001). CONCLUSIONS: Pre-ablation measurement of DeltaIMP and DeltaT can indicate the lesion size resulting after ablation in temperature-controlled, power-controlled and irrigated ablation in vitro, since DeltaT reflects cavitary cooling and to a smaller extent electrode-tissue contact, and DeltaIMP reflects only electrode-tissue contact.

Mechanisms for enlarging lesion size during irrigated tip radiofrequency ablation: is there a virtual electrode effect?

INTRODUCTION: Irrigated tip radiofrequency ablation of cardiac arrhythmias was developed to increase the size of the radiofrequency-induced lesion, since cooling of the electrode tip allows use of higher power settings. The purpose of this study was to determine if the increased lesion size during irrigated tip ablation is caused by the cooling effect solely or if increased electrical conductivity around the tip also contributes by increasing the "current-delivering size" of the tip: the so-called "virtual electrode effect." METHODS AND RESULTS: In vitro strips of left ventricular porcine myocardium and in vivo canine left ventricles were ablated. In vitro closed loop tip and showerhead irrigated tip catheters were compared. In vitro and in vivo showerhead tip catheters irrigated with solutions having different ionic content were compared. We found no difference in lesion size for closed loop and showerhead-type catheters (998 +/- 345 vs. 811 +/- 313 mm(3) during power-controlled ablation and 227 +/- 76 vs 318 +/- 127 mm(3) during temperature-controlled ablation). For irrigation with liquids having increasing ionic strength we found a decrease in lesion volume in vitro (361 +/- 249 vs. 812 +/- 229 mm(3) (P < 0.001) for power-controlled and 156 +/- 78 vs. 318 +/- 127 mm(3) (P < 0.05) for temperature-controlled ablation and nonsignificant differences in vivo. CONCLUSIONS: The mechanism for enlarging lesion size during radiofrequency irrigated-tip ablation is that higher power levels can be used. There is no virtual electrode effect caused by the highly conductive surroundings of the tip during irrigation. In vitro this effect is shown to be opposite: it decreases lesion size.

Can lesion size during radiofrequency ablation be predicted by the temperature rise to a low power test pulse in vitro?

During radiofrequency ablation sufficient power must be delivered to achieve a proper lesion, while tissue temperatures are kept below 100 degrees C. Tissue temperature and hence lesion size are determined by electrode to tissue contact and convective cooling, which vary with any obtained electrode position in the beating heart. A tool for evaluation of the combined effect of these parameters prior to ablation to predict the resulting lesion size for any actually obtained electrode position would be valuable. A low power pulse prior to ablation will cause a small increase in tip temperature, which will reflect tissue contact and convective cooling of the tip for that particular electrode position and may thereby predict lesion dimensions after subsequent ablation at that site. Ablation was performed in vitro on strips of left ventricular porcine myocardium during different levels of convective cooling, variable contact pressure, and two electrode orientations in temperature-controlled and power-controlled modes. A 4-mm tip catheter was used. Prior to ablation a 30-second, 0.6-W test pulse was delivered and the increase in tip temperature (DeltaT) registered. The study found that DeltaT and lesion size were mainly affected by the external cooling of the electrode tip and less by the electrode to tissue contact. Increased cooling caused a lower DeltaT (P < 0.001), and the effect on lesion dimensions depended on the ablation mode. For temperature-controlled ablation a larger lesion due to increased power output was obtained (P < 0.0001) for increased cooling, whereas for power-controlled ablation a smaller lesion was obtained for increased cooling (P < 0.05). DeltaT can predict lesion size, but the interpretation of DeltaT depends on the ablation mode. For temperature-controlled ablation a high DeltaT indicates a small lesion and for power-controlled ablation a high DeltaT indicates a large lesion.