Arrhythmic episodes in patients implanted with a cardioverter-defibrillator – results from the Prospective Study on Predictive Quality with Preferencing PainFree ATP therapies (4P)

The Prospective Study on Predictive Quality with Preferencing PainFree ATP therapies (4P) was a prospective, multicenter, observational study of planned 24 months duration aimed at generating real-world evidence of ventricular tachyarrhythmia management by implantable cardioverter defibrillators (ICD) under conditions of daily practice. The study is registered on clinicaltrials.gov with the reference number NCT01509378.

Eligible patients were adults who gave their written informed consent for participation in the study and implantation of an ICD based on a Class I indication for primary or secondary prevention of sudden cardiac death according to the latest published guidelines [10, 11]. Single-chamber (SC), dual-chamber (DC) and triple-chamber (CRT-D) devices, either as new implants, upgrades or replacements, were included.

Leads could be from any manufacturer. A connection to the CareLink™ network was required. Patients with permanent atrial fibrillation, a life expectancy of less than 24 months due to another non-cardiac disease or participating in another concomitant trial were excluded. The study was conducted in accordance with the Declaration of Helsinki. Ethical Review Board approval was obtained prior to study start from the corresponding institutions of all participating centers.

The primary objective of the study was to document device activity and performance in patients with a Class I indication seen in daily practice. In order to assess SST capability to detect and categorize device-based episodes, sensitivity, specificity, positive- and negative predictive values were computed based on the number device-detected, discriminated and categorized VTAs, ATP therapies, and rescue shocks. Computation of episodes considered both number of episodes and number of patients having experienced one or more of such episodes during the follow-up period. The secondary objective was to report medical outcomes in these patients: adjudicated symptomatic events; hospitalizations (all-cause, cardiac, and arrhythmia-related), deaths (all-cause, cardiac), and severe adverse events including serious adverse device effects (SADE). Finally, in addition to symptoms registered at any of the follow-up visits, patients were encouraged to consult or to inform the follow-up center if they had symptoms suggestive of arrhythmia episodes such as syncope, pre-syncope, palpitations or shock. An expert board of two experienced investigators and one external expert, not involved in the trial, analyzed these events. Symptomatic events were adjudicated and classified as VTA or not by the expert board, using the device-recorded data.

All implanted ICD devices featured enhanced detection algorithms integrated in the SST (Medtronic Inc., Minneapolis, MN, USA) package, which includes: Lead Noise Discrimination differentiates RV lead noise from VT/VF by comparing a far-field EGM signal to near-field sensing; RV Lead Integrity Alert extends the VF detection time, triggers programmable alerts and increases diagnostic data collection and monitoring in case of lead malfunction; PR-Logic and Wavelet are algorithms that differentiate ventricular from supraventricular rhythms considering either the relation between A and V EGMs, for PR-logic, or morphology of the V EGM during tachycardia compared to V EGM during sinus rhythm; T-wave oversensing withholds therapy if there is evidence that a fast ventricular rate results from double-counting due to T-wave oversensing; Confirmation + confirms the presence of an arrhythmia by comparing the rhythm cycle length to a calculated confirmation interval before a shock is delivered following the capacitor charge.

All participating centers received a recommendation for ICD programming strategies based on available evidence. Such evidence-based programming was recommended for all patients. In brief, in primary prevention, recommended cycle length for VF detection was 320 ms with an initial Number of Intervals to Detect (NID) of 30/40 [12]. In addition, FVT and VT detection was to be set OFF. The VT monitoring zone was based on a cycle length of 400 ms and a NID of 32 [12]. All SST algorithms were to be switched ON with SVT limit set at 260 ms [13‐15]. In secondary prevention, the same settings applied except the recommended cycle length for VF detection was 300 ms with initial NID 30/40 [16] and VT detection switched ON with a cycle length of 360 ms with an initial NID of 16. Primary therapy was ATP [11]. These programming settings are based on published scientific evidence [11, 13‐16] and were recommended as such in the study protocol. Accordingly, specific programming information was handed out to each implanter as part of the study documentation before trial initiation (Figs. 5 and 6 in Appendix). The protocol allowed divergences based on the judgment of the implanting or follow-up physician.

All patients who matched with the inclusion and exclusion criteria were included in the statistical analysis. Descriptive statistics were used for baseline characteristics and outcomes of interest. No imputation was done for missing data. Exploratory significance testing was performed between the three device groups (SC-ICD, DC-ICD, CRT-D), a two-sided p value of less than 0.05 being required for significance. The Bonferroni correction was applied for the post-hoc comparisons. Odds Ratios (OR) with 95% confidence intervals (95% CI) were calculated using univariate logistic regression methods for identifying predictors of technical and medical outcomes of interest. SST performance was assessed by computing sensitivity, specificity, as well as negative and positive predictive values. Time to first medical outcomes (hospitalizations, deaths) was described by Kaplan-Meier curves with Cox regression models applied for adjustment between study centres and the Hazard Ratios (HR) with 95% CI were reported. The annual rates of device therapies and clinical events were presented per 100 patient years together with the Poisson 95% CI. Treatment success was defined as the absence of VTA redetection following therapy delivery. The power calculation was based on the results in the ATP arm of the PainFree Rx II trial [11]. This trial compared the efficacy of ATP for shock prevention in 313 patients with 4230 spontaneous episodes during a mean follow-up time of 11 months. Using a translational approach, 200 patients followed during 24 months were deemed necessary to be included in the 4P study. All analyses were performed with the SAS 9.4 software package (SAS Institute Inc., Cary, NC, USA) by a senior statistician expert in the field (L.M.).

Between September 2011 and January 2014, 199 patients were enrolled in 10 participating centers in Switzerland. Three patients were protocol violators and excluded from the analysis, two of them because of permanent atrial fibrillation at baseline and one because of participation in another trial. Thus, 196 patients in three groups (48 SC-ICD, 50 DC-ICD and 98 CRT-D) were included in the analysis (Fig. 1). Detailed baseline characteristics presented in Table 1 show that CRT-D patients were older, more clinically compromised with significantly lower left ventricular ejection fraction and a significantly higher proportion of patients with symptomatic heart failure in NYHA II-IV compared to the other 2 groups.

Patients were followed for a mean (± SD) duration of 27.7 (8.6) months, range 0.8 to 44.4 months, with similar durations of follow-up in the three ICD groups (28.1, 27.2, and 27.7 months, respectively). Device programming strategies diverged from recommendations (Figs. 5 and 6 in Appendix) in all patients but one (Table 4 in Appendix); in particular, VT cycle length was programmed longer than recommended in 71.4% of the devices (mean 353 ± 27 and 368 ± 51 ms in primary and secondary prevention, respectively) and the Number of Intervals to Detect (NID) was programmed shorter than recommended in 47.4%. This phenomenon has been observed in all participating centers. Programmed mean SVT limit was consistent with recommendations (256 ± 17 and 259 ± 19 ms in primary and secondary prevention, respectively). Regarding VF cycle length, there was no significant difference between primary and secondary prevention (295 ± 22 and 294 ± 22 ms, respectively) and no significant difference between SC/DC/CRT-D. In all patients features of SST were turned “ON” by default. For further details on programming divergences, refer to the Table 4 in Appendix.

During the follow-up period, 5147 episodes were device-detected, of which 1797 (34.9%) were device-categorized as possible VTAs (Fig. 2). Of these, 523 were withheld from therapy delivery by SST (Fig. 2). As shown in Table 2, the remaining 1274 VTA episodes (1161 VT, 16 FVT, 97 VF) resulted in device therapy delivery: 1208 ATP only, 62 shock only, and 4 ATP followed by a rescue shock. The overall device-therapy (ATP ± rescue shock) delivery rate was 2.8 (95% CI 2.7–3.0) per 100 patient-years (3.3 (3.0–3.7) for SC-ICD, 6.6 (6.2–7.1) for DC-ICD, and 0.7 (0.6–0.8) for CRT-D). The ATP success rate was 99.9% (1207 successfully treated episodes of 1208 treated) and the shock success rate was 100% (60 successful shock episodes out of 60 delivered) when excluding the single patient with an electrical storm (one non-successful ATP followed by six non-successful shocks and one successful shock – counted as one episode). A total of seven inappropriate interventions occurred in 5 patients (2.6%) due to noise/artefacts (1 ATP, 1 ATP + shock, 2 shocks), non-sustained VT (1 shock), and SVT (2 shocks).

The risk of experiencing one or more VTAs was higher in patients with a history of VT/VF episodes (Odds Ratio (OR) 2.9, 95% CI 1.4–6.0, p = 0.004) and in patients with a secondary prevention indication for ICD implant (OR 2.4, 95% CI 1.3–4.4, p = 0.006). Patients treated with a CRT-D were at lower risk of such events (OR 0.5, 95% CI 0.3–0.9, p = 0.016). Consistently, delivery of ATP therapy was significantly more likely in patients with a history of VT/VF (OR 3.5, 95%CI 1.5–8.2, p = 0.003) or with a secondary prevention indication (OR 3.1, 95%CI 1.4–6.8, p = 0.005), and less likely in patients treated with a beta-blocker (OR 0.3, 95% CI 0.1–0.7, p = 0.004), with similar findings with regard to rescue shock delivery.

In total, 146 of 196 patients (74.5%) had one or more device-detected VTA during the 2.3 years of observation (Fig. 2). Of these, 44 patients (22.4%) had experienced one or more VTA episodes per patient that resulted in therapy delivery by the device: 26 patients (59.1%) with VT episodes, 6 (13.6%) FVT, and 25 (56.8%) VF (Table 2). Treatment success rates at the patient level were 100%, for both ATP and shock therapies.

During the course of the study, 215 symptomatic clinical events suggestive of spontaneous VTAs were reported in 45 patients. Using device recordings, 175 episodes (81.4%, in 39 patients) were adjudicated as appropriate VTA episodes: 156 VT, 7 FVT, and 12 VF were treated by either ATP only (125), ATP followed by rescue shock (13), or shock only (20), and 17 episodes self-terminated before therapy was delivered. The remaining episodes were not-classified (29), inappropriate interventions (7; see above), or short non-sustained ventricular tachycardia episodes (4).

Of the 196 patients included in the analysis, 19 (9.7%) died during the 2.3 years of observation, 3 in the SC-ICD group, 4 in the DC-ICD group and 12 in the CRT-D group, corresponding to annual mortality rates of 2.6, 3.5, and 5.3 per 100 patient-years, respectively (Table 3, Fig. 3a). Ten patients died of non-cardiac causes, including 2 of pneumonia, 2 after suicide, and 1 each of cancer, end-stage renal disease, stroke, pulmonary embolism, amyotrophic lateral sclerosis, and one not further specified. Seven patients (all CRT-D patients) died of cardiac causes (4 of worsening heart failure, 2 of acute myocardial infarctions, and 1 of a not shockable recurrent VF). Two patients died of unreported causes. No cases of death were temporally associated with a shock.

Overall, 123 hospitalizations of any cause occurred in 72 patients, of which 63 hospitalizations for cardiac reasons (in 47 patients), 23 hospitalizations for arrhythmia (in 14 patients). As shown in Table 3 and Fig. 3b, hospitalization rates per 100 patient-years were significantly higher in the DC-ICD group compared to the SC-ICD group (any cause, non-cardiac and cardiac causes, and arrhythmia). Overall mean length of stay was 14.3 ± 40.4 days (a single patient with long-term hospitalization) with corresponding values of 4.9 ± 4.2, 21.7 ± 68.8, and 12.6 ± 17.1 days for the SC-ICD, DC-ICD and CRT-D groups, respectively. The risk of hospitalization was increased in patients with heart failure NYHA-class III-IV (any hospitalization with Hazard Ratio (HR) 2.4, 95%CI 1.4–4.1, p = 0.002 and cardiac hospitalizations with HR 2.9, 95%CI 1.4–6.1, p = 0.004) and in patients taking an anti-arrhythmic drug (any hospitalization HR 2.1, 95%CI 1.2–3.5, p = 0.006).

Nine SADE (in 6 patients) were reported, including 5 right ventricular lead dysfunctions, 2 device dysfunctions (1 battery end-of life state and 1 device dislocation), and 2 cases of device pocket infection.

The present study has some limitations. First, ICD patients with heart failure disease and left ventricular dysfunction often present with co-existing comorbidities such as diabetes mellitus, metabolic syndrome, and moderate-to-severe renal impairment. These conditions and their therapeutic control over time may have had an important impact on hospitalization rates, mortality rates, as well as ventricular arrhythmic burden [20‐23]. Furthermore, the identification of patients at higher risk by integrating echocardiographic evaluations of cardiac function and left ventricular dimensions with neuroendocrinal and inflammatory parameters could certainly have been relevant. However, neither detailed echocardiographic evaluation nor sampling of particular hematologic markers, were defined by the study protocol. Second, patients with persistent atrial fibrillation were not included. This may have contributed to the low rate of inappropriate shocks. However, 25% of the included patients had a history of non-persistent atrial fibrillation. Third, the comparisons between the different ICD models should be considered exploratory as the study was not designed for this purpose. On the other hand, in a context of widely available strong evidence of the benefits and risks of ICD therapy, 4P offers unique real-world insights into the patient profiles, arrhythmic events and electrical therapies as they occur in daily routine and confirms that results achieved in clinical outcome trials are reproducible in conditions of daily practice. Fourth, patients included in 4P were followed during only 2 years with 44% of them receiving a new implant. Long term studies over 11 years which included patients who underwent multiple ICD device replacements suggest that the proportion of patients experiencing appropriate and inappropriate shocks may increase over time [24]. Thus, the low rates of appropriate and inappropriate shocks observed in 4P may be explained, at least in part, by the comparatively short duration of observation.