Arrhythmic risk stratification in heart failure mid‐range ejection fraction patients with a non‐invasive guiding to programmed ventricular stimulation two‐step approach

Abstract Background Although some post myocardial infarction (post‐MI) and dilated cardiomyopathy (DCM) patients with mid‐range ejection fraction heart failure (HFmrEF/40%‐49%) face an increased risk for arrhythmic sudden cardiac death (SCD), current guidelines do not recommend an implantable cardiac defibrilator (ICD). We risk stratified hospitalized HFmrEF patients for SCD with a combined non‐invasive risk factors (NIRFs) guiding to programmed ventricular stimulation (PVS) two‐step approach. Methods Forty‐eight patients (male = 83%, age = 64 ± 14 years, LVEF = 45 ± 5%, CAD = 69%, DCM = 31%) underwent a NIRFs screening first‐step with electrocardiogram (ECG), SAECG, Echocardiography and 24‐hour ambulatory ECG (AECG). Thirty‐two patients with presence of one of three NIRFs (SAECG ≥ 2 positive criteria for late potentials, ventricular premature beats ≥ 240/24 hours, and non‐sustained ventricular tachycardia [VT] episode ≥ 1/24 hours) were further investigated with PVS. Patients were classified as either low risk (Group 1, n = 16, NIRFs−), moderate risk (Group 2, n = 18, NIRFs+/PVS−), and high risk (Group 3, n = 14, NIRFs+/PVS+). All in Group 3 received an ICD. Results After 41 ± 18 months, 9 of 48 patients, experienced the major arrhythmic event (MAE) endpoint (clinical VT/fibrillation = 3, appropriate ICD activation = 6). The endpoint occurred more frequently in Group 3 (7/14, 50%) than in Groups 1 and 2 (2/34, 5.8%). Logistic regression model adjusted for PVS, age, and LVEF revealed that PVS was an independent MAE predictor (OR: 21.152, 95% CI: 2.618‐170.887, P = .004). Kaplan‐Meier curves diverged significantly (log rank, P < .001) while PVS negative predictive value was 94%. Conclusions In hospitalized HFmrEF post‐MI and DCM patients, a NIRFs guiding to PVS two‐step approach efficiently detected the subgroup at increased risk for MAE.

Such SCD frequency may justify arrhythmic SCD as a possible therapeutic target but lack of methods for the detection of the truly HFmrEF high arrhythmic risk patients lead these patients to be ignored by the current guidelines which recommend an implantable cardiac defibrilator (ICD) only for those with a depressed LVEF ≤ 35%. 8 Furthermore, these populations are heterogeneous with frequent comorbidities including hypertension, coronary artery disease, atrial fibrillation, cerebrovascular disease, obesity, and anemia. 9 Hence, these patients with competing risks demand the adequate directed toward the arrhythmic risk approaches.
Previous attempts for SCD risk estimation of patients with relatively preserved left ventricular systolic function were based on prognostic scores derived from clinical characteristics, 10 such as the presence of left bundle branch block, increased natriuretic peptide levels 11 and impaired glomerular filtration rate. 12 These initial prognostic models had a low sensitivity for the arrhythmic SCD occurrence.
We recently published results from the PRESERVE-EF study 13 addressing post-myocardial infarction (post-MI) patients with well preserved left ventricular systolic function and a mean LVEF 50.8% for the risk of arrhythmic SCD. Our two-step algorithm (non-invasive risk factors [NIRFs] guiding to PVS) detected efficiently the subpopulation at risk for major arrythmic events (MAE) that could be protected by an ICD. Thus, in the present study, we investigated post-MI and dilated cardiomyopathy (DCM) HFmrEF patients with a mean LVEF = 45% for their SCD risk with a combined NIRFs leading to a PVS two-step approach. Patients found at risk received an ICD for primary prevention of SCD and the results of their long-term follow-up are presented.

| Patients and study design
The study conforms with the principles outlined in the Declaration of Patients with ongoing myocardial ischemia anticipated to be improved with revascularization, acute myocarditis, significant valvular disease, hypertrophic or restricted cardiomyopathy, chronic atrial fibrillation, alcohol-associated disease, cardiac toxicity, malignant diseases affecting survival, end-stage renal failure on hemodialysis were excluded.
The screening was performed in two-steps. In the first step, patients underwent a NIRFs investigation with electrocardiogram (ECG), signal averaged ECG (SAECG), 24-hour ambulatory ECG (AECG), and echocardiography (ECHO), and in the second one, those found with at least one positive NIRF were submitted to PVS. 14

24-hour ambulatory ECG monitoring
During hospitalization, every patient in sinus rhythm underwent a 24-hour AECG (Spider View). The recordings were analyzed using SyneScope 3.10 software (SpiderView & Synescope 3.10, Sorin Group, Clamart, France). The events were reviewed and manually corrected after having been automatically classified. Heart rate, RR intervals, number of ventricular premature beats (VPBs), and non-sustained ventricular tachycardia (NSVT) episodes were calculated from the AECG. 25

| Second step: Programmed ventricular stimulation
After informed consent was obtained, NIRFs positive patients underwent PVS with three extrastimuli from two different right ventricle sites (Era-His, Biotronik, Berlin, Germany). Stimulation protocol in our EP Lab has been described previously. 16 DCM patients underwent additional PVS during iv isoproterenol infusion. 18 The protocol was terminated prematurely if a sustained VT (monomorphic or polymorphic) or VF was induced. The combined two-step algorithm is depicted in Figure 1.

Follow-up
Follow-up information was obtained either by having the patients visiting the outpatient arrhythmia clinic or by telephone contact. Patients with ICDs were seen at one month after discharge and then at 6-month intervals throughout the follow-up and whenever they reported palpitations or shocks. During each follow-up visit, patients were examined and devices interrogated to determine the appropriateness of any delivered therapy (antitachycardia pacing and/or shocks). ICD therapies were classified appropriate when they occurred in response to VT or VF and inappropriate when triggered by supraventricular tachycardia, T-wave over-sensing, electrode dysfunction, or other environmental interactions of any cause. The study endpoints were the occurrence of a MAE either in the form of clinical VT/VF or/and appropriate ICD activation.

F I G U R E 1 Flow chart of study and outcomes
For the present study, stored data in our EP Lab were retrospectively analyzed. In the total database (n = 416), the mean LVEF was 32.4%. To investigate the HFmrEF subgroup, all patients with LVEF < 40% were excluded (n = 326). The remaining 90 patients were candidates for analysis. However, complete AECG, SAECG, and PVS data were available only for 48 of them. This subgroup (n = 48) was finally analyzed (Figure 1).

| Statistical analysis
Continuous variables were reported as mean ± SD, with categorical variables as counts and percentages. The Shapiro-Wilk test was applied to evaluate normality of the distributions. All re-

| RE SULTS
The hospitalization cause of the target group of 48 HFmrEF patients included syncope (n = 7), acute ischemic episode (n = 14), TA B L E 1 Clinical characteristics and non-invasive indices for total patients sample and G1, G2, and G3 groups     Table 3). The PVS sensitivity was 87.5%, the specificity 70.8%, the positive predictive value 50%, and the negative predictive value 94.4%. The study flow with associated results is shown in Figure 1.

| D ISCUSS I ON
In this study, we retrospectively analyzed data from 48 hospitalized for a variety of reasons HFmrEF CAD and DCM patients with a mean LVEF = 45±5%, who were subsequently referred to our EP Lab for risk stratification for the primary prevention of SCD. We found that a multifactorial NIRFs leading to PVS two-step approach was able to separate the high from the low-risk HFmrEF population. We also provide evidence that these patients are adequately protected from MAEs by an ICD.
The causes for our HFmrEF patients hospitalization varied and they specifically included syncope, acute coronary syndrome, HF deterioration, supraventricular arrhythmia with haemodynamic destabilization and non-sustained VT. Considering the hospitalization etiologies these patients presented with, they were not the typical community HFmrEF population but rather a hospitalized HFmrEF subgroup with a higher arrhythmic and mortality risk profile. These patients also suffered frequently from diabetes and hypertension. No statistical differences were observed between groups for comorbitidies.  We risk stratified them for SCD with a combined two-step risk stratification approach. In the first-step during the non invasive screening, all the negative NIRFs patients deemed to be low risk has been chalenged recently. 28 However the LVEFcut-off point of 35% remains the principal risk stratification tool for the primary prevention of SCD with current guidelines recommending an ICD only for the post-MI and DCM patients with LVEF ≤ 35%. 8,15 The selection of this cut-off point as a marker of risk, treat the arrhythmic SCD risk as a binary state. This selection reflects the MADIT II 29 and SCD-HeFT 30 trials study design. Patients with a LVEF > 35% were ignored and this happened not because their SCD risk was insignificant, but rather from the nessecity for focusing in a subpopulation in which more frequent endpoints were expected. In fact, SCD risk is continuous through the entire LVEF spectrum with a reverse trend, the more LVEF increases the more SCD decreases, but it never abolishes even in patients with a LVEF > 50%. 6,[31][32][33] Arrhythmic SCD risk in such patients with high LVEF may originate from fibrotic scar areas of small previous myocardial infarctions or silent myocardial infarctions even without a clinical history of CAD. 34 Nowadays we acknowledge the need for protecting the HFmrEF and HFpEF patients as well, 13 which is a larger population on an epidemiological basis. 35 Populations with a relatively preserved LVEF were explored only in a few previous studies. In post-MI patients with an average LVEF = 55±10%, the presence of T-wave alternans, non-sustained VT or/and ventricular late potentials intentified a subpopulation at risk. 6 In ISAR-Risk study, an abnormal heart rate turbulence (HRT) and deceleration capacity (DC) of the heart rate identified a subgroup with severe autonomic failure that had a high mortality risk. 36 Current studies attempt to select high-risk patients for an ICD implantation based on the presence of NIRFs (REFINE-ICD) 37 and Magnetic Resonance Imaging findings (GUIDE-CMR). 38 The PRESERVE-EF study 13

Limitations of the study
The study is based on restrospective analysis of a limited number of HFmrEF patients screened for primary prevention of SCD in our EP Lab. According to their hospitalization etiology, these 48 patients represent probably a higher arrhythmic risk profile subgroup of the total HFmrEF population. Thus, our observed inducibility on PVS and MAE event rates are probably not reflecting the risk existing in the broader HFmrEF population in the community. It is likely that the incidence of positive NIRFs as well as the inducibility on PVS rates might have been higher in our seriously affected hospitalized population subgroup of HFmrEF patients, compared to the rest of the corresponding community population. While it is known that betablockers may protect against malignant arrhythmias, a significant part of our patients was undertreated, with only 60% of them receiving beta-blockers and 69% receiving angiotensin-converting-enzyme inhibitors or angiotensin II receptor blockers. However, it should be noted that no significant difference in therapies for the high risk (G3) vs the low and intermediate risk (G1 and G2) groups have been observed. Furthermore, among the examined NIRFs we did not include novel indices like DC, HRT, T-wave alternans, QT interval duration, or modern Imaging modalities from Echocardiography and Magnetic Resonance of the Heart.

| CON CLUS IONS
A combined NIRFs guiding to PVS, two-step risk stratification approach, detected efficiently those post-MI and DCM Heart Failure mid-range Ejection Fraction (HFmrEF/40%-49%) hospitalized patients, to be at risk for arrhythmic SCD.

ACK N OWLED G M ENTS
The authors thank Ioanna Kampitsa (DipTransIoL) for the editing of the manuscript.

CO N FLI C T S O F I NTE R E S T
Authors declare no conflict of interests for this article.