Myocardial scarring and recurrence of ventricular arrhythmia in patients surviving an out‐of‐hospital cardiac arrest

Prediction of recurrent ventricular arrhythmia (VA) in survivors of an out‐of‐hospital cardiac arrest (OHCA) is important, but currently difficult. Risk of recurrence may be related to presence of myocardial scarring assessed with late gadolinium enhancement cardiac magnetic resonance (LGE‐CMR). Our study aims to characterize myocardial scarring as defined by LGE‐CMR in survivors of a VA‐OHCA and investigate its potential role in the risk of new VA events.


| INTRODUCTION
Ventricular arrhythmia-related out-of-hospital cardiac arrest (VA-OHCA) is the main cause of sudden cardiac death (SCD) worldwide 1,2 with ischemic heart disease accounting for up to 80% of the cases of SCD. 3,4According to new guidelines, the recommended standard of care of VA-OHCA survivors includes routine performance of an electrocardiogram, echocardiography and a coronary angiography (CAG). 5A focused trans-thoracic echocardiography is the first-line imaging technique to assess global cardiac function and can acutely detect reversible causes of VA-OHCA. 6Although its sensitivity to detect the underlying myocardial phenotype predisposing to VA events solely relying on left ventricular ejection fraction (LVEF) as a structural abnormality is low so its role as a prognostic tool is controversial. 7,8Importantly, secondary prevention implantable cardioverter-defibrillator (ICD) patients tend to show higher LVEF at the time of ICD implantation than LVEF threshold established for ICD primary prevention indication, highlighting the limited role of conventional imaging techniques. 91][12] This imaging technique allows not only for a quantification of the burden of myocardial scar but also for a detailed characterization of several scar components: core mass, border zone (BZ) mass, and BZ channels.All these scar components can be substrates for ventricular arrhythmias and have been shown to be a strong predictors of outcome beyond LVEF in primary prevention patients. 13,14[17][18] In this study, we sought to explore in detail the characteristics of the myocardial scar of any etiology as defined by LGE-CMR and investigate its potential role in the prediction of VA events during follow-up in survivors of an OHCA.The study complied with the Declaration of Helsinki, and the local ethics committee approved the study protocol.All participants included in the study provided informed written consent.

| CAG
CAG was performed at admission in patients with high clinical suspicion of myocardial ischemia or as a part of the diagnostic workup.Visually estimated stenoses >70% and/or fractional flow reserve values of ≤0.80 were defined as significant and treated with percutaneous coronary intervention.Complete revascularization was defined at the discretion of the operator.

| Late gadolinium CMR processing
LGE-CMR was performed using a 3-Tesla scanner (MAGNETOM Trio; Siemens Healthcare), or 1.5-Tesla scanners (ACHIEVA; and, or LGE-CMR imaging postprocessing with ADAS-3D software in two VA-OHCA patients with (A) and without (B) BZ channel identification.Upper left panel: LGE-CMR reconstruction of the LV showing a lateral scar (core in red, BZ in yellow, and healthy myocardium in blue).A white line is drawn over the surface, representing the BZ channel (white arrow).The substrate progress is seen through different layers, from the endocardium (10%-40%) to the epicardium (50%-90%), with a defined BZ channel (20%).Upper right panel: LGE-CMR raw images in long (A) and short axis (B) used by the software to detect scar, core, BZ and BZ channels.Lower left and right panel: LGE-CMR reconstruction and raw images in long (A) and short axis (B) of the LV showing a lateral scar of a VA-OHCA patient without BZ channel identification.BZ, border zone; CMR, cardiac magnetic resonance; LGE, late gadolinium enhancement; LV, left ventricle; OHCA, out-of-hospital cardiac arrest; VA, ventricular arrhythmia.MAGNETOM Aera, Avanto, or Espree; Siemens Healthcare).Myocardial fibrosis was directly quantified via LGE and analyzed using a previously described semiautomated protocol using ADAS 3D software (ADAS3D). 13ne concentric surface layers (from 10% to 90%) were created automatically from endocardium to epicardium of the LV wall thickness, obtaining a 3D shell for each layer.A PSI-based algorithm was applied to characterize the hyper-enhanced area as core zone, BZ or healthy tissue using 40 ± 5% and 60 ± 5% of the maximum PSI as thresholds.

| Myocardial scar parameters
The total scar mass, BZ mass, and core mass in each shell were automatically measured using the ADAS 3D LV software.A BZ channel is defined as a corridor of BZ connecting two areas of normal myocardium flowing between two core areas, between a core area and a valve annulus or within a core area.These BZ channels are automatically detected by the software. 17,19e LGE-CMR imaging postprocessing is illustrated in Figure 1 in a patient (A) with a BZ channel (white arrow) and recurrent VA; and a patient (B) without a BZ channel and no recurrent VA.

| Study end points
The primary endpoint was recurrent VA defined as appropriate  Receiver operating characteristic (ROC) curve analyses were used to evaluate the optimal cutpoint value of scar-, core-, and BZ mass, and the presence of BZ channels for predicting occurrence of recurrent VA-that is, VA recurrence is most likely to occur in patients with these exact values or values above.The incidences of recurrent VA depending on the different scar determinants stratified by optimal cutpoint were estimated using the cumulative incidence function (CIF) with competing risk of death.Graphic presentation of the cumulative probability was done using the CIF curves, and the unadjusted differences between patients with versus without the optimal cutpoint for the different scar determinants were compared with Gray's test.A sensitivity analysis was performed on VA recurrence in the OHCA patients (n = 170) with a conventional CMR (no LGE protocol).All analyses were performed with RStudio Version 1.2.1335 (R Core Team).A two-sided p value ≤ .05 was considered statistically significant.from Barcelona, 87% male; median age 57.9 years; 29% known ischemic) (Figure 2).A total of n = 170 patients with a conventional CMR (no LGE protocol) were excluded.

| Patient sample
Clinical and demographic characteristics are summarized in Table 1.
We found no statistically significant differences in baseline characteristics between the patients with versus without recurrent VA.However, a trend toward a higher number of patients with already known coronary artery disease and prior myocardial infarction (before index hospitalization

| Delayed enhanced cardiac magnetic resonance results
The LGE-CMR substrate derived parameters are shown in Table 2.
We found a trend toward a lower mean LVEF in patients with VA recurrence (42.9 ± 16.1% vs. 52.1 ± 16.1%; p = .060),and a higher incidence of patients with recurrent VA had LVEF ≤ 35%.The LVEDV and LVESV were higher in patients with recurrent VA.The extent of LGE was larger in patients with recurrent VA compared with patients without recurrent VA (Table 2).
Patients with recurrent VA presented significantly larger scar mass (consisting of core-and BZ mass), core mass, and BZ mass compared with patients without recurrent VA (Table 2).A total of 37 BZ channels were identified in 17 (33%) patients, with a median of 2 BZ channels per patient in both groups.A significantly higher proportion of patients with recurrent VA had BZ channels compared with patients without VA recurrence with no significant differences in BZ channel mass, length, width, or area (Table 2).The findings were consistent among study sites.
In a total of n = 45 (87%) patients a 3T scanner was used.We found no differences in scar characteristics between studies performed using a 3T versus 1.5T scanners (scar mass, p = .420;core mass, p = .092;BZ mass, p = .122;BZ channels, p = .311).

| Predictors of recurrent events
The univariate hazard model analyses for recurrent VA revealed that none of the clinical baseline characteristics were associated with recurrent VA at follow-up, whereas all scar parameters were significantly associated with the primary endpoint (Table 3).
T A B L E 2 Late gadolinium enhancement cardiac magnetic resonance derived substrate parameters in the study population (n = 52).Three multivariable Cox proportional hazards regression models for prediction of recurrence of VA were created including the presence of BZ channels, adjusted by: scar mass, and LVEF (model 1); core mass, and LVEF (model 2); and BZ mass, and LVEF (model 3) (Table 3).In all models, one of the scar parameters evaluated was the  4. The CIFs showed statistically significant differences in recurrence of VA related to scar-, core-, and BZ mass; and the presence of BZ channels stratified by optimal cutpoints as shown in Figure 4.
T A B L E 3 Multivariable Cox proportional hazards regression models to identify independent variables associated with recurrent ventricular arrhythmia events in the study population.F I G U R E 3 ROC curves for the different scar determinants.BZc, border zone channels.
A sensitivity analysis was performed on VA recurrence in the OHCA patients (n = 170) with a conventional CMR (no LGE protocol) (Supporting Information: Table 1).Patients with a conventional CMR were slightly older and had a higher frequency of known CAD, whereas VA recurrence and LVEF were similar between patients with a conventional CMR versus LGE-CMR.

| DISCUSSION
The present study is to our knowledge the first to describe scar characteristics visualized by LGE-CMR in a VA-OHCA population with mostly preserved LVEF and the subsequent risk of recurrent VA at a mid-long-term follow-up.The main findings were that (i) all scar parameters were associated with VA recurrence with the presence of BZ channels being the strongest predictor, and (ii) LVEF was not predictive of VA recurrence in multivariable analyses.The results were consistent among study sites.
The prognostic stratification of new life-threatening arrhythmias in OHCA survivors remains a challenge, and there is a general consensus that the presence of severe LV dysfunction as a high-risk marker of VA and SCD is deficient. 7,8In the present study, LVEF was not an independent predictor of recurrent VA in any of the multivariate models, however, less than 50% of the patients had LVEF < 35%.This supported by previous studies reporting that LVEF lacks both sensitivity and specificity in predicting life-threatening arrhythmias. 20On the contrary, our results emphasize the potential of LGE-CMR-based scar characterization as a risk stratification tool in identifying OHCA patients with high risk of VA irrespective of etiology confirming the results of previous metanalysis. 21In this sense, our study demonstrates the predictive value of BZ channels in cardiac arrest survivors.The BZ channels within the BZ tissue are widely described as the main substrate for re-entrant VA and correlate well with the electrical circuits of slow conductivity or VT isthmuses identified with voltage mapping during scar-related VA ablation procedures. 15,16We found the presence of BZ channels to be strongly associated with recurrent VA in two out of three multivariable analyses and identified patients with recurrent VA with a 67% sensitivity and 85% specificity.Other studies from our group, using the same CMR technique as the present, have similarly demonstrated that the presence of BZ channels is predictive for VA events 6,13,19 in scenarios other than OHCA.In addition to the predictive value of BZ channels, the two previous studies from our group in patients with primary prevention showed that a cutpoint of >10 g scar mass strongly identified patients at high risk of VA. 13,19 In the present study we found a cutpoint of >21 g scar mass to be predictive of recurrent VA, that is, twice as much as previously described for patients in primary prevention.The same difference was found in the cutpoint for BZ mass.This difference might be explained by the different nature of the studied population.Our study includes patients with better LVEF in comparison with studies focusing on primary prevention patients.In this sense, it could be hypothesized that because our population is "healthier" in terms of LV function, more extensive scar characteristics might be needed to initiate VA.Hence, VA recurrence in patients with preserved LVEF may identify a healthier myocardium being less prone to arrhythmias and demanding a higher cutpoint of arrhythmogenic scar tissue, whereas the myocardium in a patient with reduced LVEF is more receptive and thus develop VA more easily.Another explanation could be that the OHCA patients in general have been more extensively revascularized and thus other substrates than scar tissue (residual ischemia) is better taken care of.Still, we found a tendency toward patients with recurrent VA more often had known coronary artery disease and prior myocardial infarction along with a higher incidence of chronic total occlusions and were more often incompletely revascularized during index hospitalization.Thus, patients with recurrent VA were in increased the risk of having residual ischemia, which might have created or accentuated an electrical instable milieu with local electrophysiological alterations with the potential of inducing VA.Nevertheless, our data show that a more detailed analysis of CMR data is informative and, in that regard, BZ channels seems to be is a reliable tool.
Risk stratification is crucial for effective treatment planning in patients surviving an OHCA.Our retrospective data shows that scar characterization with CMR may be useful in risk stratification for subsequent life-threatening arrhythmias in OHCA survivors, including those with preserved LVEF.However, the different cutpoints of the scar characteristics in our study may be the result of statistical variation between the two relatively small nonrandomized studies and thus only very rough approximations given limited numbers of patients studied.Additional studies to confirm this are needed.

| CONCLUSION(S)
All myocardial scar parameters but especially the presence of BZ channels analyzed by LGE-CMR are strong predictors of recurrence of ventricular arrhythmia in cardiac arrest survivors with LGE on CMR and have better predictive power than LVEF.

| LIMITATIONS
This study has several limitations.First, this was a retrospective study with a relatively small sample size.The study population, however, is well characterized and systematically reflects our pre-ICD CMR substrate characterization, and our results show a strong association between scar characteristics and recurrent VA in a fragile patient population with an a priori high probability of recurrent arrhythmias.A second limitation is the heterogeneity of diagnoses, that is, ischemic versus nonischemic, and the subsequent possible differences in the substrate of VA analyzed with LGE-CMR.However, the study population was well-balanced.Third, the primary endpoint was defined as the composite of appropriate ICD therapy including ATP, shocks, or both.We are aware that ICD shocks and ATP are not equivalent.Finally, although this was a multicenter study on a well-depicted OHCA population, there is a need for prospective studies involving larger patient cohorts to confirm the precision of the present values.

2 | METHODS 2 . 1 |
Study population A retrospective LGE-CMR study including consecutive patients surviving a VA-OHCA without ST-segment myocardial infarction (STEMI) at Rigshospitalet, Copenhagen University Hospital, and Hospital Clínic, University of Barcelona between 2015 and 2022.All patients included in this study underwent LGE-CMR and subsequently received a secondary prevention ICD before discharge.Both patients with and without scar were included.Importantly, patients with LGE-CMR did not necessarily have detectable or measurable LGE but underwent CMR including an LGE protocol.Patients with a conventional CMR (no LGE protocol), technically unsuitable LGE-CMR, channelopathies, and sequelae due to a mechanical valve, or extracorporeal membrane oxygenation (ECMO) were excluded.The study was designed to compare patient and LGE-CMR characteristics in VA-OHCA patients with versus without VA recurrence.
ICD therapy for ventricular tachycardia (VT) or fibrillation (VF) (antitachycardia pacing [ATP], shock, or both) or clinically detected VT or VF.Secondary end point was all-cause mortality.Inappropriate ICD therapy was not registered as an end point and was not included.Duration of follow-up was determined from the date of VA-OHCA to the time of first VA event, time of death, or end of follow-up.

2. 6 |
Statistics Demographic, procedural, and end point characteristics are presented as mean with standard deviation (SD), or median and interquartile range (IQR) for continuous variables, and frequencies and percentage for categorical variables.For Table 2, the Wilcoxon rank-sum test was used to evaluate continuous variables and the χ 2 test or Fisher's exact test (when appropriate) for categorical values.Missing values are reported.To analyze the effect of the ischemic versus nonischemic diagnosis on VA recurrence, we did a two-way analysis of variance (ANOVA).Crude and multivariate hazard ratios were obtained from the cause-specific Cox model.Variables selected in the univariate analyses (p value ≤ .05):scar-, core-, and BZ mass, and the presence of BZ channels, and LVEF (because of its established prognostic value) were entered into three multivariate cause-specific hazards models to estimate the independent effect of the scar tissue characteristics on the primary endpoint.
Between 2015 and 2022, a total of 230 VA-OHCA patients without STEMI met the criteria for secondary prevention ICD and were referred for CMR before device implantation at Copenhagen University Hospital, Rigshospitalet, Denmark (n = 97) and Hospital F I G U R E 2 Flowchart of the study population including inclusion and exclusion criteria.A total of 52 VA-OHCA patients with LGE-CMR and without STEMI were included of whom 18 had recurrent VA.CMR, cardiac magnetic resonance; ECMO, extracorporeal membrane oxygenation; LGE, late gadolinium enhancement; OHCA, out-of-hospital cardiac arrest; STEMI, ST-segment elevation myocardial infarction; VA, ventricular arrhythmia.THOMSEN ET AL. | 2289 Clínic, University of Barcelona (n = 133) of whom we included 52 VA-OHCA survivors with LGE-CMR (n = 27 from Copenhagen, n = 25 only independent predictor of recurrent VA: in model 1 the presence of BZ channels, in model 2 the core mass, and in model 3 the presence of BZ channels.The overall diagnostic performance of the different scar characteristics to differentiate patients with versus without VA recurrence are shown in Table 4.The AUC for scar mass was 0.76 (95% CI: 0.61-0.91;p = 0.001), for core mass of 0.75 (95% CI 0.59-0.90;p = .002),for BZ mass of 0.74 (95% CI 0.59-0.89;p = .002),and for the presence of BZ channels of 0.76 (95% CI 0.63-0.89;p < .001).The ROC curves for all the scar parameters analyzed are shown in Figure 3 and the best cutpoints for the identification of patients with documented VA for the different scar parameters are shown in Table

F
I G U R E 4 (A) Cumulative incidence function curve for recurrent ventricular arrhythmia depending on the presence of BZ channels.(B) Cumulative incidence function curve for recurrent ventricular arrhythmia depending on the presence of the cutpoint for scar mass.(C) Cumulative incidence function curve for recurrent ventricular arrhythmia depending on the presence of the cutpoint for core mass (D) Cumulative incidence function curve for recurrent ventricular arrhythmia depending on the presence of the cutpoint for BZ mass.BZ, border zone.
Baseline characteristics in the study population (n = 52).
Crude hazard ratios for each variable are shown to the right.
Abbreviations: BZ, border zone; HR, hazard ratio; LVEF, left ventricular ejection fraction.T A B L E 4The diagnostic performance of the different scar characteristics.Abbreviations: AUC, area under the curve; BZ, border zone; VA, ventricular arrhythmia.