Prognostic value of late gadolinium enhancement cardiac MRI for ICD therapy in non-ischaemic cardiomyopathy

All consecutive NICM patients undergoing ICD implantation for the primary prevention of SCD at Amsterdam University Medical Centre, location VUMC, the Netherlands, between January 2009 and April 2022, in whom adequate-quality CMR imaging was available, were included in the study. The study population was further refined according to the 2023 position statement of the Task Force of the Netherlands Society of Cardiology [10]. Patients who had received CRT‑D were excluded.

Patients who carried specific genetic variants associated with an increased risk of SCD, such as the phospholamban (PLN) p.Arg14del mutation, lamin A/C (LMNA) mutation, filamin C (FLNC) mutation, receptor-binding motif (RBM20) mutation, and pathogenic desmoplakin (DSP) mutation, were excluded. Additionally, patients with hypertrophic cardiomyopathy (HCM), arrhythmogenic right ventricular cardiomyopathy (ARVC), Brugada syndrome, cardiac sarcoidosis, cardiac amyloidosis or congenital cardiac disease were excluded from the analysis.

The primary endpoint of this study was the time to first appropriate ICD therapy, defined as antitachycardia pacing (ATP) and/or shock for the termination of ventricular arrhythmia. The secondary endpoint was all-cause mortality. Deaths were identified using the National Health and Social Care Information Service and the electronic medical record system.

Each patient was followed for a maximum of 5 years. This duration was selected as it aligns with established timelines for ICD studies [9]. Prolonged follow-up periods resulted in a decrease in patient numbers, partly due to the upgrade from ICD to CRT‑D in some patients. A 5-year follow-up duration allowed for adequate assessment of device efficacy and patient outcomes within a clinically relevant timeframe. A longer follow-up period has been examined and is available upon request to the corresponding author.

Routine device interrogations were performed at 10 days post-ICD implantation, 2 months post-implantation, and then every 6 months thereafter or in the event of device therapy (ATP or shock). The majority of patients were connected to remote monitoring systems, with event transmissions reviewed by specialised cardiac device technicians and electrophysiologists. ICD programming was based on standard clinical practice and was adjusted based on physicians’ discretion [11].

Patients were followed for the occurrence of the primary endpoint or death until 1 May 2024. Data from patients lost to follow-up were censored after their last clinical contact, as were data for patients undergoing ICD extraction, ICD deactivation, or upgrade to a CRT‑D.

Inappropriate ICD therapy and ICD-related complications were assessed throughout the follow-up period. Inappropriate ICD therapy was defined as device-delivered therapy (shock or ATP) triggered by non-ventricular arrhythmic causes, such as atrial fibrillation with rapid conduction, supraventricular tachycardia or lead noise. ICD-related complications were recorded and described for each individual case.

CMR scans were performed using either a 1.5 T or 3 T clinical MRI system (Siemens Avanto, Siemens Sonata, Siemens Sola, Siemens Vida, or GE Signa). LGE imaging included both long-axis and short-axis views to ensure a comprehensive evaluation of myocardial LGE distribution and extent.

Visual assessment of myocardial LGE was conducted by two independent readers. In cases of uncertainty regarding the presence and distribution of LGE, a third reviewer evaluated the LGE-CMR scan. The presence or absence of LGE was assessed for each myocardial segment using the American Heart Association (AHA) 17-segment model. The number of segments with myocardial LGE was recorded for each patient.

Additionally, myocardial LGE was categorised based on its distribution pattern, including midwall, epicardial, subendocardial, transmural, and hinge-point LGE. Midwall LGE was further classified into midwall stripe LGE and patchy midwall LGE. The localisation of LGE was identified as anterior, inferior, lateral, septal, or all-around.

Patients with LGE were compared with patients without LGE on CMR (LGE+ group and LGE− group, respectively). T‑tests were performed on normally distributed continuous variables, while Mann-Whitney U tests were used for non-normally distributed continuous variables. For categorical variables, a chi-square test was employed.

For the time-to-event analyses, cumulative incidence curves were constructed using the Kaplan-Meier method, and hazard ratios with 95% confidence intervals were calculated using Cox proportional-hazards models. Additionally, a chi-square test was conducted to examine the relationship between the extent of LGE, the distribution pattern of left ventricular LGE, and the occurrence of appropriate ICD therapy and all-cause mortality.

Statistical analyses were performed using SPSS statistical software, version 28.

A cohort of 434 patients with NICM who had received an ICD for the primary prevention of SCD was identified; 263 of these patients had undergone LGE-CMR prior to implantation. After excluding 5 patients due to poor-quality LGE-CMR scans, 258 patients remained eligible for further analysis. Patients who had received a CRT‑D were then excluded (n = 123), as were those with HCM (n = 28), ARVC (n = 6), cardiac sarcoidosis (n = 3), cardiac amyloidosis (n = 5), Brugada syndrome (n = 1), and specific genetic variants associated with an increased SCD risk, including the PLN p.Arg14del mutation (n = 3), LMNA mutation (n = 1), and pathogenic DSP mutation (n = 1) (Fig. 3). Additionally, two patients with congenital cardiac disorders were excluded: one with Tetralogy of Fallot and another with Ebstein’s anomaly.

Following these exclusions, 85 patients remained for further analysis. Among these, 41 patients had LGE on CMR (LGE+ group, 48%), while 44 did not show LGE (LGE− group, 52%). Baseline characteristics of both groups are presented in Tab. 1, revealing no significant differences between the groups except for the presence of hypercholesterolaemia. The majority of patients (n = 81, 95%) received a transvenous device, while four patients (5%) received a subcutaneous ICD (S-ICD). The median follow-up duration was 5 years (interquartile range [IQR]: 2.0–5.0 years) for both the LGE+ group (5.0 years, IQR: 1.7–5.0 years) and the LGE− group (5.0 years, IQR: 2.2–5.0 years).

Among the 85 patients with ICDs included in the study, 14 (16%) experienced appropriate ICD therapy for ventricular arrhythmias. Of these 14 patients, 5 (36%) received ATP only, 3 (21%) received both ATP and shock, and 6 (43%) received shock therapy alone. The arrhythmia underlying the therapy was VF in two patients (14%) and VT in the remaining 12 patients (86%).

The primary endpoint of appropriate ICD therapy occurred in 8 patients in the LGE+ group and in 6 patients in the LGE− group, with a 5-year Kaplan-Meier estimated cumulative incidence of 20% and 14%, respectively (hazard ratio (HR) 1.62, confidence interval (CI) 0.56–4.63, p = 0.37) (Fig. 1).

In both univariate and multivariate analyses, no significant variables were found to be associated with the incidence of appropriate ICD therapy among NICM patients (Tab. 2).

Nine patients died over a 5-year follow-up period, indicating a 2% annual mortality rate. Among the LGE+ group, 3 patients died during follow-up, compared with 6 patients in the LGE− group (5-year Kaplan-Meier estimated cumulative incidence of 7% and 14%, respectively, HR 0.60 (95% CI 0.16–2.23), p = 0.46) (Fig. 1). The causes of death for the study population are summarised in Tab. S1 of the Electronic Supplementary Material. Among the 9 patients who died during the study period, 2 deaths (22%) were attributed to respiratory causes, 2 (22%) to heart failure and 2 (22%) to oncological causes. The cause of death was unknown in 3 patients (33%).

The extent of LGE typically ranged from 1 to 4 segments (Fig. S1 in the Electronic Supplementary Material). Patients with LGE were divided into two groups based on the median number of segments with LGE: 1–2 segments and > 3 segments. No significant differences were observed between these groups in terms of appropriate ICD therapy (p = 0.37) or all-cause mortality (p = 0.41).

Additionally, the distribution pattern of LGE was assessed (Fig. S2 in the Electronic Supplementary Material). The midwall septal stripe and hinge-point LGE were the most common patterns, present in 56% and 51% of patients with LGE, respectively. Epicardial LGE was also frequent, occurring in 36% of cases. There was no significant association between appropriate ICD therapy or all-cause mortality and the presence of midwall septal stripe, hinge-point LGE, or epicardial LGE.

Inappropriate ICD therapy was observed in 9 patients (Tab. S2 in the Electronic Supplementary Material). The most common cause was supraventricular tachycardia, including atrial fibrillation, which accounted for 6 of the 9 cases (66%). Other causes of inappropriate therapy included lead fracture (n = 1/9, 11%), T‑wave oversensing (n = 1/9, 11%), and subcutaneous emphysema (n = 1/9, 11%).

Device-related complications occurred in 19 patients (22%) and are summarised in Tab. S3 in the Electronic Supplementary Material. The most frequent complications involved lead or device repositioning or replacement, occurring in 8 patients (9%). Other complications included pneumothorax in 4 patients (5%), pocket haematoma in 3 patients (4%), ICD-related infection in 1 patient (1%), a thrombotic event in 1 patient (1%), cardiac perforation in 1 patient (1%), and VF during lead positioning in 1 patient (1%).

In our study cohort, 17% of patients (14 out of 85) received appropriate ICD therapy over a 5-year follow-up period, corresponding to an annual incidence rate of 3%. This rate is relatively modest compared with rates reported in the literature. For instance, a recent pooled analysis of five primary prevention ICD trials, which included patients with both ischaemic cardiomyopathy and NICM, found a 6% annual incidence of appropriate ICD therapy in NICM patients [17]. The lower rate observed in our study may be attributed to the exclusion of NICM patients with an increased risk of ventricular arrhythmia based on specific underlying aetiologies.

In contrast, the more contemporary DO-IT (Dutch Outcome in ICD Therapy) study reported a 5.9% cumulative incidence of appropriate ICD therapy over 2 years in NICM patients, which aligns more closely with our findings. This suggests that our cohort, which excluded higher-risk patients, exhibited an incidence rate that is consistent with recent studies that have employed more stringent patient selection criteria [16].

Our study population also demonstrated a low incidence of all-cause mortality, with an annual rate of 2%. This contrasts with findings from the Danish Study to Assess the Efficacy of ICDs in Patients with Non-Ischaemic Systolic Heart Failure on Mortality (DANISH), which reported an annual mortality rate of 4%. The DANISH study, which involved over 1000 patients with NICM and an LVEF of ≤ 35%, found no significant mortality benefit from the use of ICD therapy compared with medical therapy alone [5]. The implications of the DANISH study are particularly relevant in light of our findings. Given the low incidence of both mortality and appropriate ICD therapy in NICM patients in our study cohort, the potential benefits of ICD implantation may be limited. This suggests that ICD therapy should be considered primarily for carefully selected NICM patients who are at an above-average risk for ventricular arrhythmia. The findings underscore the importance of thorough risk stratification prior to ICD implantation to avoid unnecessary procedures and reduce the risk of adverse events and complications associated with device implantation.

Previous research has shown that the presence of LGE on CMR is a significant predictor of major ventricular arrhythmic events in patients with dilated NICM [9]. A recent meta-analysis including studies published between 2008 and 2022, with a median follow-up period of 3 years, reported a pooled odds ratio of 3.99 (95% CI 3.08–5.16), highlighting that LGE assessment may be a valuable parameter for primary prevention ICD implantation in NICM patients [18]. However, in our exploratory study, we found no significant association between the presence of LGE and appropriate ICD therapy, potentially due to the limited sample size and low event rate. Approximately half of the NICM patients in our study exhibited LGE on CMR, consistent with existing literature on LGE prevalence in this patient population [18]. Additionally, the prevalence of a midwall septal stripe in the LGE+ group aligned with current findings on LGE distribution in NICM patients [19]. Although specific LGE patterns such as midwall fibrosis have been linked to an increased risk of arrhythmias, our study did not observe a significant association between specific LGE distribution patterns and appropriate ICD therapy [20].

Therefore, our study does not provide conclusive evidence regarding the utility of LGE assessment as a primary risk stratification tool for ICD implantation in patients with NICM. Notably, 14% of patients without LGE on CMR still experienced appropriate ICD therapy, highlighting the necessity of closely monitoring patients who, under the new guidelines, do not receive an ICD. This finding also emphasises the need for large randomised trials to evaluate both the risk of SCD in NICM patients without LGE and the risk-benefit profile of prophylactic ICD implantation in those with LGE. Such studies are essential to establish more definitive guidelines for ICD therapy in this population. Furthermore, we encourage other centres to replicate our analysis to enable pooled data assessments, thereby allowing for a more comprehensive evaluation of arrhythmic risk and the role of LGE in NICM patients.

Our study has several limitations. First, in line with the retrospective and exploratory nature of this study on ICD therapy and all-cause mortality in NICM patients, the cohort size and event rate are modest. Second, LGE quantification was not performed because current quantification strategies are inconsistent and lack robust validation [21‐23]. Moreover, the threshold for LGE extent and its associated risk remains undetermined [24]. Consequently, assessing the presence or absence of LGE visually was deemed practically reasonable, as endorsed by national guidelines for ICD implantation in NICM patients. Thirdly, current guidelines recommend genetic testing for diagnosis and risk stratification in NICM [25]. However, this was not routinely performed in patients included in our analysis. Fourthly, CRT patients were excluded to ensure a homogeneous cohort, as CRT is a heart failure therapy that independently modifies arrhythmic risk and mortality, potentially confounding the analysis of ICD therapy events [26, 27]. Future studies should focus specifically on the CRT‑D population. Finally, heart failure therapy has significantly improved over the years, including the standard use of SGLT2 inhibitors and angiotensin receptor-neprilysin inhibitors, which significantly improve outcomes in heart failure patients [28]. This may have influenced the results of this study significantly.