Predicting Sustained Ventricular Arrhythmias in Dilated Cardiomyopathy: A Meta-Analysis and Systematic Review

Authors, Journal, Affiliations, Type, DOI

• Authors: Arjan Sammani, Elham Kayvanpour (co-first authors), Laurens P. Bosman, Farbod Sedaghat-Hamedani, Tanja Proctor, Weng-Tein Gi, Alicia Broezel, Katrin Jensen, Hugo A. Katus, Anneline S.J.M. te Riele, Benjamin Meder, Folkert W. Asselbergs (co-senior authors)
• Journal: ESC Heart Failure
• Affiliations: UMC Utrecht / Utrecht University; University Hospital Heidelberg; DZHK (German Centre for Cardiovascular Research); University of Heidelberg Institute of Medical Biometry; Stanford Genome Technology Center; UCL Institute of Cardiovascular Science; Health Data Research UK
• Type: Systematic review and meta-analysis
• DOI: https://doi.org/10.1002/ehf2.12689

Overview

A systematic review and meta-analysis of 55 studies (11,451 patients, mean follow-up 3.7±2.3 years) examining predictors of sustained ventricular arrhythmias (VA) in non-ischaemic dilated cardiomyopathy. The crude annual event rate was 4.5%. Pooled analyses identified LGE (HR 5.55), prior sustained VA (HR 4.15), hypertension (HR 1.95), LVEF on echocardiography (HR 1.45 per 10% decrease), and LV dilatation (HR 1.10) as significant predictors. Genetic mutations in PLN, LMNA, and FLNC were associated with VA in non-pooled analyses. The authors conclude that LVEF alone is insufficient for ICD risk stratification in DCM and call for a multi-parameter DCM-specific risk calculator analogous to HCM Risk-SCD.

Keywords

Dilated cardiomyopathy; Sudden cardiac death; Implantable cardiac-defibrillator; Prognosis; Risk

Key Takeaways

Introduction

• DCM patients are at elevated SCD risk; LVEF ≤35% has been the dominant ICD selection criterion, but the DANISH trial showed no all-cause mortality benefit in non-ischaemic DCM specifically.
• An updated meta-analysis of ICD trials confirmed ICD efficacy but concluded LVEF-based recommendations from ischaemic disease cannot be straightforwardly extrapolated to non-ischaemic DCM.
• Prior systematic reviews combined arrhythmic and heart failure outcomes, limiting SCD-specific conclusions — this study focused exclusively on sustained VA as the primary endpoint.

Methods

• MEDLINE/Embase search February 2018, updated January 2020.
• 1,996 unique citations screened → 55 studies included (29 with HRs eligible for meta-analysis).
• Study quality assessed with the Quality in Prognostic Studies (QUIPS) tool independently by two investigators.
• Only unadjusted HRs with uniform predictor definitions were pooled in random-effects meta-analysis (R, meta package). Studies reporting only ORs were included in systematic review only.
• Primary outcome: sustained VA = spontaneous sustained VT, VF, resuscitated SCD, or appropriate ICD intervention.

Quality Assessment

• Highest risk of bias: study attrition, limited confounder adjustment, statistical analysis limitations.
• Overall quality of evidence: moderate.
• Studies span 1992–2019; HF treatment has changed substantially over this period, limiting comparability.

Results — Demographics & History

• Age per 10-year increase: HR 0.82 [0.74–1.00] — younger age is protective, consistent with other cardiomyopathies and primary arrhythmia syndromes. May reflect faster conduction velocity and lower Ca²⁺ handling thresholds in younger hearts.
• Male sex: HR 1.51 [0.96–2.37] — non-significant trend; different from ACM where male sex is a strong predictor; sex-based Ca²⁺ handling and QT differences may contribute.
• NYHA class III/IV vs I/II: HR 1.37 [0.77–2.46] — non-significant; significant heterogeneity (I²=65%).
• Hypertension: HR 1.95 [1.26–3.00] — significant; mechanism may involve hypertension-induced myocardial fibrosis and LV remodelling. Two of four studies pooled.
• Family history of DCM: HR 0.90 [0.52–1.54] — non-significant.
• Prior sustained VA: HR 4.15 [1.32–13.02] — strongly significant but high heterogeneity (I²=93%). All four pooled studies and three of six non-pooled studies confirmed this association. This is the secondary prevention population who should receive an ICD per guidelines.
• Syncope: non-significant (only two studies, no HRs available).

Results — Genetics

• LMNA and PLN mutations significantly associated with VA in a previously published meta-analysis (Kayvanpour et al., Clin Res Cardiol 2017); OR data, not pooled in current analysis.
• FLNC truncating mutations: three cohort studies reported frequent premature SCD and VA in 82% of participants.

Results — Electrophysiology

• Atrial fibrillation: HR 1.44 [0.77–2.70] — non-significant.
• QRS duration per 10 ms: HR 1.12 [0.95–1.32] — non-significant; significant heterogeneity (I²=84%).
• QRS >110 ms: HR 0.84 [0.56–1.25] — non-significant; direction opposite to QRS per 10 ms analysis.
• Fragmented QRS (fQRS): HR 4.11 [0.40–42.41] — non-significant; wide CI; only two studies.
• LBBB: HR 1.05 [0.53–2.09] — non-significant; significant heterogeneity (I²=81%).
• Non-sustained VT (nsVT): Not pooled (no uniform HRs); 9 of 14 studies showed significantly increased arrhythmic risk (P ≤ 0.05) — a consistent non-pooled signal.
• HRV (SDNN): Mixed results; 3 of 6 studies significant. Unstandardized definitions limit pooling.
• T-wave alternans (TWA): HR 6.5 [2.46–17.14] — significant; remaining three studies confirmed. Reflects autonomic dysfunction. Clinical utility limited by poor standardization.
• Signal-averaged ECG: Non-significant across three studies.

Results — Imaging

• LVEF per 10% decrease (all CMR + echo methods pooled): HR 1.30 [0.98–1.71] — non-significant (I²=58%).
• LVEF per 10% decrease (echocardiography only): HR 1.45 [1.19–1.78] — significant; little heterogeneity (I²=0%).
• LVEF <30%: HR 1.52 [0.78–2.97] — non-significant; very high heterogeneity (I²=96%).
• LVEF <35%: HR 2.4 [0.81–7.69] — non-significant across three pooled studies.
• LGE presence: HR 5.55 [4.02–7.67] — strongly significant; all eight studies pooled directed towards the same effect; specific LGE patterns not assessable due to non-uniform reporting.
• LGE per 5% LV mass: HR 1.61 [0.90–2.93] — non-significant; high heterogeneity (I²=91%).
• LGE combined with QRS ≥120 ms: HR 9.53 [2.84–31.98] — one study only; potential additive value.
• LVEDV/LVESV per 10 mL/m² (cMRI): HR 1.10 [1.10–1.11] and HR 1.10 [1.00–1.22] — significant (small effect size).
• Global longitudinal strain per 1% increase: HR 1.21 [0.97–1.51] — non-significant; high heterogeneity (I²=82%).

Results — Biomarkers

• BNP, ANP, eGFR, norepinephrine, potassium — not significant for arrhythmic risk.
• Plasma creatinine per μmol/L: HR 1.01 [1.00–1.02] — small but significant.
• GDF-15 per 1 SD increase: HR 2.1 [1.1–4.3] — significant in one study.

Discussion

• Promising risk factors for a future DCM-VA risk model: younger age, hypertension, prior (non-)sustained VA, decreased LVEF (echo), LV dilatation, LGE, and genetic mutations in PLN, LMNA, and FLNC.
• Hypertension's association with VA may represent an opportunity for risk modification through antihypertensive therapy — not yet investigated.
• PLN dysregulates SERCA2a/Ca²⁺ cycling → DADs/EADs → arrhythmogenicity.
• LMNA mutations destabilize the nuclear envelope and interact with cytoskeleton/extracellular matrix.
• FLNC truncating mutations → DCM/ACM phenotype overlap with LV dysfunction and frequent VA.
• TWA has strong HR but poor real-world utility due to standardization challenges.
• LVEF alone is insufficient: the current ICD threshold (LVEF ≤35%) was derived from ischaemic disease and may miss high-risk non-ischaemic DCM patients with LVEF 36–50%.
• LGE (fibrosis) and LV volumes provide additive prognostic information beyond LVEF.
• Future models should: (i) incorporate all promising risk factors, (ii) include sex, (iii) use competing risk analysis (VA vs. HF-related outcomes compete), and (iv) be analogous to HCM Risk-SCD and ARVC risk calculators.

Limitations of the document

• Only unadjusted HRs pooled — adjusted risk estimates not available due to variable confounder sets; crude HRs may over- or underestimate true associations.
• Moderate quality of evidence; heterogeneity moderate to high for most predictors.
• Studies span 1992–2019; HF management (HFrEF pharmacotherapy, device therapy) has changed substantially — event rates may differ in contemporary cohorts.
• Events per variable often <10 in included studies, limiting statistical power and confounder adjustment in original studies.
• LGE localization patterns not evaluable due to non-uniform reporting across studies.
• Competing risks (VA vs. HF outcomes) not modelled; crude HRs may misrepresent arrhythmic risk in patients also at risk for HF-related death.
• ICD use in 18/29 meta-analysed studies may underestimate true VA burden (treated events replace fatal VA).

Key Concepts Mentioned

• concepts/Late-Gadolinium-Enhancement — strongest single pooled predictor of VA in DCM (HR 5.55)
• concepts/Sudden-Cardiac-Death — primary prevention target; annual DCM VA event rate 4.5%
• concepts/VA-Risk-Stratification-DCM — this paper provides key evidence base for the concept

Key Entities Mentioned

• entities/DCM — primary subject; multi-parameter risk stratification needed for ICD candidacy
• entities/LMNA — significantly associated with VA; non-pooled but meta-analysed by reference 12
• entities/PLN — significantly associated with VA; non-pooled; 82% VA/SCD in FLNC cohort
• entities/FLNC — truncating variants: 82% VA/SCD rate in cohort studies

Wiki Pages Updated

• wiki/sources/VA-DCM-Sammani-2020.md — created
• wiki/concepts/VA-Risk-Stratification-DCM.md — created
• wiki/entities/DCM.md — updated with pooled HR data, annual event rate, multi-parameter model case
• wiki/entities/LMNA.md — updated with meta-analysis VA predictor data
• wiki/entities/PLN.md — updated with meta-analysis VA predictor data
• wiki/entities/FLNC.md — updated with 82% VA/SCD figure and meta-analysis context
• wiki/concepts/Late-Gadolinium-Enhancement.md — updated with pooled HR 5.55 in DCM
• wiki/concepts/Sudden-Cardiac-Death.md — updated with DCM annual VA event rate 4.5%
• wiki/index.md — updated
• wiki/log.md — appended