BACKGROUND

Risk stratification for ventricular arrhythmias (VA) and sudden death in nonischemic dilated cardiomyopathy (DCM) remains suboptimal.

OBJECTIVES

The goal of this study was to provide an improved risk stratification algorithm for VA and sudden death in DCM.

METHODS

This was a retrospective cohort study of consecutive patients with DCM who underwent cardiac magnetic resonance with late gadolinium enhancement (LGE) at 2 tertiary referral centers. The combined arrhythmic endpoint included appropriate implantable cardioverter-defibrillator therapies, sustained ventricular tachycardia, resuscitated cardiac arrest, and sudden death.

RESULTS

In 1,165 patients with a median follow-up of 36 months, LGE was an independent and strong predictor of the arrhythmic endpoint (hazard ratio: 9.7; p < 0.001). This association was consistent across all strata of left ventricular ejection fraction (LVEF). Epicardial LGE, transmural LGE, and combined septal and free-wall LGE were all associated with heightened risk. A simple algorithm combining LGE and 3 LVEF strata (i.e., ≤20%, 21% to 35%, >35%) was significantly superior to LVEF with the 35% cutoff (Harrell's C statistic: 0.8 vs. 0.69; area under the curve: 0.82 vs. 0.7; p < 0.001) and reclassified the arrhythmic risk of 34% of patients with DCM. LGE-negative patients with LVEF 21% to 35% had low risk (annual event rate 0.7%), whereas those with high-risk LGE distributions and LVEF >35% had significantly higher risk (annual event rate 3%; p = 0.007).

CONCLUSIONS

In a large cohort of patients with DCM, LGE was found to be a significant, consistent, and strong predictor of VA or sudden death. Specific high-risk LGE distributions were identified. A new clinical algorithm integrating LGE and LVEF significantly improved the risk stratification for VA and sudden death, with relevant implications for implantable cardioverter-defibrillator allocation.