To enhance risk stratification for sudden cardiac death (SCD), the European Society of Cardiology (ESC) has recently added the extent of late gadolinium enhancement (LGE) using cardiac MRI in the guidelines, setting a threshold at ≥ 15% of left ventricular mass. While previous studies have showed the prognostic value of LGE extent to predict all-cause mortality, the prognostic impact of additional LGE features is not well established.

We aimed to assess the incremental prognostic value of the granularity of LGE using cardiac MRI including extent, location, and pattern in patients with HCM to predict all-cause death.

Between 2008 and 2021, all patients referred for HCM assessment using cardiac MRI, without history of coronary artery disease (CAD) or clinical history of myocarditis were prospectively recruited in two French centers. The outcome was all-cause death using the French National Registry of Death. The concept of “LGE granularity” was defined as a model combining LGE extent, location, and pattern. Using nested Cox proportional hazard models, the additional predictive value was assessed by C-statistic increment, continuous net reclassification improvement (NRI), integrative discrimination index (IDI) and global Chi².

Among 2672 included patients (52±7 years, 56% males), 862 (32%) had LGE. After a median (IQR) follow-up of 9 (7–11) years, 447 (17%) patients died. The presence of LGE was strongly associated with the risk of all-cause death (log-rank P<0.001, Fig. 1A). Even after adjustment for known prognosticators, the presence of LGE was still associated with all-cause death (adjusted hazard ratio (HR) 3.96, 95% CI: 3.26–4.80, P<0.001). In the subgroup of patients with LGE (n=862), survival curves showed that the “LGE granularity model” was associated with a higher risk of all-cause death (all P<0.001, Fig. 1B). A nested Cox model adjusted on known prognosticators showed that the LGE extent, location and pattern were all independently associated with all-cause death (all P<0.001, Fig. 1C). The “LGE granularity model” combining all these LGE features showed the best improvement in model discrimination and reclassification over and above known prognosticators (C-statistic improvement: 0.90; NRI=41.9%; IDI=13.2%, Chi² global=450, all P<0.001; LR-test P<0.001).

In a large cohort of HCM, LGE granularity model (extent, location, and pattern of LGE) had an incremental prognostic value above traditional prognosticators to predict all-cause death.