Incremental prognostic value of LGE granularity risk score in non-ischaemic dilated cardiomyopathy

Abstract Introduction The prognostic stratification of non-ischaemic dilated cardiomyopathy (DCM) to predict the risk of death is mainly driven by the left ventricular ejection fraction (LVEF). Left ventricular ejection fraction (LVEF) ≤35% is currently the only indication for primary prevention with an implantable cardioverter-defibrillator (ICD) in patients with non-ischaemic dilated cardiomyopathy (DCM), despite several limitations. Beyond LVEF, several cardiovascular magnetic resonance imaging (CMR) studies have suggested the strong prognostic impact of late gadolinium enhancement (LGE). The LGE granularity risk score – including extent, location and number of areas – could improve the risk stratification in DCM patients. Objective To assess whether a novel LGE granularity risk score improves risk prediction of all-cause mortality in DCM patients, beyond traditional prognostic factors and LVEF. Methods Between 2008 and 2021, consecutive DCM patients (LVEF 50%) without prior ICD or sustained ventricular arrhythmia were included from two centres: centre 1 (N=1,452, training cohort) and centre 2 (N=216, test cohort). LGE granularity risk was defined as high-risk if at least one of the following was present: (1) LGE extent 3 segments, (2) septal LGE location, or (3) LGE in multiple areas; otherwise, patients were classified as low-risk. All-cause mortality was the primary outcome, collected from the French National Death Registry. Survival analysis was conducted in the training cohort; prognostic model performance was assessed in the test cohort. Results A total of 1,668 patients were included (age 52 ± 8 years, 54% male). Median follow-up was 9 (7–12) years. In the training cohort, high-risk LGE granularity was strongly associated with higher all-cause mortality (HR: 10.2, 95% CI: 7.9–13.3, p0.001) compared to low-risk. Similarly, in the test cohort, high-risk LGE granularity patients had a higher risk of all-cause mortality (HR: 3.76, 95% CI: 2.2-6.5, p0.001) compared to low-risk. In the test cohort, adding the LGE granularity risk score combining all these LGE features showed the best improvement in model discrimination over traditional prognosticators, including LVEF (chi-square increased from 39 to 47, p=0.006; C-statistic improved from 0.73 to 0.76). Conclusion The LGE granularity score, integrating extent, location, and number of LGE areas, was independently associated with mortality and improved prognostic accuracy beyond LVEF and traditional prognostic factors. Its performance was validated across two independent centres, supporting the need for a prospective trial to validate CMR-based risk stratification for DCM.Survival curves of all-cause mortality Discrimination assessment