Cardiac magnetic resonance imaging aids in differentiating dilated cardiomyopathy from the athlete's heart

Abstract Background Distinguishing mild dilated cardiomyopathy (DCM) from the athletes' heart is complex due to similar increases in indexed left ventricular end-diastolic volumes (LVEDVi) and decreases in ejection fractions (LVEF). Purpose To determine if Cardiac Magnetic Resonance (CMR) Imaging features can differentiate genotype- and CMR phenotype-positive DCM patients from LVEDVi and LVEF matched elite athletes. Methods We matched healthy, elite athletes from the ELITE cohort to genotype-positive DCM patients from the Amsterdam DCM registry with a 1:1 ratio. Selection criteria for DCM were LVEDVi exceeding the sex-specific upper limit of normal or reduced LVEF ≥ 40%. Matching was based on sex, LVEDVi, and LVEF using the smallest propensity score distance. Athletes with cardiovascular disease or non-hinge-point (non-HP) late gadolinium enhancement (LGE) were excluded. DCM patients with a (likely) pathogenic variant with non-DCM phenotypes (ARVC, HCM) were excluded. A multivariate backward elimination logistic regression model (1500 bootstraps) was used to differentiate DCM from the athlete’s heart, validated by sensitivity/specificity and area under the curve (AUC) analysis. Results We included 137 DCM patients (54.7% female, primary (likely) pathogenic variant: PLN 38.7%; TTN 23.4%; LMNA 7.3%, MYH7 6.6%; FLNC 5.8%, other 18.2%) and matched these to 137 elite athletes (51.1% female). DCM patients were older than elite athletes (41.2 years ±15.2 vs 27.9 years ±7.9, p < .001). Despite matching, DCM patients had a lower LVEF (51.3% [47.0, 55.7] vs 53.3% [50.8, 56.0], p = .002) and LVEDVi (101.6 ml/m2 [90.2, 116.0] vs 107.9 ml/m2 [97.6, 120.0], p < .001) than athletes. DCM patients had larger max left atrial volumes (LAV) (37.0 ml/m2 [30.8, 45.6] vs 32.8 ml/m2 [27.0, 40.8], p < .001) but smaller max right atrial volumes (RAV) (41.6 ml/m2 [33.2, 47.9] vs 55.7 ml/m2 [46.5, 66.2], p < .001). DCM patients had lower indexed LV mass (LVMi) (43.8 g/m2 [26.3, 74.2] vs 50.5 g/m2 [25.2, 91.2], p < .001. DCM patients less frequently showed HP-LGE than elite athletes (7% vs 45%, p < .001); 33% of DCM patients had non-HP LGE. The multivariate logistic regression model including age, sex, LVEDV, LVM, HP-LGE, max LAV and RAV, had an AUC of 0.966 (95% CI 0.947-0.986) with a sensitivity of 86% and a specificity of 96%. After excluding DCM patients with non-HP LGE and rematching, we retained 88 DCM patients and 88 elite athletes with similar LVEF (51.7% [47.9, 56.1] vs 52.5% [50.1, 55.9], p = 0.483) and LVEDVi (102.2 ml/m2 [90.1, 117.2] vs 108.0 ml/m2 [95.4, 121.4], p = 0.122). A multivariate logistic regression model including age, sex, LVEDV, LVM, max LAV and RAV, had an AUC of 0.936 (95% CI 0.890-0.967) with a sensitivity of 79% and a specificity of 95%. Conclusion Our findings highlight a limited number of parameters, easily obtained with CMR Imaging, which can differentiate between DCM and the athlete’s heart, with an excellent AUC in internal validation.