Abstract 4147317: Impact of the Addition of Cardiac MRI Features and Sex Interactions with a Validated Clinical Risk Score to Improve Discrimination of Risk

Introduction: The “Meta-analysis Global Group in Chronic Heart Failure” (MAGGIC) score is a validated clinical score, but does not include quantitative insights provided by cardiac magnetic resonance (CMR). We sought to determine if CMR parameters result in further discrimination of risk prediction. Methods: We evaluated 1414 ischemic and non-ischemic cardiomyopathy (ICM vs NICM) patients, who were referred for CMR between 2001-2017. The primary endpoint was the composite event of death, transplant, or LVAD. The data was split equally into development and validation cohorts using stratified randomization with presence/absence of an event and ICM/NICM as the two strata. A Cox proportional hazards regression model was first developed in the training patient cohort using only the MAGGIC score alone (baseline model). CMR measures with associated interactions with sex were then added to the baseline model in addition to ICM vs NICM categorization. The final fitted model was tested on a validation cohort of 704 patients. Uno concordance index was estimated with Wald test to assess significance. Results: There were 782 ICM patients and 674 NICM patients, with the following characteristics: 31.8% female, mean age 58±14 years, LVEDVi 131±45.0 mL/m 2 , LVEF 30±11%, LAEF 38±17%, RVEF 43±13%, LGE 8±11% and mitral regurgitant fraction 15±15%. The final model with CMR parameters is described in table 1. The MAGGIC score was a significant predictor of outcomes with a c-index of 0.635±0.0190, which increased to 0.656±0.0154 after adding CMR features and cardiomyopathy classification in the validation cohort ( p<0.001 for c-index difference ). Interaction of male sex and LVEDVi and RVEF yielded a maximum likelihood estimation (MLE) of -0.0077 (p=0.003) and 0.016 (p=0.023) respectively, suggesting attenuated impact of increasing LVEDVi and declining RVEF on risk in men compared to women. Conclusion: Integration of CMR parameters with novel sex interactions with the MAGGIC score significantly improves discrimination of risk. Further studies are needed to validate this CMR-enhanced risk score and to determine its association with adverse outcomes and differential therapeutic benefit in external patient cohorts.