Markers of Myocardial Damage Predict Mortality in Patients with Aortic Stenosis

Soongu Kwak, Russell J. Everett, Thomas A. Treibel, Seokhun Yang, Doyeon Hwang, Taehoon Ko, Michelle C Williams, Rong Bing, Trisha Singh, Shruti Joshi, Heesun Lee, Whal Lee, Yong-Jin Kim, Calvin W.L. Chin, Miho Fukui, Tarique A. Musa, Marzia Rigolli, Anvesha Singh, Lionel Tastet, Laura E. DobsonStephanie Wiesemann, Vanessa M Ferreira, Gabriella Captur, Sahmin Lee, Jeanette Schulz-Menger, Erik B Schelbert, Marie-Annick Clavel, Sung-Jin Park, Tobias Rheude, Martin Hadamitzky, Bernhard L. Gerber, David E Newby, Saul G Myerson, Phillipe Pibarot, João L Cavalcante, Gerry P McCann, John P Greenwood, James C. Moon, Marc R Dweck, Seung-Pyo Lee

Research output: Contribution to journalArticlepeer-review

Abstract / Description of output

Using machine learning, we identified four CMR markers of myocardial damage (ECV%, RVEF, LGE%, and LVEDVi) as major predictors of post-AVR mortality. Each parameter demonstrated a distinct non-linear association with mortality in the random survival forest model, most notably an ECV%>27% being associated with increased risk. These markers significantly improved risk prediction when added to the prediction model based on clinical risk factors and also showed effective risk stratification when combined into the AS-CMR risk score. The results were externally validated in a large independent cohort. These myocardial damage markers may offer major potential in optimizing the timing of AVR.
Original languageEnglish
JournalJournal of the American College of Cardiology
Early online date2 Aug 2021
Publication statusPublished - 10 Aug 2021

Keywords / Materials (for Non-textual outputs)

  • Aortic valve stenosis
  • magnetic resonance imaging
  • random survival forest


Dive into the research topics of 'Markers of Myocardial Damage Predict Mortality in Patients with Aortic Stenosis'. Together they form a unique fingerprint.

Cite this