Markers of Myocardial Damage Predict Mortality in Aortic Stenosis
- Among patients with aortic stenosis (AS) who undergo SAVR or TAVR, data from CMR added to clinical and echocardiographic data in predicting mortality after intervention.
- Increased mortality was associated with ECV% >27%, LGE >2%, LVEDVi ≤55 or >80 ml/m2, and RVEF ≤50 or >80%.
- Data from CMR might help in the determination of the optimal timing of intervention for AS.
What are the prognostically important markers on cardiovascular magnetic resonance (CMR) imaging among patients with aortic stenosis (AS) undergoing aortic valve replacement (AVR)?
A cohort of 440 patients with severe AS and planned surgical (SAVR) or transcatheter AVR (TAVR) prospectively enrolled at 10 international sites was used to form a derivation cohort in whom CMR was performed shortly before AVR. Using machine learning, a random survival forest model was built using 29 variables (12 clinical, four from echocardiography, 13 from CMR) with post-AVR death as the outcome. The model was tested in an external validation cohort of 359 patients at four international sites.
There were 52 deaths in the derivation cohort (age 70 ± 10 years, 58.9% men, 32.7% bicuspid aortic valve, 6.6% low-flow low-gradient AS; 71% SAVR, 14% SAVR plus bypass, 15% TAVR; median follow-up 3.8 years [IQR 2.9-4.6 years]) and 51 deaths in the validation cohort (age 73 years; 30.6% TAVR; median follow-up 3.3 years (IQR 1.4-4.9 years]). The four most predictive CMR markers were extracellular volume fraction (ECV%), late gadolinium enhancement (LGE), left ventricular end-diastolic volume index (LVEDVi), and right ventricular ejection fraction (RVEF). Across the whole cohort and in asymptomatic patients, risk-adjusted predicted mortality increased strongly once ECV% exceeded 27%, while LGE >2% showed persistent high risk. Increased mortality also was observed with both large (LVEDVi >80 ml/m2) and small (LVEDVi ≤55 ml/m2) LVs, and with high (>80%) and low (≤50%) RVEF. The predictability was improved when these four CMR markers were added to clinical factors (3-year C-index 0.778 vs. 0.739). The prognostic thresholds and risk stratification by CMR variables were reproduced in the validation cohort.
Machine learning identified myocardial fibrosis and biventricular remodeling markers as the top predictors of survival among patients with AS who undergo AVR, with nonlinear associations with mortality. The authors concluded that these markers may have potential in optimizing decisions regarding the timing of AVR.
Diffuse myocardial fibrosis using T1 mapping (ECV%) and replacement fibrosis using LGE both are known to be associated with prognosis among patients with AS; however, ECV% and LGE are not independent, and may be associated with other clinical and imaging factors that are related to prognosis. Machine learning is a method of data analysis that automates analytical model building, and can assess the predictive hierarchy of variables. This study, using both a derivation cohort and a validation cohort, identified four CMR variables (ECV% >27%, LGE >2%, LVEDVi ≤55 or >80 ml/m2, and RVEF ≤50 or >80%) associated with post-AVR mortality, with incremental predictive power of those variables added to clinical and echocardiographic data. With the premise that existing clinical criteria (symptoms, decreased LVEF) are imperfect for the prediction of mortality after intervention, this study reinforces that data from CMR might help in the determination of the optimal timing of intervention for AS.
Clinical Topics: Cardiac Surgery, Cardiovascular Care Team, Geriatric Cardiology, Heart Failure and Cardiomyopathies, Invasive Cardiovascular Angiography and Intervention, Noninvasive Imaging, Valvular Heart Disease, Aortic Surgery, Cardiac Surgery and Heart Failure, Cardiac Surgery and VHD, Interventions and Imaging, Interventions and Structural Heart Disease, Echocardiography/Ultrasound
Keywords: Aortic Valve Stenosis, Artificial Intelligence, Cardiac Surgical Procedures, Cardiomyopathies, Contrast Media, Diagnostic Imaging, Echocardiography, Fibrosis, Gadolinium, Geriatrics, Heart Valve Diseases, Heart Valve Prosthesis, Magnetic Resonance Spectroscopy, Risk Assessment, Stroke Volume, Transcatheter Aortic Valve Replacement, Ventricular Function, Right
< Back to Listings