Clinical risk score for cardiac death or heart failure hospitalization in moderate aortic stenosis

The association between moderate aortic stenosis (AS) and adverse cardiovascular outcomes is heterogeneous. Outcomes are likely dependent on clinical factors but no formal means of integrating these variables has been defined. This study aims to develop and validate a risk score to predict 5-year cardiac mortality or heart failure (HF)-related hospitalization in moderate AS. This was a retrospective cohort study that included patients diagnosed with moderate AS. Patients with aortic valve intervention or severe AS were excluded at baseline and censored at follow up. Multivariable Cox proportional hazard model with LASSO penalty followed by a greedy selection algorithm was used to derive a risk score, which was then externally validated for predicting the 5-year composite risk of cardiac mortality or HF-related hospitalization. The derivation cohort included 2,212 patients with moderate AS (mean age 73.4±11.0 years, 65.7% male) with median follow-up of 4.3 years (interquartile range: 1.7-5). The top 10 variables included in the risk score included 6 echocardiographic variables (left ventricular (LV) end-diastolic diameter, LV outflow tract velocity-time integral, E-wave, end-diastolic left ventricular posterior wall thickness and moderate/severe mitral regurgitation, moderate/severe tricuspid regurgitation) and 4 clinical variables (age, diastolic blood pressure, acute coronary syndrome, hyperlipidemia). The C-statistics for the score were 0.70 (95% CI: 0.67–0.76) in the internal validation dataset and 0.75 (95% CI: 0.70–0.79) in the external validation dataset (n = 1,141), demonstrating good predictive performance. This moderate AS risk score, based on demographic and clinical features, as well as conventional echocardiographic parameters, predicts outcomes in patients with moderate AS. This quantification of risk may help with shared decision-making about possible interventions, planning the frequency of follow-up, and selecting candidates for potential randomized trials in this heterogeneous population.