Development and validation of the Aortic valve Stenosis Guarded by Amplified Risk Determination (ASGARD) risk score for safe monitoring in asymptomatic non-severe aortic valve stenosis

Abstract

Abstract Aims Current guidelines recommend serial echocardiography at minimum 1–2-year intervals for monitoring patients with non-severe aortic valve stenosis (AS), which is costly and often clinically inconsequential. We aimed to develop and test whether the biomarker-based ASGARD (Aortic valve Stenosis Guarded by Amplified Risk Determination) risk score can guide the timing of echocardiograms in asymptomatic patients with non-severe AS. Methods and results The development cohort comprised 1093 of 1589 (69%) asymptomatic patients with mild-to-moderate AS who remained event-free one year after inclusion into the SEAS trial. Cox regression landmark analyses with a 2-year follow-up identified the model (ASGARD) with the lowest Akaike information criterion for association to AS-related composite outcome (heart failure hospitalization, aortic valve replacement, or cardiovascular death). Fine–Gray analyses provided cumulative event rates by ASGARD score quartiles. The ASGARD score was internally validated in the remaining 496 patients (31%) from the SEAS cohort and externally in 71 asymptomatic outpatients with non-severe AS from six Copenhagen hospitals. The ASGARD score comprises updated measurements of heart rate and age- and sex-adjusted N-terminal pro-brain natriuretic peptide upon transaortic maximal velocity (Vmax) from the previous year. The ASGARD score had high predictive accuracy across all cohorts (external validation: area under the curve: 0.74 [95% CI, 0.62–0.86]), and similar to an updated Vmax measurement. An ASGARD score ≤ 50% was associated with AS-related event rates ≤ 5% for a minimum of 15 months. Conclusion The ASGARD score could provide a personalized and safe surveillance alternative to routinely planned echocardiograms, so physicians can prioritize echocardiograms for high-risk patients.