Validation of the developed ASGARD risk score for safe monitoring of asymptomatic patients with non-severe aortic valve stenosis

Abstract Background The recommended standard monitoring of patients with non-severe aortic valve stenosis (AS) involves echocardiography surveillance at 1-2 year intervals. Since many of these costly routine echocardiograms yield no clinical consequences during a prolonged watchful waiting period, we propose to replace them with a risk stratification approach. Purpose The primary aim was to validate the ability of ASGARD risk score to a) identify low-risk AS patients who do not need new echocardiography for safe monitoring, and b) estimate optimal follow-up intervals. Methods We have developed the ASGARD risk score using data from 1196/1739 (69%) asymptomatic AS-patients with mild-to-moderate AS who were enrolled in the randomized, multicentre Simvastatin and Ezetimibe in Aortic Stenosis (SEAS) trial. The primary outcome was a composite of AS-related outcomes, aortic valve replacement, or hospitalization with heart failure during a clinically relevant 2-year follow-up. From multivariable Cox proportional hazards regression analyses, we chose the best prognostic model with the lowest Akaike information criterion (AIC) values and p-values <0.05. The risk score included five predictors: age, sex, systolic blood pressure, N-terminal pro-brain natriuretic peptide (NT-proBNP), and last year's measurement of transaortic maximal velocity (Vmax). Kaplan-Meier plot visualized event rates stratified by quartiles of risk score. We validated the ASGARD risk score in an internal validation cohort of 543 patients (31% hold-out from SEAS-population) and an external validation cohort of 69 asymptomatic out-clinic patients with Vmax ≤4 m/s from six hospitals from capital city. Results Patients from the external validation cohort significantly differed from those in the development cohort in terms of higher age (mean [SD], 71.7 [8.5] vs. 67.4 [9.8] years, p<0.001), less prevalent left ventricular hypertrophy (13.0% vs. 37.5%, p<0.001), and higher event-rates (12.5 [95% CI, 7.6-20.8] vs. 4.5 [3.7-5.4] events per 100 patient-years of follow-up, p<0.001). The ASGARD risk score showed consistent predictive performance across the validation cohorts (external validation: area under the curve: 0.82 [95% CI, 0.73-0.91]; calibration-in-the-large, p=0.24; calibration slope, p=0.15; Brier score, 0.18). The score performed just as well or better than a new Vmax measurement (Figure 1A-1B). In the three upper ASGARD score quartiles, 95% of patients had less than 5% risk of AVR or heart failure until 7.5, 16.5, and 22.5 months, respectively (Figure 1C). Conclusion Without a new echocardiogram, the ASGARD risk score identifies patients at low risk for AS-related events among asymptomatic patients with non-severe AS. The ASGARD risk score also estimates the safe watchful-waiting period of 6, 15, 21, and 24 months for the four ASGARD risk score quartiles in decreasing order.Figure 1