Abstract 295: Aortic Valve Calcium Scores Are Strong Predictors of Aortic Stenosis in a Population-Based Setting: The AGES-Reykjavik Study
BACKGROUND: The clinical utility of computed tomography (CT) derived aortic valve calcium (AVC) scores for predicting aortic stenosis (AS) has yet to be evaluated in an unreferred, population-based setting.
METHODS: The ability of AVC scoring to predict AS was assessed in the 748 members of the Age, Gene-Environment Susceptibility (AGES)-Reykjavik Study who had both continuous wave Doppler echocardiography and CT within a 48 hour period. AS was defined as a peak transvalvular jet velocity >2.5 m/s, and subclassified as mild (2.5–3.0 m/s) or moderate-to-severe (mod-severe, >3.0 m/s) AS. The discriminatory ability of AVC scoring (by Agatston method) was then evaluated using receiver operating characteristic (ROC) curves and bootstrapped C-statistics.
RESULTS: Among this sub-cohort [57% female, aged 76±6 (range 67–95) years], 336 subjects had prevalent AVC (44.9%), 28 subjects had AS (3.7%), and 9 subjects had mod-severe AS (1.2%). Those with AS had higher AVC scores than those without (1030 vs. 95 Agatston units, p<0.0001), as did those with mod-severe vs. mild AS (2032 vs. 945, p=0.002). The area under the ROC curves for AS and mod-severe AS were 0.96 (95%CI: 0.88,1.00) and 0.99 (95%CI: 0.97,1.00) respectively (Figure 1⇓). An AVC score of 265 carried 96% sensitivity and 92% specificity for detecting AS, while a score of 1127 carried 100% sensitivity and 98% specificity for mod-severe AS.
CONCLUSIONS: CT-determined AVC scores strongly predict AS in a population setting. For epidemiologic investigations that phenotype calcific aortic valve disease by CT, AVC score cutoffs of 265 and 1127 demonstrate high sensitivity and specificity for identifying AS and mod-severe AS respectively.
This research has received full or partial funding support from the American Heart Association, Pacific/Mountain Affiliate (Alaska, Arizona, Colorado, Hawaii, Idaho, Montana, Oregon, Washington & Wyoming).