Abstract
Funding Acknowledgements
Type of funding sources: None.
Background
The calcium score of the aortic valve (CaScAoV) is now recommended as a supporting tool to assist in the grading of aortic stenosis (AS) severity when echocardiographic assessment is inconclusive. However, the proposed CaScAoV cut-offs for considering severe AS "unlikely", "likely", or "very likely" have never been validated in Portuguese cohorts.
Aim
The purpose of this study was to assess the performance of the proposed CaScAoV cut-offs in identifying patients with severe aortic stenosis.
Methods
A total of 513 consecutive patients (median age 83 years [IQR 79–87], 38% males) evaluated at a single-centre TAVI-programme between Jan/2016 and Nov/2019 were retrospectively identified. Only patients with an ECG-gated cardiac computed tomography (CT) and a transthoracic echocardiography performed within a 6-month time-frame were included. Main exclusion criteria were left ventricular ejection fraction < 50%, indexed stroke volume < 35 ml/m2, previous valve surgery and
bicuspid aortic disease. CaScAoV was measured according to the Agatston method (Agatston units – AU). As previously reported, the likelihood of aortic stenosis as assessed by CT was categorized as: "very likely" (>3000 AU for men, >1600 AU for women); "likely" (>2000 AU for men, >1200 AU for women) ; or unlikely (<1600 AU for men, <800 AU for women). Diagnostic tests performance measures were calculated for each category. Separate analyses were performed for each gender.
Results
Severe AS (mean gradient ≥ 40 mmHg) was present in 422 patients (overall 82.3%: 83.1% in females and 80.8% in males), with a median transvalvular gradient of 49 mmHg (IQR 42 – 60).
Overall, the discriminative ability of the CaScAoV to distinguish severe from non-severe AS was higher in men when compared with women (c-statistic 0.86 [95%CI 0.80 – 0.93] vs. 0.72 [95%CI 0.64 – 0.80], p for comparison < 0.001). In males, the "very likely" cut-off had a sensitivity of 71% (95%CI 63 – 78%), a specificity of 81% (95%CI 65 – 92%), a positive predictive value (PPV) of 94% (95%CI 89 – 97%) and a negative predictive value (NPV) of 40% (95%CI 33 – 46%) for the diagnosis of severe AS. Conversely, in women the sensitivity was 75% (95%CI 69 – 80%), specificity was 57% (95%CI 43 – 71%), PPV was 90% (95%CI 86 – 92%) and NPV was 32% (95%CI 25 – 39%).
On the other end of the spectrum, the "unlikely" cut-off showed poor performance in dismissing severe AS, particularly in females – NPV of 43% (95%CI 25-63%) in women vs. 83% (95%CI 63-93%) in men.
Conclusion
In our population, the discriminative power of CaScAoV for identifying patients with severe AS was lower than in previously published cohorts, particularly in females. While very high CaScAoV is strongly supportive of severe AS, caution should be employed when interpreting low CaScAoV values in women, since the recommended cut-off value does not allow the safe exclusion of severe aortic stenosis.
Abstract Figure. Waterfall chart of individuals CaScAoV
Performance of Prediction Models for Diagnosing Severe Aortic Stenosis Based on Aortic Valve Calcium on Cardiac Computed Tomography: Incorporation of Radiomics and Machine Learning
