Influence of BMI on virtual coronary artery calcium scoring.

Abstract

Virtual non-contrast (VNC) coronary artery calcium scoring (CAC) may obviate the need for traditional non-contrast (TNC) CAC. There is no data on the influence of body mass index (BMI) on VNC reliability. We aimed to evaluate the influence of BMI on VNC CAC agreement with TNC. All patients who underwent sequential CAC and coronary CT angiography (CCTA) using spectral CT with TNC CAC > 0 between August 2020 and December 2021 were included. Agatston CAC scores were calculated manually by 2 blinded readers from VNC scans. A correction factor was calculated from the slope of the linear regression using the method of least squares and applied to the VNC scores. Bland-Altman plots and Cohen's weighted Kappa were utilized. We included 174 patients (57.5% female). Mean BMI was 32.6 ± 7.02kg/m2 [BMI < 30 (39.7%); BMI 30-40 (45.4%); and BMI > 40kg/m2 (14.9%)]. Mean TNC CAC was 177.8 ± 316.86 and mean VNC CAC after applying the correction factor 149.34 ± 296.73. The TNC value strongly correlated with VNC (r = 0.94; p < 0.0001). As BMI increased there was a progressive reduction in signal-to-noise ratio, contrast-to-noise ratio and coronary enhancement (p < 0.05). The degree of agreement between VNC and TNC CAC decreased as BMI increased (agreement = 91.79 (weighted Kappa = 0.72), 91.14 (weighted Kappa = 0.58) and 88.46% (weighted Kappa = 0.48) (all P values < 0.001) for BMI < 30; 30-40 and > 40kg/m2, respectively). BMI has a significant influence on the accuracy of VNC CAC. VNC CAC shows substantial agreement in non-obese patients but performs poorly in BMI > 40kg/m2. This is the first study to evaluate the influence of body mass index (BMI) on virtual non-contrast (VNC) coronary artery calcium scoring (CAC) as compared to traditional non-contrast (TNC). We retrospectively evaluated 174 patients with TNC CAC and two blinded reviewers manually calculated the VNC CAC. All cases were included without specific selection for quality. The ratio between the two directly proportional values was determined using the slope from the linear regression through the method of least squares. This correction factor of 2.65 was applied to the calcium scores obtained from VNC images. We found that VNC CAC shows substantial risk-class agreement with TNC in non-obese patients (agreement = 91.79 and weighted Kappa = 0.72) but performs poorly in BMI > 40kg/m2 (agreement: 88.46% and weighted Kappa = 0.48). These findings show the potential use of VNC CAC to avoid additional radiation in non-obese patients. However, further research on potential improvement strategies for VNC CAC in obese patients is needed.