Combining cardiac magnetic resonance and computed tomography coronary calcium scoring: added value for the assessment of morphological coronary disease?

Stolzmann, Paul ; Alkadhi, Hatem ; Scheffel, Hans ; Plass, André ; Leschka, Sebastian ; Falk, Volkmar ; Kozerke, Sebastian ; Wyss, Christophe ; Donati, Olivio

In: The International Journal of Cardiovascular Imaging, 2011, vol. 27, no. 7, p. 969-977

Ajouter à la liste personnelle
    Summary
    To investigate prospectively, in patients with suspicion of coronary artery disease (CAD), the added value of coronary calcium scoring (CS) as adjunct to cardiac magnetic resonance (CMR) for the diagnosis of morphological coronary stenosis in comparison to catheter angiography (CA). Sixty consecutive patients (8 women; 64±10years) referred to CA underwent CMR (1.5 T) including perfusion and late gadolinium-enhancement imaging as well as CS with computed tomography. Diagnostic performance was evaluated for CMR and CS separately, and for both methods combined, with CA as reference standard. Best CS threshold combined with a specificity >90% to predict significant stenosis in patients without abnormalities on CMR was determined from receiver operator characteristics (ROC) analysis. Abnormal CMR results were considered to indicate significant stenosis regardless of CS; CS above threshold reclassified patients to have CAD regardless of CMR. CA identified 104/960 (11%) coronary segments with coronary artery stenosis >50% in 36/60 (60%) patients. ROC revealed an area-under-the-curve of 0.83 (95%CI: 0.68-0.99) with the best CS threshold of 495 Agatston score (sensitivity 50%). CMR depicted 128/960 (13%) myocardial segments with abnormalities in 31/60 (52%) patients. Sensitivity, specificity, negative (NPV) and positive predictive value (PPV) of CMR were 78, 88, 72 and 90%. When adding CS to CMR, sensitivity and NPV increased to 89 and 83%, while specificity and PPV slightly decreased to 83 and 89%. Accuracy of the combined approach (87%) was significantly (P<0.05) higher than that of CMR (82%) alone. Adding CS to CMR improves the accuracy for the detection of morphological CAD