Quantification of aortic valve fibrosis and calcification by computed tomography angiography: a prospective head-to-head comparison with histology

Calcific aortic stenosis (AS) is characterized by leaflet degeneration involving fibrocalcific alterations which limit leaflet mobility and eventually lead to obstruction of flow across the aortic valve. While cardiac computed tomography (CT) has been routinely used for quantification of aortic valve calcium in individuals in whom echocardiography is inconclusive, recent studies on quantification of aortic valve fibrosis are lacking histological validation. [1,2]

We, thus, aimed to validate a non-invasive CTA-derived quantitative characterization of aortic valve tissue against histologic examination of excised aortic valve leaflets.

In this prospective study, patients underwent CTA within a month before surgical aortic valve replacement for aortic stenosis. Tomographic valve tissue components were identified by Gaussian mixture modeling applied to CTA, by automatically fitting Gaussian functions to the Hounsfield Unit (HU) histograms for blood pool, non-calcific, and calcific aortic valve tissue. Tissue composition, expressed as % [(tissue component volume / total tissue volume) × 100%], was automatically calculated. For histologic evaluation, explanted valve leaflets were stained with Movat's Pentachrome. For every leaflet, three non-consecutive 5-µm sections were obtained. Further, an automatic algorithm detecting pixel intensities was applied for color differentiation and quantification of tissue components (Figure 1). Separate measurements from all the sections of the single valve were finally averaged and compared with CTA-derived calculations.

A total of nineteen patients (63% men, 67±8 years old) undergoing isolated surgical aortic valve replacement for severe aortic stenosis with preoperative CTA were evaluated. The mean proportion of non-calcific aortic valve tissue was 68±16% in CTA and 75±3% on Movat’s Pentachrome staining (p=0.004, paired samples comparison). There was a strong correlation between the two modalities (r=0.856, p<0.001; Figure 2A). Intermodal repeatability between CTA and histology for non-calcific tissue quantification was excellent with an interclass correlation coefficient of 0.855 (95% CI 0.663 – 0.942, p<0.001), coefficient of repeatability of 16.8% and bias of -6.7% (p=0.234, Figure 2B).