HISTOLOGIC VALIDATION OF A NOVEL 3D ULTRASOUND APPROACH TO DETECTING VULNERABLE ATHEROSCLEROTIC PLAQUE

Abstract

<h3>BACKGROUND</h3> Atherosclerosis is a progressive inflammatory disease characterized by plaque formation due to subintimal lipid accumulation in the arteries, which eventually leads to cardiovascular events. Carotid endarterectomy (CEA) is commonly performed to reduce mortality due to plaque accumulation. Currently, CEA candidates are selected based on plaque burden, but ultrasound (US) pixel grayscale analysis of plaque lesions may better identify plaque composition and vulnerability to rupture. Grayscale ranges representing various tissue types have been previously defined in the literature using decades-old US systems. Direct histologic validation of these ranges has not been established. Our objective was to assess the ability of US to identify plaque tissues based on existing grayscale pixel ranges and validate them via histologic analysis of excised plaques. <h3>METHODS AND RESULTS</h3> A 3D carotid US (Philips Healthcare EPIQ Elite, XL14-3 transducer) was performed on 4 participants prior to undergoing CEA. The excised plaques were fixed, decalcified, and embedded in 2% agarose for imaging. Each plaque was sectioned every 2 mm, trichrome-stained, and histological composition was determined using QuPath, a digital image analysis software. The 3D US image was iSliced using QLab Philips software and slices were registered to corresponding histological sections. US plaque composition was analyzed using a novel, semi-automated software (IntelliPlaque) that applied the previously published grayscale ranges ex-vivo to map tissue types within the plaque. Matched paired t-tests were used to compare tissue composition between US and histology. Comparison​ of plaque tissue composition determined by US (percent fibrous, muscle, fat, blood, and calcified tissue) and the corresponding histological sections (n=27) indicated some resemblance. Percent fat-like tissue on US was not significantly different from percent foam cells on histology, indicating resemblance for this tissue type (mean difference=-0.20±0.41%, r=0.27; p=0.63). Similarly, percent muscle and percent calcium on US were not significantly different from histology (p>0.1). Percent fibrous and percent blood were significantly different, indicating poor resemblance. The weakest resemblance was between percent fat-like tissue on US and percent lipid-rich/necrotic core on histology (-19.80±3.09%, r=0.66; p < 0.0001). <h3>CONCLUSION</h3> Our findings suggest that existing grayscale ranges corresponding to the necrotic core and/or blood tissue are outdated and require re-validation using modern US technology. US pixel distribution analysis is a non-invasive, rapid assessment tool that can improve patient risk stratification beyond plaque burden alone. Updated grayscale ranges will improve the detection of plaque composition and potentially risk of plaque rupture.