Animal Models and Virtual Histology
We read with much interest in the February issue of Arteriosclerosis, Thrombosis, and Vascular Biology an article by Granada et al1 entitled “In vivo Plaque Characterization Using Intravascular Ultrasound Virtual Histology in a Porcine Model of Complex Coronary Lesions.” The results of this study showed that the Virtual Histology (intravascular ultrasound (IVUS)-VH) was not accurate in detecting the relative amount of specific plaque components in an animal model. The findings from this study are no surprise to us, as the algorithm was created from human coronary arteries; the authors failed to mention that IVUS virtual histology was not created to detect pig arterial changes that have been produced in 10 weeks of high cholesterol feeding and injection of oxidized LDL delivered by microsyringe infusion catheter. Although the authors claim that they performed frozen sections, the illustrations provided are paraffin-embedded samples and not one section of the oil red O-stained artery is provided to convince the reader that actually lipid deposition was detected in a credible manner or the lesions in any way resemble human atherosclerotic disease. Also the lesion illustrated with “myeloperoxidase stain” is actually negative. We believe that the model is not one of atherosclerosis but is rich in smooth muscle cells and proteoglycans and therefore more indicative of a restenosis model. The authors claim that fibrous tissue was 40.3±12.8% and fibrolipidic was 16.8±8.19% (not clear how this was determined), and it is unclear how the fibrous tissue was determined and by what stain. The authors do not show a single lesion that actually resembles human atherosclerotic disease and therefore it is not surprising that a technology created and based on human plaques would actually detect pig restenosis with macrophage infiltration (we assume that authors are showing their best sections).
Virtual histology is an IVUS-based technique that analyzes the backscattered ultrasound signal reflected from tissues and correlates them with a predefined database of frequency-based ultrasound parameters.2 These parameters were determined by the careful correlation of backscattered data collected from fresh ex vivo human arteries with the corresponding histology sections. In the original article by Nair et al,3 4 tissue types were defined. Regions of densely packed collagen were termed fibrous; those with significant lipid interspersed with collagen were labeled as fibrolipid (now fibro-fatty); necrotic regions comprised cholesterol clefts, microcalcifications, and were devoid of collagenous structural components; and densely packed calcium deposits were identified as calcium. One can infer from the original description of the technique that these regions of interest were homogeneous; however, human coronary plaques are very complex, and although Nair et al3 provided a gross description of the tissue type used to train the algorithm, the reality is that they trained their algorithm to detect subtleties found only in human plaques. It must be stressed that virtual histology does not detect specific chemical compounds such as lipid or collagen, rather the mixture of compounds that make up 1 of the 4 tissue types (eg, necrotic core). Only true histology, in many cases using specialized staining techniques, can truly determine plaque composition down to the specific component level.
In summary, there is no obvious reason to suggest that the Granada animal model described can accurately reproduce the human coronary plaques and therefore generate meaningful data from the pig model. This model produced plaques rich in smooth muscle cells and possibly foam cells, and it is not possible to appreciate the extent of lipid in these lesions as no specific lipid stains were shown. Also, the lesions lack necrotic core, calcification, and collagen (type I), and therefore we believe the VH-IVUS would not be able to identify the components that the system had been trained to detect. It is not surprising that the VH-IVUS did not match the histology, as the histological findings reported by Granada et al may not have existed in the model. In addition, it appears that intra- and possibly interoperator variability (data not shown) in detecting the EEL may have led to the misclassification of adventitia as calcium.
CVPath Inst, Inc received company sponsored research support from Volcano Therapeutics Inc, and Renu Virmani, MD, is a consultant for Volcano Therapeutics Inc.
Granada JF, Wallace-Bradley D, Win HK, Alviar CL, Builes A, Lev EI, Barrios R, Schulz DG, Raizner AE, and Kaluza GL. In vivo plaque characterization using intravascular ultrasound-virtual histology in a porcine model of complex coronary lesions. Arterioscler Thromb Vasc Biol. 2007; 27: 387–393.
Nair A, Kuban BD, Tuzcu EM, Schoenhagen P, Nissen SE, and Vince DG. Coronary plaque classification with intravascular ultrasound radiofrequency data analysis. Circulation. 2002; 106: 2200–2206.