Assessment of atherosclerotic luminal narrowing of coronary arteries based on morphometrically generated visual...

Determination of the degree of stenosis of atherosclerotic coronary arteries is an important part of postmortem examination of the heart, but, unfortunately, estimation of the degree of luminal narrowing can be imprecise and tends to be approximations. Visual guides can be useful to assess this, but earlier attempts to develop such guides did not employ digital technology. Using this approach, we have developed two computer-generated morphometric guides to estimate the degree of luminal narrowing of atherosclerotic coronary arteries. The first is based on symmetric or eccentric circular or crescentic narrowing of the vessel lumen and the second on either slit-like or irregularly shaped narrowing of the vessel lumens.

Using the Aperio ScanScope XT at a magnification of 20× we created digital whole-slide images of 20 representative microscopic cross sections of the left anterior descending (LAD) coronary artery, stained with either hematoxylin and eosin (H&E) or Movat's pentachrome stain. These cross sections illustrated a variety of luminal profiles and degrees of stenosis. Three representative types of images were selected and a visual guide was constructed with Adobe Photoshop CS5. Using the "Scale" and "Measurement" tools, we created a series of representations of stenosis with luminal cross sections depicting 20%, 40%, 60%, 70%, 80%, and 90% occlusion of the LAD branch. Four pathologists independently reviewed and scored the degree of atherosclerotic luminal narrowing based on our visual guides. In addition, digital technology was employed to determine the degree of narrowing by measuring the cross-sectional area of the 20 microscopic sections of the vessels, first assuming no narrowing and then comparing this to the percent of narrowing determined by precise measurement.

Two of the observers were very experienced general autopsy pathologists, one was a first-year pathology resident on his first rotation on the autopsy service, and the fourth observer was a highly experienced cardiovascular pathologist. Interobserver reliability was assessed by determination of the intraclass correlation coefficient. The degrees of agreement for two H&E- and Movat-stained sections of the LADs from each of 10 decedents were 0.874 and 0.899, respectively, indicating strong interobserver agreement. On the average, the mean visual scores were ~8% less than the morphometric assessment (52.7 vs. 60.2), respectively.

The visual guides that we have generated for scoring atherosclerotic luminal narrowing of coronary arteries should be helpful for a broad group of pathologists, from beginning pathology residents to experienced cardiovascular pathologists.