Atherosclerotic Plaque Development

Coronary atherosclerotic plaque development begins early in life and is characterized by the accumulation of lipid-laden macrophages within the intima of arterial walls.² With increasing accumulations, the lesions often progress to Stary type IV and type V atheroma, which are well-developed plaques characterized by intramural collections of cholesterol and phospholipids. These lipid collections are often covered by a thin, fibrous cap (fibroatheroma). Because of remodeling, the lesions initially have little significant luminal narrowing and are often undetected by angiography. Two thirds of individuals with acute myocardial infarctions or unstable angina have only minimal angiographic narrowing at the culprit site of occlusion. Myocardial perfusion studies that attempt to identify the hemodynamic effects of coronary stenoses may be normal and often underestimate an individual’s risk for a cardiac event.

Because these plaques are predisposed to spontaneous rupture, the lesions are often referred to as “vulnerable plaques.” Why plaques rupture is unclear, but the process is likely multifactorial and related to biomechanical stresses and localized plaque inflammation. The histologic composition of these plaques may predict eventual outcomes from CAD; screening examinations that can identify plaque morphology may provide the best assessment of risk for coronary heart disease.

When a fibrous cap ruptures, the lipid core is exposed to circulating blood, and an acute thrombogenic reaction may ensue. Advanced lesions that produce stenoses have a greater prevalence in patients with chronic or stable angina, and they are more frequently detected by traditional diagnostic techniques that either identify the stenosis or screen for their hemodynamic effects.

A strong correlation has been found between the quantitative measurements of coronary artery calcium and pathologic measurements of plaque area and volume. Rumberger and colleagues³ showed that calcium is identifiable by CT when plaque area measures 5 to 10 mm² per 3-mm-thick voxel. It has been established that as calcium increases, so does the likelihood of hemodynamically significant stenoses. Heavy concentrations of calcium suggest a greater atherosclerotic burden and a greater likelihood of hemodynamically significant stenoses. Supporting this concept is an article by Kragel and associates,⁴ who reported that atherosclerotic plaques associated with significant stenosis often contain more calcium than nonobstructive plaques.

Autopsy studies have shown that large CAC burdens correlate with greater likelihoods of significant arterial luminal narrowing, especially when distributed over multiple vessels. One such study was by Mautner and coworkers,⁵ who examined 1298 segments from 50 heart specimens and observed that 93% of arteries with stenoses greater than 75% had CAC. Conversely, only 14% of arteries with stenoses less than 25% were associated with calcification. Many other studies have shown that heavier CAC burdens were strongly associated with significant stenoses on angiography and with overall poorer patient outcomes. Calcium measurements derived from CT cannot predict site-specific stenoses. CAC measurements cannot be used to predict the site or the severity of the stenoses.⁶

Coronary calcium is a frequent constituent of vulnerable and hard plaques, and the presence and quantity of CAC correlate well with the overall severity of the atherosclerotic process, and make these lesions potentially identifiable by traditional noninvasive methods, such as fluoroscopy and CT. There are no diagnostic tests that can identify a priori vulnerable plaques that are susceptible to rupture. Postmortem analysis of coronary arteries of adults with sudden cardiac death have shown that histologically determined calcium scores for stable and ruptured plaques were similar (4.5 vs. 5.2).⁷ Despite this, plaque calcification is frequently present in most patients who have acute plaque disruption and sudden death. In addition, intravascular ultrasound examinations performed on patients with acute cardiac events (infarct or unstable angina) have shown that vulnerable plaques tend to be associated with less calcification than plaques found in patients with stable angina, and that moderate levels of coronary calcium portend a greater number of vulnerable plaques and a subsequent higher risk for sudden death. Complicating this issue further is evidence suggesting that small to moderate amounts of calcium within plaque may be associated with a more unstable plaque configuration, which may facilitate their eventual rupture, and may make plaques less tolerable to shear stresses and promote endothelial lining disruption.

Electron-Beam Computed Tomography

The electron-beam CT scanner was introduced in 1980. It generated images by having electrons sweep tungsten target rings located beneath the patient to produce x-rays that traversed the patient to be collected by solid-state detectors in the gantry above the patient. This technology, with its absence of moving parts, allowed temporal resolutions of 50 to 100 ms, which for the first time could essentially “freeze” cardiac motion.

Older electron-beam CT scanners operated at 130 kV, 625 mA to produce 50-ms to 100-ms 1.5-mm, 3-mm, and 6-mm slice thickness images at resolutions of 9.5 1 p/cm. Later model electron-beam CT scanners operated at 140 kV, 1000 mA, and were able to generate dual 1.5-mm, 3-mm, and 8-mm slices at 50-ms temporal resolution. The spatial resolution was 10 1 p/cm. This updated version provided a 100-ms mode for higher spatial resolution and a 33-ms mode for higher temporal resolution.

Multidetector Computed Tomography

Electron-beam CT scanners have now been replaced by multidetector CT scanners, which have detectors capable of generating 256 detector images of varying thicknesses with each gantry rotation. Gantry rotation times have been reduced from 1000 ms to 330 ms, and with segmented reconstruction and dual source imaging, times approximating 83 ms are possible. An added advantage with most scanners is that the information generated is as a volumetric data set, and this permits reformations at different slice thicknesses. Multidetector CT images are commonly ECG gated, and this further decreases motion unsharpness by allowing image acquisitions during the quieter phase of the cardiac cycle. The latter is particularly important in coronary calcification and CT angiography imaging.

Current multidetector CT scanners can generate images by prospective gating, wherein the scanner is activated only during the time needed to acquire an image; this is roughly at one half the gantry rotation time. The time of data collection is initiated from the R wave of the ECG, and the operator can select the desired delay. This mode is frequently used in CAC imaging because patient radiation dose can be kept to a minimum. Rapid or irregular heartbeats can affect image quality and reproducibility, however. Previous 4-detector ring scanners rotating at 0.5 seconds and programmed to provide 2.5-mm slice thicknesses could acquire data in about a 20-second breath-hold; however, current 64-detector scanners are able to reduce scanning times to 8 to 12 seconds.

Retrospective gating is the more commonly used operating mode. In this sequence, there is helical scanning of the entire heart while recording the patient’s ECG. On completion of the scan, the images are reconstructed at a preselected phase of the cardiac cycle. To avoid anatomic gaps in the data set, the pitch is set very low; the patient radiation exposure is higher. Nevertheless, the ability to reconstruct images during multiple phases of the cardiac cycle from the same high-resolution data set provides important information. Submillimeter slices 0.6 mm thick can be obtained to achieve high spatial resolution. The patient’s heart rate can be a major factor influencing image quality, however, and if the heart rate exceeds 65 to 70 beats/min, β blockers are commonly administered to allow data collection during a single heartbeat.

Calcium Score Reporting

Calcium scores are reported using the Agatston score, volume score, and mass score.⁸ The Agatston score was the initial reporting score and is used in much of the older literature. It used electron-beam CT technology to identify lesions with a threshold of +130 Hounsfield units (HU) and 2 to 3 contiguous pixels located over the course of the coronary artery. To calculate the score, a region of interest is placed around each lesion, and the area of the lesion is multiplied by a weighted factor of 1 to 4 based on the peak signal anywhere in the lesion. A weighted factor of 1 is used for a peak calcification of 130 to 199 HU; 2, for a peak calcification of 200 to 299 HU; 3, for a peak calcification of 300 to 399 HU; and 4, for a peak calcification of greater than 400 HU.

The volume score linearly interpolates the data for isotropic volumes and represents the volume (in mm³) of each lesion above the 130 HU threshold. Lesions with similar area but differing amounts of calcium may have different volume scores.⁸

The mass score uses a calibration factor derived from scanning a phantom containing a known amount of calcium. The phantom is placed in the scanning field, and a calibration factor is determined. From it, the calibration factor times the number of voxels containing threshold calcium times the volume of one voxel times the mean CT number for each lesion equates to the mass score (in mg). The total score is the sum of all individual scores.

Rumberger and Kaufman⁸ reviewed the Agatston, volume, and mass scores, and found equivalence of the three CAC scoring methods for stratification of their cohort of 11,490 individuals who had undergone electron-beam CT. Likewise, comparable agreement was found among the three CAC scoring methods over successive electron-beam CT scans. Based on phantom experiments, however, these investigators reported nonlinearity of the Agatston and volume scores with the volume score overestimating lesion volume. Using the same phantoms, mass scores were found to be linear with a few exceptions.

Standardization of Computed Tomography Scanners

To ensure that calcium scores are meaningful, it is important that CT scanners and protocols are standardized so that scores from one scanner can be compared with another. Toward that end, the Physics Task Group of the International Consortium on Standardization in Cardiac CT was formed.⁹ Using a phantom with inserts of calcium and water density material embedded in an epoxy anthropomorphic body torso, scanning algorithms for all five commercially available scanners were developed (Fig. 32-2). The manufacturers included were Toshiba, Imatron, General Electric, Phillips, and Siemens.

FIGURE 32-2 A semi-anthropomorphic thoracic phantom and three additional attenuation layers were used to represent small, medium, large, and extra-large adults. Two syringes were placed in the water-filled cardiac regions of the phantoms. The dotted circle on each phantom represents the ROI where the background noise level was measured.

(From Yu L, Primak AN, Liu X, McCollough CH: Image quality optimization and evaluation of linearly-mixed images in dual-source, dual-energy CT. Med Phys 2009; 36:1019-1024.)

Multidetector CT scanners were calibrated against the phantom for temporal and spatial resolution and noise. Minimum requirements included not less than 4 slices per rotation, rotation times less than 0.5 second, and an ability to reference an ECG signal. The target noise baseline was set at ± 20 HU for the water insert of the phantom. To accommodate different patent sizes, external circumferential rings can be added. Using consortium-developed scanner algorithms, variations of 4% for Agatston scores, 7.9% for volume scores, and 4.9% for mass scores were achieved. The calculated calcium score was within ± 5 mg of the actual calcium mass of the phantom. For the calibrations, a fixed density of 100 mg/mL of calcium hydroxyapatite was used. Subsequently, all manufacturers have now implemented these recommendations into their clinical protocols. To determine the approximate phantom size, the lateral skin-to-skin measurement width at mid-liver measured from an anteroposterior radiograph is used. A multidetector CT database registry is currently under development.