Applied Radiology

Dr. Thompson is an Associate Professor and Dr. Stanford is a Professor at Roy J. and Lucille A. Carver College of Medicine, Iowa City, IA.

Coronary heart disease (CHD) affects 1.6 million Americans each year and accounts for approximately 500,000 deaths annually. ^1,2 The pathogenesis of cardiovascular heart disease is complex, poorly understood, and is related to a multitude of predisposing factors. Even with a high aggregate risk for CHD, many individuals with coronary artery disease (CAD) will not experience a "cardiac event." Conversely, of those who experience a myocardial infarction, only approximately 50% will have a history of CAD. ³ Despite the fact that the 8-year risk of CHD for the average middle-aged person is only 1% to 5%, depending on risk factors, the enormous costs of this disease (both in lives lost and healthcare expenditures) makes the development of an effective screening program extremely important. ⁴ The results of therapies specifically aimed at risk factors, such as hypertension and hypercholesterolemia, have reduced the mortality rates from heart disease. ^1,5 Despite the enormous epidemiological efforts that target risk-factor assessment and modification, risk-factors alone predict only two-thirds of individuals who will eventually succumb to heart disease. ³ Thus, there is a critical need to develop a simple, noninvasive screening test that can provide an accurate assessment of the presence and severity of CAD, while at the same time can predict the individual risk for developing CHD.

In the last 11 years, there has been an explosion of interest and associated scrutiny examining the clinical value of using the presence of coronary arterial calcification (CAC) as a prognosticator of CHD. Since CAC has been unequivocally shown to be a marker of arteriosclerosis, considerable investigation has been performed over the last decade, evaluating CAC detection and quantification as a means to identify and predict those who are at greatest risk for CHD. A great deal of this research has focused on computed tomography (CT), specifically electron beam CT (EBCT), producing evidence that EBCT has value as a valid screening tool for CHD. Recently, the development of helical CT (HCT) technology has refocused these research interests toward helical CT, assessing its potential for CHD screening as well.

This article will briefly revisit the evidence for CAC screening as a predictor of CHD and examine the unique characteristics of EBCT that have established its superiority in CAC detection. Additionally, the authors will discuss how the rapid technological advances in HCT are creating a challenge to EBCT's role as the reference standard for CAC screening.

The predictive "value" of coronary calcium

The presence of vascular calcification is an indisputable marker for atherosclerosis. Furthermore, a strong correlation has been identified between quantitative CAC measurements and pathologic assessment of plaque volume and area. ^6-9 Detection of CAC initially performed by conventional and digital fluoroscopy subsequently identified an association between the amount of CAC and CAD ^5,10 and demonstrated a relationship between actual calcium mass measurements and the measurements of the actual histologic specimen. ¹¹

The presence of large and diffuse calcium deposits is also associated with significant vascular narrowing on angiography on a site-by-site basis. ¹² Coronary calcification, particularly when distributed over multiple vessels, correlates with angiographic results, showing a high sensitivity for atherosclerotic vascular disease (80% to 100%). ^12-24 Coronary calcification serves as a strong predictor of significant coronary narrowing. ^{1,3,7,19,20,25-35} There is also a strong correlation between individual artery CAC area measurements and plaque assessments determined by histology (r = .90, P < 0.001). ⁶ Numerous studies have also repeatedly shown that extensive CAC, particularly when distributed over multiple vessels, is associated with significant coronary narrowing, correlates well with angiography, and is also a predictor of patient outcomes. ^{1,3,7,19,20,25-35} Conversely, there is strong evidence that the absence of CAC, while not excluding the presence of coronary atherosclerotic disease, virtually excludes the likelihood of significant coronary arterial stenosis (negative predictive values = 84% to 100%). ^{13,14,18,25,33,36-41} The specificity of CAC measurements for CAD is generally lower, with reports ranging from 31% to 100%. No gender differences exist between men and women. ^12-25,42

While the presence of CAC as measured by calcium volumes and scores are reflective of atherosclerosis, can such measurements really be predictive of future coronary heart disease? Although the development of CHD is unpredictable, autopsy studies have shown a definite correlation between coronary calcium burden and the frequency of myocardial infarction. ^43-45 The causes of acute coronary occlusion due to atherosclerosis are likely multifactorial. "Unstable" plaque configurations in conjunction with localized inflammation probably act as a precipitating factor. ^2,4,46 Histologically, "soft plaques" (those with lipid cores and thin fibrous caps) are believed to be at greatest risk for rupture and subsequent vascular thrombosis. ⁴⁶ It is postulated that soft plaques are more likely unstable and prone to rupture, leading to acute cardiac syndromes, but that such plaques are not necessarily associated with significant luminal narrowing. ^4,46

Occlusive coronary disease that arises from the rupture of lipid-rich plaques occurs independent of plaque size or severity of luminal narrowing. ^2,4 In fact, up to two-thirds of patients who have acute myocardial infarction or unstable angina may have only minimal narrowing at the site of the occlusion. ² It is not clear if vascular calcium destabilizes plaques facilitating rupture or whether this marker of plaque maturity imparts plaque stability. ^4,46 It is known, however, that hard and soft plaques coexist in similar proportions. ^47-49 Therefore, the quantification of calcified plaques could serve as a surrogate measurement of the number of soft plaques. ³ Thus, it seems logical to assume that CAC could be useful as one determinate to establish individual relative risk for CHD.

There is growing consensus that the risk of developing CHD can be stratified according to CAC volume, with greater risk ascribed to those with extensive, multivessel disease. Quantitative CAC measurements by EBCT have shown that as CAC increases so does the likelihood of coronary heart disease. ¹³ Patients with EBCT calcium scores above the median (>75) were found to be six times more likely to experience a cardiac event, such as myocardial infarction and sudden cardiac death. ¹⁹ Arad et al ³³ demonstrated that individuals with calcium scores >160 were 35 times more likely to experience a cardiovascular event and that CAC was more predictive of such events than other traditional risk factors. For coronary artery calcium score thresholds of 100, 160, and 680, the sensitivities of EBCT for cardiac events were 89%, 89%, and 50%, while the specificities were 77%, 82%, and 95%, respectively. Ultimately, such observations have served as the foundation for the establishment
of CAC quantification by CT as a valid useful screening exam for CHD. ^{1,3,7,19,20,25-35} These investigators, among many others, were instrumental in developing EBCT imaging standards that have become universally applied by many investigators and screening centers around the world. ^{6-8,12-14,19,20,25-28,36,37,50,51}

Technical advantages of EBCT

The unique design of EBCT affords significant advantages over conventional CT for cardiac imaging. Having no moving parts, EBCT is capable of providing 50-msec image acquisition times. Coupled with prospective electrocardiogram (ECG) gating, EBCT can provide near real-time cross-sectional imaging of the beating heart. Possessing excellent spatial and temporal resolution, EBCT is considered by many as the reference standard in providing cardiac analysis, and, particularly, CAC quantification.

EBCT imaging protocol

Standard imaging protocol for EBCT CAC screening encompasses single breath-hold neutral axis imaging with prospective ECG gating. Implementing a 60% to 80% R-REKG interval (R-R) interval trigger point, coupled with superb temporal resolution, EBCT can provide reproducible diastolic image sets of the heart that are critical to CAC quantification. In the volume mode sequence (3-mm slice thickness at 100 msec/slice), EBCT can produce 80 images of the heart with a nominal pixel size of 0.68 mm ² (35-cm reconstruction circle) (Figure 1). To optimize temporal and Z-axis spatial resolution, a complete data set can usually be acquired with one breath. The examination time is usually <15 minutes, delivering <2.6 mSv of radiation for a complete coronary calcium study. This would equate to approximately 10.4 months of background radiation to the general public.

Several proprietary software packages enable operators to quickly detect and perform quantitative CAC measurements for all major coronary arteries. Aggregate calcium burden is usually reported as a calcium score that can be compared with normalized data according to age- and gender-matched controls. ¹³ While scoring packages allow independent manipulation of threshold criteria, most centers default to the scoring protocol established by Agatson et al. ¹³ However, increasing emphasis is being placed on the volume score, rather than on the calcium score, as a measurement of CAC burden.

The inter- and intra-observer variabilities of EBCT CAC measurements are excellent. ^13,52,53 Poorer inter-study variability exists, particularly with the identification of smaller foci of calcium. ²¹ This variability largely stems from slice misregistration due to cardiac motion. This problem has been largely rectified by manipulating threshold area or slice thickness and/or by averaging the CT density measurements of lesions (rather than using peak density of calcium deposits) to calculate calcium scores. ^54,55 Other solutions that have improved examination reproducibility have been achieved by implementing postprocessing. ⁵⁶ Similarly, averaging scores from duplicate scans has improved reproducibility and is now becoming common practice at many screening sites. Shields et al ⁵⁷ reported a reliability of 0.99 in 50 subjects who underwent dual scanning. ⁵⁸ Hernigou et al ⁵³ reported an inter-examination error rate of 7.2%. ⁵⁸

Helical CT

Since EBCT is considered the reference standard for imaging of coronary calcium, how does HCT compare? Due to a host of very significant technologic advances in recent years, the performance of HCT in CAC quantification is quickly approaching that achieved with EBCT. The most significant of these developments relate to faster image acquisition times, use of multislice systems, and institution of ECG gating. Just in the past 3 years, image acquisition times have decreased from 1 sec to 500 msec, significantly enhancing temporal resolution, image quality and scan reproducibility (Figures 2A and B). Half-scan techniques in the sequential single-slice mode have decreased imaging time from 320 msec to 125 msec. Like EBCT, HCT allows examinations to be performed using single breath-hold sequences, thereby improving Z axis resolution. Inherent advantages in image efficiency have also been realized with the introduction of multislice scanning protocols that provide greater Z axis coverage per revolution, further decreasing examination time and decreasing problems with slice misregistration.

The ability to acquire a true diastolic image data set has become a reality with development software that provides prospective ECG gating. The significance of this development has remedied a significant shortcoming of HCT--previously, nongated acquisitions that were prone to significant volume averaging and slice misregistration errors resulted in divergent CAC measurements as compared with EBCT. ⁵⁹

Gated acquisitions can be performed operating in the sequential (single-slice) mode or helical acquisition mode, and can be obtained either prospectively (similar to EBCT) or retrospectively. The latter requires postprocessing of the entire data set, allowing operators to select diastolic images individually for analysis. Since retrospective ECG-gated studies are associated with large data sets, there is a concomitant increase in total radiation dose as well as additional time requirements to perform postprocessing image analysis (Table 1). ⁶⁰ Prospective ECG gating closely resembles an EBCT acquisition, choosing a triggering point at peak diastole (usually 60% to 80% of the R-R interval). Compared with retrospective gating, this mode represents a targeted, smaller data set with an associated decrease in total radiation dose. Unfortunately, prospective gated sequential acquisitions require longer examination times to allow for table movement to occur. This, in turn, often requires two acquisitions to complete a study, potentially introducing breathing and slice misregistration artifacts. Compared with multislice acquisitions that can examine the heart in larger tissue volumes, the sequential gated acquisitions, in reality, are associated with less artifact degradation and tend to provide CAC scores that more closely approximate EBCT measurements. ^61-63 Regardless, the development and institution of ECG gating has significantly improved image quality (Figure 2) and calcium score variability compared with traditional nongated acquisitions. ⁶⁴

Early investigations that compareCAC measurements of EBCT and HCT show that both now possess comparable performance in image quality and CAC quantification. Reproducibility correlation coefficients reported by several investigators are also excellent (r = .95 to .99). ^65-69 Preliminary evidence also shows that helical CAC measurements, like EBCT, correlate well with angiography with regard to accuracy and predictive values for CAD. ^70,71 Broderick et al, ⁶⁵ using two scoring methods for HCT, reported sensitivities and specificities of 81% to 88% and 52% to 61%, respectively, as compared with obstructive disease by angiography. There was no significant difference between contiguous slice acquisitions and overlapping slice (volumetric) acquisitions. ⁶⁵ Like EBCT, negative HCT scans appear to have high negative-predictive values for CAD compared with angiography (specificity 100%; sensitivity 61%; accuracy 85%). ⁷¹

Conclusion

With significant advancements in hardware design and software technology, HCT technology has made great progress in removing many of the hurdles that initially limited its CAC screening capability, as compared with EBCT. At this time, preliminary data is beginning to demonstrate comparable performances for both modalities. Unlike EBCT, where the technology is mature and additional refinements unlikely, helical technology will continue to undergo further refinements, such as expansion of multislice configurations. This will, in turn, provide additional improvements in spatial resolution and will decrease examination times. Considering the popularity and widespread availability of spiral scanners, these inevitable refinements portend a future in which HCT will likely assume a pre-eminent role in CAC detection and quantification. AR