Beyond Coronary Stenosis: Coronary Computed Tomographic Angiography for the Assessment of Atherosclerotic Plaque Burden

Abstract

Coronary computed tomographic angiography (CCTA) is emerging as a key non-invasive method for assessing cardiovascular risk by measurement of coronary stenosis and coronary artery calcium (CAC). New advancements in CCTA technology have led to the ability to directly identify and quantify the so-called “vulnerable” plaques that have features of positive remodeling and low density components. In addition, CCTA presents a new opportunity for noninvasive measurement of total coronary plaque burden that has not previously been available. The use of CCTA needs also to be balanced by its risks and, in particular, the associated radiation exposure. We review current uses of CCTA, CCTA’s ability to measure plaque quantity and characteristics, and new developments in risk stratification and CCTA technology. CCTA represents a quickly developing field that will play a growing role in the non-invasive management of cardiovascular disease.

Introduction

Coronary artery disease (CAD) remains the leading cause of morbidity and mortality worldwide despite major pharmacologic and risk stratification advancements over the last two decades.(1) In 30–50 % of patients, the first sign of CAD is presentation with acute myocardial infarction.(2) Multiple studies have linked non-calcified atheroma to adverse outcomes.(3, 4) Interestingly, patients who experience acute MI are generally found to have previous sub-acute plaque ruptures prior to their major event.(5, 6) Therefore, the detection of these vulnerable plaques may allow for ideal risk stratification prior to a severe CAD outcome. Recent technological innovations in coronary computed tomography angiography (CCTA) have improved imaging resolution such that CCTA can now provide both visualization of the coronary lumen, and identification of at-risk non-calcified plaque.(7–10) The reference standard diagnostic tool for coronary stenosis is quantitative coronary angiography (QCA), while intravascular ultrasound (IVUS) is the standard for investigation of in-vivo intracoronary plaque characteristics. However, both modalities are unsuitable for screening purposes, as they are invasive and expensive. Although CCTA currently does not carry the spatial resolution of IVUS, it is non-invasive and technological increases have improved speed and resolution while decreasing radiation exposure.(11–14) These properties make CCTA an ideal screening modality for genetically susceptible and intermediate-risk patients, as well as a plausible method to quantitatively track the utility of atherosclerosis treatment.(15, 16)

Accurate analysis of plaque characteristics by CCTA would allow for improved risk stratification, measurement of response to therapy, and understanding of genetic markers of disease. This review will focus on CCTA in the context of non-calcified plaque, specifically: 1) characteristics and epidemiology of vulnerable plaques, 2) the ability of CCTA to identify, quantify, and assess plaque characteristics, and 3) recent technological improvements and future directions in CCTA.

Current Methods of Risk Stratification by Coronary Computed Tomographic Angiography

Technological advancements in CCTA have resulted in increasing diagnostic capability as well as an accelerating trend in utilization. CCTA is well validated for analysis of coronary artery calcium while more sophisticated techniques for plaque burden and composition remain under development.

Coronary Artery Calcium Score

Currently, CT is used to assess coronary artery calcium (CAC) score, which has been successfully used as an independent predictor of negative CAD events such as myocardial infarction (MI), revascularization, and mortality.(17–19) Detrano et al.’s study from the multiethnic study of atherosclerosis (MESA) has expanded CAC’s applicability to multiple ethnic groups including white, black, Hispanic, and Chinese. Doubling the CAC score increased the risk of a cardiac event by 18–39%. No major differences were seen between ethnic groups in the predictive ability of CAC score.(20) Additionally, CAC has been shown to add incremental value to a number of the standard measurements of cardiac risk such as JUPITER and Framingham risk scores. Blaha et al. found that in a cohort of MESA patients with CAC score >0, and meeting JUPITER inclusion criteria, the five-year number-needed-to-treat (NNT) to prevent cardiovascular disease (stroke and MI and complications) was only 30.(21) In the Heinz Nixdorf Recall Study, CAC was used to correctly reclassify patients with intermediate cardiac risk by Framingham Risk Score, resulting in a net reclassification index (NRI) of 21.7% to a low risk category (determined by CAC<100) and 30.6% to a higher risk category (determined by CAC≥400).(22) A study of the MESA population by Polonsky et al. found a similar NRI for CAC score on top of traditional risk factors (age, gender, tobacco, blood pressure, cholesterol, and ethnicity), at an overall rate of 25%.(23) However, while CAC appears to be an excellent surrogate marker for presence of disease, calcification of atherosclerotic plaque is thought to represent the endpoint of plaque maturation and lesion stabilization. Heavily calcified plaque is not the typical culprit lesion implicated in acute coronary syndrome.

Low CAC scores appear to be insufficient to fully assess risk. Multiple studies of patients with a CAC score of 0 have confirmed the presence of NCP and significant coronary artery stenosis, with prevalence of 11–13% and 0.9–3.7% respectively and severe stenosis (>70% luminal narrowing) seen in 0.4–1.5%.(24–27) CAC score of 0 has only a 45% sensitivity to predict the absence of ≥50% stenosis.(28) Additionally, within approximately 20% of asymptomatic patients with cardiac risk factors and no detectable CAC, significant stenosis is present. Within almost one third of these patients, these stenoses result in ischemia.(29) These studies show that there is a significant at-risk population that is unable to be diagnosed by CAC and traditional risk factors alone.

Measurement of Coronary Artery Plaque Stenosis by CCTA

In 2008, three major prospective multi-center trials marked the establishment of CCTA as a technology that could equal or eventually surpass invasive angiography for determining the degree of coronary artery stenosis. These three trials (CORE-64, ACCURACY, and Meijboom et al.) compared CCTA to QCA with sensitivities/specificities of 85%/90%,(30) 95%/83%,(31) and 99%/64%,(32), respectively. These multi-center trials that offered comparison to invasive angiography establish the diagnostic capability of CCTA. In general, these trials support the concept of the use of CCTA for the evaluation of patients at intermediate risk of coronary artery disease. For patients at high risk, CCTA may be unnecessary since invasive angiography serves to diagnose and treat in the same setting.

In addition to angiography, the degree of coronary stenosis by CCTA has been compared to IVUS. In a recent meta-analysis, Voros et al. identified 33 studies comparing CCTA to IVUS. They showed that CCTA was excellent for detection of coronary plaques (sensitivity 90%, specificity 92%), with increased accuracy with improving scanner technology. CCTA slightly overestimated lumen area (0.46 mm² p=0.005) compared to IVUS, but there were no significant differences between plaque area, volume, weighted mean difference of volume, and percent area stenosis.(33) Thus, CCTA’s ability to identify plaque in the coronary arteries even beyond that which causes stenosis is well established.

Beyond detection of plaque, CCTA has shown promise in characterizing the type of plaque that is present. In our laboratory using Vital Images software (Vitrea fX 6.1), the total amount of plaque can be determined by a semi-automated method (Sureplaque) that detects the inner and outer vessel walls over a length of the coronary artery. Area between inner and outer vessel walls is counted as plaque. Plaque burden is calculated by normalizing the volume of plaque within the vessel by the length of the vessel. In this system, an artery that appears normal, without visually identifiable plaque, had a total plaque burden (TPB) of approximately 6.5–7.5 mm2. Patients with increased amounts of plaque (eg figures 1a and 1b) have larger scores. The total plaque is further divided by the software into several components by HU density (soft plaque: −100–29; fibrous plaque: 30–149; calcified plaque: 150–1300). The subdivided plaque burdens in normal-appearing vessels are soft plaque burden (SPB) of <2 mm2, fibrous plaque burden (FPB) of <4 mm2, and calcified plaque burden (CPB) of <2 mm2. Figure 1a shows visualization of calcification within coronary arteries using different CCTA views, with the patient’s corresponding plaque burdens. The use of CCTA for characterizing plaque is discussed in greater detail below.

Figure 1.

a. Calcified coronary arteries. A) Agatston calcium score axial view. B) Filtered Back-projection axial view. C) Arterial cross section view. D) Orthogonal curved multiplanar reconstruction view. Green arrows: Calcified plaque in left anterior descending (LAD) artery visualized in multiple modalities. This patient’s CAC score was 1982. Total plaque burden (TPB) is 11.0; soft plaque burden (SPB) is 1.9; fibrous plaque burden (FPB) is 4.4; calcified plaque burden (CPB) is 4.7.

b. Coronary arteries exhibiting non-calcified plaque. A) Agatston calcium score axial view. B) Filtered Back-projection axial view. C) Arterial cross section view. D) Orthogonal curved multiplanar reconstruction view. Green arrows: Mixed LAD plaque.. Red arrows: Significant non-calcified plaque in LAD proximal to calcified segment. The patient’s CAC score is 54. Total plaque burden (TPB) is 10.1; Soft plaque burden (SPB) is 1.7; Fibrous plaque burden (FPB) is 5.3; Calcified plaque burden (CPB) is 3.1.

Plaque characterization with CCTA

Overview: Vulnerable Plaques and Acute Coronary Syndromes

Culprit lesions responsible for rupture, thrombosis, and acute MI are typically larger soft plaques with incomplete or no calcification representing a different segment of disease from CAC.(34–36) These plaques were found to be 1) lipid laden and non-calcified plaques (NCP), 2) thin-capped, and 3) positively remodeled(37–39) implying that vulnerable, pre-rupture plaques would share the same characteristics.

The clarity of the term “non-calcified plaque” has suffered greatly from its dual use as a descriptive and definitive phrase. Early studies focusing on CCTA alone have often defined NCP as any discernable object within the vessel wall with enhancement greater than the wall structures but less than the lumen,(26, 40, 41) effectively defining NCP as any plaque that is not completely calcified. Recent SCCT guidelines have recommended subdividing plaque type into calcified, non-calcified, or mixed.(42) IVUS comparison studies further subdivide plaque categories with definitions of NCP excluding plaques with minor calcified elements within.(4, 43) Histological comparison studies subdivide plaques by the Stary classification system, with NCP synonymous with corresponding Stary stages.(44) For the purposes of this review, we will refer to NCP as plaques that are not completely calcified, encompassing both SCCT categories of “non-calcified plaque” and “mixed plaque,” and corresponding with the characteristics of the culprit lesions discussed above.

While many studies have attempted to correlate plaque sub-types with risk factors, results have been highly inconsistent. States generally found to be associated with NCP include typical cardiac risk factors such as: hypertension,(45, 46) smoking,(45–47) obesity,(46, 48) gender,(47) diabetes,(47) CAC score,(49) high-sensitivity CRP,(50) and apolipoprotein A1;(50) as well as non-typical factors such as: pericardial fat volume,(51) epicardial fat volume,(52) and inflammatory monocyte subsets.(53) It is important to note that not all studies found statistical strength of association for these factors.(48) The variability in correlating NCP with known cardiovascular risk factors suggests a significant need for prospectively designed studies.

Recent evidence suggests that CAC is not a perfect indicator of atherosclerotic burden. In patients with zero CAC score, the rate of obstructive NCP in asymptomatic populations is estimated to be <1%. However, in symptomatic patients this percentage is much higher, suggesting that there are at-risk population subsets that would benefit from NCP screening.(26) In fact, elevated NCP burden is present in almost all patients who present with ACS.(38, 39, 54) A study by Motoyama et al. found that culprit lesions were comprised of NCP in 79% of ruptured lesions causing ACS, whereas stable lesions causing typical angina were comprised of NCP in only 9%.(55) The prospective multicenter ACCURACY trial showed that plaque composition by CCTA is related to stenosis severity by QCA; mixed plaques showed a higher rate of stenoses ≥70% (68%) compared with completely non-calcified and calcified plaque (28% and 3.6%).(56) Other studies demonstrated higher proportions of mixed plaques in patients with ACS vs. stable CAD, as well as higher proportions of NCP.(4) These studies suggest that NCP represents a clinically-significant type of measurable pathology, associated with both ACS and stenosis, and not adequately addressed by CAC score.

Coronary Computed Tomography and the Ability to Characterize Plaque Types

Ever since early multi-detector coronary CTA, researchers have been interested in the ability to non-invasively differentiate plaque types. Comparison of 4-detector row CT with histology by Becker et al. noted that CCTA could assess significant attenuation differences between lipid-rich and fibrous-rich non-calcified plaque (47±9 and 104±28 HU, respectively; p<0.01),(57). Further developments in CT technology enabled more precise differentiation between plaque types with 16-slice MDCT, although microcalcification within plaques remained difficult to classify.(58–60)

As imaging technology has improved, ex-vivo histological comparisons have become less prevalent. Despite concerns about accuracy,(61, 62) IVUS has become the gold standard for intravascular plaque analysis.(63, 64) Early comparisons of CCTA with IVUS showed promise for characterization and quantification of plaques, but results suffered from poor scan resolution and reproducibility, (65–68) especially outside of the proximal coronary tree segments.(69) Improvements in CT technology in more recent studies resulted in improving ability to quantify and distinguish plaque types,(4, 43, 70–72) especially if the measured areas were limited to regions with high scan quality, without blurring or heavy calcification.(73) Figure 1b shows CCTA visualization of calcified and non-calcified plaque within coronary arteries.

Multiple studies have attempted to determine distinct HU ranges corresponding to different histological plaque types. Mean HU densities for lipid-rich soft plaques range from 14–75 HU,(66, 74) fibrous plaques range from 67–149 HU,(75, 76) and calcified plaques range from 135–1089 HU.(77, 78) Table 1 summarizes recent studies correlating attenuation with plaque characteristics. Figure 2 shows lipid-rich and fibrous soft plaques visualized on CCTA.

Table 1.

Studies correlating attenuation level with plaque characteristics.

First Author	Year	Citation	CT	Comparison	N (lesions)	Soft/Lipid- Rich/hypoechoic (HU)	Intermediate/Fibrous /hyperechoic (HU)	Calcified (HU)
Kopp	2001	(65)	4-MDCT	IVUS	6	6 ± 28 and −5 ± 25	83 ± 17 and 51 ± 19	489 ± 372 and 423 ± 111
Schroeder	2004	(44)	4-MDCT	Histology	17	42 ± 22 (Stary III/IV)	70 ± 21 (Stary V)	715 ± 328 (Stary VII)
Sakakura	2006	(119)	16-MDCT	IVUS	23	50.6 ± 14.8	131 ± 21.0	721 ± 231
Becker	2003	(57)	4-MDCT	Histology	33	47 ± 9 (Stary IV and Va)	104 ± 28	-----------------
Schroeder	2001	(66)	4-MDCT	IVUS	34	14 ± 26	91 ± 21	419 ± 194
Xiao	2007	(120)	16-MDCT	Histology	38	53 ± 12	106 ± 17	429 ± 94
			64-MDCT	Histology	38	51 ± 13	110 ± 19	435 ± 87
Hur	2009	(72)	64-MDCT	IVUS	61	54 ± 13	82 ± 17	392 ± 155
Chopard	2010	(62)	64-MDCT	Histology	83	70 ± 41	83 ± 35	966 ± 473
				IVUS	83	67 ± 31	84 ± 40	724 ± 564
Iriart	2007	(121)	16-MDCT	IVUS	84	38 ± 33	94 ± 44	561 ± 216
Carrascosa	2003	(76)	4-MDCT	IVUS	156	75.73 ± 44.30	148.61 ± 36.54	449.07 ± 221.4
Ferencik	2006	(77)	16-MDCT	OCT	164	29 ± 43	101 ± 21	135 ± 199
Viles-Gonzalez	2004	(122)	16-MDCT	Histology (Non-human)	182	51 ± 25	116 ± 27	-----------------------
Kitagawa	2007	(75)	64-MDCT	IVUS	202	18 ± 17	67 ± 21	-----------------------
Pohle	2007	(123)	16-MDCT	IVUS	252	58 ± 43	121 ± 34	-----------------------
Carrascosa	2006	(74)	4-MDCT	IVUS	276	71.5 ± 32.1	116.3 ± 35.7	383.3 ± 186.1
Leber	2004	(67)	16-MDCT	IVUS	450	49 ± 22	91 ± 22	391 ± 156
Choi	2008	(82)	64-MDCT	IVUS	80	41.3 ± 26.4 (≥10% necrotic core volume)	93.1 ± 37.5 (<10% necrotic core volume)
							44 (26–63) (Stary V);
Galonska	2008	(78)	16-MDCT	Histology	195	50 (33–59) (Stary IV)	40 (36–67) (Stary VI); 67 (37–124) (Stary VIII)	1089 (333–1944) (Stary VII)
							34 ± 19 (Stary V); 104
Leschka	2010	(59)	DSCT	Histology	245	55 ± 12 (Stary IV)	± 84 (Stary VI); 91 ± 16 (Stary VIII)	501 ± 402 (Stary VII)
						51 ± 25	92 ± 59	968 ± 494

Figure 2.

Filtered back-projection reconstructions of coronary arteries exhibiting non-calcified plaque indicated by red arrows, with occasional calcification (not marked). A) Patient with plaque in left main coronary artery. HU of plaque is 5, consistent with lipid-rich plaque. Total plaque burden (TPB) is 7.9; soft plaque burden (SPB) is 1.5; fibrous plaque burden (FPB) is 3.6; calcified plaque burden (CPB) is 2.8. B) Patient with plaque in left main coronary artery. HU of plaque is 29, consistent with lipid-rich plaque. TPB is 6.5; SPB is 2.6; FPB is 3.5; CPB is 0.4. C) Patient with plaque in left anterior descending artery. HU of plaque is 85, consistent with fibrous non-calcified plaque. TPB is 10.1; SPB is 1.7; FPB is 5.3; CPB is 3.1. D) Patient with plaque in left anterior descending artery. HU of plaque is 85, consistent with fibrous non-calcified plaque. TPB is 8.7; SPB is 1.7; FPB is 4.9; CPB is 2.1.

Comparison is difficult between these studies, due to different modes of classification for CCTA-determined plaque types (non-calcified/mixed/calcified, fatty/fibrous/calcified, hyperechoic/hypoechoic/calcified, or stratification by Hounsfield units), which do not precisely correspond to the different modes of IVUS classification (soft/intermediate/calcified, calcified/mixed/fibrous/soft, fatty/fibrous/calcified, fibrotic/fibro-fatty/necrotic-core/calcified) or histological classification. Despite the need for future improvement of CCTA, there is a clear trend that improving technology continues to close the gap to provide a non-invasive, cost-saving alternative to expensive and invasive gold standard measurements.

CCTA and the Ability to Identify Vulnerable Plaques

As our understanding of plaque histology has improved,(79) interest has risen in the ability of CCTA to identify particularly-at-risk plaque structures. At risk plaques tend to have low x-ray attenuation (Hounsfield unit) values. Alternatively, soft plaque may had specific anatomic characteristics such as the “napkin-ring sign” that indicate a specific morphology.

CCTA quantification of plaque types subdivided by SCCT recommendation (non-calcified, calcified, partially calcified) shows excellent inter and intra-rater agreement.(80) While the categories lack the descriptive strength of pathology-based histological subtypes by Stary classification, CCTA and IVUS-based histology correspond well in some important respects. Highly vulnerable plaques such as thin-capped fibroatheromas (TCFA) are found in 32% of plaques classified as partially calcified by CCTA, versus 13% for “non-calcified,” and 8% for calcified.(43) Additionally, TCFA may be associated with the “napkin-ring sign” on CCTA, a ring-like enhancing lesion that has been implicated in rupture and ACS in multiple studies.(7–9) In these cases, the ring may be due to differential density of necrotic lipid core surrounded by thin fibrous material, as the lesion itself is typically associated with large plaques with necrotic lipid cores.(10)

More recent studies show a weak correlation between high density NCP and fibrous plaque tissue (R=0.47) as well as between low-density NCP and fibrofatty/necrotic core plaques (R=0.25), as noted in the ATLANTA I study using quantitative CCTA to IVUS.(81) Choi et al. demonstrated that plaques with greater than 10% necrotic core volume had significantly lower HU densities (41.3±26.4 vs 93.1±37.5) and significant correlations were found between HU density and percent necrotic core (r=−0.53) and fibrotic component (r=0.57), but not with fibrofatty component or calcium determined by IVUS. (82)

CCTA and the Ability to Identify Positive Remodeling

Positive remodeling, as described by Glagov et al.,(83), is the tendency of coronary arteries to undergo compensatory enlargement from early atherosclerosis. Positive remodeling is measured by dividing the cross sectional vessel wall area by a corresponding proximal and distal segment reference area or lesion area. While definition of the proximal and distal reference segments is somewhat subjective, the CCTA approach has been validated against IVUS (75, 84, 85) and has shown excellent reproducibility.(80) Positive remodeling has been associated with a high risk of ACS. (37, 38) Kitagawa et al. which showed that within NCP lesions, remodeling index was the only factor that correlated with culprit lesions causing ACS and non-culprit lesions within patients who had ACS due to plaque rupture at a different site.(86) Positive remodeling is presumed to be associated with inflammation and increased propensity for rupture. Negative, or constrictive remodeling, is associated with lesion stabilization and has been observed in plaques following statin treatment.(87)

More recent studies have examined the use of positive remodeling, which is easily measured by CCTA, as a surrogate marker for regions of vulnerable plaque. IVUS studies have demonstrated that positive remodeling is associated with increased intra-plaque lipid volume and decreased calcium and fibrous tissue, a composition typical for vulnerable plaque.(88, 89) This association is supported within the CCTA literature, as Schmid et al. demonstrated correlation between degree of remodeling and CT attenuation of plaque in a sample of 72 lesions. Positively-remodeled plaque attenuation had a significantly lower value than lesions displaying no or negative remodeling (59±22 HU versus 91±20HU, p<0.001).(90) Additionally, a study by Kroner et al demonstrated that thin capped fibroatheromas by IVUS were associated with 43% of positively-remodeled plaques by 64MDCT as compared to 4.8% of negatively remodeled lesions.(91) The authors concluded that CCTA positive remodeling was correlated with actual histological markers of plaque instability.

Future Directions in Coronary Computed Tomographic Angiography

Plaque characterization by CCTA has potential application for 1) assessing overall risk for primary prevention, 2) evaluation of ACS, and 3) following progression of disease and response to therapy. These topics are reviewed below.

Plaque assessment for primary prevention

There have been several studies looking at the prognostic value of plaque characteristics both individually and in addition to current modes of risk stratification. In 810 patients, Matsumoto et al. demonstrated that low attenuation plaque (HU <68) was associated with higher rates of major cardiac events over 2.9 years (OR:2.9 95%CI:1.2–6.7 p<0.05).(92) In a study of 517 patients referred for cardiac evaluation undergoing 16-slice MDCT and myocardial perfusion imaging, van Werkhoven et al showed that having two or more segments with completely non-calcified plaque was predictive of events (HR:5.0 95%CI:2.2–11.7 p<0.01) and that having three or more segments with mixed plaque was also predictive of events (HR:3.5 95%CI:1.5–8.1 p<0.005).(93) This same group was also followed for composite all-cause mortality, non-fatal MI, and unstable angina over a median follow-up of 22 months in 432 patients undergoing CAC and 6-slice CCTA. There was a significant incremental predictive value of CCTA in patients with a calcium score greater than 1000, particularly with regards to number of NCP’s (HR:1.3 95%CI:1.1–1.4 p=0.001) and mixed plaque-containing segments (HR:1.2 95%CI:1.0–1.4 p=0.039). The number of calcified segments was not predictive.(94)

More impressively, Ahmadi et al. recently showed that CCTA plaque type had significant predictive value in 1,102 symptomatic patients with non-obstructive (stenosis <50%) CAD followed over 78±12 months.(95) In this cohort, the rate of death increased in association with calcified plaque (1.4%) to mixed plaque (3.3%) to completely non-calcified plaque (9.6%). The risk-adjusted hazard ratio in patients with mixed plaque was 3.2 (95%CI1.3–8.0 p<0.001) and for completely non-calcified plaque the risk was twice as high at 7.4 (95%CI:2.7–20.1, p<0.001). These studies demonstrate that the presence of soft plaque and mixed plaque provided incremental value in predicting all-cause mortality over traditional measurements of risk.

Plaque assessment and acute coronary syndrome

Distinguishing between ACS and stable angina on presentation would allow for faster and more accurate treatment. Retrospective analysis of CCTA from patients presenting with ACS have shown differences between their plaque characteristics compared to those with stable angina. Volumetric plaque burden was assessed by 64-slice CCTA in 57 patients with NSTEMI compared to 19 with stable angina pectoris, finding that the mean HU density was lower in patients with NSTEMI (74 vs. 99HU p=0.02), and that the volume of calcified plaque and number of lesions with calcified plaque was decreased in NSTEMI. Stable angina patients had fewer segments with NCP, and positive remodeling was found in 19% of patients with NSTEMI as compared none in patients with stable angina.(96)

Two studies have looked at developing standardized scoring methods for detecting culprit lesions, which include plaque characteristics such as low attenuation, remodeling index, spotty calcification, and plaque volume. In one of these studies, Kim et al. combined plaque attenuation of ≤ 60 HU, remodeling index ≥1.05, and presence of NCP or spotty calcification to predict ACS patients with a sensitivity of 97.1% and specificity of 67.6%, ROC analysis showed an (AUC:0.908 p<0.001).(97) Ferencik et al. used presence of positive remodeling, spotty calcium, and stenosis length >4.5mm, finding that this was associated with OR 4.6, 95% CI 1.6 to 13.7 (AUC 0.824) for prediction of ACS over stable angina.(98)

Although prognosis before the onset of MI is ideal, using CCTA to stratify risk after diagnosis of ACS may also be of value. Stone et al. revealed that recurrence of a major cardiac event after treatment with percutaneous coronary intervention was attributed to a different, untreated vessel region almost half of the time. These untreated regions were associated with the typical characteristics of vulnerable plaque by IVUS,(3) therefore, if CCTA is able to identify co-existing vulnerable plaques, treatment decisions can be tailored to the patient’s individual risk. In a study of 312 patients, CCTA measurement of NCP volume in patients presenting with NSTEMI was a better predictor of future cardiac events over 16 months than clinical variables and left-ventricular ejection fraction (LVEF).(99) Watabe et al. showed that CCTA-defined culprit lesions prior to percutaneous coronary intervention may be predictive of myocardial necrosis and significantly greater post-procedural risk.(100)

Plaque assessment to monitor treatment progression

There is little biological evidence that statins are able to reduce the burden of calcified plaque, but earlier studies magnetic resonance imaging studies of the aorta and carotid arteries have shown reduced plaque burden in response to intensive statin treatment. (101–108) With CCTA, a number of studies have showed decreased NCP with treatment.(16, 109–111) Burgstahler et al. demonstrated that initiation of atorvastatin therapy reduces NCP volumes (24±13% p<0.05) but does not impact calcium scoring (Agatston score: 261±301 vs. 282±360 p>0.05) or total plaque burden (0.15±0.11ml vs. 0.13± 0.075ml p>0.05).(112) Inoue et al. used serial CCTA to track decreases in plaque volume in response to fluvastatin treatment, and found decreases in both low attenuating plaque volume (4.9 ± 7.8 vs. 1.3 ± 2.3 mm³, p = 0.01) and total plaque volume (92.3 ± 37.7 vs. 76.4 ± 26.5 mm3, p < 0.01).(16) While specific therapy to prevent NCP rupture has yet to be discovered, there is evidence that statins may cause constrictive remodeling and increased plaque stability, suggesting that the decrease in plaque volume in the studies above may be protective against ACS events.(87)

Technical issues for plaque characterization with CCTA

Radiation Reduction in Coronary CTA

Accurate assessment of coronary plaque by CCTA requires high spatial resolution and excellent contrast resolution. Unfortunately both “requirements” are associated with high radiation dose. In addition, the concept of CCTA monitoring of plaque healing implies multiple CCTA examinations with further radiation exposure. Already, CT use for non-cardiac applications represents the greatest contributor to medical radiation in the US which has now reached equivalence to natural background radiation.(113, 114) Radiation exposure associated with CCTA is especially concerning because of the widespread nature of coronary artery disease and high radiation doses that were used with early CT scanning devices. The sections below indicate recent developments that have allowed cardiac CT radiation exposure to transition from one of the highest dose to one of the lowest dose CT scanning applications.

CCTA presents a unique set of hurdles due to the necessity for high resolution images in an anatomical area predisposed towards significant motion artifacts. Widespread evaluation of CCTA began in 2004, with radiation dose estimates ranging from 9–21.4 mSv per scan(14) due to continual scanning throughout the cardiac cycle at maximum tube current and voltage with retrospective electrocardiogram (ECG) gating. ECG-based tube current modulation was subsequently introduced, where tube voltage was decreased by 75–96% when outside of diastole, resulting in a decrease to approximately 8–10 mSv per scan.(13) Low tube voltage protocols were the next development in patients with BMI<25, with reduction of tube voltage from 120 kV to 100 kV, resulting in a radiation dose of approximately 6.5 mSv.(13) Prospective ECG triggered gating, or “Step-and-shoot” technology, where the x-ray tube is only activated during pre-determined moments in diastole effectively took the ECG-based tube current modulation technology to its limit. While this procedure generally was only useful in patients with lower heart rates, radiation dose was decreased to 1.5–4 mSv depending on the tube voltage protocol that was used.(13) Most recently, development and improvement of multidetector CT (MDCT) with large scale detector systems has allowed single-beat acquisition of CCTA for patients with heart rate <63 bpm and decreases in exposure to the 1 mSv range.(11, 13, 14)

Image Reconstruction in Coronary CTA

More recent innovations have focused in improvement of post-processing techniques through computationally intense algorithms that improve the clarity and quality of scans to compensate for decreased scan radiation.(115, 116) Classically, image reconstruction in CT has used a Filtered Back Projection (FBP) algorithm. This mode of reconstruction creates images by combining the multiple axial 2D back-projections generated by each gantry revolution and applying filters to increase contrast and compensate for the increased signal intensity at the center of the image. FBP relies on a few simplifying assumptions, idealizing the characteristics and interactions of the x-ray beam with the patient and the detector, although leading to some inaccuracies and artifacts. FBP has the benefit of being fast with low computational load, but suffers in scanning large patients, imaging at high resolution or in areas of high bone density (e.g. the pelvis) or using when low-dose technique.(117)

Iterative reconstruction refers to a repeated process of image reconstruction that draws the reconstructed image closer to an idealized model, creating a reconstruction with less noise and higher clarity. Figure 3 shows a comparison of FBP versus iterative reconstruction, showing decreased noise in the iterative reconstruction views. This approach can compensate for lower radiation dose, making sub-millisievert CCTA possible.(118) While based on similar reconstruction theory, these techniques are not exactly the same and the literature reveals subtle differences between all of these methods. The variety and range of new techniques for radiation dose reduction are defined in Table 2.

Figure 3.

Comparison of axial and arterial cross-sectional views using different reconstruction techniques from same data set. A) Filtered back-projection. B) Adaptive Iterative Dose Reduction, mild level. C) Adaptive Iterative Dose Reduction, standard level. Notable for increased incremental image quality, decreased noise and increased smoothness in higher levels of iterative dose reduction (right) versus traditional reconstruction by filtered back-projection (left).

Table 2.

Iterative reconstruction techniques for radiation dose reduction

Abbreviation	Name	Company	# CCTA Studies	Conclusion	Citations
ASiR	Adaptive Statistical Iterative Reconstruction	GE Healthcare	5	27% dose reduction versus FBP possible with equal image quality	(124–128)
iDose	iDose Iterative Reconstruction Technique	Philips Healthcare	3	Hypothetical 76% dose reduction, improved image quality over FBP in normal and low-dose protocols	(129–131)
IRIS	Iterative Reconstruction in Image Space	Siemens Medical Solutions	2	68% dose reduction versus FBP with equivalent noise levels	(132, 133)
SAFIRE	Sonogram-Affirmed Iterative Reconstruction	Siemens Medical Solutions	2	50% dose reduction versus FBP with improved image quality	(134, 135)
ADIR	Adaptive Iterative Dose Reduction	Toshiba Medical Systems Corporation	1	Dose reduction unknown, 42% noise reduction, 70% CNR improvement	(136)
MBIR	Model Based Iterative Reconstruction	GE Healthcare	2	Dose reduction unknown, ex vivo study shows increased image quality, 51–69% CNR increase over ASiR and FBP	(137, 138)

Conclusion

New technological developments have expanded the ability of CCTA to elucidate aspects of cardiovascular disease previously only measurable through invasive modalities; however, concerns still remain about utilization and radiation exposure. New methods of reconstruction may allow CCTA to be performed with lower radiation dose and better image quality. Besides assessment of coronary stenosis, plaque composition and characteristics have begun to assume increasing importance. CCTA can thus combine features of stenosis detection by invasive angiography and plaque characterization by invasive ultrasound. Assessment of plaque by CCTA may be used to assess prognosis as well as to identify at risk plaques. Finally, the concept of medical therapy and monitoring of plaque healing and regression by CCTA has recently become possible using low radiation dose techniques.