Role of Computed Tomography for Diagnosis and Risk Stratification of Patients with Suspected or Known Coronary Artery Disease

Abstract

Cardiac computed tomography angiography(CCTA) has emerged as a powerful imaging modality for the detection and prognostication of individuals with suspected coronary artery disease (CAD). Because calcification of coronary plaque occurs in proportion to the total atheroma volume, the initial diagnostic potential of CCT focused on identification and quantification of coronary calcium in low to intermediate risk individuals, a finding that tracks precisely with the risk of incident adverse clinical events. Beyond non-contrast detection of coronary calcium, CCT employing the use of iodinated contrast yields incremental information regarding the degree and distribution of coronary plaques and stenosis, as well as vessel wall morphology and atherosclerotic plaque features. This additive information offers the promise of CCT to provide a more comprehensive view of total atherosclerotic burden as it relates to myocardial ischemia and future adverse clinical events. Further, emerging data suggest the prognostic and diagnostic importance of stenosis severity detection and atherosclerotic plaque features described by CCT—including positive remodeling, low attenuation plaque and spotty calcification—which have been associated with the “vulnerability” of plaque. We report a summary of the evidence supporting the role of CCT in the detection of subclinical and clinical CAD in both asymptomatic and symptomatic patients, and discuss the potential of CCT to augment identification of at-risk individuals. CCTA and coronary artery calcium scoring offer the ability to improve risk stratification, discrimination and reclassification of the risk in patients with suspected CAD and to non-invasively determine the measures of stenosis severity and atherosclerotic plaque features.

Coronary Calcium Scoring by CT

Pathophysiology of Coronary Artery Calcium

Coronary atherosclerosis is a complex inflammatory process that involves, in part, endothelial damage, deposition of oxidized low density lipoprotein into the intima and smooth muscle cell proliferation, macrophage infiltration and activation.^1-4 Several inflammatory mediators and chemokines contribute to plaque initiation, growth, and rupture, which is the most common proximate event behind sudden coronary thrombotic events that result in an acute coronary syndrome.² Amongst the plaque constituents, coronary calcium fares prominently, and is virtually pathognomonic for atherosclerosis⁵ (Figure 1).⁶ Calcium phosphate in a hydroxyapatite form accumulates in intimal atherosclerotic lesions.⁷ Calcification of the atherosclerotic plaque first appears in the lipid core of the atheroma, juxtaposed to inflammatory cells and occurs by an active process resembling bone formation, under the control of complex enzymatic and cellular pathways.⁸ This formation involves osteoblast-like cells, cytokines, transcription factors, and bone morphogenetic proteins (such as BMP2a, osteoprotegerin, osteopontin, osteocalcin and osteonectin), and is characterized by inflammation, lipoprotein and phospholipid accumulation, apoptosis, and finally hydroxyapatite deposition.⁹ The inciting mechanisms are not definitively understood but apoptosis of smooth muscle cells seems to be an important step, which then serves as a nidus for calcification. There is a linear relationship between coronary calcification and total coronary plaque burden on a segmental and entire coronary vessel basis. Based on histomorphometric studies, roughly 20% of the atherosclerotic plaque burden in the coronary vascular bed is calcified and these macro calcifications can be identified by non-contrast enhanced computed tomography.^7,9,10

Figure 1.

Diagram depicting the progression of coronary atherosclerosis with age. By the teenage years, there is almost universal appearance of type 3 lesions in the aorta.⁶ With increasing number of risk factors and age and ensuing endothelial damage, there is progressive growth in atheroma size. Calcification occurs in some of these lesions and leads to increasing calcium deposition with time, which is readily appreciated by non-contrast cardiac CT.

Methods of Measurement of Coronary Artery Calcium

Identification and quantification of coronary artery calcium by CT was first performed using electron beam CT (EBCT), a technology that provides the very high temporal resolution needed to image the perpetually-moving coronary arteries. This temporal resolution is facilitated by a CT scanner with a non-moving gantry using an electron beam steered electromagnetically onto tungsten rings to produce x-rays, which are then swept across the patient's heart and detected on 2 parallel static detector rings.¹¹ With the more recent introduction of multidetector row CT (MDCT), however, EBCT has been generally supplanted a scanner with superior spatial resolution. By either method, the radiation dose to detect and quantify coronary calcification is very low. Radiation exposure of coronary artery calcium scoring is ∼1 mSv, which is comparable to that of a screening mammogram (0.7 mSv).¹² With rapidly occurring further improvements in scanner hardware and software, radiation exposure will go down even further, to below 0.5 mSv.

An array of scores have been used to quantify coronary calcium content, amongst which the Agatston score remains the most commonly employed.¹¹ In this method, calcified lesions above a threshold value of 130 Hounsfield Units (HU) and area > 1 mm² (3 adjacent pixels) are detected based on their density and area. The area of calcification in each of the slices is calculated and multiplied by a weighting factor (e.g., for 130-200 HU, a factor of 1 is assigned: for 201-300 HU, a factor of 2 is assigned: for 301-400 HU, a factor of 3 is assigned: and so on). All axial scores are then summed (Figure 2). One caveat is that the Agatston is prone to overestimation of coronary calcium content, as it is susceptible to partial volume effects, motion, arrhythmias, image noise and artifacts from metallic clips and implanted defibrillator leads. To overcome these limitations, Callister proposed a volume score because the Agatston score does not take into account the depth of calcium, but rather the area only.¹³ Thus, an increase in the Agatston score over time may occur from an increase in the amount of plaque but may also occur as a result of increased plaque attenuation (density) rather than a true increase in plaque size. In contrast, the calcium volume score is independent of image overlap and slice thickness.¹³ In the Multi Ethnic Study of Atherosclerosis (MESA) study, each participant underwent an EBCT and MDCT 2 minutes apart – volume scores showed better reproducibility compared to Agatston scores.¹⁴ A third method of measuring coronary calcium content is the mass score, a technique that multiplies the lesion volume, the average CT density and a calibration factor. The method of coronary artery calcium scoring (CACS) requires an external standard calibration phantom but is the most reproducible value as it is independent of hardware or scanning protocols. Nonetheless, these newer scores have a smaller prognostic evidence base of studies and despite its aforementioned limitations, the Agatston score has been studied extensively for risk assessment in various large population-based studies and continues to enjoy the greatest popularity of use.

Figure 2.

Example of a non-contrast CT scan using a 64 slice CT scanner in an asymptomatic 43 year old patient whose only risk factor was a family history of myocardial infarction in his father at age 48. Calcified plaque is seen in all 3 coronaries (LAD= left anterior descending, RCA= right coronary artery, CX = circumflex artery), with an Agatston score of 456, which placed him in the 99^th percentile for his age, ethnicity and gender. Statin therapy was recommended for this patient.

Risk stratification, Discrimination and Reclassification of At-Risk Individuals Based Upon Coronary Artery Calcium Prevalence and Extent

An online arterial age calculator (www.mesa.nhlbi.org) is available to estimate the percentile ranking of a patient compared to the MESA data after taking into account age, gender, ethnicity and smoking status.¹⁵ This method, while clinically often employed, does not more effectively identify at-risk individuals. Indeed, absolute CACS of >300 or 400 are more predictive of major adverse cardiac events (MACE) rather than a percentile score.¹⁶ This is especially true in younger patients where the percentile score informs relative risk, whereas the absolute score gives the lifetime risk.¹⁷ CACS increases with increasing age and men. In general CAC scores in women lag behind men by approximately 10 years. The prognostic value of CACS is retained in the elderly.¹⁸

Further, the pattern of coronary artery calcification is also important, with greater risk of adverse clinical events occurring in the setting of greater numbers of calcified plaques and more coronary vessels involved, particularly if the calcification is in the more proximal segments.¹² A newly developed calcium coverage score incorporates the distribution and dispersion of calcium throughout the coronary tree.¹⁹ In a recent analysis of the MESA database with 3,398 patients followed for a median of 7.6 years, the calcium volume score was associated with MACE but calcium density was inversely associated with MACE, and it is feasible that more calcium-dense plaques might be protective because of a decrease in vulnerability.²⁰ At this time, there is no data showing that preventive treatment of patients with high CACS is associated with improved cardiovascular outcomes, although it is likely that it does so and that such demonstration with require a very large trial with a long follow up period.

An Agatston CACS of 1-99 identifies a mild amount of plaque burden, 100-399 is moderate, >400 is high and >1000 is severely elevated. A zero score carries an excellent prognosis in asymptomatic patients. Pooled analysis from 6 studies from 2003 to 2005 with 27,622 asymptomatic patients showed that 43% had a CACS of 0 and a very low rate of MACE of 0.4% over a 3-5 year follow up.^12,21 The risk ratio for MACE increased to 7.2 fold for a CACS between 400-1000 and to 10.8 fold for a score >1000, therein demonstrating a continuous graded relationship between CACS and MACE. Survival rates have also been shown to have a similar graded relationship: from a large observational study of 25,253 patients with a mean follow up of 6.8±3 years, cumulative 10-year survival was 99.4% for a CACS of 0, which dropped to 87.8% for a score exceeding 1000.²² In a systematic review of 13 studies with a total of almost 65,000 patients, 45% had a CAC of 0 and their event rate was 0.56% during 4.25 years of follow up.²³ Longer term follow up of up to 12 years²² showed a mortality rate of only 0.4%. Thus CACS = 0 now allows for an intermediate-term “warranty period” that is more predictive of salutary outcomes, as compared to other methods of risk stratification, such as a negative stress nuclear perfusion study or carotid intima-media thickness.^24,25 Importantly, while the non-contrast CACS does not permit direct visualization of coronary stenosis, the probability of coronary stenosis increases as the CACS increases (Figure 3)²⁶. As an example, Shareghi and colleagues observed that amongst 3,529 asymptomatic subjects of the Framingham Offspring Cohort, a zero CACS—while not excluding the presence of non-calcified plaque—virtually excluded the presence of significant atherosclerosis with a high negative predictive value (NPV) for stenosis as well as event risk. Patients within this study experienced an annualized event rate of 0.6% over 5 years.²⁷

Figure 3.

As the Agatston score increases, the proportion of patients with more obstructive stenoses on CCTA increases.²⁶

A major strength of CACS has been in its ability to improve reclassification and discrimination of patients into low-, intermediate- and high-risk states. As alluded to previously, the traditionally employed Framingham Risk Score (FRS) and Adult Treatment Panel-III (ATP-III) schema misclassify a significant proportion of low and intermediate risk individuals. In one study, 4,129 subjects without known coronary artery disease (CAD) were followed for 5 years.²⁸ When CACS was applied to refine the FRS risk estimate, 21.7% were reclassified into the low-risk category and 30.6 % were reclassified into the high-risk category. CACS therefore increased the receiver operating curve statistic, area under the curve (ROC AUC) from 0.681 to 0.749 compared to FRS, and from 0. 653 to 0.755 compared to ATP-III. Adding CACS to traditional risk factors in the MESA cohort including age, sex, tobacco use, systolic blood pressure, antihypertensive medication use, total and HDL (high density lipoprotein) cholesterol, race, and ethnicity resulted in a net reclassification index (NRI) of 0.25, implying that 25 percent of patients were correctly reclassified into a different risk group.^29,30

Notwithstanding the oft-observed discordance between CACS and traditional coronary heart disease risk factors, there is a significant continuous graded relationship between the prevalence of coronary calcium, mean CACS and the number of risk factors. In a recent investigation, 40% of asymptomatic men under the age of 40 show calcium deposits while the prevalence is 74% in men older than 60 with ≥3 risk factors. This relationship between coronary calcification and risk factors becomes diluted in older men >60, where the prevalence of coronary calcium deposits exceeds 80% and is not as tightly dependent on the number of risk factors. A family history of premature coronary heart disease also increases the likelihood of coronary calcification in the offspring.³¹ In the CARDIA (Coronary Artery Risk Development in Young Adults) study in younger adults aged 18 to 30 years, a higher number of risk factors was associated with higher CACS.³² As a general rule, gender differences are significant for CACS, with women, in general, lagging 10 years behind men in their plaque burden and CACS.

Current guidelines on CACS recommend it as reasonable (Class IIa recommendation) for risk assessment in asymptomatic adults at intermediate risk (10 year risk of 10-20%) and in diabetics; and, that it may be reasonable (class IIb) for patients at low to intermediate risk (6-10% risk). They recommend against screening (class III) for high risk (>20%) or low risk (<6%) patients because CACS results will not demonstrably alter the reclassification or discrimination of these patients for risk prediction. With respect to the latter group, there is one exception: the AUC (appropriate use criteria) committee judged it to be appropriate for low risk patients who also had a family history of premature CAD.¹² In fact, the clinical utility of CACS has also been studied in a myriad of patient subsets. The prognostic value of CACS holds true for diabetics who traditionally are thought to have a high burden and risk of CAD. In 900 patients with diabetes and a CACS = 0, survival was similar to nondiabetics with a zero score (98.8% vs. 99.4%).³³ Likewise, in the MESA study of 2,600 asymptomatic women with a median Agatston score of 0, the ROC AUC for CAD was increased from 0.8 to 0.835 with the addition of CACS.¹⁶ Similar data exist for the elderly and smokers as well.³⁴ One notable exception is patients with chronic kidney disease (CKD), which is associated with increased total body calcium and increased CACS as well.^35,36 In the CRIC (Chronic Renal Insufficiency Cohort) study of coronary calcification in patients with chronic kidney disease (CKD), the degree of coronary calcification and cardiac event rate increased in a graded manner with worsening of renal function as measured by the glomerular filtration rate (GFR). A GFR<30 carried an odds ratio of 1.53 in multivariate analysis. In the Hisayama study of Japanese patients, there was a strong correlation between both CAC and atherosclerotic plaque burden and CKD.³⁷ Furthermore, patients on hemodialysis have much higher CACS compared to age matched individuals and have a faster rate of CAC progression.³⁸ At initiation of dialysis, 50-60% of patients show CAC, with even higher rates in the subset of those with diabetes.

Role of Coronary Calcium Scoring in Symptomatic Patients with Acute Chest Pain

In symptomatic patients, a CACS = 0 portends a favorable prognosis: although in this population, the negative CACS may be associated with a small but non-negligible prevalence of non-calcified plaque, a finding predictive of higher adverse clinical event rates (3.6%).²⁷ Thus, while current National Institute for Health and Clinical Excellence (NICE) guidelines recommend CACS as the initial workup test in acute chest pain patients felt to have a low pretest probability³⁹, this strategy will miss a small proportion of patients whose CACS = 0 but nonetheless possess high-grade non-calcified plaque. In series studying these populations, approximately 6 to 12% of patients have only non-calcified plaque.^40,41 This strategy may be most useful in the low-risk population, as evidenced by a study of 1031 patients with chest pain evaluated in the emergency department, where 61% had a CACS of 0 and were discharged home.⁴² Of these, only 0.3% had events during 6 months of follow up. A CACS of 0 was also correlated with a normal follow up Single-photon emission computed tomography (SPECT) perfusion scan. However, there is the possibility of missing high-grade stenosis from non-calcified plaque by use of CACS only. Since calcified plaque represents roughly 20% of all plaque burden, despite the above reassuring data, there continues to be some controversy whether a CACS of 0 can be used to make discharge decisions in patients with acute chest pain.

Germane to these findings, the sensitivity and specificity for a CACS>0 for predicting any stenosis exceeding 50% is 91% and 49% respectively. However the negative predictive value to exclude high grade stenosis is very high in the setting of a CACS = 0, where the probability of >50% stenosis is <1%.¹⁰ As the CACS increases, the probability of obstructive CAD increases, e.g., for a CACS>400, the likelihood of a >50% stenosis is greater than 60% (Figure 3).²⁶

Progression of CACS

In the MESA study, conversion from a CACS = 0 occurred annually in 5% of 50 year olds and in >12% of 80 year olds. In those patients, the CAC progression was associated with 1.5-fold increased risk of hard coronary heart disease (CHD) events. Among patients with CAC at baseline, event risk was 6 fold higher when the progression rate was >300 units/year.⁴³ CACS progression rates were associated with conventional CHD risk factors such as age, male gender,⁴⁴ smoking,⁴⁵ hypertension, fasting plasma glucose,^46,47 family history of premature CAD,⁴⁸ and diabetes.^49,50 Novel markers of risk for CHD were also investigated. While microalbuminuria,⁵¹ carotid intima-media thickness,⁵² white race⁴³ and polymorphisms of the reninangiotensinmsystem genes⁵³ predict progression of subclinical coronary atherosclerosis, the C-reactive protein did not.^54,55 Interestingly, although abdominal obesity,⁵⁶ body mass index⁵⁷ and metabolic syndrome⁵⁰ were associated with progression of CACS, the levels of low and high-density lipoproteins were related just with the incidence, and not with the progression of preclinical atherosclerosis disease.⁴³

It has been described in a community based screening cohort that calcium screening led to a 3 to 7- fold increase in the use of aspirin, statins and lifestyle changes over 6 years.^58,59 However, data on the effect of lipid lowering therapy on the rate of progression of CACS has been still conflicting. Previous studies described positive correlation between statin treatment, serum lipid levels and coronary calcium progression.^13,60-62 More recently, however, several studies demonstrated that statin treatment was not able to reduce progression of the CACS^63-67, independently from the serum levels reduction of cholesterol and low-density lipoprotein.^63,64 St. Francis Heart Study is the largest prospective trial evaluating lipid-lowering therapy on CAC progression.⁶⁴ It randomized 1,005 patients with CAC equal or greater than the 80^th percentile for age and gender, to statin treatment (20 mg atorvastatin daily) versus placebo and were followed them prospectively with a repeat CACS in 4.3 years. Although it was demonstrated a reduction in the low-density lipoprotein by 43% from baseline (146 mg/dl), the CACS increased from 528 to 846 and the 3% absolute risk reduction in the composite MACE endpoint in patients treated versus placebo did not reach statistical significance (p=0.08, trend toward significance). Remarkably, in the subgroup of patients with a baseline CACS of >400 (almost half of the study population), MACE was reduced by 42% (p=0.046). This discrepancy between CACS progression and MACE reduction may be explained on the basis of plaque stabilization by statins that involves depletion of the lipid core and reduction in plaque macrophage content and inducing a fibrovascular transformation of thin-cap fibroatheroma (TCFA).² Thus, while the plaque may become less vulnerable and plaque volume may regress, the relative proportion of calcium may actually increase accounting for these observations.

Continued increased of CAC extension may indicate inefficacy of the lipid-lowering therapy and a rapid increase in score is associated with incident of myocardial infarction (MI)^57,63 beyond conventional cardiovascular risk factors and current algorithm models for CHD risk prediction.⁶⁴ Conversely, those with baselines scores of <100 and progression of <15% are at lower risk.⁵⁷ In a study of 422 patients, the highest rate of conversion from a CACS = 0 was in the fifth year and was nonlinear (i.e. 15% cumulative in the first 4 years and 25% in the fifth year), suggesting that 4 years might be the warranty period for a CACS = 0, a period longer than the 2 year “warranty period” that is generally espoused for a normal stress perfusion study.⁴⁵ However, at this time, repeat CACS testing is considered inappropriate for general use because of lack of data from large scale randomized trials showing that this strategy reduces adverse events.⁶⁴

Extracoronary Calcium

CCT is also able to identify non-coronary calcification. One example of this is mitral annular calcification (MAC), which has been shown to increase the risk of MACE. MAC increases with age and hypertension and is more common in females.⁶⁸ On transthoracic echocardiography, the prevalence of MAC in women >65 years of age without evident cardiovascular disease is as high as 40%.^68,69 MAC is also correlated with CACS,⁷⁰ CAD⁷¹ and vulnerable plaque⁷² and increases the risk of MACE.⁷³ Other cardiac and vascular foci of calcification such as aortic valve calcification (AVC), intramyocardial calcification in healed infarct scars, and thoracic aortic calcium (TAC) also provide additional risk information. However, in a recent analysis from the MESA trial, the addition of TAC, AVC or MAC to FRS did not improve cardiovascular risk prediction and that CACS alone was the strongest factor to predict incident adverse events.⁷⁴

Plaque Characterization

Since CACS does not detect non-calcified plaques, contrast-enhanced CCTA is more ideally suited for the non-invasive detection of non-calcified components of coronary atherosclerotic plaques, and in the classification of plaques along the spectrum of stable to vulnerable.^75-79 Current CT scanners are capable of visualizing not only the lumen for coronary stenosis assessment but also the vessel wall and provide information on plaque size, length, volume, geometry, composition and adverse features. CCTA has shown its ability to effectively rule out obstructive and nonobstructive CAD with an exceptionally high NPV, a finding associated with very low incident MACE rates.^80,81 CT harnesses the different tissue attenuation values of different plaque components such as fibrous tissue and lipid core and calcium to characterize plaque as calcified (CP), noncalcified (NCP) and mixed plaque (MP) (Figure 4)⁸².

Figure 4.

Examples of different patients with the 3 different major plaque types based on CTA appearance and attenuation values. A) Noncalcified plaque B) Mixed plaque which consists of calcific lesions interspersed within noncalcific lesions and C) Calcified plaque wherein most of the plaque is calcified.⁸²

Qualitative plaque measurements such as CP, NCP and MP have been demonstrated to be consistent amongst readers, with good intra- and inter-observer agreement (r>0.8).^83-87 These 3 plaque types by CCTA have different compositions when validated against virtual histology intravascular ultrasound (IVUS-VH), and may be able to effectively differentiate the risk state of specific plaque types. As an example, 32% of mixed plaques in a recent study were associated with TCFAs.⁸⁸ In a recent meta-analysis of 20 studies, CCTA had excellent accuracy in detecting plaque (AUC 0.94), sensitivity of 90% and specificity of 92%.⁸⁶ Plaque area, volume and %-area stenosis on CCTA were similar to IVUS.⁸⁹ CCTA demonstrated slight overestimation of lumen area compared to IVUS by 0.46 mm² (6.7%), likely because of partial volume effects from the contrast filled lumen. Plaques with >10% necrotic core on IVUS-VH had lower attenuation values by CT versus those that did not (43 vs. 93 HU).^84,90 Otsuka reported that CCTA overestimated calcified plaque volume by only 3% but underestimated NCP by 17%.⁹¹ Voros et al^84-86 showed that lumen area was overestimated by 22% on 64 slice CT compared to IVUS. Similarly, vessel area, NCP area and CP area were also overestimated by 19%, 44% and 88% respectively. NCP by CCTA correlated best with necrotic core plus fibrofatty tissue.

CCTA is also reproducible for the quantification of arterial luminal diameter and area, plaque size, volume, geometry and features. Although there remains some variability at smaller plaque volumes (∼10 mm³) due to the limited spatial and temporal resolution of current generation CT, CCTA measurements are very robust with larger plaque volumes (∼100 mm³).⁹² Our group has reported on the inter-observer variability of plaque quantification measures, demonstrating good correlation between different observers.^93,94 To further improve reliability of measurements and make it less time consuming, standardized automated software is capable of reducing inter-observer and improving the limits of agreement.⁸⁷ CT generally overestimates the minimum lumen area (MLA) by 27% but area stenosis is only underestimated by 5%.⁸⁶

In a comparison of CCTA to optical coherence tomography (OCT), Kashiwagi et al⁹⁵ observed several indices of plaque instability by CCTA to be associated with TCFAs, including increased remodeling indices in TCFAs (1.14 vs. 1.02) and lower attenuation values (35 vs. 62 HU). In addition, ring like enhancement (also referred to as a “napkin-ring” sign)—characterized by contrast accumulation in the periphery of the plaque but not in the central core—is observed in 44% of plaques with TCFA vs. only 4% in plaques without TCFA. As opposed to these technologies (IVUS and OCT) which allow morphological assessment, NIRS (near infrared spectroscopy) uses the infrared light absorption spectrum of cholesterol to identify the cholesterol content of plaques in the artery. CCTA has shown good correlation of plaque features compared to this newer technology also. Voros et al⁸⁴ demonstrated good spatial correlation between cholesterol localization on the block chemogram on NIRS and the presence of NCP as well as the plaque burden on CCTA. The advantage of CCTA is that it is noninvasive and lends itself to serial follow up to assess changes in plaque geometry and composition as compared to the other 3 invasive modalities.⁹⁶

Prognosis

Given the relationship of CCTA artery and plaque features to measures of plaque stability, an abundance of recent data have examined the prognostic utility of CCTA-identified stenosis and plaque constituents. In a recent meta-analysis of 18 studies with 9,592 patients with suspected CAD, CCTA robustly predicted MACE as well as all-cause mortality over a median follow up of 20 months.⁹⁷ When stratified according to tertiles of no stenosis vs. <50% stenosis and >50% stenosis, the annualized rates of MACE was 0.17% vs. 1.4% vs. 8.8%. Rates for all-cause mortality were 0.15% vs. 0.75 vs. 2.2%. Further similar results have been reported for MACE in a large, prospective, multinational, multicenter CONFIRM (COronary CT Angiography EvaluatioN For Clinical Outcomes: An InteRnational Multicenter) study of 15,187 patients without prior known CAD.⁹⁸ CHECK THIS ONE Amongst these patients, 595 MACE events (3.9%) occurred at a 2.4±1.2 year follow-up. In multivariable analyses, an increased risk of MACE was observed for both non-obstructive (hazard ratio [HR] 2.43, p <0.001) and obstructive CAD (HR 11.21, p <0.001) when compared with patients with normal CCTA. Risk-adjusted MACE increased in a dose-response relationship based upon the number of vessels with obstructive CAD ≥50%, with increasing hazards observed for non-obstructive (HR 2.54, p <0.001), obstructive 1-vessel (HR 9.15, p <0.001), 2-vessel (HR 15.00, p <0.001), or 3-vessel or left main (HR 24.53, p <0.001) CAD. Among patients stratified by age <65 years versus ≥65 years, older individuals experienced higher risk-adjusted hazards for MACE for non-obstructive, 1-, and 2-vessel, with similar event rates for 3-vessel or left main (p < 0.001 for all) compared to normal individuals age <65 years.

Our group has also examined the prognosis of CCTA luminal stenosis by a modified Duke coronary artery disease index, which integrates not only stenosis severity but also the distribution and location of plaque.⁸²Using this measure, the stenosis findings by CCTA offer improved risk stratification. Survival worsens with higher risk Duke score (96% survival for 1 stenosis>70% dropping to 85% survival for LM stenosis>50%).

Measures of stenosis severity as a function of prognosis transcend conventional definitions of anatomically severe. Lin et al⁹⁹ studied the presence and extent of nonobstructive CAD and showed that the presence of any plaque increased the risk of death beyond clinical risk factors (Figure 5).⁹⁹ Hazard ratios for all-cause mortality over 3 years in this study increased from 1.98 for the presence of any plaque, to 5.12 when >5 segments were involved with nonobstructive plaque.

Figure 5.

Presence of nonobstructive coronary artery disease is associated with poorer survival than patients who have no visible plaque on coronary CT angiography. In this study of 2,583 patients, the presence of any nonobstructive plaque carried a 1.98 fold higher risk of death at 3.1 years, compared to the group without any plaque. This risk increased further if more vessels were involved – when all 3 major epicardial coronaries showed non-obstructive plaque, the hazard ratio for death increased to 4.75, after adjustment for CAD risk factors.⁹⁹

In addition to measures of stenosis severity, CCTA offers prognostic power by atherosclerotic plaque features. As an example, culprit lesions at the time of acute coronary syndromes demonstrate higher plaque volume, remodeling indices, larger vessel areas, less overall calcification, more low attenuation plaques (<30 HU), spotty calcifications and ring-like enhancement patterns.^100-104 In a prospective study of 1,059 patients undergoing CCTA for evaluation of chest pain, 4% of patients possessed plaques with both low attenuation plaque and spotty calcification. Amongst this subset, 22% developed acute coronary syndrome (ACS) at 27 months of follow up.¹⁰⁵ In contrast only 0.5% of patients without either of these 2 features developed ACS, providing preliminary evidence that these atherosclerotic plaque features are infrequently seen in stable outpatients but when seen, do portend a worse downstream event rate.

CCTA offers prognostic value beyond that of CAC.¹⁰⁶ This is true even in patients with a zero CACS.²⁶ Furthermore, the burden of atherosclerotic involvement byCCTA correlates well with mortality and cardiac event rate. The worst prognosis is observed in left main and in 3 vessel coronary disease and then there is a graded decrease in risk with proximal left anterior descending artery (LAD) disease and further lower risk with single vessel disease.⁸² Nonetheless, even patients with nonobstructive disease (<50%) fare worse than those with normal CTAs indicating that the total plaque burden bears a close relationship to cardiac event rate and prognosis.⁹⁹ The segment involvement score and segment stenosis score and Duke CT jeopardy score are three different ways of quantifying the degree of atheroma burden and stenosis respectively in the entire coronary arterial tree.

CONFIRM is a 27,000 patient international CCTA registry involving 12 centers in 6 countries that is tracking patients who underwent a clinically indicated CTA between 2005 and 2009 and are being followed prospectively for cardiac events.¹⁰⁷ It has yielded a wealth of information regarding the relationships and interactions between risk factors and scoring systems (including FRS), lipid levels and the extent and distribution of plaque as well as clinical events. Given its unparalleled large patient cohort size and prospective design with cumulative follow up, it exists as a very robust database to evaluate the relationships between various aspects of CAD including gender, lipids, treatment patterns and prognosis. Its limitations are that it is not a randomized comparison but rather an observational registry and is therefore subject to referral and treatment biases.

Serial Progression

Data on serial coronary plaque progression evaluation by CCTA are scant but general thematic lessons are emerging. Early studies have determined that within 1-2 years, there is an observable increase in plaque volume and extent by CCTA. In a single center study of 50 patients followed for 17 months who were not treated with medical therapy for CAD, non-calcified plaque volume in the left main and proximal LAD increased from 91 to 115 mm³.¹⁰⁸ In contrast, in a separate study of patients on statin therapy, no significant changes in overall plaque volume were observed at 18 months. These neutral findings were nevertheless accompanied by a decrease in the non-calcified portion of the plaque from 42 to 30 mm³.¹⁰⁹ Concordant with this study is another investigation that comprised patients without any specific intervention to assess the natural history of plaque progression, demonstrating an increase in noncalcified plaque from 27% at baseline to 35% at 1 year of follow up, with a decrease in minimum luminal area.⁸⁷ This study advances the notion that the “percent of atheroma volume” may be a better parameter to evaluate changes in plaque volume over time relative to the extent of arterial remodeling, and may be a useful adjunct to conventional measures of luminal stenosis severity.

Role of CCTA in Symptomatic Patients with Acute Chest Pain

In patients presenting to the emergency department with acute chest pain, the evidence for CCTA has grown rapidly and now includes 4 randomized trials that have convincingly shown that a negative CCTA is associated with a very low cardiac event rate, shorter length of stay and lower rates of hospital admission as well as cost savings (Goldstein et al,¹¹⁰,CT-STAT,¹¹¹ ACRIN-PA,¹¹² ROMICAT II¹¹³). As such, CCTA allows busy hospital EDs to safely discharge such patients as opposed to the conventional strategy of admitting these patients for a ‘rule-out MI protocol’ followed by stress testing, all while decreasing resource utilization and improving efficiency. Given these data, the appropriate use criteria now classify CCTA as an appropriate test for such patients who have a low to intermediate pretest probability of having CAD.¹¹⁴

Future Developments in CT

The recently completed HeartFlow analysis of coronary blood flow using CT angiography: Next steps (HeartFlow NXT) trial¹¹⁵ enrolled 254 patients and demonstrated the potential of FFR_CT to correctly identify patients with ischemia better than a purely anatomy based evaluation. All patients underwent CTA, fractional flow reserve derived from CT angiography (FFRCT), invasive angiography and also invasive FFR as the gold standard. FFR_CT had a specificity of 79% to rule out ischemia compared to 34% for CTA and 54% for ICA. Sensitivity to detect ischemia was 86%. Similar to the Determination of Fractional Flow Reserve by Anatomic Computed Tomographic Angiography (DeFACTO) trial, the AUC for FFR_CT was approximately 20% better at 0.82 versus 0.63 for CCTA signifying the ability of FFR_CT to better discriminate lesion-specific ischemia. A strategy of revascularization based on fractional flow reserve (FFR) evaluation has been shown to decrease cardiac event rate, mortality and reduce costs by up to $3000 per patient.^116-118. Other advantages of FFR_CT are that it doesn't require additional contrast and analysis can be performed on the same dataset as acquired during the CCTA without any additional modification of the scanning protocol.

Summary

In recent years, CT has emerged as a robust technology with demonstrable utility in the evaluation of asymptomatic and symptomatic patients with suspected CAD. A plethora of population-based studies have confirmed the usefulness of CACS to offer improved risk stratification, discrimination and reclassification over conventional risk assessment of CAD. CCTA has been introduced as a more contemporary adjunct to CACS, offering the ability to determine measures of stenosis severity and atherosclerotic plaque features. In large-scale studies, these findings may enable improved risk stratification than even CACS, and may hone the ability to identify individuals at risk of incident myocardial infarction or death.

Abbreviations

CCT

Cardiac computed tomography

CAD

Coronary artery disease

EBCT

electron beam CT

MDCT

multidetector row CT

CACS

coronary artery calcium scoring

HU

Hounsfield Units

MACE

major adverse cardiac events

FRS

Framingham Risk Score

NRI

net reclassification index

LDL

low density lipoprotein

TCFA

thin-cap fibroatheroma

MAC

mitral annular calcification

AVC

aortic valve calcification

TAC

thoracic aortic calcium

CCTA

coronary CT angiography

CP

calcified plaque

NCP

noncalcified plaque

MP

mixed plaque

IVUS

VH-virtual histology intravascular ultrasound

OCT

optical coherence tomography

MLA

minimum lumen area