Article Text

Download PDFPDF

Non-invasive imaging
New frontiers in CT angiography: physiologic assessment of coronary artery disease by multidetector CT
  1. James K Min,
  2. Jorge Castellanos,
  3. Robert Siegel
  1. Cedars-Sinai Medical Center, Cedars-Sinai Heart Institute, Los Angeles, California, USA
  1. Correspondence to Professor James K Min, Cedars-Sinai Medical Center, Cedars-Sinai Heart Institute, David Geffen UCLA School of Medicine, Los Angeles, CA 90048, USA; james.min{at}

Statistics from

Since the introduction of 64-detector row CT1 ,2 scanners in 2005, coronary CT angiography (CCTA) has developed into an accurate non-invasive method for direct visualisation of coronary arteries, coronary stenoses, and atherosclerotic plaque.3 These CT scanners have embodied a combination of favourable technological characteristics—including adequate volume coverage, high temporal resolution and reasonable spatial resolution—permitting the acquisition of images of the heart with reasonable breath holds, thus providing an advantage over prior generation 16-detector row scanners. Since its introduction, CCTA has experienced rapid adoption into daily clinical assessment of patients with suspected or known coronary artery disease (CAD), with resultant offerings of societal guidance documents such as the American College of Cardiology Appropriate Use Criteria, the American Heart Association Expert Consensus Statements, and Position Statements of the European Society of Cardiology.2 ,4 ,5 Without exception, these guidance documents have focused on the use of CCTA for anatomic assessment of CAD, owing to its previously demonstrated high diagnostic performance.

In this regard, more than 100 studies have been published which have compared CCTA to quantitative coronary angiography as a reference gold standard for stenosis severity. Several pooled meta-analyses have documented the high diagnostic sensitivity and specificity of CCTA for this end point, with performance measures ranging from 91–99% and 74–96%, respectively.6 Subsequent to these single centre studies—which were unvaryingly susceptible to referral, selection, and ascertain biases—three prospective multicentre studies have been published (table 1). In the first of these studies—the ACCURACY (Assessment by Coronary Computed Tomographic Angiography of Individuals Undergoing Invasive Coronary Angiography) trial—230 patients without known CAD underwent CCTA before clinically indicated invasive coronary angiography (ICA).7 This 16-centre US based study identified only 13.9% patients with a per-patient maximal ≥70% stenosis, thus underscoring a general ability for clinicians to identify precisely individuals with high grade anatomic disease, a finding that has been confirmed by a large scale examination of the National Cardiovascular Data Registry in almost 400 000 stable patients undergoing ICA.8 Diagnostic performance of CCTA for measures of sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were 94%, 83%, 48%, and 99%, respectively.

Table 1

Diagnostic performance of CT for anatomically obstructive stenosis in individuals without known CAD

These studies are in direct accordance with the diagnostic performance of newer generation CT scanners—in particular for dual source CT—wherein the multicentre MEDIC (Multicenter Evaluation of Coronary Dual Source CT Angiography in Patients With Intermediate Risk of Coronary Artery Stenoses) trial demonstrated a sensitivity, specificity, PPV, and NPV of 95%, 91%, 71%, and 99%, respectively. In a follow-up European multicentre, multivendor trial by Meijboom and colleagues of 360 symptomatic patients with both acute and stable angina, diagnostic sensitivity, specificity, PPV, and NPV were 99%, 64%, 86%, and 97%, respectively.9 While these studies highlighted the high diagnostic performance of CCTA to detect and exclude anatomically obstructive CAD, the less-than-perfect specificity in both studies also highlighted the general overestimation of coronary stenosis severity with an undue rate of false positive interpretations. Furthermore, in a heterogeneous group of individuals with both known and not known CAD, as was seen in the CorE64 study, sensitivity and NPV were lower (85% and 83%, respectively), which may be expected with increasing disease prevalence.10

CCTA for measures of myocardial ischaemia

While proponents of CCTA have suggested that the generally high diagnostic accuracy of CCTA is sufficient to justify its use in common everyday clinical diagnosis of patients with suspected CAD, others have suggested that the inability of CCTA to discriminate individuals who do or do not have ischaemia is a current drawback to CCTA.11 Prior studies, particularly with myocardial perfusion imaging (MPI) by single photon emission CT (SPECT), have demonstrated a generally acceptable diagnostic performance for identification of patients with anatomically obstructive CAD that is coupled to a prognostic proficiency that is unsurpassed by any other CAD imaging modality.12 Further, by a large single centre registry, there appears to be a threshold above which the application of coronary revascularisation imparts a survival benefit from cardiac specific death.13 ,14 This relationship is observed both for patients without known as well as with known CAD. Supporting these findings, the nuclear substudy of the COURAGE (Clinical Outcomes Utilising Revascularisation and Aggressive Drug Evaluation) trial demonstrated greater ischaemia reduction and reduced rates of major adverse cardiac events (MACE) in patients with stable angina and moderate or severely abnormal myocardial perfusion scintigraphy (MPS) undergoing revascularisation when compared to patients receiving optimal medical treatment alone, although this study was underpowered to make definitive conclusions.15 Both of these studies are hypothesis-generating, and are pivotal to the questions being addressed by the ongoing ISCHEMIA (International Study of Comparative Health Effectiveness with Medical and Invasive Approaches) trial ( NCT01471522).

Despite its high reported diagnostic performance, the ‘real world’ accuracy of MPS is less sanguine. When corrected for referral bias, the sensitivity and specificity to identify individuals with a severe stenosis is only 65% and 67%, respectively.16 Numerous potential explanations exist to account for these findings, as MPS abnormalities can occur not only as a consequence of haemodynamically significant coronary stenoses, but also from diffuse non-obstructive coronary atherosclerosis, abnormal coronary vasomotion from endothelial dysfunction, and microcirculatory dysfunction. Furthermore, as MPS is reliant upon identification of relative differences in perfusion between myocardial territories, its accuracy is compromised in patients with multivessel CAD.17 When employing a fractional flow reserve (FFR) standard for vessel specific ischaemia among patients with multivessel CAD, MPS identifies ischaemic territories correctly <50% of the time, with both under- and overestimation of vessel specific ischaemia in 36% and 22% of cases, respectively.17 Given this, physiologic stress testing has been associated with a high rate of false positive studies in which ICAs are found to be normal.

Nevertheless, there has been considerable discussion regarding the ability of CCTA to identify haemodynamically significant CAD (figure 1). This concern stems from numerous prior studies that have demonstrated a general inability for measures of angiographic stenosis severity to serve as an efficacious barometer of physiologically significant ischaemia. In a series of elegant studies by Gould and colleagues, stenoses as low as 40% were found to be potentially ischaemic, with follow-up studies from the FAME (Fractional Flow Reserve for Multivessel Evaluation) trial substudies demonstrating an unreliable relationship between intermediate lesions (50–70% stenosis severity) and severely obstructive lesions (≥70% stenosis severity) to discriminate ischaemia.17 ,18 Importantly, measures of coronary stenosis severity appear to be additive and synergistic to ischaemia by SPECT-MPI for identification of at-risk individuals who have a worsened prognosis.19 Yet, routine assessment of individuals by multiple tests imparts ‘double dose’ radiation and increased costs, thus mitigating the efficacy of such diagnostic algorithms.

Figure 1

Discordance between anatomy assessment of stenosis severity by CT versus (A) single photon emission  myocardial perfusion imaging and (B) invasive fractional flow reserve (FFR). CCTA, coronary CT angiography.

Physiologic assessment of CAD by CCTA

To establish the evidentiary base for CCTA to assess physiologic consequences of CAD, numerous methods have been espoused for CCTA to distinguish the haemodynamic significance of CAD lesions. Coupled to anatomic CAD assessment, the goals of these techniques are to add to the assessment of coronary stenosis severity for the conception of the ‘holy grail’ of non-invasive CAD, namely, by providing an integrated anatomic–physiologic measure of CAD severity. To date, four methods have been evaluated for this goal: (1) transluminal contrast attenuation gradients20; (2) anatomic plaque features that suggest ischaemia; (3) physiologic stress testing by CT perfusion; and (4) determination of FFR21 from typically acquired CT scans (table 2). The remainder of this review is dedicated to consideration of the scientific evidence base that underlies these methods.

Table 2

Methods of physiologic assessment by CT

Transluminal contrast attenuation gradients

Contrast enhanced coronary artery opacification is reliant upon a multitude of factors, including left ventricular ejection fraction, contrast bolus rates, coronary flow velocity and pressure, iodine concentration, and contrast volumes. During performance of CCTA, iodinated contrast concentration—as measured by Hounsfield units—demonstrates a gradual diminution gradient from the proximal portion of a vessel to its most distal type. If all of the above aforementioned factors were to be kept constant, the use of this gradient has been posited to be useful for serving as a surrogate to resting coronary flow. Several ‘proof of concept’ studies have evaluated this phenomenon, with early initial positive results.

In the first of these studies, Choi et al22 examined the linear gradient of contrast in relation to coronary flow velocity—termed the transluminal attenuation gradient, or TAG—as judged by Thrombolysis in Myocardial Infarction (TIMI) flow grade and corrected TIMI frame count (cTFC) at ICA (figure 2). In 7263 segments from 370 coronary arteries, TAG decreased in accordance to TIMI grade (TIMI 3, −5.1 HU/10 mm; TIMI 2, −8.3 HU/10 mm; TIMI 1, −15.3 HU/10 mm; TIMI 0, −13.1 HU/10 mm) as well as cTFC (<22, 4.6 HU/10 mm; 22–29.9, −4.3 HU/10 mm; 30–41.9, −7.0 HU/10 mm; ≥42, −11.8 HU/10 mm) (p<0.0001 by analysis of variance (ANOVA) for both). A consistent and significant increase of TAG was also observed in relation to the degree and direction of coronary collateral flow, with TAG values ranging from −4.4 HU/mm for Rentrop scores=0 to 0.8 HU/mm for Rentrop scores=3 HU/mm. Chow and colleagues furthered confirmed the feasibility of this concept by examining attenuation gradients at two sites—before and after a coronary stenosis, with Hounsfield units normalised to the aortic contrast opacification.23 This method, termed corrected coronary opacification (CCO) was compared to TIMI flow at the time of ICA. In 104 coronary arteries, CCO differences were observed in arteries with ≥50% versus <50% stenosis. CCO differences were noted to be lower than those with normal TIMI flow (0.41 vs 0.08, p<0.001), with an overall diagnostic sensitivity, specificity, PPV, and NPV of 83%, 91%, 83%, and 91%, with good agreement with stenosis severity (κ=0.75). While demonstrating similar findings, TAG and CCO have significant differences. CCO is dependent on two measurements normalised to the aorta, while TAG relies upon 1 mm increments of the entire coronary vascular bed. In this regard, CCO may be more easily applied in daily clinical practice.

Figure 2

Example of decreasing Hounsfield unit (HU) density across the length of an artery by transluminal attenuation gradient.

Nevertheless, as described above, this technique is highly dependent on a number of scan and patient specific factors, and the variability of such a method when applied to heterogenous patient cohorts should be done with caution. One potentially useful technology improvement for TAG or CCO may lie in the introduction of wide-volume coverage CT scanners, with 256- or 320-detector row.24 These enable acquisition of the entire arterial length within a single heartbeat, thereby potentially reducing the variable that may accompany scans acquired by 64-row CT, wherein the contrast opacification is more dependent on the time required for complete heart coverage (eg, 5–7 s). However, the physiological significance of such a measure has been called into question, given the rest state in which the contrast gradients are derived. Stress induced states with normal resting perfusion have been a mainstay of physiological testing, and questions have arisen as to the feasibility of detecting haemodynamically significant CAD by a rest scan evaluation.

Anatomic plaque features for identification of ischaemia

One diagnostic advantage of CCTA over conventional ICA is the ability to identify atherosclerotic plaque even in the absence of stenosis.25 Furthermore, CCTA enables direct visualisation of coronary plaque and artery changes, including plaque composition, distribution, location, and arterial remodelling. Pertaining to plaque composition assessment, most prior studies have reported CCTA measures as non-calcified, calcified or ‘mixed’ (part non-calcified and part calcified). A limited number of studies have further dichotomised non-calcified plaque into low- and high- attenuation plaques, and have evaluated the presence of low-attenuation plaque (LAP) to describe adverse prognosis and ischaemia. Germane to the latter, Cheng and colleagues examined three atherosclerotic plaque characteristics (APCs), which may imply underlying arterial injury (figure 3).26 These APCs include LAP,20 positive arterial remodelling (PR), and ‘spotty calcifications’ within non-calcified plaque. In 49 patients with a ≥70% stenosis by CCTA who also underwent MPI, APCs were related to regional (per artery) reversible total perfusion deficit (TPD), with ≥3% considered diagnostic of ischaemia. For patients with artery specific ischaemia, the presence of a stenosis with both LAP and positive arterial remodelling was associated with reversible TPD, while the absence of artery specific ischaemia suggested plaques that did not possess these features. Importantly, a dose–response relationship was observed with an increasing number of APCs to artery specific TPD, wherein TPD increased for LAP−/PR−; LAP+/PR− or LAP−/PR+; and LAP+/PR+ (1.3% vs 3.2% vs 8.3%, p<0.001). This exciting proof-of-concept study encourages further examination of the potential of arterial plaque characteristics to specific lesions that may be ischaemia-causing.

Figure 3

Measurement of adverse plaque characteristics, including (A) low-attenuation plaque (LAP), (B) positive arterial remodelling (PR), and (C) ‘spotty calcifications’. CP, calcified plaque; NCP, non-calcified plaque.

Rest–stress CT perfusion for identification of global myocardial ischaemia

Prior preclinical studies have identified the contrast kinetics of iodinated contrast to be similar to gadolinium based agents for cardiac MRI. As such, some have proposed that in addition to rest CCTA, the performance of a pharmacologic stress CCTA may allow for comparisons of the relative myocardial perfusion from CT (figure 4). Recently, numerous single centre studies have demonstrated the feasibility of stress CT perfusion (CTP) for added incremental value for the detection of physiologically significant coronary stenosis. In the first of these exciting studies, George et al27 examined the feasibility of CT based perfusion to assess myocardial blood flow (MBF). In a canine model, these investigators noted the mean MBF to be reduced in stenosed versus non-stenosed myocardial territories (2.5 vs 8.9 ml/g/min, p<0.05), with a significant linear relationship of signal densities to MBF. Subsequent to this, Cury examined CTP using a dual source CT scanner, with improved temporal resolution. In 34 subjects, stress CTP followed by rest CT demonstrated a sensitivity and specificity of 93% and 74%, respectively, with a per vessel NPV of 98%.28 This compared favourably to SPECT-MPI, which demonstrated a sensitivity and specificity of 83% compared to ICA as a reference standard.

Figure 4

(A) High grade stenosis with total occlusion of the left anterior descending artery, and (B) corresponding regadenoson induced stress perfusion study of the basal and mid portion of the left ventricular myocardium.

Since then, numerous protocols for CTP have been evaluated, including whole heart coverage by 320-detector row CTP, time resolved dynamic CTP by dual source CT, and CTP by dual energy imaging.29 Each of these techniques has a number of advantages and disadvantages. Whole heart coverage allows for single-beat acquisition, which enables comparison of all myocardial segments simultaneously, thereby mitigating potential time dependent differences in contrast opacification of myocardial segments. Nevertheless, this technique allows only for a ‘one shot’ assessment of myocardial contrast opacification, thereby precluding the ability to assess absolute MBF. To avoid this potential limitation, time resolved dynamic CTP has been evaluated for its feasibility by assessment of contrast uptake and decrement over time. This technique appears reliable for assessing absolute MBF, but is limited by substantial increases in radiation dose, owing to its requirement for multiple acquisitions of myocardial images.30 With current and future dose reduction techniques, this technique may be a reliable indicator of myocardial perfusion but, at present, should be employed with caution. Finally, technologically feasible dual energy CTP which is either projection based (within the ‘raw’ data of CT image acquisition) or image based (a post-processing technique derived from already acquired images) may represent a technique that improves specificity of CTP by reducing shading artefacts and allowing for material decomposition of absolute myocardial iodine content within myocardial territories. At present, there are two prospective multicentre trials evaluating CTP by 320-detector row CTP and by traditional 64-detector row CTP (NCT00934037 and NCT0133418).

FFR derived from typically acquired CCTA

At present, the ‘gold’ standard assessment of the physiologic significance of CAD is invasive FFR.21 This technique, which uses a coronary pressure wire, involves measuring the coronary pressure proximal and distal to a stenosis at rest and during adenosine induced hyperaemia.31 Adding to the diagnostic accuracy of this technique is a robust evidence base to support its use to result in improved event-free survival. In the multicentre randomised controlled FAME trial, more than 1005 patients with multivessel CAD were randomised to an FFR guided versus stenosis guided revascularisation strategy, with lower rates of MACE32 observed in patients undergoing ischaemia guided revascularisation.31

In recent years, advances in computational fluid dynamics (CFD) have enabled the calculation of coronary flow and pressure from typically acquired static CT images.33 When added to standardly acquired CCTAs, these techniques allow for non-invasive FFR (figure 5). This technique, entitled FFRCT, is rooted in several scientific tenets, including not only the application of CFD solutions to CCTAs, but also accurate patient specific geometric models (derived from the CCTA), application of allometric scaling laws, and prediction of hyperaemia. The application of allometric scaling laws allows for anatomic information to serve as ‘clues’ to the physiologic processes occurring in the coronary arteries; and include ‘form-function’ relationships wherein the mass or size of an object relates to its shape, anatomy and physiology. An example of these ‘form-function’ relationships includes the observation that resting coronary blood flow is proportional—but not linearly—to myocardial mass; and that the resistance of the microcirculatory vascular bed at rest is inversely proportional to the size of the feeding vessel. These calculations would be inadequate were it not for the ability to calculate the hyperaemic state, a computation that is rooted in the scientific evidence base that underlies the use of 140 μg/kg/min for routine pharmacologic stress testing and invasive FFR.34

Figure 5

Example of (A) high grade stenosis of the left anterior descending artery by coronary CT angiography, (B) FFRCT of 0.47, highlighting lesion specific ischaemia, and (C) invasive coronary angiogram with invasive FFR of 0.55, demonstrating lesion specific ischaemia. FFR, fractional flow reserve.

The FFRCT technique has been tested in a single prospective multicentre trial to date.35 In this study, FFRCT was compared to CT stenosis for its ability to identify haemodynamically significant CAD, as referenced to an invasive FFR standard. The performance of FFRCT was substantially superior to CCTA stenosis for the diagnosis of lesion specific ischaemia, with a sensitivity, specificity, PPV, NPV, and accuracy of 88%, 82%, 74%, 92%, and 84%, respectively, for FFRCT; and 91%, 40%, 47%, 89%, and 59%, respectively, for CCTA stenosis. Further, when comparing the susceptibility of FFRCT and CCTA stenosis towards reduced diagnostic performance in the presence of common CT artefacts—such as poor contrast opacification, high calcium content, and motion artefact—FFRCT demonstrated a robustness in accuracy that was not observed for CCTA stenosis. An ongoing larger prospective multicentre trial, entitled Determination of Fractional Flow Reserve by Anatomic Computed Tomographic Angiography (or DeFACTO) has completed enrolment, and results are expected later this year (NCT01233518).


In the past several years, numerous techniques have emerged which have tested the ability of multidetector row CT to assess haemodynamically significant CAD. These techniques include measures of global myocardial ischaemia—such as CTP—as well as lesion specific ischaemia—such as observed with TAG, CCO, APCs, and FFRCT. Future multicentre studies confirming the robustness of these techniques are now needed. Yet if any or all of these techniques prove successful, the addition of them to standard CCTA may allow for an integrated assessment of the entirety of anatomic and physiologic determinants of ischaemia.

Physiologic assessment of coronary artery disease by multidetector CT: key points

  • While the diagnostic accuracy of coronary CT angiography is robust for the assessment of anatomically obstructive coronary artery stenoses in individuals without known coronary artery disease, CT stenosis cannot determine the haemodynamic significance of such lesions.

  • Numerous novel methods—including adverse plaque characteristics, contrast opacification gradients, myocardial perfusion imaging, and fractional flow reserve derived from CT—now enable the physiologic assessment of coronary artery lesions by CT.

You can get CPD/CME credits for Education in Heart

Education in Heart articles are accredited by both the UK Royal College of Physicians (London) and the European Board for Accreditation in Cardiology—you need to answer the accompanying multiple choice questions (MCQs). To access the questions, click on BMJ Learning: Take this module on BMJ Learning from the content box at the top right and bottom left of the online article. For more information please go to:

▸ RCP credits: Log your activity in your CPD diary online (—pass mark is 80%.

▸ EBAC credits: Print out and retain the BMJ Learning certificate once you have completed the MCQs—pass mark is 60%. EBAC/ EACCME Credits can now be converted to AMA PRA Category 1 CME Credits and are recognised by all National Accreditation Authorities in Europe (

Please note: The MCQs are hosted on BMJ Learning—the best available learning website for medical professionals from the BMJ Group. If prompted, subscribers must sign into Heart with their journal's username and password. All users must also complete a one-time registration on BMJ Learning and subsequently log in (with a BMJ Learning username and password) on every visit.


  1. A meta-analysis comparing MACE of radial versus femoral access for coronary procedures. Both procedures had similar MACE, but radial access had lower rates of entry site complications at the expense of higher procedural failure rates.
  2. The European Society of Cardiology recommends that cardiac CT should be considered for cardiac event risk stratification in asymptomatic patients who have intermediate CAD risk factors or have type 2 diabetes mellitus. Myocardial perfusion scintigraphy should only be considered for cardiac event risk stratification in asymptomatic patients who have a first degree relative with premature CAD.
  3. Succinct analysis of published data analysing the strength and limitations of cardiac CT angiography in the evaluation of patients with suspected CAD.
  4. Among the first studies to compare the diagnostic potential of multislice spiral CT in the diagnosis of CAD as compared to invasive coronary angiography.
  5. Among the early prospective multicentre trials that studied the diagnostic accuracy of 64-multidetector row coronary CT angiography in patients without known CAD but who presented to the emergency department with chest pain and were referred for invasive coronary angiography.
  6. American College of Cardiology registry analysis concluding that only approximately one third of patients without known CAD who underwent elective cardiac catheterisation had obstructive CAD.
  7. Among the symptomatic patients with acute or stable angina, CT coronary angiography had a sensitivity of 99% and a specificity of 64%.
  8. Myocardial perfusion imaging and multidetector CT have complementary roles in diagnosing patients with CAD.
  9. This analysis concludes that both coronary CT angiography and myocardial perfusion imaging have complementary roles in assessing patients for CAD.
  10. This observational study of 13 555 patients over an 8.7±3.3 year follow-up concluded that patients with myocardial ischaemia but no prior myocardial infarction had a greater survival benefit from early revascularisation relative to patients with a prior myocardial infarction.
  11. Landmark paper which concluded that patients with a small amount of ischaemia on myocardial perfusion imaging have a greater survival benefit if treated medically, while patients with moderate to large areas of ischaemia fare better with revascularisation.
  12. This paper studied a subset of COURAGE patients and concluded that patients with moderate to severe ischaemia on MPI who underwent percutaneous coronary intervention in addition to optimal medical therapy had a greater reduction in ischaemia compared to optimal medical therapy alone.
  13. Referral bias adjustment for SPECT greatly decreases its sensitivity from 98% to 65% while increasing its specificity from 13% to 67%.
  14. Myocardial perfusion imaging has poor concordance with intracoronary fractional flow reserve measurements.
  15. Early study showing the concordance of invasive angiography and multislice CT for the detection of CAD.
  16. Transluminal attenuation gradient can be used to improve classification of coronary artery stenosis.
  17. Corrected coronary opacification measurements may be useful in the identification of arteries with <50% stenosis or <TIMI 3 flow.
  18. Presence of low-attenuation plaque and positive remodelling on coronary CT angiography are predictive of myocardial hypoperfusion.
  19. Among the first studies to demonstrate that multidetector CT can be used to provide measurements of myocardial perfusion.
View Abstract


  • Contributors JKM designed and drafted the paper; JC and RS provided review of critical intellectual content.

  • Competing interests In compliance with EBAC/EACCME guidelines, all authors participating in Education in Heart have disclosed potential conflicts of interest that might cause a bias in the article. JKM (GE Healthcare, significant research support, TC3 Equity Interest), JC (no disclosures), RS (no disclosures).

  • Provenance and peer review Commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.