SCCT 2021 Expert Consensus Document on Coronary Computed Tomographic Angiography: A Report of the Society of Cardiovascular Computed Tomography

Since the recognition that coronary artery stenoses can produce chest pain, the imperative has been to identify through non-invasive testing both the patients whose chest pain is ischemic in etiology, and, with a view towards revascularization, the arteries and specific stenoses that are responsible for the ischemia. To fulfill this need, testing has evolved from simple exercise treadmill test (ETT) to (a) Measures estimating myocardial blood flow changes: myocardial perfusion imaging by single photon emission computed tomography (SPECT), positron emission tomography (PET), magnetic resonance imaging (MRI), (b) Measures detecting the functional consequence of reduced myocardial blood flow: stress echocardiography (SE), (c) Anatomic Imaging: CTA, and finally (d). Combination of anatomic coronary imaging coupled with physiology or perfusion: CTA derived fractional flow reserve (FFR_CT) and CTP. How these modalities compare with each other has important implications for diagnostic strategies.

The gold standard for determining ischemia has also evolved from percent diameter stenosis (DS) on invasive coronary angiography (ICA) to more physiologic measures, such as invasive fractional flow reserve (FFR) that better reflect coronary blood flow and inducible ischemia. Using DS as a reference standard often provides an inaccurate assessment of ischemia. For instance, when compared to invasive FFR ≤0.80, the sensitivity of ICA is 69%, and the specificity is 67%. Although invasive FFR was initially validated by functional noninvasive testing (SPECT and SE), this method has become a universally accepted gold standard by virtue of its strong association with outcomes.^{,  ,}  Nonetheless, %DS continues to be used much more often than invasive FFR before percutaneous coronary intervention (PCI), - In the ALKK Registry in Germany, FFR was performed in only 3.3% of 40,160 patients undergoing ad hoc PCI from 2010 to 2013. There has been an increase in invasive FFR use in the US, from 8.1% in 2010 to 30.8% in 2014, in a registry of 397,737 patients undergoing nonacute PCI. Consequently, the non-invasive imaging modalities will be compared to both %DS and FFR. The best level of evidence is provided by meta-analyses, which will serve as the basis for comparisons, with the exception of 2 recent single center studies not included in meta-analyses. The meta-analyses included patients with and without confirmed CAD and did not draw distinctions between them.

The National Cardiovascular Data Registry suggested that functional testing is suboptimal for detecting significant coronary stenoses. Of the 661,063 patients undergoing elective catheterization, 64% had testing before the invasive coronary angiogram (ICA); of those, only 51.9% were abnormal. The percentages of patients with <50% DS on subsequent ICA ranged from 55 to 56% after an abnormal exercise treadmill test (ETT), stress echocardiography (SE), single photon emission computed tomography (SPECT) and magnetic resonance imaging (MRI); for resting CTA, the percentage was 30%. In the oldest report, Fleischmann et al. evaluated 5874 patients in 41 studies from 1990 to 1997, and reported sensitivity and specificity of 85% and 77% for SE and 87% and 64% for SPECT, with 52% and 71% for exercise ECG. DeJong et al. (Table 2A), in a meta-analysis of 5088 patients in 51 studies from 2000 to 2011 evaluated MRI, SE and SPECT with >50%DS by ICA as reference. MRI was the most sensitive and specific (91% and 80%), with SE (87% and 72%) and SPECT (83% and 77%) roughly similar. Jaarsma et al. (Table 2B ), reported on SPECT, MRI and positron emission tomography (PET) in 141 per-patient studies and 70 per-vessel studies. Per-patient diagnostic odds ratio (DOR) was highest for PET (36.47) followed by MRI (26.42) and SPECT (16.31). In per-vessel analysis, PET and MRI were equal (24.74 and 24.11), while SPECT was lowest (11.75). In a meta-analysis limited to 26 studies in which CTA was compared to either ETT or SPECT in the same group of patients, Nielsen et al. (Table 2C) reported CTA sensitivities of 95–99%, specificities of 68–93% and DOR of 128–728. Corresponding ranges for ETT were 65–70%, 24–60% and 0.7–4 and for SPECT were 67–73%, 48–52% and 2–4. It is important to understand that available meta-analyses are also challenged by the small numbers of patients in some of the individual reports, potential referral bias, and often include a mixture of newer and older technology (e.g., planar and SPECT imaging). Finally, in a paper published too recently for meta-analysis inclusion, 391 symptomatic patients, 52% with intermediate and 46% with high risk pre-test probability, who were scheduled for ICA, underwent both CTA and SPECT with >50%DS by ICA as reference. Sensitivity, specificity, positive and negative predictive values were 0.92, 0.75, 0.84 and 0.87 for CTA and 0.62, 0.68, 0.74 and 0.55 for SPECT. AUC was significantly higher for CTA (0.91 versus 0.69, p < 0.001.

There have been several recent meta-analyses of the correlation between noninvasive testing and Invasive FFR ≤0.80. Takx et al. (Table 3A) compared multiple myocardial perfusion imaging modalities to FFR in 2048 patients and 4721 vessels in 37 studies. They reported the highest areas under the receiver operator characteristic curve (AUC) per patient for CTP (0.93), PET (0.93) and MRI (0.94) compared to SPECT (0.82) and SE (0.83). Similarly, the highest per vessel sensitivities were for MRI (89%), CTP (88%) and PET (84%) compared to SE (69%) and SPECT (74%). Specificities were similar for all modalities, ranging from 79% for SPECT to 87% for PET, with 80% for CTP and 84% for SE and MRI.

A second meta-analysis, analyzing 3798 patients and 5323 vessels in 23 studies, by Danad et al., (Table 3B), excluded studies in which <75% of vessels were evaluated by FFR, included CTA >50% diameter stenosis and ICA >50%DS and excluded PET, for which there were not sufficient numbers after excluding studies with <75% of vessels having invasive FFR. Sensitivity was highest for CTA and MRI in both per patient (90%) and per vessel (91%) analyses. SPECT sensitivity was the lowest of the functional tests for both patients (70%) and vessels (57%) while SE was also suboptimal (77%). ICA sensitivity was dramatically lower (69%) than for CTA even though both depict coronary anatomy. Specificity was highest for MRI for both per patient (94%) and per vessel analysis (85%), followed by the other 2 functional modalities of SPECT and SE in the 75–78% range. CTA specificity was remarkably lower (39%) than both the functional tests and ICA (66%). The likelihood ratios and AUC reflect these differences; MRI was superior for both positive and negative likelihood ratios and AUC. CTA negative likelihood was excellent as well but had the lowest per patient and per vessel positive likelihood ratio and AUC. Comparison of the anatomical modalities indicates that %DS is overestimated by CTA and underestimated by ICA, explaining the higher sensitivity and lower specificity for CTA.

A third meta-analysis of all the functional imaging modalities with considerably more patients, by Dai et al. (Table 3C) of 74 studies, included CTFFR and CTP and excluded solely anatomic CTA. As before, CTP, CTFFR CMR and PET had superior per patient sensitivity (88–90%), specificity (84–87%) and DOR (41–57). The 2 most frequently performed functional imaging modalities of SE and SPECT were the least accurate: 69% and 78% sensitivity, 77% and 79% specificity, and 7.40 and 13.40 DOR.

Finally, in the PACIFIC trial, a single center study of 208 patients who underwent CTA, SPECT, PET and ICA with FFR, CTA was 90% sensitive, 60% specific and 74% accurate, compared to 87%, 84% and 85% for PET and 57%, 94% and 77% for SPECT.

CT has 2 additional advantages in diagnosis and management of chronic stable CAD. It can prognosticate very well^{,  ,  ,  ,}  , and has the unique ability to identify adverse coronary plaque characteristics that portend adverse risk^{,  ,  ,  ,  ,  ,  ,  ,  ,  ,}  and might even influence the occurrence of ischemia (). Some of the newer value added technologies like CT-FFR and CTP (^,  ) have now been shown to improve the accuracy of CAD diagnosis over and above CTA alone.

Addition of physiologic studies to anatomic information in the same CT scan improve test performance.^, The meta-analysis by Gonzalez et al., of 1535 patients in 18 studies, compared CTA, CTP and CT-FFR. Per patient sensitivities were similar (90–94%), but specificities (43%, 77% and 72%) and DOR (9.17, 63.42 and 24.34) were lowest for CTA without a functional imaging component. Per-vessel results were much less disparate, with sensitivities of 89%, 83% and 83%, specificities of 65%, 76% and 77%, and virtually identical DOR of 19.78, 20.10 and 18.21. A more recent meta-analysis (5330 patients) comparing CTA, CTP and CT-FFR also showed improved efficacy for diagnosing hemodynamically significant CAD compared with CTA alone with higher vessel level, pooled specificity with CTP (0.86; 95% confidence interval [CI]: 0.76 to 0.93), and CT-FFR_CT (0.78; 95% CI: 0.72 to 0.83) than that of CTA (0.61; 95% CI: 0.54 to 0.68); addition of either FFR_CT, or CTP to CTA improved specificities (0.80–0.92) and superior diagnostic accuracy for CTP, FFR_CT, and combined CTA and CTP, compared with CTA. On-site FFR performed as well as off-site FFR and dynamic CTP was more sensitive (0.85 vs. 0.72), but less specific (0.81 vs. 0.90) than static CTP.

With few exceptions, these meta-analyses represent a compilation of prospective and retrospective single center studies with their implicit biases and general lack of direct inter-modality comparisons in the same group of patients. Nonetheless, they offer the most comprehensive evaluation by virtue of their large numbers, and the similarities of the findings irrespective of the inclusion criteria for the meta-analyses.

CTA is a diagnostic test and its accuracy has been established for the diagnosis of coronary artery disease (see section 2.1). It is important to distinguish between its diagnostic accuracy for atherosclerosis, obstructive coronary artery disease and angina pectoris due to coronary artery disease. Clearly, the latter also relies on the patient history and the clinical context. The SCOT-HEART,^, Cardiac CT for the Assessment of Pain and Plaque (CAPP), the Computed Tomography versus Exercise Testing in Suspected Coronary Artery Disease (CRESCENT 1), and CRESCENT II and Min et al. trials directly assessed the influence of CTA on the diagnosis of stable chest pain that was suspected to be due to coronary artery disease. All studies found that CTA was superior to functional testing or standard of care, with the SCOT-HEART trial reporting a 2-fold increase in diagnostic certainty compared to standard of care. Whilst the frequency of the diagnosis of coronary artery disease rose in all trials, the diagnosis of angina pectoris due to coronary heart disease tended to fall in the SCOT-HEART trial perhaps reflecting the absence of obstructive disease in those who were initially presumed to have angina.

The effect of CTA on subsequent clinical management is highly dependent on the population studied. In the SCOT-HEART, CAPP and CRESCENT trials, the study population consisted of patients specifically referred for the evaluation of chest pain suspected to be due to coronary artery disease, with a high prevalence of obstructive coronary artery disease. In these trials, rates of invasive coronary angiography were either reduced or unchanged. However, documentation of obstructive coronary artery disease was more frequent at the time of invasive coronary angiography, which led to a modest increase in coronary revascularizations in the short term trials. In the 5 year follow up of the SCOT-HEART trial, the apparent early increases in coronary angiography and coronary revascularization were offset by later reductions in both invasive angiography and coronary revascularization; by 5 years there was no difference in these procedures. Indeed, beyond the first year, CTA was associated with less invasive coronary angiography (hazard ratio, 0.70; 95% CI, 0.52 to 0.95; p = 0.022) and coronary revascularization (hazard ratio, 0.59; 95% CI, 0.38 to 0.90; p = 0.015). This suggests that the right patients are identified early and treated more promptly, thereby preventing progression of disease and avoiding later reinvestigation and revascularization.

Both the CAPP and the CRESCENT trials^, were designed to assess the influence of CTA on angina symptoms in comparison to a functional testing strategy. They reported reduced levels of angina after 12 months of follow-up. Similar improvements in symptoms were seen in the SCOT-HEART trial, especially in those demonstrated to have normal coronary arteries or those with obstructive disease who underwent coronary revascularization.

The SCOT-HEART and PROMISE trials were sufficiently large to assess the impact of CTA on hard clinical outcomes.^,, The PROMISE trial had a large composite clinical outcome that included all-cause mortality as well as coronary events (myocardial infarction and unstable angina). Although there was no difference in this primary outcome, CTA appeared to be associated with a lower rate of death or myocardial infarction at 12 months. Meta-analysis has reported reduced rates of myocardial infarction with CTA (hazards ratio, 0.69 [95% confidence intervals, 0.49 to 0.98]) but no effect on overall mortality. Similar reductions in myocardial infarction have also been reported in a large (n = 86,705) observational Danish registry (hazards ratio, 0.71 [95% confidence intervals, 0.61 to 0.82]). The 5-year outcome data from the SCOT-HEART trial have now confirmed these earlier promising results: hazard ratios were 0.59 (p = 0.004) for CTA compared to standard of care for the primary endpoint of death from CAD or nonfatal myocardial infarction and 0.60 for nonfatal myocardial infarction alone, without overall differences in ICA or revascularization.^,

Physiologic Basis of CAD Severity. Current standards for determining the physiologic severity of CAD are invasive fractional flow reserve (FFR) and non-invasive coronary flow reserve (CFR). Both derive from experimentally defined CFR, stenosis pressure flow fluid dynamic equations, pharmacologic stress, integrated anatomic dimensions to predict pressure gradient or relative stenosis flow reserve and FFR. Evolution from experimental to clinical applications paralleled advancing invasive and non-invasive technologies, and were validated clinically using pressure flow velocity wires, and quantitative positron emission tomography (PET). PET and/or MRI allow assessment of coronary flow reserve (CFR), coronary flow capacity (CFC), and stress MBF myocardial blood flow cc/min/g (MBF. On the other hand, functional assessment can be made using angiograms, including CTA measurements of FFR (FFR_CT), quantitative coronary angiogram FFR (FFR_QCA), quantitative flow ratio (QFR), and stenosis flow reserve (SFR). CT-FFR chiefly reflects the degree of stenosis but is also affected by the size of the coronary arteries as well as the mass of ventricular myocardium subtended by the stenosis bearing vessel and both these parameters can be accounted for by CT, which may make this parameter more meaningful.

The relative merits of these metrics depend on personal preference and available technology for invasive versus non-invasive, or directly measured physiology versus anatomy-based calculations. However, two universal characteristics provide objective comparisons and insights into the final criteria of patient benefit. The first is testretest precision defined by the standard deviation of repeat serial measurements in the same subject at the same time. The second is the imprecision of the critical threshold of any metrics in relation to the net benefit of revascularization strategy over medical therapy. Substantial literature shows the utility of CT-FFR.^{,  ,} 

Inclusion of FFR_CT. This metric involves an integration of computational fluid dynamics, in addition to the anatomical data from coronary CTA, to allow the calculation of a 3-dimensional pressure map (Fig. 3). To facilitate FFR_CT, CTA should be performed according to best practice guidelines with heart rate control and administration of sublingual nitroglycerin. Not all CTA examinations are of adequate quality for FFR_CT analysis, with artifacts related to misalignment and motion resulting in a higher likelihood of erroneous FFR_CT analysis. In clinical practice, 4–10% of CTA examinations are of insufficient quality for analysis.

There have been numerous diagnostic accuracy studies assessing FFR_CT compared to invasive FFR. The most recent is the PACIFIC sub-study which showed FFR_CT to be the most accurate modality for the discrimination of lesion specific ischemia, with significant improvement in accuracy compared to CTA, SPECT and PET alone. The area under the receiver operating characteristic curve (AUC) for identification of ischemia-causing lesions was 0.94 in comparison with coronary CTA (0.83, p < 0.01), SPECT 0.70, p < 0.01 and PET (0.87, p < 0.01). The diagnostic accuracy of 46% for FFR_CT in the “grey zone” of 0.70–0.80 has raised concerns, although the FFR_CT sensitivity of 87% for invasive FFR values of 0.70–0.80 (92) may be reassuring.

In addition to accuracy data there is growing evidence of clinical utility; the recently published 90-day outcome data from the ADVANCE (Assessing Diagnostic Value of Non-invasive FFR_CT in Coronary Care) registry with over 5000 subjects undergoing FFR_CT, demonstrated significant changes to clinical management, with more refined determination of revascularization versus medical management. Building on the 90 day experience, the 1 year clinical outcomes of the ADVANCE registry were recently published highlighting the good prognosis associated with a negative FFRct (>0.80) with significantly lower CV death and MI rate amongst those participants as compared to those with a positive FFR_CT. Moving beyond outpatient testing for stable CAD, CTA/FFRct to guide decision making in more complicated CAD was evaluated recently in both the SYNTAX II and III trial as an aide to guide complex coronary revascularization. The SYNTAX II study demonstrated that FFR_CT may enable improved treatment decision making in patients with complex multivessel CAD compared to CTA alone. The findings highlighted the ability of CTA with FFR_CT to generate a non-invasive functional Syntax score that correlates with the invasive gold standard. Introducing the derived functional data appears to moderate the disease overestimation based on anatomy alone, with good correlation between the invasive and non-invasive functional Syntax score, allowing decision making regarding complex revascularization and safe deferral at one year. Subsequently, the SYNTAX III Revolution trial randomized heart teams to determine revascularization treatment decisions based upon invasive coronary angiography vs CTA with FFR_CT as needed. It documented a high correlation between the two, with a Cohen’s Kappa of 0.82. In clinical practice FFR_CT is also being evaluated in a randomized controlled trial in the UK comparing it to NICE guided standard of care for patients with stable chest pain and will be followed by a larger international trial evaluating FFR_CT against traditional testing algorithms in the outpatient setting. These trials will be important for better definition of the clinical role of FFR_CT. At present, FFR_CT is a reasonable option for informing downstream ICA and treatment planning in patients with moderate to severe single and multivessel disease (30–90% severity) with a limited role in patients with ≥50% left main stenosis or critical triple vessel disease.^, Other approaches reporting similar results, e.g., machine learning without utilizing computational fluid dynamics, have been reported but have not yet received approval. On site techniques might improve adoption of CT-FFR more widely but these are still under development for routine clinical use.^{,  ,} 

Similar to other more established modalities, it is possible to use CT to image myocardial enhancement during hyperemia and identify functionally significant CAD (Fig. 4). Static perfusion protocols acquire a single set of images during the first pass of contrast medium through the myocardium and allow for qualitative differentiation of normal and hypo-perfused myocardium. Static perfusion imaging can be performed on most CT systems, and the radiation dose is comparable to a regular CT angiogram. A number of single and multi-center studies have shown that static perfusion imaging has incremental value over CTA for the detection of hemodynamic CAD.^{,  ,  ,  ,  ,  ,  ,}  Because dual-energy CT offers better differentiation of tissues and contrast-enhancement, these systems may provide more accurate static perfusion imaging.^, For dynamic perfusion imaging a series of (low-dose) datasets is acquired during the passage of contrast medium, from which quantitative perfusion parameters can be derived. Dynamic perfusion imaging requires either a 2nd/3rd generation dual-source or wide-detector CT system for complete myocardial coverage in 1 or 2 acquisitions, and is associated with a higher radiation exposure than static perfusion imaging. Dynamic perfusion imaging correlates well with other functional tests and provides incremental value over CTA alone for the detection of functionally significant CAD.^{,  ,  ,}  Meta-analyses indicate at least comparable diagnostic accuracy for CT-based perfusion imaging compared to other perfusion imaging modalities and perhaps a slightly higher accuracy for dynamic compared to static perfusion imaging,^,, although, no head-to-head comparison has been performed in the same cohort. There are practical advantages to a so-called stress-rest protocol, i.e. lingering contrast from CTA can be avoided by performing the perfusion scan first, yet a rest first protocol makes more sense from a clinical point of view by allowing deferral of the perfusion scan in case of a normal CTA or one showing clearly nonobstructive lesions. In one randomized study, the use of dynamic perfusion imaging after a positive CTA removed the need for noninvasive downstream testing and avoided negative invasive angiograms compared to standard care based on functional testing. A more recent study using modern scanners with whole heart coverage showed that adding stress CTP to coronary CTA better identified functionally significant CAD with only a small additional radiation dose.