Introduction

Functional stress tests have been the imaging cornerstone for the assessment of patients with suspected obstructive coronary artery disease (CAD) for several decades. Nonetheless, the diagnostic yield of invasive coronary angiography (ICA) following abnormal conventional stress testing is low, with only about 40% patients showing obstructive CAD, compared to an approximate 70% of obstructive findings among patients referred from coronary computed tomography angiography (CCTA) [1, 2]. Such a suboptimal yield of ICA, derived from inaccurate risk stratification, or on occasions from misleading functional tests, affects one of the foundations of the management of patients with suspected CAD, which is to contain the number of unnecessary ICA referrals (patients without obstructive CAD who undergo ICA). The suboptimal performance of traditional functional cardiac imaging might also be attributed to the discordance between ischemia and stenosis highlighted by several clinical studies [3, 4]. Besides, the ability of ischemia testing to predict subsequent myocardial infarction (MI) and death is currently being challenged in light of recent studies that will be discussed below.

There is an ongoing debate whether the initial diagnostic test among patients with new onset stable chest pain should be a functional test or CCTA [5]. In this regard, aside from considering the specific characteristics of individual patients such as their clinical likelihood of CAD, ECG interpretability and exercise capacity (Fig. 1), a number of questions arise related to the objective of the initial test: (1) Is it the aim of the test to rule out obstructive CAD (gatekeeper role)?; (2) Is it the purpose to predict the clinical outcome (prognostic role)?; and/or (3) Is it the search for a tool that offers a more rational assessment of the potential benefits and risks of revascularization (decision-making)?. Therefore, the aim this review is to provide a contemporary overview of these issues and discuss the emerging role of CCTA as the upfront imaging tool for most patients with suspected CAD.

Fig. 1
figure 1

Graphical representation of the considerations for the initial diagnostic strategy in patients with stable suspected CAD. *Left bundle branch block, pacemaker, resting ST-segment abnormalities

Full size image

Diagnostic yield and prognostic value of functional vs. anatomic testing in stable patients

A large number of studies using stress-echocardiography, stress-single photon emission computed tomography (SPECT), and stress-cardiac magnetic resonance (CMR) have demonstrated the relationship between the magnitude of myocardial ischemia and major adverse cardiac events (MACE) [6,7,8,9]. In this context, the number of ordered cardiac stress tests has steadily increased in the past decades, with almost 4 million tests performed annually in the United States, a figure expected to increase given the aging population [10]. However, there is a significant decline in the number of tests that are positive for myocardial ischemia in contemporary clinical practice (< 10%) [11, 12]. Likewise, the prevalence of moderate to severe ischemia in SPECT studies has also experienced a major decline, from 20.6% in 1991 to 4.6% in 2009 [11]. Such reduction in the frequency and severity of myocardial ischemia, aside from lower threshold or higher accessibility (resulting in lower risk patients undergoing testing), can be partly attributed to the reduction in the risk factor burden and in the increasing use of statins and beta-blockers, which parallels the significant decline in the rates of MI and cardiac death in the US [13]. Furthermore, the evidence indicating that noninvasive cardiac imaging promotes a significant reduction in the risk of MI or death is conflicting [14, 15].

Regarding the diagnostic yield of non-invasive testing, in a very large registry including 661,063 patients undergoing elective catheterization, the diagnostic yield of ICA for the prediction of obstructive CAD was below 50% [1]. Indeed, among patients with a positive stress test, only 41% showed obstructive CAD, and even among patients with typical anginal symptoms and positive stress tests the diagnostic yield of ICA was below 55% [16]. Notwithstanding, it should be acknowledged though that such registry excluded a large number of patients who entered the National Cardiovascular Data Registry (NCDR) CathPCI Registry such as those with known CAD and sites without diagnostic catheterization results. In addition, the definition of non-invasive tests (NIT) as low, intermediate, and high risk was left to each site interpretation; and such risk stratification is not fully comparable between NITs. Likewise, the fact that they used a low threshold for detecting obstructive CAD (diameter stenosis ≥ 50%) might lead to further discrepancies. Also, almost 60% of the patients were asymptomatic or had atypical symptoms.

The PROMISE (PROspective Multicenter Imaging Study for Evaluation of chest pain), SCOT-HEART (Scottish COmputed Tomography of the HEART), and ISCHEMIA (International Study of Comparative Health Effectiveness With Medical and Invasive Approaches) randomized trials have provided insightful data in this regard. The PROMISE and SCOT-HEART trials compared clinical outcomes of the management of stable symptomatic patients with CCTA (anatomy) vs. standard of care (functional). The PROMISE trial randomized 10,003 symptomatic patients to a strategy of initial anatomical testing with CCTA or functional testing. Although demonstrating similar clinical outcomes compared to standard of care (SOC), CCTA enabled a significantly better diagnostic yield of ICA, by leading to more ICA but with significantly lower rates of non-significant lesions [17].

In an insightful sub-analysis of the PROMISE trial, Hoffmann et al. reported that CCTA provides better prognostic information than functional testing. Furthermore, when findings were stratified into mildly, moderately, or severely abnormal (Fig. 2), CCTA showed a risk continuum compared to a normal test [mild HR 2.94 (95% CI 1.64–5.26); moderate HR 7.67 (95% CI 3.83–15.37); and severe HR 10.13 (95% CI 5.15–19.92), respectively], whereas functional testing did not [mild HR 0.94 (95% CI 0.47–1.89); moderate HR 2.65 (1.46–4.83); and severe HR 3.88 (2.58–5.85)], respectively [18]. Moreover, in terms of prediction of hard events, even mildly abnormal CT exams were associated with death or MI [HR 2.73 (95% CI 1.20–6.25), p = 0.0170], whereas only severely abnormal functional tests were associated with these hard endpoints [HR 2.13 (95% CI 1.16–3.91), p = 0.0141]. It is noteworthy that such risk continuum provided by CCTA was achieved using a categorical approach considering only lesion severity, disregarding plaque characteristics with prognostic relevance that are included within specific scoring systems such as the CT Leaman score and the Leiden CT risk score.

Fig. 2
figure 2

Graphical representation of the estimated prevalence and annual rate of hard events according to the presence of normal, and mildly, moderately, and severely abnormal functional and anatomic (CCTA) tests in patients with suspected CAD [18]. *Although not clearly established, criteria for moderate-severe ischemia is generally defined as ≥ 10% ischemic myocardium at stress-SPECT, ≥ 3/16 newly dysfunctional segments at stress-echo, or ≥ 2/16 ischemic defects at perfusion CMR or ≥ 3/16 newly dysfunctional segments at dobutamine stress-CMR [94]

Full size image

The SCOT-HEART trial included 4146 patients and demonstrated that using CCTA in addition to SOC in patients with stable chest pain resulted in a significantly lower rate of hard events (death from CHD or non-fatal MI) after a follow-up of 5 years compared to SOC alone. Of note, such gain did not result in significantly higher rates of ICA or revascularization, and was mostly attributed to the ability of CCTA to improve targeting of preventive (statin and antiplatelet) therapies [19, 20]. In keeping with this, a very large (n = 86705) Danish registry demonstrated that among patients with suspected CAD, initial evaluation with CCTA was associated with a 30% lower risk of myocardial infarction (MI) compared to patients who underwent initial functional testing [21]. Moreover, a recent meta-analysis including randomized CCTA trials showed that among patients with suspected CAD, CCTA was associated with a 30% reduction in the incidence of MI compared to functional stress imaging, CCTA patients were more likely to undergo ICA and revascularization than those evaluated with functional testing [22]. Taken all together, CCTA appears as the more effective first choice in this clinical scenario.

Accordingly, there is supportive evidence that compared to functional assessment CCTA is a more effective gatekeeper to ICA, providing improved clinical outcome including unsurpassed negative predictive value, and offering a risk continuum for hard cardiac events (Fig. 2), as well as enabling enhanced lifestyle modifications, eligibility, and adherence to statin therapy and aspirin [20, 22,23,24,25,26,27]. In terms of its economic impact, though this issue deserves specific and much more detailed analysis. According to the PROMISE data (US centers) CCTA and functional testing have similar costs at 3 years of follow-up. In contrast, European studies suggest that CCTA might be a more cost-effective strategy [28, 29]. Furthermore, in the recent multinational mostly Asian CONSERVE (Coronary Computed Tomographic Angiography for Selective Cardiac Catheterization) study the cumulative diagnostic test costs were 57% lower among patients randomized to selective (using CCTA as gatekeeper) ICA referral compared to those with direct ICA [2].

As a consequence of the above, the recent National Institute for Health and Care Excellences (NICE) clinical guideline recommended CCTA as the first-line investigation for patients with stable chest pain [30]. The recently released 2019 ESC guidelines for the diagnosis and management of chronic coronary syndromes recommend (class I) the use of either noninvasive functional imaging (particularly among patients with intermediate to high likelihood of CAD) or anatomical imaging using CCTA (particularly among those with low to intermediate likelihood of CAD) as the initial test for diagnosing CAD [31].

Targeting cause or effect: implications for evaluation and treatment

Until recently, the clinical relevance of myocardial ischemia was unchallenged, and most clinical decisions were closely related to the presence, extent, and distribution of inducible ischemia. Indeed, stress-induced ischemia is broadly used as a surrogate of the risk of MACE in patients with suspected and established stable CAD, and generally promotes and expedites the use of myocardial revascularization even in asymptomatic patients [32, 33]. However, the association between myocardial ischemia and CV events, though still under debate given the results of the ISCHEMIA trial, does not imply cause-effect [34]. The fundamentals of such uncertain association rely on different backbone concepts, summarized as follows. To begin with, almost 70% of acute thrombotic events arise from angiographically mild lesions and approximately half of acute MI occur in patients with no history of previous symptoms [35,36,37,38]. Indeed, several studies have documented the presence of high risk plaques as well as plaque rupture outside the culprit lesions as a relatively common finding in both stable and unstable patients [39,40,41,42]. In keeping with this, in the PROSPECT study (Prospective natural-history study of coronary atherosclerosis), half of the subsequent major coronary events were related to non-culprit lesions (with a mean baseline diameter stenosis of 32%), although it is worth mentioning that most of those events were rehospitalization for unstable angina [43]. On the other hand, among stable patients, 57% of cardiac events in the PROMISE trial occurred in patients with normal functional tests [18, 43]. Moreover, in the international CLARIFY (Prospective Observational Longitudinal Registry of Patients With Stable Coronary Artery Disease) registry, of the 469 hard events (CV death or MI), 58% occurred in patients without angina or ischemia [44]. Accordingly, it is difficult to relate ischemia with events.

In parallel, and as a consequence of the above, the optimal management of patients with stable CAD remains controversial, with imaging subanalysis of several trials such as COURAGE (Clinical Outcomes Utilizing Revascularization and Aggressive Drug Evaluation trial), BARI 2D (Bypass Angioplasty Revascularization Investigation 2 Diabetes), and FAME 2 (Fractional Flow Reserve Versus Angiography for Multivessel Evaluation 2) showing lack of benefit of revascularization in terms of reduction of MI or death compared with optimal medical therapy (OMT) [45,46,47]. Indeed, in the FAME 2 study, 16% of patients with normal fractional flow reserve (FFR > 0.80) had subsequent MACE [48]. Furthermore, in the recently published 10-year follow-up of the MASS II randomized trial, the presence of baseline myocardial ischemia was not identified as a predictor of MACE or of changes in left ventricular systolic function among patients with multiple vessel CAD [49]. In the aforementioned CLARIFY registry, that included 32,105 patients with stable CAD, anginal symptoms (with or without demonstration of ischemia) but not ischemia were related to death or MI [44]. Moreover, almost 60% of such events occurred in patients with neither angina nor ischemia, compared to 12% among patients with ischemia, 12% among those with angina alone, and 17% with both. These results are in keeping with the findings of the COURAGE study, where baseline ischemia did not predict events while anatomic extension and severity did [50]. Notwithstanding, none of the aforementioned studies were designed to compare outcomes between revascularization and OMT according to the extent of ischemia.

In turn, the ISCHEMIA trial was the first randomized comparison of a noninvasive ischemia-guided revascularization strategy (invasive vs. conservative) in patients with stable CAD. Of note, ischemia was severe in 54% of patients, moderate in 33%, and mild in only 12% of patients. The trial randomized 5,179 patients to ICA (followed by revascularization if needed) on top of OMT, or to an initial conservative strategy of OMT alone. After a median follow-up of 3.3 years, no differences were found between strategies both regarding the primary endpoint of death, MI, hospitalization for unstable angina, heart failure or resuscitated cardiac arrest [13.3% in the invasive group vs. 15.5% in the OMT group (adjusted HR 0.93; 95% CI 0.80–1.08)]; or to cardiovascular death or MI [11.7% in the invasive group vs. 13.9% in the conservative group (HR 0.90; 95% CI 0.77–1.06] [34]. This long awaited trial was conceived from the uncertainties raised by the COURAGE trial over a decade ago, reporting no benefit of revascularization over OMT in stable CAD. One of the main criticisms of the COURAGE trial, beyond the fact that it included a relatively low ischemic burden, was the outdated stent technology and the many exclusions following the initial angiogram. In contrast, the ISCHEMIA trial used state-of-the-art OMT and revascularization strategies. Also, randomization was performed without an initial ICA.

Will ischemia-driven intervention remain the cornerstone of CAD management?

Although myocardial ischemia has been the foundation of decision-making in patients with suspected CAD for several decades, the demonstration of improved clinical outcomes by means of revascularization, yet currently under debate after the ISCHEMIA trial, has only been observed in patients with large ischemic burden (more than 10% of ischemic myocardium) [51, 52]. As a matter of fact, in the COURAGE nuclear substudy, although a significant reduction of the ischemic burden lead to reduced unadjusted rates of death or MI, adjusted differences (controlling for randomized –revascularization-treatment) were comparable [53]. What might be more relevant, a post hoc analysis including SPECT and quantitative ICA identified anatomic but not ischemic burden as a predictor of death, MI, and non-ST elevation acute coronary syndromes [50].

Furthermore, in a study including 549,078 patients with suspected ischemia across 224 hospitals, centers with the higher rates of noninvasive cardiac imaging were not associated with a decrease in readmission rates for acute MI despite higher rates of ICA [15]. Finally, in a meta-analysis including 5 randomized studies evaluating PCI and OMT vs. OMT alone for stable CAD with documented ischemia, PCI was not associated with reductions in death or MI compared with OMT alone [54].

Overall, ischemia-driven interventions, though improving anginal symptoms, have failed to reduce the risk of hard cardiovascular events, thus disputing the role of ischemia as an optimal surrogate of CV risk and as a therapeutic target in stable syndromes. In contrast, statins, and more recently novel anti-inflammatory drugs and icosapent ethyl have shown a significant impact in the rates of death and MI [55,56,57,58]. Such divergent impact of ischemia-driven vs. atherosclerosis-driven interventions can be at least in part explained by understanding that the former approach targets only the later stages of the actual underlying process. However, the pathophysiology of anginal symptoms is likely more complex, and the degree of focal stenosis represents only one of multiple contributors. This concept was supported by the aforementioned CLARIFY registry, where anginal symptoms overruled ischemia as a predictor of hard events [44].

Accordingly, we are likely at the dawn of a new era, that comprises the convergence of three related developments that will shape our management of CAD in the years to come: (1) the unsatisfactory performance of functional tests as gatekeepers of ICA; (2) Degradation of the prognostic value of myocardial ischemia compared to anatomy relevant for clinical decision-making; and (3) Equipoise between PCI and an initial conservative strategy among most patients with stable CAD in terms of MACE and survival [59].

Ischemic burden or anatomic burden: Why not both?

There are several additional advantages of CCTA over functional tests as the initial strategy for most patients with suspected or established CAD. First, the identification of complete absence of coronary atherosclerosis (Fig. 3) provides an unsurpassed negative prognostic value, enabling an at least 5-year long safety window, with an annualized rate of events lower than 0.25% [60,61,62]. This is a critical discriminating aspect between anatomic and functional testing (Fig. 2), since patients with extensive but non-obstructive disease have a higher risk of events independent of the ischemic burden and clinical features [50]. Additionally, the presence of any plaque, even mild, is associated with all-cause mortality [63, 64]. Secondly, CCTA enables the identification of the presence, extent, and type of plaque (high-risk plaque characteristics, including: positive remodeling, low-attenuation, plaque burden > 70%, napkin-ring sign, and spotty calcification; Table 1), and the spatial distribution of nonobstructive plaque (Fig. 4) [65]. Such portrayal of the atherosclerotic burden has major prognostic implications that has gained clinical relevance in the past few years with the demonstration, as mentioned above, that patients with extensive but nonobstructive disease bear a similar prognosis than those with obstructive but not extensive disease [63, 64, 66,67,68,69]. Third, and of immediate clinical relevance, CCTA allows ruling out left main disease (Fig. 5), a high-risk subset of patients found in 5% of the ISCHEMIA trial population, in whom revascularization (regardless of the ischemic burden) is regarded lifesaving [70].

Fig. 3
figure 3

Sixty-six year-old asymptomatic female, with hypertension. She underwent an exercise (treadmill) stress test, where a brief (four beats) episode of ventricular tachycardia associated with dyspnea and without ST-T changes occurred. A CCTA was requested to rule out CAD, which was normal

Full size image
Table 1 Summary of conventional and additional anatomic and hemodynamic CCTA features with prognostic value
Full size table
Fig. 4
figure 4

Fifty-two year-old female, with hypertension. She had two episodes of typical chest pain associated with palpitations, with normal ECG and enzyme levels. She underwent a stress-echocardiogram, showing inferior-wall ischemia. CCTA demonstrated absence of calcifications and normal LAD (a) and LCX (b) arteries. The RCA (c) showed two non-significant lesions at the mid portion (white asterisk, and d) and at the distal/posterolateral branch ostia (yellow asterisk, and panel e). Albeit mild, both lesions had high-risk characteristics including positive remodeling, a low attenuation core, and napkin-ring sign; as portrayed in cross-sectional views (d and e)

Full size image
Fig. 5
figure 5

Seventy-eight year-old male, with obesity (body mass index 31 kg/m2). He has anginal chest pain and dyspnea. He underwent a rest-stress myocardial perfusion imaging SPECT (90% maximum heart rate, 7 METS), showing normal relative myocardial perfusion (a), preserved post-stress left ventricular function (b) and absence of ST-T changes (c); although ventricular premature beats and non-limiting chest pain were documented at maximal stress. Given the discordant results between SPECT images and the symptom referred by the patient during the treadmill test, an anatomic evaluation was recommended. CCTA revealed complex multi-vessel disease, including severe stenoses of the LMCA (arrows in panels d and e), proximal LAD (d) *, and distal RCA (asterisks in f), including a bifurcation lesion with significant positive remodeling (f, white asterisk). The LCX had moderate proximal stenosis (e, *)

Full size image

In other words, if a stress test shows moderate or severe ischemia one would certainly want to rule out left main disease. Fourth, CCTA has the advantage of identifying alternative causes of exertional and non-exertional chest discomfort such as coronary anomalies (including fistulae), pericardial, aortic, valve, and pulmonary disease, or a hiatal hernia. In this regard, the presence of exertional angina and nonobstructive CAD (INOCA) may have other underlying mechanisms such as microvascular dysfunction or coronary spasm that cannot be defined by CCTA and deserve advanced functional imaging or, in some instances, invasive techniques. In addition, the CT scan allows for the measurement of epicardial fat and the detection of occult (subendocardial) myocardial infarcts, both with independent prognostic value [71,72,73,74].

Nonetheless, despite its very high sensitivity and negative predictive value, CCTA has a relatively lower specificity, particularly in the presence of diffuse calcification. Stress myocardial CT perfusion (CTP) has emerged as a possible solution in this regard, and several studies have demonstrated that CTP might offer a significant incremental value over CCTA, though it demands an additional acquisition thus substantially higher radiation and contrast dose [75, 76]. In contrast, the non-invasive assessment of FFR through computational fluid dynamics obtained from conventional CCTA datasets (FFR-CT, Fig. 6), enables the assessment of the hemodynamic significance of coronary lesions from the same CT angiogram. A number of multicenter clinical studies have demonstrated that FFR-CT provides incremental value over CCTA, particularly by increasing the specificity and reducing the number of unnecessary referrals to ICA [77,78,79,80]. Indeed, in two recent head-to-head comparison studies, the accuracy of FFR-CT was at least comparable to functional tests for the assessment of the hemodynamic significance of lesions [76, 81]. In addition, computational fluid dynamics applied to CT angiograms can elucidate the mechanical forces onto individual plaques, including endothelial shear stress and axial plaque stress, which are associated with adverse outcome (Fig. 6) [82]. Non-invasive assessment of the hemodynamic stress on plaques may identify lesion-specific precursor of an acute coronary event, and improve the modest positive predictive value (< 20%) of the currently available high-risk (anatomic) plaque characteristics [43, 69, 83]. In this regard, the recently published EMERALD (Exploring the Mechanism of plaque Rupture in Acute coronary syndrome using coronary CT Angiography and computationaL fluid Dynamics) trial demonstrated improved identification of culprit lesions of future acute coronary syndromes through the integration of anatomic (including lesion severity and length, and adverse plaque characteristics) and non-invasive hemodynamic parameters (ΔFFR-CT, wall shear stress, and axial plaque stress) [84]. Hence, CCTA (with selective FFR-CT) has the potential for comprehensive and integral evaluation of the anatomic burden, tissue composition, spatial distribution, and hemodynamic impact both on a patient and lesion-specific basis, within a single exam and without the need of additional contrast, radiation, or vasodilatory stress agents [65].

Fig. 6
figure 6

Seventy-two year-old female, with hypercholesterolemia and atypical chest pain. She underwent stress-SPECT, showing normal myocardial perfusion and unspecific ST-T changes. CCTA showed mild calcification of the proximal left anterior descending (LAD) artery, and a moderate non-calcified lesion at the mid LAD (a) *. The left circumflex and right coronary arteries were normal. Non-invasive computational flow dynamics analyses were performed to rule out lesion-specific ischemia, displaying a normal FFR (b) but adverse hemodynamic characteristics including high wall shear stress (≥ 154.7 dyn/cm2, c) and axial plaque stress (≥ 1606.6 dyn/cm2, d)

Full size image

Final considerations and future perspectives

It should be emphasized that the decision for the initial diagnostic strategy depends of many factors (Fig. 1), as well as regional or institutional conditions that determine the preferred initial diagnostic test for patients with suspected or new-onset stable CAD. Indeed, the selection of the initial diagnostic test remains closely linked to individual patient characteristics, and in some situations a simple exercise ECG may be sufficient to guide patient management [85].

The role of coronary artery calcium score (CACS) among symptomatic patients with suspected CAD also merits consideration as it might potentially act as a gatekeeper for functional tests, or refine risk stratification particularly among patients with mildly abnormal or equivocal functional tests. In this regard, a number of studies have shown that the absence of calcifications (CACS 0) among symptomatic patients with low to intermediate pre-test likelihood of CAD, which is found in approximately half of these patients, is related to very low rates of obstructive CAD (< 4%) and of MACE (annual rate ~  < 0.5%) [86,87,88,89] [90].

As for the future perspectives, there are emerging approaches to plan revascularization strategies without the use of ICA. The CT-SYNTAX score, which may be calculated automatically using machine learning algorithms in the near future, offers information regarding the extent, location, length, calcification degree, and tortuosity of lesions, among other features, aiding the selection of the revascularization strategy and providing additional prognostic value in patients with complex CAD [91, 92]. This was recently shown in the SYNTAX III Revolution trial, a randomized multinational trial where in patients with complex or extensive CAD, treatment decision-making based on CCTA was found to be similar to the decision derived from conventional ICA, with a 93% agreement in the decision-making and planning [93].

Overall, in view of the better diagnostic performance and prognostic value, an (at least) similar economic burden at follow-up, the promotion of a more rational use of ICA, the aforementioned relevant additional characteristics provided (Table 1) including ruling out left main disease, and the possibility of evaluating lesion-specific ischemia from the same scan; we believe that CCTA might become frontline, comprehensive imaging tool for the majority of patients with suspected CAD.