Coronary artery disease (CAD) is the most prominent cause of death worldwide and contributes extensively to the economic burden of health care costs. To date, numerous cardiovascular imaging modalities designed to assess CAD have been introduced for clinical performance. Among those, much focus has been placed on computed tomography (CT)-based methods that include: coronary artery calcium (CAC) score, coronary CT angiography (CCTA), myocardial CT perfusion (CTP), fractional flow reserve derived from CT (FFR_CT). The current review provides an overview regarding the benefits, limitations, and future directions of CT and discusses the appropriate use of CT in detecting CAD.

CAC is a well-known proxy of coronary artery atherosclerosis, and is closely associated with total atherosclerotic burden. CAC scoring is considered a robust measure for early screening of CAD, particularly in asymptomatic individuals. CAC is generally defined as a hyperattenuated lesion above a threshold of 130 Hounsfield units (HU) with an area ≥ 3 adjacent pixels on non-contrast enhanced cardiac CT.1) Though, there are several values of CAC cut-points that are often used for the purpose of stratification in categorizing the risk of CAD: 0 (very low), 1–99 (mild), 100–400 (moderate), > 400 (severe).

CAC has been demonstrated to be one of the most robust cardiovascular risk prediction markers. Indeed, numerous prospective studies have found that a moderate-to-high CAC likely reflects an incremental predictor of future obstructive CAD over and above conventional risk factors.2) Moreover, prior studies indicate that the addition of CAC to conventional risk factors also improves classification of cardiovascular risk.3) Conversely, a zero CAC score is closely associated with a low prevalence of adverse cardiac events and can be considered protective of CAD. Sarwar et al.4) evaluated the diagnostic and prognostic performance of a zero CAC score in asymptomatic and symptomatic individuals. In this cohort, only 146 of 25903 patients (0.45%) without CAC experienced a cardiovascular event in the study period, providing evidence that a zero CAC score is associated with a very low risk of future cardiovascular events. Additionally, risk prediction is not limited to coronary events; prior studies have demonstrated a moderate predictive benefit of CAC for incident stroke.5)

In the 2013 American College of Cardiology/American Heart Association guidelines, CAC was given a class IIb recommendation for assessment of cardiovascular risk in asymptomatic adults.6) CAC ≥ 300 may be taken into consideration, similar to family history, among patients with a 5–7.5% 10-year estimated risk of atherosclerotic cardiovascular disease using the newly developed pooled cohort equations. Although CAC score may reflect a strong tool for early detection of coronary atherosclerosis, several limitations should be considered. First, CAC is extremely limited when identifying the non-calcified plaque components of CAD.7) Thus, while the quantification of coronary calcium strongly prognosticates future adverse cardiovascular events, it may not reflect events that may result from plaques that are not yet calcified. Second, radiation exposure for CAC is low but non-negligible and must be considered carefully. The typical effective doses administered for CAC scanning are 1 mSv by electron beam CT and 3 mSv by multidetector row CT.8) Given these doses, continuous efforts have been made to reduce the amount of radiation dosages without jeopardizing the assessment of CAC score by techniques such as lower tube voltage and current.

Further, a zero CAC score has been contended by some to allow stratification of individuals whose risk is sufficiently low that pharmacological therapies could be avoided. For example, it has been estimated that current guidelines would recommend statins for the majority of the United States population over age 60.9) To this end, Nasir et al.10) evaluated within the Multi- Ethnic Study of Atherosclerosis (MESA) population the degree of risk reclassification for statin eligibility due to CAC. Of the patients who were recommended for high-intensity statins, 41% had CAC = 0 and had only 5.2 events over 1000 personyears, far below the 7.5% yearly event rate attributed to them using the pooled cohort equations. Similarly, Miedema et al.11) assessed the potential of CAC score for guiding aspirin use for the primary prevention of CAD in MESA. Participants with CAC score ≥ 100 had favorable risk/benefit estimates from aspirin use, while individuals with a zero CAC score would likely experience more harm than benefit. Finally, Bittencourt et al.12) evaluated the potential benefit of a polypill according to CAC score among the MESA participants. Individuals with a zero CAC score had a very low event rate with a high projected number needed to treat. These results suggest that avoidance of therapy in subjects with a zero CAC score could allow for significant reductions in the population considered for treatment and the number needed to treat for preventing CAD events, and could potentially improve safety and cost.

While current guidelines largely support CAC score in asymptomatic individuals, some investigators have maintained that CAC has a role as a potential gatekeeper to downstream procedures in symptomatic populations. Several studies provide insight regarding the incremental diagnostic and prognostic value of CAC score in symptomatic patients.13) In a study examining the prevalence and severity of CAD in 10037 patients with chest pain, a zero CAC score showed a negative predictive value (NPV) of 96% and 99% according to 50% and 70% coronary stenosis, respectively.13)

Patients undergoing lung cancer screening by thoracic nongated CT often demonstrate calcified coronary artery atherosclerosis and recently, the combined detection of CAD and lung cancer has been encouraged.14) In particular, the use of a one-time CT scan for identifying both CAC and lung cancer may display numerous benefits, including lower overall radiation exposure. Importantly, while there is generally high correlation between CAC and low dose lung cancer CT screening tests, there are practical differences in image acquisition and radiation dose between CAC scanning and low dose CT for lung cancer screening that relate to tube voltage and current, and electrocardiography (ECG) gating. To date, no study has evaluated the prognostic utility of CAC and low dose lung cancer screening CT when performed as a single exam, although such clinical trials are currently being designed.

CCTA is a non-invasive tool that can directly visualize the coronary anatomy with a reportedly high diagnostic accuracy, rendering it useful for anatomical assessment of CAD. Previous large prospective studies have reported that CCTA accurately identifies the presence and severity of obstructive CAD. For instance, the Assessment by Coronary Computed Tomographic Angiography of Individuals Undergoing Invasive Coronary Angiography (ACCURACY) trial was the first prospective multicenter trial to evaluate the diagnostic accuracy of CCTA in symptomatic patients without known CAD.15) Among 230 patients who underwent both CCTA and invasive coronary angiography (ICA), CCTA demonstrated high accuracy for detection of obstructive CAD with very high NPV (99%) for diagnosing patients with ≥ 70% stenosis. The Coronary Artery Evaluation Using 64-Row Multidetector Computed Tomography Angiography (CORE 64) study is a multicenter international trial using 64-slice CT detectors for identifying CAD in 291 patients with calcium scores ≤ 600.16) The sensitivity, specificity, positive predictive value (PPV) and NPV of ≥ 50% stenosis in CCTA were 85%, 90%, 91%, and 83%, respectively. In comparison to the ACCURACY study, where those with severely obstructive (≥ 70%) coronary stenosis represented only 13.9 percent of the study sample, the CORE 64 study observed a higher prevalence of obstructive CAD (defined at a lower threshold at ≥ 50% stenosis) at 56%, resulted in an expectedly higher PPV at the cost of NPV when compared to the ACCURACY trial results. A third multicenter trial studying the diagnostic value of 64 CCTA by Meijboom et al.17) documented the sensitivity, specificity, PPV and NPV for detecting patients with significant CAD (i.e., ≥ 50% lumen diameter reduction) was 99%, 64%, 86%, and 97%, respectively. This study reported a high sensitivity and NPV, similar to the ACCURACY trial. Importantly, both the ACCURACY trial and the study by Meijboom et al.17) enrolled only patients without known CAD, and suggests a very robust ability of CCTA to exclude obstructive CAD in this symptomatic population.15)16)

The prognostic value of CCTA for clinical outcomes has been well established by the Coronary CT Angiography Evaluation for Clinical Outcomes: An International Multicenter (CONFIRM) registry.18) Among 23854 patients without known CAD, the incidence of mortality was significantly associated with obstructive CAD according to a per-patient, per-vessel, and persegment analysis. Further still, the absence of CAD by CCTA was associated with a favorable prognosis (e.g., annualized death rate = 0.28%). To this end, CCTA may prove useful for predicting future mortality outcomes, and may also serve an important role for effectively ruling out future cardiovascular events.

Direct head-to-head comparisons of CCTA to the more traditional used stress tests have been recently evaluated. Recently, The Prospective Multicenter Imaging Study for Evaluation of Chest Pain (PROMISE) trial aimed to compare clinical outcomes in symptomatic patients that required further evaluation to initial strategy of anatomical testing by use of CCTA or functional testing.19) The adjusted hazard ratio for a CCTA strategy, as compared with a usual-care strategy of functional testing was 1.04 (95% confidence interval, 0.83 to 1.29), with adjustment for age, sex and several cardiovascular risk factors. In this study, there was a trend towards reduced rates of adverse clinical events at 12 months for individuals undergoing CCTA. To this end, the PROMISE trial indicates an initial strategy of CCTA performance is comparable to functional testing. Based upon this and similar data, current guidelines recommend the use of CCTA to symptomatic individuals, with a general focus on those considered low-to-intermediate risk.20) Societal guidance documents also endorse the use of CCTA in patients with discordant ECG exercise and imaging results, such as those who continue to experience symptoms even with a normal ECG exercise test.

While some studies have suggested a clinical benefit of CCTA when screening asymptomatic populations, others have reported conflicting results. In general, the prognostic implications of screening for occult CAD in asymptomatic individuals with CCTA, rather than CAC, have not been addressed adequately due to their very low rate of hard events.21)22) Cho et al.23) explored risk stratification by CCTA in 7590 patients without chest pain syndrome. Although both CAC score and CCTA significantly improved the performance of standard risk factor prediction models (likelihood ratio p < 0.05 for all), the additional riskpredictive advantage by CCTA did not appear to be clinically meaningful in the general asymptomatic population. In select high risk patients, CCTA may offer benefit over CAC; among 400 asymptomatic diabetic individuals without known CAD, segment stenosis by CCTA was associated with increased major adverse cardiovascular events after adjusting for numerous CAD risk factors as well as CAC score.22) Yet, the appropriate treatment of these individuals is still unknown. In the Screening for Asymptomatic Obstructive Coronary Artery Disease Among High-Risk Diabetic Patients Using CT Angiography, Following Core 64 (FACTOR-64), a randomized controlled trial consisting of asymptomatic diabetic patients with relatively high cardiovascular risk,24) 900 individuals were randomized to receive either screening with 64-slice CCTA or guideline-based optimal care. No differences in outcomes were observed between the CCTA and guideline-based strategy, although rates of adverse events were very low and thus may have lacked power to exclude a benefit to screening. Additional studies to improve our understanding of the utility of CCTA for risk stratification appear warranted.20)

CCTA enables assessment of atherosclerotic plaque volume and characteristics in a similar fashion to invasive intravascular ultrasound (IVUS). In a meta-analysis, CCTA proved to be highly accurate for determining plaque volume and area as compared with IVUS.25) Recently, atherosclerotic plaque characteristics assessed by CCTA have been shown to be associated with ischemic lesions identified by invasive fractional flow reserve (FFR), whereby positive remodeling and low attenuation plaques were independently associated with myocardial ischemia (odd ratio: 5.30, p < 0.001, and odd ratio: 2.1, p = 0.038, respectively).26) The latter findings suggest CCTA may be a useful adjunct for patient stratification and for improving specificity and PPV. Further, these atherosclerotic features identified by CCTA also provide predictive ability for future acute coronary syndromes and other adverse cardiovascular outcomes.27)

CCTA allows detection of anatomically obstructive coronary stenoses but, similar to anatomic evaluation by coronary angiography, cannot determine alone whether a stenosis is flow-limiting. In light of this, several prior studies have aimed to combine CCTA with established functional modalities, such as single photon emission CT (SPECT) or positron emission tomography myocardial perfusion imaging (MPI).28) Although these approaches are feasible, they are generally burdensome to patients who are required to undergo at least two tests, while paying a penalty to cost and radiation exposure. In this regard, combined CCTA and CTP imaging may benefit the patient by enabling both morphologic and functional assessment of CAD into a single technique with high accuracy.

The majority of studies for CTP have employed a static single-energy approach, with comparable diagnostic performance compared to SPECT or ICA with SPECT.29)30) Rochitte et al.30) assessed the diagnostic performance of integrated CT angiography (CTA) and CTP for identification of individuals with flow-limiting CAD, as defined by ICA with a concomitant perfusion deficit on SPECT in the first multicenter study. The patient- based diagnostic accuracy of the combined CTA-CTP was 87% in all patients, and increased further to 90% and 93% when patients without prior myocardial infarction and known CAD were excluded, respectively. Subsequently, this study concluded that the integration of CTA and CTP could correctly identify individuals with obstructive CAD defined as ≥ 50% stenosis by ICA causing a perfusion defect by SPECT.

Although recent advances in CT technology have allowed the assessment of CTP, there still exist significant limitations that constrain its widespread use in daily clinical practice. Artifacts from CTP are significant, and include beam hardening, misregistration, image noise and motion artifacts. Future improvements in CT technology might help to alleviate these artifactrelated problems. Others have appropriately contended that the additional cost and safety of additional contrast and radiation exposure should be continuously considered. Further, static CTP imaging is limited by the absence of quantitative techniques, but also by the acquisition of images for a single snap shot during the early arterial phase.31)

Dynamic myocardial CTP can overcome some flaws of static CTP and has emerged as a novel non-invasive imaging technique to evaluate reduced absolute myocardial blood flow. This technique has been demonstrated to have high diagnostic accuracy for the detection and exclusion of ischemic or infarcted myocardial segments.32) Although the majority of studies have been limited by study sample size, the diagnostic performance of dynamic myocardial CTP has been to date investigated in comparison with SPECT and ICA, invasive FFR, or cardiac magnetic resonance imaging (MRI).31)32)33) Ho et al.33) evaluated the ability of stress dynamic myocardial CTP for detecting abnormal blood flow reserve and infarction in comparison to stress nuclear MPI. The results showed that sensitivity, specificity, PPV, and NPV were 0.83, 0.78, 0.79, and 0.82, respectively, when compared with SPECT MPI. While compared with ICA, the results of dynamic CTP were 0.95, 0.65, 0.78, and 0.79, respectively. Recently, Bamberg et al.32) evaluated the use of dynamic CTP for the evaluation of myocardial ischemia in comparison to cardiac MRI. In that study, the diagnostic value of CT reported a sensitivity of 77.8%, specificity of 75.41%, NPV of 91.3%, and a modest PPV of 50.6%. Further, a higher diagnostic accuracy was observed for transmural perfusion defects and infarcted segments with a sensitivity of 87.8% and 85.3%, respectively. However, numerous potential limitations exist for dynamic CTP. Perhaps most importantly, the radiation dose of performing dynamic CTP is substantially higher compared to static CTP. Prior studies have reported average doses of ~20 mSv, equivalent to nearly 4–10 times CCTA doses. Also, the relatively long breath hold times and scan duration of 30 seconds that is necessary to complete the exam renders it difficult and uncomfortable to many patients. Furthermore, table movements may result in spatial misalignment. Thus, taking these points into consideration, additional studies along with further advances in CT technology are clearly needed to resolve these limitations.

Recently, application of computational fluid dynamics to typically acquired CCTA has allowed for a novel disruptive technology to calculate FFR from static CCTA data. FFR_CT is utilized to assess the coronary anatomy without the need for additional radiation exposure, additional iodinated contrast, or administration of additional medications (Fig. 1). To date, several prospective multicenter trials have evaluated the diagnostic performance of FFR_CT compared to an invasive FFR reference standard. The Diagnosis of ISChemia-Causing Stenoses Obtained Via NoninvasivE FRactional FLOW Reserve (DISCOVER- FLOW) study initially explored the diagnostic performance of FFR_CT in 103 patients across 159 vessels.34) Anatomical obstruction was defined as a CCTA with stenosis ≥ 50%, and ischemia was defined using FFR_CT with FFR ≤ 0.80. On a per-vessel analysis, the accuracy, sensitivity, specificity, PPV, and NPV were 84.3%, 87.9%, 82.2%, 73.9%, and 92.2%, respectively. Using receiver operating characteristic curve analysis, the area under the curve (AUC) was 0.80 for FFR_CT, which was significantly higher than the AUC based on CCTA (e.g., 0.75, p < 0.001). A follow-up study–the Determination of Fractional Flow Reserve by Anatomic Computed Tomographic Angiography (DeFACTO) study–consisted of a larger population including 252 patients representing 407 vessels.35) Of these patients, 137 (54.4%) had an abnormal FFR determined by invasive FFR. On a per-patient basis, diagnostic accuracy, sensitivity, specificity, PPV, and NPV of FFR_CT plus CCTA were 73%, 90%, 54%, 67%, and 84%, respectively. Interestingly, diagnostic accuracy of FFR_CT in patients with lesions of intermediate stenosis (i.e., 30% to 70% stenosis) was significantly higher sensitivity than CCTA only (82% versus 37%), without a change in specificity. Recently, the Analysis of Coronary Blood Flow Using CT Angiography: Next Steps (NXT) study also demonstrated high diagnostic performance of FFR_CT.36) Diagnostic accuracy, sensitivity, specificity, PPV, and NPV of FFR_CT were 81%, 86%, 89%, 65%, and 93%, respectively. Compared with CCTA, FFR_CT showed a significant reduction in the rate of false positives, particularly among those with lesions of intermediate stenosis (PPV: 63% versus 37%, p < 0.001). These prospective multicenter studies underscore the robustness of FFR_CT to accurately pinpoint specific coronary lesions that cause ischemia, and extend the diagnostic paradigm beyond anatomically obstructive stenoses to lesion-specific ischemia.

Beyond improved diagnostic accuracy, FFR_CT may also prove useful for guiding clinical decision-making. Using FFR_CT to select patients for ICA and percutaneous coronary intervention (PCI) may lower health-related costs by 30% and may reduce the overall number of adverse cardiovascular events by 12% when compared to ICA or visual guidance for PCI.37) Recently, the Prospective LongitudinAl Trial of FFR_CT: Outcome and resource iMpacts study (PLATFORM) trial evaluated the utility of FFR_CT to improve patient selection for ICA among 584 patients with new onset chest pain. Patients who were randomized to FFR_CT had cancellation of ICA in 61% of all patients, with normal ICA in only 12%, as compared to 73% of those randomized to usual care.38) One potentially intriguing application of FFR_CT is its use to perform "virtual stenting" by computational modeling of coronary flow after virtual modification. Indeed, Kim et al.39) reported that there are a positive correlation between FFR_CT and invasive FFR before and after stenting, with an overall accuracy of 96%. According to the latter study, treatment planning using FFR_CT and 'virtual stenting' method can help to predict the therapeutic benefit of coronary revascularization prior to invasive coronary angiogram.

In the recent past, cardiac CT has emerged as an extremely reliable tool for detection of CAD, with disruptive technologies such as FFR_CT allowing for combined anatomic and physiologic assessment. In this review, we have provided a description of the benefits and disadvantages of several CCTA applications, having considered anatomic and physiologic assessment, as well as newer CT technologies (Table 1). Further, we have reviewed novel methods for improvement of image quality and radiation dose (Table 2). Continuing and future studies will help to define the clinical role of cardiac CT in the prevention and care of patients with cardiovascular disease.