Coronary computed tomography angiography and [15O]H2O positron emission tomography perfusion imaging for the assessment of coronary artery disease

Determining the anatomic severity and extent of coronary artery disease (CAD) by means of coronary computed tomography angiography (CCTA) and its effect on perfusion using myocardial perfusion imaging (MPI) form the pillars of the non-invasive imaging assessment of CAD. This review will 1) focus on CCTA and [15O]H2O positron emission tomography MPI as stand-alone imaging modalities and their combined use for detecting CAD, 2) highlight some of the lessons learned from the PACIFIC trial (Comparison of Coronary CT Angiography, SPECT, PET, and Hybrid Imaging for Diagnosis of Ischemic Heart Disease Determined by Fractional Flow Reserve (FFR) (NCT01521468)), and 3) discuss the use of [15O]H2O PET MPI in the clinical work-up of patients with a chronic coronary total occlusion (CTO).


Introduction
Coronary atherosclerosis is marked by a chronic inflammation of the coronary arteries leading to accumulation of lipids and inflammatory cells in the arterial wall (plaques) [1]. Development of plaques may take decades but by diminishing blood flow to the subtended myocardium can eventually lead to ischaemia causing symptoms such as chest pain and dyspnoea. It is vital to assess the presence and extent of coronary artery disease (CAD) in patients with suspected CAD in order to determine the correct diagnosis and appropriate treatment strategy [2]. The non-invasive imaging modalities, coronary computed tomography angiography (CCTA) and positron emission tomography (PET) myocardial perfusion imaging (MPI) are widely utilised to that extent and assess the anatomic severity and functional significance of CAD, respectively. In this review we will highlight the assessment of CAD by means of CCTA and [ 15

Coronary computed tomography angiography
CCTA may represent a good alternative for invasive coronary angiography (ICA), especially in patients with a low or intermediate pre-test likelihood of CAD [2]. It is an anatomical imaging modality that allows for the assessment of extent and severity of coronary atherosclerosis. A large body of evidence demonstrates that CCTA is able to exclude significant CAD with a near to absolute certainty due to its excellent sensitivity and negative predictive value [3]. Nevertheless, it is hampered by a high rate of false-positive findings and as such its specificity and positive predictive value is only moderate [3]. This is explained by the tendency of CCTA to overestimate the severity of disease due to artifacts caused by, for example, calcifications, known as 'blooming artifacts' (Fig. 1; [4]). Prospective studies have shown that patients who underwent CCTA as a first-line test were more likely to be referred for ICA and even be revascularised as a consequence compared with those who underwent a functional test or standard care [5,6]. On the other hand, the rate of non-obstructive CAD on ICA following CCTA is also higher as compared with a diagnostic strategy that utilises a functional test [5]. This highlights the limitations of CCTA since the burden of calcification seen on computed tomography does not directly relate to the degree of luminal obstruction, let alone its functional consequences. However, CCTA has justly acquired a prominent place in contemporary guidelines as a first-line test for the evaluation of symptomatic patients with a low to intermediate pre-test likelihood of obstructive CAD [2]. Accordingly, guidelines recommend a functional test in the presence of obstructive CAD on CCTA, known as the hybrid approach, as viable diagnostic strategy in order to minimise the rate of false-positive CCTA findings and as such lead to a more judicious referral for ICA [2]. Recently, a CCTA-based technique has been developed that assesses lesion-specific ischaemia, namely FFRct: fractional flow reserve derived from CCTA [7]. FFRct (HeartFlow Inc. Redwood City, USA) uses computational fluid dynamics and a 3D model of the coronary vasculature derived from standard CCTA datasets to calculate FFR [7]. Prospective trials have consistently demonstrated FFRct to accurately detect lesion-specific ischaemia [3,8,9]. The FFRct PACIFIC sub-study was the first study to compare the accuracy of CCTA, FFRct, single-photon emission computed tomography (SPECT), and positron emission tomography (PET) myocardial perfusion imaging (MPI) in a head-to-head manner and demonstrated FFRct to exhibit the highest accuracy for lesion-specific ischaemia as refereed by invasive FFR. Noteworthy, FFRct could not be obtained in 17% of the vessels [10]. Furthermore, incorporating FFRct in CCTA assessment possibly reduces healthcare costs without a penalty to clinical outcome as compared with standard care [11]. Fig. 1 demonstrates how FFRct can lead to a more prudent referral pattern for ICA. Another approach to predict the functional significance of CAD solely based on CCTA is related to parameters of severity and burden of atherosclerosis, such as total plaque volume, non-calcified plaque volume and adverse plaque characteristics that have all been linked to the presence of ischaemia [12][13][14]. These analyses are, however, time-consuming and therefore not yet applicable in daily practice. Implementation of new technologies such as machine learning may overcome this barrier [15]. Machine learning has the potential to run these analyses swiftly and with high accuracy and consistency. Future studies, such as the CONFIRM-II trial, will investigate whether machinelearning analysis provides improved diagnostic ac-curacy and prognostication compared with human readers.

[ 15 O]H 2 O PET perfusion imaging
Nuclear-based functional testing is at the heart of diagnosing CAD. For decades, the field of MPI has been dominated by SPECT. From the outset, SPECT has been the MPI workhorse. However, over the last years a switch from SPECT to PET MPI has been taking place, given the increasing availability of PET scanners and 82 Sr/ 82 Rb generators, lower radiation exposure, improved resolution, ability of PET to quantify perfusion in absolute terms (in ml/min/g) and lastly superior pharmacokinetics of the tracers used as compared with SPECT tracers [16]. There is a wide variety of PET perfusion tracers available such as 82 Rb, 13 H2O PET MPI has been employed to assess the presence of ischaemia in patients with a possible indication for percutaneous revascularisation of their CTO. Case examples of CCTA with incorporation of FFRct and the ICA result. Case 1 presents the CCTA of a patient with nonobstructive disease in the LAD, as expected owing to the high sensitivity and negative predictive value of CCTA, subsequent ICA with FFR measurements confirmed non-significant CAD. The diagnostic performance of CCTA is, however, hampered by a relatively high rate of falsepositive findings, an example is seen in Case 2. Incorporation of FFRct analysis in the assessment of CCTA can lead to a shift from false-positive results to true negatives (Case 2) and can confirm the significance of CAD as seen in Case 3. CAD coronary artery disease, CCTA coronary computed tomography angiography, DS diameter stenosis, FFR fractional flow reserve, FFRct CCTA derived FFR, ICA invasive coronary angiography, LAD left anterior descending artery which might go unnoticed on relative uptake images of PET and SPECT as these are dependent on normally perfused myocardium to serve as reference area ( Fig. 3; [18,19]). The optimal quantitative MBF cut-off to detect significant CAD has been studied by Danad and colleagues, who showed a hyperaemic MBF of ≤2.3 ml/min/g to be the optimal threshold to detect FFR-defined disease [20]. In addition to hyperaemic MBF, coronary flow reserve (CFR) can be calculated by dividing hyperaemic MBF by baseline MBF. CFR has a lower accuracy for detecting significant CAD as compared with hyperaemic MBF [20]. Dependency of CFR on both baseline and hyperaemic MBF probably contributes to this finding, as diminished CFR is not necessarily concomitant with reduced hyperaemic MBF but can be a result of high baseline values. Although CFR has been shown to be of incremental prognostic value it seems justified that for diagnostic purposes stress-only PET protocols suffice, obviating the need for baseline perfusion imaging leading to a reduction of radiation dose and scan acquisition time [21,22]. Furthermore, as recently published, [ 15 O]H2O PET derived hyperaemic MBF predicts adverse patient outcome independently of CFR in patients with suspected CAD [23].
Hybrid cardiac PET/CCTA imaging, more than the sum of its parts?
Interestingly, [ 15 O]H2O PET can be performed on hybrid PET/CT scanners which allow assessment of coronary anatomy and functional significance of observed disease within one single scanning session [24]. In the Amsterdam University Medical Center (UMC), a clinical cohort of patients with suspected obstructive CAD underwent combined CCTA and Among these patients a hybrid approach led to a higher diagnostic certainty as compared with either modality alone, mainly by reducing the rate of false-positive CCTA findings [25]. Furthermore, hybrid PET/CCTA imaging could impact clinical decision-making, wherein MPI served as a valuable gatekeeper leading to less referral of patients for ICA when an abnormal or equivocal CCTA outcome was observed [26]. However, the true additive value of hybrid imaging remained debated due to the retrospective nature and lack of an appropriate reference standard of the aforementioned studies. As such, the PACIFIC trial was designed to determine whether stand-alone anatomic assessment by CCTA or standalone functional assessment by SPECT or PET MPI was superior in terms of diagnostic accuracy and if a hybrid approach provided incremental diagnostic value [27]. A total of 208 patients with suspected CAD without a cardiac history underwent CCTA, SPECT, and PET in a true head-to-head fashion followed by ICA in conjunction with interrogation of all major coronary arteries by invasive FFR regardless of imaging findings and stenosis severity. The diagnostic performance of CCTA, SPECT, and PET when refereed by FFR measurements is displayed in Tab. 1. In summary, quantitative [ 15 O]H2O PET exhibited a significantly higher accuracy as compared with CCTA and SPECT. In addition, CCTA proved to be an ideal tool for the exclusion of significant CAD as reflected by its high sensitivity and negative predictive value. An important finding was the unexpectedly low sensitivity of SPECT as a result of a high number of falsenegative findings. The putative accuracy of SPECT derived from earlier studies is controversial due to the use of an anatomical reference standard, namely obstructive disease on ICA [28]. Furthermore, the un-favourable pharmacokinetics of SPECT tracers led to a high rate of false-negative findings when referenced by FFR (Fig. 2; [16]). The addition of functional testing to CCTA increased specificity by reducing the number of false-positive CCTA findings but came with a penalty to sensitivity as a result of false-negative MPI results [27]. As such, there is paradoxically no incremental diagnostic value of combining MPI with CCTA. The findings of the PACIFIC trial have been confirmed by the prospective Danish Study of Non-Invasive Diagnostic Testing in Coronary Artery Disease (Dan-NICAD) showing a low sensitivity of SPECT (36%) and cardiac magnetic resonance imaging (41%) MPI in patients with obstructive CAD on CCTA [29]. Interestingly, both studies have in common that FFR was used as reference standard instead of obstructive disease on ICA. A multitude of sub-studies utilised the [ 15 O]H2O PET and CCTA data obtained in the PACIFIC trial of which we will highlight a few.

CCTA derived plaque burden and morphology, more than meets the eye
As mentioned previously, CCTA allows for the assessment of obstructive CAD and in addition permits the visualisation and quantification of plaque burden and morphology. Adverse plaque characteristics such as positive remodelling, low attenuation plaque, and spotty calcification are associated with the occurrence of acute coronary syndromes [30,31]. Plaque burden and morphology harbours, beside prognostic value, information about the effect of atherosclerosis on downstream perfusion as assessed by [ 15 O]H2O PET and FFR ( Fig. 4; [12]). Driessen et al. showed positive remodelling and non-calcified plaque volume to have a detrimental effect on both hyperaemic MBF and FFR independent of lesion severity, whereas spotty calcification and low attenuation plaque negatively affected FFR but not [ 15 O]H2O PET derived hyperaemic MBF [12]. In contrast to FFR, the invasively obtained resting pressure index instantaneous wavefree ratio (iFR) showed not to be associated with highrisk plaque features [32].

Reversing the roles: invasively measured indices referenced by [ 15 O]H 2 O PET determined MBF
As mentioned previously, [ 15 O]H2O PET derived MBF is considered the reference standard for non-invasive assessment of quantitative myocardial perfusion. However, absolute coronary flow can also be invasively measured using continuous intracoronary infusion of saline, known as continuous thermodilution. Everaars [33]. This novel technique is, however, not yet used in clinical practice in contrast to

Do we need MPI in the future or can computational models do the job?
In recent years novel techniques have been developed that assess lesion-specific significance by estimating invasive FFR solely based on 3D models of the coronary vasculature and computational fluid dynamics. Advantages of these computational models are that they obviate the need to use pressure wires and in-duce hyperaemia. One of these techniques is FFRct, which was highlighted previously, another is quantitative flow ratio (QFR) which is derived from ICA cine contrast images. FFRct and QFR demonstrate a similar and high diagnostic accuracy when referenced by FFR [3,35]. In the PACIFIC population, QFR had a higher accuracy compared with SPECT and PET MPI for the diagnosis of lesion-specific ischaemia [36]. Noteworthy, QFR computation was not feasible in 48% of the vessels due to the lack of a predefined dedicated QFR acquisition protocol in the PACIFIC trial hampering a per-patient analysis. Introduction of these computational-based techniques in the clinical arena will delineate their role in the diagnostic armamentarium.

[ 15 O]H 2 O PET MPI in patients with chronic coronary total occlusion
Clinical guidelines emphasise the importance of ischaemia and viability assessment in patients with a chronic coronary total occlusion (CTO) prior to revascularisation due to the slightly increased risk of procedural complications as compared with revascularisation of non-CTO lesions and furthermore to establish an appropriate indication [37]. In the dedicated CTO program of the Amsterdam UMC, [ 15 O]H2O PET MPI is used to assess the presence and extent of ischaemia in patients with a potential indication for percutaneous coronary intervention (PCI) of a CTO.
Prior reports from this program demonstrated marked ischaemia (>10% of the left ventricle) to be present in practically all patients with a CTO irrespective of collateral status [38,39]. In fact, the median extent of ischaemia related to the CTO lesion was 24% of the left ventricle [39]. Of note, all patients had an indication for evaluation of the CTO with the majority of patients (>80%) being symptomatic. Furthermore, the extent and depth of ischaemia was observed to be more profound in patients with a CTO as compared with patients with severe haemodynamically significant lesions as determined by FFR (mean FFR: 0.55 ± 0.19) [10,40]. These findings may be expected given the absence of antegrade flow and the complete dependence of myocardium subtended by a CTO on collateral supply. However, in clinical practice it is regularly assumed that well-developed collaterals preclude stress-induced ischaemia. This assumption may be refuted and should not be used as a reason to defer a patient from revascularisation.  [42]. In contrast, if hyperaemic MBF is higher in surrounding myocardium not subtended by obstructive CAD (indicating normal functioning microvasculature), the gain in hyperaemic MBF in the CTO area that can be expected after PCI is higher as well [42].

Conclusion
Coronary CTA and MPI are established non-invasive imaging modalities to diagnose CAD with techniquedependent advantages such as the high negative predictive value of CCTA and the ability of MPI to assess the functional severity of CAD. Computational fluidbased techniques such as FFRct and QFR diversify the diagnostic opportunities available to the physician. Although novel insights and developments in the field of (non)invasive imaging are promising and might lead to a more judicious assessment of CAD, the incremental value of imaging-based treatment strategies to improve patient outcome should be carefully reviewed. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.