Fig. 31.1
Concepts for non-invasive assessment of myocardial ischemia
31.2 Perfusion CT
The strength of perfusion imaging is visualizing the myocardial blood flow on which myocardial metabolism depends. Perfusion MR uses similar concept used in nuclear perfusion imaging or perfusion cardiac magnetic resonance imaging (CMR). From the myocardial and left ventricular cavity arterial input function or time-attenuation curves, the extent of regional myocardial perfusion is calculated or compared with the other regional myocardial perfusion. Perfusion is imaged in a complete cardiac cycle (dynamic perfusion imaging) or as a snapshot (static perfusion imaging). Scanners equipped with dual energy source can be used for perfusion imaging and mostly used for static perfusion imaging (Fig. 31.2). The performance of perfusion CT for predicting functionally significant stenosis is considered to be similar to nuclear perfusion imaging, stress CMR, or stress echocardiography, and is being validated against FFR [3–5]. Standard coronary angiography can be done along with perfusion imaging, which enables simultaneous anatomic evaluation of coronary arteries with functional evaluation of heart. Therefore, perfusion CT combined with coronary CT angiography can be a one-stop shop modality that assesses both anatomical and functional stenosis within a single session [6].
Fig. 31.2
Principle of myocardial perfusion CT . The difference between myocardial blood flow correlates with the myocardial up-slope normalized by arterial input function (AIF) up-slope, area under curve (AUC) of myocardial signal intensity up to AIF peak, or myocardial peak signal intensity. The difference between normal tissue and ischemic tissue is imaged as perfusion defect (line with red arrows)
31.2.1 Technical Aspect of Perfusion CT Imaging
Hyperemia is induced by pharmacological stress agents . Intravenous adenosine is widely used in a continuous dose of 140 μg/kg/min for 2 or 3 min. Regadenoson has longer plasma half time than adenosine and is administered in a single agent. Also it is a selective adenosine 2A receptor agonist and can be used in patients with asthma or airway disease. Dobutamine, a myocardial beta-1 agonist, or dipyridamole, adenosine receptor blocker, is not commonly used (Table 31.1).
Table 31.1
Stress agents for perfusion imaging
Advantage | Disadvantage | |
---|---|---|
Exercise | Most physiological | Motion artifact → not practical for CT or MR |
Least expensive | Effort-dependent | |
Adenosine | Current de facto standard | Potential bronchospasm (not good for chronic obstructive lung disease, asthma, caffeine user) |
Tachycardia, AV block | ||
Dipyridamole | Inexpensive | Prolonged effect |
Tachycardia, AV block | ||
May require aminophylline | ||
Regadenoson, binodenoson | Bolus injection | Expensive |
Fewer side effects in COPD/asthma | Prolonged effect | |
Tachycardia | ||
Dobutamine | Physiological | Lower sensitivity/specificity |
Tachycardia | ||
Can provoke ischemia |
Static or snap-shot perfusion CT assesses myocardial contrast distribution in a single time and doable in most CT scanners with lesser radiation exposure to dynamic perfusion CT. With sophisticated mathematical modeling, dynamic perfusion CT enables direct quantification of myocardial blood flow (MBF), myocardial blood volume, and myocardial flow reserve (Table 31.2). Regarding the diagnostic performance, static perfusion CT showed sensitivity = 0.85 (95% confidence interval = 0.70–0.93), specificity = 0.81 (0.59–0.93), area under curve = 0.90 (0.87–0.92) [7]. A recent dynamic perfusion CT showed comparable performance compared to CMR (Table 31.3) [8–20]. Also perfusion CT is better suited for quantification of myocardial blood flow than perfusion MR. Based on the nuclear perfusion studies, the nominal value of resting myocardial blood flow is known to be 0.9 ml/μg/min. The cut-off value of hemodynamically significant stenosis in perfusion CT was reportedly 0.75–0.78 ml/μg/min [16].
Table 31.2
Techniques for myocardial perfusion CT
Strength | Weak points | |
---|---|---|
Static perfusion CT | • Doable in most CT scanner | • Highly affected by image acquisition timing |
• Doable with standard CCTA | ||
• Minimal radiation (1–3 mSv) | • May need multiphase reconstruction to reduce artifacts (beam hardening, motion, partial scan artifacts) | |
• TPR (trans-myocardial perfusion ratio) | ||
Dynamic perfusion CT | • Less susceptible to artifact | • Need specific scanner (256 or 320-slice, or 128-slice with shuttle mode) |
• Quantitative blood flow analysis (myocardial blood flow or flow reserve) | ||
• High radiation (>10 mSv) | ||
• Axial coverage might be insufficient | ||
• Need separated CCTA scanning | ||
• Limited clinical data | ||
Dual energy perfusion CT | • Iodine distribution map → better tissue discrimination | • Affected by image acquisition timing |
• Quantitation of myocardial blood pool | • Needs standardization of iodine map interpretation | |
• Mostly static perfusion CT |
Table 31.3
Diagnostic performance of perfusion CT
Study and publication | Techniques (Scanner) | N | Reference | Stenosis (%) | Sensitivity | Specificity | PPV | NPV | Basis of analysis |
---|---|---|---|---|---|---|---|---|---|
Rocha-Filho et al. [8] | Static (64-DSCT) | 35 | ICA | 50 | 91 | 91 | 86 | 93 | Vessel |
70 | 91 | 78 | 53 | 97 | |||||
Feuchtner et al. [9] | Static (128-DSCT) | 25 | ICA | 70 | 100 | 74 | 97 | 100 | Vessel |
Nasis et al. [10] | Static (320-MDCT) | 20 | SPECT/ICA | 50 | 94 | 98 | 94 | 98 | Vessel/territory |
70 | 79 | 91 | 73 | 93 | |||||
Carrascosa et al. [11] | Static, rapid kV switching | 25 | SPECT | – | 73 | 95 | – | – | Vessel |
Magalhaes et al. [12] | Static (320-MDCT) | 381 | ICA + SPECT,MR | 50 | 98 | 96 | 98 | 96 | Territory |
58 | 86 | 87 | 55 | Vessel | |||||
Huber et al. [13] | Static (256-MDCT) | 32 | ICA | 50 | 76 | 100 | 90 | 100 | Vessel |
Rossi et al. [14] | Dynamic (128-DSCT) | 80 | ICA | 50 | 78 | 75 | 91 | 51 | Territory |
88 | 90 | 98 | 77 | Vessel | |||||
Bamberg et al. [15] | Dynamic (128-DSCT) | 33 | FFR | 50 | 93 | 87 | 75 | 97 | Vessel |
Bamberg et al. [16] | Dynamic (128-DSCT) | 31 | Cardiac MR | – | 100 | 75 | 100 | 92 | Vessel |
Ko et al. [17] | Dynamic (64-DSCT) | 45 | ICA | 50 | 93 | 86 | 88 | 91 | Vessel |
Wang et al. [18] | Dynamic (128-DSCT) | 30 | ICA/SPECT | 50 | 90 | 81 | 58 | 97 | Vessel/territory |
Kim et al. [19] | Dynamic (128-DSCT) | 50 | Cardiac MR | – | 77 | 94 | 53 | 98 | Vessel/territory |
Wichmann et al. [20] | Dynamic (128-DSCT) | 71 | Visual assessment | 50 | 100 | 88 | 100 | 43 | Territory |
31.3 Computational Simulation of Fractional Flow Reserve
Increase of myocardial blood flow by 2 to 3-fold is required to match the increased need of cardiac output in most activities. Coronary microvessel accounts for most resistance or pressure drop in coronary circulation. The increase of myocardial blood flow is mainly controlled by decrease in microvascular resistance. Therefore functionally significant epicardial coronary artery stenosis can be defined by failure to increase blood flow during hyperemia which induces maximal dilatation of resistance vessel. Fractional flow reserve (FFR) is defined by the ratio of hyperemic coronary flow through stenotic vessel to the hypothetical normal vessel. Because flow is proportional to pressure in fixed stenosis, FFR can be measured by average pressure gradient. Pressure drop of more than 20% or FFR ≤ 0.80 is widely advocated as a gold standard of vessel-specific physiologically significant stenosis which may evoke myocardial ischemia.
FFR is measured during invasive cardiac catheterization and requires insertion of a pressure wire inserted through the stenosis. There may be and instability of measurement and signal shift. Placement of a pressure wire near the stenosis or pressure recovery zone may lead to overestimation of FFR. A non-invasive simulation of FFR would be very valuable to avoid these procedural shortcomings and the expense of pressure wire and invasive cardiac catheterization.
31.3.1 Computation of Simulated FFR
Like the other fluid systems, blood flow in the cardiovascular system is ruled by the physical laws of mass conservation, momentum conservation, and energy conservation. Therefore it can be calculated by mathematical models. For patient-specific coronary circulation, 3-dimensional numerical models based on computational flow dynamics which can compute complex flow patterns are preferred to zero dimensional models or lumped parameter model which is employed in large systemic vessels. Computational FFR is derived based on the regional physical geometry, the boundary condition which is the behavior and properties at the boundaries of the region, and the physical laws of fluid in the region.
FFR can be described as a pressure gradient across stenotic segment during maximal hyperemia. Anatomical stenosis, myocardial mass, and microvascular resistance constitute FFR value and can be calculated from patient-specific sophisticated coronary arterial anatomical model, vessel-specific myocardial mass, and microvascular resistance which determine the outlet boundary condition [21, 22]. CT images provide patient-specific anatomy model of local geometry, individual coronary artery morphology, volume, and myocardial mass. From these data, cardiac output and baseline coronary blood flow can be calculated by using allometric scaling laws [23–25]. This computational approach was derived from a general model that describes the transport of essential materials through space-filling fractal branched networks, and is based on a form-function relationship [26]. The diameter-flow rate relation is determined according to Murray’s law [27] and Poiseuille’s equation, which considers shear stress on the endothelial surface and remodeling to maintain homeostasis [28]. Morphometry laws are also adapted to obtain the physiological resistance to flow aroused by coronary artery branches [29]. Microvascular resistance at baseline and during maximal hyperemia, which is fundamental for FFR measurement, can be approximated using population-based data on the effect of adenosine on coronary flow [30] (Fig. 31.3).
Fig. 31.3
Concept of computational FFR
31.3.2 Clinical Results of Computational FFR
Landmark trials including DISCOVER-FLOW [31], DeFACTO [32], and NXT [33] showed that FFR-CT, a proprietary computational FFR, showed high diagnostic performance in discriminating ischemia in patients who had intermediate coronary artery stenosis. The NXT trial reported sensitivity and negative predictive value of FFR-CT in diagnosis of ischemia (defined as invasive FFR < 0.80) in patients with intermediate stenosis severity were 80% and 92%, respectively [33]. In a recent meta-analysis of FFR-CT based on 833 patients and 1377 vessels, FFR-CT showed a moderate diagnostic performance for identification of ischemic vessel with pooled sensitivity = 84% and specificity = 76% at a per-vessel basis [34] (Table 31.4). The PLATFORM study showed that a decision-making strategy using CCTA with FFR-CT was associated with clinical outcomes comparable to using invasive FFR and a 33% cost reduction [35]. Therefore, FFR-CT can effectively rule out intermediate lesions that cause ischemia and could also reduce the unnecessary ICA and invasive FFR.
Table 31.4
Results of non-invasive computational FFR technologies
Study | Technology | N of vessels | Sensitivity (%) | Specificity (%) | PPV (%) | NPV (%) | Accuracy (%) | Area under curve, per vessel | Correlation coefficient compared with FFR | Computation time |
---|---|---|---|---|---|---|---|---|---|---|
FFR-CT, DISCOVER-FLOW, Koo et al. [31] | Heartflow ver 1.0 | 159 | 88 | 82 | 74 | 92 | 84 | 0.90 | 0.72 | Hours, off-site |
FFR-CT, DeFACTO, Min et al. [32] | Heartflow ver 1.2 | 408 | 80 | 63 | 56 | 84 | 69 | 0.81 | 0.63 | Hours, off-site |
FFR-CT, NXT, Norgaard et al. [33] | Heartflow ver 1.4 | 484 | 84 | 87 | 62 | 95 | 86 | 0.93 | 0.82 | Hours, off-site |
FFR-CT, Kim et al. [19] | Heartflow ver 1.2 | 48 | 85 | 57 | 83 | 62 | 77 | – | 0.60 | Hours, off-site |
cFFR | Siemens cFFR ver 1.4 | 67 | 85 | 85 | 71 | 93 | 85 | 0.92 | 0.66 | <1 h, standalone |
cFFR | Siemens cFFR ver 1.4 | 189 | 88 | 65 | 65 | 88 | 75 | 0.83 | 0.59 | <1 h, standalone |
cFFR | Siemens cFFR ver 1.7 | 23 | 83 | 76 | 56 | 93 | 78 | – | 0.77 | <1 h, standalone |
Pooled analysis | – | 1330 | 84
Stay updated, free articles. Join our Telegram channelFull access? Get Clinical TreeGet Clinical Tree app for offline access |