J o u r n a l o f C a r d i o v a s c u l a r C o m p u t e d T o m o g r a p h y 8 ( 2 0 1 4 ) 2 7 2 e2 8 1

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.JournalofCardiovascularCT.com

State of the Art

Physiologic evaluation of ischemia using cardiac CT: Current status of CT myocardial perfusion and CT fractional flow reserve Andrew D. Choi MDa,b,c,*, Joanna M. Joly MDa,b, Marcus Y. Chen MDc, Wm. Guy Weigold MDa a

Cardiology Division, MedStar Heart Institute, MedStar Washington Hospital Center, 110 Irving Street, NW, Washington DC 20010, USA b Cardiology Division, Medstar Georgetown University Hospital, Washington, DC, USA c Cardiovascular and Pulmonary Branch, Advanced Cardiovascular Imaging Laboratory, National Heart Lung and Blood Institute, National Institutes of Health, Bethesda, MD, USA

article info

abstract

Article history:

Cardiac CT, specifically coronary CT angiography (CTA), is an established technology which

Received 31 October 2013

detects anatomically significant coronary artery disease with a high sensitivity and nega-

Received in revised form

tive predictive value compared with invasive coronary angiography. However, the limited

7 February 2014

ability of CTA to determine the physiologic significance of intermediate coronary stenoses

Accepted 13 June 2014

remains a shortcoming compared with other noninvasive methods such as single-photon emission CT, stress echocardiography, and stress cardiac magnetic resonance. Two methods have been investigated recently: (1) myocardial CT perfusion and (2) fractional

Keywords:

flow reserve (FFR) computed from CT (FFRCT). Improving diagnostic accuracy by combining

Cardiac computed tomography

the anatomic aspects of coronary CTA with a physiologic assessment via CT perfusion or

Computed tomography perfusion

FFRCT may reduce the need for additional testing to evaluate for ischemia, reduce

Coronary artery disease

downstream costs and risks associated with an invasive procedure, and lead to improved

Fractional flow reserve

patient outcomes. Given a rapidly expanding body of research in this field, this compara-

Ischemia

tive review summarizes the present literature while contrasting the benefits, limitations, and future directions in myocardial CT perfusion and FFRCT imaging. Published by Elsevier Inc. on behalf of Society of Cardiovascular Computed Tomography.

1.

Introduction

Cardiac CT, specifically coronary CT angiography (CTA), is a rapidly evolving technology which detects anatomically significant coronary artery disease (CAD) with a high sensitivity and negative predictive value (NPV) when compared with

invasive coronary angiography (ICA).1 However, the limited ability of CTA to determine the physiologic significance of coronary stenoses remains a limitation compared with other noninvasive methods. In current clinical practice, patients undergoing CTA with an intermediate lesion and who are found to require assessment for ischemia may undergo a

Conflicts of interest: The authors report no conflicts of interest. * Corresponding author. E-mail address: [email protected] (A.D. Choi). 1934-5925/$ e see front matter Published by Elsevier Inc. on behalf of Society of Cardiovascular Computed Tomography. http://dx.doi.org/10.1016/j.jcct.2014.06.006

J o u r n a l o f C a r d i o v a s c u l a r C o m p u t e d T o m o g r a p h y 8 ( 2 0 1 4 ) 2 7 2 e2 8 1

second test with either single-photon emission CT (SPECT), stress echocardiography, stress cardiac magnetic resonance (CMR), or stress positron emission tomography (PET). Alternatively, patients may have fractional flow reserve (FFR) measured invasively in the catheterization laboratory. Improving diagnostic accuracy by combining the anatomic aspect of coronary CTA with a physiologic assessment via CT may eliminate the need for additional testing to evaluate for ischemia, reduce the risks associated with an additional invasive procedure, and decrease downstream costs that would lead to improved patient outcomes. Two methods have been investigated recently: (1) myocardial CT perfusion (CTP) imaging and (2) FFR computed from coronary CTA (FFRCT). This review seeks to summarize the current literature and to offer a comparison of benefits, limitations, and future directions in CTP imaging and FFRCT.

1.1.

The importance of assessing ischemia

The Clinical Outcomes Utilizing Revascularization and Aggressive Drug Evaluation (COURAGE) trial, published in 2007, supplied evidence that percutaneous coronary intervention (PCI) offered no reduction in death and nonfatal myocardial infarction when compared with optical medical therapy (OMT) if decision making is based on angiographic severity of stenosis alone.2 As a result, the ability to accurately detect functionally significant stenosis using appropriate imaging modalities has taken added importance. The Fractional Flow Reserve vs Angiography for Multivessel Evaluation (FAME) study, published in 2009, extended the findings of COURAGE by using FFR, whereby functional stenosis severity was determined using an intracoronary pressure wire to quantify the trans-stenotic pressure gradient.3 In this trial, in patients with multivessel coronary disease, the decision for PCI was either determined by FFR
0.80 or decision making was on angiographic appearance alone. The use of FFR resulted in more selective stenting, with an average of 2.7 stents placed based on angiographic appearance compared with 1.9 with FFR guidance with no significant change in procedure time but reduced overall cost and contrast administration. The combined rate of death, myocardial infarction, and repeat revascularization at 1 year was significantly reduced by 30% in the FFR group. The FAME 2 trial, published in 2012, compared FFR-guided PCI plus OMT to OMT alone in the treatment of FFR-proven functionally significant stenosis, defined as FFR
0.80.4 The study was terminated early because of a significant decrease in the primary end point, a composite of all-cause mortality, nonfatal myocardial infarction, or urgent revascularization in the group receiving OMT plus FFR-guided intervention, primarily driven by decreased urgent revascularization in this group when compared with patients treated with OMT alone despite FFR
0.80. FAME and FAME 2 demonstrated that the outcome of coronary stenosis is more reliably predicted by its functional effect than by its degree of anatomic stenosis. However, routine use of invasive FFR increases case complexity and may raise the risk of catheter-based complications such as coronary perforation and dissection,5,6 hence the attraction toward an

273

efficient and accurate noninvasive means to assess the physiologic significance of anatomically significant stenoses.

2.

CTP protocol overview

CTP, like other stress imaging procedures, typically consists of 2 scans: stress and rest contrast-enhanced imaging. Stress imaging has been performed with adenosine, dipyridamole, or regadenoson as the vasodilator.7e10 On first-pass circulation, there is direct relationship between iodinated contrast within the myocardium and CT attenuation (HU); therefore, areas of decreased blood flow would have lower attenuation allowing for visual detection of a perfusion defect (Fig. 1). Similar to SPECT, reversibility or persistence of a perfusion defect from stress to rest would differentiate between ischemia and infarct. A third scan for delayed imaging may also be performed to detect myocardial viability.7 The use of dynamic perfusion imaging whereby contrast kinetics are recorded with a resultant generation of timeattenuation curves has been studied as a quantitative and semiquantitative method of perfusion quantification.11,12 Dynamic perfusion imaging may help address limitations in contrast inhomogeneity. Dual-energy CT for CTP imaging, where there is rapid switching between low and high tube potentials (kV), has been studied where iodinated contrast within the myocardium shows different spectral characteristics at the varied energy levels.13 In particular, dual-energy CT may improve perfusion quantitation by reducing beam hardening artifact. There have been multiple proposed scanning protocols, and each protocol offers advantages and disadvantages. One approach has been performing a stress scan (CTP), followed by a rest scan (CTA) 10 to 20 minutes later to improve detection of ischemia without possible contrast contamination from a rest scan and to allow for administration of nitrates during rest imaging.11,14 However, given the possibility for persistent contrast accumulation from the stress-to-rest scan in areas of decreased myocardial blood flow, this approach may decrease sensitivity for detecting ischemia and infarct. On the other hand, performing coronary CTA first would allow minimizing contrast and radiation exposure, as CTP would only be needed if intermediate stenoses or inconclusive findings were present in coronary CTA. Recent protocols use a 10- to 20-minute washout period of nitroglycerin and contrast before adenosine administration.15e17 In addition, the possible use of betablockers to lower the heart rate for coronary CTA may cause underestimation of the extent of myocardial ischemia.18 Ultimately, it may be best to select the CT perfusion protocol based on the patient’s risk profile. In low-risk patients, performing coronary CTA first and escalating to CTP in the setting of coronary stenosis may be preferred, whereas in high-risk patients, performing stress CTP first would enhance its sensitivity for the detection of ischemia. In the next section, we aim to review the salient studies on CTP to date, with an early clinical feasibility study, followed by illustrative studies showing the protocol progression and evolving gold standard using SPECT, ICA, invasive FFR, and stress CMR and finally leading to a multicenter study of CTP. Table 1 lists a summary of published studies on CT perfusion.

274

J o u r n a l o f C a r d i o v a s c u l a r C o m p u t e d T o m o g r a p h y 8 ( 2 0 1 4 ) 2 7 2 e2 8 1

Fig. 1 e A 75-year-old woman presenting with chest pain who underwent CT perfusion (CTP) imaging protocol with stress acquisition. CTP shows anterior, anteroseptal, and anterolateral perfusion defect in the (A) short axis and (B) 2-chamber views. (C) Cardiac catheterization in the same patient with significant stenosis in the left anterior descending artery.

2.1.

Initial feasibility

In 2005, Kurata et al19 published their findings of CT perfusion (LightSpeed Ultra 16; GE Healthcare, Milwaukee) using adenosine in a stress-rest protocol. All patients had SPECT and then received ICA as clinically indicated. In 36 coronary territories of 12 patients, there were 26 areas of hypoperfusion identified by CTP and 22 areas of hypoperfusion on SPECT. The agreement between CTP and SPECT was 83% (30 of 36) with significant (P < .05) concordance. However, a notable limitation was that image quality of the coronary arteries in the stress images was significantly lower than that of the rest images; only 48% of coronary segments were evaluable in the stress images vs 89% at rest (P < .05).

2.2.

CTP vs SPECT or ICA

Blankstein et al7 developed a protocol using first generation dual-source CT (Definition; Siemens Medical Systems, Germany) starting with (1) adenosine stress CTP using retrospective gating and tube-current modulation, followed by (2) rest coronary CTA using prospectively triggered image acquisition, and finally (3) delayed-enhancement (DE) imaging acquired 7 minutes after rest CTA. All patients also had SPECT and ICA (with stenosis severity assessed by blinded quantitative assessment). There were 33 patients, all aged >40 years and with high-risk features on a recent nuclear-stress test. On ICA, 25 patients (74%) had 50% stenosis in at least 1 coronary vessel. On a per-patient basis, CTP was 92% sensitive and 67% specific, whereas SPECT was also 92% sensitive and 67% specific. For stenosis 70%, on a per-patient basis, sensitivity and specificity were 94% and 41% for CTP vs 94%

and 41% for SPECT. DE was visualized in 17 of the 33 patients completing the protocol, but the quality of the images was scored lower than on the corresponding stress or rest images. Radiation exposure of the entire CT protocol (stress, rest, and DE) was 12.7 mSv vs 12.7 mSv for SPECT (rest and stress). The authors concluded that CTP demonstrated equivalence when compared with SPECT. Limitations in diagnostic accuracy of CT were attributed to limitations in temporal and spatial resolution, presence of slab and misregistration artifacts, and need for optimization of signal-to-noise and contrast-to-noise ratios. In addition, there was a selection bias given that all patients had been already known to have SPECT positivity and referred for cardiac catheterization. However, this study, along with a similar study from George et al,8 offered early proof of the feasibility of myocardial CTP imaging.

2.3.

CTP vs invasive FFR or ICA

Ko et al20 compared the diagnostic accuracy of adenosine CTP with ICA and FFR. They used a 320-detector row dynamic volume CT (Aquilion ONE; Toshiba Medical Systems, Japan) to assess myocardial perfusion. Forty-two patients with chest pain and known CAD with at least 1 50% stenosis as determined by invasive angiography (with stenosis severity assessed by blinded quantitative assessment) underwent CTP and were compared with patients with invasive FFR
0.8. When compared with FFR, CTP had a sensitivity, specificity, positive predictive value (PPV), and (NPV) of 76%, 84%, 82%, and 79%, respectively. There was moderate concordance (k ¼ 0.60) and a diagnostic accuracy of 80% between CTP and FFR methods. The combination of 50% stenosis by CTA and

34 35 30 (vs ICA) 41 33 40 50 65 37 20 101 381 Total: 906 (range: 12e381) ICA or SPECT ICA ICA or SPECT ICA or CMR FFR
0.75 FFR
0.8 SPECT FFR
0.8 FFR
0.75 ICA and SPECT FFR
0.8 ICA and SPECT 64-det. DSCT 64-det. DSCT 128-det. DSCT 64-det. DSCT 128-det. DSCT 320-det. MDCT 320-det. MDCT 128-det. DSCT 128-det. DSCT 320-det. MDCT 64-det. MDCT 320-det. MDCT 2009 2010 2010 2011 2011 2012 2012 2013 2013 2013 2013 2013 Blankstein et al7 Rocha-Filho et al33 Ho et al11 Ko et al34 Bamberg et al35 Ko et al17 George et al16 Greif et al36 Choo et al37 Nasis et al38 Bettencourt et al14 Rochitte et al21

CMR, cardiac magnetic resonance; CTA, CT angiography; CTP, CT perfusion; det, detector; DSCT, dual-source CT; FFR, fractional flow reserve; ICA, invasive coronary angiography; MDCT, multidetector CT; NPV, negative predictive value; n, number of patients; PPV, positive predictive value; pt, patient; seg, segment; SPECT, single-photon emission CT.

92 (vs ICA) 67 (vs ICA) 89 (vs ICA) 75 (vs. ICA) 96 100 100 91 95 (seg. vs ICA) 65 (seg.vs ICA) 78 (seg. vs ICA) 79 (seg. vs ICA) 97 (vs ICA) 50 (vs ICA) 95 (vs ICA) 67 (vs ICA) 95 64 84 88 95 95 95 95 72 91 81 85 97 (per vessel) 68.8 (per vessel) 95.7 (per vessel) 76.1 (per vessel) 93.1 (per vessel) 94.2 (per vessel) 90.0 (per vessel) 96.0 (per vessel) 100 92 88 100 89 (vs FFR) 83 (vs FFR) 80 (vs FFR) 90 (vs FFR) 80 74 65 86 Range: 72%e100% Range: 50%e100% Range: 65%e100% Range: 67%e100%

Not reported 85 Not reported 92 2005 2009 Kurata et al19 George et al8

16-det. MDCT SPECT 64- and 256-det. MDCT ICA and SPECT

12 27

Not reported 16.8 (64-det.) 21.6 (256-det.) 12.7 11.8 18.2 14.4 13.1 9.2 13.8 12.6 4.6 9.2 5.0 9.8 Range: 4.6e21.6

(By seg.) 90 86

(By seg.) 79 92

Per-pt NPV, % Per-pt PPV, % Per-pt specificity, % Per-pt sensitivity, % Average radiation dose (mSv) n Reference standard Scanner type Year Author

Table 1 e Summary of selected studies evaluating coronary CTA and Myocardial CTP vs ICA (>or ‡50% stenosis), SPECT, or Invasive FFR as the reference standard.

J o u r n a l o f C a r d i o v a s c u l a r C o m p u t e d T o m o g r a p h y 8 ( 2 0 1 4 ) 2 7 2 e2 8 1

275

perfusion defect on CTP was 68% sensitive and 98% specific for FFR
0.8, with 97% PPV and 77% NPV, whereas a 40 years with 2 CAD risk factors and either a positive or inconclusive exercise treadmill test deemed to have an intermediate to high probability of CAD. All patients underwent CTA and CTP (Somatom Sensation 64 Scanner; Siemens Medical Systems) as well as stress CMR (1.5 T Siemens Symphony; Siemens Medical Systems) 1 week before undergoing ICA with FFR. CTA or CTP, when compared with a reference standard of FFR
0.80, had a sensitivity, specificity, PPV, and NPV of 89%, 83%, 80%, and 90%, respectively, similar to CMR (89%, 88%, 85%, and 91%, respectively). When stress CMR was used as the reference standard, CTP was noted as having 82% accuracy. CTA analysis was limited by the fact that 33% of patients had at least 1 unevaluable segment, most commonly because of extensive calcification. Interestingly, this cohort had a higher percentage of patients (53%) with any coronary stenosis >40%, a higher proportion than typically referred for cardiac CT. Although single centered, this study used contemporary reference standards with CMR and invasive FFR. Adhering to a protocol with 100 kV-tube voltage and strict tube-current modulation led to an average effective dose of 5.0 mSv, which was substantially lower than in prior protocols. Importantly, this study enrolled patients before coronary angiography and avoided the selection bias present in the previously mentioned studies on CTP.

2.5.

CORE 320

An important recent multicenter trial evaluating CTP is named “Computed tomography angiography and perfusion to assess coronary artery stenosis causing perfusion defects by single-photon emission computed tomography” (CORE 320). Published in November 2013, CORE 320 was a large multicenter study evaluating CTP to identify flow-limiting CAD causing a perfusion defect. CT was compared with SPECT and 50% stenosis on ICA (using quantitative assessment) as the gold standard.21 A total of 381 patients from 16 centers in 8 countries underwent CTA, adenosine stress CTP, SPECT, and ICA. All CT images were acquired using a single protocol with adenosine developed for a 320  0.5 mm detector row system (Aquilion ONE). All noninvasive imaging was performed within 60 days before ICA. CT perfusion defects were interpreted using semiquantitative visual assessment and customized software by 2 independent investigators who

276

J o u r n a l o f C a r d i o v a s c u l a r C o m p u t e d T o m o g r a p h y 8 ( 2 0 1 4 ) 2 7 2 e2 8 1

were blinded to CTA findings. The average patient age was 62 years and 66% of patients were male. Nearly all patients (98%) were of intermediate to high risk for CAD by DiamondForrester score. Sensitivity, specificity, PPV, and NPV for CTA 50% stenosis and a CTP perfusion defect were 80%, 74%, 65%, and 86%, respectively, for an summed stress score of >4 as derived by CT validated against 50% stenosis on ICA with a corresponding SPECT perfusion deficit. The area under the receiver operating characteristic curve (AUC) for the primary end point (stenosis 50% in ICA with a corresponding SPECT perfusion defect) was 0.87 (95% confidence interval [CI], 0.84e0.91). The multicenter CORE 320 trial supports the findings of the single-center studies on CT perfusion. Additionally, the results of CORE 320 appear to validate the use of a rest-stress protocol. CORE 320 completed enrollment in 2011 and 5-year outcomes are expected to become available in 2016. They will help evaluate the prognostic significance of CTP. Limitations to this study include a significant number of false positives leading to a lower than expected PPV, a single-vendor protocol on a scanning platform not yet widely used, and the use of SPECT instead of invasive FFR as the gold standard for ischemia.

2.6.

CTP: current limitations and future directions

Several limitations exist with regard to CTP. The need for 2 scans increases radiation exposure with an additional contrast load. Notable artifacts include beam hardening, motion, and the need for edge-enhancement reconstruction to accurately interpret subendocardial defects. In addition, there remains uncertainty regarding the optimal timing for myocardial contrast opacification. Taken together, these challenges have limited the ability to accurately identify perfusion abnormalities as evidenced by the not insignificant number of falsepositive patients in studies such as CORE 320. As of this writing, there are at least (http://www. clinicaltrials.gov; NCT01434043, NCT01368237, NCT01334918, NCT01680081, NCT00857792, and NCT00934037) 6 studies further evaluating the validity of CTP vs various reference standards including PET, SPECT, CMR, and invasive FFR, highlighting the continuing investigation in this field. Of note, Cury et al10,22 presented preliminary results of a large multicenter study at the Society of Cardiovascular Computed Tomography 2013 Scientific Sessions that evaluated the detection of ischemia with regadenoson using CTP vs SPECT in 124 patients in a multicenter (11 US sites), multiplatform (64-slice to 320-slice) manner. They showed that CTP agreed with SPECT as the gold standard in 87% of cases and that CTP overall had 84% specificity and 90% sensitivity for detecting or excluding ischemia. A recent novel use of CTP has been in patients with coronary stents, addressing limitations caused by metallic stent material and coronary motion. Rief et al23 evaluated CTP in 91 patients with coronary stents in the diagnosis of CAD and in-stent restenosis. The authors showed that the perpatient diagnostic accuracy of CTA or CTP was 87% (95% CI, 78%e93%), which was higher than CTA alone (71%; 95% CI, 61%e80%; P < .001), primarily because of a significant (P < .001) reduction in nondiagnostic examinations.

3.

FFRCT methodology

FFRCT is a novel proprietary technology developed by a company called HeartFlow, Inc (Redwood City, California). It applies computational fluid-dynamic modeling after semiautomated segmentation of the coronary arteries and left ventricular mass. FFRCT has been validated against invasively measured FFR in 3 trials (Table 2). Figure 2 shows the correlation between CTA, ICA, and FFRCT in these studies. The principle of determining FFRCT has been described as follows24: First, an anatomic model of the coronary tree is created using CT. A mathematical model for coronary physiology to represent variables such as cardiac output, aortic pressure, and microcirculatory resistance is used along with a model for fluid dynamics. Blood is modeled as a Newtonian fluid and turbulence is modeled using Navier-Stokes equations, which are time-averaged equations of motion for fluid flow. In building the model further, image segmentation algorithms extract the luminal surface of the major vessels and branches from CT, obtaining information on topology, coronary plaque, and luminal boundary. Total coronary flow is computed based on myocardial wall volume from CT. Resistance is derived from coronary flow, whereas mean arterial pressure is taken from the patient’s mean brachial artery pressure. Coronary blood flow is simulated under conditions modeling adenosine-mediated coronary hyperemia. To obtain a solution for velocity and pressure in the coronary vasculature to represent coronary flow, solving for millions of partial differential equations simultaneously repeating this process for thousands of time intervals per cardiac cycle is required. FFRCT is determined after deriving velocity and pressure to model the mean pressure distal to a coronary stenosis and thus derive a ratio of mean pressure distal to coronary stenosis relative to mean aortic pressure. To paraphrase the description by Koo et al,25 the technology is based on 3 principles: (1) coronary supply meets myocardial demand at rest, (2) resistance of the microcirculation at rest is inversely but not linearly proportional to the size of the feeding vessel, and (3) microcirculation reacts predictably to maximal hyperemia. FFRCT
0.80 is considered diagnostic of lesionspecific ischemia. CT images are transmitted to the FFRCT core laboratory at HeartFlow, Inc, which is informed of the coronary artery stenosis by an independent scientist interpreting CTA and ICA but not involved in the analysis. FFRCT analysis requires approximately 6 hours per case.6 A variety of scanner platforms have been used for image acquisition (Lightspeed VCT, GE Healthcare; Somatom Sensation and Definition CT, Siemens; Brilliance 256 and 64, Philips, United Kingdom; Aquilion ONE and 64, Toshiba).

3.1.

DISCOVER-FLOW

The DISCOVER-FLOW (Diagnosis of Ischemia-Causing Stenoses Obtained Via Noninvasive Fractional Flow Reserve) study was the first multicenter trial that evaluated the diagnostic performance of FFRCT against ICA and invasive FFR.25 Performed at 4 sites, 159 coronary vessels in 103 patients were

91 84 92 Range: 84%e92% Det, detector; FFR, fractional flow reserve; MDCT ¼ multidetector CT; n, number of patients; NPV, negative predictive value; PPV, positive predictive value.

85 67 65 Range: 65%e85% 82 54 79 Range: 54%e82% 93 90 86 Range: 86%e93% 103 (4 sites) 252 (17 sites) 251 (10 sites) Total: 606 (range: 103e252) 2011 2012 2014 Koo et al25 Min et al6 Nørgaard et al29

64- or 256-det. MDCT 64-det. or higher det. MDCT 64-det. or higher det. MDCT

FFR
0.80 FFR
0.80 FFR
0.80

Range: 3e15 6.4 Range: 3e14 Range: 3e15

Per-pt specificity, % n Reference standard Scanner type Year Author

Table 2 e Summary of studies evaluating FFR CT.

Radiation dose, mean (mSv)

Per-pt sensitivity, %

Per-pt PPV, %

Per-pt NPV, %

J o u r n a l o f C a r d i o v a s c u l a r C o m p u t e d T o m o g r a p h y 8 ( 2 0 1 4 ) 2 7 2 e2 8 1

277

studied by CTA, FFRCT, ICA, and FFR. Ischemia was defined by either FFR
0.80 or FFRCT
0.80. There was good correlation between FFR and FFRCT on per-vessel analysis by Spearman rank correlation (0.72; P < .0001). Among the 47 vessels with 50% to 60% stenosis, 25.5% (n ¼ 12) exhibited ischemia by invasive FFR. In these intermediate stenoses, FFRCT demonstrated diagnostic accuracy of 83.0%, sensitivity of 66.7%, specificity of 88.6%, PPV of 66.7%, and NPV of 88.6% when compared with invasively measured FFR. In all 4 false-negative cases with FFRCT >0.80, invasive FFR measured between 0.75 and 0.80. When FFRCT was compared with FFR on a per-vessel basis, accuracy of FFRCT was 84.3%, sensitivity 87.9%, specificity 82.2%, PPV 73.9%, and NPV 92.2%. On a per-patient basis, FFRCT accuracy was 87.4%, sensitivity 92.6%, specificity 81.6%, PPV 84.7%, and NPV 90.9%. In contrast, when CTA was compared with FFR, CTA showed 58.5% accuracy and 46.5% PPV on a per-vessel basis, whereas on a perpatient basis, CTA showed 61.2% accuracy and 24.5% specificity, with 58.0% PPV. Overall, this trial demonstrated the concept and feasibility of FFRCT in a clinical trial setting.

3.2.

Effect on intermediate stenosis

The performance of FFRCT in patients with intermediate stenosis (40%e69%) from the DISCOVER-FLOW study was examined in a separate analysis by Min et al.26 Of the 103 patients in DISCOVER-FLOW, 66 vessels of 60 patients (58%) were identified as having an intermediate stenosis in coronary CTA. Among these 66 vessels, 31 (47%) had a FFR
0.8, whereas 34 (52%) were considered ischemic by FFRCT
0.8. As compared with invasively measured FFR, FFRCT was 86% accurate, 90% sensitive, and 83% specific, with 82% PPV and 91% NPV. There was good correlation of per-lesion FFRCT with FFR (r ¼ 0.60; P < .0001). Both FFRCT and FFR did not correlate well to percent diameter stenosis by ICA (r ¼ 0.38 and r ¼ 0.43, respectively), highlighting the limitation of percent stenosis as a measure of functional significance in these intermediate “borderline” stenoses.

3.3.

Effect of image quality

Another substudy analysis of DISCOVER-FLOW investigated the accuracy of diagnosing severe stenosis by FFRCT (
0.80) and by CTA (50% reduction in luminal diameter) with respect to image quality, heart rate, signal-to-noise ratio, and common CT artifacts including calcification, motion, and poor contrast enhancement.27 Of 103 patients, FFRCT diagnostic performance for detecting lesions with FFR
0.8 was higher across all artifact types when compared with CT’s ability to detect 50% stenosis. In the presence of calcification, accuracy of FFRCT was 85.7% vs 66.7% in CTA. In the setting of motion or misregistration artifact, accuracy of FFRCT was 90.5% vs 57.1% for CTA and with poor contrast opacification, accuracy of FFRCT was 100% vs 71% for CTA.

3.4.

Cost analysis

Hlatky et al28 performed retrospective modeling of costeffectiveness from DISCOVER-FLOW data in a study

278

J o u r n a l o f C a r d i o v a s c u l a r C o m p u t e d T o m o g r a p h y 8 ( 2 0 1 4 ) 2 7 2 e2 8 1

Fig. 2 e Comparison of (A) multiplanar reformat of coronary CT angiography (cCTA), (B) invasive coronary angiography (ICA) with fractional flow reserve (FFR) value, and (C) FFR CT showing a hemodynamically significant stenosis of the left anterior descending artery. Reproduced with permission from Dr Bon-Kwon Koo, Seoul National University Hospital and Elsevier. This figure was published in the Journal of Cardiovascular Computed Tomography (Min et al. Rationale and design of DeFACTO study. 2011;5(5):301-309),32 copyright Elsevier (2011).

sponsored by HeartFlow, Inc. This model suggested that CTA þ FFRCT would limit ICA procedures and reduce the rate of PCI from 81/100 to 49/100 with similar 1-year mortality and myocardial infarct rates, 2.6% in the ICA group and 2.3% in the CTA þ FFRCT group. These findings suggested a significant decrease in estimated initial treatment cost per patient, from $10,702 to $7674 in the patients who underwent CTA þ FFRCT, assuming the cost of FFRCT to be $1500. However, the actual cost and reimbursement model of FFRCT is yet to be determined.

3.5.

DeFACTO

After the publication of DISCOVER-FLOW, the Determination of Fractional Flow Reserve by Anatomic Computed Tomographic Angiography (DeFACTO) study, an international multicenter clinical trial at 17 centers in 5 countries, was conducted to evaluate the accuracy of FFRCT when compared with the gold standard of invasive FFR in the diagnosis of hemodynamically significant coronary stenosis.6 The primary end point was to determine whether the per-patient diagnostic accuracy of CTA þ FFRCT vs FFR
0.8 could exceed 70% using a 1-sided test at a P value < .05 with power calculation based on the DISCOVER-FLOW findings. Among 285 patients who underwent CTA, ICA, FFR, and FFRCT, 252 had evaluable CTA and FFRCT. Thirty-one patients were excluded for nonevaluable CTA, and 2 patients were excluded for the inability to colocalize the FFR wire to its corresponding location on CTA. CTA alone was 64% accurate (95% CI, 58%e70%), with 84% sensitivity, 42% specificity, 61%

PPV, and 72% NPV in predicting FFR
0.80. In an intention-todiagnose analysis, on a per-patient basis, FFRCT was 73% accurate (95% CI, 67%e78%), with 90% sensitivity, 54% specificity, 67% PPV, and 84% NPV compared with FFR
0.80 (Table 2). Direct per-vessel correlation of FFRCT to FFR was moderate (Pearson correlation coefficient 0.63; 95% CI, 0.56e0.68). In a separate patient-based analysis examining those patients with intermediate stenosis judged to be 30% to 70% by CTA, accuracy of FFRCT was 71% (95% CI, 61%e80%), sensitivity was 82% (95% CI, 63%e92%), specificity was 66% (95% CI, 53%e77%), PPV was 54% (95% CI, 39%e68%), and NPV was 88% (95% CI, 75%e95%), which was higher compared with CTA alone, with an accuracy of 57% (95% CI, 46%e67%), sensitivity of 37% (95% CI, 22%e56%), specificity of 66% (95% CI, 53%e77%), PPV of 34% (95% CI, 20%e53%), and NPV of 68% (95% CI, 55%e79%). Although in the DeFACTO trial FFRCT improved diagnostic accuracy compared with coronary CTA alone when invasive FFR
0.80 served as the reference standard, the study did not achieve the prespecified end point for diagnostic accuracy of FFRCT when compared with FFR.

3.6.

HeartFlow NXT

A third validation study, HeartFlow Next steps (NXT) was published in 2014, evaluating FFRCT to invasive FFR for 251 patients with CTA stenosis of 30% to 90% in 10 sites in Europe, Asia, and Australia.29 This study used a refined version of the FFRCT technology that included an improved ability to identify

J o u r n a l o f C a r d i o v a s c u l a r C o m p u t e d T o m o g r a p h y 8 ( 2 0 1 4 ) 2 7 2 e2 8 1

the luminal boundary and an improved physiologic model of microcirculatory resistance. Overall, there was a specificity of 79% and accuracy of 81% in predicting FFR
0.8. Twelve percent of CT scans were excluded because of image artifacts. In patients with intermediate (30%e70%) stenosis, FFRCT correctly reclassified 68% of false-positive patients as true negatives.

3.7.

FFRCT current limitations and future directions

Although the initial results of the DISCOVER-FLOW and DeFACTO studies were encouraging, the expected accuracy and specificity was lower than expected in the detection of ischemia in DeFACTO, though after technological improvements showed improved results in the HeartFlow NXT trial. There remain several limitations to the accuracy and use of FFRCT. These include the mathematical assumptions made regarding coronary flow, resistance, and luminal size as well as the expected response to hyperemia that will inevitably vary from patient to patient. Uptake of a proprietary technology with a subsequent need for off-site, supercomputer analysis may also limit widespread use. Achieving a highquality CTA with good contrast and free of motion artifact remains of paramount importance. Although substudy analysis of scans with these artifacts is promising, the number of these patients included was very small; for example the effect of significant coronary calcification (Agatson score >400) was only evaluated in 11 patients.27 There are several potential areas of future research: (1) the role of FFRCT in clinical decision making and outcomes, (2) the impact of FFRCT on downstream resource utilization, such as cardiac catheterization, (3) identifying the ideal patient profile by risk and baseline CT characteristics to perform FFRCT, and (4) prospective evaluation of cost efficiency. Some of these questions will be addressed by the ongoing (http://www. clinicaltrials.gov; NCT01943903) PLATFORM study, “Prospective Longitudinal Trial of FFRCT Outcome and Resource Impacts”, that will evaluate the use of FFRCT vs an individual institution’s standard care on the primary outcome of the rate of negative ICA, along with secondary outcomes of all-cause death, nonfatal myocardial infarction, and urgent revascularization.

4.

Concluding thoughts

Numerous clinical landmark trials (such as COURAGE and FAME) have demonstrated that the clinical benefit of revascularization is largely limited to ischemia-causing lesions and not anatomic grading alone. Hence, the continued development and validation through randomized controlled trials with hard outcomes analysis of imaging tests that can determine lesion-specific ischemia are highly important.30 The most prominent ongoing study evaluating the role of ischemia is the National Heart, Lung, and Blood Instituteefunded “International Study of Comparative Health Effectiveness With Medical and Invasive Approaches (http://www. clinicaltrials.gov: NCT01471522)” trial that aims to compare ICA and revascularization plus OMT with a conservative approach of OMT in patients with confirmed moderate or severe ischemia on stress imaging (SPECT, stress echo, or

279

stress CMR) with a primary end point of cardiovascular death or nonfatal myocardial infarction.31 Limitations in our most commonly used noninvasive methods for physiologic assessment of ischemia heighten the need for improved techniques. By adding to the established strength of anatomic assessment using CTA, detection of physiologic ischemia by CT aims to save patients an additional test, prevent unnecessary invasive procedures, and improve outcomes. For CTP, the need for appropriate CT hardware on multiple vendors, a standardization of protocol, further validation, and substantiation of multicenter prognostic data are required before widespread clinical adoptability. At the same time, as the CTP protocol can conceivably be performed on an individual institution’s site, this may lead to wider clinical uptake. FFRCT, although a very exciting concept, at present remains a proprietary experimental technology with the need to improve the diagnostic accuracy and reduce false negatives, as the overall specificity appears, at best, incrementally better than CTA alone. FFRCT, when compared with CTP, may have benefit in obtaining functional information without added radiation. FFRCT also has the attractive potential for modeling the effect of coronary intervention before invasive angiography, although this requires prospective validation. Importantly, there remains a need to devise an appropriate reimbursement model for CT scans reprocessed by HeartFlow, Inc for FFRCT measurements before approval for clinical practice. It should be noted that FFRCT is not yet commercially available in the United States. For the present, although outcomes benefits for SPECT, stress echocardiography, stress CMR, and stress PET have also not been rigorously validated in multicenter randomized trials (with the aforementioned ISCHEMIA trial ongoing), these noninvasive tests remain the preferred noninvasive physiologic tests of ischemia. CTP remains a method confined to the limited number of centers that have tested its validity, and FFRCT remains firmly investigational. However, with several prospective clinical studies underway that will include outcomes data, coupled with anticipated technological advances, CTA with the added assessment of ischemia through CTP and/or FFRCT aims to offer clinicians an evidence-based, comprehensive, anatomic- and physiologic-based alternative to our current methods to detect significant CAD.

references

1. Miller JM, Rochitte CE, Dewey M, et al. Diagnostic performance of coronary angiography by 64-row CT. N Engl J Med. 2008;359:2324e2336. 2. Boden WE, O’Rourke RA, Teo KK, et al. Optimal medical therapy with or without PCI for stable coronary disease. N Engl J Med. 2007;356:1503e1516. 3. Tonino PA, De Bruyne B, Pijls NH, et al. Fractional flow reserve versus angiography for guiding percutaneous coronary intervention. N Engl J Med. 2009;360:213e224. 4. De Bruyne B, Pijls NH, Kalesan B, et al. Fractional flow reserveeguided PCI versus medical therapy in stable coronary disease. N Engl J Med.367:991e1001. 5. Tobis J, Azarbal B, Slavin L. Assessment of intermediate severity coronary lesions in the catheterization laboratory. J Am Coll Cardiol. 2007;49:839e848.

280

J o u r n a l o f C a r d i o v a s c u l a r C o m p u t e d T o m o g r a p h y 8 ( 2 0 1 4 ) 2 7 2 e2 8 1

6. Min JK, Leipsic J, Pencina MJ, et al. Diagnostic accuracy of fractional flow reserve from anatomic CT angiography. JAMA. 2012;308:1237e1245. 7. Blankstein R, Shturman LD, Rogers IS, et al. Adenosineinduced stress myocardial perfusion imaging using dualsource cardiac computed tomography. J Am Coll Cardiol. 2009;54:1072e1084. 8. George RT, Arbab-Zadeh A, Miller JM, et al. Adenosine stress 64- and 256-row detector computed tomography angiography and perfusion imaging: a pilot study evaluating the transmural extent of perfusion abnormalities to predict atherosclerosis causing myocardial ischemia. Circ Cardiovasc Imaging. 2009;2:174e182. 9. Cury RC, Magalhaes TA, Borges AC, et al. Dipyridamole stress and rest myocardial perfusion by 64-detector row computed tomography in patients with suspected coronary artery disease. Am J Cardiol. 2010;106:310e315. 10. Cury RC K T, Feaheny K. A randomized, multi-center, multivendor study comparing myocardial perfusion imaging with regadenoson stress computed tomography perfusion and single photon emission computed tomography. Society of Cardiovascular Computed Tomography 2013 Annual Scientific Sessions. 2013;Abstract 118. 11. Ho KT, Chua KC, Klotz E, Panknin C. Stress and rest dynamic myocardial perfusion imaging by evaluation of complete time-attenuation curves with dual-source CT. JACC Cardiovasc Imaging. 2010;3:811e820. 12. Huber AM, Leber V, Gramer BM, et al. Myocardium: dynamic versus single-shot CT perfusion imaging. Radiology. 2013;269:378e386. 13. So A, Hsieh J, Imai Y, et al. Prospectively ECG-triggered rapid kV-switching dual-energy CT for quantitative imaging of myocardial perfusion. JACC Cardiovasc Imaging. 2012;5:829e836. 14. Bettencourt N, Chiribiri A, Schuster A, et al. Direct comparison of cardiac magnetic resonance and multidetector computed tomography stress-rest perfusion imaging for detection of coronary artery disease. J Am Coll Cardiol; 2013. 15. George RT, Arbab-Zadeh A, Cerci RJ, et al. Diagnostic performance of combined noninvasive coronary angiography and myocardial perfusion imaging using 320-MDCT: the CT angiography and perfusion methods of the CORE320 multicenter multinational diagnostic study. AJR Am J roentgenology. 2011;197:829e837. 16. George RT, Arbab-Zadeh A, Miller JM, et al. Computed tomography myocardial perfusion imaging with 320-row detector computed tomography accurately detects myocardial ischemia in patients with obstructive coronary artery disease. Circ Cardiovasc Imaging. 2012;5:333e340. 17. Ko BS, Cameron JD, Leung M, et al. Combined CT coronary angiography and stress myocardial perfusion imaging for hemodynamically significant stenoses in patients with suspected coronary artery disease: a comparison with fractional flow reserve. JACC Cardiovasc Imaging. 2012;5:1097e1111. 18. Koepfli P, Wyss CA, Namdar M, et al. Beta-adrenergic blockade and myocardial perfusion in coronary artery disease: differential effects in stenotic versus remote myocardial segments. J Nucl Med : official Publ Soc Nucl Med. 2004;45:1626e1631. 19. Kurata A, Mochizuki T, Koyama Y, et al. Myocardial perfusion imaging using adenosine triphosphate stress multi-slice spiral computed tomography: alternative to stress myocardial perfusion scintigraphy. Circ J. 2005;69:550e557. 20. Ko BS, Cameron JD, Meredith IT, et al. Computed tomography stress myocardial perfusion imaging in patients considered for revascularization: a comparison with fractional flow reserve. Eur Heart J. 2012;33:67e77.

21. Rochitte CE, George RT, Chen MY, et al. Computed tomography angiography and perfusion to assess coronary artery stenosis causing perfusion defects by single photon emission computed tomography: the CORE320 study. Eur Heart J. 2014;35:1120e1130. 22. Cury RC, Kitt TM, Feaheny K, Akin J, George RT. Regadenosonstress myocardial CT perfusion and single-photon emission CT: rationale, design, and acquisition methods of a prospective, multicenter, multivendor comparison. J Cardiovasc Comput Tomogr. 2014;8:2e12. 23. Rief M, Zimmermann E, Stenzel F, et al. Computed tomography angiography and myocardial computed tomography perfusion in patients with coronary stents: prospective intraindividual comparison with conventional coronary angiography. J Am Coll Cardiol. 2013;62:1476e1485. 24. Taylor CA, Fonte TA, Min JK. Computational fluid dynamics applied to cardiac computed tomography for noninvasive quantification of fractional flow reserve: scientific basis. J Am Coll Cardiol. 2013;61:2233e2241. 25. Koo BK, Erglis A, Doh JH, et al. Diagnosis of ischemia-causing coronary stenoses by noninvasive fractional flow reserve computed from coronary computed tomographic angiograms. Results from the prospective multicenter DISCOVER-FLOW (Diagnosis of Ischemia-Causing Stenoses Obtained Via Noninvasive Fractional Flow Reserve) study. J Am Coll Cardiol. 2011;58:1989e1997. 26. Min JK, Koo BK, Erglis A, et al. Usefulness of noninvasive fractional flow reserve computed from coronary computed tomographic angiograms for intermediate stenoses confirmed by quantitative coronary angiography. Am J Cardiol. 2012;110:971e976. 27. Min JK, Koo BK, Erglis A, et al. Effect of image quality on diagnostic accuracy of noninvasive fractional flow reserve: results from the prospective multicenter international DISCOVER-FLOW study. J Cardiovasc Comput Tomogr. 2012;6:191e199. 28. Hlatky MA, Saxena A, Koo BK, Erglis A, Zarins CK, Min JK. Projected costs and consequences of computed tomographydetermined fractional flow reserve. Clin Cardiol. 2013;36:743e748. 29. Norgaard BL, Leipsic J, Gaur S, et al. Diagnostic performance of noninvasive fractional flow reserve derived from coronary computed tomography angiography in suspected coronary artery disease: The NXT trial (Analysis of Coronary Blood Flow Using CT Angiography: Next Steps). J Am Coll Cardiol. 2014;63:1145e1155. 30. Arai AE. Computed tomography perfusion to assess physiological significance of coronary stenosis in the post-FAME era (Fractional Flow Reserve Versus Angiography for Multivessel Evaluation). J Am Coll Cardiol. 2013;62:1486e1487. 31. International study of comparative health effectiveness with medical and invasive approaches (ischemia). Available at: ClinicalTrials.gov Identifier: NCT01471522. 2012. 32. Min JK, Berman DS, Budoff MJ, et al. Rationale and design of the DeFACTO (Determination of Fractional Flow Reserve by Anatomic Computed Tomographic Angiography) study. J Cardiovasc Comput Tomogr. 2011;5:301e309. 33. Rocha-Filho JA, Blankstein R, Shturman LD, et al. Incremental value of adenosine-induced stress myocardial perfusion imaging with dual-source CT at cardiac CT angiography. Radiology. 2010;254:410e419. 34. Ko SM, Choi JW, Song MG, et al. Myocardial perfusion imaging using adenosine-induced stress dual-energy computed tomography of the heart: comparison with cardiac magnetic resonance imaging and conventional coronary angiography. Eur Radiol. 2011;21:26e35.

J o u r n a l o f C a r d i o v a s c u l a r C o m p u t e d T o m o g r a p h y 8 ( 2 0 1 4 ) 2 7 2 e2 8 1

35. Bamberg F, Becker A, Schwarz F, et al. Detection of hemodynamically significant coronary artery stenosis: incremental diagnostic value of dynamic CT-based myocardial perfusion imaging. Radiology. 2011;260:689e698. 36. Greif M, von Ziegler F, Bamberg F, et al. CT stress perfusion imaging for detection of haemodynamically relevant coronary stenosis as defined by FFR. Heart. 2013;99:1004e1011.

281

37. Choo KS, Hwangbo L, Kim JH, et al. Adenosine-stress lowdose single-scan CT myocardial perfusion imaging using a 128-slice dual-source CT: a comparison with fractional flow reserve. Acta radiol. 2013;54:389e395. 38. Nasis A, Ko BS, Leung MC, et al. Diagnostic accuracy of combined coronary angiography and adenosine stress myocardial perfusion imaging using 320-detector computed tomography: pilot study. Eur Radiol. 2013;23:1812e1821.