JOURNAL OF THE AMERICAN COLLEGE OF CARDIOLOGY VOL. 78, NO.

20, 2021

ª 2021 BY THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION

PUBLISHED BY ELSEVIER

ORIGINAL INVESTIGATIONS

Diagnostic Performance of Dynamic

Myocardial Perfusion Imaging Using
Dual-Source Computed Tomography
Kakuya Kitagawa, MD,a,* Satoshi Nakamura, MD,b,* Hideki Ota, MD,c Ryo Ogawa, MD,d Takehito Shizuka, MD,e
Tadahiro Kubo, MD,f Yan Yi, MD,g Tatsuro Ito, MD,h Naoki Nagasawa, RT,b Taku Omori, MD,i Shiro Nakamori, MD,i
Tairo Kurita, MD,i Jun Sugisawa, MD,j Naoki Hatori, MD,e Hitoshi Nakashima, MD,f Yining Wang, MD,g
Teruhito Kido, MD,d Kouki Watanabe, MD,k Yasuharu Matsumoto, MD,j,l Kaoru Dohi, MD,i Hajime Sakuma, MDm

ABSTRACT

BACKGROUND Single-center studies indicated a high diagnostic accuracy of dynamic computed tomography perfusion
(CTP) imaging in the diagnosis of coronary artery disease (CAD).

OBJECTIVES This prospective multicenter study determined the diagnostic performance of combined coronary
computed tomography angiography (CTA) and CTP for detecting hemodynamically significant CAD defined by invasive
coronary angiography (ICA) with fractional flow reserve (FFR).

METHODS Seven centers enrolled 174 patients with suspected or known CAD who were clinically referred for ICA. CTA and
dynamic CTP were performed using dual-source CT before ICA. FFR was done as part of ICA in the case of 26% to 90%
coronary diameter stenosis. Hemodynamically significant stenosis was defined as FFR of <0.8 or >90% stenosis on ICA.

RESULTS The study protocol was completed in 157 participants, and hemodynamically significant stenosis was detected
in 76 of 157 patients (48%) and 112 of 442 vessels (25%). According to receiver-operating characteristic curve analysis,
adding dynamic CTP to CTA significantly increased the area under the curve from 0.65 (95% CI: 0.57-0.72) to 0.74
(95% CI: 0.66-0.81; P ¼ 0.011) on the patient level, with decreased sensitivity (93% vs 72%; P < 0.001), improved
specificity (36% vs 75%; P < 0.001), and improved overall accuracy (64% vs 74%; P < 0.001).

CONCLUSIONS In this prospective multicenter study on dynamic CTP, the combination of anatomic assessment with
coronary CTA and functional evaluation with dynamic CTP allowed more accurate identification of hemodynamically
significant CAD compared with CTA alone. However, the clinical significance of this approach needs to be further
investigated, including its usefulness in improving prognosis. (Assessment of Myocardial Perfusion Linked to Infarction
and Fibrosis Explored With Dual-Source CT [AMPLIFiED]; UMIN000016353) (J Am Coll Cardiol 2021;78:1937–1949)
© 2021 by the American College of Cardiology Foundation.

From the aDepartment of Advanced Diagnostic Imaging, Mie University Graduate School of Medicine, Tsu, Japan; bDepartment of
Listen to this manuscript’s
Radiology, Mie University Hospital, Tsu, Japan; cDepartment of Diagnostic Radiology, Tohoku University Hospital, Sendai, Japan;
audio summary by
d
Department of Radiology, Ehime University Graduate School of Medicine, Matsuyama, Japan; eDepartment of Cardiology, Na-
Editor-in-Chief
tional Hospital Organization Takasaki General Medical Center, Takasaki, Japan; fDepartment of Cardiovascular Medicine, National
Dr Valentin Fuster on
Hospital Organization Kagoshima Medical Center, Kagoshima, Japan; gDepartment of Radiology, Peking Union Medical College
JACC.org.
Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China; hDepartment of Cardiovascular
Medicine, Kobe University Graduate School of Medicine, Kobe, Japan; iDepartment of Cardiology and Nephrology, Mie University
Graduate School of Medicine, Tsu, Japan; jDepartment of Cardiovascular Medicine, Tohoku University Graduate School of
Medicine, Sendai, Japan; kDivision of Cardiology, Saiseikai Matsuyama Hospital, Matsuyama, Japan; lDepartment of Cardiovas-
cular Medicine, Shioya Hospital, International University of Health and Welfare, Yaita, Japan; and the mDepartment of Radiology,
Mie University Graduate School of Medicine, Tsu, Japan. *Drs Kitagawa and Nakamura equally contributed to this work.

ISSN 0735-1097/$36.00 https://doi.org/10.1016/j.jacc.2021.08.067

1938 Kitagawa et al JACC VOL. 78, NO. 20, 2021

Diagnostic Performance of Dynamic CT Perfusion Imaging in a Prospective Multicenter Study NOVEMBER 16, 2021:1937–1949

C
ABBREVIATIONS oronary computed tomography best of our knowledge, no multicenter study has yet
AND ACRONYMS angiography (CTA) has high sensi- assessed the diagnostic value of dynamic CTP. There-
tivity and an excellent negative pre- fore, this prospective, multicenter study sought to
ATP = adenosine triphosphate
dictive value for detecting obstructive determine the diagnostic performance of dynamic CTP
AUC = area under the curve
coronary artery disease (CAD) (1,2). Despite for the detection of hemodynamically significant cor-
CAD = coronary artery disease
its usefulness in the evaluation of coronary onary stenosis defined by invasive coronary angiog-
CT = computed tomography artery morphology, CTA is limited in deter- raphy (ICA) with FFR. It was hypothesized that
CTA = computed tomography mining the hemodynamic significance of myocardial blood flow (MBF) quantification with dy-
angiography
anatomic lesions (3). Considering the namic CTP may provide an incremental diagnostic
CTP = computed tomography
growing evidence of the benefit of the value over CTA in the detection of hemodynamically
perfusion
physiology-guided coronary revasculariza- significant stenoses.
ECG = electrocardiogram
tion strategy (4,5), the noninvasive evalua-
FFR = fractional flow reserve
tion of the hemodynamic significance of METHODS
ICA = invasive coronary
coronary stenoses should be valuable. Func-
angiography
tional imaging tests to obtain such informa- STUDY POPULATION. The AMPLIFiED (Assessment
MBF = myocardial blood flow of Myocardial Perfusion Linked to Infarction and
tion include single photon emission
ROC = receiver-operating Fibrosis Explored With Dual-Source CT) study
computed tomography (CT), positron emis-
characteristic
sion tomography, and stress perfusion mag- (UMIN000016353) is a prospective, multicenter,
netic resonance imaging (6). Recently, because of noninterventional observational study involving 7
the advances of CT systems and postprocessing, CT hospitals in 2 Asian countries. This study complies
perfusion imaging (CTP) has been proposed as an with the Declaration of Helsinki. The research proto-
alternative for the functional assessment of CAD. col was approved by the central Institutional Review
Board (the Clinical Research Ethics Review Commit-
SEE PAGE 1950
tee of Mie University Hospital) and by local Institu-
There are currently 2 techniques in CTP: static and tional Review Boards, and all patients gave written
dynamic. In perfusion assessment by static CTP, informed consent.
myocardial enhancement is acquired at a predefined A total of 174 patients between 40 and 85 years of
single time point during the first pass of contrast ma- age with suspected or known CAD who were referred
terial bolus. Static CTP can be performed by 64- for clinically indicated ICA between January 2015 and
detector-row or wider CT scanners with fast rotation December 2018 were prospectively enrolled (Figure 1).
speed; however, the assessment of myocardial perfu- Clinically unstable patients, patients with known
sion with static CTP highly depends on contrast ma- history of allergy to iodinated contrast material, and
terial bolus timing (7) and cardiac motion (8) and is patients after coronary artery bypass grafting were
generally limited to a qualitative and visual evaluation excluded. Detailed study inclusion and exclusion
of enhancement in the myocardium. In contrast, with a criteria are described in the Supplemental Methods.
more technically challenging dynamic CTP, perfusion Of 174 patients enrolled, 158 patients underwent both
assessment is based on myocardial enhancement at a comprehensive cardiac CT examination and ICA
multiple time points during the first pass of the within 60 days after CT, because 16 patients did not
contrast material, which is robust with respect to bolus undergo CT (n ¼ 5) or ICA (n ¼ 11). One patient was
timing and allows for fully quantitative analysis of excluded because of a history of coronary artery
myocardial perfusion. Multiple single-center studies bypass grafting (screening failure). Therefore, 157
have demonstrated the feasibility of dynamic CTP in patients made up the final study population.
the diagnosis of hemodynamically significant stenosis CARDIAC CT. Cardiac CT examinations were per-
defined by fractional flow reserve (FFR) (9-15). High formed using a second- or third-generation dual-
diagnostic performance in single-center studies has source CT (Somatom Definition Flash or Somatom
led to increased expectations for cardiac CT examina- Force, Siemens Healthineers). Cardiac CT protocol
tions, including dynamic CTP. Nevertheless, to the consisted of: 1) a positioning scan; 2) unenhanced CT;

The authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’
institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information,
visit the Author Center.

Manuscript received February 12, 2021; revised manuscript received August 5, 2021, accepted August 25, 2021.
JACC VOL. 78, NO. 20, 2021 Kitagawa et al 1939
NOVEMBER 16, 2021:1937–1949 Diagnostic Performance of Dynamic CT Perfusion Imaging in a Prospective Multicenter Study

F I G U R E 1 Patient Flowchart

– Age: 40-85 years

Patients recruited
– Referred for ICA with
n = 174
suspected or known CAD

Excluded patients: n = 17 No CT: n = 5

No ICA: n = 11
History of CABG: n = 1

Patients who underwent

cardiac CT and ICA
n = 157 (vessels = 471)

Excluded vessels = 29 Hypoplastic RCA: n = 16

FFR not performed: n = 13

Hemodynamically Hemodynamically
significant stenosis (–) significant stenosis (+)
n = 81 (vessels = 330) n = 76 (vessels = 112)

Of 174 patients enrolled, 157 patients made up the final study population. For vessel-based analysis, ICA and FFR demonstrated hemody-
namically significant stenosis in 76 of 157 patients or 112 of 442 vessels after exclusion of 16 hypoplastic RCAs and 13 moderately stenotic
vessels without FFR. CABG ¼ coronary artery bypass graft; CAD ¼ coronary artery disease; CT ¼ computed tomography; FFR ¼ fractional flow
reserve; ICA ¼ invasive coronary angiography; RCA ¼ right coronary artery.

3) dynamic stress CTP; 4) resting coronary CTA; and 5) model combining a simplified deconvolution algo-
CT delayed enhancement. Detailed CT protocols are rithm with the maximum-slope method to obtain a
described in the Supplemental Methods. Briefly, dy- voxel-based MBF map, and has been validated in
namic stress CTP was initiated after >3 minutes of animal models (16-18). In this study, an MBF map of
administrating adenosine triphosphate (ATP) at 3-mm thickness and 1-mm increments was generated
0.16 mg/kg/min by injecting 40 mL of iopamidol. for image analysis.
Dynamic datasets were acquired in the end-systolic
INVASIVE CORONARY ANGIOGRAPHY. ICA was per-
phase for 30 seconds via electrocardiogram (ECG)-
formed according to standard techniques within
triggered axial scan mode repeated at 2 alternating
60 days of the cardiac CT. During the procedure,
table positions (ie, “shuttle mode”). Ten minutes af-
coronary arteries were visually assessed. If diameter
ter dynamic stress CTP, standard prospective coro-
stenosis of 26% to 90% was visually observed in the
nary CTA was performed at rest by bolus injection of
main coronary arteries or their branches ($1.5 mm in
26-mg-iodine/kg/s of iopamidol over 12 seconds with
diameter), FFR was measured using a sensor-tipped
the coronary arteries dilated with a nitrate. Heart rate
guidewire as far as the procedure deemed safe. The
was controlled before coronary CTA with intravenous
FFR was calculated as the ratio between the mean
injection of a b-blocker, if necessary. After CTA,
coronary artery pressure distal to the coronary ste-
additional iopamidol was injected as necessary to
nosis measured by the pressure wire and the mean
make the total injected iodine dose of 600 mgI/kg,
aortic pressure measured through the guiding cath-
and a CT delayed enhancement scan was performed
eter, recorded simultaneously under conditions of
5 minutes after coronary CTA.
maximal hyperemia induced by a continuous intra-
Quantitative analysis of dynamic CTP was per-
venous infusion of ATP (0.16 mg/kg/min for a mini-
formed at each site using commercially available
mum of 3 minutes).
perfusion software (Syngo VPCT body, Siemens
Healthineers). This software is specifically designed IMAGE ANALYSIS. Three blinded independent core
for the “shuttle-mode” dynamic CTP, uses a hybrid laboratories analyzed dynamic CTP, coronary CTA,
1940 Kitagawa et al JACC VOL. 78, NO. 20, 2021

Diagnostic Performance of Dynamic CT Perfusion Imaging in a Prospective Multicenter Study NOVEMBER 16, 2021:1937–1949

and ICA. CT delayed enhancement was analyzed at hemodynamically significant stenosis, a vessel or a
the core laboratory for dynamic CTP. patient was considered positive.
Dynamic CTP was analyzed using the MBF map by
STATISTICAL ANALYSIS. Continuous variables are
an observer blinded to the coronary CTA and ICA
presented as the mean  SD, and categorical variables
data. Regions of interest were manually placed within
are expressed as frequency (percentage). The diag-
each of the 16 American Heart Association myocardial
nostic performance was analyzed based on the area
segments (excluding an apical segment) in the short-
under the receiver-operating characteristic (ROC)
axis view on the MBF map at a minimal distance of
curve (AUC). The ROC curve was constructed in 2 ap-
1 mm from the endo- and epicardial borders to avoid
proaches: 1) dual positivity with a fixed cutoff for CTA
contamination. Absolute MBF values and those
($50% stenosis) and the best cutoff for relative MBF
normalized by remote MBF, which was the highest
after testing varying cutoffs (Supplemental Table 1);
MBF among the 16 segments, were used to assess
and 2) using a risk score based on a logistic regression
myocardial perfusion (relative MBF values) (19).
analysis with the 5-point scale stenosis degree of CTA
The presence of stenosis on CTA was visually
and relative MBF as predictor variables (21). The risk
evaluated in a joint reading by 2 observers blinded to
score was calculated by the following formula: 0.191 þ
the dynamic CTP and ICA data. Coronary segments
0.856  CTA score – 4.520  relative MBF. The Delong
with a reference diameter of $1.5 mm were assessed
test was used to compare the AUCs. Sensitivity, spec-
for the detection of stenosis using a 5-point scale for
ificity, and predictive values were calculated using a
the luminal narrowing: 0: minimal (0%-24%); 1: mild
cutoff of $50% stenosis on CTA and the best cutoff of
(25%-49%); 2: indeterminate; 3: moderate (50%-69%);
dynamic CTP using the dual-positivity approach. The
and 4: severe (70%-100%). Stented vessels were
McNemar test was used to compare the sensitivity,
evaluated for the presence of significant in-stent
specificity, and accuracy of CTA versus CTA plus CTP in
restenosis, defined as >50% reduction in the lumen
detecting hemodynamic significant stenosis. Indeter-
diameter and classified into the following 3 cate-
minate stenosis degree on CTA was considered to
gories: 0: no restenosis; 2: indeterminate; 4: reste-
be $50% stenosis. The diagnostic impact of adding the
nosis. Vessels were deemed indeterminate when it
quantitative CTP parameters to the CTA score on the
was difficult to determine the degree of stenosis
estimation of the probability of hemodynamically
caused by calcification, stent, or motion artifacts.
significant CAD was evaluated by estimating contin-
Coronary arterial segments and myocardial perfusion
uous net reclassification improvement (NRI) and in-
territories were aligned individually using the model
tegrated discrimination improvement. Furthermore,
used in a previous study, which accounts for vari-
patients were divided into 1-, 2-, and 3-vessel disease
ability in coronary anatomy (20). To combine the re-
according to the results of CTA, and the incremental
sults of CTP and CTA, the lowest MBF in the segments
value of CTP to CTA in each group was tested at the
supplied by the vessel was used for the per-vessel
vessel level using ROC curve analysis. The correlation
analysis. In the per-patient analysis, a patient was
between relative MBF and FFR was evaluated using the
considered positive only if a vessel-based dual posi-
Pearson correlation coefficient. A 2-sided P value
tivity (both CTA and CTP positive) was observed.
below 0.05 was considered statistically significant. All
The presence and extent of delayed enhancement
analyses were performed using the SPSS statistical
was visually evaluated in a joint reading by 2 ob-
package, version 23.0 (IBM) and the R statistical
servers blinded to the dynamic CTP, coronary CTA,
package, version 3.4.4 (R Foundation for Statisti-
and ICA data. Extent of delayed enhancement in each
cal Computing).
myocardial segment was assessed using a 4-point
scale: 0: none (0%); 1: 1%-25%; 2: 26%-50%; 3: 51%- RESULTS
75%; and 4: 76%-100%. For combined assessment of
myocardial perfusion and delayed enhancement, PATIENT CHARACTERISTICS. Baseline characteris-
MBF in each segment was measured by avoiding the tics of the study population are shown in Table 1.
area with delayed enhancement. Segments with a Among the 157 patients (age: 66.5  10.4 years; male:
score of 4 were excluded from this combined analysis. 71%), 54 (34%) had known CAD including a history of
The presence of stenosis on ICA was assessed in a myocardial infarction or percutaneous coronary
joint reading by 2 observers blinded to dynamic CTP intervention. Coronary artery stents were found in 63
results. Hemodynamically significant stenosis was of 442 vessels (14%) and 47 of 157 patients (30%).
defined as FFR of <0.8 or >90% stenosis in segments Pretest probability of obstructive CAD in patients
with diameter of $1.5 mm. In the presence of at least 1 with suspected CAD was 37%  26% by clinical model
JACC VOL. 78, NO. 20, 2021 Kitagawa et al 1941
NOVEMBER 16, 2021:1937–1949 Diagnostic Performance of Dynamic CT Perfusion Imaging in a Prospective Multicenter Study

There were 19 (12%) patients with a history of

T A B L E 1 Patient Characteristics
myocardial infarction, 17 of whom had received
Age, y 66.5  10.4
stents. Chronic total occlusions were found in 13
Male 112 (71)
2 vessels (2.9%) on ICA and counted as hemodynami-
Body mass index, kg/m 24.6  4.0
Hypertension 112 (71) cally significant CAD. Among the 19 patients with
Diabetes 47 (30) history of myocardial infarction, 2 patients had total
Dyslipidemia 91 (58) occlusion on ICA (10.5%; 2/19), whereas occlusions
Smoking 69 (44) were found in another 10 patients without history of
Family history of coronary artery disease 16 (10) myocardial infarction (7.2%; 10/138; P ¼ 0.614).
Creatinine, mg/dL 0.82  0.21
Known coronary artery disease 54 (34) RESULTS OF CTA, DYNAMIC CTP, AND CT DELAYED
History of myocardial infarction 19 (12) ENHANCEMENT. Among 442 vessels, minimal, mild,
Prior percutaneous coronary intervention 48 (31) indeterminate, moderate, and severe stenosis was
Symptom found on CTA in 135 (31%), 55 (12%), 137 (31%), 52
Typical 48 (31)
(12%), and 63 (14%) vessels, respectively. Of the 135
Atypical 26 (17)
vessels with minimal stenosis, 95 vessels had no CAD.
Dyspnea 17 (11)
Nonanginal 21 (13)
There were 35, 46, and 42 patients with 1-, 2-, and 3-
Medication vessel disease by CTA when indeterminate lesions
ACE inhibitor/ARB 84 (54) were considered positive. The mean MBF in all vessel
Beta-blocker 56 (36) territories was 105.5  39.4 mL/100 mL/min by dy-
Calcium-channel blocker 71 (45) namic CTP. There was a significant difference in both
Antiplatelet agent 94 (60)
absolute MBF (87.7  32.8 mL/100 mL/min vs 112.0 
Statin 92 (59)
39.3 mL/100 mL/min; P < 0.001) and relative MBF
Pretest probability (suspected CAD)
(0.60  0.17 vs 0.72  0.12; P < 0.001) between the
Clinical model, % 37  26
Clinical þ CCS model, % 39  31
vessel territories with and without hemodynamically
significant stenosis. The optimal cutoff value for the
Values are mean  SD or n (%). detection of hemodynamically significant stenoses
ACE ¼ angiotensin-converting enzyme; ARB ¼ angiotensin receptor blocker;
CAD ¼ coronary artery disease; CCS ¼ coronary calcium score.
was 116 mL/100 mL/min in absolute MBF or 0.71 in
relative MBF, both in all vessels and in vessels
with $50% stenosis on CTA. Delayed enhancement
and 39%  31% by the clinical plus coronary calcium was observed in 189 (8%) segments in 57 (36%) pa-
score model of the CAD Consortium (22). Minor tients, and the extent was 1% to 25% in 61, 26% to 50%
complications during ATP infusion were observed in 5 in 56, 51% to 75% in 15, and 76% to 100% in 57 seg-
patients (dyspnea: n ¼ 2; bronchospasm: n ¼ 1; and ments. Images of CTA, dynamic CTP, and CT delayed
others: n ¼ 2). enhancement in 2 representative cases are shown in
Figures 2 and 3.
RESULTS OF ICA AND FFR. ICA was performed at a
The effective doses for CTA, dynamic CTP, and CT
median of 3 days (interquartile range: 1-8 days) after
delayed enhancement were 3.5  2.9 mSv, 4.9  1.2
CT with no interim events. The clinical indication for
mSv, and 1.9  0.5 mSv, respectively, using a con-
ICA included change in symptom (n ¼ 88; 56%), stress
version coefficient of 0.014. Total radiation dose in
test findings (n ¼ 45; 29%), medically refractory
the whole CT examination was 12.8  4.3 mSv. Radi-
symptoms (n ¼ 16; 10%), ECG change (n ¼ 3; 2%), and
ation dose of the CT protocol and hemodynamic
others (n ¼ 5; 3%). FFR was performed for 88 vessels
response to ATP during CTP are summarized in
in 62 patients. FFR could not be performed for 13
Supplemental Table 2.
moderately stenotic vessels in 11 patients: 9 vessels
were due to technical difficulties, and 4 were deter- DIAGNOSTIC PERFORMANCE OF CTA VERSUS CTA
mined to be mildly stenotic at the time of initial PLUS DYNAMIC CTP. In the dual-positivity approach,
evaluation in the catheterization room and were ROC curve analysis at a patient level (Central
picked up on central reading. These vessels were not Illustration A) showed that CTA had an AUC of 0.65
included in the analysis. In addition, 16 hypoplastic (95% CI: 0.57-0.72) with a fixed cutoff ($50% steno-
right coronary arteries were excluded. Collectively, sis) for the detection of hemodynamically significant
ICA and FFR demonstrated hemodynamically signif- stenosis. Adding dynamic CTP by relative MBF mea-
icant stenosis in 76 (48%) of 157 patients or 112 (25%) surement to CTA significantly increased the AUC to
of 442 vessels. There were 47, 22, and 7 patients with 0.74 (95% CI: 0.66-0.81) with the best cutoff of 0.68
1-, 2-, and 3-vessel disease. (P ¼ 0.011). In the risk score approach, the AUC for
1942 Kitagawa et al JACC VOL. 78, NO. 20, 2021

Diagnostic Performance of Dynamic CT Perfusion Imaging in a Prospective Multicenter Study NOVEMBER 16, 2021:1937–1949

F I G U R E 2 A Case of Anterior Ischemia With Subendocardial Myocardial Infarction

A 75-year-old woman with nonspecific chest pain with a history of myocardial infarction and stent placement to the LAD artery. Dynamic
CTP showed reduced myocardial blood flow in the LAD territory in both the (A) short-axis view and (B) long-axis view (absolute MBF:
41 mL/100 mL/min; relative MBF: 0.25). (C and D) CT delayed enhancement revealed a subendocardial infarction in the anterior wall within
the reduced MBF area (arrows). Coronary CTA (E) showed a high-grade stenosis in the LAD artery just proximal to the stent (arrow). Invasive
coronary angiography (F) revealed >90% stenosis (arrow). CTA ¼ computed tomography angiography; CTP ¼ computed tomography
perfusion; LAD ¼ left anterior descending; MBF ¼ myocardial blood flow; other abbreviations as in Figure 1.

CTA was 0.81 (95% CI: 0.73-0.86), which was signifi- Table 2 and Supplemental Table 3. Accuracy data
cantly improved to 0.85 (95% CI: 0.78-0.90; using $70% stenosis as the CTA positive threshold,
P ¼ 0.027) by relative MBF. The influence of adding which demonstrated a suboptimal sensitivity of CTA
CTP to CTA by requiring dual positivity is shown in (53%) and combined CTA and CTP (39%), are also
JACC VOL. 78, NO. 20, 2021 Kitagawa et al 1943
NOVEMBER 16, 2021:1937–1949 Diagnostic Performance of Dynamic CT Perfusion Imaging in a Prospective Multicenter Study

F I G U R E 3 A Case of Severely Calcified Coronary Artery Without Ischemia

An 80-year-old man with nonspecific chest pain with diabetes mellitus and hypertension. (A) Coronary CTA showed mild stenosis in the right
coronary artery. Obstructive stenoses due to heavily calcified plaque were suspected in the (B) proximal LAD and (C) left circumflex artery
(arrow). (D) On dynamic CTP, both absolute and relative MBF were well above the threshold for ischemia (116 mL/100 mL/min for absolute
MBF and 0.71 for relative MBF) in all myocardial segments, although a mild decrease in MBF was observed subendocardially. (E) There was
no delayed enhancement. Invasive coronary angiography demonstrated 50% stenoses in the (F) proximal LAD and (G) left circumflex artery
(arrows), which were both negative for hemodynamic significance, with FFR values of 0.83 and 0.90, respectively. Abbreviations as in
Figures 1 and 2.
1944 Kitagawa et al JACC VOL. 78, NO. 20, 2021

Diagnostic Performance of Dynamic CT Perfusion Imaging in a Prospective Multicenter Study NOVEMBER 16, 2021:1937–1949

C E N T R A L IL LU ST R A T I O N Diagnostic Performance of Combined Computed Tomography Angiography

Plus Computed Tomography Perfusion to Identify Hemodynamically Significant Coronary Artery Disease

A Patient Level B Vessel Level

100 100

80 80
Sensitivity

Sensitivity
60 60

40 40

20 20

0 0
0 20 40 60 80 100 0 20 40 60 80 100
100-Specificity 100-Specificity

Computed Tomography Angiography: Computed Tomography Angiography:

Area Under the Curve = 0.65 Area Under the Curve = 0.70
Computed Tomography Angiography Computed Tomography Angiography
+ Computed Tomography Perfusion: + Computed Tomography Perfusion:
Area Under the Curve = 0.74 Area Under the Curve = 0.73
Computed Tomography Angiography Score: Computed Tomography Angiography Score:
Area Under the Curve = 0.81 Area Under the Curve = 0.80
Computed Tomography Angiography Score Computed Tomography Angiography Score
+ Computed Tomography Perfusion Values: + Computed Tomography Perfusion Values:
Area Under the Curve = 0.85 Area Under the Curve = 0.84
Kitagawa, K. et al. J Am Coll Cardiol. 2021;78(20):1937–1949.

The receiver-operating characteristic curve and corresponding area under the curve describing the diagnostic performance of coronary computed to-
mography angiography (CTA) alone and combined CTA plus dynamic computed tomography perfusion (CTP) with relative myocardial blood flow to
identify hemodynamically significant coronary artery disease as defined by invasive coronary angiography with fractional flow reserve at (A) the patient
level and (B) the vessel level. Adding CTP to coronary CTA significantly increased the area under the curve at both the patient and vessel levels.

presented in Supplemental Table 4. Thirty-two pa- (95% CI: 61%-82%), 75% (95% CI: 65%-84%), 73%
tients with positive CTA were correctly reclassified as (95% CI: 65%-81%), 74% (95% CI: 66%-81%), and 74%
negative by CTP, and 16 patients with positive CTA (95% CI: 66%-80%), respectively, with a significant
who had hemodynamically significant CAD on ICA improvement in specificity (P < 0.001) and accuracy
were incorrectly reclassified as negative by CTP. CTA (P < 0.001) and a significant decrease in sensitivity
and ICA findings in these 16 patients are summarized (P < 0.001). Continuous net reclassification
in Supplemental Table 5. The sensitivity, specificity, improvement and integrated discrimination
positive and negative predictive values, and accuracy improvement were 0.289 (95% CI: 0.014-0.564;
of CTA alone with a fixed cutoff ($50% stenosis) were P ¼ 0.039) and 0.029 (95% CI: -0.015 to 0.072;
93% (95% CI: 85%-98%), 36% (95% CI: 25%-47%), 58% P ¼ 0.192), respectively, when comparing combined
(95% CI: 53%-62%), 85% (95% CI: 70%-93%), and 64% CTA and CTP to CTA alone. The percentage of patients
(95% CI: 56%-71%). When relative MBF was added to with CAD whose probability of CAD increased by
CTA with a cutoff of 0.68, these values were 72% adding the results of CTP to CTA was 34.2% (26/76),
JACC VOL. 78, NO. 20, 2021 Kitagawa et al 1945
NOVEMBER 16, 2021:1937–1949 Diagnostic Performance of Dynamic CT Perfusion Imaging in a Prospective Multicenter Study

significant correlation between relative MBF and

T A B L E 2 The Influence of Adding CTP to CTA for Detection of
Hemodynamically Significant CAD
invasive FFR (Supplemental Figure 2).
The combination of dynamic CTP by relative MBF
Hemodynamically
Significant CAD on ICA
and CT delayed enhancement did not result in a sig-
Positive Negative
nificant increase in AUC at either the patient and
Patient level
vessel level compared to dynamic CTP alone (0.73 vs
CTA positive 71 52 0.73 [P ¼ 0.665] at the patient level and 0.73 vs 0.72
CTA positive/CTP positive 55 20 [P ¼ 0.237] at the vessel level).
CTA positive/CTP negative 16 32
CTA negative 5 29
DISCUSSION
Vessel level
CTA positive 99 155 AMPLIFiED is a prospective, multicenter study that
CTA positive/CTP positive 82 92 evaluated the incremental diagnostic value of dy-
CTA positive/CTP negative 17 63 namic CTP over CTA for detecting hemodynamically
CTA negative 13 175 significant coronary stenosis defined by ICA with FFR
measurement in patients with suspected or known
CTA positive indicates $50% stenosis; CTP positive indicates relative MBF
of <0.68 (patient level) or 0.76 (vessel level). CAD. ROC curve analysis demonstrated that the
CAD ¼ coronary artery disease; CTA ¼ coronary computed tomography angi-
addition of dynamic CTP to CTA provided a signifi-
ography; CTP ¼ dynamic computed tomography perfusion; ICA ¼ invasive coro-
nary angiography; MBF ¼ myocardial blood flow. cant increase of AUC for the detection of a hemody-
namically significant lesion both at the patient level
and at the vessel level. Improved diagnostic perfor-
mance with dynamic CTP was mainly obtained when
and the percentage of patients without CAD whose
multivessel disease was suspected by CTA. The
probability decreased was 80.4% (65/81).
addition of CT delayed enhancement did not improve
ROC curve analysis at the vessel level was also
the diagnostic performance of dynamic CTP.
presented in the Central Illustration B. In the dual-
The current expert consensus recommendation
positivity approach, the AUC of CTA was 0.70
states that myocardial CTP may be added to coronary
(95% CI: 0.66-0.75) and that of combined CTA and
CTA when there is a high likelihood of ischemic heart
CTP was 0.73 (95% CI: 0.66-0.77), with the best cutoff
disease, known CAD, prior coronary intervention, or
of 0.76 with no significant improvement (P ¼ 0.199).
significant calcifications (23). Our multicenter study
However, in the risk score approach, CTA had an AUC
adds to the evidence supporting this recommenda-
of 0.80 (95% CI: 0.76-0.84), and the addition of dy-
tion and suggests that CTP may be particularly useful
namic CTP by relative MBF to CTA resulted in a sig-
when multivessel disease is suspected on CTA.
nificant increase of the AUC to 0.84 (95% CI: 0.80-
0.87; P ¼ 0.002). In addition, CTP significantly DYNAMIC CTP IMPROVES SPECIFICITY FOR
increased the AUC when added to CTA at the vessel DETECTING HEMODYNAMICALLY SIGNIFICANT
level in a subgroup of patients without prior STENOSIS. Prior studies reported that CTA accu-
myocardial infarction (Supplemental Figure 1). The rately identified the presence and severity of coro-
influence of adding CTP to CTA in the dual-positivity nary stenoses as diagnosed by ICA (1,2). However,
approach at the vessel level is presented in Table 2 CTA poorly predicts hemodynamic parameters
and Supplemental Table 3. determined by FFR, with a specificity of approxi-
In Figure 4, the diagnostic performance of CTA mately 36% (9-11). This was consistent in our study,
alone and CTA plus dynamic CTP by relative MBF is with a specificity of 53% for the detection of hemo-
shown in the subgroups of patients with 1-, 2- and 3- dynamically significant stenoses by CTA. As our re-
vessel disease by CTA. The addition of dynamic CTP sults have shown, the addition of dynamic CTP to
to CTA significantly increased the AUC from 0.80 to CTA significantly increased the specificity from 36%
0.84 (P ¼ 0.041) in 2-vessel disease and from 0.65 to to 75% for the diagnosis of hemodynamically signifi-
0.73 (P ¼ 0.022) in 3-vessel disease, but not in 1-vessel cant stenoses. This increase of the specificity might
disease (from 0.84 to 0.87; P ¼ 0.265). lead to the reduction of unnecessary ICA.
Relative MBF demonstrated higher AUC than ab- Our study used a dynamic CTP technique. An
solute MBF both at the patient and vessel levels, but alternative to the dynamic approach is a static
the differences were not statistically significant (0.73 approach, where the single-shot acquisition is per-
vs 0.68 [P ¼ 0.175] at the patient level and 0.72 vs formed for the assessment of myocardial perfusion. A
0.68 [P ¼ 0.097] at the vessel level). There was a weak prospective multicenter trial using static CTP showed
1946 Kitagawa et al JACC VOL. 78, NO. 20, 2021

Diagnostic Performance of Dynamic CT Perfusion Imaging in a Prospective Multicenter Study NOVEMBER 16, 2021:1937–1949

F I G U R E 4 Value of CTP in 1-, 2-, and 3-Vessel Disease on CTA

A CTA 1VD B CTA 2VD

100 100

80 80

Sensitivity
Sensitivity

60 60

40 40

20 20
n = 95 n = 126
0 0
0 20 40 60 80 100 0 20 40 60 80 100
100-Specificity 100-Specificity

CTA: AUC = 0.84 CTA: AUC = 0.80

CTA + CTP: AUC = 0.87 CTA + CTP: AUC = 0.84
P = 0.265 P = 0.041

C CTA 3VD
100

80
Sensitivity

60

40

20
n = 126
0
0 20 40 60 80 100
100-Specificity

CTA: AUC = 0.65

CTA + CTP: AUC = 0.73
P = 0.022

Diagnostic performance of coronary CTA alone and combined CTA plus dynamic CTP with relative MBF in (A) 1-, (B) 2-, and (C) 3-vessel
disease by CTA. The addition of dynamic CTP to CTA significantly increased the AUC in 2- and 3-vessel disease, but not in 1-vessel disease.
1VD ¼ 1-vessel disease; 2VD ¼ 2-vessel disease; 3VD ¼ 3-vessel disease; AUC ¼ area under the curve; other abbreviations as in Figure 2.

that the combination of CTA and perfusion correctly acquisition protocols are limited to a qualitative and
identifies patients with flow-limiting CAD, defined visual evaluation of myocardial perfusion. On the
as $50% stenosis by ICA causing a perfusion defect by other hand, the dynamic CTP technique is based on
single photon emission CT (21). Single-shot repeated imaging over time during the inflow and
JACC VOL. 78, NO. 20, 2021 Kitagawa et al 1947
NOVEMBER 16, 2021:1937–1949 Diagnostic Performance of Dynamic CT Perfusion Imaging in a Prospective Multicenter Study

washout of the contrast agent in the myocardial tis- assess for ischemia. In fact, previous studies have
sue and vascular compartment to construct time- shown a strong relationship between ischemia and
attenuation curves from which MBF values can be unfavorable prognosis (4-6,25,26). A detailed patient
computed. In this regard, the dynamic approach has risk assessment may be useful in selecting patients
the potential to provide a more objective assessment for a closer follow-up or more aggressive medical
of myocardial perfusion compared with the static therapy, even when medical therapy is the treatment
approach using visual evaluation. Still, further of choice. Additionally, in the ISCHEMIA trial, pa-
research is encouraged to investigate whether MBF tients in the invasive-strategy group had more pro-
quantified with the dynamic approach has added cedural infarctions and fewer nonprocedural
value to the static approach (eg, identification of infarctions during follow-up, implying the impor-
diffuse ischemia using absolute MBF in patients with tance of assessing whether revascularization can be
triple-vessel disease). safely performed and the extent to which revascu-
The diagnostic performance of dynamic CTP was larization can improve myocardial ischemia (27,28).
not affected by considering or not considering the In this respect, visualization of coronary anatomy
delayed enhancement. This may be because the with CTA and quantification of ischemic burden with
viable myocardium around scar also had a lower MBF dynamic CTP could play a pivotal role in optimizing
compared to the remote myocardium, although not as therapy for patients with extensive CAD, especially
much as that of scar. Methods to adequately diagnose when optimal medical therapy has failed (29).
ischemia in the presence of scar need to be further
STUDY LIMITATIONS. First, this was performed in
investigated.
patients referred for ICA who consented to participate
CLINICAL IMPLICATIONS. Our results implied that a in the study. Selection bias may be present because
single CT examination may allow both the assessment need of ICA was evaluated by the referring physician,
of coronary morphology with CTA and the quantita- patient enrollment was not consecutive, and subjects
tive evaluation of the hemodynamic significance of included had a high probability of coronary artery
anatomic lesions with dynamic CTP. However, in a stenosis. Thus, the results are applicable only to
subgroup of patients with single-vessel lesions in populations with the same prevalence as in this
coronary CTAs, the improvement in diagnostic accu- study. In addition, all subjects were Asian and had a
racy with the addition of dynamic CTP was not sta- relatively small body weight, which made scanning
tistically significant. This may be due to the higher easier than in patients of different races with a larger
diagnostic accuracy of coronary CTA alone for single- body size. Second, although among 121 CTP scans
vessel disease. On the other hand, when multivessel performed in the second-generation scanner 26 scans
disease is suspected by coronary CTA, CTP can indi- (21%) showed incomplete z-axis coverage within the
cate which vascular territories are ischemic and may length of 73 mm, all scans performed in the third-
be useful for treatment planning, including revascu- generation scanner had complete z-axis coverage
larization. Given that this study was done on patients within the length of 105 mm. Third, combining CTA
referred to ICA rather than those referred to coronary with dynamic CTP increases the radiation exposure.
CTA, our study findings may suggest that coronary The radiation dose in this study was 4.9 mSv, which
CTA may be appropriate in this high-risk population if was comparable to that in a multicenter trial using
CTP is available. static CTP (5.3 mSv) (21). Nevertheless, more efforts in
However, the improvement in the diagnosis of technical developments are desired to decrease the
hemodynamically significant coronary stenosis with radiation exposure for an extensive use of dynamic
the addition of CTP was relatively small and of un- CTP in clinical practice. Fourth, all scans in this study
clear clinical significance. Moreover, in the recently were performed with dual-source CTs, and therefore,
published ISCHEMIA (International Study of the results are not applicable to non–dual-source CT
Comparative Health Effectiveness with Medical and techniques. The use of shuttle mode does not allow
Invasive Approaches) trial, an initial invasive strategy image capture at every heartbeat for every level of the
as compared with an initial conservative strategy did myocardium. This may lead to less precise time
not confer a reduced risk in patients with moderate to attenuation curves. Fifth, the ICA procedures were
severe myocardial ischemia, suggesting that many of often done in a typical clinical fashion, and FFR was
these patients might possibly be treated medically not performed in all vessels with intermediate ste-
without assessment of myocardial perfusion (24). nosis severity, which can introduce measurement
However, this does not mean that we do not need to selection bias in the performance of FFR. Sixth,
1948 Kitagawa et al JACC VOL. 78, NO. 20, 2021

Diagnostic Performance of Dynamic CT Perfusion Imaging in a Prospective Multicenter Study NOVEMBER 16, 2021:1937–1949

invasive FFR may not be an ideal reference standard ACKNOWLEDGMENT The authors are grateful to Dr
for myocardial perfusion imaging. For example, Kensuke Nishimiya for his help with the analysis of
invasive FFR should be questioned in the presence of invasive coronary angiography.
infarcted myocardium (30). In addition, using nitrate
FUNDING SUPPORT AND AUTHOR DISCLOSURES
to block coronary spasm when measuring invasive
FFR and not using nitrate for dynamic CTP may be Sponsored by Bayer Yakuhin, Ltd. The study sponsor was not involved
one of the reasons for the discrepancy between the in any stage of the study design, data acquisition, data analysis, or
invasive FFR and dynamic CTP. Seventh, CTP was manuscript preparation. Dr Kitagawa is the Endowed Chair for
Department of Advanced Diagnostic Imaging at Mie University Grad-
done before coronary CTA to obtain a “clean” CTP,
uate School of Medicine funded by donations from Siemens Healthcare
without any contamination from drugs to be admin- K.K. and FUJIFILM Medical Co, Ltd. All other authors have reported
istered for the coronary CTA (contrast agent, beta- that they have no relationships relevant to the contents of this paper

blockers, or nitroglycerin). In clinical practice, it is to disclose.

best to perform coronary CTA first and then decide

whether to continue with CTP. Finally, prognostic ADDRESS FOR CORRESPONDENCE: Dr Kakuya Kita-

value of dynamic CTP, which was the second gawa, Department of Advanced Diagnostic Imaging,
endpoint of this study, was not investigated. Mie University Graduate School of Medicine, 2-174
Although there have been several retrospective Edobashi, Tsu, Mie 514-8507, Japan. E-mail: kakuya@
studies related to the prognostic implication of dy- med.mie-u.ac.jp. Twitter: @KakuyaKitagawa.
namic CTP (25,26,31), data from prospective studies PERSPECTIVES
are limited. Further follow-up analysis should be
performed to confirm the safety and prognostic value
COMPETENCY IN PATIENT CARE AND
of dynamic CTP in our study.
PROCEDURAL SKILLS: Combining dynamic CT
perfusion imaging with coronary CT angiography can
CONCLUSIONS
improve the detection of hemodynamically significant
coronary stenosis compared with coronary CT
Our results suggest that the combination of anatomic
angiography alone, especially in patients with multi-
assessment of coronary arteries with CTA and func-
vessel disease.
tional evaluation with MBF quantified by dynamic
CTP allows more accurate identification of hemody-
TRANSLATIONAL OUTLOOK: Further investiga-
namically significant stenosis compared with CTA
tion is needed to evaluate the value of incorporating
alone. However, the clinical significance of this
delayed enhancement imaging along with dual-source
approach, including its usefulness in improving
dynamic CT perfusion imaging and angiography in the
prognosis and cost-effectiveness, needs to be further
assessment of patients with coronary artery disease.
investigated in further multicenter, multivendor
studies with a very large sample size.

REFERENCES

1. Haase R, Schlattmann P, Gueret P, et al. Diag- 4. Tonino PAL, De Bruyne B, Pijls NHJ, et al. tomography perfusion to detect myocardial
nosis of obstructive coronary artery disease using Fractional flow reserve versus angiography for ischemia– derivation cohort from dynamic
computed tomography angiography in patients guiding percutaneous coronary intervention. myocardial computed tomography perfusion. Circ
with stable chest pain depending on clinical N Engl J Med. 2009;360:213–224. J. 2016;80:2506–2512.
probability and in clinically important subgroups:
5. Bech GJ, De Bruyne B, Pijls NH, et al. 8. van Rosendael AR, de Graaf MA, Dimitriu-
meta-analysis of individual patient data. BMJ.
Fractional flow reserve to determine the appro- Leen AC, et al. The influence of clinical and
2019;365:l1945.
priateness of angioplasty in moderate coronary acquisition parameters on the interpretability of
2. Knuuti J, Wijns W, Saraste A, et al. 2019 ESC stenosis: a randomized trial. Circulation. 2001;103: adenosine stress myocardial computed tomogra-
Guidelines for the diagnosis and management of 2928–2934. phy perfusion. Eur Heart J Cardiovasc Imaging.
chronic coronary syndromes: the Task Force for 2016;18:203–211.
6. Hachamovitch R, Hayes SW, Friedman JD,
the Diagnosis and Management of Chronic Cor-
Cohen I, Berman DS. Comparison of the short-term 9. Bamberg F, Becker A, Schwarz F, et al. Detec-
onary Syndromes of the European Society of
survival benefit associated with revascularization tion of hemodynamically significant coronary ar-
Cardiology (ESC). Eur Heart J. 2019;41:407–477.
compared with medical therapy in patients with no tery stenosis: incremental diagnostic value of
3. Meijboom WB, Van Mieghem CAG, van Pelt N, prior coronary artery disease undergoing stress dynamic CT-based myocardial perfusion imaging.
et al. Comprehensive assessment of coronary ar- myocardial perfusion single photon emission Radiology. 2011;260:689–698.
tery stenoses: computed tomography coronary computed tomography. Circulation. 2003;107:
angiography versus conventional coronary angi- 10. Greif M, von Ziegler F, Bamberg F, et al. CT
2900–2907.
ography and correlation with fractional flow stress perfusion imaging for detection of haemo-
reserve in patients with stable angina. J Am Coll 7. Tanabe Y, Kido T, Kurata A, et al. Optimal scan dynamically relevant coronary stenosis as defined
Cardiol. 2008;52:636–643. time for single-phase myocardial computed by FFR. Heart. 2013;99:1004–1011.
JACC VOL. 78, NO. 20, 2021 Kitagawa et al 1949
NOVEMBER 16, 2021:1937–1949 Diagnostic Performance of Dynamic CT Perfusion Imaging in a Prospective Multicenter Study

11. Rossi A, Dharampal A, Wragg A, et al. Diag- 18. Rossi A, Uitterdijk A, Dijkshoorn M, et al. tomography perfusion imaging. J Am Coll Cardiol
nostic performance of hyperaemic myocardial Quantification of myocardial blood flow by Img. 2019;12:1379–1387.
blood flow index obtained by dynamic computed adenosine-stress CT perfusion imaging in pigs
26. Nakamura S, Kitagawa K, Goto Y, et al. Prog-
tomography: does it predict functionally signifi- during various degrees of stenosis correlates well
nostic value of stress dynamic computed tomog-
cant coronary lesions? Eur Heart J Cardiovasc Im- with coronary artery blood flow and fractional
raphy perfusion with computed tomography
aging. 2014;15:85–94. flow reserve. Eur Heart J Cardiovasc Imaging.
delayed enhancement. J Am Coll Cardiol Img.
2013;14:331–338.
12. Kono AK, Coenen A, Lubbers M, et al. Relative 2020;13(8):1721–1734.
myocardial blood flow by dynamic computed 19. Rossi A, Wragg A, Klotz E, et al. Dynamic 27. Shaw LJ, Berman DS, Maron DJ, et al. Optimal
tomographic perfusion imaging predicts hemody- computed tomography myocardial perfusion im- medical therapy with or without percutaneous
namic significance of coronary stenosis better than aging. Circ Cardiovasc Imaging. 2017;10: coronary intervention to reduce ischemic burden.
absolute blood flow. Invest Radiol. 2014;49:801– e005505. Circulation. 2008;117:1283–1291.
807.
20. Cerci RJ, Arbab-Zadeh A, George RT, et al. 28. Nanasato M, Matsumoto N, Nakajima K, et al.
13. Goto Y, Kitagawa K, Uno M, et al. Diagnostic Aligning coronary anatomy and myocardial perfu- Prognostic impact of reducing myocardial ischemia
accuracy of endocardial-to-epicardial myocardial sion territories: an algorithm for the CORE320 identified using ECG-gated myocardial perfusion
blood flow ratio for the detection of significant multicenter study. Circ Cardiovasc Imaging. SPECT in Japanese patients with coronary artery
coronary artery disease with dynamic myocardial 2012;5:587–595. disease: J-ACCESS 4 study. Int J Cardiol. 2018;267:
perfusion dual-source computed tomography. Circ 202–207.
21. Rochitte CE, George RT, Chen MY, et al.
J. 2017;81:1477–1483.
Computed tomography angiography and perfusion 29. Ferraro R, Latina JM, Alfaddagh A, et al.
14. Coenen A, Lubbers MM, Kurata A, et al. Diag- to assess coronary artery stenosis causing perfu- Evaluation and management of patients with sta-
nostic value of transmural perfusion ratio derived sion defects by single photon emission computed ble angina: beyond the ischemia paradigm: JACC
from dynamic CT-based myocardial perfusion im- tomography: the CORE320 study. Eur Heart J. state-of-the-art review. J Am Coll Cardiol.
aging for the detection of haemodynamically 2014;35:1120–1130. 2020;76:2252–2266.
relevant coronary artery stenosis. Eur Radiol.
2017;27:2309–2316. 22. Bittencourt MS, Hulten E, Polonsky TS, et al. 30. Bruyne BD, Pijls NHJ, Bartunek J, et al. Frac-
European Society of Cardiology–recommended tional flow reserve in patients with prior myocar-
15. Pontone G, Baggiano A, Andreini D, et al. coronary artery disease consortium pretest prob- dial infarction. Circulation. 2001;104:157–162.
Stress computed tomography perfusion versus ability scores more accurately predict obstructive
31. Meinel FG, Pugliese F, Schoepf UJ, et al.
fractional flow reserve CT derived in suspected coronary disease and cardiovascular events than
Prognostic value of stress dynamic myocardial
coronary artery disease. J Am Coll Cardiol Img. the Diamond and Forrester score. Circulation.
perfusion CT in a multicenter population with
2019;12:1487–1497. 2016;134:201–211.
known or suspected coronary artery disease. Am J
16. Mahnken AH, Klotz E, Pietsch H, et al. Roentgenol. 2017;208:761–769.
23. Patel AR, Bamberg F, Branch K, et al. Society
Quantitative whole heart stress perfusion CT
of cardiovascular computed tomography expert
imaging as noninvasive assessment of hemody-
consensus document on myocardial computed
namics in coronary artery stenosis: preliminary KEY WORDS 4-dimensional computed
tomography perfusion imaging. J Cardiovasc
animal experience. Invest Radiol. 2010;45:298– tomography, coronary artery disease,
Comput Tomogr. 2020;14:87–100.
305. multicenter study, myocardial ischemia,
24. Maron DJ, Hochman JS, Reynolds HR, et al. Initial
17. Bamberg F, Hinkel R, Schwarz F, et al. Accu- myocardial perfusion imaging
invasive or conservative strategy for stable coronary
racy of dynamic computed tomography adenosine
disease. N Engl J Med. 2020;382:1395–1407.
stress myocardial perfusion imaging in estimating
myocardial blood flow at various degrees of cor- 25. Nakamura S, Kitagawa K, Goto Y, et al. Incre- A PP END IX For an expanded Methods section
onary artery stenosis using a porcine animal mental prognostic value of myocardial blood flow as well as supplemental tables and figures,
model. Invest Radiol. 2012;47:71–77. quantified with stress dynamic computed please see the online version of this paper.