Coronary flow quantification estimated by dynamic 320-detector CT angiography: validation by ¹³N ammonia PET myocardial flow reserve

Abstract

Objectives:

Resting coronary flow index (rCFI) estimated by 320-detector low-dose dynamic coronary CT angiography (CCTA) is a direct flow quantification using intracoronary attenuation. We propose modified-rCFI from new protocol combining dynamic scan and standard CCTA using dose-modulation, and validate its consistency with quantitative values and ischemia depicted by ¹³N-ammonia PET (NH₃-PET).

Methods:

46 patients who underwent dynamic CCTA and NH₃-PET for coronary artery disease were evaluated using original rCFI in 21 patients and modified-rCFI in 25 patients. Two types of rCFI were calculated for three major coronary arteries. Myocardial blood flow (MBF) at rest and stress, myocardial flow reserve (MFR), and the presence or absence of ischemia for three major territories were depicted by NH₃-PET. Coronary territories were categorized as territories with MFR <2.0, ≥2.0, or with and without ischemia. Receiver operating characteristic analysis was performed to determine the optimal cut-off of rCFI to distinguish territories with MFR <2.0 or the presence of ischemia.

Results:

rCFI and modified-rCFI had significant positive correlations with stress MBF and MFR. The optical cut-offs of rCFI and modified-rCFI of 0.39 and 0.61 could detect territories with MFR <2.0, with AUCs of 0.75 and 0.73, sensitivities of 48 and 34%, and specificities of 97 and 98%. Optimal cut-offs of rCFI and modified-rCFI distinguished ischemic segments from non-ischemic segments, with AUCs of 0.75 and 0.91, sensitivities of 53 and 50%, and specificities of 93 and 95%.

Conclusion:

Two types of rCFI correlated with quantitative values from NH₃-PET, and were consistent with a high specificity in detecting functional ischemia.

Advances in knowledge:

rCFI can contribute as additional functional test over standard CCTA in clinical work-up.

Introduction

Coronary computed tomography angiography (CCTA) has a high diagnostic accuracy for coronary stenosis and is widely used for morphological evaluation of coronary artery disease (CAD). Based on recent guidelines, functional coronary stenosis, rather than morphological coronary stenosis, is the key in determining whether to perform percutaneous coronary interventions.^1,2 Recently, it has been reported that quantification of resting coronary flow index (rCFI) derived from whole heart dynamic angiography is useful for detecting myocardial ischemia and is associated with coronary risk factors.³ rCFI is a direct flow quantification using intracoronary attenuation and is expected to contribute as additional functional test over standard CCTA in clinical work-up. However, due to the small amount of contrast media and low-dose scanning in the previously reported method, it was difficult to obtain a stable coronary artery attenuation. Therefore, in order to compensate for this limitation, we propose a new protocol that combines low-dose dynamic scanning and standard CCTA at boost scan using dose-modulation, and the modified-rCFI from this protocol.

In addition to being able to detect myocardial ischemia in patients with CAD, nitrogen-13 ammonia positron emission tomography (NH₃-PET) allows the calculation of absolute values of myocardial blood flow (MBF). Furthermore, the ratio of MBF during vasodilation of the coronary arteries is calculated as myocardial flow reserve (MFR). MFR is an index that identifies the severity of CAD and is also useful in detecting multivessel obstructive disease and coronary microvascular disease. MFR less than two has also been established as an independent poor prognostic factor for major adverse cardiac events.^4–6 In this study, rCFI and modified-rCFI from two different protocols will be validated using MBF and MFR obtained from NH₃-PET as reference standards. Besides, we investigated the ability of rCFI to detect myocardial ischemia depicted by NH₃-PET.

Methods and materials

Study population

The study population was comprised of 46 stable patients with suspected or known CAD who underwent both dynamic CCTA and NH₃-PET within 3 months in our institute between June 2016 and December 2019. A total of six patients underwent coronary revascularization or presented myocardial infarction. CCTA and NH₃-PET examinations were performed on these patients to evaluate the current coronary artery and for ischemia assessment. 40 patients with suspected CAD, who had typical angina symptoms and clinical findings, had CCTA conducted before NH₃-PET. In these cases, mild-to-moderate coronary stenosis and high-risk plaques were reported on CCTA and additional NH₃-PET was performed to evaluate functional ischemia. Exclusion criteria were as follows: age <35 years, a body weight of less than 40 kg or more than 80 kg, allergy to iodinated contrast agents, renal dysfunction except for dialysis patients (estimated glomerular filtration rate <30 mL/min/1.73 m²), systolic blood pressure <90 mmHg, prior coronary artery bypass surgery, acute myocardial infarction (within 3 months), unstable angina (recent onset of angina within 1 month, or severe and worsening clinical symptoms), severe left ventricular dysfunction (left ventricular ejection fraction <20%), congestive heart failure (New York Heart Association Class IV), significant valvular disease, non-ischemic cardiomyopathy, and congenital heart disease. Patients with a high heart rate (>70 bpm at the time of scan), arrhythmia including frequent extra beats, or significant motion artifacts in the CCTA scan were also excluded. Stented vessels with chronic total occlusion or with diffuse circumferential advanced calcification were excluded from the analysis.

In addition, among patients with suspected CAD, 48 patients who did not have coronary artery stenosis or plaque on CCTA were enrolled as controls. In these patients, NH₃-PET was not performed. Any medical history of diabetes, hypertension, and hyperlipidemia was extracted from all the patients’ medical records and reviewed. Patient with a prior medical history, regardless of current treatment, were considered positive. Patient characteristics are summarized in Table 1. Informed written consent was obtained from each patient before the study and this study complied with the Declaration of Helsinki and the guidelines of the local institutional review board.

Table 1.

Patient’s characteristics

Variable	Patient group	Control
Number	46	48
Male	29 (63%)	25 (53%)
Mean age (range, years)	71 (36-88)	56 (40–82)
Cardiovascular risk factors (%)
Hypertension	31 (67%)	6 (13%)
Dyslipidemia	28 (61%)	21 (44%)
Diabetes mellitus	17 (37%)	6 (13%)
Never smoker or unknown	28 (61%)	38 (79%)
Positive family history	5 (11%)	10 (21%)
History of CAD (%)
History of myocardial infarction	6 (13%)	0
Previous of PCI	5 (11%)	0

CAD = coronary artery disease, PCI = percutaneous coronary intervention

Dynamic whole-heart coronary CTA protocol

All participants underwent dynamic CCTA with dose modulation using a 320-row CT scanner (Aquilion One; Cannon Medical Systems Co., Tochigi, Japan). Intravenous or oral metoprolol was administered to patients with a heart rate of ≥65 beats/min. Immediately before image acquisition, all patients received sublingual nitroglycerin 0.2 mg.

Protocol A:

Dynamic CCTA scanning was continuously performed in mid-diastole for 8–12 cardiac cycles with prospective ECG-gating axial scans after a 4 s contrast media injection (259 mgI/kg, Iopamiron 370; Bayel Healthcare, Osaka, Japan). The scan parameters were as follows: gantry rotation time, 0.275 s; detector collimation, 320 × 0.5 mm; tube voltage, 80 kV; tube current, 60 mA. This protocol was the same as previously reported³ and was carried out in 21 patients who underwent NH₃-PET and 18 controls between June 2016 and July 2017.

Protocol B:

The new protocol utilizes dose-modulation and consists of one series of standard CCTA at boost scan and low-dose dynamic imaging. The injection rate is the same, but the amount of contrast agent is more than double to provide sufficient coronary artery contrast in the dynamic data. In addition, the total examination time including test scan has been shortened. This protocol was carried out for 25 patients who underwent NH₃-PET and 30 controls between August 2017 and December 2019. First, the test scan was performed at the ascending aorta level without ECG gating. This was done at 1 s intervals from 7 s after injection of the contrast medium until the attenuation in ascending aorta reaches its peak to determine the optimal scan timing for dynamic CCTA. Based on the time-attenuation curve from the test injection, dynamic CCTA scanning was continuously performed in mid-diastole (70–80% of RR interval on ECG) for 8–12 cardiac cycles with prospective ECG-gating axial scans after a 10 s contrast media injection (259 mgI/kg, Iopamiron 370; Bayel Healthcare, Osaka, Japan). One scan of the dynamic CCTA was performed as a boost scan for standard CCTA at the peak phase of the ascending aorta, which was determined by a test injection (Figure 1). For a patient weighing 60 kg, a total of 58 ml of contrast medium was used, 16 ml for the test scan and 42 ml for the dynamic CCTA. The scan flow for Protocols A and B shown in Figure 2. The scan parameters were as follows: gantry rotation time, 0.275 s; detector collimation, 320 × 0.5 mm; tube voltage, 100 kV; tube current, 80 mA at dynamic scan, and auto radiation exposure reducing control at boost scan with a field of view of 200 mm.

Figure 1.

Volume rendering whole heart dynamic coronary CT angiography obtained with protocol B for a control case demonstrates the first pass of contrast media from the heart cavities into their normal coronary arteries with a high spatial resolution and noise reduction. Flow is travelling from the upper left to the lower right over time.

Figure 2.

Scan flow for protocols A and B. In protocol A (upper row), a dynamic scan with a small amount of contrast agent is first performed and rCFI is calculated from this dynamic data. Standard CCTA is then performed. In protocol B (lower row), a non-ECG-gate test scan is first performed to determine the timing of the boost scan. A dynamic scan is then performed, including standard CCTA. Modified rCFI is calculated from this dynamic data. CCTA, coronary CT angiography.

For both protocols, morphological coronary stenosis was analyzed using dedicated software (Ziostation 2 Phyziodynamics; Ziosoft, Tokyo, Japan), based on a combination of transverse sections and automatically generated curved multiplanar reconstruction (MPR) images of target vessels. Images were clinically interpreted by the consensus decision of two experience radiologist and one cardiologist, using the American Heart Association 15-segment model. The degree of stenosis was divided into three categories: 0–49%, 50–69%, and >70%. We recorded the machine-generated volume CT dose index (mGy) and the dose–length product (DLP) for the dynamic CCTA scan. We also estimated the effective radiation dose to the ECG-gated CCTA using an equation and the latest k-factor of 0.026⁷ : Effective radiation dose = DLP×0.026 (mSv).

Quantification of the resting coronary flow

The data set for the dynamic CCTA with 8–12 continuous cardiac cycles was transformed into dedicated software (Ziostation 2 Phyziodynamics; Ziosoft, Tokyo, Japan). Via the motion coherence image processing (MCIP),^3,8,9 the original data set was interpolated by neighboring phases up to 24–36 dynamic image sets (Figure 1). In the converted dynamic series, a spherical volume of interest (VOI) was drawn in the ascending aorta and distal portions of coronary arteries with a diameter of 2 mm (coronary segment #3 in the right coronary artery (RCA), #8 in the left anterior descending artery (LAD), #13–14 in the left circumflex artery (LCX)), and the distal site of the major branches such as the diagonal, ramus, and obscure marginal in cases where significant myocardial ischemia was proven by three-dimensional fusion image of CCTA and NH₃-PET. In coronary arteries with calcified plaque, the VOI was placed in an area that avoided those calcifications. Once the VOI was activated, the coronary contour was automatically extracted by synchronizing with the movement of the coronary artery. Each voxel in the VOI was tracked in all time phases. The CT value was then exported as a time-attenuation curve in Microsoft Excel. The slope of the dynamic curve was defined as its starting point and the peak. In accordance with the maximum slope method, the resting coronary flow index (rCFI) was generated using the ratio of the maximum upslope in the distal coronary artery and that of the ascending aorta (Figure 3). In the graph in Figure 3, if there are any deviating points, check that the VOI does not deviate from the coronary contour in the movie reconstructed by motion coherence image processing. If the VOI deviates from the coronary contour, the setting is redone again. The rCFI for RCA, LAD, and LCX was calculated, and the rCFI obtained from protocol B was named modified-rCFI. detector collimation, 320 × 0.5 mm; tube voltage, 80 kV; tube current, 60 mA.

Figure 3.

Calculation of resting coronary flow index. Graphs show time-density curves of intra coronary attenuation (HU) for the ascending aorta (blue) and the distal portion of the LAD #8 (orange) throughout all phases (upper). The maximum upslope (HU/s) for the sites was calculated from the linear upslope of the CT value, which represents the pass of contrast media. From 3 to 11 s, the maximum upslope for the ascending aorta and distal site of stenosis were 38.743 HU/s and 20.631 HU/s (lower). According to the maximum upslope method, the rCFI of LAD was calculated as 0.53 (20.631/38.743). HU, Hounsfield unit; LAD, left anterior descending artery; rCFI, resting coronary flow index.

MBF and MFR measurements using ¹³N ammonia PET

After making the necessary preparations,^10,11 patients were positioned in a PET-CT system (Biograph, mCT, Siemens Healthcare, Erlangen, Germany). For attenuation correction, a sequential CT scan with tube voltage of 120 kV, tube current of 20 mAs, and 3 mm slice collimation was acquired. Immediately after the administration of 185 MBq (5 mCi) of N-13 ammonia intravenously, ECG-gated acquisition was performed during 10 min with a 16-frame per cardiac cycle, a parallel list-mode acquisition.¹² After PET myocardial perfusion imaging was conducted at rest, an adenosine stress test was performed (administered at 0.12 mg/kg/min for 6 min). 3 min after the administration of the vasodilator, 555 MBq (15 mCi) of N-13 ammonia was infused.¹³

Images were reconstructed using Fourier transform and filtered back-projection with a 12 mm three-dimensional Hann window of the ramp filter. Automatic reorientation of the images, extraction of mean myocardial and left ventricular cavity time-activity curves, and generation of polar maps of absolute MBF and MFR were performed using a dedicated software (Syngo MI cardiology, Siemens Healthcare, Erlangen, Germany). MBF was generated from the time–activity curve of the left ventricular cavity input and myocardial uptake using a two-compartment model and dataset of list mode images obtained in the first 2 min. The repeatedly upgraded software package was used for dose correction (i.e. the difference in residual N-13 ammonia activity between resting and stress images).¹⁴ MBF in RCA, LAD, and LCX territories was calculated and regional MFR was determined as the ratio of hyperemic to resting regional MBF. Regional MFR < 2.0 was considered abnormal.^4,5

Ischemia assessment by static imaging

Two experience radiologist and one cardiologist had access to polar maps and perfusion images of 16 short-axis, 16 vertical long-axis, and 16 horizontal long-axis slices. The degree of uptake was identified using a 17-segment model and the semi-quantitative scoring system (0 = normal, 1 = mildly abnormal, 2 = moderately abnormal, 3 = severely abnormal, and 4 = complete defect). Summed stress scores (SSS), summed resting scores (SRS), and summed difference scores (SDS = SSS – SRS) were calculated. With the consensus of the observers, SDS ≥2 in the RCA, LAD, or LCX territory was considered as the positive of ischemia for each territory.⁵

Statistical analyses

Continuous data were expressed as the mean ± the standard deviation. Testing of differences between demographic and clinical data of rCFI and other factors was done using the Kruskal–Wallis test with Dunn’s multiple comparison test or the Mann–Whitney U test for continuous variables. Measures of the association among rCFI, MBF, and MFR were accomplished using Pearson’s coefficient analysis. A receiver operating cumulative curve (ROC) analysis was performed to determine the optimal cut-off of the rCFI for the detection of territory with an MFR <2.0 or ischemia, and to investigate the area under the curve (AUC), sensitivity, and specificity. All statistical tests conducted were two-sided. A p-value of <0.05 was considered statistically significant when comparing the two groups. A p-value of <0.016 was considered statistically significant when comparing the three groups. All analyses were performed using the JMP statistical software (v. 9.0; JMP, Inc., Cary, NC).

Result rCFI for coronary territory and radiation exposure

The rCFI for the controls was 0.63 ± 0.22 (RCA 0.72 ± 0.26, LAD 0.62 ± 0.18, LCX 0.55 ± 0.19). There was no significant difference in any of the two territories between the RCA, LAD, and LCX. Modified-rCFI for the controls was 0.93 ± 0.20 (RCA 1.02 ± 0.24, LAD 0.94 ± 0.18, LCX 0.83 ± 0.13). Modified-rCFI for RCA was significantly greater than that for LCX (p = 0.0023). There was no significant difference between the RCA and LAD or between the LAD and LCX.

For protocol A, the mean DLP and effective radiation dose used in this study was 569 mGy.cm and 14.8 mSv. For protocol B, the mean DLP and effective radiation dose used in this study was 330 mGy.cm and 8.6 mSv.

Correlation among rCFI, MBF, and MFR

Both rCFI and modified-rCFI had significant positive correlations with stress MBF (Pearson r, 0.305 and 0.475; p = 0.015 and p < 0.0001) (Figure 4). In contrast, there was no significant correlation between rCFI and rest MBF (Pearson r, −0.208; p = 0.101) or between modified-rCFI and rest MBF (Pearson r, −0.009; p = 0.94). As a result, both rCFI and modified-rCFI were positively correlated with MFR (Pearson r, 0.425 and 0.399; p = 0.0005 and p = 0.0004) (Table 2).

Figure 4.

Scattergram shows the results of linear regression analysis between stress MBF and rCFI (left) and modified-rCFI (right). Pearson’s correlation coefficient is provided for each plot. MBF, myocardial blood flow; rCFI, resting coronary flow index.

Table 2.

Correlation among rCFI, MBF, and MFR

	rCFI			Modified-rCFI			rCFI & modified-rCFI
	Pearson r	95% CI	p value	Pearson r	95% CI	p value	Pearson r	95% CI	p value
Rest MBF	−0.208	−0.434 to 0.042	0.101	−0.009	−0.245 to 0.229	0.943	−0.105	−0.271 to 0.068	0.233
Stress MBF	0.305	0.062 to 0.514	0.015	0.475	0.269 to 0.640	<0.0001	0.357	0.199 to 0.498	<0.0001
MFR	0.425	0.199 to 0.609	0.0005	0.399	0.189 to 0.574	0.0004	0.406	0.257 to 0.537	<0.0001

CI, confidence interval; MBF = myocardial blood flow, MFR = myocardial flow reserve;rCFI = resting coronary flow index.

Comparison of rCFI between MFR < 2.0 and>2.0

In protocol A, MFR < 2.0 was found in 27 territories, and MFR > 2.0 was found in 36 territories.

In protocol B, MFR <2.0 was found in 32 territories, and MFR >2.0 was found in 49 territories. Both rCFI and modified-rCFI for territories with an MFR <2.0 (rCFI, 0.42 ± 0.16/modified-rCFI, 0.71 ± 0.18) were significantly lower than those for territories with an MFR >2.0 (rCFI, 0.57 ± 0.15/modified-rCFI, 0.89 ± 0.17) and control (rCFI, 0.63 ± 0.22/modified-rCFI, 0.93 ± 0.20; p < 0.0001) (Figure 5). There was no difference in both rCFI and modified-rCFI between territories with an MFR >2.0 and control. The optical cut-offs for rCFI and modified-rCFI of 0.39 and 0.61 could detect territories with MFR <2.0, with AUCs of 0.75 and 0.73, sensitivities of 48 and 34%, specificities of 97 and 98%, and likelihood ratios of 17.3 and 16.8, respectively.

Figure 5.

Comparison of rCFI and modified-rCFI among territories with MFR ≥2.0 and<2.0 and controls The left shows a box-and-whisker plot for rCFI, and the right displays one for modified-rCFR. Both rCFI and modified-rCFI for territories with MFR <2.0 were significantly lower than those with MFR ≥2.0, and the controls. There was no difference between the territories with MFR ≥2.0 and the controls. *p < 0.0001 The horizontal lines indicate the maximum and minimum values, and the box indicates the 25th and 75th percentiles. MFR, myocardial flow reserve; rCFI, resting coronary flow index.

Diagnostic ability of rCFI for detection of ischemia

In protocol A, NH₃-PET detected 19 ischemic segments and 44 non-ischemic segments. In the protocol B Group, NH₃-PET detected 12 ischemic segments and 63 non-ischemic segments. rCFI and modified-rCFI for ischemia segment (0.40 ± 0.16/0.87±0.17) were significantly lower than those for the non-ischemic segment (0.55 ± 0.16; p = 0.0032/0.60±0.14; p < 0.0001) (Figure 6). ROC analysis revealed that the use of the optimal cut-offs of rCFI and modified-rCFI of 0.36 and 0.61 distinguished ischemic segments from non-ischemic segments, with AUCs of 0.75 and 0.91, sensitivities of 53 and 50%, specificities of 93 and 95%, and likelihood ratios of 7.72 and 10.50. A representative case is presented in Figures 7 and 8, Supplementary Videos 1 and 2.

Figure 6.

Scatter plot shows rCFI and modified-rCFI in ischemic and non-ischemic segments. Both rCFI and modified-rCFI were significantly lower in ischemic than non-ischemic segments (*p < 0.005; **p < 0.0001). The horizontal long line represents the mean value and the upper and lower short lines the standard error of the mean. rCFI, resting coronary flow index.

Figure 7.

A male in his 60 s with effort angina volume rendering whole heart dynamic coronary CT angiography (upper row) obtained from protocol B shows scattered coarse calcifications at the proximal left coronary artery and moderate stenosis at the proximal LAD. According to intracoronary attenuation, the coronary artery was delineated to five colored scales. Warm colors represent high CT values, and cold colors represent low CT values. Dynamic color-coding (lower row) can visualize the first pass of contrast media in the LAD, and demonstrates high attenuation with a delay at the distal site of stenosis (arrow). The modified-rCFI of the LAD is decreased to 0.61. The modified-rCFI values of RCA and LCX are 0.78 and 0.74, respectively. LAD, left anterior descending artery; RCA, right coronary artery; rCFI, resting coronary flow index.

Figure 8.

A male in his 60s with effort angina, the same case as Figure 7 stress and rest perfusion map obtained from NH₃-PET (upper row) shows the area of hypoperfusion in the mid-anterior to apex during the stress condition and fill-in in the same area during the rest condition. This is a typical finding of ischemia in the LAD territory. On the MFR map, the MFR of the LAD territory is reduced to 1.92. The MFRs of the RCA and LCX are two or more. The modified-rCFI is consistent with the MFR results. LAD, left anterior descending artery; LCX, left circumflex artery; MFR, myocardial flow reserve; PET, positron emmision tomography; RCA, right coronary artery.

Relationship between rCFI and degree of coronary artery stenosis

Both rCFI and modified-rCFI decreased with increasing degree of coronary artery stenosis. In protocol A, rCFI of coronary artery with ≥70% stenosis (n = 8, 0.33 ± 0.10) was significantly lower than that of controls (n = 54, 0.63 ± 0.22) or <50% stenosis (n = 39, 054 ± 0.16). There was no significant difference between any of the other two groups (Figure 9). In protocol B, modified-rCFI of coronary artery with ≥70% stenosis (n = 9, 0.66 ± 0.14) was significantly lower than that of controls (n = 90, 0.93 ± 0.20) or <50% stenosis (n = 48, 0.87 ± 0.21). In addition, modified-rCFI of coronary artery with 50 to 69% stenosis (n = 18, 0.74 ± 0.16) was significantly lower than that of controls or <50% stenosis. There was no significant difference between control and <50% stenosis and between 50 to 69% stenosis and ≥70% stenosis (Figure 9).

Figure 9.

Comparison of rCFI and modified-rCFI among coronary arteries with 0–49, 50–69%,≥70% stenosis and controls The left shows a box-and-whisker plot for rCFI, and the right displays one for modified-rCFR. Both rCFI and modified-rCFI decreased with increasing degree of coronary artery stenosis. rCFI of coronary artery with ≥70% stenosis was significantly lower than that of controls or <50% stenosis (*p < 0.05). There was no significant difference between any of the other two groups. Modified-rCFI of coronary artery with ≥70% stenosis was significantly lower than that of controls or <50% stenosis (**p < 0.005). In addition, modified-rCFI of coronary artery with 50–69% stenosis was significantly lower than that of controls or <50% stenosis (*p < 0.05). There was no significant difference between control and <50% stenosis and between 50 to 69% stenosis and ≥70% stenosis. The horizontal lines indicate the maximum and minimum values, and the box indicates the 25th and 75th percentiles. rCFI, resting coronary flow index.

Discussion

The present study proposes a new quantification of resting coronary flow using dynamic CCTA combined with a low dose and boost scan based on a previously reported method,³ and investigates the association with qualitative values obtained from NH₃-PET. Two types of rCFI are derived as the ratio of the upslope in the distal coronary artery CT value and that of the ascending aorta. Aorta attenuation is included in our calculation model as a sort of systemic input correction, which is based on the maximum slope method for calculation of tissue perfusion. These indexes are measured during mid-diastole with sufficiently reduced peripheral perfusion pressures, similar to the resting index such as instantaneous wave-free ratio (iFR) measured from invasive coronary catheters. Invasive iFR is calculated from a pressure range based on the linear correlation between coronary pressure and flow under specific conditions.¹⁵ On the other hand, rCFI is calculated directly from coronary flow and is more in line with this concept.

A particularly interesting result is that both rCFI and modified rCFI had significantly positive correlations with stress MBF. In contrast, no correlation was observed between these two indexes and resting MBF. As a result, these two indices showed positive correlations with MFR, and it was possible to identify a territory with an MFR <2.0 with high specificity. In a study using intravascular Doppler ultrasound technique in dogs,¹⁶ nitroglycerin increases epicardial coronary artery cross-sectional area and peak flow velocity, resulting in a 40–50% increase in coronary blood flow compared to resting state. On the other hand, adenosine does not change epicardial coronary artery cross-sectional area, but decreases peripheral vascular resistance and increases the peak flow velocity, resulting in a coronary blood flow 270% of that at rest. This is reminiscent of the arterioles to capillaries dilating effect of adenosine. Accordingly, rCFI is an indicator that reflects increased blood flow in epicardial coronary arteries that have been dilated by nitrol and have increased flow velocity. If there is a functional stenosis or endothelial damage, there will be a reduction in blood flow distal to the stenosis and rCFI will be decrease (Figure 8). On the other hand, stress MBF and MFR estimated from ammonia PET reflect increased blood flow due to adenosine dilation of arterioles to capillaries. In other words, rCFI represents the coronary macrocirculation and MFR represents its peripheral microcirculation; rCFI regulates MFR as a source of supply for the microcoronary circulation. Based on this relationship, we believe that a correlation between rCFI and MFR was obtained in this study. The rCFI and modified-CFI measurements can be used as functional parameters to determine the revascularization therapy in stenotic coronary arteries. The hybrid assessment of standard CCTA and rCFI is a feasible method that enables morphological and functional assessment in CAD.

Correlations of modified-rCFI with NH₃-PET measurements were similar to those of rCFI. Modified-rCFI was slightly superior to rCFI in detecting ischemia. Since protocol A has more than twice the amount of contrast media than protocol B, the concentration gradient in the coronary artery naturally increases. Consequently, modified-rCFI was greater than rCFI. In addition, the modified-rCFI can be measured with good reproducibility because the contrast of the coronary artery is sufficiently obtained, which makes the starting peak CT value during the dynamic series clearer. Recently, CCTA-estimated fractional flow reserve (FFR-CT) has been used as non-invasive method for detection of significant coronary stenosis.^17,18 Compared to FFR-CT, extra radiation exposure should be noted because a dynamic scan was required to calculate rCFI. The average effective radiation dose for protocols A and B were 14.8 mSv and 8.6 mSv, respectively. Protocol B is less radiation exposed than protocol A. These radiation exposures are considered within the diagnostic reference level of CCTA recently described in Western countries.¹⁹ The process of image analysis did not require supercomputers such as FFR-CT and could be analyzed immediately after imaging in a relatively short time (approximately 30 min). While protocol B assessed coronary stenosis and modified-rCFI from the same data series, protocol A utilizes two separate datasets of dynamic CCTA and standard CCTA. Protocol B is easier to handle than protocol A. The workflow in both protocols, from the start of inspection to diagnosis, takes one day. This novel technique can be added to routine clinical work-ups to predict hemodynamically significant coronary stenosis in patients.

Limitations

The main limitation of this study is the small number of patients in a single-center cohort study. Only 46 patients who underwent both CCTA and NH₃-PET were enrolled, and they were further divided into two protocols (21 for an already validated one and 25 for the one to validate). As the controls did not undergo NH₃-PET, a fully validation of this new protocol is limited. Second, both rCFI and modified-rCFI for controls tended to be greater in the RCA and lower in the LCX. The variability in the RCA was large in both indices. This may be due to the dominance of the RCA and LCX, as well as the diameter of the coronary arteries. Therefore, we think that it is necessary to correct for the normal range in each coronary artery. Third, the sensitivity for ischemia detection is low when the threshold values of rCFI and modified-rCFI are set low. The patients in this study had a high prevalence of CAD. Therefore, it is important to increase specificity and identify patients who do not need coronary revascularization. The rCFI and modified-rCFI are functional assessments that are performed in addition to the morphological evaluation, the high specificity is important in deciding in the treatment plan.

Conclusion

The present study proposes two types of resting coronary flow indices using dynamic CCTA with a 320-detector scanner. Both rCFI and modified-rCFI positively correlated with stress MBF and MFR obtained from NH₃-PET, and were consistent with high specificity in the detection of significant functional ischemia.