Normal values of myocardial blood flow measured with dynamic myocardial computed tomography perfusion

Abstract

Aims

Dynamic myocardial computed tomography (CT) perfusion (DM-CTP) can, in combination with coronary CT angiography (CCTA), provide anatomical and functional evaluation of coronary artery disease (CAD). However, normal values of myocardial blood flow (MBF) are needed to identify impaired myocardial blood supply in patients with suspected CAD. We aimed to establish normal values for MBF measured using DM-CTP, to assess the effects of age and sex, and to assess regional distribution of MBF.

Methods and results

A total of 82 healthy individuals (46 women) aged 45–78 years with normal coronary arteries by CCTA underwent either rest and adenosine stress DM-CTP (n = 30) or adenosine-induced stress DM-CTP only (n = 52). Global and segmental MBF were assessed. Global MBF at rest and during stress were 0.93 ± 0.42 and 3.58 ± 1.14 mL/min/g, respectively. MBF was not different between the sexes (P = 0.88 at rest and P = 0.61 during stress), and no correlation was observed between MBF and age (P = 0.08 at rest and P = 0.82 during stress). Among the 16 myocardial segments, significant intersegmental differences were found (P < 0.01), which was not related to age, sex, or coronary dominance.

Conclusion

MBF assessed by DM-CTP in healthy individuals with normal coronary arteries displays significant intersegmental heterogeneity which does not seem to be affected by age, sex, or coronary dominance. Normal values of MBF may be helpful in the clinical evaluation of suspected myocardial ischaemia using DM-CTP.

Graphical Abstract

Implementation of normal values for clinical evaluation of CAD. An 84-year-old woman with exertional chest pain. Top panels: coronary CT angiography shows right coronary dominance and severe stenosis in the proximal RCA. Stress DM-CTP with low absolute and relative MBF in the inferior wall. Bottom left and central panels: cut-off for mild ischaemia based on the results extracted in this study. Bottom right panel: severity of segmental ischaemia shows mild to severe ischaemia in Segments 3, 4, 5, 10, 11, and 15, corresponding with the case-specific segment allocation of the RCA. MBF, myocardial blood flow; reference MBF, the 75th percentile of flow values.

Open in new tabDownload slide

Introduction

The evaluation of symptomatic individuals suspected of coronary artery disease (CAD) is recommended by the European Society of Cardiology to include assessment of either coronary anatomical pathology, myocardial perfusion, or, if technically and logistically possible, both.¹ Myocardial computed tomography (CT) perfusion has emerged as a non-invasive method that in combination with coronary CT angiography (CCTA) can provide anatomical and functional evaluation using a single modality.^2,3 Static myocardial CT perfusion provides a single image of the relative myocardial distribution of contrast agent at one specific time point after intravenous injection. This approach requires precise timing of image acquisition, does not entail quantification of myocardial blood flow (MBF), and relies on subjective visual evaluation.³ Interestingly, the recently introduced, dynamic myocardial CT perfusion (DM-CTP) method does not seem to have these limitations; multiple images are acquired during the first pass of a contrast agent, timing is therefore less critical, quantification of absolute MBF can be achieved, and data interpretation can be largely automated.^4,5

DM-CTP appears to have incremental prognostic value over CCTA and allows for improved risk stratification of patients with CCTA-verified CAD.^6,7 These observations have to some extent paved the way for clinical implementation of DM-CTP as an identifier of stenoses with functional significance in patients suspected of CAD.^8,9 Nevertheless, to define abnormal findings by any imaging modality, values of normal MBF in healthy individuals are crucial. To this end, studies using DM-CTP have not reported data on individuals without any CAD.^10,11

The aim of this work was to evaluate the feasibility of DM-CTP in a cohort of healthy individuals with normal CCTA, to establish normal values for MBF, to assess the effects of age and sex, and to explore regional myocardial tissue perfusion and its possible relation to coronary dominance.

Methods

Study population

This study was a prospective, single-centre observational study performed from October 2020 to November 2021. The study was conducted according to the Declaration of Helsinki II and was approved by The Danish Committee on Health Research Ethics (protocol number H-19043153). Oral and written consent was obtained from all participants. Participants in the study were recruited from the ongoing Copenhagen General Population CT study in which women and men of Danish descent are drawn randomly from the Copenhagen Central Person Register and invited to undergo a research protocol consisting of a non-contrast and a contrast cardiac CT.¹² Individuals with a coronary artery calcium score (Agatston score) of zero, a normal CCTA, normal echocardiography, and age > 45 years were eligible for inclusion in the study. Exclusion criteria included (i) body mass index (BMI) > 35, (ii) diabetes, (iii) hypertension, (iv) hypercholesterolaemia, (v) current or prior moderate or heavy smoking (>20 pack years), (vi) inability to receive study information, (vii) bronchospastic lung disease [chronic obstructive pulmonary disease (COPD) or asthma], and (viii) pregnancy or active breastfeeding.

A total of 100 participants were planned enrolled for two types of study procedures. One procedure involved DM-CTP examination at rest and during adenosine stress (rest and stress DM-CTP group, n = 40), and another encompassed DM-CTP examination during stress (stress-only DM-CTP group, n = 60).

DM-CTP image acquisition

Image acquisition was performed using a 320-detector row CT scanner (Aquilion ONE/PRISM Edition, Canon Medical, Otawara, Japan).¹³ Participants were instructed to strictly avoid caffeine 16 h prior to examination. Two intravenous lines were inserted, and 12-lead electrocardiogram (ECG) was obtained, to rule out conduction abnormalities. Oral metoprolol (50–100 mg according to weight and systolic blood pressure) was administered to participants with a pre-examination heart rate above 65 b.p.m. This was done to establish haemodynamic conditions similar to that of a post-CCTA clinical scanning session. An individualized contrast volume based on estimated body water was injected at a flow rate of 6 mL/s.

After a delay of 6 s, participants were informed to hold their breath and low-dose prospective acquisitions were performed at every second heartbeat for 30 consecutive seconds with a z-axis covering the entire myocardium.¹⁴ A slow exhale and subsequent shallow breathing was recommended if the participant was unable to sustain breath hold for the full duration of the scan. Tube voltage was 80 kV and tube current was fixed at 300 mA. The scan range was 120 mm, the gantry rotation time was 275 ms, and detector collimation was 0.5 mm × 320. DM-CTP was performed as an electrocardiography-triggered prospective target scan acquired at 77–87% of the RR interval. Heart rates above 70 resulted in a scan time of 25 s to reduce radiation dose. A conversion factor of 0.019 mSv/(mGy × cm) was used to determine effective dose from dose–length product.

Intravenous adenosine (0.14 mg/kg/min) was used to induce myocardial hyperaemia, and the stress image acquisitions were initiated 3 min after starting the infusion pump. Blood pressure and heart rate was recorded before stress and 2 min after initiating the infusion of adenosine (during stress). In addition, the heart rate was recorded from the CT scan ECG gating summary. Discomfort experienced during adenosine stress along with ability to sustain breath hold for the full scan duration was recorded. Discomfort was categorized according to the following symptoms: chest pain, dyspnoea, blushing, and headache.

In the rest and stress DM-CTP group, the stress scan was preceded by a DM-CTP examination at rest performed according to the same scan protocol as the stress. A wait of 15 min between the first and the second injection of contrast agent was ensured.

DM-CTP image evaluation

For the evaluation of rest and stress DM-CTP, a low-motion image phase was determined by an automatic raw data analysis tool (phase eXact, Canon Medical Systems Corporation). Images were reconstructed at 1 mm slice thickness using iterative dose reduction [Adaptive Iterative Dose Reduction (AIDR3D)] and a kernel (FC03) with beam hardening correction. Before evaluation, all acquisitions were corrected for breathing-related displacement using a non-rigid body registration algorithm. Also, a 4D noise reduction filter (4D similarity filter) was applied. The 4D similarity filter reduces noise in dynamic CT imaging by averaging voxels with the highest probability of similar dynamic behaviour while keeping the spatial resolution.

Interpretation of the image data and calculation of MBF was performed on a workstation using dedicated software (Vitrea research, version 7.11, Canon Medical Informatics, Minnetonka, MN). The semi-automated software set the axis of the left ventricle (LV), identified the contours of the myocardium, and placed a region of interest in the ascending aorta used to derive the arterial input function.¹³

The LV myocardial tissue was subdivided into 16 segments (excluding the apical cap) according to the American Heart Association myocardial segmentation model using conventional definitions of long and short axis.¹⁵

Myocardial attenuation in the 16 segments and reference attenuation in the ascending aorta were plotted against time after subtraction of the baseline attenuation. Finally, MBF were computed based on the Renkin–Crone model and the Kety and Schmidt model.¹⁶ The Renkin–Crone model describes that the extraction of tracer from the blood to the tissue is a function of blood flow, is dependent on properties of the capillary membrane, and assumes that the tracer is diffusible. The single-tissue (two compartments) Kety and Schmidt model is based on the Fick principle according to which the rate of uptake of a tracer is equal to the difference of the arterio-venous concentration.

Definitions of absolute and relative MBF calculated on the global and segmental level are given in Table 1. Rest MBF and myocardial flow reserve results were based on measurements in the rest and stress DM-CTP group, and stress MBF results were reported as pooled stress results from both groups (rest and stress DM-CTP and stress-only DM-CTP). Since this study included only healthy individuals without any CAD, the mean of MBF in all 16 segments (global MBF) was used as reference for calculation of relative segmental MBF.

A reference range for normal relative MBF on the segmental level was defined as the relative segmental MBF minus two standard deviations (SDs). The use of a normal database is an established technique for the comparison of patient scans against reference limits using single-photon emission CT and positron emission tomography (PET) myocardial perfusion imaging.¹⁷ In addition, the use of two SDs was adapted from the general use of reference intervals where the normal range is defined as falling within two SDs of the mean, including 97.5% of observations when a one-sided cut-off is implemented.¹⁸

Statistical analysis

Using descriptive statistics, the continuous variables are expressed as mean ± SD. The Mann–Whitney U test (Wilcoxon rank-sum test) was used for comparison of non-parametric data, and Student’s t-test was used to assess differences in parametric data across groups. Categorical data are expressed as frequency (percentage), and the $χ2$ test was used to assess differences in proportion. The relationship between continuous variables was assessed using Pearson’s correlation coefficient. To compare MBF on the segmental level, a one-way analysis of variance (ANOVA) followed by Tukey’s honestly significant difference post hoc test was performed. The influence of sex, age, and coronary dominance on the segmental heterogeneity was examined using a two-way ANOVA. A binary age variable was constructed using the median age of all eligible participants. P-values of <0.05 were considered significant. All statistical analyses were performed using R statistical package, version 4.2.2 (R Foundation for Statistical Computing).

Results

Due to the COVID-19 pandemic, the planned enrolment of 100 participants was hindered, and a total of 91 healthy individuals were enrolled and underwent either both rest and stress DM-CTP (n = 34) or stress-only DM-CTP (n = 57). Of the consecutive 91 participants examined, 82 were eligible for analysis. Nine participants were excluded due to structural heart disease discovered after inclusion (n = 3), technical failure (n = 4), contrast agent extravasation (n = 1), and discomfort (severe dizziness) during adenosine (n = 1) (Figure 1).

Of the 82 participants, all were Caucasian, 72 had sedentary work, and 73 were moderately physically active (2 or 3 on a 4-point physical activity scale) in their leisure time. There were no significant differences in sex, age, BMI, prior smoking, or coronary dominance as defined by CCTA between the rest and stress DM-CTP and stress-only DM-CTP groups (Table 2). Twenty-six of 82 participants (32%) received metoprolol prior to DM-CTP. The effective radiation dose for rest and stress examinations was 4.16 ± 0.73 and 4.22 ± 0.56 mSv, respectively.

Symptoms and haemodynamics during DM-CTP adenosine stress

Sixty-seven participants (82%) were able to hold their breath for the duration of the stress scan. Frequencies of different types of discomfort experienced during infusion of adenosine were as follows: chest pain (72%), dyspnoea (43%), blushing (64%), and headache (20%). Heart rate before stress increased significantly during stress (58 ± 8 vs. 75 ± 11 b.p.m.; P < 0.01) (Table 3).

Global MBF

In the Rest and stress DM-CTP group, global MBF rose significantly from 0.93 ± 0.42 mL/min/g at rest to 3.53 ± 1.10 mL/min/g during stress (P < 0.01), resulting in a flow reserve of 4.22 ± 1.86 (Table 4). Global stress MBF results were similar in the rest and stress DM-CTP and stress-only DM-CTP groups (P = 0.78).

No correlation was found between global MBF and age at neither rest nor stress (P = 0.08 and P = 0.82), and no difference in global MBF between men and women was observed at rest (0.95 ± 0.42 and 0.92 ± 0.43 mL/min/g, P = 0.88) or during stress (3.50 ± 1.08 and 3.63 ± 1.20 mL/min/g, P = 0.61). Administration of oral beta-blockers did not influence global MBF during rest (P = 0.62) or stress (P = 0.20) and neither did the ability to sustain breath hold for the full duration of the adenosine scan (P = 0.25).

A significant correlation between heart rate and global MBF was observed during both rest and stress (rest, P = 0.036; and stress, P < 0.01).

Segmental MBF

Segmental MBF is given in Table 5. Significant intersegmental differences were observed with the lowest MBF noted in Segment 16 (apical–lateral) and the highest in Segment 2 (basal–septal) both at rest and during stress. Segmental MBF was significantly different between the segments at both rest and during stress (P < 0.01), and the heterogeneity was not related to age, sex, or coronary dominance (P ≈ 1 for all three factors for both rest and stress condition).

Relative segmental MBF at rest and during stress and cut-offs for normal relative flow are presented graphically in Figure 2 and in polar maps in Figure 3. Relative segmental MBF during stress varied between 85% of global MBF in Segment 16 and 113% in Segment 2 (difference of 28%).

Implementation of normal values for clinical evaluation of CAD

In the  Appendix, a method for the implementation of the normal values obtained in this study for clinical evaluation of symptomatic patients suspected of haemodynamically significant CAD using DM-CTP is presented.

Discussion

This is the first larger study to investigate DM-CTP in healthy, asymptomatic individuals without cardiovascular risk factors and with CCTA-verified absence of any CAD, providing several important findings. First, only one stress examination was aborted due to patient discomfort. DM-CTP therefore appears feasible and tolerable and can be implemented as stress-only protocol on top of CCTA with an added radiation dose of ∼4 mSv. Second, MBF measured with DM-CTP in healthy individuals aged 45–78 years (median age = 55) was not related to age or sex, and finally, MBF displayed regional heterogeneity, and therefore, segmental values of normal MBF appears to be needed to define normally perfused myocardium.

Influence of age and sex on MBF

No correlation between MBF and age was found at neither rest nor stress, and no differences were seen at rest or during stress between men and women.

The literature does not seem to support age-related alterations of MBF during stress in healthy individuals, at least not before 70 years of age.^19,20 However, it seems that there is a consensus regarding an increasing resting blood flow at older age, which could to some extent be attributed to increased resting myocardial work and increased blood pressure.^19,20 This probable age-related difference at rest was not detected in the present study, which might be due to age span of the participants (eight individuals of age 70 years or more).

Regarding the influence of sex on MBF, mixed results have previously been reported using other modalities. In a study of 1683 patients with signs/symptoms of ischaemia and angiographically unobstructed coronary arteries, a slightly higher resting and a slightly lower hyperaemic coronary blood flow was reported in women.²¹ Danad et al.²² reported that hyperaemic MBF measured with PET were higher in women compared with men but found no sex-dependent difference in baseline MBF after correction for rate–pressure product. Finally, Chareonthaitawee et al.²³ found that baseline MBF in women was significantly higher compared with men but this difference was not present for hyperaemic flow. In our work using DM-CTP, we did not find any differences of MBF between men and women, which appears to add to the notion that sex is not a major determinant of MBF.

Segmental heterogeneity of normal myocardial tissue

In the present study, relative segmental MBF varied from 85% to 113% compared with global MBF. These findings are in line with results by Kühl et al.,²⁴ who examined patterns of MBF using static perfusion CT and ¹³N-ammonia PET. Using both methods, they found that segments with a location in the anteroseptal wall had the highest and segments in the lateral wall had the lowest relative MBF. The range reported by Kühl et al. (25%) is also comparable with the range found in the present study (28%). Heterogeneous MBF was also reported in two ¹³N-ammonia PET studies by Porenta et al.²⁵ and Kofoed et al.²⁶ In line with the results of the current study, they both reported the highest relative MBF in the septal part of the myocardium.

Since this heterogeneity has been reported using both CT and PET, it seems that the phenomenon to some degree can be attributed to physiology. However, common methodology such as the inability of implemented models to fully account for the temporal dispersion of MBF in the baso-apical direction [the highest MBF was noted in Segment 2 (basal) and the lowest in Segment 16 (apical)] might also play a role.²⁷

Interindividual variability

In this study, pronounced interindividual variation in global MBF was seen. At least three possible explanations could be the source of this variation. First, CT as image modality and the method used for calculation of MBF might inflate small or non-existing physiological differences; however, Danad et al. performed a study of 330 patients using what is currently considered the gold standard for measurement of MBF, ¹⁵O-H₂O PET. They reported a range of hyperaemic blood flows in individuals without significant CAD (1.00 ± 0.25 mL/min/g at rest and 3.26 ± 1.04 mL/min/g during stress) which is coherent with our results, and the interindividual variability does therefore not seem to be a CT-specific phenomenon.²⁸

Secondly, a variable response to adenosine specific for each individual might cause variation in absolute MBF. Regarding a variable response to adenosine, we ensured that none of the participants had any intake of caffeine 16 h prior to the stress examination. Importantly, we found a significant correlation between heart rate and MBF during adenosine, suggesting that our MBF results reflect the haemodynamic condition of the individual.

Finally, large physiologic interindividual differences in hyperaemic MBF may exist in healthy individuals. It would be intriguing to examine this result further using the microsphere method which has been employed in a pig model by Christian et al.²⁹ The study found that CT-based myocardial perfusion imaging offered comparable results with those obtained with the gold standard microsphere method which is not feasible in humans. However, results using ¹⁵O-H₂O PET produce the same range of values, which is an indication that substantial physiologic variability exists among healthy individuals. ¹⁵O-H₂O PET is considered the gold standard for human MBF evaluation since the tracer diffuses freely and the uptake is therefore directly proportional to flow.²⁸ Overall, none of the three possible explanations can be isolated and all three might contribute to the currently observed interindividual variability.

Limitations

The primary limitation of this study is the single-centre, single-scanner, and single-perfusion CT software setup. Even though our results are in line with previous observations obtained with other modalities, specific models included in the choice of perfusion CT software might affect the degree of heterogeneity. Therefore, normal values established across vendors and kinetic models would be advantageous.

Finally, it is important to note that no consensus regarding the size of the possible effects of age and sex and MBF exist. Therefore, the study was not explicitly powered to detect these differences.

Clinical perspectives

It remains challenging to define a clinically reliable cut-off value using absolute measures to differentiate normal from abnormal MBF measured with DM-CTP.³⁰ Importantly, in recent larger DM-CTP accuracy studies, relative measures of MBF were used to discriminate normal from impaired MBF.^8,9

Furthermore, this study describes the variability and heterogeneity of normal MBF. Interpretation of DM-CTP in the context of a CCTA will therefore most likely improve the clinical information derived from the perfusion scan when utilized to determine whether a lesion results in ischaemia in the region supplied by a particular vessel suspected of obstructive CAD. Overlay of CCTA on the MBF polar map can facilitate a more vessel-specific interpretation as presented in the  Appendix. This approach is utilized in the ongoing DYNAMITE trial (NCT04709900).

Finally, despite international multicentre studies starting to appear, the DM-CTP method is mainly implemented for clinical use at sites with research activity within the field.⁹ The authors hope that the results of this study could aid in the dissemination of the method which is hampered by the lack of reference values in addition to the associated radiation dose and the technical difficulty of the scan protocol and image analysis requiring dedicated software.

Conclusions

Absolute and relative values of MBF at rest and during adenosine stress measured with dynamic myocardial CT perfusion have been defined in a cohort with CCTA-verified absence of coronary artery disease. There were no differences between the sexes and no correlation between MBF and age. MBF was heterogeneous among the 16 myocardial segments, and the heterogeneity was not related to age, sex, or coronary dominance. Normal values of regional MBF could be helpful in the clinical evaluation of CAD using dynamic myocardial CT perfusion.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.

Appendix

Implementation of normal values for clinical evaluation of coronary artery disease

References

Author notes