Coronary Artery and Sinus Velocity and Flow

Coronary Sinus Flow

Velocity mapping of coronary venous outflow is less problematic than velocity mapping in the coronary arteries because the coronary sinus has a much larger diameter (typically 7–10 mm), and also because effects of signal loss are unlikely because flow is less susceptible to turbulence. In the human heart, coronary sinus flow almost equals total coronary flow because approximately 96% of left ventricular (LV) venous blood flow drains into the right atrium via the coronary sinus. van Rossum and colleagues were the first to show that blood flow in the coronary sinus could be measured using CMR. This feasibility study assessed the ability of cine CMR velocity mapping to measure phasic and mean coronary venous outflow in the distal coronary sinus of 24 healthy subjects. The flow profiles were generally biphasic and primarily diastolic, with 37% of subjects showing some reverse flow immediately after the R-wave. The mean volumetric flow over the cardiac cycle was 144 ± 62 mL/min, similar to values reported in normal subjects using continuous thermodilution (122 mL/min). Although phasic blood flow in the sinus may have been expected to be predominantly systolic, as in other venous structures, the authors suggested that the thin, compliant walls of the sinus, together with its drainage into the right atrium, where pressure varies considerably, may be responsible for the predominantly diastolic flow profile. These findings have been supported by other independent studies, both directly, using an ultrasonic transit time technique to measure phasic volumetric coronary sinus flow in conscious dogs, and indirectly, by observing areas of signal void, caused by accelerating and turbulent flow near the entrance of the coronary sinus in the right atrium in early diastole, in conventional cine CMR in healthy subjects.

Fig. 25.1 shows the typical image plane orientation for coronary sinus flow measurements together with a typical coronary sinus flow curve.

The imaging time for the feasibility study was typically 4 minutes, and this may be inappropriate when assessing sinus flow under pharmacologically induced maximal vasodilation. Kawada and associates assessed the possibility of using a segmented k -space gradient echo CMR approach to the acquisition so that all data could be acquired in <25 seconds. This had the further advantage that the full acquisition could be performed in a single breath-hold, thereby eliminating the effects of respiratory motion. In a study of 9 healthy subjects and 29 patients with hypertrophic cardiomyopathy (HCM), data were acquired from an oblique coronal plane. Four reference and four velocity encoded views were acquired per data segment, giving a segment duration of 120 ms, but the temporal resolution was effectively improved by view sharing, a technique whereby data are generated at intermediate time points from the preceding and following data segments. Hence, depending on the RR interval, velocity maps were obtained at up to 14 phases in the cardiac cycle. The authors observed the same biphasic velocity and flow profiles in both healthy subjects and patients with HCM, with no significant difference noted between the baseline myocardial blood flow (coronary blood flow per unit mass of myocardium) in the two groups (0.74 mL/min per gram vs. 0.62 mL/min per gram). However, after intravenous administration of 0.56 mg/kg dipyridamole, the increase in myocardial blood flow in patients with HCM (0.62 mL/min per gram to 1.03 mL/min per gram) was less than that in the healthy subjects (0.74 to 2.14 mL/min per gram), resulting in significantly different coronary flow reserves for the two groups (1.72 ± 0.49 vs. 3.01 ± 0.75, respectively, P < .01). The authors concluded that, in healthy subjects, myocardial blood flow and coronary flow reserve, as measured by breath-hold phase velocity mapping, were similar to those found by other techniques; in addition, the CMR technique was able to distinguish between healthy subjects and patients with HCM. A more recent study in young HCM patients (22.3 ± 6.4 years) and age-matched subjects at risk of HCM has similarly shown that HCM patients (but not subjects at risk of HCM) have reduced myocardial blood flow response to adenosine-induced hyperemia, even in the absence of diastolic dysfunction or left ventricular outflow tract obstruction. The breath-hold approach has also been used by Kennedy and colleagues to show significant differences in the coronary flow reserve of healthy subjects (4.59 ± 0.58) and heart transplant patients with mild (2.15 ± 0.44, P < .05) and severe (2.21 ± 0.59, P < .05) coronary disease, as determined by posttransplant coronary angiography. It has also been used to show reduced coronary flow reserve in patients with high serum eicosapentaenoic acid (EPA) compared with those with low EPA, and in those with heart failure with preserved ejection fraction compared with those with hypertensive left ventricular hypertrophy and controls. At 3 T, the increased signal-to-noise ratio (SNR) available generally allows imaging of coronary sinus flow with higher spatial resolution or, particularly when used in combination with parallel imaging techniques, reduced breath-hold duration. Imaging at 3 T has been used to determine the feasibility of flow reserve quantification using regadenoson, a relatively new selective adenosine A2A receptor agonist and to investigate the role of aminophylline reversal. It has also been used to show reduced coronary flow reserve in patients with surgically repaired tetralogy of Fallot, compared with healthy volunteers (1.19 ± 0.34 vs. 2.00 ± 0.43, P = .002) and in two similar studies (one at 1.5 T and one at 3 T) to show that, whereas smokers and nonsmokers have similar baseline myocardial blood flows, smokers have a significantly reduced response to cold pressor testing ( Fig. 25.2 ). A spiral k -space coverage phase velocity mapping sequence at 3 T has allowed the acquisition of data with a spatial resolution of 0.8 mm × 0.8 mm in an 11 to 15 s breath-hold with a temporal resolution of 60 to 69 ms and has been used to show myocardial blood flow increases in response to cold pressor testing in asymptomatic women. A validation of the breath-hold approach was reported by Koskenvuo and associates, who compared myocardial blood flow measured by CMR with that measured by PET in both healthy subjects and patients with coronary artery disease (CAD). Good correlations were reported for myocardial blood flow in both subject groups (0.82 and 0.80, respectively), whereas for coronary flow reserve, the correlations were 0.76 and 0.5, respectively.

One issue with the breath-hold approach to measuring coronary sinus flow is that breath-holding changes intrathoracic pressure and affects cardiac output and venous blood flow. Schwitter and coworkers performed a validation of nonbreath-hold velocity mapping against PET. Although taking longer to acquire (typically 4 minutes), the nonbreath-hold technique has the advantages of higher spatial resolution (0.8 × 0.8 mm) and improved SNR through the acquisition of multiple averages. Furthermore, the high temporal resolution that is achievable is better suited for resolving the highly pulsatile flow profile and for minimizing the blurring effects of the extensive in-plane motion of the coronary sinus during the cardiac cycle. Unlike the initial study of van Rossum and coworkers, this free breathing study used retrospective electrocardiogram (ECG) gating, which enables data acquisition throughout the entire cardiac cycle, and respiratory ordered phase encoding, which reduces the effects of respiratory motion. Scans were performed both before and after the administration of 0.56 mg/kg dipyridamole in 16 healthy subjects and 10 orthotopic heart transplant recipients and the results were compared with PET data. The mean difference between coronary flow reserve measured by PET and CMR was 2.2%, with limits of agreement of −27.2% and 31.6%. Correlation between CMR-measured myocardial blood flow (coronary sinus flow divided by myocardial mass) and PET was good ( r = 0.93), although the slope was considerably less than unity (0.73). This underestimation of blood flow measured by CMR results from the fact that a variable part of the inferior and inferior-septal myocardium is drained by the middle cardiac vein, which either enters the coronary sinus just before its orifice or empties directly into the right atrium. If coronary sinus flow is divided instead by the mass of drained myocardium (as measured from a stack of short axis images), the correlation remains good ( r = 0.95), whereas the slope approaches unity (1.05). Fig. 25.3 shows a regression plot of myocardial blood flow (sinus blood flow per unit mass of drained myocardium) measured by CMR and PET, with baseline flows normalized for rate-pressure product, together with a Bland-Altman plot. The mean differences between PET and CMR were 3.4% ± 12.8% at resting baseline and 3.9% ± 20.2% during hyperemia and were not significant. The authors have subsequently used this technique to show that administration of 17β-estradiol over 3 months without progestin coadministration does not improve coronary flow reserve in postmenopausal women. A similar nonbreath-hold technique was used by Moro and colleagues at 3 T to show that women have a higher increase in myocardial blood flow in response to cold pressor testing than men (0.73 ± 0.43 mL/g per minute vs. 0.22 ± 0.19 mL/g per minute, P = .0012).

A validation of nonbreath-hold CMR-measured coronary sinus flow has also been performed against flow probes in dogs by Lund and associates. In this study, the correlation between coronary blood flow (measured as the sum of left anterior descending [LAD] and circumflex [LCX] coronary flow) with flow probes against coronary sinus flow measured with CMR was 0.98, with a nonsignificant mean difference of 3.1 ± 8.5 mL/min. Coronary sinus blood flow per unit mass of myocardium was 0.40 ± 0.09 mL/min per gram (CMR) compared with 0.44 ± 0.08 mL/min per gram (flow probes) and also was not significant. The authors went on to study patients with chronic heart failure and showed that coronary flow reserve was significantly reduced compared with that in healthy subjects (2.3 ± 0.9 vs. 4.2 ± 1.5, P = .01). This has been recently confirmed by Aras and colleagues (coronary flow reserve 1.45 vs. 3 in patients with chronic heart failure and healthy subjects, respectively, P < .001). Similarly, Watzinger and coworkers have shown significantly reduced flow reserve in patients with idiopathic cardiomyopathy compared with healthy subjects (2.19 ± 0.77 vs. 3.51 ± 1.29, P < .05). In all three studies, there were no significant differences in baseline flow between healthy subjects and those with disease (0.52 mL/min per gram vs. 0.46 mL/min per gram, P = not significant [NS]; 0.83 ± 0.26 mL/min per gram vs. 0.85 ± 0.30 mL/min per gram, P = NS; and 0.55 ± 0.19 mL/min per gram vs. 0.48 ± 0.07 mL/min per gram, P = NS).

These studies provide useful results, but have a number of important limitations. Cardiac and respiratory motion results in blurring of the sinus, and the small number of pixels covering the sinus (typically five across the diameter in diastole) results in considerable partial volume averaging in edge pixels, which is problematic for the accurate assessment of sinus cross-sectional area and for the determination of mean sinus flow velocity. As discussed, drainage of a variable part of the inferior and inferior-septal myocardium by the middle cardiac vein results in underestimates of myocardial blood flow measured, and although this estimate is an indicator of total coronary blood flow, no regional assessment of either flow or flow reserve is possible with this technique.

Total Coronary Flow Reserve From Measurements in the Aortic Root

In 1993, it was proposed that coronary flow reserve might be derived from flow measurements made in the ascending aorta, which is less susceptible to cardiac and respiratory motion and partial volume effects than the coronary arteries. Coronary diastolic flow, which represents the bulk of coronary flow, can be estimated as the retrograde flow in the ascending aorta during systole and diastole minus the antegrade flow during diastole. A variable velocity encoding window was implemented to maintain the accuracy of velocity measurements during periods of both high flow in systole (window = 200 cm/s) and low flow in diastole (window = 30 cm/s). Although it was suggested that the assessment of absolute diastolic coronary flow with this technique is inaccurate because of known errors associated with the velocity mapping technique, it was argued that these errors should be the same both pre-vasodilation and postvasodilation and therefore should be eliminated from the assessment of diastolic coronary flow reserve, defined in this instance as the difference (rather than the ratio) between the two measurements. In seven patients with abnormal findings on myocardial perfusion scintigraphy, the diastolic coronary flow reserve was −50 ± 76 mL/min compared with 260 ± 66 mL/min in eight healthy subjects. The principle of this technique was later refined by taking into account the motion of the coronary arteries during the cardiac cycle. A model was developed describing the flow through five transverse parallel aortic slices extending from the base of the aortic valve to above the level of the coronary ostia. This was then solved mathematically, and in five healthy subjects, it was shown that the standard error in the measurement of total coronary artery flow was approximately 90 mL/min, or 30% of total coronary artery flow. The error in coronary flow reserve is higher because the errors in baseline and maximal vasodilation flow are additive. In addition, the assumption that flow is predominantly diastolic, although true in healthy subjects, may not hold in the presence of disease, introducing further unknown errors into the technique. A more recent attempt at measuring coronary artery blood flow in this way likewise found that the technique was compromised by poor reproducibility, although significant changes in coronary blood flow with hormone replacement therapy were observed.

Direct Assessment of Coronary Artery Velocity

The problems associated with the assessment of flow velocity and coronary artery flow were discussed earlier and preclude the application of standard CMR techniques. As for coronary imaging, the major breakthrough for coronary flow techniques was the shortening of sequence duration to the extent that data could be acquired over a single breath-hold, effectively freezing respiratory motion and eliminating the resulting artifact. Further advances with navigator echo monitoring of the diaphragm position have allowed data to be acquired over multiple reproducible breath-holds or during free breathing, enabling increased spatial and temporal resolution and data averaging (resulting in higher SNRs). The following sections describe the approaches used to assess coronary flow velocity and flow.