Value of Magnetic Resonance Imaging in Assessing Patency and Function of Coronary Artery Bypass Grafts
An Angiographically Controlled Study
Background Previous studies have demonstrated the high sensitivity and moderate specificity of standard magnetic resonance (MR) spin-echo (SE) and gradient-echo (GE) techniques in predicting the patency of coronary artery bypass grafts. These techniques, however, do not provide quantitative information. Therefore, the objectives of this study were first to investigate whether MR cine GE images, performed in addition to standard SE images, have additional value for the assessment of graft patency and second to assess the graft function by measuring the flow pattern and flow rate with MR phase velocity imaging.
Methods and Results Forty-seven patients with previous histories of coronary artery bypass grafting underwent angiography and MR SE and cine GE phase velocity imaging. SE and GE images were evaluated by three independent observers blinded to the angiographic results. The spatial mean velocity and volume flow were measured and repeated for each image at consecutive 50-millisecond intervals throughout the cardiac cycle. The 47 patients had 98 proximal aortotomies, of which 60 were single and 38 sequential grafts. Seventy-three grafts were patent; 25 were occluded. Eighty-four grafts (86%) were eligible for comparison of the results of SE and GE images. Assessment of patency was inconclusive on SE images in 7 grafts (5 occluded by angiography) and on GE images in 7 grafts (2 occluded). A comparison of the results of contrast angiography and SE and GE MR imaging techniques showed that both techniques had a high sensitivity (both 98%) and somewhat lower specificity (85% and 88%, respectively) for graft patency. Combined analysis of the SE and GE images did not improve the accuracy. The strength of the interobserver agreement on GE images was good (κ=0.66), whereas on SE images the agreement was moderate (κ=0.51). Adequate MR phase velocity profiles were obtained in 62 (85%) of the 73 angiographically patent grafts. Graft flow was characterized by a balanced biphasic forward flow pattern. The volume flow of sequential grafts to 3 regions (136±106 mL/min) was significantly higher than in single grafts (63±41 mL/min, P<.01).
Conclusions Considering the good interobserver agreement and the 85% success rate of quantitative flow measurements, cine GE phase velocity mapping is a promising clinical tool in the noninvasive assessment of graft patency and function.
Previous studies have demonstrated the high sensitivity and moderate specificity of conventional MR SE and GE techniques in the evaluation of patency of coronary artery bypass grafts.1 2 3 4 5 6 7 On SE images, flowing blood is depicted as a signal void. However, calcifications, metal clips, thickened pericardium, and small pericardial collections of fluid can mimic the signal void of flowing blood. In contrast to the signal void on SE images, flowing blood on GE images is depicted as a bright signal. Therefore, it has been suggested that the specificity of MR imaging for graft patency will be improved by this technique.5 6 Another point that has implications for the clinical value of MR graft imaging is the interindividual variability in the assessment of graft patency by these techniques.
Although SE and GE images give important, noninvasive, qualitative information regarding graft flow, quantitative information is not obtained. Therefore, optimal MR imaging of coronary artery bypass grafts may consist of a standard SE examination, followed by cine GE MR imaging. The latter would be performed perpendicular to the graft at selected levels that best display the graft location and allow measurement of flow by means of phase velocity mapping. The value of such a combined approach has not yet been studied. Only one limited report has been published that describes the feasibility of flow measurements in four saphenous vein coronary bypass grafts with the phase velocity mapping technique.8
Hence, the aims of the study were first to investigate whether MR cine GE images, performed in addition to standard SE images, have additional value for the assessment of graft patency and second to assess the graft function by measuring the flow pattern and flow rate with MR phase velocity imaging.
Forty-seven patients (mean age, 66±4 years; 38 men) with previous histories of coronary artery saphenous vein grafting (mean, 4.4±3.9 years before catheterization) underwent angiography and MR SE and cine GE imaging to assess graft patency (mean time interval between angiography and MR imaging, 2.2±4.9 months). Flow pattern and flow rate were measured on simultaneously acquired MR cine GE phase velocity maps. None of the patients showed any change in clinical symptoms between catheterization and MR imaging. Forty patients underwent catheterization because of angina class II or III according to New York Heart Association criteria; 7 patients, as part of a follow-up study.
The number of proximal aortotomies and the sites of distal anastomoses were known for all patients from the surgical operative reports at the time of cardiac catheterization and MR imaging. Because of the clip artifact they induce on MR images, internal mammary artery grafts (n=11) were excluded from the study. All patients were in sinus rhythm.
Selective angiography was performed by the Judkins technique with standard catheters. A nonionic contrast agent (10 mL) was injected selectively over a period of 3 to 5 seconds. All bypass grafts or stumps were visualized in at least two projections.
All patients underwent ECG-gated MR SE and cine GE phase velocity imaging to assess graft patency. We previously reported on the application and validation of phase velocity mapping in our 0.6-T system (Teslacon II, General Electric/CGR).9 10 In brief, the technique enables one to obtain simultaneously standard MR cine GE images for visualization of anatomy and corresponding velocity maps for calculation of velocity at multiple frames throughout the cardiac cycle.
The patients were placed in prone positions on a surface coil. The grafts were identified on sagittal and transverse multilevel SE scout views. Fig 1⇓ illustrates the typical course of saphenous vein coronary bypass grafts with respect to the imaging planes.
Figs 2⇓ and 3⇓ are representative examples of a sagittal and transverse SE image perpendicular to grafts anastomosing with the LCx, LAD, and RCA. SE imaging parameters were as follows: time of repetition was equal to the length of the cardiac cycle (RR interval), time to echo was 28 milliseconds, the average of two radiofrequency excitations was taken, slice thickness was 5 mm, interslice gap was 0.5 mm, acquisition matrix was 160×256 interpolated to a display matrix of 256×256, the field of view was 30×30 cm2, and spatial resolution was 1.9×1.2×5 mm3. These images were used to acquire cine GE images and phase-encoded velocity maps in an orientation perpendicular to the proximal course of the grafts. Cine GE imaging parameters were as follows: time of repetition was 50 milliseconds, time to echo was 24 milliseconds, flip angle was 25°, and the other parameters were as in SE images. Velocity was phase encoded in the direction of the slice-selective gradient, thus measuring the component of flow perpendicular to the imaging plane. Sensitivity to velocity was adjusted to measure velocities up to 40 cm/s without aliasing.
Phase velocity imaging of grafts to the RCA was performed in a transverse plane at the level of the four-chamber view (Fig 4⇓). Velocity imaging of grafts to the left coronary artery was performed in a sagittal plane at the level of the pulmonary artery trunk; in some instances, the imaging plane was slightly rotated on the longitudinal axis (Fig 5⇓).
Angiograms were used as the gold standard for graft patency. The patency of the grafts was visually evaluated by an experienced observer with the cine angiograms. When the proximal segment of the graft, from the origin of the graft in the ascending aorta up to and including the first distal anastomosis, was filled by contrast, it was considered patent.
MR SE and cine GE images were evaluated from the hard copies by three independent observers to test interobserver agreement. They were blinded to the angiographic results but informed about the number of proximal anastomoses and the sites of distal anastomoses. Finally, differences in the interpretation of graft patency between the observers were solved by consensus. A graft was defined as patent on SE images if a signal-free lumen was seen on at least two contiguous images at a position consistent with the expected location for that graft. If an intermediate signal intensity was seen in the lumen on two contiguous SE images, which could result from slow flow or thrombus, this was noted separately and judged as inconclusive. On the cine GE anatomic images, a graft was defined as patent if it was seen as a bright blood flow signal on consecutive images throughout the cardiac cycle at the expected location. A graft was judged inconclusive if the investigator was not sure about its patency and if no consensus could be obtained between the three observers.
The proximal segments of the grafts were identified from the typical sites of the aortotomy (Fig 1⇑). The graft with the most cephalic origin from the aorta and coursing laterally from the distal main pulmonary artery or the left pulmonary artery was considered to anastomose with the LCx or marginal branches (LCx region). The next lower graft, with a more anterior course immediately leftward from the main pulmonary artery, was judged to anastomose with the LAD or diagonal branches (LAD region). Fig 2⇑ is a typical example of such a situation. The graft with the most caudal origin from the aorta coursing next to the right atrium or AV groove was considered to anastomose with the RCA or posterior descending artery (RCA region), as illustrated in Fig 3⇑.
For graft flow analysis, delineation of the graft had to be obtained on the consecutive cine GE images throughout the cardiac cycle. The MR data were transferred to a computer workstation. The anatomic image and the phase velocity image were displayed side by side. The graft cross section was outlined on a magnified anatomic image and copied to the corresponding velocity image. The spatial mean velocity was then measured. These measurements were repeated for each pair of images at consecutive 50-millisecond intervals throughout the cardiac cycle. Volume flow was calculated by integration of the average velocity multiplied by the cross-sectional area over the cardiac cycle. The data were displayed in curves of flow velocity and volume flow versus time.
To facilitate comparison of data and to allow construction of the mean saphenous vein graft flow pattern, we normalized each curve to a heart rate of 60 bpm. All systolic acquisition points were normalized to the systolic time interval of a heart rate of 60 bpm; the diastolic acquisition points, to the diastolic time interval.11
Values are expressed as mean±SD. Student’s t test was used for means of variables of the continuous type. Values of P<.05 were considered statistically significant. As a measure of interindividual agreement, κ was used. This value represents the proportion of potential agreement beyond that expected on the basis of chance.12
The 47 patients had 98 proximal aortotomies: 60 were single and 38 were sequential grafts. The first distal anastomosis was to the LAD region in 49, to the LCx region in 26, and to the RCA region in 23 grafts. All aortotomies were visualized; 73 grafts were patent and 25 were occluded. Therefore, the prevalence of graft occlusion was 26%.
MR Imaging and Patency
The imaging procedure required 45 to 60 minutes, depending on the heart rate and number of grafts that had to be visualized. Four patients (8 grafts) were unable to continue the imaging protocol because of the discomfort of lying in a prone position and were excluded from the study. Of the remaining 90 grafts, 6 were imaged on an incorrect plane by the cine GE technique. Therefore, 84 grafts were eligible for comparison of the results of SE and cine GE images.
Table 1⇓ compares the results of contrast angiography and MR imaging. On SE images, the assessment of graft patency by consensus was obtained in 77 of 84 grafts (92%). The sensitivity for graft patency on SE images was 98% (63 of 64), with a specificity of 85% (11 of 13) and a predictive accuracy of 96% (74 of 77). Of the 7 inconclusive grafts, 5 had an intermediate signal intensity and were shown to be occluded on angiography; 1 of the 2 grafts without intermediate signal intensity was occluded.
On GE images, 7 grafts were defined as inconclusive, of which 2 were found to be occluded and 5 to be patent on angiography. For the remaining 77 grafts (92%), the sensitivity for graft patency was 98% (59 of 60) and the specificity was 88% (15 of 17), with a predictive accuracy of 96% (74 of 77).
Table 1⇑ also presents the results of a combined approach in which the SE and GE images were evaluated together. Graft patency could be assessed in 81 of 84 grafts (96%). The sensitivity and specificity for graft patency of the combined approach were 98% and 76% (63 of 64 and 13 of 17), respectively. The predictive accuracy was 94% (76 of 81).
The interindividual agreement was determined according to the MR technique and imaging planes (Table 2⇓). The agreement on cine GE images was good (mean κ=0.66), whereas the agreement on SE images was moderate (mean κ=0.51). No major differences in agreement were found for the different imaging planes.
MR Velocity Imaging
Adequate MR phase velocity profiles were obtained in 62 of the 73 angiographically patent grafts (85%). Inadequate velocity images were obtained from 3 grafts. Of the remaining 8 grafts, 2 were considered to be occluded, and the imaging plane was not correct in 6. The success rate of measuring graft flow did not vary between single and sequential grafts or with the site of the first anastomosis.
Fig 6⇓ shows the flow patterns obtained by MR phase velocity imaging for single and sequential grafts. Graft flow was characterized by a balanced biphasic forward flow pattern, with one peak in early systole and one in early diastole.
Table 3⇓ shows the graft flow parameters. The volume flow of sequential grafts to three regions (136±106 mL/min) was significantly higher than in single grafts (63±41 mL/min, P<.01).
Patency of MRI
The present study demonstrates a high sensitivity and somewhat lower specificity of conventional SE (98% and 85%, respectively) and cine GE (98% and 88%, respectively) MR imaging in the prediction of coronary artery saphenous vein bypass graft patency. These results are in agreement with previously published studies.1 2 3 4 5 6 7
Remarkably, the specificity for the combined evaluation was lower than that for the SE and GE images separately. This was due to the fact that 2 grafts classified as inconclusive by each of the techniques separately were falsely considered patent on the combined approach. Moreover, of the 5 other grafts considered inconclusive on SE images, the patency was correctly assessed on the GE images: 1 as patent and 4 as occluded. In the combined analysis, however, only 1 of the 4 occluded grafts was correctly assessed. In the other 3 cases, 1 was assessed as patent and 2 as inconclusive.
From this one can derive that in the case of an inconclusive SE image, assessment of patency should be based on GE images only. The specificity would then have been 94% (16 of 17), with an unchanged sensitivity of 98% (63 of 64).
Furthermore, we can conclude from this study that grafts with intermediate signal intensity on SE images are occluded. If this knowledge were used to predict patency on SE images, the sensitivity would remain unaltered, but the specificity would increase to 89%.
The interobserver agreement for graft patency was good to very good for cine GE images and moderate for SE images. In this respect, cine GE imaging would be preferable to SE imaging for the assessment of graft patency because good reproducibility of an observation is essential for appropriate clinical decision making.
MR Velocity Imaging
In this study, quantitative information of graft patency was combined with functional information by MR phase velocity imaging. It has been demonstrated that phase velocity imaging is successful in obtaining flow profiles from patent grafts. An important contributory factor to the successful assessment of graft flow was probably the use of a surface coil. The spatial resolution of 1.9×1.2×5 mm3 permits accurate cross-sectional imaging of a bypass graft because the cross-sectional area of the graft is approximately 13-fold larger than the pixel size. Partial volume effects are reduced by the slice thickness of 5 mm. Also, the signal-to-noise ratio is improved for structures that are localized superficially, as is the case with proximal parts of bypass grafts. Moreover, the prone position on the surface coil is effective in reducing respiratory motion artifact.
We demonstrated a biphasic forward flow pattern with one peak in early systole and one peak in early diastole. No quantitative differences were found between the flow patterns of single and sequential grafts. A biphasic velocity pattern in coronary artery bypass grafts has also been reported by others using Doppler ultrasound.13 14 In contrast, native coronary flow occurs predominantly in diastole when aortic pressure exceeds left ventricular pressure. The systolic flow in grafts can be explained by passive capacitance of venous grafts during the cardiac cycle, which is likely to be greater than the passive capacitance of the native coronary artery bed. It will influence flow pattern more in the proximal than the distal part of the graft. During systole, the runoff from the distal part of the graft is hampered by the high resistance of the native vessel, whereas flow is allowed to occur in the proximal part because of the capacitance of the graft. This hypothesis is supported by others who demonstrated a higher systolic than diastolic velocity in the proximal part of coronary bypass grafts.14 In accordance with the hypothesis, the opposite pattern was demonstrated in the distal part of the graft with a higher diastolic than systolic velocity.13 Therefore, the biphasic balanced forward flow pattern in the middle part of the graft, demonstrated in this study, can be well understood.
The results of the present study are in accordance with the previously reported average flow rates for the coronary arteries as measured by a nitrous oxide desaturation technique (65 to 82 mL·min−1·100 g−1) and xenon-133 clearance of 43 to 84 mL·min−1·100 g−1.15 16 17 18 Recently, measurements by positron emission tomography with oxygen-15–labeled water revealed a basal myocardial blood flow of 1.13±0.26 mL·min−1·g−1 tissue.19 Maximum flow velocities in the present study are lower than those found by Doppler ultrasound studies (15 to 28 cm/s).12 13 This is due to the fact that in the present study the mean velocities were obtained as spatially averaged values over graft cross sections. In contrast, the Doppler ultrasound technique measures spatial peak velocity within the graft.13 14 As mentioned, flow velocities in these studies were measured at different anatomic levels of the grafts, which makes comparison even more difficult.
No differences were found in flow velocity parameters and volume flow according to the type of graft, single or sequential, or the number of coronary regions supplied, except for sequential grafts supplying three coronary artery regions. These grafts had a significantly increased volume flow, which can be explained by the fact that they have to supply a larger part of the myocardium.
Limitations and Considerations for Improvement
Internal mammary grafts cannot be adequately assessed because of the artifact generated by the metallic clips. A solution to this problem would be to use nonmetallic clips, which are already available.
GE images were performed only as a single-level multiphase cine loop, perpendicular to the plane in which the graft was expected on the basis of the SE images. We opted for this approach so that we could quantify the graft flow. This strategy provided more important additional information than graft patency alone. A multilevel GE technique would probably have increased the accuracy by allowing the investigator to assess the graft on contiguous images. Most likely, such a strategy would have reduced the number of grafts imaged at an inadequate plane.
Another important limitation is that given the relatively large voxel sizes, we cannot visualize graft stenoses. Although phase velocity imaging can be a measure of graft function, the site and morphology may also be important in the treatment strategy. Moreover, a graft stenosis or occlusion in a distal part of a sequential graft may not hamper the runoff from the proximal parts and therefore can be missed by functional assessment in the proximal part.
Improvement in imaging of the distal parts of bypass grafts can be expected from the recently introduced MR angiographic techniques developed for visualizing coronary arteries. By use of ultrafast GE techniques, two-dimensional images of coronary arteries can be obtained within a breath-hold, or three dimensional imaging can be obtained when combined with respiratory gating.20 21 22 Similarly, these approaches could be used for imaging of bypass grafts.
From the experience gained from our study, we would propose the following protocol for clinical use: first, a multislice, GE series in a transverse and sagittal plane or sequential two-dimensional breath-hold MR angiography if available, and second, phase velocity mapping at levels selected from the information obtained from the first step.
The good interobserver agreement and accuracy for cine GE images were important findings in the use of MR as a clinical tool in the noninvasive assessment of graft patency. Additional quantitative graft flow imaging was demonstrated to be feasible with a high success rate and revealed a biphasic forward flow pattern in saphenous vein grafts.
Future directions of MR imaging of bypass grafts might include volumetric flow studies, in conjunction with stress agents, to investigate the effects of graft or distal vessel stenosis on the flow reserve of the graft. Also, this technique might be combined with direct imaging of bypass graft stenoses by MR angiography,20 21 22 23 which might ultimately offer a comprehensive, noninvasive approach to patient stratification.
Selected Abbreviations and Acronyms
|bpm||=||beats per minute|
|LAD||=||left anterior descending coronary artery|
|LCx||=||left circumflex coronary artery|
|RCA||=||right coronary artery|
- Received July 10, 1995.
- Revision received September 27, 1995.
- Accepted October 4, 1995.
- Copyright © 1996 by American Heart Association
Jenkins JPR, Love HG, Foster CJ, Isherwood I, Rowlands DJ. Detection of coronary artery bypass graft patency as assessed by magnetic resonance imaging. Br J Radiol. 1988;61:2-4.
Aurigemma GP, Reichek N, Axel L, Schiebler M, Harris C, Kressel HY. Noninvasive determination of coronary artery bypass graft patency by cine magnetic resonance imaging. Circulation. 1989;80:1595-1602.
Rubinstein RI, Askenase AD, Thickman D, Feldman MS, Agarwal JB, Helfant RH. Magnetic resonance imaging to evaluate patency of aortocoronary bypass grafts. Circulation. 1987;76:786-791.
Underwood SR, Firmin DN, Klipstein RH, Rees RSO, Longmore DB. Magnetic resonance velocity mapping: clinical application of a new technique. Br Heart J. 1987;57:404-412.
Van Rossum AC, Sprenger M, Visser FC, Peels KH, Valk J, Roos JP. An in vivo validation of quantitative blood flow imaging in arteries and veins using magnetic resonance phase-shift technique. Eur Heart J. 1991;12:117-126.
Fleiss JL. Statistical Methods for Rates and Proportions. New York, NY: Wisley; 1973:143-147.
Fusejima K, Takahara Y, Sudo Y, Murayama H, Masuda Y, Inagaki Y. Comparison of coronary hemodynamics in patients with internal mammary artery and saphenous vein coronary artery bypass grafts: a noninvasive approach using combined two-dimensional and Doppler echocardiography. J Am Coll Cardiol. 1990;15:131-139.
Pitt A, Griesinger GC, Ross RS. Measurement of blood flow in the right and left coronary artery beds in humans and dogs using the 133xenon technique. Cardiovasc Res. 1969;3:100-106.
Frank MJ, Levinson GE, Hellens HL. Left ventricular oxygen consumption blood flow and performance in mitral stenosis. Circulation. 1965;31:824-833.
Simon R, Amende I, Oelert H, Hetzer R, Borst HG, Lichtlen PR. Blood velocity, flow and dimensions of aortocoronary venous bypass grafts in the postoperative state. Circulation. 1982;66(suppl I):I-34-I-39.
Pennell DJ, Keegan J, Firmin DN, Gatehouse PD, Underwood SR, Longmore DB. Magnetic resonance imaging of coronary arteries: technique and preliminary results. Br Heart J. 1993;70:315-326.