Author + information
- Received July 5, 2002
- Revision received September 24, 2002
- Accepted October 17, 2002
- Published online March 5, 2003.
- Christian M Wacker, MD*,
- Andreas W Hartlep, PhD†,
- Stefan Pfleger, MD‡,
- Lothar R Schad, PhD†,
- Georg Ertl, MD* and
- Wolfgang R Bauer, MD, PhD*,* ()
- ↵*Reprint requests and correspondence:
Dr. Wolfgang R. Bauer, Medical Clinic, Department of Cardiology, University of Wuerzburg, Josef-Schneider-Str. 2, 97080 Wuerzburg, Germany.
Objectives Evaluation of the severity of a coronary artery stenosis is of paramount importance for therapy. A relevant stenosis provokes post-stenotic microvascular dilation with capillary recruitment. This autoregulatory response was investigated in the present study by use of susceptibility-sensitive magnetic resonance imaging (MRI) without contrast agents.
Background Functional alterations of the microvascular system may be studied noninvasively and without a contrast agent by susceptibility-sensitive MRI, which is based on the paramagnetic property of deoxyhemoglobin. This effect, also referred to as the “blood oxygenation level-dependent (BOLD) effect,” is investigated by phase relaxation (T2*) measurements.
Methods In patients (n = 16) with single-vessel coronary artery disease, no history of myocardial infarction, normal left ventricular function at rest, and a positive stress echocardiogram, the susceptibility-sensitive parameter T2* was assessed in the myocardium.
Results In regions associated with the stenotic artery, T2* was significantly lower than in residual myocardium (p < 0.01). This difference in T2* increased after application of the vasodilator dipyridamole (p < 0.001). In patients being re-investigated after therapeutic interventions, the microvascular dilation was partly removed.
Conclusions For the first time, we could show that myocardial BOLD MRI detects post-stenotic capillary recruitment dependent on coronary artery stenosis.
Evaluation of the relevance of coronary artery stenosis is of paramount importance for therapeutic decision-making. A severe stenosis implies the activation of compensation mechanisms, including post-stenotic dilation of the microvascular system. This lowering of vascular resistance aims to maintain sufficient blood supply, at least under rest conditions. However, many obstacles hamper the noninvasive assessment of this autoregulatory response.
We anticipated that post-stenotic vasodilation implies capillary recruitment (1–3). Almost all (i.e., >90%) of intramyocardial blood resides in that type of vessel (4). Due to their large arteriovenous oxygenation difference, myocardial capillaries contain considerable amounts of deoxyhemoglobin. Hence, in regions with autoregulatory capillary recruitment, the tissue concentration of deoxyhemoglobin should be elevated when compared with myocardium supplied by a normal vessel. Due to its paramagnetic property and its intravascular confinement, the natural contrast agent deoxyhemoglobin may be assessed by susceptibility-sensitive, or also called blood oxygenation level-dependent (BOLD), magnetic resonance imaging (MRI).
Whereas BOLD imaging is well established in functional imaging of the brain (5–7), cardiac BOLD imaging in whole animal models (8,9)or humans (10–13)is hampered by breathing and motion artifacts, so that investigations were mainly done in isolated organ preparations. We recently developed a stable imaging technique (14)to assess the BOLD effect in the heart by obtaining maps of the phase relaxation (T2*) time. This technique is similar to that of Anderson et al. (13), who estimated myocardial iron overload.
The results of our recent studies and theoretical considerations indicated that the BOLD-related part of transverse relaxation in myocardium is mainly due to inhomogeneous pericapillary magnetic fields, or alternately, the BOLD-related part of T2* is mainly a function of the density, size, and deoxyhemoglobin content of capillaries (1,2). In the present pilot study, we exploited the dependence of T2* on parameters of the capillary system to characterize capillary recruitment in the myocardium of patients with stable angina due to single-vessel coronary artery disease (CAD). Because blood supply and oxygen demand are balanced under rest conditions, capillary recruitment in regions supplied by a stenotic coronary artery should result in an increase of tissue concentration of deoxyhemoglobin and, hence, in reduced T2*. Additionally, measurements were done after vasodilation with dipyridamole, which normally shifts the blood supply/oxygen demand balance toward hyperperfusion. In myocardium supplied by a normal coronary artery, this decreases the tissue concentration of deoxyhemoglobin and T2* increases, as previously shown in our laboratory (14). In myocardium supplied by a stenotic coronary artery, post-stenotic capillary recruitment implies only a moderate dilatory reserve (i.e., one would expect only a moderate increase in T2*).
It is obvious that verification of the aforementioned hypothesis should have significant implications for BOLD imaging as a diagnostic tool for determination of capillary reserve and, hence, evaluation of the severity of coronary artery stenosis.
Volunteers and patients
Before the patient study, 16 volunteers (20 to 59 years old; mean 31 ± 10 years) without a history of cardiovascular disease were studied by MRI.
Then, 16 patients were included (44 to 77 years old, mean 63 ± 9 years). Inclusion criteria included stable angina, single-vessel CAD on the coronary angiogram (degree of stenosis >70%; in the left anterior descending coronary artery [LAD; n = 12], right coronary artery [n = 2], and left circumflex branch of the coronary artery [n = 2]), wall motion abnormalities on the stress echocardiogram, and no wall motion abnormalities of the left ventricle at rest. A previous myocardial infarction was excluded by a medical history, evaluation of electrocardiographic (ECG) abnormalities, and past enzyme elevations. Stable angina implies that there was no ischemia under rest conditions, but that there was functional severity of the stenosis under stress conditions. The single-vessel condition assured a simple relationship between the culprit artery and supplied myocardium. The last criteria excluded that structural abnormalities (e.g., scar) were responsible for T2* alterations.
Patients underwent X-ray angiography, stress echocardiography, and MRI within 4 ± 2 days. Two patients with LAD stenosis were re-investigated—one after percutaneous transluminal coronary angioplasty (PTCA) with stent implantation and another after coronary artery bypass graft surgery (CABG). Written, informed consent was obtained from all participants, and the local ethics committee approved the study.
Cardiac catheterization was done after inguinal puncture of the right femoral artery with a 5F catheter. Coronary arteries were assessed in the typical left and right anterior oblique projections.
Echocardiography was done at rest and after administration of dipyridamole (rate of 0.56 mg/kg body weight over 4 min). After the examination, patients were antagonized with 200 mg aminophylline.
Magnetic resonance examination
Magnetic resonance examinations were performed on a 1.5-T whole-body scanner (SIEMENS Vision, Erlangen, Germany) with gradient overdrive, using the integrated body coil for radiofrequency excitation and a four-element phased-array coil for signal reception. Fast, non–ECG-triggered scout images (i.e., fast low-angle shot [FLASH]) were used to position the imaging slice at the short-axis plane of the heart between the valve system and papillary muscle.
For T2* measurements, a segmented gradient echo pulse sequence was used, which acquired 10 successive gradient echoes per radiofrequency excitation in a single breathhold, as recently proposed by our laboratory (14).
Volunteers and patients were informed about the possible side effects of dipyridamole, such as headaches, nausea, and general feelings of warmth. All participants avoided the consumption of tea and coffee, as well as medications containing aminophylline at least 12 h before the dipyridamole test protocol. Heart rate was continuously monitored, and blood pressure was measured before and after the examinations. Images were acquired repeatedly at rest and under dipyridamole-induced stress (rate of 0.56 mg/kg, infused over 4 min through an antecubital vein). Measurements were done in 1-min repetitions until the initial heart rate was re-established. Infusion was controlled by an infusion system (CAI 626P, DOLTRON AG, Uster, Switzerland), which allowed precise adjustment of the infusion rate relative to the patient’s weight.
For data evaluation, magnetic resonance images were transferred to an off-line workstation and processed using a home-written software package. In a first step, a region of interest (ROI) was placed around the complete myocardium in the first-amplitude image (which was the brightest) of the 10 echo image series. Superimposing susceptibility artifacts, mainly at the inferolateral border of the heart, could be clearly identified in the corresponding amplitude images and led to exclusion of these pixels from evaluation. To minimize the effect of ROI positioning on the final results, the ROI placement was repeated 2 to 3 times and compared by the same observer.
Maps of the apparent transverse T2* were calculated by applying a linear fit to the logarithm of the signal intensities in the ROI. The noise amplitude was first determined in a region outside the body. In the following fitting procedure, only those signal intensities that exceeded the noise amplitude threshold by a factor of 2 to 3 were used.
Average T2* values of the myocardium were calculated from the mean value of the signal intensities in the ROI. To localize myocardial areas with lower T2* values, a series of T2* maps acquired before, during, and after dipyridamole administration were displayed in a fast cine loop. In this display mode, areas of reduced or no T2* increase (i.e., T2*r) with dipyridamole were identified visually, and ROIs were placed manually. For these regions, average T2*rvalues were determined and compared with the remaining myocardium (T2*n).
To test the statistical significance of the T2* increase with dipyridamole, T2* values (i.e., T2*rand T2*n) were grouped into two sets. Baseline values of T2* were defined from the data acquired before dipyridamole administration and before there was an increase in heart rate. The second data set included all those measurements where the heart rate exceeded the baseline value by 10%. The relative change in T2* was defined as the mean T2*r− mean T2*nnormalized by T2*n. An assumption of normality was demonstrated using the Kolmogorov-Smirnov test. The significance (p value) of the difference of mean values was calculated using the paired ttest. Results are expressed as the mean value ± SD.
After dipyridamole infusion, T2* increased significantly from 35 ± 3 ms (range 28 to 40 ms) to 40 ± 4 ms (range 29 to 48 ms) (i.e., by 10 ± 5 %; p = 0.01). The T2* maps appeared homogeneous under baseline and stress conditions, and no areas with reduced T2* values could be delineated (Fig. 1). The average heart rate increased from 62 ± 9 beats/min to 84 ± 11 beats/min (i.e., by 35 ± 15 %). The mean duration of the increased heart rate was 9 ± 6 min.
Under rest conditions, regions with reduced T2* values (i.e., T2*r) were clearly detectable (Fig. 2). These areas (covering 25% to 50% of the area of the left ventricle in the imaged short-axis view) corresponded well to the regions with wall motion abnormalities. Reduced T2* values due to superimposing susceptibility artifacts at phrenicomediastinal recess (15)were clearly detectable in a few patients and led to exclusion of these myocardial areas from evaluation. The relative change in T2* was 31 ± 9% (p < 0.01), and this difference increased to 43 ± 21% (p = 0.0001) under stress conditions (Fig. 3).
The T2* increase started 3 ± 1 min after the onset of dipyridamole infusion. The average heart rate increased from 62 ± 8 beats/min to 73 ± 9 beats/min (i.e., by 19 ± 9%). The mean duration of the increased heart rate was 15 ± 6 min. The observed side effects of dipyridamole were angina (n = 5), dyspnea (n = 1), and dizziness (n = 2). In two cases, the examination had to be interrupted early due to severe angina. One patient was antagonized with intravenous aminophylline (200 mg), and two patients took nitroglycerin to overcome angina.
Two patients with stenosis of the proximal LAD were re-investigated after therapeutic interventions (10 weeks after PTCA with stenting [Fig. 4] and 20 weeks after CABG [Fig. 5]). When observing the obtained T2* maps, differences between regions of reduced T2* and regions of normal myocardium were now less pronounced than they were before the interventions.
The BOLD effect
The BOLD MRI makes use of the paramagnetic property of intravascular deoxyhemoglobin. In the presence of an external magnetic field, deoxyhemoglobin reduces the transverse spin relaxation time T2* as a function of its tissue concentration (i.e., intravascular oxygenation state) and its distribution in tissue (i.e., the arrangement and the filling state of the vessel tree in tissue).
What affects T2* in myocardium?
In contrast to that of patients, the myocardium of healthy volunteers did not show areas with reduced T2*. In normal hearts, an increase of myocardial perfusion by dipyridamole led to a significant increase in T2*, and calculated T2* maps were homogeneous under rest and stress conditions. This result is related to the fact that dipyridamole enhances coronary flow without much alteration of myocardial performance (16). This imbalance toward higher oxygen supply than oxygen demand decreases the deoxyhemoglobin concentration and, hence, the amplitude of the field inhomogeneities. The reduced T2* in myocardial segments supplied by a stenotic coronary artery and its modest response to vasodilation could, in principle, be due to structural tissue alterations and to factors related to functional changes in the coronary circulation. However, when considering the first, a different tissue composition of myocardium supplied by the stenotic artery is unlikely, as there was no history of previous myocardial infarctions in these patients, and the normal left ventricular function at rest ruled out structurally altered, viable (e.g., stunned or hibernating) myocardium. The best explanation for the decreased T2* in regions associated with a stenotic artery is that there is a higher concentration of paramagnetic deoxyhemoglobin. In principle, this could be caused by either an increase of intravascular deoxyhemoglobin or an increase of the vessel compartments containing deoxyhemoglobin. In nonischemic myocardium, however, coronary autoregulation maintains the arteriovenous difference of oxygen saturation of hemoglobin at an almost constant rate of ∼70% (17). According to the clinical selection of the patients and normal left ventricular function, the myocardium supplied by the stenotic artery was not ischemic under rest conditions (i.e., an increase of the intravascular concentration of deoxyhemoglobin can be ruled out as responsible for the reduced T2* at rest). Hence, the explanation left is that the relative intravascular volume of vessels containing deoxyhemoglobin must increase.
Evidence for capillary recruitment?
Only vessels containing a significant amount of deoxyhemoglobin are responsible for the BOLD effect. These are capillaries and veins, when arterial oxygenation is normal. Cardiac blood vessels can be divided in epicardial and intramyocardial vessels. The intracapillary volume is a fraction of the intramyocardial blood volume. Based on the anatomic data of Kassab et al. (18), Kaul and Jayaweera (4)state that this fraction is >90% of the intramyocardial blood volume. Thus, the fraction of the intramyocardial venous compartment is smaller than 10%. This implies that of all intramyocardial vessels, mainly capillaries contribute to the BOLD effect. Theoretical considerations from our group demonstrated that this capillary contribution is sufficient to be responsible for the myocardial BOLD effect, as observed in animal studies and humans (1,2). The main portion of the venous volume resides in larger epicardial veins, which, in principle, may also contribute to the BOLD effect. However, one would have expected a spotted texture of myocardium visible in amplitude images and calculated T2* maps, which was not the case. In our study, visible veins were excluded from segmentation.
When the BOLD effect is mainly related to intracapillary deoxyhemoglobin, it is evident that the reduced T2* in myocardium supplied by a stenotic coronary artery reflects the elevated relative intracapillary blood volume. This increase of the relative intracapillary blood volume is most likely a result of coronary autoregulation, which maintains sufficient blood supply by post-stenotic dilation of coronary resistance vessels and precapillary sphincters, leading to capillary recruitment. This vasodynamic response was also found by echocardiographic studies in animal models with an induced coronary stenosis (3). In summary, there is evidence that the decrease of T2* reflects the autoregulatory response to a coronary stenosis on the microvascular level. This vasodynamic response also explains the diminished increase of T2* after dipyridamole application in regions supplied by a stenotic artery. Because dilation of resistance vessels and precapillary sphincters already occurs under rest conditions in these regions, dipyridamole cannot considerably dilate these vessels further (i.e., there is no significant increase of local blood flow and, hence, no increase in T2*) (Fig. 6).
A reduction in T2* can also arise due to artifacts, as recently described (11,14,15). In our study, in a few patients, reproducible susceptibility artifacts occurred in phrenicomediastinal recess, which superimposed inferolateral sections of myocardium. This phenomenon is caused by through-plane perturbations of veins (11)and the heart-lung interface along the inferior segments of the heart (15). We found that in humans, these artifacts could be reduced by performing measurements in end-expiratory breathhold when phrenicomediastinal recess was diminished (14).
One might argue that patients undergoing MRI were selected according to their clinical and angiographic presentation and a positive stress echocardiogram. However, this was a pilot study to demonstrate that a non-contrast agent MRI technique may detect alterations of microcirculation in myocardium supplied by a stenotic coronary artery. The ischemia-related potential of this stenosis was proven by stress echocardiography, which is a well-established method in this field (19). It was not our intention to compare established techniques with the MRI technique to gain specificity or sensitivity of the latter, which requires a much higher number of patients. In this pilot study, we did not perform scintigraphy for the following practical reasons: in order to find well-defined regions with potential altered microcirculation, only patients with single-vessel disease were included. In normal clinical routine, angiographic diagnosis and therapy are performed in one session in these patients. However, MRI interrupted this procedure, and the interval between diagnosis and therapy would have been prolonged if scintigraphy were performed, which would have been in conflict with ethical and cost considerations. Two patients were re-examined after therapeutic interventions, and the obtained T2* maps were more homogeneous than they were before therapy. Of course, stenting and CABG induce susceptibility artifacts, which can affect image quality. In the present study, however, this could reliably be excluded by choosing a suitable imaging plane (i.e., below the stent), and in the future, nonmagnetic stents and clips might overcome this problem.
The authors thank Michael Bock, PhD, for initial sequence development.
☆ This work was supported by the German Cardiac Society (to Dr. Wacker), Forschungsfonds Klinikum Mannheim/Heidelberg Projekt 42 (to Dr. Hartlep), and Grant Sonderforschungsbereich 355 and Graduiertenkolleg “NMR” HA 1232/8-1 (to Dr. Bauer).
Drs. Wacker and Bauer equally contributed to this article.
- blood oxygenation level-dependent
- coronary artery bypass graft surgery
- coronary artery disease
- left anterior descending coronary artery
- magnetic resonance imaging
- percutanous transluminal coronary angioplasty
- region of interest
- phase relaxation
- Received July 5, 2002.
- Revision received September 24, 2002.
- Accepted October 17, 2002.
- American College of Cardiology Foundation
- Wu C.C.,
- Feldman M.D.,
- Mills J.D.,
- et al.
- Ogawa S.,
- Lee T.M.,
- Kay A.R.,
- Tank D.W.
- Ogawa S.,
- Tank D.W.,
- Menon R.,
- et al.
- Beache G.M.,
- Herzka D.A.,
- Boxerman J.L.,
- et al.
- Anderson L.J.,
- Holden S.,
- Davis B.,
- et al.
- ↵Tauchert M, Hilger HH. Myocardial autoregulation. In: Schaper W, editor. The Pathophysiology of Myocardial Perfusion. Amsterdam: Elsevier/North-Holland Biomedical Press, 1979:141
- Kassab G.S.,
- Lin D.H.,
- Fung Y.C.
- EPIC (Echo Persantin International Cooperative) Study Group,
- Picano E.,
- Ostojic M.,
- Sicari R.,
- Baroni M.,
- Cortigiani L.,
- Pingitore A.