topeeng.gif (8383 bytes)

[ Scientific Activity - Actividad Científica ] [ Brief Communications - Temas Libres ]

Noninvasive Detection of Myocardial Ischemia From Perfusion Reserve Based on Cardiovascular Magnetic Resonance

Nidal Al-Saadi, MD, Eike Nagel, MD, Michael Gross,  MD, Axel Bornstedt, PhD, Bernhard Schnackenburg,  PhD, Christoph Klein, MD, Helmut Oswald, PhD, Eckart Fleck, MD


Background: Myocardial perfusion reserve can be non-invasively assessed with cardiovascular magnetic resonance. In this study the diagnostic accuracy of this technique for the detection of significant coronary artery stenosis was evaluated.
Methods and Results: In 15 patients with single vessel and five patients without significant coronary artery disease the signal intensity time curves of the first pass of a gadolinium-DTPA bolus injected via a central vein katheter were evaluated before and after dipyridamole infusion to validate the technique. A linear fit was used to determine the upslope and a cut off value for the differentiation between the myocardium supplied by stenotic and nonstenotic coronary arteries was defined. The diagnostic accuracy was then examined prospectively in 40 patients with coronary artery disease and was compared with coronary angiography.
A significant difference in myocardial perfusion reserve between ischemic and normal myocardial segments (1.08± 0.23 and 2.33± 0.41; p<0.001) was found which resulted in a cut off value of 1.5 (mean minus 2 standard deviations of normal segments).
In the prospective analysis sensitivity, specificity and diagnostic accuracy for the detection of coronary artery stenosis (³ 75%) were 93%, 89%, and 91%. Inter- and intra-observer variability for the linear fit were low (r= 0.96 and 0.99).
MR first pass perfusion measurements yielded a high diagnostic accuracy for the detection of coronary artery disease. Myocardial perfusion reserve can be easily and reproducibly determined by a linear fit of the upslope of the signal intensity time curves.



In principle, the reduction of myocardial perfusion is a sensitive indicator for myocardial ischemia, as myocardial blood flow is directly correlated to myocardial oxygen supply. Such measurements are superior to coronary angiography for the detection of myocardial ischemia, as the functional relevance, rather than the morphological appearance, of a stenosis is assessed. Clinical routine measurements of myocardial perfusion are performed with single photon emission computed tomography (SPECT) or with positron emission tomography (PET). To induce myocardial ischemia in myocardial regions supplied by stenotic coronary arteries physical or pharmacological stress is applied. Sensitivity and specificity for the detection of significant coronary artery disease with SPECT or PET range from 83% to 95% and 53% to 95%. However, these techniques have a rather low spatial resolution and are not suitable for the detection of subendocardial perfusion defects, which by themselves are extremely sensitive to the occurrence of myocardial ischemia. In addition, the requirement of radioactive markers prohibits the use of these techniques for follow up examinations. The long washout time of some tracers makes repeated measurements on the same day difficult and, finally, due to the characteristics of the tracers used, SPECT imaging is limited by artifacts that arise mainly in obese patients and women especially in the inferolateral and anterior myocardial region. PET has a higher sensitivity and specificity than SPECT, but in addition to its resolution problems it is burdened by its limited availability.

Ultrafast magnetic resonance (MR) tomography allows an analysis of myocardial perfusion by the use of the first pass of a T1 shortening contrast agent bolus. The advantages of MR tomography are its high spatial resolution, the independence from radioactive tracers, and thus attenuation artifacts, as well as its capability for repeat examinations for follow-up. Several studies have shown in principle that an analysis of myocardial perfusion with MR is possible and may even permit a quantitative assessment of myocardial blood flow.
The aim of this study was to define a threshold value for ischemic regions by myocardial perfusion reserve. This was measured by cardiovascular magnetic resonance to differentiate the myocardium supplied by a stenotic coronary artery from the myocardium supplied by a nonstenotic coronary artery. Also we aimed to determine prospectively the diagnostic accuracy of this cut off value for the detection of significant coronary artery stenosis in patients with suspected coronary artery disease.



The study population consisted of 60 patients (Table 1) who were referred for coronary angiography. Written and informed consent was obtained from all patients. Initially, 15 patients with single vessel disease and five patients with chest pain, but without significant stenoses of the coronary arteries, were examined to define the cut off values in perfusion measurements for the detection of significant coronary artery stenosis (group A). This group served for the validation of the technique and for the determination of inter- and intra-observer variability. Then, the subsequent 40 patients with suspected single or double vessel disease, who were referred for a coronary angiography because of new chest pain or progressive symptoms, were prospectively examined by the use of the previously defined threshold value (group B). In this group the diagnostic accuracy of MR perfusion reserve measurement in comparison with angiography was assessed. Patients were excluded if they were <18 years old, had a history of prior myocardial infarction, unstable angina, hemodynamic relevant valvular disease, ventricular extrasystoly ³ Lown III, atrial fibrillation, ejection fraction <30%, blood pressure >160/95 mmHg or <100/70 mmHg, obstructive pulmonary disease, known claustrophobia, or a contraindication for an MR-examination such as incompatible metal implants.

Table 1

table1.gif (4598 bytes)

After the MR examination all patients underwent left-sided cardiac catheterization and bi-plane selective coronary angiography in Judkins technique. Coronary stenoses were filmed in the center of the field from multiple projections, and as much as possible overlap of side branches and foreshortening of relevant coronary stenoses was avoided. Coronary angiograms were quantitatively assessed for high grade coronary artery stenoses (³ 75% area stenosis). The examiner was blinded to the MR examination.

The patients were examined in the supine position with a 1.5 Tesla whole body MR tomograph (ACS NT, Philips, Best, The Netherlands). After two rapid surveys to determine the exact position and axis of the left ventricle, a short axis slice at the height of the origin of the papillary muscles was chosen for perfusion imaging using an ECG triggered T1-weighted inversion recovery single shot turbo gradient echo sequence (inversion pulse, pre-pulse delay 360 ms, acquisition duration 360 ms, flip angle 15°, echo time 1.7 ms, repetition time 9 ms). Slice thickness was 8 mm with a spatial resolution of 1.7 x 1.9 mm. A bolus of 0.025 mmol gadolinium DTPA/kg body weight (Magnevist, Schering AG, Berlin) was rapidly injected by hand and flushed with 10 ml of 0.9% NaCl. Sixty dynamic images (one image per heart beat) were acquired during the first and second pass of the contrast agent. The dipyridamole infusion was discontinued prematurely upon patient request or when chest discomfort indicative of progressive or severe angina, dyspnea, decrease in systolic pressure >40 mmHg, severe supra-ventricular or ventricular arrhythmias or other adverse effects occurred. Aminophylline was administered as required.

In all images the endo- and epicardial contours were traced by an examiner blinded to the angiographic results. The myocardium was divided into 6 equiangular segments. An additional region of interest was placed within the cavity of the left ventricle excluding myocardial segments or papillary muscles (Figure 1). Images acquired after premature ventricular beats or insufficient cardiac triggering were excluded from the analysis to guarantee steady state conditions. Signal intensity (SI) was determined for all dynamics and segments (Figure 2). The native SI was subtracted and the upslope of the resulting SI-time curve was determined by the use of a linear fit. The results of the myocardial segments were corrected for the input function by dividing the upslope of each myocardial segment through the upslope of the left ventricular SI curve which was regarded as a measure of the input function. Perfusion reserve was calculated by dividing the results at maximal vasodilation by the results at rest. In group A the segment with the lowest myocardial perfusion reserve within the territory of the stenotic coronary artery was defined as ‘ischemic’. All segments of the patients without significant coronary artery disease and the contralateral segment opposite to the ischemic segment in the 15 patients with single vessel disease were defined as ‘nonischemic’. The absolute upslope at rest, after dipyridamole infusion, as well as myocardial perfusion reserve of ischemic and nonischemic segments, were compared. The cut off value was defined as the mean perfusion reserve minus 2 standard deviations of all nonischemic segments. In group B myocardial perfusion reserve was calculated for all segments. If the myocardial perfusion reserve was less than the defined cut off value, the segment was classified as pathological (MPR+), if it was more than the cut off value it was defined as normal (MPR-). If at least one segment within the territory of a coronary artery was found to be ischemic, MR was regarded as positive for that region.

fig1.jpg (49361 bytes)
Fig. 1: MR images of the transit of the gadolinium bolus through the right ventricle (a), the left ventricle (b) and the left ventricular myocardium (c). Note the compact contrast bolus, with almost complete exit of the contrast agent from the right ventricle, when passing through the left ventricle.



fig2.jpg (60292 bytes)
Fig. 2: The six evaluated myocardial segments of the left ventricular myocardium with the determination of the coronary artery territories for each segment and the resulting signal intensity versus time curves of the myocardial segments each compared to the signal intensity curves of the left ventricular cavity. SI: signal intensity (arbitrary units). LAD: left anterior descending coronary artery; LCX: left circumflex coronary artery; RCA: right coronary artery



In group A no significant difference between myocardial segments supplied by stenotic coronary arteries (median area stenosis 94%) and contralateral myocardial segments supplied by normal coronary arteries was found at rest (1.6± 0.7 vs. 1.6± 0.8). After dipyridamole infusion there was a significant difference between the ischemic and the nonischemic myocardial segments (2.1± 0.9 vs. 2.9± 1.0; p<0.05) (Figure 4). Myocardial perfusion reserve after dipyridamole infusion resulted in highly significant differences between myocardial segments supplied by stenotic coronary arteries (1.08± 0.23) and myocardial segments supplied by nonstenotic coronary arteries (2.34± 0.41; p<0.001) (Figure 5). A cut off value for myocardial perfusion reserve of £ 1.5 was defined as ischemic.
In group B 60 coronary artery stenoses were found by angiography. In this group myocardial perfusion reserve was 1.16 ± 0.29 in the ischemic segments and 2.17 ± 0.62 in the nonischemic segments (p<0.001). For the diagnostic results, see table 2.


fig3.gif (5314 bytes)

Fig. 3: Inter-observer and intra-observer variability for the linear fit of the upslope



fig4.gif (3482 bytes)
fig4a.gif (4520 bytes)
Fig. 4: The absolute values of the upslope of ischemic (closed circles) and the contralateral control segments (open circles) at rest and after vasodilatation with dipyridamole (stress) in group A represented as single values and mean ± one standard deviation. The difference between the ischemic and control segments at rest were not significant.


fig5.gif (3300 bytes)
Fig. 5: Myocardial perfusion reserve of the ischemic and control segments (contralateral segments in group A -closed circles- and all segments of the five patients without significant coronary artery disease – open circles). Mean ± one and two standard deviations.


      Table 2

table2.gif (4285 bytes)



Magnetic resonance perfusion imaging can be used to detect coronary artery stenosis with high diagnostic accuracy. In this study a sensitivity of 90% and a specificity of 83% for the detection of significant coronary artery stenosis was reached in 34 patients by the use of a previously defined threshold for ischemic myocardial regions. A linear fit of the upslope of the first pass of a gadolinium-DTPA bolus before and after dipyridamole infusion enabled an easy and reproducible determination of the myocardial perfusion reserve.
In this study, a new and easy approach for the determination of myocardial perfusion reserve was used. To circumvent problems with a gamma variate function and to minimize the influence of diffusion on the results, a linear fit of the upslope of the SI- time curves, rather than mean transit time, maximal signal intensity, downslope or time to maximal signal intensity, was used for the present analysis. The linear fit was highly reproducible with an excellent inter- and intra-observer variability and could be performed in 97% of all evaluated myocardial segments.

In this study coronary angiography was used as the reference method for the detection of coronary artery stenosis. As coronary angiography detects luminal morphology rather than the functional significance of a stenosis, ‘false positive’ MR results might in fact be ‘false negative’ angiograms. Three of the seven segments that had a ‘false positive’ reduction of myocardial perfusion reserve showed one or more stenoses <75% area reduction of the corresponding coronary artery on quantitative angiography. Further two false positive segments were in one patient with diffuse atherosclerosis of the nonstenotic coronary arteries. Thus, specificity of myocardial perfusion measurements with MR may be as high as 95 % if corrected for these patients.
In the current study we have shown that MR first pass perfusion measurements yield a high diagnostic accuracy for the detection of coronary artery disease. Myocardial perfusion reserve can be easily and reproducibly determined from the upslope of the SI- time curves.


Questions, contributions and commentaries to the Authors: send an e-mail message (up to 15 lines, without attachments) to , written either in English, Spanish, or Portuguese.