Gerald M. Pohost, MD Mary Gertrude Waters Professor of Cardiovascular Medicine and Director, Center for NMR Research and Development, University of Alabama at Birmingham Mark Doyle, PhD Associate Professor of Medicine, Center for NMR Research and Development, University of Alabama at Birmingham

First, it is important to note that myocardial perfusion imaging using radionuclides has become an extremely important tool for diagnosis of ischemic heart disease. However, radionuclide imaging has relatively low resolution and the property of emitting radiation, gamma photons, or X-rays.

Several issues with regard to the way we now perform MR perfusion imaging need to be discussed. First, it is important that we deliver a very compact bolus of contrast agent as rapidly as possible so that it is delivered as a tight bolus to the myocardium so that we can see contrast between deficits in perfusion, infarcted myocardium, and normally perfused myocardium. Second, we need to achieve a substantial amount of flow reserve using a vasodilator like adenosine or dipyridamole. Third, the spatial and temporal resolution of the study is extremely important, as they influence both sensitivity and specificity.

Delivering a Compact Bolus

Typically, the contrast agent is delivered using a power injector via a catheter in the anticubital vein. Several factors determine the speed of delivery (table 1). These factors include the viscosity of the agent (lower viscosity is preferred); the catheter diameter (the larger the better); and possible limits on injection rate due to safety concerns (the agents should be delivered at the highest safe rate).

Viscosity is determined by a number of factors, including the temperature of the agent (table 2). At room temperature (68°F, 20°C), Magnevist has a viscosity (4.9 cP) nearly 2.5 times higher than that of ProHance (2.0 cP). At body temperature, 37°C, the viscosity of Magnevist is reduced by 40% (2.9 cP) and that of ProHance by 35% (1.3 cP). So, at both temperatures, the difference in viscosity is about 2.5-fold although the agents contain identical concentrations of gadolinium (0.5 mmol/mL).

When performing a myocardial perfusion imaging study, we usually do it in this order: functional imaging is performed prior to contrast injection (control state); then perfusion imaging is performed at a low contrast dose; and finally it is performed using a higher contrast dose. The higher dose is given so that the initial contrast dose does not have a major effect on images acquired with the second injection. The second injection is given during the vasodilation phase, i.e. during the administration of adenosine or dipyridamole.

Figure 1 illustrates the differences in intravascular pressure versus time of delivery of both of the agents. This graph shows that, during delivery, the pressure rises very rapidly with Magnevist at 6 mL/sec and less rapidly at 4 mL/sec, as you would expect. The pressure is substantially less with 6 mL/sec of ProHance and is the lowest with 4 mL/sec of this agent. The lower the intravascular pressure, the less the risk of rupturing the vein.

Unpublished data from our site suggests that ProHance, injected into the anticubital vein, is delivered more effectively than Magnevist because it does not disturb the continuity between needle and vein (personal communication, Ed Walsh, PhD). In my experience, there have been several instances of either vein rupture or needle perforation or dislodgment with Magnevist, but fewer with ProHance.

Myocardial Flow Reserve

As noted earlier, a second area of concern is myocardial flow reserve. Myocardial perfusion imaging is usually performed at baseline and with hyperemia induced by the agents mentioned. The flow reserve is believed to affect the accuracy of myocardial perfusion imaging studies. The first-pass MR perfusion imaging study can be used to simultaneously assess perfusion and myocardial flow reserve. In fact, we have coupled the myocardial flow reserve measurements made by MR with SPECT studies and divided the SPECT studies into low-flow reserve and high-flow reserve. As you will see, the high-flow reserve SPECT studies are substantially more accurate than the low-flow reserve studies.

The WISE Study

The NHLBI-supported WISE (Women's Ischemia Syndrome Evaluation) study was undertaken to develop new techniques to assess women with potential coronary artery disease. The techniques available were grossly ineffective and the incidence of coronary artery disease in women was consistently under-reported. Several of the techniques evaluated in the study were MR approaches. One group studied coronary artery disease using two different MR methods: myocardial perfusion imaging and MR spectroscopy to detect ischemia.

In this study, an index was measured so as to ascertain the flow reserve in the myocardium with or during myo-
cardial perfusion imaging. Both MR and SPECT perfusion studies were obtained in parallel and, in fact, in very close proximity to one another to evaluate the influence of the flow reserve for each.

The sequence of the study is shown in figure 2. First, all patients underwent coronary angiography prior to the study to determine if they had normal coronary arteries. Most of the patients in this study did have normal coronaries. Those with abnormal coronary arteries underwent phosphorus-31 study to see if they had a decrease in the high-energy phosphate, phosphocreatine, induced by handgrip stress that would suggest microvascular disease. Next the study participants underwent a variety of imaging procedures that included the injection of the bolus of MR contrast agent and the injection of Sestamibi, which stays in the myocardium long enough for the MR study to be completed. Both injections were made in very close proximity to one another, so the perfusion pattern should have been identical. At the end of the series, coronary angiography was repeated.

We began the WISE study using Magnevist, but converted to ProHance after a few weeks because of patient safety concerns. Due to the high flow rates needed for myocardial perfusion studies (4 to 6 mL/sec) and the high viscosity of the agent, Magnevist caused higher pressure, resulting in ruptured IV lines and veins. Since our switch to ProHance, we have experienced no ruptured lines and no complications associated with extravasation.

A total of 138 women were assessed to determine if they had perfusion defects. Radionuclide-gated SPECT and quantitative MR first-pass perfusion imaging were performed in all patients. Quantitative X-ray coronary angiography, which was read at a core laboratory, was used as the gold standard. Significant disease was defined >70% stenosis in any of the major coronary vessels or in a major branch. The myocardial flow reserve was categorized as adequate when there was a high hyperemic response or inadequate with a low hyperemic response. Approximately 30 patients were found to have an inadequate flow response (figure 3). SPECT perfusion imaging was shown to have higher specificity, sensitivity, and overall accuracy in patients with adequate flow reserve (figure 4). The same held true for the quantitative studies as well, although the difference between specificities and sensitivities in the inadequate versus adequate flow reserve was not significant until we looked at overall accuracy (figure 5). It is important to note that 80% of the women in this study with chest pain syndromes that were thought to be coronary in etiology had normal coronary arteries on coronary angiogram.

Figure 6 shows an example of a patient with an adequate flow reserve and one with inadequate flow reserve. The contrast on these images was turned up deliberately to see if it was possible to identify a perfusion defect. This is commonly done when reading such studies.

Overall, the conspicuity of defects was substantially greater in the high-flow reserve images than it was in the low-flow reserve images, despite the fact both had similar stenoses of a similar vascular distribution. This happened many times, and this is just one example.

This study also found several discriminating characteristics that were significantly different, including blood pressure and rate-pressure product (table 3). Interestingly, changes in blood pressure between baseline and vasodilatation were not significant predictors.

This study concluded that the accuracy of detection of significant coronary artery disease (CAD) using MR and SPECT myocardial perfusion imaging depends on the myocardial flow reserve achieved. This flow reserve can be measured using a contrast injection with MR, but cannot be measured using radionuclide methods. Therefore, accuracy for detection of significant CAD is higher in the good myocardial perfusion index group and lower in the inadequate myocardial perfusion group. The study also concluded that those with an inadequate myocardial flow reserve index are hemodynamically different at baseline (resting rate-pressure product) prior to the administration of dipyridamole.

Characteristics of an Imaging Sequence

What about the characteristics of an imaging sequence? The ideal MR imaging sequence will capture the first-pass dynamics of a contrast agent in order to be able to assess myocardial perfusion and myocardial flow reserve. It must provide good T1 contrast. It must not adversely affect or be affected by susceptibility effects, and it must image the heart in multiple slices.

The imaging approach that we now use is a hybrid gradient echo/EPI sequence. An inversion saturation pulse applied prior to imaging generally provides T1 contrast and the number of slices obtained is dependent on the heart rate. Usually these studies are done on a short-axis orientation and greater than six slices can be imaged with heart beat resolution allowing evaluation of the transmural extent of the perfusion deficits.

Conclusion

In conclusion, myocardial perfusion imaging with MR in conjunction with a tight bolus delivery using a power injector is at least as accurate as SPECT at the present time. Detection of first-pass contrast agent dynamics allows assessment of myocardial flow reserve, and that influences the accuracy of the ability to detect disease.

The most effective way of obtaining a compact bolus and avoiding extravasation or leakage of the contrast into the tissue around the vein is to use a large-bore catheter and to use the agent with the lowest viscosity available. Warming the agent lowers the viscosity and can help this process. It is also helpful to deliver the agent with a pressure-limiting power injector at the highest possible rate.

Finally, MR imaging methods and hardware are currently adequate to allow myocardial perfusion imaging. MRI hardware should continue to improve, and with further improvements and refinements, myocardial perfusion imaging and flow reserve detection should improve further.

I believe that myocardial perfusion imaging, which provides an evaluation of the amount of remaining viable but ischemic myocardium, will be the most important diagnostic imaging tool for ischemic heart disease assessment. If you want to see the dead myocardium (which we can't do anything about), then delayed hyper-enhancement is the best approach at the present time. One can use phosphorus-31 spectroscopic imaging, but that is something for the future. At the present time, just like the thallium and the Sestamibi scanning became so important and pyrophosphate and the other bone-imaging agents became less important, I think you'll find that when we have developed the optimal MRI approach for myocardial perfusion imaging of the heart, this will be the most clinically useful approach.