AUDIO TRANSCRIPT: What I’d like to talk about today is how MRI is used to quantify aortic regurgitation. I’ll start with the basic principles of how phase contrast imaging is used to measure velocity and flow in MRI. Then I’ll speak a little bit about the methods for quantifying aortic regurgitation. Finally I’ll talk about the reproducibility, accuracy and outcomes data that are available.
The MRI signal originates with the hydrogen nucleus. As you can see here, schematically, much of the hydrogen nuclei are found in water. And when you put hydrogen nucleus in a magnetic field, like an MRI scanner, the hydrogen nucleus acts like a bar magnet. It has a negative and a positive magnetic pole.
That bar magnet or hydrogen nucleus, if you will, spins around the z axis in the x-y plane. That changing magnetic field can be detected as an inducted current or voltage in a loop of wire, which is the MRI coil. The frequency of that signal is directly proportional to the magnetic field strength.
Here’s a schematic showing two water molecules, one that is stationary, say piece of muscle. The other is in an artery, and therefore is moving with blood flow. These hydrogen nuclei (yellow dots) originally start with the same signal at the same point in the x-y plane. When we try to measure flow using phase contrast imaging, what we do is we turn on a magnetic field gradient (that’s a magnetic field that varies as a function of position in the scanner). You can see here that the magnetic field gradient is higher where these water molecules are and lower downstream. Now, if we look at what happens to these hydrogen nuclei that are located at this position, you can see that they both have sped up because they’re both located in an area of higher magnetic field strength. Now, if we wait a little bit of time, what’s going to happen is these static hydrogen nuclei in the muscle are not going to move. Whereas the hydrogen that are in the water that’s flowing with the blood, they’re going to move downstream. After a certain amount of time, Dt, we’re going to apply a magnetic field in the opposite direction. If you look at what happens to the hydrogen nuclei in the static water molecule, they slow down because the magnetic field is lower, and they return back to their original phase that they had before this experiment was done. Whereas the hydrogen nuclei that have moved see a higher magnetic field the second time and are phase advanced even further. It’s this difference in phase that allows us to determine what the distance of the water molecule traveled as a function of time. If you know that distance travelled over a period of time (Dt), you can calculate a velocity. And, that’s the principle by which velocities are measured in the MRI scanner.
Now, a fundamental difference between echocardiography and MRI is that typically in echocardiography velocities and flow are measured in-plane, whereas, with MRI, they’re typically measured through-plane. What do I mean by this? Well, here you can see a three chamber of view where the scan plane is oriented perpendicular to the ascending aorta near the sinotubular junction. Phase contrast imaging gives us two images. One is a magnitude image, which shows what things look like and you can see the ascending aorta here in cross section. It’s circled in red. The second image we get is a velocity image, which shows what the velocity of each pixel is in the image, with black be being very fast in one direction, in this case towards the head, and white being fast in the other direction, namely towards the feet, and gray, meaning not very much motion at all.
What you’ll see here is that with MRI we can look at what the velocity is in each pixel. We know what the area of the pixel is, and when we multiply those together and we add them up for all the pixels, we can calculate what the flow is. That’s very different than in echocardiography, where typically the gate is placed at a single point in the vessel and the assumption is that the velocity of the blood flow is identical at all points in the vessel at that level — something called plug flow. Well, here you can see a typical example of the velocities varying within the ascending aorta. It’s not plug flow. You have much higher velocities here, in the anterior part of the aorta, than you do in other parts of the aorta. That’s one reason why MRI is more accurate than echocardiography in quantifying blood flow.
Another important difference between MRI and echocardiography is that in echocardiography, a lot of weight is put on the size of the jet. Jet morphology is important because you could have a central jet or you could have an eccentric jet and eccentric jets are much harder to evaluate with echocardiography. And, you can see here is an example of an image where the jet is eccentric. It’s actually crawling along the anterior leaflet of the mitral valve. That’s what gives you the Austin-Flint murmur. Pity the poor echocardiographer who needs to go ahead and deal with assessing how bad the aortic regurgitation is in a patient like this, for instance, that has an eccentric jet.
By the way, I’d like to point out that all the images that I’m going to show with this talk are done without contrast. There’s no reason to give MRI contrast if one simply wants to quantify the severity of aortic regurgitation.
Quantification of aortic regurgitation is fundamentally different with MRI. With MRI, we actually measure the flow just distal to the aortic valve near the sinotubular junction, and that is not dependent on the morphology of the jet. You can see that here. This is a screen capture from the analysis being done.
By the way, I’d like to point out that now it’s possible to have automated segmentation with absolutely no user input. The new advances in artificial intelligence have taken out a lot of the need for user input, for segmenting a vessels and determining flow.
Here, you can see the instantaneous flow as a function of time in the ascending aorta. You can see systole here. You can see diastole here. In a normal person, diastolic flow would be near zero during diastole here. If you integrate the area under the curve, you can see we get a regurgitant volume of about 74 ml. And, if you express it as a percentage of the total flow, it’s about 51 percent. So this would be somebody who has severe aortic regurgitation.
The classification system that is most commonly used is the same as that is used in echocardiography. Namely, if there’s less than 30 ml regurgitant volume, or the regurgitant fraction is less than 30 percent, it’s considered mild. Or, if the regurgitant volume is greater than or equal to 60 ml or greater than or equal to 50 percent, then it’s considered severe.
Now, just because we can get a number that doesn’t necessarily mean the number is correct. What evidence do we have that these numbers are any good? Well, one method of assessing how good a measurement is, is to look at its reproducibility. And, here you can see a study that was done in our office where we looked at 24 consecutive patients who had aortic regurgitation and a regular cardiac rhythm. We excluded patients, for instance, with atrial fibrillation because with atrial fibrillation the R-R interval is changing and that can lead to variations in the amount of aortic regurgitation. So, when we looked at these 24 patients, what did we find? This is a Bland-Altman plot. You can see we got wonderful correlation between two separate acquisitions. By the way, these values were all obtained automatically with no user input, and you can see that the mean difference between the two acquisitions is less than one ml and that the standard deviation was less than 3 ml. Three ml would be 10 percent of one grade of aortic regurgitation. So, you can see that the reproducibility is truly excellent and this is not just in our lab. This type of reproducibility has been found in other studies that have been published where you can see intra observer variability and Inter observer variabilities are on the order of just a few ml.
So excellent reproducibility — but, just because a measurement is reproducible, doesn’t mean it’s right. And, so therefore the question is how accurate is this measurement? Well, this is a little bit of a difficult problem to answer because the question is what’s the gold standard? Some people would say echocardiography is the gold standard. I would say that echocardiography is likely not as good as MRI at assessing regurgitant volumes. And, so therefore these types of comparative studies, which I’ve shown here, are somewhat flawed because we’re comparing MRI to echocardiography, and while in general, many of these studies show that there’s reasonably good agreement between the two tests, there’s always some difference between the two. And the question then is, which one is correct? These types of comparative studies don’t answer that question. Furthermore, I think there’s a bias in these studies. The ones that tend to get published are the ones that tend to show better correlations than what one might really find in practice. So the question is how can we go ahead and assess the accuracy of this and how significant can the differences be?
Well, I’d like to use an example here to illustrate the issue. This is a 30 year old man who was referred for surgery for aortic regurgitation. And the reason why he was referred for surgery is he had three separate echocardiograms, one that showed severe aortic regurgitation, one that showed moderate aortic regurgitation, and one that showed moderate-to-severe aortic regurgitation. And, because there was not a consistent result, the patient insisted that his aortic regurgitation be assessed with MRI.
You can see here the MRI examination showed a regurgitant volume of just 22 ml per beat, which would be mild. And, this is a patient whose MRI occurred the day after the echocardiogram that showed the aortic regurgitation to be moderate-to-severe. So, not a very good correlation. Now the question is, which test is right? I would argue that if we look at the MRI we see two separate measurements that are very reproducible. But, the question is we still don’t know which one is right. So, do we have any other information from the MRI examination that might help us? Well, we know in patients with isolated aortic regurgitation that there’s a very tight coupling between how severe the aortic regurgitation is and how big the LV volume is. This is from a paper we published several years ago where we saw a very good correlation, an r2 of 0.8, which means 80 percent of the variation of the left ventricular volume can be explained by the aortic regurgitation that was measured.
Where does this patient fall on this graph? He falls right here. So we can see that not only does he have mild aortic regurgitation by regurgitant volume, but we also see that his LV is enlarged by pretty close to what would be the expected amount. And, that people with severe aortic regurgitation tend to have not just larger regurgitant volumes but also larger ventricular volumes. So this suggests that in fact the MRI may be right. Do we have any other information available to us?
Well, there are several indirect methods for quantifying aortic regurgitation. These methods are based on the fact that there are no other valves with significant regurgitation, and that there is no intracardiac shunt. Both of these were true in this patient.
We can see there are three indirect measures that we can use to quantify aortic regurgitation. One compares the LV stroke volume to the aortic flow, another compares the LV stroke volume to the pulmonary artery flow and the last one compares the LV stroke volume to the RV stroke volume. When we look at the report that we were able to get a from the MRI measurements, we can see that the LV stroke volume differs from the RV stroke volume by 24 ml, which is very close to the 22 milliliters that we got from the direct measure of aortic regurgitation. I would argue that now we have another totally independent method that doesn’t measure use any flow data to quantify the regurgitant volume. That is another piece of evidence that leads us to believe that indeed the MRI is correct.
Are there any clinical outcomes data supporting the value of these measurements that we have with MRI? There are a few. This is one study that was published several years ago where they looked at 113 patients with at least moderate aortic regurgitation. They followed the patients for up to nine years and what they found was that a regurgitant fraction of greater than 33 percent had a reasonably good sensitivity and specificity for identifying which patients would progress to needing surgery. And, you can see that, in the Kaplan Meier plot here, if the regurgitant fraction was less than 33 percent, these patients generally did not go to surgery. If it was more than 33 percent, you can see that they eventually did go to surgery. And, there are two other studies that also have looked at clinical outcome as well.
Now, some of these studies have been done on patients who have had aortic valve replacement, and would I just like to talk briefly about assessing these patients. Just because somebody has an artificial aortic valve doesn’t mean they’re not a good candidate for MRI or for quantification of aortic regurgitation. This is a 63 year old man who had an aortic valve replacement and you can see a black jet right here in the proximal ascending aorta, which represents paravalvular leak. This black area right here is his metallic valve. So while we can’t see the valve itself, we can see the leak. You can see the leak, perhaps a little bit better on this gradient echo image here, and on the flow images here in cross section. You can see the systolic flow in the aorta, and also the diastolic paravalvular leak here in white in the anterior part of the aorta. We can quantify the leak. In this case there was a regurgitant volume of 45 ml. And, we can use the velocity information to see the velocity of flow in systole, as well as the higher diastolic velocities related to the paravalvular leak.
There have been a number of studies that have looked at these patients with a paravalvular leak who’ve had aortic valve replacement. What you see is a systematic underestimation by echocardiography in terms of the severity of aortic regurgitation. This may be because the regurgitant jet is just not well visualized, perhaps because it’s eccentric or because it may be obscured by the metallic valve. MRI is probably much better at being able to assess the severity of the regurgitation because does not depend on visualizing the jet.
So to conclude, MRI is a wonderful test for quantifying aortic regurgitation and its effect on the left ventricle. It not only can quantify the regurgitant volume and the regurgitant fraction, but it can also assess the effect of the leak on the left ventricle in terms of the end-diastolic volume, LV mass, and ejection fraction. The regurgitant volume measurement has been shown to be reproducible and accurate and it is able to predict outcomes. Unfortunately, cardiac MRI is underutilized by most physicians in assessing patients with aortic regurgitation.
Thank you very much.Back to Videos