Ultrasound Liver Elastography How I Do it - HD

Introduction to Liver Elastography

Thank you.

So we've been involved with liver elastography for some time and I think there's been, especially in the earlier applications or uses of the technology, there's been a lot of problems with reproducible results.

I think the vendors have added new additional things that we'll talk about that make this pretty consistent now, and we're getting very reliable results.

But the key thing is you have to follow an extremely strict protocol to be able to do that.

I have relationships with everybody.

Chronic Liver Disease and Fibrosis

So chronic liver disease is a substantial worldwide problem.

It's major consequences, increasing deposition of fibrous tissue within the liver leading to the development of cirrhosis with its consequences of portal hypertension, hepatic insufficiency, and HCC.

The stage of liver fibrosis is important to determine prognosis, surveillance, prioritization for treatment, and now with the potential for reversibility to follow those patients.

So we now know that the process of fibrosis is dynamic and regression of fibrosis is possible with treatment of the underlying conditions.

Limitations of Liver Biopsy

Previously, the only method of staging the degree of fibrosis was liver biopsy.

And liver biopsy is really an imperfect histological reference standard.

It does assess fibrosis and as well as great steatosis necrosis and inflammation, which we cannot yet do with ultrasound, but we're making some progress that we may be able to do that in the next several years.

Liver biopsy is invasive with severe complications and up to 1%, and again, it represents a very small portion of the liver volume.

And if you look at pathology studies, the kappa value, how often do the hepa pathologists agree it's, some of the studies are 40%.

So when we try to take our elastography data and match it up with histology data, if the histology don't agree more than half of the time we're, we're not gonna have good correlation.

So I'm hoping that at the end of the talk you'll see we're presenting a different method of using to analyze this and I think our hepatology colleagues are moving to that direction also.

Clinical Stages of Cirrhosis

So from a hepatologist standpoint, they can think cirrhosis consists of two distinct clinical stages.

We go chronic liver disease to compensated cirrhosis, to decompensated cirrhosis and death.

And it's decompensated Cirrhosis is an easy diagnosis.

They usually don't need us to tell them that these are patients that present with variceal hemorrhage, ascites, encephalopathy, and jaundice.

Their mean survival is approximately two years compensated.

Cirrhosis on the other hand, is very difficult to diagnose clinically.

Their mean survival is greater than 12 years.

And again, what they would like us to do is find these patients at these earlier stages where their mean survival is quite higher and try to prevent them from getting into the decompensated state, which has a rapid progression to death.

The other thing that they would like to know if the patient has varis or no varis because if they have varis that triples the mortality one year mortality.

So these are things that we would like to be able to do from our ultrasound examination to guide the hepatologist in treatment.

Elastography Techniques for Liver Fibrosis

If we look at elastography for liver fibrosis, there are multiple techniques, strain, which has a very limited literature and really looks at just the variability of the liver echoes.

It's really not recommended in any of the guidelines.

1D ShearWave or transient asy, they use the mechanical push.

You may know that as the fiber scan, there's point ShearWave where we're using RFI technology to generate shear waves in a very small about one cc volume.

And we get a number and then 2D ShearWave, which can be done either as a single shot where you get one image or now several vendors have real time that every second or so you get another repeat measurement.

That's over a much wider field of view that you can then put an ROI in.

And this allows us some more reproducibility 'cause we can see the variability in the liver.

We can see where there are artifacts and we'll talk about that as we progress with the talk.

And then there's MR elastography, which I'm really not gonna cover, but is out there, but is obviously more expensive and more time consuming.

Understanding Stiffness vs. Fibrosis

Key things to remember is that elastography does not measure fibrosis.

It measures stiffness.

And the stiffness is a combination of fibrosis, inflammation and, and other things.

So remember that, you have to figure out what your patient's condition is because you may overestimate the degree of fibrosis and we'll talk about things that can do that.

And again, anything that increases the hepatic pressure is going to elevate the liver stiffness and artificially give you, if you're using cutoffs, an elevated level of fibrosis that's not there.

So things like portal hypertension, hepatic congestion, or increased blood flow from food digestion increase stiffness of the liver without fibrosis.

So again, in those conditions, you're going to overestimate the degree of fibrosis if you use cutoff tables.

And inflammation is another thing.

So again, looking at liver function tests, if they're markedly elevated, you're gonna always overestimate the degree.

So this is not a very good test to use in patients with acute hepatitis.

It's really meant for chronic hepatitis without an inflammatory response.

Point ShearWave Elastography

So again, points your wave.

We're using RV acoustical radial force impulse, which is just an ultrasound pulse that's tailored to create a momentum transfer.

So the pulse is generated to apply a lot of energy to the tissue and make the tissue to move.

It has two effects.

One, it does cause the tissue to move in the plane of the RFI pulse, which is strain imaging, which we're not gonna look at for liver, but it also generates shear waves, which isn't, you can think the RFI pulse is a stone as you're dropping it into water.

And then we, the ripples are the shear waves and we're gonna measure the speed of those shear waves and that's proportional to the stiffness of the liver.

We can systematically select different parts of the liver, to do in fiber scan.

You're limited.

You don't know where you're taking your measurements.

2D ShearWave multiple measurements using this RV technology over a larger field of view, it can be done as a single image, as a single shot or performed in real time.

Many of the vendors now have real time and we'll talk about that.

I think it does help to get you more reproducible results.

It's real time imaging so you can avoid masses and large vessels.

And I'll show you that at least in the 2D.

And a lot of the vendors now have a quality map that tells you areas that there are artifacts.

So it's flagging those for you so that you can get more consistent results and avoid pitfalls.

It's a color coding so you, we can color code and thankfully all of the vendors use the same color map.

So you can look at the image and you can see how uniform the liver is.

You can pick out where there may be some artifacts.

And again, the ROI size is not fixed in all these machines.

You can have a larger, smaller ROI to fit what you think is appropriate.

Optimizing the Examination Protocol

So this is a slide from the SRU consensus that's been out for a couple years and we summarized all of the issues that you need to deal with when you're evaluating liver fibrosis.

So we have pre-test and post-test probability.

So if you have a clinic that you're only looking at cirrhotic patients, you don't want to miss any of those.

So you may want to adjust your cutoff value a little lower so you don't miss any of those.

Whereas if you are looking at everyone that walks in the door and may or may not have liver disease, you don't want to overestimate these and you may want to elevate the cutoff values.

So all of the papers have kind of picked a middle ground of these.

But again, when you review papers for cutoffs, you need to look at is their population the same as yours because you may need to adjust that cutoff for your population.

And again, we'll discuss how to deal with some of this stuff a little bit later in the talk.

But then we've got multiple etiologies that all lead to the same liver fibrosis and onto cirrhosis.

Age, gender, ethnicity and lab tests can affect our results.

We've got patient factors such as obesity, ascites medications and fasting comorbidities, acute on chronic disease and vascular congestion.

The methods of performing the exam, MR and ultrasound MR uses the bulk modulus.

We use the Young's modulus so their values are one third of our value, so they're not gonna match.

You need to remember that.

And then you've got each of the vendors equipment is slightly different.

And the kiba, the Quantitative Image Biomarker Association, which is a group but the F-D-A-R-S-N-A vendors and some clinicians have been looking at this and we have been using phantoms and figuring out how the systems are variable from each other and giving that information back to the vendors.

And they are modifying their equipment as they do upgrades so that they're becoming more and more uniform.

But I would say right now we're in the 10 to 12% variable difference if you use one machine versus another.

So about up to 12%, some of them are very close, others are not as close to the others.

And again, we have the experience of a technologist and the person reading the examination.

So what's really critical is how you perform the examination.

Patient Positioning

An intercostal approach of the right robe of the liver is preferred.

The patient should rise their head above their head and it's not this, we want them to do this.

So you actually open up the ribs and Dr. Oli says Make 'em a banana so they can really open up the lid.

That gives you a much better acoustical window.

And my thing is the RFI pulse is what really generates our shear waves.

So we need to have the best RFI pulse possible.

So again, you need to have a extremely good B mode image 'cause that's telling you you're getting sound in.

That means your RV pulse is going to get in and generate good shear waves that we can measure.

And also remember that we're tracking actually the shear waves with B mode imaging.

So bad B mode, you're gonna get bad results.

So it's very critical to get extremely good B mode image.

So I like to say optimize, find the best window that you get the best acoustical window on B mode.

And that's gonna tell you you're gonna get the energy in and you're gonna be able to track the shear waves.

Breathing and Measurement Timing

Well the measurement should be taken during a breath hole in a neutral breathing position.

As you take a breath in, you're increasing your right heart pressure and it's actually transmitted to the liver.

So you need to have in your lab a consistent way of looking at the liver as the patient's breathing and coming up that you all do the same thing.

Not take a breath in and hold it, not release it and hold it.

Everybody's gotta do the same.

The recommendation is if you look at the liver, you pick a neutral position, ask the patient to hold their breath.

It's really critical that you talk to the patient before you do the exam and to explain to them when I say stop breathing, don't take a deep breath in 'cause that's the natural response.

So we usually work with the patient to make sure they understand.

Now with the modern equipment, we need to have them hold their breath for 10 seconds.

So it's really not that bad as it was in the past.

Measurement Location

The measurement should be taken in the right lobe of the liver.

The left lobe is unreliable and we believe it's mostly because there's a lot of motion from the heart.

So the left lobe of the liver is constantly moving and you're going to get inaccurate measurements.

And we like to take it between the ribs because you can't put pressure on the liver.

So if we're looking at breast where the breast is above the river, the ribs as we apply pressure, we really change the stiffness of all the tissues because the liver is behind the ribs.

Unless you crack the ribs that you're pushing so hard, you're not gonna really apply a lot of pressure and that's why we don't like you to do a substernal approach because you can often apply significant amounts of pressure through the substernal approach.

You're going to avoid the first one and a half to two centimeters from the liver capsule because we have reverberation artifact in the 2D systems.

Now you can actually see that on that color map.

So the 2D system allows you to avoid that.

But if you're using the point ShearWave system, you don't know where that's at.

So it's really critical that you take and our protocol is to take a two centimeters below the liver capsule.

Another thing to remember, the sweet spot if you will, for all of the vendors where the RV pulse is the strongest is about four to four and a half centimeters from the transducer surface.

So the ideal patient would be that when you apply the RV pulse you would have it four to four and a half centimeters from the transducer, but still two centimeters below the liver capsule.

So obviously in patients with large BMI, that's impossible.

So you should realize, and that's situation, it's going to be harder to get more accurate measurements.

You're going to avoid large blood vessels in bile ducts.

And again, the reason here is these are regular ultrasound waves, they bounce off the vessels and we get an interference pattern.

The vendors have filters that do this, some do it better than others, but you want to try to stay at least five millimeters away.

And you have to remember that this is 3D, not 2D.

So if a vessel is right behind your plane that you're not seeing, you can get an artifact.

And again, these quality maps that we're gonna talk about with 2D help you because it actually looks at those sheer waves and says there's a problem.

And again, the RV pulse should be perpendicular to the liver capsule and that's just so we don't get any refraction off the liver capsule and we get as much energy in the liver as possible.

So you're gonna scan parallel to the rib space, perpendicular to the liver capsule, make sure your perpendicular to the liver capsule in both superior, inferior and right to left planes.

You're going to avoid imaging a depth.

The RRP pulse is attenuated as it goes in liver like everything else.

And the deeper you go, the more attenuation, the weaker the RRP pulse.

The smaller the sheer waves, the more difficult you're gonna have in measuring those.

The more air you're going to have that's gonna be worse in patients with stenotic livers or cirrhosis.

So those are the patients you're really going to struggle with.

And I would say in our practice we have to give up in about 5% of patients that we just can't do them.

They just have so echogenic livers or have such high bmi.

We just can't get enough energy in the liver to get good shear waves to be able to measure them.

Again, we want to avoid vessels, we want to avoid imaging at angles.

And again, when we're angling to the liver capsule, a lot of the energy is refracted off and we often see artifacts in the, about the 10 to 15% of each of the wings of the image.

So doing that, you're going to get bad measurements.

You wanna avoid the first one and a half centimeters to liver capsule because of reverberation artifact.

And this is just a 2D image and here you can see that we've got very high stiffness values in that first one and a half centimeters.

So again, using the 2D technique and you can see these maps, it allows you not to have fall into the pitfalls that we see because here you can clearly see that you're gonna want to do your measurements here or below factors that affect the RV pulse.

Factors Affecting the ARFI Pulse

The amount of tissue displacement again is dependent on the strength of the RRP pulse.

So my whole thing is how do we get that RV pulse?

The A IUM physics people are trying to convince the FDA that they're going to allow us to have more energy.

I think that will be a big help for us in those patients that are very difficult.

The stenotic patient and the cirrhotic patient or the very big MI patient, the RV pulse is attenuated as it travels.

Therefore measurements taking a greater LEP have less signal to noise and again, harder to do the measurements.

And again, the attenuation is greater in a stiffer liver.

So patients with advanced cirrhosis measurements will be more variable and RV pulse can be refracted off the liver capsule.

And we've got some data that we haven't published yet that shows that it's very critical in angling the probe by even 10 degrees can decrease the RV strength by more than 50%.

Importance of Standardized Breathing

And again, I think when you're starting the biggest confounding factor is breathing.

You have to control this.

When lab was having variable results and then when we talk to them, every tech was having the patient hold their breath differently.

And this can make a big, big change.

So you have to in your lab, come up with a protocol.

Ours is to try to find that neutral position and I think that's probably the best.

Some people have the patients do a little bit breathing in and hold it.

But whatever you do, you need to be consistent throughout your lab if you're gonna get consistent results.

And again, we talk to the patient and we make sure they understand what we're talking about before we do the examination.

Again, I think now with the 2D realtime system, you've got a B mode image and you can see the liver moving up and down.

You can see when you say hold your breath that the liver moves a lot and you know they've taken too big or too small of a breath or breath in or out.

So that again is very critical to do and I think that's the hardest part to doing this uniformly.

So this is me and you can see I have a nice thetic liver.

But the upper image is me doing a neutral breathing and the one on the bottom is me doing a Val Salva.

So I went from a normal stiffness value to a cirrhotic within a couple seconds by doing a Val Salva.

So breathing is very, very critical and you really need to come up with a standard way in your lab that everybody is doing it.

The same.

Additional Variables Affecting Shear Wave Speed

The same variables that will affect the sheer wave speed acute on chronic conditions and any inflammatory, any increased right heart pressure is going to give you increase stiffness values.

And if you use a table to read these, you're gonna overestimate the degree of liver fi fibrosis, multiple disease ologies, postprandial.

When you eat, you increase blood flow through the portal vein, you increase the stiffness value of the liver.

We usually like to say they need to fast for about four hours.

The other thing is our techs always say, oh, the patient had a coffee before they came, so we're gonna cancel.

So my policy is we do everybody even if they ate, because if they're normal, they're normal, we don't have to worry about it.

It only eating will only elevate the liver stiffness.

So if they elevate, if the stiffness is elevated, then we put a disclaimer in the report that says Patient eight before the exam, which may artificially elevate the liver stiffness.

And we let it up to the clinician to decide if they want to send the patient back.

Remember not to sample near or by vessels, again, the 1.5 centimeter from the glycerins capsule.

Don't sample at the edge of the sector.

Avoid rib shadows and not only rib shadows that are really shadowing dense that you don't see anything, but you may have some areas if you look at the B mode image and you notice there's a little bit of decrease in attenuation, that's some refractive shadowing from the ribs.

And again, that refractive shadowing is gonna refract the shear waves and you're gonna get bad measurements.

So you really need to have a very good bemo image and look and make sure that you don't have any, even a small amount of shadowing.

Don't use the left lobe of the liver.

And again, we talked about motion and congestive heart failure.

So this is our protocol, perpendicular to the liver capsule, one and a half to two centimeters.

We put the top of the box there.

And again, to have everybody in your lab doing exactly the same thing will really help to improve the reproducibility of these measurements.

Other things that may artificially elevate the stiffness, extra hepatic ccy stasis use of beta blockers.

'cause they may change the portal pressures.

And again, as I showed you, doing a val Salva can be very critical meters per second versus kilopascals.

Units of Measurement

So the FDA would prefer us to use meters per second because all of these systems measure everything in meters per second.

We use a conversion table with some assumptions to convert things to the kilopascals.

Everybody does it the same way.

So at this point in time it's very easy to do that.

Unfortunately, all the hepatology literature is in kilopascals and most insurance companies don't understand meters per second.

They will only accept kilopascals.

So we give both in our report.

And like I said, our hepatologists usually prefer the kilopascals giving both allows us to avoid those kind of issues.

Vendor Variability and Reproducibility

Again, this is measurements from the kiba and I just wanna show you issues that why you have variable reproducibility.

So this is in a phantom and we had a 1.1, a 2.3 and a 2.9 centimeter stiffness phantom.

And we did measurements at three, four and a half and six centimeters.

And these numbers here tell us how far off from the phantom we were.

So for this vendor, if we had a very stiff 2.9 meters per second phantom and did it at 4.5 centimeters, their value was right on.

However, if we did 4.5 in a normal stiffness, they underestimated it by 15%.

These numbers have gotten much better.

This was from the first round.

And after this bias table is hid somewhere on your machine or in the paperwork that you never look at.

But again using these, the vendors have been able to go back and adjust things.

So we're slowly getting things together.

Right now the last set of measurements suggest that the biggest difference between one vendor and another could be 12%.

And Dr. Oli and I did a study that was published in ultrasound recently that we looked at seven different vendors with 22 patients.

We had four people do them.

And what we found was the machines are absolutely very close in the low stiffness values, they get very different above 15 kilopascals or 2.2 meters per second.

So those people that have cirrhotic livers, the measurements between machines are very different, but they're cirrhotic.

They all say that they're cirrhotic.

So from a clinical standpoint, it's not so bad in the lower values, the machines tend to have much more correlation with each other.

And the reason for this again, is that we have a transducer that has a central frequency and a bandwidth and the higher frequency, they're attenuated faster.

So the average frequency that we're generating shear waves actually changes as we go through the liver.

And that causes a change in the shear waves.

And you'll see in the future machines are coming out now that the machine actually looks at the variable mid frequencies.

And now we can do what's called dispersion that we can see how the ShearWave changes with changes in the RFI pulse, which may give us additional information on steatosis and inflammation.

And that may again provide us a lot of information that the hepatologist would really like to know.

Number of Measurements and Quality Assessment

How many measurements, I have no idea how they came up with 10 for fiber scan.

I think some statistician decided this was a whatever test and we were gonna do 10.

It works.

And I think it's very important.

There were some papers out that experts have been able to get the same result doing six or seven.

So but we recommend that for point shear wave you do 10.

As we move to 2D shear wave that has a quality measure, the w new wilham guidelines says you can do five because now you can see that you're not taking measurements in areas that are artifacts.

So you can do that.

60% of the measurements are good measurements, meaning zero zero XX, X is is you didn't get a measurement or you know, color coding on 2D.

And these are the patients that have a very stiff liver and you just not gonna be able to get good measurements.

So the point here is if you did 10 measurements and six of them are bad, if you did 50 measurements to get 10 good measurements, they're still all bad.

So these are the patients you do and you say the patient's body habit is prevents us from getting an accurate measurement.

So you don't try to do 50 measurements to get 10 good ones because they're probably bad measurements anyway.

And we use the interquartile range over the median or IQR over M of less than 30% or point through three.

That tells us that the data set is very close.

And that's probably indicates a very good measurement.

And I'll tell you I like this because I can evaluate our sonographers and I can evaluate the quality of our labs.

So when we started doing these routinely a couple years ago, we were between 0.2 and 0.3.

Now we're between 0.1 and 0.2.

So with experience, our lab has gotten better.

And I can tell you my six or seven texts, I can rank them as to who does a better job because looking at over time, I can see what their IQs are.

So we have some that consistently get very low IQR and some that actually have higher interpretation.

Interpreting Results and Cutoffs

We like to use likelihood ratios and this is what we use in the SRU consensus.

And you can see that we do very well with normals and very well with cirrhotics, but the middle is very poor.

So that's why we came up with a low cutoff value below which you have a high probability of being normal, a high cutoff value above which you have a high probability of having cirrhosis.

And then this middle range, again, the new systems have this quality map.

And here you can see a velocity map and here a quality map where there are areas of red and yellow are telling you that the sheer wave quality is poor.

So you're gonna do your measurements where you're in green.

Again, this is gonna help you and this is why we're saying you only need to do five measurements because now you have that.

Again, the SRU gave the low and the high cutoff values.

These are here, we need to update these because now they're old and we've got many more vendors.

And hopefully the SRU will come up with a amendment to these.

We still gave the cutoff values based on metaverse scores based on the literature.

And we're moving away from that.

In fact, the WILHAM guidelines do not recommend you use cutoff values, although for the literature we've included those.

So how to interpret the results again below the low value, low likelihood of requiring of cirrhosis may not need follow up above 2.2 meters per second in this vendor.

High likelihood of cirrhosis in between we say at risk for progression of fibrosis and should be followed.

Reporting Recommendations

Again, we've got the SRU consensus things that what we recommend, and again, we're using the median value and we use the IQR over median as a quality measure.

So my impression in the report is in the elevation of elevated function test, eating, congestive failure, the liver may be estimated.

So I use that as a thing in my report to flag that it may not be good.

You may want to give the IQR over median, I only give it if it's abnormal 'cause I don't wanna get them confused with a lot of numbers.

But we would say the variance of measurements is large and therefore the atrophy of the measurements is in question.

And then we have these cutoff or macros in the dictation system for vendor A, what the SRU cutoffs, what the me our cutoffs are.

And I think this has really helped our clinicians because some of them want to use SRU, some of 'em want to use me.

Our, so everything is there for them to make their decision.

So again, what we've done is we've got the split into a low cutoff below itch, no or minimal fibrosis, no median treatment followed based on clinical grounds, severe fibrosis, and the mild to moderate fibrosis treatment based on followup.

This is where our harvoni appropriate patients would fall.

And then we've got the severe fibrosis cirrhosis, which now the hepatology literature is moving to, compensated advanced chronic liver disease, which is kind of a mismatch of f threes and F fours.

We're gonna start screening for HCCs.

We can use our regular B mode image to see if we see things that we can say the patient is decompensated and we can use our regular ultrasound examination to decide if there's normal or abnormal portal pressure.

So again, this is what we've done.

We do these cutoffs and kind of split them into three groups.

The what the wilham guidelines is doing, there's something called the BAO Conference.

It's for portal hypertension, but they came up with a very interesting theme that is almost the same as the SRU.

And for fiber scan is, which they use, it's 5, 10, 15, 20, 25 below five year normal five to 10.

You have a high probability of being normal, but do a test to prove it.

Between 10 and 15, you have a probability of being cirrhotic do a test to prove it.

And then above 15 you have cirrhosis.

Above 20 or 25 you have clinically advanced portal hypertension.

So it's a very nice way.

And what they've basically done is taken the SRU in that big group in the middle, split it in half and said, do another test to prove if they're low or high.

So I think we're moving into that and that's what the new Wolf M guidelines that are done and should be out and available very shortly.

Conclusion

So to conclude, detection of significant fibrosis and cirrhosis, important for diagnosis, determination of treatment, prognosis, and follow-up of chronic liver disease.

The literature supports the use of non-invasive tests.

And in fact, in Europe they did a study and now the UK Kingdom actually advises that liver biopsies for chronic liver disease not be done.

And they use this technique except in rare situations.

Again, strict protocol is very critical and you need to make sure all of your sonographers are doing things exactly the same way.

And again, both patient and scanning factors can affect results.