Beyond Morphology: Quantitative Ultrasound in the Liver - SD

Introduction to Quantitative Ultrasound in the Liver

Hi, I'm Bill Lee from UCL University College London in England, and I'm gonna talk to you today about quantification in the liver.

The title of my talk is Beyond Morphology, quantitative Ultrasound in the Liver.

Conventional Ultrasound for Diagnosing Cirrhosis

In the past, we've been quite good at diagnosing cirrhosis from normal liver, just using conventional ultrasound, particularly using high frequency probes applied close to the surface of the liver.

We can quite clearly see the difference between normal and abnormal, using the definition of cirrhosis as nodular change and regeneration within the liver.

Regenerative nodules show quite clearly and allow a fairly definitive diagnosis, but we can't do fibrosis.

Ultrasound Contrast Agents and Nodular Patterns

There is an observation that we made a few years ago using certain ultrasound contrast agents in this instance, the original contrast Leave Vista, where we can enhance the nodular pattern for reasons known not to us.

Introduction to Elastography

Recently we've seen the introduction of elastography in various forms.

A compression of manual elastography has been difficult to apply in the liver because of the intervening abdominal chest wall, but the recent development of AFI acoustic radiation force imaging has changed all of that.

How Acoustic Radiation Force Imaging (AFI) Works

If we look at these SLE and photographs of ultrasound beams, we can see How The standard beam produces a sharp focus at some depth predetermined within the liver.

Now, it's been noted that the force of the acoustic pulse can actually produce displacement within tissue.

These displacement are very small of the order of a 10th of a wavelength, or a few tens of microns.

But using tracking techniques similar to color doppler, we can detect these displacement and show them as an ela.

So by applying a push pulse, which you can see here through the acoustic field, immediately followed by a detection pulse, line by line, we can build up a full scan across the liver, which is the ELAs gram showing the degree of displacement that we've detected.

This is AFI called Virtual touch Imaging by Siemens, which is the technology that I have the most experience of, and we can see that we can detect hard here, shown as black and soft, shown as white often in areas where we can see very little on the conventional B mode ultrasound scan.

Applications of AFI in Lesion Detection

This relies on relative stiffness.

We're detecting a hard tumor in the presence of soft normal surrounding liver.

Obviously with cirrhotic liver, a hard tumor in a hard liver, we'll show no difference, so we won't get the contrast that we expect.

The main use of this technique is to provide an additional contrast to the normal B mode contrast in the liver showing us lesions that are difficult to see.

Unconventional B mode.

Here you can see a hard tumor in the hard abnormal surrounding liver.

A situation we find often in the absence of cirrhosis, patients on chemotherapy frequently have significant changes to the liver parenchyma, which diminishes the degree of contrast showing this tumor.

Again, similar example where we can see a small tumor hidden in an area of abnormal liver parenchyma.

This technique behaves very much as we would expect for detecting borders.

And here we can see an HCC where on the B mode scan, even though we can detect the tumor very clearly, it's very difficult to define, except in this area here where the exact borders of the lesion are.

Here, we can see on the aphy agram, we can see the continuation of that well-defined border through an area where we cannot see it at all.

On the conventional ultrasound scan, this is a liver abscess which behaves in exactly the opposite way.

These can often be very difficult to distinguish from tumors on bemer ultrasound.

And historically, we have applied a needle biopsy to these rather to frequently, But we can see here How The central part of the tumor is bright on the ELAs agram indicating softness and the surrounding liver parenchyma here is stiff compared to the background of a parenchyma indicating edema in the periphery of the abscess.

So a useful way of discriminating in a qualitative sense between hard tumor and soft necrotic liver tissue.

AFI in Post-Ablation Assessment

The technique of AFI was originally developed by physicists at Duke University, where the main aim was to try and determine if they could use this to establish liver necrosis following RFA and other local ablative techniques.

In my experience, this has been almost impossible.

Here you can see an HCC immediately after RFA in the liver.

There's really no discernible change on either the aphy or the B mode ultrasound.

In practice, we find ultrasound contrast the most powerful technique.

It shows us exactly where the zone of necrosis is, and we can map that very precisely to the enhancing tumor.

And if we see enhancing nodules, we know we have to go back to perform more RFA.

Conclusion on Virtual Touch Imaging

So in conclusion, the virtual touch imaging, the Siemens trade name is useful as an aid to lesion visualization.

It provides an additional form of contrast, but it's unidimensional, and we find it's only a very limited use for lesion characterization, except in the simplest way it will characterize cysts, abscesses, and heman omas.

But between different types of solid lesion, it gives very limited information and we rely on my practice in England, much more on ultrasound contrast to give us this additional information.

But the RFI image is an independent parameter both to the B mode and the contrast in Hans ultrasound.

So all in all, it gives us significantly more information.

Example of Ultrasound Contrast Superiority

This is an example of the use of ultrasound contrast, and it's often it's superiority over CT and MRI in characterizing small focal liver lesions.

Here you can see a lesion that's been detected on CT in the arterial phase.

It's strongly enhancing suspicious for an HCC, but in the portal venous phase, it's still enhancing.

It's not showing early washout.

So the differential diagnosis is between a small HCC or a high flow hemangioma.

Now, here we can see the same result after administration of ultrasound contrast.

And if we look at the video of this, we can see quite clearly that we have a blotchy peripheral enhancement immediately with centripetal filling of the central part of the lesion, the typical pattern of a hemangioma.

The other point to bear in mind this video is not running in real time, is that this entire filling in process in this case takes less than 10 seconds.

And we've discovered that the high speed kinetics of ultrasound contrast often give us information that we cannot get from CT or MRI, unless we've just happened to choose the timing of our phases by luck to be to precisely fit.

So the main use of this technique is to Detect tumor in the background presence of an abnormal liver and to distinguish rather iffy ultrasound features on B mode from tumor and no tumor.

So detection rather than characterization.

As you all know, we can use ultrasound contrast to detect perfusion and quantify perfusion both in normal liver and HCC, and this is going to be an extremely powerful technique in the future, as I'll demonstrate later in the lecture.

Quantitative Extension: Virtual Touch Quantification

Now, this technique of aphy can be extended into a quantitative field, Siemens call this virtual touch a quantification, and here we can see a region of interest box placed within the substance of the normal liver parenchyma.

The way this works is that once again, we input a push pulse, which causes displacement through the liver.

This is a longitudinal wave, which as it passes through solid tissue will generate transient sheer waves, which travel at right angles to the direction of the longitudinal pulse, and can be tracked by these minute displacement as they traverse the region of interest.

And what we're actually measuring here is not the presence or the displacement of the sheer waves, but the speed of the sheer waves as they traverse the box.

So by measuring the time it takes to traverse a given distance, we can calculate the speed Of these sheer waves.

And this is a function of the Young's modulus of the tissue IE its stiffness.

So we have a way of quantifying the stiffness of liver tissue in terms of its young's modulus.

Public Health Problems with Chronic Liver Disease

Now, the reason why we're interested in this is because in Western Europe and in the United States, there are major public health problems growing with chronic liver disease.

The main one of these is HCV, which is really a hidden epidemic.

We've no idea how many people in western countries are infected with the HCV virus, but it's probably of the order of one to 2% in some countries of the world, such as Egypt, up to 20% of the population unknown to be infected.

We also have a significant incidence of alcoholic liver disease, which is rising and with the increase in obesity in our populations, non-alcoholic fatty liver disease is also a substantial problem in Western Europe and North America.

Anything up to 50% of the adult population have a fatty liver, which is obesity related.

We know that 20% of these, even if they drink no alcohol, will progress to develop chronic liver disease in 10% of that population.

This will go on to full blown cirrhosis with parenchymal necrosis and nodule formation.

And 10% of these, which is approximately 1% of the entire population, will die from cirrhosis cause simply by fatty liver.

And this has been well documented in epidemiological studies in both North America and in Western Europe.

So it is a major public health problem.

Need for Noninvasive Tests for Liver Fibrosis

So there is a need in clinical practice to develop accurate noninvasive tests to diagnose and stage liver fibrosis, which is the first stage of chronic liver disease before patients develop full blown cirrhosis.

So we need tests that are accurate in the early stages that can show progression from one stage to the next.

That can discriminate between fibrosis, fatty change and inflammation.

The triad that pathologists can characterize on a liver biopsy.

And if we are looking at large populations, then the test should be cheap, repeatable and reproducible and be performed in any clinical environment.

FibroScan: Transient Elastography

There is already a device which uses a transient elastography, which can perform this task of di of measuring liver stiffness.

This device is called a fiber scan invented in France about 10 years ago, which uses a mechanical piston combined with an A mode ultrasound scan to drive a sheer wave impulse mechanically through the abdominal wall into the liver.

This obviously does not work so well in obese patients, or it works not at all in the presence of ascites.

This is the kind of data that is produced from FibroScan, looking at the stiffness results on the left compared with the pathological staging here using the ishaq staging scheme.

Six grades from normal to full blown cirrhosis.

And you can see there is a gentle exponential curve rising up to full blown stenosis with a lot of overlap between the individual stages, but clear separation between the normal and the more severe stages.

This is an ROC curve of the same data showing quite good results in severe fibrosis and in cirrhosis.

Study on AFI in Liver Disease Patients

We have performed a study in my own institution looking at AFI in approximately 300 patients, with a hundred normal patients and other patients with various forms of liver disease as they typically present in our liver clinic.

Please note that we had only one technical failure to achieve results in these 300 patients, and the technology has improved since these early figures which I produced.

We also had 10 patients in this study with fatty change with BMIs of greater than 40.

Now, there is a major problem in defining a normal population, as we've seen if 50% of the population have a fatty liver.

We can't call patients normal if we simply take them off the street.

So I've defined a normal population for this study as no history of liver disease, recent normal liver function tests and a normal conventional ultrasound scan.

But it's important to point out that even biopsy is not a gold standard.

A biopsy samples approximately one 50 thousandths of the liver volume, and we know that hepatic fibrosis and cirrhosis is not uniformly distributed through the liver.

So the liver biopsy as a method of determining the bulk properties of liver tissue is dominated by sampling error.

Limitations of Liver Biopsy

Here's an example of a cirrhotic liver where you can see quite clearly variations in the degree of nodularity of the liver contour frank scarring here in this area, and nodular regeneration and hypertrophy in the chordate lobe down here.

Vast variations between different parts of the liver would give you enormous variations in the degree of fibrosis.

So liver biopsy is not a true standard.

It's invasive with some risk.

As we've said, there is significant sampling error.

There's also poor agreement between pathologists and this extends to the extent there are at least three different pathological staging schemes that you will come across in the literature.

The ishaq, the vir and nodal, and these are not linear products.

These don't grade the degree of fibrosis.

They are categorical products.

They indicate specific morphological stages such as bridging fibrosis, which is either present or not present.

And because there's disagreement between pathologists in any maltus center study that is performed, we would need to have a reference laboratory with a dedicated group of pathologists working to a common standard.

This is from a study done in my own institution a few years ago, where we actually measured the amount of fibrosis within the liver using optical markers and optical quantification techniques and compared that here with the ishak stage.

And as you can see, as we go from no fibrosis to cirrhosis, if you take the column on the right, you can see in the the first few stages there is very little increase in the bulk of fibrosis as we go from one stage to the next, with greater quantities of fibrosis in the later stages.

So we end up with a sigmoid curve, Which is the true quantity of fibrosis.

AFI Measurement Protocol

So in our study population, we have a subgroup with a very recent biopsy correlation and, and of quantified the fibrosis scores according to the ishak scale, which is what our own pathologists use.

There is some argument about the measurement technique in afi.

It hasn't been completely standardized yet.

But the technique that I have used is to take measurements initially from segment five via an intercostal approach.

You want minimal pressure on the transducer to avoid artifact, a good acoustic pathway, which means a clear image.

We take six measurements, discard the highest and the lowest and average the rest.

Now, normally if we are scanning a normal patient, we will get very repeatable results with little variance and stop at this point.

But if we have a large variance or if the measurements are abnormal, then we repeat these six samples in three other segments so we can average over the liver as a whole.

There are some artifacts in the left lobe of the liver caused by cardiac pulsation, which make it not a good starting point for our measurements.

Study Data and Graphs

This is the curve for our normal population, and as you can see, we had three patients who fit our normal criteria who had unexplained readings of greater than two meters per second.

These All genuinely appear to be false positives as follow-up investigations which exclude biopsy have not shown any underlying chronic liver disease.

So this is not quite a normal curve, but it approaches it.

So this graph compares the normal curve with the curve of the cirrhotic patients.

And as you can see, although there's good separation, there is some overlap in this small area here at a level of about 1.6 meters per second.

This gets even worse if we look at patients with fibrosis or without cirrhosis.

And you can see, as you would expect, there is quite a lot of overlap down at the bottom end of the curve.

This is the same data expressed as box whisker plots, And again, you can see the normal distribution is quite tight with very little variance indeed, as is the fibrosis group, which is ishaq S two to five.

Those with cirrhosis show almost total separation with the odd outlier here at approximately two standard deviations.

This is also a group of patients presenting with abnormal liver function tests, and as you can see, approximately half of them are outside the normal range, which is roughly what one would expect.

This is the same data expressed in terms of numbers.

The mean of the normal 1.1 meters per second, the mean of the fibrosis, 1.7 mean of cirrhosis three point naught, with some differences in the standard deviations, but these differences between normal and fibrosis is highly significant at a P level of P 0 0 5.

This is a composite graph from data from my own institution, but not done prospectively, where we've, I've plotted here the ishaq stage on the bottom of the graph versus the percentage of fibrosis determined by different methods.

This curve shows the percentage of fibrosis determined by our optic chemical means showing a sigmoid curve with very little change at the lower levels.

The purple curve is FibroScan data, again, showing almost an identical sigmoid curve and the yellow triangles of my own AFI data, which again fit these plots very precisely.

So the correlation between the true percent of fibrosis within the liver and the aphy or FibroScan data is an excellent linear correlation.

The problem is with the non-linear nature of the ish AQ stage, which doesn't quite fit the direct measurement of fibrosis.

Comparisons with Other Studies

There have been several other groups who have been studying AFI in chronic liver disease, mainly in patients with HCV infection, and this is from one of them from Romania, where you can see the VIR score here On the bottom of the scale compared with the AFI measurement on the left.

And you can see the same sigmoid curve.

The MET scale has only four points compared with six on the IS aq, but they can be transposed virtually one to the other.

But you can see again, fairly tight distributions except for the cirrhotic patients and this same sigmoid curve fitting the data that I've just showed you.

Further data from Germany published in radiology two years ago, shows you also comparison between aphy and FibroScan.

And although they haven't in this publication compared a directly patient by patient, you can see the scatter plots of the data of the FibroScan on the right and the aphy on the left looking very, very similar.

And the ROC curves show effectively no significant difference between aphy imaging and transient elastography.

And again, this is Sensitivity over specificity with areas under the curve, no significant difference.

So the two techniques, transient elastography or FibroScan and the RFI quantification technique produce comparable results.

Advantages of AFI Over FibroScan

Now, the fiber scan was invented over 10 years ago, and there are now well over a hundred patients papers documenting the results with some very large well constructed series which document the effectiveness of the technique.

It has an approximately 5% failure rate in obesity, ascites, and in acute liver disease because acute significant acute inflammation will also tend to stiffen the liver FibroScan merely produces numbers.

It has no imaging component and it's a dedicated machine, which is of comparable cost to a general purpose ultrasound machine.

It does, however, produce a one dimensional bulk tissue measurement right through the liver, whereas the aphy is a small volume sampling technique.

So in order to map variation within the liver, we need to take many samples, but this is changing.

New technologies are coming which will enable two dimensional aphy mapping to give us the same kind of bulk tissue measurement that we get with FibroScan.

So to conclude that we are producing quantitative results with the aphy technique, which are similar to those produced by FibroScan.

It's less prone to failure, and it is combined with a complete conventional ultrasound assessment.

So it's effectively a one-stop shop.

Handling Unexpected Results and Future Studies

Now, what do we do with unexpected results?

This has been a problem in my own study because I didn't get IRB approval to do liver biopsy on normal patients.

And as we never take biopsies from completely normal patients, they always have to have some clinical indication just doing correlations with biopsies.

We'll take a very long time to produce a normal range.

So in any future studies particularly concerned with screening, we need to chase the false positives to determine if there are really false positives or not.

Other Emerging Technologies in AFI

Now, there are other technologies becoming available, which all work on a similar principle using acoustic radiation force imaging.

There is a French company Supersonic imagine, which produce real time quantitative mapping of tissue stiffness using the AFI technique.

This has been applied mainly to superficial structures such as the breast and thyroid, and it does have a fairly low MI input, which means that penetration into the liver is severely limited.

And here we can see A superficial ELAs, quantitative ELAs produced in the breast.

There's been a very recent study presented only a few days ago by Minish Deani from the Mass General attempting quantification within the liver using this new technique, which is, you can see produces large regions of interest or good bulk tissue measurements.

This is his ROC curve, which is as yet not particularly impressive compared to FibroScan or the other AFI technique, but this is the very first study, very preliminary data, which is really quite encouraging.

So the message here is that there are already good data produced to show that acoustic radiation force imaging can produce quantitative measures of liver stiffness, which correlate extremely well with other techniques.

There are many new technologies coming, both improvements to AFI from Siemens and from other companies using the same technology, which I think will extend this more widely to other tissues than the liver and produce more accurate and reproducible results.

Thank you.