Elastography for Evaluation of Breast Masses - SD

Introduction

Hello, my name is Barbara Kavanaugh.
I'm the director of the Division of Breast Imaging at Thomas Jefferson University Hospital in Philadelphia, Pennsylvania.
I'm gonna be speaking today about elastography for assessment of solid breast masses, and I'll be concentrating on two different types of elastography, one of which I have participated in research in multicenter clinical trials.
Thanks very much.

Elastography for Evaluation of Breast Masses: Learning Objectives

What I'd like for everyone to take away is an understanding of the physical principles of elastography imaging of the breast.
There are different types, and I'd like everyone to understand how we might use elastography in our assessment of breast masses.

Why Do We Want Another Ultrasound Modality?

We know that ultrasound is an excellent modality for evaluating the breast.
It's non-invasive.
There's no ionizing radiation, there's no intravenous injection.
It's very well tolerated by patients.
The equipment is relatively inexpensive and widely available, and when we see something on ultrasound, it's generally extremely easy to biopsy.
But the downside of ultrasound is that it has limited specificity for characterization of solid masses.
The negative predictive value is excellent.
When we use the strict criteria developed by Dr. Stavros in the mid nineties, that gives us a less than 2% rate of malignancy.
But in order to achieve that negative predictive value, we have a biopsy rate that is two to three times higher than in mammography with positive predictive values in the range of 10%.
That gives us lots of benign biopsies for every cancer we find, and that imposes a burden of morbidity on women.
So with Elastography, we're hoping to improve that a little bit with low suspicion masses that nonetheless do not meet the strict benign criteria.
We'd like to use elastography to obviate our need to do either short interval follow-up or biopsy, and even better, we'd like to take that one to 2% of lesions that turn out to be malignant when we think they're benign.
We'd like to see if we can increase our detection of cancer in that tiny subpopulation of benign appearing masses that actually are malignant.

Principles of Elastography

Now elastography is kind of like ultrasound palpation, in other words, as with clinical palpation, where we are assessing the relative firmness of normal and abnormal tissue.
With elastography, we're using ultrasound to assess relative firmness of a lesion against background breast tissue.
Now, the principle underlying all assessment of tissue stiffness is the idea that abnormal or malignant tissue is going to be more stiff than benign breast tissue.
Now, clinical palpation is subjective and elastography we are hoping produces a more quantitative assessment of tissue firmness.
How do we do it?
Quite simply, you apply force to the tissue with either an external device, typically the transducer itself, or you can apply that force to the tissue internally by sending some sort of a vibration or pulse through the tissue.
And again, typically using the transducer itself.
Once you have applied force to the tissue, you simply measure the tissue displacement induced by the force.
And based on the degree of displacement, you can estimate tissue stiffness.
The imaging modality that you can use to do this can be ultrasound or MRI or optical imaging.
Most typically ultrasound is used for elastography.
There are two basic types.
Static elastography utilizes manual pressure that's applied with the transducer.
ShearWave elastography produces tissue deformation by using a high intensity ultrasound pulse transmitted into the breast by the ultrasound transducer.

Static Elastography

I'm gonna start by discussing static elastography because at the moment that is the most commonly used method for breast elastography, the most commonly available modality on different ultrasound machines.
Now, the method of producing the elastography is the same for all of these machines in that the pressure is applied by the transducer itself.
But in the past, the research has looked at several different ways of measuring the tissue deformation.
There are many different ways I'm gonna concentrate only on two more widely available methods of measuring tissue deformation.
And we'll look at the research in both of those.
Just diagrammatically static elastography is simply manual pressure applied to the breast over a lesion.
And once the pressure is applied, the degree of deformation of the lesion is measured.
Maybe it'll squish out only a little or maybe it'll squish out really a lot.
And that's a relative measure of its firmness.
And this has been around for a really long time.
Actually, the initial apparatus being very simply a transducer applying force through an opening and modified mammography compression plate with the very early ELAs grams showing that soft squishy masses like this fibroadenoma produce no deformation to produce.
No signal on the ELAs because they're soft.
And here is another fibroadenoma, but it happens to be firm producing a signal void on the ELAs agram because it's hard, it's a lot more sophisticated.
Now again, just to review, to perform static elastography, you press with a transducer and the person performing the ultrasound is the one who presses with the transducer.
The pressure has to be very particular, not too hard, not too soft, because you want to be in the pressure range that maintains a linear relationship between the degree of pressure applied and the tissue strain.
If you press too hard, most materials including breast tissue, will start to manifest non-linear properties of tissue elasticity.
And once you lose that linear proportional relationship, you're not gonna be able to easily estimate how hard something is based on how hard you're pressing on it.
And for that reason, static elastography can be subjective.
The results are somewhat user dependent and they are difficult to reproduce from person to person and from lesion to lesion.

Lesion Size Comparison Technique

I'm gonna take a look first at the lesion size comparison technique for analyzing the degree of tissue deformation that occurs with static elastography.
And this very simply, as the name implies, involves measuring the size of your lesion on your B mode image prior to applying pressure, and then measuring the size of the lesion while you are applying pressure.
And then you can grade the lesion based on the ratio of the size pre and post strain.
And there are computer programs that will then generate an ELAs gram based on these measurements.
So here's a fibroadenoma, and the second picture shows you that same fibroadenoma while the sonographer is pressing on it.
With the transducer, it presses out a little bit.
You can see visually it's a flatter and wider and the resulting ELAs gram is fairly light, which tells you that it's soft.
Now, in order to accomplish this analysis of the size difference pre and post strain, someone has to draw a region of interest around the mass.
And this is where a little bit of user dependence comes into play.
These are just more grams showing the type of picture you get with something hard where you get a signal void and you're resulting ELA gram.
But here's the pitfall of lesion size comparison technique.
This researcher chose five expert radiologists to each independently analyze given masses with lesion size comparison technique.
So the five are labeled A, B, C, D, and E.
And this lesion, which I believe is a malignancy.
Here we have the strain image and here is the diameter, the perimeter, and the greatest diameters that were assessed by each of the radiologists.
And you can see there is marked variation in how big the lesion is and where each radiologist thought the perimeter was.
And if you have a technique that's depending on the size of a lesion, but there is huge amounts of inter observer disagreement about what that size is, I think it becomes obvious that your results will be difficult to reproduce.
And here is a benign mass.
The inter observer variability is not quite as great as it was for the malignant mass, but it still is substantial.

Combined Autocorrelation Method

Now the other widely used method for analysis of tissue deformation with static elastography is called the combined autocorrelation method.
And this is what's available on some manufacturer's equipment.
This method displays a color map of elasticity relative to the average strain of the lesion against background tissue.
Again, the images are obtained by taking a look at the lesion before pressure is applied.
And then while pressure is being applied, the elasticity image is displayed as a color map and then the lesion is graded based on this image and it's graded from one to five.
We'll go through each of those.
And in this method, the softest tissue which displays the greatest strain, is red and the hardest tissue, which does not display any strain, will be blue.
On average, breast tissue is colored green.

Grading System

Now here is grade one.
This is the lowest suspicion grade for softest tissue.
And in this particular case it's displaying a fibro adenoma.
And the fibro adenoma itself is fairly uniformly green.
Same green as background breast tissue.
And this is just a pictogram of what that looks like.
And again, a fibroadenoma, which happens to be very green, same as the green background breast tissue.
So grade one lesions are considered benign.
Now grade two lesions have a little bit of firm tissue, but not very much.
And the color map is described as a mosaic appearance.
And here's the pictogram showing you that there's gonna be some mixture of blue, but predominantly green and it's a little bit randomly distributed.
Here again, is another fibroadenoma, a little bit firmer, but a random distribution of blue and green in the color map telling you that it's fairly soft.
Most investigators have considered grade two also to be benign.
And another example of a fibroadenoma with a mosaic pattern of blue and green grade three is a little firmer.
And here the pattern is mixed blue and green but not random.
So the center of the mass is very blue and the perimeter of the mass is green matching background tissue.
So here's a mass, this is the perimeter and you can see just the inside of it is blue.
That was LCIS.
And here's another lesion.
This is an invasive cancer and mass itself, the blue part of the mass on the ELAs gram is much smaller than the mass.
The perimeter is green and these are variably assigned low suspicion or biopsy with, depending on the researcher.
Here's another mass, this is grade four.
So the entire lesion is blue, but the breast tissue surrounding it is still soft, it's green and these are always graded suspicious.
And again, grade four suspicious lesion also a malignancy.
The lesion is blue, the surrounding tissue is still soft.
And then grade five, the firmest grade, the surrounding tissue has also become firm.
So here's the pictogram showing you the mass, which is all blue and a rim of blue hard tissue in the surrounding breast tissue.
Here's our mass, which is blue, and this surrounding echogenic tissue is also blue.
And another example of a mass, which is entirely blue and surrounding breast tissue also firm and blue.

Utilizing Static Elastography in Assessing Masses

So how can we utilize static elastography when we are assessing masses on ultrasound?
When we are interpreting the research in this, I want everyone to remember what the goal is.
And remember, the goal is to decrease the rate of intervention for benign lesions, but to maintain or even increase our cancer detection rate.
Now we don't need a elastography for everything.
There are completely straightforward lesions.
Here is a obvious simple cyst.
You don't need elastography to help you with this.
And here's an irregular mass.
We also don't need elastography.
This is suspicious and has to be biopsied.
We're looking at elastography for equivocal masses.
This is a completely ISO OIC mass.
This would go to biopsy because it has no strictly benign features.
And here's a circumscribed ovoid mass that we might follow.
Well, the mass on the left is actually a cyst.
It collapsed when it was punctured.
And the mass on the right is DCIS.
So will Elastography help us avoid intervention here but still pick up this cancer?
And again, we want elastography to obviate short interval follow-up of our BIRADS three masses obviate biopsy of our very low suspicion birads four A masses, and hopefully find these 2% of low suspicion lesions that are actually cancer.
Here's static elastography for low suspicion masses.
These are not birads three, but they would be BIRADS four A.
And the one on the left, which is green elasticity score, one is a fibroadenoma and the one on the right which has an elasticity score of four, it's all blue, was actually an invasive ductal carcinoma.
And here's the other potential.
We have just a mass like area, we would take this to biopsy, but it's all green elasticity score of two.
It's actually a little bit of a mosaic, and that was adenoma versus another mass, which is entirely blue and would definitely go to biopsy because it's a score four.

Clinical Trials for Static Elastography

So how do the clinical trials Back this up in the United States?
The standard of care for breast imaging is that we need a negative predictive value of 98% for any modality that tells us to not biopsy a mass.
So if we have a mass that is even three or 4% likely to be malignant, we still send it to biopsy.
So if we're going to add a modality to our assessment of breast masses, if we want it to obviate the need for biopsy, the negative predictive value of that modality has to be 98%.
And in the clinical trials utilizing static elastography, the negative predictive values ranged from 80 to 96%, which is not quite enough for us to not do a biopsy.
Elastography greatly increased specificity, but it decreased sensitivity unacceptably for our standards.
However, most of the authors were using a cutoff score of an elasticity score three as their threshold to do a biopsy.
So what happens if we look at different scores?
Here's a bar graph of elasticity scores of 111 lesions, and let's look at how they're distributed.
Across all the scores on the left are the benign lesions and most of them are score one or two.
But notice that the benign lesions are still scattered throughout elasticity scores of three, four, and even five.
Similarly, for the malignant lesions, they're mostly score four and five, but there's a substantial number that were score two and three.
However, in this series, none of the malignant lesions had an elasticity score of one.
Another way of looking at the results of the elastography series, there have been several published and most of them had not huge numbers of cases, but on the order of a hundred to 300 different masses, the false negative rate was on the order of nine to 13% in all of them, again, using a cutoff of three as the threshold to biopsy.
I am again showing bar graphs of the distribution of the different ELASTOGRAPHY scores according to whether lesions are benign or malignant.
And light blue is benign, dark blue is malignant.
If you pick an elasticity score of one, the negative predictive value of that is very, very high.
Most research researchers we're using two and three for our standard of care.
That's probably not gonna be high enough.
But again, a score of one with very, very high negative predictive value.
And what happens if we look at assessment of breast masses utilizing both our birads category and ELASTOGRAPHY score?
Well, when we've called something birads three, if it had an elastography elasticity score of one, there were no malignancies.
Of course in this small series, there were no malignancies for any of the birads three.
But very importantly, let's look at the category of four A.
These are lesions that we biopsy, but the odds of them being malignant are probably less than 10%.
These are very low suspicion.
If you utilize your biopsy threshold as a score of one.
In other words, anything above score one goes to biopsy.
But with a score, one you don't biopsy, you're not going to miss any cancers.
And this might be where the usefulness of static elastography comes into play utilizing our visual assessment, which is our birads assessment with the elastography score.

ShearWave Elastography

Now the other type of elastography is ShearWave elastography.
It's a very different mechanism unlike static elastography where manual pressure is applied by the sonographer and while the pressure is being applied, the color map, the elasticity map is generated in ShearWave elastography.
The sonographer simply scans the breast.
There is no need to apply pressure.
You're holding the transducer just as you normally would.
And when you activate the elastography mode, a high intensity ultrasound pulse is transmitted into the breast by the transducer itself.
And then in real time a color map of tissue stiffness is generated.
So the sheer wave is generated within the tissue utilizing a high intensity ultrasound pulse, and then the resulting shear waves are imaged and an elasticity map is generated.
So just to go through the different mechanism of static elastography versus ShearWave elastography with static elastography, the sonographer applies pressure with the transducer.
And while applying pressure watches to see how much lesion deforms or changes appearance, shear wave elastography looks at a completely different property.
A high intensity pulse is transmitted into the breast tissue and that pulse generates lateral shear waves similar to if you see drops of water falling into a pond, instead of looking at what happens to the pond right underneath the droplet, we're looking at what happens to those lateral ripples that spread out, on either side of that pulse.
Now shear waves move extremely quickly through tissue, and this has been what has limited this modality until recently.
You need an extremely high fast system to image these sheer waves.
These sheer waves move at a speed that is related directly to Young's modulus of elasticity.
And young's modulus is a measure of elasticity.
It's actually a ratio of stress to strain.
In other words, for every material including breast tissue, there is a ratio of how much it will deform under stress.
The elastic modulus is expressed as units of pressure in Pascals.
And for those of us who don't want to remember so much physics, very simply, lower numbers are softer tissue, higher numbers are firmer tissue.
And then interestingly, fluid does not propagate shear waves at all.
So any structure in the breast that is fluid such as a cyst will show up as a signal void.
It has no elasticity value.
Here's a phantom with a completely ioic mass within it.
And when the elastic topography is activated, You can see that it shows up.
This happens to be stiff.
It shows up very, very clearly against the background of soft material within the phantom.
Now, sheer wave elastography is also quantitative.
So unlike static elastography, which generates a color map based on a stenographer's pressure, this is a user independent modality that actually provides a quantitative measure of tissue stiffness.
There is a bar at the side of the screen, which gives you the actual measure in kilo pascals of the tissue stiffness within this box.
In general, normal tissue fat and glandular tissue will be generally less than 70 kilo pascals malignant tissue will generally be higher than 100 kilo pascals.

How to Perform ShearWave Elastography

So how do we do transient sheer wave of imaging?
Again, just as a an image to keep in mind, similar to dropping droplets into a pond of water, the transducer generates a high energy ultrasound pulse and then it very, very rapidly images the lateral shear waves that occur as a result of that pulse.
So here's the pulse that we apply.
This is the direction of propagation of the high energy pulse transmitted into the tissue.
But here's what the transducer is then looking at is these lateral shear waves that occur.
So how actually do you do this?
Well, while you're scanning, you identify your lesion, you optimize its appearance on your B mode images, you activate the elastography, which is simply a button, and that high energy pulses transmitted and the images are analyzed instantaneously in real time.
And you can look at the color map of tissue stiffness that results.
Now, unlike the static elastography color maps here, the blue tissue is the softest and then firm tissue is gonna variably be yellow, orange, or red.
So here we have a soft benign mass and a firm malignant mass.
The color map is uniformly blue for benign tissue and it's what we call actually a tie dye appearance, a mixture of yellow, red, and orange for firm malignant tissue.
In addition to just displaying a color map, you can also see that the apparent size and shape of an image will be extremely different for heart tissue as opposed to soft tissue.
So these are two malignant masses, the appearance in B mode.
And then as compared to the appearance on the color map, which is considerably larger and more irregular for both of them.
And another example of malignant tissue with a color map looking quite a bit larger and more irregular than the beam mode image.

Quantitative Assessment in ShearWave Elastography

Now those are the subjective assessments of the mass with the color map itself.
However, you can also do a quantitative assessment.
So you activate, it's called the Q box.
It's literally a region of interest that you can place over any part of the color map that you want.
So if you place it on the stiffest portion, the most red portion, and then place another region of interest against the surrounding normal or blue tissue, you can not simply measure the actual hardest tissue, but you'll also get a ratio of the hardest to the softest tissue and again in kilo pascals and in a very, very reproducible way.

Incorporating ShearWave Elastography into Assessment of Breast Masses

So how do we incorporate sheway elastography into assessment of breast masses?
Again, remembering our goal.
We want to decrease intervention for benign masses, but we don't want to miss any cancers.
Here's a lesion which is, granted it's not equivocal on mammography or ultrasound, but it's an example of how elastography can display malignant tissue.
So this is a large, almost circumscribed mass, very characteristic of very high grade cancers through transmission on the B mode image.
And on elastography, the area around the perimeter of the mass in all of the surrounding breast tissue is markedly abnormal and markedly stiff telling us that this is malignant.
Here is a more subtle mammographic finding.
This is a focal asymmetry, which was a change from prior studies.
On ultrasound, we have a ill-defined very in homogeneous mass mixed echogenic and hypoechoic On color doppler analysis, we see that there's lots of blood flow.
So this is a suspicious lesion.
And on ShearWave elasticity, the appearance is a little bit more subtle.
It's not uniformly blue, however it's mostly blue, but with areas of pale blue and yellow.
So this is not the most suspicious appearance for ShearWave elasticy, but definitely not typically normal.
So it agrees with the memo and ultrasound assessment that this is suspicious.
So How can we use she ShearWave, elastography and ultrasound?
And I'm going to be looking at this specifically for our probably benign masses.
Masses that we would've called birads three or Birads four A.
Can we use ShearWave elastography to downgrade Birads three to a BIRADS two or can we downgrade four A to BIRADS three
For this, we need to understand clearly what a negative elastic gram is and what the negative predictive value of that is.
And in order to accomplish this, there has been a multicenter clinical trial which enrolled 2000 patients at various sites in Europe and North America.
And the collection of data is still underway, but the preliminary analysis of the first thousand cases is very promising.
The specific questions of the trial was to actually look at whether ShearWave elastography can improve sensitivity and specificity.
It also looked at whether the ShearWave ELAs maps and values were reproducible and user independent and if it could also reliably distinguish solid masses from cysts.
So the way this was done at all of these different sites, we identified masses.
We performed our preliminary B mode and color Doppler exams, and then we performed elastography three successive times.
And for each of those times we did a subjective assessment.
We did a quantitative assessment and we also evaluated perimeters and diameters to look at appearance.
And we at Thomas Jefferson University Hospital was one of the sites participating in this trial.
So of the first thousand cases that have been analyzed, it has been found that this is a highly modality.
If you look at similarity scores for elastography and for appearance, there basically was 98.4 agreement within observers about the appearance on the different ELASTOGRAPHY images that we obtained.
And again, here's the quantitative reproducibility.
This was the reproducibility of the quantitative measurement itself was almost perfect agreement.
And from this so far it appears as though shear wavy elastography is in fact user independent and highly reproducible.
So this is preliminary, because the trial needs to be completed.
But preliminarily it looks as though we can utilize ShearWave elastography as a compliment to birads assessment and downgrade BIRADS three lesions to BIRADS two, in many cases, but very importantly also take birads four masses and downgrade them to BIRADS three, but also of birads three lesions that we would've followed.
If they have an equivocal ShearWave score, we upgrade them to BIRADS four.
So the sensitivity and specificity may really improve dramatically with this.
So here's how we might use this.
Here is a low suspicion mass, but it's not fully encapsulated.
So we would call this BIRADS four A and we would bi biopsy this, but the sheer wave color map is uniformly blue.
This is a benign color map.
And again, a couple more examples of masses that are not qualified for birads three assessment.
They still would go to birad biopsy, but they're quite low suspicion with a uniformly blue color map, these could probably safely be followed.
Here's another mass.
We know that echogenic masses are benign, however, they need to be absolutely uniformly echogenic.
And this mass has a hypoechoic region within it.
So we would biopsy this and here the color map is completely blue, so potentially we could use ShearWave elastography to convert this to simply a birads three or if this does turn out to be a very robust indicator of benign tissue, we may even just call this normal.
Here are equivocal hypoechoic lesions in breast two different lesions.
They're similarly ill-defined hypoechoic regions.
On the left we have a uniformly blue color map and this turned out to be stromal fibrosis.
On the right we have a tie dye color map, a mixture of green of yellow and orange and red.
And this was DCIS.
Here is a low suspicion mass, but with one irregular margin, this would go to biopsy as a birads four A.
And here the color map agreed with our assessment.
There is pale blue and yellow in the corner that is ill-defined this turned out to be benign stromal fibrosis and pash.
So ShearWave elastography is not going to completely eliminate benign biopsies, but that's okay because if we can eliminate a significant percentage of them, we really are still advancing patient care.

Elastography and Cysts

So how about elastography and cysts?
In the last 10 to 15 years with the improvement in resolution of ultrasound equipment, we are all being bombarded with circumscribed hypoechoic masses that we think are probably cysts, but they don't fulfill criteria for simple cysts and we end up needing to do intervention on them.
So is there any way to use elastography to distinguish among these hypo coic masses, which are cysts and which are actually solid?
I'm gonna take a look at both static elastography and transient ShearWave imaging with cysts.
So we know from clinical exam that benign cysts are often very soft and compressible.
And in fact, on static elastography, these soft compressible cysts will produce negative ELAs grams.
But what we also know from clinical examination is that sometimes cysts are very firm and actually that's when they're palpable.
And similarly with static elastography, which simply measures the firmness or softness of a given lesion, sometimes the cysts are soft and sometimes they show up as hard on the ELAs.
So static elastography can sometimes help, but definitely will not always help.
Now let's look at ShearWave elastography.
Remember that shear waves don't propagate in fluid at all.
So on the color map of the ShearWave, you'll end up with a signal void wherever there's fluid.
So is this reliable?
Here's a indistinct hypoechoic mass, which is a complete signal void on the ELAs.
And here on the more recent equipment are more examples of signal voids due to cysts.
So this is a very common scenario for all of us.
We have lots of scattered simple cysts, and then among them an indistinct hypo coic mass.
Here on the sheer wave image, the color map shows us uniformly normal blue tissue in the background with a signal void where that hypo coic mass is.
And that is in fact a cyst.
Big important question is how reliable is this?
Here is another hypo coic lesion showing up as a signal void against a background of normal tissue.
And another example.
So can we look at transient shear wave elastography and say, yes, all signal voids are fluid?
No, you absolutely cannot do that.
And we have to keep coming back to the fact that breast cancer is very heterogeneous and no single imaging feature can be used to assess a mass out of context of the other imaging features.
Remembering back to this cancer that I showed earlier, this is distinctly abnormal color map in the tissue at the perimeter and surrounding this cancer.
And it's around a signal void.
So cancers can definitely give you signal voids.
The reason is probably because the tissue is so stiff that the sheer wave is propagating so quickly that the transducer simply can't pick the signal up at all, and it shows up as a signal void.
So not all signal voids are cysts.
Sometimes they're malignant.
And is there a way for us to tell the difference?
Well, here is that signal void, which was a cyst, and the background tissue is absolutely completely normal, uniformly blue.
So that's when you can look at a signal void and say, yes, it's fluid.
If the background is normal, it's fluid.
And again, here's another signal void, absolutely normal uniform blue background, that's a cyst.
And here's another signal void with a markedly abnormal background that will never be a cyst.
So a signal void against the background of abnormal elasticity is highly suspicious.
A signal void against a background of completely normal uniformly blue tissue is almost certainly cyst.

Summary

So in summary, using elastography for assessment of breast masses may improve the specificity and sensitivity of ultrasound.
That's important to know what type of elastography you are using because if you're using static elastography versus ShearWave elastography, you're going to be assessing the information differently.
It's also critically important to utilize the elastography results in combination with the gray scale image characteristics.
Elastography can potentially avoid biopsy when you have a low suspicion gray scale appearance, but it should probably not even be utilized if you have a highly suspicious gray scale image characteristic.
And as compared to static elastography, ShearWave, elastography appears to be more objective and more reproducible.
And in fact, the unique properties of shear waves, since they don't propagate in fluid, may also help us distinguish among the many hypo coic masses that are cysts and distinguish them from solid masses.