Sonoelastography 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.
And today I am gonna be speaking about the clinical
usefulness of elastography in the assessment
of breast masses.

Common Clinical Scenarios

This slide is
showing a fairly common clinical scenario.
We have a very young woman. She has a small palpable mass.
She's not exactly sure if it's new or not,
and it feels smooth and firm and mobile
and by all accounts it should be a benign fibroadenoma.
And we're faced with this decision every day.
How likely is it to be benign
and does it need to be biopsied?

Similarly, these masses in the slide are all
low suspicion lesions.
We have an ovoid parallel mass.
We have two patients, different patients, each
of whom presented with palpable nodularity that corresponded
to some vaguely in homogeneous breast tissue
without a definite mass.
And those of us who scan breast tissue all the time know
that in homogeneity
of the glandular tissue is extremely common.
And all four of these cases went to biopsy
and became one of the 1.6 million breast biopsies
that were done in the United States last year,
of which 260,000 approximately were malignant.
Now these four cases all were malignant.
We don't want to miss these.
We think that the cost to,
these patients
and to all of us is high.
If we miss malignancies at the same time,
we really don't want to be doing this many benign biopsies.

Overview of Sonoelastography

So today I'm going to be speaking about sono Elastography
for the evaluation of breast masses.
And my intent is to
understand how we can integrate elastography into the
clinical and sonographic evaluation of breast masses
in an attempt to decrease these benign biopsies
but not miss any cancers.

I'd like to disclose that I receive research support from
supersonic imagine So elastography,

Advantages of Ultrasound

primarily because ultrasound
as an imaging test has many advantages.
It's noninvasive, well tolerated by patients.
When we see a lesion on ultrasound, it's very simple
and easy to biopsy it.
We don't use injections or ionizing radiation.
The equipment itself is relatively inexpensive and mobile.
And if we use strict interpretation criteria
for our assessment of benign masses,
we will miss fewer than 2% of malignancies.

Disadvantages and Potential Impact of Elastography

Now, there is also a major disadvantage,
which is low specificity.
Even with highly skilled breast sonographers,
the positive predictive value
of a biopsy recommendation is a about 10%.
That means nine out of every 10 biopsies is benign.
That is a high burden for women
and a high burden for society.

The potential impact of elastography would be to
increase the accuracy of our assessments to
decrease benign biopsies without missing cancers.
And it could do that by potentially even obviating short
interval follow-up of probably benign lesions,
but also identifying within that category
of probably benign lesions, those
that are actually malignant.
And importantly, it could hopefully obviate biopsy
of low suspicion masses.
Masses that we categorize as birads four A.
Those are lesions that have a 10% chance of being malignant.
Additionally, for the many, many hypo masses that turn out
to be cysts, elastography could possibly confirm
that a mass is a cyst without resorting
to aspiration or biopsy.

Principles of Elastography

Now, palpation itself,
physical examination is simply qualitative elastography.
So we're taking something
that has been performed in medicine forever
and using ultrasound to apply a more quantitative,
version of it with sono elastography
or ultrasound palpation.

And some of the earliest work was performed with,
the breast compressed and an ultrasound transducer
being pressed gently against the breast
through an aperture in the compression paddle.
The mass is deformed by this pressure
and then the ultrasound transducer
or the images record the degree of deformation
of the mass in some way.
Very soft lesions as in
this fibroadenoma will become flatter
and wider when they're compressed
than when they're not compressed.
And this measurement can be used to determine the,
benign,
benign or malignancy of a mass.

Types of Elastography

Today we use primarily two types
of elastography strain elastography
where manual pressure is applied to the breast, either
by simply pressing down with the transducer
or by holding the transducer still
and allowing the patient's own physiologic motion
to compress the mass
and the degree of deformation of the tissue
that results is measured.

Alternatively, in sheer wave elastography,
an acoustic radiation force is applied via a highly focused
high energy ultrasound beam.
That beam will exert pressure on the mass
and cause mechanical vibrations
that will then propagate shear waves.
Those shear waves can be detected by ultra fast
ultrasound imaging
and the speed of propagation of those shear waves is related
to tissue stiffness
and resulting map of stiffness can be displayed.

Performing Elastography

So how do we perform elastography?
Well strain imaging simply compares the ultrasound signal
before and after application of slight compression.
So during ultrasound examination, a mass
is visualized, the appearance of the mast is optimized,
the elastography program is activated,
and a color map of the
stiffness
and softness of the tissue is displayed on the monitor.

In this particular manufacturer, the
ultrasound transducer is pressed lightly onto the breast
with other manufacturers.
The transducer is simply held steady
And as in this manufacturer, the
map is a gray scale map.
That transducer is being held steady.
The map will show relative opacity, which means
that the transducer needs to be adjusted
or it will show a gray scale map showing relative stiffness
of the mass against background.
And when the appearance is like this,
it means the transducer is being held in
a satisfactory position.

Another manufacturer utilizes a color map
and as you can see, this phantom is moving.
I'm going to play that again.
When you activate the elastography protocol,
you press gently actually sort
of bouncing the transducer against the breast
and record a cine loop.
There is a series of green bars in the image
that will let you know if the transducer is being
held appropriately or not.
And that will help assure you that the color map
that results is representative of the tissue stiffness
in these color maps, red and green are very soft.
That's background. And blue is relatively firm
and you can see this firm mass against background.

Sheer wave elastography utilizes acoustic radiation force.
So again, the B mode imaging is performed, the appearance
of the masses optimized
and then the elastography program is activated.
The radiation force is
instantly activated
and the color map is again recorded nearly instantaneously.
The color map is going to be displayed as
relative tissue stiffness in kilopascal
and there are actually quantitative measurements available
with softest tissue being in the range of zero
to 70 or 80 kilo pascal
and stiff malignant tissue being anywhere from 80
to 200 kilo pascal.

Interpreting Elastography Images

So understanding elastography images in breast imaging
strain images need to be obtained in a particular way
to be sure that the color map will be representative
of tissue stiffness.
So you're gonna set your region of interest box
to include everything from the immediate subcutaneous fat
down to the superficial portion of the pectoral muscle.
The reason for this is
that the resulting color map will help to determine whether
the force that's being applied is appropriate or not.
The subcutaneous fat should be red and green, which is soft,
and the anterior surface, the visualized portion
of the pectoral muscle should be blue.
So this would be considered a very high quality image.
And in this morphologically indeterminate mass,
which is very soft on the gram,
it's a good probability of a benign etiology,
which in this case was a fibroadenoma.

However, image quality has to be good
or the accuracy of assessment of strain images will suffer.
And at the bottom of the screen is an ELAs,
which was considered non-diagnostic.
The subcutaneous fat is indeed red
and green, which is appropriate,
but the pectoral muscle, which is immediately posterior
to this irregular mass, is displayed in varying shades
of red, green, and blue, which means that the
ELAs Graham may not be entirely appropriate
of all of the tissue.
In addition, there's a large signal void at the bottom left
hand corner of the image.
Now the mass itself is irregular and highly suspicious
and there is no question that it needs to be biopsied.
But in a more equivocal case,
this could affect your ultimate diagnosis.

So in this particular series by Dr. Chang,
these low quality images were considered inadequate
for graphic assessment.
Other images were considered simply low quality
and others were high quality in the series
of images scored low quality, they were only half
as accurate at detecting malignancies.
Then the images scored high quality.
Very important to remember when we are trying
to assess whether a breast mass needs to go to biopsy
or not, the quality of the imaging was strongly correlated
with characteristics of the mass and the breast itself.
So when masses were located superficially
or when the breast thickness was relatively thin,
elastography performed much better than with
masses in a posterior location or with very thick breasts.
In addition, smaller masses tended
to perform better on strain imaging than larger masses did.

Scoring Systems

Once allowing for image quality,
the images will be scored according to
a variety of scoring systems.
The one that is primarily used developed by Dr.
Ito some years ago uses five classifications.
Grade one is a uniformly green red image.
This indicates that the lesion is uniformly soft
and is a score of one.
Score two and three are varying degrees of mixtures of green
and blue scores four
and five indicate a uniformly blue mass score five
having blue stiff tissue surrounding the periphery
of the mass as well as within the mass itself.

A more simplified system was developed by a number
of authors who simply graded from zero
to two grade zero being uniformly green
and uniformly soft grade two being any mixture of green
and blue and grade two being uniformly blue with
or without increased stiffness in the surrounding tissue.

Other authors have utilized various ratios
of stiffness within the lesion against background
or various ratios of area of lesion before
or after compression.
In this series, using strain ratios, a region
of interest can be placed within your mass
and within adjacent subcutaneous fat.
And this was found to also correlate with tissue stiffness.

ShearWave elastography images are similarly obtained.
Once the B mode image is optimized, the region
of interest is placed over the mass.
The elastic gram sequence is activated
and a color map is displayed nearly instantaneously.
The elasticity scale is displayed in the opposite color
spectrum from strain imaging where blue will be very soft,
yellow, orange red will be very stiff.
And on machines,
that are available in Europe
and probably soon to be available in the United States,
the elasticity scale is actually displayed in quantitative
measurements of kilo pascal.

And similar to strain imaging,
ShearWave imaging will display differences in the lesion
characteristics with the gram.
Very soft lesions will be almost indistinguishable from
background tissue on the ELAs, whereas
abnormal stiff masses will be larger
and more irregular on the ELAs than they are
on the B mode image.
Additionally, ShearWave imaging is quantitative, so a region
of interest placed within the mass will actually
directly measure the stiffness in kilo pascal.
And these measurements are displayed on the left
hand side of the screen.

So here are some of the spectrum of appearances of masses
with ShearWave imaging, very soft
masses being hard
to pick out against background on the elastic gram
and other masses having varying degrees of light blue,
two shadings of yellow indicating increasing stiffness
and masses that look larger and more irregular.
With the an ELAs Graham of yellow and orange
and shading into many shades of orange
and red, there is a signal void within this large irregular
cancer, which I'm going to speak about a little later.

Clinical Findings and Research

So authors have reported in varying degrees
of detail about their findings.
But overall the authors report that benign masses have ELA
graphic scores that are softer than malignant masses.
And the authors who have looked at varying strain ratios
and area ratios have also found
that benign masses are softer than malignant masses.

However, in breast imaging we really need
a more specific definition of when
to biopsy and when to follow.
So I took a look at the
Research in a little more detail.

Again, researchers have shown improved ROC curves when they
add elastic graphic assessment to B mode features.
But how that translates into clinical day-to-day breast
imaging is not as clear.

So using these results in clinical practice, remember
that many times interpretation
with B mode ultrasound is completely straightforward
and there is no need to use elastography at all.
When we see a classic simple cyst
that will always be a simple cyst
and we don't need a elastography
to help us call that normal.
When we see an irregular spiculated hypoechoic mass
that is highly suspicious
and that will always need a biopsy no matter
what the ELAs Graham shows,

however many times BMO ultrasound presents us
with diagnostic dilemmas, mostly for masses
that are low suspicion
and in these cases a variety of ovoid,
hypoechoic masses, all of which are cancer.
We would like to biopsy all of the ones that are going
to be malignant, but if possible we would like
to exclude the many cases that look exactly like this,
that turn out to be benign.

Here is a busy slide showing results
of various strain imaging clinical trials.
They all measured sensitivity, specificity,
accuracy and positive and negative predictive values.
The number that is most useful
to us in breast imaging is sensitivity.
How likely to pick up cancer and negative predictive value.
How likely is the tool to exclude the possibility of cancer?

A word about how we practice breast imaging in the
United States for 20 or 30 years.
The threshold to perform biopsy in screening mammography
has been 2%,
and that was a threshold that was set when screening
for preclinical breast cancer was initiated.
Every screening trial has to have a threshold
below which you ignore things and above which you intervene.
And breast imaging simply chose 2%.
And when Dr.
Stavros and his investigators published their work on the
sonographic evaluation of solid masses,
they utilized the same threshold for biopsy.
So it makes sense in the United States
that we take this threshold
that has already been well established in breast imaging
and apply it to ELASTOGRAPHY as well.

And in all of these series, elastography alone
did not perform well enough
to meet our standard for breast imaging.

Now here's an investigator who very helpfully
showed a bar graph of his various malignant and benign cases
and looking just at the cancers on the left side
of the screen, you can see that with an elasticity score
of one that was that uniformly green appearance,
there were no cancers at a score of two, there were five,
and at a score of three, there were two more.

Many investigators have proposed a threshold of three
to four as the elasticity score
above which biopsy will be performed
and below which follow-up will be performed.
But multiple investigators have shown similar results that
with elasticity scores of two
and three, there were numbers
of cancers usually on the order of 10% of their series.
So for us, that won't work.

And here is the results of another investigator
who lumped together the elasticity scores of one
and two,
which is not so helpful for us
because it's starting to look like the threshold of one
to two is very important.
Nonetheless, you can see that in the middle table
tabulating the benign
and malignant results against elasticity score.
These soft lesions score one
and two included a substantial number of cancers.

What we don't know about all of these cases in all
of these papers is what the morphologic characteristics
of these soft cancers were
because we know, as I said,
that a morphologically abnormal mass, a mass
that would be characterized as BIRADS four B or C
or BIRADS five is going to go to biopsy regardless
of the graphic scoring.

So if these false negative elasticity scores had
abnormal, highly abnormal birads appearances, they would go
to biopsy anyway and they would then not be false negatives.
And that's what a lot of investigators have started to do.
Again, a number of studies that tabulated benign
and malignant against elasticity scores
but did not utilize the morphologic assessment
as another criteria.
So for these soft cancers where elasticity
was false negative, we don't know
what they looked like morphologically.

We do know from these various series,
and this is just a different way of displaying the data,
that overall elasticity scores
of one contained very few cancers.
Scores of two
and three started to have increasing numbers of cancers.
What we can also see from this data Is
that the benign lesions, lesions that turned out
to be benign comprised the multitude
of cases in most of these trials.

And here is another investigator increasing numbers
of cancers in score two and three, but none in score one.
And again, none in score one.
So I'm going to concentrate on this elasticity score one
and see if we can add that to our morphologic assessment to
decrease our benign biopsy rate.

Integration with BI-RADS Assessment

And here is a series from Dr.
Cho who is one of the investigators who has started
to correlate elasticity scores against birads appearance.
Now in his series Of about
75 or so cases,
I've tabulated the cases that were birads three
and BIRADS four A.
I did not tabulate BIRADS four B four C or BIRADS five
because all of these are gonna go to biopsy
and I don't think we need to pay attention to
what their ILE elasticity score is.

But in this subcategory of BIRADS four A masses,
none of the cases with an elasticity score
of one were cancer.
As soon as you go to score two, you start
to see some cancers.
So it seems reasonable that using a cutoff of score one
to score two, elasticity score one
with BIRADS four A, do not biopsy elasticity score two
or above, take it to biopsy, is going to decrease a lot
of biopsies without missing any cancers.

And in this particular series, half of their masses
that went to biopsy were BIRADS four A,
the other half were four B, four C, or five.
And of that half that were birads, four a half
of those could have avoided biopsy using an
elasticity score of one.
So although we're applying a very narrow category
of elasticity score against a very narrow category
of birads assessment, we're actually encompassing
the majority of masses that go to biopsy.
And this series, we would've avoided 20% of all biopsies
by utilizing elasticity.

Again, just to show Dr.
Ito's information again, he did not utilize
his birads assessments,
but at a score of one, there were no cancers.

Another series using strain ratio rather than elasticity
score found that strain ratio generally correlated well
with tissue stiffness,
but it did not eliminate all cancers from the BIRADS four.
A group one out of 23 were malignant.
That's an unacceptably high miss rate
and we would probably not correlate strain ratio
with our birads assessment.

Now the opposite has also occurred,
and this is a very important potential contribution
of elastography.
We know that when we categorize a mass as birads three
and send it to follow up,
we have a approximately 2% odds that
that mass might be malignant.
So about one in 50 of them are malignant.
It would be better to find those prospectively
and biopsy them rather than finding them at follow-up.

And in this series, looking at a number of masses that went
to biopsy, three
of the BIRADS three lesions were upgraded
to biopsy based on their suspicious elastography appearance.
And if this holds out in other trials,
this is another useful application
of elastography ShearWave.

Elastography has had limited publication,
but of the 48 masses that were studied in this trial,
28 were birads four.
And of those nearly half had elastography
ELASTOGRAPHY scores in the benign range.
So if Elastography had been utilized in addition
to the birads assessment, half
of these biopsies could have been avoided
and none of them were malignant.

A much larger series is
awaiting final statistical analysis
and publication that's going to include just short
of 2000 masses and promises to show similar results.

Clinical Outcomes and Examples

So the outcome that we hope for is that
of these similarly low suspicion masses,
these are birads four A,
they're not completely circumscribed,
but they're not highly suspicious, both likely to be benign.
Elastography looking benign might take this fiber adenoma
out of the biopsy pool
and put it into follow up, thus avoiding a biopsy
and at the same time show that this similar
four a mass is very stiff
and send this appropriately to biopsy.

So avoid a biopsy of a benign mass
but not miss the cancer elastography
with sheer wave imaging similarly can take a low suspicion
but not typically benign mass.
This is birads four A with a completely soft uniform.
ELAs could appropriately send that
to follow up rather than biopsy.

And here's just another set of cases with
negative ShearWave ELAs grams
B mode appearance is BIRADS four A.
These were all benign at biopsy,
and if elastography is utilized
with a completely normal result, these could be
followed safely without missing any cancers.

Conversely, if a mass has suspicious B mode features,
it needs to go to biopsy regardless of the gram.
In the case on the left with strain imaging, the mass is
irregularly lobulated.
There is a completely indistinct margin along one side.
This is at least birads four B, probably birads four C,
even though the ELAs is a benign score of two,
and many investigators would send this to follow up.
The B BMO appearance by itself requires
that this be biopsied.

Similarly, the case on the right
with ShearWave imaging is an irregular
in homogeneously hypoechoic mass with shadowing.
This is a birads four B or four C appearance.
This would go to biopsy even though the gram
shows soft tissue.
The most stiff area is less than 70 kilo pascal,
which would be classified as benign,
but we don't use gram only.
We use elasticity plus birads appearance,
and in fact, a birads appearance
that is suspicious like this.
We don't use the gram at all. We send this to biopsy.

Applications to Non-Simple Cysts

What about non simple cysts?
These are an increasing problem in practice as ultrasound
technology becomes very high resolution, more
and more fluid filled structures are displaying lots
of internal echoes,
thus appearing complicated rather than simple.
A lot of these go to aspiration,
and if we had a tool to distinguish these hypo cog masses
that are complicated cysts from actual solid masses, a lot
of intervention would be spared strain.

Elastography does show some different characteristics
of cyst of against solid masses,
but early work showed a lot of variability in appearance.
However, a recent series suggested
that there is an interesting aliasing artifact that occurs
with fluid-filled masses.
The appearance of blue and green
and red near the posterior surface of a lesion
was highly suggestive of a cyst in this series
and in ShearWave imaging fluid has an interesting
characteristic in that it does not
propagate shear waves at all.

So fluid in anela ELAs agram generated
with ShearWave imaging will have a kilo pascal value
of actually zero.
It will be a signal void potentially.
This can allow confident identification of
hypo coke masses that are non simple cysts
and spare them from going to intervention.

Here are some earlier grams using using ShearWave imaging
showing the hypo coke mass empty
of signal against a soft background
and another indeterminate hypo coke mass
showing complete signal void against a soft background.
Both of these were cysts
and more recently on equipment that we have been using,
we can see signal void against a uniform blue background.
And this uniform blue background is important to be sure
that the signal void is fluid,
not signal void within some abnormal mass
and has been highly reliable in correlating with a cyst.

Not all signal voids are fluid.
On the left is an irregular hypo coke mass with intense
acoustic shadowing,
and the ELAs Graham shows a large area
of extremely stiff tissue
with a central signal void in the mass itself.
We're not exactly sure why there is a signal void.
One theory is that the center of the mass is so stiff
that the sheer waves propagate so fast
that they're not visible to the transducer.
Another possibility, notice all of the attenuation
that there may be so much attenuation
of sound beam in the center of the mass,
that the ultrasound transducer cannot pick up any signal at
all, whatever.

The reason, what makes the difference
between a suspicious mass
and cyst is that the background in the presence
of cancer is highly abnormal.
On the right is a lobulated hypo mass.
The ELAs shows uniform blue background, no evidence
of stiffness in the tissue,
and a signal void in the middle of the mass.
And this is what you look
for when you're gonna call something assist
Background has to be normal.

Again, a large high grade invasive ductal carcinoma.
This was all solid.
This was not necrotic on B mode imaging on color flow.
There was color everywhere within the mass,
so it's not tissue necrosis.
It's some combination of either attenuation
of the sound beam itself
or sheer wave signals that are so fast
that the transducer can't pick them up.

Summary

So in summary, ELASTOGRAPHY shows great promise
for assessment of breast masses
because it may improve the specificity
and the sensitivity of breast ultrasound.

Elastography results must not be used alone,
however, they must be used in combination
with gray scale image characteristics.
Specifically, it appears that an elastic score
that is very soft may avoid biopsy
only when the morphologic appearance is low suspicion.
In other words, only when the mass is graded
as birads four A.
If a mass is graded as birads four B, four C,
or five, which are more suspicious appearances,
those masses need to go to biopsy regardless of
what the elastic score is.

In addition, elastography has the potential to take lesions
that we already follow, birads three lesions
that have a less than 2% chance of being malignant,
and possibly elastography can find that 2%
that are malignant and upgrade them to biopsy.

In addition, elastography may be useful in distinguishing
the increasing problem that we have
with non simple cysts appearing as hypo masses.
If Elastography can pick out the masses
that are actually complicated cysts from the solid masses,
that will spare a lot of intervention.

Thank you very much.