Recent Improvements in the Sonographic Detection of Ovarian Cancer

Introduction

Greetings.
I'm Dr. Arthur C. Fleischer, professor of radiology,
and O-B-G-Y-N, chief of Diagnostic Sonography at Vanderbilt
University Medical Center in Nashville, Tennessee.
I will be presenting a talk on
the concepts concerning the early detection
of ovarian cancer with ultrasound
and emphasize the new developments in color Doppler
and in 3D
and finally with contrasted sonography.
I hope you find this to be helpful.

Overview of the Presentation

This presentation will cover the recent improvements in the
sonographic detection of ovarian cancer.
I want to give a tribute to Dr. Barry Goldberg,
who has helped guide some of this research
and certainly has helped me in my quest to
develop new techniques in ultrasound.
This is a picture of my sabbatical at Dr.
Goldberg's. We imbibed
and drank ultrasound,
and I again am indebted to him
and his staff for giving me some of the ideas
that 20 years later I am going to present
during this talk.
Basically, this talk is going to cover concepts
concerning improved sonographic detection of ovarian cancer
using morphology, using color Doppler sonography
using 3D,
and also the new development of using contrast ultrasound
in finding differences between benign
and malignant ovarian masses.
Microbubbles can be used off-label
for general purposes in this country,
but is not FDA approved at this point in time.
We hope this will be changed in the very near future
and have microbubble technology available for
general applications.
I am an unpaid consultant to Phillips,
and I am the recipient of an N-C-I-N-I-H grant
concerning contrasted sonography.

Principles of Color Doppler and Power Doppler

Basically, I'd like to cover some of the principles
concerning color Doppler and amplitude
or power color Doppler
and discuss the role of these technologies in depiction
of tumor neo angiogenesis,
and of course, apply this to differentiating
benign from malignant ovarian lesions.
As we know, color Doppler allows us
to understand the physiology concerning blood flow
to the ovary and other structures,
and it is hoped that it improves the
differential diagnosis of ovarian masses over that
based on morphology, morphology meaning assessment of tumor,
wall tumor papillary expressions,
those are morphologic features.

Stages of Technology Adoption

Now, in any new technology, it seems that
there's an early part where it's great,
95% accurate, et cetera.
And then there's an era of
what I call inappropriate enthusiasm, where it's used
for everything and it's then an area
of time of skepticism
where we can understand it has some limitations.
And then finally, in reality, it finds
what its particular niches in the clinical world.
And I would say color Doppler has passed
through these stages
and is now a very useful technique in
differentiating benign from malignant masses
and as well as making the diagnosis of torsion,
which I will not cover in this presentation.

Technical Aspects of Color Doppler

Now, color Doppler depends highly on the angle at which
we examine the vessel,
and I would say not only does the amplitude
of flow in the vessel count in its depiction,
but the angle at which it's interrogated has a very
important role in its depiction.
And rather than quantifying the relative impedance values,
the RI, which is resistive index,
which is systolic minus diastolic over systolic,
or the pulsatility index, which is systolic minus diastolic
over the mean, we really can differentiate
high from low resistance
and also detect when there's reverse flow,
which is a very high impedance to forward flow.
Now, these flows depend on several factors intrinsic
to the tissue that we're looking at.
It depends on the downstream resistance, the vascular tone,
is the vessel constricted or is it dilated?
And finally, interstitial pressure has a very important
role as to what we see with color Doppler.
A lot of tumors have very high interstitial pressure,
and that affects the flow to the center
of the lesion, and it also parenthetically affects whether
or not chemotherapy can get into the interstitial portion of the tumor.

Tumor Vessels vs. Normal Vessels

Now, it was theorized
that if we look at the various vessels,
the one on the bottom is a tumor vessel
that lacks a muscular media,
that there is increased diastolic flow in these tumor
vessels as compared to the vessel on the top,
which has a muscular media
which constricts blood flow.
And the differences in the waveform are significant.
Now, the vessel in the middle is a vasodilated normal
vessel, which can have a very similar appearance
to either normal or neoplastic vessels.
Another way of analyzing the color Doppler signal is
by looking at the area under the curve,
which is a reflection of the red blood cells that are coming
into the field of view.
And with using power Doppler, we can get
even smaller vessels and their flow.
And this is to differentiate the
frequency based color Doppler, which is
very good, but not as sensitive as amplitude color Doppler.

Power Doppler Advantages and Disadvantages

Now, as Fleming Frostburg
showed many years ago that amplitude
or power Doppler has advantages and disadvantages.
It improves our sensitivity,
but it is susceptible to tissue motion artifacts,
and there is no aliasing with amplitude Doppler,
but there's no direction or velocity information,
and there's limited temporal resolution.
And this is going to play into when we talk about tumors
and depiction of small capillaries, that the
Doppler technique is in fact limited relative to
contrast, where you're looking at small capillary flow.

Examples of Color and Power Doppler Imaging

Now, here we have a side by side comparison of
frequency based imaging and power Doppler imaging.
And as you can see in this image,
we have a multiloculated tumor
and we have a few color Doppler signals.
But with power Doppler, we can depict the
vessels in the wall of this hemorrhagic ovarian cyst.
This is another patient whose has a solid mass,
and we can see that there are no central vessels indicating
that it was most likely a benign lesion versus a tumor which
would have central flow.
And we can see this on the power Doppler.
We can see vessels only along the periphery.
Now, basically with that lesion,
we depicted the larger vessels we could see on
the vessel staining compared to the H
and E staining, we could see pretty much
what we were looking at with the power Doppler signal.
This is a 3D representation of a
normal ovary, and we can see that there branching points
and very orderly branching of the small arterials
inside the ovary as shown histologically on the right
that have a muscular media.
Now, compare that to what Dr.
Schoenfeld showed many years ago.
In a tumor, there's arterial venous shunts.
There's a tangle of vessels, as we can see here,
and these vessels lack a muscular media
as shown histologically here.
So one would think that their Doppler signals,
and as we can see with contrast, that their entire flow pattern is different.

Neoangiogenesis and Its Role

Now, neo angiogenesis is a
very important process in normal tissues in ovulation
and placenta in wound healing, for example.
But these are regulated
after they're done, they end tumor blood flow
is not regulated and continues to progress.
Now, if we compare the very small tumor,
this happens to be a breast tumor, a few millimeters,
we can see the very abnormal vessels
that occur very early in tumor development.
And Dr. Folkman, of course, theorized
and correctly so that for a small tumor
to progress from two
or three millimeters to several centimeters, it has
to incite this abnormal vascularity.
And in fact, inside the tumor, there's areas
of abnormal interstitial pressure
and areas of focal ischemia.
We can see this on the slide, that as tumors grow,
they have an increased interstitial pressure,
which affects the vascularity in the tumor.
So in the same tumor, we may be seeing different phases
of vascularity.
Now, color Doppler I think is helpful in looking at vessel density
and arrangement in morphologically suspicious areas.
And less important are the actual impedance and velocities.
And with 3D we can depict the vessel density,
which I'll refer to later in the talk,
and quantify this very important parameter.

Physiology in Normal and Pathologic Ovaries

So what color Doppler is very good at is telling us
what the physiology is involved with the
normal ovary and ovaries that contain masses.
And of course, where there's no follicular development,
the vessels are very tightly coiled,
and the impedance value is very high, as opposed to
areas where there's a vascular arcade in a corpus luteum
where the value, there's increased diastolic flow.
We can see this visually here in this mature follicle
where there's vascularity around the periphery
and a hypovascular space, which is an area
where ovulation is going to occur.
After ovulation occurs, there's formation
of a corpus luteum,
and again, as I mentioned, there's low
impedance flow resulting from the corpus luteum
and the wall of the corpus luteum.
The vessels give a very low impedance value.
And with power Doppler, we're beginning to see the branches,
the small branches from the wall
in a corpus luteum as shown here.
And anatomically shown here is these very tiny vessels
in the wall of a corpus luteum.

Quantitation of Vessels

Now, how to quantitate these vessels has been a subject of interest in several different studies.
And basically it's been shown that there is no magic value
of RI or PI
to determine if something is benign or malignant.
And our study and others have shown that if we
look at an ROC curve receiver operating curve,
our data shows that approximately a PI of 1.0
shows areas of about 80% accuracy.
We can look at other parameters,
whether the vessels are central or peripheral,
and of course, tumors have a more common central
distribution of vessels.
The systolic velocities are difficult to determine
since we can't see the vessels in their entirety.
And it's very hard to see exactly what the angle is.
The impedance values are helpful
and the notch is really not.
And when we put all these together, we come up with ROC values, but it gets very complicated,
and we really need something
a little bit more simple in describing
benign versus malignant color Doppler features.

Importance of Morphology

We must remember that morphology is important.
Morphology or shape is important.
There's a big difference between these two individuals,
and it's based on their morphology.
Now, when we look at benign versus malignant, we know
that smooth walled structures are probably benign.
Papillary expressions and irregular solid masses are most likely malignant.
And when we add color Doppler to that,
the more regularly spaced vessels are more commonly benign.
Impedance high in benign lesions
and velocities are typically lower.
And of course, changes in size is important.
We should make sure that the mass did
is still present after one or two cycles
and not a physiologic mass.
And finally, we need to know whether
or not there's ascites, liver masses or lymphadenopathy.

Examples of Color Doppler in Ovarian Masses

So I wanna show rather than going over these
specific guidelines,
and they are guidelines, they're not certainly written in stone anywhere.
Some examples of what I've seen with color Doppler
and ovarian masses.
This was one of the first cancers we found.
This is about a two by three centimeter ovary
with an irregular solid area based on morphology.
Of course, one would be concerned about this lesion,
and we could see that it extended into the fallopian tube
on the left here.
And here's the operative specimen
with the tube clamped by the hemostat.
And in low power microscopy,
you can see that these are all very abnormal vessels
surrounding the epithelial ovarian cancer.
This is another patient that was being followed
for follicular monitoring.
There are two cysts, two follicles here,
and in this area, the ovary.
There's a solid irregular papillary expressions
with a big blood vessel that has low impedance,
and this was an ovarian, a stage one ovarian cancer.
And this is the low power showing the abnormal proliferation of epithelial cells shown in purple
in this ovarian cancer.
Now morphologically this lesion
has some suspicious characteristics.
There's irregular wall thickening, there's an area
of focal calcification,
and in the wall itself, we can see abnormal blood flow.
This blood flow shows probably from vessels such
as this that have no muscular media.
Now, what was more concerning to me was the cluster
of vessels right here surrounding the lesion.
And these are associated with arteries
that are infiltrative in the fallopian tube in
this ovarian cancer.
This a patient that presented for uterine bleeding,
and we can see that there's a big solid mass
with low impedance flow.
And this is typical of a clear cell carcinoma.
This patient has a cystic area in this ovary as well
as this papillary expressions.
Now, most of the time, papillary expressions are associated
with malignancy or borderline tumors,
and we can see blood vessels in the papillary expressions,
which histologically it confirms
conforms to this on low power
and on high power as areas of
vascular infiltration by this papillary cancer.
Now, we can differentiate the cancers from
benign lesions by showing that the lesion
does not have any significant central blood flow, such
as in this corpus luteum,
and such as in this hemorrhagic cyst that we can see
has some fine linear interfaces
inside the hemorrhagic mass.
And this is characteristic of hemorrhage.
And we can also differentiate tumors from non tumors.
This is a peritoneal cyst by showing no vascularity
around the lesion.

Data from Mid-1990s Studies

Now, our data from the mid nineties showed
that we could be very accurate.
In excluding cancer,
we had a 98% negative predictive value.
But in any disease with low prevalence,
a predictive value will be relatively high.
We had a 83% positive predictive value
in finding ovarian cancers.
And the cancers that we did find
were stage one or stage two.
Now, I must admit that these lesions do show significant variation in different studies.
And some studies which included pre
and postmenopausal women showed various statistics.
If one uses mostly post-menopausal women, the
data, as you can see in the study on the right,
shows much better specificity
and improved positive predictive value.

Clinical Usefulness of Color Doppler

So, is color Doppler clinically useful?
The answer, of course, I think it is very useful
and it's useful as an adjunct to abnormal morphology.
And we really feel that what we're looking for, clusters
of vessels in morphologically suspicious areas.
Now, the advantage of 3D allows us to see the vessel density
and their relative branching pattern as well
as their vascular caliper.
And so 3D is in fact helpful
in looking at some of these tumor vessels.
And I've shown you already that with tumors,
there's a cluster of vessels.
The vessels have very abnormal branching patterns.
And so this is one of our first images showing 3D in a tumor
in an ovarian tumor here.
And adjacent to a patient with a cyst that has very few vessels surrounding it.
So this is the niche type 3D determination.
And we can see again the difference between tumors
and the vessels in cyst with 3D.
We can see the papillary expressions
and we can see sometimes the vessels inside
the papillary expressions.
And here's a patient that has a solid area in this cystic lesion with a large feeding vessel.
And this is at various degrees of rotation.
Now, I hope to show you this on 3D
and distinguish it from a benign endometrioma.
Now this is a 3D with vessels, as you can see here,
that are very small.
They don't course to the center, but very evenly spaced.
Again, as you can see here, this is a benign
lesion of the ovary and endometrioma.
This on the other hand, has a large feeding vessel going
to the center of a solid area, which
represented an ovarian tumor.
And we can be very
specific in differentiating the two.
This is a 3D showing a volume
of a metastatic lesion to the ovary.
And you can see that there are large vessels that
we'll see on the multiplanar image that course toward
the center of the lesion.
And this is apparent here
as showing you all these large feeding vessels into the center
of this metastatic ovarian lesion.
So what we have here are multiplanar images of an ovary with central vascularity.
And here are the vessels inside of the
ovarian lesion showing
abnormal branching patterns.
So here are the vessels inside of the
metastatic lesion.
This is a lesion that has large central vessels,
and as you can see, it's a tubo-ovarian abscess,
but it points out how we can use 3D to monitor
where a needle would pass if
an interventional procedure is done in this patient.
Here's the large ovarian mass with the central vessels
feeding the walls of several small locules
and by sculpting
and doing so-called electronic bloodless surgery,
we can see inside the lesion.
So we took the top off
and are now looking down into the lesion.
We can see the locules that are present inside the lesion,
and these would contain perhaps a few ccs of pus,
but the needle would have to traverse the large vessels to get to them, and
therefore was not a very good idea to try
to aspirate and get the very small amounts of pus inside of this ovarian lesion.
So now we'll go back to where we were
and put the tissue back bloodlessly
and get out of the
tubo-ovarian abscess that I've shown you.
So with 3D we can do multiplanar imaging
and we can see these central vessels inside of
ovarian malignancies.
And this is a gray scale image now using 3D showing
the abnormal branching patterns of the tumor neovascularity.

Angiogenesis and Contrast-Enhanced Ultrasound

Now I'd like to spend some time discussing angiogenesis
and the depiction of angiogenesis with ultrasound.
As I mentioned, tumors
incite a vasculature off the host vasculature
using what's called vascular
endothelial growth factor.
And this produces very abnormal vessels
that in fact have areas of ischemia inside of the tumor.
And you can see this area of ischemia can be very large
and plays a very important role into what we see
with color Doppler.
Now, it became clear that with color Doppler,
we were in fact not seeing the capillaries that are
integral in the tumor angiogenesis process.
In order for us to get a Doppler signal, a lot of these
small capillaries would have to be flowing at the same time, a hundred
or more to give a Doppler signal.
So it was hoped that with the microbubbles,
we can in fact see the small vessels
and come to a better appreciation of perfusion as to find
as blood flow over time in a particular volume.
Now here's an animation showing what I'm talking about.
We would grab a small capillary
and we would image the microbubbles shown in blue through the blood flow inside of a tumor.
So we were interested in this in developing this technology.
And basically we did many years of mice work with tumor implanted tumors.
And of course the microbubbles, as you can see here,
are very small particles.
They can pass through even the smallest of capillaries.
And this is a picture of a microbubble when increasing the
mechanical index, the microbubble can
be destroyed.
This has a lot of potential in that if we
destroy the microbubbles
and then watch as they reperfuse, we can really come up
with an idea of how the vascular bed is how vascular these beds are.
Now this is a video showing a one centimeter implanted tumor.
And by flashing we could see the microbubbles,
as you can see here in white, in the capillaries inside of the tumor.
Now if we varied our flashing sequence, you can see in this tumor,
we could destroy the microbubbles
and watch as they come in,
or as in this case, we do continuous visualization
of the flow within the capillaries.
As the microbubbles perfuse this necrotic tumor,
we can quantitate this activity in our time activity curve, where we can
see the time to peak the peak value
and the washout phase
and the area under the curve
reflects the overall vascular volume.
And this is shown in this image with the tumor
and the time activity curve.
We can also do a maximum intensity projection image
and get an idea of what the vessel branching patterns are.
And this has been termed the microvascular imaging.
Now, tumors have approximately 5% of their total
is made up of vessels.
With this, you can see 35% interstitium about
60% of tumor cells.
And this is a beautiful electron micrograph
of the vascularity in normal tissues
with an orderly branching of arteries to arterioles,
to capillaries, to venules, et cetera.
Now, look at the tumor.
Tumor has a very abnormal vascularity,
vessels that are big and small,
and the transit time to get through something like this
would be much greater,
and that's what the microbubble technology is telling us.
This is an in vivo picture of tumor perfusion showing some of the RBCs inside of the
rat mesentery and the arteries next
to the veins, as you can see there.
But these are this is the level that we're imaging
with the microbubbles.
So what we're doing is we're looking at tumor angiogenesis in the
angiogenesis of tumors.
There's a destruction of the basement membrane
by metalloproteinases followed by a
fibroblast extravasation
through these gap junctions.
And they set up small abnormal vessels.
And as you can see, whether
or not the microbubble passes into the tissue
or not, is related significantly
to the interstitial pressure.
This has implications as to not only how we image them,
but how the chemotherapeutic agents get into the tissue
and fight the cancer.
Now, there are many applications of microbubbles,
including the therapeutic potential
for targeting the microbubble to a certain receptor,
to having drug encapsulated in a microbubble,
improving gene transfection, for example.

Studies on Contrast in the Ovary

Now, getting back to contrast in the ovary,
there have been several studies.
The first study was from Finland where they used Doppler.
And Doppler has the problem of
decreased temporal resolution and blooming.
But they did find a difference in uptake time in
tumors versus non-tumor.
And this is a follow-up of their study showing
that tumors had overall increased intensity and longer dwell time.
This is a study from France, which confirmed that.
Now our recently published study first looked at tumor morphology
and then went
into looking at the vascular areas inside of the lesion.
After we determined
where vascular regions were inside the tumor.
Definity was used in an IV setup,
one microgram per kilogram, followed by a 10 cc
solution of saline.
The images were stored on an offline system
using Q lab,
and the following parameters were quantitative,
the wash in time, the peak enhancement, the washout,
and the area under the curve, which corresponds
to the vascular volume.

Examples of Contrast-Enhanced Imaging

Now here we have a image of a cystic mass.
This is the fundamental image, the two
by three centimeter lesion.
Here's the contrasted image.
This is the microvascular
or maximum intensity projection image.
And as you can see with the contrast flowing in,
we saw no abnormal areas in the wall
of the lesion, and we could determine
that this was most likely a benign lesion.
Now here's the time activity curve.
When we look at this area of interest,
and we can see that there's relatively sharp upstroke,
but relatively little as far as peak intensity.
And this is the 3D of this lesion showing the vessels
in their distribution.
Now, contrast this to this lesion.
This is a cystic lesion with a very irregular border,
and we can see the contrast coming into this lesion,
but as soon as the contrast comes in, it leaves.
So this is a relatively fast transit time, as you can see,
it's decreasing in intensity right now.
And when we look at the time activity curve, we can see
that this lesion has this, we stated
before, a very quick uptake
and very short dwell time, as you can see now, pretty much
the lesion shows very little of the contrast.
And here's the time activity curve showing that high relatively quick uptake and short dwell time.
Now contrast that to this lesion about a two
by three centimeter ovary with a one centimeter cystic area.
Now watch in this cancer how the
contrast comes in and is
very intense over a long period of time.
As you can see here, the
peak intensity is at 27 decibels,
which is about four times higher than the others.
And there's a very long dwell time.
And this was an adenocarcinoma.
This is the ovary on the other side,
and you can see that it's enlarged.
But note how there's marked increased
vascularity in this ovary.
Again, this is a patient
with bilateral stage two ovarian cancer.
And again, what we see is high peak long dwell time.
And this is the lesion itself.
Here's the right ovary with a little cystic area in it,
and this is the left ovary.
Now, when we looked at the actual numbers, the time
to peak was not statistically different in benign versus
malignant, but when we look at the overall peak enhancement,
there's about a two
to three fold difference in benign versus malignant.
And certainly the washout time is about
three fold different.
And the area under the curve is about fourfold different
in these different parameters.

Quantitative Analysis and 3D Applications

So what we're saying is
that we can look at the microvascular,
namely the capillaries,
and make a differentiation
between benign versus malignant.
And the question is,
can we do this looking at the macrovascularity?
And we know for a long time
that we can see in malignant lesions
many more color Doppler signals.
So in trying to quantitate this, we can come up with a
vascularity index, which is percent of pixels
or voxels in 3D with color over the total
or flow index, which is power weighted pixel density,
which relates to the intensity of the signal
or the vascularity flow index.
And we can look at fractal dimensions
and give an idea of vessel branching.
And caliper. We found in our initial work a
significant difference in benign versus malignant
and the vascularity index in the power weighted index.
We also saw a very significant difference.
Now, can we do this in 3D?
And we're hoping that our data will show that
with 3D we can do this if an even better fashion because we have more representation
of the tumor vessels.
As you can see, diagrammatically here we can see more
of the vessels using 3D.
And in fact, if we magnify, we can see areas
of micro aneurysms.
We can see areas of stenosis and abnormal branching patterns
and venous lakes consistent with tumors.
And this shows us a an example of
abnormal branching patterns in an ovarian malignancy.
Abnormal branches, as you can see in this papillary cancer,
and many features
of the abnormal vascularity, as you can see here.
And you can see on this 3D video clip of
tumor vascularity with vessels that go from
dilated to stenosis to dilated
a very abnormal pattern
and a very characteristic pattern of tumor neovascularity.

Summary

So in summary, I believe that the improved resolution
with 3D and with contrast
has some significant advantages over morphology
in making the diagnosis of ovarian cancer.
We can look at of course, papillary expressions, irregular
or focal thickened wall
and presence of ascites as the morphology of ovarian cancer.
But now with color Doppler, we can look at
increased density of the vessels,
abnormal vessel morphology, branching,
and with contrast we can see abnormal washout times and greater vascular volume in tumors.
And I'd like to thank my group at Vanderbilt for
their team effort in working on this topic
and hope that you have success in finding patients with early ovarian cancer.
Thank you.