Focal Liver Lesions: Cirrhotic

Introduction to Focal Lesions in the Cirrhotic Liver

We're going to talk about a topic that I spend a lot of time thinking about and reading about and writing about.

Various people have said to me, we really need help on this topic because distinguishing among these focal lesions in the cirrhotic liver is really challenging.

I would agree with that.

Here's a little article we wrote some time ago about the non-specific to cirrhosis, focal lesions such as cysts and angios, focal fat and confluent fibrosis.

These can usually be diagnosed accurately by imaging.

Non-Specific Focal Lesions

One important thing to know about, we know that hemangiomas can be found in up to 20% of individuals at non-selected autopsies.

Yet it had been my experience that we almost never identified hemangiomas with any confidence on the advanced cirrhotic liver.

In fact, when I talked to our hepatic pathologist, they said, we almost never find hemangiomas in the cirrhotic liver.

So we did a little study to look at this and what we found is that hemangiomas, which after all are again sort of a bag of blood with no real capsule or anything within the progressively fibrotic hard liver, they get squeezed shut and they lose their characteristic vascular supply so that in the advanced cirrhotic liver, these become fibrosis and fail to be recognized as such.

But in earlier cirrhotic patients, you can still use the same criteria, and I'll show you one of those.

At least focal fat can occur.

Again, the key imaging finding is signal loss on opposed phased gradient echo imaging.

Confluent Hepatic Fibrosis

Confluent hepatic fibrosis is one I really want you to know about.

This is present in about 30% of patients with advanced cirrhosis and even more whose cirrhosis is due to primary sclerosing cholangitis.

For reasons unknown to me, it is usually in the anterior and medial segments of the liver.

Usually a wedge shaped area of abnormality in 80% of cases.

As I'm trying to convey by the drawing, there is associated volume loss with retraction of the overlying capsule of the liver.

If you can identify blood vessels within this, they will be crowded together.

Much like in bronchiectasis, the dilated bronchi get crowded within the collapsed lung on non enhanced CT.

This will be a low density lesion.

It tends to not be very vascular, but like all fibrotic tissue, it will show typically delayed persistent enhancement.

Again, that is an indication of fibrous stroma, not malignancy or benignity.

Unfortunately it has non-specific high intensity on T two weighted imaging and that's of no help in distinguishing this from tumor.

So here's a very characteristic non enhanced, arterial and late portal venous phase image.

We have a wedge shaped area of low density on non enhanced CT overlying retraction of the hepatic capsule.

Heterogeneous enhancement, delayed persistent enhancement.

There is no way in the world that this is hepatocellular carcinoma.

Now you could say that a cholangiocarcinoma could have similar appearance, but this is a patient with chronic viral hepatitis and a much more likely etiology for this finding is merely confluent hepatic fibrosis rather than cholangiocarcinoma.

Here's another one here.

Capsular retraction.

Anterior medial segment segments.

Low density on non enhanced CT heterogeneous delayed enhancement.

Here it is on MR.

Low intensity on T one weighted imaging.

Anterior medial segments capsular retraction, delayed persistent enhancement.

So very characteristic appearance.

If you see this combination of findings in a patient with chronic liver disease or known cirrhosis, confluent hepatic fibrosis, not HCC.

Focal Nodular Lesions Peculiar to the Cirrhotic Liver

Let's go on to the focal nodular lesions that are peculiar to the cirrhotic liver.

We're going to focus on regenerative nodules, dysplastic nodules and hepatocellular carcinoma.

Theory of Evolution of Nodules

Dr. Sakamoto and colleagues proposed some time ago a theory of evolution, which has been embraced by most of the academic community interested in chronic liver disease.

Everybody with cirrhosis has cirrhotic regenerative nodules a hundred percent.

Some of those regenerative nodules evolve into low grade dysplastic nodules.

A few of those evolve into high grade dysplastic nodules.

A few of those evolve into well differentiated HCCs and a few of those evolve into overt or moderately or poorly differentiated HCCs.

Now we don't know exactly what percentage of each of these goes on in that evolution.

It is clearly a tiny minority, but it's important nonetheless for us as imagers to identify these because clinical correlation is not going to make the differentiation.

Regenerative Nodules

Most regenerative nodules are not seen by imaging, certainly not by ultrasound or CT MR.

Much better.

Those that are seen on CT are those that are surrounded and set off by low density fibrotic septa, low density on arterial and portal venous phase at least, or they contain excess iron or copper rendering them hyperdense on non enhanced CT or on T two weighted MR.

The ones with excess iron are sometimes called siderotic nodules and those may be detectable.

Detectable on non enhanced CT tend to disappear into the background liver on arterial and portal venous and delayed phase imaging.

I often hear it said by people, cirrhotic nodules don't enhance.

That's not true.

All viable tissue enhances.

They just don't enhance quite as much as the surrounding liver and fibrosis and therefore they disappear into the background.

Liver siderotic nodules are in fact well seen on T two weighted and gradient echo MR.

Larger or vascular or enhancing regenerative nodules.

While uncommon are very difficult to distinguish from dysplastic nodules or HCCs though, I'm going to give you some good pearls as to how to do so in many cases.

Here's a patient who has non enhanced arterial and portal venous phase images.

Let me just concentrate on a couple of these nodules distinctly hyperdense to spleen and underlying liver on non enhanced imaging, not hypervascular.

And then sort of blending into the background liver on portal venous phase imaging, no hypervascular lesions, no washout and I'm going to call these regenerative or siderotic nodules.

Non enhanced CT nodular and heterogeneous liver.

But do I see discrete lesions?

I don't really think so.

I don't see any hypervascular foci on arterial phase nor portal venous phase.

And yet when I do a gradient echo MR, I see thousands of little distinct hypo intense foci.

These are all cirrhotic regenerative nodules visible really only on the gradient echo.

Maybe we can see a couple of these on the non enhanced CT as hyperdense foci.

T one weighted imaging, T one weighted imaging, we really don't see much.

Liver looks a little heterogeneous, but what's normal, what's regenerative nodules?

When you look at T two, you see innumerable tiny hypo intense foci.

Those are this patients innumerable tiny regenerative nodules.

Here's another 61-year-old woman.

You're going to see this same case three times in the next 20 minutes because this same woman, same MR study, has every variety of focal nodular lesion in her liver.

So here we have her regenerative nodules, which are the tiny hypo dense sub centimeter foci best seen on T two weighted imaging.

Going back to other patients, we have a non enhanced CT that shows several many typical siderotic nodules.

One here, that's what, maybe two and a half centimeters in size.

That would worry me.

The lesion is not vascular, it's not encapsulated, it has no fat, it doesn't wash out.

So I'm thinking it's probably in the benign end of the spectrum, like an unusually large regenerative nodule or maybe a dysplastic nodule but not a frank HCC.

We'll have more to say about this.

I would prefer to study this patient further with MR.

Here's multiple regenerative nodules, some of which may be seen as hyperdense foci, a non enhanced CT.

This is another non enhanced section on the same patient.

Now why am I bothering with this?

Let's say we started only with the arterial phase image.

So put this out of your mind.

Would you be inclined to say this was a hypervascular focus?

I would.

And here's a portal venous phase.

And you'd say, that's unusual.

It's persistently hypervascular.

No it isn't.

It's persistently hyperdense because of all of the iron tissue.

Now interestingly, when I moved from Pittsburgh to Stanford at Pittsburgh, our standard cirrhosis CT protocol was non enhanced arterial and portal venous phase in Japan, by the way, it's non enhanced early arterial, late arterial portal venous delayed.

But I don't want to do five separate CT scans through the liver in every patient.

I think that's a lot of radiation.

When I moved to Stanford, I found the existing protocol was arterial portal venous and delayed.

And I had to think about whether I thought that was adequate.

Suffice it to say you need to do either the non enhanced or the portal venous and I actually think it's probably better to go the Stanford way and do the delayed phase imaging because it's better for evaluating washout from HCCs, which is the most specific sign that we'll talk about.

So beware of misidentifying a hyperdense lesion as being a hypervascular lesion.

'Cause if you called me up and said, I'm looking at a small hypervascular lesion in the cirrhotic liver, I'm going to tell you it's either HCC or an AP shunt and this is neither one of those.

We're going to talk more about that in a few minutes as well.

So here are multiple hyperdense foci, non enhanced CT, none of them more than about 15 millimeters in size.

I'm comfortable calling those.

They don't enhance much nor wash out.

I'm comfortable calling those cirrhotic regenerating nodules in the same liver, same study, I see a lesion that has nodular peripheral enhancement, iso dense with blood vessels there and there.

Two of these.

And then there is progressive fill in with the lesions be the enhanced portions being essentially iso dense.

The blood vessels.

I could show you additional series and sections, but it would get too busy.

I'm comfortable calling those cavernous hemangiomas.

But in the same liver, same study, same day, heterogeneously, hypervascular, mass not seen on the unenhanced, maybe a little bit of wash out here.

This is HCC and it just emphasizes the necessity of characterizing lesions on all phases of CT or MR.

The only way you can make sense in and differentiate among these lesions.

Dysplastic Nodules

Now let's go on to dysplastic nodules, which have been called a variety of things including adenomatous hyperplasia.

This is the term you want to use dysplastic nodules.

These are considered pre-malignant though in an article I just read in the pathology literature, they said that the majority of low grade dysplastic nodules never evolve into malignancy.

So the larger or high grade dysplastic nodules are considered to be potentially pre-malignant.

We don't really diagnose these very accurately on ultrasound or CT.

MR is definitely the best way to diagnose dysplastic nodules.

They're typically iso to more commonly hyperintense on T one weighted images and hyperintense on T two, which is exactly the opposite of what we would expect for HCC or other malignant lesions.

They should not enhance much on hepatic arterial phase imaging.

And if we perform the cirrhotic protocol with Visipaque, these should retain EOS because there are functioning hepatocytes and not enough de-differentiation to cause this to be a cold lesion on the delayed scan.

So the problem is that small low grade dysplastic nodules may be indistinguishable from regenerative nodules even on histology.

And this book that I'm working on currently, we have a lot of material on pathology as well.

So I've read that stuff recently.

These are usually less than 1.5 centimeters in diameter with minimal vascularity.

And these are usually again, hypo intense on T two weighted images.

This is a spectrum.

So you are going to see high grade dysplastic nodules that are evolving into HCC.

They're typically going to be large.

They do acquire more arterial vascularity and some of those may become hyperintense on T two.

Suffice it to say that neither you nor me nor anyone else is going to confidently differentiate a very high grade dysplastic nodule from HCC and we just have to live with that.

Here's an older study.

I used to go to all the explantations in Pittsburgh and photograph the explanted specimens, which was good for learning about rad path correlation.

Here's the T one weighted image, here's the T two and we have sort of a clover leaf shaped hyperintense lesion here and another one up here.

And each of those lesions is hypo intense.

On T two we said that the clinicians, that's behaving like a dysplastic nodule, I wouldn't, biopsy embolize, ablate or otherwise treat this.

Just go on to watch his patient when he needs transplantation.

Go ahead and do it.

And here is the explanted specimen and this is that lesion.

And this was called a dysplastic nodule by the pathologist.

Here's a large lesion, hyper intense to liver on the non enhanced scan.

Not very vascular at all.

Doesn't wash out.

Now this suggests a regenerative nodule, but that's getting a little big for me to feel comfortable about just blowing that off.

So we do an MR here.

That lesion is hyper dense or bright on T one does not show signal dropout.

So it's not fat containing and that's important.

It's dark on T two.

It's not very vascular.

This is not washout by the way.

Here's a pitfall that I see.

I think what's happening is the surrounding liver is enhancing more than the lesion.

So the density or the intensity of this hasn't really changed.

So let's reiterate here.

Bright on T one, no signal loss and no focal fat dark on T two minimal vascularity equals dysplastic nodule.

That was biopsied because this patient was actually not close to needing a transplant based on the criteria that are used.

The so-called MELD score and that was confirmed to be dysplastic nodule.

Alright, let's try our hand at this one.

This is that same 61-year-old woman again.

So she has a lesion that is dark on T one, bright on T two heterogeneously hypervascular washes out on delayed phase and new criterion showing up now bright on diffusion weighted imaging.

This is HCC dark and T one bright on T two hypervascular.

This was done with Visipaque, no retained Visipaque on the delayed scan, bright on diffusion weighted.

This is HCC.

How confident are we confident enough that we're going to treat this based on imaging alone?

We're not going to biopsy it.

This will be weighed along with the other MELD criteria as far as does this patient, where does this patient fit on the transplant queue enough that we could embolize or ablate or otherwise treat this lesion?

We are not going to do a biopsy on that.

Here's the same woman.

New lesion left lobe bright on T one, dark on T two.

Not hypervascular.

Doesn't really change much in density whatsoever.

Is not bright on diffusion weighted imaging.

That's a dysplastic nodule.

And by the way, all of these lesions are proven on explanation to be what we call them.

By the way, look at some of the regenerative nodules on delayed phase Visipaque.

They take up and retain Visipaque because they have functioning hepatocytes and they make bile, but the bile can't get out very quickly.

Hepatocellular Carcinoma (HCC)

Let's go on to HCC, which is really the holy grail.

That's what we're trying to find in these patients with cirrhosis.

So in the drawing we've put it within a nodular cirrhotic liver with ascites, it's a hypervascular lesion.

It often has satellite lesions and it has this really bad habit of invading portal veins and hepatic veins.

Heterogeneously hypervascular washes out on delayed phase, usually more than two centimeters in size.

90% of the lesions are hypervascular and show up best on arterial phase imaging and wash out on portal venous phase.

This is as good as or better than a biopsy if it has those findings.

Smaller lesions may have a capsule or focal fat within the lesion to help us identify them.

We may see a so-called nodule in nodule pattern, which I'm going to show you two cases of larger lesions and we should be embarrassed to find these.

These are usually found in the non surveillance population.

So in somebody that's in a well structured surveillance program where they're getting on regular intervals, ultrasounds or CTs or MR, we should be finding HCCs in the two centimeter range.

If we don't, if somebody just shows up with a symptomatic or a lesion found without the surveillance program, they tend to be big lesions.

They sometimes have this mosaic pattern or satellite lesions and they often have invasion of portal and or hepatic veins at the time of diagnosis.

Screening and Imaging Protocols

Now people ask me, should you use CT or MR or ultrasound in screening?

And the answer is we use all three.

And it's an evolving thing.

I would say in most North American institutions we do a lot more of the surveillance with CT we tend to use MR in a problem solving modality to distinguish among focal hepatic masses.

If you're a an aficionado of MR and you want to do MR instead, that's fine.

Personally, if I had to read 15 MRs of cirrhotic livers every day like I do CTs, it would be challenging.

It's got to be multiphasic.

You need an arterial phase image at about 25 to 35 seconds.

Dual arterial or test bolus is ideal but we don't do that portal venous phase.

You want to do it about 60 or 70 seconds.

Again, you either want to do a non-contrast or a delayed or equilibrium phase.

I now do delayed phase imaging.

Very useful to evaluate washout from HCC.

You must have a rapid injection of four or five ml per second and it has to be an adequate volume.

I see outside CT scans all the time where the contrast is being administered at two ml per second and they're giving 75 or a hundred ml of contrast.

You're not going to find HCC unless it's some big ugly, very advanced incurable disease.

So if you want to diagnose curable HCC, it has to be with these techniques.

What about MR and what contrast agent to use?

Well, there's been much more published about the usual gadolinium based extracellular agents, which actually give you the optimal opacification of blood vessels and evaluation of hyper vascularity within tumors and varices and so forth.

And we virtually always do, the MRs with and without contrast in multiphasic things arterial venous and delayed.

And we do precontrast T one, T two and gradient echo imaging.

So here again is sort of an embarrassingly large HCC dark on T one heterogeneously bright on T two heterogeneously hypervascular on arterial phase washes out on delayed phase.

This again is with usual gadolinium agents and it's bright on diffusion weighted imaging.

Absolutely slam dunk HCC every time.

Here is the value of diffusion weighted imaging, which we've only been using for several years, but I've become a big believer in this arterial phase imaging.

I'm not sure I would recognize a lesion diffusion weighted imaging.

They got this little light bulb here and now I go back and look at that and then I more closely scrutinize the delayed phase and say, well I'll be darned.

There is a lesion there, it's hypervascular and it washes out.

That's a little bitty HCC.

What about the Visipaque or in Europe, the rest of the world?

Primovist, so-called hepatobiliary contrast agents.

Well I think the jury is still out a little bit.

I was lecturing over in Japan last year and one of the hospitals I was at, they do every cirrhotic liver with Visipaque.

Of course they also have some big grants from the makers of Visipaque, but I personally find it challenging in some of these cases.

I'm trying to keep an open mind, because I find it's not so easy to interpret.

In many cases it is a great agent to identify FNH.

It's a great agent to find every little metastasis within a non-cirrhotic liver.

I don't have time to talk about that right now, but there have been variable reports about its value in cirrhosis and the problem is this, there are two problems.

Number one, the cirrhotic liver doesn't take up and retain the EOS as well as the non-cirrhotic liver.

So you got all this fibrosis, you got all of these nodules and the underlying liver can be so heterogeneous as to make it difficult to recognize the HCC number two, and it's not on this slide, EO vs is a suboptimal vascular enhancing agent.

And number three HCCs are usually probably a little more than 80%, defects or hypo intense on the delayed phase imaging.

But there are some reports of heterogeneous increased uptake in retention within HCCs on the delayed phase imaging.

Usually well never as homogeneous and bright as an FNH, but enough that the persons interpreting the study we're really uncertain about the significance.

So how good are we in detecting HCC on CT or MR?

Well we should be really good for finding all masses greater than two centimeters in diameter.

We should be really good at describing the morphology and the hemodynamics of the tumors.

But the small really well differentiated HCCs are tough to identify.

They may in fact not be hypervascular.

They may in fact show up best as hypodense lesions on portal venous or delayed phase imaging.

And I'm going to show you examples of that.

I have this underlined in italics because if I see a lesion in a cirrhotic liver, a focal lesion that is encapsulated or contains fat, that is HCC until proven otherwise.

But most greater than 90% of HCCs are going to be seen as hyperdense or hyperintense on hepatic arterial phase CT or MR.

Alright, so let's try our hand at this one.

If I can see it here on CT, the top row we have no detectable lesion on arterial phase heterogeneously hypervascular, but curiously, it's only the right side of this lesion that's hypervascular.

The left side is not the lesion is encapsulated and I've said that that strongly suggests HCC as does the heterogeneous hypervascular.

Now let's look at MR.

The patient's left side has a crescentic area that's hyperintense and T one and dark on T two and we said that's the hallmark of a dysplastic nodule while the other side is dark on T one and bright on T two and hypervascular.

So we have HCC and dysplastic nodule.

This is the nodule in nodule pattern of an HCC developing within a dysplastic nodule.

I love that case.

Here's another one, non enhanced CT.

You got this yin yang sign here about the same size lesion, different patient.

The left side of the lesion is very low density because it's got lipid or fat within it.

On T one weighted image, the right side of the lesion is hyperdense.

The left side is dark and then it's a flip flop.

The left side is bright on T two, the right side is dark.

Okay, so let's do this again.

Contains fat dark on T one, bright on T two equals HCC, The right side of the lesion dark on T one, A little bit bright on T two, not hypervascular.

That's a dysplastic nodule nodule in nodule pattern of HCC within a dysplastic nodule.

But the most typical thing, and again we don't do MR, we probably do 20 CTs for every one MR.

In our cirrhotic surveillance program because most of the time we see something like this.

Heterogeneous hyper vascularity on arterial phase imaging, subtle low density lesion on non-contrast starts to wash out, probably has a little bit of a capsule.

We just tell 'em this is an HCC and we make the decision for management on that basis.

So as a little reminder about the hemodynamics of focal lesions in the liver, the normal liver has about 80% blood supply from the portal vein, 20% from the hepatic artery.

Jumping to the other end of the spectrum, a moderately or poorly differentiated HCC has essentially 100% arterial supply and no portal venous.

So as you move along this spectrum from normal through cirrhotic nodule dysplastic nodule well differentiated cancer moderately to poorly differentiated HCC, you progressively lose portal venous supply and acquire more hepatic arterial supply.

Now that should inform us as to how we interpret these studies.

Here's another patient, subtle low density lesion on non enhanced CT hypervascular on arterial phase image washes out and has a capsule HCC every time needs no biopsy needs no MR.

Just treat it.

Here's another one.

Encapsulated nearly iso dense lesion on non enhanced heterogeneously hypervascular washes out HCC every time to reiterate a lesion with a capsule in a cirrhotic liver is almost always HCC.

What is it in a young woman who's not cirrhotic?

Adenoma Here is a patient who has a lesion that's not hypervascular but shows up very well on portal venous phase imaging.

This is what we occasionally see with a well differentiated HCC because this does not, has not acquired hyper arterial supply yet.

So remember that you have to look at all three phases of imaging or four phases of imaging and the well differentiated HCC may not be hypervascular.

Here is a lesion essentially seen only on the arterial phase imaging a fairly typical moderately differentiated HCC.

So here's a lesion, we look at that and we say, yep, that's an HCC.

Good for us.

We found this when it's only 15 millimeters in diameter, this is great.

It washes out, it's got a capsule, we know it's an HCC, we can cure this.

Nope, we can't actually because what's this portal vein tumor.

So this little bastard has already invaded the portal vein and we have enhancing tumor thrombus within the portal vein that is now a stage four HCC forget transplantation.

Forget ablation.

We can still do chemo embolization but we are not going to cure this lesion.

How do we distinguish bland thrombus from tumor thrombus in this way?

Thrombus contiguous with the cancer that expands the lumen and has enhancement or, and or actually you can actually see the tumor vessels within this thrombus and even the tumor thrombus is heterogeneously hypervascular and washing out some other features.

Ancillary features a big ugly mass in the left lobe of the liver.

It's got some focal calcifications which are rare in HCC and it's got focal fat.

But this is a cirrhotic liver hypervascular foci and fat within a mass in the cirrhotic liver that washes out HCC every time.

So the pearl again, a lesion with focal fat in a cirrhotic liver is HCC.

Focal fat within a lesion in a non-cirrhotic liver is most likely an adenoma.

Alright, a little question here to engage you focal lesion that is hyperdense on non enhanced CT minimally enhances bright on T one, dark on T two is most likely what dysplastic nodule this graph is in your handout and I actually was working the last couple of days to modify this.

This is a very complicated thing to draw out.

I would add two more columns here.

Do this yourself if you would, here to the MR.

So there are two more columns I want you to add here.

One is diffusion weighted imaging.

So right along this axis here.

Next one I want you to, next column I want you to label DWI regenerative nodule negative sign dysplastic nodule negative sign HCC, either well or moderately differentiated bright arrow going up.

And then I want you to add another column over here.

It's going to be a delayed Eovist scan.

Regenerative nodule arrows going up, dysplastic nodule arrows going up either of the two HCCs arrow going down indicating that the lesion is going to be hypo intense relative to the rest of the liver.

If you keep these factors in mind, you should in fact be able to identify with some confidence nodular lesions in the cirrhotic liver.

So how do we do?

Well, we're great at lesions greater than two centimeters.

It is challenging in the screening population to pick up the sub or the one centimeter lesion and so forth.

There's no question we miss some of those in some material from Pittsburgh that we never quite got around to writing up.

We did an interesting thing.

We said alright, we know that we miss tiny HCCs but we have these so-called Milan or UCSF criteria for transplant ability of patients with cirrhosis who may have an HCC.

And without going into great detail, it has to do with the size and number of the HCCs and would maybe paradoxical seem paradoxical to you if you have a small stage one HCC in your cirrhotic liver, you actually get moved up the transplant scale.

So you actually get transplanted sooner than somebody who doesn't have HCC.

And it's appropriate because those patients actually do quite well with transplantation.

So what we did is this, we said okay, let's say it as a given, we're going to miss some small HCCs but we're going to go to all the explants and find out if we gave them the correct preoperative staging based on CT or MR.

And in over 95% of cases we did.

So I think actually we can do quite well and we can guide other therapy well as well.

So an HCC that maybe doesn't need transplantation but needs ablation or trans arterial chemo embolization, we can choose that as well.

MR is probably more accurate than CT.

It is certainly better at differentiating among dysplastic nodules.

We've already done those factors.

Pitfalls in HCC Detection

Alright, I want to finish up with some pitfalls.

FADs, transient hepatic attenuation or THIDs, transient hepatic intensity differences.

The very small peripheral wedge-shaped lesions, centimeter or so in size, they're never malignant.

You can ignore those.

It's either due to AP shunts or aberrant veins.

Larger, more rounded segmental or lobar lesions are worrisome.

In fact, a segmental or lobar are often due to tumor occlusion of the portal vein.

And I'll show you an example of that.

AP shunts are sort of the bane of our existence.

They're very common in cirrhosis.

And here's how I handle those.

If it's small, less than a centimeter peripheral nons spherical, iso dense or iso intense or portal venous in delayed phase, sometimes I'll see visible vessels into and out.

I dismiss it.

We'll just continue to see that patient in the usual surveillance program.

So here we go.

Lobar THAD, the whole right lobe of the liver in a straight line demarcation on arterial phase imaging is hyperdense to the rest of the liver.

This basically tells me the portal vein to the right lobe is being knocked out.

Here's the culprit.

We have an HCC and there that is invading the right portal vein.

You can see the thrombus here.

So if this right lobe of the liver is getting no portal venous supply, in order to stay alive, it has to recruit more hepatic arterial supply.

Another way of looking at it is the hepatic artery is delivering undiluted contrast material to the right lobe to the left lobe.

That contrast is being diluted by opacified portal venous flow.

So that is a lobar THAD think HCC in the cirrhotic liver.

Here's a patient who has a prominent left hepatic artery and an early portal venous branch returning from that lateral segment.

I am pretty sure I caused this 'cause we used to do our liver biopsies subxiphoid under ultrasound guidance and that's right where we do the biopsies.

I'm not losing sleep about that because these never amount to anything.

For the most part they go away on their own and we just say that's an AP shunt.

Probably iatrogenic due to the site of a prior biopsy.

No lesion visible on portal venous or delayed phase.

Here's another one.

Visible blood vessel draining that lesion.

Prominent artery feeding that lesion is sort of a little AV fistula or AP shunt.

Nothing to worry about.

This one's probably not iatrogenic, it's probably just the AP shunt that occurs very commonly within cirrhosis.

Here's another one.

Actually.

This patient had what?

Five more about eight of these lesions.

Triangular.

Peripheral subcapsular less than a centimeter.

Don't worry about him.

Just bring this patient back again for his regular scheduled surveillance scan.

Here's another one.

Peripheral subcapsular early draining vein.

No washout, don't list this as multifocal HCC 'cause now you've just denied this patient transplantation.

So be aware that a large percentage of patients with cirrhosis have peripheral AP shunts.

Read some more about it or review this material to make yourself more confident about that.

But don't over-diagnose these as HCC.

Performance of CT and MR

So overall, I think there's probably somewhat comparable performance of CT and MR with a slight nod to MR as it gets better and better.

MR is definitely, preference for those with an iodine allergy or known steatosis.

Definitely better at characterizing nodules.

Conversely, I prefer CT in the patient who is more acutely ill or unstable who has ascites, who's tachypneic.

Both are evolving and improving constantly.

But really we need to perform and interpret these properly in order to get the maximum benefit out of these procedures.

Surveillance Imaging for Cirrhosis

Let me just ask a question in the audience, in your own practice, what is the main surveillance imaging test for moderate to advanced cirrhosis?

How many would do ultrasound?

How many CT?

Okay, how many MR.

And how many PET CT?

We'd never do PET CT in this setting.

So here's what we do.

They do the screening, alpha fetoprotein.

There is some other markers as well.

The protein induced by vitamin K, blah, blah, blah.

Ultrasound.

We do a lot of ultrasound and it seems when you read a literature like there's this total discrepancy between what's reported from Japan and Europe and North America.

But let me try to straighten that out for you.

Ultrasound is a very good screening modality in experienced hands in relatively thin patients in not very cirrhotic livers.

Now, in fact, that describes the situation in Asia and they have really good results with ultrasound, but much of their, the great majority of their cirrhosis is due to hepatitis B, which often does not induce much hepatic fibrosis.

In America where we have larger patients, often with alcohol as part of the injury to the liver, ultrasound is more challenging.

So when I'm reading surveillance ultrasound, which we do, you know, every week when I'm on our outpatient thing, I include in my dictation the following, the liver is relatively homogeneous and it's well evaluated by ultrasound and I'm reasonably confident about that.

Once the liver becomes nodular, fibrotic scarring, what I put in my dictation there is now ultrasound has ceased to be an effective screening modality and we really need to get CT.

We often get patients referred into us who have been followed only by ultrasound, believe it or not.

And almost always, by the time this lesion, a focal lesion has been found by ultrasound, we do the CT or MR.

And the horse has left the barn.

It's a big ugly incurable lesion.

So that's my take on this.

We generally do CT or MR about every 12 months.

If somebody has merely chronic hepatitis without cirrhosis and it's a relatively smooth liver, we'll still follow those patients by ultrasound for high clinical suspicion.

An AFP that's going up, deteriorating liver functions, et cetera.

We're going to shorten the intervals and rely even more heavily on CT or MR.

Summary

So in summary, ultrasound, CT and MR are all useful in evaluation of cirrhosis.

Large or symptomatic HCCs are easily detected in stage, but unfortunately those are incurable small HCCs in a screening population are more challenging.

There is some overlap among the nodules, but I hope that the criteria that we've discussed here today will allow you to more effectively and confidently distinguish among these.

So we'll stop with that.

Thank you.