Pitfalls: Liver

Case 1: Hemangioma in Steatotic Liver

Here is a case from my own practice, stone Protocol ct.

One of my colleagues said there is a hyperdense hypervascular hepatic mass and listed a long differential diagnosis with some allusions to much more imaging that they might consider.

I would look at this and say, this mass is hyperdense relative to the underlying liver, but it's essentially isodense with blood pool on the non-contrast.

Then I see nodular peripheral enhancement that is isodense to blood pool.

I think this is hemangioma.

Your diagnosis please. What's that? How many metastasis, how many hepatocellular carcinoma? How many hemangioma? Yeah, I gave you the hint there. FNH.

Background steatosis the mass really fulfills all of the real criteria for diagnosing hemangioma, but it upsets your normal visual pattern in terms of recognizing it as a hemangioma.

I see this mistake made a lot.

You really want to compare hemangiomas to blood pool density. If it's a fatty liver that can easily throw you off.

Case 2: HCC in Fatty Liver

Here's another case from my own practice, 60-year-old woman with cryptogenic cirrhosis and here is an arterial and venous and delayed phase imaging.

This is the impression, slight interval growth in a 2.2 centimeter hypervascular lesion, blah, blah blah.

This lesion is suspicious but not classic for hepatocellular carcinoma.

Once again, some recommendations were given for other modalities for workup and I was asked to look at this separately and I said this is a typical HCC, but it's a typical HCC within a fatty liver.

It's encapsulated, it washes out.

Remember, washout means the lesion deenhances to become hypodense to liver but hypodense to normal density liver.

If the underlying liver is steatotic, which it most definitely is, it may not be capable of washing out to the point of being hypodense to fatty liver.

You must consider the background liver.

Pitfall Number One: Failure to Identify Background Steatosis

Pitfall number one, failure to identify background steatosis.

It alters the expected appearance of liver masses.

Almost all masses may appear hyperdense to a fatty liver depending on the degree of steatosis and it's likely to lead to an overly broad or even incorrect differential diagnosis as it did in the two cases that I showed you.

Steatosis is extremely common and becoming more common all the time.

We traditionally think of it as alcohol and diabetes, but increasingly the metabolic syndrome, the obesity epidemic is becoming the major cause of this.

On ultrasound, we expect increased echogenicity, poor penetration of the sound beam and poor visualization of vessel margins within the liver.

If I had to choose one imaging modality for definitive diagnosis and even quantitation of fat, it would certainly be MR.

Very severe fatty liver may give you increased signal on a T one weighted image, decreased signal on fat suppressed sequences.

But the especially good modality is in and out of phase imaging because there's generally a mixture of fat and water protons within the areas of steatosis.

Selective dropout of signal on out of phase imaging is the best manifestation imaging manifestation of steatosis on ct.

I still get a lot of funny responses from people about what criteria they use to diagnose steatosis.

In general, the non enhanced CT is the best modality among ct, decreased density less than the spleen.

On a non-contrast set of images, I'm often asked if you can diagnose steatosis on contrast enhanced imaging and I would say be very careful of doing it on arterial phase.

Be somewhat wary on portal venous phase.

The delayed or equilibrium phase imaging is probably most useful among the enhanced phases of imaging.

But if I'm going to diagnose steatosis on enhanced ct, I wanna see the liver be substantially less dense than the spleen maybe in the range of 25 or 30 Hounsfield units less.

An important visual clue of steatosis to distinguish it from mass is the presence of blood vessels running through the lesion undisturbed.

They're not deviated, they're not pinched, they're not occluded.

Steatosis is often in a geographic area.

It can be rounded and here's a good little pearl for you.

It tends to occur around hepatic ligaments and vessels and may easily resemble metastases.

I've seen dozens of such cases incorrectly called metastatic disease.

Focal and Multifocal Steatosis Examples

Here's a patient who has both diffuse and focal steatosis.

The liver in general on this non enhanced scan is less dense than the spleen and some areas are even more hypodense, which are just relatively more severe foci of steatosis.

When the area is quite distinctly different than the rest of the liver, it will certainly give you pause.

Is this a mass or is this steatosis?

But I would again urge you to look for that sign of blood vessels running through the area of interest in an undisturbed fashion that was focal fatty infiltration. No mass.

Here is multifocal fatty infiltration having both a geographic distribution but also a tendency to be more prevalent surrounding major blood vessels within the liver.

Again, a very characteristic distribution.

Here's another example of multifocal steatosis that might be mistaken for liver masses.

This was a cancer patient and we wanted to make a definitive diagnosis.

We said this is most likely steatosis and therefore if you want confirmation, the method of confirmation is going to be not biopsy, but rather MR being sure to get in and out of phase imaging in phase imaging out of phase imaging selective dropout of signal from each of these hypodense foci definitive for multifocal fatty infiltration definitive to the point of being better than liver biopsy because biopsy can have sampling error.

Case: Focal Fat vs Hepatocellular Carcinoma

Alright, here's a patient who we are following in our cirrhosis clinic with periodic imaging and I am showing you a cirrhotic liver with signs of portal hypertension.

I want to know whether you think this is focal fat or hepatocellular carcinoma.

Simple binary decision here. How many go for focal fat? What's that? Okay, you can actually vote on this focal fat or HCC one or two. All right, focal fat.

But what would be the definitive thing to do?

Here's a T two weighted image.

Most tumors are going to be hyperintense on T two weighted imaging.

This is not, but the go-to modality is going to be in phase and out of phase imaging and there is selective signal dropout.

This is focal steatosis, not HCC.

We did not go on to biopsy this and this patient was just maintained on the follow up list awaiting other criteria for liver transplantation which occurred way down the road.

Focal fatty infiltration, simulating tumor.

Oh gee, this looks very familiar.

See time and again you'll see this finding.

Focal Sparing in Fatty Liver

Okay, here we have a focal hyperdense lesion or lesions, in the liver.

One in the lateral segment, one in the under surface of the medial segment.

What do you like here? This patient also has some chronic liver disease.

Is it HCC? Is it metastasis? Is it FNH or are these two foci normal liver? Vote please. All right, good. Normal liver.

The areas of interest are actually what they're supposed to be.

They look like they're enhancing about the same amount as the spleen.

It's the normal density difference of liver rather than spleen and it's the underlying liver that is abnormal.

A little pearl here. Focal areas of sparing the undersurface of the left lobe and the area around the gallbladder fossa are favored areas for focal sparing.

This patient had liver disease due to non-alcoholic steatohepatitis, which is epidemic in our population and which will shortly in my judgment become the number one cause for liver cirrhosis and HCC in North America and Europe.

Ultrasound Equivalent of Focal Sparing

Here is the ultrasound equivalent of this.

We expect the steatotic liver to have increased echogenicity, poor sound transmission and so forth.

But you may perceive this as being a hypoechoic mass here on the under surface of the left lobe and here surrounding the gallbladder fossa.

But again, everything is relative.

If the liver, if the fatty liver is too echogenic, maybe the normal liver is going to look sonolucent relative to the fatty liver, but it's about what you'd expect relative to the kidney and other internal standards.

Focal fatty infiltration and focal sparing are a perfect match for each other.

Ironically, they sort of occur in the same areas, and having to do probably with alterations in venous drainage.

Here is focal steatosis in that undersurface of the left lobe and around the gallbladder fossa the same thing we saw with focal sparing.

Don't be surprised to see that.

Diseases Simulating Fatty Liver

Let's go on to other diseases that may simulate fatty liver.

I've showed you a lot of cases and you got the diagnosis correct, perhaps sort of prejudiced by me to by saying that focal steatosis may mimic tumor.

Conversely, I've seen a lot of mistakes made when radiologists have looked at a CT scan that shows a, an enlarged and low density liver and they say paragraph impression steatotic liver.

Usually that's correct, but sometimes not.

It might be diffuse tumor.

I see this particularly with lymphoma, some metastases and some hepatocellular carcinoma.

Neither of these tends to be completely uniform, but it certainly can be widespread.

You may see it in diffuse infection almost always in my experience in immunosuppressed patients such as AIDS or transplant recipients.

This can be anything in the microbiological spectrum, it can be viral microbacterial, fungal tb, et cetera.

Diffuse acute liver injury may give you diffuse low density due to widespread hepatocellular necrosis.

You can get it from severe hepatitis mushroom poisoning.

The most common cause for acute liver failure in America is Tylenol overdose.

I see about 12 patients a year go to liver transplantation from Tylenol overdose, usually not in a suicide attempt, so be very careful about Tylenol.

Radiation hepatitis can give you widespread low density in the liver.

Here is a big swollen liver and the spleen is enlarged as well.

But this is a patient with known melanoma.

There were no other known predisposing factors for steatosis.

We biopsy the liver and this was all melanoma.

You can't see focal lesions and I've seen this in diffuse lymphoma and diffuse breast cancer metastases as well.

Keep that in mind in patients with cancer.

Challenge Case: Steatosis vs Metastases

Alright, here's another challenge case.

45-year-old woman with a history of breast cancer and she's got multifocal liver lesions, unenhanced scan contrast enhanced scan and I want you to vote on steatosis or metastases while you're voting.

I'll point out as it says in the caption that I can see some blood vessels in this area of abnormality.

I guess we could debate whether they're running through undisturbed or not.

There was one vote, okay, It was a voting block. Did you see that? It was a, all right, so these are metastasis and I know that in this case because we happened to do an FDG PET scan on her and you see the FDG avid metastases diffusely in the right lobe with liver.

Again, a repeat caution.

Breast metastases in particular can diffusely infiltrate the liver.

Todd showed you a case of pseudo cirrhosis due to widespread metastases from breast cancer.

Be aware of patients with breast cancer mimicking steatosis.

Pitfall Number Two: Cyst vs Too Small to Characterize vs Something Bad

Pitfall number two cyst versus too small to characterize versus something bad.

This is a topic I'm particularly passionate about.

I really believe we need to provide a more useful diagnosis than hypodense liver lesions too small to characterize.

That's a total wastebasket terminology.

This is a patient with colon cancer and our first preoperative staging exam.

Now some of these lesions I think we can clearly say are cysts, pencil, thin wall water density, no mural, nodularity and so forth.

Others we're not really going to be able to measure water attenuation because of partial volume averaging.

Now we pick this up in our outpatient imaging center.

Ultrasound is right across the hall and rather than just say can't rule out metastases, recommend repeat imaging at six months or something.

Again, I think that's in general a cruel thing to do to patients.

We just called a referring physician wielder across the hall to ultrasound and every one of these lesions had diagnostic criteria for representing cyst, simple cyst and therefore we were able to dictate both studies as indicating multiple benign cyst of varying sizes, no evidence of metastatic disease.

I think we should try to be as specific as possible.

Hepatic Cysts: CT Criteria and Differential

Hepatic cysts, CT criteria, water density, no enhancement of any portion of it, no visible wall, no more than about two septa off, often of different sizes.

Differential diagnosis, yes we can see cystic tumors and abscesses and hematomas, the latter usually following trauma or surgery.

The problem as we know is that small cysts often don't measure water attenuation due to partial volume averaging.

What do you do for problem solving other than just saying I'm not sure what I'm looking at.

Sonography is really good.

No debris, no mural nodularity, no wall thickening, good acoustic enhancement and so forth.

MR is certainly good as well though frankly it's more expensive and we do a lot more in a way of ultrasound of that.

Pearls for Small Hypodense Lesions

Here's some pearls that I apply in my practice every day.

A small hypodense lesion less than two centimeters in diameter in a non cancer patient is essentially always benign.

My dictation will say by imaging criteria, indeterminate but statistically very likely to be benign.

That is a very different statement than I can't rule out HCC metastases and a whole bunch of other life-threatening situations.

Lesions less than blood density on an unenhanced CT scan are almost always benign.

If you happen to have a non-contrast CT scan, I really don't think you need to hedge about these.

It's either gonna be a cyst or a biliary hamartoma or a thrombosed angio or if it's near or fat density, it's gonna be an angiomyolipoma basically not to worry about if you see multiple slightly complex lesions that are small all being less than 15 millimeters, these are probably biliary hamartomas and I've said this many times in lectures and people have said to me, I'd never make a diagnosis of biliary hamartomas.

They're out there.

I'm sure you see these in your practice.

Cystic metastases can be from any primary tumor, but the ones that are especially likely to cause these are sarcomas including just tumors, tumor or metastases from cystic primary tumors such as the ovary or pancreatic mucinous cyst, adenocarcinomas and squamous cell carcinomas.

Almost all of these will show some complexity on CT ultrasound or MR.

Once again ultrasound is a very good problem solver at showing subtle mural nodularity.

Frankly, in the old days of doing eight and 10 millimeter thick CT sections, this was much more of a problem for me than it is now where we can do thin sections, multiplanar, reformations and so forth.

I don't struggle with this nearly as often on CT as I did maybe 15 years ago, but still you'll run into cases like this 60-year-old man who has squamous cell carcinoma of the tongue and he's got these multiple low density liver lesions.

I'm showing you three of them here and I want you to vote on cysts mets. We're not sure. All right, most of you have voted for mets, some admitted you weren't sure And let's go.

But if you go back and look at this, this lesion while being low density is not quite as low density as the water in the stomach nor the bile in the gallbladder nor the renal cyst.

Also for a lesion that size or this size, I really expect for that to be very sharply demarcated.

Then you wanna look for other subtle findings as well.

Upstream actually downstream from this lesion I can see subtle dilation of the intrahepatic bile ducts, not a finding that I expect to see for a cyst of that size.

I think there are several findings here that correctly indicated the likelihood of these representing cystic or necrotic metastases.

Squamous cell carcinoma, you'll see low density involved metastatic lymph nodes, liver lesions, sometimes even adrenal lesions.

Metastatic Sarcoma Examples

This is a typical appearance of metastatic sarcoma where yes there are large cystic or necrotic regions, but hopefully nobody would miss the enhancing mural nodularity once again there's obstruction of intrahepatic bile ducts as a clear indication of an invasive process, not a benign process.

Another patient with metastatic sarcoma on mr.

Again, I don't really care that this looks like pretty unremarkable fluid.

The fact that there is a visible enhancing wall, there's a debris level within this.

Those are very worrisome findings.

In fact, I'd say diagnostic findings, particularly the mural nodularity that this cannot be a benign lesion but rather represents metastasis.

Effect of Anti-Neoplastic Agents on Metastases Appearance

Here's a case that I really want you to pay attention to 'cause it's really emblematic not just of this specific entity but this newer generation of anti-neoplastic agents that deprive tumors of their blood supply.

Some of the old criteria like the simple recist criteria for determining whether a patient's doing better or worse on therapy and sometimes even a matter of recognizing what is really a metastasis versus a cyst had been thrown into question or confusion.

I am actually reading this scan in our outpatient center one day with a resident and he says, I see a big cyst in the liver, it measured near water attenuation.

I said, let's make sure we're looking at prior scans and here's a scan from 11 months earlier and this nobody in the room would call that a benign lesion.

This is in fact a metastasis from a gist.

When the patient went on Gleevec, there was such effective devascularization of the lesion, that it really reverted to being almost a water density, sharply defined lesion that could be mistaken for a cyst.

Here's the same sort of situation again, I'm reading in the outpatient center one day and we've got this scan that I am reading and we have the original CT scan right here.

I'm reading this and am I going to say these are incidental cysts, they measure water density or these mets that are getting worse or these mets that are getting better.

Sounds like a crazy choice they have to make, huh?

This is the original CT scan and this is the scan I'm reading after I think a couple of months of therapy.

Your votes please.

This one's a little complex so hang in there with me.

It's an important point, believe me. Alright, these are mets that are getting better.

The original scan that I had to read was a portal venous phase only scan when we had an in-between scan to read after just one round of chemotherapy, we could see on the appropriate arterial phase imaging because GI stromal tumors are hypervascular tumors, we could see multiple solid enhancing lesions that disappeared on a portal venous phase of imaging.

We could see that this is a solid though less vascular metastasis as is this.

This patient had multiple metastases.

Now with further treatment even on biphasic imaging that we performed, we actually only saw several remaining lesions, each of which could easily be mistaken for a simple cyst.

Once again, I know you've heard this several times over the last two days, but it's critically important.

You gotta dig through the jacket.

I know it's not a film jacket anymore, but you gotta do your homework and really figure out the course of this patient's medical care over more than just the most recent prior scan, there were actually dozens of hypervascular mets that disappeared or became cystic and this is a good response even though we saw more lesions than we did on the initial CT scan.

Here's another patient in this case with a small bowel GI stromal tumor.

There is only one meta, this is that patient actually only one metastasis is seen on the original scan and that again was done with only a portal venous phase of imaging and undoubtedly we underestimated the extent of metastases.

Here's a woman with ovarian carcinoma.

She has multiple lesions, didn't measure quite water density but they're awfully low density on this portal venous phase imaging, but they were somewhat irregular and complex and these in fact do represent metastatic foci from a cystic primary ovarian malignancy.

Hemorrhagic Cyst vs Cystic Neoplasm

Here's a word of caution.

If you have a cyst in any organ, liver and kidney, most often that undergoes hemorrhage.

That can be very difficult to distinguish from a cystic neoplasm.

If you looked only at the contrast enhanced scan, you might say this looks like an enhancing mural nodule and enhancing mural nodules are always neoplastic.

The principle is true.

But if you happen to look at the non enhanced scan, this is hyperdense material hyperdense to muscle or liver, this is actually clotted blood.

I'm not here to say that this is a really easy confident diagnosis to make.

It is decidedly difficult to distinguish hemorrhage within a cyst from cystic neoplasm.

I often will see cases with multiple low density lesions in the liver and people will say multiple simple cyst, autosomal dominant polycystic disease and if they've heard of biliary hamartomas, they'll throw that into the differential diagnosis but in fact they generally don't look alike.

Autosomal Dominant Polycystic Disease

The diagnosis of autosomal dominant polycystic disease is not based on the number of cysts.

The cyst in polycystic disease are usually innumerable and a varying size and usually in large and distort the liver.

The lesions are often of different density or intensity on MR which is more sensitive.

They may even have calcified walls due to prior episodes of bleeding within the cyst.

If you don't make the diagnosis based on the number of cyst, what do you make it on?

The appearance of the cyst within the liver is enumerated there.

The presence of cyst in other organs such as the kidney, pancreas, seminal vesicles and so forth, family history is important.

There are genetic markers as well.

This patient has multiple cysts in the liver, but notice how varying in size they are.

There are almost always cysts that vary from less than a centimeter to many centimeters.

Notice that the cysts, some have irregular calcifications in the wall from prior bleeding episodes and there are cysts in the kidneys as well as there usually are.

They're not always. And this is absolutely diagnostic of autosomal dominant polycystic disease.

MR is much better than CT at showing the complexity of the fluid.

On this T two and then T one weighted image, this particular cyst is relatively high intensity on both characteristic of hemorrhage.

Some of the cysts look just like simple fluid and then there's essentially every shade of gray in between.

Notice again the distortion of the liver and the presence of innumerable cysts of varying size.

Case: Biliary Hamartomas

Here's a 60-year-old man with vague abdominal pain and he has liver lesions and I want you to tell me what you think he has.

This is a non-contrast scan.

This is enhanced, looks like it's in the portal venous phase of imaging and I won't show you a whole lot of images through the rest of the liver, but there were dozens of these lesions and I'll show you the MR two.

What the hell? What do you think those represent?

Alright, so we got three polycystic liver and one biliary hamartomas.

Well the answer is biliary hamartomas.

Let's go back to the basic principles.

If this was polycystic liver disease, almost certainly some of these cysts would've been more than 15 millimeters in size.

There's no distortion of the liver, there's no alteration of liver function.

All of these lesions are less than 15 millimeters in size.

That is essentially diagnostic of biliary hamartomas.

Bile duct hamartomas, no other organs or family history of polycystic disease.

They can range in a number from few to innumerable.

They're invariably small.

There actually will be some irregularity in the shape of these lesions.

They are very bright on T two weighted imaging.

Frankly, MR is not especially helpful in narrowing the differential diagnosis.

Once you have a CT scan, here's another, actually that may be the same case, but again very bright on T two weighted imaging.

There's a little bit of subtle irregularity to the outline of some of these cystic collections.

These go by a variety of names including bile duct hamartomas or adenomas or Von Meyenburg Complex benign proliferation of bile ducts and stroma.

They're invariably small.

They're essentially never more than 15 millimeters in diameter.

In fact I think this is actually a common cause of the low density lesions that we see so often and dismiss as too small to characterize.

It is common patients who have other manifestations of congenital fibro polycystic disease, but I don't have time to go into that right now.

They're often too small to get a valid measurement of attenuation on CT ultrasound or MR.

Counterintuitively. These lesions are often quite echogenic on ultrasound.

Now if you think about it, there's a little bit of fluid in these lesions.

They're small and there's a variable amount of fibrotic tissue in the wall.

That creates a good environment for vibration on ultrasound.

These lesions can look quite echogenic on ultrasound, I'll show you that.

Conversely, on CT or MR, you may mistake this for polycystic disease, but again, the uniformity of size, the small size of the lesions, the irregular wall thickening would be rare in an autosomal dominant polycystic liver disease or multiple simple cysts.

These are hard to biopsy and in most cases we just say this is what it is and we don't recommend any additional workup.

Here's another patient with biliary hamartomas.

There is some irregularity to the outline of each of these lesions.

They're generally not perfectly spherical.

This is the same day ultrasound, minutes afterwards.

I remember talking to the fellow about this case and he said, I saw some of those cysts in the liver, but not nearly as many on ultrasound, which really surprised him.

He was concerned because he saw all of these echogenic lesions in the liver.

Well in fact they're the same thing.

All of these lesions are biliary hamartomas.

Again, the smaller ones or the ones that have a lot of fibrotic nodularity in the wall will appear echogenic on ultrasound.

We probably saw, I dunno, 10 times as many lesions on CT as we did on ultrasound in this particular case.

Here's another example I've seen, I don't know, dozens of these over the years.

MR does give one additional piece of information in the differential diagnosis Notice on this MRCP, these lesions do not communicate with the biliary tree.

That helps to distinguish this in this asymptomatic woman from Caroli's disease when, which as you know, you can get cystic dilation of the intrahepatic bile ducts.

But in Caroli's disease, those cyst in quotes communicate with the bile ducts, whereas they do not in cases of biliary hamartomas.

Summary

In summary, know how to recognize steatosis and its mimics and we need to try to be more helpful than hypodense liver lesions too small to characterize.

I really think we can do better and owe it to our clinicians and the patients, to do that.