ACR TI-RADS: Tips, Pitfalls, and Prospects - HD

Historical Naming of the Thyroid Gland

When Thomas Wharton named the thyroid gland, he was referring to the appearance of an ancient Greek shield that you see here on the left.

You might say that it doesn't look like a thyroid gland, but if you turn it around like that and superimpose a thyroid gland, I think he was right on track.

Disclosure

This is my disclosure.

I didn't know anything about college football until I moved to Alabama and roll tide.

Rationale for Studying Thyroid Nodules

So why are we doing this? Dr. Desser and her talk eloquently described some of the rationale for what we're doing.

There's lots of nodules around, mostly benign, few that are detected by palpation.

I'll say from my own point of view, I was never able to feel the thyroid nodules that my instructors said were present.

Autopsy, the study from the 1950s found thyroid nodules in about 50%.

And ultrasound, as you heard, 67, 68%.

So very high prevalence of thyroid nodules of which most are benign, which ones deserve attention.

Ultrasound Features of Thyroid Nodules

Dr. Desser showed you some of the ultrasound features that have been described over the years, and I'm gonna talk about the same features, but from the point of view of the investigators who brought them to light.

And one of the first articles that looked at this was from the AJR in the AJR back in 2002 from the South Korean group who identified these four features, microcalcifications irregularity of shape, hypoechogenicity, and height greater than width, the same as taller than wide.

And you'll see these features repeated over and over in all the other subsequent investigations and guidelines.

SRU Consensus Conference

The SRU jumped into the fray back in 2004, 2005 when we had a consensus conference.

They locked a bunch of us in a room for about a day and a half in Washington DC until we came up with some ideas about what to do with thyroid nodules.

And this was the table that was result in from that.

And again, you'll see some of the same themes repeated over and over, microcalcifications solidity and so on.

Inspiration from BI-RADS

As this was going on, many people were looking at the success of BI-RADS from the ACR in breast imaging.

And so it made sense possibly to apply this to the thyroid gland as well, and call it TI-RADS standing for Thyroid Imaging Reporting and Data System.

Early TI-RADS Publications

The first group that actually latched onto this and published it was Horvath Group from San Diego, Chile when they published this article in 2009, and they called it TI-RADS.

And this is a table from their paper.

I won't go through all the features, but many of the things that again, are repeated through multiple guidelines.

You will see here, hypoechoic and so on, with increasing risk of malignancy as you go through the stages.

And they went through TI-RADS two all the way up to TI-RADS six.

That's another thing you'll notice in the various risk stratification systems.

Some people decide to go just in ordinal numbers, 1, 2, 3, 4, 5, and so on.

Some had subgroups, as well, as you'll see here, this was from AL in radiology in 2011.

Again, they decide to call it TI-RADS.

I'm aware of at this point, there are at least five or six different systems out there, all called TI-RADS, which has led to some confusion in the ultrasound community that I'll talk about a little bit more later.

But the same feature showing up again and again, solidity decreased echogenicity marked decreased echogenicity, micro location or irregularity of margin, micro calcifications taller than wide.

And these are the risk of malignancies associated with their categories.

And as I just mentioned, they as well as some other groups decided to sub stratify one of the levels in this case, level four into three sub levels A, B, and C.

American Thyroid Association (ATA) Guidelines

But it's not as if at this point that anybody had found some other sign of malignancy that nobody else had discovered the American Thyroid Association in 2015.

And their guidelines took a slightly different approach, which is pattern based, and they describe multiple patterns.

As you go from bottom to top in the chart, you increase the risk of malignancy.

And by patterns I mean that you look at a thyroid nodule and you have to match it to one of these patterns.

And unfortunately, in some cases, you, a particular thyroid nodule will not match any one of these patterns.

And then you're not sure what to do.

Korean TI-RADS (K-TIRADS)

The South Korean groups, the Korean Society of Thyroid Radiology and the Korean Society of Radiology, and they've done some incredible work in this regard, published their version of TI-RADS, the they termed K-TIRADS last year in 2016.

And this is their management guidelines with increasing risk, high risk of malignancy at the top and less risk at the bottom.

And again, based on very much the same features that I've describing for you in all the other guidelines.

EU-TIRADS

The most recent TI-RADS was published in just a few months ago in August of this year by the Europeans.

It was a refinement of the French TI-RADS, the EU and his colleagues published several years ago.

But here's another TI-RADS in this case called EU-TIRADS and very similar things.

They're going through five levels going from normal to high risk with very similar characteristics to the other risk stratification systems.

Here's their flow chart.

On the high risk side, they're talking about at least one feature that's highly suspicious.

And if there's no highly suspicious features, they stratified this way.

Over-Diagnosis of Thyroid Cancer

And as you also heard from Dr. Desser, the as all this was going on, there was increasing realization of a over-diagnosis of thyroid nodules in the fact that thyroid nodules often had an indolent course.

And in this article from the New England Journal in 2014, this was highlighted showing that following implementation of a screening program for thyroid nodules in South Korea, the thyroid cancer incidents, which was mostly papillary shot up, but the thyroid cancer mortality stayed very low through this period.

And this article that was also published in the New England Journal of Medicine in 2016 said that over-diagnosis of thyroid cancer, which they defined as diagnosis of thyroid nodule can, tumors that wouldn't have left alone have led to symptoms or death accounted for as much as 70 or 80% of thyroid cancer cases in women, and 45% of cases in men in the US between 2003 and 2007.

That's a lot of money being spent.

ACR Projects on Thyroid Nodules

So a few years ago, the ACR embarked on a three phase project.

The first was to develop recommendations for what to do with incidentalomas in the thyroid gland.

And Dr. Desser showed a slide from that article.

That effort was led by Jenny Wang at Duke.

The second was to develop a standard lexicon for describing thyroid nodules on sonography that was led by ED Grant.

And then the last one, developing recommendations for management, which nodules to biopsy.

The lexicon article is the front page is shown here.

And this form the basis of TI-RADS, again, based very much on these same features.

What we did was divide them into five categories, composition echogenicity, shape margins, and echogenic foci.

And the ones I have outlined here in red are the ones that are most suspicious.

Development of ACR TI-RADS

Phase three was to develop a ACR TI-RADS, which was our version of the risk stratification system.

And I'm immensely grateful to all the people on my committee, several of whom are here with me today, who did a wonderful job in keeping me on the straight and narrow.

As this effort went along.

Our aims were to base our system on the ACR lexicon, make the system easy to apply across a wide gamut of practices.

We realized that not everybody does a lot of thyroid ultrasound in their practice.

There are some practices that do relatively little of it, and we wanted to make the system easy for them to use as well.

We wanted to classify all nodules.

I mentioned a moment or two ago that the ATA system isn't able to classify all nodules.

Our motto was essentially no nodule left behind.

And then we want, of course, to be as evidence-based as much as possible.

And as Dr. Desser mentioned, to identify biologically significant cancers that are more likely to harm the patient.

Our principles, first of all, we wanted to make it points based, not pattern based, avoiding the possibility of not classifying some nodules.

We started out working with half points, but we realized that doing the math was harder.

So we decided not to do half points.

We said that we weren't going to do A, B and C subcategories, not that they're bad in and of themselves, where we thought that would it would be simpler not to have them.

We would have no category for a normal thyroid.

And those of you who do a lot of thyroid ultrasound, that you never see a normal thyroid gland anyway.

Measuring Thyroid Nodules

One of the things Dr. Desser mentioned that we dealt with is how to measure thyroid nodules.

And there are a lot of different ways to do this, but in ACR TI-RADS in the sagittal plane, we take the longest measurement from end to end, and that may not be parallel to the skin surface in the axial plane.

We take the longest measurement, again, that often won't be parallel or won't necessarily be parallel to the skin surface, and then perpendicular measurement to that.

Why is size important to begin with?

But one of the articles that was used as a basis for size criteria and several other guidelines was this one from 2005.

And this was a graph from that article that looked at papillary follicular thyroid cancers and said that there was a takeoff at about a threshold of about two centimeters.

But we also had an insight in our deliberations that ultrasound sizing and pathologic sizing may not be the same thing.

And it turns out that ultrasound results in larger measurements with a discrepancy of about 0.7 centimeters.

That is when we measure something, it's larger than what's seen pathologically.

And we took that into account in coming up with our size cutoffs.

Remember that these are guidelines and not standards.

ACR TI-RADS Feature Categories

So getting back to the feature sets, we decide to take the various signs that I've been talking about and group them into these four categories going from benign to highly suspicious.

And we, our challenge was to figure out how to turn that into something that could be used practically in the ultrasound lab.

And this is what we came up with.

This is the ACR TI-RADS chart.

And I'll note down here, this is a picture of my dog, Sam for Samantha, which you could take any way you want, the significance of that picture being there.

But here's the ACR TI-RADS chart.

Here's the five categories, composition, echogenicity, shape margin, and echogenic foci.

So you go through these one by one, starting off with composition.

You pick one and it has points associated with it.

Same with echogenicity, same with shape, same with margin.

The exception is echogenic foci because unlike in the other categories where the features are mutually exclusive, here they are not.

For example, you can have comet tail artifacts and rim calcifications or peripheral calcifications in the same nodules.

You pick all the ones that apply.

When you once you've done that, you take all the points and add 'em up, and you come up with one of these five levels going from zero points all the way to seven points or more.

And each of those translates to a risk level ranging from TI-RADS 1 to TI-RADS 5 shown here.

For TI-RADS 1 or TI-RADS 2 nodules, we say that no follow up or FNA is necessary for TI-RADS 3, TI-RADS 4 and TI-RADS 5, depending on the maximum diameter of the nodule, either you do an FNA or you do a follow up.

If it's less than the threshold needed for follow up, no further follow up is recommended.

Down here at the bottom are a series of notes that relate to the various feature categories.

For example, what do you do if you have a nodule that is calcified and you can see what it looks like internally.

And for that you assume that the nodule is solid, even though you can see that with certainty.

This chart was meant to be used, either viewed on the screen electronically or put up in the reading room.

And as I'll mention in a moment, it's also useful to have in the ultrasound scanning rooms.

Risk Data for ACR TI-RADS

These are numbers from Dr. Middleton's work, which was done sort of in parallel with ours.

That informed some of the decision making in ACR TI-RADS.

And these are the aggregate risks going from TI-RADS 1 to TI-RADS 5, and you see those increasing.

And here is similar data.

Data shown a slightly different way going from zero points to 10 points.

And these are the risk of malignancy, which you see go up as you'd hope.

Growth and Follow-Up Recommendations

What about growth? We define growth in the same way that the ATA does.

That is a 20% increase in at least two nodule dimensions with a minimal increase of two or a 50% or greater increase in volume.

Volume sounds like it might be difficult to calculate, but actually most ultrasound equipment now lets you do that quite readily.

So it's an option. As for follow up, there was not a lot of good data in the literature to inform this.

So this was really expert opinion.

We said for a TI-RADS 5 nodule that didn't meet the criterion of size for FNA, that annual follow up for five years would be required.

For TI-RADS 4, we did slightly less 1, 2, 3, and five years and TI-RADS 3, 1, 3, and five years.

And that was partially informed by work that showed that especially for benign looking nodules that remain static for five years, that they're highly likely to be benign.

Active Surveillance

Active surveillance was something we discussed as well because as Dr. Desser mentioned, a lot of small cancers are really indolent.

Patients die with them, but not of them.

So it's an option for patients with less than one centimeter carcinomas, micro carcinomas that are either proven or highly likely to be malignant, and they may elect to undergo watchful waiting versus surgery or possibly ablation.

And it's important when you see one of these two report the proximity to the trache esophageal groove because that has an implication for prognosis.

Special Considerations

I'm often asked what to do about previously sampled nodules that have already been Bethesda classified.

We decided not to go down that rabbit hole and encompass it in ACR TI-RADS.

There are good guidelines for doing that.

You've already heard about molecular testing in practice, often this, most often this is going to be driven by a physician request.

What about multinodular glands?

If you have multiple nodules of similar characteristics or they're confluent or diffusely, heterogeneous usually don't have to do an FNA.

And here's an example of such a gland.

You could probably put a cursor around something like this and invent a nodule.

But it's really hard in these cases to know where one nodule starts and the other stops.

Lymph nodes are something we didn't deal with directly either.

So they're not formally encompassed in ACR TI-RADS.

The however their appearance may influence the decision whether to biopsy a nodule.

And you may do thyroid ultrasound may or may not incorporate a comprehensive look at the lymph nodes.

You can do that at the time of the initial scan, the time of the FNA or subsequently after malignancy is confirmed.

Tips and Pitfalls in Thyroid Ultrasound Reporting

So in the last couple of minutes, what I want to do is some tips and pitfalls for you.

The first one is a free tip.

It's not counting as one of my five that is user reporting template.

I found using reporting template in our voice dictation system makes reporting these nodules a lot faster.

And I'm sure that there is work underway to automate this to some extent.

Maybe not to the extent of doing computer aided diagnosis, but essentially adding up the points for you.

So here's the first tip.

Formally report no more than four nodules with the highest point scores that warrant FNA or follow-up ultrasound.

You don't have to report every nodule in the thyroid gland.

Doing that would make the report much too long.

So we decide in ACR TI-RADS to stop at four.

The second recommend FNA for no more than two nodules with the highest point scores that warrant FNA, there's little to be gained by biopsying three, four or more nodules in a given thyroid gland.

Third tip, don't use the term dominant nodule.

I admit that I used to use this term, I used it in a different way than most.

I used it to refer to the most suspicious nodule, but it's often used to describe the largest nodule, which may have nothing to do with whether that nodule warrants further attention.

I saw a patient last week with an outside sonogram that reported a dominant nodule of two millimeters because it was the largest nodule to highlight that tip.

This is an important one.

I suggest that you begin with an overview scan and make sure that your sonographers know what to look for.

That is before starting to number and take images of nodules, look through the whole gland and sort of get the lay of the land and figure out which nodules might require more attention.

And if you have the ACR TI-RADS chart in the ultrasound scanning rooms, they can refer to it as they decide which nodules to take pictures of.

And the last one is compare with the oldest previous scan.

Sometimes changes whether in architecture or size are more evident when you compare with the oldest scan rather than the most recent one.

Common Pitfalls

And quickly, a few pitfalls.

Over calling spongiform is one thing I see most often.

We like to call nodule spongiform because it means we don't have to do anything with them.

So it's easy to do, but it really has to have the cystic spaces in more than at least 50% of the nodule, which I don't think is the case in this one.

The second one is over calling punctate echogenic foci.

I'm pointing to a small white dot.

You can find these white dots all over the place.

They're part of a speckle pattern.

In fact, you can find them in normal thyroid gland.

So you really want to call punctate echogenic foci, discreet round clearly visualized echogenic foci.

Extra thyroidal extension is also easy to over call.

Here's one such nodule.

And it looks like it might be bulging beyond the thyroid margin.

The thyroid doesn't have a true capsule, but just bulging doesn't mean that there is actual extra thyroid extension.

So in order for me to call it, I wanna see actual invasion of surrounding structures.

Fourth pitfall is to obsess over features.

For example, you don't have to spend a lot of time deciding if the margin is lobulated or irregular.

They get the same number of points.

Yes, it's important to be as accurate as possible, but don't go crazy over it.

And finally, it's sometimes really hard to tell a very hypoechoic from cystic nodules.

This one was a papillary thyroid cancer that was hypoechoic and not cystic.

What helps there is often putting color on.

And if you see color in the middle of a nodule, obviously it's not cystic.

Future Prospects for Thyroid Imaging

So what are the prospects for the future?

There's a lot of work going on comparing ACR TI-RADS to other systems.

You heard about elastography, scoring I just mentioned will be incorporated into voice recognition.

And you heard about artificial intelligence will hopefully relieve us of the burden of classifying all these nodules in a few years.

And the last one I'll mention is harmonizing.

I said there are at least five or six different TI-RADS is out there, if that's a word.

And there are multiple other systems that's confused the community as a whole.

And everybody has a stake in saying, well, my system is better than yours.

It would be helpful, I think, if there were some effort to try to get harmony between these systems.

And I've actually been crazy enough to try to undertake such an effort by writing to the leads of some of the other societies to see if they'd be interested in the least talking.

The response so far has been very positive.

I haven't, I'm just starting on this effort and I have no idea where it's going to go, but I think it's worth a try.

So with that, thank you very much for your attention.