Pitfalls: Breast Imaging, Part 1

Definition of Pitfalls

To review, the definition of a pitfall, as we've heard this morning, as a hidden or not easily recognized, danger can be thought of as a landmine or hazard or other risk.

In my field, I think of it as a booby trap.

And it either comes in the form of a false negative, which usually refers to the sensitivity of the exam.

And that may lead to a misdiagnosis with potential for delayed treatment or morbidity or mortality or a false positive, which often refers to the specificity of the exam.

And that can lead to an incorrect diagnosis with the potential for overtreatment and unnecessary costs or anxiety.

And I think of them either as perceptual errors, interpretive errors, and technical errors.

In the literature, they've all types of errors have been attributed to pitfalls.

These are the kind of categories of errors that I'll be reviewing in my talk.

Objectives

The objectives today will be a case-based review of some common perceptual interpretive and technical pitfalls in breast imaging.

And this first talk will be in the context of screening.

And I'll try to provide along the way some useful references for evidence-based algorithmic practice.

And most of these come from the ACR appropriateness criteria that are available online.

Caveats

A few caveats. This is not an encyclopedic review.

I may not be able to prevent you from being named in a lawsuit, and it is essential to reflect on and learn from your own personal pitfalls.

Audience Poll

Quickly, just an informal poll, and this'll get everyone used to using your audience response system.

Clickers, how often do you read mammograms?

Always, often, sometimes, rarely, or never.

Okay, it looks like we have a mix, but many people clicked on that category five, I think that was Number five.

Never half of you never read mammograms, so this is great.

I'm gonna click back to our normal slides and we will not do more audience response, just given the time constraints, but you'll see the answers and to all the questions that are incorporated into my talk.

Given that about half of you don't routinely read mammograms, I wanted to do a quick review of the BI-RADS assessment categories.

BI-RADS Review

BI-RADS, again, stands for Breast Imaging Reporting and Data System.

BI-RADS zero is reserved for the screening category.

If you feel like the images are incomplete and the patient needs to come back for additional imaging, BI-RADS one and two are associated with a screening recommendation.

Basically, return to annual screening.

BI-RADS three probably benign, means that there's less than 2% chance of cancer.

And that's associated with a short-term followup recommendation.

BI-RADS four five, both associated with a biopsy recommendation and BI-RADS six, that's reserved for a known biopsy proven malignancy.

And breast imaging, unlike other parts of radiology, really has this requirement for recommendations being closely linked to all assessments, which is one reason why I'm attracted to the field.

Overview of Talk

To give you an overview of today's talk, we'll go over three categories of pitfalls seen in the screening setting.

The first will be focused on poor mammographic positioning.

Poor Mammographic Positioning

And let's start with this screening case.

This was a woman came in for routine screening.

She had had a prior history of right breast cancer and mastectomy.

This is a left screening mammogram, two views, MLO and CC.

This was called a BI-RADS zero technical recall.

I just wanted to ask the audience without necessarily needing for you to respond, why is this a technical recall?

Is it that there's no inframammary fold, an inadequate visualization of the pectoralis muscle or the posterior tissue, all three, or you have no idea?

I just wanna give you a chance to look at that picture.

And then hopefully you recognize that there are multiple problems with this medial lateral oblique view.

Similarly, the CC view was also a technical recall.

Was it for excessive exaggeration, inadequate posterior tissue?

Both of the above. Was there a skinfold or you have no idea?

Both A and B were technically inadequate.

This woman came back for additional views.

Here's that inadequate MLO view.

And here is a repeat MLO view that now gives us a chance to see a speculated mass.

And this kind of speaks to the importance of a technically adequate view.

Then we looked back at her 2011 mammogram, and were able to then see in retrospect, the speculated mass that had not been identified and that had grown in the interval.

We were looking for this mass on the CC view.

We had our current study in our prior study.

Neither one shows the mass.

And luckily the CC view had also been called technically inadequate for excessive exaggeration, that nipple is not midline and there is not enough posterior tissue.

This was repeated, and you can clearly see the mass in the medial breast now.

And if you look at the two CC views, you can see that we barely were able to capture the anterior aspect of those speculations on our initial technically inadequate view.

And trying to see the architectural elements that are different between the two views.

You see this lobulated fat that is positioned differently between the two views.

And that enabled us to send the ultrasound technologist to look in the upper inner left breast and find this what turned out to be an invasive breast cancer.

That case kind of gives us a context in which to review the goals of appropriate mammographic positioning.

First, the pectoralis muscle should have a wide should be wide with a convex anterior margin.

And that suggests that the patient is relaxed.

The pectoralis muscles should be visualized to the posterior nipple line, which is this perpendicular line to the pectoralis muscle.

The inframammary fold should be clearly visualized and smooth.

The nipple should be up and not sagging.

And on the CC view, ideally the nipple is midline, and the posterior nipple line is within a centimeter between the CC and MLO views.

And those positioning goals are clearly articulated by the ACR.

But also reviewed in this article in Radiologic Clinics of North America 2010.

And this table from that article shows the eight categories for clinical image evaluation.

And there are many, but I'm focusing today on positioning.

And this gives you a list of 10 potential deficiencies that are really useful when you decide to look back at your own mammograms in your own practice.

And here are some cases from our own practice when we first became interested in this positioning issue.

Here we have poor visualization of posterior tissue.

You see that there's definitely more than a centimeter of that posterior nipple line between the CC and MLO view.

The inframammary fold is obscured.

And on this view, we have multiple problems.

The CC view, there's excessive exaggeration.

The MLO view, the nipple is sagging.

The IMF is inadequately visualized, if at all.

And in this case, the woman is positioned too high on the image receptor.

You have a bunch of upper abdominal fat on the image.

On this view, we have skinfold that big fold in the upper part of the image sagging breast, inadequate IMF and inadequate pec.

And then finally this view, we have portions of the breast that are actually cut off, meaning that the patient's breast size was not appropriately matched with the size of the image receptor.

Just briefly, the pearls for positioning would include the fact that it's important to remember that they're very well-defined ACR standards, but adherence to the standards is really dependent on both the radiologists and the technologists.

I would recommend considering an audit of positioning deficiencies in your own practice, and given especially the fact that it's the most common reason for ACR accreditation failure, but also because we wanna make sure we take good care of our patients.

That's a brief review of how mammographic positioning could be an easy pitfall, and one that's very important to correct.

Dense Breasts

Second topic in the screening setting is in the category of dense breasts.

Here's a scenario that we would all like to avoid.

A 49-year-old woman with a palpable right breast mass came in in March 2012, and she had had a negative screening mammogram in January.

This is just two months later, she's coming in with a clinical symptom.

Here is her mammogram from when she presented with this palpable mass.

You see a BB placed over her palpable area of concern.

And here are the two views.

And what is the breast composition?

Is it largely fatty, scattered fibroglandular, heterogeneously dense, or extremely dense?

I think no one would have an issue identifying this woman as extremely dense.

And what is your next step?

She's in your clinic for the diagnostic workup.

We always start our diagnostics for any palpable abnormality in a woman of this age with additional mammographic views.

These are spot magnification views, and you see, again, the BB demarcating her palpable abnormality.

And we don't have better definition of a mass, but we see kind of a little bit of suggestion of distortion.

We see a little increased density compared to the very high density of her background breast.

And there are also calcifications associated with this palpable abnormality.

And at this point, your next step should include a breast ultrasound.

And again, in the diagnostic setting, this is routinely done in clinic.

Ultrasound targeted to her palpable area of concern shows this very large irregular mass that's hypoechoic and has this thick echogenic rim, very large.

BI-RADS assessment, again, we reviewed one through five.

And this was considered highly suggestive of malignancy, mostly based on the sonographic features.

And the woman underwent biopsy, which showed invasive ductal carcinoma, and then for staging, had this MRI.

And you see this very extensive abnormal enhancement that is much easier to appreciate than on her mammogram.

This case very easily highlights the fact that dense breasts limit mammographic sensitivity.

You had a false negative screening mammogram.

And two months later, she's biopsy proven with extensive invasive disease.

The fact that dense breasts limit mammographic sensitivity has become a focus of the breast cancer advocacy community in the last few years.

And that's why I wanted to do a quick review.

Categories of breast density ranging from almost entirely fatty to extremely dense and density matters because it can mask breast cancer.

'Cause cancer usually shows up as a white abnormality on a mammogram.

And there's also evidence to suggest an association between increased density and increased risk.

Density has also become important because there are now mandatory density notification and insurance laws in states across the United States, California being one of them.

And finally, other modalities are available that can be used in conjunction with mammography.

And those include ultrasound tomosynthesis, which is a 3D mammogram and MRI.

Here's just a lay of the land in terms of the legislative landscape.

All of these states that are in red have enacted the law in blue.

The law has been introduced in white.

They're working on a bill in black, there's no action, I'm sorry, pink.

The law is enacted. So you see California here, pink is the law is enacted, and it was effective April 1st.

It's becoming an issue that we all have to deal with because we as radiologists are now mandated to communicate this not only in our reports, but directly to patients.

Why am I talking about this in the context of a pitfall lecture?

First of all, 50% of women are considered dense under these legislative efforts.

And that includes the categories of heterogeneously dense or extremely dense.

That's 2 million women yearly in California.

And what are the pitfalls that I'm referring to?

Well, decreased mammographic sensitivity can increase the likelihood of a false negative result.

And supplemental screening, especially for average risk patients, can increase the likelihood of a false positive result.

It's really critical to know the appropriate recommendations to make in this particular population.

Here are some pearls to remember.

The benefits of harm and harms of supplemental screening are the same as those for screening in general.

And they're based on the prior probability of disease and the value placed on increased cancer detection, given the risk of false positive exams.

Women at high risk for breast cancer are recommended to undergo annual screening breast MR as an adjunct to annual screening mammography regardless of density.

That's very important to remember.

However, there's limited data on supplemental screening for average risk women for both ultrasound and tomosynthesis.

The trials that are available show a similar degree of benefit, meaning added cancers identified, but a very different degree of harm.

Very different number of false positives, specifically with ultrasound screening.

Breast ultrasound, there are four times more false positives than with mammography compared to tomosynthesis, where there's about 30 to 40% fewer false positives than mammography.

And in general, I'm part of the California Breast Density Information Group, which is a working group of academic radiologists and private practice radiologists who have gathered together to talk about this issue over the last year.

And we came up with these recommendations.

First, we wanna help referring physicians and patients put density associated risk in context.

Just so that you're aware, although the published literature often reports a four to six times increased relative risk in the context of density, that's generally comparing those with the fattiest breasts to those with the densest breasts.

Comparing two extremes, you have that four to six times increase in relative risk.

We, however, thought it would be more useful to compare the heterogeneously dense and extremely dense categories to average density women.

Heterogeneously dense breasts are associated with a relative risk of 1.2 compared to average density, extremely dense, a relative risk of two compared to average density women.

That's similar to the risk associated with having one first degree relative with unilateral post-menopausal breast cancer.

That's important information just to help patients who are getting this letter put that risk in context.

The other thing we recommend is using this notification as an opportunity to elicit any red flag risk factors, because we know that women at high risk will benefit from supplemental screening MRI.

Here's a list of the red flag risk factors.

These are also the same risk factors that the NCCN suggests would make for an appropriate referral to a high risk oncologist.

And then finally, remember that screening recommendation should be based on risk, not just on density.

And it's important that we can reassure patients at average risk and encourage annual mammographic screening, even though that they have dense breasts.

Subtle Findings

And finally, I'd like to go on to the third and final section of subtle findings.

The chief of breast imaging at Stanford, Deb Ikeda, published this article in Radiology in 2003, and she reviewed with her colleagues almost 180 missed breast cancers.

And this is a table of the factors that are all interpretive detection and technical that were associated with those misses.

I'm gonna focus here on these detection factors and give you some examples of cases that highlight these subtle findings.

The one view finding, I don't know if anyone can see it yet, but here it is hard to know without looking at priors if this is a new finding.

She has extremely dense breasts.

But because she had a history of lymphoma and radiation to the chest, she also underwent a screening breast MRI.

And here's the axial view rotated so that it looks like the CC view.

And you clearly see this invasive cancer that presented as a one view finding on mammo calcifications can be very difficult to see, easily overlooked.

They were detected. And here's a magnified image of these calcifications.

They can be very subtle, but if they're new, especially if they're new or have suspicious morphology or distribution, they should be recalled and ultimately biopsied.

Here's a finding at the fibroglandular tissue edge.

Again, extremely dense breasts.

This woman happened to have had a surgery for a benign finding.

There's a scar marker here near this palpable abnormality.

Here's a BB denoting a palpable abnormality.

But it turns out that this subtle area at the edge of the fibroglandular tissue on MR is a very large aggressive triple negative breast cancer.

Here's another finding at the image edge and on the CC view, you can hardly see this mass, but clearly on the MLO view, speculated mass, even without priors, this woman would be recalled for additional imaging.

Distracting lesions are very important, especially mammography.

'Cause you're always, generally, unless the woman has had a mastectomy, you're looking at two breasts and you're looking for symmetry.

You can be easily distracted by this finding, which turned out to be a sebaceous cyst.

And this mass could distract you potentially from seeing this.

That initially was thought of as a technically inadequate view with a skinfold.

But on repeat imaging and magnified views revealed architectural distortion with calcifications that turned out to be invasive ductal carcinoma.

An algorithm for dealing with mammography in general to make sure that you don't miss subtle findings.

Number one, compare with priors whenever possible.

And in your institution, if you are able to establish workflows such that the patient's priors arrive before the patient arrives and can be uploaded into your PACS system so they're easy to access and review that really makes that possible.

The second is to try to adhere to standard operating procedures whenever possible.

That includes standard hanging protocols, standard search patterns, and standard workups.

And that's actually very easy to implement if you have standardized reporting.

I think mammography specifically, and now breast ultrasound and MRI has been able to go a long way in structured reporting in part because of the BI-RADS lexicon.

And it's exciting to see that structured reporting is now becoming more of a standard across radiology.

And finally, auditing your practice is critical in breast imaging.

It's federally mandated, but it's still something that you need to do.

I think the mandate just requires that you do an audit.

It doesn't necessarily require that you're meeting benchmarks or that the data that you generate is good.

But doing an audit is a very important way to identify your pitfalls and then learn from them and improve.

And I wanted to highlight the fact that it's really important to reflect on both good and bad calls.

Usually people are working in a group, some people may be working independently, but you can learn collectively and not just from the data, from the numbers, but also from image review.

If there are ways that can be incorporated into your practice on a regular basis, I think you and your patients will benefit.

Case Examples of Subtle Findings

Here are a couple cases 52-year-old asymptomatic woman referred for routine screening.

These are some subtle findings.

If you notice this one view asymmetry on the CC view on the right at screening, ideally you would recall this woman for additional views and an ultrasound.

She came, she had our standard workup for a one view asymmetry, which included a spot view in the CC view.

And ultimately we saw a spot view in the MLO view that allowed us to direct the ultrasound technologist to the right breast 12 o'clock position.

And we saw this irregular hypoechoic mass with an echogenic rim, highly suggestive of malignancy biopsy proven another asymptomatic woman baseline mammogram.

She has scattered fibroglandular densities, but we see even in the absence of priors, this asymmetry, again, it's a one view finding.

Even though there's no correlate on the MLO view, this should be recalled.

And I gave away the answer.

Additional mammographic views now show actually two adjacent, partially obscured irregular noncalcified masses in the upper outer left breast.

And we could direct our sonographer to see both of these areas.

They were deemed highly suggestive of malignancy.

And since they were very easily seen by ultrasound, they underwent ultrasound guided biopsy And pathology confirmed invasive ductal carcinoma of both lesions.

Again, that presented as a one view asymmetry at screening and resulted in a multifocal invasive ductal carcinoma diagnosis.

And here's another case, 63-year-old woman history of left breast cancer status post lumpectomy eight years ago.

She also referred for routine screening.

And this kind of highlights the distracting multiple lesion type case that I referred to before.

Here she has many things that can distract us.

First of all, here's her lumpectomy scar with some underlying architectural distortion.

Here is an area that she was actually able to palpate, and she wrote that on her intake sheet.

This in the past had been biopsied and shown to be fat necrosis.

That all made sense.

This is a circumscribed mass that had been present and stable on multiple priors.

That was something that was dismissed.

But if you aren't careful, you can miss this very subtle new architectural distortion.

And luckily the radiologist who read this was not distracted.

And recalled this patient, again, that is a one view finding of new architectural distortion, can be very easily overlooked, especially given all those additional distracting lesions.

Here's a spot magnification view where you see that very subtle distortion, and now you're starting to see maybe some speculation.

And here's that oval, circumscribed, benign, stable mass.

And this subsequently, this woman subsequently underwent ultrasound.

We see this hypoechoic taller than wide mass with shadowing and a very thick echogenic rim.

This patient happened to have extremely large breasts that were predominantly fatty, and that resulted in a technically very challenging biopsy.

It was a very deep lesion and a very fatty breast.

And I did the biopsy. And despite multiple efforts to reposition the patient, it was very technically challenging.

Now, I knew I was in the correct spot because as you can see, there's a little post biopsy marker that we place in the breast.

Doesn't project well on this MLO view, but I knew I was in the correct location.

But because I knew it was a very technically challenging biopsy, I would not accept anything but an invasive cancer diagnosis result on the pathology.

And the pathology came back benign, fibro fatty tissue.

And just like for all of our cases we're required to do rad path correlation on all of our biopsy recommendations.

And I knew very quickly that this was a discordant pathology result, and so recommended her for surgical excision.

Her surgical preoperative localization was also technically very challenging.

I wasn't involved in this case, but because she had very fatty breasts, the wire that was placed under ultrasound guidance got pretty much sucked into her breast all the way to the medial breast.

Although the cancer was speared the cancer, I just gave away the punchline.

But we were able to help guide the surgeon to remove this and get a final path diagnosis of invasive ductal carcinoma.

Summary

Just to summarize the three categories of pitfalls that you often see in the screening setting.

Mammographic positioning deficiencies, a source of increased recalls and false negatives are pretty easy to identify with those 10 clearly articulated potential positioning deficiencies that the ACR publishes.

And because they're easy to identify and they're so critical to rectify it's worth considering an audit of your group's positioning practices.

Second, I reviewed the importance of breast density, especially in the current climate.

And everyone understands why dense breasts limit mammographic sensitivity and that supplemental screening modalities may be considered for women at elevated risk.

But it's important to consider the risks and benefits of supplemental screening, particularly for average risk women.

And finally, subtle findings will always be subtle.

I would just leave you with the pearl to maintain a high index of suspicion, rely on standard operating procedures and consider regular audits.

Thank you very much.