Navigating Threshold Diagnoses of the Breast

So in very layman terms, if you think of a patient who presents with a breast mass or on radiology on screening mammogram, there's a mass, we don't know benign, malignant, or anything more than just there's a mass. So we have to get a biopsy. We have to get tissue, and it's on the pathologist to make that precise, accurate diagnosis to guide the clinical team.

So what are we really talking about here? So back to pathologists rendering an interpretation. It is an interpretation. So it is through the lens, through the prism of that individual. That's one element.

The second element is that there are certain diagnoses in the breast that just have inherent diagnostic variability. And part of that is the fact that we have an imperfect tool, and that is the eye. Right? So the eye can only see that far, we cannot see molecular alterations with just the naked eye. So some of those entities have overlapping morphology, they have overlapping immunophenotypical findings, they have overlapping molecular findings, and again, back to the eye, not being able to quite discern between one entity and the other.

And in general, when we look back over the past fifty years of breast care, we've really changed the way we manage our patients. In the 1990s, we were doing mastectomies for anything and everything, and now we've shifted to breast conservation surgeries, to sentinel lymph node biopsies rather than axillary lymph node dissection. Even trials at this time that are surveillance only trials for people with ductal carcinoma in situ of the breast. That means we're leaving this non obligate precursor lesion inside the woman or the man's breast and we're observing only. So things have shifted quite a lot, but it's really dependent back on us as pathologists to make that diagnosis to help our clinical colleagues to know which route to go.

So I mentioned a couple of entities here, and I'll just go back and review those. One of them was ductal carcinoma in situ, low grade ductal carcinoma in situ. That is a threshold lesion. It has overlapping features with atypical ductal hyperplasia. Some of us in the pathology world, my colleagues know, we refer to those as ADH and DCS.

And our criteria for those two lesions, how to discern them adequately and accurately, that criteria was developed in 1982 and later on with an update in 1990. And now it's 2025. So what we used to do before fifty years ago, forty years ago, doesn't quite give us enough information in the era that we are currently, which is the era of de escalation of treatment for breast cancer patients. So how does that again relate to threshold lesions? Why is that important?

Because we are increasingly asked to give more and more details to continue sub stratifying into more and more diagnostic buckets all these lesions and providing more detail to our clinicians so they can pick the correct route of treatment. And I'll just stop there and give you an opportunity because I realize I covered a lot of different topics in a very short amount of time. So I'm ready for a question if you have one.

Get to know that, build that brain bank of images, and then be prepared that depending on the practice you go to, where you are, you may have a substantial number of those threshold lesions, perhaps even up to fifteen percent. So a fraction of what you're going to see will fall into that, oh my god, I'm not quite sure. Is it ADH or is it DCIS? So I think, of course, know the good examples because we always compare abnormal to normal. So you have to know in your mind what abnormal looks like to make that determination.

And that's how I approach my career. That's how I approach my training. And then it was that extra layer of learning, and I would say that learning never really stops. It continues. It's a lifelong learning journey of what it means to make the diagnosis and how to fine tune and how to find that sweet spot of feeling comfortable with uncertainty. And that's what really threshold lesions are. They're uncertainty, and you, as a breast pathologist, have to find that comfort level.

Is this ADH as you've been talking about? How might one start to navigate that well?

We certainly have them here in Mayo and all of our campuses. Get that CK five, keratin five, get that estrogen receptor stain, and see what the lesion is staining like. Of course, there are other things you can try, levels. We like to think of lesions as two d, but in reality, they're three d. It's a formalin fixed paraffin embedded block that is approximately four or five millimeters, so there's definitely that Z plane, right?

Let's see how the tissue, how the lesion develops. Is it going to open up? Is it going to get smaller? Did we pretty much excise it with the biopsy or there's more that we're not seeing? So get those levels, get that basic workup.

That's the number one. People say, my more senior colleagues will say, I want to sleep on it. Right? So just think of the next day, look at those stains, and come back with fresh eyes. And I like to also say it's most important to get the diagnosis accurate rather than to rush it.

So let your clinical colleagues know that there may be a little bit of delay on the case because we're working it up. That is perfectly appropriate. So that's the basics. Now what else can you do to help yourself? And there are a few things you can do to help yourself.

Number one, don't be a hero. Don't commit to a diagnosis that is not quite applicable to this patient. So if you need to have a descriptive diagnosis on your final diagnosis, then utilize that. What are some examples? We keep talking about ADH and low grade TCS.

We know there's going to be a subset of those that will define accurate reproducible classification. So when you're faced with one of those, because it's a matter of when, not if, when you're faced with one of those, use a descriptive diagnosis. For example, atypical intraductal epithelial proliferation, C comment, or at least atypical ductal hyperplasia, or atypical ductal hyperplasia bordering ductal carcinoma in situ. Let your colleagues know that this is not your straightforward lesion. And that is a way to help yourself because we don't want to overcommit, and we don't want to be too brave with our diagnosis because you can imagine the significance of carrying diagnosis of cancer that comes with a lot of repercussions.

Right? A a lot of anxiety for the patient. There's a lot of treatment that's involved, and so on and so forth. So you're helping yourself, you're helping the patient, you're helping your clinical colleagues by admitting the challenges of the lesion. For more laboratory education, including a listing of conferences, webinars, and on demand content, visit MayoClinicLabs.com/education.

We've removed ourselves from being too aggressive in surgery, in oncology, in radiation treatment. And so in pathology as well, what we used to by the page criteria that we mentioned 1982 or so, 1985, around that time, that criteria tells us that as long as we have two millimeters to two duct involvement by these low grade neoplastic proliferation of cells, then we feel comfortable calling it DCS. In reality, I will tell you, myself included, my senior colleagues in pathology, we actually allow for more. We're pushing that border. We're moving that bar.

We allow for three millimeters, maybe four millimeters. We like to quote see more, and there's not really a a good definition of more look what more looks like, but we are pushing that bar. We're being safer. We are noticing that with all the advancement that's happened in the past, again, forty, fifty years in breast medicine, breast cancer mortality increases. So we've been excellent in catching smaller and smaller lesions.

We're talking about couple millimeters. Our radiology colleagues are phenomenal. They catch small calcifications, small lesions. We excise them. We treat the patient, and yet there's no really effect on mortality from invasive breast cancer.

And that's really what we're trying to do. We're trying to prevent breast cancer by treating ADH, atypical ductal hyperplasia, and low grade DCS ductal carcinoma in situ, the nonobligate precursor. Right? So maybe we're overtreating or maybe we're overdiagnosing these lesions and being a little too aggressive. So back to don't be a hero.

You're doing service to yourself and the patient by being careful and admitting challenges. And that I would also go on a little bit of a side note here and mention when you admit challenges, when you have a common that explains, hey, this is an in between lesion, this is a borderline, a threshold lesion, you prevent the misconception that another pathologist is wrong. Because oftentimes what patients do, and they're extremely smart, they will go to another institution, they will seek a second opinion. And depending on how the other pathologist is trained, remember, going back to this is just an interpretation, depends on the person interpreting it. Right?

The pathologist. Depending how that person is trained in their personal approach and understanding of the literature, they may be just a little more aggressive. And then two different diagnosis may happen. And then the patients may think, well, you know, one pathologist is wrong. Which one is it?

These are two separate lesions. And there's a lot more understanding than that that we don't convey, that these are related lesions from the low grade neoplasia pathway and that they're threshold lesions. Right? So by being careful and not being a hero, you're helping your colleagues and prevent the misconception of somebody, a pathologist, being wrong.

There's the other pathologist who, again, if the case goes out, the material is sent to another institution. There's the trainee. I talk to myself sometimes. I use certain terminology and certain sentences. So in a situation where I get a phone call, anybody asking about my report, I can tell you exactly what I saw because I understand what my words mean to me.

We speak to a wide range of people, the nurses or the techs who code these lesions. Right? They have to have a drop down menu. They have to code everything. It's very complicated.

So every so often, things will be conveyed in a way that's perhaps suboptimal because it's not a diagnostic bucket, and we like buckets because they follow an algorithm, a treatment algorithm. But we also have to admit that breast pathology has those threshold diagnosis and lesions, and it's just by playing the numbers. Eventually, you'll be faced with one of them.

And you could be the next person who cures cancer, I don't know. But I do think there is going to be a shift, and I think part of that is AI. We've seen AI help us on many different things and fronts, in breast pathology included in Mayo Clinic Florida. We do automated interpretation with AI for some of the biomarkers in breast cancer, for example, proliferation index, MIB. In Rochester, they do estrogen receptor, progesterone receptor HER2.

So it's extremely helpful because it provides that precision. It is not a subjective, it is an objective tool. It's reproducible, and it provides precision. I think there's going to be an improvement in the way we render diagnosis just by incorporating some of those AI tools. But I also think that there has to be an added layer, perhaps incorporating molecular into our diagnosis and not just immunophenotypic findings and morphology.

I think there has to be an added layer, and I know there are a couple of companies out there working on that. And hopefully in a couple of years, if we do this again, I may have some more information for you that, you know, we can share. But as of right now, work in progress.

And I think that pathology is one of these careers that I think there's a perception among medical students of, in the world of AI, you know, what is a pathologist going to do? What are your thoughts as an atomic pathologist in AI? You know, your answer there, you highlighted, was really talking about it kind of almost behaving as an adjunct of kind of integrating additional information, then you as the pathologist overseeing that. But what do you think that future interface is going to look like?

It's threatening our jobs. So in other words, this will be the solution to our problems, and it was not. And I kind of hear a little bit of that with AI where AI is out and it's going to solve all of our problems and pathologists will be out of jobs. Right? Something that people forget is that it is the breast pathologist, and because we're talking about breast pathology, but it's the breast pathologist who teaches that AI.

So who builds the ground truth for the AI? It is that pathologist. And I've had a couple of projects where we sit down and we annotate and we teach the algorithm what cancer looks like, what benign looks like. So there has to be a human element. There has to be a human element. And I think we need a lot of prospective data before we can remove the human element. Until then, I think we'll be okay.