Breast MRI Screening of High Risk Patients

Screening High-Risk Patients with MRI

Now we're going to talk about screening of high risk patients and look at the literature and also look at the kind of lesions that we would expect to find at high risk screening.

I'm going to show you examples from our practice. And many of these are in situ cancer, which of course is important to detect.

When we talk about MRI as a screening tool, I think we have to acknowledge that MRI is actually the most sensitive test we have for detecting breast cancer. It's more both invasive and in situ. It's the best test for detecting small cancers. It's better than mammography and ultrasound in this task, because I'm going to show you.

And its application is at the present time restricted to patients who are at high risk of developing breast cancer. And we'll define that. I suspect that in the future we will expand the list of patients for whom screening with MRI will be important.

Evidence from Mammography Screening

But as an overview, we have high quality evidence from a lot of randomized control trials as far as mammography screening is concerned across the United States and Europe, with over half a million subjects and all. But one showed statistically significant reduction in breast cancer mortality amongst those invited for screening mammography.

But none of them were designed to study high risk women. And we know that with mammography screening, we can organized mammography screening, we can observe downshifting of disease. As you can see here before and after screening, the percentage of in situ lesions is much higher mammography screening. And the amount number of patients with large cancers is significantly reduced.

We know, however, that the sensitivity of mammography decreases as the amount of breast tissue, so-called breast density increases. And this has been shown to be the case. And so the sensitivity for mammography screening in the general population with dense breast tissue is significantly lower. 62% is number here in this study.

Screening Strategies for High-Risk Women

So what are the screening strategies for women who are high risk? We know many of these women develop cancers at a young age when their breasts are dense, but they're higher risk for developing breast cancer.

Well, it's based upon less rigorous scientific evidence. And we know there's a recognizable group of very high risk women who have an early age diagnosis of breast cancer, that these tumors tend to grow rapidly and respond poorly to treatment if they're in advanced stage.

So the strategies we've adopted to define these women are more intuitive. However, we have done so, and we have various risk estimation models that are currently used to designate those women who at higher risk, mostly based on family history, and most of them do not include breast density at mammography as a risk factor, despite the fact that the density of breast tissue mammography has been shown to be a definite risk factor for subsequent breast cancer development.

Evidence for Breast MRI in High-Risk Screening

So the evidence we have in terms of breast MRI is, we have a lot of studies now that have been done comparing the sensitivity of detection of breast cancer in this high risk group of women with ultrasound and MRI. And I'm showing you some of the studies here.

The this is a large Dutch trial. This is a Canadian trial of trials in Germany. This is the UK trial. And if you look at the overall sensitivity of mammography across these trials, ultrasound and MRI, you'll see that MRI is clearly the winner here. And that mammography and ultrasound have very very low sensitivities of less than 40%.

If we look at the analysis of all 10 studies with 4,600 plus women screened. The prevalence of cancer detected with MRI that was not seen in mammography was 3%. And the sensitivity of all studies was 81%. And I think today we would say the sensitivity will be higher than this with modern MRI.

And adding ultrasound to the mix didn't make any difference, didn't help very much at all.

These are the kind of small cancers that MRI. If it's done well, we can find small cancers. We can find cancers that are smaller than five millimeters in our own high risk screening studies, which are currently unpublished, but will be presented at ASCO this year in our high risk screening study.

With screening high risk women every six months with MRI and annually with mammography. The average size of cancer detection in that study was seven millimeters.

So we can find very small small cancers. This was an MR biopsy lesion. It was about three to four millimeters in size. There's invasive cancer. Grade three, the kind of very small cancers that we really need to find at high risk screening.

Detection of In Situ Cancers

We can look at it used to be said that MRI wasn't good at finding in situ cancer. That really is not true today. And this study of Christian Kuhl, it was published in the Lancet, showed that MRI was very sensitive, was more sensitive than mammography in finding high grade in situ lesions as well as lower grade.

And that MRI, the statement was made should no longer be regarded as an adjunct mammography, but as a distinct method to detect breast cancer in its earlier stage. And I would agree, we find a lot in our clinical practice of in situ cancer that is mammographically occult.

The descriptors on the lexicon for a non mass enhancement, which early cancer we want to find are two. They are distribution and internal enhancement pattern. And here we see the distribution of focal linear segmental. These are probably the most important descriptors that would suggest an in situ cancer because of you following the anatomy of the ducts, linear, linear branching or segmental distribution.

And at three Tesla, we are now more confident in our diagnosis of in situ cancer. Because we can actually see the enhancing ducts themselves as I'm showing you here.

So this is an area of clumped enhancement. And you can see there is at least one major duct perhaps two with branching ducts enhancing all the way to the nipple. This has made a big difference in our degree of confidence because we're able to resolve ducts at three tesla, which we can't really do at 1.5.

So at 1.5 we may say the segmental enhancement, but now we can actually see enhancing TDLU and enhancing ducts, which really shows great promise for more specific diagnosis of non mass enhancement.

In the future, we would not expect to see this enhancing ducts in patients with just normal parenchymal enhancement. We don't see the enhancing ducts there, we just see the terminal ductal lobular units and the parenchymal enhancing. But when you see enhancing ducts which have this linear pattern, then you can be more confident that you're dealing with a true abnormality here.

The internal enhancement pattern is the other descriptor which you need to use these here are homogeneous, heterogeneous, clumped and clustered ring. These are the only four now that will be in the new lexicon.

And I suspect that clustered ring is actually a high resolution clumped enhancement. Because we actually then able to resolve more specifically the ducts and the terminal ductal lobular units at when we see clustered ring enhancement than we just see a clumpy enhancement.

And you can see here on this MR, where we have a low bar distribution of enhancing ducts and TDLUs. It's very similar to the appearance that we see when we do histologic three large section when we can see TDLU and ducts themselves. If you compare this one image with the other, so clustered ring has been added to the lexicon.

Examples of Screen-Detected Cancers

Okay, some examples of screen detected cancers here. Here we have an in situ cancer clustered ring, little ring-like structures in a segmental area. This is an example of a screening case where we only have actually one duct that I could find that's enhancing. Here's the duct itself. This is and here we have enhancing little ring-like structures which could be little terminal ductal units. This is DCIS case.

Remember, when you're screening, you have to look at every slice. You can't go through very quickly. You could miss this. This is only seen like on one and a half slices on our exam. So this is just one ductal structure, which we biopsy in our MR then was in situ cancer high grade, the first post contrast subtraction is often the best way of seeing these lesions because the parenchyma hasn't enhanced as much on your first post contrast subtraction.

And here you can see another example of sort of a clumpy enhancement and an extending anteriorly individual ducts can be resolved at three Tesla. Here's an example, sagittal reformatting showing the extent of the in situ cancer. And here you can see that we can resolve ducts themselves and we can see ductal enhancements that makes our diagnosis more specific.

Here's a screening patient just to show the mammography alongside the MRI, right mastectomy five years ago. And here we find a little focus of enhancement. It looks irregular. It's around five millimeters or just under this was biopsied and it's a four millimeter high grade in situ cancer. This is actually a vessel on end here. So these small little three four millimeter lesions are the ones we want to find. It can be difficult if you've got a lot of other foci to deal with and a lot of parenchymal enhancement. Again, the first post contrast is probably the most useful. The subtraction imaging.

And then look at the kinetics of the lesion as well. Here's another one. This patient is also 45, had a prior cancer here. She's had reconstruction of her right breast. We're looking at the left breast and again, what we see this is a source non subtracted image. We see some linear enhancement here. We have little focus of enhancement here, and then a little sort of linear and definite linear enhancement here. She had sequential screening. This wasn't there before. Here's her cc mammogram. We can't see anything abnormal. Here's just the one slice that shows this linear kind of branching. And then another focus anteriorly MR. Biopsy was done on this patient. You'll notice that there's no color on the angio on the angio map. This lesion did not enhance more than threshold, but it's linear. It got a biopsy. You see linear enhancement or clumped enhancement or clustered ring enhancement. You need to biopsy it. And this was high grade in situ cancer, 25 millimeters in extent. Found at screening.

MR show you another example. This patient had history of benign breast biopsies on the left for atypical lobular hyperplasia. Here's a mammogram, and here's the MRI linear enhancement. That's all. Just that little bit of linear enhancement small. Here. You can see it again on a slightly thickened. These are slightly thickened images. Be careful with in situ cancer. Because often you might see the sort of more obvious spot of enhancement. And then you might overlook some surrounding tissue. Actually this corresponds with this area on the sagittal views. And we thicken up the slides. We see another more anterior lesion, so you have to look more carefully. This is actually a segmental area of non mass like enhancement found at screening. Solid low to intermediate grade in situ cancer.

So these are the kind of little cancers that we're looking for. Here's a couple more screen detected in situ cancer. Here's a little clump enhancement here. And then some linear this linear enhancement. This was invasion, this was in situ. And here's another example of a 68-year-old with segmental clumped enhancement. Linear. This is 1.5 Tesla, not three T but here you can see some linear enhancement. And then also seary foci high grade in situ cancer.

This is a patient with a nightmare mammogram. Look at the density of the breast. But here's her MR, her subtraction imaging. And we can see that she's got some non mass like enhancement laterally. This is in our screening group here. You can see it on the first post contrast subtraction image. A little clumpy area of enhancement. And here you can see it on the sagittal and the crl. It's helpful when you see it corresponding in size and location approximately with the additional imaging. And here is her angio map. And it's got a heterogeneous pattern with mostly persistent, but some plateau type enhancement. And this is in situ lesion, again associated with a two millimeter invasion at final excision. This was biopsied. All of these lesions I'm showing you are all biopsied under MR guidance. Most of the lesions that we send to MR biopsy where we don't find any ultrasound correlate actually in situ.

You know non mass lesions, 32 personal history. These are all screen detected cancers in our practice. Here you can see invasive cancer and some of these invasive cancers in these high high grade women, particularly triple negative cancers, for example, may look like sort of oval masses and may fool you into thinking they're fibroadenomas, but they're not. Here. You can see the kinetic data wash out curve. IDC grade three here. Totally. These are all totally invisible on mammography.

Here's an example. We've been screening every six months in our practice. And this was a case where this little focus was overlooked, was not diagnosed, but six months later had grown and was definitely abnormal at that time. And had a sort of site wash out curve here. Plateau mostly. So this is kind of the natural history of tumor growth. We didn't call it here, but we did call it a six month interval. So if you have these little foci and you're not comfortable with them, as I mentioned earlier, you can bring the patient back in the optimal time, the menstrual cycle and re-image and put a needle into them if need be. Or do a six month follow up. And look for any change. This was a microinvasive ductal cancer.

Clinical Trials and Cancer Yield

There's a very nice trial, recent trial called the AVA trial published in JCO in 2010, where they looked at the cancer yield a screening study women who have more than a 20% lifetime risk. So this would correlate with the risk that the American Cancer Society recommends for high risk screening in this country.

1,679 screening rounds, annual mammography, MRI and ultrasound every six months. And here you can see the cancer yield from MRI, MRI plus ultrasound. So ultrasound didn't do much for us. And MRI plus mammography. Mammography helped a little. So the cancer yield again, supporting the clinical. What's known clinically as the MRI is the most sensitive method. And here you see the 27 cancers. MRI found 93% mammography. 33, ultrasound 37. So added value of mammography and ultrasound is very small. In fact, some investigators in Europe are postulating that we shouldn't even use mammography. And we should just screen with MRI because they're concerned about radiosensitivity of the breast in these younger women.

So MRI sensitivities have an increased cancer yield. We detect cancer at an earlier stage. And as the DUTCH study showed that we have more than a 50% decrease in the positive nodes compared to the control group in the women that were screened. And here are examples of two more women in our study, little rim enhancing mass. And here's another invasive cancer that we found.

Specificity has been variable in terms of recalls and biopsies, but overall I would say that a specificity in all studies is reasonably high. And certainly in our practice specificity of MR biopsy. The cancer detection rate, the PPV three is comparable and if not slightly better than that from mammography and certainly better from that for ultrasound screening.

Cancers in Young Women and BRCA Mutations

Just a word or two about these cancers diagnosed in young women below the age of 40. And for five to 10% of them have germline mutation. 16% of African American women less than age 35, BRCA one when they're diagnosed below the age of 40. There's a high prevalence, interestingly of Ashkenazi type mutations among Hispanic women as well. And 60% of the carriers have no family history, first and second degree, which I found to be surprising.

These tumors in BRCA one tumors have a distinct phenotype. They have a distinct MR phenotype as well. They tend to be medullary or atypical medullary. And they have a high mitotic rate as evidenced here by all this washout within the lesion. Aneuploid, high proliferation fraction, very aggressive tumors, estrogen receptor negative and no HER two gene amplification. And they often have TP 53 mutations.

So on MRI they tend to present as masses. They often can look at very round masses. They do look round or oval on ultrasound and they may fake you into thinking they might be benign. But they fortunately have very high uptake of contrast on MRI. And this interesting paper in 2008, looking at familial breast cancer phenotypes, showed that the posterior location in 67% of these cancers in high risk women, which again makes them more difficult to detect mammographically because they have posterior location. Many of these are basal like cancers.

And so we have to remember in a lot of studies now being done looking at the phenotype, the features of lesions, comparing with them with a clinical molecular and genetic phenotypes. And this is certainly true for this group of lesions. It's very interesting to compare. There's a study in radiology by Francesca Sardanelli, who's done a lot of the screening work in Italy. And looking at the incidence of triple negative breast cancer, showing that in the BRCA one group is a very high incidence of triple negative 67% in BRCA one patients, BRCA one mutations. Where this is not true in the rest of the high screening cohort, where most of the cancers are not triple negative, more than half and not triple negative.

Updated Guidelines for MRI Screening

The updated guidelines the American Cancer Society are recommending MR screening for patients with the genetically predisposed to develop breast cancer first degree relatives of those women who may for one reason or another choose to be not tested approximately 20 to 25% or greater estimated lifetime risk by model patients with radiation therapy to the chest. Usually those are young patients with diagnosis of lymphoma, beginning at age 30 as long as the patient's in good health. Those are the current recommendations.

Just to take this another step, we know this is the best method for diagnosing early cancer. So Francesco Sardanelli said last year that if a breast MRI's negative, the added diagnostic value of mammography and ultrasound is not significant and that we're thinking too much about the old paradigm for screening and high risk screening as an adjunct to screening mammography. Whereas really it's an independent tool for diagnosing breast cancer.

And should we be using MRI alone and could we use MR as a population based screening tool? And this is work from Christian Kuhl where she does a three minute breast MRI so called, I mean I think we couldn't do it as that fast, but certainly you could do a limited study where you just did the T one post contrast limit the number of acquisitions and so look and see this is her work showing normal or abnormal three minute study, abnormal cancer, normal, abnormal normal, you get the idea.

Certainly we could only do this if the cost came down. And this is a topic of considerable interest in our practice and trying to look at ways where we could generalize more generalized MR screening to certainly to that intermediate group of women who are not eligible for MR screening given current guidelines either costs, these are the costs of MRI compared with the costs of mammography. But bear in mind, if the group is, if the high risk group, the yield is much greater with MRI. So that offsets the cost sum.

These are the estimated costs of implementing the breast MR screening guidelines as they currently exist in the US 1.4 billion a year would be the cost of screening the high risk group of women with 429 million for follow-up costs.

Other Groups Benefiting from MR Screening

Now, just in the last few minutes, I just want to talk a minute about the other groups of women who might benefit from MR screening. In addition to the women who have documented high risk and who have fit the current guidelines, the American Cancer Society said there was insufficient evidence to recommend for against screening of women who had LCIS ALH ADH, high risk histologic diagnoses from prior breast biopsies. Women who had very dense breasts mammography, and women who had a personal history of breast cancer.

In this group, I think we have some evidence that more of these women should be screened and we currently are screening in our own practice. Our oncologists recommend screening surveillance with MRI for all women who are diagnosed with breast cancer prior to age 40 respective of what their risk factors come out to be. And there are some papers that support that.

And so should all women with a personal history be screened? And these are just some examples of some scenarios where this kind of screening might be appropriate. Women with multifocal disease treated with breast conservation women with a diagnosis of ILC women whose cancer was mammographically occult at diagnosis. There's a paper by Sandra Brennan in AJR in 2010 looking at this, looking at women who had no family history but all had a personal history of breast cancer, so they wouldn't have fit into the conventional high risk estimates of being risky enough.

And she showed that. And so she did MR screening in this group and found malignancy in 12% with a positive predictive biopsy of 39%. That's pretty good. So this suggests that this group of women might be appropriate for surveillance, at least for a while.

And this other paper, there's a degree of background enhancement. You know, I've talked a lot about that today affect the detection of small cancers at screening. In other words, how limited are we? And this is a paper. These are our classifications of minimum, mild, moderate, and marked. There's a paper in the European Journal of Radiology looking at this showing that sensitivity was reduced in women who had moderate marked enhancement some. So we have to do our best to sort of minimize that if we can.

Role of Breast MRI in Borderline Lesions

And then finally, a subject, dear to my own heart, the role of breast MR imaging for predicting malignancy at histologically borderline lesions diagnosed core needle biopsy. You know, it was a I was giving a lecture last year and there was an old surgeon sitting in the back and he heard me talk about staging and he got up and he said, you know, Dr. New said, he said, this is all very interesting about staging. Why can't you do something about these high risk lesions I keep having to operate on and I don't find any cancer? Right? And so he has a point, you know, we surgeons don't like operating on women. They don't have cancer. And we know that women who have high risk lesions from prior breast biopsy, ADH ALH, whatever, most of them do not have cancer. 80% of them do not have cancer at final surgery.

So can MRIs sort out those ones that really do need to have surgery? Because they do have cancer than those who just have high risk lesions. This study from Dr. Ponzone in Italy looked at this 32 high risk lesions and found that MR was pretty good at doing this. And there was only one false negative, which she didn't detect on MR. And that was a low grade DCIS, which you we could argue about.

And there's another paper my doctor Sung saying that, why don't we screen with MRI patients with a history of high risk women? And she also had pretty good results. 20% of the cancers that she found in follow up were found by MRI alone. So these are two groups of women, I think the intermediate risk group of women where MRI might be useful in directing patient management. MRI in this study resulted in a 4.5% incremental cancer detection rate. So maybe we could triage patients from surgery to watch for waiting if MRI is negative.

Summary and Recommendations

And finally, it's important to do a breast MR audit. If you're doing MR screening, we want to find an MR. A high percentage of small node negative cancers expect a higher yield than with mammography and ultrasound alone. And perhaps appropriate benchmarks based upon literature might be a cancer yield at our initial examination of 22 per thousand and subsequent examination 10 per thousand at an acceptable cost measured by recall rate and by biopsy recommendations.

So that's a summary of cancer screening, how to do it, how to find small, early lesions. Many of these are going to be in situ. And I think we need to get the word out in the radiology community more than we currently have to try and identify this women who are truly high risk and might benefit from this kind of screening with MRI. Thank you very much.