So, first, the evolution of therapy. It's been really exciting to see and to be breast imagers working in these multidisciplinary communities, where we've moved from the radical mastectomy to Bernie Fisher, showing that lumpectomy with wide margins and whole breast irradiation in a select group of patients is equivalent to mastectomy. That really opened up a whole new world. And it was such an exciting time. Can you imagine being an imager then, where you were finding cancers that were so much smaller with screening mammography? Patients were having options with that small cancer not to have their whole breast removed. As the imager, you were working so closely with the surgeons who were trying to preserve the breast to make sure that with your imaging tools of mammography and then ultrasound, you were defining the extent so they could be more precise in who was a good candidate for breast conserving surgery and who really needed to have a mastectomy as the first surgery. Today, it's a whole new world. We have a lumpectomy with no ink on tumor for invasive disease, and at most centers, two millimeters for DCIS. Partial breast radiation is being offered more and more. And we've extremely relaxed the criteria for patients that can opt for breast conserving surgery. No longer are we in the days of, it must be a very small tumor. You can't have two sites of disease. We've really relaxed those criteria, which behooves us to have very precise imaging tools to guide this new era in therapy. The problem is that the evolution of imaging, this advanced MRI, the superior diagnostic accuracy, it's been out of sync with this evolution of treatment. So if we go back to the days where women were all getting radical mastectomies, we couldn't support that with mammography or ultrasound or MRI. But as we moved into what we now see as the traditional approach to lumpectomy, breast conserving surgery, and whole breast radiation, we really were seeing the incredible power of the mammogram and the ultrasound to guide that process. Now that we've moved into modern precision treatment where we're trying to be much more precise and individualized and targeted with our treatment, we need to bring MRI into that. But we've sort of stayed back in many, many centers with mammo and ultrasound. I think in hindsight, we moved to measuring the impact on treatment of advanced imaging with MRI prematurely. And we relied on observational retrospective studies. There was a selection bias inherent to those. The patients that were undergoing MRI were different than the patients that were not undergoing MRI. And correlation is not causation. So this whole domain of MRI led to more mastectomies, was your classic error that you make when you think that correlation is causation. In hindsight, we also performed prospective trials before clinical centers were meeting what we think of as quality standards. Socomi's trial is an excellent example of that. There was no MRI guided capability in those centers. How could we imagine that we would have positive outcomes for a trial that didn't have what we now require centers to have to be accredited by the ACR to perform breast MRI? We also applied requirements for evaluation of new treatment. There's one set of metrics that you would use and research methods you would use to evaluate new treatment. We applied that to evaluating new diagnostic tests. In addition, diagnostics were being paired with fairly sloppy therapeutics and that was unlikely to have the impact for our patients that we wanted. Christian Kuhl has spoken very eloquently about this, pointing out that the University of Oxford Center for Evidence-Based Medicine acknowledges how different we must be when we are measuring the impact of treatment and the difference in how we evaluate a diagnostic test. Clearly MRI as a diagnostic test meets level 1A criteria by the University of Oxford Center for Evidence-Based Medicine. If we then try to take that clearly improved diagnostic test and apply it to treatment where we have such wide variation in surgical styles and performance with or without MRI, that's unlikely to translate into those outcomes we want. That's going to be much more complicated and has to follow a different pathway than just giving someone MRI and seeing what happens at the end of the day. Reoperation rates by surgeons range widely from 0 to 70% in multiple large multi-center trials. So this very widely acknowledged and accepted variation in re-excision rates just as a start is a challenge to then try to use MRI to impact and change patient outcomes. So that sounds a little depressing, but it's going to get a little bit better. Let's look at a few cases for those of you who are just interested in if I'm using MRI in my practice, how do I approach the patient that has a recently diagnosed breast cancer and is in that preoperative phase? So this is a 49-year-old woman. She presented with a palpable thickening in her left breast. The BB shows the area that she was feeling. And we can see that she has this irregular mass. It was deep to the nipple in the central portion of the breast. It's very obvious on ultrasound as well. This is now biopsy and we have infiltrating lobular carcinoma. So we have a 49-year-old patient with heterogeneously dense breast tissue infiltrating lobular carcinoma. We're not going to leap to MRI because we want to do a thorough diagnostic mammography and ultrasound exam first. And if we do our diagnostic evaluation, and you might even have caught your eye that besides the obvious palpable lesion, there was a satellite lesion that's just at the edge here and that we can demonstrate. This is still within the equivalent of a quadrant or in a center. So we still have unicentric disease even if we don't know what this is. This hasn't been biopsied yet. But the question is, would you, just by a raise of hands, how many of you would proceed with biopsying both at the same time versus biopsying the one and then opting for MRI? So just show of hands that would biopsy both. And that would wait, do the one, and then do the MRI. All right, so most would biopsy both. I think this is a case-by-case, knowing your surgeons, knowing your practice, how you manage these patients. You're in ultrasound. You're right there. You can sample both and confirm them. In some centers, they're going to sample this to have that known cancer and then have the MRI to have more planning. And that's really where it gets into what I think is the fun part of all of this, to really be using our brains, having a good sense of our local surgical practices, what they're looking for, how we need to define the extent of the disease and move forward. In this woman, the MRI also showed that she clearly had multicentric disease and was not going to be a candidate for breast conservation. Of all the areas, I think, that are most complicated in the preoperative setting, it's this basic concept with BI-RADS. When I have a patient with a biopsy-proven cancer, when is it, overall, the exam, a BI-RADS 6? And when is it a BI-RADS 4? In some centers, it will routinely assess a patient with a current breast cancer diagnosis as 6. And you could imagine that that could be a confusing communication to the referring providers because 6 is, they have cancer, there's nothing else that needs to be done. And it would also render it impossible to do any kind of auditing because you wouldn't be able to distinguish between the patient that had the one already known biopsy-proven cancer, that's your BI-RADS 6 lesion, and other sites or other areas that you were recommending biopsy on. Christophe Lee, Wendy Demartini, and others at the ACR Breast MRI with Guided Biopsy course actually studied, you know, what kinds of cases were most challenging. And definitely the diagnostic setting was more challenging for participants in this course, radiologists out in practice that are doing breast MRI. And one of the more challenging areas was in this preoperative setting. So the rule that we use is the most suspicious finding determines the BI-RADS score. I actually erased this slide. It used to say, most suspicious finding trumps, exclamation point, and I decided we're going to use another verb. It's going to determine our BI-RADS score. Patient with biopsy-proven cancer, that's your BI-RADS 6 lesion and a separate lesion that is suspicious for but that is not known to be cancer, that's going to be a BI-RADS 4 or 5 as your final overall assessment. So there's a highest order assessment and recommendation where 4 or 5, recommending biopsy, is a higher order than 0. We don't use 0 very much with MRI and very little in the preoperative setting, but there might be cases where you wanted more imaging, such as an ultrasound, before making a final assessment of that MRI. A BI-RADS 3, also rarely used in this setting, but that would be the next order. And then BI-RADS 6. BI-RADS 6 is saying, known cancer, proceed with treatment. Nothing else to do that's actionable on this MR. And then 1 and 2 wouldn't be used. So we have a checklist. We confirm that we know the location, the features, and the size of the biopsy-proven cancer. That's our BI-RADS 6 lesion. We check the history, the prior imaging. We want to make sure that we don't assume that that big enhancing mass on the MRI is the biopsy-proven cancer. Maybe it was a patient that had a small cluster of calcs that were DCIS and that mass has actually not been identified yet by conventional mammography and ultrasound imaging. We then ask the question, is our MRI finding consistent with the already biopsy-proven known cancer in the location, in the features, and in the size? And is there another lesion that warrants evaluation? And this concept of, is there another lesion, is basically, is there a lesion that's distinct from the known biopsy-proven cancer? Clearly if the lesion is suspicious and it's in the other breast, that's going to be a BI-RADS 4 or 5, not a 6. If it's in the same breast but it's in a different quadrant or the equivalent of a different quadrant, again, that would be actionable. Just because something is enhancing and looks suspicious does not mean that it is cancer. You want to have either targeted ultrasound-guided biopsy or MR-guided biopsy to distinguish an enhancing lesion that's malignant or benign. And then, is it more than two centimeters from the known cancer? And it's this last one that you can use some of your judgment on a case-by-case basis, but it's a good rule of thumb. So with the E4112 study, which I'll show a little bit of the details about in a moment, we took that and we adjusted it for our clinical practice. So if a patient undergoes MRI and the findings that we see on MRI are completely concordant with the standard imaging, that is our BI-RADS 6, if there are additional findings that if those additional findings were malignant, they would change from breast conserving to a mastectomy, that is going to be a BI-RADS 4. We want to flag that case as such. If additional findings, if malignant, would not convert the patient to a mastectomy and they're within two centimeters of the known cancer, we flag the size differential to the surgeon, but we still refer to that as a BI-RADS 6 because we don't think that that necessarily is going to need further intervention. But if it's more than two centimeters away from the already documented malignancy, we're going to flag that as a BI-RADS 4. The surgeon may decide in reviewing the case that something that's three centimeters away in a specific patient with a specific breast size, depending on the size of the primary known cancer, that that doesn't actually warrant more imaging or more image-guided biopsy and they'll just take a larger sample. And that's where you need to know your surgeon and have those conversations. At our center, when there are additional findings that were extra mammary, such as suspicious axillary lymph nodes or other extra mammary findings, we flag that to our surgeons, but we don't give that a BI-RADS 4. We assess that as a BI-RADS 6 with a descriptor and flag of those extra mammary findings. As George Bernard Shaw said, the single biggest problem in communication is the illusion that has taken place. Sometimes we think our reports are being understood by the surgeon for our intentions. If we have the luxury of being participants in multidisciplinary conferences, that's a great time to make sure that our communications are clear. If we're in a center where we have engagement with the surgeons in person, that's terrific. But I really think that much of the pre-op MRI setting and the challenges have been in the communication. And so the more locally you have an understanding with your referring surgeon, the assessments that you have will be influenced by your individual referring surgeon's preference. And the caveat, never to make a decision to convert a patient who is destined for breast conserving surgery to mastectomy based on imaging alone. The pathology confirmation is really essential. In a rare case, a patient may opt, may decide to undergo mastectomy, but we don't want them to be doing that because of an appearance on MRI. We'll talk a little bit about that as well. So this is another case. This woman was 45 years old. There's a mass on mammography and ultrasound. It was biopsy confirmed IDC with also ductal carcinoma associated with it. And this is a bit of a flag when we have associated DCIS. On the mammogram, we could see some calcifications extending about two centimeters from the mass that was biopsied. So again, the question is, do we biopsy these calcifications now that are about two centimeters away from the known IDC mixed with in situ disease? In this patient, we had the MRI first after we confirmed that this was cancer. And this is the appearance of that MRI. So this is our BIRED6 lesion, the biopsy proven IDC with associated DCIS. There was extensive clumped non-mass enhancement extending all the way from inside the nipple back to the pectoralis muscle. It did not extend into the chest wall. And then also a small lesion in the opposite breast. This small spot here was a very clear intramemory lymph node on the MRI. So that was not assessed as suspicious. But clearly, this non-mass enhancement was, as was this focus of enhancement in the right breast. This was biopsied under, was identified under ultrasound and biopsy and shown to be a fibroadenoma. This area was biopsied and showed to be extensive high-grade DCIS and a decision for mastectomy in this patient was made. We identified an area of the non-mass enhancement that was clearly in a different quadrant from the known infiltrating ductal carcinoma. So let's look at the research, which has been very consistent. It's amazing how many decades people have been studying and presenting in a diversity of settings the same results. And this is the very basic question. Will MRI show us additional disease within the breast with cancer after a mammographic and traditional ultrasound evaluation is done? The numbers vary because in each of the studies, the patients that were selected varied. And the definition of additional disease within the breast varied. But there's a very consistent theme. In every case, additional disease is identified. And in most, especially of the more recent studies that have been done, they were very strict in their criteria for defining additional disease in the same breast. So there should be no question about this at all. Sometimes when there's confusion, and the confusion stems from the impact on surgery, people think that, oh, MRI and the pre-op setting, everyone's all over the place. This is extremely consistent findings that you will. It is the most sensitive tool we have to identify additional disease in the ipsilateral breast and the contralateral breast. The same, and people continue to study this. At this point, we're starting to see an appropriate maturity and evolution of the right types of studies because we really don't need to study anymore, will MRI find more disease in the ipsi and contralateral breast? We know that. We need to understand better how to partner in a multidisciplinary approach in using this fantastic tool in a way that improves our patients' outcomes and supports less aggressive more precise treatments. We also know, because sometimes people worried when they started to see these studies, well, these are at the academic centers with highly specialized breast imagers with the best of the breast MRI, but we've had many, many centers, and we're grateful for this, that took the time to report on their experience in community practices when they had sequential patients having preoperative MRI. Very consistent findings, and in general, we're going to see roughly 5% to 10% of patients will have additional cancer within the ipsilateral breast and 3% to 4% in the contralateral breast. Given this, why has it been such a rocky road to implementation? Why is it that neither the ACR, nor the SBI, nor USOB, nor the NCCN say this should be done? This should be part of the preoperative pathway, diagnostic pathway for our patients. And each of them use the same expression, acknowledgment of sparse quality data on long-term outcomes. They sort of support the use. They support consideration of the use, but they're reluctant, understandably, to say this is a diagnostic test that the woman who's been recently diagnosed with breast cancer should undergo. We certainly have some extremely strong studies to help guide us in how to make decisions for these patients. Chris Cheung Kuhl's excellent paper on the impact of preop breast MRI and MRI-guided surgery on diagnosis and surgical outcomes of women with invasive breast cancer with and without a DCIS component is really used by many centers where the surgeons don't understand still the high, high impact of MRI in the patient with DCIS. We still have this outdated urban myth that mammography is a tool for DCIS, MRI is for invasive disease. MRI in patients with intermediate and high-risk DCIS can be a very powerful tool. And Janice Sung and the team at Memorial Sloan Kettering looking at the effect on surgical and long-term outcomes, showing that re-excision rates were reduced, mortality was not, which we wouldn't expect in this patient population. But this is part of the confusion that many people have where they have a plethora of studies that say it finds more disease, but it doesn't change re-excision rates, it doesn't change recurrence rates, it's actually not impacting the patient. And we go back to the study designs, what's needed, and the influence of that imaging on the behaviors of the surgeons, of the radiation oncologists, of the medical oncologists. We're still pairing advanced imaging with fairly old approaches. So if we're in a center where they have very large margins, everyone's getting whole breast radiation, that was an approach based on just having mammography and ultrasound available. Now we can do better, we think, with this advanced imaging. I'd like to finish with three trials that I think are the types of trials we need to see more of. This Swedish trial, a prospective multicenter randomized trial, there's going to be very rich data that we'll continue to learn from. Lower re-operation rates in the MRI group versus the no MRI group, 5 versus 15 percent, and no difference in the mastectomy rates. Francesco Sardinelli from University of Milan, the MIPA trial, preliminary results of over 2,400 patients, the MRI reduced the re-excision rate from 13.4% to 8%, very similar to the Swedish results. And what was also interesting is they looked at not only the patients that had more extensive surgery but also less extensive surgery after the pre-op MRI. And so there was a balance there. And some of the studies only looked at patients that were upgrading their surgery but didn't look at patients that were having less aggressive surgery in this setting. ECOG-ACRIN decided to have a trial. My co-chair in this was a fantastic surgeon, Seema Khan, and we were addressing how we could have a trial that really combined advanced imaging with advanced diagnostics. In this particular case, the 12-gene oncotype DCIS expression assay. In this trial, we discovered several things that we thought was important. One, this was community practices that were engaged in this. They performed at a very high quality level in performing the MRI. Just about 19% of patients had additional biopsy performed, so over 80% just proceeded with their surgery, and 32% positive predictive value. So we felt like this was a performance in the diagnostic space that was robust and reproducible. We thought it was important to see of those patients who were destined to have breast conserving surgery and opted for mastectomy as the first surgery, during this trial we stopped and assessed the surgeons, the patients, the data that was available at multiple time points on the road from first hearing that they had cancer to having their final surgery. And that was important because we wanted to see when and how these decisions were being made. Of the 54 patients, out of our total of the 339 that entered the trial, of those 54 that went to mastectomy as their first surgery, half were based on the MRI and half on patient preference. 5% was because of a genetic history that was discovered after they entered the trial, and 3% a contraindication to radiation. Of those patients that then had mastectomy after a first attempted wide local excision, in 10 patients it was because of a positive margin at the attempted excision, and one it was patient preference. So this was the first prospectively collected multicenter data evaluating the impact of both breast MRI and a multigene assay on surgical treatment and on radiation treatment, and we want to see more of these studies that are truly multidisciplinary, looking at the impact on the advanced imaging on treatment choices. Approximately half of the 16% of women who were eligible for breast conservation based on conventional imaging proceeded to mastectomy, required this based on MRI findings, and about half opted to have a mastectomy based on personal choice. Among the women with DCIS who were candidates for breast conserving surgery and had MRI, successful breast conserving surgery was achieved in 81%, and radiation therapy was avoided due to a low DCIS score in 27% of those patients. So this is a new paradigm in using our advanced diagnostics, our advanced imaging, and careful cooperation and communication amongst all the members of the multidisciplinary team. This is where we can move to. I also thought it was important to note that 97% of women who had wide local excision as their first surgery after they were cleared by MRI achieved success with wide local excision as their final surgery. So to be able to tell a woman after a clear MRI, they have a very high likely opportunity to be a good candidate for wide local excision. Some women will make a decision to proceed with mastectomy because they're worried of how likely will they have a successful outcome, and almost 80% that was a successful wide local excision was with the very first attempt. So we know that MRI has superior diagnostic accuracy to define extended disease. That's unquestioned. We have lack of synchronization of precision imaging with precision treatment, and that slowed effective clinical implementation. As our treatment paradigms for breast cancer continue to evolve in this era of less is more, active engagement of breast imagers and the multidisciplinary team is imperative in our guidelines, in our clinical case conferences, in our referring provider and patient education and communications. Last slide, should all women eligible for breast conserving surgery have preoperative MRI? At this time I think it's a local question. There is no rule for all the centers out around the globe that have the access to MRI. Know your multidisciplinary team, provide open access to the exam and to biopsy, audit your practice, share your results with your referring providers early and often, and if you're just starting out, really think about the biology of the tumors, infiltrating lobular carcinoma and intermediate high-grade DCIS, your patients with dense breast tissue, mammographically occult primaries likely will have the highest impact in your early adoption and transition into this. So when someone is diagnosed with breast cancer, I think the first thing to do is to take a pause and look at her initial imaging. And you can look at the screening and diagnostic mammogram, and perhaps there's another cancer that you may find. And here's a 52-year-old woman who had a new diagnosis of right breast cancer, and she has a heterogeneous complex breast parenchymal pattern. And on second look and review of her mammogram, we see a stable fibroadenoma, and slowly scanning through her tomosynthesis slices, we actually saw two additional lesions, had to convince ourselves that, in fact, perhaps there was a speculated mass here and an architectural distortion out here. So we did an ultrasound, a targeted ultrasound, and scanned pretty carefully to find these two separate lesions. I can tell you that she had a screening ultrasound before that, and these lesions weren't seen probably because they were kind of close together. And so it was just thought that it was just some normal tissue. And so this turned out to be two additional grade 2 infiltrating ductal cancers. We can use ultrasound for problem solving as well. This is a 64-year-old who had a palpable neck mass. She had an abnormal level 4 neck node, and there was a core needle biopsy that was done, and it was shown to be a metastatic adenocarcinoma, and there were no squamous or glandular features. But on physical examination, she had erythema and subtle skin thickening, so a mammogram was done. And you can see that there is an enlarged intramammary lymph node as well as an abnormal axillary lymph node. And looking and comparing with her priors, you can also see that there's some increased subtle focal asymmetry in the upper outer quadrant of the right breast. So we went ahead and did the ultrasound, and it was pretty easy to find the intramammary lymph node here and the abnormal axillary lymph nodes. There's no fat. There's simply pathologic lymph nodes, but we didn't stop, and we kept looking. And posterior in her right breast, at 10 o'clock, we see hypoechoic shadowing mass, and that was the primary cancer. This was a grade 2 infiltrating ductal cancer. She had neoadjuvant chemotherapy and had mastectomy and complete response. So I think that this cancer probably was not included on her mammogram. So what is the role of ultrasound in preoperative clinical staging? I mean, there are certainly advantages of doing a complete whole breast ultrasound in these patients. We can detect multifocal and multicentric disease. Wendy Berg showed this way back in 2000 prior to the widespread use of MRI that we can see invasive tumor foci much better than mammography, 94 percent of the time on ultrasound versus 81 percent on mammo. We also see infiltrating lobular cancer better on ultrasound as well. Ultrasound has the advantage that we can assess the regional nodal basins, including the axilla, the infraclavicular, and internal mammary regions. And of course, ultrasound is well-tolerated by the patients. It's relatively inexpensive. We have real-time evaluation. And if we see something, we can often just do a pretty quick biopsy. Tissue sampling is easy to do and easy to schedule. The disadvantage, of course, is that the sensitivity is not as high as other modalities, such as MRI, in general. There are also false positives. And there's really no clear benefit in finding these additional lesions as far as long-term outcomes, recurrence, and survival. So here's a 75-year-old woman with a new diagnosis of right breast cancer and dense breasts. And you can see that there's a spiculated mass in the upper aspect of the breast, probably out here in the lateral aspect of the breast, but masked by the dense breast tissue. She had dense breast, so we did a whole breast ultrasound as well. And we found a second lesion in the left breast, posterior, and it turned out to be back here. But we did not see this initially on the mammogram. And so we biopsied this, and this was a grade 2 infiltrating lobular cancer, in addition to the grade 2 infiltrating ductal cancer that she had on the right side. So most of us are familiar with the TNM staging. It's easy to find in review articles and in textbooks. But just be aware that last year it was updated. And the eighth edition includes clinical staging and prognostic factors. Clinical staging is demarcated with a C. That means the clinical stage prior to surgery or neoadjuvant systemic therapy. And so this actually acknowledges, I think, that imaging does play an important role in clinical staging. Pathologic staging is demarcated with a P, and that's after surgery. And if the patient had neoadjuvant therapy, that's demarcated with a Y, so you might see a PY in the stage. So tumor grade receptor status and genomic assays was also added to the new TNM staging. And this all reflects the changes in knowledge in the biology of breast cancer. And that's optional if it's available. So here's a 53-year-old woman with a new palpable right breast mass. And you can see it here. It's pretty obvious on her ultrasound that she likely has a cancer. This is BI-RADS 5. This lymph node on her mammogram doesn't look terrible, but it looked abnormal on her ultrasound. The hilum is displaced. The cortex is thickened. We did a whole breast ultrasound to stage as well, and we found additional lesions. Another one on the right and two on the left. So technically, you might say that this woman could be a candidate for five biopsies. But there's no way we're going to do that, right? There's no way she's going to tolerate it, and there's no way that I'm going to tolerate that. So what do you do in this case, right? So we just did what we thought was the index lesion, and then we did the lymph node, and we picked the most suspicious lesion on the left. And they were all positive. She had a grade 3 triple negative cancer on the right with axillary metastases. And then on the left, she had a grade 3 DCIS. And that was enough information for the patient, as well as her surgeon and oncologist, to make the appropriate treatment and management decisions. So the downside of this extra imaging is the false positives. So the more we look, the more we're going to see, and we're going to get some false positives. This is a recent case of mine, a 74-year-old with a screen-detected papillary infiltrating ductal cancer. Here, there was this subtle oval mass that was actually speculated, seen really on the tomosynthesis views. And so we did our ultrasound, and we found additional lesions. There was a second nearby hypoechoic lesions. And then at 12 o'clock, just one hour away from the index lesion, there was a second spot. So we thought, at least I thought, that she was going to have multifocal disease. But it turned out she had a papilloma and fibrocystic changes. But the advantage of this is that when we did the ultrasound, we could do the biopsy right away. So it didn't take that much time, and it didn't delay her treatment decisions. And the surgeon was very confident, then, that they could just take a small amount of tissue and have negative margins. So the big competitor is MRI, of course, as Connie showed, that there's high sensitivity. There are disadvantages. Our surgeons really don't love MRI. We do it sometimes. There have been reports that there's increased mastectomy rates in these patients because of the additional lesions we might see on MRI. And then you have to have a second look ultrasound, and then if you don't see it, then you have to schedule them for MRI-guided biopsy. And these patients are already anxious, and this could go on for a week or two. And by the time they're through it, they'll just say, I'm done, and they might have a mastectomy. And that's something that we have seen has occurred. So it's time-consuming, it's expensive, and again, no clear benefit on the long-term outcomes. So in our practice, our surgeons will order it sometimes. They order it really for the tricky cancers, the infiltrating lobular cancers, cancers in BRCA-positive patients, high-risk patients, young patients, and anyone who might choose to have partial breast irradiation. So this is an interesting study that looked at the incremental cancer detection rate in women with newly diagnosed breast cancer. And this came out of MD Anderson a few years ago, and they do a lot of whole breast ultrasound staging in their patients. And obviously, they see a lot of breast cancer patients. And so the incremental cancer detection rate that they reported was 16% in the ipsilateral breast and 4% in the contralateral breast. So hopefully, you guys have good memories or you're paying attention to Connie's talk, because this is almost exactly similar to what studies of MRI incremental cancer detection rates show. Now, this is a single study and hasn't been reproduced. And again, we know that ultrasound is operator-dependent. And I think that, you know, in very skilled hands and experienced hands, it's possible perhaps to do this kind of work and use ultrasound for staging, clinical staging. This study is more recent out of European Radiology that looked at tomosynthesis combined with automated whole breast ultrasound and compared it to MRI. So they had 73 patients with newly diagnosed breast cancer and 160 lesions in total. So a pretty small study. And the overall accuracy of MRI was 93% sensitivity and specificity, 96% and 89% higher than ultrasound. But ultrasound combined with tomosynthesis was not that far behind at 90%, 92%, and 87%. So not as good, but not too far off. Ultrasound missed two infiltrating ductal cancers and four DCIS. And so the authors concluded, and I think reasonably so, that ultrasound may be an alternative to MRI, particularly in patients with low DCIS burden or patients who can't tolerate MRI for reasons that we know. Okay, what about ultrasound and neoadjuvant therapy? The ACR and NCCN recommend MRI assessment to look at pathologic tumor response. MRI is good, but it's not perfect, and the guidelines are not universally accepted. There's the issue of false positives, so we don't do it all the time. And in our practice, mammography and ultrasound is commonly used. It's really important to use the same imaging modality before treatment as after treatment. And it didn't take our surgeons long to learn that, because they would do an MRI in advance, and then they would just do an ultrasound afterwards. And you know, you have to compare apples to apples. There's no way we could tell, you know, what really was happening in these patients. So it's important to do the same modality before and afterwards. This study was a review, and so it was a meta-analysis of multiple studies looking at the pathologic complete response rate and the accuracy of various modalities, MRI, mammo plus ultrasound, ultrasound and mammo, and then they compared it to clinical breast exam, again, MRI being superior in assessing pathologic complete response, but the combination of mammo and ultrasound was not that far behind. So 84 percent for MRI and 80 percent for mammo plus ultrasound, 79 percent accuracy for ultrasound, and 74 percent for mammo alone. So here's a patient who had partial response. She was a 49-year-old with a new palpable 6-centimeter triple negative high-grade infiltrating ductal cancer, and she has this very large cancer, and her mammogram looks great, right? After neoadjuvant chemotherapy, it looks really clear. Here's her pretreatment ultrasound. We thought this cortex looked a little bit prominent, and so this lymph node was biopsied, but fortunately for her, it was negative. So here's her post-treatment ultrasound. We see the biopsy marking clip. You see a hypoechoic lesion extending a little outwards with some little extensions here, and on final pathology, she had a residual 3.5-centimeter tumor. So this could be fibrosis or it could be residual tumor. It's hard to tell on imaging. Here's another patient with complete pathologic response, another triple negative. They tend to do better with neoadjuvant chemotherapy. Palpable mass, speculated on mammography, looks ugly on ultrasound, irregular margins, microlobulated margins as well, and afterwards, here is her clip. It looks pretty good on her mammogram, although there's some tissue in here, so it's hard to know what's going on. Here's the ultrasound, the hypoechoic mass, but luckily for her, she did have complete pathologic response and had breast conservation. So some tumors shrink with or without fibrosis, or most tumors you can usually see if they have complete response, a little hypoechoic area if you scan carefully. So what about ultrasound and axillary lymphadenopathy? Just to review, abnormal lymph nodes have focal cortical thickening, usually greater than three to four millimeters. They have a narrow or obliterated fatty hilum, round shape, and they're hypoechoic. The advantage of ultrasound is that it's real-time evaluation. You can biopsy it and leave a clip in place. In my opinion, ultrasound is superior to mammo and even MRI, just because you can do the biopsy and look at the cortex really well. So here is the needle that's leaving, the introducer needle that's leaving the clip in place, and here's another abnormal lymph node with a clip. So in our practice, we do corneal biopsy for the axillary lymph nodes that look abnormal. We use a 14-gauge, and we use the open trough technique so we don't fire into the breast. Some practices I know use FNA, but FNA does have a higher false negative rate, and so you may have to do repeat biopsies. You have to also have a skilled cytopathologist who will read the FNA. Before I do any axilla, I open up my color box really wide, and I make sure that there's no vessel sneaking there. Usually we're not as deep as the axillary vessels, but I always want to be sure it's not there. And even if there's a significant branch, all I really do is I just angle away from it and make sure that it's not in my field of view, because I know that if I can watch that tip at all times and that vessel is not anywhere angled away from it, I won't hit it. So I always do it so we don't fire in the breast. We do always leave marker clips in the axilla. There's a lot of different clips to leave in place. I like the wider ones because they seem to be easier to see. So the cue or the basket or the spiral ones are a little bit easier to identify on ultrasound. And it can be hard to do. This, believe it or not, is the clip. I didn't have a Cine to show you, but this was the original abnormal lymph node in the axilla. And here is the clip as well. I always tell the residents, it reminds me of the fetal spines that we looked at, you know, and like six-week embryos. But anyways, that's what it looks like. But we've been able to see them, and we scan really slowly. You have to make sure that your ultrasound technique is optimized, and we haven't had to go so far as to, I've heard some people have had to go and do CT for these clips. We haven't had to do that, but we just take our time and scan, and we've been able to find them. So in these patients with new breast cancer and no suspicious axillary lymph nodes on mammography or PE, on physical exam, the question is, is axillary ultrasound necessary in these patients? Okay, so do we do axillary ultrasound only for biorads 4C and 5 masses that we're pretty sure are cancer, or do we do it in all patients with newly diagnosed breast cancer? Again, this is controversial, and it really depends on your practice, what your surgeons want. So you have to have an open discussion with your surgeons and your breast cancer care team. Certainly, as Connie mentioned, a less is more approach is in style in breast cancer care in general. So let's take a pause and talk about less is more. I always thought it was attributed to the beautiful lady on the left, and that's Coco Chanel. You know, take off one accessory before leaving the house, and you'll look much better. It's a good rule of thumb. But actually, I did some research, and it was attributed, that phrase was actually attributed to a guy of her generation, Mies van der Hoff. And he was born in Germany, and he was the director of the Bauhaus movement and leader of 20th century modern architecture. And interestingly, he also coined the phrase, God is in the details, which I think really applies to breast imagers, because we're really detail oriented. And this is some of his work. I don't know if any of you guys recognize this building, but I think I passed it on my way this morning on the bus, because he lived in Chicago. And this is in downtown Chicago, and these are some of the houses that are in the area here. All right, so that's just a little trivia. So why is less is more important, particularly for the axilla? Well, this landmark study, the ACASOG Z11 trial, was published in 2011, and basically showed that there was a number of patients who did not need axillary lymph node dissection, even if they had positive sentinel lymph nodes. So basically what they did was a prospective study of nearly 900 patients with early breast cancer, T1 or T2 breast cancers, and one to two positives sentinel lymph nodes. And they randomized them to axillary node dissection and no axillary node dissection, and found that there was no difference in survival, first at six years, and then they followed them for a full 10 years. And so overall survival and recurrence was similar. So pre-Z11, what we used to do was if there was no palpable axillary lymph nodes, they could either go right to sentinel lymph node biopsy, and if it was negative, then they wouldn't have the axillary node dissection, and if it was positive, they would go ahead and have that procedure. We also were free to add ultrasound, and if it was negative, they'd go for the sentinel lymph node biopsy, and if it was positive, they would typically go right to axillary node dissection. Post-Z11 looks really simple, where you have no palpable axillary nodes, T1, T2, T tumor, and you do a sentinel lymph node biopsy, and if it was only through a positive, you would have an axillary lymph node dissection, and these patients right here with one or two positive nodes would not have to have axillary lymph node dissection. I think the flaw of the Z11 is that they only used physical exam to, as their definition of a clinically negative axilla, and they did not include ultrasound, and our surgeons were a little uncomfortable because they admitted that, you know, they're good at doing the breast exam and looking at the axilla, right, on PE, but they also know that they can miss significant adenopathy, particularly in heavier patients, and so they really weren't all that comfortable with just following this religiously, and there have been several studies since the Z11 came out showing that ultrasound can identify patients with higher disease burden, and this is just one, and for the sake of time, I'm just showing one, and this basically looked at patients with T1 and T2 tumors who had metastatic lymphadenopathy diagnosed via ultrasound corneal biopsy, and then they looked at the rate of having two or more malignant nodes on final pathology, which was the magic number based on the Z11, and if there was a single abnormal node on ultrasound, there was a 31 percent of having more than two malignant nodes on final pathology, and if there were multiple abnormal nodes on ultrasound, then there would be a 70 percent chance of having more than two malignant nodes, and they concluded that 3 to 5 percent of patients with abnormal axillary ultrasound in these T1 and T2 tumors met the Z11 criteria, so not that many patients with positive axillary ultrasound. So post-Z11, the algorithm is a little bit more complicated. There are certainly some practices and surgeons who do not want to do ultrasound and really follow the Z11 and just go straight to sentinel node biopsy and then decide whether or not axillary lymph node dissection is needed, so you need to talk to your surgeons. You can do ultrasound, and if you do, which we do in our practice pretty freely, if it's negative, they can just go and follow the Z11 way. If it's positive, and there's one or two positive nodes on biopsy, then these patients may go on to sentinel lymph node biopsy, and then if there's nothing additional found, they would not have axillary lymph node dissection, but if there's more than three or more nodes on ultrasound, and at least one of them is biopsied, and the patients have a higher clinical stage or more aggressive tumors, then some patients would go straight to axillary lymph node dissection. So basically, by doing ultrasound, you might avoid two procedures in patients. Okay, so what about axillary ultrasound and neoadjuvant systemic therapy? We know that neoadjuvant therapy can eradicate nodal disease in 40 to 70 percent of patients, but the question is, what do we do with patients with biopsy-proven positive axillary nodes who convert to normal on imaging and clinical exam post-therapy? Do these patients need axillary lymph node dissection? Because we don't know. Perhaps full dissection wouldn't provide an oncologic benefit to these patients. So the thought was maybe we could do sentinel lymph node biopsy in these patients, but the problem is that the sentinel lymph node biopsy can miss the original positive node in up to 20 or 23 percent of the time. Okay, so the positive node that we biopsy on ultrasound may not actually be the sentinel lymph node. Okay, so that's a take-home point. And this was work done by mostly colleagues in MD Anderson, and that they found that the false negative rate of sentinel lymph node biopsy was 5 to 20 percent, and then the ACASOG Z1071 trial showed that the false negative rate of sentinel lymph node biopsy in these patients was close to 13 percent. And in general, a false negative rate of less than 10 percent for sentinel lymph node biopsy has been the accepted standard. So we also live and work in an era where there's a lot of information, and I think that more information is good, especially if we want to have more precision and personalized medicine treatment. So we want as much information as possible. So we need to, you know, figure out what's going on with these patients, and we don't certainly want to leave any biopsy-proven masses in the breast after a neoadjuvant therapy. So what's the answer to this is to do a targeted axillary dissection, and that's basically localizing the clipped node. So that's why you have to leave a clip in the node, and then you might be able to localize it, or you should localize it after neoadjuvant therapy. So this is the algorithm. You have a pretreatment ultrasound with a positive biopsy or FNA. You clip the abnormal node. It responds, hopefully, to neoadjuvant therapy, and then they would have a targeted axillary dissection and sentinel node biopsy. If it's positive, then the patients would go on to axillary node dissection, and if negative, patients would not have axillary node dissection, and this is really a game changer because, again, more patients can be spared the potential morbidity of a full axillary lymph node dissection. So here are some examples. A 46-year-old with a grade 3 infiltrating ductal cancer and a positive axillary node. This is her original mammogram with a clip in the cancer and the positive node. Here's the positive node on ultrasound. There's her cancer. She responded really well with neoadjuvant therapy, and we can see the clips there with nothing around it, and so we had to localize this. Here's the clip, and there's our localization wire and the hook right there, so we were able to localize it. Our surgeon did a specimen radiograph just confirming that we got the clip out. There was a nice study in academic radiology last year that looked at axillary localization techniques. It is possible to put a wire in, but they also have these smart markers, radioactive seeds and magnetic seeds that you can also put into place. We haven't gone that route yet. We probably will do so, just for workflow and convenience for everybody and not having to do too many localizations early in the morning, but that's our main reason. It's not because we can't see the clip, and in some ways, it's very convenient for us because we just leave the clip at the time of the initial biopsy, and then we can find it. Okay, I think I have a little bit of time left, so I just want to talk briefly about MRI-directed ultrasound because, again, it is the hardest type of ultrasound, in my opinion, that we do, and we often do it in these patients with newly diagnosed breast cancer. So it's great when I am on the biopsy service, and it's a second-look MRI-directed ultrasound, and I read the original MRI. Like I'm always really happy because I've already invested, you know, time and energy into these complicated cases, but most often, that's not the case, so you have to spend time and really carefully review the MRI. This is not the kind of study that the tech should do, and if they say it's normal, you let them go. Like you have to go in and scan and spend time with these patients. It's time-consuming, and you have to use the right probe. You have to use the right, you know, optimal ultrasound settings, and you have to remember that the morphology of the lesion on ultrasound may be different than that on MRI, and there's a lot of differences in patient positioning because of the differences of the prone scanning for MRI and supine for ultrasound. There was a really great article, a very detailed article published in Radiographics in 2013 by a group in South Korea going through it all, so I encourage everybody to browse through this article, and remember that with ultrasound, all layers are flattened and widened, especially if there's a lot of fatty tissue, and that the lesion may also flatten out, okay? And the lesions that are going to change in position the most are the ones sort of in the middle of the breast. They can be anywhere. The ones underneath the skin probably aren't going to move. They're going to stay under the skin, and the ones next to the pectoralis muscle, they're going to stay in the same position. Remember that lesion, if it's surrounded by fat on MRI, it's most likely going to be hypoechoic or isoechoic and surrounded with fat by ultrasound. If the lesion is surrounded by glandular tissue on MRI, it's most likely going to be surrounded by glandular tissue on ultrasound, okay? And again, central lesions and fatty breasts are probably not going to correspond to the MRI depth at all. So this is a case that I'm really proud of that I did maybe a month ago, 49-year-old screening ultrasound with a grade two infiltrating ductal cancer, fairly aggressive, and she had dense breast tissue. She had a screening ultrasound detected the mass here. You can see it's hypoechoic. And then she, it was tricky, and so she had an MRI as well afterwards, and we found this other little mass on the posterior aspect of her breast. And she had a lot of fatty tissue anterior on the T1. And it was close to the pectoralis muscle, and that's a relief, because then you know where you're going to focus your scanning. And knowing this, that everything is going to compress, I'm looking for something that is sort of behind a patch of glandular tissue, right? So there's the subcutaneous fat, glandular tissue, and then the pectoralis muscle. So I'm going to focus here, and I don't know, which one is it? Is it this? Kind of hard to tell, but in real-time scanning, carefully. You guys see it? It's right here. And we have to scan, get away from it, go back to it, convince ourselves it's real. And it was real, and we did a core biopsy, and it was an infiltrating ductal cancer. And these are the kind of cases where the surgeon is, you know, really happy, and we show our value, and they're impressed, and the patients like it too. So in summary, whole breast ultrasound is feasible for problem-solving, local regional staging, and to evaluate tumors after neoadjuvant therapy. Axillary ultrasound is controversial. I know the preference of your breast surgeons. I do think more information at the time of diagnosis is good. Remember the clipped abnormal and biopsy proven positive axillary lymph node may not be the sentinel node. MRI-directed ultrasound, use those tips and tricks. So why do I have a stealer here, okay? It's not because Rita's up next, and she's from UPMC, and it's football season, or the fact that I grew up in Pittsburgh and also went to Pittford Medical School. But this guy, his name is Cordell Stewart, he is a true utility player. He played quarterback in his pro career and threw like 100 touchdowns, but also played wide receiver, and he made a bunch of catches and made touchdowns playing as a pro football player. And so there are utility players in sports, and ultrasound is the utility player in breast imaging and the newly diagnosed breast cancer patient. Thanks. So this is not a 25-year-old technology. We don't have a lot of studies to demonstrate consistency or even how to interpret it, honestly yet. But it's a really intriguing technology for a lot of reasons, and I'm going to run through those today with you. So this is the outline of the presentation that we're going to do. So let me just start with some basics, because my guess is a lot of people in the audience haven't used this technology yet. So let's just talk through it a little bit. It uses iodine-based contrast. We use the same dye that we use in CT. We inject the same dose as CT, so it's 1.5 mils per kilogram body weight, usually given at 3 mils a second. The time to image is generally two and a half to three minutes that most of us start imaging. This hasn't been solidified yet as the optimal time to do it, but that's when most people are beginning to do the images. So you basically bring the patient into the mammogram room, put a little IV in their arm, give them the dye, we wait two or three minutes, and then we start acquiring the mammogram pictures. The total exam time is under 10 minutes. It's very easy to correlate with standard images, and in fact, you get low-energy mammographic images that look exactly like a regular mammogram, because they are a regular mammogram. Biopsy methods are under development right now, so a lot of the cautions that Connie talked about, about early MR research, we have to hear those cautions and make sure we don't stumble through the same problems again with CEM. There is a little bit higher risk of contrast reaction than MRI, but it's still under 1%, and most reactions are mild treated with Benadryl. The pros of CEM, you use a mammogram unit. Now it has to be upgraded to have the appropriate detectors and software, but it might be something that doesn't need a forklift upgrade, depending on what you've got. The images are available immediately, just like doing a mammogram picture. This is, I think, very important. There's decreased sleepless nights for the patient, because you can read this test literally immediately, and they're relatively easy to interpret, so you can potentially give the patient a lot of the information if this pans out right on that same visit. It's easy to correlate with a mammogram, because we get the low KEE images. It eliminates the access problems that we have with MRI, especially in underserved areas. There's no issues with claustrophobia. The cost to the patient and organization will likely be much lower. There is no CPT code for this yet, but the cost of this test is going to be lower, and the training for the staff is very straightforward. The cons of it is that there is a cost to upgrading the equipment, or possibly a forklift upgrade. You have to have a power injector and be able to do estimated GFR in the breast imaging suite, which are two unusual things, and you need training for your staff around that. You have to be prepared for contrast reactions, so getting some training around that would be very important for your techs, your nurses, and your faculty, and we mentioned before the contrast reactions. So the two initial trials that came out, Roberta Jong, who's sitting in the audience in the back, did one of them. It's very important, and John Lewin did the other. These came out a long time ago, in 2003, and the interesting thing about these studies is that the results that they found in these small cohorts are still valid today after multiple trials. I think that these were important studies demonstrating, explaining to us to look for enhancement on CEM, basically ranking it as strong, moderate, or weak by John's study, and they showed mostly that cancers are medium to strongly enhancing above background, and that DCIS is more weakly enhancing. They also showed that people who don't have breast cancer typically don't have enhancement. What about patient preference? You know, it's really important to our patients. They're our customers, and we want to not only deliver our results accurately and quickly, but we also want the patient to have a pleasant experience, and so a lot of patients give feedback that an MRI, especially if you're not doing an abbreviated study, can be relatively tiring and hard on them, and so this study looked at that. It was a Likert questionnaire to patients, and CEM performed better in this cohort of 49 women for comfort of compression, comfort of IV injection, anxiety, and overall preference, and the reasons that they gave was that it was a faster test, a more comfortable test, and a less noisy exam. Now Lexicon is under development today. We are using primarily MRI terminology currently for CEM interpretation. This study looked at a cohort of patients. You can see here a pretty big study, 200 and some benign lesions and over 400 cancers in this trial. We're looking at kappa evaluation between radiologists for assignment of lesions, and there was a high kappa score to this, and then the cancers were primarily exactly what you would see on an MR, an irregular mass or a speculated mass or one with ill-defined margins with heterogeneous enhancement that was intense, and I'll show you what I mean by intense in a minute. The interesting thing about non-mass enhancement was that the only criteria for malignancy was the intensity of the enhancement, not distribution or anything along those lines. One thing that I think we have to interrogate further with CEM is the notion of rim enhancement. That's a very suspicious finding on MRI, but we're not quite so sure that that is valid for CEM, and in this study, it was not an independent criterion for malignancy, so watch out for the rim enhancing lesion on CEM. This retrospective study was looking at background parenchymal enhancement, so they took almost 300 women with elevated risk, and they rated the background parenchymal pattern for CEM and MRI. They rated it in the same categories that you're accustomed to for MRI. Just like MRI, most patients have minimal or mild background enhancement, and the reader agreement here was moderate, it's 0.66, and this is what it looks like. So we have minimal, mild, moderate, and marked. You can translate your MR skills to this directly. This is a patient who has heterogeneously dense breast tissue, and I would say she has mild to moderate background parenchymal enhancement and a strongly enhancing irregular mass in the superior right breast. Good luck reading that. This is way easier to read. Here are the cranial caudal images. You can see that this is a very difficult cancer to identify on the low energy images, and here it is strongly enhancing. This is an invasive ductal carcinoma with DCIS. Here's an example of moderate enhancement. This patient also has heterogeneously dense breast, but a pretty easily identified lobulated mass in the superior left breast, and it's moderately enhancing on the contrast enhancement mammogram. This has maybe mild background parenchymal enhancement, and here are the cranial caudal views demonstrating the same thing. This was a fibrotinoma. Here's weak enhancement. Weak enhancement's difficult. It's just above background or at or slightly above background. These are the ones that are harder to identify. This person has almost no background parenchymal enhancement. We can see that this small mass with some calcifications picked up on the mammogram is barely enhancing on the contrast enhanced mammogram. Here it is on the cranial caudal view. This was an invasive ductal carcinoma. So everything, all cancers are not strongly enhancing. Now let's look at some accuracy considerations. First, let's start off by comparing contrast mammography to mammography, which we're very comfortable with. This was a retrospective single cohort study that was very instructive. This study demonstrated that most invasive cancers had strong to medium enhancement, something that you would translate pretty easily over your MR knowledge, and that DCIS had less enhancement. It's weaker enhancement on CEM, it appears. This is an interesting point. In this trial, 100% or 33 lesions were benign if they had no enhancement, and 60% of enhancing lesions on MR, or sorry, on CEM were actually benign. So if you think about that opposite, everything that enhances is not cancer, but if something isn't enhancing, so far the early literature is suggesting that there's going to be a very strong likelihood that's not a breast cancer. And think about that. If I showed you some lesion that I detected on a mammogram or an ultrasound, and you wanted to give it a low BI-RADS-IV score, and then I showed you an MRI and there was no enhancement, would you dismiss that lesion? Those are the sorts of thoughts that we want you to start to have. So this study showed that the accuracy of CEM was higher than digital mammography. Here's another study looking at CEM, this time using mammography and ultrasound. There were 80 cancers and 62 benign lesions in this cohort, and a reader went through sequentially, reviewed the mammogram and the ultrasound, and then weighted and re-rated the same studies with CEM, and they showed improved sensitivity, specificity, and area under the curve. Again, they showed in this instance about 30% of benign lesions had some enhancement on CEM, so there are false positives with this technology. This is an interesting study. This was almost 1,000 patients, just short of that, recalled from screening for whatever, who had a CEM at the time of recall. Seventy of those patients, almost 10%, not quite, had a CEM-only detected lesion at a separate site that turned out to be cancer, or 54% of them were cancer. So there were 38 cancers detected at another site in this group of people recalled. Most of them were invasive. I thought this was interesting data. Now, CEM currently in the United States is approved as a diagnostic tool, and so one way to think about it, if you're considering implementing this technology, is to put it into play in places like this, the recalled patient, the patient with a lump. That's where it's approved. We can use it. We have the ability in the country to use it for screening off-label, but this is what it's approved for. Here's a patient who was recalled from screening with a focal asymmetry in the left upper outer quadrant. I think everybody would probably recall this. We didn't have any priors on this lady. Here's her CEM. This turned to she underwent a biopsy of that area, and it was pseudoangiomatous stromal hyperplasia. So if I showed you this first and then went back to the mammogram, you know, I wonder if we can get to the point where we can start to dismiss things like this. Here's a patient who was recalled from screening. She has a focal asymmetry in the posterior aspect of the left upper inner quadrant. Here is a circle or some arrows on it showing it's more speculated when we look up close at it, and there was a second very subtle lesion detected under the skin that here on the non-MAG view, as you can see. We went to ultrasound and found two suspicious lesions here at 11 o'clock, the area that was recalled. Here's the second area that was detected on the recall just below the skin in the 8 o'clock axis of the breast. Patient underwent two-site ultrasound-directed core biopsy. Here's her CEM. This suspicious mass is enhancing, strongly enhancing in the upper inner quadrant, but the other thing isn't enhancing at all. This lesion turned out to be an invasive ductal carcinoma, and the 8 o'clock lesion was fat necrosis. Here's this patient's MRI, axial, sagittal. Here's the index lesion at 11 o'clock containing a clip, and there's a little tiny bit of enhancement in the 8 o'clock axis, but a very, very correlative study, and you can see here that the amount of enhancement in this lesion is really subtle, and some of that could even be biopsy change. So we undertook a study trying to improve specificity along the lines of less is more. I thoroughly believe that, and I think one of the goals that we have as breast imagers is really to try and focus on accuracy, not sensitivity. Our job is to find as many important cancers as we can, but also not overutilize, not put our patients through the stress and anxiety of additional testing and additional biopsy if we don't have to, and so searching for modalities that allow us to really tailor that is important, I believe. So we undertook a retrospective review of the first 60 biopsy lesions in a consecutive accrual that's still ongoing. We had eight radiologists who had never seen CEM before actually read these cases, so we asked them to read the tomo first and rate the case, and then an ultrasound rate the case, and then we gave them the CEM, and they re-rated the case, and here are our ROC curves. So we showed that the sensitivity of CEM was significantly higher than mammogram and tomosynthesis, and the specificity of CEM was higher than ultrasound, and so the area under the curve was higher, was statistically significantly higher for CEM. So this begins to speak to the question that possibly we could be using this tool in the diagnostic setting prior to doing biopsy to be able to eliminate some of those low probability lesions that if I gave you an MRI, you would scratch your head and think twice about putting a needle into them, but here you have it right in the clinic at the time you're thinking about the procedure. So I think that's pretty intriguing. Here's a couple examples from that trial. This patient has heterogeneously dense breast. She gets recalled for this really subtle asymmetry on the left breast. Now that I've circled it, you might see it. Here's the ultrasound. I mean, how often do you get an ultrasound like this? My God, which one is it? So now you go back, and we finally decided it was this lesion, which we did an ultrasound directed biopsy of. Here's her CEM. Now this is not an easy CEM to look at. If I showed you a patient with marked enhancement, that's not an easy MRI to read, but there's literature out there for MRI demonstrating that the background parenchymal enhancement actually doesn't lower your sensitivity, and I think we'll probably find the same thing in CEM. There's nothing different here. These all look the same. Waldo is not to be found. That area that we were looking at in the mammogram and the ultrasound is not standing out. On this study, we thought it was somewhere in here, and it turned out to be just usual type ductal epithelial hyperplasia. Here's another example patient in this trial. She's recalled for a more suspicious-looking mass. This is a little irregular mass. Now, I don't think anybody would not biopsy a suspicious speculated mass if it didn't enhance on CEM. I wouldn't recommend that. This is going to be more of a 4C or a 5 lesion, but just to show you what it looked like, here it is at ultrasound. It's not quite as suspicious on the ultrasound. In fact, it's almost isoechoic to the surrounding fat. Here it is on the TOMO images, or sorry, here it is on the CEM images, and it's enhancing. So this is moderately enhancing irregular mass. This is a small invasive ductal carcinoma. How does this technology compare to MRI? It is our gold standard. I think Connie did a beautiful job outlining the stability and the wonderful sensitivity and accuracy we have with MRI, but it is expensive. We're using gadolinium. That's depositing now in people's brains. They're worried about it. We're having to hand out information to every patient who gets gadolinium because there's a black box warning from the FDA now. You're competing with other resources to get your patient in there. There are issues with MRI. Even with abbreviated MRI, I think we have access issues. So how does CEM stack up? All of these studies so far, looking at either whether they looked at sensitivity and specificity or area under the curve, are demonstrating equivalent performance between these two technologies. Now, this is the beginnings of the chart that Connie showed earlier. We are not in the position where we can begin to stop using MRI. I think we need a lot more research and a lot more effort put into this new technology, but I'm really beginning to be excited about this, thinking that this really could be something that we could put in our armamentarium. How about lesion classification versus MRI? This study had 263 patients. It was enriched with 177 cancers. Have a look at this comparability. Sensitivity, the same. Specificity, slightly higher for CEM. Positive predictive value, negative predictive value, false negative rate, area under the curve, all similar. This study showed, and we've seen in multiple other studies, that CEM appears to perform relatively comparable to MRI thus far and may have a slightly better specificity. Why is that? Probably because we don't have multiple phases. The spatial resolution is not as good. We don't have all the different sequences to look at. This is a snapshot in time. You're getting two views of a breast instead of thousands of slices to look through. But I personally think that that could be something we really want to think about because, again, we in our specialty need to work on accuracy, not sensitivity. Here's a patient from our study. She has a pretty obvious asymmetry in her breast. We can see that the skin's a little bit thick. There's a little bit of interstitial edema going on here, and she has a bigger regular mass. Here is her sagittal MRI. This is a delayed axis MRI demonstrating her large cancer. And here's her CEM image. So not maybe the detail, the spatial resolution of the MRI, but it gives you exactly the same information. What about cancer size accuracy? That's very important, especially as patients are wanting better cosmetic outcomes. We really want to make sure we're telling our surgeons how much breast tissue to remove. So there are two studies published to date. Mark Globes and Eva Fallenberg have both looked at this. You can see these are pretty comparable studies. Mark demonstrated that pretty much both studies were equal for size, comparing each other to pathology, and there was really no benefit of MR over CEM. And Eva demonstrated basically the same thing. These are millimeters of difference and all very close to pathology. So it appears to be able to demonstrate extent of disease for index very well. And here's an example from our trial. This is an invasive ductal carcinoma in this patient, that it is heterogeneous enhancing mass, probably moderate to strong enhancement, and a second lesion is identified inferior and medial to the index. And here's the MRI demonstrating the same thing. And we can see here's the index lesion, and here's the satellite. This lesion is not enhancing as strongly on the MR, nor is it enhancing as strongly on the CEM. Very comparable. And here's that patient I showed you before with the invasive ductal carcinoma and the moderate background parenchymal enhancement. This is her MRI MIP. She ended up getting a false positive biopsy of the contralateral breast for this clumped enhancement that was just benign breast tissue. And you never would have called that on this study. So these are the possibly the ways that we're going to see a little bit of incremental improvement. We undertook a study comparing MRI contrast mammography and molecular breast imaging in patients newly diagnosed with breast cancer for staging. We had 100 women that were recruited into this trial. It was a prospective trial. There was 110 index lesions in the cohort. We really didn't have any difference across any of these modes for detection of the index lesion. The CEM did a better job of size estimates. It only overestimated extent of disease in 11 percent, whereas MR had almost a quarter of them. And a lot of those are ER positive tumors. We know that MRI tends to overestimate disease in ER positive breast cancers. Now, many people don't do MRI for luminal A breast cancers, but luminal B, ER positive luminal Bs, we can overestimate extent. The positive predictive value for additional lesions in this study was 52 percent, whereas MRI was only 28 percent, and MBI about the same. And here's a case from our study. This lady has an irregular mass in her breast. Here's her mammogram. Here's the ultrasound. It's, if you're not careful, you could almost dismiss this lesion, but it's actually round, which is a very suspicious sonographic finding, and it is not circumscribed. Here is her contrast-enhanced mammogram demonstrating mild background parenchymal enhancement with moderate enhancing mass that looks pretty much the same as the mammogram. Here's the molecular breast imaging demonstrating the same thing, and her MRI. Here's another patient from that study that we did. She had a contralateral breast cancer, and this small enhancing mass was identified in the right breast. And here are her other two studies demonstrating no corresponding area of enhancement or uptake on the MBI. That was a false positive by MR. She underwent biopsy. This is a patient we just had to demonstrate extent of disease. She has a pretty easily identified cancer. I don't think anybody needs CEM to see this breast cancer, I hope. But if you notice carefully, there's a lot of other changes on this mammogram. The skin's a little thick. There's actually quite a large area of asymmetry in the breast that once you sit back a little bit away from the monitor, you're going to pick it up. But look at the CEM here. This was an invasive ductal carcinoma with extensive intraductal component. And here it is on the cranial caudal view. So you can see, once you sit back and look at this, the extent of this process. You could easily target in on this and completely get a force for the trees problem, but the CEM really helps to lay out that extent in this study. Now how about screening? Can we use it for screening to identify people so that we can triage our patients better? Some work has been done to date. The first was a high-risk screening trial looking at 307 women out of Memorial Sloan Kettering. There were three cancers in this cohort. CEM missed one of the DCIS cases. There was two-year follow-up and five additional screen-detected cancers were detected by two-year follow-up, but none of them were considered interval cancers. This study showed that the positive predictive value and specificity was comparable. This was sort of a proof-of-concept screening trial. Two other trials have come out. This is the first one looking at 611 patients for intermediate risk, which includes family history, personal history of breast cancer, or breast density as an independent risk marker. And their results were pretty significant. Significantly higher sensitivity for CEM over mammography, but lower specificity than mammography. Probably could be because we're learning, could be because we have those enhancing lesions, as we talked about in the earlier studies that we're picking up on CEM, but the negative predictive values were very good. Ultrasound in this cohort revealed 73 additional lesions beyond the index, all that were suspicious and they were all benign if they weren't picked up on the CEM test. The incremental increase in cancer detection with CEM was 13 per thousand, which is very good. We're really thinking along the lines of screening with MR, where the cancer detection rate is somewhere between 13 and 18 per thousand. Here's another study that just came out of Memorial Sloan Kettering, almost a thousand screened women with at least one year of follow-up. Again, family history, personal history of breast cancer, or just having dense breasts was what included these women in this study. Most of the patients were assigned a BI-RADS 1 or 2 with CEM, 92% of them. 3% were assigned a BI-RADS 3, and just 5% were given a biopsy recommendation. The cancer detection rate in this intermediate risk cohort was 15.5 per thousand, with a positive predictive value of biopsy of around 30. The low KEV images only found 50% of these breast cancers. We've just started a prospective trial at my institution. I'm doing it with Dr. Berg and one of our junior faculty, Ismael Wahid. It's a prospective accrual of 1,600 women with a personal history of breast cancer, and we're doing this trial because I'm embarrassed to tell you that when we look at the false negative, we review our false negatives every six months, and what we've been noticing consistently for the last several years is that a portion of those women, in our practice about 25%, have a history of breast cancer and have undergone breast conservation therapy. All of you sitting in the audience know how hard those cases are to read. There's distractors all over the place, and so we're really searching for a tool that will improve care in those women. So this is a prospective trial. Pretty much anybody who comes through the door who's had breast cancer, we're enrolling. We have two readers independently reading in opposite order, tomosynthesis-based screening and CEM. Primary outcomes are going to be cancer detection rate and false recall rate with all, you know, additional analysis of the biologic profiles of the cancers, et cetera. Here's an example so far from the trial. This patient underwent breast conservation therapy in 1994 in the right breast. She presents for screening, and we can see that it's just difficult to look at these cases. She also had a screening ultrasound, and this vague area was identified on the screen, ultrasound by the technologist. She was recalled and underwent this diagnostic ultrasound where this highly suspicious, very ill-defined mass was identified. Here's her CEM study. So the same area that was identified on the ultrasound is identified on the CEM. This is ILC, classic type ILC, and if I back up and show you the mammogram, this is that breast cancer. I wouldn't, we didn't call it. I don't know if you would, but we missed this on her regular mammogram. So what about response to chemotherapy? There are just a couple studies so far, first looking at 33 women and demonstrated that the mean size difference to pathology using CEM and MRI was equivalent. Bhavika Patel looked at 65 women, also who had neoadjuvant chemotherapy and demonstrated really no difference in the ability of these technologies to predict complete imaging response, nor in their ability for staging or for residual disease after chemotherapy was provided. So very preliminary data. This study is given to me by Bhavika. This is a 60-year-old patient with a HER2 positive breast cancer who had a complete PCR. So here's the index lesion prior to treatment, and here is her contrast enhanced mammogram. We can see the area that had been present before and the MRI. So in conclusion, this is a very new technology. I really think it's one of the most exciting things we've seen come down the pike for a while. It really may turn out to be a very efficient and accurate tool for diagnostic images so that we can detect more breast cancers, especially those that are biologically important. It seems so far to be comparable to MRI for higher risk screening and extended disease evaluation, but we need more data, and we just started an EconAkron trial to evaluate that. CEM so far appears comparable to MRI in extended disease evaluation as well, and it may be able to improve our positive predictive value. Thank you very much.

breast cancer

diagnosis

treatment

imaging technologies

mammography

MRI

contrast-enhanced mammography

breast-conserving surgery

multidisciplinary teams

surgical precision

clinical trials

diagnostic accuracy