Doris R. Brown, MD, PhD, assistant professor of Radiation Oncology at Wake Forest School of Medicine, discusses areas for advancement in breast radiation therapy including how to treat different patients and improving efficiency without decreasing efficacy. Dr. Brown also explores regional nodal irradiation in early-stage breast cancer.
Submit a question to Doris R. Brown, MD, PhD
View Doctor Profile DORIS R. BROWN: So I'm going to give you an update on radiation. And every year, it seems that right before this symposium, comes out the press about how we are undoubtedly killing people and hurting people and everything else. So I just want to put it back into perspective for you. And also, this year, we get to pick on the surgeons, because, finally, they got blamed for something besides me. And also, I'm really lucky because our plastic surgeons are so good. You've seen the outcomes that they have. Those are typical. Those are not, these are the best cases we're showing you. So I am someone who advocates for radiation when it's needed and usually reassure my patients that we do what's most important for you because of your cancer. I've got colleagues that will make you look beautiful down the road. It's going to take a little time. We'll hold your hand, we'll get you there, but we have great faith in our outcomes. So I don't have to say that I'm worried about it, because I put it in their hands. So today, I just want to update you on some advancements in breast cancer radiation and what are the areas that we look at. So every time, when we're coming into meetings and we're looking at where to go from here, the first goal is to eliminate or decrease the radiation treatment. So for this symposium, I just want to talk to you about where we are for DCIS, because that's a big hot topic right now, particularly in the news. And then also, what do we do about this regional nodal irradiation-- the one to three node-positive. It's always an ongoing debate. And unfortunately, I don't think it's going to end anytime soon, but I'll give you my perspective of the recent literature and where we're going from there. So nothing to disclose. So let's talk about DCIS. So in October of last year, the 20 year update of the Swedish DCIS study came out. And it told us similar to what the other updates did was that for about 1,000 patients, they have breast-conserving surgery with or without radiation. Our median follow-up was 17 years that compared to no radiation, that the radiation group had decreased risk of local recurrence and that those that do recur, in the absence of radiation, about half of them are going to be invasive-- the other half are going to be in situ-- and that if you look at the number at risk of cumulative incidence here, that we're decreased risk of having an ipsilateral cancer but we don't affect contralateral cancers, which is not surprising. So we also have, from 2010-- and it should be updated. We're hoping in 2016 will be the manuscript about DCIS. So this is the Oxford meta-analysis group, and they had a big input of data. That's why they're going to delay the analysis by a little bit, because we have the Swedish study that just updated. And then, there's a couple of others that are updating. So it looks like the Oxford analysis will be not on the five year schedule just because of how much data that they anticipate having to analyze. But based on the last analysis, if you look at those that had radiation versus those that don't, radiation decreases the risk by about half, which is significant. The benefit seems to be a little greater in women who are older, probably because they have real DCIS. The women who are younger probably have that environment where they're going to go on to get invasive cancers. So we may not be as helpful in those women, and they also tended to have larger tumors. So there's a lot of things with the older data as to how many of them really had invasive cancers that were hidden or microinvasion. And you may ask, well, why has that happened? Well, even in our experience on the most recent study that we're accruing now, our rate of microinvasion on Central path review is one out of seven so far. So even the best pathologists, when they're getting more sample and they're looking at more of the tumor, are finding more microinvasion. So a lot of the DCIS literature is going to be affected by how good the pathology was. But based on what we have, we know that women with small low-grade tumors-- they still had negative surgical margins-- still benefited in the risk of local recurrence. So their conclusions are that radiation has no effect on mortality, at least breast cancer-specific mortality, all-cause mortality, as well. So now, the question is, what should we do for DCIS? So there was a SEER analysis that was recently reported. And right after it, the press took off, because in the article, they made a one line statement about maybe we are over diagnosing breast cancer. DCIS is not really breast cancer. Therefore, the surgeons probably don't even need to do anything. We don't need to do mammograms on women-- the statement was so broad that it took off in the news. We don't need to do mammograms, we probably don't need to operate all these women. And then, they went further to say that the surgeons are actually harming these patients and that the mortality-- which I'm sure Dr. Levine and Dr. McNatt will tell you-- is not the number that the papers quoted-- that we should not do anything for women with DCIS any longer. It's interesting, because I'm going to tell you about this same group in another paper that they reported the same year. But they're now saying that, maybe, we have now harmed-- so again, that radiation, the women fared worse. And we have killed people by doing radiation. I'll tell you that's probably not the case. But it's the result of a statistician who took some data and made it look a certain way. So this group looked at breast cancer mortality after ductal carcinoma in situ. They used SEER analysis. So they had 100,000 women. So that's what really drove all of the press was that it was 100,000 women, so it must be true. What they reported, interestingly, was follow-up mean was about nine years. They're quoting data at 20 years, so certainly not realistic for what their data was supporting. And then, they also were reporting the mean. So instead of reporting medians, like most papers, they were reporting means. And they reported the standard error of the mean. So most of you are not statisticians, but if you want to make your numbers look really good, you look at the error of the error, and the bars go from this, to this, to this. So you can make numbers look really good and sound good. But here was a 20 year survival based on estrogen receptor status. So these are women who are ER-positive, and then, these are women who are ER-negative. And then, survival wise, they blew this curve up so that that looks like there is some kind of difference. But you'll notice that we're talking about 96% survival. And that's at 20 years. But notice that they started with 46,000 patients here, and their data for 10 years is only 109 patients. So that tells you the quality of the data and what they're reporting. And that just shows you how you can possibly be skewed by reporting-- most places report a median. But if your mean is off, you can very much skew the data based on that. And that's what the statisticians in this paper did. It's interesting, because the same group used a different database, and they reached the complete opposite recommendation in the same year. So in this case, they compared invasive cancers and outcomes from a database of about 3,000 women compared to DCIS. So these were recurrence-free survival. This is recurrence-- of in-breast recurrence after a diagnosis. So in all of these, people got radiotherapy. So whether you were invasive or you had DCIS, if you had an early invasive cancer stage 1A, and you compared it to DCIS, if you gave them radiation, look at that. You have 89% to 90% recurrence-free survival. If you give them no radiation and no tamoxifen, those numbers drop down into the low 90s and low 70s. So they concluded that the evidence here suggests that the natural history of DCIS is the same as invasive, at least early stage stage 1 breast cancer-- that the two look, basically, the same. And they advocated that the overall risk of in-breast recurrence after DCIS is higher, because a lot of DCIS patients, at least in their database, were no longer getting any treatment. So they weren't getting hormones, and they weren't getting radiation. But if you adjust for that-- so you look at, based on treatment, their outcomes are stage for stage the same for DCIS and stage 1. So we should use the word cancer, because the outcome for these women is going to be the same. And we should treat them the same. So where do you go from there? They used two different databases, and they have two extremes. One is treat them just like stage 1, which is what we've been doing for the last 40 years. Or treat them as if it's not a cancer at all and do nothing, perhaps not even operate. So the answer is, we've got a study for it. So I want to introduce you to a new concept. It's called the Oncotype DX DCIS Assay. So this assay is a modified panel from the Oncotype DX, so the invasive cancer panel. It looks at genes that are both markers of proliferation, mostly-- so your Ki-67, your Cyclin B1-- but also, ability to get out-- so Survivin, the MYBL2, as well as the progesterone receptor. So it was quote, "validated." I'll tell you that that is up for debate, which is why we currently have a study ongoing to actually validate it. So they looked at an ECOG study, that I'm going to go over, that was nonrandomized. It was a prospective study to look at excision without radiation, and then, they retrospectively looked at two parts of it to validate it. The study itself-- patient characteristics were mostly older women and 60%-- greater than 60 was 50% of the population. Tamoxifen use was very low. Margin status were wide margins. I mean, if you add this up, almost everyone had over three millimeters margin. That was part of the study requirement on central review. And that the majority of them were going to be intermediate risk. So what this means is that the Oncotype DX may not apply to your 30-year-old patient with DCIS that's high-grade. But we're trying to find that answer out. So based on what we have for this study, we know that their risk, over time, goes up. And as you would expect, without radiation, we usually quote women a risk of an ipsilateral breast cancer of about 25% to 30%, which is what the study showed us. And that that risk probably depends on the biology of the disease, which is why we have this study. So using this group, if you group it into a high, intermediate, or low, you can get a curve similar to what some of you may have seen for an Oncotype DX curve that predicts whether radiation will benefit them or not. So this tells you the risk of any breast cancer event, either the DCIS or invasive. So I'll tell you that the study is out there. It's utilization is in the single digits. Why? Because most people don't think it's validated. And the second reason is that people have already started adopting what to do for DCIS without knowing how to handle it. And now that we have two very controversial papers from the same group as to whether don't treat it at all or to treat, what we do is we are very supportive of the ECOG-ACRIN study. So we don't like to be hypocrites. So if we really think that there is two possibilities, and there's equipose as to what to do, if my therapy may cause harm, then let's make sure that we find out who needs a therapy and who doesn't. So this is a well-designed study. So it eliminates a lot of the problems in our prior DCIS studies. One, all women get an MRI. So you get an idea of the whole argument about whether DCIS is in a milieu effect-- so the whole breast is at risk, therefore, they need whole breast treatment-- or whether it's a local effect. So you're going to look at everybody with an MRI, so you'll hopefully get and weed out anyone who has mutlicentric DCIS. So the in-breast recurrence rate should be low, because we've screened for those women who only have one area by imaging characteristic. We also look to see the other breast. That's the other benefit of this. So you get an idea of how many women with DCIS in one breast are going to have DCIS in the other or an invasive cancer. Those women can go on to whatever surgery they choose. If they choose a mastectomy, they will still be followed for a recurrence. If they go for a wide local excision, even though the current guidelines from SGO and ASTRO and the others say that we can achieve negative margins on ink, this study is based on prior studies that we do need to have a two millimeter margin. So women go back until they have a two millimeter margin. And then, they have Central path review, which is good because of our cancers that we've submitted, the first seven cases we've had one that has had invasion on Central review. We've had two that have had microinvasion on our local review and were confirmed centrally. So our rate of upstaging, even with very well-selected patients, has actually been kind of high. And then, once they've gone through all of these hoops, they get the Oncotype DCIS test done. If their score is 39 or below, they don't get any radiotherapy, but they do get endocrine therapy. And then, their DCIS will be sent for more genetic studies. And then, those that have a score of 39 or above are getting radiation treatment. So we're now trying to use individualized medicine for our DCIS patients. So I really advocate participation on this study. Almost every woman who comes through the door, every woman gets offered this study. Not everyone will qualify, but we definitely try to get people on this study, because I think this is where the answers to what to do with DCIS is. Certainly, we know DCIS probably is not all the same. We've done the triple negative DCIS study. We've done a HER2-positive DCIS study. We're waiting for all of those results. But as of now, this is where we are. So then, the next big topic is, what do we do with the node-positive women? And I'm not going to talk about the pre-op or neoadjuvant chemotherapy issue. That's a whole other talk. I'm just going to talk about the one of three node-positive women. So regional nodal radiation-- so what do we know from B04? That's our workhorse of what we base decisions based on. So we know that women who had a radical mastectomy versus the total mastectomy radiation versus total mastectomy. If you look at women who have negative nodes versus those women who have positive nodes, women with positive nodes and had radiation did better than those women who did not. If you look at those between node-positive and node-negative group, we didn't see any difference. If you look at the clinically node-negative patients, despite an 18% axillary rate failure with just the total mastectomy, there really wasn't a difference in disease-free survival or overall survival. When you take all of the data, and you do the-- you look at the Oxford meta-analysis, and you look at, where do women recur-- so these were the women who were node-positive who had a mastectomy and an axillary dissection. If you look at local regional recurrences, first with radiation and without radiation, you're going to have a significant difference, and this is at 10 years. If you look at any first recurrence between radiation and no radiation, there's a difference here. So we have a 10 year absolute gain of 10.6% with radiotherapy. And if you look at breast cancer mortality at 20 years-- so if you look at those who had radiotherapy and those that did not, we have an absolute gain of 8.1%. If you're a statistician, and you do the numbers, and you make it sound really good, for every one local regional recurrence we prevent at 10 years, we will save one person's life at 20 years. That's a high bar to set and to stay at. So all of the talks you've heard before have talked about differences in median survival-- that you're talking about breast cancer-free survival, disease-free survival. We have an overall survival benefit. So now that we have said that, everyone now puts the bar for radiation is if you can't have an overall survival benefit, you don't need to do it. I'm going to argue differently. I'm going to argue that, in every other site of the body where we add radiation, if we can give you a local regional benefit, we add that modality to it. We don't necessarily have to have an overall survival benefit. But we've set a bar that, unless a paper can say that I give you a survival benefit, let's no longer do radiation for the node-positive women. So it's a really high bar to reach. And we're not always reaching it now, because we've now got systemic therapy that has finally made it to a point where our local failures are starting to come down, and distant failures are starting to come down. So this just shows, again, that if you look at the node one to three positive women, that reduction is still significant. If you look at the question about, what do we do in this setting now, where we have axillary nodal dissections versus just sentinel nodes. Well, we do have some of that data that Oxford has published. If you look at axillary sampling with and without radiation, in that case, you still see a benefit, even in the absence of axillary lymph node dissection. If you look at breast cancer mortality, you still see a difference. Although, this difference is not quite as strong. But if you look at those who did not get axillary nodal dissections, look at just those that have axillary sampling, the benefit is 12%. So now that we're in the error of Z11, the question is, should we continue to do the radiation that was published based on studies that were done in the 1970s and 1980s? So what did they do there? Regional nodal radiation. So should we still be doing that, or should we only do breast-only radiation? So indications for treatment in the post-mastectomy setting-- in the four or more positive nodes seem to be a slam dunk-- 51% versus 10% from the Danish study. What happens is the question of what to do with the one to three positive nodes? So if you look at local regional recurrence, we do have a benefit. Overall survival-- we had a benefit. So we should say, well, we always have a benefit. Well, the problem is that those were older studies. Chemotherapy is better, surgery is better. Undoubtedly, the Oxford meta-analysis is driven by European and Canadian data, where our surgeons will argue their surgery is not as good as ours. So we should not adopt what they have to do to make up for bad surgery. So in order to answer this question, we now have the MA.20 trial. So the MA.20 trial was a very simple question. It was one that could not be done in the US. That study closed, because people just already believed in either treating nodes or not. But the MA.20 study was breast-conserving surgery. They actually did have axillary dissection, because this was in the era before Z11. And then, they had systemic therapy, and they were stratified by their number of nodes positive-- how many were involved or how many were removed, what type of chemotherapy they had, hormonal therapy. And they were randomized to treat the breast-only or treat the breast plus regional nodes. Very simple question. Nothing to do with surgery at all. Just a radiation question. And what did we find? Well, the 10 year results have just come out. In overall survival, although there is a trend, there was not a significant benefit in overall survival. So the press immediately got this and said, you don't need to do radiation any longer. Well, what did I say? We had the bar set of overall survival. No one else has to do overall survival except for radiation for regional nodal radiation. But they missed the point that disease-free survival is better. Local regional disease-free survival is better. And distant disease-free survival is better. All those curves are separating and continue to separate. At some point in time-- remember, based on the Oxford meta-analysis-- when we prevent a recurrence at 10 years, we didn't see the survival benefit until 20 years. So if we continue to follow these patients, and we get the 20 year results, the whole way that we approach things may change. We may now say that we have definite evidence that you need to do radiation. So until then, I think we have to still rely on our older data and know that this data is maturing. It's showing the trend that we expected it to do and that if we wait 20 years, we probably will see the survival advantage. If you look at subgroup analysis, the women who seemed to benefit the most tend to be those women who were ER-negative, PR-negative, and there's some question about grade, as well. So it may also be that we need to start customizing what we do as far as regional nodal by some kind of test-- a molecular test. So the Oncotype DX test is being used on a study looking at one of three positive nodes. It was a test designed for chemotherapy, but one of the analysis that will happen is those women who got radiation or not, it wasn't dictated by this study. So some women will have gotten radiation and some will not. So we may get a little glimpse of whether a test will actually be able to predict for node-positive one to three women, whether or not they need radiation. So the other study that actually does show a survival benefit is the EORTC study. So this was published side by side with the MA.20. This was node-positive or a very high-risk node-negative patients. These were the T3 patients who had medial tumors. They either got radiation to the regional nodes or not. And then, you had to be N0 or N2, and you could actually be mastectomy or breast-conserving. The MA.20 was breast-conserving only. And what this shows is that they have power at 10 years for regional radiation versus no regional radiation. The follow-up isn't long enough to use the data at this point. So their statisticians were very honest. They basically said that those curves start coming together, we don't know whether that's real or not. So you cannot use that data. You have to use the point that we are powered to look at. And for those women, they actually had an improvement in distant disease-free survival, as well as an overall survival gain. It's very small, but it's there. And it is anticipated to continue to show a benefit as they collect more data at those time points. If you look at where do people fail on this study? So they looked at local regional, and it was a question about adding internal radiation or not. So most of their failures, you'll see, are in the medial supraclavicular, and then, the distant disease was higher. So why does a recurrence matter? So what does a recurrence cost? So a local recurrence estimates, depending on where you are in the world, about $19,000 to $37,000. So that's just cost to the patient. A distant recurrence, that's about $11,000 up to $34,500 per month, depending on how many new drugs you can be on at one time. So one of the new estimates-- so if you do 34,500 times 12, you'll get a very big number. So if I can prevent a local recurrence or a distant recurrence, even if I cannot prevent or show you an overall survival advantage, I think that that's worth something. If you ask my patient, or if you-- and nothing about radiation prevents the contralateral breast. So how much does a contralateral breast cancer cost? About $19,000. So a local recurrence is going to cost you more than a contralateral breast cancer. When you ask my patient, they're going to say it's priceless. And why is that? So why do we have to wait 10 years to find that benefit at 10 years translates to a 20 year overall survival advantage? Well, it's this-- so if you look at Kaplan-Meier curves, based on recurrence, only 50% of local recurrences are salvageable. And of those local recurrences, the time to death was about-- in 50%, took about five years. The local recurrence to a distant recurrence, again, takes time, but a distant recurrence to death is much faster. So in all of these studies, if our local recurrence rate goes down, then we're expecting a distant recurrence rate to go down. And that can translate to overall survival, but we need time. So did this-- how did any of this end the debate? So there was an article that actually was published by Poortmans a year prior to publishing his study, the EORTC study, that said that the debate is ended. Because the simple answer is that, in the 1950s and '60s, we were here. We have high-risk disease, no effective chemotherapy. So the benefit of local control didn't matter, because they were going to die of systemic disease. Once we started to get chemotherapy, and systemic disease started coming up, local failures start mattering. The only time when local failures don't matter is when you have such good chemotherapy that patients will live forever with a distant recurrence. We are not there yet. We are somewhere in this area here. We're probably reaching the point here where, with molecular characteristics of tumors, we can decide who needs additional therapy and who doesn't. We can start decreasing our therapies and wait for the day when we have highly effective systemic therapy that won't cause a second malignancy, because we always get blamed for second malignancies. But remember, systemic therapy's dirty little secret is that they have systemic cancers, as well. So we have to weigh those two. What I advocate is that we minimize both treatments. So you take the same approach we've done with Hodgkin's disease where you try to decrease both arms until you're at the least amount of toxicity and the least amount of mortality and morbidity to a patient so that they have less long-term effects. So radiation benefits all patients. The question is, is the magnitude worth the risk of toxicity? There's not a single paper out there that shows you that radiation won't benefit them in some way. It's either a local recurrence decrease, distant recurrence, or an overall survival. The elimination of radiation in DCIS may be possible. So we ask that you support the ACRIN study and help us to get that answer. And regional nodal radiation early stage breast cancer does decrease the risk of breast cancer recurrence, does decrease the risk of breast cancer mortality, but not overall survival, at least not yet. So in a few years, we'll probably be able to tell you that. And we will be back on the podium to say that we need to treat nodes. Until then, it's still going to be a debate depending on where you are. Thank you. [APPLAUSE]