Era of Hope 2011 Plenary Sessions

Session Title: Opening Session – The State of the Science

Title of Presentation: Perspectives on the Survival of Women with Breast Cancer 2011

Mark Pegram, MD, University of Miami: Good evening and welcome. My name is Mark Pegram; I’m the Chair Elect of the Integration Panel for fiscal year 2012 and I want to be the first to extend you a warm welcome to South Florida and to the Era of Hope. When you plan a meeting in South Florida in August, the one thing you are guaranteed is a warm welcome and I mean that literally.

I’m pleased to introduce our first speaker tonight, John Glaspy, who is the Director of the Clinical Research Unit at the Johnson Comprehensive Cancer Center at UCLA and also Director of the Women’s Cancer Program at UCLA. And he’s going to discuss for us incidents, risks, treatment effects, and changes in mortality in—in breast cancer; John.

John Glaspy, MD, MPH, UCLA School of Medicine: Thank you Mark. So I’m going to try to weave together what Dr. Lee was going to talk about in terms of the epidemiology of breast cancer, where we’re at, and then transition into some specific comments on what’s happening to survival in breast cancer and to what factors might we attribute those changes and what does this mean for the future and try and do all that in 40 minutes.

So the epidemiology of breast cancer, it’s—it’s well known that there are very large international variations in the reported rates of breast cancer and that’s an important qualification because obviously different countries have different thresholds for identifying and reporting cases of breast cancer and that might account for some of the variation but these are very large variations, five- or sixfold differences, and unfortunately the United States comes out at the head of the list in terms of the—the number of breast cancers per unit population of women, who are diagnosed every year with—with breast cancer in this country.

Does that translate into international variations into mortality? It does, but they’re somewhat blunted by the fact that the case fatality rates are very different in different countries. The—the mortality incidence rate ratios are .19 in North America, essentially the United States and Canada and .69—about sixfold higher in Africa. Now part of this may be due to only very serious cases of breast cancer being reported in some countries and therefore a more fatal type of breast cancer being reported more often. It—but it also probably has to do with differences in treatment and—and the support that’s available in different countries.

So 21% of all cancers diagnosed in women worldwide are breast cancers. There’s about a million cases a year on this planet of breast cancer being reported. There’s sixfold differences in high risk versus low reported incidence—countries. The rates in some of these low incidence countries are increasing; they tend to increase as countries industrialize and Japan, Singapore, and Korea have had rates that have gone up by two- to threefold in the past 40 years.

Finally, 45% of the incidence cases and 55% of breast cancer deaths are occurring in low and middle income countries. And this is important for us when people say things like eradicate breast cancer, this is going to be part of the challenge and the problem; a lot of the breast cancer mortality is taking place in parts of the planet that have limited resources.

This is the—the famous sort of comparisons that are always brought out to try to ask the question whether this is genetic or environmental, the international variations in breast cancer incidence. So when women in low-risk countries migrate to higher-risk countries, the question is what happens to their risk? And these are the changes in breast cancer rates with age in—in women who are in Japan and women who are in Japan and then within a generation of living in Hawaii, what happens to their risk and then women who are Japanese in—in extraction but live in San Francisco. This is always held out as evidence that part—a significant part of this international variation in breast cancer risk has to do with environmental factors that are—that—that happen in the countries of the high incidence.

There are indeed now let’s bring it down to the United States. There’s 194,000—195,000 give or take cases of breast cancer in the United States annually. Of these, all but 2,000 are happening in women; the other 2,000 are happening in men. It’s 26% of all the cancers diagnosed in women. It’s the most common cancer in women in the United States. It’s got a twofold greater incidence than lung cancer does in—in that group of people, and there are 40,000 deaths per year in the United States from breast cancer.

The established risk factors for breast cancer, this is a disappointing list primarily because these are not risk factors that are—the serious ones—are not subject to intervention. You don’t pick your age of menarche or menopause generally. You don’t choose your productive game plan around minimizing breast cancer risks. You don’t pick your family, your—your family history, and you don’t pick how dense your breasts are on mammography necessarily. It’s primarily obesity and alcohol that are risk factors that are subject to intervention. These are not going to be easy interventions to address. And the risk factors are relatively small. They’re not very high—1.2 to 1.5-fold differences in—in risk in non-alcohol users versus high alcohol users for instance.

A word about hormone replacement therapy. A lot of those risk factors are—have to do with the effects of estrogen in the body. The biggest risk factor of gender has to do with the hormones in the body. And so the question was—when we finally got good data about the role of hormone replacement therapy in supporting the health of post-menopausal women, the data suggested that estrogen in combination with progesterone, which was one of the combinations studied to address the potential for uterine cancer risk with unopposed estrogen alone, showed an increase in breast cancer risk in women who were treated with the combination, and this was not balanced by the promised health benefits in cognition and cardiovascular health. And this has made a major public health impact. This is something that—that has very quickly pulled us back from a very high proportion of women who were on estrogens. Menopause was being treated as though it was a treatable disease with hormone replacement therapy, and now many women have chosen not to take post-menopausal replacement therapy.

Now Fran on the—on the video said that we’ve learned that breast cancer isn't one disease. That’s going to be the theme of what I—of my comments today. That’s true, not just for the treatment and the biology, which we’re going to get into a little bit in a minute; it’s true for the epidemiology in the public health aspects as well. There are a couple of different ways to break breast cancer up. One of the ways that is the most time-honored and that we have a lot of data for because it’s been around a long time is simply histology; how does it look under the microscope? And although as I’ll show you in a few minutes, breast cancer rates are rising over time in the United States, it is primarily in the lobular sub-set of breast cancer where this is happening. These are cancers that tend to be hormone receptor positive and hormonally driven. Their rates have risen between the mid-1980s and the beginning of the 21st century by 65% compared to 3% for ductal cancers.

Now let’s go back and look at some of the risk factors in epidemiology with just that simple cut of ductal versus lobular. So if we look at ductal versus lobular and remember lobular is the cancer that’s increasing, although ductal cancer is still more common, lobular cancer is the one that’s going up—the—we see sort of a difference in the effects of hormone therapy on risk of lobular and ductal cancer. So here if there’s a number that’s less than one, the risk is lower than—with the let’s say the estrogen—E is estrogen—than it is without the estrogen. For ductal cancers, there really isn't a difference in risk with estrogen alone, but for lobular cancer there is a statistically significant increase for lobular cancer, and this is—the same is true for estrogen plus progesterone and that actually is where a lot of this change in lobular cancer risk may be coming from. It may have to do with a hormone replacement therapy, and if it did then we should start to see that go away now that there’s been a change in public health attitudes towards menopause.

Alcohol and breast cancer—it’s the dietary factor, the environmental factor, that’s most often linked to breast cancer risk. It’s a fairly modest relationship; one drink conferring in a 10% increase in risk, two drinks a day a 21% increase in risk. And it’s—there’s a hormonal link here. Circulating estrone levels are higher in women who drink alcohol than women who don’t, and this may all flow back into that whole estrogen story.

Now let’s go back and do that by histology. So if we look at where the alcohol-induced cancer risk is—is going it’s—it looks like it’s going to lobular cancers, not the ductal cancer, so this cancer that’s been increasing by 65% and appears to be linked to hormone replacement is also linked to alcohol intake and the ductal cancers not so much.

Now the challenge of breast cancer prevention; many of the risk factors for breast cancer have been identified. None of them are very strong risk factors and when you address any individual woman you have several scores on several of these factors. Some will be high and some will be low, and you can't just add them up and come up with a risk. There are models that allow us to calculate what the risk for the average woman in the group will be but that doesn’t tell us what an individual woman in that group risk will be. And many of these factors especially the ones that we really want to intervene with that have great relative risks of fourfold or more like BRCA status, like age, like gender, we don’t have any control over. It’s the—the ones we have control over, at least some of the ones that we have control over are relatively low risks and so the—the idea of an intervention is—has been lucid and we’ll talk a little bit more about that in a moment.

Back now to the theory that there’s—not a theory but the theme that there’s more than one disease nested in breast cancer, there are many different types. You can type them out histologically and we’ve been doing that. You can type them out molecularly. I think Dr. Ashworth is going to get into this with you a little bit and you will know this but you can do very sophisticated genomic arrays of breast cancers and when you do you find they sort into, depending on who is doing the sorting, five different sub-types, six sub-types. It depends if you’re a splitter or a lumper. When you do that though, you end up with basal cell breast cancer which is usually triple negative—estrogen, progesterone, and HER2/NEU negative, HER2 positive breast cancer is a discrete biologic sub-type, and then a couple of sub-types that have estrogen receptors and appear to at least in some instances be hormonally driven and those get called luminal. And we developed this mindset recently in oncology that the luminal breast cancers are easily tamed and—and we sort of have an answer for those and we’ve been called to pay most of our attention to the triple negatives. We have something we can do for the HER2 positives. As I’ll show you, the luminal breast cancers are still a big problem for us in terms of mortality in this country.

But you can type them out that way and you can—you can carry this forward into breast cancer risk assessment by instead of chopping up the tumors, you test the genes of each individual woman and do SNP analyses and develop a risk model based on genetics for a woman’s risk of breast cancer. And we’re going to talk about how powerful that’s been in a couple of minutes.

Now let’s instead of doing the histologic type, let’s do a split of triple negative breast cancer versus ER positive. This is roughly luminal versus basal but we don’t have array data on a large numbers of women, and so you have to do this sort of indirectly. So when we look at estrogen receptor positive breast cancer, these parody issues of women not having children versus having children and the number of children she has and how young she is when she has her first child that are all protected from breast cancer, they play out well in triple negative breast cancer—I mean in ER positive breast cancer and there—there is a link there, the weak one we’ve talked about. But if we now look at triple negatives, there does not appear to be that same—in that same link; in fact, it’s sort of reversed. And the more children, the higher the risk, which is a giant clue but—from God for—for something that—that we need to pay attention to in terms of trying to figure out this problem, but I don’t think anyone has figured out what the—what the clue is telling us yet.

Oral contraceptive use and the risk of triple negative breast cancer, there is a link between oral contraceptive use and the risk of triple negative breast cancer. There is not a link in women with estrogen receptor positive breast cancer, a little counterintuitive. This is a hormonal product that’s being given to patients that modulates hormones and yet it appears to impact risk of triple negative but not ER positive breast cancer. And when we lump them all together and treat them all in a one-size fits all approach, we miss the subtleties of—of this. It’s going to be important going forward to do more and more of the epidemiology on—on a sub-set basis.

So finally for figuring out someone’s risk of breast cancer, assuming we had something we could do about the—the risk when it was modest but increased in an individual woman, we have the Gail Model, which is based on epidemiologic factors that we’ve talked about and sorts women into groups that—where we know they’re average risk, and we have these SNP assays and the two together don’t perform really any better than either one alone at sorting women into risk groups. So we’re not where we need to be yet on being able to harness the power of molecular biology to identifying women at risk for breast cancer, letting alone the need to have some better strategies for what we do when we identify them.

DCIS—DCIS is pre-invasive breast cancer. This is a cancer that’s 100% curable, and we in the cancer treatment community are enthusiastic about curing it—sometimes perhaps too enthusiastic about curing it, and there are a lot of issues about whether some of these women are getting overtreated in terms of the amount of treatment they’re getting. It does identify a group of women who are at risk for breast cancer, and this is the sub-set of women where people talk about using drugs like Tamoxifen and aromatase inhibitors in—or other serums in—in approaching their cancers.

So to finish the epidemiology part of this and then we’re going to move onto mortality and what’s happening in breast cancer over time. If breast cancer evolves as it very likely does through a progressive dysplasia in a breast duct that ends in breast cancer that hasn’t yet penetrated the wall of the duct and as invasive, there are a lot of places for us to be thinking about intervening either to prevent the disease or to treat it. And I’m going to go through what evidence we have in a broad-brush overview of how well we’re doing and—at those and maybe where we need to go. This doesn’t show up very well but the panel on the—on the right is breast cancer; the panel on the left is colorectal cancer, and this is the incidence in the other parts of the world. We’re going to do the United States in a minute. And it doesn’t really matter which line is Australia and which one is England and which one is Norway, et cetera; what matters is that all those lines are going up.

In all the countries outside the United States where we have good reliable data, breast cancer rates over time expressed per unit population are rising and have been rising for quite a while. On the bottom is mortality; this is the number of women who die of breast cancer and that in all of these countries is going down slowly, but going down over time. And that’s all good news. And the question is why? Is this due to treatment? Is it because we’re getting better at diagnosing breast cancers early? Are we diagnosing more easily to cure breast cancers, et cetera? Those are all questions that I’m going to try to address from the data available.

What about the United States? So if we do this in a one-size fits all, which is how epidemiology is usually done, the—the graph on the left is the changes in reported incidents of breast cancer—reported death from breast cancer in the United States over time, and it’s going down, the same as it was in the other parts of the developed world that I showed. This is good news. It appears to be decreasing over time, but slowly. If we look at the other panel—now these are split into estrogen receptor positive and negative cases; this doesn’t control for the HER2 driver, sub-type of breast cancer but that’s evenly split approximately between the two groups and so what this really is—is a very crude back of the envelope comparison of luminal breast cancer to triple negative breast cancer, and the decrease in mortality over time in the United States has been in the luminal breast cancer. So it—it has not really been in the triple negative breast cancer, and the difference between the two is statistically significant.

Now this is a little bit complicated. The time at the bottom is not now years; it’s the time since an individual woman was diagnosed with breast cancer then her risk of dying is the Y axis. And what you see here is the—there’s a big early bump in three of these curves, and those are all the triple negative ER negative breast cancers. ER negative breast cancers if they’re going to kill—kill relatively early and there hasn’t been a big change in cohorts of ’90 to ’93, ’94 to ’97, ’98 to 2002. They are all within error of each other. The three curves at the bottom that aren't dotted are the—are the curves for death from ER positive breast cancer and there’s an early peak and then a later peak and there does appear to be a difference in the cohorts of years, similar to what you saw on the previous slide that isn't—that isn't expressed quite—in quite the same way that the difference has been in the ER positive patients.

Now this is an even more complex slide. So the question is now not how long since the woman was diagnosed, but what year was she diagnosed and how did that affect her risk of dying of breast cancer. If you look at the—the graph labeled All, which is the one size fits all, all the breast cancers are in the pool here—here we go, you see that over time the ER positive breast cancers are going—are going down in terms of the—the case of fatality and the ER negatives are not going down. And if you look within the ER positives, which is the second graph that—the one on the bottom left, in the ER positives that fall is primarily in young women that is women who are not 70 years of age or older. There hasn’t been that same progress in older women which is not what we usually think when we think about breast cancer in women 70 or older that’s ER positive. We usually think that we now have really good treatments for them, and those treatments work very well. It hasn’t—it’s not something that we can show in the—in the data. And for the ER negatives, it really doesn’t matter what year you were diagnosed. There’s not been evident progress in mortality rates.

Before we go any farther I want to clarify here; when we see a difference in survival or people talk about a difference in survival, there are three possible things we could be talking about. The first would be no change in cure rates. You’re talking about eradicating breast cancer and to me that means cure. But you shift the survival curve enough to the right that it’s statistically demonstratable that it didn't arise by randomness, that—that some intervention was—was the reason that this happened. That would mean for prevention, delaying the onset of the disease so if we were looking at the Tamoxifen breast cancer prevention trials, one of the big controversies was—is all it does is delay the diagnosis in women and not really change whether they get breast cancer or not if you’re talking about prevention, and if you’re talking about treatment of established breast cancer you’re shifting the survival curve to the right. We don’t have anywhere near the data we need to know how often curve shifts in survival, that are statistically significant, are clinically significant enough that we can be confident that fewer women ultimately are dying of breast cancer. That’s the gold standard, and that’s the second thing you might be talking about; you’re talking about actually curing the disease or actually preventing it. So a woman who would have gotten breast cancer never ever gets breast cancer. And a woman who has breast cancer is treated and never ever dies from breast cancer. And that—that’s the gold standard. It’s very hard with the data available. I’ll tease this point out a little bit to see how often that’s happening if at all.

And the last is artifacts that happen in epidemiologic studies. So if you are a screening country and you find breast cancer a couple of years before it would be found in a non-screening country you will see a longer survival. Or if you’re in the same country and screening is increasing over time, you will see people living longer, women living longer, because they were diagnosed earlier. You’re cheating and starting the clock early.

It’s also well established in the prostate cancer venue, and most of us think this probably happens in breast cancer as well that there is such a thing as trivial cancer that is being diagnosed more and more often but if left alone wouldn’t cause trouble. And if you diagnose a lot of those, your case fatality rate will go down because you’re diagnosing cancers that aren't—that aren't potentially fatal. So those are the artifacts.

Now, when I started training in cancer and people were talking about how we were going to set up systems for addressing solid tumors like breast cancer, everybody pointed to pediatric acute leukemia and said do what they did. And the curve on the top is the—is the 5-year survival. In that disease that’s a surrogate marker for cure over time and at the time, which was the late 70s, early 80s, they were having incredible success. Everything they tried to do made dramatic improvements in survival of those children. And these incremental gains, it was believed, would lead to a cure and so we built a whole system around this incrementalist slow—do one thing at a time and see what happens, and then do another thing and—and very quickly you’ll get to 100%. Now we’re going to look how well that’s worked out for us in cancer, and then look back at what’s happening in ALL because it’s instructive.

Before we—before we look at the survival data though I think it’s important—on the bottom, advances in cancer research prevention treatment are always what we assume is causing any improvement in survival that we see in breast cancer. But there are other potentials; I’ve already told you that we appear to be shifting to a different histology of breast cancer that’s ER positive and making most of our progress in ER positive breast cancer. And that is going to affect how successful we look. If we simultaneously cause something that we have a good treatment for, treatments will look better and better even if they’re not getting better.

The lead time—rapidity of diagnosis, how old people are, and how sick they are—there are racial and socioeconomic factors we’ll comment on, access to quality care; there’s some data and we’ll comment on it, diet and lifestyle, genetic background including BRCA patients use of hormone replacement therapy and what’s rarely talked about is acute and long-term toxicities of therapy. I don’t mean this in a cynical way but we have to be very careful that if we define success as somebody lives out their life and dies of something else that—that something else they die of isn't a result of our treatment and then we give credit that we’ve cured breast cancer, right. So we’ll talk a little bit about that.

Now where—if we’re having an impact and I think we are having an impact, where is that coming from? Most of the data points to two areas that probably have improved breast cancer survival. One is screening—mammography—and the other is adjuvant therapy for breast cancer, treatment that a woman receives to prevent recurrence. So top left panel is the proportion of women who are being screened on a regular basis; this is focused on women from 40—40 years of age and older. I don’t want to get into the 40- to 50-year old business right now but that’s just the way they cut the data, but it’s—screening is increasing. The use of adjuvant chemotherapy and adjuvant estrogen-directed therapy for ER positive cancers is increasing and that’s the—the graph on the right. And then from this data, these authors constructed several models and asked the models to show if this is the rate at which people are being screened or getting therapy, and this is the reported impact on survival. What would our survival curves be looking like and those are shown in the bottom right.

And one of those curves is actually a real curve; it’s the black one. And so the models kind of predicted what was happening in real life. And then they asked the models what would happen if neither one happened? And you can see breast cancer mortality is rising if we don’t do either one and it will come down modestly if we did both.

So this is part of the picture of what’s going on.

Now others—this is the Norwegian data—others have looked at this and said let’s be a little more thoughtful about this. There are a lot of things going on over time and not just increasing mammograms and in Norway they have groups—large groups of women who are never screened and they looked at them and their mortality from breast cancer—it is going down over time as well. And then they looked at their screened populations and their mortality is going down more. So in a sort of crude subtraction way, it’s a pretty good start. They came to the conclusion that there’s been about a 30% reduction in breast cancer mortality in Norway and about a third of that is due to—is due to breast cancer screening, and the rest is just due to they were lucky enough to be around at a later point in time when something else was different.

Socioeconomics and race, you—you—this is complex but every time it’s looked at, black women have a lower survival with breast cancer, a higher chance of dying when they get it than white women do. This has been attributed to differences in access to modern top-level care and early diagnosis and appropriate therapy. But the—nobody had quite addressed it to sort out what—whether this has something to do with the difference in the biology that breast cancer in black women manifests or a difference in health care. We know that black women for instance are more likely to get triple negative breast cancer, which we’ve already shown isn't—isn't doing well from a mortality standpoint at this point.

These authors went and—and asked the question where were these women cared for, what hospital? And then they divided the hospitals by the percent of women in the hospital who were black. So they compared the women that were—less than 10% of the women in the hospital were black and all the way up to more than 50% of the women in the hospital were black, and this explained about 30%—36% the variation in mortality. So both factors appear to be evident, there’s been maybe a difference in biology and most clearly a potential that there’s a difference in access to services and to treatment. It also is on the positive side evidence that treatment can affect the mortality of breast cancer, which is good—that we can do that to some extent. But it’s very sobering that—that we—it only explains 36% of the variation.

Now we’ll get more applause. So there’s usually not a free lunch and the adverse consequences of intervention may reduce survival or compromise quality of life and there are two I think I’ll point to that both have to do with the heart. The graph on the left is the—is the now famous follow-up of women who received anthracyclines as part of their adjuvant therapy or did not and looking at how their hearts age over time. We all—if we live long enough presumably will get heart failure but it happens in a more rapid way in women who have been historically exposed to anthracyclines. And this is one of the things that’s at the center of this raging anthracycline debate; there’s clearly a price to be paid and the fight is whether there’s a benefit and if the benefit is worth it.

The other curve has been very interesting to watch evolve. This is the left side versus right side breast cancer heart attack studies. So when a woman receives radiation for a left-sided breast cancer, she gets some radiation to her heart and over time theoretically that could increase the rate of which she develops coronary artery disease. And there had been a paper in this same journal that published this one 5 years earlier that said we looked; it’s good—we’ve looked out for 10 years and there’s no difference. And now they come back and say okay, when—when we got to 15 years, there is a difference and the difference is continuing to get wider.

These long-term effects that can affect mortality and won't be seen in breast cancer death rates is very important, and one of the themes I’d like to make is that long-term careful follow-up and knowing what is causing survival changes and if there are survival changes and if we’re talking about cure or simply a shifted curve is going to be increasingly important.

So if only approximately half—we’ll give the Norwegians a little bit of credit—of the observed 1% per year reduction in breast cancer mortality is due to advances in prevention and diagnosis and/or treatment, why not just stay the course? Why don’t we just do what we’re doing and eliminate breast cancer mortality in our lifetimes? Well first of all it’s not clear that half of it is due to treatment. It might be more modestly one-third; secondly we don’t know that any of this decreased mortality is due to reductions in case fatality rates or is it simply a delay? Are we kicking the can down the road? And you—we all know people—I’m going to end by talking about somebody that they go a long time and die of their breast cancer. That’s not cured; that’s not eradication.

And multiple factors are going to conspire to make this strategy not work. The magnitude of the advances will decrease over time and I’ll prove that to you in a minute with our pediatric colleagues. The smaller and smaller gains that we’re going to make over time are going to require larger and larger sample sizes to prove. And they’re going to have less and less clinical impact and cost more and more money. There’s a problem that I think Dr. Ashworth is going to talk about that there’s a burden of marginally effective therapies that are statistically significant in their impact but the question of their clinical significance is not fully answered. And the ethical issues that women have to receive all of these things before they can be enrolled in an innovative clinical trial—that is going to—is going to slow progress as well. So progress will feedback to slow progress, and these accumulating burdens of toxicities we’re only beginning to understand are—are going to threaten to increase late mortality which is going to affect what we try to do.

So here’s the slide I led with. This is the optimistic 1980s just follow the ALL people. Here’s—here’s what’s happening to them over time. So since the 1980s they’ve been doing about what we’ve been doing. They’ve been doing larger and larger clinical trials with smaller and smaller effect sizes and they’ve turned now their attention to breaking the disease up into biologically discrete units and calling those individual diseases and looking for big hits just like we’re doing in breast cancer. It’s statistics. I—I hate statistics but—but you have to understand them a little bit. And this is why it doesn’t—it’s not feasible to look for tiny advances over time. If the numbers at the bottom are the effect size, so if something is—reduces mortality by 20% that would be a .8. If that’s the effect size you want to detect this is the number of patients you need to enroll and it approaches an asymptote pretty quickly when you get above a 20% change.

And this is assuming that you’re going to wait until everybody has passed away before you do your analysis. If you want to analyze it earlier, you have to increase the sample size dramatically more. It’s assuming no one will drop out. If they do and they always do, you have to add a lot more. The sample sizes become huge as the effect sizes you’re looking for go down. So this is the dose-dense, the Q2 week versus Q3 week, overall survival impact of early breast cancer treatment with the adjuvant chemotherapy. This was clearly statistically significant but it took us 2,000 patients to find this difference of every 2 weeks versus every 3 weeks. It was an immense investment of time and resources; a lot of women they could have put on other trials and were put on this one to—to get the answer and you can look at those two curves and—and we have to get passed the—the tyranny of P-values. A P-value of .05 doesn’t mean it’s important to humanity; it means it probably didn't arise by randomness alone. That’s all it means. We’re supposed to then look at the data and then decide if it’s important.

And then because I knew there were going to be Naval people here, there’s—there’s another sort of approach. This is my favorite clinical trial. I think it was the first controlled clinical trial. James Lind recognized it. About as many British sailors died of scurvy during the—one of their wars with France that preceded his—his work—I don’t remember which one—as die of breast cancer every year. And it was a huge problem, so he got 12 sailors and, you can't read this, but he just gave two of them vinegar. He made two of them drink sea water, he made two others do something called elixir of vitriol, but he doesn’t tell you what that is, and I can't remember what the other two got. And then two of them got oranges and lemons and the clinical trial was over pretty quick. He had an answer because he had found something very fundamental about the cause of the disease and when you really have a huge effect size, you don’t need very many. I don’t think the most positive clinical trial in history having six cohorts and twelve patients is something we’re likely to repeat in breast cancer, but it’s worth aspiring to.

So here’s a not-all-one-size-fits-all approach. Here’s the—the thin hair data for HER2 positive breast cancer, for adjuvant Herceptin. They were able to demonstrate their—an impact with 200 patients, so you could have done 10 of these with the women that were enrolled in that one every 2-week versus every 3 clinical trial. It’s a modest investment of time and resources and I’m going to—I argue that—that is a better place for us to put our shackles. So Bevacizumab also known as Avastin in metastatic breast cancer, another source of great controversy, the two curves on the top and the one curve on the bottom left are the effects on progression-free survival. These are the—there—it’s pretty clear from those curves that it’s not really curing people, okay. There’s no question that it’s simply shifting the survival curve, a progressive-free survival curve rather—that all of them ultimately progress. But if you look at the survival curves for either trial—I only put the one for the second trial on here and that’s the curve at the bottom right—you can't show a difference.

And this has some relevance to how we move forward. This is a very expensive drug. It has an effect on progression-free survival. We can't demonstrate the effect on overall survival. This has become a huge political issue. There are doctors and patients who are convinced it has helped their patients or them, and it’s a big mess throughout—in a tough time economically.

So impacts have changed the paradigm. If we do small trials aiming at much larger effect sizes, if we try and de-scurvy instead of—you know dose down—what does that—the disadvantages? Well first, there are business models that you need to change. The cooperative groups have a big investment in this superstructure aimed at very large trials. Pharma has all their business models directed at—at big drugs and OIC—is my word—the Oncology Industrial Complex; there’s a huge—there’s a huge industry and it’s you know—I don’t mean it entirely pejorative in—in the status quo in cancer treatment. And all of that will need new business models if they’re going to be partners in this. We need adjustments in the regulatory environment for—for this to work and we’re going to have to agree that we’re going to miss some of these small, incrementalist, fine-tuning in our therapy as we try and move forward and look for a Vitamin C.

And finally we don’t have the tissue banks we need for this. We don’t—we’ve known for decades that anti-estrogen therapy was effective in treating hormone receptor bearing breast cancer and we don’t have tissue that we’ve stored from before and after—from that therapy in large amounts around the country that we can go query and ask what—what’s the molecular change and how do we prevent that. The advantages are it’s faster; it’s got potential for big revolutionary advances, and if we do it right we won't run out of money before we—we cure.

You know this week if it’s taught us anything it’s that finally people are going to be talking about—about the deficit and nobody wants the ceiling to cave in. And so there’s going to be a tremendous pressure put on every nickel that—of the public dollar that’s spent on anything. And we—we need to—you know we need to be smart in planning so that breast cancer research continues to do what it’s done and more.

So I’ll conclude by saying breast cancer has been diagnosed more frequently; the case—the mortality rate is falling modestly. It’s in part due to advances in treatment. The old model for addressing breast cancer, the one size fits all and incremental doesn’t work epidemiologically or therapeutically anymore. And it’s unlikely we’re going to make any advances in prevention or treatment if we stick to that old model. And we’re going to need—we should be thinking about a lot of small exploratory trials powered to rule out huge effect sizes. And everybody has their person, so I want to close with this.

This is Laura Ziskin who is another woman we lost just a few weeks ago to metastatic breast cancer. She did her best to fight back against cancer in general with some of her fund-raising and—and endeavors just as—as you all have. And she captured the hearts of every cancer researcher I know and had access to the best that the country had to offer. And she had luminal-a breast cancer, and she had her cancer repeatedly biopsied and arrayed and it never changed one bit of its nice luminal-a, this isn't the fatal kind of breast cancer’s mark, and we lost her. And one time we were talking about access, and she said to me access to what? And I—I want to end with that note, not that—to be pejorative about the hard work we’ve all done and the progress that’s been made, but sort of a call to action that we need to do a lot better. And I’ll stop there.