pearls to take back in the primary care setting of things you don't want to miss in terms of thinking

Transcription

1 RACHEL C. JANKOWITZ, MD 1 Good morning everyone. Thanks for coming out this morning. So today I'm going to be talking to you about breast cancer risk assessment hopefully from the perspective that I can just give you some pearls to take back in the primary care setting of things you don't want to miss in terms of thinking about breast cancer risk. So I'm going to start by talking a little bit about the epidemiology of breast cancer and then move into risk assessment for the individual patient and then finally touch on some of the things Darcy talked about with hereditary breast cancer and genetic testing. So I think we are all pretty familiar with standard breast cancer risk factors, you know some of the most common being age as a continuous risk factor for breast cancer, things like previous chest irradiation in the setting of Hodgkin's disease and all of the other things listed on this slide. However when we think about breast cancer risk I like to think about it in two, in sort of two categories, risk that is modifiable and risk that is not. And in yellow we have things that are somewhat modifiable in our life that can definitely affect breast cancer risk and something that you can think about counseling your patient to decrease breast cancer risk. So it's estimated that over a quarter of all new cases of breast cancer in the United Kingdom in 2010 could be attributed at least partially to lifestyle factors. Obesity plays a role in up to 23% of postmenopausal breast cancer cases and I'm not saying that obesity was the only thing that caused the breast cancer in that individual obviously, but certainly a factor, one of many factors but one that's modifiable potentially. Physical activity is another one, and alcohol is one that has been shown in many, many studies to increase breast cancer risk in sort of a dose related way. So the more you drink the more it increases risk. So these are things we can talk about with our patients. And

2 RACHEL C. JANKOWITZ, MD 2 actually there have been some studies in the literature looking about what doctors talk to their patients about in terms of breast cancer risk and commonly these two topics are not brought up. So family history is talked about a lot, hormone replacement therapy is touched on but for whatever reason doctors don't usually talk to patients a lot about these modifiable risk factors. So this is a metaanalysis of the association between BMI and breast cancer risk in postmenopausal women. You can see with all the many studies that have been done the forest plot that the relative risk for most of the trials shows increased relative risk with increasing with a higher BMI. Some of the biggest trials are in the bigger boxes. And this is a more recent study showing that there is an overall nonlinear dose response with BMI and breast cancer risk in postmenopausal women, so it's not completely linear the response but the risk certainly goes up with increasing BMI. The vertical bars are confidence intervals so you know there are wide confidence intervals here but the trend is certainly in that direction. So finally this one other risk factor is premenopausal weight gain. So there was a recent study showing that if you gained weight when you were premenopausal it increased risk a lot in a short term way. But Dr. Arndt is going to cover this topic in more detail so I'm going to skip this part. Okay, so now getting into risk assessment for the individual patient. So risk assessment for the individual patient is tricky and difficult and we hear terms floating around like moderate risk, high

3 RACHEL C. JANKOWITZ, MD 3 risk, hereditary risk, familial risk and sometimes it's hard to sort through those terms and understand what they mean. So when we talk about moderate risk of breast cancer we are talking about women who we consider maybe a 15 to 20% lifetime risk of breast cancer based on risk assessment tools that are largely dependent on family history such as Claus, BRCAPRO, BOADICEA and Tyrer-Cuzick. But in actuality no one in America is at particularly low risk of breast cancer because the baseline population risk in the United States is 1 in 8, which is about a 12% risk of breast cancer. So we have to take - we have to keep that background level of population risk in mind. Other things that moderately increase risk of breast cancer are high risk proliferative pathologies such as the in-situ pathologies atypia and LCIS and now more and more is coming out about breast density and breast cancer risk. So what kind of risk prediction models do we use in the clinic? Some of the most common ones that you hear about are the GAIL model, Claus and BRCAPRO which are more trained to flesh out more hereditary causes of breast cancer. The Tyrer-Cuzick model and a newer model called the Breast Cancer Surveillance Consortium Risk Calculator which incorporates breast density. So the GAIL model is an extremely easy tool to use, if you just Google GAIL model it will bring you immediately to the website, it's free and anyone can use it. So you just answer quickly 7 questions and it gives you a 5 year and a lifetime risk estimate of breast cancer. So this model is not validated in anyone who has DCIS or LCIS so it won't let you answer the questions if you say yes to

4 RACHEL C. JANKOWITZ, MD 4 number 1. But the GAIL model is the most classic risk prediction model that was used for all of the Breast Cancer Prevention trials in terms of eligibility. And in order to go on the Breast Cancer Prevention trials you had to have 5 year risk of breast cancer of more than 1.7% by the GAIL model. Now does that mean that every patient that has a 5 year risk by the GAIL model of 1.7% should go on medical prevention therapy? No, because it turns out that a 5 year risk of 1.7% is exactly the average risk for a 65 year old woman. But there was a very nice paper in 2011 that showed that when you look at that 5 year risk that you can calculate quite quickly you can just sort of quickly estimate the benefit of SERMs like Tamoxifen or Raloxifene in terms of breast cancer prevention. So this color coded chart shows you and this is in women with a uterus that according to their 5 year GAIL model score as it increases the evidence goes higher and higher in terms of favoring the use of SERM therapy to lower breast cancer risk. And you can see that without a uterus on the next slide the risk really starts - I mean the benefit really starts to go up with SERM therapy in terms of even with lower 5 year GAIL scores. So my suggestion number 1 to you is if you want to quickly think about someone who might be - who might benefit from medical prevention therapy quickly do a GAIL score and if their score is - their 5 year risk is greater than 2 1/2% even if they do have a uterus, going back to the last slide, that might be somebody you might want to refer to our high risk clinic to talk about the benefits of medications like Tamoxifen or Raloxifene because that's where you get into the range where the evidence favors the benefits over the risks of the therapy.

5 RACHEL C. JANKOWITZ, MD 5 Here is another risk evaluator called the Tyrer-Cusick model. So the Tyrer-Cusick Risk Evaluator is a little bit more cumbersome to do. I think it's less easy to do in a busy primary care setting because it takes me at least 10 minutes to do for each patient. I can show you some of the things that go into the Tyrer-Cusick model, it's more detailed, it takes into account height and weight, high risk proliferative pathologies and then it also very interestingly builds a little family tree when you plug in the family history. However you do need to know the ages of the unaffected individuals in the family so you really do have to sit down and do a little family tree and ask them how old is your paternal grandmother, how old is your maternal grandmother? How old is your - how many aunts do you have, how old were they? Did they or did they not have breast and ovarian cancer? But when you plug in all of that information it starts to build a family tree and then it gives you, you know, a risk of breast cancer and it also gives you the risk of having a BRCA gene mutation. So I find this part of it actually more handy, that patients are often impressed by how low their risk of having a BRCA gene mutation is. The model also knows whether anyone in the family has been BRCA tested already and you can reflect that. And so this is sort of the end page you get for each patient. So patients like it because it gives them this pretty picture at the end. However there is a problem I believe with the Tyrer-Cusick model and it definitely increases risk of - it overestimates risk of breast cancer in women with atypia. So that was demonstrated in a study in So this study looked at the performance of Tyrer-Cusick in comparison to GAIL for women with atypia and you can see that the predicted number of breast cancer in 10 years in 330 women with atypical hyperplasia the Tyrer-Cusick said that 59 of them would get breast cancer and GAIL said that 30 of them would get breast cancer. And when you look at the observed number of breast

6 RACHEL C. JANKOWITZ, MD 6 cancers that occurred the GAIL model was actually dead on with how many people would get breast cancer but the Tyrer-Cusick model significantly overestimated that number. The other problem is that neither model was particularly concordant on an individual level. So although GAIL did a good job of saying how many people were going to get breast cancer in that cohort, on the individual level for each individual patient the concordance was pretty low. So we have to keep in perspective that none of these models are perfect and you can calculate them and give someone a projected range of risk but you have to counsel them that these numbers need to be taken with a grain of salt, particularly for women with high risk proliferative pathologies. So the authors of that paper said then perhaps - their conclusion was that perhaps with women with atypia we should not be using either the GAIL model or the Tyrer-Cusick model and instead we should quote them exactly what we saw in our study, and that came down to the fact that they saw about a 1% per year risk of breast cancer with women with atypical hyperplasia and that' soften a very common number that we use in the setting of LICIS as well, a lot of the studies have estimated about a 1% per year risk. So if you want - you know it's better just a quick and dirty thing just to tell the patient if they have either atypia or LICIS it's reasonable to increase - to estimate that their breast cancer risk is on the order of 1% per year. Now getting into breast density which I know a lot of other people are talking about in this talk as well. This is in this morning's session as well. We know that breast density increases breast cancer particularly if you are in the highest risk category. And some studies have suggested this can be up to a 4 to 5-fold elevation in risk. And unlike some other risk factors breast density can be modifiable

7 RACHEL C. JANKOWITZ, MD 7 with prevention therapy. So this was a really interesting study which was embedded in the IBIS-I Breast Cancer Intervention Study and it randomized women to either - that trial was one of the original prevention trials where they randomized women to either Tamoxifen or placebo for 5 years. But what they did in this little case control study within that study was that they looked at the breast density at baseline in the women before and after they took their Tamoxifen. So it was really just baseline and then 12 to 18 months later after they were on their Tamoxifen or placebo. And aside from have atypic or LCIS the breast density was the single most important factor for predicting risk of breast cancer on that small study. And the actual predictive value of the density increased when they adjusted in multivariant analysis for the BMI in the presence of atypia. And so the odds ratio for developing breast cancer that was associated with that highest density category that's what we see on the mammogram at points when it says extremely dense 76 to 100% density, so that actually went up from 2.83 in univariant analysis to 3.48 in the multivariant analysis. So suggestion number 2 to you in your practice is consider supplemental screening with ultrasound in extremely dense patients. So you know there is not that many of them when you see your mammogram reports come through, you see a lot heterogeneously dense but you don't see that many that come through that say extremely dense and I do believe that those women are at increased risk of breast cancer, and you can talk to them about the pros and cons of supplemental screening and I'm not going to get into too much detail about that but ultrasound does pickup more breast cancer when added to the mammogram in extremely dense breast cancer patients.

8 RACHEL C. JANKOWITZ, MD 8 So the final point about that trial in the Tamoxifen arm 46% of the women had a greater than 10% reduction in their breast cancer risk and a 63% reduction in their breast cancer risk - I'm sorry, so half the women had more than 10% reduction hat should say in their density and that correlated to a 63% reduction in their breast cancer risk. So women who did not have a 10% reduction in their breast density on Tamoxifen actually got no risk reduction from the Tamoxifen which was a very interesting signal in that trial indicating that perhaps the women that were dropping their density on Tamoxifen were the ones getting the most risk reduction. So this is that final model I had mentioned to you, the Breast Cancer Surveillance Consortium Risk Calculator. So this is basically a modified GAIL and you answer these questions. Another very quick and easy model to do in the clinic and it incorporates their breast density and so that is one I think that is easier and more doable in a quick way in the middle of clinic. So finally the last point, hereditary breast cancer and genetic testing. I know Darcy just talked a lot about this topic but sometimes these terms get confusing and what does it mean to have familial risk versus hereditary risk, so I just wanted to review that quickly. So familial is defined as a multifactorial disease with polygenic factors such as hormone exposure, environmental factors, lifestyle factors contributing to risk. Hereditary cancer means that there are germline mutations in high-penetrance or moderate-penetrance cancer susceptibility genes and we think that they cause at least 10% of breast cancer. So classic characteristics of hereditary breasts and ovarian cancer families include early onset, young age, bilateral disease, male breast cancer, multiple affected family members and a combination family history of ovarian, primary peritoneal, pancreatic and

9 RACHEL C. JANKOWITZ, MD 9 prostate cancer. So if you think you see those other cancers in combination with breast cancer at young ages that's when the red flag should be going off that we need to be thinking about a hereditary process. So the third suggestion is you know pay attention to these clues for hereditary breast and ovarian cancer syndrome when you are seeing these patients in clinic. And these are the testing criteria according to NCCN and they reflect those clues that we just talked about, but it is hard to like open NCCN every time you are seeing a patient and see if they formally meet criteria for testing and you probably don't have time to do that on a daily basis. So just you know keep the clues on the following slide in mind and but just so you know these are the testing criteria. And these are the testing criteria for women who have family history only as opposed to the previous slide was women who are affected by breast cancer, this is the testing criteria just so you have them handy. So if the criteria are met what do you do? And I think Darcy just touched a lot about that. You go - if there is not a familial mutation known do you start with BRCA testing or do you go to a multigene panel? And that's sort of what's coming in the future. So traditional cancer genetic testing is usually done in a serial fashion and you think about what's the most likely mutation. You test for that first. The advantage to doing it that way is it's more specific, you have more time for pretest counseling for each individual gene. The disadvantage is it's time consuming and it's expensive and sometimes you only have one shot with insurers so sometimes they won't let you do a panel after you've already done BRCA. So the disadvantage also is that individuals with mutations in some of the more

10 RACHEL C. JANKOWITZ, MD 10 moderate penetrance genes are less likely to have a striking family history like a BRCA family would. So the next issue is that DNA sequencing cost has decreased significantly with next generation sequencing technologies and multigene panels are now available whether we like it or not. They include the high-penetrance genes but they also include a lot of other genes with that increased cancer risk in a more moderate way, but the problem is there is a lack of clear guidelines about how to take care of people who have some of these other genetic mutations. So when we think about the other breast cancer genes we know that BRCA is not the whole story anymore, just recently we had a big New England Journal of Medicine article on PALB2. You can see that many, many genes are included on this slide and I don't have time to go through them all but I'm going to touch on just a couple. But we are somewhat soon I think going to be - and I borrowed this slide from Miranda, one of our genetic counselors, so I have to give her credit for helping me with some of these slides. But you know we are going to be drowning in next generation sequencing data before we know it. So again the gene panels can include many genes that increase risk of other cancers besides breast cancer and the guidelines are unclear. There are clear guidelines for some of the genes such as BRCA, PTEN, STK11 AND p53 and they do clearly include the option of prophylactic surgery such as mastectomies, however for the other ones there are no management guidelines for some of them. So when you think about the genes that are included on these panels they are usually divided into

11 RACHEL C. JANKOWITZ, MD 11 high-penetrance genes, meaning you know markedly elevated risk, moderate-penetrance genes and unknown. And os for the ones that are - we know a lot about their phenotype we have clear guidelines and you can see over here that BRCA 1 and 2 are included on that slide. You can see that as of now PALB2 is still not made it into the newest iteration of the guidelines yet despite the New England Journal article and then some of these ones down here are still really a gray zone. We can see that the high risk genes confer up to an 85% risk of breast cancer, the moderate risk genes you are talking more of a 20 to 50% lifetime risk and familial risk being more 12%. So this is just my slide that I made up about genetic risk not to miss in your clinic because when you are thinking about syndromes that cluster with breast cancer, and I didn't include BRCA because I think I already did the HBOC BRCA slide a few slides ago. But when you are thinking about does that cancer go with breast cancer and which syndrome is that again so Cowden syndrome is PTEN, breast, thyroid, endometrial. So if you hear breast, thyroid, endometrial in a family think Cowden syndrome, think about referring for PTEN testing. Peutz-Jeghers which is the hereditary polyposis cause benign polyps in the GI tract, hyperpigmented macules on the lips and oral mucosa and can have a number of cancers including colon cancer and breast in combination. Li Fraumeni can cause just about any kind of cancer but the big buzz words when you think about that in a family history are sarcoma, a brain tumor, adrenocortical tumor and breast tumor and leukemia. And that also young lung cancers. So in terms of the Lynch syndrome we all know about the colon cancer, endometrial, gastric, ovarian, pancreatic. Breast cancer is a little bit less clear about how much it clusters with Lynch syndrome. And then the hereditary diffuse gastric syndrome is the combination

12 RACHEL C. JANKOWITZ, MD 12 of gastric cancer with lobular breast cancer and that's due to the CDH1. So just something to think about if you hear these cancers clustered in a family history. These are some of the panel options that are out there and all of the genes they include. You can see as Darcy said that some of these panels can be quite costly and the insurance coverage varies according to what the patient's insurer is. So we have to have informed consent and we strongly believe in our department that we should not be offering these panel tests without the guidance of our genetic counselors due to all the limitations that I just reviewed. I just wanted to quickly do an example of how much trouble and sort of uncertainty you can get yourself into when you think about ordering one of these panel tests. So I saw a lady in the breast cancer prevention clinic due to a family history of breast cancer. She herself had not had any breast biopsies or surgeries, she was relatively healthy in her other breast cancer risk factors. This is her pedigree and this is our patient who is unaffected by cancer but she had two sisters who were affected by breast cancer and it turns out that her one sister had a CHEK2 mutation and that had been documented and at the time that I saw her, her other sister who had breast cancer had CHEK2 testing but we didn't know the result yet was going to be. So what does this mean for our patient? /She has a first degree relative with one of these moderate-penetrance gene mutations, how does that affect her breast cancer risk? So CHEK2 is one of these mutations, it's involved in DNA damage signal pathways and they have been associated as modifiers of breast cancer risk. This is the most common one that has been seen.

13 RACHEL C. JANKOWITZ, MD 13 So the lifetime breast cancer risk in an unaffected woman with a CHEK2 mutation and no family history of breast cancer is about 20 to 25%. However the risk in a woman with a CHEK2 mutation is not determined solely by the presence of this mutation, the penetrance of the mutation is also determined by whether she has a family history of cancer. And this is very common for a lot of these moderate-penetrance genes because they don't work alone, they work together. So there will be a higher breast cancer risk if the mutation is present with family history like this patient. And this is a graph illustrating that, that the lifetime risk goes up according to the amount of family members that are affected. So in other words if the person has the mutation, no family history she's down in a 20% range, but as you can see here with both a first and second degree relative the risk is up more in the 40% range, so that's pretty significant. The other tricky part is that if my patient tests positive for the same mutation that her sister has I can pretty clearly tell her then okay you have you know upwards of a 40% lifetime risk of breast cancer. However if she tests negative for that mutation that's where it gets very murky because it's not entirely clear that family members that don't have these mutations that their affected family members do have are not still - they may still be at elevated risk because of the fact that they are modifiers of risk. And they modify risk of cancer imparted by other genes. So even if she tests negative for the mutation that her sister has she's probably still at elevated risk of breast cancer and we would recommend addressing this level of risk purely based on her family history. So if you calculate her Claus score it's 19.8% which is very high and a Claus of 20% is where we consider the use of screening MRI. So in actuality the testing did not change my management of this patient at all whether she tests positive or negative - well if she tests positive I'll be able to tell her her risk is

14 RACHEL C. JANKOWITZ, MD 14 definitely higher based on her two affected family members. But even if she tests negative I'm still going to screen her as a high risk patient because at this time I can't clearly tell her she's not at elevated risk. So that's where it gets a little bit tricky. We recently also saw sort of a similar pattern with the PALB2 mutation that breast cancer risk in PALB2 is a lot higher than we thought it was. It's a partner and localizer of BRCA2 and we previously thought it conferred about a 2 to 4-fold elevation in risk in non-mutated - in the mutation carriers. It's rare. This study looked at 14 centers and in those 14 centers there was 175 families with at least one member with a PALB2 mutation and the bottom line is the result of the trial showed that it was a cumulative risk of breast cancer in a carrier of about 35%. But again here we see again that the risk varied by family history. So with no family history it was 33%, but it was up to 58% if they were a carrier with two first degree relatives with breast cancer. So this is now approaching the level of a BRCA mutation here and so, so this is sort of a big deal. And it hasn't been incorporated into as of yet into the guidelines but this, a patient with this type of mutation is going to be eligible for the same type of screening and risk reduction options in my mind as a BRCA carrier. So this is the end of my slide. I'm actually approaching the end of the slides but finally I'm just going to talk about the fact that it is difficult to ball park risk, individualize risk even for the BRCA mutation carriers. What we do is we tell them about an average risk per decade based on their - based on their current age. And this study in JCO illustrated that pretty nicely. You can see overall the BRCA1 risks are higher than the BRCA2. And there are people working on looking at modifiers of risk in BRCA 1 and 2. The largest one is the CIMBA organization and our high risk registry is

15 RACHEL C. JANKOWITZ, MD 15 sending DNA and family history to this worldwide organization to try to clarify risk better for BRCA mutation carriers. And as of yet we have submitted over 300 samples to CIMBA and just this month CIMBA came out with their first paper showing that actually the type and the location of BRCA1 and 2 mutations does track with higher risk of breast versus ovarian cancer, so it's really exciting to see this data finally start to come out. It's a little bit too early to be using clinically but it's nice to see that and hopefully for future generations we will be able to characterize risk better on an individual level. And then just a final word about variants. When you see BRCA mutations and variants of uncertain significance they normally, the vast majority of them get reclassified as benign polymorphisms so do not immediately jump to the conclusion that a patient with a variant is going to have the same risk as a patient with a BRCA mutation. The variants are reclassified over time and so they should check in annually with their genetic counselors. And so these are sort of all the options we have for risk assessment and handling of patients at increased risk of breast cancer and just to remind you that if a woman has a BRCA mutation doing a BSO can reduce her risk of overall mortality in the setting of BRCA1 by almost 80% and BRCA2 by 53%. So just getting their ovaries out whether they do or do not decide to do the prophylactic mastectomies has a huge impact on their overall mortality so that's why we don't want to miss those patients.

16 RACHEL C. JANKOWITZ, MD 16 And so these are my conclusions. Some breast cancer risk factors are modifiable. We need to do a better job of individualizing breast cancer risk prediction and hopefully in the future the models that include breast density, things like SNPs will help improve our risk prediction over time. And risk evaluation with multigene panels at this time should really be undertaken with extreme caution and the involvement of a genetic counselor and we really want to identify the BRCA mutation carriers so we can improve their outcomes. Thank you.