Myriad Oncology Live episodes are recordings of an open-forum webinar hosted by Dr. Thomas Slavin. The opinions and views expressed in this recording do not necessarily represent those of Myriad Genetics or its affiliates. To participate in a future recording, visit myriad-oncology.com/myriad-oncology-live for a list of dates, times, and subjects.
Additional resources cited in this episode:
Hu et al NEJM - A population based study of genes previously implicated in Breast Cancer 2021. https://www.nejm.org/doi/pdf/10.1056/NEJMoa2005936
Palmer et al. JNCI paper - Contribution of Germline Predisposition Gene Mutations to Breast Cancer Risk in African American Women. https://pubmed.ncbi.nlm.nih.gov/32427313/
Antoniou et al NEJM - Breast caner risk in families with mutations in PALB2 2014. https://pubmed.ncbi.nlm.nih.gov/25099575/
Yang et al 2020 PALB2 paper Dr. Couch referenced
BCAC study - NEJM 2021 https://www.nejm.org/doi/full/10.1056/nejmoa1913948
Narod editorial to these two papers - CARRIERS and BCAC. https://www.nejm.org/doi/full/10.1056/NEJMe2035083?query=recirc_curatedRelated_article
BCAC Study paper titled: Breast Cancer Risk Genes - Association in More than 113,000 Women CARRIERS study is titled - A Population Based Study of Genes Previously Implicated in Breast Cancer
0:00:11.5 Thomas Slavin: Welcome. This episode of Inside The GENOME is a recent recording of Myriad Oncology Live, a webinar hosted by me, Dr. Thomas Slavin, Senior Vice President of Medical Affairs at Myriad Oncology. The opinions and views expressed in this recording do not necessarily represent those of Myriad Genetics or its affiliates. To participate in a future recording, please visit Myriad Oncology for a list of dates, times, and subjects. I look forward to exploring the world of genetics with you all.
0:00:40.0 TS: Hello everyone, welcome to Myriad Oncology Live. Thanks for taking time during your day to treat yourself to a little education. A little housekeeping to start and then we'll introduce our special guest today. So today we are talking about hereditary genes with limited guidelines. I thought about some of the recent work that's been done in unselected populations for hereditary cancer and reached out to Dr. Couch. So Fergus Couch will be joining us today, so thank you so much. For those of you who don't know, there are few people on the planet [chuckle] that are influencing hereditary genetic testing as much as Dr. Couch is, period. He's definitely in the tippy top of that category, so we are very privileged to have him come on today and share his expertise and thoughts on some of the genes that we all test and deal with on a sometimes daily basis.
0:01:53.8 TS: And then next week, we will get into pros and cons of germline testing for all breast cancer patients. This one will be a fun one as well. So we'll have pros versus cons, although I didn't really set it up as pros versus cons, but we'll get into both. We're gonna be joined on that call by two people. We'll have Dr. Mark Robson, who many of you know from Memorial Sloan Kettering. He is an absolute expert and really the clinical translational expert... Or interface, I should say, clinical translational interface of what we're doing in the research labs and commercial testing labs and what it actually means for patients. So we're very fortunate to have him.
0:02:41.4 TS: And then also Dr. Paul Baron, who is a breast surgeon in New York, who's very closely associated with the American Society of Breast Surgeons, and has been an advocate for moving genetic testing down into a lot of breast cancer patients. We'll also talk about OlympiA, which is a new study. So if you haven't heard what that is, don't worry, you can just show up next week, you'll learn all about it. (EDITED) And then in August, we're going to talk about polygenic risk scores. Holly Pederson from the Cleveland Clinic is coming back for this one 'cause people really enjoyed her on the last time we touched on polygenic risk scores. And then we showed some work across all ancestries at ASCO, so we'll dive into that a little bit. But if this is your first time on this webinar, these are completely open-ended, they're just an open door for discussion. Feel free to ask literally any single question. The focus is themed and today's is just really around hereditary cancer genes, but if you have a burning question about Oncotype DX or gene expression profile for breast cancer or something, feel free to also ask that. And then a little, last housekeeping, we are now recording these sessions just for people that can't attend, and then we've been... Oh, I hit the wrong link. We are posting these now on the podcast that I have, and it's a little messy but I think people can figure it out. So anything that says "Myriad Live" on here is just a recording of the podcast that... Or this webinar series.
0:05:00.2 TS: And so there's some really good ones, of the recent ones that we've been doing. So I'm glad that we're actually preserving them in some way, shape, or form. Things that don't say "Myriad Live" are just the standard podcast. So the Myriad Live ones are 45 minutes, so they're a little longer for sure, the ones that don't say "Myriad Live" are just me sitting down with experts in the field talking about something of interest. We just posted a few, this one with Dr. Karen Hurley around psychosocial aspects of hereditary genetic testing. And then this one I did a while ago, we just posted it, but it's very good. If you don't know about breast cancer gene expression essays, this is an A to Z coverage. Dr. Adam Brufsky is a absolute expert in this field, he's been involved in it a long time.
0:05:50.6 TS: And so you can ask questions, I encourage you all to just unmute yourself, ask a question, and we will get it answered. If you don't feel comfortable asking a question, you are free to type it in the chat, and I believe I'm trying to... Let me stop sharing real quick, but I believe Shelly is on, thank you, Shelly. So Shelly will be fielding questions there if you send them through the chat. And then I guess the last bit of housekeeping is, congratulations to Shelly. So Shelly took my old job. [chuckle] So she went to... So now, Shelly is the Vice President of Medical Affairs for Myriad, so all things medical affairs-related on the oncology side, and then I moved onto Chief Medical Officer. So now... And we've been thinking, Shelly and I have been talking as it pertains to this webinar that maybe we'll open it up across kind of all things instead of just cancer focus.
0:06:54.7 TS: So be on the lookout for that. We're still kind of crafting what that might look like, but we definitely have a lot of prenatal expertise here in-house because we have a lot of folks from council that was acquired some years ago. We have a very robust pharmacogenomics program here, so I think there's a lot of opportunity for some cross-education. I still think that the main... We'll still have it largely oncology-focused, whether it's just unaffected gene carrier, hereditary tumor, affected, etcetera, so be on the lookout for that as we go forward. So without further ado. Thank you, Dr. Couch for coming on. I don't know if maybe you wanna... You wouldn't mind giving the folks a little run down about what you do and your role at Mayo, just to orient folks. I'm sure many people on the call have some familiarity with you.
0:07:57.4 FC: Yeah, I've been in this game, in the pre-disposition game from the early days, I was a postdoc, actually postdoc fellow with Barbara Weber and Francis Collins in the days of the BRCA1 search, and then we went on and worked with Myriad actually on the BRCA2 search and have stayed in the field all along through that, so I've led a number of the GWAS studies that were done, and particularly in BRCA1 and triple negative breast cancer. And then I've sort of gone back to my roots of germline genetics and have been working on a lot of different data sets. And most recently, it was this New England Journal paper that we're gonna talk about today, which is the big population-based series that we ran, which is about 30,000 cases and 3000 controls. And we did all that testing in my own lab, and then I think that's probably the last large study that'll ever be done in an individual academic lab because the companies are doing so much of this these days that no single investigator can keep up. So I think academic researchers are becoming more data analyst-driven experimentalist rather than direct genetic experiments in their lab.
0:09:07.1 FC: But yes, still working on this, still working a lot on VUS, Variants of Uncertain Significance in BRCA2 and a whole range of other genes. We have functional assays running for those, and we're trying to classify a whole bunch of those guys with the ACMG type guidelines. We're involved in ClinGen. I lead the HBOC VCEP, which is the Hereditary breast, Ovarian and Pancreatic cancer VCEP for classification variants in those genes, in the genes associated with those cancers. And involved in the BRCA1 and 2 VCEP, through ClinGen as well. So just lots of genetic stuff with a few functional assays thrown in. Yeah, so...
0:09:46.7 TS: Yeah, busy guy, [chuckle] to say the least. And I always remember, I think the first time I met you, I brought up, oh yeah, I learned the Couch model when we were trying to figure out risks of being a carrier of mutations. Or that was actually the Couch model. No that was more risks of breast cancer, lifetime risks, yes.
0:10:09.9 FC: Yeah, that's right.
0:10:10.2 TS: Yes, and that was... It predated Tyrer-Cuzick. Correct, or was it around the same time.
0:10:15.9 FC: Oh, yeah, it was way back at the time of the empirical models that Myriad had back in those days as well, and those were the two main models that people would use to estimate risks. Yeah, when the genes are first being identified, we run them through a fairly large number of high-risk probands with a lot of family history data, and we're just able to make some simple estimates. And I think they worked out quite useful... They turned out to be quite useful in those early days when there were no better models available, but of course they've been superseded by much more complex models for many years now.
0:10:51.2 TS: Yeah, no, and you've done just so much. And the functional work that you've done, you've had some pretty good papers over the last six months on the functional side. So people who haven't seen those. You can search Dr. Couch's name and his lab is well known for really picking apart the nitty-gritty of these, particularly moderate penetrants of BRCA1 and 2 and other mutations. You've done a lot of work in the HRD pathway, period. Yeah, we're honored to have you on, so thank you.
0:11:24.0 FC: Thank you.
0:11:24.4 TS: I will pull up... I'm gonna share my screen and pull up the paper. Let me see, I have it here. Where is it? Here, carriers. So does it look really odd to people or does that look normal?
0:11:42.3 FC: No, that looks good.
0:11:43.0 TS: Okay good.
0:11:44.3 Speaker 3: It's a little small.
0:11:44.8 FC: It's a little small but I think you can make it out.
0:11:46.3 Shelly: It needs to be bigger, for sure.
0:11:49.0 TS: Okay, let me stop sharing. Let me try one more time. Let me try that. Is that better?
0:11:56.5 FC: Yes, that's good.
0:11:57.9 TS: Great, okay. So yeah, if people haven't seen this thing, Shelly, I saw you put the citation in the chat, this is really sentinel work. And I was thinking maybe we could start with this paper and then we can move over with some of the significance, two of the... Yeah the bridges. I think from a... What was your nidus for really doing this work in the first place?
0:12:27.3 FC: So I think the story is fairly obvious to most people is that we know that the risks that are out there for BRCA1 and 2 and some of the other genes are predominantly derived from high risk probands because these are the people who tend to get tested. They fulfill NCCN criteria, but then there was a couple of JCO papers and a paper that we actually had as well over the last two or three years, really showing that as many as 50% of people with mutations in those genes were completely missed by NCCN guidelines.
0:12:58.7 FC: So that led us to believe that in fact, the risk estimates that were out there and were being used weren't really appropriate for those many people who wouldn't necessarily have a strong family history or a particularly young age of onset of disease. What we really needed was a large population-based, totally unselected series of patients to really estimate what those risks would be and would they be dramatically different from the high-risk population risks. You could envisage at one point where you might have a patient walks in the door and if they have a strong family history, you give them one set of risks and if they have no family history at all, you give them a different set of risks, it could be two sheets stuck on your wall sort of thing, so that was the mindset there. And we were lucky enough to be able to team up with you can see it amongst all the authors and where they're from, but many of the very large cohorts, true epidemiology cohorts, where you essentially enroll 50 to 100,000 people at baseline, all unaffected and you follow them through time, 20, 30 years, and some number of them will get breast cancer, so these are breast cancers arising in, essentially, the general population.
0:14:06.7 FC: You've got perfectly matched controls or unaffected people who match up with them based on age, family history, a whole bunch of other criteria, and so they truly do reflect the underlying US population and so that's what... We were lucky enough to do that. We teamed up with about 8 or 10 of those groups, so the California Teachers Study, the Nurses' Health study, the American Cancer Society, CPS2 and 3, the multi-ethnic cohorts, Black Women's Health study, you can see them all listed there that Tom just brought up. A bunch of them there, and many of you will recognize those names because they're very well-known in the cancer epidemiology community. So yeah, we were able to pull all those together. Luckily enough, they had blood or buckle samples from the vast majority of their breast cancer cases and many of the controls.
0:14:58.2 FC: And so we were able to obtain all that DNA and bring it into my lab and test it for a panel of predisposition genes, and we used actually a QIAGEN amplicon-based panel. And we'll admit up front that maybe it's not quite as perfect as the Myriad approach or some of the other companies in the way they do it, because you guys have so much effort put into quality control and all that sort of stuff 'cause you have to from a clinical perspective. I think we did pretty well. We've done a number of pilot studies with the panel and show that we could essentially find 100% of the mutations in two or three different blinded studies, so that was good. The one thing, if we do miss something, it's probably one of these large genomic rearrangements that the commercial entities are better at finding.
0:15:45.9 TS: And I think if I remember, you accounted for that in this...
0:15:50.5 FC: Yeah, exactly. We try to do it a couple of different ways to try and account for that. And so we were able to basically estimate frequencies of mutations in the different genes and then also estimate risks, and so just simple case control association studies, adjusting for things like age of diagnosis or age of last follow-up, a family history of breast and ovarian cancer. Ethnicity, ethnicity was a big thing here, and we did it a couple of different ways, you can see it right in middle of this screen right now, they're nearly all non-Hispanic white, as you might imagine, but there were a fair few blacks in there as well, about 4000 cases and about 5000 controls. So we had a separate study published about a year and a half ago now in GNCI, looking at those 4000 black cases and the 5000 controls and just the frequency mutations and the risks in the black population. So this study was all of them thrown in together, and then we did sort of a sensitivity analysis where we restricted it to the non-Hispanic whites just to see if there was any difference, and the reality was there was no difference. You couldn't really see much of a change when you subset to the specific non-Hispanic white population.
0:17:09.4 TS: Yeah, and what do you think were the biggest findings in the... All this work, it's a monster.
0:17:16.7 FC: Yeah, exactly, it's a monster, but there weren't really any sharks. That table you just had there, the list of genes, that one. Yeah, that kinda tells you most of what we found. First of all, we confirm that in the general population, again, not specifically selected for a particularly young age of onset or family history, you still find about 1% of cases of BRCA1 and about 1% of cases of BRCA2 mutations, so about 2% when you combine the two genes. CHEK2 is also around 1%, ATM is just slightly less than 1%. These are the four main genes and then you throw PALB2 in there at about a half a percent, 0.46% on this table. So those four genes together, you're at 4.5-5% of all unselected cases with breast cancer in the general population is gonna have a mutation, one of those five genes, and then the other genes are just...
0:18:14.0 TS: Yeah, which is a lot. Yeah.
0:18:15.3 FC: Yeah, it is quite a lot really, it was surprising to us that it was so high. When I started the study, I thought it might come in around 2%, but here we are at somewhere between 5% and 6% for all the genes in there, and we did restrict it. You can see the list there. We restricted it to about 12 genes that we thought were directly clinically relevant, essentially actionable breast cancer genes, there's a lot of other genes in the supplements that we don't need to look at today, but things like NBN, MRE11, RAD50 which we showed in this paper and in a few other papers that they really don't confer an increased risk of breast cancer.
0:18:53.0 TS: I'll pull it up, I do have the supplement here too, just... Also because I feel like I tricked people a little bit in the sense that I said this was rare genes. [chuckle] I know we're talking about standard genes but yeah, we can pull that up towards the end and look through that for sure, just to make sure.
0:19:09.3 FC: Yeah, sure. So there were genes like that, the mismatch repair genes were on the panel and we have some data there, but they're just exceedingly rare to find mutations in those genes in the breast cancer group. So, the take home message with the frequency, which was around 5-6% in cases, and around 2% in controls, and that's equally surprising, that you could have as much as 2% of totally unaffected women at an average age of let's say, 52-55 walking around on the street from these cohorts and they have mutations in these genes. So, there's just huge numbers of women out there that are carrying these things and don't even know it because we're not doing any population testing.
0:19:49.2 TS: Yeah.
0:19:49.7 FC: So, that was a surprise to begin with, and then as you go to the right-hand side with the odds ratio calculation, that's really telling you the approximation of risk. So, if you have a mutation in this gene like ATM, you have a 1.82 fulled increased risk over the general population. That's kind of the idea there, over the null. A couple of surprises there, obviously BRCA1 and BRCA2 were high risk genes, as you could see there, 7 and 5, and then PALB2 was at 3.8, and that's a little bit lower. Some of you might know there was a paper published, I think was late last year by a huge international consortium that we were involved in, and I think Myriad was too, but basically a consortium looking about 800 PALB2 carriers mainly from families, and they found that PALB2 mutations were really very strong. They had an odds ratio, or relative risk in that study of between 7 and 8. So, we come along with the general population here, and it's only 3.8 which is about half that.
0:20:51.9 TS: Yeah.
0:20:53.9 FC: Kind of makes some sense, actually, if you think about it, right? You would expect some drop in the risk, some attenuation of the risk when you go to the general population because you're not seeing those strong family history effects.
0:21:03.3 TS: Yeah.
0:21:03.6 FC: So, it's come down to moderate risk gene in this setting. But certainly, a high risk family setting.
0:21:10.6 TS: Yeah, I'm sorry to interrupt. The only thing I was gonna add there was, this was surprising to myself, and I know Alicia's on too, Alicia Hughes, who you may know.
0:21:19.8 FC: I saw her there. Yeah.
0:21:20.7 TS: Yeah, because she had that paper with Allison and some of the other people on this call that use multi-variable logistic regression to control for multiple things including family history, and actually the odds ratios here look very similar to that paper, that Korean 2017 paper, which is nice validation.
0:21:40.1 FC: Yeah. Yeah.
0:21:42.5 TS: Yeah.
0:21:42.7 FC: Yeah. So, I think that big study from that large consortium on PALB2, there's gotta be some flaws in it. I do think that it's enriched for family, strong family history positives. There's sometimes, you just can't adjust for these things enough if the original ascertainment is really biased. So, I think the scores in that other paper are just too high, and then we've got this, which is more representative of the general population. So, BRCA1, BRCA2, PALB2, all more or less behaved as we expected. CHEK2 was still around 2.5, which converts to about a 25% lifetime risk, which made sense, but ATM was a little bit lower than we thought. We thought ATM would look like CHEK2, but it's a little bit lower, 1.8 and some people use an odds ratio of 2.0 as sort of a clinical threshold for calling something actionable. It's pretty close at 1.8. I think you probably wouldn't change the approach to ATM at this point. Elsewhere in the paper, we did do modeling of the lifetime risks and you still get ATM at about 20% or a little bit higher. It's right around 18%, I think.
0:22:49.6 TS: Yeah. I think it's...
0:22:51.9 FC: So, yeah, there it is you can see...
0:22:55.3 TS: Yeah. This is a great figure. It's very nice to see this represented.
0:23:00.7 FC: Yeah. So, the dotted lines represent the general population that don't have mutations and predisposition genes, and then the black line represents the ones that have mutations and what their cumulative risk would be by age. And you can see that the ATM does squeeze up there to about 20%. So, with the mammography and MRI screening guidelines generally set at 20%, these five genes all still qualified for that. So, that was good and is essentially a validation of what we're doing today, but knowing now that what we had been doing for the high risks still applies in the general population, which is an important step. So, confirmatory to some extent. And then we...
0:23:40.1 TS: Yeah, and this was... Sorry, go ahead.
0:23:43.0 FC: No, please. Go ahead.
0:23:44.3 TS: I was just gonna say, this was so nice to see. Also, it's very similar to the Bridges BCAC data, because it's really showing that we probably shouldn't be running around telling everybody that they're 80% chance of breast cancer with a BRCA1 mutation, which has been strongly perpetuated, and... Someone can correct me if I'm wrong, but I think even the current NCCN guidelines say something like 60-80% or something. I mean, very high.
0:24:10.5 FC: Yeah. I think it's 69 for BRCA2 and 71 for BRCA1, or something like that.
0:24:16.5 TS: Yeah.
0:24:16.8 FC: And that's come out of...
0:24:17.4 TS: So I mean, it depends on the population.
0:24:21.0 FC: Right, but I think a lot of that has come out of some of the CIMBA consortium studies and some of the European groups where they tracked patients over time. Prospective sort of studies, but I do think they're a little bit inflated at times. Certainly playing with polygenic risk scores, you can see a number BRCA1 and BRCA2 carriers that have lower risk down around... They can go as low as 40% or so. So yeah, I think this was appropriate for, again, the general population and not the enriched group.
0:24:51.5 TS: Yeah. And I guess one thing that's confusing to me as we go forward as a genetics community is how do we start making the distinction between who's the high risk population and the general population? Because here we are, we're gonna see probably massive changes to who's recommended for genetic testing with the Olympia. Well, I guess that's affected though, but that will likely trickle down in some fashion to the unaffected. So, I think we'll probably start seeing an unaffected population that is just a little bit wider than we've been used to in the past, and so it's gonna be hard to know what kind of risk estimates to place on people without some sort of more complex modeling.
0:25:38.5 FC: Yeah, complex modeling or if we do want to keep it simple, we just have to rely on family history, which of course is what we were doing 25 years ago. So we're kinda going back to our origins in some ways, except we do have all the numbers now, which ever side of the line you fall on, we have the numbers to give you sort of detailed risk estimates, but identifying which bucket you fall into will probably be a family history based. Predominantly a family history-based model.
0:26:03.6 TS: Yeah, let's pause for a second. I know there's been some things on the chat. I don't know, Shelly, I know you've been posting some things.
0:26:12.5 FC: Yeah, you posted a PALB2 paper. That was the original PALB2 paper. There was a follow-up in 2020, I think it was.
0:26:18.7 Shelly: Okay, I'll look for it, sorry. I can't find it as you're referencing.
0:26:23.7 TS: Oh yeah, no problem. Sorry, Shelly. I do see a question here from Robin Palmer. "Other than BRCA genes, what genes do you consider early onset breast cancer genes?" So that's a good question. This figure hints at that a little bit.
0:26:42.1 FC: Yeah. And it's interesting because we looked at the average age of onset across mutations and all these various genes, and really, BRCA1 and BRCA2 were the only young onset ones. The average age... Most people would know this, but the average age for BRCA1 is about 42. The average age for BRCA2 is about 48, for breast cancer, that is. But then, in these other genes, we're finding them at 52 to 55 in this population, in this unselected series, which is a little bit older, and we actually also did some modeling of how frequent the mutations were at different ages.
0:27:20.1 FC: So if you look at BRCA1 and 2, the majority of the mutations are occurring in young women, 30, 40, 50 years of age, and then the mutations are quite rare in affected women over the age of 70, let's say 60, 70. You don't find very many. We know that there's definitely an age effect for those genes. But when we looked at the other genes, we only see a slight age effect of ATM and CHEK2, and we saw absolutely no age effect of PALB2, which is kind of interesting. This is the same frequency at age 65 as there was at age 40. We really couldn't see the difference...
0:27:55.8 TS: We're working on a paper that we're seeing the same right now, with Mary Daylilly and Allison Kurian.
0:28:04.3 FC: Right. Yeah, so there really doesn't seem to be an age differential there for PALB2, which I don't fully understand, but that's what we're all seeing, yeah. So after that, then the other genes that might have been young onset, there were strong effects. We went looking for more triple-negative type genes, and triple-negative being a relatively early onset type of breast cancer, mainly because of course we all think of it has been driven by BRCA1, but there's a bunch of other genes involved. And so, we ended up bringing in BARD1, RAD51C, and RAD51D. And there are very, very rare mutations in those genes, but when they do occur, they confer increased risk.
0:28:46.5 FC: So at the bottom of the table, you can see RAD51D there, it's almost four, an
observation of almost four, yeah, right there, which is quite strong. And we've seen that in number of different studies now, not just this one. It seems to hold up. That's ER-negative, and then the triple-negative, we couldn't calculate, there just wasn't enough events. But certainly, in ER-positive, it's relatively flat at 1.6. It just doesn't cause ER-positive disease. The RAD51C is kind of a moderate-risk gene and in the triple-negatives, ER-negatives as well, and then BARD1 up at the top of the table, same idea. So those three genes seem to partner up with BRCA1. If you think of four genes that cause triple-negative diseases, it's those four guys plus PALB2 to some extent. And so you can see PALB2 with ER-negatives, they're 9.2 and 13 and triple-negatives. It's a very strong effect. And BRCA2...
0:29:48.9 TS: And how do you think we'll incorporate this ultimately? How do you think this can move into clinical utility? Because, yeah, it is complex when you start talking about risk for triple-negative.
0:30:00.8 FC: It's extremely hard for someone to counsel a patient, "Oh. Your risk for ER-negative is this and your risk for positive is that." I don't know if we can counsel at that level. I think at least at the front end of counseling, we probably just have to talk about the genes overall.
0:30:22.1 TS: Because it's tough. I mean, right now, if you think about all the women today, they're getting results with positive RAD51C and D results, and they're being told that they really don't have a risk for breast cancer, and that to focus on the ovarian cancer risk and consideration of risk reducing salpingo-oophorectomy. Here they're at an increased risk for triple-negative breast cancer, and should we be screening them with MRI every so often. Yeah, things that are just better able to pick up triple-negative breast cancers, potentially.
0:31:01.4 FC: Yeah, we did some modeling of these genes and they do come in right around 10%, a little bit higher, 10% lifetime risk, and when you consider that triple-negative is a very aggressive former breast Cancer. You know, we use that threshold of 20% from mammography and MRI screening, but for something that we know is gonna cause a more aggressive form, we drop that number to 10%, let's say, and then capturing these guys. There's still a lot of questions to answer in terms of utilization at the clinical level.
0:31:34.4 TS: Because, yeah, if you use a correlative cancer, like ovarian cancer, that triple-negative breast cancers arguably may be not quite as bad as ovarian, but it's pretty bad, and our threshold for action on ovarian seems to be more around 5% to 8%, 10% for sure, risk.
0:31:53.8 Speaker 5: Dr. Couch, did the subjects in these studies include DCIS patients as well as invasive?
0:32:00.9 FC: We did it both ways. We put in the DCIS with the invasive, and then we just did invasive alone, and we didn't see much of a difference. In part because we have very few DCIS in there. We're actually working on another paper on DCIS right now, where we took the DCIS and just looked at them alone, and we have about 5000 of them. And it's actually kind of a slightly interesting story. One of the genes that doesn't show up on this paper that we didn't show at least in this table, was MSH-6. So everybody knows that as a mismatch repair gene, colorectal cancer, ovarian cancer, endometrial, etcetera. We didn't see it, but in the companion paper, the other paper from the Bridges group in Europe, they saw MSH-6 as a moderate risk breast cancer gene. But we didn't see it overall in the carriers, but now that we've been looking at DCIS, we're actually seeing it as a moderate risk gene for DCIS, which is really kind of strange. So, the story is not completely told about MSH-6 at this point in breast cancer, there seems to be something going on there. We look to the ones that we have, and they don't have personal or family history of colorectal, endometrial, or ovarian, they're breast only, with MSH-6 mutations, and there's an increased risk in the DCIS, and, I don't know what that means. Maybe...
0:33:25.6 TS: Yeah, that's really interesting. I don't know. Yeah, I think back, even when I was working with you with our SIMPLEXO group, and we had some interesting MSH-6s, and we went and pulled tumors and ultimately, I don't think we did much with them, maybe I certainly...
0:33:40.2 FC: No we didn't, we never completed that, yeah. But there's something going on there about breast cancer, and MSH-6, that, it's just...
0:33:47.5 TS: Yeah and, we just had a bilateral breast cancer paper, and MSH-6, off the top my head was the only odd significant gene in that as well.
0:33:57.8 FC: Very good.
0:33:58.2 TS: Which is... Jeff Whitesell was the first author on that paper. So yeah, no, there's something there. Do you think that something is a polygenic risk scores? Even how those might... Polygenic risk scores, I don't know, maybe they're even interacting in some fashion with the ER negativity, probably not, because this is more going down HR pathways, to HRD pathways. But yeah, I don't know any thoughts there, about how we'd be using those in the future?
0:34:30.7 FC: I don't know. These mutations are so rare it's quite hard to study it. I think if anybody could do it, it would be you guys. Yeah, PRS likely does modify these risks, so it would be interesting to see what they do in those rarely mutated genes. You guys have published on PRS in the more frequently mutated genes, and we've just had... With Pete Craft on the same topic. So it's different things to what you guys saw. But what about these really rare ones, the BARD1, RAD51C/D, MSH-6. What about those guys? That's why we have to keep doing genetics, because we actually still have stories that we haven't figured out yet.
0:35:09.7 TS: Yeah, and I've been personally thinking lately. You look at a gene like RAD51D, even in ER positive. I mean that's similar to a snip, if not higher. If you think of a snip odds ratio usually being around 1.3 or so, 1.1, 1.2, 1.3. So we probably do need to think about RAD51D carriers in the context of other genomic and clinical and family history variables ultimately to better understand their risk for breast cancer.
0:35:41.1 FC: No, I agree. Yeah, for sure.
0:35:43.5 S5: Dr. Couch, I know your study and the BCAC study included RAD50, but you didn't pull it out into this table three that we're looking at now. And I wondered your thoughts about RAD50, since it...
0:36:02.9 TS: Yeah and I have the...
0:36:04.1 S5: Back and forth.
0:36:04.2 TS: I'll stop share there. Let me reshare, I do have this supplement.
0:36:06.5 FC: So there was the three genes that formed a protein complex, MRE11, RAD50, and NBN. And they were all just completely flat for us, we couldn't see anything there at all. Yeah, there's just so many table... Numbers in this table, it's hard to figure out.
0:36:25.1 TS: Yeah I can't remember where T value I guess, is the...
0:36:29.5 FC: Yeah, so... But anyway, yeah, those three genes are really very flat, they don't do anything for breast cancer. And Bridges saw that, we saw that. Other people have seen it. I think even you guys have reported that previously as well. So I think it's a common story now in these large studies that those 3 genes really are not driving breast cancer risk at all. But there's still...
0:36:58.9 TS: Yeah I mean, anything else on your mind? Of these other genes that... We talked about MSH-6.
0:37:07.4 FC: That was probably the most interesting one. CDH1, of course did pop up, again it's very, very rare. But we did see a few, certainly in the ER positive setting, as you would imagine, think of it as lobular breast cancer. But, we saw a few with it too, and certainly there was an increased risk with CDH1 mutations and interestingly, the majority of our CDH1 mutations carriers in this population-based study, did not have any personal or family history of gastric cancer. So that stories' been spinning...
0:37:40.3 TS: Yeah, that's interesting.
0:37:41.0 FC: After a while, a few years, everybody said, "Well, it's really a gastric gene, and the breast cancer is incidental," it doesn't look like it. It looks like there is a standalone breast cancer risk there that can occur in the absence of gastric, and that just super complicated counseling as a result of that. Because you really have two pathways that you could go down with the mutation and CDH1.
0:38:06.0 TS: Yeah, and that one's always been personally, CDH1, is the bane of probably many people's existence on this call. But that one's been so tricky because there has been the mantra, or maybe there are just some lobular breast cancer families, but gosh, if you think about it, if there's a problem with E-cadherin, which is what the gene does, and so it's a cell adhesion, you'd think that it would translate pretty relevantly to the gastric mucosa as well. It's just so hard to, from a pathway standpoint, try to say that somebody would be safe from gastric cancer if you only see lobular breast in their family.
0:38:46.7 FC: Yeah, that's right.
0:38:49.4 Speaker 6: So I have a question for Dr. Couch. Your study and the BCAC study represent enormous amounts of work, very large sample sizes to be doing this testing. There are some other studies out there, I think the Life Pool study, some other studies that are attempting to do similar kinds of things. There's our work with the multivariate logistic regression model, and in some cases, when you compare all of these different studies to each other, there's a fair amount of agreement, but there's always cases where there's very poor agreement, either the magnitude of the risk is off by one study finds two-fold another study finds eight-fold. NBN, we had a literature that suggested that NBN was associated with breast cancer risk, now we have multiple studies that don't find that.
0:39:52.4 S6: Given that even with these really quite massive studies that are being done, we're having trouble nailing a lot of this down. What's the prospects for ever getting some certainty on some of these genes, both in terms of whether or not they're legitimately associated with risk and what the magnitude of that risk is? 'Cause there's a big difference between two-fold increased risk and five-fold increased risk.
0:40:16.5 FC: Yes, absolutely. I think a lot of it is geographical difference, which you could actually translate into a PRS. So, I think PRS is gonna be the way to answer these questions. We're gonna need massive numbers because some of these mutations are really quite rare as you know, but if you put a PRS with these things, it should help us to shuffle out. There may be people, as Tom was just pointing out an odds ratio of 1.6 might actually be meaningful. You throw a PRS on that, there might be 5 or 10% of those carriers that might be above two. That sort of idea. So, we can't figure that out today. We just don't have the numbers, but as we go forward and accumulate more data, I think we will be able to, so within a couple of years. People like you guys and some of the other larger companies will be able to get to that point. There's still a number of questions out there, like on this table, FANCM has been proposed as a triple negative gene, maybe. It's hard to tell, the data's kind of bounced back and forth a bit.
0:41:16.3 FC: The Bridges guys had FANCC. And if you look at our data in this paper on ER negative, it's 1.6 and the confidence interval's easily bridge 2, and then it's 3.1 and triple negatives. So, maybe they were onto something there, and our numbers just weren't big enough to be confident. What else? MUTYH, that's the one I think we're fairly clear just like NBN, there's really not much going on for breast cancer for MUTYH. Yeah. And then MSH2 and MSH6. We talked about MSH6 already. So, MSH2, just so rare. It's hard to study it in this setting.
0:41:56.3 TS: What do you think about NBN in the context of being of Finnish ancestry? Maybe having some unexplained haplotype or something.
0:42:07.6 FC: Yeah. You're probably familiar with Ken Offit's data for Memorial, right? Ken is seeing this... How do I best describe this? He's seeing like a multi-cancer phenotype. So, when he looks at a whole bunch of families with NBN mutations, he doesn't just see overt breast cancer or ovarian or whatever, but when he looks at all the different cancers together, he can actually see an increased frequency over what would be expected. So, I don't think he's really figured out what's driving that, but it still raises the question of what about a poly-cancer sort of approach, multi-cancer kind of approach for NBN with the risk for any particular disease being very, very low.
0:42:54.9 TS: Yeah. Yeah and then how do you make that clinically actionable?
0:42:58.4 FC: I have no idea.
0:42:58.5 TS: Then it becomes a whole different question, but maybe in the future of early cancer detection, that may play a role.
0:43:06.4 FC: Awesome. Good. Yeah.
0:43:07.1 S5: And so Where we are now with the data that we have, do you have an opinion on whether NBN, MRE11, and RAD50 should be removed from the panels that are currently available given the clinical uncertainty?
0:43:23.7 FC: Well, removing them from the panels is tricky, just technically for you guys to do that, you have to re-develop the whole panel. So, I get that. Well, we certainly shouldn't be attempting to counsel people based on any results that come out of those genes. So does that mean we should mask the results, does that mean that we should provide the results back to the care provider and to the patient, but explained that it doesn't mean anything in terms of breast cancer risk? Again, I'm not a clinician, I'm not a counselor, so I'm not gonna recommend which direction to go with that, but it's very clear, it's not driving breast cancer risk, that's for certain.
0:44:02.1 TS: Yeah. No. That's really interesting. I wanted to pick your brain about ATM and CHEK2. Why do you think they're driving ER positive breast cancer so strongly? I mean, I haven't seen much hypotheses there but...
0:44:18.3 FC: No, there haven't been many. It's a very different biology, if you think of it, right? So we think of the other genes as being homologous recombination repair, DNA double-strand breaks and replication repair. So, BRCA1, BRCA2, the RAD genes, PALB2, they're all involved in homologous recombination. ATM and CHEK2 are not. They're upstream kinases that help to, sensually to either detect the damage in the cell, which is ATM. It's kind of a sensor and it activates many other genes downstream, but not just DNA repair genes. It activates cell cycle. It basically stops the cell cycle, stops the cell from proliferating and replicating, presumably to allow for [0:45:02.6] ____. It also regulates the [0:45:04.0] ____ and a number of other things. CHEK2 the same way. So, we think of them more as cell cycle genes, cell cycle regulators in response to DNA damage.
0:45:14.0 FC: So, the biology is very different. Now, why that translates into only ER positive cancer? I really have no idea. It's pretty clear that the tumors that have those germline mutations do lose the second allele. So, the vast majority of them are homozygous. They're LOH on one allele and a mutation on the other. So they've lost both copies. So, there are certainly driver genes. I just have no idea what the underlying biology would be.
0:45:42.6 TS: Yeah, I think Karen Maxwell was doing some transcriptome analysis on some carriers. I sent her some with Jeff White, so before left City Hope, I'll have to check in on [chuckle] with her, with what's going on there. But I've also... I've often thought, there's a unique opportunity here in ATM and CHEK2 that we're not really taking advantage of, which is, these genes are not high enough risk, I would argue, and many others do, and there's no guidelines to support for bilateral mastectomy. But at the same time, these people are probably uniquely suited for chemo prevention, and if we could, whereas a BRCA1 or 2 carrier still has a high risk for an ER negative breast cancer, you can really mitigate the majority of the breast cancer risk in a ATM or CHEK2 carrier with chemo prevention. So I feel like where we could... We have the opportunity to do a better job with chemo prevention in these carriers and make that the focus instead of trying to counsel why not to remove breast or the benefits of...
0:46:52.0 FC: Yeah, and there's some interesting... In the therapeutic realm, there's some interesting data now with ATR inhibitors in an ATM setting. So, for therapeutics downstream.
0:47:03.1 TS: Yeah, that's a good point. Are there any other genes on here that we...
0:47:08.0 FC: Well there was a question here about PMS2 actually in the chat there. We actually were reluctant to show it, to be quite honest, because as you guys know, there's a lot of phenocopies of PMS2, there's a lot of pseudo-genes in the genome, and it can be quite tricky to actually measure this properly. So we really didn't wanna make a big deal out of that one, and it's very rare in our setting anyway. So I would say we probably don't have the final story on PMS2 either. Again, it'll be groups like you guys who do a very specific assay to detect that those long... The larger re-arrangements can occur, but also in the very specific, correct, location and not in any of the pseudo-genes. So that's the other one that stands out to me and then, yeah, that's...
0:48:08.7 TS: What if...
0:48:10.9 FC: And P10 is sitting there, but again, we had zero in some parts.
0:48:15.7 TS: Yeah, it's just so rare.
0:48:17.3 FC: The categories there are just so rare, but it's clearly a substantially increased risk of breast cancer.
0:48:23.2 TS: Yeah, and only a few SDK11s too.
0:48:26.2 FC: Yeah, so rare.
0:48:28.5 TS: And MUTYH, did you parse out the common founders from frameshift? I know we did that in the SIMPLEXO data, and there was a signal, but it was only with like five cases, and it didn't didn't have the founders for breast cancer. I don't know if you've ever really looked at that, I just always thought that was interesting.
0:48:49.8 FC: Actually, no, we just kinda took it at phase value here, and just record what was out.
0:48:54.0 TS: Because 80% of these would theoretically be the founder, I would think. One or two, yeah interesting. Any questions from the audience? It's been a great discussion.
0:49:11.4 FC: Do you wanna switch over to that Bridges paper or...
0:49:14.4 TS: Yeah, we can show that. Just so people are familiar with it. We don't have to, I have a few minutes left, but yeah, good to talk about. Here's the... It's gonna look strange again. Let me redo that.
0:49:29.4 FC: Yeah, so their study was almost twice the size of ours. It's like 60,000 cases, 60,000 controls. They did about 10 or 12,000 of those individuals from Malaysia actually, so they have an interesting mix of populations in there too. They more or less found what we found the same gene, the five main genes, and then they also saw BARD1, RAD51C, and RAD51D. You can see it right there in the abstract. So very similar results but the two extra genes that they've found that they suggested might have something going on was MSH-6 and FANCC. And those are the two we've commented on already.
0:50:07.0 TS: Those are here, yeah. That's our odds ratio.
0:50:10.4 FC: Yeah.
0:50:13.5 TS: Yeah, all studies.
0:50:13.6 FC: Definitely opens the possibility that there's something going on there. But hard to study just because they're rare in this setting. The other thing they did in this paper is they studied missense mutations a lot. We more or less restricted, in our paper, we restricted to very clear clinically pathogenic variants in the various genes. They did sort of an exploratory study of missense variants as well, which most of us, I think would call predominantly VUS. And they did some modeling of that, trying to show that there was an increased risk associated with certain missenses and things like that, which makes sense. It wasn't anything particularly unusual about what they reported. Yeah, there's there lifetime research...
0:51:02.2 TS: Yeah and, which look very similar to your data.
0:51:06.8 FC: Yep, that's right.
0:51:06.9 TS: And how did you end up publishing at the same time in the same journal. You knew this was going on. Did the journal sync you up a little bit?
0:51:14.9 FC: We actually didn't know that they had submitted to New England Journal, we knew they had completed their study and they were submitting somewhere, and vice versa for them. And it just turns out that we both happened to submit to the same journal, and the editors pulled them together in time. They were actually, I believe, a little bit ahead of us, so they were at the journal maybe a month or two before us, but the editors decide to sort of hold them and publish them back to back, which makes a lot of sense.
0:51:41.1 TS: Wow, so you had no idea, even... Until it was published that they were coming out the same...
0:51:44.6 FC: No, I didn't know. Yeah.
0:51:45.2 TS: Oh, wow.
0:51:45.2 FC: And the editorial from Steven Narod, we had no idea who was doing that either.
0:51:50.7 TS: Oh wow. Yeah, I thought his editorial was good. Again, I think, I was surprised that really nobody in these papers or the editorial got into PRS modification since it is... Has, definitely emerging as a strong helper in this whole quest to figure out what someone's risk is.
0:52:11.9 FC: Yeah, yeah. I think the way it works in academics is that people just keep it to the smallest simplest study or story. Both BCAC and ourselves had PRS papers in the works already at that time. So we felt that any discussion of the PRS belonged in those PRS papers and so we just didn't get in here. This was really just about establishing risks in the general population, just keep it focused, but I fully agree with you. I think the PRS is really currently the big deal, and for some people, maybe the next big deal in genetics is then how to apply that, how to make sure that more people get a PRS screen and a score so that we can modify... Individualize their risk is, if you will.
0:52:58.8 TS: Yeah, yeah. Just checking the chat for any other... "Please show the title of this paper." Oh, thanks, Shelly. You put it in the citation. Any other questions? We have a few more minutes. I just wanna do around the horn. This was great discussion.
0:53:19.1 Susana: I do have a question for Dr. Couch. My name is Susana, I'm with the genetic counseling team here at Myriad and I am in charge of reviewing a lot of the literature that has to do with a RAD51 C and D, and BARD1, so I've read a lot of your papers, and we talk a lot about some of the cancer risk estimates that their papers come up with when comparing two controls in ExAC and those type of databases, and what are your thoughts about that? That it's not like a matched control type of situation?
0:53:57.6 FC: Right, in these two papers, it was a match control, so it was fairly clean here, but in previous papers you're right, we've compared against things like Nomad, ExAC, that sort of stuff. What we've really done, we're pretty careful with that, we go through and we clean all the data ourselves, we really evaluate to make sure it's not just a calling mistake or something like that, and there's limitations to what you can do with that public domain data. Absolutely, but we do try to go through it and clean it a bit, and then the other thing is that we take the control data and we use it for estimation of risks in ovarian first, and of course, we know that these are high risk ovarian genes, RAD51C and D, and so when we saw the high risk for ovarian up around 8 or 10 odds ratio, then we felt that the controls were behaving themselves, and so then we could apply the same controls in this case in the setting of breast cancer.
0:54:49.8 Susana: I see.
0:54:50.6 FC: So we almost kind of controlled it in the background a little bit to make more certain that we could use it in that way, but you have to be cautious with this, you can't just grab the data and just throw it into a model, you have to be way more thoughtful about it than that.
0:55:06.5 Susana: Go it. So it's just a matter of resources and just how rare this...
0:55:10.0 FC: Spending time with the data and really understanding it and cleaning it, and I think we would all admit that there's limitations to that sort of public domain data set, there are maybe errors in it, we just hope that they're not big enough errors that they really dramatically change results. We don't think they do. They're more or less in line with these big population-based studies, which had matched controls in them. So I think in the end, it was an okay thing to do.
0:55:37.0 TS: Yeah, some of the trickier things in the controlled data sets that we've worked with in the past have been... And we as in everyone, I guess, but also me and you have been... Yeah, when you get into really TP53 in particular, there's probably a lot of muddling by clonal hematopoiesis.
0:55:53.9 FC: Yeah absolutely.
0:55:54.8 TS: In some of these genes in these databases.
0:55:58.0 FC: Yeah, it can get very messy. You gotta be super careful. In our paper, we didn't talk about it here, but we restricted our calling of variants in TP53 and NF1 in our paper here because we were worried about these clonal hematopoiesis events, so we made it more conservative as to what we considered a pathogenic or a deleterious mutation one of those genes. Just straight type...
0:56:22.4 TS: What do you think about NF1 as a whole? We should have talked about that, but... 'Cause that one's on the...
0:56:27.7 FC: Because you talk to Kate Nathanson at Penn and she's quite expert in NF1 stuff and neurofibromatosis in particular, broad spectrum of phenotypes, obviously all the way from a cafe au lait to lots of tumors and... I don't know, it's a tricky one. She always says that you cannot have an NF1 mutation without some sort of neurofibromatosis phenotype.
0:56:50.1 TS: Something, yeah.
0:56:51.1 FC: I don't know.
0:56:51.5 TS: It's been my kind of experience too, that's usually been my mantra through the city, of course, to some extent that I've really been impressed yet with someone that had zero findings of NF1 that really had a true NF1 mutation that wasn't maybe clonal hematopoiesis or something.
0:57:06.1 FC: Exactly, right. So we've been sort of... We didn't talk a lot about NF1 in our paper. We kind of stayed quiet in the discussion about it because we didn't... We're being more and more cautious about that particular gene because of, number one, we don't know the phenotypes, we don't know if they have neurofibromatosis because it's just a large collection from other sources. And number two, because of the clonal hematopoiesis possibilities. So there's this big question mark, I think, hanging over that gene and breast cancer. It's a complicated one.
0:57:36.2 TS: Yeah, definitely. Well, we're at time. I cannot thank you enough...
0:57:40.9 Shelly: Jay, can you answer one more chat question?
0:57:43.7 TS: Yes.
0:57:44.0 Shelly: There was a question about Myriad's plans to add PRS for mutation carriers other than CHEK2?
0:57:50.3 TS: Oh yeah, so we have a paper, maybe just, I think it just came out actually on ATM, clearly, if you take anything away, I think from some of this conversation, one of the big things is we probably need more data to fine-tune risk, and some of that is just gonna come from polygenic risk scores. You're seeing a lot of papers now starting to come out on carrier modification, I think you had the first one, Fergus, on the CHEK2 1100delC paper years ago in Genetics and Medicine.
0:58:40.2 FC: That's right.
0:58:40.6 TS: Probably the first paper maybe in the space, so... It's definitely gonna be part and we have to sort out how we're gonna bring it out and kind of present to the world, but yeah, ultimately I think it's gonna keep expanding, and if I have my way, it'll be in... We'll hopefully be able to solve the MSH-6 questions with these kind of more complex models and everything, hopefully over time, it's gonna take time. Yeah exactly.
0:59:03.8 FC: It's all about volume of data, it's all about volume of data 'cause massive numbers needed to do this. Yeah.
0:59:08.2 TS: Yeah, yeah. So really exciting. So thank you so much, Fergus, for coming on. Can't thank you enough.
0:59:12.5 FC: Thank you.
0:59:12.5 TS: And... Yeah, hopefully everyone learned something. I know I did. So it was good. [chuckle] Alright, have a great day everyone.
0:59:21.5 FC: Okay, thanks, bye.
0:59:21.6 TS: Thank you. Bye.