HR-Positive Metastatic Breast Cancer — An Interview with Dr Seth Wander on Optimizing Biomarker Assessment and Related Treatment Decision-Making

Welcome to oncology today, optimizing biomarker assessment and related treatment decision making for patients with hormone receptor positive metastatic breast cancer. This is medical oncologist Dr. Nioh Love, and I met with Dr. Seth Wander from the Massachusetts General Hospital in Boston to discuss the role of biomarker assessment in the management of hormone receptor positive disease. To begin, I asked Dr. Wander to comment on how he approaches the use of tissue and liquid biopsy in patients with metastatic breast cancer. I'm going to answer this two ways. I'm going to certainly play devil's advocate on both sides. In my own practice, if somebody is diagnosed with metastatic disease, they virtually always have a biopsy proving it at the time of metastatic diagnosis. So we already have that tissue. So I will typically send that tissue for sequencing because you saw there were some discordant cases, for example, where picturese A's picked up on solid but not on liquid. I will do a concurrent liquid biopsy at baseline to kind of set the stage for where we're starting with all these genes. Typically for progression, I tend to prefer liquid biopsy for all the reasons that we talked about. We may get a solid biopsy if, for example, the patient was hurt to zero and I'm trying to figure out if they're eligible in the future for TDXT. We may get a solid biopsy if there's like oligo progression in one area and not others to figure out if the receptor status has changed. And if I get another biopsy, I'll send it for sequencing. I think the bigger question that you're alluding to is should we be doing repeat solid biopsies because of the discordance that can occur. And I think, I think if your suspicion is high enough or you're really looking for other treatment options, it's not unreasonable. But of course, it's putting the patient through more invasive procedures. I like to think, and I do believe this, that the technology is getting better and better and better. The threshold of detection for the liquid assays is getting lower and lower and lower. And I hope that over time, the discordance that we're seeing, which I showed you a few different papers, it was on the order of anywhere from 10 to 30%. That actually, that number is going to shrink and shrink over time because the liquid biopsies are actually moving at the speed of light. So I think the general consensus of if you have to choose liquid as sort of safer and better and easier is true. But I don't think we should assume that every single mutation will always be caught on liquid biopsy. That's the way that I would kind of hedge on that. Yeah, that's sort of news to me. Another thing you mentioned was, you know, I always hear the word epigenetic. And to be honest, I don't know if the only one who doesn't really understand sort of what that means biologically. But what does it mean? Are there markers that sort of evaluate that? Yeah, it's really, it's a, I think a frontier area, not just for breast cancer, but many different tumor types. The way that I think of epigenetics is the architecture of the chromatin. So remember, most of the DNA, the vast majority of the DNA in the chromosomes is non-coding. There's no genes in there. But it impacts the shape that the chromosomes take and it's packed so tightly together that all of that non-coding DNA shapes the architecture and the ability of the transcriptional machinery to get into the gene right to turn it on. So epigenetics refers to methylation changes on the DNA that opens it or closes it and allows the machinery to get into the genes to turn them on or blocks the machinery and turns it off. And what we've started to appreciate is that a ton of oncogenic events, turning this gene on or turning that gene off, may not be detected by conventional sequencing, which is looking for a hit in the DNA code. It might be some change in the methylation markers on the DNA around it that changes the architecture and blocks the transcription or encourages the transcription of the genes. And so some newer technology, and I would love to come back and talk about it, we can actually start to profile the methylation changes across the genome by the blood. And that will then tell you whether gene A, RB, P10, pick your gene is turning on or turning off. And that would be totally undetectable, for example, by targeted sequencing of different genes. So I call this the sort of, we used to call it the dark matter of CDK resistance. When I was a postdoc and we were working in the lab, there'd be 25 to 33% of patients. We could not find anything on solid biopsy, on liquid biopsy. There was no RB, there was no AKT, there was no RAS. But the patient was resistant to CDK. So we hypothesized that there's some other events, whether it's the DNA architecture, the methylation changing, expression of the gene, whether it's post translational changes to the protein, whether it's happening, you know, through some other pathway that we don't know. And so I think we're actually going to start to uncover a lot of that dark matter of resistance, looking at RNA-based changes, looking at methylation, epigenetic changes. So really cool technology coming along to help us do that with the blood biopsy. Another thought I had as you're going through that talk was, I don't have to put it exactly other than I was thinking a lot about prostate cancer and PSA and how helpful that is clinically. And what I saw in your talk in a number of places was what looks to be the opportunity to really start to look at how therapy affects biomarkers again, something kind of new, at least to me. But this TK level thing sounded really interesting. Can you just kind of go back through that and talk about where you see that heading? It seems like it would have an important clinical role. I agree. We're jealous of our friends in geoncology because I think you would agree with me and you know more about all the different cancer types than anybody that I know. And so I'm curious how you're going to respond to this. I think PSA is the best easily accessible biomarker in any tumor. Var none. I can't think of another example. We're 99% of the patients express it right and it's a dynamic change. So we're jealous of that in breast cancer, right? We've got CA 153 and CA 2729. But they're nowhere near sensitive or specific as PSA. So we have an unmet need as do many other tumor types. And the TK test, the thymidine kinase test, which we've worked with them to help develop this is really cool. And I'll show you a case, a couple of cases in a few minutes on that. So the way I think about TK is it's almost meal like K I 67, right, which is a good metric of sort of how the tumor cells are dividing. But you can test it in the blood. So the higher the thymidine kinase level in the blood, the more cell cycle you have, the more G1 to S, you know, transition you have. So think about a drug like a CDK inhibitor, which acts by blocking cells in G1. They can't divide. You should expect your TK, your thymidine kinase level to plummet if you're stopping cell division, right? And so TK is a beautiful test for any kind of drug where you're actually measuring the amount of tumor cell division. And you want to do it in a quick non invasive way. And the test comes back in a couple of days, like it's not a three week or six weeks send out. I'll show you some data that we've been developing from different patients. But to me, there's at least two ways to think about a TK test, for example, an ER positive breast cancer. And we talked about one of them today. You can develop with three time points, baseline, halfway on cycle one, and off drug at the beginning of cycle to the pattern. If there's no drop in cell division, no drop in TK, those patients are not responding to the CDK. And they're likely to progress earlier. You can say this patient's going to get an inferior outcome most likely compared to somebody who has a deep drop. In vomiting kinase. The rebounders are interesting. The rebounders are saying they have some response to CDK. But when you're on that off week, they're starting to already, the tumor cells are starting to divide. And so they're somewhere in between. So the first thing you can do at the TK level is look at the patient within a month and say, I can pretty accurately predict where you're going to fall on this PFS curve for Mona Lisa 2 or Paloma 2 or pick your trial. The second thing that's interesting, which we didn't talk about yet, but make sense. Let's say like Serena 6, now the patient's doing well. They're on AICDK. You could check a TK every two and three months. Once it starts rising, once that TK level starts to rise, it's sort of like CTDNA, the ESR1 rising. But we're not just looking at ESR1. We're looking at total tumor bulk. And my colleague Cynthia Ma at WashU is doing great work on this as well. She has beautiful data from patients who they have a TK level that starts to rise months in advance of a scan, right, showing any kind of clinical progression. So I could imagine a future state where you take the TK level to predict kind of who's going to do well on a CDK inhibitor from the beginning. We've been developing clinical trials to think about how to escalate therapy in patients who don't get a good drop. And then you're using it to monitor. And once you start to see the rise, now we're going to check CTDNA. See is there ESR1? Is there AKT? Is there P10? Is there something we can target when the TK results? So it's like a global tumor marker that I think you're right. PSA is the best kind of analogy to that because it's kind of a universal thing, right? It should be reliable if the cancer cells are dividing. So you know, kind of a follow up to that would be is this specific to endocrine therapy of breast cancer? Is it just basically a good tumor marker for any therapy breast or otherwise, even outside of breast? I think it could be a good tumor marker outside of breast because again, that G1 to S checkpoint is not unique to breast cancer. Anytime you have tumors that are dividing, you're going to have thymidine kinase levels in the blood. Now, as you know, I'm sort of a one trick pony. Like I only do my work in breast cancer. So I'm not as familiar with what kind of data has been collected. Immediately, you think about other hormone dependent, either gynecologic or prostate. But I agree with you. I think even a non-hormonally dependent tumor where you have drugs that are blocking cell division, it would work. The thing that probably wouldn't work is that rebound pattern. The reason the rebound thing works is because you're three weeks on one week off. You're getting a look at what the tumors do. during the off week. If you don't have a three on one off, you could still use TK just as a measurement of tumor division, but you can't generate that unique profile. So that's I think unique to the CDK, but you could still follow TK just like you would follow CTDNA. I think both CTDNA globally, like the amount of CTDNA in the blood and TK are future biomarkers to look at the activity of a cancer, right? And if you just set aside the presence or absence of ESR1 or the presence of absence of PTEN or AKT, just the amount of CTDNA, if there's more CTDNA in the blood, that's a bad sign, right? That's a sign that there's more cancer cells. I happen to think probably the TK is a little more sensitive than that, but it's the same idea and it's the same idea as the PSA, a universal kind of biomarker that could be used to find subclinical progression before the patient symptomatic, before there's millions of cancer cells growing on the skin. So a couple specific questions that came out of your really great slides are looking at NGS. So I mean, I know it's not the the emphasis here, but I have to ask you just looking at the algorithms you had there. In terms of looking for BRCA, if you have a negative NGS, do you need to do germline? It's a technically yes, because those NGS tests are not specifically designed for germline assessment. The implications in my experience are if it's not detected on NGS, the likelihood of having a germline mutation is exceedingly low. And typically I also tell the fellows, if you have NGS and you see a BRCA, where the elelic frequency is on the order of 30 to 60%, it's a really good chance that they have germline, right? Because it's probably there and about half the cells. But that's, but that we can't use that. We still have to confirm it with the sort of clear approved germline, you know, germline test. But you have somatic BRCA. So if I see a BRCA with an elelic fraction of 1% or less than 1%, that's almost definitely not germline. And it's probably acquired in the cancer. And we have data coming along that acquired, you know, HRD changes like acquired BRCA. They can still be sensitive to PARP inhibitors as well. We've had some trials looking at that. So, but yes, I take your point and I agree that typically there's a pretty good correlation on somatic sequencing for what you might expect with like a germline change. A couple of other things that sort of popped up on your chart there that I thought was interesting. Have you ever had a patient have with an FGFR alteration or a ret, a fusion that you treated with target of therapy? I didn't think about that. I haven't. And I've treated thousands of patients. One of these days I'm going to come back to you and say, yes, I found N-TREC. I found my N-TREC patient or I found the REB patient. I have had patients with very high TMB. I have had a couple of patients with high MSI. I have had FGFR patients. Although again, sometimes the FGFR is an amplification event. Sometimes it's a point mutation. We have some nice data that we're generating looking at resistance to CDK inhibitors and PI3K pathway inhibitors with FGFR. I think that's actually really underappreciated target in ER positive breast cancer. I have used immune therapy in those cases sometimes where the TMB is high or the MSI is high. I would love to use the N-TREC inhibitor if I could find it. I hear about it in tumor board. You know, once in a while I'll do a bunch of tumor boards and I'll hear about it. It's tough though because, and I'm sure many of your other colleagues who you've done these sessions with will say this, I find TMB not to be the best predictor of response to like immune therapy. It's one of those things where, you know, if I have a patient who's got a TMB of 15 or 20, it's a little bit high. The track record in ER positive breast cancer really isn't great if you look across the whole population. I have a couple of patients where the TMB is like in the hundreds. I feel a little bit more excited about using it if it's really high. If the cutoff is 10 and I have a patient who's in like the 80 to 100 plus range, I'm saying, okay, this patient's way off the chart like standard deviation. But I have used some of those tumor agnostic. They hurt two kinase blockers and somebody with an acquired her to mutation. Those are acquired changes that provoke resistance to endocrine and CDK. So I sort of hesitate to bring this up because you already talked about so many trials, but there was one trial you didn't talk about because you were focused on metastatic disease. But I want to ask you about it in terms of a lot of the concepts that you just talked about. And of course that's the Ladeira study where you didn't have ESR1 mutations and yet you have a SIRD. I don't know whether it was cause of efficacy or tolerability or whatever, but I mean I know we're not here to talk about Ladeira necessarily, but I do think it kind of relates, you know, sort of getting back to the kind of original SIRD papers comparing SIRDs to other therapy, where people were looking for it to be better. And I was sitting there going, well, it looks like it says good. So how do you explain Ladeira where they don't have ESR1 mutations and yet did better? Yeah, it's a great, great question. I'm going to answer this a couple of different ways. So Ladeira is an interesting study, of course. Gira Deskrant could be a tough conceptual drug for our fellows and for our community practice doctors. And here's why think about, for example, Pembrolysiumab in triple negative breast cancer. Pembrolysiumab to get it approved in metastatic disease, you have to be PDL1 positive, but in adjuvant, neo adjuvant, keynote, you don't need to test PDL1. Same drug, same disease in the early stage setting, there's no biomarker. In the metastatic setting, there is a biomarker. That's conceptually a little difficult. Imagine Gira Deskrant in a relatively near future state where EVERA, because of the data you and I just reviewed, it gets an ESR1, you need ESR1 to give Gira Deskrant Everalimus in the metastatic setting. But Ladeira, per the point you just made, there was no stratification for ESR1. All comers did very well. So you could have a situation where if you have early stage disease, you could give Gira Deskrant to kind of anybody, but in metastatic disease, you might only be able to give it for you or someone. I don't know if that's the case. I'm not on those panels. I'm just, I guess, I'm guessing. Now, why would it work? I'm going to be, it's cynical is the wrong word. I think if Ladeira had been sort of weekly positive or negative, but the ESR1 subset did well, everybody would have said, "Oh, that's obvious. Those drugs only work with ESR1." But since Ladeira worked, everybody says, "Oh, that's obvious. It's preventing the emergence of ESR1." And that's why it works, right? It's all like 2020 hindsight. My feeling about Ladeira is the combination of a few different things. I think it is a better estrogen blocker for lack of a better word or degrader than Tomoxifen or an AI. To your point, I think these drugs are better tolerated, so there's more treatment persistence. We saw that in Serena 6. And I think it's preventing the most common bypass pathway, which is the acquisition of ESR1. So between those three things, it's a better drug. It's better tolerated, so people stay on it longer. And it's preventing the key mechanism of resistance. Now, one of the things I worry about, and this is not a reason not to use the drug, this just means we have to do our work in the academic setting, what comes out of that? Who progresses on Ladeira and what does their tumor look like? Almost certainly it won't be ESR1, right? Because you're blocking those clones. It's going to be the stuff you were just asking about. Her two acquired mutations, RAS, pathway alterations, AKTP10, ER loss, probably looking more like a triple negative. We need to see, that doesn't mean we shouldn't use it because you're hopefully curing more patients. When you have progression, the phenotype of the tumor changes. The best example of that is if you and I have been talking five to ten years ago, and you asked me before CDK inhibitors, what's the average time on full vestrine to the second line metastatic? I would say, you know, 12 months, like 10 months. If you ask me now, two months, right? Three months, we did that by using CDK inhibitors. We changed the genomic structure of the tumors, and we actually made them more virulent for lack of a better word. No, I remember when those papers came out, and I was like, I thought full vestrine was a good drug, but not in that situation, the two month PFS. One other sort of quick thing, I'm always having a problem trying to figure out when we're writing stuff with term to use related to AKTP alterations. And I noticed that one of your slides you had the term PAM pathway, PAM pathway, PAM. What does that stand for? That's P-I-3K AKT M-TOR pathway. So that's basically saying, "Oh, okay, perfect, good, I like it." I think about that, Neil, because of the get a drug, the Gadada Lissib is a P-I-3K AKT M-TOR, so they call it a PAM inhibitor. Huh, wow, that's good, I like it. Okay, another question is, you know, this really cool slide there, either, it was incredible. I know you sort of made it up where you see the mutation and status of a patient over time, and I realized that you created this as sort of an artwork thing, but it's just so informative to be able to look at this, and I'm curious whether you think in the future, this is the kind of thing we're going to see on a lab result. Yes, this is actually from a Garden Test. What you see on the Garden Report is if you scroll down and we can do this as an exercise one day, for each gene, they show you, if you have multiple Garden Test, they'll show you at each time point what the level of the gene was, but then you can look at the whole picture, like the whole spectrum of mutations, and what you see here is, as time goes on, and I put the part at the bottom, which is diagnosis, resistance, and stay antiestrogens, when you treat with the drug, you see that the patient gets more complex mutations and the new colors that come up is each one of different mutations. Each one of those is different and I'm just tagging which one if you look on the key, it tells you which mutation is which color. So the green, the light green is ESR1, the dark blue is picked RCA. The dark blue you notice is there the whole time and the frequency which is the thickness of the color is about the same for the dark blue, the picked RCA. But the light green didn't exist at time zero, diagnosis, but now is a reasonable frequency because that's acquired, right? So trunkal is the ones that are present at baseline acquired are the new colors that are present on the far right. I'm hesitant to bring this up, I'll send this to you because you could take this to a whole other level. We did a paper that's in review, it's about to be published hopefully and I think it's going to be an NPJ precision oncology where we looked at patients who had at least 10 different variants in ESR1. So they had 10 different types of ESR1 mutations and you should see these plots in those patients where they may have dozens of different mutations at one time, right? And the software generates this plot with all different colors and all different things and you can see when you treat the patient and you retest after treatment with certain drugs, some of the mutations collapse and disappear, some of them increase and we were simply trying to demonstrate the complexity of the clonal architecture on and off anti-estrogens and on and off chemo and on and off ADCs. So this one has three genes to begin with and maybe 10 genes later. Imagine a patient who has a report with 30 genes or 50 genes, right? Trying to decode the complexity of that is very interesting. So a couple other things and then again we'll get to your cases but I agree with you, that presentation is San Antonio from Serena where you saw the allelic fraction drop down was very impressive and again it seems like it's something that has clinical relevance now. I mean I don't know do you look at the, do you get serial assays and people you already are treating with the, assert to see whether it's dropping or not. It kind of looks like a tumor marker. Oh, I completely agree and I actually think to the point that you brought up earlier, if we were to test total burden of CTDNA, allelic fraction of ESR1 or thymidine kinase level, they would all, if I showed you a plot of all three of those things, they would all look exactly the same because they're all measuring the same outcome which is sort of dynamic on treatment changes in tumor burden. We've started to recheck CTDNA on treatment. I think once we have a kind of a formal FDA approval for Serena 6, it'll be a bit easier to kind of get that testing done all the time. Right now, you worry a little bit about will it get reimbursed at the same rate, do we have to work with the company to sort of get things paid if the patient's not obviously progressing. I think it's going to turn out to be a great biomarker, any of those things but certainly dynamic ESR1 changes. To me, everybody looks at that plot and says, wow, look, it's 100% median reduction. So cool, I look at that plot and say, what about the five or 10 patients who didn't drop to zero? What's going on with them? We have some ideas as to what might happen to them, but I've talked to my friend Dr. Bedard and the team at Serena 6 to say, well, let's do a deep dive on the seven or eight patients who didn't get to zero and try to figure out what the resistance mechanisms are because you'll learn a lot about who maybe shouldn't get an oral surgery. Another thing I was thinking about is we've been doing a series now on CTDNA and oncology in multiple tumors, including breast. And I kind of put over on this side those type of assays, like Signetara, and assays like ESR1, but as you work one through this, they kind of are coming together in my own mind. So it's almost like it seems like it's sort of the same concept, just much more targeted to one spit, like you were saying, like PSA and prostate. So do you see a role for quote, you know, an eight, well, let's just say an assay like Signetara and breast cancer? Yes, definitely. And we've got very exciting data for breast cancer. The problem is, I think you'll agree with those assays, at least in breast cancer, is we know that prognostically it's very powerful. Like if your Signetara is positive in the early stage setting, you're likely to have a recurrence, we don't know what to do about it, like what is the action ability? You could imagine a situation in the not too distant future where we're monitoring for MRDs with something like Signetara. If flips positive, we then immediately test for ESR1. If it's positive, go to an oral surgery, just like, you know, with the Lidera, you know, through the Lidera or Ember 4 or elegant, the idea being we're monitoring the patient, we've detect tumor DNA, we figure out if it's driven by ESR1 and we switch them to an oral surgery without anything on a scan, right, with no abnormalities detectable yet on imaging. That to me would be a great use of this technology moving earlier in time for patients. So one other question, you were talking about these new AKT inhibitors, this one without a name, RLY2606 as an example, and having just done a webinar last night on hyperglycemia with AKT peak 3 agents that was pretty complicated, I was just curious, one thing he didn't show, I don't think, was toxicity. You kind of wonder, do you see less toxicity since it's targeting the mutation as opposed to the wild type? Yeah, great question. And some of them are getting names. It's hard to keep up with exactly which one is getting which name because now they're moving into phase 3. There's a bunch of these next generation PI3K mutant specific and now AKT mutant specific drugs. And yes, I didn't show the toxicity just in the interest of time because we were doing biomarkers, but yes, the hyperglycemia rates are dramatically lower, dramatically lower, because the hyperglycemia is the wild type peak 3CA side effect using the first generation agents. So some of these studies and drugs were even tested in patients with diabetes, right, and they were doing okay. So the newer drugs have a much better therapeutic window, right, like much less off target wild type. But the downside maybe is that because it's so specific to the mutation, you're going to be able to get secondary mutations in AKT, P10 and maybe even other picked 3CA mutations that could cause resistance. We don't know that for sure yet, we're working on it. That's the flip side of that is the gedatalisive data where it's not mutant specific. It goes across PI3K, AKT and M2R, but because it's IV, you have less hepatic metabolism and some of the side effects. It actually also has lower hyperglycemia rates, even though it's not a mutant specific isoform. The downside is it's three weeks on IV one week off, so we have to kind of work our heads around that in the clinic. So the newer drugs, whether they're the IV formulation or the oral mutant specific to your point, are going to have a lot better toxicity window and we're going to be able to give these drugs to a lot more people. So one final sort of global question that we've talked a lot about over the last few years and I just kind of rethinking is the clinical situation. We have both, I guess you say a PAM alteration, is that what you say nowadays? Sure. And ESR1, both, which is not rare at all, it does happen. And which way are you going to go? And we've talked about, we've done surveys trying to figure out what people do clinically, but sort of relevant to that. Again, last night when we were talking about AKT and all, and we were going back to the original studies, I was looking at how it was going, wow, those hazard rates are really impressive, particularly when you treat earlier, like with the end of a list of where you started first line. And then I was thinking about what we see with SIRDS and ESR1 mutant. So I know it's indirect comparison, but just kind of curious because I know there are a lot of factors people consider in making that decision. And of course, when you have both, I don't know if that's different, but you have an inherent feeling that whether or not one has more efficacy than the other. Yes, it is de-key question right now in clinic and we're actually generating some data on this with institutional experience and real world data. And I think that'll complement the surveys that your team has done very nicely to try to answer that question, which doesn't matter which one you give first. Can you give them in sequence? I'll answer this a couple of different ways. First, I think it depends a lot on the patient, right? What are their medical comorbidities? How bad is the disease? What rate is the progression? I wouldn't use the alleleic fraction to make the decision because almost always the pick 3CAA is going to be a higher alleleic fraction. And some might argue that the lower alleleic fraction is the new thing. That's the thing that's causing the progression. So you could argue it both ways. I tend to prefer doublet therapy for patients who had a less robust control on their frontline drugs for patients who have more disease and you need to be a little bit more aggressive. I tend to prefer the oral-served monotherapy for patients with maybe less aggressive progression. But patients who had a much longer duration on their frontline therapy, the side effects are obviously much easier and the monitoring is much easier on the oral-served than to put them on the Pam pathway inhibitor. The last thing I always say about this is it's a huge problem right now, but it's temporary because you have these next Gen PI 3K inhibitors coming. They're going to be combined with next Gen A and T-Estrigen. So we're going to have perfect combos where we have a good oral cert or a or a surm or a serian or something combined with a good next generation p_m_p_i_3_k_a_k_t_ inhibitor so we have this moment where this is going to be a big issue or trying to write some papers and look at this and then five years from now you and i are going to be talking to like remember when we had to choose between these things that was weird right like that you know we that was like a crazy time uh. when we had to make those those hard choices so just a couple other quick questions i i just forgot i was going to ask you about where we get to the cases a couple of your publications of which they're sure are a lot you have like a publication house up there but anyhow a couple of them i wanted to ask you about one is in a journal i'm just curious espw a deep learning enabled tool for precision based use of endocrin therapy in resource limited settings this is like a_i_ for what we're talking about here a colleague of mine works in a_i_ at m_g_h_ and um. asked for some assistance on the breast cancer side that's a really cool thing what they're doing is they have digitized pathology for example from a Haiti where the patient gets biopsy and it's very hard for them to get er you know p_r_ her two testing done in in resource limited setting so they've created an a_i_ algorithm where they've trained based on thousands and thousands of samples they can take a snapshot a picture of a slide from a tumor no stain no e_r_ stain no p_r_ stain and the a_i_ algorithm can tell you with pretty high confidence whether it's e_r_ positive and so you don't give the you know the wrong to moxifin or or nastras all or whatever to a patient who's the our negative and you are more confident that you're giving it to the right so i was sort of a consultant on that paper just to kind of from the breast cancer side help them understand the right questions but the the real the the the other authors did the a_i_ work on that for us hopefully that'll be published in a pure e-journal soon that's interesting do you see a_i_ being a very assistance in this type of realm in terms of you know taken met a profiles of uh. biomarkers and providing some kind of input to in oncologist i think it will i think it will be helpful especially as we have more and more genes and we have more and more types of testing so imagine a situation where you have multiple targets you're trying to decode across different platforms like t_k_ or like r_n_a_ and like d_n_a_ and you can you can create an algorithm to help you predict you know which patient would do better on this drug versus that drug uh. i i think it's i think we're only scratching the surfaces to how those decision support tools might be deployed particularly in in in practices where you know with with all of your other colleagues you the speed with which it's moving with all these new approvals and g_i_ and thoracic in myeloma and melanoma right in lymphoma you have to have some support to to read the clinical data pathology data the genomic data and to understand these are my choices and these are the outcomes right with the top choice or the middle choice or the bottom choice so i i think there's there's great potential for for this sort of technology with complex data input you know i think about all the stuff you just went through and think about a general medical oncology's got to you know figure out what to do about blot knee o'agents therapy of bladder cancer et cetera real challenge one other paper caught my attention i didn't get a chance to read it but i'm real curious about it good but several several papers actually same topic comparative efficacy of first versus second-line c_d_k_ inhibition yeah i have a great um. student is rotating with us and she's very good at doing some of these meta analyses we were trying to take a look at kind of the sonia approach right that do you have to use a c_d_k_ inhibitor in the front line versus later setting and we're trying to do meta analyses of of how of how patients do and i think i think the point here is we as a as a community of drifted toward the idea of deploying c_d_k_ inhibitors in the front line setting across the board there are probably situations where patients for example who maybe a higher risk of toxicity or may have less aggressive more indolent bone only disease could get away with going with endocrine therapy alone um. neal one of the things that you'll be interested here is we've proposed a trial we're trying to get a funded where we use the t_k_ test and if you have deep suppression of the t_k_ you pull back the c_d_k_ inhibitor because maybe those are highly endocrine sensitive patients who don't need the financial toxicity in the monitoring for c_d_k_ so that meta analysis was meant to kind of examine and recapitulate the sonia data but sort of set the stage for this argument that if we had a better biomarker we could predict who doesn't need but all that cost right of of a front line c_d_k_ inhibitor for years when they might do well on a_i_ alone more more than the median so something like t_k_ to me makes a lot of sense to say this is the group who's exquisitely sensitive and maybe doesn't need it and that paper was sort of a look back at traditional studies to see whether it mattered whether we deployed at first or second what about and maybe you can just kind of give me without doing a paper maybe would be cool to do a paper but when i saw the title what i was thinking was comparing first line c_d_k_ to c_d_k_ after c_d_k like how much how do you compare the benefit of the post monarch approach to using it up front yep yeah and i think you know the post monarch the way that i think about those studies maintain post monarch we had some retrospective data before post monarch they all suggest that continued blockade with c_d_k_ inhibitor has some biologic activity but the magnitude of that effect was was limited i think you would agree was on the order of a couple of months for the most part so if you look at what the other way to answer that question too is when you look at ember three was kind of interesting the inlunaster in abemocyclib combo forty percent of those patients never had a prior c_d_k_ sixty percent of those patients did have a prior c_d_k_ and the magnitude of benefit on that arm wasn't hugely different was like nine months versus nine point i think was five or seven or something so to answer your question there it didn't seem like the magnitude of deploying abemocyclib with inlunaster in the second line setting had a had a huge difference whether or not you received it the first time which surprised me i thought that the patients who'd never had abemocyclib would have really had a lot more benefit than the patients who had previously had a c_d_k_ inhibitor in the front line setting maybe the sir takes away some of that if you use a better antiestrogen again maybe getting too far out there but again these things pop into my mind it kind of reminds me the other issue here is is abema a better c_d_k_i then you know for example in terms of efficacy compared particularly uh. to palibo again i sort of flashed on uh. alch where you know a lot of people have been getting uh. elect anib and then lorlatin ed came along that looks better and yet you do see responses after elect anib to lorlatin ed but now everybody's given lorlatin ed up front so do you think abema is a better c_d_k_ i often i used to think a lot about this uh. when we were first doing the work in the lab about c_d_k_ resistance and and we we knew that abema had different dosing schedule potential c_n_s penetration had a different kind of pharmacokinetic profile where it hits some other targets outside of four six different you know as we said different side effects uh. i do think it's a little bit of a different drug uh. and the whole idea of post monarch and are in our retrospective data was to sort of show that you could salvage patients who had maybe progressed on palibo or i bow with abema um. laboratory modeling of that has been really difficult it's been really hard to prove in the lab that you can grow resistant to palibo and then be sensitive to abema at least in cell culture which again is obviously not as complex as a patient so i do think that there are differential activities with abema i think we can see some of that in the post monarch in the ember three um. data but i also think we have newer drugs coming next generation c_d_k_ four two drugs that are really active much more so than abema after progression on palibo or on ribo these drugs are moving from phase one phase two into phase three in combination with his next generation anti-estrogens we have a really cool study we're about to open any minute now looking at a next gen c_d_k_ four two inhibitor with genin tech in combination with gira destined and this drug has tremendous activity after progression on palibo ribo or abema um. in laboratory uh. models and in phase one so these newer drugs are hitting c_d_k_ two which i think is the key target after progression on c_d_k_ four six and abema has a little bit of that but not as much as the new ones always just going to say the point you made us a great one about it or latin and elective these new c_d_k_ four two inhibitors which were sort of proven in lab and phase one models of c_d_k_ four six resistance guess whether being moved frontline in precisely the same logic for some better drug move it front line sort of hit the target harder to begin with so we'll see we'll see if it holds up in breast cancer i think it will all right let's get into your case is starting out with the sixty five year woman so this is a case of a sixty five year old patient well-controlled hypertension hypothyroidism comes in with bone pain she hasn't had a mammogram but a while pet scan with diffuse acyst lesions and an f_d_g_a_v_ right breast mass that's measuring four centimeters with lymphadenopathy she gets a bone biopsy that's e_r_ positive p_r_ positive her to one plus we do baseline sequencing of the tissue from the bone nothing actionable we do baseline c_t_d_n_a_ she's got a low t_m_b_ stable microsatellite no actionable alterations we put around let's resolve a ribo we see significant improvement in the bones reduced f_d_g_ uptake the breast mass shrinks the regional lift node shrink she has sclerotic changes over time in the bone she gets great control for five years and then she has isolated progression after five years in one spot in the pelvis. Now in this situation I tend to try to call my friends in radiation to see if we can get more time. So we gave her XRT to that area. She continued Lettruzel and Rybo for another six to six to nine months. Then she had more progressions and a couple of bone lesions and a couple of spots in the liver. Now we repeat the CTDNA. The TMB is still below 10, but now she's got two ESR1 mutations, D538 and Y537. The elelic fractions are you know less than 5%. She has a amplification in cycling E1, which we've seen as a resistance driver to CDK. So we decided to put her on elicestrant. Second line, she has good control with stability on the PET scan for nine months. She has some further bone and liver progression with some new symptoms. Sure TMB is nine. We still see the ESR1, but look the elelic fractions actually lower on the ESR1 after the elicestrant. We see some P53 and now FGFR changes and we put her on TDXT because now she's symptomatic. She's got a 1+IHC her 2. She's growing in the liver. We've already given her multiple lines of endocrine and CDK therapy. She does not have a PI3K pathway and she didn't have an actionable FGFR that we saw. So we put her on TDXT. So a very common situation. We also brings up the issue how to look at the VAF. Is 5% or 3% is that high? Is it low? Is it more about how it changes? How do you look at that? Yeah, I put that on here to sort of prompt I think this discussion. It can be affected by tumor purity, right? So you can see if you take multiple samples, little changes in VAF. I hesitate to sort of make big decisions based on the VAF. I think the point here is a low VAF, even if it's 1%, doesn't mean that you can't go after the target, right? Because what you're seeing in the blood may not be the actual VAF and all the individual cancer cells because not all of them are shedding DNA into the blood. I do think to your point changes in time in the VAF is perhaps more important than what the number is itself. So for example, if you have a VAF that's rising over time in the same patient, it suggests that that population is becoming more and more pronounced or dominant. And look here, when we targeted the ESR1, we actually saw a drop in the VAF, right? And then we see other things emerging, like P53, etc. So I think the VAF is less important than the concept of present or absent. I think the VAF is more interesting in the same patient when you're tracking it over time. But I also think with the new data you and I looked at from Serena 6, that we may have a future where we can use kind of on-target drops in VAF to kind of convince ourselves that our missile guidance is correct. We're hitting the target correctly, right? By seeing the drop in the VAF. So another thing that's interesting about this is this patient had a good response for five years to CDKI. I'm curious what fraction of patients like this, I mean she was the novo and all, but just what fraction like this do go for five years and whether it helps predict. I mean she had kind of a typical response, but how much of a difference in response do you see as it gets longer and longer out? Yeah, I think she's also a great example of that. Emerald subset data, right? Where the patients with the longer duration on the frontline CDK, you know, are the great candidates for ill-assestorant. You made the point and I agree with it that a denovo never before treated strongly ER positive, bone only patient is most likely to have the best duration. This also brings up the question you had brought up a few minutes ago. Might she have been okay on an AI alone for a number of years, right? And we could have spared all the monitoring and the toxicity on the CDK. Maybe, maybe she could have been. I would say, you know, the median time on frontline CDK kind of all comers would be two to three years. I've had patients who progress within two to six months who are just up front refractory. I've had a few patients who I've inherited from my colleagues. They're more than a decade. They've been on the drugs since it was FDA approved and they still have metastatic disease on the scan. They're not cured. You know, if we stop it, they'll they have disease that that would grow. A patient like this, I think, is on the good is really on the good end probably in the top 20% or the top 15% but I've seen longer, you know, in probably 10% or less of the patients, but she's in the she's in the upper quartile, I think. So also, I'm going to see cases like this. I'm used to the next step of being in the patient had a great response to TDXD. Did she? Yeah. So she's doing well on TDXD and as you as you guessed, you know, these patients were chemo naive, right? Even after several years of therapy, they tend to robustly respond. And again, this brings up the question though, in clinic, who could be okay on Cape, right? With with just oral therapy versus who needs IVTXD. The data suggests, of course, from the destiny breast trials that the TDXD is better, but it's the same question with Sonya. Does everybody need TDXD or can some people with this kind of disease get away with less? So, I mean, this lady was 65, but why did why did you choose TDXD, which did you feel she needed, which is symptomatic or? Yeah, she was, yeah, she was starting to have more symptoms and we wanted to sort of use what we thought was going to be the most likely to drive a quick like deep response. All right. Let's hear about this 55 year old woman. So 55 year old patient with rheumatoid arthritis doing well not on therapy. She had a T1N1 ER positive tumor in the past got surgery, TC chemo remediation, let's resolve for five years and she's been off for two years now. So she had kind of standard frontline treatment, has refractory hip pain, had imaging concerning for bone lesions and lung nodules, FES. We did an FES pet, which is the ER signaling pet scan that we can talk about that was positive. Bone biopsy again, ER positive breast cancer PR negative hurt to zero. Sequencing showed the E45, E545K sort of common pick 3CA mutation with a 30% alleleic fraction P53 CT DNA with a TMB just over the cup point at 11 stable microsatellite. We see the same pick 3CA, same TP53. She goes on AI ribo with zymeta does, okay, but not a huge response. She gets about eight months with stability, has worsening disease at that point. We repeat the CT DNA. She's got the same TMB, the same pick 3CA. Now she's got a couple different P53. She's got an RB mutation and an ESR1 with a 1% alleleic fraction. So she's acquired RB and ESR1. So she's got second line therapy. So we decided to give her full vestrant with Kopeva assertive, which we can talk about. She's got improvement in multiple areas and she said good control for about a year. Then she gets pet CT, new liver lesions, bone disease, the lung is stable. Her TMB is creeping up. She's 15. Pick 3CA hasn't changed as it usually doesn't. We have multiple P53, RB and NF1, but we don't see the ESR1 anymore. Liver biopsy again, metastatic breast cancer, her to zero. You could tell I'm looking for the ultra lower, the one plus. So I did a repeat liver biopsy. So we give her chemo with Kopeva. She's got a good response for about eight months. Then we give her Sassatuzumab and that's where we're at now. So she got one line of chemo. We couldn't give her TDXD. She was her to null Zediro. We did an ADC with Sassatuzumab in the fourth line setting with some improvement. So why does she lose her ESR1 on Cappy? Yeah, I think it's a good question. The elelic fraction to begin with was quite low. It could have been that it was cleared with the full vestrine and with the Kopeva assertive. It could be that it's still there. It's just below the threshold of detection that whatever's growing is kind of the nastier NF1, P53, RB, mutant cells. I think when the elelic fraction is exceedingly low, like a fraction of a percent, you're right at that cut point of whether they can report it or not. So you can see it fluctuate a little bit. It clearly to the point we were making over time is not rising. Nor would I be super excited about giving Endocrine monotherapy after she's already been through multiple lines of doublet therapy without a ton of response. I mean, the other reason with her compared to the first case, she was only on her frontline CDK for less than a year without robust response on her PET scan. So if you look at the Emerald subset data, even if she had the ESR1, she would have fallen into the category of patients who did kind of not as well as the patients who had longer durations, right? Yeah, I was going to ask you about that. I'm actually, it was eight months that she progressed. Looks like she had a good response to Kopeva. So do you see the same thing with Kopeva assertive that with shorter time on CDK, they don't do as well or is it more about serve? Yeah, I don't think we have that data for Kopeva. It's a great question. I think anecdotally I don't see quite the same correlation. And also it's a doublet and you're targeting a specific alteration like the Pick 3 CA. So I don't think I would view it the same way. If somebody had a reduced response on frontline, I don't think I would hesitate to use the Kopeva. The only time I might hesitate is if they're really in like visceral crisis. And I need to quickly debulk them where I would use conventional chemo or TDXD. So is that one of the reasons that you use the Kopeva in this patient that you know you were less enthusiastic about using a serve because of the short time on CDK? Right. Yeah, so I was less enthusiastic about single agent endocrine therapy, even if you had sort of a higher ES or want to lealic fracture. I wanted to use a doublet. You kind of had the, you know, you can almost sense that this patient isn't quite as endocrine-sensitive based on what the scan looks like and how long the control is on the front line. So you were going to comment on FES pub, had them curious when you use it. Yeah, we're starting to use a lot more of it. We're generating some data on this right now where you use it. I think when you have imaging findings at baseline that you're not sure about. So, you know, FDG can light up for a lot of different reasons, right? Inflammation, autoimmune stuff, wound, you know, healing infection, cancer. FES is highly, highly specific for breast cancer. And so if you're trying to get a sense of the amount of disease or you have indeterminate findings like lung nodules that you can't access with, you know, with bronchoscopy, I've also found in patients who had subtle CT findings and subtle FDG that the FES is more sensitive. I had a patient with an oligo lesion in the sternum on a CT that I couldn't figure out. Even on the FDG, it wasn't great. I did an FES. She had more disease in other areas that didn't show up on the FDG or on the CT. So the problem with FES is you can't really repeat it if they're on certain drugs. Like full vestric, the washout is six to eight weeks. So you can't take someone off medicine for six to eight weeks. On an AI, you can do it any time. On a SIRM, I think it's a few weeks, a couple of weeks. So depending on the drug they're on, there's a washout period and that can really limit it. So it's a lot of times it's baseline. You're trying to figure out what's going on if there's new imaging findings or the extent of diseases and clear. All right. Let's finish out with your 78-year-old woman. So I have a 3A and a 3B and I had a feeling you would like the TK. So I brought you two variants on the TK testing. So this is a 78-year-old patient with a fib on ELEQUIS. She had a T1N0 ER positive tumor, got surgery, chemo, radiation, and an AI. She had a recurrence six years later and got a NAST result. This was an in-brest recurrence with a right mastectomy and switched over to full vestrant. She then, this is in the setting of a recurrence. She then came back with back pain, on full vestrant. She had multifocal liver and bone lesions and the liver biopsy showed ER positive PR-negative HER21+. sequencing of the liver negative, CT DNA with a TMB of 14 MS stable, P10 loss, NF1 and P53 mutations with an FGFR low level amplification. So we put this patient, my colleagues did, on Lettrezol and a Bemicyclub. She progressed within three months, okay. Now check out the TK value here. So her TK level, the Thymidine kinase in the blood, at baseline was high at 400, when they put her on a Bemicyclub. She did not clear her TK. She did not go undetectable. And look, she has the P10 and the NF1 mutations, which are known drivers of resistance to CDK inhibitor. So if I had looked at her CT DNA and said, you know what, I'm worried about these mutations. We have data in the lab and we have data in other patients that they don't really do well. And her TK showed no drop and she progressed within the first eight to 12 weeks. So this is an example of an unsuppressed TK level with genomic findings correlated to CTK resistance. Now I'm bringing you to this abstract here. This is one of our fantastic fellows who's working with me, Elena Michaels. She's doing a really interesting project where she's laying the DNA results, the circulating tumor DNA, on top of the TK, to show in patients who have known drivers of resistance to CDK what the pattern of TK is. And this patient was a perfect example of a patient who had a known resistance driver, no suppression of TK, and rapid progression clinically. So I think that shows potentially the utility of the TK test. Now let me show you the flip side. So here's three B different patients, same series, okay? 70 year old patient healthy comes in with diploplium and back pain. CT shows bony lesions, calvaryl mass with some soft tissue extension. Petscan with mildly avid soft tissue nodule on the left breast and lytic lesions in the thoracic spine. Bone biopsy, ER positive, PR positive, her too low. The baseline CTDNA TMB is around 10, MS stable, nothing actionable. She gets a chyphoplasty in radiation and she goes on lectures on a BEMMA. So she has some problems on the BEMMA, has to dose adjust, gets down to 50 milligrams, and then has an excellent response and she's still responding. So look at her TK, this is also very interesting. So she has a high TK, she starts CDK, she drops to almost undetectable. They hold her drug to dose adjust and look what happens when they hold her drug. She pops up to 160. Then they get her back on drug at a better dose and she goes back down. So this is a patient who has a deep drop in TK. It's sustained, she has no resistance mechanisms on her CTDNA and when you're adjusting the dose, it's so sensitive that even when you're adjusting the dose, when she's off for a couple of weeks, you can see it flip back up and then it flips down again. So I'm just looking, yeah, so I'm curious about the Diplopliya, whichever retinol met. I think she had some soft tissue extension that was pushing a little bit on the, on like the occipital, so they did radiation. Yeah, they covered it with radiation. Well, interesting. You know, I've been flashing a lot on the whole idea of, you know, what we used to have of neo-adjuvant therapy to try to see whether people respond or not. And now I'm watching this thing and we don't need, we get, this is blood tests. You don't need tissue. That's right. So it's kind of like similar to how we used to, K67 and all that, but now we're looking at the blood. So one final question, I've got to throw out there before. One final question, which is, I mentioned we were doing this thing with CTDNA and I did a program with Dr. Puzdai from Yale. And when I was looking up his CV, I saw this really cool paper he did. I don't know if you were aware of it. When they looked at seer data and they determined what fraction of breast cancer deaths are patients who originally started out with stage one. Are you familiar with that? I don't think I've seen that data. He's a phenomenal expert on this topic too and he's working with us on the Thymidine kindness as well. What was the result of the seer analysis? Yeah, so what would you guess? What's your guess? No, no. All the people who die of breast cancer, what fraction of them were originally stage one. My gut tells me, I'm just thinking about my own practice, that it would be relatively low, probably on the order of 10 to 20%. But it might be a surprising number. 23%. So on the upper end of that. Yeah, but I mean, I think because you see this, there's some biology there that even though you caught it early, there's still some sort of propensity for those patients. Yeah, well that's kind of his point that even though it's a very low risk situation, there's so many people who are stage one that it's a quarter of breast cancer death. So my question to you is, think about all the technology you've just been discussing. How would you apply it to that specific problem? Finding those stage one patients who maybe we need to approach differently. Yeah, and I think you brought up a great concept earlier about AI. There may be a way to look at those patients and look at their genomic profiles with some of these newer sequencing platforms and start to derive a signature for patients who have higher propensity for recurrence, right? Ankyotype kind of gets at that, but it's a small number of genes and its gene expression. And when you were telling me the story, I was just thinking about a few patients I've had who had very low archetypes, right? They had stage one or stage two disease. Their archetype was very low. We forego chemotherapy. We give them endocrine therapy, but they recur. Now it hasn't happened a lot, but it's happened a couple of times. And the point that I teach the fellows when we're sitting in clinic and dealing with this rare but sad situation is that archetype tells you the overall risk of recurrence. If it occurs, it doesn't tell you how it's going to behave because I've had patients with very low archetype who when they recur, it's not an indolent necessarily ER sensitive tumor that does well for 10 years. It can be an extremely aggressive tumor. And I think that's the point you're making here just because it's stage one when they recur. It doesn't mean that they're not going to die from, unfortunately, from the breast cancer. That original biology doesn't dictate the course of the disease. So we may be able to use some of these deeper sequencing tools, even in the primary tumor where there may be tiny, tiny fractions of these bad mutations to start to say, "Uh oh, like this person might have the capacity for a more difficult recurrence later on." And to your point, maybe we shouldn't be thinking about them like a stage one. Maybe they need more chemo or maybe they need 10 years on endocrine therapy or maybe they need signetare monitoring, right? Or something like that, even though you wouldn't think necessarily from the pathology and the anatomy that they should get there. It's a very interesting finding and kind of matches to some extent that feeling with the archetype of those rare recurrences with the archetype of five, right, or something like that. This concludes our program. Reload financial relations up to the faculty can be found in the program notes. Note that the use of agents and treatments that are not approved may be discussed in this program. So check out the package inserts for more information. Special thanks to Dr. Wander and thank you for listening. This is Dr. Nielo for oncology today, optimizing biomarker assessment and related treatment decision making for patients with hormone receptor positive metastatic breast cancer.