Case (Dr Ibrahim): A man in his mid-80s with metastatic castration-resistant prostate cancer (mCRPC) whose disease progressed through multiple therapies is found to have CD274 (PD-L1) amplification, microsatellite instability (MSI)-high status and mutations in ATM and EGFR

DR LOVE: I thought we could start out with really, I think, a signature case related to our topic today. This is sent in by a general medical oncologist in a small community in Indiana, and here’s the case: 84-year-old man with a history of metastatic castrate-resistant prostate cancer who received all standard therapies, leuprolide, abiraterone, enzalutamide, docetaxel and finally cabazitaxel, which he was presenting with.

At that point he had a large pelvic mass, obstructive uropathy, renal failure and hyperkalemia and presented to the emergency room. Called this oncologist. And at that time, they were saying in the ER, “This man either has to get dialyzed or go to hospice.” And, of course, that’s a pretty difficult decision in a patient like this. But as the doctor, the oncologist was thinking about it and seeing the patient. He checked a liquid biopsy that he had sent 3 weeks previously and found some pretty interesting findings in it. Specifically, PD-L1 amplification, CD274, we’ll talk a little bit about that. I’ve heard about MSI, of course, but supposedly this patient had both, which I didn’t even know occurred. One question the doc’s saying is, “How often do you see microsatellite instability to start with in prostate cancer?”

Another question was, now this man, because of his age and his tenuous situation, actually, amazingly, he recommended dialyzing the patient and started him on pembrolizumab, which he told me was kind of interesting to talk to the family, because they thought he was going to head to hospice. But he explained what these findings were and the hope that maybe the pembrolizumab would help the patient. They went along with it. He was temporarily dialyzed. Now he has had 1 dose of therapy, and he is off narcotics. there’s been no imaging, but clinically, he’s improving.

But anyhow, kind of an amazing case in terms of that. He also brought up the question — he used pembro — but what about ipi/nivo, which we’re hearing a lot about? We’ll talk about today MSI. But also, and I think this really starts to get into some of the complexities of dealing with patients and these kinds of assays, the liquid biopsy showed an ATM mutation. He’s saying, “What about a PARP inhibitor in a patient like this?” The liquid biopsy also showed an EGFR R1068 mutation. I’ve heard of exon 20 insertions and a few other weird ones, but I’ve never heard of this one. He wants to know, what about an EGFR TKI?

And then the other question he had is, suppose the patient just had the PD-L1 amplification without MSI. Would you still expect a checkpoint inhibitor? Emmanuel, pretty simple case — just kidding — but anyhow, really, this is what it’s all about. Any thoughts about it?

DR ANTONARAKIS: I’ve looked at a lot of genetic reports over the years, and this one is one of the most complex I’ve seen in a long time. And there’s some interesting things here and some interesting things that don’t really jive as well, which makes it even more challenging. The thing that jumps out at me is the MSI-high designation. That’s there. But the paradox is, when you look down that list, you don’t see any mutations in the canonical mismatch repair gene. There’s nothing in MSH2 coming up. There’s nothing in MSH6. There’s nothing in MLH1, and there’s nothing in PMS2. Yes, you’re seeing the MSI-high designation on the report, but you’re not seeing the genomic mutation that supposedly gives rise to that.

Effect of CD274 amplification on response to immune checkpoint inhibitors; activity of PARP inhibitors in patients with ATM mutations

DR LOVE: Now that you’re at checkpoint central of the universe at Hopkins, I’m curious — of course we heard about the CD274 amplification thing in Hodgkin lymphoma. That’s supposedly why they’re — but then there were reports, Andy, that this was found rarely in a bunch of other solid tumors, and I think they had, like, 12 patients or something that were treated, and a bunch of them responded. Is that really playing out that way?

DR MCKENZIE: Yes, so far. Those data are very sparse. Like you said, it’s a really rare population of solid tumor patients that have amplification of the gene that encodes PD-L1 CD274. But it seems like those patients do respond very well to checkpoint inhibitors, at least in that small patient population.

DR LOVE: Ian, what about this issue that he brings up of maybe not right away, but just more theoretically, an ATM mutation. At least that was seen in the liquid biopsy. Any thoughts about that and the theoretical relevance of a PARP inhibitor?

DR KROP: I mean, I think building on the proven benefits of PARP inhibitors with the BRCA1 and BRCA2 mutations, where clearly there’s a synthetic lethality that happens, because the tumors are dependent upon PARP-mediated DNA repair, because their homologous recombination pathway through BRCA 1 and 2 is deficient. There’s certainly interest in whether other genes involved in DNA repair may confer sensitivity to PARP inhibitors. In breast cancer, there are trials going on now to see whether PARP sensitivity is mediated when you have things like ATM mutations. I don’t know what the data are in prostate cancer yet, if they have actual clinical data on the sensitivity when you have a mutation like that. But I think, again, certainly mechanistically it makes sense to explore it. And at least in breast cancer, that’s being done in the setting of a clinical trial.

DR LOVE: I was mentioning that a couple times now we’ve gotten people from different disciplines together, and it’s always great, because now I can just turn to the person sitting next to you and say, “What’s the answer to his question?”

DR ANTONARAKIS: There’s an interesting feature here, in fact. This person has 3 ATM mutations, not just 1. Two of them are frameshifts, and the allele frequencies are 1.5% and 1.0%. That suggests that they’re somatic. And then the third mutation is a splice site mutation, and the allele frequency of that one is 42%. Whenever you see an allele frequency that is between 40% and 60%, this is virtually always a germline mutation.

DR LOVE: Hmm. Wow.

DR ANTONARAKIS: And when I look at this report, I’m thinking to myself this guy has a germline ATM mutation. We would have to prove it by doing a saliva sample or a blood sample, but with an allele frequency of 42% it’s almost certain. And then the second issue is that if we believe that he has a microsatellite instability, multiple genes, including the ATM gene, have microsatellite tracks within them, that the microsatellite instability can cause a secondary mutation in the ATM gene, which is not a driver. It’s a passenger, and it’s a consequence of the MSI-high status, not a driver.

DR LOVE: Wow, that is really interesting. But just to get a more fundamental question — we’ll get into this later when you go through the data, but do we know about response to PARP inhibitors with patients with prostate cancer and germline ATM?

DR ANTONARAKIS: The responses to PARP inhibitors with ATM mutations are much, much more limited — less than 10% of the time. We do occasionally see profound responses, but in prostate cancer most of the profound and durable responses to PARP inhibitors are in BRCA1 and BRCA2 patients.

DR LOVE: Andy, we’ll talk later about the setup at Sarah Cannon and how you’re reaching out to oncologists like this, but I’m sure you’re hearing about tons of EGFR mutations. What’s R1068? Is that an activating mutation you would think about a TKI in?

DR MCKENZIE: It’s one that we’d have to do a little bit more digging on, because as you mentioned, it’s not one that comes to top of mind when you think of EGFRs, but I will say that it is a premature stop codon there at position 1068. And usually premature stop codons, depending on the protein and depending on where it exists in the protein, cause truncation and in some cases loss of function. I’d have to dig a little deeper on that to see if that’s something that would be activating or not.

I’ll point out, though, that this report also has mutations in IDH1, FGFR2, other genes that are known to be mutated in cancers and have molecules that are in development. This is, as was mentioned earlier, a very complex case that I think opens up a lot of opportunities, especially from a clinical trial point of view.

DR LOVE: And this patient could be potentially involved in 5 different kinds of trials, theoretically. Who knows? Maybe he’ll respond and be able to get different kinds of therapies.

Advantages and limitations of multiplex genomic testing

DR LOVE: I presented this case really to really get into the key question I have here today, representing general medical oncologists and also patients, Johanna, which is, this would be an example of looking for a needle in a haystack and finding a pretty good one.

DR BENDELL: Yes.

DR LOVE: If it really plays out that way. Of course, this is a very dramatic example. We’re going to talk in a second about some of the data we got in this survey we did of 77 oncologists. But I just want to get all your take right now, and then we’ll come back at the end and see after we go through all this, do we still all feel the same way? Is it worth it? I mean, I don’t want to get exactly into a cost/benefit thing, but when you look at cost/benefit, it’s always the issue about the value. What’s the benefit? Then you can look at the cost. And I think it also has to be taken into context with a lot of the other expenditures that we make, which, whether that’s right or not I don’t know, but it’s a reality.

Johanna, is it worth it to be looking for needles in a haystack like this?

DR BENDELL: I think yes, but I don’t know what I would say in the process of looking for the needle in the haystack. It is for some of the needles in the haystack, and you’ve heard me say before, catching NTRK fusion, it’s like winning the Powerball, right? Very rare, but if you hit it it’s amazing.

DR LOVE: And we saw a bunch of these in the survey.

DR BENDELL: Yes.

DR LOVE: NTRK fusions that were treated.

DR BENDELL: Yes, and this gentleman as well. And I think where some of the benefit lies is not only for the individual patient, right, to be profiled and actually learn to see, what are potential driver mutations? What are potential actionable mutations with approved therapies? Or potential clinical trials in the future.

But I also think for the population as a whole, because a lot of the data that we have on mutations and fusions and expressions all come from, for instance, TCGA, which is all off of primary tumors when the patients are first diagnosed. What happens to patients’ tumors along the way? We’ve seen multiple examples where we might accumulate mutations or develop fusions, and I still think we don’t really understand what happens with those patients’ tumors as they go and what might be potential actionable things that happen along the way.

I think in terms of the broader population, where we have a lot of these large real-world evidence databases that are also encompassing the molecular profiles that are done both at baseline and longitudinally, if we’re able to aggregate that data and take a look at that data in conjunction with how patients are doing, I think that that might give us even more information on how to potentially approach treating patients’ cancers as a whole. And what I mean is benefiting the whole population of patients with cancer.

DR LOVE: We’ll actually look and see how this is playing out. I was very surprised, actually, in our survey, how many people actually responded to PARP inhibitors — breast and prostate. How many people with MSI they found. Like you said, NTRK fusions, there’re a bunch of them there.

I don’t know, Ian, I have to admit, probably people would be surprised. I’m kind of a supercloset skeptic about stuff including targeted therapy, because it’s great, but is it really going to end up curing people? I don’t know. What are we achieving with this? At a human level, sure, it’s great to see a partial response, but I want to be convinced. Cases like this are pretty helpful, though. Any thoughts about this?

DR KROP: No. I mean, I think a healthy degree of skepticism is appropriate here. There’s the natural inclination for all of us — we have a new technology, of course we’re going to use it. And I think that’s not the best use of resources. It’s not the best thing for patients. And we have to be skeptical about it.

But it was interesting, a couple years ago at San Antonio Breast Cancer meetings there was a big debate, a staged debate, on whether we should be doing some type of gene sequencing on all patients with metastatic breast cancer. And 2 very accomplished faculty members were presenting, and it really was hard for the person on the pro side to make a compelling argument that back in 2017 we should be doing sequencing on everybody, because there were very few examples, at least in breast cancer, in which patients found something that led to a therapeutic benefit.

But I think that that’s changing. I think partly because the number of needles is increasing. As we get more targeted therapies, there’s more chance that if we find a particular niche population of a mutation in a patient that you can actually do something about that, and clearly the number of targeted therapies is increasing. I think that’s going to help. And I think we kind of get data from — to me, this is a great opportunity for real world data because typically we use these drugs as monotherapy, it’s easier to figure out what something’s working than when you use 3 different chemotherapies plus a targeted therapy. We’re using these as monotherapies. A lot of time when the drugs work they actually have objective responses as opposed to stable disease.

I do think this is an opportunity for there to be some collection of real-world data to start looking at these populations. Because it’s hard to do in the setting of a basket trial when these things are rare, I think this is an opportunity.

But again, I think skepticism is important. This particular case is really good for that patient, but it’s by far the minority right now. But I think that certainly there’s lots of opportunity for advancing this. And again, as more targeted therapies become available, it’s going to make it easier.

Role of serial testing to identify variations in genomic mutations or aberrations over time; potential benefit of liquid next-generation sequencing (NGS) panels

DR LOVE: One final point, Emmanuel, kind of thinking a little bit about what Johanna said about really learning more about the disease and thinking about this case where a liquid biopsy was done, where you can do multiple liquid biopsies over time and start to see changes, what do we know — I mean, obviously it’s different with different cancers — about the dynamics of how genomic alterations change over time?

DR ANTONARAKIS: Right. The advantage of a liquid biopsy, of course, is you, don’t need a solid tumor biopsy. With prostate cancer, a lot of our men have bone disease only. Those biopsies, outside of highly specialized centers, are very, very difficult. Oftentimes all we have is a liquid biopsy. The other beauty is that, as you said, the mutations can change over time. The importance of that depends on the actual mutation. In prostate cancer, for example, the ATM mutations and the BRCA1 and BRCA2 mutations typically don’t evolve. They’re there from the very beginning. And if they’re in the primary, they’re going to be in the met. And if they’re not in the primary, they usually don’t develop in the met.

With mismatch repair mutations, about one third of them are acquired. They’re not present in the primary, but they are present in a met or in a circulating tumor DNA. If you’re looking for a BRCA and all you have is the prostatectomy, that’s probably good enough. If you’re looking for an MSI-high or a mismatch repair mutation in prostate cancer, then you really do want to get either the met or a liquid biopsy. It’s gene dependent and context dependent.

DR LOVE: Andy, the kind of paradigm that I started thinking about, starting out with lung related to liquid biopsies in general was, if you see it, believe it. If you don’t see it, it’s probably true but not necessarily. You need to get tissue. Is that still the case in general?

DR MCKENZIE: I think the first part of that is definitely true. If you see it, you can believe it. The second part of if it’s not there be skeptical, I think the assay sensitivities are getting a lot better. But there is still the notion that tissue biopsies are the gold standard for these types of NGS tests. And I’ll kind of go back to something that was said earlier, that in prostate, seeing things like resistance mutations pop up in the androgen receptor is something that you can also monitor over time that’s kind of in a similar vein to what you saw in lung cancer — that now you can monitor these mutations as they emerge, and resistance to therapy.

Now, what that means for actual changing care and therapies? I think that’s yet to be determined, but at least now you have a biological reason for why a patient might be progressing that you can see in a mutation report.

DR ANTONARAKIS: Could I – 1 thing, Neil. The problem with liquid biopsies, or one of them, is that there is this phenomenon called clonal hematopoiesis, also called CHIP, C-H-I-P. And these are mutations that occur in the leukocytes of these patients. And if the positive test should always be believed, is what we just heard, I’m not 100% sure that’s always the case, because you don’t always know if that mutation is coming from the tumor DNA or from a leukocyte. And in prostate cancer patients, many of whom are older men, this hematopoiesis CHIP condition becomes more prevalent with age. I have often found p53 mutations, for example, in liquid biopsies, and I’m really not sure if that p53 mutation is coming from a leukocyte or from the prostate cancer tumor DNA. That is just an additional potential concern or caveat with the liquid biopsies.

DR LOVE: Wow, I did not know that.

DR MCKENZIE: That’s a great point, but I was coming at it from a technical point of view. That mutation exists. It’s there. But if we don’t know where the origin —

DR ANTONARAKIS: If it’s a leukocyte, it’s irrelevant for the patient’s cancer.

DR MCKENZIE: Right. Exactly.

DR LOVE: Wow. I’ve never heard that one. Incidentally, what’s an AR, androgen receptor amplification?

DR ANTONARAKIS: Right. There’re many ways that the androgen receptor tries to overcome —

DR LOVE: Because this patient had that.

DR ANTONARAKIS: Yes. Hormone therapy. It basically means the gene itself is amplified 10-fold, 100-fold. You have more copies of the AR gene itself.

DR LOVE: Are they more sensitive to hormonal therapy?

DR ANTONARAKIS: They’re typically more insensitive.

DR LOVE: Insensitive. Hmm.

DR ANTONARAKIS: Yes. It’s a resistance mechanism for AR-targeted therapies.

DR LOVE: That would make sense. This man’s already had 3 or 4 forms of therapy.

Use of multiplex genomic assays by community oncologists and implications for clinical practice

DR LOVE: Alright, let’s talk about this survey that we just completed of 77 US-based medical oncologists, trying to figure out what they’re doing in their own practices in metastatic colorectal, prostate and breast cancer. The docs were randomly recruited from across the United States and come from different backgrounds in terms of their geographic locations and practice types.

One thing we asked them that I thought was kind of interesting, I’d seen this in an NCI survey that they did about genomics, was how comfortable they feel in decisions about patient treatment and management with some of the things that we’re going to talk about today, one being MSI testing. We also asked them about genomic tissue assays and also liquid biopsies. And we said, are you not at all confident? A little confident? Moderately confident? Or very confident? I’m going to say, I mean, in general, I think the answer ought to be very confident, and yet it seems that a substantial minority do not feel that way. Hopefully programs like this can contribute to that. And we’ll talk today a little bit about needs in terms of education.

When you look at the practices of these 77 docs, they see a lot of patients. I think it’s probably a typical general medical oncology audience. But, I mean, the number of patients here, particularly prostate, I don’t know why, but for some reason, the fact that they have 30 patients in their practice in the last couple years that they’ve seen with metastatic prostate cancer, I mean, I guess it’s just the way things are nowadays. You can look at breast cancer, in a way, as 3 diseases, based on ER and HER2 and, of course, the prominent one is ER-positive, HER2-negative, but the other two minorities are substantial, triple-negative and HER2-positive. Here’s what we really want to find out about, which is, what are they doing in terms of genomic assays? This begins to get into the issue of their use of multiplex testing.

And you can see that there’s a substantial number of patients who have had tissue-based assays, not as many as you might expect who’ve had liquid biopsies, but not an inconsequential number. We also asked them, what type of assay? And this is kind of consistent with what we’ve been hearing in our conferences. A lot of use of FoundationOne assays, but also Caris^®. And you can see some of the others that they’re utilizing.

We asked them a couple other general questions. We’re going to get into what they said about the specific modalities, but a couple teasers. First we asked them to cite an example of a patient with metastatic colorectal cancer where they found an actionable target.

And you can see, the 2 most common ones that are identified are MSI, BRAF — I guess we can talk about that in terms of do you do that as a one-off test, or do you do it as part of multiplex and what the implications are. Also interesting, HER2, that they’ve got a bunch of people with HER2. We know that’s not uncommon. But 3 NTRKs there. In prostate cancer, again, a couple more NTRKs, MSI, BRCA. Kind of similar in terms of the actionable things that they found.

And then finally in breast cancer, PIK3 mutations. Again, we can talk about the different ways that that’s brought up. But now, of course, with the option of PI3 kinase inhibitors like alpelisib, that becomes relevant. But also, you see more BRCA. Still MSI and NTRK. It’s kind of interesting, NTRK, a fair number of patients. We’ll talk more about the specifics, but I’m just kind of curious, Emmanuel, in general, in terms of some of the numbers I just went through, is that what you were expecting or not?

DR ANTONARAKIS: I was surprised by the range of things that they found. It means that these people are testing a lot. I mean, the fact that they found NTRK mutations, which in these 3 cancers is probably 1% of each cancer type, to find 2 or 3 means that they’re testing 300 people. Or maybe they got lucky, or maybe their patients got lucky. But yes, the usual suspects are all there. The BRCAs are there. The MSIs are there. I think those are the two across the board most important ones. And then each cancer type has a few additional disease-specific alterations.

Optimal timing of genetic testing for patients diagnosed with metastatic disease

DR LOVE: Another issue, I think, Ian, that we’re going to kind of get into, and we’ll show some data as we go through this, is when to do an assay like that. Should it be done at the time of first diagnosis of metastatic disease? Should you know what a typical first- and second-line therapy is, no matter what you see? Should you just hold off on doing it? Any comments in breast cancer about how you think it through and how it’s different based on a different subtype?

DR KROP: It’s changed a lot, because 2 years ago there was no reason, really, to do it other than identify patients for clinical trials. Now we have 2 approved drugs. We have alpelisib for PIK3CA mutants in ER-positive disease, and we have 2 PARP inhibitors for patients with BRCA1 or 2 mutations. I think our general practice now has been for ER-positive, HER2-negative disease to start to test patients during their first line of therapy, that if they have a PIK3CA mutation, that you’re queued up to offer alpelisib in the second line. And, similarly, to know their germline BRCA status relatively early for both ER-positive and triple-negative cancers.

I think definitely the recommendations now are to do testing early for ER-positive and triple-negative disease. I think the bigger question, as we were alluding to it earlier, is, when do you start doing serial testing? And right now in breast cancer, I’m not sure that’s going to be incredibly impactful, but it may be in the future.

DR LOVE: That’s really amazing. I have the opportunity to talk to investigators all the time. I’ve talked to many people about MSI. I never heard what Emmanuel just said, which is still ringing around in my head and kind of bothering me in a way, because there’re a bunch of people in here who were MSI high who responded to checkpoint inhibitors. This is for real. This has a real clinical implication. They’re not likely to get a checkpoint inhibitor if they’re not MSI high. In terms of timing, it seems like it makes more sense to do it closer to the time you might want to treat.

DR ANTONARAKIS: Yes. I mean, there’s many other ways to check for MSI. A cheap way is to look for the proteins. You can do immunohistochemistry for the 4 proteins. It probably costs $10 a pop. You do 4 of them, that’s 40 bucks. You can find out in 2 days.

DR LOVE: But then there’s 10 thousand other things there, and I don’t know what the story is, what they’re changing. I don’t know whether we know at all. What about this question of when, Andy? Any thoughts about it?

DR MCKENZIE: One of our leaders in our drug development unit, Dr Skip Burris, published a paper, kind of an editorial, this past year around who and when people should get tested. And in the context of access to clinical research, we’re big proponents of anyone in the metastatic setting getting that type of broad testing. But again, that’s if you have this access to clinical research programs that you can match patients to clinical trials for emerging new targeted therapies. For the general medical oncologists that don’t have access to robust research programs, it might be a bit more limited in what these tests actually show for clinical actionability.

DR LOVE: Although when you look at the list of patients that they have, where from their point of view they had actionable mutations, I didn’t see anybody write — oh, and they went on a trial and responded, too. I mean, it might have happened, but this is all off label, maybe not even off label in terms of MSI. It’s on label. But it’s all approved drugs that they’re using with this. But the problem is, I guess, Johanna, generally it’s not that productive. We’ll talk as we go through today how often you really find something you can do something about. What’s your own philosophy when you have patients with colorectal cancer — metastatic disease. When do you start thinking about multiplex testing?

DR BENDELL: I tend to do multiplex testing when I first see them in the metastatic setting. Of course we’ll always check for MSI status early on to help make adjuvant decisions, but if somebody comes in with metastatic disease, I’ll do a multiplex testing at baseline, because I think if you don’t have the data, you’ll get your expanded RAS analysis. You’ll get your BRAF. You’ll get your microsatellite instability. And I think one thing that we’ll probably talk about a little bit more is, does your assay actually include the right tests? For instance, we’re talking about HER2 status, and it was interesting to see HER2 being picked up in the colorectal. Now, there are sometimes HER2 mutations, but when we think about treatment of HER2 for patients with colorectal cancer, we’re actually looking at overexpression or amplification more than necessarily a mutation.

DR LOVE: The breast cancer concept.

DR BENDELL: Yes. We’ll actually be doing IHC to look at colorectal cancer. Also with NTRK, right? You might catch some NTRK on some of the commercial assays, but some of the NTRK fusions you might actually have to order special testing for. RNA-seq is becoming a big thing that people are looking at testing, to actually look at what proteins are being transcribed? And I think that that’s probably the next generation of testing that we’re going to see developed.

Presentation (Dr McKenzie): Overview of multiplex genomic assay methodology

DR LOVE: Maybe you could take a crack at talking about how you would explain to general medical oncologists or somebody like me kind of how these assays are done. And really what’s going on at Sarah Cannon right now, which I think is really an outstanding operation in terms of the whole concept of bringing this into practice. If you want to go through that?

DR MCKENZIE: Great. Very happy to. Thanks, Neil. Briefly, I wanted to just discuss the overview that we’ve got here of multiplex genomic assays and methodologies, use cases, limitations and cases where we think that this has a lot of really great clinical utility.

Very briefly we’ll just go through some of the very commonly used NGS platforms. We’ll talk through serial NGS testing. What that looks like, the rationales, benefits, limitations. Talk a little bit about the usage of NGS and how that’s grown over time, and studies that have come out analyzing patients who get matched to targeted therapies based on these NGS reports and what that looks like.

And then we’ll finish up with kind of where this is going, the horizon here of these novel biomolecules that are being analyzed. Things like exosomes and circulating tumor cells and things like that in the liquid biopsy space. It’s a really exciting time in the NGS world as more and more companies are coming up with better and better technologies to analyze tumor DNA.

This table here shows some of the most commonly used NGS platforms, including things like Foundation Medicine, that was discussed in that survey, Caris and others. And the big takeaway from this is that they’re each a little different. Each test offers a different gene set. They analyze the genes a little bit differently, and they also differ in the coverage. Some of them offer whole exonal coverage, and others are just hotspots. And knowing what you’re ordering when you go into a test is really important in understanding what the test resulted back. For instance, if you order a test that only tests for fusions in 1 or 2 of the NTRK genes, you’re only going to get results back for those 2 genes, and you might miss whatever might be present in that third NTRK gene.

I think that’s another thing for general oncologists to be aware of, is that these tests vary in cost, the sample requirements. You have some new technologies coming out that are offering 2- or 3-day turnaround times on samples smaller than a grain of rice. And I think the jury’s still out on how well those assays perform, but the market is getting inundated with these companies coming out with the next and best types of tests. Turnaround time is always something that we’ve heard is a barrier to ordering these. Folks want the answers as quickly as they can get them.

And then also you have to be aware of, is my assay detecting specifically germline mutations, as was pointed out earlier? Some of these can be detected, and even though it’s a somatic assay, these germline mutations can be detected. And end users have to be aware of that and at least savvy enough to question whether or not that particular mutation in BRCA or ATM could be a germline mutation.

And the last thing I’ll add is, most of the companies that we talk about do DNA-based next-generation sequencing. But as we’ve talked about earlier, and will further, RNA sequencing is emerging as a very important tool to detect complex fusions in large genes, especially in things like NTRK. And I’ll go down just a small bit of a rabbit hole there with NTRK to show what that means and what it looks like.

This slide here is meant to show why certain tests do and don’t detect fusions in genes like NTRK. NTRK is a very large gene, and the gene is primarily composed of these huge intronic regions. In the DNA you’ve got introns and exons, and it’s only through splicing that you get rid of the introns and you get exons. MRNA is the target that you would want to sequence to get rid of all of those big complex introns.

Just to follow along on the slide here, that top line is the entire NTRK1 gene, and these vertical black lines are the exons. And you can see the space between these vertical lines and these large intronic regions. And that’s replicated in NTRK1, 2 and 3.

NTRK-based sequencingdoesn’t do a good job in covering these really long sequencing regions that need to happen in order to detect these complex fusions. And companies have done a really great job of optimizing their assays for some but not all of the NTRK genes. And I’m just using NTRK as an example here. But realizing those limitations, they’ve started adding in RNA-based sequencing as a part of their routine offerings.

The chart on the bottom right just shows which companies offer complex fusion detection and in which NTRK genes they will call those variants. And as you can see, just because someone says I ordered NGS on my patient, depending on the vendor, you’re getting a very different result back, especially when it comes to detecting NTRK fusions. Again, start with the end in mind when you’re thinking about ordering these tests. What do I want to detect? And if NTRK is something that you’re looking for, then you have to make sure that you’re ordering a test that covers these and with high degrees of fidelity.

This is a question that we get asked routinely. Should I do a tissue test? Should I do a liquid-based test? And as was discussed earlier, I think there are pros and cons to each of these. Right now from an FDA perspective, we’ve got FDA-approved tissue-based tests. The liquid biopsies are continuing to emerge as incredibly important but don’t have yet that level of scrutiny on them.

Just to kind of walk through the pros and cons here, tissue biopsies, they’re clinically validated. They are invasive, though. Very difficult to perform serially, as you need new and fresh tissue biopsies to do that. One of the things that we’ve learned is, tumor heterogeneity gets lost in some of these tissue-based tests, as you’re detecting mutations in one tumor and beyond that in a very narrow slice of that one tumor. And you don’t get the whole picture of metastatic disease when taking a look just at a tissue biopsy. It’s also impractical for periodic monitoring of treatment response but does allow for histological diagnosis and staging, and as we discussed earlier, it’s still the gold standard for tumor characterization.

Liquid biopsies, on the other hand, are less invasive, easily repeatable. They have the potential to reveal tumor heterogeneity and give you this kind of comprehensive complete picture of the disease.

Pros and cons on each, and I think in the community setting, at least with our general medical oncologists that we work with, the liquid biopsies are gaining a lot of momentum, because it’s easy. You can do it in the office with a quick blood draw.

I will say that there are specific-use cases, I think, when liquid biopsies are most meaningful, particularly when you don’t have any tissue, you’ve exhausted tissue or you’ve got insufficient tumor within the sample that you did biopsy. Or if you’ve got patients with bone-only disease or are wanting to kind of regularly monitor response to certain therapies, and you can see the emergence of resistance mutations.

There are use cases for that. And as we’ll discuss, beyond the detection of actionable mutations, there’s emerging literature regarding other uses of liquid biopsies, particularly in prognosis determination, certainly as we discussed in the selection of treatment and monitoring of treatment efficacy in relapse.

But also, there’s a big space in early detection. These are your at-risk population of patients who don’t yet have cancer. Liquid biopsies are emerging as a potential use for early detection of cancers. That’s still very early work, but I think something we’ll see in the clinic in the future.

Another question that we get asked a lot is, “The liquid biopsy result that I got is different than the tissue biopsy result that I got. What gives?” And the simple answer that I try to give back is, there’re 2 separate tests — the technologies are different. The analytes are different. And you should expect some discordance there. The issue comes when 1 test detects a driver mutation and the other doesn’t detect a driver mutation. Which one do we believe? But I think that the overarching story here is, discordance is to be expected based on the technologies.

I’ve pulled a couple of papers here just to show that in 1 case the percent of tissue alternations found in cell-free DNA was between 20% and 30%. There’s a large discordance there, but like I said, I think the technologies and the actual tests themselves are the reason why those are different.

And it’s what we talked about earlier, that believe what’s there but be skeptical of things that might not be there. And kind of the theme that we give back when doctors have these questions for us.

Rationale for the use of serial testing and liquid assays to identify variations in genomic mutations or aberrations over time

DR MCKENZIE: Switching to another controversial area here, serial testing. When do I do it? Why would I do it? What does the data say right now about the use of serial NGS testing?

And we talked about this a little bit before in the overview, but it has the ability to reveal resistance mechanisms or emerging targeted therapy targets. And I’ll show a couple of slides on predictors of progression.

And there’s some studies that came out that we’ll talk about that show just the presence of and the amount of circulating tumor DNA can be a predictor of progression, maybe in the absence of or before radiologic progression.

Here are a couple of examples of what serial testing can do. The upper left plots there are the allelic fraction of circulating tumor DNA for a couple of targets in EGFR, the emergence of an EGFR resistance mutation, and that causing a doctor to switch treatment to a third-generation EGFR tyrosine kinase inhibitor.

And those are kind of the classic examples, but the bottom left graph also shows that that can be expanded, not just in lung cancer but also in breast cancer, where you have the emergence of one of these PIK3CA mutations that is a known target for alpelisib, the newly approved PIK3CA inhibitor. And the emergence of this mutation came up before in this study. There was evidence of radiologic progression, and this was a predictor of that progression. This gets back to the abilities. Some of these tests can actually predict progression before you get the scans back.

The case on the right is one from our institution where this doctor, over the course of a year, ordered 6 different molecular profiles. And this was their own doing. This wasn’t encouraged for them to do. And, I think, after speaking with this physician, they were looking for resistance mutations to come up. This was an EGFR-positive patient that was responding quite well to EGFR-directed therapy but then started progressing, and they wanted to know why. And they weren’t able to see any of the conical gatekeeper resistance mutations pop up.

But what emerged really late, if you look the right side of this graph here, what emerged later is amplifications in both EGFR and MET, which have both been associated with acquired resistance to EGFR inhibitors. Now this doctor has a rationale for why their patient quit responding, but more than that is considering a clinical trial option of a dual inhibitor strategy with an EGFR inhibitor and a MET inhibitor based on the amplification of MET there.

Again, just showing how serial NGS testing can reveal novel mutations that can change and augment treatment decisions there. A lot of pioneering work has been done in showing that tumors are very dynamic, and they change through time and therapy.

This is a lot of work that Dr Charlie Swanton has done in the UK just showing that the tumors evolve through therapy and over time. As was alluded to earlier, there are a set of truncal mutations that occur in a lot of cancers, but as they metastasize and progress, these branch mutations occur, and that evolution is something that we’re trying to figure out how to target.
And this is another instance where serial testing — maybe not over time, but through one tumor — can show you differences in mutation landscapes within the same tumor, which just makes this even more complex, as you’re looking at a tumor evolving and how to target that.

Differentiating between germline and somatic mutations in cancer tissue

DR LOVE: I’m going to go around the table and let each one of you ask a question or make a comment related to the data. I want to actually start out by asking you this. You were talking about the issue of differentiating germline from somatic, and before, Emmanuel was talking about allele number. Is that the way you do it? Or is there another way?

DR MCKENZIE: We take into consideration 2 things when we look at if a mutation is a potential germline mutation: the purity of the sample and the allelic fraction. If you have a very high tumor purity and detect a mutation at a high allele frequency, it’s harder for us to know if that’s a germline or not. However, if you have a relatively normal mixed purity of sample and a high allelic fraction, it’s almost guaranteed to be a germline mutation. Yes, allele fraction does determine a lot of what we consider to be germline in these tests.

DR LOVE: But, I mean, the sine qua non would be germline testing?

DR MCKENZIE: Correct. Yes. That’s what our advice is. If ever you see something that’s skeptical in a gene that’s known to be associated with hereditary cancer syndromes and at a suspicious allele fraction, we suggest getting that confirmed.

DR LOVE: Emmanuel, any comment?

DR ANTONARAKIS: There’s another potential clue, which is, it turns out that the type of mutations that occur in the germline are often different from those that occur in tumors. There are databases online. You can look up the exact mutation. For example, the one that I use is called ClinVar — C-L-I-N-V-A-R. ClinVar. And that’s a germline database. For BRCA2 and BRCA1, these mutations have been extensively described over the past decade and a half. And there’s certain mutations in BRCA2 you see an allele frequency of 42%. You’re suspicious. You’re not quite sure. You look it up on ClinVar. It’s been reported 14 times. You look it up in a somatic database like COSMIC or cBioPortal. It’s at zero times. Even before doing your germline tests, which you should do anyway, you kind of have a suspicion that it’s going to be germline. It’s also the nature of the mutation that can tip you off.

DR LOVE: Interesting.

DR MCKENZIE: Agree.

Perspective on the use of tissue versus liquid biopsy for genetic testing

DR LOVE: Johanna, any questions or comments?

DR BENDELL: First comment: That physician was not me who sent all of those tests, just to let you know. And I guess my question to you, Andy, is the one that you have alluded to the answer, but you really didn’t give one, because I know that you’re trying to slip out of it, is, there’s this common argument — I see a lot of GI cancers, especially colon cancer — we always usually have access to tissue to be able to send off for profiling. And then there’s Dr Johnson, who works 2 offices down from me, who’s a lung cancer doctor, and as much as she wants to get tissue, she oftentimes doesn’t have access to it. She sends a lot of liquid biopsies.

And when we look across our community at Sarah Cannon, we see this mishmash of tissue and liquid. And it also is sometimes practice dependent, what people are sending off. We have 1 practice in particular that sends off a lot of liquid biopsies.

My question to you is, if you have a tissue sample, would you say send that tissue and not the liquid? Or would you say pick one?

DR MCKENZIE: Yes, I still think it’s dependent upon when that biopsy was taken. If it’s an archival tissue biopsy from 5 or 6 years ago and they’ve had 3 or 4 lines of therapy, it might not be the best thing to send off for testing. And if they don’t have disease that you can easily biopsy, in that case I would certainly advocate for a liquid-based biopsy to give you a current snapshot. If the biopsy is fresh, that would always be our preference, is to do a fresh biopsy.

But that’s another thing that we’ve kind of tracked in our practices over time, is the time between when a biopsy was taken and when a test was ordered, and when these NGS-based tests kind of first hit the market, we were seeing folks use 3- or 4-year-old biopsies. And in the past several years we’ve seen that time go down to months, if not within the past year. And we’re excited to see that, as I think that education is getting out there. But it’s a great point, Johanna, that the fresher tissue biopsies are always the preference. If you don’t have one of those, the liquid is probably the better way to go.

DR BENDELL: Thank you. And also, I guess also with RNA-seq. That’s even more important, right? The freshness of the biopsy?

DR MCKENZIE: Yep. That’s right. And the technologies that are out there in the liquid space are lacking when it comes to their ability to detect changes in the RNA. RNA is a bit more labile and doesn’t tend to be amenable to testing in the blood. Again, fresh tissue, especially in the context of RNA sequencing, yup.

Sensitivity of liquid biopsies in detecting tumor mutations

DR LOVE: Ian, anything you want to ask in terms of the stuff that we’ve just been covering?

DR KROP: Yes, I think you’ve pointed out very well the pros and cons of liquid versus tissue. In breast cancer we have a lot of patients who have bone-only disease. I think that was one of the examples you mentioned. We are doing more liquid biopsies. And we’ve said now multiple times about this idea of if you see something, it’s there probably. If you don’t see something, it may not be absent.

Is there some way of looking at a liquid biopsy report and get a sense of how likely it is to pick up a particular mutation? I mean, I had a patient who has clinically HER2-positive cancer. She didn’t have biopsiable disease. We got a liquid biopsy. And it showed a few mutations, allele frequenciesof, like, less than 1%. And the patient called me up and said, “What’s going on? It’s not showing that my cancer’s HER2 amplified anymore. Is my cancer now HER2-negative?” And I worried that because there probably wasn’t a lot of DNA there, or tumor DNA in the biopsy, is that why they didn’t have the sensitivity to detect the HER2 amplification at that point? Is there something you can look at and say, “This is probably going to be reliable or not”?

DR MCKENZIE: That’s a great question. But what you alluded to, the allelic fraction, can tell you a lot about what’s going on in your sample. If you have one, as was presented earlier, where one of the alleles was detected at 40-some-odd percent and the other ones at 0.01%, that test has gone well and has detected circulating tumor DNA, and you can rely on those results.

It’s ones where you have very low allelic fractions where you’re uncertain if the enrichment techniques actually worked to collect as much DNA in there. And then also if your tumor is actually shedding the DNA to begin with. Because that’s what these assays rely on, is the tumor shedding the DNA into the blood stream. And that usually happens under two circumstances, either a highly metabolic tumor or a tumor that’s actively undergoing apoptosis and dying and releasing that into the bloodstream.

You can kind of squint and say, “Alright, is this disease, like, really aggressive?” And if it is, it should be producing a lot of DNA into the bloodstream. If it’s not, if it’s kind of indolent, then you might not expect there to be a whole lot of DNA in there. And in those cases, what comes back is maybe not the best picture of what that tumor actually is carrying.

Available data exploring the impact of NGS use on patient outcomes

DR MCKENZIE: The last part here should go pretty quickly. Usage and outcomes.

This is really just a snapshot on the bottom left, in this kind of orange and black graph, showing the utilization of NGS-based testing in community oncology practices since 2012. And you can kind of see this exponential rise in the utilization there. The black caps are plasma-based NGS tests, and the orange bottoms there are the tissue based.

It’s being done on a whole variety of tumor types. That’s what that colorful graph in the middle is.

And as I alluded to earlier, the time between when a biopsy is taken and when the test is performed is shrinking. Which shows that people are using fresher biopsies. And, Neil, you alluded to this earlier, there’s a significant portion of general oncologists who still feel like they aren’t very confident in interpreting the results that come back on an NGS report. That was shown several years ago, and I think still persists today. There’s an education gap there in what to do with these reports when I get them back.

And the last bit that I want to focus on is a question that was posed earlier, is it worth it? And what data out there shows me if it’s worth it or not?

And just to show a quick snapshot of some of the clinical trials that have been done, matching patients to targeted therapies and looking at responses there, this slide here I’ll just talk through that. A couple of studies very early on looked at time to treatment failure, either on a matched targeted therapy or looking at the therapy immediately prior to. And what people saw is that when you gave a matched targeted therapy, there seemed to be an increased time to treatment failure in these early studies. And that’s what kind of spurred on the field to say, “We should do some large-scale clinical trial matching studies where we take a specific mutation and match it to a targeted therapy.”

And those are your studies, as I’ll show in the next slide, like the NCI-MATCH study, the ASCO TAPUR study and a couple of others that are out there that are actively analyzing that question. When I have next-generation sequencing data, what does that mean for overall survival and progression-free survival in a matched targeted therapy population? And those are, like I say, ongoing studies, for the most part.

There have been some readouts from a couple of these, and I’ll show you just a couple of quotes from a couple of studies here. The WINTER study, genomic and transcript mechanalysis are both useful for therapy selection. Higher degrees of DNA and RNA matching independently were associated with longer PFS and OS. From the NCI-MATCH study, T-DM1 was well tolerated in GE and gastric cancers, and there were some responses that were seen as well. This is all data that’s emerging, trying to prove out whether or not it is worth it to do these large-scale NGS studies and match your patients to therapies that might be approved in other indications.

Selection of a multiplex genomic assay for detecting genetic alterations

DR LOVE: Just getting back to when you were talking earlier about the various assays that are available, I have a pretty simple question, which is, how do you pick which one you use? It kind of seems like part of this is, like, maybe over the heads of a lot of people and they’re kind of trusting a brand or something. They’re not necessarily understanding all the exact —

DR MCKENZIE: Yes. Uh huh.

DR LOVE: They want to turn to people like you and say, “What do we do?”

DR MCKENZIE: Uh huh. Yes, that’s a great question, because I think you have to start with the end in mind. What am I expecting to detect with this report? If it’s in a case like secretory breast cancer or a sarcoma, you want to make sure that you’re testing for all of the NTRKs that are possibly out there. You want to make sure that your test covers all of those.

In cases where that’s super rare and you might not be expecting to see that, I can understand someone who wouldn’t go the extra mile there. But the theme that we’re seeing is tests that are what we call comprehensive, not hotspot tests, and then also ones that are starting to include RNA sequencing are the preferred vendors that we want to try stick towards. And most of them are starting to incorporate RNA sequencing as a part of their standard offerings.  

DR LOVE: Emmanuel, we’re kind of in general in the mode of, what do you do? And I appreciate all the refinitiesof what you just talked about, but I still can’t figure out what to do. Any advice?

DR ANTONARAKIS: The other consideration is, is the test FDA approved and is it generally reimbursed? And the Foundation Medicine NGS test has been FDA approved for about a year and a half. And as long as you indicate somewhere in your note that one of the reasons you’re doing the test is for MSI, since we have a tumor type agnostic FDA approval of pembrolizumab in those patients, you’re using an FDA-approved test with an FDA-approved indication, I cannot remember the last time I’ve had a patient getting a bill from Foundation Medicine for their assay. That goes a long way, because I can tell you, the first time a patient gets a bill, that’s the last time that physician’s going to order that test. It’s a huge disincentive.

DR LOVE: Could you just clarify the FDA approval status of both liquid and tissue multiplex assays?

DR MCKENZIE: Yes, sure. I think there are now 3 tissue-based tests that are FDA approved. The MSK-IMPACT^TM test is FDA approved. Foundation’s CDx^TM is approved. And then I think one is Oncomine^TM is FDA approved. That one has a smaller gene list than these large tests do. And in the liquid space we haven’t seen one that’s been FDA approved yet. And that’s still on the horizon.

But that’s the FDA kind of status there for tissue-based testing. And I don’t know if there’s a different thought there or something that I don’t know about, but that’s what our landscape is.

Implications of multiplex genomic assay results for therapeutic decision-making

DR LOVE: Again, moving beyond whether you can get it paid for in terms of just putting that aside in terms of what might be optimal, Johanna, any thoughts about that?

I mean, I kind of think a little bit, again, this idea of, for example, with breast cancer the 21-gene Recurrence Score^®, Oncotype, there are multiple genomic assays used in that situation. But the 21-gene assay is by far utilized more commonly, and part of that is, they have stronger data. That they put out a lot of data. They’re in the academic sphere. Is that something that’s part of how do you factor? Again, putting aside whether you get it paid for, is it advantageous to use one or another of these, either tissue-based assays or liquid assays?

DR BENDELL: I think part of it is what results you’re looking for. For some of those Oncotype assays, you’re looking for a risk of recurrence. Should I use chemotherapy? Should I not use chemotherapy? And these are different decision points. And I think that that’s 1 set of tests. And the other set of profiling tests is, what am I going to do with it? And is it going to help me decide between standard therapies, or is it going to be something where again, we’re looking for that needle in the haystack, and we’re going to say “Hey, would this patient potentially be matched to a different FDA-approved targeted therapy that I might want to try?”

And I think with some of the data that Andy just presented, which suggested that if you match a patient’s profile to the type of treatment that that patient’s going to get, will that result in better outcomes? And I think you have to take that very much with a grain of salt, because we’ve seen profiles where you can use a treatment and, almost tumor agnostically, the patient’s likely going to receive benefit. Like microsatellite instability. Like NTRK fusions.

But then we have other ones where it’s not clear-cut. PARP and BRCA is a very, very interesting — the pancreas data, very different than the ovarian data. And there’s still a lot of debate within the pancreas community about how much benefit olaparib will give a patient with pancreas cancer as opposed to other types of tumors.

PIK3 mutations, another one that’s out there where very common mutations in colon cancer — have we really seen efficacy of PIK3 inhibitors in patients with colon cancers? Very, very mixed. I think you have to really consider before taking an FDA-approved targeted agent and giving it to your patient who has a mutation that you might pick up on one of these assays really what the data is that’s out there about using that particular targeted agent in that particular tumor type.

DR KROP: I think there is this fundamental issue here of, Johanna’s saying that the provider needs to know the data for using a particular drug in a particular mutation. How do they get that data?

I mean, if you look at a Foundation report, you see there’ll be a particular mutation and there’ll be some curated potential therapies, but there’s no differentiation of this one’s got an 80% response rate and this ones got a 10%. It’s just a list. Is there a way to get that information for the average provider to look at a Foundation report or a Guardant or anything else and say, “This is something I should act on, and this is not”? Because right now we can’t get that —

DR ANTONARAKIS: I think that’s the job of the molecular tumor board. I mean, to be honest with you, I get a Foundation report. It’s 35 pages. I look at page 1, page 2 and I throw away the rest.

DR BENDELL: Yes.

DR ANTONARAKIS: The matching trials is completely useless.

DR KROP: Yes.

DR ANTONARAKIS: It’s uniformly useless. First of all, because they include trials that are not relevant. Or they include, let’s say, 15 PARP inhibitor trials, and then you’re looking down this list, which one should I choose? They’re all reasonable.

I think that’s where having a molecular tumor board where you can have molecular pathologists, clinicians with genetic knowledge, germline clinic people there and then to give that clinician a recommendation. And ordering the likelihood of benefit, like that case we saw in the beginning. The first thing I would do for that patient is, I would try pembro, because it looks like he’s MSI high. The second thing might be to target the ATM mutation. The third thing might be to target the IDH1 mutation. In that order, right? The clinician needs to know. If they start with an IDH1 inhibitor in a patient who’s MSI high, they’re wasting their time.

DR KROP: Right. At Hopkins you have that. What about the oncologist in Peoria who just gets the report and doesn’t have an academic center to back them up? How do you get that molecular tumor board information that he or she can actually use?

I mean, again, I think this is an unmet medical need. We’re bringing in a test, and how you act on it is not clear at this point.

Specific biomarkers and genetic alterations routinely evaluated in clinical practice

DR LOVE: One thing we asked about was, it’s become the kind of the traditional workup in terms of biomarker evaluation in a patient with metastatic disease. Interesting that when we ask about pan-RAS status, while almost everybody says they do assess, it’s certainly been standard for a while. It’s interesting, for some reason they don’t all do it first line. Which I think is kind of interesting, although maybe they think it’s not going to affect their decision. They don’t need it.

HER2 status, actually, almost a third of them don’t assess HER2 status. And we know there’s targetable therapy available. I think that’s kind of interesting. BRAF, it’s interesting. They assess for BRAF, but again, not necessarily first line. And you can question — I certainly hear investigators talking about considering BRAF status in terms of first-line therapy.

They pretty much all assess MSI status, again, not necessarily first line. You can kind of question why. And then we get into some really interesting stuff.

We asked them about the last patient in their practice who they treated who had an MSI-high colorectal cancer, and you can see that almost all these people got an anti-PD-1 antibody. Various lines of therapy, usually not first line, as you could imagine, since there are a bunch of effective therapies available.

And then the thing that really was interesting was of the 44 patients where we have information at this point, in other words, it’s not too early — 41 have responded to a checkpoint inhibitor. I don’t know what that means, but it’s interesting, that’s for sure.

And it’s interesting. Consistently we saw this with MSI and PARP, that the response rates are high. Now, it’s not a clinical trial, but it’s a random bunch of oncologists. I don’t know what it means. We can talk about it.

Effects of up-front multiplex genomic assay testing on outcomes for patients with mCRC

DR LOVE: We also asked them about whether or not they generally order a multiplex genomic assay in patients with metastatic colorectal cancer. And almost all of them do, and about half of them do at first diagnosis, as you were discussing, Johanna. We asked them before about their whole experience. Now we said, “Okay, last patient that you sent a multiplex assay who had metastatic colorectal cancer, and were the findings useful?” And amazingly, in half the cases they say yes. Now maybe this is up front, and they got a BRAF or something, is maybe not this needle in the haystack thing. But when you look at what the findings are, you see a couple MSI, a couple HER2, et cetera. But again, the RAS status is also something, which, I guess that’s really more up-front therapy.

Did the patient respond? These are in that subset of patients, and when they had a treatment that was determined, they would have been in those half of the doctors, most of the time they responded to treatment.

Now again, this is not like doing a formal clinical research trial. It’s just a snapshot. But the good thing about this is, it’s kind of fresh.

We asked them about the last patient who had a liquid assay with colorectal cancer, and you can see a lot fewer of those were helpful in these patients, and a lot fewer experienced response. This is another last patient thing we do that is always interesting.

The last patient in your practice who died of metastatic colorectal cancer. Median age of 65. Median duration from diagnosis of metastatic disease to death 24 months. Of course it’s a little bit skewed, because they died. It’s not necessarily a typical patient. But the question we wanted to know was, in that patient did they get a multiplex genomic assay? And you can see that most of them did. It looks like it’s almost about two thirds of them did have a multiplex assay. Kind of a statement in terms of what’s going on in clinical practice.

And we also asked, okay, in this patient, this patient who died in your practice, again, somewhat skewed, was, were the findings in the multiplex assay helpful? And in 12 of them they were. Not the majority, but kind of surprising the patient died either way. And you can see a bunch of those patients, again, from the point of view of the doc, responded to treatment.

Before we start going through the data, any thoughts, Johanna, about the picture this paints of management of metastatic colorectal cancer in a general medical oncology setting?

DR BENDELL: It’s funny, because even though the numbers are still not 100% or 90% towards more up-front testing to help guide clinical decisions, they’re certainly better than they used to be. I was actually rose-colored glasses, really happy to see that.

DR LOVE: I was surprised.

DR BENDELL: Yes.

DR LOVE: Up front.

DR BENDELL: Yes, yes. And the number of physicians that are looking at HER2 testing and thinking across the spectrum of relevant clinical findings that they might find from the testing. I was very impressed with the survey results, that they are testing and they are looking for the appropriate things and they are potentially trying to treat based on them.

Response to immune checkpoint inhibitors in patients with MSI-high mCRC

DR LOVE: Any thoughts, Emmanuel, again, it’s just a sample, a small number, all kinds of biases, et cetera, but we saw this also with the other cancers, a lot of responses in MSI high. Kind of more than you see in the trials. And maybe that’s some kind of artifact or something, but any thoughts about that?

DR ANTONARAKIS: Yes, I think we have to worry about recall bias. In other words, if a patient responds, you’re more likely to remember them.

DR LOVE: Right.

DR ANTONARAKIS: And if a patient doesn’t respond, you try to tuck that away from your memory, even if it was last week. A lot of these physicians are seeing 5, 6 patients a week that they’re doing NGS on. And if you’ve treated one guy with pembro and he responded, and another guy the same week and he hasn’t responded, the one you’re going to remember first is the one who responded.

I worry that these estimates are inflated, but the published literature tells us that 50% of these patients respond. That’s a high proportion. And your data suggests, like, 85%, which doesn’t seem realistic.

DR LOVE: Any thoughts about that, Ian?

DR KROP: I think that there are selection biases to these kinds of analyses, but the bottom line is, there are real, clinically meaningful responses in nontrivial percentages of patients. And this is a test that’s easy to get, and I think that the data are quite compelling.

DR LOVE: I mean, we did say “the last patient.” It may be that they substituted another patient, but technically if they answered it correctly it should be random. But who knows, there may be some other kind of bias. When you see the spectrum of issues that came out here, does that mirror what you see at Sarah Cannon at a larger level?

DR MCKENZIE: I think so. We have a very diverse group of doctors, both in age and training backgrounds, and the uptake in testing for some has been very easy to convince, and others we’re still working on that, from an education standpoint, when and where it’s most appropriate to do that, especially in colon cancers.

DR BENDELL: I will also argue that with the checkpoint inhibitors and the perception, I completely agree with recall data, but it’s funny, because even in my own recall data, and as I was listening to the survey results, I was thinking about my last patients who’ve received immunotherapy checkpoint inhibitor therapy with MSI. And I’m actually usually very surprised when I find somebody who doesn’t respond. And I wonder if part of that is due to when we report out response rates on clinical trials they’re by RECIST criteria, where we have to have a 30% shrinkage in disease to call that a response. And I’m wondering in our own minds when we see patients in the clinic, we’re not sitting there calculating out their RECIST. We’re seeing if the patient’s doing well. And I’m wondering if that might have something to do with it as well, because when you think about the number of patients with microsatellite-instable cancers that have progression as their first response, that’s, I think, a much smaller number than you would think of 50%.

DR LOVE: If you think about a waterfall plot, if the patient shrinks 20%, and then they do better, and they’ve responded, although, as you say, maybe it’s more reflective of a clinical benefit. That maybe it’s higher than we thought.

Presentation (Dr Bendell): Biomarker assessment in mCRC; role of multiplex testing

DR LOVE: Why don’t we go through some of the data, specifically on colorectal cancer, and put it out on the table. Because I know from our other experiences with multiple tumors that the other investigators are not necessarily always on exactly the same page about what’s going on. Maybe we can get everybody up to speed.

DR BENDELL: Really exciting about colorectal cancer, and it’s even more complicated now for our doctors that treat other cancers such as lung cancer, but we actually have different subgroups, depending on their molecular profile at baseline, that might have a different treatment algorithm. And what I want to point out here is, the patients, if they have RAS mutations, they might be treated in a different way than patients who have BRAF mutations than patients who have RAS wild-type tumors. Our choices of therapy may change or differ based on the molecular profile that we find at baseline for patients.

Let’s talk a little bit about what that is. In profiling our patients with metastatic colorectal cancer, we’re looking at, first of all, RAS status. And it used to be KRAS status, but now it’s expanded RAS status, because we found that there’s other mutations in RAS, including NRAS, where that might affect the patient’s response to EGFR inhibitors with their treatment.
And, of course, BRAF-mutated colorectal cancer patients, who we’ll talk about later, who will have some very specific treatment recommendations based on that mutational status.

And then we have microsatellite instability versus microsatellite stability mismatch repair that we’re looking at. Nowadays we’re starting to look at HER2 status, because we found that HER2 treatment may actually result in a very good and prolonged response for patients who do have HER2-mutated disease.

Let’s talk about a really interesting test that we can do for our colorectal cancer patients, which just is in the history itself of left- versus right-sided cancers. And what we’ve seen is, we’ve always known that right-sided colon cancer, that includes the transverse colon, which is only about 2% of colon cancers. They do worse prognostically overall than patients with left-sided tumors.

And why is that? There’s some very beautiful work, actually, out of Johns Hopkins, that has looked at the cancer microbiome, and looking at the bacteria in those different areas of the colon that might affect not only the molecular profile but development of these colon cancers. And we’ve certainly seen different molecular profile analyses on right-sided tumors versus left-sided tumors.

And what we’ve seen, and here’s the overall prognosis based on tumor location for colon cancer patients, is again, you can see here that overall survival, left sided versus right sided, it’s about one third better for the left-sided than the right-sided cancers. And we’ll talk a little bit about how that might make an influence in terms of choice of treatment.

Actionable genetic alterations in mCRC; detection of BRAF and HER2 mutations and implications for practice

DR BENDELL: What is the baseline guidance for biomarker testing for colon cancer? What should we get if we’re not ordering a multiplex analysis? What specific things should we be looking for at baseline? And that includes, again, expanded RAS, BRAF and microsatellite stability.

Let’s talk first about BRAF-mutated colon cancer. BRAF V600E mutations occur in about 8% to 10% of patients with metastatic colorectal cancer. And interestingly enough, when we started to see BRAF inhibitors being developed for treatment of patients with melanoma, we said, “Oh, this is great. We can maybe use these BRAF inhibitors for patients with BRAF-mutated colon cancer.” And when we gave them, we saw very small responses for the colorectal cancer patients. And there was some beautiful work that was done on both sides of the pond, both in Boston as well as in Amsterdam, trying to figure out why that was happening.

And what we saw was, in colon cancers, patients with BRAF mutations, when they’re treated with BRAF inhibitors, upregulate the EGFR receptor on the surface of their cancer cells, which becomes an escape mechanism for BRAF-mutated colorectal cancer cells to escape the BRAF inhibitors. And what we’ve learned is, we probably have to combine treatment with a BRAF inhibitor and an EGFR inhibitor to treat patients with BRAF-mutated colon cancers.

And I’m going to, as an aside here, I’m also going to say this is a really important learning for us too, because not all BRAF mutations across tumors are the same. And we have to remember that when we see a certain mutation on our molecular profiles, it may not be the same in different tumor types.

What does that all mean in terms of our treatment? We looked in terms of adding an EGFR inhibitor to a BRAF inhibitor for patients with BRAF-mutated colon cancer. This is the SWOG study that looked at vemurafenib, a BRAF inhibitor, plus cetuximab, plus irinotecan in the second-line setting for BRAF-mutated colorectal cancer patients versus cetuximab and irinotecan alone.

And you can see here this primary endpoint of progression-free survival, a significant advantage to using the BRAF inhibitor plus EGFR inhibitor plus chemotherapy for these patients. And this is one of the original trials that really showed that difference in a randomized fashion.

Even more recently we looked at a completely targeted approach through the BEACON trial, which looked at binimetinib, which is a MEK inhibitor, that’s along the RAS/RAF/MEK pathway, encorafenib, a BRAF inhibitor, and cetuximab, the EGFR inhibitor, for patients with BRAF-mutated colorectal cancer. And this was a randomized study that looked at the triplet therapy versus the doublet therapy of the BRAF inhibitor plus EGFR inhibitor versus dealer’s choice of irinotecan-based therapy with an EGFR inhibitor.

And we saw some very beautiful data come out of this, with this overall response rate for the triplet of 26%, median overall survival of 9 months compared with 5.4 months, with that hazard ratio of almost 0.5. Interestingly enough, even the doublet had very good responses, as well as benefit in overall survival, to treat those patients with BRAF V600E mutations.

We see here a very specific molecular profile that needs a very specific treatment.
Patients who have BRAF-mutated colorectal cancer in general have a poorer survival than BRAF wild type. They have about half the overall survival. What about HER2?

I brought this figure up here because we think about HER2 in breast cancer, and that’s where we looked at a lot of the data, and how can you tell by immunohistochemistry whether or not a patient has HER2 1+, 2+ or 3+? And one thing that we learned was a very important lesson, which I just bring up here, is that HER2-expressing GI cancers look much different under the microscope than HER2-expressing breast cancer. When you’re looking at that pathology report, and it’s really relying on the pathologist to have experience in looking at HER2 staining of GI cancers as opposed to breast cancer, where GI cancers are going to have this luminal staining that’s going to look very different than the breast cancer patients.

And here is an assessment of HER2 status in metastatic colorectal cancer that came from the study that looked at the combination of trastuzumab and lapatinib for patients with HER2-positive disease. And this is how they defined, by both IHC or FISH analysis, a HER2-positive colorectal cancer patient.

And this is the HERACLES trial itself, and you can see here this is a waterfall plot that those of us who treat patients with solid tumors love to see. It almost looks like a leukemia waterfall plot, where you see all of these beautiful responses for patients who had HER2-positive colorectal cancer being treated with both trastuzumab and lapatinib. And you can see here about a 40% response rate, but even more importantly than that, on the spaghetti plot, this is the Italian study, that on the right side those are really prolonged responses that you saw.

This is the combination of trastuzumab and pertuzumab for patients with HER2-positive, treatment-refractory metastatic colorectal cancer. And this is one of the basket trials that Dr McKenzie was alluding to in his presentation about taking patients with HER2-positive disease, tumor agnostic, and treating them with anti-HER2 therapies. And you can see here, just like we saw with the HERACLES trial, this response rate of almost 40% for patients with HER2-positive disease, and these were also prolonged responses.

Rationale and indications for the assessment of MSI and DNA mismatch repair (MMR) status; emerging role of tumor mutational burden

DR BENDELL: What about MSI or mismatch repair? When we look at guidelines for testing for patients with colorectal cancer, we’ve got several different ways to look at microsatellite instability status or MMR proficiency or deficiency for patients. And probably the way that’s most commonly used, and probably the cheapest way to do it, is by immunohistochemistry, where you look at 4 mismatch repair proteins and you see by immunohistochemistry, do you have presence of all of those proteins? If you have presence of all of your mismatch repair proteins, then you’re what we call mismatch repair proficient. Meaning you probably don’t have microsatellite instability.

I say that because when you read the reports, sometimes you have to look, because if you’re not realizing what the test is looking at and you see the presence of all of these proteins, some may think that means you have microsatellite instable disease, and I’ve seen that misconception, or misperception, in some of the interpretations of these reports. But what you’re actually looking at is a loss of 1 of these proteins.

Some also will do it by polymerase chain reaction, or PCR testing, looking for microsatellites or number of microsatellites that you see within a sample. When you have microsatellite instability, you see a lot more of various sizes of microsatellites when you do PCR. And you can test for it that way. When you look at concordance between IHC- and PCR-based testing, they are over 90% concordant with each other. They should be fairly interchangeable in terms of pathology tests.

And then now that a lot of us are using next-generation sequencing, we’re also using next-generation sequencing to look at microsatellite status. And you can use that as well.

Again, what’s the difference? You’ll hear microsatellite instability. You’ll hear mismatch repair. You’ll hear them used interchangeably. And I think most of it has to do with what test you used to determine, whereas deficient mismatch repair, which is akin to microsatellite instability, just means you looked by immunohistochemistry for the loss of 1 of those mismatch repair proteins.

What are some of the testing strategies we use? And again, this is just a look, pictographically, looking at the immunohistochemistry versus looking at all of these different microsatellite pieces, the variation in sizes and differences there.  

What does it mean to have microsatellite instability and a BRAF mutation? We get asked this question a lot. And when we think about somebody having microsatellite instability, we say, “Okay, do these patients have germline mutations?” Which makes you think of Lynch syndrome, which makes you think of having to look at all of the other family members for the presence of microsatellite instability. It also makes you think about having to look at the patient in terms of other endoscopic surveillance that you might have to do, versus patients that have somatic mutations. And what we’ve seen as one way to differentiate, very roughly, is by the presence or absence of a BRAF mutation.

We’ve found that patients with somatic microsatellite instability — this is not germline — sporadic microsatellite instability usually have a BRAF mutation that goes along with it. And the reason why is — the mechanism by which they develop this microsatellite instability is by what we call hypermethylation along their DNA. And we also see that the BRAF also gets hypermethylated and is mutated. And that’s how you get the correlation between BRAF mutation, microsatellite instability, and that’s a somatic rather than a germline.

Why do we use checkpoint inhibitors for patients with microsatellite-instable or deficient mismatch repair colorectal cancer? This is the way I simplistically explain it to myself.

When we replicate our DNA, we can make mistakes along the way. We have the mismatch repair proteins that help us correct those mistakes along the way. If we are lacking in those mismatch repair proteins, then when you replicate your DNA, you’re going to make mistakes. And when you make mistakes, you get mutations. When you get mutations, you get more frequency of developing different types of cancers. But also, when you get more mutations, you can potentially form what we call neoantigens, or abnormal-looking proteins, that then the immune system will react to and see and react against. And when you look under the microscope at a colon tumor who has microsatellite instability, you’ll see a lot of T cells in the area, because they’re sensing all those neoantigens that are there in the tumor and are ready to react against them.

And when you add a checkpoint inhibitor like PD-1 or pembrolizumab, you actually turn on the immune system to be able to react to all of those neoantigens, and you see these beautiful responses, as in the original data set, looking at pembrolizumab for patients with MMR-deficient colorectal cancer.

We’ve also seen this for nivolumab alone, but even more impressive, adding the CTLA4 inhibitor ipilimumab on top of nivolumab has given us what I think is the most impressive data for patients with microsatellite-instable colorectal cancer. Where you see patients at 1 year, about 86% of the patients are surviving out to 1 year with using the combination of nivolumab and ipilimumab.

And it’s the combination of nivolumab 3 and ipilimumab 1, which is different than the regimen that we use for melanoma, which is ipi 3, nivo 1, which could potentially result in more toxicity than the other way around.

And, interestingly, when you look at subgroup analysis, you say, “Does it matter if the patient has a KRAS mutation? Does it matter if they’re germline or somatic microsatellite instability? Does it matter what their PD-L1 expression was?” And in general, the answer is no. The combination of nivo plus ipi or nivo alone has about the same efficacy across all of those subtypes.

Role of liquid versus tumor biopsies for detecting genetic alterations

DR BENDELL: And this is talking about the liquid biopsies, and I think we talked about this with Dr McKenzie: What do we use? It’s a joke that somebody did — call it the stethoscope for the next 2 thousand years.

And let’s take a look at what other alterations we might be looking at aside from what we talked about for patients with colorectal cancer. And you can see we’re looking at RAS mutations, but we also see a fair number of patients with PIK3 kinase mutations, loss of PTEN. We see the BRAF V600E, microsatellite instability.

But I want to point out here a couple of things: One is the polymerase epsilon mutation. Patients who have polymerase epsilon mutations tend to have high frequency of mutations and might be patients who would respond to checkpoint inhibitors.

We’re also looking here — there’s 1% of gene fusion. We do also see NTRK fusions for patients with colorectal cancer. That’s something to remember as well, when you’re looking across the results that you might get in profiling of some of these patients.

This is one of the ways that we can look at RAS mutations and some of the data that’s been published looking at liquid biopsies for colorectal cancer. And this is using beaming technology through liquid biopsies to look at RAS mutation status within patients with colorectal cancer. And this is just small populations. It’s about 238 patients looking at plasma versus tissue results using beaming technology to detect RAS mutations.

And you can see here a positive agreement of about 92%, a negative agreement of 94% and overall percent of agreement of 93%. Being able to look at liquid biopsies with this technology to detect RAS mutations is viable.

This is another publication looking at, could you use liquid biopsies to look at microsatellite-instability status? And there’s some controversies to how you might interpret the results here, but I do think that what this tells us is that we can find patients with microsatellite instability using liquid testing. The question is, is it ready for prime time, and can it replace tumor analysis? And I think for right now tumor analysis, again, remains the gold standard, because you may miss some of them using a liquid biopsy. But that is a possibility. It’s a 95% positive predictive value.

This is also looking at prognosis. The presence of circulating tumor DNA from the colon tumors and prognosis, and as Dr McKenzie alluded to earlier in his presentation, you can see that for patients who are circulating tumor DNA-positive, they tend to have a poorer prognosis, as you have those elevated levels of circulating tumor DNA.

We’re going to start to look at this in terms of maybe should we use a different treatment for these patients? Should those patients be put on clinical trials looking at things like personalized cancer vaccines to help prevent them from having that re-emergence of their cancer in the future?

And this is a study that’s ongoing, COLOMATE, using a liquid biopsy in colorectal cancer to then select a molecular targeted therapy. This is an umbrella trial using the Guardant360^® NGS assay that’s ongoing. We’ll wait to see what data this trial may show us. 

MSI/MMR and tumor mutational burden as predictors of benefit from immune checkpoint inhibitors

DR LOVE: Could you just comment on why you’re thinking about that on the issue of correlation between MSI testing? We had a case of a patient who had negative IHC. The doc was suspicious because of the family history, and the patient ended up having germline testing, which showed Lynch with MSH6.

How do you explain that case? And what do we know about different assays?

DR BENDELL: Yes, a really interesting case, and it’s funny. We had that case in Florida, and probably about 2 years ago in Brazil I was giving a talk and somebody stood up and said, “I had a patient that came back microsatellite stable and then ended up having microsatellite-instable cancer when I retested.” And so what you have to realize is, these assays, it’s 95%.

You do have a failure rate of some of these assays. And I think what’s really interesting about that case in Florida, as well as the case in Brazil, is that it was the astuteness of the clinician to say, “Wait a minute, this doesn’t really make sense. We’ve got this huge family history. The tumor’s poorly differentiated. They saw lymphocytes. This really feels like a microsatellite instable cancer. Could we be potentially missing the result on this test, and maybe it is time to retest?” And, of course, they saw on the retesting that the patient was microsatellite instable. For some of those cases where you get back something where it just doesn’t jive, it may be worth rechecking.

DR KROP: In a related question, would it be easier to just look at tumor mutation burden as a way to — because that’s at least 1 of the downstream effects of MSI deficiency. And also, would it take into account other potential mechanisms for hypermutation? Is that an easier route?

DR BENDELL: Easier, but it takes a little bit longer and it’s more expensive. It just depends on your approach to testing, right? the most common test that people are going to do when they’re looking at microsatellite status for patients with colon cancer is the immunohistochemistry. They’re going to run it. They’re going to get the assay results in a couple of days. If people are running next-generation sequencing frequently on their newly diagnosed patients, then certainly you could use tumor mutational burden there to take a look at — and thinking oh, then this patient might have microsatellite instability. And probably because you used a different assay, you might see that microsatellite instability pop up on your next-generation sequencing. But I don’t know that I would say that means that you shouldn’t do IHC and next-gen sequence everybody.

DR ANTONARAKIS: I just want to make a point. In a perfect world, you would do all 4 things on every patient. There’re 4 ways to assess MSI/dMMR. The first is to look for protein loss by IHC. The second is to look for a gene mutation in 1 of the 4 genes. The third is to look at the MSI status. And the fourth is to look at the TMB, tumor mutational burden. Your confidence in using a PD-1 inhibitor would go up if all 4 of those were abnormal — if you have a protein loss of MSH6 and you find a mutation of MSH6 that confirms a protein loss and the person has an MSI-high status and their TMB is, like, 30 mutations per megabase.

DR LOVE: How low would it have to be to be suspicious that it wasn’t MSI? How low?

DR ANTONARAKIS: Most of these platforms either just detect it as being MSI high, intermediate or low.

DR LOVE: No, I mean tumor mutation burden.

DR ANTONARAKIS: It’s different for every cancer, but if the TMB is, like, less than 10 mutations per megabase, I would consider that a low likelihood of a PD-1 inhibitor working despite an MSI high.

DR LOVE: You were mentioning some other finding that may be related to sensitivity of checkpoint inhibitor.

DR BENDELL: Yes, polymerase epsilon. That is a gene that can be associated when it’s mutated with high mutational burden with other tumors. People have been looking at the utilization of checkpoint inhibitors for patients who have polymerase epsilon mutations across different tumor types. It’s pretty rare, though.

Emerging role for the assessment of neoantigen load

DR LOVE: Are there other ways to assess sensitivity to checkpoint inhibitors? Kind of like what Ian — is there a shortcut where you can say, oh, like, TMB is less than 10, no matter what’s going on, it’s not going to be helpful. Are there any new markers like MSI that are tumor nonspecific?

DR ANTONARAKIS: The best way to do it, Neil, would be to look at not just the tumor mutational burden but the neoantigen burden. Now, this is complicated, and it will be costly. Not every mutation will result in a theoretical neoantigen, okay? And there’s been new data showing that if you have a frameshift mutation, that creates a lot more antigens than a point mutation. If you have a frameshift mutation, that’s usually caused by a small insertion or a deletion of a nucleotide that is not a multiple of 3. If you have an insertion of 1 nucleotide or a deletion of 2 nucleotides, because a codon is composed of 3 nucleotides, right? If you have an insertion or deletion that’s not a multiple of 3 you’ll get a frameshift, meaning that the open reading frame of the DNA will be shifted, and you get all these jumbled amino acids, and then you’ll get a premature termination codon.

Almost all frameshift mutations result in a termination of the protein prematurely. But before the termination, you get all these amino acids, usually between 10 and 20, all of which are not present anywhere in the wild-type cell. And each of those amino acids, with 1 single mutation, becomes its own theoretical neoantigen.

DR LOVE: Is there an assay for neoantigen levels?

DR ANTONARAKIS: Not a commercial assay, but there are multiple companies working on that. The Johns Hopkins group has coined this term called frameshift burden instead of just tumor mutational burden, because the frameshift mutations are highly antigenic compared to, let’s say, a point mutation, where 1 single amino acid changes and that’s it.

Case (Dr Bendell): A man in his mid-50s with mCRC, Lynch syndrome and MSI-high status achieves a partial response to pembrolizumab as second-line therapy

DR LOVE: Why don’t we finish out with your cases.

DR BENDELL: Absolutely.

This is a 56-year-old gentleman who came to his primary care physician with a history of abdominal bloating, and he was found to be anemic, sent for endoscopy, and the colonoscopy showed a mass in his ascending colon. This is a right-sided tumor. Biopsy showed a poorly differentiated adenocarcinoma. And profiling showed RAS wild type and microsatellite instable. Profiling confirmed Lynch syndrome. He was BRAF wild type, and then we did germline mutation analysis that showed Lynch syndrome.

And of note, this is really interesting, when he was diagnosed with that, he got all his kids checked, and his brother as well. And his brother ended up having Lynch syndrome and was diagnosed with a GE junctional cancer, which was locally advanced. And unfortunately he ended up passing away —

DR LOVE: Wow.

DR BENDELL: — from his treatment, but we did catch another cancer in the family.

CT scan showed omental metastases. Started on FOLFOX/bev. And then he had initial partial response, changed to maintenance and then had disease progression after 10 months. And then he came and was started on pembrolizumab, and he had a partial response to pembrolizumab, continues on pembro today, 2 years in, has seen 1 grandchild be born, 1 daughter get married, 1 daughter’s engaged and continues.

And actually he just told me he was working 100 hours a week at a local auto plant. But now he has decided to move back to regular hours. He can enjoy more time with his family.

DR LOVE: Has he had any autoimmune issues?

DR BENDELL: No. He has done very well.

DR LOVE: Just cruised right through it?

DR BENDELL: He has, knock wood, been cruising right through it.

I have another patient who is very similar to him who has developed some rheumatoid arthritis-type symptoms in her hands. We treat her with occasional steroid packs. But she actually wants to stay on her immunotherapy, and I think that could be another interesting discussion of do you ever stop? And when should you stop? A lot of the trials now say stop at 2 years. But that’s an interesting debate. But these are his scans, and it shows his omental disease and the shrinkage as he was treated with pembro.

Case (Dr Bendell): A woman in her early 60s with microsatellite stable mCRC and a BRAF V600E mutation receives the combination of cetuximab, encorafenib and binimetinib after disease progression on FOLFOXIRI/bevacizumab

DR LOVE: Do you want to go on to the next one?

DR BENDELL: Sure. This is a 61-year-old woman found to have a mass in her ascending colon. Again, routine colonoscopy. Biopsy adenocarcinoma. Microsatellite stable. RAS wild type. BRAF V600E mutation. Scans had abdominal lymphadenopathy. She was started on FOLFOXIRI/bevacizumab. We have seen for the patients with BRAF-mutated colon cancer that we have to be more aggressive with up-front therapy.

And this is data from the TRIBE study, another Italian study, which showed us that the combination of FOLFOXIRI plus bev is better for the patients with BRAF-mutated colorectal cancer than just FOLFIRI plus bev alone.

DR LOVE: That’s just 1 of the reasons I was surprised that not everybody got BRAF up front.

DR BENDELL: Yes, Yes.

DR LOVE: I guess not everybody in practice buys into that.

DR BENDELL: Yes, but you really should. And people worry about giving the kitchen sink approach up front, but if you look at the trial that you give the kitchen sink for 6 months and then you go into maintenance, there is a quality of life that does come there.

She got switched to maintenance therapy, then had stable disease and then progressed, and then she was started on the combination of cetuximab, encorafenib and binimetinib, had a PR to that regimen that continues after 6 months. But very interestingly, her treatment is complicated by side effects that we see with this triplet combination, of the acneiform rash, nausea and loose stools. And this is response in her lymphadenopathy as well as her primary tumor as early as 2 months into treatment and then at 4 months.

Case (Dr Bendell): A man in his early 40s with HER2-positive mCRC receives trastuzumab and pertuzumab on a clinical trial after disease progression on FOLFIRI/bevacizumab and attains a partial response

DR BENDELL: Case number 3 is a young gentleman who presented with hematochezia, right-sided abdominal pain, had CT scans for the pain, showed lung nodules, multiple liver lesions, distal sigmoid thickening, some lymph nodes. Colonoscopy showed this rectosigmoid junctional mass. And liver biopsy was done that showed this moderately differentiated adenocarcinoma with a RAS mutation. Got FOLFOX plus bev, then maintenance. Progressed after 10 months, then moved to FOLFIRI plus bev, with initial stable disease, then progressive disease. And then had molecular profiling by next-generation sequencing done, as well as HER2 IHC, and was found to be HER2 3+ by IHC and was started on trial with trastuzumab and pertuzumab and had a PR which continues 6 months later. And these are his baseline liver lesions on scan, and here’s the liver lesions after the trastuzumab and pertuzumab, and you can see that difference there.

DR LOVE: Any tolerability issues? Any diarrhea?

DR BENDELL: Very mild. Very mild, if any. Doing very, very well.

DR KROP: The diarrhea from trastuzumab and pertuzumab is much more pronounced with chemotherapy than without. Once they get to maintenance HP, it’s usually pretty well tolerated.

DR LOVE: Yes, I just recently started to hear that. I mean, I don’t know, I’m sure everybody’s known it except me for a long time, but I thought that was really interesting.

Any other questions?

DR KROP: Is that patient HER2 amplified?

DR BENDELL: All we checked was the IHC. We ended up not doing the FISH, because we just tested the IHC and saw that positivity.

DR KROP: On the next-gen sequencing, does it show up?

DR BENDELL: No.

DR LOVE: Why were you asking?

DR KROP: Just curious. Because again, for patients who aren’t having everything done, how much can you get by with just doing the next gen?

DR LOVE: Right.

DR KROP: Because if it’s highly amplified, then it should show up.

Novel HER2-targeted therapies under evaluation for patients with mCRC and HER2 alterations

DR LOVE: Maybe if something hot comes up, you’ll see more research in this. There’s a fair number of people out there.

DR BENDELL: It’s funny, because when you look in the Phase I unit, where I spend a lot of my time, there I think one of the hottest areas right now is development of new HER2 therapies. Novel HER2 antibody-drug conjugates, novel HER2 bispecifics that hit the HER2 receptor in 2 different epitopes to really grab onto the HER2 receptor, where we’ve seen some beautiful data — the antibody-drug conjugates, the DS-8201 showing some really beautiful disease response and control with breast cancer as well as some gastric cancer and colon cancer.

DR LOVE: And those were, like, IHC 1+ and 2+.

DR BENDELL: And yes, even in the low response — what we’ve been trying to do is tell folks to actually look at IHC who have the ability to get those patients into clinical trials, because the 1+ and 2+ patients are having potential treatment options now. And we even have HER2 CAR T that is open at our center too.

DR LOVE: HER2 CAR T.

DR BENDELL: The way that it works is that the CAR T is actually — it creates an FC receptor binder. And what you do is, you infuse the cells, but you use trastuzumab infusions to turn on.

DR LOVE: Wow.

DR BENDELL: What happens is, the CAR T only binds to the tumor. When the antibody is infused, the CAR T hits the FC portion of the trastuzumab, and the trastuzumab hits the tumor cell. When the trastuzumab’s not present, the CAR T just floats around waiting for the antibody infusion to come.

DR LOVE: Wow. You’re doing that?

DR BENDELL: Yes.

DR LOVE: Have you seen any responses?

DR BENDELL: There’re only a handful of patients on trial. The first patient was a breast cancer patient who it may have been a little too late in her disease course to show the response. And, actually, we just lymphodepleted our first HER2-positive colon cancer patient. She should be infused with her cells probably in about 2 weeks.

DR LOVE: And you get the cytokine release neurologic stuff? Same thing?

DR BENDELL: With the patient that we treated, no. The idea is that you shouldn’t, because the CAR is only activated at the time that you infuse the trastuzumab.

DR LOVE: Wow.

DR BENDELL: Super exciting technology going on.

Analysis of the gut microbiome and strategies to enhance the immune response to cancer

DR LOVE: I have to ask, any time I hear the word microbiome I can’t let it go by. I’m just kind of curious, anything new in terms of assays to study the microbiome that you guys are looking at?

DR ANTONARAKIS: It’s the wild west. Everybody’s got their favorite species. Some of them are supposed to be proimmunogenic. They stimulate the antitumor immune response. Some dampen the antitumor immune response. The one that we’ve been studying at Johns Hopkins is called akkermansia — A-K-K-E-R-M-A-N-S-I-A. Akkermansia.

DR LOVE: But, I mean, is it just a case of getting some fecal samples and trying to grow it?

DR ANTONARAKIS: No. It’s a PCR-based test to look at the genome of the organism. You’re looking for the genetic material from the bacteria.

DR LOVE: You do that on —

DR ANTONARAKIS: On a rectal swab.

DR LOVE: On a rectal swab.

DR ANTONARAKIS: Yes, yes. I don’t think that’s ready for clinical decision-making yet.

DR LOVE: Hmm. It sounds interesting, though. All right.

DR KROP: But as just said, it’s incredibly exciting stuff, and I think all of us who are doing immunotherapy studies are collecting microbiome samples —

DR LOVE: But is that the way you’re doing it, with rectal swabs?

DR KROP: We do stool samples and then do the sequencing off of the stool samples.

DR BENDELL: And there’s even companies out there that are already automatically trying to engineer bacteria to give them in combination with immunotherapies. We have one with a bifidobacteria. But the other piece is the fecal transplants. If you take the stool or the microbiome, I’ll say it that way, from the patient who responds to immunotherapy, give it to a patient who’s not responding, they’ve seen anecdotally patients start to respond again. And this is where they give it in the capsules that the patient takes orally and then maybe changes outcomes.

Genetic alterations assessed in routine practice, including MSI status, BRCA mutations and AR-V7 variants; role of multiplex testing

DR LOVE: Let’s move on now and talk about prostate cancer. And again, just to start out in terms of our survey, we asked people whether they’re doing MSI testing in metastatic prostate cancer, and they go, “Yes. It has definitely reached practice.” Most of them do it at first diagnosis, but quite a few wait, which you can understand. Particularly, some of them wait for castrate-resistant disease. But in any event, they are testing.

Then we asked them, “The last patient in your practice with MSI-high disease that you treated, did they get a PD-1 antibody?” Most of them say, “Yes.”

You can see generally this is third- or fourth-line therapy, and yet in spite of that, 12 out of 17 of the patients responded. Another question we asked them was whether in general they’re doing multiplex genomic testing in patients with metastatic disease, and it looks like about 75% are. Again, a little bit of a spectrum about when, but most of them, a majority of them, are testing.

We asked them about the last patient where they did a tissue-based multiplex genomic assay. And you can see that the majority of patients did not have it affect their treatment decision, but on the other hand, maybe a quarter of them did. It’s not a small number. And, interestingly, most of the findings that they felt were going to affect their treatment were related to BRCA.

We asked them whether or not they responded, and again, we can look at this later in terms of what it was, and most of them responded. But I think more importantly, here’s the blood-based assay. I think even a lower number that they were felt were productive. Again, a couple of BRCA patients.

A couple of people responding. We asked them whether or not they do germline BRCA testing in patients with metastatic prostate cancer. A pretty basic question, that when we ask at our meetings, I would say the vast majority of people say, “Yes.” But now we’re going out to a random sample, and interestingly there’s not a small number who say no, which is interesting. there’s a spectrum of, I guess, people being aware of the potential issues of PARP inhibitors.

We asked them about the last patient in their practice with metastatic prostate cancer who had a BRCA mutation that was treated. And as you can imagine, most of that was PARP inhibitors. It was generally the third- or fourth-line therapy. And, interestingly, getting back to our discussion before about the MSI, 25 out of 29 people, from the point of view of the oncologist, responded.

Now again, that might not be a RECIST response. It might be a clinical response, but from their point of view, it was a response.

I thought you’d be interested in this, Emmanuel, AR-V7 testing. We can talk about where that is today, but we asked them whether they’re generally utilizing it, and about a third of them are. And for the ones who are, they’ve used it, it looks like quite a bit. Eight times in the last couple years. And we asked them a little bit about how it affected what they did, and we’ll talk a little bit more about where that is and where it’s heading.

The final thing we asked them, as we did for all 3 of these situations, was the last patient with prostate cancer who died in their practice. You can see the median age is higher than it was for colorectal cancer, 78. At least in this group, the median survival was greater also, more than 3 years. And half of these patients had multiplex genomic testing as part of their care.
We asked whether it affected treatment, and for the majority of them it did not affect their therapy. I guess before we start getting into the data, I’m just kind of curious, Emmanuel, anything about these data that jumped out to you?

DR ANTONARAKIS: I think it was kind of as expected. The BRCA mutations and the MSI were the 2 most actionable things.

DR LOVE: Anyone else have any thoughts or comments about what we just saw there in prostate? Andy, any thoughts?

DR MCKENZIE: I was just curious, the germline BRCA testing, is that what you would expect in — it says here that most patients got it when they were discovered to have castration-resistant disease. That’s when they went and confirmed that. Is that what you would expect there?

DR ANTONARAKIS: The current NCCN guidelines, and for the past 2 years, have recommended that every patient with metastatic disease, whether it’s hormone sensitive or castration resistant, should undergo germline testing, because there’s a 12% chance of having a positive germline HR mutation, not just BRCA2, but BRCA2 plus 10 other genes. Yes, I was a bit surprised. It seemed low.

DR LOVE: I guess you could say, are you testing for the patient or the family? If it’s the patient, I mean, sure. They could go in a clinical trial, but they typically wouldn’t get a PARP inhibitor, I don’t think, even if they’re BRCA germline, as first-line therapy. What do you think?

DR ANTONARAKIS: It wouldn’t be first line. It would have to be off label, because we don’t have any PARP inhibitors right now for prostate cancer which are FDA approved.

DR LOVE: But also, I mean, just putting aside the issue of FDA approval. I mean, even if you did have it approved, I don’t think you would use it before you use androgen deprivation, would you?

DR ANTONARAKIS: You would not, no.

DR LOVE: You could make the argument for practical purposes it doesn’t matter, but with that high though of germline BRCA, you have the family to think about. It might be a couple years, and a family member could develop a cancer and die of it.

I remember I was talking to you, I think I was interviewing you once, and you were saying something about it shouldn’t be called BRCA. It should be called PRCA, prostate.

DR ANTONARAKIS: It’s just as prevalent in prostate cancer as it is in breast, but it was discovered in breast cancer first. It just happened to be called B-R-C-A instead of P-R-C-A.

Presentation (Dr Antonarakis): Biomarker assessment in mPC; detection of DNA damage repair abnormalities and prediction of benefit with PARP and PD-1 inhibitors

DR LOVE: Right. With that, why don’t we go through some of the slides, particularly related to this particular issue.

DR ANTONARAKIS: I’m going to cover a few issues first, just 1 slide about the landscape right now for hormone-sensitive and castration-resistant prostate cancer, then talk about the prevalence. I’m going to focus on DNA repair genes. There’re certainly other genes that are actionable, but I think for this talk I’ll focus on those. And then specifically home in on the BRCA1, 2 and ATM mutations as well as the mismatch repair mutations, less on the latter, more on the BRCA and ATM. And then to talk about the 3 PARP inhibitors that we have data for, although I must say at the beginning we don’t, at this point, have an FDA-approved PARP inhibitor for prostate cancer, and then some conclusion statements.

The treatment landscape for metastatic hormone-sensitive prostate cancer and metastatic castration-resistant prostate cancer has changed a lot. I joined faculty in 2010, and at that point in time we had 1 drug, docetaxel, and we now have 8 drugs plus 1 more, which is pembro. We have a total of 9 drugs, 1 of which is only approved for the MSI high.

And these boxes here represent the clinical states. We tried to think of prostate cancer in terms of these states rather than staging. And all the green ones are the hormone sensitive, and all the orange ones are the castrate resistant.

Typically what happens is, a patient gets diagnosed with localized disease. They have primary therapy, either surgery or radiation. Some of them may have a relapse, which is detected by PSA rising alone. And then depending on whether you decide to use androgen deprivation therapy for those rising PSAs, you can either get a nonmetastatic castrate-resistant state, whereby you use the hormone therapy in someone who doesn’t yet have mets and they develop castration resistance but without mets. Or you can wait until they develop their natural history and develop that first met, and then you treat them with ADT for the first time.

Let’s start with the bottom one. For the nonmetastatic castration-resistant prostate cancer, these are patients who have a biochemical recurrence. They get ADT, then they have castration resistance, no mets. In the last 2 years we’ve had 3 new options for these patients, which are those 3 AR antagonists that you can see there — apalutamide, enzalutamide and darolutamide. They’re basically all versions of the same thing. And if a patient presents with metastatic hormone-sensitive disease, in other words, they have not had any hormone therapy yet, they can either get hormone therapy alone or hormone therapy plus docetaxel up front or hormone therapy plus 1 of 3 novel hormonal therapies.

Eventually, all patients do progress to both metastatic and castration-resistant disease, and all the androgen receptor therapies are still available, plus 2 taxanes. One is docetaxel. The second is cabazitaxel. We also have an immunotherapy called sipuleucel-T, which is an autologous dendritic cell-based immunotherapy, FDA approved in 2010, and then radium-223, which is a bone-targeting radionuclide. And then at the bottom there, pembrolizumab for the 2% of people that are MSI high, low proportion.

Landscape of DNA damage repair mutations in newly diagnosed and metastatic prostate cancer

DR ANTONARAKIS: Let’s now talk about the frequency of DNA repair gene alterations in localized disease and in advanced disease. There are 2 main ways that a cancer cell repairs its DNA damage.

One is by repairing 1 strand at a time. That’s called single-strand DNA repair. And there are 3 main pathways. The most important for diagnostic and therapeutic indications is the MMR pathway, which is highlighted in orange. And those are the 4 genes that we all know, and if there’s a germline mutation in those genes, they have Lynch syndrome.

And then there are 3 main ways that a cell can fix double-strand DNA breaks. The most important, by far, is homologous recombination, which is the orange one, and the BRCA1 and 2 genes, perhaps ATM and a number of other genes are involved there.

In localized prostate cancer — this is one of the first studies. This was from the TCGA. Again, these are men that have nonmetastatic disease. They just have a prostatectomy, and the genomic sequencing is done from the prostatectomy. And in this study, 8% of those patients with localized disease had a biallelic inactivation. This meant that both copies of the DNA repair gene were mutated or inactivated.

And that’s important, because in order for PARP inhibitors to work, and we’ll talk about this later, at least in theory the biallelic inactivation is very important, because if you have 1 normal copy of the gene and 1 abnormal copy of the gene, the normal copy can rescue the abnormal and make PARP inhibitors ineffective.

But 8% of guys with localized disease, just keep that number in mind, 8% to 10% have a mutation. And as you can see, BRCA2 is the most common, BRCA1 and ATM are number 2 and number 3.

In metastatic castration-resistant disease, the incidence is about double, about 20% to 25% of patients that have metastatic castration-resistant prostate cancer, when they have a biopsy of the metastatic site, will have 1 or more DNA repair gene mutations. And in the metastatic setting, the most common one was BRCA2, many of which are germline as well, and followed by ATM. Eight percent to 10% in localized disease, 20% to 25% in advanced disease — about twice as common in advanced disease.

Now, this is the paper we were discussing previously with germline testing. This was a study published in the New England Journal now 3 years ago, where they took almost 700 patients unselected for family history, unselected for Gleason score, just 700 consecutive patients being seen at 7 different institutions. And they just did germline testing on all of them. And the surprising thing, and I believe the reason why this was published in the New England Journal, is because 12% of men had 1 or more germline mutations in 1 of the DNA repair genes. That’s 1 in 8 men.

Typically on a Tuesday I see 24 to 25 patients. Three of those are going to have a germline mutation. That’s a fairly high prevalence.

As you can see on the pie chart, of those who tested positive, 44%, almost half, were BRCA2. And then if you add ATM, CHEK2 and BRCA1, you get 75%. If someone asks you what are the 3 to 4 most common genes that are mutated in the germline, you would say BRCA1, 2, ATM and CHEK2.

In localized prostate cancer only 4% of these patients have a germline mutation. This does appear to be enriched in men that have metastatic disease, either de novo or who subsequently develop metastatic disease.

Rationale for the use of PARP inhibitors for prostate cancer with BRCA or ATM mutations

DR ANTONARAKIS: Now I want to spend 1 slide talking about this concept of synthetic lethality. People might have heard about it. It seems a bit confusing. And this is the mechanism by which PARP inhibitors are thought to work in patients that have either BRCA1 or 2 mutations.

And the term was coined actually in 2005, with 2 consecutive Nature papers, and there’s a figure here that tries to boil it down. The concept about synthetic lethality is that 1 plus 1 equals kill. You need to cripple both the single-strand repair pathway and the double-strand repair pathway in order for that cell to die. And, of course, the PARP inhibitor cripples the single-strand pathway.

If you look at the cell on the left, which is in green, this cell has an intact BRCA1 and 2 function. Their double-strand repair is working beautifully. If you give this patient a PARP inhibitor, you’re going to cripple their single-strand pathway, but the double strand will be able to rescue that function. The cell will not die. The drug will not work.

If you go to the right-hand side, with the red cell, this is a patient that has 2 hits in BRCA2. You can see those 2 red marks, 1 in each allele. This patient has a completely inactivated BRCA2 gene. His homologous recombination function, ie, double-strand repair, is wiped out. If you then cripple his single-strand repair pathway by giving a PARP inhibitor, that cell has no way to repair its DNA damage and has no choice but to die due to catastrophic DNA damage that accumulates.

In the middle scenario, and this is where it gets a little bit tricky, this is an example of a patient that has a monoallelic BRCA alteration. What that means is, we talked about this at the very beginning, this is a person who has 1 mutated copy of BRCA2, but the second one is wild type. This patient may not respond to a PARP inhibitor because his double-strand repair pathway can be rescued by the normal copy of the BRCA2 gene.

And this is actually interesting, because when we get these Foundation Medicine reports or Caris reports, they don’t typically report what’s going on with the second allele. They just report what’s going on with the first allele. And this is an unmet medical need for us to make decisions. But that’s the theory of synthetic lethality.

Genetic testing of primary and metastatic tumors for alterations in the DNA repair and androgen receptor (AR) signaling pathways

DR LOVE: Maybe you can stop there and see if there are any questions.

DR BENDELL: When you see the difference between primaries and metastases, what’s your opinion then on testing, right? Because you want to have baseline testing on everybody to find microsatellite instability, et cetera, but do you then, in your practice, look further down the line?

DR ANTONARAKIS: Right, that’s a great question. And the question that arises, or reading between the lines is, when do these mutations occur? Are they occurring in the primary, but those patients are more likely to then develop mets? Or is it that they are sometimes not found in the primary, and then they are then acquired in the metastatic clones? And the only way you can answer that is by taking the same patient and doing a paired biopsy of the primary site and a metastatic site several years later and looking for concordance or discordance, for example, in BRCA2 or ATM. And those types of studies are hard to do, because they require multiple biopsies, although they don’t have to be done synchronously.

And the long story short is that most of these BRCA1, BRCA2 and even ATM mutations are early truncal events that are present in the primary and are not necessarily acquired over time in the metastatic clone. The prevailing hypothesis right now is that if you find a BRCA mutation in the met, it was probably there in the primary. If you don’t have a met and all you have is a primary and you do the sequencing on that, you’ll find 95% of them.

DR MCKENZIE: What if you take a step outside of the DNA damage repair pathway? What about other mutations in prostate cancer, for instance in the androgen receptor, and would you recommend, to Dr Bendell’s question, repeat testing in the context of acquired resistance in those types of scenarios?

DR ANTONARAKIS: Right, you raise a very important issue. Unlike the case with DNA repair mutations, AR alterations become more and more prevalent with time, and they become more and more prevalent with the increasing number and types of prior hormone therapies.

In a patient who has received, let’s say, 1 androgen deprivation therapy, he may or may not have an AR mutation. Some of them will have AR amplification. Some of them might have AR splicing variants like AR-V7. By the time you receive the second-line hormone therapy, let’s say abiraterone, third-line hormone therapy like enzalutamide, the chance of finding an AR amplification is about 70% to 80%. The chance of finding an AR mutation which activates the ligand binding domain, which causes promiscuous activation by other ligands other than testosterone, is high. And the chance of getting these constitutively active splicing variants which do not even require the ligand to stimulate transcription of genes also goes up. That’s a very important consideration. And the good thing is, you can do that very nicely from liquid biopsies.

Clinical significance and genetic detection of AR splice variants

DR LOVE: Can you explain a little bit more about what the splice variants are? And what the clinical significance at this point is, if any?

DR ANTONARAKIS: Right. Splicing variants are a big interest of mine, as you know, Neil. I’vebeen working on this for about 5 years now. And these are interesting variants where if you sequence the gene of the androgen receptor, you don’t find the mutation. The gene itself is wild type. What happens is, after the transcript is created, of course in the pre-mRNA there’re introns and there’re exons. And the introns are supposed to be spliced out, and all the exons are supposed to basically join together.

The androgen receptor gene has 8 exons and has 7 introns between them. Now, once in a while, one of those introns does not get spliced out. It is retained, okay? And typically that really shouldn’t matter that much, except some of these introns have cryptic exons within them. And some of those cryptic exons can actually lead to the coding of an altered protein that has a premature stop codon.

What these splicing variants, especially AR-V7, lead to is, they lead to an AR protein product which is truncated, the first half of the protein, which we call the end terminal, is intact, but the right-hand side of the protein, which we call the C terminal, is missing. The reason why that’s important is because the androgens bind to that C terminal. On the face of it, you would expect these AR variants like AR-V7 to be inert, inactive, but it’s the exact opposite. They’re actually constitutively active, which means that they can function as transcription factors and do good things to the cancer, bad things to the patient, even without testosterone binding to them. And the corollary to that is, none of the antiandrogens work in these patients, because the antiandrogens either bind directly or indirectly to that ligand binding domain, which is also missing in those splice variants. These patients do better with chemotherapy, like taxanes.

DR LOVE: And is it actually playing out in practice? Are there any assays now that are validated and clinically useful?

DR ANTONARAKIS: Yes, most of the assays are using liquid biopsies, and they’re usually using circulating tumor cells. Now, one interesting thing about these assays is that they’re not DNA based. You can’t get a Guardant360 or even Foundation Medicine, for that matter, and find an AR-V7. You have to do RNA sequencing from circulating tumor cells or PCR on RNA, which is called RT PCR, reverse transcription PCR.

The second way that you can find them is, you can look for the truncated protein in circulating tumor cells. You can develop an immunohistochemical or immunofluorescence antibody that binds to the sequence in the protein, which is only found in the variant but not found in the wild type.

And there’s a company, and they have the first and currently only commercial assay for AR-V7, and I think that’s the one that most of your responders actually ordered.

DR LOVE: And do you think the data supports using it right now in clinical practice outside a trial setting?

DR ANTONARAKIS: I think so, but it also depends on the context of use and lines of prior hormone therapies. If you have a patient who is hormone sensitive who has never received any androgen deprivation and has never received any antiandrogens, the chance of finding an AR splice variant in this patient is close to 0%, because their AR has never been inhibited. That cell has no reason to make an alternative form of AR to continue to survive.

Once a patient has received enzalutamide or abiraterone, but not both, the chance of finding an AR splice variant goes to about 15% to 20%. And those patients, if they’re positive for AR-V7, should go to a taxane next, like docetaxel. Once that patient has received both enzalutamide and abiraterone, then that person’s chance of having an AR splice variant like AR-V7 is, like, 30% to 40%. The more and the stronger the prior lines of hormone therapy, the more you get this adaptive response where the cell wants to maintain its addiction to AR, but it can’t use the canonical AR. It uses a deviant form of AR, an aberrant form, which is the AR-V7.

Results of the CARD trial: Cabazitaxel versus enzalutamide or abiraterone for patients with mCRPC previously treated with docetaxel who experience rapid disease progression on an AR-targeted agent

DR LOVE: The clinical situation you hear this being discussed in is the patient who’s had either enzalutamide or abiraterone for metastatic disease. They may have responded. Now they’re progressing. The question is, do you go to the other one? Or do you go to, say, chemotherapy? However, there was just a trial presented at ESMO where all those patients did better with cabazi. It was randomized between that and just giving them the chemo. Do you think that is going to push away the clinical relevance of this? Or do you think it’s still relevant?

DR ANTONARAKIS: Well it may. That study that you mention is called the CARD trial. It’s actually published now in the New England Journal as well. And this was a study where basically all patients have previously received abiraterone, but they had all progressed within 12 months. It was a little bit artificial and stacking the deck in favor of cabazitaxel.

DR LOVE: Oh, I didn’t realize that they progressed within 12 months.

DR ANTONARAKIS: Yes, that was an eligibility criteria.

DR LOVE: Huh. That’s, like, not typical though, right?

DR ANTONARAKIS: It’s not typical.

DR LOVE: Wow.

DR ANTONARAKIS: The median PFS for first-line abi or enza is 18 months.

DR LOVE: Wow, I did not realize that.

DR ANTONARAKIS: It was a good study. And it was really looking at those patients that have primary AR therapy resistance. What should they do next? Should they get a second AR agent, or should they get a taxane? The caveat, which I think the authors may have pointed out in their discussion, is that these patients were basically all relatively AR insensitive. I have many patients in my clinic who have 3-year responses or 5-year responses to abiraterone. And this patient probably will respond to enzalutamide on the back end if they’ve had such a deep and durable response to the first one.

DR LOVE: And I don’t imagine they have tissue or blood from that study.

DR ANTONARAKIS: They do, actually. They have circulating tumor cells in a subset —

DR LOVE: Huh.

DR ANTONARAKIS: And they do plan to prospectively evaluate AR-V7.

DR LOVE: Wow, that is really interesting.

Efficacy of olaparib for mCRPC

DR LOVE: Let’s talk about PARP inhibitors in prostate cancer.

DR ANTONARAKIS: Alright, the first one that made a splash was olaparib. And this is now 4 years ago, January 28, 2016, the FDA actually came up with this breakthrough designation for olaparib. And at that time there was interest in both BRCA1 and 2, as well as ATM being on the label. And again, a breakthrough designation does not mean that there’s an approval. It just means that the FDA is willing to consider an accelerated approval if there’s a positive Phase III study in those patients.

And that breakthrough designation was based on this 49-patient study published in The New England Journal of Medicine by Joaquin Mateo and Johann de Bono. And this was a genomically unselected patient population. They all had to have metastatic castration-resistant prostate cancer. They all to have a baseline metastatic biopsy. But then they were treated with olaparib regardless of their mutation status. And then in a post hoc fashion there was a whole exome sequencing done.

And the interesting thing about this study was that virtually all the responses were in those patients who had 1 or more somatic or germline homologous recombination mutations. Sixteen of the 49 patients had a mutation. Those are clustered on the left.

And in those 16 patients the response rate was 88%, which is very, very high. This was a composite response based on either PSA dropping or tumor shrinking or both. In the biomarker-negative group, in other words, those patients that had no mutations, only 6% responded.

And if you look at the progression-free survival and overall survival, again, this is a single-arm study split by whether or not the patients had or did not have a homologous recombination mutation. And the blue line represents the patients that did have a mutation. And the progression-free survival and overall survival was greater.

This led to the FDA breakthrough designation. This was a nonrandomized study. There was no control arm.

Now, one of the things that we became interested in at Hopkins is to distinguish the effectiveness of these drugs in BRCA1 and 2 versus ATM. And after that New England Journal of Medicine paper was published, we and others began to use these drugs off label.

And we published this paper after we had done this 23 times, with our first 23 patients. These are all off-label use of olaparib. And the patients in light or dark blue are patients that have BRCA2 or BRCA1 mutations. And the 5 patients in the yellow or orange are the patients that had ATM mutations. And as you can see very clearly on the waterfall plot showing PSA responses, all the PSA responses that we saw were in the BRCA2s and BRCA1s, and almost all the patients with ATM had basically progression as their best response. And if you look on the right-hand side, which is the Kaplan-Meier curve of rPFS, you can see the PARP inhibitors really work well in the blue line, which is the BRCA1 and 2, but they all had progressed within 3 months’ median with the ATM group. We actually had a signal here, which was a bit sobering and somewhat depressing, that even though the FDA included ATM on their breakthrough status for this particular drug that it may not pan out.

Now, fast-forward to the pivotal Phase III study, which was called PROfound. This study looks a bit complicated. There were basically 2 cohorts. Cohort A was the primary cohort of interest, and this was the one that the FDA was going to either approve or disapprove the drug on. And cohort A was patients that had either BRCA1, BRCA2 or ATM mutations. You have to have 1 of those 3. It could be germline. It could be somatic. It didn’t matter. And the central lab that did this was Foundation Medicine, and they were then randomized to receive either olaparib as a monotherapy or physician’s choice between abiraterone or enzalutamide, either one.

There was also a second cohort, which was cohort B, that allowed inclusion of patients with 12 other homologous repair genes, which are listed across the bottom, BARD1, BRIP1, CDK12, et cetera, et cetera. And in those patients, the same thing happened. They were randomized either to receive olaparib or physicians’ choice of AR therapy. The primary endpoint, and this is very important, was only for cohort A. And the rest was exploratory. The primary endpoint was rPFS, and it was met.

This was presented at ESMO by Maha Hussain, has not yet been published, and there was at least a doubling of median radiographic progression-free survival with a hazard ratio of 0.34, highly statistically significant. And the overall survival also trended in favor of olaparib, although it did not meet the statistical threshold on this interim analysis. But pretty compelling results there.

One of the things that I found the most interesting was the breakdown in the progression-free survival according to the gene. If you look at BRCA2, you see a very big and profound difference favoring olaparib, with a 10.8-month median PFS versus 3.4 months. If you look at ATM, you actually see no difference. You see 5.3 versus 4.7, not statistically different. What was even more surprising to me, no clear difference in BRCA1, either.

Although we lump these genes together because they sound the same, they’re actually functionally different. And if I was interpreting this, I would say that the main benefit was driven by primarily, or maybe even exclusively, the BRCA2 patients.

Now, there are these rare genes, RAD51B and RAD54L, which if you speak to a DNA repair biologist, he’ll tell you that these are very critical to homologous repair function. And there did seem to be a very big difference in those patients favoring olaparib, but those genes are quite rare in prostate cancer, more like in the 1% range. Maybe it looks like, from this pivotal Phase III trial, that the majority of the benefit is in BRCA2.

It’ll be interesting to see what the FDA says about that, because again, their primary endpoint encompasses all 3 genes. it would kind of be unprecedented for them to change their mind and say, “We’re only going to approve this for BRCA2,” although they have every right to do so. Or they could say this might be approved for all 3 genes. We don’t know what’s going to happen.

Available data with rucaparib, niraparib and talazoparib for mCRPC

DR ANTONARAKIS: The second one is rucaparib, and in October of 2018 there was another breakthrough designation for rucaparib in prostate cancer. Interestingly, by now the FDA had gotten a bit smarter, and the data had been emerging, and they excluded ATM on this breakthrough designation. This one specifically says, “Treatment of BRCA1 and 2 metastatic CRPC.” Maybe serendipitously, this study might be more positive because they’re excluding the ATM patients. Rucaparib is a stronger PARP1 and PARP2 enzymatic inhibitor, but it’s not a stronger PARP trapper. The significance of PARP trapping in humans is really unclear.

And they have a Phase II and a Phase III program. Their Phase II program is called TRITON2. This is a third-line study. Patients have to have received both an AR therapy and a chemotherapy. TRITON3, which we don’t have results on, is a second-line study, and patients have received only an AR therapy.

And the gene list at the bottom, which is overlapping with but not identical to the gene list with olaparib, includes BRCA1, 2, ATM, plus 12 other genes. Again, these are not identical, and there’s no uniformity here.

These are the preliminary interim data of the TRITON2. This is a Phase II study. They have to have received an AR therapy plus a chemotherapy. This shows a waterfall plot for objective response. We have to remember that not all prostate cancer patients have soft tissue disease. This only incorporates those patients that do. And while this looks quite impressive to the naked eye, if you look at the top table, which is hard to see, most of the benefit comes from BRCA1 and 2. You can see an overall response rate of 44% in the BRCA1 and 2, only 9% in the ATMs, 0% in the CDK12s, 0% in the CHEK2s.

The pattern is repeating itself that again, with a different PARP inhibitor, the same overall conclusion that the majority of the benefit is coming from BRCA1 and 2. This study did not separate those two. We don’t know if that’s actually being driven by 2 rather than 1. But what is clear is that ATM, CDK12 and CHEK2 are not responding.

And that is true as well when you look at PSA responses. Again, when you look at this waterfall plot, you see that most of the bars are going downwards. But PSA responds, BRCA1, 2, 52%. ATM only 3%. That’s pretty low. CHEK2, 14%. CDK12, 7%. Once again, the chance of having a PSA response to rucaparib as a second PARP inhibitor is greatest in BRCA1 and 2, less in ATM, CDK12 and CHEK2.

And then the third one with the breakthrough designation October 3, 2019, niraparib. Again, ATM was left off of the breakthrough designation language. Again, just BRCA1- and 2-mutated prostate cancer. This can be germline or somatic.

And again, this PARP inhibitor, not only is it a strong PARP1, 2 enzymatic inhibitor, it also causes the so-called PARP trapping, which means that the PARP gets stuck on the DNA, is unable to dissociate, and supposedly induces more double-strand DNA breaks by getting trapped on the DNA. Nobody knows if that’s important in humans, but the drug companies think it is.

And the study’s called GALAHAD, and the interesting thing about this study is the gene list on the bottom. In this study, the gene list was a lot more restrictive. It only consisted of 6 genes. BRCA1 and BRCA2 were the ones that were part of the primary analysis, and then there were 5 additional genes. ATM, as you can see, is in there.

Now, one distinction with this study is that the investigators required a biallelic inactivation. Two copies of the gene had to be mutated or lost. In the other two studies, that was not required. This study required screening more patients to find those that were biomarker-positive, because a single mutation, let’s say, in 1 allele of BRCA2, even if it was clearly pathogenic, was not enough. They had to have 2 hits.

And here the orange bars are the BRCA1/2, and the blue bars are the non-BRCA. And in this study, they did not tell you which patients were which bars. But as you can see, the best 7 or 8 responses. This shows the objective responses were all in the BRCA1/2s. Very few responses in the blue patients, which are the non-BRCAs. Same when you look at PSA, with 1 exception of this blue guy who may have had some other gene mutation that was undisclosed. Almost all of the other more profound PSA responses were in the BRCA1 and 2.

Three different studies coming almost to the identical conclusion, which is that in prostate cancer, at least, the benefit is largely driven by BRCA2, perhaps 1 — we’re not sure about that. Not ATM, and probably not the other genes.

My conclusions from this are that somatic and germline DNA repair gene mutations are common in prostate cancer. They affect about 8% to 10% of localized disease, 20% to 25% of advanced disease. These homologous repair mutations, primarily BRCA2, may sensitize to PARP inhibitors. There is also data that I did not show that these mutations may also sensitize to platinum drugs, which we don’t typically use in prostate cancer, and also radium-223, which is a double-strand DNA damaging radiopharmaceutical product. If you have a drug that damages DNA and your cancer cell is not able to fix the DNA damage, it might be more effective. And olaparib and rucaparib are likely to be the first PARP inhibitors to receive FDA approval. Neither one has yet received FDA approval in prostate cancer. These are still off-label use. And there’s a third one underway, which is niraparib, and now there’s a fourth one called talazoparib.

And the unanswered questions are whether these drugs work better in germline versus somatic. The hypothesis would be, if you have it in your germline, that means that every tumor cell has it. If you have it in the somatic, only a fraction of tumor cells might have it. And the second thing is, if you have a germline first hit, you’re more likely to have a somatic second hit, that biallelic thing I kept talking about. Whereas if it’s somatic, a lot of the somatics are just monoallelic. The word on the street is, and we have not seen the data yet, that you are probably more likely to respond if your first mutation is a germline mutation, although that has not been proven.

Role of radium-223 dichloride for patients with mPC; novel combination approaches with radium-223 under investigation

DR LOVE: Before we get into your cases, I want to give you all a chance to question Emmanuel. To start off, what did you say about radium-223 in this?

DR ANTONARAKIS: Radium-223, when I talk to my patients about it, I call it a liquid radiotherapy, and then they look at me as though they’re very confused. The reason I call it a liquid radiotherapy is, it’s an injectable drug. It’s injected into the vein, and it’s a radioactive particle, which is radium-223. Radium is a calcium mimicker. If you go back to your tenth-grade chemistry periodic table, group 2, it’s the alkali earth metals. Calcium, strontium, samarium and radium are in the group 2. All those have very similar chemical structures.

When you inject radium-223, which is radioactive, into the blood stream, the bone can’t tell the difference between radium and calcium. The bone thinks that it’s calcium, and it uses calcium to bind into areas of bone where there’s basically osteoblastic activity — the tumor cells are eating away at the bone. Calcium is going in to replace those pockets left by the tumor cells, but now radium is being incorporated as well. Radium is an alpha particle. There’s 3 types of radiotherapy particles. Alpha, which has the shortest wavelength but is the most potent. Beta, which has an intermediate wavelength and intermediate potency. Gamma, which goes the furthest but is very weak.

Radium is an alpha particle. When that radium binds to the bone at the sites of bone metastases, a single radium molecule is able to basically cause double-strand DNA damage to kill 1 cancer cell, whereas it takes about 100 beta particles to kill a cancer cell. Very, very potent double-strand DNA killing. If you have a patient with a BRCA2 mutation, that person’s cancer cells are not able to fix double-strand DNA damage because their BRCA is deficient. If you give that patient a drug that induces double-strand DNA damage and that cancer’s unable to repair the damage so it can stay alive, it will die. There’s multiple publications now showing that in patients who happen to have DNA repair mutations this drug works better. Obviously it was not approved for those patients. It was approved for all men with bone mets.

DR LOVE: That’s really interesting. I would guess that ALSYMPCA, I’m guessing they didn’t look at BRCAness or anything like —

DR ANTONARAKIS: They did not. That study was reported in 2013. We weren’t doing genomics back then.

DR LOVE: Because the weird thing about that was, there was a survival benefit.

DR ANTONARAKIS: There was. There was.

DR LOVE: Hmm. Wow. That is very interesting. If it really turns out that that’s maybe where radium-223 is working, I guess it would really affect the whole algorithm. What about using radium and PARP together?

DR ANTONARAKIS: We have to consider the toxicities. We don’t know what’s going to happen there. That’s one possibility. The other possibility is, we know that ATM mutations are even more radiosensitive than BRCA2 mutations. There’s this new class of drugs called ATR inhibitors. And there’s this new thought about combining radium-223 with these ATR inhibitors. There could be new forms of synthetic lethality that are created by combining 2 drugs that both cause DNA damage.

DR LOVE: Wow.

Sensitivity of BRCA1, BRCA2 and ATM mutations to PARP inhibitors

DR LOVE: The big question I have after your talk is when again, I haven’t heard before about BRCA1 and BRCA2 and ATM, that granularity. Do you see that in breast cancer or ovarian cancer?

DR ANTONARAKIS: In breast cancer, the trials were exclusively enrolling patients, at least the first ones, with germline BRCA1 or 2 mutations. They were excluding the ATMs. And as far as I remember, the initial studies, they were not including people with somatic mutations only.

DR LOVE: But did the BRCA1 respond?

DR ANTONARAKIS: They did. In breast cancer, they did. Yes.

DR KROP: I mean, in fairness to that first study, the number of BRCA1 patients was like —

DR ANTONARAKIS: Very few, yes.

DR KROP: — 15 or something. But it’s interesting. In the 2 published trials in breast cancer of talazoparib and olaparib, 1 drug shows preferential activity in BRCA1. The other shows in BRCA2. It’s unclear whether that’s real or just a fluke. But it definitely, in germline BRCA mutation breast, both BRCA1 and BRCA2 are sensitive to those drugs.

DR LOVE: Huh. And we don’t know anything about ATM?

DR KROP: They’re just now — as you were alluding to, they’re just now doing studies of somatic BRCA mutations and sensitivity to PARP inhibitors and the same kind of genes that were tested in these trials that were just mentioned. These other genes involved in DNA repair, they’re just now testing to see if those would be sensitive in breast to PARP inhibitors. It’s an open question.

DR LOVE: What about ovary? Do you know whether BRCA1 and BRCA2 respond the same?

DR ANTONARAKIS: Germline ovarian BRCA1 deficient do respond to PARP inhibitors.

DR LOVE: Hmm. Any questions?

DR MCKENZIE: I’ve got one. We talk about mutations like they’re all the same, BRCA mutant, ATM mutant…do you see a difference in the types of mutations in BRCA1 or BRCA2 and how the specific mutations might differ between histologies and then also in response?

DR ANTONARAKIS: Yes. I have not seen that data, although I would love to see it. This raises 2 questions — a question to your question. One is, is it possible that biallelic mutations are the norm with BRCA2 but perhaps not with ATM? Is it possible that a prostate cancer patient or a breast cancer patient with an ATM mutation may not have a second hit, whereas most of the BRCA2 mutations do have a second hit, and that’s what’s driving the difference?

The other thing is, as you pointed out, a missense mutation. No matter how pathogenic it is, it just changes 1 single amino acid. Whereas a truncating mutation, if it happens early on in the reading frame of that gene, will wipe out 50%, 60%, 80%, 90% of that protein. It would seem to me intuitively that an early frame shift or nonsense mutation or a whole genomic deletion should cause more synthetic lethality than a clearly pathogenic point mutation. But I don’t think any of the companies have analyzed their data in that granular a fashion, but I would love to see it.

DR MCKENZIE: Likewise.

DR KROP: Can I just ask 1 quick question? I didn’t even appreciate the biallelic versus monoallelic differences, which are really interesting. Is the idea that the monoallelic loss is a tumor driver through haploinsufficiency, or is it just that it’s not relevant in that cancer?

DR ANTONARAKIS: Right. If the first hit is a somatic hit, then that cancer might have developed independently of the BRCA mutation. If the first hit is a germline hit, the vast majority of those patients when you do somatic sequencing, you find a second hit — 80%, let’s say. And even in the ones that you don’t, most people think that you’re missing it. It’s there somewhere. It might be an LOH. It might be some other genomic rearrangement that you can’t pick up. In the case where the first hit is germline, most people do believe that the reason why that patient got cancer is because they got a second hit at the somatic level.

Now, we do sometimes see double somatic BRCA mutations in prostate cancer, where you see 2 hits. They’re on different alleles. You do a germline test. The germline test is completely normal.

DR BENDELL: It was interesting, too: In your data, it really looked like the trial that was — you would think stacked the deck with the biallelic, didn’t really show much difference.

DR ANTONARAKIS: Did much better than the other two. I agree. Which is interesting.

DR BENDELL: Yes.

DR ANTONARAKIS: Perhaps in real life, as Neil would say, it may not matter if you have the 2 hits or not.

Activity of PARP inhibitors in patients with breast, ovarian, prostate and pancreatic cancers with BRCA mutations

DR LOVE: I’ve asked this question before, which is, is BRCA the new MSI or the new NTRK? And Johanna gave a good example of BRAF, where you don’t see the same thing in colon cancer as you do in melanoma and lung cancer, for that matter. We’re already hearing some differences in how people respond. I don’t even have a really good feeling for how effective PARP inhibitors are in these three, and let’s bring in ovary as well. They’re studied in different ways. Ovary is like maintenance after platinum.

First of all, do you think you could begin to look at BRCA as, like, an MSI? In other words, tumor nonspecific? And what are the key elements in whether somebody’s going to clinically respond? Do we know that?

DR ANTONARAKIS: The short answer is, I would say yes. And fortunately, these mutations, at least in some of the common epithelial tumors, are very much more prevalent than NTRK or IDH1 mutations, let’s say.

The challenge is that we may have shot ourselves in the foot in the way that we have designed these studies in tumor-type specific indications. Whereas if we had, at the very beginning, done a basket trial, let’s say, for any histologic type with a pathogenic BRCA1/2 and then they got PARP inhibitor and objective response was the primary endpoint. Then you could have envisioned a tumor type-agnostic approval. But we have not done that, and I’m not aware of any studies that are doing that.

DR LOVE: I mean, do you have any intuitive sense if you were to compare these 3 cancers, and let’s bring in ovarian cancer — pound for pound — how a BRCA germline patient would respond to a PARP inhibitor in all those diseases?

DR ANTONARAKIS: I would say generally yes, and I would add pancreatic cancer to that list as well.

DR LOVE: Sure, of course. But you think that, for example, a patient with pancreatic cancer kind of inherently has the same sensitivity to a PARP inhibitor as a patient with ovarian cancer?

DR ANTONARAKIS: I don’t treat those other cancer types, but if I had to give you my — my best guess, I would say that — and this is going to be a pure guess — across cancer types, I think about 40% to 50% will have some type of benefit, and the benefit on average would last somewhere in the 6-month range to 9-month range.

DR LOVE: Do you have any — as long as you’re postulating things, it’s really interesting when we go to a general medical oncology audience and we present a BRCA-germline breast cancer case, the most common answer is platinum followed by PARP maintenance, even though it’s never been studied in breast cancer. But they like that model, that concept. Do you have any sense about what the best way is to utilize these agents? Should it be as monotherapy or as maintenance after platinum, for example?

DR ANTONARAKIS: I’d love to hear Ian’s point, but in prostate cancer we kind of got lucky because we don’t use platinum in this disease.

DR LOVE: Right.

DR ANTONARAKIS: When these PARP inhibitors were developed, a lot of people said, “This is just an expensive platinum.” Ie, do we really need to give a drug that’s going to probablycost $10,000 a month, or can we just give carboplatin to these patients? That hypothesis has never been proven, but I think in other cancer types where both drugs are FDA approved and commonly used, my experience talking to my colleagues is that there is definite cross resistance between platinum agents and PARP inhibitors.

The most striking evidence of that are these so-called reversion mutations, which are also called restoring mutations. And this is one of the most fascinating things in biology that happens, is that the mutant form of the gene converts back to wild type under the selective pressure of the PARP inhibitor therapy or the platinum therapy.

And what is going to blow you mind even more, Neil, is that this can happen even when the first hit is a germline mutation. You can have a germline BRCA2 mutation in a breast cancer patient. You give that patient a PARP inhibitor. They respond for 9 months. They develop a secondary resistance. You take a biopsy of the resistant clone, and you sequence that gene. And that gene is wild type. It’s back in frame.

DR LOVE: What has happened?

DR ANTONARAKIS: What has happened is that 1 of the clones in that tumor has developed a mutation that restores the function of BRCA such that synthetic lethality is no longer applicable.

DR LOVE: Huh.

DR ANTONARAKIS: That cell is no longer synthetic lethal with a PARP inhibitor.

DR LOVE: Incidentally, before Ian comments, is there any interest in looking at platinum agents again now in prostate? If you go back, look just at the 25%, would you see something different?

DR ANTONARAKIS: In 2009, Cora Sternberg and Oliver Sartor did a study with a platinum agent —

DR LOVE: Right.

DR ANTONARAKIS: — called satraplatin.

DR LOVE: Right, I remember that.

DR ANTONARAKIS: And it’s a fascinating study. It was published in JCO, and then the drug company went out of business because it was negative. Fifteen percent to 20% of patients had a profound response to this oral platinum.

DR LOVE: Huh.

DR ANTONARAKIS: That was in the days when we were not even thinking about genomic testing. I’ve always said to Oliver and Cora, if we had done that study now and we had selected the HR-deficient patients, maybe done a head-to-head comparison of olaparib versus satraplatin, both drugs might have been equivalent.

DR LOVE: Wow. That’s interesting.

DR ANTONARAKIS: That’s pure speculation.

DR LOVE: Ian, any comments?

DR KROP: No, I think this is actually a very topical discussion now about in breast, we did our trials of PARP monotherapy versus chemotherapy. They’re quite active — the response rate is about 60% for both drugs, with good progression-free survivals and relatively good tolerability. But we see the very impressive data they’re getting in ovarian cancer with maintenance therapy, and there’s a real interest in potentially trying to test that approach in breast cancer. Because again, we took this trial approach because it was straightforward of just chemotherapy versus PARP.

But it doesn’t mean that it’s the optimal way to do it. And I think there’s real interest now in going back now that the drugs are approved and then looking to see whether we can get further benefit from that. I think it’s a very reasonable strategy. It just needs to be tested. Because you’re right. I think in the community people are doing different things, how they’re sequencing platinums versus PARP, and there is a real potential for negative effects because of these reversion mutations and other resistance mechanisms. I think it would be really helpful to have definitive data that we’re doing it the best we can for the patients rather than just do it empirically.

DR LOVE: And we’re not talking much today about pancreatic cancer, but as you mentioned, PARP’s very much on the table. And there, from what I’m hearing from investigators, the trial is a little bit complicated really because of the chemotherapy and because they get up-front platinum, FOLFOX, as part of their up-front therapy — or FOLFIRINOX, more likely. And the problem that the GI people have is, there was no arm where they used chemotherapy maintenance.

It kind of didn’t ask the critical question. What are you doing nowadays about BRCA testing and PARP in pancreatic cancer?

DR BENDELL: I mean, we test for it for pancreatic cancer patients. And more because it might define what we’re going to use in our first-line therapy, where we would want to include a platinum as part of first-line therapy. But I think that the question that’s been alluded to is, is the PARP just an expensive oral platinum? There’s some arguments that people that are pro-PARP in the maintenance setting for the patients with BRCA pancreatic cancer say, “It’s a quality of life issue. It’s being able to put somebody on oral agent. It’s less toxic than FOLFIRINOX potentially, and FOLFOX.”

And that’s why we would use it. But we never actually look to see and compare versus chemotherapy, that if you just continue the platinum agent, would those patients do better than the PARP inhibitor? I think for pancreas cancer patients in particular, we see the PARPs work, but the PARPs seem to have less activity than they do in other tumor types. Again, pancreas cancer is the hard one. It’s very heterogeneous. There’s a lot of issues with the microenvironment. A lot of different contributing factors, which may play a role in the fact that PARPs don’t seem to be as active. But we still have to figure out why.

Homologous recombination deficiency and response to PARP inhibitors

DR LOVE: Andy, another term you hear, and it’s actually in a bunch of these trials, particularly in ovarian cancer, is the concept of HRD. Like, what it is, and what are you actually measuring?

DR MCKENZIE: Yes, that’s a good question. Homologous recombination deficiency is HRD. And I might defer to some other colleagues who use these in the clinic, but it’s basically just another measure of how well your cells can repair DNA. And the specifics of those tests I’m not as clear on, and I might defer to some others on the panel here. But that’s basically what it is, how well does your DNA get repaired by its intrinsic DNA damage repair machinery?

DR LOVE: Is it actually looking at that functionally? Or is it just a bunch of mutations that kind of correlate with that function?

DR ANTONARAKIS: It’s the latter. It’s either mutations or sometimes RNA analysis. And it’s basically a measure of genomic scarring. How much scarring do you have in your genome at a particular point in time? The more scarring you have, the more double-strand breaks and rejoining of chromosome ends you have, the higher that HRD score is. It’s also called an LOH score — loss of heterozygosity score.

DR LOVE: Yes, I’ve heard that term “scarring.” But again, like, what are they measuring? They’re not measuring the scar. Are they measuring mutations?

DR ANTONARAKIS: Yes, they are. Yes. Either mutations or transcripts from these genes.

DR KROP: But that’s a real limitation. Theoretically that’s a real limitation of the assay, because, as you were asking, is this measuring functional recombination? And it’s not. It’s measuring what happened in the past, in terms of the cells’ ability to repair its DNA.

And as you already heard from Andy, that’s something that can be in flux. And the cancer may have been HR deficient in the past, but then under selective pressure it’s now regained its ability to do homologous recombination. You can’t tell that by looking at the HRD assay because the scar, the effects are still there, but functionally the cell is able to perform homologous recombination and therefore may not be sensitive to the drugs like PARP inhibitors.

A truly functional assay is something that requires live cells and much more complex and not something that we’ve been able to do at scale yet. But that, theoretically, is what you’d really want to do, is to say, “Right now, is the cell able to do HR, not what it used to be able to do?”

DR LOVE: I mean, I thought the most interesting, encouraging thing about the survey was how many people responded to PARP inhibitors. You look at trials, whatever, trials are trials, but to see people in the community giving these drugs to these patients that wasn’t even done 2 years ago, and they’re responding. It’s amazing.

Case (Dr Antonarakis): A man in his mid-40s with mCRPC and a BRCA2 mutation attains a complete response to olaparib

DR LOVE: Let’s go through a couple of your cases, Emmanuel.

DR ANTONARAKIS: These are all cases that I saw in clinic. They’re all absolutely real cases. This is the first guy, an African American man, actually. He was diagnosed himself at a very young age, 45, which is, like, 2 decades earlier than the mean. His PSA was 6.9 at diagnosis, had a high Gleason score of 8 and he had T3a disease, which means he had extracapsular extension. No lymph node metastasis and no distant metastasis. Interestingly, this patient’s mother was diagnosed with breast cancer at 39, and she passed away in her 40s from metastatic breast cancer. Neither his mother nor the patient nor anybody else in his family had had any type of germline or somatic testing before they met me.

But when I met him, he had already had a prostatectomy and failed that. Had salvage radiotherapy, failed that. His PSA had been rising. His local oncologist put him on ADT, which worked for about 20 months, a bit more than a year. And then when I first met him, he was basically nonmetastatic, castration resistant. His PSA was rising. It was 15 on the day that I met him. Again, he had not had any genomic testing. And we did a scan, and for the first time we found pulmonary mets. And this is a bit interesting, and I’ll get back to it, because this is an unusual site for a first metastatic spread. His bones were clear, and the rest of his lymph nodes were clear, and he had, like, 20 small cannonball lesions in both lungs.

And in those days, I was using this color genomics assay, which has a panel of 20 genes. This is a germline test. They do it from saliva. And as you can see there, it’s kind of small to see, he has a pathogenic germline BRCA2 mutation. It’s a frameshift. This is a protein truncating mutation, pathogenic.

We don’t know if his mother had it. I presume that she did, but she had already passed away. This is an African American guy with a germline BRCA2 mutation. It does occur in African Americans, not just Caucasians, not just Ashkenazi people.

DR LOVE: Is it less common?

DR ANTONARAKIS: It’s a bit less common, but not that much less common. And I gave him off-label olaparib. You can see his PSA was 15. This is a real scale.

Those are the real dates at the bottom. The reason why it changes from blue to purple is because he started going to a different lab halfway through. And over the course of 6 months, this man had a complete PSA response just with PARP inhibitor, nothing else.

DR LOVE: What was his prior therapy systemically?

DR ANTONARAKIS: He had just had ADT. Just ADT.

DR LOVE: But he progressed after ADT.

DR ANTONARAKIS: Yes, he was castration resistant. This was pretty profound. He tolerated the drug well. He had a little bit of nausea. No neutropenia at all. No weight loss. No cytopenias at all.

And you can see that these are the real dates, January of 2019 and October 28^th, 2019. And I didn’t circle the pulmonary mets, but the radiologist called it a CR — complete response. PSA went to zero, and all his pulmonary mets disappeared. He’s still on the drug. He’s been on it for 9 months, and his PSA is zero. He’s tolerating it well.

DR LOVE: Just out of curiosity, any tolerability issues?

DR ANTONARAKIS: Some nausea. If he takes it with food, the nausea is better. And that’s it.
He’s very young, though. He’s very young.

DR LOVE: How often to you see a response like this?

DR ANTONARAKIS: About 50% of the time.

DR LOVE: Really?

DR ANTONARAKIS: Yes. With BRCA2.

DR LOVE: With BRCA2 specifically. And this is a syndrome of pulmonary mets that’s specific?

DR ANTONARAKIS: We actually became interested in this phenomenon because we had not seen pulmonary mets as a presenting sign of metastatic disease that often. And we went basically back into our archives and we asked, “What are the genomics of these patients?” We actually ended up finding about 20 patients over a 10-year period who had presented with the first line of mets being pulmonary only. No bone. No lymph node.

And the pie chart on the bottom shows that 50% of these patients have a DNA repair deficiency. The yellow, which is 25%, is MMR, which is 2% in the unselected patients but is a quarter of these guys. And then the gray one, which is the bottom right-hand corner, that’s HRD, including BRCA.

If you have a guy that presents with pulmonary mets, he has a 50% chance based on this single Hopkins study to have an actionable DNA repair gene mutation, either a mismatch repair mutation or, as in this patient’s case, a homologous repair mutation.

DR LOVE: And, of course, you’re going to tell us why.

DR ANTONARAKIS: I have no idea.

DR BENDELL: Did you ever do full profiling on him?

DR ANTONARAKIS: We did not. We did not.

DR LOVE: What were you thinking?

DR BENDELL: There’s been some question when you have even HRD patients that they could have a higher rate of mutations than others, and are these patients ones that you would think of using an immunotherapy? And there’ve been studies that have looked at combinations of immunotherapy with PARP inhibitors. Is this something that you would think about at some point?

DR ANTONARAKIS: Yes, this patient had pulmonary-only mets. They were very small. They were about a centimeter, and he did not want to run the risk of a pneumothorax. He passed on the research biopsy. All we have on this man is germline.

Case (Dr Antonarakis): A man in his mid-60s diagnosed with de novo mPC and MSI-high status, high tumor mutation burden and an MSH2 mutation experiences an excellent response to pembrolizumab after multiple lines of therapy

DR ANTONARAKIS: Now, this is also another African American patient. He was diagnosed at 66, which is a bit more usual. And he was one of these guys that presented out of the blue with metastatic disease. And he had high-volume disease, which is classified as more than 4 bone mets or visceral mets. His Gleason score was extremely high, 5 + 4 = 9. And he also had something that was very unusual, which was a ductal carcinoma. These are prostate cancers that arise in the prostatic ducts, not in the acini. No family history of cancer whatsoever, and he had a big family.

And as we are doing more and more these days based on the CHAARTED study that Chris Sweeney published, we did ADT plus 6 cycles of docetaxel. He did respond — had a pretty profound response — however a year and a half later developed a rising PSA and progression in bone. He then took abiraterone, which is the usual next step. Responded to abiraterone for a year.

DR LOVE: Wow.

DR ANTONARAKIS: Yup. We talked to him about chemotherapy again. He was very reluctant, because he didn’t have such an easy time with the docetaxel. He wanted to try enzalutamide back to back. His AR-V7 was negative. We thought it was a reasonable thing. He did not respond at all to enzalutamide. Developed further progression in bone and now developed a very bulky retroperitoneal mass. His PSA was 150. Gleason 5 + 4 ductal carcinoma, African American guy, no family history.

And this is his Foundation Medicine report, genomic testing. the first thing you’ll see is it’s MSI high. The second thing you’ll see is that the tumor mutational burden, TMB, is 44 mutations per megabase. Let me put this in context. Only 3% of prostate cancer patients have a TMB greater than 20 mutations per megabase. This is on the far right-hand part of the bell-shaped curve. Forty-four mutations per megabase. That’s very high.

And if you look at that list — I've helped you by underlining it. You can see that there’s a loss of function MSH2 mutation. On this patient’s immunohistochemistry, he had protein loss of MSH2 and 6, which is what you would expect.

This guy has an MMR-deficient, microsatellite-high and high-TMB tumor. He’s got all 3. He’s got the trifecta. Germline was negative. This was not a germline mutation. It’s not Lynch.

There it shows his 1-year response to abiraterone, then got enzalutamide for 3 months, had no response whatsoever and then when his PSA was 150, we got standard of care pembro for him because of the MSI high, and you can see the date there, October 16^th, the last time I saw him. His PSA went down to zero within 3 doses, and he’s had a complete response.

DR LOVE: Wow. Again, any tolerability issues?

DR ANTONARAKIS: No tolerability. Let me show you the objective.

DR LOVE: Wow.

DR ANTONARAKIS: That big mass in the center, it’s hard to see there. It’s not really down to zero, but I guess the radiologist called it a PR.

Tolerability? No. No autoimmune phenomena. He is developing a bit of fatigue — he’s been a little bit more tired. But nothing else.

Association of ductal and intraductal prostate cancer with a high prevalence of DNA damage repair gene mutations

DR ANTONARAKIS: Now, the reason that I mentioned the ductal is because again, when you see these rare histological variants, you’ve got to wonder whether there’s something else going on. We then, prompted by this patient, went back and looked at 50 consecutive cases with ductal carcinoma. We have a pathologist called Jonathan Epstein who does all these weird secondary cases. And Jonathan had 50 cases in his archives with ductals, either from prostatectomy or needle biopsies. And we actually did not whole exome but targeted panel on these guys.

And interesting result. Once again, that 50% number keeps coming up. Half of these patients with ductal carcinoma of the prostate, most of which have Gleason scores of 9 or 10, had a DNA repair mutation. In this one, MMR was 14% and HRD was 31%. Again, weird clinical manifestation, like pulmonary mets. That might enrich. And then weird histological variants like ductal. That might enrich.

The last thing I wanted to say about this case is there’s something else, which is a hybrid, which is called intraductal carcinoma. Intraductal carcinoma does not arise in the ducts. It arises in the acini, but it spreads into the duct. The pathologist is able to say, “This is an acinar carcinoma with intraductal spread.” It’s also called intraductal carcinoma sometimes.

This was another study from Hopkins where we took 150 patients — these were all unselected — and we published this paper shortly after The New England Journal of Medicine paper looking at germline mutations. And this was the first 150 patients in my clinic after that Pritchard paper was published. And in his paper it was 12%. In our hands it was 14%. It’s the same makeup, BRCA2, ATM, CHEK2 and BRCA1.

But when you look at the intraductal versus no intraductal at the table at the bottom, so if the patient had intraductal histology, which is the minority of them, there was a 40% chance that he had a germline mutation — 40%. In those that had no intraductal, it was pure acinar without any ductal and no intraductal, the prevalence was 9%. This again suggests that another histological variant, intraductal carcinoma, might also enrich. In this case, it enriches for germline mutation. There’s all these clinical clues that we can use to increase our chances of finding an actionable mutation.

Predictors of benefit with immune checkpoint inhibitors

DR LOVE: That’s really interesting, because it seems like it attempts to find a tissue or other assay to predict sensitivity, particularly to checkpoint inhibitors. Kind of doesn’t seem to be going too far beyond PD-1 and TMB, and even those aren’t all that tremendously exciting.

Anything you’ve heard about, Andy, in terms of prediction of benefit to checkpoint inhibitor that’s new?

DR MCKENZIE: Yes, I think we’re better at predicting lack of response to immunotherapies than we are at predicting response to immunotherapies. And I think back to things in the lung cancer world especially, where you’ve got STK11 and KRAS co-occurring mutations, and then also this new gene KEAP1 that keeps coming up. And those are associated with lack of benefit.

I think that the tumor microenvironment is going to be the key where we really discover pockets of patients who have a distinct tumor microenvironment that are primed to respond to immunotherapies. And I don’t think we’ve done all the work there yet. It’s something that’s on the horizon.

DR LOVE: Here’s a very simple question. And probably I’m just not smart enough to understand the obvious answer, but if you have a patient with a known BRCA germline mutation, strong family history, pretty classic situation and the patient gets lung cancer, do they respond to a PARP inhibitor?

DR ANTONARAKIS: You need to have a piece of the lung tissue and look for the mutation in there.

DR LOVE: If it’s germline, it’s going to be in there, right?

DR ANTONARAKIS: No, not —

DR LOVE: No?

DR ANTONARAKIS: No. It could be a sporadic lung cancer that occurred in a germline BRCA2-positive patient.

DR KROP: The mutation will be there, it just won’t be biallelic.

DR ANTONARAKIS: Right.

DR KROP: Yes. You won’t have LOH. You won’t have loss of the other —

DR ANTONARAKIS: Neil, this brings up an important point. Not every cancer in a patient with a germline DNA repair mutation is due to an alteration in the second copy of that gene. You can still get sporadic cancers, even in Lynch syndrome. For example, I have seen patients with prostate cancer that have germline MSH2 or 6 mutations, and their prostate tumor has the germline mutation but does not have a second hit in that same gene in the prostate cancer.

DR LOVE: How do you tell that?

DR ANTONARAKIS: You look for that second allele.

DR LOVE: On what? Like, an NGS?

DR ANTONARAKIS: Yes.

DR LOVE: You look for the second allele.

DR ANTONARAKIS: Right. Do you remember Knudson’s 2-hit hypothesis? Alfred Knudson.
What Knudson hypothesized, which is — it seems like it’s bearing out — is that you need 2 hits. The first hit, typically in the setting of cancer, occurs in the germline DNA. But then a cancer does not develop until a somatic cell, let’s say a colon cell, develops a second hit in that same gene. And then that second gene inactivates the function of, let’s say, the mismatch repair protein, and then that person gets a colorectal cancer. If you don’t find that second hit in the tumor, the somatic hit, you cannot assume that that cancer was related to the germline predisposition.

DR LOVE: What exactly would the NGS assay say?

DR ANTONARAKIS: Let’s say you have a frameshift mutation in MSH2. The NGS assay would have a second mutation somewhere else, or it would have a deletion of the other allele. It would say loss of heterozygosity.

DR LOVE: Do you think the average general medical oncologist would see that on a report?

DR MCKENZIE: They would see it. On one of the cases that we reviewed earlier, I think there was an example of that.

DR ANTONARAKIS: The ATM case.

DR MCKENZIE: ATM. It had an ATM mutation, another ATM mutation, and then the splice site variant. they listed them one right after the other on the Foundation Medicine report.

But you’re bringing up the exact right question, and that’s most people see the report and see the gene name and that’s enough. And I think what we’re kind of talking about is diving deeper into what the specific mutations are in that gene and, even deeper, what are the allelic fractions of those mutations? Are all the minutia involved in here that are all very clinically important and very difficult for a general medical oncologist to get just from one of these reports?

DR LOVE: But let’s move on now and talk about breast cancer.

Role of MSI testing and multiplex genomic assays in mBC

DR LOVE: And I have to say that I guess because there’s many things going on with breast cancer, many — HER2 and ER — everything, somehow MSI was a little bit off my radar. I mean, I guess not all people are tested but here. More than half the people test for it. 

We asked them to tell us about the last patient in their practice they had with metastatic breast cancer who had MSI-high disease. And you can see, mainly it’s triple-negative. I don’t know whether that’s because they’re looking for it more in triple-negative, or you can comment on that when you go through that. And most of these people, the treatment they got was a PD-1 antibody. They got it anywhere from first line to way later than that.

But again, interestingly, at least in their viewpoint of the 24 people that were evaluable, 19 from the point of view of the doc responded to therapy.

We also asked them whether or not they — just like we asked with the other 2 tumors, but of course we broke it down into the 3 subsets — do they generally order multiplex genomic testing at some point in their patients with metastatic disease? And you can see that they order it in all 3 subtypes more than half the time. A little bit more in triple-negative, but they still order it in the other 2 substantially.

We asked when they do genomic testing. And again, you could think about okay, maybe do it right up front just to get it started, or wait until you’re close to when you think you might want to use it therapeutically or when you don’t really have a great option. In ER-positive, HER2-negative metastatic disease, certainly you have at least first-line hormonal therapy with a CDK. I think people would get that almost regardless of what their genomic status was. But anyhow, you can see when they order the testing.

The same thing with HER2-positive. When they order it there — I would have thought it would have been a little bit less than the other two, because there's many HER2 options.
And then in triple-negative, where almost half the people do multiplex testing on first diagnosis.

We asked them to tell us about the last patient in their practice with breast cancer where they did multiplex testing. And you can see, actually, in most of the cases, they didn’t feel it helped make the decision or make a decision. But when it did, you can see the markers that they looked at — either immune, PIK3 or BRCA, for example, and whether or not they responded.

We asked about liquid biopsies, and it was even less frequently utilized and perceived to be less benefit in breast cancer.

Assessment of PIK3CA and BRCA mutations and assays for ESR1

DR LOVE: Is PIK3 mutation status a standard in metastatic ER-positive and HER2-negative? And almost everybody says yes. And again, some up front, but a lot later on, which is when a drug like alpelisib might be considered.

We asked how many patients they had that they had PIK3 mutations? I think at this point it’s pretty standard, certainly in patients after they’ve gotten a CDK inhibitor. Interesting that about a third of these patients were PIK3-positive. That’s a little bit lower, I guess, but not that far off for a small sample.

And you can see that almost all these patients who did have the PIK3 mutation got alpelisib, which kind of makes sense. If they’re ordering it, you would think they would use it. And you can see the line of therapy and just in a few of these patients, actually, 4 out of 6 of them responded.

We also asked about — and Ian’s going to talk about — ESR1 assays, and on average they ordered it or have seen it in their assay results in 10 patients. It’s about split between positive and negative. But it did affect their treatment, and Ian will talk about how conventionally people respond to that in terms of selection of therapy.

We asked about BRCA testing again in this subset. And it looks like everybody is testing — and mostly it’s up front. They’re kind of getting it out of the way even though I’m not — we can talk about how relevant it is.

With HER2 they’re also ordering BRCA, although I don’t think PARP inhibitors are approved in HER2 right now, but they are ordering it. And kind of it does make sense when you hear people talking about using it. And certainly they’re doing it in triple-negative disease. We asked them what kind of BRCA testing they utilize.

You can see a substantial number just do BRCA-specific germline testing. And Ian can talk about what he thinks would be optimal. Others do panels, multigene panels and either BRCA-specific somatic, which I’m not sure exactly what that would be, maybe HRD.

And we asked about the last patient in their practice who had a BRCA mutation. You can see mainly these are patients with triple-negative breast cancer — it’s kind of interesting to look at this, really, and think about how it’s playing out in the triple-negative population. And you can see that about two thirds of these patients actually got a PARP inhibitor, usually second line or beyond. And, interestingly, again, of the 34 patients where you have some input about response, 29 responded. It seems to have been an action that was positively received.

We asked about PD-L1 assays in these patients, and a lot of people don’t get them at all. I think about almost half the people don’t. I don’t know exactly what the value would be at this point of PD-1.

Presentation (Dr Krop): Biomarker assessment in mBC — ER-positive and triple-negative disease

DR KROP: I was going to talk about 5 different biomarkers in breast cancer. And I think what’s interesting is that all 5 of these just started to become clinically relevant in the last couple years. This is a field that’s moving quite quickly in breast cancer.

And just to start, this is just a sense of the landscape of genetic alterations in breast cancer. These are about 8,600 advanced breast cancer cases for which a genomic profile was done through the Foundation platform. And the blue bars are the nucleotide substitutions.

And you can see on the left there, p53, PIK3CA are by far the most common alterations. But beyond that there’s a smattering of genes with much lower prevalence. And that’s just to kind of give you a sense of where we are in breast cancer right now.

As you’ve already heard from other speakers, we can also do circulating tumor DNA assessments of DNA alterations in breast cancer. And we know that that’s able to provide useful information in about 85% of patients with metastatic disease, and you also heard that these are generally concordant with matched tissue, as long as they are taken at around the same time.

We have the technology to do genomic assessments in breast cancer. But up until recently, there really wasn’t a strong clinical utility for doing this, but that’s changed, and it changed largely, at first at least, because of the development of PI3 kinase inhibitors.

Just to remind you, PI3-kinase is an important protein that is, along with its downstream factors of AKT and mTOR, a key regulator of survival and proliferation in breast cancer, as well as other cancers. And in ER-positive breast cancer, this pathway gets hyperactivated as a way to become resistant to endocrine therapy, and, as you’ve already heard, about 40% of patients with hormone receptor-positive breast cancer had an activating mutation in the catalytic domain of PI3 kinase, and that’s the PIK3CA gene.

For all these reasons, there was clear rationale for trying to target PI3 kinase in breast cancer, particularly ER-positive breast cancer. But I can speak from personal experience, as someone who’s been in development of these drugs: It was frustrating for many years, because the prior generation of these agents was toxic, and the efficacy was underwhelming.

But now we do have a drug that’s approved, and that’s alpelisib, and that approval was based on the SOLO1 trial, which looked at patients who had hormone receptor-positive, HER2-negative breast cancer, the most common type of breast cancer, that had already progressed on an aromatase inhibitor. And these patients were randomized to fulvestrant and other endocrine therapy with either placebo or alpelisib. And we had 2 cohorts in the trial, one where patients who did not have detected PIK3CA mutations and the other ones who did.

And what was found was that for the first time, a PI3 kinase inhibitor led to substantial improvements in progression-free survival. But, interestingly, it was only in the patients who had PIK3CA-mutant breast cancers. The ones who did not have detectable mutations really did not derive significant benefit.

This led to the approval of alpelisib in this setting, and along with it a companion diagnostic test. This is called the therascreen^®. This is a real-time PCR-based assay that picks up 11 of the most common hotspot mutations. I think in practice it’s perfectly reasonable to use a dedicated assay like this or one of the common types of NGS assays that we’ve already been talking about, such as Foundation or Guardant. But certainly this is now a key reason to start uniformly checking at least PIK3CA mutation status, if not a mutation screen, at least in patients with ER-positive, HER2-negative cancers.

Another interesting alteration that’s been found and that’s worth discussing now is when people were going back and looking at both hormone therapy-naïve cancers as well as hormonal therapy progressing cancers, they noticed a significant difference in the mutation profile, and that was principally because of the development of ESR1 mutations.

And that’s shown on the left there. Virtually undetectable in hormone therapy-naïve patients and at least 20%, and actually more, as I’ll show you, in patients who progressed on hormonal therapy.

These mutations are clustered in the ligand-binding domain of the estrogen receptor. In preclinical data, these mutant estrogen receptors are able to develop estrogen-independent transcriptional activity, suggesting that they could be a cause of resistance to hormonal therapy. And consistent with that, if you look at who has these ESR1 mutations, again, you can see if you haven’t been exposed to an aromatase inhibitor the rate is pretty low, about 7%.

But if you look at patients who progressed on an adjuvant aromatase inhibitor, the rate jumps to 30%. And then if you’ve had progression in the metastatic setting, probably a slightly stronger selective pressure. You see the rate goes up above 40%. These are quite common in the endocrine-resistant population.

And again, probably the best data that these are associated with resistance came from this retrospective analysis of a trial called SoFEA, which was a simple trial comparing an aromatase inhibitor, which basically inhibits the production of estrogen, versus another hormonal therapy, fulvestrant, which not only blocks the estrogen receptor but also degrades the estrogen receptor and in preclinical models is able to still do that even in the presence of an ESR1 mutation.

And you can see on the left the Kaplan-Meier curves show a substantial superiority of fulvestrant in those patients who have an ESR1 mutation compared to the aromatase inhibitor, whereas those patients who don’t have an ESR1 mutation, on the right, the 2 drugs are essentially equivalent.

It’s clear these mutations are pretty common, but for them to become clinically useful we need to figure out a way to overcome the resistance associated with these mutations. And right now there are not any prospective data on how to do this, but I think certainly there's a lot of interest in looking at drugs that degrade estrogen receptor. Both fulvestrant, which is the currently approved one, as well as there’s a number of oral SERDs, as they’re called, that are in clinical development.

There are also preclinical data to suggest a number of alternative ways to inhibit the function of this mutant estrogen receptor. They include PI3 kinase inhibitors. There are some interesting kind of synthetic lethality data suggesting that blocking CDK2 or CDK7 also could be beneficial, and there's also interest in looking at bromodomain inhibitors.

There’s a number of potential pathways that are being tested right now, both preclinically and starting to be tested in the clinic, as ways to overcome these ESR1 mutations. But right now, if you find a mutation, I think there’s a general sense that you might want to avoid an aromatase inhibitor and prefer fulvestrant, although this approach hasn’t been tested prospectively as of yet.

PTEN loss and implications for therapeutic decision-making for mBC

DR LOVE: I’m going to stop and ask, any questions at this point? Emmanuel?

DR ANTONARAKIS: In the PI3 kinase inhibitor study, you showed the biomarker-positive and -negative group. In the negative group there was also a numerical improvement with the inhibitor, but not statistically significant. And I wonder whether that might be driven by those patients that have PTEN inactivation or loss, because that would also theoretically activate AKT pathway.

DR KROP: Right, and amplifying on that, the pathway gets hyperactivated in estrogen receptor-positive breast cancer even without detectable alterations, genomic alterations in the pathway. I think your point is well taken. There could be some signal there. It’s certainly not at the level that we see in the mutants. And these are toxic drugs, I think. For small benefits, it’s probably not going to be worth the risk/benefit balance.

I think there’s certainly been hope that we can come up with more targeted approaches to this pathway that can take advantage of the ability to block the pathway even when it’s not altered in those patients who don’t have mutations. But right now, again, we’re fighting against the issue that they are fairly toxic drugs. We can use mTOR inhibitors in the kind of scenario that you have mentioned, although the loss of PTEN hasn’t been shown to be a strong predictor of benefit of any of these drugs.

DR BENDELL: The PTEN loss data for those patients, it doesn’t look like the alpelisib will help?

DR KROP: It hasn’t, as far as I know, hasn’t been looked at —

DR BENDELL: Okay.

DR KROP: — with alpelisib specifically. We have a little bit of trouble in breast cancer really determining PTEN loss, because a lot of times it doesn’t seem to be genomically — it can be through methylation or other means that really can only be tested by protein, and the protein assays for PTEN aren’t great. Right now, at least, it seems that the mutations are the primary biomarker of sensitivity.

DR ANTONARAKIS: The reason I ask is because in prostate cancer there was a study with ipatasertib —

DR KROP: Yes.

DR ANTONARAKIS: — which is an AKT inhibitor.

DR KROP: Right.

DR ANTONARAKIS: And it was a randomized Phase II study. They got abiraterone with or without the AKT inhibitor. Overall study was negative. But in the subset of patients that have PTEN deletion, which in prostate cancer occurs 40% of the time, there was a PFS benefit. And with the AKT inhibitor, in prostate cancer we don’t have activating PI3 kinase mutations as often. That was the reason for the question.

DR KROP: No, I think it’s a great question, and we actually have tested now AKT inhibitors, both ipatasertib and capivasertib, which look promising. And the other interesting thing that speaks to your question, I should have mentioned earlier, is that there are at least some preclinical data that suggest that loss of PTEN tends to be more activating of PI3 kinase beta as opposed to alpha, and alpelisib is an alpha-selective inhibitor, which has the benefit of having better toxicity profile than a pan inhibitor, but it may be a weakness when they’re talking about a PTEN loss-driven cancer. That’s a good point. I should have mentioned that.

ESR1 mutations and resistance to hormone therapy

DR MCKENZIE: I’ve got just 1 quick question about the setting that you put forward was post-CDK inhibitor getting tested for things like ESR1 mutations and others. Is there any rationale for detecting maybe RB1 loss and seeing if that has a connection with resistance to CDK inhibitors? Any thoughts on that?

DR KROP: Yes. Just to be clear, though, the setting that alpelisib was looked at for ESR1 mutations is progression after a hormonal therapy, not necessarily CDK inhibitor therapy. But it’s a hugely important clinical question in breast cancer: Why do cancers become resistant to CDK4/6 inhibitors, since they are such a mainstay of our treatment right now? And RB loss does occur. It seems to be, I think — somewhat to the surprise of us — it seems to be a relatively low frequency event. It’s, like, 5% or 6% of patients seem to develop RB loss as a way to escape the effects of CDK4/6 inhibitors.

There are other mechanisms that have been started that look like they may be more common. Amplification of CDK6 has been seen. But it doesn’t look like there’s 1 clear dominant escape pathway from CDK4/6 progression. That’s mainly looked at through circulating tumor DNA-type assays. We’re not able to look at protein. We don’t have a lot of tissue after progression on CDK4/6. That’s something that’s just now being looked at. But again, everybody agrees that we need to figure this out, because once patients progress on CDK4/6 inhibitors, the ability for further hormone therapy to be effective is diminished.

DR LOVE: That’s interesting, though. We saw in the survey a third of them had already used alpelisib. You can be sure that almost all those people had CDK inhibitors.

DR KROP: Right.

DR LOVE: And yet that’s not where the data is, actually, so…

DR KROP: There were patients on CDK4/6 inhibitors who had previously progressed on CDK4/6 on the SOLO1 trial, but it was a relatively small number.

DR ANTONARAKIS: I’ve always been amazed by the similarities between breast and prostate cancer. And I couldn’t help to wonder when you were talking about the ESR1 mutations, which are in the LBD —

DR KROP: Yup.

DR ANTONARAKIS: — just like in the AR, androgen receptor, the activating mutations are also in the ligand binding domain. But my question was, do those mutations cause promiscuous activation by other steroid ligands? Or do they cause constitutive activation in the absence of ligand?

DR KROP: Yes, yes. When you were talking, I was thinking the exact same thing. Boy, there’re sure some impressive parallels there.

As far as I know, the ESR1 mutants, most of them, at least the ones in the ligand binding domain, don’t need any ligand. They’ve been tested in charcoal-stripped media, and they’re fairly transcriptionally active without anything. The mutations seem to — there’s — I can’t remember exactly the steric effects, but somehow lock the protein into a transcriptionally active domain. That is my understanding. It doesn’t need a steroid round.

DR BENDELL: The PIK3 mutations and the ESR1 mutations are mutations that are mechanisms of resistance. Do you then serially test? Or when to you profile?

DR KROP: Yes, the ESR1s are a little different, because we’re not absolutely sure what to do with them yet, whereas with PIK3 mutations we have a drug. I think that’s an important question, and right now the field is just trying to come to grips with it since alpelisib just got approved pretty recently. But I think what most people are doing is doing the genomic assays in the first line while their patients are getting CDK4/6 inhibitors and then using that to drive your decision-making for second-line therapy, whether you do alpelisib or not.

Theoretically, you might like to know whether they have —

DR BENDELL: Yes.

DR KROP: — an ESR1 mutation, but in practice, since they generally arise after aromatase inhibitor therapy, you’re, for the most part, going to use fulvestrant anyway, because you’re not going to go back to another aromatase inhibitor.

DR BENDELL: then the PIK3 mutations are occurring before the CDK4/6?

DR KROP: Yes, it’s hard.

DR BENDELL: Yes.

DR KROP: The PIK3CA mutations are truncal.

DR BENDELL: Okay.

DR KROP: Those are not selective. It’s the ESR1 mutations that are selective, for the most part. PIK3CA are truncal. You could check their primary cancer and you’ll pick it up.

DR MCKENZIE: One last question about the ESR1 mutations. In your graph it showed the prevalence of those showing up over time, but 7% of them existed before treatment. Are those germline? And is there any evidence of any germline ESR1 mutations that might indicate use of fulvestrant and SERDs earlier in treatment?

DR KROP: I haven’t heard of a germline ESR1 mutation. It’s an interesting question. Could that be a transforming oncogene in and of itself? I haven’t heard that. That 7%, I don’t know whether those patients could have had tamoxifen, which is another hormonal therapy that still is potentially selective for ESR1 mutation development. The pure primary cancer, never seen a treatment population, the ESR1 mutation rate is a couple percent.

DR MCKENZIE: I see.

DR LOVE: Is there a one-off assay for ESR1?

DR KROP: Not that I know of. You could develop one. You can do a CT DNA —

DR LOVE: Right.

DR KROP: There’s not a commercial one that I know about. People have them in their labs.

DR LOVE: From a practical point of view, is there an advantage to doing a multiplex assay, for example, NGS, in a patient, let’s say, progressing on a CDK/hormone, ER-positive, HER2-negative, as opposed to a one-off PI3K?

DR KROP: Yes. I think that’s a great question, and I think in practice most of us just do a Foundation-type panel because then you get, in addition to your PIK3CA, you could potentially pick up a BRCA mutation. You can pick up these other mutations like AKT that may give you access to a clinical trial of a particular targeted therapy. But in truth, for the vast majority of people, especially who don’t have access to, as Johanna was saying, a big clinical trial infrastructure, you’ll be fine with just doing a targeted assessment of PIK3CA. But for most of us it’s just as easy to get the panel.

DR LOVE: I keep thinking about that case that we started this thing off on. When I saw that presentation a couple years — I think it was in SITC, of the non-Hodgkin patients who had CD274, I was thinking to myself, the answer to the question I just asked you maybe ought to be NGS. Even if it’s 1 in 10,000 that you pick that up.

DR KROP: Right. If it was you, you’d want to know if you happen to have a very targetable — and the MSI is another potential —

DR LOVE: Although, theoretically you could get that in a one-off assay.

DR KROP: Right. But it would be a separate one-off.

DR LOVE: Right.

DR KROP: Yes.

DR LOVE: In lung the issue comes up where there’s many things they want to test for. There’san issue of how much tissue’s available. Is that ever an issue with some of these other tumors?

DR MCKENZIE: I’ll let the clinicians speak to that, but I think certainly so. And I think the NCCN guidelines, at least in lung cancer, since there are many, have now included saying language like “Get these genes tested in the context of a broad genomic assay.” But I don’t think the other disease types have gotten there yet. But I’d be interested to hear clinical experience there.

DR BENDELL: I mean, exhaustion of tissue is always an issue, sorry. But it is, I mean —

DR LOVE: In colon also?

DR BENDELL: Even in colon. I mean, unless you’ve got a primary, but then sometimes the primary’s so old that you want the core. And then how many cores do you get when you do your sampling, and what do you end up spending your tissue money on?

Efficacy of PARP inhibitors in patients with BRCA mutations

DR LOVE: Let’s keep going, though, with breast cancer, because it really is 3 diseases. But you had some comments about BRCA and PARP inhibitors. I think that’ll complement what Emmanuel was presenting well. Can you go through that?

DR KROP: Sure, sure. Again, this is another biomarker that’s now become very clinically relevant because of the availability of PARP inhibitors. We’ve already talked about the rationale and the mechanism of action of PARP inhibitors in patients who have a BRCA mutation through the synthetic lethality pathway. I won’t go over it again. But in breast cancer this approach has been tested in a couple large Phase III trials.

The first was OlympiAD, which looked at patients who had germline BRCA1 or BRCA2 mutations — they didn’t include ATM here — who’ve had up to 2 prior lines of chemotherapy. And they were randomized to single-agent olaparib versus physician’s choice of chemotherapy. And the trial was clearly positive, with a 42% improvement in progression-free survival of olaparib over chemotherapy.

And a very similar trial was done with talazoparib, the EMBRACA trial, and the results are actually strikingly the same. The hazard ratio for PFS is both in the mid-0.5s. Response rate is both about 60%. There did seem to be a little bit greater toxicity, hematologic toxicity, with talazoparib, perhaps fitting with the fact that it’s a little bit more potent in trapping. But the efficacy looks essentially interchangeable.

I think based on these data it’s very clear that this should be a standard of care for patients who have a germline BRCA mutation, and it requires testing for this. Exactly when you test, I think, you could argue for ER-positive patients it’s still beneficial — we haven’t tested this formally, but for most patients you’re going to start out with hormonal therapy just because the toxicity profile is favorable. But I think you’ll want to know their BRCA status pretty soon after the diagnosis of metastatic disease. You know that this is an option as opposed to first-line chemotherapy. And, of course, we’ve talked about this whole issue of sequencing with potentially platinums and maintenance things, but those right now should be as part of trials, not for clinical practice, in my view.

I think this field has changed dramatically. A couple years ago we talked about that there really wasn’t a strong rationale for doing genomic testing in breast cancer, and now we have 2 targetable mutations by approved drugs. There’s a number of agents in development that look quite promising. We already talked about the AKT1 and PTEN pathway alterations being targetable by a couple fairly promising AKT inhibitors. ERBB2 mutations are pretty rare in breast cancer, but they do occur, and they do seem to be sensitive to potent HER2 inhibitors like neratinib. We talked about ESR1 mutations, which right now we’re not sure how to necessarily overcome their effects, but we have some promising leads.

And then there’s a number of other more rare things, such as FGF receptor amplification. And there’s a long series of other potential targets. And then there’s this observation of somatic mutations in BRCA1 and 2 as well as other DNA repair proteins. And that’s now being tested with PARP inhibitors in breast cancer and other diseases as well, as you’ve already heard. Again, the landscape has changed quite dramatically.

HER2 expression in HER2 nonamplified cancers; activity of trastuzumab deruxtecan in HER2-low breast cancer

DR KROP: I just wanted to touch on a couple other nongenomic biomarkers. One is looking at HER2 expression in HER2-nonamplified cancers. And you may wonder why we’re talking about this, since it’s very clear that the conventional HER2-directed therapies do not work in HER2-negative cancers. That’s been shown fairly conclusively, but there are new drugs.

And Johanna mentioned this before: There’s 1 antibody conjugate that appears particularly potent. This is DS-8201, or trastuzumab deruxtecan, which is another HER2-directed antibody-drug conjugate. It’s got a more potent payload than the current drug, T-DM1. It’s got more of them. The drug-to-antibody ratio is about 8. And, interestingly, this payload is membrane permeable, meaning that once it goes inside of a targeted cell, the payload can actually leak out through the membrane and hit a neighboring cell. And that’s in a target-independent fashion. If you have a heterogeneously expressing tumor, the target-negative cells can still potentially be killed by this agent.

And we have a small amount of clinical data. These were data presented from the Phase I trial of this agent by Shanu Modi. And this was looking at what we call HER2-low patients. These have moderate levels of protein by IHC. One-plus, 2+, but they’re not FISH amplified. In clinical practice, these are HER2-negatives. And we just thought of these as just as HER2-negative as an IHC 0 patient, because there was no clinical utility of knowing exactly how much HER2 there was if they were FISH-negative.

But what was found was that this drug, DS-8201, had a response rate of over 40% in this population. Duration of response was almost 10 months. PFS was over 7 months. this is a clearly active drug in this population, which previously wasn’t even recognized as a subtype of breast cancer.

This is now being tested in a randomized Phase III trial in HER2-low patients comparing 8201 versus physician’s choice of chemotherapy. And if this is positive trial, then again, it starts a whole new subset of breast cancer, and it’s a very common subset of breast cancer. Really potential for broad clinical impact.

PD-L1 and other biomarkers predictive of benefit from immunotherapy in triple-negative breast cancer (TNBC)

DR KROP: And then lastly, we’ve kind of already talked about predictors of immunotherapy. In breast cancer we’re certainly no better than anybody else, probably worse, in identifying predictors of benefit. But it’s worth talking about PD-L1 itself, because in breast cancer the 1 indication for immunotherapy is in the first-line triple-negative setting, and that was determined by this trial called IMpassion130, which looked at newly diagnosed triple-negative patients with metastatic disease and randomized them to nab paclitaxel with or without the PD-L1 inhibitor atezolizumab.

And this is the latest updated survival data, and what they found was that in those patients who were PD-L1-positive, there was a substantial survival benefit. And based on these data, the drug has recently been approved in the PD-L1-positive population. Knowing that, we need to decide whether we’re testing PD-L1 optimally or not.

There are 2 commonly used PD-L1 tests in the clinical practice. One is 22C3 and the other SP142, which is what was developed in this trial on the Ventana platform. And the question is, is one of these better at this purpose of identifying patients who’re most likely to benefit from a PD-L1 inhibitor?

And I think there were some really interesting data at ESMO from Hope Rugo, and they basically looked retrospectively in purely post-hoc nonplanned analysis at the samples from IMpassion130, the trial I just mentioned. And you can see on the left that SP142 detects a substantially smaller percentage of patients as being PD-L1-positive than 22C3, in the green there.

And that’s because, as you can see in that pseudo-Venn diagram in the middle that both antibodies detect about 45% as positive, but 22C3 detects an additional 35% of tumors. It appears that 22C3 is more sensitive, but what really matters is, how well does it predict benefit? And that’s where these data are really helpful. They looked at how the different subgroups, defined by these antibodies, benefited from the drug.

And you can see in the top row, those are the patients whose tumors were positive by both assays, and you can see a clear benefit, a hazard ratio of about 0.6, for benefit of atezolizumab. But in the second row are those that are only 22C3-positive, that extra 35% that that antibody’s picking up that SP142 doesn’t pick up.

And you can see there you really lose most of the signal for atezolizumab benefit and no benefit if you’re negative for both antibodies on the bottom there. This really would suggest, at least from the data we have from IMpassion130 and atezolizumab, that probably SP142 is the best antibody to be using right now. But I think clearly there’s room for us to improve, and I would love to hear my colleagues who use immunotherapies much more than us humble breast cancer doctors do if you have comments about that.

DR BENDELL: It’s really interesting many times it depends on the cutoff.

DR KROP: Right.

DR BENDELL: Like in gastric cancers — it’s funny, I get frustrated because I’ll order a PD-L1 IHC, and I’ll get the results. And it’ll just say CPS greater than or equal to 1, and I’m like, actually, I sort of care about 1, but I really care about 10. And they’re not reporting it out. And it’s interesting, and I’m sure they’ll present data on this as well with the SP142, is that as the percentage rises, do you split the curves further? And what’s your cutoff? And does that make more of a difference?

DR MCKENZIE: Wasn’t there a difference too in that study depending on where the PD-L1 was expressed? If it was expressed on the tumor cells or if it was expressed on the infiltrating lymphocytes?

DR KROP: With SP1, that’s a good question. There’s another data set that with SP142, essentially all the cancers that are positive in the tumor are also positive in the stroma. And you don’t pick up any additional tumors by looking specifically at the tumor cells themselves. Most of the positive cancers are in the stroma. And then there’s a few percent that are also — 9%, I think — that are also in the tumors, but it’s almost 100% overlap. I don’t know what the data are from the other tumor types about how that works. It doesn’t look like that is going to help.

DR MCKENZIE: I see. Okay.

DR BENDELL: And there’s been much question about correlation with PD-L1 levels but also tumor mutational burden. Did they show data there?

DR KROP: They haven’t reported TMB. I don’t think they reported TMB in IMpassion130. We’re waiting for, like, a full correlative analysis presentation. They haven’t given that yet. It’s basically just been mostly the clinical data.

Case (Dr Krop): A woman in her late 50s with de novo TNBC receives atezolizumab/nab paclitaxel as first-line therapy

DR LOVE: We’ll go through these couple of cases fairly quickly, just to demonstrate examples, and then we’ll spend the rest of our time going through your stuff and the Sarah Cannon experience. How about this 58-year-old lady?

DR KROP: Yes. This is one of the first patients I treated with a checkpoint inhibitor. She’s a 58-year-old high school teacher, no family history, no past medical history, and unfortunately presented with de novo triple-negative disease. Multiple liver lesions. Biopsied the liver lesions. It’s triple-negative, as was her breast. And it was positive for PD-L1 by the SP142 assay, which is what our pathologist had set up.

We started her, around that time, on the IMpassion130 regimen, nab paclitaxel/atezolizumab. She tolerated it very well initially. She had a near CR on her first staging just 3 months later, and we continued treatment. And then about another month later she started developing diarrhea, which was new. Some abdominal pain. We stopped the atezolizumab and did the usual workup for diarrhea, and her colonoscopy was read as consistent with checkpoint-related colitis. We put her on prednisone. Fortunately, her symptoms didn’t improve immediately but gradually improved. Her son was getting married over the summer. She made it very clear that we needed to get her symptoms to stop prior to that, which fortunately we were able to.

We then tapered the steroids off after 6 weeks. She still was in a CR. We restarted the atezolizumab last month, just recently. We stopped the taxane. She was starting to get a little neuropathy. And again, she had a CR.

And so far she has been doing well. It’s only been another month. She still gets some diarrhea if she eats the wrong thing, but otherwise she is doing well. This was, overall, a success story for us with immunotherapy. The first patient I’ve treated outside of a trial.

DR LOVE: Bad luck with that colitis, though.

DR KROP: Yes. The statistics say it’s not common, but we definitely have seen it.

DR LOVE: There it is.

DR KROP: Yes.

DR LOVE: Any other questions?

DR BENDELL: I was going to say, it’s just interesting, though, because the people who get colitis, at least in my experience, you get it, you treat it with a nice steroid taper, you’re able to start them back up again. And they’ll probably get it back again, and you just do these —

DR LOVE: Really.

DR BENDELL: — do these pulses of steroids.

DR LOVE: Repeated?

DR BENDELL: Yes.

DR LOVE: Huh.

DR KROP: I know that as long as it’s not Grade III or IV that it’s reasonable to restart, but I haven’t had experience kind of doing it cyclically —

DR BENDELL: Yes.

DR KROP: But you’re saying that’s worked for people.

DR BENDELL: It’s interesting, too, because once the patients have gotten it they know what they’re looking for the next time around. You usually catch it a little bit sooner.

DR LOVE: Interesting.

DR BENDELL: But you do your slow taper, get them down to 10 of prednisone, Grade 1, and then you —

DR KROP: Yup.

DR BENDELL: — just go back again.

DR LOVE: Huh.

DR ANTONARAKIS: The challenge with this is, you don’t know how much the atezo is contributing, right?

DR KROP: Right.

DR ANTONARAKIS: It’s not as satisfying in the sense that you might have had a miraculous response to nab —

DR KROP: Right.

DR ANTONARAKIS: — on its own. Although, since it was a CR and she had liver mets, you have to believe that it was somehow —

DR KROP: But there are patients — I mean, treatment naïve. Triple-negative breast cancer is a chemo-sensitive cancer. Again, point’s very well taken. This could all be chemotherapy, it’s just usually that’s not durable. Time will tell with that. But you’re right, we can’t tell the relative contributions.

Case (Dr Krop): A woman in her early 60s with ER-positive, HER2-negative mBC and a germline BRCA2 mutation whose disease progresses through multiple therapies attains a partial response to olaparib

DR KROP: This is one of my favorite patients who is a soccer player and coach, still at age 60, who 10 years ago presented with a left-sided breast cancer. This was not at our institution. High risk. She’s ER-positive, HER2-negative. She was treated surgically and found to be multiple nodes positive. She got standard chemotherapy at that time, anthracycline and taxane, postmastectomy radiation. She was put on adjuvant hormonal therapy.

And then about 7 years later, while she was still on the aromatase inhibitor, she presented with mental status change, which is unusual for ER-positive breast cancer. She was found to have a frontal mass, which was given — it was symptomatic and solitary. It was resected. Still showed ER-positive, HER2-negative breast cancer, as before. She also was found to have some bone lesions. The resection brain cavity was radiated stereotactically, and we put her on standard treatment with fulvestrant and palbociclib.

But she actually progressed relatively quickly, and we switched her to exemestane/everolimus and blew through that. Now had visceral metastases. That was when she came to us, and we said, “You can go on capecitabine as the standard therapy, or a clinical trial.” She went back to her oncologist and was on capecitabine for a short period of time.

But meanwhile she had signed up for a research study we do with our own mutation panel, similar to the Foundation panel, and it came back showing a BRCA2 mutation, a deleterious BRCA2 mutation. This is one of the founder Ashkenazi mutations. And she actually was adopted. And she had told us that she had had genetic testing when we first met her, but we didn’t have the results. And it turns out that she was mistaken. She didn’t have it. then when we found this, this was a surprise. But she was still doing okay on the capecitabine. She remained on that.

But in November of last year she progressed, and we started her on olaparib then, and she has had a PR and is tolerating it well. Her counts have been fine despite her pretreatment. And she remains in a PR.

This was a little bit of a wakeup call for me that again, I’ve got to be more careful about confirming that people had germline testing. But it is nice that you can still pick it up on somatic testing as well. And again, it’s obviously provided her another treatment that we wouldn’t have had otherwise.

DR LOVE: This BRCA2 with a bunch of numbers after it. Is that mainline BRCA2 or some unusual variant?

DR KROP: No, that’s one of the founder Ashkenazi mutations, it’s very well characterized. And it’s a frameshift.

DR ANTONARAKIS: There are 2 ways to classify the nomenclature of mutation. One is at the nucleotide level —

DR KROP: Right.

DR ANTONARAKIS: — and one is at the protein level.

DR KROP: This is the newer one.

DR ANTONARAKIS: This, at the protein level, is the Ser1982Argfs*22 mutation, which is present in 5% of all Ashkenazi people, whether or not they have cancer. One in 20 Ashkenazi people have this. This is the most common founder mutation in Ashkenazis. And if you look this up on ClinVar you have, like, 100 hits, and if you look this up on somatic databases you have zero hits. You could infer from the tumor testing that this was 100% germline.

The use of molecular tumor boards to review molecular profiling data and facilitate therapeutic decision-making

DR LOVE: Alright, let’s finish out, Andy. We asked Andy rather than presenting individual cases to kind of go through the Sarah Cannon experience. I don’t know of any other place in the United States that’s kind of carried out this model as far as you all have. I’m really curious to hear about it today.

DR MCKENZIE: Great. I mean, this has been a really great day, productive and very educational for me.

My thought always goes back to, how do we disseminate this type of information at large to our general medical oncologist colleagues who are fighting the good fight out in their communities, and how do we arm them with this information? And Sarah Cannon has taken a different approach to what we call molecular oncology support services. These are activities that our personalized medicine program puts together and offers to doctors out in their community oncology practices.

We started this work a couple of years ago based on data like this, which is — this is data taken from our drug development unit, where we took a look at the extreme responders that we saw on clinical trials and kind of bend them into which types of therapies they had received. A lot of our extreme responders, as we call them, received targeted therapies or immunotherapies.

And this kind of let us know that when we can really with high fidelity match a patient and their molecular profile to the right targeted therapy or immunotherapy, we have the potential to have some pretty significant responses. But it was, again, how do we communicate that information to our community medical oncology partners?

And we set out on an endeavor to create these molecular oncology support services, and at the foundation of that was access to the molecular profiling data. As most of you know, a lot of this testing occurs not in a centralized laboratory within our own institution, but we order these tests from commercial laboratory vendors that we have spoken a lot about today. They each do things a little bit differently, and the first challenge was, how do we coral all of this data so that we can have really high actionability on this data? And we’ve put together a lot of foundational IT infrastructure that I’m not going to talk about that connects all the pipes but then also allows us to communicate with these doctors out in the field through a virtual molecular tumor board type of delivery model.

This is kind of how that data flows. On the bottom there, the big 3 circles, is the typical treatment pattern here that you have, or activity that a doctor has. They order the NGS report. They receive those NGS results then create some sort of treatment plan. That’s kind of the 3 buckets that happen.

Our services come in and, like I said, harmonizing the data and then creating this learning IT infrastructure based on the recommendations that our molecular tumor board gives and in the responses that we collect afterwards for those patients. It all goes into this kind of learning system, IT system, that always gets better and smarter as we put in more data.

The ins and outs of how this works — we run what we call a molecular tumor board, and it’s basically variance annotation consulting work. A doctor can reach out and say, “What does this particular BRCA mutation mean? Is it inactivating? Is it one that would be sensitive to a PARP inhibitor?” And we do the work on the back end to annotate that variant for them.

And moreover, they can send us very complex NGS reports that we’ve seen here today, and we basically take those reports and run them through a sieve and say, “Worry about these. Don’t worry about these mutations, as far as actionability goes.” And that’s what the main gist of the program is. And like I said, we result that information back into this digital platform that our doctors have access to through throughout the network.

And that platform is called Genospace, there at the bottom. And Genospace is — they do all of the work in harmonizing the data and then providing a secure communication platform that we converse with our doctors back and forth with.

This is what we do in these molecular tumor boards or cancer conferences. We make a commitment to our doctors that we review every single NGS report that they order. It doesn’t matter the complexity. We review a lot of KRAS colon cancer patients. But we also review a lot of very complex cases.

And we decided to do this, because the academic model of picking a few interesting cases and talking about them in depth wasn’t something that our oncologists were interested in. They wanted to know from my patients that I’m seeing this week, what can I do for those patients?

And we do these molecular tumor boards at their individual clinics all across our network. That includes in Tennessee, in Florida, Denver and other places. And we do it regularly, occurring every 2 weeks or every month, depending on their caseloads, and we talk through suitability of patients for clinical trials in our network, targeted therapies, of course contraindications, potential germline considerations and others. All of that gets distilled into these tumor boards that we deploy locally, where they’re ordering the tests.

We do follow the AMP ASCO guidelines for variant calling and descriptions all the way from tier 1 to tier 4 FDA approval to likely benign variants. And that’s just a way to standardize things that our doctors can be assured that whether it’s me or someone on my team who’s analyzing these variants that we’re all following the same standard procedures to do that.

And then just a couple of examples I wanted to show here of how this molecular tumor board is delivered, what it looks like and the results from there. This is just a really quick head and neck cancer case that has 2 interesting HER2 mutations. But they both exist in the ligand-binding domain, and this is not a canonical HER2-positive as far as overexpression or amplified, but these are discrete point mutations. And the thought was, based on the literature, that these might be activating and potentially sensitive to inhibition with a targeted therapy. The patient enrolled on a targeted therapy clinical trial and had a very robust response. And that quote there is from the research note there that the patient returned back to his hobby of woodworking and reports a better quality of life overall. And this was based off of a discussion that we had with the doctor that decided to modulate their treatment plan based on the detection of this noncanonical HER2 mutations.

Similarly, we talk through cases like this where we have a variant that hasn’t been described before. This is an FGFR2 fusion. It’s fused with this gene VCL, and at the time there were no reports in the literature about this particular fusion, but we knew from other literature, fusions that occurred in that region were likely oncogenic. And we discussed a clinical trial. It was a targeted therapy with activity against FGFRs. And this patient also had a very robust response to a pretty common drug, regorafenib, in response there.

Those are the types of actions that we take out of these molecular tumor boards. And it works extremely well at an organization like Sarah Cannon, because we’ve got access, and broad access, to clinical research, even for doctors out in their community setting. And having access to these clinical trials really makes the difference with these molecular tumor boards, because we can discuss therapies that they otherwise might not have broad availability to.

And the last bit here, this is an actual — this is actually the My Pathway study that was discussed earlier. And the reason I bring it up is because we were a part of the molecular tumor board for the My Pathway study to actually determine whether or not a mutation was appropriate or not for the My Pathway study — representatives from Sarah Cannon were. This is an example of how you can actually take a clinical trial and execute a molecular tumor board for that specific clinical trial and hedge the bets for response based on an expert group of reviewers taking a look at these patients.

These are some direct outcomes from our experience of molecular tumor board. This graph on the left is the number of reports ordered per month, that line. And the vertical line in the middle there is when we initiated the molecular tumor boards at this 1 site, and then we measured, 6 months before and 6 months after, the number of molecular reports that were ordered, just as a way to see if we were encouraging doctors to order more just based on the fact that we were reviewing these cases. And what we saw was, when you add that support into their clinical decision-making, they’re more likely to order the test again in the future.

And this was a really important finding for us, that the doctors really wanted to do this but didn’t have the support in place to make the most of those reports when they ordered them.
And then at the bottom it just shows the types of studies that patients enroll to. We had an 18% accrual rate to clinical trials for folks who went through this program, which is fairly high, considering the national average. And they went on targeted therapies predominantly, but also some immunotherapy studies and others that you see there.

It looks like when you put these molecular oncology support services on top of a very busy community oncology setting, you can actually change practice patterns a little and encourage clinical research. And that was what our kind of main focus was for that.

This goes to show — that was 1 clinic example. We’ve done this kind of analysis in our network at large. That arrow there on the upper-left graph just shows when our molecular tumor boards were initiated network wide and the ensuing increase in molecular profiling that has occurred.

Now, that’s not necessarily causal. There’s a lot of things that happened in, when is that, late 2016/early 2017 from a targeted therapy and immunotherapy standpoint that all contributed to that. But it was a serendipitous time to actually start these molecular tumor boards.

And we see people profiling in a myriad of different disease settings. It’s not just for lung cancer. And we’re very excited about that uptick. And again, these are kind of physician behavior graphs here at the bottom. More physicians are ordering these tests, and those physicians that are ordering them are ordering more of them year over year.

And that was the takeaway, at least from our experience at Sarah Cannon. When you employ molecular tumor boards and molecular oncology support services to our community docs, it seems to be something that they gravitate towards, and they really want more of this kind of educational support.

And that’s it.

DR LOVE: Any comments or questions? This gets into the central issue of how do we improve what’s out there right now, particularly in terms of actions? I have a thought, but I’ll let you take your shot first, and then we’ll finish. Anybody have any thoughts about it?

DR BENDELL: I think we’re very lucky, but then, you ask, how is this scalable?

DR LOVE: Exactly.

DR BENDELL: I think we’re getting about 500 a month of new molecular reports. And I think in our database overall, we have just under 30,000 molecular reports. How do Andy and his team keep up? And you start to think about at some point whether or not you have to start to triage some of those reports out and accessibility.

DR MCKENZIE: Yes, and we’re very fortunate, too, that we work with some pretty smart and sophisticated software engineers that have built our genomics platform Genospace that now is, like I said, kind of learning from what we’ve put in there that when it sees a KRAS-mutated colon cancer it knows that we’ve got these studies available for those if they’re microsatellite stable or if they’re not and can kind of triage that decision support for us without us having to do a lot of the heavy work over and over again.

DR BENDELL: Where AI comes in.

DR MCKENZIE: That’s right.

DR LOVE: Subsequent to the recording of this program, on December 20, 2019, the FDA granted accelerated approval to trastuzumab deruxtecan for patients with unresectable or metastatic HER2-positive breast cancer who have received two or more prior anti-HER2-based regimens in the metastatic setting.

Also, on December 16, 2019, the FDA approved enzalutamide for patients with metastatic castration-sensitive prostate cancer.