Dr. Alan H. Bryce, MD, presented “Therapeutic Use of Genetic Testing in Advanced Prostate Cancer” at the 26th Annual Perspectives in Urology: Point-Counterpoint, November 11, 2017 in Scottsdale, AZ
How to cite: Bryce, Alan H. “Therapeutic Use of Genetic Testing in Advanced Prostate Cancer” November 11, 2017. Accessed Jul 2020. https://grandroundsinurology.com/Genetic-Testing-in-Advanced-Prostate-Cancer/
Alan H. Bryce, MD, argues that the urologic community should expand genetic testing in prostate cancer patients. With expansive, quality data, urologists will be able to better recognize aberrations in patients’ genes. Also, this would be a step closer to gaining FDA approved biomarkers for prostate cancer.
Therapeutic Use of Genetic Testing in Advanced Prostate Cancer
Okay, so this is the data from the Cancer Genome Atlas, TCGA, so hopefully you’ve heard of this. It’s a big nationwide effort looking at a variety of different cancers, really to try to get the whole spectrum. And what they do is they gather hundreds of cases in each of the cancers and do a comprehensive genetic profile so the DNA/RNA analysis, germline, and try to characterize cancers in a publicly available database, this is a research tool, hopefully to point people in the right direction and say, all right here is an encyclopedia to start with what does prostate cancer look like, what genes are involved? This should be hypothesis generating, and when you look at this, and May contributed a lot to this, and a lot of institutions did. Some of these things are already known, right? So we already know that ETS fusions, you know, the – – ERG and that family of mutations is very common in prostate cancer, so that is here. Then we see other recurrent families, a significant proportion of patients with mutations at SPOP and also in the – – genes, IDH, over here we see the P10 family, right, and P10 feeds into the Pi3 kinase pathway. And this becomes a—this is kind of the—this is the zoo of different types of prostate cancer if you will. I mean you look at this and you recognize with these different families, these different clusters happening, this ought to lead to different disease phenotypes. The genotype should at some level translate into phenotype, and if we’re doing a really good job, one would hope that this starts to inform how we’re treating patients. Not every patient is the same, that there is something we can say beyond Gleeson to start to talk about how we can approach prostate cancer.
That sounds intuitive, but the fact of the matter is we’re not doing this. The fact of the matter is they do this much better in almost every other cancer. But as we see, you know, and we’re familiar with this from having lived this for so many years, prostate cancer lags behind the other big four tumors in terms of how we are incorporating biomarkers, how quickly we get to new therapies, in terms of the amount of research that gets applied and how we bring that benefit to our patients.
So we’ve already looked at some of this data. This is just a different table organizing it a bit differently but again emphasizing the point. You know, here is TCGA is primary prostate cancer, prostatectomy sample is what I was just showing you. This is that metastatic prostate cancer database, and one of the things—this is the data we showed you earlier, one of the things that gets emphasized is the fact that as you move from primary prostate cancer to metastatic prostate cancer, the rates of almost all of these mutations increases dramatically. Okay, so cancer is not stable. It is inherently unstable and it evolves with time. It evolves from the time it goes from primary to metastatic and across the timeline of the metastatic disease it continues to evolve, so patients who get biopsied prior to first-line therapy for metastatic prostate cancer, have less mutated disease than patients who get biopsied after three, four lines of therapy. With every step in the process, the cancer looks different, and accumulates greater and greater mutational burden.
So this is data coming out of one of the stand up to cancer dream teams looking at metastatic castration resistant prostate cancer, so now we’ve moved from TCGA which is primary cancer to looking at germ line defects in metastatic disease, so really what the cancer started with, to now looking at what are we seeing in the castration resistant population. And this is immediately getting much more complex. Right? I mean these are cumulative mutations as the patient progresses through their disease course. At the very top here, and I know this is small, at the very top here is operations in the androgen receptor. Most of these are going to be amplifications of the AR, and then the ETS fusions, the P10 deletions, these are the things that are well known, these are the things that are going to be most common.
But other things that start to pop out, I’m sorry, this ends up being quite small. There are other pathways that are recurrently mutated here. You start thinking about Pi3 kinase mutations or P10 deletions happening in over 20%. If you start talking about AR amplification, we start talking about defects in DNA repair, so BRCA, ATM, some of these others, we start talking about defects in cell cycle regulation, CDK N2A, other familiar genes. We start talking about defects in cell signaling and these are all aberrations I’m highlighting here that in fact have targeted drug therapy available that have been applied in other malignancies. These are all defects for which these targeted therapies are typically not considered in prostate cancer. So these aberrations that we talk about are not unique in prostate cancer, and this is one of the key concepts I think that we have to understand is you know there’s only a certain number of pathways that drive our cells. There’s only a certain number of recurrent aberrations that are going to drive all malignancies whether you are talking about prostate cancer, or breast cancer, or colon cancer, or whatever else.
Prostate cancer is unique in that a healthy prostate cell is highly dependent on the AR and as it turns out the AR is probably the most powerful super regulatory gene in the human genome, it directly controls the transcription of more genes than any regulatory gene yet identified, which is why in prostate cancer aberration in a single pathway is enough to induce oncogenesis initially but as I say as we progress through the natural history of prostate cancer, the disease starts to evolve and I’m going to touch on this later when we talk about the—or I believe it’s tomorrow when we talk about other AR targeted therapies, but the disease evolves and really starts to move away from dependence on just AR. Okay, it starts to really move down the pathway towards being a generalized carcinoma not unlike the other malignancies, and that is what these highlighted regions are really talking about. These are common pathways that are aberrant in other cancers, lung, bladder, and what not, and pathways for which targeted therapies are potentially available. At the end of the day leaving AR and ETS fusions out of it, half of all patients will have some kind of targetable aberration if you add this all up, not mutually exclusive, right, a patient may have multiple pathways that are aberrant, but ultimately at least half of our patients have something going on and the fact of the matter is we never look for these, okay, not as standard of care.
So if we dig into this further, what we are seeing here is, A, we see copy number alterations so that it is more or fewer copies of a relevant gene suggesting the pathway is going to be more active or less. We see in B gene fusions. This is something that is often left out of the conversation, and when you’re talking about all of these gene panels, what we’re really talking about is point mutations when you do genome testing, but genome testing for the most part is going to miss fusion events, but it’s just because of the way the technology works, when we read across the genome, these sequencers they use inserts, and with the length of the insert you will pick up a point mutation but it is not going to read from Gene A to Gene B to tell you that these two genes are fused together. That information if it is picked up is generally discarded by the algorithms so if you want to find fusions at the very least we need to do transcriptomics, that is RNA sequencing, or if we know what we’re looking for we can do FISH testing. We can do break apart probes, and in full disclosure I have some patents on some break apart probes on some fusions that we found, but this family of aberrations can be drivers and it’s something that is not looked at in typical panels so you think about CML, Philadelphia chromosome, it’s a unique fusion protein that was discovered because it was visible on karyotype, but what we are finding now in the clinic when we’re doing this in the research setting is that a lot of malignancies will have driver fusions beyond what is known by any normal panel, but the fact of the matter is we’re not looking for them. Okay? So a driver fusion would be when you have a constitutively active gene, something that should be running in a cell, and the promoter region of that will then get fused onto an oncogenic gene, which normally should not be active, so when you think about Philadelphia chromosome BCRABL what you have is the promoter region for BCR driving ABL, ABL driving oncogenesis, so ultimately it is a question of dysregulation of a pathway, and what we find here is that fusion events are common in prostate cancer. In fact, Levy Garaway’s [phonetic] group has described this process that they call chromothripsis, and what we find in prostate cancer is that these events will happen very rapidly, typically all at once, you know, evolution and cancer cells is not a continuous smooth process, it’s really more of a staccato process. It happens in large leaps and bounds rather than at any kind of definable steady state. And fusion events will happen in a prostate cancer cell where there will be multiple fusions across the genome from chromosome 1 to 3 and that same partner in 3 to 6 and the same partner in 6 back to 1, the second partner in 1 over to 9 and 9 over to 23 or 21, and so you’ll end up with 10 fusions at one event happening immediately, and these can then end up being drivers.
What they looked at in this data set was they said which one of these events seems to be associated with aggressiveness, with enrichment in metastatic disease versus early disease, and no surprise P53 pops out, and P53 is the single most mutated gene across all cancers, right? This is a cell cycle regulator. When you lose it, you have increased mutational rate, and of course the AR pops out, but in other things reliably. I mean this passes the sniff test of things you would expect, BRCA, MML2, APC, and what not. Other things don’t appear to be relevant, but ultimately then again you can start to talk about what are the driver mutations, right? Amplifications, you’re talking about AR gene fusion is quite common, and then this list here breaking down bit further what I was talking about of actionable mutations in metastatic CRPC. We’re not looking for this, and in most prostate cancer patients no one looks for this information. But I always emphasize, you know, you can’t find it if you don’t look for it, and the eyes can’t see what the mind doesn’t know. If you don’t first become familiar with the data, become familiar with the aberrations that are accumulating, become comfortable with the data, then frankly presenting the data makes no difference at all. Repetition matters, familiarity matters. We need to know what we’re looking at when we see it.
So biomarker driven treatment is common in essentially every other cancer. When we think about the big four, breast cancer, prostate cancer, colon cancer, lung cancer, these four tumors represent 48% of all cancer diagnoses in the United States, excluding just the little skin cancers. Okay, so in breast cancer, you have ER, PR, Her2. In colon cancer KRAS, BRAF, NRAS. Lung cancer, a variety of markers EGFR, ALK, ROS-1. Melanoma only kills 8,000 people a year, a fraction of what we see with prostate cancer, has a long risk of relevant known biomarkers. In prostate cancer today there is not a single FDA approved biomarker because we haven’t incorporated this into the way we do clinical trials, into the way we think about our patients. We’ve been doing docetaxel for 20 years now, and we have a dozen negative docetaxel plus X studies where we added a second drug, and I would argue with you that this is a failure of trial design if you go back to the early Her2 studies in breast cancer, you are talking about 15% of patients having this aberration and if a patient has Her2 amplification in breast cancer, the likelihood of response is dramatically higher, several fold higher than if they are negative. And if they had done an unselected study in breast cancer then that study would have been at least 3,000 patients in order to try and detect a clinically significant difference when we talk about applying P value, and that is ultimately the trick in order to get FDA approval if you don’t have the biomarker. If you have the biomarker, then they can get away with a 500-patient study, and you have drug approval, and yet in prostate cancer we’re not incorporating biomarkers into any of our studies, so I would argue that progress in prostate cancer has been held back by nihilism, by conservatism, by a field where we say, look, we’re doing well enough, you don’t need to push, and you know, these patients they’re old, it’s all right, don’t hurt them. Right? We’ve heard this.
So how are we going to get there? I mean we’ve heard about BRCA. We’ve heard some about ARV7. ARV7 I mean if you haven’t heard about it, the issue here is that the androgen receptor can undergo splice variation where part of the receptor will be truncated. This was initially described by our colleagues at Mayo Clinic in Rochester, you know, back in the 2000—I think it was about 2004 that the first description of this came out, but here we are 2017 and you’ve only really been hearing about this in the conferences for about two years now, and so what ends up happening with ARV7 it’s an inducible biomarkers, something that can come and go in a prostate cancer cell. When you have a full-length AR, which is always present, it’s floating in the cytoplasm, and it needs to complex with testosterone in order to migrate into the nucleus, bind to the DNA and cause cell proliferation and growth. Okay? So the androgen receptor is a cytoplasmic protein that exhibits its activity by binding to DNA whereas ARV7 doesn’t need to complex with testosterone, can independently form a heterodimer either with another truncated ARV7 or over here with the full-length AR, and cause activation of DNA replication, so this is why ARV7 matters. It means that the prostate cancer can continue to use the AR pathway independent of the ligand, independent of testosterone, and so this leads to a predictable outcome, and this is a variety of small studies, all small studies, all single institution using a variety of different platforms to test for ARV7, but what you see here is that, and look here PSA response in patients with and without ARV7. The response in patients with ARV7, the abiraterone enzalutamide is essentially zero, here again 7%, here’s 0%, first is what we expect, this 40 to 70% in patients without ARV7, and in chemotherapy it has little impact. ARV7 does not predict for response to chemotherapy. This is the number we expect for cabazitaxel in the second-line setting if you go back to the TROPIC study. ARV7 appears to be a relevant biomarker for the prediction of response to androgen-directed therapy versus chemotherapy in prostate cancer, and yet the first clinical trial that tried to test this failed for lack of accrual, so this is how the field handicaps itself. We need to drive forward looking at biomarkers in order to better select care and risk stratify patients, but when the trial comes up we fail to accrue it.
I’m not going to go over this. You’ve already seen it in the previous talk, BRCA is a relevant target. So the question here, and this is what I’m getting at, this is how I think about cancer, and I’m going to show you this more specifically in tomorrow’s talk about AR-directed therapies. Traditionally, very simplistically we used to think about linear evolution in cancer, you have an initiating event, you have this carcinoma in situ population, this early stage population, but as the disease starts to accumulate mutations and become more aberrant, then you start to get more aggressive disease. But of course the reality is not like that. I think we all know this intuitively, right, evolution is branched, A can go in this direction, it can go in that direction, it can go over here, and then you get further populations and what you end up with once you get a tumor that is a centimeter in size you’ve got a billion cancer cells in there. And they’re not all identical. They’re cousins, and what you have then is heterogeneity, and this has been clearly demonstrated across a variety of malignancies now. In the early days of genomics, showing heterogeneity in one cancer was enough for a New England Journal paper, although one would think it’s intuitive, but now it’s demonstrated. We could say it’s true. And so this is a demonstration then of prostate cancer tumor heterogeneity, this is a patient in whom tissue was taken from multiple metastatic sites, you can see it over here, and all of the different sites were then sequenced, and this was done across a variety of patients, and the reason I show this is one of the immediate questions that come up is people say, fine, you biopsy a tumor, you found a mutation, but you know, you showed one site, and the other metastatic site has a different mutational profile, how are you going to use that information. This is the nihilism part of it. There’s too much data. You can’t possibly dig into this, but what we see from patients like this in fact is that while inter-tumoral heterogeneity does exist, the amount of heterogeneity between the different tumors in the patient is far less than the amount of heterogeneity that you see between patients, so Mr. Jones, Mr. Smith, their tumors look very different, but if you look at the five tumors within Mr. Jones, the fact of the matter is, yes, they are different, but they are not that different, and it goes back to the fact that tumor evolution is a cumulative process. The early mutations are the most likely to be drivers, and the fact of the matter is those early mutations don’t disappear, so they are still relevant.
Is it perfect data? No. We heard yesterday a lot of arguments about why it’s impossible to use the data meaningfully. Yes, it’s difficult, but it is not meaningless because it is—there are recurrent aberrations, we can start to develop themes, we can start to think about major pathways, and we can star to direct our research appropriately if we’re looking at the cumulative data.
This is a demonstration of ten patient stalking about the same thing, talking about how do the mutations happen and what these kind of charts are showing is here on the trunk, here are the mutations that are seen as common amongst all of the different metastatic sites, and we start to draw these lineage maps saying here is how the tumor evolves, here is another branch, here is another branch, this is really the reality of that branch versus linear evolution I’m talking about, but again if you look across, you can’t read it, but if you look across at the branches, the trunks that is for all of the patients, each line here is a different patient, you see in fact that the trunks are very different, again emphasizing the point that interpatient heterogeneity is far greater than intratumoral heterogeneity within a patient. Subclonal populations do share characteristics so even one biopsy is information we can act on.
And if you haven’t seen this picture, it is something that you do need to see. This was I think a seminal paper when it first came out. This was a series of rapid autopsies, so patients died of metastatic prostate cancer. They were taken for rapid autopsy. All of their metastatic sites were excited and then the clonal populations within the tumors were examined. Now, you know, for years we’ve heard arguments, some people would say, look lymph nodes can’t metastasize, all metastases come from the primary, take the primary out and you’re not going to get further metastases. XYZ, lots of arguments like that, and I think on the surface they were always to simplistic and didn’t make much sense. Cancer is not that fragile, but nevertheless, the arguments persisted. What this is showing is the migration of metastases, clonal populations so you go back to those branch maps I was showing you. We can demonstrate lineage relationships, familial relationships with those kinds of maps, and you look at this patient here and it’s a remarkable picture or this one, and what this is demonstrating is metastasis 1B came off very early from the prostate cancer, but then metastasis 1B ended up seeding a number of other populations. Some of these populations eventually fed back and repopulated the prostate, which then sent out further metastases, which then refed the prostate and fed other metastatic sites. Okay, so the point is the metastatic, the proliferative process is an interplay between all of the sites of disease. There is no magic in lymph nodes. There’s no magic in any of the metastatic sites. Where the cancer lives is not the key to how it behaves. So what does this mean in my clinic? There’s over 100 or over 200 approved targeted cancer therapies, approved, when you talk about those that are in Phase 1 clinical trials, a large part of my footprint is in the Phase 1 clinic, we can expand that list dramatically potentially targeted therapies that are approved, but when you talk about non-AR-targeted therapies in prostate cancer, you have got none, even though at least half of all prostate cancers will have aberrations that should be exploitable by currently available therapies for other cancers.
So the question is do I do tumor sequencing? Yes, I do. I approach tumor sequencing like people vote in Chicago, early and often. Okay? I like to look at tumor evolution. I look at the primary. I look at their first metastasis. I look at their next metastasis after every line of therapy and what I know from having done that over and over is that the tumors do evolve, that these mutations are cumulative, and that you start to get a sense of which one of these aberrations are happening early, which pathways are more relevant in hormone sensitive disease versus castration resistant disease versus post chemo, post abi, post enza, pick your drug, and this is important because for the medical oncologists, the treatment decision is not just about what I want to do today. When I’m thinking about my patient, I’m thinking about their entire disease course, I’m thinking about an entire treatment arc from today to five years from now, seven years from now, ten years from now when we’ve been through multiple lines of therapy. Because I need to think about how to sequence the drugs in the optimal way, and that is what I’m going to talk about more in tomorrow’s talk, okay? To an extent. We want to get the most out of the cumulative course of care, not just out of this step.
Clonality and evolution are long recognized clinical realities. We know this in the clinic, you know, you give a patient docetaxel and 5 of their 10 tumors shrink, 5 of them grow, these are different cell populations.
So the only way we are going to understand what are the differences between these populations is by examining them, right? Not just by watching them in our imaging studies.
This is Max Plank, you know, Nobel Prize winner if physics, German scientist had a lot of very good quotes back in the day and I find it useful to go back to. “When you change the way you look at things, the things you look at change.” If you don’t think about the genomic profile of a cancer, then you only have one cancer, prostate cancer is one monolithic disease without a recognition of the sub-populations and the sub-types that exist within that monolith. “Experiment is the only means of knowledge at our disposal. Everything else is poetry and imagination.” We are not trying to be snake oil salesman, just making things up so that it sounds good. What we’re trying to do is actually develop a database of knowledge that can lead to hypothesis-driven research. To get to experimentation, to get to definable knowledge that is translatable, and this last one I think is important. This is the only that gets paraphrased as science only advances one death at a time, “A new scientific truth does not triumph by convincing its opponents and making them see the light, but rather because its opponents eventually die and a new generation grows up that is familiar with it. So this is unfortunately all too true, right? A lot of times we stick with what we learned early on, or more to the point researchers are like parents, right, we all think that our baby is the prettiest, so we invested 30 years of our life proving a certain point, and then that point changes because new research comes around and it’s very unlikely that researcher is willing to give up on what they believe, right?
We don’t have to be stuck in that paradigm, and cancer therapy evolution is far too quick for that. In prostate cancer we have new drugs being approved every single year. Prostate cancer has dominated ASCO for the last five years or so with plenary after plenary after going decades with very little attention whatsoever, so sequencing certainly is about moving forward quickly. It’s certainly something that I’m not saying is easy to do, but it is something that we need to do in order to move the field. I know I’m out of time.
Ask the same question again, which of these pathways is commonly aberrant in prostate cancer? All right, very good. Something I said stuck. It’s good. All right, so the ARV7 question again. Which of the following is true regarding the impact of ARV7 on the response to therapy in castration resistant prostate cancer? I was right, no impact on the response to chemotherapy.