Faculty Presentation — Dr Antonarakis

DR ANTONARAKIS: I’m going to jump right in and start talking about the different types of DNA damage repair pathways in cancer and in prostate cancer specifically. There are 2 main ways that DNA repair occurs. The first is through a single-strand repair, and the second is through a double-strand repair. DNA can either fix its damage 1 strand at a time or 2 strands at a time.

There are 3 primary pathways involved in single-strand DNA repair. The first is the mismatch repair pathway, which we all know about because of the Lynch Syndrome and also because these patients are sensitive to PD-1 inhibitors, et cetera. The second is the nucleotide excision repair pathway, and the third is the base excision repair pathway. And this third one is important, because PARP1 and 2 are the main enzymes involved in that. For the audience to remember, single-strand repair, predominantly mismatch repair, and base excision repair, number 1 and number 3 are the most important and therapeutically relevant.

DNA can also be repaired 2 strands at a time –– these are the so-called double-strand repair pathways –– by far the most important mechanism here is the homologous recombination repair. And the genes that do this are the Fanconi anemia genes, the BRCA1 and BRCA2 genes. And then a list of other genes where we think they might be involved but we’re not 100% sure, including ATM, CHK2, PALB2.

I added CDK12 on this list. We previously thought this was a homologous repair gene. We are now thinking a bit differently about this, and I’ll get back to that at the end.

And then there are 2 other pathways which are more error prone –– the nonhomologous end joining or the translesion synthesis. These pathways only kick, to a large degree, when homologous recombination is crippled, typically by a mutation. And while homologous recombination is a high fidelity repair pathway, the last two, nonhomologous end joining and translesional DNA synthesis, so they make errors when they’re fixing double-strand damage. And again, they only kick in when the top one, homologous recombination, is deficient.

I now want to talk about how common these gene mutations are, and it turns out that the prevalence is a little bit different than localized prostate cancer and in metastatic prostate cancer. First I want to focus on localized nonmetastatic disease. There have been many studies published –– this was one of the nicest ones that I’ve seen –– where, in this case, 477 samples from localized nonmetastatic prostate cancer underwent either whole genome or whole exome sequencing. And the 5 genes that pop up, which add up to about 9%, are the DNA repair genes. These are, in fact, all homologous repair genes, and the ones that jump out at me are BRCA2, ATM and CDK12. Again, when you add them up, approximately 9% to 10% chance of having a DNA repair gene in a localized nonmetastatic prostate cancer.

Our group, led by a junior faculty member, Dr Catherine Marshall, decided to look at the publically available databases from TCGA and to ask a question, is there a relationship between Gleason score and these homologous repair gene mutations? The pie chart on the left shows that out of the publically available localized prostate cancer data from TCGA — and there were thousands of cases there, 8% of them have at least 1 DNA repair mutation — and the pie chart shows where that 8% comes from –– so 2.7% is ATM, 2% is BRCA2, 1.5% is CDK12 and 0.7% is BRCA1.

On the right, the bar graphs show the prevalence of any DNA repair mutation, shown in blue, or specifically BRCA1/2 and ATM, only those 3 shown in yellow. And Gleason grade group 1 is Gleason 3 + 3. Gleason grade group 2 is 3 + 4. Grade group 3 is 4 + 3. Grade group 4 is Gleason 8, and Grade group 5 is Gleason 9 and 10. What you can see is that there’s a stepwise increase, both in the total number of DNA repair mutations and BRCA1/BRCA2 and ATM specifically, up until about Gleason grade group 3, and then it plateaus.

The teaching point for me is, once you get to a Gleason score of 4 + 3 or above, then you have a higher risk of having a DNA repair mutation. And above that it doesn’t really seem to enrich any further. If you were designing a trial to enroll patients with localized prostate cancer, let’s say using a PARP inhibitor, you probably would want your eligibility to focus on, of course, Gleason 8 through 10, but also potentially the Gleason 4 + 3 = 7.

These repair defects are about twice as common in metastatic prostate cancer compared to localized. In this initial study — and there have been several others — we see a prevalence overall of about 20% to 25%, again in metastatic biopsies.

The most common gene here, BRCA2, and followed by ATM and BRCA1. Recurring themes keep coming up over and over again: BRCA2, ATM and BRCA1 are prevalent, about twice as common in metastatic disease than localized disease.

I know want to shift to discussing germline mutations. Up until this point, I’ve been talking about tumor mutations. What about the inherited germline mutations? And this field began with Colin Pritchard’s paper in the New England Journal 4 years ago, showing that in 692 unselected prostate cancer patients with metastatic disease, either presenting with metastatic disease or developing metastatic disease after primary local therapy, a surprising 12% of those men had at least 1 inherited germline DNA repair mutation. And this caught me by surprise and caught many of us by surprise. And again, the most common genes were BRCA2, about 5% of the total, CHEK2, ATM and BRCA1.

At Johns Hopkins we decided to follow up on this and working with a very talented fellow, Pedro Isaacson Velho. We looked at our own series of 150 consecutive patients, and we saw a very interesting association between a histological type of prostate cancer called intraductal carcinoma, which occurs about 20% of the time, and the presence of these germline DNA repair mutations.

The distribution was roughly the same as the Pritchard paper, with BRCA2 being the most common, followed by ATM, but if you look at the purple box on the bottom right-hand side of the screen, in patients who had intraductal prostate cancer on their histology, a whopping 40% had a germline DNA repair mutation. If they didn’t have any intraductal histology — in other words, if they had just pure acinar adenocarcinoma without any variant features, only 9% of those had a germline mutation. Based on this study and others, the NCCN guidelines have now adopted this and strongly recommend germline testing in anyone with any Gleason score if they have intraductal histology on their pathology.

And then another rare variant that’s even more rare is the pure ductal carcinoma. These are prostate cancers that do not arise in the acini, but they arise in the ducts themselves. These comprise only about 1% of all prostate cancers. They’re typically linked with low PSA production and hormone insensitivity. However, in this analysis of 50 consecutive ductal cases that was conducted by Dr Michael Schweizer, we found that half of these men, 49%, have either a mismatch repair mutation or homologous repair mutation.

These ductal prostate cancers are also enriched for some other weird stuff, specifically PI3 kinase pathway mutations, which we typically see more of in breast cancer, and also WNT pathway mutations like APC, and CTNNB1, which we most commonly see in colorectal cancer. These are definitely unusual genetic subtypes of prostate cancer.

One slide that I wanted to bring to everyone’s attention, and I always use this when I teach my fellows, is that homologous recombination mutations can be a consequence of mismatch repair mutations. This is an example of a real patient from my clinic.

As you can see there on the top, he has a mismatch repair mutation, MSH2, a very high mutational load and an MSI-high genotype. He also has 5 other mutations, as you can see on the bottom, including a BRCA2 mutation, a FANCM mutation, ERCC4, ERCC5.

And the teaching point here is, this patient is unlikely to benefit from a PARP inhibitor, because the driving event in this patient is not the BRCA2 mutation, it’s actually the mismatch repair mutation. This person, in my opinion, should be preferentially treated with a PD-1 agent not a PARP inhibitor.

I now want to review some of the national guidelines for germline testing as well as somatic genomic testing. And these are literally hot off the press, NCCN prostate cancer guidelines from March 16, 2020, Version 1.2020.

The first slide that I’m showing here are the current NCCN prostate panel recommendations for germline inherited genetic testing. I’ve underlined what I think is the most important: Any patient with a high-risk or a very high-risk prostate cancer or a patient with regional lymph nodes or a patient with M1 metastatic disease should undergo germline genetic testing.

If you reverse the coin on its head, this is anyone except for the intermediate- and low-risk prostate cancers. It’s basically anyone other than a Gleason 6 and some Gleason 7s.

If you don’t fit into that category but you have an Ashkenazi Jewish background, you need to get tested. If you have a family history of someone else in your family with a BRCA1 or BRCA2 or a Lynch mutation and you’ve got prostate cancer, even if it’s low risk, you should get tested.

And then anyone who has a strong family history of prostate cancer, defined as 2 first-degree relatives, for example, a brother or a father, with prostate cancer diagnosed below age 60, other than Gleason 3 + 3. Or if you have 3 or more family members with any other cancer diagnosed before age 50. These include breast, colorectal, endometrial, gastric, melanoma, ovarian and pancreatic.

But the punchline here is, anyone with metastatic disease should have germline testing. Anyone with high-risk or very high-risk localized disease should also undergo germline testing.

The second slide focuses on tumor-based somatic testing. And again, this is from the same guidelines recently published on March 16, 2020. And here it’s relatively simple: Anyone that has a diagnosis of metastatic prostate cancer. And then you should also consider it in anyone who has a regional prostate cancer –– meaning pelvic lymph node metastasis. This is a fairly broad group. And again, the main reason for this at the present time, even the PARP inhibitors are not currently approved as of today’s date, is because of the investigational use of PARP inhibitors that could be considered or the off-label use of a PARP inhibitor such as olaparib, or the use of a platinum chemotherapy for these homologous repair deficient. And, of course, the other reason is the FDA approval of pembrolizumab for MMR-deficient prostate cancers, which can, again, only be determined through the tumor testing. Just to summarize this one, somatic testing for anyone with metastatic prostate cancer, and strongly consider it in addition for anyone that has lymph node-positive disease as well.

You asked me to review some of the current testing platforms, and this is just my own list. Others may not fully agree with this list, but the key point is that we are now doing more and more panel testing with next-gen sequencing rather than looking at individual genes like we used to do 5 years ago or 10 years ago.

Looking at the germline component, this typically can be done either from blood, where we get the DNA from leukocytes, or from saliva. And the most common platforms that I have seen and that I use myself are the ones I listed. They have a multigene panel with 84 genes, which has about a 2-week turnaround time, and they do it from saliva.

In terms of the somatic testing, of course this needs to be done either by a tissue biopsy or a blood-based circulating tumor DNA. Let’s focus on the tissue biopsy first. I think the industry leaders right now are Foundation Medicine with their FoundationOne^® CDx platform, as well as Caris Life Sciences^®. And then there are few other panels: Tempus is one that is gaining popularity, PGDx and also multiple inhouse panels. As far as blood-based circulating tumor DNA, I think the Guardant 360 is the market leader and with FoundationOne liquid being a close second, as well as certain others.

In my practice, I always go after tissue first if I can. If the patient doesn’t have a metastatic biopsy that’s amenable, I like to use their primary prostate tumor, as long it was not more than 5 years old. And if they neither have a metastatic biopsy nor a primary tumor that’s accessible, then I will do a circulating tumor DNA in that setting.

I now want to move into the synthetic lethality hypothesis and to begin to talk about PARP and PARP inhibition. The word “synthetic lethality” was actually coined in 2005, 15 years ago. And the way I like to think about synthetic lethality is, 1 + 1 = cell death. You need to wipe out both single-strand DNA repair and double-strand DNA repair to lead to cancer cell lethality.

Let’s go through this diagram, because I think it’s quite informative. On the left-hand side I’m showing a cell that does not have a BRCA mutation –– this is a wild-type cell. This cell has a normal single-strand repair pathway and normal double-strand repair pathway. If you give a PARP inhibitor to that cell, nothing is going to happen. The cell is going to survive, because even though PARP is inhibited, the BRCA function is normal and it’s going to be able to fix double-strand DNA breaks that way.

In the middle scenario I’m showing a cell that has a single mutation in BRCA2, for example. This can either happen, for example, because of a germline mutation without a second hit or it can happen in a tumor cell that has a single BRCA mutation but not a double hit. In this case, the normal copy of the BRCA gene is sometimes able to rescue the deficient copy.

Even when you use a PARP inhibitor in this patient, some BRCA function will be maintained, and this cell may also survive because you did not achieve synthetic lethality, either.

On the right-hand side I show what would be expected to happen if you had a double mutation in BRCA2. This can either happen because of a patient with a germline first mutation with a second somatic hit in their tumor, or in some cases you can have a double somatic mutation that’s only found in the tumor, which is not inherited.

In this case, homologous recombination function is completely abolished. And, therefore, when you knock out PARP in this patient, you will not allow DNA repair to ensue, because both the single-strand function by PARP and the double-strand function by BRCA have been wiped out. Therefore, the cell will die. This is what we call synthetic lethality.

I want to talk briefly about PARP biology. I know this is a clinical audience, but just a few key points. Poly-ADP ribose polymerase is what it stands for. As we mentioned before, this is involved in single-strand DNA break repair through a process called base excision repair.

In the diagram on the bottom I show the PARP enzymes: PARP1, PARP2 and PARP3 as these little yellow and orange balls. And these get attached to single-strand DNA breaks. As I’m showing there on the bottom, you see the double helix. There’s a nick on 1 strand –– that’s what a single-strand DNA break looks like –– and then PARP enzymes get recruited to these double-strand breaks.

After PARP gets recruited, it then recruits other DNA repair proteins, as I’ve shown on the bottom right, and it causes these so-called branched side chain polymers to accumulate, and this is called PARylation, and this causes other DNA repair enzymes to be recruited to this site.

When you inhibit the PARP enzyme through drugs such as PARP inhibitors, as I’m showing on the diagram there –– there’s a red X on top of the PARP –– the PARP enzymatic activity or catalytic activity gets inhibited so the PARP is unable to recruit the other DNA repair enzymes onto that single-strand break. And then what happens is, during the S-phase, which is the DNA synthesis phase of cell cycle, those single-strand breaks get converted to double-strand breaks. By inhibiting PARP, it leads to a double-strand DNA break.

There’s a second theory that is getting increasing attention, which is this concept of PARP trapping. Some PARP inhibitors, not all of them, cause the PARP enzyme to bind onto the DNA, and it cannot come off. It cannot disassociate. It gets stuck there. It’s trapped. And when PARP gets trapped on the DNA, again, during the S-phase, synthesis phase, it can cause collapse of the replication fork and an ensuing double-strand break.

Not only is PARP enzymatic inhibition leading to double-strand breaks, but also, too much PARP binding and getting stuck on the DNA, not being able to disassociate from the DNA, which is called trapping, can also lead to double-strand repair deficiency.

This table here shows 5 different PARP inhibitors that have been tested in prostate cancer, and what I’m showing here is the distinction between the catalytic enzymatic activity and the PARP trapping activity. If you look at PARP1 IC50, the lower the number, the greater the inhibition. You can see that out of these 5 PARP inhibitors here, the strongest PARP1 enzymatic inhibitor is talazoparib, with the lowest IC50, second best being rucaparib.

When you look at PARP trapping, the more the pluses the better the trapping, so talazoparib, as you can see, has 4 pluses. That means that this is, at least theoretically, the strongest PARP trapper, and veliparib is the weakest PARP trapper.

Preclinically, it has been suggested that both the enzymatic inhibition and the PARP trapping are important. Clinically, I’m not myself convinced whether the trapping itself has clinical implications, but it might.

I did want to take a moment and talk about this interesting gene, at least in my opinion, called CDK12. We used to think of CDK12 as one of the homologous repair deficiency genes –– in ovarian cancer it might still be. In prostate cancer this really functions a little bit more differently.

And I was honored to write this editorial for The New England Journal of Medicine, which is based on a paper by Chinnaiyan’s group, published in Cell, and this figure summarizes the key points.

CDK12 mutations are found in approximately 7% of metastatic prostate cancers and only 1% of localized prostate cancers. They are very often biallelic mutations where both copies of the gene are mutated. And interestingly enough, these are never found in the germline DNA. Some people have hypothesized that if these mutations were in the germline, they would be so catastrophic that the embryo would not survive, that they would be embryonic lethal. That’s just a presumption.

In prostate cancers that have biallelic CDK12 mutations, instead of getting a homologous repair deficiency, you get this unusual pattern of genomic instability that is characterized by these tandem duplications, shown on the top right hand of the figure, where a piece of the chromosome, as shown in orange or purple, is duplicated exactly once, and the duplicated segment lies adjacent to the part that was duplicated.

Now, going to the bottom left of the figure, if these duplications occur within the coding regions of genes, in other words within exons, they lead to what is known as a fusion gene, where 2 protein coding sequences are juxtaposed to each other. These fusion genes are highly antigenic, because they create an amino acid sequence that is not present anywhere in the wild-type cell. And then if you move to the bottom right-hand part of the panel, these antigenic fusion genes sequences, if the tumors are exposed to PD-1 inhibition, could lead, at least in theory, to a favorable response to PD-1 blockade. Not PARP inhibition, however.

And this is the Cell paper that led to this hypothesis. If you look at the 2 graphs on the left, the blue color shows the CDK12 mutations, and these patients have a very high number of antigens from gene fusions. They also have high numbers of antigens overall, second only to the MMR deficient, which is even higher.

And on the right-hand side, this shows 4 patients with metastatic castration-resistant prostate cancer who received the PD-1 inhibitor, that initial paper, and 2 out of the 4 responded. It looked like about a 50% chance of responding to a PD-1 inhibitor if the tumor had a CDK12 mutation.

We became interested in figuring out how common this was in other cancer types, and we worked with Foundation Medicine, and we looked at 142,000 consecutive cancers. They’ve got an amazing database.

And what we showed is that it was perhaps not by accident that this was discovered in prostate cancer first. This is the most prevalent cancer type for this particular mutation. About 6% in the Foundation data set, but there are 10 other cancer types that have a prevalence between 1% and 5%, as I’ve outlined in the red box there, including ovarian cancer, gastric, cervical, salivary, breast, endometrial, bladder and colorectal. These prevalences are kind of on par with MMR deficiency, actually, with many cancer types.

And then what we recently published is, we looked at 60 consecutive prostate cancer patients with CDK12 mutations, and of those 60 patients, 9 received a PD-1 inhibitor, of course off label. And the waterfall plots are shown on the left. We found that 3 out of the 9 patients responded by PSA criteria. On the right we show the swimmers plots, and you can see that 1 of the responses is lasting more than 15 months. That patient had both a PSA response as well as an objective response. And then we had 2 other patients who either had a PSA and/or objective responses is lasting at least 6 months. Not everybody responds with CDK12 mutations to PD-1 agents, but probably these patients are not going to benefit from PARP inhibitors.

And so the way that I’m viewing prostate cancer in 2020, Neil, is, I’m thinking of this as a new molecular classification of the disease, just like we’re doing in non-small cell lung cancer, for example, or breast cancer, and this really hasn’t happened before in the prostate space. And I think we can identify at least 5 genomic subsets to date.

The top, of course, is the most common. These are the prostate cancers driven by the androgen receptor. And these patients typically fare well with androgen receptor blockers. The second are these homologous repair-deficient cancers that may respond from PARP inhibition. The third are the MSI-high or MMR-deficient prostate cancers. These might respond to PD-1 blocking agents. And the fourth, as we just discussed, are these CDK12 mutations that may also respond to PD-1-based agents. And the final group are these so-called neuroendocrine or small cell or AR-independent prostate cancers, and these are very hard to treat, typically do not respond to hormone therapies or taxane chemotherapies and might require other unique approaches.

In summary, I hope that I’ve convinced you that DNA repair mutations, both somatic mutations and germline mutations, are present in about 10% of localized prostate cancers and up to 25% of advanced prostate cancers. The early data seemed to suggest an enrichment in patients that have intraductal carcinoma or ductal carcinoma. If you ever see that, you should always be thinking about DNA repair mutations.

The NCCN guidelines now recommend germline genetic testing for all patients with high-risk, lymph node-positive or metastatic prostate cancer. Somatic genomic testing is indicated for all patients with metastatic prostate cancer. The concept of synthetic lethality means blocking both single-strand DNA repair using a PARP inhibitor and double-strand DNA repair by having a homologous recombination mutation.

It turns out that not all of these gene mutations are created equal, in the sense that some might predict favorable responses to PARP inhibition like BRCA1 and BRCA2, but not others. And so we have to find other ways to functionally characterize HR deficiency beyond just single genes. And finally, the field is paying a lot of attention now to these CDK12 mutations, which don’t appear to sensitize to PARP inhibition but might sensitive to PD-1 inhibition.

Discussion with Dr Love

DR LOVE: A couple of questions: Are there any data on checkpoint inhibitors in patients with these other tumors and whether responses have been seen?

DR ANTONARAKIS: The efficacy of checkpoint blockade, specifically PD-1 inhibition, has been quite sobering in prostate cancer, Neil. In unselected patients overall, we recently published the results of the KEYNOTE-199 study. This looked at 258 patients with metastatic castration-resistant prostate cancer. Unfortunately, we saw a 4% objective response rate and a 9% PSA response rate in the overall population.

Now, there are some data that are emerging suggesting that in addition to the mismatch repair mutations that we all know about, perhaps there is some enrichment for response to PD-1 agents in patients that have BRCA1 and BRCA2 mutations and perhaps ATM mutations as well.

This has not really been borne out in other cancers, like breast cancer or ovarian cancer –– somehow this seems to be unique to prostate cancer –– but I’m not sure why that is the case, why a BRCA2 patient or ATM-deficient patient might respond better to a PD-1 blocker. 

But the CDK12 thing appears to be real. It does not guarantee a response to a PD-1 agent, but of course an MMR-deficient mutation doesn’t guarantee a response either. There’s so much hype about MMR mutations predicting favorable response to PD-1 agents. But in prostate cancer, the response to a PD-1 agent in a mismatch repair-deficient case, it’s only 40%. That still means that 60% of prostate cancer patients with a mismatch repair mutation will not respond favorably to a PD-1 agent.

DR LOVE: Another question about the CDK patients. What do you see in terms of the PD-1 levels and tumor mutation burden?

DR ANTONARAKIS: The CDK12 mutations typically do not give rise to a high tumor mutation burden in the classical sense. The way that we are currently measuring and quantifying tumor mutational burden is, we’re looking at small mutations called point mutations or small insertions and deletions.

What we are not capturing are these big genomic rearrangements or fusion genes. And because CDK12 deficiency doesn’t lead to an increased number of point mutations or small insertions and deletions, what it does do is, it leads to increased numbers of gene fusions.

We might be miscoding these as low-TMB tumors, but they might in fact be high-TMB tumors. If we measure TMB in a different way, it also includes the gene fusions. But in the classical sense of the way that most assays measure TMB, tumor mutational burden, these would not be considered to be TMB high. With respect to your second question on PD-L1 presence or absence, yes, these tumors do tend to have higher expression by immunohistochemistry of PD-L1, at least much more so than unselected prostate cancer patients.

DR LOVE: Another question I have at a more basic biologic level is, sometimes you see NGS reports — I think I saw one on one of your patients –– where they say, “This looks like a BRCA germline.” How do you tell on a somatic assay like that whether it’s germline or not?

DR ANTONARAKIS: That’s a great question, Neil. I think that we should all be discouraged from using a somatic assay to figure out whether the mutation is a germline or not, but it can give us some hints that we can then confirm by doing dedicated germline testing.

There are 2 hints, Neil. One is the so-called allele frequency, or sometimes it’s called tumor allele frequency. This tells you what percentage of the cancer cells have that mutation –– it’s also called the mutation allele frequency. If the allele frequency is roughly 50%, this strongly implies that it’s a germline mutation. Most somatic mutations, the allele frequency is somewhere between 1% and 10%.

Now, not all of the NGS platforms report the allele frequency. Some of them do. Almost all of the circulating tumor DNA assays do report the allele frequency, but not all of the biopsy based.

If you see a BRCA2 mutation and you see an allele frequency of 49% or 51%, that doesn’t prove but strongly suggests that this is a germline mutation, because it means that half of the cells have it and half of the cells don’t.

The second way is the –– and this is a little bit more complicated –– there are databases that document and catalogue germline mutations. And it turns out, and nobody really knows the answer –– I certainly don’t –– there are certain hotspot mutations, for example in the BRCA2 gene, that almost never occur at the somatic level, but they almost always occur at the germline level. Many of those are genes that are Ashkenazi founder mutations.

There are certain mutations that, when you look them up in the databases, they have been reported 30 times in germline databases and 0 times in somatic databases. That’s another way that you can determine that, but that requires work on the part of the clinician and knowledge of where to search for them.

DR LOVE: The last thing I want to ask you about is what’s new, how things are going, so to speak?

When you went through your talk, you referred to a slide that had the NCCN guideline for March the 16^th, and I was thinking to myself, “Hmm, I wonder if that really applies right now,” even though it’s pretty recent, because things are changing quite a bit.

First of all, I’m just kind of curious in general, how has this situation affected your practice, and particularly your management of prostate cancer?

DR ANTONARAKIS: The coronavirus epidemic/pandemic has certainly changed things dramatically. In fact, the NCCN panel has issued new guidelines that were posted on how the practice guidelines are now to be adopted or changed in the setting of COVID-19.

I’ll just give you some general highlights from that panel and also my own practice. I think patients who do not need to be seen should not be seen in person. For example, the gentleman with a rising PSA that is just getting a PSA once every 3 months, he does not have to come in and see you.

The patients that are receiving intermittent hormone therapy probably could skip a dose of their leuprolide. It may not make a big difference.

Patients who are receiving long-term hormone therapy could receive longer leuprolide injections, for example — 6-month injections of leuprolide instead of 3 months or 12-month implants instead of 3 months, so trying to make them come into the hospital less often to get their androgen deprivation therapy.

In terms of chemotherapy, if it can be avoided, it should. One potentially controversial statement that we put out from the NCCN panel under the COVID-19 situation is that in the hormone-sensitive setting, when you are thinking about augmenting their systemic therapy and you have the option between ADT plus docetaxel or ADT plus abiraterone, apalutamide, enzalutamide, under the COVID-19 pandemic we are recommending against docetaxel in those patients.

Clearly, in the metastatic castration-resistant setting, if symptoms are there, specifically bone symptoms, we can justify giving the chemotherapy. But what we are doing here at Hopkins, Neil, is that if a patient is coming in for their chemotherapy, they’re meeting with the nurse, getting their infusion, but we are still meeting with them through a teleconference or videoconference.

In other words, even though they are there in the hospital already, instead of having 2 interactions with the care providers they’re getting their IV and their infusion by the nurse. They are then going home, and then we’re doing the teleconference with the physician on the same day to limit the number of actual human interactions.

In terms of our clinical trials, Neil, we have temporarily put on hold almost all of our therapeutic studies, with the rare exception of studies that have a very high theoretical risk of benefit versus risk — and I can tell you in prostate cancer, there are very few such studies –– but any study that was enzalutamide plus something else or docetaxel plus something else or hormone therapy plus something else, all those have been really put on hold, because we can’t demonstrate the additional investigational agent is going to have such an increased benefit compared to just the standard of care alone.

DR LOVE: I’m kind of curious what your sense is –– things are changing so rapidly –– in terms of the needs that are out there right now, particularly for general medical oncologists, which, as you know, is our primary focus, and they’re trying to keep up with every –– it’s almost every single cancer, and up until recently had so many new therapies and approvals, is a real challenge.

And it’s fine, the NCCN can put out these general guidelines, people can talk about general guidelines, but it almost seems like the entire algorithm for oncology is up for play now. That we have to relook at all the decisions we’re making and figure out, as you just said, is something like docetaxel in a patient up front, metastatic disease in prostate cancer, is that a rational therapy when you have oral therapies?

Any sense about what kind of interest there is, a general medical oncologist, in finding out what investigators are doing right now? Are a lot of people asking you?

DR ANTONARAKIS: I think there is interest from the community as to what is being done in academic centers. But I think that sometimes what we are doing here may not apply to them. We have 2 considerations: One is, what’s happening with our standard of care? And the second is, what’s happening with our clinical trials? And by the way, all the nontherapeutic trials, like imaging trials, biomarker trials, those are the first two to shut down.

One thing that I think is relevant across cancer types that we do not have the answer to, Neil, is whether or not the type of immunosuppression that occurs with more cytotoxic chemotherapies puts patients at risk for coronavirus. As you know, most cytotoxic agents induce neutropenia rather than lymphopenia, and I have not seen any good studies on that, so the jury is out. When we give our typical cytotoxic agents, does neutropenia increase the risk of getting a viral coronavirus infection? I don’t think we have the answer to that.

There’s also some interesting data about the potential skewing of deaths between male patients and female patients. These data came merely from China and from Europe, suggesting that if you got coronavirus infection and if you were a man you had a higher chance of having a fatal form than if you were a woman.

In the United States, the death rates so far seem to be evenly split, almost 50/50. But could there be a male predisposition per deaths? Is that related to comorbidities? Is that related to other factors like smoking, preexisting pulmonary disease? Those questions remain to be answered.

DR LOVE: One final question. I’m just kind of curious what you’re seeing in terms of the spectrum of response in your patients to the threat of COVID and how they respond when you discuss alternatives to try to navigate around that.

I was thinking particularly of situations where a treatment might have borderline benefit. How much do you find people kind of concerned about COVID as opposed to their cancer? I’m sure there’s a big spectrum, but what have you observed?

DR ANTONARAKIS:  At this moment, Neil, it depends on where the patient lives. Most of my patients from New York City are very happy not to travel outside their home and to skip their leuprolide injection for 3 months, especially the older men. And what I always tell those patients is, if you’re over the age of 70 or 75, even if you miss a 3-month leuprolide, it’s going to be 6, 9, 12 months before your testosterone starts to increase anyway, so it’s much safer to stay at home.

Patients from areas where the incidence is lower and perhaps they’re not listening to the news as much and they’re less scared, they’re more reluctant to even, for example, skip a leuprolide shot. And many times I instruct my scheduling staff to cancel the visit or do it by tele-visit, then all of a sudden the patient shows up and he’s there and he wants his leuprolide and he doesn’t seem as concerned.

It really is patient dependent. And I think it’s interesting, right now the patients seem to be dictating more which ones are being done by telemedicine rather than the physicians, although we are trying to do as many as we can from telemedicine.

Faculty Presentation — Dr Sartor

DR SARTOR: I’ll be talking about PARP inhibition in prostate cancer today. Very, very timely topic and one that has gathered a great deal of interest recently.

First of all, I’d like to give you a little bit of background, talk about the underlying pathophysiology and the genetic alterations in prostate cancer. Briefly discuss the mechanisms of action in PARP inhibition and some of the monotherapy studies, moving on to the Phase III study, which will shortly be appearing in The New England Journal of Medicine. And then a little bit about future considerations in combination studies, which is a big topic as we go forward.

Initially, I’d like to talk about targeting the DNA repair pathway, and I’ll be talking about PARP inhibition, but there’s going to be a whole wide variety of other targets. PARP is just one. We potentially have ATR inhibitors, ATM inhibitors, DNA-P kinase inhibitors and, of course, be able to target mismatch repair, not so much by inhibiting those enzymes but by looking at PD1 inhibition. What we’re broadly talking about is a subset of the broad field of DNA repair inhibition.

One of the things that we’ve learned, and this has been from beautiful science, is that DNA repair defects are going to be present in about 20% of advanced prostate cancer patients — BRCA2 being most common, but also ATM, BRCA1, CDK12, MLH1 — and as we have begun to study this disease in more detail, we’ve unequivocally demonstrated that DNA repair is an important part of the pathophysiology in an important subset of patients.

And what that does is give PARP inhibition an opportunity to have activity within these selected groups. PARP plays a key role in DNA repair, particularly single-strand DNA repair. It actually binds to areas of DNA damage and serves as a platform to recruit DNA repair proteins. It’s an enzyme, and it adds the poly-ADP-ribose units to a variety of target proteins and is part of the sensing mechanism in the early stages of DNA repair. This is called PARylation.

The most important of these enzymes — and by the way, there are a variety of PARP enzymes — but the most important one is PARP1, and when you inhibit PARP1 it leads to a series of double-strand DNA breaks that are unable to be repaired in cells that have no homologous recombination repair mechanism, such as a BRCA-mutated cell.

When you have a BRCA mutated cell and you create a double-strand break, encouraging PARP inhibition, you create cell death, and this is called synthetic lethality and is very, very important. Because actually, as it turns out, PARP inhibition does not play a large role in inhibiting the DNA repair if, in fact, you have an intact homologous repair pathway.

First I want to talk about the monotherapy trials in prostate cancer. We’re going to talk briefly about the PARP inhibitors first of all, and there’s a wide variety now –– there’s olaparib, veliparib, talazoparib, niraparib and rucaparib, and all of these are potent inhibitors of PARP1 and, to some extent, PARP2, and they have various effects on what we call PARP trapping, which is not only the inhibition of the PARP enzymes but the trapping of the actual PARP on the DNA, and that may or may not play a role –– we’re not certain at this particular point in time.

I’m going to be covering a variety of data and in particular those with olaparib, rucaparib and niraparib. And each of these are moving forward in the metastatic CRPC space. And each of these have had breakthrough therapy designations by the FDA.

The first trial, and this was in the New England Journal, a very important trial called TOPARP-A, and it used olaparib. And what they basically did was to look at individuals who had DNA repair defects, in particular BRCA2, ATM, as you can see on the left-hand side, FANCA, CHEK2, and they simply categorized those individuals who were responders versus nonresponders. And what they found is that about 88% of the patients who were biomarker-positive, found to be DNA repair defects, actually had some form of response to olaparib, whereas only 6% of the patients who were biomarker-negative had a response. That was important.

And it actually showed progression-free survival benefit and overall survival benefit in these patients. That created a lot of excitement back in 2015 when this New England Journal article came out.

The TOPARP-A was followed by TOPARP-B, driven out of the UK, and here we began to get more detail about changes in PSA, changes in baseline RECIST measurements, and clearly what was going on is that BRCA1/BRCA2 patients were dominating in response, but also responses were noted in PALB2, a little lesser extent in ATM, CDK12 and other.

And they began to look quantitatively at a whole variety of responses, RECIST, PSA and CTC conversion. What they found was, when you looked at the RECIST and PSA responses, it was mainly in that BRCA1/BRCA2, less so in ATM and not really in CDK12, but a little bit in PALB2, rarely in others.

Rucaparib is being tested in a trial called TRITON2. Now, one of the important elements that I should have explained perhaps a bit a moment ago was, when you look at the olaparib, it was all somatic alterations, getting tissue. In the TRITON2 trial, there were germline alterations that were allowed. And as you can see on the right-hand side is BRCA1/BRCA2, ATM, CDK12 and some others. These individuals had already progressed despite AR-directed therapies such as abiraterone and enzalutamide and a prior taxane.

And what they found was, BRCA1/BRCA2 again were the predominance of the patients, also a little BRIP1 thrown in. Not a lot in CDK12. Not a lot in ATM. Perhaps a little bit in FANCA. But the bottom line is, they were showing objective responses as well as PSA responses, and some of these were quite long.

With niraparib, there’s a different strategy toward the acquisition of the biomarker. Here they use ctDNA, circulating tumor DNA, and this was men with histologically, cytologically confirmed metastatic CRPC, again progressing after at least 1 line of a taxane-based therapy or an AR-targeted therapy.

When you look to the niraparib data, again you find the changes predominantly in BRCA1/BRCA2, occasionally in some of the others. But the good news is, you were seeing activity with this agent as well. And this is a ctDNA assay.

If we look at a brief summary, what we can see is that the BRCA genes are active in olaparib, rucaparib, niraparib studies and we’re getting objective responses, whereas with ATM and CDK12 responses are much diminished. There are occasional responses in other genes. I mentioned the PALB2 a little bit earlier.

Side effects. Haven’t really mentioned side effects today, but what I’ll say is, these are very similar to what we see in the nonprostate cancer patients: anemia, some degree of thrombocytopenia, a little bit of neutropenia, some fatigue, some nausea, rare vomiting, but it can occur. There are GI side effects and myelosuppression side effects, and these are fairly typical of what we’ve seen in other disease states.

It’s a Phase III — this was presented at ESMO, and it is also going to be published in the New England Journal. It’s been accepted in the New England Journal. One of the things that they did was look only at somatic patients, and they screened 4,426, which is a lot of patients got screened. Importantly, they only had DNA results from 63%, about 2,793. You have to realize that in about 37% of patients you’re going to try to get tissue, you’re going to try to get results, but you’re not going to be successful due to the quality of the tissue.

Now, 17.5% of their screened patients had a DNA repair defect, and BRCA2, which is most predominant, was in 6.1% of the screened patients. This was presented by Maha Hussain at ESMO, and as you can see, I’m one of the middle authors that have participated on the Steering Committee here.

Key eligibility criteria was progression after a prior novel hormonal agent such as abiraterone or enzalutamide. And it’s divided into 2 cohorts, Cohort A and Cohort B. It was stratification by previous taxane.

The patients were randomized to either olaparib 300 mg/BID or physician’s choice. And the primary endpoint was radiographic progression-free survival. Very importantly, at progression, the patients were allowed to cross over to olaparib. When we look at overall survival, we have to realize that a number of patients have crossed over.

The primary endpoint was radiographic progression-free survival. And here you’re looking at Cohort A, which is BRCA1/BRCA2 and ATM. It’s unequivocally positive. Hazard ratio 0.34, good confidence interval, p-value is less than 0.001. Clearly a positive study in Cohort A.

If you look at the interim overall survival, again we see a very positive trend, and this is actually in patients who were crossing over. As it turns out that about 80% of the patients in Cohort A crossed over. Here the hazard ratio is 0.64, did not meet the actual primary endpoint, because the alpha spend at the interim was 0.01, and you can see here the p-value is 0.0173. Formally it was not positive when it came to overall survival. But in my mind, it was quite significant because of this crossover as well.

If we get Cohort A and Cohort B combined, again you came out with a favorable hazard ratio of 0.67. I think that this will lead to FDA approval. Time to pain progression in Cohort A clearly positive.

When you look at the summary by the various cohorts here, you can see that you’ve got good hazard ratios with rPFS. The overall response rate, 33% versus 2.3%, and the overall survival as I just covered.

These are positive in Cohort A. In Cohort B, we’re clearly going to need to get a little more data.

One of the things that I thought was interesting about this olaparib Phase III was when they began to break it down by gene, they clearly saw that the BRCA2 was positive. CDK12, the confidence intervals were large and there really was not a lot of rPFS benefit. ATM, not a lot of rPFS benefit. Interesting, BRCA1 not a lot of rPFS benefit, and CHEK2 confidence intervals were large again. RAD51B and RAD54L are not shown here, but PALB2 also seemed to be positive.

We’re beginning to dissect out and look at these genes in some detail that might actually play a role in predicting responsiveness. And when we look at these Phase III trials, a variety of them that are looking at PARP inhibitors today, PROpel, KEYLYNK, and the PROpel is using olaparib and abiraterone. The KEYLYNK trial’s looking at olaparib/pembrolizumab. TRITON3 is going to be looking at rucaparib. TALAPRO is going to be looking at talazoparib and enzalutamide. And then MAGNITUDE is going to be looking at niraparib and abiraterone.

A lot of these Phase III trials are ongoing. But the one that is reporting is going to be the one with olaparib.

And there’s going to be a variety of AR inhibition immunotherapy trials. This is an area of active exploration. I might add, in very exploratory trials, we’re beginning to have a variety of the radiopharmaceuticals looked at as well.

One of the things I think is quite interesting is that we can image PARP expression by using F-18-olaparib or other PARP inhibitors that are bound to an imaging agent. And there’s a little bit of data now beginning to look at radiopharmaceutical therapeutics with binding to PARP activity. That’s another very interesting area.

In summary, DNA repair gene defects, both inherited, which we’ve known for a while in somatic, are common in metastatic prostate cancer, maybe about 20% of patients. The PARP inhibitors as monotherapies have high levels of antitumor activity in those who have BRCA2 and other selected genes. I think we have a lot of work to do with the optimal predictivebiomarkers. Is it tissue? Is it ctDNA? Is it germline? What type of assay? Exactly which genes? Is monoallelic or biallelic alterations required? What about mutational signatures? We have a lot to do in terms of the optimal predictive biomarker assays.

The combination of PARP inhibitors are in Phase III right now with AR pathway inhibitors and immune checkpoint inhibitors. And I think we're going to have an FDA approval this year, perhaps 2 FDA approvals this year.

There’s a lot of progress in this field. I’m excited to play a part.

Discussion with Dr Love

DR LOVE: A couple of follow-up questions, Oliver. You referred to the trials looking at combinations with, for example, hormonal therapy. And one of the trials I noticed there was talazoparib. And I didn’t see any information there on the monotherapy work that’s been done, if any, with talazoparib. Is there anything that we know about that clinically at this point?

DR SARTOR: Yes, very, very preliminary data with talazoparib. First of all, we know that it’s very active in PARP trapping plus a good PARP inhibitor. It’s quite potent in terms of the actual number of mg that are used. It does clearly have some activity. The big trial that’s going to be done in combination with enzalutamide and that trial is well underway. They initially had to do an analysis of interactions between the two, and there actually was an interaction requiring some dose modifications. But now the large trial is well underway. I’ll simply say that talazoparib has activity as monotherapy, but it’s a large exploration trial that’s going to be looked at in terms of enzalutamide in combination.

DR LOVE: Another thing I wanted to ask you about was when you talked about screening there for the PROfound trial, you mentioned that more than a third of the patients didn’t have tissue. Was this people –– were people with bone mets? Or they couldn’t — what was the clinical scenario?

DR SARTOR: Okay, not quite that they didn’t have tissue, Neil. What they lacked was a tissue readout. Many of the patients had tissue, but it failed in quality control. A lot of times these were prostate biopsies maybe from some years before, and they were sent off to Foundation Medicine. But when Foundation Medicine did the analysis, they could not get a quality readout, so they failed quality control. Tissue was available but actual data regarding the homologous recombination repair defects or not.

DR LOVE: Any thoughts about the best way to get tissue nowadays, a patient who maybe only had a prostate biopsy, just has bone mets? And what about liquid biopsies?

DR SARTOR: The liquid biopsies looked at specifically in niraparib trial, and they’re working really hard to get that up and going. What I’ll say is that if they are successful, and I think that there is still an if right now, a qualification, that it’ll be tremendous. Now, in the TRITON trial, I was careful to point out that germline defects were allowed. And that’s actually very important, because germline is easy to get. In terms of what tissue should you get, fresh tissue is always better. But as you know, many of these patients are bone only and getting tissue is quite problematic. I’ll simply say do the best you can, but ctDNA is a potential path forward and germline is a potential path forward as well in terms of the rucaparib.

DR LOVE: Another thing I wanted to ask you about is your personal experience with PARP monotherapy, whether on trial and off trial. And I guess the first question I have is, trying to look beyond the numbers, can you think of patients you’ve seen where you’ve given them PARP monotherapy and you really feel very confident they’ve had an antitumor response?

DR SARTOR: Yes, absolutely. And this is unequivocally active agents. Now, my experience has predominately been in the BRCA2 setting where I’ve seen these dramatic responses. One of the things that’s interesting, and this is not published — we’re going to be writing it up –– is to take patients who’ve been on platinum, and you’re probably aware that platinum has activity within this subset of patients, but a lot of times the patients become platinum intolerant either from allergic-type reactions or issues that may have occurred with bone marrow suppression.

We’ve added olaparib in that setting and have actually seen some very nice responses. These are not platinum resistant, but they are platinum pretreated. They could play a role in terms of maintenance going forward.

DR LOVE: Interesting.

Another thing I’m curious about is your experience with toxicity. I don’t know which agents you’ve used. A lot of people have had experience with olaparib. But one of the issues, and I think you referred to it with cytopenias. And I’m particularly curious about patients with prostate cancer because of the fact they have often bone mets, maybe marrow compromised. I don’t know if you have patients who have preexisting cytopenias from marrow involvement. What happens when you give them PARP inhibitor?

DR SARTOR: Yes, a great question, Neil. And that has been the number 1 problem. I’ve not had much trouble with nausea or GI effects, but the cytopenias can be significant. And we’ve had both anemia problems –– I’ve got a guy right now who’s responding to PARP, but I’m also giving him transfusions every 2 weeks because his red count just cannot tolerate the degree of PARP inhibition. But he’s responding very well, so I’ve been reluctant to withdraw the agent, and his counts are too low –– he’s also thrombocytopenic, so I’m titrating him down to platelets around 40,000, giving him red cells, and he’s borderline neutropenic as well. The cytopenias are a real issue and clearly needs to be evaluated carefully in prostate cancer.

DR LOVE: Maybe we could just shift over for a couple of minutes and get your thoughts in general in terms of managing GU cancers, particularly prostate cancer, in the current environment of social distancing. And what I’m hearing is, everybody’s trying to figure out ways not to have patients come in to be seen, not to get infusions, not to get diagnostic tests. How is this, putting aside the COVID-specific issues, but how has this situation affected the way you make your oncology decisions?

DR SARTOR: Oh, great question, Neil, and it’s pretty unique times. Everybody is finding their way forward. I’ve been instantly communicating with a variety of my friends, but it turns out we’re all pretty much doing the same thing. Number 1 is, we’re minimizing any of the routine visits, including even routine labs.

If the patients don’t need to be seen by a physician, don’t need to be seen in a medical facility, then we’re trying to defer that. And many patients who have relatively routine monitoring may be responding to longer-term ADT, abiraterone/enzalutamide, darolutamide, et cetera. Those types of patients we’re really trying to defer this type of visit.

Number 2, there are some patients who require their ongoing either radiation for palliation or they require ongoing chemotherapy. And we’re doing it. Every patient who comes in the clinic gets screened, both by temperature and questionnaire. If there’s anything suspicious are being evaluated by a physician. But those patients that need care are getting care within our center. We’ve got a lot of the virus around New Orleans. We have a lot of the virus around Tulane Medical Center. We’re being as cautious as can be. We are masking both the patient and provider right now and minimizing visits. But the key thing is, we are providing what we believe is necessary care to those who require it.

DR LOVE: I’m curious if you start to get into the granularity of decision-making, often times in oncology there are multiple options that are similar, maybe even identical, and I’m wondering again how some of these might play out. You and I just went through the topic of, for example, PARP inhibitors, in this case in prostate cancer. But I could see situations where people maybe are trying to decide between therapies. For example, let’s say a patient with metastatic prostate cancer who has a BRCA germline mutation who maybe has already been through primary endocrine therapy for metastatic disease. Would the fact that a PARP inhibitor would be oral kind of move it up earlier? And is that kind of some of your thinking?

DR SARTOR: Yes, absolutely, Neil, and I agree. And what we're doing is, we’re trying to defer the cytotoxic chemotherapy in particular. And PARP inhibitors would fall into that category. In addition, the hormonal therapies, where we’re trying to squeeze as much out of the hormones as we possibly can. And quite frankly we’re deferring chemotherapy right now because of the immunosuppressive effects plus the frequent visits and monitoring that is required.

DR LOVE: I’m going to say that in a way it’s almost like we instantly have to come up with a new set of guidelines in a way, or certainly, I think, docs in practice have always looked to investigators who are thinking these things through.

Let me give you an example of another decision: The patient who presents with metastatic prostate cancer. I’ve been hearing from you all that something like docetaxel is no better or worse than, say, an oral endocrine approach. Are you leaning more away in that situation from chemotherapy?

DR SARTOR: Absolutely. And now we have a multiplicity of FDA-approved agents. We have abiraterone, enzalutamide, apalutamide, all approved in that up-front metastatic, hormone-sensitive setting. And absolutely, I’m moving away from chemotherapy when I have a choice.

DR LOVE: Any other clinical scenarios that you’ve encountered in the last couple of weeks that you’re managing differently or maybe everybody you’re managing? I mean, I assume you’re –– are you already using a lot of telemedicine with your patients?

DR SARTOR: Yes, absolutely. The good news is that I have almost all my patients’ cell phones in my cell phone. We use a lot of text. Some email. Some phone calls, some telemedicine in the formal sense of it. But the key thing we’re doing right now, Neil, is trying to maintain communication, particularly with our critically ill patients, because they need communication and guidance, and we’re trying to provide it.

For the patients who are more in the monitoring scenario, quite frankly I’m telling them, “Check back with me in a month. I’ll tell you what I’m doing in a month.” Right now we’re trying to prioritize those patients that are acute and deprioritize those patients that are not.

DR LOVE: Are you trying to get away from specific drugs right now, for example, IOs? I would guess they’re a neutral issue about COVID. Or do you see it that way? Neutral?

DR SARTOR: Yes, I see the IOs as being neutral. We’re continuing our bladder and renal patients with the IOs.

DR LOVE: I was going to ask you about the renal, too. I would assume that procedures like transplant are, like, pretty much on hold?

DR SARTOR: Yes, absolutely. Our transplant team is pretty active right now. And there are some that are absolutely essential, where transplant is really the only option. And, of course, these are for the hematologic malignancies, and they are proceeding with transplant, but that’s only life or death. I mean, some patients, they don’t get transplanted, they don’t live.

Faculty Presentation — Dr Agarwal

DR AGARWAL: I am going to discuss novel combination approaches with PARP inhibitors in metastatic castrate-resistant prostate cancer and the rationale behind ongoing research. Before I delve into the trials using PARP inhibitors, let me briefly review its mechanism of action.

This review describes the pathways leading to PARP activation and how PARP inhibitors may lead to cell death. PARP is recruited to the site of single-strand DNA damage, which leads to the activation of base excision repair pathway and ultimately cell survival. In the presence of PARP inhibitors, an alternate mechanism of DNA repair mediated by homologous recombination repair pathway gets activated, and cell is still able to survive. The DNA repair mechanism remains intact. However, in the presence of PARP inhibitors, in underlying homologous recombination repair deficiency, the DNA is not able to repair, which leads to accumulation of double-strand DNA breaks, and that leads to cell death.

Alternatively, certain PARP inhibitors, such as talazoparib, traps PARP at the site of DNA damage, and that prevents the DNA replication and leads to replication fork instability and cell death. And this can happen regardless of homologous recombination repair defect.

Let’s move on to the rationale for combining PARP inhibitors with androgen access inhibitors in nonhomologous recombination repair-mutated metastatic castrate-resistant prostate cancer. There are multiple nonclinical and translational studies which suggest the potential for synergism when PARP inhibitors are combined with AR signaling inhibitors. And there have been multiple publications, as mentioned here, but mostly what we have observed based on those data is that inhibition of AR signaling suppresses the expression of genes associated with DNA damage response.

Also, the PARP activity has been shown to support the function of androgen receptor, suggesting that coblockade of PARP may synergize with androgen receptor-directed therapy. In addition, clinical resistance to AR blockade has been shown to associate with codeletion of RB1 and inactivation of BRCA2, which basically means inactivation of BRCA2 may lead to PARP inhibitor sensitivity.

Based on multiple laboratory translation data, this trial was conducted and reported last year. This was a Phase II trial of the combination of olaparib, a PARP inhibitor, with abiraterone, an androgen receptor, signaling inhibitor by virtue of targeting the intratumoral production of testosterone by inhibiting the enzyme CYP17 hydroxylase. In this trial, 140 patients were randomized to the combination of abiraterone with olaparib versus placebo plus abiraterone. The key eligibility criteria are here, and they included patients who were candidates for treatment with abiraterone, and they must have had received docetaxel chemotherapy for metastatic castrate-resistant prostate cancer. And patients needed to have a decent performance status.

The primary endpoints included radiographic progression-free survival but also the assessment of adverse events and number of patients who progressed on these individual combinations. Secondary endpoints included overall survival, second progression or PFS2, time to first subsequent therapy and so on.

Let’s look at the baseline characteristics of this patient population. And to make sure that these 2 arms were even as far as disease and patient characteristics were concerned, and especially from the perspective of underlying HR mutation status. The median age was similar. The baseline PSA was higher numerically in patients who are on combination arm, thus tilting the balance against the combination arm. However, as far as homologous recombination repair mutation status was concerned, they were pretty evenly balanced in both arms. Equal number of patients had received prior cabazitaxel, and to bring to your attention that docetaxel was required before patients went on to be registered on this clinical trial. One hundred percent of patients had received docetaxel in metastatic castrate-resistant prostate cancer.

This is the Kaplan-Meier curve showing the radiographic progression-free survival in the intent-to-treat population. And we can see here that radiographic progression-free survival was significantly improved on the combination arm versus the control arm with abiraterone therapy alone. The hazard ratio for radiographic progression-free survival was 0.65, favoring the combination therapy arm, which basically translates to 35% reduction in risk of disease progression in those patients who were receiving combination therapy with olaparib and abiraterone.

Let’s move on to the Phase III trials. Multiple Phase III trials have been started based on the strong laboratory and translational data and the very exciting Phase II data of the combination abiraterone plus olaparib. The results we just discussed. Let’s look at those Phase III trials.

The first trial is the PROpel Phase III trial, which is basically using the same combination which was used in the Phase II design. The difference is, this is a much larger trial with the registration intact. In this trial, 720 patients are being randomized to abiraterone with olaparib versus abiraterone plus placebo. The key eligibility criteria include progressive metastatic CRPC. Patients could have received docetaxel chemotherapy in metastatic castration-sensitive prostate cancer, but no patient is eligible if the patient has received any therapy in the castrate-resistant prostate cancer setting. The main stratification factors are bone versus visceral metastasis and whether the patient received docetaxel chemotherapy in the metastatic castrate-sensitive prostate cancer setting or not. The primary endpoint is radiographic progression-free survival, and there are several secondary and safety endpoints.

The next trial is the MAGNITUDE trial, which is a large Phase III trial of 1,000 patients, which is randomizing patients to abiraterone with niraparib, another novel PARP inhibitor, to abiraterone plus placebo. Again, if you look at the eligibility criteria, they are very similar, with some differences, which are subtle, such as progressive metastatic CRPC. Patients are required to have castrate-resistant prostate cancer, which is metastatic. However, patients could have received not only docetaxel but also novel AR inhibitor in the castration-sensitive prostate cancer setting. In the castrate-resistant prostate cancer setting, patients are allowed to have less than 4 months of abiraterone therapy. Meaning patients who have started abiraterone in metastatic castrate-resistant prostate cancer setting, they are allowed to enroll on this clinical trial as far as the duration of abiraterone is not more than 4 months.

Notably, the 2 cohorts which are being tested separately in this trial, Cohort 1 patients are positive or have to be positive for DNA repair defect in the tumors, while Cohort 2 patients do not have to harbor any homologous recombination repair-rated mutations. The primary endpoint is radiographic progression-free survival, and there are multiple secondary endpoints, which include overall survival and time to symptomatic progression and time to initiation of cytotoxic chemotherapy.

The final trial in this group is another Phase III trial, known as TALAPRO-2 trial, which is testing the combination of enzalutamide plus talazoparib at 0.5 mg daily, not the traditional doses of talazoparib, which is 1 mg daily. And these patients are being compared with patients who are being randomized to enzalutamide at standard doses plus placebo.

Again, there are very similar things among these 3 trials, which include progressive mCRPC. These patients have to have progressive castrate-resistant prostate cancer. They cannot have a very symptomatic disease. No prior treatment for castrate-resistant disease is allowed. However, 1 difference in this trial is, this trial allows prior treatment with abiraterone in metastatic castration-sensitive prostate cancer setting. And of course prior treatment with docetaxel is also allowed in metastatic castration-sensitive prostate cancer setting, as allowed by other 2 trials, which were PROpel and MAGNITUDE trials.

There are 2 different cohorts which are being tested. One cohort is of those patients who are unselected for DNA damage repair defects or homologous recombination repair defects in their tumors. And the cohort second, which will have 268 patients, they will have to have homologous recombination repair defect in the tumors to be eligible for trial. The Cohort 1 is a large cohort of 750 patients, and these patients do not have to have a DNA repair defect. And the other cohort is of 268 patients who have to have DNA repair defect. And patients would be stratified in the Cohort 1 based on prior treatment with novel hormonal therapy, which is abiraterone in this case, or taxanes versus no, and whether they have DNA repair mutation status available, meaning are they deficient or not deficient versus unknown.

This is a unique aspect of this trial, which allows every patient regardless of DNA repair defect in the Cohort 1. And there is exclusive cohort of DNA repair defect-positive patients who will be in Cohort 2. Radiographic progression-free survival is the primary endpoint, and secondary endpoints include overall survival and, again, time to PSA progression, time to initiation of cytotoxic chemotherapy and in addition to multiple other secondary endpoints.

Let’s look at those trials and how they compare with each other. Key eligibility criteria remain very similar. No prior therapy for metastatic castrate-resistant prostate cancer in all these 3 trials, except in MAGNITUDE trial, which allow 4 months or less of abiraterone therapy in the castration-refractory metastatic setting. And these patients have been allowed to be randomized to abiraterone with niraparib versus abiraterone plus placebo. Otherwise, other trials, like PROpel and TALAPRO-2 trial, do not allow any treatment in metastatic CRPC setting.

The sample size of all these trials are relatively large, with 1,018 patients in TALAPRO-2 trial, to 720 patients in the PROpel trial. The stratification factors, they vary but not much, and mostly whether they received intensified therapy in metastatic castration-sensitive prostate cancer or not, with docetaxel chemotherapy in the PROpel trial and in the context of TALAPRO-2 trial where patients received novel hormonal therapy or taxane versus not. And please know that novel hormonal therapy in TALAPRO-2 trial remains only CYP17 inhibitors such as abiraterone or patients who received experimental therapy with orteronel, or TAK-700. But patients who have received apalutamide or enzalutamide in the castration-sensitive metastatic prostate cancer setting are not eligible to enroll on the TALAPRO-2 trial. The primary endpoint for all these 3 trials is radiographic progression-free survival.

Before I move forward with other combination therapy, I would like to briefly mention the results of TALAPRO-1 trial, which is a Phase II study of 100 patients with metastatic castrate-resistant prostate cancer who are harboring DNA damage repair mutations. The prelim results were presented by Dr Johann de Bono in GU ASCO 2020. And results were very exciting.

Overall, the responses were, if you just include all patients, regardless of what type of DNA repair defects they have, whether they had BRCA1, BRCA2, PALB2, ATM and many others. The overall objective responses was 25%. However, if you look at BRCA1 and BRCA2 mutations, the objective responses were much higher, at 68%, and much lower, still present, in at least 1 patient who had ATM mutation, had objective response, out of 15 patients with ATM mutation.

This was something which I thought was novel, because so far we have not seen any responses with PARP inhibitors in patients harboring ATM mutations in the tumor. Interestingly, the PSA responses were also present, of the same magnitude, as we saw with the objective responses. Patients who had BRCA1 and BRCA2 mutations, the PSA responses were 64%, and they went to 33% in patients who had PALB2 mutation.

Let’s look at the progression-free survival of this cohort. Overall, the radiographic progression-free survival in all patients, and please note that at the time of interim analysis, when the results were presented, the data were available only for 53 patients out of 100 patients who were enrolled or who were going to be enrolled on the trial. Of these 53 patients, regardless of their DNA repair defect or type of DNA repair defect, the median radiographic progression-free survival was 5.6 months. But again, interestingly, the median radiographic progression-free survival was much higher, at 8.2 months, in patients who harbored BRCA1 and BRCA2 mutations. And they were lower in other mutant types. Although it is hard to come up with a median progression-free survival for each of these subcohorts according to the type of DNA repair defect, because the sample size of each one of those become very small.

Now let’s look at the TALAPRO-2 trial. The safety and tolerability of enzalutamide plus talazoparib combination. This was a uniquely done Phase I trial which was embedded within a Phase III trial so that we didn’t have to do 2 different trials with 2 different protocols. In this clinical trial, which is called Part 1 TALAPRO-2 trial, patients who had progressive mCRPC were treated with enzalutamide at standard dose with talazoparib at 1 mg.

The primary endpoint of this trial was to confirm the right dose of talazoparib when used in combination with enzalutamide. And the trial allowed decreasing the dose of talazoparib to 0.5 mg, depending upon the PK data and the toxicity data. The eligible populations were very similar to the randomized portion of the TALAPRO-2 trial, which I have just described. It included metastatic CRPC, and patients could not have received any therapy in the CRPC setting.

As we can see here, the toxicities, the hematologic toxicities, which are traditionally associated with talazoparib, were more pronounced in the talazoparib at 1 mg dose with enzalutamide at 160 mg dose compared to talazoparib at 0.5 mg in combination with standard dose of enzalutamide. We hardly see many hematologic toxicities at 0.5 mg of talazoparib versus 1 mg of talazoparib when combined with enzalutamide. The question is why it may have happened, or we were seeing higher hematologic toxicity with talazoparib.

These are the PK data. We can see here talazoparib exposure, when combined with enzalutamide 160 mg daily, was actually higher than the talazoparib exposure reported in the EMBRACA study when talazoparib was used as monotherapy at 1 mg daily. And this likely happened because of the inhibition of drug efflux transporters such as P-glycoprotein by enzalutamide. The geometric mean ratio of talazoparib 1 mg dose normalized 80 under the curve in the first 24 hours, at week 9, when combined with enzalutamide, compared with that reported in talazoparib monotherapy trial, was 157%.

We can see, like, trough levels of talazoparib, the AUC of talazoparib, were much higher when compared with enzalutamide at 1 mg daily compared to when talazoparib was used at 0.5 mg daily. And then when talazoparib was used at 0.5 mg daily in combination with enzalutamide standard dose, the drug levels were very similar to what we see with talazoparib at 1 mg daily as a monotherapy.

And it makes perfect sense why we were seeing higher hematologic toxicity when talazoparib was being used at 1 mg in combination with enzalutamide. And those toxicities improved rapidly, actually. We did not see the level of toxicity close to what we saw with 1 mg daily at 0.5 mg daily. Interesting data to share, and basically it tells us the importance of the pharmacokinetic studies when we are doing these combination trials.

Notably, if you look at the PSA reductions or PSA responses they were very similar, whether talazoparib was used at 1 mg daily with enzalutamide or talazoparib was used at 0.5 mg with enzalutamide. And based on these data, the Phase III dose for talazoparib was chosen as 0.5 mg instead of 1 mg in combination with enzalutamide. The TALAPRO-2 trial Part 2, which is the randomized portion of the trial, is using this dose of talazoparib 0.5 mg with enzalutamide standard dosing.

Let’s look at another novel combination which is being used in combination with PARP inhibitor, which is the combination of PARP inhibitor with checkpoint blockade. Before we talk about those clinical trials, let’s look at the biological rationale for combining these 2 agents. We can see here that besides the direct cytotoxic effect of PARP inhibitors, PARP inhibitor can also exert antitumor immunity.

When double-strand DNA breaks are induced by PARP inhibitors, these generate the cystosolic double-strand DNA fragments that binds to cGAS, and that leads to activation of STING, or we call it stimulator of interferon genes. When STING gets activated, it leads to activation or increased production of type I interferon. And when that happens, that leads to increasing recruitment of T cells, but also, at the same time, we also see higher upregulation of PD-L1, programmed death-ligand 1.

This is very interesting to see that PARP inhibitors cannot only lead to upregulation of STING pathway and type I interferon production, but at the same time it also activates those pathways which are inhibitory to T cells. And this makes sense to us based on these data that when we are using PARP inhibitor from the perspective of immunotherapy. We combine them with checkpoint inhibitors so as to continue with the positive aspects of PARP inhibitors and immune therapy, but we also get to inhibit the negative aspect of PARP inhibition-associated immune upregulation by using checkpoint inhibitors.

Based on these rationales, studies have been started, and one of the first studies, studies like PARP inhibitors and PD-1 inhibitors have been started. The first of these studies was the combination of durvalumab with olaparib. This was a Phase II trial reported by Dr Karzai around 2 years ago. In this study, patients who had progressive metastatic CRPC and who had progressed on enzalutamide and/or abiraterone were enrolled on this trial. This was a single-arm study, and they received durvalumab with olaparib. Durvalumab was standard dose, 1,500 mg intravenously every 4 weeks, and olaparib at 300 mg twice daily. The primary endpoint was clinical efficacy, and secondary endpoints were response rates, safety, duration of response and PSA response.

Interestingly, there was no limitation on previous standard therapies for metastatic CRPC population, so the patient could have received any number of prior therapies in castration-sensitive or castration-resistant setting.

Let’s see what the responses were. We can see here about 9 out of 17 patients had a PSA response of 50% or more. Nine patients had PSA responses. And here in this waterfall plot of PSA responses, we can see many of these patients also had RECIST responses. Those patients were colored by green, or who had stable disease colored as blue here, or patients who did not have any response colored as red also did not have any PSA response. Obviously that data are encouraging. Many of these responses were durable. Obviously encouraging data.

Let’s look at the biological basis why these patients would have responded. Let’s look at those 9 patients who had a PSA 50% response. And many of them also had radiographic responses. Four of these 9 patients had DNA damage repair defects in their germline. Two of the patients had somatic mutations in their DNA repair pathway, and 2 of these patients did not have any known mutations. Obviously it looks like most of the responses were happening patients who were already harboring DNA repair defects, either in germline or somatic mutations inside the tumors.

The probability of progression-free survival at the median follow-up was close to 8 months or 9 months, so probability of PFS based on DNA damage repair status was higher in patients who were harboring either germline or somatic mutations versus patients who did not harbor any of these mutations. The progression-free survival was longer in patients who were harboring DNA damage repair defects.

Let’s move on to the next trial. KEYNOTE-365 trial with pembrolizumab plus olaparib. This is also a single-arm study, Phase II trial. Patients who were eligible — and eligibility was defined as patients who had received docetaxel chemotherapy in the metastatic CRPC setting and at least 1 NHT in the metastatic CRPC setting.

These eligible patients would enroll on this single-arm study with pembrolizumab 200 mg every 3 weeks, which is the standard dose of pembrolizumab with olaparib 400 mg bid. The endpoints were safety and PSA responses.

Patients who had RECIST measurable disease, 14% of patients had objective responses. Only 1 out of 13 patients who did not have RECIST measurable disease had response, so 8% response. If you combine all patients, 12% of patients in this study had objective responses.

Again, we are seeing some signal of responses. Based on this earlier data a Phase III randomized trial has been started, which is randomizing patients who have progressive metastatic CRPC, have received treatment with either abiraterone or enzalutamide but not both, who had disease progression on docetaxel in castrate-resistant setting. These patients are being randomized to the standard dose of pembrolizumab with olaparib versus abiraterone or enzalutamide and alternate NHT, something we have not received before. And the primary endpoint of the trial is overall survival and radiographic progression-free survival.

To conclude, based on the prelim data, combining olaparib and abiraterone, and strong biological rationale, multiple Phase III trials have been started in combination of several PARP inhibitors, various PARP inhibitors, and novel hormonal therapy, which includes abiraterone or enzalutamide. PROpel and MAGNITUDE trials are combining a PARP inhibitor with abiraterone, and the TALAPRO-2 trial is combining enzalutamide with talazoparib. All these Phase III trials are investigating whether these combinations of PARP inhibitor with novel hormonal therapy are only efficacious in patients who have HRR mutation or these combinations may be effective in patients who do not harbor these DNA repair defects.

As we discussed, strong biological rationale exists for investigating combination of PARP inhibitors and PD-1/PD-L1 checkpoint inhibitors. However, we are consistently seeing, at least in the context of combination of immunotherapy with PARP inhibitors, based on the earlier data, prelim data, that not all patients are responding. In fact, a small minority of patients have responded so far. And patients who did not harbor DNA repair defects, the likelihood of responses was much lower. Even an earlier study of combining abiraterone with PARP inhibitor, not everyone responded, although radiographic progression-free survival was remarkably improved with the combination abiraterone/olaparib compared to abiraterone alone. I think one of the most challenging aspects of ongoing research and investigation is to find out, what are the molecular biomarkers we can use in our patient population which will allow us to help select patients who are more likely to respond to these therapies?

Discussion with Dr Love

DR LOVE: I just want to ask you a couple of follow-up questions. I was curious, you were talking about the biologic mechanisms, and I know that these PARP inhibitors have slightly different mechanisms. Some have PARP trapping, et cetera. When you think about the synergy that you’re looking for and the idea of blocking resistance, is one, theoretically, PARP inhibitor more likely to be synergistic than another based on the mechanisms?

DR AGARWAL: Based on the laboratory data, the translation data, it seems like talazoparib is unique amongst all the PARP inhibitors, that not only it inhibits the enzyme poly ADP-ribose polymerase, but it also traps the enzyme PARP at the site of the single-strand break on the DNA. And I think that may allow talazoparib to be more efficacious, especially in those patients who do not harbor DNA repair defects in their tumor or in the germline.

But so far, as I said, we have not compared these PARP inhibitors with each other. Time will tell. We have Phase III trials going on. We’ll see how these PARP inhibitors pan out clinically.

DR LOVE: Another thing I wanted to ask you is, it’s been a little challenging for me to figure out comparative toxicity of some of these agents. And in breast cancer you have both olaparib and talazoparib approved, and it’s kind of been hard for me to separate them out in terms of toxicity. But I have heard some breast cancer investigators talk about alopecia with talazoparib. I don’t know with these men whether or not you’ve noticed that at all.

DR AGARWAL: Again, I will go with the reports in breast cancer literature. We have not observed something similar in patients with prostate cancer. As we gain more experience with these various PARP inhibitors in our patients we’ll have a better idea down the line.

DR LOVE: Another quick question. I’m just kind of curious in terms of the cytopenias that you see in patients with prostate cancer there. You see more bone mets than some of the other cancers, and marrow involvement the PARP inhibitors have been used for. How much of a problem are cytopenias with these men?

DR AGARWAL: Interestingly, we have not seen a higher level of cytopenia with PARP inhibitors in patients with prostate cancer than what has been reported in patients with breast cancer, even though patients are older, actually, than breast cancer population, in my view. I think this could be related to prior use of chemotherapy in patients with prostate cancer, which is not allowed by any of these ongoing trials in the castrate-resistant prostate cancer setting. But so far so good. I think we are not seeing any wrong signal here.

DR LOVE: The last thing I want to ask you about is taking care of patients with prostate cancer, taking care of patients in general in oncology nowadays in the era of social distancing. In terms of your center, how are you approaching the decisions you’re making about these patients in terms of bringing them into the clinic, the use of telemedicine? How has your approach to these patients changed in the last few weeks?

DR AGARWAL: One of the most welcome steps taken by Medicare and CMMS in the last month was along telehealth for our patients. We are seeing patients through telephone or video conferencing, definitely not as good as seeing them in person. We lose a lot of clinical sense of what is happening, but for all patients who are not on chemotherapy, who are on oral hormonal therapies, who are on clinical trials with oral pills, again, we are trying to minimize the number of visits for them. Monthly visits required by the clinical trials have been allowed by the study monitors, and pretty much all clinical trials have allowed us to see patients remotely for those visits, which do not mandate patients to come here.

Many of the biomarker studies are being delayed because of that. But I think this is all being done for patient safety. And this is a challenge, I would admit, but whatever we can do to allow patients to be away from the hospital, I think we are doing it.

DR LOVE: I’m curious, for example, a decision like the treatment of a man who presents up front with metastatic disease. Have you changed your approach to those patients recently? In the past we’ve heard investigators say it’s kind of a coin flip between docetaxel and hormonal therapy. Is it shifting away from chemotherapy, this decision?

DR AGARWAL: We were already shifting away from docetaxel chemotherapy in metastatic castration-sensitive prostate cancer, especially after approval of apalutamide and enzalutamide for our patients. That was already happening. I think that shift has been expedited by the ongoing COVID-19 crisis. Giving chemotherapy to anyone at this point of time when other alternative options are available which do not induce myelosuppression makes sense to me. I think the shift was already happening, and it will be complete by the time we are out of this crisis.