Read the Full Video Transcript

Johann Sebastian de Bono: We have the question before us today, who benefits from PARP inhibition? I do want to share my disclosures. My institution has a patent on DNA repair defects on PARP inhibition, and I have developed three of the four main PARP inhibitors from my early clinical trials, olaparib, talazoparib, and niraparib.

Here is my talk overview. What do we think we know? What else do we need to know? What should we do in the interim? So these are basics. And forgive me for starting with basics. Prostate cancer is not one disease. It is a heterogeneous disease of many subtypes.

And in fact, even in one tumor, as depicted by the M&Ms or Smarties, as they're called in the UK, sweets, there are many different types of cells in one tumor, including stroma. And we must not forget the fourth dimension of time. That actually during treatment, over time, the tumor is changing and evolving.

What do we think we know? Well, it's quite clear now to us that prostate cancer has DNA repair defects. Many different types of DNA repair defects, as you've heard already. Don't forget MMR and CDK12 alterations and immunotherapy. Although I will not discuss that today.

You do have to think about these defects as being germline or somatic, clonal or subclonal. Impacting different parts of the DNA repair machinery. Including, as you've heard from Joaquin, homologous recombination. But it's important to note that PARP inhibitors do work not only against tumors with homologous recombination defects, but also other defects, as I'll explain in my talk just now.

What do we know about PARP inhibitors? What do we think we know? Well, PARP enzymes are key to repairing, gluing back DNA double-strand breaks, DNA damage. So if we inhibit PARP, what we get is DNA damage. We get double-strand DNA breaks. Now, that is good, but it's also bad. And I'll come back to that. And toxicity of PARP inhibitors later on. And Elena, who will follow after me, will talk about that further.

But what is key is that actually if you prevent the gluing back of those DNA double-strand breaks, if you already have a DNA repair defect, to put it simply, you get synthetic lethality in a tumor cell–restricted fashion because the tumor cell cannot repair well. And that's why you get tumor cell selectivity.

And in the carriers of BRCA2 in the normal cells, you don't get a lot of toxicity because you still have the other allele. So you need to lose both alleles of the genes of interest, usually to get a sensitivity to PARP inhibition. And obviously, you've heard about PARP and BRCA2 from Joaquin.

Now, what I've been asked to address today was what about the other genes beyond PARP? And simply, it's complicated. And I'm not going to go through them all today. But if you look at this figure on the bottom right, there's a lot of genes.

But let me put it simply. There are homologous recombination repair genes. These include BRCA2, BRCA1 losses in ovarian, prostate cancer, although it's the commonest loss in ovarian cancer, a big difference. But please don't forget about PALB2. I'll come back to that in a minute.

Then there are other genes that have been called HRD genes incorrectly, like ATM. ATM is not an HRD gene. But ATM loss, undoubtedly in this CRISPR screen from the Daniel Desrosiers lab, you can see ATM, big hot spots. ATM, complete loss, definitely sensitizes PARP inhibition. In many labs, they've all shown the same thing. But you have to have complete protein loss for loss of function.

Fanconi anemia genes, FANCA genes. Rare in prostate cancer, but when completely lost, also sensitized, as also shown in this CRISPR screen. But surprisingly, in this CRISPR screen, there was a gene that pulled out that, again, is not an HRD gene. And that's the RNase H2 complex that removes RNA from DNA. And I'll come back to that in my talk. I think this is quite important in prostate cancer.

But a key point is that we need to lose both alleles, usually to sensitize. And none of the trials we have done really, that we've been talking about today, have really done that, I think, quite enough. And this is, as you've heard from Niven, very difficult to do, particularly from plasma DNA.

Now, when you have BRCA, it's easy because most of the time, as we've shown in our paper by Carreira et al., if you lose one allele in a tumor, you almost always have biallelic loss. So BRCA is easier. But with ATM, when you detect an alteration, it's really complicated because you don't know if it's truly a loss of function, as you heard from Joaquin.

Now, when you see an alteration, unless it's clearly a frameshift or a truncating mutation, and you often usually don't know about the second allele—what's happening to that—unless you do ATM immunohistochemistry, which is what we did in our TOPARP trial, which is why we saw responders when selecting for those patients.

But let me reassure you, every lab that has done these screens has shown PARP inhibitors work for ATM loss cancers, including prostate cancer. Now, what about how we analyze these tumors? Well, Joaquin has mentioned that we need functional assays, but these are not yet good enough.

And remember, we need functional assays not just of HRD but also of the other repair defects that are not HRD. There are quite a lot of studies looking at archeological scars, like the Myriad assay. But our experience to date is that these don't perform well in prostate cancer, unlike ovarian cancer. And the RAD51 IHC assay that I've worked on with Joaquin and Violeta Serra in Barcelona has limitations with a significant failure rate.

OK, so let me talk about three genes, PALB2, ATM, and RNASEH2B today. Well, PALB2 should really be known as BRCA3. It's key to BRCA2 function. It's not commonly lost in prostate cancer. When it's altered, it's usually a germline mutation. But when it's biallelically lost, you get prolonged responses, as you see in this curve.

These patients had PALB2 in our TOPARP-B trial. And you see that the tumors that have biallelic loss largely do well and have more than six months on trial. But what was surprising to me in this study was that not all patients with a germline mutation and PALB2 alteration had biallelic loss.

And this is one of our patients. You see here two years of response to single-agent olaparib in this trial. This is worthwhile trying in these patients. What about ATM? Believe me, if you get the ATM assay right, patients respond to single-agent PARP inhibition.

Here is one of our responders in TOPARP-B. This is Nina Tunariu, our radiologist, who's amazing and helps us analyze the whole body MRI and CT scans. This is a responding patient here in soft tissue disease. ATM is not an HRD gene, as I mentioned earlier. It detects DNA damage. But more of that I think another time.

And this is our, I guess our talazoparib trial, which, again, confirmed that both PALB2 and ATM with a different drug get responses if you have these alterations in CRPC with PSA and RECIST responses. BRCA2 in blue, PALB2 in red, ATM in orange. Hopefully you're not colorblind.

OK. In PROfound, we also have confirmed that ATM loss sensitizes to PARP inhibition. Remember, this trial did not select for biallelic loss, as I mentioned earlier. Unfortunately, in PROfound, we did not get any PALB2 patients recruited.

But actually, we did see some data on CDK12 that I'm a bit confused about even today, suggesting that the CDK12-altered tumors may have some benefit on PARP inhibition. But that may be because they have second hits on other genes like BRCA2, which we have seen that sensitize to PARP inhibition.

Now, let me talk briefly about RNASEH2B. So RNASEH2B is a gene that sits right next to RB1. And we hypothesized that if you lost RB1, you also lost RNASEH2B. Well, actually, we were wrong. But actually, we did find quite a lot of RNASEH2B loss in many prostate cancers.

And actually, what we've shown in our TOPARP trial is that when you give PARP inhibition, these subclones are removed. Is this good for the patient or not? The honest answer is I don't know. We need more trials to show that.

But I can tell you we clear subclones of blue cells that have no RNASEH2B, which is one of the main sensitizer gene losses in that screen I showed you earlier from the Zimmerman et al. Nature paper. And we need more data on this. But subclone eradication may lead to patient benefit.

Now, what else do we need to know? Well, I have to ask apologies for those that have heard me being very critical of the PROpel and TRITON—and TALAPRO2—trials. But I remain very critical. Until I see overall survival benefit in the non-HRD tumors, in the non-BRCA tumors, I will not believe the data.

This is the data for those tumors in PROpel. This is the BRCA and HRD, and you see quite different survival curves. Our assays are not good enough, as you've heard already. We have false positives and false negatives. We need to find out better why these tests are, or how these tests are performing or poorly performing, and to improve these tests.

The other big gorilla in the room, which I think is going to become a big problem for us, is secondary myelodysplasia and leukemia. And this is underappreciated. You're going to hear more about this from Elena. These drugs are DNA-damaging drugs.

As GU urologists and oncologists, think of testicular cancer etoposide. You are giving chronically a drug for long periods of time that damages DNA. We're going to start seeing a lot more of this. And be careful, because the more the patients, and the longer they're on these drugs, this is going to become a big problem.

By the way, I am waiting for the big kettle bell, because actually, I've been asked that I should want to hear it today. What else do we need to know? Can clearing subclones be beneficial? But also, is there something else we've missed? Are PARP inhibitors doing something else that will benefit prostate cancer subjects suffering from prostate cancer?

Maybe we've missed something. And I've deliberately put my titles to say, I'm not saying we know it all. But I'm saying based on what we know, we cannot argue for giving PARP inhibitors to everyone, as some have said to us in publications and at meetings.

So what should we do in the interim? Well, I think we have to very carefully consider the genomic data we are getting. These are complex analyses. And if you don't understand them, please do ask for advice. Look at our two papers we've published on this subject if you want to think further about this question. And they're listed here below. Reviewing the data. This is not just from me, but many people.

And finally, in my opinion, it's not just the BRCAs that benefit. It would seem to me our regulators in Europe and the FDA have gone to polar opposites. Thank you. That's a lovely sound. I want it louder. But we should treat not only BRCA2, but PALB2, ATM, biallelic loss, rarer subsets.

But if you can't get PARP, use carboplatin. Patients do benefit. But also, if you progress on PARP, you usually get reversions, you get HRD back, and you're unlikely to benefit from a PARP1 inhibitor through olaparib, for example, or from platinum after olaparib. So thank you for your attention.