Heather Cheng: So now, providing the splitting perspective is our next speaker, Dr. De Bono, who is from Malta.

Johann De Bono: Thank you. Thank you, [inaudible] and Silke for organizing the meeting. Good afternoon. Good morning if you're further west, and good evening if you're further, I guess, east from us. Wherever you are, thank you for joining. So I want to I guess argue the case. These are I guess my disclosures.

Should we be lumpers or splitters? Well, I guess it depends on the question. What is the question you're trying to address? I don't think that we can really refute the fact that actually prostate cancer is highly heterogeneous, and depending on what question you're addressing, you may want to be a lumper, you may want to be a splitter. For abiraterone, we lumped when we designed the Abiraterone trials. I think for part based on the question we're addressing, we had to split because we're pursuing a synthetic little strategy. But overall, our primary priority for our patients is first, do no harm.

Now what about the underlying biology? PARP inhibition was really developed based on the premise that actually we can get synthetic tumor cell kill, i.e. kill the tumor cell, spare the normal cell, which to me remains the holy grail of cancer medicine. Now remember, you have to reduce both alleles of the tumor suppressor gene, BRCA, ATM to sensitize. And most of these studies have not looked for bi-allelic loss. And indeed, for some of the trials we're talking about, the assays were done on plasma DNA, which cannot detect gene alterations like BRCA2 HOMDELs, because the tumor fraction in the plasma is far too low to really elucidate that.

Now you've heard eloquently from Noel how different niraparib and olaparib are, but actually, in fact, in the clinic... And I did phase ones on both these drugs. I really led the niraparib and olaparib phase ones, they actually work both pretty well in the BRCA patients.

And I want to show you these data. These are not my data. These are data from Yves Pommier, an outstanding scientist at the NCI. And you can see sensitization, some differences, but actually in fact, for example, look at the BRCA. Both drugs sensitized. And in fact, veliparib, which is actually not even a PARP trapper, in fact, actually still, also has anti-tumor activity, although maybe much weaker than olaparib and niraparib.

And in fact, please note that actually, if you down regulate nonhomologous end joining, for example, Q70 top of the slide in black here is Yves has shown that by down regulating NHEJ, you can actually cause resistance PARP inhibition.

Now, what does AR blockade do? Well, Charles Soros has told us that actually abiraterone AR blockade actually impacts, down regulates NEHJ, which might even cause resistance to PARP inhibition. So be careful what, what this can lead to.

And we know now that for, well, there's no doubt at all that actually the benefit of these drugs is primarily in the DNA repair defective patients. This is one of the first ever prostate cancer patients to get a PARP inhibitor, a three-year response for this gentleman who was actually an actor on Star Wars. He spoke openly about his disease on olaparib because of his BRCA alteration.

Now what about the clinical trial data? Well, TOPARP-A I think remains in CRPC, really the primary study for single agent PARP inhibition that actually did not select patients. And actually almost all, bar one, of the responding patients, single agent PARP inhibition had a DNA repair defect. Let me be very clear. We did not see PSA falls in men without a DNA repair defect apart from one.

Now, what does that say to me? The argument that actually PARP inhibition blocks the R signaling, for me, doesn't hold. I have studied PARP inhibitors in the lab. I don't see blockade of ER signaling by PARP inhibitors. So tenant one of the PROpel study for me, doesn't hold. We don't really see tumor cell kill in the lab by blocking PARP inhibition, through AR. Now what we've seen in the TOPARP-B trial that Joaquin Mateo, who is doing a Ph.D., in my lab at the time, is now in Barcelona, has shown is that there are super responders and responders that are less impressive. And the super responders actually are the BRCA2 HOMDELS. They actually are, are in study for threefold longer than even the BRCA2 mutation and het loss. Now that's quite important because the BRCA2 HOMDELS are the guys that are hardest to detect. And a lot of the assays, particularly plasma let's say, actually miss these because of that low tumor fraction issue. And I'll come back to that later.

Our trials with TOPARP led us to design PROfound. The PROfound trial, and I think the one thing I can claim is actually I chose the title of PROfound. I liked PROfound and PROpel was named after PROfound, I guess, pro standing for prostate. So that one thing I can claim for sure.

But PROfound led to the approval of olaparib for the cohort a primarily, I guess, because that's what it's powered on, but in the US for cohort A and B and Europe, for just the BRCA. I'll be honest, in my opinion, the drug probably works best for somewhere in between those two regulatory approvals, because there are some genes that cohort B where PARP inhibition, in my opinion, does not work, but I have no doubt in my mind that the ATM bi-allelic-loss tumors can respond to PARP inhibition.

So what about PROpel and MAGNITUDE? Let's go back to what Noel was saying, and I'll give you my bias. Now, I may be wrong, but I've been working on PARP inhibitors since 2005, so hopefully I do have some expertise in this space. Well, Kim Chi led on the MAGNITUDE trial, Shaneen Sandhu, I think I saw was the last author. She trained with me as well. And I think MAGNITUDE, for me, gave me no surprises. The activity was primarily in the biomarker positive. Now is niraparib that different from olaparib? To be honest, I don't think so. I mean, it is different. Don't get me wrong, but I don't think it's that different. And I don't think that explains what we're seeing in PROpel, in my opinion. Clearly in the biomarker negative patients in this trial, we have seen no benefits, but we have seen benefit in the biomarker positive, most impressively in the BRCA patients.

So again, no surprises, but let's now discuss PROpel because that's where the issue really arises. Is there really a rationale for PARP inhibition to work in prostate cancer? Noel said to us, "Okay, the combination is because PARP inhibition inhibits AR or AR blockade causes BRCA-ness or HR defects." Well, think about it. If AR blockade causes BRCA-ness, loss of BRCA, what's going to happen? Your response rate to the combination, because they're on a BRCA loss is going to increase to 80%. Is that what PROpel showed us? I don't think so. Although I'll be honest, PROpel has not yet presented all the data. We have to see more. But this was one of the slides that Dr. Saad showed at GU ASCO. These are two main mechanisms of action that predicated this trial design AR blockade or induction of BRCA-ness to quote [inaudible] statement. But I would actually propose to you that these actually hypothesis are both incorrect and that the PROpel data prove that they're incorrect as do the PARP data.

So if you inhibit AR blockade by PARP inhibition, should you not see PSA falls? We don't, unless you have the DNA repair defects. If you don't regulate HRD with PARP inhibition, should you not have a huge increase in response rate? And you don't in any of the studies we've seen to date. Now, is there some other kind of fortuitous mechanism action? Now I can't exclude that. And in fact, we've published that actually on Abiraterone, tumors can evolve sub subclones that lose chromosome 13, lose the RB locus. We know that from the work of Charles Soros. So, so maybe subclones are escaping that PARP can hit. And I'm sure that is the case for a small portion of tumors. But even so that will still require, in my opinion, molecular stratification for chromosome 13.

So you've heard already about PROpel. I don't have to describe it further to you. These are the PROpel data. This is the investigator assessment of rPFS has ratio 0.66. Look carefully at this rPFS curve. What does this remind me of? Over a decade ago we published that platinum sensitivity and PARP sensitivity are fairly similar. Okay. Similar mechanisms of action.

My dear friends, Cora and Oliver, who I have utmost respect for, published many years ago that this poor man's platinum satraplatin okay, causes very similar in molecularly unselected prostate cancer, hazard ratio 0.67, 0.66, 0.67, I think we're all happy with that. Again, no survival benefit in unselected patients. Now, what does this mean? Well, for me, and my bias platinum and PARP inhibition have similar mechanisms of action and if you don't select, that's what you're going to see. But clearly let's wait and be patient for the PROpel overall survival data. We haven't yet got enough follow up.

One further comment and Niraj Agarawl I think at GU ASCO did raise this question and there was an interesting discussion between him and Tia Higano and Niraj gave this comment, and I agree with Niraj and that's why I was raising the question about the BRCA HOMDELs. Are any of the non HRR, and by the way, I don't like the term HRR. ATM loss does not cause HRR. Let's be very clear, and hopefully Colin will mention that. Non HRR in this study, I am sure mean false negatives. I have absolutely no doubt. And if you want to have beer over it, delighted. Remember a lot of these tumors were genomic analyzed from plasma and plasma misses a lot of BRCA HOMDELs. So be very careful. Here's the response rates from PROpel. If Abby, which I know very well causes BRCA-ness what would you expect, for the combo? A response rate of 80% or 70%? Okay. Look at my data, our data from the olaparib studies. Are we seeing that? No, the response rate increases by 10%.

Is it really causing BRCA-ness? In my opinion, I don't think so. So what are my recommendations today? We clearly need more data. Should we lump? Should we split? Based on the initial premise of synthetic lethality, we should definitely split. Is there some other lucky serendipitous fortuitous mechanism of action? Possibly, maybe, evolving sub subclones that lose RRB1, but that would still require, in my opinion, splitting. And finally remember our priority as physicians is do no harm. These drugs do carry risk. They cause DNA damage in men that already have a germline repair defect. There is a risk of leukemia, mild dysplasia. I have had the unfortunate experience of seeing men rapidly die of leukemia, following PARP inhibition, and women with ovarian cancer. So at present, I would recommend we should split unless we see more data. So thank you for your attention.