Rosland Eeles: Thanks very much, Robin. Thanks for the organizers for this fantastic meeting and a plea because I think we really need your help, as you'll see as I go through this talk. So, I was given the task of looking at which men need, well, really, testing, first of all. I think the genetic counseling mechanisms are well tested, very good guidelines, particularly in Europe for how we counsel. In some countries, of course, we're now looking at a lot of direct-to-consumer testing, and that's going to come like an oncoming train if we don't sort out the guidelines, so this is very helpful.

So, first thing is, what are the questions? Well, the first question is what's the spectrum of genetic predisposition that we currently know? And I'm also going to show you what's coming like an oncoming train, and I do think we're going to be in a very different position in the next consensus meeting, possibly even next year, with a much better test that we can offer.

The next thing is, if we have a better test, who should we offer it to? The costs of testing are going down all the time, and this is going to lead to consumer-driven testing, and we have to really have guidelines to be able to offer testing, particularly as the costs drop.

Then, the next thing is, if we find mutations, how do we manage men at a higher risk? I'm going to show you some data that are coming out in two weeks. Just to plea, please do not tweet it. It's embargoed, but I didn't want to lose the opportunity to show it to you. We've got the interim data from the IMPACT study, and I'll show you that. And the last thing is, putting everything together. What's everything going look like possibly even in a year's time?

So, to answer those questions, the first thing is what's the spectrum of genetic predisposition? If we lined you all up in a room, there's a normal distribution of risk. Some men are at very high risk, but on average, one-in-eight risk in your lifetime. So, prostate cancer predisposition has a very similar distribution to things like height and other normal distributions that you see that is studied with genetic studies, and you can split up this risk just by doing a single PSA test at 60. If it's less than one, you can split the group up like this, but with genetic testing now, we can split the group up into percentiles. If there are 100 men here, you can find the man on the right, and he's at 40 times higher risk of prostate cancer than that man on the left. We can already do this test, but what we don't know is how we should screen those individuals. Should you even stop screening the man on the left? That's not known, and those studies are ongoing.

The next thing is, what's the type of genetic predisposition that we might be able to test for, and in this disease, we have mainly contributions from common variants. They've been found by large genome-wide association studies, and I'm going to give you the latest update on these, and then most of the next set of contribution is from the middle set. They're relatively less common. We're talking about allele frequencies, approximately one in 100 ATM. For example, in the British population, is mutated in one in a 100 individuals, so it's not that uncommon, but it's not very common, and they have moderate risks in the main, usually about two, two and a half, some of them even as high as six- to eight-fold if you take BRCA2 for example. And this is the summary of where I think we are today, and we're going to be in a slightly different position, I think, in a year.

So, if you look at the contribution of genetic predisposition, so, say you have a family history of prostate cancer, and so, from the family history alone, you know that the unaffected first-degree relative will be at increased risk. Then, what's the component of the risk, and could you do a test for that? Well, the first thing is, there are common variants, and from genome-wide association studies for many groups, now there are over 40 of these. A lot of them are done by the PRACTICAL Consortium. I'll show you a slide with that in a minute. We can account for just over a third of the predisposition, and we already have a profile. It's available in a saliva test, and we've got a very nice sequencing test that we developed with Eureka, now Thermo Fisher, in the US, and that's been led by Zsofia Kote-Jarai in my lab, 170 snip variants that you can do on a saliva test. So, you could profile the men in that field, and you could find the guy at the right who's at 40 times higher risk than the man on the left, and he's at about five times higher risk on average than the man in the middle.

So, that's the common variation. What about rare variation? I think Colin's talked a lot about this both in tumors and how likely you are to find it in the germline, and as Colin outlined, there are several of these. Most are DNA repair. The sort of outlier as HOXB13 that interacts with androgen signaling, and obviously, they then have therapeutic implications, but the big question at the moment for somebody testing is, what should you put into your test? And you'll see, when I summarize everything at the end of the talk, it depends where you are and what your company can offer or what your lab could offer.

Now, what's coming? There's a paper that's just been submitted by my colleague, Chris Haiman from the PRACTICAL Consortium, and he's been looking at ethnic minority groups. Large components of these are of African or African American origin, and the answer is that the results are extremely interesting. He's found a lot of new hits, and it is not all covered by the current profile. So, what's becoming increasingly clear is that in a year's time, the profile is going to be a lot better than it is now.

The other thing is that if you've got panel tests that might test an individual, you want to find those people that have the rarer mutations, but they have significant implications, both a risk. On average, these risks are much higher than with the common variants on their own. Whereas with common variation, you're profiling a whole population to stratify it.

Now, what about finding the common variants? Have they been done, and most of this work's been done by the PRACTICAL Consortium. I would like to acknowledge the many people in the PRACTICAL Consortium are here. Large case-control studies. It's 133 groups, and we have both Caucasian and ethnic minority groups within it. There's now over 200,000 samples in this consortium, so it's got the power to find these common variants. The latest paper, which is by Fred Schumacher, last year, published in Nature Genetics, found that final 170-snip common variant profile that we can use in Caucasians. It's very important to note this does not apply if you've got a Black patient, as we've recently found from Chris Haiman's data. So, if you have a White population, then you profile them for these 170 snips like that man in the field. The one on the right, the one in the top 1% of the profile of the population will be about 5.7-fold risk compared with the average of the population, and the relative risk in the top 10% on the right is 2.7-fold.

So, if that was breast cancer, you'd offer earlier screening, and at the moment, we're doing studies in London to look at whether mass testing of the population... We're testing 5,000 men in GP practices. We're doing intensive screening in them, which now it does include MRI. The old screening algorithms and impact did not include MRI, but I think they now should, and we should have those results by the next consensus meeting.

So that gives you a feel for the fact that test is going to involve common variation and rare variation. I think it has to be in one test, which means you have to have a sequencing test and that's what Zsofia Kote-Jarai together with Eureka has developed. It's where you actually sequence to find the common variants, and you can sequence gene sequences that in the same test.

So, the next thing is, who do you test? Well, if you take a man that has a family history of three cases of prostate cancer at any age, these are data from the UK, and as Colin mentioned, we tried to dovetail our talk so we showed you different data to give you an overall picture. This is from the United Kingdom Genetic Prostate Cancer Study, over 300 collaborators in the UK. Again, I'd like to acknowledge many of them are here. And what Dan [inaudible 00:08:52] did in my lab was, he used the BROCA panel. It's mainly a DNA repair gene panel developed by Mary-Claire King's group. At that stage, it was the 22-gene panel. It's not the new 60-gene one, and what he found, the bottom line is that three cases of prostate cancer in the family history, any age, about 7% of men will have a germline mutation in a DNA repair gene. That looks about the same ballpark as Colin showed.

The next thing is, do they have different disease? Well, if you look at their disease that they had, the men that had the germline mutations, they're more likely to be node-positive, more likely to have metastatic disease. And I think that's not really any surprise. Now, having seen the data like those from [inaudible 00:09:35] and Mateo and Johann de Bono and from Colin Pritchard, showing in maybe about one in six men with metastatic, castrate-resistant prostate cancer have a germline DNA repair gene mutation.

So, the next question is, what about young men, and also, is it all just the genes that we've seen on the lists that were shown to us? The other question is, is it odd other genes other than BRCA1, BRCA2, ATM, important to put into the panel eventually? I feel very strongly like we've heard before, that mismatched repair genes should be in this panel. They're important for treatment. I've also seen dramatic responses in men. In one man I treated a week after the FDA approved PD-L1 inhibitors in tumors that are genetically unstable, had dramatic response, and he had no other options for treatment. Now, 18 months out, no disease, who has MSH6 germline mutation.

So, the question, then, what we asked was, what other rare variants might we want to put into the panel, and does this apply to single young cases? Again, this was down in my lab when he was doing his PhD with Zsofia and myself, and he looked at a much wider gene panel, 175 genes, most of them DNA repair. And this was the sample set, and it was very, very important to sequence controls because, as Colin alluded to, if one in 100 people in the UK carry an ATM mutation, one in 100 prostate cancer cases are going to have an ATM mutation, one in 100 cases of brain tumor are going to have an ATM mutation. So, the question is, is it causative? So, it's very, very important to do controls and see if the mutation rate is higher and whether it's associated it with more aggressive disease on a case-case analysis. And indeed, that's what we found. So, we found that that was the case, and also, it's enriched if you looked at the BRCA panel. So, in other words, that does pick up very nicely, most of the mutations but not all of them.

And this is the overall result. Again, these were published last year, and these are in the European Urology, and you can see here a spectrum of genes. It's the 20 genes altogether that are mutated in association with risk of disease. And very interestingly, not all of them are associated with disease aggressiveness. So, some are associated with risk, and some are associated with more aggressive disease. For example, BRCA2, I think that's no surprise, but what was a surprise to us was that there's a common mutation in CHEK2 in Europeans about between half and 1%, and if you have the non common mutation, the non-1100 del C, then that is associated with aggressive disease which you do not see with the common mutation, so it might not only be gene-specific. It might be mutation-specific. That's when it's gets really complicated. And these are the survival curves.

So, you might be in a situation now where a man comes to you. They've got a family history of prostate cancer, and you may be able to offer a panel test, including common variants, and they may also interact. We now know that the common variants modify the BRCA2 risk. So, many of us, Heather Chang, for example, with Colin has set up risk in it. We have to look at how we manage these men, and this has to be done in studies.

And now just to finish, I'm going to just show you some very overarching results from a study that's coming out. This is the interim analysis of the IMPACT study on screening of BRCA2 mutation carriers. This is embargoed, if you don't mind, so please don't tweet it. It's coming out in European Urology on the 14th of September, but I thought you really should see this in a meeting like this. It's silly not to show it. And many of you know the IMPACT study algorithm. It's annual PSA screening initially and BRCA1 and 2 carriers. That cohort is now closed. We're doing followup, and this is the impact collaborators, 65 centers, 20 countries. Again, many of you here, thank you very much. You'll have seen the full paper. And this paper shows the four-year screening data from over 2,900 men, and they're both BRCA1 mutation, BRCA2 mutation carriers, all pathogenic, no VUSs, and also, negatives, so predictive test negatives, and these are the bottom line.

So, the basic answers are that if you have a PSA of over three in this study, you go to biopsy. Now, of course, a lot of people now will have had an MRI to biopsy them, and we're currently looking at that MRI data, but you didn't need an MRI to have the biopsy in those days because the study started 2006, so the chance you'd have a biopsy is 5%, so we're not over-biopsying. 112 cancers were seen. And the positive predictive value, if you came to biopsy and had a BRCA2 mutation was 31%, and you can see the control PPB there, and the cancer incidence rate in BRCA2 was significantly higher than in controls.

Now, what happens if you match the data, as far as we could, to the ERSPC Goteborg Study? And the answer is that the PPV of biopsy was significantly higher than in the Gotenborg Study, the PSA screening study and also, if you just look at higher PSA.

What are the tumors like? Well, the answer is that with BRCA2 mutation carriers, the tumors are significantly found at younger age, 61 versus 64 years in the controls. They are more likely to be of higher Gleason score. So, for example, 12% of the BRCA2 carriers had a Gleason score of eight or above, whereas 7% in controls, and on the NICE guidelines, they are more likely to be in the intermediate- or high-risk category, i.e., we are finding more aggressive tumors. This is not like population PSA screening. In fact, it's completely flipped on its head. You're more likely to find aggressive disease if you screen BRCA2 mutation carriers. And these are the pie charts to put this in a pictorial form. So, you can see here that with BRCA2, you have a high or intermediate percentage of 77% versus the control's 40%, but with BRCA1, we are not seeing this result. Now, we don't have enough followup yet to definitely say that we do not see this with BRCA1 mutation carriers, but at the moment, it looks like this is only significant for BRCA2 mutation carriers.

So, this is the bottom line to take away, and again, the slides will be on the video, but what I've said is, they occur younger, they're more aggressive, and they're more likely to need treatment.

So, this is the summary slide of where I think we are, and the answer is, it depends where you are, because in England, at the moment, despite the 100,000 genomes project, we've now got seven genomic medicine hubs that have started, we can only get BRCA1 and 2 germline mutation testing in a prostate cancer case if we have certain criteria that are mainly based on breast cancer. At the moment, testing is not in the new testing directory, and in the Cancer Genetics Group, we're moving hard to try and push it in together with [inaudible 00:17:20].

So, if you're in the private sector, or if you want a country like the US, you would, I think, get panel testing, and the main questions there are, what should be in the panel? So, like should you put ATM into the panel, and if you do, what is the relative risk? Well, PRACTICAL will tell you the answer, I hope, again in about six months' time. And so this is a very fast-moving field.

But the next thing is, does it alter treatment? And there is a page from Johns Hopkins showing, if you have a BRCA1 or 2 mutations, you're twice as likely to progress. So, that's going to drive testing, particularly if you want to consider somebody for active surveillance. And I think the area where we have the least data at the moment is, should you biopsy everybody if they have a BRCA2 mutation? Should you do mass snip testing in the general population? And if so, how should you manage those individuals? And that's the basis of studies such as barcode profile impact that are ongoing. And a huge thank you to everybody who's collaborated in these studies. It's taken a lot of work, and thank you for your efforts.