Emmanuel Antonarakis: Thank you Aurelius and Silke, and thanks to the Lugano team, the administrative and logistical team has done a great job as well, so we thank them. So 12 minutes to talk about immune checkpoint inhibitors. These are my disclosures. So I'm going to review all five of these topics and I'll have to go rather quickly, but I think I'll leave you with some take-home messages and let's begin. So first, CTLA-4 blockade. We now know that CTLA-4 is primarily expressed on the Tregs, the regulatory T-cells. So when we inhibit CTLA-4, we are indirectly inhibiting T regs rather than specifically activating CD8 T-cells. And we've had now two-phase three studies. The post-chemotherapy ipilimumab study. In this study, patients got eight grade radiation to bone lesion followed by the ipilimumab or placebo. Overall survival benefit was almost met.

It's the most positive negative study you're ever going to see with an upper limit of hazard ratio, 1.00, p-value 0.053, but the PFS was significant, did not lead to an FDA approval or an EMA approval. In the pre-chemotherapy situation, the OS difference was less pronounced, although the PFS difference was still there. And interestingly 23% of patients here had a confirmed 50% response. So clearly there is activity, but it wasn't enough to meet any regulatory authorities. Now moving to PD-1, PD-L1, here we are actually activating CD8 T-cells. And this is the largest monotherapy study of ipilimumab, the KeyNote-199 that I had the pleasure to lead with Yohan. Unfortunately, overall response rate only 5% and PSA response rate only 7%. So this objectively looks worse than ipilimumab for example, as a monotherapy. We did make some insights here. There were six extreme responders in this trial. Two of them had HRR mutations, as you can see here. Two of them had these interesting dominant negative p53 mutations, hold that thought for a moment.

And the other two didn't have any appreciable DNA alterations, although in Yohan's lab, he detected mismatch protein loss on IHC. So one of these is a mismatch repair deficient patient. And there were a bunch of other abstracts trying to see whether the DDR or HRR mutations enriched for response and they did slightly enrich, 15% for BRCA one, two, and ATM versus 3% without or 17% response versus 4% without, but even in the MSI-high, it was disappointing. Of the six MSI-high patients, only one of the six responded. So a 16% response rate, so slight enrichment, but it's not the key to the response. There was also a numerical trend towards greater objective response rates in the higher TMB and higher PD-L1 staining but again, when you look at these plots, you can see the differences, not very profound. Now, what about combining CTLA-4 blockade plus PD-1 blockade. So a bit better, but still very modest I would say.

Objective response rate, 18%, PSA response rate, 14%. Now, interestingly, this study did delve into the biomarkers and here when they split the TMB into above it versus below the median, they did see an improvement with ipi-Nivo in those with TMB above the medium. And there was also a slight hint of activity, although not statistically significant in the HR deficient patients showing numerically better responses. Of course, all of us in this audience know about the tumor type agnostic approval of pembrolizumab for the MMR, or MSI-high patients. But how common is this in prostate cancer? It's not that common, 2.5 to 4%. So in this study, it was two and a half percent of the patients. In Wassim Abida study, it was 3.1% when they added the MSI-intermediate plus the MSI-high. The composite was 32 out of a thousand patients. So 3.1%. So keep that number in mind. It's exciting when you see it, but it's rare. And even when you see it, it's actually not that exciting because these are 11 patients that received the PD1 or PD-L1 inhibitor at Memorial.

And if you can look at these swimmers plots, two of them had responses more than a year, but the other nine didn't. And when you compare this to mismatch repair deficient colorectal cancer or endometrial cancer, the majority of those patients respond for two years, three years, sometimes five years. We don't see that in prostate, even in the mismatch repair deficient. Now, one thing that we studied when I was still at Johns Hopkins is the type of mutation. And we had this hypothesis that a missense mutation may change a single amino acid. Whereas a frameshift mutation will lead to this downstream changing of every amino acid after the frameshift until you get to the stop codon.

And we looked at 21 consecutive patients at Hopkins with MMR deficient prostate cancer and we showed that the TMB itself is numerically prognostic, but not statistically prognostic for PFS or OS. But when we specifically quantify the number of frameshift mutations divided by a total mutation burden, which we coined the frameshift mutation burden, we showed that those who have high frameshift mutations, which we think are more neoantigenic, had a better response than those with lower. So perhaps it's not just as simple as looking at TMB, but maybe the quality of those mutations in the character, and maybe the frameshifts respond better than just the missense, needs to be further confirmed.

And then in 2018, the [inaudible] group discovered this CDK12 mutation, which is present in about 7% of prostate cancers. I was asked to write an editorial, and this is a figure from that editorial. But basically, this is a gene mutation. When it occurs in a biallelic fashion, it leads to this very unique and unusual type of genomic instability, which is not an HR deficiency, but it's this focal tandem duplication signature whereby a piece of the genome is duplicated exactly once. And if that duplicon just happens by chance to be within a coating region within an exon, it causes these fusion genes and these fusion genes accumulate non-discriminately throughout the genome.

And again, because there are fusions, they lead to this antigen amino acid sequence that doesn't exist in the native cells. So this is the hypothesis and there's been a number of single institutions, small studies, 12 patients here, 20 patients there showing that yes, these CDK12 mutant patients, which are 7% of the total may respond, let's say equally to the MMR D patients. But again, only very few of these, maybe 10 or 15% have responses that are durable beyond a year.

Now, what about the TMB above 10 indication? This was the only study that I could find from a Foundation Medicine and Neeraj Agarwal was the last author in this study. So how common is TMB above 10 in prostate cancer? It's 6%. So in this study, they had to look at about 10,000 patients to find 600 that had TMB above 10, so 6%. And in those 6%, they had 33 that received an immune checkpoint inhibitor and 574 that received the taxane for mCRPC. And on the top, they're showing that in the TMB below 10, the taxane outperformed the immune checkpoint inhibitor in a retrospective series, but that in the TMB above 10, the pembrolizumab outperformed in this retrospective series, the taxane.

So the suggestion here was that in fact, if you do have a TMB above 10, you should consider using pembrolizumab. There's a caveat here and the caveat was in the supplement, and that is what proportion of TMB above 10 are not mismatch repair? And it turns out that the vast majority of these, more than 90% are mismatch repair deficient. And I would argue that the other 10% are probably mismatch repair deficient that were under-reported or not diagnosed. So it's unclear whether this TMB above 10 adds anything above and beyond the mismatch repair deficient.

So this is Antonarakis’ list of biomarkers. It's a long list. Each one is individually quite rare, but maybe when you add them all up, the top four, I think I would say are the most established and then the rest are more question marks and we have a lot more to learn here. So let's talk about some of the combinations. So first of all, PD1 inhibition plus enzalutamide. I'm only going to talk about the pivotal studies, there are many, many others that I'm sorry, I don't have time to talk about, but here we must discuss the IMbassador250 led by Dr. Powles and Dr. Sweeney.

That's just the goat bell, by the way. No difference in overall survival or radiographic progression-free survival. So on the face of it, a dead negative study, why was it published in Nature Medicine? Because of all the biomarkers that they did. Interestingly enough, and this is the really important piece. Patients benefited, again in retrospective fashion in these biomarker subsets, if they had high PDL1. Secondly, if they had high CD8s in their tumors. Thirdly, if they had high TMB above the median. Fourth, and interestingly, if they had Wild-Type AR compared to AR mutations and fifth, a bit controversial, but interesting also, the p53 mutations. So this is the second or third time that we're seeing that certain types of p53 mutations may actually increase immunogenicity.

What about immune checkpoint inhibitors plus PARP inhibitors? There's a rationale behind it. I won't go into it, has to do with DNA damage and [inaudible] response. This was the first study we thought initially that this might work in everybody, biomarker selected and unselected. But if you look at the Kaplan-Meier curves, it turns out that the duration of benefit is much greater in the whole molecule for combination deficient. Merck did this study with Evan Yu, KEYNOTE-365. They did show a 12% PSA response rate, 7% objective response rate in unselected patients. This gave them the enthusiasm to do a Phase III study called the KEYLYNK-010, pembro plus olaparib versus abi-enza. And unfortunately, the press release came out last month. This study was negative, stopped for futility. So another huge immune checkpoint-based study, which was negative. And finally, I've got 30 seconds left. Immune checkpoints plus kinase inhibitors.

This is Neeraj Agarwal's data, cabozantinib plus immune checkpoints. Again, there's various hypotheses here about inhibiting Myeloid-derived suppressor cells skewing the macrophages into more of an M1 phenotype. And this was his clinical data from 40 patients. Objective response rate of about 20%, PSA response rate of about 15 to 20%. And this has led to the Phase III study CONTACT-02. Randomized study of cabo plus atezolizumab versus abi-enza remains to be reported. So in conclusion, no immune checkpoint inhibitors currently approved by the EMA or FDA for molecularly unselected patients. pembrolizumab can be used for those with MMR deficiency or TMB above 10. There are multiple Phase III studies ongoing. The ones that have been reported so far have been negative, unfortunately, and we do need better biomarkers to select patients for immune checkpoint benefit. This is my new home now, the Masonic Cancer Center. Thanks very much.