 And thanks to the organizers for inviting me here to speak today after this excellent lineup of, as I mentioned before, SITC All-Stars, it's a little intimidating to talk after people who have such positive data since so much of what I've done in this area has been somewhat less positive, let's just say that. Also these folks are true All-Stars. I'm at best a SITC member and I'm not sure my membership is up to date so I have to work on that. We'll check on you. Yes. Okay, good. So I've heard a lot about PD1 and PDL1. I thought I'd focus a little bit on biomarker development in that area as opposed to talking about some of the past. Just as a quick summary of where I think we stand right now is we have an early signal of efficacy based on just phase one trials at the moment. One of the key questions with the data that we've seen is, is the clinical benefit that we've seen in some ways a reflection of patient selection or will the stable disease that we've seen in some of these patients translate into improved survival and pivotal trials? That certainly was the case as Pam mentioned with epilimumab. We'll have to see if that's the case with PD1 and PDL1. And also very important, as she mentioned as well, is how many of these responses are durable when we stop therapy? We talk about the tail of the curve which is what we all care so much about. Those folks who actually enter a remission and come off treatment, are those responses as durable with PD1, PDL1 as we see with IL2 and IPI. Toxicity is an issue particularly when you talk about combinations which is obviously the next step. But what I was asked to focus on is biomarker development and one fundamental question is are we ready to guide clinical development or pivotal trials based on biomarkers, particularly PDL1 expression which has received some attention early on in the development of these drugs. That was largely based on the Topolian New England Journal study that was published a while back where we looked at a small, relatively small subset of patients on that trial looking at tumor, immunohistochemical expression of PDL1 on the tumor. Essentially in that experience if there was not PDL1 picked up on the tumor, there were no responses in that group. But it must be emphasized a very small group of patients, only 42 patients. Also a very small group of kidney cancer patients. So we obviously need more data. One of the interesting things in my mind about ASCO this year was we started to get some more data about the impact of PDL1 expression on response. This is data from Genentech and their phase one trial of their PDL1 antibody looking at PDL1 staining on the infiltrating cells in the tumor. And what you see here is essentially while responses seem to be higher in PDL1 positive tumors. We're still seeing responses in PDL1 negative tumors. So in my opinion, we need a lot more work in this area. We also, I don't think we're ready to exclude patients who are PDL1 negative particularly in kidney cancer for reasons that I'll explain. So here's some of summary data that we've seen so far. I mentioned the PD1 and PDL1 data. There was also another smaller study, smaller analysis of the melanoma patients on the original pivotal trial looking at 34 patients in that setting. Interestingly, those that were PDL1 positive were the response rate was higher than we saw on the overall population. But once again, slightly in a different tumor type, responses being seen in PDL1 negative patients. And we'll talk about the combination in a second. So there's quite a few unanswered questions. I'll try to take just a few of them as far as the biomarker development, translational questions which we hope we can address in the coming years. The first one that I decided to pick out is PDL1 expression uniform and stable in an individual tumor and from primary to metastases. We've all, this has been cited multiple times at almost every meeting. We cite the British group's work from Girlinger et al. from New England Journal about tumor heterogeneity. My opinion this is going to have a large impact on biomarker development with these targeted immune therapies. Gordon already showed this slide. I'll make a slightly different point than he made with looking at, this is Sabina Senioretti's lab looking at his PDL1 IHC antibody, looking at concordance in some patients between primary and metastases, but discordance in other cases. We've started to look at a group of primary and matched met pairs. These are not biopsies samples. These are surgical excisions on patients. We obviously have to do this in a much larger group of patients and we're glad to collaborate with folks who can provide samples. But we've looked at in those 34 patients, we had 10 patients who were PDL1 positive by their assay, but in seven out of 10 of those patients, the met was positive and three, it was negative. Also we saw some patients where the primary was negative and the met was positive. So there may be some discordance in patients. Recently she found that in the primary, PDL1 staining was quite heterogeneous. It tended to be in areas of high nuclear grade, but in the met, the staining was more homogeneous. So how does this impact biomarker development? It's possible in our early experiments we need to be sampling both the primary and the metastases and I would argue that we need to consider all these trials are hard to execute, trying to get excisional tissues so we can see a more comprehensive view of what the tumor is doing. Another question is what factors, in addition to PDL1, the expression can reliably predict response to treatment. So for example, why do some PDL1 negative patients actually respond to a PD1 antibody? I think Gordon mentioned this as well and it may be as simple potentially as PDL2. And Sabina has shown in the case example of one patient's tumor in the bottom here that does not have PDL1 on its surface, but does express PDL2. So could that be a possible explanation for why PDL1 patients don't respond, are they just expressing PDL2, that needs to be explored, this antibody needs to be refined and tested in much larger groups of patients. Another unanswered question is predictive marker expression influenced by prior VEGF therapy. I think it's an interesting question and does it predict for a response to this therapy? So for example, are VEGF therapies which are so common, more active in PDL1 positive patients? I sort of alluded to the IL2 select trial that we mentioned before as a sort of a negative experience in biomarker development and immune therapy and kidney cancer. Here's a look at the same cohort of patients we looked briefly at whether PDL1 expression and B7H3 expression, this was the help with Eugene Kwan and Brad Liebowicz at the Mayo Clinic. And essentially in our hands, PDL1, when it was up in your tumor, B7H3 is up in your tumor, you are more likely to respond to IL2. Obviously that needs to be confirmed in a prospective data set, but more important to the point that I was trying to make, most of these patients did not get cured with IL2, surprisingly, but they went on to other therapies. And one of the things we tried to look at is whether staining impacted their response to those other therapies, and it turns out that group was probably a little bit less likely to benefit from VEGF TKI's, other groups have published similar results in this area. GSK presented a trial at ASCO, which I would have shown you as the next slide, except Dr. Schweri forgot to give it to me, unfortunately. But it shows a similar experience. Thank you, Dr. Schweri. All right, moving on, so he's busy. So PD1 blockade, another question which I'd like to see us pursue is PD1 blockade better applied in the treatment naive setting or in the VEGF pathway resistant setting. One trial that Mike Atkins is trying to launch is trying to get at some of these questions, try to help address some of these issues, both clinical and translational. This is a phase two trial where patients would get their primary tumor sampled, they'd also get a baseline tumor biopsy of a metastasis that would be excisional, they'd be then randomized to VEGF TKI or a PD1 antibody, and then at the time of progression undergo a second excisional biopsy in the case of the VEGF TKI patients, they probably could be biopsied because that would then allow them to go on to get the PD1 antibody which patients are obviously highly motivated to get. This trial, this type of trial design is hard to execute, but I think given the interest in PD1 antibodies at the moment, it might be possible, it might help address some of these issues not just on predictive biomarkers and the impact of VEGF therapy on those biomarkers, but some of the mechanistic questions that have been alluded to about why some of these agents work and why they may not work in some patients. Last question that I had here is, what are some of the mechanisms of resistance? Obviously, just like with IL2 and epilimumab, the patients getting a long benefit with PD1, PDL1 or obviously just a subset of the entire group, why is that? How can we look at both innate and acquired resistance to PD1 pathway blockade? There are quite a few potential mechanisms of innate resistance. This is sort of a long list. I won't go over it all because it's just I don't have too much time, but I'd like to focus on one of the potential mechanisms for innate resistance. Could it be that in tumors where exhausted T cells or inhibitory T cells predominate that approaches to activate T cells like IL2 or target inhibitory molecules like Tim3 and lag3 as you've heard from other speakers or deplete regulatory T cells as epilimumab might do, could those things be necessary in those patients to overcome innate resistance? This is a look, once again, from Sabina Senuretti, looking at co-expression of T cell inhibitory pathways in kidney cancer specimens. This is looking at T cells. Essentially what this shows is T cells in the tumor both expressing PD1, Tim3, at the same time, so it's conceivable that you need to be blocking both of these antibodies, both of these targets, or other targets as other people have mentioned, to get a full response. So we do see this in some specimens. Genentech has made a great effort at trying to look at some of these mechanistic questions in their recent Phase 1 trial that I mentioned before, where patients with access to tissue were asked to undergo serial biopsies. This is looking at a serial biopsy of a kidney cancer patient. Unfortunately, many of our patients don't have easily biopsiable disease, but this person obviously did with a mass under their arm. You can see the response happening fairly quickly, but also probably more interesting is you can see some changes, which are characteristic of a responding patient with T cell CD8-positive infiltrate in this specimen sample taken at the second biopsy. Obviously, coming up with situations where we can do that effectively in kidney cancer is going to be harder than it is, say, in melanoma, but another potential place where we could test it is in the neo-adjuvant setting, as Chuck mentioned, and I think as Naomi Haas mentioned earlier this morning, and that's something we're trying to develop in the cooperative group setting, looking at a neo-adjuvant application of PD1 blockade. So this has been mentioned several times. I won't belabor this, because this is the recent New England Journal paper of looking at combinations. The point is not that this is the greatest data since oncology started, because it is. But more importantly, the biomarker part of this story, we saw increased responses than with single-agent PD1. We saw increased toxicity, as been pointed out by at least one speaker. But I wanted to focus a little bit on the biomarker piece of this story. As I mentioned to you before, PDL1 expression has been associated with higher response rates, but it's not perfect at excluding non-responders. One of the interesting pieces of this trial was that the response rate in PDL1 negative and PDL1 positive patients was robust in both groups, and that begs the question which maybe we can address with the speakers here is, you know, what's going on there? In a PDL1 negative patient, what is CTLA 4 blockade doing to that tumor microenvironment that are making those patients responders? That was an exciting application in my mind of biomarker development that hopefully will lead to future insights. So as we move forward, you know, there's a lot of exciting new things to do, as many people have said. It's great that people are showing interest in immune therapy, and I certainly hope it continues. A lot of different combinations that make sense, but only with rational trial design and biomarker development are we going to be able to move this path forward, particularly when we have active agents in the clinic when you think about clinical development. So we're going to have to narrow our population, narrow our application of these agents to the proper patient in order to move these combinations forward, in my opinion, so that we can achieve Dr. Finke's dream of more cures for kidney cancer with targeted immunotherapy. Thank you very much for your attention.