 It's my great honor and privilege to introduce Dr. David McDermott from Beth Israel Deaconess Medical Center. I think one of the leading proponents of immunotherapy over the last 10 or 15 years to talk to us about T-cell checkpoints inhibitors, a new hope. David? Thank you, James. I don't think I really know the answer to that question that we just had. I'm not sure I'll know it even better. I have a better answer after my talk, hopefully I'll have an answer. So I'd like to thank the organizers for inviting me here to what my daughter calls the homeland. I'd like to thank my cousin Ray McDermott for the invitation. This might be the first meeting in scientific history where there are ever two McDermott's in the same symposium, which I think that's a good thing for us as a clan. So one of the things I also find interesting about the talks that I get to give, I'm always glad to do it, but when I give the talks on the American side, my title is usually fairly straightforward. When I've given European talks, I always put a question mark at the end of my title. I'm not sure why that is, but I think it has something to do with Dr. Gore, anyway. All right, so we're here to talk a little bit about anti-PD-1 and a little bit of background on the pathway. This is really a pathway that's commonly used more in the setting of infection. As you see here in an antigen presenting cell, to activate a T-cell, you need two signals to shut off a T-cell, which you often want to do once the infection is controlled. You also need two signals, and there are these negative regulators or immune checkpoints that can shut off T-cell function. It turns out that cancers can sort of take over this negative regulatory pathway by expressing PD-L1 on their surface as shown here, and we now have monoclonal antibodies, actually several of them in clinical development that can block this interaction between the tumor expressing PD-L1 and the T-cell expressing PD-1. So instead of the T-cell becoming deactivated by this interaction, if you have a monoclonal antibody interrupting this interaction, you can actually activate T-cells leading to tumor cell killing. We've known for a long time in kidney cancer that PD-L1 plays an important role in the development of the disease. Work from the Mayo Clinic and other places have suggested tumors that express PD-L1 do worse. They're more aggressive. They're shorter survival. So being able to interrupt this interaction may have an important role in some patients with this disease. So when we talk about what we're treating, when we're treating patients with PD-L1, where we're seeing the most activity, I think we're probably seeing the most activity in what Tom Gajewski calls an inflamed tumor. And what does he mean by that? Well a tumor that's expressing a lot of innate immune signals, that's expressing cytokines that are drawing T-cells to the area of the tumor, but importantly it's also expressing negative regulators that block the action of this immune infiltrate. So you have immune infiltration plus defense. We could think of it as barbed wire, the PD-L1 expression on the tumor, and it's probably in these inflamed tumors where we're seeing a single agent activity. As Bernard likes to point out and presented this probably a hundred times, there'll be more new data fortunately at ASCO this year. This is the phase one data for a very small cohort of patients on a very large phase one trial with Nolamab. The important points here is it's generally a tolerable agent grade three and four toxicity was not dramatic. We did see deaths, three deaths due to pneumonitis, although now with an improved recognition that this can happen those deaths are becoming less common. But still need, there are a lot of autoimmune like toxicities that need to be monitored very closely. But we also saw an efficacy signal. Now this is probably the best where we're going to see with single agent PD-1 activity in kidney cancer. Meaning once you go from phase one to phase three you get less selected patients. So it wouldn't be surprising if the response rates and PFS declined. But what's probably still going to exist is this tail of the curve. Maybe not as robust as we saw it in the phase one trial, but there's still patients out surviving off drug what we'd like to call treatment free survival. Meaning these patients out here at two years this is where treatment stops. And you can see five or six patients who are on this study who've gone at least a year without being treated with anything and have yet to progress. So hopefully those patients have entered a remission of their disease, which I find pretty exciting. So what are the unanswered questions why alluded to one of them? Is the clinical benefit that we see in the phase one just a reflection of patient selection? Or will the stable disease patients actually lead to an improved survival in phase three? We'll have to see. That was certainly the case as James mentioned with epilimumab the CTA4 blocking antibody melanoma. That also has a very low response rate in melanoma but was able to improve median overall survival in large part because it created a substantial amount of stable disease that lasted for a while. We'll see if that's the case with PD-1 blocking agents. Another important question for us is how many of the responses will be durable off therapy? Obviously some of them are but will be they as long lasting as we see with IL2 and epilimumab which last years in the rare patients that have those responses. We don't know that yet because as I mentioned on this curve before while there are some patients off treatment a lot of these patients in response are still on drug. So we don't know how many of those people when they come off drug will eventually progress. Also will some of these uncommon toxicities prove vexing? They have with other checkpoint inhibitors particularly in kidney cancer you might worry about things like nephritis and lung cancer pneumonitis is a big issue because many of the patients getting treated with these agents come to treatment with decreased baseline lung function, cough, all that. So it makes it harder to treat those patients with this class of drugs. But most importantly since the benefit is probably only going to be in a subset of patients how can we identify those patients before treatment? Bernard also alluded to this in his talk and I'd like to take on a few of the questions as far as developing predictive biomarkers. After a talk like Charlie's you become a little bit discouraged that we could actually develop predictive biomarkers given some of the difficulties with tumor heterogeneity but we tried you know we're trying to address these in our clinical trials and we'll talk about you know what we learned so far. So one question is does PD-L1 expression alone predict for response? As Bernard mentioned in his talk the early data suggested that it might this is from Suzanne Topolian's New England Journal paper from 2012 was suggested that if you had PD-L1 expression expressing on the surface of your tumor you had a chance at responding to a PD-1 antibody but that if you did not there were no responders in that group but this analysis as Bernard pointed out was a very small number of patients 42 patients only five of them whom had kidney cancer. So what have we learned since then? Well when you think about how we might treat patients with immunotherapy for kidney cancer one of the things that I think is probably going to be true when we look you know five years from now for these patients with inflamed tumors who are PD-L1 positive they're probably going to be sensitive to single agent PD-1 or PD-L1 blockade and that's probably where we're going to use these drugs but the story of PD-L1 as Bernard alluded to is not a clear one this is data that Dan Cho presented at ASCO last year this is looking at the Genentech PD-L1 antibody in kidney cancer patients I'm in a 47 kidney cancer patient so a slightly larger group several interesting aspects of this one the response rate with the PD-L1 agent was somewhat less than we see with PD-1 we'll see if that maintains itself in future trials but importantly there was a small group of patients who are PD-L1 negative that respond and we've seen this in other studies as well so if a tumor is PD-L1 negative by immunohistochemistry it doesn't necessarily mean they can't respond to a PD-1 drug where we also seen that well we I mentioned the kidney cancer data with PD-L1 there's been a recent publication of data in melanoma which suggests a similar story once again a very small number of patients so we don't want to over conclude from this analysis but here the response rate higher than one might expect in PD-L1 positive patients so it seems like PD-L1 on the tumor might enrich for patients likely to respond but still a fair number of patients who don't respond so should we exclude those patients going forward that's an area of debate there are many reasons why PD-L1 negative tumors might respond to a PD-1 antibody here's just a list of some of them I think Charlie's talk focused on one of the major players in this which is PD-L1 expression is heterogeneous not just throughout the primary tumor but between the primary and the met and that's something we're looking into that also the diagnostic antibodies are not necessarily ready for prime time there needs to be a lot of work with them none of them are as yet commercially available for therapeutic use but they may become available when some of these drugs are available as a companion diagnostic and there's there's other other issues here as well which I won't go into but I'll try to pick off some of these issues one at a time looking at this area of tumor heterogeneity so if you ask the question is does PD-L1 expression reliably predict responders I think at this point it doesn't there are several interests other interesting things that are being looked at but most of that data isn't published looking at for example the not just PD-L1 expression on the tumor but looking at the immune infiltrate in the tumor looking at gene expression in the tumor those are all things that are being looked at but have yet to be published in conjunction with these trials so getting back to the tumor heterogeneity question will it complicate biomarker development this is the paper that launched all of this concern we started looking at whether or not there was an impact of discordance of PD-L1 expression in our group at Harvard from Savina Senior Reddy's lab Marcella Kalea presented data at GUASCO this year looking at a very small group of patients 34 who we had both a large piece of their primary tumor and a large excision on a biopsy of their metastasis we're starting to think that in looking at PD-L1 expression need probably large samples to look at this because the expression of PD-L1 is heterogeneous it's also inducible in our cohort 29% of these patients were PD-L1 positive in the primary however only seven of the 10 that were positive in the primary were positive in the metastasis and there was one patient that was positive in the metastasis but not in the primary so in and as I mentioned before the PD-L1 expression at least in our hands was very heterogeneous and importantly it tended to be positive in areas of high nuclear grade so getting back to one of the questions that was in the introductory questions about whether poor risk patients will respond to these agents the general thinking is that immunotherapy in the past has been best in patients with relatively indolent low grade disease that may not be the case with PD-1 and PD-L1 agents in fact if expression is highest in patients with most aggressive disease it may be those patients where these agents would be most active and that would be a particularly interesting and potentially important finding for our patients we obviously need to do a lot more work with that we also found that the PD-L1 expression at least in our hands was more homogeneous in the metastasis and here's some examples of that here's looking at PD-L1 expression in the primary tumor and the metastasis you see some concordance here but here's a tumor where it's we thought it was negative in the primary and positive in the metastasis and positive here in the area of the highest nuclear grade and here's just a look at the data in a similar way looking at the discordance in some patients between the primary and the metastasis so if you're asking the question will tumor heterogeneity complicate bio market development for at least this biomarker the answer is probably yes and we need a lot more data a lot more patients and potentially bigger samples to look at than just needle biopsies of patient metastatic disease another important question with the excitement someone call it hype around PD-1 PD-L1 a lot of patients with kidney cancer have been calling around I don't know if you've been getting these calls asking whether non-clear cell patients can enroll in these trials and for reasons that aren't entirely clear to me although it probably goes back once again to the historical whereas we generally think of non-clear cell patients as less responsive to IL-2 and less responsive to interferon they've been excluded from trials of PD-1 and PD-L1 so for the most part we don't know much about these patients we there are a few patients like this treated in the Genentech trial but they've been excluded from the BMS trial so the question is should they be excluded so Andre Fay who's a fellow working at Harvard with Tony Chewary and Savina seniority looked at PD-L1 expression in non-clear cell renal cell carcinoma here's a group of about 90 patients with non-clear cell disease looking at expression as it relates to grade and stage but also histology in these patients and what you see here at least with this antibody relatively low amounts of expression in chromophobe and papillary tumors and I realize this is a ridiculously small number of patients but of the seven patients with the translocation kidney cancer three of those had higher levels of PD-L1 expression so this is the kind of data hopefully we'll move in the future to allowing patients who had traditionally not had what we would consider immunoresponsive tumors on some of these trials going forward if their tumors are expressing PD-L1 and I think that's a strategy that many sponsors of these trials are starting to take in their development is identifying patients based on expression of PD-L1 and hopefully that will happen in kidney cancer particularly since some of these diseases are really hard to treat affect young people translocation kidney cancer is something we obviously need to do a lot more work on so should non-clear cell patients be excluded from trials I would say we should rethink that and we should try to potentially develop trials and non-clear cell patients to answer these questions so another question is can biomarkers guide frontline and combination trial development this is something that I've heard Bernard talk about before at meetings and he has advocated for you know should we be doing this this is the compares study that he mentioned earlier and Tony Chewary presented a subsequent analysis looking at PD-L1 tumor expression on the outcomes of those patients in this trial and I thought this was result was pretty interesting and what he looked at this is just part of the data what they looked at was looking at PD-L1 we know it's a prognostic marker for overall survival but how did it impact response to VEGF TKI inhibitors and what you see here is looking at PD-L1 expression here is patients with low PD-L1 expression on their tumors here are high and both groups have been treated with either sunitinib or pazapinib and what you see here at the outcomes of patients with PD-L1 high tumors here do much worse with VEGF TKI's than those with PD-L1 low expression I think this is interesting for several reasons obviously once again I think I've said this like six times now you don't want to make too much out of this because the numbers of PD-L1 positive patients in this trial are relatively small but I think this might allow us a direction potentially of getting these agents into the front line you know so for example should we enrich our trials for patients that are PD-L1 positive more importantly if if the agents are more active the PD-1 and PD-L1 drugs turn out to be more active in in patients who typically don't respond to VEGF TKI's that I think that would be a truly important finding for our patients we need treatments for these people are these people the patients that Bernard was mentioning who have refractory disease this would be not only an exciting way to get these treatments to the proper patients but maybe maybe identify patients who should get VEGF TKI's first versus those who should not so right now I would think that biome we should be using biomarkers to get these agents into the front line the faster we can get them there the better because ideally people getting them front line some of those patients would not need subsequent treatment so looking at PD-L1 expression and response I think the bigger problem with the field right now is not that PD-L1 expression isn't helpful I think it is helpful I think the bigger problem is regardless of whether you have PD-L1 up on your tumor or not still the majority of patients don't respond to this treatment which is what we've typically seen with many immune therapies the bigger issue is that even in the best case scenario the response rates maybe two and three times in some studies higher for PD-L1 positive patients but still oftentimes the majority of patients don't benefit so going forward we talked about how inflamed tumors will likely be treated with single agent treatment for those tumors that are inflamed meaning they have PD-L1 expression on their surface but don't respond which is quite common what else what are we going to do for those patients and I would argue that we probably need to be thinking about combination approaches in those patients there are quite one of the interesting things about combinations is one PD-1 and PD-L1 are relatively less toxic than prior forms of immunotherapy so you can imagine combinations of at least two agents with PD-1 antibodies but another important thing is many of the things here on my list and this list could be twice as long of things that you might want to combine with a PD-1 blocking strategy many of these agents are either FDA approved or in clinical development many of these agents have very limited single agent activity and are looking for something to combine with so for example we talk about different strategies that might work well with PD-1 maybe we should be eliminating T-regs through antibodies that targeting CTLA-4 or Gitter should we be inhibiting myelodoride suppressor cells this is work that's a lot of it's been done at the Cleveland Clinic with Brian and Jim Finke looking at should we be combining VEGF TKI's or other agents like HDM-2 antagonists that might exclude myelodoride suppressor cells from the tumor should we be supporting effector cells with agents like IL-2 there are newer versions of IL-2 in development that don't drive regulatory T cells as much at least in preclinical models and there are other more selective T cell agonists like IL-15 and IL-21 that could be tested here should we be supporting the adridic cell functions with GMCSF those very interesting randomized phase 2 data of epilimumab and GMCSF in melanoma which showed a survival advantage in those patients with a combination so it could that be something we test in kidney cancer as well you know going back to these are the original questions that we had I don't think that one of the exciting things about immunotherapy is that the responses that we see are not necessarily disease specific they may be able we may be able to take something we learn from say melanoma and bring it to kidney cancer and have some positive effect and vice versa and there are obviously other checkpoint inhibitors that are in clinical development already that may make sense to combine so here's an example of once again a very small study this here in melanoma looking at PD1 blockade with the volumab and CTLA4 blockade with epilimumab and this data is you know some of the best you know stage 4 oncology data that I've ever seen this is looking at patients who hadn't had treatment before so this is probably a very selected group of patients this is looking at their tumor shrinkage from the startup treatment and what you're seeing here is the majority of patients almost 50 percent are having a response but not only that the depth of response is pretty impressive and the reason for that we get excited about depth of response at least in immunotherapy land is because these patients that often go on to have remissions that last years so small numbers of patients very well selected patients but there's a signal here of increased effectiveness of combining two checkpoint inhibitors in melanoma here not only are the responses deep but they're happening quickly here the response is happening within 12 weeks of starting treatment but as with many combinations they are obviously more toxic and that's going to play a big role in developing these across tumor types here's looking at the concurrent treatment and the sequence of these agents we're trying to understand how best to give them at the current time but you can see grade three and four toxicity of 50 percent which is non-trivial and you know patients need to be carefully instructed and very closely monitored with this combination because it is toxic but because of the clinical efficacy this combination is already in development in lung cancer and in kidney cancer and it's moving quickly to phase in phase three trials in melanoma so one of the more interesting findings in this trial not only was the efficacy but here looking at pdl1 expression in response we talked already about how pdl1 expression seems to enhance your likelihood of responding to one of these agents here looking from that combination study admittedly very small numbers so we're not going to get too excited but the response rate in pdl1 negative patients was as good as it was in pdl1 positive patients if this data is real it suggests that adding epilimumab is doing something to the tumor microenvironment that makes tumors more likely to respond could it be depleting regulatory T cells that's one possibility we don't really understand it but if this if we can take pdl1 negative patients and make them respond by combinations I think that would be an exciting prospect to move the field forward not only when you talk about immune checkpoints the the number of immune checkpoints or these so-called breaks on the immune system is enormous this is a just a figure from a nature reviews article but interesting thing about many of these is they can be targeted with monoclonal antibodies and many of these antibodies are already in the clinic in phase one so here's an example of preclinical data that led to a phase one trial that's ongoing this is from Chuck Drakes group at Johns Hopkins this is looking at the synergy between pd1 blockade and another immune checkpoint called lag 3 this is looking in a mouse colon cancer model mc38 in the control mice they all do poorly fairly quickly adding lag 3 there's some delay in tumor progression adding pd1 there's further delay the combination seemed to be better and now that's combination is being tested in a phase one trial for multiple solid tumors so when you look at the current pd1 trials in kidney cancer there are quite a few of them if you go to asco this year you'll see data on the phase two nevolumab trial the biomarker study this combination trial of not only pd1 with tkis but pd1 with epilimumab will be presented as bernard mentioned the phase three trial will have to probably wait until next year to see that and there are several interesting trials that are being launched looking at tkis and the murk pd1 antibody which is at least as active as the bms per pd1 antibody and melanoma and lung cancer we're going to get our first taste of it here in kidney cancer and see how active it is and pdl1 in combination with bevacizumab and a randomized phase two design against sunitinib and here you can't really see it but there's the lag pd1 antibody so a lot of trials with data coming quickly it'll answer some of these questions both clinical and some biomarker development questions so if i had to look forward a year and make some predictions not that you asked for my opinion but i'm going to give it about these pd1 pathway inhibitors i i think several of them will probably gain fda approval and several solid tumors that does not mean that the kidney cancer trial will be positive i have no idea it may not be but several of these trials will likely be positive and several of these drugs will likely be available it's also likely true the pdl1 expression both either in the tumor or in the infiltrate of the tumor will predict for a higher likelihood of benefit to these agents but the majority of unselected ppi treated patients will require salvage treatment and why is that well that's because of this which is the non-inflamed tumor which probably represents the majority of solid tumors out there in in most cancer clinics and these are tumors this is back to the gaevski model again this is a tumor that has a high expression of vascular markers a lot of infiltration by macrophages maybe these myelo derives suppressor cells fiber blasts not a lot of inflammation not a lot of chemokines not a lot of recognition by the immune system not a lot of lymphocytes in this tumor so you're not getting the effector cells at the tumor you're not getting pdl1 expression and this is the more common problem that we deal with so going forward we talked about different ways we might improve our immunotherapy the biggest challenge for us next will be how do we treat these non-inflamed tumors the tumors that are being completely ignored by the immune system and one of the possibilities is we're going to have to do something to induce anti-tumor immunity there are many different strategies to think about how we might do this um increasing antigen presentation by combining immune therapies with things like radiation for example or intra-tumoral interferon using demethylating agents to increase antigen expression it may turn out that some of these tumors with higher antigen expression may be more sensitive to to immune therapies we also may want to improve the focus of the immune response through vaccines there are two phase three trials with dendritic cell based vaccines in kidney cancer if those were to be positive that would be an obvious next step for combinations in kidney cancer but here's a look at how sort of predicting um you know how we go forward is challenging as charlie mentioned we don't have great mouse models um for kidney cancer we certainly don't have great mouse models for immunotherapy and kidney cancer this is a rancor based model that's the slides come to me from gordon freeman and what he wanted to to look at is how do his pd1 and pdl1 agents work in a rancor model and what you see here with both single agent mouse at bulimumab and mouse pd1 very few responding patients it was only when he combined these drugs with irradiated rancor given as a vaccination that he saw activity um so while and if you had just based on gordon's models you probably never would have developed a pd1 or epilimumab and kidney cancer but it turns out it might be a useful model for studying resistance and it may be that you know vaccine plus checkpoint inhibition as you see here but with both mouse ipi and mouse pdl1 there was a slightly better outcome in that setting as well so more to come um so but ultimately i still think this is going to be a subset story meaning at the moment it's the minority of patients that will benefit so it's up to us to identify the patients who have the right immune response the right tumor factors that allow them to benefit from immune therapy and it's there that we should focus our efforts still treating the most appropriate patients not treating everyone and this is an opportunity for us in the translational world to identify those patients i think if we do this i think we can we can achieve dr albijay's goal of more cures for patients with metastatic kidney cancer through targeted immunotherapy thank you very much for your attention that's very much indeed david don't stay there stay there i want to ask you a question at least if no one else does um that was wonderful um are there any questions from the audience tim obrani going to the microphone yes tim sensational talks guys really really enjoyed those can i suggest a really crude way of switching on the immune system in a tumor yeah and that's to embolize it there's some really interesting case control data that suggests that tumor embolized not only they have less bleeding during surgery but you you get an incredible immune reaction when you embolize a tumor fever and i just wonder whether that would be a very crude but very simple way of switching on immunity in a tumor um i think the short answer is yes and i think the short answer is it should be tried um but i think that the practical issue is the hard one which is how do you get someone to sponsor a trial like that because so for example um you know you mentioned embolization the the folks who would argue for that radiation does the same thing that focus radiation will kill a tumor um release antigen and produce responses there have been anecdotal stories of that effect so it's a similar concept where you hit one tumor release antigen and you produce a a scopol response elsewhere but actually organizing a randomized trial to do that is a challenge not to say it's not possible um but there's it requires independent resources and very organized very committed uh individuals so for example we're trying to get a radiation plus il2 trial going for years based on just single institution data that suggests that's a good idea it's been very hard to fund pay for the radiation you know because if it's considered research it doesn't get paid for so what you'd need is if it's you you would need to handle some of those practical issues so for example at least in the u.s. we can't um do something that's not standard of care um and not have a way of paying for it does that make sense just do the things the drug companies want us to do i'm not arguing what's right the enemy is the disease i'm with you i'm not i'm i'm just saying from a practical point of view so for example getting back to my radiation example i'm as far as i'm concerned that trial should happen and it should be done but what i'm saying is from a practical point of view for a trial to happen it has to be supported at least some way at least in a small amount and so for example in the case of radiation and i think this would be true with your ablation idea is if the ablation is not considered clinically necessary then at least in the u.s. who pays for it so if no the answer is nobody then the trial has to pay for it so then the trial organizers have to find a way to pay for it and that's where a lot of these ideas fall short i'm not saying they shouldn't be done they absolutely should be done and it may turn out that the crudest methods are the most approach the most efficacious because it we have all these very sophisticated vaccines that are being developed where we think we know the antigen that we need to target it may turn out that the antigen that's important for one person to the next is totally different and you know producing a storm of antigen may be the way to go i'm just saying from a practical point of view it's been very hard for us to get those trials off the ground but i'm with you they should happen okay one last question i'm going to ask if i may because i don't think there's any other so melanoma now for two or three years i guess we've had chat points inhibitors which we can use in the clinic and at least a single agent so they seem to be pretty well tolerated along with targeted agents at the same time and quite quickly we've moved i think in melanoma to a paradigm where we're actually trying to use upfront immunotherapy if we can potential durable benefits and then save our targeted therapies for salvage bearing in mind some of the limitations we've heard about for targeted therapies already do you think in the next five years we're going to see that kind of transformation in kidney cancer not as quickly um what i like to see it sure but there but there are some issues so in the melanoma world there are many more believers in immunotherapy than there are the kidney cancer worlds um you know and there was so there's a longer tradition of immunotherapy going back to interferon and hydrocyl 2 there are more people who want to believe that immunotherapy is an active approach that advocate for it that push back a little bit on the b-raff um targeted agents and now we actually have some data that if you start in melanoma with b-raff you might be preventing some patients from ever going on to get immunotherapy because in the case of melanoma the disease is very often very aggressive and was failing as as bernard pointed out with failing vege of tkis there's some patients who fail b-raff therapies you stop b-raff and they go downhill very quickly so there's a there's a sense that there's a window as you pointed out in melanoma that you need to act the patients are young the patients are motivated they're willing to tolerate a lot of toxicity there many of their doctors believe in immunotherapy that's different than the kidney cancer world we need to change that but one of the ways we're going to change it is if we can identify the patients who should be getting these drugs not arguing that immunotherapy for everybody but immunotherapy for the right patient once we if we can do that effectively which we've not been able to do with it be we've not been able to do with il2 coming up with selection criteria then we can get them in the hands of people frontline i think we'll make progress for for a group of patients okay thanks david that that was wonderful okay so um the this is the end of the session the next session actually starts at heart bus three and it's a satellite symposium so what i'd like to do is ask you all to thank uh charlie swanson and david but for two wonderful talks