Alicia Morgans: Hi, this is Alicia Morgans, Associate Professor of Medicine and GU Medical Oncologist at Northwestern University in Chicago in the U.S. I am so excited to have here with me today, Professor Tom Powles, who is the Director of the Barts Cancer Centre in London, in the UK, as well as Professor Matt Galsky, who is a Professor of Medicine at Mount Sinai in New York. Thank you both so much for being here today.

Matthew Galsky: Thank you.

Thomas Powles: Thank you for inviting us.

Alicia Morgans: Wonderful. So I wanted to talk with the two of you today about some exciting data that came out of ASCO, the virtual meeting, 2020, and let's start with IMvigor010, which is a study of adjuvant atezolizumab in patients who have muscle-invasive urothelial cancer. Dr. Powles, you were one of the lead authors on this trial, can you tell us a little bit about the study design and why it was so important, what you found?

Thomas Powles: Thanks, Alicia. So Joaquim Bellmunt is the Chief Investigator, Maha Hussain presented it at ASCO, and I'm on the steering committee and there are a number of people who have been involved in the studies, not least the Roche group, which has done work in this area. It's not the result we were looking for. So the result we were hoping for, I was hoping for, was as we moved these drugs earlier into the disease setting, into the non-muscle-invasive and muscle-invasive disease, I was hoping, because we know the immune infiltration is higher, and the PD-L1 expression is higher in muscle-invasive disease compared to the metastatic disease, I was hoping we'd see a bounce in response higher, survival signals and significant benefits associated with adjuvant therapy. This was an adjuvant study of atezolizumab versus best supportive care and on a one-to-one randomization.

The trial was a year of atezolizumab, which I think is an appropriate amount. And the population was quite a high-risk population, mainly T3 or T4 disease. Some patients with residual disease T2, often neoadjuvant chemotherapies, but essentially it was a proper high-risk population, and the results for disease-free survival with disappointing negative with a hazard ratio of 0.89. When one looks at the shape of the curves, they go apart a little bit and then come back together again, making a sort of a thin banana type shape, which is not what you want in a clinical trial when you're hoping for longterm durable benefit. And so it suggests that longer follow-up probably wouldn't be beneficial.

Surprisingly, in the biomarker-positive subgroup, the hazard ratio was 1.0, and I'm really struggling with that conceptually because we know that in the frontline metastatic setting, biomarkers are important for selecting patients to benefit. And so, one would expect a little bit of enrichment. Now I know that there's statistical noise associated with this, but one would have expected the numbers to be going in the right direction. And again, there wasn't this tail that we expected.

And then the final issue is around the survival signal of 0.85 premature. To some extent from the follow-up being currently modest, but obviously it won't be statistically significant because of all of the outliers being spent on disease-free survival, but there is an exploratory survival signal that we should keep an eye on in the future. The subset analysis, perhaps some benefit in the T4 population on the forest plot analysis, but very wide hazard ratios. And there's a bit of a [inaudible] forest plot analyses I'm always nervous about. So in summary, for me a real disappointment. I think that there's nothing wrong with the drug atezolizumab, it has performed well in urothelial cancer. There was nothing wrong with the trial design, as I can see it, the population's pretty clean, as Maha presented. And so, really complex for the future.

Alicia Morgans: But really an important study to complete, I think, and an important negative to be found because this has been an area of a lot of interest. As we treat patients, particularly with neoadjuvant chemotherapy and they go to cystectomy. If they have residual disease, I think we all as a field, want to do something more if we can. So this, I think, was a really important contribution to the field that at least this approach is not the approach that we hoped it would be. And there are actually several additional trials that are looking at adjuvant checkpoint inhibitors in this similar patient population. Dr. Galsky, I'm wondering what your thoughts are on those other studies that are looking at a similar question.

Matthew Galsky: So I would say, first, to your point that the fact that this trial was completed, we probably shouldn't gloss over. We have had struggles completing adjuvant studies in this disease for decades, and this trial was a large randomized trial and it was fully accrued and that's an incredible landmark in this disease and shows that this can be done.

The question about what this means for the field in general? I was sort of thinking about this as, as Tom was discussing it and I think I have the same things in mind that he was saying, which is that you can boil this down into whether or not this is a failure of drug, of trial, or of concept. And so is this a failure of drug? Does P1 versus PD-L1 matter? I think most of us feel probably not, but we don't know that a hundred percent. And so that's a reason to keep these other trials going and finish them up and answer the question definitively.

The second, is it a failure of trial? Is it simply that the cohort of patients that were enrolled were balanced in a way that doesn't allow one to detect the signal adequately and that's just bad luck and that happens sometimes. And that's another reason why we should do the other trials.

And then the third is this a failure of concept and we've seen in model systems in small randomized studies that giving some neoadjuvant treatment might be necessary and might be better than just giving adjuvant treatment with immune checkpoint blockade. That's been seen in melanoma. It's been seen in both mice and in human studies in small cohorts. It's been shown in glioblastoma. And so if that's at play, then, unfortunately, these other trials will not be successful either. But the only way to know that is to finish them and analyze them.

Alicia Morgans: I agree. And I'm very glad that we, as a field of physicians, treating patients with urothelial cancer and the patients themselves have been able to complete this trial because as you said, this has actually been a challenge of ours in the field of urothelial cancer, bladder cancer, upper tract disease. So this is definitely important data and negative data can be equally informative. And so Doctor Powles, before we move onto the next trial that we wanted to talk about, some biomarker data from IMvigor130, what would be your take-home message from IMvigor010?

Thomas Powles: So the first message is an important one, is we nailed, really nailed an adjuvant study and we recruited 800 patients. It happened fast. We did a clean study and we've gotten a really good result. It wasn't the result we were looking for but we answered a really important scientific question. Remember, we've done a whole series of chemotherapy adjuvant studies over the last 20 years and to some extent, neoadjuvant studies, which on the whole have been unsuccessful and they've been unsuccessful because we haven't recruited, which has meant we haven't been able to persuade the community to give adjuvant or neoadjuvant chemotherapy consistently. Even neoadjuvant chemotherapy, we struggled. And I think that this trial shows us that we can recruit successfully and answer important questions. The important thing about that, it allows other people to look at what we've done and then say, yes, we can invest in urothelial cancer, emotionally and effort-wise, to try and get a positive result.

And we will get positive adjuvant and neoadjuvant studies in the future. Perhaps not today, perhaps not this year, but we will in the future, with all of the drugs that have become available to us and all of the opportunities and the neoadjuvant opportunities. I predict we will have positive studies in the not too distant future. So this will go down as the first of a number of crucial steps and the investigators and the company and the patients should all... It was a huge effort and I think it's a really important step. I think that the principle of where we go from here, Matt sort of described very nicely. There's a lot of uncertainty. I'm less positive about the peri-operative space than I was a year ago, obviously, but I still think, as I've said previously, that we will get there in the end with a different combination.

Alicia Morgans: Great. Well, thank you for sharing the data that you and the team have completed and presented at ASCO. I certainly appreciate it. And we're going to switch gears slightly to talk about IMvigor130 and some of the biomarker data that Doctor Galsky has actually presented at ASCO this year. And just to remind everyone, this is the study of metastatic urothelial carcinoma and we looked in this setting, at a combination of platinum-based chemotherapy with atezolizumab or atezolizumab as single agent. And Doctor Galsky, you had some interesting data that I think is important to share regarding biomarkers and it might help us understand and inform our patient selection in terms of atezolizumab monotherapy. Can you share that?

Matthew Galsky: So I wanted to focus on one finding in particular, in the context of some of the treatment paradigms that are evolving, like maintenance therapy and whether or not we should be combination chemotherapy and immune checkpoint blockade upfront. Certainly, what's missing in that dialogue is the role for single-agent immune checkpoint blockade up front and sort of where is that left? And so this is exploratory data. It's exploratory in the sense that the analyses weren't predefined and it's exploratory in the sense that we looked at an overall survival endpoint when the study hasn't fully readout for overall survival.

But I want to focus on one group and we know from multiple studies that TMB and PD-L1 expression convey independent prognostic or predictive information. One would think those two are well correlated with one another, but it turns out they're not, for unclear reasons. And they convey independent information. If you look at patients in the IMvigor130 study, who have both high PD-L1 expression and high TMB, which is about 15% of the population, and you look at the arm that received a T-cell single-agent versus chemotherapy alone, the hazard ratio in that group in favor of a T-cell is 0.22. And so there's probably a group that might benefit from upfront immune checkpoint blockade. Again, this is exploratory. It's a small set in the context of the larger study, but I think we need to pay attention to what these biomarker exploratory analyses are telling us in terms of how we think about future paradigms.

Alicia Morgans: So thank you. So, Doctor Powles, what is your thought? That's an impressive hazard ratio, I think.

Thomas Powles: Yeah, that's tricky results.

Alicia Morgans: A subset analysis, exploratory, of course.

Thomas Powles: We did some work in the neoadjuvant setting with the same, and we built those models and if you build TMB and PD-L1 together, and you look at CD8s and there are other TFA [inaudible] ratios. I think that there are a lot of overlapping biology between CD8, PD-L1, but the TMB does seem to be slightly different. And we did some work with 211 with that as well. And it doesn't surprise me what Matt has shown. In the back of my mind, there's always been that independence there. One of the key issues around Matt's data, which he's probably underselling because he does to much of his data, is that actually the important thing around it is it's randomized. And so all of the work we've done before, there's always been this hypothesis around whether TMB, kind of unstable, would be better with chemotherapy. Same about PD-L1. We know with 211, that was the case.

And when you're getting into values of 0.2s, you can let that drift a few points in the wrong direction. It might be 0.4, but it's still a really, really good result. And I think what this is telling us, which I think Matt is saying, but I'd like to hear what he thinks, is that actually, we will have frontline biomarkers in the future, which will be determining much more, how we use therapy than we are currently. Because currently I'm not convinced the global community, in the same way as we're not totally bought into neoadjuvant chemotherapy, although some of us are using biomarkers in the frontline. I think many people still say gem/cis [inaudible] and see what happens after. But when you get the biology down to 0.2s, you're beginning to sort of really say, well, no, but we do need to be doing this and of course, other trials can then look at this in more detail.

Alicia Morgans: What are your thoughts, Doctor Galsky?

Matthew Galsky: I completely agree. And I think that even though the switch maintenance strategy allows one to not think about what sequence matters and at the end of the day, it might not matter. These data do provide a little bit of food for thought, in terms of maybe what you pick first might matter to some patients.

Alicia Morgans: I think particularly in the context of being potentially able to spare patients from chemotherapy, this would be transformative in patients' lives for sure and would provide some direction for the physicians who were trying to treat them too. So thank you for sharing this. And I really do look forward to hearing more on it, seeing the publications, of course, that will come out of this. And of course, I'm sure hearing at future meetings about other analyses that you and the team are doing. So thank you both so much for your time today. Congratulations on your work. Negative or positive, this work really informs the field and is important for people who are trying to deal with this disease and those who care about them. So thank you both.

Matthew Galsky: Thank you.

Thomas Powles: Thank you very much and see you soon.