Alicia Morgans: Hi, my name is Alicia Morgans, and I'm a GU medical oncologist at Dana Farber Cancer Institute in Boston. I'm so excited to talk today with a good friend and colleague, Dr. Matt Galsky, who's a professor of medicine and the director of GU Medical Oncology at Mount Sinai in New York. Thank you so much for being here with me today, Matt.

Matthew Galsky: Thank you.

Alicia Morgans: Wonderful. I wanted to talk with you a little bit about an oral presentation that you did at ESMO 2021, really digging into an exploratory analysis within IMvigor130, where you looked at the differential responses that patients may have had, whether they were treated with cisplatin versus carboplatin within that study. Of course, the other part of that was the addition of atezolizumab in the treatment arm. So can you tell us a little bit about what you and the team investigated, maybe starting with just a brief overview of the schema of IMvigor130 and the purpose of the overall trial before you get into that?

Matthew Galsky: Absolutely. IMvigor130 was a randomized phase III study of patients with metastatic urothelial cancer who were treatment-naive, and it was really asking two separate questions. One question was whether or not we should combine platinum-based chemotherapy with atezolizumab versus giving platinum-based chemotherapy alone, like we've been doing for decades. The other question was, should we give immune checkpoint blockade alone versus platinum-based chemotherapy? So this analysis really focused on the chemo combination issue, and the reason that this analysis was pursued was because in the analysis of chemo combined with the Atezo versus chemotherapy alone, the progression-free survival endpoint was met, there was a significant improvement in PFS, the overall survival has not yet met statistical significance, but even if you look at the PFS benefit, it's relatively small compared to what we see with some other tumor types like lung cancer, when you combine chemotherapy and immune checkpoint blockade.

So this really raised the question, why are we seeing these differences in urothelial cancer compared to lung cancer or other cancers? On a subgroup analysis of that primary endpoint, one of the things that jumped out, remember subgroup analyses are hypothesis-generating, one of the things that jumped out was that the effect size for PFS, and OS, actually, with cisplatin-based chemotherapy, with adding Atezo to cisplatin-based chemotherapy, seem to be much greater than when adding Atezo to carboplatin-based chemotherapy. And so we really wondered why that might be mechanistically, and we pursued a series of investigations to try and dissect that better.

Alicia Morgans: So first, that's really, really interesting, because of course, cisplatin is our favorite chemotherapy when we're treating urothelium carcinoma. And I don't know that anyone has really done a deep dive into why, mechanistically, that might be the treatment effect that we see when we're just doing normal treatment of patients with urothelium carcinoma. So I think this is extremely interesting from that perspective. And then, certainly, it's interesting from a mechanistic perspective in combination with atezolizumab. So what is it that you saw? What investigations did you do?

Matthew Galsky: People talk about this tail on the curve with cisplatin-based chemotherapy versus carboplatin-based chemotherapy, and as you point out, in model systems, the differences between cisplatin and carboplatin have been pursued for decades, but we don't have a great number of studies using human tissue to try and understand durable responders to cisplatin-based chemotherapy. And so the first thing that we did was really just look whether or not the baseline tumor microenvironment might matter with regards to cisplatin versus carboplatin-based chemotherapy, both on the control arm of the study and on the Atezo combination arm.

And really interesting, what we found was that in tumors with high levels of PD-L1 expression, and this nuance is important because we talk about PD-L1 expression, there are different assays with different therapeutic antibody staining different cell populations. Remember, SP142 is specifically focused on infiltrating immune cells, PD-L1 staining on immune cells. So in tumor specimens with baseline increased PD-L1 expression using this assay, patients treated with carboplatin-based chemotherapy, didn't matter what the baseline PD-L1 expression was. Made no difference in terms of the survival outcomes. But in patients treated with cisplatin-based chemotherapy, remember, alone, on the control arm, in patients with high PD-L1 expression, you had a markedly different long-term outcome compared to patients with low levels of PD-L1 expression. And that really indirectly raises the concept, are durable responses to cisplatin-based chemotherapy related to the immunomodulatory effects of cisplatin?

When you do the same analysis with the Atezo combination arm, you actually see the same thing, but it's magnified a bit. So now, not much of a difference with PD-L1 expression when you look at carboplatin plus Atezo, but with cisplatin plus Atezo, the curves splay even further apart. That really led to the next set of questions, which is, what's happening in the tumor microenvironment that might be related to immune modulation with cisplatin that makes it pair well with the atezolizumab? And unfortunately, on this study, we didn't have serial tumor tissue. We didn't have post-treatment or on-treatment tumor biopsies. But we did have serial peripheral blood mononuclear cell collections.

So what we did was take cycle three 1 PBMCs and cycle one day one PBMCs, and those underwent single-cell RNA sequencing. And so, in total, 142 specimens underwent single-cell RNA sequencing, which is a pretty large cohort for single-cell RNA-seq. And those cells underwent graph-based clustering to identify the major cell types in the peripheral blood based on gene expression. And then we looked at what genes in gene sets were modulating on treatment with cisplatin versus carboplatin. And ultimately, what you see is a market up-regulation of gene sets involving innate immunity with cisplatin-based chemotherapy cycle three day 1 compared to baseline versus carboplatin. So specific immunomodulatory effects related to cisplatin that may reflect why cisplatin leads to better durable outcomes and pairs better with atezolizumab.

Alicia Morgans: That's fascinating. And like you said, that actually is a very big cohort that you were able to perform that RNA-seq, and I'm very glad you were able to get it over time. Because just to remind everyone the tissue that you had when you were looking at the PD-L1 expression, that was really a baseline pretreatment sample. So really fascinating, and I think that this will be so interesting to look at in paper form when you and the team get this out and are really able to show us with the complete analysis that you're not able to present in a shortened abstract form at a meeting, because I think this will be, as you said, hypothesis generating for future studies. And along those lines, what does that mean for the field in terms of future studies? Where do you go from here, in terms of investigating this and in terms of thinking of future combinations, perhaps with atezolizumab, for patients that may be more effective, at least positive in terms of a clinical trial like this?

Matthew Galsky: A few considerations. One, is that the overall survival data from IMvigor130 hasn't yet read out in final form. So I think in the context of those results and in the context of the subgroups, I think we'll have to go back and ask ourselves again, is there a role for chemo combination, specifically cisplatin-based combos with Atezo. The second is that, as you know, there are other phase III studies that haven't read out yet that are actually enriching for cisplatin-treated patients are only treating groups with cisplatin, with or without immune checkpoint blockade. That will be another test of this concept in a prospective fashion. And then I think the third issue is, how do we take this information to inform combinations in the future? And for that, I think we really need to understand precisely what cisplatin is doing in the tumor microenvironment to develop, so-called, rational combinations, and there's work being done now in model systems to compliment this, work in human, sort of a reverse translational approach, that will hopefully inform those combinations for the future.

Alicia Morgans: That's fantastic. Well, I always love talking with you, Dr. Galsky. I feel like I learned so much and each contribution that you make certainly helps me rethink the way that things are done in the field. So thank you so much for your expertise and for sharing your work today.

Matthew Galsky: Thank you. Always a pleasure to speak to you.