Alicia Morgans: Hi, this is Alicia Morgans. I'm a Medical Oncologist at Northwestern University, and I am so excited to have here with me today a good friend and colleague, Ronac Mamtani, who's an Assistant Professor of Medicine at the University of Pennsylvania, where he's a GU Medical Oncologist in the Division of Hematology/Oncology. And his friend and colleague, Dr. Ravi Parikh, who is an Instructor of Medical Ethics and Health Policy, also at the University of Pennsylvania in the Perelman School of Medicine. He's also a practicing GU Medical Oncologist. Guys, thanks so much for joining me today.

Ronac Mamtani: Thank you so much for having us, Alicia.

Alicia Morgans: Great. Ronac, we've talked over the years. You've done a lot of work, and now are doing this work with Ravi as well, but looking at what I really think of, and that you've described as cancer outcomes and effectiveness research that I think has the possibility to really influence policy guidelines and everything else and the way that we actually use the data that we have. Can you tell us a little bit about some of the work that you did recently published, looking at the effectiveness of first-line immune checkpoint or IO therapy versus carboplatin chemotherapy for patients with metastatic urothelial cancer?

Ronac Mamtani: Absolutely, Alicia. Our research program really tries to focus on generating real-world evidence of effectiveness. And this work is really exceptional work that was actually led by Dr. Emily Feld. She is currently a third-year fellow at Penn. For those that are listening and don't know her, you really should get to know her.

Alicia, this study compared immunotherapy with carboplatin-based chemotherapy in patients with metastatic bladder cancer. And I think the listeners know that the study was really motivated by a complete lack of data that directly compared these two treatment strategies in bladder cancer. Despite these approvals for immunotherapy more than two years ago, which was clearly a pivotal moment in the history in our community, clinical decisions still relied on cross-trial comparisons. In fact, really the only real guidance for physicians was last year's FDA safety alert and label revision. And that was based on some of the results from IMvigor130 that was presented a few weeks ago. Dr. Feld and I felt this was the right time for a comparative effectiveness study.

Alicia Morgans: Can you tell us a little bit about the dataset that you used to do this. I think that's always really important when you're thinking about this kind of work. And what were the outcomes? What were you looking for?

Ronac Mamtani: Yeah, absolutely. This was a retrospective cohort study among patients in Flatiron Health. Flatiron is a really unique dataset. It is an oncology-specific electronic medical record. We felt, Dr. Parikh, Dr. Feld and I felt that the database really offers many important strengths when compared to our traditional datasets like claims data or registries. For example, it has real-time data with minimal lag. It has detailed treatment information. And it's reliable. So, that's the dataset that we used. What we did was, with a primary outcome of overall survival, we compared first-line immunotherapy initiators with first-line chemotherapy initiators. And we used an advanced statistical technique called propensity score analysis, specifically inverse probability of treatment waiting, to really try to address imbalances and confounders.

Alicia Morgans: So I think that I really want to emphasize that you did that sort of advanced technique and analysis, which is I think one of the reasons why this data is so valuable. Because when you don't have a randomized study, you can have actually have patients who are very, very different, getting different treatments because of unmeasured confounders. And when you do something like you did, this propensity weighting and matching, you can actually sort of adjust for that and then actually then compare outcomes between groups even though you have a non-randomized trial.

Ronac Mamtani: Absolutely. But I think for the listeners we need to emphasize that look, this is not a magical solution. There's no replacement for randomized data. But these are the type of data that can really complement clinical trials, and we just have to be transparent that this methodology doesn't do anything about unmeasured confounders, which may not be recorded by the dataset and may influence treatment choice. But you're exactly right.

Alicia Morgans: Absolutely. Well, and thank you for making sure that that is clear. And so if it's not measured, we're not going to be able to do it. But this is a method to try to make things a little bit more balanced. And so I do appreciate that you guys did that. And what did you find? Was this consistent with what current guidance is regarding these populations?

Ronac Mamtani: Yeah, it's a great question. So we had about 2,000 patients. It's a good size. And the model that we used really resulted in near-complete balance of the covariate. So the methodology worked. And what we found is lower median overall survival in immunotherapy group versus the carboplatin group. So nine months versus 11 months. But what was really interesting is that the curves crossed around 15 months. And as such, the survival rates in the immunotherapy group were a little bit lower at the one-year mark versus chemotherapy. And this actually corresponded to an increased hazard of death, so with a hazard ratio of 1.37. And I think what's so interesting about this analysis is this provides some insight into that recent FDA safety alert and the IMvigor130 data that were presented.

Alicia Morgans: Absolutely what I find so fascinating is this is sort of what we would expect to happen, but no one's actually shown in real-world populations that for those patients who respond, they can have these durable responses that are going to be longer than chemotherapy. But this is just a really beautifully performed study to really demonstrate that in a real population.

Ronac Mamtani: Right. yeah, exactly. And when you look at beyond 12 months ... so first 12 months you see a higher risk of death with immunotherapy. Beyond 12 months that's when you see and appreciate the tail of the curve that we've all been excited about with immunotherapy. That survival was improved in the immunotherapy group with a three-year overall survival rate of about 30% versus 13% for chemotherapy. And that hazard ratio was 0.5. So it's complicated, but it's just an interesting analysis and really exciting results.

Alicia Morgans: Absolutely. And I think that what's also important is that your group also tries to understand in the real world what do changes in policy that are affected by this kind of work, what actually happens when these things happen? So the FDA put out an alert and really suggested to us that if a patient's chemotherapy eligible, we should be using chemotherapy based on work like yours that helps to demonstrate that those patients are going to have a superior survival with chemotherapy upfront. But Dr. Parikh, you've actually done some work too, to look at how clinicians have responded to that guidance that came down from the FDA. What was that study?

Ravi Parikh: Sure. So this is work that we published recently in JAMA that essentially performed a quasi-experimental study looking at trends in immunotherapy and chemotherapy prescribing before and after the FDA label restriction that you just referred to.

And I think it's really important as a more general point is that these immunotherapy agents in bladder cancer received approval based under the accelerated approval program, based on single-arm Phase II trials, at least in first-line use in metastatic and advanced urothelium carcinoma. And this has come under a lot of scrutiny recently because the accelerated approval provides earlier access to these agents, which is a good thing and a good thing for many other drugs that have been approved through the program. But it comes at the expense that we don't necessarily have this gold standard randomized evidence when these drugs are coming to market.

And in some rare cases like what was seen in urothelial cancer, single-agent immunotherapy in the first-line setting actually came under some safety concerns when confirmatory Phase III testing was run that led to this label restriction against single-agent immunotherapy for PD-L1 negative patients. And so in our study we looked at what actually happened in the real world before and after that trial.

Alicia Morgans: So what did you find? Because oncologists are not always, not oncologists specifically, but doctors, in general, are not always so nimble in responding to new data. Now, this I think as a label restriction or label guidance, my hope would be that we could be more rapid in responding to that. But that's what you were testing. How quickly could we actually change practice based on this guidance? What did you find?

Ravi Parikh: Sure. I should say that we had the same intuition that you did that even though this was a label restriction, because immunotherapy at the time of the label restriction represented over half of first-line systemic agents that were used in the real world in the United States, we were expecting that it would take some time to respond to this. I mean, this was a major change in oncologists' practice.

And actually what we found was that within about just a little over six months after the label restriction, use of immunotherapy declined by nearly 50% even when adjusting for patient and tumor-specific factors as well as geographic factors. And I think it really shows that, to your point about oncologists being nimble, oncologists, even in the absence of published Phase III evidence, really responded quickly to this FDA label restriction. And arguably in favor of what the evidence showed and responded very quickly to some of these safety concerns. I think generally from a policy standpoint, this bodes somewhat favorably for the accelerated approval program that oncologists respond so quickly to this and respond nimbly to emerging safety alerts. But I think for the bladder cancer-specific community, I think it illustrates how rapidly our practice shifted after the FDA label restriction.

Alicia Morgans: Which is really encouraging for patients, I think. Because bladder cancer has been, if nothing else, an area of many shifts in the last few years, and there are many more to come. So it's encouraging that physicians are listening, that patients are getting what they need. And I'm just wondering from the two of you, are there any limitations, anything that we should think about as we think about this kind of data analysis? And then of course, what are the strengths and take-home messages that we should all consider?

Ravi Parikh: I think that in terms of some limitations, the dataset that we're using, as with any real-world data, is certainly subject to selection effects and selection bias in terms of what treatments oncologists choose for their patients. While we have tried to adjust for all of those effects in all the analyses we've run, unmeasured confounding is a huge problem in real-world data. I think in terms of one take-home that I've taken from this is that in terms of looking at trends in prescribing patterns and real-world sort of real-time impacts of policy changes and label changes from the FDA, this is arguably one of, if not the best, sort of data source in general to actually look at some of these prescribing changes, and I think that as some of these policies and label restrictions come out in more areas of genitourinary oncology in the future, we should really be capitalizing on the power these datasets offer in terms of looking at real-world changes and disseminating how oncologists are responding.

Alicia Morgans: I agree. And Dr. Mamtani, Ronac any closing thoughts from you or messages for the listeners?

Ronac Mamtani: First it was so much fun to work with Emily and Ravi. I think they're rising stars. But I think the one take-home message is that we can really use real-world data for many reasons and many purposes other than just comparative effectiveness research. We can use real-world data to provide sort of real-time insight into policy impact and hear an FDA warning. That postmarketing studies are critical and for everyone to remember some questions cannot and won't be answered by clinical trials.

Ravi Parikh: Alicia, if I could make one quick point. I think the other thing as someone early in their academic career that was really heartening for me is just how valuable collaborations are between academia, policymakers and industry. In the case of the JAMA paper where our co-first authors and co-senior authors were Blythe Adamson and Aaron Cohen were both working at Flatiron and where one of our co-authors, Sean Khozin is actually involved in the FDA. I think that the design of this analysis and sort of the kind of dissemination and interpretation of the findings that we saw were particularly powerful because we had inputs from a lot of stakeholders that were affected by the label of restriction.

Ronac Mamtani: Yeah. This is a multidisciplinary team, Alicia, that involved our lead biostatistician also at Penn, Rebecca Hubbard. So it's been a team effort, and it's fun science and we're so appreciative of the opportunity to tell you about it.

Alicia Morgans: Wonderful. And I'm really appreciative of the opportunity and chance to learn from you and from your team with so much expertise. I am certain that you'll continue to work together and with all of your collaborators over time to help us really understand what's happening in the real world and understand this sort of real-world clinical effectiveness type work in the bladder cancer space. So thank you so much for your time today.

Ravi Parikh: Thank you.

Ronac Mamtani: Thank you.