Leslie Ballas: Hi, I'm Leslie Ballas. I'm a radiation oncologist at Cedars‑Sinai in Los Angeles, and I am so pleased to welcome Dr. Daniel Joyce, an assistant professor in the Department of Urology at Vanderbilt University to discuss his recent paper, Cost‑Effectiveness of Trimodal Therapy and Radical Cystectomy for Muscle‑Invasive Bladder Cancer.

Dr. Joyce, thank you for joining.

Daniel Joyce: Thanks so much for having me. Pleasure to be here.

Leslie Ballas: Why don't we have you present a few slides on the paper and its findings.

Daniel Joyce: Yeah, thanks. So the impetus or rationale for this study really came from this really nice retrospective multicenter study that was published in 2023. And just to remind those who may not know, in 2019 there was a systematic review that was done comparing radical cystectomy to trimodal therapy. And what they found was that actually radical cystectomy had improved cancer‑specific and overall survival compared to trimodal therapy. We are dependent on retrospective data to help us understand oncologic outcomes in this space. Some will remember nearly 20 years ago, we tried to accrue patients in the SPARE trial, which would've randomized patients to their trimodal therapy and radical cystectomy and it didn't accrue. And since the failure of that trial, I think the idea of having prospective randomized data to help guide us here is no longer feasible. And so we really are dependent on retrospective data.

This 2023 study that was published really turned everything on its head, brought a lot of things into question. It looked at a very specific population of patients. So these are patients who would've been good for either radical cystectomy or trimodal therapy. And specifically on average they were aged 71 years and they had clinical stage T2 to 4A, no nodal mets, no metastases, and they had a solitary tumor of less than seven centimeters with no or unilateral hydronephrosis. Adequate bladder function and a lack of multifocal or extensive CIS. And what they found using very sophisticated statistical modeling is that there was no difference in cancer‑specific or overall survival between these two treatment options.

Which led us to ask the question, well if this is true, if both of these treatments are equally good at treating cancer, how do they compare from a value standpoint? Should we be directing patients more towards trimodal therapy with the idea that most patients want to keep their bladder? And so the objective of our study was to compare the cost‑effectiveness of these treatments using a micro simulation model, and we wanted to account specifically for downstream costs and also toxicities from these treatments. And of course we did this from a US healthcare perspective.

I'll talk a little bit about what a micro simulation model is for those who may not be familiar. Basically what you're doing is you're taking a hypothetical patient and you're giving them one of two treatments, either trimodal therapy or radical cystectomy, and then you've created a statistical model that lets them progress through their hypothetical life experiencing all the things that might happen to a patient that has received that treatment. And when we say at each point along that timeframe, you are capturing things like costs both to the patient and to the healthcare system as well as quality of life. And quality of life, you want to account for both the quality of life but duration of that quality of life. And to do that, we use a term called quality‑adjusted life years, or QALYs.

Quality of life in these cost‑effectiveness analyses basically rates life on a value scale from zero to one, zero being dead, one being perfect health. You can then use that rating, so say it's 0.8, and if a patient experiences that rating of health for one year their QALY would be 0.8. So 0.8 of that quality‑adjusted life year.

We wanted to base our micro simulation model on the findings from this new data that we had, this new retrospective data. So specifically we wanted to use cancer outcomes that were equivalent between these treatments and we treated our model like the patients in that study. And what you get when you run this micro simulation model is an overall cost‑effectiveness outcome called an ICER, or incremental cost‑effectiveness ratio. All that is it's the difference in costs between the two treatments over the difference in effectiveness, which is the QALYs, the difference in QALYs.

And so you can see here at five years, and again, we use five years because the retrospective data have had follow‑up through five years and so this is the time period in which we have the most robust data, you can see the costs were significantly higher for trimodal therapy than radical cystectomy. However, trimodal therapy did provide more QALYs for patients or improve their quality of life compared to cystectomy.

Unfortunately, that improvement in quality of life did not mean it was cost‑effective. So you can see the ICER there of $464,000. We typically in cost‑effectiveness analysis research consider a willingness‑to‑pay threshold, so the amount our society is willing to pay to say that a treatment is worth it or cost‑effective, that's typically around $100,000. And so we're over four times that limit here at five years.

We did, as a sensitivity analysis, expand out our time horizon to 10 years of follow‑up, now understanding that 5‑ to 10‑year range of toxicities and costs we know less about and have less good data to support. And so we didn't want to say this was our absolute outcome due to the uncertainty. However, if you use what data we do have and expand it out to 10 years, you can see that despite the increase in QALYs that you're continuing to see with trimodal therapy it still did not make it cost‑effective.

These cost‑effectiveness acceptability curves that you see at the bottom are nice because they tell you kind of, well, what if we increase that willingness‑to‑pay threshold? What if we as society say, ah, let's say it's $200,000. Does that change anything? And you can see even when we expand up to $200,000, radical cystectomy is still the most cost‑effective option in the majority of these simulations that we're running.

That is where most people drop a cost‑effectiveness analysis paper, and I always caution people in doing that because that's actually the least interesting outcome of our paper and of any cost‑effectiveness analysis in my opinion. I think the real meat of these studies and the real value in them comes in the sensitivity analyses. And so we run a series of one‑way sensitivity analyses including each variable that goes into our model. And what you can do is as you vary one of those parameters you can see does that change anything? Does that all of a sudden flip our model and make trimodal therapy the more cost‑effective option?

And what you can get out of those sensitivity analyses is these diagrams called the tornado diagram and it looks more complicated than it is. It basically just lists the variables on the left and then they're ordered in the importance that they have on our model outcomes. So the wider bar you see here, the more likely that variable is to affect outcomes, and then you can see that as we vary that variable, either reducing it or increasing it, we can change the ICER or that final outcome of whether it's cost‑effective. And what you see here is not super surprising, but that cost really is the driver in this model. Now, there are some long‑term toxicities also that can impact the model, but really not enough to change the model outcome.

And indeed, when we looked at thresholds, so a point at which one of these variables flips the model to favor trimodal therapy, what we found was that if the initial cost of trimodal therapy would have to be reduced to about $17,000—remember, in our base case or our analysis it was around 40 grand. If you flip that and say, okay, what about cystectomy? Well, cystectomy would have to more than double, increasing to $48,000, and in our base case it was $19,000.

Additionally, interestingly, we found that the risk of metastasis after these treatments also had an impact. So if you could say that trimodal therapy has a better metastasis‑free survival by 11%, and remember in our base case we treated these as equal treatments, but if trimodal therapy really was better then it would also become the most cost‑effective option. So I think these are where some of the nuances of, okay, what is it that makes trimodal therapy not cost‑effective and how can we influence that to help improve value both for our healthcare system but also for patients as well.

Leslie Ballas: Thank you, Dr. Joyce. That was very informative. Can you tell me, how did you account for complications, perioperative complications in your model? I mean, how do you account for 3% mortality, for example, and the cost that has or the variation in hospital stay based on complications? A paper from 2019 from the Mount Sinai group showed that cost varies based on complications in the perioperative setting between $9,000 and $20,000 per patient. And so if you could address that for me.

Daniel Joyce: Yeah, so a lot of the probabilities of things like cystectomy death, receiving cystectomy after trimodal therapy—a lot of those probabilities came straight from the multicenter retrospective study, so we used the values that they reported.

As far as complications, this gets a little trickier, but we have actually very good perioperative complication data for radical cystectomy, and so we use that based on robotic‑assisted radical cystectomy data to inform those probabilities.

The long‑term toxicity data, which we can get into, is a little bit trickier and is definitely an area where we need more robust prospective evaluations of what those outcomes are. But again, a lot of this, the nice thing about cost‑effectiveness analysis is that you can understand the uncertainty of the variables that you're putting in by varying those one‑way sensitivity analyses. So if you have an outcome like, let's say, bowel complication after cystectomy, you can vary that probability anywhere from zero to 100%. We typically don't do that. We typically do it within a plausible value. But you could, if you wanted to, say everybody gets a bowel complication—how does that affect your outcome?

We can see that in our findings, especially those short‑term toxicities, which makes sense, they don't really have that much impact on overall treatment value. And the reason for that is that it's the long‑term quality‑of‑life gains that you get from a treatment and also long‑term costs that accrue over time that really start to impact things. So a lot of these short‑term toxicities, which might favor using cystectomy in a lot of these patients because it's a short‑term thing—yes, there are costs associated with it, but really it's the long‑term follow‑up that impacts these patients most.

Leslie Ballas: And I guess to sort of build on what you're saying and sort of a similar question—so then how do you account for the 17% of patients or so that would have renal failure after cystectomy and the long‑term costs of that?

Daniel Joyce: Great point. You could argue, what about the renal failure from chemotherapy also? We actually did not model that; we have limited long‑term toxicity data in this population. Of the long‑term toxicity data that we have, we lumped them together in either a urologic complication or a bowel complication, and then we ranked them as minor and major. Interestingly, we really wanted to do our due diligence and understand, man, if we combined all of these complications and figured out if they had every complication that is possible from both of these treatments, did that impact model outcomes? And it didn't. And I think it didn't because the costs are so different and are really the driver in what's making trimodal therapy not cost‑effective.

There are, of course, differences in these toxicities that are going to be really important to understand. We have no real prospective patient‑reported outcome measures that are of high quality for trimodal therapy. We certainly don't have them in a population like this of patients that are good candidates for both treatment options. So a real area of need that we need further work to figure out.

Leslie Ballas: The BC2001 trial did have a patient‑reported quality‑of‑life functional assessment of patients in the long term, which is a trimodal therapy study.

Daniel Joyce: Yes, I totally agree. Out to five years, I believe, they looked at quality of life. We chose to actually look at the MGH experience given that those same investigators were the ones that were involved in the multicenter retrospective study, which also had very robust quality‑of‑life assessments. I think the difference is that you don't have a matched cohort of patients for those quality‑of‑life assessments, so how can you compare similar patients receiving cystectomy to those with trimodal therapy in order to understand the quality‑of‑life trade‑offs?

It is also worth noting that a lot of these questionnaires do not translate specifically into utility values. So utility values, remember, are those zero‑to‑one assessments of quality of life that can be used to create QALYs. It's a little bit tricky in the statistical modeling. It's also difficult to validate cross‑talking or mapping between those questionnaires. There are some quality‑of‑life instruments where it has been done and validated, and so sometimes we can do it but not always. And so we're often dependent on proxies of utility values based on what we're seeing from probabilities of developing a toxicity from these treatments.

Leslie Ballas: Yeah, that's the thing that always is confusing to me when I see these QALYs reported and they don't seem—I mean they're not based on patient‑reported quality‑of‑life outcomes. As you described it, it's a scale based on complications and death. Yet it's not really taking into account what patients are valuing.

Daniel Joyce: Great question. So this is a little bit of the nitty‑gritty of cost‑effectiveness analysis, but it's helpful to know what a utility value is and how we get that. There are a number of different methods to get to that utility value, like a time‑trade‑off or standard‑gamble approach. There are also questionnaires that you can use to create utility values that are quality‑of‑life questionnaires. So the EQ‑5D‑5L is one of these instruments where they ask you to rate your quality of life in five different domains. How you answer that question then gets interpreted from a societal perspective, and so you have the patient who's experiencing that treatment—so you'll say, I'm going to survey this patient who had trimodal therapy longitudinally through five years.

They have their assessment and there's also a visual analog scale to that assessment. That visual analog scale is kind of like how they feel about their life right then. The other five‑domain questionnaire is kind of how they're feeling about different domains of quality of life, and then you get health states based on those answers. So there's a number of different possible health states you can get based on how you answer the questions, each of those five‑domain questions. They then use a cohort in the US that does not have that disease, does not experience that treatment, and they rank their life—how they view life—and they again use different methodologies in how to do that.

But what they do is create a mapping system between that EQ‑5D answer from the specific patient and the society. It's important to know the difference in the visual analog scale and the societal preference. So there are patient preferences for their quality of life, and then there are the society's preferences for that quality of life. The society's preferences for quality of life are typically what we're using in cost‑effectiveness analysis. And the reason for that is that we can have a metric that's really easy to compare across disease spaces.

And so what we often rely on is that metric that's taken from a similar disease state, and we use that—now the probabilities of experiencing that health state come from patient‑specific data that are from those patients who are receiving that treatment, if that makes sense.

Leslie Ballas: Yeah. And I get that. It's just hard when to translate this into talking to that patient that's in front of you and saying, hey, society puts this price on saving your bladder. Because to the patient, that might be the only thing that matters.

Daniel Joyce: I'm so glad you brought that up and it is a really, really important distinction to make. A cost‑effectiveness analysis is not used to help a patient make a decision. It can be. It can be. So if you look at quality‑of‑life trade‑offs, so those QALYs specifically, that is very important for the patient, and you could use that—you could use that in our study—and it would not say to give them cystectomy; it would say trimodal therapy is probably better for you because you see quality‑adjusted life years are greater in trimodal therapy.

So you have to be really careful about how you're interpreting these studies. This is not a shared decision‑making tool in and of itself. It can be used to inform shared decision‑making based on these prolonged outcomes that we can kind of project for patients what their life's going to look like downstream. And that's why I say that cost‑effectiveness ICER, that overall outcome, is often not that helpful because of that. Because just because something's not cost‑effective doesn't mean we shouldn't be using it.

Other countries feel differently; they will often use that ICER to guide funding reimbursement for specific management strategies. Here in the U.S., we have a lot of ethical implications with that. It doesn't really jibe with our overall feeling towards healthcare, and nor do I think it should. What I think it does do, though, is allow us as clinicians to see something's really increasing cost here. Does it justify that incremental increase in effectiveness or quality‑adjusted life? And if it does, then what can we do to make it better? What can we do to decrease those costs? If it doesn't, then we do really need to say, maybe this isn't a treatment that we're worth exploring.

A great example of this is EV‑Pembro in metastatic urothelial carcinoma. That is not cost‑effective, I can tell you right now. I haven't done the study, but I guarantee it's not cost‑effective, but there is no one on the planet who would not offer that to their patients. And I feel similarly about this study. Trimodal therapy is something absolutely that our patients want, that we need to be offering to our patients. I think it's worth knowing, though, that it is driving up the cost of the most expensive cancer to treat over a patient's lifetime, and so if we can do things to modify that cost to bring it down to therefore increase the availability and access for our patients, I think that's worthwhile.

Leslie Ballas: I think that's a great place to end. That's really helpful information and I really enjoyed getting to hear about this paper and understanding some of the nuances. So thank you again, Dr. Joyce, for your time and for sharing your expertise with us.

Daniel Joyce: It's my pleasure. Thanks so much.