The Uromigos Episode 131: The NORSE Study—Erdafitinib vs Erdafitinib/Cetrelimab

Dr. Jonathan Rosenberg gives an overview.

Episode Transcript

Tom:
Welcome everybody to one of our ESMO podcasts. We’re joined by Jonathan Rosenberg. Jonathan, I’m going to ask you to introduce yourself. I’m going to talk about an abstract with an FGFR inhibitor in combination with a PD-1 inhibitor. The data I’m presenting makes it slightly complicated.

So, I guess you’re going to introduce yourself then maybe have to ask me questions, but we’ll just see how we go. So, Jonathan, why don’t you fire away and then I’ll dive in. If you’ve got any major conflicts of interest in this area, which you probably do, just let us know.

Jonathan Rosenberg:
Yeah. So, thank you. I’m Jonathan Rosenberg. I’m a medical oncologist focusing on urothelial cancer at Memorial Sloan Kettering in New York. And I’m the chief of genitourinary oncology there. I’ve consulted for multiple companies that are involved in FGFR inhibitor development and have led a study very similar to what Tom was going to talk about using a different agent, rogaratinib, as opposed to erdafitinib, which is what we’ll be talking about primarily today.

Tom:
And Jonathan, you combine with atezo. Is that correct?

Jonathan Rosenberg:
Correct. So that was the phase two study looking at atezolizumab and rogaratinib in cisplatin-ineligible patients using a different biomarker strategy. So erdafitinib is generally employed in patients with mutations or fusions, whereas rogaratinib was being developed in patients with overexpression by mRNA detection, as opposed to mutation.

And that captured almost all the mutation patients, but also included patients who had high levels of overexpression which was thought perhaps to be driving the benefit in a larger group of people to target FGFR pathways. So that study was atezo.

Tom:
And Jonathan, you had about 25, this is really annoying, because you completely stole my thunder, but I’m going to keep going anyway.

Jonathan Rosenberg:
I love it.

Tom:
There was about 25 patients and you had response rate of 58%. Do you just want to talk a little bit about the immune infiltration in that study and whether or not you feel that individuals with FGF, and we’re talking about FGF2 and FGF3, is that correct? With your study? How did you pick your patients?

Jonathan Rosenberg:
Yeah, so this was actually just FGFR3 …

Tom:
Gene expression.

Jonathan Rosenberg:
… and one, but mostly three, essentially. FGFR1 and three overexpression in that trial. But functionally, it was almost all FGFR3 because that’s the primary FGFR overexpression modality in bladder cancer.

Tom:
And Jonathan, the relationship between FGF and immune parameters in your study.

Jonathan Rosenberg:
Yeah. There’s not a lot of data that was actually garnered in that trial yet. And what was interesting in that trial, I think was the anti-correlation with response with PI 3-kinase pathway mutations in bladder cancer, which are fairly prevalent. So, if you had one of those alterations, you seemed to be much less likely to respond.

Tom:
We did a study called BISCAY. In BISCAY we enrolled patients whose cancer had progressed after platinum-based chemotherapy. And they had, using foundation analysis and FGF2 or three alteration, we showed response rates for single agent 4547, which is AstraZeneca’s FGF inhibitor, but 30%, which actually is not unreasonable.

Remember erdafitinib is 40% as a single agent, but we then combined it with durvalumab, and we didn’t show a bounce in response rate.

Why do you think there’s a discrepancy between what you showed in your trial, Jonathan, and then the no bounce in BISCAY? By the way, in the BISCAY trial the numbers were small again. It was about 20 in both arms. And in that study, we show low PD-L1 expression, and we showed a lack of immune infiltration in the patients with FGF alterations compared to the remaining population.

Jonathan Rosenberg:
I think the big difference between the trials beyond the biomarker selection strategy was the line of therapy, untreated patients versus prior platinum. And I don’t know how much that might’ve affected this or perhaps selected out different populations despite having the mutation. But we know that as disease becomes more resistant, in general the response rates tend to go down.

And I do wonder whether the tumor microenvironment was changed somewhere along the way. It is an interesting discrepancy, but I suspect line of therapy and the type of patients who went in may have something to do with this.

Brian:
And pretty small numbers, right, Tom? You said only 20 patients.

Tom:
Yeah. But it was … Because I’m going to present this NORSE data right now, Brian. I’m going to do it right now.

Brian:
I’m waiting with bated breath.

Tom:
So NORSE is a randomized phase two trial of erdafitinib or erdafitinib plus cetrelimab, which is a PD-1 inhibitor. And this is for patients with metastatic or locally advanced urothelial cancer and FGFR alterations.

And it’s the interim results of the study, essentially, we’ve combined the two drugs together and we’ve compared it in a one-to-one randomization versus erdafitinib alone in first-line metastatic urothelial cancer patients who are ineligible for cisplatin-based therapy and have FGFR alterations.

It was a planned interim analysis essentially. And we ended up with about 20 patients for efficacy evaluation and 25 patients for safety evaluation in both arms. The characteristics of the patients were relatively well balanced. Most patients had FGFR mutations rather than fusions. And the response rate was 68% in the combination arm and 33% in the single agent arm.

And that’s, as I said before, about 20 patients in each arm. 21% CR rate, that’s only four patients, but it’s still 21% in the combination arm. And in the combination arm, 90% of patients got disease control. So, progression of disease only in 10% of patients.

Brian:
So, Tom, do you think this is somehow [crosstalk 00:07:03]

Tom:
… interrupted.

Brian:
… mechanistically different than the other combos? Or is it just small numbers and patient selection and the like? What’s different to account for these results?

Tom:
Well, I think it’s different from BISCAY, but I don’t think it’s different from Jonathan’s study, which is one of the reasons, we’ll talk with Jonathan all about it, it was one of the reasons I was really interested in his opinion on this.

Because when we started on this journey, we’ve shown that combining with chemotherapy, for example, hasn’t proven to be spectacularly successful. And the BISCAY trial tried four or five different combinations with targeted therapies, and none of them were particularly successful.

And so, when Jonathan study came out, because you’d expect, erdafitinib is a single agent, you’d probably expect somewhere between 30 and 40% response rates, in my opinion, in the frontline setting, because it’s quite difficult to get in control of the disease. Although, this is a selective population because of FGF. And I actually think this 68% is higher than the 30% we saw with a single agent.

And so, under those circumstances, I do seem to think that actually we’ve done now a randomized trial that shows some addition of the immune drug on top of the single agent targeted drug. And Jonathan’s study also, a 58% response rate does seem quite high.

The intriguing thing about this sort of figure is really chemotherapy has got about a 45% response rate. So, it does seem higher than chemotherapy. And so, if you were to compare yourself to chemotherapy, it would be reasonable to think that actually you’d have a higher response rate. And we know the responses with immune therapy are likely to be more durable than that we see with chemotherapy alone.

The follow-up of this trial doesn’t give us that information, but the spider plot does show some durability of those responses. Not complete, but it’s not a V-shaped curve, which is often the case of what we see with chemotherapy. And indeed, with erdafitinib.

Brian:
What’s the plan for further … Is it fully accrued now, or what’s the plan?

Tom:
No, this trial is going to be ongoing. But the issue with this trial that’s complicated is we’ve currently got 20 in each arm. We talk about the plan being for 45 in each arm to complete. But the challenge with that is 45 patients in each arm is unlikely to get FDA or EMA approval.

So, we’ll be in a position where we have more robust data, which is terrific, but is it practice changing data? And we don’t know. So, do we need to launch a randomized phase three? And that’s the question which I think that needs to be answered.

Jonathan Rosenberg:
I think the data is supportive enough based on these two phase two studies, different drugs, different biomarker selection strategies, but 58, 68% is not sort of what we’d expect. And it’s in the range of what we’ve seen with EV and pembro, which has ignited a firestorm of excitement over that combination. So, I actually feel like the strategy is probably worthwhile pursuing in the first-line setting for these patients who can’t get cisplatin at this point.

Brian:
Question for both of you. Because you’ve talked about different selection strategies for FGF. Is there an accepted standard? Is there one better than the other? Is one more practical if this does, say move forward and get used in the community?

Tom:
Well, I’m confident mine’s better than Jonathan’s.

Brian:
I highly doubt that, but yeah.

Tom:
Jonathan, what do you think?

Jonathan Rosenberg:
Well, I think that the mutation status was clearly a more accepted as a way of testing for FGFR3 activation. But I do think there’s some merit to the overexpression idea because there are tumors that are probably addicted to FGFR without mutations.

There is transcriptional upregulation that leads to constitutive signaling in some of these tumors and the results in using the different biomarker strategy actually allow you to treat a lot more patients than if you’re looking for mutations alone.

The mutation frequency is what, 20% at best? Maybe a little higher. Whereas if you’re looking at overexpression, based on RNA scope, which was the biomarker that was used with rogaratinib, you’re looking at 40% of patients. So, you may be doubling that.

And if you think they derive similar benefits, but you can’t detect them with mutation, then you’re missing a group of patients who might benefit from this approach. So, I don’t know that it’s better, but it casts a wider net.

Tom:
Jonathan, in BISCAY we looked at using [inaudible 00:11:54]. We looked at circulating ctDNA. We looked at circulating FGFR alterations, which had a really strong correlation between tumor and circulating. I think that’s point number one.

And then point number two is we also showed RNA analysis correlating with mutation expression. My inclination at the moment is that if this program is going to be successful, we probably ought to think about moving towards a circulating type of biomarker. What’s your feeling on that?

Jonathan Rosenberg:
We’re seeing in some preliminary data that the frequency of FGFR alterations in ctDNA seems to be higher than what we might expect, and whether or not these are sub clonal alterations that aren’t drivers for the majority of disease has me somewhat concerned.

But I think it’s a very useful biomarker to identify patients. But remember that almost all the trials that have been done with this had been looking at the tumor DNA. And we really need to look at a ctDNA in the same manner, because it’s really a different biomarker.

Tom:
The adverse event profile grade three or four adverse events with monotherapy was 38%. And with the combination was 50%. The commonest adverse events, hyperphosphatemia, stomatitis, diarrhea, nail abnormalities, dry mouth. What’s your feeling on the safety profile of the single agent and the combination compared to chemotherapy? Can patients stay on these drugs for months and months on end?

Jonathan Rosenberg:
I’ve found the FGFR inhibitors much more toxic than I might’ve expected, but it’s not the kind of toxicity that gets you in the hospital. It’s the kind of toxicity that just makes you kind of miserable. And for patients, that’s a real issue. So, I think they all have issues.

These are non-selective FGFR inhibitors. I think the next generation of FGFR inhibitors that are FGFR3 selective might have certain advantages if they can be developed and actually have, say pharmacologic profiles and might be a better tolerated drug.

But patients don’t like their fingernails falling off. They don’t like stomatitis, and these are often and hard things to manage. But I think the safety from an overall toxicity profile of the single agent drug, in terms of the things we associate with chemotherapy and other oncologic therapies is actually good. But I don’t know that patient quality of life is as good as we’d like it to be.

Brian:
I have a big picture philosophical question. So, we talked about chemotherapy-free regimens. FGF IO, erdafitinib IO. Do you think in three to five years when we’re doing this, that any chemotherapy-free regimen will be in use in the frontline for metastatic urothelial cancer?

And parenthetically, when IO first came in, it was like, well, we can just use IO upfront. And then we sort of found out that yeah, chemotherapy is important, and maintenance is probably better. So where do you think we’re headed in that regard?

Jonathan Rosenberg:
I think we need something that is cytotoxic in the first-line setting, and whether it’s traditional chemotherapy or an antibody drug conjugate, or even an FGFR inhibitor as these are cytotoxic for those patients who are sensitive. Those drugs, I don’t think are going to be missing from the first-line setting.

It isn’t necessarily going to be platinum-based chemotherapy, but I do think that this was a rapidly progressive disease for many patients. And without something that can cytoreduce people, you’re going just continue to see failures in the first-line setting. Tom, what do you think?

Tom:
Yeah. I think if chemotherapy were invented tomorrow and Larry [Einhorn 00:15:51] who gave our podcast, who was in his late twenties or early thirties invented cisplatin tomorrow, I think there’d be a lot of excitement in urothelial cancer right now about getting that initial control.

There’s no doubt, Jonathan, that you’re absolutely right. Getting that initial control is very important. Because you don’t get two good goes at urothelial cancer in my experience, the best way to do it is to get the frontline therapy right. And chemotherapy is really good at getting that control.

Now, EV plus pembro may be a little bit better than that, but I suspect if patients were to progress or their cancers were to progress on that, I suspect from there chemotherapy will have a role subsequent to that. I think the disappointment is that we haven’t been able to combine chemotherapy successfully with immune therapy.

Brian:
I was going to ask, is the cytotoxic agent … Can these agents replace chemo, or would chemo still have a role? Are we sort of moving towards triplets in the frontline like we are in RCC? Or is that just going to be too toxic?

Jonathan Rosenberg:
I don’t think we’re going to see triplets with novel agents and chemotherapy in the first-line setting at the moment. I wonder whether we’ve got the wrong chemo backbone, but I don’t know that …

Tom:
I agree with that.

Jonathan Rosenberg:
… a second bite at the apple here to try to figure that out on a large scale. And so, I don’t know that there’s anything inherently different about bladder cancer than lung cancer in many ways, except the fact that they’re using pemetrexed carboplatin while we picked gemcitabine and cisplatin as the backbone of our chemotherapy.

So it may be that we’re damaging the tumor immune microenvironment with the cytotoxics. And so, I’m cautiously optimistic that there may be a way to do that, but I don’t know that we’re going to get there given the current landscape.

Tom:
Brian, my feeling on this combination is that it’s relatively easy to give, not without toxicity exactly as Jonathan has said. But what I have noticed is it’s really good at getting control of disease. And so, it feels a bit like, you know when you start, and you both will know this, when you start immune therapy in urothelial cancer, in some patients, when it’s by itself, some patients just grow and there’s not much you can do about it. You can’t really predict who they’re going to be, and sometimes it goes wrong.

When you give this combination, erda and cetrelimab, almost all patients you get quite good control of disease. And so, it feels a bit like chemotherapy, and I feel more reassured by this. And I think this would be gem carbo upfront. That doesn’t mean gem carbo wouldn’t have a role to play potentially in the future, as I’m sure EV would have a role to play plus or minus SG and other combinations.

But my feeling is that this combination is active in urothelial cancer. We don’t have that many combinations or active drugs in urothelial cancer. So, I think it’d be disappointing if we didn’t continue to pursue this because either it’s not common enough for a biomarker perspective, or we’re not confident enough to be able to complete a randomized trial.

Remember these trials have been running for some period of time and we’ve only got 40 patients in each arm. And so, we have to take part, we have to recruit patients into these trials because otherwise we might end up letting this target go completely.

Brian:
So, Tom, I was going to ask, if EV pembro leapt to a phase three, why not this combination? Let’s put accrual aside. I know it’s important, but it has similar response rates. It’s chemotherapy-free, seems to be well tolerated, et cetera.

Tom:
Yeah. And you go against all-comers. I wouldn’t confine myself to the carboplatin population. I don’t think that distinction has been particularly helpful. It looks like immune therapy is more active in the cisplatin-eligible than the cisplatin-ineligible population, as you’d expect. The problem with the carboplatin population is many of these patients have got aggressive disease, performance status of two because of the disease, and liver and bone metastasis, and nothing works very well on that population.

The cisplatin-eligible population, the young man or woman aged 45 with a few lymph nodes relapsed after a cystectomy, that patient does better with immune therapy than the carboplatin patient with liver and bone metastasis. So, this carboplatin distinction has been really unhelpful for drug development in urothelial cancer. We need to move away from it.

Brian:
So where does this go next, Tom, then I guess? Finishing accrual, 20 more patients. You said it’s at the 20-ish mark going to 45, but then what after that?

Tom:
I think it’s a difficult decision needs to be made about whether we launch a rival randomized phase three study. And then the decision needs to be, is that a neoadjuvant study? Is it an adjuvant trial? Or is it a frontline metastatic trial? I don’t know what you think, Jonathan.

Jonathan Rosenberg:
I think moving into the peri-operative space is a lot more complicated and I would probably favor the metastatic setting at this point. The other issue with accrual to an FGFR inhibitor trial is the fact that you’re only really able to accrue 20% of patients if you’re pursuing a mutation strategy. And so, these are more challenging studies, unfortunately just based on the biomarker selection, however they’re doable and they need to be done.

And I agree that I wouldn’t stray too far from the data you have. And I would probably do a first-line trial if this were my decision to make comparing to chemotherapy. And I agree I’d be agnostic as to platinum eligibility. And we need to take these concepts forward, whether it’s erdafitinib or one of the newer agents, or heaven forbid, rogaratinib will never get resurrected.

These are active drugs in the disease. And the combinations, well, I don’t think we’ve proven that FGFR is causative for this sort of immune desert phenotype. I do think that clinical data is so compelling that we really … It’s incumbent upon us to really push these trials forward.

Tom:
Jonathan, it’s been fantastic. Brian, any questions?

Brian:
No. Was there anything else FGF wise at ESMO? Any other complimentary data or was this sort of the standout for FGF?

Tom:
I think this is it at the moment, Brian.

Brian:
Congratulations, Tom.

Jonathan Rosenberg:
Great work.

Brian:
Jonathan, thanks for joining us. Always a pleasure.