Personalized Therapy with Erdafitinib: A Biomarker-Targeted Therapy for Urothelial Cancer - Arlene Siefker-Radtke

August 26, 2020

Arlene Siefker-Radtke joins Tom Keane to discuss personalized therapy for the treatment of metastatic urothelial cancer with erdafitinib. Dr. Siefker-Radtker outlines the importance of understanding the underlying biologies and mutation patterns that encompass the multiple diseases that makeup bladder cancer. When taking this into consideration, she gives an overview of the tolerance and safety profile of erdafitinib in clinical trials, as well as discusses its limitations as a preferred agent in treatment. 


Arlene O. Siefker-Radtke, MD, Professor of Genitourinary Medical Oncology, Division of Cancer Medicine, The University of Texas MD Anderson Cancer Center

Thomas E. Keane, MBBCh, FRCSI, FACS,
Professor and Chairman of Department of Urology at the Medical University of South Carolina

Read the Full Video Transcript

Tom Keene: Hello everybody. This is Tom Keene coming to you from UroToday. Today, we have a real treat. We have Professor Seifker-Radtke who is coming to us from the division of cancer medicine at the University of Texas at MD Anderson Cancer Center where she is a Professor of Genitourinary Medical Oncology.

Arlene Siefker-Radtke: Well, thank you, Tom. It's a real pleasure here today to present on personalized therapy with erdafitinib, the first biomarker targeted therapy for the treatment of metastatic urothelial cancer. Bladder cancer is no longer just one disease. Bladder cancer is actually composed of multiple diseases which differ in their underlying biology and also may differ in their mutation patterns, which can play a role in this biology.

In particular, FGFR3 mutations do appear to play a role in the development and biology of bladder tumors. For quite a long time, in fact, more than 15 years, there's been a dual track concept of how bladder cancers develop. The majority of tumors develop along this papillary pathway as seen in the figure on the left, where the lining of the bladder develops hyperplasia and then forms these polyploid or stalk-like lesions.

These are often very low grade and low stage tumors. And it occurs in about 80% of bladder cancers that develop in patients. But over time, additional instability occurs and a fraction of those patients transform into the invasive tumor that kills.

There is a group of patients who de novo have bladder tumors that arise with dysplasia and carcinoma and situ early in the pathway. We typically see early RB and p53 mutations and that group of patients has a very high frequency of transforming into an invasive disease which can kill or take a person's life.

Intriguingly, FGFR3 mutations appear to follow the similar pathway, where we see this hyperplastic and then low grade, low stage lesions having the highest frequency of FGFR3 receptor alterations, occurring and over 60% of tumor specimens.

A small fraction of those tumors may transform with additional mutations including p53, deletions of chromosome nine, and become the invasive tumor that may take a person's life. When we look at the flat dysplastic or carcinoma in situ pathway, it does appear that there's a very low frequency of FGF receptor 3 alterations in that group of patients.

Using gene expression profiling, we have learned some insights into the different types of bladder cancer. And this is the same gene profiling that's been done in breast cancer and is akin to the breast cancer TCGA data where we see both luminal and basal tumors.

Some of the earliest data, both from the TCGA, MD Anderson, Lund groups and UNC suggested that bladder tumors differentiate into two basic clusters, a basal cluster, and a luminal cluster.

Also using gene expression profiling, we found that we could differentiate tumors that not only differ in their prognosis but may also predict one's benefit from treatment.

This is some early data from Choi et al, published in Cancer Cell. And when we looked at the subtypes of bladder cancer, the basal subtypes shown in the survival figure with the red line had very poor overall clinical outcomes overall in the setting of muscle invasive disease.

And each of these sub-types had different factors associated with them. With the basal subtype having high proliferation and high levels of stemness and the worst clinical outcomes. And intriguingly, the FGFR3 mutations appeared to occur in this luminal or as we now call it luminal one or luminal papillary subtype.

This is some early data from MD Anderson, where we looked at patients treated with neoadjuvant chemotherapy. And as we can see in the figure on the right, these very poor clinical outcomes when treated with cisplatin-based chemotherapy with the survival figure on the left, we can see that basal subtype outlined in red now has the best clinical outcomes in the setting of neoadjuvant cisplatin-based chemotherapy.

And this led us to suggest back in 2015 that perhaps using these different signatures and different types of bladder cancer, we could start to predict which patients would benefit from which therapies. With the basal signature appearing chemo-sensitive, having an immune signature associated with that group and then in the luminal subtype we saw strong FGFR3 signature suggesting this may be the group of patients who benefit most from an FGF targeting strategy.

Around the time that we were developing FGFR3 inhibitors, I happened to speak to my colleague, Dr. David McConkey, because I was noticing that patients with FGF receptor 3 alterations did not appear to be responding well to immune checkpoint inhibition. When he went back and looked at the different subtypes, it does appear that this luminal papillary or luminal 1 figure shown on the left appears very immunologically cold.

Whereas, the different subtypes, including a basal subtype appeared very hot and there was a group with intermediate markers that we wondered, could they have the just right amount of immune cells needed to achieve an immune response? When we looked at the FGF receptor 3 amplification, mutations and fusions, we see high enrichments in this luminal 1 or luminal papillary signature.

So when we look at the origins and transcriptional control of basal versus luminal bladder cancer, the presence of these mutations and alterations could potentially predict where the tumor is rising from the urothelial lining. It does appear that cells arising more from the umbrella layer have more PPAR-gamma and FGFR3 expression, which appears more consistent with the luminal subtype.

Whereas, the basal group of tumors shares a lot of factors associated with the STEM cell layer. So it's becoming increasingly apparent that bladder cancer is complex, with multiple tumor types, it's no longer just one disease. Since then, there's been additional data here from Seiler et al, suggesting that patients again with that basal subtype shown in red derive the most clinical benefit from a neoadjuvant cisplatin-based chemotherapy.

And the increase in complexity is seen by the further inclusion of additional signatures that might begin to predict which patients respond not only to FGF targeting agents, but possibly ERBB2 signaling in the luminal pathway, stromal factors in this luminal infiltrated group of tumors, and different signatures in basal as well.

And more recently, we've even seen the inclusion of a neuroendocrine signature. So there's clearly multiple different types of bladder cancer. And as we begin to understand the different flavors of tumor, we may be able to predict the best treatments for a particular patient.

The early evidence did suggest that FGFR was a target or at least a potential target in this immunologically cooled tumor. We saw the first results presented at ASCO in 2018 of a phase two study of erdafitinib in patients with metastatic surgically unresectable bladder cancer and FGF receptor alterations. This data was recently published in the New England Journal of Medicine in 2019.

When we look at the historical data with advanced urothelial cancer, second-line treatment resulted in very poor clinical outcomes with response rates of around 10% and a median overall survival less than a year. With immune checkpoint inhibitors, there's clearly improved outcomes and benefit. However, with an objective response rate of around 20% and an overall survival of approximately 10 months, it's clear that not all patients benefit from this treatment.

And with the idea that there may be subtypes that might benefit more or less from immunotherapy, it could be an unmet need to achieve targeting in a particular immunologically cooled phenotype. Erdafitinib is an oral pan FGF 1 through 4 inhibitor with inhibitory concentrations in the single-digit nanomolar range.

There was evidence of early clinical activity and metastatic unresectable urothelial cancer, and in other histologies including cholangiocarcinoma in the setting of an FGF receptor alteration.

The clinical trial was initially designed as a randomized trial. At the time this therapy was being developed, we hadn't yet had an FGFR targeted strategy that could be given in a continuous fashion. Dovitinib and BGG398 were both given on an intermittent schedule due to the toxicity observed with treatment.

Therefore, this study randomized patients to a 10-milligram daily dose, seven days on, seven days off, and then a continuous regimen of six milligrams a day. After enrolling a group of patients who were randomized to each of these, we then looked at toxicity and potential activity to choose a regimen for the Phase II trial.

As the trial proceeded, it appeared that the continuous dosing regimen appeared to have good toxicity and a good efficacy profile. And we also found with additional modeling that we could increase the dose to eight milligrams a day with uptitration to nine milligrams a day based on target phosphorus levels at day 15.

This regimen three was selected for the Phase II clinical trial with 99 patients enrolled. The primary endpoint was looking for an objective response in these FGFR3 altered tumors. Looking at the baseline characteristics, we do see a very high percent, nearly half of patients having more than two prior lines of treatment.

As expected in this patient cohort who've received previous treatment, there was a high rate of visceral metastases in 79% of patients, and 53% of patients had a creatinine clearance less than 60mls per minute. So clearly poor prognostic factors that have typically been associated with core overall prognosis.

The objective response rate on this trial did meet its primary endpoint with an investigative reported response rate that was confirmed with repeat imaging of 40%. 76% of patients overall had at least some reduction in the sum of the target lesions of their tumors.

As treatment was given, the responses to erdafitinib were maintained. And at the time of the data cutoff, we can see the longest patient who is still ongoing treatment of around 20 months. And I still have a patient who continues on erdafitinib today with continued control of their disease.

Looking at this forest plot, we tried to determine if there were factors associated with worse or better clinical outcomes based on response. And there were some trends to having a poor performance status doing worse, but again, very small numbers of patients. So the confidence intervals overlapped.

But I do think it is important to note that there was no difference in response based upon the level of kidney function. No significant difference based on the number of lines of prior therapy. And although it's not included here, there was also no significant difference based upon response in the setting of visceral versus non-visceral metastasis.

This is a patient whose outcomes I shared at the time of the meeting. This patient, when they enrolled, had very rapidly progressive liver metastases that were causing clinical symptoms. In fact, he told me that if he did not feel better in two weeks, he was not going to return for another clinic visit since he realized he was dying of cancer and he had young children that he needed to spend time with.

I told him, that was fine. If it's not improving, if you're not feeling better in two weeks, then it would be clear to me that the treatment wasn't working. Sure enough, within two weeks he felt better, returned to his clinic visits, and as you can see with these restaging scans, he ultimately had an 82% reduction in his very numerous measurable liver metastases.

The progression-free survival of patients on the trial was 5.5 months which appeared favorable compared to historical data of second-line strategies and the median survival of 13.8 months also appeared favorable compared to other treatments at this time interval.

As we can see here, there are treatment-related effects, but very few patients actually discontinued due to a treatment-related adverse event. We did see a high frequency of hyperphosphatemia but keep in mind, we actually increase the dose for uptitrated patients trying to aim for a degree of hyperphosphatemia as a potential surrogate showing suppression of FGFR3. We did see evidence of mucositis in patients, diarrhea, GI effects, in addition to hand-foot syndrome.

There were some treatment-related adverse effects of clinical importance that are shown here. Not only the hyperphosphatemia and skin and nail events, but there is also central serous retinopathy, which is a known class effect of inhibitors of the MAPK pathway. As a result, patients were actively monitored for this with baseline screening via an ophthalmology exam, and then they underwent Amsler grid testing at every clinic visit, and if there were any abnormalities noted, treatment was held and the patient was referred to ophthalmology for additional imaging looking for central serous retinopathy.

We did see evidence of grade three retinopathy in about 3% of patients. However, this rarely led to discontinuation. And in most patients, we held the treatment and they were able to resume therapy.

So in conclusion, this trial did meet its primary endpoint with a 40% objective response rate and the progression-free survival and median overall survival of 13.8 months does appear favorable compared to other therapies that have been reported to date.

It did appear that erdafitinib was well tolerated with a safety profile that allowed continuous dosing and uptitration based on serum phosphate levels and as a result, not only was erdafitinib granted breakthrough therapy designation, but in April of 2019, it was granted accelerated approval in the second line treatment of patients bearing FGFR3 receptor alterations.

There's additional studies ongoing with a Phase III trial of erdafitinib versus chemotherapy or pembrolizumab and a Phase IB/II trial combining erdafitinib with an immune checkpoint inhibitor.

There has currently been a lot of debate in the field, whether a mutant or altered FGFR3 urothelial cancer would have greater benefit from an immune checkpoint inhibitor or from treatment with an FGF targeted therapy. This is again based on the finding that FGF receptor 3 amplifications, mutations and fusions appear enriched in this luminal 1 or luminal papillary tumor that may have a lower response rate to an immune checkpoint inhibitor.

We retrospectively looked at patients enrolled in the clinical trial of erdafitinib and checked the investigator reported response rate to prior immune checkpoint inhibitor. 22 patients enrolled in the Phase II dose had received prior immunotherapy and there was only one investigator response reported to the prior immunotherapy. This response was not durable with the time to next treatment of approximately 10 months following their partial response.

When we compare that to the objective response rate of erdafitinib in this cohort which was 59%, it does suggest that potential that an FGF altered tumor may benefit more, at least based on response compared to a patient receiving immunotherapy in that setting.

However, there is a limitation. This is based on small numbers, retrospective review and it's also possible that excellent responders to immunotherapy may not have progressed at the time the clinical trial was ongoing.

There has recently been alternative data presented from Galsky et al, suggesting FGFR3 mutations may respond similarly to an immune checkpoint inhibitor. This is based on data from the IMvigor210 or second-line Atezolizumab and the Checkmate 275 trial of nivolumab. They went back to look at these tumors, and in the Atezolizumab or IMvigor210 cohort, 49 patients had an FGFR3 receptor alteration, and in the Checkmate 275, 15 patients had an FGF receptor 3 alteration. They reported a response rate of around 21% in this cohort, which is as similar to the response rate to immunotherapy in general across the population of bladder tumors.

The limitations, again include the small numbers, the potential enrichment for PD-L1 positive cohorts on this trial, and there was no information provided on durability of response or overall survival and the potential inclusion of other mutations that may be driving the tumor. There were also FGF receptor alterations were determined using a different strategy compared to patients involved in the trial of erdafitinib, and they may include FGFR3 mutations of uncertain significance since not all were definitively known as driver or hotspot mutations.

So clearly there's still room for debate and we need to explore further whether an FGFR patient responds better to immunotherapy or erdafitinib. This question is actually being tested in the clinical trial 4, where patients who have failed frontline chemotherapy for metastatic unresectable urothelial cancer and have an FGFR3 alteration present will be randomized to pembrolizumab or erdafitinib providing the data that we need to determine if these FGFR3 altered tumors benefit a bit more from immunotherapy or an FGF targeted therapy.

Regardless, FGF receptors are a clear target in the treatment of urothelial cancer. The presence of the mutation occurs in about 15% to 20% of metastatic bladder tumors and an even higher frequency in patients with upper tract tumors and earlier stage disease. It is a new class of agent with an alternate toxicity profile compared to immunotherapy, and I think may meets an unmet need in the treatment of our bladder patients.

Another important question being discussed in the community is whether FGF receptor three inhibitors can potentiate the effects of an immune checkpoint inhibitor. If these tumors really are immunologically cold, by targeting them, can we reawaken the tumor and make it more susceptible or more likely to benefit from an immune targeting strategy?

We recently presented some early evidence based on clinical samples from patients treated with an FGF receptor 3 targeting antibody, including patients with wild-type urothelial cancer. We did see responses that appeared enriched in the luminal papillary subtype, and in addition, when we looked at the responders and compared the post-treatment biopsies to the pretreatment biopsies, we saw evidence of increased expression of genes associated with an inflammatory response.

These were biopsies taken post-treatment with Vofatamab and prior to receiving pembrolizumab. So it does appear that there is that potential where treatment may alter the immune signature of these tumors. And the hope is that we could create a more favorable immune signature that could potentiate the response to immunotherapy. There has been early evidence suggesting we can shift subtype as well with chemotherapy. And this is some early data that we had presented and published in European Urology looking at patients treated with neoadjuvant chemotherapy. When we compared the subtypes before treatment and after treatment, we saw enrichment for this luminal 2 or p53-like signature, which was associated with a lot of stromal markers.

So it is clear bladder cancer is not one disease, and we are just beginning to understand the biology of the different subtypes of bladder cancer. My hope is, in the future we will be able to choose the best treatment for patients based upon biology, understand resistant mechanisms of each subtype, and by use of this, build more rational combinations and sequences, ultimately realizing personalized medicine for our bladder cancer patients.

Tom Keene: That was great. I have one question or a couple of questions, but one that really intrigued me, visceral and non-visceral mets, both groups respond to erdafitinib. How is that-?

Arlene Siefker-Radtke: That is correct.

Tom Keene: ... A third of that, that one treatment is actually able to do both of those types of mets. What's your feeling on that?

Arlene Siefker-Radtke: Well, I think the interest in the visceral versus non-visceral metastasis question really arose from the immune checkpoint data, which suggested that node-only metastases did appear to respond better to an immune checkpoint inhibitor compared to patients with visceral metastases.

And in fact, the presence of liver metastases has been associated with a very poor or abysmal prognosis in the setting of an immune checkpoint inhibitor. So having a treatment that works in rapidly progressive visceral metastases I think reflects an unmet need as well. And to have an oral agent that targets this, I think would be a very attractive option for our patients.

Another thing that's attractive about this strategy, it avoids the renal toxicity that has been associated with platinum. And we saw similar responses, whether the GFR was above 60 or less than 60. So patients who may not be platinum candidates may still have erdafitinib as an option for their FGFR3 altered tumor.

Tom Keene: And then, at the present time, realizing the fact that not all the studies have been done and there's a picture emerging however that if you, the basal response when you said there's a better ... They are the worst type tumor to have, but then when you give them neoadjuvant therapy, chemotherapy, you may get subsequent response rates.

Do you think that same process would be going on dealing with FGFR? That you can basically convert something from a very bad prognostic picture to an improved prognostic picture?

Arlene Siefker-Radtke: Well, that is certainly the hope behind a lot of this work. Can we change the signature or change the tumor, making it more favorable to response to a particular treatment? I would say, we still don't have evidence confirming this, but that's a lot of the reasoning behind the strategies combining FGF targeting agents with an immune checkpoint inhibitor in the hopes that we'll get the benefits of both, whether additive or potentially synergistic, if we can increase the immune response by targeting these immunologically cold tumors.

Tom Keene: It's a huge field. I mean, looking at the data that you presented and the variety of different types of sub-tumors, if you like, that we never knew existed before. I mean, you've got years of work to go on this, but it's a tremendous beginning. And I would ask you one last question, today is accepting the fact that not everything is clear, what would be the ideal patient that you would want to put on, who has a particular type of bladder cancer that you could use this therapy for?

Arlene Siefker-Radtke: Well, at the moment, erdafitinib is only approved second-line patients who failed prior treatment and have a preselected FGFR3 alteration that was included in the clinical trial. So we are currently limited based on approval strategies, but I have to ask myself, if I had a patient who was platinum ineligible and a known FGF receptor alteration or had visceral metastases in the setting of failing a prior chemotherapy in the known FGF alteration, would I want to pick erdafitinib or an FGF targeted strategy earlier compared to immunotherapy based on current clinical data?

Some might argue for that, but clearly we have a lot of work to do to prove these hypotheses.

Tom Keene: A lot of work is better than having no hope.

Arlene Siefker-Radtke: I like that. I prefer more work over less hope.

Tom Keene: Exactly. But it's been a real pleasure hearing this presentation, thank you very much for the time.

Arlene Siefker-Radtke: It's my pleasure.