Read the Full Video Transcript

Ashish Kamat: Hello and welcome to UroToday's Bladder Cancer Center of Excellence. I'm Ashish Kamat, Professor of Urologic Oncology at the MD Anderson Cancer Center. And we're joined today by Tahlita Zuiverloon, who is an Assistant Professor in the Department of Urology at the Erasmus MC Cancer Institute in Rotterdam.

Tahlita has been involved with bladder cancer research and bladder cancer network and group activities for many, many years. And Tahlita, it was really great to see this publication that came out under your guidance and under your leadership that looks at non-muscle invasive bladder cancer. And addresses a very important point, which is can we actually predict differential response to intravesical BCG?

So thank you for taking the time to join us today and share your work with our audience. And with that, I'll hand the stage over to you.

Tahlita Zuiverloon: Well thank you, Ashish, for the kind introduction and having me here. Yes, so today I will be talking about our recent publication, which was in Science Translational Medicine, where we investigated whether we could find molecular subtypes that were predictive of response to BCG.

We had two objectives in this study and the first one was to identify the molecular features of high risk non-muscle invasive bladder cancer predictive of BCG response. And of course logically from that we wanted to see that if we found druggable genes and pathways, if we could use these for novel alternative bladder sparing treatment options.

So if we have a look at the study design, five European urological centers were included, and we included patients with high risk non-muscle invasive bladder cancer that received adequate BCG. We performed the central pathology review, and the pre-BCG tumors and the post-BCG recurrences were analyzed by whole-transcriptomic sequencing. The endpoints were progression-free survival and high grade recurrence-free survival, both according to the EAU guidelines. And here you can see we had firstly the discovery cohort of 132 patients and a validation cohort B with 151 patients.

So the first thing that we found were three different subtypes, and we named these the BCG response subtypes, so BRS1, BRS2, and BRS3. And interestingly what we found was that these three subtypes had a differential response to BCG. And importantly the most important one was BRS3, where we could see that the patients had a poor progression-free survival, so a poor response to BCG.

Then of course we looked into the differences in expression of genes between the different subtypes and firstly focused on BRS3. And interestingly, what we saw, was that there was an overexpression of EMT genes, basal genes, HLA and immune genes. So that was a little bit of a paradoxical finding for us, because we would've expected that immune genes would not be upregulated.

So to investigate this in more depth, we performed an immune deconvolution analysis. And what we found, if we look here, you can see BRS1, BRS2, and BRS3. So what we found was that there was an upregulation of cells, or an increased level of B cells, macrophages, tumor associated macrophages, CD8 cells, but also T regulatory cells.

And we had a more in-depth look at these cells and what we found was mainly that there is an increased immune suppressive profile. Here we predicted as well with a proteomic analysis where we did immune histochemistry on CD8, we looked at T regulatory cells, macrophages, and stromal proteomics. And here you can nicely see that in BRS3 there is an increase in T regulatory cells, the macrophages and stromal markers.

Then what we found very interesting was also looking into only the BRS3 group of patients, and then looking into the responders, which are depicted in blue, and the progressors. And you can nicely appreciate here that within BRS3 we also find that there is an upregulation of T regulatory cells and macrophages in patients that develop progression. So we concluded that the immune profile that we saw in the beginning, that it's an immunosuppressive profile, and that fits with non-response to BCG.

So we also looked at BRS1 and BRS2 patients, which is depicted here. Where you can nicely see that in BRS1 these patients have a good response to BCG, and this could be explained by overexpression of genes associated with autophagy, protein secretion and antigen processing, so also BCG processing. BRS2 had more luminal markers that are upregulated, FGFR3, and also MYC targets.

So then of course as a clinician for me it's important to know, so what does this mean clinically? So, as you know, patients are currently being stratified according to clinical pathological parameters, and the high risk patients are divided into high risk and very high risk patients. So this is our cohort according to the clinical pathological stratification.

And what we then did is that we added our molecular stratification to the clinical pathological stratification. And there were two very interesting findings here. First of all, when we added the molecular stratification to the high risk patients, we could see that patients with very high risk disease and BRS3 that they did really bad. And mainly in the first two years you can see that 50% of the patients develop progression, if you compare this to the clinical pathological stratification. So these patients should really be offered a radical cystectomy.

Then another interesting finding is if we look at BRS1 and BRS2 combined with very high risk clinical pathological stratification, you see that the these patients actually fare very well. So we should keep in mind, according to the guidelines, we would offer these patients a radical cystectomy while they do actually relatively well.

So the next thing for us was, "Okay, we have these findings. And we would like to translate these findings into the clinic." And performing whole transcriptomic sequencing on all patients of course is costly and it's time-consuming. So we were approached by InnoSIGN, a spinoff company of Philips, and they developed the OncoSIGNal platform.

So what we tried to see, if the OncoSIGNal platform could be used to identify BRS1-BRS2 versus BRS3. And indeed we were able to identify this with an accuracy of 0.87.

So the next steps for us will be that we will be validating this OncoSIGNal in a prospective Dutch national cohort.

So the last thing that we looked at was post-BCG recurrences, to see if we could see an enrichment of BRS3. So here you can see that we found BRS3 in 29% of the patients in the pre-BCG setting, and then we found an increase to 64% of the patients being BRS3 in the post-BCG setting. So this actually means, or we hypothesized, that there must be a clonal expansion of BRS3 cells during BCG treatment which causes resistance and these patients to fail treatment.

Then we also did immune convolution, and indeed if we compare pre and post-BCG we see again increase of macrophages, tumor associated macrophages, CD8 cells, T regulatory cells. So the immune suppressive profile is also enhanced post-BCG.

Then finally, and this is some work we are currently busy with, is we looked into genes upregulated post-BCG versus downregulated post-BCG to identify novel druggable targets. And some interesting findings that we are currently also investigating are TGF-beta, DDR2. And this is what we are currently working on.

So to conclude on this, is that we found three different molecular subtypes, BRS1, BRS2 and BRS3, with a divergent response to BCG. For us, it's very important to work or to continue to work on BRS3, and especially combining this with the clinical staging.

And this is what we will be doing in the next study, that's the IMPASSE study, that's a Dutch national study which started in March. Where we will be including patients throughout the Netherlands with high risk of muscle invasive bladder cancer, which of course is a real world cohort. And from these patients we will validate externally the BRS molecular signatures, and we will also be working on the different druggable targets in collaboration with the Hub Organoids in the center in Utrecht.

So thank you.

Ashish Kamat: So that's great. Thank you, Tahlita, for sharing your work with us. This is really exciting data that's come out.

As you know, of course, because I've known you for many years now, our group and many of our collaborators have been looking to see how we can predict response to BCG now for over two and three decades. And the understanding has morphed to realizing that it's not necessarily just the tumor per se, but it's what goes on in the stroma, the immune environment. And you've shown that beautifully in your findings.

A couple of questions for you. Of course, looking at the EAU risk stratification with the very high risk and the high risk, that classification was for patients who are not treated with BCG, right? And you are showing now that you can add to that and prognosticate patients. Do you have any insight into how your classification system or BRS3, BRS1, BRS2 might correlate with a response to chemotherapy? Or is that something you're looking at going down the road?

In other words, is this specific to BCG because of the immune signature? Or is this also a general phenomenon that will translate to a response to intravesical chemo?

Tahlita Zuiverloon: Yeah. So yeah, thank you for this question. So that's something indeed we looked at. So we have multiple cohorts, for example, of upper tract to start with, but also with muscle invasive bladder tumors that have been treated with immunotherapy.

And indeed we find the same. These are small cohorts, but you can see the trends towards BRS3 not responding to the immunotherapy. So that's also something we are looking into, but that's really preliminary at the moment.

Ashish Kamat: And any signal when it comes to chemo, intravesical or luminal chemo? Is it also a BSRS3 even though it was a developer BCG? Is that a general phenomenon you're seeing or is that something that remains to be investigated?

Tahlita Zuiverloon: Yeah, that remains to be investigated. We didn't look into that. We don't have cohorts available with sequencing data treated with intravesical chemo.

Ashish Kamat: Okay. Any insight into how you might be able to use this from urine assays? Have you done any work looking at cytology, or shed DNA, urinary DNA, et cetera?

Tahlita Zuiverloon: Well, we didn't do this specifically in this study, but of course that'ssomething coming from Professor Zwarthoff who was my predecessor. And this is something we can look into, for sure.

And I mean, there are multiple RNA based tests as well where you can predict recurrences and outcome and also looking into cell-free tumor DNA. So that's something we would like to look into, for sure.

But I think it's important that we first find which genes are specifically associated with developing progression. So that's something we are doing now by looking in depth into organoids, if we can really recapitulate what we found in the patients.

Ashish Kamat: Right. No, excellent. Now, again, if you look at the data that you presented and the work that you've done, how would you recommend... with the caveat of course that this needs to be validated, but how would you recommend that something like this, and assuming you validate it or you build upon it, is used in the clinic by people when they're counseling patients?

Tahlita Zuiverloon: Yeah. So the ideal for us would be that we validate the OncoSIGNal platform quickly, because it's a consumer ready assay. And if that works, then we would advise at least the urologists in the Netherlands to start using it.

And let's say a patient presents with high risk non muscle invasive bladder cancer, we do a TURBT, we also perform OncoSIGNal on the tissue. And depending on what the outcome is, of course, of the pathological report combined with the OncoSIGNal outcome, I think we should treat the patients accordingly.

So especially in the very high risk group I think this is important, because you can nicely see very high risk BRS1 and BRS2, these patients, they do very well. And when we saw this we were really shocked, because we also offer these patients a cystectomy. So that's something that really needs to change, in my opinion.

Ashish Kamat: Yeah. No, and again, work that has come out from our group has actually shown that even in the very high risk group, if you select the patients appropriately and give them BCG, the progression is 14%, not 40%, which is what correct in Richard Sylvester's paper...

Tahlita Zuiverloon: Correct.

Ashish Kamat: Because again, those patients did not get the adequate BCG.

Tahlita Zuiverloon: Right.

Ashish Kamat: I think one of the strengths, of course, of your work is that it's a relatively homogeneous cohort. The patients all got adequate BCG, you did a well annotated clinical cohort, it's not mishmash of people. So we can sort of get a lot of information from your study, which was very well done. And again, congratulations on that.

Could you share with us a little bit about the OncoSIGNal? You said it's commercially available and people can use it right now.

Tahlita Zuiverloon: Yeah. So OncoSIGNal was developed by Philips, and now they have a spinoff company, InnoSIGN, that took over OncoSIGNal. And it's a qPCR based assay where they look at seven or eight, they're still developing it further, different pathways.

So they look at MAP kinase, for example, the estrogen receptor, Notch, Hedgehog. So these things they look into, these are generally associated with progression. And, for example, in breast cancer it's already being used. And it's really easy, it's a relatively easy assay.

Ashish Kamat: But how does it exactly correlate with your sequencing data? Because it wasn't developed by your sequencing data, right, it's informed-

Tahlita Zuiverloon: No, correct. So we did two things. So first of all, you can analyze sequencing data with their algorithms for OncoSIGNal. That's what we started with. And when we saw that the signal corresponds to the outcome, so patients having a BRS3 OncoSIGNal and also BRS3 whole-transcriptomic sequencing data, that correlated very nicely.

And then we said, "Okay, then we should also look into RNA." Because that's the assay is made for. So we extracted RNA from these patients and we send it to Philips by then. And they did again the analysis, and it fit perfectly. So that's how we performed the analysis.

Ashish Kamat: Great, excellent. Once again, congratulations on this body of work. This is a great step forward in this whole field of looking at response to BCG, this, the UROMOL data, the cytokine separate data, all of this together.

At one point, we obviously hope we can sit with a patient and say, "Based on this, we can strongly recommend go forward with BCG or strongly recommend against it." Because right now, again, as you showed in your curves, there's not a clear separation just using clinical parameters. And sometimes you get surprised and patients actually do really well with BCG when you expect them not to.

So this is a really great effort. And thank you again for taking the time and sharing with all of us.

Tahlita Zuiverloon: Yeah, thank you for having me.