Biomarkers and Next Generation Sequencing in Advanced Bladder Cancer - Noah Hahn

December 16, 2021

In this 2021 LUGPA presentation, Noah Hahn presents "Biomarkers and Next-Generation Sequencing in Advanced Bladder Cancer Patients". Dr. Hahn discusses the differences in prognostic biomarkers and predictive biomarkers before discussing PD-L1 on the immunohistochemistry side, sequencing biomarkers, tumor mutational burden, and microsatellite instability or mismatch repair deficient tumors.


Noah M. Hahn, MD, Professor of Oncology and Urology, Department of Oncology and Urology, Johns Hopkins University School of Medicine, Deputy Director, Johns Hopkins Greenberg Bladder Cancer Institute at the Johns Hopkins School of Medicine, Baltimore, MD. 

Read the Full Video Transcript

Noah Hahn: So I'm going to cover biomarkers and next-generation sequencing. My disclosures are listed here.

And I think, first, let's just talk about what a biomarker is. I just put up here a definition. And this is actually provided by the NIH, in one of their workshops back in 2016.  And I think the key characteristics is that a biomarker is defined, it's not an abstract thought. It can be measured. And it's an indicator of a biological process or response to an intervention. And those are true, no matter what the biomarker is that we are talking about.

Some examples of biomarkers that we use every day are outside of the cancer sphere. But these would be things like troponin for acute myocardial ischemia, a pulmonary function test for looking at obstructive airway disease, EEGs for seizures, C. Diff toxins for infections, and more recently, with the pandemic, I think we are all familiar now with COVID antigen PCRs, as well as rapid antigen tests.

However, when we talk about biomarkers, and we're talking about cancer and other therapies, predictive versus prognostic is something I think we all get hung up on a little bit. I don't want to focus so much on the curves right now. I'm going to come back to these after an example.  But, prognostic biomarkers sort of tell us a good disease from a bad disease. They don't necessarily tell us how to guide therapy. Whereas, predictive biomarkers are generally inherent to very specific therapies, and in some cases, can actually separate out outcomes based on therapy approaches, rather than just portend a good versus bad risk disease.

So, these are the biomarkers I'm going to try to cover in the next 10 minutes. Quite a bit. So let's dive right in first with PD-L1 on the immunohistochemistry side.  So, PD-L1 has been much maligned over the last four to five years. Sometimes we've liked it, and sometimes we haven't. But I just want to go through real quick, what we've seen in bladder cancer and why this is still part of our armamentarium.

If we go back to the first studies with immune checkpoint inhibitors that were done in advanced bladder cancer patients that did look at this, these were in their early studies, the phase one studies for atezolizumab, and then the phase two study with pembrolizumab, Keynote052, both of these studies looked at their immunohistochemistry stain for PD-L1 protein staining. And in both studies, it looked like there was a trend towards increased responses with increased PD-L1 positivity.

Now, I think it's important when we talk about positivity with PD-L1 markers in any of these antibodies, that each of these antibodies are separate. And this is really important, it's not a one-size-fits-all. And what wound up happening is that each of the five drugs that were subsequently approved in bladder cancer developed their own antibody, with the exception of the SP263 antibody for both avelumab and durvalumab. And what you can see is there are different cutoffs, different things that make them positive or negative, immune cell staining, tumor cells staining, both staining, either/or, in various cutoffs, 1%, 5%, 25%, et cetera. But what we saw from this, in general, when you take this as a whole, is that it looked like there were more responses in patients that are PD-L1 positive by the marker of choice for the drug that was used.

However, we thought we had that figured out going in. We were like, "Great, we can figure out patients that we can respond to and maybe not." So, these drugs were first approved after platinum therapy. And then around 2018, this marker positivity became really an utmost priority when the FDA changed the labels on atezolizumab and pembrolizumab in the frontline setting, and they actually then required patients to be PD-L1 positive, or for people to say they couldn't get any therapy or any chemotherapy, regardless of status. Well, why was this?

Well, this was because of two ongoing phase three trials that were looking at chemotherapy plus immunotherapy, versus chemotherapy alone. And in both of these trials, highlighted here in the middle, there was an immunotherapy alone arm, atezo, and pembro in the respective studies.

Well, the FDA made that label change based on some interim analyses that they were following, and shown here are the outputs for both studies at the end of the day. On the left, the ImVigor 211 study with atezo, and on the right, Keynote-361 with pembro. Both studies were negative. However, if you look in the bottom left-hand corner highlighted in purple is the PD-L1 negative analysis for the atezolizumab monotherapy, versus the chemotherapy control. And in this case, the blue arm is actually the treatment of atezolizumab. And so, it caused pause for us because it looked like these patients were possibly doing worse than with traditional chemotherapy. And that prompted the label change where they said, "Wait a sec, if we're going to be doing this against chemotherapy, we, at least, need to see them PD-L1 positive." It didn't answer the whole question, however, because, again, these trials were overall negative.

So, in this particular case with PD-L1, this wound up being sort of a prognostic biomarker, rather than a predictive biomarker of success, with atezolizumab and pembrolizumab respectively in these trials.

Well, the story's not over yet though with PD-L1. So this last summer, and again, this is all in the metastatic setting, in August, late August, the FDA made another label change on pembrolizumab. And in this particular setting, it's kind of interesting. They voted to change the accelerated approval for pembrolizumab in the frontline metastatic urothelial cancer population to full approval. And they now required none of that PD-L1 status. They simply said that the only people we should offer pembrolizumab therapy to are those who are completely ineligible for any chemotherapy, regardless of PD-L1 status.

There have not been any label changes to atezolizumab yet.  However, I think many of us anticipate this may follow suit. I think they are waiting on some more mature follow-up with the ImVigor 211 study. So it's been a back and forth with PD-L1, and I would tell you, for myself, in my own practice now in the advanced setting, I do not really use it much to incorporate into who I treat and who I don't treat. I think we've started giving more chemotherapy upfront and reserving the checkpoint inhibitors as an isolated therapy for patients that are pretty frail, who are not going to be chemotherapy candidates. But this is a moving target, and it does keep changing.

Sorry, I got a little bit too far ahead of myself. So, let's talk a little bit about some of the sequencing biomarkers, however. And I'm going to go through a couple of these, one by one, FGFR, tumor mutational burden, then the MSI story.

When we look at platforms that are available now in commercial labs across the country, there are a whole number of panels that are out there. And I've listed them here in this one table. They are pretty similar, but there are a couple of differences. There are differences in the number of genes reported on each of these particular panels. Some of these panels report tumor-only mutations, so they only require you to send an FFPE tissue to them. Whereas some of them also require some type of a germline sample, either peripheral blood, saliva, something else. And in those particular assays, in this case, Tempus and MSKCC, they report somatic mutations, meaning, mutations that are unique to the tumor, not something that has been inherited with us. And then there are varying degrees of whether they report fusions or transcripts across each of these particular different assays.

Well, the reason this is important for bladder cancer and, particularly, I think, as we're thinking about earlier stage disease is that FGFR3, as most of us know in bladder cancer, is a highly frequently mutated gene, particularly in early-stage disease. Shown on the left is just sort of a cartoon schematic of the actual gene itself. It's a transmembrane domain, a tyrosine kinase receptor. In low-risk patients, these low-grade TA papillary tumors, the FGFR3 mutation rate is extremely high. Even in intermediate, it's about two-thirds, but as we go to high-grade disease, it starts coming down a little bit better. So you kind of see that in that middle, just kind of a simplistic graph, that as T-stage and grade goes up, we see the decreasing percentage of FGFR mutation. And shown on the right is actually data from the TCGA Project, which muscle-invasive bladder cancer, we saw about 14%, 15% of patients with FGFR3 activating mutations, and about 2% with FGFR3 fusions.

Well, the reason that's important for bladder cancer is that it has resulted in actual therapy. So shown here is the response data for erdafitinib from their BLC2001 study, which resulted in its FDA approval in 2019. So this is an oral drug that inhibits FGFR, it has a response rate of 40% in patients with metastatic disease with activating FGFR3 or FGFR2 alterations. And so this was approved, based on this response rate in the advanced metastatic patients in a trial of about 99 patients.

Now, relevant to bladder cancer, I showed you the FGFR3 story, but I highlighted in green here, again, going back to TCGA, there are many other mutations that are present in frequencies that are relevant and, I would say, represent a decent portion of bladder cancer patients, so 10% to 15% in several of these other targets. So these are all drugs that are either approved for other cancers or are in development for bladder cancer patients, that are trying to capitalize in sort of this model of FGFR3 of getting the right drug to the right patient for their mutational profile.

So, shifting gears again quickly, but let's talk a little bit about tumor mutational burden. This whole concentration on tumor mutational burden comes about by this idea, again, about more shots on goal, so to speak, that every mutation that our immune system sees is potentially antigenic and stimulatory to the immune system. I mean, in theory, it shouldn't be there, and it can be recognized as foreign. So, tumors with higher tumor mutations, just by sheer numbers, are presenting on their surface in the context of MHC to our immune system, many more peptide sequences, again, that may stimulate the immune system.

So, in bladder cancer, from the atezolizumab trials that were done, they looked at this, and there did seem to be an association with higher tumor mutational burden in tumor response in the metastatic setting, shown here by the bar graphs on the left. And then, I think what's a little bit easier to see is the actual quartiles on the survival curve on the right. The red line at the top is essentially the upper quartile tumor mutational burden. So, higher tumor mutational burden patients lived longer with atezolizumab treatment.

This has been looked at across many tumors. And in 2020 last summer, pembrolizumab was actually approved by the FDA, in a sort of pan-cancer approach for tumors with high tumor mutational burden, in this case, as measured by the Foundation Platform, which had metastatic disease and prior therapy. And in their trial, they saw a response rate of about 30% in TMB high patients, compared to about 6% in TMB low patients.

However, the story is not quite as clean as that. We would like it to be that way. And this is where it gets kind of complicated and tricky, both for everybody, medical oncologist, urologists, and the like. The tumor mutational burden story is not something that is conserved across every single cancer. So, this is a really nice paper that was published a little earlier this year, in 2021. And what they looked at in this, these are all patients who were treated with immune checkpoint inhibitor therapies across about 20 different tumor types. They sequenced all of these particular cancers and then profiled responses according to tumor mutational burden.

But also what they did is they correlated the CD8 T-cell score, from a gene expression standpoint, with the tumor mutational burden, and by algorithms that assessed the tumor mutations as to whether or not they could be, what we call, neoantigens, stimulating the immune system because of this. In the pink, on the plot on the left, those are cancers in which the tumor mutational burden or neoantigen loads seemed to correlate with the CD8 T-cell score. Whereas most of the tumors below that in the gray, there really was not a correlation. Now, bladder cancer is actually in one of these tumors on the top. So that's kind of good for us. It seems like we are a high TMB tumor and does correlate with neoantigen load. But it's not one size fits all, so this is really context-dependent.

And then moving on again quickly, I'm just going to cover a couple of things here because it's quick. The MSI story is similar to TMB, although a different story in that, pembrolizumab was approved, again, in a tumor agnostic fashion, for patients with microsatellite instability or mismatch repair deficient tumors. This is an important part of the DNA repair complex. And in their particular trials, patients that had MSI tumors did better with pembrolizumab, compared to chemotherapy. And this was panned out from [inaudible 00:14:44] and Hopkins and many others in colorectal cancer as well.  The relevance for this is that there are mismatch repair and MSIs, particularly in upper tract tumors. So this is something where I think we may have a potential lynch syndrome index patient that we need to look at, because that may shape where we go down the road, both perioperatively or others in early-stage disease.

And then lastly, as I'm up against time, I'm just going to talk with you real quick about circulating tumor DNA. This is really exciting, but I do not think we are quite ready to make decisions yet. This is based on the concept that tumors shed DNA into the circulation. There has been some really nice work done by Alex Wyatt's group out of Washington and their Rapid Autopsy Program, as well as collecting from patients that are still alive that have done some concordant studies between ctDNA mutations and the actually matched tissue. And then shown on the far right, which is really pretty exciting, is some of the work that has come out of Lars Dyrskjot's group over in the Netherlands that has looked at ctDNA prior to cystectomy, and then afterward. And it really does appear that the survival and also recurrences seem to track with detectable ctDNA sort of load. And so that has spawned several trials trying to look at, are these the patients we should focus on?

So in summary, we'll just kind of go through it. It's a lot to cover. There are a lot of tissue-based nextgen sequencing markers in development. PD-L1 is prognostic but not predictive. I think ctDNA is exciting. And I think the next generation sequencing is really going to become part of the standard workflow. I believe in non-muscle and muscle-invasive bladder cancer as these therapies move forward.

So I'll stop there. I have a tremendous team that I want to acknowledge, as well as my other co-chairs and panelists for this session. So thanks for that.