The Potential Influence of Circulating Tumor DNA in Treatment Decisions for Bladder Cancer - Petros Grivas
June 27, 2023
Petros Grivas, MD, PhD, Professor, Clinical Director of Genitourinary Cancers Program, University of Washington, Associate Member, Clinical Research Division, Fred Hutchinson Cancer Research Center.
Ashish Kamat, MD, MBBS, Professor, Department of Urology, Division of Surgery, University of Texas MD Anderson Cancer Center, President, International Bladder Cancer Group (IBCG), Houston, Texas
ASCO GU 2023: Cell-Free Urinary Tumor DNA to Detect Minimal Residual Disease Prior to Repeat-Transurethral Resection of Bladder Tumor in Non–muscle-Invasive Bladder Cancer: A Prospective Study
Cell-Free Urinary DNA Predictive Impact on Patients with Non-Muscle Invasive Bladder Cancer - Kyle Rose
Ashish Kamat: Hello, and welcome to UroToday's Bladder Cancer Center of Excellence. I'm Ashish Kamat, Professor of Urologic Oncology at MD Anderson Cancer Center.
It's a distinct pleasure to have once again on the stage someone who is well-known to all of our audience, listeners, viewers, Professor Petros Grivas, who is Professor of the Department of Medicine in Medical Oncology at the University of Washington, and really needs no introduction when it comes to the field of bladder cancer.
Petros, with all the work you've done in bladder cancer, one of the things that really is very important for our viewers and listeners, and all the way from residents, all the way up to practitioning oncologists, is the concept of minimal residual disease. That's why when I reached out to you and asked you if you would enlighten us on this topic and you said yes, we were excited. So Petros, welcome, and take it away.
Petros Grivas: Thank you so much, Ashish, for inviting me. It's always a pleasure to be here and discuss with the audience, really exciting to discuss about minimal residual disease. These are my disclosures and I can say research consulting and trials.
I can say that the field is really, really exploding in terms of the role of circulating tumor DNA. The overall promise here is can we detect cancer earlier in terms of screening, for example, or recurrence of disease after definitive therapy in the occult micrometastatic disease level, to introduce some effective and safe therapies in order to improve outcomes, and improve cure rates, and improve survival. That's your overall overarching goal with minimal residual disease detection.
Let me go with some terminologies here. Circulating tumor DNA, we talk about it all the time, is derived from cancer cells, and is released to the circulation as part of setting maybe tumor cell necrosis, or from viable cells, but it's something that we want to understand better in terms of the actual source.
Cell-free DNA is also deliver from non-cancer cells is a more generic term. This is free of molecular alteration and consists of longer base per lengths compared to ctDNA. So ctDNA is a tumor- specific.
Minimal residual disease, let's call it MRD for this talk, is this state of persistent micrometastatic disease after the definitive treatment. That's what we try to discuss today.
Here the big question we know we have to answer is, when we talk about cell-free DNA, as I mentioned, is shed from dying cells, what's the proportion of circulating tumor DNA as part of the cell-free DNA that we're collecting in a blood, in the plasma sample? And we have to have, of course, sensitive and specific techniques that can detect this tumor-derived DNA, even in small amounts, small quantities, because circulating tumor DNA represents less than 0.01% of total cell-free DNA so we have to have very sensitive and specific techniques.
Back in the day, this was probably five years ago, we tried to look at circulating tumor DNA as a method to describe genomic alterations, in particular as therapy settings, and correlate those alterations with response to treatment. Here there is an example of the Guardant360 ctDNA assay with 73 genes, cancer-related, and I have a couple of clinical examples in the middle of the slide, the middle part, is courtesy from the legend, Dr. Nick Vogelzang who, as you may know, passed away recently. We honor his memory.
So Nick sent this case with me. This is a patient who had chemotherapy with dose-dense MVAC for four cycles, had a nice partial response and you see that the variants there of genomic alterations is closing down, meaning many of those clones that are eradicated with chemotherapy. But then the patient had progression and some of those clones may show up again. Then the patient had a beautiful response to checkpoint inhibition and stayed there. The question is, can you correlate, can you evaluate correlations between the presence of circulating tumor DNA minimal residual disease as a quantity, as a quantitative measure, but also look at genomic alterations that may give you hints about mechanisms of resistance?
Similarly in that concept on the right part of the slide, courtesy of Dr. Chris Holmes, this is a patient who had a beautiful response to gemcitabine-cisplatin chemotherapy and you see all these genomic alterations, like a rainbow, different colors went down and really, really contract to a very small tiny amount, which correlates with mass radiological response, but then suddenly, the patient had progression and you see more alterations showing up.
So again, this is hypothesis-generating. This is not something we use in practice to guide therapy selection right now, but definitely, I think it erased the hypothesis that we can use ctDNA to evaluate mechanism of resistance.
Having said that, we use ctDNA in clinical practice, at least in my practice, to detect genomic alterations that might be useful for treatment selection. For example, erdafitinib is an FGFR inhibitor that inhibit its FGFR family and specifically it's FDA-approved for patients with platinum refractory disease with tumors harboring FGFR2 or FGFR3 activating mutation of fusion. That can be the detected in the tumor tissue, which is an early angiogenesis. But sometimes you may not see it in tissue, you may see it in blood. So I'm doing both cancer tissue and ctDNA metastatic disease as my routine practice right now in order to evaluate the potential detection of those alterations and also additional alterations for clinical trials. So I'm using ctDNA in clinical practice. The question that has to be answered is the MRD question. Can you use ctDNA to monitor patients for resistance and guide therapy selection based on that? And the MRD questions is the subject of clinical trials.
Speaking about that concept of MRD, Dr. Lars Dyrskjøt and colleagues in Denmark have done a wonderful job, Dr. Christensen working with that team, and did longitudinal ctDNA analysis in patients with localized muscle-invasive bladder cancer. They try to evaluate the presence of minimal residual disease, MRD, based on circulating tumor DNA in a patient-specific tumor-informed assays, a 16-plex PCR amplicon assay that was informed by the tumor tissue alterations from that particular patient. If you have this tumor-informed assay, you can design this amplicon assay to detect ctDNA in the blood.
What they saw with this serial ctDNA assay was that the circulating tumor DNA is very prognostic. So if you look at the recurrence-free survival data on the left before neoadjuvant chemotherapy, on the middle before radical cystectomy and the right after cystectomy, the presence of circulating tumor DNA in the context of MRD is highly prognostic. So presence detectable ctDNA correlates associated with much higher risk of recurrence of death. Clearance or not detectable ctDNA correlates with a longer recurrence-free survival and potential these patient may be cured.
The question is how do we use this information to make clinical trial designs and potential improved outcomes? If you look at the lower part of the slide, you see that the on average there is a lead time about three months between the detection of MRD in the blood, in the plasma and radiologic metastasis. So the big question is does this, let's say three-month lead time, make a difference? Do we benefit patients by changing treatment or introducing treatment earlier than later or we need to await for the metastasis to show up on the CAT scans. Those questions are being answered in ongoing clinical trials.
If we think about those concepts going forward here, I want to show a few more examples. If you give neoadjuvant chemotherapy, you may see differences, change in the kinetics or dynamics of ctDNA. In some of those patients, the ctDNA may go down and the question is, if that happens, does this, let's say lowering or decrease or clearance ideally of ctDNA, does it correlate with pathologic downstaging and pathologic complete response? Again, small numbers here with hypothesis generating, but there seems to be some correlation, earlier association between clearance of circulating tumor DNA and pathologic downstaging, pathological complete response and better outcomes so the much lower chance of recurrence, much lower chance of cancer-specific death if you have clear associated DNA. Again, small numbers, but very interesting.
Similar data by Dr. Szabados and Dr. Powles in the ABACUS trial. This was a neoadjuvant trial looking at atezolizumab as neoadjuvant therapy for cisplatin-ineligible unfit patients who undergo radical cystectomy and you see similar data here. That ctDNA detectability, the red line, is associated with much higher chance of recurrence or death. However, on the contrary, you have ctDNA clearance, undetectable ctDNA, people do much better in terms of recurrence-free survival, the blue line versus the red line.
So different datasets point in the same direction that I think many of us feel very, very enthusiastic about the very promising role of ctDNA-based MRD to potentially guide therapy decisions, but you have to prove whether it's clinically useful, meaning validate this very promising data from Dr. Christensen, Dr. Dyrskjøt and Dr. Szabados and Powles in prospective trials.
The question is how do we validate this data? What are the questions we can ask in prospective clinical trials? These are just ideas that were discussed by Dr. Christensen in a beautiful paper and the potential desire we can think of is a diagnosis of muscle-invasive. You can potentially certify patients, both ctDNA-negative or positive, and you may reach clinical trials for ctDNA-positive patients. You can evaluate them, see if their ctDNA clears or not, and titrate chemotherapy based on that or change chemotherapy based on that. Again, these are clinical trial designs ideas for clinical trials.
On the right side of the slide, after radical cystectomy, you can potentially titrate or select adjuvant therapies based on the presence of MRD on ctDNA. You see that you might think about trial designs focusing in reaching ctDNA-positive patients and those escalate or deescalate based on that presence or clearance. Again, all these are ideas or concepts that have to be validated prospectively in clinical trials.
We have three adjuvant clinical trials that have been contacted and I can tell you very briefly the adjuvant atezolizumab versus observation. This was a negative trial. There was no improvement in disease-free survival or overall survival in that node with atezolizumab adjuvantly, but I'll show you some very interesting data, a ctDNA data from that trial. There was an adjuvant nivolumab versus placebo that showed disease-free survival benefit with nivolumab. We're still waiting for OS data, but nivolumab is approved by the FDA for patients with high pathologic risk, high pathologic stage and high risk of recurrence after radical surgery and this is being discussed with the patients. We're waiting for the third trial, maybe the tiebreaker, the AMBASSADOR trial led by Dr. Apolo in the NCI with adjuvant pembro versus observation. So three trials in that context, one negative, one positive and one pending, but from you can learn so much even from negative trials.
This is what the work by Dr. Powles, myself and others, I was published in Nature about a year ago. It shows that we can learn a lot from many trials. This is data from adjuvant atezolizumab versus observation and you see very similar data that I showed you before. The ctDNA detectability potential for prognosis, if it's detectable, much higher chance of recurrence or death compared to patients who have ctDNA-negative. Having said that, you may argue that about 30% of patients with ctDNA-negative status may still have recurrence. So it's not perfect, but definitely there is an association with a lower risk of recurrence of death with ctDNA-negative patients. Disease-free survival data on the left, overall survival data on the right. So you see definitely a putative prognostic role of ctDNA there.
What about predictive? Can the presence of ctDNA associate or predict benefit with adjuvant atezolizumab? That's a promise here. That's a hypothesis. You see left, disease-free survival, right overall survival and you see that the presence of ctDNA is associated with benefits, of survival benefit with atezolizumab. So it's an exploratory subset. This is not a prominent point, this is not practice-changing, but raises a very interesting hypothesis that presence of ctDNA MRD may be associated with survival benefit with atezolizumab. So the question is if you reach your trial with selected patients based on ctDNA, can you show benefit?
That's exactly the design of IMVigor 011. This trial is ongoing. It's very exciting, promising trial. Patients who may or not have received neoadjuvant chemotherapy based on the pathologic states and risk of recurrence, they get followed by ctDNA. If they are positive for MRD, then only then they get randomized to atezo versus placebo. That trial I think is going to answer this very, very important question and aims to validate the promising exploratory analysis from IMVigor 010.
Now, very similar concept in the TOMBOLA trial by Dr. Lars Dyrskjøt in Denmark. This is a little bit different design. Patients gets serial ctDNA MRD assay using the Signatera, both trials using the Signatera assay. What you see here is if the patient becomes positive may get atezolizumab. If negative, they get observation. That's what I think is very important as a trial design to look at the potential clinical utility of ctDNA.
Very interesting trial by Dr. Galsky and colleagues in Alliance. This is, again, the patients who may or not have received neoadjuvant chemotherapy and based on the pathologic stage and risk of recurrence, had radical surgery, they get ctDNA. If they're positive, then they get randomized to nivolumab anti-PD-1 or nivolumab plus relatlimab and ALAC3 antibody. So pretty much checkpoint inhibition, single agent or escalated combination immunotherapy for those ctDNA-positive patients. Those who are ctDNA-negative, they get randomized to nivolumab or surveillance, which I think is a very important question to see whether it's ctDNA-negative patients, you need checkpoint inhibitors. So we're very excited to see these trials and I think we have to accrue in order to evaluate the clinical utility of ctDNA.
To conclude, ctDNA is a very powerful and promising biomarker in muscle-invasive disease. It can risk-stratify patients before chemotherapy and cystectomy. It can be used for early diagnosis of MRD before chemo, before cystectomy and after cystectomy and potentially monitor response to therapy. The question is, can we demonstrate clinical utility? Meaning, can we see that if we use ctDNA, we impact the outcomes? Do the patients live longer? But that's a hypothesis. We have to prove it. So can we improve overall survival, disease-free survival, quality of life, or about cost effectiveness? In IMVigor 011, TOMBOLA and MODERN trials are going to answer those questions so hopefully they accrue well.
I would like to thank you so much, Ashish, for inviting me to discuss this topic.
Ashish Kamat: Thank you, Petros. That was a very, very succinct presentation of a complicated topic, which I expected you to do, of course.
A couple of questions, Petros, and talk to our audience because many of them are coming in and very interested in the basic concepts, too. First off, what is your opinion on the different types of ctDNA assay? We have those that are tumor-specific, for example, the Natera. Then we have one which essentially is based on overall mutations that you detect. Well, do you have an opinion on which one might be better? And if so, why?
Petros Grivas: Ashish, that's a great question and we do not have enough data to compare tumor-informed assays like the Signatera with tumor-uninformed or tumor-agnostic assays that do not require tumor tissue for a particular detection assay in the blood. And I can see pros and cons in both approaches.
I think it depends also on the feasibility. In bladder cancer, in muscle-invasive disease, usually we have enough tissue to look at the tumor genomic sequencing and design this amplicon assay like Signatera does. At the same time, if you go to non-muscle invasive disease, CIS, carcinoma in-situ only, there you have challenges you do not have, usually may not have enough tissue to inform your ctDNA assay. So maybe you may want a sensitive tumor-agnostic assay there in the absence of available tissue.
There are other discussions about the time that is needed to develop an assay based on the tumor tissue and the urgency to treat. You are in a curative-intense setting so you want to avoid delays, obviously the faster, the better, but I think we still have way to go to understand better the pros and cons. And, of course, it comes down not only to the time to develop the assay, but also the performance characteristics of the assay, sensitivity, specificity, positive and negative predictive value. So I think the jury is still out there and I think we will have to evaluate the data and be open-minded in either approach.
Ashish Kamat: Right. And currently both of them should be looked at equivalently in the research setting and hopefully, we'll have some studies that are able to compare and contrast the two. It's always hard when you have competing assays of these natures.
Petros, in the studies that you presented and, again, people will say, well, does the ctDNA add anything over the actual PCR data? And if so, how much is the incremental benefit in the neoadjuvant studies that you mentioned?
Petros Grivas: That's a great question, Ashish, and I'm being asked this question in almost every talk I give.
It's a very valid question because the data that I showed you from Dr. Christensen and Dr. Szabados in the slides look very compelling. And the first time you look at the data says, wow, the patients with positive ctDNA after therapy, they do very, very poorly and we need to develop more therapies for those patients.
The second question that comes up is, if you have positive ctDNA after neoadjuvant chemotherapy, are we going to cure or not this patient with cystectomy? That really, really makes a challenge because the question is are we able to make decisions with ctDNA right now?
Personally, I have a high bar on defining clinical utility and I would like more prospective trials to validate this data. I'm not comfortable yet to base a clinical decision about, for example, cystectomy or not or chemo or adjuvant immunotherapy or not based on ctDNA. I won't wait for the ongoing trials that I showed you.
But I think my prediction is in a few years from now ctDNA is very likely, it's more likely than not to be incorporated in what you said. We're going to design trials and evaluate the role of local therapy based on ctDNA, dose escalation or the escalation of therapy. Even in the adjuvant setting, I think it's going to be used, pathologic stage, pathologic complete response versus not, downstaging and ctDNA are going to be in the future be used. I just want to wait to prove clinical utility in trials.
But I know that people do it in different sites and I think if people do it, we have to learn from it. So I would love to have a registry and capture whatever data's out there to inform these future decisions.
Ashish Kamat: Petros, currently, if someone were to use this in the clinic, you would rate the pathologic downstaging of the actual surgical specimen as being more informative than the ctDNA? Is that correct? Is that right?
Petros Grivas: I think we have more data, especially in the context of neoadjuvant chemotherapy, Ashish, from the SWOG 8710 trial by Dr. Grossman and others, that if you have a pT0N0, let's say ypT0N0 after neoadjuvant chemotherapy, you have an 85-plus chance of cure. So if you reverse it, you have less than 15, 10% chance of recurrence. I think that's a very strong data in terms of pathologic complete response in the context of neoadjuvant chemotherapy, 85-plus chance of cure.
Having said that, we don't have that strong data in the context of neoadjuvant immunotherapy. Neoadjuvant immunotherapy is still experimental. It's undergoing phase III trials. Phase II trials look very promising, but we have to wait for phase III trials for neoadjuvant immunotherapy and we do not know yet whether PCR, pathological complete response, is predictive of overall survival in that context. I think it will be great to see that data in the ongoing phase II and phase III trials and also interrogate ctDNA in that context.
So I think we were I would say pathological complete response have very strong prognostic value after neoadjuvant chemotherapy. I want to see more data with neoadjuvant immunotherapy. And again, ctDNA looks very, very promising. It's not perfect, but I think in the future we're going to have more data and I view it as potentially complementary to pathologic response rate and downstaging.
Ashish Kamat: Yeah, and that's something very important for our listeners and audience to keep in mind. It's not ready yet for primetime use. It can be used to inform clinical decisions, but we shouldn't be basing our decisions entirely on that because it hasn't been validated yet, which is very important.
Last question, Petros, and not to put you on the spot, but obviously with the approval that we have of nivo based on the CheckMate study and, of course, with IMVigor 11 and the data from 10, would you advise our listeners that have a patient that they're considering for nivo, would it benefit the patient to get the assay done in a non-research setting or based on the NIVO data, would you say, well, consider for adjuvant nivo based on the CheckMate studies, independent of circulating DNA?
Petros Grivas: Great question. I will give you the short answer and then I will explain.
The short answer is, as of today, I'm making or at least discussing the data with nivolumab with the patient regardless of the ctDNA. In other words, I'm not yet using ctDNA to inform a decision about adjuvant nivo or not. And there are a couple of reasons I'm not there yet. I think we'll get there eventually, but the reason I'm not there yet is number one, in the nivolumab trial, CheckMate 274, we have not seen yet any new data that can be particular with that trial. Would love to see the data. I'm not sure whether we will or not because I don't know if the data's available.
The data from the atezo trial looked very compelling, no doubt, I was a co-author. Obviously, I'm positively biased and enthusiastic about it. But if you look at the ctDNA-negative population, Ashish, there was about 30% percent of patients with negative ctDNA that had recurrence. At the same time, I'm not entirely sure whether I have proven and I have not proven yet, I would say, the therapeutic survival benefit with anti-PD-1 versus no anti-PD-1 in the adjuvant setting based on ctDNA.
So I think it's very promising. I think the MODERN trial by Dr. Galsky and Alliance is going to answer the question very well because you have ctDNA-negative patients, nivo versus surveillance and you have combo versus single-agent nivo in ctDNA-positive. I think it's a very nicely designed trial and hopefully will answer the questions in the context also of the TOMBOLA and the IMVigo 011.
But as I mentioned before and to conclude with my short answer, very enthusiastic, very promising. But so far, I'm having a discussion with a patient about the CheckMate 274 data, but I'm not sending ctDNA assay to inform the adjuvant decision of nivolumab yet.
Ashish Kamat: Okay, great. That's all of what I wanted to hear from you and I'm glad we are on the same page. We usually are, especially when it comes to science and cuisines as well, which is, you know.
Petros, again, thank you so much for taking the time. You and I will, I'm sure, continue to chat, but let's close this in the interest of time for our audience. Thank you again.
Petros Grivas: Thank you so much for having me. Great discussion.