Dec 21, 2022
JCO PO author Dr. Mark Stewart, PhD, Vice President, Science Policy at Friends of Cancer Research, shares analysis on clinical trials and the association between CT-DNA and outcomes in lung cancer. Host Dr. Rafeh Naqash and Dr. Stewart discuss dataset metrics, identification of biomarkers, timepoints, and checkpoint inhibitors, development of new medicines, and novel technologies measuring CT-DNA. Click here to read the article!
TRANSCRIPT
Dr. Rafeh Naqash: Welcome to ASCO's JCO Precision Oncology Conversations, where we bring you highlights and overview of precision oncology. Episodes will feature engaging conversations with authors of clinically relevant and highly significant articles published in JCO Precision Oncology. These articles can be accessed at: ascopubs.org/journal/po.
Hi, I'm Dr. Rafeh Naqash, Medical Oncologist, and Assistant Professor of Medicine at the OU Stephenson Cancer Center, and you're listening to JCO Precision Oncology Conversations podcast.
Today, I am delighted to be talking with Dr. Mark Stewart, Vice President Science Policy at Friends of Cancer Research. We'll be talking about their group's recent paper in JCOPO, titled, 'Changes in Circulating Tumor DNA Reflect Clinical Benefit Across Multiple Studies of Patients With Non-Small-Cell Lung Cancer Treated With Immune Checkpoint Inhibitors'.
At the time of this recording, my guest and I have no relevant disclosures.
Welcome to the podcast, Mark.
Dr. Mark Stewart: Thanks. Thanks so much for having me here.
Dr. Rafeh Naqash: I am excited to discuss this new publication that your group has come out with, and I see this is a significant effort involving both academia and industry, and also your organization. Could you tell us about what led to this work, and then we can go into finer details about what your findings were?
Dr. Mark Stewart: Sure. I might start with just briefly describing our organization. We are a non-profit patient advocacy organization in Washington, DC. And our mission as an organization is really to help accelerate development and access to new medicines. And we do this by doing horizon scanning to see kind of what issues, or emerging technologies, are coming down the pipe that might have implications in oncology.
And we bring together experts from universities, government, industry, patient advocacy, to help develop evidence-based policies that can pave the way for future discoveries, but also accelerate the pace of scientific progress. I think that's really embodied in the manuscript that we recently published, and that we'll be talking about today.
Dr. Rafeh Naqash: Awesome. And I have come across some of the other phenomenal work that your organization is doing, and this is definitely a step forward in developing more personalized therapies, and trying to identify relevant biomarkers so that we can treat patients and their cancers appropriately.
Moving forward, could you tell us some of the main findings from this publication, and then we'll take a deeper dive in trying to understand the kind of data that you used and the kind of analysis that was done? But just as an overview for our listeners, could you briefly explain some of the major findings from this publication?
Dr. Mark Stewart: Sure. Maybe in my last answer, I could have provided some rationale in terms of what led us to even initiate this project. So, I think if you look at trends in oncology drug development over the past decade, the use of precision medicines led to really incredible outcomes for patients. I think that's led to really transformative medicines. And with these therapies, if you look at the drug development paradigm, they've often used these expedited development programs at the agency, and that's largely driven by our improved understanding of the biology of the cancer, and the natural history of the disease, and the ability to use endpoints that can read out earlier, that provide us insights into whether a drug is working or not.
And as we continue to improve the available treatments that patients have, we're seeing the shift where drugs are starting to be investigated in patients that have earlier disease. And because of that, the length of time it takes to understand whether a drug is working or not can often take much longer because of the follow-up time needed to read out endpoints like progression-free survival, or overall survival. And so, the ability to identify new novel biomarkers that can help us understand whether a drug is working or not sooner would certainly provide a lot of value to drug development, it could help expedite the development of new, innovative therapies for cancer patients.
And I think like everyone else, when we first saw a lot of the emerging data coming out on the role of circulating tumor DNA and its potential role in clinical research and care, I think there was a lot of excitement around the potential. And when you look at all these exploratory studies, you see a lot of potential for using ctDNA changes to signal whether a drug is working. But a lot of these initial studies were conducted independently, using different methods, different technologies for measuring ctDNA, and this really left questions about the applicability of this from a more rigorous standpoint, and how this might be applied in a regulatory setting.
And so, we brought together a diverse group of scientific leaders to really design a comprehensive plan that could leverage these prior studies, and then bring them together in an effective manner to where we can increase our power, strengthen our understanding of this association that we see between changes in ctDNA and outcomes. And so, this manuscript is really a first iteration of several data readouts we have planned over the next year. This first study included five different clinical trials of patients with lung cancer that were treated with immune checkpoint inhibitors. And I think one of the figures that I think is most interesting to see in this, is actually our FIG 1., where it's a swimmer plot, where it really shows the differences across all these different trials, and it also exemplifies some of the challenges that we had to overcome when we brought all these datasets together, but yet still arrive at some meaningful data to address our core question of whether changes in ctDNA can predict treatment outcomes.
Dr. Rafeh Naqash: So, definitely an exciting endeavor that you guys have brought forward this aspect of ctDNA. So, in my other life I'm involved in a lot of early-phase clinical trials, and my patients often ask me, "How is it that you're going to assess our response? How long will we be on treatment?" And currently, as most of our listeners know, the standards are primarily using CT scans, or other imaging modalities to assess for responses, and with the innovation that's being made in the field of circulating tumor DNA, it's definitely transforming the world and innovations on the precision oncology side. For example, in GI cancers, there's been significant development, there are clinical trials in this setting. In the lung cancer setting, there are clinical trials trying to assess ctDNA-related changes, and treatment changes associated with those alterations, or decrease, or increase, in the ctDNA.
So, based on what you describe here from the understanding that I have reading through this interesting paper, you used, as you mentioned, previously-conducted clinical trials. Could you tell us a little more about what kind of clinical trials were these, and for non-small cell lung cancer, and what kind of therapies patients were treated with in these clinical trials, and how did you determine which clinical trials you would use for this project?
Dr. Mark Stewart: Sure. When we first launched this project, we basically put a call out to various drug developers, academic investigators, to see what types of data were even available. And at that time, maybe not a surprise to many, there was a lot of ongoing clinical trials that were measuring ctDNA in patients with lung cancer. And because of that, we found a number of clinical trials that kind of fit a general criteria that we felt could be brought together as part of this initial analysis. We tried to keep the criteria broad, we didn't want to be overly exclusive. So, we didn't limit the studies to having to have used a specific assay, for instance, and we thought that that was an important component here because we wanted to understand whether this phenomenon of changes in ctDNA being associated with outcomes isn't necessarily due to a specific technology. And I think the strength of this becoming a potential endpoint is that it is something that can be reproducible and repeated regardless of the kind of technology that was used.
The clinical trials, it’s also important that they had multiple timepoint collections while the patient was on treatment because there's still questions remaining around what's the right time to collect ctDNA, and which timepoints are most correlative to long term outcomes. And we thought those were important things to begin to investigate in our study. And so, it was really the availability of data that naturally led us to first starting in lung cancer.
Dr. Rafeh Naqash: Sure. Thank you for that explanation, Mark. And to the best of my understanding, when I read through some of the details in the manuscript, there's a mention that these were five clinical trials. Patients had been treated with either immunotherapy alone, or immunotherapy with chemotherapy, both being the standards of care, depending on some of the other biomarkers and disease burden. But from a ctDNA perspective, it seems that you were trying to include mostly patients that had at least two assessments done at baseline, no earlier than 14 days, and at least one within the first 70 days of treatment initiation. Is that a fair understanding from what has been described in the manuscript?
Dr. Mark Stewart: Yes. And we didn't limit as well to a specific immune checkpoint inhibitor. Again, for us, our goal was to cast a broad net and try and include as many clinical trials as we could in this first analysis.
Dr. Rafeh Naqash: Right. And I guess, based on the data that you have used in this project, there's obviously heterogeneity with the trials, with the treatment, with probably different lines of therapies that the patients had been treated with, so, that is probably why you were trying to explore different metrics of this ctDNA change. And there's a couple of metrics, it seems, that you and your team have assessed here; one of the metrics that seemed to stand out was this metric where you used three different categories of patients that, depending on the change in the ctDNA, whether they had a decrease, an intermediate category, and a category that had an increase.
So, Mark, based on the findings on the paper, it seems that you're using different ctDNA-based metrics to assess changes in responses and survival. Based on the methodology, what was the most appropriate, or the strongest one that you were able to identify that was associated with differences in survival, and responses as you've sort of explained in the manuscript? Could you describe that briefly for us?
Dr. Mark Stewart: When initiating the study, there were a lot of unknowns in terms of how we would be able to bring together these different datasets, particularly knowing that they use different ctDNA assays that potentially included targeted panels, and whole genome sequencing, and also had different kind of readouts and metrics that were used. And so, early on we explored various metrics. And the one that rose to the top was a variant allele frequency, which is simply the number of mutant alleles divided by the total number of mutant and wall-type alleles. And there's different ways you can report that out -- it could be a mean, it could be a median, or a maximum. While we didn't necessarily see large differences between those, we did observe that consistently, maximum VAF correlated with overall survival and was one that we continued to use as a primary analysis in our manuscript.
Dr. Rafeh Naqash: So, Mark, based on some of the analysis you've done here, was a landmark timepoint used to compare survival for the patients because there could be a difference in the number of timepoints of ctDNA assessment for different patients, which would, in turn, mean that some patients could have been treated longer versus some other patients? So, did you try to limit that heterogeneity by performing a landmark analysis on this cohort?
Dr. Mark Stewart: Yes. That was one of the approaches we had to take, given the heterogeneity across the different studies that we included in this analysis. In FIG 2., again, you can see across the five different studies how there are different numbers of ctDNA timepoint collections, but also that they were collected at different timepoints. And so, to try and create a more equal playing field and informative analysis here, we did use a landmark of 70 days here and used ctDNA timepoints that were around that landmark when we were looking at the association between changes at that particular time to overall survival, or progression-free survival.
Dr. Rafeh Naqash: So, Mark, based on what you've shown here as far as overall survival is concerned, could you tell us about how the specific ctDNA metric that you used was able to compartmentalize patient survival on checkpoint inhibitors?
Dr. Mark Stewart: Sure. As I'd mentioned previously, in this analysis, we used a three-level variable to differentiate response to treatment using ctDNA. This three-level ctDNA metric represented patients that had a decrease in ctDNA from baseline, an intermediate change, or an increase. And so, for each of these three ctDNA metrics, we were able to bucket patients into these three categories based on the percent change and varying allele frequency. And so, those that had the 50% decrease in ctDNA were bucketed in the 'decrease' category, and those that had a 50% change in the positive direction were bucketed in the 'increase', and then all the patients that remained were placed into the 'intermediate' category. And as you can see from the Kaplan-Meier Curves, I think a quite robust differentiation between those three groups.
Dr. Rafeh Naqash: Right. And another interesting finding on one of the forest plots that you've shown is that patients who smoked had a better survival or better outcome with therapies. And I remember a couple of years back reading an interesting paper in Science, if I remember correctly, from a group at Sloan Kettering, showing the increased neoantigens and mutational burden related to smoking, that probably predicts better responses. Is there any other interesting aspect to this from a ctDNA standpoint that your group was able to identify, or is looking at, at least in this comparison of smokers versus never-smokers?
Dr. Mark Stewart: Yeah. Actually, when we took into account all the different clinical variables that were included in these datasets, we actually saw an association between patients that were smokers or had smoked at one point, and the ctDNA levels that were present. Despite that smoking may impact the levels of ctDNA present, I think, the fact though, that we are trying to determine whether a treatment is working, or not, based on a change, or a delta, between a baseline and a subsequent timepoint, it shouldn't really matter necessarily what your baseline is, so long as you're still able to observe a decrease or an increase.
Dr. Rafeh Naqash: From, I guess a futuristic perspective since this project primarily involved pooling of data from five different clinical trials, is there a plan to validate some of this in an independent cohort of patients in the next year to two years?
Dr. Mark Stewart: Yes. As I mentioned, the CT Monitor project that we've developed into multiple different modules, and the modules have been broken up basically in terms of when data is available in different clinical trials. And so, we've continued to work with different drug sponsors and academic investigators to put together data use agreements, and I'm excited to say that we have three different modules that we plan to read out over the course of this next year. We have a module that includes patients with lung cancer that are treated with a TKI - Tyrosine Kinase Inhibitor. We have a second module that is an additional analysis in patients with lung cancer treated with immune checkpoint inhibitors. I think a key thing to highlight there is that the studies that we'll include in this next round are all randomized clinical trials, and so, we'll be able to look at the ability to differentiate between two treatments using ctDNA, and understanding its kind of predictive nature as a potential endpoint. And then the third module is really kind of a catch-all that includes multiple different cancer types, and where patients have been treated with either a Tyrosine Kinase Inhibitor or an immune checkpoint inhibitor.
And across all of those studies, we have about 22 different clinical trials that'll be included in those analyses that represent over 3000 patients. And so, I do think we'll be able to validate the findings that we've been able to show in this latest publication, and also address some additional questions. Many of these studies are newer, so I think people have learned from some of the earlier clinical trials that included ctDNA. So, in these latest studies you see many more collection timepoints. You also see earlier collection timepoints even before the first RECIST measurement. So, I think we'll be able to understand, again, get greater granularity on, what are the optimal timepoints for collecting ctDNA; how early can ctDNA predict clinical outcomes; and finally, just to add to that body of evidence that can hopefully help provide confidence to the community that this is an objective and meaningful measure, and that could hopefully be used in a regulatory standpoint as a potential early endpoint that could serve as a basis for a drug approval.
Dr. Rafeh Naqash: Those are all awesome, phenomenal things that need to be accomplished, definitely, and your organization is leading these efforts in a one-of-a-kind public-private partnership. And I completely agree with you that this could be a very important biomarker metric for patients who are treated with novel therapies, including checkpoint therapies, Tyrosine Kinase Inhibitors, or even potentially, on clinical trials.
From a practical application standpoint, Mark, has your organization also made any efforts in terms of trying to see how we can try to get both FDA and insurance approval for patients who are treated on standard of care therapies but would benefit from having serial, you know, every four to six months ctDNAs done? Because I try to do that sometimes in my practice and do face occasional challenges from insurance companies where they're not willing to pay for repeated ctDNA assessment. And I try to space it out when I'm not getting CT scans, you know, alternate it with CT scans. But is there any kind of an effort in that direction where from a regulatory standpoint, in the next one to two years, we can make progress so that clinicians are able to order ctDNA on patients, whether they're treated on checkpoint therapies or other targeted therapies?
Dr. Mark Stewart: That's a critical issue that you raise, and certainly a potential barrier that can prevent patients from benefiting from this novel technology. To date, our efforts have been mostly focused in kind of the use of ctDNA in the research space, but I think that they're not mutually exclusive. I think one important way to ultimately get coverage for these tests is to have that evidence base that really shows the utility that it is providing benefit to patients. I look at this as kind of a first step to getting there, and I think once you have this evidence, and also you have the right evidence that we know payers want to see, that hopefully, it can help address those potential concerns.
Dr. Rafeh Naqash: I could not agree with you more. And your work, definitely, is a step forward in the right direction, and I think will lead to a lot of exciting subsequent things, as you mentioned.
I congratulate you and your group again, on this exciting project, and I appreciate that you chose JCO Precision Oncology as the final destination for this phenomenal work.
Dr. Mark Stewart: Thank you for having me. And certainly I just want to extend gratitude to my fellow authors. I think we have around 34 authors on this manuscript, and I think it just, demonstrates the recognition that this truly does require collaboration across different stakeholders, and look forward to next steps.
Dr. Rafeh Naqash: Thank you so much, Mark.
Thank you for listening to JCO Precision Oncology Conversations. You can find all our shows, including this one, at: asco.org/podcasts, or wherever you get your podcasts.
To stay up-to-date, be sure to follow and share JCOPO content on Twitter with the handle: @JCOPO_ASCO.
All JCOPO articles and series can be found at: ascopubs.org/journal/po.
The purpose of this podcast is to educate and to inform. This is not a substitute for professional medical care and is not intended for use in the diagnosis or treatment of individual conditions.
Guests on this podcast express their own opinions, experience, and conclusions. Guest statements on the podcast do not express the opinions of ASCO. The mention of any product, service, organization, activity, or therapy, should not be construed as an ASCO endorsement.
Guest Bio
Mark D. Stewart, PhD, is Vice President, Science Policy at Friends of Cancer Research in Washington, DC, an advocacy organization that drives collaboration among partners from every healthcare sector to power advances in science, policy, and regulation that speed life-saving treatments to patients.