Impact of ctDNA MRD Testing on Clinical Outcomes in NSCLC: Precision Oncology Breakthrough vs Early Clinical Evidence (Round 1)​

To hear the arguments related to testing beyond standard imaging in NSCLC, see Round 2.

Round 1: Does early detection of recurrence using MRD testing translate to improved outcomes in patients with lung cancer? And I'm going to turn it over to our first speaker, Dr. Garon. 

Dr. Garon: Thank you very much. Thank you for the kind introduction, and thank you for talking  long enough so I could get some coffee and—it's still pretty early in Los Angeles. So, and thank you all for being here. So, we're first going to talk about detection of recurrence with ctDNA MRD testing and how that translates to improved clinical outcomes. So, whenever you see a paper—or, for me, sometimes a grant on lung cancer—the sort of first sentence is always the sort of “lung cancer is bad” sentence. Sort of talking about the overall poor outcomes in lung cancer. And, of course, there are a few reasons for that. One of them is that it is a disease that tends to be diagnosed at an advanced stage. But another reason for that is that the outcome for a given stage is often not particularly good. Here, you can see sort of a real-world assessment of patients with early-stage disease. Now, remember, lung cancer is a disease that tends to be diagnosed at advanced stages. But there still are patients—and hopefully with increases in CT screening, there will be more patients—who do have early-stage disease. But what you can see is, even in our best cases—the yellow bars here are three-year recurrence-free survival, and the blue bars are five-year recurrence-free survival—and what you can see is about 15% of the patients, even in our best group, are having, you know, either recurrence or death within the five years after their surgery. When you get into, for instance, stage III disease, what you see is that really two-thirds of the patients are anticipated to have disease come back. So, the sort of issue of disease recurring after curative intent therapy is a particularly big problem in lung cancer. So, it's always interesting, I always like to test the fellows on why we do staging. And their answer is always so that we know how to treat people, But, actually, the reality is that staging is designed so that you can identify uniform populations of patients largely so that you can interpret clinical trial data. And so, for that reason, these staging curves that you see for sort of survival over time are finely tuned to achieve a certain expectation of survival. So, I always like to contrast this because, if you look at some histologies—breast cancer is one that I always think about—oftentimes, our good prognosis post-surgical group is similar to their bad prognosis group. So, here, you do see that you have a very good outcome, particularly if you look at staging 8, which is over closer to me here, with the people who have this stage IA disease. But you can see that once you get to stage II, you're getting to about a 50% chance of having—of being alive five years out. And stage III, you can get substantially lower than that. Now, unlike the last slide—this is looking at overall survival—unfortunately, some of these people have had recurrence of disease and are just being managed by systemic therapy. So, some of the earliest studies looking at circulating tumor DNA in real clinical populations are here of a tumor-informed assay. This is from Charlie Swanton's group. And what you're looking at here on the left are individual patients. And what they did is they looked at a sequence of the tumor. Then, they looked in circulating blood for evidence of circulating tumor DNA. They categorized that as having at least two of the mutations that were seen in the tumor, being able to be identified in circulating tumor DNA. Blue is for a sort of clonal versus red here being not clonal. There are some mutations that are basically seen in almost all of the cancer cells, some of them in less. But when you look, what they found was that they were able to basically identify evidence of disease substantially prior to evidence of clinical worsening of disease or radiographic worsening of disease. So, here, this is an evaluation in patients who have early stage non-small cell lung cancer. Many of the studies are conducted this way. And so, we'll talk about it before. It's basically looking at ctDNA at a landmark period. So, you'll have sort of, prior to the treatment, you'll have evaluated for ctDNA, identified whether there is ctDNA there. And then, there's a period of time that we're looking for. A certain number of weeks after a surgery, for instance, a window of like six weeks after—like, a couple weeks after the surgery to evaluate whether disease comes back. And what you can see here is that the outcomes are quite disparate. You can see overall survival, which is panel A. But really, in many respects, although we always describe overall survival as the gold standard, in this case, what you're really testing for is recurrence-free survival. Does the tumor come back? Because scientifically that's the point of the assay is to evaluate whether there's residual disease there, and does the tumor come back? And you can see the solid line is the patients who don't have circulating tumor DNA, whereas the dotted line is. So, there's a big difference in outcome. This is not lung cancer data here. This is actually GI cancer. There were two studies out of Japan, one in colorectal cancer, one in other GI tumors. This is using the Signatera assay and looking at outcomes basically—whether or not they, in blue, did not have evidence of circulating tumor DNA or, in red, did have evidence of circulating tumor DNA. And here, what you see is some real-world experience in monitoring ctDNA in an early stage, non-small cell lung cancer. Again, they looked for whether or not, if they had positive ctDNA in a specific timeframe, how did the patients do? And you can see, in blue, these are the people who did not have circulating tumor DNA positive. In red, these patients did have circulating tumor DNA. And you can see, obviously, the outcomes are quite disparate. And so, very clearly, you are able to identify a group of patients who do substantially worse if they have circulating tumor DNA that is identified after a potentially curative approach. So, I will pass the clicker on. 

Female [off camera]: We've got some really great energizing music for the con. It's too early, still 3 am. 

Dr. Raez: Thanks to the organizers for invitation. We're going to talk about why we don't do MRD yet. You know, it's hard for me because I have to disclose that I do research for Natera. I do research for Personalis/Tempus, and I do research for Guardant Health because we want them to be successful for the benefit of our patients. We need these technologies to really have better tools to fight cancer. And today, we're going to talk about the limitations of the technology, basically. I cannot say that it's not useful, but we are not maybe there yet. So, everybody has a publication like this, you know, and different tumor types. So, I guess everybody in the room is familiar that if your ctDNA is positive after surgery, or your ctDNA is positive when you start therapy, you don't do one. So, everybody knows that. So, we have tons of publications. So, it's a prognostic, in other words. It's a prognostic, no? But the reason why they don't want to incorporate this technology in our daily tools is because we don't need prognostic tests, you know? So, as Dr. Garon said, you are stage IV, that's a good—I know that your prognosis is terrible. So, we need this to have some, something, some tool that can make us change the prognostic. So, that's why the bar for the MRD is we need to show that this can improve survival, okay? So, we don't need another prognostic. We already know that these patients are in stage IV. What we need to show is finding this in the blood, we can improve survival of the patients. And that's why that's the part that we are trying to reach. And we have several publications. Now, this is actually Dr. Leighl. This is your study from—it was presented in ASCO last year. This is one of the examples of how can make. So, it's basically when you need to use this tool to show how we can intervene or we can do something or we can detect the cancer early. And this is one of the options. These are patients with stage I. And, basically, you see they measured the MRD before the surgery and after surgery. And you have these two options, you know, that if they’re CT negative, it’s the standard of care. Because for these patients with stage I, we don't give chemo. But if this is positive, so that they have a bad prognosis, maybe we can change the prognosis, escalating the therapy compared with observation. So, and then you can see the challenges, no? Okay, we're going to try to, in these patients that are positive, to escalate therapy. Okay, so we test the patients. And look at here, we test the patients. And yeah, we find the ctDNA positive in 22.7%. So, this is a tool that only works for 22.7%. We need a tool that works for 100%. So, that's a challenge. First of all, we need to be able to identify the MRD in the patient. So, otherwise, how are we going to randomize them and intervene? So, in a lot of the studies, you see that's the problem. We use all the tissue for pathology. Then, we use the tissue for NGS. When you order the MRD, a lot of times we say, "Oh, there is no more tissue." So, I'm not saying this is a problem of the test. I'm saying it's a problem that there is no more tissue. So, we really now have to go back to our surgeons or pulmonologists and say, “Hey, now we not only need tissue for the NGS, now we need tissue for the MRD.” So, you need to do one more biopsy, one more pass. I see that's a challenge. And then, you see that there is a—you see the MRD detected. This is the detection and preparation-free survival, recurrence-free survival. And this is the nondetected. Okay, you see the difference between both of them, if we detected or not. But this is the group that you didn't detect it. So, these guys are supposed to be cured, all of them. And you see, no. So, that's why, even if you don't detect here, there are recurrences. So, that's why the test is not 100% successful. The test said, “No, you don't have MRD. Go home.” And then, you see the cancer comes back. So, that's another challenge that we have in all of these studies when we discuss. This one was discussed in the other room earlier today. I don't know if you were in the GI session, because this is a good example of an interventional test. This was presented in ASCO this year. And the people in colon cancer are a little bit ahead of us because they start to do this earlier, before us. That's why this is a good study, you know, that basically you have people with stage III colon cancer. You're supposed to give chemo to everybody after surgery. What they're trying to do here is they're trying to—based on the MRD, you de-escalate, so you give a less toxic chemo, or you escalate, and you give a more toxic chemo. So, to make the story short, because you already saw this presentation in the other room, there was no difference. There was no difference if I am using the MRD or I am not using the MRD. Actually, this was a very frustrating result for all of us because when we went to ASCO, we thought, “Oh, it is going to be positive.” It was not positive yet. So, we need to learn what we need to do to improve this. But, the fact is, it was a negative study. And I think the speaker this morning did a good job giving you some subset analysis and example, that in what settings this can be positive and what settings it cannot be positive, you know? And one thing that we have to do is, as I told you in the beginning, we don’t have to blame the test. I will give you an example. Maybe there is not enough tissue.  So, it has nothing to do with the technology or its components. If there is not enough tissue, you cannot do the MRD. But maybe the key model that we’re using is very lousy. So, what are the chances, if you put a patient with metastatic disease on FOLFOX or FOLFOXIRI, how many people do you cure? You don’t cure anybody. You only prolong survival. So what good is it to detect cancer earlier if I give a treatment that doesn’t change the outcome? So, that is why another factor that we have is that the treatments are no good. The treatments are no good. As long as we don’t have good treatments, we cannot change the story of these patients. So, it's not like the MRD is bad. It's like the MRD needs a good treatment. The MRD needs tissue. The MRD needs many more things to be successful.