Alex Wyatt: Thanks very much. And I've been coming here since 2012, so it has been a pleasure over the years to see so much good science here.
Oliver Sartor: It is. I'm appreciative of the PCF for all of the funding that they've provided and so much knowledge that we've gained because of their funding. Circulating tumor DNA, and gosh, this is a big field. We can focus on the idea of circulating tumor DNA within the radiopharmaceutical sciences, how they merge, how they're distinct, how their knowledge, and how much we're learning. So let me start off by saying, let's talk about baseline prognostic information from the circulating tumor DNA within the radiopharmaceutical trials. And let's start there and then we'll get involved.
Alex Wyatt: Yeah, so I guess a good place to start is to begin with the fact that in some cancers we're talking about ctDNA in the MRD setting and we're talking about just detecting ctDNA and that being prognostic. In advanced prostate cancer, it's not just about detection, it's about how much ctDNA you have. And so what we and others have seen over the years is that the more ctDNA you have in the blood, the more aggressive that seems to represent as a cancer. And so certainly across several radioligand trials now, we're beginning to see that prior to treatment, those patients with the highest amounts of ctDNA in the blood, they have the poorest outcomes overall. And that's generally been seen across treatments. But we have this curious finding in the ANZUP TheraP trial, where actually it was the patients with the lowest ctDNA had really superior outcomes on radioligand therapy. And perhaps that's not quite as strong a relationship in the context of chemotherapy.
Oliver Sartor: What's a little bit interesting is that in these trials you have image-based biomarkers and the ctDNA. Talk a little bit about the relationship between the circulating tumor DNA and the image-based biomarkers that some people may be more familiar with.
Alex Wyatt: Yeah, absolutely. So it's a good grounding to think about fetal cell-free DNA in the context of a pregnant mother. So after a few weeks gestation, you have huge amounts of fetal placental cell-free DNA circulating in a mother's bloodstream. And what you have there, is a rapidly proliferating, invading piece of tissue. And so if you go to cancer now, high ctDNA, people used to think, "Well, it's coming from dead cells, so it must represent the cancer that's dying from treatment or the dying portion of the cancer." What we now think is that that's representing inherently the proliferative capacity of a cancer. When we see high ctDNA, it's representing proliferative cancer cells. And so what we've seen now in the context of the ANZUP TheraP trial, is that patients with really high FDG total metabolic tumor volume will have the highest levels of ctDNA. And conversely, there seems to almost be an inverse relationship with PSMA PET avidity. So those patients with high PSMA SUV mean actually have a little bit lower ctDNA, so almost this yin-yang relationship between FDG and PSMA avidity.
Oliver Sartor: I think this is really important. Most people look at a PSMA PET scan and they say, "Oh, the higher the volume and more the uptake, perhaps we're looking at truly the worst patients." But what you're saying is it's the FDG total tumor volume that is most likely to be correlated with the ctDNA. And because we're not looking typically at the FDG PET, we're missing this relationship. So we have the ctDNA giving an additional view beyond the PSMA PET itself.
Alex Wyatt: Absolutely. So the FDG ctDNA correlation is not perfect, but what we certainly see is that those patients with really high ctDNA, they almost all have very, very high FDG metabolic tumor volume, which we would associate with poor prognosis as well, right?
Oliver Sartor: Yeah. Have you looked by any chance at any organ-specific involvement, if the patient may have had a liver metastasis or lymph node metastasis today, very in accordance with what our expectation may be, that higher ctDNA for the liver and lower from the nodes? Any chance to look at that?
Alex Wyatt: Yeah, so we've looked at that over the years in a variety of different cohorts actually. And what you tend to see is that those patients with liver metastasis will have the highest levels of ctDNA.
Oliver Sartor: Yeah.
Alex Wyatt: Now, I actually think probably this isn't reflective of the environment, as much as it is the cancer that is able to colonize the liver is very aggressive. And so I don't know if it's the liver that is somehow interacting and producing ctDNA, but that cancer [inaudible 00:05:18].
Oliver Sartor: No, I agree completely. And in the old days, we would try to culture these cancer cells out of people, and we found that some of the cancer cells would grow on plastic, and that was like, "Oh my god, that is the nastiest cell ever." It's not the plastic, it's the cells that can grow on the plastic, the cells that grow in the liver that are the problem.
Alex Wyatt: Yeah.
Oliver Sartor: So we get a baseline, we have a relationship to TTV in the FDG sense, but not so much in the PSMA sense. We have a stratification of the biomarker for prediction of those who will and will not respond. Let's move on to talking about a response biomarker. Let's talk about the changes you see and the implications of those changes for the prognosis of the patient.
Alex Wyatt: Absolutely. So there's been several correlative efforts in the last three to five years across different treatment classes in prostate cancer showing that if you have an early on-treatment plasma cell-free DNA measurement, there's a very good relationship between the change in ctDNA fraction from baseline, to say end of cycle one, two or three, a good relationship between that change and the depth and duration of response to treatment. And that relationship has also been observed in other solid cancers. And I've actually been lucky to be involved in an effort with the RECIST working group to try to define how we could standardize ctDNA within a type of RECIST framework. Now that's still ongoing, but I think prostate cancer is actually poised to benefit some of the most from this effort, because obviously, as you know, defining so few patients are RECIST-evaluable if they're bone-only disease and so forth. And then PSA as an early marker of response is a little bit unreliable. Sometimes you need several measurements to really get an idea if there's a real change. There's a lot of noise after that first month.
So what we see is that patients that have really longer responses, you almost get clearance in that first cycle. So you can have high ctDNA and then it disappears. And it's reflecting the fact that the disease is under control. So it's a molecular evidence of response. And so we've seen that in, Chris Sweeney has published on that in some of the phase three trials that he's worked on, Johan de Bono in PSMAfore, which I'm sure you were a big part of as well. And so with these correlative studies we've really begun to establish that this is a potential early response biomarker. And so I'm really excited by that, because many of the Australian trials have collected on treatment samples as well, and I think we'll have an opportunity to define what is a threshold for a clinically meaningful change. And maybe that even opens up the door to things like dose adaptation and changes once we can understand if we have a real time marker for response.
Oliver Sartor: Now we're talking about response, we're talking about four weeks, six weeks, eight weeks, 12 weeks, what timeframe are we typically talking about?
Alex Wyatt: So my experience has been that as early as the end of the first cycle, you have a really reliable output. So in this case it will be six weeks, but I think we can look earlier actually. It's just the cycles make for a very practical schedule for blood collection. So I suspect that's probably going to remain the easiest, the least burden on patients to collect at their time when they're coming into clinic.
Oliver Sartor: I want to do a little bit of a thought experiment. So I'm taking somebody with a high circulating tumor DNA fraction, and then they have undetectable at that first measurement. And then I'm taking somebody with a relatively low circulating tumor DNA fraction baseline, and they're also undetectable. Do those two patients have a pretty similar outcome now? Or is it still dictated in part by the original measurement, which is now part of the history of that individual patient?
Alex Wyatt: Yeah, that's a great question. And actually what we do in fact see that those patients that have high ctDNA initially and clear their ctDNA, they convert to the good outcomes. So yeah, it's a very powerful marker.
Oliver Sartor: One of the things that I really like is to be able to get a rapid readout for the benefit of the patient, and particularly when using exploratory combination therapies and other therapies. I could see this being adapted to really give a nice early readout to the combinations that we all want to explore, but which we don't have the two and three years of follow-up to really give because the field is moving so fast. So this early assessment could really be valuable in the clinical trial setting in addition to patient care.
Alex Wyatt: Yeah, absolutely. I think it could be even a trigger that we can test in trials as a point where we can intensify or even de-escalate on this time point, so yeah, I'm excited by that. And I think there is even potential for an intermediate endpoint in future subject to a lot of validation. So yeah, I'm very excited about it as well.
Oliver Sartor: Yeah, that'd be great. One of the things in the vast world of precision medicine in which we live, we talk about the PI3 kinase mutations, we talk about the Parkin mutation, the RB losses. And so how do you capture all of this heterogeneity, and particularly things that might be lost like RB, are they related to the ctDNA fraction? So I'm just trying to take our precision medicine background and shove it into this new framework.
Alex Wyatt: Yeah, I've been thinking about this a lot recently, because the way ctDNA came into everybody's clinic was to test for things like BRCA2 defects. And we know that most of those defects they arise very early on. Obviously a germline alteration is the first event. And so we know for the most part that those variants don't change over time. And there's obviously some exceptions to that, but that was how ctDNA testing first really entered everybody's understanding. Now we're using it more and more, as you said, to look for tumor suppressor loss for example, like RB1 or TP53, which we know is associated with aggressive disease. We're using it to look for androgen receptor changes, PTEN loss, which might be associated with the PI3 kinase pathway inhibitor activity.
But those changes I think are less reliably captured in archival pre-treatment tumor tissue. And so there's more and more emphasis now, I think, not just on looking in the ctDNA of a CRPC sample a patient wants, but maybe even multiple times to understand how certain clones are selected by our treatment. And so we've even seen in the context of radioligand treatment that you can select for clones that are missing RB1 or TP53 in a small percentage of patients. So even though you're not explicitly targeting the androgen receptor, you can still select for those clones that have some lineage plastic capacity.
Oliver Sartor: It's really interesting, we're moving away from the percentage decline, the percentage improvement to an absolute absence to be able to pull out the best prognostic patients. This is really laudable and some goal that I think we can achieve. So it's hard to talk about a typical trial, but imagine just for a moment that you have a, "typical trial" for metastatic therapy, maybe the PLUTO trial might be an example. How many of these patients are actually hitting the undetectable in that six week timeframe after treatment? Is this 2% or is it 20%? Is it 50%? What numbers should we think about?
Alex Wyatt: Yeah, I mean, obviously that's going to depend on where the trial is in terms of lines. So the PLUTO trial is a good example that's probably the sweet spot, one line exposed to prior androgen receptor pathway inhibitor, that are otherwise eligible for docetaxel treatment. In that type of situation, we'll probably see 20 to 30% of patients where we won't be able to detect any ctDNA on treatment. Now, to some extent, that's a subjective measure, because different tests have different sensitivity, if you collected more blood, maybe you can see evidence if you looked for more mutations. But certainly it represents a very, very steep decline. And in the remainder, there's obviously some where you see almost no change, and those patients are going to unfortunately progress very rapidly. And then somewhere there's a decline, but not to zero. And what we suspect is happening there is you have disease control for the most part, but there's still some lesions that are able to elude the bottleneck of the therapy, and that in some cases represents disease that's less probably expressing PSMA, less vulnerable.
Oliver Sartor: If we went to six weeks and we did a simple histogram between the undetectables, the detectables, I'm going to come up with a number zero to 30, and then these individuals that are higher, I would imagine that that simple stratification would set the future for that patient. You wouldn't know the future of that patient with a very simple histogram, not just the undetectables, but the other ones as well.
Alex Wyatt: Yeah, absolutely. It's surprisingly accurate.
Oliver Sartor: Yeah. I love this work. It is so amazingly insightful. It's going to make us better physicians when we learn truly how to use it, and better clinical trialists. Alex, thank you for all you do. Really, really appreciate your insights and your work in the field.
Alex Wyatt: Thank you, Oliver, and thanks for all your support over the years.