The future of liquid biopsy
Physician Ash Alizadeh has seen the future of disease diagnosis and monitoring.
It is coursing through every patient’s veins. Traditionally, biopsies have required invasively gathering tissue – from a lung, a liver, or a fetus. Now it’s possible to look for disease without surgery. The DNA is sitting there in the bloodstream, Alizadeh tells host Russ Altman, as they preview the age of liquid biopsies on this episode of Stanford Engineering’s The Future of Everything podcast.
Transcript
[00:00:00] Russ Altman: This is Stanford Engineering's The Future of Everything, and I'm your host Russ Altman. I thought it would be nice to revisit the original intent of this show. In 2017, we wanted to create a forum to dive into and discuss the research my colleagues across the campus are doing in science, technology, medicine, and other areas. My goal was to show you that these are people working hard to improve the world. The university has a long history of doing work to impact the world. And it's a joy to share with you the motives and the work of these colleagues as they try to create a better future for everybody. I hope you'll remember that when you think about universities and their role in society. I also hope you'll walk away from every episode with a deeper understanding of the work that's in progress here and that you'll share with your friends and family as well.
[00:00:50] Ash Alizadeh: Now immunotherapy is going to become a cornerstone of the frontline treatment of patients as we start removing elements of, you know, these toxic chemotherapies and often we used to use them in combination with radiation. Can we start shaping that towards more directed strategies? Can we match the microenvironment with a particular therapy that might not be as toxic? And that's, so that's emerging work that I think liquid biopsies will be an important part of.
[00:01:20] Russ Altman: This is Stanford Engineering's The Future of Everything, and I'm your host Russ Altman. If you're enjoying the show or if it's helped you in any way, please share it with friends, colleagues, and family. Personal recommendations are a great way to spread news about the podcast.
[00:01:35] Today, Ash Alizadeh will tell us how our ability to detect the DNA of cancer in blood is revolutionizing our ability to detect cancer and follow treatments. It's the future of liquid biopsy. Before we get started, a reminder to tell friends, family, and acquaintances about The Future of Everything podcast. Personal recommendations are what it's all about. I'm excited to announce that today we're going to do something for the first time on The Future of Everything. We have an audio question from a listener. We're going to play that question and we're going to ask our guest to help answer it. That's coming later in the second segment.
[00:02:19] So you know when you get cancer there's often a biopsy, usually a small or serious surgery in which you're opened up and they take a piece of tissue and they use that tissue, uh, to understand what kind of cancer it is and to choose the therapy. In the last few years though, liquid biopsy has become a new thing. This is where they take a sample of blood, uh, either it's to detect cancer in the first place, but currently, it's mostly used after you've been diagnosed to follow the effects of treatment. Is there still signs of disease or are we seeing no trace of cancer DNA in the blood? Well, this has continued to develop. There are companies that offer the basic technology. But we're pushing the frontiers and especially with this new field of immune-oncology, where we're treating cancer by becoming allies with the immune system. There are new possibilities and new ways to use liquid biopsies to advance our understanding of cancer and its treatment.
[00:03:19] Well, Ash Alizadeh is a faculty member at Stanford University in Oncology and Medicine, who's an expert at liquid biopsy. He's going to tell us how it works, how it's being used now, and how it may evolve in the future.
[00:03:33] Ash, to start with, we really should define, what is meant by liquid biopsy?
[00:03:39] Ash Alizadeh: Oh, well, Russ, uh, the idea is that you access the tissue you're interested in through a body fluid of some kind, most commonly the blood, for sake of measuring things that you would traditionally rely on through, uh, an invasive strategy. Uh, whether it's a surgical procedure or a minimally invasive radiographic procedure where a needle is put into the tissue of interest. So whether using blood, urine, um, cerebrospinal fluid, bronchial lavage fluid or other fluids that are obtained without a surgical procedure or breaking a barrier to get into the tissue, uh, to kind of get a non-invasive window into that tissue. That's the broad idea.
[00:04:31] Russ Altman: And so following up on that, when we do a biopsy, you know, usually if like you have a lung, somebody saw something on your X-ray and you're getting a lung biopsy, you usually go to the spot where you saw the abnormality so that you can grab a piece of tissue. But very often the fluid that's being accessed is the blood. And so my question is, why can we reliably see, for example, a lung cancer by taking a sample of blood from the arm or, you know, or a liver cancer or even a bloodborne, bloodborne cancer is easier to understand because it's sitting there in the blood. But do we have access to these other more localized cancers even through the blood?
[00:05:10] Ash Alizadeh: We do. Um, and just thinking beyond cancer, the ideas for this actually started before cancer was being looked at in a liquid fashion. The other two major applications are for prenatal diagnostics. So when you're trying to avoid doing an amniocentesis or chorionic villus sampling of placental tissue and the early fetal tissues to get an idea of whether there's a fetal genetic abnormality, that can be done through the blood too. Or for organ rejection.
[00:05:43] Russ Altman: Yup.
[00:05:43] Ash Alizadeh: The idea of, you know, instead of going into the heart and plucking a piece of the heart out to see if it's not being rejected by the recipient of a um, organ graft, heart transplant, uh, to get at the blood and see if the DNA of the donor of that organ is spilling into the blood as a reflection of the tissue injury from inflammation that's rejection. So that's the basic concept, but yeah, localized cancers can be detected in the blood from a range of cancer types and data is emerging about the clinical utility of doing that for preventative screening type strategies or for monitoring a definitive procedure that a surgeon or radiation oncologist does to eradicate a local tumor.
[00:06:32] Russ Altman: And the general idea is that there are cells that are part of the tumor that are just basically dying and releasing their cellular contents into the blood. And we're able to sensitively detect mostly DNA, is it, are we still talking mostly about DNA?
[00:06:46] Ash Alizadeh: Yeah, I think it's still mostly DNA and that's the basic idea. For example, in the prenatal world, the early work showed that within a few weeks of pregnancy, like at six weeks of pregnancy, as much as six to ten percent of the DNA in maternal circulation is fetally derived. But if you look at the cells in the blood, you don't find, let's say you use chromosomes,
[00:07:11] Russ Altman: Yeah.
[00:07:11] Ash Alizadeh: And you have a male fetus and a, of course, female pregnant mom, you don't see six percent of the cells being male. It's just that you have a rapidly dividing and dying tissue that sheds a lot of DNA, and that DNA comprises a large fraction of the DNA molecules in circulation. Um, so that's the basic idea.
[00:07:34] Russ Altman: Great. And so, of course, for the prenatal, it makes sense that the baby's DNA, although it's related to the mom's, is not identical. There's usually a dad involved. And so that those differences could be detected. Now for cancer, is it a similar thing? The cancer, uh, the DNA of the cancer diverges, even though it's derived from that patient, it has diverged enough so that you and your colleagues can tell the difference between what we might call the normal DNA from the patient and this abnormal, worrisome DNA.
[00:08:03] Ash Alizadeh: Yeah, that's a great question. The, there are differences and some of those differences are the genetic differences that we look for in mutations. And so if we know that certain cancers have particular types of mutations that are very characteristic, let's say, and, you know, the very famous genes like, uh, P53 and KRAS and EGFR and others, that hotspot mutations or clustered mutations are very common, uh, and seen in a large fraction of patients, you could look for that. So that's one type.
[00:08:41] Russ Altman: Yep.
[00:08:41] Ash Alizadeh: So genetic mutations, point mutations. Then there's structural alterations like copy number aberrations that, an extra copy of a chromosome. The same way as you might detect a down syndrome fetus in a mom by seeing extra chromosome twenty-one, could you do the same thing for say, a HER2 amplified breast cancer or an androgen receptor amplified prostate cancer, something like that copy number. And then the other class is to not look at the mutations and genetic alterations, but instead look at the epigenetic DNA changes, um, which are characteristic by DNA methylation differences, CPG methylation, and there are companies that work on this technology and we've been working in this area as well. Or, uh, an entire different class is to look at fragmentomic features. So how are the DNA fragments that derive from the tumor different than the non-cancer derived fragments in their size and other. So these are, there is a whole range of DNA differences between tumor derived molecules and non-tumor derived molecules that we leverage for this, you know, is it cancer or not question.
[00:09:51] Russ Altman: Great. Great. Okay. So now that, so we've established that there are the differences and I do want to spend a little time 'cause this is amazing that we can do this. What do, how do we, how does it actually work? How does the technology for detection of these differences? And of course, I'm guessing there are several technologies cause you just mentioned several differences and we could, but maybe at a high level or the most common, how do they most commonly do this very sensitive detection of DNA differences?
[00:10:16] Ash Alizadeh: Sure. So let's take the example of, um, mutation based differences. And mutation-based differences, so let's say a patient comes to our clinic with an established cancer. Um, and let's say it's an advanced cancer. We want to know about certain mutations that are, um, informing for therapy selection and we didn't have a chance to measure them in the tumor tissue or there was too little tumor tissue. We could take a blood sample and send it to a company like Guardant or others and, um, ask are there genetic mutations in these genes. And the high, there's very high positive predictive value of those mutations found in the blood. Let's say if there's an EGFR mutation, we can avoid going back and redoing a biopsy in a patient or more quickly get the answer from the blood than the tissue. The negative predictive value isn't perfect. So we often go back to the tissue if it's a black and white kind off decision making. So for genotype directed, um, uh, selections, that's not a clinical scenario.
[00:11:16] Russ Altman: Yep.
[00:11:16] Ash Alizadeh: Um, and then for monitoring, we could also draw blood before surgery and then after a surgery. And send it to a company like Natera or others where, um, they make personalized panels for individual patients and monitor residual disease with much more sensitivity than this genotyping exercise that I mentioned, to say, is there residual disease in a way that the surgeon did their work, but unfortunately, there is residual disease that should make us think about adjuvant maneuvers. Or if there is not, maybe we shouldn't be using these toxic adjuvant therapies that we would otherwise. The way that generally works, Russ, is the companies receive the blood sample, they separate the blood sample into the cellular and acellular fraction, the plasma, the juice, that's on top of the cells. And that juice has a lot of proteins and other things but it has some DNA in it. Usually about a few thousand cells worth of DNA per milliliter, that is compared with a few million cells in the cellular fraction. So about a thousand fold difference in DNA, but even though there is a thousand fold less DNA in the juice than there is in the bottom, it's worth the squeeze because that juice, um, is where the molecules tend to hang out that come from these dying cells, not in the cells.
[00:12:36] Russ Altman: Right. Great. Now, so, thank you very much. We see the scenario here and you've mentioned a few companies, which means that has been reduced to practice and they're doing it. But I know that in your lab, you're pushing the frontier. So tell us what are the capabilities that are not yet present from these companies that you're, you and your colleagues are trying to create for a kind of the next generation of liquid biopsy.
[00:13:00] Ash Alizadeh: Yeah, so we have spent a lot of time on building technologies for genetic mutation-based detection, whether it's for early detection or for monitoring or minimal residual disease. And some of those methods have been licensed to companies that are using them and products on the marketplace already. Um, but the most recent, um, types of things we've been doing that I'm more excited about, is to go after the non-mutant molecules and see what we can get out of them. Um, so one, and I kind of alluded to this, was there are methods to use fragmentomics and methylation differences. So we've leveraged those methods with DNA methylation and fragmentomic differences to detect things with DNA. And then another group of studies that we've done has been focused on cell free RNA. So just like cells that when they die spill their DNA molecules, they also spill their RNA molecules. And as you know from the old early genomics days where we relied, we relied on techniques like DNA microarrays to measure RNA gene expression differences, you can make those same measurements in the blood plasma from cell free RNA and we have a paper accepted that will be coming out soon.
[00:14:10] Russ Altman: And just to remind people the RNA is interesting for many reasons, one of which is it shows what these cells are actually making at that moment or in that period of time. Whereas DNA is a capability that may or may not be exercised by the cell. The RNA means the cell is putting resources into creating that molecule, which I am imagining is extremely useful for you and your colleagues as physicians to know what's actually going on in the tumor.
[00:14:36] Ash Alizadeh: Yes, absolutely. And so you could see non genetic things through use of, um, RNA. You could do pharmacologic measurements of, you know, as we use drugs, often, uh, cells will adapt and become resistant. And much of that is not genetic. It's epigenetic and manifest in RNA so that there's some cool things you can do there. And then with RNA, you can overcome some of the technical limitations of DNA where the best sensitivity, uh, is achieved by a tumor informed strategy. So just this is an important concept where, um, imagine where's Waldo, that, that game. And it's much harder to play that game if you don't know what Waldo looks like, right?
[00:15:24] Russ Altman: I agree.
[00:15:25] Ash Alizadeh: And that same principle applies to cancer detection and cancer monitoring. Uh, so having an initial fingerprint of the tumor by sequencing either a tumor tissue or a blood sample before treatment. And using that fingerprint to monitor has a statistical, molecular, and biological advantage in being able to see Waldo in a much larger population of molecules than without prior knowledge. That's called the tumor informed strategy.
[00:15:55] Russ Altman: So the idea is by taking advantage, but you have a sample, you can use that to create a very specific marker that you have some confidence will be exactly right for detecting the recurrence of that cancer.
[00:16:09] Ash Alizadeh: That's right. So for example, the companies like Natera, Personalis, Foresight, and others, when they make personalized panels for monitoring of cancers, they take the original tumor, find the set of somatic alterations and then monitor those alterations in the blood as a personalized fingerprint, a very bespoke exercise.
[00:16:28] Russ Altman: Now, let me ask about that. I'm sorry to interrupt, but we all hear about cancers evolving over time, that they're not stable. And so, it, does that mean that these personalized tests only have a limited lifetime before they start not working because the cancer has kind of escaped. Waldo has put on a new shirt and therefore it's much harder to find Waldo now because it's now blue and white stripe and not red and white stripe.
[00:16:53] Ash Alizadeh: Yeah. So for the genetic alterations, what we know from multi region sequencing and diagnosis relapse studies is that even though there is evolution that's happened, if you find the set of genetic lesions that are ancestral and clonal, that they don't get erased. They, um, so Waldo is still going to be, I don't know, having a certain number of features, um, there may be new ones that emerge that if you hold on to the core set of Waldo-ness, that's unlikely to change.
[00:17:21] Russ Altman: So maybe his hat changes, but the red and white shirt is unlikely to go away.
[00:17:25] Ash Alizadeh: And so all of these outfits that I mentioned, leverage that, uh, knowledge to cherry pick the juiciest of the juicy. And then track those such that you guard against the clothing change that you mentioned. But they have this limitation in their tumor informed you need a baseline sample. And of course, this doesn't apply to the early detection problem. RNA helps you get around that to a certain extent. Methylation does too, but not quite to the, uh, robust analytical limits that we can get to with RNA.
[00:17:57] Russ Altman: And one quick question on the RNA and the methylation is, one of the beautiful things about DNA is, you know, we know it lasts for a hundred a million years. It's a very stable molecule, and so you can kind of knock it around and it will give you a good signal. Is that, are we able to do the same kinds of things with methylation or an RNA or do we have to kind of treat them a little bit more gently?
[00:18:16] Ash Alizadeh: Yeah. So for RNA, just so, you're right. DNA is much more bulletproof in that regard, but even while it's bulletproof, for the species that are cell free and in the blood, their half-life is quite short on the order of an hour or less.
[00:18:33] Russ Altman: Okay.
[00:18:33] Ash Alizadeh: So the molecule you're measuring in the, at any given time was shed half an hour, forty minutes, maybe an hour before. And that's what you're looking at. Most of the time because the enzymes will otherwise chew it up.
[00:18:45] Russ Altman: I see. So it is late breaking news.
[00:18:48] Ash Alizadeh: Late breaking news in that sense. RNA on the other hand is much more labile. You're right. And it is broken down. And we, you know, the reason our paper took so long and almost eight years of work for several graduate students and postdocs that have spent over that time to bring it to fruition, is to overcome those technical issues and to kind of deal with a lot of the pre analytic issues, the laboratory issues that arise. But despite that, you can still, worth a squeeze once you overcome those issues.
[00:19:19] Russ Altman: And is the methylation signal, which is a, for those who don't know, it's a modification to the DNA that's made during life usually, but can persist. So is that a pretty stable signal?
[00:19:30] Ash Alizadeh: Yes, it is. And in fact, you know, there are some very beautiful papers that have shown that there is a memory of the epigenetic change that can persist over long periods of time, whether it's cancer or non-cancer. That's a heritable, um, almost Lamarckian kind of concept of epigenetic change that the last two generations of cell division that we rely on for inferring that it came from a certain cell type.
[00:19:57] Russ Altman: This is The Future of Everything more with Ash Alizadeh next.
[00:20:15] Welcome back to The Future of Everything. I'm Russ Altman. I'm speaking with Ash Alizadeh at Stanford University. In the last segment, we talked about the basics of liquid biopsy, how it's already being used for many cancers. In this segment, we're going to do a couple of things. First, we're going to get a question from a listener, and then we're going to discuss the emerging area of immunotherapies and how liquid biopsies can play a role there.
[00:20:38] As I promised at the top of the episode, we actually have a question from a listener today, so we're going to listen to that question and then ask Ash to answer it. Here it comes.
[00:20:49] Daniel Kim: Hi, Russ. My name is Daniel Kim. I'm an assistant professor at UC Santa Cruz. Big fan of your Future of Everything podcast. So my lab has been developing an RNA liquid biopsy blood test for cancer early detection using the latest nanopore sequencing technology and machine learning to detect cancer at the earliest stages. And so people say that early detection is a holy grail, but my question for you is, what are your thoughts on the value of a pre cancer early detection blood test and subsequent cancer prevention and how achievable this might be in the next few years? Thanks so much.
[00:21:21] Russ Altman: So there we have it, Ash. A colleague from Santa Cruz wondering about detection of cancer in the first place.
[00:21:28] Ash Alizadeh: Yeah. I, as I said, very cool question. I'm a fan of Daniel's work as well. The, I guess the idea is, if we can detect a subset of early-stage cancers with liquid biopsies, can we do even better and go even to the pre-cancer stage? And I think it's a cool idea and we and others have tackled this question, uh, a little bit.
[00:21:51] Russ Altman: What does pre-cancer mean, just for those of us who don't think about it all the time?
[00:21:55] Ash Alizadeh: So when you go to your, for your colonoscopy, or when a woman goes for her mammogram, um, there are findings that can identify a precursor lesion that has some risk of becoming a cancer. Whether it's an adenoma that can become a carcinoma or whether it's in situ abnormalities, CCIS that can evolve into an infiltrating ductal carcinoma of the breast. Those are types of pre cancers and the same exists for almost every tissue in the body where there is a early change that has a certain risk and sometimes higher risk features, where we can kind of act on. I think it's a cool idea, especially if we have ways to do something about it. Um, meaning not just learning bad news about a pre cancer, but, um, what's the value of bad news if you can't do something about it?
[00:22:45] Um, when you have a colonoscopy and your, um, gastroenterologist, uh, finds a polyp that they deem suspicious and they take it. They've kind of solved the problem already. Um, and so with liquid biopsies, we have to have that same tool. I think it's a cool idea that Daniel has, but honestly, I've been, I think we should be able to walk before we run efficiently. And so far, the early detection performance of liquid biopsies, in my opinion, have been rather disappointing. Um, that the diagnostic performance for early-stage cancers has been quite low sensitivity, and we have yet to see any clinical studies where the intervention based on liquid biopsy results, improved cancer outcomes. Um, we know from studies done in England, for example, a huge study of a quarter million British women that use protein, old fashioned protein measurements for early detection of ovarian cancer, is that you can migrate the stages, you can detect them earlier. But unfortunately, even though you found them earlier, when you, the surgeon goes in to do their thing in a quarter million women, you don't see a benefit of that maneuver. There's not a survival advantage. So I don't know.
[00:24:00] Russ Altman: Gotcha.
[00:24:01] Ash Alizadeh: I think the idea of doing that same thing with liquid biopsies has not been done. And the diagnostic performance is, even though there's a lot of promise, I think much of it is sadly hype. Um, and I would not advise routine use of these tests, even though the, um, there's been strong pressure to for the US preventive Services Task Force and others to kind of get involved in trying to offer these tests to increase, um, use of early strategies to intervene and prevent the rising rate in certain cancers. So I think until we do that efficiently, going to the pre cancer stage is a little premature, but we need to do research.
[00:24:44] Russ Altman: Very good. And thank you to Daniel for our first ever listener question during an episode. So going back to what I wanted to talk about, uh, kind of building on this, is, um, we've all heard about the role of the immune system in, uh, fighting cancer. And as you know very well, there are these now immune therapies, which are sometimes miraculous in their ability to cure cancer, but have rocky, have a rocky road still because they don't always work. Is there an interaction between the liquid biopsy of technology and improving the efficacy and utility of these immune therapies?
[00:25:21] Ash Alizadeh: Yeah, so that's a cool idea. There definitely looks to be emerging data on this. Um, I'll give you an example. So I'm a lymphoma medical oncologist, and one of the most exciting emerging, and now not just emerging, kind of established therapy for patients with relapse of their aggressive lymphomas is to use cell therapy with engineered T cells. Well, they are called chimeric antigen receptor CAR T cells, and these have been dramatic in that they, um, take a subset of our patients who recur with their disease and cure a significant fraction of them.
[00:25:58] Russ Altman: And T cells are these immune cells that can be engineered to basically recognize the cancer of interest.
[00:26:04] Ash Alizadeh: Exactly. And the way that works, Russ, is a patient comes, we collect their blood, uh, we, uh, infect those blood cells with the T cells with a virus. Either a retro or lentivirus that has a transgene that will be expressed on those T cells to see a marker on the surface of the B cells. So we arm them with the knowledge that we engineer in and they go in and they, it's quite exciting when they first go in. They expand and attack and patients can get quite sick. But it's a one and done thing most of the time where this happens and the engineered cells go in and clean house and the patients are cured instead of many cycles of chemotherapy. So we did a study to say, okay, if patients are getting this therapy, can we watch not just the tumor, but the effector cells? And we built an assay called simultaneous tumor and effector profiling. We kind of cheated because this, this, uh, engineered T cell has some DNA that other cells don't have. And that's the retrovirus or lentivirus that we infect the cells with. So we look for that DNA in cell free and cellular.
[00:27:07] Russ Altman: It's opportunism, not cheating.
[00:27:08] Ash Alizadeh: It's opportunism. Um, so we could see that in quite dramatics fashion. Even after the cells leave the circulation to go into the tumor to do their job because they're turning over, they shed their DNA and you can watch what's going on. Um, and it seemed valuable in that we found that the expansion kinetics of that DNA, and the drop in the tumor DNA, work together in a way to best predict what's going to happen for patients. So that's,
[00:27:35] Russ Altman: So you need these T cells to proliferate. And if you don't see that you get nervous.
[00:27:39] Ash Alizadeh: That's right. Um, now, um, that is one such idea. And when we first saw that, of course, you know, here you were putting in millions of cells per kilogram. They all have this engineered thing. And so maybe it's cheating, there's just so much of it that you can see. Um, so we've looked at other types of immunity in this context, and there are, you know, aside from engineered T cells, there are other immunotherapies that where we use antibodies to boost the T cells to do the right thing. Uh, and some of those antibodies are called checkpoint blockade antibodies, antibodies that take key breaks on the system of T cells to not get too excited. Um, but cancers have figured out how to step on those brakes and keep the T cells from doing the right thing. And if we come in and we get in the way, they can do the right thing. So we've been studying that same effect in the context of immunotherapy, and we're seeing some interesting things. Um, and then often, uh, a lot of antitumor immunity relies on, uh, tumor infiltrating lymphocytes, cells that already live in the tumor. They're trying to do the right thing. Uh, sometimes they do the wrong thing.
[00:28:52] They actually stimulate the tumor. Some of them are good. Some of them are bad. Can we get a window into these cells and what they're doing through the course of therapy? I'll tell you an example from, um, a rare type of cancer called Hodgkin lymphoma. Has a long history, um, and a lot of that at Stanford, we cure a lot of patients with this disease with combination chemotherapy. And now we know that immunotherapy should be part of that up front, these checkpoint antibodies. So, um, the problem with this disease is when you look under the microscope, the pathologist can diagnose it, but the cells that define this, let's say a patient comes to me in my clinic, they have a football size mass in their chest, if the pathologist looks under the microscope, what fraction of the cells do you think in there are cancer? They're really a small number. They're about a percent.
[00:29:42] So this football, one percent of this football is cancer, the rest of it is inflammatory. And a lot of it is T cells. So we, um, there is a paper some years ago from a group in Belgium that was looking in pregnant women getting screened for fetal aneuploidies, the chromosome abnormalities I mentioned, Down syndrome, etcetera. And instead of finding Down syndrome, they found a bunch of stuff abnormal in the blood. And most of these patients that had abnormalities ended up having Hodgkin's disease. And we thought, well, wow, that's really high amounts of DNA. How could that be? So we studied a group of patients with Hodgkin's disease, and we saw that there is so much DNA that it's easier to diagnose this cancer and genetically profile this cancer from the blood then the tumor tissue. That there's way more, there's like an order of magnitude more DNA that's tumor derived percentage wise in the blood than the tissue.
[00:30:33] Russ Altman: Even though it's only one percent of the football, so to speak.
[00:30:37] Ash Alizadeh: Yes.
[00:30:38] Russ Altman: It's representation in the floating around DNA is much higher than one percent.
[00:30:43] Ash Alizadeh: Exactly. And we figured out part of the puzzle for why that may be. And we think that there is a special inflammatory environment that kind of helps move this DNA out. And we found some of the cells that do this. They're called dendritic cells that kind of decorate the lymphocytes, the cancer cells and kind of help exaggerate this effect in the blood. Um, and in doing that, we profiled a few hundred patients. We had a paper last year in Nature that we looked at this difference and we found that their disease can actually be defined into two genetic groups. And we submitted this from hundreds of patients from US and Europe, um, to the journal and they said, wow, this is super interesting, but wouldn't it be interesting to find other non-genetic differences between these that could help us identify how to target the differences between them. And I thought, oh, wow, this is checkmate. How are we going to do this? You know, it's so, uh, how are we going to, so, a couple of really talented folks, um, uh, the lead author, Stefan Alig, who's now a faculty member in Germany, and Mohammad Esfahani, who's now a faculty member here at Stanford, kind of took this checkmate problem and attacked the problem again using liquid biopsies. But, um, instead of using the DNA to find the mutation differences, they took these two genetic subgroups and say, can we find the non-genetic differences?
[00:32:02] Russ Altman: And just to be clear, it's useful because if there are two subpopulations of Hodgkin's disease, very often that means two different treatments, uh, optimal treatments. And so you'd like to figure out if you can tell the difference.
[00:32:15] Ash Alizadeh: Exactly. So what they did was they, um, they used a method that we developed, Mohammed helped develop the year before called EPIC-Seq. This is epigenetic inference of gene expression, uh, from DNA fragmentomics. And the basic idea is this is poor man's RNA from DNA, where we look at DNA fragmentation in particular parts of the genome, in particular, the beginnings of genes, transcription start sites. And we can look in other regions like, uh, enhancers and other regions to kind of infer expression from how the DNA is broken.
[00:32:55] Russ Altman: Right.
[00:32:55] Ash Alizadeh: So a gene that's off, the gene that's off, the DNA is generally very stereotyped in its size distribution of fragmentation profile, where a gene that's on and the trans, the, uh, chromatin has to remodel for RNA polymerase to come in and transcribe the gene, the DNA fragmentation is quite different. We can measure that very precisely in the entropy of the DNA molecules. And in doing that, we can use DNA as a poor man's RNA.
[00:33:23] Russ Altman: And the reason you're calling it a poor man's RNA is you're not actually measuring the RNA, but you're seeing these signals in the DNA that allow you to predict the RNA levels.
[00:33:32] Ash Alizadeh: Exactly. Um, and honestly, I was a little surprised because transcriptome wide, you can measure that with about a Pearson correlation of point nine, around thousands of
[00:33:42] Russ Altman: So these predictions are extremely good.
[00:33:45] Ash Alizadeh: They're pretty good. They're pretty good. Um, of course, as the genes become more lowly expressed, the problem becomes harder. But what these guys did was to take that insight, take the existing DNA we had from these cases and take the two subtypes that I mentioned genetically defined said, are there chromatin differences? And they found a few hundred genes that were different. And lo and behold, the results were kind of surprising. They weren't, for a large subset of genes, this was not about gene expression differences in the tumor cells, but from the tumor microenvironment. And I said, how could this be? You know, like, this can't be true. That we are seeing the tumor microenvironment, and it took these guys a bunch of time, almost a year, to prove to me that we were really seeing the tumor microenvironment by going to the tumor tissues by using separate methods, T cell receptor profiling. So that's quite exciting. We're seeing the tumor microenvironment. We're not only seeing genetic differences in the tumor cells that distinguish the subtypes, but differences in the tumor microenvironment that might be relevant.
[00:34:49] Russ Altman: Right. And by microenvironment, just to clarify, these are all of the kind of host derived, um, inflammatory and fibrous cells that create the tumor and that from which the cancer cells basically derive support and nourishment. And it turns out that you're seeing differences in these two different types and that raises the possibility of new treatments.
[00:35:12] Ash Alizadeh: That's right.
[00:35:13] Russ Altman: Again, not only go after the cancer, but maybe disrupt the environment that's supporting the cancer.
[00:35:20] Ash Alizadeh: Exactly. That's the idea. And we're hoping to do that because now immunotherapy is going to become a cornerstone of the frontline treatment of patients as we start removing elements of, you know, these toxic chemotherapies, um, and often we used to use them in combination with radiation. Can we start shaping that towards more directed strategies? Can we match the microenvironment with a particular therapy that might not be as toxic? And that's, so that's emerging work that I think liquid biopsies will be an important part of.
[00:35:52] Russ Altman: Thanks to Ash Alizadeh, that was the future of liquid biopsy.
[00:35:56] Thank you for tuning into this episode. Don't forget, we have zillions of episodes in our back catalog. You can listen to all of them if you want, and you'll be up to date on The Future of Everything. Please remember to hit follow in whatever app you're listening to, to make sure you're always alerted, usually on Fridays, to all the new episodes. You can connect with me on social media platforms like LinkedIn, Threads, Bluesky, and Mastodon. I'm @RBAltman on most of them, but @RussBAltman on Threads. You can also follow Stanford Engineering on social media, @StanfordENG, or @StanfordSchoolofEngineering.
