290 ‒ Liquid biopsies for early cancer detection, the role of epigenetics in aging, and the more

1. 0:00 – 0:08

   Intro

   1. PAPeter Attia0:00

      [upbeat music] Hey, everyone. Welcome to the Drive podcast. I'm your host, Peter Attia. [upbeat music]

2. 0:08 – 19:57

   Alex’s background in applying engineering to problems in medicine

   1. PAPeter Attia0:10

      Hey, Alex. Great to be sitting down with you here today. I kinda wish we were doing this in person. It's, uh, 'cause we haven't seen each other in person in a few months, and even that was sort of a chance meeting. So, um, I guess by way of background, you know, you and I go back over 20 years now. I guess it's 25 years that we both started med school together. [chuckles] It's hard to believe it's been that long, huh?

   2. AAAlex Aravanis0:34

      Yeah, I know. It s- seems like a million years ago, but, but also seems like yesterday. Uh, yeah, those were good times.

   3. PAPeter Attia0:40

      Alex, you know, y- y-- one of the things I remember when we first met was that we pretty much clicked over the fact that we were both engineers coming in and, and we had a good group of friends that I remember in medical school, and the one thing we had in common is not one of us was a pre-med. We were all kind of, you know, whatever the term was they used to describe as non-traditional path to medical school. Um, so let's talk a little bit about just, you know, briefly your, your background. You know, you, so you, you came in as a electrical engineer, and then you did a PhD in, in a, in a, in a lab of a, a very prominent scientist by the name of Dick Chen. Um, maybe tell folks a little bit about what you did in that work and what it was that got you, you know, excited enough about science to, to deviate off the, the traditional MD path.

   4. AAAlex Aravanis1:25

      Yeah. So, uh, so as you said, um, uh, my PhD was in electrical engineering. Um, and Stanford, you know, ha- has a cool configuration on the campus where the engineering school's literally a- across the street from the medical school. And so over time, I became more and more interested in, uh, applying, you know, signal processing techniques, circuit design, imaging, AI, things like that, but the problems in medicine that were more interesting to me than, uh, some of the, you know, t- traditional engineering products and, and things like that. And, uh, um, met a world-famous neuroscientist, name as you mentioned, uh, Dick Chen, uh, who's very interested in fundamental questions about the quantum unit of communication in the brain, which, which is, uh, the individual synaptic vesicle, and there was a question of just, you know, what did it look like and how did it operate. And, um, it was the beginning for me of just applying these engineering tools to really important questions in biology and helping answer them. Uh, that first story was a, was a great, um, article in Nature where we, you know, definitively answered the question of how that quantum is, uh, is transmitted between cells. Um, and then went on to do, uh, several other projects like that.

   5. PAPeter Attia2:37

      Can you say a little bit about that? How is that information transmitted?

   6. AAAlex Aravanis2:41

      It was really fun to come up with these problems with an engineering and, and communications background. Um, but you know, like if you look at a central neuron in the brain, um, and you look at the rate at which, uh, you know, information is transferred, uh, it seemed to be much faster than the number of synaptic vesicles in the terminal, right? So there was this, well, there's only 30 synaptic vesicles in the terminal by like an electron microscope, yet, you know, you're seeing, you know, hundreds of transmissions, you know, over a few seconds. So how is that possible? Um, and there were various theories. Um, there was an individual vesicle that was fusing and staying fused and pumping neurotransmitter through it without collapsing, and that's how you could get these, you know, so much more rapid puffs. And, and we, we came up with a cute term, which was called kiss and run-

   7. PAPeter Attia3:30

      [chuckles]

   8. AAAlex Aravanis3:30

      ... to, to explain that phenomenon, and it again helped answer this fundamental question of how did the brain get so many small neurons, yet able to transmit so much, uh, information per, per individual connection.

   9. PAPeter Attia3:45

      So Alex, if you think about all the things that you learned during your PhD, um, I mean, I guess one of the benefits of doing it where you did it in the lab you did it in was you overlapped with some other really thoughtful folks, in- including a previous guest on the podcast, Carl Desroche. Um, w- what do you think were the most important things you learned philosophically, um, not necessarily technically, that are serving you in the stuff we're gonna talk about today? So we're gonna talk today a lot about liquid biopsies. We're gonna talk a lot about epigenetics. Um, we're gonna talk a lot about certainly technologies that have made those things possible. Um, and when you think back to your background in double E, w- what were the transferable skills?

   10. AAAlex Aravanis4:35

       Y- so I think one of them is a sen- a saying in engineering, which is if, if you can't build it, you don't understand it, right? So simply understanding a description of something is not the same as you can build it up from scratch. And so you can't always do that in biology, but you can do experiments where you're testing the concept of can I really make it work? Um, and so I think that was an engineering concept that served me well a lot. Um, uh, y- you know, another, it's not exclusive to engineering, but was being very first principled, right? Um, you know, do we really understand how this works? Um, in that particular lab, um, there's a big emphasis on, you know, doing experiments where you always learn something, right? Um, where, you know, regardless of whether or not it confirmed or rejected your hypothesis, um, you learn something new about the system. Don't do experiments where, you know, you may just not learn anything. Um, and so that, that was a very, uh, kind of powerful way to think about things.

   11. PAPeter Attia5:35

       So we'll fast-forward a bit just for the sake of time. There's obviously an interesting detour where after I go off to residency and after you finish your PhD, we still find ourselves back together side by side-

   12. AAAlex Aravanis5:47

       Yep

   13. PAPeter Attia5:47

       ... in the same company for four years, um, which again, brought m- many funny stories, [chuckles] including my favorite is you and I getting lost in the middle of-

   14. AAAlex Aravanis5:59

       Yeah

   15. PAPeter Attia5:59

       ... Texas. Actually, not in the middle of Texas, but just outside of El Paso and nearly running out of gas. I mean, [chuckles] this was no cell signal. We, uh, we, we were, we were in trouble, but we somehow made it out of that one together.

   16. AAAlex Aravanis6:14

       Yes. Yes.

   17. PAPeter Attia6:14

       [chuckles]

   18. AAAlex Aravanis6:15

       Yeah, no, I remember that, that, uh, us Californians thought that there must be a Starbucks within, you know, ten, fifteen miles out in the middle of West Texas, and it turns out you can go hundreds of miles without a Starbucks. [chuckles]

   19. PAPeter Attia6:27

       That's right. Passing a gas station with an eighth of a tank saying, "We'll stop at the next one" can be a strategic error. [chuckles] Uh, there was also the time you bailed me out when I forgot my cufflinks because you had some dental floss. Do you remember that?

   20. AAAlex Aravanis6:44

       Oh.

   21. PAPeter Attia6:45

       I don't know if you remember that.

   22. AAAlex Aravanis6:45

       There you go.

   23. PAPeter Attia6:46

       Yeah.

   24. AAAlex Aravanis6:46

       Yeah.

   25. PAPeter Attia6:47

       You, you had some dental floss-

   26. AAAlex Aravanis6:48

       Total MacGyver move.

   27. PAPeter Attia6:49

       That's right. Yeah, total MacGyver move. But anyway, let's, let's fast-forward to all that stuff. So, um, I don't know what year it is. It's gotta be circa what? Twenty twelve. When do you end up at Illumina for the first time?

   28. AAAlex Aravanis7:03

       Uh, early twenty thirteen.

   29. PAPeter Attia7:05

       Okay. Talk to me about that role. What, what was it that you were recruited to Illumina to do? And maybe just tell folks who don't know what Illumina is a little bit about the company as well.

   30. AAAlex Aravanis7:16

       Yeah. So today, um, Illumina, uh, is the largest maker of DNA sequencing technologies, right? So when you hear about the human genome being sequenced, um, you know, uh, things like, you know, uh, expression data or any seq, um, most liquid biopsies, most, you know, tumor sequencing, finding, um, you know, genetic variants in, in kids with rare disease, most of that is done with, uh, Illumina technology. Um, so they also make the chemistries that process the DNA, the sequencers that, uh, generate that information, and also the software that helps, uh, analyze it. So, um, uh, really took that tool from a very niche research technology to a standard of care in medicine and, uh, you know, hundreds of thousands of publications and tremendously adv- has been advancing science. Yeah.
3. 19:57 – 24:05

   A primer on human genetics, and the history and current landscape of DNA sequencing

   1. PAPeter Attia19:58

      So let's talk a little bit about where we are today. To my recollection, the last time I looked, to sequence, uh, uh, to do a whole genome sequence today is on the order of $1,000, five hundred to a thousand. Is that about accurate?

   2. AAAlex Aravanis20:17

      Yeah. That's way too expensive, uh, Peter. Um, today, a couple hundred dollars. Yeah.

   3. PAPeter Attia20:23

      Wow, okay. So a couple hundred dollars today. And I, I feel like I looked at this on a graph a while ago, and it was one of the few things I noticed that was improving faster than Moore's law. Um, maybe tell folks what Moore's law is, why it's often talked about. I think everybody's heard of it. Um, and, and maybe talk about the, the step function that sort of out... It, it, it's basically, if I'm looking at it correctly, there were two Moore's laws, but there was something in between that sort of became even a bigger improvement. But, but maybe tell folks what, what, what Moore's law is, first of all.

   4. AAAlex Aravanis21:03

      Yeah, yeah. So it's, it's, it's not like a law, like the, a law of physics or something like that. But it w- became an industry tr- trend in microprocessors, and it ha-- and what it refers to is the density of, um, transistors, uh, you know, on a microchip, right? Um, and, um, the cost of the amount of computing power, um, per, per amount of transistors. And, um, that geometrically decreased, um, kind of in a steady way. Um, actually don't remember the exact number, if it's, like, doubling every, you know, two years or something like that. But there was a, a geometric factor to it that the industry followed for, for decades. Um, you know, it's not quite following that anymore. I mean, transistors are getting down to, like, you know, the, the atomic scale, but, um, you know, it went way faster than, than people had, had, uh, had envisioned.

   5. PAPeter Attia21:58

      Yeah, I mean, it's basically, it basically started in the late '60s and, as you said, it went until it hits the limits of atomic chemistry.

   6. AAAlex Aravanis22:06

      Yeah. And so that, that relentless push is, is what made, uh, you know, uh, you know, all the, you know, the whole software engineering high-tech industry possible.

   7. PAPeter Attia22:15

      So back to my question, which is if you just look at the cost of sequencing from 2000 till today, it's sort of like two curves, right? There's, like, the, the relentless curve that, you know, we-- gets to where we are in two thousand thirteen, but then there was another big drop in price that occurred after that. And, and I'm, I'm guessing that had to do with shotgun sequencing or the commercialization of it. I mean, not the concept of it, which already existed. But is that, does that, does that, does that sound right?

   8. AAAlex Aravanis22:46

      Yeah. So, um, so really starting, so when Illumina really started to deliver, um, uh, the higher throughput next-generation sequencings, it, it brought along a, a new, uh, a new faster curve, um, because of the miniaturizations, right? So this ability to sequence billions of fragments in a small area, um, and Illumina just continued to, and, you know, I, I was privileged to be a big part of this, uh, this effort, um, continuing to drive the density down, the speed of the chemistry up. Um, all the associated optics, uh, engineering software around it, um, you know, drove that much, much faster than Moore's law reduction in cost.

   9. PAPeter Attia23:26

      And so it, h- how many... Were, were other companies involved in the, uh, the culmination of next gen sequencing?

   10. AAAlex Aravanis23:33

       Yeah, many. Um, and some of them are still around. Um, uh, you know, none nearly as successful as Illumina, but, uh, you know, also some important players there. Yeah.

   11. PAPeter Attia23:44

       And today that's the industry standard. I assume there's no sequencing that's going on that, that isn't next gen.

   12. AAAlex Aravanis23:51

       No, the vast majority is, is next gen sequencing. Um, you know, there, there's niche applications, um, uh, where there's other approaches, but in, yeah, the ninety-nine percent of the data being generated is some version of next generation sequencing.

4. 24:05 – 44:14

   The advent and evolution of liquid biopsies for early detection of cancer

   1. PAPeter Attia24:06

      You mentioned a moment ago that part of the effort to bring you to Illumina was, um, you know, presumably based on not just your, um, your, your innate talents, but also the fact that you came from a somewhat clinical background as well. You're an MD and a PhD, and if their desire is to be able to branch out into clinical applications, that, that would make for a natural fit. So where in that journey did the idea of liquid biopsies come up and, you know, maybe talk a little bit about the history of, of one of the companies in that space that we're gonna talk about today.

   2. AAAlex Aravanis24:41

      Yeah, yeah. So to start with that, I should talk about first, um, tumor sequencing, right? Which, which predated liquid biopsy. And so, um- Uh, a couple companies, most notably Foundation Medicine, um, developed, uh, using Illumina technology, developed tumor sequencing, right? Um, so there had been some academic work, but they tried to develop it and were the first to do it successfully as a, as a clinical, uh, as a clinical product. And, um, so what you can imagine is, you know, there's these genes that are implicated in cancer that often get mutated. Um, they're, you know, knowing which mutations a tumor has, has big implications for prognosis, but also for treatment, right? So, um, over time, we have more and more targeted therapies where, you know, um, uh, you know, if your tumor has a very particular mutation, um, you know, it's more likely to respond to, to certain drugs that target that type of, um, tumor. Um, and at the time, uh, as, you know, more and more of these mutations were identified that could be important in the treatment of a tumor, it was becoming, um, impractical to, say, do a PCR test for every mutation, right? So imagine there's a hundred potential mutations you'd like to know about, uh, if a patient has in their, you know, in their tumor, in their lung cancer, um, you know, doing each of these individually, again, a lot of expensive, um, expense, a lot of false positives. And so, um, what companies like Foundation Med say, "Hey, why don't we just sequence all of those positions at once, given next-generation sequencing?" So, um, so they would make a panel to sequence, say, five hundred genes or a few hundred genes, the ones that are most important, uh, in, in, um, in most solid cancers, and then they would sequence them, and then in one test, they would see the vast majority of the potential, uh, mutations that could be relevant to treat- treatment for that, um, for that cancer patient. Um, and so that is still a, a very important tool in, in oncology today. Um, a large fraction of tumors are sequenced, and that's what a- allows people to get, you know, um, access to many types of drugs, right? Um, uh, you know, many of the targeted, um, uh, therapies for lung cancer, melanoma, or you hear about things like, you know, microsatellite instability or high mutational burden, that all comes from, you know, tumor sequencing. Um, now, um, in a, you know, so o- once that was established, then, uh, a few folks, most notably at, um, at Johns Hopkins, but also other places, started to ask the question, well, you know, could we, could we sequence the tumor from the blood, right? And, and you might say, "Well, hey, you have a tumor in your lung. Why would sequencing, you know, the blood be, be relevant to looking at the tumor?" Well, it, it turns out, um, there is tumor DNA in the blood and, you know, this is interesting. So in the late '40s, um, it was first identified that there was DNA in the blood outside of cells, so-called cell-free DNA. Um, and then in the '70s, it was noticed that, um, cancer patients had a lot of DNA, uh, outside their cells in the blood, and that some of this, you know, was likely, uh, from tumors, right? From the cancer itself, right? And, um, i- if you know anything about tumor biology, you know that cancer cells are constantly dying, right? So you think of cancers as growing very quickly, and that's true, but they actually, um, are dying at an incredible rate because it's disordered growth, right? So many of the cells that divide are, you know, have all kinds of genomic problems, so they die, or they're cut off from vasculature. So, um, the, the crazy thing about a tumor is, yes, it's, it's growing in, you know, fast if it's an aggressive tumor, but also the amount of cell death, um, within that tumor is just, uh, you know, i- is very high. And every time one of those cells die, um, some of the DNA has the potential to get into the bloodstream. And so it was this insight along with the tumor sequencing that said, you know, "Hey, what if we sequence this cell-free DNA? Could we end up sequencing some of the tumor DNA or the, the cancer cell DNA that's in circulation?" Um, and so, um, early, early results, particularly from, um, uh, this group at Johns Hopkins, um, began to show that indeed that was possible. Um, and, uh, and then a few companies, again, using Illumina technology, and then we started doing it at Illumina also, our own liquid biopsy, um, assays and tests and technologies, um, developed what, what became liquid biopsy. Um, in this context, it was for late-stage cancer, so it was for patients who diagnosed with a cancer. You wanted to know, did their tumor have mutations? And you could do it from the blood. Um, there was a big benefit, which was, you know, as you know, for, for lung cancer, taking a biopsy can be a very dangerous proposition, right? Um, you can cause a pneumothorax. You can, you know, you know, someone land someone in the ICU. You, you can have, you know, in rare cases, you know, um, uh, you know, can lead to death in that type of procedure. And so the ability to, you know, get the mutational profile from the blood, um, was really attractive. And so that started many companies, um, down the road of developing these liquid biopsies for late-stage cancers.

   3. PAPeter Attia30:13

      Let's talk about a couple things there. Uh, tell me the typical length of, uh, a cell-free DNA fragment. How many base pairs or what's the range?

   4. AAAlex Aravanis30:24

      It depends on the exact context, but around a hundred and sixty base pairs, right? So that's, you know, a hundred and sixty letters of the ATCG code. Um, and there's a very particular reason it's that length, which is that, um, again, if, if you, you know, pull the string on the sweater, you unwind the, um, uh, the chromosome, right? And you keep doing it till you get down to something around a hundred and sixty base pairs. Um, what you find is that the DNA, right, it's not just naked, it's wrapped around something called a nucleosome, which is, um, an octamer, uh, or eight of these histone, um, proteins in a, in a cube, and the DNA is wrapped around it twice. And that's the smallest unit- um, of chromatin, uh, you know, of the, you know, of this larger, uh, chromosome structure. And so the reason it's 160 bases is that's more or less the geometry of going around, um, twice. And so, um, DNA can be cleaved by enzymes in the blood, but that nucleosome protects, um, the DNA from being cut to anything smaller than about 160 base pairs.

   5. PAPeter Attia31:34

      And does that mean that the cell-free DNA that is, um, found in the blood is still wrapped around the nucleosome twice? Like, it's, it's still clinging to that, and that's what's protecting it from being cleaved any smaller?

   6. AAAlex Aravanis31:49

      That's right.

   7. PAPeter Attia31:50

      You mentioned that obviously the first application of this was presumably looking for ways to, you know, figure out what the mutation was of a pa- a person with late stage cancer without requiring a tissue biopsy. Presumably by this point, it was very easy to gather, you know, hundreds of 160 base pair fragments and use the same sort of mathematics to reassemble them based on the few overlaps to say, "This is the actual sequence," because presumably the genes are much longer than 160 base pairs that they're looking at.

   8. AAAlex Aravanis32:29

      That's right. So by this point, um, in, um, yeah, 20, you know, uh, 2014, 2015, um, the informatics was quite sophisticated. So, um, you know, you could take, um, a large number of DNA sequences from fragments and easily determine, you know, which gene, uh, it was associated with.

   9. PAPeter Attia32:52

      And at some point, I recall in here, and, and I had a discussion on the podcast maybe a year and a half ago, two years ago, with Max Dean, another one of our med school classmates-

   10. AAAlex Aravanis33:04

       Yeah

   11. PAPeter Attia33:04

       ... about looking at recurrences in patients who were clinically free of disease. So you took a patient who's had a resection, plus or minus some adjuvant chemotherapy, and to the naked eye and to the radiograph, they appear free of disease. And the question becomes, you know, is that cancer recurring? And the sooner we can find out, the better our chance at treating them systemically again, because it's a pretty well-established fact in oncology that the lower the burden of tumor, the better the response, the lower the mutations, the less escapes, et cetera. And so did that kind of become the next iteration of this technology, which was, if we know the sequence of the tumor, can we go fishing for that particular tumor in the, in the cell-free DNA?

   12. AAAlex Aravanis34:00

       Broadly speaking, there's kind of three applications for, for looking at tumor DNA in the blood. One is screening, which, which we'll talk about later, which is, you know, people who don't have cancer, or 99% who don't, and trying to find the individual who has cancer, an invasive cancer but doesn't know it. Um, there's this application of what we call therapy selection, which is you're a cancer patient trying to decide which, which targeted therapy would be best for you. And then you're, this other one you mentioned is a third application we call, um, often minimal residual disease. We're looking at, um, uh, monitoring your response, which is you're undergoing treatment, and you wanna know is the amount of tumor DNA in the blood, you know, undetectable? I- is it, you know, and also its velocity, is it changing? 'Cause as, as you mentioned, that could tell you, um, is your treatment working? The tumor DNA burden or load is going down. Um, is it undetectable and you're potentially cured, that there's no longer that, a source of tumor DNA in your body? Um, or, you know, is it present even after a, a, um, uh, a, you know, a treatment with intent to, to cure and that, and the presence of that tumor DNA still means basically, um, and, and we appreciate this now, um, unfortunately you have not been cured or that patient hasn't been cured 'cause there is some nidus of tissue somewhere that still harbors these mutations and therefore is the tumor, even if it's not detectable by any other means.

   13. PAPeter Attia35:32

       So at what point does this company called GRAIL that, uh, uh, we're gonna talk about, at, at what point does it come into existence, and what was the impetus and motivation for that as a distinct entity outside of Illumina?

   14. AAAlex Aravanis35:50

       So there were several, um, kind of technological and scientific insights that, that came together, uh, along with, as often is this case, um, some really, um, bold, um, entrepreneurs and, and investors. Um, so, um, you know, the, the use of this liquid biopsy technology in late stage cancers, um, you know, it was clearly possible to sequence tumors from the blood. Um, uh, and, and it was clearly actually the tumor DNA, and it was useful for, for cancer patients. Um, so we knew that, you know, there was tumor DNA. We knew it could be done. Um, but what the field didn't know is, you know, could you just see this in early stage cancers, right? You know, localized cancers that were small. Um, not a lot of data on that, but, you know, clearly there, there was the potential. Um, there was also a really interesting incidental, uh, s- set of findings in a completely different application called non-invasive prenatal testing. So, um, a- again, totally different application, but, um, uh, you know, it was discovered, um, you know, principally by a, um, a scientist in Hong Kong named Dennis Lo that, um- You could see fetal DNA in the blood, uh, or more specifically placental DNA, uh, in the blood, and it was also cell-free DNA. And, um, what he developed actually along with, um, uh, you know, one of our professors at, at Stanford, Steve Quake, was a technique to look for trisomies in the blood based on this placental or fetal DNA, and this is called non-invasive prenatal testing. And so what you do is you sequence the cell-free DNA fragments in, um, uh, in a pregnant woman. Um, you look at the DNA, and if you see extra DNA, for example, at, you know, the position of chromosome twenty-one, well, that indicates that there are, you know, there's tissues in, in the woman, you know, presumably the fetus or placenta that's giving off, um, extra chromosome twenty-one. And so this ended up being an incredibly, um, sensitive and specific way to test for, you know, uh, the presence of trisomies, you know, chromosome twenty-one, eighteen, thirteen, early, uh, in, in pregnancy. And, you know, it's, it's had a tremendous impact. Um, I, I was also involved in, in subsequent iterations of the test in the United States. It decreased amniocentesis by about eighty percent 'cause the test is so sensitive and specific, um, as a screen that, um, you know, many, many women have now, you know, not had to undergo, you know, amniocentesis and, and the risks around that. So, um, again, totally different application of cell-free DNA. Um, but what happened is, um, during the early commercialization of about the first few hundred tests, um, uh, the companies pioneering this, and one of them was a company called Veranada, uh, that, that Illumina acquired, began to see, um, in rare cases, very unusual, um, DNA patterns, um, where they-- it wasn't just a chromosome twenty-one or eighteen or thirteen, but you know, what, what's often called chromo, um, tripsis, um, which is, um, many, many, um, aberrations a-across, uh, the chromosomes, right? Um, and mo- the woman, um, the two women who really, um, did this anal-analys- analysis and, um, really brought both Illumina and the world's attention to it were, um, Meredith Hawks Miller, a pathologist and, um, uh, and lab director at, at this, uh, Illumina-owned company, Veranada, and, uh, another bioinformatics, um, scientist, Daria Chudova. And, um, what they showed is u-ultimately that, uh, these women actually had cancer. Um, and so they were getting-- They were young, you know, you know, women of, um, uh, childbearing age. Um, there was ultimately, they ultimately had healthy children. Um, but they had an invasive cancer, and it was being diagnosed in their cell-free DNA by this non-invasive prenatal test. Um, and, you know, as they began to show these patterns from people, it became, you know, clear that they were clearly cancer, right? So a-again, you know, um, you know, um, you know, uh, if you have many, many chromosomes that are abnormal, that-that's just not com-compatible with life, right? Or a fetus, right? Um, and so when you saw this just genome-wide, um, uh, chromosomal changes, uh, it was very clear that, um, you know, th-these were inc- you know, were incidentally finding cancer in these women. So anyway, what I wanted to-

   15. PAPeter Attia40:42

       But Alex, let's, let's, let's talk a little bit about that actually because-

   16. AAAlex Aravanis40:45

       Yeah

   17. PAPeter Attia40:45

       ... I wanna dig into that. It's so, it's so interesting. Um, so let's take a step back. So again, whenever you say we're sampling for cell-free DNA, we should all be keeping in the back of our mind we're looking for these teeny, tiny, little hundred and sixty base pair, base pair fragments wrapped around little nucleosomes. Now, when-- let's just go back to the initial use case around trisomy twenty-one. With a hundred and sixty base pairs, is that sufficient to identify any one chromosome? I mean, presumably you're also sampling maternal blood, so you know what the maternal chromosomes look like, and you're presumably juxtaposing those two as your control. Is that, I assume that that's part of it?

   18. AAAlex Aravanis41:31

       No, it, it's, uh, not quite. So it's all mixed together. So in a, in a, uh, in a pregnant woman's blood, uh, the maternal blood, it's a mixture, right? So you have cell-free DNA from, you know, the, the majority of the cell-free DNA is from her, um-

   19. PAPeter Attia41:45

       Is hers

   20. AAAlex Aravanis41:45

       ... her own cells and tissues.

   21. PAPeter Attia41:47

       Yep.

   22. AAAlex Aravanis41:47

       Um, and then you have, um, superimposed on that a bit of cell-free DNA from mostly the placenta. Um, and so what you're seeing is this mix of cell-free DNA. And so, um, uh, and then what you do is you sequence. There, there's different ways to do it, but the most common way is, you know, you, you do shotgun sequencing, and you sequence millions of, of these fragments, and every time you sequence a fragment, you place it in a chromosome based on its sequence. So you have, you know, your first fragment, and you say, "Hey, when I compare this to the draft human genome, this goes on chromosome two." And then you, you sequence your sec- you know, third fragment, and you say, "Hey, you know, this sequence looks like chromosome fourteen." And you keep kind of putting them in the chromosome buckets. And what you expect, you know, if, if every tissue in, i-is, is, um, has an even chromosome distribution, you know, or two chromosomes, is that that profile would be flat, and each bucket would be about the same level. Um, but what you see in, um, a woman carrying a, a, a fetus that has a trisomy, like chromosome twenty-one-

   23. PAPeter Attia42:54

       You'll, you'll see, you'll see fifty percent greater in the chromosome twenty-one bucket.

   24. AAAlex Aravanis42:59

       Yeah, yeah. You, you actually see more like five or ten percent, because again, remember, um, ninety percent of it might be f- maternal blood, right? So that's all gonna be even. But within the-

   25. PAPeter Attia43:08

       Yep

   26. AAAlex Aravanis43:08

       ... ten percent fetal, right, that's gonna be, um, you're gonna have an extra fifty percent. So, so the total might be an extra-

   27. PAPeter Attia43:14

       It's only five percent or whatever

   28. AAAlex Aravanis43:15

       ... five or ten percent. Um, but that's eas- very, but that's a whopping big signal and very easy to detect

   29. PAPeter Attia43:21

       Isn't, isn't it interesting? It just gives a sense of how large the numbers are if a 5% delta is an off-the-charts unmistakable increase in significance. I wanna make sure, again, people understand what you just said, 'cause it's, it's very important. Because the majority of the cell-free DNA belongs to the mother, and because the fetal/pesle- placental cell-free DNA is a trivial amount, even though by definition a trisomy means there is 50% more of one chromosome, you've gone from two to three copies, in the fully diluted sample, that might only translate to a few percent, but that's enough given the large numbers that you're testing to be a definitive statistically significant difference that triggers a positive test.

   30. AAAlex Aravanis44:12

       Yep. Uh, y- well put. Yes.

5. 44:14 – 51:05

   The role of cell-free DNA in cancer detection: how incidental findings in non-invasive prenatal testing led to the development of liquid biopsies

   1. PAPeter Attia44:14

      Alex, let's talk a little bit about some of the... I wanna come back to the story, 'cause this is clearly the beginning of the story. Um, but let, let's come back to just a couple other housekeeping items. You know, a moment ago, you, we talked about cell-free DNA in the context of tumor. Someone might be listening to us thinking, "Wait, guys, you just said that the majority of the cell-free DNA is from this mother. 99.9% of the time she doesn't have cancer. Where is that cell-free DNA coming from?"

   2. AAAlex Aravanis44:39

      Again, you know, uh, when cells, um, are destroyed, right, e- either through necrosis or apoptosis or just, you know, there's a lot of cell turnover, right, of cells that replicate, right, especially epithelial cells, blood cells, uh, and so on. Um, um, as those, again, as the kind of natural biochemistry destroys them, some of the DNA from the nucleus ends up in circulation, right, uh, again, in these, uh, where they're wrapped around these nucleosomes. So it's essentially, um, you know, cell death and cell turnover is the source of it. And since, you know, again, at any one time, there's, you know, millions of cells dying and being turned over, there's always, um, some, some base level cell-free DNA in the blood.

   3. PAPeter Attia45:23

      And again, I don't know if you've ever done the calculation. If not, I don't mean to put you on the spot, but do you have an approximate guess for how many base pairs of cell-free DNA are floating around your body or my body as we sit here right now?

   4. AAAlex Aravanis45:36

      So, uh, what I can say is i- if you took, uh, a 10 mil blood tube, right, which is a lot of what these, in these tests use, right, and you remove all the cellular DNA, remember, there's a, there's a ton of DNA in the cells in circulation

   5. PAPeter Attia45:48

      Sure, the white blood cells, the red blood cells, et cetera. Get, get rid of all that. Yep

   6. AAAlex Aravanis45:52

      It is huge, huge amount. Um, you know, you probably have on the order of a, a few thousand cells worth of, uh, of cell-free DNA in a 10 mil, um, blood tube, um, which, which isn't a lot, but, but, uh-

   7. PAPeter Attia46:04

      Well, h- hang on. That sound, that's, just to make sure I understand you, you're saying a few thousand cells worth, with-

   8. AAAlex Aravanis46:11

      Yes

   9. PAPeter Attia46:11

      ... where each cell would be three billion base pairs.

   10. AAAlex Aravanis46:14

       Yes. Yes.

   11. PAPeter Attia46:16

       Wow. I mean, on the one hand, it doesn't sound like a lot because there are billions of cells. On the other hand, it still sounds like a lot. I mean, that's, that's still a big computational problem.

   12. AAAlex Aravanis46:28

       Where it becomes challenging is, is when we get into early detection, right? Where if you think about it, um, for any position in the genome, right, you only have, you say, um, a few thousand representations of it because there's only a, a few thousand cells.

   13. PAPeter Attia46:45

       Mm-hmm.

   14. AAAlex Aravanis46:45

       And so that starts to limit your ability to detect, you know, events that occur at, you know, one in a million or one in 100,000.

   15. PAPeter Attia46:53

       So Alex, do you recall these, um, incident cases of the pregnant mothers who ultimately through this t- and again, I guess we should probably go back and re-explain that because, uh, it's such an important and profound discovery. There were a handful of cases where in the process of screening for trisomies, they're discovering not that the mother has additional chromosomes that can be attributed to the fetus, but that she has significant mutations across a number of genes that also are probably showing up in relatively small amounts because they're not in all of her cells. Is that, is that correct?

   16. AAAlex Aravanis47:41

       Yeah. Yeah. Uh, wh- which is y- you'd, you could, so you might expect a flat pattern, right?

   17. PAPeter Attia47:47

       Yep

   18. AAAlex Aravanis47:47

       In, in the majority of cases. Um, or you know, what, um, uh, when the fetus has a trisomy, you see these very well-known, um, accumulations, mostly in 21, but occasionally in 18 or 13.

   19. PAPeter Attia48:00

       Yep.

   20. AAAlex Aravanis48:01

       Um, and instead what you see is, um, just increases and decreases, monosomies and trisomies across many, many chromosomes, right? Which is just not compatible with, you know, uh, life even as a fetus. Um, and so, but there is a, um, a biology where you do see these, you know, tremendous changes in the chromosomes, and that's often in the case of cancer, um, many, many cancers.

   21. PAPeter Attia48:27

       And, and do, do you recall what those cancers turned out to be in those young women? I mean, I assume they were breast cancers, but they could've been lung cancers, anything?

   22. AAAlex Aravanis48:36

       Yeah, so Meredith and, and Daria, they published a paper in JAMA, which, um, for anyone interested, which details these, um, 10 or so cases, uh, and what happened in, in each of them. It was a mix. Um-

   23. PAPeter Attia48:48

       Hmm

   24. AAAlex Aravanis48:49

       ... I think there was a neuroendocrine, um, uterine, some GI cancers. Um, it, it was a smattering of, of different things.

   25. PAPeter Attia48:57

       And what was the approximate year of that? We'll try to find that paper and link to it in the show notes.

   26. AAAlex Aravanis49:01

       In 2015. It was 2015 in JAMA.

   27. PAPeter Attia49:04

       Got it. I mean, that seems unbelievable. Uh, of course, one doesn't know the contrapositive. One doesn't know how many women had cancer but weren't captured. But is it safe to assume that the 10 who were identified all had cancer?

   28. AAAlex Aravanis49:25

       Yes. Yes. So, so we-

   29. PAPeter Attia49:26

       So there were no false positives, we just don't know how many false negatives there were.

   30. AAAlex Aravanis49:30

       Right. Yeah. So i- this is one of the, the things that, uh, kind of contributed to the, to the, uh, evidence that cancer screening, uh, might be possible using cell-free DNA, which is these incidental findings. As I mentioned earlier, you know, we already knew that, yes, tumors do put cell-free DNA, um, into the bloodstream, but this was a profound demonstration that, um, [lips smack] uh, you know, in, in actual clinical practice you could find undiagnosed cancers in asymptomatic individuals, um, and that it was highly specific. M- meaning that when it, when it was found using this method, um, [lips smack] you know, it, it al- almost... Well, I think in those initial ones it was every case, but almost every case turned out to have cancer, right? Now, to your point, it's not a screening test, right? Because, you know, even in relatively healthy and, and, and, you know, w- women of childbearing age, um, you know, a population of 100,000, you know, uh, you know, you expect epidemiologically, you know, 10 times or, or so, or 50 times that number of cancers, right, you know, o- over a year or so. So, um, you know, clearly you're missing the majority of, of cancer, so it's not a screening test.
6. 51:05 – 56:48

   The development of a universal blood test for cancer detection and a discussion of specificity of tests

   1. PAPeter Attia51:05

      So what was the next step in the process of, um, systematically going after addressing this problem?

   2. AAAlex Aravanis51:13

      So, uh, myself and, and some other folks at Illumina, a- along with, uh, w- with the two, two scientists I mentioned, Meredith and, and Daria, um, and then also in particular the, the CMO at the time, um, [lips smack] uh, Rick Klausner, um, who had a very, um, long history in, in cancer research, um, and in cancer screening. He was the previous NCI director, uh, uh, under Bill Clinton. So that's the N- National Cancer Institute at the NIH, um, under Bill Clinton, and he was the CMO at Illumina t- at the time. And we started to talk more and more about, you know, um, [lips smack] you know, what would it take, uh, to develop or, you know, determine the feasibility of a universal blood test for cancer based on this cell-free DNA technology? Um, [lips smack] and, you know, being very first principle, I, I really asked the question, well, why is it in f- f- you know, 50 years of many companies and a tremendous amount of academic research, no one had ever developed a, a broad-based blood test for cancer? You know, not just many cancers, you know, let alone any cancer. And really the only example is, um, PSA, and again, the false positive rates there are, um, are so high that it- its benefit to harm has, has, you know, been questioned many times and, and that's why it's no longer, you know, um, you know, doesn't have a USPSTF, um, grade A or B anymore. Um, uh, and the fundamental reason is specificity. So, um, there's lots of things that are sensitive, meaning that there are proteins that accumulate, um, biochemistries, metabolites that go up in cancer. Um, but the problem is they go up in a lot of benign conditions, right? So y- you know, a big benign prostate spews out a lot of PSA. Um, [lips smack] uh, and pretty much every other protein or metabolite does that. And so what, you know, the biomarkers to date were all very sensitive, but all had false positive rates of, you know, say, 5 or 10%. And so if you're imagining screening, you know, the whole population, y- you can't be working up one of 10 people for a potential cancer. And so the key technological thing to solve was, well, how do you have something that has a 1% false positive rate or a half percent false positive rate? 'Cause that's what you need to get to if you wanna do broad-based cancer screening in relatively healthy, asymptomatic people. And this is why, um, we thought it might be possible with cell-free DNA because the tumor DNA could be more specific, right, than, um, proteins and other things that are, you know, are common in, in b- benign disease. And so that was the reason to believe. Um, the things we didn't know is, well, how much DNA does a early stage tumor pump out? If it doesn't pump out any, well, there's, there's nothing to detect. The other is the heterogeneity, right? Um, cancer is not like infectious disease or where, you know, you know, there, there's one very identifying antigen or, or sequence, right? Every tumor i- is, is truly unique, right? So even two lung cancers that are both the same histological subtype, right, they can share very few mutations or none, right? So you can have two squamous cell lung cancers that, you know, honestly don't have a single, um, shared mutation. So now you need to look at, you know, hundreds or thousands or even of millions of positions to see enough potential changes. And this is where, again, NGS, um, was a really good fit, which is how do you overcome the heterogeneity that you need to now look for a disease that isn't defined? I can't tell you these three mutations are the ones you need to find for this cancer. No, there's, you know, a huge, um, [lips smack] uh, set of different ones for every cancer. And then that got us thinking, well, look, i- in addition to sequencing many positions and sequencing very deeply and using cell-free DNA, um, we were gonna need to use AI or machine learning, um, because, um, we had to learn these complex associations and patterns that, um, you know, no human being could curate thousands and, of, um- You know, different mutational profiles and try to find the common signals and, and so on. And so what, what, what emerged over the course of a year is, um, look, this might be possible, but we're gonna have to enroll very large populations just to study and find the, the, the, the signals and develop the technology. And then we're gonna need, you know, very large studies to actually pr- you know, do interventions and prove, prove it clinically valid, that, that it actually works. Um, we're gonna have to use NGS and sequence, you know, broadly across the whole genome, uh, and only then might it be possible. Um, and so the company at the time, you know, uh, you know, decided, and this was a, a board-level decision, that, you know, ultimately, um, this made more sense as an independent company given, um, the amount of capital that was gonna be required, given the scientific and technical risk, um, given the kind of people that you would need to recruit, uh, who are passionate about this, that, um, it made sense to do as a, a separate company. And so the CEO at the time, Jay Flatley, in early 2016, you know, announced, uh, the founding of the company and then spinning it out of, uh, Illumina, and, and I was, um, I had the honor of being, uh, one of the co-founders of it.

7. 56:48 – 1:05:27

   Advancements in cell-free DNA analysis and development of a multi-cancer screening test at GRAIL

   1. PAPeter Attia56:48

      So let's go back to 2016. You guys are now setting up new shop, you've got this new company, it's called GRAIL. You've brought over some folks like you, um, from Illumina, and presumably you're now also recruiting. What is the sequence of the first two or three problems you immediately get to work on?

   2. AAAlex Aravanis57:10

      You know, I, I, so I wrote the, the, the, uh, the starting research and development plan. And, um, the way I wrote it was, um, you know, we needed to evaluate, um, every potential feature in cell-free DNA, me- meaning that, um, any known method of looking for cancer in cell-free DNA, we needed to evaluate. That if we were gonna do this, um, and recruit these, these cohorts and all these altruistic individuals, and we were gonna spend the money to do this, we needed to not look at just one method or someone's favorite method or whatever they, they thought might work, we needed to look at every single one. And so that's what we did. We developed an assay and software for mutations, for, and then a bunch of other things, chromosomal changes, changes in the fragment size, and, and many others. And we said, look, we're gonna test each one of these head-to-head, and we're gonna test them in combination, and we're gonna figure out the best way to do this. Um, and, um, you know, w- we even had a, a mantra that, that, um, uh, that, that Rick came up with that I thought was very helpful, which is, we're either gonna figure out how to do this or we're gonna prove it can't be done. Um, uh, and so I, I think that was very helpful in thinking about how to do these initial experiments. So it was a lot of building these assays. Um, and then, um, we needed a massive datasets to train the machine learning algorithm, so we had this study called the CCGA, the Circulating Cell-free Genome Atlas, where we recruited 15,000 individuals, um, with and without cancer, of every major cancer type, um, in most cases, you know, um, hundreds. Um, and then we did, uh, and then we tested all of these different, um, methods, um, the ones I mentioned and, and also importantly, a methylation-based assay. And we did blinded studies, uh, to compare them and see could any of them detect a large fraction of the cancers? Did any of them have the potential to do it at high specificity? Because that's what we would need if we were gonna deve- develop a, uh, a universal test for cancer that could be used in a broad population.

   3. PAPeter Attia59:19

      So let's kinda go back and talk about a few of those things there 'cause there was, there's a lot there. So you said upfront, "Look, we're going to make sure that any measurable property of cell-free DNA, we are measuring, we are quantifying it, we are noting it." And so [clears throat] we talked about some of them, right? So fragment length, that seems relatively fixed, but presumably at large enough dataset, uh, or sample size rather, you're gonna see some variation there. Does that matter? Um, the actual genetic sequence, of course, that's, th- that's your bread and butter, you, to be able to measure that. Um, you also mentioned, of course, something called methylation, which we haven't really talked about yet, so we should explain what that means. Um, w- were there any other properties besides fragment length, sequence, and methylation that I'm missing?

   4. AAAlex Aravanis1:00:16

      Yeah. There, there were, there were several others. One was chromosomal changes. So as w- we mentioned, in cancer, um, this, um-

   5. PAPeter Attia1:00:23

      Mm-hmm. Got it. Yep

   6. AAAlex Aravanis1:00:24

      ... the, the numbers of chromosomes, uh, often changes. So many cancers, and this is wild, will often double the number of chromosomes. So, so you can go from twenty three-

   7. PAPeter Attia1:00:34

      Wow

   8. AAAlex Aravanis1:00:34

      ... to, uh, you know, double or even triple the number. Um, but these chromosomes are not normal, right? S- so y- you'll often have arms or, or the structures of chromosomes will get rearranged. Um, uh, and so there's a way to look at that also in the cell-free DNA by, like as we mentioned in the noninvasive prenatal testing, where you look at the amount of DNA per chromosome or per part of chromosome. So we looked at, uh, what's called these, um, chromosomal abnormalities. Um, we also looked at cell-free RNA. So it turns out, um, there's also RNA from tumors in circulation.

   9. PAPeter Attia1:01:12

      Now, now how, how stable is that, Alex? I was under the impression that [clears throat] RNA wouldn't be terribly stable, uh, unlike DNA, which of course is double strand and quite stable. Um, s- so how, how, how do you capture cell-free RNA?

   10. AAAlex Aravanis1:01:29

       Yeah. So, so naked RNA is, is not very stable. Um, however, there's, um, there's proteins that if the RNA is bound to, um, and, and one type is called an Argonaute protein, if the RNA is bound to it, um, it is protected. There's also, um, uh-

   11. PAPeter Attia1:01:46

       And this is-- I assume this is typically messenger RNA that's been in the process of being transcribed, but somewhere along the way before translation occurs, the, there's the disruption to the cell that results in lysis or something, and so the, the, the, the... You're just basically getting the cell-free RNA because you happened to catch it at that point. It was, it was in a, it was a replicating cell or something? Or it was just translating protein or...?

   12. AAAlex Aravanis1:02:13

       Yeah. Or during apoptosis, it's somehow, during some kind of program cell death, it's, it's being digested or, or bound.

   13. PAPeter Attia1:02:21

       Mm.

   14. AAAlex Aravanis1:02:21

       Um, you know, um, you know, the, the amount relative to the amount of cell death is low, so presumably most of the RNA, it is destroyed, but a, you know, enough of it does get protected and bound to proteins. Um, you know, whether or not it's kind of cellular detritus or garbage or it's intentional, you know, is kind of a, a different question, but it is present. Um, there's also vesicular structures, so l- little bubbles of, of membrane that the RNA can be contained in. Um, uh, the most common one is referred to as an exosome, um, which are these little vesicles in, in circulation. So, um, in a variety of different ways, you can have messenger RNA and other types of RNA preserved outside of cells in, in circulation. Um, and so we, we looked at that also.

   15. PAPeter Attia1:03:13

       Wow. How long did it take to, to quantify all of these things? And presumably, I think you sort of alluded to this, but we're not just looking at any one of these things. You're also asking the question, can combinations of these factors add to the fidelity of the test, correct?

   16. AAAlex Aravanis1:03:32

       Yeah. So this initial research phase, um, took, uh, close to three years, um, you know, cost hundreds of millions of dollars. Um, we had to recruit, you know, um, the largest cohort ever for this type of study, the, the CCGA study as I, as I alluded to. Um, and there were different phases. There was a discovery and then multiple development and validation phases. Um, we had to make the world's best assays to look at each of these features. Um, and then we had to process all of those samples and then analyze them and, you know, we did it in a very rigorous way where the final testing was all done blinded and the analysis was all done blinded so we could be sure that, um, you know, the, the results, you know, um, were not biased. Uh, and then we compared them all, and we also compared them in combinations, and we used sophisticated m- machine learning approaches to really maximize the, the use of the, of each individual type of data from, from each, you know, whether or not it was mutations or chromosomal changes or methylation.

   17. PAPeter Attia1:04:38

       So you mentioned that the CCGA had fifteen thousand samples. How many of those samples were cancers versus controls? What was the distribution of those?

   18. AAAlex Aravanis1:04:48

       I believe it was, uh, it was about sixty percent cancer ve-versus, uh, versus con- versus controls, yeah, forty percent.

   19. PAPeter Attia1:04:57

       You sort of alluded to it, but just to be sure I understood, y- y- you're obviously starting first with, um, a biased set where you know what's happening, and then you're moving to a blinded, unbiased set for confirmation. Is that effectively the way you did it?

   20. AAAlex Aravanis1:05:10

       Yeah. Yeah. It's often referred to as, um, like a, um, a training set and a test set.

   21. PAPeter Attia1:05:16

       Yeah.

   22. AAAlex Aravanis1:05:16

       Yeah.

   23. PAPeter Attia1:05:17

       And so tell, tell us what emerged, Alex. What, you know, when it was all said and done, when you had first and foremost identified every single thing that was measurable and knowable... Oh,

8. 1:05:27 – 1:11:00

   DNA methylation explained

   1. PAPeter Attia1:05:27

      I-- sorry, before we do that, I keep taking us off methylation. Explain methylation, because that's, of the, of all the characteristics, that's the one I don't think we've covered yet.

   2. AAAlex Aravanis1:05:37

      Yeah. So, so DNA methylation is a chemical modification of the D- DNA. So in particular at the C in the ATCG code, the C stands for a cytosine, so that's a particular, um, nucleotide or, or base in, in DNA. Um, uh, the, the, um, mammalian biology can methylate. It means that it can add a methyl group, but a methyl group is just a, a single carbon atom, right, um, with, with three hydrogens and then bonded to that cytosine. Um, and so that's what DNA methylation is. So to say a cytosine is methylated, it means that it has that, um, you know, single, you know, methyl group, uh, bonded to it. Um, uh, turns out that there's about twenty-eight million positions in the human genome, um, that can be methylated. It, it usually occurs at, um, what's called CpG sites, which is, um, uh, if you, if you go along one strand of DNA, this is not pairing of the DNA, but one strand, um, where a C, um, or a, a G, uh, follows a C. So that's what a CpG is. It's a C with a phosphate bond to a G. Um, and so at those positions in the genome, there are enzymes that can methylate the cytosine and demethylate it. Um, and there's again about twenty million, um, of those sites, you know, out of the thirty, uh, out of the three billion, um, you know, overall, um, bases in the, in the human genome. Um, these chemicals, these chemical modifications are really important because they affect things like gene expression. Um, they're, um... It's, it's, it's one of the more important classes of something that's called epigenetics, which is, um, changes that are outside of the genetics or outside of the code itself. You know, um, as you know, the Uh, the DNA code is the same in most cells of the human body, but obviously the cells are quite different. So a T-cell is very different than a neuron. Um, and other than the T-cell receptor, all, all of the, the genes are the same, right? The code is the same. So why are the cells different? Well, it's the epigenetics. So, so things like which, um, parts of the gene are methylated or which ones are associated with, um, histones that are, um, kind of blocking access to the DNA, that that's what ultimately determines which genes are transcribed, um, which proteins are made, and why cells take on very different morphology and, and properties, right? So it's a, it's... The methylation is a very fundamental code, um, for controlling it. Um, so I call the, the epigenetics kind of the software of the genome, right? The, the genetic code is kind of the hardware, um, but how you use it, which gene you, genes you use when, what combination, that's really the epigenetics.

   3. PAPeter Attia1:08:33

      And what is the technological delta or difference in reading out the methylation sequence on those CPG sites relative to the ease with which you simply measure the base pair sequences? So right, you can measure C-G-A-T, C-C-A-T-G, et cetera, but then is it the same, in the same readout do you also acquire which of the Cs are methylated or are you doing a separate analysis?

   4. AAAlex Aravanis1:09:07

      Yeah. So, so, um, th- there's different technologies to do that. For cell-free DNA, um, where you usually want very accurate sequencing of billions of these or, you know, many hundreds of millions of these small fragments, um, the way it's done is, um, and this adds complexity to, to the, to the chemistry, is you pre-treat the DNA in a way that encodes the methylation status in the ATG sequence, and then you just use a sequencer that can only see ATCG. But because you've encoded the information, you can then de-convolute it and infer which sites were methylated. Um, uh, to, just to be a little more specific, um, there's chemicals that will, for example, um, deaminate, uh, a, a cytosine that's not methylated, and then that deaminated cytosine, um, effectively turns into a uracil, which is a fifth letter in RNA. And then when you copy the DNA and you amplify it prior to the sequencing, it amplifies as a T because a U, um, when it's copied by, um, our DNA polymerase, becomes a T. And then you end up with a sequence where you expect to see Cs and you see a T, and if you see a T there, then you know that, aha, um, this must have been an unmethylated site.

   5. PAPeter Attia1:10:26

      That came from a U and the U is an unmethylated C. Brilliant.

   6. AAAlex Aravanis1:10:30

      C.

   7. PAPeter Attia1:10:31

      Yep.

   8. AAAlex Aravanis1:10:31

      Right. And if the C, um, was not changed, then you say then that must have been a site that was methylated. Yeah.

   9. PAPeter Attia1:10:38

      Because you'll see Gs opposite them.

   10. AAAlex Aravanis1:10:41

       Right. Uh, well, a- and-

   11. PAPeter Attia1:10:42

       Oh, sorry. If the, if the C was methylated, you'll see the G opposite because you won't turn it to the uracil.

   12. AAAlex Aravanis1:10:47

       Right. Right.

   13. PAPeter Attia1:10:48

       Yep. Yep. Brilliant.

   14. AAAlex Aravanis1:10:49

       And so, um, uh, yeah, so that, that technique's called bisulfite sequencing. Um, there's other, there are other, there are other ways to do it, but that's the predominant method.

9. 1:11:00 – 1:17:22

   Optimizing cancer detection with methylation analysis of cfDNA in small blood samples

   1. PAPeter Attia1:11:00

      All right. So now back to the question I started to ask a minute ago, but then realized we hadn't talked about methylation. So you've, you've now me- you've come up with all these different things that you can do with this tiny amount of blood. Um, because again, you, you talk about 10 ml. You know, in the grand scheme of things, that's a really small amount of blood, right? That's two small tubes of blood. Very easy to do. Um, presumably there was an optimization problem in here where you min-max this thing and realize, well, look, this would be easy to do if we could take a liter of blood, but that's clinically impossible. Um, yeah, it would be nice to Theranos this, uh, quote-unquote, and do this with a finger stick of blood, but you're never gonna get the goods. So i- it-- did you sort of end up at 10 ml by, um, by acci- I mean, was it, was it just sort of an optimization problem that got you there, as the most blood we could take without being unreasonable, um, but yet still have high enough fidelity? And, and maybe asked another way, can you get better and better at doing this if, if, if, if you were taking eight tubes of blood instead of two?

   2. AAAlex Aravanis1:12:09

      Yeah. S- so, um, there's a couple considerations. One is the practical one, right? Y- you know, you need a format that's, that's, you know, to the extent your standard phlebotomy a- and, and standard volumes that don't require, um, that, you know, um, that are, that are below the volumes at which, you know, you could put someone in jeopardy. Uh, you know, that's a big practical issue. But it actually turned out that what ultimately limited the sensitivity of the, of the test, um, was the background biology.

   3. PAPeter Attia1:12:42

      Mm-hmm.

   4. AAAlex Aravanis1:12:42

      So for, for cancer screening, for, for broad-based cancer screening, more blood would actually not help you.

   5. PAPeter Attia1:12:48

      Interesting.

   6. AAAlex Aravanis1:12:48

      Um, now there's other applications for, uh, monitoring or, or the therapy selection where you're looking for a very particular target, um, in a, you know, in someone who already knows ca- has cancer and you know what kind of cancer, and there you could perf- improve your sensitivity. But just for, um, c- cancer screening, uh, y- you're, you're usually not limited by the amount of blood.

   7. PAPeter Attia1:13:12

      Wow.

   8. AAAlex Aravanis1:13:13

      Yeah.

   9. PAPeter Attia1:13:14

      A- and so did methylation turn out to be the most predictive element at giving you that very, very high specificity, or was it some combination of those measurable factors?

   10. AAAlex Aravanis1:13:28

       Yeah. So it was pretty unexpected. I w- I would say going into it, most people thought that, uh, the mutations were gonna be the most s- you know, the most sensitive method. Um- Some of us thought that, uh, the chromosomal changes were gonna be the most sensitive. I would say that the, the methylation signals were kind of the dark horse. Uh, there was just, um, let's say, uh, I had to, um, uh, you know, I had to fight several times to keep it in the running. [laughs] Um, but, uh, um, but again, w- we, we really took a let the data, um, tell us what the right thing to do is, not biases from other, you, you know, o- other experiments. Let's do this in, in a comprehensive, rigorous way. And in the end, um, the methylation performed by far the best. Um, so it was the most sensitive, so it detected the most cancers. Importantly, it was very specific, so it could actually, um, it actually had the potential and ultimately did get to less than 1%, um, false positive rate. And then the methylation had this other feature, um, which was very unique, which was that it could predict the type of cancer. What was the original, um, [lip smacks] um, what we call now the, um, the cancer site of origin. Um, [lip smacks] uh, you know, what, what organ or tissue did it originally or originate fro- from. Um, interestingly, adding additional, adding them all together didn't improve on the methylation, and I, I can explain why-

   11. PAPeter Attia1:14:57

       Hmm

   12. AAAlex Aravanis1:14:57

       ... now in hindsight. You might have thought, "Hey, m- more types of information and signal are better," but, uh, it, it actually didn't. So, so we ended up with one clear result that the methylation patterns in the cell-free DNA were the, you know, most useful and, and information, and adding other things didn't, was not gonna help the performance.

   13. PAPeter Attia1:15:18

       And why do you think that was? Because it is a little counterintuitive. Um, there are clearly examples I could probably think of where you can degrade the signal by adding more information, but I'm curious if you have a biologic, teleologic explanation for why one and only one of these metrics turned out to be the best, and any additional information only diluted the signal.

   14. AAAlex Aravanis1:15:44

       Yeah. So, so it comes down to, this is a good engineering principle, right? If you wanna improve your prediction, y- you need an additional signal that carries information and is independent from your initial signal. If it's totally correlated, then it doesn't actually add anything, right? So, um, let's take an analogy. Let, let's say you're on a freeway overpass and you're developing an image recognition for Fords, right? Um, [lip smacks] and you say, "Okay, what I'm gonna start initially with is, uh, an algorithm. It's gonna look for a blue oval, you know, with the letters F-O-R-D in it," right? Okay. Um, so that's pretty good. Now, now let's say you say, "Okay, I know that some Fords also have the, the number 150 on the side, you know, F-150, so I'm going to add that," right? Well, if you think about it, if your algorithm based on the blue oval is already pretty good, adding the 150 is not gonna help, right? Because whenever the 150 occurs, the blue oval is also al- always there.

   15. PAPeter Attia1:16:48

       [laughs]

   16. AAAlex Aravanis1:16:49

       Now, if, y- you know, if, if, if the blue oval wasn't always there or there were Fords that didn't have the blue oval, then, then some other signal could be helpful. And so that's kind of what, what ended up happening, is that the methylation signal was so much more prevalent and so much, uh, more, um, distorted in cancer that everything else, um, uh, didn't really add because, you know, um, any time you could see one of the others, you could also see many more abnormal methylation fragments.

10. 1:17:22 – 1:32:57

    The importance of understanding sensitivity, specificity, positive predictive value, and negative predictive value in cancer screening

    1. PAPeter Attia1:17:22

       Yeah. That, that, that's really fantastic. I guess I also want to, again, just go back and make sure people understand [lip smacks] the, kind of the, the mission statement you guys brought to this, which was high specificity is a must. Um, so you know, people have heard me on the podcast do this before, but just in case there are people who, um, haven't heard this example or forget it, I sometimes like to use the metal detector analogy in the airport to help explain sensitivity and specificity. So sensitivity is the ability of the metal detector to detect metal that should not go through. And let's be clear. It's not that people in the airports care if your phone is going through or your laptop or your watch or your belt. They care that you're bringing guns or explosives. That's why we have metal detectors. Or knives or things of that nature. [lip smacks] That's why the metal detector exists. So it has to be sensitive enough that no one carrying one of those things can get through. On the other hand, specificity would say, so if you're optimizing for sensitivity, you make it such that you will detect any metal that goes through that thing, and by definition, you're gonna be stopping a lot of people. You're gonna stop everybody from walking through. If their, if their zipper is made of metal, you'll stop them. So you-

    2. AAAlex Aravanis1:18:51

       Or a prosthetic or a big belt or, yeah, boots or, yeah.

    3. PAPeter Attia1:18:55

       Yeah. Anything.

    4. AAAlex Aravanis1:18:55

       Right.

    5. PAPeter Attia1:18:56

       You got a little metal on your glasses, you're gonna get stopped. So you have to dial the thing in a way so that you have some specificity to this test as well, which is I can't just stop everybody. In an ideal world, I kinda want everyone to make it through who's not carrying one of those really bad things, and we're defining bad thing by a certain quantity of metal. And therefore, your specificity is to kind of say, "I don't want my test to be triggered on good guys," right? "I want my test to be triggered on bad guys." Now, when you guys are designing a test like this, like the GRAIL test, I guess I should just go back and state that anybody who's ever been through two different- airports wearing the exact same, you know, clothing and realizes sometimes it triggers, sometimes it doesn't. What you realize is not every machine has the same setting, and that's because the airport, the people at TSA, they turn up or turn down the sensitivity, and that changes the specificity as well. How deliberately do you, when you're setting up this assay, have the capacity to dial up and down sensitivity and specificity? So while I understand your mandate was a very high sens- uh, specificity test, where was the, um, the, the control or manipulation of that system, if at all?

    6. AAAlex Aravanis1:20:20

       Yeah. Um, so there's a threshold in, in-- It's complex, but there's the-- conceptually there, there's a threshold inside the algorithm, right? So, so you can imagine, um, that, um, you know, after you, uh, you have this comprehensive, um, uh, map of all these different types of methylation changes that can occur in the fragments of hundreds of examples of every cancer type, uh, and then you compare it to all the methylation changes that can occur outside of cancer, which we haven't talked about, which is very important, right? So, you know, most of the methylation patterns are pretty similar in, in similar cell types across individuals. But there are changes that, that occur, that occur with age or ethnicity or environmental exposure, um, and, and, and so on. And, you know, what you'd like is those t- two populations to be completely different. Um, but it turns out there is some overlap, right? So, so there are fragments that occur in cancer that, you know, can occur outside of cancer. Um, and um, the algorithm in a very complex state space is, is trying to separate these populations. Um, and whether or not y- you're going to, um, uh, call something as a potential cancer and say a cancel signal is detected is whether or not you-- the algorithm thinks is it associated with this cancer group or is it associated with a non-cancer group? Um, but again, there's some overlap, um, you know, be- between these. And so where you set that overlap, like in the border between individuals who don't have cancer, but have for whatever reason an abnormal le- level of fragments that kind of look cancerous, um, you know, that will determine your specificity. So there is a dial to turn where you can increase the stringency, um, call fewer false positives, but then you will start to miss s- some of the true positives. Now, w- what was so great about methylation is that these populations were pretty well separated, better than anything the world had ever seen before, which is why you could get high specificity and still pretty good sensitivity. Um, but yes, the- there is some overlap, um, which means you have to make a trade-off and, and dial it in.

    7. PAPeter Attia1:22:45

       And inside the company, is there sort of a specific discussion around the trade-offs of it's better to have a false positive than have a false negative? Like we-- let's use the example you brought up earlier, right? So, so prostate-specific antigen, uh, is kind of the mirror image of this, right? It's a highly, highly sensitive test with very low specificity. It's obviously a protein, so it's a totally different type of assay, right? It's a, you know, it's a far cruder test, of course. Um, but the idea is, in theory, and of course, I could give you plenty of examples, um, someone with prostate cancer is going to have a high PSA. So you're not going to miss people with cancer. But as you pointed out earlier, you're going to be catching a lot of people who don't have cancer, and it's for that reason, as you said, that it is no-- there is no longer a formal recommendation around the use of PSA screening. It has now kind of been relegated to the just talk to your doctor about it. Um, and of course, um, you know, the thinking is, look, there are too many men that have undergone an unnecessary prostate biopsy on the basis of an elevated PSA that, you know, really should have been attributed to their BPH or prostatitis or something else. Uh, so notwithstanding the fact that we have far better ways to screen for, for prostate cancer today, you know, that's a test that is highly geared towards never missing a cancer. Um, in its current, uh, format, um, under low prevalence populations, which is effectively the population it's being designed for, right? This is designed as a screening tool. Um, it seems to have better predictive po-- negative predictive value than positive predictive value, correct?

    8. AAAlex Aravanis1:24:30

       Uh, it's pretty high in both because negative predictive value also is re- is related to, uh, is also related to, to prevalence. Um, but, um, but, y- you know, well, just to put some numbers out there, right? So in the CCGA study, but then importantly in an interventional study called Pathfinder, um, you know, the, a positive predictive value is around forty percent, right? Which-

    9. PAPeter Attia1:24:56

       Is-- That's, that's all stages?

    10. AAAlex Aravanis1:24:59

        Um, y- yeah. So that's a- all cancers, all stages, right?

    11. PAPeter Attia1:25:04

        Yep.

    12. AAAlex Aravanis1:25:04

        It's a population study. So it's what, it's whatever natural-

    13. PAPeter Attia1:25:07

        Whatever shows up. Yep

    14. AAAlex Aravanis1:25:08

        ... whatever natural, um, set of cancers and stages occur in, in that group, right? So that was about, um, sixty-five hundred individuals.

    15. PAPeter Attia1:25:17

        And do you, do you recall, Alex, what the prevalence was in that population? Was it a low-risk population?

    16. AAAlex Aravanis1:25:24

        Yeah. So it was, it was a mix of a, uh, a slightly elevated risk population and then a, um, average risk population. So ju- just, just in terms of risk, and I think you'll, you'll appreciate this, um, you know, I, I think of anyone over fifty as, as high risk, and, and that's where the majority of these studies are, are happening, right?

    17. PAPeter Attia1:25:43

        Yeah.

    18. AAAlex Aravanis1:25:43

        So, I mean, ju-- I mean, age is your single biggest risk, risk factor for, um- Uh, for, for, for cancer, right? Um, you know, you, you have like relative to the pop- the population over 50 is about a 10X increased risk, uh, re- relative to the population under 50.

    19. PAPeter Attia1:26:01

        And age, age 55 to 65 is the decade where cancer is the number one cause of death.

    20. AAAlex Aravanis1:26:08

        Yeah, sort of, i- i- it's, yeah, I would say in, in developed nations, I mean, th- that's, that's actually increasing, right? I mean, we're making such incredible progress on metabolic disease and cardiovascular disease. Um, so I mean, cancer in the developed world is predicted to become, surpass, um, [lips smack] you know, cardiovascular disease as the number one killer. Um, so, but, um, a- anyway, older populations are at, I, I wouldn't call them low risk, I'd call them, you know, kind of a- average risk for that age group, which is still relatively high for the overall population. But, but it, but it was a mixed, um, yeah, prevalence, um, you know, um, uh, a bit l- a bit less than 1%. Um, some of these studies do have a, a healthy bias vol- you know, a healthy volunteer bias to them.

    21. PAPeter Attia1:26:56

        That's interesting. So in a prevalence, in a, in a 6,500-person cohort with a prevalence of 1%, which is pretty low, the positive predictive value was 40%. Um-

    22. AAAlex Aravanis1:27:11

        Yeah, that's right.

    23. PAPeter Attia1:27:12

        What was the sensitivity for all stages then? It must have been 56. I'd have to, I mean, it's easy to calculate if I had my spreadsheet in front of me, but it's, it's gotta be 60% or higher sensitivity, and specificity has gotta be close, is, you know, over 99% at that point, right?

    24. AAAlex Aravanis1:27:30

        Yeah, yeah. Th- those are the, those are the rough numbers. Yeah, that's right. Um, so about half of, s- some of the important statistics there, right? So about, um, half of the cancers that manifested, um, over the lifetime of the study were detected, um, by the test. The test actually, um, doubled the number of cancers in that interventional study than were, were detected by standard of care screening alone. So in the, the interventional study, the pa- the Pathfinder study, patients or the enrollees were, uh, getting standard of care screening, right, you know, according to guidelines. So y- you know, mammography, um, [lips smack] uh, importantly cervical cancer screening, and then, uh, you know, colonoscopies or, or stool-based testing ba- based on guidelines, right? Um, and so a number of the cancers that the GRAIL Galleri test detected were also detected by standard of care, which, which you would expect. Um, [lips smack] but the total number of cancers found was, was about doubled, um, with the addition of the, uh, you know, of the Galleri test. Um, and that was predominantly cancers where there, there isn't a screening test, uh, for, um, [lips smack] which is going back to the positive predictive value. Just, you know, the, the positive predictive value of most screening tests is low single digits.

    25. PAPeter Attia1:28:49

        That's right.

    26. AAAlex Aravanis1:28:49

        So we're talking about, so, so-

    27. PAPeter Attia1:28:51

        Yeah

    28. AAAlex Aravanis1:28:51

        ... I mean, you probably have the experience more than I have, but I, many, many times, uh, a, a, a female colleague, friend, or someone's wife calls and said, "You know, I got a mammography and, um, uh, they found something. I'm gonna have to go for a follow-up and a biopsy and so on." And literally, you know, 19 times out of 20, it's a false positive, right?

    29. PAPeter Attia1:29:14

        Right.

    30. AAAlex Aravanis1:29:14

        Um, and so, you know, that's one where we've acce- we've accepted, for better or worse, a huge false positive rate, um, to catch, you know, uh, to catch some cancers, right? And, and that's why there's, there's a fair amount of debate around, uh, mammography. But again, that's a positive predictive value of about four and a half percent. So, so, you know, most people get, the vast majority of people who get an initial positive, they, they're not gonna end up having cancer, you know, and, you know, um, [lips smack] you know, but, you know, s- still potentially worth it. Um, now we're talking about something where we're approaching one or two, uh, positive tests will ultimately lead to a cancer diagnosis w- that's, that's potentially actionable. So it's, um... I think sometimes when people hear 40%, they say, "Gee, that means there's still, uh, um, a fair amount of, um, uh, people who are gonna get a, a, a positive test," meaning a cancer signal detected and ultimately not, but again, for a screening test, that's incredibly high yield. Yeah.
11. 1:32:57 – 1:41:05

    The performance of the GRAIL Galleri test and its ability to detect various types and stages of cancer

    1. PAPeter Attia1:32:57

       too. Um, let, let's talk a little bit about some of the really interesting stuff that, [clears throat] that, that emerged in the his- various histologies and the various stages. Um, and I've had some really interesting discussions with your colleagues. Uh, I guess just for the sake of completing the story, [clears throat] you're no longer a part of the company GRAIL. Um, so, uh, maybe just explain that so that we can kinda get back to the GRAIL stuff. But, but just so the people understand kind of your trajectory, um, yeah.

    2. AAAlex Aravanis1:33:27

       Yeah, no, we should do that. So, uh, yeah, so I was at Illumina, and then I helped spin off GRAIL as the, um, co-founder, led the R&D and, and clinical development. Uh, and then, um, I actually went back to Illumina as the chief technology officer. We're running all of the company's research and development, really, really fantastic, uh, fun job. Uh, subsequently, Illumina acquired, uh, GRAIL. Um, and so, uh, now a solely owned, um, subsidiary of Illumina. Uh, and then that was, uh, about, uh, almost three years ago. And then, uh, recently I, I left Illumina to, um, start a new company, uh, where I'm the... a really interesting biotech, uh, company that I'm the CEO of. Um, yeah, so no longer actively involved in, in either company. Um, have great relations with all my former colleagues and, uh, excited to see their progress. Um, I should also say that I, I am still a shareholder also of Illumina, just for full disclosure.

    3. PAPeter Attia1:34:26

       Yep. Thank you. Um, [clears throat] so kind of going back to, you know, you have a number of colleagues, as you said, who are, who are still at GRAIL who, who I've gotten to know. Um, and you know, one of the, one of the things that really intrigued me was, again, some of the histologic differences and the stage differences of cancer. So if you look at the, um, uh, the, the sort of opening data, a few things stood out, right? So there were certain histologies that if you took them all together by stage, um, didn't look as good as others. So for example, talk a little bit about prostate cancer detection using the, uh, Galleri test.

    4. AAAlex Aravanis1:35:13

       So I think what you're referring to is, um, you know, uh, there's a big, um, [lips smack] uh, uh, a very wide variety of, or of different performances in different cancers. So they're all highly specific, so very low false positive rate 'cause there's only one false positive rate for the whole test, w- which, which is probably worth us spending some time on later. Um, but for example, sensitivity to many GI cancers or certain histologies of, of lung cancer, um, the test is very good at detecting earlier stage, um, localized cancers. Um, particularly in prostate cancer and hormone receptor positive breast cancer-

    5. PAPeter Attia1:35:52

       Mm-hmm

    6. AAAlex Aravanis1:35:52

       ... um, you know, the detection rate is, is lower for, like, stage one cancers. Um, [lips smack] but I think the, the, but this gets to a very important issue, which is what is it that you want to detect, right? So do you want to detect everything that's called a cancer today or, or is what, what you want to detect is you want to detect cancers that are gonna grow and ultimately cause harm, right? So the weird thing about cancer screening in general is there's both over and under-diagnosis, right? So there's, um, most small breast cancers and most d- DCIS and most even small prostate cancers will never kill the patient or cause morbidity, but there is a small subset that will. Um, and so for mo- for those we, you know, have decided to, um, again, go for a trade-off where we'll often resect things and go through treatments just to make sure that smaller percentage is removed even though we're removing a ton of, um, other, you know, quote, "cancers" that, that are unlikely to ever, uh, proceed into anything dangerous. Um, on the flip side, 70% of people who die of cancer, they die from an unscreened cancer. So there's huge under-diagnosis, right? So again, look, we should, we should remember that. 70% of people who ultimately die, you know, o- of cancer on, on their death certificate, um, they die from a cancer where there was no established screening, uh, you know, um, you know, pr- prior to something like, uh, G- GRAIL's Galleri. So we have this weird mix of, there's a lot of cancers where we're probably, where we know we're over-diagnosing, um, but we're doing it for the right, you know, for a, a defensible trade-off, and then there's a huge number of cancer deaths occurring where, you know, there's zer- essentially zero di- diagnosis. Um, but back to the ones where there's, um, [lips smack] you know, uh, under-diagnosis, right? Um, y- you know, it gets back to what does it mean to have tumor DNA in your blood? Right? So the measuring and detecting a cancer from tumor DNA in your blood is a functional assay. So for, for, to get tumor DNA in your blood, you have to have enough cells, they have to be growing fast enough, dying fast enough, and have blood access. So those are the things that you require. Now, if you have a tumor that's small, encapsulated, not growing, well, guess what? It's not gonna have DNA in the blood. So unlike an imaging assay, which is anatomical, this is really a functional assay. You're, you're querying for whether or not there's a cancer that has the mass, the cell activity and death, uh, and access to the blood to get and manifest its DNA into the blood. So, so it's really stratifying cancers on whether or not they have the activities. Now, interestingly, this functional assay is very correlated with ultimate, um, mortality. So, um, there's a really nice set of data that, that the GRAIL put out where you look at Kaplan-Meier curves. So over the course of the CCGA s- study, which is now going out, I don't know, five-plus years, you can say, well, what do survival curves look like if you were positive, right, your test was detected, versus your test was, uh, negative, meaning your cancer was not detected by the GRAIL test, and there's a b- big difference. So basically, if your cancer was undetectable by the GRAIL test, um, you have, um, uh, a very, um, good outcome, right? Much, much better than the general population with that cancer. So this suggests two things. One is, um, you know, A, those cancers may not have actually been dangerous because, um, uh, there's, there's not a lot of mortality associated with them, right? And maybe that's also why they couldn't put their tumor DNA in the blood. The other is whatever the existing standard of care is, it's working well, right? Now, if you look at all the, the cancers in the Kaplan-Meier curve that were detected, um, they have a lot of mortality associated with them. And so w- what it's showing is that it's the dangerous cancers, the cancers that are accounting for the majority of mortality, those are the ones that the test is detecting. Um, and again, this biological rationale makes a lot of sense, which is, okay, a tumor that grows fast can get its DNA in the blood, well, that's probably also a dangerous tumor that is gonna become invasive and, and, and spread. So again, it's, it's a functional assay. So if your test is detected by one of these tests like the Galleri test, it's saying something about the tumor that is very profound, w- which is that it's active enough to get its signal into the blood, and it's, you know, very likely, if untreated, um, to ultimately, you know, be, be associated with morbidity and, and potentially mortality. Um, and so I think it's an open question of, you know, these, these, these tumors that aren't detectable and that are in, um, cancers we know there's a lot of indolent, uh, disease, um, what does it really mean that the test is low sensitivity for that, right?

12. 1:41:05 – 1:54:45

    Do early cancer detection methods, like liquid biopsies, translate to improvement in overall survival?

    1. PAPeter Attia1:41:06

       Yeah. I would say that when I went through these data, and I went through every single histology by stage, um, I did this exercise probably eighteen months ago, um, the one that stood out to me more than any other was the sensitivity and specificity discrepancy, well, I should say the sensitivity discrepancy between triple-negative breast cancer and hormone-positive breast cancer. You alluded to this, but I want to reiterate the point because I think within the same, quote, unquote, "disease of breast cancer," we clearly understand that there are three diseases. There's estrogen positive, there's HER2/neu positive, there's triple negative. Those are the defining features of three completely unrelated cancers, with the exception of the fact that they all originate from the same mammary gland. But that's about where the similarity ends. Their treatments are different, their prognoses are different, and to take the two most extreme examples, you take a woman who has triple positive breast cancer, i.e. it's estrogen receptor positive, progesterone positive, HER2/neu positive, you take a woman who has none of those receptors positive, the difference on the Galleri test performance on stage one and stage two, so this is cancers that have not even spread to lymph nodes, the, the, the hormone positives were about a twenty percent sensitivity for stage one, stage two, and the triple negative was seventy-five percent, uh, sensitivity for stage one, stage two. And so th- this underscores your point, which is the triple positive cancer is a much, much worse cancer, and that at stage one, stage two, you're detecting seventy-five percent sensitivity, um, portends a very bad prognosis. Now, I think the really important question here, and I, I mean, I believe that this is being asked, is does the ability to screen in this way lead to better outcomes? So, I mean, I will state my bias because I think it's important to put your biases out there, and I've stated it publicly many times, but I'll state it again. My bias is that yes, it will. My bias is that early detection leads to earlier treatment, and even if the treatments are identical to those that will be used in advanced cancers, the outcomes are better because of the lower rate of tumor burden. And by the way, I would point to two of the most common cancers as examples of that, which are breast and colorectal cancer, where the treatments are virtually indistinguishable in the adjuvant setting versus the metastatic setting, and yet the outcomes are profoundly different. In other words, when you take a patient with breast or colorectal cancer And you do a surgical resection, and they are a stage three or less, and you give them adjuvant therapy, they have far, far, far better survival than those patients who undergo a resection but have metastatic disease and receive the same adjuvant therapy. It's not even close. And so that's the reason that I argue that the sooner we know we have cancer and the sooner we can begin treatment, the better we are. But the skeptic will push back at me and say, "Peter, the only thing the GRAIL test is going to do is tell more people bad news." So we'll concede that people are going to get a better, more relevant diagnosis, that we will not be alerting them to cancers that are irrelevant and over-treating them, and we will alert them to negative or more harmful cancers, but it won't translate to a difference in survival. So what is your take on that, and how can that question be definitively answered?

    2. AAAlex Aravanis1:45:11

       So the-- It's a very important question, um, and over time, I, it, it will be definitively answered, so, um, we should talk about some of GRAIL's, uh, GRAIL studies and how they're, how they're going about it. Um, [lips smack] so, you know, the, the statistics are very profound, like you said. So most solid tumors, five-year survival when disease is localized, um, you know, hasn't spread to another organ, you know, seventy to eighty percent five-year survival. Um, you know, less than twenty for, for metastatic stage four disease. So it's, it's very-- that correlation of, um, [lips smack] you know, stage at diagnosis versus five-year survival is, is night and day. And obviously, you know, everyone would want them and their loved ones and most people in the, you know, in the, um, localized disease, um, um, category. Um, you know, now there's an academic question like you're saying, which is, okay, well, that's true, uh, but, you know, you know, does that still-- but does that really prove that if you find people at, at localized disease through this method, as opposed to all the variety of methods that happens today incidentally, um, [lips smack] that, uh, you will have the same outcome? And, and sure, I, I guess you could come up with some very, you know, theoretical possibility that somehow that won't, um, and but, um, that doesn't seem, uh, ve-very likely. And I think it gets to a fundamental question of, well, are, are we gonna wait, you know, decades to, to see that? And in the meantime, um, you know, uh, give up the possibility, which is probably likely that finding these cancers early and, and intervening early will change outcome. Um, so I'm, I'm all for, and, and I think everyone is, you know, bigger and more definitive studies over time. But the, the idea that we're never gonna do that study or just take kind of a nihilistic point of view that until it's done, uh, we're not gonna find cancers early and intervene, I, I just, uh, you know, I, I don't think it's conscionable to, to, to do that, especially when the, the false positive rate's low. Um, I think there's a few other ways to come up-- come at it, which is, I mean, um, you know, if, if what you said was, was really true, it-- once it's, you know... So, so take-- So I've, I've, um, I've met some of the folks and, and been called by them wh- the, the GRAIL test has found a positive. Um, uh, I can think of a former colleague and, and the test found an ovarian cancer. Um, uh, do you think when she went to her OBGYN and said, "Look, the, the test said that I have potentially an ovarian cancer," and they did an ultrasound, and they found something, um, that the, you know, OBGYN said, "You know what? Since this was found through a new method, let's not intervene." So, so the-- there's a malignancy, you know, um, likely, you know, you know, it is an ovarian cancer. Um, we know what the natural history is, but we're not gonna intervene. Um, similarly with cases of pancreatic cancer, head and neck, or things like that. So I, I don't understand the logic because today, people do show up, although often, um, [lips smack] it's not very often with early-stage versions of these disease, ovarian, pancreatic, head and neck and things, and we treat them. So why, why is it you wouldn't treat them if you could find them through this modality? Um, I, I just don't know of any GI surgeon who says, you know, um, you know, uh, "Well, you're one of the lucky people. We found your pancreatic cancer at stage one, too, but we're not gonna treat it, um, because there isn't definitive evidence over decades that, uh, mortality isn't, isn't better." So I, I get the academic point and, you know, GRAIL and, and others are investing tremendous amount to increase the, the data. But, um, uh, the, the idea that we have this technology, and we're gonna allow huge numbers of cancers to just progress to late stage before treating, I, I, I don't think that's the right balance of, uh, potential benefit versus, um, uh, you know, kind of burden of evidence. You know?

    3. PAPeter Attia1:49:21

       So, so is there now a prospective real-world trial ongoing in Europe, though, to actually begin-

    4. AAAlex Aravanis1:49:29

       There is. There is.

    5. PAPeter Attia1:49:29

       Yeah. So let's, let's talk a little bit about that.

    6. AAAlex Aravanis1:49:32

       Yeah. So, um, the NHS, um, has been piloting the GRAIL test, uh, in a population of about a hundred and forty thousand. Um, uh, the, um, they've, they've done, I think-- So, so it involves sequential testing, uh, I think at least two tests, and then they look at, um, outcomes. And it's a, it's an interventional study with return of results. Um, and they're looking for a really interesting endpoint here, right? So mortality takes time, right? So I mean, some cancers, I mean, to ultimately see whether or not, you know, getting diagnosed at a different stage and the intervention changes that, you know, that, that could take one or, or in some cases two decades. Um, but they came up with a really interesting surrogate endpoint, which is reduction in stage four cancers. So, so here's the logic-

    7. PAPeter Attia1:50:22

       Yeah

    8. AAAlex Aravanis1:50:22

       ...which, uh, it makes a lot of sense, which is If you deplete-- if, if people stop getting diagnosed with stage four and, and say a big reduction in stage three cancer, then doesn't it stand to reason that ultimately you will reduce mortality, right? So, so if you remove the end-stage version of cancer, which, you know, kills most people, um, and that you know that you have to pass through, most people don't die of stage two cancer. They, they die of, um, uh, if they were diagnosed with stage two, they die because it turned out it wasn't stage two and it, and it spread, right? So if you, if you, if you do a study and within a few years, um, when you're, when you're screening people, you-- there's no more, and let's take the extreme, stage four cancer, then you've stage shifted the population, right? Um, and you're kind of eliminating late stage metastatic cancer. You know, while we're waiting for that to read out, you know, my, my personal belief is, you know, the potential benefit of finding cancer i-i-is so significant, testing now, um, you know, uh, you know, for, for many patients makes sense. Uh, and then, um, I think this endpoint of stage four reduction-

    9. PAPeter Attia1:51:39

       Yeah, that's a clever, clever endpoint. Yeah. Uh, y-y-you know, one of the things that, uh, uh, I know that, that a lot of the, the, the folks who oppose cancer screening tend to, tend to, to cite is that a number of cancer screening studies do not find an improvement in all-cause mortality, even when there's a reduction in cancer-specific mortality. So, hey, we did this colonoscopy study, or we did this breast cancer, you know, screening study, and it, it indeed reduced, you know, breast cancer deaths, but it didn't actually translate to a difference in all-cause mortality. Um, I've explained this on a previous podcast, but it is worth for folks who didn't hear that to understand why. Um, to me, that's a very, very, um, [lips smack] oh, how can I say this charitably? Um, it's a very misguided, uh, view of the literature because what you fail to appreciate is those studies are never powered for all-cause mortality. And i-if you reduce breast cancer mortality by forty percent or thirty percent, that translates to a trivial reduction in all-cause mortality because breast cancer is still just one of fifty cancers. And even though it's a relatively prevalent cancer, over the period of time of a study, which is typically five to seven years, the actual number of women who were going to die of breast cancer is still relatively small compared to the number of women, period, who were going to die of anything. And I, in previous podcasts, have discussed that it's very difficult to get that detection, uh, within the margin of error. And so if you actually wanted to be able to see how that translates to a reduction in all-cause mortality, you would need to increase the size of these studies considerably, even though really what you're trying to do is detract-- detect a reduction in cancer-specific mortality. I say all of that to say that I think one of the interesting things about the NHS study is it is a pan-screening study, and to my knowledge, it's the first. In other words, it has the potential to detect many cancers, and therefore you have many shots on goal. Potentially, this could show a reduction in all-cause mortality and not just cancer-specific mortality. Again, I, I would have to see the power analysis, but I wonder if the investigators thought that far ahead. Do you know?

    10. AAAlex Aravanis1:54:08

        I mean, they're gonna follow these patients long term. Um, uh, so they will get, uh, the, the-- be able to have the data on, you know, mortality. Um, I don't know if it's powered for all-cause morta- uh, all-cause. I mean, it's-- I, I think that's unlikely just for, for the reasons you said, which is that the, the numbers would be really high. Um, I mean, you know, again, if you're powering it to see a, a reduction in stage four over a couple years, um, that, that may not be enough.

    11. PAPeter Attia1:54:41

        Yeah. Yeah. Interesting. Well, time will tell. Alex,

13. 1:54:45 – 2:01:58

    The role of epigenetics in aging

    1. PAPeter Attia1:54:45

       I wanna pivot, if we've got a few more minutes, to a topic that you and I spend a lot of time talking about these days. Um, and so by way of disclosure, you, you sort of noted that you've, you know, left, uh, Illumina somewhat recently. You've started another company. Uh, I'm involved in that company as both an investor and an advisor, and it's an incredibly fascinating, um, uh, subject. But, um, o-one of the things that, that we talk about a lot is the, going back to this, the role of the epigenome, right? So, uh, I think you did a great job explaining it and putting it in context, right? So we've got these three billion base pairs, and lo and behold, some twenty-eight million of them also happen to have a methyl group on their C. Um, I'll fill in a few more details that we didn't discuss on the podcast, but just to, to throw it out there. As a general rule, when we're born, we have kind of our max set of them, and as we age, we tend to lose them. So in a-- as, as a person ages, the number of those methylation sites goes down. You obviously explain most importantly what they do, what we believe their most important purpose is, which is to impact gene expression. It's worth also pointing out that there are many hallmarks of aging. There are many things that are really believed to be at the fundamental level that describes why you and I today look and function entirely different from the day we-- from the way we did when we met twenty-five years ago. We're just, we're, we're half the men we used to be. Um, I could make a Laplace Fourier joke there, but I will refrain. Um, so I guess the question is, Alex, where do you think methylation fits into the biology of aging? That's a, that's a macro question, but...

    2. AAAlex Aravanis1:56:52

       Yeah, yeah. So, so you talked about the hallmarks of aging. Um, uh, you know, and, uh, it, it was, uh, who's the author who, who... I think it was H- Han Raham, um, came up with that about 10 years ago, this hallmarks of aging. And he recently gave a talk where he talked about perhaps, uh, methylation is the hallmark of hallmarks of, of, of aging. Um, and w- what he's referring to is the, the mounting data that, um, the epigenetic changes are, are the most descriptive of aging, um, and are becoming more and more causally linked to, uh, to, uh, to, to, to aging events. Yeah.

    3. PAPeter Attia1:57:35

       There's lots of data that show that people of comparable age but different health status, for example, smokers versus non-smokers, people who exercise versus people who don't, people who are obese versus people who are not, can have very different methylation patterns. Um, there's also some data that look at centenarians, um, relative to non-centenarians. Um, and obviously that's a complicated analysis because by definition there's a difference in age. But you get a sense of different patterns of methylation. And clearly centenarians, we've established long ago, do not acquire their centenarian status by their behaviors, right? Just look at Charlie Munger and Henry Kissinger, uh, you know, two people who recently passed away at, at basically the age of 100, despite no evidence whatsoever that they did anything to take care of themselves. Um, so clearly their biology and their genes are very protective. Um, as you said, there are a bunch of these hallmarks, right? I, I think the original paper talked about nine, and that's been somewhat expanded. Um, but you, you, do-- you share that view, I suppose, that the epigenome sits at the top and that potentially it's the one that's impacting the others. So when we think about mitochondrial dysfunction, which no one would dispute, mine and yours are nowhere near as good as they were 25 years ago. Our nutrient sensing pathways, inflammation, all of these things are moving in the wrong direction as we age. How do you think those tie to methylation and to the epigenome and to gene expression by extension?

    4. AAAlex Aravanis1:59:23

       You know, I think maybe let's reduce it to, like, a kind of an engineering framework. So i- if we took, um, Peter's epigenome from 25 years ago when I first met you, right? And we knew for, you know, every cell type and every cell, um, what was the methylation status at all 28 million positions. W- we had recorded that, and we took yours, uh, today, where, um, most of those cells have deviated from that, and we could flip all those states back, right? We said, look, you know, um... And, and that's, that's kind of how I think about it, is the sites don't go away, just whether or not they have the methyl group or not, um, changes. Um, and some places gain it, some places lose it. Um, if we could flip all those back, would that force the cell to behave like it was 25 years ago? Um, express genes, uh, the fidelity with which it controlled those genes, uh, the interplay between them, would it, would it, um, would it be reprogrammed back to that state? Um, and so, you know, that I think is a really provocative hypothesis. We don't know that for sure, but there's more and more evidence that that might be possible. And so to, to me, that's the, the burning question is, now that we have the ability to characterize that and we know what it looks like, um, in a higher functioning state, which is, you know, correlates with youth, and we are gaining technologies to be able to modulate that, um, and actually change the epigenome, as opposed to modifying proteins or gene expressions, but actually go in and remethylate and demethylate certain sites, um, can we kind of reprogram things back to that earlier state? And, and if it is the root level at which things are controlled, will you then get, um, uh, you know, the, uh, all of the other features that, that the, the cell had and the organism had? So I think that's the-- that's a really exciting question to, to answer. Just 'cause if the answer is yes or even partially yes, then it gives us a really concrete way to go about this. Um, and so I, I think you, you alluded-- We talked about the hallmarks, and the hallmarks are complex and interrelated. What, what I like about the epigenome is we can read it out, and we're gaining the ability to modify it directly. So really, it's the m- more fun- the most fundamental level at which all of these other things are controlled. It, it gives us, and again, maybe back to the early discussion, a very straightforward engineering way to go about this.

14. 2:01:58 – 2:09:48

    How cell-free DNA methylation patterns can help identify a cancer’s tissue of origin

    1. PAPeter Attia2:01:58

       Let's talk a little bit about how that's done. A, a year ago, you were part of a pretty remarkable effort that culminated in a publication in Nature. Um, if, if I recall, it sequenced the entire human epigenome. So if we had the Human Genome Project 24 years ago, uh, roughly, we, we had the, uh, the, the Epigenome Project. Can you talk a little bit about that and, um, maybe explain technologically how that was done as well?

    2. AAAlex Aravanis2:02:28

       Yeah. So i- in the development of the, the GRAIL Galleri test, right, um- There was a key capability that, that we knew was gonna be important for a multi-cancer test, right? So, so very different than most cancer screening today, which is done, you know, one cancer at a time, right? Um, so if you have a blood test and it's gonna tell you there's a cancer signal s- present and, and this person should be worked up for cancer, you'd really like to know, well, where does that cancer likely reside, 'cause that's where you should start your workup, and you want it to be pretty accurate. So if the algorithm detects a cancer, um, and it's really a head and neck cancer, you'd like the test to also say it's, uh, it's likely head and neck and then do an endoscopy, uh, and not have to do lots of whole body imaging or, uh, you know, a whole body PET/CT or things like that. So we developed something called a, a cancer site of origin, and so today the test has that. If you get a signal detected, it also predicts where the cancer is, and it gives, like, a top two choice, top two choices, and, um, it's about 90% accurate, accurate in, in doing that. But how does that work? Um, and, you know, you know, the, the physicians and patients who have gotten that have described it as kind of magic, that it detects the cancer and predicts it. Um, and it's based on the methylation patterns, right? So methylation is what determines cell identity and cell state, right? So again, DNA code is more or less the same in your cells, but the methylation patterns are strikingly different. When a cell replicates, why does it continue to be the same type of cell? When an epithelial cell replicates, same DNA as a T cell or a heart cell, but it, but it doesn't become those, it stays. It's because the methylation pattern, those exact, uh, methylation states on the 28 million are also replicated. So just in the same way DNA has a way of replicating the code, there's an enzyme that looks and, and copies the pattern to, to, to, to the next cell. And so that exact code determines, you know, again, is it a colonic epithelial cell or a fallopian epithelial cell or, or, or what- whatever it is. And so we knew that the only way to make a, um, a predictor in the cell-free DNA is to have that atlas of all the different methylation patterns. And so with, uh, a collaborator, a guy named Yu- Yuval Dor at, at Jerusalem University, um, we laboriously got surgical remnants from, from healthy individuals. He developed protocols to isolate the individual cell types, uh, of, of most of the cells that get transformed in cancer. Uh, and then we got pure methylation patterns where we sequenced the whole-- like, like sequencing the whole genome, sequenced the whole methylome of all those cell types, and we published that a year ago as the first, uh, atlas of the human methylome in, in all of the major cell types. And so for the first time we could say, "Hey, this is the code which makes you, um, beta islet cell in the, in the pancreas that, that makes, uh, insulin versus something else." Um, you know, interestingly, you know, there's only one cell in the body where the insulin promoter is not methylated, right? And, and that is the beta islet cell. Every other single cell, that promoter is heavily methylated 'cause it shouldn't be making insulin. Um, and so it's those kinds of signals, um, that when you have the cell-free DNA and you look at the methylation pattern, allows the algorithm to predict, hey, this isn't just methylation signal that looks like cancer. The patterns and what's methylated and what's not methylated looks like, um, colorectal tissue or a, or a colorectal cancer, and that's how the algorithm does it. And so, um, this atlas again is, was a real breakthrough for, for diagnostics, and it made cancer site of origin useful. It's also being used for, for lots of MRD, um, uh, or those monitor, cancer monitoring tests too because it's so sensitive. Um, but it also brought up this interesting possibility, which is if you're gonna develop therapeutics or you want to, say, rejuvenate cells or, or repair them that, that have changed or become pathologic, what if you compare the methylation pattern in the good state versus the bad state? Does that then tell you the exact positions that need to be fixed? And then with another technology which can go and flip those states, um, will that, um, you know, reverse, uh, or kind of rejuvenate the, the cell to, to the, um, to the original or desired state?

    3. PAPeter Attia2:07:13

       So Alex, unlike the genome, which doesn't migrate so much as we age, I mean, obviously it accumulates mutations, but with enough people, I guess we can figure that out pretty quickly. Do you need longitudinal analysis of a given individual, i.e. within an individual, to really study the methylome? Do you need to be able to say, boy, in an ideal world, this is what Peter's epigenome looked like when he was one year, you know, at birth, one-year-old, two, three, four, fifty years old, so that you could also see not just how does the methylation site determine the tissue specificity or differentiation, um, but how is it changing with normal aging as well?

    4. AAAlex Aravanis2:08:03

       Yeah, yeah. Um, I think a lot of it is not individual s- specific. So I mean, I'll give, give you an example. So, um, d- done a fair amount of work in T cells, and if you look at, say, exhausted effector T cells versus naive memory cells where, you know, uh, younger individuals tend to have more of those and it gives them more reservoir to do things like fight disease, fight cancer, um, there's very distinct methylation changes. Certain genes get You know, uh, uh, methylated or demethylated. Um, and those changes seem to be, uh, again, very correlated with this, um, change in, um, T cell function. And so my guess is, my, my belief is that those represent fundamental changes as the T cell population gets aged, and you end up with more and more T cells that, relatively speaking, are useless.

    5. PAPeter Attia2:09:01

       Mm-hmm.

    6. AAAlex Aravanis2:09:02

       And so, you know, if you wanted to rejuvenate, um, the, the T cells, repairing those methylation states is something that would benefit, um, everyone. Now, th- there are definitely, um, a small percentage of methylation sites that, that are probably kind of, uh, you know, drifting or degrading, and those could be, um, individual, um, uh, specific to individuals. There's some gender-specific sites for sure. There's some ethnic ones. Um, but the major- the big, big changes, um, seem to happen more with loss of function, you know, big changes in age, um, that are probably common across, um, you know, individuals or, you know, in the case of cancer, we also have profound changes.

15. 2:09:48 – 2:21:30

    Cellular and epigenetic reprogramming and other exciting work in the field of aging

    1. PAPeter Attia2:09:48

       When you think about, um, this space, a term comes up. I mean, if, if folks have been kind of following this, they've probably heard of things called Yamanaka factors. Um, in fact, a Nobel Prize was awarded, uh, to Yamanaka for the discovery of these factors. Can you explain what they are and what role they play in everything you're discussing?

    2. AAAlex Aravanis2:10:15

       Yeah, yeah, yeah. So what, uh, what, what Yamanaka and colleagues discovered is that, um, if you take, uh, you know, fully differentiated cells, for, for example, fibroblasts, um, and you expose them to a s- particular cocktail of four transcription factors, um, that the cell reverts to a stem cell-like state, um, and these are called, um, induced pluripotent stem cells. So again, you, you subject, um, a differentiated cell that, you know, is a, kind of a, a mature cell of a particular type. I think most of their work was in fibroblasts. And the cell, when it's exposed to these, um, transcription factors, many, many-- A- and these transcription factors are, are kind of, uh, powerful ones at the top of the hierarchy. They unleash a huge number of, of changes in, um, gene expression. Genes get turned on, get turned off. And then ultimately, if you keep letting it going, you end up with something that, um, is a, is a, is a type of stem cell. And why this was so exciting is it gave the, the possibility to create, um, stem cells through a manufactured process. So, you know, as you know, there's a lot of controversy about getting stem cells from embryos, um, or other sources. Um, this, you know, created a way now to create stem cells and use them for medical research, um, you know, by just taking an individual's own cells and kind of dedifferentiating it back to a, back to a stem cell.

    3. PAPeter Attia2:11:49

       And how much did that alter, um, the phenotype of the cell itself? In other words, the fibroblast has a bunch of phenotypic properties. What are the properties of a stem cell, um, and how much of that is driven by the change in methylation? In other words, I'm trying to understand how these transcription factors are actually exerting their impact a- throughout this regression, for lack of a better word.

    4. AAAlex Aravanis2:12:20

       Yeah. So most of those, um, we ca-- So, like, we refer to, um, cell type specific features as somatic features, like, um, you know, a, a T cell receptor, right? That that's a feature of a T cell, or a, a dendrite or an axon would be for a neuron, or an L-type calcium channel for, for a, for, you know, a cardiac myocyte. So those are, you know, very cell type specific features. So if you turn on these Yamanaka factors and you go back to a pluripotent stem cell, um, you lose most of these, right? And, and the r- and that word pluripotent means the potential to become anything, um, at least in, you know, at least in, in theory. So, um, you lose most of these cell type specific features. So the, the use of the iPSCs is then to redifferentiate them, right? And, and that's what people have been, uh, attempting to do, and it opened up the ability to do that, which is you create this, um, stem cell that now potentially has the, the ability to be differentiated into something else. You give it a different cocktail, and you try to make it a neuron or, um, you know, a muscle cell, right? And then use that in a, in a tissue replacement therapy, and there's a lot of research on that and a lot of groups trying to do that. You also asked about what is the relationship between that and the epigenetics and methylation state. Um, that has not been well explored, and that's something that, um-

    5. PAPeter Attia2:13:45

       Hmm

    6. AAAlex Aravanis2:13:45

       ... I'm, I and others are excited to do because it could be that, um, the, it, it's, you're indirectly affecting the epigenome with these Yamanaka factors, and that if you translated that into an epigenetic programming protocol, you could have a lot more control over it. Um, because one of the, the challenges with the Yamanaka factors is, um, if you, if you do this for, for long enough, eventually the stem cell becomes something much more like a cancer cell, um, and, and just becomes kind of unregulated growth. And so, um, again, huge breakthrough in, in the, in learning about this kind of cell reprogramming and dedifferentiation, but our ability to, to use it in a practical way for, for tissue and cell replacements is not there. My, my hope is that by converting it to an epigenetic level, um, it'll be more tractable.

    7. PAPeter Attia2:14:44

       You mentioned that this is typically done with fibroblasts. I assume the experiment has been done where you sprinkle Yamanaka factors on cardiac myocytes, neurons, and things like that. Do they not regress all the way back to pluripotent stem cells?

    8. AAAlex Aravanis2:14:59

       Um, I think to varying extents. And so, um, uh, I mean, if you truly have a pluripotent stem cell, it, I, I guess in theory it shouldn't matter where it came from, right?

    9. PAPeter Attia2:15:10

       Yeah.

    10. AAAlex Aravanis2:15:10

        Because it's pluripotent, so with, um, developmental factors, right? So I mean, where did it-- where did your first neurons came, come from? You, you had a stem cell, and then, you know, in the, um, you know, in the embryo or the fetus, there were factors that then coaxed that stem cell to become, you know, these other types of cells and tissues, right? So if it's truly pluripotent, then, um, you should be able to do that. Now, I think you're getting at something which is different, which is called partial reprogramming. And, um, he and, and, and his, um, the people have followed his work have, have what, what they, what they're trying to do is things which is kind of s- stop halfway. So what if you took a, um, uh, you know, a heart cell or a T cell that's function-- you know, that's lost a lot of function, and you give it these Yamanaka factors, but, you know, you stop it before it really loses its cell identity. Will it have gained some properties of its higher functioning youthful state-

    11. PAPeter Attia2:16:12

        Mm-hmm

    12. AAAlex Aravanis2:16:12

        ... without having, having lost it? And so there, there's some provocative papers out there on this. Um, uh, there's a guy at, uh, the Salk, Juan Carlos de Mont- Mont- del Monte, who's, um, uh, done some work on this and some very provocative results in mice of doing these partial reprogramming protocols and, you know, rejuvenating, you know, again, it's mice, so all the usual caveats, but getting str- you know, very striking improvements in function, in eyesight, cognition, again, in these kind of m- you know, mouse met- metrics. So, um, certainly interest in, in trying to understand how that would, um, how that might be able to translate to humans. The, again, the worry there would be that if you don't control it, then you could make, you know, uh, essentially a tumor, right? So, um, you know, it's opened up that whole area of science that it's possible to, um, do these kinds of dramatic dedifferentiation, how to really harness that in a context of, you know, human rejuvenation. Uh, we don't know how to do that yet, but, but there's a lot of, uh, a lot of people trying to figure that out.

    13. PAPeter Attia2:17:25

        If, if you had to guess, with a little bit of optimism, but not like pie in the sky optimism, where do you think this field will be in a decade? Which, you know, there was a day when a decade sounded a long time away. It doesn't sound that long anymore. Decades seem to be going by quicker than I remember. Um-

    14. AAAlex Aravanis2:17:47

        Right

    15. PAPeter Attia2:17:47

        ... so it, it's, you know, it's gonna be a decade pretty soon. Um, but that's still a sizable amount of time for the field to progress. What do you realistically think can happen with respect to addressing the aging phenotype, uh, vis-a-vis some method of reversal of aging, some truly geroprotective intervention?

    16. AAAlex Aravanis2:18:16

        Yeah. Yeah. So I'm optimistic, and I'm a believer. Um, I think for specific organs and tissues and cell types, there will be treatments that rejuvenate them. So, you know, I, I don't-- It's hard to see in a decade that there's just a, you know, complete [laughs] rejuvenation of every single cell and tissue in, in a human. But, you know, joint tissues, um, you know, the retina, you know, um, you know, immune cells, um, we're learning so much about the biology, um, related to r- rejuvenation and healthier states of them. And then in combination with that, the tools to manipulate them, which is equally important. You, you could understand what the biology is but not have a way to intervene. The tools to go in and, um, edit these e- at a genomic level, to edit it at an epigenetic level to change the state, um, and the delivery technologies to, um, uh, to get them to very specific tissues and organs is also pr- progressing tremendously. So I, I definitely see a world in ten years from now where we may have rejuvenation therapies for, you know, osteoarthritis, um, rejuvenation for various retinopathies, where we can rejuvenate whole classes of immune cells that, um, make you more resistant to disease, more resistant to cancer. Um, uh, yeah, I think, I think we'll see things that, uh, will have real benefits in, in improving, uh, healthspan.

    17. PAPeter Attia2:19:54

        Alex, this is, uh, this is an area that, um, I think truly excites me more than anything else in all of biology, which is to say I don't think there's anything else in my professional life that grips my fascination more with this, uh, than this question, right? Namely, if you can revert the epigenome to a version that existed earlier, can you take the phenotype back with you? And that could be at the tissue level, as you say. Could I make my, my joints feel the way they did twenty-five years ago? Um, could it make, you know, my T cells function as they did twenty-five years ago? Uh, and obviously, one can extrapolate from this and think of the entire organism. Um, so anyway, I'm, I'm excited by the work that, that you and others in this field are doing, um, and grateful that you've, uh, you've taken the time to, to talk about something that's really no longer your main project, but something for which you provide, uh, probably as, as good a history of as anyone vis-a-vis the liquid biopsies, and then obviously a little bit of a glimpse into the problem that obsesses you today.

    18. AAAlex Aravanis2:21:07

        Awesome. Uh, well, fun chatting with you as always, Peter. Um, yeah, glad op-- glad to have the opportunity to, to d- dive in deep with this. There are, there aren't many places to do this. Thank you.

    19. PAPeter Attia2:21:19

        Yeah. Thanks, Alex. [outro music]

Episode duration: 2:21:30