The Peter Attia Drive - #267 ‒ The latest in cancer therapeutics, diagnostics, and early detection

00:00:00.000 Hey, everyone. Welcome to the drive podcast. I'm your host, Peter Atiyah. This podcast,

00:00:16.580 my website, and my weekly newsletter all focus on the goal of translating the science of longevity

00:00:21.580 into something accessible for everyone. Our goal is to provide the best content in health and

00:00:26.780 wellness, and we've established a great team of analysts to make this happen. It is extremely

00:00:31.720 important to me to provide all of this content without relying on paid ads to do this. Our work

00:00:37.000 is made entirely possible by our members. And in return, we offer exclusive member only content

00:00:42.760 and benefits above and beyond what is available for free. If you want to take your knowledge of

00:00:47.980 this space to the next level, it's our goal to ensure members get back much more than the price

00:00:53.260 of the subscription. If you want to learn more about the benefits of our premium membership,

00:00:58.080 head over to peteratiyahmd.com forward slash subscribe. My returning guest this week is Dr.

00:01:06.540 Keith Flaherty. Keith was a previous guest on episode number 62 of the drive way back in July of 2019.

00:01:14.380 Keith is currently the director of clinical research at the Massachusetts General Hospital

00:01:18.740 Cancer Center and a professor of medicine at Harvard Medical School. And he serves as the

00:01:23.760 editor-in-chief of clinical cancer research, a very prestigious journal. His research focuses

00:01:29.320 on understanding novel, molecularly targeted therapies in cancer. Within this field, his focus

00:01:35.140 has been on the development of response and predictive biomarkers to define the mechanism of action

00:01:41.100 and resistance of novel therapies, as well as to identify the optimal target populations.

00:01:45.920 In this episode, we start by looking at some of the statistics around the prevalence of cancer as

00:01:50.720 we age. This really highlights the importance of this topic. And although we don't spend a lot of

00:01:54.620 time on it, because I think intuitively people understand that, I do think it's important for

00:01:58.280 people to understand progress. We then, of course, shift to what has been done and what has not been

00:02:04.060 done over the past several decades. And there have been some very notable improvements in cancer

00:02:09.440 therapy over the last 10 years, which we highlight. From there, we shift our focus to looking at what is on the

00:02:15.040 horizon and what the future of cancer therapeutic holds, both in the short term and in the long term.

00:02:20.060 And I think that even within a five-year period, there are some incredibly exciting things that

00:02:25.240 look to build on the successes of the past decade. We also talk about liquid biopsies, which, of course,

00:02:31.600 play a very important role in early diagnosis of cancer. And we talk about the state-of-the-art

00:02:37.800 today, but again, what we think it's going to be in the future. And this is something that you've

00:02:43.240 probably heard me talk about in the past. Liquid biopsies have the potential to diagnose cancer

00:02:48.560 from a simple vial of blood, where they not only can determine if a cancer is present in an early

00:02:53.940 stage, but also identify the possible tissue of origin. And a lot has changed since Keith and I

00:02:59.120 initially spoke over four years ago, which is why I thought it made sense to have him back and talk

00:03:03.840 about these things again. And I will say this conversation was illuminating to me, and it certainly

00:03:08.500 won't disappoint those of you who are interested in cancer. So without further delay, please enjoy

00:03:13.140 my conversation with Dr. Keith Flaherty.

00:03:20.320 Hey, Keith, great to be back with you again. Hard to believe it was almost exactly four years ago that

00:03:26.260 we sat down in Boston to do what will be part one of this discussion. But, you know, we've got a lot

00:03:32.800 more listeners now, and it's not like some of that content isn't still relevant today. So we'll

00:03:38.320 probably talk a little bit about some of the things we spoke about then, but there are a number of

00:03:43.100 things that I'm excited to discuss with you that we haven't talked about, and I suspect that will make

00:03:47.820 up the lion's share of our discussion. So thanks again for making time.

00:03:51.120 Yeah, thanks, Peter. It's a great pleasure to talk with you again. And yeah, you're right. Four years,

00:03:56.120 a lot has happened. You know, in therapeutic development, maybe you could have said four years ago that

00:04:01.580 some of the things that have played out would have played out. But on the diagnostic side,

00:04:06.040 that's, I think that's probably where four years ago I was, I didn't quite have the crystal ball

00:04:11.400 vision as to how things would develop there. So, and of course, those two areas are like tightly

00:04:16.560 related in oncology. So excited to dig back in. Let's just start, maybe give folks a short background,

00:04:22.600 a little bit about what you're doing and why it is that you're, at least in my view, in a great

00:04:26.960 position to talk about cancer in a way that is more than just an inch wide and a mile deep, which is

00:04:33.800 the general nature of the field, but several inches wide maybe and several, you know, in a mile deep.

00:04:39.780 Like, what is it about your background? Right. I'll try to hit some highlights here that make

00:04:43.640 that point. So my medical oncologist, I've been in the field now for 23 years, which is a relevant

00:04:48.960 number because of the fact that the first translation of molecular insights, you know, specifically genetic

00:04:54.260 insights into cancer biology, you know, really became therapy starting that year. That's when

00:04:59.540 Matt Niberg-Lievek was first in patients and was kind of a revolutionary moment. My career started

00:05:05.720 right then and there. Myself interested as a medical oncologist in trying to translate science to

00:05:10.140 medicine in very much that way, like taking insights in terms of the genetic makeup, like the mutations

00:05:15.740 and then sort of mutational architecture, if you will, of cancers and trying to translate that

00:05:20.160 understanding into therapy. So I've been doing that for 23 years. Like anybody in the academic

00:05:24.820 medical world and oncology, I had focused on specific cancer types of melanoma and kidney cancer. And both

00:05:31.020 of those I chose because of the molecular insights that existed at the time that felt like they were

00:05:35.580 at least beginning to be ripe for translation. So I did that work for about a decade at University of

00:05:40.700 Pennsylvania, moved to Mass General, part of the Harvard University umbrella, Harvard Medical School,

00:05:46.680 to build a clinical program focused on therapeutic development much more broadly across cancer.

00:05:52.760 And then I think as we'll talk about this interplay between therapeutics and molecular

00:05:57.800 understanding and ultimately diagnostics, basically built a translational research group surrounding

00:06:03.120 therapeutic development, what I refer to as bedside to bench translational research to sort of

00:06:07.100 understand mechanisms of action, mechanisms of resistance. In other words, when drugs work,

00:06:11.260 why? And if they don't work, you know, why not? And use those insights to try to kind of accelerate or drive

00:06:16.800 the whole process forward. And then I'll just throw in also that over the past 10 years, I've co-founded

00:06:22.280 now a handful, a little bit more than a handful of biotech companies with Loxo Oncology being the first

00:06:27.160 10 years ago when I was acquired four years ago and Scorpion Therapeutics being the most recent. And I'm actually

00:06:32.560 sitting in the offices of Scorpion Therapeutics at the moment. And so through those channels, I guess I would say

00:06:39.400 it's my job to keep, you know, a steady eye on new therapeutic concepts that could be ready for

00:06:46.620 prime time to move forward. And then again, trying to translate micro-understanding into tools that we

00:06:51.780 can actually use for real patients, aka diagnostics. And Keith, I think it's always worth repeating to

00:06:56.140 folks what it is about cancer that's, as far as the big four chronic diseases, I call them the four

00:07:01.420 horsemen in my book. There's something about cancer that's particularly damning, which is when you look at

00:07:08.440 the other two chronic diseases that are huge killers, which are cardiovascular disease and

00:07:13.920 neurodegenerative disease, they increase in their severity exponentially as you age. And they don't

00:07:21.700 really become a dominant source of mortality until people are in the seventh and eighth decade of life.

00:07:28.900 And that's not true for cancer. In fact, I have a little table in front of me that I had one of our

00:07:32.360 analysts pull together that I think is just, we'll put it in the show notes. It's very powerful,

00:07:35.300 right? So it's sort of looking at people in 10-year increments, just from 25 to 34, 35 to 44,

00:07:41.420 et cetera, all the way up to north of 85. And it lists the percentage of people in each age category

00:07:47.720 that die in response to cancer. And here's what's interesting is that number peaks in the middle,

00:07:53.840 right? So at 25 to 34, it's 6%. 35 to 44, it's 13%. Think about that for a minute. That is a staggering

00:08:01.360 number for people so young. By the time you get up to 45 to 54, it's 23%. 55 to 64, 30%. 65 to 74,

00:08:11.920 31%. And then paradoxically, it begins to come down after that because those other diseases are taking

00:08:18.660 off. Another way to look at this is where does cancer rank in cause of death for all causes by

00:08:27.300 decade. And if you go in those same buckets, starting at 25 to 34, it goes from third, third,

00:08:32.900 second, first, first, second, third. In other words, it's always first to third. There is no other

00:08:40.580 disease that always ranks in the top three cause of death for every age. That's it. Full stop, period.

00:08:49.540 It's cancer. And so it's the second leading cause of death overall. We can talk a lot about those stats,

00:08:55.200 but there's nobody who's listening to this podcast whose life has not been affected by cancer. That

00:08:59.920 wouldn't be possible. I don't think you could come up with an example of someone who doesn't know

00:09:03.660 someone who's at least had cancer and very likely died as a result of it.

00:09:07.920 Now, that's a good reminder. You know, one interesting thing, maybe just to break that

00:09:12.540 data down a little bit, you know, people think of pediatric cancers. Of course, those are like,

00:09:17.160 if your life has ever been touched by a kid with cancer, there's like almost nothing more

00:09:22.160 jarring, you know, almost seemingly unjust, if you will, about a child being diagnosed with cancer.

00:09:28.140 For children, cancer is quite rare. It really occupies an enormous amount of mindshare.

00:09:32.760 But then as you go into the decades that you were summarizing, what's interesting is to reflect on

00:09:37.200 the cancer types, you know, that kind of lead the way. So brain tumors, leukemia, melanoma, the most

00:09:43.380 deadly form of skin cancer, you know, one of the cancer types I mentioned that I've been focused on my

00:09:47.440 career long. You know, that really jumps up in those 20s, 30s, 40s. Those cancer types kind of

00:09:53.740 lead the way in kind of the younger population. And there's some interesting implications there

00:09:58.440 in terms of like, well, what causes those cancers? And some people, you know, so vulnerable to them.

00:10:03.000 And then, you know, carcinogen-induced cancers, well, melanoma, of course, is there's a carcinogen

00:10:07.460 called ultraviolet light. That's a carcinogen for skin cancer, including melanoma. But like smoking-related

00:10:13.280 cancers, for example, you know, those really start to jump up in later decades. And then you've got,

00:10:18.840 I mean, everyone's aware of this, but obviously lung cancer leads the way there. But there's a

00:10:22.240 smoking footprint for a bunch of other cancer types that people don't think about so much.

00:10:26.300 Head and neck cancer is one that I think is, you know, relatively not top of mind for people. But,

00:10:30.600 you know, even when you get to bladder cancer, which you think, well, how does smoking cause bladder

00:10:34.220 cancer? And it's not the sole cause, but it's certainly a big contributor. You know, these sort of

00:10:38.360 smoking-related cancers, they take exposure, obviously, and a bit of time to accumulate

00:10:43.080 their population impact ultimately. And just one other thing that I would kind of throw in there,

00:10:48.140 because I'm sure we're going to talk about the really population-prevalent cancers, breast cancer,

00:10:52.140 prostate cancer, lung cancer, colorectal cancer, the big four. So breast cancer and prostate cancer

00:10:58.020 are not related to, well, obviously ultraviolet light or smoking so much, and a little bit of

00:11:02.780 smoking influence on breast cancer risk. But there, it's, you know, this really interesting interplay

00:11:07.540 between these hormone receptors, hijacked or co-opted in a way. And you think about the way

00:11:13.580 in which those cancers form, I think it best fits your age. You know, cardiovascular disease and

00:11:19.080 neurodegenerative disease, I would argue, there's something at play there that's similar to these

00:11:24.000 hormone-driven cancers, which are very age-related. So breast cancer and prostate cancer really pick up

00:11:28.620 in those later decades of age. So it's just kind of interesting to reflect on kind of the how and why

00:11:33.440 different cancers kind of feature in those different decades of age. And that has tons

00:11:37.620 of ramifications in terms of how we think about screening, which I'm sure we'll get into.

00:11:41.740 Yeah, for sure. So let's add a little more color to that, Keith. So you mentioned the big four,

00:11:46.600 lung being number one. I think a breast and prostate is kind of number two and colorectal number four.

00:11:52.140 And then if you add a fifth in pancreatic, you now account for a little over 50% of all cancer deaths.

00:11:58.560 So, you know, we'll talk about incidence, but we're going to talk about mortality. And at the end of the day,

00:12:02.340 just five cancers account for half, a little over half of all cancer death in the United States.

00:12:07.920 That's one point I'd make. Second point I'd make that's very interesting when it comes to breast

00:12:12.280 and prostate is on the one hand, we have this clear understanding of the role of hormones. And yet,

00:12:18.060 as you point out, the implicated hormones are actually at their lowest levels when these cancers

00:12:24.200 typically come on board. So we talk about the relationship between testosterone and dihydrotestosterone,

00:12:29.560 MDHT and prostate cancer. And yet when men have their highest levels of these hormones

00:12:34.700 in their twenties, in their thirties, and even in their forties, that cancer is never to be found.

00:12:40.020 That cancer shows up only when those hormones are long gone, you know, not gone completely,

00:12:44.560 but of course, greatly diminished. And the same is true with breast cancer, right? We see the incidence

00:12:48.400 of breast cancer going up, but it's not necessarily hitting at the peak level of estrogen in women.

00:12:53.220 There's more complexity to it, but again, it speaks to just how much is going on beyond the surface

00:13:00.580 and the first order of thinking. Yeah, no, that's a great point. And actually it's kind of very

00:13:05.200 tempting to insert here, you know, sort of deep cancer biology principle. If you look at other cancers

00:13:11.860 than breast and prostate, the cancer types where we've really gone the deepest in our understanding

00:13:17.960 of what causes them to become cancers in the first place, and even in terms of translating that in

00:13:21.960 terms of therapies, it's really been around the growth factor receptor and downstream of growth

00:13:28.100 factor receptors. Like, so on the surface of cancer cells, and then internally, that's where the action

00:13:32.900 has been. And here's the point, which you just made about the hormone receptors, is that basically

00:13:37.480 cancer cells, quote unquote, figure out how to become independent of the actual growth factors

00:13:42.540 themselves. So they basically turn through genetic mutation or alteration, they turn on these surface

00:13:48.160 receptors or the immediate downstream signaling molecules from those surface receptors. It's

00:13:53.020 mutations there. That's the absolute, like, nidus, if you will, the hotspot of where most cancers,

00:13:58.520 not all, but where most cancers actually get their kind of oncogenic drive, you know, the mutations that

00:14:03.660 drive cancer. Again, really analogous to the comments you, or the reality and the comments you made

00:14:08.580 about prostate cancer and breast cancer in terms of basically that the circulating hormone levels at the

00:14:13.800 time those cancers manifest are low, what's happened is cancer cells have wired themselves in a way to

00:14:18.920 be sort of autonomous or independent of those ligands, but still using the receptors and their

00:14:23.900 downstream consequences to drive those cancers. And I wonder if there's a parallel between the

00:14:28.240 following observation, which is a prostate cancer that develops in the presence of low testosterone,

00:14:34.440 all things equal, is a worse prostate cancer. So there's that paper in the New England Journal of

00:14:38.080 Medicine, God, it's probably been 15 years now, that demonstrated very clearly that the lower the

00:14:43.400 testosterone at the time of diagnosis of prostate cancer, the worse the outcome. Very counterintuitive,

00:14:47.980 right? Everybody thinks testosterone is causing prostate cancer, when in fact, what we're seeing is,

00:14:52.740 so I wouldn't interpret that to mean testosterone potentially has zero role, or that high testosterone

00:14:57.480 is protective, although some have argued that. What I would argue is your point, which is the cancers

00:15:02.640 that grow without the hormone are worse. And therefore, the parallel, if you will, is the ERPR negative breast

00:15:11.380 cancers are worse than the ERPR positive breast cancers. Those cancers, those typically hormone

00:15:18.620 sensitive or driven cancers that proliferate, whether it be the initiation or proliferation without their

00:15:24.000 respective hormones, tend to be the harder ones to combat. That's right. So in those instances,

00:15:29.860 those cancers have actually figured out how to essentially replace the function by virtue of

00:15:34.580 accumulation. I use language like they figured out. I think you might remember from our conversation.

00:15:38.120 Yes, we like to anthropomorphize cancer.

00:15:39.880 Exactly. I was just going to say, I love anthropomorphizing cancer. Actually, from a therapeutic

00:15:43.500 development perspective, it's like the easiest mindset to sort of adopt in terms of thinking about

00:15:48.440 how cancers solve the problems that they need to solve to become cancers. Like, it's kind of scary language

00:15:52.800 to use, I realize. But when you're trying to then reverse that or intercept that, it becomes a little bit

00:15:57.300 useful. So anyway, the point is that there's a constellation of mutations that can turn on

00:16:01.120 essentially the downstream pathways, for example, estrum receptor. So what's driving then a so-called

00:16:07.220 triple negative breast cancer, so lacking hormone receptors and lacking HER2, which is a well-established

00:16:11.620 surface target, a growth factor receptor on normal cells and on cancer cells, including breast cancer

00:16:16.780 cells. For breast cancers that have figured out how to become breast cancer, what you see in their

00:16:22.300 genetic makeup, is that they basically still are dependent on the same sort of cellular processes.

00:16:28.320 They still have to kind of regulate the same downstream programs. They just do it through

00:16:33.040 a variety of means and ones that become very challenging to directly target, to like to

00:16:38.160 intercept those. And so the point you're making about breast cancer, I'll just maybe complete a

00:16:41.820 little bit this way, which is you emphasize that their prognosis is better, right? So in other words,

00:16:46.820 before ever talking about therapy, those breast and prostate cancers that form and stick with breast

00:16:51.820 cancer, because that spreads across younger ages, you know, a bit more than prostate cancer does,

00:16:56.120 their prognosis is better. So before even coming to the issue of treatment, but their treatability

00:17:00.580 is also far greater because we've got drugs... Yeah, we have more targets.

00:17:04.100 Exactly. And intercepting those hormone receptors specifically, which are, by the way, inside of

00:17:08.520 cells, just a little nuance to separate from the comments I made about sort of the surface growth

00:17:12.740 factor receptors. Those drugs, we've had them for a long time, but also like serious advances have been

00:17:17.660 made applying new chemistry strategies to developing even better and better versions of those.

00:17:23.540 So what we're witnessing, let me just make this point in relation to pancreatic cancer, which I'm

00:17:27.840 glad you called out, is that I might've used this term four years ago. We have this scenario in cancer

00:17:33.280 where there's kind of a distribution of haves and have-nots. And what I mean by that is we have

00:17:37.680 patients whose prognosis to start with is better and whose therapy advances are really accelerating.

00:17:43.200 And like making certain cancers, and hormone receptor positive breast cancer is quite a good

00:17:48.700 example of this, where additional drugs now have been successfully developed as combinations since

00:17:53.900 we spoke four years ago, even now FDA approved and on the market. And so the outcomes of those patients

00:17:58.240 just continue to be distanced from cancers like pancreatic cancer, where first off, the lethality of

00:18:04.120 pancreatic cancer per case diagnosed, the case fatality rate so-called, is far higher than these other

00:18:09.640 cancers, right? So it doesn't even come close in terms of number of cases diagnosed to breast and

00:18:14.640 prostate or lung for that matter. But per case diagnosed, the likelihood that it's going to be

00:18:19.400 fatal ultimately is inordinately high. That's a prognostic issue. Those are aggressive cancers,

00:18:24.280 but also our therapy advances have been like really quite minimal, which is to say all we've got are the,

00:18:30.420 what I refer to as classical chemotherapy drugs of kind of the pre-2000 era, which have a modest

00:18:36.040 impact at all. So talking about haves and have-nots, pancreatic cancer unfortunately remains very much

00:18:41.040 in that kind of have-not end of the spectrum. Yeah, a couple other points to make just on the broad

00:18:45.580 contours of cancer. One of the other carcinogens we haven't really discussed, which is essentially the

00:18:50.820 second most prevalent environmental trigger of cancer after smoking is obesity. And we can certainly

00:18:56.940 debate whether it's obesity per se, which I don't think it is. In other words, I don't think it's

00:19:01.480 adiposity. I think it is inflammation and growth factors that come with obesity, namely insulin,

00:19:08.020 probably IGF-1, not to mention the inflammation that is part and parcel with that, which I assume

00:19:13.240 is in some way impairing the immune system and things of that nature. So that's another example

00:19:18.180 where you have a lot of these cancers. Let's think of those top five, breast, prostate, pancreatic

00:19:24.220 are clearly linked. Colorectal as well. I'm not sure about lung. So lung might not be as related

00:19:30.140 to obesity as the other four, but there are also many other cancers that fall outside of the top

00:19:35.680 five lethal, where we do tend to see this relationship. To my last count, I think there

00:19:39.460 were about 25, 26, maybe 27 cancers that have a pretty tight relationship to that. So that's not

00:19:45.840 only something that's increasing in terms of societal prevalence, but you might argue that that also takes

00:19:51.920 a while to sort of kick in. Yeah, totally right. Thank you for inserting that because it really is,

00:19:57.840 it's so easy to think about ultraviolet radiation and skin cancers, so easy to think about smoking.

00:20:03.480 I mean, now that we understand, you know, when we sequence an individual cell or a population of

00:20:07.920 cancer cells, sequence their genome, and we can now see the footprint, if you will, the damage

00:20:12.800 that those types of carcinogens induce, obesity is unquestionably another, that third highest ranking,

00:20:20.820 quote unquote, carcinogen. But the way it does it is certainly more complicated. And it is,

00:20:25.100 as you're saying, sort of, it's systemic. You know, I really do latch on to that literature

00:20:29.940 that you alluded to regarding insulin signaling, the body's metabolic response to obesity. I mean,

00:20:37.180 even you can study this just in a fed versus fasted state, like in a short-term setting. But when

00:20:43.100 you're, someone is obese, there are metabolic adaptations, if you will, that the body sort of

00:20:48.920 attempts to make. And I would draw the analogy to, you know, where we started with the hormone

00:20:53.180 receptor-driven cancers, so breast and prostate. You know, it's a different phenomenon to a degree,

00:20:58.820 but basically, you know, insulin growth factor, IGF, you alluded to, and then its receptors on cells,

00:21:04.580 which are sort of ubiquitous on all cell types, and certainly the cell types that, for which we've

00:21:08.620 got epidemiologic evidence that those cancers are more common in the obese population. You know,

00:21:14.320 what you can say from laboratory data is that the signaling that happens through insulin

00:21:19.580 signaling in cells. It's tightly tied to what we kind of talked about already, which is that sort

00:21:25.700 of growth factor receptor pathway. It is ultimately part of that same biology. There's a pathway that

00:21:31.220 connects those surface receptors into cells that then regulate how the mitochondria act as the power

00:21:37.220 stations, if you will, inside of cells. So the so-called PI3 kinase pathway, well described as being a

00:21:42.920 driver in cancer. That pathway is basically being chronically driven in that setting of high insulin

00:21:48.660 levels, high insulin growth factor circulating levels. So exactly what threshold level poses risk

00:21:55.600 and over what period of time, like, I would say those dots haven't been fully connected. But the

00:21:59.400 epidemiology is undeniable, and I would say the laboratory data that supports, you know, that

00:22:02.780 connection is also undeniable. So, you know, there's something about that pushing, that driving. It's, I mean,

00:22:07.960 like chronic inflammation, as you cited, which itself, by the way, of course, is a direct causal factor

00:22:12.820 for certain cancers. An organ site or tissue site where there's chronic inflammation, well described

00:22:18.420 that cancers can arise in that setting. It's a similar phenomenon, basically. They kind of keep

00:22:23.200 whipping the horse, if you will, in a way. And cells will, you know, ultimately, through genetic

00:22:28.200 alteration, still basically respond to that environmental stress, and cancers ensue.

00:22:34.940 So I remember in January of 2000, I'm in my last year of medical school, and I trekked across the

00:22:40.620 country from California to Bethesda to go and spend four months as a medical student rotating

00:22:46.980 on the immunotherapy service with Steve Rosenberg. And this was the opportunity and dream of a

00:22:53.860 lifetime for me. I had read his book, The Transformed Cell, many, many times as a medical

00:22:58.360 student and wanted to basically go and learn what I could. And I look back at that, I think I was there

00:23:03.780 from January until April of that year. I mean, literally one of the most joyful examples of

00:23:09.260 pure bliss. I've told this story before, but I think I told this story when I had him on the podcast.

00:23:14.240 You know, I had found a friend I could stay with in Bethesda, but it turned out in the four months I

00:23:19.080 was there, I was probably only there eight times. I didn't leave the hospital. I literally had a cot

00:23:25.980 where I slept, and I just didn't want to be out of there. And I wanted to be as close to the lab,

00:23:30.580 as close to the clinic as possible. But I'll never forget one of the most, you know, insane things

00:23:35.020 that he said in the first week that I was there. He said, looking back over the past 30 years,

00:23:41.220 we have basically made no progress in the long-term management of metastatic epithelial cancers.

00:23:49.900 Translating that into English, if you had a solid organ tumor that had spread to a distant site in

00:23:56.260 1970, the chance that you were going to be alive in 10 years was the same in the year 2000. And that

00:24:02.640 was basically zero. Now there were a couple of small exceptions, and they happen to be the cancers

00:24:07.460 that you and he are both interested in. There were about 10 to 15% of patients could achieve

00:24:13.140 a solid durable remission at the time to a high-dose interleukin-2, but that was not appearing

00:24:18.680 to be the case for any other epithelial tumor. And there was still absolutely no sense of why that

00:24:24.080 wasn't the case for the other 90% of patients who had metastatic renal cell carcinoma and melanoma.

00:24:29.660 How are we doing today, 23 years later? Do you have a sense of how much bigger that number is?

00:24:36.420 So if we went from 0% 10-year survival in 1970, and I'm using 10-year to really try to get out some

00:24:42.840 of the median survival extension stuff, but if we were 0% survival for a solid organ tumor in 1970,

00:24:49.760 call it 1% in the year 2000 on the back of the few cases of RCC and melanoma, what are we at today?

00:24:58.880 I think 15% is a conservative number. Some would make a case for 20%. I think the problem is there

00:25:04.840 that we need a little bit more time. Some of the newest therapies, you know, are only-

00:25:08.000 We don't have the 10 years.

00:25:09.080 Yeah, we don't have the 10 years. But if you track their five-year, three- and five-year outcomes,

00:25:12.620 if you will, you'd like to think that they're going to get us to 20%. Just to clarify what we mean by

00:25:18.420 using the term metastatic. So clinically overt, detectable metastatic cancer means that you're

00:25:24.660 picking it up radiographically or clinically. That's what that term means. Now, the fact is

00:25:29.000 that cancers are found, when they're found at what's thought to be an earlier localized site,

00:25:34.340 you know, it's very, very common that cancers will have spread to so-called regional lymph nodes,

00:25:38.220 like through lymphatic channels to the closest lymph node basin, wherever that may be. This is not true

00:25:43.040 for all cancers, but it's certainly true for the common epithelial cancers you're focused on

00:25:46.300 in this question. And so basically the point to emphasize there is that spread to lymph nodes

00:25:51.940 is properly called metastatic. It's a jargon term. We don't think about that as being metastatic.

00:25:57.880 Sort of the analogy, right, is like when people leave a city in an airport and go to another airport

00:26:03.960 that's in another city, we don't call it spread until they leave the airport and go to the city proper,

00:26:10.020 even though they've clearly demonstrated the capacity to go from their house to the airport,

00:26:14.960 hop on an airplane, go to another airport. And once they step foot out of customs and collect

00:26:19.960 their bags, well, now we can say they've really spread. The other point I was just going to quickly

00:26:23.320 make is that it's feasible to surgically remove regional lymph nodes along with the primary site

00:26:27.760 of disease in the vast majority of cases. And so because of that sort of historical standard

00:26:32.800 practice of surgical resection, including regional lymph nodes, we think of surgeries that can encompass

00:26:38.240 all of that as basically being one treatment. And then those patients, we're going to come to this,

00:26:42.300 I think probably a little bit further along in the conversation, but they're thought to not have

00:26:46.640 metastatic disease. But what I mean by that is overt. So like you can detect it clinically or

00:26:51.460 radiographically. How do we know that some of those people actually have metastatic disease at that

00:26:56.440 time? Well, by following them, five and 10 years, well, not even that long. I mean, one, two, and three

00:27:00.360 years is in fact enough for most of the aggressive cancers. You do the surgery, you clean the slate,

00:27:05.440 you do scans of various kinds, you see nothing. And basically a substantial fraction of those

00:27:10.820 patients, depending on the cancer type and depending on how much, you know, the features of their

00:27:14.080 primary tumor, as well as lymph node involvement, a substantial fraction of those patients, let's go

00:27:18.240 with 30, 50%, kind of a typical range, over those few to several years of follow-up will manifest

00:27:24.300 metastatic disease. Well, they always had it. They had it from before the surgery was ever done.

00:27:28.580 And as you said, cancer cells, or the analogy of the traveler, left the airport. They lodged in a

00:27:34.680 distant site. We just didn't have the methods to find it. So time would tell that in fact,

00:27:39.560 in retrospect, they had had it. This is where actually some huge advances have been made in

00:27:43.460 terms of blood-based detection of those instances. We're not perfectly good at that now, but there

00:27:48.360 have been substantial improvements in the technology for detecting, particularly circulating tumor DNA

00:27:54.500 in instances where people have only microscopic metastatic disease. That's the term I wanted to

00:27:58.860 kind of insert in this conversation is microscopic metastatic disease. We'll come back to this, but let me

00:28:03.620 just pick up on this theme again for evident overt metastatic disease when you can see it on scans or

00:28:10.020 clinically witness it. That's where those numbers pertain that we're talking about, like getting in

00:28:14.160 this 15 moving towards 20% range. 10% on an absolute scale. Let's go with the idea that we're on the path

00:28:21.000 with available therapies that have just recently been introduced, included towards that 20% number.

00:28:26.420 Half of that advance has come from one therapeutic modality, PD-1 antibody-based immunotherapy. A

00:28:33.000 single approach has accounted for half that number. It's astounding. And then what about the other

00:28:38.800 half? That has come from a repertoire of these so-called molecularly targeted therapies that

00:28:44.260 intercept those genetically altered drivers that I alluded to some minutes ago, these surface receptors

00:28:49.600 and their downstream signaling molecules. Those individual drugs have picked off as small as 0.2%

00:28:55.600 of the cancer population in the rarest instance up to a couple few percent. But you add them all up

00:29:02.460 and those have produced also long-term survivors now by historical standards. So that 10-year

00:29:08.260 numbers, it's an astoundingly long survival by historical standards because metastatic cancer

00:29:12.840 will prove fatal in nearly everybody untreated within that timeframe, even the most quote-unquote

00:29:17.980 indolent cancers. Yeah. So let's just kind of go back and restate the important part of that. So

00:29:22.620 basically 1970 to the year 2000, zero progress has been made. 2000 until now, we've probably been able

00:29:31.380 to make a small dent in that. Half of that dent has been on the back of Keytruda. How many drugs are

00:29:37.460 in the other half of that? So we mentioned Gleevec a minute ago. That was probably the first.

00:29:42.480 There have been 52 FDA approvals, but that's against 19 unique mechanisms, right? So there's a lot

00:29:48.760 of meat-to-ism. I mean, this is true in all therapeutics, not just oncology. I tend to then

00:29:52.720 go down to that number. 19 unique mechanisms. And even there, there's some overlap. So like

00:29:57.680 different molecular targets, but like for the same population, like within a given pathway

00:30:01.600 inside a cell, there's instances where we've actually successfully drugged one component and

00:30:06.620 its immediate downstream component and the immediate downstream component yet again. That would then

00:30:11.620 count as three on that list of 19, but really they're very overlapping. So if I were to really

00:30:17.740 boil it down, we're kind of in the 10 range in terms of truly unique molecular targets. And Keytruda

00:30:24.320 does have company. So this target of Keytruda, just for those who are trying to keep up with the jargon

00:30:29.720 as they do their Google searching on any of these topics, the target of Keytruda is P, capital P,

00:30:35.980 capital D, hyphen one. That's a surface receptor on certain immune cells, particularly on these CD8

00:30:43.040 positive T lymphocytes that can kill tumors directly. But there are other immune cells that

00:30:47.300 express PD-1 and that's a break on those immune cells. So the antibodies that block that break

00:30:52.740 are the so-called PD-1 antibodies. There are five of them that are FDA approved and Keytruda or

00:30:56.960 Pembrolizumab is that's the dominant one that made it to market first and also in the broadest number of

00:31:03.540 cancer populations. When I said me-tooism before, there's lots of me-tooism in that space.

00:31:08.620 And then is anti-CTLA-4 still being used or is that mostly just being used in melanoma?

00:31:14.480 What's the prevalence of its susceptibility versus that of PD-1?

00:31:19.560 Four cancer types now, and most would argue a fifth, have evidence that adding CTLA-4 to PD-1

00:31:25.480 as another block, so that's another break on the immune cells, on those same T cells. It was actually

00:31:29.840 discovered before PD-1 as a target and the therapy was advanced against it a little earlier than PD-1.

00:31:36.380 But a much smaller percent of cancer patients get a benefit from that drug. There's some evidence

00:31:40.600 that it actually can act kind of independently, sort of exert its own benefit. But I'm ascribing

00:31:45.300 that 10% number. By the way, there's some real math behind that. It's not totally like just a

00:31:49.480 gestalt number of patients who get long-term benefit from PD-1. If you add in CTLA-4,

00:31:54.700 you're in the 1% range in terms of the addressable population for CTLA-4 antibody therapy who derive

00:32:01.740 then 10-year type benefits. So PD-1 is doing really the heavy lifting.

00:32:06.760 I think it's probably worth just sort of explaining immunotherapy again. We have an entire podcast

00:32:11.760 dedicated to that. When I sat down with Steve Rosenberg, you and I spoke about it briefly

00:32:15.780 four years ago. But I think given that the immune system is, I was about to make a joke that wasn't

00:32:22.820 intended to be a joke. I was about to say it's not innate, as in our understanding. But of course,

00:32:27.580 it is innate in its physiology. But given that I think that people don't necessarily completely

00:32:31.560 understand the nuances of the immune system, given that it's played such an important role

00:32:36.920 in cancer optimism over the past two decades, and given that it's probably about to play more of an

00:32:44.420 important role as we go forward, I think it's worthwhile for the listener and viewer to understand

00:32:50.140 how the immune system works with respect to cancer. Because when we talk about TIL,

00:32:55.180 when we talk about checkpoint inhibitors, which we've already touched on, I don't want people

00:32:58.980 to be lost. So unfortunately, this is one of those moments in this podcast where you got to buckle your

00:33:03.200 seatbelt up a little bit, but it pays dividends because you become a very educated consumer of

00:33:08.000 how these drugs work.

00:33:08.900 Let me kind of layer the onion this way, which is, I find it most useful, and I'd say this even in

00:33:14.620 talking to patients, my lay audience, if you will, to start with the concept that the immune system

00:33:19.800 needs to find levers that it can grab onto, as in differences, things that are fundamentally

00:33:24.620 different than normal cells. Our immune system is trained in fetal development to do exactly that,

00:33:30.440 and quote-unquote only that, except for the fact that we unfortunately hold on to self-recognizing

00:33:34.840 immune cells, and those can cause autoimmune disease, which is not the topic of our conversation

00:33:38.960 today. But basically, consider what's different about cancer cells. You know, what have we learned

00:33:44.080 over the decades on that topic? There are a variety of differences. I'm going to start here

00:33:49.200 because it's going to give a little bit of chronology in a way. We began to understand

00:33:53.260 some time ago that a common feature of cancer cells is that they behave like they're sort of

00:33:58.720 progenitors or precursors. Like in development, all mature cells in the body come from a stem cell

00:34:04.420 of some sort, and there's lineages and different types of stem cells. But ultimately, you see cancers

00:34:10.720 actually adopt sort of a biological behavior that's like backing up, if you will, in the

00:34:15.820 developmental process. This is just a consequence of the genetic alterations, sort of the combination

00:34:20.560 lock, as I often refer to it, of genetic alterations that can lead to cancers. That's one of the programs

00:34:25.380 that they typically adopt. And it turns out that developmental cells have surface proteins, surface

00:34:31.500 markers, if you will, that are not expressed in fully mature tissues. And the immune system can see

00:34:37.660 those. So that's well-documented. And Steve Rosenberg's early successes, actually, were

00:34:41.920 identifying those immune cells that existed in people that could recognize those types of antigens.

00:34:47.580 These are referred to as cancer testis antigens. So just think of that as kind of this developmental

00:34:51.600 sort of biology. It turns out there are also, interestingly, some what we refer to as lineage

00:34:56.680 antigens. So like surface markers that tag a certain cell type that the immune system, interestingly,

00:35:02.560 can recognize, even though we think of those as being more like self. But we see that. We see evidence

00:35:07.240 that the immune system reacts to those. And that there are cell therapies, as you were alluding

00:35:12.500 to before, that also take advantage of that. The big discoveries of the recent several years have

00:35:18.040 been that carcinogens cause mutations in genes, that then those genes encode first RNA and then

00:35:25.440 proteins. And the altered amino acid sequence of the protein, that can be recognized. So those are

00:35:30.940 intracellular, almost always, those proteins. But we have a machinery in our cells, all cells in the

00:35:36.240 body, including those cells that go on to become cancer, that basically breaks down those proteins

00:35:41.180 as they age and will present a representative set, if you will, of those broken down protein

00:35:47.020 fragments or peptides to present them, meaning on the cell surface, in the context of these molecules

00:35:52.780 we refer to as major histocompatibility receptors, as they're kind of alluded to. But the idea is that

00:35:57.460 they're trying to show the wares, if you will, the inner contents of a cell to the immune system.

00:36:02.460 Because we think that because of that, that virally infected cells, you know, have an infection

00:36:07.120 inside. We think this is how this machinery was ever, you know, how it ever evolved in the first

00:36:12.020 place. And so kind of showing the inner contents, if you will, as a way of being able to let the

00:36:15.820 immune system know that there's a virally infected cell. Well, that same machinery exists again in every

00:36:19.840 cell. And by the way, if cells stop doing that, there's a branch of the immune system, stop presenting

00:36:24.720 antigens at all. Antigen is a new word. I meant to introduce that. Antigen means a difference,

00:36:30.800 like a protein fragment that's being presented and seen as different. We call that an antigen,

00:36:35.360 and it can come in these different categories that I'm talking about. So basically, if a cell,

00:36:39.100 if a cancer cell were trying to hide itself, if you will, by not expressing these receptors to

00:36:44.540 present antigens, then there's actually a branch of the immune system that's basically natural killer

00:36:48.380 cells, as they're called. They're very primordial immune cells that are supposed to just swoop in

00:36:51.820 and kill those cells. And we have evidence that that does occur.

00:36:54.240 So let's just pause here, Keith, to make sure people are following the anthropomology of this.

00:36:59.180 Basically, you have a row of homes, and each person in their home is responsible for demonstrating

00:37:07.560 the contents of their home. So they reach inside, and they pull out various items from their home,

00:37:13.380 and they leave them on the curb. And the military is coming down the street, inspecting the contents on

00:37:19.920 the curb. And they're just making sure that it's all stuff that we've pre-agreed is safe, right?

00:37:26.960 That's right.

00:37:27.300 So they don't know the entire repertoire of what could be presented, but they have a very clear

00:37:31.860 list of what is acceptable. And they're basically just identifying anything that is not on the

00:37:37.140 acceptable list. And if anything shows up and it's not on the acceptable list, the house is destroyed.

00:37:42.380 Furthermore, if you leave nothing on your curb, either because you're too incompetent,

00:37:47.500 or you're nefarious and you're trying to hide what's in your home, there's another branch of

00:37:52.260 the military that comes along and just blows up your house. So failing to play the game

00:37:56.620 results in a loss of home.

00:37:58.820 That's right. Well said. So that's the beginning. I mean, it's this kind of sampling, if you will,

00:38:03.060 like you said, of the inner contents. That's important to recognize because if you start

00:38:07.440 with this core principle that cancer is a quote-unquote genetic disease, meaning that mutations

00:38:11.540 that happen in key genes that disable cells' ability to repair DNA damage as a common feature

00:38:18.440 of cancers, for example, or mutations that activate some of those surface receptors or

00:38:22.740 downstream signaling molecules that we talked about before, those mutations we've learned

00:38:26.980 in recent years can be seen as different. So they began to increase the toolbox, if you will,

00:38:32.580 of handles that the immune system can latch onto. So if you think about it that way, the cancers

00:38:36.620 you know, begin to form potentially, if they're witnessed by the immune cells as having a

00:38:42.320 difference early, we have lots of evidence that they can be eliminated. And there's actually

00:38:45.920 indirect negative evidence, if you will, that people who have profoundly compromised immune

00:38:50.280 systems will pop up with cancers. I mean, if you give people seriously high-dose immunosuppressive

00:38:54.740 medication for various other medical conditions, you will see cancers just sprout up quickly and then

00:39:00.220 certainly over time as well. This immune surveillance concept is an inordinate amount of evidence

00:39:04.780 in support of this idea that if at least keeping them down, you know, proto-cancers down, if not

00:39:10.280 outright eliminating them, is that's just part of life on planet Earth in the cosmic storm, if you will,

00:39:14.900 with UV radiation as being, you know, one carcinogen I mentioned. Well, actually gamma radiation

00:39:19.420 coming through the atmosphere is also a cause of DNA damage. We have to try to repair that damage

00:39:25.560 inside of cells. I mean, when I say we, again, using the anthropomorphic inside of a cell's inner

00:39:30.160 workings here. But if the repair can't happen, we have this other mechanism of immune

00:39:34.220 surveillance basically to wipe it out. The reason why I wanted to just kind of spend enough words

00:39:38.660 on this concept is that basically people have to understand that by the time they're diagnosed with

00:39:42.620 cancer, something's gone wrong. The system didn't work to detect, you know, in this surveillance mode

00:39:48.600 the forming cancer. It didn't eliminate it. How can that be? Well, it turns out for every process that

00:39:55.820 activates the immune system in response to an infection, let's go with it, idea that that's the

00:39:59.620 primary function of the immune system in terms of how it is that we ever got out of the swamp in the

00:40:03.620 first place evolutionarily. There's a break on the immune system. Like you can't just elaborate

00:40:08.100 immune response, you know, indefinitely. I mean, imagine having the flu forever, like just dumping

00:40:12.960 cytokines or immune system hormones into the bloodstream, cranking up body temperature, consuming

00:40:18.600 a ton of metabolic resources in fighting infection and feeling bad as a consequence when you have the

00:40:23.560 flu. You can't do that indefinitely. You've got to stop immune responses. And so we have mechanisms to

00:40:29.920 do that. Turns out very elaborate set of mechanisms to do that. And cancers have just ever so craftily

00:40:35.400 figured out how to basically kind of reach into the genetic code, the blueprint, and co-op mechanisms

00:40:41.460 that will basically impede immune system recognition and response. That's the PD-1, PD-L1 story. So PD-1,

00:40:49.840 we've talked about, that's the target of Keytruda. But what cancers do, it's a nasty little trick,

00:40:54.880 is they've figured out how to, not all cancers, but the ones that are most responsive to Keytruda,

00:41:00.900 they have figured out how to express on their surface, the foot that presses on the brake.

00:41:05.860 Okay. So that's called PD-L1. So program death hyphen L1, ligand 1, which basically reaches across

00:41:13.420 to PD-1 on T cells and tells them shut down, basically. Mission accomplished. Don't need to do

00:41:18.800 anything here. And so a lung cell, an alveolar lung cell that ultimately becomes cancer, is not

00:41:25.280 supposed to be expressing that protein on its surface, right? It's not supposed to be regulating

00:41:28.840 the immune system. That's not the natural job of a lung alveolar cell. But a cancer that arises from

00:41:34.620 that cell, in many instances, basically, quote-unquote, figures out how to express that protein.

00:41:40.000 And so then blocking the interaction of the foot with the brake, that's the magic. There it is.

00:41:44.640 Now, that's just one mechanism. But as I said, it's actually produced a bigger incremental benefit in

00:41:49.380 the cancer population than any single mechanism we've ever discovered in all of cancer biology

00:41:54.440 research and therapeutic development history. So it's a pretty powerful one. But I'll just conclude

00:41:59.060 with this statement that there are other mechanisms by which the immune system can be suppressed. In fact,

00:42:04.460 there's entire cell types in the immune system repertoire have a dampening effect on immune system

00:42:09.880 response. And cancers can recruit them into their so-called microenvironment and create this very

00:42:16.120 adverse environment for the T cells that could otherwise attack and kill. So it's almost like

00:42:22.020 assembling a force field by virtue of inviting in these non-cancerous cells. This is like the cancer

00:42:28.140 cells recruiting in these suppressive immune cells. So this is some of what we're up against. I mean,

00:42:33.640 I just want to make it clear how kind of complicated it is. Yes, we're super grateful to have had this

00:42:38.900 kind of eureka moment with the success of PD-1 drugs. But cancers have co-opted multiple mechanisms

00:42:44.640 by which they defend themselves in terms of trying to close the gap then and use this immunotherapy

00:42:49.440 concept much more broadly in cancer is going to require us to develop the understanding of, okay, well,

00:42:55.460 which tricks are being pulled and how to be able to really target those very specifically. We can't

00:43:00.780 disable people's immune systems. Like, that's not okay. And so we do need a fair amount of precision

00:43:06.180 and figuring out kind of the sweet spot, if you will, in terms of what mechanisms cancers are using

00:43:12.040 for this purpose. All the things that Steve talked about when I interviewed him last year,

00:43:17.640 the one that I was most blown away by, which spoke to my time away from the trenches, time away from what

00:43:25.080 you're doing day to day, was that roughly 80% of epithelial tumors had novel neoantigens. Now,

00:43:34.480 again, if you said that at a party, that would go over everybody's head and it wouldn't resonate as

00:43:38.820 a particularly insightful thing to say. But in light of what you've just said, let's make sure

00:43:44.060 people understand what that means and how shocking that is relative to where we were 20 years ago.

00:43:51.000 Just to put this in context, when I finished my time at the NCI and went back to finish medical

00:43:56.380 school and applied to residency, you know, you talk in residency interviews, you're talking about what

00:44:01.280 you're obsessed with and what you're interested in. And I talked a lot, I mean, that my time at NIH was

00:44:05.860 such a formative part of my education. And I can't tell you how many people I interviewed with

00:44:12.500 that just laughed in my face and said, this immunotherapy stuff is nonsense. Like it's

00:44:17.700 totally irrelevant. What are you talking about, kid? Like you're going to sell yourself to us as an

00:44:22.360 interesting person that we should let into our program. And you're talking about that crap. Like

00:44:26.920 it literally means nothing. Okay. It works on melanoma. Who cares? Okay. Why is the fact that

00:44:34.340 80% of epithelial cancers have novel neoantigens, a totally staggering feature that had people

00:44:40.840 understood that 20 years ago, maybe more than just a handful of people would have found immunotherapy to

00:44:45.420 be a very promising field? Let's break that down a little bit, you know, kind of the biology behind

00:44:50.620 this. So mutations accumulate in a cell that's going to become cancer. Fair number. I mean,

00:44:55.600 we're talking, you know, never less than dozens in the quote unquote, most genetically simple

00:44:59.920 cancers, but you're typically into the hundreds and thousands. And basically not all of those have

00:45:05.780 a consequence to the point about these antigens. So some of them are in parts of the genome that

00:45:10.220 don't even encode proteins, in which case they're not going to ever become the types of antigens you're

00:45:14.320 alluding to. The ones that are translated into proteins, again, those proteins age, they get broken

00:45:19.360 up in the proteasome presented in the context of these MHC molecules that I referred to before on the

00:45:25.020 cell surface. But that's done differently in each of us. And so basically, we don't show our entire

00:45:30.180 wares, we show selected representation of them. In other words, these MHC molecules, you know, you

00:45:35.000 inherit kind of half of your set from your mother and half of your set from your father, they have a

00:45:39.320 ability to grab just certain protein fragments and present them. When I say grab, they're like actually

00:45:44.320 loaded onto those by a cellular machinery that's quite elegant. And so the point is, we have this

00:45:49.080 repertoire of showing meaner contents. And so only certain mutated genes that translate into what's

00:45:55.680 called the mutated proteins or altered proteins, only certain of those can actually be presented

00:45:59.980 out of the very large number of mutations that actually exist. But what is astounding is that,

00:46:04.740 as you say, 80% of... So when you use epithelial cancers, just remind people, like breast, colon,

00:46:10.380 prostate, lung... It's not leukemia and lymphoma.

00:46:12.740 And brain tumors. Right, yeah. Not leukemia and lymphoma or brain tumors. And melanomas,

00:46:16.460 as we mentioned before, this actually come from melanocytes, which are of neural crest origin,

00:46:20.000 which are actually share common features with brain tumors. But basically, just to be complete,

00:46:23.840 it's all the rest of cancer. What's astounding is that basically you can find evidence that these

00:46:29.740 mutated proteins are being presented in the vast majority of these common cancers. And here's the

00:46:34.460 point. We are born with... Well, born and during early development after birth, we elaborate this

00:46:42.060 very impressive repertoire of T cell receptors that sit on the surface of T cells that can recognize

00:46:47.940 exactly these altered proteins. Like with just one amino acid substitution present in the peptide

00:46:53.800 fragment, we've got that repertoire. You know, the proof that these antigens are antigens, I mean,

00:46:58.340 like to meet the definition of antigen, you have to find in a human being that the immune system can

00:47:02.500 actually see it in the context of it being presented on these MHC complexes. And it turns out that

00:47:07.280 it's kind of a lock and key concept, essentially. It has to structurally work out that the protein

00:47:12.500 fragment is being presented, you know, in the T cell receptor docking in and seeing that version,

00:47:17.440 but not the unmutated version, where that difference is enough to basically tell the T cell,

00:47:21.960 go, kill. Yeah. So these exist. Like this first began to be described in earnest about a decade ago.

00:47:28.640 You know, of course, we were sequencing cancer genomes a ton at the time. And we figured out as PD-1

00:47:33.620 and CTLA-4 were being clinically developed, we began looking then in retrospect and seeing,

00:47:39.460 you know, it's these tumors that have a ton of these mutations. They're the ones that are responding

00:47:43.500 like much more likely than other cancer patients slash cancer types. And so guess what? It's the

00:47:50.340 ultraviolet radiation associated cancers that have like just enormous amounts of mutations in total

00:47:55.520 and the presence of actually significant numbers, usually oftentimes dozens of these mutated

00:48:01.200 neoantigens. That's the jargon term that we're coming towards here. And so that explains why the

00:48:06.620 response rates are so high in those cancers. Smoking-related cancers then account for just

00:48:11.280 about all the rest of where PD-1 has been efficacious. Like we didn't know this when PD-1

00:48:16.240 and CTLA-4 antibodies were first being developed, that it was this interplay that we could then just

00:48:21.600 add one drug that blocks this foot on the brake, as I mentioned before, and bam, you unleash these

00:48:27.440 pre-existing T cells against these presented antigens. But that does explain a ton of the benefit that

00:48:32.940 we've covered already with this so-called immune checkpoint antibody approach. So that's fascinating.

00:48:38.320 I think what you're coming to is then what else can we do with that information? And so what Steve,

00:48:42.580 I'm sure, talked about with you is, well, we can actually engineer immune cells to attack these

00:48:48.020 things. Basically, potentially overwhelm other ways that the immune system, that cancers try to

00:48:53.540 protect themselves from the immune system, which is what cell therapy of various kinds can do.

00:48:58.560 So basically, we're still in the early days of elaborating this understanding that, yes,

00:49:03.400 the vast majority of cancers have these alterations that the immune system can actually recognize.

00:49:08.640 Let me just finish with this one very nuanced point. We have learned that some mutated new antigens

00:49:13.560 will cause a much more robust immune response than others. In other words, they're not all the same

00:49:17.680 in terms of the type of immune response that can be elicited. And there's an argument that many

00:49:23.200 have made, in terms of thinking about cancer biology and evolution and coexistence of this

00:49:28.580 immune surveillance system, that basically the mutations that we end up seeing in diagnosed

00:49:34.060 cancers are ones that aren't particularly well-recognized. They don't produce powerful

00:49:38.880 immune responses. The ones that produce powerful immune responses, well, guess what? Those cancers

00:49:42.760 never became cancers in the first place. They got wiped out. So there's this notion that basically

00:49:46.740 you have to be able to fly under the radar. You can build yourself as a cancer cell with a certain

00:49:51.760 repertoire of mutations, provided that none of them are powerfully immunogenic.

00:49:56.880 We could talk about this forever. I'll say a couple more things on it so that we can move on to talk

00:50:00.680 about T cell therapies, where I think we're going next. Again, I think to put a bow on this, the way I

00:50:05.620 think about this is that through all of recorded human history, there have been very, very rare

00:50:13.100 reportable incidents of spontaneous regressions of solid organ metastatic, you know, these epithelial

00:50:20.200 tumors, right? Where, you know, Steve Rosenberg writes about one, which was the patient who got

00:50:25.240 him to completely change his career. It's the 1960s. He's a resident at the Brigham. A patient

00:50:30.340 comes in who 10 years earlier had been sent home to die with metastatic gastric cancer throughout his

00:50:37.440 liver. You know, they took his stomach out to paleatum and he should have been gone in three months. He

00:50:42.240 shows up 10 years later with a gallbladder that needs removing, not a shred of cancer. Clearly a

00:50:47.800 spontaneous remission. There's an example of someone who made not so much and so significant

00:50:55.200 of an antigen that it got wiped out before it got anywhere. This one actually got all the way to the

00:50:59.940 promised land, but somehow at that point, the immune system said, I recognize it and there are

00:51:06.320 enough of us that recognize it and we're going to wipe this thing out. And then what basically happened,

00:51:11.640 and it took 20 years, yeah, almost 20 years, right? Was figuring out that if you just dump enough

00:51:18.400 interleukin-2, which is candy to T-cells, you're going to pick up the next threshold, which is in

00:51:26.180 melanoma, in renal cell. At the time, we didn't know why, but as you point out, they just have so

00:51:31.480 many freaking mutations that you're bound to just stoichiometrically come up with an antigen that's

00:51:37.940 going to be your lottery ticket. If we just dump enough interleukin-2 on, we're going to flip the

00:51:42.760 next threshold. And then of course, the checkpoint inhibitor takes it one step beyond that, which is,

00:51:49.260 okay, you clearly don't have enough for spontaneous mutation, spontaneous response. You don't even have

00:51:54.000 enough that if I just gave you IL-2, but if I give you a more sophisticated help turning down the

00:51:59.800 suppressor, now it's going to work. But to really unlock this, to basically say we could make 80% of

00:52:06.940 cancer, gone. Just gone. Imagine that. And by the way, it might be more because maybe you can induce

00:52:12.460 mutations. We're going to come to that in a moment. If we just wanted to take 80% of cancer deaths off

00:52:17.240 the table, we have to be able to find out who is that perfect soldier down there that's really,

00:52:23.000 really, really outnumbered and make more of them. So what does that look like?

00:52:27.480 You know, just to connect the dots from early versions of cell therapy to where we are now,

00:52:31.820 you know, Steve Rosenberg's work was so-called adoptive T-cell therapy. Let's not focus on that

00:52:36.700 jargon term so much, but basically doing a surgery to remove a single site of cancer,

00:52:40.840 metastatic cancer, removing the immune cells that had found their way into that cancer,

00:52:45.360 which turns out actually is some of them are seeing antigens they're specific for. Others are

00:52:50.420 just trafficking through, and they're kind of bystanders, as it turns out. But in any case,

00:52:53.640 immune cells don't traffic in high numbers through all cancers, but certain, quote-unquote,

00:52:58.580 immunogenic cancers, yes, they do. Melanoma, again, being near the top of the chart there.

00:53:03.480 What Steve was doing through the 90s, and certainly by the time you got there,

00:53:07.520 was taking those immune cells, isolating them from that patient's tumor, and simply expanding

00:53:12.020 them. No genetic anything. It was just grow these till.

00:53:15.160 Exactly. Yeah. To a number that when he infused them back could then, you know, systemically,

00:53:20.560 could then traffic through the body and destroy people's cancers. Now, not all the time, but a

00:53:25.660 significant minority of patients could be cured that way. That's still true today. And by the way,

00:53:30.360 we are right on the verge of that, just that approach alone, no genetic manipulation becoming

00:53:35.720 an FDA-approved therapy, finally, for melanoma, which is where Steve had had the most consistent

00:53:39.820 success back in those years. He's tried it in many different cancer types. Now, what's been learned

00:53:44.920 along the way is exactly what we've already summarized, which is, you know, this idea of

00:53:48.680 antigen specificity that you can find, you know, this kind of, what is it that the immune system is

00:53:52.540 seeing? These infiltrating immune cells in tumors, what are they looking at? And then taking that

00:53:57.180 knowledge to basically now begin to engineer immune cells, generally starting with the patient's own

00:54:02.060 immune cells, so not coming from tumor anymore, but just basically collecting them from the blood,

00:54:07.120 you need a fair number of them. But because of genetic engineering advances, cellular genetic

00:54:12.400 engineering advances, we can now basically swap in, swap out, essentially whatever we like.

00:54:17.540 And in these immune cells that we want to direct against cancers, we can basically introduce

00:54:22.780 the recognizing piece, if you will. If we sequence a patient's cancer, which we do as part of routine

00:54:28.460 standard care these days, but you do it a little bit deeper, a little bit more in a systematic,

00:54:33.880 thorough way, basically, let's go with that number, 80% of patients, you know, we can identify

00:54:38.300 a mutated antigen that will only be in the cancer cells and introduce into their immune cells a surface

00:54:45.220 recognizer, if you will. I'm just being vague about the term to not get lost in too much jargon all at

00:54:50.480 once, and then basically just dial up that number of cells in the laboratory and then infuse them

00:54:55.580 back like a blood infusion, which is how cell therapy is given. And that's the approach that

00:55:00.140 translates, you know, connects the dots that we've covered. We are not doing that today, to be very

00:55:04.440 clear. The cell therapy advances beyond just simply expanding the tumor-infiltrating immune cells or

00:55:10.980 lymphocytes, beyond that approach, the engineering that's being done right now are against surface

00:55:16.780 lineage markers. So on B cells for lymphoma primarily, but some leukemias and now multiple

00:55:23.200 myeloma as well. Basically, we are wiping out the cancer cells that arise from that population and the

00:55:29.360 normal ones, okay, just to be clear. What we were talking about is a very elegant, very tumor-specific

00:55:35.980 cell therapy strategy, which you can readily envision sort of taking the field, if you will. But where we are

00:55:42.460 right now in cell engineering is going after common surface markers in cell populations that we can

00:55:48.280 quote-unquote afford to get rid of. So eliminating B cells is actually not a great long-term thing,

00:55:53.080 but you can actually survive without your B cells. These are antibody-producing cells for those who don't

00:55:57.400 track immunology. So the poster child for this, of course, is CD19. And as you said, every B cell is

00:56:04.540 walking around with this marker on it. We don't have to get into why it's called CD19.

00:56:08.540 And when a subset of B cells go on to become lymphoma that is otherwise unresponsive to other

00:56:15.100 treatments, lo and behold, you could wipe out, you could basically send in someone that's going to

00:56:20.460 target every CD19. You'll get rid of the bad guys. You'll get rid of some good guys. On balance,

00:56:25.820 it's worth it for sure. But yes, what we're talking about here is a next layer of sophistication

00:56:32.960 because, for example, if a patient has metastatic lung cancer, it's not an option to wipe out all

00:56:40.580 of the lungs. But it's also a more complicated problem. Like there are many cancers for which

00:56:45.680 you could completely live without the organ. You don't need your colon. You don't need your breast.

00:56:50.480 You don't need your prostate. You don't even need your pancreas. I mean, I'll give you an example.

00:56:54.620 I think you may remember this. I have a friend with Lynch syndrome. It was unbeknownst to him

00:56:58.560 because he was adopted. Developed colon cancer in midlife. Again, a great surprise when you're 40

00:57:04.560 to develop a stage three colon cancer, but later developed a pancreatic adenocarcinoma. I sent him

00:57:10.640 to see an excellent doc who I had trained with and he was inoperable. So everybody who is familiar with

00:57:16.920 pancreatic cancer understands inoperable, locally advanced pancreatic cancer is a six to 12 month

00:57:23.940 prognosis. But because he had Lynch, I mean, this was 2012, maybe, maybe 2013. It was just around the

00:57:34.140 time that a paper had come out in the New England Journal of Medicine that had mentioned, hey, if you

00:57:40.100 have mismatched mutation genes, you might be a candidate for this new anti-PD-1. And this story

00:57:46.620 is a happy ending in that, sure enough, he got the anti-PD-1, went into a complete remission,

00:57:53.340 but now needs insulin because his immune cells destroyed every pancreatic cell in his body,

00:58:00.260 not just the cancer, but the non-cancer. Do we not have enough novel proteins on breast cells or

00:58:06.760 prostate cells that the CD19 approach is going to work anywhere else? Is that a one hit wonder?

00:58:11.220 Well, I wouldn't say that, but it's true that we're, you know, these T cells are so powerful.

00:58:15.320 You need to find handles, again, on the surface that are truly specific for cancer cells. So I was

00:58:21.960 honing in on, you know, what is truly specific for cancer cells, these mutations. It's a big hill to

00:58:26.600 climb that I haven't, we haven't gotten into in terms of developing personalized engineered T cell

00:58:31.680 therapy for the entire global cancer population. There's a cost issue there. There's a technology

00:58:36.500 issue to some degree as well. But in any case, in the meantime, what is the field focused on right now?

00:58:41.680 It's trying to identify those surface markers, proteins that are truly specific to cancer. And

00:58:47.920 that's what we've been struggling with because there are other therapeutic modalities that don't

00:58:50.900 require quite as much specificity. Like you can direct chemo even, you know, an antibody that's on

00:58:55.780 the back end of it has chemotherapy drugs that get, you know, sort of ingested by the cancer cell and

00:58:59.940 have a more localized effect. Radionuclide, so like really powerful radiation emitters on the back end of

00:59:05.480 such molecules also. And it's clear, you know, clinical data and now some improved drugs even

00:59:10.480 that are what I'm describing, that there's like a spectrum of selective expression that is not

00:59:16.240 needed there. But for cell therapy, you need it. Otherwise, again, you're going to obliterate every

00:59:21.200 single cell in the body. And here's the problem. Cancers come from us. When they're reading the

00:59:25.560 blueprint, if you will, and translating certain genes into RNA and then proteins, that comes from the same

00:59:31.900 genetic blueprint that our normal cells have. And so finding such proteins, this has been a real,

00:59:38.380 not just technology gap, but it's been a real conundrum. And like feeling blindly and just trying

00:59:43.160 a bunch of things because of the power of the killing potential of these immune cells, there's

00:59:47.920 every constituent in the field has no appetite for that. This has been where the field has been

00:59:52.420 anguishing most in terms of trying to understand if there's more CD19-like opportunities, but on common

00:59:58.820 epithelial cells where we can't destroy the normal version. We need to get to this greater

01:00:03.420 specificity. Let me just insert one final thought here, which is that the cell engineering field has

01:00:08.560 certainly advanced to the point of being able to create bifunctional recognizing elements to these

01:00:14.560 surface recognizing receptors where basically both of the targets have to be present. So like an

01:00:19.700 and switch, as it's called, like the Boolean and or. Instead of just creating a cell that goes

01:00:23.680 after CD19, you create a cell that goes after CD19 and CD20, and you only ever kill a cell that's got

01:00:29.220 both. Actually, that's not a perfect example because CD19 and CD20 are almost always co-expressed

01:00:33.360 on B cells. But in any case, you get my point that there is a fair amount of work going on right now

01:00:38.000 to try to find pairs of proteins that might only be expressed on certain cancers that might start to

01:00:44.220 give us the opportunity to take this same basic approach that's more readily scalable than the

01:00:49.800 more personalized. Well, you've got this genetic makeup of your cancer cells, and we're going to

01:00:54.600 zero in on a personalized approach that's specific to your immune system type and to that mutation.

01:00:59.320 In theory, it can be done, but we have to drive down cost of manufacture. A lot has to happen for

01:01:04.960 that to be remotely feasible, economically manageable. Let's go back to TIL for a moment. You mentioned that

01:01:10.960 they're on the cusp of receiving an FDA approval for the treatment of metastatic melanoma. So again,

01:01:16.640 just to bring people back to what that means, that means a patient with metastatic melanoma who

01:01:21.140 presumably has progressed through all other non-cell therapies and still has harvestable tumor.

01:01:26.920 This is a very important feature of TIL. You actually have to be able to surgically pull out

01:01:31.500 a large enough sample of a tumor. So a patient, for example, has cancer that has spread to their

01:01:36.480 lung. They have to actually undergo lung surgery and take out a wedge or lobe or whatever amount is

01:01:41.380 necessary. That tumor is taken immediately to the lab where all the lymphocytes that are there

01:01:47.700 are expanded and expanded and expanded. And I forget, it's been so long since I've been at it,

01:01:53.140 I think they want to get to at least 10 to the 9. Is that the order of magnitude? Okay. You expand this

01:01:58.000 to about 10 to the 9 cells and they're re-infused, usually with interleukin. And again,

01:02:03.580 you're looking for this response. It sounds great in theory. Why doesn't it work every single time?

01:02:09.220 You've clearly identified lymphocytes that know how to go there and presumably that's half the battle.

01:02:14.860 Why doesn't this work every single time? Again, it goes back to defense mechanisms to a degree. I

01:02:19.640 mean, I talked about that kind of layering of the onion metaphor before. There are layers of force

01:02:24.960 fields. It is the case that there's actually direct mechanisms that can impede killing at the tumor

01:02:31.340 cell level. And I talked about like PD-L1 being expressed on cancer cell surface. Well, it turns

01:02:35.720 out even intracellular mechanisms, so the way in which interferon, which is an immune system hormone

01:02:41.620 that basically triggers cell death, it's part of the killing process that when CDA positive T cells are

01:02:47.020 trying to kill a cell, it's like a virally infected cell, or in this case, a cancer cell.

01:02:51.120 And it turns out that to become cancers, successfully become cancers in the first place,

01:02:55.480 that you can find direct evidence, certainly in melanoma, of that cancer intracellular

01:03:01.320 pathway itself being altered, that the immune cells are actually unable to do the killing

01:03:06.560 because that cell is no longer sort of sensitive to immune cell-mediated death, which is a very

01:03:12.120 nasty little trick and one that can't be overcome just by dumping in more immune cells. At least we

01:03:16.540 don't have direct evidence that it can. This also can cause resistance to PD-1 antibodies. We've

01:03:21.320 demonstrated in melanoma and a handful of other cancer types now. So you have to start inside the

01:03:26.380 cell in terms of ways in which cancers have evolved an ability to resist immune recognition that they're

01:03:32.080 contending with. But then you've also got, you know, that recruitment of suppressive immune cells

01:03:36.800 that I alluded to before, which especially very antigenic cancers very commonly do that, need to do

01:03:42.620 that. And you can't overwhelm them just by introducing more CDA positive T cells by and large. And so in those

01:03:49.420 cancer types where those so-called myeloid cells are very, very predominant, this cell therapy has just not

01:03:55.040 taken hold at all. And then there are trafficking issues, basically features of cancer microenvironments,

01:04:01.660 some of them related to oxygen tension, some of them related to nutrient availability, that make

01:04:06.240 it very challenging for immune cells to persist, you know, multiply and do their killing work. And so

01:04:12.620 we've known for such a long time, cancer cells are metabolically inefficient, that they're living in

01:04:17.620 this incredibly harsh environment as a very low oxygen gradients. People didn't really understand like

01:04:22.680 the why of that. What you're describing, of course, stems in part from the Warburg effect,

01:04:26.400 which had always been, people had always thought, well, it must be that cancers can't undergo oxidative

01:04:31.900 phosphorylation. And that's why they're doing this inefficient thing. But to your point, between the

01:04:37.200 substrate argument, going through reams of glucose leaves you more building blocks, which is what they

01:04:42.600 need more than ATP. And then on top of that, you're lowering the pH, you're creating this incredibly

01:04:48.180 harsh microenvironment. It seems like there's every reason in the world from a natural selection

01:04:52.700 standpoint for cancer to do that. This is my point, is that I think it's the why of it. Well,

01:04:57.400 cancer cells like ultimately figure out how to sufficiently thrive, if you will, in such harsh

01:05:03.020 environments. Well, who can't survive in that very harsh environment? Well, immune cells. The idea that

01:05:08.320 this is all, much of this has to do with creating, conditioning this harsh environment as a force field.

01:05:14.820 This is another thing we are not addressing by virtue of just dumping in more immune cells.

01:05:19.840 You know, my argument is when I talk about multimodality therapy for cancer, it's, you know,

01:05:24.520 about targeting those mechanisms that we can address inside the cancer cell. It's about modulating the

01:05:28.760 environment metabolically, actually even fixing to a degree this oxygen, paucity of oxygen in the

01:05:35.720 pockets of that, as well as manipulating these other cell populations like immune cells. And it turns

01:05:40.900 out even like fibroblasts, for example, like get recruited into certain cancers, most notably

01:05:44.960 pancreatic cancer. They seem to be part of the force field against the immune system as well. So we have

01:05:49.620 to kind of knock down the force field. I mean, it's just the Star Wars analogy, right? You've got to take

01:05:53.660 out the moon that generates the force field around the Death Star before you send in your fighters to

01:06:00.760 actually try to destroy it. And so to me, we're on the verge of understanding kind of the hierarchy

01:06:06.500 of this biology and how to think about both diagnosing and then treating at this level. But

01:06:12.300 the toolbox has to elaborate, you know, much more completely. It's just, in my strongly held view,

01:06:17.440 it's not going to be four therapeutic maneuvers all in column A or four in column D. It's like one

01:06:23.920 from A, one from B, one from C, one from D. That's the type of four drug regimen that's going to eradicate

01:06:28.920 cancer. And it's not going to be one cocktail for all patients.

01:06:32.500 Let's dig a little deeper into that in terms of what the next five years might hold for us,

01:06:36.040 right? So if we're sitting down again in five years to talk about the success of the previous

01:06:41.100 five years, what's likely happened? How much further have we gone in immunotherapy, just in

01:06:46.500 straight up activating T cells, either through adoptive cell therapy via genetic engineering to

01:06:55.400 take peripheral blood lymphocytes and turn them into or engineer them into TIL. Let's put that as a

01:07:01.300 category of therapy. Let's talk about other ways to identify checkpoints or checkpoint inhibitors and

01:07:09.940 or combat the tumor suppressor cells. So call that the sort of tumor suppressing environment,

01:07:16.260 go after it. How much of it is going to be in the metabolic environment or the interstitial

01:07:22.620 micro environment and targeting the hostility? And then how much of it is going to be inducing

01:07:28.300 mutagenesis? So again, you and I spoke about this probably a couple of years ago, and I ended up,

01:07:34.660 I think I was able to keep it in the book. I know there's always so much pressure you're trying to

01:07:37.720 chop stuff out of the book. So I don't remember if this finally made it in, but I at least referenced

01:07:41.720 one study that had taken patients, I think with lung cancer, none of them had any PD-1 activity.

01:07:48.740 And then a course of platinum-based chemotherapy all of a sudden rendered a subset of them

01:07:53.160 to now be susceptible to it. In other words, using a conventional chemo increased immunosusceptibility,

01:08:00.100 even though the conventional chemo itself wasn't particularly responsive. And again,

01:08:04.160 there's lots of ways to go about doing that. Paradoxically, you could almost imagine

01:08:08.560 taking a cancer cell and exposing it to more mutation-forming insult. And again, I'm sure there's

01:08:15.280 other ideas, but keeping the timeframe short, which is five years, what are we going to need to do to

01:08:19.680 double the response rate, the durable response rate? Within a five-year horizon, I think let's

01:08:24.560 look backwards briefly. Over the past eight years, we have exhaustively tried to find other gas pedals

01:08:31.340 and brakes on immune cells, CDA positive T cells, most notably. And we know what those gas pedals

01:08:36.120 and brakes are on those cells. And we have tried drugging those typically on top of PD-1 antibody

01:08:41.420 therapy. And that has almost completely systematically failed. Now it doesn't interestingly produce horrific

01:08:46.840 toxicity. In other words, the immune system doesn't get so hyperactivated, like that's not the problem,

01:08:51.040 but it just hasn't moved the needle. Now I will caveat that by saying that those approaches have

01:08:56.600 been used without any notion of trying to like zero in on individual patients and sets of patients for

01:09:04.400 whom that new immunologic mechanism was hypothesized to be uniquely suited. In other words, we've been

01:09:10.540 throwing a lot of spaghetti at the wall and hoping things would stick by just treating a broad array of

01:09:15.680 different cancer patients with absolutely no molecular selection, even though there are certainly

01:09:19.900 there were and remain hypotheses along those lines that were never really tested. So anyway, just trying

01:09:25.120 to hyperactivate T cells with drugs, I would say we've kind of played that out. And it's hard to

01:09:29.040 imagine the only way to revolutionize that would be what I just alluded to, which is really tightening

01:09:33.080 or sharpening our lens, if you will, and focusing on applying those drugs in very specific patient

01:09:39.560 populations. Beyond that, there's what I would say a related class of therapies, the metabolism

01:09:44.060 targeted therapies, and epigenetic targeted therapies, which have been exploding in terms

01:09:48.420 of understanding how the blueprint, the genetic blueprint is sort of folded up and unfolded,

01:09:53.720 the regulators of that, and the way in which cancers, many cancers, like need to figure out

01:09:58.320 how to co-opt or take over the function of some of those folders and unfolders. And so there's been

01:10:04.580 a real explosion in novel very early in development drugs in that class. And it turns out, interestingly,

01:10:11.440 that altered metabolism, so the Warburg phenomenon that you alluded to, and the regulators of that

01:10:16.180 switch, those have become elucidated in a much more complete way in fairly recent years. Many people

01:10:22.120 would have thought, well, you can't target metabolism and get away with that, right? Because every cell in

01:10:25.940 the body needs to be able to regulate its metabolism in a, you know, kind of condition-dependent way.

01:10:31.100 That's true. But cancers really do, they very much depend on this metabolic dysregulation. And we think

01:10:36.900 that we're on to some of the unique regulators that cancers particularly co-opt. I would pay a ton of

01:10:43.080 attention. I mean, our group, our therapeutic development work is really quite focused in that

01:10:46.960 area. Can you give us a bit more of a sense, Keith, of what that looks like? So we know that just from

01:10:51.500 a glycolysis standpoint, we know cancer basically is a one-trick pony, most cancers, right? They're

01:10:56.000 turning glucose into pyruvate all day, every day, independent of how much fatty acid is available and

01:11:01.920 independent of how much oxygen is available. And they have perfectly healthy mitochondria. People

01:11:06.040 used to hypothesize the mitochondria were defective. That's the way they were doing it. No evidence that

01:11:09.780 that's the case. So let's just play out what you're saying. You could take something really

01:11:13.800 draconian and say, okay, there's an end. So we're not going to interfere with any enzyme that turns

01:11:18.220 glucose into pyruvate. That would be a bad idea because you have to do that if you're healthy. So where else

01:11:24.020 could you target where you disproportionately hurt a cancer cell without hurting a non-cancer cell that's

01:11:30.020 undergoing glycolysis? Our group just published a paper on this topic just five months ago,

01:11:34.960 looking quite broadly to understand these metabolic regulators and which ones cancers seem to

01:11:39.980 selectively use. And interestingly, this analysis was focused on immune cell recognition versus lack

01:11:46.940 of recognition, kind of the interplay between these two things. So we already laid out the argument of

01:11:51.720 the idea that cancers, it seems, in part adopt this inefficient metabolic strategy because it allows them to

01:11:57.720 kind of suck in available nutrients and keep them away even from immune cells. So we were trying to

01:12:02.660 unpack that. And basically, when you look in an unbiased way at all of the gene products that are

01:12:08.600 expressed in cancer cells differently than normal cells, what you see is that it's outside the

01:12:13.940 mitochondria. So inside the mitochondria, I'm 100% with you. Basically, you can't poison the factory

01:12:18.780 in that way. But it turns out that not only the function of mitochondria, but also just like the

01:12:23.040 production of mitochondria. So mitochondrial biogenesis, it's called. Like there's many

01:12:26.460 different mitochondria per cell. Different cell types need different numbers of them based on

01:12:31.440 their metabolic demands. So cancer cells will actually regulate the amount of mitochondria they

01:12:36.220 have through these outside of mitochondria programs, if you will, transcription factors in many cases that

01:12:42.220 regulate the program in the genome. This is the nuclear genome, not the mitochondrial genome that

01:12:47.600 regulate this process. There's some switches there. And one of those switches basically jumped out of

01:12:52.800 this analysis as like the top differentiator, if you will, expressed in cancers and not in others.

01:12:58.160 Now, it's the type of molecule that historically has been thought to be challenging to create a drug

01:13:02.180 against. There actually is a proto-drug against it that's still preclinical. But moving forward,

01:13:07.020 we've been collaborating academically with that company. And so early days in terms of knowing

01:13:11.520 whether this is really going to bear out. But these are the types of insights we just didn't have

01:13:15.060 five and certainly 10 years ago that there might be ways to actually kind of laser in on the

01:13:20.680 regulators and metabolism that cancers are most potentially vulnerable to. And I'm not suggesting

01:13:25.140 these are going to be standalone approaches, as I said before. It's rather than a potent... Yeah,

01:13:29.020 they're going to potentiate these other therapies. And let me just kind of make this statement to that

01:13:33.000 point. When we look at what drives resistance to both targeted therapy, so those molecules I referred

01:13:39.040 to before, these surface receptor and downstream molecules that have been successful and extend

01:13:43.600 people's lives with cancer, and immunotherapy, and we look at common themes in terms of resistance,

01:13:48.760 this metabolic switch, like using oxidative phosphorylation when they weren't using it

01:13:53.660 before, that's a very common theme in what we call the persister cell population in both therapy types.

01:13:59.620 And so the idea that you would then potentiate simply what we've already got with this class

01:14:03.220 of therapies go from 20% of cancer patients having long-term survival to 40% of them making that number

01:14:07.920 up just by figuring out this piece of the puzzle. I think that's very much in view. Now, we might have

01:14:13.880 to toggle upstream, downstream, like play with where it is that we're ultimately poisoning this

01:14:19.260 process. And we may have to do it just periodically. Like in other words, not constant like drug exposure

01:14:23.520 all the time to be able to get away with it, which is a common theme in terms of thinking about four

01:14:27.960 drug regimens for cancer. But let's come to your idea of actually taking advantage of this very delicate

01:14:34.400 balance, if you will, where cancer cells have accumulated genetic alterations to a degree that's

01:14:39.900 supposed to be intolerable for a cell's survival. In other words, if you can't repair mutations and

01:14:45.840 alterations that have been caused, let's say, by acute exposure to something like radiation, for

01:14:50.720 example, where you get a lot of mutations all at once, we have repair mechanisms. But if they don't

01:14:54.620 do their job, then a cell basically has a program by which it commits suicide, so-called programmed cell

01:14:59.900 death. And basically, cancer cells live dangerously on the edge, if you will, in having accumulated

01:15:05.540 these mutations in certain cancers, like with your friend, with Lynch. I mean, wow, the number of

01:15:10.160 mutations that accumulate because of the defective machinery is just off the charts, like ultraviolet

01:15:14.820 radiation-associated skin cancers, also off the charts. In any case, the point is that we know that

01:15:20.360 actually, if you introduce more mutations into those cells, like in the laboratory, like you push

01:15:26.160 them over the edge. There's a limit to how much, to what they can handle. So how about combining that

01:15:31.680 concept with what we were talking about before, immune system recognition of mutated proteins,

01:15:35.980 and just say, hey, okay, you want mutations? And again, going back to my anthropomorphic,

01:15:40.600 now we're talking to the cancer cell. You want lots of mutations because it helps you dial the

01:15:44.640 combination lock, if you will, and become a cancer? Fine. You know, we're going to not just double,

01:15:48.880 we're going to 10x the number of mutations you have, both to increase immune recognition and possibly

01:15:54.120 also just simply push them towards self-death. Yeah. That concept is behind platinum-based

01:15:59.380 chemotherapy's effectiveness in cancers that are somewhat deficient in repairing their genomes,

01:16:04.740 right? So that's a link that we've known about now therapeutically for a number of years. PARP

01:16:08.600 inhibitors, that's a DNA damage repair enzyme, PARP. And inhibiting its function can push certain cancers

01:16:14.960 that are close to the edge, if you will, over the edge. So there's already some direct evidence that we

01:16:18.940 can get that benefit. The immunologic piece, that requires another layer of complexity, which is that

01:16:24.480 basically you would need to introduce the mutations and ones that are shared across the whole population

01:16:31.120 of cancer cells. And if not the whole, then nearly the whole. So the immune system actually,

01:16:35.420 interestingly, is able to elaborate immune responses that become broader, right? So this is,

01:16:40.040 you know, sort of epitope spreading, as it's called, where the immune system latches onto a certain

01:16:43.640 antigen in mounting an initial immune response, but then actually can bring in reinforcements that are

01:16:48.700 recognizing other antigens and create a more sort of polyclonal response, what was initially a

01:16:53.220 monoclonal response. So that's part of innate immune function, but there's pretty good experimental

01:16:57.380 evidence that you have to start with something that's at least shared in 95, 98, maybe even 99%

01:17:02.820 of cancer cells. So this is where the idea of like, you know, using, let's say, radiation to treat like

01:17:07.580 a single site of metastatic cancer in someone who's got 20 sites, we've tried this. It hasn't worked.

01:17:12.940 There are very sporadic cases where it actually can trigger a much more profound immune response that's

01:17:18.240 actually systemic, like that goes after all the tumor sites, but that's quite rare.

01:17:23.400 Again, just for folks to make sure we're following, the reason it would be rare is

01:17:26.640 if you only introduce a whole bunch of mutations to 10% of the tumor, you might generate a new immune

01:17:32.820 response. You might kick the tumor over the edge either by having so many mutations that it all

01:17:38.140 undergoes programmed cell death, or it now finally rises to the level of detection, but that's not

01:17:43.640 sufficient enough across the entire organism. Yeah, it won't clear the rest of the tumors.

01:17:47.780 Then the medical oncologist, you can imagine this is where my mind commonly goes in, that we have to

01:17:52.240 come up with a systemic approach. There's some really fascinating data that a colleague of mine at

01:17:57.120 Mass General is about to publish. And it would suggest that basically you can incubate cancer cells,

01:18:04.040 and by incubate, I mean actually in living being, with mutation-inducing drugs, aka chemotherapy drugs,

01:18:10.860 certain chemotherapy drugs. But for that to work, you need to actually be, you have to kind of pin

01:18:15.880 them down with another therapy first. So some of the therapies we've talked about already that like

01:18:19.960 actually are effective, partially effective for a period of time, months to many months in some cases

01:18:26.360 before resistance might manifest to some of these targeted therapy approaches. If you pin them down

01:18:31.380 with that therapy and incubate in, you know, these chemotherapy drugs that basically start to dial in more

01:18:38.080 and more mutations, at least in mouse models, it would appear that actually you can buy the time that you need

01:18:44.040 to be able to actually introduce new mutations and have that trigger immune recognition and make even, you know,

01:18:50.500 PD-1 antibody-based therapy much more effective. So we're going to try that idea in human beings, basically

01:18:56.100 taking so-called oncogene-targeted therapy, backbone treatments, and then using what are called

01:19:01.560 alkylating-agent chemotherapies, which are the ones that can introduce new mutations most commonly. And

01:19:06.840 even at somewhat low doses, it would appear, you potentially can introduce the mutations without

01:19:10.740 having some of the deleterious effects that chemotherapy drugs are well known to cause.

01:19:14.560 Keith, I'm going to change gears for a moment only because I know we have a very short clock today

01:19:19.120 relative to how long you and I could normally speak. And I want to talk about another very important

01:19:23.260 topic. So to introduce it, let me share with you some stats that you know better than I do,

01:19:28.940 but I'll let you interpret the stats for the listener. If you take a person with stage three

01:19:33.900 colon cancer, so this person has cancer in their colon, it's even spread to the lymph nodes of the

01:19:40.080 colon, but to the visible eye, it has spread no further into, there's no radiographic evidence that

01:19:45.360 it's anywhere else. You're going to put that patient on a fancy regimen of chemotherapy. I don't

01:19:49.820 have to spell out full Fox and all that stuff, but there's a regimen of chemotherapy you'd put that

01:19:53.840 patient on. How many of those patients are going to be alive in five years? 60, 70% of them?

01:20:00.440 Yeah, that's about right. Again, it all depends on the size of the initial tumor and other features,

01:20:04.420 but that's about right. Yeah.

01:20:05.360 Let's now take that same patient in a way, except he also has cancer that has spread to his liver.

01:20:13.620 So you're going to go ahead and cut the colon out, take those lymph nodes out. But on the CT scan,

01:20:17.900 you're going to notice that he's also got metastatic cancer. So one of them is stage three,

01:20:22.260 one of them is stage four. We're going to give that stage four patient the same chemotherapy.

01:20:27.380 We're going to give them the same drugs. But in five years, somewhere between none and a few

01:20:33.780 percent of those patients will be alive. And if you wait to 10 years, it's none.

01:20:37.020 Yeah. Yeah.

01:20:37.600 What's a decent explanation for that observation? Which, by the way, if we had more time,

01:20:41.780 we could tell the same story for every cancer, basically.

01:20:45.200 In other words, what I refer to as microscopic residual disease, why is it that we're actually able

01:20:49.160 to eradicate microscopic residual disease with the same drugs that don't do the job when you have-

01:20:54.120 Macro disease. Yeah.

01:20:55.380 Yeah. Like I was going to say-

01:20:56.260 In other words, why does it work when you have hundreds of millions or billions of cells,

01:21:00.560 not all clumped together, but sort of diffuse, but when you have like a hundred billion cells

01:21:05.400 and they're like in big visible clumps, the same drugs just fail?

01:21:09.940 There's, I would say, two prevailing explanations. There are hypotheses. I mean, frankly,

01:21:15.140 because they'll become explanations once we actually connect the dots and really prove that

01:21:19.600 we can demonstrate our knowledge by curing more patients with this. So one is basically just a

01:21:24.540 clonal heterogeneity concept. So basically as cancers evolve, we used to think that cancer cells

01:21:29.600 were kind of identical clones of one another. Like they're just a massive number of absolutely

01:21:34.400 identical cells. That in the beginnings of cancer, that is largely true. But as cancers continue to

01:21:40.300 evolve in our bodies, they actually keep mutating. And so you start establishing subclones. You can

01:21:46.080 have a dominant subclone. That's typically the case. Like that might even be 99% of cells.

01:21:50.120 And then in that remaining 1%, you might have 10, 20 subclones. We've proven now that certain

01:21:55.480 therapies actually are able to pick off the 99%. They leave the 1%. And then somewhere in that 1%

01:22:00.440 is a clone that has a resistance mutation, like already in it to the drug that we're giving.

01:22:05.800 So there's a clonal heterogeneity hypothesis that I would say is quite strong at this point,

01:22:10.500 because of some of the evidence I just alluded to, that that's a big part of the problem.

01:22:14.140 If you nip it in the bud, if you will, with offering the same therapy when there's not so

01:22:18.440 much clonal heterogeneity, that represents a curative opportunity. And so when I was talking

01:22:22.940 before about these so-called oncogene targeted therapies, which is not all of the targeted therapy

01:22:27.440 successes we've had, but the ones that go after these mutated activated proteins,

01:22:31.640 those growth factor receptors and downstream ones in particular, it's very clear. You can cure a

01:22:37.500 trivial fraction of patients with overt metastatic disease, and you can cure a pretty substantial

01:22:41.140 fraction of patients in the so-called adjuvant setting, so microscopic residual disease setting.

01:22:45.560 And so we have direct evidence that this phenomenon occurs, but you're asking the why question.

01:22:50.900 It's, we think, some contribution or some part explanation from having to do with lack of

01:22:56.640 clonal heterogeneity. The other is the secondary immune response concept, right? That basically

01:23:02.160 all successful curative cancer therapies actually do trigger immune recognition through what's

01:23:08.880 referred to as immunogenic cell death, so that you're killing the cells directly with your drugs,

01:23:13.180 but that the mop-up work, if you will, of actually eradicating every last single cell

01:23:17.500 is the immune system's job. It's a concept that was first introduced when we had just these

01:23:22.560 conventional chemotherapy drugs from the 1900s. And now we actually have more and more evidence that

01:23:28.240 our sort of more elegant molecularly targeted drugs actually engender these types of better

01:23:33.440 immune system recognition phenomena as part of their mechanism of action. And because we've directly

01:23:39.380 demonstrated that, like better immune recognition in patients who are receiving these therapies,

01:23:43.080 looking at biopsies compared to pretreatment, and that that happens rather quickly, you know,

01:23:47.680 I think it's reasonable to then overlay that on top and say, well, yeah, extending your spontaneous

01:23:52.860 remission starting point from a while ago in discussion, that basically you're allowing this kind of

01:23:58.580 tipping point phenomenon to occur. Yes, you're directly killing cells with these drugs, that's

01:24:02.720 true, but the eradication piece is ultimately an immune system phenomenon.

01:24:07.100 And I think there's kind of a hybrid there too, right, Keith, which is that in the micrometastasis

01:24:11.720 environment, in the adjuvant setting, you have less capacity for the tumor to create the hostile

01:24:17.140 environment in which to impair the immune system from mopping up the damage. It all favors keeping

01:24:25.380 the cancer cell on its heels. And the way to do that is to just have as little of them as possible

01:24:30.740 is going to increase our odds. You still have to win. I mean, you still have to kill because if you

01:24:35.540 don't, it will get back onto its toes. That's right. I mean, this is what we're talking about is behind

01:24:39.840 this massive wave of enthusiasm, and it's legitimate enthusiasm, not hype, that early detection is going

01:24:48.000 to allow our same toolbox of drugs to be massively more effective. That's the only reason I posed the

01:24:52.920 question. You're taking the cue and you're running with it. Right. We're pretty terrible at that as it

01:24:57.460 stands right now. Many of your audience know, basically, we can only screen, we only have real

01:25:03.100 direct evidence of effective screening for a few cancer types. I mean, cervical cancer, for sure,

01:25:09.360 but we could also hopefully eradicate cervical cancer by getting everybody vaccinated. HPV vaccine.

01:25:13.780 Exactly. But cervical cancer screening absent a vaccine is quite effective. So that's one. Breast

01:25:18.820 cancer, for sure. We can cut the risk of breast cancer death by about a third with mammography.

01:25:23.500 That's not a very inspiring number. I mean, I absolutely suggest that everyone who's eligible,

01:25:28.140 who you care about, you should strongly insist that they get mammograms. Colonoscopy and other

01:25:34.140 less invasive means of detecting colon cancer can reduce risk of colon cancer death by about 25% to

01:25:39.620 30% ballpark. I mean, the most optimistic estimates would be about a third also. And there again, I'd say,

01:25:45.620 well, that's a real number. I've had my first colonoscopy, and I'll keep doing them. But that isn't,

01:25:50.740 again, particularly inspiring. And despite lots of effort and, I would say, lots of controversy,

01:25:55.600 prostate cancer screening is really quite poor. It's almost like just non-randomly assigning people

01:26:00.520 to get prostate biopsies, you know, getting PSA tests, basically. It's not a great way of

01:26:04.480 detecting cancers and certainly not potentially dangerous prostate cancers.

01:26:08.460 But I would add something to that, Keith, which is that all of those three, the three last ones,

01:26:12.940 which are three big cancers, those are three of your big five, they all have something in common.

01:26:18.400 So if you look at mammography, infrequent colonoscopy, and PSA, I would make a case that all of those

01:26:25.180 are not great screens by themselves. And I'm sure you would agree. So in other words,

01:26:29.720 people often confuse, and unfortunately, this is true of physicians and policymakers more than it

01:26:35.500 is patients, because I think the patients are looking to those of us who think about this for

01:26:40.260 input. Patients confuse, or policymakers rather, and physicians confuse the statistics you rattled off

01:26:46.220 as proof positive that early screening doesn't justify the cost.

01:26:50.040 A different way to say it is, no, mammography used in isolation, which has its blind spots,

01:26:58.940 is not a monotherapy. PSA by itself, as you said, is shy of a random number generator.

01:27:06.720 But that doesn't mean that adding ultrasound or MRI to the breast surveillance program won't

01:27:13.580 dramatically, by stacking tests with different sensitivities and specificities, right?

01:27:18.480 Mammography, exceptional for small calcified lesions, works poorly in hyperglandular tissue.

01:27:26.540 The exact opposite is true with the MRI. Similarly, with the PSA by itself, virtually meaningless,

01:27:32.740 but PSA density, PSA velocity now adds much more specificity. Furthermore, you start to add things

01:27:40.580 like a 4K, and if the risk is high enough, you get a multi-parametric MRI. I'll tell you this,

01:27:46.160 Keith, I mean, I'm not telling you anything you don't know, but I think, again, just for listeners,

01:27:49.600 in 10 years, I have not had one patient get a prostate biopsy that wasn't warranted. I'm not

01:27:55.840 a superstar. It's not like I've got some... No, it's just that we're doing this, we're not just using

01:28:01.240 PSA. Sometimes we've had patients who only get picked up on PSA velocity. Their PSA is not high

01:28:07.880 enough to trigger the 4K. No one would go and do anything based on their... So I get a little

01:28:14.840 frustrated when the medical community that's anti-early screening or screening and early

01:28:19.880 detection pooh-poohs it based on what I still think are impressive numbers, the number you state,

01:28:25.220 because that's sort of like saying, you know, there's too many fatalities in cars we shouldn't

01:28:29.720 drive. It just doesn't make any sense. It's like, yeah, there are fatalities in driving. Let's figure out

01:28:34.960 ways to drive better. You can put a seatbelt on, you could not drink while driving, and you could

01:28:40.500 mind the speed limit. That's a totally different situation than saying we're going to abandon all

01:28:45.800 those things. So... You know, you're reminding me that I, in my world, you know, being an oncologist,

01:28:50.000 I don't have to contend with the community you're referring to. I take those numbers as being like

01:28:54.820 absolute support and endorsement. But part of what I'm getting at is the remaining unmet need. And it's

01:28:59.600 into that massive unmet need that there has been just an enormous advance in terms of methods,

01:29:04.960 for detecting single alleles, single fragments of genes in the bloodstream. So it turns out that

01:29:12.220 normal cells shed DNA in the bloodstream. It is digested and broken down reasonably quickly,

01:29:18.340 but not immediately. Cancer cells do this also, as it turns out. And the more cancer you have in

01:29:23.780 your body, of course, the more... It's not so obvious, but it is true that the more cancer that

01:29:27.800 is in the body, the more of the copies of cancer DNA that will actually be shed in the bloodstream.

01:29:32.500 But sequencing technologies have advanced to a degree that now, you know, from a single

01:29:39.200 10 milliliter tube of blood, and particularly one collected over time, so kind of analogous to your

01:29:44.320 PSA velocity example, where you're sampling at multiple time points. If you sample at multiple

01:29:49.480 time points now and subject those, I don't mean just to the methods that are being commercialized now,

01:29:54.840 but are being commercialized. I mean, since we talked four years ago, what felt like very much a

01:29:58.860 research method is now emerging as a real clinical option. There's methods now that can find cancers

01:30:05.000 at an earlier point, and a broad array of cancers, like way beyond just the cancer types that we're

01:30:09.040 talking about. The ones for which we have screening methods, I mean. So really, you know, almost pan-cancer

01:30:14.600 tests. But in R&D mode, right behind them are 10x, 100x more sensitive methods that are absolutely

01:30:21.740 going to move the needle in terms of our ability to find cancers at a microscopic point. Now, here's the

01:30:27.220 problem. The problem is, at a microscopic point, what do you do? Where do you direct the scalpel?

01:30:32.400 I mean, this is a fundamental conundrum. So you overlay on top of what I just described, the fact

01:30:37.300 that on circulating tumor DNA, as it's called, you can actually do more than just sequence for

01:30:43.700 mutations to find that it's circulating tumor DNA as opposed to normal DNA. You can also look at what

01:30:48.820 are called methylation patterns, which has to do with this kind of like folding and unfolding of a

01:30:52.700 blueprint, these molecular modifications that exist in certain cell types. And basically, if you find

01:30:58.080 mutated sequence of DNA, and it's got the methylation pattern of a colon epithelial cell, guess what

01:31:04.560 cancer you probably have. And while you can't direct the scalpel right away, obviously, you can do a

01:31:08.120 colonoscopy. But similarly, for breast cancers and others, where you can then start to focus your

01:31:12.300 attention, right, with imaging analysis to try to detect the cancer, or maybe not the moment that the

01:31:16.920 blood test is positive. Maybe it's going to take you 6 months, 12 months, 18 months of continued

01:31:21.060 surveillance, and then you'll find it at a much earlier point than you ever would have found it

01:31:24.980 based on our other methods. So that's one sort of paradigm. And that's where we are right now with

01:31:29.960 the adoption, early adoption of methods as they exist now that are getting rapidly better in terms

01:31:36.000 of increased sensitivity. So there's been a real explosion in terms of investment in this area,

01:31:39.900 and now scale up of technologies that are commercially relevant. But the other concept I wanted to just kind

01:31:45.280 of weave in here is, I think, where you kind of started this set of questions, which is that

01:31:49.680 basically, in certain instances, we're going to find targetable mutations. And by that, I mean,

01:31:54.660 with drugs or with immunotherapies, where basically, you know, we'd say, well, look, we see it in the

01:31:59.700 blood. We actually, with our best available scanning technology, we can't actually see it in a way to

01:32:04.220 direct a scalpel, but we actually know what drug to give you to eradicate your trivial amount of cancer.

01:32:10.720 You know, using the analogy that we started with here, which is like clinically overt

01:32:14.540 metastatic disease versus, you know, microscopic disease that remains after surgery, aka adjuvant

01:32:20.540 setting. But now finding cancer is at a point where there's many fewer of these cells, and where,

01:32:25.160 you know, the defense mechanisms, the force fields, the heterogeneity that we talked about before don't

01:32:28.700 exist. And so this is, I mean, a real reason for optimism. I should just highlight that there's

01:32:33.140 two applications here. One is actually to do much more precise therapy in the post-surgical setting.

01:32:39.880 So really figuring out, you know, right after surgery, who still has microscopic disease in

01:32:44.340 them? Who doesn't? That's a much easier problem. I mean, I actually talked with Max Dean about that

01:32:48.760 problem, and he's one of the pioneers in that field. And of course, not to minimize the amazing

01:32:54.400 breakthroughs there, but there you know what you're looking for. You've taken out the patient's lung

01:32:58.860 cancer. You know exactly how that lung cancer differs from a non-cancer lung cell. And you're out

01:33:05.920 there looking, and you're right. I mean, this now becomes the most elegant way for post-treatment

01:33:11.100 surveillance. But it's what we started with that is the much more difficult problem, and frankly,

01:33:17.300 the most important problem. I mean, if you solve this problem, I don't know that the other things

01:33:22.300 matter anymore. Oh, no, no. If you solve this problem, you win the game. We've always been stuck

01:33:26.380 in this mode in cancer research and therapeutic development, kind of start with the worst case

01:33:30.740 scenario, if you will, right? And then... And that's where you have to learn. That's right. And that

01:33:34.340 the same therapies that are somewhat effective in the overt metastatic setting are much more

01:33:38.400 effective in the so-called adjuvant or post-surgical setting. And we have every reason to believe

01:33:41.840 they're going to be at least as effective, arguably more, when we're down to two logs,

01:33:46.360 three logs, even fewer cells at the time that we're finding and intercepting the cancer, when

01:33:50.820 the immune system is still quite competent, is still actually doing much of its job.

01:33:55.000 What's the state of the market today, Keith, in terms of tests that people listening to this

01:34:00.700 can actually get as part of a cancer screening protocol? So GRAIL has a commercially available

01:34:06.580 kit. It's not over the counter. You need to get it through a physician. What are some other tests

01:34:10.880 out there? And in your view, how close are we to these tests being an imperative part of cancer

01:34:18.220 screening? I think we're not quite at the imperative point, but we're certainly at the point where I

01:34:22.300 would say it's reasonable to get a test. If you consider yourself an early adopter and you know who you

01:34:27.100 are, it's reasonable to get a test. And just quickly to answer your question, first part of

01:34:30.980 your question, there's a test that was initially developed by a company called Thrive that was

01:34:34.300 acquired by Exact. They have a commercially available test as well. And then another company called Delphi

01:34:39.100 have a commercially available test, which all have performance characteristics that are

01:34:42.800 in the same realm in terms of supporting their current clinical use. But here's the concern that many

01:34:49.460 people have, which is that basically we're at a state in the field where finding people who are

01:34:55.220 blood positive, blood test positive could lead to a large degree of anxiety in terms of then you do

01:35:02.780 standard radiographic assessment, you don't find the problem. And then basically the medical community,

01:35:07.340 because we're not talking about oncologists who are doing the tests, the medical community really

01:35:11.600 hasn't had time to really kind of work out the kinks of how do we manage this situation? And so

01:35:16.560 you can almost argue that there need to be generalists who really develop expertise, content knowledge,

01:35:23.160 have a network of specialists that they can work with to be able to kind of catch these patients.

01:35:28.480 And these tests were launched before that was really created. So that's just my PSA here,

01:35:32.640 my public service announcement, in terms of the fact that if you just get a test and you're not in

01:35:37.360 the hands of someone who can manage a positive test, that is at least anxiety provoking. We've,

01:35:43.320 with collaborators of ours, we've actually been doing like direct head-to-head comparison analyses,

01:35:47.880 like of how much lower can we go in terms of, you know, amount of tumor DNA in the blood that can be

01:35:52.960 detected with, you know, what are currently R&D methods, but are readily scalable. We're really

01:35:57.500 at a hundred X better. At that point, you're talking about a finger drop of blood. If you're

01:36:01.720 a hundred X better than 10 ML, you mean? Okay, that's true. You could take it in that direction,

01:36:06.340 but we actually... You would say, no, stay with 10 or 20 ML and we are way more sensitive.

01:36:11.200 Yes. Oh, exactly. Exactly. So that's the point. And then do serial analyses,

01:36:15.060 go with high risk populations to prove this point if you like, but we can readily envision how it is

01:36:19.460 that we basically start to capture, you know, a much bigger section of the population. It's a

01:36:23.640 little hard to estimate right now, basically, until we do more studies.

01:36:26.580 What's differentiating these companies? So if you just look at Delphi and Grail. So Grail,

01:36:32.120 to me, I have no interest. I have no affiliation with any of these companies. We use Grail in our

01:36:36.860 patients. When we do, we don't do this in everybody for exactly the reasons you've stated. You got to be

01:36:42.040 able to tolerate the noise that may come of these tests. That's sort of our view of aggressive

01:36:46.740 cancer screening in general. But when we use it, we do use Grail. And that's largely based on the

01:36:50.940 affiliation with Illumina, which is if you've got the best sequencing company in the world that

01:36:55.140 created the engine for this thing, then that sort of makes sense to me. But what else differentiates

01:37:00.400 these companies? I mean, there's three kind of aspects to circulating human DNA that we now know

01:37:04.920 if you pay maximal attention to, you can increase your sensitivity. You can find more cancers.

01:37:10.040 So I started with mutations. That's kind of the first principle. Then comes the so-called

01:37:14.840 fragment length. So fragmentomics, if you will, or its own field, separate from first-generation

01:37:20.440 genomics. And that Delphi basically came out of that scientific discovery that circulating

01:37:25.740 DNA basically comes in different fragment sizes than normal cell DNA. And that basically,

01:37:31.980 this is a kind of population phenomenon. You're measuring multiple or many, many circulating

01:37:36.180 DNA fragments and tuning your algorithm ultimately to be able to kind of find the sweet spot of

01:37:41.800 differentiation. That's definitely part of the formula. And I would argue that just everybody's

01:37:45.860 going to rise to that inclusion of that method. And then that methylation aspect that I talked

01:37:50.380 about before, that's the other feature that has been a differentiator in terms of kind of the first

01:37:55.360 marketed products in this class, even across these three companies. But there are 10 more companies

01:37:59.940 coming right behind. So basically, the first one is kind of not valuable for pan-screening

01:38:05.600 because that's how you're checking for recurrence when you know the mutation, right? I mean, we're not

01:38:09.240 going to be able to screen people on the basis of guessing cancer mutations, are we?

01:38:14.560 Some would argue we can. I mean, the cost of sequencing continues to nosedive, believe it or not,

01:38:18.980 still going down lower and lower costs per unit of sequencing done. There are some who argue

01:38:24.800 actually, no, no, we just go after, you know, pick a number. You know, the thousand most common

01:38:28.400 cancer mutations, the 10,000 most common, the 100,000 most common cancer mutations that we

01:38:32.800 actually can... Yeah, I mean, I guess if you had KRAS and you had... And P53. Yeah, P53 and BREC.

01:38:37.820 P53 is mutated in 50% of all cancers. Now it turns out that there's hundreds of different

01:38:41.660 things with three mutations. It's a big gene. But still, people used to object to that concept based

01:38:46.660 just on a sequencing cost argument. But that, I think, is becoming less and less relevant. So that

01:38:51.740 first feature doesn't require that customization, which is what's being done in the post-surgical

01:38:55.940 setting. You alluded to that, but just to make sure that people understand that... Yeah, good point.

01:38:59.820 When you know what mutations exist in the resected tumor, it does allow you to create

01:39:03.400 very, very sensitive tests for that patient. And that's what's being done commercially are these

01:39:08.260 bespoke assays based on what comes out in the surgical specimen. But when you don't know what

01:39:12.320 you're looking for, the argument is if you do enough sequencing, you'll find them.

01:39:15.880 Are there companies or is that all done in the lab right now? Are there companies that are

01:39:19.300 actually taking... No. It's not yet commercialized.

01:39:21.680 No. But there's a real... I mean, I mentioned this briefly before. There's just a real scale

01:39:26.060 up in investment in this area, which is incredibly heartening to see. I mean, I used to complain

01:39:29.980 for so much of my career about the fact that diagnostics just didn't get the same investment

01:39:34.880 that therapeutics got because the return on investment just fundamentally different, like

01:39:39.080 fundamentally different between the two domains. And yet, as clinicians, we need the diagnostics.

01:39:44.100 We can't even think about therapeutics until we basically diagnose.

01:39:47.660 That's the irony of it, right? Is that people talk about, oh, we'd really love to lower

01:39:50.680 healthcare costs. Yeah. You need earlier and better diagnostics. It's just such a no-brainer.

01:39:56.520 I think it's wise to be upset about the cost of oncology therapeutics that are adding no value.

01:40:02.360 But you spend a tenth of that on the diagnostics, you make that problem irrelevant.

01:40:06.340 That's right. And then the durations of therapy that we need to give people to have curative outcome.

01:40:10.640 You solve so many problems by virtue.

01:40:12.500 You just say we're turning this into a lock and key model from diagnostic to therapeutic. Keith,

01:40:18.020 we're just about out of time. So I want to kind of end with a question that you may not be able to

01:40:21.420 answer, but it's worth asking anyway. I think of you and people I know like you as the most remarkable

01:40:26.580 oncology advocates, meaning I know that if one of my patients comes down with cancer, I can call you up

01:40:35.620 and say, Keith, I've got this woman. It's a very unusual breast cancer. It's her two new positive,

01:40:41.940 but ERPR negative. There's something funky about it going on. Who do you like? Who should she be

01:40:48.540 seeing? And let's say there's another situation where a traditional therapy is failing and you're

01:40:53.760 going to point me in the direction of where there's a clinical trial that's promising, not just a phase

01:40:58.120 one that's like probably got no hope, but here's a phase two that really has some hope. Okay. So

01:41:02.940 there should be an entire industry of Keith Flaherty's who are there to be consulted by

01:41:11.600 families who find themselves in this situation. Because again, we come back to how we started this

01:41:17.120 discussion. There's nobody listening to us right now or watching us right now who hasn't been touched

01:41:23.460 or will not be touched by cancer. And even if it's a cancer that ultimately doesn't kill them, which

01:41:28.880 again, in about half the cases, it won't actually kill you, you will need help navigating the system.

01:41:34.960 And the disparity in cancer care in this country, and probably in most countries, is significant.

01:41:41.700 And therefore it does matter who you know. It does matter which expert points you to the best

01:41:48.180 treatment center. Because I can't clone Keith and a dozen other people that I know that I can pick up

01:41:54.920 the phone and call. What does that look like? What can somebody do when they get that bad news?

01:42:00.360 This drives me crazy, what you're talking about, in terms of access to expert opinion when you need

01:42:05.260 expert opinion, and particularly for complex, unique outlier cases, if you will. And you don't know as a

01:42:10.100 patient or a family member whether you're dealing with a middle-of-the-road case. How would you know?

01:42:14.280 So first off, we need to pool our insights, if you will. Break down the silos of hospitals and centers

01:42:20.120 and universities and whatever, and pool our opinions. That's point number one. Point number

01:42:26.440 two is, we need to leverage technology for this purpose. People get all excited about artificial

01:42:30.960 intelligence in terms of how it's informing chemistry advances and the like. And I'm excited

01:42:36.080 about those things too. And then other aspects of biology discovery and all that. But this is the

01:42:40.400 most obvious use, basically, right? Is that you start to build the database, essentially, of opinions

01:42:46.180 that I and others offer to specific cancer cases based on certain aspects of their diagnosis?

01:42:52.500 And the patterns are, these are not hard for a machine to figure out. I mean, a human could figure

01:42:57.280 out. It's sort of codifying what you do, what you do very easily as the teaching set for the AI.

01:43:03.520 Exactly. And what I'm getting at is within the 95%, you know, boundary of typical cases,

01:43:09.360 the decision support can really be based on, you know, the last hundred cases that I and my,

01:43:14.280 you know, other melanoma colleagues have seen that are just like this. And then the edge cases is

01:43:18.640 where we need to apply our specific attention. I think there are actually enough of us to handle

01:43:22.200 the edge cases. The problem is, like, the way our system works is, like, nobody knows what their

01:43:25.960 complexity of their diagnosis is. Everybody's seeking the same level of care and sort of decision

01:43:31.720 making without that understanding. And we can get way ahead of this and be transparent in explaining,

01:43:37.520 like, look, you know, here's why we're saying you've got a very typical case. We have a ton of

01:43:41.380 outcome data. We know what therapy is the very best. You know, the issue of therapeutic access

01:43:46.040 and investigational therapies that are crossing the divide and are showing real responses in real

01:43:50.380 human beings and should be considered as a certain priority, maybe not the top priority,

01:43:54.380 but a backup option or something, that's also not rocket science. And you shouldn't have to get on

01:43:59.580 an airplane and never go see anybody. But even on the Zoom screen, I mean, honestly, this is,

01:44:02.980 we are so inefficient in terms of how it is that we disseminate information. It drives me crazy.

01:44:08.320 There are very few entities, but are few working on this problem and kind of see it this way.

01:44:13.360 You throw some technology at this problem, I think this goes away.

01:44:16.460 So what is the best thing that one could do now? What are the companies that are out there that

01:44:19.840 are trying to do this now that are reputable in your mind?

01:44:22.500 N of One's been at this for a long time. This is still a model that they're, I think, quite good at,

01:44:27.400 but not the accumulating of the database and the, again, figuring out how to focus attention,

01:44:32.620 if you will. There's a company I know called X-Cures that's doing exactly this kind of work,

01:44:37.100 but it's still at the helping individual patients navigate level right now.

01:44:41.140 Yeah. It's not the full insight machine.

01:44:43.440 That's right. But it's going to come. I mean, this is another area that is underinvested.

01:44:47.340 We'd really like to see this because we, otherwise, as you know, we're kind of blowing

01:44:50.520 the bank on a very efficient system as it stands right now. And it's not scalable,

01:44:54.440 certainly not globally scalable.

01:44:56.260 Well, Keith, this was fantastic. Obviously we're going to sit down in four years again and

01:45:00.100 talk about the last four years, which will be going forward from here. And I have to say,

01:45:04.900 I find myself quite optimistic about what I see happening. And I think we'll be talking about

01:45:10.080 some big wins in four years. Again, I don't think we're quote unquote curing cancer, but I think

01:45:14.440 we're going to get a lot better at detecting it earlier, which gets people into a treatment pipeline

01:45:18.140 sooner. And I think we're going to continue to see probably incremental ways to harness the immune

01:45:23.180 system. That's probably where I see a lot of optimism. And again, I think that's in combination

01:45:27.600 with other traditional therapies and non-traditional therapies, such as the metabolic ones you

01:45:31.560 mentioned. Totally agree. The good news is that the technology curve continues to bend upward,

01:45:37.020 right? And so we talked about sequencing technology as an example there, but cell engineering advances,

01:45:42.160 ability to take new molecular targets and rapidly cycle that through to, you know, new drugs,

01:45:48.780 small molecules, antibodies, the like. All of these advances are converging in a way that if we keep

01:45:54.180 talking at four-year increments, the pace of progress is going to be substantially greater per unit time.

01:46:00.180 We've already witnessed that. There's no question that four-year increments over my career, that's been true.

01:46:05.500 It's reflected just in approval of drugs by the FDA for cancer. But now the convergence of diagnostics

01:46:10.940 and therapeutics, that's what's finally coming into view. That piece has really, I would say, been largely

01:46:15.820 missing. But if you link up all that we talked about today, I think that's the take-home message,

01:46:20.980 really, is it's the crossing of those wires that's what's really going to massively get us towards the path

01:46:26.820 of having many, much, much higher percentage of patients who are 10-year survivors to use that

01:46:30.800 benchmark again?

01:46:31.980 Well, Keith, thanks again. It's been great speaking with you. And I will speak a lot more in the next

01:46:35.340 four years, but I look forward to hosting you again back in four years and having the next increment

01:46:39.540 of time discussion.

01:46:40.620 Absolutely. Really enjoyed it. Be well.

01:46:42.220 Thank you for listening to this week's episode of The Drive. It's extremely important to me to

01:46:47.720 provide all of this content without relying on paid ads. To do this, our work is made entirely

01:46:52.700 possible by our members. And in return, we offer exclusive member-only content and benefits above

01:46:59.120 and beyond what is available for free. So if you want to take your knowledge of this space to the

01:47:03.460 next level, it's our goal to ensure members get back much more than the price of the subscription.

01:47:08.160 Premium membership includes several benefits. First, comprehensive podcast show notes that detail

01:47:14.940 every topic, paper, person, and thing that we discuss in each episode. And the word on the street

01:47:20.560 is nobody's show notes rival ours. Second, monthly Ask Me Anything or AMA episodes. These episodes are

01:47:28.640 comprised of detailed responses to subscriber questions typically focused on a single topic and are designed

01:47:34.600 to offer a great deal of clarity and detail on topics of special interest to our members. You'll also get

01:47:40.140 access to the show notes for these episodes, of course. Third, delivery of our premium newsletter,

01:47:46.060 which is put together by our dedicated team of research analysts. This newsletter covers a wide range of

01:47:51.820 topics related to longevity and provides much more detail than our free weekly newsletter. Fourth,

01:47:57.960 access to our private podcast feed that provides you with access to every episode, including AMA's

01:48:04.320 sans the spiel you're listening to now and in your regular podcast feed. Fifth, the qualies, an additional

01:48:11.740 member-only podcast we put together that serves as a highlight reel featuring the best excerpts from

01:48:18.020 previous episodes of The Drive. This is a great way to catch up on previous episodes without having to go

01:48:23.140 back and listen to each one of them. And finally, other benefits that are added along the way.

01:48:27.960 If you want to learn more and access these member-only benefits, you can head over to

01:48:32.480 peteratiamd.com forward slash subscribe. You can also find me on YouTube, Instagram, and Twitter,

01:48:39.220 all with the handle peteratiamd. You can also leave us a review on Apple Podcasts or whatever podcast

01:48:46.000 player you use. This podcast is for general informational purposes only and does not constitute the practice

01:48:52.120 of medicine, nursing, or other professional healthcare services, including the giving of medical advice.

01:48:57.300 No doctor-patient relationship is formed. The use of this information and the materials linked to this

01:49:03.600 podcast is at the user's own risk. The content on this podcast is not intended to be a substitute for

01:49:09.540 professional medical advice, diagnosis, or treatment. Users should not disregard or delay in obtaining

01:49:15.160 medical advice from any medical condition they have, and they should seek the assistance of their

01:49:20.120 healthcare professionals for any such conditions. Finally, I take all conflicts of interest very

01:49:25.480 seriously. For all of my disclosures and the companies I invest in or advise, please visit

01:49:31.080 peteratiamd.com forward slash about where I keep an up-to-date and active list of all disclosures.

01:49:38.120 I'll see you next time.

The Peter Attia Drive - August 21, 2023

#267 ‒ The latest in cancer therapeutics, diagnostics, and early detection ｜ Keith Flaherty, M.D.

Episode Stats

Length

Words per Minute

Word Count

Sentence Count

Misogynist Sentences

Hate Speech Sentences

Summary

Transcript